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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
1edb8be9-9f77-40b5-8909-3638403e0e69
|
pcace-a-statistical-approach-to-ranking
|
2112.15571
| null |
https://arxiv.org/abs/2112.15571v1
|
https://arxiv.org/pdf/2112.15571v1.pdf
|
PCACE: A Statistical Approach to Ranking Neurons for CNN Interpretability
|
In this paper we introduce a new problem within the growing literature of interpretability for convolution neural networks (CNNs). While previous work has focused on the question of how to visually interpret CNNs, we ask what it is that we care to interpret, that is, which layers and neurons are worth our attention? Due to the vast size of modern deep learning network architectures, automated, quantitative methods are needed to rank the relative importance of neurons so as to provide an answer to this question. We present a new statistical method for ranking the hidden neurons in any convolutional layer of a network. We define importance as the maximal correlation between the activation maps and the class score. We provide different ways in which this method can be used for visualization purposes with MNIST and ImageNet, and show a real-world application of our method to air pollution prediction with street-level images.
|
['Seth Flaxman', 'Esra Suel', 'Sílvia Casacuberta']
|
2021-12-31
| null | null | null | null |
['air-pollution-prediction']
|
['miscellaneous']
|
[ 3.85114878e-01 2.81988263e-01 2.88292140e-01 -6.28077030e-01
1.29927501e-01 -5.86421967e-01 4.76447880e-01 3.88522834e-01
-5.86027205e-01 5.32994449e-01 2.94688016e-01 -6.88234925e-01
-3.82832795e-01 -7.23946810e-01 -4.87776756e-01 -5.59844851e-01
-1.30907163e-01 2.22964864e-02 1.01033784e-01 -5.64840920e-02
1.72819436e-01 8.03425193e-01 -1.45148218e+00 6.09809756e-01
5.11767328e-01 1.07522702e+00 -8.66758525e-02 7.11190820e-01
-1.30644470e-01 6.81211233e-01 -7.46638060e-01 -2.45613962e-01
1.83082059e-01 -4.75810736e-01 -1.06509316e+00 -8.49247649e-02
7.18051434e-01 6.42066523e-02 2.65053511e-02 1.31128395e+00
2.21252307e-01 1.72596827e-01 8.07199121e-01 -1.21705997e+00
-5.61057448e-01 4.82953757e-01 -1.96735561e-01 8.29047620e-01
-3.33454072e-01 2.57401168e-01 1.17699873e+00 -4.41651165e-01
3.57802808e-01 1.45912397e+00 5.06292582e-01 3.82910907e-01
-1.44461679e+00 -3.26877683e-01 4.20863509e-01 1.44396439e-01
-1.06602538e+00 -1.54858172e-01 3.50855201e-01 -8.63651037e-01
7.86830485e-01 6.66552305e-01 8.45077693e-01 7.62547553e-01
-1.52348906e-01 3.61096770e-01 1.23871684e+00 -4.24221128e-01
4.54390436e-01 2.16906682e-01 5.25464416e-01 7.83310473e-01
5.77586532e-01 -7.17611983e-02 -1.92484245e-01 1.65653780e-01
1.05051899e+00 5.77518381e-02 -9.72106233e-02 -4.82582999e-03
-1.26793158e+00 8.17105532e-01 9.26737726e-01 4.40875918e-01
-2.56456554e-01 5.35339534e-01 1.27323017e-01 2.19619840e-01
5.73901236e-01 7.70990193e-01 -4.48393226e-01 3.05257320e-01
-6.96702898e-01 1.94816217e-01 6.11112595e-01 3.17791253e-01
6.73502505e-01 1.30757987e-01 -2.62740344e-01 6.53941035e-01
2.87210435e-01 1.95212662e-03 3.42280716e-02 -9.99926329e-01
3.03190053e-01 9.31910932e-01 -1.75447285e-01 -1.03795838e+00
-4.79941666e-01 -4.88100797e-01 -8.60701740e-01 1.06363440e+00
7.69612908e-01 -1.47841588e-01 -8.86145532e-01 1.43946731e+00
-1.40705720e-01 -3.85452732e-02 -1.68430403e-01 9.30028379e-01
9.35767949e-01 3.18822742e-01 3.98022830e-01 4.32762414e-01
1.62509000e+00 -7.43029416e-01 -2.89878100e-01 -3.70634675e-01
4.61271286e-01 -3.28060597e-01 1.15971994e+00 1.63466141e-01
-8.68697226e-01 -5.58078766e-01 -1.12563431e+00 1.46989385e-02
-9.53290701e-01 2.03488007e-01 4.84917074e-01 6.26320422e-01
-1.07185817e+00 8.42936277e-01 -8.38589072e-01 -4.06192958e-01
7.02819109e-01 3.52828801e-01 -3.05953205e-01 4.43762600e-01
-1.08583403e+00 1.10743570e+00 6.40326798e-01 2.79118419e-01
-4.94373769e-01 -5.12766898e-01 -5.87272167e-01 4.56544518e-01
-9.25912932e-02 -7.51410425e-01 1.20161605e+00 -1.37418664e+00
-7.26970851e-01 8.94750178e-01 -8.19277465e-02 -4.57627326e-01
3.35428029e-01 3.78973335e-02 -3.51177901e-01 8.43425915e-02
-2.80454755e-01 8.96856844e-01 5.03315330e-01 -1.23958826e+00
-8.64321649e-01 -2.77778238e-01 5.20013213e-01 7.28304386e-02
-5.42246878e-01 5.83727239e-03 -3.45852643e-01 -5.83763659e-01
1.15635999e-01 -9.19826686e-01 -5.25304735e-01 5.52518129e-01
-4.76832420e-01 -1.90927938e-01 6.40151203e-01 -5.24479270e-01
1.01274133e+00 -2.15967417e+00 -6.43486977e-02 4.72285271e-01
7.61953533e-01 3.95900071e-01 -4.19737473e-02 -2.30172113e-01
-5.68851948e-01 6.89323664e-01 -3.84032607e-01 -5.59423529e-02
5.93596473e-02 2.62269855e-01 -1.37580723e-01 2.83921093e-01
5.97680330e-01 8.87733519e-01 -1.02428067e+00 -3.31009835e-01
3.24569136e-01 7.01874793e-01 -2.74390727e-01 -1.48975611e-01
-9.06069577e-02 3.12320858e-01 -3.98761392e-01 1.44949540e-01
3.31182867e-01 -4.80975658e-01 1.40993604e-02 -2.28534922e-01
-3.26914668e-01 2.10492179e-01 -1.18872762e+00 8.08456481e-01
-7.75158107e-02 1.32315052e+00 -4.91542928e-02 -8.55184317e-01
5.93966246e-01 1.21619247e-01 5.71150817e-02 -7.47219801e-01
1.49172559e-01 1.90674849e-02 3.48862767e-01 -5.22182882e-01
2.74713159e-01 6.35884926e-02 4.45642591e-01 4.09556031e-01
-2.53979087e-01 3.43145728e-01 3.74421567e-01 -1.30133018e-01
1.10173154e+00 -3.15111279e-01 3.46496463e-01 -5.94434321e-01
3.42646211e-01 1.08066015e-01 6.14656769e-02 6.79443598e-01
-7.92217776e-02 8.43444109e-01 1.05015004e+00 -9.16327655e-01
-1.05434537e+00 -1.01959217e+00 -3.77975069e-02 1.01858759e+00
-2.53146261e-01 -2.15180174e-01 -1.01727867e+00 -6.04613304e-01
-3.41857783e-02 4.92518693e-01 -1.09965694e+00 1.88081846e-01
-5.73739111e-01 -9.68252420e-01 4.15388376e-01 7.09544301e-01
3.81760389e-01 -1.33394098e+00 -1.05335355e+00 -9.10046101e-02
1.13966748e-01 -8.81368279e-01 5.31698689e-02 4.63859767e-01
-1.11265671e+00 -1.34154785e+00 -5.73960423e-01 -6.92308187e-01
1.16442537e+00 1.47454008e-01 1.52281427e+00 4.77071941e-01
-3.65586638e-01 -1.57641638e-02 7.86636397e-02 -7.48025358e-01
-2.71895140e-01 1.84037164e-01 -5.01451075e-01 -1.14479251e-01
2.86638170e-01 -3.86635274e-01 -9.51396406e-01 4.60620485e-02
-1.19785452e+00 1.74017847e-01 6.71097279e-01 4.28778261e-01
4.71310169e-01 2.02631772e-01 7.13064224e-02 -1.02228367e+00
1.01179135e+00 -2.88099825e-01 -6.68702841e-01 2.22575426e-01
-5.17084658e-01 3.53991061e-01 4.76355255e-01 -9.77755487e-02
-5.57173789e-01 1.14158981e-01 -2.10354760e-01 -6.25226796e-02
-5.03788054e-01 4.38332647e-01 8.45705271e-02 -3.38838361e-02
8.52408409e-01 -3.66198897e-01 -3.19821090e-01 -5.94122708e-01
3.63438904e-01 2.06202015e-01 3.48938286e-01 -2.39154294e-01
7.21210063e-01 5.46408713e-01 2.48801455e-01 -8.56511176e-01
-7.17394471e-01 -2.88302094e-01 -9.07818735e-01 -1.99571192e-01
1.22133112e+00 -5.06837785e-01 -9.58280206e-01 1.34601751e-02
-1.34370303e+00 -4.02613044e-01 -3.16471338e-01 2.23629341e-01
-4.66441028e-02 -7.71073997e-02 -1.71378955e-01 -8.94937515e-01
-1.09541178e-01 -1.18478382e+00 7.45765448e-01 2.52509058e-01
-5.35990834e-01 -1.39094710e+00 -1.73366532e-01 -9.26515833e-02
5.77020288e-01 5.10840774e-01 1.41439283e+00 -5.30310273e-01
-5.57930171e-01 1.27873212e-01 -7.80931652e-01 4.83778268e-01
1.60065927e-02 8.10202658e-02 -1.21431434e+00 1.52872175e-01
-2.41337314e-01 1.82136998e-01 1.27955329e+00 4.49645907e-01
1.69674098e+00 -6.24460280e-01 -2.65045941e-01 3.72884244e-01
1.36976707e+00 5.73889501e-02 6.99935496e-01 3.97732377e-01
7.77307332e-01 9.40886319e-01 -1.94264874e-02 -6.20635115e-02
6.06076159e-02 5.16284704e-01 8.25442135e-01 -8.08447897e-01
-1.59048095e-01 1.02604978e-01 -1.14087872e-01 2.79261380e-01
-3.35030586e-01 -3.25863957e-01 -1.06647670e+00 5.28976619e-01
-1.86760020e+00 -9.14591849e-01 -4.55336601e-01 1.93863213e+00
4.29102421e-01 3.53544891e-01 1.17770676e-02 3.06091011e-01
6.54833317e-01 1.49955228e-01 -3.01929474e-01 -5.07028818e-01
5.23659103e-02 2.56096393e-01 4.75536644e-01 4.75114524e-01
-1.27498496e+00 4.17101681e-01 7.08906603e+00 4.40353662e-01
-1.19450557e+00 -1.56835735e-01 1.19163120e+00 -1.27891116e-02
-1.74684286e-01 -2.04939798e-01 -3.79223824e-01 3.44456851e-01
8.03216338e-01 3.84263247e-01 2.75473148e-01 7.05067456e-01
4.08042550e-01 -1.07182987e-01 -1.34913814e+00 8.71638358e-01
-3.89492720e-01 -1.52672589e+00 -5.81209324e-02 5.56147248e-02
6.23880088e-01 9.75164697e-02 2.78597176e-01 -3.17234039e-01
3.87968242e-01 -1.58373821e+00 7.40784228e-01 6.72325373e-01
5.08198500e-01 -4.82049257e-01 5.45234680e-01 1.23424421e-03
-1.17501307e+00 5.54978214e-02 -3.13910991e-01 -3.12109649e-01
-2.55398989e-01 6.40973985e-01 -9.17882919e-01 -1.37096047e-01
9.32684362e-01 3.83883774e-01 -9.89917219e-01 1.30355275e+00
-3.50865483e-01 6.44832432e-01 -1.54587835e-01 -2.62289971e-01
4.24233615e-01 -6.55353367e-02 2.46716633e-01 1.46399009e+00
1.67038992e-01 -1.28098145e-01 -3.69438946e-01 1.06213498e+00
-5.69794662e-02 -2.73326129e-01 -5.36197305e-01 -1.35912433e-01
3.28087844e-02 1.49301529e+00 -1.31892133e+00 -4.36753631e-01
-2.66752094e-01 6.52194142e-01 4.03828502e-01 5.95378458e-01
-4.93523210e-01 -4.24554825e-01 9.97974992e-01 3.37191373e-01
1.07682422e-01 -3.54686052e-01 -7.72262573e-01 -5.64023077e-01
-1.04708798e-01 -5.81362188e-01 2.61021584e-01 -9.62624609e-01
-1.10155880e+00 7.84955740e-01 5.45622548e-03 -1.01189995e+00
1.56729400e-01 -1.26713145e+00 -8.64493132e-01 1.03260756e+00
-1.52471507e+00 -5.27155697e-01 -4.66989815e-01 4.72548865e-02
2.80557930e-01 8.98453891e-02 8.69873405e-01 2.14244619e-01
-5.05521357e-01 2.80067045e-02 5.89681864e-02 4.71757621e-01
5.75013421e-02 -1.69622350e+00 8.95960271e-01 6.56377017e-01
4.01814371e-01 7.64768422e-01 9.29795504e-01 -1.01705439e-01
-5.73363245e-01 -1.20340240e+00 9.00966227e-01 -6.51365280e-01
5.45538902e-01 -2.95741767e-01 -7.86771238e-01 4.85468507e-01
3.64210725e-01 1.23354025e-01 6.71477258e-01 1.98487148e-01
-3.33670586e-01 -1.38639256e-01 -9.23190057e-01 8.96889925e-01
8.32341135e-01 -2.81848043e-01 -5.02633393e-01 1.95739597e-01
4.77004319e-01 -9.58095714e-02 -4.14232165e-01 1.88985124e-01
5.39358914e-01 -1.04325926e+00 1.05091488e+00 -8.81839395e-01
5.26149333e-01 -5.54430425e-01 2.58158743e-01 -1.48673093e+00
-4.61701035e-01 1.34989947e-01 2.31034458e-01 7.10642517e-01
8.01778793e-01 -6.37580633e-01 7.94295549e-01 5.68334520e-01
-3.63028049e-02 -7.85165548e-01 -7.62372077e-01 -4.51807052e-01
-8.56435075e-02 -3.56273055e-01 5.65712631e-01 7.91594446e-01
-3.55713844e-01 3.47525775e-01 8.97355601e-02 1.87694728e-01
3.46879303e-01 -3.61033171e-01 2.06496820e-01 -1.70634806e+00
2.13542618e-02 -9.54107523e-01 -5.12880504e-01 -5.68593442e-01
-1.72650635e-01 -8.77368152e-01 -3.96125764e-02 -1.89672804e+00
1.78061575e-01 -2.03760237e-01 -6.93862379e-01 7.56723642e-01
-1.66017190e-01 6.66290283e-01 2.88857013e-01 -2.63420939e-02
-4.84623790e-01 -1.00791447e-01 1.12148976e+00 -2.92375445e-01
8.71793851e-02 -1.76256612e-01 -9.37333047e-01 9.66294289e-01
9.63428497e-01 -5.45933783e-01 -3.21737379e-01 -7.38909781e-01
5.98283827e-01 -7.14700878e-01 9.78734910e-01 -1.21219873e+00
7.82112405e-02 -1.91620201e-01 9.23101962e-01 -3.10943961e-01
1.69068992e-01 -9.72910881e-01 1.13117151e-01 6.57712758e-01
-7.20075846e-01 5.25595844e-01 3.41185778e-01 3.09850514e-01
-1.58138365e-01 -3.35770100e-01 8.17041159e-01 -4.33070958e-01
-7.85790026e-01 1.16975673e-01 -5.05756438e-01 -4.90830687e-04
5.46730280e-01 -8.34188387e-02 -5.59348404e-01 -1.38873443e-01
-9.12397027e-01 3.54421660e-02 1.60586342e-01 4.28545207e-01
5.80581903e-01 -1.35133922e+00 -4.38840836e-01 7.18817785e-02
9.56109986e-02 -2.08645523e-01 -1.17509790e-01 5.80591202e-01
-8.56710613e-01 4.33579952e-01 -4.02662724e-01 -4.94831532e-01
-1.48542631e+00 7.71735013e-02 5.77984691e-01 -2.41430759e-01
-3.99618417e-01 7.70356953e-01 5.48406184e-01 1.05463155e-01
2.84400970e-01 -6.74985826e-01 -7.01679945e-01 1.18786909e-01
6.78019464e-01 3.96472633e-01 2.65141279e-01 -4.60034728e-01
-3.61452371e-01 2.43365914e-01 2.66822547e-01 -9.83799174e-02
1.39228010e+00 1.89805329e-01 -1.64294735e-01 5.00070393e-01
1.13972855e+00 -5.01691341e-01 -1.27490950e+00 1.48482665e-01
2.73054421e-01 -1.93300724e-01 2.77090985e-02 -8.62250566e-01
-1.29094362e+00 1.42383468e+00 1.01225924e+00 6.89030409e-01
9.63716507e-01 -6.42367154e-02 1.31712973e-01 6.08129025e-01
-5.05810320e-01 -1.02338648e+00 9.37848166e-02 6.28391623e-01
8.54721248e-01 -1.20068538e+00 -4.14692275e-02 -1.55210629e-01
-3.76516342e-01 1.33374023e+00 4.63707328e-01 -1.26716524e-01
7.41278887e-01 1.29540309e-01 1.72808990e-01 -5.87819040e-01
-4.88114804e-01 -5.67382753e-01 6.78219914e-01 5.46836913e-01
7.49726534e-01 1.61404267e-01 4.17457670e-02 1.81691825e-01
-2.84927726e-01 -2.17226759e-01 1.43775806e-01 5.88159502e-01
-8.47939074e-01 -8.05308402e-01 -3.43302757e-01 6.60206735e-01
-5.58078349e-01 -3.01542252e-01 -6.60782576e-01 7.14110136e-01
4.27375883e-01 7.27215052e-01 2.89962173e-01 -2.27760747e-01
4.11025047e-01 1.78499110e-02 1.97508723e-01 -6.99116647e-01
-7.11170971e-01 -3.94567609e-01 1.07110351e-01 -2.73764342e-01
-3.92189533e-01 -2.90958405e-01 -1.14558411e+00 -1.79507911e-01
1.94826603e-01 5.17672580e-03 8.11355770e-01 1.01582289e+00
-3.62201147e-02 9.92079735e-01 4.48782556e-02 -6.76158011e-01
1.10581875e-01 -7.02592075e-01 -4.16537285e-01 5.01690865e-01
6.38578475e-01 -3.79586995e-01 -5.03721535e-01 1.63678318e-01]
|
[8.882684707641602, 5.486849784851074]
|
5f3fd0a0-00a7-4642-adf0-318fb91ad03b
|
invisible-backdoor-attack-with-dynamic
|
2211.10933
| null |
https://arxiv.org/abs/2211.10933v2
|
https://arxiv.org/pdf/2211.10933v2.pdf
|
Invisible Backdoor Attack with Dynamic Triggers against Person Re-identification
|
In recent years, person Re-identification (ReID) has rapidly progressed with wide real-world applications, but also poses significant risks of adversarial attacks. In this paper, we focus on the backdoor attack on deep ReID models. Existing backdoor attack methods follow an all-to-one or all-to-all attack scenario, where all the target classes in the test set have already been seen in the training set. However, ReID is a much more complex fine-grained open-set recognition problem, where the identities in the test set are not contained in the training set. Thus, previous backdoor attack methods for classification are not applicable for ReID. To ameliorate this issue, we propose a novel backdoor attack on deep ReID under a new all-to-unknown scenario, called Dynamic Triggers Invisible Backdoor Attack (DT-IBA). Instead of learning fixed triggers for the target classes from the training set, DT-IBA can dynamically generate new triggers for any unknown identities. Specifically, an identity hashing network is proposed to first extract target identity information from a reference image, which is then injected into the benign images by image steganography. We extensively validate the effectiveness and stealthiness of the proposed attack on benchmark datasets, and evaluate the effectiveness of several defense methods against our attack.
|
['Cairong Zhao', 'Cheng Deng', 'Duoqian Miao', 'Dongsheng Li', 'Shuguang Dou', 'Xinyang Jiang', 'Wenli Sun']
|
2022-11-20
| null | null | null | null |
['image-steganography', 'open-set-learning']
|
['computer-vision', 'miscellaneous']
|
[ 2.22792551e-01 -3.06846648e-01 4.28701192e-02 -2.80219197e-01
-3.58668298e-01 -1.05202460e+00 6.51440024e-01 -1.88160494e-01
-3.74112815e-01 7.33972847e-01 -8.65594074e-02 -7.39485398e-02
5.54741807e-02 -1.18499899e+00 -8.24213207e-01 -9.11301136e-01
-1.22962177e-01 3.88231933e-01 4.98317853e-02 -3.69477063e-01
-9.41643026e-03 3.65768820e-01 -1.31835985e+00 1.89501364e-02
5.62943876e-01 5.50597191e-01 -4.77711052e-01 3.38847488e-01
1.68609753e-01 1.75371408e-01 -9.91263926e-01 -7.77631104e-01
7.17974663e-01 -5.02538443e-01 -5.15328407e-01 -5.30215740e-01
4.75597233e-01 -5.54785013e-01 -9.14720893e-01 1.37319207e+00
7.32371032e-01 -1.62179336e-01 3.69453818e-01 -1.58735156e+00
-8.67889404e-01 5.81675231e-01 -5.01386821e-01 3.06579471e-01
5.54945707e-01 2.46485770e-01 4.17223334e-01 -5.90718269e-01
3.99128556e-01 1.43819475e+00 5.91517746e-01 8.78096163e-01
-1.01477814e+00 -1.48259044e+00 1.30257547e-01 5.00494659e-01
-1.86005056e+00 -3.95288855e-01 5.17154336e-01 -9.33872908e-02
2.26023883e-01 5.73094845e-01 4.71499413e-01 1.40021122e+00
1.21750310e-01 3.19991022e-01 1.20915782e+00 -1.92148745e-01
-8.58230069e-02 3.85685980e-01 1.86024532e-01 3.80844325e-01
6.72641158e-01 7.32129216e-01 -2.75114924e-01 -5.71980536e-01
5.90236068e-01 1.68037504e-01 -4.78787422e-01 -3.01534459e-02
-1.08302474e+00 8.86139750e-01 3.92189592e-01 -9.96747538e-02
5.20263575e-02 2.97245546e-03 2.90176660e-01 3.63692760e-01
-6.28434122e-02 1.26700941e-02 -2.74524856e-02 3.89575750e-01
-4.22939330e-01 5.04152596e-01 9.04366970e-01 6.04130983e-01
8.49567056e-01 -1.77866995e-01 -1.59179717e-01 2.57333457e-01
2.31029138e-01 8.59674692e-01 5.51535368e-01 -2.98456639e-01
3.87088031e-01 3.10868323e-01 -6.74810447e-03 -1.42817879e+00
-1.86408181e-02 -6.14692390e-01 -9.59234238e-01 -1.01713955e-01
4.70360279e-01 -1.39452651e-01 -7.75148928e-01 2.08117080e+00
6.99544430e-01 7.50453353e-01 2.83492237e-01 7.44066060e-01
7.98699498e-01 6.72082603e-01 -7.96631947e-02 -4.12788168e-02
1.57130277e+00 -6.31332755e-01 -5.94137311e-01 -1.47952914e-01
4.45469230e-01 -5.30014515e-01 5.57764173e-01 7.26873353e-02
-4.47941720e-01 -5.08953393e-01 -1.20474589e+00 3.05318058e-01
-5.43446720e-01 -3.26752335e-01 2.77599156e-01 1.28250694e+00
-7.65851617e-01 4.27451916e-02 -2.56611466e-01 -2.56608903e-01
3.44535351e-01 7.88524032e-01 -7.64652789e-01 -3.66589934e-01
-1.79001617e+00 4.37088072e-01 5.47377229e-01 3.08406502e-01
-1.14784074e+00 -4.89223003e-01 -1.01291335e+00 1.88982394e-02
3.31075072e-01 -5.69553435e-01 7.41423249e-01 -7.19063997e-01
-1.09215486e+00 8.72457862e-01 -4.31371592e-02 -6.34072006e-01
5.90604603e-01 1.42766656e-02 -8.17206025e-01 -7.12604001e-02
1.87304780e-01 3.91952127e-01 1.02122521e+00 -1.31717110e+00
-3.52814853e-01 -5.96935809e-01 2.56022990e-01 -8.45576823e-02
-4.41898018e-01 7.46231079e-02 -2.02705637e-01 -7.38044739e-01
-1.90221056e-01 -1.14913642e+00 -2.03329101e-02 -2.63803393e-01
-6.69113934e-01 1.03593193e-01 1.27878523e+00 -5.26918054e-01
1.04039717e+00 -2.23991203e+00 -4.11583871e-01 5.97102106e-01
1.37969792e-01 6.28400207e-01 -2.30093226e-01 3.35283458e-01
-2.33666554e-01 2.24959701e-01 2.93546915e-02 -1.14057079e-01
-8.68657529e-02 3.17759335e-01 -6.78022921e-01 6.73706889e-01
-3.63577813e-01 7.79959500e-01 -8.54072392e-01 -3.92347634e-01
3.66810597e-02 6.31289482e-01 -3.55987549e-01 1.74492076e-01
3.61334831e-01 6.00383103e-01 -3.91999274e-01 5.58834672e-01
1.14139986e+00 3.10093653e-03 3.53123881e-02 -6.18714988e-02
2.46925607e-01 -1.88830018e-01 -1.32409918e+00 8.11488748e-01
-1.23803420e-02 2.31078133e-01 -2.32749075e-01 -6.83070898e-01
8.90651703e-01 3.48343164e-01 1.04151070e-01 -6.06744170e-01
1.66941807e-01 1.43694431e-01 2.00414121e-01 -2.18731493e-01
3.39918524e-01 6.26169220e-02 -2.99736768e-01 5.24290621e-01
-2.38604769e-01 8.32930744e-01 -2.50544339e-01 1.18902326e-01
8.63596678e-01 -2.79767483e-01 1.36917159e-01 -8.26771706e-02
1.02561271e+00 -2.42902994e-01 8.63000154e-01 1.04848778e+00
-3.79013300e-01 5.33016324e-01 1.84916884e-01 -6.42267466e-01
-7.80629218e-01 -1.11338913e+00 -2.44735897e-01 6.00287855e-01
7.09240258e-01 -3.13693136e-01 -8.74705672e-01 -1.11868310e+00
3.03154349e-01 2.14660242e-01 -7.01423705e-01 -4.81768340e-01
-7.12772012e-01 -7.85007000e-01 1.22754896e+00 5.94245642e-02
1.12454998e+00 -7.28789926e-01 9.30698216e-02 6.27106279e-02
-3.57315689e-01 -1.13600671e+00 -7.75096357e-01 -7.58304775e-01
-2.13360742e-01 -1.36208045e+00 -7.16991663e-01 -8.68325055e-01
8.98551643e-01 5.47287405e-01 3.89403790e-01 2.42470309e-01
-2.75740027e-01 1.74400926e-01 -5.48911579e-02 -4.16255474e-01
-4.72618729e-01 -1.45459071e-01 5.39948463e-01 6.61025286e-01
5.29685199e-01 -2.65290767e-01 -7.93511212e-01 8.41983318e-01
-8.18893313e-01 -4.34012383e-01 3.64630073e-01 9.06130314e-01
3.78354490e-01 4.55746502e-01 6.77953362e-01 -8.34583044e-01
3.79396647e-01 -5.93308568e-01 -5.89504182e-01 3.29753786e-01
-2.84734726e-01 -6.84691891e-02 7.50423670e-01 -8.69007826e-01
-6.76847279e-01 -5.71764670e-02 -2.25054801e-01 -4.71266091e-01
1.26837594e-02 7.41300136e-02 -6.46876156e-01 -6.46809399e-01
5.73012829e-01 6.15380406e-01 2.70055309e-02 -1.72275022e-01
1.48719504e-01 7.40985870e-01 9.57523346e-01 -4.62916672e-01
1.59805536e+00 7.09067464e-01 -3.18260081e-02 -5.40700674e-01
-3.82221192e-01 -1.86089486e-01 -8.66031200e-02 -1.18529990e-01
5.85907519e-01 -1.02463067e+00 -1.15183997e+00 1.02291811e+00
-1.10560143e+00 1.90066114e-01 1.04552142e-01 4.15960848e-01
1.46406293e-01 7.30897367e-01 -3.69098186e-01 -5.37191987e-01
-5.62672436e-01 -1.17900789e+00 6.52572691e-01 5.82945406e-01
7.96798989e-02 -8.42826843e-01 -7.97351822e-02 3.29603493e-01
2.81723499e-01 5.75063944e-01 6.04655564e-01 -1.14146399e+00
-6.56193078e-01 -6.30103648e-01 -1.32900730e-01 1.70265555e-01
3.52216452e-01 -4.62297469e-01 -9.85872865e-01 -7.86700368e-01
1.36835575e-01 -4.05198336e-02 7.12895393e-01 -2.08081722e-01
1.01229465e+00 -7.99812555e-01 -6.68191254e-01 9.06838119e-01
1.15071046e+00 2.37583533e-01 7.90825009e-01 6.20037735e-01
8.24470758e-01 5.45516491e-01 4.73386765e-01 4.00168747e-01
6.25907540e-01 7.14585125e-01 3.56648386e-01 2.48619057e-02
2.58673191e-01 -5.00636160e-01 4.24705923e-01 1.29056573e-02
2.07555071e-01 -3.78094435e-01 -5.67483366e-01 4.22271729e-01
-1.55934453e+00 -1.09049737e+00 1.41762763e-01 2.57342219e+00
7.57609487e-01 7.94074982e-02 9.87441391e-02 3.73855680e-02
1.35960007e+00 1.71671674e-01 -5.38158953e-01 -8.31433609e-02
-1.86562076e-01 8.14476833e-02 6.71852589e-01 3.03226590e-01
-1.20088804e+00 8.38761628e-01 5.38271427e+00 7.16132939e-01
-1.21579218e+00 1.52345598e-01 5.26384592e-01 1.83687419e-01
-1.76376715e-01 1.97282538e-01 -1.26977003e+00 8.18737507e-01
6.18283868e-01 -4.63615447e-01 4.07354742e-01 6.51917279e-01
-2.57658243e-01 3.54597390e-01 -1.16525662e+00 1.13176751e+00
3.24008912e-01 -1.12893701e+00 3.10800523e-01 4.71529454e-01
6.22544587e-01 -5.88279068e-01 4.10233021e-01 3.56858224e-01
2.09660992e-01 -9.44498062e-01 4.34650809e-01 6.10913150e-02
9.86857355e-01 -1.05453336e+00 8.60322893e-01 3.38355571e-01
-1.26815569e+00 -2.87999481e-01 -4.86881644e-01 2.66237944e-01
-1.37265041e-01 8.19961280e-02 -5.32244325e-01 7.29479194e-01
5.94515920e-01 1.91448107e-01 -4.92927551e-01 1.00571752e+00
-2.81455398e-01 3.57986510e-01 -5.01690626e-01 2.43257165e-01
8.52930844e-02 2.47970238e-01 6.82726860e-01 9.21686292e-01
3.11535209e-01 2.08259284e-01 2.27641121e-01 7.34076619e-01
-2.98606634e-01 -5.38939796e-02 -7.89749146e-01 2.72807062e-01
9.35110569e-01 9.47279513e-01 -3.12998235e-01 -2.37324774e-01
-4.16798294e-02 1.03173876e+00 -2.57291108e-01 3.41678679e-01
-1.01108479e+00 -6.59041703e-01 1.03420472e+00 2.90813986e-02
9.11057964e-02 -6.95508346e-02 3.30101907e-01 -1.37983835e+00
-8.41001347e-02 -1.20678830e+00 7.37132013e-01 -1.01390153e-01
-1.23830903e+00 4.86914158e-01 -5.03600501e-02 -1.30666041e+00
-1.13881484e-01 -1.45173490e-01 -8.90056610e-01 9.18306530e-01
-1.53435481e+00 -1.33957529e+00 -3.48504514e-01 1.13799906e+00
-1.06325753e-01 -3.82132798e-01 9.88734543e-01 4.06506926e-01
-8.00263464e-01 1.51368558e+00 1.15692683e-01 6.99563861e-01
8.54897320e-01 -6.26628995e-01 7.86757171e-01 1.16616571e+00
2.90154256e-02 1.01265132e+00 4.94962424e-01 -7.72738576e-01
-1.52579319e+00 -1.22192848e+00 6.84075058e-01 -2.36819997e-01
2.46192366e-01 -7.35236347e-01 -9.87801015e-01 9.06712592e-01
-2.28061914e-01 2.68075407e-01 6.69182539e-01 -4.37242389e-01
-8.33472133e-01 -3.21684420e-01 -1.60669470e+00 4.49391067e-01
8.55142593e-01 -6.78739190e-01 -3.87495697e-01 2.52672136e-01
6.45502865e-01 -4.54443157e-01 -4.76733327e-01 3.86936903e-01
7.83398271e-01 -5.66921175e-01 1.37355244e+00 -3.97083372e-01
-2.62089640e-01 -4.09273088e-01 5.27636260e-02 -9.21809018e-01
-1.30667746e-01 -8.81863058e-01 -1.39625818e-01 1.41577196e+00
-2.36396119e-01 -1.39284754e+00 9.56358433e-01 2.68758953e-01
5.32379448e-01 -1.85579598e-01 -1.17365539e+00 -1.06047404e+00
-4.09467407e-02 1.37911752e-01 1.26704943e+00 1.15463984e+00
-4.58521724e-01 -1.88493684e-01 -6.91870868e-01 9.76580620e-01
1.21883857e+00 -1.66445971e-01 1.15757322e+00 -1.20073628e+00
-2.74368107e-01 1.24030419e-01 -8.16381574e-01 -7.52672374e-01
2.72066057e-01 -7.78285921e-01 -3.45992476e-01 -7.85170615e-01
2.15472430e-01 -6.28152549e-01 -4.54430252e-01 5.22710443e-01
-3.07883769e-01 6.62598610e-01 1.96820885e-01 4.55176383e-01
-7.08734915e-02 4.11105514e-01 9.34459150e-01 -3.83017391e-01
-4.18646894e-02 1.71512112e-01 -7.81208277e-01 3.49866480e-01
6.59007132e-01 -7.23348618e-01 -4.70224887e-01 -9.73221734e-02
-1.64573997e-01 -1.03467852e-01 7.47527361e-01 -1.12467527e+00
3.90872508e-01 9.46317837e-02 4.36243057e-01 -3.63242716e-01
2.67383188e-01 -7.05781877e-01 4.14443523e-01 8.92210901e-01
6.64026514e-02 -2.46732067e-02 1.81897014e-01 7.09259629e-01
6.73624873e-02 7.10181370e-02 7.13855624e-01 2.33167931e-02
-8.02368879e-01 6.15316033e-01 1.88212529e-01 -1.27633378e-01
1.30378580e+00 -5.33664763e-01 -8.46125901e-01 -3.07297856e-01
-5.45143485e-01 4.33672518e-01 4.86990690e-01 5.86815536e-01
7.36446619e-01 -1.38266993e+00 -9.24283087e-01 8.08845818e-01
2.68080086e-01 -3.09001595e-01 5.53479612e-01 3.00475895e-01
-4.16460037e-01 2.55087376e-01 -3.04077089e-01 -4.35957253e-01
-1.62869930e+00 8.27992797e-01 5.55525243e-01 -1.18752159e-01
-5.16956091e-01 7.74880290e-01 6.87572181e-01 -4.74964261e-01
2.13961542e-01 5.97163320e-01 -2.34312296e-01 -3.05504322e-01
9.83921230e-01 2.44326547e-01 -4.29552257e-01 -1.10302889e+00
-4.91686940e-01 6.16020739e-01 -5.20208061e-01 7.32354224e-02
6.52745962e-01 -1.16554491e-01 -2.39423454e-01 -4.03006196e-01
1.27854657e+00 1.22913472e-01 -7.27871060e-01 -4.33595717e-01
-4.19549465e-01 -8.70345473e-01 -4.35194254e-01 -5.07497549e-01
-9.23762858e-01 5.48367500e-01 9.95580912e-01 9.09182131e-02
8.77108634e-01 -3.23622406e-01 1.33002293e+00 4.82032835e-01
6.49997771e-01 -4.99513328e-01 -1.31312072e-01 2.50178903e-01
4.64710951e-01 -1.12518704e+00 -1.41916260e-01 -4.61632878e-01
-2.43501768e-01 8.36621463e-01 7.11870134e-01 -2.89274156e-02
6.46574557e-01 -5.88681139e-02 7.78369457e-02 2.06633493e-01
-1.77007496e-01 1.61059633e-01 -1.28268644e-01 8.20916176e-01
-6.01109028e-01 4.33673449e-02 -1.19247988e-01 5.82382441e-01
-3.89735222e-01 -1.88235641e-01 5.13320506e-01 6.95627749e-01
-3.14413868e-02 -1.49160230e+00 -8.96724880e-01 -1.20648667e-01
-6.44779861e-01 1.35630727e-01 -2.98915476e-01 7.38844037e-01
3.80320668e-01 8.78152013e-01 -1.66749269e-01 -8.73776495e-01
1.11418292e-01 -1.98056117e-01 2.12819442e-01 -4.13024485e-01
-6.54900432e-01 -4.75792140e-01 -2.16537505e-01 -2.42076442e-01
1.24580711e-01 -3.77882600e-01 -1.08700049e+00 -7.95793116e-01
-4.35817033e-01 3.86192709e-01 5.27801871e-01 7.88655341e-01
3.54718745e-01 -5.76035902e-02 1.20126593e+00 -5.09546161e-01
-7.26491332e-01 -4.12109882e-01 -4.11357462e-01 3.08802068e-01
4.46144611e-01 -6.20133400e-01 -4.30917859e-01 -2.94348538e-01]
|
[13.213784217834473, 1.0628199577331543]
|
8c7fcc4f-8983-46eb-9c2b-243957a9123c
|
experimental-assessment-of-polynomial
|
2011.08520
| null |
https://arxiv.org/abs/2011.08520v1
|
https://arxiv.org/pdf/2011.08520v1.pdf
|
Experimental assessment of polynomial nonlinear state-space and nonlinear-mode models for near-resonant vibrations
|
In the present paper, two existing nonlinear system identification methodologies are used to identify data-driven models. The first methodology focuses on identifying the system using steady-state excitations. To accomplish this, a phase-locked loop controller is implemented to acquire periodic oscillations near resonance and construct a nonlinear-mode model. This model is based on amplitude-dependent modal properties, i.e. does not require nonlinear basis functions. The second methodology exploits uncontrolled experiments with broadband random inputs to build polynomial nonlinear state-space models using advanced system identification tools. The methods are applied to two experimental test rigs, a magnetic cantilever beam and a free-free beam with a lap joint. The respective models of both methods and both specimens are then challenged to predict dynamic, near-resonant behavior observed under different sine and sine-sweep excitations. The vibration prediction of the nonlinear-mode and state-space models clearly highlight the capabilities and limitations of the models. The nonlinear-mode model, by design, yields a perfect match at resonance peaks and high accuracy in close vicinity. However, it is limited to well-spaced modes and sinusoidal excitation. The state-space model covers a wider dynamic range, including transient excitations. However, the real-life nonlinearities considered in this study can only be approximated by polynomial basis functions. Consequently, the identified state-space models are found to be highly input-dependent, in particular for sinusoidal excitations where they are found to lead to a low predictive capability.
|
['Malte Krack', 'Matthew S. Allen', 'Jean-Philippe Noël', 'Simon Peter', 'Matthew R. W. Brake', 'Ali Tatar', 'Gleb Kleyman', 'Maren Scheel']
|
2020-11-17
| null | null | null | null |
['cantilever-beam']
|
['miscellaneous']
|
[ 4.24716324e-01 -1.25663340e-01 -7.95674324e-02 4.22061205e-01
-5.48005402e-01 -5.96380949e-01 3.45696002e-01 -3.49133551e-01
1.44040018e-01 6.58815622e-01 -4.63665336e-01 -1.48934603e-01
-8.76665890e-01 -3.42643440e-01 -3.27858388e-01 -9.57315326e-01
-2.80749388e-02 2.29690507e-01 1.30260155e-01 -4.29178029e-01
1.96525961e-01 5.68650961e-01 -1.65195692e+00 -3.82156402e-01
5.97332597e-01 8.23151231e-01 9.24882144e-02 6.87594175e-01
5.54148436e-01 2.85567313e-01 -4.95635420e-01 5.63064277e-01
-1.63148679e-02 -5.08029640e-01 -5.55765569e-01 8.93321261e-02
-1.79533377e-01 -3.69688123e-01 -1.30357042e-01 7.91114628e-01
6.17063224e-01 2.37286463e-01 7.26501524e-01 -9.68487978e-01
-3.31665128e-01 2.48315662e-01 -1.10233359e-01 2.91032661e-02
5.33196926e-01 2.83943802e-01 6.81850612e-01 -1.05317867e+00
4.18601096e-01 7.22399533e-01 9.52393591e-01 5.11179686e-01
-1.90849745e+00 -4.87301111e-01 -6.57193005e-01 -9.95202363e-02
-1.41635621e+00 -4.08315927e-01 1.55096674e+00 -7.14637220e-01
7.80507863e-01 4.45859492e-01 4.57285315e-01 7.34463155e-01
3.88413191e-01 -1.53695807e-01 1.27844810e+00 -6.59923732e-01
1.22465014e-01 1.87347189e-01 4.01407838e-01 3.21901947e-01
9.58090499e-02 7.26169050e-01 -2.39928842e-01 -3.49846184e-01
1.13814259e+00 -3.23354691e-01 -6.24275684e-01 -3.11335772e-01
-8.47299457e-01 5.23785949e-01 7.38700852e-02 9.16869938e-01
-5.92748940e-01 -9.67826843e-02 1.06910236e-01 4.40799385e-01
-4.62868214e-02 6.57675505e-01 -1.93202808e-01 -1.94385886e-01
-8.95845056e-01 3.09503049e-01 8.79628897e-01 3.46361130e-01
7.00001001e-01 7.26207793e-01 2.35535651e-01 6.85564399e-01
3.99463475e-01 8.10394883e-01 5.11548102e-01 -7.63213158e-01
-1.32267982e-01 3.23587388e-01 4.78658974e-01 -1.07295215e+00
-8.86670351e-01 -3.97732049e-01 -7.37681150e-01 2.07236394e-01
5.33751190e-01 -3.30275983e-01 -6.20455563e-01 1.57430327e+00
2.52857298e-01 -9.13086757e-02 2.96773165e-01 9.90432858e-01
5.78310072e-01 7.54393518e-01 -4.67159152e-01 -8.96174133e-01
1.16647112e+00 -3.01539361e-01 -1.08636475e+00 1.27062142e-01
1.50274679e-01 -7.78043866e-01 1.31293511e+00 3.66611093e-01
-1.09233952e+00 -8.28200877e-01 -8.95506978e-01 7.51319826e-01
1.35052294e-01 2.95187384e-01 -3.05362314e-01 4.86034989e-01
-6.94100380e-01 6.75846398e-01 -1.01221704e+00 -1.41451329e-01
-7.50694871e-01 2.18110532e-01 -7.52608804e-03 6.21109426e-01
-1.48072159e+00 9.23852205e-01 3.19944382e-01 3.90492409e-01
-4.71682906e-01 -1.00797594e+00 -3.55790824e-01 7.68816248e-02
1.03569716e-01 -2.78924555e-01 1.21074808e+00 -5.69903851e-01
-2.12326002e+00 1.86303973e-01 -2.41845429e-01 -2.56298799e-02
7.20225498e-02 1.34673845e-02 -7.84678757e-01 3.74863952e-01
-3.18526387e-01 -5.65936148e-01 1.00363386e+00 -1.52463293e+00
2.37688914e-01 3.01724583e-01 -4.96426940e-01 -2.43773952e-01
-4.71224010e-01 -1.44948483e-01 3.81959349e-01 -4.75376576e-01
4.45402533e-01 -1.06842661e+00 -6.21223375e-02 -6.69238567e-01
-5.94086707e-01 1.73585340e-01 9.79788840e-01 -4.41272646e-01
1.54795718e+00 -2.00065804e+00 4.80081746e-03 6.56209707e-01
-2.87431359e-01 2.56812960e-01 2.58516282e-01 1.25674832e+00
-3.68668735e-01 -1.52730912e-01 -2.54845023e-01 4.35700297e-01
-1.16537422e-01 -1.00202203e-01 -2.78756022e-01 3.80339652e-01
4.83541012e-01 5.50290525e-01 -3.88891101e-01 1.15148157e-01
2.35896602e-01 4.97927606e-01 -5.03732800e-01 2.58986920e-01
2.48762533e-01 7.36954689e-01 -3.70997757e-01 5.98859191e-01
3.83468777e-01 -4.71049584e-02 1.00530326e-01 -7.94729590e-01
-6.31210029e-01 -1.41931131e-01 -1.51139867e+00 8.55494201e-01
-5.04372537e-01 4.64833945e-01 4.76031542e-01 -1.31493497e+00
1.34711063e+00 9.08913732e-01 6.48553431e-01 -4.42246944e-01
1.79897696e-01 7.73150265e-01 2.89846629e-01 -8.79334748e-01
1.06892429e-01 -5.33823133e-01 1.05693445e-01 2.42668405e-01
8.41427594e-02 -4.28349167e-01 1.10362405e-02 -5.08806944e-01
5.93392611e-01 6.15664423e-02 2.72745252e-01 -6.73946321e-01
1.03900695e+00 1.66382819e-01 4.39498335e-01 2.92302072e-01
2.52122015e-01 3.52268308e-01 2.48522341e-01 -8.21702033e-02
-9.26261663e-01 -7.57172167e-01 -5.64321995e-01 4.93351728e-01
2.03883663e-01 2.44205758e-01 -5.70534647e-01 6.12472594e-01
-7.70295188e-02 7.31940567e-01 -1.91984251e-01 -6.75246000e-01
-8.63982737e-01 -6.65021598e-01 3.99002343e-01 2.46940196e-01
-2.03425735e-01 -8.76558363e-01 -7.75762320e-01 6.30048990e-01
-6.91822991e-02 -8.87640119e-01 -1.01722680e-01 3.60243171e-01
-8.72555256e-01 -1.15804470e+00 -5.22881389e-01 -7.76436031e-01
5.90256691e-01 -5.38063705e-01 7.33307362e-01 -8.85393918e-02
-3.09433490e-01 7.50844657e-01 -9.47825909e-02 -1.92454457e-01
-8.78351927e-01 -2.74349868e-01 6.01955533e-01 1.81105912e-01
-1.93533301e-01 -7.25148857e-01 -5.07405937e-01 8.32849026e-01
-7.32093155e-01 -5.28035998e-01 5.98302126e-01 1.18397033e+00
5.98666966e-01 5.24448395e-01 1.02100277e+00 -4.20227200e-01
8.30738485e-01 -1.84395894e-01 -8.55978966e-01 -1.48858935e-01
-7.17386603e-01 -2.16536388e-01 9.80684876e-01 -9.70271945e-01
-1.00821590e+00 8.62752125e-02 -1.10967703e-01 -6.34306312e-01
-2.10543290e-01 9.40926909e-01 6.06581233e-02 -3.27705532e-01
8.80403697e-01 5.99141657e-01 6.38381243e-01 -6.08117044e-01
-2.54139751e-01 6.66163981e-01 9.74220455e-01 -7.35600889e-01
1.24376261e+00 -1.28121570e-01 3.86411309e-01 -1.32942522e+00
-1.05160773e-01 -3.36485445e-01 -4.08372551e-01 -5.90358317e-01
2.13518336e-01 -4.07411575e-01 -1.11019540e+00 7.96481311e-01
-7.67668426e-01 -2.69812226e-01 -5.07945061e-01 8.28655660e-01
-8.74813974e-01 1.59888282e-01 -7.38519847e-01 -1.48016524e+00
-5.36675394e-01 -1.10272944e+00 6.25500739e-01 3.34754586e-01
-4.74697620e-01 -1.11918294e+00 2.30022222e-01 -3.10734119e-02
7.46337295e-01 7.46819615e-01 1.08956301e+00 -3.71993929e-01
-3.31578888e-02 -5.24923265e-01 7.34388888e-01 2.49220014e-01
2.11422116e-01 4.89970297e-01 -8.09185028e-01 -6.65819824e-01
6.46352470e-01 -1.53792739e-01 -3.81367728e-02 5.25154173e-01
1.56970173e-01 -2.10475937e-01 -1.58735916e-01 8.06743558e-03
1.84844220e+00 4.31280166e-01 3.09593022e-01 -1.39143970e-02
3.95917684e-01 7.07006812e-01 5.38218439e-01 2.22194329e-01
-4.08490121e-01 8.34641039e-01 4.72901225e-01 -5.45737073e-02
2.56896347e-01 5.44450656e-02 4.28806812e-01 1.05331886e+00
-2.82114625e-01 1.02822594e-01 -1.05922747e+00 5.01037180e-01
-1.21490741e+00 -9.22171235e-01 -5.74500442e-01 2.20119452e+00
8.63550901e-01 1.08794980e-01 1.89904273e-01 8.65859151e-01
7.67454445e-01 -3.95228833e-01 -5.13635159e-01 -3.75880688e-01
6.56872839e-02 2.88748920e-01 1.20988600e-01 4.87641931e-01
-7.12608695e-01 1.03525013e-01 6.65828705e+00 4.03476715e-01
-1.67353857e+00 -3.80476952e-01 -1.80407658e-01 2.20303133e-01
-6.66715503e-02 -3.55622582e-02 -7.28427231e-01 5.01189649e-01
1.34175849e+00 -3.85136455e-01 5.12507319e-01 6.42741680e-01
6.96013153e-01 1.18349260e-02 -7.33083963e-01 5.45921147e-01
-4.34786171e-01 -1.10798597e+00 -4.57914174e-01 -1.12315483e-01
6.54997349e-01 -7.67202973e-01 5.62958159e-02 -4.03068289e-02
-7.03523755e-01 -7.24898100e-01 6.99136734e-01 1.01752841e+00
9.26729083e-01 -6.58800304e-01 6.98472917e-01 7.88235784e-01
-1.06857359e+00 -4.27548796e-01 6.86715078e-03 -1.66194454e-01
7.11370111e-01 6.44074202e-01 -3.25072736e-01 6.43128216e-01
3.35042536e-01 2.36118808e-01 2.58224726e-01 6.78217709e-01
2.93510407e-01 1.23886275e+00 -4.86382723e-01 -9.30259898e-02
-1.86726183e-01 -2.89410681e-01 8.74007046e-01 6.18505955e-01
5.39402664e-01 3.29174131e-01 7.31889755e-02 1.25059021e+00
9.40193355e-01 -1.17435530e-01 -2.13061839e-01 -1.65953845e-01
5.43981433e-01 1.08734989e+00 -1.24038905e-01 3.26353431e-01
-2.71002412e-01 -2.15342175e-02 -6.50039852e-01 6.04707301e-01
-7.64886677e-01 -5.60047448e-01 3.59279096e-01 6.18063450e-01
1.44932747e-01 -3.87503117e-01 -3.80564630e-02 -5.67485809e-01
-1.16777405e-01 -6.61060631e-01 -2.67841369e-01 -6.15379155e-01
-1.12180281e+00 5.17897129e-01 4.83740151e-01 -1.70179760e+00
-9.38632071e-01 -6.63695514e-01 -8.87884259e-01 1.14531803e+00
-1.27486742e+00 -8.22357178e-01 -1.56004772e-01 5.91535687e-01
-4.50885333e-02 -3.69005650e-02 9.84026432e-01 4.19177890e-01
-5.83476424e-01 2.81373054e-01 3.78718793e-01 -8.93857628e-02
3.03184986e-01 -9.11914408e-01 -3.70393306e-01 6.39761746e-01
-6.82460606e-01 9.59126592e-01 1.30356061e+00 -3.32699209e-01
-1.71381485e+00 -6.32553577e-01 4.88466650e-01 1.63829014e-01
8.75206590e-01 1.84584726e-02 -1.48283899e+00 3.87527347e-02
-1.29225135e-01 1.76745683e-01 3.90535831e-01 -6.51448548e-01
4.79613483e-01 -1.51926288e-02 -1.05137348e+00 2.14891493e-01
1.93932772e-01 -6.06078506e-01 -7.29557812e-01 5.30854091e-02
3.13115537e-01 -4.62218255e-01 -1.50066972e+00 7.22266495e-01
5.40419698e-01 -5.54665089e-01 8.74991298e-01 -2.05479890e-01
2.11891666e-01 -5.92747569e-01 1.48482054e-01 -1.31849456e+00
-4.97341335e-01 -1.09873521e+00 -3.78882378e-01 1.37566590e+00
3.73679549e-01 -1.09362924e+00 4.99847054e-01 4.90665138e-01
-8.54895711e-02 -7.59255767e-01 -9.48567510e-01 -1.23548329e+00
7.89139867e-02 1.22485757e-02 5.05045541e-02 6.76970482e-01
3.00062984e-01 2.70994961e-01 -1.60804093e-01 5.07250190e-01
5.02638817e-01 2.94427782e-01 5.60858667e-01 -1.38937676e+00
-1.49034947e-01 -1.66991323e-01 -1.95805043e-01 -5.73482096e-01
-1.38582319e-01 -1.55158848e-01 2.56761640e-01 -7.73266494e-01
-4.89484668e-01 -3.14970523e-01 -2.30379507e-01 2.34794900e-01
-7.22266808e-02 2.41563246e-01 -2.50943244e-01 5.16082168e-01
9.25755322e-01 4.48265016e-01 1.03994751e+00 1.02185898e-01
-5.74906766e-01 5.41408420e-01 1.78128593e-02 5.56883752e-01
7.97440469e-01 -2.11163774e-01 -6.34958148e-01 3.55210692e-01
7.74027482e-02 5.29972374e-01 6.18162513e-01 -1.37349904e+00
2.40263537e-01 -2.05964282e-01 -1.74946278e-01 -5.57156026e-01
1.45803511e-01 -1.21464360e+00 9.79379177e-01 8.78610015e-01
-5.55084385e-02 -1.02166116e-01 3.29804778e-01 4.49955940e-01
-5.39576590e-01 -2.66574949e-01 1.01528192e+00 4.70059454e-01
-2.68110275e-01 -5.53545237e-01 -6.80698156e-01 -2.66544223e-01
7.97995865e-01 -7.04061508e-01 -5.49766719e-02 -2.93591768e-01
-9.68411386e-01 -2.65255392e-01 2.49351710e-01 -6.67304918e-02
2.59135872e-01 -1.32578945e+00 -5.30210555e-01 6.09212339e-01
-2.82197654e-01 -3.02796572e-01 5.04661918e-01 1.26733506e+00
-3.68628621e-01 6.08260691e-01 -9.73152518e-02 -9.71903920e-01
-1.05632877e+00 4.48714644e-01 7.01358736e-01 -8.05878043e-02
-1.86209083e-01 2.58120477e-01 -3.65334481e-01 -3.75669360e-01
-4.16645139e-01 -4.24676806e-01 -3.22634071e-01 5.43962233e-02
7.07353875e-02 5.00675678e-01 1.97757170e-01 -8.88636351e-01
-2.03110188e-01 1.20803916e+00 7.39351213e-01 -7.99705088e-02
1.15753448e+00 2.78150924e-02 -1.28897028e-02 7.82907665e-01
1.08820319e+00 7.32174292e-02 -1.04642165e+00 -2.17881650e-01
-1.79255292e-01 8.61647800e-02 -3.11383847e-02 -5.79007030e-01
-9.10692394e-01 4.48474288e-01 5.83421946e-01 9.03458059e-01
1.45291495e+00 -3.93394738e-01 4.60491985e-01 2.15700746e-01
2.65168268e-02 -9.74655867e-01 2.48419736e-02 4.04999197e-01
1.06044924e+00 -6.25414610e-01 -2.59529036e-02 -6.01553142e-01
-4.20595556e-02 1.49295485e+00 5.02610564e-01 -2.25372419e-01
8.59184623e-01 4.67722565e-01 4.98834290e-02 -4.98414785e-02
-4.49087530e-01 2.87242591e-01 5.19285977e-01 4.00331408e-01
5.18570483e-01 -3.84166062e-01 -6.28459096e-01 6.99098527e-01
-1.32893160e-01 1.60417542e-01 5.22821724e-01 1.04074991e+00
-4.24095839e-01 -8.37109923e-01 -1.01521504e+00 1.46123737e-01
-3.97416472e-01 4.29222643e-01 7.08987489e-02 1.15329731e+00
-5.05097389e-01 1.03710437e+00 6.47465661e-02 -3.57147068e-01
1.07373178e+00 2.65744060e-01 -4.60097864e-02 -2.00025469e-01
-6.00984931e-01 4.20407921e-01 -6.20367341e-02 -3.86102408e-01
-5.34897745e-01 -5.57489634e-01 -1.32768238e+00 -7.67133338e-03
-1.08808184e+00 3.03474128e-01 4.40029263e-01 5.81832051e-01
8.46557245e-02 7.34864652e-01 9.14876282e-01 -1.12382913e+00
-1.05518579e+00 -1.00985384e+00 -7.74595380e-01 2.60193974e-01
5.35644829e-01 -8.69478941e-01 -7.86144674e-01 1.74097463e-01]
|
[5.853240489959717, 3.0029609203338623]
|
76f6873b-6374-49aa-9b90-e1f84fb1aeff
|
deep-pixel-wise-binary-supervision-for-face
|
1907.04047
| null |
https://arxiv.org/abs/1907.04047v1
|
https://arxiv.org/pdf/1907.04047v1.pdf
|
Deep Pixel-wise Binary Supervision for Face Presentation Attack Detection
|
Face recognition has evolved as a prominent biometric authentication modality. However, vulnerability to presentation attacks curtails its reliable deployment. Automatic detection of presentation attacks is essential for secure use of face recognition technology in unattended scenarios. In this work, we introduce a Convolutional Neural Network (CNN) based framework for presentation attack detection, with deep pixel-wise supervision. The framework uses only frame level information making it suitable for deployment in smart devices with minimal computational and time overhead. We demonstrate the effectiveness of the proposed approach in public datasets for both intra as well as cross-dataset experiments. The proposed approach achieves an HTER of 0% in Replay Mobile dataset and an ACER of 0.42% in Protocol-1 of OULU dataset outperforming state of the art methods.
|
['Anjith George', 'Sebastien Marcel']
|
2019-07-09
| null | null | null | null |
['face-presentation-attack-detection']
|
['computer-vision']
|
[ 4.84987199e-01 -3.20694059e-01 -3.59340869e-02 -1.47046655e-01
-6.03996098e-01 -8.84541392e-01 6.24351025e-01 2.43577920e-03
-5.48414230e-01 3.80488366e-01 -1.64583862e-01 -7.18057096e-01
1.59499258e-01 -6.74201369e-01 -5.60874045e-01 -7.62318194e-01
-1.52252927e-01 -5.26062787e-01 -1.15690291e-01 3.26111093e-02
2.22910672e-01 9.34355140e-01 -1.27856469e+00 4.87799019e-01
3.63389449e-03 1.60800242e+00 -7.38429964e-01 9.23638821e-01
3.92866313e-01 2.76073605e-01 -9.71785724e-01 -8.66720855e-01
4.77818936e-01 7.03638270e-02 -4.53219295e-01 -1.49734452e-01
6.81276321e-01 -6.98791683e-01 -6.80155933e-01 7.78589308e-01
1.07258785e+00 -5.51153421e-01 2.75967091e-01 -1.44617140e+00
-3.00679475e-01 2.56877899e-01 -9.58580196e-01 5.27441859e-01
5.36834598e-01 1.72946557e-01 3.23786855e-01 -6.05769575e-01
2.45254755e-01 9.53943193e-01 6.10467136e-01 6.94850922e-01
-9.71144676e-01 -1.15722358e+00 -5.57148814e-01 1.86057806e-01
-1.55089843e+00 -6.30817652e-01 7.78265774e-01 5.02978601e-02
8.04279983e-01 1.13296628e-01 1.15258612e-01 1.37208986e+00
1.78054586e-01 4.90124762e-01 1.22634554e+00 -4.03452665e-01
-2.32950002e-02 -3.51094152e-03 8.33722949e-02 6.08103037e-01
5.68986118e-01 1.16140522e-01 -7.56048262e-01 -2.83703029e-01
4.53975737e-01 1.02720261e-01 -8.40741247e-02 4.72785652e-01
-5.76934934e-01 5.01168072e-01 1.02918409e-01 2.22887009e-01
-3.88046592e-01 2.88475752e-01 6.41119719e-01 2.59845465e-01
3.04770749e-02 -5.52286565e-01 -4.07872349e-01 -2.49094993e-01
-9.08629596e-01 4.99277227e-02 5.36492825e-01 5.48957467e-01
1.83149815e-01 3.02222878e-01 -6.85504004e-02 4.35421050e-01
4.07097191e-01 7.75836825e-01 3.42726320e-01 -2.94673502e-01
4.74303484e-01 4.10459340e-01 -2.96978951e-02 -1.06548548e+00
-1.72575504e-01 -2.96964832e-02 -8.96419287e-01 1.81279659e-01
4.57608104e-01 -5.28463900e-01 -1.04129291e+00 1.49790680e+00
1.88183919e-01 6.56455994e-01 7.42318481e-02 2.84826636e-01
7.21781254e-01 3.19913536e-01 2.98235804e-01 -1.02959245e-01
1.64156771e+00 -1.70278117e-01 -6.49728000e-01 5.58182120e-01
3.61490697e-01 -1.00522494e+00 5.77441335e-01 5.81323206e-01
-8.79492819e-01 -3.81798863e-01 -1.34799373e+00 3.77197593e-01
-4.35578704e-01 1.40612155e-01 5.02643049e-01 1.97721577e+00
-1.11182761e+00 1.89773083e-01 -5.92118084e-01 -3.31931442e-01
1.15317559e+00 1.09196067e+00 -7.39403605e-01 2.26603616e-02
-9.72090304e-01 2.81658798e-01 -1.49908112e-02 1.69423133e-01
-7.54868686e-01 -4.36513692e-01 -6.72540724e-01 4.62097451e-02
-3.97311598e-02 1.64720923e-01 1.14949441e+00 -6.47963524e-01
-1.46548998e+00 8.42735946e-01 -3.93669270e-02 -7.25789607e-01
3.15864950e-01 -2.50316560e-01 -8.66541803e-01 3.30169052e-01
-5.69698513e-01 4.22390521e-01 1.01943779e+00 -8.14128995e-01
-4.22749877e-01 -5.36178529e-01 1.17210992e-01 -4.39961106e-01
-7.52547085e-01 5.31232476e-01 -4.97387499e-02 -5.27941346e-01
-2.56283879e-01 -9.43413556e-01 5.50838590e-01 5.52504882e-03
-2.33775973e-01 -7.49398693e-02 1.75957632e+00 -8.36619496e-01
1.01626921e+00 -1.99607635e+00 -7.53430545e-01 3.49737585e-01
-4.53634858e-02 8.93913150e-01 1.32900067e-02 4.91998225e-01
-1.49288684e-01 3.38187158e-01 8.84939730e-02 -2.21657783e-01
-1.66612506e-01 -2.12507471e-01 -2.67978996e-01 7.71601260e-01
3.18349570e-01 7.62146175e-01 -2.22524881e-01 -9.37181562e-02
1.77062705e-01 1.05621767e+00 -1.59730613e-01 1.18025608e-01
3.99649948e-01 1.16109170e-01 -1.90335006e-01 1.15194356e+00
1.28297317e+00 1.47367507e-01 1.72398791e-01 -2.58922637e-01
2.21438393e-01 -6.46263286e-02 -1.26296282e+00 1.17379630e+00
-2.71631390e-01 8.26054752e-01 4.26158048e-02 -5.42372704e-01
6.52827203e-01 8.23412180e-01 1.99769393e-01 -7.87342370e-01
7.13946164e-01 -8.73502530e-03 1.19271487e-01 -4.56856936e-01
1.06246032e-01 2.73664534e-01 1.01242483e-01 6.83206558e-01
7.96907097e-02 9.86340284e-01 -2.52261549e-01 9.57730785e-02
1.20964420e+00 -1.66130036e-01 1.62708625e-01 -8.02272037e-02
7.10577250e-01 -8.55172276e-01 1.33890852e-01 6.53912544e-01
-5.99442422e-01 2.16641590e-01 2.71203637e-01 -6.97085679e-01
-7.01998174e-01 -8.49505544e-01 -1.55651003e-01 7.15363681e-01
-9.44668204e-02 -3.68440181e-01 -1.07842183e+00 -1.07896686e+00
-1.28189802e-01 -1.97702944e-01 -7.13468552e-01 1.12152390e-01
-6.85898483e-01 -6.89045608e-01 1.42066860e+00 4.93296504e-01
1.09006643e+00 -1.01262891e+00 -9.02451694e-01 -2.04266652e-01
1.80149049e-01 -1.54818034e+00 -2.41749734e-01 -3.74553531e-01
-4.14728105e-01 -1.29639816e+00 -6.76814139e-01 -5.06206870e-01
5.77393651e-01 2.44007930e-01 4.12963182e-01 4.99900281e-01
-6.74696863e-01 4.98089135e-01 -1.89114571e-01 -5.86038828e-01
-5.85679039e-02 -1.24963839e-02 1.27023116e-01 5.98917484e-01
6.51418090e-01 -3.60860288e-01 -1.03559041e+00 2.39124104e-01
-1.12031424e+00 -7.97931194e-01 3.21856499e-01 6.38680398e-01
-2.70561606e-01 1.25016689e-01 6.09042406e-01 -7.55473435e-01
7.01755464e-01 -2.37280712e-01 -5.86034358e-01 2.19676033e-01
-4.21220601e-01 -4.32483763e-01 2.90717870e-01 -5.98275125e-01
-9.74210024e-01 1.92872733e-01 -7.28504360e-02 -1.13002984e-02
-5.52609146e-01 6.28165603e-02 -4.62953210e-01 -7.12037265e-01
4.97542471e-01 1.61164999e-01 -2.19016239e-01 -2.27540225e-01
-3.83027233e-02 1.07937133e+00 4.14163142e-01 -4.24801111e-01
9.09638226e-01 5.66866398e-01 2.63652295e-01 -1.11283815e+00
-2.98896562e-02 -2.42648855e-01 -3.31311852e-01 -2.48707250e-01
4.77341771e-01 -7.94308782e-01 -1.56865048e+00 1.02934134e+00
-1.16962326e+00 2.00358063e-01 7.59521365e-01 -5.59549518e-02
5.41446619e-02 7.95350492e-01 -7.44185865e-01 -1.22069418e+00
-8.22276711e-01 -1.15713775e+00 1.11857426e+00 4.33731854e-01
4.05254867e-03 -6.04061306e-01 -3.01292419e-01 4.53920245e-01
7.86399841e-01 5.71834922e-01 1.97859943e-01 -7.32713759e-01
-4.45819616e-01 -9.00753438e-01 -5.50580442e-01 1.52520895e-01
3.37405384e-01 5.55919856e-03 -1.62013745e+00 -5.73174775e-01
-2.42785230e-01 -3.65544319e-01 5.61179459e-01 1.55525833e-01
1.28020275e+00 -4.82710928e-01 -4.13753949e-02 5.01182258e-01
1.58368063e+00 2.62186319e-01 1.11628139e+00 3.15553218e-01
4.11240399e-01 3.33656222e-01 2.82364488e-02 7.13055849e-01
-3.04093845e-02 6.47976279e-01 2.75331259e-01 1.19014631e-03
7.66748264e-02 1.02673799e-01 2.48507708e-01 -2.43171602e-01
3.59442271e-02 -5.45610547e-01 -9.00414586e-01 2.20488921e-01
-1.41027987e+00 -1.04058850e+00 1.13453515e-01 2.24859619e+00
3.15345287e-01 8.74926045e-04 3.76883119e-01 7.48865128e-01
7.34859228e-01 -1.96536314e-02 -1.96471944e-01 -7.63332903e-01
-1.64862812e-01 9.37859654e-01 7.45645523e-01 1.51363760e-01
-1.31965590e+00 8.03527594e-01 5.59652424e+00 6.03427947e-01
-1.48861623e+00 1.80572897e-01 1.02951300e+00 1.34071708e-01
5.25298417e-01 -5.38280547e-01 -9.12087739e-01 5.74135900e-01
1.41760027e+00 3.14450800e-01 9.07151699e-02 4.59229678e-01
8.62734914e-02 -1.26581505e-01 -6.67677283e-01 1.43249917e+00
3.16207290e-01 -1.47020388e+00 -5.06407171e-02 4.51867878e-01
4.00875032e-01 -4.34697211e-01 5.79226315e-01 -3.60752881e-01
-3.11400503e-01 -1.31518030e+00 1.66374400e-01 -1.88971177e-01
1.11228526e+00 -1.18649948e+00 1.22538030e+00 -2.73080051e-01
-1.28121722e+00 -3.58768702e-02 1.04292564e-01 2.00050309e-01
1.32878730e-02 -1.50165290e-01 -9.25310731e-01 2.64079243e-01
6.25366509e-01 3.58461179e-02 -5.56567729e-01 8.93221498e-01
1.04933880e-01 9.70198929e-01 -5.97279251e-01 1.94746211e-01
3.30925025e-02 6.79117680e-01 5.80199882e-02 1.45163691e+00
4.64559227e-01 1.16788745e-01 -2.19813004e-01 -1.15547515e-01
-4.81008112e-01 2.71985587e-02 -7.15929985e-01 -7.44618401e-02
4.24300283e-01 1.32103038e+00 -8.55683148e-01 3.14486586e-02
-3.80273730e-01 1.14887285e+00 -3.56171578e-01 2.42849633e-01
-8.23068917e-01 -6.26513898e-01 5.15663266e-01 1.56555682e-01
3.99655372e-01 -2.61819571e-01 -4.04840678e-01 -8.33749771e-01
1.81019992e-01 -1.04002321e+00 6.62807703e-01 -1.89037360e-02
-7.82804310e-01 7.86487937e-01 -2.13620320e-01 -1.05241883e+00
-1.55534551e-01 -8.79266977e-01 -6.28430724e-01 8.92746747e-01
-1.60113120e+00 -1.72411728e+00 -3.40239137e-01 9.36503232e-01
1.58334494e-01 -6.20667934e-01 9.55996633e-01 5.74560821e-01
-7.20182478e-01 1.51879466e+00 -4.70802486e-01 6.72390640e-01
5.43884456e-01 -7.39286184e-01 5.65317392e-01 1.18611598e+00
6.23953491e-02 8.97553623e-01 3.20516646e-01 -5.16768396e-01
-1.69086158e+00 -7.57930219e-01 5.28012931e-01 -2.63593495e-01
3.44590187e-01 -4.51502532e-01 -5.62831581e-01 2.39540145e-01
5.70429385e-01 2.33390033e-01 1.20303655e+00 -3.94133896e-01
-8.52548897e-01 -3.14936638e-01 -1.82623065e+00 4.65263933e-01
3.79567027e-01 -7.46014059e-01 3.44391048e-01 -6.71330616e-02
1.01563372e-01 -2.12390736e-01 -7.16300964e-01 3.78774196e-01
1.19909143e+00 -9.14864480e-01 9.47170377e-01 -4.32882041e-01
-1.77361965e-01 -2.34888494e-01 -3.53502661e-01 -1.14444442e-01
4.12256747e-01 -1.23648441e+00 -5.25206923e-01 1.46960354e+00
3.33559245e-01 -4.30258244e-01 1.38345301e+00 7.29732156e-01
7.34233439e-01 -3.48646551e-01 -1.12388170e+00 -5.72501481e-01
-3.58879685e-01 -5.33698618e-01 5.78320265e-01 5.97831845e-01
-3.01609606e-01 -1.43783301e-01 -6.34257674e-01 6.68623328e-01
9.47180271e-01 -6.66445076e-01 7.80613065e-01 -1.03550029e+00
6.83863238e-02 7.48254068e-04 -1.05542338e+00 -2.40660533e-01
-5.51540293e-02 -1.47728667e-01 -6.14159942e-01 -4.89327550e-01
2.65733153e-01 1.88016668e-01 -5.46375573e-01 7.03560770e-01
3.40887129e-01 1.15067124e+00 2.52803892e-01 -3.30609381e-01
-1.99125186e-01 -6.41478375e-02 1.98190749e-01 -6.81498572e-02
-1.38903586e-02 3.15296575e-02 -5.18852353e-01 3.67173463e-01
1.22689939e+00 -2.41554886e-01 -2.48368099e-01 -1.44711835e-02
-2.39248008e-01 -2.14252427e-01 3.19360942e-01 -1.12553930e+00
1.51589110e-01 4.81589884e-02 8.42238724e-01 -6.10436499e-01
2.90274948e-01 -9.89851952e-01 -1.13053396e-01 5.71602225e-01
-1.75576702e-01 5.54393232e-01 7.58488595e-01 4.26909536e-01
2.48372182e-01 1.50967821e-01 7.95060158e-01 3.48613650e-01
-4.03730959e-01 4.68493551e-01 -2.43006259e-01 -5.27515352e-01
1.02354133e+00 -7.14045942e-01 -5.64552844e-01 -3.60537797e-01
-3.49876076e-01 -5.12646258e-01 1.96262076e-01 4.00534272e-01
9.73770618e-01 -1.05697477e+00 -5.60857356e-01 5.84133983e-01
4.86580990e-02 -9.22463119e-01 3.48307639e-01 3.11673701e-01
-7.49296546e-01 4.33763057e-01 -5.78671694e-01 -4.14412439e-01
-2.23513603e+00 2.35821277e-01 1.51361451e-01 2.30409592e-01
-3.49363446e-01 6.36739552e-01 -3.52994472e-01 2.34966129e-01
4.94734079e-01 3.18471968e-01 -1.62041754e-01 -2.86251575e-01
1.39391565e+00 3.23622912e-01 4.96898115e-01 -9.52347934e-01
-5.28147757e-01 3.27472597e-01 -3.54381710e-01 -3.73807728e-01
9.82220411e-01 1.23034589e-01 1.15390897e-01 -4.59751129e-01
1.44625628e+00 8.69626775e-02 -9.85009491e-01 1.00088216e-01
2.59643383e-02 -7.05643594e-01 1.20581709e-01 -9.75315332e-01
-1.15702558e+00 9.75098252e-01 1.47411418e+00 1.49242312e-01
1.21634686e+00 -5.74906170e-01 1.06256843e+00 5.05084813e-01
2.10297331e-01 -5.61359763e-01 4.87533920e-02 -5.20394593e-02
3.76369029e-01 -1.29040182e+00 -8.83781239e-02 -3.11262250e-01
-3.03239018e-01 1.27465796e+00 4.83003646e-01 7.46457651e-02
1.02323353e+00 5.70629656e-01 2.24705130e-01 -1.31159604e-01
-4.10848796e-01 1.10239610e-01 2.24721789e-01 9.76117313e-01
4.69149709e-01 -1.62357792e-01 -6.63034990e-02 1.92346215e-01
1.51863441e-01 1.39931262e-01 5.59438229e-01 1.38666272e+00
3.12636457e-02 -1.39511609e+00 -4.02322829e-01 9.51098353e-02
-1.37015224e+00 1.10714152e-01 -4.71125066e-01 6.48154438e-01
1.55300394e-01 1.22909462e+00 -2.71974713e-01 -5.60422957e-01
-2.47905143e-02 8.81823972e-02 4.26004618e-01 4.47565084e-03
-9.33386862e-01 7.27579072e-02 -3.33542041e-02 -4.28572685e-01
-5.55846632e-01 -3.78764093e-01 -7.18914449e-01 -6.88458383e-01
-2.03017756e-01 -2.83768833e-01 1.07950485e+00 8.65941763e-01
5.96491873e-01 -3.23071890e-02 8.36485445e-01 -5.15207589e-01
-2.33830392e-01 -7.21167147e-01 -3.43312770e-01 2.92990804e-01
4.46937650e-01 -3.76711756e-01 1.68685749e-01 2.44535297e-01]
|
[13.070781707763672, 1.1432968378067017]
|
475409d4-4376-4b56-b267-2903f935878d
|
tad-transfer-learning-based-multi-adversarial
|
2210.15700
| null |
https://arxiv.org/abs/2210.15700v1
|
https://arxiv.org/pdf/2210.15700v1.pdf
|
TAD: Transfer Learning-based Multi-Adversarial Detection of Evasion Attacks against Network Intrusion Detection Systems
|
Nowadays, intrusion detection systems based on deep learning deliver state-of-the-art performance. However, recent research has shown that specially crafted perturbations, called adversarial examples, are capable of significantly reducing the performance of these intrusion detection systems. The objective of this paper is to design an efficient transfer learning-based adversarial detector and then to assess the effectiveness of using multiple strategically placed adversarial detectors compared to a single adversarial detector for intrusion detection systems. In our experiments, we implement existing state-of-the-art models for intrusion detection. We then attack those models with a set of chosen evasion attacks. In an attempt to detect those adversarial attacks, we design and implement multiple transfer learning-based adversarial detectors, each receiving a subset of the information passed through the IDS. By combining their respective decisions, we illustrate that combining multiple detectors can further improve the detectability of adversarial traffic compared to a single detector in the case of a parallel IDS design.
|
['Wim Mees', 'Tayeb Kenaza', 'Jean-Michel Dricot', 'Thibault Debatty', 'Richard Bauwens', 'Islam Debicha']
|
2022-10-27
| null | null | null | null |
['network-intrusion-detection']
|
['miscellaneous']
|
[ 1.95237041e-01 -9.53049958e-03 1.51577219e-01 8.84868857e-03
-3.90443951e-01 -6.89220488e-01 9.65646505e-01 -8.47778320e-02
-5.12982130e-01 4.31062996e-01 -3.59871626e-01 -8.20634007e-01
1.59832241e-04 -1.06707966e+00 -6.95340037e-01 -4.52574760e-01
-4.68203515e-01 7.24967957e-01 5.83106816e-01 -7.66345620e-01
3.21971923e-01 1.16586828e+00 -1.20557034e+00 4.17291611e-01
1.20385721e-01 6.41165316e-01 -7.24384665e-01 9.35250401e-01
-7.06187636e-02 8.25168133e-01 -1.24062574e+00 -2.27502868e-01
6.07955754e-01 -3.35446745e-01 -5.54677665e-01 -2.33831450e-01
4.05820131e-01 -4.72392619e-01 -7.52944171e-01 8.78401995e-01
5.68202436e-01 9.29587614e-03 6.97008491e-01 -1.63990176e+00
-3.26773226e-01 5.65356970e-01 -1.28214210e-01 6.15361810e-01
2.66675383e-01 4.87679124e-01 5.28708518e-01 -2.24485949e-01
2.29541376e-01 1.44433248e+00 6.55045569e-01 9.18070376e-01
-1.27723932e+00 -1.07004702e+00 5.66433519e-02 7.35230595e-02
-1.04817545e+00 -5.21813035e-01 9.44615185e-01 -2.30178475e-01
9.54066873e-01 1.55851141e-01 -9.17611346e-02 1.51446688e+00
3.83823812e-01 4.22585130e-01 9.80977416e-01 -5.98413765e-01
3.52641225e-01 3.76104802e-01 -7.20994398e-02 5.62159836e-01
2.38975227e-01 8.48330021e-01 3.15212250e-01 -4.15201187e-01
5.17182946e-01 -6.44101724e-02 2.09424317e-01 -1.47768363e-01
-4.43342507e-01 1.08215189e+00 6.53083980e-01 6.12945616e-01
-4.04493213e-01 1.98510662e-01 7.39431202e-01 8.33862066e-01
2.69889325e-01 7.93819368e-01 -5.24800241e-01 2.03763202e-01
-4.74630892e-01 1.55543908e-01 1.09263098e+00 4.26653326e-01
4.74209428e-01 7.12004066e-01 -2.80256361e-01 2.96593159e-01
2.36732867e-02 5.47341943e-01 1.59291491e-01 -4.22033548e-01
1.08338796e-01 4.22000587e-01 7.18015879e-02 -9.12098527e-01
-5.23635983e-01 -3.36968273e-01 -4.21660990e-01 7.49438763e-01
4.38479662e-01 -6.14952564e-01 -9.27639723e-01 1.59248412e+00
2.15636328e-01 3.67315471e-01 1.85660779e-01 3.35541070e-01
-4.85644490e-02 4.53849018e-01 2.66928256e-01 1.34087771e-01
1.03280795e+00 -5.14944315e-01 -2.18081772e-01 -2.43168443e-01
5.55199742e-01 -5.91598451e-01 5.47257900e-01 2.25454256e-01
-5.41949332e-01 -5.93082428e-01 -1.17349744e+00 1.03347635e+00
-8.70286345e-01 -7.19059765e-01 1.79583743e-01 1.43896723e+00
-8.40809107e-01 5.89468241e-01 -6.53936923e-01 -2.98392475e-01
3.79174620e-01 6.35336220e-01 -1.15655690e-01 2.26932600e-01
-1.58765590e+00 1.18154490e+00 2.89129794e-01 -5.84167004e-01
-1.45711911e+00 -4.59938645e-01 -4.83106107e-01 1.23392202e-01
3.64513844e-01 -3.99919927e-01 1.35713863e+00 -1.05024052e+00
-1.31251860e+00 8.00997615e-01 5.30621469e-01 -9.13060784e-01
4.98136967e-01 4.23984155e-02 -9.97637153e-01 2.16334730e-01
-2.50254363e-01 9.50983837e-02 1.26991987e+00 -1.42346895e+00
-5.97116530e-01 -1.85170114e-01 3.82180870e-01 -5.20675480e-01
-5.91515064e-01 4.86255199e-01 2.92937189e-01 -5.21897793e-01
-7.44890034e-01 -8.43542457e-01 -4.66234297e-01 -3.74105453e-01
-4.64871138e-01 1.16113894e-01 1.49655604e+00 -1.04863606e-01
9.54989791e-01 -2.13737011e+00 -2.28015959e-01 5.17636895e-01
8.53990391e-02 1.14700556e+00 -3.05835664e-01 5.51643372e-01
-3.86378855e-01 3.53708416e-01 -4.62660901e-02 2.66747251e-02
1.96505353e-01 2.45372698e-01 -5.18690288e-01 4.19537008e-01
5.90634167e-01 5.31683505e-01 -7.55308867e-01 -3.18104215e-02
5.83391249e-01 3.06572407e-01 -5.40776372e-01 4.21557784e-01
-1.42407402e-01 5.01676381e-01 -7.94061482e-01 4.80774730e-01
4.64282155e-01 3.84073675e-01 -5.87892719e-02 2.61672646e-01
2.58331686e-01 7.63379931e-02 -9.14839327e-01 4.58857387e-01
-4.45650548e-01 4.80607331e-01 4.91501987e-02 -1.30189610e+00
1.02677214e+00 4.38546747e-01 5.18454969e-01 -8.88812423e-01
5.63429475e-01 6.85829595e-02 4.54049140e-01 -1.27686217e-01
-1.50752723e-01 -4.54763681e-01 -3.93042594e-01 7.28832603e-01
1.60504747e-02 3.74228358e-01 -3.22236889e-03 9.67082232e-02
1.89848006e+00 -7.43839443e-01 2.30190963e-01 1.09632492e-01
9.80877697e-01 -1.95284367e-01 1.24968998e-01 1.29563451e+00
-5.76241434e-01 -1.93406105e-01 5.55030465e-01 -7.92084217e-01
-1.14538693e+00 -1.20630777e+00 2.58439127e-02 1.22701681e+00
-3.83701622e-01 5.67168603e-03 -9.41181004e-01 -1.24820852e+00
2.99622774e-01 9.69422817e-01 -7.03115702e-01 -8.78479481e-01
-8.98971379e-01 -6.96009457e-01 1.42074037e+00 4.69452590e-01
4.82777953e-01 -1.42990386e+00 -5.51843941e-01 3.60256791e-01
6.77190542e-01 -1.10234272e+00 4.05210480e-02 4.56232041e-01
-2.96073645e-01 -1.25627470e+00 -8.97504911e-02 -5.26553392e-01
3.27523887e-01 -1.28775463e-01 1.03831530e+00 1.78955019e-01
-4.37725514e-01 5.01065016e-01 -5.36849260e-01 -7.54660249e-01
-1.21384358e+00 1.15538232e-01 5.90228558e-01 5.81805557e-02
6.94185078e-01 -5.67131102e-01 -1.59289479e-01 4.11918372e-01
-1.14107561e+00 -1.10903478e+00 6.41787648e-01 8.25097501e-01
-2.53675431e-01 2.91027099e-01 8.81665945e-01 -1.11008501e+00
8.73319268e-01 -6.35191917e-01 -6.57449663e-01 -7.56508410e-02
-6.38963401e-01 1.41182750e-01 1.33426166e+00 -7.12364018e-01
-7.92324126e-01 -3.43360603e-01 -3.70257407e-01 -6.75442636e-01
-6.47791624e-01 -1.45391211e-01 -1.61004603e-01 -7.02436268e-01
1.29545593e+00 1.66233063e-01 6.74415752e-02 -1.10486105e-01
2.63525248e-01 6.55419230e-01 1.73814714e-01 -7.11399913e-01
1.36395192e+00 2.59260088e-01 1.44814014e-01 -6.97168350e-01
-5.31435907e-01 -1.40122294e-01 -4.07172799e-01 -2.31068552e-01
3.11634243e-01 -1.85304552e-01 -7.60762930e-01 5.22986472e-01
-7.99011707e-01 -3.50017458e-01 -3.86011571e-01 -1.18426554e-01
-5.74730158e-01 1.38162166e-01 -7.81646967e-01 -7.88005471e-01
-3.03976744e-01 -1.21001315e+00 5.01637280e-01 -2.45598610e-02
1.43461013e-02 -1.17949009e+00 4.29820091e-01 -1.01458624e-01
9.00673211e-01 3.74015301e-01 8.98143172e-01 -1.70587218e+00
-1.71531692e-01 -8.98062527e-01 5.15846871e-02 6.76682889e-01
1.01603143e-01 4.19208109e-02 -1.13836908e+00 -5.87287903e-01
1.33983761e-01 -1.13417700e-01 5.82545877e-01 -1.26523699e-03
8.63382697e-01 -3.94743413e-01 -5.15513778e-01 3.90845567e-01
1.21100330e+00 6.94550991e-01 6.24644637e-01 8.25053632e-01
5.14962614e-01 3.88069838e-01 2.47085854e-01 3.22050244e-01
-5.92846811e-01 5.85397303e-01 6.16031289e-01 -1.91445008e-01
1.93168446e-01 6.88089151e-03 5.56405544e-01 -4.32460159e-02
3.33260983e-01 -4.98119742e-01 -1.09631574e+00 7.95712620e-02
-1.34597600e+00 -1.16783512e+00 3.07575941e-01 2.08417439e+00
2.50591278e-01 8.37809145e-01 4.70929950e-01 4.13241386e-01
7.77981281e-01 1.55032337e-01 -5.31572104e-01 -9.60925519e-01
2.51313925e-01 7.15350926e-01 7.04766393e-01 4.11967456e-01
-1.52089107e+00 1.08520341e+00 7.25385666e+00 5.54114997e-01
-1.08150804e+00 -5.62570281e-02 2.13279635e-01 1.44392237e-01
3.59807551e-01 -1.69370577e-01 -6.91351712e-01 3.19945514e-01
1.77895617e+00 -1.92371562e-01 3.93581837e-01 1.06183863e+00
-7.33896121e-02 5.99570334e-01 -1.15010679e+00 9.25361812e-02
-3.07238325e-02 -1.00459707e+00 3.47138226e-01 2.68521369e-01
3.97973627e-01 -8.68167207e-02 3.42489362e-01 9.52559114e-01
7.38399208e-01 -1.01230037e+00 2.98080035e-02 1.87885299e-01
3.03454340e-01 -1.21747172e+00 8.98567557e-01 3.28452468e-01
-7.70231307e-01 -4.60830241e-01 -1.36915594e-01 -1.20265767e-01
-1.81639552e-01 7.81612378e-03 -1.14747894e+00 2.67966270e-01
2.93704003e-01 1.34301875e-02 -6.22756779e-01 7.26581931e-01
-1.18532524e-01 8.83658290e-01 -2.85156131e-01 -9.30566899e-03
7.06975698e-01 3.51995230e-01 8.04944277e-01 1.35065174e+00
-2.58924991e-01 -1.96799815e-01 5.79487741e-01 6.02189302e-01
-4.93012331e-02 -2.12925449e-01 -1.11174273e+00 -1.45270959e-01
5.67859352e-01 1.02487826e+00 -4.84883577e-01 -3.29029590e-01
-2.39159703e-01 7.32354581e-01 2.09275395e-01 2.60181129e-01
-1.01408887e+00 -5.91992438e-01 9.87632751e-01 1.32508233e-01
3.38414013e-01 1.72921032e-01 1.47178054e-01 -7.12749779e-01
-4.70578194e-01 -1.32053041e+00 5.05499661e-01 1.13801361e-04
-1.67349505e+00 6.48136497e-01 -3.76246087e-02 -1.12824965e+00
-4.64868844e-01 -8.80423188e-01 -1.16567850e+00 8.38918746e-01
-1.11177754e+00 -9.55480039e-01 2.23612487e-01 7.63049781e-01
1.76698893e-01 -8.53364885e-01 1.10959744e+00 3.72181386e-01
-8.30671489e-01 1.04490018e+00 6.14638887e-02 6.54851198e-01
7.13007689e-01 -1.18861282e+00 8.81544054e-01 9.10433531e-01
-9.92025957e-02 5.08076906e-01 9.45592165e-01 -4.40993339e-01
-1.04343069e+00 -1.25177431e+00 -5.09482399e-02 -6.89858258e-01
1.03878534e+00 -2.07571596e-01 -9.56206799e-01 1.00650072e+00
-8.58225599e-02 1.01514302e-01 6.51378036e-01 2.01117471e-02
-7.21213520e-01 -1.56064123e-01 -1.70006871e+00 7.70151138e-01
4.55182105e-01 -6.04994893e-01 -6.47801876e-01 3.88289273e-01
6.48669899e-01 1.17760420e-01 -6.25477016e-01 3.79749596e-01
2.13107839e-01 -1.14143741e+00 1.35503018e+00 -1.37966037e+00
-3.19656730e-02 -1.47186860e-01 -7.08158612e-02 -1.39388824e+00
-4.97766852e-01 -5.25435805e-01 -3.84251326e-01 1.10579789e+00
1.80597767e-01 -1.02754080e+00 7.89270937e-01 1.54233128e-01
8.15653205e-02 -3.86469603e-01 -9.10485327e-01 -1.07208896e+00
5.38615346e-01 -3.69714499e-01 3.80465388e-01 9.57648039e-01
-2.23196507e-01 3.41787636e-01 -2.89716750e-01 4.84504461e-01
6.67949617e-01 -5.77133656e-01 9.87707317e-01 -1.10957682e+00
-3.41980159e-01 -5.43091774e-01 -9.94003892e-01 -7.20969215e-02
4.06604737e-01 -5.40843010e-01 -2.01573700e-01 -5.99897146e-01
-3.99906307e-01 -4.61968720e-01 -8.15744340e-01 6.23902500e-01
-9.15517882e-02 9.69926417e-02 1.42255247e-01 -1.17165163e-01
-2.98419833e-01 1.33673042e-01 6.38570726e-01 -3.23874563e-01
6.22573197e-02 2.63416797e-01 -6.13377810e-01 7.38332391e-01
1.12024105e+00 -7.60373294e-01 -1.29018217e-01 2.41135985e-01
-5.74816227e-01 -1.17748156e-01 3.10251325e-01 -1.37910247e+00
1.62366256e-01 3.16664241e-02 5.44222116e-01 9.43474099e-02
2.04498157e-01 -8.66499186e-01 -3.49239945e-01 1.04873013e+00
-2.51008064e-01 6.51330352e-02 5.59519053e-01 5.29784441e-01
3.14460248e-02 -1.74031258e-01 1.16320503e+00 -2.29437947e-01
-7.47289121e-01 2.43140504e-01 -9.00593579e-01 9.59536433e-02
1.49738538e+00 1.60509095e-01 -4.20047671e-01 -2.86698997e-01
-6.17285073e-01 -1.65531747e-02 4.41259265e-01 4.62692142e-01
4.21881497e-01 -9.47401643e-01 -7.15248287e-01 4.88812327e-01
5.97873442e-02 -9.32165504e-01 -1.10252745e-01 2.03438342e-01
-4.87551570e-01 5.04875243e-01 -7.36863077e-01 -1.55837119e-01
-1.13467491e+00 1.38242781e+00 7.98216701e-01 -5.22332132e-01
-4.09513086e-01 6.23673916e-01 -1.23079993e-01 -6.12414479e-01
2.16458097e-01 5.56019485e-01 -5.30119762e-02 -3.75741810e-01
7.47538507e-01 2.67051309e-01 1.10818125e-01 -3.65977287e-01
-5.27305245e-01 5.61626256e-02 -5.95263720e-01 1.00892007e-01
7.40863502e-01 6.33006215e-01 2.90664643e-01 5.52424006e-02
1.00281894e+00 -1.76185966e-01 -7.72212088e-01 -2.40364000e-01
1.18350007e-01 -3.91614527e-01 -1.35956243e-01 -7.74414539e-01
-1.04321897e+00 7.63618767e-01 7.90440142e-01 7.38028288e-01
1.14232671e+00 -5.22621393e-01 6.13565266e-01 7.62338161e-01
4.27854002e-01 -5.69678366e-01 3.49462420e-01 6.82847917e-01
3.34174752e-01 -1.13945413e+00 -3.15088928e-01 -2.89057996e-02
-2.65839547e-01 1.13194549e+00 8.06474745e-01 -7.27129221e-01
6.82163417e-01 5.55061281e-01 1.29996583e-01 -1.72954857e-01
-6.48726404e-01 -2.25772470e-01 -8.25306028e-02 1.03877115e+00
-1.06713325e-01 6.84185326e-02 2.26300120e-01 -3.46979350e-02
1.63769111e-01 -5.66133976e-01 4.97672737e-01 1.29071641e+00
-7.36076832e-01 -1.47506237e+00 -8.58643174e-01 3.78126800e-01
-5.54560721e-01 1.98616043e-01 -9.17673171e-01 1.08723307e+00
-5.13100289e-02 1.17389333e+00 6.64410926e-03 -8.34264755e-01
6.86704576e-01 3.18133652e-01 2.29788318e-01 -5.27433872e-01
-1.11319745e+00 -5.38641453e-01 9.80156884e-02 -5.14580965e-01
1.31901994e-01 -3.26705962e-01 -8.38390350e-01 -7.44162798e-01
-7.61420503e-02 1.32493913e-01 4.89020079e-01 1.13624346e+00
2.03190655e-01 1.16260302e+00 1.13059556e+00 -9.41306472e-01
-1.20875263e+00 -9.21041548e-01 -4.60279644e-01 5.82159281e-01
4.15526003e-01 -8.31235826e-01 -6.86643779e-01 -4.81457055e-01]
|
[5.51564884185791, 7.566676616668701]
|
2dea7051-53c3-4f1d-9097-debdf20442fe
|
a-generative-model-for-user-simulation-in-a
| null | null |
https://aclanthology.org/E14-1066
|
https://aclanthology.org/E14-1066.pdf
|
A Generative Model for User Simulation in a Spatial Navigation Domain
| null |
['Mark Steedman', 'Aciel Eshky', 'Ben Allison', 'Subramanian Ramamoorthy']
|
2014-04-01
| null | null | null |
eacl-2014-4
|
['user-simulation']
|
['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]
|
[-7.413116931915283, 3.5917794704437256]
|
5b4dd31d-6ebb-4793-9b66-995275c84656
|
generalizable-lightweight-proxy-for-robust
|
2306.05031
| null |
https://arxiv.org/abs/2306.05031v1
|
https://arxiv.org/pdf/2306.05031v1.pdf
|
Generalizable Lightweight Proxy for Robust NAS against Diverse Perturbations
|
Recent neural architecture search (NAS) frameworks have been successful in finding optimal architectures for given conditions (e.g., performance or latency). However, they search for optimal architectures in terms of their performance on clean images only, while robustness against various types of perturbations or corruptions is crucial in practice. Although there exist several robust NAS frameworks that tackle this issue by integrating adversarial training into one-shot NAS, however, they are limited in that they only consider robustness against adversarial attacks and require significant computational resources to discover optimal architectures for a single task, which makes them impractical in real-world scenarios. To address these challenges, we propose a novel lightweight robust zero-cost proxy that considers the consistency across features, parameters, and gradients of both clean and perturbed images at the initialization state. Our approach facilitates an efficient and rapid search for neural architectures capable of learning generalizable features that exhibit robustness across diverse perturbations. The experimental results demonstrate that our proxy can rapidly and efficiently search for neural architectures that are consistently robust against various perturbations on multiple benchmark datasets and diverse search spaces, largely outperforming existing clean zero-shot NAS and robust NAS with reduced search cost.
|
['Sung Ju Hwang', 'Minseon Kim', 'Hyeonjeong Ha']
|
2023-06-08
| null | null | null | null |
['architecture-search']
|
['methodology']
|
[-1.67712048e-02 -7.44920790e-01 1.72184333e-01 -1.30903870e-01
-1.08595026e+00 -9.53036070e-01 4.96654421e-01 -2.73524016e-01
-4.58056301e-01 4.58860070e-01 -1.36632055e-01 -2.65833944e-01
-1.57002702e-01 -4.87641186e-01 -1.05138206e+00 -9.32915688e-01
9.66981500e-02 2.08954047e-02 1.89292923e-01 -3.72643083e-01
2.37583369e-01 4.78616923e-01 -1.56869006e+00 -2.55766362e-01
8.14481914e-01 1.00453961e+00 3.72467153e-02 6.59419775e-01
2.57442325e-01 4.13581163e-01 -7.56467581e-01 -4.78347003e-01
6.85249686e-01 -1.38474435e-01 -4.98242617e-01 -5.01269579e-01
8.71491373e-01 -3.77690077e-01 -6.45586789e-01 1.48835862e+00
8.72891486e-01 2.21854404e-01 2.49253720e-01 -1.20552611e+00
-9.42101479e-01 4.47669625e-01 -2.95268059e-01 4.97227877e-01
-2.57300381e-02 6.64780378e-01 9.85526741e-01 -8.55090618e-01
2.97896206e-01 1.02350318e+00 6.44331515e-01 7.87232220e-01
-1.22176993e+00 -9.42638218e-01 6.04694068e-01 2.98892051e-01
-1.33574283e+00 -1.03667426e+00 6.92911685e-01 -9.09967870e-02
1.06460130e+00 4.33250338e-01 9.28886235e-02 1.56548846e+00
1.17184490e-01 6.42431676e-01 7.31847048e-01 -8.05739239e-02
6.28168583e-01 -2.58841366e-01 1.00639373e-01 5.38274229e-01
4.39506322e-01 2.58697748e-01 -4.87214208e-01 -2.16574669e-01
4.50069129e-01 -4.37562801e-02 -4.72460955e-01 -4.32770878e-01
-1.08177114e+00 5.68109214e-01 6.81096315e-01 1.37608677e-01
-3.25414449e-01 2.85304576e-01 5.78394234e-01 7.11371481e-01
-4.18560505e-02 9.43454444e-01 -5.62677622e-01 -8.70442390e-02
-5.80392420e-01 2.72360802e-01 7.12413490e-01 8.80077481e-01
4.69874799e-01 6.86638474e-01 -1.08029224e-01 7.67803252e-01
2.11130921e-02 6.75376832e-01 6.78738177e-01 -8.33703816e-01
5.90586185e-01 3.75649214e-01 -1.78920045e-01 -1.19738591e+00
-1.39657721e-01 -6.88304067e-01 -9.79493380e-01 2.36865968e-01
1.27578840e-01 -2.39947796e-01 -1.03300858e+00 2.16728258e+00
2.59201199e-01 4.60965484e-01 2.57528812e-01 1.01406753e+00
8.01074266e-01 5.17467797e-01 -2.70355970e-01 3.41626108e-02
1.02060854e+00 -1.06011975e+00 -3.61047506e-01 -6.21541560e-01
2.33906969e-01 -4.16495889e-01 1.25701356e+00 2.16133058e-01
-1.08630633e+00 -3.64345163e-01 -1.27308440e+00 2.83353090e-01
-2.85661966e-01 -2.96485275e-01 2.46772930e-01 8.09879601e-01
-1.11338496e+00 4.39597726e-01 -9.15528774e-01 -2.20107213e-02
3.51429164e-01 4.92177784e-01 -3.29588860e-01 -5.12164272e-02
-1.11295784e+00 7.14862585e-01 1.30112961e-01 2.70052463e-01
-1.40362501e+00 -7.08399832e-01 -7.39402294e-01 3.02318037e-01
6.30090952e-01 -5.82296848e-01 1.09078884e+00 -8.89596045e-01
-1.48769557e+00 2.56113827e-01 4.57216315e-02 -5.75497746e-01
3.27165306e-01 -3.78689945e-01 -4.79537368e-01 -6.16126582e-02
-2.56422669e-01 2.81650275e-01 1.09216917e+00 -1.15284383e+00
-2.04836652e-01 -2.96504080e-01 2.16743246e-01 9.16665420e-02
-8.23137224e-01 1.83368236e-01 -7.44256616e-01 -8.58432889e-01
-1.37977675e-01 -1.18516922e+00 -4.21995789e-01 -1.61712185e-01
-5.42600453e-01 3.17294478e-01 6.10694170e-01 -4.50190276e-01
1.05551505e+00 -2.23140788e+00 3.28564048e-01 1.94988579e-01
1.37402773e-01 6.49204135e-01 -6.35489523e-01 1.54872658e-02
-1.39776953e-02 2.86943793e-01 -1.35205090e-01 -2.62331724e-01
1.61818415e-01 4.47414182e-02 -6.32587016e-01 4.37240928e-01
1.50962844e-01 8.92584264e-01 -7.42960811e-01 -1.21104188e-01
-1.59688085e-01 5.19018292e-01 -6.13640130e-01 3.47328216e-01
1.51112126e-02 3.13761890e-01 -4.19628739e-01 7.72174180e-01
4.20681089e-01 -1.87224314e-01 3.11611649e-02 -4.91879061e-02
1.86952323e-01 1.44425791e-03 -1.09566140e+00 1.38660693e+00
-5.53657651e-01 6.30258083e-01 2.88545996e-01 -8.93262625e-01
8.24032366e-01 1.88667685e-01 1.16857469e-01 -8.90617132e-01
1.37388408e-01 2.08603010e-01 4.45671156e-02 -1.52873546e-01
2.64247000e-01 4.87840116e-01 -5.91357388e-02 5.84486723e-01
-9.86254513e-02 1.93996385e-01 -3.72374132e-02 -8.66308361e-02
1.59560442e+00 -2.39439741e-01 2.47910656e-02 -5.07543534e-02
4.17587727e-01 -2.94481277e-01 7.70178556e-01 1.03710103e+00
-3.70581448e-01 7.51600146e-01 1.30800426e-01 -6.09707534e-01
-9.38602686e-01 -9.67697620e-01 1.90817028e-01 1.27244294e+00
2.76690125e-01 -2.22203463e-01 -8.33766937e-01 -5.76195359e-01
-3.31388205e-01 4.44378316e-01 -6.12723410e-01 -5.89296520e-01
-6.39615059e-01 -8.41950536e-01 7.10686445e-01 3.96843523e-01
4.69102591e-01 -1.03332698e+00 -8.20002079e-01 -7.00914348e-03
5.21190315e-02 -1.06718731e+00 -7.44275033e-01 1.35188773e-01
-6.54444754e-01 -1.02187359e+00 -5.14211178e-01 -8.50956738e-01
7.77137876e-01 4.50455576e-01 9.99117792e-01 2.40098447e-01
-4.53536212e-01 2.88595498e-01 -2.29736641e-01 -1.46089643e-01
-3.04325730e-01 1.96062043e-01 4.51256335e-01 1.71161760e-02
-1.86910838e-01 -6.68949664e-01 -7.50932992e-01 4.23262000e-01
-1.11618972e+00 -4.74549532e-01 7.07320571e-01 1.00108063e+00
6.43205822e-01 -3.39371674e-02 5.80036342e-01 -4.83721167e-01
9.82573688e-01 -6.49780393e-01 -6.88273787e-01 5.76227546e-01
-9.01034653e-01 2.82979786e-01 1.04778516e+00 -7.33562350e-01
-6.18759871e-01 -1.35885835e-01 -1.34558037e-01 -9.87550557e-01
-1.29727662e-01 3.75092506e-01 -1.23869605e-01 -4.54427451e-01
9.71513212e-01 5.86925983e-01 -8.23028609e-02 -5.03068686e-01
1.24387175e-01 1.98719248e-01 9.09818530e-01 -6.07793987e-01
1.29281890e+00 1.32177828e-03 -1.24648266e-01 -4.98319179e-01
-6.63034320e-01 -1.45967662e-01 -2.86416739e-01 1.94028169e-01
3.31590712e-01 -9.07517076e-01 -4.23022062e-01 6.39009655e-01
-8.16558599e-01 -1.22184008e-01 4.06245105e-02 6.71036318e-02
-3.65770072e-01 4.15615857e-01 -4.41893220e-01 -5.18023491e-01
-9.25453126e-01 -1.60086179e+00 5.25394320e-01 3.21920425e-01
9.08512250e-02 -7.52278566e-01 1.44606397e-01 7.82455951e-02
8.94688010e-01 2.69153804e-01 8.65233123e-01 -8.31507981e-01
-7.31177568e-01 -4.37345535e-01 2.77259387e-02 3.97558689e-01
4.02200557e-02 1.27830327e-01 -8.94870043e-01 -8.08213830e-01
2.24902201e-03 -5.73707819e-01 6.93021417e-01 9.03802663e-02
1.33757663e+00 -8.44935119e-01 -3.27744149e-02 1.15861535e+00
1.50914359e+00 1.61063984e-01 3.70176941e-01 6.40956998e-01
5.58089256e-01 5.17292693e-02 3.37102771e-01 1.73454776e-01
3.34190205e-02 7.33245194e-01 8.38355184e-01 1.73782021e-01
2.72882115e-02 1.05089068e-01 6.26812935e-01 8.33035588e-01
1.32799506e-01 -1.54064938e-01 -9.24596071e-01 7.23092914e-01
-1.73379493e+00 -9.02462423e-01 6.69244051e-01 2.22957540e+00
8.86963010e-01 2.17446655e-01 -1.01167582e-01 -1.31525919e-01
6.12142563e-01 3.56503338e-01 -1.32137549e+00 -4.10390079e-01
-1.31374121e-01 1.35488614e-01 3.94853383e-01 -6.23969873e-03
-1.02031589e+00 7.20248640e-01 6.81395912e+00 4.87057805e-01
-1.32741165e+00 -5.15373871e-02 5.33377767e-01 -6.00140095e-01
-2.45561630e-01 -2.82950729e-01 -5.97514987e-01 5.50806642e-01
1.02894533e+00 -3.77062172e-01 8.11881542e-01 1.16645372e+00
-4.11357582e-01 6.85649574e-01 -8.86296093e-01 1.01485884e+00
1.41923100e-01 -1.39287543e+00 -1.69918500e-02 -2.64267892e-01
1.01796854e+00 4.65934187e-01 6.54581368e-01 3.92667592e-01
5.05887866e-01 -1.11798239e+00 7.35496402e-01 3.32560539e-01
6.50075376e-01 -7.28552282e-01 5.96396506e-01 2.84301192e-01
-9.56898749e-01 -4.31927562e-01 -3.74507457e-01 4.03223127e-01
-3.91151339e-01 -5.64255007e-03 -5.45751035e-01 1.93424955e-01
9.61299717e-01 6.88735843e-02 -8.55884016e-01 1.21422398e+00
-5.05436212e-02 5.31798899e-01 -2.67758518e-01 -1.14875644e-01
3.75084877e-01 1.81854516e-01 6.99371338e-01 8.13043535e-01
5.01735926e-01 -3.06523532e-01 6.63504526e-02 6.75502241e-01
-2.74276167e-01 -5.25893830e-02 -5.81248462e-01 4.96524833e-02
9.63139892e-01 9.96336997e-01 -3.44141364e-01 8.49626213e-02
-3.42545182e-01 7.42234528e-01 5.27692020e-01 5.03464460e-01
-9.45415556e-01 -5.47348976e-01 1.23316932e+00 -4.79048908e-01
4.29133117e-01 -1.80868745e-01 -2.42435127e-01 -1.26009095e+00
3.03286761e-01 -1.60372233e+00 3.60229552e-01 -2.74350584e-01
-1.31416917e+00 1.03620970e+00 -3.07099104e-01 -1.09831786e+00
-2.97042608e-01 -2.93481827e-01 -8.75250936e-01 5.67228019e-01
-1.46826184e+00 -8.98175418e-01 -2.09216952e-01 7.90629983e-01
6.43278539e-01 -6.53032959e-01 9.87529814e-01 6.03522398e-02
-1.32028759e+00 1.29637694e+00 4.67974693e-01 2.76091322e-02
8.10987771e-01 -1.01948988e+00 1.07353675e+00 1.36584926e+00
3.31919849e-01 1.11621356e+00 8.74597251e-01 -2.58425027e-01
-2.03618765e+00 -1.11051631e+00 3.98141593e-02 -3.04521084e-01
7.59214878e-01 -1.87308118e-01 -1.12769783e+00 4.73546505e-01
-6.58996915e-03 3.42432171e-01 4.75980908e-01 1.17500700e-01
-8.85673761e-01 -3.42359513e-01 -9.14834440e-01 1.09642100e+00
1.11995125e+00 -5.09316385e-01 -2.43037850e-01 2.60515690e-01
9.57181394e-01 -5.52658796e-01 -6.98692381e-01 3.79349530e-01
4.04593736e-01 -7.72670984e-01 1.29992461e+00 -8.49748969e-01
1.01073407e-01 -2.66987115e-01 -2.17185915e-01 -1.46278965e+00
-3.43740731e-01 -1.02284622e+00 -4.65747118e-01 9.39917445e-01
5.26408494e-01 -8.34204018e-01 5.81160605e-01 8.59484732e-01
-1.90144032e-01 -8.90044868e-01 -1.03396988e+00 -1.04341352e+00
6.29787520e-02 -2.25760639e-01 1.03202665e+00 7.63889551e-01
-6.36756063e-01 -1.21021792e-01 -5.40724277e-01 4.97596174e-01
6.65698946e-01 2.22074285e-01 9.14855063e-01 -8.33479106e-01
-2.45091960e-01 -6.66707993e-01 -3.49017620e-01 -5.87939739e-01
3.40850800e-01 -5.79814494e-01 3.54967535e-01 -9.57734644e-01
8.19501281e-03 -4.99288410e-01 -7.22749591e-01 6.83283746e-01
-4.96237844e-01 1.67112768e-01 1.74794868e-01 4.36735094e-01
-7.14494586e-01 5.97286582e-01 6.40658557e-01 -2.81516641e-01
-5.53260967e-02 -3.14008072e-02 -1.05766869e+00 5.08228064e-01
8.60204458e-01 -4.27507430e-01 -4.96719718e-01 -9.86801386e-01
2.73776412e-01 -4.23735648e-01 3.53819937e-01 -1.33407676e+00
6.06219053e-01 -3.16899955e-01 1.19100660e-01 2.14750953e-02
2.05941752e-01 -7.63321459e-01 1.22178614e-01 4.29551750e-01
-4.61520016e-01 4.86548960e-01 3.50164473e-01 8.45117807e-01
-1.64405629e-01 -2.30773926e-01 8.49458575e-01 7.69716054e-02
-9.09512520e-01 6.78993821e-01 2.39889219e-01 2.17012346e-01
1.17040253e+00 -6.40021339e-02 -7.19874084e-01 -3.03499937e-01
-3.68946791e-01 3.48013610e-01 6.76094413e-01 6.96520448e-01
7.64635265e-01 -1.25365806e+00 -6.91215694e-01 3.63942415e-01
6.11389913e-02 5.26984558e-02 3.84973198e-01 4.54318434e-01
-4.48401064e-01 3.44989032e-01 -3.44372004e-01 -3.47790509e-01
-1.30251360e+00 8.26062560e-01 4.48937744e-01 -1.17201090e-01
-5.04942775e-01 1.11137044e+00 8.96360874e-02 -4.21419144e-01
6.96367502e-01 1.52731806e-01 1.23719819e-01 -4.45925474e-01
7.48344004e-01 1.24130152e-01 2.52476275e-01 -4.86933172e-01
-3.88847113e-01 4.92975950e-01 -2.93999642e-01 2.57582366e-01
1.48583567e+00 2.21609436e-02 1.49295732e-01 3.10883578e-02
1.14921713e+00 -1.39827922e-01 -1.65308356e+00 -5.41248024e-01
-2.17116624e-01 -5.29428840e-01 1.25051409e-01 -6.27717137e-01
-1.50198662e+00 6.71569288e-01 6.88650250e-01 1.23878621e-01
1.37567496e+00 -3.59845489e-01 9.79330420e-01 9.92117584e-01
3.05365831e-01 -9.26921487e-01 3.24283153e-01 6.09757185e-01
1.02458096e+00 -1.11418736e+00 -1.57756075e-01 2.38128066e-01
-6.31510615e-01 9.12339211e-01 9.27580714e-01 -1.65236384e-01
3.86562824e-01 3.14603209e-01 1.54434338e-01 -7.31590576e-03
-1.16071916e+00 1.90480292e-01 3.21100324e-01 7.18725860e-01
-2.44104445e-01 -3.08385193e-01 4.25257415e-01 4.19753402e-01
-2.13445053e-01 -7.29202032e-01 2.10712194e-01 9.84406352e-01
-2.81017840e-01 -7.73467898e-01 -3.99295121e-01 2.16779083e-01
-6.68176770e-01 -1.16914347e-01 -3.22781831e-01 3.28495830e-01
-4.38127190e-01 8.67656410e-01 -2.57974178e-01 -6.55077338e-01
4.45897400e-01 2.69849189e-02 9.92759690e-02 -3.08866173e-01
-7.30892479e-01 -4.80996966e-01 -2.32733250e-01 -7.90615201e-01
2.28188217e-01 -6.25137150e-01 -6.93818390e-01 -3.02871108e-01
-2.55561054e-01 -1.02426752e-01 8.33413541e-01 7.83424020e-01
7.91146457e-01 4.32276309e-01 9.52300131e-01 -6.65227175e-01
-1.36325824e+00 -5.38192213e-01 3.17659527e-02 4.29021031e-01
6.66771650e-01 -4.87989247e-01 -4.80538696e-01 -1.81950837e-01]
|
[5.560215950012207, 7.951155185699463]
|
0b41c858-f852-4f60-87a5-2f7b8272f050
|
habicrowd-a-high-performance-simulator-for
|
2306.11377
| null |
https://arxiv.org/abs/2306.11377v1
|
https://arxiv.org/pdf/2306.11377v1.pdf
|
HabiCrowd: A High Performance Simulator for Crowd-Aware Visual Navigation
|
Visual navigation, a foundational aspect of Embodied AI (E-AI), has been significantly studied in the past few years. While many 3D simulators have been introduced to support visual navigation tasks, scarcely works have been directed towards combining human dynamics, creating the gap between simulation and real-world applications. Furthermore, current 3D simulators incorporating human dynamics have several limitations, particularly in terms of computational efficiency, which is a promise of E-AI simulators. To overcome these shortcomings, we introduce HabiCrowd, the first standard benchmark for crowd-aware visual navigation that integrates a crowd dynamics model with diverse human settings into photorealistic environments. Empirical evaluations demonstrate that our proposed human dynamics model achieves state-of-the-art performance in collision avoidance, while exhibiting superior computational efficiency compared to its counterparts. We leverage HabiCrowd to conduct several comprehensive studies on crowd-aware visual navigation tasks and human-robot interactions. The source code and data can be found at https://habicrowd.github.io/.
|
['Anh Nguyen', 'Thieu Vo', 'Huynh Thi Thanh Binh', 'Dzung Nguyen', 'Baoru Huang', 'Minh Nhat Vu', 'Toan Tien Nguyen', 'An Dinh Vuong']
|
2023-06-20
| null | null | null | null |
['human-dynamics', 'visual-navigation']
|
['computer-vision', 'robots']
|
[-5.55465281e-01 -2.13353172e-01 2.60890692e-01 1.23332553e-01
8.42460468e-02 -3.69142324e-01 7.36664712e-01 -1.51279539e-01
-7.25889444e-01 6.98392749e-01 1.70155659e-01 -3.80320340e-01
2.16110468e-01 -5.55786133e-01 -4.22095060e-01 -5.66040814e-01
-4.35132354e-01 4.10279125e-01 6.33650720e-01 -9.48531985e-01
5.23983762e-02 2.27849901e-01 -1.90721178e+00 -3.97748828e-01
1.06276464e+00 6.02132499e-01 4.22894508e-01 8.35450888e-01
1.06584810e-01 7.78196931e-01 -5.43872595e-01 -1.67647436e-01
2.81837791e-01 -4.83734339e-01 -3.84553611e-01 -4.36668396e-01
-1.21615313e-01 -2.20889807e-01 -8.24803352e-01 8.83169711e-01
1.03500319e+00 3.23515892e-01 6.26931787e-01 -1.98652887e+00
-7.46691465e-01 7.40366578e-02 -5.30711830e-01 3.31484467e-01
9.02232289e-01 7.30578661e-01 4.70322877e-01 -8.85440350e-01
7.22617447e-01 1.49209797e+00 4.94575024e-01 8.45880568e-01
-6.92979336e-01 -6.81362212e-01 3.54058534e-01 2.43815109e-01
-1.36821437e+00 -3.32534850e-01 4.17552620e-01 -5.53612590e-01
1.05453968e+00 -3.55237387e-02 1.24019468e+00 1.33003652e+00
2.94513255e-01 9.87282515e-01 9.09855485e-01 -2.91654110e-01
5.15225410e-01 -2.21815750e-01 -7.27992281e-02 8.90855491e-01
4.59257215e-01 3.76092702e-01 -8.07487369e-01 4.89723496e-02
7.59220958e-01 -3.11109960e-01 -3.73910367e-01 -1.05339313e+00
-1.29963529e+00 5.65972984e-01 7.32324779e-01 5.90071417e-02
-3.49942118e-01 3.65114778e-01 4.69760716e-01 6.10185154e-02
7.99802691e-02 1.00002371e-01 3.18427235e-01 -4.92649764e-01
-5.53741381e-02 8.63738894e-01 8.14775169e-01 1.23246968e+00
3.80792856e-01 1.33550256e-01 -7.87561685e-02 4.33979213e-01
3.10032606e-01 9.47826326e-01 2.93057889e-01 -1.13945961e+00
2.03903466e-01 6.07647955e-01 4.55567718e-01 -1.26940846e+00
-6.57983124e-01 1.44443050e-01 -7.68001854e-01 7.36496091e-01
3.99693072e-01 -3.19928885e-01 -7.84921646e-01 1.76694775e+00
5.72672248e-01 3.70261483e-02 2.18656629e-01 1.44667637e+00
1.15033090e+00 4.87086624e-01 1.78519651e-01 2.16099858e-01
1.16936541e+00 -1.30913854e+00 -7.33757794e-01 -4.49839026e-01
5.61979830e-01 -2.33306408e-01 1.21006429e+00 -2.00094772e-03
-1.01923192e+00 -2.50579894e-01 -1.02066314e+00 -7.73736686e-02
-5.05907893e-01 -4.18632179e-01 7.37371266e-01 5.05498707e-01
-1.20718622e+00 -1.80894524e-01 -9.60804522e-01 -8.46554399e-01
2.24217653e-01 1.70974836e-01 -4.33518440e-01 6.26761541e-02
-1.21313655e+00 1.11887181e+00 -5.71263842e-02 9.14350972e-02
-1.21911705e+00 -2.47584790e-01 -9.68054712e-01 -3.55127633e-01
6.07976258e-01 -1.02561295e+00 1.48503864e+00 -2.85099536e-01
-1.57193804e+00 7.61558175e-01 -2.57815540e-01 -2.94839323e-01
1.00569320e+00 -2.61458755e-01 -2.07404539e-01 -1.66373655e-01
2.80374110e-01 7.15591431e-01 1.40500948e-01 -1.60520685e+00
-6.83044851e-01 -2.95425206e-01 3.00535083e-01 6.84156239e-01
-5.63550070e-02 -3.45278472e-01 -8.82877171e-01 -3.22446972e-01
-3.30755264e-01 -1.24897754e+00 -4.71834540e-01 3.54023665e-01
-3.34392160e-01 -1.06390774e-01 6.80886745e-01 1.26120159e-02
9.80859935e-01 -1.75012028e+00 5.60766101e-01 -1.79050595e-01
4.62346315e-01 3.25483710e-01 -5.01140915e-02 7.00918078e-01
7.55033672e-01 -1.95525616e-01 -1.88399538e-01 -5.65345705e-01
4.36623663e-01 1.67426690e-02 -2.26944522e-03 5.32345355e-01
-5.06355226e-01 1.14243317e+00 -1.27786231e+00 -3.78061891e-01
2.57536799e-01 6.66535795e-01 -5.25193512e-01 1.32603168e-01
-9.18660909e-02 5.99582374e-01 -4.83141869e-01 6.31449819e-01
6.43492758e-01 -8.27654153e-02 -6.50984887e-03 4.98081505e-01
-3.69948536e-01 -3.06652129e-01 -9.70417261e-01 1.92458630e+00
-2.48070568e-01 7.38591194e-01 2.19750419e-01 -4.09412891e-01
5.84020317e-01 -3.40798795e-02 2.97039211e-01 -1.11633563e+00
3.59024376e-01 1.93226412e-01 -1.80808790e-02 -5.73228657e-01
8.39693904e-01 3.15551728e-01 -2.14373022e-01 2.87786990e-01
-3.48697990e-01 -1.81345761e-01 2.22015738e-01 3.05698663e-01
1.17104971e+00 2.99880385e-01 4.57186013e-01 -3.27392787e-01
3.34484637e-01 5.19828081e-01 4.89034742e-01 9.21729267e-01
-8.90864849e-01 3.75716448e-01 1.07112102e-01 -4.21201408e-01
-8.60407531e-01 -1.17070830e+00 4.09882039e-01 1.08448124e+00
1.22972178e+00 -3.97473872e-01 -8.06953192e-01 -2.80353218e-01
2.39035100e-01 5.24704039e-01 -8.14376712e-01 -3.89496833e-02
-6.28277600e-01 -5.52673876e-01 7.13455319e-01 3.53661954e-01
8.04245412e-01 -1.38337624e+00 -1.49709380e+00 1.32204248e-02
-2.62217373e-01 -1.12208891e+00 -2.60088593e-01 -2.10127801e-01
-1.57825887e-01 -1.20541155e+00 -1.02872562e+00 -8.55960548e-01
3.53121668e-01 9.56340075e-01 1.11502552e+00 3.56208593e-01
-3.44418913e-01 7.83088624e-01 -4.54089493e-01 -6.02265954e-01
-2.45073378e-01 -3.00654415e-02 3.88353616e-01 -6.55032039e-01
4.31111515e-01 -3.02229792e-01 -9.73964870e-01 6.33197188e-01
-3.70663315e-01 2.22824976e-01 2.25569621e-01 5.83764911e-01
2.10585475e-01 -4.86078501e-01 2.24113867e-01 -1.86390340e-01
8.98850262e-01 -4.98831987e-01 -5.62924922e-01 -3.04483026e-02
-1.92087919e-01 -3.68623465e-01 4.29212004e-01 -3.88205975e-01
-9.22637045e-01 -2.71678060e-01 1.44518599e-01 -9.71502289e-02
-2.46609986e-01 3.03509593e-01 -3.45491022e-02 -2.79656410e-01
7.82473803e-01 2.32609451e-01 7.38932937e-02 7.58706629e-02
2.89185107e-01 4.97155249e-01 6.41100109e-01 -5.05174696e-01
7.11769640e-01 7.07519412e-01 -8.34437385e-02 -9.94751096e-01
-3.47253270e-02 -2.32989833e-01 -4.60115284e-01 -5.80160081e-01
1.03267908e+00 -1.03642273e+00 -1.32455552e+00 8.58113825e-01
-9.32177246e-01 -8.68180692e-01 7.80885145e-02 3.25349003e-01
-7.60293782e-01 3.42577040e-01 -4.37951624e-01 -1.09012187e+00
-1.06112011e-01 -1.45152164e+00 1.00466001e+00 6.24337792e-01
-1.94104075e-01 -9.03959155e-01 3.50962758e-01 -2.17165872e-02
5.06649911e-01 3.13241422e-01 3.91370863e-01 -4.68531922e-02
-6.53341174e-01 1.20516792e-01 -1.27885222e-01 -7.16888726e-01
-2.18165994e-01 -3.30976933e-01 -6.10989511e-01 -4.94786859e-01
-5.10356724e-01 -4.20644015e-01 5.39456546e-01 3.27688903e-01
2.57285476e-01 4.03011531e-01 -8.08075190e-01 5.18389404e-01
1.03206193e+00 5.44224977e-01 5.62023282e-01 9.53456044e-01
5.99389255e-01 4.38210398e-01 7.84928739e-01 6.53800249e-01
1.12465835e+00 9.61538970e-01 6.79400802e-01 -5.00283018e-02
-1.23462960e-01 -3.81882936e-01 2.59453297e-01 5.90645790e-01
-3.96319300e-01 -7.56420195e-01 -1.45826340e+00 5.85680068e-01
-2.17851543e+00 -9.01460648e-01 -1.03244931e-02 1.78718233e+00
8.21804777e-02 1.89843196e-02 5.05611360e-01 -1.71134472e-01
6.13277018e-01 1.46207780e-01 -8.37645769e-01 3.83146927e-02
-2.78136224e-01 -6.11545980e-01 2.20927969e-01 5.20797491e-01
-8.91232729e-01 1.29689741e+00 6.12653351e+00 3.35797668e-01
-8.19406688e-01 5.10933846e-02 -8.92952606e-02 -9.62512940e-02
-9.89259183e-02 -1.52686596e-01 -6.25751317e-01 4.09642726e-01
3.93571943e-01 -4.81307864e-01 5.19426644e-01 8.49758327e-01
3.30011129e-01 -5.74423969e-01 -7.93831885e-01 1.10775483e+00
1.09399959e-01 -9.91489112e-01 -1.69775654e-02 2.11223423e-01
4.94354099e-01 1.34486616e-01 2.34641731e-01 4.67674702e-01
7.63706267e-01 -1.09585500e+00 1.06900311e+00 3.77190322e-01
3.55811775e-01 -7.67358661e-01 5.61048985e-01 5.37264407e-01
-1.39198518e+00 -1.53302401e-01 -4.09026653e-01 -4.21885073e-01
6.23525858e-01 -1.89777464e-01 -4.31221515e-01 4.05475408e-01
1.16982841e+00 5.53076923e-01 -3.87701929e-01 1.35202110e+00
-2.91003175e-02 -2.25408420e-01 -2.07583115e-01 -6.20678365e-01
4.28242743e-01 -2.68146485e-01 9.21895564e-01 1.02060187e+00
4.14117515e-01 2.45082200e-01 3.51874173e-01 6.03702664e-01
3.12025964e-01 -5.19322902e-02 -1.12434793e+00 2.62430787e-01
6.82994723e-01 7.22778022e-01 -7.50834882e-01 -1.22038096e-01
-2.96766222e-01 1.21905208e+00 4.96184021e-01 5.31274736e-01
-1.10881126e+00 -4.61876631e-01 1.18230295e+00 -8.01967233e-02
3.12451776e-02 -7.32555091e-01 -4.66627814e-02 -1.09378600e+00
4.61686263e-03 -8.42066288e-01 -8.64685178e-02 -1.01759613e+00
-8.79425108e-01 8.99311602e-01 4.44021672e-02 -1.32584238e+00
-1.09343201e-01 -4.01268154e-01 -4.14792746e-01 5.68540215e-01
-1.25990450e+00 -1.14936960e+00 -1.14595366e+00 5.38263619e-01
3.97941530e-01 -3.12457770e-01 7.17512429e-01 2.60304864e-02
-5.71484864e-01 5.47271490e-01 -8.10037851e-02 -7.97479898e-02
5.26188493e-01 -1.00134504e+00 1.00958061e+00 8.57563257e-01
-4.09607053e-01 4.67614263e-01 1.00314057e+00 -7.90409386e-01
-1.71874797e+00 -5.57413876e-01 3.96363020e-01 -6.45229757e-01
4.75672185e-01 -6.89361572e-01 -5.86578012e-01 5.47294736e-01
3.87771308e-01 -3.26961540e-02 2.75346041e-01 -4.10692275e-01
3.03366836e-02 5.30657649e-01 -9.19566393e-01 1.50855660e+00
2.04750037e+00 -1.97311252e-01 -2.47322515e-01 -6.07195310e-02
7.46024787e-01 -8.03213239e-01 -3.13386112e-01 2.16760322e-01
7.77856708e-01 -1.39388597e+00 1.07835722e+00 -2.13770062e-01
3.46331671e-02 -6.79908872e-01 -2.05257490e-01 -1.61717224e+00
-3.22218359e-01 -6.51793003e-01 -1.62741482e-01 6.59600675e-01
9.29026902e-02 -7.19941258e-01 7.48414755e-01 4.53325689e-01
-1.54887602e-01 -5.97208023e-01 -8.91751945e-01 -9.64843333e-01
3.39690782e-02 -1.74801394e-01 5.90959847e-01 4.98113722e-01
2.25746796e-01 5.26941679e-02 -5.49371302e-01 1.90837026e-01
6.07830942e-01 -2.54743874e-01 1.43239188e+00 -9.49549675e-01
1.61823586e-01 -6.25849485e-01 -6.67465150e-01 -1.34615242e+00
3.13784093e-01 -5.97700655e-01 3.98544014e-01 -1.92164695e+00
1.55181125e-01 -4.63759720e-01 3.09238285e-01 1.77604362e-01
-2.78799295e-01 1.85529798e-01 5.70466697e-01 2.64361680e-01
-9.77729261e-01 1.19044828e+00 1.54227877e+00 -7.20140804e-03
-4.75239456e-01 -4.59528387e-01 -5.66827178e-01 6.90418780e-01
6.84914887e-01 -2.14265734e-02 -5.67952752e-01 -6.71523511e-01
8.45247693e-03 1.65341944e-01 6.31833196e-01 -1.41911387e+00
8.15240204e-01 -2.24571303e-01 3.46148945e-02 -4.57425267e-01
6.03188276e-01 -6.54611468e-01 1.13035150e-01 7.76771784e-01
6.48073182e-02 6.97412372e-01 3.30989331e-01 8.36867571e-01
-3.75194736e-02 3.58124822e-01 4.54713702e-01 -2.05188245e-01
-1.32319713e+00 3.38541925e-01 -8.93630862e-01 4.67682838e-01
1.36578655e+00 -5.07203937e-01 -7.18859017e-01 -6.59863710e-01
-2.98437834e-01 6.69170141e-01 1.13967299e+00 6.68714166e-01
7.14953959e-01 -1.24630332e+00 -3.77707660e-01 -1.09510042e-01
4.06107068e-01 4.46522562e-03 4.08507943e-01 6.87530398e-01
-9.20169771e-01 4.74812239e-01 -5.39106488e-01 -6.67289972e-01
-1.09796607e+00 5.73407292e-01 4.17128682e-01 2.08370507e-01
-9.40039158e-01 7.05772042e-01 4.67477590e-01 -6.61509573e-01
4.82375413e-01 1.59670804e-02 -1.23177223e-01 -4.46438015e-01
5.59630930e-01 5.93577325e-01 -5.04256964e-01 -1.01872671e+00
-7.56928205e-01 7.14669049e-01 4.54820991e-01 -3.45476210e-01
1.06261146e+00 -5.64747453e-01 4.31001127e-01 3.57666194e-01
6.66546285e-01 -1.50219291e-01 -1.47223413e+00 2.28278413e-02
-2.99753487e-01 -3.85998756e-01 -2.87123352e-01 -5.74955165e-01
-7.30360866e-01 7.66597211e-01 7.25232303e-01 -8.90342370e-02
6.99796557e-01 -1.63549691e-01 7.24505484e-01 6.72777474e-01
1.16816390e+00 -7.40736663e-01 2.92271942e-01 9.16073442e-01
9.72609162e-01 -1.38040447e+00 -3.89190316e-01 -3.56652021e-01
-1.11849999e+00 5.10052502e-01 1.23367500e+00 -1.48463741e-01
5.13934314e-01 4.61082727e-01 4.33028072e-01 -3.19562018e-01
-5.65651238e-01 -5.67721307e-01 -2.30841905e-01 1.15473044e+00
1.09378852e-01 -5.92050329e-02 1.93680972e-02 4.90814656e-01
-3.59855652e-01 4.65886258e-02 6.25309408e-01 1.31656528e+00
-4.20031130e-01 -5.68892062e-01 -2.33049870e-01 -4.06267792e-01
2.05627143e-01 2.35741466e-01 -4.76524681e-01 1.20792603e+00
-2.20073625e-01 1.06131470e+00 -3.38742509e-03 -4.83408898e-01
7.19874918e-01 -4.32655126e-01 4.13013428e-01 -1.35681868e-01
-5.36478698e-01 -4.33286548e-01 1.76803377e-02 -8.26506615e-01
-3.17164093e-01 -4.72939014e-01 -1.60037804e+00 -7.96839595e-01
1.41492575e-01 6.75572604e-02 4.44576144e-01 5.91391802e-01
6.88989699e-01 4.41232294e-01 -9.82732698e-02 -1.57157743e+00
9.82055962e-02 -4.97874022e-01 -3.42013627e-01 4.03839648e-01
3.70139092e-01 -1.46851373e+00 -2.98135906e-01 -4.53798264e-01]
|
[4.63317346572876, 0.68758624792099]
|
3ded8005-8ea5-4e95-bc0b-0bf1820cf6b9
|
evaluating-prompt-based-question-answering
|
2305.12900
| null |
https://arxiv.org/abs/2305.12900v2
|
https://arxiv.org/pdf/2305.12900v2.pdf
|
Evaluating Prompt-based Question Answering for Object Prediction in the Open Research Knowledge Graph
|
There have been many recent investigations into prompt-based training of transformer language models for new text genres in low-resource settings. The prompt-based training approach has been found to be effective in generalizing pre-trained or fine-tuned models for transfer to resource-scarce settings. This work, for the first time, reports results on adopting prompt-based training of transformers for \textit{scholarly knowledge graph object prediction}. The work is unique in the following two main aspects. 1) It deviates from the other works proposing entity and relation extraction pipelines for predicting objects of a scholarly knowledge graph. 2) While other works have tested the method on text genera relatively close to the general knowledge domain, we test the method for a significantly different domain, i.e. scholarly knowledge, in turn testing the linguistic, probabilistic, and factual generalizability of these large-scale transformer models. We find that (i) per expectations, transformer models when tested out-of-the-box underperform on a new domain of data, (ii) prompt-based training of the models achieve performance boosts of up to 40\% in a relaxed evaluation setting, and (iii) testing the models on a starkly different domain even with a clever training objective in a low resource setting makes evident the domain knowledge capture gap offering an empirically-verified incentive for investing more attention and resources to the scholarly domain in the context of transformer models.
|
['Sören Auer', 'Moussab Hrou', "Jennifer D'Souza"]
|
2023-05-22
| null | null | null | null |
['general-knowledge', 'relation-extraction']
|
['miscellaneous', 'natural-language-processing']
|
[ 1.45315722e-01 5.56524038e-01 -6.20342731e-01 -1.63350180e-01
-8.46594691e-01 -7.47404993e-01 9.54331279e-01 1.38544515e-01
-3.02715123e-01 6.83163941e-01 4.01692450e-01 -6.51177526e-01
-5.96104920e-01 -8.53460014e-01 -7.16295600e-01 5.93023258e-04
1.09660529e-01 9.29427564e-01 3.88187736e-01 -2.38240927e-01
2.85034031e-01 4.17456418e-01 -1.18328822e+00 1.66899204e-01
8.11299443e-01 7.96622217e-01 1.22390591e-01 3.83131087e-01
-4.70958017e-02 1.07058883e+00 -3.50388646e-01 -9.09261703e-01
1.23818792e-01 1.99003760e-02 -1.32498550e+00 -1.18946351e-01
7.68656075e-01 -2.31825057e-02 -4.85801011e-01 4.99958217e-01
2.29247168e-01 1.68181613e-01 7.80406713e-01 -1.02736115e+00
-1.10667872e+00 1.11535680e+00 -1.62166208e-01 6.30013704e-01
3.42742354e-01 9.32780057e-02 1.51253402e+00 -7.36905694e-01
1.01756155e+00 1.07215476e+00 9.94595170e-01 1.04089156e-01
-1.11316562e+00 -7.03541696e-01 2.78848648e-01 2.34929562e-01
-1.21914780e+00 -6.21757507e-01 4.46270883e-01 -4.58011091e-01
1.59535015e+00 1.95093289e-01 3.64317238e-01 1.35953081e+00
-2.26408303e-01 7.11728156e-01 1.07011497e+00 -5.49443483e-01
-2.65707880e-01 3.63457859e-01 2.37661615e-01 7.08433270e-01
4.94241714e-01 -8.18876550e-02 -7.22723007e-01 -6.57620430e-02
4.55241352e-01 -5.04924536e-01 -1.44389659e-01 1.38294563e-01
-1.18971109e+00 6.88326478e-01 1.05174385e-01 6.71233535e-01
-3.09244931e-01 -1.00001797e-01 4.74179059e-01 2.85488009e-01
6.75279677e-01 8.99375856e-01 -9.49265063e-01 -5.06185174e-01
-9.89668012e-01 2.75642395e-01 1.37202168e+00 1.23693788e+00
4.14372653e-01 -7.93708116e-02 -2.29156137e-01 8.24382365e-01
4.36442979e-02 2.69996554e-01 4.63373721e-01 -7.56381810e-01
8.19399118e-01 5.54625750e-01 -3.33725750e-01 -9.16283727e-01
-3.44140381e-01 -7.03148544e-01 -2.60613590e-01 -5.08131564e-01
7.83665776e-01 -8.56766328e-02 -6.36518180e-01 1.58597541e+00
-9.84839723e-03 1.52276218e-01 6.38197809e-02 3.08327705e-01
1.14420557e+00 4.16786939e-01 4.32790369e-01 7.29927886e-03
1.38288522e+00 -6.33962512e-01 -4.51667219e-01 -5.29630601e-01
8.72841239e-01 -7.86048532e-01 1.39221287e+00 3.30561489e-01
-8.67646635e-01 -3.26770633e-01 -7.77777314e-01 -3.45908374e-01
-7.28192747e-01 6.19717501e-02 9.49323356e-01 6.19162917e-01
-1.04352319e+00 5.90614319e-01 -6.20203555e-01 -8.74636710e-01
4.91060525e-01 1.17721938e-01 -3.29983473e-01 -1.86272323e-01
-1.31578183e+00 1.25973701e+00 4.32433039e-01 -4.52042252e-01
-4.52114701e-01 -1.10252500e+00 -8.63345265e-01 3.05982977e-01
7.70487368e-01 -6.16030753e-01 1.22050059e+00 -6.77603900e-01
-1.23646617e+00 1.11479115e+00 -1.97846871e-02 -3.59215051e-01
2.98274100e-01 -6.98015839e-02 -4.24002826e-01 -1.47950232e-01
4.18363243e-01 7.26332292e-02 4.24044281e-01 -9.76840377e-01
-6.04366422e-01 -2.20132470e-01 1.83067486e-01 1.58933386e-01
-7.86762595e-01 1.71744004e-01 -4.61314827e-01 -6.88058317e-01
-3.97090822e-01 -6.08758032e-01 4.00286287e-01 -6.91948831e-01
-4.56907213e-01 -8.00431490e-01 8.22646379e-01 -6.66716039e-01
1.32725906e+00 -1.75569665e+00 -1.45953611e-01 1.14604823e-01
2.47186929e-01 8.78546610e-02 2.24645268e-02 5.79816520e-01
-7.07541257e-02 5.70447862e-01 3.30719709e-01 -2.68857688e-01
2.93366641e-01 2.45857313e-01 -5.50522566e-01 2.34773103e-02
1.91661373e-01 1.08789682e+00 -8.79637003e-01 -7.81269431e-01
-1.54197916e-01 8.67440030e-02 -3.84271622e-01 -1.71683893e-01
-3.20918411e-01 -8.59180763e-02 -6.09573305e-01 7.12791681e-01
5.73802218e-02 -5.87130308e-01 3.77345592e-01 -2.66878337e-01
-3.94049436e-02 7.97409654e-01 -6.92668915e-01 1.41885209e+00
-4.28282320e-01 8.46887589e-01 -3.59304488e-01 -1.11125064e+00
5.96028566e-01 2.86980063e-01 4.77651119e-01 -6.95572734e-01
-1.19795613e-01 2.40591660e-01 -1.31122582e-02 -6.94392979e-01
6.25002801e-01 -4.51926559e-01 -5.94728105e-02 4.42051649e-01
4.43015993e-01 -2.13312104e-01 3.19403499e-01 4.24898207e-01
1.57711291e+00 3.31820041e-01 9.03851166e-02 -2.79640704e-01
8.70261639e-02 2.77548552e-01 2.88584083e-01 8.22479546e-01
1.61682978e-01 1.94269307e-02 4.76348966e-01 -3.70670785e-03
-8.75449061e-01 -7.01681733e-01 -3.68056655e-01 1.50317502e+00
-2.48025984e-01 -7.30585039e-01 -2.55701870e-01 -1.00621986e+00
3.14327866e-01 1.20036256e+00 -5.22106528e-01 -4.15888876e-02
-5.42576253e-01 -7.69119561e-01 9.11500692e-01 6.20125055e-01
3.53013486e-01 -8.91978860e-01 -5.02887778e-02 1.79642767e-01
-2.06084788e-01 -1.66318619e+00 -6.51859120e-02 2.64444321e-01
-7.48410761e-01 -1.14566743e+00 -2.76246548e-01 -6.56967700e-01
5.13782129e-02 -1.02626979e-01 1.61045969e+00 -3.81677188e-02
9.42391232e-02 7.80455887e-01 -5.18962622e-01 -5.28942287e-01
-3.62193674e-01 6.98264539e-01 -2.05016375e-01 -6.36167049e-01
7.09828854e-01 -5.55386007e-01 -6.79488704e-02 1.67425722e-01
-5.70066929e-01 -7.13763908e-02 6.72511935e-01 6.70296729e-01
2.59413570e-01 3.92502844e-01 9.09225285e-01 -1.17574787e+00
8.09403479e-01 -8.12454522e-01 -1.34986073e-01 4.74003166e-01
-9.73939002e-01 -9.98508558e-02 6.07086360e-01 -4.82591838e-01
-1.12074375e+00 -5.81947505e-01 1.71343058e-01 -3.49815860e-02
1.99357606e-02 9.94868457e-01 7.08312378e-04 -4.37144339e-02
8.14605534e-01 -7.53483251e-02 -3.56345505e-01 -5.95595241e-01
4.73366886e-01 5.88563979e-01 4.29026604e-01 -1.34287024e+00
9.63927388e-01 1.78445101e-01 1.96005739e-02 -6.89729750e-01
-1.18142080e+00 -3.08067173e-01 -6.51583850e-01 9.89230797e-02
4.61296022e-01 -8.01886857e-01 -6.98780119e-01 -4.76053543e-02
-6.75474167e-01 -7.26885378e-01 -4.41051573e-01 3.81314576e-01
-4.00798559e-01 3.48608375e-01 -5.90504289e-01 -4.51130360e-01
-2.24325538e-01 -7.07782269e-01 8.83128345e-01 -6.17025718e-02
-3.82008463e-01 -1.46290183e+00 -1.52495608e-01 6.57256484e-01
3.27006131e-01 6.32104799e-02 1.33032668e+00 -1.28610480e+00
-3.94733757e-01 -3.45851660e-01 -2.89481074e-01 2.62794644e-01
9.62007269e-02 3.64975706e-02 -9.70027149e-01 -1.98749173e-02
-2.09661812e-01 -6.22861147e-01 6.26890123e-01 -6.68328330e-02
1.04596817e+00 -3.96009296e-01 -5.91981828e-01 3.32250476e-01
1.18796575e+00 -1.45445898e-01 2.81661749e-01 6.09141707e-01
6.53674364e-01 4.36326385e-01 5.35389900e-01 1.95077136e-01
8.94311190e-01 7.22431540e-01 3.01135574e-02 1.21798337e-01
-2.90050954e-01 -3.60803217e-01 2.84871459e-01 5.15581608e-01
-1.87206596e-01 -5.06674230e-01 -1.28359139e+00 8.85173440e-01
-1.67321491e+00 -1.01688063e+00 -9.61373299e-02 1.88103151e+00
1.33628881e+00 4.39232290e-01 2.12855533e-01 -2.14772195e-01
2.28338405e-01 -5.67778759e-02 -3.59843135e-01 -1.69784367e-01
-1.19211026e-01 6.19158149e-01 6.48506105e-01 2.45489031e-01
-8.44295621e-01 1.30667758e+00 6.75033379e+00 1.11133885e+00
-9.35785174e-01 1.77935228e-01 3.73837113e-01 -1.72330782e-01
-3.54352415e-01 3.23946327e-01 -1.11695898e+00 3.10878366e-01
1.10948277e+00 -4.29958522e-01 5.23907840e-01 7.70429373e-01
-1.53737962e-01 9.22466889e-02 -1.52217710e+00 5.51456273e-01
8.73459429e-02 -1.38978517e+00 8.07116479e-02 2.51101851e-01
3.65074456e-01 2.20699459e-01 1.31489471e-01 8.72978032e-01
7.01194108e-01 -1.27819824e+00 7.05020308e-01 4.49894249e-01
8.07410359e-01 -3.79388183e-01 5.43385327e-01 4.87583280e-01
-8.71669114e-01 -1.04015050e-02 -1.13862813e-01 -1.05255172e-01
-1.73095897e-01 5.07543862e-01 -1.20259297e+00 6.85204327e-01
7.56731510e-01 1.03075337e+00 -1.05315077e+00 3.70224327e-01
-1.01344205e-01 1.22157359e+00 -4.34962839e-01 1.27255455e-01
2.45126978e-01 2.29140192e-01 4.84605134e-01 1.64414036e+00
7.91625008e-02 9.76150706e-02 7.06367418e-02 9.94398475e-01
-4.49825674e-01 9.84266400e-02 -7.38302231e-01 -5.71190059e-01
6.61834478e-01 1.26738000e+00 -4.53035533e-01 -4.86314833e-01
-6.21667385e-01 3.21130991e-01 5.51338732e-01 3.16352338e-01
-5.84208846e-01 -3.13227713e-01 6.31650630e-03 5.00701487e-01
4.75203484e-01 -2.94122957e-02 -5.26311994e-01 -1.22519124e+00
3.77901681e-02 -9.37671483e-01 7.22044766e-01 -7.80441999e-01
-1.73457801e+00 2.40527049e-01 3.66997212e-01 -4.63545620e-01
-4.25872207e-01 -8.33660722e-01 -3.44990671e-01 8.96744728e-01
-1.66437173e+00 -1.47441232e+00 -1.70517534e-01 6.01792753e-01
3.94600809e-01 -3.76089990e-01 8.03746521e-01 3.57417762e-01
-5.40066957e-01 7.21130252e-01 -1.07386298e-01 2.54095674e-01
9.75003481e-01 -1.38368559e+00 2.83173651e-01 7.48331130e-01
2.69480169e-01 8.43395710e-01 5.88764668e-01 -8.67090046e-01
-1.43399775e+00 -8.22072685e-01 1.29252911e+00 -1.18894601e+00
1.19532180e+00 -2.51217782e-01 -9.71065402e-01 1.43375945e+00
2.14694783e-01 -3.09993654e-01 7.61726618e-01 8.69221747e-01
-5.53824246e-01 7.16162771e-02 -1.03377092e+00 3.86313409e-01
1.34775758e+00 -6.55843318e-01 -9.43852425e-01 6.78125679e-01
4.66739684e-01 -3.84581625e-01 -1.62177968e+00 4.96153444e-01
3.50678682e-01 -4.19772148e-01 9.80672419e-01 -8.99432421e-01
6.49943531e-01 2.47958124e-01 -2.46333525e-01 -1.03349018e+00
-6.08891368e-01 -5.56284070e-01 -2.75222868e-01 1.81410658e+00
4.71923679e-01 -5.42622030e-01 6.07531309e-01 7.22050369e-01
-2.79575735e-01 -8.85953128e-01 -7.91339457e-01 -8.46960783e-01
4.55856740e-01 -6.55372560e-01 5.09038687e-01 1.42761779e+00
2.78211802e-01 7.46449053e-01 2.84075201e-01 -8.17542672e-02
2.75500149e-01 1.34913139e-02 7.77487993e-01 -1.41702807e+00
-4.10402566e-01 -6.23567164e-01 -1.55199766e-01 -9.02114332e-01
6.01835012e-01 -1.36904514e+00 -3.19729418e-01 -1.70324576e+00
3.41603845e-01 -8.65942657e-01 -1.14316940e-01 9.59098876e-01
-2.14509383e-01 -1.82606027e-01 -2.51700673e-02 2.38033891e-01
-4.43536729e-01 1.96556553e-01 9.41391826e-01 -6.57701027e-03
1.08073734e-01 -2.07728356e-01 -1.36812186e+00 6.48953259e-01
4.32856202e-01 -2.89491624e-01 -5.91771543e-01 -3.22731942e-01
5.45068145e-01 -2.52528816e-01 4.48984325e-01 -4.60715413e-01
1.47513643e-01 -2.55682796e-01 2.47806057e-01 -1.03703253e-01
1.11238271e-01 -7.91594267e-01 -1.50624573e-01 -7.47620985e-02
-3.06126595e-01 -1.77160814e-01 5.48762679e-01 4.27741021e-01
1.34306505e-01 -3.09170067e-01 3.15133303e-01 -2.35729992e-01
-8.39666545e-01 9.19533968e-02 -2.15268403e-01 7.17344344e-01
5.42005837e-01 -4.61728871e-01 -5.94696283e-01 -2.49072894e-01
-4.90753084e-01 -7.10238740e-02 2.95165926e-01 5.83829403e-01
1.12499222e-01 -9.22363698e-01 -8.75328243e-01 -2.99978077e-01
2.67920315e-01 -1.75075576e-01 -2.62164891e-01 9.38421130e-01
1.00546107e-01 6.19755149e-01 2.76136518e-01 -3.05054694e-01
-1.02017343e+00 4.33214247e-01 2.05522448e-01 -8.50203931e-01
-6.95665538e-01 8.54966342e-01 -7.69756958e-02 -3.58615577e-01
5.59627190e-02 -3.24302942e-01 -1.38018802e-01 5.64234070e-02
-2.12628394e-01 3.37105691e-01 2.74779111e-01 -4.37906384e-01
-4.05243278e-01 2.71316588e-01 -1.99672237e-01 6.32662773e-02
1.66050398e+00 1.38857946e-01 -2.31065545e-02 4.74403530e-01
8.81930888e-01 4.24595416e-01 -6.68580651e-01 -6.38541996e-01
4.01610523e-01 -2.65691340e-01 2.17665434e-01 -1.35915685e+00
-7.75124669e-01 3.93039346e-01 -1.57114327e-01 2.88696229e-01
8.61472905e-01 4.11740303e-01 5.44670939e-01 4.80097026e-01
2.40325898e-01 -1.03983462e+00 -1.66656505e-02 7.22261250e-01
8.14190209e-01 -9.75700140e-01 4.80134189e-01 -3.86113256e-01
-7.14850008e-01 1.06145799e+00 6.35734916e-01 1.56364426e-01
6.11762643e-01 2.37393051e-01 -4.85042125e-01 -7.04439223e-01
-9.17223454e-01 -3.03014182e-02 5.42771220e-01 5.78560054e-01
6.99446082e-01 -2.48103812e-01 -2.53162570e-02 8.74808371e-01
-6.24157012e-01 1.39122859e-01 2.43365154e-01 7.51755357e-01
-5.33650443e-02 -8.34265292e-01 6.17908128e-03 8.11061025e-01
-6.80538535e-01 -4.63178158e-01 -6.92219615e-01 1.34275913e+00
8.73037353e-02 9.93453026e-01 -2.10249960e-01 -3.50199282e-01
5.10799170e-01 3.12749207e-01 5.76584160e-01 -1.02232826e+00
-8.99824202e-01 -3.74515146e-01 6.77264154e-01 -2.45891541e-01
-1.98521912e-01 -8.86114120e-01 -9.38193202e-01 -5.15800536e-01
-4.23261970e-01 1.26984045e-01 3.38847160e-01 1.24134755e+00
4.52010900e-01 6.36316597e-01 -1.02434196e-01 -3.61090213e-01
-6.30970955e-01 -1.28152442e+00 -4.89998072e-01 5.38754582e-01
-1.09517582e-01 -6.87689185e-01 -3.41266304e-01 9.66378301e-02]
|
[9.948092460632324, 8.512968063354492]
|
395fc019-46a4-4dea-b1ba-cda4a9506e45
|
solving-single-objective-tasks-by-preference
| null | null |
https://openreview.net/forum?id=HJxV5yHYwB
|
https://openreview.net/pdf?id=HJxV5yHYwB
|
Solving single-objective tasks by preference multi-objective reinforcement learning
|
There ubiquitously exist many single-objective tasks in the real world that are inevitably related to some other objectives and influenced by them. We call such task as the objective-constrained task, which is inherently a multi-objective problem. Due to the conflict among different objectives, a trade-off is needed. A common compromise is to design a scalar reward function through clarifying the relationship among these objectives using the prior knowledge of experts. However, reward engineering is extremely cumbersome. This will result in behaviors that optimize our reward function without actually satisfying our preferences. In this paper, we explicitly cast the objective-constrained task as preference multi-objective reinforcement learning, with the overall goal of finding a Pareto optimal policy. Combined with Trajectory Preference Domination we propose, a weight vector that reflects the agent's preference for each objective can be learned. We analyzed the feasibility of our algorithm in theory, and further proved in experiments its better performance compared to those that design the reward function by experts.
|
['Feng Chen', 'Shangqi Guo', 'Jinsheng Ren']
|
2019-09-25
| null | null | null | null |
['multi-objective-reinforcement-learning']
|
['methodology']
|
[-1.52840093e-02 -2.22967952e-01 -2.66243815e-01 -2.21122548e-01
-4.20565426e-01 -6.23578548e-01 -2.09404305e-02 -5.14949076e-02
-7.46594191e-01 1.10589933e+00 1.14723057e-01 -6.44949451e-02
-7.96150446e-01 -4.67309356e-01 -3.60061884e-01 -9.17958140e-01
-1.12405054e-01 6.36352599e-01 1.53436614e-02 -3.10681313e-01
6.69512928e-01 2.09546864e-01 -1.20895767e+00 -3.33711982e-01
1.31785953e+00 1.01817715e+00 5.14845610e-01 4.24525142e-01
3.37205604e-02 4.60545272e-01 -6.50126517e-01 -3.27449918e-01
1.33818090e-01 -1.08455211e-01 -7.99360216e-01 2.75255442e-01
-6.00311637e-01 -2.79125929e-01 3.20815623e-01 1.18078351e+00
5.11999071e-01 4.25470591e-01 6.34707212e-01 -1.54479373e+00
-5.22240937e-01 5.60909867e-01 -6.50516212e-01 -1.13777392e-01
9.37604904e-02 2.98878133e-01 1.18642187e+00 -2.34113142e-01
2.17840999e-01 1.11564302e+00 -2.42459252e-02 5.48003018e-01
-1.08470368e+00 -2.02454895e-01 5.46812892e-01 2.71374941e-01
-1.01872540e+00 8.80034268e-02 8.08746040e-01 -3.48627865e-01
2.83761173e-01 1.94572300e-01 5.20717859e-01 1.03219652e+00
2.44607776e-01 8.50494564e-01 1.05379224e+00 -5.46882022e-03
4.81628180e-01 3.14888239e-01 -1.97907507e-01 2.97312856e-01
3.32375944e-01 3.47431183e-01 -2.06514820e-01 -1.14658192e-01
5.00779390e-01 4.74413596e-02 -5.05556345e-01 -6.87325835e-01
-9.84116673e-01 7.50330985e-01 2.49767989e-01 1.31785944e-01
-5.42047441e-01 2.24681467e-01 1.03053972e-01 2.83456773e-01
-2.57482305e-02 8.75239193e-01 -4.70304579e-01 -1.70719609e-01
-4.21137094e-01 2.56716013e-01 4.77876574e-01 6.11058235e-01
6.15509808e-01 6.40039593e-02 -4.37684566e-01 5.59554040e-01
3.57756734e-01 2.05565706e-01 1.76909909e-01 -1.25689387e+00
4.89592046e-01 4.93466824e-01 9.74665582e-01 -8.49612355e-01
-3.59304070e-01 -9.03098464e-01 -4.66119051e-01 6.94540381e-01
6.22771025e-01 -6.11158431e-01 -3.67063522e-01 1.82508576e+00
1.87722385e-01 -3.43370676e-01 -1.81773286e-02 1.42288494e+00
-2.31976599e-01 4.74619031e-01 4.76260073e-02 -4.32996750e-01
8.71399283e-01 -7.98457801e-01 -8.09338808e-01 -2.42524832e-01
1.50762469e-01 -4.76008445e-01 1.19519043e+00 4.32381809e-01
-9.08377409e-01 -6.65391609e-02 -8.69878829e-01 6.97708309e-01
1.50924651e-02 5.30355945e-02 5.66143513e-01 5.19089460e-01
-7.55483091e-01 6.49055958e-01 -2.31642425e-01 4.00452651e-02
1.55538455e-01 6.39071345e-01 2.13113323e-01 3.91750574e-01
-1.10616469e+00 1.06915450e+00 4.18668449e-01 1.61105216e-01
-1.05567920e+00 -4.22851980e-01 -2.32275501e-01 2.72615433e-01
1.15154970e+00 -6.22229517e-01 1.39568913e+00 -1.19521117e+00
-1.78658903e+00 1.36699408e-01 2.04800218e-01 3.50492820e-02
8.04329515e-01 -1.32534429e-01 -2.08473593e-01 -1.95073947e-01
-7.05153961e-03 2.79599875e-01 1.01053619e+00 -1.58192325e+00
-8.67557526e-01 -1.39531061e-01 5.34892023e-01 5.58860421e-01
-6.12491965e-01 -3.13192271e-02 -7.87696242e-02 -3.02173913e-01
-5.86037636e-01 -7.41597772e-01 -6.61800385e-01 -2.17996299e-01
-2.60899365e-01 -1.85021579e-01 3.87589484e-01 -1.62488446e-01
1.33898127e+00 -1.85340393e+00 5.24284661e-01 2.79733539e-01
2.32820809e-01 6.79612234e-02 -2.55237937e-01 2.57063329e-01
4.23095822e-01 1.39404938e-01 -2.26604268e-01 -7.81781375e-02
3.13888252e-01 2.89582670e-01 -1.65402368e-01 3.64369363e-01
1.76164925e-01 6.41043425e-01 -1.22064114e+00 -3.74159783e-01
-1.05462953e-01 -3.00477892e-02 -5.49446464e-01 3.74668419e-01
-4.97694701e-01 4.82037157e-01 -1.13551164e+00 5.19624770e-01
3.70433927e-01 -2.93909520e-01 2.84838885e-01 7.22209588e-02
-2.30461314e-01 -3.03333044e-01 -1.27885878e+00 1.26351666e+00
-4.56151873e-01 -3.68518080e-03 3.66560906e-01 -1.10650110e+00
8.72198939e-01 2.45546907e-01 7.25379646e-01 -4.34664369e-01
4.86427903e-01 3.38639289e-01 2.28599936e-01 -6.24019146e-01
6.26547575e-01 -1.31654814e-01 -6.60548136e-02 4.58282918e-01
-3.05088609e-01 6.74722418e-02 1.64450869e-01 -2.45890424e-01
7.00375974e-01 2.40871727e-01 2.42635667e-01 -5.38975239e-01
5.21473706e-01 2.97897477e-02 7.86386728e-01 7.05166996e-01
-4.59377319e-01 1.44814342e-01 7.13098884e-01 -1.48034036e-01
-5.81536651e-01 -7.61297286e-01 1.91344693e-01 9.02578592e-01
5.84778607e-01 2.11135581e-01 -3.23323131e-01 -7.46246040e-01
-2.53243297e-02 6.94837213e-01 -4.83783126e-01 -1.17994897e-01
-3.19059402e-01 -8.20336938e-01 -9.23881531e-02 1.79750368e-01
3.49847198e-01 -8.43698502e-01 -1.12887621e+00 4.70389664e-01
-1.54166996e-01 -7.15253532e-01 -1.07215381e+00 2.01139405e-01
-6.34868622e-01 -1.01949084e+00 -1.00482380e+00 -3.07870954e-01
8.08877766e-01 1.20987087e-01 9.30409551e-01 -6.68128282e-02
3.63286622e-02 4.46321040e-01 -2.84988701e-01 -2.27995694e-01
1.44202486e-01 -7.40882084e-02 1.38219342e-01 4.96392608e-01
-1.99117362e-01 -5.21419883e-01 -6.96677327e-01 5.34594178e-01
-9.45120692e-01 -2.93700814e-01 6.80500746e-01 7.59479344e-01
2.64505118e-01 4.38925654e-01 8.74670267e-01 -3.16021234e-01
1.16183710e+00 -4.66699690e-01 -1.01885116e+00 5.81455410e-01
-5.72181523e-01 5.58850050e-01 9.02794659e-01 -7.81740427e-01
-1.16163409e+00 4.02562283e-02 4.31546241e-01 -2.30041578e-01
2.75349021e-01 5.73529065e-01 -3.97969157e-01 -7.71106407e-02
1.29169658e-01 -3.83449607e-02 -1.00124476e-03 -2.02397063e-01
2.18905643e-01 4.17828560e-01 8.50044638e-02 -1.20221829e+00
6.89390898e-01 8.22732151e-02 1.84684679e-01 -3.12149584e-01
-8.08339536e-01 -4.10321318e-02 -8.11723396e-02 -7.19004452e-01
7.13413715e-01 -2.71743536e-01 -1.49675381e+00 1.10508077e-01
-1.19884765e+00 -2.27208346e-01 -1.51815131e-01 5.63732922e-01
-6.77612126e-01 1.48968801e-01 1.68960646e-01 -1.33358455e+00
1.16392843e-01 -1.29356349e+00 3.23824227e-01 5.08925200e-01
-1.22509435e-01 -9.28980350e-01 1.78455766e-02 9.17675197e-02
6.67455852e-01 2.04438284e-01 8.49078536e-01 -1.52183503e-01
-7.45477438e-01 2.39658549e-01 -7.96838552e-02 -7.24910647e-02
8.54104012e-02 -3.23924907e-02 -4.20558155e-01 -2.35852927e-01
1.73837990e-01 -3.57087523e-01 4.92387503e-01 4.12425160e-01
1.29038203e+00 -6.25180185e-01 -1.16596982e-01 3.23407024e-01
1.66446018e+00 7.07795203e-01 2.07289502e-01 5.01986146e-01
2.45912254e-01 9.24543083e-01 9.26789224e-01 8.21159542e-01
2.76893884e-01 7.49765694e-01 9.88882363e-01 2.04116166e-01
7.24960029e-01 7.43301213e-02 4.22401339e-01 1.76478609e-01
-4.65162575e-01 -5.36835790e-01 -5.48146725e-01 5.51675141e-01
-2.36351299e+00 -9.38036025e-01 1.82746887e-01 2.37146950e+00
7.72885144e-01 1.15806744e-01 4.43221688e-01 -6.98923245e-02
8.20352018e-01 -6.77194372e-02 -8.78119111e-01 -4.72459197e-01
9.04716030e-02 -3.62357646e-01 5.93920767e-01 6.17510617e-01
-7.42938697e-01 4.35764670e-01 6.01330948e+00 8.17661762e-01
-1.12788165e+00 -1.51682496e-01 6.58862770e-01 -3.92219007e-01
-6.91800535e-01 -4.02523614e-02 -5.67110300e-01 6.40274048e-01
2.78150469e-01 -7.88340509e-01 9.57129836e-01 6.62041485e-01
4.89744782e-01 -1.92340806e-01 -8.17628384e-01 7.16917634e-01
-4.49602664e-01 -7.11264968e-01 -3.50328863e-01 3.10967803e-01
7.16639459e-01 -5.97567916e-01 3.14228773e-01 1.70544818e-01
7.21455097e-01 -1.02841711e+00 8.28802705e-01 5.89425564e-01
3.91245425e-01 -1.01710665e+00 6.10476971e-01 5.93971968e-01
-9.88835990e-01 -5.23465037e-01 -1.98497310e-01 -1.62179857e-01
3.68685991e-01 6.03252470e-01 -2.92683661e-01 7.10076571e-01
2.70328641e-01 3.73545468e-01 1.75332934e-01 1.32136691e+00
-3.21517646e-01 -2.42775530e-02 -1.33818969e-01 -6.22799158e-01
5.29995799e-01 -5.31438589e-01 7.21415877e-01 4.71987307e-01
5.92249155e-01 8.71393364e-03 3.95057648e-01 1.10475039e+00
1.98514134e-01 -6.32538423e-02 -3.07777882e-01 -1.74773902e-01
3.27224433e-01 1.32059264e+00 -5.81310093e-01 6.60859197e-02
6.08817814e-03 7.99798131e-01 3.98344338e-01 6.15133107e-01
-1.21537459e+00 -3.19832921e-01 8.82254183e-01 -3.48307729e-01
1.62091747e-01 -1.91941455e-01 -2.83916384e-01 -9.86279845e-01
2.07814835e-02 -7.35166252e-01 1.73428491e-01 -3.40762645e-01
-1.38835323e+00 5.35929739e-01 -2.83919666e-02 -1.39778101e+00
-2.71551614e-03 -5.84202945e-01 -7.50134110e-01 8.18910956e-01
-1.77667987e+00 -2.88742661e-01 9.65084136e-02 5.85892439e-01
2.85631061e-01 -3.08141205e-02 3.53609264e-01 2.64722288e-01
-7.62106657e-01 3.06771040e-01 2.65247703e-01 -5.63549280e-01
5.15721202e-01 -1.24993515e+00 -6.81713700e-01 5.14833689e-01
-4.94239926e-01 2.98619777e-01 9.48476255e-01 -3.76799554e-01
-1.43951833e+00 -7.57465065e-01 6.40370965e-01 8.68496299e-02
9.67596292e-01 1.62424475e-01 -4.56967801e-01 1.46620050e-01
1.35138258e-01 -3.80263239e-01 3.72911602e-01 -9.08815190e-02
2.55678713e-01 -2.23924085e-01 -1.17080462e+00 9.11750495e-01
9.85654116e-01 8.92444402e-02 -3.72570127e-01 -1.02880802e-02
5.94158649e-01 -6.48661554e-02 -6.09594107e-01 2.78043002e-01
4.58903700e-01 -6.16726398e-01 8.74391198e-01 -8.29591036e-01
4.85383898e-01 -4.75443333e-01 -7.23626241e-02 -1.83274317e+00
-4.36162978e-01 -8.18294048e-01 5.97799197e-03 9.73185420e-01
4.02163625e-01 -7.90670753e-01 6.13094568e-01 8.64562809e-01
9.48108081e-03 -8.93442631e-01 -8.48280013e-01 -1.21015120e+00
-1.22629091e-01 9.59458128e-02 7.15015709e-01 6.89476907e-01
1.44496769e-01 3.01720709e-01 -8.15918207e-01 2.39316940e-01
9.49043453e-01 3.59190226e-01 2.31854230e-01 -1.12922299e+00
-6.08697176e-01 -8.00846279e-01 5.01640737e-01 -1.01917875e+00
9.33695436e-02 -4.46196944e-01 3.35112542e-01 -1.41262257e+00
-1.59600358e-02 -6.66479051e-01 -5.34779847e-01 2.69993007e-01
-3.06220353e-01 -5.72441697e-01 3.45671624e-01 4.04639542e-02
-7.73512602e-01 9.48040307e-01 1.79135311e+00 -2.49369666e-01
-3.61691624e-01 2.54829347e-01 -9.93580043e-01 4.94943887e-01
1.13192666e+00 -5.98537445e-01 -6.51822269e-01 -5.63705444e-01
2.59199381e-01 6.45817995e-01 -3.04903667e-02 -6.27681017e-01
2.51051784e-01 -1.19000173e+00 -1.86753795e-01 -2.07152575e-01
4.39861268e-01 -1.23506665e+00 8.45036879e-02 4.83123779e-01
-3.58587563e-01 -5.64052574e-02 -3.20654452e-01 5.12131512e-01
-9.10711735e-02 -5.94805658e-01 5.71562529e-01 -1.13780327e-01
-5.93577266e-01 4.69889075e-01 -4.82228696e-01 2.53186822e-02
1.25489736e+00 7.64622092e-02 -2.75810421e-01 -4.79330212e-01
-5.47969043e-01 8.79877329e-01 1.50682107e-01 3.27779651e-01
5.28541565e-01 -1.09877729e+00 -5.33680439e-01 -4.86560285e-01
-2.23916128e-01 -3.42949688e-01 1.15133058e-02 8.41244340e-01
2.06121519e-01 3.86248231e-01 -4.54899788e-01 -4.10410054e-02
-9.16172326e-01 6.45396709e-01 4.31818247e-01 -6.56905890e-01
1.94603354e-01 3.77975702e-01 5.61037809e-02 -1.51803613e-01
4.02512908e-01 -6.88668266e-02 -5.58170259e-01 2.07203865e-01
2.19870701e-01 5.54317474e-01 -5.69959044e-01 -2.72802301e-02
-3.26077193e-01 6.57402933e-01 2.68772691e-01 -3.43210936e-01
1.30434644e+00 -2.60362864e-01 1.22019917e-01 8.12484622e-02
7.03415394e-01 -1.40158087e-01 -1.57221127e+00 1.62057169e-02
1.79721743e-01 -7.08294153e-01 5.79347722e-02 -9.71462548e-01
-1.08929431e+00 5.92148542e-01 1.97027773e-01 3.55723888e-01
1.33166993e+00 -4.91945177e-01 4.64236528e-01 4.52885091e-01
7.42390871e-01 -1.37797868e+00 4.78211939e-01 5.74063420e-01
8.28536212e-01 -1.23497212e+00 -2.90944934e-01 -9.81760100e-02
-9.78537619e-01 1.09292948e+00 7.72326946e-01 -6.30324185e-02
3.73948604e-01 5.94919212e-02 -1.72381550e-01 6.08870871e-02
-6.55860722e-01 -4.47487056e-01 2.04777420e-01 5.15481532e-01
4.95344261e-03 2.22167209e-01 -7.46884704e-01 7.31684268e-01
3.78574610e-01 1.40672937e-01 5.00243127e-01 8.39293063e-01
-6.75657809e-01 -1.58583713e+00 -5.82703233e-01 2.83645213e-01
-2.98446834e-01 3.54643553e-01 -1.45552441e-01 3.46307993e-01
4.10783151e-03 1.22260571e+00 -2.91806459e-01 -1.90861091e-01
3.11318576e-01 -4.18739319e-01 4.15495783e-01 -2.16481119e-01
-6.44245684e-01 -1.30129000e-02 8.11368451e-02 -5.34650207e-01
-3.49413574e-01 -3.83900762e-01 -1.15808034e+00 -2.46734813e-01
-7.40925297e-02 4.06200141e-01 5.02376676e-01 9.38501894e-01
1.17860056e-01 8.09428453e-01 9.97068465e-01 -6.30611897e-01
-1.16397703e+00 -2.78710335e-01 -6.74138010e-01 8.33687186e-02
3.55770230e-01 -9.24252927e-01 -4.05556560e-01 -5.27594686e-01]
|
[4.394063472747803, 2.485564708709717]
|
5c503cab-7030-458d-a17d-cefc0de2f08b
|
scalable-deletion-robust-submodular
| null | null |
https://icml.cc/Conferences/2018/Schedule?showEvent=1927
|
http://proceedings.mlr.press/v80/kazemi18a/kazemi18a.pdf
|
Scalable Deletion-Robust Submodular Maximization: Data Summarization with Privacy and Fairness Constraints
|
Can we efficiently extract useful information from a large user-generated dataset while protecting the privacy of the users and/or ensuring fairness in representation? We cast this problem as an instance of a deletion-robust submodular maximization where part of the data may be deleted or masked due to privacy concerns or fairness criteria. We propose the first memory-efficient centralized, streaming, and distributed methods with constant-factor approximation guarantees against any number of adversarial deletions. We extensively evaluate the performance of our algorithms on real-world applications, including (i) Uber-pick up locations with location privacy constraints; (ii) feature selection with fairness constraints for income prediction and crime rate prediction; and (iii) robust to deletion summarization of census data, consisting of 2,458,285 feature vectors. Our experiments show that our solution is robust against even $80%$ of data deletion.
|
['Morteza Zadimoghaddam', 'Amin Karbasi', 'Ehsan Kazemi']
|
2018-07-01
| null | null | null |
icml-2018-7
|
['data-summarization']
|
['miscellaneous']
|
[ 2.92559117e-01 2.08101407e-01 -3.31580788e-01 -4.99198228e-01
-1.05095911e+00 -9.14806962e-01 -3.72225642e-02 6.14832938e-01
-7.31269538e-01 9.76772189e-01 4.56550896e-01 -2.82990307e-01
-2.89640248e-01 -8.31386805e-01 -7.47645199e-01 -7.13125646e-01
-5.95366895e-01 4.15928334e-01 -3.60022277e-01 -3.27940024e-02
1.50330171e-01 6.08886182e-01 -1.11050224e+00 -1.86287481e-02
8.75957668e-01 1.05897820e+00 -4.08792675e-01 3.52725565e-01
6.38160348e-01 4.13756341e-01 -3.66776139e-01 -8.72860849e-01
1.07650268e+00 1.73933655e-01 -6.99641705e-01 -1.51392832e-01
2.76495010e-01 -9.82827961e-01 -3.68833840e-01 1.26296723e+00
6.84170842e-01 1.61314189e-01 4.58089292e-01 -1.76878524e+00
-5.35326958e-01 7.92644680e-01 -1.10412920e+00 -1.06098495e-01
2.40113676e-01 -1.17708586e-01 8.04787576e-01 -6.12435758e-01
7.60314524e-01 1.00376213e+00 6.09187126e-01 4.54552323e-01
-1.49498165e+00 -8.47632229e-01 5.59664192e-03 -3.11753094e-01
-1.85833383e+00 -8.04588437e-01 3.58368605e-01 -1.85168728e-01
5.78205884e-01 1.16375375e+00 -3.36251641e-03 5.48547089e-01
-9.22269523e-02 8.83952558e-01 4.72203255e-01 2.18514517e-01
4.53587502e-01 4.19952899e-01 5.77647574e-02 2.14460209e-01
1.06388688e+00 -1.70502663e-01 -5.08365452e-01 -1.44162750e+00
2.28613131e-02 2.59410113e-01 -3.12406123e-01 -5.72425067e-01
-6.03117764e-01 1.03732467e+00 -3.01656369e-02 -6.27317131e-01
-4.70613182e-01 2.11751059e-01 3.32742244e-01 2.63859272e-01
6.49994791e-01 1.51087448e-01 -6.55352175e-01 2.97212601e-01
-1.00819647e+00 9.76083159e-01 7.53528893e-01 1.26418543e+00
5.69468975e-01 -9.70857218e-02 -2.17228815e-01 4.03861105e-01
1.94080994e-02 8.24933767e-01 -4.14350368e-02 -1.07527530e+00
1.03304958e+00 3.46010774e-01 6.08611882e-01 -1.42656267e+00
-2.79616684e-01 1.46408215e-01 -9.65921223e-01 -3.69455032e-02
2.48780698e-01 -8.80212545e-01 -1.79314166e-01 1.99033153e+00
6.18273020e-01 -3.08404982e-01 -1.48719952e-01 6.55366838e-01
3.45817059e-01 4.83305871e-01 -3.10204700e-02 -6.15556359e-01
1.00543523e+00 -9.62762162e-02 -5.56380093e-01 -8.02346915e-02
6.52860105e-01 -4.89130169e-02 -7.14961765e-03 2.65737981e-01
-1.39522684e+00 4.71020162e-01 -5.94152033e-01 -2.25989804e-01
-2.48025522e-01 -4.26674813e-01 7.87150145e-01 1.17468333e+00
-7.97422588e-01 2.07374722e-01 -5.76628447e-01 3.19667794e-02
1.06595266e+00 6.50014460e-01 -8.29225183e-01 -1.29870787e-01
-9.84127641e-01 1.31237224e-01 1.36114098e-03 -2.45337903e-01
-6.18682206e-01 -1.10907626e+00 -9.87998247e-01 3.44168544e-01
3.82192522e-01 -5.34833491e-01 6.24783337e-01 -5.86717546e-01
-1.93873599e-01 7.73489833e-01 -2.96675056e-01 -6.45746291e-01
9.60110605e-01 2.20338106e-02 -1.36735633e-01 1.51890889e-01
3.29354644e-01 2.71831959e-01 7.05913484e-01 -1.33929753e+00
-8.84396613e-01 -9.71074641e-01 -3.17910910e-01 2.58365005e-01
-9.54015315e-01 2.62883067e-01 4.61617783e-02 -8.56242061e-01
-2.87492603e-01 -6.79193735e-01 -7.45932639e-01 2.11486131e-01
-6.70803249e-01 4.11829770e-01 7.09518850e-01 -1.10173011e+00
1.46032345e+00 -2.23380733e+00 -2.35394776e-01 6.59935594e-01
2.74213016e-01 -8.65538120e-02 -3.05944592e-01 6.57558084e-01
1.70920521e-01 4.75846797e-01 -5.95014036e-01 -6.77145958e-01
1.62640154e-01 -2.58002698e-01 -4.98068452e-01 1.00541985e+00
-4.90619779e-01 7.60547996e-01 -6.80546761e-01 -6.60393611e-02
-4.72270697e-01 1.67794079e-02 -8.06853831e-01 -9.03442204e-02
2.46837303e-01 -2.04734191e-01 -5.01418293e-01 8.42379451e-01
1.57873404e+00 4.09558654e-01 4.17191476e-01 4.97416914e-01
3.30080956e-01 -2.60019153e-01 -1.58444369e+00 1.27662134e+00
1.74796820e-01 1.11630522e-01 9.45827544e-01 -7.41761863e-01
5.92154920e-01 1.05717197e-01 8.33616793e-01 -3.28629345e-01
-6.19241074e-02 -3.80654745e-02 -6.50240779e-01 -3.05385411e-01
8.70555282e-01 1.26971126e-01 -5.87413013e-01 9.83006895e-01
-6.17964566e-01 2.60692179e-01 -2.96415359e-01 4.80164587e-01
1.24179685e+00 -8.78539801e-01 3.48624170e-01 -3.95836174e-01
-2.46609017e-01 -3.13098073e-01 9.83405948e-01 1.01503491e+00
-2.89679468e-01 7.47680128e-01 6.74230814e-01 -4.83190000e-01
-7.39422321e-01 -9.90788281e-01 -4.94982861e-02 1.21660399e+00
6.44237548e-02 -2.48627931e-01 -6.29919231e-01 -9.79627550e-01
1.06331837e+00 8.47686529e-01 -6.60627246e-01 7.94943795e-02
-5.78673184e-02 -1.14558995e+00 6.24181211e-01 3.14646930e-01
1.66006520e-01 -3.42208952e-01 -4.16897297e-01 -5.07470891e-02
-2.78755814e-01 -8.05238545e-01 -9.63155389e-01 -1.04542524e-01
-6.59850895e-01 -9.14572775e-01 -4.74297136e-01 -3.42695653e-01
1.21384490e+00 5.03805518e-01 5.71861029e-01 5.32745663e-03
-4.63928610e-01 4.33545411e-01 1.01332165e-01 -7.57693172e-01
2.88164049e-01 -3.63662001e-03 2.39846796e-01 3.74033719e-01
3.77396494e-01 -6.25579000e-01 -6.33377016e-01 -1.20850123e-01
-1.23962414e+00 -6.44124568e-01 -1.87398553e-01 3.01933646e-01
4.84582573e-01 3.60450059e-01 7.35237062e-01 -1.20209980e+00
8.50772083e-01 -1.05179954e+00 -6.30810022e-01 1.69077650e-01
-4.10339534e-01 -4.19605702e-01 4.66128975e-01 -5.13403490e-02
-9.07673240e-01 3.82021695e-01 2.19440401e-01 8.36411044e-02
1.50929824e-01 2.77929991e-01 -6.54121459e-01 -2.43934572e-01
4.22042668e-01 2.38261893e-01 -6.69749156e-02 -3.63874555e-01
4.54774410e-01 9.71082389e-01 2.79375255e-01 -5.82639217e-01
8.15065324e-01 9.15930212e-01 6.79934472e-02 -8.14108253e-01
-1.04296133e-01 -4.05816525e-01 -1.80717930e-01 5.25354028e-01
1.79700077e-01 -1.10765302e+00 -9.58576202e-01 6.10018492e-01
-8.61381829e-01 -7.14180022e-02 -4.78891075e-01 2.01321002e-02
-3.25634062e-01 7.22971797e-01 -2.68373847e-01 -1.16336274e+00
-6.66516364e-01 -6.24435008e-01 7.85138369e-01 -8.89261663e-02
-9.82168615e-02 -5.70394874e-01 -1.15829252e-01 6.16245985e-01
2.81939715e-01 7.47848451e-01 6.12192154e-01 -9.60655451e-01
-4.62235004e-01 -9.06675279e-01 -1.55848086e-01 1.21402971e-01
4.26152684e-02 -4.61831838e-01 -8.19562674e-01 -8.57368350e-01
-1.63297519e-01 -3.62241596e-01 7.54757345e-01 3.89580667e-01
1.83264804e+00 -1.34336996e+00 -3.78849685e-01 9.47239876e-01
1.38962471e+00 -8.57012644e-02 6.30227983e-01 -9.60168466e-02
3.06663543e-01 7.79363513e-01 6.52915359e-01 1.58852911e+00
6.60836220e-01 2.76679426e-01 6.43861592e-01 1.86486781e-01
9.81660366e-01 -3.81061673e-01 1.24217920e-01 -2.24431425e-01
1.26740828e-01 -4.51941639e-01 -5.38908541e-01 9.84999359e-01
-2.06934404e+00 -1.14298403e+00 1.82343319e-01 2.77216268e+00
8.16608608e-01 -5.66006899e-01 4.67603415e-01 -2.51677066e-01
5.62283814e-01 3.79430592e-01 -7.43163526e-01 -6.66733980e-01
-3.30650359e-01 -3.23354565e-02 1.42613864e+00 3.07926297e-01
-1.25936651e+00 3.26262295e-01 6.38646460e+00 9.56485331e-01
-3.80480349e-01 1.39438480e-01 1.27554238e+00 -9.00107622e-01
-8.97846937e-01 -3.12109232e-01 -6.57907605e-01 6.29483640e-01
4.85299736e-01 -7.56381094e-01 7.19356060e-01 9.92919743e-01
8.90289396e-02 -8.09290931e-02 -8.40982378e-01 9.42686617e-01
1.26916677e-01 -1.40869224e+00 -9.58438441e-02 6.03994071e-01
1.17598414e+00 -1.95196733e-01 2.79662311e-01 -1.89808846e-01
6.40416503e-01 -1.10005891e+00 5.04890800e-01 3.68922323e-01
9.99802649e-01 -1.42797208e+00 5.43248296e-01 5.42990685e-01
-8.59762371e-01 -4.73776013e-01 -6.06926262e-01 -2.19773571e-03
1.48657694e-01 6.95950627e-01 -2.77271301e-01 5.76705337e-01
8.22615802e-01 -8.37358925e-03 -2.99321264e-01 9.16297555e-01
3.13140839e-01 4.04298067e-01 -8.87653708e-01 2.01159731e-01
-1.77158445e-01 -1.15435813e-02 6.51772201e-01 1.16202116e+00
4.30694997e-01 4.69723195e-01 -4.92975058e-04 5.88161945e-01
-6.16713941e-01 2.54374236e-01 -9.94808435e-01 3.09772551e-01
1.03631997e+00 1.11023438e+00 -6.96460679e-02 1.30417943e-01
-1.05693743e-01 9.25147951e-01 1.91633046e-01 4.32503909e-01
-5.98687291e-01 -6.15291595e-01 1.35258973e+00 3.60480636e-01
4.14386272e-01 1.29732564e-01 -7.50625372e-01 -1.04444063e+00
2.60027617e-01 -8.06232452e-01 8.58841538e-01 -5.95248975e-02
-1.34625208e+00 -2.46206298e-01 -8.00596625e-02 -6.15548730e-01
-2.43315827e-02 1.52016282e-01 -5.15778363e-01 8.92893016e-01
-1.09124386e+00 -9.48168635e-01 2.62141019e-01 8.41543019e-01
-3.82092863e-01 -2.27336124e-01 6.88724518e-01 2.92311311e-01
-5.54921448e-01 1.27824771e+00 6.94543421e-01 1.53125182e-01
4.15423214e-01 -9.93753970e-01 3.73040110e-01 1.22737384e+00
-5.46068549e-01 6.97241306e-01 4.79948848e-01 -7.75036037e-01
-1.71432424e+00 -1.53876293e+00 1.21063817e+00 -4.10244137e-01
1.33107886e-01 -7.87938833e-01 -4.89502639e-01 1.00615442e+00
-2.05477417e-01 3.59205872e-01 1.08339918e+00 -2.22547755e-01
-3.38929802e-01 -4.69691306e-01 -2.32415915e+00 4.48222667e-01
1.17753863e+00 -1.58937678e-01 1.96181089e-02 4.26785439e-01
6.69568479e-01 -2.72127241e-01 -7.64410257e-01 9.52521563e-02
5.63671172e-01 -5.61870575e-01 1.13746357e+00 -9.68600154e-01
-3.30220349e-02 1.36084810e-01 -5.16087294e-01 -8.30457568e-01
-3.83247077e-01 -1.21066046e+00 -1.84152365e-01 1.52995729e+00
4.03453887e-01 -8.45841825e-01 9.65611100e-01 1.81243551e+00
8.64297986e-01 -4.21021134e-01 -1.21246791e+00 -5.60598373e-01
1.41528055e-01 -3.01429003e-01 1.20621324e+00 1.19449210e+00
8.87298808e-02 -7.26936221e-01 -1.03819191e+00 5.49570501e-01
1.18293297e+00 1.34786174e-01 1.02362537e+00 -8.66133273e-01
-1.33743715e-02 4.23234366e-02 -2.26559043e-01 -4.86986548e-01
1.35547027e-01 -7.51680315e-01 -3.71735722e-01 -1.15419412e+00
6.26536489e-01 -4.53659981e-01 -1.57203168e-01 8.81568551e-01
-2.09594667e-01 1.24825165e-02 2.50218600e-01 -3.50971550e-01
-5.56889355e-01 5.27671754e-01 2.80296445e-01 -2.36700043e-01
-2.06256673e-01 3.84098411e-01 -1.57837260e+00 3.71000707e-01
6.06536508e-01 -8.17976534e-01 -3.73481274e-01 -4.70107943e-01
3.57041806e-01 2.17779025e-01 2.55316228e-01 -3.42866123e-01
1.36583745e-01 -8.39355648e-01 3.84349883e-01 -5.63021243e-01
1.18287809e-01 -1.20448494e+00 3.09380591e-01 5.16448796e-01
-4.69031274e-01 -7.66864195e-02 1.35758873e-02 6.83755457e-01
2.99547046e-01 -1.36175632e-01 4.45390761e-01 2.34819986e-02
-2.44654998e-01 8.89832497e-01 -2.18162969e-01 3.32780123e-01
1.44793808e+00 -4.33417223e-02 -5.16328871e-01 -7.45404661e-01
-3.18149179e-01 8.94412577e-01 5.13483822e-01 8.81696716e-02
7.49426365e-01 -1.26073730e+00 -1.17127061e+00 2.45522991e-01
8.70179608e-02 1.04998671e-01 4.96579647e-01 3.49285305e-01
-1.36322260e-01 1.48788884e-01 4.31006812e-02 4.11467999e-01
-1.53516150e+00 8.54811847e-01 -1.97425440e-01 4.20739762e-02
-1.21161744e-01 9.98945415e-01 8.77733678e-02 -4.77801263e-01
4.23533082e-01 2.81723291e-01 4.66557175e-01 3.00675720e-01
9.94140267e-01 9.89383459e-01 -1.13936812e-01 -5.29996395e-01
-5.92481554e-01 -3.50754052e-01 2.16433294e-02 -1.02806248e-01
1.68707430e+00 -5.37868559e-01 -3.65107775e-01 -5.46656430e-01
1.27832997e+00 5.56850731e-01 -1.00307870e+00 -9.93284434e-02
-3.97905201e-01 -1.14214361e+00 -3.01258028e-01 -5.41782141e-01
-1.49068928e+00 1.50939927e-01 1.87223375e-01 1.88922420e-01
1.31164563e+00 -4.39881891e-01 1.11095202e+00 1.43293887e-01
7.50037491e-01 -1.16705549e+00 -1.02835786e+00 -3.72305512e-01
8.28265965e-01 -1.12870800e+00 4.33144808e-01 -2.39933714e-01
-7.37297118e-01 5.33641875e-01 2.86997378e-01 -4.26624939e-02
9.15360093e-01 4.94317263e-01 -3.89280945e-01 2.76790142e-01
-7.22579539e-01 4.03701514e-01 -2.01637462e-01 9.42408442e-01
-3.98224384e-01 5.35167456e-01 -4.82175559e-01 1.33528221e+00
3.69212069e-02 -1.13893166e-01 6.53413236e-01 1.13174820e+00
-2.28052318e-01 -7.75854170e-01 -4.90073949e-01 8.81407499e-01
-8.28361392e-01 -1.22356817e-01 -5.07746577e-01 2.49613076e-01
8.56064186e-02 1.07037485e+00 -2.20367331e-02 -6.77912086e-02
1.11424811e-01 -3.05792183e-01 -9.71776024e-02 -4.19952899e-01
-7.50418186e-01 -4.40321654e-01 7.67714828e-02 -7.52273321e-01
2.83841997e-01 -1.16078913e+00 -9.39127326e-01 -9.24700499e-01
-2.71575898e-02 1.36203751e-01 4.45966244e-01 3.82949203e-01
8.80525410e-01 -6.20661616e-01 1.23883379e+00 -3.69134307e-01
-1.17574084e+00 -2.74364859e-01 -1.22329009e+00 5.78216851e-01
4.08807516e-01 2.21803337e-01 -3.71183425e-01 -3.17451477e-01]
|
[6.480630874633789, 5.207437038421631]
|
6d20cc38-32b1-4b62-9eb0-92d7f8eb07b9
|
unit-based-speech-to-speech-translation
|
2305.15405
| null |
https://arxiv.org/abs/2305.15405v1
|
https://arxiv.org/pdf/2305.15405v1.pdf
|
Unit-based Speech-to-Speech Translation Without Parallel Data
|
We propose an unsupervised speech-to-speech translation (S2ST) system that does not rely on parallel data between the source and target languages. Our approach maps source and target language speech signals into automatically discovered, discrete units and reformulates the problem as unsupervised unit-to-unit machine translation. We develop a three-step training procedure that involves (a) pre-training an unit-based encoder-decoder language model with a denoising objective (b) training it with word-by-word translated utterance pairs created by aligning monolingual text embedding spaces and (c) running unsupervised backtranslation bootstrapping off of the initial translation model. Our approach avoids mapping the speech signal into text and uses speech-to-unit and unit-to-speech models instead of automatic speech recognition and text to speech models. We evaluate our model on synthetic-speaker Europarl-ST English-German and German-English evaluation sets, finding that unit-based translation is feasible under this constrained scenario, achieving 9.29 ASR-BLEU in German to English and 8.07 in English to German.
|
['Eunsol Choi', 'David Harwath', 'Anirudh Srinivasan', 'Anuj Diwan']
|
2023-05-24
| null | null | null | null |
['speech-to-speech-translation']
|
['speech']
|
[ 6.97574556e-01 5.12286067e-01 -7.40410462e-02 -6.88668072e-01
-1.68265259e+00 -6.43307269e-01 8.50054741e-01 -1.65565088e-01
-4.72912401e-01 7.12097824e-01 4.70731676e-01 -8.01355064e-01
6.46273196e-01 -3.46688062e-01 -9.07004297e-01 -4.23951209e-01
3.76963288e-01 8.63990247e-01 -1.52857140e-01 -2.73852468e-01
-3.74937952e-01 -1.78759061e-02 -8.23486209e-01 5.42492390e-01
8.77788901e-01 4.68880981e-01 4.04773623e-01 1.09957707e+00
-1.05025746e-01 5.86377978e-01 -6.58880115e-01 -5.91741443e-01
3.92186403e-01 -1.18975282e+00 -6.40427411e-01 2.80844897e-01
1.62719622e-01 -3.85184959e-02 -2.15247318e-01 9.75674510e-01
5.66016734e-01 -2.43610248e-01 5.91836154e-01 -6.41627192e-01
-8.15876901e-01 9.43123162e-01 -8.64539146e-02 -1.44325448e-02
3.31091553e-01 -9.87089425e-02 1.04761994e+00 -1.24234116e+00
7.43023336e-01 1.29064429e+00 3.66167992e-01 5.84948242e-01
-1.70298421e+00 -4.23432022e-01 -3.53184760e-01 -2.87622869e-01
-1.40112340e+00 -1.26851296e+00 4.58891183e-01 -3.70867044e-01
1.52853680e+00 3.01806718e-01 1.58017159e-01 1.45661330e+00
1.79632053e-01 5.90632021e-01 1.06317222e+00 -9.91315365e-01
2.70206600e-01 4.31507379e-01 -4.31992263e-01 3.68783742e-01
-2.54803449e-01 3.63594055e-01 -6.84814572e-01 1.87850315e-02
4.35284644e-01 -7.53191948e-01 9.77580715e-03 1.19668588e-01
-1.64675653e+00 9.02134597e-01 -1.41211674e-01 4.03761804e-01
-3.76510412e-01 5.87615632e-02 4.89884377e-01 1.07452095e+00
8.58813584e-01 1.64225370e-01 -5.37355483e-01 -2.08562002e-01
-1.28162408e+00 -3.08916360e-01 8.84995937e-01 1.25005054e+00
6.90655053e-01 5.44215858e-01 1.12764649e-01 1.02150476e+00
3.02609503e-01 1.10257602e+00 8.84495080e-01 -6.61050975e-01
8.05334151e-01 -1.91176623e-01 -1.17017351e-01 -2.65572071e-01
2.71934450e-01 -4.26730543e-01 -5.02233803e-01 -3.12391341e-01
-8.22332427e-02 -4.10406262e-01 -9.13607419e-01 1.75953162e+00
3.16546969e-02 -1.42675057e-01 7.16418922e-01 4.86524105e-01
3.55159551e-01 1.18357956e+00 -2.57595360e-01 -5.40762961e-01
1.07980895e+00 -1.11759913e+00 -8.90500724e-01 -5.60824394e-01
7.77294457e-01 -1.15103209e+00 1.04730332e+00 -5.96181825e-02
-1.45006442e+00 -7.82910407e-01 -9.95379865e-01 5.16713969e-02
-1.73709020e-01 4.53550667e-01 -3.51985514e-01 6.75386488e-01
-1.34418666e+00 2.24789590e-01 -8.99244606e-01 -5.51338077e-01
-4.09154147e-01 2.96299398e-01 -5.51077962e-01 2.20070586e-01
-1.33001041e+00 1.15850258e+00 4.03207511e-01 -2.80466825e-01
-1.02141225e+00 -1.27765223e-01 -1.19859898e+00 -1.19729467e-01
-1.01501770e-01 -5.43841898e-01 1.59428000e+00 -1.46914065e+00
-2.01164246e+00 8.16467702e-01 -6.47040904e-01 -7.47103035e-01
2.89234787e-01 2.88600139e-02 -7.52916992e-01 -7.21015269e-03
2.99914777e-01 4.93507445e-01 1.06595814e+00 -1.10733485e+00
-6.39164448e-01 -1.55954942e-01 -6.92006111e-01 3.75794172e-01
-1.72032803e-01 5.27366757e-01 -1.58255592e-01 -8.38003457e-01
1.67456001e-01 -8.61971319e-01 1.10226870e-02 -6.28106713e-01
-2.98093945e-01 1.44712776e-01 4.64425653e-01 -1.17056298e+00
1.18324542e+00 -2.03414011e+00 5.00033677e-01 6.78936020e-02
-4.57741320e-01 2.60066390e-02 -5.16446650e-01 8.15238118e-01
-2.98384279e-01 -1.47927716e-01 -5.34596980e-01 -1.04415429e+00
7.96176419e-02 5.18126190e-01 -5.65507710e-01 3.68346423e-01
3.82929623e-01 9.70683753e-01 -8.39152038e-01 -2.94179887e-01
1.69789225e-01 4.46164906e-01 -3.58922780e-01 4.89901334e-01
-5.30879421e-04 5.20873189e-01 1.94500238e-01 5.15035272e-01
1.99189156e-01 4.93239790e-01 3.55726659e-01 2.10913971e-01
-2.21994549e-01 1.08899188e+00 -6.42430723e-01 1.79506385e+00
-8.56654048e-01 8.61585617e-01 2.23637506e-01 -1.10301125e+00
1.04152679e+00 8.02741170e-01 1.24921851e-01 -7.85346210e-01
4.45772633e-02 8.32356632e-01 5.34318872e-02 -2.77563870e-01
3.30381632e-01 -5.10830939e-01 -2.73645371e-01 5.04410326e-01
6.73754871e-01 -5.45972228e-01 2.16132645e-02 -1.49568275e-01
9.62027431e-01 1.38873056e-01 2.70860612e-01 -2.19356090e-01
3.77079666e-01 -2.81917062e-02 2.50480801e-01 3.42064172e-01
1.31231740e-01 6.54526711e-01 1.21865079e-01 3.58052738e-02
-1.42525375e+00 -1.27666831e+00 1.80056900e-01 1.15437400e+00
-5.50196350e-01 -4.47777003e-01 -1.08810925e+00 -4.94437069e-01
-5.17225325e-01 1.30308974e+00 -2.52271056e-01 -1.03425823e-01
-7.69801259e-01 -3.18447977e-01 1.05452728e+00 1.05473533e-01
-6.05758615e-02 -9.50799584e-01 1.54621392e-01 6.67157650e-01
-5.35319626e-01 -1.33781242e+00 -1.00750959e+00 5.63223362e-01
-7.22410619e-01 -2.80743420e-01 -8.42306435e-01 -1.20957720e+00
6.56946659e-01 -1.08429231e-01 1.06050229e+00 -7.54830062e-01
3.05494308e-01 2.04623848e-01 -4.21883792e-01 -1.63584813e-01
-1.50636113e+00 1.83016166e-01 4.91867214e-01 3.04251522e-01
4.97082680e-01 -4.70808446e-01 1.43144578e-01 3.61120373e-01
-7.25650132e-01 3.90586928e-02 8.41975093e-01 9.17721927e-01
6.37738407e-01 -5.63024521e-01 5.64225078e-01 -5.57575166e-01
7.61088490e-01 -3.23667705e-01 -4.75227654e-01 2.40242943e-01
-5.65665841e-01 1.75135583e-01 7.73877025e-01 -4.87329811e-01
-8.38216782e-01 3.46345127e-01 -3.96100670e-01 -3.61746311e-01
-3.74762230e-02 5.30467868e-01 -3.63411367e-01 4.47261274e-01
9.38915730e-01 8.02263439e-01 4.96245325e-02 -4.19348687e-01
6.46234572e-01 1.45406234e+00 7.41444707e-01 -3.79149437e-01
1.03654742e+00 -1.03773035e-01 -8.19468200e-01 -1.09432578e+00
-2.45943651e-01 -4.47422773e-01 -8.69017780e-01 1.71115264e-01
1.09615886e+00 -1.22667694e+00 2.75825322e-01 1.35547504e-01
-1.60079062e+00 -4.88344401e-01 -5.27686298e-01 7.72309184e-01
-7.49791503e-01 3.09696853e-01 -6.79980397e-01 -6.27538979e-01
-4.68971759e-01 -1.27039397e+00 1.37039602e+00 -6.21885955e-01
-4.92301673e-01 -9.88016486e-01 4.84777987e-01 2.79724538e-01
5.10473073e-01 -3.83978307e-01 5.73920786e-01 -9.77442682e-01
-8.69095549e-02 -2.31645241e-01 1.65487766e-01 1.03352201e+00
4.17991430e-01 -3.37732255e-01 -8.84882987e-01 -4.47291136e-01
2.35799894e-01 -1.57811493e-01 4.09233779e-01 1.05230965e-01
-9.98312607e-02 -5.51394403e-01 1.42882630e-01 5.59113860e-01
1.04223537e+00 2.94185072e-01 4.72341001e-01 -8.62317719e-03
3.96319598e-01 6.52309537e-01 1.90869570e-01 -2.29225770e-01
3.45522344e-01 7.44596183e-01 -2.99008280e-01 -2.13797525e-01
-4.07134533e-01 -6.56838834e-01 1.37810528e+00 1.91025579e+00
5.15334189e-01 -2.48671681e-01 -7.73435295e-01 8.67265999e-01
-1.46019745e+00 -6.77933037e-01 1.33261532e-01 2.29754806e+00
1.13782477e+00 2.23070964e-01 3.13819526e-03 -1.72754839e-01
7.53069878e-01 -4.18720730e-02 -7.13321716e-02 -8.11662793e-01
-1.45383596e-01 4.90318149e-01 5.29540718e-01 1.16719043e+00
-7.81977355e-01 1.35504937e+00 6.12940741e+00 6.78358078e-01
-1.23972678e+00 6.44927979e-01 5.07951140e-01 1.84853286e-01
-4.07184124e-01 1.95231199e-01 -6.47790253e-01 2.84561455e-01
1.94852412e+00 -3.00250113e-01 7.61795461e-01 5.87964714e-01
3.65209818e-01 5.29996514e-01 -1.35000992e+00 9.11115170e-01
3.07083279e-01 -1.02325618e+00 1.38482183e-01 -9.59443226e-02
6.73522592e-01 5.44751227e-01 -2.43992835e-01 3.47355574e-01
2.86881655e-01 -9.21511233e-01 1.07170010e+00 -8.80824327e-02
1.44122481e+00 -5.28828263e-01 5.51898539e-01 4.75840032e-01
-1.06701374e+00 5.42538702e-01 -4.00331527e-01 2.40883425e-01
3.70190799e-01 2.52561897e-01 -1.20181561e+00 6.61242902e-01
7.08649904e-02 5.62479854e-01 -1.92175824e-02 9.45335254e-02
-3.86082023e-01 1.11312568e+00 -3.23652059e-01 9.65523645e-02
2.39092991e-01 -4.15634453e-01 7.86978900e-01 1.77514672e+00
7.75542140e-01 -4.84437913e-01 1.71039179e-01 6.46589816e-01
-1.39768273e-01 3.95508498e-01 -7.52989829e-01 -4.24333662e-01
4.69060689e-01 6.42764449e-01 -4.41037327e-01 -6.13287628e-01
-4.00962889e-01 1.64071739e+00 -1.90614183e-02 5.11592805e-01
-4.68410730e-01 -6.26802385e-01 5.88680744e-01 1.03301987e-01
2.22983345e-01 -4.92019922e-01 -9.19817239e-02 -1.37050247e+00
7.59729967e-02 -1.21305025e+00 -3.28024209e-01 -6.60689414e-01
-9.85835969e-01 1.33525193e+00 -3.20237666e-01 -1.17770302e+00
-9.57650840e-01 -3.62639397e-01 -3.71790677e-01 1.29882908e+00
-1.24637222e+00 -1.30670130e+00 5.48017144e-01 4.55532223e-01
1.08699787e+00 -6.19646847e-01 1.21765578e+00 3.12310547e-01
-1.31242126e-01 7.49037206e-01 4.69104797e-01 3.44395876e-01
9.12100792e-01 -1.22828460e+00 1.17587590e+00 1.14527214e+00
6.24606431e-01 5.48151016e-01 8.24274898e-01 -4.89948899e-01
-1.26129413e+00 -1.23715150e+00 1.74887264e+00 -6.81990564e-01
9.77723241e-01 -1.08407605e+00 -5.50074637e-01 9.97259974e-01
7.73217499e-01 -3.93084466e-01 6.61288559e-01 -2.44080544e-01
-2.37416744e-01 -7.08882883e-02 -7.12072492e-01 6.57697082e-01
8.11763287e-01 -1.10746574e+00 -9.48830545e-01 4.62023348e-01
1.15156865e+00 -3.32859099e-01 -7.04419553e-01 -2.63408329e-02
2.43663758e-01 -3.31090152e-01 6.69857144e-01 -5.06817222e-01
3.06149334e-01 -2.48593971e-01 -7.63344646e-01 -1.83500791e+00
1.42752275e-01 -1.18892407e+00 2.66012311e-01 1.07313979e+00
1.05148828e+00 -7.07422733e-01 1.29870623e-01 -1.35571286e-01
-5.98272741e-01 -1.95593864e-01 -1.41050661e+00 -9.03548360e-01
2.42825255e-01 -4.65966225e-01 3.13559026e-01 7.89930999e-01
1.17294863e-01 9.09574628e-01 -4.51390207e-01 1.45938918e-01
4.25039023e-01 -2.75336206e-01 8.25608790e-01 -6.79107368e-01
-6.56065285e-01 -9.81465727e-02 -2.05811813e-01 -1.35308862e+00
3.69440258e-01 -1.26268971e+00 4.98013645e-01 -1.37355995e+00
-3.36535513e-01 1.61379680e-01 7.63492584e-02 4.64586407e-01
2.82007158e-01 2.50755399e-01 -9.64294747e-02 2.33070418e-01
-1.55949667e-01 6.86486244e-01 6.27466857e-01 -1.95072904e-01
-1.73487872e-01 -4.98979865e-03 -3.83454591e-01 1.67407319e-01
6.15971446e-01 -7.42967606e-01 -3.88095379e-01 -6.22053683e-01
-1.78907886e-01 3.70191842e-01 -1.82832673e-01 -7.22105384e-01
7.18503520e-02 2.11909544e-02 -8.98351595e-02 -2.15378270e-01
3.02117288e-01 -7.35197484e-01 7.68567920e-02 4.83156681e-01
-4.85226005e-01 3.62330139e-01 1.36968881e-01 2.35439748e-01
-5.29498756e-01 -1.04340248e-01 8.18823159e-01 6.03955388e-02
-8.48432258e-02 -1.70274183e-01 -7.35252202e-01 -8.85816142e-02
6.83208585e-01 -1.53869301e-01 1.36806861e-01 -6.73492849e-01
-9.07323062e-01 -2.71403849e-01 3.65856230e-01 5.85261881e-01
4.86750275e-01 -1.43208575e+00 -1.42466080e+00 8.49908292e-01
2.01403826e-01 -5.47956407e-01 -5.14411509e-01 6.96827769e-01
-4.21208709e-01 7.05347359e-01 2.95302331e-01 -6.95245087e-01
-1.30888712e+00 1.86097726e-01 2.89465040e-01 -3.24202850e-02
-2.79694825e-01 8.26309860e-01 -6.07621931e-02 -1.03034985e+00
2.82960455e-03 -3.89506608e-01 6.05513871e-01 -1.94053605e-01
2.47403741e-01 -1.60018831e-01 4.26354945e-01 -1.23454881e+00
-2.49865457e-01 2.16068178e-01 1.84016861e-02 -9.94413793e-01
1.16302931e+00 -4.47447836e-01 -9.51574221e-02 6.72318161e-01
1.47588193e+00 4.46413666e-01 -7.57393003e-01 -4.41122115e-01
1.89956263e-01 1.11609243e-01 -1.32870480e-01 -7.13812292e-01
-3.42267722e-01 9.65588093e-01 3.36081445e-01 2.57125497e-02
9.13075984e-01 -5.70681617e-02 1.02016366e+00 5.26064456e-01
3.33117634e-01 -1.34142125e+00 -2.57786512e-01 8.18608701e-01
8.78261030e-01 -1.10392118e+00 -7.07466245e-01 2.34069563e-02
-6.28374577e-01 1.03647316e+00 -1.73339710e-01 3.09506860e-02
4.44003642e-01 4.32990938e-01 6.17840648e-01 3.25378895e-01
-8.97167206e-01 -1.43853068e-01 3.77986372e-01 4.73935336e-01
7.05837250e-01 2.92601317e-01 -1.66304141e-01 2.56526887e-01
-5.89275241e-01 -3.33463907e-01 3.47600192e-01 6.64765656e-01
-4.10519809e-01 -1.68072295e+00 -3.90165985e-01 -8.30134302e-02
-3.11875880e-01 -6.53001606e-01 -6.84328079e-01 4.35629308e-01
-2.96156317e-01 1.23321331e+00 -3.76379080e-02 -5.31473339e-01
4.02812660e-01 5.34016311e-01 2.32334375e-01 -1.04122150e+00
-4.75780487e-01 6.75636590e-01 3.99716586e-01 -3.70285809e-01
-1.56677723e-01 -6.44296646e-01 -1.01592302e+00 -3.04877013e-03
-3.28436375e-01 4.83398378e-01 1.11949265e+00 9.63433802e-01
3.85332584e-01 5.27582824e-01 8.79422545e-01 -7.29881883e-01
-9.72584784e-01 -1.38292778e+00 -1.52948782e-01 1.39704555e-01
5.40723503e-01 2.03176290e-01 -4.32885498e-01 5.41822374e-01]
|
[14.535393714904785, 7.145258903503418]
|
4deb0894-1045-40e9-b99a-b84ae900638c
|
revisiting-shadow-detection-a-new-benchmark
|
1911.06998
| null |
https://arxiv.org/abs/1911.06998v3
|
https://arxiv.org/pdf/1911.06998v3.pdf
|
Revisiting Shadow Detection: A New Benchmark Dataset for Complex World
|
Shadow detection in general photos is a nontrivial problem, due to the complexity of the real world. Though recent shadow detectors have already achieved remarkable performance on various benchmark data, their performance is still limited for general real-world situations. In this work, we collected shadow images for multiple scenarios and compiled a new dataset of 10,500 shadow images, each with labeled ground-truth mask, for supporting shadow detection in the complex world. Our dataset covers a rich variety of scene categories, with diverse shadow sizes, locations, contrasts, and types. Further, we comprehensively analyze the complexity of the dataset, present a fast shadow detection network with a detail enhancement module to harvest shadow details, and demonstrate the effectiveness of our method to detect shadows in general situations.
|
['Pheng-Ann Heng', 'Chi-Wing Fu', 'Xiaowei Hu', 'Tianyu Wang', 'Qiong Wang', 'Yitong Jiang']
|
2019-11-16
| null | null | null | null |
['shadow-detection']
|
['computer-vision']
|
[ 6.46842360e-01 -3.62799048e-01 2.44624034e-01 -5.66014290e-01
-4.63049896e-02 -5.61156929e-01 2.74439991e-01 -5.19851029e-01
1.01677917e-01 1.00754154e+00 2.11009800e-01 -5.76752365e-01
4.99020070e-01 -5.86546361e-01 -3.01585436e-01 -9.94713962e-01
-1.94710582e-01 3.34608644e-01 1.15597689e+00 -2.53421247e-01
6.77822530e-02 6.58977687e-01 -1.40941584e+00 9.63159278e-03
1.01791763e+00 8.87602150e-01 7.68009484e-01 8.54769945e-01
2.09974289e-01 6.23615563e-01 -1.09068215e+00 1.08775022e-02
2.49479398e-01 -3.45083594e-01 -4.11705673e-02 1.99566707e-01
3.78673464e-01 -6.60661221e-01 -6.97390735e-01 5.37535310e-01
7.42054701e-01 -8.58107582e-02 4.07344639e-01 -1.48249280e+00
-2.31053054e-01 -1.07605837e-03 -4.27837878e-01 3.02236229e-01
4.97104555e-01 9.14970487e-02 6.09025598e-01 -9.42192256e-01
3.99639875e-01 1.17326498e+00 6.46439910e-01 -3.87798250e-02
-7.25204289e-01 -9.08940494e-01 3.26582730e-01 3.25710624e-01
-1.09557199e+00 -3.60616922e-01 7.16268837e-01 2.88969129e-01
4.16990519e-01 5.08086383e-01 9.20043945e-01 1.35210514e+00
4.20578688e-01 9.95787084e-01 1.79351485e+00 -6.77345395e-02
4.76955660e-02 9.13854539e-02 3.87626588e-02 8.86301160e-01
5.22246420e-01 1.97517827e-01 -5.44327259e-01 -1.86299190e-01
4.66665566e-01 2.12139383e-01 -7.77793467e-01 -5.28951228e-01
-1.10768282e+00 4.24761623e-01 6.20724261e-01 -2.47109219e-01
9.02772322e-02 9.97901894e-03 3.54841584e-03 -2.39277542e-01
4.41671573e-02 -8.72816145e-02 -3.23630482e-01 3.79410565e-01
-7.67957866e-01 6.11531921e-02 1.04534173e+00 1.32815564e+00
7.49962032e-01 4.40844372e-02 -4.32828456e-01 5.22959888e-01
5.86435162e-02 1.50705004e+00 -1.19726934e-01 -7.50451326e-01
2.86299109e-01 4.22139317e-01 1.91568196e-01 -1.29050410e+00
-5.81120431e-01 -3.09047401e-01 -9.86527681e-01 1.69121236e-01
1.30451456e-01 -1.48201302e-01 -1.10245216e+00 1.24577153e+00
3.79941285e-01 3.41391683e-01 1.36414424e-01 8.86560321e-01
1.09777498e+00 6.33453250e-01 -4.27468240e-01 -3.27376395e-01
1.17181659e+00 -9.65514004e-01 -8.51213276e-01 -6.46045804e-01
-1.71367541e-01 -9.49428678e-01 1.35093939e+00 1.27323240e-01
-1.07026726e-01 -2.71198332e-01 -1.14858794e+00 1.89078927e-01
-5.37047148e-01 2.48463273e-01 1.10366380e+00 6.60454929e-01
-6.26636028e-01 -5.67272827e-02 -2.60401011e-01 -5.23969352e-01
5.97985446e-01 -1.16391838e-01 1.97167322e-01 -4.02562052e-01
-1.01030743e+00 6.32322431e-01 9.43325534e-02 3.62333357e-01
-1.28498316e+00 -3.49529058e-01 -5.59529126e-01 -1.69460531e-02
1.07063258e+00 -2.86514372e-01 9.36468005e-01 -3.90619457e-01
-1.10337853e+00 3.01113993e-01 -3.55217963e-01 -2.54714340e-01
5.71219921e-01 -3.46753359e-01 -6.55077696e-01 -1.76863536e-01
1.20496072e-01 5.72111830e-03 6.83560252e-01 -2.13202667e+00
-6.44702017e-01 7.64921447e-03 2.88529396e-01 4.72574979e-01
5.21531813e-02 -1.51946455e-01 -6.49565339e-01 -5.44258595e-01
9.35727134e-02 -1.23829043e+00 -3.45486522e-01 9.28808674e-02
-8.88504624e-01 5.29425800e-01 1.47733426e+00 -3.26675773e-01
1.22624266e+00 -2.08182454e+00 -6.69339955e-01 2.44122282e-01
2.02909112e-01 1.68634221e-01 3.19492996e-01 5.52215517e-01
6.16815269e-01 -4.82046038e-01 -7.70152271e-01 -2.44436711e-01
-3.65627073e-02 7.55653560e-01 -8.43198359e-01 4.63469863e-01
-4.64817494e-01 7.74746835e-01 -8.77866626e-01 -9.09498394e-01
3.36401850e-01 2.42556691e-01 1.15813144e-01 5.43239772e-01
-1.26666114e-01 -3.42715867e-02 -4.80109870e-01 1.30599129e+00
1.11972845e+00 -1.96188599e-01 3.68551344e-01 -1.96199223e-01
1.77337939e-03 -1.52486756e-01 -1.25887620e+00 9.80674148e-01
-3.40482354e-01 1.33368862e+00 1.95351288e-01 -7.79112950e-02
6.08228505e-01 -2.45444492e-01 2.17686847e-01 -5.61214864e-01
-5.07108085e-02 1.37202203e-01 -6.88355789e-02 -3.62265527e-01
6.47730231e-01 1.72892451e-01 -1.23182073e-01 4.74572122e-01
-7.70455301e-01 -3.59801114e-01 -1.87839791e-01 4.26035166e-01
1.39874232e+00 -5.24588563e-02 3.79144192e-01 -3.03586036e-01
1.28953189e-01 1.48737252e-01 5.65805674e-01 9.51220274e-01
-3.60748112e-01 8.47112477e-01 9.10502076e-02 -2.09014431e-01
-3.10946852e-01 -1.34487689e+00 -2.17806965e-01 1.21932578e+00
1.04913735e+00 -1.88699186e-01 -3.37681681e-01 -7.90722013e-01
2.89282233e-01 4.32184160e-01 -5.66401064e-01 1.57729104e-01
-5.95040381e-01 -1.20351732e+00 5.92632115e-01 5.31360865e-01
1.08188820e+00 -1.25981522e+00 -1.26419461e+00 -2.82591313e-01
-3.85773361e-01 -1.53014338e+00 -2.90202707e-01 3.46111685e-01
-2.28246972e-01 -1.41366136e+00 -4.32151049e-01 -5.14321387e-01
5.54799259e-01 1.43181992e+00 1.41793740e+00 4.92055833e-01
-7.70651639e-01 2.03856587e-01 -4.11999404e-01 -9.56641734e-01
1.20370112e-01 -4.99022335e-01 -4.19346914e-02 -1.43006310e-01
-4.21204627e-01 -4.93419051e-01 -9.57896888e-01 8.29814851e-01
-6.88535929e-01 3.59394908e-01 1.00837934e+00 6.94327235e-01
2.27922469e-01 2.55158573e-01 -2.14735106e-01 -1.11199486e+00
3.71536613e-01 -2.48139352e-01 -6.20112896e-01 4.48655844e-01
-6.76347613e-01 -5.51408291e-01 4.47452635e-01 -4.84046154e-02
-1.54123724e+00 3.49105112e-02 4.85087037e-01 7.93801621e-02
-2.32245043e-01 -2.46070012e-01 -5.82514524e-01 -3.35234821e-01
4.97818798e-01 3.15305978e-01 -7.51245856e-01 -5.43636642e-02
2.39362314e-01 5.29291272e-01 7.19122529e-01 -3.78270745e-01
1.37692142e+00 1.13711405e+00 1.70264840e-01 -1.04157114e+00
-1.15917194e+00 -4.81727362e-01 -5.98684490e-01 -4.41112608e-01
2.65355796e-01 -7.59760022e-01 -4.44138885e-01 7.63096273e-01
-8.73858273e-01 -9.86312687e-01 1.29009023e-01 -1.04938209e-01
-4.47925627e-02 6.18003249e-01 -2.98975259e-02 -9.25769985e-01
-1.22075580e-01 -8.70531499e-01 1.27928472e+00 3.89757723e-01
5.49371421e-01 -8.03850591e-01 -1.81025509e-02 9.48892608e-02
3.77508640e-01 4.97787386e-01 4.99222815e-01 1.99803993e-01
-1.17481327e+00 1.42586932e-01 -8.80079746e-01 4.27528061e-02
3.54065657e-01 -2.80182734e-02 -1.24980259e+00 -1.40863419e-01
-3.63830656e-01 -9.22313407e-02 1.08091283e+00 3.06868613e-01
1.39462233e+00 -1.27973603e-02 -8.70325804e-01 7.75621712e-01
1.56504345e+00 6.42489791e-02 6.32702887e-01 1.27818193e-02
8.94576252e-01 2.91786969e-01 1.19343817e+00 5.85257828e-01
4.03230667e-01 4.14392471e-01 5.80384672e-01 -7.92719960e-01
-5.34673333e-01 9.58430469e-02 2.32056573e-01 2.03571334e-01
-1.76664487e-01 -8.78454983e-01 -8.83377850e-01 1.83932975e-01
-1.62677848e+00 -8.54536831e-01 -3.84285897e-01 1.68374395e+00
2.62866080e-01 4.51874346e-01 -3.34949940e-01 -2.56507304e-02
4.34646308e-01 7.81511784e-01 -8.47454786e-01 4.40440774e-01
-7.48660803e-01 -9.43867937e-02 1.02573478e+00 4.53733355e-01
-9.14453506e-01 1.27366400e+00 7.38492155e+00 6.47153080e-01
-8.13786030e-01 -2.00247288e-01 2.26152793e-01 3.91590834e-01
-3.39372128e-01 3.64586025e-01 -6.61193252e-01 4.19820368e-01
1.49854183e-01 2.13633433e-01 3.63517702e-01 8.28748107e-01
8.84779915e-02 -1.05307305e+00 -3.80473554e-01 9.04415131e-01
3.41216177e-01 -1.14622509e+00 -3.38668764e-01 7.36397952e-02
7.68824756e-01 1.28595635e-01 -1.40970439e-01 2.01749206e-01
5.90274096e-01 -9.19680655e-01 4.48463231e-01 2.70261407e-01
9.06958520e-01 -2.10644275e-01 8.11106205e-01 2.72361726e-01
-1.55417311e+00 -2.55997717e-01 -4.65894043e-01 -1.96979925e-01
2.98060983e-01 9.43004429e-01 -1.16777766e+00 5.56356609e-01
1.00898778e+00 4.77619171e-01 -8.62767160e-01 9.27687883e-01
-4.08106208e-01 6.55633211e-01 -4.64979768e-01 -1.95913687e-01
-7.89262503e-02 -1.74776375e-01 3.00536484e-01 1.61684859e+00
-1.31912548e-02 4.39954460e-01 5.44632435e-01 1.21927813e-01
8.05670619e-02 -4.02784139e-01 -7.27906168e-01 3.76325816e-01
7.23493993e-01 1.51169467e+00 -1.24382961e+00 -4.92827564e-01
-1.77394763e-01 1.20296729e+00 -1.50411814e-01 5.91924071e-01
-1.38891757e+00 -4.30465311e-01 6.72569394e-01 4.12965938e-02
-1.46360293e-01 -4.42845106e-01 -3.66464883e-01 -9.32439089e-01
3.86180133e-02 -7.17273414e-01 3.40226516e-02 -1.18742228e+00
-7.70890355e-01 4.24819231e-01 9.75662246e-02 -1.05535650e+00
6.26513660e-01 -6.69696033e-01 -9.11951482e-01 3.34226221e-01
-1.80039573e+00 -1.19150960e+00 -1.60197806e+00 7.61913002e-01
7.00371206e-01 2.65455619e-02 6.16275012e-01 1.00231208e-01
-5.37846327e-01 1.65145040e-01 3.06208789e-01 9.57194045e-02
1.07969224e+00 -1.34285235e+00 3.96918267e-01 9.16712523e-01
-1.44340187e-01 2.46052369e-01 1.01391697e+00 -7.72599518e-01
-1.57362449e+00 -9.77669895e-01 1.59975126e-01 -4.53944355e-01
4.37581360e-01 -8.58191013e-01 -5.96704900e-01 3.45689207e-01
1.31874889e-01 9.41982269e-02 4.15891796e-01 -3.07766423e-02
-8.11756626e-02 -3.20758283e-01 -9.58324909e-01 7.95430899e-01
1.71463335e+00 -3.07939649e-01 -8.55326951e-02 7.68795192e-01
7.06542373e-01 -8.36986065e-01 -2.91135311e-02 7.44409144e-01
8.95860791e-01 -1.69162869e+00 1.05679166e+00 1.92651391e-01
2.79023461e-02 -5.37698746e-01 -6.07565522e-01 -1.08462942e+00
-1.24097675e-01 -4.75424558e-01 -3.15182626e-01 1.05993068e+00
2.17020884e-02 -7.05546200e-01 8.93576145e-01 -1.52151182e-01
-4.95959729e-01 -9.83673573e-01 -5.65074921e-01 -6.12041831e-01
-1.00132644e+00 -3.71808290e-01 7.41366565e-01 2.44215712e-01
-7.86655545e-01 2.12808967e-01 -6.13312960e-01 6.90541625e-01
7.67498493e-01 1.07390559e+00 1.41316938e+00 -1.15328109e+00
-1.29623756e-01 4.09346521e-02 -2.67301165e-02 -1.23608017e+00
9.46962237e-02 -8.93641934e-02 7.56491840e-01 -1.93478918e+00
6.47990048e-01 -9.03445423e-01 -2.56923903e-02 3.05233747e-01
-4.72366035e-01 4.11388785e-01 7.43956789e-02 3.55990797e-01
-7.37103999e-01 6.06767833e-01 1.40454614e+00 3.43082957e-02
9.48350206e-02 3.12888682e-01 -3.47691923e-01 9.26455438e-01
6.90182865e-01 -2.20450833e-01 -6.44198239e-01 -2.28928164e-01
-3.30958158e-01 -1.79171667e-01 6.23418391e-01 -1.32615483e+00
-7.76973367e-02 -8.36664200e-01 7.76568651e-01 -1.17596996e+00
8.48682225e-01 -9.65400517e-01 -4.80978936e-02 5.84865868e-01
4.62576658e-01 -2.59676993e-01 -1.34071270e-02 9.28418219e-01
2.58222371e-01 5.28349102e-01 7.52629936e-01 -2.50134133e-02
-1.20438409e+00 4.08895493e-01 -1.55842274e-01 1.03216410e-01
1.11795926e+00 -4.34558421e-01 -6.80412352e-01 -5.70638776e-01
3.55960317e-02 3.00873369e-01 7.93430209e-01 1.05479620e-01
8.63432705e-01 -8.78723741e-01 -2.83019602e-01 1.46044776e-01
2.33432218e-01 5.19115739e-02 1.28873348e-01 5.78948677e-01
-5.89272916e-01 1.66125774e-01 -1.61252245e-01 -6.15360260e-01
-1.62456751e+00 2.73890436e-01 2.34482914e-01 -4.69108112e-02
-8.86308193e-01 7.05924869e-01 8.44222128e-01 -2.27860600e-01
4.38533276e-01 -4.47333276e-01 3.24484706e-01 -3.52284044e-01
1.66712359e-01 4.79058921e-01 -2.30392173e-01 -5.45356512e-01
-5.08087754e-01 4.18704122e-01 7.05784261e-01 1.96066424e-01
8.31568837e-01 -2.96580613e-01 1.11630403e-01 4.37201560e-01
5.33072233e-01 4.57312346e-01 -1.48891366e+00 -1.80990547e-01
-3.99288744e-01 -9.52361465e-01 -4.15094137e-01 -9.28429544e-01
-8.72779429e-01 5.75751960e-01 6.48220956e-01 1.15642324e-01
1.39108157e+00 1.29517734e-01 1.07667744e+00 8.24564159e-01
7.15008855e-01 -1.12283361e+00 3.22633684e-01 7.50462949e-01
8.36036682e-01 -1.61152279e+00 6.84669137e-01 -1.00521696e+00
-7.35293508e-01 7.75456250e-01 6.23823643e-01 6.60129040e-02
7.57682323e-01 7.25262403e-01 2.70075589e-01 -2.68144637e-01
-2.62464315e-01 -5.00599384e-01 -7.06182495e-02 9.30353165e-01
-2.23637074e-01 5.26503623e-01 1.85889646e-01 1.56695142e-01
-3.67913991e-01 -5.13755858e-01 6.45163894e-01 1.14674842e+00
-9.75480735e-01 -6.20231926e-01 -7.75563717e-01 5.78596592e-01
2.70204544e-01 -6.88857138e-02 -8.60660136e-01 9.90214527e-01
2.55855858e-01 1.12600183e+00 -5.31974912e-01 -3.31719488e-01
3.32896173e-01 -6.90748334e-01 2.80635774e-01 -4.97513771e-01
1.98961958e-01 -1.64537281e-01 2.30753392e-01 -7.64572561e-01
-2.79675633e-01 -5.31865597e-01 -1.35824645e+00 -2.97689825e-01
-3.89537334e-01 -1.07263379e-01 5.30880272e-01 9.25451577e-01
2.32189540e-02 7.01801896e-01 7.22769320e-01 -1.21862912e+00
4.79607843e-02 -7.89209485e-01 -8.43392193e-01 3.98021415e-02
5.58033884e-01 -1.27361965e+00 -3.77174199e-01 -1.24840006e-01]
|
[10.83560848236084, -4.096983432769775]
|
ac25ddb5-6092-4ffd-9a2f-00246c699245
|
deepfakeart-challenge-a-benchmark-dataset-for
|
2306.01272
| null |
https://arxiv.org/abs/2306.01272v2
|
https://arxiv.org/pdf/2306.01272v2.pdf
|
DeepfakeArt Challenge: A Benchmark Dataset for Generative AI Art Forgery and Data Poisoning Detection
|
The tremendous recent advances in generative artificial intelligence techniques have led to significant successes and promise in a wide range of different applications ranging from conversational agents and textual content generation to voice and visual synthesis. Amid the rise in generative AI and its increasing widespread adoption, there has been significant growing concern over the use of generative AI for malicious purposes. In the realm of visual content synthesis using generative AI, key areas of significant concern has been image forgery (e.g., generation of images containing or derived from copyright content), and data poisoning (i.e., generation of adversarially contaminated images). Motivated to address these key concerns to encourage responsible generative AI, we introduce the DeepfakeArt Challenge, a large-scale challenge benchmark dataset designed specifically to aid in the building of machine learning algorithms for generative AI art forgery and data poisoning detection. Comprising of over 32,000 records across a variety of generative forgery and data poisoning techniques, each entry consists of a pair of images that are either forgeries / adversarially contaminated or not. Each of the generated images in the DeepfakeArt Challenge benchmark dataset has been quality checked in a comprehensive manner. The DeepfakeArt Challenge is a core part of GenAI4Good, a global open source initiative for accelerating machine learning for promoting responsible creation and deployment of generative AI for good.
|
['Dayou Mao', 'Alexander Wong', 'Carol Xu', 'Hossein Aboutalebi']
|
2023-06-02
| null | null | null | null |
['data-poisoning']
|
['adversarial']
|
[ 5.0488478e-01 9.7303346e-02 4.9850482e-01 9.5219128e-02
-1.0225378e+00 -9.2484432e-01 1.1398598e+00 -3.7586734e-01
6.6898842e-03 6.1221170e-01 4.8318598e-01 -2.2392406e-01
3.2491454e-01 -8.1724501e-01 -8.9983076e-01 -8.8344026e-01
1.8993768e-01 5.3038502e-01 -2.2056603e-01 -1.6357957e-01
1.8379526e-01 5.5231255e-01 -1.5263126e+00 5.0800788e-01
4.9806693e-01 8.2801574e-01 -3.3338514e-01 9.5451051e-01
2.8382158e-01 9.3063533e-01 -1.2740608e+00 -1.1826601e+00
2.5623187e-01 -8.2305610e-01 -8.0957836e-01 4.5727071e-01
6.2286723e-01 -3.5331649e-01 -5.5416101e-01 1.0230197e+00
8.8737601e-01 -1.4908201e-01 6.0202450e-01 -1.7485578e+00
-1.1637673e+00 5.0898749e-01 -3.7188953e-01 2.6965860e-01
1.8490028e-01 9.4743508e-01 8.5500085e-01 -9.0729254e-01
8.9063895e-01 1.5445404e+00 4.9722180e-01 7.4234116e-01
-1.2749563e+00 -7.0325744e-01 -5.7332718e-01 1.4449334e-01
-1.0829097e+00 -7.5358719e-01 6.0189831e-01 -4.6458021e-01
8.3849031e-01 3.9508927e-01 5.5358356e-01 1.7204218e+00
2.5288370e-01 1.0559322e+00 8.9248592e-01 -3.4625325e-02
2.8964126e-01 -1.6886394e-02 -6.1695397e-01 5.6431407e-01
8.9697219e-02 3.2452896e-01 -4.4971386e-01 -3.3622786e-01
5.1062649e-01 -2.5853422e-01 -1.9689760e-01 -3.4941204e-02
-1.0773298e+00 1.0759119e+00 3.8394794e-01 -4.3772295e-02
-3.9757419e-01 3.9851478e-01 4.6715140e-01 1.4119306e-01
4.5773444e-01 8.3475137e-01 2.7988833e-01 -1.8958057e-01
-6.8300903e-01 7.6460302e-01 6.5490758e-01 7.9431230e-01
3.6044291e-01 6.0960847e-01 -3.6403313e-01 7.2534800e-01
3.6460605e-02 6.6356152e-01 3.3571738e-01 -8.1916231e-01
3.3158892e-01 5.6377226e-01 -2.3032548e-02 -1.1226546e+00
2.8361782e-01 -2.5969380e-01 -9.2026162e-01 3.5703665e-01
2.2419657e-01 -1.6729370e-01 -1.1701616e+00 1.4306848e+00
2.0135696e-01 -4.8980698e-02 2.2272198e-01 8.8094074e-01
1.0165588e+00 7.6655728e-01 4.6934448e-02 1.7310826e-01
1.1881746e+00 -8.5634911e-01 -5.2709019e-01 -4.0914130e-01
2.1061181e-01 -9.7933352e-01 9.5508474e-01 6.7068332e-01
-1.0918397e+00 -1.6746196e-01 -9.3493861e-01 -5.6272060e-02
-5.3285903e-01 -4.3863085e-01 4.2205864e-01 8.6987495e-01
-9.1462582e-01 3.4027132e-01 -2.0912485e-01 -1.5576975e-01
1.0159024e+00 -3.5288714e-02 -4.1692883e-01 -4.7612169e-01
-1.0468960e+00 7.5644904e-01 2.4470289e-01 -6.9327675e-02
-1.3635461e+00 -7.0696425e-01 -5.5167168e-01 -3.4090701e-01
2.9723945e-01 -6.3275814e-01 1.0833750e+00 -1.0768673e+00
-7.6630104e-01 1.1035746e+00 4.3338630e-01 -6.6767049e-01
9.8378593e-01 -2.3618658e-01 -6.0402089e-01 -3.7030790e-02
1.1817164e-01 8.3967209e-01 1.3546852e+00 -1.3377956e+00
-3.7606001e-01 -2.7161878e-01 -2.9888836e-01 -2.0246583e-03
-1.9710557e-01 1.1395710e-01 -2.8201047e-01 -1.0933352e+00
-6.0931814e-01 -9.5965171e-01 1.6694942e-01 -1.5550543e-01
-9.1340524e-01 -1.8263564e-01 1.2670039e+00 -7.0553154e-01
7.3061550e-01 -2.1214468e+00 1.1981678e-01 -1.8843822e-01
3.9648598e-01 5.8616966e-01 -2.7683678e-01 5.2469641e-01
-1.3715057e-01 6.3092238e-01 -2.8145733e-01 -3.7775332e-01
6.3886411e-02 1.1952096e-01 -7.9865098e-01 4.0516073e-01
5.3293473e-01 1.1842670e+00 -9.0497810e-01 -1.0181072e-01
1.0734436e-01 6.8876463e-01 -4.2053574e-01 3.2246235e-01
-4.9218476e-01 1.8368393e-01 -1.4524899e-01 8.5373467e-01
3.7177831e-01 -3.2856863e-02 -5.2516896e-01 -1.4524558e-01
1.7794423e-01 -3.3215534e-02 -6.9147992e-01 1.2703253e+00
1.7861033e-02 7.9894841e-01 -1.8952636e-01 -3.3838344e-01
7.2173011e-01 3.6616826e-01 8.1990033e-02 -7.1421897e-01
2.1911786e-01 -4.1228067e-03 -8.8507056e-02 -3.4440115e-01
7.1609789e-01 -1.3891311e-02 -1.5844455e-01 8.7568945e-01
-4.8778825e-02 -5.3740948e-01 2.3940569e-01 5.7535100e-01
1.6210115e+00 1.6713809e-02 -1.1946905e-01 1.9589730e-01
4.0074468e-02 3.2025775e-01 1.5255238e-01 7.1577495e-01
-1.0103399e-01 9.9497813e-01 3.2844484e-01 -6.6484153e-01
-1.4832047e+00 -1.1508994e+00 2.3087029e-01 8.4222740e-01
-2.1924147e-01 -2.5315151e-01 -9.5260441e-01 -6.3711262e-01
7.6436453e-02 9.2984527e-01 -7.5832075e-01 -5.9402794e-01
-4.3374071e-01 -8.7846404e-01 1.1894298e+00 1.5849584e-01
7.0286369e-01 -1.8087487e+00 -4.4647864e-01 1.7338212e-01
-2.8031871e-01 -9.3792337e-01 -5.1983225e-01 -2.1375220e-01
-2.1071336e-01 -1.1565959e+00 -7.1684575e-01 -3.8060588e-01
3.7819719e-01 2.1596862e-01 1.4659001e+00 2.3987566e-01
-9.4889486e-01 5.0668901e-01 -4.4480014e-01 -6.4987481e-01
-1.2726047e+00 -2.9288870e-01 -3.0581200e-01 5.3696811e-02
-8.9276627e-02 -2.0622808e-01 -6.0597342e-01 2.0429911e-01
-1.3877103e+00 -3.0387197e-02 4.4383577e-01 6.2853718e-01
3.0696115e-01 1.5953491e-02 6.3788164e-01 -9.7651458e-01
1.0265977e+00 -7.3816717e-01 -2.7086207e-01 1.9040737e-01
-4.1366231e-01 -1.9592182e-01 6.0501361e-01 -4.4831055e-01
-8.2957685e-01 -3.2678607e-01 -1.4630058e-01 -7.0568979e-01
-2.7074805e-01 1.0484244e-01 -3.2671657e-01 1.7113127e-01
1.0025505e+00 4.2588297e-01 -1.2389590e-02 -1.5920594e-01
7.6857597e-01 6.1231804e-01 1.0680350e+00 -2.6450339e-01
9.9673647e-01 2.2933102e-01 -1.7425032e-01 -8.5393524e-01
-4.2944232e-01 1.7328633e-01 1.8725793e-01 -3.8860598e-01
7.8328818e-01 -6.3249433e-01 -3.8234511e-01 1.0681641e+00
-1.0898316e+00 -1.7936413e-01 -3.7065247e-01 -5.0226355e-01
-4.9135995e-01 1.4178854e-01 -5.4252702e-01 -7.1354300e-01
-7.2336870e-01 -1.2292997e+00 9.3795764e-01 2.7320376e-02
-4.1826758e-01 -6.2552053e-01 2.6350481e-02 7.7747333e-01
4.0630373e-01 7.3209786e-01 1.0197468e+00 -7.1994370e-01
-7.0431411e-01 -4.5636111e-01 -1.8549395e-01 4.3727627e-01
7.0881560e-02 2.2966318e-01 -1.0656457e+00 -1.9658697e-01
-2.1115121e-01 -8.4388906e-01 6.9740075e-01 -2.3208775e-01
8.8546121e-01 -7.2206670e-01 5.2298371e-02 4.2510673e-01
1.1289525e+00 4.0514651e-01 1.2126039e+00 1.9543734e-01
8.5062265e-01 3.2502347e-01 1.5670674e-01 5.5043292e-01
-2.4391230e-02 3.7783191e-01 8.4359807e-01 -1.3926032e-01
-5.7125890e-01 -4.1328463e-01 1.8109131e-01 8.8510655e-02
1.8833175e-01 -9.7039288e-01 -8.8014847e-01 7.0912701e-01
-1.5541490e+00 -1.3571383e+00 -6.7481376e-02 1.8913854e+00
7.9906058e-01 -5.6480821e-03 2.9286617e-01 1.9052608e-01
7.0632219e-01 2.0049487e-01 -7.8153467e-01 -5.6900126e-01
-4.6093386e-01 2.1373443e-01 1.8664135e-01 -6.6160552e-02
-1.1234235e+00 9.9368519e-01 6.2061563e+00 9.2444819e-01
-9.1822028e-01 -3.3818614e-02 1.0328696e+00 -1.0258255e-01
-4.4344217e-01 -4.6481383e-01 -4.5144635e-01 6.9625682e-01
7.1634716e-01 -1.9185922e-01 8.1292689e-01 7.5840336e-01
1.7696528e-03 -2.4748178e-02 -9.2733365e-01 9.6052152e-01
4.4159359e-01 -1.7395549e+00 2.9318029e-01 2.0575295e-01
9.8198920e-01 1.7455909e-02 5.2769077e-01 3.6960579e-02
8.5897136e-01 -1.3274311e+00 9.7964215e-01 1.8155563e-01
8.0967480e-01 -9.6657926e-01 4.7798064e-01 9.9360190e-02
-4.5342544e-01 1.0338433e-01 -1.2008268e-01 3.3960351e-01
6.8173677e-02 5.1325333e-01 -9.9852180e-01 -4.8586298e-02
6.6171640e-01 3.1804660e-01 -6.5275228e-01 8.4556288e-01
-2.5642082e-01 6.7516983e-01 5.5210173e-02 2.8327176e-01
2.6049000e-01 1.0879182e-01 8.6162376e-01 9.7447550e-01
9.2810243e-02 -1.9312876e-01 -3.4581167e-01 1.3105941e+00
-5.8277041e-01 -4.1261649e-01 -8.3093923e-01 -8.0003679e-01
4.3379828e-01 1.2240484e+00 -6.3570130e-01 -1.6626577e-01
-8.9515731e-02 1.2286501e+00 -3.8348675e-02 2.4359339e-01
-9.3451488e-01 -8.7868750e-02 7.3112202e-01 1.2742826e-01
2.2828732e-01 2.1185890e-01 -2.6948798e-01 -6.9459069e-01
-2.2875822e-01 -1.5419501e+00 2.6654845e-01 -1.0549667e+00
-1.4761653e+00 9.6459854e-01 -2.7824801e-01 -7.6025498e-01
-5.8358300e-01 -7.3327996e-02 -6.9095862e-01 8.9916044e-01
-7.0785880e-01 -1.3735528e+00 -3.6427286e-01 6.8508863e-01
7.1341580e-01 -5.6498885e-01 7.6616722e-01 5.7409540e-02
-5.8653033e-01 5.0510305e-01 -1.2893498e-01 2.2980899e-01
6.4422059e-01 -9.0512037e-01 1.1513897e+00 1.2484666e+00
3.4320888e-01 1.8738207e-01 8.4059304e-01 -9.2984116e-01
-1.5699862e+00 -1.4497739e+00 4.7947836e-01 -7.2775370e-01
5.2800393e-01 -3.7259963e-01 -7.4212646e-01 5.9561092e-01
4.3326494e-01 -2.0818996e-01 2.4906150e-01 -7.3173052e-01
-5.8015627e-01 3.1533268e-01 -1.4330045e+00 7.5198555e-01
9.2752582e-01 -4.3455291e-01 -3.7267303e-01 4.4809088e-01
6.5414959e-01 -2.8656289e-01 -4.4583464e-01 1.4398998e-01
1.9740111e-01 -9.7440755e-01 1.0366585e+00 -5.7072890e-01
1.0702276e+00 -1.4116366e-01 1.6969904e-02 -1.3676233e+00
-2.8612208e-01 -1.0372787e+00 -1.3145287e-01 1.4893939e+00
9.7478203e-02 -2.8272620e-01 8.3419555e-01 7.9825699e-01
-2.2820558e-01 -4.0628168e-01 -6.2861961e-01 -5.3968000e-01
3.0591138e-02 -1.9074094e-01 6.5932709e-01 8.1210184e-01
-7.5942409e-01 2.7578366e-01 -7.1974379e-01 -2.6906738e-01
7.9101360e-01 -7.2219633e-02 1.1520660e+00 -7.4460226e-01
-2.1764325e-01 -5.3912228e-01 -5.3740817e-01 -3.2151562e-01
-2.9875085e-01 -7.0315516e-01 1.5010443e-01 -1.3969275e+00
2.0146638e-01 -1.5976588e-01 3.8472864e-01 6.8541002e-01
-3.0018985e-01 9.5369941e-01 3.3699152e-01 3.8106877e-01
-2.7851528e-01 4.0541893e-01 1.1596574e+00 -3.5485384e-01
2.6222980e-01 -9.0355821e-02 -8.9451772e-01 4.2520493e-01
6.2162447e-01 -5.3460354e-01 -3.7877885e-01 -4.2530918e-01
3.5064861e-01 -3.1351748e-01 8.9051014e-01 -9.2205459e-01
-2.3197317e-01 -1.2588258e-02 5.1259565e-01 -4.8123455e-01
3.2409844e-01 -4.7380072e-01 6.2428451e-01 4.4024587e-01
-2.3356204e-01 2.8223091e-01 2.6329565e-01 5.3197503e-01
1.9267147e-05 3.1582005e-02 8.3517081e-01 -3.4245080e-01
-7.6490402e-01 2.7240351e-01 -3.8902774e-01 5.2618206e-01
1.0300978e+00 -2.7418688e-01 -8.6891288e-01 -5.8559781e-01
-3.9200017e-01 -1.8861137e-01 5.7626963e-01 8.3442187e-01
7.8820032e-01 -1.3104336e+00 -9.4623905e-01 1.8686958e-01
1.6498539e-01 -7.9185873e-02 1.8014500e-01 1.4646432e-01
-5.0886232e-01 -1.0186126e-01 -2.2827055e-01 -2.2010344e-01
-1.3060150e+00 7.5303280e-01 1.5501021e-01 7.1915695e-03
-7.2892612e-01 1.0529218e+00 2.3657337e-01 2.2086198e-02
1.2716578e-01 4.0714034e-01 4.2154813e-01 -1.8928082e-01
7.7123427e-01 6.1172187e-01 2.1784319e-01 -8.6785799e-01
-1.1437195e-01 -3.1257790e-01 -1.1691562e-01 -1.1882746e-01
1.3293205e+00 2.6752949e-01 2.2409638e-02 -2.9764870e-02
1.0369599e+00 -7.6103158e-02 -1.3214847e+00 5.1348846e-02
-4.3169278e-01 -4.2661211e-01 -3.6794122e-02 -1.3960167e+00
-1.3196825e+00 7.2298646e-01 4.6643227e-01 4.0450230e-01
9.8270279e-01 1.2711304e-01 9.7688508e-01 -6.7373790e-02
1.7190292e-01 -5.6877679e-01 5.3809106e-01 1.7935587e-01
1.4745058e+00 -1.0587938e+00 -2.4205787e-02 -1.0959272e-01
-9.8859996e-01 6.7120755e-01 5.0636160e-01 -1.2772512e-01
-1.1763111e-01 3.5789055e-01 8.7831914e-02 -3.5787082e-01
-7.4299580e-01 1.3639988e-01 2.4793877e-01 7.7562571e-01
1.2994613e-01 2.3542253e-02 4.1413462e-01 2.7272624e-01
-4.9970451e-01 -1.6730198e-01 4.9018216e-01 7.6208633e-01
-1.3489094e-01 -9.5944840e-01 -6.2357986e-01 4.6347389e-01
-5.0879705e-01 -2.9995331e-01 -1.1343832e+00 7.3511392e-01
1.7500396e-01 1.1402403e+00 9.0104364e-02 -4.1652682e-01
1.3082707e-01 1.1362078e-02 4.2916933e-01 -2.9417998e-01
-1.0327622e+00 6.8500568e-03 2.0431037e-01 -5.3118181e-01
1.0348036e-02 -5.4079396e-01 -8.2288414e-01 -8.5165107e-01
-3.0209761e-02 -3.1203988e-01 7.1569717e-01 7.8677005e-01
5.1409042e-01 3.5050836e-01 4.7529805e-01 -6.0556018e-01
-3.5925710e-01 -6.7584890e-01 -2.5976452e-01 8.4694028e-01
7.2078206e-02 -1.9420101e-01 -3.6190537e-01 2.9481980e-01]
|
[12.443496704101562, 1.0989665985107422]
|
184705b0-ed15-49c7-a9e0-3a0c043e1320
|
hierarchical-multi-scale-attention-networks
|
1708.07590
| null |
http://arxiv.org/abs/1708.07590v2
|
http://arxiv.org/pdf/1708.07590v2.pdf
|
Hierarchical Multi-scale Attention Networks for Action Recognition
|
Recurrent Neural Networks (RNNs) have been widely used in natural language
processing and computer vision. Among them, the Hierarchical Multi-scale RNN
(HM-RNN), a kind of multi-scale hierarchical RNN proposed recently, can learn
the hierarchical temporal structure from data automatically. In this paper, we
extend the work to solve the computer vision task of action recognition.
However, in sequence-to-sequence models like RNN, it is normally very hard to
discover the relationships between inputs and outputs given static inputs. As a
solution, attention mechanism could be applied to extract the relevant
information from input thus facilitating the modeling of input-output
relationships. Based on these considerations, we propose a novel attention
network, namely Hierarchical Multi-scale Attention Network (HM-AN), by
combining the HM-RNN and the attention mechanism and apply it to action
recognition. A newly proposed gradient estimation method for stochastic
neurons, namely Gumbel-softmax, is exploited to implement the temporal boundary
detectors and the stochastic hard attention mechanism. To amealiate the
negative effect of sensitive temperature of the Gumbel-softmax, an adaptive
temperature training method is applied to better the system performance. The
experimental results demonstrate the improved effect of HM-AN over LSTM with
attention on the vision task. Through visualization of what have been learnt by
the networks, it can be observed that both the attention regions of images and
the hierarchical temporal structure can be captured by HM-AN.
|
['Bai-Ling Zhang', 'Shi-Yang Yan', 'Wenjin Lu', 'Jeremy S. Smith']
|
2017-08-25
| null | null | null | null |
['hard-attention']
|
['methodology']
|
[ 3.89392883e-01 -1.72918320e-01 4.11935635e-02 -4.80648503e-03
-2.70534873e-01 9.03092604e-03 3.80128860e-01 -4.88377094e-01
-4.92584318e-01 6.69368386e-01 2.43164271e-01 -6.29443908e-03
1.30749550e-02 -4.73317355e-01 -5.65295756e-01 -1.09557164e+00
1.83966547e-01 -1.32386684e-01 5.38689017e-01 -1.12732932e-01
4.04328406e-01 4.82682407e-01 -1.73692536e+00 4.87498969e-01
8.58069420e-01 1.07883275e+00 6.83793724e-01 8.07156324e-01
-1.90815404e-01 1.20128798e+00 -6.09033346e-01 3.69395167e-01
-1.37839988e-01 -7.53057778e-01 -6.68778896e-01 6.94721565e-02
-2.55741358e-01 -1.49711564e-01 -3.55908453e-01 8.64741981e-01
5.13533294e-01 4.02540028e-01 6.26999259e-01 -6.84961259e-01
-6.52936041e-01 3.61615002e-01 -7.64470398e-01 5.37152588e-01
1.20320823e-02 1.90285116e-01 6.76800489e-01 -7.70869851e-01
2.67708182e-01 1.40960133e+00 3.52020293e-01 6.39445484e-01
-8.17389965e-01 -4.69023019e-01 4.05403674e-01 7.71908760e-01
-1.20220780e+00 -7.85032213e-02 8.06070924e-01 -4.58269984e-01
1.10565591e+00 1.11374669e-01 6.23316646e-01 1.21628714e+00
7.35977054e-01 1.01220870e+00 1.25416410e+00 -4.39362496e-01
3.58760618e-02 -1.33872658e-01 1.15378998e-01 5.31778336e-01
-3.87147546e-01 1.12632141e-01 -3.91496003e-01 2.27364406e-01
1.07190943e+00 2.96083212e-01 -1.86608627e-01 2.53889441e-01
-7.55574882e-01 5.03273070e-01 7.37213731e-01 7.02475846e-01
-5.25474131e-01 2.28832886e-01 5.69749355e-01 -8.78726877e-03
2.75649399e-01 5.77497371e-02 -3.41058165e-01 3.10530216e-02
-6.56578720e-01 -4.48264092e-01 9.13854986e-02 5.17906904e-01
5.36018610e-01 4.40622926e-01 -3.68436635e-01 7.95411527e-01
3.31743032e-01 4.67997819e-01 1.17643666e+00 -6.58971310e-01
1.33898556e-01 7.30782986e-01 -1.07340932e-01 -8.03494155e-01
-5.68515062e-01 -2.32390642e-01 -1.34722364e+00 2.30061546e-01
1.37154087e-01 -1.06904872e-01 -1.19515228e+00 1.48188543e+00
1.27648070e-01 3.60294640e-01 2.57316142e-01 9.65033829e-01
7.47784019e-01 1.03598738e+00 2.94229239e-02 -4.98815954e-01
1.32194209e+00 -1.08512306e+00 -9.80435371e-01 -2.66739041e-01
2.38242134e-01 -4.33275163e-01 8.98977578e-01 2.94692099e-01
-7.84003079e-01 -9.87137556e-01 -9.07705188e-01 1.30051319e-02
-4.01722163e-01 3.10797781e-01 3.16793144e-01 -3.25765498e-02
-9.60908830e-01 7.56863594e-01 -1.02633500e+00 -4.47965384e-01
2.41356120e-02 3.53266209e-01 6.54064566e-02 4.05776411e-01
-1.55438769e+00 9.03269291e-01 6.45537198e-01 8.26622188e-01
-8.63230467e-01 8.88177082e-02 -5.62863052e-01 3.33210789e-02
3.46643329e-01 -3.90098333e-01 1.14330363e+00 -1.34852707e+00
-1.91720605e+00 2.25440845e-01 -2.18778208e-01 -4.93817389e-01
9.04158205e-02 -2.70821184e-01 -2.29519397e-01 3.17661881e-01
-2.63688147e-01 5.15658081e-01 9.06377673e-01 -8.30745339e-01
-5.89947402e-01 -4.39412713e-01 -2.50874490e-01 2.89316475e-01
-3.46714467e-01 2.50615746e-01 -2.84301311e-01 -7.12627530e-01
9.06640440e-02 -9.06098962e-01 -4.06600028e-01 -5.50141931e-01
-2.01265395e-01 -4.43898976e-01 1.03738213e+00 -8.50673318e-01
1.37691784e+00 -2.04521632e+00 3.90765905e-01 -8.83581787e-02
-3.04078877e-01 4.86252248e-01 4.13560420e-02 2.92789638e-01
-1.36112705e-01 -7.94621184e-02 -2.22854868e-01 1.41067967e-01
-4.83303159e-01 3.74516696e-01 -2.32705370e-01 1.78405866e-01
1.92366809e-01 9.55546737e-01 -6.20820701e-01 -6.03361070e-01
3.87748152e-01 5.32114685e-01 7.62103312e-03 3.85008752e-01
-1.85860187e-01 6.83416247e-01 -4.74975497e-01 3.17609042e-01
1.74483538e-01 -3.36127222e-01 5.33030219e-02 -9.19721350e-02
-2.74234265e-01 -1.46221191e-01 -9.24164414e-01 1.34980917e+00
-4.56694365e-01 6.91480935e-01 2.93761268e-02 -1.11830628e+00
1.16700780e+00 6.10470951e-01 4.90476787e-01 -8.50219905e-01
2.68941879e-01 -2.48574354e-02 1.39818013e-01 -8.99801075e-01
2.73215532e-01 -1.52547881e-01 1.61770090e-01 1.66513488e-01
-1.52777478e-01 1.91359267e-01 -6.08074851e-02 -2.28504822e-01
7.20652103e-01 3.63853097e-01 2.48310164e-01 -1.02698922e-01
9.02565062e-01 -4.59769696e-01 7.12609291e-01 5.63996673e-01
-1.97439238e-01 3.82895142e-01 2.07492501e-01 -5.13938546e-01
-1.03504419e+00 -4.22584713e-01 1.55058131e-01 1.21935260e+00
4.77944389e-02 -2.22801026e-02 -6.38073146e-01 -3.72577280e-01
-4.38564181e-01 3.31378430e-01 -7.50338972e-01 -3.18323731e-01
-7.09494650e-01 -7.70219862e-01 3.86777520e-01 8.13848317e-01
8.42928410e-01 -1.82789791e+00 -1.00082517e+00 3.78089219e-01
-1.35999039e-01 -9.76497293e-01 -3.61521930e-01 1.74469009e-01
-1.02622795e+00 -8.02572548e-01 -8.97079170e-01 -7.62454748e-01
3.44792038e-01 -9.55687761e-02 3.68800789e-01 -1.05464935e-01
-2.75724411e-01 1.43686563e-01 -4.96465296e-01 -2.72156030e-01
-2.14221731e-01 3.36355641e-02 1.98810194e-02 3.74452293e-01
2.47300014e-01 -7.37506866e-01 -5.65836668e-01 3.85685295e-01
-9.26885843e-01 1.56701311e-01 9.76428568e-01 9.07914639e-01
4.33195472e-01 3.51711959e-02 7.15145588e-01 -4.13124353e-01
4.92703736e-01 -1.12704784e-01 -6.15786672e-01 3.67952347e-01
-3.12466055e-01 4.15573210e-01 1.01992631e+00 -6.29556537e-01
-1.29149711e+00 1.76616058e-01 -1.83347523e-01 -7.28729546e-01
-2.01244414e-01 5.88674188e-01 -6.18228409e-03 1.88188955e-01
3.06299806e-01 7.34340250e-01 -1.38709977e-01 -3.49454582e-01
1.59869537e-01 8.75284791e-01 3.57853264e-01 -2.67377198e-01
1.58364296e-01 3.11176300e-01 -5.89358695e-02 -1.19494081e+00
-7.45933473e-01 -4.52724338e-01 -7.90201247e-01 -4.49985117e-01
1.41965723e+00 -5.70877969e-01 -9.87560093e-01 9.09267366e-01
-1.51437271e+00 -3.87295514e-01 -3.50666977e-02 6.45405769e-01
-6.58889234e-01 4.09384131e-01 -1.00148642e+00 -1.30173981e+00
-5.33556938e-01 -1.06469214e+00 8.39519858e-01 6.08255982e-01
1.27625629e-01 -9.71598804e-01 -1.58314899e-01 1.73494935e-01
3.53358954e-01 1.15199469e-01 8.20244014e-01 -3.38863075e-01
-6.50804341e-01 2.27431208e-01 -8.79948884e-02 4.68479961e-01
4.26046737e-02 2.23601729e-01 -9.71403897e-01 -1.26802206e-01
4.50324029e-01 -3.25092554e-01 1.14785087e+00 7.86223531e-01
1.25770760e+00 -3.35704237e-01 -1.97945639e-01 2.69417375e-01
1.27543664e+00 8.29124987e-01 9.47830498e-01 3.39795172e-01
6.77824497e-01 5.84553361e-01 5.25663793e-01 4.75022852e-01
-1.91944912e-02 5.55160463e-01 3.85152370e-01 -2.00491637e-01
1.93355069e-01 -6.44427687e-02 6.81425691e-01 1.20095026e+00
-5.17497003e-01 9.64301522e-04 -6.90318763e-01 3.76591980e-01
-2.23242116e+00 -1.05586195e+00 -2.57618260e-02 1.92481613e+00
7.30329037e-01 3.33173007e-01 -8.64912849e-03 -2.00019628e-02
1.08230758e+00 2.65343517e-01 -9.20907676e-01 -6.39304221e-01
-1.35063648e-01 -3.76467593e-02 2.77054191e-01 4.48843718e-01
-9.28701282e-01 8.82395387e-01 5.66430998e+00 8.92472446e-01
-1.42422855e+00 -1.49093404e-01 6.27287686e-01 1.79342777e-01
3.09054703e-01 -2.87450612e-01 -8.66782665e-01 4.84461427e-01
1.10388088e+00 8.81202593e-02 2.87110716e-01 5.32485187e-01
6.76338255e-01 -1.73542708e-01 -6.79742932e-01 9.36283708e-01
-4.65248339e-02 -8.41329575e-01 -1.69596672e-02 -1.58682913e-01
5.03834963e-01 -1.04684629e-01 -3.51264291e-02 2.92717904e-01
2.40673557e-01 -9.39551055e-01 3.48868400e-01 9.63623226e-01
4.81162578e-01 -7.26376295e-01 9.50210035e-01 5.74494600e-01
-1.50920951e+00 -5.21533787e-01 -4.44135934e-01 -1.63834184e-01
6.76467866e-02 3.30916584e-01 -5.61396480e-01 5.96687675e-01
8.49631727e-01 8.64267468e-01 -4.20018405e-01 8.40446293e-01
-4.58968312e-01 5.74417114e-01 -1.65376827e-01 -2.84265697e-01
4.82262731e-01 -3.09626937e-01 3.75224978e-01 1.13910043e+00
3.06905389e-01 3.64631653e-01 1.75006643e-01 7.77114332e-01
3.22730094e-01 1.04921028e-01 -6.78205609e-01 -7.16878548e-02
1.09331021e-02 1.27233315e+00 -9.15054440e-01 -4.53482091e-01
-2.27855399e-01 9.94729996e-01 2.34113351e-01 6.16900682e-01
-9.31749821e-01 -4.92592007e-01 1.31843790e-01 -3.01053017e-01
5.68539917e-01 -2.58928418e-01 -9.45994332e-02 -8.94530416e-01
-3.41205448e-02 -6.31742120e-01 4.92233127e-01 -1.16294563e+00
-9.45576727e-01 7.32764840e-01 -2.23354161e-01 -1.07331669e+00
-2.27836713e-01 -4.32042629e-01 -6.10274076e-01 8.85669768e-01
-1.23267782e+00 -8.97737861e-01 3.02995611e-02 6.10486329e-01
1.01296163e+00 1.60584562e-02 5.50019324e-01 -1.61222797e-02
-9.00411904e-01 2.09649816e-01 2.75463700e-01 1.38121411e-01
4.05197352e-01 -9.73822355e-01 -1.97006632e-02 8.94701838e-01
-1.83848038e-01 4.42692965e-01 5.74001253e-01 -6.48140430e-01
-9.09014404e-01 -1.03745794e+00 4.52602148e-01 -3.59879099e-02
7.40994036e-01 -2.70500094e-01 -1.17192483e+00 5.96312523e-01
5.18010914e-01 -2.13188380e-01 1.94606230e-01 -2.58285731e-01
7.61090219e-02 -2.25127071e-01 -6.48195565e-01 5.38325489e-01
6.26813471e-01 -6.69036388e-01 -7.73236156e-01 2.19433621e-01
8.53638411e-01 -2.03691542e-01 -8.37086201e-01 4.87512767e-01
5.60305715e-01 -9.77463484e-01 6.07938826e-01 -5.37128866e-01
4.83152956e-01 -3.72269869e-01 1.51799053e-01 -1.23125768e+00
-4.74552333e-01 -5.05279243e-01 -1.61187261e-01 1.03069901e+00
2.38135934e-01 -6.98715150e-01 5.43287635e-01 3.60403836e-01
-2.33202934e-01 -9.19598699e-01 -9.02953088e-01 -8.26730669e-01
-1.60170779e-01 1.58401281e-02 5.22799976e-02 4.08707440e-01
-5.70012219e-02 7.38234282e-01 -5.36109805e-01 2.19006553e-01
1.55009329e-01 1.04904264e-01 1.79832309e-01 -1.05126178e+00
-2.93818325e-01 -4.65865463e-01 -3.21188390e-01 -1.25330794e+00
1.83302388e-01 -3.41083407e-01 3.30650777e-01 -1.62597764e+00
1.50818720e-01 2.04125583e-01 -7.54902661e-01 4.75946009e-01
-2.89162785e-01 -1.28685534e-01 1.71676710e-01 3.76452297e-01
-6.20256126e-01 8.21454108e-01 1.49145424e+00 -1.30473569e-01
-4.59797442e-01 1.27458632e-01 -1.16794139e-01 7.54581511e-01
7.64895141e-01 -2.70989895e-01 -3.83660316e-01 -1.03009373e-01
-1.56718999e-01 3.90075058e-01 1.58983275e-01 -1.04921806e+00
5.05904436e-01 -2.55224138e-01 4.13945854e-01 -7.83724487e-01
3.75550479e-01 -8.71509910e-01 -8.43660012e-02 6.75977349e-01
-3.75921994e-01 1.64374128e-01 2.75604069e-01 4.97718096e-01
-4.07531232e-01 -1.74732700e-01 8.38179648e-01 -4.22323704e-01
-8.19080651e-01 1.06804334e-01 -8.86289895e-01 -2.57481992e-01
1.03462172e+00 -2.95033842e-01 -1.70469582e-01 -3.12653661e-01
-9.32249367e-01 2.84436822e-01 -1.03496820e-01 3.38370919e-01
8.34476233e-01 -1.34087145e+00 -4.33572352e-01 2.14315042e-01
-3.09661597e-01 -3.56406495e-02 6.25700355e-01 9.79356587e-01
-1.23270690e-01 6.88857853e-01 -4.34414029e-01 -6.73282981e-01
-1.31792414e+00 8.38503003e-01 5.43743491e-01 -4.86097246e-01
-5.18618107e-01 6.72307253e-01 3.84927571e-01 -5.92625625e-02
1.89716503e-01 -5.81307769e-01 -7.85702646e-01 4.99730445e-02
5.97589910e-01 3.26698661e-01 -2.08552480e-01 -5.87705076e-01
-2.98837900e-01 8.32193553e-01 -1.58863455e-01 -7.04160184e-02
1.33280492e+00 -2.38676801e-01 -3.05833340e-01 9.72531497e-01
8.74545634e-01 -6.10549450e-01 -1.49806178e+00 -1.33081838e-01
1.18968427e-01 2.20048070e-01 -1.45983800e-01 -5.10492265e-01
-9.78445172e-01 1.22943544e+00 7.31634855e-01 3.75020266e-01
1.41668832e+00 -2.83164918e-01 8.10253143e-01 3.58700871e-01
-1.11286528e-02 -1.40133429e+00 3.85343909e-01 8.08973432e-01
1.00251329e+00 -1.02466142e+00 -3.12705547e-01 8.54202732e-02
-6.25189245e-01 1.42179322e+00 9.14995432e-01 -7.52150789e-02
4.24436808e-01 1.12808384e-01 6.57459199e-02 -3.52767808e-03
-9.18795168e-01 -3.58667463e-01 1.81835771e-01 2.70520508e-01
3.30826372e-01 -2.85383493e-01 -1.82115287e-01 4.58315134e-01
3.20878983e-01 2.33769670e-01 3.93550783e-01 8.38455200e-01
-7.37478316e-01 -7.00995743e-01 -4.36555654e-01 2.80834168e-01
-3.52774978e-01 -6.52395561e-02 -3.47763211e-01 4.72447634e-01
-3.69638018e-02 8.58353138e-01 2.63763275e-02 -5.51261663e-01
2.44899720e-01 1.47181138e-01 2.86342889e-01 -3.89556438e-01
-5.52700281e-01 2.83432037e-01 -3.52305830e-01 -4.08164978e-01
-6.48429692e-01 -3.79630387e-01 -1.58369458e+00 2.97668189e-01
-4.53386664e-01 2.32533887e-01 3.86305392e-01 1.26152658e+00
2.65883833e-01 9.51658010e-01 6.33900046e-01 -9.37135220e-01
-3.73475313e-01 -1.28401768e+00 -5.40564299e-01 1.69213131e-01
3.94821703e-01 -4.69817489e-01 -3.11590672e-01 4.60495427e-02]
|
[8.040557861328125, 0.6490355730056763]
|
518b47db-3132-4fc9-a013-627134e051b0
|
attention-guided-generative-models-for
|
2110.06393
| null |
https://arxiv.org/abs/2110.06393v1
|
https://arxiv.org/pdf/2110.06393v1.pdf
|
Attention-guided Generative Models for Extractive Question Answering
|
We propose a novel method for applying Transformer models to extractive question answering (QA) tasks. Recently, pretrained generative sequence-to-sequence (seq2seq) models have achieved great success in question answering. Contributing to the success of these models are internal attention mechanisms such as cross-attention. We propose a simple strategy to obtain an extractive answer span from the generative model by leveraging the decoder cross-attention patterns. Viewing cross-attention as an architectural prior, we apply joint training to further improve QA performance. Empirical results show that on open-domain question answering datasets like NaturalQuestions and TriviaQA, our method approaches state-of-the-art performance on both generative and extractive inference, all while using much fewer parameters. Furthermore, this strategy allows us to perform hallucination-free inference while conferring significant improvements to the model's ability to rerank relevant passages.
|
['Bing Xiang', 'Zhiheng Huang', 'Davis Liang', 'Peng Xu']
|
2021-10-12
| null | null | null | null |
['triviaqa']
|
['miscellaneous']
|
[ 3.53332758e-01 5.77090561e-01 3.70247304e-01 -3.40671688e-01
-1.69848764e+00 -7.13784993e-01 7.98201442e-01 -8.71142372e-02
-3.50034148e-01 7.72186637e-01 6.71210229e-01 -5.81733525e-01
1.37768745e-01 -8.75647068e-01 -9.63326871e-01 -1.16458602e-01
4.48551536e-01 8.21803868e-01 2.04314053e-01 -6.53659165e-01
2.25292251e-01 -5.38738072e-02 -1.35478973e+00 6.16815805e-01
1.29625010e+00 5.70637167e-01 2.72035033e-01 1.17891467e+00
-4.83957440e-01 1.33189034e+00 -8.75451326e-01 -9.50614810e-01
-1.65137932e-01 -9.11189198e-01 -1.45304513e+00 -2.37704828e-01
5.54181516e-01 -6.44643128e-01 -5.21553934e-01 6.71270967e-01
5.16808987e-01 2.37687200e-01 6.12312376e-01 -5.96821964e-01
-1.18621707e+00 6.59031451e-01 -5.31194918e-02 3.63343954e-01
6.68249369e-01 3.66300851e-01 1.60293925e+00 -9.51333523e-01
4.46037263e-01 1.38053715e+00 2.32205003e-01 7.16955125e-01
-1.34016192e+00 -1.94186643e-01 -5.75154535e-02 5.44096291e-01
-8.46011817e-01 -6.93021774e-01 4.99088645e-01 8.58628079e-02
1.38028896e+00 4.49967295e-01 3.04971993e-01 9.94216979e-01
9.30787474e-02 1.18033469e+00 7.26180375e-01 -4.05428320e-01
3.07748206e-02 -2.73099422e-01 -2.64708325e-02 7.96133757e-01
-3.25136244e-01 -3.44846308e-01 -4.15764987e-01 -7.26493746e-02
4.52317119e-01 -3.95800620e-01 -4.46442813e-01 1.59725487e-01
-1.01909196e+00 9.36534464e-01 5.44619977e-01 6.76559731e-02
-4.28508282e-01 2.93773293e-01 1.97731137e-01 4.64501292e-01
4.22517538e-01 9.72306132e-01 -4.46186423e-01 -2.44290531e-01
-7.81479776e-01 4.11019176e-01 1.00987494e+00 7.98483610e-01
6.99451864e-01 5.76198287e-03 -7.02617228e-01 7.91524410e-01
6.25309944e-02 6.72179818e-01 3.43099684e-01 -1.23692143e+00
6.90541267e-01 3.41822505e-01 1.30586158e-02 -5.96070647e-01
1.54058216e-02 -6.70088947e-01 -3.76621336e-01 -3.92261684e-01
3.86920452e-01 -9.73544270e-02 -1.10911632e+00 1.78690565e+00
-1.79908052e-03 -9.36449096e-02 3.25303823e-01 6.04844570e-01
8.35903287e-01 9.18876708e-01 1.06426805e-01 1.70833215e-01
1.58914924e+00 -1.14044762e+00 -7.04949617e-01 -5.24399459e-01
7.00866044e-01 -6.80701613e-01 1.43342566e+00 1.92012593e-01
-1.48320293e+00 -5.36602020e-01 -7.46594787e-01 -6.32318020e-01
9.84882638e-02 -2.51724899e-01 2.43369132e-01 3.22912216e-01
-1.09405124e+00 4.08993065e-01 -6.54664993e-01 -1.67855725e-01
4.78284061e-01 1.46780297e-01 -4.22866419e-02 -3.45860958e-01
-1.14261508e+00 1.10621607e+00 2.10588917e-01 -9.78205353e-02
-1.15357947e+00 -7.86961436e-01 -8.04797590e-01 4.71564382e-01
5.05356193e-01 -1.43479717e+00 1.75336456e+00 -6.00026011e-01
-1.69649053e+00 5.94273806e-01 -6.22138798e-01 -5.96010387e-01
1.03865482e-01 -5.86997926e-01 -2.71093190e-01 6.29320323e-01
1.27483159e-01 8.16626728e-01 7.33429849e-01 -1.03187072e+00
-2.80878305e-01 -1.97980747e-01 4.93171245e-01 2.17091337e-01
-2.53296822e-01 -1.14266284e-01 -4.60718989e-01 -3.77112329e-01
-1.06488317e-01 -6.83762252e-01 -6.00513779e-02 -5.35930216e-01
-5.04791677e-01 -5.04839361e-01 4.42166597e-01 -1.17705345e+00
1.19893157e+00 -1.65676200e+00 3.99470627e-01 -3.26839924e-01
3.12779903e-01 3.51358473e-01 -5.05602241e-01 6.62895381e-01
2.89508909e-01 1.52543619e-01 -4.89451230e-01 -4.80542451e-01
1.87575787e-01 3.51118416e-01 -7.01832414e-01 -2.72294641e-01
8.11348915e-01 1.65920424e+00 -1.01287937e+00 -4.11779195e-01
-2.53455907e-01 2.68869340e-01 -9.74030852e-01 5.86174309e-01
-8.55715215e-01 3.68902564e-01 -4.36066687e-01 2.70509660e-01
1.46301806e-01 -6.00508332e-01 7.45427459e-02 6.07204996e-02
5.32712340e-01 1.05225956e+00 -2.11702645e-01 1.87653661e+00
-8.26117158e-01 6.75499439e-01 -1.02811925e-01 -7.12222517e-01
6.00049138e-01 3.40694934e-01 -3.55252236e-01 -1.12741172e+00
-6.09092861e-02 1.25553176e-01 1.36188567e-01 -7.65411377e-01
8.05401742e-01 -2.66497374e-01 -7.14802044e-03 5.64885080e-01
4.71519798e-01 -3.38322043e-01 3.64037305e-01 6.44920349e-01
1.22081327e+00 3.31020176e-01 -1.73870334e-03 -4.24903585e-04
4.53787386e-01 3.79204974e-02 1.76014397e-02 8.46417964e-01
3.51153016e-01 6.98331475e-01 4.73239839e-01 9.81122628e-02
-1.05510712e+00 -1.32280159e+00 3.74467790e-01 1.26073432e+00
-3.54113460e-01 -5.05954087e-01 -8.57081175e-01 -8.07070732e-01
-2.17253760e-01 1.30395710e+00 -6.12358987e-01 -3.76714647e-01
-1.02459407e+00 -3.55822623e-01 8.44144046e-01 6.26123130e-01
4.66447502e-01 -1.13190269e+00 -4.14309680e-01 3.58356655e-01
-8.22456419e-01 -1.02086532e+00 -4.61949676e-01 -4.28922698e-02
-9.76526141e-01 -7.03470051e-01 -8.57971013e-01 -6.14101708e-01
4.95697439e-01 5.61710484e-02 1.82656646e+00 1.78614378e-01
1.09768875e-01 4.67937618e-01 -3.81330907e-01 -1.92090273e-01
-6.86454594e-01 5.51771402e-01 -7.22688913e-01 -1.97788715e-01
2.13643640e-01 -6.59131110e-01 -5.42342484e-01 -7.86605775e-02
-9.96500134e-01 -2.33415533e-02 7.91853249e-01 9.56816435e-01
3.65912944e-01 -8.77799690e-01 1.02058291e+00 -8.59789133e-01
1.02673137e+00 -5.46725273e-01 -2.85719484e-01 4.30317938e-01
-2.88447738e-01 6.33981407e-01 8.08066607e-01 -5.71215488e-02
-1.30841446e+00 -6.86900377e-01 -6.80001199e-01 -2.48858467e-01
-9.06679109e-02 6.01354361e-01 -1.93038106e-01 3.78010780e-01
7.27188349e-01 5.37558675e-01 3.46618257e-02 -5.52771628e-01
8.70485425e-01 3.69369835e-01 8.19226861e-01 -6.68524146e-01
7.48765469e-01 1.39005795e-01 -3.77170742e-01 -5.17861784e-01
-1.26783276e+00 -2.70105153e-01 -1.98755413e-01 1.44394934e-01
1.12193727e+00 -7.54261136e-01 -7.78048992e-01 -4.65213694e-02
-1.46559668e+00 -2.41464630e-01 -4.03701246e-01 9.42116231e-02
-5.33678114e-01 3.63828391e-01 -8.82946074e-01 -7.51356721e-01
-6.14338219e-01 -8.51809502e-01 1.00510490e+00 1.80141181e-01
-3.92877996e-01 -1.01204205e+00 2.18942180e-01 9.69514132e-01
6.58639610e-01 -2.47750536e-01 1.37562907e+00 -8.63319933e-01
-1.04692698e+00 1.87299386e-01 -1.74374968e-01 3.86995077e-01
-2.18773976e-01 -5.40311158e-01 -1.14997280e+00 -4.01098765e-02
1.09411791e-01 -6.33464873e-01 1.16557920e+00 -1.12543993e-01
1.05185270e+00 -6.78206146e-01 4.83897440e-02 3.46735567e-01
1.06577539e+00 6.51430935e-02 1.07719600e+00 4.66078259e-02
6.52460456e-01 5.25857985e-01 1.07589744e-01 -1.87455509e-02
8.45098555e-01 4.00385529e-01 4.45568532e-01 3.38256955e-02
-4.52931970e-01 -6.32968962e-01 3.70880455e-01 1.16325676e+00
2.10045055e-01 -4.78139371e-01 -8.15678000e-01 9.50186551e-01
-1.51130533e+00 -1.14793837e+00 -7.10187107e-02 1.98005557e+00
9.80909348e-01 5.07342890e-02 -1.02142021e-01 -2.42672488e-01
1.67945251e-01 2.17070609e-01 -5.07846296e-01 -5.88614404e-01
-4.76276949e-02 8.62232327e-01 -1.85951322e-01 8.07035029e-01
-5.30822396e-01 9.38594937e-01 6.63713455e+00 7.17440605e-01
-6.23139441e-01 2.15447038e-01 4.89642501e-01 -9.05229524e-02
-1.06088877e+00 1.13608919e-01 -6.35257840e-01 2.00414106e-01
1.29590464e+00 -3.21245492e-02 4.35276002e-01 3.39242518e-01
-3.16397399e-01 -6.99845841e-03 -1.14399219e+00 5.08566678e-01
4.92182255e-01 -1.43229985e+00 4.98316854e-01 -1.43155381e-01
6.21534586e-01 4.51946110e-02 8.94276574e-02 7.66452253e-01
5.09706855e-01 -1.13802087e+00 4.92313743e-01 5.58996856e-01
5.20887554e-01 -8.16616476e-01 6.19994044e-01 4.89912361e-01
-5.86435497e-01 -5.11474460e-02 -2.62897789e-01 -1.37464181e-01
6.66097403e-01 3.89256179e-01 -1.04139948e+00 7.50094831e-01
2.78382391e-01 1.22243412e-01 -6.79884493e-01 8.08057010e-01
-6.85307503e-01 9.97722924e-01 -1.17264360e-01 -2.07538038e-01
3.21877986e-01 8.96077380e-02 5.40892124e-01 1.02858019e+00
2.03515619e-01 2.09288090e-01 -2.82959908e-01 1.19731545e+00
-5.31832397e-01 -8.17300081e-02 -3.35267335e-01 -2.25835830e-01
2.66135633e-01 8.66421461e-01 5.60041703e-03 -5.72429657e-01
-3.38039368e-01 1.29445422e+00 6.70301437e-01 5.40362477e-01
-7.30712533e-01 -6.70459628e-01 3.62567604e-01 -5.28451279e-02
6.65131509e-01 -2.28297666e-01 -1.62444636e-01 -1.33477056e+00
3.24265137e-02 -1.27439523e+00 3.70766312e-01 -1.01025152e+00
-1.13055313e+00 8.11229408e-01 -1.56080544e-01 -6.17004693e-01
-7.41352499e-01 -2.68107533e-01 -5.72089970e-01 1.08509040e+00
-1.75721037e+00 -1.16755116e+00 5.10020740e-02 4.46740657e-01
7.06146896e-01 7.22459108e-02 9.70094681e-01 2.50633627e-01
-1.62154898e-01 6.36210501e-01 -1.28028080e-01 1.32957160e-01
5.96977413e-01 -1.37497985e+00 8.16901565e-01 9.65490699e-01
6.03736639e-01 8.41809630e-01 6.01472318e-01 -4.06541467e-01
-1.42446053e+00 -8.94872427e-01 1.26295412e+00 -1.02805078e+00
4.66508865e-01 -3.67790043e-01 -1.19195545e+00 8.27616453e-01
8.80685270e-01 -6.30763292e-01 6.75785184e-01 4.11598325e-01
-6.41818285e-01 1.17374964e-01 -6.98554873e-01 7.42645264e-01
9.01362121e-01 -9.81283247e-01 -1.31148052e+00 1.82215497e-01
1.16705155e+00 -3.46969068e-01 -6.86668217e-01 1.31795630e-01
2.28344589e-01 -7.95506060e-01 9.56172407e-01 -8.92306864e-01
1.00458252e+00 -1.19758025e-01 -5.51838167e-02 -1.37303531e+00
-3.76117140e-01 -8.25739205e-01 -5.30604482e-01 1.07973075e+00
7.57393897e-01 -5.64877033e-01 6.10303342e-01 2.99149901e-01
-5.21195889e-01 -8.22385371e-01 -7.91774452e-01 -6.27940118e-01
4.35555547e-01 -2.03997090e-01 6.80606604e-01 4.71568346e-01
-1.55964300e-01 1.17161369e+00 -2.45326996e-01 -1.04159713e-02
3.98311347e-01 1.48057193e-01 8.08780670e-01 -8.16633523e-01
-8.12761128e-01 -2.80947417e-01 2.02551633e-01 -1.88353074e+00
2.20739171e-01 -1.10514688e+00 1.57467619e-01 -1.98460507e+00
1.69750214e-01 1.22855872e-01 -1.72029343e-02 3.23135883e-01
-6.94576859e-01 1.79355949e-01 3.97798717e-01 4.53371145e-02
-8.23743582e-01 9.08932686e-01 1.56045818e+00 -8.43238384e-02
2.35955864e-01 -3.86438340e-01 -1.03911138e+00 2.28788391e-01
6.46933317e-01 -4.38584417e-01 -6.00145042e-01 -1.18348253e+00
4.37333316e-01 3.19554687e-01 4.69847381e-01 -8.17270398e-01
2.46406049e-01 3.14779848e-01 1.90594971e-01 -5.63289344e-01
5.47947466e-01 -1.30908221e-01 -4.62410271e-01 3.17335606e-01
-6.13182485e-01 3.25109065e-01 3.52664798e-01 5.50078332e-01
-4.33511823e-01 -2.73213446e-01 4.39433038e-01 -2.99699575e-01
-3.82924974e-01 -7.20450729e-02 -2.77246773e-01 8.56435835e-01
3.14435393e-01 1.50103390e-01 -7.40636289e-01 -9.03700531e-01
-5.63498855e-01 4.06754464e-01 1.02822371e-01 3.48939955e-01
6.69990420e-01 -1.04917192e+00 -1.12657034e+00 -6.18040375e-02
7.51318261e-02 -1.06102385e-01 3.04650217e-01 6.71077967e-01
-3.56461823e-01 8.03696930e-01 8.48834366e-02 -3.96300614e-01
-9.42694068e-01 3.51405919e-01 2.96825081e-01 -6.49775922e-01
-4.17983472e-01 1.19061399e+00 3.37754756e-01 -5.89388371e-01
-2.63102263e-01 -3.06159407e-01 -2.65565682e-02 -9.55350921e-02
5.77576339e-01 2.33442008e-01 1.98757544e-01 -9.69295725e-02
-3.50179151e-02 1.49898067e-01 -4.43539113e-01 -4.44544196e-01
1.08737016e+00 -1.14800513e-01 4.97470377e-03 4.90682907e-02
1.30356264e+00 1.35379702e-01 -1.04460514e+00 -2.26216346e-01
-4.43975776e-02 -1.93496197e-01 -1.79977149e-01 -1.14688540e+00
-5.07861912e-01 1.26473963e+00 -1.33836970e-01 2.16878295e-01
9.81611311e-01 2.51286566e-01 1.33899105e+00 6.85819864e-01
5.63071556e-02 -6.33770525e-01 5.25864422e-01 9.57113326e-01
1.07533169e+00 -1.06057119e+00 -5.43359220e-01 -9.21500027e-02
-5.83083928e-01 8.34041715e-01 5.36700785e-01 -7.96222761e-02
-1.45177558e-01 -9.93988290e-02 5.37973945e-04 -1.94034278e-01
-1.16352832e+00 -4.33020622e-01 4.97135967e-01 4.16449070e-01
4.82743889e-01 -2.22572863e-01 1.28609361e-02 4.61349815e-01
-4.30691063e-01 -1.20651118e-01 3.67001563e-01 6.55447304e-01
-5.91947734e-01 -1.12314832e+00 -8.88979584e-02 4.76910770e-01
-7.85542309e-01 -6.59347892e-01 -3.01843703e-01 5.08905351e-01
-5.75426638e-01 1.15551972e+00 9.24612880e-02 -4.63739522e-02
4.00346845e-01 5.28056145e-01 8.46102715e-01 -6.74742043e-01
-7.53927112e-01 -2.91377336e-01 6.01908803e-01 -4.21519458e-01
8.38223025e-02 -2.35188380e-01 -1.13368571e+00 -1.85667068e-01
-3.86468977e-01 4.73016530e-01 2.46020243e-01 1.22363067e+00
8.54827464e-01 9.14077997e-01 2.36367121e-01 -1.06941618e-01
-9.35116470e-01 -1.10445035e+00 1.49165124e-01 3.68101954e-01
4.57160830e-01 -1.30406143e-02 -1.92392215e-01 8.01703632e-02]
|
[11.283794403076172, 8.085051536560059]
|
06462998-0d84-4599-9a68-4977e955e313
|
using-meta-knowledge-mined-from-identifiers-1
| null | null |
https://aclanthology.org/2021.acl-long.545
|
https://aclanthology.org/2021.acl-long.545.pdf
|
Using Meta-Knowledge Mined from Identifiers to Improve Intent Recognition in Conversational Systems
|
In this paper we explore the improvement of intent recognition in conversational systems by the use of meta-knowledge embedded in intent identifiers. Developers often include such knowledge, structure as taxonomies, in the documentation of chatbots. By using neuro-symbolic algorithms to incorporate those taxonomies into embeddings of the output space, we were able to improve accuracy in intent recognition. In datasets with intents and example utterances from 200 professional chatbots, we saw decreases in the equal error rate (EER) in more than 40{\%} of the chatbots in comparison to the baseline of the same algorithm without the meta-knowledge. The meta-knowledge proved also to be effective in detecting out-of-scope utterances, improving the false acceptance rate (FAR) in two thirds of the chatbots, with decreases of 0.05 or more in FAR in almost 40{\%} of the chatbots. When considering only the well-developed workspaces with a high level use of taxonomies, FAR decreased more than 0.05 in 77{\%} of them, and more than 0.1 in 39{\%} of the chatbots.
|
['Henrique Ferreira', 'Gabriel Malfatti', 'Maira de Bayser', 'Melina Guerra', 'Mauro Pichiliani', 'Julio Nogima', 'Heloisa Candello', 'Ana Appel', 'Victor Henrique Alves Ribeiro', 'Paulo Cavalin', 'Claudio Pinhanez']
|
2021-08-01
| null | null | null |
acl-2021-5
|
['intent-recognition']
|
['natural-language-processing']
|
[ 4.91992868e-02 7.12018728e-01 3.80584180e-01 -3.47240269e-01
-3.78900379e-01 -5.64690113e-01 6.38815463e-01 7.31222238e-03
-5.53389609e-01 6.98485196e-01 2.52717167e-01 -2.82395124e-01
-2.60515362e-01 -4.28824514e-01 -5.01591086e-01 -3.85283053e-01
1.08012214e-01 4.60795492e-01 2.28706643e-01 -4.78934735e-01
4.80354190e-01 -9.20236632e-02 -1.57026875e+00 7.49872565e-01
7.47040391e-01 6.17604673e-01 3.25279862e-01 5.97385824e-01
-4.02460694e-01 1.08822584e+00 -1.25389862e+00 -5.89876413e-01
-1.07514225e-01 -3.00060093e-01 -1.03651524e+00 -3.37277949e-01
1.11120872e-01 -2.29443535e-01 -1.71402439e-01 8.25841069e-01
1.69607908e-01 3.63274738e-02 7.14155436e-01 -1.22854412e+00
-4.20861214e-01 1.17949748e+00 -7.14837313e-02 -1.27468249e-02
7.23156095e-01 7.52303973e-02 1.15334225e+00 -6.51600957e-01
7.54103005e-01 1.00061142e+00 8.60119998e-01 6.42396748e-01
-1.14932370e+00 -5.66236436e-01 -5.90466522e-02 -1.31700383e-02
-1.18517625e+00 -3.22989255e-01 3.73920321e-01 -7.30041265e-01
1.57154548e+00 2.52688080e-01 5.40256143e-01 1.18966305e+00
-1.40061658e-02 4.73696858e-01 7.83740580e-01 -7.35732079e-01
1.35990173e-01 6.41340792e-01 5.73153138e-01 4.78775859e-01
1.84237868e-01 -1.51131183e-01 -4.03218567e-01 -3.36499691e-01
6.83245420e-01 -1.73162088e-01 -8.54704157e-02 4.33141291e-01
-1.03387344e+00 8.80792975e-01 -1.16708362e-02 8.91128123e-01
-2.02277154e-01 1.74161702e-01 5.85203767e-01 4.43400979e-01
4.93419647e-01 9.82723176e-01 -4.17565525e-01 -9.80407119e-01
-3.83622825e-01 1.04473330e-01 1.38061023e+00 1.06188214e+00
6.93870783e-01 -7.10847080e-02 -6.01578923e-03 1.15644622e+00
-1.18892141e-01 -8.86436403e-02 6.82737410e-01 -1.12179554e+00
6.30563974e-01 1.15721190e+00 2.47130573e-01 -8.05181146e-01
-6.21718347e-01 -3.01089793e-01 -2.93029696e-01 9.03659388e-02
6.50805235e-01 -4.12574410e-01 -4.92468297e-01 1.57956493e+00
-5.90311959e-02 -4.43971843e-01 -4.17791121e-02 4.19393241e-01
6.13064110e-01 2.87846953e-01 -2.47052565e-01 -2.40400031e-01
1.46909440e+00 -6.03342533e-01 -7.38360167e-01 -3.06517899e-01
1.08956349e+00 -7.81494975e-01 1.17643785e+00 2.97626883e-01
-6.67497456e-01 -3.90851706e-01 -1.11017215e+00 4.67298597e-01
-3.17406178e-01 -8.69412497e-02 5.53641677e-01 9.78456855e-01
-9.71283913e-01 7.54767478e-01 -5.36370337e-01 -3.16569477e-01
1.72945112e-01 5.78366160e-01 -4.13603932e-01 4.56171185e-02
-1.04637182e+00 1.16571593e+00 3.78689438e-01 -3.96513909e-01
-4.76792872e-01 -8.66223574e-01 -7.52381504e-01 8.36888105e-02
4.93184447e-01 -2.45529115e-01 1.41770029e+00 -8.37991655e-01
-1.32235229e+00 6.99302852e-01 2.52403080e-01 -3.50284666e-01
3.98791909e-01 -3.71898264e-01 -1.55940548e-01 -2.55865365e-01
6.59240410e-02 2.68692076e-01 4.12307709e-01 -8.23388517e-01
-6.89638615e-01 -6.28720562e-04 6.70825005e-01 -8.95543471e-02
-6.58488393e-01 1.54445380e-01 -3.72287333e-02 -2.80604243e-01
-2.12205797e-01 -9.68627810e-01 1.05572194e-01 -4.53845590e-01
-2.06144661e-01 -5.27985990e-01 5.43293536e-01 -6.66184843e-01
1.25606871e+00 -2.01024866e+00 4.31504399e-02 -1.67481199e-01
5.33381283e-01 2.03981459e-01 -8.53590518e-02 8.33736241e-01
-8.51230025e-02 3.40367913e-01 1.19795986e-01 -3.52282554e-01
2.34697595e-01 2.16197118e-01 1.40296340e-01 1.67504415e-01
2.00959221e-01 4.37400311e-01 -7.89580166e-01 -1.54375046e-01
1.25670418e-01 1.33421272e-01 -5.19319236e-01 2.17185333e-01
-1.39065772e-01 -2.24835396e-01 -1.92360923e-01 1.96993545e-01
2.56264582e-02 -1.48611292e-01 2.47286543e-01 4.02358949e-01
-2.01519117e-01 8.17504883e-01 -9.40741003e-01 1.38580513e+00
-9.72682714e-01 9.87119377e-01 -2.47534573e-01 -7.23444641e-01
9.48255360e-01 7.07184434e-01 1.69701055e-01 -3.69619757e-01
4.73313369e-02 3.88656616e-01 5.56599498e-01 -5.48261166e-01
4.97816831e-01 -1.41296819e-01 -1.44058883e-01 8.26943874e-01
2.45083705e-01 -4.59497571e-02 3.33283365e-01 3.63967381e-02
1.68713796e+00 -2.54913419e-01 3.12397718e-01 -4.82351743e-02
4.26585138e-01 4.90574539e-02 2.92280912e-01 9.42220986e-01
1.23383820e-01 6.48860991e-01 9.76649165e-01 -6.11519635e-01
-9.65228736e-01 -3.06622654e-01 -6.13676980e-02 1.29474604e+00
-4.54207063e-01 -6.56234682e-01 -1.04380488e+00 -9.62293506e-01
-1.59892604e-01 9.83220160e-01 -5.97190857e-01 -3.73682141e-01
-5.63229918e-01 -5.64248979e-01 1.03524137e+00 4.98318851e-01
2.09664553e-01 -1.36377251e+00 -7.84548640e-01 4.25163388e-01
-1.81664795e-01 -1.13779092e+00 -3.04878075e-02 5.53054810e-01
-6.62684679e-01 -1.10304630e+00 -4.82625544e-01 -4.99749511e-01
3.56193721e-01 -2.82391667e-01 1.02599490e+00 5.23399651e-01
-1.55204967e-01 1.34932483e-02 -8.05588722e-01 -4.54119831e-01
-1.05795491e+00 2.20123887e-01 1.72009114e-02 -5.10141313e-01
5.23579001e-01 -6.37277722e-01 2.12004647e-01 4.68961179e-01
-4.40454692e-01 -5.92822954e-02 4.28976953e-01 1.05494368e+00
-6.05270386e-01 -2.60322630e-01 6.44572794e-01 -1.16911840e+00
8.48136663e-01 -4.05430228e-01 -4.37129170e-01 -6.58185035e-02
-6.46524668e-01 2.23383173e-01 6.30105615e-01 -7.73024619e-01
-7.90424168e-01 -3.06121647e-01 -3.08331072e-01 -1.97900370e-01
-3.82952005e-01 5.22105575e-01 3.06617051e-01 1.65397733e-01
9.71845329e-01 -4.27277572e-03 4.90291826e-02 -6.10884726e-01
-1.88606195e-02 1.07362759e+00 4.73777130e-02 -4.87614870e-01
3.63496870e-01 -1.80399314e-01 -6.01418138e-01 -9.56050158e-01
-3.40324461e-01 -5.61158299e-01 -3.29515219e-01 -2.23368600e-01
6.25327647e-01 -3.17433715e-01 -9.83184457e-01 3.49381864e-01
-1.52350152e+00 -4.70662534e-01 -2.33440265e-01 5.34057975e-01
-5.36327243e-01 2.55068421e-01 -6.06565535e-01 -1.15118349e+00
-5.23180440e-02 -1.34535122e+00 7.64176548e-01 -1.11705356e-03
-1.11726451e+00 -8.21039021e-01 -1.82979386e-02 4.40254420e-01
5.12917519e-01 -7.88963586e-02 1.19252443e+00 -1.39846539e+00
7.20802024e-02 -4.67807144e-01 -3.21869575e-03 6.97906971e-01
2.44779855e-01 -2.12710917e-01 -1.13359785e+00 1.48971021e-01
2.54242420e-01 4.09954116e-02 2.08479181e-01 -3.02599043e-01
3.61762077e-01 -5.11416376e-01 -3.03035885e-01 -1.42691255e-01
8.58918846e-01 7.69818425e-01 6.42194867e-01 3.40366602e-01
3.87561828e-01 8.40297341e-01 4.70298290e-01 5.17692029e-01
1.30251065e-01 9.32970405e-01 4.90469158e-01 5.99703312e-01
1.11741573e-01 2.95704365e-01 7.04351842e-01 8.20213914e-01
-3.20423722e-01 -2.18307137e-01 -1.20771086e+00 6.70629025e-01
-1.63350141e+00 -8.54780614e-01 -4.10224915e-01 1.77979386e+00
1.03550529e+00 3.13226730e-01 1.77960649e-01 3.61980557e-01
8.20604861e-01 -1.43368587e-01 -1.43253375e-02 -7.17831492e-01
5.15272021e-01 7.89001361e-02 3.61509398e-02 2.64610678e-01
-7.84819543e-01 5.84415257e-01 6.08863688e+00 5.27176619e-01
-6.68502927e-01 2.36836106e-01 1.97644204e-01 -3.28762494e-02
4.07991968e-02 1.07799314e-01 -6.25737429e-01 6.03446603e-01
1.29443705e+00 8.41893926e-02 5.75844646e-01 1.10902071e+00
-2.65560001e-01 -3.75571072e-01 -1.45598769e+00 8.09536278e-01
9.32561681e-02 -1.41867363e+00 -7.39237547e-01 7.64092207e-02
5.86678028e-01 -4.38032299e-02 -5.08592188e-01 6.25173986e-01
4.43367064e-01 -1.07081163e+00 6.91263497e-01 3.58437896e-01
5.50263464e-01 -4.76501465e-01 1.36548674e+00 9.16654885e-01
-5.52300513e-01 -1.22703962e-01 -2.77657241e-01 -7.44233251e-01
-1.13847658e-01 4.88052130e-01 -1.33369184e+00 4.24507499e-01
7.38596022e-01 3.76156390e-01 -3.12008262e-01 7.77274966e-01
-1.66337937e-01 6.82585359e-01 -5.83533466e-01 -7.10394382e-01
9.89506692e-02 -7.66059710e-03 4.79381114e-01 1.32653117e+00
-1.30326655e-02 -1.02004178e-01 -3.03667963e-01 9.82846797e-01
-7.08602509e-03 -2.51232266e-01 -5.90240836e-01 -3.76666784e-01
5.65216660e-01 1.05345905e+00 -4.55444366e-01 -3.00692648e-01
-3.05873573e-01 7.75133014e-01 1.69233426e-01 9.10419375e-02
-7.82862306e-01 -9.72425520e-01 7.81749308e-01 4.90414537e-02
1.46756396e-01 2.11340949e-01 -3.53201389e-01 -5.92995584e-01
2.87313104e-01 -1.06396294e+00 1.23831658e-02 -6.55641675e-01
-1.01167440e+00 8.33509386e-01 1.57648817e-01 -9.81322765e-01
-7.32994080e-01 -7.18253613e-01 -6.43263996e-01 9.62639093e-01
-4.26486999e-01 -6.58157110e-01 -3.78905177e-01 -9.23740342e-02
5.19456208e-01 -3.38381350e-01 1.08066452e+00 3.44955146e-01
-2.99094588e-01 5.94376087e-01 -2.81801045e-01 3.65695834e-01
6.09413743e-01 -1.30501819e+00 4.12826151e-01 2.80311525e-01
2.19490543e-01 1.19942832e+00 9.46536720e-01 -3.32606345e-01
-1.04697025e+00 -6.58314824e-01 1.03962970e+00 -7.25569069e-01
6.43005788e-01 -5.64955175e-01 -9.57485735e-01 9.11459148e-01
3.40343177e-01 -5.61897397e-01 4.18517321e-01 6.16348624e-01
-4.85093087e-01 3.99091363e-01 -1.17916620e+00 5.62803447e-01
1.06728876e+00 -7.48243093e-01 -1.13789320e+00 4.48560685e-01
8.14705491e-01 -4.10052478e-01 -9.60810602e-01 -5.08717634e-02
6.64265037e-01 -1.16141772e+00 6.75070643e-01 -5.81746042e-01
6.11255586e-01 -1.00459578e-03 1.32922381e-01 -1.18280005e+00
-4.46549430e-02 -4.67816710e-01 6.10061400e-02 1.48636580e+00
4.90066469e-01 -9.15265083e-01 6.53924167e-01 5.51641762e-01
-4.59350854e-01 -5.40734947e-01 -1.11801362e+00 -7.53516436e-01
-1.47279082e-02 -6.44144654e-01 3.20262223e-01 8.38544250e-01
7.39315093e-01 3.24751109e-01 -7.28580430e-02 -3.47984940e-01
-2.15078697e-01 -5.15727639e-01 6.46395922e-01 -1.20384300e+00
-5.17643452e-01 -4.18910414e-01 -7.14701295e-01 -5.70984244e-01
1.27820030e-01 -7.13564992e-01 2.31711701e-01 -1.62243330e+00
7.71320835e-02 -3.60964358e-01 1.19058669e-01 9.66710448e-01
7.52255991e-02 1.26769170e-01 1.91280544e-01 3.38358358e-02
-3.02006662e-01 -3.04647069e-02 4.69692200e-01 -5.82267977e-02
-1.54176593e-01 1.06949933e-01 -6.69381440e-01 9.94353771e-01
6.95742249e-01 -9.97806549e-01 -1.99480116e-01 -2.20362574e-01
3.43504995e-01 1.21147528e-01 1.30044624e-01 -1.11450934e+00
6.28296733e-02 1.00737683e-01 -2.32894808e-01 -2.46840432e-01
5.16908467e-01 -7.61547089e-01 2.54716069e-01 6.53424442e-01
-4.38120961e-01 1.33362813e-02 2.84054190e-01 1.42211974e-01
-6.16554469e-02 -9.55896735e-01 2.63665259e-01 -1.88460737e-01
-4.18968141e-01 -9.96606410e-01 -1.01919067e+00 1.99356824e-01
9.44061697e-01 -4.59985197e-01 -4.95198727e-01 -4.20455307e-01
-5.50220847e-01 -2.00214133e-01 4.05776232e-01 6.48895383e-01
1.57726049e-01 -6.40966177e-01 -2.81593293e-01 8.84733796e-02
1.64767474e-01 -2.93074101e-01 -4.43376824e-02 1.02604604e+00
-5.33760667e-01 4.29729909e-01 -2.56844997e-01 -3.62533391e-01
-1.40791202e+00 -5.24404198e-02 3.74693632e-01 -2.56512433e-01
-4.58205372e-01 1.02285063e+00 3.89141552e-02 -6.18110418e-01
3.45601171e-01 -6.48915052e-01 -1.91537634e-01 2.26335675e-01
4.69689220e-01 5.10809481e-01 2.72514403e-01 -2.17541263e-01
-6.21310234e-01 1.41493276e-01 -2.39255130e-01 -3.37498128e-01
1.42293823e+00 5.93575597e-01 -1.69648439e-01 8.97378862e-01
1.12670720e+00 1.85454004e-02 -7.78664649e-01 4.42162365e-01
4.47249562e-01 -1.20921835e-01 -4.99262840e-01 -9.95994627e-01
-4.13288355e-01 8.10912311e-01 1.27520129e-01 6.61575675e-01
4.26668108e-01 8.90573189e-02 2.31220126e-01 6.37699544e-01
7.55468130e-01 -9.95625257e-01 1.96782783e-01 1.02160549e+00
9.96367574e-01 -8.50149095e-01 -1.75411120e-01 -2.69505650e-01
-7.55131841e-01 1.26641321e+00 5.15893161e-01 -1.28747299e-01
1.23093024e-01 4.16869581e-01 1.45086601e-01 -3.80322129e-01
-9.58465457e-01 2.98000097e-01 -9.81738791e-02 7.01473176e-01
6.43152356e-01 -7.97362104e-02 -3.77805233e-01 1.10257828e+00
-2.95599371e-01 -2.18160495e-01 8.98589313e-01 9.73130286e-01
-3.47562551e-01 -1.07513177e+00 -3.99769813e-01 7.59216666e-01
-4.29333180e-01 -1.88368171e-01 -6.01692379e-01 1.20707560e+00
1.84348091e-01 1.14649713e+00 1.29393965e-01 -8.78444135e-01
5.01013041e-01 8.40092480e-01 8.97793472e-02 -1.16459882e+00
-1.49571157e+00 -3.78809571e-01 7.90162623e-01 -5.07208467e-01
-1.03800088e-01 -6.63870990e-01 -1.27316594e+00 -1.46965876e-01
-6.55564666e-01 2.31276467e-01 7.72033215e-01 1.26928508e+00
1.12900898e-01 7.78499782e-01 1.34164289e-01 -6.04342699e-01
-8.10813010e-01 -1.52856445e+00 -4.18043107e-01 2.13353455e-01
3.67003344e-02 -8.90663385e-01 -7.41770506e-01 -2.44325772e-02]
|
[12.52093505859375, 7.787801265716553]
|
1baaffdc-34ce-4445-8bbe-fc36d07ebf12
|
a-survey-of-software-defined-smart-grid
|
2306.14697
| null |
https://arxiv.org/abs/2306.14697v1
|
https://arxiv.org/pdf/2306.14697v1.pdf
|
A Survey of Software-Defined Smart Grid Networks: Security Threats and Defense Techniques
|
Smart grids are replacing conventional power grids due to rising electricity use, failing infrastructure, and reliability problems. Two-way communication, demand-side administration, and real-time pricing make smart grids (SGs) dependent on its communication system. Manual network administration slows down SG communication. SG networks additionally utilize hardware and software from several vendors, allowing devices to communicate. Software-defined SGs (SD-SG) use software-defined networking (SDN) to monitor and regulate SG global communication networks to address these concerns. SDN separates the data plane (routers and switches) from the control plane (routing logic) and centralizes control into the SDN controller. This helps network operators manage visibility, control, and security. These benefits have made SDN popular in SG architectural and security studies. But because SD-SGs are vulnerable to cyberattacks, there are concerns about the security of these SD-SG networks. Cybercriminals can attack software-defined communication networks, affecting the power grid. Unauthorized access can be used to intercept messages and introduce false data into system measurements, flood communication channels with fraudulent data packets, or target controllers, a potential single point of failure, to cripple SDN networks. Current research reflects this paradigm as defense and security against such attacks have developed and evolved. There is a need for a current study that provides a more detailed analysis and description of SD-SG network security dangers and countermeasures, as well as future research needs and developing threats for the sector. To fill this void, this survey is presented.
|
['Janise McNair', 'Sharon Boamah', 'Dennis Agnew']
|
2023-06-26
| null | null | null | null |
['security-studies']
|
['miscellaneous']
|
[-3.20077628e-01 -3.95708345e-02 -3.97710800e-01 -1.19302273e-01
3.64233375e-01 -1.18188989e+00 3.53648782e-01 -1.21430166e-01
3.34872395e-01 9.29935098e-01 -1.59206763e-01 -7.15021193e-01
-5.81143685e-02 -1.30018711e+00 3.05037111e-01 -9.64356244e-01
-5.06577909e-01 2.61230767e-02 2.95836210e-01 -3.31714824e-02
2.72580862e-01 9.23043787e-01 -7.40552068e-01 -3.59442651e-01
5.03532171e-01 1.09070575e+00 -3.93600553e-01 2.12279558e-01
6.04229458e-02 5.33250511e-01 -1.46104395e+00 5.35982311e-01
3.40732306e-01 -4.34159428e-01 -7.44548082e-01 1.51336446e-01
-6.31347954e-01 -6.15474403e-01 1.03394590e-01 1.33567536e+00
1.19664550e-01 -1.04736373e-01 -1.86086714e-01 -2.03323054e+00
-4.24588323e-01 6.12273037e-01 -5.05234241e-01 2.86888093e-01
2.97634304e-01 9.44720626e-01 9.04482126e-01 3.20645571e-02
4.26045984e-01 5.59937954e-01 3.52951735e-01 1.92162976e-01
-1.54542673e+00 -1.38368714e+00 4.07007523e-02 1.57982409e-01
-1.10213447e+00 -3.48782808e-01 4.59994495e-01 -3.23592648e-02
1.34788346e+00 3.33190590e-01 8.35310698e-01 6.56554103e-01
4.45385575e-01 -1.80984765e-01 1.01411593e+00 -2.12677047e-01
6.97575331e-01 5.54021671e-02 4.47231203e-01 -2.16654301e-01
1.00924313e+00 2.07507983e-01 1.61855385e-01 -2.05215916e-01
8.01591277e-01 -3.98117363e-01 -9.62924123e-01 -1.31826952e-01
-7.77996182e-01 9.88220990e-01 1.97978050e-01 9.61982965e-01
-4.45867687e-01 -1.79037899e-01 8.03715289e-01 5.70298314e-01
-1.18152224e-01 8.07302654e-01 -7.39585340e-01 -3.20587069e-01
-5.80829203e-01 -5.30310094e-01 1.36618614e+00 7.54133642e-01
2.62749821e-01 9.51909661e-01 6.30357206e-01 -1.54658034e-01
4.87094671e-01 6.54822826e-01 3.95437181e-02 -7.71091521e-01
-2.07616985e-01 1.93722188e-01 -4.37425673e-02 -1.04599118e+00
-6.96255445e-01 -4.34672177e-01 -1.03481400e+00 8.70485604e-01
2.79213011e-01 -3.75564933e-01 -1.25301272e-01 1.26214433e+00
2.15801429e-02 3.67106646e-02 2.73882803e-02 7.22288191e-01
-1.58297345e-02 7.69064844e-01 -4.26547527e-02 -4.54441786e-01
1.11371446e+00 -4.97643590e-01 -1.18993008e+00 2.92495549e-01
5.81689835e-01 -6.82035208e-01 4.83630627e-01 4.22581732e-01
-8.49185407e-01 -3.18753831e-02 -1.60056329e+00 8.77473176e-01
-4.58076745e-01 -8.03600907e-01 2.39249513e-01 1.12900293e+00
-1.15915024e+00 3.93245518e-01 -1.08994627e+00 -6.13334954e-01
2.51545340e-01 4.58720505e-01 -4.34866585e-02 5.16542554e-01
-1.28092003e+00 1.19613314e+00 2.93073475e-01 -1.43497169e-01
-7.35908210e-01 -5.97271502e-01 -8.81373167e-01 2.91720599e-01
3.52492303e-01 -3.54725718e-01 1.02593064e+00 -5.98308027e-01
-1.53227830e+00 -5.17066754e-02 7.25895345e-01 -6.92153215e-01
-2.90688872e-01 5.05651116e-01 -1.44332421e+00 4.21027005e-01
3.01569015e-01 -4.93091613e-01 4.37263668e-01 -8.80218685e-01
-6.21728718e-01 -1.44876972e-01 -5.58387078e-02 -6.34516180e-01
1.39279276e-01 3.71998340e-01 1.18950641e+00 -3.41816187e-01
2.67056935e-02 -2.58842021e-01 -2.65950173e-01 -2.69794077e-01
-5.47229588e-01 -1.28432050e-01 1.84733045e+00 -2.59464353e-01
9.69983220e-01 -2.03784013e+00 -8.34442616e-01 7.92139649e-01
4.17482585e-01 6.21802986e-01 3.93376797e-01 6.00888014e-01
-3.22717011e-01 4.38658237e-01 2.27946550e-01 4.34292436e-01
3.30630749e-01 4.26239401e-01 -3.80614072e-01 9.80234265e-01
-2.43047148e-01 4.51430976e-01 -6.07364833e-01 4.42081213e-01
7.21202374e-01 2.68443257e-01 1.65106639e-01 -1.30509824e-01
1.57053307e-01 7.49406099e-01 -3.89471084e-01 5.71287215e-01
9.80661273e-01 -5.12611091e-01 6.08806491e-01 -3.41113538e-01
-6.52978003e-01 7.16146111e-01 -1.28354549e+00 6.93432450e-01
-2.09886760e-01 4.16560799e-01 6.76013827e-01 -1.29280710e+00
6.76815331e-01 6.36261284e-01 5.46442032e-01 -6.69608533e-01
3.32698435e-01 2.05174029e-01 -3.22941504e-02 -1.09355062e-01
-3.44407618e-01 -7.52342343e-02 -3.92169468e-02 1.13447082e+00
-1.37560129e-01 -1.87723473e-01 6.98874965e-02 9.59690809e-02
1.31387687e+00 -4.58319396e-01 5.61842978e-01 -7.75854528e-01
8.25392544e-01 -1.54709443e-03 9.11603391e-01 3.29917014e-01
-5.45347273e-01 -6.84148252e-01 4.95813608e-01 -5.71949005e-01
-6.27355456e-01 -1.23923612e+00 -9.68048126e-02 1.61014974e-01
4.52651754e-02 -6.36475682e-01 -3.46741408e-01 -7.61714578e-01
-8.25455636e-02 1.59004366e+00 3.86556715e-01 -2.77378678e-01
-4.19042945e-01 -4.76743639e-01 4.57283467e-01 4.37628150e-01
7.21884549e-01 -8.35541248e-01 -7.92518675e-01 6.91335022e-01
4.05200332e-01 -1.27006066e+00 -2.75673389e-01 2.66226649e-01
-5.03644109e-01 -1.65219593e+00 4.14507955e-01 -2.90238589e-01
7.14959204e-01 2.48564735e-01 1.04198468e+00 2.58615404e-01
-1.37829810e-01 3.23088586e-01 -2.39557907e-01 -6.91503882e-02
-5.84060788e-01 -4.29434031e-01 4.89619404e-01 -2.86205500e-01
6.78981841e-01 -1.20241809e+00 -4.41219091e-01 6.10328734e-01
-5.88097632e-01 -3.94855499e-01 -1.82502389e-01 2.90500641e-01
-7.74643049e-02 8.79153371e-01 1.13592470e+00 -4.10400689e-01
5.12167692e-01 -4.32425529e-01 -1.57021677e+00 -1.80047765e-01
-1.19109070e+00 -6.22848511e-01 1.26119196e+00 2.61689246e-01
-4.58415240e-01 -7.18956113e-01 1.16334118e-01 2.39475556e-02
-5.17306745e-01 1.27082855e-01 -6.57826781e-01 -6.47739768e-01
-4.84368131e-02 3.02211624e-02 5.09055078e-01 -1.27022952e-01
-8.44037607e-02 4.45160985e-01 2.38761809e-02 -8.13028403e-03
1.27271914e+00 6.70386672e-01 -4.22708094e-02 -9.59363461e-01
-2.26123169e-01 1.62462041e-01 -1.34160087e-01 -1.07925922e-01
8.23755383e-01 -5.30291855e-01 -1.17685950e+00 4.50647682e-01
-1.06577289e+00 -9.91360247e-02 -3.25590223e-01 5.64101636e-01
1.27749309e-01 3.65692139e-01 -1.03962994e+00 -2.39564478e-01
-5.12738228e-01 -1.20939875e+00 -2.57355481e-01 4.85525280e-01
-6.57564640e-01 -1.15651095e+00 -3.72130275e-01 6.02722540e-02
1.25625944e+00 4.27749455e-01 8.03736508e-01 -8.81301999e-01
-1.03065658e+00 4.83236946e-02 -3.24211299e-01 7.49006450e-01
9.48928118e-01 2.09077492e-01 -5.27849734e-01 -9.50892866e-01
6.79903984e-01 3.02722782e-01 -5.81962466e-01 3.70661259e-01
7.46538281e-01 -6.76497102e-01 -1.92711383e-01 7.30837166e-01
1.70116377e+00 7.95790553e-01 4.68954325e-01 6.61186755e-01
2.31055617e-01 9.35776159e-02 1.96799561e-01 6.16117358e-01
2.07299173e-01 6.10222816e-02 7.49699950e-01 -2.57226855e-01
4.43951815e-01 4.18932259e-01 3.14020693e-01 7.95547426e-01
3.45930755e-01 -7.87772015e-02 -5.61645508e-01 -6.52136952e-02
-8.49774182e-01 -9.49338436e-01 -3.31174940e-01 1.88943744e+00
1.88798867e-02 6.40057564e-01 -1.63703486e-02 4.03245062e-01
7.39440322e-01 -2.73374617e-02 -6.07393682e-01 -5.04415035e-01
-1.92510948e-01 4.26440120e-01 8.74329507e-01 2.91909188e-01
-5.14719188e-01 3.13052624e-01 5.76576805e+00 3.00267011e-01
-1.30315185e+00 2.28412524e-01 2.90619463e-01 3.75193387e-01
1.63240716e-01 3.50058138e-01 -5.50682068e-01 6.12933815e-01
1.23833001e+00 -9.64439392e-01 4.85398144e-01 5.72978973e-01
1.12352479e+00 -1.58059120e-01 -7.56716251e-01 6.45745218e-01
-4.29988235e-01 -1.39825845e+00 -3.24166030e-01 3.38165760e-01
5.58556378e-01 6.54283091e-02 -7.15359688e-01 -1.19672880e-01
7.22350478e-01 -6.37598336e-01 1.98708609e-01 -2.13915035e-01
3.47705096e-01 -1.16167629e+00 7.78891265e-01 -1.92180425e-01
-1.29323673e+00 -1.97561219e-01 2.30982065e-01 -4.12128925e-01
8.25177789e-01 6.75621748e-01 -2.66098499e-01 7.95462906e-01
8.11367095e-01 6.01916552e-01 2.59784050e-02 7.01826215e-01
-7.29367495e-01 9.57645059e-01 -3.80646318e-01 3.56770843e-01
3.02437097e-01 -5.83871424e-01 7.13496745e-01 2.58749217e-01
-1.28525630e-01 1.87770322e-01 4.29458439e-01 1.25722694e+00
3.56136948e-01 -7.03155279e-01 -6.33719921e-01 -8.30638036e-02
8.04194570e-01 1.44506204e+00 -1.17482340e+00 -2.53377378e-01
-9.24084187e-01 4.34057295e-01 -8.20242584e-01 4.63473737e-01
-7.62969911e-01 -8.96604717e-01 1.38042736e+00 1.92166403e-01
-5.60794063e-02 -6.12030745e-01 -6.93215668e-01 -9.27024484e-01
-5.56088805e-01 -1.18843710e+00 4.53288078e-01 -4.58389342e-01
-1.57318532e+00 5.05295515e-01 -3.03118020e-01 -1.23665309e+00
4.86137606e-02 -4.49921414e-02 -9.66970325e-01 7.19490409e-01
-1.25395441e+00 -6.68330193e-01 1.08600065e-01 9.01461899e-01
1.87283084e-01 -1.14006475e-01 1.11471748e+00 3.15317541e-01
-5.87879002e-01 7.71828070e-02 -7.35950330e-03 4.47303653e-01
1.97989821e-01 -8.24549437e-01 3.67805749e-01 1.21662295e+00
-5.49815595e-01 3.58188808e-01 6.71767354e-01 -5.28533220e-01
-1.47375667e+00 -6.70136154e-01 3.53188753e-01 3.06691051e-01
1.26268709e+00 -3.20292860e-01 -7.15060472e-01 1.06176543e+00
9.00649011e-01 -9.94958282e-02 9.84090626e-01 -4.61453408e-01
2.63370126e-01 -2.90484518e-01 -1.66887367e+00 3.03733855e-01
1.85345888e-01 -5.39653778e-01 -4.31230515e-01 1.16557091e-01
5.32890081e-01 2.89369375e-01 -1.01062751e+00 -1.69583205e-02
-9.59405750e-02 -9.20166016e-01 4.61219043e-01 -1.93297401e-01
-9.09883976e-01 -7.92223632e-01 -3.56543399e-02 -1.65872490e+00
-3.19600940e-01 -1.11961019e+00 1.60202742e-01 1.29002881e+00
2.13422272e-02 -1.83031416e+00 4.49632943e-01 5.11723697e-01
-1.84526414e-01 6.67345747e-02 -1.03888690e+00 -1.13962400e+00
-2.70750403e-01 -3.02420288e-01 1.38279045e+00 1.70779371e+00
7.77483940e-01 3.76428515e-01 4.20545429e-01 1.06595397e+00
1.23513567e+00 9.87869874e-02 5.06301641e-01 -1.09656334e+00
4.23539013e-01 -5.30553460e-01 -6.52175546e-01 -4.44681078e-01
-3.64423916e-03 -3.90960991e-01 -8.39740336e-01 -1.43233514e+00
-9.48532462e-01 -2.77474195e-01 -1.57943383e-01 5.67348421e-01
1.02618504e+00 -1.61932290e-01 5.48796356e-01 -5.43304458e-02
-1.64107427e-01 2.02259198e-01 8.89807880e-01 -5.08617125e-02
4.24339324e-02 7.31917992e-02 -5.45961380e-01 6.91519439e-01
1.61182058e+00 -2.12715805e-01 -5.19656777e-01 -6.83373399e-03
-3.79821211e-01 3.31014693e-01 2.56158054e-01 -1.10994494e+00
5.69299281e-01 -2.11405799e-01 2.74512053e-01 -4.64836717e-01
-3.82480711e-01 -1.64696479e+00 5.21075308e-01 1.06269777e+00
6.07502401e-01 4.56056416e-01 -5.73577173e-03 -5.75063452e-02
-9.08298418e-02 3.56855690e-01 1.21383047e+00 1.71776973e-02
-3.93302023e-01 1.46725699e-01 -1.42381990e+00 -2.43932918e-01
1.64623046e+00 -3.57181579e-01 -7.61569262e-01 -4.07693505e-01
-1.03153598e+00 8.11610579e-01 5.42427540e-01 3.39241415e-01
3.81011814e-01 -8.74243438e-01 -3.40091176e-02 1.06400251e+00
-7.04416275e-01 -4.40637201e-01 -1.05761833e-01 1.00921047e+00
-6.58085048e-01 6.00089192e-01 -5.01059175e-01 -4.31643307e-01
-9.54407096e-01 3.89907449e-01 7.13391602e-01 -1.09204262e-01
-8.41757596e-01 3.18134278e-02 -2.03661829e-01 3.39127146e-02
-9.64911580e-02 -4.17139292e-01 -1.57308541e-02 -1.89247251e-01
7.92415321e-01 8.86851430e-01 -1.46858871e-01 -3.43966573e-01
-4.08595413e-01 -4.79335412e-02 1.52372364e-02 4.62108612e-01
1.15018368e+00 -6.45670533e-01 -9.48963702e-01 1.83630679e-02
8.07971835e-01 -2.13286072e-01 -5.20055652e-01 4.57812667e-01
3.05434793e-01 -6.99445069e-01 3.72637004e-01 -8.55175316e-01
-1.76053727e+00 -1.75311361e-04 2.30861679e-01 1.23848224e+00
1.18254960e+00 -3.96307915e-01 7.33590126e-01 -3.07609290e-01
1.17777920e+00 -1.02038288e+00 -2.99326539e-01 2.90095806e-01
4.45719883e-02 -3.96871656e-01 -2.02377513e-01 -7.05109715e-01
2.06026621e-02 1.43362629e+00 9.19054091e-01 -1.05466414e-02
1.29923284e+00 9.58539784e-01 8.86138380e-02 -5.27540982e-01
-5.77579319e-01 5.43325245e-01 -1.27988350e+00 1.01081991e+00
-2.93707222e-01 3.19516599e-01 -3.67827117e-01 4.82373655e-01
-3.95789888e-04 1.85054064e-01 1.32677257e+00 1.03906250e+00
-3.33843112e-01 -1.35052061e+00 -5.79555273e-01 4.87077981e-01
-7.90804565e-01 3.02772254e-01 4.45477515e-01 9.15624261e-01
-3.89259234e-02 1.67589605e+00 1.64420858e-01 -2.26050280e-02
3.48127723e-01 -3.58054847e-01 -3.65423679e-01 -3.83824348e-01
-6.22811139e-01 -5.04629463e-02 2.81472147e-01 -8.25427592e-01
-5.74471755e-03 -5.15272141e-01 -1.67238522e+00 -1.09473526e+00
-6.83592618e-01 7.42823958e-01 6.19424641e-01 7.28731155e-01
2.45186448e-01 6.42254055e-01 1.10914576e+00 -3.34529817e-01
-7.36563981e-01 -4.45118904e-01 -1.46434736e+00 -1.39182433e-01
4.82989281e-01 -4.33991402e-01 -1.18299866e+00 -5.11252642e-01]
|
[5.913158893585205, 2.607565402984619]
|
5671a576-a98c-4b3f-b7ac-45306510b5db
|
a-novel-approach-for-generating-customizable
|
2212.06701
| null |
https://arxiv.org/abs/2212.06701v1
|
https://arxiv.org/pdf/2212.06701v1.pdf
|
A Novel Approach For Generating Customizable Light Field Datasets for Machine Learning
|
To train deep learning models, which often outperform traditional approaches, large datasets of a specified medium, e.g., images, are used in numerous areas. However, for light field-specific machine learning tasks, there is a lack of such available datasets. Therefore, we create our own light field datasets, which have great potential for a variety of applications due to the abundance of information in light fields compared to singular images. Using the Unity and C# frameworks, we develop a novel approach for generating large, scalable, and reproducible light field datasets based on customizable hardware configurations to accelerate light field deep learning research.
|
['Vidhi Chhabra', 'Aloukika Patro', 'Toure Smith', 'Julia Huang']
|
2022-12-13
| null | null | null | null |
['unity']
|
['computer-vision']
|
[-8.80985856e-02 -8.77862155e-01 1.66029513e-01 -5.69753587e-01
-3.18024099e-01 -3.77484828e-01 4.17417288e-01 -3.30396116e-01
-1.88019469e-01 9.04878318e-01 -2.63699889e-01 -2.65146285e-01
-1.62923876e-02 -1.01861191e+00 -7.38565266e-01 -8.46654773e-01
5.07770240e-01 2.26019830e-01 5.01568258e-01 -7.29764923e-02
3.97440970e-01 8.27154279e-01 -1.75981426e+00 2.79016584e-01
5.29615104e-01 1.19749117e+00 4.34371173e-01 3.27958792e-01
-1.33379519e-01 7.77356803e-01 -2.71666259e-01 -1.43844396e-01
5.32602847e-01 8.23094393e-04 -5.02956450e-01 -9.90324467e-02
9.82617319e-01 -7.25948393e-01 -5.30062616e-01 1.00633657e+00
8.28350723e-01 4.67908867e-02 2.30219275e-01 -1.19359052e+00
-6.93192482e-01 1.65655926e-01 -6.60568893e-01 3.95004958e-01
2.57155895e-01 6.90706074e-01 7.04176009e-01 -8.00475538e-01
6.93274379e-01 9.45906460e-01 4.52690572e-01 4.38167959e-01
-1.16982687e+00 -9.45343375e-01 -3.78106117e-01 3.55392694e-01
-1.11673510e+00 -5.02011120e-01 1.08324623e+00 -5.50541878e-01
9.61778104e-01 -3.04414004e-01 8.26587737e-01 9.94568229e-01
5.31053960e-01 3.86131346e-01 1.27526331e+00 -4.99634027e-01
2.76754826e-01 -8.20583627e-02 -2.48260126e-02 7.50627339e-01
3.24146688e-01 1.80602998e-01 -6.14274144e-01 1.20297216e-01
1.14121366e+00 2.43733317e-01 -2.14552552e-01 -2.89486319e-01
-1.47561920e+00 5.74266374e-01 5.37284315e-01 2.19647825e-01
-2.74351984e-01 3.66422474e-01 2.02984318e-01 1.31623466e-02
3.24026018e-01 3.70888114e-01 -3.36584896e-01 4.71822992e-02
-6.73746228e-01 3.20645124e-01 4.09527153e-01 1.07278633e+00
1.28265512e+00 6.26564398e-02 -1.38558000e-01 6.79279625e-01
1.41288325e-01 7.14506984e-01 4.52101976e-02 -1.04945588e+00
4.74627130e-03 6.38574064e-01 -3.87407653e-02 -9.45515633e-01
-5.81942618e-01 -2.23780796e-01 -9.81785536e-01 6.19349539e-01
3.86710554e-01 -1.04315039e-02 -8.30196798e-01 1.53411305e+00
3.84793192e-01 4.85675901e-01 -1.47761017e-01 9.75487351e-01
1.35241425e+00 4.84995902e-01 -2.00311899e-01 5.38622700e-02
1.01789701e+00 -5.84186137e-01 -3.37903619e-01 -3.22905816e-02
3.52694422e-01 -1.07788932e+00 1.30215478e+00 5.44696629e-01
-1.15113652e+00 -5.63221693e-01 -7.82085478e-01 -5.63811243e-01
-2.08137453e-01 3.81805226e-02 1.42430854e+00 4.00350600e-01
-1.08102489e+00 3.68153960e-01 -4.63638186e-01 -2.54199654e-01
8.46941173e-01 3.21532488e-01 -1.84142381e-01 -4.28907007e-01
-6.99004948e-01 3.74185145e-01 1.77763507e-01 -1.70126200e-01
-9.11927879e-01 -1.05087328e+00 -5.10791481e-01 -2.38366023e-01
2.91542634e-02 -8.55955184e-01 1.07311559e+00 -3.41529101e-01
-1.49157870e+00 9.47459579e-01 1.54623881e-01 -5.37811741e-02
1.56404972e-01 2.85136908e-01 -3.58431250e-01 1.45340890e-01
-5.28131574e-02 6.89533472e-01 8.42948139e-01 -1.04552233e+00
-7.42770255e-01 -2.39906490e-01 4.63025630e-01 -3.12126428e-01
-2.75983751e-01 1.93365425e-01 -2.91192591e-01 -1.48539945e-01
-1.08768441e-01 -9.45968807e-01 -3.61237735e-01 4.26717371e-01
-1.58268988e-01 -2.38850310e-01 9.72202897e-01 9.86199453e-02
8.20319712e-01 -2.08586454e+00 -4.49101001e-01 -1.59463231e-02
7.00068533e-01 3.83860767e-01 2.26412266e-02 1.18175365e-01
1.38841614e-01 -3.17664236e-01 2.59825379e-01 2.25000270e-02
-2.67552018e-01 2.49597192e-01 -3.24801207e-01 5.58876336e-01
-2.17319489e-01 7.88083613e-01 -8.48057926e-01 -5.07755041e-01
7.44350731e-01 5.31404078e-01 -8.47441435e-01 1.03699952e-01
-4.32527423e-01 7.31096983e-01 -4.52568918e-01 7.70742655e-01
9.54091787e-01 -6.11885071e-01 -4.44866478e-01 -6.43089533e-01
-4.52815533e-01 -6.30933733e-04 -1.22354817e+00 1.86484516e+00
-6.21421635e-01 8.78858745e-01 4.39874604e-02 -5.96113384e-01
8.33298922e-01 -3.01935039e-02 8.38684261e-01 -7.55101085e-01
2.52848536e-01 2.39254266e-01 -8.54732841e-02 -5.10167301e-01
4.23238814e-01 -1.36280134e-01 5.58032095e-01 6.11553729e-01
2.15289712e-01 -4.19108927e-01 4.59454030e-01 1.13303311e-01
1.16184521e+00 -1.89205870e-01 -2.14045957e-01 -3.32734257e-01
2.97836125e-01 -2.15964213e-01 5.75998425e-01 6.68359578e-01
-2.99916595e-01 6.55331790e-01 -1.03987411e-01 -1.05417430e+00
-1.15945399e+00 -7.94225514e-01 -6.95700467e-01 8.86366129e-01
3.66963267e-01 -3.88361514e-01 -3.95164430e-01 -2.83908308e-01
-1.72312316e-02 1.39183104e-01 -5.94381578e-02 3.01137298e-01
-4.50515628e-01 -8.92453909e-01 2.05259249e-01 4.78299111e-01
8.09924722e-01 -1.10230410e+00 -6.92947507e-01 2.01497842e-02
1.00140765e-01 -1.57947528e+00 -1.70112178e-01 -1.44356996e-01
-5.98024666e-01 -1.00328386e+00 -3.87839854e-01 -7.21327603e-01
5.49935162e-01 8.44292819e-01 1.39113522e+00 1.56007752e-01
-7.53506958e-01 4.22754198e-01 -9.45597738e-02 -6.50359511e-01
2.35576518e-02 -2.53264934e-01 1.64845034e-01 -4.34020087e-02
5.13612509e-01 -7.12887943e-01 -1.09598994e+00 1.55927882e-01
-9.99141097e-01 3.60781401e-01 6.41251564e-01 6.97569013e-01
7.45942533e-01 1.75024793e-02 2.03474373e-01 -7.47560263e-01
3.55277807e-01 -2.87101716e-01 -1.16761911e+00 1.51777893e-01
-6.02670550e-01 -8.34865570e-02 6.95526540e-01 -4.01085168e-01
-1.16733074e+00 -3.64027470e-02 -2.08865225e-01 -3.36904049e-01
-4.07334656e-01 1.08684577e-01 3.10855452e-02 -9.61408198e-01
9.45025086e-01 -7.68817216e-02 -4.25922900e-01 -3.20708066e-01
2.40075529e-01 5.92881382e-01 5.51629245e-01 -8.62466812e-01
7.55980313e-01 7.15342224e-01 5.09478509e-01 -9.87822294e-01
-8.14178824e-01 -4.88314748e-01 -5.73509872e-01 -3.78451288e-01
7.73055553e-01 -8.87094617e-01 -9.37331438e-01 6.96626127e-01
-1.42358160e+00 -2.65688032e-01 -2.04162896e-01 6.83280885e-01
-4.45202261e-01 2.29945406e-01 -4.51127440e-01 -3.25007379e-01
-4.71314162e-01 -1.45465696e+00 1.28126347e+00 5.55394173e-01
4.21341270e-01 -9.47267234e-01 1.78842574e-01 3.86637568e-01
5.82553566e-01 2.71145795e-02 9.83873785e-01 3.20796579e-01
-1.38500893e+00 4.48356867e-02 -6.84537828e-01 3.04792136e-01
6.91243932e-02 3.12706470e-01 -1.15457582e+00 -3.38585198e-01
-2.26915225e-01 -7.16672122e-01 4.76069003e-01 5.34431636e-01
1.66841865e+00 3.13756138e-01 -4.27511543e-01 1.27436638e+00
1.66147506e+00 3.64902280e-02 9.37126040e-01 1.97520420e-01
8.89118969e-01 1.13144971e-01 -3.01230289e-02 7.17046618e-01
1.98701397e-01 6.60990238e-01 3.80535632e-01 -3.72940242e-01
-4.67623472e-01 8.97731930e-02 -1.68488964e-01 6.39763713e-01
-2.47441113e-01 -6.20677657e-02 -8.98151517e-01 2.84576118e-01
-1.32067096e+00 -1.09861791e+00 -5.09919286e-01 1.96004367e+00
8.35628390e-01 -1.22010395e-01 -3.82946968e-01 8.59054457e-03
3.75004768e-01 1.12829104e-01 -7.12450147e-01 1.88311040e-01
-2.59831190e-01 4.68785197e-01 4.42949980e-01 4.25598361e-02
-8.01074445e-01 9.07473862e-01 7.45501280e+00 8.51948321e-01
-1.70919788e+00 1.94713026e-02 4.10999030e-01 -1.08317822e-01
-4.66958970e-01 1.11226194e-01 -1.04172873e+00 5.91008425e-01
2.87255883e-01 -1.29497409e-01 5.86011708e-01 8.57835829e-01
1.69373691e-01 -1.62097722e-01 -1.03637552e+00 1.61460125e+00
-2.15333104e-01 -1.77716434e+00 -6.57111928e-02 1.13302767e-01
1.08519697e+00 5.18610597e-01 8.59239027e-02 -1.22690000e-01
2.87421793e-01 -8.20676565e-01 2.74147093e-01 5.04400969e-01
1.11347914e+00 -3.89559329e-01 3.11330259e-01 2.57853270e-01
-8.36070001e-01 2.12845206e-01 -8.35460365e-01 -3.44053432e-02
7.36617371e-02 1.25846148e+00 -5.59857488e-01 2.64095366e-01
1.10733700e+00 8.56654167e-01 -6.22029483e-01 1.28132463e+00
2.07170397e-02 6.22026324e-01 -4.66598600e-01 -2.97890510e-02
-5.59960231e-02 -3.38824421e-01 1.15916230e-01 8.30207705e-01
3.16522956e-01 6.04866967e-02 2.47687101e-01 1.19412744e+00
-3.35362792e-01 -1.60594180e-03 -8.94253552e-01 2.50958055e-01
2.18337730e-01 1.71700740e+00 -6.87660396e-01 -2.94906809e-03
-9.98662353e-01 4.21076626e-01 3.13292712e-01 4.01979923e-01
-7.05614388e-01 -3.20137978e-01 7.64985919e-01 4.56422180e-01
-1.70514151e-01 -5.14608622e-01 -4.38166082e-01 -1.32900929e+00
1.66942962e-02 -5.81411958e-01 8.08231905e-03 -1.23074460e+00
-1.52554464e+00 4.43616509e-01 -1.23671077e-01 -1.13622141e+00
1.39253825e-01 -8.91149282e-01 -7.56093860e-01 8.03054214e-01
-1.83571434e+00 -1.35115647e+00 -9.43625867e-01 9.43293929e-01
9.61574093e-02 -4.29363012e-01 4.25829113e-01 6.74157977e-01
-4.68558043e-01 2.16151699e-01 3.03659558e-01 -5.67927957e-02
7.99282908e-01 -9.68372643e-01 3.12167436e-01 7.80377805e-01
2.82263875e-01 5.43159783e-01 3.42793703e-01 -3.92242551e-01
-1.80092537e+00 -9.92635608e-01 3.95310879e-01 -4.28612858e-01
5.48300028e-01 -3.88315767e-01 -8.06235135e-01 6.02550924e-01
2.76800513e-01 7.50070035e-01 6.76634789e-01 -6.99554309e-02
-1.18500918e-01 -6.49982095e-01 -1.13793874e+00 5.35878539e-01
1.25167108e+00 -5.98691165e-01 4.42695394e-02 7.70940065e-01
3.75478268e-01 -4.84212577e-01 -9.01007533e-01 4.99760926e-01
3.54402363e-01 -1.32866061e+00 1.12630713e+00 -1.89712912e-01
5.92624784e-01 -3.44234198e-01 -1.74662143e-01 -1.16874385e+00
-5.15145242e-01 -7.09290147e-01 6.44013984e-03 1.16900420e+00
-2.38746237e-02 -5.77258766e-01 9.25499678e-01 6.20323718e-01
-1.75847322e-01 -4.98216778e-01 -7.64029264e-01 -5.09022772e-01
-9.20135379e-02 -5.00123501e-01 8.57594371e-01 7.55348146e-01
-6.32115901e-01 4.12201852e-01 -3.08203369e-01 -5.46612293e-02
8.76055062e-01 6.36781752e-01 1.10751855e+00 -1.41610301e+00
-1.97003752e-01 -3.18189889e-01 -5.97500205e-01 -1.13584280e+00
1.68305844e-01 -9.89255250e-01 -7.91633204e-02 -1.47212112e+00
3.14904928e-01 -9.62543428e-01 -1.08956173e-01 3.53813469e-01
7.21754506e-02 5.31660914e-01 -8.14375579e-02 2.47310638e-01
-7.12377191e-01 4.01441216e-01 1.63517594e+00 -1.58708304e-01
1.19370103e-01 -1.78439543e-01 -4.57565129e-01 6.82946861e-01
5.12814641e-01 -1.73697397e-02 -4.22167867e-01 -7.97643960e-01
4.41236526e-01 -3.99442136e-01 5.48034608e-01 -1.38456714e+00
4.25578296e-01 -5.64098120e-01 6.09635592e-01 -4.74045217e-01
1.94437280e-01 -7.53931522e-01 1.62427172e-01 -7.05018500e-03
2.57262774e-02 -1.83252618e-01 1.20009020e-01 1.40560642e-01
-6.34985790e-02 -1.41064733e-01 1.12040830e+00 -3.21979076e-01
-1.05013812e+00 9.96578634e-01 1.11601345e-01 3.56363915e-02
1.04955935e+00 -4.19142703e-03 -8.30436409e-01 -8.23595971e-02
2.14600652e-01 -1.42804816e-01 6.08126760e-01 1.50050536e-01
6.66691661e-01 -1.53271866e+00 -6.09395027e-01 6.96107388e-01
4.16867852e-01 3.81125182e-01 3.62350017e-01 4.60799009e-01
-8.91853809e-01 3.52490425e-01 -7.13621676e-01 -9.53965604e-01
-8.99605572e-01 5.32834351e-01 1.80969059e-01 2.95485973e-01
-1.10408640e+00 5.30930042e-01 4.48660225e-01 -1.45386353e-01
-3.58820409e-02 -4.55693990e-01 1.39325514e-01 -6.48251355e-01
8.82392883e-01 3.33674610e-01 3.60266000e-01 -4.50652599e-01
-1.37392670e-01 9.59057868e-01 1.06748812e-01 4.43420768e-01
1.46330655e+00 4.31546569e-02 -2.13428259e-01 -6.66093305e-02
1.05244637e+00 7.94845298e-02 -1.30630660e+00 -3.91882151e-01
-6.47544742e-01 -9.27451789e-01 6.51647031e-01 -4.47078377e-01
-1.49792254e+00 1.11772335e+00 7.82148898e-01 -5.60785756e-02
1.21697760e+00 1.51848257e-01 1.08010054e+00 5.25444508e-01
7.74749398e-01 -1.05799639e+00 9.75814536e-02 5.15153766e-01
6.93372309e-01 -1.52786529e+00 1.72528271e-02 -3.18159372e-01
-1.28143266e-01 1.18419147e+00 7.76229262e-01 1.68154821e-01
7.94836879e-01 6.36116326e-01 1.93401620e-01 -3.79264712e-01
-6.95973039e-01 -2.27427915e-01 -3.67546380e-02 9.56772208e-01
5.54310083e-01 -2.18947113e-01 1.08570093e-03 1.00732282e-01
-5.68594970e-02 5.44251323e-01 5.32895803e-01 7.74529517e-01
-4.97029424e-01 -1.05220604e+00 -2.44297996e-01 6.20068491e-01
-3.44442308e-01 -1.76553950e-01 1.50923491e-01 2.76726574e-01
4.40179795e-01 8.30184340e-01 -4.52774949e-02 -3.57400537e-01
1.86372146e-01 -5.22333562e-01 8.88757229e-01 -4.99662578e-01
-1.69598877e-01 -2.38059014e-01 -4.97169971e-01 -6.54396176e-01
-5.60315609e-01 -1.87811598e-01 -1.03474462e+00 -8.35504889e-01
-1.72985688e-01 -5.20415306e-01 7.69696534e-01 8.72933924e-01
5.09740174e-01 1.96499422e-01 6.93607152e-01 -1.16474414e+00
3.54447751e-03 -7.02591300e-01 -7.14602351e-01 6.34041667e-01
1.23386852e-01 -9.93283093e-01 -2.58674286e-02 1.99513440e-03]
|
[9.604804039001465, -2.6068248748779297]
|
4a81901b-5ded-4c32-b27c-f751cc1165af
|
transferable-deep-learning-power-system-short
|
2303.07138
| null |
https://arxiv.org/abs/2303.07138v1
|
https://arxiv.org/pdf/2303.07138v1.pdf
|
Transferable Deep Learning Power System Short-Term Voltage Stability Assessment with Physics-Informed Topological Feature Engineering
|
Deep learning (DL) algorithms have been widely applied to short-term voltage stability (STVS) assessment in power systems. However, transferring the knowledge learned in one power grid to other power grids with topology changes is still a challenging task. This paper proposed a transferable DL-based model for STVS assessment by constructing the topology-aware voltage dynamic features from raw PMU data. Since the reactive power flow and grid topology are essential to voltage stability, the topology-aware and physics-informed voltage dynamic features are utilized to effectively represent the topological and temporal patterns from post-disturbance system dynamic trajectories. The proposed DL-based STVS assessment model is tested under random operating conditions on the New England 39-bus system. It has 99.99\% classification accuracy of the short-term voltage stability status using the topology-aware and physics-informed voltage dynamic features. In addition to high accuracy, the experiments show good adaptability to PMU errors. Moreover, The proposed STVS assessment method has outstanding performance on new grid topologies after fine-tuning. In particular, the highest accuracy reaches 99.68\% in evaluation, which demonstrates a good knowledge transfer ability of the proposed model for power grid topology change.
|
['Kai Wu', 'Peiyuan Sun', 'Zijian Lv', 'Xin Chen', 'Zijian Feng']
|
2023-03-13
| null | null | null | null |
['feature-engineering']
|
['methodology']
|
[-8.41622293e-01 -8.37218106e-01 9.03222710e-03 -1.88103035e-01
-7.37898171e-01 -7.94001102e-01 4.69166547e-01 3.02383780e-01
4.10699099e-01 1.21617270e+00 -2.09031656e-01 -2.82929659e-01
-5.62708735e-01 -1.06147492e+00 -3.64846706e-01 -1.07426190e+00
-9.23538327e-01 4.59761411e-01 -7.96608329e-02 -4.13022637e-01
-4.78595607e-02 9.15370822e-01 -1.07724309e+00 -2.64056146e-01
1.37947404e+00 1.14611089e+00 -2.84000393e-02 2.39615873e-01
1.97899222e-01 4.74001408e-01 -1.12255204e+00 5.03903329e-01
1.74786672e-01 -3.49910021e-01 -6.37260795e-01 -1.73525363e-01
-1.26859844e-01 -4.84928861e-02 -5.02644658e-01 1.32190716e+00
8.22260439e-01 3.33922863e-01 6.25609756e-01 -1.49187636e+00
-6.59459054e-01 6.32568002e-01 -6.03572071e-01 9.08550978e-01
4.70292538e-01 4.64137524e-01 7.22841203e-01 -5.61853349e-01
2.28697628e-01 9.60866690e-01 7.33423650e-01 -2.47151762e-01
-1.20793831e+00 -7.72205830e-01 1.94042534e-01 9.08390820e-01
-1.39526403e+00 2.76459992e-01 1.12749684e+00 -7.34937131e-01
1.15562236e+00 2.64046583e-02 8.99758160e-01 4.54193175e-01
8.38501275e-01 5.99783242e-01 1.20469785e+00 1.12777732e-01
3.53147060e-01 2.18291357e-02 2.87466235e-02 2.04458803e-01
8.07510689e-02 1.23798326e-01 -8.74572173e-02 1.88982695e-01
3.27591807e-01 -5.33783853e-01 -6.69002116e-01 -4.51107115e-01
-7.91920245e-01 7.65379488e-01 9.47355032e-01 7.09189236e-01
-1.42031565e-01 -4.37947541e-01 8.15005660e-01 6.54206097e-01
5.02290487e-01 4.07229275e-01 -9.09720778e-01 -2.15991020e-01
-7.92315006e-01 -2.42371801e-02 5.22095025e-01 5.68259597e-01
3.40536654e-01 1.32405126e+00 7.18701258e-02 4.35387254e-01
-1.56405225e-01 7.18061745e-01 7.59593070e-01 -3.91694635e-01
1.97849676e-01 6.13324881e-01 7.51078501e-02 -1.10759401e+00
-8.67826760e-01 -1.03647709e+00 -1.29931056e+00 5.22631764e-01
7.51876682e-02 -3.62225085e-01 -4.68721122e-01 1.70237279e+00
1.72960594e-01 1.61350623e-01 8.72883759e-03 5.90026498e-01
6.01975977e-01 1.19760728e+00 -2.51045853e-01 -7.73353517e-01
9.12455380e-01 -2.40696415e-01 -1.02988696e+00 4.67104048e-01
5.44764280e-01 -2.61104733e-01 7.85074413e-01 4.98955011e-01
-9.55246031e-01 -7.79884338e-01 -1.36519837e+00 6.70810401e-01
-6.56386197e-01 2.74659395e-02 1.85906559e-01 2.70413309e-01
-1.11199498e+00 8.21051419e-01 -6.75950289e-01 -2.32487828e-01
2.01140046e-01 3.02190483e-01 -5.39179482e-02 5.36388934e-01
-1.95089877e+00 1.35259235e+00 8.15469384e-01 1.65410921e-01
-9.11791325e-01 -1.02146661e+00 -8.17560077e-01 2.42289305e-01
2.25117058e-01 -3.02821994e-01 8.63065600e-01 -4.95584697e-01
-1.54582810e+00 -7.82014355e-02 1.37064278e-01 -6.18510783e-01
3.40831310e-01 4.18475837e-01 -1.16609216e+00 1.83638468e-01
5.36511838e-02 -4.86727983e-01 6.16297722e-01 -1.12982130e+00
-8.13551128e-01 -2.45208681e-01 -3.57840449e-01 2.00656205e-01
-3.65744799e-01 -5.02144277e-01 6.00390673e-01 -5.88505864e-01
-4.05914262e-02 -3.10007751e-01 6.65073991e-02 -5.27723491e-01
1.24949012e-02 -7.84959495e-01 1.38047755e+00 -1.06363165e+00
1.26327610e+00 -1.78674710e+00 1.30105764e-01 3.95458728e-01
-7.51898214e-02 4.89054441e-01 2.12267324e-01 5.13463199e-01
-3.94285619e-01 -3.11500043e-01 -1.37614310e-01 5.90968072e-01
3.50534230e-01 1.53646752e-01 -3.47676098e-01 5.22790253e-01
2.04497129e-02 9.17306304e-01 -1.04015613e+00 1.58464089e-01
8.34700286e-01 2.14592069e-01 1.06903262e-01 -1.96281359e-01
1.42272394e-02 7.35632658e-01 -2.57839859e-01 2.01258868e-01
9.26460385e-01 -8.94582644e-02 1.38097331e-01 -5.43663442e-01
3.63801443e-03 -1.18126392e-01 -1.25825286e+00 1.35949922e+00
-5.41866541e-01 6.03641629e-01 1.46366343e-01 -1.66913509e+00
1.11869144e+00 3.05994034e-01 9.03975904e-01 -1.14647281e+00
7.11923763e-02 -1.26546351e-02 1.13634996e-01 -2.45315805e-01
-9.85870063e-02 2.53569558e-02 -1.01462372e-01 2.08749771e-01
4.43687588e-01 -4.52189893e-01 3.96515071e-01 1.33106142e-01
4.93891478e-01 -1.16122700e-01 4.45287317e-01 -1.04257989e+00
1.05673265e+00 -8.79047737e-02 1.06670415e+00 1.67332590e-02
-3.20017219e-01 -4.22976822e-01 3.78929138e-01 -5.82973301e-01
-7.54668951e-01 -1.44292009e+00 -6.99340761e-01 2.00067267e-01
1.76971868e-01 -1.38558596e-01 -3.10786903e-01 -5.97777307e-01
3.30311507e-01 1.13310194e+00 -3.64177197e-01 -5.08249938e-01
-5.20081580e-01 -1.25102961e+00 -3.41959819e-02 7.67614603e-01
6.35781944e-01 -6.74280107e-01 -1.05933517e-01 5.79698920e-01
1.47850156e-01 -8.99903297e-01 5.93713969e-02 1.97161943e-01
-6.15453601e-01 -1.12192225e+00 -3.53910089e-01 -9.37191725e-01
3.13243181e-01 -5.33884168e-01 1.07951772e+00 -3.62965316e-01
-1.44705445e-01 5.91179878e-02 -7.56621286e-02 8.13190565e-02
-2.78874993e-01 -7.22071379e-02 8.10499430e-01 -2.84088314e-01
1.07666016e-01 -9.33616281e-01 -2.70482272e-01 2.59871215e-01
-3.05408031e-01 -4.96256799e-01 -2.11831275e-02 1.07401264e+00
4.21420276e-01 1.42357099e+00 1.40799749e+00 1.02882825e-01
7.57880211e-01 -6.07068658e-01 -1.28183854e+00 3.85827199e-02
-1.03017247e+00 -4.58958477e-01 1.49869180e+00 -1.59528136e-01
-9.59582746e-01 -4.49467719e-01 -1.12492904e-01 -3.24912548e-01
1.45301467e-03 7.21837759e-01 -6.61931157e-01 -3.10207367e-01
2.32610390e-01 5.04121244e-01 -1.47396997e-01 -6.01120770e-01
1.79802433e-01 4.12689269e-01 6.41712904e-01 -3.96294534e-01
1.33386254e+00 2.63409261e-02 3.01961660e-01 -7.59581447e-01
-2.58543819e-01 6.49953857e-02 -9.67795014e-01 -2.00043723e-01
5.19423664e-01 -8.91037345e-01 -9.73503232e-01 8.53410304e-01
-4.75031823e-01 -2.48971060e-01 -5.04824638e-01 3.07461768e-01
-2.40418762e-01 6.52462304e-01 -8.47669065e-01 -4.18049365e-01
-5.05920529e-01 -1.30665326e+00 2.46700868e-01 3.38729590e-01
3.14149201e-01 -1.75722277e+00 -1.42305031e-01 -3.18660975e-01
4.48470533e-01 7.87940383e-01 1.32225990e+00 -5.49102008e-01
-1.50625244e-01 -2.17916770e-03 5.44211417e-02 5.31326592e-01
5.42609632e-01 -9.62336920e-03 -5.24633944e-01 -1.17284477e+00
1.61883116e-01 1.69173703e-02 -2.38634646e-02 5.73151588e-01
1.02564490e+00 -2.55522609e-01 -2.19454154e-01 6.51402473e-01
1.72585094e+00 8.77962947e-01 7.86086991e-02 4.00622010e-01
5.36882520e-01 -1.18096210e-01 2.58548200e-01 5.42876720e-01
5.70363641e-01 5.52524805e-01 1.79595724e-01 -2.10784346e-01
1.85077235e-01 1.51105464e-01 4.56259876e-01 1.36823511e+00
6.05624914e-01 -1.14063248e-02 -9.37111795e-01 7.12721586e-01
-1.50709713e+00 -8.75835836e-01 -2.33010545e-01 1.77137578e+00
5.39184809e-01 3.82042944e-01 -1.00129031e-01 6.20702624e-01
4.73514289e-01 4.23846059e-02 -1.03150070e+00 -6.56246483e-01
-5.79434156e-01 6.59759939e-02 1.06854048e-02 4.89712328e-01
-9.67401385e-01 2.63591528e-01 5.24910355e+00 1.14159727e+00
-1.26950479e+00 6.52980134e-02 4.38456804e-01 2.52075642e-01
6.42915219e-02 -5.16460478e-01 -3.43449503e-01 8.65686357e-01
9.80781555e-01 -1.16298163e+00 3.77077371e-01 5.33079565e-01
6.45983160e-01 -1.34617314e-01 -8.30150127e-01 9.38938200e-01
-1.03190228e-01 -1.13047588e+00 1.99571609e-01 -2.24665031e-01
1.21914124e+00 1.21947572e-01 -8.07219520e-02 3.91873896e-01
3.29903245e-01 -7.65325189e-01 3.50764424e-01 4.49069381e-01
7.63693750e-01 -1.16463923e+00 1.02815735e+00 1.79546535e-01
-1.84571123e+00 -5.66485941e-01 -3.21572721e-02 -3.05850822e-02
5.53888738e-01 8.78023624e-01 -5.67994952e-01 1.52272260e+00
1.04561996e+00 1.37380624e+00 -6.87659323e-01 9.65477705e-01
-3.24588150e-01 7.57128537e-01 -3.31042379e-01 3.51182520e-01
2.66846716e-01 -4.53814983e-01 7.36926854e-01 6.77421689e-01
3.16127270e-01 -1.33571982e-01 6.75627947e-01 6.36479497e-01
2.02115864e-01 -2.25309312e-01 -3.53859961e-01 4.65724438e-01
6.19632006e-01 1.31344604e+00 -4.58975375e-01 -6.04725063e-01
-7.73081183e-02 2.96208650e-01 -1.63362101e-01 5.77236891e-01
-7.98005521e-01 -6.06334388e-01 7.60733783e-01 -2.70932317e-01
9.57047120e-02 -1.35094449e-01 -4.21180934e-01 -1.19582522e+00
-4.06013168e-02 -7.31308520e-01 6.38641834e-01 -9.20515954e-01
-1.81894290e+00 6.76080644e-01 1.59968585e-01 -1.50141656e+00
-4.04837251e-01 -3.63160819e-01 -1.11842155e+00 1.17217517e+00
-1.62781119e+00 -8.68014872e-01 -1.43234342e-01 7.59960473e-01
6.99871302e-01 -5.40203452e-01 6.18013322e-01 2.68720597e-01
-9.64242041e-01 4.87819672e-01 8.52675259e-01 2.29828954e-01
-1.06122442e-01 -1.70476675e+00 2.62085617e-01 1.12685788e+00
-4.38476115e-01 -4.49160747e-02 7.19860256e-01 -5.28642476e-01
-1.51967502e+00 -1.15135849e+00 1.47101507e-01 -9.26390290e-02
1.08011234e+00 -1.22318655e-01 -1.45391297e+00 4.45054650e-01
7.04964936e-01 -3.14598228e-03 2.73537397e-01 -3.54209334e-01
3.88469428e-01 -6.41477346e-01 -1.46880305e+00 1.67425394e-01
5.25304675e-01 -4.33776319e-01 -9.19579744e-01 3.33446056e-01
1.58397168e-01 -2.08625078e-01 -1.72200787e+00 8.41809571e-01
-2.46588513e-01 -1.80569872e-01 8.27710152e-01 -2.92361915e-01
-4.96871501e-01 -8.76500010e-01 8.83268937e-02 -2.29088736e+00
-8.17814648e-01 -6.63464308e-01 -4.44219112e-01 1.25886881e+00
9.75667611e-02 -1.21167564e+00 1.05787732e-01 -1.87102720e-01
-1.93300650e-01 -6.71466112e-01 -1.27777052e+00 -1.10699296e+00
7.96613157e-01 -6.71554506e-02 1.23124981e+00 1.66286767e+00
4.53826666e-01 3.09801191e-01 1.91627279e-01 6.84554636e-01
8.50721121e-01 4.33752418e-01 1.97671071e-01 -1.31752217e+00
3.39121640e-01 -7.12235749e-01 -6.97114289e-01 -2.43964657e-01
4.43806887e-01 -1.04125512e+00 -4.83230770e-01 -1.93247485e+00
-3.25238198e-01 -2.29504809e-01 -8.16866994e-01 3.25507820e-01
-7.22158551e-02 -1.01446010e-01 8.63505006e-02 -8.64102915e-02
-8.11771899e-02 1.14006388e+00 1.07965565e+00 -4.70110953e-01
-5.98279834e-02 -8.93331319e-02 2.43356094e-01 4.57973778e-01
1.39540792e+00 4.12608325e-01 -6.25907004e-01 -1.29150659e-01
-1.87394291e-01 3.02990645e-01 -1.04148582e-01 -1.41797221e+00
1.26070842e-01 9.20050591e-02 8.54359269e-01 -1.06248772e+00
-2.38594279e-01 -7.66420007e-01 3.60507280e-01 9.01926160e-01
3.28934520e-01 6.69058740e-01 7.12842345e-01 4.02183771e-01
-3.39586437e-01 4.90482152e-01 9.30634618e-01 2.42278904e-01
-1.28014386e+00 4.05510962e-01 -6.77898228e-01 7.80460611e-02
1.27409852e+00 -5.78487664e-02 -3.34993273e-01 -3.21970403e-01
-9.13293183e-01 1.01195371e+00 2.22588345e-01 7.04559684e-01
3.52319568e-01 -1.51655829e+00 -7.07809329e-01 4.75089967e-01
-2.59055406e-01 -3.21574152e-01 3.67268831e-01 8.11790884e-01
-2.69058883e-01 4.10277873e-01 -4.61596251e-01 -9.58434701e-01
-6.43541932e-01 6.64339185e-01 9.12121236e-01 -1.15119979e-01
-8.84935200e-01 2.50827342e-01 -2.65497625e-01 -3.96696299e-01
-1.35403052e-01 -3.51445258e-01 -5.69372356e-01 4.24697936e-01
2.97345132e-01 6.58785224e-01 5.81973910e-01 -7.35939384e-01
-3.02950263e-01 7.51833618e-01 4.44460005e-01 5.73314667e-01
1.29084086e+00 -3.86826664e-01 -1.36755511e-01 5.60894012e-01
1.22858620e+00 -5.59254467e-01 -1.21627927e+00 -1.74701586e-01
2.88918447e-02 -1.48047790e-01 4.52279389e-01 -1.38549078e+00
-1.66313398e+00 8.37218702e-01 9.44496393e-01 5.67380965e-01
1.32632565e+00 -6.58618212e-01 7.61691630e-01 -1.10905105e-02
8.99593115e-01 -1.21993744e+00 -5.58202028e-01 6.23824656e-01
7.37254798e-01 -8.34337592e-01 -1.45653933e-02 3.36783797e-01
-3.10619593e-01 1.17394590e+00 6.99336112e-01 -9.62945749e-04
9.19986784e-01 5.03782034e-01 -8.30496773e-02 -2.83062900e-03
-6.79557920e-01 2.59559780e-01 2.75609910e-01 8.82738650e-01
-1.40056267e-01 2.43758082e-01 -1.68672845e-01 2.52427042e-01
-3.75449032e-01 -2.06651375e-01 5.55832922e-01 6.84831858e-01
-2.45579615e-01 -5.63903332e-01 -2.97476023e-01 6.43123448e-01
-2.00622976e-01 4.85415667e-01 4.13799584e-01 8.05532336e-01
1.12141937e-01 9.39959586e-01 6.02252632e-02 -2.54630655e-01
5.06183207e-01 -4.06816714e-02 2.35223711e-01 -4.84336279e-02
-5.04667640e-01 -2.24607274e-01 -2.64735937e-01 -3.21854740e-01
1.79648608e-01 -6.96182787e-01 -1.44458163e+00 -4.70016181e-01
-2.44165644e-01 8.56735349e-01 4.16885495e-01 9.40839291e-01
1.01348348e-01 9.82571959e-01 1.29143775e+00 -7.89910257e-01
-6.91435874e-01 -1.00553989e+00 -1.02310598e+00 3.91702652e-01
4.20623094e-01 -1.10486186e+00 -6.48595870e-01 -3.92758667e-01]
|
[5.9634504318237305, 2.6055026054382324]
|
05fd7bd6-8a19-4a2c-a843-950c9ac1c6cb
|
self-supervised-learning-of-event-based
|
2106.01862
| null |
https://arxiv.org/abs/2106.01862v2
|
https://arxiv.org/pdf/2106.01862v2.pdf
|
Self-Supervised Learning of Event-Based Optical Flow with Spiking Neural Networks
|
The field of neuromorphic computing promises extremely low-power and low-latency sensing and processing. Challenges in transferring learning algorithms from traditional artificial neural networks (ANNs) to spiking neural networks (SNNs) have so far prevented their application to large-scale, complex regression tasks. Furthermore, realizing a truly asynchronous and fully neuromorphic pipeline that maximally attains the abovementioned benefits involves rethinking the way in which this pipeline takes in and accumulates information. In the case of perception, spikes would be passed as-is and one-by-one between an event camera and an SNN, meaning all temporal integration of information must happen inside the network. In this article, we tackle these two problems. We focus on the complex task of learning to estimate optical flow from event-based camera inputs in a self-supervised manner, and modify the state-of-the-art ANN training pipeline to encode minimal temporal information in its inputs. Moreover, we reformulate the self-supervised loss function for event-based optical flow to improve its convexity. We perform experiments with various types of recurrent ANNs and SNNs using the proposed pipeline. Concerning SNNs, we investigate the effects of elements such as parameter initialization and optimization, surrogate gradient shape, and adaptive neuronal mechanisms. We find that initialization and surrogate gradient width play a crucial part in enabling learning with sparse inputs, while the inclusion of adaptivity and learnable neuronal parameters can improve performance. We show that the performance of the proposed ANNs and SNNs are on par with that of the current state-of-the-art ANNs trained in a self-supervised manner.
|
['Federico Paredes-Vallés', 'Jesse Hagenaars', 'Guido de Croon']
|
2021-06-03
| null |
http://proceedings.neurips.cc/paper/2021/hash/39d4b545fb02556829aab1db805021c3-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/39d4b545fb02556829aab1db805021c3-Paper.pdf
|
neurips-2021-12
|
['event-based-optical-flow']
|
['computer-vision']
|
[ 5.24189949e-01 -2.02893302e-01 4.18066859e-01 -2.65698522e-01
5.92722669e-02 -4.99593645e-01 6.12309813e-01 -4.67749760e-02
-1.10767674e+00 8.19853067e-01 -2.10996211e-01 -2.60306580e-04
-1.36803493e-01 -6.62787497e-01 -9.68110263e-01 -8.99631381e-01
-1.87768489e-01 3.63962315e-02 5.35315156e-01 3.64581198e-02
2.61095226e-01 7.27540195e-01 -1.78730834e+00 2.68297762e-01
5.33609688e-01 1.26641774e+00 3.50372881e-01 8.75133216e-01
-4.61721867e-02 8.48897636e-01 -4.99170810e-01 -1.20956585e-01
2.85971045e-01 -5.11587381e-01 -2.19479769e-01 -3.62744629e-01
2.61132121e-01 -1.09594740e-01 -5.31025767e-01 7.19061673e-01
5.65442741e-01 4.98899398e-03 6.39992356e-01 -1.18936920e+00
-2.59982556e-01 4.23121005e-01 -9.19443667e-02 4.34991509e-01
-1.56502724e-01 5.60592353e-01 5.24976015e-01 -6.60337389e-01
6.00113034e-01 7.71957397e-01 7.30541110e-01 8.24580133e-01
-1.26012099e+00 -6.83897972e-01 4.21799384e-02 1.58576250e-01
-1.03219926e+00 -7.27562606e-01 6.58938348e-01 -3.65963310e-01
1.25546288e+00 -2.33882576e-01 1.10826719e+00 1.14962864e+00
3.83263111e-01 5.12692213e-01 1.07641697e+00 -1.51033849e-01
7.92525411e-01 -1.31970018e-01 -5.42431790e-03 6.13301277e-01
3.10653418e-01 9.51273143e-02 -8.71720314e-01 2.85913765e-01
1.09719515e+00 -1.60067119e-02 -1.93896174e-01 -1.43473879e-01
-1.16320479e+00 3.66152108e-01 5.86358964e-01 3.38700771e-01
-3.00029069e-01 6.16382301e-01 2.72636861e-01 3.85293067e-01
-9.91203710e-02 5.01641691e-01 -3.10221076e-01 -3.48317355e-01
-1.16075587e+00 -1.67391449e-01 9.71238852e-01 4.22971576e-01
8.04425001e-01 3.75424355e-01 -1.72980819e-02 3.90403271e-01
1.22208260e-01 3.19474488e-01 6.85898364e-01 -1.17584419e+00
9.96307731e-02 5.83899975e-01 -1.48700044e-01 -5.35076559e-01
-5.75665474e-01 -4.96525288e-01 -1.00688684e+00 5.33449829e-01
7.16097176e-01 -2.72007078e-01 -9.30857599e-01 1.85814238e+00
-1.73174679e-01 4.82869685e-01 2.38977477e-01 9.89769936e-01
4.56130892e-01 6.71783268e-01 -1.28265575e-01 -3.81513447e-01
1.01238573e+00 -7.03003645e-01 -5.56749225e-01 -4.16413665e-01
2.32366070e-01 -2.70814568e-01 8.69035482e-01 4.22064185e-01
-1.37323618e+00 -4.72099572e-01 -1.15045607e+00 -1.41367495e-01
-4.22166139e-01 5.53071834e-02 6.53213561e-01 4.39496636e-01
-1.18507326e+00 9.18677568e-01 -1.36554897e+00 -3.78528833e-01
7.45636106e-01 7.46590137e-01 -2.01506376e-01 3.86886328e-01
-8.09451222e-01 7.36634016e-01 1.71203494e-01 2.54700661e-01
-9.34635520e-01 -7.46599019e-01 -3.85644555e-01 1.47437602e-01
-1.66775778e-01 -9.08296645e-01 9.92220819e-01 -1.14977086e+00
-1.90228570e+00 6.90588236e-01 -3.25580746e-01 -8.40663612e-01
3.85771066e-01 6.68561757e-02 -4.16546874e-02 2.17951372e-01
-3.95063996e-01 9.34478045e-01 8.36459160e-01 -8.53499532e-01
-4.15879190e-01 -3.68379384e-01 -5.41115254e-02 -8.50561708e-02
-7.21889615e-01 -1.64015606e-01 -1.56149819e-01 -3.58796060e-01
-4.26050499e-02 -7.84131289e-01 -1.65599182e-01 4.67877120e-01
1.48535073e-01 1.76593542e-01 6.90584183e-01 -3.18044238e-02
8.38025987e-01 -2.04242039e+00 2.51964301e-01 -3.97119038e-02
6.65995106e-02 3.33881915e-01 -8.68840069e-02 1.66865394e-01
1.72582224e-01 -4.12543505e-01 -5.39286494e-01 -5.16523242e-01
-2.69889027e-01 4.33034748e-01 -1.98849842e-01 3.80332857e-01
4.56683010e-01 9.90324140e-01 -7.31531560e-01 -2.24717021e-01
9.14604813e-02 7.54406631e-01 -5.87588072e-01 2.23154575e-01
-2.22394541e-01 7.32494533e-01 -1.23368345e-01 3.94371390e-01
2.13929951e-01 -3.48808527e-01 -1.10059783e-01 -2.09977522e-01
-5.76174974e-01 3.16932797e-01 -1.10954309e+00 2.00027394e+00
-5.79600990e-01 9.79043543e-01 9.36838090e-02 -1.14815855e+00
1.02461326e+00 8.67321715e-02 6.15236342e-01 -1.03092825e+00
2.02907518e-01 3.50709587e-01 2.80674815e-01 -3.38952243e-01
4.01696414e-02 9.19845514e-03 3.60918730e-01 4.17844713e-01
4.06645983e-01 1.39209181e-01 3.40747148e-01 -3.51444446e-02
1.40378237e+00 3.35325837e-01 -1.41543671e-01 -1.91142872e-01
2.99598753e-01 -2.24902660e-01 4.90996301e-01 7.23376095e-01
-9.48536396e-02 5.67414045e-01 4.57050145e-01 -3.82362306e-01
-1.08637106e+00 -1.17890370e+00 -1.31475195e-01 8.01532269e-01
-5.80804422e-02 -3.70765738e-02 -6.66249335e-01 6.35087192e-02
-2.24875376e-01 1.95201382e-01 -3.53917181e-01 -1.23668380e-01
-6.76787674e-01 -9.70894396e-01 7.51290202e-01 5.67392588e-01
4.83442545e-01 -1.29573250e+00 -1.58503795e+00 5.62188685e-01
3.62318873e-01 -1.34049249e+00 1.16794571e-01 8.39012861e-01
-1.00806701e+00 -7.69900501e-01 -6.21104240e-01 -7.06039011e-01
6.79615796e-01 -2.84278214e-01 8.15328181e-01 -4.07148868e-01
-4.82531846e-01 2.32233346e-01 4.99629043e-02 -4.11749959e-01
4.73197773e-02 8.66549164e-02 1.67753249e-02 2.16607258e-01
5.97619899e-02 -1.28157306e+00 -8.52005959e-01 7.86631927e-02
-1.14119267e+00 7.43405670e-02 6.72537386e-01 6.97602510e-01
6.57606006e-01 -3.02419305e-01 5.33642352e-01 -7.10833967e-01
2.83710629e-01 -2.01529667e-01 -7.69959509e-01 1.68151557e-02
-5.08995593e-01 4.22019631e-01 1.05022228e+00 -7.44669378e-01
-7.97807455e-01 5.98475575e-01 -1.41627446e-01 -4.30927932e-01
4.09966009e-03 2.19084710e-01 3.08639139e-01 -4.59387064e-01
7.32979417e-01 3.02932113e-01 6.86495155e-02 -5.43655306e-02
3.87876332e-02 2.92280465e-01 7.24612057e-01 -3.53775263e-01
5.06614089e-01 8.35473597e-01 3.79160196e-01 -6.62250996e-01
-4.75885719e-01 -1.98344857e-01 -5.74098706e-01 -3.52362633e-01
6.78879321e-01 -7.47927427e-01 -1.09976864e+00 8.72966588e-01
-1.36076272e+00 -7.21068919e-01 -7.03700721e-01 4.66156751e-01
-7.62343645e-01 -2.29596227e-01 -7.48300433e-01 -8.41768980e-01
-2.99051851e-01 -9.72159684e-01 6.89611018e-01 6.54910505e-01
1.50829718e-01 -8.84663284e-01 2.76382416e-01 -2.38237515e-01
8.22584927e-01 1.36993989e-01 5.28232872e-01 -4.02898937e-01
-7.45678484e-01 2.11992562e-01 -2.89941519e-01 2.20641419e-01
-1.01052850e-01 6.62893355e-02 -1.29017544e+00 -1.76227704e-01
1.42968804e-01 -3.79660845e-01 1.26389611e+00 4.74921435e-01
1.10068905e+00 -2.14464322e-01 -4.19326238e-02 9.78045404e-01
1.65007043e+00 1.61595136e-01 7.49616385e-01 2.15429008e-01
5.08122444e-01 5.46970308e-01 -3.69987547e-01 5.10710299e-01
3.46579760e-01 3.30638617e-01 6.32992566e-01 1.24597095e-01
-1.73894137e-01 2.80497912e-02 5.69170833e-01 9.37827229e-01
-2.25882873e-01 -3.07425875e-02 -7.48896062e-01 4.95448142e-01
-1.92127991e+00 -9.10954237e-01 1.32910281e-01 2.17404890e+00
9.67152655e-01 3.35925132e-01 2.73791961e-02 3.30450654e-01
3.71497840e-01 1.64723545e-02 -9.15357053e-01 -4.52336997e-01
-4.25548553e-01 5.25716245e-01 5.11806369e-01 2.47780561e-01
-7.42427945e-01 7.92808354e-01 5.52961969e+00 1.38394982e-01
-1.60212898e+00 5.42176217e-02 3.36878657e-01 -4.06930596e-01
-6.62200972e-02 -2.79069524e-02 -8.38447928e-01 5.77100456e-01
1.34686339e+00 2.14550391e-01 9.31912780e-01 2.41289601e-01
2.89956301e-01 -1.97838202e-01 -1.28186476e+00 1.11366880e+00
-1.40382633e-01 -1.54302752e+00 -1.50238588e-01 -2.15113312e-01
6.63958669e-01 3.11331391e-01 -7.60380924e-02 -3.35629694e-02
-5.93142621e-02 -1.01794004e+00 6.58295870e-01 9.78928924e-01
5.06034672e-01 -3.21848482e-01 4.70764816e-01 3.22909862e-01
-1.05368340e+00 -3.79717261e-01 -3.56796980e-01 -4.68399972e-01
1.57535002e-01 8.68488193e-01 -2.67375290e-01 2.88234395e-03
8.26034546e-01 9.21334982e-01 -5.49171627e-01 1.30080986e+00
5.80385141e-02 4.73636270e-01 -6.58922791e-01 -3.22648793e-01
1.60602123e-01 -1.65015802e-01 3.86329144e-01 1.29306424e+00
3.32234919e-01 7.50981737e-03 -4.22745258e-01 1.14848530e+00
-1.47981137e-01 -3.32819611e-01 -5.39088190e-01 -1.17525078e-01
4.12055492e-01 1.32071793e+00 -8.50806653e-01 -2.39254143e-02
-3.94309163e-01 8.42924178e-01 4.08450902e-01 3.98680896e-01
-5.69603145e-01 -4.58307117e-01 5.36229491e-01 1.61636263e-01
5.03898323e-01 -3.40524167e-01 -5.79068601e-01 -1.09958661e+00
1.78573787e-01 -2.27571189e-01 1.31970849e-02 -7.54009604e-01
-1.00790894e+00 6.65298760e-01 -5.63655138e-01 -9.98526037e-01
-4.30155069e-01 -8.06711495e-01 -7.00645566e-01 3.99498641e-01
-1.76999569e+00 -7.67375052e-01 -4.33784574e-01 8.13274205e-01
1.08856820e-01 1.22235622e-02 7.24769115e-01 3.46759349e-01
-6.05278194e-01 3.67947727e-01 1.77775715e-02 4.21684124e-02
5.28328240e-01 -8.71550024e-01 1.99617937e-01 1.15137684e+00
2.36833274e-01 4.14898574e-01 4.69594449e-01 -6.50796369e-02
-1.73366010e+00 -1.06889415e+00 5.81205487e-01 -4.66485210e-02
8.85283709e-01 -5.67382693e-01 -8.19680274e-01 4.11374301e-01
5.40219583e-02 5.11480868e-01 1.71586379e-01 -4.72735852e-01
-4.11483765e-01 -6.17134571e-01 -9.76829886e-01 5.60516059e-01
1.21411276e+00 -5.18295884e-01 -2.28775948e-01 -9.37785953e-02
3.27025831e-01 -1.28276110e-01 -6.25926614e-01 4.08052713e-01
8.18389058e-01 -1.17316794e+00 7.46825933e-01 -2.72006392e-01
4.77886617e-01 -3.49697739e-01 5.55214807e-02 -1.10337639e+00
6.19581863e-02 -8.26888323e-01 -4.34176981e-01 9.22927737e-01
3.11566263e-01 -7.86330044e-01 9.80146289e-01 3.59283715e-01
-2.39873439e-01 -8.67850721e-01 -1.19132471e+00 -6.09005034e-01
-8.00653994e-02 -2.78842598e-01 -8.15756768e-02 3.48541021e-01
-1.83358222e-01 2.69767076e-01 3.23731080e-02 4.54108305e-02
6.04264021e-01 -7.99396411e-02 2.99244523e-01 -1.27793014e+00
-4.05286312e-01 -5.61117649e-01 -6.23893619e-01 -9.88458574e-01
1.01266369e-01 -7.44761825e-01 2.16923967e-01 -1.36603236e+00
-6.05055541e-02 -2.97770828e-01 -4.62268084e-01 6.19269133e-01
3.65914971e-01 3.66222560e-01 2.82045364e-01 2.42158398e-01
-5.85536420e-01 3.51102829e-01 9.27128673e-01 1.08860247e-01
-3.20311069e-01 -1.59238905e-01 -1.99231684e-01 5.37304103e-01
7.34870791e-01 -5.31448722e-01 -3.71057868e-01 -7.07826853e-01
3.81721467e-01 -5.17024063e-02 6.53796613e-01 -1.62059081e+00
1.12916768e+00 2.46093333e-01 4.43414360e-01 -6.33701235e-02
5.32116413e-01 -8.15856934e-01 -1.17417602e-02 6.26801491e-01
-3.67852300e-01 1.35118082e-01 2.08832368e-01 3.67576540e-01
-1.82299823e-01 -2.22524315e-01 9.36404705e-01 -2.04105839e-01
-5.32685816e-01 2.29138628e-01 -6.07851982e-01 1.81508407e-01
7.77402163e-01 -5.73006511e-01 -5.38201630e-01 -7.55669177e-02
-5.47351003e-01 -9.59018916e-02 3.47832829e-01 6.17784262e-03
6.07043922e-01 -7.97762454e-01 -3.65296632e-01 5.69626808e-01
-1.85377508e-01 7.76685998e-02 -1.83196157e-01 8.89373302e-01
-3.98293108e-01 2.98545092e-01 -9.16863680e-01 -7.39494801e-01
-5.58399379e-01 1.25944957e-01 4.67479885e-01 -5.52484170e-02
-2.51920998e-01 7.76071012e-01 -1.38225526e-01 -1.82169080e-01
5.35705209e-01 -4.53935534e-01 -6.38418794e-02 -2.94221612e-03
2.95407504e-01 3.37663531e-01 1.96254551e-01 -8.04229975e-02
-4.05673712e-01 6.94363654e-01 2.28352264e-01 -2.71829635e-01
1.64278281e+00 1.01637796e-01 -1.92380771e-01 7.89592743e-01
1.02148771e+00 -5.13442934e-01 -1.81856406e+00 2.34124623e-02
-1.26102373e-01 2.71251738e-01 5.45464605e-02 -6.94921315e-01
-1.32628632e+00 1.23850119e+00 7.93207288e-01 9.89973545e-04
1.40808082e+00 -2.83482015e-01 7.91293740e-01 6.14723861e-01
3.47552419e-01 -9.68651772e-01 2.88786709e-01 7.22412288e-01
3.37486774e-01 -8.76814425e-01 -3.72257113e-01 9.58938822e-02
-1.82995677e-01 1.34144580e+00 5.98809838e-01 -5.19375503e-01
6.15574479e-01 8.47579241e-01 -1.48833200e-01 1.07779518e-01
-1.08197105e+00 -1.67651594e-01 -1.24391861e-01 5.33089221e-01
2.03988239e-01 -4.15762693e-01 -9.07921717e-02 3.76190156e-01
-2.51248050e-02 4.56796706e-01 5.22046268e-01 9.35277820e-01
-3.89400840e-01 -8.67612302e-01 1.27860770e-01 4.92985785e-01
-3.84405077e-01 -1.04703672e-01 -1.15716770e-01 4.58037406e-01
6.02858290e-02 4.98889089e-01 4.32589561e-01 -2.73144573e-01
2.38112658e-01 4.74226773e-02 7.37209320e-01 -3.53995204e-01
-8.10451210e-01 -2.21653074e-01 -3.42303425e-01 -7.16925859e-01
-7.73935378e-01 -5.37821054e-01 -1.71465409e+00 2.21928768e-02
8.59455243e-02 -2.76669264e-01 1.10620308e+00 1.09255612e+00
5.01447618e-01 6.60615325e-01 4.32544291e-01 -1.03834820e+00
-2.61316210e-01 -5.20897031e-01 -1.16309419e-01 9.49385539e-02
5.43763340e-01 -3.31875741e-01 -5.09501874e-01 2.98478037e-01]
|
[8.232086181640625, 2.3732922077178955]
|
e5673ae5-5c2c-4922-98b6-3199728913be
|
stprivacy-spatio-temporal-tubelet
|
2301.03046
| null |
https://arxiv.org/abs/2301.03046v2
|
https://arxiv.org/pdf/2301.03046v2.pdf
|
STPrivacy: Spatio-Temporal Privacy-Preserving Action Recognition
|
Existing methods of privacy-preserving action recognition (PPAR) mainly focus on frame-level (spatial) privacy removal through 2D CNNs. Unfortunately, they have two major drawbacks. First, they may compromise temporal dynamics in input videos, which are critical for accurate action recognition. Second, they are vulnerable to practical attacking scenarios where attackers probe for privacy from an entire video rather than individual frames. To address these issues, we propose a novel framework STPrivacy to perform video-level PPAR. For the first time, we introduce vision Transformers into PPAR by treating a video as a tubelet sequence, and accordingly design two complementary mechanisms, i.e., sparsification and anonymization, to remove privacy from a spatio-temporal perspective. In specific, our privacy sparsification mechanism applies adaptive token selection to abandon action-irrelevant tubelets. Then, our anonymization mechanism implicitly manipulates the remaining action-tubelets to erase privacy in the embedding space through adversarial learning. These mechanisms provide significant advantages in terms of privacy preservation for human eyes and action-privacy trade-off adjustment during deployment. We additionally contribute the first two large-scale PPAR benchmarks, VP-HMDB51 and VP-UCF101, to the community. Extensive evaluations on them, as well as two other tasks, validate the effectiveness and generalization capability of our framework.
|
['Shuicheng Yan', 'Mike Zheng Shou', 'Jussi Keppo', 'Pan Zhou', 'Xiangyu Xu', 'Jiahe Li', 'Jia-Wei Liu', 'Hehe Fan', 'Jun Liu', 'Ming Li']
|
2023-01-08
| null | null | null | null |
['facial-expression-recognition', 'video-understanding']
|
['computer-vision', 'computer-vision']
|
[ 3.77474606e-01 2.57498417e-02 -3.34889233e-01 -1.43907323e-01
-6.36415064e-01 -9.36533034e-01 3.00492167e-01 -1.49785981e-01
-5.72555065e-01 5.18370628e-01 3.69074583e-01 -3.83966476e-01
1.09163150e-01 -6.11278355e-01 -8.38119924e-01 -8.73924375e-01
-1.90297917e-01 -3.70120376e-01 1.78939566e-01 9.37267169e-02
1.98413734e-03 7.02470362e-01 -1.10291135e+00 3.12824905e-01
6.89830661e-01 1.16263974e+00 -6.16473436e-01 3.13964933e-01
3.69205743e-01 9.28694785e-01 -5.22581637e-01 -8.73439670e-01
9.15924728e-01 -6.81564808e-02 -4.99525458e-01 2.14074865e-01
5.13604939e-01 -8.98959994e-01 -1.11582530e+00 1.01801753e+00
4.31614488e-01 9.88489389e-02 4.82004210e-02 -1.63674188e+00
-6.65811062e-01 3.91001284e-01 -8.02640259e-01 4.39435504e-02
2.10575417e-01 7.73661494e-01 7.58226395e-01 -4.54696536e-01
3.77175331e-01 1.12482309e+00 7.56379545e-01 7.08396912e-01
-1.08227539e+00 -8.03391933e-01 4.40599740e-01 2.24735200e-01
-1.32193267e+00 -6.78734481e-01 6.70258343e-01 -2.62557507e-01
6.88163221e-01 5.81386507e-01 5.33451080e-01 1.43656325e+00
5.62921241e-02 9.91370261e-01 9.03786421e-01 1.21089198e-01
2.09846139e-01 -1.24720633e-01 -1.43159717e-01 4.64281738e-01
2.48199239e-01 1.44101426e-01 -5.50539255e-01 -4.29209888e-01
8.49785388e-01 1.69626370e-01 -6.50183260e-01 -6.44250751e-01
-1.11328423e+00 6.61071420e-01 1.42568007e-01 -3.14302504e-01
-1.64545506e-01 2.94893593e-01 7.95215964e-01 4.22236681e-01
1.84637055e-01 2.40412369e-01 -3.48405153e-01 -1.17832117e-01
-4.40904558e-01 2.99020559e-01 6.19128764e-01 1.08939707e+00
3.79888862e-01 -1.14990793e-01 -6.36267483e-01 2.50376195e-01
1.33399041e-02 2.52444148e-01 3.60167801e-01 -1.14356661e+00
9.51074600e-01 3.21978986e-01 6.52627572e-02 -1.44264221e+00
8.97588730e-02 -1.66737270e-02 -9.58676279e-01 1.29514024e-01
5.05608439e-01 -2.58785665e-01 -6.79410815e-01 1.92127359e+00
3.39742035e-01 4.13382500e-01 1.03194453e-01 9.18845713e-01
3.63791555e-01 3.22532475e-01 2.01926142e-01 -1.63019344e-01
1.32529187e+00 -9.20595586e-01 -6.50373101e-01 -7.44863600e-02
6.07389033e-01 -2.40178928e-01 1.06458473e+00 2.27285564e-01
-9.22209084e-01 -1.81537524e-01 -8.81724834e-01 -3.61277908e-02
-1.64444104e-01 5.38656227e-02 6.05690897e-01 8.94362152e-01
-7.14310825e-01 4.32538062e-01 -1.10213959e+00 -8.53652582e-02
9.90500510e-01 5.61521232e-01 -7.33115494e-01 1.07816100e-01
-1.22043169e+00 1.14286117e-01 2.54753351e-01 -5.65737262e-02
-8.28659773e-01 -7.39449024e-01 -9.17924821e-01 1.62084460e-01
7.29020357e-01 -5.83016217e-01 1.01994121e+00 -8.83310497e-01
-1.45522070e+00 7.90056765e-01 1.61600858e-01 -9.29340184e-01
1.18183219e+00 -2.23748952e-01 -3.99635583e-01 3.76618564e-01
-1.39287010e-01 4.23234314e-01 1.09432995e+00 -9.58659410e-01
-6.71089828e-01 -4.91876960e-01 4.41332161e-01 1.45067602e-01
-9.44189727e-01 -2.35372055e-02 -7.35179126e-01 -1.12919152e+00
-2.38439485e-01 -8.97999108e-01 -3.56538445e-01 6.26665294e-01
-6.12755418e-01 3.10180128e-01 1.14423347e+00 -6.94175661e-01
1.24896681e+00 -2.69106817e+00 -1.58993199e-01 1.40418082e-01
5.80196619e-01 6.09904766e-01 -1.93741679e-01 1.66675985e-01
-1.78030804e-01 3.04983944e-01 -2.54497051e-01 -4.58877236e-01
4.59690429e-02 1.10021748e-01 -6.87911928e-01 8.19547534e-01
1.83847770e-01 8.67461085e-01 -8.86021495e-01 -3.67479116e-01
2.86650091e-01 3.98907840e-01 -7.50408590e-01 2.43899859e-02
-3.93027291e-02 5.73592901e-01 -7.89690495e-01 6.54196024e-01
9.22355890e-01 6.40067384e-02 2.67984450e-01 -2.11897776e-01
1.52760208e-01 -6.89564720e-02 -9.70172822e-01 1.45012915e+00
-6.90779388e-02 4.70276564e-01 5.32055534e-02 -8.91690850e-01
4.87111181e-01 2.50217319e-01 7.63042629e-01 -5.52286088e-01
1.08401597e-01 -1.60625666e-01 -4.05712575e-01 -4.68559235e-01
3.72703344e-01 3.35992485e-01 -2.37550706e-01 3.04969877e-01
-4.57111955e-01 5.56775451e-01 -2.44113579e-01 4.37248051e-02
1.49550772e+00 3.35827805e-02 2.92410314e-01 2.25304887e-01
6.15174770e-01 -4.70078468e-01 1.06578052e+00 7.93697059e-01
-7.51118660e-01 6.19526029e-01 9.27400768e-01 -6.07316315e-01
-7.34749675e-01 -9.30964112e-01 1.92323878e-01 7.00267613e-01
3.31317902e-01 -5.46756089e-01 -7.87107825e-01 -1.24608302e+00
3.05423290e-01 3.62952530e-01 -5.86853802e-01 -4.86325771e-01
-8.21504831e-01 -4.12932456e-01 1.03638887e+00 5.32512307e-01
9.49768066e-01 -6.99813902e-01 -7.23806262e-01 -4.74225767e-02
-2.60630220e-01 -1.48254573e+00 -9.42753971e-01 -2.85833955e-01
-5.88317692e-01 -1.19652379e+00 -5.25594950e-01 -2.59459287e-01
6.62356675e-01 4.43586826e-01 5.75702488e-01 -7.22364932e-02
-9.93803069e-02 4.63339657e-01 -3.80022734e-01 -1.09041356e-01
-1.48357183e-01 -6.84963837e-02 8.94755274e-02 5.27779698e-01
3.15117061e-01 -7.59873629e-01 -7.34634697e-01 4.76582676e-01
-1.01461244e+00 -2.95845687e-01 4.00735527e-01 6.62498474e-01
7.81375229e-01 3.19793373e-01 -3.74507196e-02 -7.38770962e-01
4.13097441e-01 -3.39754552e-01 -6.73528135e-01 2.14074418e-01
-9.12272260e-02 -3.72415006e-01 8.73288274e-01 -5.48180521e-01
-7.86819160e-01 3.53513896e-01 4.05256748e-02 -1.09524083e+00
-1.03416048e-01 -9.55985412e-02 -7.56447434e-01 -4.55472797e-01
3.16063523e-01 3.99820507e-01 1.38318852e-01 -2.82700807e-01
3.11425835e-01 4.01228309e-01 7.55641937e-01 -4.45594549e-01
9.97094214e-01 8.34479928e-01 6.49615303e-02 -6.15982890e-01
-4.03468072e-01 -1.25924304e-01 -3.36732984e-01 1.37928948e-01
7.69926786e-01 -1.04514229e+00 -1.10964155e+00 8.61894011e-01
-1.01075923e+00 -1.41064063e-01 -5.19718111e-01 2.50221103e-01
-7.16307461e-01 1.01137757e+00 -4.70259786e-01 -6.26535058e-01
-1.83740944e-01 -1.21768844e+00 9.33085918e-01 3.99745554e-02
2.97304820e-02 -6.07384861e-01 -3.53284270e-01 4.10970122e-01
3.87940779e-02 6.35457456e-01 5.88001251e-01 -5.83668649e-01
-8.34404588e-01 -3.76603007e-01 -1.85624093e-01 5.04814863e-01
2.43472561e-01 -1.44977167e-01 -9.25482571e-01 -6.03134990e-01
1.80877745e-01 -3.27870995e-01 8.84244502e-01 9.50078964e-02
1.99079955e+00 -9.74186957e-01 -2.82711715e-01 1.23216736e+00
1.11086440e+00 1.19571798e-01 9.40466940e-01 4.27181244e-01
9.16589856e-01 2.88268209e-01 6.32797003e-01 8.63478065e-01
2.12802261e-01 7.16599822e-01 7.28287220e-01 -9.81088076e-03
1.07351273e-01 -4.80769902e-01 5.71962893e-01 5.16644400e-03
1.17797442e-02 -5.27968585e-01 -4.34269100e-01 3.54025602e-01
-2.03989601e+00 -1.03261983e+00 2.36129880e-01 2.42716265e+00
6.70022905e-01 1.05135189e-02 3.83821428e-01 9.36106369e-02
6.41427934e-01 6.04184210e-01 -7.94083178e-01 4.03749682e-02
-2.58231610e-01 -1.56817570e-01 1.04650104e+00 7.65589764e-03
-1.56115472e+00 9.57318425e-01 5.22978401e+00 8.43484342e-01
-1.03034186e+00 9.22118127e-02 7.44793773e-01 -3.17607611e-01
-1.69602334e-01 -6.41423464e-02 -4.98202026e-01 6.64518893e-01
4.36587334e-01 -6.11965768e-02 4.40235376e-01 7.61004806e-01
1.99037686e-01 4.67548132e-01 -1.03993654e+00 1.14738250e+00
-1.61206543e-01 -1.35131896e+00 9.86658558e-02 2.84396499e-01
3.24679881e-01 -3.40968579e-01 3.09173763e-01 1.89378541e-02
2.93941617e-01 -7.72016048e-01 7.11670101e-01 2.40931451e-01
9.32562053e-01 -9.29942667e-01 3.81198585e-01 8.13309625e-02
-1.05950642e+00 -4.09455329e-01 -3.33566368e-01 1.75893471e-01
5.15613444e-02 2.05006465e-01 -1.17083877e-01 6.58753157e-01
8.27595353e-01 7.83282280e-01 -3.92336577e-01 8.98575723e-01
-3.66822422e-01 4.79533434e-01 -3.04754138e-01 4.98954237e-01
2.73386627e-01 1.54441688e-02 7.79482663e-01 8.81079733e-01
1.36208221e-01 2.38108590e-01 4.42827754e-02 5.69131672e-01
-3.43176514e-01 -1.02654941e-01 -7.26118863e-01 -1.83855146e-01
5.82713366e-01 8.71384501e-01 -2.54387826e-01 1.60331756e-01
-5.09819925e-01 1.32539856e+00 1.72204524e-01 3.77124697e-01
-1.16853702e+00 -3.74421179e-01 1.41477501e+00 1.43943012e-01
5.26257157e-01 -1.51654646e-01 -1.65930420e-01 -1.51870251e+00
5.07579148e-01 -1.30250192e+00 5.67480505e-01 8.52481648e-03
-1.17795551e+00 2.60557622e-01 -1.95967004e-01 -1.51033819e+00
2.35028803e-01 -3.23906869e-01 -4.66234773e-01 3.90163392e-01
-1.36831653e+00 -1.18442774e+00 -2.35712826e-01 1.05426276e+00
1.46211013e-01 -2.02918306e-01 6.57788455e-01 4.37243551e-01
-8.34503114e-01 1.32540047e+00 3.68925817e-02 5.56604862e-01
5.44494033e-01 -7.52708793e-01 7.62642801e-01 1.10646594e+00
-1.14152562e-02 5.03531635e-01 3.34237844e-01 -5.25271177e-01
-1.63264477e+00 -1.51271796e+00 4.35262650e-01 -5.01026511e-01
6.11422360e-01 -6.54560924e-01 -8.73259485e-01 9.08094764e-01
-2.01769292e-01 5.21715462e-01 6.22010827e-01 -4.69335198e-01
-6.02518678e-01 -2.44874865e-01 -1.36052597e+00 8.93166244e-01
1.26838815e+00 -6.29728079e-01 -1.84262037e-01 4.65102166e-01
9.64170456e-01 -5.33839166e-01 -7.38468170e-01 3.92048031e-01
6.51373804e-01 -1.19475305e+00 1.20725620e+00 -8.34287107e-01
1.92206666e-01 -3.85682106e-01 -1.73267350e-01 -7.11976349e-01
-1.68951541e-01 -1.36759830e+00 -5.11527479e-01 1.32744372e+00
-1.17663845e-01 -9.00142968e-01 1.17800403e+00 8.44224334e-01
2.38460392e-01 -6.06464982e-01 -1.17355847e+00 -1.11943591e+00
-1.78461328e-01 -3.72310251e-01 1.03705800e+00 1.05815768e+00
-4.00778472e-01 -4.56308335e-01 -9.32267427e-01 5.09846509e-01
7.27579594e-01 -3.40671360e-01 1.07621014e+00 -4.82743263e-01
-4.65584785e-01 -2.44418353e-01 -7.65429974e-01 -1.13998652e+00
3.39105725e-01 -3.36032599e-01 -1.70905307e-01 -6.59344554e-01
-3.96426506e-02 -3.16076487e-01 -4.27562982e-01 8.40957582e-01
-4.12507057e-02 1.29151180e-01 3.36735755e-01 3.50758970e-01
-4.65885937e-01 7.61278212e-01 1.15412903e+00 -1.99698985e-01
-1.36266917e-01 2.46831432e-01 -8.28168154e-01 7.27891207e-01
7.43278503e-01 -4.74487484e-01 -7.04467297e-01 -6.56340599e-01
-3.31554025e-01 -2.84632593e-02 7.49083459e-01 -1.01408529e+00
8.69843364e-02 -3.75588417e-01 1.15077689e-01 -1.41809344e-01
2.99405813e-01 -1.14970791e+00 9.83758047e-02 5.23663819e-01
-3.13044310e-01 -4.44005691e-02 2.45960131e-01 9.98661697e-01
-1.45080283e-01 3.11021060e-01 8.11680555e-01 2.33000219e-02
-7.81853557e-01 8.98972392e-01 -1.92102343e-01 1.37536481e-01
1.50005925e+00 -4.91036147e-01 -4.78206545e-01 -4.62342352e-01
-4.30758238e-01 4.25286829e-01 8.63322079e-01 4.73710448e-01
5.73979259e-01 -1.26588321e+00 -4.56511945e-01 3.88597220e-01
2.32360601e-01 -1.39874190e-01 4.44098502e-01 8.92263114e-01
-3.93219709e-01 1.93257153e-01 -2.81550318e-01 -1.94781065e-01
-1.32967722e+00 8.45529020e-01 3.52272689e-01 -2.00920030e-01
-9.79098678e-01 8.47430527e-01 5.45252442e-01 -7.45522529e-02
6.31852627e-01 -2.07594603e-01 2.02416196e-01 -2.53130168e-01
6.44031882e-01 2.72879094e-01 -1.55064642e-01 -6.24526799e-01
-4.93812412e-01 2.94364840e-01 -1.76416561e-01 2.76704371e-01
1.11924601e+00 -2.23133355e-01 1.36740610e-01 -4.59241807e-01
1.27102876e+00 1.09455660e-01 -1.82582784e+00 -1.87629074e-01
-2.67532974e-01 -1.18975353e+00 -1.64961949e-01 -3.52614284e-01
-1.55834448e+00 6.50803745e-01 4.74183351e-01 7.34228492e-02
1.40299451e+00 -4.64274764e-01 1.13312876e+00 3.78219783e-01
2.59251237e-01 -8.76863360e-01 -4.33944911e-02 1.60393596e-01
5.96392989e-01 -9.87622917e-01 -1.88742764e-02 -6.84667230e-01
-7.10225582e-01 7.79265881e-01 7.19987094e-01 5.84291816e-02
4.37268645e-01 1.40751213e-01 -1.33739009e-01 2.16813281e-01
-4.64776903e-01 2.19463974e-01 -1.43868431e-01 8.02042663e-01
-2.67720163e-01 -8.66835341e-02 -1.28381476e-01 8.03370655e-01
2.11008951e-01 5.45185953e-02 4.29923147e-01 1.14167738e+00
2.63322324e-01 -1.08596146e+00 -1.61120117e-01 2.45047122e-01
-6.73800766e-01 9.98622626e-02 -4.47598785e-01 7.68502474e-01
1.94933459e-01 6.85410917e-01 -9.58801135e-02 -6.12050533e-01
4.40940261e-01 -4.34948564e-01 2.46097237e-01 -2.49028191e-01
-5.49235165e-01 -2.77169287e-01 1.51154790e-02 -1.10481119e+00
-1.01621121e-01 -8.73978257e-01 -8.48379791e-01 -6.47096395e-01
1.19438812e-01 -2.31709555e-01 -5.17348235e-04 6.97742641e-01
7.39611924e-01 1.94595233e-01 8.71484637e-01 -3.99748772e-01
-1.00947511e+00 -5.30259497e-02 -6.14477038e-01 4.70017552e-01
5.52033424e-01 -3.98233145e-01 -3.60926360e-01 3.20761949e-02]
|
[5.83485221862793, 6.745745658874512]
|
c32ffd4f-67b1-40d0-b557-f5f7c86c5427
|
conrpg-paraphrase-generation-using-contexts
|
2109.00363
| null |
https://arxiv.org/abs/2109.00363v1
|
https://arxiv.org/pdf/2109.00363v1.pdf
|
ConRPG: Paraphrase Generation using Contexts as Regularizer
|
A long-standing issue with paraphrase generation is how to obtain reliable supervision signals. In this paper, we propose an unsupervised paradigm for paraphrase generation based on the assumption that the probabilities of generating two sentences with the same meaning given the same context should be the same. Inspired by this fundamental idea, we propose a pipelined system which consists of paraphrase candidate generation based on contextual language models, candidate filtering using scoring functions, and paraphrase model training based on the selected candidates. The proposed paradigm offers merits over existing paraphrase generation methods: (1) using the context regularizer on meanings, the model is able to generate massive amounts of high-quality paraphrase pairs; and (2) using human-interpretable scoring functions to select paraphrase pairs from candidates, the proposed framework provides a channel for developers to intervene with the data generation process, leading to a more controllable model. Experimental results across different tasks and datasets demonstrate that the effectiveness of the proposed model in both supervised and unsupervised setups.
|
['Jiwei Li', 'Chun Fan', 'Fei Wu', 'Qinghong Han', 'Xiaofei Sun', 'Qing He', 'Xiang Ao', 'Yuxian Meng']
|
2021-09-01
| null |
https://aclanthology.org/2021.emnlp-main.199
|
https://aclanthology.org/2021.emnlp-main.199.pdf
|
emnlp-2021-11
|
['paraphrase-generation', 'paraphrase-generation']
|
['computer-code', 'natural-language-processing']
|
[ 5.75306118e-01 1.61792368e-01 -1.65918663e-01 -5.82999170e-01
-7.52448380e-01 -5.12213767e-01 6.98813796e-01 2.54941225e-01
-5.36889210e-02 7.30187416e-01 4.00505245e-01 -1.22838564e-01
-1.51568636e-01 -8.86017263e-01 -8.90723526e-01 -2.95872688e-01
6.11925900e-01 3.19682479e-01 1.01764016e-01 -2.62408733e-01
7.39179790e-01 8.41787644e-03 -1.87103999e+00 6.26076758e-01
1.51953471e+00 7.16666818e-01 6.76123679e-01 2.63953060e-01
-3.05882841e-01 7.89283693e-01 -5.68278968e-01 -5.98579824e-01
8.11553970e-02 -8.07814538e-01 -8.78323495e-01 -2.77732462e-02
2.46803924e-01 7.39901215e-02 3.07850718e-01 1.04767001e+00
2.47592270e-01 -1.65078267e-01 5.99451482e-01 -9.45589781e-01
-8.60736787e-01 9.58997726e-01 -1.43431768e-01 -3.08231935e-02
8.75564456e-01 2.26708502e-02 1.42047358e+00 -1.13278091e+00
5.80292702e-01 1.10670745e+00 3.33392113e-01 6.08074605e-01
-1.23142350e+00 -4.50339317e-01 1.94077264e-04 2.80164152e-01
-1.10424399e+00 -4.26936120e-01 9.60114837e-01 -4.55360800e-01
7.91936100e-01 3.02313298e-01 6.78874373e-01 1.34257150e+00
1.50043219e-01 7.34065056e-01 1.15882611e+00 -8.56598020e-01
3.87407064e-01 5.68095267e-01 2.91380972e-01 4.16781217e-01
3.40551764e-01 -3.15026313e-01 -6.91806674e-01 -3.47535491e-01
4.00156707e-01 4.33196779e-03 -2.65898347e-01 -3.65187913e-01
-1.17218900e+00 8.56927514e-01 1.51106596e-01 3.75411540e-01
-2.80181676e-01 -1.76733941e-01 2.32078001e-01 4.77276206e-01
3.22229087e-01 9.59964871e-01 -1.58917308e-01 -8.13876763e-02
-1.08817053e+00 3.56192380e-01 7.53945947e-01 1.26676726e+00
8.70059073e-01 -3.31487864e-01 -5.70882916e-01 1.03822768e+00
3.52808446e-01 5.24365067e-01 8.12433720e-01 -6.39416039e-01
8.23947132e-01 8.96335840e-01 4.75553453e-01 -8.46732497e-01
2.01553240e-01 -2.92089313e-01 -4.76549208e-01 -5.92799820e-02
9.24107656e-02 -1.15143526e-02 -6.33828580e-01 1.79095328e+00
-1.00038528e-01 6.99568689e-02 1.90774977e-01 5.87881148e-01
6.09674156e-01 5.34903109e-01 -1.39020368e-01 -2.32915029e-01
1.20493627e+00 -9.03169155e-01 -5.79380751e-01 -2.72115678e-01
3.46186370e-01 -7.82082021e-01 1.62202656e+00 1.99095488e-01
-1.16211367e+00 -7.82295585e-01 -1.23959732e+00 6.36100248e-02
-6.99913409e-03 3.54190052e-01 3.70737195e-01 5.36829591e-01
-9.79887724e-01 7.24877119e-01 -5.15610993e-01 -5.56152761e-01
1.83789968e-01 9.30453539e-02 -2.60202680e-02 3.49649489e-02
-1.24002123e+00 7.68748045e-01 4.80144978e-01 3.83915752e-02
-7.86352277e-01 -3.88719827e-01 -7.41044998e-01 3.32644761e-01
9.73437354e-02 -1.09411466e+00 1.12801445e+00 -1.32156956e+00
-1.68033910e+00 7.47577190e-01 -3.84856582e-01 -4.47482198e-01
3.27623576e-01 -4.38215047e-01 -2.11933367e-02 8.52700174e-02
3.04801464e-01 3.49332392e-01 1.01083899e+00 -1.09574342e+00
-5.50928235e-01 -8.71241465e-02 5.15750870e-02 3.66423249e-01
-6.17055833e-01 5.03337309e-02 -1.41917169e-01 -5.63997924e-01
-6.66323453e-02 -8.39950681e-01 -1.80484012e-01 -3.90658110e-01
-5.52526176e-01 -3.30866128e-01 3.75843823e-01 -6.03343666e-01
1.36141622e+00 -1.77469742e+00 3.82047683e-01 1.93598434e-01
-5.83877638e-02 1.73859805e-01 -9.53053981e-02 7.31806576e-01
-1.05912425e-01 7.51045123e-02 -2.83620775e-01 -4.12379891e-01
1.15180649e-02 -1.16435565e-01 -6.52306557e-01 -2.95802712e-01
4.21545774e-01 8.80088031e-01 -1.21037114e+00 -4.61581916e-01
-2.85278298e-02 -2.56898463e-01 -5.64549387e-01 7.18769014e-01
-3.40524405e-01 3.20373178e-01 -6.39954507e-01 2.75997490e-01
3.50588620e-01 -2.72687703e-01 2.32718065e-01 1.60587296e-01
1.14951335e-01 6.17066622e-01 -9.89208102e-01 1.83016443e+00
-8.55814755e-01 2.19518095e-01 -3.78318846e-01 -9.54334438e-01
1.24839234e+00 2.51411319e-01 7.53439497e-03 -3.66383612e-01
-2.28001326e-01 4.22724038e-01 -2.41959944e-01 -7.47894526e-01
5.08768559e-01 -7.93974176e-02 -2.53427565e-01 7.61240184e-01
4.87908684e-02 -4.21779990e-01 4.96657640e-01 3.15788478e-01
1.04654479e+00 1.94244996e-01 3.39554787e-01 -1.91112995e-01
8.35104346e-01 -4.72867526e-02 3.96811396e-01 9.89461422e-01
3.40010434e-01 6.25403941e-01 4.78415728e-01 -2.66315073e-01
-1.05454910e+00 -1.06428325e+00 1.72640413e-01 7.38700569e-01
2.27910176e-01 -5.04108548e-01 -8.29151750e-01 -7.55592287e-01
-3.29142869e-01 9.90520060e-01 -5.00257611e-01 -3.18796277e-01
-5.60104609e-01 -3.98909867e-01 2.76621819e-01 3.03424299e-01
2.98602045e-01 -1.47213662e+00 -4.54389602e-01 2.05813810e-01
-4.81210351e-01 -8.17714989e-01 -3.90370399e-01 -8.13677460e-02
-9.47682202e-01 -8.93838227e-01 -6.11531913e-01 -9.69639957e-01
9.41780448e-01 3.21173847e-01 1.23583913e+00 1.47317499e-01
2.42732823e-01 1.64718777e-01 -6.70513391e-01 -2.63268560e-01
-1.04688919e+00 3.21200222e-01 -2.17896104e-01 1.22733399e-01
3.95015448e-01 -8.83511066e-01 -5.03610671e-01 1.09826677e-01
-8.21978569e-01 5.06808996e-01 7.58721769e-01 1.00241399e+00
4.08354998e-01 -3.83558899e-01 1.07468891e+00 -1.07235050e+00
1.22637689e+00 -5.61717093e-01 -3.94579649e-01 7.31489718e-01
-8.49820375e-01 3.44841838e-01 1.15626478e+00 -3.05273771e-01
-1.31889355e+00 4.07291278e-02 1.36559114e-01 -9.80987921e-02
-1.25545532e-01 3.91939044e-01 -1.78406745e-01 3.83406132e-01
8.37891519e-01 6.94400728e-01 -3.61586511e-02 -4.99099910e-01
6.79187536e-01 9.35335875e-01 1.87090516e-01 -6.99189544e-01
8.28292847e-01 2.19235346e-01 -4.48072523e-01 -4.64774489e-01
-7.66380847e-01 -4.09537137e-01 -5.26202679e-01 -2.71670781e-02
7.00478971e-01 -8.57400358e-01 6.08876459e-02 1.38542980e-01
-1.41041815e+00 1.41418755e-01 -4.12723273e-01 2.14520693e-01
-6.45971119e-01 6.58363104e-01 -3.85616481e-01 -6.45618379e-01
-7.82471776e-01 -1.16238165e+00 1.15588593e+00 3.16623747e-01
-5.25817335e-01 -5.88486373e-01 1.97020769e-01 5.98860323e-01
3.95009428e-01 -7.78178275e-02 1.21697044e+00 -7.69597530e-01
-6.66465878e-01 -1.97776645e-01 -5.05071357e-02 5.42284966e-01
4.57268894e-01 -2.57801320e-02 -8.18414032e-01 -9.98063385e-02
2.14915425e-01 -6.23335421e-01 5.97391784e-01 4.93349333e-04
1.04459417e+00 -3.42668325e-01 -2.50389636e-01 2.19239131e-01
1.11193252e+00 -5.31859621e-02 6.94783211e-01 1.43233642e-01
4.36291218e-01 7.09303439e-01 8.22961867e-01 5.03895342e-01
2.19796196e-01 7.68982053e-01 1.40624762e-01 1.31853983e-01
3.81846428e-02 -7.63920903e-01 5.14283001e-01 1.05693650e+00
9.43904296e-02 -8.17925557e-02 -4.01728988e-01 7.01692104e-01
-2.02583027e+00 -1.14583147e+00 -9.15677398e-02 2.35111189e+00
1.17853105e+00 2.80701160e-01 -3.80465053e-02 1.53398607e-02
7.79447734e-01 2.62341220e-02 -2.81967342e-01 -5.07782876e-01
1.13851875e-01 5.00909686e-01 -2.43671477e-01 3.45906168e-01
-5.87214470e-01 8.97734225e-01 5.76551104e+00 9.01106775e-01
-8.52694333e-01 -1.48380801e-01 4.11307156e-01 8.62221345e-02
-7.87112772e-01 4.32260424e-01 -7.94764936e-01 8.32304537e-01
6.15696073e-01 -6.48065448e-01 4.24983531e-01 9.98411715e-01
6.64960563e-01 6.12702407e-02 -1.33135033e+00 6.95839703e-01
2.31150031e-01 -1.22339511e+00 5.35264850e-01 -4.44748878e-01
7.96359420e-01 -3.90945703e-01 -2.08895460e-01 2.37643629e-01
2.58045137e-01 -8.08401823e-01 7.77138114e-01 5.99809825e-01
5.07951379e-01 -5.05888045e-01 6.34246528e-01 6.54162765e-01
-1.03217947e+00 -1.53064415e-01 -6.37792706e-01 -1.26686662e-01
8.69468078e-02 7.01197326e-01 -1.15551269e+00 7.83635974e-01
1.64259031e-01 8.62689734e-01 -8.40385139e-01 9.66806889e-01
-8.03022563e-01 6.58983111e-01 6.23102933e-02 -4.95643944e-01
-2.50517905e-01 -4.95943487e-01 4.11896914e-01 1.19203866e+00
4.93583173e-01 -4.28331256e-01 -7.78180659e-02 1.37776923e+00
2.45120488e-02 3.99572521e-01 -7.18646944e-01 5.69675826e-02
7.20067322e-01 1.24791956e+00 -3.96528333e-01 -4.37153578e-01
-3.09671640e-01 1.09647131e+00 4.94993001e-01 2.49804363e-01
-7.64097810e-01 -6.27717257e-01 1.44489512e-01 1.34899795e-01
1.64839610e-01 2.12105379e-01 -4.46801037e-01 -1.35814285e+00
5.34646332e-01 -9.81108606e-01 7.23499358e-02 -8.55260670e-01
-1.36363375e+00 6.50568485e-01 1.30192772e-01 -1.51734436e+00
-4.43072140e-01 -1.88428164e-01 -1.19555891e+00 9.49640810e-01
-1.39685404e+00 -1.16922724e+00 -4.55876082e-01 3.55006456e-01
9.30651724e-01 -3.07541758e-01 8.08509171e-01 -7.34176785e-02
-4.06336844e-01 4.91949201e-01 -1.96637452e-01 -2.99736798e-01
7.05528617e-01 -1.37836969e+00 5.97548485e-01 9.75404263e-01
3.87204707e-01 1.25143743e+00 6.36845231e-01 -6.74668372e-01
-1.08190858e+00 -1.04665649e+00 1.22489333e+00 -4.44304615e-01
6.29427314e-01 -5.98317385e-01 -7.62934148e-01 3.01946193e-01
4.08466041e-01 -6.60091877e-01 6.65317774e-01 2.89688185e-02
-2.21703812e-01 -2.32267365e-01 -8.75911832e-01 6.98442876e-01
1.11524415e+00 -5.40351272e-01 -1.20715356e+00 5.65605044e-01
7.94042408e-01 -2.92754848e-03 -4.52093482e-01 6.57506362e-02
4.10588652e-01 -1.06880033e+00 6.68253005e-01 -5.25343597e-01
1.04284322e+00 -2.05585510e-01 1.95132315e-01 -1.37003028e+00
-2.23605707e-01 -6.95748031e-01 8.79428312e-02 1.44641864e+00
7.86195517e-01 -4.23051268e-01 6.56481087e-01 5.66080213e-01
-7.50630572e-02 -7.95181692e-01 -6.96244717e-01 -7.29511201e-01
-2.27754280e-01 -2.38067601e-02 6.75746322e-01 6.75228000e-01
4.78165984e-01 7.65830338e-01 -4.38102692e-01 -5.11843190e-02
4.51366931e-01 6.13099158e-01 9.46254253e-01 -1.00404108e+00
-7.35245466e-01 -1.93588272e-01 5.91595024e-02 -1.40040350e+00
3.08681369e-01 -1.07136023e+00 2.08902881e-01 -1.55903018e+00
5.46601236e-01 -3.92112195e-01 -2.84206182e-01 2.17811078e-01
-5.47678351e-01 -3.15745533e-01 1.67651996e-02 4.19893742e-01
-5.72191656e-01 6.35945499e-01 9.74052310e-01 -1.62011143e-02
-3.11879098e-01 4.20847207e-01 -9.52835858e-01 5.54791152e-01
6.88062787e-01 -5.35860598e-01 -8.98809791e-01 -4.68176186e-01
3.72606963e-01 4.81050573e-02 2.16554001e-01 -7.71747649e-01
1.52647719e-02 -3.16334635e-01 -1.27560228e-01 -2.02427834e-01
8.10564607e-02 -6.63128674e-01 1.37515850e-02 3.20704937e-01
-7.18144536e-01 1.84518069e-01 -3.12436789e-01 6.17911935e-01
-3.20597321e-01 -9.31892157e-01 6.32352531e-01 -1.56300813e-01
-4.10300553e-01 -3.05503219e-01 -2.12172404e-01 2.29153588e-01
8.93115163e-01 -3.34776968e-01 -1.70677155e-01 -4.10462201e-01
-3.51827353e-01 8.12906921e-02 6.75046504e-01 7.46914327e-01
7.63803542e-01 -1.23182559e+00 -7.26259530e-01 2.56044596e-01
4.55963612e-01 -1.29779652e-01 -2.33672276e-01 2.61757016e-01
-2.41030157e-01 2.76572466e-01 -4.23931703e-02 -4.72689360e-01
-1.14944375e+00 4.51477766e-01 -8.59847665e-02 -6.51132405e-01
-2.50937998e-01 7.34482408e-01 8.96541774e-02 -2.86478847e-01
-1.30531400e-01 -5.34731567e-01 -3.73386264e-01 -2.13945493e-01
4.68068302e-01 1.47923697e-02 4.48052697e-02 -1.94846272e-01
-5.40065207e-02 4.31481510e-01 -2.91254103e-01 -1.39641777e-01
1.28546989e+00 1.11390720e-03 -1.52550265e-01 3.77051711e-01
6.92986071e-01 2.23361820e-01 -7.95617878e-01 -4.62513827e-02
3.51903439e-01 -4.77748990e-01 -6.31359220e-01 -6.25726342e-01
-4.29943472e-01 6.52301610e-01 2.00607553e-01 4.14506435e-01
1.09807467e+00 -5.51107414e-02 7.80853271e-01 4.08842534e-01
6.24799311e-01 -9.81665492e-01 4.67106760e-01 2.20082596e-01
9.99198258e-01 -1.08145225e+00 -8.81035551e-02 -5.89531243e-01
-7.91414380e-01 1.04609334e+00 5.74883044e-01 -2.44971484e-01
5.92885539e-03 -2.92738110e-01 -1.92369223e-01 -5.61544970e-02
-1.00350142e+00 1.35739475e-01 1.56453758e-01 4.04714316e-01
5.21375537e-01 -1.91499323e-01 -8.41049492e-01 8.27238560e-01
-2.74207801e-01 2.11080000e-01 7.15535462e-01 7.84959793e-01
-5.63169718e-01 -1.58730876e+00 -7.60910660e-02 5.65438271e-01
-1.71178207e-01 -4.25196111e-01 -6.32309973e-01 9.67552885e-02
3.35060246e-02 1.12036896e+00 -4.53198522e-01 -2.50064850e-01
4.59774286e-01 2.09905237e-01 3.45075399e-01 -1.30636978e+00
-7.25142360e-01 -4.05062139e-01 2.77953476e-01 -3.36944729e-01
-2.87950307e-01 -4.73429859e-01 -8.52720559e-01 4.69347060e-01
-6.47928596e-01 5.36486804e-01 3.75093073e-01 1.06168056e+00
6.87205493e-01 1.50744215e-01 1.15604436e+00 -5.32389164e-01
-1.04773748e+00 -1.14854324e+00 -3.02772075e-01 1.01420283e+00
-2.99186051e-01 -3.98878127e-01 -4.33783978e-01 3.59493613e-01]
|
[11.657265663146973, 9.288808822631836]
|
84858811-11ad-4069-a2c4-4201eb6be177
|
identifying-water-stress-in-chickpea-plant-by
|
2104.07911
| null |
https://arxiv.org/abs/2104.07911v3
|
https://arxiv.org/pdf/2104.07911v3.pdf
|
Intelligent Monitoring of Stress Induced by Water Deficiency in Plants using Deep Learning
|
In the recent decade, high-throughput plant phenotyping techniques, which combine non-invasive image analysis and machine learning, have been successfully applied to identify and quantify plant health and diseases. However, these techniques usually do not consider the progressive nature of plant stress and often require images showing severe signs of stress to ensure high confidence detection, thereby reducing the feasibility for early detection and recovery of plants under stress. To overcome the problem mentioned above, we propose a deep learning pipeline for the temporal analysis of the visual changes induced in the plant due to stress and apply it to the specific water stress identification case in Chickpea plant shoot images. For this, we have considered an image dataset of two chickpea varieties JG-62 and Pusa-372, under three water stress conditions; control, young seedling, and before flowering, captured over five months. We have employed a variant of Convolutional Neural Network - Long Short Term Memory (CNN-LSTM) network to learn spatio-temporal patterns from the chickpea plant dataset and use them for water stress classification. Our model has achieved ceiling level classification performance of 98.52% on JG-62 and 97.78% on Pusa-372 chickpea plant data and has outperformed the best reported time-invariant technique by at least 14% for both JG-62 and Pusa-372 species, to the best of our knowledge. Furthermore, our CNN-LSTM model has demonstrated robustness to noisy input, with a less than 2.5% dip in average model accuracy and a small standard deviation about the mean for both species. Lastly, we have performed an ablation study to analyze the performance of the CNN-LSTM model by decreasing the number of temporal session data used for training.
|
['Tapan K. Gandhi', 'Rohan Wadhawan', 'Shiva Azimi']
|
2021-04-16
| null | null | null | null |
['plant-phenotyping']
|
['computer-vision']
|
[ 4.33856875e-01 -3.26628745e-01 4.29783091e-02 1.76348209e-01
-8.51609334e-02 -8.39877605e-01 1.03871904e-01 6.51927650e-01
-1.36857644e-01 3.60853821e-01 -6.33666396e-01 -5.63759804e-01
-2.68046290e-01 -9.34184611e-01 -4.25610662e-01 -8.52709293e-01
-5.18800616e-01 -3.58407609e-02 3.06095421e-01 -1.99352056e-01
-9.34790298e-02 9.42982435e-01 -1.44692624e+00 2.35972509e-01
5.96921921e-01 9.22874093e-01 9.41672921e-01 7.66884327e-01
1.19616605e-01 1.67063594e-01 -7.84910977e-01 4.36499059e-01
-1.14764310e-01 -2.00632274e-01 -8.04605663e-01 2.94757634e-01
2.09360644e-01 -3.27341169e-01 8.21255296e-02 5.28833508e-01
7.26183593e-01 -2.48861119e-01 2.04833552e-01 -1.00978136e+00
-8.52847874e-01 3.91963214e-01 -8.39183807e-01 1.73836008e-01
-8.74072388e-02 4.29328352e-01 4.42158103e-01 -4.54141378e-01
6.81027174e-01 8.35327327e-01 8.76654148e-01 1.27968073e-01
-1.62871885e+00 -1.48958072e-01 1.03446327e-01 2.65314281e-01
-1.28815854e+00 -1.84848122e-02 2.11139515e-01 -5.70943236e-01
1.11942255e+00 3.74148451e-02 7.66021788e-01 8.22735846e-01
3.15386713e-01 5.54268003e-01 9.67196584e-01 -2.91751564e-01
2.94182450e-01 -4.60962355e-01 -2.42876127e-01 4.98620093e-01
-1.25472946e-02 1.94847509e-01 4.09856588e-02 1.86346561e-01
7.02728271e-01 2.60754302e-02 -4.08998460e-01 -1.61399961e-01
-9.15728569e-01 3.81517917e-01 8.70903254e-01 6.97892845e-01
-6.27515852e-01 -3.14999260e-02 8.16129804e-01 1.35451555e-01
3.32110763e-01 5.16953170e-01 -1.23762321e+00 1.15931034e-01
-9.54158306e-01 -5.03443666e-02 7.02166915e-01 3.72130573e-01
2.73991942e-01 3.47503066e-01 -2.09044963e-01 9.11504507e-01
-4.25969958e-02 6.11906230e-01 1.68223158e-01 -7.95700967e-01
-3.57443243e-01 5.51607788e-01 1.52887851e-01 -1.22321594e+00
-6.75696492e-01 -2.76326180e-01 -9.70988035e-01 4.66863930e-01
5.14247000e-01 -2.61818198e-03 -1.30602920e+00 1.83882213e+00
-7.80961663e-02 1.97809841e-02 -9.47338045e-02 6.38572872e-01
5.95922351e-01 8.73367846e-01 4.17599797e-01 -3.57068896e-01
1.31619322e+00 -1.90618902e-01 -5.11352718e-01 -1.97791576e-01
8.02968442e-01 -6.53067112e-01 1.07246161e+00 3.86368185e-01
-7.49793708e-01 -5.20768404e-01 -1.04828608e+00 4.88102615e-01
-8.83334160e-01 6.78806067e-01 5.80187380e-01 4.57328290e-01
-9.66198325e-01 1.06156743e+00 -9.17475998e-01 -1.14125121e+00
5.88819206e-01 3.02340716e-01 -6.62824988e-01 -3.23110237e-03
-8.50498319e-01 9.63745117e-01 5.47534108e-01 8.49388778e-01
-1.18352997e+00 -9.51877713e-01 -4.44722742e-01 5.35338402e-01
3.02334219e-01 3.72285582e-02 8.73553574e-01 -5.42050064e-01
-1.64184153e+00 1.10450304e+00 3.48566286e-02 -2.91747481e-01
-2.86669672e-01 -4.95629311e-02 -4.06950086e-01 4.04241495e-02
9.55729559e-02 6.32337570e-01 3.68845344e-01 -1.24795735e+00
-4.36418951e-01 -5.95568359e-01 -1.92666739e-01 -5.44966400e-01
-3.29185903e-01 2.53125373e-02 1.33950159e-01 -3.51458579e-01
3.79784435e-01 -1.20050561e+00 -2.68909872e-01 3.50441784e-01
-2.74909943e-01 4.82696414e-01 1.39044631e+00 -8.56871545e-01
6.30912423e-01 -2.10595083e+00 9.18451250e-02 -1.95530921e-01
-1.28369749e-01 1.13224435e+00 -6.25129998e-01 4.55757976e-01
-3.69963318e-01 3.97707701e-01 -4.69850123e-01 9.94469896e-02
-5.03594875e-01 4.39006954e-01 -2.12985188e-01 4.53543037e-01
5.11675835e-01 6.94605649e-01 -7.80091166e-01 1.09932333e-01
5.30094862e-01 5.32239914e-01 2.31575742e-01 1.45134807e-01
-8.96818638e-02 2.91913122e-01 1.10632047e-01 1.14520490e+00
1.01556396e+00 -2.60465052e-02 3.64416599e-01 -1.92628458e-01
-7.02359915e-01 -3.76049727e-01 -6.73606813e-01 1.52582085e+00
-5.04959822e-01 7.17546105e-01 3.97370756e-01 -1.10088110e+00
9.86925602e-01 5.58091342e-01 7.29968429e-01 -4.39381331e-01
-7.98479915e-02 3.04447830e-01 2.49055594e-01 -3.68965030e-01
-5.15723974e-02 3.49776477e-01 5.01948297e-01 -2.14685872e-01
2.60530680e-01 -7.29756728e-02 3.84919316e-01 -1.15003332e-01
1.23568296e+00 2.28562608e-01 8.67661610e-02 -4.37232524e-01
4.63610709e-01 4.75915186e-02 7.18960047e-01 4.33188587e-01
-8.73780310e-01 5.29456019e-01 6.48974955e-01 -4.18321133e-01
-8.34796906e-01 -8.13333333e-01 -2.75238544e-01 1.02581584e+00
-3.10484409e-01 -1.03555612e-01 -3.46312553e-01 -2.70820171e-01
1.60170525e-01 5.13689637e-01 -7.84768522e-01 -3.37423176e-01
-2.51829892e-01 -1.09510970e+00 7.55707741e-01 6.30954862e-01
7.34344065e-01 -1.43352580e+00 -1.02295649e+00 5.62957227e-01
1.21455595e-01 -1.27298152e+00 5.49245119e-01 8.08806598e-01
-7.43307710e-01 -8.70768130e-01 -8.26854169e-01 -6.07338428e-01
1.75224766e-01 3.36014539e-01 9.54218030e-01 -4.38717790e-02
-9.45658743e-01 -2.11229213e-02 -4.06226754e-01 -3.52274597e-01
-1.83647931e-01 3.10700923e-01 -3.81156176e-01 -5.55905342e-01
1.89311132e-01 -9.18745518e-01 -5.54074943e-01 2.56144017e-01
-6.59794688e-01 -3.39768410e-01 5.62159300e-01 1.06848121e+00
4.77908581e-01 -4.20821235e-02 6.13159716e-01 -3.77559781e-01
-6.26028106e-02 -4.29651082e-01 -8.48416865e-01 5.18261731e-01
-5.14101207e-01 -5.62965155e-01 6.20321274e-01 -3.78328174e-01
-7.33517945e-01 4.31745380e-01 1.49397135e-01 -2.25755543e-01
-7.34890163e-01 1.01619637e+00 -2.30794385e-01 -2.21369699e-01
7.44823456e-01 9.90420859e-03 -5.78721687e-02 -2.93560207e-01
1.74442768e-01 3.42687368e-01 7.15970755e-01 -1.67696089e-01
5.58758140e-01 4.18198615e-01 4.83239740e-01 -1.12950754e+00
-7.93029606e-01 -4.62049454e-01 -8.81166637e-01 -3.46863091e-01
8.82863283e-01 -4.99474585e-01 -1.17621577e+00 1.12683892e+00
-1.04614580e+00 -7.34350085e-01 -1.23330325e-01 2.08271876e-01
-2.87464708e-01 2.54437357e-01 -7.63508201e-01 -7.64898241e-01
-4.91669804e-01 -9.26619053e-01 1.03630757e+00 3.87557566e-01
2.95356829e-02 -9.52470005e-01 5.33078909e-02 -4.06721741e-01
7.54165530e-01 8.38253677e-01 8.94922674e-01 -7.27051198e-02
-3.38486843e-02 -5.28378904e-01 -7.43221343e-01 2.23546103e-01
5.17281890e-01 8.55612457e-01 -1.24848795e+00 -4.29436833e-01
-2.87088394e-01 -5.00580192e-01 7.81524718e-01 9.61899936e-01
1.11461306e+00 4.61850017e-01 -3.51747751e-01 6.15387797e-01
1.67019761e+00 4.34535295e-01 6.47707164e-01 3.83014947e-01
6.63532078e-01 6.98996902e-01 6.44466758e-01 5.38255632e-01
-3.52695525e-01 4.71489966e-01 1.06245017e+00 -4.65822577e-01
7.03858286e-02 3.67025793e-01 1.97443545e-01 2.10365444e-01
3.06404140e-02 -5.43553293e-01 -1.19537866e+00 9.08471167e-01
-1.62631977e+00 -1.04555953e+00 -5.04955351e-01 1.98401320e+00
4.67270195e-01 -9.24963728e-02 -1.32140383e-01 4.68586832e-01
5.27404070e-01 2.53116369e-01 -6.95813000e-01 -6.01709664e-01
-5.11373281e-01 2.15255305e-01 8.23429704e-01 3.61380689e-02
-1.30754530e+00 1.17926264e+00 6.54862309e+00 1.68848276e-01
-1.68533230e+00 -2.26277739e-01 4.87589657e-01 2.89217263e-01
5.56840897e-01 3.52868408e-01 -3.12423050e-01 3.13531458e-02
1.17436707e+00 2.44755968e-01 2.16923237e-01 6.41771317e-01
4.97439235e-01 -3.84401292e-01 -6.30243242e-01 2.91837066e-01
-4.20072258e-01 -9.28850055e-01 -5.40431857e-01 4.36768420e-02
3.78987908e-01 1.44339010e-01 2.22717077e-02 1.42523915e-01
2.05414847e-01 -9.96941864e-01 -1.76291796e-03 1.21338539e-01
7.33758807e-01 -3.56634021e-01 6.58201754e-01 1.76203892e-01
-1.30372930e+00 -3.13409358e-01 -5.66789865e-01 9.47483908e-03
1.70057975e-02 8.07675898e-01 -7.23667860e-01 6.32424474e-01
9.57596481e-01 8.63864064e-01 -8.33441675e-01 9.17581379e-01
1.05509177e-01 9.37703669e-01 -3.42956960e-01 4.92880732e-01
3.08723778e-01 8.40149000e-02 2.50201076e-01 1.06118524e+00
6.00438237e-01 -9.73083079e-02 3.56204987e-01 8.25657129e-01
4.70748782e-01 -1.10220388e-01 -5.51904798e-01 -4.57161397e-01
2.19158933e-01 1.58472800e+00 -1.16721988e+00 2.31886227e-02
-6.88469931e-02 1.16587245e+00 2.45870464e-02 2.48449236e-01
-4.68252301e-01 -4.23538804e-01 4.58153784e-01 -3.07188362e-01
5.80240846e-01 -3.86711955e-01 -2.18312055e-01 -6.95745766e-01
-3.04428339e-01 -5.96685112e-01 1.20600536e-01 -1.02644539e+00
-9.30484891e-01 3.70897412e-01 -2.31054172e-01 -6.47422314e-01
5.94387092e-02 -1.04696786e+00 -7.92634964e-01 1.24909306e+00
-1.14322162e+00 -1.40029132e+00 -7.86804557e-01 -5.46242781e-02
3.35878670e-01 8.44069868e-02 1.56112289e+00 1.58903152e-01
-1.08022070e+00 -1.26591315e-02 2.35379413e-01 -3.24932605e-01
5.55797219e-01 -1.29190266e+00 4.55150813e-01 1.14850628e+00
-3.41306776e-01 2.04067409e-01 5.77355206e-01 -6.29784405e-01
-1.12226248e+00 -1.15317190e+00 7.90516078e-01 7.80784786e-02
5.69314659e-01 -5.90667352e-02 -1.25758266e+00 4.92750973e-01
1.17629178e-01 2.89954573e-01 5.16783178e-01 1.20698310e-01
-1.42012492e-01 -1.90635234e-01 -1.37008512e+00 4.73273307e-01
5.66925764e-01 -5.26329637e-01 4.30262744e-01 4.66204286e-01
4.71156329e-01 -1.28278717e-01 -1.32384384e+00 8.92152190e-01
6.19019806e-01 -7.40514815e-01 8.38293016e-01 -3.89014661e-01
3.26388687e-01 -4.90524799e-01 -3.34713340e-01 -1.50479496e+00
-8.92885447e-01 -2.68361002e-01 1.38125613e-01 1.52663493e+00
1.70112416e-01 -2.22100243e-01 5.01147568e-01 6.92972764e-02
-1.04626484e-01 -4.74384874e-01 -3.96760315e-01 -9.25551653e-01
1.58343360e-01 -7.78806061e-02 3.24118286e-01 9.91217792e-01
-3.89153421e-01 -2.79798955e-01 -1.52211905e-01 5.67608714e-01
7.72656053e-02 9.15620849e-02 3.22536260e-01 -1.31966722e+00
7.10393954e-03 -4.29367363e-01 -5.70752740e-01 -1.17661379e-01
1.38486519e-01 -5.60231268e-01 2.22765148e-01 -1.45565498e+00
-6.60743415e-02 -8.52953121e-02 -3.68858397e-01 9.68311071e-01
-1.46448776e-01 -7.18080625e-02 4.36781831e-02 -1.02083944e-01
5.04199803e-01 1.08232819e-01 9.87242103e-01 -1.34016976e-01
-2.98038930e-01 3.40916775e-02 -2.59555519e-01 3.63497525e-01
1.22795796e+00 -3.21580052e-01 -4.48042870e-01 -5.30320704e-01
-1.81762561e-01 -2.91156843e-02 7.67375767e-01 -1.16306543e+00
-4.18615639e-01 -3.33402961e-01 5.09228170e-01 -6.38612151e-01
1.45567283e-01 -8.05450857e-01 9.78340060e-02 9.26209211e-01
-3.41178775e-01 1.43189609e-01 9.09387529e-01 1.53488457e-01
6.33578375e-02 -3.05832207e-01 9.05587018e-01 -9.78361517e-02
-9.42203462e-01 2.93739170e-01 -9.31920350e-01 -6.40821874e-01
1.00862813e+00 7.20175654e-02 -5.21712840e-01 2.52471142e-03
-9.26330030e-01 7.47896954e-02 1.44137427e-01 3.69838238e-01
1.19988307e-01 -9.67311800e-01 -6.79945469e-01 4.14428860e-02
2.00782821e-01 -4.91045833e-01 4.65152800e-01 8.74176979e-01
-9.35803235e-01 4.59862262e-01 -9.13504243e-01 -1.03164411e+00
-1.11303592e+00 7.40575492e-01 3.98927331e-01 -1.24443270e-01
-2.59254396e-01 6.15218818e-01 -1.57215580e-01 -5.18582880e-01
1.58822294e-02 -6.39955223e-01 -3.58884990e-01 1.35388896e-01
1.76378831e-01 2.90541112e-01 3.62421602e-01 -5.39866924e-01
-4.72363025e-01 1.93372548e-01 2.29486853e-01 1.88705996e-01
1.45265388e+00 1.18539087e-01 -2.13175640e-01 8.56943905e-01
8.89584064e-01 -7.13082075e-01 -1.04643464e+00 2.15791434e-01
3.37837964e-01 -3.68155777e-01 3.36471111e-01 -1.16024888e+00
-1.30624282e+00 9.68299568e-01 1.45788717e+00 7.32177913e-01
1.52471495e+00 -6.15468681e-01 5.22687435e-01 5.07659674e-01
-2.01460570e-01 -6.88198268e-01 -5.32451689e-01 6.12702787e-01
9.21773374e-01 -1.24721265e+00 -2.76880562e-01 -3.38965446e-01
-9.00706276e-02 1.09506488e+00 8.16209793e-01 5.73599432e-03
6.95286095e-01 4.10015255e-01 3.11441392e-01 -2.62755990e-01
-9.44056213e-01 -4.77352440e-01 -3.53890419e-01 1.13166404e+00
7.87265420e-01 1.34137183e-01 7.71471625e-03 -1.21726088e-01
3.50915492e-01 2.69701689e-01 5.80352187e-01 1.23472297e+00
-2.88012922e-01 -8.41857374e-01 -1.79521531e-01 1.37100071e-01
-4.11857426e-01 4.04915027e-02 -5.51056981e-01 7.80149817e-01
6.69722110e-02 8.59914362e-01 -8.01724717e-02 -1.99093133e-01
5.14496803e-01 1.73682105e-02 3.19444805e-01 -5.34586132e-01
-6.98790729e-01 4.30115848e-04 -3.96067500e-01 -4.98196542e-01
-3.96144241e-01 -7.75506496e-01 -7.50568509e-01 -4.61808980e-01
-5.85260153e-01 -5.56503832e-01 7.85963833e-01 8.61128509e-01
5.75504363e-01 9.06832159e-01 6.55746520e-01 -1.06919670e+00
-1.26233801e-01 -1.12209082e+00 -8.28185201e-01 4.79639880e-02
1.52079672e-01 -5.86839795e-01 -5.59136830e-02 1.52789742e-01]
|
[9.159419059753418, -1.570779800415039]
|
1b2c8393-09b0-487c-bb5f-40d50e1d7997
|
anticipatory-music-transformer
|
2306.08620
| null |
https://arxiv.org/abs/2306.08620v1
|
https://arxiv.org/pdf/2306.08620v1.pdf
|
Anticipatory Music Transformer
|
We introduce anticipation: a method for constructing a controllable generative model of a temporal point process (the event process) conditioned asynchronously on realizations of a second, correlated process (the control process). We achieve this by interleaving sequences of events and controls, such that controls appear following stopping times in the event sequence. This work is motivated by problems arising in the control of symbolic music generation. We focus on infilling control tasks, whereby the controls are a subset of the events themselves, and conditional generation completes a sequence of events given the fixed control events. We train anticipatory infilling models using the large and diverse Lakh MIDI music dataset. These models match the performance of autoregressive models for prompted music generation, with the additional capability to perform infilling control tasks, including accompaniment. Human evaluators report that an anticipatory model produces accompaniments with similar musicality to even music composed by humans over a 20-second clip.
|
['Percy Liang', 'Chris Donahue', 'David Hall', 'John Thickstun']
|
2023-06-14
| null | null | null | null |
['music-generation', 'music-generation']
|
['audio', 'music']
|
[ 6.62223577e-01 1.62617788e-01 1.72154844e-01 -7.31369555e-02
-8.43868077e-01 -9.21631098e-01 1.11047518e+00 -1.05080627e-01
-3.59069481e-02 6.12346113e-01 8.73458683e-01 1.67425573e-01
-1.80367202e-01 -5.84834158e-01 -7.34978139e-01 -6.11778080e-01
-3.48727137e-01 8.23617816e-01 -2.44229689e-01 -1.36056721e-01
1.74039096e-01 2.42276207e-01 -1.45845854e+00 5.14907420e-01
3.15142572e-01 3.60716373e-01 2.48818710e-01 1.23145187e+00
2.47300714e-01 1.25276339e+00 -8.56395543e-01 5.70362695e-02
7.02247396e-02 -9.27894175e-01 -3.14640164e-01 7.18166977e-02
-8.26402605e-02 5.67078255e-02 -1.94078282e-01 3.83972079e-01
5.10210872e-01 5.21660388e-01 8.15474749e-01 -1.48556578e+00
-5.01805782e-01 1.34736121e+00 -1.10285267e-01 -3.00349861e-01
7.23624706e-01 2.81722307e-01 1.08942127e+00 -5.85130870e-01
8.49421144e-01 1.27541006e+00 6.42321885e-01 6.37339830e-01
-1.77586257e+00 -7.21986532e-01 3.70895937e-02 -4.49186951e-01
-1.26453614e+00 -7.66359031e-01 7.58795798e-01 -6.16887152e-01
7.68318176e-01 3.72738332e-01 1.00376403e+00 1.49801183e+00
1.90910682e-01 1.03777051e+00 6.64904475e-01 -4.98566806e-01
4.91090983e-01 -6.54537320e-01 -3.76935571e-01 1.21111413e-02
-4.87000644e-01 4.34391260e-01 -1.05826175e+00 -4.15459722e-01
9.96614814e-01 -2.21822262e-01 -2.25003377e-01 2.47981295e-01
-1.66834295e+00 6.08305275e-01 -2.59219408e-01 1.22426704e-01
-7.81917453e-01 7.66921580e-01 1.33353770e-01 3.04530978e-01
9.42304283e-02 9.10924315e-01 -4.06239957e-01 -5.46325743e-01
-1.26532781e+00 6.74685657e-01 1.12214577e+00 1.16633463e+00
1.20779961e-01 3.55110407e-01 -7.76516974e-01 2.47372076e-01
1.58212617e-01 4.69481230e-01 3.23175043e-01 -1.11410892e+00
8.05470571e-02 -2.43504420e-01 4.26273137e-01 -4.95679259e-01
1.95942223e-02 -3.10479701e-01 -7.03619063e-01 2.23656878e-01
5.50889492e-01 -3.37138653e-01 -9.52774346e-01 2.20367217e+00
3.00183855e-02 6.16202235e-01 -1.10366307e-01 3.88400018e-01
8.94043148e-02 9.25035954e-01 3.79145116e-01 -6.02763116e-01
8.38522315e-01 -5.60850382e-01 -7.91776240e-01 -3.56734693e-02
-3.03407311e-02 -1.09259963e+00 1.05393767e+00 7.14972317e-01
-1.52427626e+00 -7.70076096e-01 -8.09988379e-01 3.06569397e-01
6.37686789e-01 4.08550538e-02 7.62177169e-01 5.07112853e-02
-8.49569201e-01 8.80640447e-01 -1.02861404e+00 7.09995851e-02
-9.22144502e-02 9.05407965e-02 1.72562897e-02 7.28881180e-01
-1.02641141e+00 2.85294384e-01 2.15029240e-01 -1.18547842e-01
-1.59830403e+00 -9.55663919e-01 -4.26235110e-01 1.13258906e-01
1.82843417e-01 -5.45741141e-01 1.83724201e+00 -8.99455369e-01
-1.61797261e+00 6.11680567e-01 -2.10623935e-01 -5.68883061e-01
6.28518283e-01 -5.22555768e-01 -4.38480735e-01 -1.74547210e-01
2.24095598e-01 5.83229005e-01 1.26391470e+00 -9.87643301e-01
-4.48726505e-01 1.70746088e-01 -3.68998945e-01 7.36394674e-02
2.40000427e-01 1.80043191e-01 -3.71930093e-01 -1.16747808e+00
-1.25175506e-01 -1.32110131e+00 -4.46770847e-01 -6.12499297e-01
-7.98539519e-01 -1.67713821e-01 1.84402391e-01 -3.73884171e-01
1.55690730e+00 -2.35763717e+00 5.80502212e-01 4.83272411e-02
-1.81068927e-01 -5.81844568e-01 -2.27541715e-01 8.62641692e-01
-4.23674345e-01 1.73832983e-01 9.28299353e-02 -4.98897284e-01
2.91109800e-01 4.42855619e-02 -1.05855191e+00 1.39713168e-01
1.56290352e-01 7.67724931e-01 -9.18014526e-01 -2.41526872e-01
-2.25541592e-01 3.17022413e-01 -6.88584507e-01 6.37543440e-01
-9.07152534e-01 8.50213647e-01 -4.03292596e-01 3.59317333e-01
-2.52415150e-01 -1.16256051e-01 1.97097495e-01 2.34406382e-01
-4.29190546e-01 4.71691012e-01 -1.29515600e+00 1.87716293e+00
-3.48003864e-01 3.70050848e-01 -2.78392404e-01 -1.51756033e-01
9.60126042e-01 1.00067449e+00 5.55651546e-01 2.12974846e-02
-1.85532514e-02 -1.40050966e-02 1.74004287e-01 -1.90338582e-01
6.29898608e-01 -4.96763259e-01 -5.54243326e-01 8.62785578e-01
-8.25559497e-02 -8.52821112e-01 4.17331368e-01 2.65979677e-01
1.27812839e+00 5.76767266e-01 3.47725958e-01 1.17397569e-01
-1.97780937e-01 -4.16500308e-02 7.08220482e-01 1.08096552e+00
1.50998726e-01 8.39483798e-01 7.52790511e-01 -2.48125866e-01
-1.29956937e+00 -1.54419947e+00 3.89973730e-01 1.16064334e+00
-5.79588592e-01 -7.90963113e-01 -3.55711013e-01 1.20088279e-01
-2.04778284e-01 1.30236018e+00 -6.61258161e-01 -5.07169515e-02
-8.53625894e-01 -4.55160201e-01 5.43991864e-01 7.18614042e-01
-1.71986163e-01 -1.60645270e+00 -5.68009973e-01 7.28323340e-01
-1.92102417e-01 -6.06656134e-01 -9.23807025e-01 4.03524697e-01
-5.65322816e-01 -5.94512999e-01 -4.47136819e-01 -5.86951613e-01
1.32377461e-01 -7.14255154e-01 1.55891204e+00 -2.02152476e-01
-2.04037756e-01 4.30272073e-01 -4.89170663e-02 -7.71390915e-01
-8.70263994e-01 -6.27563298e-02 1.07441492e-01 -1.23066525e-03
-2.71393657e-01 -1.05507410e+00 -2.97974318e-01 4.80891131e-02
-9.27640080e-01 4.96066362e-01 2.68943191e-01 7.54010141e-01
8.19007635e-01 -1.82837993e-01 6.64769292e-01 -5.32028913e-01
8.07881653e-01 -2.77157277e-01 -3.27464521e-01 1.34352529e-02
-1.64831460e-01 1.53292581e-01 3.92003119e-01 -1.20038855e+00
-1.09326351e+00 4.33723778e-01 2.80294925e-01 -4.96881872e-01
-4.70530465e-02 4.87097979e-01 2.28866432e-02 9.80209172e-01
8.12253773e-01 9.75753963e-02 -2.99035251e-01 -1.03595130e-01
5.35142541e-01 -3.59024145e-02 1.03481460e+00 -9.46457207e-01
8.09523106e-01 2.41244018e-01 3.46234106e-02 -4.86793905e-01
-5.50553799e-01 4.66723293e-02 -4.55555320e-01 -4.91856068e-01
8.91329050e-01 -9.92536366e-01 -8.85330677e-01 2.66528219e-01
-1.26331580e+00 -8.43466640e-01 -9.71263945e-01 6.22403026e-01
-1.33842134e+00 -4.70398724e-01 -8.48816216e-01 -1.18381631e+00
-1.94830462e-01 -5.88794470e-01 1.15153265e+00 2.25076452e-01
-1.41441858e+00 -5.95916569e-01 5.00632823e-01 -4.35492933e-01
7.38175362e-02 6.90444887e-01 8.11757624e-01 -5.49361348e-01
-6.80945933e-01 -3.69749337e-01 8.90912294e-01 -8.05734321e-02
-7.18070716e-02 5.62218130e-01 -8.58144164e-01 2.28011478e-02
-8.05821642e-02 -2.90497988e-01 5.88587463e-01 6.06043458e-01
6.29504144e-01 -1.10326365e-01 -2.11871862e-01 3.17963958e-01
8.85275245e-01 6.46964490e-01 7.22090960e-01 -1.15532562e-01
2.95567453e-01 4.34594661e-01 5.29114068e-01 8.28849852e-01
-2.72468776e-01 4.83088762e-01 -2.88681686e-02 2.99847424e-01
-1.50689974e-01 -1.13452733e+00 6.31573319e-01 8.25333834e-01
-4.56690460e-01 -5.20060956e-01 -7.58492410e-01 7.25832105e-01
-1.91934848e+00 -1.61994946e+00 -1.78276286e-01 2.24008679e+00
1.15698385e+00 2.39047557e-01 1.36194050e-01 3.12523454e-01
7.53742099e-01 1.63919553e-01 -3.78000528e-01 -1.92561418e-01
-1.18309163e-01 6.86556518e-01 -1.13595970e-01 5.29423654e-01
-1.03405011e+00 9.29273427e-01 7.45758057e+00 6.63322449e-01
-8.40214312e-01 -2.44868308e-01 4.44086403e-01 -5.09238780e-01
-5.57430625e-01 2.63874650e-01 -5.83186030e-01 4.49556142e-01
1.04260588e+00 -5.06223500e-01 7.05843091e-01 5.05468309e-01
7.44060814e-01 1.83130220e-01 -1.63852429e+00 6.41714096e-01
-3.72111142e-01 -1.32289171e+00 1.04403488e-01 -1.52217224e-01
9.86953914e-01 -4.85597283e-01 2.36551106e-01 4.15071666e-01
1.02350330e+00 -1.13982725e+00 1.34087074e+00 1.09113896e+00
7.84456611e-01 -6.79612160e-01 -3.38743508e-01 2.53411144e-01
-1.10339224e+00 -3.78127466e-03 3.65958571e-01 -4.87090141e-01
7.82864809e-01 4.10326451e-01 -9.94599819e-01 -3.80541012e-02
1.44971952e-01 5.17446280e-01 2.78885849e-02 7.11922228e-01
-7.88303077e-01 1.13372445e+00 -1.99954584e-01 1.06940605e-02
-1.81240350e-01 -1.82095021e-01 7.72661865e-01 9.79811549e-01
6.24082923e-01 1.81936443e-01 1.66781098e-01 1.07512701e+00
1.12884499e-01 -2.28408337e-01 -5.08858025e-01 -6.19798481e-01
6.74285591e-01 9.78972912e-01 -6.45882070e-01 -5.38697243e-01
2.05433935e-01 9.57545459e-01 -1.73394427e-01 4.78762001e-01
-9.99544621e-01 5.95421530e-03 4.61292148e-01 7.15718716e-02
2.02617720e-01 -5.55975258e-01 -4.28275943e-01 -9.63177025e-01
-4.91683811e-01 -9.51347649e-01 1.84568867e-01 -1.39064336e+00
-1.17064750e+00 5.15110791e-01 -2.63741873e-02 -1.25918305e+00
-1.16824818e+00 2.26030409e-01 -1.04639542e+00 9.26705122e-01
-3.50038081e-01 -9.83236074e-01 1.20555416e-01 5.60348988e-01
6.85284734e-01 -1.74176507e-02 1.22397208e+00 -1.73104197e-01
-1.16169751e-01 -1.81281775e-01 -4.79609132e-01 -1.33942738e-01
8.37181330e-01 -1.39233136e+00 6.86444581e-01 8.23800802e-01
6.35454595e-01 7.11132407e-01 1.21882260e+00 -1.01648712e+00
-1.09722173e+00 -8.23709130e-01 1.20657134e+00 -7.26714909e-01
8.77907038e-01 -3.91958207e-01 -4.30612028e-01 1.01261365e+00
3.73643965e-01 -6.38189852e-01 6.37835443e-01 1.53675213e-01
2.82680057e-02 3.66014779e-01 -3.87659669e-01 1.06900287e+00
1.12498295e+00 -5.17498553e-01 -7.56922483e-01 1.58201501e-01
9.43903565e-01 -4.58595425e-01 -6.21973157e-01 2.29956537e-01
8.26909006e-01 -6.54038668e-01 7.94625878e-01 -7.87336588e-01
7.65780091e-01 -5.16377151e-01 -3.01852860e-02 -1.36203742e+00
-4.05154973e-01 -1.58656716e+00 -1.96587831e-01 1.44242394e+00
4.75473851e-01 2.42499366e-01 5.31294882e-01 7.38684893e-01
-1.87791809e-01 -2.18921620e-02 -4.90717351e-01 -6.71427011e-01
-3.61167222e-01 -7.88490355e-01 6.16245687e-01 7.18684316e-01
-1.45301551e-01 3.77727300e-01 -6.51436388e-01 1.45110711e-02
5.38714170e-01 2.90299863e-01 9.47724998e-01 -1.12307596e+00
-8.84105504e-01 -2.08647162e-01 7.69673735e-02 -7.18482733e-01
-8.55004601e-03 -7.78412998e-01 3.78547907e-01 -9.94755447e-01
-1.59107745e-01 -2.82185823e-01 -3.56009305e-02 4.70668018e-01
-8.84994119e-02 1.69329979e-02 6.14025831e-01 3.91840011e-01
-1.01068690e-01 6.65471017e-01 1.26703775e+00 -5.55764735e-02
-7.17864096e-01 3.16945702e-01 -5.16738415e-01 8.13772261e-01
5.41715980e-01 -6.20774627e-01 -6.51700735e-01 -1.06242724e-01
4.40783411e-01 7.11176634e-01 2.89942801e-01 -1.05652153e+00
3.87511015e-01 -3.87728453e-01 5.23537099e-01 -5.69437027e-01
3.94938529e-01 -1.58224955e-01 1.07718670e+00 3.31515312e-01
-9.58563566e-01 3.51393849e-01 2.66623124e-03 6.14623785e-01
-2.27159485e-01 9.02950466e-02 2.55162954e-01 -1.06248952e-01
-2.85719573e-01 1.56415135e-01 -6.95910811e-01 2.42655110e-02
7.98685014e-01 2.42487147e-01 2.61556417e-01 -8.07206810e-01
-1.50704718e+00 -2.10586786e-01 1.52028844e-01 3.40705603e-01
3.99468839e-01 -1.45655298e+00 -8.44050825e-01 6.41171932e-02
-2.85432070e-01 -9.08386111e-02 -9.39173475e-02 5.72099984e-01
-1.43930510e-01 -7.25148246e-02 -1.03974221e-02 -5.01096547e-01
-9.12806273e-01 5.58557034e-01 -1.89422369e-01 -3.50106180e-01
-5.39849043e-01 8.38506460e-01 4.68264632e-02 1.60207555e-01
1.59782842e-01 -3.63004416e-01 1.14995465e-01 2.21585691e-01
6.17777705e-01 8.74411538e-02 -5.79950929e-01 3.81601341e-02
1.43851474e-01 6.21991195e-02 2.94260651e-01 -1.04081762e+00
1.39559829e+00 1.47353783e-01 -4.55565192e-02 1.23234367e+00
4.45008844e-01 4.66316223e-01 -1.50849450e+00 3.43498796e-01
1.44323051e-01 -5.07623032e-02 -4.93854642e-01 -8.45140815e-01
-3.68266225e-01 3.31366152e-01 -1.06596693e-01 2.49773160e-01
1.07981551e+00 -1.91260856e-02 5.12653291e-01 -3.77607383e-02
3.72866422e-01 -8.95190775e-01 3.24569166e-01 4.97012556e-01
1.27253771e+00 -3.95924062e-01 -3.44970495e-01 -6.53491393e-02
-7.95106292e-01 9.70942676e-01 6.87458217e-02 -3.87015015e-01
4.78783190e-01 7.85182893e-01 -3.01051810e-02 2.52950117e-02
-1.45684874e+00 -1.00809470e-01 1.12541653e-01 4.13600713e-01
7.74422884e-01 3.48025113e-01 -2.09864467e-01 7.43400633e-01
-7.38955677e-01 3.45503569e-01 5.27813613e-01 8.07452619e-01
3.98603231e-02 -9.94325817e-01 -4.08746928e-01 6.54409006e-02
-3.23871344e-01 -4.08812881e-01 -3.07501048e-01 7.48641074e-01
6.33311719e-02 1.06309712e+00 3.89110714e-01 -3.03304434e-01
2.19162956e-01 3.35184097e-01 4.96004522e-01 -7.43941724e-01
-6.76441252e-01 7.45009422e-01 3.62817317e-01 -4.32600170e-01
-3.46003354e-01 -1.21896446e+00 -1.29360735e+00 -2.42338162e-02
1.03484467e-01 2.39792034e-01 3.50554138e-01 5.44293702e-01
1.15692116e-01 9.81216490e-01 6.81172073e-01 -1.12719357e+00
-4.34226096e-01 -1.15469420e+00 -7.35109985e-01 6.63674116e-01
4.61859256e-02 -4.76989010e-03 -4.42894548e-01 8.65187645e-01]
|
[15.750670433044434, 5.753371715545654]
|
c0756db4-8393-4af2-befa-8a48e7fac7c9
|
ada-vad-unpaired-adversarial-domain
| null | null |
https://ieeexplore.ieee.org/document/9746755
|
https://sigport.org/sites/default/files/docs/ADA-VAD_ICASSP2022_Poster_v2.pdf.pdf
|
ADA-VAD: Unpaired Adversarial Domain Adaptation for Noise-Robust Voice Activity Detection
|
Voice Activity Detection (VAD) is becoming an essential front-end component in various speech processing systems. As those systems are commonly deployed in environments with diverse noise types and low signal-to-noise ratios (SNRs), an effective VAD method should perform robust detection of speech region out of noisy background signals. In this paper, we propose adversarial domain adaptive VAD (ADA-VAD), which is a deep neural network (DNN) based VAD method highly robust to audio samples with various noise types and low SNRs. The proposed method trains DNN models for a VAD task in a supervised manner. Simultaneously, to mitigate the performance degradation due to back-ground noises, the adversarial domain adaptation method is adopted to match the domain discrepancy between noisy and clean audio stream in an unsupervised manner. The results show that ADA-VAD achieves an average of 3.6%p and 7%p higher AUC than models trained with manually extracted features on the AVA-speech dataset and a speech database synthesized with an unseen noise database, respectively.
|
['Jong Hwan Ko', 'Jiho Chang', 'Taesoo Kim']
|
2022-04-22
| null | null | null |
icassp-2022-4
|
['action-detection', 'activity-detection']
|
['computer-vision', 'computer-vision']
|
[ 1.06653765e-01 -4.21190500e-01 3.90702605e-01 -1.28268853e-01
-1.22431636e+00 -5.77392340e-01 5.19427001e-01 -3.16013455e-01
-2.19718024e-01 6.11680388e-01 4.51076835e-01 -3.69637281e-01
2.32009083e-01 -4.66876447e-01 -3.46456915e-01 -8.75910103e-01
1.41802937e-01 -1.00208297e-01 1.54575318e-01 -9.81308073e-02
-3.36685568e-01 6.31710529e-01 -1.52816844e+00 2.38574818e-01
6.14750206e-01 1.27537477e+00 2.68241405e-01 9.19001520e-01
-4.21933383e-02 5.64207256e-01 -1.26971912e+00 -7.00359195e-02
4.06778336e-01 -6.70732915e-01 1.06058056e-02 4.60768752e-02
2.77029783e-01 -2.52064615e-01 -7.18743742e-01 1.30104506e+00
1.23363781e+00 3.13327402e-01 7.89855599e-01 -1.06911886e+00
-2.42558479e-01 3.44569385e-01 -2.74997115e-01 5.78774691e-01
2.74616897e-01 4.83456612e-01 5.55312395e-01 -1.09558451e+00
1.36770844e-01 1.25206447e+00 3.92926604e-01 8.09815824e-01
-9.33176816e-01 -7.98448980e-01 -1.38631761e-01 2.10691273e-01
-1.25623560e+00 -9.65269327e-01 1.24375522e+00 -2.32975334e-01
8.32842708e-01 2.60464668e-01 2.09745437e-01 1.66874516e+00
-7.33611062e-02 8.21646750e-01 1.03278553e+00 -3.15725088e-01
4.99990880e-01 -3.55886370e-02 -4.56338346e-01 -9.42552090e-02
-2.75304824e-01 2.09041759e-01 -6.00345969e-01 -8.94645974e-02
4.72656310e-01 -5.90565085e-01 -3.11289251e-01 7.08758384e-02
-8.30749035e-01 4.97136891e-01 4.69873361e-02 2.98702627e-01
-6.22240067e-01 -2.70198643e-01 7.27228165e-01 5.54181039e-01
5.74852049e-01 2.22225308e-01 -5.12650847e-01 -3.54489326e-01
-7.65978813e-01 4.69067842e-02 6.56607985e-01 6.92516804e-01
-1.90604001e-01 1.06618989e+00 -3.34803820e-01 1.32615054e+00
3.58373433e-01 8.31606805e-01 7.76587009e-01 -7.83392429e-01
4.94477004e-01 -1.76544785e-01 4.72312979e-02 -6.77618325e-01
-7.54953474e-02 -9.17424262e-01 -9.24984157e-01 4.29721296e-01
2.88816839e-01 -3.72808367e-01 -1.07418597e+00 1.72506249e+00
3.50798130e-01 3.73154908e-01 5.63603222e-01 8.78037333e-01
8.59804451e-01 9.48074341e-01 -7.19212145e-02 -6.70070171e-01
1.06002033e+00 -8.06412697e-01 -1.22753251e+00 -4.04778719e-01
-9.89881381e-02 -1.10079432e+00 1.15963411e+00 7.73886919e-01
-1.09616613e+00 -7.74897277e-01 -1.21778679e+00 3.08044314e-01
-1.09651856e-01 -5.04489802e-02 -2.37577632e-01 1.01894879e+00
-8.18283856e-01 -1.33413421e-02 -6.02219343e-01 7.06617311e-02
4.52747494e-01 7.82978609e-02 -2.31035352e-01 3.97378765e-02
-1.31407571e+00 6.39839828e-01 -5.78739308e-03 1.34383321e-01
-1.52559364e+00 -3.22748184e-01 -7.09720552e-01 -7.63695389e-02
2.13615209e-01 -1.62281379e-01 1.50519073e+00 -1.03848076e+00
-1.86363328e+00 4.46488738e-01 -1.24151610e-01 -7.65639484e-01
6.57522678e-01 -2.21031502e-01 -1.16646433e+00 1.21807829e-01
-1.92732260e-01 -3.58828120e-02 1.44079423e+00 -1.08467281e+00
-3.04733217e-01 -1.18024580e-01 -6.71259999e-01 3.99917781e-01
-3.00797313e-01 2.20609173e-01 -1.03319593e-01 -1.00644600e+00
2.69394845e-01 -3.60069096e-01 7.79633000e-02 -1.40828505e-01
-2.06308216e-01 -9.04550180e-02 1.17496192e+00 -1.16629362e+00
1.07667887e+00 -2.57761836e+00 -1.65902644e-01 1.01881318e-01
-5.42365201e-02 1.04474020e+00 -1.93862006e-01 5.78265823e-02
2.30312310e-02 -3.27492774e-01 -2.79052585e-01 -2.50793934e-01
5.82524315e-02 2.24009599e-03 -2.69362658e-01 4.02684152e-01
3.17627102e-01 1.96699411e-01 -8.90859127e-01 -2.50228643e-01
3.94448996e-01 5.84483266e-01 -1.46439418e-01 6.12317383e-01
-9.81786922e-02 5.41074932e-01 1.45979635e-02 6.44252717e-01
8.60071421e-01 8.00804794e-01 -1.88460469e-01 2.96685891e-03
1.80184439e-01 4.28740352e-01 -1.35088944e+00 1.40681601e+00
-5.42613149e-01 8.90274942e-01 7.42652655e-01 -8.17511976e-01
1.35505950e+00 8.27165604e-01 1.29501969e-01 -7.63173580e-01
3.00364465e-01 4.79378521e-01 2.61172891e-01 -4.13641244e-01
-3.52857374e-02 -1.57840952e-01 2.27955863e-01 -1.89662114e-01
2.84121931e-01 -4.61579680e-01 -5.74077904e-01 -2.46983811e-01
1.39104068e+00 -4.64335263e-01 2.39989161e-01 7.68018737e-02
7.62220919e-01 -7.25305378e-01 7.29442179e-01 6.55736387e-01
-8.62323225e-01 6.84155345e-01 1.46736577e-01 1.88870266e-01
-9.26092565e-01 -1.33884490e+00 -1.57106131e-01 7.19056845e-01
-3.01322818e-01 1.55617416e-01 -8.29218686e-01 -3.26904833e-01
-3.89999986e-01 8.62064600e-01 -7.04114586e-02 -2.78952390e-01
-4.59591925e-01 -2.16924533e-01 1.00265670e+00 4.17670965e-01
7.84558713e-01 -1.14241624e+00 1.74791813e-01 4.72659379e-01
-1.13063380e-01 -1.24223125e+00 -4.56359237e-01 4.04221267e-01
-3.65230471e-01 -4.88077790e-01 -9.34773505e-01 -1.12553751e+00
4.85246927e-02 1.75202146e-01 7.04370916e-01 -6.87371731e-01
1.28347918e-01 1.77232236e-01 -4.33098584e-01 -7.27912009e-01
-1.11349654e+00 -2.99130559e-01 6.12079263e-01 2.86388040e-01
5.04531801e-01 -7.61733770e-01 -5.51110506e-01 4.44540620e-01
-6.77882254e-01 -5.04693270e-01 4.59671974e-01 1.10154891e+00
5.13099611e-01 4.69061494e-01 1.17099988e+00 -2.35551342e-01
1.02166510e+00 -5.81950545e-01 -4.59070742e-01 -2.88858443e-01
-1.51454628e-01 -4.64588046e-01 9.39738393e-01 -7.54838169e-01
-1.29710662e+00 -2.82662790e-02 -7.75517106e-01 -8.39229822e-01
-4.58986163e-01 2.43939281e-01 -9.88427818e-01 2.70029306e-01
1.07037258e+00 4.66677517e-01 -8.43880046e-03 -6.53998375e-01
1.78512901e-01 1.36668670e+00 9.19680536e-01 -1.58577383e-01
1.06345904e+00 -8.55323821e-02 -3.49001437e-01 -1.20443881e+00
-4.08519983e-01 -4.84464020e-01 -2.07217485e-01 -1.64915979e-01
6.78035796e-01 -1.14090943e+00 -1.21603729e-02 1.04759598e+00
-1.19485176e+00 -2.44275242e-01 -1.24849521e-01 7.68795669e-01
-3.39616895e-01 2.60436952e-01 -5.20960391e-01 -1.24823749e+00
-5.56008279e-01 -1.20757282e+00 5.56720674e-01 9.67431068e-02
-2.48011947e-02 -5.09506285e-01 -1.82009175e-01 4.35841590e-01
5.45470536e-01 1.19306467e-01 5.13640225e-01 -1.17540133e+00
2.14499254e-02 -2.76231706e-01 3.31144512e-01 1.39367867e+00
5.94020605e-01 -2.49749690e-01 -1.72755075e+00 -1.78297639e-01
5.45381665e-01 -6.39713779e-02 4.11486804e-01 5.32757223e-01
1.03606415e+00 -3.32098752e-01 2.74175525e-01 3.76586854e-01
7.93665648e-01 8.53303492e-01 6.33017480e-01 2.23308541e-02
3.85308802e-01 2.49617130e-01 7.28300810e-01 2.55865365e-01
-6.30355358e-01 4.97456163e-01 4.71097589e-01 1.48079386e-02
-6.26495719e-01 -3.05544794e-01 5.49913585e-01 1.24040735e+00
5.95973194e-01 -6.94663942e-01 -6.44026518e-01 7.69709885e-01
-1.17594218e+00 -9.13632333e-01 1.38094410e-01 2.16787958e+00
8.27817500e-01 5.21380186e-01 2.28444219e-01 9.26293075e-01
9.44525182e-01 3.39152098e-01 -8.30889463e-01 -5.77913404e-01
-5.06936014e-01 5.35607040e-01 1.54859141e-01 4.36400384e-01
-1.05379426e+00 7.45510578e-01 5.54358816e+00 1.23961401e+00
-1.17470574e+00 2.51953244e-01 4.92247134e-01 -2.89164744e-02
3.87167721e-03 -8.33807707e-01 -1.80039048e-01 5.91421664e-01
1.20865214e+00 -1.58868507e-02 5.15159726e-01 1.07805562e+00
6.01849973e-01 4.05105323e-01 -7.26679385e-01 1.18790984e+00
-7.16049299e-02 -6.83029652e-01 -2.37836197e-01 -1.39118776e-01
5.21598101e-01 -8.42648596e-02 4.50007200e-01 4.14701492e-01
-4.47594412e-02 -9.25926328e-01 6.78179622e-01 1.98279079e-02
8.63235533e-01 -8.94283891e-01 8.37672293e-01 3.63400310e-01
-1.00353360e+00 -7.37581030e-02 -1.65409833e-01 6.33430406e-02
1.74597243e-03 5.10096490e-01 -1.32571900e+00 1.62520498e-01
6.13661408e-01 1.34786323e-01 1.35573387e-01 1.05924320e+00
-3.43001068e-01 1.31125891e+00 -1.03654958e-01 -3.77269611e-02
-1.82705317e-02 1.52850181e-01 1.17361593e+00 1.30471003e+00
3.81770790e-01 -1.38558879e-01 -4.19072211e-01 3.10292453e-01
-4.25376475e-01 5.73795363e-02 -7.15513349e-01 -4.25522961e-02
1.05317318e+00 7.78369367e-01 -1.65686861e-01 -3.77107672e-02
-2.27510825e-01 1.01435256e+00 -3.96022409e-01 4.80690837e-01
-8.29790652e-01 -5.98975658e-01 1.07959819e+00 -8.97549167e-02
2.85736442e-01 -2.28339866e-01 -1.43927857e-01 -6.25538588e-01
1.73029184e-01 -1.36214292e+00 -4.46145870e-02 -8.74811590e-01
-1.42730618e+00 7.49475718e-01 -5.35907090e-01 -1.65352643e+00
-1.91519469e-01 -5.81923366e-01 -7.84092367e-01 1.13466835e+00
-1.29687893e+00 -6.48613572e-01 -1.18024580e-01 7.44632244e-01
1.12079108e+00 -7.44705915e-01 8.22388649e-01 5.84145844e-01
-5.80450177e-01 9.23922896e-01 4.34424818e-01 4.42327976e-01
7.75988579e-01 -1.10313976e+00 5.78944564e-01 1.41467416e+00
8.37989897e-02 1.95774347e-01 9.97599304e-01 -3.11330229e-01
-1.15202141e+00 -1.30135286e+00 3.99762064e-01 3.84480096e-02
4.03489113e-01 -5.81397116e-01 -1.07833648e+00 1.61620513e-01
2.82470882e-01 2.40104139e-01 5.90879321e-01 -4.68793422e-01
-2.62358218e-01 -4.80490893e-01 -1.45307779e+00 4.73179460e-01
8.96417201e-01 -7.82619298e-01 -8.56552243e-01 7.19233379e-02
1.06428170e+00 -5.68198085e-01 -6.40048087e-01 2.96351045e-01
1.32994667e-01 -6.91763341e-01 9.15445983e-01 -3.87971640e-01
-1.75372675e-01 -5.88333011e-01 -5.38979053e-01 -1.70237827e+00
7.73524959e-03 -9.32353139e-01 -3.34385574e-01 1.67871356e+00
3.00692528e-01 -5.96613050e-01 3.30040932e-01 -5.44226319e-02
-5.90930343e-01 -1.44078404e-01 -1.40507889e+00 -1.06111646e+00
-1.60087451e-01 -9.03102517e-01 5.79897821e-01 6.91432297e-01
-4.59092438e-01 2.59866536e-01 -4.92185265e-01 6.15946591e-01
4.40052629e-01 -9.48303640e-01 6.95131361e-01 -7.75326967e-01
-2.56575257e-01 -2.36967951e-01 -6.25944078e-01 -9.65777695e-01
1.11557901e-01 -4.38266873e-01 3.22461128e-01 -1.27341378e+00
-8.79255235e-01 6.04807176e-02 -6.86909735e-01 -7.24587291e-02
-6.46594241e-02 -7.17458501e-02 -1.56031042e-01 -1.35550037e-01
-5.88763319e-02 7.86938906e-01 1.08280098e+00 -4.21446770e-01
-4.49447006e-01 5.05118251e-01 -4.40740645e-01 7.45376229e-01
9.63979363e-01 -4.70314056e-01 -6.01671338e-01 -1.28064200e-01
-6.87851191e-01 2.43974283e-01 -8.23312905e-03 -1.32476032e+00
5.76856658e-02 8.07451829e-03 3.60495776e-01 -5.69062531e-01
8.50056112e-01 -7.74692118e-01 -6.80144131e-02 3.61199260e-01
-2.41675720e-01 -4.87779796e-01 4.02152777e-01 7.16446757e-01
-5.53190529e-01 7.20517635e-02 1.07131517e+00 9.60000083e-02
-5.52925825e-01 4.79146652e-03 -7.31350303e-01 4.89512496e-02
7.20172107e-01 -1.38913572e-01 -3.19687158e-01 -7.54544735e-01
-5.91292024e-01 -3.47953618e-01 -1.94303602e-01 4.46171373e-01
7.73287416e-01 -1.34010291e+00 -8.10249209e-01 4.94405448e-01
-4.25859317e-02 7.72818401e-02 3.85564089e-01 3.39484215e-01
-3.15060884e-01 -2.79701930e-02 -4.89179567e-02 -4.93192106e-01
-1.39220047e+00 4.92794573e-01 6.27520144e-01 4.21411753e-01
-2.60444164e-01 1.10380089e+00 -9.89280418e-02 -7.05176368e-02
7.42106020e-01 -2.41003439e-01 -7.55288452e-02 -1.91598028e-01
6.41223907e-01 5.35558701e-01 4.99894023e-01 -7.97601402e-01
-3.65466833e-01 -8.25854987e-02 2.58789867e-01 -3.71108830e-01
9.86265481e-01 -6.10824078e-02 5.19343972e-01 3.95208269e-01
1.15972865e+00 5.22399127e-01 -1.40244436e+00 -4.68127251e-01
-3.02431792e-01 -4.23414409e-01 3.95280957e-01 -9.91837859e-01
-9.95346367e-01 1.07025671e+00 1.15456104e+00 2.78474182e-01
1.48141861e+00 -3.32926333e-01 9.02297318e-01 1.42896501e-02
-4.20966633e-02 -1.23764920e+00 3.04358274e-01 3.38677913e-01
1.04076648e+00 -9.69991326e-01 -4.80598837e-01 9.71823279e-03
-7.50755012e-01 9.01057720e-01 6.11067116e-01 1.61399111e-01
6.57412529e-01 3.75630766e-01 6.79320455e-01 4.80463386e-01
-4.78155881e-01 -2.15210065e-01 1.87418520e-01 1.17558920e+00
-6.29740953e-02 4.68025506e-02 1.27738610e-01 8.37151587e-01
-2.56409019e-01 -4.72921431e-01 2.43565559e-01 6.60148144e-01
-5.54350615e-01 -8.49055767e-01 -7.70398438e-01 3.82405400e-01
-6.44298792e-01 -7.70491138e-02 -3.55316341e-01 3.18846345e-01
-3.67222242e-02 1.61188543e+00 -7.69036189e-02 -7.28148520e-01
6.51354790e-01 3.38460714e-01 -3.06025762e-02 -3.61780882e-01
-4.76939708e-01 5.46351612e-01 1.57824919e-01 2.39389632e-02
-1.19072266e-01 -6.71633661e-01 -9.48202014e-01 5.85284941e-02
-2.88075089e-01 2.86108870e-02 8.66569579e-01 7.30655611e-01
2.26042226e-01 8.10051084e-01 1.18712711e+00 -5.15963376e-01
-7.38440990e-01 -1.39490974e+00 -8.01576674e-01 2.69352436e-01
7.83346653e-01 -5.02740681e-01 -9.48533177e-01 -1.32850215e-01]
|
[14.903179168701172, 6.103391647338867]
|
b91c77e4-6267-44ff-a3f1-912116173966
|
simultaneous-fidelity-and-regularization
|
1804.04522
| null |
https://arxiv.org/abs/1804.04522v4
|
https://arxiv.org/pdf/1804.04522v4.pdf
|
Simultaneous Fidelity and Regularization Learning for Image Restoration
|
Most existing non-blind restoration methods are based on the assumption that a precise degradation model is known. As the degradation process can only be partially known or inaccurately modeled, images may not be well restored. Rain streak removal and image deconvolution with inaccurate blur kernels are two representative examples of such tasks. For rain streak removal, although an input image can be decomposed into a scene layer and a rain streak layer, there exists no explicit formulation for modeling rain streaks and the composition with scene layer. For blind deconvolution, as estimation error of blur kernel is usually introduced, the subsequent non-blind deconvolution process does not restore the latent image well. In this paper, we propose a principled algorithm within the maximum a posterior framework to tackle image restoration with a partially known or inaccurate degradation model. Specifically, the residual caused by a partially known or inaccurate degradation model is spatially dependent and complexly distributed. With a training set of degraded and ground-truth image pairs, we parameterize and learn the fidelity term for a degradation model in a task-driven manner. Furthermore, the regularization term can also be learned along with the fidelity term, thereby forming a simultaneous fidelity and regularization learning model. Extensive experimental results demonstrate the effectiveness of the proposed model for image deconvolution with inaccurate blur kernels, deconvolution with multiple degradations and rain streak removal.
|
['Ming-Hsuan Yang', 'WangMeng Zuo', 'David Zhang', 'Lei Zhang', 'Dongwei Ren']
|
2018-04-12
| null | null | null | null |
['image-deconvolution']
|
['computer-vision']
|
[ 3.10284048e-01 -5.99390209e-01 3.44277084e-01 -3.02946597e-01
-5.04639566e-01 -3.62681836e-01 3.00807506e-01 -4.79178429e-01
6.18201564e-04 1.03044808e+00 1.61477178e-01 7.31119439e-02
-2.05199346e-01 -4.11116093e-01 -6.65621579e-01 -1.09074426e+00
4.76564199e-01 4.64741439e-02 8.64949822e-03 4.04546745e-02
1.50000244e-01 2.92224169e-01 -1.50485098e+00 -4.66629267e-02
1.47449350e+00 7.57974207e-01 8.17322552e-01 7.42973983e-01
-9.35046896e-02 8.97462666e-01 -5.02205074e-01 1.48559004e-01
2.27481827e-01 -5.04219115e-01 -3.01314235e-01 6.94439232e-01
4.88817185e-01 -6.91684008e-01 -3.40590805e-01 1.38304913e+00
1.98484465e-01 6.99456334e-02 6.91393852e-01 -7.14135349e-01
-8.28389406e-01 -4.60280217e-02 -7.14687705e-01 1.33301750e-01
1.12327844e-01 2.03184187e-01 4.24568802e-01 -9.33733821e-01
1.21733241e-01 9.45794046e-01 6.20239377e-01 2.62261063e-01
-1.24990320e+00 -3.32976490e-01 1.06464557e-01 1.60928607e-01
-1.33602130e+00 -6.13298953e-01 7.69157767e-01 -6.10969067e-01
9.32456702e-02 3.10194999e-01 4.88269746e-01 6.00828886e-01
5.17383888e-02 4.02301133e-01 1.58475900e+00 -1.92705065e-01
2.10633382e-01 5.63277770e-03 3.89026195e-01 4.13593292e-01
5.27405918e-01 3.22143435e-01 -1.87692747e-01 -1.11773960e-01
9.18339014e-01 3.04829299e-01 -1.16372311e+00 -1.98214322e-01
-8.96159172e-01 4.27053809e-01 4.63235110e-01 1.42485723e-01
-5.87795973e-01 -1.00751251e-01 -1.81543753e-01 1.62256092e-01
5.21635592e-01 2.18126535e-01 -2.37550825e-01 3.31675977e-01
-1.37067509e+00 2.13727832e-01 6.52016580e-01 6.61000252e-01
1.12054563e+00 3.61173630e-01 -3.75296533e-01 1.00590765e+00
5.91168880e-01 8.28950822e-01 3.50073189e-01 -9.83171403e-01
1.29079789e-01 4.59066220e-02 1.04931724e+00 -6.51378810e-01
8.30563530e-02 -5.75238824e-01 -1.29638648e+00 5.59755564e-01
4.63439822e-01 -1.61970295e-02 -1.17894351e+00 1.52660680e+00
2.38972828e-01 8.77025545e-01 2.98704267e-01 1.54194117e+00
5.18028319e-01 7.57581174e-01 -2.26378903e-01 -7.88276494e-01
1.18000066e+00 -9.89458919e-01 -1.17753923e+00 -5.33209264e-01
-2.15782329e-01 -1.10064650e+00 8.46303225e-01 4.15121496e-01
-1.06526148e+00 -6.68813467e-01 -1.15068734e+00 -4.42951405e-03
3.30011427e-01 4.09659386e-01 2.12354690e-01 4.25717264e-01
-9.59913671e-01 5.50758064e-01 -6.49684370e-01 -7.32441917e-02
1.26770809e-01 -1.99781910e-01 -1.13549918e-01 -7.73286819e-01
-9.45654392e-01 1.01219308e+00 1.49241999e-01 8.48983407e-01
-1.22661412e+00 -6.84048474e-01 -4.90553558e-01 -9.53600332e-02
6.81547374e-02 -1.06623626e+00 9.05800164e-01 -1.13089991e+00
-1.40862882e+00 3.17564338e-01 -2.78188258e-01 -1.44574150e-01
6.32420421e-01 -5.50696969e-01 -4.66392547e-01 1.68904394e-01
-3.80149297e-02 -3.64208400e-01 1.63742685e+00 -1.90289140e+00
-2.96855599e-01 -2.66546607e-01 -4.07886039e-03 5.81604779e-01
3.06719154e-01 -2.00829297e-01 -2.17620254e-01 -6.87297881e-01
2.92194039e-01 -5.27500331e-01 -8.90934095e-02 2.21367598e-01
-2.46607855e-01 7.47690141e-01 7.57712662e-01 -1.23413372e+00
1.00481308e+00 -2.35980177e+00 1.97115064e-01 -2.24499881e-01
2.81662047e-01 3.50105613e-01 -5.10100983e-02 1.24581084e-01
2.92277969e-02 -2.68548340e-01 -9.27566290e-01 -4.10328537e-01
-3.42379779e-01 4.27163541e-01 -4.46649373e-01 8.21701705e-01
-5.18533885e-02 3.05027455e-01 -9.41777647e-01 4.70237099e-02
4.19884801e-01 7.98853040e-01 -3.58833116e-03 7.62098670e-01
-6.75490275e-02 7.88213730e-01 -3.00109863e-01 4.94435221e-01
1.47511721e+00 -1.49389997e-01 -5.58715761e-02 -6.14800572e-01
-3.37260425e-01 -3.48090470e-01 -1.38322961e+00 1.36983156e+00
-6.37871981e-01 3.55164587e-01 7.38317847e-01 -8.60865891e-01
8.28096092e-01 3.72098386e-01 1.31005878e-02 -9.37757641e-02
-1.94525808e-01 3.84701788e-01 -3.20285827e-01 -7.65293837e-01
3.46434772e-01 -7.12278903e-01 8.22093546e-01 1.10451944e-01
-1.78714231e-01 -2.05499217e-01 -2.42811322e-01 -1.45919815e-01
8.50817144e-01 1.12538420e-01 9.35071558e-02 -1.12549126e-01
6.00159168e-01 -2.58764237e-01 6.06937706e-01 7.94383109e-01
-1.18098497e-01 1.18188703e+00 -3.51706296e-01 -6.85662031e-02
-1.03756714e+00 -9.95021999e-01 -2.56343335e-01 4.54086632e-01
5.06679177e-01 2.53836811e-01 -6.65845275e-01 -1.34788662e-01
-1.45420611e-01 6.52001679e-01 -2.78842658e-01 1.15481773e-04
-3.46225709e-01 -1.19252372e+00 7.63701051e-02 -5.58458120e-02
9.26189005e-01 -6.44197404e-01 -4.62119281e-02 2.04134881e-01
-3.66148472e-01 -1.06421351e+00 -4.52798277e-01 -1.24176383e-01
-9.23870921e-01 -1.33999252e+00 -1.06830382e+00 -6.55030072e-01
8.80076587e-01 9.71437275e-01 7.62128115e-01 8.03427622e-02
-2.32205063e-01 2.16847554e-01 -2.62315601e-01 1.04861356e-01
-4.33683813e-01 -1.08114505e+00 -1.97886210e-02 7.09878981e-01
-4.00953382e-01 -7.14638650e-01 -8.25504899e-01 1.14067972e-01
-1.10099661e+00 2.59541363e-01 7.75523543e-01 1.01812708e+00
3.77996922e-01 5.34866929e-01 1.47535130e-01 -7.59877026e-01
5.30836403e-01 -5.51538646e-01 -6.91815376e-01 3.30162048e-01
-6.27884150e-01 -5.45301437e-02 5.24499059e-01 -5.13582945e-01
-1.73087668e+00 3.50539456e-03 3.12707633e-01 -8.59519243e-01
-3.11010689e-01 6.31717265e-01 -2.67336369e-01 -2.39045188e-01
5.94628751e-01 6.63119912e-01 -9.67738591e-03 -9.77345586e-01
4.21492785e-01 5.68205774e-01 6.66657269e-01 -3.26522022e-01
1.21367013e+00 6.30665720e-01 -2.17435583e-01 -1.00274289e+00
-8.74318123e-01 -6.96362793e-01 -3.47400725e-01 -2.34238923e-01
6.42444909e-01 -1.27983725e+00 -1.18305199e-01 1.04800260e+00
-1.28640378e+00 -3.14698458e-01 -8.78528059e-02 6.86826408e-01
-3.41870546e-01 9.03782964e-01 -5.93499839e-01 -1.10691619e+00
-1.44550443e-01 -9.03865159e-01 6.76135600e-01 4.65115160e-01
6.44697964e-01 -9.35062587e-01 -6.72772676e-02 5.31466901e-01
5.73354602e-01 6.46557137e-02 5.97212732e-01 2.99108654e-01
-8.97151887e-01 1.82847157e-02 -6.29002571e-01 8.50508451e-01
6.33978665e-01 -2.99160063e-01 -1.11063421e+00 -3.82143855e-01
7.60941565e-01 1.30996600e-01 9.88882065e-01 6.23365402e-01
5.81812263e-01 -5.49113154e-01 1.26744390e-01 7.99533367e-01
1.84316313e+00 -2.22616076e-01 7.17441678e-01 6.93193302e-02
9.13620889e-01 3.23475182e-01 6.35597527e-01 4.74241316e-01
9.09451023e-02 4.59151834e-01 5.53214252e-01 -1.11087039e-01
-5.42903244e-01 2.97164619e-01 3.17281753e-01 8.57858121e-01
-3.10222238e-01 -1.93443984e-01 -4.67122793e-01 6.11648440e-01
-1.84629607e+00 -8.67848933e-01 -5.10932863e-01 2.53763819e+00
9.51059878e-01 -3.62736881e-01 -5.75945079e-01 1.20617021e-02
8.88311505e-01 3.26369941e-01 -5.10602474e-01 2.81675696e-01
-3.91608834e-01 -9.61808637e-02 5.79661489e-01 9.66571629e-01
-8.77424777e-01 5.45594275e-01 5.51795912e+00 5.99229634e-01
-1.04924297e+00 2.43387997e-01 1.31448999e-01 3.36160064e-01
-3.61939669e-01 2.99140632e-01 -4.12513137e-01 7.54805148e-01
4.90455300e-01 -3.14290598e-02 1.00891161e+00 1.89458162e-01
8.44330490e-01 -3.34062546e-01 -6.07768714e-01 1.18514168e+00
-5.51913269e-02 -8.50206614e-01 -1.88661460e-02 -1.81371570e-01
8.71930361e-01 -2.09602833e-01 -9.78614688e-02 -1.19269855e-01
2.03455567e-01 -8.06650996e-01 5.63153386e-01 1.36057448e+00
6.11660480e-01 -2.08790198e-01 6.99194789e-01 6.29108548e-01
-7.78527379e-01 1.01145077e-03 -4.83987719e-01 -1.72057122e-01
2.55686581e-01 1.25216269e+00 -2.87764817e-01 9.95273888e-01
6.42697692e-01 8.51022184e-01 -1.78603709e-01 1.59139836e+00
-6.84310794e-01 8.09514642e-01 -4.88310866e-02 8.52888525e-01
-1.39055759e-01 -8.33942115e-01 8.63484502e-01 9.56087708e-01
5.97115815e-01 5.16798079e-01 6.37960806e-02 9.15190876e-01
1.84910789e-01 -3.46169382e-01 -2.38595203e-01 2.64373183e-01
2.19687879e-01 1.19855738e+00 -1.25308067e-01 -2.31785759e-01
-3.86269689e-01 1.42920792e+00 -1.37242109e-01 1.08477294e+00
-7.59441614e-01 8.06084722e-02 8.56358230e-01 5.76827750e-02
2.53651768e-01 -2.24653050e-01 -1.16163932e-01 -1.47967112e+00
8.89222249e-02 -6.42512500e-01 -1.99818790e-01 -1.29844511e+00
-1.55103838e+00 4.39063400e-01 -2.84121454e-01 -1.35576963e+00
3.24153900e-01 -4.01094019e-01 -6.78910673e-01 1.51805353e+00
-2.17774439e+00 -1.18765354e+00 -8.39299321e-01 4.83299136e-01
4.62462455e-01 2.18900681e-01 5.22638917e-01 3.58450353e-01
-6.06938541e-01 -3.14611524e-01 7.86530733e-01 -4.86826539e-01
9.46138561e-01 -1.24976516e+00 -4.83522683e-01 1.19915974e+00
-4.58783686e-01 5.17768860e-01 1.28889656e+00 -7.21262455e-01
-1.31048286e+00 -1.35276473e+00 3.87420952e-01 1.61067232e-01
5.77853441e-01 1.49559513e-01 -1.49925268e+00 3.39854598e-01
1.21377833e-01 4.37060624e-01 1.68951169e-01 -4.38541740e-01
-2.09671035e-01 -3.32685500e-01 -1.12893879e+00 2.03520700e-01
5.40690720e-01 -5.88301003e-01 -6.62258804e-01 5.80470264e-01
4.85180855e-01 -4.40936685e-01 -5.49933910e-01 4.68343407e-01
1.88596502e-01 -1.10194600e+00 1.10498857e+00 -1.36368781e-01
2.68761009e-01 -9.97504413e-01 -2.55269110e-01 -1.50867188e+00
-5.82463145e-01 -4.40889686e-01 -4.95210439e-01 1.43625474e+00
-1.74489498e-01 -4.68892664e-01 2.46134520e-01 4.68686461e-01
-2.67131418e-01 -7.85842165e-02 -5.76562881e-01 -7.07125783e-01
-3.70586842e-01 -2.72431085e-03 2.37396747e-01 1.02452886e+00
-8.76415968e-01 1.47358418e-01 -9.40016985e-01 1.11992931e+00
1.15279388e+00 3.11951816e-01 4.74254906e-01 -1.14344239e+00
-4.25020814e-01 -2.62491163e-02 1.19242072e-01 -1.20327961e+00
-1.32125795e-01 -3.21988702e-01 6.70780659e-01 -1.84904706e+00
2.69061804e-01 -3.64918530e-01 -3.03758770e-01 -5.76294586e-02
-5.68356931e-01 1.11716352e-01 -1.03226863e-01 8.63249362e-01
-3.68130095e-02 7.34534979e-01 1.41636539e+00 -1.52128384e-01
-7.99626559e-02 2.82061726e-01 -4.78201240e-01 6.91408634e-01
5.01863718e-01 -2.77504683e-01 -5.50372064e-01 -6.80409968e-01
-1.78030491e-01 3.09539914e-01 7.62829483e-01 -9.31898654e-01
2.88925290e-01 -4.23316956e-01 2.31062502e-01 -1.61346763e-01
5.60959220e-01 -8.44342589e-01 4.97174263e-01 -6.56229351e-03
1.26855507e-01 -8.79163504e-01 2.20397208e-02 1.00586581e+00
-3.86639297e-01 -3.92717749e-01 1.05862677e+00 -2.94348210e-01
-5.30717075e-01 2.53504843e-01 -3.05711359e-01 -3.55689198e-01
4.89569485e-01 -2.73477733e-01 -5.14426053e-01 -2.91720033e-01
-8.06782663e-01 -5.94491661e-02 5.98539531e-01 -2.03048456e-02
7.78642118e-01 -9.48255539e-01 -1.00993729e+00 -1.43831840e-03
-1.23951659e-01 1.70343351e-02 8.12469780e-01 8.64320457e-01
-6.01613402e-01 -1.67045102e-01 -3.51778157e-02 -2.57327348e-01
-1.10330331e+00 6.23700261e-01 6.36501849e-01 6.12245090e-02
-5.63056588e-01 6.75075233e-01 6.40345991e-01 -1.63895581e-02
-6.27810657e-02 -1.96828365e-01 -2.38080084e-01 -3.40873212e-01
7.02149808e-01 5.19684017e-01 -1.06496401e-01 -7.82038689e-01
9.95281786e-02 5.72386563e-01 2.20312774e-01 8.41504931e-02
1.23441327e+00 -8.12870979e-01 -4.03037906e-01 5.73679507e-01
7.85857677e-01 8.56631920e-02 -1.66534877e+00 -3.49139035e-01
-4.59174752e-01 -7.14002013e-01 5.91411471e-01 -1.04556572e+00
-1.09711838e+00 7.55924702e-01 8.94276023e-01 1.37159973e-01
1.41106379e+00 -4.51170921e-01 4.90698397e-01 -1.79005131e-01
4.25606780e-02 -5.88742077e-01 -1.47764996e-01 3.77777338e-01
1.11624849e+00 -1.19876444e+00 1.66020021e-01 -5.34298241e-01
-2.42149293e-01 1.00680244e+00 3.78459185e-01 -1.63816467e-01
6.33358538e-01 -4.20498215e-02 1.65457755e-01 1.80498883e-02
-1.76377594e-01 -1.67209327e-01 2.62914181e-01 4.81543005e-01
1.12854086e-01 -7.89821148e-04 -2.86204785e-01 5.64867914e-01
4.04208839e-01 2.25594625e-01 7.08844066e-01 6.92267358e-01
-7.57480919e-01 -7.50450730e-01 -1.02091157e+00 1.96939066e-01
-1.09376431e-01 -3.10580313e-01 2.87443072e-01 9.77173820e-02
2.08761588e-01 1.29506135e+00 -4.06777233e-01 1.21228747e-01
1.70531690e-01 -2.30283365e-01 4.94479835e-01 -6.27386034e-01
9.40083191e-02 3.14327568e-01 -7.15259388e-02 -2.52669483e-01
-6.67656600e-01 -5.63436925e-01 -8.57426167e-01 -1.28257245e-01
-5.08173943e-01 1.83736727e-01 5.83741248e-01 1.04492593e+00
9.87258330e-02 4.76120949e-01 6.78354979e-01 -9.13741946e-01
-4.74418312e-01 -1.12874317e+00 -1.21527731e+00 1.67375162e-01
1.06105518e+00 -6.20670497e-01 -9.51783478e-01 4.10580635e-01]
|
[11.535736083984375, -2.679246425628662]
|
a8459d1d-3449-4278-aff2-82b4473e4908
|
incremental-learning-on-food-instance
|
2306.15910
| null |
https://arxiv.org/abs/2306.15910v1
|
https://arxiv.org/pdf/2306.15910v1.pdf
|
Incremental Learning on Food Instance Segmentation
|
Food instance segmentation is essential to estimate the serving size of dishes in a food image. The recent cutting-edge techniques for instance segmentation are deep learning networks with impressive segmentation quality and fast computation. Nonetheless, they are hungry for data and expensive for annotation. This paper proposes an incremental learning framework to optimize the model performance given a limited data labelling budget. The power of the framework is a novel difficulty assessment model, which forecasts how challenging an unlabelled sample is to the latest trained instance segmentation model. The data collection procedure is divided into several stages, each in which a new sample package is collected. The framework allocates the labelling budget to the most difficult samples. The unlabelled samples that meet a certain qualification from the assessment model are used to generate pseudo-labels. Eventually, the manual labels and pseudo-labels are sent to the training data to improve the instance segmentation model. On four large-scale food datasets, our proposed framework outperforms current incremental learning benchmarks and achieves competitive performance with the model trained on fully annotated samples.
|
['Wing-Kwong Chan', 'Chong-Wah Ngo', 'Yu Cao', 'Huu-Thanh Nguyen']
|
2023-06-28
| null | null | null | null |
['instance-segmentation', 'incremental-learning']
|
['computer-vision', 'methodology']
|
[ 4.38379019e-01 3.04814965e-01 -6.46754205e-01 -8.37145030e-01
-8.43248785e-01 -5.75315833e-01 -2.44668514e-01 8.43323469e-01
-4.13429260e-01 4.02875215e-01 -1.96520224e-01 -8.60424191e-02
2.17974290e-01 -9.29789722e-01 -9.27920878e-01 -7.48719156e-01
1.26204178e-01 8.97234261e-01 2.40767188e-02 1.26168609e-01
1.75401092e-01 -1.00723229e-01 -1.44599354e+00 5.36822021e-01
1.13728189e+00 1.36336100e+00 5.19969881e-01 7.45689392e-01
-3.99058253e-01 7.37292886e-01 -4.80974376e-01 -2.40558639e-01
4.35189635e-01 -5.38287640e-01 -8.45888674e-01 5.34513354e-01
2.31147692e-01 -5.21100402e-01 5.30508101e-01 1.31119156e+00
2.95238882e-01 -3.70442346e-02 4.67038393e-01 -1.23318565e+00
-4.46872979e-01 1.26652539e+00 -4.94833946e-01 3.11019514e-02
1.57017007e-01 6.15547001e-02 8.97212505e-01 -7.03143537e-01
2.91104913e-01 8.57801855e-01 8.09489548e-01 3.92829597e-01
-9.62772191e-01 -3.87645006e-01 5.99098265e-01 1.06751710e-01
-1.07439482e+00 -2.50590265e-01 7.76912630e-01 -3.08727801e-01
7.00662851e-01 1.42413512e-01 9.15975928e-01 5.58321536e-01
-1.82637334e-01 1.24760270e+00 8.49712014e-01 -3.53955388e-01
5.53993821e-01 2.88520399e-02 6.15878940e-01 6.08201981e-01
1.70427561e-01 -2.50838876e-01 -4.78476025e-02 3.47833663e-01
3.97989035e-01 1.30403683e-01 1.15934787e-02 -2.61134952e-01
-8.67968023e-01 1.09990132e+00 7.29115665e-01 4.53301370e-02
-5.12251556e-01 -3.49270463e-01 5.71875274e-01 1.21792555e-01
7.44080007e-01 4.37202960e-01 -1.12062871e+00 3.38064611e-01
-1.05255020e+00 7.57443532e-02 1.03549659e+00 1.10749793e+00
7.87354827e-01 -2.54297763e-01 -3.09441775e-01 8.76983225e-01
4.05891359e-01 2.39983529e-01 3.14942449e-01 -9.01391089e-01
4.09187794e-01 1.05914629e+00 2.84557462e-01 -8.20046604e-01
-8.35285366e-01 -5.08517504e-01 -7.80347824e-01 -1.35106415e-01
7.60514796e-01 -2.05336288e-01 -1.20095468e+00 1.25341856e+00
8.54269505e-01 -1.12039730e-01 -6.09006472e-02 9.84736741e-01
1.14868760e+00 8.26315582e-01 5.87667227e-01 -4.18932527e-01
1.34155774e+00 -1.64792049e+00 -6.53543115e-01 -3.76637518e-01
7.58977234e-01 -7.09351897e-01 9.28571880e-01 7.20617950e-01
-9.97863054e-01 -9.34538901e-01 -1.05663240e+00 -1.62416965e-01
-3.89459252e-01 3.45778763e-01 8.21133614e-01 5.19843638e-01
-7.41704106e-01 5.65488994e-01 -8.25693667e-01 -6.40176311e-02
5.95875859e-01 4.79654729e-01 2.87998319e-01 -2.80500442e-01
-8.92767847e-01 2.54473299e-01 1.03281629e+00 4.36138123e-01
-9.68178511e-01 -7.53378987e-01 -1.25425529e+00 -6.53253347e-02
6.15686834e-01 -3.58455360e-01 1.59115314e+00 -1.52293479e+00
-1.40797043e+00 9.95380282e-01 1.99340448e-01 -5.44389546e-01
4.98433799e-01 -1.78131431e-01 -5.16891293e-02 -1.46000832e-01
3.78733933e-01 1.17711425e+00 8.64523530e-01 -1.24628782e+00
-1.11637187e+00 -3.33302885e-01 3.06534529e-01 3.87610584e-01
1.77548975e-01 -4.21335965e-01 -3.53142828e-01 -2.52004594e-01
2.59540379e-01 -6.95837379e-01 -5.40873468e-01 -2.71178454e-01
-4.70539242e-01 -5.92181861e-01 2.69575775e-01 -6.97213829e-01
1.04587305e+00 -2.01921606e+00 -1.12260439e-01 -2.10667133e-01
-8.98353104e-03 1.77449554e-01 -7.87959695e-02 -2.40911305e-01
6.05300777e-02 -1.37777284e-01 -1.75473407e-01 -1.24163754e-01
1.10239878e-01 2.26267606e-01 3.72267187e-01 2.68042862e-01
1.55180424e-01 9.87419069e-01 -1.29644859e+00 -6.48140609e-01
2.81210989e-01 -8.31110589e-03 -6.05923772e-01 4.68192548e-01
-8.54438484e-01 5.49496889e-01 -4.01458800e-01 9.99408603e-01
8.47395062e-01 -3.71485293e-01 2.08495319e-01 -5.75759470e-01
-1.06762536e-01 -6.90157861e-02 -1.17493272e+00 1.83228946e+00
-1.25609383e-01 -3.68237555e-01 1.37539282e-01 -1.30228162e+00
8.10131550e-01 -2.25436725e-02 6.22729123e-01 -7.30968416e-01
5.11404157e-01 2.74182618e-01 -1.19202957e-01 -7.89423645e-01
5.30064404e-01 2.03367993e-01 -4.36276287e-01 1.49160951e-01
1.59727886e-01 -8.38274509e-02 8.67998123e-01 -2.89155275e-01
4.29682583e-01 6.97501302e-01 3.90393317e-01 -4.03914958e-01
5.22483110e-01 6.34295702e-01 8.46069872e-01 6.25303268e-01
-5.39078474e-01 2.28954837e-01 9.84009951e-02 -1.11224711e+00
-1.17480671e+00 -6.01450384e-01 1.01666808e-01 1.61447692e+00
1.78569928e-01 -2.31248531e-02 -1.18292189e+00 -1.15255010e+00
-2.48270348e-01 4.46345270e-01 -6.30280256e-01 -2.58775819e-02
-5.63023865e-01 -1.07711387e+00 -2.25114062e-01 6.94078922e-01
6.10046089e-01 -1.40785408e+00 -6.95040643e-01 5.88148594e-01
-4.57842618e-01 -8.09863091e-01 -5.59135497e-01 6.58291519e-01
-8.77212942e-01 -1.25772238e+00 -6.55182779e-01 -1.29643714e+00
9.45390105e-01 3.90599184e-02 1.51341736e+00 1.94267988e-01
-1.92786098e-01 -1.92236558e-01 -7.06201792e-01 -6.66673064e-01
-5.49925923e-01 5.36666632e-01 -3.28319967e-01 -2.16628283e-01
7.66289294e-01 2.57153600e-01 -1.07922387e+00 3.93395960e-01
-8.46586883e-01 2.73893565e-01 3.39360058e-01 6.30329967e-01
1.28740978e+00 3.35478514e-01 6.70429885e-01 -1.22374260e+00
1.46655038e-01 -5.12575388e-01 -5.74989438e-01 2.03273952e-01
-3.95714849e-01 -1.80179521e-01 8.63415301e-01 -6.15242660e-01
-8.20393860e-01 7.40362525e-01 -5.33819973e-01 2.74266511e-01
-3.95803690e-01 5.08156478e-01 -2.50175297e-01 3.26014757e-01
5.03636718e-01 -2.50456542e-01 -3.43660772e-01 -4.54034388e-01
4.24876601e-01 4.24180418e-01 3.44327360e-01 -4.96464223e-01
-2.48644184e-02 -5.37412241e-02 -4.87483472e-01 -1.99727699e-01
-1.72502244e+00 -7.36718774e-01 -1.00247180e+00 -1.52217239e-01
1.02510381e+00 -8.74961615e-01 -4.15930688e-01 6.68225706e-01
-6.78156018e-01 -9.64426517e-01 -5.38020551e-01 2.60212570e-01
-4.66785103e-01 1.92365143e-02 -7.55529284e-01 -6.67803764e-01
-6.23947740e-01 -1.19498277e+00 1.20672929e+00 5.06884038e-01
3.01148221e-02 -8.09229136e-01 -2.20206395e-01 5.37931263e-01
8.58854726e-02 3.55035096e-01 8.00903201e-01 -6.46660209e-01
-1.46056563e-01 -1.56454518e-01 -1.51428625e-01 3.97172213e-01
1.81909248e-01 -1.53622046e-01 -8.00753117e-01 -3.37790102e-01
2.36963168e-01 -6.53650582e-01 7.31019557e-01 9.21194136e-01
1.32313251e+00 -2.38783106e-01 -2.52329528e-01 6.12712204e-01
1.55517709e+00 5.14577985e-01 1.08213440e-01 4.29584980e-01
7.32748687e-01 5.31151831e-01 1.12623787e+00 2.56675333e-01
7.02729583e-01 1.51035398e-01 6.73382938e-01 -3.41996133e-01
4.81126010e-02 -2.04990685e-01 2.10469007e-01 1.07642686e+00
3.35884571e-01 -3.02587360e-01 -5.35381079e-01 5.90215087e-01
-1.92048109e+00 -3.58665109e-01 -3.08637589e-01 1.92992735e+00
9.18617487e-01 2.78837919e-01 3.97922039e-01 3.95711392e-01
5.82487226e-01 -2.49873638e-01 -1.10860217e+00 -4.83353406e-01
3.60011131e-01 4.47242558e-02 5.06771028e-01 3.33229959e-01
-1.67379999e+00 9.76761878e-01 5.91509676e+00 6.34166360e-01
-7.28895783e-01 1.08496450e-01 1.51627338e+00 2.06382602e-01
2.67313838e-01 -6.16150320e-01 -1.06607008e+00 4.88866478e-01
9.56047118e-01 4.33544934e-01 3.89370680e-01 1.19350994e+00
1.57421187e-01 -2.67779350e-01 -1.37909722e+00 6.42404795e-01
8.23552981e-02 -7.58288205e-01 -1.23005219e-01 -4.04358268e-01
9.43072557e-01 -7.46463239e-02 -1.34030223e-01 7.48552918e-01
6.18399978e-01 -7.32441723e-01 8.38544548e-01 1.66765302e-01
5.24806678e-01 -8.30506384e-01 9.90319252e-01 6.35925889e-01
-1.36358047e+00 -3.79532874e-01 -7.28296459e-01 -1.91645399e-02
1.23846292e-01 7.65150249e-01 -8.44082296e-01 4.73753780e-01
7.86112785e-01 4.54247564e-01 -5.99092841e-01 1.12070131e+00
-2.28044972e-01 4.79798257e-01 -3.86733592e-01 7.38534480e-02
5.32147944e-01 -4.34330881e-01 -4.20143396e-01 1.09709156e+00
1.34837970e-01 5.76965362e-02 1.00675178e+00 6.22081339e-01
-1.66812956e-01 4.26943570e-01 1.20717278e-02 1.65719822e-01
-5.11667319e-03 1.62275684e+00 -1.34191978e+00 -5.26727915e-01
-2.64922589e-01 1.07112706e+00 4.78872210e-01 -5.09273075e-02
-8.39583099e-01 2.38041386e-01 -1.04537509e-01 -2.14031905e-01
3.19803447e-01 2.91230828e-01 -3.06855619e-01 -6.85526431e-01
-2.29070559e-01 -8.55827928e-01 6.61000609e-01 -5.28649688e-01
-1.34662700e+00 5.43590009e-01 -2.35961840e-01 -8.50217044e-01
-1.38543814e-01 -5.97875714e-01 -2.19728202e-02 3.89482886e-01
-1.60104370e+00 -1.24295938e+00 -6.50437653e-01 1.74343690e-01
1.16961014e+00 4.25254762e-01 8.70166361e-01 4.24968272e-01
-6.48510218e-01 4.83361632e-01 -1.40457094e-01 2.32214510e-01
1.84897229e-01 -1.62814641e+00 3.27226371e-01 4.83541995e-01
-7.74397179e-02 -1.72484159e-01 5.33799112e-01 -6.67512178e-01
-1.07155693e+00 -1.47975302e+00 6.74640715e-01 -2.15778187e-01
2.56191224e-01 -4.31934804e-01 -7.36232996e-01 4.85795945e-01
-1.62944332e-01 1.24388203e-01 7.67103970e-01 -1.13595389e-01
1.45140067e-01 -1.46422714e-01 -1.51672745e+00 -4.93839663e-03
8.75777900e-01 2.54106939e-01 -2.61002600e-01 7.27161884e-01
9.71868277e-01 -7.22305000e-01 -9.04326856e-01 4.40840721e-01
4.73566711e-01 -5.33257186e-01 6.95139885e-01 -4.53833044e-01
4.86365676e-01 -6.78571612e-02 7.77372494e-02 -1.28317821e+00
-4.66446519e-01 -1.27971336e-01 -2.82700155e-02 1.18794477e+00
6.80789113e-01 1.12746246e-01 1.06688738e+00 6.55765057e-01
-2.07062215e-01 -9.62846160e-01 -2.92736083e-01 -2.73398846e-01
2.03098878e-02 -2.78801292e-01 8.47632229e-01 1.00602949e+00
-2.73490101e-01 3.35339665e-01 -6.15934506e-02 5.30563481e-02
6.56916082e-01 4.73831892e-01 5.71915984e-01 -1.20261514e+00
-1.05254091e-01 -1.70787632e-01 2.55486310e-01 -1.42921519e+00
-1.77512333e-01 -9.29085135e-01 5.96613526e-01 -1.83624685e+00
2.42258936e-01 -7.56686807e-01 -4.19635028e-01 5.99083185e-01
-5.20669818e-01 4.03023660e-01 -7.16744363e-02 3.27841286e-03
-1.09542274e+00 -1.25016451e-01 1.51277220e+00 -5.65397859e-01
-6.77499413e-01 2.41402492e-01 -7.12845385e-01 8.49619150e-01
1.08893669e+00 -4.93178278e-01 -5.47693908e-01 -6.28439724e-01
4.17007476e-01 -7.93370008e-02 -1.58950552e-01 -8.76185596e-01
-9.79312602e-03 -9.37964991e-02 5.77015698e-01 -8.89648795e-01
-2.31372476e-01 -9.56348956e-01 6.74450845e-02 5.68918467e-01
-5.13449728e-01 -1.34935915e-01 -3.63390446e-02 3.60906065e-01
2.34350517e-01 -6.02914691e-01 7.47157872e-01 -7.21515238e-01
-6.48886681e-01 6.79396272e-01 5.83252348e-02 -3.43366228e-02
1.01981640e+00 -2.33615279e-01 1.36574030e-01 2.27394387e-01
-1.18709517e+00 6.11649156e-01 2.09606990e-01 2.63914347e-01
1.56124264e-01 -1.19166565e+00 -7.95820832e-01 2.68785894e-01
5.82294725e-02 7.63852239e-01 2.87421227e-01 5.95075786e-01
-6.18176281e-01 1.92682624e-01 1.04028974e-02 -8.27161789e-01
-9.01899815e-01 1.16953743e+00 2.50003397e-01 -6.32338166e-01
-6.14633083e-01 9.98990119e-01 1.91954359e-01 -6.29877388e-01
4.32138979e-01 -9.99989033e-01 -5.15380442e-01 2.51557261e-01
7.09975004e-01 2.23066941e-01 1.67238370e-01 -5.76054394e-01
1.03332780e-01 2.87336409e-01 -2.54280597e-01 7.62354076e-01
1.37421834e+00 -2.13242486e-01 3.70263904e-02 5.16089261e-01
9.49282646e-01 -8.17800820e-01 -1.74055123e+00 -2.54029721e-01
4.08254489e-02 -4.20087390e-02 -8.16074666e-03 -1.16615772e+00
-1.38266706e+00 4.60103661e-01 9.11418676e-01 5.32556713e-01
1.18335736e+00 -8.28015059e-02 1.07772529e+00 6.51816875e-02
5.00708938e-01 -1.54139447e+00 -1.43482059e-01 3.89952183e-01
4.19322640e-01 -1.63172126e+00 -1.34350613e-01 -6.53623104e-01
-5.04437447e-01 9.92831528e-01 7.29020059e-01 -2.46205047e-01
5.56619346e-01 3.78636271e-01 2.56442010e-01 -2.93760002e-01
-1.70246363e-01 -1.33540288e-01 1.39796272e-01 7.59683430e-01
3.66356790e-01 4.46512908e-01 -3.19474965e-01 8.88533056e-01
-3.26562226e-01 2.48685047e-01 -6.95121735e-02 6.06689513e-01
-7.01837063e-01 -8.47713113e-01 -2.28532508e-01 6.59908235e-01
-5.14817774e-01 3.42904143e-02 -1.34078085e-01 5.04388511e-01
7.98322797e-01 1.10087872e+00 1.11121088e-01 8.82183909e-02
3.33342493e-01 -3.24036255e-02 5.10893524e-01 -8.20792913e-01
-8.42543006e-01 3.36980730e-01 8.75899941e-02 -3.12414259e-01
-7.81747043e-01 -5.56941748e-01 -1.37381911e+00 3.58876176e-02
-6.06609643e-01 2.66423613e-01 6.73092127e-01 9.69205856e-01
-3.04172099e-01 7.40366638e-01 7.22896099e-01 -1.03325021e+00
-3.81257623e-01 -1.06214547e+00 -3.48204285e-01 5.28506994e-01
1.27991304e-01 -3.56190532e-01 -6.52795136e-02 4.38445240e-01]
|
[9.713847160339355, 0.5356376767158508]
|
311ad86d-5e10-414d-ae3d-cc9093c31599
|
isar-imaging-analysis-of-a-hypersonic-vehicle
| null | null |
https://ieeexplore.ieee.org/document/9552517
|
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=9552517
|
ISAR Imaging Analysis of a Hypersonic Vehicle Covered With Plasma Sheath
|
In this article, a hypersonic target electromagnetic (EM) scattering echo model combined with the inhomogeneous zonal medium model (IZMM) and the classical scattering
center model (SCM) is proposed with a distributed satelliteborne
array radar as the detection platform. A parallel physical
optics (PO) method is used for multiview inverse synthetic aperture radar (ISAR) imaging of a moving hypersonic target covered
with plasma sheath based on the analysis of high-resolution range
profile in the S-X ultrawideband range, reconstructing 2-D EM
scattering echo data (the target) and motion compensation. The
results show that the surface of the inhomogeneous plasma sheath
flow field is an excitation layer with random and irregular
fluctuation characteristics, which increases the false scattering
centroid of the 1-D range profile of the hypersonic target and
can interfere with and disrupt the radar localization of the
target along the radial direction. In addition, shallow scattering of
EM waves occurs in the plasma sheath, and the average signal
intensity of the target can gradually reduce from 0.5 × 10−5
at 60 km and 20 Ma to 0.1 × 10−5 at 30 km and 20 Ma,
with a fivefold weakening of the overall scattered echo signal.
In particular, the faster the hypersonic target travels at 30-km
altitude, the weaker the imaged scattered echo signal becomes,
with the average intensity of the imaged signal weakening by
approximately threefold from 15 to 25 Ma. This study provides
considerable technical support and data assurance to establish
a synthetic aperture radar (SAR) automatic target recognition
(ATR) database, and the findings of this study can be used as a
reference for fine-structure feature analysis of hypersonic targets
for feature extraction and the classification and identification of
targets.
|
['Bian Zheng']
|
2021-10-09
| null | null | null |
journal-2021-10
|
['motion-compensation']
|
['computer-vision']
|
[ 0.55951536 -0.45245016 0.6713532 0.08018668 -0.34091377 -0.6057434
0.42850733 -0.96770823 -0.13593327 0.59817755 -0.27061206 -0.2914349
-0.5346201 -0.64174074 0.10941912 -1.2759154 -0.27300957 0.46700305
-0.04275592 -0.37128034 0.11698871 1.0885347 -1.482499 0.09945097
0.94781005 1.1970807 0.3281705 0.29250118 0.21524882 0.608422
-0.60302955 0.40454006 0.22276066 -0.24284418 -0.03029599 0.04027914
0.17331447 -0.02019093 -0.41520524 1.3122666 0.35320902 0.17739472
0.81892985 -0.572595 -0.31521863 0.03553086 -0.7547414 0.47421706
-0.24481696 0.14799626 0.26285735 -1.263251 0.35754308 0.43100727
0.5596489 0.20323808 -0.6324157 -0.52199864 -0.63235205 0.29241744
-1.510937 -0.33414856 0.83140606 -0.6069278 0.5798229 0.8166329
0.60536724 0.54229367 0.46740326 -0.17760628 1.2517011 -0.28923965
0.05098277 0.06308025 0.31380314 0.34280717 0.7205279 0.8699413
-0.1824294 -0.45069 0.33746222 -0.18449655 -1.1208928 -0.0128963
-0.7782112 0.63220453 0.1458705 0.56253856 -0.83519703 -0.48036093
-0.26565936 0.347488 0.51495093 0.7464607 -0.34131488 0.08046784
-0.89648056 0.20256896 0.63031965 0.07892109 0.7030821 0.73474896
0.25272587 0.5171916 0.40715358 1.8737763 0.20746674 -0.52281827
-0.18498042 0.2397576 0.48984557 -1.0551507 -0.3718539 -1.0202307
-0.85870534 0.4037731 0.08964312 -0.7557854 -0.7520213 1.2663436
0.37973806 0.32566187 0.5559672 1.4559784 0.6922447 1.0010487
-0.6807187 -0.6702574 1.2016639 -0.18777141 -0.7586403 -0.5731755
0.6450574 -0.94722027 0.04350619 0.14332744 -0.57297474 -0.2821215
-1.0439857 1.2837071 0.24356869 -0.08297443 0.3800451 0.37733865
-0.5921612 0.37257388 -0.5076546 0.08786117 -0.33795404 -0.23775703
-0.17982055 -0.34273574 -1.3164366 0.94395804 -0.26968306 0.68769765
-0.5490097 -1.1109809 -0.5321344 -0.18262415 -0.11538529 -0.4580593
0.6278885 -0.9622394 -1.1957886 0.5328966 0.03924939 -0.18165706
-0.0791418 -0.12816514 -1.1407094 0.44385847 0.07353939 -0.6234354
0.7381179 -1.4620026 -0.21847472 -0.8137625 -0.6883878 0.3050674
0.27795723 -0.07997923 0.71714747 -0.17127989 0.5028762 -0.9545466
-0.20249538 -0.7384387 -0.07146777 0.5944412 1.4350252 -0.38982883
0.7305791 -2.4349492 -0.06468773 0.4649323 0.13113174 0.48085877
-0.27045608 0.56811386 -0.17412247 -0.6692564 -0.48123944 0.81639934
-0.72047615 -0.41140923 -0.7840469 1.005354 -0.3162468 0.5370813
-0.5459794 0.5209738 0.04080411 0.376096 0.08249993 0.22788252
0.17966479 0.4400871 -0.6704896 0.509264 1.6560544 0.02921672
-0.15534213 -0.37132803 -0.90358484 -0.5620339 -1.1613703 0.51372045
-0.1181383 0.7080496 0.94464314 -0.68813705 1.3353887 0.19592562
0.5184648 -0.819691 -0.11451934 0.2945758 0.41845748 -0.8338508
0.28486276 -0.81507653 0.32436717 0.28491557 -0.75556695 -0.43920147
-0.5832754 -0.07992335 1.2248598 -0.49252602 -0.10098014 -0.6305067
0.68296665 0.42806402 0.4236626 0.5936428 0.3520106 0.3178369
-0.14747958 -0.5136176 -0.5787844 -1.036595 -0.611434 0.02164278
0.45301798 0.34916338 -0.14912489 -0.02514697 -0.01090743 0.94010186
-0.269392 -0.4709546 -0.36868784 -1.4078065 0.34422106 -0.14126682
0.69665116 -0.9807258 -0.6931544 -0.07561808 -0.2543508 -0.9901185
0.22752112 -0.00872124 -0.7179487 -1.2768838 -0.30546844 -0.18892443
0.5229912 0.76987165 0.7146588 -0.18927346 -0.5968392 0.5407191
-0.6143525 -0.48888898 -0.26243663 -1.0042499 0.17306435 0.4748858
0.29793 -0.38549948 -0.30678812 0.5644993 -0.78147864 -0.24357082
0.52831846 0.8730164 0.08098235 0.36424065 0.3607865 -0.37390685
0.19234632 -0.439481 -1.1095396 -0.25984955 -0.34693787 -0.39801365
0.5116047 -0.34702337 -1.5489167 -0.7126717 0.38396254 -0.3487349
-0.26689088 0.8951908 0.04515404 -0.61305404 0.8941743 0.7718244
0.20342281 -0.08554197 -0.3401723 0.6487478 0.27654397 0.23479187
1.52561 0.9161459 0.40900412 -1.8332844 -0.88577044 -0.5436347
-0.04552808 -0.5079886 0.61730945 -0.75192106 -0.36054936 0.6850351
-0.84291524 -0.34514448 -0.17744485 1.3362873 0.05406436 0.55682623
-0.29839465 -1.0638267 -0.48654097 -0.36052755 0.55481434 0.20656238
0.08352091 -0.8037795 0.43460423 0.43273434 0.84500384 0.21486583
0.7950202 -0.11083344 -0.6811321 -0.3183437 -0.25373325 0.13603635
0.00727733 -0.2653521 -0.6103702 -0.52796084 1.0819778 0.01826809
0.75847584 0.88441914 0.10121024 -0.02102215 -0.65452087 1.0679989
1.7622175 0.57288027 0.67099756 0.27834734 0.2939029 0.50536525
1.0876929 0.6445715 -0.5948409 0.40785125 0.47408265 -0.26706773
0.27331004 0.90870595 0.38149047 0.4180756 -0.30356646 0.02188292
-0.78842807 0.24209045 -1.2705992 -1.3383827 -0.82043475 2.3298957
-0.04285271 -0.59773546 -1.0964894 -0.26656333 0.50290763 0.318961
-0.27767876 0.19225189 -0.4772746 0.28999865 0.5359415 0.7451554
-0.3769005 0.39540234 5.09138 0.2863744 -1.757922 -0.2181209
-0.3623707 0.14299604 -0.59202415 0.05849507 -0.5640604 0.22280565
0.7985695 -0.21005937 0.31849813 0.32739854 0.4204343 -0.43632203
-0.1612743 1.0843912 0.1340467 -1.1624279 -0.01598259 0.24081717
0.5322446 0.48486876 0.09475204 0.11197852 -0.38343716 -0.69816905
0.2913322 1.0139841 0.8083149 -0.7635767 1.1036465 0.5563484
-0.899863 0.17179897 -0.5146341 -0.19013639 0.58842623 1.3306568
-0.69521725 0.9240481 0.6505596 0.38331357 0.14217253 0.78378695
0.12362099 0.5974072 -0.3534455 0.03189062 0.32034898 -0.96031475
1.4544686 0.64293873 0.948881 0.9794971 -0.06777985 0.9774711
0.59623533 -0.20366146 -0.937198 -0.11449336 0.26895666 1.6410205
-0.3484899 -0.06019635 -0.20218867 0.34338102 -0.7158339 0.7687207
-0.96268696 -0.56993544 0.88843906 0.5430859 0.3764283 -0.29024592
-0.05051436 -1.080776 -0.24242453 -0.4561799 -0.1171739 -1.1228039
-1.1151879 0.9287437 -0.02371501 -1.2512976 0.06595816 -0.6499979
-0.92962515 1.2651485 -1.4030215 -0.64379567 -0.7867861 0.40953857
-0.3141358 -0.34137025 0.9143876 -0.03934552 0.05278181 -0.57904047
0.44052097 -0.17402866 0.28476968 -0.4204453 -0.3314359 0.85422224
-0.48564714 0.27900383 1.1618676 -0.8947051 -1.817523 -1.2327173
0.2985388 -0.06977559 0.75040525 0.08170959 -0.7198671 0.37510124
0.13180579 0.08533152 0.72796446 -0.27507892 0.16087307 -0.0323918
-1.1765382 0.38463223 0.49680272 -0.02697794 -0.83281636 0.327169
0.1617761 -0.04904876 -0.54103905 1.4427766 0.2717676 -1.0796227
0.6181351 -0.37597132 -0.31434894 -0.56499225 -0.4543586 -1.4787289
-0.7178962 -0.20621386 0.34467646 0.587126 0.45551705 -1.0802293
0.6799269 -0.4159453 -0.4217855 -0.41062072 -1.0038359 -0.65085787
-0.33914596 -0.21048576 -0.24624573 1.021765 -0.10975 0.47883397
-0.02662535 1.2769079 1.1829712 0.7412401 0.3484387 -1.5949149
-0.07914703 0.24046315 -0.08829093 -0.5607893 0.03550606 -0.8090591
0.13346358 -1.2580113 -0.09943448 -0.4483506 0.19389217 -0.4739802
0.45060313 0.10133383 -0.09475376 0.36775783 0.2737708 0.78412217
1.0895356 -0.11710428 -0.03842186 0.3418447 -0.07857716 0.7046058
0.51251584 -0.4294887 -0.13015684 -0.07769543 0.12877211 0.44536448
0.369948 -1.4022235 0.14882459 -0.34596744 0.54694676 -0.73994315
0.4864844 -0.82246006 0.5801188 0.6233315 0.82525444 -0.5959105
0.2940717 0.330241 -0.45510298 -0.4366567 1.2868105 -0.08513466
-0.44883546 0.27902183 -0.93338376 -0.02372863 1.288709 -0.08555234
-0.39997864 -0.2882444 -0.51388067 -0.11228126 0.3264612 -0.15949701
0.9422738 -0.8492679 -1.0896987 0.6882705 0.03709448 -0.46952352
1.028896 1.2985768 -0.5045933 0.52692634 -0.0144532 -0.6083126
-1.2619377 0.10991938 0.9332104 0.07692064 -0.72119725 0.63907015
0.53930897 -0.44835192 -0.57746327 0.31200933 -0.41952112 -0.04977717
1.0114772 0.45882088 0.06864767 -0.63586956 -0.37654808 1.0155474
0.24600093 -0.08810961 1.2986648 -0.02396308 -0.58081853 0.29770187
0.6255188 0.6192149 -0.8604344 0.03508942 -0.5130665 -0.3448022
0.63028896 -0.80715317 -0.7888835 0.30719486 0.4272951 0.3526922
1.0544747 0.11194952 0.34478825 0.31596345 0.44962752 -0.5400708
-0.2187265 0.99687725 0.94487184 -0.47545904 -0.1317454 -0.5789358
-0.5616972 1.3369249 0.45854178 -0.18507914 0.94760466 0.7629421
0.426396 -0.71502465 -0.6058569 0.1277541 0.09252023 0.54532784
-0.04295363 0.10892706 -0.24598049 0.46946734 -0.01128927 -0.32576245
0.87754357 0.55702645 -0.98990834 -0.28886506 -1.1487068 0.3956571
-0.24292956 -0.14766411 0.05046436 0.57858586 -0.10176703 0.8226554
0.29909065 -0.07015405 0.39866695 0.01088742 0.30800033 -0.45788574
0.2598619 0.0737736 0.1336876 -0.1594225 -0.31131095 -0.6514446
-1.3367538 -0.04257444 -0.57172066 1.0738639 0.48639318 0.78556824
0.18479688 0.2753584 0.95938057 -0.8492172 -0.627679 -1.2225688
-1.5643917 -0.29250005 0.47419503 -0.7102996 -1.4463503 -0.66037077]
|
[6.81529426574707, 1.0786495208740234]
|
a1edc43b-3059-4d30-bcf4-3898b47c6a8a
|
combining-recurrent-convolutional-and-1
| null | null |
https://openreview.net/forum?id=yWd42CWN3c
|
https://openreview.net/pdf?id=yWd42CWN3c
|
Combining Recurrent, Convolutional, and Continuous-time Models with Linear State Space Layers
|
Recurrent neural networks (RNNs), temporal convolutions, and neural differential equations (NDEs) are popular families of deep learning models for time-series data, each with unique strengths and tradeoffs in modeling power and computational efficiency. We introduce a simple sequence model inspired by control systems that generalizes these approaches while addressing their shortcomings. The Linear State-Space Layer (LSSL) maps a sequence $u \mapsto y$ by simply simulating a linear continuous-time state-space representation $\dot{x} = Ax + Bu, y = Cx + Du$. Theoretically, we show that LSSL models are closely related to the three aforementioned families of models and inherit their strengths. For example, they generalize convolutions to continuous-time, explain common RNN heuristics, and share features of NDEs such as time-scale adaptation. We then incorporate and generalize recent theory on continuous-time memorization to introduce a trainable subset of structured matrices $A$ that endow LSSLs with long-range memory. Empirically, stacking LSSL layers into a simple deep neural network obtains state-of-the-art results across time series benchmarks for long dependencies in sequential image classification, real-world healthcare regression tasks, and speech. On a difficult speech classification task with length-16000 sequences, LSSL outperforms prior approaches by 24 accuracy points, and even outperforms baselines that use hand-crafted features on 100x shorter sequences.
|
['Christopher Re', 'Atri Rudra', 'Tri Dao', 'Khaled Kamal Saab', 'Karan Goel', 'Isys Johnson', 'Albert Gu']
|
2021-05-21
| null | null | null |
neurips-2021-12
|
['sequential-image-classification']
|
['computer-vision']
|
[ 2.60353208e-01 -3.89547199e-01 -2.61033595e-01 -2.56067425e-01
-3.28732044e-01 -3.80906552e-01 6.82340026e-01 -1.51662037e-01
-6.22079492e-01 6.41236544e-01 3.52501608e-02 -8.84536922e-01
-4.05537814e-01 -5.67086577e-01 -1.01356578e+00 -7.52119005e-01
-6.30942941e-01 2.21527386e-02 -3.36285025e-01 -4.76461738e-01
1.19823754e-01 5.25295138e-01 -1.15298653e+00 1.12300731e-01
5.61427832e-01 1.32737577e+00 2.18004286e-01 7.62426734e-01
-5.59101701e-02 1.19777298e+00 -3.26211959e-01 1.23068862e-01
4.36707318e-01 -3.56231689e-01 -8.63934219e-01 -2.56138682e-01
8.17436725e-02 -2.47530147e-01 -9.19508696e-01 5.92417836e-01
4.15374666e-01 4.32675391e-01 4.93349016e-01 -1.05586481e+00
-9.91403341e-01 5.49157917e-01 -3.49773556e-01 6.21554494e-01
-1.06793553e-01 5.69209516e-01 8.45941842e-01 -4.83830035e-01
5.29484451e-01 1.03324592e+00 1.17181611e+00 7.29994059e-01
-1.44378805e+00 -7.79297054e-01 5.05384386e-01 1.10219702e-01
-1.04207945e+00 -3.15679580e-01 5.22737384e-01 -3.51920605e-01
1.82199705e+00 1.95236832e-01 6.69243634e-01 1.44673634e+00
7.63582170e-01 9.00200784e-01 1.01193106e+00 -5.60940616e-02
2.94923365e-01 -5.56073070e-01 3.08997869e-01 8.42536926e-01
-3.20049584e-01 3.69241476e-01 -4.70548213e-01 8.23504233e-04
9.97668445e-01 5.32148778e-01 -2.68869340e-01 -1.56800747e-01
-1.35206592e+00 7.90992022e-01 5.18372178e-01 4.43644106e-01
-5.14941514e-01 7.12755501e-01 6.73103094e-01 7.62806714e-01
3.42282504e-01 7.05665708e-01 -7.57326186e-01 -2.79925853e-01
-8.19348335e-01 1.62935063e-01 6.60327852e-01 8.31595182e-01
4.62857306e-01 7.12826669e-01 -8.22656527e-02 7.13161588e-01
-2.90485173e-01 5.84778368e-01 1.09565246e+00 -1.07375717e+00
2.19997957e-01 1.90168962e-01 -1.63260028e-01 -7.14785635e-01
-7.60495245e-01 -8.18630636e-01 -1.45213127e+00 -2.33201310e-01
1.01426467e-02 -3.06975573e-01 -1.23639929e+00 2.19635391e+00
-4.16743129e-01 6.10776067e-01 1.11307904e-01 4.97260153e-01
2.50270635e-01 1.09295881e+00 -7.83764720e-02 -4.61013496e-01
8.08520317e-01 -9.12780941e-01 -6.43524706e-01 -2.83691853e-01
7.41549194e-01 -2.39310443e-01 9.08503950e-01 2.53812462e-01
-1.26749289e+00 -6.13298118e-01 -9.41793680e-01 -9.28197522e-03
-4.71521914e-01 -7.51779079e-02 8.59434605e-01 1.74330220e-01
-1.59478199e+00 1.24836099e+00 -1.35091639e+00 -2.97889024e-01
1.97201878e-01 5.83600223e-01 7.24549545e-03 2.95266867e-01
-1.40968871e+00 9.73021626e-01 -1.54553233e-02 3.00481915e-01
-1.02306414e+00 -1.16996098e+00 -8.98576975e-01 2.35311668e-02
1.88968360e-01 -7.85701394e-01 1.65524554e+00 -7.63823986e-01
-1.65948176e+00 4.25436467e-01 -1.25151008e-01 -1.40199697e+00
2.28123784e-01 -1.24189861e-01 -6.05923951e-01 -1.86844379e-01
-1.41277075e-01 6.80888593e-01 9.27888930e-01 -3.27869952e-01
-3.42301428e-01 8.66755843e-02 -1.69501901e-01 -2.07347870e-01
-3.20234388e-01 -2.67491043e-01 -4.77690957e-02 -9.21843767e-01
6.97527900e-02 -1.13156259e+00 -6.81746066e-01 -1.86383620e-01
-2.35022515e-01 -2.13923603e-01 7.51087844e-01 -5.99154413e-01
1.59938490e+00 -2.05058265e+00 2.16783062e-01 7.73473382e-02
3.27340007e-01 3.66498798e-01 -4.77265537e-01 6.58685684e-01
-6.44888341e-01 1.27965197e-01 -3.09266180e-01 -3.60945940e-01
1.23788364e-01 3.71528178e-01 -6.63279831e-01 3.78110319e-01
3.29802334e-01 1.31452024e+00 -7.59170353e-01 1.42021090e-01
1.73493117e-01 4.33637708e-01 -3.25229138e-01 -1.65063530e-01
-3.26687366e-01 1.25597760e-01 -2.11711839e-01 1.35616034e-01
1.39755100e-01 -7.22807765e-01 9.12225097e-02 -7.33605474e-02
-2.17930943e-01 5.88267982e-01 -6.11427128e-01 1.83730590e+00
-6.13754272e-01 8.86843503e-01 -3.24841976e-01 -1.48809302e+00
5.80800951e-01 3.24347079e-01 8.73652458e-01 -1.06377590e+00
1.22706629e-01 2.55474091e-01 1.67567015e-01 -4.17168170e-01
2.34625116e-01 -2.13074371e-01 -1.18660234e-01 7.66385317e-01
1.67295292e-01 -4.71162684e-02 1.18015006e-01 1.00525148e-01
1.51165795e+00 -9.57630128e-02 1.67259991e-01 -1.45163476e-01
1.49418414e-01 -3.35988402e-01 3.37937295e-01 1.02739561e+00
-1.10597618e-01 2.57815093e-01 3.49684983e-01 -7.20900595e-01
-1.30737102e+00 -1.06549299e+00 6.19906597e-02 1.12313771e+00
-5.01425505e-01 -3.00663173e-01 -3.78280908e-01 -1.01568684e-01
-1.40945213e-02 5.99571228e-01 -9.77805138e-01 -4.08401340e-01
-8.95622551e-01 -7.95637965e-01 6.74231112e-01 8.53604853e-01
4.46239263e-01 -1.24011886e+00 -8.46786559e-01 6.62025034e-01
-3.89756039e-02 -7.93793321e-01 -7.65626132e-01 7.30225205e-01
-9.73697066e-01 -5.87779880e-01 -8.17499995e-01 -7.90477216e-01
2.29115844e-01 6.62427098e-02 1.15218532e+00 -1.47027746e-01
-3.22587907e-01 4.72138375e-01 1.19556829e-01 -2.43419006e-01
-2.39987284e-01 2.65757173e-01 5.28180063e-01 -1.32699490e-01
2.57233474e-02 -9.22852933e-01 -6.94080949e-01 -9.34831798e-03
-9.68293369e-01 9.98981819e-02 6.19797707e-01 1.07586944e+00
5.18683493e-01 -2.13204250e-01 6.24241233e-01 -4.33115661e-01
9.27678943e-01 -4.28593367e-01 -4.37553793e-01 1.42026037e-01
-6.88408732e-01 5.26854157e-01 7.64097154e-01 -8.73492718e-01
-4.50852424e-01 -3.88579294e-02 -2.05141649e-01 -9.66958642e-01
3.70163202e-01 7.80575752e-01 7.38526940e-01 9.42106470e-02
5.40467739e-01 9.05821919e-01 5.09595394e-01 -2.55808651e-01
3.50174457e-01 8.68246928e-02 5.71657121e-01 -5.27897120e-01
3.62068504e-01 3.69022787e-01 -1.92717854e-02 -7.53321111e-01
-6.93514049e-01 -9.45268050e-02 -2.31656224e-01 2.08984882e-01
6.31294131e-01 -8.13288391e-01 -1.04557323e+00 6.99843407e-01
-1.01110923e+00 -9.30030942e-01 -6.33097112e-01 4.23603207e-01
-9.63174284e-01 -1.46550713e-02 -1.28564179e+00 -7.46519625e-01
-5.34592867e-01 -8.54969144e-01 6.90685332e-01 -8.07403326e-02
-2.14034334e-01 -1.04785073e+00 3.68822590e-02 -7.30565786e-01
8.49804401e-01 1.88797563e-01 1.11475742e+00 -4.67372268e-01
-3.23469162e-01 2.26118956e-02 8.44691619e-02 6.60851657e-01
-2.31497437e-02 -2.62078911e-01 -5.06636679e-01 -4.16792810e-01
3.81466001e-01 -2.20641002e-01 1.01990843e+00 9.12414372e-01
1.45783937e+00 -5.30653417e-01 -4.10741806e-01 7.03587234e-01
1.22261143e+00 7.43793011e-01 4.60110575e-01 1.12077385e-01
3.43442172e-01 5.96653819e-02 -1.11486405e-01 3.55193198e-01
2.94392288e-01 2.25577950e-01 2.26641148e-01 -1.15459614e-01
3.34788620e-01 -7.82428086e-02 6.37419701e-01 1.17957103e+00
-5.63470758e-02 -1.42639831e-01 -9.18644607e-01 4.78323221e-01
-1.92636549e+00 -1.24187219e+00 3.46347421e-01 1.80355489e+00
8.11990380e-01 3.59781533e-01 1.15359187e-01 2.60188133e-02
3.51811796e-01 4.51235384e-01 -1.17663777e+00 -4.83932227e-01
-2.09705248e-01 6.70927286e-01 6.27421141e-01 2.54694134e-01
-1.09346616e+00 6.92247391e-01 6.97524834e+00 6.85675383e-01
-1.72312367e+00 9.28885266e-02 8.86871219e-01 -5.78630209e-01
-4.52833287e-02 -6.70593500e-01 -5.55124402e-01 3.45400989e-01
1.70146966e+00 -2.52099484e-01 8.25614512e-01 4.74693835e-01
3.02789807e-01 5.32330155e-01 -1.42744231e+00 1.23110425e+00
-1.51930511e-01 -1.82193375e+00 -1.15252763e-01 2.66268682e-02
7.98206210e-01 6.15805805e-01 7.06272125e-01 7.84352779e-01
6.12688184e-01 -1.37938571e+00 6.29103839e-01 6.18251562e-01
7.67606258e-01 -4.47409719e-01 2.34844446e-01 3.44430894e-01
-1.18015480e+00 -5.32970309e-01 -5.37062809e-02 -3.93417865e-01
2.25573391e-01 1.77713722e-01 -4.10167933e-01 1.45968810e-01
7.65998065e-01 1.11859405e+00 -1.51359409e-01 5.76944530e-01
4.64096636e-01 6.59805954e-01 -5.16291678e-01 -3.39156806e-01
9.10588562e-01 -1.02458354e-02 2.06653744e-01 1.15288377e+00
3.96984607e-01 3.29108566e-01 -8.26736093e-02 8.26517105e-01
3.79001983e-02 -3.73440951e-01 -6.74748838e-01 -5.36550224e-01
2.17676356e-01 6.18471742e-01 -3.15139085e-01 -4.57522690e-01
-3.62242311e-01 8.92333031e-01 4.00245011e-01 6.74405634e-01
-1.00973701e+00 -3.17263901e-01 1.04262555e+00 -5.80344386e-02
5.74745297e-01 -6.35913908e-01 -6.20880015e-02 -1.09980822e+00
-7.55336583e-02 -9.99786377e-01 3.10857981e-01 -7.81688511e-01
-1.38117218e+00 8.95849526e-01 -3.22235703e-01 -1.36275578e+00
-8.32346380e-01 -7.05751061e-01 -2.84100890e-01 8.87268484e-01
-1.25367641e+00 -5.77692866e-01 3.88511628e-01 7.83426583e-01
7.27254808e-01 6.91786921e-03 8.46201658e-01 1.52930826e-01
-7.07096636e-01 5.16211808e-01 4.46753711e-01 2.03606039e-01
1.13461174e-01 -9.69453752e-01 1.00017893e+00 7.25774646e-01
3.73978689e-02 1.11579108e+00 6.48663521e-01 -4.09433037e-01
-1.65068662e+00 -1.20594573e+00 6.41970932e-01 -2.34828472e-01
1.09831071e+00 -3.69986415e-01 -9.54703450e-01 1.10103095e+00
3.13635468e-01 4.39622924e-02 2.79558301e-01 4.13161330e-02
-5.10251045e-01 -8.74367952e-02 -6.64842427e-01 7.23824203e-01
1.33362079e+00 -8.63937318e-01 -4.00684744e-01 3.47057432e-01
9.61737394e-01 -6.39115691e-01 -8.00845921e-01 3.60815555e-01
7.12194145e-01 -6.42763317e-01 1.15520561e+00 -1.16703153e+00
6.37012005e-01 1.88966647e-01 -8.72971639e-02 -1.57515430e+00
-3.96546751e-01 -1.05140483e+00 -4.05142814e-01 2.95630783e-01
4.81059432e-01 -8.34368169e-01 5.31745791e-01 4.59339559e-01
-4.34747100e-01 -1.22448719e+00 -9.10826504e-01 -1.11974847e+00
4.62831289e-01 -6.71242714e-01 5.58420181e-01 9.79179382e-01
2.94072777e-02 2.74532825e-01 -5.35018504e-01 -6.82353675e-02
8.87116417e-03 8.59642029e-02 1.25660777e-01 -8.82980525e-01
-3.54847521e-01 -9.36430871e-01 -9.74567384e-02 -1.51159072e+00
2.24614754e-01 -7.67092943e-01 2.14996785e-02 -1.21281123e+00
-2.13174328e-01 -3.94015193e-01 -8.09660673e-01 5.97478449e-01
1.58350393e-01 -2.01053068e-01 2.11150914e-01 6.45506084e-02
-4.78517324e-01 4.65974361e-01 9.52789843e-01 -2.26149067e-01
-2.37353176e-01 -7.33348951e-02 -5.35231054e-01 4.18825179e-01
6.90186203e-01 -1.52638584e-01 -5.81926227e-01 -6.91967368e-01
8.57809186e-02 5.44892013e-01 3.31542134e-01 -1.05763555e+00
5.17908931e-01 -2.14936495e-01 2.07930878e-01 -6.18461430e-01
5.75442135e-01 -5.78933775e-01 2.30488852e-01 9.71273839e-01
-7.26235271e-01 7.62312055e-01 3.79095227e-01 6.45146489e-01
3.20088081e-02 4.30085510e-01 5.61416805e-01 -2.50982136e-01
-7.18852878e-01 7.13540673e-01 -6.74942195e-01 -1.77292228e-02
6.32027090e-01 1.06817484e-02 -2.59363532e-01 -6.89157128e-01
-9.37554419e-01 2.26331025e-01 -2.87904054e-01 6.25823557e-01
4.40754116e-01 -1.28938615e+00 -4.15924489e-01 3.44411433e-01
-3.34282964e-01 -3.42893034e-01 4.90936995e-01 1.01561892e+00
-2.01361910e-01 8.56296778e-01 -2.07986802e-01 -7.85795510e-01
-5.18900156e-01 8.05092096e-01 8.22595179e-01 -5.06113529e-01
-7.21825123e-01 8.73720050e-01 1.59849554e-01 -3.75586718e-01
3.41219932e-01 -1.11538923e+00 3.11724961e-01 -1.39522791e-01
4.58355963e-01 9.21760723e-02 2.32223775e-02 -1.54394194e-01
-2.39840046e-01 4.52536732e-01 -2.95394599e-01 -8.85462910e-02
1.73318315e+00 -1.63905751e-02 1.51060864e-01 8.56909513e-01
1.50626075e+00 -8.89943480e-01 -1.49759471e+00 -6.14533067e-01
-1.22048147e-01 4.48665857e-01 -2.08981484e-01 -6.59915626e-01
-1.15439177e+00 1.02023387e+00 6.46963716e-01 3.83039355e-01
1.24617505e+00 -3.22480142e-01 1.07641768e+00 7.28285491e-01
2.34294206e-01 -1.00312448e+00 3.00011903e-01 1.05530989e+00
8.27140033e-01 -9.23772693e-01 -4.41883415e-01 3.77595365e-01
-4.11078334e-01 1.06078494e+00 3.80365163e-01 -3.97785664e-01
1.03274882e+00 5.10945678e-01 -1.71375602e-01 -1.06637880e-01
-1.36813843e+00 8.00543725e-02 -2.76605971e-03 2.67475992e-01
3.61566573e-01 -5.19732293e-03 -6.93390891e-02 5.93957603e-01
-3.64005506e-01 2.04284310e-01 2.92666107e-01 1.05114782e+00
-7.19895735e-02 -6.31532550e-01 1.85629562e-01 7.73051977e-01
-2.69382477e-01 -4.02898520e-01 2.80075014e-01 8.44470620e-01
-3.64357352e-01 4.64396745e-01 5.51517725e-01 -6.12260640e-01
1.57500833e-01 2.31953248e-01 3.47698152e-01 -2.89001316e-01
-7.63817012e-01 -1.62032340e-02 -2.85156727e-01 -8.08857262e-01
-2.71939605e-01 -8.73432755e-01 -1.28493488e+00 -6.31503403e-01
8.06197897e-02 -1.26344681e-01 5.48782349e-01 1.09839463e+00
6.37280643e-01 8.76416862e-01 5.69706202e-01 -8.28221798e-01
-9.54530060e-01 -8.91854405e-01 -4.15901989e-01 1.71209961e-01
9.90925193e-01 -2.74907023e-01 -2.05025271e-01 -1.13146212e-02]
|
[7.430160045623779, 3.3743021488189697]
|
56807d9d-4ffa-438c-b818-19f58a69f31f
|
image-forensics-detecting-duplication-of
|
1802.06515
| null |
https://arxiv.org/abs/1802.06515v3
|
https://arxiv.org/pdf/1802.06515v3.pdf
|
Image Forensics: Detecting duplication of scientific images with manipulation-invariant image similarity
|
Manipulation and re-use of images in scientific publications is a concerning problem that currently lacks a scalable solution. Current tools for detecting image duplication are mostly manual or semi-automated, despite the availability of an overwhelming target dataset for a learning-based approach. This paper addresses the problem of determining if, given two images, one is a manipulated version of the other by means of copy, rotation, translation, scale, perspective transform, histogram adjustment, or partial erasing. We propose a data-driven solution based on a 3-branch Siamese Convolutional Neural Network. The ConvNet model is trained to map images into a 128-dimensional space, where the Euclidean distance between duplicate images is smaller than or equal to 1, and the distance between unique images is greater than 1. Our results suggest that such an approach has the potential to improve surveillance of the published and in-peer-review literature for image manipulation.
|
['M. Cicconet', 'H. Elliott', 'D. Wainstock', 'M. Walsh', 'D. L. Richmond']
|
2018-02-19
| null | null | null | null |
['image-forensics']
|
['computer-vision']
|
[ 5.37642241e-01 -3.53250772e-01 -2.71893889e-01 -3.80499661e-01
-5.27816892e-01 -7.65933454e-01 3.71840209e-01 1.83059946e-01
-5.46413779e-01 5.76305509e-01 -3.59257847e-01 -4.74622756e-01
-2.39886716e-01 -3.90428066e-01 -1.01421022e+00 -5.19536734e-01
-2.92008482e-02 2.93018103e-01 -2.49229390e-02 2.15219140e-01
9.65458333e-01 1.05438077e+00 -1.51009583e+00 -2.88241021e-02
4.75436270e-01 7.39955366e-01 1.27872452e-01 8.70096684e-01
-2.57206917e-01 1.47555530e-01 -8.62193644e-01 -4.20761138e-01
6.51301503e-01 -5.57024181e-01 -7.38577425e-01 3.30423042e-02
8.61831605e-01 -3.79744947e-01 -1.17219836e-02 1.19985175e+00
4.37436372e-01 -1.74045086e-01 5.12929082e-01 -1.41298687e+00
-8.05810988e-01 2.66218543e-01 -8.71745944e-01 4.63017792e-01
3.11466753e-01 1.91917986e-01 3.42475235e-01 -8.04789901e-01
1.16175306e+00 1.02690744e+00 5.54418087e-01 9.56764221e-02
-1.12129557e+00 -8.05169582e-01 -4.71558779e-01 2.17700228e-01
-1.50410187e+00 -2.34905392e-01 4.77979064e-01 -4.62847978e-01
8.78289223e-01 2.15220928e-01 7.23909795e-01 4.94866133e-01
5.26839674e-01 7.38834171e-03 8.08342874e-01 -6.73178196e-01
1.45504966e-01 -6.32363465e-03 -4.14806426e-01 6.28367126e-01
6.84352875e-01 -1.66304365e-01 -6.22235596e-01 -1.28659561e-01
8.11822951e-01 -6.85195252e-02 -1.11722372e-01 -6.62556887e-01
-1.35487390e+00 6.31593645e-01 3.36322963e-01 4.60796207e-01
-3.72395843e-01 1.94607943e-01 2.22190544e-01 3.95879269e-01
8.66542980e-02 1.15117621e+00 -8.86993632e-02 -3.45756739e-01
-1.44700861e+00 3.08315337e-01 5.91208518e-01 9.64342773e-01
7.99573660e-01 -2.89142519e-01 3.01847190e-01 1.41520366e-01
6.88236265e-04 2.55408645e-01 5.45534074e-01 -1.20776820e+00
2.28659809e-01 5.95698774e-01 1.72283929e-02 -1.51835370e+00
-1.74739197e-01 -3.91644686e-02 -5.25853932e-01 3.62567127e-01
4.81432974e-01 1.75992996e-01 -5.53713024e-01 1.19293094e+00
3.73343080e-01 1.08274721e-01 -2.38321871e-01 6.14704847e-01
4.34243679e-01 3.59503239e-01 -2.80776650e-01 -1.07435510e-01
1.06058860e+00 -7.69052327e-01 -6.50159001e-01 2.05931067e-01
6.50848448e-01 -1.14926016e+00 6.50657177e-01 4.58820373e-01
-1.30657673e+00 -4.50526804e-01 -1.42592502e+00 -2.60736436e-01
-7.65846550e-01 2.41979137e-01 5.62874936e-02 6.27537787e-01
-1.00339103e+00 7.74505019e-01 -5.22716761e-01 -3.81654084e-01
5.32961607e-01 2.71651715e-01 -5.81435680e-01 -2.02686116e-01
-6.81669414e-01 9.99693334e-01 3.56233954e-01 -5.94865680e-02
-3.86810601e-01 -7.82771230e-01 -6.95946932e-01 4.46513444e-02
1.96529239e-01 -3.03257585e-01 8.02477539e-01 -1.09630382e+00
-1.09325182e+00 1.22759962e+00 3.84095684e-02 -4.54317033e-01
6.79500818e-01 2.08292395e-01 -2.30654523e-01 2.96258807e-01
3.02867025e-01 8.59570026e-01 1.17368400e+00 -1.00260580e+00
-6.93450987e-01 -5.21432817e-01 -3.06853026e-01 1.45560503e-01
-3.89609754e-01 3.45517159e-01 -4.29614365e-01 -6.20466828e-01
-9.27975122e-03 -1.12048745e+00 6.72357529e-02 6.98066533e-01
-3.70619655e-01 3.18848453e-02 1.06019974e+00 -5.86512327e-01
1.12445056e+00 -2.37600303e+00 -1.78014001e-04 4.00248706e-01
2.75838763e-01 4.91306067e-01 -2.33425498e-01 3.66636544e-01
-4.46874559e-01 4.69184339e-01 -3.23760509e-01 1.16444640e-02
-3.87786001e-01 -1.91577971e-01 2.03575771e-02 8.83068144e-01
2.62452424e-01 6.28166676e-01 -8.93855035e-01 -4.66196775e-01
2.24210143e-01 3.97394329e-01 3.44061339e-03 -2.82111168e-02
1.99617103e-01 1.01075709e-01 -2.28089504e-02 4.77205873e-01
1.13128734e+00 -1.06354572e-01 2.52009898e-01 -5.37247099e-02
-5.07191479e-01 -8.20627138e-02 -1.32560587e+00 1.64911902e+00
1.01890832e-01 1.14455009e+00 -1.92289054e-01 -8.64727914e-01
9.29360807e-01 -1.27044305e-01 4.24524158e-01 -5.11993766e-01
8.75068530e-02 3.91148418e-01 1.67768836e-01 -5.09734213e-01
7.04582810e-01 4.00144398e-01 3.21824878e-01 6.57725811e-01
-1.95368528e-01 -3.01803678e-01 6.23083472e-01 1.91513360e-01
1.01783240e+00 8.85513723e-02 2.12305516e-01 -1.37401566e-01
3.50354373e-01 1.27969950e-01 2.32877582e-01 8.35711062e-01
-4.21008050e-01 7.95408487e-01 7.48008549e-01 -4.32500988e-01
-1.48214233e+00 -6.35570109e-01 -2.02212185e-01 6.08523071e-01
3.31231087e-01 -9.52222422e-02 -8.10617089e-01 -3.47262233e-01
3.58771443e-01 5.86401582e-01 -5.15050352e-01 -3.37946713e-01
-6.68297350e-01 -1.23274468e-01 7.93309927e-01 2.31797874e-01
4.22822893e-01 -9.55051780e-01 -1.09209275e+00 2.50841100e-02
2.73594379e-01 -8.39470506e-01 -6.57191277e-01 -1.81548223e-02
-7.81580031e-01 -1.31374586e+00 -1.02800727e+00 -9.29821074e-01
9.43174779e-01 5.78683317e-01 7.40025103e-01 2.21785307e-01
-7.71011770e-01 6.35290593e-02 -3.60118635e-02 -3.08510125e-01
-4.31306779e-01 8.80731866e-02 -8.49745572e-02 -2.58269399e-01
5.92414260e-01 -5.32295145e-02 -5.46534717e-01 2.63000160e-01
-1.08997321e+00 -2.85455674e-01 5.11627853e-01 5.64006269e-01
5.15340567e-01 2.16146093e-02 2.76822537e-01 -2.73425132e-01
9.21346128e-01 -2.38208085e-01 -7.33179450e-01 4.59339052e-01
-6.73027158e-01 -8.55165571e-02 4.58259463e-01 -5.26499689e-01
-5.58851063e-01 2.98123956e-01 6.09176219e-01 -7.69934893e-01
-1.68328792e-01 4.62784559e-01 2.00018793e-01 -5.85695267e-01
6.64561749e-01 1.73623368e-01 3.32433730e-01 -1.38367891e-01
4.97315913e-01 6.69619083e-01 6.86164379e-01 -9.59755704e-02
6.41122878e-01 5.00732362e-01 1.41312733e-01 -8.10365260e-01
1.16755843e-01 -4.97712523e-01 -8.40999603e-01 -3.36050063e-01
7.88821697e-01 -5.49195230e-01 -4.67695147e-01 5.55431128e-01
-1.18946803e+00 1.95102677e-01 -1.47778064e-01 6.09245241e-01
-3.64704370e-01 7.09202945e-01 -8.96424279e-02 -3.23051155e-01
-1.64683789e-01 -1.44632792e+00 7.88310587e-01 3.39150816e-01
-4.07514989e-01 -3.95970255e-01 -9.75833610e-02 2.00434282e-01
4.66261744e-01 1.72335684e-01 9.62761879e-01 -6.76919580e-01
-5.89473426e-01 -5.99916697e-01 -3.86479199e-01 1.54697835e-01
1.60424396e-01 6.09200716e-01 -2.94753999e-01 -3.91329736e-01
-1.52669027e-01 -1.31337628e-01 4.17402923e-01 3.92414004e-01
1.31214511e+00 -1.14462212e-01 -4.14720535e-01 5.48952281e-01
1.17083514e+00 4.96681720e-01 6.69616520e-01 7.48040676e-01
4.08573180e-01 5.29356182e-01 4.57482845e-01 3.83439809e-01
8.58502835e-03 5.44436693e-01 3.30247611e-01 -1.33084543e-02
-1.28523409e-01 1.68707401e-01 -2.04785496e-01 3.23332042e-01
2.05212012e-01 -3.06755513e-01 -1.02270484e+00 6.78931475e-01
-1.44118249e+00 -8.30290973e-01 -1.43148303e-01 2.41790199e+00
6.93498850e-01 8.09434354e-02 -2.12686554e-01 -4.97436933e-02
1.11592162e+00 -1.35788321e-01 -6.99647665e-01 -7.19998062e-01
-7.08646700e-02 2.36613862e-02 9.41282988e-01 -5.58932051e-02
-9.33759689e-01 6.25761092e-01 6.72426367e+00 6.58896089e-01
-1.38772213e+00 -4.38030720e-01 5.86538374e-01 -1.20127529e-01
7.37101957e-02 -6.40128180e-02 -4.40569311e-01 5.77016711e-01
7.34287679e-01 -2.69619048e-01 3.93383741e-01 7.22343087e-01
1.23572737e-01 -4.37912226e-01 -1.01095045e+00 1.06071711e+00
5.04050076e-01 -1.61060834e+00 7.87777302e-04 2.22282156e-01
8.60160530e-01 -4.00016665e-01 2.69569248e-01 -4.29942936e-01
1.56146763e-02 -1.01024342e+00 5.23517609e-01 4.99085426e-01
8.82457376e-01 -8.21776688e-01 6.00563526e-01 -1.27515271e-02
-6.72096193e-01 3.12798351e-01 -3.27264696e-01 2.14337900e-01
-2.02501789e-01 3.77392590e-01 -9.42287743e-01 3.48645866e-01
8.59229088e-01 6.44645154e-01 -7.66802192e-01 1.35568452e+00
1.06692933e-01 9.16636214e-02 -3.49496841e-01 -2.67392043e-02
1.48125812e-01 -2.88519025e-01 5.67136824e-01 1.11919463e+00
5.44916272e-01 -4.88291681e-01 -1.35044903e-01 8.84780407e-01
-3.62237304e-01 1.31031126e-01 -9.21793103e-01 -4.63947773e-01
5.92205346e-01 1.13138914e+00 -1.02202535e+00 -4.30785209e-01
-2.66768277e-01 9.53379989e-01 -2.41277386e-02 1.41155988e-01
-5.55892467e-01 -1.03411269e+00 4.23888415e-01 -9.64400694e-02
6.17485106e-01 -2.60618120e-01 -4.81642663e-01 -6.32514894e-01
1.77057862e-01 -9.92401779e-01 1.47014663e-01 -1.02823794e+00
-8.21139753e-01 2.43693829e-01 8.96500647e-02 -1.15061963e+00
-3.04315180e-01 -6.27812266e-01 -4.49400365e-01 8.26778710e-01
-1.07757568e+00 -6.51224673e-01 -2.44872957e-01 2.08181411e-01
3.39953274e-01 -4.58886325e-01 4.46293145e-01 2.06064522e-01
-4.65597808e-01 5.86149573e-01 5.99523485e-01 8.33696276e-02
9.67953563e-01 -7.82834053e-01 2.91952729e-01 8.46863210e-01
-1.98325321e-01 8.32989693e-01 6.21909916e-01 -7.32708037e-01
-1.53307736e+00 -7.51632094e-01 1.09916663e+00 -4.58908863e-02
5.55010140e-01 -9.74911789e-04 -9.06790376e-01 4.63637084e-01
3.71924490e-01 -1.33910835e-01 5.15402615e-01 -6.14728272e-01
-3.33947182e-01 1.29211158e-01 -1.43760037e+00 6.64347887e-01
5.87641776e-01 -4.23075914e-01 -4.46701646e-01 2.76847124e-01
1.46297619e-01 -3.66000652e-01 -8.94018769e-01 -6.32560328e-02
6.87806368e-01 -6.93505943e-01 9.35211182e-01 -6.61479414e-01
8.49213839e-01 -3.76584858e-01 2.14726955e-01 -1.04371107e+00
-4.25584763e-02 -4.45126057e-01 2.88267195e-01 7.96699822e-01
1.88125491e-01 -3.56424630e-01 7.33308554e-01 5.84815621e-01
1.08639471e-01 -4.97220218e-01 -1.15784979e+00 -7.14224994e-01
9.94697735e-02 1.57957897e-01 6.69042349e-01 1.12628496e+00
-1.38036773e-01 -2.14495987e-01 -5.78537472e-02 -2.05521554e-01
3.42148423e-01 1.48822227e-02 9.36055839e-01 -9.60103214e-01
4.59126025e-01 -9.39219654e-01 -9.70453560e-01 -5.41429043e-01
6.69583604e-02 -6.76473081e-01 6.48144856e-02 -1.33607244e+00
1.12744085e-01 -9.04896408e-02 1.21439584e-01 3.08649838e-01
2.36124396e-01 2.68140227e-01 1.45835236e-01 3.32289577e-01
-2.95069635e-01 6.26348332e-02 1.00985551e+00 -2.48114079e-01
-3.72832455e-02 -5.56154370e-01 -4.27722692e-01 4.79443401e-01
8.95082414e-01 -8.31842244e-01 -9.94411707e-02 -3.61563355e-01
3.53091806e-01 -2.47416556e-01 1.74367204e-01 -9.59707320e-01
4.56967801e-01 -1.05139203e-01 5.89497209e-01 -6.58057988e-01
-1.08170383e-01 -8.51574659e-01 2.00885415e-01 6.51413143e-01
-6.85853124e-01 6.50923491e-01 2.69268066e-01 4.45943534e-01
-9.18169916e-02 -7.42365777e-01 7.34736383e-01 -3.03547025e-01
-5.59666276e-01 -6.77185282e-02 -5.66064835e-01 -2.31874064e-01
1.33442068e+00 -6.02395058e-01 -3.21512610e-01 -1.62852690e-01
-1.02115482e-01 -5.33312671e-02 7.18061030e-01 5.68489134e-01
6.28196776e-01 -1.04453719e+00 -5.86723566e-01 3.03597003e-01
1.37425110e-01 -2.42343843e-01 1.34746339e-02 5.97694695e-01
-1.22659612e+00 2.23540097e-01 -6.46693647e-01 -5.84283650e-01
-1.51429439e+00 7.28632450e-01 3.92323405e-01 5.67692444e-02
-3.83032203e-01 6.23235762e-01 -4.16929394e-01 -3.62381637e-01
3.07991356e-01 -2.17923410e-02 1.25508651e-01 -1.14927210e-01
5.47806978e-01 6.21778607e-01 3.07039857e-01 -7.21922159e-01
-3.65176231e-01 4.49820429e-01 -3.88267964e-01 -4.78923470e-02
1.04855239e+00 1.03751615e-01 -4.04814333e-01 1.58534944e-01
1.60187697e+00 -3.41994643e-01 -8.65596354e-01 9.61794853e-02
3.43945362e-02 -9.25598502e-01 5.39401099e-02 -4.08418745e-01
-1.04937065e+00 7.26221800e-01 9.24307406e-01 2.05651253e-01
7.24026680e-01 -3.22452813e-01 3.93768013e-01 6.02006018e-01
-2.36059010e-01 -1.43923700e+00 1.39221653e-01 1.84507221e-01
9.63946819e-01 -1.10805738e+00 4.75305498e-01 -2.69278456e-02
-2.89146692e-01 1.45010722e+00 5.31856179e-01 -1.14657439e-01
4.91589189e-01 1.62400588e-01 -2.18428075e-02 -3.04907292e-01
-1.45528033e-01 4.41502273e-01 9.37230065e-02 5.27649999e-01
3.38700324e-01 -2.77160555e-01 -5.36071420e-01 -4.94207233e-01
-1.97899997e-01 3.77194345e-01 7.03113317e-01 1.54155278e+00
-2.92277634e-01 -7.19705701e-01 -6.96616173e-01 5.92070878e-01
-2.80174047e-01 -8.66502002e-02 -6.82590008e-01 8.02374780e-01
1.57230839e-01 6.21621490e-01 4.82449293e-01 4.88309599e-02
1.79632425e-01 -1.22122325e-01 5.04846931e-01 -5.78000732e-02
-6.08562589e-01 -1.21234506e-01 -4.28152531e-01 -2.83480942e-01
-5.63947976e-01 -7.91022301e-01 -9.06405270e-01 -4.71313924e-01
-2.51927733e-01 -3.01951230e-01 1.27963746e+00 7.10066915e-01
7.90830195e-01 2.72264779e-01 4.57932353e-01 -9.07382429e-01
-4.97797310e-01 -6.15815818e-01 -5.28075933e-01 4.25677180e-01
1.22265428e-01 -5.54026425e-01 -3.59880179e-01 4.09567505e-01]
|
[12.035745620727539, 0.8549147844314575]
|
79c7c612-48f0-4fc9-bb36-6695735e55cd
|
unsupervised-image-matching-and-object
|
1904.03148
| null |
http://arxiv.org/abs/1904.03148v1
|
http://arxiv.org/pdf/1904.03148v1.pdf
|
Unsupervised Image Matching and Object Discovery as Optimization
|
Learning with complete or partial supervision is powerful but relies on
ever-growing human annotation efforts. As a way to mitigate this serious
problem, as well as to serve specific applications, unsupervised learning has
emerged as an important field of research. In computer vision, unsupervised
learning comes in various guises. We focus here on the unsupervised discovery
and matching of object categories among images in a collection, following the
work of Cho et al. 2015. We show that the original approach can be reformulated
and solved as a proper optimization problem. Experiments on several benchmarks
establish the merit of our approach.
|
['Patrick Perez', 'Yann Lecun', 'Kai Han', 'Francis Bach', 'Jean Ponce', 'Minsu Cho', 'Huy V. Vo']
|
2019-04-05
|
unsupervised-image-matching-and-object-1
|
http://openaccess.thecvf.com/content_CVPR_2019/html/Vo_Unsupervised_Image_Matching_and_Object_Discovery_as_Optimization_CVPR_2019_paper.html
|
http://openaccess.thecvf.com/content_CVPR_2019/papers/Vo_Unsupervised_Image_Matching_and_Object_Discovery_as_Optimization_CVPR_2019_paper.pdf
|
cvpr-2019-6
|
['single-object-discovery']
|
['computer-vision']
|
[ 0.38673636 0.06548534 -0.465888 -0.5092344 -0.4405604 -0.49993882
0.7754321 0.18152449 -0.535016 0.32062954 -0.0683106 -0.03137114
-0.21099372 -0.4310087 -0.41867235 -0.8268449 -0.06307613 0.26528516
0.2827017 0.14087428 0.28643408 0.43445745 -1.6852611 0.09887974
0.6566105 0.9721534 0.1356323 0.16099073 -0.07432003 0.81745315
-0.36393112 -0.4218334 0.44436064 -0.19585422 -1.0348736 0.7709958
0.58850205 0.03943158 -0.14381093 1.2736195 0.24128775 0.14501855
0.5985842 -1.2803603 -0.31878242 0.4384933 -0.6451692 0.06543349
-0.04643017 -0.0980339 1.1738597 -0.9886193 0.4782397 0.9152097
0.54474765 0.32236642 -1.3708395 -0.46194738 0.20388378 0.33686522
-1.3597077 -0.484503 0.78443176 -0.7049057 0.49926642 0.19028652
0.29951653 0.7516678 -0.33262035 1.1757255 1.1756532 -0.6557089
0.24365515 0.16684954 0.48235315 0.6533888 0.3113713 -0.0103396
-0.36119136 -0.04304716 0.48800132 0.36976403 -0.15945467 -0.77794874
-1.2081859 0.92265487 0.5325235 0.35745895 -0.12548092 -0.27429268
0.3966787 0.19565964 0.57008827 0.8126812 -0.41924936 0.0352904
-0.88437474 0.03619884 0.7621425 0.91314 0.64402646 -0.13681686
0.07995732 1.0033327 0.14517173 0.08938468 0.39148098 -0.99239826
0.04006265 0.8367332 -0.1589881 -0.8287969 -0.28843778 -0.50816226
-1.0482936 0.1131539 0.4433223 0.20529744 -0.8800618 1.5297823
0.28054315 0.25163004 -0.23756751 0.81734353 0.70719826 0.304757
0.01974578 -0.40150413 1.0343373 -1.1744668 -0.6585403 -0.19815321
0.35099682 -0.69404066 0.6953846 0.540599 -0.6825527 -0.45143896
-0.7804546 0.13687566 -0.4352827 0.08889268 0.92568284 0.46686098
-0.87988967 0.4622089 -0.8545244 -0.52832615 0.79333586 0.42473668
-0.48030832 -0.16166975 -0.7155332 0.7999787 0.49945232 0.22202992
-0.54649246 -0.33158118 -0.81213963 -0.14964241 0.8351231 -0.4081385
1.2462314 -1.1690258 -1.2309549 1.2008469 0.02424527 -0.58077914
0.40209126 -0.26499566 -0.2005088 0.06377024 0.1359011 0.45762497
0.8959284 -1.0903926 -0.8801716 -0.4113765 0.08650117 0.01846764
-0.69672257 0.40362173 -0.5922319 -0.75921714 0.25357726 -1.0605686
-0.44742975 0.03751041 -0.4426886 -0.5865674 0.8688166 -0.2944757
1.0939162 -2.2506545 0.27151096 0.11752899 0.560959 0.40511134
0.30618396 0.20072402 -0.15377015 0.16065691 -0.82077354 -0.5782948
-0.10028629 0.28369108 -0.22883548 0.7328406 0.38464528 0.75664216
-0.9993538 -0.5257406 0.17678355 0.18372066 -0.34585205 0.3033951
-0.04769072 0.49333194 -0.35959932 0.7184165 0.3225153 -0.5200707
0.15241922 -0.03685621 -0.11806045 0.16042778 -1.2228004 1.4942966
0.03122406 0.8901638 0.11536601 -1.712636 0.7203559 0.13081066
0.7953206 -0.36524558 0.12393726 0.1142143 0.09753274 -0.5967089
0.2792276 -0.17652366 -0.1686933 0.39679393 0.18968675 0.12919247
0.42754874 0.2014975 0.9375022 -0.18834415 0.48586795 -0.38140595
0.5490505 0.01880484 0.5418715 0.8481818 -0.2547424 0.558535
0.22076833 -0.5072817 -0.87992287 -0.8802764 -0.3134694 1.0353726
0.05402412 -0.49993333 -0.61390686 -0.9662838 -0.03263219 0.03511204
-0.54240704 -0.05892477 -0.61316466 -0.5479678 0.32765394 0.47997156
0.56581086 -1.1596348 -0.60766876 -0.13753088 -0.11821507 -1.264161
-0.21829087 0.30259427 -0.87186974 -1.3391498 -0.7813746 -1.0459414
0.9233151 0.6038634 1.3662794 0.37760657 -0.52507126 0.5653478
-0.49548033 -0.5869728 -0.11014009 0.23123945 0.2225826 0.4047875
0.44172573 -0.4756584 -0.21235964 0.4525603 -1.164487 0.05751744
0.56958556 0.8832585 0.71205 0.19507518 0.5122422 -1.2858837
0.2184268 -0.5840509 -0.88899523 0.08305642 -0.85456103 -0.02528224
0.49375626 -0.26982942 -0.76321054 0.5165191 0.09698227 -0.40891117
-0.4108024 0.5735208 -0.37963492 -0.13463712 0.24546617 0.23226944
0.01939509 -0.80426025 0.35743552 0.70579857 0.58852774 -0.32993668
1.0995196 0.58069557 -0.09226546 -0.8852964 -1.1901605 -0.9865943
-1.0008205 -0.11990763 0.8282454 -0.7259527 -0.47265983 0.41319287
-0.8606679 -0.24263942 -0.26797178 0.41972524 -0.46171921 0.5264569
-0.31684473 -0.67503595 0.05576169 -1.0178928 0.8871318 0.27784702
0.05188004 -0.9514106 0.02349612 0.41555876 0.15332174 0.02138353
0.52394074 -0.88096493 -0.62121415 -0.31743136 -0.22014911 0.5793736
0.42376494 -0.07983355 -0.94326824 -0.23874368 0.18786296 -0.40230757
1.0921463 0.08613198 1.4405564 -0.28677016 -0.14734927 0.7221927
1.3901587 -0.26286408 0.28150862 0.45766038 0.78140765 0.8334979
0.7621659 0.22968031 0.16223311 0.81010556 0.5488323 -0.14057365
0.07683358 0.02852167 0.02870171 0.84039557 -0.24421754 0.25798512
-0.9909442 0.59084886 -2.04684 -1.1299024 -0.19413976 2.3039432
0.8192719 0.08676452 0.31830427 0.38277158 0.7174452 0.10725296
-0.2813354 0.16310553 -0.30494508 0.15854985 0.48104867 0.0249091
-1.5294898 0.8800668 6.620429 0.6381728 -1.0757214 0.07304895
0.5735841 0.26058063 0.26985162 -0.02512871 -0.62540454 0.47457078
0.43783355 -0.10204133 0.14587262 0.95918083 -0.11383227 -0.0790529
-1.1957445 1.1477753 0.20565543 -1.2111928 -0.31937978 0.1486576
1.0158997 0.08756406 0.04009383 0.29924148 0.19286382 -1.027895
0.47653642 0.20555055 0.49855986 -0.44669056 0.7769054 0.58048016
-1.2317351 -0.13934869 -0.38313293 -0.28545764 -0.05281878 0.6253301
-0.61969995 0.43455413 0.8553605 1.1218497 -0.7879589 1.5385609
-0.40173757 1.0277191 -0.33195668 0.24974896 0.17213404 -0.19424613
0.43795434 1.3317472 -0.34258342 0.11904417 0.662695 0.5618731
-0.23937954 0.0114479 -0.6246596 -0.12647875 0.18603534 1.4871038
-1.0287001 -0.45635417 -0.57033026 0.86887985 0.6515062 0.2345356
-0.513419 -0.21873882 0.47226676 -0.00937053 0.3737054 -0.34330007
-0.32517526 -1.3368257 0.11107489 -0.974522 0.58218056 -0.15282078
-1.4685155 0.3020576 0.01836065 -1.4300333 -0.09283759 -0.677822
-0.34386963 0.33026338 -1.644489 -1.0072253 -0.45433882 0.5657569
0.8181006 -0.25262025 0.81847817 0.5531213 -0.78268445 0.33947188
0.19856735 0.567761 0.7161457 -1.4999928 0.08394101 0.99967605
0.81356543 0.5214182 0.56729406 -0.21010327 -1.3390967 -1.0910193
0.9674285 -0.50828195 0.6810489 -0.4307693 -1.088298 0.73031986
0.21423033 0.2651836 0.72235173 0.15213045 -0.47167686 -0.16583882
-0.7273242 0.4322123 0.9751212 -0.49710014 -0.6178254 0.5166544
0.40268624 -0.21597633 -0.7698733 0.53130203 0.24388611 -1.0133104
0.964991 -0.68613654 0.41239074 -0.20364964 0.06957339 -1.0320989
-0.25001788 -0.7350082 -0.22159462 1.377723 0.12745567 -0.5080164
0.7211697 0.42008325 -0.07267889 -0.7982652 -0.611258 -0.96914506
-0.20909819 -0.45898983 0.1049176 1.1362057 0.04722361 0.4193893
-0.3888375 -0.01874833 1.0447987 0.27171382 0.862489 -1.5531253
-0.18097667 -0.74986863 -0.5773055 -1.1498677 0.4121113 -1.086565
0.33103454 -1.3592268 0.43560314 -0.54057944 -0.64208305 0.45646098
-0.3285802 0.6591588 0.3411071 0.7077507 -1.0160801 0.3599707
0.7114294 -0.29535112 0.11890242 0.4160327 -0.7100247 1.019858
0.8764676 -0.5351689 -0.14120305 -0.34161034 0.09541015 -0.5733296
0.4628899 -0.77516884 0.44930297 -0.2600555 -0.01407124 -0.45430893
0.02711992 -1.1796453 -0.28278494 0.2691741 -0.47875664 -0.1518214
-0.19102634 0.61256474 -0.5554143 -0.38842747 0.7864294 -0.11969291
-1.0022272 0.42533332 -0.16255611 0.07162285 1.0285808 -0.10348874
0.1381829 -0.11253224 -0.5837053 0.392888 0.39845276 0.42266512
0.46363804 -1.2284019 -0.6120334 0.01914128 0.37102428 0.24859613
-0.28660768 0.98263466 -0.06728029 0.47603047 0.03835301 -0.9149466
-1.387557 0.6149127 0.165633 -0.34464175 -0.6878609 0.70523334
0.3289875 -0.33191857 0.60761327 0.21987426 -0.4651619 -0.02976724
0.52447903 0.2750248 -0.08492688 -0.6715687 -0.32212216 0.42964312
-0.10619096 0.33496037 1.5260547 -0.18088154 -0.24806328 0.6872919
1.149868 -0.06868187 -1.147969 -0.633268 0.79258734 -0.4341368
-0.06117785 -0.46182427 -1.1994855 0.7376353 0.45307618 0.518067
1.1907996 0.3195007 0.42623278 0.5923118 0.34311214 -1.2358148
0.06970685 0.56375396 0.4759416 -1.798432 0.15702847 -0.5683963
-0.3228349 0.99952686 0.5281257 -0.30889952 0.7612048 0.13939714
-0.10178141 -0.19360068 -0.48692903 -0.66128105 0.5758414 0.57083726
0.464138 -0.12298968 -0.35834652 0.56129336 0.20112711 -0.14495288
0.26334053 1.1519413 -0.5224146 -1.1796572 -0.19953631 0.5625426
-0.55601853 -0.05735746 -0.52391785 0.6496145 0.19565906 0.9634901
-0.01181195 -0.15306872 0.16938008 -0.00929147 0.3208673 -0.9848263
-0.5866484 -0.02207817 -0.20152728 -0.51802915 -1.0237072 -0.502856
-0.9812153 0.31494087 -0.477425 0.14624444 0.51325226 1.1985068
0.08495618 0.24296507 0.9506793 -0.7001878 -0.69339395 -0.9291156
-0.6471594 0.68410236 0.3202368 -0.7975061 -0.39651847 0.53280777]
|
[9.465348243713379, 2.3677542209625244]
|
106770b0-0a19-4af6-9050-c5afb5db1817
|
graph-neural-network-aided-exploratory
|
2304.04497
| null |
https://arxiv.org/abs/2304.04497v1
|
https://arxiv.org/pdf/2304.04497v1.pdf
|
Graph Neural Network-Aided Exploratory Learning for Community Detection with Unknown Topology
|
In social networks, the discovery of community structures has received considerable attention as a fundamental problem in various network analysis tasks. However, due to privacy concerns or access restrictions, the network structure is often unknown, thereby rendering established community detection approaches ineffective without costly network topology acquisition. To tackle this challenge, we present META-CODE, a novel end-to-end solution for detecting overlapping communities in networks with unknown topology via exploratory learning aided by easy-to-collect node metadata. Specifically, META-CODE consists of three iterative steps in addition to the initial network inference step: 1) node-level community-affiliation embeddings based on graph neural networks (GNNs) trained by our new reconstruction loss, 2) network exploration via community affiliation-based node queries, and 3) network inference using an edge connectivity-based Siamese neural network model from the explored network. Through comprehensive evaluations using five real-world datasets, we demonstrate that META-CODE exhibits (a) its superiority over benchmark community detection methods, (b) empirical evaluations as well as theoretical findings to see the effectiveness of our node query, (c) the influence of each module, and (d) its computational efficiency.
|
['Won-Yong Shin', 'Ming Li', 'Cong Tran', 'Yu Hou']
|
2023-04-10
| null | null | null | null |
['community-detection']
|
['graphs']
|
[ 1.27579376e-01 2.17379570e-01 -2.04673246e-01 3.78295705e-02
-2.83514678e-01 -7.41512299e-01 4.55246866e-01 6.88576043e-01
-1.76136643e-01 5.79686880e-01 8.31668545e-03 -4.11708206e-01
-4.68686193e-01 -9.15807426e-01 -5.29258370e-01 -5.16790807e-01
-9.21831369e-01 8.23140740e-01 1.16833396e-01 1.80494800e-01
2.84901738e-01 4.36349928e-01 -7.92830944e-01 -2.33128741e-01
9.88740981e-01 6.21565998e-01 -1.89028516e-01 3.97017479e-01
-1.04071125e-01 6.94971681e-01 -1.14920422e-01 -3.99547786e-01
2.29096919e-01 -2.32448690e-02 -1.00928712e+00 2.72113889e-01
4.04440053e-02 -6.56282678e-02 -5.43559372e-01 1.16947532e+00
1.93237633e-01 -7.25335628e-02 3.53805542e-01 -1.57215488e+00
-5.01631737e-01 9.42673028e-01 -8.65310848e-01 1.70159906e-01
2.74368733e-01 2.18095347e-01 1.42415643e+00 -8.42318118e-01
1.02287579e+00 1.17056406e+00 9.41858470e-01 3.02905198e-02
-1.70507455e+00 -6.95031941e-01 2.21308336e-01 -1.18756190e-01
-1.76084208e+00 -2.59894431e-01 1.07819939e+00 -5.80401719e-01
5.07366180e-01 -1.15044087e-01 8.01583230e-01 9.46439505e-01
-3.11281621e-01 6.45755529e-01 6.48371637e-01 -3.83604579e-02
3.28151584e-01 1.59130186e-01 -5.45391813e-02 8.99688184e-01
4.65478122e-01 -1.41920939e-01 -3.15892786e-01 -5.74352622e-01
8.52540970e-01 1.19849525e-01 -2.83132941e-01 -9.40397382e-01
-1.22028494e+00 1.06092966e+00 1.05066192e+00 6.92986622e-02
-3.02470595e-01 2.08559707e-01 5.52572370e-01 3.58114570e-01
5.14590085e-01 3.97280216e-01 -1.32469729e-01 2.60557860e-01
-9.00911331e-01 -1.53375685e-01 1.24949312e+00 8.74500275e-01
8.08098316e-01 -2.33488932e-01 3.28985900e-01 5.74826539e-01
5.44516444e-01 -9.47301835e-02 -2.38819301e-01 -5.98579884e-01
5.84025145e-01 9.47465837e-01 -3.88352662e-01 -1.42288697e+00
-3.95857215e-01 -7.51998782e-01 -1.39662004e+00 -2.99129456e-01
3.67204636e-01 -1.65326729e-01 -2.10664451e-01 1.81976748e+00
5.62500000e-01 3.45972598e-01 -2.15202659e-01 6.09483063e-01
6.22022808e-01 2.03858972e-01 -2.01539904e-01 -5.12863928e-03
1.10206544e+00 -9.40781236e-01 -1.75883502e-01 3.07351421e-03
5.26904821e-01 -1.06774248e-01 5.73930383e-01 2.71138698e-02
-8.05114865e-01 -2.39526015e-02 -8.47183943e-01 2.42369249e-01
-3.22720140e-01 -1.33842945e-01 8.50146174e-01 4.44450289e-01
-1.33741140e+00 6.18947268e-01 -7.88814127e-01 -6.24300241e-01
8.51884484e-01 4.74947631e-01 -5.05823910e-01 -9.77869481e-02
-8.71106625e-01 -8.60628299e-03 5.12260556e-01 1.82255179e-01
-8.86436760e-01 -6.19185448e-01 -9.85695302e-01 4.76229161e-01
9.56511140e-01 -7.50298738e-01 3.84032369e-01 -7.56885648e-01
-8.50143790e-01 6.16123378e-01 -3.35264653e-02 -4.25870389e-01
4.72852290e-01 2.40028217e-01 -1.89768121e-01 5.63644886e-01
2.75419086e-01 4.86432761e-01 4.46161687e-01 -1.25351322e+00
-2.37074524e-01 -2.53122985e-01 -2.74163974e-03 -3.73242944e-02
-5.25278986e-01 -2.27331549e-01 -6.47626936e-01 -5.53111792e-01
4.75663334e-01 -9.62055445e-01 -4.81018096e-01 6.17936790e-01
-1.01553822e+00 -3.26713413e-01 7.58953393e-01 -4.47046697e-01
1.26580548e+00 -1.99412668e+00 1.78301990e-01 9.93757844e-01
1.22353649e+00 -1.83501363e-01 -3.08965564e-01 6.38955891e-01
-5.82600832e-02 5.86472452e-01 -3.60355288e-01 -2.70663798e-01
-3.08026016e-01 -3.14866722e-01 8.23235959e-02 6.00693226e-01
3.75946492e-01 7.77811944e-01 -1.20827103e+00 -6.69779181e-01
-1.48181707e-01 3.99626791e-01 -7.56239712e-01 -1.03424251e-01
1.06370181e-01 3.08903962e-01 -5.29832482e-01 1.05884159e+00
5.35243452e-01 -1.04896414e+00 8.52928877e-01 -1.59776032e-01
2.25524053e-01 3.81470434e-02 -1.29647386e+00 1.34259701e+00
1.22289017e-01 7.16365397e-01 5.91482341e-01 -1.15291762e+00
7.35854268e-01 1.86543748e-01 7.74457574e-01 -2.50584427e-02
-1.09541327e-01 5.44027016e-02 1.72950953e-01 -2.59323478e-01
1.40446633e-01 3.82947385e-01 1.55638292e-01 8.29775512e-01
-7.79686943e-02 5.29172063e-01 4.01532918e-01 9.79310989e-01
1.64861751e+00 -5.67586780e-01 1.54418916e-01 -3.11634570e-01
3.50185305e-01 -1.45020857e-01 5.21959305e-01 6.85467720e-01
-2.79280066e-01 2.34553143e-01 1.00608599e+00 -2.67017037e-01
-9.63901699e-01 -1.19445646e+00 1.17137782e-01 6.69077098e-01
3.23240399e-01 -4.50627953e-01 -4.85121131e-01 -8.45493078e-01
3.61242443e-01 -1.49541959e-01 -6.56339347e-01 1.01489602e-02
-4.42148209e-01 -6.29984021e-01 4.86776173e-01 3.58083397e-01
3.77440304e-01 -9.10506964e-01 1.16783097e-01 1.76757678e-01
-3.70125055e-01 -1.10854805e+00 -7.76358187e-01 1.13169849e-01
-1.17621315e+00 -1.56543589e+00 -2.85316825e-01 -9.29506660e-01
1.14503264e+00 4.92509991e-01 1.13447797e+00 6.60809457e-01
-2.67928064e-01 3.68285447e-01 -8.35275128e-02 4.61584359e-01
-9.39429402e-02 5.11714458e-01 1.41482979e-01 1.90425396e-01
1.67256787e-01 -1.17199862e+00 -6.92737877e-01 2.76120454e-01
-6.22619271e-01 -2.18904287e-01 8.05134833e-01 8.08036208e-01
2.23235726e-01 3.34733665e-01 6.13103151e-01 -9.84471500e-01
8.57135475e-01 -9.92795110e-01 -7.09454894e-01 1.95480272e-01
-7.74945617e-01 -3.09543852e-02 4.03214276e-01 -4.56988126e-01
-3.68567914e-01 -4.04396281e-02 4.04849470e-01 -4.55436677e-01
3.42972368e-01 1.05109346e+00 -1.65397018e-01 -2.19978690e-01
4.40109313e-01 2.27432638e-01 4.49474275e-01 -3.82602721e-01
1.67690679e-01 4.26207066e-01 3.81052911e-01 -5.79124629e-01
1.28090632e+00 6.59875691e-01 2.56091356e-02 -8.76018167e-01
-4.56949085e-01 -7.67870724e-01 -8.74863744e-01 -2.48874873e-01
3.21277827e-01 -9.98500645e-01 -1.01362991e+00 2.43085306e-02
-8.39157701e-01 -6.11251332e-02 6.99968785e-02 3.01247388e-01
1.07101701e-01 7.81615853e-01 -9.79965150e-01 -7.78769493e-01
-4.83367950e-01 -8.83281589e-01 5.36956429e-01 -6.97462484e-02
-1.44944459e-01 -1.39558864e+00 2.46061031e-02 3.77495497e-01
2.46143803e-01 5.39994717e-01 1.03076410e+00 -8.30832601e-01
-1.13522875e+00 -3.07117641e-01 -7.39951730e-01 -1.77631319e-01
1.72041103e-01 2.07118332e-01 -5.35578012e-01 -8.01206112e-01
-7.86333323e-01 -2.89968371e-01 7.30409145e-01 1.65191010e-01
1.10415280e+00 -3.69874835e-01 -8.07343960e-01 7.26522923e-01
1.50483799e+00 -4.51320767e-01 3.71226788e-01 8.74718577e-02
8.16213489e-01 5.44205487e-01 -6.52239993e-02 3.76277834e-01
4.58616138e-01 2.27390140e-01 7.04918563e-01 -2.21393436e-01
2.54123986e-01 -5.59267223e-01 7.65358359e-02 8.67729604e-01
1.66886210e-01 -2.20561206e-01 -1.06326187e+00 7.73089290e-01
-1.78413796e+00 -9.32319522e-01 -1.75433099e-01 2.11048794e+00
7.13591993e-01 3.32091063e-01 5.09364724e-01 -4.85335514e-02
1.05214977e+00 1.23987399e-01 -9.31735933e-01 2.59273678e-01
-3.06130778e-02 -2.12806925e-01 2.10195020e-01 1.52886912e-01
-1.12131560e+00 6.10343575e-01 5.77803278e+00 5.43179333e-01
-8.31570625e-01 -6.70171827e-02 6.82064533e-01 8.16722512e-02
-3.50743741e-01 3.84672105e-01 -3.47990394e-01 3.72364610e-01
3.78114134e-01 -1.21006608e-01 6.12720907e-01 9.10643995e-01
2.74913684e-02 2.59637535e-01 -1.13659132e+00 6.52749658e-01
-2.78096825e-01 -1.60057032e+00 -8.15229192e-02 6.51503921e-01
6.37487352e-01 3.05406779e-01 -3.07673156e-01 1.88075632e-01
5.85729480e-01 -9.93419707e-01 3.19992453e-01 2.25339741e-01
9.54730570e-01 -6.56763375e-01 5.24518728e-01 3.26436549e-01
-1.64193070e+00 -2.41929621e-01 -2.47137398e-01 2.20662802e-01
6.03016652e-02 9.92603421e-01 -9.09740210e-01 5.45706809e-01
5.68855226e-01 9.86906528e-01 -5.47163248e-01 1.36458492e+00
-6.99968636e-02 8.19208324e-01 -5.98974943e-01 -8.25971887e-02
2.81998068e-01 -3.57243001e-01 8.55378687e-01 9.18996811e-01
6.67516068e-02 -4.38437164e-01 4.75766540e-01 1.52124071e+00
-6.46395683e-01 -1.10858589e-01 -5.88974357e-01 -5.43192923e-01
9.97559071e-01 1.55093682e+00 -1.21474409e+00 8.78018737e-02
-2.20946401e-01 6.92767024e-01 8.13383043e-01 4.96392220e-01
-4.64332104e-01 -7.22235501e-01 4.76458013e-01 5.08846581e-01
3.04121315e-01 -1.76495999e-01 6.33398443e-02 -1.04466939e+00
1.15992337e-01 -6.47888362e-01 5.64144194e-01 -2.73563474e-01
-1.54519331e+00 4.24811959e-01 -3.19433272e-01 -1.03619277e+00
2.08874136e-01 -1.85276628e-01 -1.04047120e+00 5.92236102e-01
-1.19240355e+00 -1.04958045e+00 -3.97448808e-01 3.17608744e-01
-8.41859728e-02 -2.78776139e-01 4.63017732e-01 4.32817250e-01
-9.73736584e-01 7.12186098e-01 2.96141952e-01 7.57053792e-01
2.93991894e-01 -1.29202962e+00 4.47670013e-01 7.80731142e-01
-2.29465682e-02 9.34980929e-01 2.26524428e-01 -9.30094659e-01
-1.39778268e+00 -1.18808889e+00 7.28024840e-01 -1.94716543e-01
1.07685459e+00 -7.45457232e-01 -9.73427415e-01 6.29331887e-01
-1.72693372e-01 3.22087795e-01 6.34128988e-01 6.01773918e-01
-4.38056260e-01 -1.73531964e-01 -1.26234722e+00 5.93306243e-01
1.19753432e+00 -6.09977245e-01 1.03787161e-01 3.50666404e-01
7.70933330e-01 2.87919134e-01 -1.03531647e+00 1.85555339e-01
4.74803239e-01 -6.81366265e-01 1.01904976e+00 -4.45184290e-01
5.17806947e-01 -3.15503329e-01 2.61900425e-01 -9.75539207e-01
-4.71909672e-01 -7.88172603e-01 -4.04612273e-01 1.30435324e+00
4.60906863e-01 -6.39370799e-01 1.17265892e+00 2.06047446e-01
5.29998422e-01 -8.80311191e-01 -8.81295383e-01 -5.65952480e-01
-2.32483283e-01 9.69192088e-02 5.10905206e-01 1.43957543e+00
7.63359517e-02 5.84934533e-01 -2.99332058e-03 3.18254620e-01
1.10837269e+00 9.53768492e-02 7.91268945e-01 -1.87261498e+00
-2.71400154e-01 -6.39335513e-01 -3.37726057e-01 -1.02578843e+00
8.51762742e-02 -1.11598992e+00 -2.86691993e-01 -1.51317108e+00
6.99711680e-01 -9.27919924e-01 -1.84622779e-01 2.85863429e-01
-7.49731250e-03 3.04935239e-02 -1.96031690e-01 4.98462647e-01
-1.05013728e+00 5.20680189e-01 9.32058990e-01 -3.26899171e-01
-2.13241994e-01 1.24089785e-01 -9.34471548e-01 4.57301140e-01
5.02289057e-01 -5.90767443e-01 -4.17542189e-01 -3.86025421e-02
6.64993763e-01 2.53978670e-01 6.41173959e-01 -8.14717293e-01
5.89270890e-01 2.72611678e-01 1.81744099e-01 -6.57145739e-01
4.61339280e-02 -7.57125199e-01 5.38910031e-02 7.73751616e-01
-1.71392724e-01 -5.57055101e-02 -1.60376862e-01 1.15360653e+00
-9.72876549e-02 2.79889926e-02 5.05418181e-01 1.32064372e-02
-2.57344127e-01 6.16505146e-01 -2.54860282e-01 2.22775236e-01
8.70113075e-01 -5.03556311e-01 -2.43937969e-01 -4.65150148e-01
-6.84300959e-01 7.81331360e-01 5.72419882e-01 2.10850462e-02
6.00861669e-01 -1.17126691e+00 -7.93222189e-01 7.17071295e-02
2.34366924e-01 1.96867868e-01 2.66102180e-02 1.22123492e+00
-5.33763111e-01 -1.68677419e-02 2.90914208e-01 -8.76090169e-01
-1.33263099e+00 5.83051860e-01 3.12387854e-01 -6.25912428e-01
-7.27466643e-01 8.40354025e-01 1.43889207e-02 -8.55777025e-01
5.56035042e-01 1.29327387e-01 -4.62795459e-02 8.37800279e-03
1.30114658e-02 2.78154880e-01 -4.15112048e-01 -2.54829675e-01
-4.51674461e-01 9.92085263e-02 -2.25853488e-01 2.80163527e-01
1.42023528e+00 -2.41171643e-01 -4.70212311e-01 1.29752278e-01
1.16661441e+00 -7.82216266e-02 -1.10720944e+00 -6.81353629e-01
3.56299549e-01 -3.51549089e-01 -7.93899372e-02 -5.18853664e-01
-1.40010798e+00 6.00596368e-01 1.44420862e-01 4.02564377e-01
8.52302134e-01 9.84344259e-02 5.85044324e-01 6.71509504e-01
3.08245093e-01 -8.58194590e-01 4.10170734e-01 3.07725728e-01
4.60983604e-01 -1.43206906e+00 2.14590997e-01 -6.18403554e-01
-9.62087512e-02 1.02114582e+00 4.94732440e-01 8.31957757e-02
1.11370242e+00 -5.94505556e-02 -5.01735568e-01 -6.56647086e-01
-8.72225881e-01 7.72525966e-02 -1.68636329e-02 5.33304334e-01
7.24866986e-02 -2.13212267e-01 3.92744035e-01 3.76867294e-01
2.06088975e-01 -3.17457616e-01 4.26377773e-01 7.62808084e-01
-2.67957598e-01 -8.15801680e-01 -4.54364009e-02 8.48309338e-01
-1.51105538e-01 -1.83408752e-01 -8.14832866e-01 9.43829894e-01
-3.92728478e-01 8.71738970e-01 7.39059448e-02 -4.24233943e-01
-2.13665634e-01 -4.60963726e-01 -9.98087078e-02 -6.14834666e-01
-4.99768466e-01 -1.19161576e-01 6.03083856e-02 -3.72937143e-01
-1.54795825e-01 -6.01964533e-01 -9.08678710e-01 -7.90924132e-01
-6.91830933e-01 4.35296685e-01 4.01983023e-01 6.91373467e-01
6.58549488e-01 2.86820889e-01 9.47938323e-01 -6.52745247e-01
-4.66937274e-01 -8.88037741e-01 -7.15025723e-01 2.88858533e-01
2.91245401e-01 -5.71696222e-01 -7.11408675e-01 -3.98195386e-01]
|
[7.155932426452637, 5.977251052856445]
|
8821419c-f5b9-4dd3-95ae-f8f3cca2796f
|
interpretable-machine-learning-for-science
|
2305.01582
| null |
https://arxiv.org/abs/2305.01582v3
|
https://arxiv.org/pdf/2305.01582v3.pdf
|
Interpretable Machine Learning for Science with PySR and SymbolicRegression.jl
|
PySR is an open-source library for practical symbolic regression, a type of machine learning which aims to discover human-interpretable symbolic models. PySR was developed to democratize and popularize symbolic regression for the sciences, and is built on a high-performance distributed back-end, a flexible search algorithm, and interfaces with several deep learning packages. PySR's internal search algorithm is a multi-population evolutionary algorithm, which consists of a unique evolve-simplify-optimize loop, designed for optimization of unknown scalar constants in newly-discovered empirical expressions. PySR's backend is the extremely optimized Julia library SymbolicRegression.jl, which can be used directly from Julia. It is capable of fusing user-defined operators into SIMD kernels at runtime, performing automatic differentiation, and distributing populations of expressions to thousands of cores across a cluster. In describing this software, we also introduce a new benchmark, "EmpiricalBench," to quantify the applicability of symbolic regression algorithms in science. This benchmark measures recovery of historical empirical equations from original and synthetic datasets.
|
['Miles Cranmer']
|
2023-05-02
| null | null | null | null |
['interpretable-machine-learning']
|
['methodology']
|
[-1.38760209e-01 -3.24874103e-01 -2.35585675e-01 -3.17750961e-01
-8.21490228e-01 -3.97449106e-01 4.97369081e-01 -1.26610905e-01
-1.62618726e-01 8.78855407e-01 -7.46908784e-01 -5.19606113e-01
-1.15353160e-01 -6.26189351e-01 -9.99658883e-01 -8.59624922e-01
-4.09790814e-01 1.00029564e+00 -1.17279189e-02 -4.05892789e-01
2.31018588e-01 8.19656789e-01 -1.63757956e+00 1.02200843e-01
7.94347644e-01 7.82009661e-01 -2.95954496e-01 8.41921747e-01
1.32238165e-01 8.32963705e-01 -5.29786050e-01 -2.56538570e-01
-6.90604374e-02 -4.27479804e-01 -5.06370425e-01 -1.02076757e+00
-8.15577433e-02 2.10917786e-01 -1.09898247e-01 8.18425834e-01
5.39240837e-01 2.35992204e-03 6.91703022e-01 -1.55679464e+00
-2.03251213e-01 8.72923136e-01 -5.28752744e-01 1.59613550e-01
3.26385051e-01 4.43823993e-01 8.09219003e-01 -6.37485743e-01
5.09034157e-01 1.39461672e+00 9.99411404e-01 1.63247332e-01
-1.84625435e+00 -9.94147062e-01 -5.70848286e-01 8.41141045e-02
-1.76236153e+00 -4.54903603e-01 4.17609334e-01 -4.96481031e-01
1.44889855e+00 7.02188015e-01 6.18482590e-01 1.29398739e+00
4.74530995e-01 5.96261799e-01 7.60251820e-01 -2.36241624e-01
4.62353855e-01 -4.67192940e-02 1.26868576e-01 9.10874784e-01
4.91957478e-02 5.53097785e-01 -6.52487278e-01 -9.07806396e-01
7.10219264e-01 -4.80085075e-01 -3.30935530e-02 -4.05810297e-01
-1.08268678e+00 8.77284110e-01 9.31619257e-02 9.67256550e-04
-2.50499785e-01 6.99641347e-01 5.83958149e-01 1.87538624e-01
3.54395837e-01 9.81910229e-01 -8.78766358e-01 -6.86289907e-01
-9.66659307e-01 8.16842556e-01 1.04119468e+00 7.78995156e-01
7.64900804e-01 3.92229766e-01 1.50964960e-01 6.41709685e-01
8.12440440e-02 5.15524328e-01 5.51848233e-01 -1.33526790e+00
-3.60393584e-01 6.72309101e-01 -2.37023726e-01 -8.46597075e-01
-6.66233838e-01 -4.94047165e-01 -6.49080098e-01 4.13218945e-01
1.19155273e-01 -2.73632020e-01 -4.92218584e-01 1.40956140e+00
4.47408706e-01 4.77752537e-01 4.03170362e-02 6.94174349e-01
6.83613420e-01 6.83181763e-01 4.39739525e-02 -2.91580647e-01
6.88460588e-01 -7.85385907e-01 -2.21049428e-01 3.40930969e-01
7.92660296e-01 -4.07066584e-01 9.64058995e-01 7.17095017e-01
-1.15254009e+00 -3.30572397e-01 -1.01411963e+00 -1.04260676e-01
-6.32305264e-01 7.56022856e-02 1.10513592e+00 3.77693504e-01
-9.15844142e-01 1.08122170e+00 -1.14899182e+00 2.84190446e-01
2.50207007e-01 6.68903828e-01 3.28398347e-02 4.78563488e-01
-1.09366107e+00 7.76291013e-01 3.15871924e-01 -1.19912475e-01
-7.35975027e-01 -1.19890296e+00 -8.01105142e-01 8.46705120e-03
3.61262143e-01 -6.85883462e-01 1.24955213e+00 -1.12536323e+00
-1.86847568e+00 9.17686164e-01 -3.36017787e-01 -4.82381254e-01
3.88846457e-01 2.78450977e-02 -5.08999467e-01 -2.46500164e-01
-4.45371121e-03 2.41685271e-01 8.74738991e-01 -7.14153945e-01
-2.03314815e-02 -5.54963708e-01 -9.38504159e-01 -4.82184410e-01
3.11978281e-01 4.42950726e-01 -4.69613343e-01 -6.98450863e-01
-2.30769172e-01 -1.12599587e+00 -4.13005173e-01 -2.69183815e-01
-2.25218892e-01 -2.34202906e-01 6.40663028e-01 -5.79218209e-01
1.26811290e+00 -1.97529137e+00 7.36832917e-01 5.82906425e-01
2.75500149e-01 3.55997160e-02 2.68225700e-01 1.63703993e-01
-6.28335476e-01 2.33878046e-02 -5.16068757e-01 4.68828902e-02
8.39771032e-02 6.78037852e-02 -4.39612031e-01 5.98990798e-01
4.11218345e-01 1.09060442e+00 -8.91634166e-01 -4.65588778e-01
-1.73460409e-01 3.65052253e-01 -4.76871967e-01 4.31405418e-02
-7.85447121e-01 4.21090901e-01 -6.36291862e-01 8.40641499e-01
5.03217638e-01 -4.67652410e-01 2.63909250e-01 3.67285252e-01
-3.37021619e-01 -1.53671466e-02 -1.02179849e+00 1.64731634e+00
-3.84528488e-02 6.00684583e-01 -1.16531566e-01 -8.97626162e-01
1.18753767e+00 -2.68988520e-01 5.29366016e-01 -2.89317936e-01
3.46207023e-01 5.01777411e-01 -9.91609618e-02 -2.73856372e-01
4.74564701e-01 4.82586503e-01 -3.09620470e-01 3.81621629e-01
-1.49586368e-02 -5.82850337e-01 2.20482484e-01 -1.97955035e-02
1.33886707e+00 8.55590940e-01 3.83283615e-01 -3.56708825e-01
6.93929613e-01 4.90533829e-01 3.88616830e-01 5.64163506e-01
5.55953026e-01 8.16465542e-02 9.52223837e-01 -5.88690519e-01
-1.13503718e+00 -1.00927591e+00 -4.63860542e-01 1.46565449e+00
-3.56956661e-01 -5.67611277e-01 -4.26176995e-01 1.04932651e-01
5.76756418e-01 1.15271008e+00 -5.22886932e-01 -3.63865614e-01
-5.89097381e-01 -9.02430713e-01 1.10576594e+00 2.50327289e-01
1.03130743e-01 -1.12961602e+00 -5.98353505e-01 1.49506569e-01
3.21946293e-01 -5.09329140e-01 2.20570192e-01 5.14342606e-01
-7.17675745e-01 -9.62218881e-01 -2.82739937e-01 -3.77243072e-01
2.37107024e-01 -5.86395979e-01 1.33480620e+00 3.39183658e-02
-8.68800104e-01 -1.01093724e-01 5.09915464e-02 -4.23347354e-01
-8.33382845e-01 2.54414707e-01 3.75057943e-02 -3.88107777e-01
5.18425226e-01 -8.45690429e-01 2.94840276e-01 1.62452981e-01
-3.55769366e-01 3.72611620e-02 1.03211991e-01 1.07806718e+00
7.88006783e-01 -3.12742978e-01 1.14127338e-01 -3.35202187e-01
5.49812078e-01 -6.57478809e-01 -1.44455194e+00 2.85852551e-01
-6.10651314e-01 6.05396867e-01 7.33759820e-01 -5.09322405e-01
-5.13360918e-01 3.01790476e-01 7.78525919e-02 -9.10201371e-01
2.92087615e-01 5.02006590e-01 2.31887400e-01 -2.93695450e-01
8.59813094e-01 4.63458061e-01 3.16842973e-01 -3.54639351e-01
2.49594569e-01 4.60564435e-01 7.64517605e-01 -1.15473056e+00
4.14236546e-01 -1.51330158e-01 4.45893168e-01 -9.15723920e-01
-1.61430255e-01 7.04844994e-03 -2.67971396e-01 -3.94115746e-02
4.92029488e-01 -8.65092099e-01 -1.32595479e+00 6.84012771e-01
-1.11603391e+00 -8.31595242e-01 -1.33695886e-01 2.06547633e-01
-7.89164364e-01 -1.10831022e-01 -3.70826095e-01 -6.48548365e-01
-3.07528079e-01 -1.21754444e+00 1.22839141e+00 1.94101587e-01
-7.08179653e-01 -6.45364702e-01 4.42648351e-01 -2.76730120e-01
2.48598844e-01 7.42732286e-01 9.99451756e-01 -6.55021429e-01
-3.30927938e-01 -4.13020290e-02 4.21837457e-02 8.55931491e-02
-6.88447356e-01 9.10679340e-01 -7.66560376e-01 7.77726695e-02
-2.79415518e-01 -5.16590536e-01 6.39158607e-01 3.27862501e-01
1.56393528e+00 -6.50368109e-02 -8.15948606e-01 1.32764828e+00
1.04657876e+00 8.77984464e-02 6.03975117e-01 4.29490596e-01
4.75257605e-01 2.59110451e-01 5.26506722e-01 5.92869520e-01
6.42462894e-02 7.53397226e-01 2.69094467e-01 2.49517784e-02
7.59425938e-01 1.63604572e-01 4.99701768e-01 4.74716812e-01
-3.69888365e-01 3.52805763e-01 -1.21335614e+00 -1.21498026e-01
-2.03234649e+00 -9.62124050e-01 -5.97158432e-01 2.01675153e+00
1.43525088e+00 -8.94575939e-02 2.71495432e-01 -1.75461680e-01
4.04866874e-01 -3.95697802e-01 -1.12662518e+00 -7.70105779e-01
-2.01229379e-01 9.09682333e-01 6.76136076e-01 4.27444428e-01
-6.98718488e-01 1.24176419e+00 7.15970230e+00 1.11093855e+00
-1.45479059e+00 -1.75919011e-01 6.98889077e-01 -3.40478778e-01
-4.93472107e-02 -4.83024940e-02 -8.26040864e-01 4.42388326e-01
1.50890803e+00 -4.95891541e-01 9.25578654e-01 1.26299524e+00
2.54170988e-02 -3.08498647e-02 -1.20224893e+00 7.43863165e-01
-2.29951367e-01 -1.55939937e+00 -4.99569058e-01 -2.37191573e-01
4.44412798e-01 3.68717283e-01 -7.64838755e-02 6.73982263e-01
6.94319189e-01 -1.46214354e+00 7.15998888e-01 7.73505628e-01
9.03663397e-01 -1.21242952e+00 2.07425743e-01 3.48675042e-01
-6.77277029e-01 1.02083735e-01 -2.18225002e-01 -1.06877768e-02
-3.50741804e-01 3.58782113e-01 -6.95412278e-01 3.11080039e-01
9.60029900e-01 8.10086071e-01 -7.32264340e-01 4.78641570e-01
-6.94273710e-02 6.22663498e-01 -5.43308735e-01 -3.77596915e-01
-1.61003947e-01 -4.68745857e-01 6.36313260e-01 1.33480346e+00
1.89004749e-01 2.44032647e-02 -3.31670463e-01 1.63578856e+00
3.67095500e-01 -1.10719919e-01 -5.13140976e-01 -2.62319177e-01
4.78453189e-01 1.09589279e+00 -3.65951747e-01 -3.48967671e-01
2.33766258e-01 6.30611479e-01 2.24252328e-01 2.79723227e-01
-1.40526211e+00 -5.94638705e-01 8.75592530e-01 -9.18765028e-04
3.23658347e-01 -4.84821826e-01 -3.95214826e-01 -9.85461950e-01
-5.31674743e-01 -1.64530551e+00 1.00751430e-01 -1.11552000e+00
-7.48024285e-01 3.59408796e-01 1.41488746e-01 -5.51325679e-01
-7.97420144e-01 -8.95686269e-01 -3.64238381e-01 9.66613352e-01
-8.39012325e-01 -6.69048727e-01 -2.40689829e-01 6.14994526e-01
-4.03627194e-02 -6.22833610e-01 1.08266914e+00 -1.81024313e-01
-1.08034873e+00 7.19034910e-01 5.46332955e-01 -3.33722234e-01
4.89894152e-01 -9.84367073e-01 6.11940563e-01 2.41517857e-01
-3.89656305e-01 8.80567491e-01 1.02301586e+00 -7.87302077e-01
-1.94779730e+00 -8.71172011e-01 4.79179561e-01 -2.92603165e-01
1.16837013e+00 -4.27617669e-01 -1.06794500e+00 8.37227941e-01
-3.13089937e-01 -7.78897256e-02 4.00565922e-01 1.01246700e-01
-3.51450384e-01 -2.48344392e-02 -9.49146450e-01 7.18397260e-01
8.09652507e-01 -1.10455364e-01 -1.95739090e-01 4.01259482e-01
7.83698678e-01 -8.13036621e-01 -1.25666404e+00 2.74617940e-01
7.40691245e-01 -9.23964918e-01 1.24418104e+00 -7.54837811e-01
8.22031558e-01 -2.15290979e-01 -2.79825367e-02 -9.19931054e-01
-7.03585297e-02 -1.24549198e+00 -5.33195615e-01 6.78931713e-01
6.00595176e-01 -9.76785600e-01 4.21155542e-01 4.24267530e-01
1.44548723e-02 -8.74479949e-01 -8.14034700e-01 -9.32731986e-01
1.15704626e-01 -3.85587692e-01 1.00161064e+00 1.09406066e+00
-2.72483528e-02 2.62377530e-01 -4.88511622e-02 4.50006686e-02
5.09362578e-01 1.73435152e-01 1.33528352e+00 -1.14185226e+00
-1.04516792e+00 -9.30375218e-01 -3.31052870e-01 -4.70897734e-01
5.41998684e-01 -1.09548104e+00 -2.20645472e-01 -1.58017963e-01
-2.59897977e-01 -4.07643586e-01 1.50931999e-01 6.93286240e-01
1.84154779e-01 -1.71759203e-01 -4.17722464e-01 2.20640510e-01
-1.39723510e-01 6.15440011e-01 6.88387930e-01 1.79466009e-02
-4.54929173e-01 -1.05636142e-01 -3.34786713e-01 8.44111562e-01
6.24135077e-01 -8.18546057e-01 2.62119949e-01 1.55401766e-01
5.38294911e-01 1.35474712e-01 6.84282303e-01 -8.48611474e-01
-7.39121251e-03 -3.55703533e-01 5.42613268e-01 -6.74021125e-01
3.36521178e-01 -2.46404186e-01 7.27469385e-01 6.31952643e-01
-3.10181975e-01 3.96172017e-01 5.82056046e-01 7.03023793e-03
1.13536045e-01 -1.40263692e-01 7.96207786e-01 -1.86566692e-02
-3.65602612e-01 -4.01466452e-02 -2.15174779e-01 8.67998451e-02
1.19960690e+00 1.07429035e-01 -1.75550804e-01 -7.18137575e-03
-5.91557920e-01 3.42757553e-01 6.55544758e-01 1.24999270e-01
2.37625301e-01 -9.98619854e-01 -9.37000692e-01 2.93005407e-01
-1.65010005e-01 -2.33439580e-02 -4.50122505e-01 1.09463453e+00
-1.06964469e+00 1.80925936e-01 2.08032113e-02 -8.15155566e-01
-1.21669102e+00 6.18918180e-01 5.40231586e-01 -4.13163871e-01
-4.69433695e-01 9.34721589e-01 -6.33982480e-01 -6.28931880e-01
-2.63844598e-02 -3.51869613e-01 3.41310114e-01 -3.80030215e-01
4.57965046e-01 7.62374997e-01 -1.89618394e-01 -2.64945239e-01
-4.22257453e-01 4.54142720e-01 3.68340433e-01 3.49525991e-03
1.81956851e+00 8.57210755e-01 -8.02632034e-01 8.40305686e-01
1.16410518e+00 -1.86925218e-01 -7.91526079e-01 1.86092705e-01
1.92486525e-01 1.27747685e-01 1.33204246e-02 -8.22721183e-01
-6.86268568e-01 3.85404795e-01 9.56494957e-02 -1.13624260e-01
9.41481471e-01 -9.40814987e-02 3.49830151e-01 8.15066814e-01
2.54014999e-01 -8.15617383e-01 -3.31193060e-01 6.62640691e-01
1.21655953e+00 -8.03756952e-01 5.41133463e-01 1.29682384e-02
-3.95356745e-01 1.35712028e+00 4.15515810e-01 -3.40901941e-01
3.89522195e-01 7.79421151e-01 -5.59047699e-01 -2.70504057e-01
-9.00611639e-01 3.03494841e-01 1.25084370e-01 3.06296527e-01
2.62222648e-01 4.20272984e-02 -2.53828801e-02 7.37159431e-01
-6.90219820e-01 3.25568348e-01 6.68228343e-02 7.22704351e-01
-5.42546585e-02 -1.11096847e+00 -7.22473681e-01 3.42381090e-01
-9.57581326e-02 -2.13039592e-01 -5.00241220e-01 1.03107321e+00
-7.68838674e-02 6.07206859e-02 7.00383261e-02 -3.38244021e-01
-4.02124561e-02 1.15284212e-01 6.50352120e-01 -1.81751877e-01
-8.75566959e-01 -8.05529878e-02 2.07019612e-01 -8.17950368e-01
2.78313547e-01 -9.72934127e-01 -1.51294613e+00 -7.00456738e-01
6.46577030e-03 1.01872750e-01 9.57112253e-01 7.26459384e-01
5.02452970e-01 5.59606671e-01 3.14617306e-01 -1.10656011e+00
-7.49471545e-01 -6.14098251e-01 -4.59852040e-01 -4.67157930e-01
-1.42645231e-02 -6.39825642e-01 -4.35437411e-01 -1.85118511e-01]
|
[8.486225128173828, 6.810619354248047]
|
15173198-2239-419f-afa5-b954371cff31
|
2nd-place-solution-for-visda-2021-challenge
|
2110.14240
| null |
https://arxiv.org/abs/2110.14240v1
|
https://arxiv.org/pdf/2110.14240v1.pdf
|
2nd Place Solution for VisDA 2021 Challenge -- Universally Domain Adaptive Image Recognition
|
The Visual Domain Adaptation (VisDA) 2021 Challenge calls for unsupervised domain adaptation (UDA) methods that can deal with both input distribution shift and label set variance between the source and target domains. In this report, we introduce a universal domain adaptation (UniDA) method by aggregating several popular feature extraction and domain adaptation schemes. First, we utilize VOLO, a Transformer-based architecture with state-of-the-art performance in several visual tasks, as the backbone to extract effective feature representations. Second, we modify the open-set classifier of OVANet to recognize the unknown class with competitive accuracy and robustness. As shown in the leaderboard, our proposed UniDA method ranks the 2nd place with 48.56% ACC and 70.72% AUROC in the VisDA 2021 Challenge.
|
['Qiang Wang', 'Pengfei Xu', 'Tengfei Xing', 'Yueming Zhang', 'Xingxu Yao', 'Xiangyu Yue', 'Shanghang Zhang', 'Sicheng Zhao', 'Xiaolin Song', 'Haojin Liao']
|
2021-10-27
| null | null | null | null |
['universal-domain-adaptation']
|
['computer-vision']
|
[-4.53758948e-02 -2.67586678e-01 -2.18123689e-01 -2.21819788e-01
-7.64701307e-01 -9.78779197e-01 8.86296034e-01 -1.25005543e-01
-2.83470571e-01 7.87065804e-01 -9.07045156e-02 -1.22381374e-01
2.44361743e-01 -3.11900020e-01 -6.02233052e-01 -6.34586334e-01
1.55286491e-01 4.78884995e-01 5.04933000e-01 -1.34211570e-01
-2.49077082e-01 5.63735425e-01 -1.36424732e+00 5.68742990e-01
7.90034652e-01 1.38139021e+00 -1.19726479e-01 5.59836805e-01
-1.27395675e-01 4.96873140e-01 -7.12612927e-01 -3.49798292e-01
3.42370957e-01 -3.69807899e-01 -8.11448157e-01 4.46758559e-03
8.32656384e-01 -2.77826339e-01 -3.20171386e-01 8.31484735e-01
6.92919075e-01 1.63488820e-01 1.19994426e+00 -1.40669405e+00
-8.59124184e-01 -3.42488009e-03 -6.68850899e-01 4.79137152e-01
1.42322928e-01 2.68988073e-01 8.48765969e-01 -1.10849226e+00
1.03091621e+00 1.09015465e+00 6.21011376e-01 7.14154243e-01
-1.49779773e+00 -9.61279869e-01 3.75081152e-01 4.85130161e-01
-1.34189439e+00 -3.18185806e-01 7.40835309e-01 -7.21715033e-01
1.04751289e+00 -1.07492134e-01 1.75857812e-01 1.71031129e+00
1.06893413e-01 1.00716186e+00 1.33263898e+00 -3.98622364e-01
3.10196489e-01 4.45581526e-01 2.34804377e-01 3.61747593e-01
2.69475251e-01 3.07466835e-01 -4.80741084e-01 -5.08848839e-02
5.36398828e-01 -3.80389214e-01 9.42379087e-02 -1.08941650e+00
-9.88359928e-01 7.69407749e-01 5.79613686e-01 1.24619557e-02
-1.31706074e-01 -4.26505297e-01 7.48653114e-01 4.60370153e-01
5.24482191e-01 5.28786540e-01 -6.18737996e-01 1.71455875e-01
-6.06396854e-01 1.38714597e-01 4.41812962e-01 9.94989395e-01
5.96392393e-01 2.09985957e-01 -3.45372319e-01 1.11637926e+00
2.06376463e-01 6.30420148e-01 6.59840882e-01 -4.65880096e-01
3.39800298e-01 5.77194452e-01 -1.12292297e-01 -7.25708306e-01
-5.16762376e-01 -7.27870643e-01 -7.73932934e-01 6.26917124e-01
7.23221898e-01 -1.09702079e-02 -1.27458990e+00 1.70664322e+00
3.86612505e-01 -4.59042825e-02 2.83799887e-01 8.22421730e-01
9.87823308e-01 4.97948706e-01 4.17042226e-01 1.25425681e-01
1.36988151e+00 -1.03826523e+00 -3.61362070e-01 -4.33001012e-01
3.71476829e-01 -5.98125994e-01 1.07077253e+00 4.04500872e-01
-4.42698628e-01 -9.20412123e-01 -1.40943229e+00 -7.34419599e-02
-7.11379886e-01 2.99245209e-01 2.62345731e-01 5.01778603e-01
-7.63500333e-01 1.58641994e-01 -5.94291389e-01 -7.53519535e-01
6.65954590e-01 8.93288031e-02 -6.89299405e-01 -1.82435423e-01
-1.15033352e+00 9.42113876e-01 4.68825370e-01 -6.55828655e-01
-1.17925811e+00 -8.87631595e-01 -7.73528337e-01 -6.85893744e-02
2.55896628e-01 -6.39121413e-01 1.16242182e+00 -1.27091968e+00
-1.47925651e+00 1.18644977e+00 8.14966634e-02 -6.82700992e-01
5.26852071e-01 -2.27677733e-01 -7.83540547e-01 1.77654296e-01
1.42065242e-01 7.72343874e-01 1.05158937e+00 -1.19369650e+00
-7.63858199e-01 -4.47349966e-01 -4.39517200e-01 1.31717086e-01
-3.44480097e-01 -1.99425846e-01 -2.36323133e-01 -7.64066339e-01
-2.13956565e-01 -7.86822796e-01 2.02018812e-01 7.84621388e-02
-1.65209725e-01 -2.93085217e-01 1.08385754e+00 -7.05861568e-01
1.10354412e+00 -2.45654583e+00 5.94447963e-02 2.00585380e-01
4.06349242e-01 6.28393948e-01 -3.41009259e-01 9.39612240e-02
-4.35774654e-01 -3.11735302e-01 -1.97702214e-01 -1.57842964e-01
1.17395818e-01 -6.52778819e-02 -4.13029611e-01 3.20579290e-01
4.63836759e-01 8.31333399e-01 -7.56098866e-01 -2.39104390e-01
1.68082297e-01 1.47217885e-01 -3.29056978e-01 2.67627001e-01
5.03204716e-03 3.80917788e-01 -2.67185271e-01 7.45704830e-01
9.52788830e-01 -2.86641091e-01 3.70891899e-01 -2.94811726e-01
1.56822875e-01 -3.23743816e-03 -1.06395483e+00 1.55749989e+00
-2.42735595e-01 8.21188927e-01 -2.60656655e-01 -7.79128015e-01
1.20968759e+00 -1.09613501e-01 2.34118327e-01 -1.05967879e+00
-2.92655602e-02 2.12385103e-01 -6.12709969e-02 -9.99469608e-02
2.01314956e-01 1.52110895e-02 -3.25589925e-01 -8.71848464e-02
8.42615485e-01 1.93516731e-01 -3.04339491e-02 2.38122880e-01
1.10813212e+00 2.74003416e-01 8.62638235e-01 -3.31125766e-01
5.78972161e-01 1.26475230e-01 4.59890366e-01 6.95409417e-01
-7.58292258e-01 7.27569044e-01 6.14616513e-01 -5.97076952e-01
-1.22957873e+00 -1.52376604e+00 -1.74463972e-01 1.51364875e+00
-3.37811783e-02 -4.36156318e-02 -4.39334095e-01 -1.24983478e+00
2.54447430e-01 6.82184577e-01 -8.16013575e-01 -4.23986256e-01
-6.34297580e-02 -2.93812990e-01 7.15973020e-01 6.94422543e-01
5.31390488e-01 -8.33341181e-01 -2.03258947e-01 3.64018567e-02
1.59073412e-01 -1.27616894e+00 -3.02726001e-01 4.57637906e-01
-5.11357188e-01 -1.08151495e+00 -9.62628543e-01 -8.83649588e-01
3.65649670e-01 6.02825806e-02 1.20207405e+00 -7.40352809e-01
-1.33901417e-01 3.39341104e-01 -3.99654508e-01 -3.93104136e-01
-2.50655353e-01 3.71365249e-01 1.60056189e-01 1.76643297e-01
5.14602482e-01 -3.85602474e-01 -3.16352606e-01 4.45879400e-01
-6.74362898e-01 -1.51055932e-01 6.17524862e-01 1.03232968e+00
7.73647547e-01 -5.00531018e-01 7.36929119e-01 -9.43807423e-01
5.82499564e-01 -6.10049129e-01 -6.66619241e-01 2.76321352e-01
-7.37573743e-01 1.58500243e-02 7.49228835e-01 -5.69722474e-01
-1.12688470e+00 4.29832518e-01 1.62313938e-01 -7.48872221e-01
-5.32273054e-01 1.88542865e-02 -3.99574846e-01 -2.94242054e-03
1.29865384e+00 3.16590786e-01 -6.61696419e-02 -5.74060559e-01
4.93896574e-01 6.28636241e-01 8.52841079e-01 -3.43265504e-01
9.45310652e-01 4.22804385e-01 -2.14471161e-01 -7.41289198e-01
-6.94809854e-01 -6.35920644e-01 -9.14767563e-01 5.49529493e-02
8.80781710e-01 -1.17928278e+00 -1.34414181e-01 7.23944008e-01
-9.46774542e-01 -4.89010245e-01 -2.91752398e-01 2.98628688e-01
-3.79861832e-01 2.51231194e-01 3.41277383e-02 -2.86872059e-01
-2.96263486e-01 -8.59087229e-01 7.03772366e-01 2.63324410e-01
-1.86669156e-01 -8.92114639e-01 3.90769094e-01 6.13915063e-02
4.42054242e-01 3.50635171e-01 7.24416375e-01 -1.35734916e+00
8.85170847e-02 -1.15253525e-02 -6.51895344e-01 6.17774069e-01
1.56832054e-01 -2.02866137e-01 -1.30117238e+00 -4.70066965e-01
-6.19885921e-01 -5.06376445e-01 1.00112545e+00 3.32821488e-01
9.42885041e-01 1.29207015e-01 -4.36808735e-01 7.64809430e-01
1.14306128e+00 2.50077277e-01 4.88645136e-01 5.65222383e-01
6.07221067e-01 2.52994984e-01 6.22362614e-01 3.92624259e-01
4.02438164e-01 9.28091347e-01 2.90876240e-01 -2.54803926e-01
-6.18763506e-01 -4.38449651e-01 4.44196254e-01 1.91389218e-01
4.03143227e-01 -1.79690748e-01 -1.12518835e+00 7.61985540e-01
-1.69059575e+00 -6.64622366e-01 4.67744954e-02 2.16104913e+00
5.27218103e-01 2.16998890e-01 6.53871179e-01 -4.09423918e-01
6.96348727e-01 4.51884270e-02 -9.04900849e-01 -4.36623603e-01
-4.39309984e-01 2.31110990e-01 5.97292364e-01 1.25355810e-01
-1.55670559e+00 1.14024103e+00 6.35880661e+00 1.08293712e+00
-1.06902647e+00 2.14624345e-01 4.45502460e-01 1.32469088e-01
2.19955310e-01 -3.87750149e-01 -8.13306332e-01 3.94338608e-01
1.02869165e+00 -1.34957343e-01 1.51769325e-01 1.23953080e+00
-4.60836112e-01 2.09299579e-01 -1.05207288e+00 1.03396177e+00
6.10317476e-02 -1.04829180e+00 1.63709074e-01 -2.68426836e-02
8.89994562e-01 2.00250551e-01 2.78777421e-01 9.01343167e-01
3.89754236e-01 -8.98075461e-01 5.24178326e-01 1.38337001e-01
1.24110723e+00 -7.46839821e-01 7.10779786e-01 -1.28426582e-01
-1.16044855e+00 -1.36852369e-01 -4.28919286e-01 2.40523592e-01
-2.64431149e-01 2.10992992e-01 -1.07311785e+00 5.06188273e-01
8.97635698e-01 7.57975399e-01 -9.27127242e-01 1.09708142e+00
-1.75529361e-01 6.54909730e-01 -1.87784344e-01 3.10336530e-01
1.48152530e-01 3.54927987e-01 7.07302034e-01 1.40164745e+00
-6.99598938e-02 -3.66077125e-01 9.11501795e-02 4.61909384e-01
-3.05345446e-01 5.22984602e-02 -8.90133679e-01 2.37724818e-02
4.43574607e-01 1.17607260e+00 -3.75258982e-01 -4.24588203e-01
-4.36953843e-01 1.27583456e+00 5.18436849e-01 4.53597277e-01
-8.51083994e-01 -4.24881071e-01 1.02376127e+00 3.69207859e-02
6.64133132e-01 -8.07496905e-02 -2.72435606e-01 -1.21408379e+00
-2.16974914e-01 -1.09442818e+00 9.51478899e-01 -6.79998755e-01
-1.72318411e+00 7.66598761e-01 4.42996174e-02 -1.60577583e+00
-2.57143676e-01 -1.07572901e+00 -2.09404454e-01 7.01248169e-01
-1.74873340e+00 -1.39234614e+00 -3.52661043e-01 7.78100669e-01
6.19207859e-01 -7.09972918e-01 8.46249282e-01 3.00089598e-01
-3.94857496e-01 1.19935215e+00 6.16563439e-01 2.74561226e-01
1.30617619e+00 -1.33206248e+00 6.29334331e-01 7.87176311e-01
2.08767205e-02 1.52108327e-01 4.11455572e-01 -4.45682317e-01
-8.23769033e-01 -1.27136803e+00 4.87444371e-01 -5.88995337e-01
7.61995971e-01 -5.47059238e-01 -1.16323614e+00 7.39738345e-01
1.71845570e-01 4.14797038e-01 6.27822399e-01 6.96133971e-02
-1.14775348e+00 -3.52248549e-01 -1.27570260e+00 2.70279616e-01
9.42782879e-01 -5.92546761e-01 -7.32127130e-01 1.73305999e-02
5.62377214e-01 -4.84668344e-01 -7.40506470e-01 2.54683882e-01
6.70866787e-01 -7.96673238e-01 1.11189759e+00 -1.03149068e+00
3.45227808e-01 -4.26391780e-01 -2.81256318e-01 -1.54643059e+00
-7.25471258e-01 -1.75928354e-01 -1.14489481e-01 1.27162516e+00
4.54575777e-01 -7.05921888e-01 4.27199662e-01 6.46083802e-02
-4.98332530e-02 -1.66880220e-01 -1.12013674e+00 -1.12280500e+00
2.76434928e-01 -2.92009860e-01 3.53038162e-01 1.10978651e+00
-2.86681712e-01 6.31651878e-01 -3.88920367e-01 1.37460321e-01
5.63532591e-01 -1.49473250e-01 9.43358243e-01 -1.40879834e+00
-1.43863767e-01 -3.95350754e-01 -6.40158117e-01 -9.04102147e-01
1.57895625e-01 -1.05478346e+00 -2.73849726e-01 -1.14720094e+00
2.19071448e-01 3.22067104e-02 -9.13743615e-01 6.72260880e-01
-2.37141382e-02 2.62701750e-01 3.29165339e-01 2.42714465e-01
-9.45200205e-01 5.67988932e-01 8.96712244e-01 -2.56381601e-01
-2.02254757e-01 -2.01108620e-01 -7.90770531e-01 5.60684919e-01
6.80838704e-01 -4.07916248e-01 -2.85251230e-01 -2.97345847e-01
-3.25927347e-01 -5.77729106e-01 4.60542351e-01 -1.23624706e+00
-1.65176094e-01 -7.29748011e-02 8.57599258e-01 -4.12914127e-01
8.67763311e-02 -8.37419569e-01 -8.51688609e-02 3.09545577e-01
-3.13438594e-01 -1.41544743e-02 7.16662049e-01 6.59461558e-01
-1.57396629e-01 2.95369178e-01 1.21087205e+00 3.19607764e-01
-1.47187448e+00 4.87870984e-02 -2.59068847e-01 2.97753513e-01
1.16066122e+00 -1.57529801e-01 -7.22812474e-01 -9.19415653e-02
-7.24435031e-01 2.66216040e-01 3.54255199e-01 7.98324883e-01
4.72928226e-01 -1.61707246e+00 -8.43733907e-01 3.34318042e-01
7.93417096e-01 -4.77984905e-01 4.49222833e-01 3.93087119e-01
-2.48875722e-01 4.12227571e-01 -8.42195809e-01 -6.23810709e-01
-1.15479279e+00 6.43168211e-01 4.22501266e-01 -5.67489803e-01
-2.92903036e-01 9.10752356e-01 6.01226151e-01 -6.13800228e-01
4.90322299e-02 4.72272746e-02 -3.73880506e-01 2.96631426e-01
6.31289482e-01 3.62887323e-01 2.14930862e-01 -5.89785516e-01
-7.86200345e-01 2.55144060e-01 -3.20164263e-01 2.02207491e-01
1.05706143e+00 -4.44379523e-02 5.61370313e-01 2.80703187e-01
1.27177620e+00 -2.65930355e-01 -1.46668196e+00 -4.93692607e-01
-6.25637323e-02 -2.79972374e-01 -3.63103449e-02 -1.48750627e+00
-7.14166045e-01 9.82644379e-01 1.31888998e+00 1.49530582e-02
1.19482994e+00 3.39531451e-02 3.67204607e-01 1.32221088e-01
-2.49128286e-02 -1.19068348e+00 1.65853538e-02 6.97891057e-01
9.72011507e-01 -1.54743969e+00 -4.14522216e-02 -2.64457073e-02
-1.20011413e+00 9.90404606e-01 1.08467567e+00 -6.65185899e-02
3.31694186e-01 -1.00974686e-01 3.99925560e-01 3.38485949e-02
-7.62751937e-01 -4.25972044e-01 7.86262274e-01 1.29374647e+00
9.26682130e-02 1.75519139e-01 -5.91024682e-02 8.57485712e-01
3.14602226e-01 -2.23055154e-01 -2.73606852e-02 4.21914399e-01
-3.55255783e-01 -9.67861831e-01 -2.94248164e-01 2.46112585e-01
-1.36905208e-01 1.26622915e-01 -6.27162755e-01 1.06520462e+00
-6.95792818e-03 5.94341338e-01 9.32359546e-02 -5.89666724e-01
6.89600527e-01 4.77717072e-01 1.97850853e-01 -4.37496841e-01
-4.07192081e-01 -1.04808643e-01 -3.49994674e-02 -4.84836787e-01
-3.47080566e-02 -5.86173415e-01 -8.78554761e-01 1.33759201e-01
2.09385455e-01 -3.26713860e-01 3.93685579e-01 7.56344974e-01
7.87681639e-01 5.22588909e-01 5.23268819e-01 -5.81412673e-01
-6.70036018e-01 -9.63581443e-01 -5.54361403e-01 6.17174208e-01
4.08816844e-01 -1.14338744e+00 -1.00941285e-01 -5.35714999e-02]
|
[10.153226852416992, 2.7321789264678955]
|
3ac10931-05fe-4302-8fea-097fa32a1c30
|
3d-high-resolution-cardiac-segmentation
|
1902.11000
| null |
http://arxiv.org/abs/1902.11000v1
|
http://arxiv.org/pdf/1902.11000v1.pdf
|
3D High-Resolution Cardiac Segmentation Reconstruction from 2D Views using Conditional Variational Autoencoders
|
Accurate segmentation of heart structures imaged by cardiac MR is key for the
quantitative analysis of pathology. High-resolution 3D MR sequences enable
whole-heart structural imaging but are time-consuming, expensive to acquire and
they often require long breath holds that are not suitable for patients.
Consequently, multiplanar breath-hold 2D cine sequences are standard practice
but are disadvantaged by lack of whole-heart coverage and low through-plane
resolution. To address this, we propose a conditional variational autoencoder
architecture able to learn a generative model of 3D high-resolution left
ventricular (LV) segmentations which is conditioned on three 2D LV
segmentations of one short-axis and two long-axis images. By only employing
these three 2D segmentations, our model can efficiently reconstruct the 3D
high-resolution LV segmentation of a subject. When evaluated on 400 unseen
healthy volunteers, our model yielded an average Dice score of $87.92 \pm 0.15$
and outperformed competing architectures.
|
['Daniel Rueckert', "Declan P. O'Regan", 'Stuart A. Cook', 'Giacomo Tarroni', 'Juan J. Cerrolaza', 'Carlo Biffi', 'Antonio de Marvao']
|
2019-02-28
| null | null | null | null |
['cardiac-segmentation']
|
['medical']
|
[-7.70384967e-02 1.02196865e-01 1.88174337e-01 -4.76339608e-01
-8.40475559e-01 -5.55918276e-01 6.17304407e-02 5.32702822e-03
-4.61935431e-01 7.05489635e-01 -1.13810390e-01 -2.73171484e-01
-8.27464238e-02 -5.88775814e-01 -4.00753856e-01 -7.18175054e-01
-2.67225385e-01 1.09476840e+00 1.27589926e-01 4.33659047e-01
-2.65155077e-01 6.48922801e-01 -6.92197323e-01 -1.76572874e-01
7.21848309e-01 5.92323780e-01 4.84758705e-01 8.67597103e-01
2.07581580e-01 5.77238441e-01 -2.46014684e-01 -1.24673173e-01
1.95500806e-01 -7.78628409e-01 -8.58660460e-01 2.22092554e-01
4.48454469e-01 -9.27462041e-01 -2.11894393e-01 8.01532805e-01
8.20540130e-01 -5.68251833e-02 6.27557516e-01 -5.19354761e-01
-3.26004297e-01 4.83848572e-01 -4.58696067e-01 7.36585259e-01
-2.69018590e-01 6.13355860e-02 8.86997163e-01 -7.22816288e-01
9.49786246e-01 4.40727264e-01 7.25514352e-01 6.34300530e-01
-1.62728369e+00 -1.55747607e-01 -4.45041776e-01 -3.50978017e-01
-1.03116977e+00 -2.34409705e-01 8.73084247e-01 -8.43809187e-01
7.43843079e-01 1.16866872e-01 1.00468099e+00 5.73275745e-01
5.39768994e-01 3.78631860e-01 1.15422952e+00 9.35481489e-02
5.50835654e-02 -2.29603410e-01 -2.28532180e-02 7.52688885e-01
3.53571653e-01 -1.14539396e-02 1.63028300e-01 -1.93561614e-01
1.33356285e+00 1.04374439e-01 -6.43976569e-01 -8.35001349e-01
-1.54323292e+00 9.46843743e-01 4.39999044e-01 7.03540802e-01
-7.42074430e-01 6.60968274e-02 3.12767565e-01 -1.96800698e-02
3.46077740e-01 4.48533744e-01 -2.40734905e-01 3.64996083e-02
-1.40775311e+00 1.74725741e-01 3.63139421e-01 2.34058768e-01
2.00299621e-01 2.88544148e-01 5.58485016e-02 6.92663133e-01
5.34022331e-01 4.76680517e-01 5.31984806e-01 -1.42491448e+00
-3.67492554e-03 2.05795512e-01 -1.09789088e-01 -7.88009167e-01
-6.32324219e-01 -5.39527357e-01 -1.16421604e+00 7.22580031e-02
4.41326559e-01 -1.18302308e-01 -1.01186860e+00 1.28359938e+00
5.09473145e-01 -5.00264168e-02 -2.22995490e-01 1.46062517e+00
9.57929969e-01 3.55136126e-01 1.29822958e-02 -7.35698164e-01
1.09885108e+00 -4.98480469e-01 -6.33706808e-01 -1.10123783e-01
5.64442039e-01 -3.57981056e-01 5.43241143e-01 3.27012576e-02
-1.40075171e+00 -4.44914281e-01 -9.50890958e-01 1.03634983e-01
5.06551683e-01 -7.47760534e-02 1.93306491e-01 5.97457409e-01
-9.80239570e-01 9.64750707e-01 -1.44917238e+00 2.86356956e-01
6.58977211e-01 1.47326171e-01 -5.02401173e-01 -1.00063659e-01
-9.43283975e-01 8.89910519e-01 1.74713776e-01 1.84059441e-01
-9.49515522e-01 -9.91962969e-01 -7.47947156e-01 -9.66829285e-02
7.10429475e-02 -9.25473511e-01 9.35247600e-01 -1.63636565e-01
-1.19111848e+00 1.26009488e+00 2.05319762e-01 -4.91559207e-01
6.55231476e-01 7.21412301e-02 -2.31675450e-02 7.43953347e-01
2.02136114e-01 5.84796429e-01 7.89346874e-01 -1.11948800e+00
2.34652534e-01 -5.94969869e-01 -5.41438520e-01 1.20759778e-01
4.76378471e-01 -1.85840309e-01 4.44510132e-02 -4.18269277e-01
7.83419311e-01 -8.36871147e-01 -5.29455841e-01 -1.55296981e-01
-2.19607279e-01 6.24848366e-01 5.45826912e-01 -1.23736405e+00
9.38199043e-01 -1.85087419e+00 2.13248521e-01 4.53302525e-02
8.94936979e-01 1.71270579e-01 4.85257834e-01 -3.63864869e-01
-1.41071692e-01 3.49118471e-01 -5.72943270e-01 -2.17805021e-02
-4.08313721e-01 1.68374106e-01 2.92389065e-01 7.42021382e-01
-2.14608058e-01 1.04005766e+00 -8.76768947e-01 -8.94902527e-01
4.30452943e-01 8.33555937e-01 -5.07343471e-01 4.33638990e-01
2.85417020e-01 1.13629687e+00 -5.02276361e-01 2.89569974e-01
4.96324390e-01 -6.06931925e-01 6.43294692e-01 -2.05166683e-01
2.18013190e-02 -2.32398510e-04 -7.65838921e-01 1.82821143e+00
3.48607078e-02 3.18857938e-01 3.06061655e-01 -1.10147965e+00
6.50939941e-01 7.88079381e-01 9.92068410e-01 -6.09267831e-01
3.13124329e-01 2.59651899e-01 3.05500150e-01 -4.55894381e-01
-2.52286345e-01 -9.97741222e-01 2.57764518e-01 6.57579005e-01
-2.56351158e-02 -5.71201682e-01 -4.93545160e-02 -7.47205596e-03
8.94411445e-01 1.03006117e-01 1.71522960e-01 -2.75184959e-01
3.19585949e-01 -2.80162722e-01 8.29101801e-01 6.05898857e-01
-7.04046130e-01 1.10402691e+00 3.95471901e-01 -9.62248564e-01
-1.44046319e+00 -1.33932495e+00 -4.62272823e-01 -3.07818558e-02
-3.08714122e-01 -1.17007131e-02 -8.08441997e-01 -6.60266042e-01
-2.32660338e-01 3.78172606e-01 -3.88227731e-01 3.89412105e-01
-9.82559443e-01 -9.20743465e-01 4.36515421e-01 5.44686079e-01
2.39046186e-01 -8.79162669e-01 -1.53782058e+00 5.64970255e-01
-5.59840322e-01 -1.15051341e+00 -1.87913388e-01 1.06160246e-01
-1.61562717e+00 -1.09584379e+00 -1.36183703e+00 -5.66216171e-01
5.16351879e-01 -2.58849084e-01 1.68691158e+00 9.68014300e-02
-7.12134540e-01 9.50068608e-02 8.29693303e-02 2.90596727e-02
-5.09397686e-01 -3.14839035e-01 8.34903941e-02 -3.15587997e-01
-4.09181923e-01 -8.66638362e-01 -8.19779575e-01 1.20794483e-01
-5.29770494e-01 2.88895480e-02 5.50515413e-01 1.01179504e+00
1.34889698e+00 -1.25870049e-01 2.97687143e-01 -8.23919237e-01
1.60833508e-01 -1.91954255e-01 -5.84483743e-01 9.17949248e-03
-6.69878244e-01 -1.44734427e-01 2.59509027e-01 -2.35832073e-02
-8.11793149e-01 3.14162998e-03 -3.06703150e-01 -7.21543014e-01
-2.17599481e-01 3.12283486e-01 3.68195355e-01 1.24134235e-01
5.92285931e-01 4.10040677e-01 3.30527812e-01 -3.66231740e-01
2.36895289e-02 8.02507997e-02 5.14102399e-01 -3.25465143e-01
5.30791283e-01 5.17023981e-01 2.49728292e-01 -9.11903560e-01
-5.02337098e-01 -1.98813677e-01 -1.16931009e+00 -2.77974248e-01
1.51853728e+00 -6.53952777e-01 -5.01564085e-01 1.26464903e-01
-7.87960887e-01 -4.21895415e-01 -4.45050567e-01 8.88760805e-01
-7.63683259e-01 6.84806228e-01 -8.98954928e-01 -4.42271590e-01
-6.85113788e-01 -1.42732728e+00 8.23663592e-01 -7.62062445e-02
-2.33836785e-01 -1.07306623e+00 3.32953364e-01 5.89815795e-01
4.68476921e-01 8.19346130e-01 1.09511995e+00 -3.45367283e-01
-6.34458482e-01 -5.90877011e-02 8.58869329e-02 5.98646402e-01
-6.75398409e-02 -4.08129603e-01 -5.76829851e-01 -2.23074228e-01
3.93325955e-01 -2.47638583e-01 5.28343260e-01 1.31147647e+00
9.04651642e-01 2.39589199e-01 -4.27472442e-02 6.92994595e-01
1.26119184e+00 2.05156833e-01 2.50472963e-01 -2.64912546e-01
8.36325705e-01 2.88235247e-01 1.74717084e-01 2.66206443e-01
2.01233879e-01 3.16278785e-01 3.25931102e-01 -9.28838328e-02
9.26110987e-03 8.33167285e-02 -3.43244731e-01 1.04254162e+00
-3.80352020e-01 2.81134069e-01 -1.15397823e+00 6.62850082e-01
-1.22679830e+00 -9.08491671e-01 -4.08832252e-01 2.05239534e+00
6.77199066e-01 1.45665765e-01 1.93281129e-01 9.97171029e-02
4.71440941e-01 3.51516485e-01 -4.74196404e-01 -2.70203277e-02
1.01205654e-01 2.65301526e-01 1.65172428e-01 6.03394985e-01
-1.03362489e+00 2.93839425e-01 6.91562557e+00 -2.26643786e-01
-1.11430848e+00 2.52735138e-01 7.80588210e-01 -3.39116454e-02
-3.59327346e-01 -1.77414000e-01 -2.29780942e-01 3.09331119e-01
1.02433646e+00 1.96184531e-01 4.86008637e-02 6.54131353e-01
1.05647646e-01 -3.98091897e-02 -1.00112092e+00 1.03669107e+00
-1.88161090e-01 -1.47517443e+00 -4.55086499e-01 2.03737766e-01
5.11198699e-01 2.96682030e-01 -2.07162589e-01 -3.97190005e-02
-3.62931877e-01 -1.25471926e+00 1.86443776e-01 4.48746741e-01
9.78964567e-01 -4.22061950e-01 6.99629366e-01 5.94470680e-01
-6.74938440e-01 4.88210678e-01 -1.43728539e-01 3.49235415e-01
5.24625540e-01 8.56430709e-01 -8.69845510e-01 3.47713381e-01
7.63754427e-01 3.99547994e-01 -7.80093595e-02 7.01828003e-01
-1.03586074e-02 6.66516483e-01 -2.65975088e-01 5.98129809e-01
6.39829189e-02 -3.82744879e-01 8.34016025e-01 6.95256114e-01
2.24699691e-01 5.10079920e-01 1.61711335e-01 1.31415725e+00
9.61196572e-02 -1.56287886e-02 -5.09640992e-01 -9.31369215e-02
3.79664786e-02 1.25336635e+00 -1.01914859e+00 -4.16319907e-01
-2.08652914e-01 8.66953671e-01 -3.89220417e-02 1.89888149e-01
-6.77121043e-01 2.30778307e-01 2.33754087e-02 5.84913254e-01
3.06124955e-01 -2.77409077e-01 -2.41449669e-01 -1.32043755e+00
-6.75023571e-02 -6.75391972e-01 3.88791800e-01 -6.41930997e-01
-9.21344101e-01 7.11043835e-01 -4.09031771e-02 -7.19467163e-01
-5.18449843e-01 -2.08690926e-01 -3.03973526e-01 1.16030788e+00
-1.21453655e+00 -6.80989563e-01 -1.01253547e-01 2.49890491e-01
1.81621045e-01 1.69837236e-01 1.07045984e+00 4.14474070e-01
-2.01764643e-01 1.65252620e-03 -2.43004158e-01 4.69599932e-01
2.81195372e-01 -1.59419036e+00 1.34774148e-01 7.05492198e-01
2.17590258e-01 5.74162900e-01 5.75343966e-01 -6.63534641e-01
-9.38752592e-01 -7.42462337e-01 1.01611328e+00 -3.66604745e-01
-5.99236600e-02 3.31227988e-01 -1.00165546e+00 8.88200939e-01
-1.32872656e-01 7.69200921e-01 9.25246298e-01 -2.89759308e-01
1.39947817e-01 2.60826230e-01 -1.43518543e+00 6.20171838e-02
4.51544285e-01 -3.52107704e-01 -6.80828929e-01 -8.87548029e-02
2.58595973e-01 -8.26281846e-01 -1.66752398e+00 4.03689444e-01
6.14565730e-01 -1.26062727e+00 1.24312198e+00 -3.87938470e-01
5.66930056e-01 -2.96045393e-01 9.95579641e-03 -9.71749842e-01
-3.98753226e-01 -2.50236839e-01 -1.43564165e-01 3.63427252e-01
1.07854187e-01 -4.13199037e-01 9.61235344e-01 6.77343488e-01
-2.04345837e-01 -8.54860783e-01 -9.24837470e-01 -3.24362695e-01
3.85080904e-01 -1.35278866e-01 7.55831972e-03 1.16012406e+00
-5.30242741e-01 2.97265142e-01 -2.27943197e-01 7.29062334e-02
1.20970750e+00 4.50964659e-01 3.12305577e-02 -1.50171721e+00
-4.69584584e-01 -3.08058560e-01 -3.65946710e-01 -6.90347731e-01
1.40878959e-02 -1.02251041e+00 -1.03000566e-01 -1.62010586e+00
4.03220743e-01 -3.11768860e-01 -2.88288176e-01 1.74341686e-02
-1.01181589e-01 4.98393565e-01 1.23664960e-02 3.12357217e-01
-1.07522383e-01 2.61341184e-01 1.70670748e+00 1.01816118e-01
-1.30600214e-01 -1.05090104e-01 -2.87071854e-01 8.13966215e-01
6.24907076e-01 -6.17038369e-01 -4.31971908e-01 -4.33128476e-01
-3.00168265e-02 9.80262458e-01 4.80624557e-01 -9.13061082e-01
-2.65376836e-01 2.01082975e-01 9.87613201e-01 -8.48826408e-01
2.76574850e-01 -7.32159197e-01 3.13105673e-01 6.64342225e-01
-1.79612175e-01 6.60753548e-02 -2.58481681e-01 2.23421812e-01
-2.02598557e-01 -1.24757297e-01 1.25642085e+00 -7.69751906e-01
-6.63065836e-02 6.64492428e-01 -4.54145998e-01 4.67488348e-01
8.26264918e-01 -1.71206832e-01 5.84056616e-01 -1.70487836e-01
-1.41938818e+00 -1.21944835e-02 3.25278640e-01 -2.96301872e-01
8.72823536e-01 -1.09620667e+00 -8.67104828e-01 2.32778549e-01
-5.43490708e-01 5.78364491e-01 7.13496983e-01 1.38541842e+00
-1.08812630e+00 4.11797345e-01 -4.03468698e-01 -1.11229289e+00
-1.04660547e+00 3.45814049e-01 9.54882681e-01 -6.65478349e-01
-1.30350864e+00 7.69830883e-01 1.06862761e-01 -5.61482251e-01
-2.77374297e-01 -2.17813864e-01 -8.07974786e-02 -1.48025125e-01
2.94409037e-01 1.26401767e-01 -7.90400133e-02 -8.85993481e-01
-3.70082438e-01 7.62133420e-01 1.68924391e-01 -1.58548579e-01
1.35491002e+00 -2.27318496e-01 3.24194096e-02 5.35828888e-01
1.10279131e+00 -3.78474057e-01 -1.24868500e+00 -1.12775015e-02
-2.89160192e-01 -3.35782111e-01 5.91813028e-01 -5.97163081e-01
-1.38736594e+00 1.26921296e+00 6.70814514e-01 6.82975352e-02
8.27010036e-01 -1.92642778e-01 1.05753398e+00 -1.43775776e-01
-1.49713298e-02 -7.25329161e-01 -1.58997968e-01 -6.93222061e-02
5.74434996e-01 -1.38351786e+00 1.46258384e-01 -2.00646847e-01
-7.24630177e-01 9.91247058e-01 1.30986571e-01 -1.52049005e-01
6.21508420e-01 1.72070786e-01 4.16125268e-01 -5.28887510e-01
-3.67131710e-01 1.50563836e-01 3.40845287e-01 5.43625236e-01
7.13892043e-01 1.65368959e-01 -2.85245121e-01 2.70274520e-01
1.15616299e-01 -5.57804219e-02 4.37222600e-01 8.93256962e-01
-2.55150616e-01 -8.07978690e-01 -1.32468805e-01 4.11643893e-01
-1.10354030e+00 5.47957905e-02 2.51487851e-01 7.14452684e-01
-9.20231193e-02 2.90786564e-01 6.46062940e-02 3.24474812e-01
3.08607239e-02 4.81857449e-01 8.48035395e-01 -4.93131816e-01
-4.08412516e-01 5.71909726e-01 -3.17353040e-01 -4.67818826e-01
-3.88927460e-01 -8.95079315e-01 -1.41367161e+00 -3.58397923e-02
1.44151598e-01 1.24051869e-01 5.93782067e-01 6.78122401e-01
1.64023831e-01 7.15246439e-01 4.60377008e-01 -7.18068898e-01
-4.59019840e-01 -8.09676349e-01 -8.89133155e-01 3.24120432e-01
5.90116143e-01 -3.52178305e-01 -1.51749551e-01 3.25692266e-01]
|
[14.002737998962402, -2.4329380989074707]
|
1e8ae94e-a3e0-486d-81a1-82086b1395a5
|
hierarchical-discriminative-learning-improves
|
2303.01605
| null |
https://arxiv.org/abs/2303.01605v1
|
https://arxiv.org/pdf/2303.01605v1.pdf
|
Hierarchical discriminative learning improves visual representations of biomedical microscopy
|
Learning high-quality, self-supervised, visual representations is essential to advance the role of computer vision in biomedical microscopy and clinical medicine. Previous work has focused on self-supervised representation learning (SSL) methods developed for instance discrimination and applied them directly to image patches, or fields-of-view, sampled from gigapixel whole-slide images (WSIs) used for cancer diagnosis. However, this strategy is limited because it (1) assumes patches from the same patient are independent, (2) neglects the patient-slide-patch hierarchy of clinical biomedical microscopy, and (3) requires strong data augmentations that can degrade downstream performance. Importantly, sampled patches from WSIs of a patient's tumor are a diverse set of image examples that capture the same underlying cancer diagnosis. This motivated HiDisc, a data-driven method that leverages the inherent patient-slide-patch hierarchy of clinical biomedical microscopy to define a hierarchical discriminative learning task that implicitly learns features of the underlying diagnosis. HiDisc uses a self-supervised contrastive learning framework in which positive patch pairs are defined based on a common ancestry in the data hierarchy, and a unified patch, slide, and patient discriminative learning objective is used for visual SSL. We benchmark HiDisc visual representations on two vision tasks using two biomedical microscopy datasets, and demonstrate that (1) HiDisc pretraining outperforms current state-of-the-art self-supervised pretraining methods for cancer diagnosis and genetic mutation prediction, and (2) HiDisc learns high-quality visual representations using natural patch diversity without strong data augmentations.
|
['Todd C. Hollon', 'Honglak Lee', 'Daniel A. Orringer', 'Christian W. Freudiger', 'Asadur Chowdury', 'Akhil Kondepudi', 'Xinhai Hou', 'Cheng Jiang']
|
2023-03-02
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Jiang_Hierarchical_Discriminative_Learning_Improves_Visual_Representations_of_Biomedical_Microscopy_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Jiang_Hierarchical_Discriminative_Learning_Improves_Visual_Representations_of_Biomedical_Microscopy_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['whole-slide-images']
|
['computer-vision']
|
[ 7.2912419e-01 1.5383075e-01 -2.7130568e-01 -1.8346517e-01
-1.0724519e+00 -5.1778543e-01 5.6537837e-01 5.6194985e-01
-2.1623354e-01 6.3096392e-01 2.7543467e-01 -3.1786376e-01
-6.3180290e-02 -5.4646128e-01 -8.1059420e-01 -1.1987801e+00
7.2512731e-02 6.0983390e-01 -1.6645376e-02 1.4293177e-01
-3.6462851e-02 6.4872199e-01 -1.2359885e+00 7.1212614e-01
5.7982892e-01 8.9315844e-01 2.8045952e-01 1.0428145e+00
1.4454342e-01 7.6693016e-01 -5.2098316e-01 1.4970498e-01
5.4323696e-02 -6.6305369e-01 -8.7818378e-01 2.2924125e-01
7.5899547e-01 4.1970439e-02 -3.0983868e-01 7.9048640e-01
4.8374382e-01 -3.7103361e-01 1.2853491e+00 -9.2621583e-01
-9.7651297e-01 -6.3953742e-02 -8.3989835e-01 5.4899734e-01
-6.8193130e-02 5.7426333e-01 1.0909572e+00 -6.0836613e-01
1.1846560e+00 8.9995617e-01 8.1453013e-01 7.1042097e-01
-1.9154791e+00 -3.3133230e-01 -2.0294689e-01 -5.8863927e-02
-1.2333544e+00 -3.4479928e-01 5.2877349e-01 -8.6688364e-01
1.1173887e+00 2.5452539e-01 8.5956818e-01 1.2904698e+00
5.4221910e-01 7.2741443e-01 1.2699441e+00 -3.6366236e-01
2.2816239e-01 -7.9954201e-03 1.9403228e-01 9.7141874e-01
2.2133565e-01 1.6077468e-01 -3.3347023e-01 -4.1920862e-01
9.5824873e-01 4.9992442e-01 -5.1867610e-01 -6.7568892e-01
-1.4454038e+00 8.6357570e-01 6.8531770e-01 4.6090013e-01
-5.3773083e-02 -4.5394129e-03 3.3944157e-01 3.1145105e-01
3.0299407e-01 5.7586211e-01 -3.3516344e-01 4.3660980e-01
-9.2461067e-01 -1.4162849e-01 4.8423314e-01 4.9695781e-01
9.1248804e-01 -1.4145181e-01 -4.1824701e-01 6.9756746e-01
3.4129884e-02 4.4983798e-01 5.7990360e-01 -6.5713829e-01
-2.0931564e-01 9.3191993e-01 -2.9899222e-01 -6.7922604e-01
-4.2516568e-01 -4.5081750e-01 -1.1429014e+00 2.8782943e-01
3.5439223e-01 1.5385233e-01 -1.5165334e+00 1.4830838e+00
2.2963803e-01 1.5008649e-01 2.1276073e-01 6.3403237e-01
1.1550683e+00 4.2403254e-01 1.2801002e-01 -3.2968327e-01
1.4068075e+00 -6.3451028e-01 -1.9301017e-01 -5.4962818e-02
8.0395383e-01 -4.8975402e-01 9.9773884e-01 1.3877960e-01
-7.4208432e-01 -3.7452233e-01 -1.0079639e+00 -2.7171168e-01
-4.2659277e-01 1.3179788e-01 4.9004081e-01 1.6289265e-01
-1.2219567e+00 4.5597586e-01 -9.5388120e-01 -5.7795930e-01
1.1464289e+00 3.0796394e-01 -7.5880772e-01 -3.1294402e-01
-4.2716980e-01 5.6771082e-01 -1.2400497e-01 -3.6085576e-01
-1.3083053e+00 -1.3645111e+00 -8.8743299e-01 -8.1840865e-02
-2.9469898e-01 -9.0223336e-01 6.9762069e-01 -1.1128298e+00
-8.6462373e-01 1.6465825e+00 -2.1899566e-01 -2.1056311e-01
2.6713121e-01 4.6367052e-01 -1.0922561e-01 6.4578778e-01
2.9478353e-01 9.0863281e-01 8.7799108e-01 -1.5422995e+00
-3.6180979e-01 -6.2053609e-01 -4.0120068e-01 2.0130211e-02
-1.5958573e-01 -4.9533185e-01 -2.4810734e-01 -6.8668574e-01
-1.5842661e-01 -6.5946984e-01 -3.6006269e-01 4.7134960e-01
-5.4585916e-01 8.8562146e-02 8.7444896e-01 -4.6267101e-01
4.8462683e-01 -2.3773267e+00 2.3572110e-01 1.5558895e-01
5.7692462e-01 1.0505201e-01 -2.5537366e-01 2.1668747e-01
-2.4104905e-01 -1.3559778e-02 -3.4186086e-01 -3.7352502e-01
-4.4866553e-01 3.2198700e-01 -9.8493703e-02 8.0177104e-01
5.4625541e-01 1.2317923e+00 -1.0140535e+00 -8.2020509e-01
-3.3260453e-03 7.0326716e-01 -4.2229864e-01 5.2974713e-01
-1.5796156e-01 6.9982159e-01 -1.5797421e-01 1.1567806e+00
3.6890545e-01 -1.1147207e+00 2.5578195e-01 -5.5947649e-01
3.1473258e-01 -1.8301347e-01 -3.2028791e-01 1.6889025e+00
-3.3570023e-03 5.8112401e-01 8.3094098e-02 -1.2710377e+00
5.2073753e-01 1.4874710e-01 7.8958881e-01 -4.2548224e-01
-2.4583291e-02 -3.0524479e-03 -1.4100361e-01 -6.3472468e-01
-2.9144010e-01 -3.9755231e-01 3.7055558e-01 2.4414961e-01
4.9659449e-01 -1.5910578e-01 2.5215644e-02 2.4209452e-01
1.5383124e+00 -1.3670123e-01 5.3642851e-01 -3.9303720e-01
1.2881728e-01 4.0169847e-01 7.1316195e-01 7.0629191e-01
-4.0055394e-01 9.6149385e-01 7.7891833e-01 -4.1853315e-01
-1.0808743e+00 -1.3817534e+00 -4.8809415e-01 7.9037338e-01
-9.9770926e-02 1.4406081e-02 -2.6126319e-01 -9.4054967e-01
4.5619634e-01 -1.8639128e-01 -1.2477170e+00 -8.7771468e-02
-2.3416252e-01 -9.2351061e-01 4.4238842e-01 6.1131358e-01
-1.0247900e-01 -8.5222590e-01 -5.8873421e-01 9.7841686e-03
2.8548968e-01 -7.9284555e-01 -3.4824976e-01 6.9038349e-01
-7.3395222e-01 -1.5908817e+00 -9.5542288e-01 -1.1587888e+00
1.2607937e+00 4.0677521e-01 1.3146553e+00 1.3736622e-01
-1.2778243e+00 7.0752943e-01 -1.3954477e-01 -2.7116337e-01
-5.7802665e-01 -3.8758397e-01 -3.3672193e-01 -5.8600262e-02
2.3592257e-01 -5.8640140e-01 -9.4188714e-01 -1.4233853e-01
-9.1809708e-01 -5.7541694e-02 8.5279596e-01 1.5901651e+00
1.2685724e+00 -3.9297569e-01 3.7861326e-01 -1.2989478e+00
4.2453278e-02 -5.6603539e-01 -2.2552906e-01 4.4294921e-01
-4.4887313e-01 -1.2587106e-01 6.4938229e-01 -4.5349535e-01
-5.5728066e-01 1.7012760e-01 2.0972592e-01 -8.9214146e-01
-2.8131109e-01 4.0596545e-01 2.5957423e-01 -2.7537507e-01
1.0095381e+00 3.1366688e-01 5.4030102e-01 -1.4122601e-02
1.4024843e-01 4.2369607e-01 5.8599490e-01 -2.6851115e-01
4.5203900e-01 9.1692859e-01 2.6930171e-01 -8.6453760e-01
-6.3510376e-01 -7.9530054e-01 -6.2566274e-01 2.8299907e-01
9.9180597e-01 -1.0772139e+00 -5.2264225e-01 2.1031441e-01
-6.1763668e-01 -6.2034315e-01 -5.2477396e-01 2.2635432e-01
-5.3831935e-01 4.5649648e-01 -8.7524146e-01 -4.2355764e-01
-4.4945991e-01 -1.0479105e+00 1.4490645e+00 1.6785951e-01
-2.4885426e-01 -1.2222284e+00 2.5653359e-01 4.1153458e-01
1.2698384e-01 6.4698356e-01 1.4072937e+00 -4.7540930e-01
-3.9897817e-01 -2.9376049e-02 -3.5376930e-01 2.6812646e-01
5.5479044e-01 7.9558812e-02 -1.0982100e+00 -7.5200969e-01
-4.4277316e-01 -8.4182900e-01 1.2081029e+00 5.5335510e-01
1.2125202e+00 -2.7547637e-01 -8.6059439e-01 9.9892074e-01
1.6193311e+00 -2.0107676e-01 3.8776249e-01 -1.7304590e-01
8.6999756e-01 5.1578838e-01 6.9114409e-02 1.2113380e-01
2.7195102e-01 3.0839202e-01 3.0899656e-01 -8.8284302e-01
-3.3912912e-01 -1.7863182e-02 6.1021123e-02 3.6483043e-01
1.9004750e-01 9.5584668e-02 -8.4046268e-01 8.8486981e-01
-1.6010994e+00 -7.9721898e-01 2.8298962e-01 1.9527702e+00
1.1462379e+00 -3.5441995e-01 1.6745086e-01 -7.6429002e-02
4.3798047e-01 -4.1569095e-02 -9.1781586e-01 6.7357600e-02
-2.9762414e-01 4.8630494e-01 3.3151814e-01 1.8359417e-01
-1.2010795e+00 5.7071531e-01 6.3866482e+00 5.2841473e-01
-1.3606955e+00 -3.0821905e-02 1.1609756e+00 -3.3868469e-02
-2.3639876e-01 -3.6051235e-01 -5.4277843e-01 1.7485793e-01
5.0186861e-01 1.5191987e-01 7.6720469e-02 6.9506466e-01
-2.0853724e-01 1.4872363e-01 -1.5447116e+00 1.1423215e+00
2.7437434e-01 -1.7708174e+00 2.7452919e-01 3.3430570e-01
8.0325705e-01 1.5502751e-01 2.8847969e-01 4.9001900e-03
4.3781316e-01 -1.5408617e+00 -2.7801776e-01 3.8831156e-01
1.3064741e+00 -3.8024685e-01 6.1667496e-01 -1.4326680e-01
-8.7923288e-01 -5.9328057e-02 -4.2600933e-01 6.1895013e-01
-4.8989898e-01 5.8504111e-01 -1.1435401e+00 4.5612523e-01
7.8705877e-01 1.1719418e+00 -7.3160779e-01 7.6920927e-01
2.9015723e-01 5.9544218e-01 7.7174783e-02 3.0959079e-01
1.8271808e-01 1.5919383e-01 2.2657716e-01 1.5541819e+00
6.8472996e-02 -1.5485977e-01 2.7865064e-01 9.6663886e-01
-5.6144513e-02 -9.5485330e-02 -7.7924049e-01 -2.8549537e-01
1.5967923e-01 1.4233073e+00 -8.0051029e-01 -3.2894769e-01
-3.5490751e-01 9.8361826e-01 6.0808879e-01 4.2449296e-01
-2.7438748e-01 4.8130807e-02 7.6689750e-01 2.5909767e-01
5.3223044e-01 3.7866521e-01 -2.1786751e-01 -1.1220527e+00
-3.8188931e-01 -1.0323483e+00 6.8856728e-01 -5.7906193e-01
-1.8889377e+00 3.2824117e-01 -4.6594614e-01 -1.3442146e+00
-9.9669129e-02 -9.2508703e-01 -5.2994937e-01 6.0888141e-01
-1.6721178e+00 -1.5658467e+00 -3.8532645e-01 8.0884701e-01
1.4432925e-01 -2.8854212e-01 1.3594744e+00 -1.7846380e-01
-4.2432824e-01 6.9819617e-01 3.0923313e-01 2.5749072e-01
7.7214152e-01 -1.5367099e+00 -1.5555276e-01 3.6355320e-01
1.6853268e-01 6.2214535e-01 2.4077226e-01 -4.6344715e-01
-1.5451648e+00 -1.2891481e+00 4.5571890e-01 -5.4559231e-01
5.2962595e-01 -1.5131491e-01 -1.0788044e+00 5.4862863e-01
8.4413990e-02 8.6181450e-01 1.3203831e+00 -3.2498103e-02
-6.7752570e-01 -1.6012846e-01 -1.4231730e+00 5.4328740e-01
8.1821507e-01 -7.3327917e-01 -4.2911470e-01 7.0440364e-01
2.0404644e-01 -2.3023207e-01 -1.0752786e+00 4.5601508e-01
3.8381606e-01 -7.6847130e-01 1.0889108e+00 -7.7098274e-01
5.5341774e-01 -2.8321984e-01 -1.5693736e-01 -1.2803814e+00
-6.1099547e-01 -2.0484029e-01 -1.3661745e-01 7.0831978e-01
1.9540766e-01 -4.1022527e-01 9.4402355e-01 -6.8199016e-02
-1.4872111e-01 -1.2409159e+00 -1.0733289e+00 -4.9728945e-01
3.1322673e-01 5.3442603e-01 -9.2367129e-03 1.0955040e+00
1.1820143e-01 3.7780532e-01 1.6853228e-01 1.3694020e-01
7.8685904e-01 3.0711430e-01 6.0396558e-01 -1.3017240e+00
-6.6636974e-01 -4.4534236e-01 -8.4952581e-01 -6.4540160e-01
4.3513186e-02 -1.1678126e+00 9.0263309e-03 -1.5994103e+00
7.1175265e-01 -3.1370354e-01 -5.9068561e-01 6.9978184e-01
-3.5332543e-01 5.6997138e-01 -3.4171325e-01 4.9113280e-01
-5.2132845e-01 8.5547961e-02 1.3653804e+00 -7.7736914e-01
4.7522433e-02 -5.7123607e-01 -9.1618162e-01 3.0191159e-01
3.7505054e-01 -2.2399443e-01 -3.9775661e-01 -1.7802998e-01
-1.9474502e-01 -9.8177560e-02 7.4398488e-01 -8.4596550e-01
4.5431409e-02 -1.3650297e-01 1.0907975e+00 -1.6000909e-01
2.8739026e-01 -5.4437339e-01 -6.8960123e-02 7.2303426e-01
-3.0667955e-01 -2.3310415e-01 -2.1383680e-02 8.9674067e-01
-1.0804051e-01 2.4276239e-01 1.1800604e+00 -4.3216389e-01
-3.0407634e-01 4.4245917e-01 -1.8964489e-01 1.5688576e-01
9.9498749e-01 -4.2623094e-01 -6.2809902e-01 1.0359777e-01
-5.2150702e-01 4.0385809e-02 9.0934473e-01 -1.2254775e-01
7.1528316e-01 -1.0069036e+00 -7.7071828e-01 3.9482510e-01
7.0426977e-01 4.1195345e-01 3.2072163e-01 9.8297423e-01
-5.4344887e-01 4.5953110e-02 -3.0229330e-01 -1.2236654e+00
-1.3444008e+00 5.4173672e-01 3.4529060e-01 -6.0547245e-01
-8.9543772e-01 1.0961955e+00 8.1090754e-01 -2.9567963e-01
-5.0556394e-03 -1.7433934e-01 -2.0723680e-01 -1.5511319e-01
5.5147129e-01 -2.3952328e-01 -8.6408091e-04 -5.7627559e-01
-4.3598491e-01 5.9912527e-01 -5.4200512e-01 4.3367013e-01
1.4842443e+00 4.4519964e-01 -8.1544526e-02 5.3831398e-01
1.4938241e+00 -2.1648352e-01 -1.3885888e+00 -1.8989240e-01
-3.7677425e-01 -2.0669012e-01 -5.4380442e-03 -8.5095102e-01
-1.0070577e+00 7.9970104e-01 7.4366510e-01 -1.0183932e-01
9.1434711e-01 2.7619863e-01 4.5351696e-01 2.9464674e-01
7.0802592e-02 -7.3363602e-01 3.6765164e-01 1.6858548e-01
6.7722136e-01 -1.5379547e+00 2.3883604e-01 -2.9798755e-01
-6.2892520e-01 1.0380658e+00 5.1373011e-01 -1.9508676e-01
4.8096484e-01 4.2246139e-01 3.4811416e-01 -5.3997767e-01
-8.3103853e-01 -2.5918886e-01 2.5137872e-01 1.1192738e+00
5.8697814e-01 -2.4856726e-02 2.5229833e-01 2.2020522e-01
2.7334040e-01 4.5494203e-02 1.5287112e-01 1.1557634e+00
-8.0922417e-02 -7.2788334e-01 -1.0626776e-01 8.9454550e-01
-3.6545849e-01 -1.1820297e-01 -4.8191667e-01 6.0760301e-01
1.3418728e-01 4.9936959e-01 3.6131287e-01 -9.8296739e-02
-8.6773150e-02 -1.8985987e-01 7.8233081e-01 -9.3319798e-01
-5.2925086e-01 1.1320268e-01 -4.6805242e-01 -3.0520362e-01
-4.1820338e-01 -4.2286962e-01 -1.2715149e+00 1.3861518e-01
1.0174551e-01 -7.5714655e-02 6.3499756e-02 7.4140847e-01
7.7534825e-01 6.5225142e-01 6.7770243e-01 -8.8945299e-01
-2.4811243e-01 -7.5644928e-01 -6.5043426e-01 8.0390394e-01
8.2376391e-01 -5.6448942e-01 -5.1897579e-01 3.6776963e-01]
|
[14.963303565979004, -2.88017201423645]
|
e339cc3e-5fd9-4314-b458-7980a0791d62
|
correction-of-cloud-removal-by-fusing-multi
|
1707.09959
| null |
http://arxiv.org/abs/1707.09959v1
|
http://arxiv.org/pdf/1707.09959v1.pdf
|
Correction of "Cloud Removal By Fusing Multi-Source and Multi-Temporal Images"
|
Remote sensing images often suffer from cloud cover. Cloud removal is
required in many applications of remote sensing images. Multitemporal-based
methods are popular and effective to cope with thick clouds. This paper
contributes to a summarization and experimental comparation of the existing
multitemporal-based methods. Furthermore, we propose a spatiotemporal-fusion
with poisson-adjustment method to fuse multi-sensor and multi-temporal images
for cloud removal. The experimental results show that the proposed method has
potential to address the problem of accuracy reduction of cloud removal in
multi-temporal images with significant changes.
|
['Qing Cheng', 'Zhiwei Li', 'Xinghua Li', 'Chengyue Zhang', 'Huanfeng Shen']
|
2017-07-25
| null | null | null | null |
['cloud-removal']
|
['computer-vision']
|
[ 3.69602203e-01 -1.27618325e+00 4.04017031e-01 -1.64627716e-01
-7.79774785e-01 -6.81999266e-01 3.53539139e-01 -8.82595330e-02
-3.68202299e-01 7.76425421e-01 -5.23737967e-01 -2.74559826e-01
-3.12981844e-01 -9.82962251e-01 2.82663144e-02 -1.11052012e+00
-1.02155589e-01 1.30471066e-01 4.52389956e-01 -3.10363799e-01
3.02177548e-01 1.07458043e+00 -1.81903601e+00 1.96369022e-01
1.12978458e+00 8.85953665e-01 8.07941020e-01 6.80046976e-01
-1.50148012e-02 2.69102484e-01 -3.05325687e-01 3.08188796e-01
6.27861142e-01 -2.01992333e-01 -3.61797661e-01 3.46889704e-01
2.99811006e-01 -1.97697222e-01 3.48570049e-01 1.17176735e+00
5.28729737e-01 1.32281199e-01 6.81193173e-01 -9.41720665e-01
-4.16907728e-01 -9.28134173e-02 -1.27943408e+00 8.69415820e-01
-2.55240887e-01 1.92740094e-02 2.54018456e-01 -1.13391268e+00
2.51626581e-01 9.68566239e-01 7.52696514e-01 -1.74476564e-01
-5.80477357e-01 -6.95496738e-01 1.47334218e-01 1.62936673e-01
-1.74001741e+00 -2.32587561e-01 2.91636348e-01 -5.39876640e-01
9.41244423e-01 6.30343318e-01 9.22747552e-01 -2.74264365e-01
3.54442805e-01 1.51589155e-01 1.76088464e+00 -5.08089304e-01
-1.50669128e-01 -1.26912311e-01 1.00236580e-01 -2.01123819e-01
7.86194384e-01 1.03497669e-01 4.50914353e-02 -1.55594781e-01
5.44040561e-01 5.47777295e-01 -1.15956880e-01 4.38944340e-01
-6.76766157e-01 8.54346812e-01 3.78281116e-01 6.12502635e-01
-8.08995008e-01 1.03520788e-01 3.40705663e-02 1.37971237e-01
1.01915348e+00 6.64613023e-02 -3.79340768e-01 3.63755733e-01
-1.33638752e+00 6.03446484e-01 -1.16797850e-01 9.49014962e-01
6.87826574e-01 4.35763270e-01 1.10296234e-01 4.61194396e-01
2.89784759e-01 1.35277760e+00 7.68799558e-02 -8.38343561e-01
2.18145803e-01 2.46881261e-01 4.08135802e-01 -9.85859931e-01
2.03882474e-02 -2.63376713e-01 -9.38191116e-01 3.47286135e-01
-4.22553957e-01 2.18398347e-01 -9.33156550e-01 3.94871950e-01
3.28621238e-01 3.70215297e-01 1.38780713e-01 8.79010916e-01
6.11788452e-01 9.58667815e-01 1.26988292e-01 -7.71069467e-01
1.15898967e+00 -5.91285944e-01 -1.15405750e+00 -1.04860455e-01
-1.42556489e-01 -1.04000986e+00 4.34228361e-01 2.14581504e-01
-7.37736404e-01 -2.09296077e-01 -5.15297592e-01 5.53549528e-01
-6.58830762e-01 -7.95498714e-02 5.20150721e-01 4.72813576e-01
-9.82026994e-01 3.14487547e-01 -6.39241159e-01 -3.38797987e-01
1.46572784e-01 -4.86023575e-02 1.35261700e-01 -1.69753700e-01
-8.38116407e-01 1.08128858e+00 4.16668206e-01 5.53006649e-01
-5.18710375e-01 -5.30188322e-01 -5.08965969e-01 -2.42307082e-01
1.03751265e-01 -3.00006598e-01 9.04788435e-01 -8.89553010e-01
-7.42122889e-01 7.56364584e-01 -4.68942136e-01 -2.17642233e-01
3.30581218e-01 -1.55209787e-02 -6.49931371e-01 4.85930741e-01
3.27871799e-01 -6.98773712e-02 9.07668352e-01 -1.58902299e+00
-8.83507550e-01 -5.40314436e-01 -5.38371921e-01 4.12672818e-01
2.16345057e-01 5.61970472e-01 -1.14735186e-01 -7.32065678e-01
4.58853215e-01 -8.45672667e-01 -4.48770881e-01 -1.82596982e-01
1.77279025e-01 3.37789476e-01 1.42316091e+00 -6.61900282e-01
9.92546856e-01 -2.04124713e+00 -5.00849187e-01 1.98759720e-01
-5.52724116e-02 7.14907467e-01 3.20067331e-02 4.85694945e-01
4.18755747e-02 4.89359230e-01 -6.47075713e-01 -2.72355855e-01
-8.19585741e-01 3.17405581e-01 -3.23679447e-01 8.53039861e-01
1.47379890e-01 4.99098182e-01 -6.81884110e-01 -5.91233909e-01
7.54296005e-01 6.51899755e-01 2.95734018e-01 -1.34733960e-01
6.22069836e-03 7.09531188e-01 -5.49060285e-01 1.01053286e+00
1.77561200e+00 1.83393985e-01 -1.00101672e-01 1.69341639e-01
-8.78270984e-01 -4.42943305e-01 -1.16621554e+00 7.92857111e-01
-2.51650631e-01 5.29413700e-01 4.41210955e-01 -4.43445235e-01
7.42039621e-01 4.23283547e-01 6.20610297e-01 -5.37859321e-01
-1.41789485e-02 5.23852408e-01 -3.62744957e-01 -8.37568521e-01
9.87667143e-01 -5.90195239e-01 4.79090452e-01 4.93205972e-02
-8.52896035e-01 -8.73027861e-01 -1.59406200e-01 -1.37388900e-01
2.11694717e-01 2.36979383e-03 5.09978533e-01 -2.72255391e-01
5.64947844e-01 4.89479989e-01 6.74407899e-01 6.70614064e-01
-4.23780620e-01 6.96098208e-01 -7.09656715e-01 -5.50603509e-01
-1.09007168e+00 -6.14840329e-01 -3.30958962e-01 4.30253416e-01
2.92241603e-01 2.46859133e-01 -1.67793334e-01 2.18470484e-01
1.42839491e-01 4.11869794e-01 -3.51715535e-01 6.20429456e-01
-3.64278883e-01 -1.29863620e+00 1.62122399e-01 1.46981046e-01
8.74631822e-01 -8.51268768e-01 -7.27913439e-01 2.26732790e-01
-5.11063278e-01 -1.25802374e+00 1.79641470e-01 -3.25276673e-01
-1.35739720e+00 -8.73367548e-01 -6.19108617e-01 -2.87023306e-01
2.86346555e-01 1.59077001e+00 9.64431047e-01 5.56293607e-01
-4.58228052e-01 2.32092038e-01 -7.94052005e-01 -8.49912286e-01
6.39895871e-02 -4.66528684e-01 -2.09446207e-01 2.82765646e-02
4.39457178e-01 -6.39029801e-01 -5.20348847e-01 1.49827167e-01
-1.19536018e+00 -2.75768876e-01 2.88243622e-01 1.29307166e-01
9.92053866e-01 6.57904327e-01 6.69584200e-02 -5.97104311e-01
1.72606021e-01 -5.62354267e-01 -1.16030753e+00 2.75412440e-01
-7.05207586e-01 -8.42149317e-01 -9.31788012e-02 6.55071437e-02
-1.26304662e+00 1.46503508e-01 4.09386277e-01 -5.76080084e-01
-2.15617314e-01 7.35170603e-01 2.78375000e-01 -6.40535057e-01
3.28611672e-01 5.18919408e-01 -3.07785541e-01 -5.38459301e-01
-3.15598324e-02 8.53627264e-01 3.40869904e-01 -1.66886300e-01
1.06735802e+00 1.12731326e+00 1.67450994e-01 -1.41541088e+00
-5.58834195e-01 -1.33917022e+00 -8.29883516e-01 -6.17165387e-01
1.03284633e+00 -1.26982021e+00 -1.82191685e-01 9.81143475e-01
-1.21271098e+00 -1.56956464e-02 1.27109528e-01 6.43041790e-01
1.61671340e-02 4.22912270e-01 -1.29849583e-01 -1.64566708e+00
-5.78149199e-01 -8.36496472e-01 1.11922908e+00 1.90483689e-01
8.13080788e-01 -8.18629205e-01 9.76879671e-02 4.30465519e-01
8.08517337e-01 6.65017486e-01 9.42029466e-04 2.34770551e-01
-9.75983918e-01 -9.02263969e-02 -5.36175847e-01 2.01090008e-01
3.72259915e-01 5.44144213e-01 -1.24795187e+00 -2.65762299e-01
2.60415703e-01 3.56964111e-01 9.99918759e-01 9.22653794e-01
9.43980992e-01 -3.88742499e-02 -3.01825821e-01 8.23087990e-01
2.40109515e+00 3.03651601e-01 8.74542356e-01 6.14869475e-01
5.82975030e-01 4.12841022e-01 1.20668054e+00 6.43943071e-01
2.47168139e-01 3.16390127e-01 8.25029492e-01 -3.40736389e-01
1.06486939e-01 8.05057347e-01 -2.67845124e-01 7.95764863e-01
-8.04299712e-01 -2.94834048e-01 -9.70727265e-01 9.27316785e-01
-1.75980985e+00 -1.39677250e+00 -1.13666606e+00 1.86273396e+00
2.57379472e-01 -7.17849731e-01 -1.95522606e-01 9.55058783e-02
1.00522947e+00 3.31441253e-01 -1.87954351e-01 -4.87096421e-02
-4.58236486e-01 2.51688451e-01 1.19273996e+00 6.26390278e-01
-1.25838161e+00 9.34413314e-01 7.17735338e+00 4.79740798e-01
-1.28184986e+00 7.24584341e-01 -3.07343639e-02 2.35084444e-01
-3.76773357e-01 4.61497046e-02 -6.55718505e-01 2.64925927e-01
7.00582385e-01 -2.09847823e-01 8.80818069e-02 2.13484213e-01
9.01823342e-01 -9.40235734e-01 2.62402356e-01 1.00376046e+00
-6.41741455e-02 -1.14107740e+00 6.60073981e-02 1.31171197e-01
1.15131891e+00 5.73059499e-01 -5.38476333e-02 -4.23059642e-01
1.03560956e-02 -7.88251579e-01 6.91682398e-01 8.08181345e-01
7.51965165e-01 -6.22387648e-01 9.33617294e-01 3.92150670e-01
-1.76875639e+00 1.47534490e-01 -5.61183989e-01 -4.46504772e-01
1.63035661e-01 9.79186893e-01 -3.89092475e-01 1.17724681e+00
1.06780362e+00 7.55316079e-01 -3.32048744e-01 1.47234011e+00
1.21138774e-01 2.01695189e-01 -3.81831020e-01 4.14291859e-01
3.43700022e-01 -8.01907718e-01 6.10794127e-01 1.20628917e+00
7.86435425e-01 8.82974982e-01 1.65970847e-01 4.72170651e-01
7.50323772e-01 -2.60404544e-03 -9.57198083e-01 1.32427230e-01
7.14045346e-01 1.05259132e+00 -7.49946713e-01 -4.38717663e-01
-3.53232771e-01 6.26854360e-01 -7.41306245e-01 2.84051597e-01
-6.67299211e-01 -4.95735519e-02 6.57172740e-01 1.94284350e-01
5.27639389e-01 -5.93608797e-01 -5.98591566e-01 -1.13712382e+00
1.88728906e-02 -5.30996919e-01 2.88213164e-01 -1.23421586e+00
-9.04659152e-01 5.80226898e-01 5.75891435e-01 -1.65241849e+00
2.94520468e-01 -2.70766646e-01 -6.77474439e-01 1.49396098e+00
-2.40726447e+00 -1.49151194e+00 -8.66959929e-01 7.63434350e-01
5.48901796e-01 1.97861120e-02 7.70830274e-01 2.27902934e-01
-2.13822588e-01 -5.44720292e-01 4.19788867e-01 -4.74627107e-01
2.68162668e-01 -9.21457231e-01 4.60069440e-02 1.56918538e+00
-6.33866668e-01 2.02542990e-01 9.16547835e-01 -9.43251967e-01
-1.03433824e+00 -1.58286285e+00 8.76155674e-01 4.70983535e-02
2.97369987e-01 3.38756472e-01 -9.64471221e-01 5.99312365e-01
2.86716938e-01 3.46540332e-01 5.02396405e-01 -5.46911955e-01
4.34547737e-02 -4.99247432e-01 -1.53037167e+00 -3.48623991e-02
2.58794665e-01 -3.42749745e-01 -3.26072633e-01 5.61544180e-01
3.88295352e-01 -2.01675221e-01 -7.53632069e-01 7.00614989e-01
4.02536064e-01 -1.15846360e+00 8.34420443e-01 3.20763916e-01
1.07451871e-01 -8.23070526e-01 -4.31891143e-01 -1.06171489e+00
-4.20310169e-01 -1.79146513e-01 6.54004335e-01 9.42420244e-01
-1.51024908e-02 -5.57139635e-01 2.26895064e-01 1.42612159e-01
-3.61921303e-02 2.15945393e-01 -1.01482725e+00 -9.91118729e-01
9.65581834e-02 -3.72980565e-01 6.52655780e-01 1.21372843e+00
-1.02484643e+00 -5.41787922e-01 -3.41505259e-01 8.18276227e-01
9.40543592e-01 4.69446152e-01 4.87168789e-01 -1.49712348e+00
4.17374402e-01 -8.94532949e-02 -1.92859396e-01 1.75944775e-01
-2.51862347e-01 -2.43141308e-01 7.09721074e-02 -1.66689014e+00
4.09922808e-01 -5.61535776e-01 -7.28726462e-02 2.77795851e-01
-2.93836355e-01 5.97014964e-01 3.55436802e-01 7.45572090e-01
-2.32939608e-02 3.82650703e-01 1.40172589e+00 -1.32425845e-01
-8.42282921e-02 1.09271713e-01 -6.65463656e-02 4.99021441e-01
1.08506644e+00 -8.62539589e-01 -1.94955140e-01 -7.66873837e-01
1.61097646e-01 1.17374852e-01 4.72325087e-01 -1.05500841e+00
-4.76125069e-02 -9.41790104e-01 4.95328084e-02 -1.34663510e+00
4.30608869e-01 -1.24366975e+00 6.39644027e-01 5.29382348e-01
6.59718096e-01 5.34817517e-01 4.58206654e-01 5.49186826e-01
-5.10842562e-01 -2.40772754e-01 1.21599233e+00 -6.28786027e-01
-9.73972797e-01 4.92948711e-01 -7.40392327e-01 -6.84050441e-01
1.25522923e+00 -4.92813647e-01 -2.89549619e-01 -1.54452011e-01
-7.19518185e-01 1.20438792e-01 5.65884531e-01 -1.09806314e-01
8.43674779e-01 -1.02726138e+00 -1.06418216e+00 -3.96950990e-02
2.50827909e-01 2.82755364e-02 4.54262376e-01 1.01890051e+00
-1.12119114e+00 2.11553842e-01 -3.24583620e-01 -8.04202318e-01
-1.67759418e+00 3.33695650e-01 6.78432524e-01 -2.41629034e-02
-5.07697165e-01 6.36816204e-01 -2.61915952e-01 -2.29619190e-01
-5.81625581e-01 -3.28989655e-01 -3.09058398e-01 9.33297873e-02
6.58018887e-01 4.20751750e-01 4.40234065e-01 -9.45720673e-01
-6.25200331e-01 9.39705670e-01 5.00641584e-01 -3.72123122e-01
1.63495433e+00 -5.68400264e-01 -7.02636957e-01 7.62981832e-01
5.99266469e-01 2.30610650e-02 -8.34131718e-01 -3.00661832e-01
-3.35808784e-01 -1.07362127e+00 4.63414550e-01 -5.62998414e-01
-1.33146894e+00 7.87909389e-01 9.38333035e-01 2.65814275e-01
1.60919964e+00 -5.97431660e-01 3.77469093e-01 1.27136722e-01
3.67843240e-01 -9.05708969e-01 -5.63396633e-01 6.47294700e-01
9.13943648e-01 -1.51537824e+00 7.18247414e-01 -8.11354101e-01
-4.63581204e-01 1.02007246e+00 2.63871759e-01 -7.75373355e-02
1.04311955e+00 2.93695182e-01 3.35012406e-01 -5.02315402e-01
-3.89908433e-01 -7.03859270e-01 -1.60386175e-01 8.82211626e-01
2.09699690e-01 4.30032462e-01 -4.23177481e-01 -4.44966137e-01
4.45330977e-01 8.50717798e-02 8.30426693e-01 1.30668163e+00
-6.93723023e-01 -8.38552654e-01 -1.36189675e+00 3.76542598e-01
-4.98091251e-01 -3.97682726e-01 -5.05186282e-02 8.11225772e-01
3.32892984e-01 1.32270706e+00 2.43773773e-01 -3.25925387e-02
-1.04038380e-02 -3.08381289e-01 3.06970894e-01 -6.48259342e-01
-7.04583287e-01 7.06956208e-01 -3.27596158e-01 -1.31219581e-01
-1.56386817e+00 -1.15976083e+00 -8.05593371e-01 -6.29218519e-01
-7.67090082e-01 1.80920422e-01 9.41373348e-01 9.10943806e-01
-3.74027528e-02 4.23663080e-01 9.46937621e-01 -1.22258472e+00
-5.82232885e-02 -1.11708581e+00 -1.27837050e+00 -1.45091698e-01
7.69652843e-01 -6.56285048e-01 -4.56599325e-01 1.96026921e-01]
|
[9.766403198242188, -1.7598479986190796]
|
4a57a840-9e4f-40e7-ba6d-3a8d033b5622
|
arrhythmia-classifier-using-convolutional
|
2202.12943
| null |
https://arxiv.org/abs/2202.12943v1
|
https://arxiv.org/pdf/2202.12943v1.pdf
|
Arrhythmia Classifier Using Convolutional Neural Network with Adaptive Loss-aware Multi-bit Networks Quantization
|
Cardiovascular disease (CVDs) is one of the universal deadly diseases, and the detection of it in the early stage is a challenging task to tackle. Recently, deep learning and convolutional neural networks have been employed widely for the classification of objects. Moreover, it is promising that lots of networks can be deployed on wearable devices. An increasing number of methods can be used to realize ECG signal classification for the sake of arrhythmia detection. However, the existing neural networks proposed for arrhythmia detection are not hardware-friendly enough due to a remarkable quantity of parameters resulting in memory and power consumption. In this paper, we present a 1-D adaptive loss-aware quantization, achieving a high compression rate that reduces memory consumption by 23.36 times. In order to adapt to our compression method, we need a smaller and simpler network. We propose a 17 layer end-to-end neural network classifier to classify 17 different rhythm classes trained on the MIT-BIH dataset, realizing a classification accuracy of 93.5%, which is higher than most existing methods. Due to the adaptive bitwidth method making important layers get more attention and offered a chance to prune useless parameters, the proposed quantization method avoids accuracy degradation. It even improves the accuracy rate, which is 95.84%, 2.34% higher than before. Our study achieves a 1-D convolutional neural network with high performance and low resources consumption, which is hardware-friendly and illustrates the possibility of deployment on wearable devices to realize a real-time arrhythmia diagnosis.
|
['Zhi Qi', 'Hao liu', 'Junguang Huang', 'Zhiqing Li', 'Ninghao Pu', 'Ao Wang', 'Hanshi Sun']
|
2022-02-27
| null | null | null | null |
['arrhythmia-detection']
|
['medical']
|
[ 6.84813708e-02 -1.48473725e-01 -1.86291456e-01 -3.11650097e-01
-2.92603076e-01 1.20624721e-01 -3.45245123e-01 2.88884908e-01
-6.01538301e-01 7.96077788e-01 -2.79319465e-01 -2.88791806e-01
-2.53689915e-01 -9.15476024e-01 -2.42304668e-01 -7.97998726e-01
-2.74703115e-01 7.33786263e-03 1.12118889e-02 4.22691181e-02
-1.50116876e-01 5.74791431e-01 -1.55632699e+00 2.12760150e-01
8.34578693e-01 1.46807694e+00 2.40195826e-01 4.41322088e-01
1.63131967e-01 3.18716288e-01 -8.33515584e-01 -4.43476051e-01
4.77482453e-02 -5.29951572e-01 -3.02273214e-01 -3.53468806e-01
6.49427176e-02 -5.56017935e-01 -3.13550800e-01 9.45905328e-01
1.07537460e+00 -3.51132661e-01 4.29216027e-01 -6.92179739e-01
-2.41918251e-01 6.39920771e-01 -2.99251795e-01 2.84463465e-01
-2.93317348e-01 -3.17844981e-03 5.17524600e-01 -4.65025455e-01
1.41354799e-01 6.45894825e-01 9.18298542e-01 5.42058468e-01
-7.60095596e-01 -8.61186862e-01 -5.02682865e-01 4.78095531e-01
-1.73085141e+00 -2.20187902e-01 8.78999233e-01 -9.01934132e-02
6.05327785e-01 4.21627909e-01 1.02797389e+00 9.60976303e-01
2.44489074e-01 4.59691495e-01 6.54973388e-01 -3.33118945e-01
2.72590488e-01 5.56899197e-02 -1.60121337e-01 5.74888647e-01
6.24112368e-01 -1.78746238e-01 -1.54365957e-01 7.03744218e-02
8.69971097e-01 4.31490928e-01 -4.43813741e-01 2.60526210e-01
-1.15982723e+00 5.93019605e-01 6.69603467e-01 7.46229231e-01
-4.02058899e-01 1.54721051e-01 7.13338912e-01 2.63935365e-02
2.86751330e-01 4.32302803e-01 -4.38533694e-01 -4.26641434e-01
-9.21547592e-01 3.35602532e-03 4.70173120e-01 4.06641930e-01
1.99724868e-01 3.83681864e-01 -2.38794118e-01 9.42377090e-01
-1.42300259e-02 5.00684440e-01 8.01427722e-01 -7.19701827e-01
2.24478543e-01 6.31826580e-01 -3.34803998e-01 -1.26455486e+00
-7.74363339e-01 -1.14289653e+00 -1.83214188e+00 -3.33694816e-02
2.78390825e-01 -1.08157001e-01 -4.41543043e-01 1.52494764e+00
6.71324432e-02 1.17765419e-01 6.70316964e-02 1.10836947e+00
8.95104587e-01 5.46878517e-01 4.24787402e-02 -3.93925399e-01
1.63739097e+00 -4.05372977e-01 -9.47386801e-01 2.28963763e-01
7.61454523e-01 -3.96542311e-01 1.07214141e+00 7.86604166e-01
-9.49370623e-01 -7.81509995e-01 -1.37174284e+00 8.21410790e-02
2.58917380e-02 7.38256276e-01 6.23806834e-01 8.60216558e-01
-8.13238025e-01 9.19589281e-01 -6.70024812e-01 -8.52187350e-02
8.96906435e-01 3.72315645e-01 -4.08293307e-02 2.43176967e-01
-1.33651388e+00 4.03542191e-01 5.89033663e-01 2.45143846e-01
-4.38234329e-01 -4.68407661e-01 -3.82232666e-01 4.51106936e-01
-3.93409170e-02 -5.65909863e-01 7.96094954e-01 -7.16333508e-01
-1.42196059e+00 5.00840425e-01 2.25844085e-01 -7.17592657e-01
4.00151581e-01 -1.01081476e-01 -7.23264992e-01 3.22479457e-01
-3.16266865e-01 4.09157693e-01 7.23607361e-01 -3.54220271e-01
-5.24919629e-01 -4.52657282e-01 -1.82524383e-01 -3.98276141e-03
-1.14736021e+00 -4.11429375e-01 -1.58539549e-01 -8.41972768e-01
1.55334994e-01 -6.31913364e-01 -1.59871802e-01 3.39534611e-01
-2.21971184e-01 -7.83661604e-02 5.58944762e-01 -7.25837469e-01
1.67996490e+00 -2.44775748e+00 -2.31255203e-01 6.61614770e-03
4.24781024e-01 7.77187288e-01 2.66473293e-01 -1.77609414e-01
1.54968172e-01 1.99443206e-01 -3.08415383e-01 -1.16027847e-01
-3.86491656e-01 2.16094758e-02 1.83283165e-02 3.58568817e-01
9.79309231e-02 7.40251601e-01 -4.71337587e-01 -4.54524696e-01
1.58526108e-01 7.80958831e-01 -4.67605650e-01 -2.89894044e-02
2.23976031e-01 2.94561505e-01 -3.85673046e-01 5.34866810e-01
7.90419042e-01 -1.40302360e-01 2.72797644e-01 -4.15595859e-01
1.28152609e-01 1.56571776e-01 -9.86185670e-01 1.81394088e+00
-3.64478052e-01 4.87606168e-01 -3.59661788e-01 -1.26909125e+00
1.34684455e+00 4.92214203e-01 4.84383553e-01 -9.16901827e-01
4.03696984e-01 4.79731530e-01 2.14351386e-01 -6.01719260e-01
5.44848433e-03 -1.05388403e-01 2.10447162e-02 -2.71593742e-02
-1.86283097e-01 2.88268209e-01 -3.02457716e-02 -2.93101639e-01
1.00354183e+00 -1.92969173e-01 2.33059019e-01 -2.84643501e-01
5.60237288e-01 -4.70053345e-01 7.69277275e-01 5.28208852e-01
-2.09362522e-01 5.38988233e-01 2.50548303e-01 -9.90342259e-01
-9.40822005e-01 -7.25178897e-01 -4.51571107e-01 1.75429717e-01
5.75375929e-02 -5.29256463e-01 -7.62108445e-01 -2.44498074e-01
-2.56748736e-01 1.21788777e-01 -2.22864166e-01 -4.56400275e-01
-7.10935473e-01 -1.01461375e+00 1.07796907e+00 4.43501353e-01
1.07345986e+00 -1.00026059e+00 -1.14287090e+00 4.60585594e-01
-2.41914764e-01 -1.01390898e+00 1.16521567e-01 -2.88485792e-02
-1.39045811e+00 -7.12661207e-01 -9.75752234e-01 -7.75366604e-01
2.36022145e-01 -2.46429890e-01 8.14444125e-01 2.79330015e-01
-6.30662799e-01 -5.65002143e-01 -2.52227426e-01 -4.95262384e-01
-2.01599985e-01 4.76240605e-01 2.45526806e-01 2.90579312e-02
1.82998881e-01 -7.75470376e-01 -1.08296502e+00 -4.41747457e-02
-6.96636200e-01 3.49785127e-02 8.62413228e-01 6.55454159e-01
5.38394034e-01 2.43432865e-01 9.93645430e-01 -5.75098336e-01
5.41917622e-01 -1.20745927e-01 -4.43864822e-01 1.40727619e-02
-7.06288338e-01 -2.64460534e-01 1.00886476e+00 -4.26887542e-01
-4.89177674e-01 7.29846805e-02 -6.71585321e-01 -3.76899689e-01
-8.70917514e-02 3.15959007e-01 -1.69058099e-01 1.77560478e-01
7.05464423e-01 2.48972505e-01 1.38771161e-01 -5.34278572e-01
-1.71447009e-01 9.68745291e-01 2.92668730e-01 -1.28215447e-01
3.02465320e-01 5.89960068e-02 3.02436709e-01 -8.97916198e-01
-3.69971901e-01 -2.13923659e-02 -2.86526203e-01 -8.86601955e-02
8.82255852e-01 -9.57415879e-01 -1.08835888e+00 4.42499280e-01
-1.12524509e+00 5.31194285e-02 -3.79661471e-01 7.37722516e-01
-1.23156488e-01 4.35183555e-01 -7.51496971e-01 -7.84514070e-01
-1.06086826e+00 -1.08132100e+00 6.52049959e-01 1.94901764e-01
-1.02171741e-01 -4.30405527e-01 -3.60920787e-01 -7.82717094e-02
7.48242557e-01 4.54940081e-01 9.18472230e-01 -2.82638937e-01
-3.11981857e-01 -2.31966361e-01 -2.30548069e-01 5.33286750e-01
1.43224165e-01 -3.38704586e-01 -9.75711703e-01 -3.54562283e-01
2.82675475e-01 -3.62972952e-02 6.83826804e-01 4.31156844e-01
1.85508454e+00 -4.30314213e-01 -2.76269525e-01 7.84744680e-01
1.31012785e+00 4.87647235e-01 1.00588834e+00 1.18266143e-01
4.31921750e-01 4.45817318e-03 2.73899406e-01 6.56291723e-01
-1.78207308e-02 6.41417027e-01 4.45605129e-01 -3.75792146e-01
-1.43041357e-01 2.40061924e-01 -1.37187809e-01 1.17529261e+00
-3.52374822e-01 -6.05955794e-02 -7.25414932e-01 2.85571426e-01
-1.44019926e+00 -7.85594404e-01 -1.87165901e-01 2.36344934e+00
9.95612502e-01 3.35323453e-01 1.75579693e-02 8.02257299e-01
5.52887738e-01 -1.81765005e-01 -4.19385701e-01 -5.20913005e-01
-4.38098498e-02 6.01078868e-01 1.66338459e-01 -2.62016356e-01
-1.06013215e+00 8.06260034e-02 4.67488432e+00 1.08907282e+00
-1.46853924e+00 2.18117982e-01 1.04347801e+00 -2.63084471e-01
3.55802953e-01 -5.95220625e-01 -7.16617763e-01 6.58905149e-01
1.21284020e+00 2.13340089e-01 6.09048419e-02 7.99207032e-01
1.47933573e-01 2.74699986e-01 -6.58009589e-01 1.47200620e+00
-2.12252513e-01 -1.37519109e+00 6.23982660e-02 -9.48054567e-02
2.24623039e-01 -3.63030910e-01 -5.86925261e-02 2.05563098e-01
-9.61688519e-01 -9.32888567e-01 3.40678781e-01 4.63509768e-01
1.37894118e+00 -1.06524682e+00 1.06356215e+00 4.55581665e-01
-1.11969340e+00 -2.92753786e-01 -6.11635983e-01 -2.53560483e-01
1.86169738e-04 1.27216053e+00 -6.84767842e-01 5.00086665e-01
1.02166462e+00 6.92266643e-01 -4.23456013e-01 1.22531736e+00
1.60883740e-01 7.52100587e-01 -3.58057976e-01 -3.52556825e-01
-2.58040816e-01 -4.69261818e-02 3.19942892e-01 1.10967457e+00
6.89007282e-01 2.05501258e-01 -1.09061241e-01 6.18250370e-01
-3.03425193e-01 2.78046966e-01 -4.13231909e-01 3.00672174e-01
4.01072532e-01 1.21554077e+00 -6.59068108e-01 -5.04797935e-01
5.56869507e-02 8.10798347e-01 -5.05515374e-02 -1.85923785e-01
-1.04626012e+00 -1.12434888e+00 3.71400386e-01 1.35997996e-01
5.31389341e-02 -1.97500158e-02 -5.16615689e-01 -1.03228939e+00
3.16567123e-01 -7.47336447e-01 1.67735711e-01 -2.83697248e-01
-7.71877527e-01 9.15610373e-01 -4.92739230e-01 -1.65416515e+00
9.19824094e-02 -3.95908266e-01 -2.39306793e-01 6.67591214e-01
-1.35157502e+00 -5.13028622e-01 -6.02347016e-01 2.51636416e-01
3.84166181e-01 -1.24410413e-01 1.21305776e+00 9.58269179e-01
-6.18656933e-01 9.38848495e-01 1.13644898e-02 1.37151897e-01
5.20502865e-01 -7.66818404e-01 -9.80470981e-03 6.26948655e-01
-2.80808154e-02 5.40472567e-01 1.07370399e-01 -3.35026145e-01
-1.11968553e+00 -1.16811991e+00 8.59876990e-01 3.47746283e-01
-5.12194708e-02 -2.85162270e-01 -9.92486060e-01 -1.60497114e-01
-7.90646523e-02 3.13961357e-02 6.47840559e-01 -2.42806211e-01
9.57587361e-02 -8.51986051e-01 -1.19128263e+00 5.69695473e-01
9.64677453e-01 -2.42475569e-01 -1.74197331e-01 2.22152457e-01
5.75963080e-01 -2.90901303e-01 -1.11798501e+00 8.22544217e-01
7.54960477e-01 -9.81111348e-01 9.64194000e-01 3.35162245e-02
4.89685863e-01 -2.39634961e-01 1.10217437e-01 -8.05887401e-01
-2.91202992e-01 -4.13220227e-01 -2.70891964e-01 1.14107275e+00
2.33358011e-01 -7.50653148e-01 6.82624221e-01 1.29567534e-01
-1.62230626e-01 -1.27851546e+00 -1.11240399e+00 -6.87234223e-01
-2.19364077e-01 -3.34300220e-01 6.38026655e-01 7.50711441e-01
-1.08762972e-01 1.82881922e-01 -4.53747869e-01 -1.66862041e-01
5.47404230e-01 -9.89973545e-02 2.43791595e-01 -1.51585472e+00
-2.88783371e-01 -5.09033024e-01 -7.30866253e-01 -8.55363369e-01
-5.41684508e-01 -8.04595232e-01 -3.68476659e-01 -1.29625070e+00
-1.27301231e-01 -7.48410761e-01 -5.76002002e-01 5.23892641e-01
1.24190658e-01 7.90712118e-01 -4.08388153e-02 2.39773452e-01
-2.77786911e-01 4.84652609e-01 1.18504727e+00 -2.27340594e-01
-2.86663353e-01 2.13240176e-01 -6.06688499e-01 7.00972617e-01
1.18774295e+00 -3.91459823e-01 -3.41770202e-01 -3.55460376e-01
5.24061471e-02 2.09839165e-01 2.51072943e-01 -1.69646680e+00
1.23183027e-01 6.09635472e-01 8.56416106e-01 -4.14809912e-01
3.86988103e-01 -7.79150367e-01 3.50374103e-01 1.13696742e+00
-1.41384467e-01 -1.14958666e-01 2.52942681e-01 2.70870745e-01
-2.39662647e-01 -1.15255639e-01 7.22837925e-01 7.48241544e-02
-2.60312617e-01 4.31953698e-01 -2.43888482e-01 -2.80352920e-01
9.23414886e-01 -1.78807095e-01 2.17149593e-02 4.11802493e-02
-7.34833181e-01 -4.05617476e-01 -1.24756120e-01 1.45547479e-01
8.25705230e-01 -1.35994518e+00 -6.17697895e-01 2.80547231e-01
-8.34152624e-02 7.83576444e-02 5.24526834e-01 9.19729352e-01
-8.62846375e-01 3.94682050e-01 -4.69949991e-01 -7.25700259e-01
-1.06748521e+00 3.13028842e-01 3.54461879e-01 -2.06716910e-01
-1.08804345e+00 4.72746402e-01 -3.00566137e-01 3.62103820e-01
4.05733407e-01 -6.26365066e-01 -3.70521039e-01 -4.46757264e-02
7.47719586e-01 4.42103803e-01 4.44393575e-01 -5.18455133e-02
-4.13489759e-01 7.02924907e-01 2.20769465e-01 5.87523639e-01
1.24381018e+00 1.94841936e-01 -8.14276040e-02 2.28482828e-01
1.24802363e+00 -3.12690169e-01 -8.13119173e-01 1.07886106e-01
-3.26078057e-01 -2.74061203e-01 1.08836576e-01 -7.08534837e-01
-1.50487876e+00 1.30285323e+00 1.25798655e+00 2.00084224e-01
1.71133578e+00 -5.15936196e-01 1.37678015e+00 3.85639489e-01
5.27703941e-01 -7.25600541e-01 -1.18843928e-01 2.22533345e-02
6.21885955e-01 -8.41270864e-01 1.04293682e-01 -1.55369520e-01
-2.53364623e-01 1.52709341e+00 2.68588305e-01 -1.42223477e-01
6.15862727e-01 2.67557055e-01 -9.26499218e-02 1.20592900e-01
-1.88689947e-01 1.65524736e-01 7.83427358e-02 4.25608844e-01
3.85525227e-01 2.13031605e-01 -7.84129262e-01 9.81795430e-01
-1.12705670e-01 4.58978117e-01 2.83921450e-01 4.88882959e-01
-3.59066844e-01 -1.03471196e+00 -3.11014894e-02 7.75430799e-01
-8.67763221e-01 -7.74568915e-02 3.81418288e-01 3.93658608e-01
5.07023752e-01 7.13656545e-01 -1.72096863e-02 -5.70624888e-01
2.34670877e-01 -6.84226900e-02 2.57973582e-01 -1.30076259e-01
-5.64636886e-01 -6.69763312e-02 -1.46091014e-01 -3.69568437e-01
-3.19924563e-01 -1.04323491e-01 -1.21388257e+00 -1.38159037e-01
-2.70319790e-01 6.73577487e-02 8.63239527e-01 6.21062756e-01
7.11025834e-01 9.55744505e-01 5.14702022e-01 -4.55686718e-01
-5.29881239e-01 -8.98002565e-01 -6.45024359e-01 3.21646631e-01
2.33488232e-01 -4.49820608e-01 -5.55292331e-02 -6.86094239e-02]
|
[14.05824089050293, 3.232600450515747]
|
66a06533-6bb0-4a47-b795-d2ce8dbcac40
|
mandarin-singing-voice-synthesis-with
|
2209.10446
| null |
https://arxiv.org/abs/2209.10446v1
|
https://arxiv.org/pdf/2209.10446v1.pdf
|
Mandarin Singing Voice Synthesis with Denoising Diffusion Probabilistic Wasserstein GAN
|
Singing voice synthesis (SVS) is the computer production of a human-like singing voice from given musical scores. To accomplish end-to-end SVS effectively and efficiently, this work adopts the acoustic model-neural vocoder architecture established for high-quality speech and singing voice synthesis. Specifically, this work aims to pursue a higher level of expressiveness in synthesized voices by combining the diffusion denoising probabilistic model (DDPM) and \emph{Wasserstein} generative adversarial network (WGAN) to construct the backbone of the acoustic model. On top of the proposed acoustic model, a HiFi-GAN neural vocoder is adopted with integrated fine-tuning to ensure optimal synthesis quality for the resulting end-to-end SVS system. This end-to-end system was evaluated with the multi-singer Mpop600 Mandarin singing voice dataset. In the experiments, the proposed system exhibits improvements over previous landmark counterparts in terms of musical expressiveness and high-frequency acoustic details. Moreover, the adversarial acoustic model converged stably without the need to enforce reconstruction objectives, indicating the convergence stability of the proposed DDPM and WGAN combined architecture over alternative GAN-based SVS systems.
|
['Yi-Wen Liu', 'Hsin-Min Wang', 'Yu Tsao', 'Yin-Ping Cho']
|
2022-09-21
| null | null | null | null |
['singing-voice-synthesis']
|
['speech']
|
[ 1.18229806e-01 3.84348482e-01 5.40673375e-01 -2.84685474e-02
-1.36004841e+00 -4.81858939e-01 3.69966984e-01 -1.02056754e+00
1.22539565e-01 6.11672521e-01 4.46245342e-01 4.35025059e-02
-3.70898545e-02 -4.77137476e-01 -6.41858339e-01 -9.55562532e-01
3.30864757e-01 2.43961826e-01 -2.20485538e-01 -2.27026701e-01
-3.75776857e-01 3.94004524e-01 -1.53732133e+00 1.26244789e-02
8.63420844e-01 9.28077757e-01 2.49751642e-01 1.20370185e+00
3.24548662e-01 7.37109423e-01 -9.09926891e-01 -4.14950460e-01
4.06063497e-01 -9.16232347e-01 -2.57140934e-01 1.62748769e-02
6.18816197e-01 -3.96825880e-01 -4.24347609e-01 8.94761622e-01
1.07995689e+00 5.45519590e-01 6.70614541e-01 -7.15197563e-01
-7.07304180e-01 7.53378689e-01 6.07880615e-02 -3.21740866e-01
3.35794836e-02 6.32291317e-01 1.37604105e+00 -1.05226374e+00
4.32557285e-01 1.25402486e+00 6.39130890e-01 8.67440999e-01
-1.17758584e+00 -6.82532012e-01 -5.38168132e-01 -1.32136136e-01
-1.22799802e+00 -7.79060245e-01 1.14077771e+00 -2.34154686e-01
6.01553738e-01 5.52827775e-01 6.26847088e-01 1.18563867e+00
-1.95543110e-01 7.12315500e-01 9.18224096e-01 -4.56849635e-01
1.20435372e-01 -8.28972906e-02 -5.10076344e-01 3.78117800e-01
-5.86297452e-01 4.18377787e-01 -7.89652646e-01 2.49483511e-02
7.81899571e-01 -5.69113314e-01 -2.41900846e-01 2.43183315e-01
-9.49374080e-01 6.27254307e-01 2.61888355e-01 5.64845026e-01
-6.00996912e-01 4.28442031e-01 2.55705237e-01 2.68957168e-01
2.62605429e-01 6.61903679e-01 9.54443812e-02 -3.32578927e-01
-1.49358833e+00 4.55048412e-01 7.13139713e-01 6.87991023e-01
-3.41012701e-02 1.31513453e+00 -5.42006314e-01 1.26513100e+00
4.92067516e-01 7.04541087e-01 5.18663108e-01 -1.41841531e+00
2.89470047e-01 -2.39412934e-01 -1.76252931e-01 -5.98081827e-01
1.85203403e-01 -9.43475842e-01 -8.83752525e-01 3.88781339e-01
1.51735842e-01 -3.20986688e-01 -7.45318413e-01 1.95978510e+00
2.65053242e-01 4.08371091e-01 2.00084791e-01 1.04887438e+00
9.07836139e-01 1.10559785e+00 -2.88032919e-01 -2.14920416e-01
8.85806501e-01 -1.23994851e+00 -1.14426219e+00 1.64288417e-01
-1.83618948e-01 -9.54478681e-01 1.43148780e+00 5.43745458e-01
-1.64363050e+00 -1.25880742e+00 -1.07224262e+00 -6.66181743e-02
4.66170639e-01 2.10927024e-01 -2.24825040e-01 7.24622011e-01
-1.30634975e+00 7.88773358e-01 -5.60167372e-01 3.97047937e-01
4.03617956e-02 2.19477788e-01 -3.26868221e-02 5.01060486e-01
-1.18862069e+00 3.64262313e-01 2.17963979e-01 3.26942265e-01
-1.35326159e+00 -9.18075919e-01 -5.93231678e-01 6.35363683e-02
-1.05055436e-01 -1.06711137e+00 1.36774969e+00 -8.90151262e-01
-2.43486166e+00 3.12969893e-01 1.15124501e-01 -4.11086947e-01
4.94397014e-01 -3.09956461e-01 -4.78624165e-01 3.55119593e-02
-3.15751433e-01 5.68645239e-01 1.37817085e+00 -1.36003613e+00
-3.92851233e-01 -1.06785342e-01 -4.74653423e-01 4.52041179e-01
-3.69127631e-01 -9.11088437e-02 -1.43368378e-01 -1.15227997e+00
-2.48213142e-01 -8.76614034e-01 5.49535118e-02 -6.35114372e-01
-5.71497679e-01 -4.31775972e-02 7.59634376e-01 -1.18194354e+00
1.18688262e+00 -2.28273726e+00 6.56643689e-01 9.43015516e-02
2.81674252e-03 5.27702570e-01 -2.29147956e-01 2.69644588e-01
1.13204056e-02 -2.63466746e-01 -4.23308134e-01 -8.08551073e-01
2.95323908e-01 7.89306685e-02 -5.03397822e-01 2.36865327e-01
1.74405612e-02 8.32045376e-01 -5.82164347e-01 -3.03389102e-01
2.56312907e-01 8.11119020e-01 -8.67402256e-01 8.72746468e-01
-1.57693148e-01 9.73656952e-01 6.26499532e-04 5.25648713e-01
2.59318560e-01 7.42903173e-01 -9.29513201e-02 -6.08173832e-02
8.08764100e-02 2.40401760e-01 -1.18149006e+00 1.84145164e+00
-7.48016953e-01 2.62671322e-01 7.12590992e-01 -5.64929128e-01
1.22608364e+00 7.91162312e-01 3.20353389e-01 -1.94688335e-01
2.76102126e-02 5.68136394e-01 1.97310269e-01 -2.52910584e-01
6.22315824e-01 -7.18679249e-01 1.92776680e-01 -7.26118833e-02
5.12426019e-01 -8.13278198e-01 -3.91660035e-01 -3.93342078e-01
8.35788310e-01 3.59151095e-01 -2.46524930e-01 -1.30739614e-01
6.66659057e-01 -5.92440009e-01 6.00349963e-01 4.15895492e-01
5.12044691e-02 9.27557170e-01 -9.11482424e-02 3.07099402e-01
-1.23124945e+00 -1.36512876e+00 1.84518725e-01 9.83543396e-01
-6.55208230e-01 -1.95476010e-01 -1.24581540e+00 -1.31464720e-01
-3.39136213e-01 1.10412896e+00 -7.20493942e-02 -6.46733167e-03
-7.96222866e-01 -5.21236099e-02 1.13648391e+00 2.85229415e-01
2.48826697e-01 -1.50560117e+00 6.40541539e-02 5.50103784e-01
-1.86034307e-01 -7.65578747e-01 -9.25372064e-01 -1.00830905e-01
-5.90605795e-01 -3.16322803e-01 -1.18933415e+00 -8.67237926e-01
-1.04903594e-01 -5.94916403e-01 8.19140911e-01 -5.56517065e-01
1.13279231e-01 3.47672522e-01 -5.93677349e-02 -2.34099716e-01
-1.11015236e+00 4.38002287e-04 5.21972060e-01 3.33592147e-01
-5.13030767e-01 -1.12840462e+00 -6.24262333e-01 1.45849228e-01
-9.75404084e-01 -1.69267744e-01 2.88949400e-01 1.23252380e+00
8.42599571e-01 1.35810703e-01 9.50257480e-01 -4.05657917e-01
9.87646222e-01 -7.49433711e-02 -4.94988173e-01 -2.11741075e-01
-4.13411617e-01 -2.90570557e-01 1.08351064e+00 -4.81044948e-01
-1.21509159e+00 -1.43414944e-01 -9.88633096e-01 -9.17930484e-01
-1.02486219e-02 1.07835695e-01 -5.90041518e-01 2.06795558e-01
4.98955190e-01 4.54606026e-01 6.03517964e-02 -6.21361315e-01
6.29076123e-01 1.02104855e+00 1.19043291e+00 -5.29173136e-01
9.35499668e-01 -2.22908333e-01 -1.10810153e-01 -1.07502007e+00
-4.24537361e-01 -1.33074015e-01 -2.93005079e-01 -3.75817329e-01
9.45604265e-01 -8.52513313e-01 -7.27480173e-01 6.20652616e-01
-1.07693207e+00 -5.01510322e-01 -8.11127841e-01 5.68007290e-01
-1.09880817e+00 3.75093877e-01 -8.58493447e-01 -1.19425082e+00
-8.65973353e-01 -1.29604423e+00 9.24653053e-01 4.22179587e-02
-2.37641841e-01 -7.67735302e-01 4.06028479e-01 7.71343827e-01
6.22841835e-01 4.15219784e-01 6.49649084e-01 -2.93556005e-01
-2.46825144e-01 -4.96939830e-02 6.93580747e-01 1.25296795e+00
-1.45259455e-05 4.41790223e-02 -1.42721641e+00 -3.14392030e-01
4.52971727e-01 -3.10319304e-01 5.85867465e-01 5.72078347e-01
7.28999674e-01 -4.82729226e-01 7.26264834e-01 8.00065100e-01
1.00919175e+00 2.83444196e-01 7.01767266e-01 -3.28092545e-01
8.90349627e-01 6.07359886e-01 4.16984797e-01 2.68769056e-01
-2.24427551e-01 8.10506165e-01 3.87401551e-01 -1.22414023e-01
-8.75836968e-01 -6.80286229e-01 7.94482052e-01 1.72530079e+00
-2.59994537e-01 -2.82043576e-01 -2.53295094e-01 4.52805787e-01
-1.32819033e+00 -9.87877905e-01 8.97149295e-02 2.13816047e+00
8.06516111e-01 -1.87140539e-01 2.49923766e-01 5.67071676e-01
6.48963928e-01 4.17614669e-01 -5.20758212e-01 -6.81247234e-01
-2.06406102e-01 6.14095867e-01 -8.75023082e-02 5.96779108e-01
-7.04811096e-01 9.21815097e-01 5.45912743e+00 1.35778773e+00
-1.03000855e+00 3.83366376e-01 3.62689346e-01 -3.48122030e-01
-5.60738146e-01 -4.52662110e-01 -6.59381866e-01 3.00364435e-01
1.34497070e+00 1.94034040e-01 1.06284893e+00 7.06332386e-01
4.71477687e-01 8.26137960e-01 -7.43912637e-01 8.27893734e-01
-5.76182865e-02 -1.10251272e+00 2.91250367e-03 5.00783809e-02
7.79626667e-01 -2.03162208e-01 5.46634793e-01 3.83154452e-01
-1.48589369e-02 -1.17610252e+00 1.25813019e+00 6.38306916e-01
1.30413449e+00 -9.45425808e-01 3.91297668e-01 3.98564696e-01
-1.05222154e+00 8.51044655e-02 1.24289609e-01 4.01192963e-01
5.18677890e-01 2.72718579e-01 -9.11244154e-01 6.05769038e-01
2.47934058e-01 1.49886176e-01 2.27695182e-01 5.20223200e-01
-3.53936642e-01 1.35143197e+00 -1.92132890e-01 2.56156683e-01
2.74306506e-01 -3.30374897e-01 1.22303927e+00 1.09839618e+00
6.70082450e-01 -1.33579239e-01 -3.34546775e-01 1.15962291e+00
-3.27625215e-01 2.62253016e-01 -5.05264342e-01 -3.53313267e-01
2.95028448e-01 1.11535847e+00 1.09053209e-01 -2.76777074e-02
1.23182878e-01 9.61882710e-01 -5.44662327e-02 3.38036031e-01
-8.41709971e-01 -3.51308405e-01 7.15520740e-01 1.81742415e-01
3.60664099e-01 -2.94920236e-01 -8.22794586e-02 -7.38173187e-01
-1.04556039e-01 -1.33748019e+00 -1.29593402e-01 -7.55460143e-01
-1.14080048e+00 1.05834317e+00 -6.34421885e-01 -1.21393502e+00
-7.14865208e-01 -2.94040531e-01 -8.12550366e-01 1.17200267e+00
-1.16542470e+00 -1.37744319e+00 1.87513128e-01 5.89620769e-01
8.42786551e-01 -6.85044050e-01 9.60395932e-01 3.91801924e-01
-3.24757785e-01 7.93422759e-01 3.31624389e-01 -1.59402847e-01
6.12424433e-01 -1.42400801e+00 3.83372188e-01 9.18157876e-01
3.51111680e-01 2.65171438e-01 9.18472886e-01 -3.62672508e-01
-1.37140226e+00 -1.20561814e+00 4.25706595e-01 -1.08202405e-01
4.81282175e-01 -2.18711704e-01 -7.64454186e-01 1.32100940e-01
5.54648221e-01 -2.27035910e-01 6.33127570e-01 -3.39766234e-01
9.33252461e-03 -3.02096516e-01 -1.19450140e+00 5.83588362e-01
8.16135466e-01 -6.52342558e-01 -6.42437518e-01 -2.07511298e-02
9.09487009e-01 -3.35088253e-01 -1.17133927e+00 4.02252942e-01
4.06332493e-01 -9.22715902e-01 9.74187732e-01 -3.62316012e-01
5.06373048e-01 -4.53891098e-01 -4.57373142e-01 -1.60531151e+00
-2.67671436e-01 -1.39699268e+00 -2.89625138e-01 1.77089536e+00
4.01627332e-01 -1.27738640e-01 3.97238493e-01 1.36700213e-01
-7.78958201e-01 -4.78173375e-01 -1.22989476e+00 -8.42518210e-01
2.61232406e-01 -5.90882838e-01 5.35204649e-01 4.70002979e-01
-5.31814933e-01 4.63766187e-01 -9.95271921e-01 1.55762538e-01
7.36530244e-01 -1.38552278e-01 9.84875321e-01 -6.49687946e-01
-9.41437960e-01 -4.41570044e-01 -3.29347118e-03 -8.29760432e-01
2.83715665e-01 -9.34911370e-01 3.79830271e-01 -1.18552423e+00
-5.73054731e-01 -1.49315894e-01 -4.22118634e-01 -7.28196353e-02
-1.84261128e-01 4.85659957e-01 5.00552654e-01 6.09102100e-02
5.18907718e-02 1.16310322e+00 1.69538534e+00 -8.85714442e-02
-4.28022504e-01 4.23862070e-01 -2.81690449e-01 6.51400447e-01
5.35911202e-01 -3.27296346e-01 -3.44214648e-01 -1.40749127e-01
-5.14180541e-01 7.26325989e-01 1.47688657e-01 -1.19361758e+00
1.03907868e-01 3.37775320e-01 -5.36563694e-02 -4.43143368e-01
1.03657365e+00 -6.03602409e-01 5.73872745e-01 2.71052420e-01
-4.28485930e-01 -4.37031716e-01 4.72596698e-02 4.68384802e-01
-5.85066438e-01 -9.67217013e-02 1.11069942e+00 1.97711855e-01
-1.31983003e-02 8.94299969e-02 -2.50492960e-01 7.74381235e-02
5.69325209e-01 -4.42012362e-02 3.70756507e-01 -5.62599957e-01
-9.10546958e-01 -4.32818681e-01 -6.30854890e-02 2.69804418e-01
6.38300717e-01 -1.55232894e+00 -1.12245905e+00 2.79685915e-01
-3.43742341e-01 -7.83732533e-02 6.37876093e-01 5.80808222e-01
-4.74526584e-01 2.06622615e-01 -1.59259439e-02 -2.98541486e-01
-1.13269520e+00 1.76991493e-01 5.59765816e-01 -2.43754461e-01
-7.04455256e-01 9.13069427e-01 1.02364138e-01 -6.61366880e-01
4.31929559e-01 1.44912256e-02 -3.65506820e-02 -3.09188783e-01
1.71136916e-01 6.49343312e-01 -1.74240783e-01 -6.91777110e-01
1.52872175e-01 3.39760691e-01 5.71572423e-01 -6.59856379e-01
1.36782885e+00 -5.31700589e-02 1.23297393e-01 5.50448835e-01
9.25942898e-01 7.72641540e-01 -1.41482437e+00 6.79605231e-02
-5.12597978e-01 -2.22702473e-01 4.05414730e-01 -7.02336907e-01
-1.20463085e+00 9.19035494e-01 4.21132505e-01 -3.19502875e-02
1.21669126e+00 -3.77518892e-01 1.30917847e+00 -4.74997647e-02
-5.75138964e-02 -1.06002259e+00 -2.79861339e-03 3.34382415e-01
1.37612247e+00 -5.96453428e-01 -7.43811727e-01 1.48399072e-02
-9.78057861e-01 9.56381440e-01 1.83609515e-01 -3.47430885e-01
5.36972165e-01 3.88272375e-01 2.81853884e-01 2.65607983e-01
-3.89268517e-01 1.27008796e-01 6.34667456e-01 5.54590404e-01
3.85341674e-01 1.73709586e-01 -8.05393904e-02 1.08337808e+00
-7.73607492e-01 -9.21480134e-02 -1.50266173e-03 -9.11891554e-03
-2.06212461e-01 -1.01040876e+00 -5.29902399e-01 -8.43050629e-02
-7.36102164e-01 -3.10046136e-01 -1.79577932e-01 2.57283390e-01
1.25229806e-01 1.16950834e+00 -2.84410685e-01 -6.43797219e-01
5.77477217e-01 3.35467786e-01 3.96842092e-01 -2.71565914e-01
-1.16853356e+00 7.13813066e-01 1.67124674e-01 -2.55588055e-01
-1.42520040e-01 -4.30268586e-01 -1.01052046e+00 -9.84881073e-02
-1.97724760e-01 2.81460464e-01 7.10086524e-01 6.00902617e-01
1.84977859e-01 1.15652013e+00 1.06846845e+00 -8.01822603e-01
-9.92290378e-01 -1.36061406e+00 -1.10985124e+00 3.11469495e-01
2.52282679e-01 -7.93357566e-02 -5.00285268e-01 1.51433021e-01]
|
[15.515559196472168, 6.167105197906494]
|
42b84c4e-d2f0-46f8-9b56-ce97949c57df
|
decomposing-normal-and-abnormal-features-of
|
2011.06224
| null |
https://arxiv.org/abs/2011.06224v1
|
https://arxiv.org/pdf/2011.06224v1.pdf
|
Decomposing Normal and Abnormal Features of Medical Images for Content-based Image Retrieval
|
Medical images can be decomposed into normal and abnormal features, which is considered as the compositionality. Based on this idea, we propose an encoder-decoder network to decompose a medical image into two discrete latent codes: a normal anatomy code and an abnormal anatomy code. Using these latent codes, we demonstrate a similarity retrieval by focusing on either normal or abnormal features of medical images.
|
['Ryuji Hamamoto', 'Tatsuya Harada', 'Yusuke Kurose', 'Ryuichiro Hataya', 'Kazuma Kobayashi']
|
2020-11-12
| null | null | null | null |
['content-based-image-retrieval']
|
['computer-vision']
|
[ 5.62546551e-01 2.47152030e-01 -4.91149843e-01 -4.74274337e-01
-6.01885200e-01 -3.58128458e-01 5.48588336e-01 3.64126116e-01
-2.00859625e-02 1.64797723e-01 7.05029607e-01 -3.22198682e-02
-1.35793179e-01 -6.10769570e-01 -2.66446471e-01 -7.81955719e-01
-1.96998313e-01 4.30358797e-01 -5.55704534e-02 3.66047561e-01
-1.67209670e-01 1.83418557e-01 -1.30967426e+00 7.30822623e-01
7.24081039e-01 9.74671423e-01 1.04967803e-01 5.12149751e-01
-2.95170788e-02 8.23877811e-01 -4.50170726e-01 -2.70695031e-01
8.01094994e-02 -8.10057819e-01 -9.30050910e-01 2.14230061e-01
2.03670725e-01 -4.21090692e-01 -5.92404008e-01 1.46147835e+00
1.61053106e-01 -3.22568893e-01 8.80208254e-01 -1.02799833e+00
-1.27812731e+00 3.48839551e-01 -2.26412073e-01 5.36269322e-03
5.48419058e-01 -2.08797261e-01 1.31717396e+00 -5.39475262e-01
6.18848979e-01 1.01905036e+00 4.84985352e-01 4.68301564e-01
-1.08543539e+00 -3.31116974e-01 -1.04429305e-01 2.25930467e-01
-1.35616207e+00 -3.20173651e-01 5.69793224e-01 -5.68079174e-01
5.69540501e-01 2.82965213e-01 9.31359529e-01 8.87935460e-01
9.05354321e-01 9.76711452e-01 7.39275277e-01 -1.56863645e-01
-2.52561029e-02 -2.04699591e-01 -2.04535872e-02 1.15292966e+00
2.28788882e-01 1.38559923e-01 -2.76938349e-01 -4.70549375e-01
7.20407546e-01 6.25230670e-01 -4.41085100e-01 -3.37994903e-01
-1.48938274e+00 1.16106319e+00 4.39334661e-01 3.64514887e-01
-4.07957524e-01 1.67252887e-02 6.76670849e-01 3.09417874e-01
1.37219384e-01 8.69957507e-02 2.43473157e-01 2.01094568e-01
-9.29338634e-01 -2.77579397e-01 5.52961767e-01 9.22787130e-01
2.77427077e-01 -2.37122789e-01 -2.25493476e-01 8.18273246e-01
7.96226025e-01 4.49075907e-01 1.15356815e+00 -7.82779396e-01
-1.14322295e-02 7.68196523e-01 -4.88853157e-01 -1.43777025e+00
-9.61064696e-02 -3.82385194e-01 -1.26654375e+00 -2.93918788e-01
-5.74325144e-01 6.37918353e-01 -1.20510018e+00 1.66240227e+00
-8.27226415e-02 1.70955390e-01 3.53295624e-01 7.47476280e-01
1.17632127e+00 5.15033245e-01 5.10143582e-04 -3.29288840e-01
1.52831233e+00 -1.16114140e+00 -1.15374219e+00 4.75710854e-02
3.87353241e-01 -9.40875232e-01 4.78023022e-01 1.56644970e-01
-1.36708796e+00 -4.08892483e-01 -1.19793797e+00 -2.10749075e-01
-9.00467932e-02 1.57223344e-01 7.54379809e-01 2.00377569e-01
-9.28147554e-01 3.12638074e-01 -1.09520364e+00 -1.12582624e-01
2.88881123e-01 2.41632715e-01 -7.19643056e-01 -8.91265571e-02
-1.12853718e+00 6.45303905e-01 1.96940362e-01 7.83679262e-02
-1.11513567e+00 -3.01587105e-01 -1.22750139e+00 7.34822378e-02
-3.42116207e-01 -8.60387504e-01 1.21490049e+00 -7.89956629e-01
-1.17295539e+00 1.37481594e+00 -1.93140298e-01 -2.45915368e-01
2.61784762e-01 3.72678429e-01 -8.41363788e-01 7.46474087e-01
5.19182384e-01 6.35492146e-01 8.35336506e-01 -8.82910073e-01
-5.04386783e-01 -2.29774222e-01 -2.60257348e-02 1.94114655e-01
-3.62319231e-01 -1.18844308e-01 -4.38366741e-01 -9.41485465e-01
8.17219079e-01 -7.51336753e-01 -2.75990814e-01 3.91503841e-01
-4.90298629e-01 1.41653031e-01 5.26276052e-01 -6.06772065e-01
1.17751145e+00 -2.49220562e+00 3.16289306e-01 2.95155764e-01
8.55793297e-01 -2.13420391e-01 -1.81618258e-01 2.10994899e-01
-5.34425557e-01 4.07759286e-02 -3.51949751e-01 -1.62814409e-01
-1.06163442e-01 4.11374807e-01 -2.12791920e-01 6.90698147e-01
-4.46274392e-02 1.06507921e+00 -1.03715634e+00 -8.88115823e-01
-1.36022091e-01 3.34295779e-01 -5.26381731e-01 3.69026572e-01
4.87516493e-01 3.50559324e-01 -5.12245357e-01 7.93289125e-01
4.29481596e-01 -5.89839697e-01 1.31203711e-01 -4.16496813e-01
3.15900385e-01 1.85221404e-01 -5.22163570e-01 2.01971960e+00
-1.56398803e-01 3.31389487e-01 1.53023124e-01 -1.04921865e+00
6.33119702e-01 7.23892391e-01 9.63889062e-01 -4.57623690e-01
1.00438140e-01 3.28623414e-01 -1.87926795e-02 -6.18558466e-01
2.15413719e-02 -5.59580922e-01 -1.99090958e-01 4.11310703e-01
1.52847692e-01 5.97719587e-02 4.55664322e-02 2.73212552e-01
1.32109892e+00 -4.65532392e-01 3.88603061e-01 -1.63576007e-01
7.40919530e-01 -2.01320544e-01 5.81304371e-01 3.76139075e-01
-2.81778991e-01 7.61736929e-01 6.10711575e-01 -6.70926630e-01
-9.39256489e-01 -1.66233206e+00 -4.81073946e-01 4.35647756e-01
2.13730738e-01 -6.98841274e-01 -3.83736789e-01 -7.68398166e-01
-5.69510087e-02 -1.48072287e-01 -8.35167706e-01 -8.99369717e-01
-1.92168355e-01 -5.15969217e-01 6.23406947e-01 5.09796381e-01
5.38686752e-01 -7.53599286e-01 -5.69826186e-01 -3.92855480e-02
-2.78671354e-01 -7.23179758e-01 -1.05702841e+00 1.42762810e-01
-7.59807944e-01 -9.85278308e-01 -1.02807295e+00 -1.23708677e+00
1.21807694e+00 1.41894236e-01 1.07805872e+00 4.35623407e-01
-7.70357251e-01 4.87041384e-01 -5.99892259e-01 1.69737414e-01
-7.97443330e-01 -2.12035835e-01 3.30392234e-02 1.82080150e-01
9.98211130e-02 -6.99250519e-01 -9.05771434e-01 1.28626749e-01
-1.24996591e+00 4.04323399e-01 1.04219532e+00 1.00208127e+00
7.95301139e-01 -1.75204530e-01 -5.52825034e-02 -7.46452749e-01
5.99727631e-01 -6.21496439e-01 -5.78085380e-03 5.24936855e-01
-6.52141869e-01 2.23663330e-01 4.59522247e-01 -2.65430719e-01
-4.70808834e-01 2.53312200e-01 -1.28759831e-01 -5.56240678e-01
-3.08531467e-02 7.25858092e-01 7.84043595e-02 -4.02803794e-02
3.80800217e-01 7.14726508e-01 1.57999218e-01 -3.70273978e-01
3.58389735e-01 8.19589436e-01 6.88588202e-01 -1.99199572e-01
4.45841879e-01 6.28408134e-01 7.45557249e-02 -3.39402854e-01
-8.25794995e-01 -5.17566621e-01 -6.99790359e-01 9.55762267e-02
1.28544831e+00 -8.35346043e-01 -3.90364558e-01 -1.20935731e-01
-1.01372349e+00 4.13384676e-01 -4.99037892e-01 6.21401310e-01
-5.73552907e-01 8.23056996e-01 -1.06317055e+00 1.62094668e-01
-4.21672672e-01 -1.41596091e+00 1.26706219e+00 -1.30626559e-01
-3.01832408e-01 -1.13253748e+00 2.57271528e-01 -1.02234311e-01
2.58531421e-01 2.25142077e-01 1.31407201e+00 -6.42605603e-01
-3.04307550e-01 -6.26769066e-01 -2.67577082e-01 2.67667443e-01
5.97940803e-01 -7.13479742e-02 -2.86816597e-01 -4.33102161e-01
2.92243659e-01 -2.18445078e-01 8.44175220e-01 2.20601484e-01
1.06288874e+00 -4.36636031e-01 -4.73026276e-01 1.06128168e+00
1.24060178e+00 2.60474294e-01 2.70280540e-01 -2.01830536e-01
4.75473315e-01 1.66272759e-01 4.33541201e-02 2.93770939e-01
4.68080878e-01 4.34735864e-01 2.40092725e-01 -3.70497614e-01
-1.57718658e-01 -2.69745111e-01 1.99614316e-01 1.44833302e+00
1.03319153e-01 2.92104959e-01 -9.81636345e-01 4.67786193e-01
-1.70685780e+00 -7.83620119e-01 1.73613355e-01 1.66827977e+00
1.03841460e+00 -1.24747954e-01 -4.83883291e-01 -2.34477296e-01
7.62885630e-01 2.41320431e-01 -5.07900417e-01 -1.92877471e-01
1.56175226e-01 -6.18435703e-02 1.36587352e-01 2.66351491e-01
-1.02442098e+00 3.24920893e-01 8.56212616e+00 6.77843988e-01
-1.26196635e+00 1.53996944e-01 4.53141212e-01 1.46334976e-01
-7.17561126e-01 -1.81240663e-01 -1.40204027e-01 4.95580763e-01
7.13665366e-01 -6.08052433e-01 3.38238925e-02 9.10104275e-01
-3.75385195e-01 4.28302765e-01 -1.26546657e+00 1.18889654e+00
5.40777981e-01 -1.38901377e+00 8.45427871e-01 8.63336474e-02
5.59638619e-01 -3.79128456e-01 3.08660924e-01 5.45612872e-02
1.73965484e-01 -1.10611796e+00 5.99721611e-01 7.11430967e-01
1.24126375e+00 -2.21405059e-01 7.25269794e-01 7.07466677e-02
-1.59256446e+00 -7.40718469e-03 -3.38903219e-01 4.58687693e-01
-6.16084039e-02 2.93101013e-01 -2.56791502e-01 6.57239139e-01
3.24199528e-01 1.14941430e+00 -4.67993081e-01 9.51025486e-01
-1.96236953e-01 1.68240979e-01 1.47199661e-01 7.11385190e-01
2.64786214e-01 -8.93857628e-02 2.78871179e-01 1.10940611e+00
3.33446413e-01 1.43806085e-01 6.20806634e-01 1.02533984e+00
5.19215362e-03 7.34585226e-02 -6.40431762e-01 -8.22666362e-02
-2.95115076e-02 1.29931545e+00 -9.43675101e-01 -6.19024098e-01
-4.76005703e-01 1.47975397e+00 -2.44556680e-01 9.24763232e-02
-7.99969792e-01 -3.79857540e-01 3.80866617e-01 -1.89303398e-01
-1.77260593e-01 2.37458035e-01 -1.24817550e-01 -1.49443078e+00
-5.61094917e-02 -7.47118771e-01 6.09801054e-01 -6.83980584e-01
-1.34276879e+00 9.55284715e-01 1.43396333e-01 -1.72735882e+00
-1.58717245e-01 -4.25715208e-01 -4.60257977e-01 5.61979234e-01
-1.33180213e+00 -1.28214800e+00 -4.91465002e-01 7.92967975e-01
2.06417441e-01 -5.10001004e-01 1.32010090e+00 5.20275891e-01
-1.55594304e-01 6.16895199e-01 1.23815082e-01 4.02388185e-01
6.19055688e-01 -1.27119946e+00 6.83872923e-02 4.29442555e-01
1.74383167e-02 1.00700176e+00 1.92519069e-01 -6.05953217e-01
-9.07598972e-01 -7.97458112e-01 7.70441413e-01 7.24101067e-02
4.72163856e-01 -6.68641999e-02 -7.22780347e-01 8.16239536e-01
2.61551678e-01 3.43489647e-01 1.27186120e+00 -3.28569323e-01
-4.66059446e-01 7.52452910e-02 -1.05238390e+00 5.30941486e-01
9.10134971e-01 -7.98621535e-01 -9.60845768e-01 5.36396921e-01
9.29656982e-01 -4.53077018e-01 -1.13183749e+00 6.05236530e-01
6.70628607e-01 -6.48889124e-01 9.82255220e-01 -5.94347894e-01
8.11423779e-01 1.00397561e-02 -1.64274246e-01 -1.03052926e+00
-7.52087772e-01 -1.08793184e-01 4.93835695e-02 5.38489759e-01
1.95106208e-01 -5.00139117e-01 7.35981464e-01 3.34031880e-01
-3.04580837e-01 -9.62413192e-01 -1.01064193e+00 -5.91423810e-01
-1.18817352e-01 5.23844063e-02 4.29285258e-01 1.16986763e+00
3.23736846e-01 9.90381464e-02 -1.25015318e-01 -9.24489200e-02
3.41968060e-01 4.08224702e-01 -1.85801342e-01 -1.09328318e+00
-1.75816655e-01 -5.91891229e-01 -9.62689877e-01 -1.04286945e+00
1.73888177e-01 -1.55007863e+00 6.08274676e-02 -1.62349546e+00
8.22767138e-01 5.63108921e-03 -6.39264822e-01 7.57970631e-01
1.32437676e-01 5.65942466e-01 -1.17678717e-01 7.77757525e-01
-8.54253471e-01 5.64116240e-01 1.18272161e+00 -3.72611195e-01
2.59084940e-01 -1.74770430e-01 -5.67990065e-01 8.48425031e-01
3.77126038e-01 -5.89282990e-01 -3.22494447e-01 -3.27278018e-01
3.11267197e-01 4.37259763e-01 1.02935150e-01 -9.14686918e-01
2.92888343e-01 2.72523481e-02 2.19471201e-01 -4.09118593e-01
1.10144965e-01 -9.30159330e-01 1.08193524e-01 1.06590927e+00
-7.77727008e-01 3.49212140e-01 -1.97993234e-01 4.94098663e-01
-9.59788084e-01 -2.10617691e-01 5.25538087e-01 -3.90840769e-01
-3.67243141e-01 6.50795698e-01 -6.46006763e-01 -3.27988386e-01
9.61143911e-01 -2.00280964e-01 -5.84226586e-02 -3.06943744e-01
-1.32940483e+00 7.53720924e-02 3.21641624e-01 3.93493772e-01
1.15766990e+00 -1.99101734e+00 -6.71841800e-01 6.44932330e-01
4.48344529e-01 -3.05919409e-01 3.21062148e-01 1.15769637e+00
-9.12511349e-01 5.53993046e-01 -5.58233321e-01 -6.93814099e-01
-1.46994555e+00 6.08253896e-01 3.49565625e-01 -4.34747845e-01
-8.25824618e-01 6.76380038e-01 4.96034235e-01 -1.00583605e-01
2.29180261e-01 -2.41262734e-01 -2.73875207e-01 1.60312410e-02
4.60163027e-01 -3.42667997e-01 -3.35766613e-01 -8.71207416e-01
-4.21146005e-01 7.77363658e-01 -2.41496846e-01 2.99640396e-03
9.74899888e-01 5.18922918e-02 -6.74525857e-01 3.37183356e-01
1.80618322e+00 -4.49629016e-02 -5.52988827e-01 -4.04517978e-01
-1.77801907e-01 -4.42303419e-01 -1.67606156e-02 -3.73516142e-01
-1.17488980e+00 8.34301412e-01 8.12303722e-01 2.32772425e-01
1.23380184e+00 2.97123313e-01 9.12953019e-01 2.12253973e-01
1.34220243e-01 -6.19956493e-01 2.49396175e-01 2.73521185e-01
9.72402453e-01 -1.00856149e+00 1.88383907e-01 -3.72272730e-01
-6.59759939e-01 1.20737541e+00 1.27352700e-02 -1.97430566e-01
9.71383154e-01 1.61810145e-01 1.30488440e-01 -5.81973016e-01
-8.03568363e-01 -8.34392458e-02 6.14411533e-01 2.42269561e-01
5.22491872e-01 1.00813165e-01 -4.36931163e-01 5.27923286e-01
-1.01021342e-01 -3.42587605e-02 1.03586935e-01 9.47600722e-01
-2.53041774e-01 -1.04678047e+00 -1.79908723e-01 6.32207513e-01
-8.20485055e-01 -2.60214567e-01 -2.31989026e-01 5.58472611e-02
3.12077552e-01 4.13070470e-01 1.23415835e-01 -5.80306888e-01
1.19131051e-01 1.00241853e-02 2.68915683e-01 -1.06041157e+00
-3.65208834e-01 1.16003700e-01 -5.26285648e-01 -6.64860904e-01
-3.08745384e-01 -3.94873619e-01 -1.31397343e+00 1.21404223e-01
-5.82817085e-02 4.12688375e-01 3.69009793e-01 7.40360677e-01
2.14999348e-01 7.86112785e-01 4.93280113e-01 -3.29048246e-01
-4.99061972e-01 -6.62882984e-01 -6.32892787e-01 8.18176270e-01
5.32239735e-01 -4.49220955e-01 -3.64891142e-01 4.11523908e-01]
|
[14.600654602050781, -1.742917776107788]
|
05b59ae1-32e6-49cd-add8-db12073afac3
|
inferring-player-location-in-sports-matches
|
2302.06569
| null |
https://arxiv.org/abs/2302.06569v1
|
https://arxiv.org/pdf/2302.06569v1.pdf
|
Inferring Player Location in Sports Matches: Multi-Agent Spatial Imputation from Limited Observations
|
Understanding agent behaviour in Multi-Agent Systems (MAS) is an important problem in domains such as autonomous driving, disaster response, and sports analytics. Existing MAS problems typically use uniform timesteps with observations for all agents. In this work, we analyse the problem of agent location imputation, specifically posed in environments with non-uniform timesteps and limited agent observability (~95% missing values). Our approach uses Long Short-Term Memory and Graph Neural Network components to learn temporal and inter-agent patterns to predict the location of all agents at every timestep. We apply this to the domain of football (soccer) by imputing the location of all players in a game from sparse event data (e.g., shots and passes). Our model estimates player locations to within ~6.9m; a ~62% reduction in error from the best performing baseline. This approach facilitates downstream analysis tasks such as player physical metrics, player coverage, and team pitch control. Existing solutions to these tasks often require optical tracking data, which is expensive to obtain and only available to elite clubs. By imputing player locations from easy to obtain event data, we increase the accessibility of downstream tasks.
|
['Sarvapali D. Ramchurn', 'Timothy J. Norman', 'Joseph Early', 'Tim Matthews', 'Ryan J. Beal', 'Gregory Everett']
|
2023-02-13
| null | null | null | null |
['pitch-control', 'sports-analytics']
|
['audio', 'computer-vision']
|
[-1.63389713e-01 -5.32250963e-02 -1.85233042e-01 1.22694537e-01
-6.96825087e-01 -6.47892833e-01 5.87293983e-01 4.50827926e-01
-7.77706206e-01 9.37318087e-01 3.09843779e-01 4.38166112e-02
-4.87363160e-01 -8.83023739e-01 -9.82386649e-01 -4.22493219e-01
-4.22430009e-01 1.09511101e+00 5.02863705e-01 -5.47606647e-01
2.30522245e-01 4.38960910e-01 -1.50788128e+00 1.59235194e-01
4.02508467e-01 4.51591104e-01 9.69627574e-02 1.17672896e+00
4.04691964e-01 1.62625325e+00 -7.90856659e-01 6.92251474e-02
4.74819869e-01 -1.84977025e-01 -6.82490826e-01 4.60271537e-02
6.58348724e-02 -4.93595302e-01 -8.07827473e-01 4.03906137e-01
4.73062098e-01 8.00788641e-01 5.61878085e-01 -1.76797855e+00
-2.92622477e-01 5.63677788e-01 -9.64444041e-01 5.02663553e-01
2.11825386e-01 7.06136227e-01 6.95530713e-01 -2.21390985e-02
6.43360257e-01 7.99263060e-01 1.02792549e+00 1.60655573e-01
-1.09993732e+00 -4.66593504e-01 1.38231173e-01 4.57092166e-01
-1.35602129e+00 -4.21169728e-01 2.75281042e-01 -7.06155896e-01
1.21966779e+00 1.03912801e-01 6.45059824e-01 9.63888049e-01
1.96364537e-01 5.70946872e-01 7.11217999e-01 4.68338653e-02
2.39369795e-01 -4.78361905e-01 8.45942367e-03 5.76065779e-01
-7.13313892e-02 2.43005782e-01 -1.04986453e+00 -6.73549399e-02
8.45198572e-01 1.52376533e-01 4.30056602e-01 -4.93670441e-02
-1.45808613e+00 7.82607734e-01 3.83154720e-01 -6.57071590e-01
-6.55739188e-01 4.93458509e-01 4.02269065e-01 1.97700948e-01
4.44936782e-01 5.18182218e-01 -2.09108025e-01 -7.74564147e-01
-7.68370330e-01 7.88237631e-01 7.54363179e-01 9.05766070e-01
5.25529981e-01 1.76828891e-01 -8.39608237e-02 4.94041204e-01
-9.22152251e-02 4.16752130e-01 7.85544813e-02 -1.22736394e+00
6.86918139e-01 5.66365361e-01 4.88406748e-01 -8.78168225e-01
-1.03715622e+00 -2.64644325e-01 -5.35683572e-01 5.20022094e-01
8.79271328e-01 -8.50306153e-01 -9.01002824e-01 1.78581810e+00
6.12858176e-01 8.84826660e-01 -2.38476284e-02 1.24117947e+00
6.51796937e-01 6.01374388e-01 -1.33999586e-02 2.14483276e-01
1.17290223e+00 -9.49824989e-01 -3.53629678e-01 -6.08539283e-01
7.92403162e-01 -4.62388605e-01 5.41002989e-01 2.16794401e-01
-9.49574292e-01 -2.45710567e-01 -7.14821339e-01 7.02600926e-02
-3.21957290e-01 -1.56739831e-01 6.98819101e-01 5.83450608e-02
-7.18791425e-01 3.77162069e-01 -1.18762410e+00 -1.60571426e-01
1.54668808e-01 6.21527851e-01 -3.49104911e-01 1.23583637e-01
-9.49280143e-01 1.15224874e+00 3.41079012e-02 1.55310363e-01
-1.29437029e+00 -8.95891547e-01 -9.89664555e-01 -2.98001200e-01
7.51033366e-01 -5.32560408e-01 1.30789638e+00 -3.63036811e-01
-1.23291755e+00 2.55653948e-01 1.27257615e-01 -7.48212218e-01
4.91021484e-01 -1.27681261e-02 -1.93750918e-01 -3.22420776e-01
5.22773683e-01 4.43951100e-01 3.89654785e-01 -8.85670304e-01
-1.10395312e+00 -3.67927790e-01 3.01138908e-01 5.92496932e-01
1.91033185e-01 -9.33241323e-02 -2.11816564e-01 -2.76288509e-01
-2.07924873e-01 -1.35086119e+00 -5.49741507e-01 -4.24676716e-01
-4.32489544e-01 -1.54892327e-02 5.65124273e-01 -8.52472246e-01
8.56670797e-01 -1.88381863e+00 5.05068600e-01 1.75260574e-01
3.78172040e-01 -4.35493350e-01 -3.08633566e-01 7.08133519e-01
4.69131112e-01 -6.17375493e-01 1.47503406e-01 -4.89768982e-01
2.36118749e-01 2.85962462e-01 -1.96640454e-02 8.27152431e-01
-1.35643944e-01 8.80016625e-01 -8.05465460e-01 -1.97694376e-01
3.55856299e-01 2.28186369e-01 -4.23651606e-01 -1.18677497e-01
-3.69118094e-01 6.04855597e-01 -1.96377322e-01 5.20778418e-01
1.34226903e-01 -4.50645760e-02 -2.72942870e-03 4.09730047e-01
-4.09963310e-01 1.83381379e-01 -1.44032109e+00 1.90353584e+00
-3.59205395e-01 8.00193012e-01 2.94010043e-01 -7.71845460e-01
5.80365896e-01 1.36020720e-01 1.03840113e+00 -6.85741127e-01
2.72649694e-02 -2.75073290e-01 2.78636068e-01 -5.22739589e-01
1.10541177e+00 1.56741381e-01 -6.15122437e-01 8.00014794e-01
-3.49635601e-01 1.16904005e-01 5.59751630e-01 1.68328822e-01
1.74561775e+00 2.13296688e-03 -2.30731085e-01 1.94824368e-01
-4.63191420e-01 9.11404848e-01 5.93704820e-01 9.68898058e-01
-2.23309115e-01 5.69526136e-01 4.99859661e-01 -7.02410817e-01
-1.25304043e+00 -8.77040088e-01 5.54195940e-01 1.54866982e+00
3.43163550e-01 -2.18665570e-01 -5.24175584e-01 -1.56637788e-01
2.09143773e-01 2.85019547e-01 -6.66888773e-01 -1.63352594e-01
-8.12819660e-01 -1.07199645e+00 6.96360648e-01 6.62424564e-01
1.21949345e-01 -1.10713542e+00 -8.21830750e-01 5.41873157e-01
-3.55897218e-01 -1.05365086e+00 -4.22858357e-01 1.23105593e-01
-1.32831037e-01 -1.17144454e+00 -3.59910905e-01 -4.25436437e-01
2.00656652e-01 3.32774557e-02 1.18251336e+00 2.18198542e-02
-5.27729452e-01 2.47006983e-01 -2.11049438e-01 -6.17011786e-01
-8.27837288e-02 4.59938973e-01 4.12515432e-01 -2.18516022e-01
5.36799431e-01 -4.37147647e-01 -4.97303933e-01 2.48875469e-01
-3.77160877e-01 1.44835711e-01 3.35211962e-01 5.31836569e-01
5.39856315e-01 1.63915068e-01 6.02626562e-01 -5.17497599e-01
5.41916013e-01 -1.08119798e+00 -6.29713714e-01 -2.82898128e-01
-1.40156515e-03 -3.13922286e-01 3.83613080e-01 -6.88864887e-01
-6.32259905e-01 2.34053895e-01 3.09344321e-01 -2.21910492e-01
-4.25683141e-01 6.31143332e-01 4.30489153e-01 1.73755497e-01
8.13232660e-01 -1.31544936e-02 3.91896576e-01 -9.20176134e-02
1.99077070e-01 5.34512103e-01 9.31365609e-01 -5.39525330e-01
5.98839700e-01 6.24605417e-01 2.79345870e-01 -8.28068376e-01
-6.87893033e-01 -4.61165011e-01 -4.77441221e-01 -5.77950120e-01
7.78525770e-01 -1.22846663e+00 -1.41106606e+00 5.26233196e-01
-8.94273698e-01 -1.17900229e+00 -4.46384221e-01 7.30369866e-01
-8.51503611e-01 -1.10751167e-01 -9.01542127e-01 -7.87428260e-01
2.42044762e-01 -9.91044819e-01 1.11857760e+00 2.65349269e-01
-3.10949296e-01 -9.16274011e-01 4.49227631e-01 6.31884992e-01
9.23389047e-02 5.72468877e-01 1.22402146e-01 -4.32795107e-01
-6.19100511e-01 -1.99907899e-01 1.51529521e-01 -7.48044610e-01
-1.14102550e-01 -3.28502178e-01 -4.73033935e-01 -3.60224545e-01
-7.15993941e-01 -2.24261269e-01 5.97451031e-01 9.03351247e-01
7.48511791e-01 -6.64141327e-02 -5.15110850e-01 3.11922342e-01
9.28830385e-01 -6.90137446e-02 4.03531492e-01 7.63212562e-01
8.49796593e-01 8.06630909e-01 8.17146778e-01 8.44060183e-01
1.09990168e+00 9.54673409e-01 6.97625041e-01 -2.26733714e-01
-2.00169474e-01 -9.95809510e-02 3.21228772e-01 9.64781940e-02
-2.37069011e-01 -6.42799377e-01 -1.06120503e+00 1.10104311e+00
-2.47529888e+00 -1.24024093e+00 -5.73562980e-01 1.97333944e+00
3.72301310e-01 5.09179644e-02 8.18801403e-01 -1.97540402e-01
6.52813494e-01 -9.18030273e-03 -7.41130888e-01 -2.29311451e-01
2.34828684e-02 -2.88016140e-01 1.10684681e+00 6.29853249e-01
-1.20645261e+00 7.87350833e-01 5.97712278e+00 5.45899630e-01
-5.36276639e-01 1.87740296e-01 1.97540998e-01 -1.04831612e+00
4.86642897e-01 -1.24660194e-01 -7.82173991e-01 6.11330390e-01
1.34237766e+00 1.52217433e-01 8.18903208e-01 3.20685089e-01
7.50083625e-01 -4.05140996e-01 -1.02177918e+00 6.42133355e-01
-1.34551242e-01 -1.55821407e+00 -8.78187060e-01 4.31900203e-01
7.75725126e-01 5.51856399e-01 6.71512261e-02 4.70901519e-01
1.19663453e+00 -1.29292119e+00 7.70335793e-01 7.80717373e-01
5.02003312e-01 -1.13472688e+00 7.34972715e-01 8.01822484e-01
-1.19181609e+00 -2.78842866e-01 -1.33537859e-01 -9.29119229e-01
6.02480412e-01 -1.20580327e-02 -8.85045290e-01 1.37835234e-01
8.24860573e-01 7.26537883e-01 -2.52236754e-01 1.10100007e+00
3.46514374e-01 3.53121012e-01 -8.41988206e-01 9.12346169e-02
4.81441081e-01 -1.86319128e-01 7.45309770e-01 6.97994232e-01
1.98440313e-01 1.56234086e-01 8.07288408e-01 3.96432251e-01
1.05749145e-01 -5.37413120e-01 -7.57543743e-01 4.34727848e-01
6.80537999e-01 1.08166385e+00 -6.66263402e-01 -9.38806981e-02
-3.65293652e-01 6.36943400e-01 5.34756005e-01 2.55396932e-01
-1.05484605e+00 -1.89744532e-01 1.23406255e+00 3.98566186e-01
1.53677300e-01 -4.92638022e-01 -2.74913728e-01 -5.75815856e-01
-1.29454866e-01 -6.95721686e-01 4.66675609e-01 -6.52238548e-01
-1.24199080e+00 1.67541653e-01 2.39112210e-02 -1.28760386e+00
-8.81939054e-01 -2.56413817e-01 -6.96328878e-01 6.68274164e-01
-1.08301628e+00 -1.40783799e+00 -3.56541365e-01 4.59851593e-01
4.95447367e-01 -3.81033748e-01 4.02305305e-01 3.17612886e-01
-7.83427775e-01 1.25525653e-01 2.66158491e-01 3.19116801e-01
4.55471158e-01 -1.23402596e+00 5.76929867e-01 8.38467240e-01
2.57626295e-01 -9.97930467e-02 9.70079541e-01 -9.68618453e-01
-1.55082679e+00 -1.29550123e+00 2.72291154e-01 -1.05441630e+00
8.48713875e-01 -2.11765513e-01 -4.96614009e-01 1.11074042e+00
1.05115939e-02 4.62505361e-03 4.64665830e-01 2.90549576e-01
3.48436266e-01 2.13935509e-01 -7.82548070e-01 6.97627842e-01
9.80697930e-01 -2.44699627e-01 -1.42289966e-01 6.04091465e-01
2.71340847e-01 -9.48400259e-01 -9.73382890e-01 7.74006173e-02
4.29396033e-01 -6.27753198e-01 1.11078560e+00 -7.34578252e-01
3.16877007e-01 -6.38028681e-01 1.40666649e-01 -1.66064465e+00
-3.48185807e-01 -5.32623887e-01 -1.98972583e-01 8.55982065e-01
4.89435136e-01 -2.68428236e-01 1.21507382e+00 8.41665387e-01
-3.70597720e-01 -1.84732422e-01 -1.06527722e+00 -6.10176384e-01
-2.94054411e-02 -4.72791493e-01 8.66162658e-01 7.52882302e-01
6.92206025e-02 2.39551082e-01 -8.95567179e-01 6.90906763e-01
7.14992464e-01 -3.23991701e-02 1.21611440e+00 -1.14347041e+00
-4.90001798e-01 -1.36450186e-01 -5.72025478e-01 -8.74731898e-01
3.93869519e-01 -4.54984069e-01 3.55619967e-01 -1.62555337e+00
2.01242074e-01 -7.49267280e-01 1.51457489e-01 4.78905290e-01
2.12782845e-01 6.62620187e-01 -1.95604749e-02 2.04255693e-02
-9.62428510e-01 2.55026132e-01 8.61734748e-01 -2.17483237e-01
-3.90673786e-01 1.12035675e-02 -2.01580361e-01 7.29380250e-01
9.15707469e-01 -6.14000797e-01 -2.59704620e-01 -9.18364346e-01
2.56206781e-01 4.39139932e-01 7.94058979e-01 -1.15882874e+00
7.44844079e-01 -6.44912183e-01 2.70766288e-01 -5.82395971e-01
1.01331115e+00 -6.37528479e-01 4.82409120e-01 2.47971237e-01
-2.16538414e-01 3.86450231e-01 2.67426550e-01 8.73834729e-01
1.98470373e-02 -1.47075519e-01 2.57739946e-02 -1.99106082e-01
-1.07564306e+00 4.24369931e-01 -7.64711618e-01 1.38613939e-01
1.38701487e+00 -2.45173991e-01 -6.58328295e-01 -6.95348203e-01
-9.07128751e-01 7.53917217e-01 4.17442322e-01 3.46293151e-01
2.78853238e-01 -9.99737501e-01 -1.14510810e+00 -1.11634061e-01
-9.01052579e-02 2.55766094e-01 6.48129344e-01 1.05708706e+00
-5.89862108e-01 -1.38732478e-01 -2.59021729e-01 -6.57893598e-01
-1.23515272e+00 2.02982768e-01 2.43108973e-01 -1.06199935e-01
-8.03444266e-01 5.96113741e-01 -3.35096151e-01 -6.34766519e-01
-2.94856727e-02 3.77421593e-03 -2.41887584e-01 -2.99766976e-02
5.17307639e-01 7.93936908e-01 -7.69124255e-02 -9.58245993e-01
-1.94652811e-01 1.17487334e-01 1.19652152e-01 -2.14778215e-01
1.66453302e+00 -2.22672313e-01 2.08522990e-01 4.74269748e-01
3.28498453e-01 -1.70151949e-01 -1.90296197e+00 -2.96666403e-03
-1.15001328e-01 -3.77381802e-01 1.11267254e-01 -5.10661662e-01
-9.03815150e-01 3.67166966e-01 2.53594697e-01 4.25704420e-01
3.87260079e-01 1.63432866e-01 8.91991258e-01 1.66242585e-01
5.73743045e-01 -1.19612741e+00 -1.83772162e-01 6.31217599e-01
4.64181930e-01 -1.35586774e+00 7.87372608e-03 9.71025750e-02
-8.96907449e-01 5.11275887e-01 1.01846743e+00 -3.07019353e-01
1.90999255e-01 8.61334980e-01 -4.23550121e-02 -4.58309770e-01
-1.05199349e+00 -3.33019078e-01 -1.86039865e-01 9.37241375e-01
-1.85536623e-01 3.29151988e-01 5.52136004e-01 4.42570537e-01
-6.29663646e-01 -3.11876357e-01 1.09694684e+00 9.75728750e-01
-2.74153233e-01 -5.11624038e-01 -5.59115887e-01 7.38143504e-01
-9.07312930e-02 2.62370497e-01 -1.10808067e-01 8.23559046e-01
1.49596468e-01 1.19062281e+00 7.42658556e-01 -3.39807212e-01
5.15461683e-01 -5.07099092e-01 2.34512225e-01 -6.69709980e-01
-6.68697059e-01 -3.74468148e-01 4.40642387e-01 -4.40854430e-01
-2.27271661e-01 -9.97366965e-01 -1.44824219e+00 -1.13343811e+00
-1.48612112e-01 -4.32215184e-02 4.60853577e-01 1.04757285e+00
5.20600438e-01 9.44901764e-01 1.20946318e-01 -1.18567574e+00
-6.99728951e-02 -1.08387291e+00 -6.38349056e-01 2.82846153e-01
4.83011454e-01 -1.05929041e+00 2.08900049e-02 -4.50848266e-02]
|
[5.788835525512695, 0.6786837577819824]
|
286a280b-4f71-4921-94c3-a0cc5e22768b
|
adversarial-continual-learning-for-multi
|
2107.08751
| null |
https://arxiv.org/abs/2107.08751v4
|
https://arxiv.org/pdf/2107.08751v4.pdf
|
Adversarial Continual Learning for Multi-Domain Hippocampal Segmentation
|
Deep learning for medical imaging suffers from temporal and privacy-related restrictions on data availability. To still obtain viable models, continual learning aims to train in sequential order, as and when data is available. The main challenge that continual learning methods face is to prevent catastrophic forgetting, i.e., a decrease in performance on the data encountered earlier. This issue makes continuous training of segmentation models for medical applications extremely difficult. Yet, often, data from at least two different domains is available which we can exploit to train the model in a way that it disregards domain-specific information. We propose an architecture that leverages the simultaneous availability of two or more datasets to learn a disentanglement between the content and domain in an adversarial fashion. The domain-invariant content representation then lays the base for continual semantic segmentation. Our approach takes inspiration from domain adaptation and combines it with continual learning for hippocampal segmentation in brain MRI. We showcase that our method reduces catastrophic forgetting and outperforms state-of-the-art continual learning methods.
|
['Anirban Mukhopadhyay', 'Camila Gonzalez', 'Marius Memmel']
|
2021-07-19
| null | null | null | null |
['continual-semantic-segmentation']
|
['computer-vision']
|
[ 5.37909269e-01 1.96137324e-01 -3.03060144e-01 -4.62965548e-01
-8.43101025e-01 -6.09797299e-01 3.97712201e-01 4.98889416e-01
-9.64709699e-01 9.51606929e-01 8.16895738e-02 -2.98568666e-01
-1.31342128e-01 -6.19649470e-01 -9.95534301e-01 -7.25589693e-01
-4.97122929e-02 5.95089674e-01 3.44178736e-01 -4.04526219e-02
-1.09271720e-01 4.90334123e-01 -1.01021159e+00 4.66353655e-01
8.09000492e-01 8.86449695e-01 1.19116172e-01 2.01466903e-01
-1.54732512e-02 7.50747263e-01 -3.83434325e-01 -2.99688220e-01
4.52568680e-01 -2.93525994e-01 -1.03661096e+00 1.85632035e-01
3.44101250e-01 -5.45628548e-01 -4.67086494e-01 9.84358370e-01
3.77029270e-01 5.65749779e-03 4.80701596e-01 -1.02038395e+00
-6.58521652e-01 4.20629382e-01 -6.00574315e-01 5.13363659e-01
-4.26829942e-02 2.61855036e-01 5.64535260e-01 -5.42368114e-01
8.71747792e-01 6.32753193e-01 8.03983748e-01 9.30588186e-01
-1.70897782e+00 -7.09880531e-01 2.55152285e-01 8.24227408e-02
-1.05322754e+00 -6.17733240e-01 6.93247855e-01 -4.09236491e-01
4.81831193e-01 1.38651520e-01 7.63521910e-01 1.47670269e+00
5.18504918e-01 9.69488323e-01 1.40617287e+00 -3.25179994e-01
6.13521814e-01 2.45817289e-01 1.62171334e-01 5.06814957e-01
1.57256976e-01 2.79776961e-01 -7.48148918e-01 -3.25915158e-01
5.88498473e-01 2.82901913e-01 -1.63039491e-01 -7.95057237e-01
-9.02489126e-01 7.37045228e-01 3.83920580e-01 3.57859701e-01
-3.71609777e-01 -1.05998330e-01 5.52592099e-01 7.85405099e-01
4.40435797e-01 5.04755795e-01 -6.60531998e-01 4.06673044e-01
-1.35177338e+00 1.74559414e-01 4.78480250e-01 6.01700068e-01
5.92452526e-01 -2.08030537e-01 -1.33932784e-01 5.36062598e-01
-2.68789262e-01 2.02878311e-01 7.84864962e-01 -7.00469971e-01
3.93850356e-02 1.07306935e-01 -8.82464051e-02 -6.41403913e-01
-3.88168037e-01 -4.64070499e-01 -8.27421844e-01 3.59615237e-01
5.31966805e-01 -1.47290617e-01 -1.21826267e+00 2.02961278e+00
3.31188738e-01 3.58480692e-01 -1.25572652e-01 6.17950737e-01
3.16628188e-01 1.36727812e-02 4.87520903e-01 -3.52154464e-01
1.16486657e+00 -6.35387361e-01 -7.00034678e-01 -5.21990001e-01
4.24844265e-01 -3.57308954e-01 9.92643058e-01 4.43500966e-01
-9.66729999e-01 -5.04380949e-02 -1.04844713e+00 -7.48192370e-02
-2.74190724e-01 -5.76574564e-01 6.18756652e-01 7.13579059e-01
-1.12942886e+00 8.52904022e-01 -1.33909345e+00 -1.02995679e-01
1.05488241e+00 5.57049692e-01 -5.34614503e-01 -2.55341023e-01
-1.19691932e+00 9.20050621e-01 3.85182649e-01 -3.54767561e-01
-1.06825924e+00 -1.15681243e+00 -6.35816991e-01 -1.61396757e-01
5.49144030e-01 -8.79502594e-01 1.15579581e+00 -1.29422379e+00
-9.37768996e-01 1.14343107e+00 9.47498977e-02 -1.16873431e+00
1.09004056e+00 -1.94248021e-01 -1.99632585e-01 2.29712591e-01
1.87276050e-01 6.78415477e-01 1.36927080e+00 -1.01514184e+00
-3.64324629e-01 -5.92228830e-01 -1.50137082e-01 1.16945043e-01
-5.95634520e-01 -5.41087687e-01 -1.87044710e-01 -7.56133020e-01
1.13486741e-02 -8.84128749e-01 -4.02149111e-01 5.02497613e-01
-2.50197202e-01 3.45273793e-01 8.22385311e-01 -8.21105123e-01
8.39207292e-01 -2.38950825e+00 8.74340683e-02 1.02739520e-01
5.69684386e-01 2.39980340e-01 -1.04702614e-01 -1.07777670e-01
-1.99814156e-01 -1.14770062e-01 -6.34500921e-01 -6.25795007e-01
-5.80887258e-01 3.26888829e-01 -2.71241784e-01 7.35619485e-01
4.32651378e-02 8.99502039e-01 -1.00418961e+00 -5.09488642e-01
-8.71428922e-02 3.64315718e-01 -8.92619729e-01 -5.01162489e-04
-2.82488614e-01 8.67593527e-01 -3.32591921e-01 4.70349997e-01
8.58823538e-01 -2.42659390e-01 1.58784822e-01 7.69833401e-02
1.50368825e-01 -7.21067190e-02 -5.04058063e-01 2.24064064e+00
-2.57689893e-01 3.68863434e-01 -4.08214331e-02 -1.32919157e+00
4.24683034e-01 3.51431847e-01 9.17887151e-01 -8.85988176e-01
-1.16189912e-01 4.07161295e-01 -2.62217075e-01 -2.80264497e-01
3.47349308e-02 -7.46259987e-01 -2.16444090e-01 4.62986678e-01
2.05059811e-01 1.12813763e-01 -3.77955794e-01 2.67775416e-01
1.47641718e+00 -1.30806133e-01 3.38009864e-01 -3.37262243e-01
4.65762615e-02 7.38747120e-02 8.50302756e-01 1.07029545e+00
-5.55666268e-01 5.62479556e-01 5.15586734e-01 -6.83016300e-01
-1.01609147e+00 -1.18825030e+00 -2.45675787e-01 8.91451359e-01
-1.95161581e-01 2.17047751e-01 -6.80132329e-01 -1.08521819e+00
1.99544713e-01 8.77458632e-01 -1.01192462e+00 -7.23690450e-01
-5.95474839e-01 -8.56263638e-01 3.19702774e-01 2.27246374e-01
3.56134325e-01 -8.41687918e-01 -8.88028383e-01 2.74917990e-01
-4.71748337e-02 -7.77658582e-01 -6.11306965e-01 4.80101109e-01
-1.19909275e+00 -9.68718469e-01 -8.44768882e-01 -5.59331238e-01
6.04119718e-01 2.02719152e-01 1.19156158e+00 -1.34278893e-01
-5.42878509e-01 6.38053834e-01 -7.62620419e-02 -4.69615877e-01
-4.38921273e-01 3.63336593e-01 -5.01843961e-03 6.24604784e-02
1.87539563e-01 -9.98898447e-01 -8.65375698e-01 -2.19296515e-01
-1.24337900e+00 -2.26011485e-01 5.31080425e-01 1.22875750e+00
8.17950249e-01 -7.29842335e-02 8.86912763e-01 -1.62283397e+00
3.63093823e-01 -7.60527968e-01 -4.55885410e-01 2.52585709e-01
-8.30713928e-01 1.58363611e-01 5.46070695e-01 -5.51697612e-01
-8.64454687e-01 4.71128315e-01 -1.99034691e-01 -6.46341741e-01
2.61647254e-02 3.48867893e-01 9.89211351e-02 -7.57956654e-02
8.87140334e-01 3.93087208e-01 3.51437151e-01 -4.29151833e-01
4.17054564e-01 1.29823193e-01 6.73841059e-01 -3.00046384e-01
6.18292749e-01 8.37785959e-01 -1.41822964e-01 -6.19005799e-01
-1.08940554e+00 -2.54687697e-01 -8.59895945e-01 5.11831194e-02
8.13262939e-01 -9.23613489e-01 -3.32606047e-01 5.64876914e-01
-9.25598443e-01 -3.77270013e-01 -8.94051373e-01 1.18020378e-01
-6.25723481e-01 2.62259662e-01 -3.67905200e-01 -3.59103709e-01
-3.79250914e-01 -8.75599861e-01 6.30242944e-01 -1.68503940e-01
-2.80603439e-01 -1.02888978e+00 -5.46180643e-02 1.70203447e-01
4.71500009e-01 3.84005308e-01 1.07029390e+00 -9.43363786e-01
-5.65205216e-01 -1.78673819e-01 1.90493882e-01 2.34565839e-01
1.42263710e-01 -9.17251229e-01 -9.23354328e-01 -8.02357018e-01
4.97904003e-01 -5.49951017e-01 1.21315122e+00 4.11052227e-01
1.39310157e+00 -4.44096118e-01 -4.02874768e-01 6.94214106e-01
1.50176120e+00 5.43632470e-02 6.17339849e-01 4.27184284e-01
4.23245609e-01 4.67156827e-01 3.56016010e-01 3.45377654e-01
1.41269207e-01 2.10267007e-01 3.23669195e-01 -2.22325459e-01
-1.62072629e-01 -1.96822271e-01 -2.03040123e-01 3.05818439e-01
5.57654083e-01 2.44978994e-01 -1.07420766e+00 7.69311011e-01
-1.66863298e+00 -8.12250316e-01 5.27240634e-01 2.42511415e+00
1.29978776e+00 1.51635051e-01 5.13121188e-02 -2.89153550e-02
4.19452727e-01 6.67256042e-02 -1.13549852e+00 -5.79834059e-02
4.36101072e-02 3.24556142e-01 8.11004758e-01 3.34085435e-01
-1.33770239e+00 8.87048900e-01 6.23587608e+00 7.78964996e-01
-1.32606483e+00 6.81113303e-01 9.76099491e-01 -5.17965138e-01
-4.60752904e-01 -1.89282149e-01 -3.14189821e-01 3.94865453e-01
1.01396215e+00 -2.44001970e-01 5.00880301e-01 8.01309705e-01
-1.20219819e-01 -8.23187754e-02 -1.24610090e+00 7.93130875e-01
-8.45463946e-02 -1.50931919e+00 -1.39396220e-01 1.37875536e-02
7.35662103e-01 2.12518677e-01 5.06784499e-01 2.52505720e-01
3.78041357e-01 -8.72626185e-01 6.24294698e-01 4.13538605e-01
9.54642475e-01 -6.73469067e-01 3.15550417e-01 6.07445776e-01
-4.58460808e-01 -3.24450821e-01 -2.46453002e-01 5.14865518e-01
-2.75388104e-03 8.56731236e-01 -9.52850997e-01 3.59592348e-01
5.61431170e-01 5.36760807e-01 -4.26465601e-01 1.02026880e+00
4.53795269e-02 5.34149468e-01 -2.05038562e-01 5.65256000e-01
1.21846505e-01 2.68377155e-01 5.90247393e-01 9.12245274e-01
1.62186876e-01 3.17378119e-02 1.51755750e-01 6.06869578e-01
-2.74134189e-01 9.94341820e-02 -8.29902411e-01 2.21426353e-01
5.03193557e-01 5.77938020e-01 -7.01289594e-01 -1.81255966e-01
-3.08219314e-01 1.40012729e+00 6.51579380e-01 6.13016728e-03
-5.42402744e-01 2.88304120e-01 4.82819736e-01 4.50350195e-01
2.47594714e-01 -1.35883167e-01 -6.37708247e-01 -1.19979846e+00
-5.23795784e-02 -9.93238032e-01 9.26775932e-01 -1.17379382e-01
-1.48631191e+00 5.94973624e-01 -2.09790498e-01 -8.91303480e-01
1.82106998e-02 -1.24562427e-01 -2.57483453e-01 7.24691272e-01
-1.73520672e+00 -1.13592362e+00 2.15181783e-01 9.88111019e-01
4.28539574e-01 -1.56589225e-01 1.00486052e+00 4.60137665e-01
-2.43063316e-01 9.13139284e-01 2.87634730e-01 6.00360287e-03
9.25216317e-01 -1.17719877e+00 2.32912034e-01 7.44991124e-01
1.29438698e-01 5.79750121e-01 7.04972684e-01 -8.53626490e-01
-1.10005748e+00 -1.11375141e+00 7.79377043e-01 -3.54662955e-01
5.75345159e-01 -3.54748100e-01 -1.23271799e+00 8.75753701e-01
-1.56896055e-01 4.84925777e-01 7.97489166e-01 1.21735282e-01
-4.13739294e-01 -2.72639424e-01 -1.52517653e+00 4.65107381e-01
9.56896842e-01 -6.14110768e-01 -6.84106171e-01 4.58728522e-01
8.27421129e-01 -3.46497267e-01 -5.40646374e-01 1.02972165e-01
3.98523211e-01 -8.87870908e-01 1.04097819e+00 -9.06495273e-01
3.28672856e-01 2.76257366e-01 1.42104387e-01 -1.36197519e+00
-5.87720312e-02 -7.34782934e-01 -1.74604565e-01 7.01828599e-01
3.14010829e-01 -7.58883595e-01 9.48322952e-01 9.86841202e-01
1.84444338e-01 -6.47028029e-01 -1.50801694e+00 -8.67563665e-01
6.13387287e-01 -2.30558708e-01 5.14694870e-01 1.21256363e+00
-3.04891616e-01 1.24695957e-01 -5.96882939e-01 1.37674958e-01
9.57754314e-01 3.36429626e-02 2.47974336e-01 -1.18167639e+00
-3.65942746e-01 7.20530674e-02 -2.07216308e-01 -8.00271034e-01
1.83379814e-01 -1.13432252e+00 -1.32090777e-01 -1.02017879e+00
2.10963577e-01 -7.04713702e-01 -8.35185349e-01 7.73779988e-01
-1.09046929e-01 1.25581011e-01 7.17516989e-02 4.29385006e-01
-5.57786405e-01 5.16540229e-01 1.28196740e+00 -2.21186042e-01
-3.32943678e-01 3.17730755e-02 -9.08101499e-01 4.78949696e-01
8.22549641e-01 -9.78905141e-01 -6.26897633e-01 -5.35434127e-01
-1.09632462e-01 1.95046037e-01 4.23776537e-01 -9.66901302e-01
5.35312414e-01 1.49920993e-02 3.60150278e-01 -2.78266460e-01
2.19273373e-01 -9.73039329e-01 3.09138447e-02 7.70103872e-01
-5.98945200e-01 -3.54471564e-01 2.85906613e-01 1.03054428e+00
1.01407461e-01 -5.83982058e-02 1.10767591e+00 -5.45231938e-01
-5.03210604e-01 7.21614957e-01 -2.75887638e-01 2.94802725e-01
1.05554259e+00 1.06270082e-01 1.33322537e-01 -9.68998671e-02
-1.26518798e+00 1.33928925e-01 4.08919781e-01 1.73958555e-01
5.73679149e-01 -1.03981805e+00 -6.98340356e-01 3.78850400e-01
-3.04793194e-02 -1.62429698e-02 6.41077876e-01 9.02365208e-01
-6.24592006e-02 3.20896916e-02 -4.88680899e-01 -4.08944458e-01
-1.05349219e+00 8.91112447e-01 3.68301541e-01 -6.04959548e-01
-8.65556180e-01 9.91952002e-01 2.61347711e-01 -2.79704005e-01
1.43627554e-01 2.13004962e-01 1.56887218e-01 3.14299613e-01
5.71330070e-01 -1.27429381e-01 3.33729655e-01 -3.82711156e-03
-3.75703037e-01 -2.54858196e-01 -8.39621663e-01 -2.59838760e-01
1.74273109e+00 -2.38430306e-01 1.10875908e-02 4.20881689e-01
1.15308201e+00 -2.74791181e-01 -1.72160101e+00 -6.68028235e-01
8.42682943e-02 -5.98693252e-01 2.59294778e-01 -1.02876413e+00
-1.14556050e+00 7.56264925e-01 9.36218858e-01 -1.05021790e-01
1.17868209e+00 -1.08551368e-01 1.11822712e+00 2.54796684e-01
6.37016535e-01 -1.19494247e+00 1.44770250e-01 1.23140715e-01
7.47619390e-01 -1.31773627e+00 3.21515985e-02 1.34162689e-02
-5.48473358e-01 8.40784729e-01 1.65686250e-01 -8.38101730e-02
9.28138435e-01 5.08179009e-01 1.08121829e-02 -1.79075763e-01
-6.04374945e-01 2.05393344e-01 1.58689935e-02 7.25816965e-01
-6.38846532e-02 7.68163614e-03 -2.13561103e-01 6.29582763e-01
1.15041323e-01 3.64599466e-01 3.63694787e-01 1.23737788e+00
-1.72023684e-01 -1.24446571e+00 -2.05548286e-01 6.18263841e-01
-7.79948950e-01 -2.86998332e-01 -5.51274754e-02 5.08503616e-01
1.70726970e-01 4.16118175e-01 -1.15198165e-01 -9.11763236e-02
2.69755244e-01 2.76453257e-01 4.40430641e-01 -6.47079110e-01
-4.56710875e-01 -1.84845895e-01 -2.74334252e-01 -5.64172089e-01
-7.57411867e-02 -1.12368548e+00 -9.84099329e-01 -2.67585456e-01
1.52352139e-01 -1.78921878e-01 3.35039705e-01 9.25140858e-01
3.73433143e-01 4.78563666e-01 5.55491090e-01 -3.24175000e-01
-8.38247180e-01 -2.48277515e-01 -6.17087841e-01 5.28903008e-01
9.11171913e-01 -5.34741938e-01 -1.46506891e-01 2.09058121e-01]
|
[14.607062339782715, -1.8672319650650024]
|
eaefbb05-adc0-408c-a59b-53093432bfbe
|
insights-from-insurance-for-fair-machine
|
2306.14624
| null |
https://arxiv.org/abs/2306.14624v1
|
https://arxiv.org/pdf/2306.14624v1.pdf
|
Insights From Insurance for Fair Machine Learning: Responsibility, Performativity and Aggregates
|
We argue that insurance can act as an analogon for the social situatedness of machine learning systems, hence allowing machine learning scholars to take insights from the rich and interdisciplinary insurance literature. Tracing the interaction of uncertainty, fairness and responsibility in insurance provides a fresh perspective on fairness in machine learning. We link insurance fairness conceptions to their machine learning relatives, and use this bridge to problematize fairness as calibration. In this process, we bring to the forefront three themes that have been largely overlooked in the machine learning literature: responsibility, performativity and tensions between aggregate and individual.
|
['Robert C. Williamson', 'Christian Fröhlich']
|
2023-06-26
| null | null | null | null |
['fairness', 'fairness']
|
['computer-vision', 'miscellaneous']
|
[ 1.08007513e-01 7.73506820e-01 -1.02658403e+00 -6.18117392e-01
-4.41367328e-01 -4.92556542e-01 4.75907266e-01 5.52914202e-01
-4.40640360e-01 6.03707016e-01 9.49068785e-01 -8.96583736e-01
-4.31418240e-01 -5.73126078e-01 -2.72671878e-01 -2.66927660e-01
4.10830528e-01 6.19538017e-02 -7.52504349e-01 -2.74081618e-01
3.64846945e-01 7.85012543e-02 -1.26707661e+00 1.86258614e-01
1.25103772e+00 6.18995488e-01 -8.25344622e-01 1.27531931e-01
-1.75353974e-01 1.71706438e+00 -1.88012376e-01 -1.10299838e+00
2.57154796e-02 -4.95024025e-01 -9.49744225e-01 -1.54383272e-01
3.60519916e-01 -3.18833411e-01 -2.97495037e-01 9.06100869e-01
1.76831543e-01 -3.26989472e-01 8.19126785e-01 -1.26186061e+00
-1.12617433e+00 9.24863875e-01 -1.39115438e-01 5.71979471e-02
1.60118520e-01 7.33588785e-02 1.39015138e+00 -3.19752516e-03
4.12620395e-01 1.40986693e+00 9.52534676e-01 6.75778925e-01
-1.22064221e+00 -4.53070819e-01 1.79861769e-01 -2.63452213e-02
-6.44978702e-01 -6.57566965e-01 5.48972905e-01 -9.62425172e-01
4.45864946e-01 5.18986285e-01 9.14324462e-01 8.06330085e-01
1.37580484e-01 6.66927516e-01 1.18917179e+00 -5.41602850e-01
5.97018972e-02 2.34377265e-01 3.02114040e-01 4.85372305e-01
6.46552563e-01 5.58923304e-01 -3.20267379e-01 -4.58024532e-01
5.83116710e-01 4.87606794e-01 -6.24969862e-02 1.16442181e-02
-7.97172368e-01 1.36736035e+00 4.58671510e-01 1.07973404e-01
-4.33830947e-01 2.47536793e-01 3.22786361e-01 5.95427811e-01
6.38686180e-01 3.31912428e-01 -2.40791917e-01 -2.32153088e-01
-3.63703489e-01 3.70893717e-01 8.60496998e-01 1.67887852e-01
4.50212121e-01 -2.04955801e-01 -2.12775916e-01 5.86217701e-01
6.79030716e-01 2.66163558e-01 -1.13028221e-01 -1.53725803e+00
2.60503322e-01 9.42472816e-01 3.57601613e-01 -8.72440755e-01
-1.72502980e-01 -3.03421259e-01 -3.64285529e-01 6.63750410e-01
8.12914312e-01 -3.62169027e-01 -1.34213969e-01 1.79866326e+00
1.50106177e-01 -6.00780070e-01 2.44658753e-01 9.19218004e-01
9.92117524e-02 -1.83930337e-01 5.48142612e-01 -1.21850230e-01
1.11076701e+00 -6.14289403e-01 -6.26626015e-01 -4.02580261e-01
7.96357870e-01 -5.04680574e-01 1.26352847e+00 1.94777027e-01
-1.19528973e+00 1.95507169e-01 -6.29571021e-01 -4.58922565e-01
-1.33883417e-01 -6.80145502e-01 1.14173210e+00 1.06364691e+00
-6.51635826e-01 1.02149856e+00 -4.27121460e-01 -2.12332770e-01
1.07507861e+00 -2.90427566e-01 1.36846498e-01 1.17222197e-01
-1.22090030e+00 1.21919692e+00 -1.85190529e-01 8.32457319e-02
-1.67233944e-01 -8.29608977e-01 -6.89352334e-01 -1.02739513e-01
4.56697226e-01 -9.08480883e-01 1.33005917e+00 -1.60473561e+00
-8.38152349e-01 1.12062192e+00 1.31194741e-01 -5.39857388e-01
9.42745686e-01 -2.86598891e-01 -3.83563936e-01 -4.87652719e-01
-1.02092111e-02 -4.44274545e-02 1.06477909e-01 -1.14516652e+00
-7.82789350e-01 -5.41792333e-01 3.13934892e-01 1.16013102e-01
-1.24333583e-01 4.09325212e-01 8.82689774e-01 -5.98748088e-01
-8.27327520e-02 -6.37311935e-01 -1.77113891e-01 9.14102420e-02
-3.20330560e-02 -3.68260622e-01 -2.49196533e-02 -4.35629785e-01
1.52396452e+00 -2.12607503e+00 -2.07814649e-01 1.50310427e-01
5.95922530e-01 -1.80761546e-01 4.82787758e-01 5.55581152e-01
-1.14508025e-01 4.76204336e-01 -2.54596472e-01 -1.16370462e-01
5.54359674e-01 4.12502438e-01 -2.41197020e-01 6.49921954e-01
-1.31188959e-01 1.16947842e+00 -8.50964069e-01 -4.50048476e-01
-8.83369241e-03 3.24404687e-01 -6.61432564e-01 -1.09519258e-01
1.11391306e-01 1.12786889e-01 -3.76680583e-01 8.79168332e-01
2.93630153e-01 -7.49971122e-02 3.85231286e-01 2.26256430e-01
-3.37487042e-01 5.20554245e-01 -5.56356609e-01 1.06859028e+00
1.91884488e-02 4.99416530e-01 1.48126617e-01 -9.22883689e-01
6.00676358e-01 4.71924156e-01 2.81726897e-01 -5.45742691e-01
2.56701916e-01 1.40184313e-01 4.47902411e-01 -6.25245988e-01
7.42877424e-02 -8.24453473e-01 -8.71550813e-02 9.87052321e-01
-5.32936931e-01 -6.88618869e-02 -4.54970092e-01 7.24544153e-02
5.35830081e-01 -1.73083857e-01 6.45707011e-01 -6.30225301e-01
6.32043928e-02 -2.63531413e-03 8.36346865e-01 5.88379443e-01
-7.26082325e-01 1.14379078e-01 8.96108270e-01 -8.65497351e-01
-8.71756673e-01 -1.14288151e+00 -4.52378571e-01 1.25725114e+00
-2.51729906e-01 1.72941014e-01 -7.98298657e-01 -7.22551107e-01
7.72979975e-01 7.22704470e-01 -9.62061942e-01 -1.65089637e-01
-1.03793822e-01 -6.30437016e-01 5.09987652e-01 2.95346349e-01
-1.40262172e-01 -8.12799811e-01 -9.20048177e-01 -2.17743278e-01
-1.42103121e-01 -4.68466371e-01 -1.01714805e-01 -1.22787468e-01
-8.31321478e-01 -1.39514351e+00 -1.03119321e-01 -1.78736255e-01
1.37407452e-01 -2.32774261e-02 1.50033712e+00 7.34133661e-01
-6.17003515e-02 5.70218027e-01 -1.43467531e-01 -1.10602832e+00
-8.32873762e-01 -3.30978781e-01 -8.60377029e-02 -2.51544535e-01
7.96384394e-01 -6.21272564e-01 -5.79059422e-01 -5.52473068e-02
-8.70695233e-01 -1.15885474e-01 1.61907688e-01 6.15808964e-01
-1.25932887e-01 -4.95498061e-01 1.08169067e+00 -1.45199513e+00
8.75303388e-01 -1.01383436e+00 -2.86300212e-01 2.85751581e-01
-1.35675848e+00 -1.48844898e-01 5.79752922e-02 2.36547943e-02
-1.07628548e+00 -8.00510824e-01 2.70754546e-01 2.64781266e-01
1.14681959e-01 6.50478244e-01 -5.46556041e-02 2.82516833e-02
1.02378368e+00 -6.89197719e-01 6.69689775e-01 -3.86691093e-01
6.32112026e-01 7.39474118e-01 2.34716371e-01 -6.11729980e-01
3.80615145e-01 5.56698322e-01 -3.65340829e-01 -1.49996415e-01
-1.17417526e+00 1.81372285e-01 -3.56835634e-01 -3.02722096e-01
8.00878584e-01 -8.33001196e-01 -1.13561702e+00 4.52422537e-02
-6.52838469e-01 -3.73704106e-01 -6.80702507e-01 6.35466754e-01
-5.78208447e-01 5.87102398e-02 -4.89790499e-01 -1.34835112e+00
-9.19594839e-02 -5.87517083e-01 1.57893255e-01 2.85056263e-01
-8.85864258e-01 -1.40474772e+00 4.41696756e-02 8.50381494e-01
5.42739093e-01 7.73753941e-01 1.15564513e+00 -6.39054775e-01
-3.61901164e-01 -2.59532422e-01 -9.78905894e-03 1.94011182e-01
3.32092226e-01 -1.21196762e-01 -1.19196868e+00 3.03187460e-01
3.49007577e-01 -5.60030282e-01 8.19178462e-01 3.05494606e-01
6.63096666e-01 -7.84135222e-01 1.49185181e-01 3.11873376e-01
1.53600049e+00 3.45807130e-05 4.77125049e-01 3.17330569e-01
4.93352443e-01 1.30646563e+00 5.24339855e-01 6.03404939e-01
1.03769743e+00 8.46846104e-02 4.45813209e-01 -2.56518692e-01
4.74847615e-01 -3.29649478e-01 -1.62354987e-02 5.27784489e-02
-3.06863397e-01 5.47068775e-01 -1.21877027e+00 3.63219649e-01
-2.08849669e+00 -1.34038973e+00 -2.23254263e-01 2.17737103e+00
1.05259001e+00 1.82039887e-01 4.33360577e-01 1.31102160e-01
3.04824442e-01 3.00981164e-01 -4.76189315e-01 -9.34574306e-01
6.35337532e-02 -3.31439644e-01 4.62197810e-01 9.70126688e-01
-8.77706945e-01 6.80357993e-01 7.55785513e+00 8.04378092e-02
-5.32436252e-01 1.43016756e-01 1.03172231e+00 -2.58129656e-01
-1.21803737e+00 9.81439054e-02 1.58801123e-01 3.55415016e-01
5.59960365e-01 -4.29756552e-01 6.60544455e-01 6.64207637e-01
3.86651635e-01 -2.95915246e-01 -1.33135784e+00 4.39367861e-01
-3.24013919e-01 -1.11241412e+00 -4.59655017e-01 3.25731754e-01
8.26837063e-01 -7.63301179e-02 2.28844374e-01 -1.07254572e-01
7.82530248e-01 -1.62569964e+00 9.58497643e-01 8.09909523e-01
4.12805587e-01 -5.06557763e-01 7.01946497e-01 1.59562722e-01
-2.16588706e-01 -6.08763635e-01 -1.48203418e-01 -8.91071081e-01
1.67331755e-01 6.15799606e-01 -6.11132532e-02 1.94953248e-01
2.81942457e-01 5.74199259e-01 3.86583917e-02 3.35372865e-01
3.57114784e-02 3.62370670e-01 3.15257281e-01 3.67243677e-01
1.03612892e-01 -4.06251013e-01 1.51352286e-01 8.19172204e-01
-3.76006097e-01 2.82561213e-01 -3.52584660e-01 1.23411262e+00
2.21069809e-02 -1.07025400e-01 -7.00011373e-01 -4.78384316e-01
4.56687361e-01 8.53833616e-01 -1.40839487e-01 -7.93018267e-02
-7.66158283e-01 3.00379723e-01 2.35982373e-01 3.53756487e-01
-6.62331462e-01 2.27315724e-01 1.48311865e+00 1.22938588e-01
-6.54281318e-01 1.88453168e-01 -9.52197313e-01 -1.04030073e+00
-1.98496431e-01 -1.00844133e+00 5.32115042e-01 -1.14540853e-01
-1.41571462e+00 -3.51189703e-01 -4.40697968e-01 -5.99002540e-01
-2.64268041e-01 -4.34813738e-01 -6.42090082e-01 1.04538238e+00
-1.60106552e+00 -1.12674391e+00 -3.20164599e-02 3.73976380e-01
-2.60990858e-01 3.97279561e-02 1.03917968e+00 7.40214586e-02
-3.99451762e-01 5.77264607e-01 9.24727991e-02 2.52415147e-03
5.34328043e-01 -1.33181226e+00 5.39301097e-01 3.63505781e-01
-1.36252493e-01 6.84334457e-01 4.48673218e-01 -5.50781071e-01
-1.04652810e+00 -4.66554284e-01 1.36359203e+00 -1.04601896e+00
7.67592728e-01 5.53791597e-02 -8.51443529e-01 8.93524766e-01
8.21417943e-02 -2.22524419e-01 1.28093016e+00 6.26489580e-01
-5.70934772e-01 -2.68303175e-02 -1.45368493e+00 6.63907349e-01
1.19786775e+00 -1.06616247e+00 -5.40785193e-01 5.99810630e-02
7.25708544e-01 -5.85180484e-02 -1.06306493e+00 -4.29705083e-02
1.26963449e+00 -1.32947242e+00 9.14629400e-01 -1.07167292e+00
6.84859216e-01 4.93109107e-01 -2.05138266e-01 -7.72630453e-01
-3.50834489e-01 -7.23514915e-01 3.60907733e-01 1.28331125e+00
5.26108623e-01 -1.05578613e+00 6.16666913e-01 1.54843235e+00
9.21876654e-02 -9.86166179e-01 -7.99982667e-01 -1.61765769e-01
7.82587588e-01 -4.94168311e-01 9.68980253e-01 1.57237649e+00
7.79366493e-01 -3.55157942e-01 -1.43605828e-01 -4.68641251e-01
7.30514884e-01 1.17661897e-02 4.10140812e-01 -1.88078725e+00
-3.70497331e-02 -7.98924744e-01 -8.11615288e-02 -1.71585120e-02
1.14692062e-01 -9.54151213e-01 -3.89529198e-01 -1.42254663e+00
4.38370019e-01 -6.59373939e-01 -4.52316374e-01 3.60620826e-01
-1.94920555e-01 -1.58283383e-01 4.92531329e-01 4.03502077e-01
-1.90780357e-01 8.24348256e-02 9.27613497e-01 3.16384017e-01
-5.00438102e-02 7.53786713e-02 -2.02194405e+00 1.06720328e+00
1.05557275e+00 -5.12838542e-01 -3.25764716e-01 -4.12349850e-01
5.33519804e-01 1.52076622e-02 6.57921672e-01 -1.67715982e-01
-2.07533449e-01 -1.12850881e+00 9.36040431e-02 5.38134634e-01
-1.54397279e-01 -1.02962852e+00 2.97349654e-02 6.11430824e-01
-8.00660193e-01 -3.17786224e-02 -2.01733634e-01 2.83104211e-01
9.12121683e-02 -1.00909740e-01 6.85638607e-01 -3.12877953e-01
1.75538287e-02 -3.28179933e-02 -1.09299205e-01 5.40011883e-01
7.62402475e-01 -1.49779856e-01 -7.11712420e-01 -5.43548584e-01
-6.37459040e-01 3.92600626e-01 9.13304985e-01 2.97300577e-01
9.21072438e-02 -1.35433936e+00 -6.79861128e-01 -9.69000999e-03
-9.85327084e-03 -4.69893754e-01 1.44657716e-01 8.51277292e-01
-2.57088691e-01 4.24729735e-01 -1.30703464e-01 2.59984672e-01
-8.23465705e-01 5.03644407e-01 7.57096231e-01 3.13936681e-01
-2.87299067e-01 5.52136481e-01 9.51039940e-02 -2.49392852e-01
2.82330304e-01 -2.62576014e-01 -2.09314197e-01 3.94740134e-01
3.75868022e-01 7.43803442e-01 -7.39098966e-01 -5.47248304e-01
-3.25434923e-01 2.48064443e-01 4.47426230e-01 -2.20376775e-01
1.01838028e+00 -3.92373174e-01 -4.58083779e-01 7.52218544e-01
8.99835587e-01 1.40624210e-01 -1.22373521e+00 -7.47139081e-02
4.76280540e-01 -8.00220788e-01 -2.55937874e-01 -1.04073644e+00
-5.53384066e-01 8.03313315e-01 4.30893481e-01 6.45528257e-01
8.04225385e-01 1.40722856e-01 2.35707790e-01 -1.73172832e-01
-1.51943699e-01 -1.34676147e+00 -4.11783040e-01 -8.42626393e-02
8.87239456e-01 -1.46628582e+00 1.93810407e-02 -1.01984449e-01
-9.99492228e-01 7.06260502e-01 3.04141968e-01 4.85542193e-02
8.02979231e-01 1.02152079e-01 5.12060165e-01 -6.64594918e-02
-6.61904395e-01 -2.16223553e-01 4.41818684e-02 5.55215120e-01
9.40032601e-01 6.35807157e-01 -9.36879933e-01 1.01651776e+00
-4.20590818e-01 3.73224586e-01 3.19049090e-01 6.91502571e-01
-5.67479908e-01 -1.25660133e+00 -2.93223143e-01 5.18661261e-01
-1.17011523e+00 -7.69627169e-02 -5.55959225e-01 6.95207298e-01
3.97775441e-01 1.04010487e+00 1.21596225e-01 -2.59423286e-01
2.22066492e-01 2.56448895e-01 2.11174399e-01 -2.60781467e-01
-9.25058365e-01 -1.89856201e-01 2.72163630e-01 -6.67397499e-01
-6.20622754e-01 -8.31549525e-01 -8.67749155e-01 -1.01408744e+00
1.93591461e-01 1.27457798e-01 2.54150301e-01 1.05842912e+00
2.59735018e-01 7.37185627e-02 4.86556739e-01 -6.23324402e-02
-8.74107301e-01 -4.69832808e-01 -5.96901536e-01 3.35125148e-01
7.93138385e-01 -3.31409097e-01 -5.38619637e-01 -3.44325989e-01]
|
[8.806891441345215, 5.53641414642334]
|
57a999b0-356c-440c-b082-785002bf4922
|
umduluth-cs8761-at-semeval-2018-task-9
|
1805.10271
| null |
http://arxiv.org/abs/1805.10271v1
|
http://arxiv.org/pdf/1805.10271v1.pdf
|
UMDuluth-CS8761 at SemEval-2018 Task 9: Hypernym Discovery using Hearst Patterns, Co-occurrence frequencies and Word Embeddings
|
Hypernym Discovery is the task of identifying potential hypernyms for a given
term. A hypernym is a more generalized word that is super-ordinate to more
specific words. This paper explores several approaches that rely on
co-occurrence frequencies of word pairs, Hearst Patterns based on regular
expressions, and word embeddings created from the UMBC corpus. Our system
Babbage participated in Subtask 1A for English and placed 6th of 19 systems
when identifying concept hypernyms, and 12th of 18 systems for entity
hypernyms.
|
['Ted Pedersen', 'Arshia Z. Hassan', 'Manikya S. Vallabhajosyula']
|
2018-05-25
| null | null | null | null |
['hypernym-discovery']
|
['natural-language-processing']
|
[-6.52952641e-02 3.65305215e-01 -4.55725253e-01 -1.13494933e-01
4.89864312e-02 -4.32259083e-01 9.19533610e-01 7.87574887e-01
-1.03593242e+00 8.95022571e-01 3.63203824e-01 -5.46095908e-01
-5.66769302e-01 -1.14274478e+00 6.68920204e-02 -3.70488852e-01
-4.23495710e-01 9.94023442e-01 5.08015566e-02 -7.93223023e-01
1.64470658e-01 3.60295564e-01 -1.76156175e+00 1.42511636e-01
6.06428921e-01 5.60999334e-01 5.41067794e-02 4.48980302e-01
-7.18365252e-01 1.62592679e-01 -5.92566967e-01 -4.96682227e-01
2.59404629e-01 2.22713500e-01 -1.10118961e+00 -5.90075910e-01
4.44099635e-01 3.33973944e-01 -4.21382725e-01 1.14430320e+00
4.22593772e-01 2.93302953e-01 5.16709983e-01 -1.48123109e+00
-5.52015126e-01 8.94850194e-01 -8.59898776e-02 4.09076869e-01
6.64788663e-01 -5.90871334e-01 1.95544946e+00 -9.05396998e-01
9.83959794e-01 1.16541803e+00 2.50645369e-01 5.21872163e-01
-8.60371470e-01 -9.31541145e-01 -3.03025991e-01 5.22993505e-01
-1.66855419e+00 1.41054079e-01 -1.95510596e-01 -1.49676800e-01
1.77287412e+00 3.45661253e-01 8.58693600e-01 6.94517553e-01
-3.41018409e-01 1.75589904e-01 5.85277617e-01 -1.08373988e+00
8.46269876e-02 3.74604054e-02 8.67253482e-01 3.87394100e-01
1.05408180e+00 6.77826256e-02 -1.54941484e-01 -7.92845845e-01
2.99899608e-01 -9.89522859e-02 -2.37214878e-01 -1.08124614e-01
-1.03344750e+00 9.64256287e-01 1.52185366e-01 8.43073070e-01
-4.18772072e-01 -1.24861553e-01 3.95811945e-01 2.82287240e-01
-1.41802952e-01 1.67305601e+00 -5.46145141e-01 -1.03003621e-01
-4.37027395e-01 5.75569332e-01 1.13837266e+00 1.23750234e+00
8.69096339e-01 -4.91575241e-01 1.47764787e-01 1.09287941e+00
9.77635384e-02 2.13170886e-01 1.09641838e+00 -3.79699796e-01
2.77603995e-02 9.96407926e-01 -7.39026517e-02 -6.13132894e-01
-6.23417020e-01 1.10077128e-01 -6.48892298e-02 -2.50102043e-01
-2.95985699e-01 1.91744223e-01 -1.04211354e+00 1.45041907e+00
3.47562581e-01 7.87037760e-02 4.03670609e-01 4.89075452e-01
1.30977881e+00 5.63484788e-01 3.82402688e-01 -1.25338525e-01
1.94321740e+00 -3.58563364e-01 -8.38100135e-01 -7.83708785e-03
7.75864720e-01 -7.06018448e-01 8.34191978e-01 8.63382295e-02
-4.94687259e-01 2.10515842e-01 -1.13208747e+00 8.72843266e-02
-1.42384827e+00 -6.61952198e-01 8.18503678e-01 6.92326009e-01
-7.13080287e-01 3.25616986e-01 1.54124433e-02 -1.08933342e+00
-1.59480795e-01 3.40436071e-01 -5.91836512e-01 -7.46971443e-02
-2.06699228e+00 1.57377160e+00 1.43323720e+00 -8.15739155e-01
-2.74732202e-01 -7.85835326e-01 -1.17349160e+00 6.11394159e-02
3.13762754e-01 -3.90386879e-01 9.71701920e-01 -9.38429236e-02
-3.12132865e-01 1.11479068e+00 -5.17358147e-02 -6.69175804e-01
-6.17158234e-01 -7.93659687e-02 -1.66523278e+00 -1.41281532e-02
3.04125249e-01 4.62821335e-01 2.44885996e-01 -1.13559532e+00
-1.00807321e+00 5.96650839e-02 2.41595849e-01 4.15689737e-01
-8.71090353e-01 1.82206839e-01 -1.60904266e-02 -5.17666876e-01
6.77791908e-02 -7.04546154e-01 -2.30355769e-01 -7.21666396e-01
-3.73367459e-01 -6.40330434e-01 6.23341858e-01 -2.66464174e-01
1.77596962e+00 -1.76098728e+00 -4.82670128e-01 7.19403386e-01
5.61747193e-01 5.61466098e-01 -2.57344335e-01 8.40208888e-01
-8.73256564e-01 5.72682858e-01 1.16547875e-01 7.07270741e-01
5.26989065e-02 8.19969952e-01 -3.95337284e-01 -1.66391507e-01
-1.27632543e-01 6.45104706e-01 -1.34295487e+00 -4.96773452e-01
1.22294836e-01 -1.35684893e-01 -3.83629724e-02 -6.58627078e-02
1.06582336e-01 -9.35415387e-01 -7.09107891e-02 7.02342927e-01
1.87072888e-01 -9.84421815e-04 5.34966528e-01 -1.80182189e-01
1.50411446e-02 7.33723581e-01 -1.00092256e+00 1.11669421e+00
-6.45538449e-01 5.18396258e-01 -6.93557858e-01 -6.45361900e-01
7.51483917e-01 9.43164945e-01 6.72211826e-01 -4.98463362e-01
-1.84969343e-02 7.06394374e-01 3.47975999e-01 -9.08977568e-01
9.63001788e-01 -2.39528701e-01 -1.60742328e-01 3.89656752e-01
5.30807197e-01 -1.50675997e-01 7.37254977e-01 3.80813032e-01
1.56108260e+00 -7.51025021e-01 1.31671941e+00 -4.92982417e-01
1.81846455e-01 1.29826441e-01 4.11963224e-01 6.59414351e-01
7.26381689e-02 1.70861885e-01 5.54991364e-02 -6.98480308e-01
-1.15274751e+00 -1.22913873e+00 -6.80910707e-01 1.20308220e+00
2.16851130e-01 -1.39604163e+00 -4.58517931e-02 -5.62446535e-01
4.46526825e-01 1.15098059e+00 -5.45111358e-01 -1.92641973e-01
-3.27146858e-01 -7.01309860e-01 1.08548200e+00 9.22288746e-02
-2.37729684e-01 -1.26723123e+00 -7.90604055e-01 1.42459348e-01
-8.51279497e-02 -1.10730195e+00 -3.15221846e-02 6.42508209e-01
-4.54942167e-01 -1.34916604e+00 -2.17412248e-01 -1.07195497e+00
3.57118905e-01 9.33984965e-02 1.62828064e+00 3.17041397e-01
-8.65170121e-01 1.66119710e-01 -9.08182263e-01 -7.19476998e-01
-1.64839193e-01 2.99576581e-01 5.54980636e-01 -8.03155005e-01
1.31721437e+00 -5.83894730e-01 -3.77019122e-02 -4.78665866e-02
-1.16254079e+00 -6.80266380e-01 2.78748065e-01 9.22009349e-01
1.34634778e-01 1.08393706e-01 2.85860121e-01 -1.19773030e+00
1.16056752e+00 -7.98772812e-01 -8.14226940e-02 8.21219444e-01
-1.34861648e+00 3.15919250e-01 3.97777036e-02 -4.56849635e-01
-5.52152693e-01 -2.00409487e-01 -6.85508847e-02 3.12860496e-02
-1.16410322e-01 6.71497703e-01 -7.82620907e-02 -4.10962366e-02
1.04163206e+00 -2.60962486e-01 -6.51944160e-01 -2.52719939e-01
9.17633832e-01 7.78720200e-01 6.08057559e-01 -8.11538339e-01
7.45942712e-01 1.10278748e-01 -2.30608974e-02 -1.27872539e+00
-3.96087825e-01 -1.39869535e+00 -5.87088168e-01 1.67509317e-01
6.77951276e-01 -6.25513017e-01 -4.09272850e-01 -4.13328916e-01
-1.30160010e+00 7.56965101e-01 -4.94187832e-01 6.07678354e-01
7.57430792e-02 4.05509979e-01 -1.67556033e-01 -6.78283095e-01
-6.01941049e-01 -1.55955434e-01 6.96525574e-01 7.94921592e-02
-1.09578836e+00 -9.62273479e-01 4.90328133e-01 -1.38078973e-01
-2.32919790e-02 -2.43434329e-02 1.55983412e+00 -1.70438874e+00
4.49488223e-01 -7.14300692e-01 3.18104401e-02 -1.18269496e-01
4.01445180e-01 -6.19685724e-02 -7.77982533e-01 -3.16049904e-03
-4.51982558e-01 -2.53444284e-01 7.56566882e-01 -2.18550056e-01
6.93324864e-01 -5.88796973e-01 -6.65576577e-01 3.71309221e-01
1.59174788e+00 5.93591154e-01 7.86835670e-01 7.35222816e-01
5.76822400e-01 8.07482123e-01 5.96655488e-01 5.16590416e-01
1.29730940e-01 4.53350365e-01 3.41884196e-01 2.44284660e-01
4.52213734e-01 -2.95224458e-01 -3.60449970e-01 7.11920321e-01
5.37016615e-02 -2.63750166e-01 -1.32546234e+00 1.07217622e+00
-1.56330287e+00 -1.06220782e+00 -3.26267183e-01 2.10124922e+00
1.09020948e+00 -1.77024409e-01 -6.38029426e-02 1.13839664e-01
9.63810802e-01 6.19163252e-02 -1.28625825e-01 -5.58790863e-01
-3.15282673e-01 9.42715287e-01 7.44794965e-01 5.31704903e-01
-8.55482936e-01 1.24936807e+00 6.80420780e+00 8.76017511e-01
-2.99144328e-01 1.21427001e-02 -3.70546997e-01 -6.40904307e-02
-6.74466133e-01 4.70221609e-01 -9.01530266e-01 1.25313133e-01
6.38589680e-01 -8.26413810e-01 2.93988526e-01 7.47715116e-01
-7.81685412e-01 1.09378152e-01 -1.08087099e+00 1.14752543e+00
2.64680296e-01 -1.06214511e+00 5.44932306e-01 -5.77348657e-02
6.05911911e-01 8.32632929e-02 -4.56093341e-01 3.46660674e-01
7.83517957e-01 -1.41854107e+00 -3.23894322e-02 -2.13504694e-02
8.46012533e-01 -4.87735987e-01 7.68658757e-01 -1.30775748e-02
-1.42385995e+00 -1.54749468e-01 -7.41051555e-01 -4.81008030e-02
3.06974679e-01 6.21806681e-01 -1.58882987e+00 5.72125494e-01
6.55040383e-01 2.94164836e-01 -3.36384237e-01 1.36593020e+00
-7.10930824e-01 2.40872025e-01 -5.71084559e-01 -3.68354231e-01
2.24644050e-01 2.58502394e-01 8.35440338e-01 1.49769247e+00
3.71903211e-01 2.43930340e-01 -5.56715727e-02 5.63086689e-01
-7.36654550e-02 3.48440319e-01 -9.20580089e-01 -4.41931725e-01
1.18882775e+00 1.58611083e+00 -3.03018510e-01 -5.51283538e-01
-2.02319160e-01 5.02623737e-01 -3.16887125e-02 3.80421877e-01
-3.38588893e-01 -1.08304369e+00 1.19957352e+00 -1.19277857e-01
-2.48589620e-01 2.18396634e-01 -2.79158235e-01 -9.34510231e-01
-2.42907599e-01 -6.86875463e-01 1.18350196e+00 -6.65504396e-01
-1.70406568e+00 9.27571952e-01 2.22480804e-01 -1.15667915e+00
-5.41772842e-01 -1.00490284e+00 -5.72373688e-01 9.47669446e-01
-1.03416944e+00 -5.80310345e-01 -9.57060605e-02 3.67843956e-01
3.46502960e-02 -4.24986064e-01 1.64753640e+00 3.38088125e-01
9.23145637e-02 4.97730464e-01 -2.48336673e-01 1.66657954e-01
6.81009352e-01 -1.45656204e+00 7.14124620e-01 4.94858921e-01
6.80958211e-01 1.25800061e+00 8.41327965e-01 -8.99202228e-01
-4.91176218e-01 -6.19412601e-01 1.94811976e+00 -5.05814016e-01
9.38814163e-01 -1.67333886e-01 -1.01576924e+00 4.22421992e-01
3.19351107e-01 -2.22929850e-01 1.05240130e+00 7.40059733e-01
-9.40430701e-01 2.23855808e-01 -1.02425158e+00 7.76238441e-01
1.15407860e+00 -7.17347026e-01 -1.37620842e+00 3.39747936e-01
1.08905089e+00 3.13146263e-02 -1.04261017e+00 6.73770785e-01
6.28450990e-01 1.03495643e-01 1.10300481e+00 -1.30101120e+00
-5.18870950e-02 -2.97365218e-01 -3.88374388e-01 -1.49270248e+00
-3.19016725e-01 -5.98783791e-01 -3.55577916e-01 8.45913529e-01
9.52957988e-01 -5.46634614e-01 4.23721105e-01 6.40806377e-01
2.39230424e-01 -4.55723286e-01 -1.03411174e+00 -1.30215824e+00
-1.08840112e-02 -3.46534252e-01 1.14788115e+00 1.63786983e+00
1.21957612e+00 5.89372635e-01 1.14472352e-01 -1.36113599e-01
1.75545976e-01 -2.21098661e-01 8.98317844e-02 -1.59996092e+00
2.56213337e-01 -6.39984250e-01 -8.35220337e-01 -3.61844271e-01
3.49207610e-01 -1.22882581e+00 7.33976718e-03 -1.49182689e+00
6.77487254e-02 -2.95109421e-01 -4.61229563e-01 8.08169663e-01
-2.02944785e-01 -2.59766817e-01 -7.52999783e-02 -3.89454067e-02
-4.12391633e-01 2.41034120e-01 4.04816628e-01 -9.91693735e-02
3.06171719e-02 -5.31601965e-01 -5.13079584e-01 5.11636317e-01
7.05401838e-01 -6.86662078e-01 -3.84316236e-01 -1.42697126e-01
8.18490982e-01 -5.93455195e-01 2.63241269e-02 -5.34764886e-01
3.61077547e-01 -2.54801989e-01 -3.58406514e-01 -2.31751837e-02
2.00744241e-01 -1.07278836e+00 -6.60684034e-02 5.37348449e-01
-5.18466473e-01 5.52085102e-01 3.26453261e-02 2.24223003e-01
-3.15825313e-01 -9.50456500e-01 3.43987972e-01 -2.83212751e-01
-1.50766921e+00 5.99035807e-02 -4.05495644e-01 4.83649075e-01
9.10126746e-01 -2.71406472e-01 -4.91677672e-01 -2.11922005e-01
-4.41390932e-01 4.36748981e-01 1.31617144e-01 9.31898594e-01
7.63741672e-01 -1.61584938e+00 -4.02760088e-01 -1.65067881e-01
1.04175723e+00 -5.75726926e-01 -6.39310360e-01 -9.74249691e-02
-6.14018857e-01 5.48062623e-01 -2.66632050e-01 3.14282417e-01
-1.47565997e+00 4.52207983e-01 1.47522002e-01 -1.27743438e-01
-4.19519097e-01 9.81380582e-01 -1.51161328e-01 -7.09267259e-01
3.07338059e-01 -1.53743818e-01 -7.19900012e-01 3.61099660e-01
5.55987000e-01 4.98394817e-01 1.04151778e-01 -6.57623768e-01
-7.16324925e-01 2.43972212e-01 -9.71909761e-02 -5.01383424e-01
1.04984319e+00 4.51999843e-01 -7.26684630e-01 5.98348737e-01
1.21732557e+00 -4.25155640e-01 5.39622664e-01 -3.70204300e-01
9.33587492e-01 -4.35156524e-01 -5.76510012e-01 -8.76139939e-01
-4.16135371e-01 2.95123845e-01 4.53235894e-01 5.07410824e-01
6.85435772e-01 1.34533003e-01 9.83588874e-01 9.85199511e-01
5.58874846e-01 -1.28306472e+00 -5.09481311e-01 1.04460740e+00
5.47545731e-01 -7.98015773e-01 2.47482844e-02 -5.63196540e-01
-4.91218895e-01 1.27556801e+00 8.64538848e-01 1.30649284e-01
9.85936761e-01 7.71056488e-02 6.80112392e-02 -7.96996415e-01
-6.90270424e-01 -7.69779682e-01 4.58379477e-01 6.17309988e-01
6.28730655e-01 4.11836147e-01 -1.18663621e+00 6.07092083e-01
-5.09102762e-01 -5.27901053e-01 2.76360542e-01 7.86340177e-01
-6.41265213e-01 -1.38206029e+00 8.23393390e-02 8.39795768e-01
-2.46263534e-01 -1.07727110e+00 -7.13812649e-01 8.51511836e-01
5.79756200e-01 7.77154922e-01 1.45267278e-01 -6.05449736e-01
3.74793112e-01 4.47855413e-01 6.85666874e-02 -1.29330468e+00
-5.87415576e-01 -7.82804906e-01 7.12581515e-01 -2.27729127e-01
-2.13409498e-01 -4.20717388e-01 -1.62688410e+00 -8.84099677e-02
-6.39814854e-01 6.65817261e-01 3.21925431e-01 1.05018342e+00
-5.27929030e-02 2.24542078e-02 9.03869867e-02 3.83704334e-01
-3.26598912e-01 -1.09036994e+00 -8.78587425e-01 9.47071016e-01
-2.97385216e-01 -6.25668526e-01 -2.39925578e-01 -3.21333259e-01]
|
[9.864885330200195, 8.77247428894043]
|
5fa7208e-385f-43c2-ba40-7ec84d2c73cc
|
geometric-models-for-temporally-attributed
|
2108.12239
| null |
https://arxiv.org/abs/2108.12239v1
|
https://arxiv.org/pdf/2108.12239v1.pdf
|
Geometric Models for (Temporally) Attributed Description Logics
|
In the search for knowledge graph embeddings that could capture ontological knowledge, geometric models of existential rules have been recently introduced. It has been shown that convex geometric regions capture the so-called quasi-chained rules. Attributed description logics (DL) have been defined to bridge the gap between DL languages and knowledge graphs, whose facts often come with various kinds of annotations that may need to be taken into account for reasoning. In particular, temporally attributed DLs are enriched by specific attributes whose semantics allows for some temporal reasoning. Considering that geometric models and (temporally) attributed DLs are promising tools designed for knowledge graphs, this paper investigates their compatibility, focusing on the attributed version of a Horn dialect of the DL-Lite family. We first adapt the definition of geometric models to attributed DLs and show that every satisfiable ontology has a convex geometric model. Our second contribution is a study of the impact of temporal attributes. We show that a temporally attributed DL may not have a convex geometric model in general but we can recover geometric satisfiability by imposing some restrictions on the use of the temporal attributes.
|
['Jeff Z. Pan', 'Ana Ozaki', 'Camille Bourgaux']
|
2021-08-27
| null | null | null | null |
['knowledge-graph-embeddings', 'knowledge-graph-embeddings']
|
['graphs', 'methodology']
|
[-1.27751842e-01 8.62685740e-01 -2.75313914e-01 -4.94351208e-01
1.15599565e-01 -6.30463183e-01 8.60362530e-01 5.80658019e-01
-8.98841619e-02 4.87760931e-01 2.29845688e-01 -1.78824738e-01
-9.59146321e-01 -1.39903700e+00 -6.97519839e-01 -4.24796164e-01
-5.66658974e-01 7.76705027e-01 6.63188577e-01 -5.77485085e-01
-1.74367607e-01 8.12218130e-01 -1.85614538e+00 3.06711704e-01
7.50907898e-01 6.63732886e-01 -1.64946228e-01 2.73719877e-01
-3.07810962e-01 8.78147125e-01 -5.61590120e-02 -5.64859867e-01
1.80979017e-02 2.74637714e-03 -1.10441351e+00 -1.87383853e-02
2.34653756e-01 2.20203534e-01 -1.60322756e-01 1.13243449e+00
-2.11507380e-01 2.77862191e-01 5.88121593e-01 -1.84925592e+00
-6.93821847e-01 7.85668552e-01 2.68773705e-01 -1.26388669e-01
6.05879962e-01 -4.48579758e-01 1.19793081e+00 -3.88125896e-01
1.07948101e+00 1.23616612e+00 6.01170540e-01 4.92663622e-01
-1.39359653e+00 1.14385135e-01 1.20576404e-01 6.36733294e-01
-1.58395326e+00 -2.14120522e-01 6.31060779e-01 -4.48274791e-01
8.06253493e-01 4.49058890e-01 6.97646081e-01 8.61383498e-01
-2.39949580e-02 4.46591109e-01 1.06274736e+00 -7.63784945e-01
4.71695811e-01 5.00177205e-01 5.91895938e-01 8.67836118e-01
7.67294168e-01 5.15725724e-02 -4.15837526e-01 -1.66262373e-01
4.24622238e-01 -2.76421160e-01 -2.74397194e-01 -9.19329822e-01
-8.95186663e-01 8.72595966e-01 3.13932449e-01 7.87186563e-01
-6.50715232e-02 1.59294143e-01 3.48101676e-01 2.43426725e-01
8.82894546e-02 3.49217027e-01 -3.55588108e-01 3.94307584e-01
-2.63106495e-01 4.73352462e-01 1.05882561e+00 1.40334690e+00
6.52539909e-01 -4.08630520e-01 2.03263849e-01 3.10738832e-02
2.38609910e-01 1.34844869e-01 -1.69673532e-01 -9.60316837e-01
-7.59940222e-02 1.02502716e+00 3.52569848e-01 -1.04653645e+00
-6.98419750e-01 -1.22099735e-01 -3.15113306e-01 -1.00414902e-01
6.39908612e-01 7.81550765e-01 -1.15707695e-01 1.94896901e+00
3.81846607e-01 8.21022317e-03 4.36095417e-01 5.02036572e-01
3.23968172e-01 1.77467585e-01 3.40278335e-02 -3.87896866e-01
1.52315497e+00 -2.09249601e-01 -8.17996502e-01 2.86639541e-01
9.69107389e-01 2.07132220e-01 1.00060105e+00 3.22259396e-01
-1.13728023e+00 5.99079318e-02 -9.88020599e-01 -1.75231934e-01
-8.13198686e-01 -5.21910787e-01 7.00852275e-01 7.38216877e-01
-9.61942911e-01 5.12381136e-01 -6.63933933e-01 -7.24202156e-01
2.92251427e-02 1.16491549e-01 -5.27849913e-01 -9.05921608e-02
-1.43174255e+00 1.04902685e+00 6.65042043e-01 -1.32625774e-01
-5.74428380e-01 -4.14681047e-01 -9.44931030e-01 4.20917347e-02
1.02197766e+00 -6.27707064e-01 8.73278677e-01 -6.35661662e-01
-8.49562049e-01 1.19212747e+00 6.56494945e-02 -7.12616801e-01
5.31978190e-01 3.44918400e-01 -1.10076869e+00 2.17540294e-01
9.21082944e-02 -9.73475575e-02 2.40437999e-01 -1.12324071e+00
-5.17689109e-01 -7.50722408e-01 9.29908454e-01 -2.68761098e-01
-2.68037289e-01 -1.79983869e-01 -7.20614344e-02 -5.03451079e-02
1.60289213e-01 -8.23807478e-01 4.11527753e-02 -1.53755993e-01
-3.48914623e-01 -5.47437191e-01 4.58751202e-01 9.80468690e-02
1.28245425e+00 -2.08382177e+00 3.93510133e-01 4.61659312e-01
2.07435608e-01 -2.99575865e-01 3.71762246e-01 6.26605928e-01
-3.48965600e-02 3.88073027e-01 -1.27914503e-01 4.10036184e-02
7.23458588e-01 7.37626433e-01 -3.91400278e-01 6.86031938e-01
-2.39333957e-01 6.48743629e-01 -8.70876670e-01 -6.98735297e-01
1.90127701e-01 2.11975306e-01 -7.76348293e-01 -2.45206922e-01
-9.38424408e-01 1.28125930e-02 -4.42600101e-01 3.68100822e-01
5.90194821e-01 -1.11189321e-01 6.26236677e-01 -3.43087375e-01
-3.32721651e-01 4.00575884e-02 -1.40183854e+00 1.65532398e+00
-2.56594658e-01 -2.31559314e-02 -1.76185489e-01 -8.53566527e-01
7.53641248e-01 4.16232795e-01 4.06090558e-01 -6.35643542e-01
-1.57067813e-02 2.65405297e-01 -4.05098259e-01 -7.94725060e-01
6.01645052e-01 -4.92795557e-01 -4.29506004e-01 2.56173164e-02
-1.99521095e-01 1.85221866e-01 3.78581017e-01 2.96270967e-01
9.08918977e-01 3.82679045e-01 3.89902830e-01 -8.04451823e-01
8.26881051e-01 3.92376184e-01 7.87715435e-01 3.56809318e-01
3.70491147e-02 -6.66359020e-03 8.68158519e-01 -6.16084278e-01
-9.43882763e-01 -1.18576992e+00 -5.24600983e-01 8.47375154e-01
4.79301453e-01 -8.03205192e-01 -5.72049975e-01 -4.71870333e-01
-1.16332993e-01 1.16862500e+00 -7.95272708e-01 -4.23307449e-01
-3.11015844e-01 -2.15135917e-01 8.85276258e-01 5.00361979e-01
1.11356318e-01 -6.15204811e-01 -1.00425482e+00 9.25259367e-02
-1.08337924e-01 -1.30711591e+00 1.56538934e-01 4.98516075e-02
-7.04840779e-01 -1.55980539e+00 3.83166492e-01 -3.94460946e-01
4.88714874e-01 -4.27055329e-01 1.10228205e+00 1.36067405e-01
3.69679704e-02 8.19906592e-01 -5.89580417e-01 -3.62708777e-01
-3.22903335e-01 -2.75281370e-01 3.60322446e-01 2.55557954e-01
4.81835455e-01 -5.58727145e-01 -5.88433966e-02 1.93574414e-01
-1.40953314e+00 8.37279037e-02 -3.03455114e-01 2.30413124e-01
6.29895091e-01 4.13208425e-01 1.99139893e-01 -9.49878693e-01
1.49551407e-01 -4.27993625e-01 -8.38247597e-01 6.16727710e-01
-5.89784682e-01 7.70044208e-01 6.01892650e-01 -5.51452972e-02
-1.12544119e+00 -3.25909704e-01 4.11311686e-01 -3.72105956e-01
-1.66047737e-01 9.32368040e-01 -6.09082997e-01 2.90942669e-01
6.89122617e-01 -2.11408809e-01 -2.23750681e-01 -2.99983561e-01
5.49377978e-01 2.90945768e-02 4.11274374e-01 -1.20036876e+00
5.77684939e-01 9.38777804e-01 9.14278805e-01 -7.08340049e-01
-6.27481937e-01 -1.24503665e-01 -7.38065541e-01 -8.51404890e-02
9.65896606e-01 -3.00668210e-01 -1.08930051e+00 -3.27655643e-01
-1.00585353e+00 -1.68298587e-01 -6.96288764e-01 2.17742667e-01
-9.95327175e-01 2.98388243e-01 -2.79172182e-01 -9.59678352e-01
4.41837162e-01 -7.96446204e-01 4.58996654e-01 -3.25390518e-01
-2.43338078e-01 -1.30901527e+00 7.15056285e-02 -6.66016415e-02
7.53117800e-02 6.41710758e-01 1.58788991e+00 -9.55834389e-01
-4.17845875e-01 -1.31295979e-01 1.26136437e-01 -1.77509025e-01
-1.95917413e-01 -8.35330039e-03 -6.69194460e-01 7.08126929e-03
-3.42068970e-02 2.64398217e-01 1.20587923e-01 5.98040083e-03
6.41649544e-01 -6.41665220e-01 -3.56005192e-01 3.07521254e-01
1.90702069e+00 6.47517741e-02 6.99530721e-01 5.74508965e-01
3.24451655e-01 8.75611305e-01 5.77801645e-01 2.50875086e-01
7.15561330e-01 1.06985927e+00 7.31172502e-01 7.17636645e-01
2.78950870e-01 -1.97569460e-01 1.44475117e-01 2.55947977e-01
-4.28712696e-01 8.00534189e-02 -1.03358769e+00 6.49062753e-01
-1.94346881e+00 -1.21223176e+00 -7.67435789e-01 2.47411156e+00
7.23069608e-01 -5.38106402e-03 2.86898643e-01 2.34010682e-01
7.79538929e-01 -1.25596955e-01 7.76992813e-02 -5.05248725e-01
-3.15408468e-01 1.98647659e-02 4.53352809e-01 9.49901581e-01
-6.66147113e-01 5.64403713e-01 5.45974636e+00 1.87791169e-01
-5.20579994e-01 3.17043543e-01 -5.40387392e-01 9.51392949e-02
-8.46456707e-01 5.27796686e-01 -5.61879396e-01 1.83955580e-01
9.99301732e-01 -5.68089128e-01 2.85053641e-01 7.60309458e-01
-8.03564787e-02 8.94029066e-02 -1.58468461e+00 3.40530396e-01
-1.94062799e-01 -1.31723630e+00 1.11492045e-01 3.12458009e-01
4.83551919e-01 -6.05750084e-01 -3.60524774e-01 1.13044776e-01
3.51654261e-01 -9.48468387e-01 1.16245878e+00 1.05248702e+00
6.36482179e-01 -8.98754060e-01 7.22566128e-01 2.17941076e-01
-1.42881525e+00 -2.23134503e-01 -2.70984173e-01 -1.60827748e-02
1.08833328e-01 5.00636160e-01 -4.43162948e-01 1.22008026e+00
4.76685822e-01 5.44429660e-01 -3.30529302e-01 6.72173262e-01
-1.32650286e-01 -8.77647195e-03 -3.70375574e-01 3.27895641e-01
1.24630235e-01 -4.70174372e-01 6.85474217e-01 8.69121075e-01
3.72583479e-01 1.92588389e-01 -5.83345257e-02 1.21738923e+00
3.67804587e-01 -7.56755993e-02 -9.08528149e-01 1.50391355e-01
4.43085819e-01 5.71663678e-01 -5.94238400e-01 -2.73169458e-01
-5.66813290e-01 4.38185304e-01 9.13639665e-02 2.50285387e-01
-9.56810474e-01 6.30852506e-02 6.51359856e-01 6.77829146e-01
-1.66837089e-02 -1.10463679e-01 5.55573851e-02 -1.15293145e+00
1.00413002e-01 -3.04239124e-01 9.23802614e-01 -8.64292264e-01
-9.62698042e-01 5.07737577e-01 6.75701201e-01 -1.00895619e+00
-1.91013813e-02 -6.42772615e-01 4.90193348e-03 4.39988792e-01
-1.45140040e+00 -1.16400826e+00 -1.64618641e-01 1.17064238e+00
-2.76178658e-01 4.22289550e-01 1.12896311e+00 -4.48734649e-02
-2.09118277e-01 8.22703466e-02 -3.27297062e-01 -3.57459962e-01
2.34934449e-01 -1.46013439e+00 -3.69972020e-01 8.76989245e-01
8.82476121e-02 6.71929657e-01 1.20708895e+00 -4.49715734e-01
-1.70662105e+00 -1.10878837e+00 1.42750299e+00 -6.16206646e-01
7.35232115e-01 -8.34892243e-02 -1.01218724e+00 1.49638641e+00
-2.08094195e-01 3.53596300e-01 4.10537899e-01 2.23385721e-01
-8.48189175e-01 -3.54455709e-01 -1.43508232e+00 5.85403144e-01
1.40214598e+00 -6.70956969e-01 -9.39283192e-01 1.19060196e-01
7.71856606e-01 -2.22052485e-02 -1.17164552e+00 5.24781823e-01
2.91456610e-01 -1.28928471e+00 8.58113647e-01 -7.89014697e-01
-2.05430493e-01 -7.24294603e-01 -5.25519311e-01 -8.15376878e-01
-2.53495812e-01 -2.98471600e-01 -2.96073914e-01 1.01692450e+00
1.31270647e-01 -8.41545522e-01 3.68340731e-01 9.29732382e-01
-2.95318335e-01 -2.86388010e-01 -1.11306906e+00 -1.36707926e+00
5.00108227e-02 -7.54453361e-01 8.77385378e-01 1.12540424e+00
8.12855244e-01 -2.05547050e-01 1.16614290e-01 4.78440017e-01
7.79264808e-01 1.47672370e-01 7.38798231e-02 -1.94505429e+00
-1.10769652e-01 -4.52400982e-01 -9.03177142e-01 -3.81174944e-02
3.40394795e-01 -1.34616506e+00 -3.38410735e-01 -1.58411336e+00
-2.73567110e-01 -6.65287018e-01 8.97727981e-02 5.88953078e-01
8.38461280e-01 -2.19577178e-01 -1.00841872e-01 3.80131826e-02
-7.51805604e-01 2.70155638e-01 7.85044372e-01 -1.15058832e-02
-6.58724383e-02 -4.27713543e-01 -4.87977952e-01 8.60088527e-01
6.00998938e-01 -3.26328337e-01 -6.88092053e-01 -1.70888901e-01
1.13453782e+00 2.58222014e-01 7.05800891e-01 -8.67523134e-01
5.19324899e-01 -5.19083917e-01 -7.45665252e-01 1.09274387e-01
1.80732399e-01 -1.41734064e+00 8.96099567e-01 3.14417601e-01
-4.08311456e-01 -1.78173482e-01 3.02116610e-02 5.82423627e-01
-2.47799270e-02 -5.10308504e-01 6.38155043e-01 -1.56221926e-01
-8.45834076e-01 5.69114387e-02 -1.03619725e-01 -9.40691316e-05
1.46321535e+00 -2.87458003e-01 -2.07365081e-01 1.58806145e-01
-1.28829992e+00 -6.61567003e-02 9.33477759e-01 1.06830388e-01
3.20269704e-01 -1.45694911e+00 -1.25158548e-01 -1.69126093e-01
6.53788686e-01 -2.82884780e-02 1.75884634e-01 1.07816243e+00
-3.75456154e-01 5.92625916e-01 -1.22145951e-01 -2.72945851e-01
-1.00467777e+00 1.23643684e+00 5.31226218e-01 -5.70787080e-02
-8.41347635e-01 3.73451233e-01 -5.02524413e-02 -1.38188079e-01
2.30393052e-01 -7.30588138e-01 -1.84783727e-01 1.99250221e-01
2.68729925e-01 5.24976075e-01 6.32206500e-02 -8.13704312e-01
-8.48155379e-01 4.85330671e-01 4.45666999e-01 -7.46414214e-02
1.31241977e+00 -2.50550807e-01 -5.95560253e-01 8.18495333e-01
7.40666449e-01 1.83703721e-01 -5.87606132e-01 -3.29536706e-01
5.13673961e-01 -7.60087594e-02 -3.35237592e-01 -4.21383590e-01
-6.19452536e-01 3.83169085e-01 -6.49778768e-02 9.57597256e-01
8.84898961e-01 5.27270555e-01 2.68282574e-02 1.60365358e-01
1.03623772e+00 -1.00603509e+00 -3.92990828e-01 3.99690628e-01
8.92200828e-01 -3.91508698e-01 -1.98471993e-01 -8.15497816e-01
-3.34701300e-01 1.27522099e+00 1.45158350e-01 3.12758144e-03
4.75432336e-01 2.91295737e-01 -6.49766803e-01 -6.96617246e-01
-6.62864268e-01 -4.98960674e-01 3.07774190e-02 6.09920144e-01
-3.94966714e-02 3.14240783e-01 -5.55256784e-01 9.83703256e-01
-2.22894654e-01 1.88851789e-01 8.77589583e-01 1.04183817e+00
-2.98285306e-01 -1.09591508e+00 -3.71011764e-01 -1.21739961e-01
-2.20262140e-01 1.78416759e-01 -1.80040121e-01 1.16579700e+00
4.88555998e-01 8.93173575e-01 1.88920200e-01 -2.64732659e-01
6.19297802e-01 4.16586578e-01 9.66723502e-01 -5.42572558e-01
-3.92243685e-03 -7.06558347e-01 2.92143166e-01 -4.93026555e-01
-7.53906965e-01 -5.66186011e-01 -1.59605157e+00 -5.57940900e-01
-2.07889140e-01 5.01427293e-01 1.95011318e-01 1.22929823e+00
3.95052880e-02 3.81258577e-01 -4.16727588e-02 -1.14998929e-01
-2.71642059e-01 -3.68701071e-01 -1.12819266e+00 6.15260899e-01
2.03309387e-01 -8.12978566e-01 -4.87246543e-01 6.76565543e-02]
|
[8.724778175354004, 6.905439376831055]
|
a0c78263-bda1-47f6-8cd6-2f99b7c3bc52
|
cross-modal-learning-for-audio-visual-video
|
2104.04598
| null |
https://arxiv.org/abs/2104.04598v2
|
https://arxiv.org/pdf/2104.04598v2.pdf
|
Cross-Modal learning for Audio-Visual Video Parsing
|
In this paper, we present a novel approach to the audio-visual video parsing (AVVP) task that demarcates events from a video separately for audio and visual modalities. The proposed parsing approach simultaneously detects the temporal boundaries in terms of start and end times of such events. We show how AVVP can benefit from the following techniques geared towards effective cross-modal learning: (i) adversarial training and skip connections (ii) global context aware attention and, (iii) self-supervised pretraining using an audio-video grounding objective to obtain cross-modal audio-video representations. We present extensive experimental evaluations on the Look, Listen, and Parse (LLP) dataset and show that we outperform the state-of-the-art Hybrid Attention Network (HAN) on all five metrics proposed for AVVP. We also present several ablations to validate the effect of pretraining, global attention and adversarial training.
|
['Ganesh Ramakrishnan', 'Preethi Jyothi', 'Rishabh Dabral', 'Jayaprakash Akula', 'abhishek', 'Jatin Lamba']
|
2021-04-03
| null | null | null | null |
['video-grounding']
|
['computer-vision']
|
[ 4.99076396e-01 2.46026561e-01 1.19003460e-01 -4.68751818e-01
-1.52857327e+00 -6.54291153e-01 7.07325935e-01 2.81775333e-02
-2.45436162e-01 3.37529987e-01 4.93034840e-01 -1.43908873e-01
9.51734185e-02 -4.25076842e-01 -1.03876400e+00 -4.30701792e-01
-5.60009897e-01 2.22177580e-01 3.19027483e-01 3.03932074e-02
-1.00780003e-01 2.81033844e-01 -1.63866782e+00 8.87440920e-01
8.28055516e-02 1.24880743e+00 -6.44488335e-02 1.38184786e+00
-8.14694613e-02 1.21375704e+00 -5.98689795e-01 -3.28806818e-01
9.66609456e-04 -6.62493169e-01 -1.07245183e+00 -2.76970286e-02
7.89999008e-01 -2.56756157e-01 -3.65642577e-01 6.72811210e-01
6.47076368e-01 1.82350665e-01 4.30355906e-01 -1.44822907e+00
-5.73063254e-01 9.09373641e-01 -4.65996355e-01 5.53348780e-01
7.09482253e-01 6.67382330e-02 1.21037090e+00 -7.68120527e-01
6.51907206e-01 1.36789882e+00 7.80165553e-01 6.14597321e-01
-1.08861387e+00 -6.87108576e-01 4.43787992e-01 5.44128358e-01
-1.15263391e+00 -6.51934564e-01 1.00984621e+00 -5.59513450e-01
1.11943865e+00 3.22849870e-01 4.51778442e-01 1.58355904e+00
-1.37371734e-01 7.87261486e-01 6.62350297e-01 -6.07027590e-01
1.90071352e-02 -1.88112333e-01 1.18506864e-01 6.37260377e-01
-8.32803726e-01 2.78885305e-01 -8.53930116e-01 3.95737495e-03
4.53310519e-01 -4.32157129e-01 -3.17029804e-01 -2.70345390e-01
-1.06169283e+00 7.23033309e-01 3.27478260e-01 2.93809563e-01
-2.34107226e-01 5.09125054e-01 9.61059391e-01 3.36910874e-01
2.25426853e-01 9.90498960e-02 -3.25251698e-01 -1.06194057e-01
-1.02921128e+00 2.14571301e-02 3.71587455e-01 8.43049884e-01
4.32290286e-01 3.54903907e-01 -5.00010550e-01 6.12481952e-01
3.15524578e-01 2.17921987e-01 3.05166066e-01 -1.28956294e+00
6.25073910e-01 -1.22119285e-01 -1.29142329e-01 -7.76962698e-01
-2.89878428e-01 5.20796403e-02 -4.52936560e-01 1.28385305e-01
1.58679634e-01 -2.34011739e-01 -1.05923522e+00 2.05886269e+00
1.10696644e-01 6.17898524e-01 1.51466027e-01 7.47301996e-01
1.27306712e+00 9.06235635e-01 5.69855809e-01 -1.43955216e-01
1.15902722e+00 -1.08458579e+00 -8.18125784e-01 -2.92317361e-01
4.86009680e-02 -4.96966660e-01 9.41105783e-01 2.38721207e-01
-1.31606460e+00 -9.36661184e-01 -1.00450492e+00 -9.00104269e-02
-2.62972057e-01 5.24778711e-03 4.13589805e-01 4.11089778e-01
-1.22841918e+00 5.32841742e-01 -9.06885326e-01 -3.23465437e-01
2.47459620e-01 2.67862201e-01 -4.42059875e-01 2.46375114e-01
-1.29437840e+00 2.95322269e-01 3.53826612e-01 5.94008304e-02
-1.64707208e+00 -6.00173771e-01 -1.10526776e+00 1.48863629e-01
2.69929022e-01 -5.00089586e-01 1.32263827e+00 -1.65145433e+00
-1.45160234e+00 1.07348204e+00 -8.47408548e-02 -7.12558091e-01
2.02762425e-01 -4.81446445e-01 -4.94207084e-01 8.21797371e-01
5.00131026e-02 1.11861193e+00 1.22504878e+00 -1.34552681e+00
-6.87344670e-01 -1.09550998e-01 2.86134332e-01 3.02580837e-02
7.72908553e-02 3.32573354e-01 -7.98878491e-01 -6.91937089e-01
-3.19551468e-01 -7.09383607e-01 3.37845653e-01 -1.69304132e-01
-4.24031824e-01 1.23936264e-02 8.46682847e-01 -9.80316818e-01
1.13678122e+00 -2.36113000e+00 5.11449099e-01 -1.82282746e-01
-1.18434705e-01 8.18780139e-02 -4.41174746e-01 4.88219649e-01
-6.46666765e-01 2.42932945e-01 -4.82220016e-02 -7.13044643e-01
-1.25049487e-01 2.29789898e-01 -5.13976693e-01 2.50788331e-01
5.01934826e-01 5.67800760e-01 -8.51966321e-01 -6.53584301e-01
1.98988095e-01 7.89785743e-01 -7.00835764e-01 6.31070197e-01
-1.79785147e-01 6.05098188e-01 1.84166133e-02 7.97095001e-01
1.81287736e-01 2.36228615e-01 1.87690452e-01 -3.16066474e-01
5.01496643e-02 2.22284421e-01 -9.42443013e-01 1.94497490e+00
-5.23391306e-01 1.06716847e+00 4.28734481e-01 -1.04306364e+00
4.62874681e-01 7.62877941e-01 4.65718299e-01 -5.58162749e-01
1.58175781e-01 -3.49379718e-01 -4.20328200e-01 -6.32228076e-01
2.49057591e-01 -2.92201675e-02 -3.56872886e-01 -2.82195564e-02
7.93329954e-01 3.65935564e-01 6.40348345e-02 3.91390443e-01
1.12158799e+00 3.99408460e-01 5.52605465e-02 2.78170973e-01
6.97060823e-01 -3.10002357e-01 6.30150914e-01 7.68609583e-01
-4.48501617e-01 9.18172538e-01 7.65983701e-01 -2.68022150e-01
-7.22792506e-01 -1.18471408e+00 2.71500736e-01 1.63488686e+00
-1.35566443e-01 -5.40227115e-01 -7.78561234e-01 -8.01282942e-01
-4.03635293e-01 6.72416747e-01 -9.51920867e-01 -1.14625320e-02
-6.73820436e-01 -8.44841376e-02 7.66232848e-01 9.46671069e-01
2.68762350e-01 -1.36115384e+00 -6.49976075e-01 9.26560387e-02
-4.85651851e-01 -1.31412899e+00 -4.44020867e-01 4.04028684e-01
-4.84686464e-01 -1.14208221e+00 -5.05259871e-01 -9.01189506e-01
-4.69953939e-02 -7.50477985e-02 1.42679560e+00 -3.30060035e-01
-6.25142530e-02 1.13951254e+00 -7.28441834e-01 -2.16263428e-01
-6.12093031e-01 -2.26374164e-01 -4.16263312e-01 1.27428502e-01
-2.52754465e-02 -8.20901394e-01 -3.17873359e-01 5.58319278e-02
-8.27716172e-01 -1.82270072e-02 3.66315097e-01 7.92784214e-01
8.03299904e-01 -2.52432168e-01 4.75543022e-01 -6.78043664e-01
2.51440912e-01 -4.98510242e-01 -4.09911901e-01 2.28644803e-01
5.65958880e-02 -1.48784995e-01 4.36487675e-01 -4.56548005e-01
-9.88500178e-01 2.64291883e-01 -4.02639359e-01 -1.00585437e+00
-4.77724522e-01 3.83752763e-01 -3.04839075e-01 1.92078039e-01
4.83721495e-01 -3.05521581e-02 -3.95772934e-01 -3.37098241e-01
5.54605901e-01 4.07982975e-01 1.08224452e+00 -4.57790881e-01
4.68361020e-01 4.57380682e-01 -2.02189416e-01 -7.60637879e-01
-7.53179073e-01 -3.66426736e-01 -6.65513277e-01 -5.02434194e-01
1.47059512e+00 -1.21303105e+00 -6.08049929e-01 1.67681828e-01
-1.31645989e+00 -4.56182182e-01 -2.60412335e-01 2.46460289e-01
-9.09997702e-01 3.27286243e-01 -6.37082040e-01 -8.08780015e-01
-2.51075268e-01 -9.64705288e-01 1.37943959e+00 3.79741974e-02
-2.40856633e-01 -8.89353573e-01 2.36515835e-01 4.14212704e-01
-3.61723676e-02 6.73210084e-01 7.85959661e-01 -7.66310096e-01
-5.24225533e-01 -6.70783520e-02 -1.02254182e-01 2.94303179e-01
-3.32067668e-01 2.59257168e-01 -1.47489965e+00 -8.33690986e-02
-2.74045378e-01 -5.71389496e-01 9.79008257e-01 4.44779456e-01
1.08335018e+00 -2.20873415e-01 -5.45080565e-02 7.98266470e-01
1.36490309e+00 2.87870020e-01 5.92716992e-01 3.15852612e-01
7.10480869e-01 4.58878696e-01 4.97707069e-01 3.08374405e-01
2.05872372e-01 7.58842647e-01 8.54216218e-01 -7.21184984e-02
-4.72850084e-01 -3.15957844e-01 6.83141828e-01 5.88398516e-01
2.69449875e-02 -4.15217161e-01 -8.17098141e-01 9.28586781e-01
-1.76026750e+00 -1.23203003e+00 1.28988981e-01 2.00552559e+00
5.49143493e-01 2.12194130e-01 2.32639313e-01 2.90981829e-01
7.44848967e-01 4.41088796e-01 -2.18493998e-01 -7.70580888e-01
7.35523459e-03 3.89655292e-01 1.93902552e-02 5.25000095e-01
-1.62655532e+00 8.48745346e-01 6.76110601e+00 4.05665874e-01
-1.02931786e+00 4.42239106e-01 4.01520938e-01 -1.65698990e-01
3.98785621e-03 -2.12798908e-01 -4.84574586e-01 1.53554648e-01
1.41069007e+00 4.18273658e-01 4.32925999e-01 7.13354349e-01
-6.29691556e-02 1.99516580e-01 -1.44825542e+00 9.83286083e-01
2.48214021e-01 -1.17392635e+00 2.89722532e-02 -5.33998668e-01
2.93011248e-01 1.24930218e-01 -1.62152633e-01 5.62968969e-01
5.82707822e-02 -8.49585056e-01 1.11681259e+00 3.75940502e-01
7.33039021e-01 -8.47539186e-01 5.10082543e-01 -2.40310371e-01
-1.60048676e+00 -2.10066795e-01 1.94873706e-01 3.84515315e-01
4.30456281e-01 -1.42999440e-01 -4.12527204e-01 7.08126307e-01
1.07503831e+00 7.26491630e-01 -5.22124827e-01 6.80065036e-01
-3.18058670e-01 8.03011596e-01 -1.56151280e-02 6.62185729e-01
3.66824061e-01 3.38192344e-01 7.39626467e-01 1.57259274e+00
5.57827614e-02 -2.55527675e-01 1.25082329e-01 6.30930960e-01
4.82556550e-03 -1.26487449e-01 -5.20013154e-01 -1.36555985e-01
3.14038843e-01 9.69541013e-01 -5.93621433e-01 -1.97516307e-01
-6.55282140e-01 9.94153678e-01 2.30405912e-01 4.23866183e-01
-1.31825113e+00 -2.49591187e-01 7.41730034e-01 -1.22628659e-01
8.36358786e-01 4.32333164e-02 2.53736705e-01 -8.72365296e-01
-1.94167331e-01 -8.54007125e-01 8.13467026e-01 -1.22459269e+00
-9.55072165e-01 9.51297641e-01 1.21946402e-01 -1.08412158e+00
-6.51354969e-01 -3.19332749e-01 -8.08853209e-01 5.54010928e-01
-1.49922562e+00 -1.37650585e+00 -3.06138724e-01 9.35844600e-01
8.37435842e-01 -3.91076654e-02 9.56453979e-01 5.01782775e-01
-5.29315531e-01 6.73359215e-01 -3.50332201e-01 4.63841647e-01
6.49312079e-01 -1.27159476e+00 2.55006224e-01 1.02669930e+00
4.96417403e-01 -5.30100614e-02 6.90797508e-01 -3.42580229e-01
-1.13633287e+00 -1.21412444e+00 5.80904305e-01 -3.57661009e-01
6.97690308e-01 -4.23076153e-01 -8.18004966e-01 1.20393908e+00
7.00148106e-01 1.05170952e-02 7.97307551e-01 1.28165498e-01
-5.90175211e-01 -1.42225936e-01 -8.28155577e-01 2.60122716e-01
8.87725174e-01 -9.84159172e-01 -7.25642025e-01 2.98857331e-01
1.03966975e+00 -3.86328846e-01 -8.01625013e-01 3.38844061e-01
4.72780257e-01 -1.24276865e+00 1.29708397e+00 -8.72422874e-01
6.34239137e-01 -8.06198716e-02 -4.12404388e-01 -8.39237928e-01
-2.79424042e-01 -8.53032529e-01 -1.78667098e-01 1.76688814e+00
2.11190447e-01 1.00777447e-01 3.93487602e-01 1.38879046e-01
-4.83194500e-01 -2.82883465e-01 -1.14943790e+00 -5.83811581e-01
-1.90603524e-01 -8.50730300e-01 1.38673652e-02 8.11585784e-01
-2.39047334e-01 5.18251717e-01 -6.76049948e-01 5.94825864e-01
3.44549149e-01 5.58983684e-02 5.10677576e-01 -8.77584815e-01
-5.02320290e-01 -1.97380111e-01 -6.35125399e-01 -5.32430470e-01
4.78817493e-01 -5.71407676e-01 1.84564844e-01 -1.30447221e+00
-1.81758571e-02 2.61848241e-01 -4.20321882e-01 5.92653096e-01
7.51839280e-02 2.60169327e-01 3.44811320e-01 -6.03355318e-02
-9.61860120e-01 3.30324739e-01 5.55797458e-01 -2.89688796e-01
-1.00353532e-01 -3.31362933e-01 -4.12473798e-01 6.80908442e-01
4.78506088e-01 -5.42519808e-01 -4.83494788e-01 -5.39678156e-01
2.65318472e-02 5.50648868e-01 5.88656783e-01 -1.25551856e+00
-2.62228791e-02 3.04354221e-01 2.58604169e-01 -6.72136068e-01
6.16536558e-01 -8.01465213e-01 1.27728119e-01 7.08878711e-02
-6.14128828e-01 1.61301672e-01 6.09015763e-01 8.94964814e-01
-6.40115142e-01 -1.36340484e-02 8.56167316e-01 -4.16204259e-02
-9.67362165e-01 -8.28898996e-02 -5.29126108e-01 2.68770486e-01
1.03344059e+00 -6.82650181e-03 -8.26818123e-02 -6.95142984e-01
-1.39970291e+00 1.67590961e-01 1.66025069e-02 5.02973020e-01
5.45384526e-01 -1.40734458e+00 -5.66143811e-01 7.00280294e-02
1.98585670e-02 -4.06073511e-01 4.20588464e-01 7.26059854e-01
-4.42251652e-01 3.77240777e-01 -2.71294802e-01 -7.87701905e-01
-1.69109547e+00 9.42013919e-01 3.38979155e-01 -2.45495796e-01
-4.96629626e-01 1.06606114e+00 1.38483226e-01 -6.30350262e-02
8.16688657e-01 -3.14399004e-01 -3.85592431e-01 3.21070164e-01
6.10983253e-01 1.52889624e-01 2.15217695e-02 -8.73825610e-01
-5.14497161e-01 4.88797635e-01 2.24657327e-01 -3.35771471e-01
1.30126429e+00 -1.80638194e-01 3.81466985e-01 6.11972094e-01
1.25115561e+00 3.32762115e-02 -1.39318788e+00 1.23930983e-02
-3.11852604e-01 -9.50713381e-02 1.19209394e-01 -7.12864876e-01
-1.27456331e+00 1.14753687e+00 8.54274035e-01 4.89886612e-01
1.60536158e+00 3.27310383e-01 5.49626291e-01 -9.19692665e-02
-1.70103654e-01 -7.81371117e-01 1.09491013e-01 2.78003275e-01
1.03479648e+00 -1.03490973e+00 -3.91305417e-01 -2.87631214e-01
-8.29368711e-01 1.11735463e+00 3.94656867e-01 -5.65271042e-02
6.00568712e-01 2.52755791e-01 1.87702522e-01 -8.90931040e-02
-9.15545821e-01 -4.12451297e-01 3.88525814e-01 9.14854586e-01
4.50840622e-01 -3.11243206e-01 3.48876238e-01 7.33060956e-01
2.44231388e-01 -2.51951784e-01 3.07458013e-01 9.92295444e-01
-3.51383388e-02 -8.98238003e-01 -3.59386533e-01 -3.59557122e-01
-8.38010132e-01 -1.57244623e-01 -7.10387647e-01 1.06285298e+00
1.54480383e-01 1.02310276e+00 3.45670670e-01 -5.60330510e-01
3.14141363e-01 3.85540903e-01 4.46044028e-01 -4.40518320e-01
-9.30842876e-01 3.39016318e-01 3.87341827e-01 -9.88443613e-01
-9.36710596e-01 -6.68488503e-01 -1.12823725e+00 2.61135519e-01
-3.72745446e-03 1.34536326e-02 4.30354357e-01 7.28441656e-01
4.45626318e-01 1.11091971e+00 4.95139092e-01 -1.22873282e+00
2.11422425e-02 -9.00577605e-01 -1.74899086e-01 5.59078097e-01
4.76799995e-01 -5.91639519e-01 -4.45966661e-01 5.67165792e-01]
|
[10.076506614685059, 1.0087207555770874]
|
fd1368b4-7971-41f7-b3ff-e7df548867cf
|
improving-mutual-information-estimation-with-1
|
2303.06992
| null |
https://arxiv.org/abs/2303.06992v1
|
https://arxiv.org/pdf/2303.06992v1.pdf
|
Improving Mutual Information Estimation with Annealed and Energy-Based Bounds
|
Mutual information (MI) is a fundamental quantity in information theory and machine learning. However, direct estimation of MI is intractable, even if the true joint probability density for the variables of interest is known, as it involves estimating a potentially high-dimensional log partition function. In this work, we present a unifying view of existing MI bounds from the perspective of importance sampling, and propose three novel bounds based on this approach. Since accurate estimation of MI without density information requires a sample size exponential in the true MI, we assume either a single marginal or the full joint density information is known. In settings where the full joint density is available, we propose Multi-Sample Annealed Importance Sampling (AIS) bounds on MI, which we demonstrate can tightly estimate large values of MI in our experiments. In settings where only a single marginal distribution is known, we propose Generalized IWAE (GIWAE) and MINE-AIS bounds. Our GIWAE bound unifies variational and contrastive bounds in a single framework that generalizes InfoNCE, IWAE, and Barber-Agakov bounds. Our MINE-AIS method improves upon existing energy-based methods such as MINE-DV and MINE-F by directly optimizing a tighter lower bound on MI. MINE-AIS uses MCMC sampling to estimate gradients for training and Multi-Sample AIS for evaluating the bound. Our methods are particularly suitable for evaluating MI in deep generative models, since explicit forms of the marginal or joint densities are often available. We evaluate our bounds on estimating the MI of VAEs and GANs trained on the MNIST and CIFAR datasets, and showcase significant gains over existing bounds in these challenging settings with high ground truth MI.
|
['Alireza Makhzani', 'Roger Grosse', 'Greg Ver Steeg', 'Marzyeh Ghassemi', 'Sicong Huang', 'Rob Brekelmans']
|
2023-03-13
|
improving-mutual-information-estimation-with
|
https://openreview.net/forum?id=T0B9AoM_bFg
|
https://openreview.net/pdf?id=T0B9AoM_bFg
|
iclr-2022-4
|
['mutual-information-estimation']
|
['methodology']
|
[ 3.05244207e-01 8.56346339e-02 -3.90626162e-01 -2.09212765e-01
-1.41828251e+00 -5.73676109e-01 5.08842289e-01 -1.89283371e-01
-4.52855289e-01 1.11087251e+00 -9.92791951e-02 -3.54892462e-01
-2.75476754e-01 -7.75342405e-01 -1.13445985e+00 -1.00819242e+00
-8.49024057e-02 8.66799414e-01 3.59772108e-02 1.89879745e-01
1.32294476e-01 2.20256567e-01 -1.25192130e+00 -3.03321242e-01
8.86918843e-01 1.08411300e+00 2.06693888e-01 9.27318156e-01
8.48784596e-02 5.01306176e-01 -4.87641633e-01 -3.27692807e-01
9.09299031e-02 -7.69379377e-01 -7.92824626e-01 -2.47805804e-01
2.14670986e-01 -4.62809741e-01 -2.00494900e-01 1.11403000e+00
2.21650437e-01 9.74393934e-02 1.34192324e+00 -1.43197238e+00
-3.67741466e-01 7.10661709e-01 -8.11424315e-01 4.08865273e-01
-1.53642282e-01 -3.10060680e-01 1.20027542e+00 -5.51527143e-01
3.73501688e-01 9.82539475e-01 6.31956041e-01 5.25986671e-01
-1.40074909e+00 -6.72178388e-01 1.62010238e-01 3.29472631e-01
-1.64751220e+00 -3.65824878e-01 5.47067463e-01 -3.68252754e-01
8.01773667e-01 1.63626030e-01 3.45809221e-01 1.02501893e+00
1.37460932e-01 1.14288712e+00 9.35812116e-01 -6.29905343e-01
4.72699791e-01 2.14592174e-01 2.77423233e-01 6.94134712e-01
4.18804467e-01 4.51127347e-03 -4.89886850e-01 -5.10261774e-01
9.56864238e-01 -1.42973825e-01 -5.16213536e-01 -3.68594795e-01
-8.76293600e-01 9.47439671e-01 1.85321003e-01 8.05924162e-02
-1.49766698e-01 5.99653244e-01 5.25734387e-02 -5.42437509e-02
6.37434721e-01 -2.26344988e-02 -5.15582800e-01 -4.36288625e-01
-1.13842809e+00 1.38942078e-01 1.02874708e+00 9.21363413e-01
9.05983448e-01 -1.64826974e-01 -1.79492980e-01 7.02068746e-01
4.62703347e-01 7.57501006e-01 1.13717085e-02 -1.20753598e+00
5.33130467e-01 -2.28899837e-01 3.68327916e-01 -4.76222992e-01
7.24038631e-02 -5.64629138e-01 -8.96935463e-01 -1.08246982e-01
6.18407786e-01 -1.94441229e-01 -1.01613104e+00 2.38221717e+00
2.82224417e-01 3.98778707e-01 -8.50792229e-02 6.10662639e-01
1.63585067e-01 8.71726513e-01 -2.69179881e-01 -3.89052957e-01
1.05103183e+00 -7.28080571e-01 -5.60487986e-01 -2.23186702e-01
3.55523944e-01 -1.75190896e-01 9.35050607e-01 4.50090945e-01
-1.29149461e+00 -2.61110440e-02 -1.06199229e+00 -3.59269045e-02
9.96021777e-02 -8.28404576e-02 7.29396343e-01 7.69289613e-01
-1.08846962e+00 8.23664248e-01 -1.21345222e+00 3.17156054e-02
5.62377095e-01 3.94633472e-01 6.96500093e-02 -1.30275398e-01
-8.74748945e-01 5.80252469e-01 1.81590751e-01 9.39058810e-02
-1.04075444e+00 -7.85273850e-01 -7.60446548e-01 2.00174063e-01
3.44121009e-01 -8.64107370e-01 1.34811103e+00 -8.41540456e-01
-1.50301301e+00 3.79638553e-01 -4.37538654e-01 -5.05174935e-01
3.80723327e-01 -1.06906526e-01 2.21050918e-01 3.13029647e-01
-1.45080656e-01 4.20673519e-01 7.47234762e-01 -1.11442780e+00
-1.76472932e-01 -3.63015711e-01 6.19879812e-02 9.55285728e-02
-2.32578605e-01 -4.75912869e-01 -5.44494450e-01 -2.83886075e-01
9.31912754e-03 -8.30454111e-01 -1.05678111e-01 -9.71600227e-03
-5.74144304e-01 -7.46235177e-02 2.91637719e-01 -4.72391814e-01
1.09567344e+00 -1.79597855e+00 2.40977600e-01 3.97591203e-01
3.15329403e-01 -1.38979927e-01 1.94885477e-01 2.29751229e-01
4.05254930e-01 3.51716429e-01 -6.03114307e-01 -4.55274522e-01
3.27370107e-01 4.03805912e-01 -7.27139115e-02 5.21911800e-01
-5.70336431e-02 9.21548307e-01 -7.83000469e-01 -5.07356048e-01
3.50814648e-02 7.16274142e-01 -7.34170020e-01 5.93210123e-02
-1.50657058e-01 2.85666823e-01 -4.48593020e-01 1.41125157e-01
8.05501342e-01 -7.51399100e-01 1.98087022e-01 -2.30982438e-01
3.88731360e-01 3.23206723e-01 -1.05168819e+00 1.45837224e+00
-7.09702373e-01 5.70225894e-01 2.14447737e-01 -1.22934234e+00
1.83755517e-01 2.38094434e-01 3.28464150e-01 -2.08625525e-01
2.61524469e-01 6.49159327e-02 -1.91903010e-01 9.38260704e-02
6.15564808e-02 -4.76441503e-01 -5.04030176e-02 5.60952783e-01
1.75925061e-01 8.88298005e-02 2.11639583e-01 5.88283360e-01
1.18741298e+00 9.15127173e-02 2.20235795e-01 -4.32229459e-01
7.92157650e-02 -5.66405118e-01 4.17102993e-01 1.00050414e+00
-4.40291129e-02 7.19916046e-01 6.52580321e-01 4.11773771e-01
-1.02543294e+00 -1.63929880e+00 -4.86354381e-01 6.20678365e-01
1.54692516e-01 -2.05582261e-01 -9.82437909e-01 -6.44410610e-01
-3.94207299e-01 7.99722135e-01 -7.58652031e-01 -2.88842600e-02
-6.99187070e-03 -1.30465186e+00 2.62944132e-01 5.55921257e-01
5.09952366e-01 -3.56928080e-01 -2.64412940e-01 1.13540873e-01
-3.73070389e-01 -1.02511334e+00 -7.10030794e-01 4.41503853e-01
-8.34170341e-01 -9.77150083e-01 -7.91581511e-01 -2.58604258e-01
5.40448129e-01 -5.96316867e-02 1.28688657e+00 -1.70255348e-01
-1.29338518e-01 5.60464680e-01 4.86086234e-02 -2.65574276e-01
-3.62826169e-01 1.14540458e-01 -1.97082162e-01 -2.36368567e-01
1.13173641e-01 -7.96127439e-01 -7.46903419e-01 1.92747593e-01
-8.90338302e-01 9.28186998e-02 6.25993669e-01 9.13124025e-01
7.68847764e-01 -9.22690555e-02 5.57457805e-01 -8.74214172e-01
1.83744341e-01 -7.89770722e-01 -7.39721477e-01 2.72771955e-01
-7.48116076e-01 6.45955682e-01 3.78534079e-01 -4.12453741e-01
-1.09697735e+00 -2.68464327e-01 -4.28573519e-01 -5.35322547e-01
1.63063183e-01 4.24696505e-01 -3.28044474e-01 1.85999602e-01
3.68256152e-01 2.52836626e-02 -3.58240634e-01 -4.40447390e-01
3.21001470e-01 5.00996113e-01 6.43563867e-01 -8.92433286e-01
6.15883768e-01 4.85666424e-01 1.38938159e-01 -5.81617951e-01
-1.05790365e+00 -4.08660084e-01 -3.56527656e-01 1.22522868e-01
6.63783669e-01 -7.98869789e-01 -7.67856419e-01 2.95985937e-01
-9.21480119e-01 -5.94352841e-01 -3.10386181e-01 6.16711080e-01
-7.66473174e-01 4.07127887e-01 -8.21012616e-01 -1.18284011e+00
-3.74653190e-01 -1.04160035e+00 1.13371944e+00 3.04235872e-02
1.89161822e-01 -1.28617871e+00 1.34958535e-01 1.05051093e-01
2.13054642e-01 1.19668990e-01 8.71642709e-01 -3.57306451e-01
-7.46478558e-01 -7.59379640e-02 -2.41246656e-01 4.24227357e-01
-1.05494432e-01 -1.11059606e-01 -1.01524389e+00 -1.59684971e-01
-2.06596740e-02 -2.74515152e-01 1.17787933e+00 1.00276220e+00
1.30392432e+00 -4.77446765e-01 -5.01448452e-01 6.97247505e-01
1.64552450e+00 -3.52496617e-02 5.93109906e-01 -2.35033169e-01
5.17790318e-01 -1.58752069e-01 2.69304097e-01 6.70200109e-01
4.55838352e-01 6.30487680e-01 2.59957105e-01 2.35088378e-01
1.47876143e-01 -2.50650018e-01 4.15639102e-01 8.66973639e-01
-5.84885776e-02 -6.04157090e-01 -4.14245248e-01 6.13721132e-01
-1.63022661e+00 -1.01838601e+00 8.01899731e-02 2.54386902e+00
1.20344961e+00 8.70961919e-02 1.03183910e-01 1.28398553e-01
5.62302530e-01 -6.97972849e-02 -9.35381413e-01 -3.16539370e-02
1.09244734e-01 5.39950252e-01 7.67665863e-01 9.27904487e-01
-8.43968332e-01 4.24424499e-01 6.56537151e+00 1.14000988e+00
-4.60654378e-01 5.71285605e-01 9.58389163e-01 -3.07873309e-01
-7.73068964e-01 -6.09936006e-02 -1.01458514e+00 5.55677593e-01
1.23781347e+00 -1.26468986e-01 6.26253545e-01 6.96041882e-01
-2.98586369e-01 -5.24113953e-01 -1.28555715e+00 9.53805447e-01
-1.67991683e-01 -1.20455515e+00 -4.40041155e-01 5.45328557e-01
8.13552380e-01 3.10363472e-01 1.94272473e-01 2.31335163e-01
6.61365628e-01 -8.74206483e-01 5.51448941e-01 4.41227496e-01
8.37374032e-01 -9.83458579e-01 5.60654640e-01 5.93094468e-01
-1.04922652e+00 4.03908879e-01 -4.30663496e-01 8.63662288e-02
1.79411411e-01 1.14845705e+00 -5.61762393e-01 3.46151382e-01
3.92250210e-01 3.79186779e-01 -1.74462760e-03 8.16999674e-01
-1.95137396e-01 1.07627559e+00 -1.02618802e+00 1.72798131e-02
5.23414416e-03 -5.09221852e-01 4.63753253e-01 1.00082445e+00
4.02469456e-01 3.69264744e-02 -1.07361883e-01 1.26669192e+00
-3.96680653e-01 -2.40993664e-01 -1.96090832e-01 -3.32540907e-02
5.88685930e-01 1.04507780e+00 -7.81369567e-01 -3.96053225e-01
-2.47633293e-01 1.20226359e+00 4.01197970e-01 5.79396307e-01
-1.17587054e+00 -2.06958517e-01 7.46121764e-01 -1.60890236e-01
6.28042936e-01 -2.19844177e-01 -1.16181597e-01 -1.27289891e+00
9.73815247e-02 -3.59594405e-01 3.39909315e-01 -3.52779657e-01
-1.25632131e+00 2.71641582e-01 5.04155993e-01 -5.90202212e-01
-7.44509220e-01 -4.51936632e-01 -5.07076085e-01 9.51116025e-01
-1.50663185e+00 -7.43756056e-01 6.36327416e-02 3.71007353e-01
2.64242589e-01 4.43729907e-01 6.56663537e-01 2.50020683e-01
-7.33985364e-01 8.21082473e-01 6.64989412e-01 -1.41055003e-01
8.48496184e-02 -1.45877445e+00 3.95456612e-01 6.99599087e-01
4.67272490e-01 3.55111361e-01 7.37916172e-01 -4.11202252e-01
-1.35351825e+00 -7.01988459e-01 3.07056695e-01 -4.27623123e-01
5.61674714e-01 -5.29050350e-01 -7.70762801e-01 8.50364923e-01
-4.35488112e-02 1.32406920e-01 6.36775613e-01 2.09414244e-01
-1.78502932e-01 1.50293201e-01 -1.19388986e+00 2.71752775e-01
1.09623086e+00 -4.45949763e-01 1.17584206e-02 5.25878131e-01
5.04481018e-01 -1.81019336e-01 -9.31285977e-01 1.95351273e-01
5.27106106e-01 -9.84835625e-01 1.01673329e+00 -2.19018966e-01
2.57610857e-01 4.60412260e-03 -4.89962220e-01 -1.16894507e+00
6.85284138e-02 -5.72809696e-01 -5.85826099e-01 1.24123001e+00
6.01913571e-01 -5.39215326e-01 8.55083048e-01 7.50077426e-01
1.43227264e-01 -1.00718558e+00 -1.05473077e+00 -9.93742049e-01
3.48159999e-01 -8.64095509e-01 3.67949158e-01 4.23569560e-01
-2.03365847e-01 3.51001173e-01 -3.28890145e-01 3.25974613e-01
1.07777023e+00 -1.02256782e-01 4.00207967e-01 -1.09017956e+00
-9.65514600e-01 -3.60646397e-01 -1.64149746e-01 -1.47492564e+00
2.42347479e-01 -8.69311631e-01 1.40307143e-01 -1.52922177e+00
7.68972158e-01 -4.11525100e-01 -3.01858127e-01 7.41421208e-02
-3.19357246e-01 2.94481814e-01 -1.51742280e-01 1.68885395e-01
-5.83031952e-01 7.62938917e-01 8.23902905e-01 -3.41091789e-02
9.75123122e-02 1.34850100e-01 -6.29247069e-01 7.48220742e-01
6.59741819e-01 -4.78575677e-01 -7.57595599e-01 -1.48469910e-01
2.95334488e-01 9.37372446e-02 4.26557004e-01 -9.71327782e-01
1.08729012e-01 -6.36047721e-02 3.61220688e-01 -6.48593307e-01
6.03460550e-01 -4.46099699e-01 9.12422165e-02 1.39545113e-01
-2.94470817e-01 -5.13001084e-01 -1.36930530e-03 9.10179138e-01
2.07451820e-01 -5.85319340e-01 8.61560702e-01 2.08749752e-02
-1.08796142e-01 6.57902956e-01 -2.16673911e-01 5.39426267e-01
5.73606610e-01 6.98540881e-02 -7.58355260e-02 -8.07406485e-01
-5.83532155e-01 -1.55068431e-02 3.39101821e-01 -4.19730395e-01
5.66590965e-01 -1.30605435e+00 -5.72813094e-01 -4.33067307e-02
-2.79090852e-01 1.88149139e-01 1.66613981e-01 1.01628673e+00
-1.01772234e-01 1.14328645e-01 4.60637450e-01 -6.58741653e-01
-8.31897259e-01 4.48442429e-01 3.28703851e-01 -5.17929375e-01
-3.16463470e-01 1.05087793e+00 7.41868079e-01 -2.59067491e-02
1.50500670e-01 -3.21515083e-01 4.82680887e-01 -3.50768209e-01
6.44505739e-01 3.24435741e-01 -8.70361403e-02 -2.24248677e-01
-2.34295726e-01 4.15800363e-01 -1.90485254e-01 -5.99432826e-01
1.14250255e+00 -4.56100076e-01 1.94004163e-01 7.83373594e-01
1.67867255e+00 -1.85278878e-01 -1.59134662e+00 -3.61122072e-01
-2.50347137e-01 -2.65627593e-01 5.25993168e-01 -6.57940567e-01
-1.14737952e+00 1.06826234e+00 5.13527930e-01 -4.16625962e-02
1.14828968e+00 4.39340711e-01 8.55483592e-01 3.49804312e-01
4.80343997e-01 -9.12664413e-01 -1.35748625e-01 1.82463974e-01
4.17403966e-01 -1.08422422e+00 4.07662056e-02 -2.92628974e-01
-3.22272599e-01 7.22071230e-01 1.82142720e-01 1.21615954e-01
8.78647149e-01 6.68336570e-01 -6.22308552e-01 2.34724190e-02
-8.00758302e-01 -2.91920125e-01 4.20489043e-01 4.89712983e-01
3.32759321e-01 1.47973122e-02 -1.73747335e-02 5.90755105e-01
-8.07338804e-02 -9.02016982e-02 4.00326550e-01 7.17149377e-01
-4.57582027e-01 -9.89801586e-01 -5.70857562e-02 7.17095137e-01
-8.01262617e-01 -5.16367912e-01 6.16118722e-02 6.34392738e-01
-4.20605004e-01 8.46581817e-01 2.23441750e-01 -4.52689640e-02
-3.95870656e-01 2.62077618e-03 1.03847313e+00 -2.96904683e-01
2.96325475e-01 1.04978666e-01 -1.36139998e-02 -3.98471385e-01
-4.46858138e-01 -6.79050028e-01 -1.27283990e+00 -5.35296500e-01
-7.46849835e-01 3.44062567e-01 8.83814514e-01 1.32295799e+00
1.88301593e-01 3.16142052e-01 5.88470221e-01 -9.14836049e-01
-7.31932163e-01 -9.17598248e-01 -8.04645956e-01 2.62299785e-03
4.13935006e-01 -7.35522330e-01 -6.98519886e-01 -1.25999808e-01]
|
[7.127038955688477, 3.9796037673950195]
|
c496fdd8-16ae-45a3-a28b-9ead4f19d41b
|
sparsely-constrained-neural-networks-for
|
2011.04336
| null |
https://arxiv.org/abs/2011.04336v2
|
https://arxiv.org/pdf/2011.04336v2.pdf
|
Sparsely constrained neural networks for model discovery of PDEs
|
Sparse regression on a library of candidate features has developed as the prime method to discover the partial differential equation underlying a spatio-temporal data-set. These features consist of higher order derivatives, limiting model discovery to densely sampled data-sets with low noise. Neural network-based approaches circumvent this limit by constructing a surrogate model of the data, but have to date ignored advances in sparse regression algorithms. In this paper we present a modular framework that dynamically determines the sparsity pattern of a deep-learning based surrogate using any sparse regression technique. Using our new approach, we introduce a new constraint on the neural network and show how a different network architecture and sparsity estimator improve model discovery accuracy and convergence on several benchmark examples. Our framework is available at \url{https://github.com/PhIMaL/DeePyMoD}
|
['Remy Kusters', 'Gijs Vermarien', 'Gert-Jan Both']
|
2020-11-09
| null | null | null | null |
['model-discovery']
|
['miscellaneous']
|
[-4.43859175e-02 -4.57042605e-02 -3.89713168e-01 -2.99657822e-01
-7.88730383e-01 -3.30109566e-01 5.68887353e-01 -2.78379053e-01
-1.38414726e-01 8.79403234e-01 2.16321334e-01 -1.11000217e-01
-4.65189040e-01 -4.84349310e-01 -6.80257261e-01 -7.20073700e-01
-3.66970211e-01 2.82700330e-01 -1.09303802e-01 -1.08635955e-01
1.00189492e-01 7.26572573e-01 -1.36288977e+00 3.13317746e-01
5.14038384e-01 1.19319785e+00 6.36002980e-03 5.38743734e-01
2.88962632e-01 7.91676044e-01 -8.92898738e-02 1.64317861e-01
6.02193296e-01 -3.00040841e-01 -5.44772387e-01 -3.27810735e-01
4.90393549e-01 -2.20944390e-01 -8.77210319e-01 8.15898418e-01
3.95681500e-01 2.33517408e-01 6.44434094e-01 -1.21380568e+00
-5.21000326e-01 2.70350069e-01 -4.89789397e-01 6.21774614e-01
-1.54131604e-02 3.41387698e-03 9.31989312e-01 -1.15814662e+00
6.66006744e-01 7.03409731e-01 1.18143392e+00 4.32219535e-01
-1.51592374e+00 -4.96692955e-01 -2.49346886e-02 6.07192516e-02
-1.65332961e+00 -7.13597536e-01 9.11064088e-01 -5.14844179e-01
9.66253042e-01 3.52869749e-01 6.60827398e-01 1.30953264e+00
2.46996462e-01 6.23723090e-01 1.06989372e+00 -4.48496975e-02
3.58098328e-01 -6.74825758e-02 1.47952840e-01 6.33544981e-01
3.73987287e-01 3.34770977e-01 -5.08206487e-01 -5.23699641e-01
1.17623258e+00 1.62899956e-01 -3.51948798e-01 -5.67656815e-01
-8.65428984e-01 1.11808801e+00 3.48553509e-01 5.39475620e-01
-4.93062824e-01 3.39886546e-01 2.13668019e-01 4.34971303e-01
6.28959537e-01 6.70073986e-01 -7.42472112e-01 -1.10939503e-01
-1.27968073e+00 2.88915515e-01 9.99656081e-01 6.92653656e-01
9.35089469e-01 4.16578293e-01 6.32266179e-02 7.45066822e-01
-4.77173291e-02 2.50562102e-01 4.06655312e-01 -1.39332199e+00
-2.23474447e-02 1.87703028e-01 2.58285645e-02 -1.25953960e+00
-5.29185653e-01 -7.46749222e-01 -1.22204614e+00 1.03865094e-01
5.00331104e-01 -5.46591103e-01 -6.62678063e-01 1.55143034e+00
2.80737013e-01 5.38319588e-01 -3.36364418e-01 9.96278942e-01
6.11940682e-01 5.80817759e-01 -4.86718744e-01 -1.45980179e-01
5.79167783e-01 -5.95898569e-01 -3.96485299e-01 1.48341926e-02
6.53364241e-01 -3.69066149e-01 6.60996377e-01 5.40845513e-01
-1.09489048e+00 -2.47306868e-01 -7.27385998e-01 5.22218682e-02
-2.40414217e-01 3.46286483e-02 8.93591940e-01 8.97879675e-02
-1.24825919e+00 1.07677495e+00 -1.02182901e+00 -2.50467002e-01
6.26601994e-01 4.73156691e-01 -4.28815186e-01 -1.02168983e-02
-9.62724268e-01 8.21836352e-01 -1.89977914e-01 2.26206288e-01
-8.15502524e-01 -1.11372828e+00 -8.42900515e-01 -4.58488911e-02
4.45633754e-02 -7.31808186e-01 1.08826089e+00 -1.12987864e+00
-1.20201397e+00 5.93131244e-01 -2.90163100e-01 -7.03549683e-01
3.68020862e-01 -1.69731572e-01 -2.92003155e-01 1.38162121e-01
1.88528776e-01 3.08273852e-01 1.04056191e+00 -1.14337742e+00
-3.71289939e-01 -1.07604660e-01 -3.50763977e-01 -3.23708594e-01
-5.97192869e-02 -7.24946111e-02 5.49585093e-03 -5.75798154e-01
1.90644532e-01 -7.55874753e-01 -7.40685701e-01 1.34434894e-01
-1.97191522e-01 1.27251759e-01 7.35609770e-01 -5.25268853e-01
1.17342222e+00 -2.05272555e+00 1.93717957e-01 4.56156552e-01
5.42383611e-01 -2.46607792e-02 -2.37165675e-01 4.78634447e-01
-3.70121568e-01 -2.27107219e-02 -5.85789740e-01 -3.53818148e-01
-3.40519011e-01 1.39569923e-01 -3.50958854e-01 9.71502542e-01
5.26132226e-01 7.48025179e-01 -7.21418500e-01 -1.00249946e-01
3.37389708e-02 9.42574620e-01 -5.73908389e-01 6.00857250e-02
1.02615925e-02 6.93539500e-01 -5.12603462e-01 6.05573654e-01
5.62879264e-01 -6.57723486e-01 -1.18603297e-02 -1.09686755e-01
-1.40813902e-01 1.12952158e-01 -1.37886822e+00 1.82144475e+00
-4.38875794e-01 8.34374905e-01 4.13373709e-01 -1.36967874e+00
9.65937078e-01 3.07514429e-01 1.12458122e+00 -4.66274142e-01
2.36201584e-01 3.78841430e-01 -1.88720729e-02 -5.26300728e-01
1.48620293e-01 -4.43289056e-02 2.74962753e-01 2.08599702e-01
9.29425210e-02 7.55747110e-02 1.01499045e-02 -6.61079213e-02
1.57201266e+00 1.90595746e-01 2.02087358e-01 -6.18690372e-01
2.84998000e-01 3.42995286e-01 6.38376951e-01 7.42249012e-01
9.64586437e-02 9.05317485e-01 4.71443117e-01 -8.01696956e-01
-1.27665865e+00 -6.62369370e-01 -5.63205123e-01 7.27464795e-01
-4.07653481e-01 -3.18879396e-01 -3.93159926e-01 -2.52822429e-01
2.07160577e-01 3.09213191e-01 -8.98624420e-01 -3.00740004e-02
-7.93347299e-01 -8.06560576e-01 5.15519559e-01 4.11710650e-01
-6.24378920e-02 -8.00957024e-01 -5.51931262e-01 5.46647832e-02
1.91279531e-01 -9.60682571e-01 -1.16228834e-01 6.42541766e-01
-9.99569297e-01 -9.75039899e-01 -6.54425681e-01 -5.87083995e-01
7.41004825e-01 -2.06991564e-02 1.08083844e+00 7.85180647e-03
-5.62964320e-01 3.29568565e-01 -5.25455549e-02 7.03982338e-02
5.88300312e-03 2.94515997e-01 2.11707696e-01 1.64302662e-01
2.72628069e-01 -1.17411351e+00 -5.29973924e-01 -1.03205882e-01
-7.54005909e-01 -1.88431352e-01 3.57720286e-01 8.51369083e-01
7.80530989e-01 -2.86410451e-01 4.48649704e-01 -6.77987635e-01
5.99006176e-01 -1.07676089e+00 -7.62316108e-01 -3.84508520e-01
-3.82995784e-01 2.08097011e-01 7.06182122e-01 -5.39658070e-01
-4.90893692e-01 4.62203443e-01 -1.25445858e-01 -1.04611385e+00
-2.07287088e-01 8.33905995e-01 4.22384262e-01 -3.36505502e-01
8.29630613e-01 2.03631192e-01 2.46294782e-01 -7.73087800e-01
5.23664095e-02 5.81484251e-02 3.61227214e-01 -5.97725391e-01
8.15574884e-01 7.04283535e-01 4.71552491e-01 -9.35928524e-01
-8.78556132e-01 -5.79598248e-01 -8.51128101e-01 1.77010089e-01
2.44233474e-01 -1.05475056e+00 -5.20002723e-01 1.06679730e-01
-1.03308368e+00 -6.40496254e-01 -7.32132673e-01 6.20454967e-01
-5.70456803e-01 3.62914726e-02 -5.00642538e-01 -6.26893759e-01
-1.31374031e-01 -6.19825959e-01 9.38928664e-01 -1.51900247e-01
-4.98508394e-01 -1.10262299e+00 5.64138889e-01 -2.80574977e-01
7.77457595e-01 5.15517294e-01 4.50099111e-01 -6.59231842e-01
-5.00304878e-01 -3.31115097e-01 -1.23628333e-01 3.14454645e-01
1.54161537e-02 8.93384889e-02 -9.94371235e-01 -3.08271050e-01
3.38756710e-01 -1.56926885e-01 9.84823883e-01 9.42122936e-01
1.24013805e+00 -4.51210082e-01 -2.77070910e-01 1.26499486e+00
1.59898293e+00 -3.47679973e-01 3.83792430e-01 1.11574046e-01
7.42685080e-01 3.52941990e-01 5.55797108e-02 6.75898671e-01
4.47163172e-02 4.14062172e-01 2.98857510e-01 -2.73253798e-01
1.01515390e-01 -5.08954674e-02 3.42205673e-01 7.72788882e-01
-1.99826047e-01 2.94618517e-01 -1.04513276e+00 7.09871233e-01
-1.94646299e+00 -1.10097182e+00 -3.15709382e-01 1.94129729e+00
6.93377376e-01 -2.52585351e-01 2.53594041e-01 -5.17113460e-03
3.82508159e-01 3.12396854e-01 -7.19507933e-01 -2.35146154e-02
-3.37794036e-01 4.51616377e-01 7.02925026e-01 6.54982150e-01
-1.17168498e+00 5.98600566e-01 7.13568878e+00 5.03675282e-01
-1.44437742e+00 2.52616644e-01 5.48441470e-01 -5.60200036e-01
-3.04673731e-01 -9.52984616e-02 -9.14990664e-01 2.20776811e-01
1.09911025e+00 -1.52955696e-01 5.29109538e-01 8.72420073e-01
5.46729088e-01 2.11461827e-01 -1.18976319e+00 9.20013607e-01
-1.11988969e-01 -1.74249160e+00 -4.03007865e-01 1.95074394e-01
8.57146502e-01 8.16766500e-01 2.01499328e-01 -8.06177184e-02
3.21938008e-01 -1.45419002e+00 4.06263262e-01 7.86724865e-01
8.68729353e-01 -3.42650145e-01 3.40523958e-01 3.00844163e-01
-1.04236794e+00 -1.51689753e-01 -3.66368175e-01 -4.47936952e-01
2.08511599e-03 8.65348876e-01 -4.29785460e-01 6.59935847e-02
8.00794482e-01 1.21429110e+00 -2.87715524e-01 1.13753462e+00
3.27158451e-01 7.83895373e-01 -6.77759588e-01 2.96101540e-01
3.21547031e-01 -2.81440914e-01 7.41728008e-01 1.24971414e+00
5.76483786e-01 4.29273769e-02 1.67682901e-01 1.13559377e+00
1.59344986e-01 -8.66372809e-02 -1.04454541e+00 6.41764998e-02
2.84585565e-01 1.28074598e+00 -5.02965808e-01 4.45598140e-02
-5.04504561e-01 5.98028660e-01 3.79175842e-01 7.59644568e-01
-6.99501336e-01 -7.05576763e-02 9.59537268e-01 4.85567659e-01
5.98225415e-01 -5.87742150e-01 -4.59526062e-01 -1.41314030e+00
1.48498401e-01 -9.37307000e-01 2.53544718e-01 -4.90847379e-01
-1.38895690e+00 5.00315964e-01 -4.09966409e-02 -1.12254739e+00
-2.58485258e-01 -6.00964248e-01 -6.27784491e-01 9.02859151e-01
-1.43076456e+00 -7.84624636e-01 -1.46000981e-01 7.44173408e-01
3.60277563e-01 -2.55206734e-01 7.87749708e-01 3.42022777e-01
-6.26871824e-01 1.38660610e-01 5.68960905e-01 1.43524140e-01
2.07176730e-01 -9.11941409e-01 3.27795953e-01 7.29990542e-01
2.48019740e-01 6.02473617e-01 7.87707567e-01 -4.88364577e-01
-1.53661668e+00 -9.95671451e-01 6.34299755e-01 -6.20818794e-01
1.02970254e+00 -2.56983787e-01 -1.12482238e+00 7.91464090e-01
-5.39231785e-02 5.58717310e-01 5.72851181e-01 1.97979182e-01
-2.94723332e-01 -5.81500344e-02 -9.92108643e-01 1.83148935e-01
9.65349317e-01 -5.79803944e-01 -1.83340341e-01 3.80993128e-01
3.78926158e-01 -2.92641014e-01 -1.03235900e+00 4.82434899e-01
4.89271194e-01 -8.47793460e-01 1.01601768e+00 -8.21366012e-01
5.88106751e-01 -2.04478458e-01 -2.67957836e-01 -1.16373169e+00
-5.62721848e-01 -9.90583479e-01 -6.59757972e-01 4.26604301e-01
5.44764519e-01 -7.41701841e-01 9.88967419e-01 5.53935409e-01
-1.33526161e-01 -1.28235328e+00 -1.08468866e+00 -8.01063776e-01
3.57395917e-01 -4.30244178e-01 3.03143799e-01 1.16929960e+00
-2.16169760e-01 1.28573403e-01 -4.79303688e-01 2.35770956e-01
5.58271229e-01 6.06093109e-02 5.39396107e-01 -1.36633170e+00
-3.69316310e-01 -5.51943541e-01 -2.79206842e-01 -9.49190319e-01
2.50953943e-01 -9.66141939e-01 -2.14677587e-01 -1.21775019e+00
-1.42204747e-01 -5.89855552e-01 -3.43122721e-01 4.34721529e-01
4.51549321e-01 2.24138677e-01 -1.82258636e-01 4.27684665e-01
-2.46422723e-01 5.92547953e-01 9.59841788e-01 1.07699171e-01
-1.95521504e-01 -2.00482477e-02 -6.21420145e-01 7.37185001e-01
9.84768093e-01 -8.30615759e-01 -1.91204458e-01 -4.81626511e-01
3.11450899e-01 5.65145612e-02 7.50640810e-01 -1.02942693e+00
3.10009927e-01 -3.54721338e-01 4.99047101e-01 -2.11210281e-01
5.29973447e-01 -8.42372298e-01 3.63225847e-01 3.49593133e-01
-3.54971260e-01 1.55915067e-01 2.44211987e-01 3.95835817e-01
-2.40284875e-01 -1.61359176e-01 7.58570671e-01 -8.83157328e-02
-4.16521072e-01 6.44544840e-01 -3.29284430e-01 2.18123496e-01
6.58714116e-01 -1.03039049e-01 1.03072152e-02 -4.19062585e-01
-8.83724093e-01 -3.70466784e-02 4.59478676e-01 9.67460796e-02
5.28257370e-01 -1.30643451e+00 -9.28113580e-01 4.51517761e-01
-4.00524378e-01 8.16813484e-02 -1.91523209e-02 1.25026512e+00
-3.35663378e-01 2.76059687e-01 -2.12622702e-01 -6.40708864e-01
-8.04149449e-01 3.20720702e-01 7.52418995e-01 -2.11676117e-02
-8.83150935e-01 8.50546181e-01 -1.16454780e-01 -3.04660916e-01
1.52415633e-01 -3.91316265e-01 8.54010358e-02 -7.47809708e-02
2.75881052e-01 3.75236243e-01 -7.42577985e-02 -7.68003702e-01
-2.70294636e-01 5.31554282e-01 3.33765924e-01 9.83374342e-02
1.93689406e+00 2.34513611e-01 -1.46919772e-01 6.87348962e-01
1.60796297e+00 -1.51973128e-01 -1.46329558e+00 -2.62853324e-01
-1.30292252e-02 -3.76613498e-01 2.33145773e-01 -2.45349273e-01
-1.33947110e+00 6.86172009e-01 3.04551631e-01 3.80372673e-01
8.34726036e-01 -3.29801030e-02 5.61011195e-01 7.21656561e-01
3.97550799e-02 -8.33415866e-01 -2.01079637e-01 7.72290826e-01
1.01719797e+00 -1.30366886e+00 1.46266118e-01 -6.86168373e-02
-2.77727723e-01 1.04699183e+00 2.54347652e-01 -8.86911094e-01
1.27364969e+00 5.34859478e-01 -1.78501844e-01 -5.41489840e-01
-8.52276087e-01 -4.29222398e-02 3.07193696e-01 5.56492031e-01
4.86160278e-01 -3.43212783e-01 7.38889948e-02 5.69003761e-01
-1.31203353e-01 2.94722080e-01 3.81954044e-01 7.79287398e-01
-1.87903598e-01 -7.90709436e-01 -1.67194411e-01 8.07651520e-01
-4.58823830e-01 -3.41157526e-01 -1.91885948e-01 7.54040182e-01
-1.44259080e-01 3.49865705e-01 1.34606123e-01 -1.09883964e-01
5.29032946e-02 -3.85110751e-02 3.32236588e-01 -5.87927818e-01
-4.83529001e-01 -4.07155696e-03 6.94342628e-02 -8.04468572e-01
-4.93758023e-01 -9.31577563e-01 -9.63427126e-01 -3.58540326e-01
-3.86610925e-02 1.60160974e-01 4.55338746e-01 8.28805685e-01
7.07074761e-01 9.63298976e-02 6.35208666e-01 -1.09119189e+00
-6.01931095e-01 -6.76268995e-01 -6.38162374e-01 5.12000620e-02
8.08182120e-01 -5.71164250e-01 -7.30248749e-01 3.40867154e-02]
|
[6.63385534286499, 3.5245778560638428]
|
1c38895a-8083-459d-9bdd-33ff62b270e1
|
allo-centric-occupancy-grid-prediction-for
|
2301.04454
| null |
https://arxiv.org/abs/2301.04454v1
|
https://arxiv.org/pdf/2301.04454v1.pdf
|
Allo-centric Occupancy Grid Prediction for Urban Traffic Scene Using Video Prediction Networks
|
Prediction of dynamic environment is crucial to safe navigation of an autonomous vehicle. Urban traffic scenes are particularly challenging to forecast due to complex interactions between various dynamic agents, such as vehicles and vulnerable road users. Previous approaches have used egocentric occupancy grid maps to represent and predict dynamic environments. However, these predictions suffer from blurriness, loss of scene structure at turns, and vanishing of agents over longer prediction horizon. In this work, we propose a novel framework to make long-term predictions by representing the traffic scene in a fixed frame, referred as allo-centric occupancy grid. This allows for the static scene to remain fixed and to represent motion of the ego-vehicle on the grid like other agents'. We study the allo-centric grid prediction with different video prediction networks and validate the approach on the real-world Nuscenes dataset. The results demonstrate that the allo-centric grid representation significantly improves scene prediction, in comparison to the conventional ego-centric grid approach.
|
['Christian Laugier', 'Anne Spalanzani', 'Lukas Rummelhard', 'Rabbia Asghar']
|
2023-01-11
| null | null | null | null |
['video-prediction']
|
['computer-vision']
|
[-2.22995758e-01 1.24417461e-01 1.43808335e-01 -2.86867827e-01
1.11759022e-01 -2.66104043e-01 9.12756681e-01 -1.28670752e-01
-2.22625777e-01 9.72386658e-01 2.45510787e-01 -2.03956679e-01
1.97288021e-02 -1.12621582e+00 -6.45501375e-01 -6.47138298e-01
-1.45832241e-01 6.20060205e-01 1.03491974e+00 -3.15364987e-01
1.65377513e-01 5.91048658e-01 -1.90897131e+00 1.04089849e-01
6.36254132e-01 6.31895721e-01 6.24544144e-01 6.65557265e-01
7.05689192e-02 1.17929542e+00 -9.16750953e-02 -9.19547528e-02
2.87395775e-01 1.56535245e-02 -3.88354093e-01 2.10797414e-01
3.62764567e-01 -2.96542615e-01 -7.57528305e-01 9.85147834e-01
1.93791658e-01 6.47124171e-01 3.56412500e-01 -1.42874801e+00
2.28766635e-01 1.29189059e-01 -4.35206056e-01 3.49788874e-01
2.31186807e-01 2.30675474e-01 3.80653292e-01 -5.12867153e-01
1.08074772e+00 1.38053811e+00 6.67519569e-01 3.59442562e-01
-9.87542987e-01 -6.46298766e-01 4.89421427e-01 1.02722132e+00
-1.56784129e+00 -5.04143059e-01 7.43123651e-01 -5.56178927e-01
8.42398167e-01 3.02902907e-01 9.36119914e-01 8.11095595e-01
6.58906281e-01 4.47990000e-01 8.52178216e-01 1.33275762e-02
4.23664391e-01 1.57145992e-01 5.36864102e-02 4.30643678e-01
7.10543096e-02 3.36211592e-01 -3.71330619e-01 1.23598225e-01
5.39908469e-01 -1.42742498e-02 -4.84041609e-02 -9.00288820e-01
-1.06966352e+00 6.18122935e-01 4.56562102e-01 -1.88379526e-01
-8.40768695e-01 1.26237035e-01 2.83032775e-01 -1.55147329e-01
6.95103347e-01 -1.27368629e-01 3.25358822e-03 -2.84656137e-01
-7.74819553e-01 5.85024834e-01 5.44268608e-01 1.08318841e+00
1.05247700e+00 2.65086234e-01 -7.66434520e-02 4.73383635e-01
9.07816514e-02 3.98303807e-01 1.38024390e-01 -1.13669407e+00
2.00975940e-01 3.42892319e-01 4.17108923e-01 -1.37456298e+00
-6.63709641e-01 -5.10766804e-01 -9.19757783e-01 5.16003728e-01
2.48199925e-01 -4.10745740e-01 -7.88148522e-01 1.58480525e+00
7.35617459e-01 9.97001469e-01 1.52449548e-01 8.14418316e-01
3.92832488e-01 9.46705520e-01 7.58889765e-02 -2.82222539e-01
9.93479490e-01 -1.34240067e+00 -7.75170028e-01 -2.23093048e-01
6.37113690e-01 -4.61938202e-01 3.33817244e-01 1.37383461e-01
-6.48846447e-01 -6.75103366e-01 -7.58568108e-01 4.33732033e-01
-5.48680365e-01 -4.48509127e-01 2.94658840e-01 3.80156428e-01
-1.30279934e+00 1.08836882e-01 -7.74229288e-01 -6.44266546e-01
1.70798913e-01 6.42221943e-02 -2.92003661e-01 -2.56210297e-01
-1.07164466e+00 1.25365758e+00 5.36717832e-01 -2.87274700e-02
-9.23474014e-01 -7.07859516e-01 -9.46072400e-01 -2.01737024e-02
3.69451016e-01 -5.78214169e-01 1.03085721e+00 -7.08192468e-01
-1.15656865e+00 2.13628292e-01 -5.60954034e-01 -9.52569544e-01
8.58117938e-01 -8.53967518e-02 -5.25885940e-01 -3.27338994e-01
1.36773050e-01 7.82496631e-01 3.33484411e-01 -1.49291360e+00
-1.13261127e+00 -6.10107742e-02 3.44430983e-01 7.40341365e-01
3.68269265e-01 -5.13291478e-01 -5.98049998e-01 -1.00600459e-01
-6.08662926e-02 -1.23691571e+00 -7.72552848e-01 -2.33493328e-01
-1.80838406e-01 -3.56314033e-02 1.29215324e+00 -3.49939048e-01
1.08524191e+00 -2.01970124e+00 -4.75435585e-01 1.98694810e-01
2.44798750e-01 2.00791523e-01 5.60413152e-02 5.00652552e-01
6.09325096e-02 -3.72262925e-01 3.61848086e-01 -3.28974992e-01
-1.06452495e-01 4.74439561e-01 -2.68277764e-01 5.58371425e-01
-6.25523567e-01 6.55475259e-01 -1.00852776e+00 -4.32030082e-01
9.65442359e-01 7.54904866e-01 -6.90059483e-01 -1.05134495e-01
-1.72943965e-01 5.36044478e-01 -4.25773501e-01 -2.73219720e-02
1.00316727e+00 1.51806146e-01 1.11979000e-01 7.43990019e-02
-6.89044893e-01 2.97199041e-02 -1.15363181e+00 1.14465070e+00
-3.58794898e-01 1.10304010e+00 -9.09493715e-02 -5.83650649e-01
7.09343076e-01 1.76448226e-01 4.93421167e-01 -1.03486872e+00
-1.83973402e-01 -2.54073918e-01 -1.78406611e-01 -1.83405161e-01
9.72895920e-01 2.26326808e-01 1.56636700e-01 -1.15360804e-02
-7.44144917e-01 1.44185036e-01 3.45811546e-01 1.05666548e-01
1.01497984e+00 5.37071526e-02 2.63106376e-01 -2.66640395e-01
5.76152503e-01 4.73560035e-01 8.24335277e-01 8.80301595e-01
-4.51008976e-01 4.05706286e-01 2.29665726e-01 -1.04469156e+00
-1.31895411e+00 -7.86097527e-01 3.43760513e-02 5.92239201e-01
8.29838514e-01 -3.36374104e-01 -8.23557317e-01 -2.99776345e-01
-2.44795740e-01 9.88930762e-01 -5.86504161e-01 1.09089233e-01
-7.16438353e-01 -5.89400768e-01 -1.43664867e-01 3.27070132e-02
7.10498214e-01 -7.65409708e-01 -9.08347905e-01 5.55347323e-01
-3.17687154e-01 -1.38730145e+00 -6.48219883e-02 -5.36876380e-01
-3.17208499e-01 -1.14833260e+00 -1.62166193e-01 -3.20602357e-01
5.30323744e-01 9.05071497e-01 8.37012649e-01 -1.29309356e-01
1.02367505e-01 2.58372575e-01 -3.09916139e-01 -3.89432847e-01
-3.22764874e-01 -1.96568683e-01 2.46413648e-01 2.54921019e-01
3.60763550e-01 -5.99708915e-01 -8.37342680e-01 5.90301692e-01
-3.17635685e-01 7.81332374e-01 -3.73907946e-02 3.92311931e-01
6.28513277e-01 5.59223711e-01 3.20077837e-01 -8.88585150e-01
1.50925681e-01 -8.35899055e-01 -8.95331621e-01 -1.73814982e-01
-4.61023152e-01 -5.14753997e-01 5.71017981e-01 5.67310713e-02
-1.17784894e+00 2.51851022e-01 -3.94378304e-02 -4.77500558e-01
-3.75890136e-01 8.78697485e-02 -2.64191460e-02 -1.53197005e-01
3.93360227e-01 2.74123341e-01 -2.12517515e-01 -6.61275908e-02
1.49044991e-01 3.06733936e-01 4.80513871e-01 6.05893992e-02
8.50520909e-01 7.97052920e-01 2.36987859e-01 -1.10770941e+00
-4.49850798e-01 -4.92817163e-01 -6.14533603e-01 -9.28191245e-01
7.92781532e-01 -1.18923128e+00 -6.14190817e-01 4.18531179e-01
-1.32318652e+00 -5.30141056e-01 -2.20756799e-01 5.79071820e-01
-7.85483360e-01 3.46278727e-01 -1.57556996e-01 -8.70868385e-01
1.96530044e-01 -1.06139970e+00 6.29498065e-01 2.37487644e-01
-5.94901554e-02 -1.08881307e+00 4.65818822e-01 8.88980404e-02
5.61231732e-01 4.43796039e-01 4.11094546e-01 -1.29923373e-01
-1.06365275e+00 -9.11449268e-02 -1.54264539e-01 -3.65127295e-01
-1.67235322e-02 -1.51350617e-01 -8.37451935e-01 -1.74168423e-01
-2.26558939e-01 6.17535710e-01 5.61935723e-01 4.88043785e-01
8.23347569e-01 -3.28857183e-01 -7.33100474e-01 4.61940080e-01
1.35979366e+00 5.65547347e-01 9.27429795e-01 7.06383944e-01
6.85460746e-01 6.39941275e-01 1.07600594e+00 5.60517251e-01
8.71924341e-01 1.03458405e+00 7.83594131e-01 3.71219702e-02
-2.09681407e-01 -2.16223240e-01 7.80292004e-02 4.36231345e-01
-1.11889057e-01 -5.99392593e-01 -1.07690799e+00 7.78577387e-01
-2.36364174e+00 -1.37187314e+00 -3.63095313e-01 1.97062278e+00
-2.58311927e-01 1.06850699e-01 -2.84647532e-02 -2.27027774e-01
9.06544983e-01 3.85410875e-01 -5.09802341e-01 -2.50480890e-01
-7.79884160e-02 -8.22896242e-01 7.63652921e-01 9.78316188e-01
-1.13385332e+00 1.35825014e+00 6.31203508e+00 8.70589852e-01
-9.44854140e-01 3.50691259e-01 4.77099687e-01 9.09374561e-03
-5.09791896e-02 9.01342258e-02 -9.48893011e-01 6.49574161e-01
1.08981872e+00 -4.30161297e-01 2.43942603e-01 8.44793558e-01
9.33009624e-01 -4.39611375e-01 -5.68307221e-01 8.78615201e-01
-1.47871464e-01 -1.61699402e+00 4.25531194e-02 3.11976433e-01
8.68460715e-01 3.89480233e-01 -1.54151231e-01 1.79382712e-01
5.84942341e-01 -6.51868820e-01 7.23206639e-01 8.24849784e-01
1.76975995e-01 -8.75320375e-01 7.55739152e-01 6.38078094e-01
-1.47623205e+00 1.29581243e-01 -5.48156381e-01 -3.62394869e-01
6.91559732e-01 2.69758224e-01 -1.12398863e+00 4.16382283e-01
5.69873512e-01 8.82435381e-01 -4.12879258e-01 1.43854463e+00
4.39927667e-01 2.77098715e-01 -3.30138713e-01 2.83357412e-01
5.92815995e-01 -3.56849164e-01 1.04208803e+00 1.04782176e+00
4.36931461e-01 1.44773647e-01 3.74810845e-01 2.16737494e-01
6.12058163e-01 -4.69324589e-02 -1.10963416e+00 8.17184627e-01
4.73450273e-01 8.60776842e-01 -5.58885038e-01 -5.47399104e-01
-4.36968625e-01 5.47457457e-01 1.73510432e-01 6.03957534e-01
-9.38456476e-01 2.92762190e-01 1.09348130e+00 4.94536638e-01
3.06748152e-01 -4.43699062e-01 2.58589014e-02 -8.39229882e-01
-3.48781914e-01 -2.15929389e-01 -4.41872189e-03 -9.65798676e-01
-4.54507291e-01 9.34529305e-01 1.98625848e-01 -1.59544539e+00
-5.12589753e-01 -1.27237543e-01 -6.58256114e-01 5.68561018e-01
-1.68836987e+00 -1.17240703e+00 -6.66965663e-01 6.33422911e-01
9.14744139e-01 -2.21163273e-01 4.31326210e-01 2.36025766e-01
-4.12631363e-01 -1.03490390e-01 4.20965135e-01 -4.38142121e-01
1.72528520e-01 -8.93002152e-01 5.38601995e-01 1.07744682e+00
-2.60896623e-01 -2.25199372e-01 1.24511564e+00 -8.17257166e-01
-9.71715748e-01 -1.64097941e+00 8.52260590e-01 -3.31662089e-01
4.17524457e-01 -1.47761449e-01 -7.11096525e-01 8.71874034e-01
2.36080915e-01 1.57216322e-02 1.75862208e-01 -2.30434895e-01
1.98704660e-01 -2.59301662e-01 -9.81837273e-01 9.90111411e-01
9.65227306e-01 -1.00607388e-01 -3.62819876e-03 3.64438802e-01
4.77770776e-01 -4.83267695e-01 -2.97547638e-01 4.13087308e-01
6.66659236e-01 -1.34146965e+00 8.24122310e-01 -7.74600506e-02
-1.36472821e-01 -5.95054209e-01 -2.08268330e-01 -1.42417991e+00
-7.15937793e-01 -3.45299035e-01 3.16298082e-02 7.44852602e-01
9.17032808e-02 -7.08773077e-01 1.19421816e+00 6.80702865e-01
-3.92012477e-01 -3.27095836e-01 -1.25697386e+00 -5.42281330e-01
-3.60099941e-01 -6.19226038e-01 7.87011385e-01 6.75370097e-01
-2.63536990e-01 -1.53084807e-02 -8.95579338e-01 5.89883983e-01
6.56417251e-01 -1.54843077e-01 1.31306183e+00 -1.12777054e+00
3.56689453e-01 -1.34006277e-01 -1.03326023e+00 -1.03921056e+00
2.13782251e-01 -2.57865101e-01 5.86512387e-02 -1.64675975e+00
9.70960632e-02 -6.33264065e-01 -1.46796238e-02 -3.63984890e-02
2.23011807e-01 3.15501302e-01 2.32148305e-01 1.38126448e-01
-9.97984946e-01 7.50750005e-01 1.17816067e+00 1.29481107e-02
-1.96364567e-01 -6.10142061e-03 5.95426597e-02 8.81808579e-01
9.17313516e-01 -2.61976391e-01 -7.40517020e-01 -2.96232164e-01
6.16526790e-02 1.64823636e-01 4.71225679e-01 -1.63421154e+00
6.78069472e-01 -4.45658535e-01 3.31488326e-02 -1.21272898e+00
8.78870904e-01 -1.14671946e+00 1.04201448e+00 4.48058099e-01
2.31951773e-01 1.48485199e-01 3.80355388e-01 8.77549529e-01
-2.73453712e-01 2.27962092e-01 7.09169149e-01 -1.44446298e-01
-1.50118268e+00 4.94636565e-01 -9.11047518e-01 -5.08159995e-01
1.61510658e+00 -7.11880505e-01 -3.72088820e-01 -7.45430052e-01
-7.73089170e-01 6.16916180e-01 7.30338216e-01 5.99732459e-01
5.39106846e-01 -1.29921794e+00 -5.80792129e-01 2.18498498e-01
-5.31729348e-02 -4.36061248e-02 9.24827278e-01 6.96725070e-01
-8.60341489e-01 6.40060961e-01 -4.10979837e-01 -7.19601452e-01
-1.42518198e+00 5.93273163e-01 4.94322568e-01 -2.62772650e-01
-9.72981811e-01 1.10530831e-01 9.38112497e-01 -3.19468409e-01
-1.10272668e-01 1.78391695e-01 -6.07144773e-01 -7.95574784e-02
6.30347192e-01 6.24977767e-01 -2.42168158e-01 -1.44893444e+00
-1.48088172e-01 5.10513246e-01 -8.79228786e-02 1.45564703e-02
1.10841405e+00 -1.05813313e+00 1.65770426e-01 4.02312279e-01
6.05456710e-01 -1.79457083e-01 -1.67565048e+00 -2.61731029e-01
-3.68922949e-01 -6.56945765e-01 1.54758438e-01 -2.32223973e-01
-8.80389631e-01 5.80464363e-01 8.63723993e-01 1.48076624e-01
5.77943981e-01 -4.50880915e-01 6.62966192e-01 2.52906680e-01
9.08378899e-01 -1.03996992e+00 -5.92021942e-01 9.52209830e-01
6.22308791e-01 -9.00340617e-01 -1.58946633e-01 -6.51575625e-01
-6.58909678e-01 8.17531645e-01 9.30088043e-01 -7.75647983e-02
8.32481205e-01 -1.23125643e-01 1.44943912e-02 -1.13357469e-01
-1.08733654e+00 -3.68350148e-01 -1.63307548e-01 8.14607084e-01
-4.10115331e-01 2.34706864e-01 2.33282670e-01 -1.44389495e-01
-4.86200541e-01 -1.23900428e-01 8.36201012e-01 6.98559761e-01
-6.05368018e-01 -6.12521708e-01 -3.07792157e-01 3.20082784e-01
1.03121765e-01 2.17182174e-01 3.32447946e-01 8.44847083e-01
3.30885082e-01 9.80588377e-01 6.16261303e-01 -4.21624124e-01
2.66903967e-01 -2.40156189e-01 -7.45207295e-02 -2.80466795e-01
-1.20248448e-03 -9.92508158e-02 3.64491373e-01 -8.13923717e-01
-3.44432890e-01 -9.12824452e-01 -9.05459583e-01 -8.63937914e-01
2.61904653e-02 2.54760981e-01 5.60090840e-01 1.00345778e+00
6.41981363e-01 5.48688591e-01 5.50454617e-01 -1.31857824e+00
4.80806857e-01 -6.94618404e-01 -5.60350358e-01 1.18007325e-01
5.23489714e-01 -9.26983654e-01 -2.62470301e-02 -1.54280141e-01]
|
[5.9248223304748535, 0.8181715607643127]
|
62efb6ee-8efd-41a8-9b0b-8cb2a3653793
|
precise-stock-price-prediction-for-optimized
|
2203.01326
| null |
https://arxiv.org/abs/2203.01326v1
|
https://arxiv.org/pdf/2203.01326v1.pdf
|
Precise Stock Price Prediction for Optimized Portfolio Design Using an LSTM Model
|
Accurate prediction of future prices of stocks is a difficult task to perform. Even more challenging is to design an optimized portfolio of stocks with the identification of proper weights of allocation to achieve the optimized values of return and risk. We present optimized portfolios based on the seven sectors of the Indian economy. The past prices of the stocks are extracted from the web from January 1, 2016, to December 31, 2020. Optimum portfolios are designed on the selected seven sectors. An LSTM regression model is also designed for predicting future stock prices. Five months after the construction of the portfolios, i.e., on June 1, 2021, the actual and predicted returns and risks of each portfolio are computed. The predicted and the actual returns indicate the very high accuracy of the LSTM model.
|
['Saikat Mondal', 'Abhishek Dutta', 'Sidra Mehtab', 'Jaydip Sen']
|
2022-03-02
| null | null | null | null |
['stock-price-prediction']
|
['time-series']
|
[-5.88321924e-01 -6.78011850e-02 -2.93945849e-01 -1.35643795e-01
-4.97955233e-01 -7.40144253e-01 7.02035666e-01 -3.08830440e-01
-3.30553025e-01 8.09370518e-01 5.09418070e-01 -5.94021857e-01
-4.24130797e-01 -1.12198281e+00 -5.31012356e-01 -4.41123039e-01
-3.18002015e-01 1.92317888e-01 -1.05991244e-01 -2.29243953e-02
6.55471921e-01 4.38144445e-01 -1.14205408e+00 2.73219258e-01
3.49913090e-01 1.74735999e+00 3.20035219e-01 2.78702885e-01
-3.27561885e-01 7.33382881e-01 -4.99996722e-01 -8.69019151e-01
7.85945952e-01 1.43739367e-02 -8.58089924e-02 -5.20607173e-01
-3.77266079e-01 -7.71676242e-01 -1.76581562e-01 9.87065673e-01
6.19690344e-02 1.09050246e-02 4.99613702e-01 -6.79758787e-01
-5.65673351e-01 1.19248736e+00 -3.32951277e-01 3.88812780e-01
-4.58453298e-01 -6.04314506e-02 1.44718707e+00 -1.08057272e+00
1.06123284e-01 4.54967737e-01 6.07458472e-01 6.03200272e-02
-8.09371054e-01 -9.30137634e-01 2.56157462e-02 6.62239566e-02
-9.58786607e-01 -2.34956548e-01 6.60433233e-01 -7.16580331e-01
9.58217382e-01 1.39812052e-01 9.64460373e-01 5.96423864e-01
9.02727723e-01 2.96058118e-01 9.94278729e-01 -1.82120025e-01
1.81312457e-01 2.22948536e-01 -2.96862125e-01 -9.78462473e-02
7.71103859e-01 4.52274650e-01 -4.38667148e-01 -1.40937954e-01
5.41008472e-01 3.14562649e-01 2.14523897e-01 2.42649108e-01
-1.23948538e+00 7.47268021e-01 1.80178270e-01 4.98242736e-01
-1.05692446e+00 2.78373629e-01 1.26923978e-01 3.84578466e-01
5.55366039e-01 5.05386531e-01 -7.73354828e-01 5.16888220e-03
-1.20157647e+00 1.98769823e-01 7.00907528e-01 4.93800879e-01
2.66571283e-01 5.71555614e-01 -2.22691670e-01 3.44398856e-01
2.17105433e-01 7.77874768e-01 5.75024843e-01 -9.93445694e-01
6.82985544e-01 3.29040349e-01 6.25086606e-01 -1.10247147e+00
-1.78157404e-01 -9.73316252e-01 -5.59155345e-01 1.24737121e-01
4.20399249e-01 -4.39906150e-01 -4.66554075e-01 1.19455230e+00
-5.86094022e-01 1.83267206e-01 2.54886568e-01 1.62016273e-01
1.00728460e-01 1.07457197e+00 -1.04111105e-01 -3.70635122e-01
9.63798046e-01 -5.99965394e-01 -7.08992660e-01 -3.58855277e-01
1.07271433e-01 -5.10318995e-01 1.81177214e-01 4.17384535e-01
-1.12515378e+00 -1.61720619e-01 -8.07039738e-01 9.29502666e-01
-4.24433887e-01 1.62849322e-01 5.02816379e-01 3.20066124e-01
-7.53137529e-01 1.05600154e+00 -6.04632437e-01 6.85298264e-01
1.31607160e-01 4.15462822e-01 2.64018297e-01 8.85199249e-01
-1.46655571e+00 1.26097322e+00 7.22321808e-01 3.70440423e-01
-7.21923292e-01 -6.18274927e-01 -1.55655384e-01 4.59544063e-01
3.27944867e-02 5.66003174e-02 1.32455420e+00 -1.01087117e+00
-1.24914265e+00 1.12098359e-01 5.53965688e-01 -1.07578039e+00
4.46560532e-01 -2.28479549e-01 -4.37525451e-01 -4.08863783e-01
-1.84384063e-01 -1.72856510e-01 4.71689790e-01 -7.81599045e-01
-1.18234050e+00 -1.21869221e-01 -1.96745783e-01 -1.56119436e-01
-8.47184002e-01 1.38535663e-01 1.82065353e-01 -9.34424460e-01
-8.70109499e-02 -7.02644229e-01 -2.40570217e-01 -7.97216296e-01
-4.02267762e-02 3.26763004e-01 -2.00389978e-02 -1.50938690e+00
1.42685866e+00 -1.69273818e+00 -5.19533873e-01 4.87092644e-01
-1.72393173e-01 3.33990194e-02 2.15979040e-01 3.10224414e-01
-3.85724127e-01 2.29075029e-01 6.32846430e-02 1.54714137e-01
2.99547255e-01 1.26194516e-02 -1.05718887e+00 2.11961627e-01
5.04240319e-02 1.13146508e+00 -4.80664402e-01 2.72732764e-01
2.66995970e-02 -9.92533043e-02 -1.31956980e-01 3.49118114e-01
-2.44976431e-01 8.29741582e-02 -4.49216872e-01 4.72937435e-01
3.92508894e-01 -1.21909395e-01 9.14007351e-02 1.59719706e-01
-6.05361402e-01 5.46465337e-01 -7.58982241e-01 6.15761101e-01
-3.65933955e-01 3.62757593e-01 -5.31912088e-01 -6.54216528e-01
1.19182014e+00 5.15272319e-01 6.75530434e-01 -9.37652230e-01
2.18016584e-03 7.32627094e-01 1.67895451e-01 6.21763356e-02
6.10279024e-01 -4.81726110e-01 -2.08461955e-01 7.89261162e-01
-2.32102811e-01 2.89247245e-01 5.78447059e-02 -5.34585714e-01
6.36801302e-01 7.50577152e-02 2.34929383e-01 -3.05525661e-01
1.06677428e-01 -3.10313672e-01 7.80497849e-01 3.37284565e-01
2.65593618e-01 1.39850259e-01 6.18500292e-01 -6.41457438e-01
-9.37213361e-01 -9.02346194e-01 4.22192477e-02 6.72747254e-01
-7.93383896e-01 4.13523138e-01 -3.76767546e-01 -3.28059763e-01
2.64691740e-01 1.24629271e+00 -6.64538383e-01 8.01477581e-02
-4.54270273e-01 -1.00469685e+00 1.94962680e-01 4.37766403e-01
4.76968110e-01 -1.20528364e+00 -9.30590093e-01 5.49066722e-01
1.48039982e-02 -6.07243299e-01 -2.45379612e-01 2.55241454e-01
-9.13732529e-01 -7.07470357e-01 -1.11388004e+00 -1.62863702e-01
2.60761589e-01 -3.55540514e-01 1.10878742e+00 -4.33360517e-01
6.97956204e-01 -2.58716643e-01 1.04329698e-01 -1.17278183e+00
6.29647300e-02 2.75035501e-02 1.92303956e-02 2.76618004e-01
6.32323399e-02 -3.97007823e-01 -5.50454855e-01 1.35573903e-02
-4.10447806e-01 -1.80156261e-01 7.07285225e-01 6.38757944e-01
4.75357771e-01 4.33731467e-01 8.68688762e-01 -3.33374143e-01
5.69890618e-01 -6.68972611e-01 -1.24450719e+00 4.82336611e-01
-1.08171749e+00 6.94679469e-02 4.24077183e-01 -3.80984724e-01
-1.21744895e+00 -2.32434258e-01 6.94229081e-02 -2.39529327e-01
8.10701609e-01 9.74609554e-01 2.59000033e-01 2.94278294e-01
-1.28231987e-01 3.41846526e-01 -3.96572679e-01 -7.53968954e-01
-9.50460285e-02 4.84523594e-01 6.21506333e-01 -3.21857452e-01
8.76020670e-01 4.84667718e-02 -2.56597728e-01 -8.03385228e-02
-8.75636339e-01 3.25162590e-01 -5.29697061e-01 -4.90815669e-01
3.37402791e-01 -9.00596917e-01 -4.18766111e-01 6.51877403e-01
-8.39385986e-01 -4.30514067e-01 -5.79051614e-01 7.81306863e-01
-4.34694350e-01 -3.14745784e-01 -4.70727891e-01 -1.32036817e+00
-8.73301268e-01 -7.69291878e-01 1.25274703e-01 2.07338795e-01
-3.12712967e-01 -1.10038066e+00 -3.17505263e-02 -7.87028819e-02
8.17793846e-01 4.74038959e-01 8.99588287e-01 -1.00065327e+00
-6.57388210e-01 -5.31805754e-01 -1.48523793e-01 6.72860444e-01
3.84501457e-01 2.12112963e-01 -5.84204972e-01 -1.61258340e-01
4.55618411e-01 3.77600461e-01 7.79105902e-01 7.12778330e-01
6.59663439e-01 -8.28906178e-01 3.33850831e-01 4.43645477e-01
1.52001786e+00 9.35994804e-01 5.79502165e-01 9.54064488e-01
2.02031806e-02 9.44490194e-01 7.32020795e-01 8.23247612e-01
3.23227942e-01 2.74202168e-01 3.64049613e-01 4.28265452e-01
5.47506392e-01 -4.29505736e-01 6.65863395e-01 1.25137794e+00
-2.59603053e-01 2.05867693e-01 -1.08065808e+00 6.06487930e-01
-1.51297891e+00 -9.82854307e-01 2.67558753e-01 2.49496841e+00
5.88317156e-01 7.90098310e-01 8.33836719e-02 9.01838243e-02
4.52061474e-01 3.86409760e-01 -5.80097139e-01 -3.09781879e-01
-2.16573685e-01 4.52654511e-02 1.26214218e+00 2.61367857e-01
-9.16620970e-01 6.10372126e-01 7.32231283e+00 4.03056264e-01
-1.15574396e+00 -1.52350590e-01 1.16590524e+00 -3.48157078e-01
-8.29629064e-01 1.94170609e-01 -1.03327668e+00 1.02838767e+00
1.55382526e+00 -9.46603537e-01 4.44866419e-01 8.16105545e-01
3.75648826e-01 3.37138586e-02 -3.78457963e-01 4.07483637e-01
-3.67824733e-01 -1.62110782e+00 3.85623015e-02 4.99544650e-01
8.98458958e-01 5.58497682e-02 5.09721756e-01 8.79856274e-02
4.90755409e-01 -8.20775092e-01 1.29901350e+00 1.50351381e+00
4.01670486e-01 -1.29577041e+00 1.12379682e+00 4.09381151e-01
-1.13546503e+00 -6.63862646e-01 -3.81956488e-01 -2.13476643e-01
2.28019476e-01 8.99452329e-01 -2.23959327e-01 3.49642873e-01
7.63472855e-01 5.80098987e-01 -3.19163650e-01 7.67441273e-01
-4.59414572e-02 5.54362714e-01 -9.38280299e-02 -1.02192961e-01
3.77934992e-01 -5.91655850e-01 1.55112624e-01 7.58291304e-01
1.09870005e+00 -1.63085952e-01 -5.58340013e-01 8.62692893e-01
-4.81747724e-02 -1.37185931e-01 -5.96975446e-01 -4.08852875e-01
4.50441301e-01 7.79812872e-01 -3.07807028e-01 -1.13191314e-01
-5.83062887e-01 2.93480419e-02 -1.18803047e-01 3.60096663e-01
-5.69832802e-01 -3.71556461e-01 5.68915963e-01 7.35122487e-02
4.61872399e-01 -2.96144307e-01 -9.24596488e-01 -9.21863139e-01
-1.07481591e-02 -7.48367786e-01 2.04146370e-01 -5.52741528e-01
-1.11081064e+00 4.44842905e-01 6.26294091e-02 -1.29029417e+00
-6.77363992e-01 -6.38597369e-01 -8.96197081e-01 1.14634299e+00
-1.54574144e+00 -4.99609053e-01 6.65297031e-01 -3.20152380e-02
7.14732185e-02 -8.78409982e-01 5.33117056e-01 -5.05313762e-02
-6.57009184e-01 1.16114222e-01 8.27916324e-01 2.92450905e-01
7.25815399e-03 -1.23405898e+00 1.10035336e+00 9.06013787e-01
4.13513593e-02 4.30613786e-01 5.54260373e-01 -9.53631639e-01
-9.46597934e-01 -1.15630913e+00 1.36088216e+00 -8.97149593e-02
1.19305861e+00 1.07177049e-01 -7.30501294e-01 1.03904533e+00
1.57936960e-01 -6.23974562e-01 5.94616592e-01 -4.87537324e-01
1.98613331e-01 -6.69907570e-01 -9.04803514e-01 5.23195088e-01
1.72404632e-01 -2.57732481e-01 -7.44871497e-01 -9.13151130e-02
4.12552029e-01 4.76973467e-02 -1.35810518e+00 3.72051626e-01
1.03705525e+00 -9.10476863e-01 1.02796090e+00 -3.12380165e-01
1.54410660e-01 4.02070940e-01 -3.62580568e-01 -1.38860214e+00
-4.89939004e-01 -5.38484335e-01 -3.38783443e-01 1.34829342e+00
7.33917892e-01 -1.23288202e+00 8.02336872e-01 1.06370783e+00
3.55182707e-01 -9.51885879e-01 -1.15067208e+00 -8.43031108e-01
2.98207194e-01 -3.65616411e-01 1.28551483e+00 5.30869067e-01
-4.23281461e-01 -5.21988571e-01 -6.24007285e-01 -9.94296893e-02
9.37381864e-01 4.40950602e-01 3.12727481e-01 -1.28114569e+00
-3.78239192e-02 -8.24723423e-01 2.39282876e-01 -3.39212984e-01
1.69408888e-01 -5.13467908e-01 -4.20171618e-01 -1.45768249e+00
-1.39640778e-01 -2.13539153e-01 -1.09069884e+00 3.64660084e-01
2.19580054e-01 -1.89956009e-01 4.55821157e-01 4.86308217e-01
3.69752616e-01 6.35057032e-01 8.63329947e-01 -1.18542202e-01
7.81001300e-02 5.08976996e-01 -9.13398445e-01 9.04126704e-01
1.34806597e+00 -4.60362792e-01 7.61287808e-02 -2.90594220e-01
5.77712119e-01 3.29605430e-01 -4.33397666e-02 -7.67668009e-01
-6.36197180e-02 -7.17937589e-01 6.54822946e-01 -1.01107168e+00
2.58838892e-01 -6.74368024e-01 8.54120970e-01 9.02205825e-01
-3.74829799e-01 6.50031984e-01 1.00790203e-01 2.21708640e-01
-3.18463057e-01 -5.31356335e-01 4.86141354e-01 -3.64848763e-01
-4.39434975e-01 2.40728304e-01 -5.00447214e-01 -2.34848797e-01
9.78179574e-01 -1.15641870e-01 1.96607620e-01 -4.37738538e-01
-6.79121375e-01 2.71268368e-01 1.82756782e-02 3.66838515e-01
6.77464187e-01 -1.50663126e+00 -8.13440084e-01 -2.95607344e-04
-3.99661839e-01 -7.46307075e-01 -2.48782560e-01 2.23285586e-01
-4.56840962e-01 8.03395927e-01 -4.40773726e-01 4.18413967e-01
-4.20640677e-01 2.46214032e-01 5.93960285e-01 -9.36053395e-01
-3.71502519e-01 5.46705008e-01 1.14756487e-01 2.18328446e-01
1.73241109e-01 -4.81462598e-01 -6.17310643e-01 7.60417104e-01
7.48792112e-01 4.62098747e-01 6.25625327e-02 -5.58899879e-01
-1.28672630e-01 4.95809823e-01 2.31597751e-01 -4.78008509e-01
1.99269819e+00 -5.60223199e-02 -2.84916669e-01 8.49348545e-01
8.05217326e-01 -2.38527447e-01 -1.43462324e+00 -2.38423645e-01
9.85350013e-01 -5.35181344e-01 7.06552044e-02 -7.80286372e-01
-1.63859439e+00 6.83928788e-01 3.35564107e-01 1.61269769e-01
1.01514125e+00 -7.34724164e-01 1.06248677e+00 2.59497672e-01
5.69017529e-01 -1.32407737e+00 -3.08874696e-01 5.99052250e-01
1.19520783e+00 -5.44015527e-01 4.94784750e-02 5.50768435e-01
-6.02340579e-01 1.10938835e+00 -9.48735103e-02 -2.69329846e-01
1.00686395e+00 2.34484136e-01 -9.59284604e-02 1.73813984e-01
-9.22455966e-01 2.81912565e-01 4.62002873e-01 1.14819005e-01
7.20620230e-02 3.52255464e-01 -2.24228248e-01 1.27732670e+00
-4.81530726e-01 -9.71091762e-02 5.55540085e-01 5.89726388e-01
-5.04568160e-01 -8.51293564e-01 -5.15103102e-01 1.06048787e+00
-1.16628754e+00 -2.18915388e-01 -1.30466282e-01 4.21000600e-01
-3.22932035e-01 5.58107615e-01 3.68972689e-01 -4.64934617e-01
5.60074329e-01 2.29079381e-01 -2.98220068e-01 -2.62209386e-01
-9.04416144e-01 9.69965756e-02 1.58298910e-01 -1.27371356e-01
1.52936131e-01 -8.82436752e-01 -1.06420374e+00 -4.82526451e-01
-3.09460163e-01 1.61189899e-01 6.03932917e-01 8.38415682e-01
-2.36990675e-02 3.61414641e-01 1.17162561e+00 -9.97389197e-01
-1.23557484e+00 -8.07478964e-01 -1.09071815e+00 -2.65798301e-01
2.42419183e-01 -4.21900868e-01 -6.14669800e-01 -2.47669175e-01]
|
[4.5837016105651855, 4.109222888946533]
|
92cd9d70-46f3-4093-a4d8-502c940a874a
|
deep-stable-multi-interest-learning-for-out
|
2304.05615
| null |
https://arxiv.org/abs/2304.05615v1
|
https://arxiv.org/pdf/2304.05615v1.pdf
|
Deep Stable Multi-Interest Learning for Out-of-distribution Sequential Recommendation
|
Recently, multi-interest models, which extract interests of a user as multiple representation vectors, have shown promising performances for sequential recommendation. However, none of existing multi-interest recommendation models consider the Out-Of-Distribution (OOD) generalization problem, in which interest distribution may change. Considering multiple interests of a user are usually highly correlated, the model has chance to learn spurious correlations between noisy interests and target items. Once the data distribution changes, the correlations among interests may also change, and the spurious correlations will mislead the model to make wrong predictions. To tackle with above OOD generalization problem, we propose a novel multi-interest network, named DEep Stable Multi-Interest Learning (DESMIL), which attempts to de-correlate the extracted interests in the model, and thus spurious correlations can be eliminated. DESMIL applies an attentive module to extract multiple interests, and then selects the most important one for making final predictions. Meanwhile, DESMIL incorporates a weighted correlation estimation loss based on Hilbert-Schmidt Independence Criterion (HSIC), with which training samples are weighted, to minimize the correlations among extracted interests. Extensive experiments have been conducted under both OOD and random settings, and up to 36.8% and 21.7% relative improvements are achieved respectively.
|
['Liang Wang', 'Shu Wu', 'Zhenxi Zhu', 'Zhaocheng Liu', 'Qiang Liu']
|
2023-04-12
| null | null | null | null |
['sequential-recommendation']
|
['miscellaneous']
|
[-7.18188137e-02 -2.48349145e-01 -2.66875803e-01 -4.71779346e-01
-7.17547536e-01 -2.33094409e-01 3.74660641e-01 -3.75228599e-02
-8.34764466e-02 8.37172389e-01 5.11946917e-01 2.50915527e-01
-2.54890382e-01 -8.32507968e-01 -6.16832495e-01 -7.29587615e-01
-2.36515269e-01 3.46429974e-01 1.00370750e-01 -1.53831556e-01
3.14134359e-01 3.56583446e-02 -1.47515726e+00 4.38841283e-01
1.04048836e+00 1.19161773e+00 1.76293805e-01 1.72897458e-01
-1.11787573e-01 9.26420629e-01 -5.85259259e-01 -1.45012707e-01
1.81014717e-01 -4.32402849e-01 -5.38254023e-01 -1.27335936e-01
5.87307848e-02 -2.37419367e-01 -7.20760971e-02 8.71589303e-01
4.58293885e-01 5.29565156e-01 7.16823578e-01 -1.35926461e+00
-1.13978243e+00 7.42522299e-01 -7.97838330e-01 1.66557550e-01
8.26181769e-02 -4.70344424e-01 1.44338262e+00 -1.31000090e+00
2.43166834e-01 1.22250962e+00 5.59919715e-01 1.24828964e-01
-9.68789339e-01 -8.73167634e-01 5.00230312e-01 3.26516956e-01
-1.33594179e+00 -2.96493107e-03 7.70246983e-01 -2.48896126e-02
5.45827746e-01 3.51006001e-01 5.66283226e-01 1.23620021e+00
2.55942881e-01 1.29875970e+00 7.53907561e-01 3.08078714e-03
1.28020067e-02 5.01917481e-01 2.77167857e-01 9.89433303e-02
4.99142468e-01 8.75385012e-03 -3.60054493e-01 -2.32725471e-01
4.24688220e-01 4.84721541e-01 -3.50381941e-01 -2.58748502e-01
-1.04540861e+00 9.66118276e-01 3.38905931e-01 5.74905515e-01
-4.05894458e-01 -4.46127623e-01 2.26141587e-01 4.59200829e-01
8.65444899e-01 3.79248500e-01 -4.91479069e-01 2.22684905e-01
-1.06090105e+00 3.97165298e-01 5.00316679e-01 9.79154289e-01
6.90580904e-01 -9.47349519e-02 -4.43101346e-01 1.01595163e+00
6.38601840e-01 2.39503056e-01 8.92673254e-01 -4.41677660e-01
4.12110716e-01 5.46971560e-01 3.04421306e-01 -1.55678129e+00
-4.03550148e-01 -1.09027636e+00 -1.13620281e+00 -2.78641343e-01
8.46879631e-02 -3.01086187e-01 -2.85502493e-01 1.49924278e+00
2.78385282e-01 5.49350739e-01 2.00365067e-01 9.62624550e-01
9.13928568e-01 7.59230971e-01 -2.33256608e-01 -6.25218689e-01
7.51025796e-01 -1.07891798e+00 -5.09165347e-01 -1.58839732e-01
4.38505322e-01 -7.69425273e-01 9.84501958e-01 5.29871583e-01
-1.05881691e+00 -7.79667258e-01 -1.01172078e+00 1.84357375e-01
-8.08218867e-02 7.56278336e-02 3.92181784e-01 1.02368936e-01
-5.59850216e-01 8.38851810e-01 -2.22614616e-01 1.58785656e-03
4.54324722e-01 3.73081893e-01 9.69230756e-02 -1.45037994e-01
-1.51910186e+00 5.09122491e-01 2.57787406e-01 8.09216723e-02
-5.26530027e-01 -8.85949433e-01 -6.41936958e-01 2.02426016e-01
5.24996758e-01 -3.21286768e-01 8.89923155e-01 -1.13384306e+00
-1.10058904e+00 2.41019297e-02 -2.33638600e-01 -1.97619289e-01
4.45644855e-01 -4.84321475e-01 -9.39896524e-01 -4.18868184e-01
3.09543401e-01 -3.36078145e-02 7.45312572e-01 -1.39419580e+00
-1.05854583e+00 -1.91148445e-01 -2.48073936e-01 2.58504748e-01
-5.05586505e-01 -1.40223816e-01 -2.44886369e-01 -8.50208759e-01
5.17578274e-02 -6.18516326e-01 -2.45656088e-01 -4.89523977e-01
-2.54594117e-01 -4.59377050e-01 8.40748727e-01 -3.75374556e-01
1.50289297e+00 -1.96605933e+00 -1.15534827e-01 2.28947014e-01
2.73503333e-01 2.16871649e-01 -3.71596456e-01 3.38894203e-02
-4.82316837e-02 3.77674140e-02 4.19482976e-01 -1.37721106e-01
-1.88423604e-01 1.10547189e-02 -3.52926373e-01 2.74985611e-01
-4.04152088e-03 6.27440929e-01 -1.09792984e+00 -2.70460099e-01
3.81831117e-02 3.76336157e-01 -7.73708761e-01 3.13072115e-01
-1.52823627e-01 2.82874823e-01 -7.21986949e-01 4.44448709e-01
1.13980854e+00 -5.86791277e-01 1.32753015e-01 -3.54736865e-01
-1.76489532e-01 2.75494516e-01 -1.40045977e+00 1.11262143e+00
-7.09885120e-01 7.59530738e-02 -4.36187446e-01 -1.11702228e+00
1.01503563e+00 1.71866417e-01 7.65849948e-01 -4.98579830e-01
1.67068675e-01 -4.86769527e-02 -1.69173300e-01 -2.92842776e-01
8.08552384e-01 -6.13767505e-02 7.39473011e-03 5.11874318e-01
-5.94451837e-03 5.96895516e-01 -1.26943111e-01 2.54965294e-02
5.94933331e-01 -8.27499777e-02 3.72390658e-01 -2.52238601e-01
6.38414502e-01 -2.71071434e-01 8.13426495e-01 8.60268712e-01
7.86668360e-02 6.77812099e-01 2.66444504e-01 -2.12984815e-01
-7.94772744e-01 -9.80106056e-01 1.42875493e-01 1.13078427e+00
4.71664399e-01 -1.64670855e-01 1.32669330e-01 -1.00175166e+00
-4.63567227e-02 9.56963062e-01 -6.24920964e-01 -3.83671910e-01
-2.18028367e-01 -9.49917078e-01 -1.12453371e-01 2.09074259e-01
3.25064480e-01 -9.34081316e-01 2.34227568e-01 5.31920671e-01
-4.05931503e-01 -5.33472538e-01 -6.87914550e-01 -2.72202864e-02
-7.97709823e-01 -9.23647165e-01 -9.89743173e-01 -5.98132551e-01
5.29427171e-01 8.57147574e-01 1.16275358e+00 -1.62045419e-01
2.85917222e-01 -8.73747617e-02 -7.36561179e-01 -2.11625934e-01
-2.52833832e-02 1.41334943e-02 1.83211088e-01 4.04587954e-01
5.97834110e-01 -6.74362719e-01 -5.66664338e-01 5.66429496e-01
-7.56347001e-01 -3.64970744e-01 6.13955855e-01 1.27700114e+00
6.21016264e-01 3.21920872e-01 9.95843649e-01 -1.03542304e+00
7.70601153e-01 -1.19581318e+00 -1.09738722e-01 3.33552718e-01
-1.02648616e+00 -1.32836606e-02 1.23847210e+00 -5.79169929e-01
-1.18874013e+00 -6.28572941e-01 -5.31037562e-02 -6.21107757e-01
-7.35256914e-03 6.59979820e-01 -2.62319207e-01 3.39802295e-01
3.87628555e-01 3.39717954e-01 -3.21862996e-01 -6.17594481e-01
2.97193885e-01 7.28663325e-01 -7.70898834e-02 -2.25950241e-01
6.13255203e-01 2.51513153e-01 -3.75845551e-01 -7.34789550e-01
-1.47217798e+00 -7.61118889e-01 -3.57061744e-01 3.77914421e-02
2.23861605e-01 -1.10593283e+00 -3.59170973e-01 2.72415936e-01
-9.14702475e-01 3.68353516e-01 -2.56547779e-01 9.31944907e-01
-1.94990724e-01 5.90498805e-01 -1.88320190e-01 -8.19800317e-01
-4.90724027e-01 -6.86192632e-01 5.93780518e-01 5.55520654e-01
-1.68522298e-01 -9.70021129e-01 -2.15820894e-02 1.77881941e-01
3.72848928e-01 -1.48162246e-01 7.49207139e-01 -1.13853467e+00
-1.85340554e-01 -2.11485088e-01 -3.10696781e-01 4.46800470e-01
3.89408886e-01 -2.11013183e-01 -5.42957067e-01 -4.62986290e-01
9.74140987e-02 -1.89435750e-01 8.05475354e-01 3.98074567e-01
1.16420424e+00 -5.74328303e-01 -2.76073992e-01 3.70453060e-01
1.21380913e+00 2.01376706e-01 5.82691252e-01 1.61402985e-01
5.62928677e-01 2.56204814e-01 1.09091663e+00 9.92448568e-01
5.69461942e-01 5.05793333e-01 3.75053048e-01 -9.40690562e-03
4.04042363e-01 -2.26278499e-01 4.24311459e-01 1.05786705e+00
-2.77121365e-01 -5.85008025e-01 1.06753349e-01 4.42327619e-01
-1.95901680e+00 -1.20945656e+00 -1.68434441e-01 2.41151404e+00
6.61815464e-01 1.29654586e-01 2.57674396e-01 -1.35877997e-01
7.58559465e-01 3.90931189e-01 -9.00508404e-01 5.44997454e-02
-1.31063163e-01 5.87145835e-02 2.74638563e-01 1.45976573e-01
-1.12753963e+00 5.90258121e-01 5.34478617e+00 9.86241221e-01
-9.82194364e-01 1.03875294e-01 6.19757533e-01 -4.81905669e-01
-6.00515783e-01 -2.99055934e-01 -9.18923974e-01 5.36116421e-01
5.25759876e-01 -5.36256969e-01 9.92205590e-02 1.00960493e+00
5.71571171e-01 9.71670300e-02 -6.93270087e-01 7.88462579e-01
2.83299208e-01 -9.18091238e-01 1.81603953e-01 -8.91317613e-03
8.54247451e-01 -3.03204544e-02 1.90700859e-01 6.64729893e-01
5.73241472e-01 -5.69590926e-01 3.55512977e-01 8.70933473e-01
2.72500336e-01 -1.11279774e+00 9.57374215e-01 4.02709961e-01
-1.41225553e+00 -1.34004951e-01 -9.29654360e-01 1.32868484e-01
-8.83211382e-03 9.72541213e-01 -3.59295934e-01 8.65779400e-01
6.80579066e-01 1.32026005e+00 -1.90982208e-01 1.00019705e+00
4.22779471e-02 6.18217945e-01 -1.26057506e-01 -3.50406855e-01
1.55053437e-01 -4.15952742e-01 6.73289001e-01 8.48630309e-01
8.99234056e-01 -4.34678718e-02 2.14053035e-01 8.29379380e-01
-9.71927345e-02 4.78901982e-01 -1.91162467e-01 9.02654305e-02
2.46405631e-01 1.33563864e+00 -1.59946471e-01 -4.13387716e-01
-6.15857482e-01 1.11824811e+00 2.70560652e-01 2.32252419e-01
-9.80740845e-01 -3.60508263e-01 7.38319457e-01 1.16491199e-01
5.87769389e-01 1.59132421e-01 -3.33533362e-02 -1.54160619e+00
5.65443560e-02 -7.40332127e-01 5.10500669e-01 -4.04319376e-01
-1.73896182e+00 4.00465369e-01 -2.84924060e-01 -1.79981685e+00
-5.93579300e-02 8.39739740e-02 -8.59511018e-01 8.98155391e-01
-1.63605094e+00 -9.11739528e-01 -8.33563656e-02 5.59655726e-01
7.15686917e-01 -3.67047399e-01 5.64231396e-01 4.30355698e-01
-6.11052394e-01 9.45098758e-01 7.88143754e-01 -3.14591169e-01
8.09695780e-01 -1.05439353e+00 -4.37839329e-02 4.35720742e-01
5.79179078e-02 6.81894243e-01 4.92910504e-01 -4.80537057e-01
-9.89980578e-01 -1.58949161e+00 1.13072324e+00 1.16716251e-01
5.04337311e-01 -4.25831880e-03 -1.04704046e+00 4.45366383e-01
-1.20839491e-01 9.85844210e-02 1.13497496e+00 4.40122724e-01
-2.38559633e-01 -1.44322276e-01 -1.10017633e+00 4.96875912e-01
9.40103769e-01 1.08412672e-02 -6.76604390e-01 2.26297200e-01
6.41910195e-01 2.21780449e-01 -9.65291977e-01 3.34161878e-01
7.14792252e-01 -1.16411281e+00 1.17220259e+00 -9.30847287e-01
6.40451491e-01 -1.48264706e-01 -1.56119347e-01 -1.80333567e+00
-1.11060309e+00 -1.59997642e-01 -3.52888227e-01 1.39252353e+00
5.16079545e-01 -3.02526593e-01 7.20678687e-01 2.57639647e-01
2.91531943e-02 -9.10498023e-01 -4.94397134e-01 -7.24479675e-01
-3.76209281e-02 -2.32912570e-01 6.82055771e-01 9.85993743e-01
1.47393495e-01 6.62431896e-01 -1.06287766e+00 2.41880596e-01
5.80544293e-01 6.35308266e-01 4.92923945e-01 -1.44367290e+00
-4.43322003e-01 -3.00046355e-01 4.82026860e-03 -1.47949708e+00
-1.72307733e-02 -8.54241312e-01 -1.05185091e-01 -1.46175742e+00
1.31430700e-01 -5.22003949e-01 -9.17130113e-01 9.38281938e-02
-3.90924066e-01 5.16232848e-03 1.00669473e-01 3.76367748e-01
-1.11239529e+00 9.12502885e-01 1.32052279e+00 -1.39154106e-01
-4.60493028e-01 8.08367252e-01 -1.31275225e+00 7.70881414e-01
7.98827529e-01 -4.64382470e-01 -4.99688029e-01 7.95696378e-02
7.40623996e-02 -2.07953565e-02 -9.32297707e-02 -9.51501608e-01
-2.32510120e-02 -2.35453159e-01 7.75281310e-01 -9.68253553e-01
7.03089833e-02 -8.99718285e-01 5.35582751e-02 1.54422730e-01
-5.36940932e-01 -3.51030707e-01 -1.41202688e-01 7.69167781e-01
-2.51992345e-01 -3.45350474e-01 7.81594455e-01 -2.24891633e-01
-5.96035957e-01 5.44599056e-01 -3.65655303e-01 3.05873573e-01
8.73974919e-01 3.15840244e-02 7.98141807e-02 -4.43209261e-01
-8.70907426e-01 4.81384635e-01 -2.65079707e-01 7.63554335e-01
6.84638917e-01 -1.81463850e+00 -8.13173771e-01 1.83232576e-01
1.67854711e-01 -1.05266109e-01 6.32689416e-01 6.29330575e-01
5.35286605e-01 2.71351188e-01 1.44804016e-01 -1.34365082e-01
-1.20472801e+00 6.44800067e-01 6.57042563e-02 -7.22949445e-01
-1.08822167e-01 8.86708260e-01 2.76508152e-01 -3.46917897e-01
-1.51559906e-02 -2.82357872e-01 -7.65256166e-01 6.42200351e-01
7.38349736e-01 5.94861031e-01 -1.60578042e-01 -6.55695617e-01
-1.40329987e-01 3.31591547e-01 -5.78827262e-01 4.62842464e-01
1.61665368e+00 -5.38131952e-01 1.24479316e-01 6.76958442e-01
1.47421587e+00 5.42578921e-02 -9.81342077e-01 -7.65956044e-01
-2.36355811e-01 -5.57641566e-01 1.58819601e-01 -6.72159493e-01
-1.21992373e+00 6.86494708e-01 4.01653409e-01 4.25816894e-01
9.58906114e-01 -1.14504896e-01 1.05893588e+00 2.95042217e-01
1.94824621e-01 -1.14684176e+00 1.16095230e-01 8.62089515e-01
9.54013705e-01 -1.35580540e+00 8.39888528e-02 -7.24309012e-02
-7.84171581e-01 9.48806345e-01 7.24837184e-01 -2.00817853e-01
1.08748281e+00 -1.91868573e-01 -3.53862077e-01 1.54649019e-01
-6.78534389e-01 -5.03915884e-02 5.36190093e-01 4.79927242e-01
4.53593910e-01 -5.65600805e-02 -4.35872585e-01 1.32770920e+00
1.63392723e-01 2.21240237e-01 2.92546034e-01 3.80387872e-01
-6.18648171e-01 -9.94255543e-01 -1.77956954e-01 1.07308722e+00
-5.00239372e-01 -2.45749891e-01 -2.44856309e-02 3.71975482e-01
3.84104788e-01 1.17971551e+00 3.90565433e-02 -8.03241611e-01
5.06330192e-01 -2.71889448e-01 -1.30126953e-01 -5.59574306e-01
-4.71260697e-01 3.56365174e-01 -1.01985350e-01 -3.58629078e-01
-3.50670576e-01 -7.74904966e-01 -1.05566823e+00 -1.48596630e-01
-7.59155154e-01 5.75821102e-01 -1.33421421e-01 1.05193782e+00
4.80619431e-01 6.82020903e-01 1.42214870e+00 -7.02180386e-01
-7.86844850e-01 -1.15913045e+00 -1.08211207e+00 4.68200684e-01
1.41686782e-01 -8.36097240e-01 -6.29338861e-01 -4.07747656e-01]
|
[10.168803215026855, 5.557633876800537]
|
f4f0f26b-c43f-44f0-9e12-daaca0cf5374
|
slim-u-net-efficient-anatomical-feature
|
2302.11524
| null |
https://arxiv.org/abs/2302.11524v1
|
https://arxiv.org/pdf/2302.11524v1.pdf
|
Slim U-Net: Efficient Anatomical Feature Preserving U-net Architecture for Ultrasound Image Segmentation
|
We investigate the applicability of U-Net based models for segmenting Urinary Bladder (UB) in male pelvic view UltraSound (US) images. The segmentation of UB in the US image aids radiologists in diagnosing the UB. However, UB in US images has arbitrary shapes, indistinct boundaries and considerably large inter- and intra-subject variability, making segmentation a quite challenging task. Our study of the state-of-the-art (SOTA) segmentation network, U-Net, for the problem reveals that it often fails to capture the salient characteristics of UB due to the varying shape and scales of anatomy in the noisy US image. Also, U-net has an excessive number of trainable parameters, reporting poor computational efficiency during training. We propose a Slim U-Net to address the challenges of UB segmentation. Slim U-Net proposes to efficiently preserve the salient features of UB by reshaping the structure of U-Net using a less number of 2D convolution layers in the contracting path, in order to preserve and impose them on expanding path. To effectively distinguish the blurred boundaries, we propose a novel annotation methodology, which includes the background area of the image at the boundary of a marked region of interest (RoI), thereby steering the model's attention towards boundaries. In addition, we suggested a combination of loss functions for network training in the complex segmentation of UB. The experimental results demonstrate that Slim U-net is statistically superior to U-net for UB segmentation. The Slim U-net further decreases the number of trainable parameters and training time by 54% and 57.7%, respectively, compared to the standard U-Net, without compromising the segmentation accuracy.
|
['Subir Kumar Saha', 'SH Chandrashekhara', 'Kashish Verma', 'Deepak Raina']
|
2023-02-22
| null | null | null | null |
['anatomy']
|
['miscellaneous']
|
[ 3.14227045e-01 5.56321740e-01 -1.45151585e-01 -2.49858737e-01
-4.05459493e-01 -5.54767728e-01 -1.17775232e-01 -7.60555565e-02
-4.63453114e-01 5.36925495e-01 -2.09250942e-01 -5.92844188e-01
-7.47298077e-02 -6.82209373e-01 -8.31648648e-01 -6.20976210e-01
-2.25184828e-01 2.17696741e-01 3.94001245e-01 -5.40470593e-02
8.52114055e-03 5.55356562e-01 -7.94674695e-01 -1.67607460e-02
1.25971162e+00 1.09681189e+00 5.64332426e-01 5.66222370e-01
-3.49181831e-01 6.06009543e-01 -3.31210792e-01 -3.40064943e-01
3.10328573e-01 -7.33922958e-01 -1.12722707e+00 -8.26307982e-02
5.23796558e-01 -3.38090330e-01 -4.46990967e-01 1.25876307e+00
6.72582448e-01 2.26894006e-01 3.85077566e-01 -5.04428744e-01
-6.63098335e-01 5.85383415e-01 -6.32030785e-01 6.45942509e-01
-1.19103521e-01 -4.34086099e-02 5.55277705e-01 -3.27963740e-01
9.71103668e-01 1.05649102e+00 1.08949804e+00 7.38625586e-01
-1.18561149e+00 -6.55438721e-01 -1.92212611e-01 -1.37627602e-01
-8.75801980e-01 1.22251108e-01 6.72928333e-01 -3.12528878e-01
3.47608656e-01 5.77256918e-01 7.02670038e-01 8.02149892e-01
3.64182532e-01 1.07976747e+00 9.65598881e-01 -3.05414110e-01
-5.83016202e-02 -2.17651054e-01 3.25955719e-01 1.11919415e+00
2.70259529e-01 -3.02125532e-02 2.69038111e-01 2.73073930e-02
1.60418403e+00 -2.28194296e-02 -4.41107690e-01 -5.83903074e-01
-8.32570255e-01 5.99153876e-01 8.99429083e-01 7.20002472e-01
-1.67601198e-01 -5.80993388e-03 5.80276549e-01 -1.27329439e-01
2.03339919e-01 7.07414091e-01 -2.02505037e-01 1.18338682e-01
-7.39899039e-01 -9.71449390e-02 4.14480716e-01 1.00155592e+00
4.03928697e-01 -2.32274562e-01 -4.92694914e-01 9.14924264e-01
-8.96899030e-02 1.22667678e-01 7.79446959e-01 -1.07140410e+00
3.68676960e-01 7.03020751e-01 -2.40346074e-01 -8.09377313e-01
-7.22093046e-01 -6.35103464e-01 -1.10126877e+00 -2.25354671e-01
6.07030690e-01 -3.03023934e-01 -1.59651828e+00 1.37604773e+00
2.63388723e-01 3.42009753e-01 -2.45052055e-01 1.12508130e+00
1.10236216e+00 2.99633920e-01 -4.05502357e-02 -7.21196011e-02
1.28915036e+00 -1.05417717e+00 -8.62134755e-01 -3.00103843e-01
6.17638707e-01 -5.77361465e-01 9.51715112e-01 -1.81097407e-02
-1.28947425e+00 -5.53311288e-01 -9.44937646e-01 -2.22833484e-01
-7.63653144e-02 2.38141969e-01 7.94975519e-01 8.04372251e-01
-9.44279552e-01 9.49032962e-01 -1.24763870e+00 -1.78637117e-01
6.52328968e-01 5.58117032e-01 -3.13175291e-01 6.34277239e-02
-1.07243276e+00 9.41970527e-01 5.58801651e-01 4.73438233e-01
-4.09199804e-01 -7.27375507e-01 -1.30282712e+00 2.56192386e-01
3.75564605e-01 -5.20769477e-01 1.29437482e+00 -7.50957966e-01
-1.37111866e+00 7.70791531e-01 -3.34182344e-02 -4.18208182e-01
9.64271486e-01 1.28798276e-01 -5.73708350e-03 2.85025775e-01
6.16973341e-02 4.21265990e-01 4.40524012e-01 -1.44556558e+00
-4.73208666e-01 -4.52012807e-01 -8.40670541e-02 6.38680607e-02
-1.04742497e-01 -2.84113616e-01 -9.35041308e-01 -7.25168347e-01
8.37783217e-01 -9.34753239e-01 -5.34043312e-01 9.11834277e-03
-2.77295619e-01 1.83372483e-01 8.79243433e-01 -1.00150919e+00
1.25403559e+00 -2.12074065e+00 -5.43183200e-02 3.16026300e-01
3.50606889e-01 6.21889591e-01 1.66426480e-01 -5.36585271e-01
9.24717113e-02 1.39179811e-01 -6.11767828e-01 -1.43573731e-01
-3.96580994e-01 7.26203859e-01 2.35738769e-01 5.56593895e-01
-8.35429430e-02 1.14695108e+00 -1.21357501e+00 -8.44404459e-01
3.46542567e-01 1.59534484e-01 -6.02875710e-01 2.18219787e-01
3.29852611e-01 7.84130573e-01 -4.66814011e-01 7.22984374e-01
8.90134931e-01 -3.31210077e-01 3.32537413e-01 -2.17099339e-01
1.27343938e-01 -1.45419613e-01 -1.15007424e+00 1.79399490e+00
-4.76675540e-01 3.67467582e-01 4.63832527e-01 -1.03050673e+00
7.85717130e-01 2.11610615e-01 5.59820950e-01 -9.80591416e-01
3.84517074e-01 6.37718201e-01 2.55184621e-01 -7.35793114e-01
3.06965113e-01 -1.28295273e-01 1.56244084e-01 2.35286690e-02
3.45199704e-01 -9.11640450e-02 3.91179562e-01 -2.07545340e-01
7.50100076e-01 7.39277378e-02 1.96213752e-01 -4.75170702e-01
5.54946661e-01 -2.87125468e-01 7.00101614e-01 1.02525449e+00
-6.31485045e-01 8.26293945e-01 5.34352362e-01 -5.18571198e-01
-7.55531251e-01 -9.73502040e-01 -7.24692822e-01 4.80760634e-01
8.49384964e-01 4.04735416e-01 -8.74639690e-01 -9.14696515e-01
5.53081557e-02 2.98308492e-01 -9.42899764e-01 3.84544209e-03
-1.04528260e+00 -8.60132158e-01 6.88534915e-01 6.97636187e-01
6.58477843e-01 -1.29163873e+00 -6.91523314e-01 2.45134339e-01
-2.52321661e-01 -9.98248458e-01 -6.40591443e-01 2.98343092e-01
-1.27801740e+00 -1.21056759e+00 -1.19181120e+00 -1.17389035e+00
1.14130747e+00 2.20760028e-03 7.94628799e-01 -1.55467847e-02
-4.34122503e-01 -1.79106474e-01 9.81045328e-03 -8.54316503e-02
-4.61004138e-01 3.49496789e-02 -4.46421593e-01 -3.95211875e-01
-7.55372643e-02 -1.55807346e-01 -7.88113177e-01 3.72318149e-01
-9.88470733e-01 3.65037657e-02 6.72768772e-01 1.31127191e+00
6.53269589e-01 -3.05267215e-01 1.85399488e-01 -1.13738418e+00
5.54769337e-01 -2.54520863e-01 -4.10999358e-01 1.88677907e-01
-4.61741626e-01 -1.86447993e-01 4.32474524e-01 -5.01380026e-01
-1.14012849e+00 -1.31242558e-01 -1.77331164e-01 -4.49737012e-01
1.07344411e-01 3.16639006e-01 5.11012077e-01 -5.47616482e-01
5.50557017e-01 2.36417904e-01 1.68986171e-01 -4.52417046e-01
4.74206470e-02 4.39141631e-01 8.89050901e-01 -5.19966245e-01
1.69221923e-01 4.03686076e-01 -5.06382026e-02 -6.51615083e-01
-6.85529828e-01 -8.03547919e-01 -6.47548795e-01 -1.30231567e-02
8.46929014e-01 -4.05902594e-01 -6.02662981e-01 2.87885249e-01
-1.05516958e+00 -4.36255544e-01 -2.61903465e-01 4.27562475e-01
-4.85900700e-01 5.54902256e-01 -1.41222823e+00 -4.68036890e-01
-6.16783857e-01 -1.47661412e+00 7.01484740e-01 8.32593381e-01
-3.54235107e-03 -1.03413022e+00 -3.05032492e-01 4.16799217e-01
3.93492818e-01 4.29749876e-01 8.86422038e-01 -5.12567401e-01
-2.07019657e-01 -2.14113906e-01 -6.71160579e-01 3.58909994e-01
3.33062321e-01 -5.89230478e-01 -8.48358870e-01 -2.14282736e-01
1.09632023e-01 -2.96531487e-02 9.03015316e-01 9.29288924e-01
1.67433631e+00 5.64542003e-02 -3.90170097e-01 1.07627177e+00
1.28639066e+00 5.76116085e-01 4.81405109e-01 3.72704238e-01
7.23349452e-01 2.97389001e-01 4.57834035e-01 3.35658900e-02
-5.82718365e-02 1.81480601e-01 5.32162070e-01 -8.79508793e-01
-1.39560357e-01 8.20169821e-02 -4.50917006e-01 5.13454974e-01
-3.34983677e-01 4.12292570e-01 -8.96290243e-01 7.97456324e-01
-1.81350505e+00 -5.11568666e-01 -7.20304847e-02 1.94416797e+00
1.13128161e+00 2.21030906e-01 -2.86833823e-01 -3.76774698e-01
6.67998493e-01 1.75818168e-02 -5.42501628e-01 -3.98043185e-01
2.58422107e-01 2.50284135e-01 9.63937104e-01 5.02381146e-01
-1.39021480e+00 6.53795063e-01 6.28502655e+00 7.96007633e-01
-1.31229496e+00 9.82837528e-02 9.53777194e-01 3.09824705e-01
5.37733622e-02 -4.60654825e-01 -4.98627514e-01 4.11454082e-01
1.19743526e-01 2.39653200e-01 7.60062560e-02 7.49519706e-01
-1.94869284e-02 -1.16232425e-01 -7.94069529e-01 8.30498815e-01
-9.45577845e-02 -1.57327282e+00 -3.46215904e-01 -1.10547125e-01
9.35362637e-01 -4.95138131e-02 -2.04795152e-01 3.71215552e-01
1.59247108e-02 -1.03243721e+00 3.63861829e-01 1.52055189e-01
8.16491663e-01 -5.68577170e-01 1.37989056e+00 2.20413029e-01
-9.28489685e-01 3.20257917e-02 -3.19732636e-01 3.50727707e-01
-6.47761375e-02 1.93139032e-01 -8.48489404e-01 6.17931902e-01
8.49461973e-01 1.89278111e-01 -4.67020541e-01 1.26435614e+00
1.19203530e-01 3.84322643e-01 -2.73088723e-01 2.14868039e-01
7.44922698e-01 -4.58683103e-01 3.43588948e-01 1.57940590e+00
3.83418798e-01 2.34549657e-01 2.28453726e-01 1.09052241e+00
-1.81997254e-01 5.79484664e-02 -3.81590009e-01 3.43995601e-01
3.65123302e-02 1.23444748e+00 -1.09544480e+00 -3.97219688e-01
-1.49865821e-01 1.00380361e+00 1.19751997e-01 3.72261882e-01
-6.06514931e-01 -3.77272487e-01 1.13165257e-02 -1.06368087e-01
2.32276723e-01 1.89934954e-01 -8.01173151e-01 -8.58328879e-01
2.38591805e-02 -2.90428281e-01 5.02747357e-01 -2.28409126e-01
-8.93499196e-01 6.40626073e-01 -2.45326266e-01 -1.05122268e+00
9.92905870e-02 -5.82079232e-01 -5.77086866e-01 9.20117199e-01
-1.64126706e+00 -1.04994166e+00 -3.81306857e-01 1.53014809e-01
4.43972915e-01 4.84184325e-01 5.51512241e-01 4.42125052e-01
-5.36694527e-01 6.41533494e-01 2.40073562e-01 5.03708720e-01
5.85941076e-01 -1.67982590e+00 1.82175040e-01 8.01667392e-01
-5.48148453e-01 9.48850036e-01 5.46646893e-01 -7.92982817e-01
-1.01668823e+00 -8.99829030e-01 6.25759125e-01 4.46996139e-03
5.12841702e-01 -3.23073678e-02 -9.51994002e-01 7.93356121e-01
-1.33197039e-01 6.79580450e-01 4.38290775e-01 -1.93754509e-01
4.03206468e-01 2.33689174e-01 -1.46824932e+00 5.94476998e-01
8.24253201e-01 1.91466864e-02 -5.62581658e-01 2.47688126e-02
5.31586111e-01 -1.40858889e+00 -1.13760602e+00 7.27648795e-01
5.77262521e-01 -7.98028469e-01 1.10831285e+00 -3.12580973e-01
6.71456635e-01 4.53726165e-02 5.88651538e-01 -1.10720539e+00
-2.95213044e-01 -4.67556953e-01 -1.39273882e-01 6.78038836e-01
2.24390820e-01 -4.70586002e-01 1.13021839e+00 5.86216807e-01
-6.02114320e-01 -9.95453119e-01 -1.04464948e+00 -4.33016896e-01
9.87239764e-04 1.34779904e-02 2.48555504e-02 1.16414297e+00
-3.08694765e-02 -3.91322106e-01 4.14751563e-03 2.49570068e-02
6.22675300e-01 2.48874903e-01 3.05814654e-01 -9.82707500e-01
-1.03413038e-01 -7.00953543e-01 -1.26226649e-01 -1.05634892e+00
-3.95610213e-01 -9.54028249e-01 1.32292911e-01 -1.67246652e+00
1.21872015e-01 -6.61462963e-01 -3.80723834e-01 3.75901818e-01
-6.00791752e-01 3.40742916e-01 2.19221070e-01 8.71954188e-02
-1.53007448e-01 2.36886069e-01 2.42533040e+00 -1.66123644e-01
-5.65676630e-01 3.27505618e-01 -3.00996065e-01 8.66001129e-01
4.68654364e-01 -5.19220717e-02 -1.19505458e-01 -1.27673402e-01
-6.59877300e-01 4.11264360e-01 1.14823364e-01 -5.93577147e-01
8.44327211e-02 5.53569086e-02 7.16849744e-01 -5.41265547e-01
-9.15933102e-02 -8.46491396e-01 -2.41896823e-01 9.46845829e-01
-1.34116143e-01 -5.16015410e-01 5.03828228e-01 1.79348350e-01
-3.52571130e-01 -6.38222814e-01 1.04938734e+00 -6.17623031e-01
-7.07892716e-01 2.84657776e-01 3.21220309e-02 -1.72394916e-01
9.55939710e-01 -4.83385980e-01 3.46032828e-02 1.57194689e-01
-1.08411694e+00 5.77717245e-01 1.75148800e-01 6.02312796e-02
4.99764889e-01 -9.78692591e-01 -1.77444264e-01 4.66441065e-01
-2.67413437e-01 7.86167264e-01 5.80510974e-01 1.22096038e+00
-1.32801700e+00 4.87313062e-01 -3.63683224e-01 -7.62833953e-01
-1.09043634e+00 2.72835314e-01 8.14511657e-01 -7.88148642e-01
-1.06508648e+00 1.06865108e+00 4.34336215e-01 -9.24322784e-01
3.85411024e-01 -7.75846362e-01 -4.55208600e-01 -1.70778081e-01
2.74676085e-01 3.56317937e-01 1.09403953e-01 -3.81670177e-01
8.79149046e-03 3.14826876e-01 -2.47449428e-01 5.48628747e-01
1.07929122e+00 -1.33351758e-01 -1.15625143e-01 -7.93225616e-02
9.78967428e-01 -2.95648336e-01 -1.37859094e+00 -3.09479058e-01
6.73969463e-02 -3.19415569e-01 2.26406693e-01 -7.79228926e-01
-1.28304970e+00 5.60919225e-01 8.39326978e-01 1.06369928e-01
9.54724014e-01 -1.98784381e-01 1.20100081e+00 -9.91741121e-02
2.21532911e-01 -1.04923725e+00 -2.90511638e-01 4.45629150e-01
6.11178577e-01 -1.32556331e+00 -1.55463889e-01 -6.87556207e-01
-4.95139688e-01 1.28641236e+00 9.22686279e-01 -3.27904344e-01
3.20022970e-01 2.40656167e-01 3.19913626e-01 -2.49594837e-01
5.43351889e-01 -3.43120587e-03 4.74086702e-01 2.48473212e-01
3.94671649e-01 2.19941717e-02 -8.01187694e-01 7.16776848e-01
-4.43058610e-02 3.67815085e-02 5.42878628e-01 1.01993215e+00
-2.52716631e-01 -7.17993498e-01 -5.41698158e-01 6.22661710e-01
-8.37565005e-01 1.20456062e-01 2.22752631e-01 9.62172091e-01
2.99345881e-01 1.96334258e-01 1.68688484e-02 1.67153388e-01
3.98024708e-01 -2.52103478e-01 6.41312838e-01 -4.83971864e-01
-7.41075993e-01 4.89751428e-01 -2.12495834e-01 -6.16729736e-01
-1.35887489e-01 -3.22632968e-01 -1.59976494e+00 5.72347715e-02
-3.78777742e-01 -5.03763929e-02 5.06041884e-01 9.72058713e-01
-1.74091250e-01 1.04258311e+00 2.56459445e-01 -1.07542360e+00
-5.99188805e-01 -1.06533861e+00 -6.80470765e-01 6.75284803e-01
5.18956423e-01 -5.93359470e-01 -6.96787909e-02 -8.61073285e-02]
|
[14.541326522827148, -2.6660821437835693]
|
4147e864-d338-4b2e-8419-3a439835db6a
|
coda-an-end-to-end-neural-program-decompiler
| null | null |
http://papers.nips.cc/paper/8628-coda-an-end-to-end-neural-program-decompiler
|
http://papers.nips.cc/paper/8628-coda-an-end-to-end-neural-program-decompiler.pdf
|
Coda: An End-to-End Neural Program Decompiler
|
Reverse engineering of binary executables is a critical problem in the computer security domain. On the one hand, malicious parties may recover interpretable source codes from the software products to gain commercial advantages. On the other hand, binary decompilation can be leveraged for code vulnerability analysis and malware detection. However, efficient binary decompilation is challenging. Conventional decompilers have the following major limitations: (i) they are only applicable to specific source-target language pair, hence incurs undesired development cost for new language tasks; (ii) their output high-level code cannot effectively preserve the correct functionality of the input binary; (iii) their output program does not capture the semantics of the input and the reversed program is hard to interpret. To address the above problems, we propose Coda1, the first end-to-end neural-based framework for code decompilation. Coda decomposes the decompilation task into of two key phases: First, Coda employs an instruction type-aware encoder and a tree decoder for generating an abstract syntax tree (AST) with attention feeding during the code sketch generation stage. Second, Coda then updates the code sketch using an iterative error correction machine guided by an ensembled neural error predictor. By finding a good approximate candidate and then fixing it towards perfect, Coda achieves superior with performance compared to baseline approaches. We assess Coda’s performance with extensive experiments on various benchmarks. Evaluation results show that Coda achieves an average of 82% program recovery accuracy on unseen binary samples, where the state-of-the-art decompilers yield 0% accuracy. Furthermore, Coda outperforms the sequence-to-sequence model with attention by a margin of 70% program accuracy. Our work reveals the vulnerability of binary executables and imposes a new threat to the protection of Intellectual Property (IP) for software development.
|
['Haolan Liu', 'Yuandong Tian', 'Huili Chen', 'Farinaz Koushanfar', 'Xinyun Chen', 'Jishen Zhao', 'Cheng Fu']
|
2019-12-01
| null | null | null |
neurips-2019-12
|
['computer-security']
|
['miscellaneous']
|
[ 4.95869339e-01 -1.23047881e-01 -7.27289855e-01 5.17877042e-02
-6.93628132e-01 -8.33368361e-01 2.08942682e-01 7.00982660e-02
1.15021087e-01 2.32351556e-01 -1.84918106e-01 -1.31277800e+00
4.57441509e-01 -7.52487004e-01 -1.06413591e+00 -1.58228859e-01
1.66462943e-01 -2.03174483e-02 3.00987512e-01 -1.91891909e-01
6.40170217e-01 2.80159473e-01 -1.24544537e+00 4.03808951e-01
9.99897003e-01 7.91767240e-01 4.50495705e-02 8.71578991e-01
-5.73495850e-02 8.03950548e-01 -7.03309298e-01 -7.33679473e-01
3.17390978e-01 -2.17118442e-01 -8.95588815e-01 -2.81539112e-01
1.55072987e-01 -6.55691683e-01 -4.28366393e-01 1.63359165e+00
-1.05151303e-01 -6.86327696e-01 4.74295259e-01 -1.35434341e+00
-9.37511444e-01 8.76407027e-01 -8.09218943e-01 1.72423705e-01
1.87132061e-01 5.65108001e-01 9.46236491e-01 -6.60913765e-01
3.62589449e-01 1.04422534e+00 5.91013134e-01 7.22509861e-01
-1.21794713e+00 -7.68417478e-01 -6.09554239e-02 9.13857222e-02
-1.15103626e+00 -2.73385376e-01 6.79081202e-01 -5.31998277e-01
1.34629643e+00 3.50150406e-01 1.59998268e-01 1.23560393e+00
5.94815373e-01 5.46635628e-01 8.21011603e-01 -2.20633358e-01
2.38706604e-01 1.51439071e-01 6.00828230e-01 9.21532929e-01
5.22717893e-01 4.67350692e-01 6.53666779e-02 -4.73882109e-01
3.42604145e-02 1.39069185e-01 -3.38677317e-01 1.12171263e-01
-7.80752063e-01 7.96468556e-01 3.90125275e-01 1.18137911e-01
3.19788605e-02 2.91495323e-01 7.99311757e-01 5.01624405e-01
-3.97902764e-02 6.52035534e-01 -4.79122847e-01 -3.26879919e-01
-9.98621821e-01 4.74615172e-02 8.70062351e-01 7.73171782e-01
5.69832504e-01 4.67597663e-01 1.11990131e-01 1.98361278e-01
4.63635772e-01 5.20840228e-01 7.15611696e-01 -3.91533524e-01
8.23440135e-01 8.38597775e-01 -4.25388038e-01 -1.18772078e+00
1.18327059e-01 -3.97909373e-01 -7.07453430e-01 5.67966521e-01
1.53975800e-01 8.67802128e-02 -8.22702050e-01 1.43126380e+00
-1.01955324e-01 1.00749418e-01 1.14762299e-01 4.14566249e-01
3.87904942e-01 6.98296607e-01 -2.60330737e-01 -2.51222979e-02
1.44303071e+00 -9.97232318e-01 -3.33199859e-01 -5.63844204e-01
6.74875617e-01 -4.59954411e-01 1.07363212e+00 3.36663216e-01
-6.82743907e-01 -4.44632649e-01 -1.61575997e+00 1.64284036e-01
-2.63770908e-01 5.22316918e-02 3.22477609e-01 1.11013365e+00
-8.98420036e-01 5.30511796e-01 -9.28622842e-01 3.07441145e-01
5.13659418e-01 4.43167180e-01 -2.10393533e-01 9.39301848e-02
-7.99213648e-01 3.94700021e-01 5.51270366e-01 -6.75772950e-02
-1.35117555e+00 -7.45772779e-01 -1.02735198e+00 2.80760497e-01
3.67339522e-01 -3.30970794e-01 1.17898452e+00 -1.07831740e+00
-1.40276241e+00 5.84151149e-01 -1.95990324e-01 -7.29310691e-01
2.22531959e-01 -1.16295688e-01 -5.65556407e-01 -1.45219624e-01
-3.55740376e-02 3.33827781e-03 1.32226133e+00 -1.14157093e+00
-4.58157182e-01 -1.94570780e-01 9.87036899e-02 -6.32640004e-01
-5.91051280e-01 1.95588410e-01 -1.53639421e-01 -7.15872824e-01
-4.35063332e-01 -1.00580645e+00 -4.02644463e-03 -3.81220877e-01
-7.61253715e-01 2.05833316e-01 1.18427336e+00 -1.09419441e+00
1.84053564e+00 -2.33566570e+00 1.01766065e-01 1.85133561e-01
5.08317530e-01 7.72807121e-01 -1.26202747e-01 5.51359020e-02
-5.02818465e-01 5.86974680e-01 -6.01381838e-01 -5.08678108e-02
-2.10539140e-02 -2.65391558e-01 -1.04014599e+00 3.91192406e-01
2.96586335e-01 1.23963630e+00 -7.90721714e-01 -1.61613598e-02
-2.00109273e-01 8.95040184e-02 -8.24404597e-01 2.39062309e-01
-5.04660964e-01 -4.20670025e-02 -3.12403232e-01 9.07711387e-01
7.00426579e-01 -4.15052593e-01 1.76903397e-01 -8.18552077e-02
1.03766046e-01 4.40449566e-01 -5.35688877e-01 1.06487250e+00
-5.66803694e-01 8.88908863e-01 -2.60197997e-01 -7.61814594e-01
7.82063425e-01 2.59838756e-02 -2.73564637e-01 -4.10378844e-01
2.22986147e-01 3.92347187e-01 2.71176219e-01 -4.52381492e-01
6.25399768e-01 1.64788947e-01 -3.50364298e-01 8.06612670e-01
-4.02858853e-01 7.16717169e-02 -2.52064049e-01 2.11743876e-01
1.48275137e+00 -3.67694907e-02 4.80060518e-01 6.43502772e-02
6.57288671e-01 7.35083744e-02 4.32056367e-01 6.26600266e-01
-1.22714482e-01 2.32350901e-01 9.23120975e-01 -3.85490566e-01
-1.10219502e+00 -7.24308848e-01 1.83422998e-01 6.93796873e-01
1.59339830e-02 -5.83414018e-01 -1.14121604e+00 -1.24057388e+00
-1.49344072e-01 1.05403459e+00 -5.39551258e-01 -7.01843560e-01
-8.99251580e-01 -6.79888070e-01 1.11550820e+00 6.67130291e-01
4.85052794e-01 -8.61883700e-01 -5.93455255e-01 -2.70738034e-03
-8.49124268e-02 -9.48145986e-01 -8.39779973e-01 8.99872556e-02
-7.41719007e-01 -1.18534732e+00 -1.53970152e-01 -6.65542543e-01
8.54950845e-01 1.60794482e-01 8.32351565e-01 5.68558753e-01
-1.65121421e-01 -2.50255704e-01 -2.49173611e-01 6.05282094e-03
-1.20121372e+00 1.76166177e-01 -1.13211177e-01 -1.51183799e-01
4.26189572e-01 -4.10668731e-01 -1.19110651e-01 1.53865397e-01
-1.09307587e+00 -1.44333705e-01 6.26567245e-01 9.62447703e-01
2.30923936e-01 4.37240601e-01 2.61450410e-01 -9.01990116e-01
5.16507328e-01 -5.62788844e-01 -1.04267848e+00 2.18024909e-01
-7.59290218e-01 3.87732744e-01 1.23267555e+00 -5.12799680e-01
-7.91172802e-01 -1.11466885e-01 -3.46108079e-01 -6.01276040e-01
1.56168327e-01 4.61189538e-01 -4.26990598e-01 -1.51705638e-01
8.38627875e-01 4.84535515e-01 9.30295587e-02 -2.18495890e-01
4.33991849e-02 8.92048478e-01 7.33258665e-01 -5.18763661e-01
1.30790329e+00 7.88743719e-02 -8.49931687e-02 -4.65978384e-01
-3.19358826e-01 2.08951280e-01 -3.51148188e-01 2.92989463e-01
6.83612347e-01 -5.27786136e-01 -8.01984906e-01 7.82414973e-01
-1.38284791e+00 -3.08939278e-01 1.59844741e-01 -1.97089285e-01
-1.17319100e-01 9.37533915e-01 -8.33293974e-01 -5.05815566e-01
-5.98991692e-01 -1.96007884e+00 9.42847729e-01 -6.79359585e-03
-3.18598896e-01 -6.63463116e-01 -2.36978799e-01 3.93983871e-01
3.22755933e-01 1.42642096e-01 1.62947130e+00 -7.45739222e-01
-8.08503389e-01 -3.38315904e-01 -2.25593910e-01 6.24512911e-01
8.13633874e-02 1.52661875e-01 -8.27310503e-01 -5.12617290e-01
1.87145472e-01 -1.85568094e-01 6.31071329e-01 -2.58343726e-01
1.31166804e+00 -7.14325190e-01 -3.70801777e-01 9.26251113e-01
1.37485600e+00 4.29488540e-01 7.46163130e-01 2.91511625e-01
8.21166337e-01 1.61171228e-01 3.65469247e-01 1.64798602e-01
2.44764052e-02 5.53542256e-01 9.37192380e-01 4.63088214e-01
-1.04863398e-01 -5.91983378e-01 9.54732537e-01 7.75268853e-01
4.34865713e-01 -1.86960325e-01 -1.10461044e+00 3.84641916e-01
-1.33772659e+00 -7.43378103e-01 -1.74236491e-01 2.33709097e+00
8.21784258e-01 4.77694839e-01 -5.41112795e-02 3.54037255e-01
7.03506887e-01 1.17058456e-01 -7.39792645e-01 -7.86663473e-01
3.05604845e-01 2.91986316e-01 7.81701267e-01 5.07030547e-01
-9.04527187e-01 9.06685591e-01 5.56300545e+00 9.73040164e-01
-1.33189976e+00 1.33253306e-01 5.95175982e-01 3.74203026e-01
-5.37546933e-01 3.60399663e-01 -1.03272927e+00 7.98259139e-01
1.34527206e+00 -3.19004178e-01 7.92454481e-01 1.21866035e+00
-4.26561415e-01 4.24565911e-01 -1.15350926e+00 6.50489211e-01
1.95045993e-01 -1.28803813e+00 -1.96482334e-02 2.49992952e-01
4.66830403e-01 -1.25972882e-01 3.89477938e-01 4.93633747e-01
3.67072344e-01 -1.22816503e+00 9.39728081e-01 -4.01655473e-02
1.15765572e+00 -8.32140744e-01 6.58661366e-01 3.97024035e-01
-1.13235641e+00 -4.84540194e-01 -1.95179686e-01 1.76143885e-01
-9.84150693e-02 2.83027977e-01 -9.84988511e-01 3.15341592e-01
4.25808549e-01 5.62960327e-01 -8.96123886e-01 5.32832801e-01
-5.14404833e-01 9.00491893e-01 1.76624462e-01 -8.96547958e-02
1.90689206e-01 2.73488194e-01 7.75426686e-01 1.17083776e+00
3.74194384e-01 -3.40824634e-01 -6.41179234e-02 1.24070799e+00
-3.42955589e-01 -4.54822510e-01 -6.52393341e-01 -5.29000342e-01
4.66812581e-01 8.99067998e-01 -6.15678072e-01 -1.73328787e-01
-3.64689648e-01 1.00611186e+00 2.72731125e-01 1.71701089e-01
-1.15787101e+00 -6.39326215e-01 7.73468912e-01 5.22429124e-02
4.24730927e-01 -1.20999351e-01 -5.73710203e-01 -1.25232065e+00
1.46916956e-01 -1.55952799e+00 1.85885429e-01 -4.54281121e-01
-7.36650348e-01 9.43038821e-01 -2.01883718e-01 -1.18206930e+00
-4.04941052e-01 -7.92394817e-01 -6.18890703e-01 8.17745864e-01
-1.31151652e+00 -1.00052595e+00 -1.80040207e-02 -1.82346571e-02
6.76562369e-01 -5.52243829e-01 6.65306449e-01 1.41322464e-01
-8.50283504e-01 1.12039912e+00 -6.17165864e-02 4.23857480e-01
8.00683200e-02 -9.81532931e-01 1.24743426e+00 1.50854313e+00
-2.09729552e-01 9.58844602e-01 5.33257544e-01 -1.10431826e+00
-1.88148522e+00 -1.29958475e+00 7.02784657e-01 -5.84476590e-01
1.05940866e+00 -4.88282949e-01 -1.14128160e+00 8.67178142e-01
1.40136024e-02 -8.68094638e-02 4.54364538e-01 -5.29951036e-01
-9.85444963e-01 1.89495072e-01 -1.00446987e+00 6.87803745e-01
6.78565085e-01 -8.70088518e-01 -4.64665204e-01 7.84485117e-02
1.20410287e+00 -3.90491188e-01 -5.05201161e-01 1.16661191e-01
5.07013619e-01 -8.97700548e-01 8.96130383e-01 -7.46677458e-01
1.04714012e+00 -5.38068533e-01 -2.85143197e-01 -9.46718574e-01
-2.10192189e-01 -8.11536670e-01 -6.69837892e-01 1.19392622e+00
5.18845379e-01 -7.39618897e-01 6.41249180e-01 3.82946372e-01
-1.24600835e-01 -7.80644238e-01 -7.22027540e-01 -8.94906461e-01
1.61588758e-01 -6.74945354e-01 9.38294113e-01 7.22145319e-01
-2.39782743e-02 2.34326318e-01 -4.07551140e-01 4.30656165e-01
5.58728039e-01 2.51360536e-01 6.92884803e-01 -6.05283082e-01
-9.72309232e-01 -6.13154292e-01 -3.52784425e-01 -1.06884205e+00
6.82383657e-01 -1.17083681e+00 -4.29804958e-02 -6.09786630e-01
3.38317424e-01 -2.25176185e-01 -1.02436002e-02 7.18330145e-01
-2.91666567e-01 9.79535729e-02 1.92835420e-01 2.16106847e-01
-3.18043046e-02 1.88177302e-01 5.11713088e-01 -5.96673429e-01
4.54934612e-02 1.05500497e-01 -9.84441578e-01 6.72339022e-01
6.85972095e-01 -7.77619362e-01 -3.53960752e-01 -4.26535100e-01
4.19844091e-01 1.57126635e-01 5.08875251e-01 -8.27217340e-01
3.94047163e-02 -1.77746072e-01 -2.62174428e-01 -3.64867389e-01
-1.83024019e-01 -8.09677303e-01 1.75189540e-01 1.05243576e+00
-1.51046455e-01 3.45761508e-01 2.76074350e-01 6.19571924e-01
-4.13595214e-02 -7.83931494e-01 8.82270575e-01 1.13953106e-01
-6.42481148e-01 1.74004585e-01 -3.46769780e-01 5.13992757e-02
1.08071709e+00 -2.53172517e-01 -6.33331120e-01 8.16247389e-02
3.87120023e-02 -2.79216468e-01 8.93049836e-01 6.27573371e-01
6.78189993e-01 -9.69367921e-01 -4.54442322e-01 6.16004348e-01
9.80639085e-02 -4.15601492e-01 -6.62737042e-02 5.46942532e-01
-7.40159333e-01 4.51095551e-01 -1.56900957e-02 -3.13812494e-01
-1.56273544e+00 1.12750673e+00 2.95319021e-01 -5.43127835e-01
-4.41764504e-01 5.73893189e-01 2.84097791e-01 -2.85176873e-01
-7.66698942e-02 -3.88752520e-01 1.68398336e-01 -6.25446558e-01
8.85390818e-01 2.54001826e-01 1.39449388e-01 -6.81469977e-01
-4.21888232e-01 2.41194427e-01 -4.19602931e-01 3.87249023e-01
1.12340379e+00 3.43118340e-01 -4.35377330e-01 -1.55203238e-01
1.50246656e+00 2.76485324e-01 -9.03274894e-01 8.28537915e-04
8.34268332e-02 -5.96249044e-01 5.44231990e-03 -7.34802008e-01
-1.07835734e+00 1.07449746e+00 2.34638304e-01 1.15128204e-01
1.18919230e+00 -3.48446280e-01 1.23842621e+00 2.66374528e-01
3.78775358e-01 -3.40582520e-01 2.36299947e-01 6.48512244e-01
4.69315290e-01 -9.76532698e-01 -1.64986178e-01 -3.73370737e-01
-3.61508787e-01 1.32081306e+00 6.71034932e-01 4.22566235e-02
3.22821289e-01 8.21895242e-01 -4.10371482e-01 1.25444695e-01
-6.20671093e-01 5.10685980e-01 2.63568938e-01 5.61158895e-01
-6.86786789e-03 5.30731976e-02 6.09743334e-02 9.24024045e-01
-2.52554148e-01 -2.34841213e-01 8.35950375e-01 1.01529491e+00
-1.83694690e-01 -1.25305748e+00 -4.98372972e-01 4.85219240e-01
-7.46628225e-01 -4.97558832e-01 -3.30747992e-01 4.59117532e-01
-1.70065150e-01 8.09157252e-01 -4.33267325e-01 -9.92209077e-01
-1.49718039e-02 -1.79130398e-02 5.15999086e-03 -6.74014449e-01
-8.84773672e-01 -4.62270945e-01 -1.28745630e-01 -5.08862853e-01
5.85978627e-01 -4.23810899e-01 -1.16628051e+00 -5.53024590e-01
-2.55747527e-01 -1.72238991e-01 5.83797216e-01 7.54152477e-01
6.83872521e-01 7.46350288e-01 7.10694790e-01 -5.15442431e-01
-1.01522470e+00 -4.69062746e-01 -5.35719767e-02 1.92619443e-01
6.51742518e-01 -2.53332675e-01 -5.05763173e-01 2.67756552e-01]
|
[7.058852195739746, 7.820750713348389]
|
9707fede-911e-448e-b0e4-0447f1f50ebe
|
steadyflow-spatially-smooth-optical-flow-for
| null | null |
http://openaccess.thecvf.com/content_cvpr_2014/html/Liu_SteadyFlow_Spatially_Smooth_2014_CVPR_paper.html
|
http://openaccess.thecvf.com/content_cvpr_2014/papers/Liu_SteadyFlow_Spatially_Smooth_2014_CVPR_paper.pdf
|
SteadyFlow: Spatially Smooth Optical Flow for Video Stabilization
|
We propose a novel motion model, SteadyFlow, to represent the motion between neighboring video frames for stabilization. A SteadyFlow is a specific optical flow by enforcing strong spatial coherence, such that smoothing feature trajectories can be replaced by smoothing pixel profiles, which are motion vectors collected at the same pixel location in the SteadyFlow over time. In this way, we can avoid brittle feature tracking in a video stabilization system. Besides, SteadyFlow is a more general 2D motion model which can deal with spatially-variant motion. We initialize the SteadyFlow by optical flow and then discard discontinuous motions by a spatial-temporal analysis and fill in missing regions by motion completion. Our experiments demonstrate the effectiveness of our stabilization on real-world challenging videos.
|
['Shuaicheng Liu', 'Ping Tan', 'Jian Sun', 'Lu Yuan']
|
2014-06-01
| null | null | null |
cvpr-2014-6
|
['video-stabilization']
|
['computer-vision']
|
[-1.52351066e-01 -4.32128757e-01 -2.88046986e-01 3.32852788e-02
-4.37936671e-02 -5.17659068e-01 3.48520070e-01 -3.43227297e-01
-1.97641432e-01 7.94386089e-01 4.43332762e-01 9.09918472e-02
1.27840236e-01 -3.97686660e-01 -6.01309478e-01 -8.72188270e-01
-6.16512299e-02 -6.68277204e-01 7.16897130e-01 1.81455025e-03
2.22523123e-01 4.12989825e-01 -1.04869807e+00 5.64175099e-03
7.78696835e-01 6.59601271e-01 3.00708324e-01 5.57821870e-01
-4.69136797e-02 9.71172333e-01 -2.84810603e-01 1.82472408e-01
3.15312028e-01 -5.90478241e-01 -7.84603834e-01 6.58504486e-01
7.59080827e-01 -6.37562394e-01 -7.11143136e-01 1.15535367e+00
5.60824238e-02 5.09957314e-01 3.84036498e-03 -1.18578780e+00
-5.08063316e-01 2.13506013e-01 -8.78717542e-01 5.57188809e-01
6.42760336e-01 4.96152222e-01 3.37787032e-01 -8.19184065e-01
1.01291049e+00 1.46416104e+00 6.22933328e-01 5.76098442e-01
-1.18744659e+00 -2.67266661e-01 6.07260108e-01 5.34128360e-02
-1.10852027e+00 -5.84100962e-01 5.86153865e-01 -5.13371050e-01
3.49405855e-01 3.07661325e-01 8.17892790e-01 6.16972804e-01
3.15798610e-01 7.87602067e-01 3.40112507e-01 -1.24353833e-01
4.12764139e-02 -7.71403372e-01 3.98820899e-02 8.94582033e-01
8.93375352e-02 -3.45038585e-02 -5.00818193e-01 -2.20892683e-01
1.38802373e+00 2.70250440e-01 -9.57899392e-01 -4.83929634e-01
-1.78780043e+00 4.03105289e-01 3.41401488e-01 2.54567921e-01
-3.52552295e-01 2.72653759e-01 3.51860404e-01 1.33443788e-01
3.55039477e-01 2.44730394e-02 -6.31977469e-02 3.68996188e-02
-1.15195811e+00 2.46125981e-01 2.85639912e-01 9.32863772e-01
1.03039765e+00 2.63822287e-01 -6.14179134e-01 2.95757920e-01
3.99807930e-01 4.31588203e-01 5.03461361e-01 -1.70990396e+00
1.37790322e-01 3.35796326e-01 5.00992417e-01 -1.23370564e+00
-1.40472323e-01 5.03711179e-02 -9.89256978e-01 4.74353544e-02
3.45175147e-01 -2.79376715e-01 -6.52419448e-01 1.55092549e+00
6.98137760e-01 1.15531158e+00 -1.33895740e-01 1.14920735e+00
5.26288986e-01 9.13168669e-01 -3.05031985e-02 -8.71576905e-01
8.65292549e-01 -1.13667285e+00 -1.19096744e+00 3.18139970e-01
3.99831325e-01 -7.98421383e-01 8.18365932e-01 1.36493728e-01
-1.28973567e+00 -6.57797396e-01 -8.00091743e-01 5.80547750e-02
3.69613916e-01 -1.77345201e-01 2.45120898e-01 -9.36459284e-03
-1.07777715e+00 8.20161879e-01 -1.24148619e+00 -1.86998114e-01
1.18866362e-01 4.10624556e-02 -4.89070863e-01 6.24790192e-02
-1.04673207e+00 2.49480382e-01 2.57911384e-01 3.46272051e-01
-6.81465209e-01 -8.10640216e-01 -1.08518147e+00 -9.54780802e-02
2.34001935e-01 -8.04282188e-01 1.09951437e+00 -1.16785014e+00
-1.53636611e+00 4.49931860e-01 -9.09179091e-01 -2.84814119e-01
7.03408480e-01 -4.40408498e-01 -3.62814575e-01 1.92218006e-01
1.48312360e-01 3.78139049e-01 1.11034071e+00 -1.15438509e+00
-5.35899758e-01 2.61176854e-01 -1.83886811e-01 1.50235653e-01
-2.27053061e-01 1.04636431e-01 -8.07868779e-01 -9.11358953e-01
2.47413740e-01 -8.91793668e-01 -4.93788332e-01 3.86424333e-01
-2.92915821e-01 1.48518354e-01 1.35246968e+00 -4.55899119e-01
1.78752613e+00 -2.44153929e+00 2.97257513e-01 -2.75734067e-02
3.03107798e-01 5.00103176e-01 -2.08626643e-01 2.96599232e-02
3.78668471e-03 -6.00710623e-02 -2.73022324e-01 -2.18019009e-01
-6.54639125e-01 1.41202301e-01 -4.51336384e-01 6.57664239e-01
1.20248139e-01 7.19915509e-01 -1.30214953e+00 -6.34058774e-01
4.74079221e-01 5.42133331e-01 -5.27409673e-01 2.45190620e-01
-4.86039817e-02 9.28813338e-01 -5.80582559e-01 4.47831988e-01
9.85389948e-01 -2.09818482e-01 -3.80028933e-02 -2.23925114e-01
-5.72275639e-01 -3.70340228e-01 -1.50549543e+00 1.90249240e+00
1.16685174e-01 6.60681546e-01 4.27666962e-01 -4.74263608e-01
6.72192395e-01 1.93358064e-01 8.93514752e-01 3.44443806e-02
1.15008667e-01 -7.96377286e-02 -4.68816668e-01 -8.01864445e-01
7.41578460e-01 3.70423257e-01 4.70535904e-01 4.26057056e-02
-2.87568927e-01 1.85255602e-01 3.57984215e-01 2.44739667e-01
8.23918104e-01 4.21151578e-01 1.29196504e-02 -4.60707814e-01
1.07314849e+00 -8.33877102e-02 1.20717812e+00 3.54168355e-01
-6.50841832e-01 1.01736701e+00 2.09393620e-01 -5.37167549e-01
-7.30618417e-01 -1.08169532e+00 1.93155259e-02 5.48433959e-01
8.10172081e-01 -7.20101595e-01 -7.70912290e-01 -3.58848870e-01
-9.00678188e-02 -1.98519081e-01 -4.95700806e-01 7.16052600e-04
-1.08529067e+00 -3.40054542e-01 3.42914462e-02 3.18046421e-01
7.06362724e-01 -7.54835546e-01 -5.70362926e-01 4.65240628e-01
-4.42393780e-01 -1.07809496e+00 -1.25477791e+00 -7.64242589e-01
-1.11495006e+00 -1.14969766e+00 -1.06621349e+00 -9.89835203e-01
7.38308430e-01 9.48060811e-01 8.89097810e-01 4.81133521e-01
2.57461652e-04 2.06252441e-01 -1.63496539e-01 5.49874306e-01
-2.06913024e-01 -3.69255692e-01 2.34622523e-01 5.04172742e-01
-9.42445397e-02 -3.45152527e-01 -9.38273609e-01 5.29268801e-01
-1.12029922e+00 -1.39969755e-02 -2.30093017e-01 8.12938035e-01
9.88498569e-01 -2.64141828e-01 9.43809226e-02 -4.73004431e-01
4.26496357e-01 -3.91210705e-01 -6.53419316e-01 2.26550490e-01
2.04381272e-01 -9.39677283e-02 5.33299506e-01 -6.41799748e-01
-1.04006529e+00 3.10996383e-01 1.97342306e-01 -1.19195402e+00
8.37188065e-02 2.17274919e-01 -4.97213453e-02 -1.68681562e-01
2.97297120e-01 3.70074958e-01 2.76714414e-01 -2.82517254e-01
4.60232675e-01 1.45278022e-01 9.90025163e-01 -3.86634827e-01
7.49259531e-01 9.42365944e-01 1.54314384e-01 -9.59982872e-01
-6.27803326e-01 -5.06841958e-01 -9.55740392e-01 -4.75767821e-01
1.04221416e+00 -9.14526820e-01 -8.37048292e-01 6.77645385e-01
-1.26508188e+00 -3.03258359e-01 -1.66637525e-01 7.38159716e-01
-4.76164579e-01 1.07675314e+00 -7.80284524e-01 -5.02658844e-01
-1.36809587e-01 -1.06034207e+00 8.38955462e-01 6.22860193e-01
-2.45351598e-01 -1.18962634e+00 3.81153315e-01 -3.41099203e-01
1.62873909e-01 4.83531147e-01 8.35075080e-02 5.65992713e-01
-8.63820910e-01 2.64259875e-01 9.31150392e-02 2.27363989e-01
4.48785663e-01 6.59283042e-01 -4.59351093e-01 -4.90713954e-01
-3.73074859e-02 3.31392288e-01 9.48008299e-01 8.24176133e-01
9.70231831e-01 -1.46480858e-01 -4.90383983e-01 9.67761576e-01
1.29234076e+00 7.41682723e-02 6.80718362e-01 3.56584489e-01
8.60264838e-01 2.33128205e-01 7.75171340e-01 4.02357489e-01
6.72444478e-02 5.42042434e-01 2.76014358e-01 -1.62769601e-01
-2.31206059e-01 -3.27840507e-01 5.31278372e-01 8.05819631e-01
-1.86438411e-01 6.14213049e-02 -4.47108537e-01 6.97585940e-01
-2.27856207e+00 -1.33385503e+00 -6.39585376e-01 2.16837502e+00
6.24921441e-01 -1.51686877e-01 6.63149506e-02 -4.84545268e-02
1.13446808e+00 5.12974739e-01 -4.91842270e-01 1.43803179e-01
-3.72948259e-01 -4.28065091e-01 4.24300343e-01 9.03179824e-01
-1.29824853e+00 9.77244258e-01 6.88843107e+00 3.89557183e-01
-1.35321808e+00 -6.68537468e-02 3.10750008e-01 -1.79640040e-01
-2.25813642e-01 1.30007774e-01 -5.61827481e-01 8.52105439e-01
1.78526863e-01 -2.68260986e-01 1.29396483e-01 5.28910875e-01
9.41921771e-01 -3.02461416e-01 -6.31645620e-01 1.04441345e+00
-1.47100538e-01 -1.67858589e+00 2.73247734e-02 -2.58412719e-01
1.02921224e+00 -3.09182942e-01 -9.07489434e-02 -3.41341168e-01
1.33389637e-01 -5.21370769e-01 6.78573012e-01 7.67656803e-01
5.31908810e-01 -4.52031374e-01 1.95198923e-01 1.62885398e-01
-1.66476965e+00 1.11638382e-01 -4.64156091e-01 3.15847807e-02
8.05280447e-01 6.49136007e-01 2.89058447e-01 6.48483634e-01
6.45132065e-01 1.49336934e+00 -2.14810371e-01 1.30433297e+00
-2.91964132e-02 3.20287764e-01 -6.79781735e-02 5.75031281e-01
3.17251056e-01 -6.33633733e-01 7.97552645e-01 1.10847509e+00
4.54056203e-01 3.22056532e-01 4.05421942e-01 5.26593447e-01
2.18726501e-01 -7.28259757e-02 -4.80157018e-01 5.77612817e-01
4.71001118e-01 1.18968868e+00 -4.82811630e-01 -5.21116734e-01
-5.65885484e-01 1.23424780e+00 -1.04979672e-01 6.87127948e-01
-8.93480122e-01 -2.37135768e-01 1.17495906e+00 2.25631356e-01
2.54585892e-01 -5.78311801e-01 1.70214161e-01 -1.77494311e+00
8.97065178e-02 -1.66078091e-01 3.04307252e-01 -8.90010118e-01
-1.07304525e+00 5.55842936e-01 -1.65778995e-01 -1.83415842e+00
4.39134724e-02 -7.18869194e-02 -1.00107348e+00 8.29324543e-01
-1.65591967e+00 -5.69483638e-01 -6.11814439e-01 1.12494576e+00
5.71922779e-01 2.93436646e-01 1.39612466e-01 1.79171905e-01
-8.15156162e-01 8.59095156e-02 2.41868809e-01 2.12506995e-01
9.30049419e-01 -7.37960935e-01 2.09222198e-01 1.28713369e+00
-2.52841204e-01 7.61731863e-01 5.92651069e-01 -8.06943893e-01
-1.22366214e+00 -1.34992969e+00 6.85240865e-01 -1.63780391e-01
6.59418106e-01 2.44114637e-01 -1.42593896e+00 6.83132470e-01
3.99361014e-01 6.15394831e-01 1.39924452e-01 -7.78631747e-01
1.78028345e-01 7.23535344e-02 -9.60235775e-01 5.68524778e-01
1.24812555e+00 -1.86662018e-01 -4.53050762e-01 4.67266217e-02
6.92178190e-01 -6.49686873e-01 -7.75281250e-01 2.36700282e-01
2.81622261e-01 -9.10080969e-01 9.32059765e-01 -5.37267208e-01
2.73943454e-01 -1.10927069e+00 2.39881933e-01 -1.16889453e+00
-7.01185882e-01 -1.35969579e+00 -4.67983931e-01 1.23911572e+00
-3.85089427e-01 -2.13280052e-01 7.61001587e-01 4.60393012e-01
-1.41508654e-01 -4.35511500e-01 -7.22461045e-01 -9.17400181e-01
-1.14086613e-01 4.58200313e-02 4.48421866e-01 1.17771256e+00
6.02557836e-03 -2.59549826e-01 -4.24106807e-01 3.11952949e-01
6.96795404e-01 9.12241172e-03 7.98764348e-01 -9.62500453e-01
1.31893262e-01 -3.33531827e-01 -4.40136045e-01 -1.57286787e+00
4.36106592e-01 -3.74725670e-01 5.89965731e-02 -1.35562265e+00
-3.23318169e-02 -1.55935828e-02 -1.02438115e-01 3.26740295e-02
-5.90543091e-01 3.50262746e-02 3.06251884e-01 7.42444694e-01
-4.61843610e-01 6.53984547e-01 1.67807353e+00 3.79222743e-02
-5.04561067e-01 -2.32251391e-01 7.62273818e-02 9.47247088e-01
4.53263551e-01 -9.76042300e-02 -3.91347528e-01 -6.81686878e-01
-4.96005058e-01 2.89402753e-01 2.52853841e-01 -9.87135828e-01
2.40964800e-01 -8.06927562e-01 3.14252615e-01 -5.04093051e-01
3.35302092e-02 -6.61759496e-01 2.88954079e-01 5.64519465e-01
-1.19395636e-01 3.15311193e-01 -2.50280816e-02 6.56034410e-01
-4.52482909e-01 2.02813577e-02 9.06269431e-01 -4.23871167e-02
-1.07395804e+00 6.88446224e-01 -4.88320768e-01 -1.13258705e-01
1.35136938e+00 -4.32205826e-01 -1.19181380e-01 -2.74562657e-01
-9.50204611e-01 5.13169944e-01 8.75726461e-01 3.85764539e-01
6.87764287e-01 -1.67382669e+00 -5.98413765e-01 3.73130709e-01
-3.12389374e-01 2.19740540e-01 4.81278270e-01 9.46592093e-01
-8.72343838e-01 -7.59266242e-02 -2.31118172e-01 -8.79500210e-01
-1.33156598e+00 8.01007032e-01 5.80189109e-01 1.86428592e-01
-9.59384263e-01 6.64236665e-01 4.30192173e-01 3.93259943e-01
9.39617530e-02 -4.66498196e-01 -2.06339329e-01 -1.70914859e-01
1.01650739e+00 5.47706783e-01 -4.47686166e-01 -8.79420757e-01
-4.13023889e-01 9.54555869e-01 3.72983187e-01 -2.80924160e-02
8.75354111e-01 -7.72551358e-01 -3.26580584e-01 2.94044495e-01
1.13527441e+00 1.74011096e-01 -2.14948654e+00 -1.06781319e-01
-2.79679120e-01 -1.13112974e+00 -9.01530385e-02 3.04406673e-01
-1.63737047e+00 4.19591665e-01 2.66814560e-01 1.33978069e-01
1.13220751e+00 -3.74035388e-01 1.05778265e+00 -5.41420057e-02
7.79490396e-02 -7.70002663e-01 -1.52513199e-02 7.24872947e-01
6.28902853e-01 -8.94725621e-01 7.86031261e-02 -7.16672361e-01
-5.23463845e-01 1.30023146e+00 6.96277082e-01 -4.66359496e-01
8.16479802e-01 1.41575590e-01 1.55063555e-01 3.37111712e-01
-5.87369561e-01 -9.42684785e-02 -4.59557623e-02 5.94171703e-01
4.42706168e-01 -4.10220563e-01 -5.03889501e-01 5.69444820e-02
5.82652569e-01 4.25742358e-01 8.33136976e-01 1.09304321e+00
-4.09506053e-01 -9.12036479e-01 -6.39999807e-01 -2.38772735e-01
-2.02299416e-01 2.69111902e-01 1.17243595e-01 3.66731614e-01
6.59875199e-02 9.73349333e-01 3.83909941e-01 -2.08138287e-01
3.35889816e-01 -2.75367469e-01 3.02382618e-01 -2.89885342e-01
-3.76190752e-01 4.18360978e-01 -4.66425121e-01 -9.50813234e-01
-8.42706740e-01 -8.33451331e-01 -1.51422298e+00 -6.39130533e-01
-1.62504777e-01 4.29056399e-02 -1.44683808e-01 5.71610868e-01
4.70058262e-01 3.60085398e-01 8.49315047e-01 -1.01189613e+00
2.50401720e-02 -5.82817137e-01 -5.62423527e-01 7.13924348e-01
8.80054057e-01 -6.79492950e-01 -2.22114280e-01 8.38094950e-01]
|
[10.650837898254395, -1.4319148063659668]
|
1729cb9d-9291-4086-8828-e2bbc275146c
|
spcl-a-new-framework-for-domain-adaptive
|
2111.12358
| null |
https://arxiv.org/abs/2111.12358v2
|
https://arxiv.org/pdf/2111.12358v2.pdf
|
SPCL: A New Framework for Domain Adaptive Semantic Segmentation via Semantic Prototype-based Contrastive Learning
|
Although there is significant progress in supervised semantic segmentation, it remains challenging to deploy the segmentation models to unseen domains due to domain biases. Domain adaptation can help in this regard by transferring knowledge from a labeled source domain to an unlabeled target domain. Previous methods typically attempt to perform the adaptation on global features, however, the local semantic affiliations accounting for each pixel in the feature space are often ignored, resulting in less discriminability. To solve this issue, we propose a novel semantic prototype-based contrastive learning framework for fine-grained class alignment. Specifically, the semantic prototypes provide supervisory signals for per-pixel discriminative representation learning and each pixel of source and target domains in the feature space is required to reflect the content of the corresponding semantic prototype. In this way, our framework is able to explicitly make intra-class pixel representations closer and inter-class pixel representations further apart to improve the robustness of the segmentation model as well as alleviate the domain shift problem. Our method is easy to implement and attains superior results compared to state-of-the-art approaches, as is demonstrated with a number of experiments. The code is publicly available at https://github.com/BinhuiXie/SPCL.
|
['Mingjia Li', 'Shuang Li', 'Binhui Xie']
|
2021-11-24
| null | null | null | null |
['synthetic-to-real-translation']
|
['computer-vision']
|
[ 4.69526738e-01 2.96220090e-03 -5.07779717e-01 -6.71835601e-01
-7.20813632e-01 -7.31441855e-01 4.11092788e-01 1.42104402e-01
-2.61047572e-01 5.91743827e-01 -1.59937307e-01 1.34740129e-01
-9.82841775e-02 -8.25319529e-01 -5.86711764e-01 -8.27446282e-01
5.10737598e-01 5.31050503e-01 4.42318290e-01 7.97441825e-02
2.81675279e-01 3.75151038e-01 -1.39318645e+00 2.59757787e-01
1.11120093e+00 1.04267883e+00 4.42240745e-01 -4.70413687e-03
-3.19924206e-01 1.91512764e-01 -4.95300591e-01 -1.10419514e-02
2.60197431e-01 -4.21444356e-01 -9.12721872e-01 4.20014352e-01
5.64521909e-01 3.77829233e-03 -5.07300049e-02 1.47979712e+00
2.50479281e-01 1.80031314e-01 7.50295341e-01 -1.12687230e+00
-7.11521208e-01 2.77512997e-01 -6.97922885e-01 1.27840400e-01
-1.08124211e-01 -7.58106858e-02 1.06800461e+00 -8.49445045e-01
5.57453275e-01 1.16166759e+00 2.97617078e-01 5.36593139e-01
-1.25286102e+00 -7.26096511e-01 5.58405221e-01 2.91111976e-01
-1.41083622e+00 -2.55522430e-01 1.12290144e+00 -4.23640966e-01
3.46399277e-01 3.75920869e-02 5.43411791e-01 9.77364421e-01
-3.33274156e-01 1.03342128e+00 1.12871170e+00 -3.77079070e-01
3.55749369e-01 2.07182795e-01 3.49806607e-01 4.56555396e-01
2.02992886e-01 9.13497643e-04 -3.49104553e-01 8.35558549e-02
8.92208695e-01 1.16346851e-01 -1.85302094e-01 -8.97520959e-01
-1.01670039e+00 8.84705067e-01 7.72072852e-01 4.10171896e-01
-2.98421234e-01 -2.55500197e-01 1.68343171e-01 -4.77923714e-02
5.81831872e-01 2.97955394e-01 -5.47409356e-01 3.12748104e-01
-1.06306911e+00 1.44204542e-01 4.08965796e-01 8.53931725e-01
1.19255388e+00 -1.61165684e-01 -7.73071591e-03 1.07355642e+00
3.24985802e-01 3.61162007e-01 6.06517136e-01 -1.04367507e+00
3.14535558e-01 8.10895741e-01 -1.63326517e-01 -9.02265847e-01
-2.97909290e-01 -4.71637219e-01 -5.82020462e-01 1.51866972e-01
7.19338715e-01 1.32090569e-01 -1.22530460e+00 1.62981582e+00
5.56622446e-01 2.47136161e-01 6.19226955e-02 1.10292637e+00
7.24613547e-01 4.99638826e-01 2.33502135e-01 1.35517299e-01
1.31390047e+00 -1.15470827e+00 -3.18590820e-01 -7.61970401e-01
3.77034754e-01 -6.59210682e-01 1.11737430e+00 3.74702252e-02
-7.19572842e-01 -7.30527580e-01 -9.40758348e-01 -2.61399131e-02
-4.29482400e-01 2.10134298e-01 4.73530203e-01 3.98629487e-01
-6.77426100e-01 3.56996775e-01 -7.40909934e-01 -4.66463596e-01
7.77693450e-01 2.06577048e-01 -2.89335459e-01 -2.57904649e-01
-1.10701263e+00 4.86867666e-01 8.40645194e-01 -3.84298339e-02
-6.57729387e-01 -5.73042095e-01 -9.18423474e-01 4.42444831e-02
3.25155675e-01 -4.03524786e-01 1.20613575e+00 -1.50117266e+00
-1.31965101e+00 1.08430898e+00 -2.45203435e-01 -1.02068640e-01
3.04331094e-01 -2.84873843e-02 -2.72430956e-01 3.19733024e-01
4.86453682e-01 1.01194286e+00 9.92663741e-01 -1.35444176e+00
-7.62333512e-01 -5.65067708e-01 -1.01236008e-01 4.25192505e-01
-3.83105844e-01 -2.44664535e-01 -6.35977328e-01 -8.13658476e-01
5.79813659e-01 -8.83522570e-01 -2.98367232e-01 2.32067913e-01
-3.22612524e-01 -1.51868135e-01 9.70140815e-01 -4.42400247e-01
8.20863068e-01 -2.39774013e+00 2.32102685e-02 1.59772411e-01
3.29985283e-02 4.02943373e-01 -9.16731134e-02 -1.48262046e-02
-4.20736261e-02 -2.02258974e-01 -6.74381673e-01 -6.10747468e-03
-1.20668903e-01 3.12587559e-01 -2.53529072e-01 4.35138762e-01
3.48974347e-01 7.88384795e-01 -9.09715116e-01 -5.58551431e-01
3.45410943e-01 3.00529093e-01 -3.90894592e-01 1.01915330e-01
-2.89051950e-01 8.02158952e-01 -8.59107614e-01 6.87910318e-01
8.39363813e-01 -3.86252463e-01 1.70795083e-01 -7.26835430e-02
1.70691743e-01 1.32762939e-01 -1.20767581e+00 1.84313273e+00
-7.30893314e-02 4.02661949e-01 4.66332324e-02 -1.49098432e+00
1.13718021e+00 -2.48489939e-02 3.07150990e-01 -9.02747273e-01
2.50298325e-02 3.42536360e-01 -1.76510260e-01 -1.57192349e-01
3.47848892e-01 -1.53341845e-01 -2.44627669e-01 1.68269768e-01
2.91492958e-02 -1.33727878e-01 2.88778823e-02 -1.46975398e-01
4.79318440e-01 3.08028281e-01 3.89234602e-01 -3.02967131e-01
6.13887310e-01 2.73637742e-01 8.97514105e-01 4.43786263e-01
-4.47659820e-01 7.21569657e-01 1.17956817e-01 -2.03982249e-01
-7.55408764e-01 -1.20914340e+00 -3.34353745e-01 1.14136386e+00
7.51440942e-01 9.51993242e-02 -9.48926389e-01 -9.57542181e-01
9.99388844e-02 6.20831907e-01 -3.60930830e-01 -2.34378070e-01
-5.22996008e-01 -5.17302573e-01 1.79130614e-01 7.72479773e-01
8.42075288e-01 -1.02366865e+00 -3.81262749e-01 1.37763858e-01
-2.35617921e-01 -1.09501874e+00 -3.93669099e-01 2.19711155e-01
-1.15341258e+00 -9.62376058e-01 -9.63214397e-01 -1.17759955e+00
9.66375053e-01 3.94381136e-01 8.26720178e-01 -3.22824240e-01
-2.04625025e-01 1.71898827e-01 -3.77692878e-01 -1.26103804e-01
-9.12759006e-02 2.13988975e-01 -2.84156054e-01 9.48547199e-02
7.00865328e-01 -3.55177671e-01 -7.10912287e-01 5.89154363e-01
-8.86500120e-01 -7.21912086e-03 5.31313837e-01 8.76570106e-01
1.01767886e+00 -4.38108742e-02 5.80452502e-01 -1.08030355e+00
2.45835111e-01 -5.35387695e-01 -6.05907559e-01 2.08166406e-01
-4.58215594e-01 -6.65983651e-03 6.16645098e-01 -2.90925354e-01
-1.18801630e+00 3.80167872e-01 1.04274474e-01 -2.83616453e-01
-6.26888752e-01 2.95844853e-01 -4.50710475e-01 6.75090030e-02
7.06600308e-01 2.26476148e-01 -1.99551299e-01 -5.68759441e-01
5.28422296e-01 7.26268768e-01 5.73299229e-01 -7.16244578e-01
7.38298178e-01 5.84415793e-01 -3.55289966e-01 -7.03660786e-01
-9.88803923e-01 -7.34103918e-01 -9.96234715e-01 1.53757939e-02
7.74177074e-01 -9.51955676e-01 1.89784259e-01 5.36349893e-01
-8.05264235e-01 -3.54610533e-01 -3.57308447e-01 2.98629045e-01
-5.67877114e-01 4.81996864e-01 -2.87292421e-01 -3.01377237e-01
-2.13753395e-02 -1.17893982e+00 1.11430144e+00 5.58359921e-01
-2.16078416e-01 -1.12218046e+00 -1.55335769e-01 4.64786410e-01
8.50335956e-02 2.45427862e-02 8.62638354e-01 -7.02170610e-01
-5.36402166e-01 -1.40786424e-01 -5.02021730e-01 3.99832010e-01
3.08360755e-01 -2.28782430e-01 -1.07379293e+00 -3.09756696e-01
-3.81375179e-02 -2.13592872e-01 9.59367931e-01 4.64669138e-01
1.26125062e+00 1.29395336e-01 -5.61205089e-01 5.40466368e-01
1.20648408e+00 1.39553681e-01 3.64321202e-01 4.85642791e-01
6.79690599e-01 7.50445008e-01 1.09322917e+00 2.01541260e-01
3.85704458e-01 7.60093689e-01 2.52231479e-01 -4.32829142e-01
-3.33364248e-01 -3.00975263e-01 6.71262667e-02 3.29628915e-01
4.33642656e-01 -1.06262006e-02 -9.82306719e-01 7.00293362e-01
-1.91116118e+00 -5.25949061e-01 1.54717967e-01 2.03922176e+00
7.77352870e-01 6.44141436e-03 3.56240124e-02 3.76934931e-02
1.05402935e+00 1.67237625e-01 -8.69336247e-01 -4.88033742e-02
-1.32123321e-01 1.07814603e-01 4.27427590e-01 2.85209388e-01
-1.37352288e+00 1.36163306e+00 5.52333403e+00 9.47954953e-01
-1.28519714e+00 -8.19753483e-03 6.31928861e-01 3.31701100e-01
-9.17443261e-02 -1.22315921e-02 -7.93231487e-01 5.36874115e-01
3.72850001e-01 1.47677716e-02 1.37106776e-01 1.05018544e+00
-4.43771034e-02 -1.28923625e-01 -1.02848232e+00 8.62878621e-01
-4.62339260e-02 -9.63494837e-01 7.52887651e-02 -1.04101531e-01
8.43603551e-01 -7.28185996e-02 1.73705131e-01 1.44016683e-01
1.38634592e-01 -7.94378340e-01 6.93969548e-01 9.20379087e-02
6.94274783e-01 -6.07228577e-01 3.95919889e-01 3.83667260e-01
-1.21027172e+00 -3.64481024e-02 -5.31043470e-01 8.35305825e-02
-7.07331300e-02 5.99056482e-01 -7.24493921e-01 4.68684405e-01
6.68394029e-01 8.63733947e-01 -4.81971830e-01 1.02977633e+00
-4.92525667e-01 4.87827241e-01 -1.25316516e-01 3.98141593e-01
4.11532611e-01 -2.17752308e-01 3.55284631e-01 1.02305949e+00
1.82252198e-01 -1.26799151e-01 5.14608502e-01 9.26839411e-01
-6.99689388e-02 9.40520093e-02 -3.25785100e-01 4.86590713e-02
6.33690894e-01 1.16870606e+00 -9.73579168e-01 -3.88346285e-01
-3.68678123e-01 1.11961162e+00 3.73692900e-01 6.76766574e-01
-7.06660986e-01 -3.12177867e-01 7.97164083e-01 2.95654908e-02
5.67244232e-01 -1.58831060e-01 -5.43719053e-01 -1.16893673e+00
9.49806124e-02 -6.65516078e-01 6.01018012e-01 -4.39055800e-01
-1.38093805e+00 3.82496238e-01 -1.57511625e-02 -1.33940887e+00
-1.10110819e-01 -6.42679036e-01 -4.43163484e-01 8.54111075e-01
-1.87460506e+00 -1.15862620e+00 -4.65086102e-01 6.77939892e-01
6.51646256e-01 -5.59408665e-02 7.51873255e-01 2.78316170e-01
-4.68108684e-01 6.53383076e-01 3.15914959e-01 2.37531573e-01
8.88921857e-01 -1.07016456e+00 2.19445542e-01 8.48230302e-01
1.80156767e-01 4.02081341e-01 4.17340428e-01 -5.31248569e-01
-6.44602358e-01 -1.21257269e+00 4.88719881e-01 -1.60338014e-01
5.09028494e-01 -1.72201782e-01 -1.16271544e+00 5.81779361e-01
-2.52357125e-01 2.57581860e-01 7.34295905e-01 -6.69061672e-03
-5.25897861e-01 -1.93551525e-01 -1.25728166e+00 3.60026509e-01
9.39139009e-01 -5.21389008e-01 -6.79469228e-01 1.53647974e-01
3.57197315e-01 -3.32883030e-01 -6.38898313e-01 3.65279853e-01
3.04480553e-01 -7.67205775e-01 9.35533345e-01 -2.33698562e-01
2.29319721e-01 -6.10195339e-01 -1.42454401e-01 -1.29796505e+00
-4.16383386e-01 1.45145282e-01 2.41336361e-01 1.29547906e+00
3.77337784e-01 -7.36285329e-01 1.01633680e+00 4.64894205e-01
-1.54680684e-01 -4.23096120e-01 -9.06376183e-01 -8.33830118e-01
3.04562688e-01 -2.72966832e-01 4.91866887e-01 1.05487573e+00
-2.69985259e-01 3.15058649e-01 5.92791252e-02 3.68461460e-01
7.12572098e-01 6.51712954e-01 5.30806243e-01 -1.39474964e+00
-1.59492001e-01 -5.14460504e-01 -4.76245701e-01 -1.34252632e+00
4.02564019e-01 -1.10101247e+00 1.93208694e-01 -1.55389988e+00
2.20159739e-01 -7.46993303e-01 -5.40493190e-01 6.49623513e-01
-3.17787558e-01 3.56500983e-01 1.72286704e-01 4.38124985e-01
-5.75687945e-01 4.86570418e-01 1.40620065e+00 -2.94152707e-01
-1.78953290e-01 2.41446178e-02 -8.82442713e-01 8.07272732e-01
1.00520313e+00 -5.02713919e-01 -4.85155106e-01 -4.83158678e-01
-5.61174154e-01 -3.73426139e-01 3.99568319e-01 -9.67826307e-01
1.29460365e-01 -1.60210431e-01 5.75001001e-01 -3.25358927e-01
2.24358886e-01 -8.50587547e-01 -1.69547051e-01 2.49453932e-01
-2.31396005e-01 -7.53411055e-01 2.11521119e-01 5.54328203e-01
-5.58801830e-01 -2.96846062e-01 1.18340933e+00 -1.18188567e-01
-1.35448456e+00 2.58876890e-01 -5.47021665e-02 1.72855020e-01
1.16054809e+00 -5.96589744e-01 -2.97822297e-01 6.83423728e-02
-6.69517756e-01 3.34441841e-01 8.42549384e-01 4.25958455e-01
4.52463269e-01 -1.20034409e+00 -3.68600428e-01 4.32479829e-01
4.59066421e-01 3.17453444e-01 3.95961702e-01 5.70303619e-01
-2.68169910e-01 2.99199194e-01 -4.26574141e-01 -8.98214579e-01
-1.12454104e+00 5.03953338e-01 3.34763199e-01 6.74100651e-04
-5.42929351e-01 9.40952420e-01 8.44078422e-01 -5.93510091e-01
1.60104468e-01 -1.07772246e-01 -2.06234649e-01 1.01891331e-01
2.87160188e-01 1.72677726e-01 -1.25872746e-01 -7.81051397e-01
-4.26610142e-01 1.00208879e+00 -2.65064865e-01 2.15510517e-01
1.03234005e+00 -3.34993094e-01 1.00200929e-01 2.67151713e-01
1.21849597e+00 -3.08236718e-01 -1.62868357e+00 -6.33412778e-01
1.56431645e-01 -5.20115256e-01 1.37576845e-03 -8.23312461e-01
-1.19060266e+00 9.76446331e-01 8.02199721e-01 -1.62495479e-01
1.32003725e+00 3.49283159e-01 7.33654499e-01 1.04852788e-01
2.78881937e-01 -1.25856245e+00 1.19144067e-01 4.04622793e-01
4.01799917e-01 -1.49661410e+00 -1.57135040e-01 -7.34456122e-01
-8.86849165e-01 9.95764315e-01 6.29169822e-01 -2.18099758e-01
4.33902234e-01 -2.15024948e-01 3.74400765e-01 -1.29157037e-01
-1.47695309e-02 -3.07126641e-01 4.48698014e-01 7.17330933e-01
3.26493979e-01 2.39993602e-01 -2.66238526e-02 5.02448738e-01
8.58507827e-02 -1.57997310e-01 1.38013186e-02 8.74616146e-01
-6.11287594e-01 -1.32664883e+00 -3.44150782e-01 1.10222273e-01
-2.01390341e-01 4.80217226e-02 -4.18197215e-01 7.16034174e-01
1.06544338e-01 8.37140381e-01 1.40813857e-01 -3.45589742e-02
3.02369893e-01 3.45329531e-02 3.04380238e-01 -8.00313294e-01
-1.64687738e-01 2.89234161e-01 -2.53954351e-01 -4.85617608e-01
-5.18550515e-01 -6.62396967e-01 -1.57841635e+00 2.13823035e-01
-1.66313469e-01 9.94298309e-02 4.13199455e-01 8.50470662e-01
3.64634246e-01 3.90037090e-01 5.26159406e-01 -7.72566557e-01
-4.14156705e-01 -6.08885646e-01 -6.77724481e-01 5.23767471e-01
1.33087307e-01 -9.01863337e-01 -1.35291353e-01 7.32966214e-02]
|
[9.676398277282715, 1.306711196899414]
|
5251eeaf-ff48-49ce-9ec2-4103c79b9bba
|
graph-boosted-active-learning-for-multi
| null | null |
https://link.springer.com/chapter/10.1007%2F978-3-030-88361-4_11
|
https://link.springer.com/content/pdf/10.1007%2F978-3-030-88361-4_11.pdf
|
Graph-boosted Active Learning for Multi-Source Entity Resolution
|
Supervised entity resolution methods rely on labeled record pairs for learning matching patterns between two or more data sources. Active learning minimizes the labeling effort by selecting informative pairs for labeling. The existing active learning methods for entity resolution all target two-source matching scenarios and ignore signals that only exist in multi-source settings, such as the Web of Data. In this paper, we propose ALMSER, a graph-boosted active learning method for multi-source entity resolution. To the best of our knowledge, ALMSER is the first active learning-based entity resolution method that is especially tailored to the multi-source setting. ALMSER exploits the rich correspondence graph that exists in multi-source settings for selecting informative record pairs. In addition, the correspondence graph is used to derive complementary training data. We evaluate our method using five multi-source matching tasks having different profiling characteristics. The experimental evaluation shows that leveraging graph signals leads to improved results over active learning methods using margin-based and committee-based query strategies in terms of F1 score on all tasks.
|
['Christian Bizer', 'Anna Primpeli']
|
2021-09-30
| null | null | null |
international-semantic-web-conference-2021-9
|
['entity-resolution']
|
['natural-language-processing']
|
[ 2.02749759e-01 6.67397499e-01 -1.27816129e+00 -5.03972590e-01
-1.69985485e+00 -4.74112362e-01 6.02639139e-01 9.76734042e-01
-3.79807204e-01 8.70121419e-01 3.30525666e-01 1.27089605e-01
-5.19812286e-01 -8.24465275e-01 -8.78242671e-01 -9.06962156e-02
-2.48852074e-01 8.15737307e-01 4.45930004e-01 -1.56939045e-01
3.78115475e-02 2.50523806e-01 -1.04367721e+00 4.51385647e-01
1.10156882e+00 6.63641572e-01 -2.23687142e-01 1.35904297e-01
-5.60899138e-01 1.07349539e+00 -4.22005445e-01 -8.36808205e-01
3.01866710e-01 -1.26267344e-01 -9.20728505e-01 -4.89371896e-01
7.62445748e-01 2.17815027e-01 -1.08171061e-01 6.63359821e-01
5.91284096e-01 2.55335942e-02 4.44442600e-01 -1.29089367e+00
-4.89445210e-01 1.31448960e+00 -5.78063309e-01 3.65728885e-01
7.33152628e-01 -4.14295614e-01 1.67087030e+00 -1.10657847e+00
9.60404932e-01 9.60306525e-01 8.24052811e-01 3.19915235e-01
-1.50672019e+00 -7.79824317e-01 3.56805086e-01 3.01453412e-01
-1.41947341e+00 -9.50073600e-01 8.13778877e-01 -2.38224626e-01
8.98434281e-01 4.14469689e-01 1.06155358e-01 1.02366590e+00
-3.82103980e-01 1.11190486e+00 5.66177607e-01 -8.32968652e-01
2.56357700e-01 4.26986009e-01 4.53963637e-01 7.46173680e-01
8.15210998e-01 -1.17619351e-01 -1.22411108e+00 -8.13262999e-01
1.75996989e-01 -1.65461972e-01 -9.56681669e-02 -8.02848399e-01
-1.01526546e+00 8.10460925e-01 3.22907418e-01 2.62961954e-01
-3.96395892e-01 -1.39110819e-01 1.91979650e-02 3.04528534e-01
6.66110873e-01 8.50243032e-01 -5.70110738e-01 4.04938757e-01
-1.08889401e+00 -7.26110488e-02 1.17066169e+00 1.39599657e+00
1.10451365e+00 -5.16218424e-01 -3.34926188e-01 9.16322231e-01
5.56894124e-01 2.29561538e-01 -5.40191829e-02 -7.05577731e-01
1.04378927e+00 1.23184907e+00 1.23995349e-01 -8.93585920e-01
-4.40383554e-01 -3.78266305e-01 -2.80191749e-01 -2.55213648e-01
3.22056741e-01 -1.69880956e-01 -3.93795639e-01 1.69631314e+00
5.44197083e-01 3.46631482e-02 5.94367459e-02 4.60507244e-01
1.30383801e+00 1.01221427e-01 3.46330851e-01 -5.27517617e-01
9.94571865e-01 -8.40964735e-01 -8.07200134e-01 -3.72971117e-01
7.56207943e-01 -5.28103232e-01 5.19926488e-01 -1.62673309e-01
-8.18391204e-01 -7.02705011e-02 -1.13419294e+00 1.34161830e-01
-4.92785066e-01 4.12033778e-03 9.68591511e-01 5.48717678e-01
-6.27675533e-01 3.51768792e-01 -6.92624807e-01 -4.09499526e-01
5.86139679e-01 3.27580839e-01 -4.48218584e-01 3.59414704e-02
-1.54730678e+00 7.93941498e-01 3.82689059e-01 -5.11199296e-01
-1.99702591e-01 -1.19338727e+00 -9.28855062e-01 1.80604115e-01
9.61065531e-01 -5.90742350e-01 9.77748752e-01 -6.93418384e-01
-6.74004197e-01 9.84734893e-01 -2.28258505e-01 -5.87271094e-01
4.86208826e-01 -1.41695321e-01 -7.04984128e-01 -5.46490736e-02
3.92763644e-01 2.31068701e-01 3.56173962e-01 -1.26561606e+00
-5.93146026e-01 -3.35696012e-01 1.15585268e-01 1.49455398e-01
-3.36389452e-01 -9.49706789e-03 -6.64296210e-01 -4.87197816e-01
-8.25491827e-03 -7.31959581e-01 -1.47011518e-01 -1.34812623e-01
-4.68632013e-01 -5.85337579e-01 3.63375962e-01 -2.83341795e-01
1.56364167e+00 -1.67774427e+00 -1.26560986e-01 3.96231592e-01
4.78807628e-01 1.63477883e-01 1.38872176e-01 8.17161083e-01
7.20760301e-02 1.34359464e-01 -1.31180780e-02 -6.53418183e-01
1.27148122e-01 -1.36061355e-01 -2.41998255e-01 2.75034130e-01
1.12183496e-01 1.10996664e+00 -9.89832819e-01 -1.13666761e+00
-3.93507153e-01 2.85668788e-03 -1.15043081e-01 2.61514246e-01
-2.15417817e-01 -3.75532694e-02 -4.63116080e-01 8.83676887e-01
3.74771059e-01 -6.25018716e-01 5.65232098e-01 -1.41290292e-01
1.14724152e-01 3.37600291e-01 -1.30250371e+00 1.95106483e+00
-3.31661463e-01 3.85700107e-01 -4.47983220e-02 -7.85112262e-01
9.98068631e-01 3.58505934e-01 7.76550233e-01 -8.55506361e-01
-5.03549874e-01 3.89145494e-01 -4.07256573e-01 -3.95748705e-01
5.18474162e-01 6.12834513e-01 -3.12023610e-01 5.40109634e-01
2.95137733e-01 7.38989234e-01 4.97561663e-01 7.93563902e-01
1.25345981e+00 -4.24664803e-02 6.96721733e-01 -1.17418863e-01
3.75004858e-01 1.37466952e-01 9.79177117e-01 1.17983937e+00
7.63187930e-02 2.97793150e-01 1.86941087e-01 7.19102025e-02
-5.79032779e-01 -9.25529301e-01 -7.02638999e-02 1.43581676e+00
3.56954157e-01 -8.09975922e-01 -2.70736307e-01 -1.06287241e+00
3.58708262e-01 5.33082485e-01 -5.03279865e-01 1.00185946e-02
-7.05634773e-01 -8.62303257e-01 5.21328628e-01 3.96396071e-01
1.20585598e-01 -7.98721910e-01 3.33970599e-02 2.54570782e-01
-2.42843524e-01 -9.81889606e-01 -3.41521204e-01 1.46175444e-01
-6.43512249e-01 -1.45398784e+00 -2.42641076e-01 -6.14606380e-01
4.06232834e-01 -6.99292943e-02 1.59558749e+00 -6.83359150e-03
-2.50642598e-01 6.75451577e-01 -3.39277208e-01 -6.03264630e-01
-1.78350732e-01 7.64313579e-01 -3.28272223e-01 6.30280599e-02
7.31696308e-01 -3.72969657e-01 -3.24102640e-01 2.16618463e-01
-4.26639169e-01 -2.02699408e-01 7.09639907e-01 5.12882233e-01
6.79758430e-01 -4.20844734e-01 1.03606761e+00 -1.85720181e+00
4.58824992e-01 -1.03457355e+00 -6.34894371e-01 9.77189958e-01
-1.37097239e+00 2.84025639e-01 2.11508989e-01 -3.65252435e-01
-1.24734867e+00 2.87367582e-01 3.96596700e-01 1.44651100e-01
1.51774749e-01 5.64057946e-01 -3.69642526e-01 -7.58929476e-02
1.16732895e+00 -4.39624548e-01 -6.71917975e-01 -6.27199292e-01
3.34882528e-01 5.11207819e-01 3.79064351e-01 -4.97731090e-01
9.96446788e-01 2.77320713e-01 -1.32765114e-01 -2.00833172e-01
-1.01580286e+00 -9.99508858e-01 -8.07830811e-01 -6.22715652e-02
3.59522223e-01 -1.15621901e+00 -3.00963998e-01 2.21745856e-03
-8.25254440e-01 6.27416447e-02 -3.14369887e-01 4.45247799e-01
-2.64279217e-01 1.47264600e-01 -3.58952314e-01 -8.15752268e-01
-6.06564283e-01 -4.80755240e-01 8.71229351e-01 2.79149443e-01
-4.07954961e-01 -1.18057251e+00 5.33896983e-01 6.57790363e-01
4.29592162e-01 3.16118926e-01 6.65299416e-01 -1.33861160e+00
-9.61059153e-01 -3.52424413e-01 5.95450141e-02 -8.40597093e-01
1.98248148e-01 -4.82210189e-01 -8.94550323e-01 -2.66742051e-01
-7.49819398e-01 -4.30156589e-01 9.22466576e-01 -2.02457368e-01
4.20601994e-01 -4.98548985e-01 -8.19953144e-01 3.57458562e-01
1.44544578e+00 4.15898003e-02 2.90180504e-01 4.62945700e-01
7.44694769e-01 5.82768798e-01 8.92485917e-01 2.37305343e-01
6.26522303e-01 1.05124927e+00 4.66640890e-02 -5.04481733e-01
-2.31785625e-01 -5.74366033e-01 -4.47384901e-02 4.25832868e-01
3.20527554e-01 -3.72614771e-01 -1.05243123e+00 6.49087131e-01
-2.29002166e+00 -1.01329529e+00 -1.32698178e-01 2.31094623e+00
1.35690236e+00 1.98377505e-01 7.73587599e-02 -6.14634268e-02
9.33164597e-01 2.01645374e-01 -7.55478680e-01 3.51374418e-01
-3.18302065e-01 1.04570538e-01 6.93678319e-01 4.94637221e-01
-1.24720275e+00 6.28845513e-01 5.45750475e+00 7.09379077e-01
-6.08034432e-01 2.55892694e-01 1.35597020e-01 -9.66108441e-02
-3.94283324e-01 3.71309102e-01 -1.27319252e+00 3.44686866e-01
7.36023784e-01 -4.12766069e-01 -5.42234071e-02 9.16718125e-01
-4.85826731e-01 8.23187679e-02 -1.48905337e+00 8.62672925e-01
5.85056357e-02 -1.64461970e+00 -1.93493009e-01 -3.19081731e-02
7.52424359e-01 9.46345255e-02 -4.30832744e-01 3.55208635e-01
6.54086232e-01 -7.46911287e-01 5.05718052e-01 7.72625685e-01
6.57078266e-01 -4.34186071e-01 5.55109441e-01 4.31755543e-01
-1.31583941e+00 7.08377287e-02 4.65892284e-04 6.50042117e-01
1.14713855e-01 6.87649429e-01 -1.18063426e+00 8.97591770e-01
4.54656392e-01 5.91793060e-01 -8.97190571e-01 1.21250927e+00
-3.74780856e-02 7.63051331e-01 -2.69508183e-01 1.38332874e-01
-4.96896982e-01 3.35490227e-01 6.32634699e-01 1.47919154e+00
-2.61117965e-01 -8.58244225e-02 3.31292689e-01 7.82735765e-01
-6.14400387e-01 4.35389549e-01 -5.55907786e-01 -4.50379625e-02
1.19874656e+00 1.30358887e+00 -3.90151620e-01 -1.77751288e-01
-6.33105755e-01 4.70362693e-01 7.51618266e-01 3.33723187e-01
-4.26000327e-01 -4.58095938e-01 5.41398674e-03 3.71621847e-01
-1.30606731e-02 1.86140314e-01 -2.09003195e-01 -1.36476576e+00
1.10684624e-02 -5.99075913e-01 1.14081597e+00 -2.36021593e-01
-1.44441700e+00 1.99319050e-01 2.96960324e-01 -9.79911149e-01
-4.47821498e-01 2.26353928e-01 -4.54827487e-01 6.17709458e-01
-1.60547042e+00 -1.30012167e+00 -3.01665485e-01 5.24826527e-01
2.81469643e-01 -5.54006577e-01 9.06454265e-01 7.45972693e-01
-6.03907406e-01 1.28640473e+00 -4.00244346e-04 7.12338448e-01
1.20072448e+00 -1.50133133e+00 4.94281381e-01 9.34608519e-01
7.08736658e-01 9.34450388e-01 3.30402434e-01 -7.75542915e-01
-1.42495680e+00 -1.01601911e+00 1.40630174e+00 -6.96640611e-01
4.53810692e-01 -7.19891250e-01 -1.11153233e+00 8.42742205e-01
6.70351982e-02 1.36256903e-01 1.17536509e+00 8.02119315e-01
-7.74077296e-01 -3.33237708e-01 -1.21907032e+00 2.09429637e-01
1.34991384e+00 -6.02082014e-01 -6.39647603e-01 3.12198341e-01
6.60915613e-01 -3.23478460e-01 -1.25169313e+00 6.06012166e-01
2.54095793e-01 -4.75508600e-01 1.13412273e+00 -8.04666877e-01
-8.69188458e-02 -1.33958146e-01 -2.19139308e-02 -9.11063492e-01
-4.55503076e-01 -7.36685932e-01 -7.41461694e-01 1.95699346e+00
1.24929464e+00 -6.57086611e-01 9.14515793e-01 8.67531478e-01
1.52404219e-01 -5.07370532e-01 -7.80923784e-01 -4.94432539e-01
-3.85872245e-01 1.01313978e-01 6.94819033e-01 1.48570263e+00
2.49793559e-01 6.47430122e-01 -2.13807479e-01 2.14214310e-01
1.06917048e+00 5.89851141e-01 7.96203494e-01 -1.65457428e+00
-3.76388848e-01 3.57586406e-02 -1.01760715e-01 -6.40343726e-01
3.22320223e-01 -1.18018246e+00 -3.66185576e-01 -1.56264901e+00
3.53001058e-01 -1.25100374e+00 -3.84681672e-01 6.67856336e-01
-4.11823750e-01 6.15507970e-03 -4.30662408e-02 6.21233463e-01
-1.33690631e+00 4.45168056e-02 2.27903605e-01 -3.03199708e-01
-2.98768550e-01 1.36599183e-01 -9.37719941e-01 2.56958008e-01
6.35943174e-01 -7.36230671e-01 -5.07571876e-01 -1.29563078e-01
6.04699910e-01 2.98778862e-01 -2.08096877e-01 -5.52682281e-01
9.26885307e-01 -3.36065888e-01 3.01706344e-01 -4.43853885e-01
9.15673673e-02 -6.28228366e-01 4.73039657e-01 -1.66053578e-01
-1.04226804e+00 -3.97169232e-01 -1.93749726e-01 9.11829293e-01
-1.26649216e-01 -1.69406265e-01 4.26303089e-01 -6.09463826e-02
-9.28134799e-01 4.57735628e-01 2.10596114e-01 6.89239621e-01
7.29328454e-01 4.47558984e-02 -6.95280850e-01 -1.33635879e-01
-6.50145113e-01 4.19731200e-01 3.67911875e-01 6.21375978e-01
1.35871053e-01 -1.45268047e+00 -8.54780734e-01 -2.23322153e-01
7.99093127e-01 -8.04781169e-03 -2.29083896e-01 7.30245829e-01
2.14463532e-01 4.49769437e-01 1.92852229e-01 -3.28909427e-01
-1.58263171e+00 4.28498089e-01 1.68224573e-01 -6.19794488e-01
-3.38725477e-01 6.60443783e-01 -2.19886243e-01 -4.81813371e-01
5.28695345e-01 5.56855381e-01 -6.25267088e-01 5.90160251e-01
2.92488754e-01 5.10178089e-01 5.97139716e-01 -3.45698327e-01
-6.32806480e-01 8.12821910e-02 -3.10449690e-01 1.54551402e-01
1.33435512e+00 -1.88226938e-01 1.02165364e-01 5.53178906e-01
1.02935469e+00 7.16594934e-01 -7.31325984e-01 -1.07573259e+00
9.70296681e-01 -3.84415001e-01 -2.05779687e-01 -9.69520688e-01
-1.07055998e+00 -1.56692434e-02 3.75890583e-01 1.79664403e-01
7.37083137e-01 3.29868644e-01 4.78413135e-01 5.88873982e-01
7.67845929e-01 -1.18246973e+00 -8.60158503e-02 -7.09060729e-02
4.98484999e-01 -1.65030551e+00 4.84809339e-01 -7.65996933e-01
-4.74690586e-01 6.09206200e-01 6.58553064e-01 3.29062492e-01
4.70331341e-01 3.22988957e-01 5.56583665e-02 -4.28797811e-01
-1.07368791e+00 -3.00025582e-01 6.25695169e-01 5.21769941e-01
5.39868116e-01 -3.40143703e-02 -4.78243083e-01 6.14047766e-01
2.34085709e-01 -2.70874888e-01 1.56143367e-01 1.10440254e+00
-2.72733957e-01 -1.47872019e+00 -2.08091304e-01 5.63947082e-01
-5.15747726e-01 -1.49066478e-01 -1.00149858e+00 7.26792336e-01
7.86112472e-02 1.16627228e+00 -1.26353562e-01 -5.41499350e-03
3.71130079e-01 2.30375230e-01 3.52996469e-01 -8.49252641e-01
-9.08845127e-01 -4.61706012e-01 7.92204976e-01 -5.79971015e-01
-7.51684606e-01 -7.10295320e-01 -1.18862486e+00 -1.39839798e-01
-7.64300406e-01 5.74810266e-01 2.10031822e-01 5.82947254e-01
7.44156301e-01 -8.34954605e-02 4.62544173e-01 1.32882997e-01
-2.32245788e-01 -7.29293823e-01 -3.24602216e-01 5.49369991e-01
2.89566845e-01 -7.75217950e-01 -1.79044962e-01 -1.59498349e-01]
|
[9.364476203918457, 8.618194580078125]
|
222c6d5d-a661-4132-b3da-57e70890e4a0
|
tyolov5-a-temporal-yolov5-detector-based-on
|
2111.08867
| null |
https://arxiv.org/abs/2111.08867v2
|
https://arxiv.org/pdf/2111.08867v2.pdf
|
TYolov5: A Temporal Yolov5 Detector Based on Quasi-Recurrent Neural Networks for Real-Time Handgun Detection in Video
|
Timely handgun detection is a crucial problem to improve public safety; nevertheless, the effectiveness of many surveillance systems still depends of finite human attention. Much of the previous research on handgun detection is based on static image detectors, leaving aside valuable temporal information that could be used to improve object detection in videos. To improve the performance of surveillance systems, a real-time temporal handgun detection system should be built. Using Temporal Yolov5, an architecture based on Quasi-Recurrent Neural Networks, temporal information is extracted from video to improve the results of handgun detection. Moreover, two publicly available datasets are proposed, labeled with hands, guns, and phones. One containing 2199 static images to train static detectors, and another with 5960 frames of videos to train temporal modules. Additionally, we explore two temporal data augmentation techniques based on Mosaic and Mixup. The resulting systems are three temporal architectures: one focused in reducing inference with a mAP$_{50:95}$ of 55.9, another in having a good balance between inference and accuracy with a mAP$_{50:95}$ of 59, and a last one specialized in accuracy with a mAP$_{50:95}$ of 60.2. Temporal Yolov5 achieves real-time detection in the small and medium architectures. Moreover, it takes advantage of temporal features contained in videos to perform better than Yolov5 in our temporal dataset, making TYolov5 suitable for real-world applications. The source code is publicly available at https://github.com/MarioDuran/TYolov5.
|
['Leonardo Chang', 'Cuauhtemoc Daniel Suarez-Ramirez', 'Miguel Gonzalez-Mendoza', 'Mario Alberto Duran-Vega']
|
2021-11-17
| null | null | null | null |
['image-augmentation']
|
['computer-vision']
|
[ 1.51328044e-02 -1.40252665e-01 -1.85311139e-01 1.12077389e-02
-5.87542295e-01 -4.67692941e-01 3.23698163e-01 -2.36573160e-01
-6.14628255e-01 2.75183648e-01 -1.03427604e-01 -2.72825330e-01
3.89749222e-02 -9.35299754e-01 -8.16816568e-01 -8.11556041e-01
-1.59819156e-01 -1.26896063e-02 7.60235846e-01 -3.35913062e-01
-1.50561765e-01 3.86176795e-01 -1.45348704e+00 3.17059636e-01
4.70909774e-01 1.03210354e+00 2.69662976e-01 9.23864245e-01
4.30810273e-01 9.98847842e-01 -7.70820498e-01 -5.54573774e-01
4.08663124e-01 -3.64021689e-01 -4.82883096e-01 -1.53492033e-01
3.36922407e-01 -6.55451179e-01 -8.50697041e-01 9.76513445e-01
5.14952719e-01 1.42057523e-01 2.99854070e-01 -1.07141244e+00
-3.86450082e-01 5.81021249e-01 -5.16149223e-01 5.20093322e-01
4.04658407e-01 5.93119264e-01 5.32452345e-01 -4.72953558e-01
5.08365095e-01 1.09476423e+00 8.29492688e-01 6.72842383e-01
-8.57648790e-01 -7.93141186e-01 1.89795718e-01 4.37155098e-01
-1.34508371e+00 -3.85747552e-01 5.63327432e-01 -6.70857072e-01
8.83057773e-01 2.41588891e-01 7.80327797e-01 1.52345252e+00
1.11807380e-02 6.76366627e-01 6.14005506e-01 -2.14467764e-01
-1.12154275e-01 4.38309163e-02 1.71606854e-01 8.32402885e-01
-2.11161226e-02 2.96680510e-01 -3.41025174e-01 1.89283043e-01
9.06478226e-01 3.52868676e-01 -1.24160416e-01 1.60039201e-01
-9.89911795e-01 9.25135314e-01 4.21843082e-01 4.94448990e-01
-2.94697613e-01 2.98264802e-01 5.19291043e-01 2.18444958e-01
2.45817870e-01 2.46975884e-01 -2.32765242e-01 -1.63485423e-01
-8.32862139e-01 1.01952672e-01 2.28977978e-01 9.85136807e-01
4.51573193e-01 2.94989258e-01 -6.07815683e-01 5.89444637e-01
-1.53806016e-01 8.78417313e-01 2.00330958e-01 -8.83910656e-01
5.08087337e-01 7.13271976e-01 -7.56521448e-02 -1.04798949e+00
-5.74194789e-01 -1.52628988e-01 -7.49500573e-01 1.03381812e-03
5.25964439e-01 -3.31407130e-01 -9.57686067e-01 1.64316869e+00
9.82778370e-02 1.01190388e-01 -1.64579630e-01 8.67362261e-01
8.48912418e-01 8.33646417e-01 -4.88180034e-02 -1.34228438e-01
1.60342121e+00 -8.24822009e-01 -5.24605155e-01 -1.02441251e-01
5.27104497e-01 -5.37374079e-01 7.66446888e-01 2.73411930e-01
-7.34559119e-01 -5.31693697e-01 -7.16287553e-01 1.48876727e-01
-3.39421332e-01 4.40949142e-01 3.39856893e-01 7.10881352e-01
-9.67827082e-01 3.04408967e-01 -1.08296168e+00 -3.54007006e-01
1.87879503e-01 2.66097128e-01 -2.59827316e-01 -4.00758535e-02
-1.29016829e+00 8.59890878e-01 3.99355888e-01 3.22949111e-01
-1.10128677e+00 -1.97034642e-01 -1.01082385e+00 -1.83291957e-01
8.02987695e-01 -2.24215508e-01 1.18181682e+00 -5.54237783e-01
-1.15855718e+00 6.05173230e-01 -1.09769002e-01 -6.97979748e-01
7.28198528e-01 -1.12502664e-01 -3.30204725e-01 3.64976108e-01
6.72547817e-02 4.99691397e-01 9.93651032e-01 -6.58456504e-01
-6.83030367e-01 -3.39375734e-01 3.26204419e-01 -2.06065178e-01
-4.55958933e-01 3.66173148e-01 -9.23050761e-01 -6.92027748e-01
-5.27206600e-01 -1.17745781e+00 -4.98009026e-02 -2.56655037e-01
-3.18973392e-01 -2.82403827e-01 9.63176310e-01 -1.20266879e+00
1.31611454e+00 -2.06698370e+00 9.91793200e-02 -1.15375914e-01
3.90397981e-02 6.34207249e-01 -8.10374394e-02 3.46481562e-01
1.30002826e-01 -2.22752199e-01 -1.19875878e-01 -1.39863625e-01
-3.91395718e-01 -5.97878657e-02 9.31868237e-03 3.03975701e-01
3.15223336e-01 6.91283226e-01 -9.09693241e-01 -4.71659034e-01
4.39209640e-01 6.65494561e-01 -5.34962118e-01 2.29040742e-01
-1.79550737e-01 4.01803046e-01 -3.87076586e-01 8.04787278e-01
3.50932389e-01 -2.21288502e-02 7.74609596e-02 -1.17109664e-01
-4.31581885e-01 -2.88387805e-01 -8.96511495e-01 1.42708933e+00
-9.53699648e-02 7.40468264e-01 -2.69572139e-02 -8.16993475e-01
7.11601853e-01 5.96693277e-01 6.91387832e-01 -7.63643682e-01
5.63403666e-01 -2.88974464e-01 -1.59953177e-01 -7.43153512e-01
5.79360306e-01 3.17056000e-01 -7.95477331e-02 8.51776749e-02
5.26565909e-02 3.02772909e-01 7.42677510e-01 2.21613556e-01
1.38364959e+00 -5.78842275e-02 -2.06206426e-01 1.31174877e-01
2.90116280e-01 6.13209195e-02 6.33608758e-01 8.51813734e-01
-1.92968890e-01 4.95785236e-01 5.84139407e-01 -4.76008415e-01
-1.01128733e+00 -7.52095222e-01 9.57387462e-02 1.08924782e+00
7.48266056e-02 -3.42948556e-01 -1.00875163e+00 -6.43646717e-01
-2.90015608e-01 4.38385814e-01 -8.23669255e-01 -3.52673501e-01
-8.20091367e-01 -7.77727067e-01 8.31398964e-01 8.14862788e-01
7.04317451e-01 -1.30145359e+00 -1.08225620e+00 1.31280512e-01
-4.63595390e-01 -1.17650509e+00 -5.43347299e-01 -4.48485576e-02
-4.20514494e-01 -1.39190114e+00 -8.56925726e-01 -4.30162579e-01
3.19055766e-01 3.87137234e-01 6.50698125e-01 -4.16833498e-02
-6.76561773e-01 3.13652426e-01 -6.59133136e-01 -4.19099480e-01
-2.72280812e-01 1.38859913e-01 1.45527020e-01 -2.03654706e-01
3.83534819e-01 -8.88845623e-02 -4.04701948e-01 4.30487216e-01
-8.81408870e-01 -1.69457197e-02 4.98325557e-01 7.28766024e-01
-7.15694670e-03 -3.56761403e-02 2.88685530e-01 -3.82984012e-01
1.55817270e-01 -3.02235603e-01 -8.52560699e-01 1.88665375e-01
-6.66937307e-02 -2.08864570e-01 5.55808127e-01 -6.58403397e-01
-8.26347053e-01 1.19950958e-01 -4.53267574e-01 -8.74472380e-01
1.41891241e-01 2.99619604e-02 1.62846908e-01 3.00753176e-01
6.56063080e-01 3.18951190e-01 2.83790454e-02 -4.19736385e-01
-5.87784871e-03 3.96252871e-01 6.40801430e-01 -1.12208188e-01
5.60355067e-01 3.53315622e-01 -4.68426377e-01 -1.03561819e+00
-6.73407555e-01 -4.75665510e-01 -4.82857138e-01 -4.02767628e-01
1.20211112e+00 -8.39072227e-01 -9.36315536e-01 7.33016431e-01
-1.02934039e+00 -3.86629164e-01 6.13576025e-02 5.54081321e-01
-2.20645651e-01 2.40745917e-01 -8.36716831e-01 -1.05125713e+00
-4.39820588e-01 -1.24677455e+00 1.02196693e+00 1.83982775e-01
1.11906692e-01 -5.07132590e-01 -1.20934784e-01 3.66542429e-01
2.83497125e-01 2.98150331e-01 3.28096926e-01 -3.55821729e-01
-5.13416886e-01 -4.66896296e-01 -3.28402311e-01 3.75212461e-01
1.01199239e-01 1.17692746e-01 -1.02066839e+00 -5.28597414e-01
-2.17528075e-01 -4.49547172e-02 1.17561579e+00 6.07403040e-01
9.34131444e-01 -2.98291206e-01 -4.41565663e-01 4.44069237e-01
1.02798927e+00 7.79709816e-01 6.45022273e-01 3.21805418e-01
6.67076409e-01 3.50629300e-01 6.29534066e-01 5.51252484e-01
3.28740090e-01 9.07396376e-01 5.52447975e-01 -1.32830203e-01
-1.31054834e-01 -4.94328439e-02 5.75419426e-01 4.43841398e-01
-4.99942988e-01 -2.00202331e-01 -9.66000021e-01 6.12918317e-01
-1.91117358e+00 -1.45956326e+00 1.61628529e-01 2.05854297e+00
5.01688838e-01 6.00337908e-02 6.57834113e-01 2.78784752e-01
9.43046868e-01 2.29044482e-01 -3.46562386e-01 4.71436046e-02
-1.77572407e-02 -3.11615169e-01 5.48895836e-01 1.72975540e-01
-1.47585428e+00 9.53868091e-01 5.48012018e+00 6.20786428e-01
-1.25219560e+00 1.45406991e-01 3.69196147e-01 -3.16203564e-01
5.43216944e-01 -2.95666456e-01 -9.54115987e-01 8.96254361e-01
8.29892278e-01 3.05768132e-01 3.29194367e-01 8.73457909e-01
2.99686462e-01 -1.28932059e-01 -8.34023893e-01 9.26229298e-01
2.11027980e-01 -1.24029636e+00 -2.19003290e-01 -9.01167318e-02
3.83687437e-01 -1.99260600e-02 1.82139892e-02 4.55408663e-01
1.66593120e-01 -7.86083281e-01 8.36280286e-01 5.46896458e-01
9.01361108e-01 -8.52684557e-01 1.04352748e+00 4.57654715e-01
-1.40520775e+00 -3.41267347e-01 -2.60887384e-01 6.14862740e-02
1.33954391e-01 2.39601627e-01 -7.70807922e-01 4.96373504e-01
1.21368790e+00 6.40251875e-01 -6.01503253e-01 7.71804452e-01
-1.68109268e-01 4.95831311e-01 -3.21565956e-01 -7.12503791e-02
2.85361409e-01 2.33810805e-02 5.92712522e-01 1.55029154e+00
3.20643842e-01 2.13380083e-01 2.75809109e-01 3.90486300e-01
6.82516620e-02 -3.81626874e-01 -8.74150634e-01 2.05539182e-01
4.60112035e-01 9.89350319e-01 -6.34437144e-01 -4.42057610e-01
-3.13565701e-01 7.50686109e-01 1.09601011e-02 6.97896183e-02
-1.37784672e+00 -4.47321594e-01 3.84778410e-01 1.68817058e-01
3.95785838e-01 -1.44166142e-01 2.39163786e-01 -1.28212416e+00
1.45718023e-01 -9.14952934e-01 6.64803386e-01 -4.62024063e-01
-7.03863621e-01 9.27907348e-01 2.18706191e-01 -1.22241044e+00
-4.55865681e-01 -5.72344005e-01 -4.74504352e-01 4.14562672e-01
-8.46835077e-01 -1.29042733e+00 -5.06259143e-01 6.95585310e-01
6.73766434e-01 -1.15193598e-01 4.75068271e-01 5.86258054e-01
-1.00692058e+00 6.84741974e-01 -1.72794461e-01 7.07064927e-01
4.84991193e-01 -7.44149864e-01 2.91344732e-01 1.16498029e+00
-2.54926056e-01 3.61240447e-01 6.81437135e-01 -5.79088509e-01
-1.48670220e+00 -1.32063663e+00 3.84162784e-01 -6.43711090e-01
4.84971493e-01 -2.50699878e-01 -7.77943432e-01 7.89851904e-01
8.98047984e-02 -2.00128153e-01 8.61906186e-02 -2.46884018e-01
-3.58974069e-01 -8.96825120e-02 -1.05697608e+00 4.20598686e-01
1.10388827e+00 -3.44243467e-01 -3.94999117e-01 2.91775584e-01
6.79653108e-01 -4.28298473e-01 -6.26502573e-01 4.97566640e-01
6.08255327e-01 -9.95184660e-01 9.53098774e-01 -1.46740869e-01
3.10954303e-01 -1.77636221e-01 1.42742068e-01 -9.03340936e-01
-2.67799467e-01 -2.68499494e-01 -1.19700797e-01 1.04460847e+00
1.93821773e-01 -5.40838897e-01 8.45427155e-01 2.51108885e-01
-5.42410575e-02 -4.73820448e-01 -8.64275038e-01 -1.06618059e+00
-3.04510117e-01 -4.82401848e-01 1.85083792e-01 7.72193849e-01
-1.20970197e-01 1.30963549e-01 -9.29896653e-01 2.11249620e-01
4.87695426e-01 1.82967838e-02 7.19742358e-01 -8.30821037e-01
-2.53823906e-01 -3.99191409e-01 -5.46510756e-01 -8.81608665e-01
-2.78868794e-01 -4.88109291e-01 4.05112430e-02 -1.29341078e+00
4.09945637e-01 -1.17775388e-01 -1.20509081e-01 8.81226599e-01
-2.24196464e-02 4.89911139e-01 4.16535884e-01 -3.98078822e-02
-5.54359317e-01 3.30493242e-01 7.49043047e-01 -3.85576785e-01
-3.25562567e-01 1.18897267e-01 -1.80780515e-01 7.77076483e-01
8.82912874e-01 -4.71507728e-01 -6.66397661e-02 -6.42880380e-01
-1.65285066e-01 4.61177379e-01 6.43699706e-01 -1.10859346e+00
3.33414406e-01 8.86152149e-04 3.66528958e-01 -7.85919666e-01
7.24291086e-01 -6.16422653e-01 1.16078265e-01 9.34504867e-01
6.08370043e-02 2.52938032e-01 3.28167558e-01 4.27070946e-01
-1.29763827e-01 -1.57289907e-01 9.12233710e-01 -1.79990634e-01
-8.52743506e-01 3.35031331e-01 -6.25378728e-01 -2.16920063e-01
1.20984542e+00 -2.60013461e-01 -4.01914150e-01 -1.96633667e-01
-6.22676909e-01 3.17524344e-01 2.75500566e-01 6.77987874e-01
5.49298346e-01 -1.03837264e+00 -6.84429109e-01 5.08863516e-02
6.88937977e-02 -2.20342249e-01 5.23619533e-01 9.40448880e-01
-4.25932854e-01 5.89893937e-01 -2.37888694e-01 -7.54088283e-01
-1.58451796e+00 7.95406401e-01 2.42526963e-01 -2.04650909e-01
-7.91272402e-01 9.29099917e-01 4.11000289e-02 -1.49272382e-01
3.63846600e-01 -3.41010302e-01 -4.03493911e-01 2.38834575e-01
7.72645175e-01 5.54643810e-01 -3.98472510e-02 -6.47716522e-01
-4.56060112e-01 5.60397029e-01 5.17936647e-02 8.17079395e-02
1.29668438e+00 3.77856851e-01 1.66647866e-01 1.67390153e-01
8.23116004e-01 -2.06129670e-01 -1.35194123e+00 -1.48644686e-01
-9.71889198e-02 -4.20459032e-01 -2.26250753e-01 -5.72872996e-01
-1.21523130e+00 7.25784361e-01 7.12766051e-01 1.71737924e-01
1.25522411e+00 -3.73198860e-03 6.86947584e-01 5.45402408e-01
4.55581725e-01 -8.75720620e-01 1.92395046e-01 6.25985503e-01
8.35439861e-01 -1.21531498e+00 -3.22478324e-01 -1.85933352e-01
-6.40429914e-01 8.83419275e-01 6.72693789e-01 -6.98580146e-02
2.12476864e-01 3.30128670e-01 -2.86820997e-02 -2.56415278e-01
-6.17485404e-01 -3.16523790e-01 1.15531124e-01 4.22644526e-01
1.67821527e-01 -5.98843694e-02 -1.39435753e-02 5.16183972e-01
-2.20992602e-02 -1.35751471e-01 3.05193096e-01 9.60322678e-01
-3.27494115e-01 -5.65753698e-01 -5.53878248e-01 4.18391705e-01
-5.59042752e-01 1.11225605e-01 -2.61974484e-01 8.91076386e-01
2.99055547e-01 1.10481906e+00 6.22997992e-03 -5.93390286e-01
5.54933250e-01 -1.57687008e-01 3.15675944e-01 -3.76870215e-01
-7.12334335e-01 -8.03304184e-03 2.70957574e-02 -5.28816104e-01
-3.12286317e-01 -6.28783405e-01 -8.93914521e-01 -4.56201375e-01
-3.91983747e-01 4.15515825e-02 4.33042645e-01 8.36685181e-01
-1.70555990e-02 8.10377419e-01 4.27760124e-01 -9.56264555e-01
-2.75436550e-01 -1.20306337e+00 -2.26716489e-01 1.57281026e-01
3.61673236e-01 -7.72730052e-01 -1.41403982e-02 2.09095865e-01]
|
[8.057464599609375, 0.6200469732284546]
|
29129db4-3a4c-4b18-8617-12076f40f7bb
|
a-normalized-gaussian-wasserstein-distance
|
2110.13389
| null |
https://arxiv.org/abs/2110.13389v2
|
https://arxiv.org/pdf/2110.13389v2.pdf
|
A Normalized Gaussian Wasserstein Distance for Tiny Object Detection
|
Detecting tiny objects is a very challenging problem since a tiny object only contains a few pixels in size. We demonstrate that state-of-the-art detectors do not produce satisfactory results on tiny objects due to the lack of appearance information. Our key observation is that Intersection over Union (IoU) based metrics such as IoU itself and its extensions are very sensitive to the location deviation of the tiny objects, and drastically deteriorate the detection performance when used in anchor-based detectors. To alleviate this, we propose a new evaluation metric using Wasserstein distance for tiny object detection. Specifically, we first model the bounding boxes as 2D Gaussian distributions and then propose a new metric dubbed Normalized Wasserstein Distance (NWD) to compute the similarity between them by their corresponding Gaussian distributions. The proposed NWD metric can be easily embedded into the assignment, non-maximum suppression, and loss function of any anchor-based detector to replace the commonly used IoU metric. We evaluate our metric on a new dataset for tiny object detection (AI-TOD) in which the average object size is much smaller than existing object detection datasets. Extensive experiments show that, when equipped with NWD metric, our approach yields performance that is 6.7 AP points higher than a standard fine-tuning baseline, and 6.0 AP points higher than state-of-the-art competitors. Codes are available at: https://github.com/jwwangchn/NWD.
|
['Lei Yu', 'Wen Yang', 'Chang Xu', 'Jinwang Wang']
|
2021-10-26
| null | null | null | null |
['small-object-detection']
|
['computer-vision']
|
[-2.31781334e-01 -1.61539257e-01 5.44746742e-02 -2.33741790e-01
-9.07542527e-01 -5.36816537e-01 3.45600128e-01 3.19144189e-01
-5.71676970e-01 2.10925445e-01 -2.73653537e-01 -1.88802525e-01
3.16238075e-01 -6.94263995e-01 -8.04076552e-01 -6.75207913e-01
8.02786276e-02 2.46136904e-01 9.40843761e-01 -2.19353363e-02
5.17387390e-02 4.01192665e-01 -1.33609164e+00 -1.51451141e-01
6.72472715e-01 1.43195128e+00 1.95008129e-01 4.76116538e-01
5.98200671e-02 1.24352194e-01 -6.71658754e-01 -3.61785889e-01
7.24754035e-01 1.32264579e-02 -2.22529806e-02 -1.60742868e-02
7.32058764e-01 -4.96274292e-01 -3.77855957e-01 1.28114033e+00
7.39725411e-01 -7.34061822e-02 7.28056550e-01 -1.18473172e+00
-7.31856227e-01 5.56231797e-01 -9.67988193e-01 5.36482871e-01
1.16623953e-01 1.02198087e-01 1.00925744e+00 -1.22903681e+00
3.03400695e-01 1.34035468e+00 7.82203853e-01 4.88922119e-01
-1.11107004e+00 -6.77831233e-01 1.85639873e-01 7.86931813e-02
-1.79596937e+00 -2.17904136e-01 3.70583802e-01 -2.65083075e-01
5.05785704e-01 2.85176486e-01 2.87562340e-01 7.23256528e-01
5.75381331e-02 9.19991136e-01 7.05807030e-01 -3.25511754e-01
2.16677576e-01 2.68183529e-01 7.47045428e-02 7.32525110e-01
7.46161282e-01 -2.25884661e-01 7.10324794e-02 -2.30140612e-01
6.06206119e-01 3.19328696e-01 -8.78788754e-02 -6.41174972e-01
-1.19477320e+00 8.32134187e-01 7.81991959e-01 1.64380148e-01
4.43553599e-03 2.77391165e-01 2.18588158e-01 6.41507953e-02
3.39076847e-01 1.00342140e-01 -1.87119916e-01 1.90919518e-01
-5.36581755e-01 2.19544590e-01 5.85043073e-01 1.00024152e+00
4.97591585e-01 -2.95130253e-01 -4.93024111e-01 6.46479547e-01
2.91220218e-01 7.44599640e-01 2.64475256e-01 -5.84610939e-01
5.47408342e-01 6.45419657e-01 3.31220657e-01 -1.02498126e+00
-3.97945225e-01 -4.57787305e-01 -5.16379952e-01 2.30804294e-01
6.60485923e-01 1.30090313e-02 -8.10043335e-01 1.54176843e+00
6.29174590e-01 -6.72803670e-02 -2.13541150e-01 9.73482609e-01
7.99191952e-01 4.24906522e-01 -1.42353296e-01 5.57756983e-02
1.41630924e+00 -7.80122638e-01 -2.67966568e-01 -2.99652189e-01
6.39283240e-01 -7.85916388e-01 1.05387414e+00 8.80423412e-02
-8.92660737e-01 -5.86804450e-01 -1.22768831e+00 1.00056767e-01
-3.63519073e-01 1.73213974e-01 3.74777377e-01 7.22299874e-01
-6.49859130e-01 2.74719179e-01 -7.31869757e-01 -5.17509401e-01
6.09463274e-01 2.06809327e-01 -9.82982218e-02 -8.10779780e-02
-7.77532637e-01 7.83463180e-01 4.63105261e-01 -1.88588008e-01
-7.41940677e-01 -5.79513013e-01 -7.20120370e-01 4.45529595e-02
7.46893287e-01 -4.48098958e-01 1.22571361e+00 -1.23245597e-01
-7.09277511e-01 8.38617682e-01 7.91567788e-02 -6.84460044e-01
7.90416658e-01 -1.50235549e-01 -1.77105755e-01 -1.98919084e-02
3.00187856e-01 5.61153114e-01 8.73278081e-01 -1.05407965e+00
-9.21220243e-01 -4.73033518e-01 1.08076178e-01 4.79019061e-03
-7.07517862e-01 8.64305794e-02 -6.98506892e-01 -6.46654427e-01
3.33113402e-01 -7.88719893e-01 -2.94926345e-01 6.20974243e-01
-5.72200358e-01 -5.80608666e-01 1.04310894e+00 5.89508414e-02
1.41382325e+00 -2.30571532e+00 -4.01280850e-01 6.09319359e-02
5.26168048e-01 3.35750908e-01 -1.44562945e-01 -2.49326434e-02
3.91225517e-01 7.77265206e-02 -1.03479691e-01 -3.41001689e-01
2.02962309e-01 -1.05056107e-01 -1.01878144e-01 7.42405057e-01
-7.54366815e-02 7.54549026e-01 -9.74002421e-01 -5.40734887e-01
2.70075262e-01 3.97730559e-01 -5.62173910e-02 -5.55121414e-02
2.22636536e-02 -1.95058540e-01 -7.13356316e-01 1.01917529e+00
9.48708057e-01 -1.66367650e-01 -3.80026758e-01 -3.53771359e-01
-1.38641238e-01 -1.26834184e-01 -1.48445833e+00 1.31172979e+00
-6.15836941e-02 4.11196172e-01 -1.14493772e-01 -7.85414279e-01
1.08464980e+00 -7.40141720e-02 3.53717238e-01 -4.90626425e-01
3.76090050e-01 4.20544147e-01 2.34223194e-02 -1.40484557e-01
2.21458763e-01 9.92014483e-02 -2.00625256e-01 3.48542750e-01
-2.85788208e-01 2.02798136e-02 5.30311167e-01 2.82259792e-01
1.34151804e+00 -4.10573959e-01 4.32698041e-01 -1.46343768e-01
2.31422812e-01 -3.28029931e-01 7.17382252e-01 1.10858667e+00
-5.38695693e-01 8.65054131e-01 2.99309224e-01 -4.09567177e-01
-9.23034251e-01 -1.34546745e+00 -5.29225469e-01 9.99364078e-01
6.06087446e-01 -4.88710195e-01 -7.54757881e-01 -9.01784599e-01
5.20070076e-01 2.23435163e-01 -7.65275717e-01 -1.93746731e-01
-4.20318812e-01 -9.98823702e-01 4.65871751e-01 8.19986165e-01
4.59037542e-01 -5.52654445e-01 -9.17362928e-01 1.73262611e-01
1.30468458e-01 -1.31614304e+00 -7.58653522e-01 1.55288443e-01
-6.86045647e-01 -1.04212308e+00 -9.11797822e-01 -5.39830744e-01
6.76872611e-01 7.17001140e-01 8.45424533e-01 -5.32606058e-02
-6.10421240e-01 1.21240959e-01 -4.08963650e-01 -7.61377096e-01
1.48935337e-02 -2.14478090e-01 3.51378918e-01 -3.44837122e-02
5.75835586e-01 -2.70828247e-01 -9.58223879e-01 6.63075805e-01
-7.50051677e-01 -4.67702627e-01 7.97172785e-01 5.09272873e-01
8.31699789e-01 -7.23785982e-02 2.41956919e-01 -5.98519027e-01
3.25379699e-01 -3.62408936e-01 -8.17227662e-01 1.58622652e-01
-4.51156676e-01 -2.13740692e-02 4.66516167e-01 -7.30641484e-01
-4.65296835e-01 9.27292928e-02 2.28635788e-01 -6.42191231e-01
7.96526521e-02 -9.77367982e-02 -1.88425183e-01 -3.36589724e-01
7.68757105e-01 -2.84804888e-02 -4.88816857e-01 -5.28977931e-01
2.46494427e-01 7.25274622e-01 4.61346984e-01 -2.88594395e-01
1.09183037e+00 8.66826296e-01 -1.17026091e-01 -5.37767529e-01
-1.02532351e+00 -9.05141532e-01 -6.06610775e-01 6.61794618e-02
4.98371214e-01 -8.38309765e-01 -5.78026712e-01 2.93934226e-01
-9.49872255e-01 5.97186666e-03 -3.99338424e-01 4.02170032e-01
-1.56813338e-01 3.75093907e-01 -3.83890390e-01 -8.01177800e-01
-4.65857089e-01 -1.00048566e+00 1.13545370e+00 3.48660618e-01
1.34146899e-01 -4.99961048e-01 -2.12071955e-01 1.74254015e-01
3.57623726e-01 4.70039189e-01 3.75681728e-01 -6.68974042e-01
-4.69157159e-01 -8.01399946e-01 -6.11277521e-01 2.62484401e-01
6.45282073e-03 -1.10973217e-01 -7.45905280e-01 -4.96070683e-01
-7.78414011e-02 -2.31209755e-01 1.22245526e+00 5.08523166e-01
1.07963502e+00 3.16589400e-02 -6.66927755e-01 5.85623920e-01
1.44659519e+00 -4.43027057e-02 3.33444446e-01 5.79864383e-01
6.48953140e-01 1.79220945e-01 8.91336143e-01 7.51256049e-01
2.19359681e-01 7.83488810e-01 6.30360842e-01 6.39642328e-02
-1.69866815e-01 -7.65751451e-02 3.20280045e-01 1.93682060e-01
5.73387928e-02 -3.66967708e-01 -8.40582848e-01 5.80124676e-01
-1.94967210e+00 -7.98228502e-01 -1.93369851e-01 2.39294791e+00
4.44665164e-01 6.70340180e-01 4.08432394e-01 8.03254470e-02
8.20010126e-01 1.09014802e-01 -7.37231314e-01 -5.03950426e-03
-1.34623483e-01 -2.98044682e-01 8.41972888e-01 -4.55355495e-02
-1.48130059e+00 7.40617573e-01 5.55390310e+00 1.06587791e+00
-8.00602138e-01 2.86222011e-01 4.27977443e-01 -3.40004861e-01
2.51376808e-01 -2.82252878e-01 -1.32192409e+00 5.18939018e-01
4.58284110e-01 -2.10714877e-01 -2.78838068e-01 1.20286322e+00
2.09138483e-01 -1.14468209e-01 -1.14429200e+00 1.05261612e+00
5.59820645e-02 -8.92671704e-01 3.46916728e-02 -1.37745566e-03
6.64155185e-01 1.79011598e-01 3.15931886e-01 2.19742134e-01
1.18501782e-01 -6.51440024e-01 9.22542274e-01 -4.42965142e-02
6.26224101e-01 -5.92896402e-01 8.20501745e-01 2.50560135e-01
-1.59993637e+00 -2.79876858e-01 -1.06814837e+00 2.44455710e-01
4.82704453e-02 7.79235125e-01 -7.51911461e-01 -4.74609761e-03
1.06011057e+00 4.73106265e-01 -7.78149962e-01 1.58615911e+00
-9.17016715e-02 3.66658419e-01 -7.81042516e-01 -3.02052408e-01
3.32875520e-01 1.27439320e-01 7.94835031e-01 1.22458816e+00
4.87777054e-01 5.05125336e-02 3.26269656e-01 9.41168666e-01
-3.58492702e-01 1.38362557e-01 -3.95230591e-01 4.05957401e-01
7.51918554e-01 1.44854164e+00 -1.00950134e+00 -3.19384515e-01
-5.60819507e-01 8.09885263e-01 2.77009279e-01 5.17068096e-02
-1.07390416e+00 -5.57933688e-01 6.48969293e-01 2.19946086e-01
8.14774930e-01 9.93723273e-02 -1.79170787e-01 -9.80747402e-01
5.64194679e-01 -4.60033774e-01 4.78781432e-01 -4.08634543e-01
-1.31371701e+00 4.39729899e-01 -1.85021728e-01 -1.48868263e+00
4.01269525e-01 -6.83344662e-01 -8.20273399e-01 3.01706493e-01
-1.27843297e+00 -9.98415530e-01 -4.95868325e-01 4.76734817e-01
4.94347066e-01 1.00463241e-01 3.77628058e-01 3.77720624e-01
-6.58853710e-01 9.85884428e-01 4.69905674e-01 3.46109211e-01
8.14670980e-01 -1.34587884e+00 5.34441173e-01 9.33444738e-01
2.10002512e-01 3.30724150e-01 8.52766752e-01 -3.39848489e-01
-1.26169908e+00 -1.27910495e+00 3.67339879e-01 -7.04317987e-01
8.33036482e-01 -5.49153984e-01 -7.61322975e-01 4.14722711e-01
-5.58796465e-01 8.63207281e-01 3.21868837e-01 -2.46247411e-01
-5.95021129e-01 -3.22207570e-01 -1.25615001e+00 4.08140302e-01
1.07384658e+00 -9.48631614e-02 -3.60111028e-01 4.47496504e-01
7.61898339e-01 -2.74756849e-01 -6.17835462e-01 6.58234417e-01
6.04806721e-01 -1.06767273e+00 1.17039621e+00 -9.68241990e-02
-2.50969887e-01 -5.53944230e-01 -3.53647739e-01 -8.62530231e-01
-3.31788659e-01 -4.86964077e-01 -3.10599446e-01 1.09905493e+00
4.01884407e-01 -5.82880378e-01 6.96047843e-01 3.25273782e-01
-5.71202822e-02 -9.34137940e-01 -9.79462504e-01 -1.27746022e+00
-1.14912912e-01 -4.74406540e-01 4.43771571e-01 4.02067244e-01
-1.69023558e-01 1.11110136e-01 -4.62977588e-02 3.29664528e-01
1.05817854e+00 2.12768838e-01 7.21423090e-01 -1.39992678e+00
-1.73422590e-01 -5.59551716e-01 -8.55092168e-01 -1.23164523e+00
-5.07538557e-01 -5.42560697e-01 2.35097274e-01 -1.37248659e+00
5.70906103e-01 -6.45782351e-01 -7.45784461e-01 4.16741312e-01
-3.05960655e-01 7.28042722e-01 3.20525438e-01 2.56739169e-01
-1.08521247e+00 5.42573988e-01 1.02569592e+00 -2.83312976e-01
-7.28704110e-02 2.11352527e-01 -6.80109978e-01 9.67758000e-01
7.32750118e-01 -7.14071095e-01 8.95392373e-02 -3.55458796e-01
-7.04837590e-02 -5.29362142e-01 3.49487245e-01 -1.32761419e+00
3.18582624e-01 2.00753257e-01 3.54087740e-01 -6.50252402e-01
2.00077072e-01 -6.92474365e-01 -5.42522967e-01 5.69539487e-01
4.00624499e-02 -2.49997333e-01 1.41821742e-01 7.98033535e-01
2.61293314e-02 -1.77299395e-01 1.01597631e+00 3.18934992e-02
-8.43509674e-01 5.03461421e-01 4.54315096e-02 9.84847471e-02
1.45009506e+00 -4.11037385e-01 -3.87073517e-01 -6.70459345e-02
-4.43411827e-01 3.49507540e-01 4.35331851e-01 3.52126509e-01
7.34199107e-01 -1.44482327e+00 -8.36489618e-01 -1.71627596e-01
4.74413335e-01 9.86794662e-03 -1.69477463e-01 1.10363317e+00
-2.96006471e-01 3.54228586e-01 1.24629743e-01 -6.72357500e-01
-1.54063213e+00 5.95568419e-01 2.41337374e-01 -3.23994644e-02
-6.59730434e-01 1.19668686e+00 5.32600939e-01 -1.76056802e-01
5.65072596e-01 -5.56214333e-01 1.09872706e-01 -6.20184988e-02
8.47708344e-01 6.11332834e-01 -2.09236126e-02 -5.20033598e-01
-5.36520660e-01 7.36275315e-01 -2.53751516e-01 4.33456630e-01
1.10851634e+00 -1.08589739e-01 3.31860453e-01 2.69952446e-01
1.09944868e+00 -1.17403395e-01 -1.38840806e+00 -4.78331327e-01
-1.40986554e-02 -6.13579512e-01 -2.84092706e-02 -2.45893642e-01
-9.74183798e-01 8.00400019e-01 1.01127112e+00 3.20120722e-01
7.96754360e-01 5.59250355e-01 1.01041245e+00 5.96403301e-01
4.43313032e-01 -1.00915051e+00 1.08066879e-01 2.64669925e-01
6.79367006e-01 -1.62933743e+00 1.36210471e-01 -4.71872866e-01
-1.77404925e-01 9.82497513e-01 7.81612992e-01 -1.94583595e-01
5.78045189e-01 5.18650055e-01 5.02877310e-02 -2.59778276e-02
-3.84986490e-01 -5.51733375e-01 3.54825526e-01 4.67239678e-01
1.02282070e-01 3.21152285e-02 -3.05299729e-01 5.37646592e-01
1.14161968e-01 -3.33007008e-01 3.63869160e-01 8.73403013e-01
-1.02608752e+00 -6.11234903e-01 -6.89928472e-01 5.74629068e-01
-5.48236430e-01 1.24087650e-02 -2.53023297e-01 8.04856181e-01
2.98297703e-01 8.78789186e-01 1.20579995e-01 -4.36690301e-01
5.96416175e-01 -5.68775654e-01 3.39333862e-01 -7.00588107e-01
-4.42000963e-02 1.16405323e-01 -4.17849571e-01 -7.02298105e-01
-5.43190427e-02 -5.43969572e-01 -1.32070315e+00 -2.21809357e-01
-7.63723433e-01 -4.35752235e-02 3.74540806e-01 6.21456683e-01
9.23853666e-02 1.48398280e-01 5.86683214e-01 -9.18981612e-01
-1.03027904e+00 -1.03263986e+00 -7.16284037e-01 4.48543251e-01
2.61324912e-01 -8.14437211e-01 -5.39473593e-01 -3.79399627e-01]
|
[8.688347816467285, -0.44020959734916687]
|
477c287f-f9fe-4053-9f65-5744d46ca2ef
|
cross-task-attention-mechanism-for-dense
|
2206.08927
| null |
https://arxiv.org/abs/2206.08927v1
|
https://arxiv.org/pdf/2206.08927v1.pdf
|
Cross-task Attention Mechanism for Dense Multi-task Learning
|
Multi-task learning has recently become a promising solution for a comprehensive understanding of complex scenes. Not only being memory-efficient, multi-task models with an appropriate design can favor exchange of complementary signals across tasks. In this work, we jointly address 2D semantic segmentation, and two geometry-related tasks, namely dense depth, surface normal estimation as well as edge estimation showing their benefit on indoor and outdoor datasets. We propose a novel multi-task learning architecture that exploits pair-wise cross-task exchange through correlation-guided attention and self-attention to enhance the average representation learning for all tasks. We conduct extensive experiments considering three multi-task setups, showing the benefit of our proposal in comparison to competitive baselines in both synthetic and real benchmarks. We also extend our method to the novel multi-task unsupervised domain adaptation setting. Our code is available at https://github.com/cv-rits/DenseMTL.
|
['Raoul de Charette', 'Tuan-Hung Vu', 'Ivan Lopes']
|
2022-06-17
| null | null | null | null |
['2d-semantic-segmentation']
|
['computer-vision']
|
[ 1.85046792e-01 -2.31315896e-01 1.02171630e-01 -6.30676448e-01
-1.22274649e+00 -3.40712219e-01 6.41809762e-01 1.35805467e-02
-5.69277704e-01 5.86672068e-01 2.55728096e-01 1.01085581e-01
-6.82400391e-02 -5.15011847e-01 -9.32360470e-01 -5.08377254e-01
-3.49028893e-02 4.33484763e-01 2.98478603e-01 4.12053987e-02
1.99276522e-01 2.22936049e-01 -1.32327354e+00 4.94189590e-01
1.06670964e+00 1.06600153e+00 5.70619166e-01 5.64064741e-01
6.26892969e-02 5.72475970e-01 -2.31505617e-01 -3.38189363e-01
3.23643863e-01 9.10076126e-02 -9.64119375e-01 9.31591764e-02
7.08801210e-01 -7.61867017e-02 -8.58273506e-02 8.53318453e-01
8.43166113e-01 4.29851770e-01 6.17013395e-01 -9.70064640e-01
-4.36066598e-01 3.73537950e-02 -1.05633545e+00 1.92396775e-01
1.02259576e-01 -1.48909390e-02 1.10602438e+00 -1.13181102e+00
5.28619945e-01 1.19370961e+00 6.51005745e-01 2.37821177e-01
-1.31430840e+00 -5.52537203e-01 5.80875576e-01 3.55488777e-01
-1.41997814e+00 -4.26017612e-01 1.02352226e+00 -3.58572751e-01
1.01228654e+00 -1.93960145e-01 3.51252645e-01 1.39704943e+00
1.49932638e-01 1.22346282e+00 1.28601122e+00 -1.86220527e-01
9.75812599e-02 -1.27252981e-01 -9.86987725e-02 4.69826072e-01
1.41754031e-01 -1.47238150e-01 -6.13774717e-01 8.93543065e-02
7.87555456e-01 -2.37163231e-01 -1.40539512e-01 -5.87889016e-01
-1.21808863e+00 6.30730212e-01 5.26429474e-01 2.83315152e-01
-5.31497300e-01 3.41813207e-01 5.06435812e-01 8.25431049e-02
9.95776653e-01 2.99362123e-01 -7.92310417e-01 6.25973418e-02
-8.82030308e-01 2.47589946e-01 3.42658401e-01 9.10893857e-01
8.06031525e-01 -7.35095441e-02 -3.98956895e-01 1.18347013e+00
3.31701130e-01 5.25555074e-01 3.04298937e-01 -9.73830223e-01
5.51602662e-01 1.77470684e-01 4.13911194e-02 -9.45105731e-01
-5.93994558e-01 -6.94836617e-01 -8.08024347e-01 1.16270157e-02
3.73317450e-01 -2.49327585e-01 -8.77052963e-01 1.85131168e+00
3.71959537e-01 5.52434921e-01 -2.35786125e-01 8.93012583e-01
8.00753355e-01 3.96645099e-01 1.44015774e-01 3.57891560e-01
1.33071172e+00 -1.39207172e+00 -3.84934634e-01 -8.58343303e-01
4.84729856e-01 -8.44164670e-01 1.00529826e+00 4.59112734e-01
-9.70050812e-01 -7.77156830e-01 -8.23317230e-01 -3.82192343e-01
-2.21012458e-01 2.78723449e-01 8.45157146e-01 2.43360475e-01
-1.00746274e+00 4.61875588e-01 -9.84069049e-01 -2.70126194e-01
8.31440628e-01 2.28123143e-01 -2.80282289e-01 -2.78670818e-01
-8.33725393e-01 5.73691607e-01 2.16206253e-01 7.32155964e-02
-9.45417583e-01 -7.17243493e-01 -9.56294119e-01 -1.98300883e-01
4.03380305e-01 -9.68930423e-01 1.10277057e+00 -7.37622082e-01
-1.26744175e+00 1.15549505e+00 -3.51291984e-01 -2.62607485e-01
5.29169261e-01 -7.48733521e-01 -3.74024622e-02 5.08598052e-02
2.96320677e-01 8.26712132e-01 8.40803921e-01 -1.31074393e+00
-5.21690845e-01 -5.26476860e-01 -1.04436673e-01 4.22261506e-01
-1.55518040e-01 -2.07964733e-01 -7.76935756e-01 -9.04674828e-01
1.10040106e-01 -9.83813703e-01 -4.86300379e-01 4.64502759e-02
-4.20146734e-01 -5.80334663e-02 5.68593919e-01 -5.75550795e-01
6.82745874e-01 -2.18228412e+00 4.42909867e-01 -1.46049753e-01
1.26464203e-01 3.75671051e-02 -3.94365460e-01 1.03380479e-01
2.65918765e-02 -1.08611666e-01 -5.07427931e-01 -1.07336891e+00
-8.38316604e-02 3.35802287e-02 1.16392016e-01 5.05886257e-01
4.05029744e-01 1.14656115e+00 -8.05178940e-01 -4.79859114e-01
3.75439495e-01 5.86981297e-01 -5.10518134e-01 2.30568126e-01
-2.92059988e-01 9.19472992e-01 -7.02528954e-01 6.16873145e-01
9.31151211e-01 -5.28588235e-01 -1.00932263e-01 -3.71989369e-01
8.78889337e-02 2.68248409e-01 -1.19967139e+00 2.53637052e+00
-9.20099556e-01 6.30886734e-01 3.39780480e-01 -1.10801208e+00
8.01911592e-01 1.41218588e-01 4.63784426e-01 -9.73049104e-01
2.19227877e-02 2.17786327e-01 -4.24437761e-01 -3.99146885e-01
4.75233823e-01 1.02697514e-01 -6.72525093e-02 3.35912585e-01
2.15439186e-01 -3.60658526e-01 1.59201026e-03 7.86967054e-02
8.56365979e-01 4.96622056e-01 1.95455506e-01 -4.42648172e-01
4.26361561e-01 -3.69825751e-01 5.13279021e-01 6.20210230e-01
-9.08535346e-02 7.59218395e-01 4.23123799e-02 -1.31762043e-01
-8.48499477e-01 -9.32376683e-01 -2.33958647e-01 1.24458921e+00
3.01269382e-01 -7.96456411e-02 -2.57841796e-01 -6.23075545e-01
5.47171533e-02 5.61977446e-01 -6.27330840e-01 7.77552575e-02
-5.64112902e-01 -9.46169853e-01 2.83467144e-01 6.19628489e-01
7.80068398e-01 -7.79585361e-01 -5.77170908e-01 1.80355266e-01
-4.09013659e-01 -1.69503152e+00 -3.85911316e-01 3.67000133e-01
-8.72323036e-01 -9.14802074e-01 -1.09237218e+00 -7.88746536e-01
3.69245917e-01 4.35755640e-01 1.36308002e+00 -2.27523446e-01
-1.87013343e-01 6.12933338e-01 -2.68713862e-01 -3.34653974e-01
2.49854207e-01 4.27125454e-01 -3.84689718e-01 2.46764645e-01
7.31519312e-02 -8.47955644e-01 -8.05082321e-01 3.35106939e-01
-6.64810419e-01 2.69600898e-01 7.08161175e-01 7.33460188e-01
8.64155352e-01 -4.74953800e-01 5.38250506e-01 -9.15478170e-01
2.71069020e-01 -5.49913466e-01 -4.12665606e-01 1.46835044e-01
-2.94856876e-01 1.34643450e-01 3.22791845e-01 -3.98044735e-02
-1.33721411e+00 5.72300367e-02 -2.46245265e-01 -4.36667055e-01
-3.36903960e-01 2.54582375e-01 -2.73113638e-01 -2.38144353e-01
4.66279596e-01 9.19334888e-02 -3.69872540e-01 -6.28520072e-01
3.10726047e-01 2.18016073e-01 3.50111485e-01 -9.20087337e-01
5.91028929e-01 7.77370870e-01 1.21496201e-01 -8.94290924e-01
-1.15316236e+00 -7.37440526e-01 -8.17268074e-01 -7.01479614e-02
9.74224508e-01 -1.44620180e+00 -2.93933213e-01 8.03475678e-01
-1.29991007e+00 -7.99166620e-01 -3.61541994e-02 5.40758550e-01
-6.24436200e-01 3.45885158e-01 -4.28639621e-01 -6.43858373e-01
-2.16868892e-01 -1.17822158e+00 1.69845963e+00 5.92627637e-02
3.74091044e-02 -1.38321173e+00 1.40992343e-01 5.51292300e-01
4.04980898e-01 3.62916559e-01 4.68863338e-01 -5.36051571e-01
-7.67843485e-01 2.93233037e-01 -5.47146440e-01 3.23105097e-01
1.49936303e-02 -4.17727113e-01 -1.36743927e+00 -3.41742128e-01
-1.37401432e-01 -7.14557469e-01 1.41316640e+00 7.84997284e-01
1.43878806e+00 3.63878965e-01 -4.80774313e-01 1.01359475e+00
1.52141178e+00 -1.63402453e-01 5.26594698e-01 3.62628132e-01
1.01630771e+00 7.78239727e-01 7.51995385e-01 4.09395069e-01
7.66808212e-01 8.68514478e-01 4.03609902e-01 -4.67834383e-01
-2.81056374e-01 8.49854723e-02 -1.59535371e-02 4.74879354e-01
-1.36513516e-01 -3.83988529e-01 -9.72188771e-01 7.86609411e-01
-1.89398575e+00 -6.74906909e-01 -2.60841489e-01 1.98217618e+00
5.18891513e-01 1.60702392e-01 -2.05780528e-02 -2.04048887e-01
4.86790508e-01 5.47336757e-01 -6.27795041e-01 -1.14723109e-01
-3.72945935e-01 4.49687481e-01 5.12947738e-01 4.65509385e-01
-1.59279060e+00 1.05030906e+00 5.23292971e+00 1.05466473e+00
-1.02046490e+00 4.77943152e-01 9.24839795e-01 -2.02572107e-01
-3.84437442e-01 -3.05859208e-01 -6.55068338e-01 1.75981537e-01
5.20121753e-01 1.00260489e-01 7.02194795e-02 6.68340266e-01
8.29806328e-02 -3.63553315e-01 -1.01918054e+00 1.12871110e+00
5.85773997e-02 -1.20395148e+00 -2.71339923e-01 -7.26161674e-02
1.16130126e+00 5.82696915e-01 1.55318454e-01 1.29865110e-01
1.94543004e-01 -8.37240994e-01 7.60970116e-01 2.99289346e-01
6.11064851e-01 -4.45114136e-01 5.36816895e-01 -4.93553281e-03
-1.53559017e+00 6.98951408e-02 -9.52387825e-02 7.02261552e-02
3.45553786e-01 8.90127957e-01 -2.82691896e-01 1.01410139e+00
8.27800930e-01 1.23892021e+00 -6.47294044e-01 1.11061621e+00
-2.14158490e-01 2.74990767e-01 -2.95418769e-01 3.07793975e-01
2.50380814e-01 -1.89449072e-01 5.82971156e-01 1.46880066e+00
2.75814682e-01 -2.31741428e-01 3.57953310e-01 7.29399502e-01
-8.46472010e-02 1.88938260e-01 -5.43349087e-01 3.96431714e-01
2.24890947e-01 1.25507843e+00 -8.29784930e-01 -2.38742694e-01
-6.48377717e-01 1.33881474e+00 4.82940525e-01 5.04660249e-01
-8.53779018e-01 -1.22267842e-01 8.39314520e-01 -7.44734928e-02
5.12755692e-01 -5.46323061e-01 -5.84627867e-01 -1.17951047e+00
1.80806458e-01 -4.94343132e-01 3.87554973e-01 -6.12638772e-01
-1.31248963e+00 5.22471845e-01 4.88727205e-02 -9.85665143e-01
3.28049101e-02 -5.25858700e-01 -6.24995291e-01 7.83859909e-01
-2.06991863e+00 -1.43774617e+00 -5.67488194e-01 6.70400202e-01
8.35022628e-01 8.47028568e-02 5.32140791e-01 7.64410019e-01
-5.05428076e-01 4.58374739e-01 2.35173423e-02 -9.60046947e-02
9.51147735e-01 -1.25568724e+00 7.79697001e-01 8.96042347e-01
2.92363077e-01 2.56770756e-02 3.98679584e-01 -5.10212243e-01
-9.94704902e-01 -1.27265823e+00 4.82992500e-01 -4.02882218e-01
4.54887003e-01 -6.26149654e-01 -1.03087485e+00 7.18050480e-01
1.36909246e-01 1.26021162e-01 6.01723135e-01 2.63962030e-01
-3.80605847e-01 -1.13083674e-02 -7.50822127e-01 3.33141327e-01
1.40546846e+00 -5.24895906e-01 -9.11086425e-02 4.96213824e-01
7.19417095e-01 -5.61407804e-01 -7.60713398e-01 5.96346438e-01
3.60412478e-01 -1.10557437e+00 1.31819081e+00 -1.39871612e-01
4.64521974e-01 -1.34252295e-01 -3.33356947e-01 -1.40553653e+00
-3.63831669e-01 -3.99655968e-01 2.44852006e-02 1.01284027e+00
4.54919726e-01 -5.90738595e-01 7.21666336e-01 2.08181024e-01
-4.75506932e-01 -7.60003388e-01 -9.32395160e-01 -7.96854734e-01
1.76965430e-01 -7.72082269e-01 2.73095280e-01 8.12647521e-01
-6.99781597e-01 4.98900294e-01 -5.29377699e-01 2.33303308e-01
8.61600816e-01 2.59260505e-01 8.20621669e-01 -1.17805445e+00
-5.56477487e-01 -4.78548855e-01 -1.82869241e-01 -1.46167600e+00
2.49863788e-01 -9.55747843e-01 2.64898296e-02 -1.74130547e+00
1.29726112e-01 -4.47254330e-01 -2.87994653e-01 3.65683556e-01
-2.66648680e-01 3.20979267e-01 2.11609993e-02 6.81498125e-02
-9.36768234e-01 9.36340988e-01 1.41888142e+00 -9.24349949e-02
-2.12044761e-01 3.14952363e-03 -6.85634851e-01 6.85651958e-01
7.74117529e-01 -3.76918912e-01 -3.80106211e-01 -1.01764202e+00
5.94164021e-02 -3.10730368e-01 6.52480304e-01 -1.16487503e+00
1.41225055e-01 2.71249507e-02 3.73954147e-01 -4.29603338e-01
6.94424152e-01 -6.39534473e-01 -1.18048407e-01 8.46243501e-02
-2.59188324e-01 -2.93057770e-01 5.64987421e-01 6.81305110e-01
-2.52757102e-01 1.90090194e-01 9.20526266e-01 -4.14600372e-02
-1.14627683e+00 6.08387828e-01 1.42087534e-01 3.97935629e-01
1.03925169e+00 -8.69359151e-02 -9.82631966e-02 -2.38787413e-01
-7.21119463e-01 5.54090321e-01 3.23802799e-01 4.59364086e-01
6.39627039e-01 -1.14332783e+00 -8.76335919e-01 1.06790699e-01
3.15014482e-01 4.40107316e-01 6.98691368e-01 9.86725330e-01
-2.00480834e-01 4.52835292e-01 -1.99485794e-01 -7.92110145e-01
-1.03892076e+00 1.31010890e-01 3.62636358e-01 -4.62240487e-01
-5.62351048e-01 1.31100011e+00 6.69859111e-01 -4.52702761e-01
1.20178901e-01 -2.75421441e-01 -5.29440306e-03 3.45199294e-02
2.11685061e-01 2.63372511e-01 1.57632753e-01 -5.25358915e-01
-5.72055697e-01 9.76903260e-01 -7.98082948e-02 -4.52056713e-02
1.46767771e+00 -3.39004964e-01 1.98260501e-01 5.15805960e-01
1.36609495e+00 -1.43295184e-01 -1.66675651e+00 -4.56305295e-01
4.67222221e-02 -5.67124307e-01 3.46782178e-01 -5.88402987e-01
-1.20468247e+00 1.10120058e+00 5.02936304e-01 -3.68855238e-01
1.31851876e+00 1.84831440e-01 8.27715397e-01 1.58971623e-01
4.67036486e-01 -1.05155909e+00 5.29269159e-01 4.99668002e-01
9.28984821e-01 -1.77140462e+00 -7.07883462e-02 -6.07327878e-01
-8.02491188e-01 6.27761662e-01 7.98625648e-01 -2.61946559e-01
8.50205123e-01 1.39186099e-01 -1.60275996e-01 -2.78689235e-01
-7.54359066e-01 -5.92669070e-01 4.82414156e-01 7.53553569e-01
6.57664776e-01 -1.22857042e-01 1.01362489e-01 4.75737125e-01
3.03773791e-01 -2.06124652e-02 8.54945779e-02 7.54149973e-01
-8.12972188e-02 -1.12002051e+00 -8.06206688e-02 2.36550853e-01
-4.20566708e-01 -1.51068196e-01 -1.13078982e-01 5.75825274e-01
5.70428669e-02 8.06786597e-01 2.65626580e-01 -2.14374900e-01
3.13292354e-01 -8.09069127e-02 5.98579764e-01 -7.78737962e-01
-3.97814512e-01 2.11233184e-01 1.86372563e-01 -8.42130005e-01
-7.04632938e-01 -8.09430063e-01 -1.04452145e+00 9.34476852e-02
9.14132502e-03 -2.13957176e-01 6.39141917e-01 8.56239617e-01
6.56792581e-01 8.50149095e-01 5.14528453e-01 -1.12085044e+00
-6.84795752e-02 -9.63905752e-01 -4.15859610e-01 4.06772405e-01
2.30207279e-01 -9.37578380e-01 -1.11942589e-01 -1.38702914e-01]
|
[9.626075744628906, 1.2132996320724487]
|
6d59123a-2d3e-4e8e-b4eb-790844f53f3b
|
bigcolor-colorization-using-a-generative
|
2207.09685
| null |
https://arxiv.org/abs/2207.09685v1
|
https://arxiv.org/pdf/2207.09685v1.pdf
|
BigColor: Colorization using a Generative Color Prior for Natural Images
|
For realistic and vivid colorization, generative priors have recently been exploited. However, such generative priors often fail for in-the-wild complex images due to their limited representation space. In this paper, we propose BigColor, a novel colorization approach that provides vivid colorization for diverse in-the-wild images with complex structures. While previous generative priors are trained to synthesize both image structures and colors, we learn a generative color prior to focus on color synthesis given the spatial structure of an image. In this way, we reduce the burden of synthesizing image structures from the generative prior and expand its representation space to cover diverse images. To this end, we propose a BigGAN-inspired encoder-generator network that uses a spatial feature map instead of a spatially-flattened BigGAN latent code, resulting in an enlarged representation space. Our method enables robust colorization for diverse inputs in a single forward pass, supports arbitrary input resolutions, and provides multi-modal colorization results. We demonstrate that BigColor significantly outperforms existing methods especially on in-the-wild images with complex structures.
|
['Sunghyun Cho', 'Seung-Hwan Baek', 'Jonghyun Kim', 'Sehoon Kim', 'Hwayoon Lee', 'Seongtae Kim', 'Kyoungkook Kang', 'Geonung Kim']
|
2022-07-20
| null | null | null | null |
['colorization']
|
['computer-vision']
|
[ 5.27521372e-01 9.30314660e-02 2.45098367e-01 -1.92121446e-01
-6.04630530e-01 -7.74853885e-01 5.94000518e-01 -6.35367990e-01
-1.85275927e-01 6.76773071e-01 1.67407185e-01 -1.22513920e-01
4.49108034e-01 -9.45689678e-01 -8.55255723e-01 -7.92102993e-01
5.17589450e-01 -8.86789057e-03 -8.11651051e-02 -2.08145767e-01
8.56344327e-02 3.52881372e-01 -1.40508890e+00 1.85058683e-01
9.14068878e-01 6.12014711e-01 3.69459063e-01 7.09100604e-01
-7.41561577e-02 5.73541582e-01 -5.62201321e-01 -4.53427285e-01
4.25795466e-01 -8.02152038e-01 -4.10526037e-01 3.14926118e-01
5.62068939e-01 -4.88908172e-01 -3.29684049e-01 1.11208844e+00
2.78133124e-01 -5.32284714e-02 5.94519198e-01 -1.24165225e+00
-1.13790178e+00 4.16182846e-01 -8.57160211e-01 -3.71846884e-01
2.85682380e-01 3.92268181e-01 8.64482343e-01 -8.19734454e-01
6.95201337e-01 1.34726071e+00 2.72279382e-01 7.77672112e-01
-1.55028498e+00 -8.28164637e-01 2.77708799e-01 -1.97315961e-01
-1.30050862e+00 -2.60559648e-01 1.02129400e+00 -3.44131887e-01
3.08326989e-01 2.72690356e-01 8.14440131e-01 1.21148169e+00
-6.27494976e-02 6.54651582e-01 1.24172997e+00 -3.95106584e-01
2.53336698e-01 -1.77621901e-01 -7.08049059e-01 7.63824046e-01
2.76901901e-01 -1.44150741e-02 -5.13041139e-01 2.02976584e-01
1.23792279e+00 2.95662314e-01 -5.03846049e-01 -3.53535891e-01
-1.20465791e+00 7.16463506e-01 6.55922890e-01 5.12259230e-02
-3.20155978e-01 5.90847611e-01 -2.64722914e-01 -3.99966585e-03
2.99960494e-01 5.64154208e-01 7.36768320e-02 -3.61406431e-02
-9.57481861e-01 -3.30596715e-02 5.05483031e-01 9.39542890e-01
1.11509526e+00 3.47423494e-01 -3.14083189e-01 6.99946702e-01
1.70340553e-01 8.25842857e-01 8.38213041e-02 -1.08372295e+00
3.02749485e-01 5.48856914e-01 9.74329412e-02 -9.42057967e-01
2.81072035e-02 -3.07207555e-01 -1.13952506e+00 5.15569508e-01
2.71077454e-01 -2.10977823e-01 -1.30998981e+00 2.03999758e+00
2.88910896e-01 1.28016146e-02 7.78980255e-02 9.98742700e-01
4.81472045e-01 9.55048144e-01 -5.06007671e-02 2.64802277e-01
1.32989097e+00 -1.09804392e+00 -6.24483824e-01 -4.18407023e-01
2.72623431e-02 -8.11613083e-01 1.40099800e+00 3.71674687e-01
-1.07496405e+00 -4.57539558e-01 -1.06955338e+00 -2.69826770e-01
-1.79843605e-01 2.12950423e-01 5.93638718e-01 7.82091558e-01
-1.10456038e+00 1.88980281e-01 -6.38503551e-01 -1.34795219e-01
3.57714415e-01 -6.77210540e-02 -3.53272974e-01 -4.01596457e-01
-8.45001578e-01 3.68592352e-01 3.82780045e-01 1.61455825e-01
-9.85143542e-01 -6.51052713e-01 -8.78926992e-01 2.34990329e-01
3.33199680e-01 -8.84160817e-01 7.56983817e-01 -1.13175893e+00
-1.92616069e+00 6.04925811e-01 2.02830229e-02 1.24629505e-01
5.59575498e-01 -8.17977414e-02 -1.30588487e-01 2.21835777e-01
-4.87164967e-02 9.11367714e-01 1.33524442e+00 -1.67849171e+00
-2.93142676e-01 -1.20582022e-02 2.46604636e-01 5.60538694e-02
-3.25006902e-01 -3.64326537e-01 -8.36153746e-01 -8.58211279e-01
-6.84735999e-02 -9.08372819e-01 -2.37220973e-01 2.45992154e-01
-7.41783440e-01 5.68988502e-01 8.24519157e-01 -4.88673806e-01
9.47838545e-01 -2.32784796e+00 2.87830621e-01 2.40992531e-01
4.04837191e-01 1.29977554e-01 -4.51254249e-01 5.68384409e-01
1.81497317e-02 1.85998574e-01 -3.78046662e-01 -4.65198845e-01
3.86427110e-03 3.63464832e-01 -3.43678147e-01 2.29552045e-01
4.24127460e-01 9.78956819e-01 -8.34342659e-01 -3.51133525e-01
2.42396146e-01 8.59286427e-01 -8.34781349e-01 4.01793122e-01
-2.99721003e-01 7.03933835e-01 -3.61524493e-01 3.97554755e-01
1.10946846e+00 -3.42574030e-01 2.79378772e-01 -3.31074834e-01
-3.15246955e-02 -4.21159118e-01 -1.02504253e+00 1.99663329e+00
-6.88557565e-01 5.24326146e-01 2.05321506e-01 -6.06591225e-01
1.04388988e+00 -2.82885823e-02 7.53647909e-02 -7.29986012e-01
8.41959864e-02 1.96863100e-01 -2.74409890e-01 -9.14308056e-02
5.47743142e-01 -1.97220400e-01 -1.64826170e-01 5.68661332e-01
-1.50012538e-01 -3.50811929e-01 2.45249853e-01 3.03374797e-01
7.54624188e-01 3.37356359e-01 -1.82224154e-01 5.17623052e-02
3.73555303e-01 -5.20073354e-01 4.64283586e-01 4.66659397e-01
5.22052288e-01 1.20660949e+00 7.06839561e-01 -2.39117652e-01
-1.23747468e+00 -1.16476166e+00 3.98298502e-01 9.41890538e-01
4.47361320e-01 -4.30481583e-01 -8.32423031e-01 -3.85121614e-01
-1.98887825e-01 6.18420064e-01 -8.37956727e-01 -1.73479319e-01
-6.66754484e-01 -2.89739609e-01 4.31850374e-01 3.97843480e-01
7.31426954e-01 -8.71642947e-01 -7.45050311e-01 3.17092575e-02
-1.46122396e-01 -1.08492529e+00 -7.67421961e-01 -9.61879194e-02
-4.89647269e-01 -1.00803661e+00 -1.17061365e+00 -5.87288678e-01
1.00066948e+00 3.43777061e-01 1.00362897e+00 -9.52708274e-02
-4.03887361e-01 2.55493134e-01 -4.46273446e-01 -2.63041183e-02
-3.13394219e-01 -2.13435084e-01 -4.62014645e-01 4.21656609e-01
-3.45311105e-01 -7.50215650e-01 -1.06051815e+00 2.04411641e-01
-1.36832690e+00 6.93648696e-01 8.03307772e-01 1.04328299e+00
4.89836335e-01 -2.97090292e-01 7.99857378e-02 -1.08643162e+00
3.69648397e-01 -1.61891162e-01 -6.94331944e-01 4.54310685e-01
-3.91702682e-01 4.22560185e-01 8.63250315e-01 -4.11702156e-01
-1.27212846e+00 7.95643628e-02 6.86568860e-03 -5.87226152e-01
6.99109957e-02 2.27698922e-01 -2.84069836e-01 -1.38079152e-01
4.38496530e-01 4.55156952e-01 5.15507050e-02 -5.57045579e-01
9.27002788e-01 1.88973233e-01 6.25929952e-01 -7.68095672e-01
1.23759699e+00 6.76484287e-01 1.23065621e-01 -5.04001141e-01
-4.89381969e-01 2.19015688e-01 -4.04813796e-01 -3.96125689e-02
9.99381363e-01 -9.61805224e-01 -5.59055567e-01 4.92625952e-01
-1.05656374e+00 -4.38524783e-01 -2.67826468e-01 1.52038008e-01
-4.50177431e-01 3.58942360e-01 -5.06935060e-01 -6.31215811e-01
-3.87386650e-01 -1.19797599e+00 1.27880275e+00 4.04360175e-01
3.18750888e-01 -6.55474901e-01 -5.21330722e-03 -7.75662959e-02
6.42143726e-01 7.90471971e-01 7.84142673e-01 4.65472996e-01
-9.49226737e-01 1.47085367e-02 -6.06181264e-01 1.85622305e-01
3.12096655e-01 1.98645085e-01 -8.74240637e-01 -4.18450981e-01
-5.08751512e-01 -3.12237531e-01 1.01786602e+00 9.67403427e-02
1.35282004e+00 -4.61704642e-01 7.91580789e-03 1.13164949e+00
1.75181973e+00 1.79803252e-01 7.93854713e-01 8.32523704e-02
1.00445008e+00 3.78836513e-01 1.26552075e-01 6.12671852e-01
1.60153374e-01 4.47736442e-01 5.15986323e-01 -7.21999764e-01
-4.44322944e-01 -6.30134404e-01 3.34171236e-01 4.09348875e-01
-7.42929056e-02 -4.85456288e-01 -4.46265370e-01 3.32332224e-01
-1.45596409e+00 -7.69077778e-01 3.36777538e-01 1.84424758e+00
8.64628434e-01 -3.38021159e-01 -9.94426534e-02 -2.56669044e-01
7.85370588e-01 3.74171913e-01 -6.28646493e-01 -1.58120424e-01
-3.09994251e-01 4.21412438e-01 3.85713845e-01 4.40519452e-01
-6.45171225e-01 9.79797363e-01 5.89117384e+00 8.39146614e-01
-1.38120139e+00 -2.16799509e-03 8.42095435e-01 -1.68305933e-01
-1.02558482e+00 9.83496457e-02 -2.12236345e-01 5.36760628e-01
2.13119134e-01 9.44923386e-02 6.57123566e-01 6.48707449e-01
-7.06794262e-02 -3.49532701e-02 -8.59137356e-01 1.33810771e+00
7.16854483e-02 -1.34459746e+00 3.48969311e-01 1.68065652e-01
9.62201953e-01 -4.64503527e-01 5.69216669e-01 -6.32984191e-02
4.63274360e-01 -1.08782244e+00 9.88352060e-01 4.90507573e-01
1.50728202e+00 -7.75370002e-01 9.03987363e-02 -2.94175800e-02
-1.19279671e+00 9.02935937e-02 -3.40263337e-01 2.76018620e-01
3.19615364e-01 5.52530646e-01 -4.32930946e-01 5.26532173e-01
4.36002761e-01 5.67068636e-01 -6.31916761e-01 6.84294164e-01
-5.30527711e-01 4.15988684e-01 -2.91835070e-01 1.38366178e-01
2.99535334e-01 -3.69985789e-01 1.65403202e-01 1.15849590e+00
7.64074385e-01 8.11082944e-02 7.88528100e-03 1.39912641e+00
-8.31354409e-02 -2.33404800e-01 -5.33592463e-01 -2.93400586e-01
3.75718534e-01 1.33341384e+00 -6.08895004e-01 -1.77966207e-01
-3.14159125e-01 1.35780191e+00 1.64003164e-01 8.56803715e-01
-9.97125268e-01 -6.15899503e-01 5.75816393e-01 1.27137797e-02
5.10243058e-01 -3.18044186e-01 -1.36613935e-01 -1.46855581e+00
-1.01168640e-01 -8.37355256e-01 -3.75398882e-02 -1.02274978e+00
-9.97089326e-01 8.43897820e-01 -1.85475603e-01 -1.23134840e+00
-2.53207773e-01 -5.85896492e-01 -7.16827273e-01 1.01484835e+00
-1.66631842e+00 -1.63639235e+00 -6.91177309e-01 7.27148473e-01
9.83213857e-02 5.12390919e-02 7.21836329e-01 1.88576609e-01
-5.82472265e-01 7.09658086e-01 2.70586669e-01 8.84077698e-02
8.28391254e-01 -1.28951263e+00 4.85806704e-01 1.15688288e+00
5.17603494e-02 5.57955325e-01 5.44528484e-01 -4.00183558e-01
-1.74193263e+00 -1.01760697e+00 -2.45754527e-05 4.87648174e-02
1.90235704e-01 -7.06422985e-01 -6.19435310e-01 3.57893944e-01
6.00068092e-01 -1.28512841e-03 5.43351352e-01 -3.14073116e-01
-8.01386774e-01 -2.44506657e-01 -1.01615584e+00 7.60584414e-01
1.01020288e+00 -5.67333519e-01 1.06391698e-01 -2.06930831e-01
6.41396284e-01 -2.56859481e-01 -5.91728449e-01 1.85694788e-02
6.25345826e-01 -1.03950751e+00 9.60092425e-01 -9.10730734e-02
6.99017406e-01 -7.45500028e-01 -1.78345829e-01 -1.31682277e+00
-5.17188668e-01 -9.28276539e-01 2.11862803e-01 1.21674621e+00
2.89226323e-01 -5.55258870e-01 7.03605533e-01 5.01825035e-01
-8.77269357e-02 -4.23488498e-01 -5.51571190e-01 -4.70399737e-01
-3.81347649e-02 -8.45466629e-02 9.16705310e-01 6.55817389e-01
-4.42338347e-01 1.18585512e-01 -7.37864614e-01 -5.80945164e-02
7.64771044e-01 5.73069215e-01 9.98727739e-01 -7.79463351e-01
-5.93724370e-01 -3.52281243e-01 -1.87926888e-01 -1.20916414e+00
-1.46794468e-01 -5.34342885e-01 7.19515532e-02 -1.43696845e+00
2.74716586e-01 -4.07066286e-01 -2.41176113e-02 5.70958734e-01
-3.01755875e-01 7.79514909e-01 5.69561899e-01 9.50203836e-02
-4.30967242e-01 8.19006801e-01 1.86010683e+00 -2.79204190e-01
8.95669162e-02 -6.35047853e-01 -9.51692879e-01 3.47190052e-01
6.72520578e-01 -1.01413094e-01 -6.13705754e-01 -7.04802215e-01
4.42722231e-01 -8.53941888e-02 4.42960739e-01 -8.81715119e-01
-1.05210096e-01 -3.33504587e-01 7.43479729e-01 -4.02305633e-01
5.33702314e-01 -6.88344002e-01 6.12309515e-01 4.30636853e-01
-7.26118535e-02 -9.99300256e-02 8.50969031e-02 7.22161293e-01
-3.35203946e-01 8.43502581e-02 9.85671639e-01 -2.56333321e-01
-6.58975482e-01 4.03818190e-01 -1.37010012e-02 7.42946640e-02
8.90375376e-01 -3.25828940e-01 -4.33232099e-01 -5.55738449e-01
-4.74018842e-01 -1.00682847e-01 1.00387180e+00 2.43518978e-01
8.36278439e-01 -1.55991554e+00 -7.26008117e-01 5.79822242e-01
2.54961222e-01 1.68807179e-01 5.56603193e-01 4.03262854e-01
-9.72099662e-01 1.24239493e-02 -5.47460794e-01 -5.14823198e-01
-8.66058350e-01 6.76930189e-01 5.53758964e-02 -1.39214871e-02
-7.07285166e-01 9.52344775e-01 1.00186932e+00 -9.52940248e-03
-1.36665866e-01 -3.37995976e-01 1.60586640e-01 -2.38917455e-01
4.88477677e-01 -5.00315912e-02 -4.47084606e-01 -4.51035678e-01
1.44863687e-02 8.80190790e-01 1.53790295e-01 -3.80180806e-01
1.30568957e+00 -2.43117452e-01 -2.08463073e-01 4.82135564e-02
1.32643533e+00 2.88406789e-01 -1.85325491e+00 -1.91291332e-01
-7.26608634e-01 -9.95197713e-01 -1.03594422e-01 -6.92023218e-01
-1.47529876e+00 1.06510234e+00 5.50504088e-01 -5.00168353e-02
1.45408356e+00 -2.54603684e-01 8.71601760e-01 5.80276642e-03
3.54253650e-01 -8.10133576e-01 5.04571378e-01 2.40158558e-01
9.30725813e-01 -8.60054374e-01 -1.76141471e-01 -4.36111599e-01
-8.13049734e-01 1.22145772e+00 5.42755663e-01 -1.90724432e-01
3.27589780e-01 1.25426024e-01 -8.72235280e-03 -4.46732454e-02
-5.09424984e-01 -1.78975776e-01 3.58614504e-01 7.12327540e-01
2.91963547e-01 1.06273614e-01 2.69601922e-02 1.89935803e-01
-9.16494504e-02 -2.36519724e-01 6.30746543e-01 6.04639828e-01
-1.81482926e-01 -1.32887900e+00 -4.62679774e-01 -9.84954983e-02
-2.20220283e-01 -2.27830872e-01 -3.20596784e-01 6.12514019e-01
1.98159337e-01 7.21956015e-01 6.66267946e-02 -4.34129357e-01
1.67218838e-02 -2.53581226e-01 6.84488535e-01 -4.80360359e-01
-9.39340815e-02 4.87840205e-01 -1.88995332e-01 -7.14711130e-01
-2.94045419e-01 -1.68487862e-01 -8.67886066e-01 -3.79853398e-01
-2.42209826e-02 -5.32662831e-02 5.90941906e-01 3.22361320e-01
5.79892814e-01 5.36052644e-01 7.25667536e-01 -1.05242288e+00
-5.07825203e-02 -7.47846246e-01 -8.26051235e-01 6.03485227e-01
4.18593884e-01 -5.70050955e-01 -1.51462704e-01 1.96265280e-01]
|
[11.507645606994629, -0.8668014407157898]
|
2e499fa8-631a-496f-be37-b77f4585ca17
|
extrapolative-controlled-sequence-generation
|
2303.04562
| null |
https://arxiv.org/abs/2303.04562v3
|
https://arxiv.org/pdf/2303.04562v3.pdf
|
Extrapolative Controlled Sequence Generation via Iterative Refinement
|
We study the problem of extrapolative controlled generation, i.e., generating sequences with attribute values beyond the range seen in training. This task is of significant importance in automated design, especially drug discovery, where the goal is to design novel proteins that are \textit{better} (e.g., more stable) than existing sequences. Thus, by definition, the target sequences and their attribute values are out of the training distribution, posing challenges to existing methods that aim to directly generate the target sequence. Instead, in this work, we propose Iterative Controlled Extrapolation (ICE) which iteratively makes local edits to a sequence to enable extrapolation. We train the model on synthetically generated sequence pairs that demonstrate small improvement in the attribute value. Results on one natural language task (sentiment analysis) and two protein engineering tasks (ACE2 stability and AAV fitness) show that ICE considerably outperforms state-of-the-art approaches despite its simplicity. Our code and models are available at: https://github.com/vishakhpk/iter-extrapolation.
|
['Ankur P. Parikh', 'He He', 'Richard Yuanzhe Pang', 'Vishakh Padmakumar']
|
2023-03-08
| null | null | null | null |
['drug-discovery']
|
['medical']
|
[ 7.34824955e-01 1.58609927e-01 -1.26300976e-01 -3.72983366e-01
-7.20730186e-01 -8.78490150e-01 3.54639649e-01 3.73626590e-01
-2.56290495e-01 1.49552798e+00 -1.23182155e-01 -4.85796720e-01
1.72646612e-01 -5.62782109e-01 -1.17469561e+00 -7.56336808e-01
1.24612093e-01 6.90874219e-01 -5.68303801e-02 -5.25470674e-01
3.63236368e-01 3.22392315e-01 -1.39275229e+00 3.17814380e-01
1.30216694e+00 4.36249137e-01 2.40314201e-01 5.69912970e-01
1.10978633e-01 4.04296815e-01 -5.57722688e-01 -2.31421307e-01
1.24513194e-01 -9.22419727e-01 -6.86285973e-01 -2.29025558e-01
8.78309160e-02 -3.08480393e-02 3.14070195e-01 9.90382016e-01
7.00381994e-01 7.41160139e-02 8.36682677e-01 -1.00945485e+00
-7.01423705e-01 4.24885929e-01 -3.28415126e-01 -9.58106071e-02
4.55799907e-01 4.69403833e-01 9.21297371e-01 -8.48215759e-01
8.96345735e-01 8.32688868e-01 4.99913037e-01 8.46744299e-01
-1.69147873e+00 -5.73719263e-01 -3.01069673e-02 -7.79351592e-02
-1.22400415e+00 -4.33137864e-01 3.92184675e-01 -5.11560023e-01
1.06105912e+00 3.58522415e-01 5.11340916e-01 1.10165143e+00
2.30978355e-01 5.30648828e-01 9.98037875e-01 -3.41107786e-01
6.56760991e-01 9.91716608e-02 -3.60657543e-01 3.88151884e-01
8.81367028e-02 6.21327572e-02 -6.25822842e-01 -4.96949941e-01
2.54401267e-01 -2.33998194e-01 -3.98738921e-01 -4.93976533e-01
-1.26968002e+00 9.09738481e-01 4.03204449e-02 -3.93395126e-02
-6.42842531e-01 -6.54958934e-02 2.24682704e-01 1.61395952e-01
2.80603766e-01 1.01486957e+00 -7.94456303e-01 -3.82482648e-01
-5.85115135e-01 7.26948440e-01 7.89847314e-01 9.52906370e-01
6.09920859e-01 -1.13489494e-01 -1.76047534e-01 7.42560625e-01
-2.09054798e-01 1.53704420e-01 3.11951935e-01 -6.77107155e-01
1.88090891e-01 4.29420441e-01 4.30636436e-01 -3.68319720e-01
-1.73299670e-01 -3.90664220e-01 -5.36973536e-01 2.91259825e-01
5.71576416e-01 -3.59019995e-01 -9.65841055e-01 1.93628848e+00
4.42281634e-01 -1.09672166e-01 1.54823542e-01 7.83415496e-01
3.91143471e-01 7.68145263e-01 3.07211876e-01 -5.71809530e-01
1.08169174e+00 -7.05462992e-01 -5.29266477e-01 -6.98033050e-02
7.19788730e-01 -6.96440101e-01 1.21431649e+00 6.02621555e-01
-1.03205800e+00 -3.57234180e-01 -1.09838200e+00 3.10843110e-01
-2.71767050e-01 -4.85304222e-02 5.85145354e-01 5.90880096e-01
-6.71792686e-01 9.85178351e-01 -5.93624532e-01 -1.36708766e-01
3.33215684e-01 4.78202790e-01 -2.29770750e-01 1.63326532e-01
-1.21170533e+00 7.58124352e-01 6.77269697e-01 -1.51406869e-01
-7.89945424e-01 -1.06554294e+00 -5.70139945e-01 -1.79143369e-01
4.36261117e-01 -6.70957923e-01 1.29751992e+00 -1.04775918e+00
-1.55370486e+00 4.24054205e-01 -1.98142618e-01 -5.07969081e-01
6.67559624e-01 -1.16301529e-01 1.00610815e-01 -2.98454821e-01
-7.88634643e-02 8.23829949e-01 5.59611082e-01 -9.13240314e-01
-3.21987271e-01 -1.31506249e-01 -1.17046870e-01 1.33526772e-01
-5.12693673e-02 -7.59200950e-04 9.13326889e-02 -6.58341408e-01
-4.76921141e-01 -1.18496895e+00 -6.82292044e-01 -1.10945627e-01
-5.71368277e-01 -1.85184911e-01 2.35777438e-01 -6.67060316e-01
1.20943058e+00 -1.59945583e+00 4.81841683e-01 3.02637726e-01
-2.63863523e-03 5.12858927e-01 -6.42949045e-02 7.11144865e-01
-4.42088753e-01 2.04724282e-01 -6.65785909e-01 3.43003571e-01
-2.67383814e-01 -1.73594400e-01 -1.72417760e-01 1.71007693e-01
3.83133501e-01 9.05850887e-01 -1.02371681e+00 -8.15602988e-02
-2.40488485e-01 4.52971429e-01 -7.77743936e-01 2.10550666e-01
-1.01158369e+00 9.66847062e-01 -5.29676735e-01 3.70224625e-01
4.72891957e-01 -3.79452646e-01 3.99689078e-01 4.45330776e-02
-6.74656332e-02 1.19654931e-01 -6.62905037e-01 1.63944769e+00
-1.25467703e-01 1.27754614e-01 -6.29725933e-01 -6.22234404e-01
1.08409226e+00 2.04295486e-01 4.50428456e-01 -3.28068286e-01
-2.93229502e-02 3.59076589e-01 4.28122789e-01 -1.66972056e-01
2.62333512e-01 -2.80604720e-01 1.36593372e-01 4.39393759e-01
-4.27027643e-01 -1.77913025e-01 4.16361988e-01 1.38123855e-01
1.09517097e+00 6.73039377e-01 4.81067717e-01 -2.24546939e-01
4.52035278e-01 3.27404141e-01 8.25640559e-01 4.16181386e-01
5.51695414e-02 5.97095668e-01 8.57454300e-01 -2.60978311e-01
-1.61628568e+00 -7.56443024e-01 5.18395156e-02 8.79809380e-01
-2.78888911e-01 -6.03148699e-01 -9.81558621e-01 -6.32536590e-01
3.12508084e-02 9.56818879e-01 -4.64081228e-01 -3.57713729e-01
-6.15456343e-01 -9.05900776e-01 4.09305841e-01 3.37254524e-01
-1.85863122e-01 -1.09971690e+00 -5.17553449e-01 4.47058469e-01
-1.73356399e-01 -6.48223758e-01 -7.92158008e-01 1.61844105e-01
-5.47920585e-01 -7.65843034e-01 -9.18147981e-01 -6.73965573e-01
7.33695447e-01 -4.11255836e-01 1.02873874e+00 -1.75896175e-02
-3.53577077e-01 -4.38221365e-01 -3.85780305e-01 -6.50971949e-01
-8.76486182e-01 1.11959048e-01 6.04875572e-02 -3.49008650e-01
2.90526479e-01 -6.22603893e-01 -7.98520803e-01 3.80969405e-01
-8.83508980e-01 3.69772494e-01 5.31385362e-01 1.16267443e+00
9.14105952e-01 -3.33982229e-01 1.09496653e+00 -1.10799015e+00
9.14792717e-01 -3.69996518e-01 -6.27212882e-01 2.96873540e-01
-7.76629150e-01 3.71267796e-01 1.00291598e+00 -6.82495594e-01
-9.84390557e-01 4.80778307e-01 -4.08566684e-01 9.47492849e-03
-2.77043171e-02 5.23249805e-01 -1.99615538e-01 1.34827003e-01
1.01159763e+00 4.17807102e-01 3.45589787e-01 -2.97000349e-01
2.54300326e-01 5.62830508e-01 9.70953181e-02 -7.20947564e-01
4.17734355e-01 -5.03808931e-02 1.99973788e-02 -6.44368887e-01
-4.70332831e-01 8.20364989e-03 -2.77085125e-01 4.15061377e-02
5.99254012e-01 -7.00169265e-01 -8.42198193e-01 4.17818218e-01
-9.61957335e-01 -5.95659912e-01 -3.41467947e-01 2.15204805e-01
-9.81660604e-01 3.29171807e-01 -4.17950064e-01 -6.58932626e-01
-5.93775332e-01 -1.30973756e+00 9.05295789e-01 1.46179780e-01
-6.70070350e-01 -5.46306372e-01 1.14752889e-01 2.88116932e-01
2.84498036e-01 6.22356236e-01 1.15839553e+00 -8.15606356e-01
-5.71016848e-01 6.72854707e-02 3.10273737e-01 3.22172463e-01
2.94956863e-01 -5.53416274e-02 -4.62850571e-01 -3.50733936e-01
-4.45102990e-01 -5.11534333e-01 6.04600728e-01 2.50929743e-01
1.01961136e+00 -3.30778092e-01 -3.57953995e-01 3.28085840e-01
1.09811831e+00 5.56793690e-01 7.11275160e-01 1.91575319e-01
3.11502397e-01 5.64236581e-01 1.07297564e+00 6.64838552e-01
-1.83641061e-01 8.10163677e-01 1.33512974e-01 6.85554221e-02
1.69818565e-01 -4.86299038e-01 3.23572874e-01 1.19520910e-01
-3.51070687e-02 -4.70591128e-01 -8.33369553e-01 4.46367979e-01
-1.94214952e+00 -8.61162722e-01 -9.66545716e-02 2.28252721e+00
1.62339747e+00 8.23908970e-02 3.01173061e-01 -2.31138363e-01
7.05733836e-01 -3.93300235e-01 -1.27822232e+00 -5.62767148e-01
-1.55278563e-01 4.45832819e-01 3.80698264e-01 3.35949123e-01
-7.56071210e-01 9.85226989e-01 5.79373789e+00 9.61530149e-01
-9.69303548e-01 -3.54852080e-01 1.12110722e+00 -7.67727895e-03
-5.65879166e-01 1.40162662e-01 -7.91712999e-01 6.18847668e-01
1.11694503e+00 -3.61745864e-01 4.78342295e-01 6.54985726e-01
5.34214854e-01 1.43226609e-01 -1.14868140e+00 5.14653563e-01
-3.37491542e-01 -1.58634973e+00 1.27509579e-01 1.01531997e-01
8.84283125e-01 -4.14437503e-01 1.22060768e-01 6.10521389e-03
3.39798242e-01 -1.25739193e+00 5.27114034e-01 5.28844893e-01
9.45926070e-01 -9.82302547e-01 4.54105884e-01 5.03841758e-01
-5.17025769e-01 2.65322894e-01 -1.91853985e-01 2.57879525e-01
2.29705274e-01 5.72474718e-01 -1.18004692e+00 2.13770345e-01
2.79888093e-01 5.04334390e-01 -2.75398195e-01 7.15607405e-01
-2.57663071e-01 5.47620296e-01 -2.37236261e-01 -4.36677903e-01
-6.54859319e-02 -4.20205444e-01 5.08641958e-01 9.22473252e-01
3.60001415e-01 8.31341594e-02 5.73671274e-02 1.14210296e+00
-1.83535069e-01 3.76740605e-01 -4.76071507e-01 -4.18269336e-01
3.47298205e-01 7.98354983e-01 -4.50250775e-01 -2.58587569e-01
3.64525802e-02 1.05662262e+00 2.11101592e-01 2.97989726e-01
-1.02615094e+00 -6.76571548e-01 9.01118159e-01 1.57545373e-01
4.58903879e-01 -4.87303585e-02 -1.57930300e-01 -7.47746706e-01
7.68436259e-03 -1.37260425e+00 1.94497049e-01 -6.79099023e-01
-1.06161201e+00 5.00044644e-01 -2.93082654e-01 -1.11065078e+00
-5.03154039e-01 -4.51245040e-01 -3.08297813e-01 1.11963570e+00
-1.16693974e+00 -7.43138373e-01 1.62916109e-01 -5.26086465e-02
7.25479364e-01 -3.33268531e-02 7.29607582e-01 7.93091953e-02
-2.59713024e-01 8.09722900e-01 3.55134338e-01 -5.05534887e-01
8.33731532e-01 -1.06452727e+00 8.56395841e-01 4.85246807e-01
-4.63809818e-01 1.00214112e+00 1.26759124e+00 -9.98636067e-01
-1.32128131e+00 -1.17055750e+00 7.99046993e-01 -4.03396249e-01
4.21059310e-01 -4.19496983e-01 -9.42442477e-01 3.34396631e-01
7.46138692e-02 -4.94928926e-01 7.96322644e-01 -3.37503791e-01
-1.30049765e-01 4.15013105e-01 -1.28202558e+00 8.22571218e-01
1.23827434e+00 6.75984621e-02 9.89520699e-02 6.30135953e-01
9.82359052e-01 -4.30909067e-01 -9.76308763e-01 4.76016045e-01
5.74909329e-01 -7.61375368e-01 9.16372240e-01 -1.17368901e+00
6.17361665e-01 -6.25344217e-01 5.51514477e-02 -1.64155304e+00
-2.20285490e-01 -9.96625185e-01 -5.14330948e-03 8.90945971e-01
1.09352100e+00 -6.50033653e-01 1.06664717e+00 5.72257936e-01
1.23243630e-02 -1.11703181e+00 -5.88844121e-01 -9.09070492e-01
3.02745104e-01 2.00236380e-01 7.73671746e-01 7.60882080e-01
2.70867020e-01 5.25279164e-01 -4.87211198e-01 -3.08019489e-01
3.98090720e-01 4.79969755e-02 6.87719882e-01 -7.87012637e-01
-6.22127891e-01 -2.74009317e-01 -7.95021653e-02 -1.10038269e+00
5.33126593e-02 -8.43333542e-01 2.30124474e-01 -1.25173068e+00
2.38544434e-01 -4.68641579e-01 1.24382190e-02 4.92226064e-01
-4.84426737e-01 -9.36987158e-03 -4.94603403e-02 -5.11112399e-02
-1.94338635e-01 7.60459006e-01 1.25741315e+00 -8.91809314e-02
-4.66933876e-01 9.35537964e-02 -9.69504595e-01 2.05442384e-01
1.23684096e+00 -4.84964997e-01 -5.34166157e-01 3.17994267e-01
5.14407277e-01 5.74368611e-02 -8.08384046e-02 -5.80568016e-01
-2.08891898e-01 -4.73275244e-01 2.18414322e-01 -2.89306641e-01
3.45800787e-01 -2.97626346e-01 5.30668259e-01 6.91595495e-01
-7.02284575e-01 1.36127532e-01 2.16558307e-01 6.38312936e-01
3.02732497e-01 -2.59833544e-01 7.88725555e-01 -1.43297523e-01
-1.98618010e-01 3.09752733e-01 -4.47362304e-01 9.54431891e-02
1.21799028e+00 -1.78782880e-01 -3.32738787e-01 -4.59718615e-01
-6.65312529e-01 1.76911578e-01 7.76940465e-01 3.00232768e-01
8.11687887e-01 -9.47947085e-01 -9.60503817e-01 6.32091835e-02
4.15858030e-01 -4.90623601e-02 -8.43893066e-02 6.97597980e-01
-6.42275035e-01 3.79358441e-01 -8.87016803e-02 -4.17224199e-01
-1.37760997e+00 6.40328586e-01 3.36666256e-01 -7.02545512e-03
-2.91705489e-01 8.05005133e-01 2.26752490e-01 -6.26576900e-01
-1.29778638e-01 1.46540301e-02 9.99390930e-02 -3.29918444e-01
4.74601418e-01 1.38044655e-01 1.08136967e-01 -2.50141919e-01
-1.99777767e-01 2.47383595e-01 -3.81152183e-01 1.06055863e-01
1.45145059e+00 4.28347468e-01 -5.41063212e-02 1.99170485e-02
9.37621772e-01 -1.75574869e-01 -1.35919440e+00 1.59611721e-02
1.24711134e-01 -4.74120319e-01 -5.12651145e-01 -1.16503918e+00
-4.75147605e-01 4.27325398e-01 5.51724434e-01 -3.50580335e-01
8.87392700e-01 -6.66131675e-02 8.40706646e-01 4.44345772e-01
3.16552460e-01 -9.62884068e-01 -3.08592413e-02 2.57794946e-01
9.06477809e-01 -9.08267617e-01 1.51764210e-02 -4.70293969e-01
-9.54694092e-01 8.01677883e-01 5.82243979e-01 3.03861201e-01
1.46725357e-01 1.16844095e-01 -2.05136344e-01 1.29285529e-01
-1.02100003e+00 -6.64837891e-03 -1.72140852e-01 6.11539960e-01
8.05212855e-01 6.43154904e-02 -7.05859542e-01 2.34214857e-01
-2.16956586e-01 2.09597006e-01 5.94281197e-01 1.19047308e+00
-4.17911679e-01 -1.55067706e+00 -1.67562544e-01 4.95031893e-01
-5.08507133e-01 -3.88906181e-01 -7.66405523e-01 1.90469399e-01
-2.95809042e-02 9.16846335e-01 -5.08452594e-01 -2.10899949e-01
4.20331359e-01 1.58914715e-01 5.09022832e-01 -4.74807590e-01
-5.36937237e-01 -3.66047882e-02 3.92695367e-01 -1.71756059e-01
7.61555135e-02 -8.75845850e-01 -1.48877871e+00 -2.46561125e-01
-4.36118901e-01 5.89992523e-01 6.67676091e-01 4.52084869e-01
7.98923016e-01 4.98178512e-01 4.63818908e-01 -3.73801172e-01
-7.49118388e-01 -7.61020720e-01 -1.92960665e-01 4.61964995e-01
1.98999971e-01 -3.82239610e-01 9.93291661e-03 4.27827328e-01]
|
[4.726588249206543, 5.601823806762695]
|
41cc94c4-2ff6-4881-ba81-fdc807562956
|
humans-in-humans-out-on-gpt-converging-toward
|
2303.17276
| null |
https://arxiv.org/abs/2303.17276v1
|
https://arxiv.org/pdf/2303.17276v1.pdf
|
Humans in Humans Out: On GPT Converging Toward Common Sense in both Success and Failure
|
Increase in computational scale and fine-tuning has seen a dramatic improvement in the quality of outputs of large language models (LLMs) like GPT. Given that both GPT-3 and GPT-4 were trained on large quantities of human-generated text, we might ask to what extent their outputs reflect patterns of human thinking, both for correct and incorrect cases. The Erotetic Theory of Reason (ETR) provides a symbolic generative model of both human success and failure in thinking, across propositional, quantified, and probabilistic reasoning, as well as decision-making. We presented GPT-3, GPT-3.5, and GPT-4 with 61 central inference and judgment problems from a recent book-length presentation of ETR, consisting of experimentally verified data-points on human judgment and extrapolated data-points predicted by ETR, with correct inference patterns as well as fallacies and framing effects (the ETR61 benchmark). ETR61 includes classics like Wason's card task, illusory inferences, the decoy effect, and opportunity-cost neglect, among others. GPT-3 showed evidence of ETR-predicted outputs for 59% of these examples, rising to 77% in GPT-3.5 and 75% in GPT-4. Remarkably, the production of human-like fallacious judgments increased from 18% in GPT-3 to 33% in GPT-3.5 and 34% in GPT-4. This suggests that larger and more advanced LLMs may develop a tendency toward more human-like mistakes, as relevant thought patterns are inherent in human-produced training data. According to ETR, the same fundamental patterns are involved both in successful and unsuccessful ordinary reasoning, so that the "bad" cases could paradoxically be learned from the "good" cases. We further present preliminary evidence that ETR-inspired prompt engineering could reduce instances of these mistakes.
|
['Vincent Wang-Maścianica', 'Philipp Koralus']
|
2023-03-30
| null | null | null | null |
['common-sense-reasoning']
|
['reasoning']
|
[-5.72639257e-02 7.32121348e-01 2.85435438e-01 -3.70321542e-01
-4.94179875e-01 -5.88932097e-01 9.17457461e-01 1.10370211e-01
-2.77534723e-01 6.79394901e-01 5.40769219e-01 -8.32867563e-01
-4.31106687e-01 -9.97981191e-01 -6.94001496e-01 -2.91310847e-01
2.22304195e-01 8.45054686e-01 4.12595719e-02 -5.30683041e-01
7.96292603e-01 1.84376404e-01 -1.07672668e+00 6.91426158e-01
8.79573643e-01 5.71111143e-01 -5.53738624e-02 8.54449689e-01
4.82021868e-02 1.54106712e+00 -7.43151844e-01 -1.04618561e+00
-3.75147946e-02 -3.32668334e-01 -8.85224879e-01 -4.99653637e-01
3.17987144e-01 3.34601812e-02 -3.83148313e-01 9.02550638e-01
5.31882823e-01 5.84102012e-02 9.64284301e-01 -1.01653707e+00
-1.22835684e+00 1.24701762e+00 -2.36499190e-01 2.83238441e-01
7.20982492e-01 5.71575463e-01 1.03167474e+00 -7.62741804e-01
4.30060506e-01 1.99887598e+00 8.02723587e-01 6.95345581e-01
-1.50701678e+00 -6.08735263e-01 -1.35695815e-01 2.49697845e-02
-1.12849212e+00 -2.61411399e-01 2.28775784e-01 -5.88243067e-01
1.38262951e+00 2.25992844e-01 7.03539789e-01 1.48131120e+00
7.22253323e-01 6.09732330e-01 1.68193483e+00 -6.31255865e-01
4.60686415e-01 9.05798823e-02 2.23845318e-01 5.54498732e-01
4.78238851e-01 4.39273715e-01 -6.71162605e-01 -3.45005006e-01
8.45933437e-01 -5.21408379e-01 6.95057167e-03 4.71254915e-01
-1.43256402e+00 8.83387148e-01 4.16629136e-01 4.55860317e-01
-4.26713616e-01 2.69785076e-01 1.89347968e-01 3.38700235e-01
1.86715081e-01 9.33070183e-01 -6.50646150e-01 -3.76006901e-01
-7.12956905e-01 8.11999261e-01 1.23442352e+00 6.83327377e-01
1.92859724e-01 7.85756558e-02 -4.14129257e-01 6.75514996e-01
2.74157315e-01 6.90602422e-01 6.98529601e-01 -1.28393734e+00
4.87088054e-01 4.78509516e-01 3.61390591e-01 -1.46117496e+00
-7.25630105e-01 -5.12160063e-01 -4.43993300e-01 1.56894073e-01
8.17453206e-01 -7.53318816e-02 -5.45227766e-01 1.96239340e+00
-3.86899352e-01 -6.13792300e-01 3.17223340e-01 6.72764719e-01
4.64021176e-01 6.42483354e-01 4.37948793e-01 3.80520895e-02
1.52296436e+00 -5.11395991e-01 -4.18361783e-01 -7.95545995e-01
8.99403572e-01 -8.13340366e-01 1.40113461e+00 6.05307877e-01
-1.26411307e+00 -4.35847640e-01 -6.70664787e-01 -1.14289872e-01
-1.34974852e-01 -1.89232469e-01 7.42896497e-01 6.63217425e-01
-1.16603589e+00 6.51814044e-01 -2.55473763e-01 -3.12037259e-01
3.88279408e-01 -7.63414279e-02 1.21846810e-01 -2.06572756e-01
-1.46726477e+00 1.51144218e+00 5.28366745e-01 9.14134905e-02
-6.67837203e-01 -4.74777073e-01 -4.41364616e-01 3.13418299e-01
3.59865457e-01 -8.26722801e-01 1.50327182e+00 -7.45923877e-01
-1.26015699e+00 1.06840301e+00 -1.67631749e-02 -4.90446597e-01
6.92958117e-01 -2.30454773e-01 -6.65009260e-01 -2.29928359e-01
2.11478800e-01 5.48805773e-01 5.33774674e-01 -8.92891645e-01
-2.18841180e-01 -4.93131801e-02 1.25097126e-01 5.13155498e-02
5.70856571e-01 2.05650285e-01 7.19904780e-01 -7.90942609e-01
1.35156348e-01 -6.78429961e-01 -2.40415167e-02 -2.90227085e-01
-4.39075857e-01 -6.47288799e-01 -1.59420282e-01 -6.54805005e-01
1.09552455e+00 -1.97559273e+00 -1.78628489e-01 2.58156508e-01
4.89293188e-01 1.78451892e-02 -1.17145278e-01 5.29238164e-01
-1.36488572e-01 5.03934681e-01 1.85438320e-01 2.27749303e-01
7.40017593e-01 3.05737257e-01 -7.09134877e-01 -7.48885497e-02
2.20420271e-01 1.22829938e+00 -1.01328337e+00 -3.94505709e-01
-6.24455176e-02 -7.22617432e-02 -7.99234688e-01 -1.12881750e-01
-4.11873579e-01 -1.15049845e-02 -2.78122306e-01 4.19530988e-01
1.21631354e-01 -5.03034472e-01 9.02575701e-02 2.28513062e-01
-5.78458197e-02 7.89538622e-01 -6.18532002e-01 9.63768005e-01
-2.59628803e-01 6.60721660e-01 -6.48596704e-01 -5.85692286e-01
8.74001205e-01 3.00966084e-01 -7.30919957e-01 -1.02702105e+00
3.10254157e-01 4.55819786e-01 7.25429654e-01 -5.66364229e-01
5.45212805e-01 -8.12711298e-01 -4.01899725e-01 6.57674372e-01
-7.68208876e-02 -5.40313721e-01 2.45904580e-01 4.11156446e-01
1.09616983e+00 -1.74316123e-01 4.45744991e-01 -5.61254263e-01
3.10530365e-01 1.18204527e-01 4.67052728e-01 1.38704097e+00
-3.50059010e-02 2.64225185e-01 1.02385151e+00 -5.57677865e-01
-1.10493731e+00 -1.03596306e+00 -1.03620417e-01 1.21025717e+00
-3.86024565e-01 -4.69430268e-01 -7.43802249e-01 -1.44234374e-01
1.12001877e-03 1.90048575e+00 -6.70536041e-01 -3.91087055e-01
-4.76741344e-01 -6.39619708e-01 9.07451570e-01 4.99661744e-01
5.14137447e-01 -1.33729506e+00 -7.80023277e-01 4.00871783e-01
-2.36096114e-01 -9.00977671e-01 1.79307386e-01 1.38437245e-02
-7.61923194e-01 -8.85117233e-01 -3.02207559e-01 -2.33899817e-01
4.82996225e-01 -3.36580336e-01 1.38794065e+00 2.00511321e-01
-1.98116563e-02 1.72247097e-01 -2.59777099e-01 -4.66496021e-01
-8.90511334e-01 -4.85217214e-01 1.32912993e-01 -7.36808658e-01
6.27880454e-01 -4.30426329e-01 -2.54137963e-01 3.97738159e-01
-4.05333668e-01 2.75656164e-01 7.16392159e-01 8.76040220e-01
-2.79489011e-01 -2.27216352e-02 4.38568741e-01 -7.01754391e-01
1.01831567e+00 -3.82711202e-01 -3.04722875e-01 2.98388183e-01
-6.46828949e-01 2.88233221e-01 5.52076280e-01 -5.41521192e-01
-1.23331034e+00 -1.17623508e+00 9.31690782e-02 2.62317479e-01
-5.63616604e-02 4.68245506e-01 2.95676231e-01 2.61368543e-01
1.24478054e+00 1.81890488e-01 -4.22556341e-01 -1.37330294e-01
3.32536221e-01 3.96026760e-01 4.56022620e-01 -1.29543698e+00
5.63074946e-01 -2.71599054e-01 -4.01336253e-01 -3.15431118e-01
-1.19423532e+00 5.51047266e-01 9.23366100e-02 2.29794011e-02
7.23244786e-01 -8.15302372e-01 -1.09296978e+00 3.43811184e-01
-1.29840446e+00 -7.14920223e-01 -4.02801633e-01 3.87603849e-01
-7.93789387e-01 -3.90560664e-02 -8.83323669e-01 -9.57303762e-01
-1.64233521e-01 -1.00821412e+00 6.42229259e-01 2.75280058e-01
-1.13063037e+00 -1.01581812e+00 -1.10241190e-01 4.80673552e-01
5.52293360e-01 -1.42303482e-01 1.52220225e+00 -1.16622031e+00
-2.96538502e-01 -1.35595322e-01 -3.51569891e-01 2.23202676e-01
-4.80851948e-01 7.51872966e-03 -7.62600422e-01 1.62080035e-01
3.32051963e-01 -5.80901265e-01 4.81558502e-01 1.14184514e-01
6.78742409e-01 -6.30581319e-01 -1.21209949e-01 -1.29260011e-02
1.10608864e+00 3.25591803e-01 7.59171128e-01 2.65431732e-01
2.66062468e-01 6.74195170e-01 2.40463212e-01 2.01389551e-01
4.91455853e-01 2.73024708e-01 -3.87097359e-01 6.65169954e-01
7.74125084e-02 -6.56850636e-01 5.63697457e-01 6.40366495e-01
-3.29795003e-01 -4.30224270e-01 -1.38236344e+00 1.06009170e-01
-1.71863782e+00 -1.23865366e+00 -1.54848784e-01 1.93371439e+00
1.00651503e+00 7.29670167e-01 -2.64936954e-01 8.16126689e-02
5.81840098e-01 -2.22428776e-02 -2.36617386e-01 -9.20454681e-01
-2.85303861e-01 2.60758579e-01 2.05377340e-02 5.80231488e-01
-2.39426270e-01 1.00663781e+00 7.53352451e+00 7.88523972e-01
-7.02349901e-01 -4.71845903e-02 7.75973320e-01 7.70473480e-03
-5.26227176e-01 7.03715906e-02 -4.90879893e-01 4.26462322e-01
1.28078055e+00 -5.32163858e-01 5.63924432e-01 6.67948604e-01
1.84373200e-01 -4.86681402e-01 -1.42845726e+00 9.01285470e-01
-2.96863914e-02 -1.13536024e+00 9.99762043e-02 -1.48801148e-01
6.05161548e-01 -3.23165745e-01 3.29578482e-02 8.56056929e-01
9.90492940e-01 -1.29755545e+00 1.39936554e+00 6.45990670e-01
2.74785906e-01 -4.06187803e-01 7.94491529e-01 6.70868635e-01
-1.84129387e-01 -2.61974216e-01 -5.96653283e-01 -9.07002509e-01
-5.74329216e-03 6.73109770e-01 -6.41781449e-01 -8.69174004e-02
3.97889525e-01 2.72680134e-01 -7.16371000e-01 3.87469977e-01
-9.03407454e-01 8.47570360e-01 -2.46063739e-01 -4.30114001e-01
1.03219248e-01 2.16829479e-01 5.33976674e-01 1.18511772e+00
1.83399826e-01 5.11343360e-01 -4.05091107e-01 1.48876739e+00
1.33903965e-01 -2.33358636e-01 -3.15540254e-01 -3.33837420e-01
5.80715299e-01 7.76034534e-01 -7.28055000e-01 -6.43504798e-01
-5.93753569e-02 4.44405705e-01 4.09954518e-01 3.55142951e-01
-8.00827086e-01 -4.54452708e-02 1.74729228e-01 -4.56802137e-02
-9.53691304e-02 -2.48764697e-02 -6.10900283e-01 -1.27789676e+00
-2.80098289e-01 -1.17064250e+00 1.44913808e-01 -1.47663486e+00
-1.65259397e+00 4.27471459e-01 6.97850585e-02 -3.13224107e-01
-3.19861025e-01 -9.05432522e-01 -6.52118623e-01 1.06920195e+00
-6.31179631e-01 -5.51035702e-01 1.23483993e-01 2.35343367e-01
3.49260271e-01 5.55217592e-03 8.55844557e-01 -2.83384025e-01
-2.76927888e-01 3.95397782e-01 -3.47363740e-01 1.87795132e-01
5.78377783e-01 -1.24829853e+00 7.89408028e-01 5.29978573e-01
-5.37523292e-02 1.07250881e+00 1.11303437e+00 -5.98528683e-01
-7.94591188e-01 -4.38801944e-01 1.34871483e+00 -1.01943696e+00
9.90886211e-01 -2.20710933e-01 -8.58063459e-01 1.00973523e+00
-1.11271767e-03 -5.46812713e-01 4.69702482e-01 2.60317057e-01
-8.03909123e-01 4.56890315e-01 -1.20884085e+00 1.07107604e+00
1.11370373e+00 -6.75987244e-01 -1.69597208e+00 4.28918004e-01
6.23033226e-01 -4.31404352e-01 -6.77587032e-01 -1.57190591e-01
5.82034290e-01 -1.18726563e+00 8.47822428e-01 -7.04039156e-01
1.05140924e+00 1.46691322e-01 -3.49055707e-01 -1.42271316e+00
-7.52643049e-01 -6.01255357e-01 2.20279053e-01 9.83868778e-01
5.90013623e-01 -1.07308936e+00 2.51051337e-02 1.15935361e+00
3.19187558e-04 -4.16477710e-01 -7.67687917e-01 -6.19136810e-01
6.15956366e-01 -6.67686224e-01 3.55773777e-01 8.39441478e-01
3.14751297e-01 4.78392750e-01 2.07357183e-02 -1.80949032e-01
6.11335218e-01 -1.05221733e-01 3.24999511e-01 -1.29697239e+00
-5.47483325e-01 -7.83107162e-01 -2.10995644e-01 -6.72559977e-01
7.90566877e-02 -9.29949045e-01 -2.22513974e-02 -1.38500512e+00
2.45690256e-01 -3.75886708e-01 1.85474023e-01 5.96934736e-01
-2.64366865e-01 -2.45213524e-01 5.34195125e-01 3.54903758e-01
-2.81844616e-01 1.25981420e-01 1.29756105e+00 1.90263346e-01
1.69293746e-01 -4.04647827e-01 -1.18872869e+00 1.09153831e+00
8.26034427e-01 -6.39428556e-01 -1.89402893e-01 -3.46147358e-01
1.20234215e+00 1.77877188e-01 8.41311276e-01 -9.38652337e-01
5.93499243e-02 -3.52667928e-01 6.76384807e-01 -8.41219947e-02
-8.42589587e-02 -3.63283098e-01 8.93210173e-02 6.68462574e-01
-8.21981668e-01 2.14957163e-01 3.07524800e-01 3.42643112e-01
2.47685954e-01 -1.67908877e-01 4.68810111e-01 -5.52342653e-01
-5.35161197e-01 -8.18192661e-01 -7.22834826e-01 5.93154550e-01
5.58111012e-01 -3.52666751e-02 -1.06793785e+00 -2.75998294e-01
-8.49991977e-01 1.93575136e-02 2.57689118e-01 2.90028989e-01
3.98814976e-01 -1.00391257e+00 -8.82283449e-01 -4.91071492e-02
-1.07911728e-01 -3.16942394e-01 1.77269623e-01 9.99214947e-01
-6.25962317e-01 6.43422663e-01 -2.35140413e-01 -1.21243440e-01
-4.30982381e-01 5.25081992e-01 4.02222484e-01 -2.60331482e-01
-5.43254316e-01 1.08975732e+00 3.58073205e-01 -4.49406952e-01
-3.04981112e-01 -6.64655209e-01 2.83769011e-01 -1.66163951e-01
5.48078060e-01 3.63666475e-01 -3.92531455e-01 -2.05324590e-01
-3.00102174e-01 3.45169812e-01 -2.43983790e-01 -2.73541152e-01
9.47784364e-01 1.04049236e-01 -3.22697073e-01 7.82134175e-01
3.96353006e-01 -1.08943716e-01 -6.89750552e-01 -1.11870781e-01
1.38694257e-01 -2.99635798e-01 -3.23249847e-01 -1.61107624e+00
-1.62828833e-01 9.34696078e-01 -2.45924011e-01 2.40428448e-01
4.04732168e-01 1.74041286e-01 2.90010542e-01 7.11416066e-01
8.23217452e-01 -8.93824339e-01 6.84689134e-02 7.88481534e-01
1.28477788e+00 -7.66348362e-01 -5.86417466e-02 1.25545993e-01
-7.17365503e-01 1.02015889e+00 5.79605401e-01 -2.65981555e-01
6.63173525e-03 -3.33335772e-02 -1.70211673e-01 -3.78131866e-01
-1.38405764e+00 4.98434424e-01 -4.30054888e-02 3.39383900e-01
5.70087373e-01 4.67672318e-01 -4.90676761e-01 9.19518828e-01
-9.69879389e-01 2.37983227e-01 6.61804199e-01 8.70823085e-01
-4.33834881e-01 -5.49614608e-01 -7.21622229e-01 4.58561122e-01
-2.50545710e-01 -4.64208066e-01 -4.08834100e-01 8.94610763e-01
1.12799414e-01 9.13525701e-01 2.12163907e-02 -3.08706194e-01
2.51818061e-01 5.17174840e-01 6.76356673e-01 -5.58636725e-01
-7.96081543e-01 -3.10852617e-01 4.52330530e-01 -4.93736893e-01
-4.75960560e-02 -4.75658715e-01 -1.19020271e+00 -9.54325318e-01
-4.91818972e-02 3.09757348e-02 4.09698300e-02 1.15126073e+00
8.22422430e-02 4.44818109e-01 -2.86446959e-01 -5.84188819e-01
-1.05626822e+00 -1.14081168e+00 -5.93572319e-01 2.47651577e-01
-1.28015399e-01 -5.83415926e-01 -8.30948532e-01 -2.47182027e-01]
|
[9.742290496826172, 7.506002902984619]
|
7a8bc600-3b76-401e-b00b-a85a82f319c3
|
towards-clustering-friendly-representations
|
2106.09874
| null |
https://arxiv.org/abs/2106.09874v1
|
https://arxiv.org/pdf/2106.09874v1.pdf
|
Towards Clustering-friendly Representations: Subspace Clustering via Graph Filtering
|
Finding a suitable data representation for a specific task has been shown to be crucial in many applications. The success of subspace clustering depends on the assumption that the data can be separated into different subspaces. However, this simple assumption does not always hold since the raw data might not be separable into subspaces. To recover the ``clustering-friendly'' representation and facilitate the subsequent clustering, we propose a graph filtering approach by which a smooth representation is achieved. Specifically, it injects graph similarity into data features by applying a low-pass filter to extract useful data representations for clustering. Extensive experiments on image and document clustering datasets demonstrate that our method improves upon state-of-the-art subspace clustering techniques. Especially, its comparable performance with deep learning methods emphasizes the effectiveness of the simple graph filtering scheme for many real-world applications. An ablation study shows that graph filtering can remove noise, preserve structure in the image, and increase the separability of classes.
|
['Ling Tian', 'Guangchun Luo', 'Zhao Kang', 'Zhengrui Ma']
|
2021-06-18
| null | null | null | null |
['graph-similarity']
|
['graphs']
|
[-7.12807663e-03 -2.50250399e-01 -1.56057671e-01 -2.75388986e-01
-4.26633894e-01 -7.78698504e-01 4.69686061e-01 1.83540300e-01
-1.06404208e-01 4.62948419e-02 4.48940098e-01 -2.45393328e-02
-3.24498773e-01 -5.73833346e-01 -3.89678091e-01 -1.19081044e+00
-7.63191581e-02 2.71266490e-01 -2.90046372e-02 1.76010564e-01
1.01065353e-01 7.00968504e-01 -1.49758506e+00 4.48657453e-01
8.61805081e-01 6.89439595e-01 1.91853434e-01 1.23864628e-01
-2.76217580e-01 3.67644221e-01 -3.83730859e-01 6.11775592e-02
2.84359127e-01 -4.41055149e-01 -6.46130204e-01 5.40506661e-01
4.34602827e-01 1.55609146e-01 -6.13722563e-01 1.36995554e+00
2.11041003e-01 4.05340731e-01 7.84348071e-01 -1.16731870e+00
-5.89825034e-01 4.91039127e-01 -7.42651165e-01 -1.46680087e-01
4.17675860e-02 -1.53955817e-01 1.01721954e+00 -7.73272038e-01
6.22556448e-01 1.15595508e+00 6.56575859e-01 3.18422526e-01
-1.71749187e+00 -5.56991816e-01 2.21552983e-01 1.21537633e-01
-1.51506114e+00 -3.31547469e-01 1.11702454e+00 -5.06377876e-01
4.67893749e-01 3.91096979e-01 6.61769867e-01 8.06417584e-01
-7.26956204e-02 1.05712855e+00 9.40719664e-01 -2.06074744e-01
4.57229465e-01 -1.64162993e-01 5.13687074e-01 4.78823394e-01
3.52521002e-01 -3.32895637e-01 -3.21542829e-01 -1.53151840e-01
5.50029159e-01 3.34839165e-01 -5.73027670e-01 -1.14240026e+00
-1.21166372e+00 8.29063833e-01 6.95634305e-01 5.99717498e-01
-1.54683620e-01 -5.16945571e-02 3.92703474e-01 1.31189615e-01
3.47111046e-01 4.69899297e-01 8.66700709e-02 3.69186610e-01
-1.21795917e+00 -8.80440399e-02 5.48783362e-01 8.71758282e-01
7.80347109e-01 3.10067628e-02 -1.41957700e-01 7.44215250e-01
1.84684351e-01 2.39154458e-01 3.90134335e-01 -1.01858222e+00
1.22214131e-01 9.18202162e-01 -2.98552513e-01 -1.28209376e+00
-4.73976731e-01 -6.87882364e-01 -1.41182876e+00 5.80385402e-02
5.01991391e-01 2.10184753e-01 -9.49149907e-01 1.63764095e+00
1.72572464e-01 2.12687507e-01 -6.56895638e-02 8.99322450e-01
8.21658015e-01 4.05706167e-01 -1.27876088e-01 -3.06831211e-01
9.25496101e-01 -7.38389075e-01 -7.95976102e-01 -1.15347937e-01
4.88700151e-01 -5.36446393e-01 9.86790419e-01 4.28113520e-01
-6.48903668e-01 -5.76773345e-01 -9.55552161e-01 3.40028666e-02
-3.10618967e-01 2.27392539e-01 8.66496921e-01 5.50401151e-01
-9.69152153e-01 6.94463253e-01 -9.37245488e-01 -5.94401717e-01
5.58224678e-01 4.64812487e-01 -6.82872057e-01 -2.20834389e-01
-5.80193102e-01 1.97350785e-01 4.39183980e-01 -9.78562166e-04
-6.02958381e-01 -3.39218467e-01 -6.98058903e-01 2.33461708e-01
2.40169659e-01 -7.16972411e-01 4.37786847e-01 -9.13843095e-01
-9.87591028e-01 7.62180686e-01 -3.41396868e-01 -2.01013952e-01
2.08440766e-01 -1.07503161e-01 -2.87616223e-01 4.31653500e-01
-7.59363174e-04 3.18440825e-01 1.19393194e+00 -1.54047012e+00
-2.90695310e-01 -6.22408330e-01 -3.93537551e-01 2.38598347e-01
-7.56967783e-01 -1.54679477e-01 -8.44236135e-01 -8.19991589e-01
7.33601809e-01 -9.06163454e-01 -4.05377537e-01 -1.77050292e-01
-4.64882016e-01 -1.36611387e-01 1.16387141e+00 -6.30686104e-01
1.33474910e+00 -2.71656489e+00 4.75530863e-01 5.75864792e-01
7.03451455e-01 7.23906383e-02 -1.48883477e-01 5.50996721e-01
-3.95329833e-01 5.15302904e-02 -3.80787998e-01 -2.55525887e-01
-1.23180717e-01 2.12926622e-02 -3.97081792e-01 9.12272513e-01
-1.74081713e-01 7.28701472e-01 -9.15514886e-01 -3.78619075e-01
3.35068882e-01 5.96102595e-01 -5.23252249e-01 7.18651786e-02
2.35145822e-01 4.78753537e-01 -3.50088775e-01 3.07188153e-01
7.45235443e-01 -4.18850809e-01 4.06603187e-01 -5.38014174e-01
1.91804022e-01 -6.71291491e-03 -1.42930150e+00 1.90066540e+00
1.75410360e-01 6.95470214e-01 4.18526202e-01 -1.39845049e+00
7.61698484e-01 -1.30186127e-02 8.38346303e-01 -2.13326097e-01
1.00006171e-01 -4.99547906e-02 4.44920212e-02 -1.33305341e-01
1.18218914e-01 1.39326498e-01 9.80314091e-02 4.60006624e-01
1.83818850e-03 -1.16947256e-02 1.79631263e-01 4.70733970e-01
1.02760911e+00 -2.02422231e-01 6.52499646e-02 -7.41571724e-01
3.80460322e-01 1.66875161e-02 5.27988434e-01 6.25067174e-01
-1.90240264e-01 9.06487107e-01 3.05633336e-01 -2.82186776e-01
-7.16707230e-01 -1.17691028e+00 -7.83581212e-02 8.36219490e-01
3.63698512e-01 -6.27197146e-01 -1.01079893e+00 -8.48643422e-01
4.73386608e-02 4.96031880e-01 -6.41742587e-01 -5.59109926e-01
-4.39433068e-01 -8.43071282e-01 1.71272233e-01 5.85744917e-01
4.56874430e-01 -5.65538585e-01 -1.28489316e-01 -1.07331663e-01
-3.11161906e-01 -8.99727166e-01 -6.25717103e-01 2.50496864e-01
-1.05285418e+00 -1.07911408e+00 -4.16100055e-01 -1.02210486e+00
1.13195944e+00 1.11048460e+00 9.20824468e-01 2.06747651e-01
-8.92948955e-02 4.69373137e-01 -3.42243046e-01 1.56505704e-01
-1.67712748e-01 6.57649040e-02 1.79967150e-01 2.64227808e-01
6.96062088e-01 -7.20908284e-01 -6.21860623e-01 2.53713191e-01
-1.00861967e+00 -5.92144877e-02 4.23379868e-01 8.59273195e-01
6.76669836e-01 7.41076589e-01 2.54870504e-01 -7.72118211e-01
6.42305076e-01 -1.44298822e-01 -2.73657769e-01 1.88512638e-01
-4.28377241e-01 5.78331985e-02 9.63509917e-01 -3.14270735e-01
-5.59385300e-01 5.06772816e-01 4.07932609e-01 -7.47718692e-01
-3.62704784e-01 4.59388614e-01 -4.11680162e-01 -8.33771843e-03
6.09436035e-01 4.15439546e-01 2.63552010e-01 -6.67932808e-01
6.38414383e-01 5.98860085e-01 5.79088926e-01 -4.19498414e-01
1.16314936e+00 9.34402049e-01 1.26709817e-02 -1.14051032e+00
-6.56888485e-01 -9.20850039e-01 -1.12830353e+00 -1.09197974e-01
7.87343860e-01 -1.03033459e+00 -4.99547660e-01 1.66242197e-01
-7.11527228e-01 -7.65740871e-02 -2.80125797e-01 4.24835950e-01
-3.96504819e-01 8.15694034e-01 -3.62079710e-01 -5.55625439e-01
-1.04487605e-01 -9.98877883e-01 9.63027060e-01 7.17711821e-03
-2.86737502e-01 -1.03853881e+00 -2.30642587e-01 2.33003274e-01
7.17654882e-04 1.09284885e-01 1.04093528e+00 -5.63753426e-01
-4.99957234e-01 -2.53223091e-01 -2.13160887e-01 2.33382300e-01
4.32712942e-01 -6.44160062e-03 -9.95812953e-01 -8.23828399e-01
1.20062925e-01 5.06070033e-02 1.32602024e+00 4.75770712e-01
1.38707924e+00 -6.33107796e-02 -6.17068708e-01 9.25367594e-01
1.31864583e+00 -7.88834319e-03 4.66131032e-01 2.60629207e-02
1.18706286e+00 7.26568103e-01 1.47625655e-01 6.59755245e-02
9.86449942e-02 5.20369709e-01 2.34735340e-01 -3.46779466e-01
-2.70067334e-01 -6.70142099e-02 3.40834737e-01 1.04076111e+00
7.93996602e-02 1.12497643e-01 -1.00023532e+00 5.61239541e-01
-2.10795593e+00 -1.04243898e+00 -3.38245094e-01 2.39456606e+00
4.42589313e-01 -2.11625710e-01 3.01026404e-01 3.46525133e-01
8.62253726e-01 2.02675626e-01 -5.28303027e-01 2.09427357e-01
-3.31672579e-01 -1.67088479e-01 2.90134162e-01 2.32599989e-01
-1.50064492e+00 9.76589322e-01 6.45588064e+00 9.11238730e-01
-1.18594396e+00 -2.74830729e-01 4.52932000e-01 3.91208380e-03
-2.60248750e-01 1.45062979e-03 -3.57738107e-01 2.82857835e-01
3.87879938e-01 -2.98129022e-01 7.04493284e-01 8.41709554e-01
1.49558231e-01 2.41319343e-01 -1.21348679e+00 1.18513060e+00
9.58660617e-02 -1.30274785e+00 3.05927843e-01 1.51388779e-01
6.89476788e-01 -1.03829928e-01 1.04474761e-01 -3.60993408e-02
3.60280335e-01 -1.06949520e+00 4.69607770e-01 2.70121455e-01
5.14915168e-01 -9.33930695e-01 3.50003839e-01 2.92014301e-01
-1.30857968e+00 -1.74944833e-01 -6.73160076e-01 1.16795458e-01
-2.87833691e-01 9.57328498e-01 -5.16517103e-01 6.35143816e-01
8.20532680e-01 1.09213030e+00 -8.08713675e-01 1.09070373e+00
2.97937775e-03 6.32145107e-01 -3.01023304e-01 4.01933163e-01
1.67115211e-01 -7.93497145e-01 5.58108568e-01 1.26442742e+00
1.74983963e-01 1.65615659e-02 4.12338108e-01 9.42685425e-01
-7.03403875e-02 2.69065291e-01 -8.80813420e-01 -2.37397000e-01
3.11060011e-01 1.37089324e+00 -1.27855229e+00 -1.49922535e-01
-4.31661963e-01 1.02859020e+00 3.54254246e-01 7.12085247e-01
-4.22055900e-01 -3.95899117e-01 6.33755445e-01 1.10867143e-01
4.16510522e-01 -6.77830398e-01 -6.79493129e-01 -1.36106670e+00
-5.15570231e-02 -1.01515496e+00 6.13440812e-01 -3.22192192e-01
-1.38908350e+00 4.27566558e-01 -2.45344549e-01 -1.45815670e+00
1.02464564e-01 -4.72779810e-01 -5.82052410e-01 5.63893795e-01
-9.48437274e-01 -1.19857049e+00 -4.63414133e-01 8.87445092e-01
3.02708715e-01 -2.25516215e-01 6.88208461e-01 1.11466102e-01
-7.71851718e-01 3.38232309e-01 6.52935803e-01 3.66483688e-01
7.22973704e-01 -1.41424465e+00 1.61982656e-01 1.19665205e+00
6.03324413e-01 1.12840891e+00 5.67842305e-01 -6.67596757e-01
-1.70628762e+00 -1.26112545e+00 3.34038168e-01 -4.52134013e-01
6.11477494e-01 -6.59878314e-01 -1.23889625e+00 5.56113541e-01
2.62726903e-01 -3.12923454e-02 9.33698654e-01 3.20144027e-01
-6.04767382e-01 -2.36698076e-01 -7.82615125e-01 7.32114971e-01
1.13969195e+00 -6.50437176e-01 -5.65142751e-01 4.53428417e-01
4.39958453e-01 1.36633947e-01 -6.95776105e-01 2.87366033e-01
2.84492493e-01 -1.02728975e+00 1.17183697e+00 -7.46597707e-01
1.39664263e-01 -5.81162274e-01 -2.43834153e-01 -1.40224242e+00
-9.91381884e-01 -6.38316989e-01 -2.27748930e-01 1.22601283e+00
8.08950700e-03 -3.09926808e-01 8.84052396e-01 5.63905716e-01
1.62494499e-02 -3.75727147e-01 -6.39338851e-01 -9.95314121e-01
1.09715849e-01 -2.43861899e-01 5.06225467e-01 1.40865612e+00
2.48234376e-01 4.53828037e-01 -3.07498500e-02 3.23679060e-01
9.40534592e-01 6.69898331e-01 8.69707704e-01 -1.51416385e+00
-1.26266882e-01 -7.30321407e-01 -4.86147821e-01 -1.01445282e+00
4.32434440e-01 -1.21048880e+00 -1.27479360e-01 -1.74015701e+00
4.09249514e-01 -1.97729334e-01 -4.35672790e-01 3.64212602e-01
-3.44323218e-01 2.12850675e-01 2.37033501e-01 6.11681938e-01
-6.90256953e-01 6.37629867e-01 1.08793819e+00 -3.71587515e-01
-3.32051516e-01 -1.31078258e-01 -9.48812425e-01 6.73774779e-01
6.77999914e-01 -1.89763650e-01 -6.07108355e-01 -3.53272408e-01
-8.99820179e-02 -5.60143054e-01 1.79127917e-01 -1.08041060e+00
3.50059509e-01 -7.84588382e-02 4.90716189e-01 -5.96406639e-01
8.01161155e-02 -1.02867711e+00 2.42514819e-01 3.33978117e-01
-9.57417637e-02 -3.49902332e-01 1.27629519e-01 7.55845606e-01
-4.02054638e-01 1.94637954e-01 9.46457326e-01 2.58819852e-03
-7.37059116e-01 2.27884546e-01 -2.25617617e-01 -1.44763201e-01
1.04037130e+00 -5.20728767e-01 -1.16336852e-01 -4.68070596e-01
-7.56243587e-01 1.90240458e-01 8.26985478e-01 3.59238148e-01
5.87246358e-01 -1.45239782e+00 -4.96664226e-01 4.13093567e-01
6.39995635e-02 5.88486344e-02 6.46575242e-02 8.97235513e-01
-2.38496527e-01 2.00694337e-01 -2.89242100e-02 -9.35396194e-01
-1.31771898e+00 1.09855068e+00 6.26815259e-02 7.67399296e-02
-1.00071716e+00 6.75473034e-01 7.34580576e-01 -1.65989518e-01
3.50160092e-01 -1.51974261e-01 -2.87209868e-01 2.33704850e-01
4.34352577e-01 3.31197560e-01 3.51781137e-02 -7.77348995e-01
-3.95898908e-01 7.20589042e-01 -6.12055510e-02 3.27475518e-01
1.53324032e+00 -2.60740519e-01 -4.38002527e-01 3.14280033e-01
1.23582935e+00 9.04448256e-02 -1.14712977e+00 -1.47278011e-01
8.44709352e-02 -5.65664947e-01 2.55366445e-01 -1.73473954e-01
-1.38167834e+00 1.00575840e+00 3.32437873e-01 5.84945679e-01
1.34134185e+00 5.15178554e-02 5.51734328e-01 4.52441543e-01
2.12019250e-01 -1.10659528e+00 7.81244272e-03 2.07583666e-01
8.19280148e-01 -1.08413625e+00 1.67508468e-01 -7.63396680e-01
-4.25662607e-01 1.17659867e+00 2.98115075e-01 -2.27357000e-01
7.79714227e-01 7.33496761e-03 1.44127533e-01 -4.44749773e-01
-1.64185494e-01 -2.94070929e-01 6.75206661e-01 7.54592419e-01
4.00279254e-01 1.12012792e-02 -4.28382680e-02 4.71848160e-01
-1.92432418e-01 -5.69977582e-01 3.41118366e-01 4.64143276e-01
-3.55050445e-01 -8.79951298e-01 -5.25427520e-01 5.23785889e-01
-1.46294117e-01 8.95387307e-02 -8.69036317e-01 5.99393249e-01
-2.18897909e-01 1.04761350e+00 8.58193263e-03 -4.73387808e-01
1.03583418e-01 -2.45758444e-02 2.44954258e-01 -7.06096530e-01
-3.67370009e-01 6.18154466e-01 -3.80098909e-01 -8.23137224e-01
-5.79094768e-01 -7.58952618e-01 -1.30262399e+00 -2.06152856e-01
-2.06020936e-01 2.88401872e-01 2.65448779e-01 7.84441888e-01
6.35447562e-01 4.80884880e-01 6.71442866e-01 -7.74455130e-01
-2.72067517e-01 -4.29153353e-01 -8.50216866e-01 1.01401913e+00
2.65844405e-01 -6.41009331e-01 -4.84964013e-01 2.93536276e-01]
|
[7.968088626861572, 4.10377311706543]
|
ad23964e-263c-4fff-941e-26e3df0821e1
|
probabilistic-robust-linear-quadratic
|
2105.07668
| null |
https://arxiv.org/abs/2105.07668v2
|
https://arxiv.org/pdf/2105.07668v2.pdf
|
Probabilistic Robust Linear Quadratic Regulators with Gaussian Processes
|
Probabilistic models such as Gaussian processes (GPs) are powerful tools to learn unknown dynamical systems from data for subsequent use in control design. While learning-based control has the potential to yield superior performance in demanding applications, robustness to uncertainty remains an important challenge. Since Bayesian methods quantify uncertainty of the learning results, it is natural to incorporate these uncertainties into a robust design. In contrast to most state-of-the-art approaches that consider worst-case estimates, we leverage the learning method's posterior distribution in the controller synthesis. The result is a more informed and, thus, more efficient trade-off between performance and robustness. We present a novel controller synthesis for linearized GP dynamics that yields robust controllers with respect to a probabilistic stability margin. The formulation is based on a recently proposed algorithm for linear quadratic control synthesis, which we extend by giving probabilistic robustness guarantees in the form of credibility bounds for the system's stability.Comparisons to existing methods based on worst-case and certainty-equivalence designs reveal superior performance and robustness properties of the proposed method.
|
['Sebastian Trimpe', 'Matthias Neumann-Brosig', 'Alexander von Rohr']
|
2021-05-17
| null | null | null | null |
['robust-design']
|
['miscellaneous']
|
[ 1.00433946e-01 2.28903458e-01 -1.96783125e-01 1.48794994e-01
-1.10981536e+00 -7.58109987e-01 5.95969379e-01 3.10854226e-01
-1.03222296e-01 1.08197260e+00 -2.08972827e-01 -4.32736993e-01
-8.57847691e-01 -6.96281195e-01 -8.06994140e-01 -1.06043017e+00
-3.38481106e-02 2.24521637e-01 2.19635502e-01 6.34889528e-02
2.01000869e-01 6.08456671e-01 -1.30549371e+00 -6.72728598e-01
9.54970539e-01 1.06285095e+00 -1.52906865e-01 5.04160404e-01
5.97952187e-01 3.03412914e-01 -4.35598969e-01 4.17460389e-02
2.58604318e-01 -1.14142001e-01 -3.47188525e-02 -1.69582918e-01
-1.75713733e-01 -1.16409391e-01 -6.86514601e-02 1.26589477e+00
6.74108624e-01 4.37303305e-01 9.34004068e-01 -1.31428766e+00
-1.86856225e-01 3.47710192e-01 -3.14893872e-01 -3.31113666e-01
2.03440681e-01 1.53241515e-01 6.98210478e-01 -5.83382905e-01
1.31376669e-01 1.43429375e+00 6.45516217e-01 2.70603687e-01
-1.43395388e+00 -6.34258509e-01 1.61304906e-01 -3.76448274e-01
-1.55688286e+00 -4.14859086e-01 4.47081387e-01 -8.36181045e-01
2.29639411e-01 4.76613268e-02 1.27269387e-01 8.44056487e-01
6.78717017e-01 6.06963813e-01 1.24561489e+00 -3.49594384e-01
6.14760935e-01 1.25713721e-01 -2.46581852e-01 4.89711434e-01
8.12293768e-01 6.81492746e-01 -1.22899376e-01 -3.27712387e-01
8.49828959e-01 -3.08240533e-01 -4.30497676e-01 -7.70639122e-01
-1.08349323e+00 9.17364657e-01 9.11399052e-02 -2.29705170e-01
-4.85815912e-01 5.11421442e-01 1.81908071e-01 -1.83800429e-01
4.32512850e-01 4.87856627e-01 -3.14474702e-01 -7.15015307e-02
-7.40548730e-01 7.07140923e-01 1.00540495e+00 9.09116447e-01
1.55156165e-01 4.08376575e-01 -4.76068258e-01 2.15728492e-01
8.54335368e-01 8.70748162e-01 -1.30851224e-01 -1.04700744e+00
2.57557929e-01 3.18920501e-02 8.03383172e-01 -9.95143652e-01
-6.36002794e-02 -5.16867697e-01 -6.96910739e-01 7.33477414e-01
5.00181913e-01 -5.55247664e-01 -8.94155443e-01 1.88400352e+00
4.12028611e-01 2.12837085e-01 1.66960239e-01 5.98650932e-01
-2.82897323e-01 8.24979246e-01 -1.82591572e-01 -6.23081446e-01
9.29786623e-01 -1.86221108e-01 -9.85349119e-01 1.65251300e-01
-1.02169208e-01 -6.31432354e-01 4.97184068e-01 5.45930266e-01
-9.60420310e-01 -1.71264917e-01 -1.22351336e+00 6.90834999e-01
-4.56298590e-02 4.61025015e-02 -7.68117234e-03 9.67021406e-01
-7.29695797e-01 7.87327349e-01 -1.04532468e+00 -6.64077401e-02
-4.33819629e-02 4.36930150e-01 1.44043535e-01 3.12268585e-01
-1.25147879e+00 1.34096122e+00 5.88961482e-01 4.16572720e-01
-1.08155942e+00 -8.41169357e-01 -9.48429227e-01 -4.70187142e-02
9.49430168e-01 -6.19387984e-01 1.45531428e+00 -1.77850887e-01
-2.23833251e+00 -2.00414345e-01 1.17870666e-01 -5.29160500e-01
7.28652298e-01 -4.32037026e-01 -1.94545686e-02 2.78914999e-02
-2.22922429e-01 1.37921974e-01 1.36420000e+00 -1.27566850e+00
-5.52283645e-01 2.70411223e-02 -9.13249999e-02 -8.72221291e-02
2.78768927e-01 -7.33008906e-02 -7.70845115e-02 -7.06043661e-01
-8.16381574e-02 -1.25929248e+00 -6.02513194e-01 9.85493809e-02
-5.25376022e-01 -2.15425149e-01 6.63202941e-01 -4.50927794e-01
1.26146603e+00 -1.71654320e+00 3.15517277e-01 5.22485435e-01
-1.46645814e-01 3.03051710e-01 6.17384493e-01 6.06990039e-01
1.69117913e-01 6.39808476e-02 -3.60278755e-01 -4.07420471e-02
5.71938515e-01 -2.69727260e-02 -6.44080400e-01 9.23377454e-01
6.19037569e-01 2.66707718e-01 -8.42443943e-01 -1.25853673e-01
5.37022769e-01 4.26557094e-01 -2.99000829e-01 1.22713208e-01
-3.46285015e-01 3.95555109e-01 -7.94368684e-01 2.92358518e-01
2.87508398e-01 2.18937069e-01 -5.26025221e-02 -7.99900964e-02
-3.42070311e-01 -3.00329089e-01 -1.59917724e+00 9.73195374e-01
-4.26068783e-01 2.28489697e-01 3.96122575e-01 -7.20073462e-01
9.04965580e-01 5.66345632e-01 2.25174710e-01 3.71707737e-01
4.76760834e-01 2.24602968e-01 -1.25861630e-01 -6.13587052e-02
2.93375611e-01 -3.86419684e-01 -3.59226257e-01 1.55048355e-01
-1.47021860e-01 -9.35794950e-01 3.95930298e-02 -1.27311483e-01
6.70483232e-01 3.97638887e-01 5.43455303e-01 -7.84512818e-01
5.46386182e-01 -2.86962092e-01 7.86034822e-01 4.83508438e-01
-3.16337079e-01 1.56461403e-01 6.18203759e-01 3.80537838e-01
-9.12593305e-01 -1.12333548e+00 -1.23076826e-01 2.11317733e-01
-1.25582129e-01 -8.85177031e-02 -5.88339746e-01 3.57584283e-02
2.86962509e-01 9.97535408e-01 -5.64982057e-01 -3.94921511e-01
-1.86003923e-01 -3.94854307e-01 2.30444804e-01 5.01571596e-01
1.98117900e-03 -1.59444585e-01 -5.08319497e-01 4.90687549e-01
5.47988951e-01 -7.51587629e-01 -3.18798393e-01 1.23566642e-01
-7.11736381e-01 -1.04884923e+00 -8.45060229e-01 -8.32894742e-02
4.91589129e-01 -3.95056367e-01 4.53573704e-01 -7.33187556e-01
-5.16477711e-02 6.19823992e-01 1.99803054e-01 -9.13262486e-01
-5.53818166e-01 -3.42801571e-01 6.89371824e-01 -4.40286286e-02
-6.07500494e-01 -2.03271374e-01 -2.65597939e-01 3.69827718e-01
-7.48045146e-01 -4.18655545e-01 4.78976071e-01 8.78822088e-01
7.53400505e-01 7.90439546e-01 7.78014123e-01 -5.64888239e-01
9.03270900e-01 -4.85628039e-01 -1.65758240e+00 1.70961648e-01
-9.20076013e-01 3.60869199e-01 6.54550910e-01 -4.77405280e-01
-1.22999907e+00 3.72053862e-01 3.75211924e-01 -7.09150553e-01
1.82006091e-01 7.08112657e-01 -3.32208544e-01 -3.76650915e-02
5.48368692e-01 -2.94202954e-01 3.90962332e-01 -1.13143250e-01
5.86899936e-01 3.52902979e-01 5.72898984e-01 -1.01657057e+00
1.08430135e+00 8.49783942e-02 5.59160233e-01 -6.91259921e-01
-6.79646254e-01 -2.60213494e-01 -3.75047892e-01 -2.85613537e-01
6.25173211e-01 -8.00806999e-01 -1.07816839e+00 3.42959017e-01
-8.54924977e-01 -8.52471143e-02 -2.01564118e-01 6.91874504e-01
-1.05742598e+00 5.28697707e-02 -3.19782436e-01 -1.76763272e+00
-3.05639971e-02 -1.34660983e+00 8.98095131e-01 3.28516573e-01
-2.13697448e-01 -1.18192327e+00 1.24953605e-01 -2.69749075e-01
3.06844592e-01 9.16339993e-01 7.61863291e-01 -3.27437639e-01
-5.14488459e-01 -5.09174466e-01 9.87968817e-02 4.81338859e-01
6.86341822e-02 3.92689884e-01 -7.45401502e-01 -4.85392243e-01
3.16238046e-01 6.60091639e-02 4.78755802e-01 7.60107875e-01
3.87386292e-01 -4.59010899e-01 -4.68601823e-01 -1.34544641e-01
1.61237347e+00 3.22089881e-01 2.78992444e-01 2.98030078e-02
4.35207188e-01 8.68971884e-01 9.10825312e-01 6.79168105e-01
-1.05599523e-01 5.29641211e-01 4.23757643e-01 5.49930155e-01
5.17014503e-01 -2.63509512e-01 5.44985592e-01 4.62577045e-01
1.91428870e-01 -2.17459545e-01 -8.73825669e-01 4.42115933e-01
-2.13689470e+00 -7.99980164e-01 3.30308974e-02 2.80513906e+00
8.89152169e-01 3.58770251e-01 6.46777451e-02 3.15905690e-01
1.14698970e+00 -1.96528405e-01 -6.21544659e-01 -2.22708732e-01
3.16554636e-01 -8.16594437e-02 9.95743454e-01 6.07745886e-01
-1.13830638e+00 3.89874309e-01 6.71731377e+00 1.06057954e+00
-7.84741998e-01 -2.44332865e-01 3.37761879e-01 1.65660828e-01
-4.42659408e-02 9.26520154e-02 -1.06275010e+00 4.24082726e-01
1.22539699e+00 -8.51359248e-01 2.62373567e-01 8.65386724e-01
8.63465071e-01 -3.23432863e-01 -9.83488798e-01 5.64532340e-01
-3.90596956e-01 -9.76086259e-01 -3.60232532e-01 1.28046185e-01
1.08554280e+00 -6.63662136e-01 4.59296286e-01 2.68693060e-01
6.43129230e-01 -9.91635323e-01 9.71898675e-01 9.79874849e-01
5.59872866e-01 -1.18080103e+00 8.29494953e-01 3.16777647e-01
-1.03524470e+00 -3.28131109e-01 -1.08636186e-01 6.89779446e-02
5.29030263e-01 9.50553834e-01 -7.76251495e-01 7.45832086e-01
1.47524923e-01 4.56963092e-01 4.22478504e-02 1.41458786e+00
-4.45207566e-01 5.82120359e-01 -5.43291509e-01 -1.92263037e-01
1.07023939e-01 -2.48150289e-01 1.01020849e+00 7.34577417e-01
5.61131775e-01 1.01632960e-02 3.68070185e-01 1.04628277e+00
3.52572083e-01 -3.01043689e-01 -7.00373828e-01 -3.47038805e-01
7.08337426e-01 7.32515872e-01 -5.03320396e-01 -1.24479443e-01
6.55732006e-02 1.07756101e-01 -2.98768193e-01 3.36612523e-01
-7.82250464e-01 -5.56288064e-01 5.90044141e-01 -1.96348861e-01
3.35040420e-01 -5.24971783e-01 -2.28047252e-01 -4.51752931e-01
-1.25073507e-01 -7.79368639e-01 1.24223493e-01 -5.03905833e-01
-1.32777393e+00 1.35278314e-01 5.57546377e-01 -1.46461296e+00
-7.34037697e-01 -7.03259230e-01 -4.72061038e-01 1.08133698e+00
-1.17539775e+00 -8.84791791e-01 5.92963934e-01 1.97009429e-01
3.08379829e-01 5.10459840e-02 5.51329136e-01 -2.74214149e-01
-7.61761904e-01 1.65015593e-01 6.96589887e-01 -5.29533982e-01
8.02506864e-01 -1.33883309e+00 -2.45758697e-01 1.13366938e+00
-6.69430315e-01 7.49132276e-01 1.45263553e+00 -8.55352759e-01
-1.59066021e+00 -1.23090780e+00 1.41678661e-01 -4.35121119e-01
1.26103008e+00 8.82500634e-02 -6.87322736e-01 3.31724107e-01
-1.65127456e-01 -1.84551790e-01 9.95042324e-02 -2.47595251e-01
6.22779131e-02 -4.73113097e-02 -1.29304397e+00 7.03855872e-01
6.46885037e-02 -3.86898667e-01 -6.98233485e-01 8.92175175e-03
8.87414098e-01 -4.82741654e-01 -1.17953706e+00 6.36121631e-01
5.04350483e-01 -1.49811611e-01 8.28473032e-01 -3.32114279e-01
-1.76903710e-01 -9.47659492e-01 -1.83313638e-01 -1.67939341e+00
-1.07345775e-01 -1.29279494e+00 -3.24771196e-01 1.30123961e+00
4.73714054e-01 -6.81844890e-01 2.72185922e-01 8.84164989e-01
-2.05447853e-01 -6.36988342e-01 -1.01803577e+00 -1.30487502e+00
3.30042809e-01 -4.64493662e-01 7.96582326e-02 3.19143087e-01
3.89993042e-02 -6.06292672e-02 -3.54305476e-01 8.24858963e-01
1.04317224e+00 -2.08731547e-01 4.11305249e-01 -1.32520747e+00
-1.32318556e-01 -5.34962237e-01 -3.13493162e-01 -4.28057164e-01
3.38844448e-01 -8.68572593e-02 8.86356413e-01 -1.27881384e+00
-4.07419264e-01 -3.07714403e-01 -2.66933471e-01 -2.37448923e-02
-2.55681723e-01 -2.80369312e-01 4.86560464e-02 -1.38180941e-01
-2.83148229e-01 8.93131793e-01 9.20446217e-01 7.72734880e-02
-2.92447150e-01 6.12506568e-01 -3.42333198e-01 6.75216973e-01
7.67233193e-01 -3.66491050e-01 -7.20424712e-01 2.94258565e-01
2.30356559e-01 3.04839730e-01 1.94421381e-01 -1.14718902e+00
4.24901009e-01 -3.94110918e-01 -3.32054347e-02 -6.35998845e-01
1.72677398e-01 -9.15756941e-01 4.63616908e-01 5.93253493e-01
-2.88855523e-01 -1.83149576e-01 4.24912006e-01 1.29516315e+00
-7.12328404e-02 -3.57906640e-01 1.00066030e+00 3.30896020e-01
-1.03021696e-01 1.74700961e-01 -8.63586128e-01 -1.94826767e-01
1.14485443e+00 1.57177076e-01 -2.24515460e-02 -6.13974452e-01
-6.76943660e-01 3.46459627e-01 2.03291714e-01 1.96004048e-01
3.23752314e-01 -1.17682934e+00 -4.48825806e-01 -3.98564935e-01
-1.92825869e-01 4.55916375e-02 -3.08228447e-03 6.50654554e-01
-5.33767305e-02 6.69103265e-01 1.54269829e-01 -6.77340925e-01
-6.01983309e-01 6.68536961e-01 2.46559441e-01 -6.13278598e-02
4.55359519e-02 4.80526358e-01 -4.00339849e-02 -8.53014067e-02
2.02008024e-01 -5.67118227e-01 1.00302704e-01 -7.01077357e-02
5.14597833e-01 7.37883508e-01 -5.23235910e-02 -3.87149870e-01
-1.38114125e-01 4.84457165e-01 4.01302963e-01 -5.65125823e-01
8.82466078e-01 -2.75330156e-01 1.16206676e-01 7.45524108e-01
5.91421843e-01 -6.48058355e-02 -1.87031424e+00 -6.24854490e-02
3.82734179e-01 -2.78227895e-01 4.08891976e-01 -7.10891724e-01
-6.38214767e-01 5.64588666e-01 6.46283805e-01 1.07483134e-01
6.73309565e-01 -6.00146770e-01 1.78706944e-01 3.52154464e-01
5.34541786e-01 -1.16501367e+00 -2.26666465e-01 5.62273383e-01
9.65359867e-01 -9.15895760e-01 2.27329001e-01 -4.76228178e-01
-3.99625719e-01 1.03061891e+00 2.18760535e-01 -4.45469588e-01
8.70018244e-01 5.27440846e-01 -3.67429972e-01 4.05089438e-01
-7.73570120e-01 -6.16414733e-02 5.10905623e-01 6.65207624e-01
6.38792515e-02 1.47952914e-01 -2.20163733e-01 6.64847434e-01
2.28930265e-01 -5.59983663e-02 5.14734924e-01 1.11646318e+00
-5.62184095e-01 -9.01535153e-01 -8.67727637e-01 1.98306456e-01
-4.75915462e-01 2.51882643e-01 3.20316106e-01 7.86740243e-01
-4.62126821e-01 1.19299126e+00 -4.04098064e-01 4.28010412e-02
5.91389298e-01 1.11570917e-02 3.93564850e-01 -6.47463500e-01
-6.81955814e-02 3.15794289e-01 2.22390637e-01 -4.59371299e-01
-1.97911695e-01 -7.54721522e-01 -1.02142525e+00 -1.82969108e-01
-8.38245869e-01 2.95863479e-01 7.79616892e-01 9.77130353e-01
2.51551807e-01 6.11997187e-01 5.60738623e-01 -1.05949581e+00
-1.28887701e+00 -7.34005570e-01 -7.85126984e-01 -4.96118605e-01
2.83762395e-01 -1.27670419e+00 -5.51252127e-01 2.51833238e-02]
|
[5.076535224914551, 2.4790050983428955]
|
bcb4a796-130e-46c3-92d5-a2b4994a63c5
|
sun-exploring-intrinsic-uncertainties-in-text
|
2209.06442
| null |
https://arxiv.org/abs/2209.06442v2
|
https://arxiv.org/pdf/2209.06442v2.pdf
|
SUN: Exploring Intrinsic Uncertainties in Text-to-SQL Parsers
|
This paper aims to improve the performance of text-to-SQL parsing by exploring the intrinsic uncertainties in the neural network based approaches (called SUN). From the data uncertainty perspective, it is indisputable that a single SQL can be learned from multiple semantically-equivalent questions.Different from previous methods that are limited to one-to-one mapping, we propose a data uncertainty constraint to explore the underlying complementary semantic information among multiple semantically-equivalent questions (many-to-one) and learn the robust feature representations with reduced spurious associations. In this way, we can reduce the sensitivity of the learned representations and improve the robustness of the parser. From the model uncertainty perspective, there is often structural information (dependence) among the weights of neural networks. To improve the generalizability and stability of neural text-to-SQL parsers, we propose a model uncertainty constraint to refine the query representations by enforcing the output representations of different perturbed encoding networks to be consistent with each other. Extensive experiments on five benchmark datasets demonstrate that our method significantly outperforms strong competitors and achieves new state-of-the-art results. For reproducibility, we release our code and data at https://github.com/AlibabaResearch/DAMO-ConvAI/tree/main/sunsql.
|
['Yongbin Li', 'Min Yang', 'Luo Si', 'Fei Huang', 'Binhua Li', 'Xiangpeng Wei', 'Bowen Li', 'Binyuan Hui', 'Lihan Wang', 'Bowen Qin']
|
2022-09-14
| null |
https://aclanthology.org/2022.coling-1.471
|
https://aclanthology.org/2022.coling-1.471.pdf
|
coling-2022-10
|
['text-to-sql']
|
['computer-code']
|
[ 1.27830684e-01 4.58907962e-01 -4.11451273e-02 -9.40153658e-01
-1.22663569e+00 -8.33610833e-01 1.00049399e-01 2.15390697e-01
-2.10979149e-01 4.47747946e-01 2.00491980e-01 -3.37561309e-01
-3.61199319e-01 -9.92384970e-01 -1.28631222e+00 -3.66465628e-01
4.09748346e-01 5.47168076e-01 2.22125039e-01 -7.93541074e-02
6.56642765e-02 1.16865426e-01 -1.46919906e+00 5.50398886e-01
9.76150393e-01 1.22597623e+00 2.02725321e-01 3.26371491e-01
-5.19587636e-01 7.20278740e-01 -4.92676079e-01 -7.59397686e-01
1.16744436e-01 2.19547153e-02 -9.17731285e-01 -6.09811723e-01
4.98777717e-01 -1.13878958e-01 -3.88070554e-01 1.49655890e+00
3.61099958e-01 1.13492489e-01 2.26948023e-01 -1.09447563e+00
-1.15471947e+00 1.23233902e+00 -1.72738656e-01 7.38572329e-02
1.26659364e-01 -1.20944023e-01 1.40227425e+00 -9.02963638e-01
3.62917453e-01 1.69895327e+00 4.56161886e-01 5.27989149e-01
-1.15579355e+00 -7.72937775e-01 4.78160590e-01 2.49395609e-01
-1.27327633e+00 -3.92285585e-01 8.18629980e-01 -9.07323882e-03
9.13218737e-01 3.45961362e-01 -2.54452586e-01 1.33076882e+00
1.13566972e-01 9.51169848e-01 6.89013064e-01 -2.59847790e-01
2.03605279e-01 1.07209072e-01 3.77628207e-01 7.92056978e-01
2.01305315e-01 2.11838439e-01 -6.22833908e-01 -2.01119974e-01
3.16783518e-01 -2.96940863e-01 -1.09075725e-01 -3.86666983e-01
-8.74743462e-01 1.03948617e+00 4.17823166e-01 1.59861565e-01
5.43425269e-02 2.79352874e-01 4.21301663e-01 2.71019548e-01
2.40172893e-01 7.09927857e-01 -8.92008960e-01 -9.91911292e-02
-4.15824801e-01 2.15185061e-01 6.38940454e-01 1.17613852e+00
7.36557186e-01 -1.02565259e-01 -3.08147937e-01 7.69438446e-01
3.95113260e-01 2.07396090e-01 5.42920589e-01 -1.08641911e+00
8.40520322e-01 6.69886589e-01 -2.22199112e-01 -9.04162765e-01
-3.47348213e-01 -3.85709107e-01 -6.20897472e-01 -2.94654459e-01
3.87909472e-01 -8.10672268e-02 -8.04752827e-01 2.13327432e+00
2.63837606e-01 6.00959966e-03 2.48559743e-01 6.00248396e-01
1.10061717e+00 5.69900155e-01 4.54820469e-02 2.32653022e-01
1.34470391e+00 -8.13006461e-01 -6.39462948e-01 -6.14091754e-01
6.27606392e-01 -4.59233880e-01 1.43415594e+00 1.76466018e-01
-8.93759906e-01 -5.40586352e-01 -9.17162359e-01 -3.34460020e-01
-5.33665776e-01 -1.04237035e-01 5.02973974e-01 5.16120434e-01
-4.89537507e-01 6.86064601e-01 -8.09660137e-01 -7.21145421e-02
2.59660542e-01 2.81511426e-01 -3.56395513e-01 -1.75859556e-01
-1.67729616e+00 7.54645586e-01 8.58718753e-01 1.00497887e-01
-4.82454240e-01 -8.89763415e-01 -1.00272715e+00 3.08452845e-01
7.33575940e-01 -5.50331891e-01 1.27475882e+00 -4.72063154e-01
-1.24769604e+00 4.76963013e-01 -2.47242421e-01 -2.81084239e-01
2.84067601e-01 -3.67566168e-01 -2.58939892e-01 -1.66589141e-01
2.83967592e-02 7.09366620e-01 4.96204376e-01 -1.26677537e+00
-4.36094075e-01 -4.72125202e-01 1.70597851e-01 -2.20019501e-02
-2.36800328e-01 1.04564473e-01 -5.49577534e-01 -5.17413557e-01
5.25550127e-01 -6.56612456e-01 6.39893161e-03 -1.16968602e-01
-7.38734543e-01 -4.51602161e-01 3.93956542e-01 -5.31961977e-01
1.12383962e+00 -2.18248296e+00 2.43519396e-01 -1.93456709e-02
6.49740323e-02 -5.36024943e-02 -4.49852467e-01 2.15476856e-01
-1.78398237e-01 4.12229568e-01 -3.83088976e-01 -2.64213324e-01
3.37062091e-01 5.45288205e-01 -6.52985215e-01 -6.44545406e-02
6.13117218e-01 9.64276314e-01 -7.99122632e-01 -4.50383693e-01
-7.76846632e-02 2.91926265e-01 -6.07938826e-01 4.59955037e-01
-6.04153454e-01 1.62777677e-01 -6.45888507e-01 7.67218232e-01
8.34886014e-01 -1.33876413e-01 1.19286940e-01 -3.89466107e-01
4.13335860e-01 6.98217094e-01 -1.26568258e+00 1.83266962e+00
-3.25362831e-01 2.93230359e-02 -2.24388078e-01 -1.07335675e+00
1.14929736e+00 -3.64758074e-02 6.72184676e-02 -6.82779908e-01
-3.64876539e-02 3.25436071e-02 -9.80109125e-02 -5.18228471e-01
4.23583180e-01 2.09119961e-01 -4.64823216e-01 1.79206550e-01
3.03021789e-01 -1.32071078e-01 2.06698179e-02 1.81289345e-01
9.67939258e-01 2.98189700e-01 -1.76681131e-02 -2.48930246e-01
1.55691147e-01 -3.33976746e-01 9.56802905e-01 9.38076913e-01
-1.10895790e-01 7.42513418e-01 9.44720864e-01 -2.36478969e-01
-6.48378372e-01 -1.45981777e+00 -3.76150250e-01 1.38882399e+00
7.46492669e-03 -4.25124705e-01 -7.12519288e-01 -9.01996017e-01
2.18586922e-01 1.15816855e+00 -8.17987263e-01 -4.50080246e-01
-4.76280898e-01 -7.16746807e-01 9.14691210e-01 7.50984848e-01
2.85691470e-01 -8.57346117e-01 -2.16674551e-01 2.07185626e-01
-3.58728290e-01 -1.13429642e+00 -3.43420327e-01 7.30658770e-01
-8.42990696e-01 -1.04879200e+00 1.08106963e-01 -5.94464958e-01
4.67594683e-01 -2.76404619e-01 1.34438729e+00 -1.97669402e-01
2.69544739e-02 4.92750220e-02 -3.31023604e-01 -4.00431693e-01
-4.62229997e-01 1.93537578e-01 -9.60710719e-02 -4.01556492e-01
5.11324704e-01 -4.69579309e-01 -8.73346031e-02 1.63256332e-01
-1.25489080e+00 -1.52793735e-01 3.44453603e-01 9.11582947e-01
8.46968472e-01 1.62808761e-01 6.32713616e-01 -1.07454753e+00
6.62997246e-01 -7.33717322e-01 -7.15521216e-01 6.34977341e-01
-7.00101674e-01 8.48530173e-01 6.28768742e-01 -4.15973425e-01
-1.15631247e+00 1.21451236e-01 -1.35128066e-01 -5.23187518e-01
-1.51980653e-01 6.89610183e-01 -7.64634728e-01 3.92220557e-01
6.40025735e-01 1.30650382e-02 -3.59488219e-01 -7.16573060e-01
8.18210781e-01 4.78280276e-01 6.66730046e-01 -1.12731910e+00
6.78530753e-01 -7.04721222e-03 -3.21618915e-01 -6.75157458e-02
-1.18860233e+00 5.29093631e-02 -4.10991997e-01 3.58713478e-01
7.03135192e-01 -6.98831737e-01 -4.21776205e-01 1.45883873e-01
-1.31645548e+00 1.01663470e-01 -2.72536188e-01 2.04938546e-01
-2.94046015e-01 2.64927417e-01 -6.73701763e-01 -5.45789003e-01
-2.03943565e-01 -1.34323466e+00 9.40198302e-01 2.92967051e-01
-1.25586718e-01 -7.36304581e-01 -1.34133533e-01 4.32173312e-01
3.16064477e-01 -1.06790185e-01 1.53408659e+00 -1.17307162e+00
-6.45996690e-01 5.62986471e-02 -3.05204481e-01 3.97910923e-01
1.44597879e-02 6.40710592e-02 -1.19743490e+00 9.11704004e-02
5.53750470e-02 -4.69378650e-01 1.14379883e+00 1.16466455e-01
1.65146649e+00 -3.31327945e-01 -4.55187857e-02 6.65004313e-01
1.29611933e+00 -1.32903326e-02 5.23696601e-01 2.89017051e-01
6.61143780e-01 9.07568693e-01 4.92767453e-01 2.90023208e-01
5.83012879e-01 3.10244054e-01 8.52472365e-01 4.93237436e-01
2.03867406e-01 -6.46797478e-01 3.09121013e-01 8.17914844e-01
6.86667502e-01 -2.94945776e-01 -7.77861416e-01 4.68995422e-01
-1.92471421e+00 -5.19960105e-01 2.49054074e-01 2.04534721e+00
1.08015847e+00 1.76609859e-01 -5.06063640e-01 -2.83318967e-01
7.34375834e-01 3.07628721e-01 -8.79785120e-01 -5.15166640e-01
-2.26136163e-01 1.72961697e-01 3.36161464e-01 5.17062962e-01
-1.09424341e+00 1.14925814e+00 5.78154421e+00 8.11242998e-01
-7.38380194e-01 1.70391556e-02 6.75442636e-01 -2.30262235e-01
-8.81625116e-01 -4.25267499e-03 -1.11907029e+00 3.48465502e-01
1.05226719e+00 -1.06928967e-01 4.50096697e-01 1.01832998e+00
-2.43036523e-01 1.74496695e-01 -1.44210160e+00 5.12766957e-01
-2.01641381e-01 -1.28872085e+00 1.53880104e-01 -3.44717711e-01
3.29393417e-01 1.84020028e-01 2.07503885e-01 5.52706182e-01
6.13022625e-01 -1.13435996e+00 7.37584352e-01 6.53761744e-01
5.29769123e-01 -8.13918650e-01 7.19820321e-01 2.42800072e-01
-1.07512248e+00 -2.17325315e-01 -7.54795849e-01 4.01988596e-01
-2.35836074e-01 8.15788031e-01 -4.60916966e-01 7.09289253e-01
1.01720273e+00 2.70164996e-01 -8.56214941e-01 3.41617286e-01
-3.27932358e-01 3.93259674e-01 -3.99726331e-01 2.12607626e-03
1.60729270e-02 3.75178345e-02 2.79242367e-01 1.00282240e+00
3.60401303e-01 -2.19560161e-01 -1.56630352e-01 1.53112066e+00
-4.43627089e-01 -3.16695869e-02 -5.51434457e-01 -1.38670012e-01
9.01407421e-01 9.60597694e-01 -3.29074822e-02 1.99251063e-03
-3.28811407e-01 5.82047284e-01 8.26771259e-01 1.86413556e-01
-8.70104313e-01 -4.42429781e-01 8.17932129e-01 -3.78918946e-01
2.92776644e-01 2.01519534e-01 -4.17120934e-01 -1.20178294e+00
5.08086443e-01 -1.02090991e+00 6.51213050e-01 -7.72276819e-01
-1.58181536e+00 6.44743025e-01 7.71251991e-02 -6.55283332e-01
-4.01436120e-01 -7.55808830e-01 -4.94082093e-01 8.34054351e-01
-1.42448950e+00 -1.03167939e+00 1.61214963e-01 3.66123348e-01
3.01570147e-01 -1.87926337e-01 1.04933000e+00 -4.89708297e-02
-6.59293950e-01 9.30599928e-01 1.98326603e-01 2.85882860e-01
7.21017897e-01 -1.23785710e+00 6.13387942e-01 9.81625140e-01
2.80401856e-01 7.74832308e-01 6.08289957e-01 -5.87975919e-01
-1.55858576e+00 -1.19153631e+00 8.78951967e-01 -6.70074224e-01
8.61065984e-01 -3.81217033e-01 -1.38319087e+00 8.25607479e-01
7.82987569e-03 1.54249489e-01 5.88414311e-01 4.06318456e-01
-1.01759565e+00 -2.44614646e-01 -1.23206127e+00 4.08873141e-01
9.20402348e-01 -7.08036005e-01 -8.29387069e-01 2.45952919e-01
1.47436345e+00 -5.30771911e-01 -1.07698214e+00 5.99122703e-01
3.76441330e-01 -9.03486133e-01 8.39272082e-01 -1.07805026e+00
3.97884786e-01 -1.04496896e-01 -8.37315142e-01 -1.20929909e+00
-1.13719001e-01 -2.30987832e-01 -2.67764449e-01 1.53904748e+00
6.93092346e-01 -7.17191577e-01 7.41551876e-01 1.01014280e+00
-2.08517179e-01 -8.46547067e-01 -1.31827724e+00 -8.77738237e-01
6.47132874e-01 -8.26701820e-01 1.02785587e+00 8.38445187e-01
-1.23161800e-01 1.18598394e-01 -3.56430225e-02 7.15480149e-01
5.75761139e-01 1.00623548e-01 1.23004884e-01 -1.24730361e+00
-5.39008856e-01 -3.13642293e-01 -6.15647212e-02 -9.35789049e-01
5.34178019e-01 -9.65857446e-01 1.48668200e-01 -1.16647255e+00
1.03397854e-01 -4.47246909e-01 -6.46529377e-01 6.10423923e-01
-3.64220917e-01 -5.48173368e-01 2.24935204e-01 -1.91407666e-01
-4.53872561e-01 5.25683045e-01 8.97166610e-01 -2.15256527e-01
1.14452302e-01 -9.29814875e-02 -1.29595518e+00 5.72979212e-01
9.48342085e-01 -8.60273182e-01 -4.27185476e-01 -1.00406408e+00
6.03953898e-01 -9.85234678e-02 2.13905513e-01 -6.65989995e-01
2.26374447e-01 -1.42112941e-01 1.07894763e-01 -5.28361380e-01
2.95167238e-01 -6.62461817e-01 -1.53033480e-01 2.25891322e-01
-8.30355287e-01 8.17941204e-02 2.83986866e-01 6.47010326e-01
-2.51125515e-01 -6.51496351e-01 6.14218175e-01 -2.16340110e-01
-5.90658426e-01 2.33259499e-01 1.97195262e-01 5.85450411e-01
4.86024559e-01 5.09266496e-01 -7.36079216e-01 -1.48408204e-01
-4.20997739e-01 3.38063329e-01 1.04736201e-02 8.09807301e-01
5.80464602e-01 -1.25375450e+00 -6.68720603e-01 3.18073750e-01
4.38767195e-01 4.42936927e-01 1.43178180e-01 1.53731301e-01
-1.88734680e-02 5.85647762e-01 1.65115908e-01 -4.55387473e-01
-8.67040753e-01 5.61885357e-01 5.25677025e-01 -3.43866110e-01
-2.05362514e-01 1.14204776e+00 2.84706861e-01 -1.17583859e+00
6.47857487e-01 -7.25489914e-01 1.16558122e-02 -1.48159772e-01
3.40689510e-01 9.17018056e-02 2.10414097e-01 -1.68532893e-01
-5.21784544e-01 2.16285765e-01 -3.05281371e-01 1.01733662e-01
1.28758192e+00 2.71955822e-02 -1.36539996e-01 4.67196763e-01
1.24573338e+00 -1.12555355e-01 -1.23744488e+00 -6.09585047e-01
3.67349505e-01 -3.43243897e-01 5.85017027e-03 -9.98910606e-01
-1.21153545e+00 1.04998362e+00 5.80302835e-01 6.66762590e-02
9.54282939e-01 2.87303388e-01 5.71456492e-01 9.34029281e-01
-1.65280551e-02 -1.09920466e+00 -6.06279038e-02 6.49677396e-01
1.05244136e+00 -1.42193472e+00 -3.83641183e-01 -3.70258719e-01
-6.90858424e-01 1.21470547e+00 8.55415523e-01 1.76141113e-01
5.03874540e-01 5.01354635e-01 7.60945752e-02 7.28384852e-02
-1.07700062e+00 1.31484613e-01 2.90779799e-01 6.17659092e-01
4.51446593e-01 5.20519465e-02 2.26152524e-01 1.32577693e+00
-2.16316879e-01 -3.72657448e-01 3.70979846e-01 6.59143090e-01
-4.90993470e-01 -1.28173292e+00 -1.72339454e-01 4.16094065e-01
-4.32240337e-01 -4.52770174e-01 -3.08968902e-01 6.39967501e-01
-4.50600199e-02 1.00837231e+00 8.40108320e-02 -4.43698823e-01
4.57119524e-01 5.10564685e-01 2.28478760e-01 -5.79436064e-01
-4.41729158e-01 -3.13113183e-01 2.55311370e-01 -8.02530527e-01
1.83075607e-01 -5.06651402e-01 -1.47536051e+00 -8.00992399e-02
-1.70783684e-01 6.85247034e-02 7.93525457e-01 8.98365676e-01
7.36219347e-01 7.22682178e-01 4.28605616e-01 -1.34886252e-02
-1.23283672e+00 -8.90241504e-01 -4.53307390e-01 1.66957900e-01
7.90547766e-03 -5.76031208e-01 -3.48191053e-01 -4.09377575e-01]
|
[10.02431869506836, 7.916052341461182]
|
96fe82cc-4620-4d3f-9eca-1066889ff1a5
|
negation-detection-in-dutch-clinical-texts-an
|
2209.00470
| null |
https://arxiv.org/abs/2209.00470v1
|
https://arxiv.org/pdf/2209.00470v1.pdf
|
Negation detection in Dutch clinical texts: an evaluation of rule-based and machine learning methods
|
As structured data are often insufficient, labels need to be extracted from free text in electronic health records when developing models for clinical information retrieval and decision support systems. One of the most important contextual properties in clinical text is negation, which indicates the absence of findings. We aimed to improve large scale extraction of labels by comparing three methods for negation detection in Dutch clinical notes. We used the Erasmus Medical Center Dutch Clinical Corpus to compare a rule-based method based on ContextD, a biLSTM model using MedCAT and (finetuned) RoBERTa-based models. We found that both the biLSTM and RoBERTa models consistently outperform the rule-based model in terms of F1 score, precision and recall. In addition, we systematically categorized the classification errors for each model, which can be used to further improve model performance in particular applications. Combining the three models naively was not beneficial in terms of performance. We conclude that the biLSTM and RoBERTa-based models in particular are highly accurate accurate in detecting clinical negations, but that ultimately all three approaches can be viable depending on the use case at hand.
|
['Saskia Haitjema', 'Miguel A. R. Rios', 'Sebastiaan R. S. Arends', 'Myrthe M. Hemker', 'Marijn Schraagen', 'Sander C. Tan', 'Leon C. Reteig', 'Bram van Es']
|
2022-09-01
| null | null | null | null |
['negation-detection']
|
['natural-language-processing']
|
[ 4.00249213e-01 2.13039353e-01 -3.38018984e-01 -4.19911802e-01
-1.16071749e+00 -6.35991454e-01 3.26227307e-01 1.12574637e+00
-8.35923791e-01 8.63917112e-01 3.71321678e-01 -8.55824172e-01
-4.64861631e-01 -5.58081508e-01 -2.07576826e-01 -3.99683177e-01
9.78685245e-02 6.56676412e-01 3.42876285e-01 -8.35702196e-02
5.99229448e-02 1.78842172e-01 -1.19319999e+00 8.92492652e-01
7.07483888e-01 7.43484199e-01 -1.13866806e-01 6.89771414e-01
-1.91370696e-01 1.17370486e+00 -4.23529983e-01 -4.57566619e-01
-2.59480596e-01 -6.77923024e-01 -8.65607738e-01 -3.57314557e-01
1.35420427e-01 -1.98431417e-01 3.02854359e-01 8.04321408e-01
6.24281764e-01 -2.15812892e-01 6.74053490e-01 -6.19960546e-01
-8.17503780e-02 5.80759108e-01 1.65876061e-01 3.42935085e-01
6.13643765e-01 -5.42235672e-02 8.42157841e-01 -6.41781509e-01
9.92220879e-01 9.19617414e-01 1.04905963e+00 6.63124681e-01
-1.09822261e+00 -4.25159365e-01 4.71543195e-03 -4.36863489e-02
-1.10707414e+00 -4.39071417e-01 1.72015131e-02 -5.57725012e-01
1.48147035e+00 3.01535964e-01 5.42691231e-01 6.77643955e-01
5.06570518e-01 4.25329357e-01 1.21222436e+00 -7.33133614e-01
2.63378918e-01 2.24052057e-01 2.65368849e-01 7.65750945e-01
5.45947611e-01 -8.03349391e-02 -2.25859717e-01 -1.02054536e+00
3.60781848e-01 -1.74730122e-01 -1.61758021e-01 6.91873953e-02
-1.24855185e+00 7.24366903e-01 1.43798694e-01 7.69657910e-01
-4.77647156e-01 -2.35593036e-01 9.03658926e-01 3.42969179e-01
2.80000627e-01 5.65628231e-01 -8.13600779e-01 -1.76160857e-01
-1.26663327e+00 2.38490209e-01 1.09028745e+00 5.62574565e-01
-1.45371497e-01 -3.69734108e-01 -3.47459197e-01 8.79701078e-01
3.09596777e-01 2.35856235e-01 8.02889645e-01 -8.49827468e-01
2.60247529e-01 7.74627864e-01 1.20230064e-01 -8.12563181e-01
-8.38902235e-01 -1.89666912e-01 -4.69485939e-01 -2.75558114e-01
5.12682140e-01 -2.35672638e-01 -1.09579051e+00 1.52667618e+00
2.02477798e-01 -3.79917204e-01 3.89350712e-01 5.22819340e-01
9.71919060e-01 2.88589913e-02 5.23402691e-01 -5.06726444e-01
1.72918534e+00 -2.12609008e-01 -1.18992078e+00 -1.31917849e-01
1.68815267e+00 -1.00169814e+00 5.50303519e-01 4.82607603e-01
-9.76006031e-01 7.67341256e-02 -9.72767949e-01 8.19847882e-02
-4.01871711e-01 1.01587079e-01 1.21111006e-01 6.42851770e-01
-9.68328178e-01 6.65340304e-01 -1.07883358e+00 -6.35747373e-01
3.03303063e-01 4.99586463e-01 -4.00623053e-01 -2.65827663e-02
-1.25840485e+00 1.30186200e+00 3.84999663e-01 5.50067313e-02
-2.30771527e-01 -2.91864604e-01 -9.76879597e-01 -1.75649121e-01
4.28700238e-01 -6.92120671e-01 1.65678787e+00 -5.57109058e-01
-6.75728202e-01 1.19592047e+00 -2.86383063e-01 -5.85735083e-01
4.55992073e-01 1.13548674e-01 -4.77388501e-01 1.23896398e-01
1.39718801e-01 2.78102428e-01 1.57022625e-01 -8.42278421e-01
-7.65329480e-01 -4.00405169e-01 -3.43658090e-01 2.53642257e-02
-2.04478242e-02 3.49764496e-01 1.06137708e-01 -5.25525093e-01
-3.64594832e-02 -8.33717287e-01 -3.54899406e-01 -1.46841943e-01
-2.74934202e-01 -4.45382118e-01 3.46617222e-01 -6.91624880e-01
1.59543145e+00 -1.77978277e+00 -6.10238791e-01 1.38223886e-01
3.46950322e-01 6.15891159e-01 1.90478951e-01 4.79785860e-01
-1.33146957e-01 5.75348854e-01 -1.32330731e-01 1.68478742e-01
-4.02904958e-01 4.53174382e-01 6.84738085e-02 2.84891337e-01
3.75249237e-01 8.83734763e-01 -1.10888100e+00 -9.75448608e-01
8.19044039e-02 5.06077290e-01 -5.28356731e-01 -2.12576851e-01
-1.40957430e-01 5.20111471e-02 -5.58547854e-01 7.15948522e-01
9.08861111e-04 -4.60791886e-01 7.44353414e-01 -1.38692930e-02
1.81006804e-01 9.43841398e-01 -8.80908370e-01 1.23528326e+00
-1.34023130e-01 2.26226375e-01 3.23007368e-02 -6.84929013e-01
6.04593456e-01 9.18586433e-01 3.38232279e-01 -5.85440636e-01
-6.77304436e-03 7.83609033e-01 4.04297024e-01 -8.38208497e-01
1.62592188e-01 -7.32782304e-01 7.26049244e-02 3.46702516e-01
-2.39712700e-01 -1.07944966e-03 3.60941380e-01 1.03324838e-01
1.32384503e+00 4.41661254e-02 1.02244508e+00 -3.40039551e-01
4.93425459e-01 5.40625632e-01 9.19127762e-01 7.73305833e-01
-3.33078980e-01 5.00830114e-01 6.34749413e-01 -4.09228832e-01
-4.94147092e-01 -4.17362124e-01 -5.21469057e-01 7.03907371e-01
-6.87282085e-01 -6.21506751e-01 -5.20321071e-01 -9.33886349e-01
-1.86666325e-01 7.84499168e-01 -6.10218763e-01 3.13888267e-02
-5.98515272e-01 -8.86573434e-01 8.30063283e-01 5.46577096e-01
-2.75580317e-01 -1.15717685e+00 -8.85049403e-01 6.65335894e-01
-2.81678170e-01 -1.05520594e+00 -2.41163671e-01 6.45622253e-01
-1.07741451e+00 -1.40341783e+00 -6.60640478e-01 -8.04047704e-01
4.51888770e-01 -4.27278608e-01 1.22195494e+00 3.32434446e-01
-2.52598524e-01 3.31397265e-01 -4.93749082e-01 -6.23473585e-01
-9.81355667e-01 4.46547754e-02 -1.43947601e-01 -7.98279881e-01
1.02286994e+00 2.89022829e-02 -4.77928400e-01 -2.50461092e-03
-1.13360000e+00 -4.42878157e-01 5.90201735e-01 9.90632594e-01
6.65330052e-01 -3.97844523e-01 7.05055177e-01 -1.59600210e+00
9.31040943e-01 -2.99845994e-01 -5.23489006e-02 2.85101533e-01
-1.07168031e+00 1.25976920e-01 3.42275381e-01 -2.09176376e-01
-6.49019897e-01 2.79748887e-02 -4.88771886e-01 1.77891046e-01
-2.37232000e-01 1.04582071e+00 3.96726072e-01 2.42960870e-01
1.03944087e+00 -1.63502321e-01 1.41147330e-01 -4.15232658e-01
-2.42787823e-01 8.38938236e-01 -9.17010680e-02 -1.40757725e-01
-8.44616294e-02 1.81002855e-01 -4.34812680e-02 -6.04317367e-01
-9.98749077e-01 -5.17244160e-01 -4.70912248e-01 1.65512487e-01
9.24461842e-01 -6.73152030e-01 -4.84665573e-01 -3.72585543e-02
-9.95315492e-01 -3.37433219e-01 -4.73744273e-01 6.24775589e-01
-2.47137547e-01 4.09041524e-01 -9.63155508e-01 -7.50641167e-01
-4.99712944e-01 -1.12678289e+00 9.37615395e-01 -2.28666991e-01
-9.58941162e-01 -1.22351468e+00 2.59566277e-01 7.55002573e-02
1.59438476e-01 3.32699299e-01 1.40852320e+00 -1.54726386e+00
3.09588820e-01 -5.40607393e-01 -2.42570136e-02 2.68295884e-01
4.59777772e-01 -1.00437976e-01 -6.77426040e-01 -1.34776533e-01
2.91670054e-01 -3.68170738e-01 7.73230553e-01 4.19863611e-01
5.28739750e-01 -3.07453930e-01 -5.89622676e-01 -1.36826113e-01
1.32501912e+00 6.21539712e-01 3.65631461e-01 3.86123627e-01
3.68019044e-01 5.73848367e-01 6.76666677e-01 1.82304814e-01
3.53569716e-01 3.47285062e-01 -1.92872882e-01 -1.45426050e-01
3.53972940e-03 2.46560457e-03 9.58416685e-02 8.74852955e-01
9.12150815e-02 -1.00082457e-01 -1.36142349e+00 6.37429953e-01
-1.82795370e+00 -6.51459396e-01 -3.16713989e-01 2.09556341e+00
1.12311471e+00 5.50616384e-01 1.33329937e-02 3.48380893e-01
3.33168387e-01 -3.48945677e-01 -1.97491914e-01 -6.20434284e-01
4.45550047e-02 2.15765089e-01 4.54946756e-01 3.89704287e-01
-9.06910121e-01 4.52823400e-01 7.21815681e+00 4.73215163e-01
-9.20785964e-01 2.81318247e-01 4.68733579e-01 -8.50794688e-02
-8.50883722e-02 -1.48947418e-01 -8.40679348e-01 2.63020664e-01
1.22764730e+00 8.00248235e-02 -4.51977283e-01 5.73816419e-01
4.99640971e-01 -3.04084927e-01 -1.38589275e+00 4.88008857e-01
-8.91741216e-02 -1.32588780e+00 3.70435268e-02 1.30122885e-01
4.33250964e-01 -6.40663877e-02 -2.56347448e-01 2.91443557e-01
1.62669241e-01 -9.83819664e-01 2.44833112e-01 4.02783483e-01
8.85967314e-01 -1.28072113e-01 1.42771542e+00 3.07934254e-01
-6.79869235e-01 2.46905237e-01 2.14108415e-02 1.96394641e-02
5.70938475e-02 6.28229082e-01 -1.22485924e+00 5.51995218e-01
5.22911966e-01 4.13492143e-01 -4.59290057e-01 1.01920402e+00
-2.17316840e-02 7.58268893e-01 -3.45523924e-01 -1.53727636e-01
3.46509516e-01 2.90661663e-01 3.44341874e-01 1.66067529e+00
1.03214033e-01 3.20914060e-01 2.23062605e-01 3.83286744e-01
1.29348248e-01 3.58957946e-01 -6.40586376e-01 -1.79420128e-01
2.93215752e-01 9.65821385e-01 -1.03701663e+00 -6.86300337e-01
-4.88435119e-01 3.53401899e-01 -3.00819073e-02 9.58434939e-02
-3.25729936e-01 -2.47459263e-01 2.14882977e-02 5.99683523e-01
2.78007895e-01 4.47437942e-01 -2.97243029e-01 -9.20603812e-01
1.93936471e-02 -1.23306334e+00 9.79953349e-01 -5.09545267e-01
-1.22493553e+00 7.92856038e-01 -9.57813933e-02 -1.25788391e+00
-7.83111691e-01 -7.95228124e-01 1.55420350e-02 7.09930539e-01
-1.38258028e+00 -7.57357299e-01 1.64089501e-01 2.92833537e-01
2.13863164e-01 1.76599011e-01 1.30716670e+00 2.92166382e-01
-3.40099603e-01 4.86998707e-01 -1.00230388e-02 3.16087633e-01
1.10578144e+00 -1.26749766e+00 -6.09031133e-02 3.94922048e-01
-1.36498481e-01 9.40008163e-01 7.44483769e-01 -9.30652976e-01
-6.99207664e-01 -9.33576047e-01 1.63560343e+00 -5.63542187e-01
4.47725832e-01 1.30683616e-01 -1.20827734e+00 7.82734573e-01
-1.04328141e-01 -2.06637025e-01 1.13050139e+00 1.47933692e-01
-2.42396325e-01 4.10957128e-01 -1.35530627e+00 3.67166847e-01
5.79180956e-01 -5.37362874e-01 -1.12672210e+00 6.14989817e-01
4.99456495e-01 -3.50899607e-01 -1.15645874e+00 7.34647512e-01
4.84891444e-01 -8.06791782e-01 4.76556450e-01 -9.78071570e-01
2.95639187e-01 -9.21143368e-02 -4.48313355e-02 -7.59759009e-01
-6.49346858e-02 -1.49954230e-01 1.92755342e-01 8.21665406e-01
1.08083403e+00 -7.39983916e-01 7.02991009e-01 8.94037366e-01
1.41464353e-01 -1.03737795e+00 -9.72135365e-01 -3.63587707e-01
2.10466474e-01 -5.51563740e-01 -2.30670329e-02 1.06601584e+00
4.70962286e-01 3.65374416e-01 2.99711257e-01 -2.69468486e-01
4.54075634e-02 -2.11518511e-01 -9.78442356e-02 -1.24718559e+00
-2.51131970e-02 -2.85214961e-01 -4.36199158e-01 -2.77402073e-01
-2.98755467e-01 -1.04317486e+00 1.30501678e-02 -2.07625508e+00
3.84746790e-01 -4.59685057e-01 -4.96039987e-01 1.01048052e+00
-2.85602391e-01 1.68385461e-01 -2.32413575e-01 2.48825341e-01
-5.53002357e-01 -2.55208671e-01 7.58880138e-01 5.49464226e-02
-4.19725180e-01 2.68829819e-02 -7.99131095e-01 9.57779408e-01
5.68800151e-01 -1.08313417e+00 -8.23700354e-02 -7.09135309e-02
5.28670907e-01 2.28508487e-01 1.70469642e-01 -5.45076907e-01
3.64396304e-01 1.46448418e-01 4.03568625e-01 -4.46689069e-01
-5.20136319e-02 -6.82998657e-01 -1.79218411e-01 1.01907110e+00
-6.75363600e-01 2.11841717e-01 4.86095846e-01 4.36891079e-01
-2.96088576e-01 -5.79466581e-01 4.94612902e-01 -5.19066989e-01
-1.16146870e-01 -4.13853526e-01 -9.14148390e-01 3.48232031e-01
7.51094520e-01 -2.64725149e-01 -1.94305211e-01 -2.13838458e-01
-1.17936087e+00 2.57789552e-01 2.53059119e-01 2.78292969e-02
5.62072635e-01 -8.48151982e-01 -7.26803958e-01 -1.06477968e-01
2.74862677e-01 -1.60078168e-01 6.90688705e-03 1.16712880e+00
-6.41220510e-01 9.00057852e-01 2.48326898e-01 -5.87100804e-01
-1.63691199e+00 6.05133355e-01 5.12218773e-01 -9.07756448e-01
-3.87250215e-01 4.80328053e-01 -2.36668050e-01 -4.92324233e-01
2.79593766e-01 -8.53910446e-01 -3.34310174e-01 3.86947155e-01
5.13693154e-01 3.55562977e-02 7.43638575e-01 -4.32482153e-01
-8.38331759e-01 1.53611064e-01 -5.31004786e-01 -2.38974020e-01
1.16676021e+00 3.20999205e-01 -2.02912927e-01 6.63456261e-01
7.78815687e-01 -1.05185993e-01 -7.80018941e-02 -2.93760866e-01
5.96692502e-01 1.18278459e-01 1.54378921e-01 -1.22592163e+00
-3.37998658e-01 5.96545100e-01 5.25002182e-01 3.56271297e-01
9.69118834e-01 -2.29253694e-01 4.09226328e-01 4.96485412e-01
1.02676086e-01 -1.05073965e+00 -4.71168578e-01 2.83024341e-01
3.68251592e-01 -1.25944638e+00 1.48887306e-01 -3.28155786e-01
-7.39299655e-01 1.07164800e+00 4.32383008e-02 3.32423985e-01
7.49673903e-01 4.67035949e-01 4.83867794e-01 -5.88145733e-01
-1.04774344e+00 -1.42981231e-01 2.86234617e-01 3.81328285e-01
1.14139807e+00 -6.52980208e-02 -9.52992082e-01 7.34204471e-01
1.52209520e-01 4.94105339e-01 5.58138549e-01 1.41988456e+00
-1.86699346e-01 -1.41278219e+00 -3.47379267e-01 1.22623682e+00
-1.16736865e+00 -4.09358501e-01 -5.05577087e-01 9.54066873e-01
1.88334614e-01 1.08533239e+00 -3.69350791e-01 -1.20520599e-01
5.88778317e-01 6.31054759e-01 3.16238999e-01 -1.02031565e+00
-1.09932363e+00 3.73978466e-01 6.49581015e-01 -3.84898573e-01
-7.32785404e-01 -7.49471903e-01 -1.31255782e+00 3.08449119e-01
-5.97816110e-01 3.39744925e-01 3.85768145e-01 1.11338580e+00
3.55401307e-01 5.93906879e-01 -3.28900933e-01 1.48189247e-01
-7.51806319e-01 -1.05718529e+00 -2.56699443e-01 1.42178193e-01
4.47966993e-01 -3.51035267e-01 -2.89342403e-01 9.39239413e-02]
|
[8.44442367553711, 8.759385108947754]
|
2f053290-1cb1-4b58-9ec3-d2d685107e4e
|
a-survey-of-deep-visual-cross-domain-few-shot
|
2303.09253
| null |
https://arxiv.org/abs/2303.09253v1
|
https://arxiv.org/pdf/2303.09253v1.pdf
|
A Survey of Deep Visual Cross-Domain Few-Shot Learning
|
Few-Shot transfer learning has become a major focus of research as it allows recognition of new classes with limited labeled data. While it is assumed that train and test data have the same data distribution, this is often not the case in real-world applications. This leads to decreased model transfer effects when the new class distribution differs significantly from the learned classes. Research into Cross-Domain Few-Shot (CDFS) has emerged to address this issue, forming a more challenging and realistic setting. In this survey, we provide a detailed taxonomy of CDFS from the problem setting and corresponding solutions view. We summarise the existing CDFS network architectures and discuss the solution ideas for each direction the taxonomy indicates. Furthermore, we introduce various CDFS downstream applications and outline classification, detection, and segmentation benchmarks and corresponding standards for evaluation. We also discuss the challenges of CDFS research and explore potential directions for future investigation. Through this review, we aim to provide comprehensive guidance on CDFS research, enabling researchers to gain insight into the state-of-the-art while allowing them to build upon existing solutions to develop their own CDFS models.
|
['Zhaoxiang Zhang', 'Zhi Gong', 'Junsong Fan', 'Yuxi Wang', 'Lijuan Duan', 'Wenjian Wang']
|
2023-03-16
| null | null | null | null |
['cross-domain-few-shot', 'cross-domain-few-shot-learning']
|
['computer-vision', 'computer-vision']
|
[ 1.42334074e-01 -1.21804193e-01 -6.16226673e-01 -9.35186148e-01
-1.04118013e+00 -5.97563267e-01 3.40286314e-01 -8.17271024e-02
-1.39172539e-01 7.88414419e-01 -2.17433885e-01 -4.12040532e-01
-1.89993940e-02 -8.56028676e-01 -6.18748248e-01 -6.07485056e-01
6.03940003e-02 6.95229828e-01 6.68538928e-01 3.91251221e-02
-6.56629354e-02 5.54035068e-01 -1.67900956e+00 4.06246096e-01
6.36731267e-01 1.07939434e+00 1.74173072e-01 6.13087952e-01
-4.00685877e-01 4.71118331e-01 -1.02437997e+00 -5.88187218e-01
2.63823897e-01 -3.48953277e-01 -6.56693757e-01 -2.28552371e-02
4.76076543e-01 -4.23867166e-01 -2.74565697e-01 9.69280481e-01
7.14061737e-01 3.82776529e-01 1.01340222e+00 -1.73431253e+00
-9.20331776e-01 5.01991868e-01 -5.30606270e-01 6.66212559e-01
-6.71217591e-02 6.41830191e-02 8.63331616e-01 -6.15886390e-01
7.26778686e-01 1.00222898e+00 7.93423474e-01 8.23676705e-01
-1.11687934e+00 -8.35500062e-01 2.45787486e-01 4.77861285e-01
-1.09992540e+00 -4.19149250e-01 4.50968862e-01 -3.04378092e-01
9.53065574e-01 2.20398176e-02 6.09357834e-01 1.35869062e+00
-8.94857385e-06 1.25547516e+00 8.18767130e-01 -3.95919025e-01
6.79483652e-01 2.94377208e-01 6.72692120e-01 1.05610415e-01
1.75394744e-01 1.48735046e-01 -5.23578942e-01 1.10547252e-01
4.22222376e-01 -1.42642722e-01 1.56374827e-01 -9.60785031e-01
-6.29296958e-01 9.85623658e-01 4.22548473e-01 4.29427505e-01
1.32693365e-01 -5.61499558e-02 5.15048444e-01 3.81539434e-01
7.29508102e-01 1.64739132e-01 -6.86686337e-01 -3.49170178e-01
-1.12421894e+00 4.69986767e-01 8.16707790e-01 1.35567486e+00
4.22582746e-01 3.59169900e-01 -2.61763632e-01 9.54229355e-01
1.62116550e-02 2.65262187e-01 2.03184471e-01 -9.15412009e-01
2.27363989e-01 1.44845648e-02 -1.28668800e-01 -2.47169703e-01
-1.05489224e-01 -4.54347581e-01 -3.16542715e-01 2.18128562e-01
5.64927936e-01 -3.29800248e-01 -1.18028247e+00 1.57374811e+00
1.87810823e-01 5.60481429e-01 4.71317694e-02 5.42339444e-01
9.76868153e-01 5.75503767e-01 2.53476292e-01 -8.39918479e-03
9.45889115e-01 -8.23944211e-01 -4.28768486e-01 -2.51628131e-01
3.50900263e-01 -5.80258250e-01 1.03935933e+00 2.14910209e-02
-6.99265480e-01 -6.06220245e-01 -1.17418706e+00 1.99424073e-01
-8.59083533e-01 -6.18395448e-01 8.60893607e-01 9.54389215e-01
-7.51504779e-01 7.46261001e-01 -9.07358050e-01 -8.48471999e-01
8.17976832e-01 1.46866649e-01 2.17394859e-01 -3.76525015e-01
-1.29388130e+00 9.17657733e-01 1.85485035e-01 -5.95089674e-01
-8.52777481e-01 -1.23638749e+00 -8.14985394e-01 3.33865255e-01
3.92514467e-01 -4.43249136e-01 1.99307585e+00 -5.30832708e-01
-1.26542473e+00 7.18739629e-01 1.30379766e-01 -4.16125625e-01
3.65063846e-01 1.36072055e-01 -6.61116540e-01 -9.50345397e-02
2.43228406e-01 7.89588630e-01 5.29472053e-01 -1.05760849e+00
-1.03481150e+00 -7.97567517e-02 -3.60985138e-02 3.17237824e-02
-2.70285547e-01 3.28325033e-01 -3.87690067e-01 -5.76813936e-01
-5.50137758e-01 -5.86423635e-01 1.03523538e-01 1.13611460e-01
-1.53801069e-01 -2.13595450e-01 1.00521243e+00 3.18601215e-03
9.41006720e-01 -2.16581583e+00 -3.37754905e-01 -2.48489618e-01
1.39631167e-01 4.49660808e-01 -1.21302836e-01 4.31322277e-01
-2.00063169e-01 -2.53011227e-01 -1.78162202e-01 -1.35051444e-01
1.33763507e-01 3.28266233e-01 -4.70110953e-01 3.26332927e-01
2.48220995e-01 1.06604159e+00 -1.06385612e+00 -1.65153101e-01
3.52009714e-01 2.94150054e-01 -2.44079083e-01 1.67344689e-01
-2.41930723e-01 4.55939710e-01 -3.19573998e-01 8.92751038e-01
7.06453741e-01 -4.06763285e-01 -1.42925128e-01 2.04371149e-03
-8.86550397e-02 -4.99181189e-02 -1.07218766e+00 1.52665889e+00
-1.31148830e-01 8.14082563e-01 -3.85437757e-01 -1.35017300e+00
7.34531701e-01 1.43908784e-01 5.57180166e-01 -6.54996753e-01
1.98572069e-01 1.61736727e-01 1.48936301e-01 -2.49191046e-01
1.68355703e-01 -4.96372014e-01 -1.99524507e-01 2.29216382e-01
5.90483665e-01 1.13269491e-02 4.07189459e-01 1.61356553e-01
1.09013712e+00 7.32792392e-02 4.86985654e-01 -9.68248323e-02
-1.15730017e-01 4.22730036e-02 5.52967668e-01 1.04164481e+00
-6.60688519e-01 5.66166818e-01 4.19006854e-01 -4.34767544e-01
-9.93697166e-01 -1.64328575e+00 -4.46144938e-01 1.45467579e+00
6.35984242e-02 8.76538306e-02 -7.56071925e-01 -8.86768758e-01
2.90220946e-01 1.05580938e+00 -6.69402242e-01 -2.37357125e-01
-1.15258396e-01 -1.11672163e+00 6.05581999e-01 1.11636388e+00
3.09578717e-01 -9.19407904e-01 -5.52402735e-01 7.45532736e-02
2.10652694e-01 -9.20252264e-01 3.60893123e-02 5.43475866e-01
-6.73929632e-01 -1.23896945e+00 -1.33881247e+00 -9.62761939e-01
1.51152566e-01 3.99868608e-01 1.11307967e+00 -4.94527817e-01
-5.57615161e-01 5.40054500e-01 -4.09533054e-01 -6.87384546e-01
-2.27737159e-01 2.70436913e-01 -1.34046212e-01 -4.08988982e-01
9.25911248e-01 -4.90699857e-01 -3.87912035e-01 2.46832058e-01
-5.66975415e-01 -3.49518150e-01 2.81281263e-01 9.06305254e-01
3.98656547e-01 -1.28508031e-01 9.38158214e-01 -1.24766374e+00
7.19327569e-01 -8.22981894e-01 -6.04072988e-01 4.23150688e-01
-8.19555998e-01 -3.48215222e-01 3.48759800e-01 -3.92302334e-01
-1.44075716e+00 -2.02071458e-01 -2.31844455e-01 -3.75877291e-01
-5.17060161e-01 1.49828762e-01 -6.85685202e-02 -1.63803533e-01
7.25992322e-01 -1.02711245e-02 -3.21395069e-01 -7.54028022e-01
6.73027754e-01 9.16113496e-01 4.16780293e-01 -6.08989179e-01
5.66111207e-01 2.20238835e-01 -3.94661516e-01 -8.40239525e-01
-1.13563859e+00 -7.70398438e-01 -7.74838507e-01 -3.01348090e-01
7.90116489e-01 -6.74078286e-01 -1.76443458e-01 6.37110353e-01
-9.65267539e-01 -4.08196867e-01 -7.31571436e-01 3.54743928e-01
-7.43975401e-01 1.26789451e-01 -9.19009328e-01 -5.93147337e-01
-1.23786717e-03 -1.18840778e+00 8.77709925e-01 5.46544552e-01
-1.54355854e-01 -1.06533504e+00 1.35892630e-01 1.37451738e-01
5.99191427e-01 2.98435567e-03 1.14095092e+00 -1.05667770e+00
-3.20320308e-01 -2.69388825e-01 -4.83625561e-01 2.65989333e-01
1.37578174e-02 -9.74015053e-03 -1.30523396e+00 -2.26147249e-01
-2.58539051e-01 -4.69555348e-01 7.84508228e-01 6.22339010e-01
1.14101458e+00 6.67768836e-01 -5.16361117e-01 5.44515312e-01
1.20953691e+00 5.51965237e-01 3.29088777e-01 1.87844291e-01
2.98812002e-01 4.38341051e-01 8.09738696e-01 3.43145728e-01
1.94328472e-01 5.17671168e-01 9.45993587e-02 -1.06205575e-01
-5.47839642e-01 -1.16312439e-02 -6.89108670e-02 8.58918190e-01
2.55336195e-01 -6.19554281e-01 -1.13146698e+00 4.37698305e-01
-1.65038657e+00 -8.87894094e-01 1.09918423e-01 2.07625389e+00
6.20726347e-01 2.36622617e-01 1.20750055e-01 -9.62971523e-02
1.06626916e+00 4.56164666e-02 -1.01066935e+00 -2.21902952e-01
8.48979354e-02 5.38061082e-01 4.61057216e-01 9.14137736e-02
-1.26738262e+00 1.06584466e+00 7.69892406e+00 1.01912224e+00
-9.47979510e-01 2.12181076e-01 7.55268276e-01 -2.68895745e-01
1.26649663e-01 -1.91503331e-01 -1.08327293e+00 5.18561542e-01
9.79292691e-01 -3.43680650e-01 1.57447264e-01 1.08477223e+00
-3.34850878e-01 -1.10047258e-01 -1.36508465e+00 7.79571533e-01
5.84291480e-02 -1.28637147e+00 -2.53457546e-01 -1.78783938e-01
8.30571890e-01 3.58895451e-01 2.87619770e-01 9.74545062e-01
4.15099680e-01 -5.69701970e-01 4.71264690e-01 2.10408181e-01
1.17279875e+00 -6.44666374e-01 5.74529946e-01 2.27275804e-01
-1.20786905e+00 -1.05993010e-01 -6.36210442e-01 -1.12780116e-01
2.77162880e-01 3.69761080e-01 -7.89842725e-01 4.25467134e-01
8.75742972e-01 8.69278967e-01 -5.01025736e-01 1.47243500e+00
5.23951240e-02 7.87702382e-01 -1.05650142e-01 -2.11495925e-02
1.98355287e-01 2.14740276e-01 2.57590830e-01 1.21177745e+00
3.40230584e-01 -7.32009113e-02 2.18528956e-01 7.98297524e-01
-1.14903331e-01 -1.01839945e-01 -4.95889664e-01 -1.75289944e-01
5.75552225e-01 8.65303099e-01 -1.09288895e+00 -6.10788047e-01
-1.11488068e+00 8.03340971e-01 2.38599822e-01 5.70829690e-01
-9.01553333e-01 -7.14303434e-01 8.49967301e-01 2.92136800e-02
3.66783768e-01 2.30272591e-01 -2.38565519e-01 -1.13170922e+00
-3.65846336e-01 -4.28820848e-01 7.04414845e-01 -5.80893755e-01
-1.97052646e+00 2.12140694e-01 4.94328648e-01 -1.21216762e+00
-1.55945599e-01 -7.23697543e-01 -7.65472770e-01 6.09684944e-01
-1.62147045e+00 -1.00289512e+00 -1.98402256e-01 3.30618024e-01
9.55697596e-01 -5.18352091e-01 8.12886119e-01 7.64566958e-01
-5.40491879e-01 6.94943070e-01 6.13631308e-01 1.36318877e-01
1.18340552e+00 -1.21444094e+00 9.04146552e-01 4.80134428e-01
6.05977587e-02 3.84138137e-01 6.89683437e-01 -8.67725372e-01
-8.46910894e-01 -1.13387728e+00 5.07606149e-01 -1.92209318e-01
8.32231939e-01 -3.98696840e-01 -1.01380193e+00 8.24114680e-01
1.03219479e-01 3.55078608e-01 1.03044701e+00 3.03174734e-01
-3.52076411e-01 -2.12013409e-01 -1.24438274e+00 3.02216649e-01
1.13130546e+00 -2.50127882e-01 -6.33188486e-01 2.95493811e-01
5.14730871e-01 -4.41487193e-01 -7.28498042e-01 3.76796037e-01
5.38120449e-01 -9.80565786e-01 9.46822941e-01 -1.04123735e+00
7.69235864e-02 1.27906442e-01 -2.15161130e-01 -1.45748079e+00
-4.03181732e-01 -2.84369104e-02 -3.30241650e-01 1.36713171e+00
3.26482773e-01 -4.97013688e-01 1.04436851e+00 8.00624847e-01
-2.91936219e-01 -8.05703700e-01 -9.70937192e-01 -1.27310276e+00
5.12850642e-01 -8.11657965e-01 6.10925972e-01 9.07126546e-01
1.65998582e-02 2.91129023e-01 -1.59081921e-01 -2.83164740e-01
6.68961644e-01 1.55750811e-01 5.12856007e-01 -1.30494416e+00
-1.92261383e-01 -4.05616194e-01 -6.30574226e-01 -1.06835377e+00
1.47728339e-01 -1.06359720e+00 1.66088089e-01 -1.46507812e+00
4.13144618e-01 -3.73737484e-01 -5.98012328e-01 4.23711151e-01
3.21756564e-02 3.42841744e-01 2.64776528e-01 -1.11645818e-01
-7.80333459e-01 5.42047858e-01 9.51529622e-01 -1.50489703e-01
5.18164523e-02 4.19811815e-01 -6.90693736e-01 6.45867884e-01
9.25422966e-01 -3.88085246e-01 -7.60890901e-01 -2.61273712e-01
-3.17848265e-01 -1.85944438e-01 -2.04325989e-02 -1.14565217e+00
2.14926213e-01 -1.90858752e-01 7.35719442e-01 -6.06395364e-01
2.69463539e-01 -5.93152940e-01 -2.88224906e-01 1.53840303e-01
-2.85454214e-01 -3.60862106e-01 2.71064341e-01 8.08750629e-01
-1.11971729e-01 -4.68564957e-01 1.24405921e+00 -1.44124627e-01
-1.33362865e+00 5.22854626e-01 -4.97551829e-01 4.37876284e-01
1.59170735e+00 -2.18632802e-01 -3.69075894e-01 -2.82829851e-01
-8.08608830e-01 2.58762568e-01 3.14535618e-01 6.89882457e-01
3.37588042e-01 -1.34547353e+00 -3.40248406e-01 1.14466347e-01
4.83868033e-01 -2.52834767e-01 3.62599671e-01 3.59743148e-01
-1.62027106e-01 3.51107031e-01 -3.49503428e-01 -4.91061836e-01
-1.10446346e+00 6.61224425e-01 1.20309949e-01 7.77730122e-02
-5.76418817e-01 1.02824628e+00 4.12047923e-01 -4.65080172e-01
5.63169241e-01 -1.83528274e-01 -1.55274436e-01 1.33948907e-01
5.45362473e-01 6.69206500e-01 8.58299136e-02 -1.66786745e-01
-3.41032296e-01 1.58084169e-01 -2.70644188e-01 2.04023182e-01
1.49552965e+00 -7.36652389e-02 5.99803627e-01 8.32165360e-01
1.20074821e+00 -5.56964278e-01 -1.49738085e+00 -3.23794872e-01
1.83887973e-01 -4.49115098e-01 -2.13602543e-01 -1.08997130e+00
-1.11054003e+00 1.17426455e+00 7.84592271e-01 6.77128583e-02
8.66502106e-01 3.73239726e-01 9.23643649e-01 1.55720755e-01
2.79971719e-01 -1.30535376e+00 -3.86604406e-02 5.23057163e-01
2.52583891e-01 -1.30009019e+00 -2.41318047e-01 -4.62095410e-01
-5.90260804e-01 1.21691620e+00 7.42817461e-01 -9.31119695e-02
1.00720537e+00 5.71048558e-01 -5.17017394e-02 -7.01981038e-02
-4.95895743e-01 -3.50295961e-01 1.34499148e-01 1.38307798e+00
3.89374554e-01 -7.26239011e-02 2.81480044e-01 8.26589823e-01
5.96402660e-02 1.68210164e-01 3.29081655e-01 1.05453348e+00
-6.25211000e-01 -1.32047248e+00 -1.44530103e-01 9.01575089e-01
-1.51430115e-01 -7.90908560e-03 -1.53983980e-01 8.16492856e-01
2.80606657e-01 7.41327345e-01 2.99733758e-01 -1.66923389e-01
4.02050734e-01 4.35910881e-01 6.24354422e-01 -1.12920606e+00
-1.11751407e-01 -4.08433005e-02 -1.09534554e-01 -2.73059726e-01
-1.16941564e-01 -8.04687917e-01 -1.10528195e+00 -3.91206920e-01
-3.79010648e-01 -9.07321572e-02 5.95828712e-01 9.58935976e-01
2.47517049e-01 7.34504640e-01 1.85082689e-01 -6.95923030e-01
-6.64763868e-01 -7.18606472e-01 -9.30994272e-01 4.79750223e-02
5.66037148e-02 -1.09917176e+00 4.53585163e-02 3.24550718e-02]
|
[9.95096206665039, 2.8570895195007324]
|
39d4ab56-bbc7-49ca-8cb6-aa69ca7fecc7
|
samsung-r-d-institute-poland-submission-to
| null | null |
https://aclanthology.org/2021.wat-1.27
|
https://aclanthology.org/2021.wat-1.27.pdf
|
Samsung R&D Institute Poland submission to WAT 2021 Indic Language Multilingual Task
|
This paper describes the submission to the WAT 2021 Indic Language Multilingual Task by Samsung R&D Institute Poland. The task covered translation between 10 Indic Languages (Bengali, Gujarati, Hindi, Kannada, Malayalam, Marathi, Oriya, Punjabi, Tamil and Telugu) and English. We combined a variety of techniques: transliteration, filtering, backtranslation, domain adaptation, knowledge-distillation and finally ensembling of NMT models. We applied an effective approach to low-resource training that consist of pretraining on backtranslations and tuning on parallel corpora. We experimented with two different domain-adaptation techniques which significantly improved translation quality when applied to monolingual corpora. We researched and applied a novel approach for finding the best hyperparameters for ensembling a number of translation models. All techniques combined gave significant improvement - up to +8 BLEU over baseline results. The quality of the models has been confirmed by the human evaluation where SRPOL models scored best for all 5 manually evaluated languages.
|
['Paweł Przybysz', 'Marcin Chochowski', 'Marcin Szymański', 'Adam Dobrowolski']
| null | null | null | null |
acl-wat-2021-8
|
['transliteration']
|
['natural-language-processing']
|
[-3.93300727e-02 -1.57812595e-01 -2.86688730e-02 -2.79113322e-01
-1.41127467e+00 -1.09878409e+00 1.01858890e+00 -1.22313976e-01
-5.54148853e-01 1.30573189e+00 2.76555240e-01 -9.47234392e-01
1.55271128e-01 -3.45941335e-01 -6.56526327e-01 -2.80562162e-01
1.91481590e-01 1.25440955e+00 -4.15254757e-02 -7.38993406e-01
2.66769260e-01 3.02666038e-01 -5.46488047e-01 3.44290882e-01
1.29493535e+00 9.74024460e-02 2.98231393e-01 7.97615170e-01
1.91014558e-02 2.99244046e-01 -6.08751118e-01 -7.13896155e-01
3.42915624e-01 -8.45570564e-01 -1.22420764e+00 -2.16290995e-01
3.66574138e-01 -1.39662961e-03 1.39875561e-01 8.18632483e-01
6.07955456e-01 -2.11216837e-01 6.89609766e-01 -5.30366838e-01
-7.53120780e-01 9.40619528e-01 -4.18602407e-01 2.85762906e-01
4.00778145e-01 -2.70223647e-01 7.53167212e-01 -1.06887734e+00
9.66643572e-01 1.26062036e+00 5.40405631e-01 3.25387239e-01
-1.18521202e+00 -4.29444402e-01 -4.88173127e-01 2.16015965e-01
-1.24616981e+00 -4.12980229e-01 1.26663148e-01 -2.65126109e-01
1.44909275e+00 2.19755054e-01 2.42000461e-01 8.93190503e-01
3.34726065e-01 4.58919108e-01 1.56015933e+00 -1.11290860e+00
-6.73905089e-02 6.60502374e-01 -1.75867513e-01 3.75404745e-01
2.34847572e-02 -1.64403707e-01 -2.14640856e-01 -1.82890251e-01
5.44021666e-01 -8.66844296e-01 1.79941833e-01 2.23464206e-01
-1.69604695e+00 6.51903749e-01 -1.20691605e-01 6.44399285e-01
-3.27217162e-01 -3.95929515e-01 6.58841610e-01 9.88712668e-01
6.66507423e-01 5.51187277e-01 -1.00176132e+00 -4.27757770e-01
-8.49072158e-01 -3.14947031e-02 7.52323687e-01 1.00711262e+00
6.14313304e-01 -2.42018048e-02 1.88417420e-01 1.35810876e+00
8.89645740e-02 9.03165877e-01 7.82983482e-01 -3.26343119e-01
1.04922712e+00 2.37315357e-01 2.86869168e-01 -2.63474077e-01
-1.22932941e-01 -2.39785537e-01 -3.95586669e-01 -2.63232619e-01
5.09296536e-01 -6.84624970e-01 -1.09191298e+00 1.50115764e+00
2.45346069e-01 -5.99996328e-01 6.02503955e-01 7.10788667e-01
2.98337102e-01 9.69515264e-01 -9.66358483e-02 -4.09704953e-01
1.16250801e+00 -1.20389426e+00 -6.75318837e-01 4.49302681e-02
7.64090538e-01 -1.80791450e+00 1.10772407e+00 3.66294503e-01
-1.15902030e+00 -7.24194288e-01 -9.84109342e-01 3.26597653e-02
-3.93418133e-01 6.00314915e-01 2.80139625e-01 7.68061876e-01
-1.10189891e+00 4.79140997e-01 -8.32479000e-01 -9.04244781e-01
-2.93615460e-01 5.53205848e-01 -5.55549204e-01 -4.57713529e-02
-1.36234665e+00 1.45716178e+00 7.12716579e-01 -2.16400802e-01
-6.42547905e-01 -1.59517098e-02 -4.17355269e-01 -5.40448606e-01
-2.76223868e-01 -3.60401392e-01 1.11948669e+00 -1.28650868e+00
-1.88013446e+00 1.05470443e+00 -1.07253850e-01 -4.80371982e-01
6.11051798e-01 -4.07442927e-01 -7.05477893e-01 -2.57717550e-01
2.15921327e-01 4.17056799e-01 4.77987468e-01 -7.67336965e-01
-7.46696949e-01 -2.90868789e-01 -3.13085735e-01 5.40257514e-01
-1.60514653e-01 6.50426745e-01 -2.85611361e-01 -6.08763576e-01
-2.14623418e-02 -1.16445506e+00 -3.86462733e-02 -1.20138943e+00
-2.31530964e-01 8.34928651e-04 6.65066898e-01 -1.37858844e+00
1.00270605e+00 -1.80088139e+00 4.50549334e-01 7.13395746e-03
-7.88049042e-01 3.98429483e-01 -2.46868894e-01 9.16856408e-01
-5.90313412e-02 -8.14097971e-02 -5.68815432e-02 2.36823335e-02
-5.80986701e-02 3.56935710e-01 -2.16548786e-01 2.90716738e-01
1.36472136e-01 8.86741817e-01 -7.20612466e-01 -3.55685115e-01
2.26424962e-01 3.96138430e-01 2.34522112e-02 8.73873942e-03
-1.09694555e-01 6.07596219e-01 -4.61145975e-02 6.39638901e-01
5.67630172e-01 4.94353235e-01 3.85315150e-01 1.86541602e-01
-3.11691225e-01 8.29012513e-01 -8.07399690e-01 1.78528452e+00
-6.59763813e-01 6.30448520e-01 -2.89399058e-01 -7.72762299e-01
1.16868484e+00 6.10237360e-01 1.28564117e-02 -7.52970457e-01
1.99826345e-01 9.94899392e-01 3.04549128e-01 -4.80605900e-01
6.96648777e-01 -1.81872606e-01 -9.98970717e-02 6.96139872e-01
4.31770384e-01 -3.10982198e-01 4.49453473e-01 -2.29641378e-01
5.09481609e-01 5.10874569e-01 4.74907309e-01 -6.73471749e-01
7.76993215e-01 5.47402561e-01 3.14146042e-01 3.47460896e-01
-2.56359689e-02 4.58468944e-01 1.27794579e-01 -4.38285172e-01
-1.57195115e+00 -9.65995967e-01 -1.07099682e-01 1.27349520e+00
-3.48461717e-01 -2.95964420e-01 -9.83115137e-01 -8.36884797e-01
-6.73858523e-01 9.18789864e-01 -2.06784010e-01 2.73580343e-01
-1.14605367e+00 -1.14308572e+00 6.45056546e-01 -9.38738808e-02
5.21750450e-01 -9.68337774e-01 1.36388451e-01 4.44575071e-01
-5.45568109e-01 -1.07253480e+00 -5.02437055e-01 3.04905981e-01
-1.15775239e+00 -4.13807333e-01 -1.05939198e+00 -1.16108990e+00
3.51323664e-01 -1.91196144e-01 1.20099127e+00 -5.42036414e-01
1.62661999e-01 -1.68168753e-01 -4.19515312e-01 -4.63912398e-01
-1.12284935e+00 6.35016441e-01 2.24315807e-01 -6.01045549e-01
5.53614378e-01 -3.42191815e-01 -8.65296945e-02 4.97764140e-01
-3.42862457e-01 2.76686121e-02 8.55963349e-01 7.92424619e-01
3.19227874e-01 -4.08756524e-01 5.68337619e-01 -9.50619400e-01
7.68347621e-01 -1.82215258e-01 -4.32323217e-01 6.21948540e-01
-6.31564558e-01 8.39068145e-02 6.06193066e-01 -5.63921809e-01
-1.13053942e+00 5.56301363e-02 -3.93557280e-01 4.99056756e-01
-1.25001431e-01 4.11862433e-01 -1.69945836e-01 -5.69546521e-02
1.04064000e+00 4.43408072e-01 -3.09020370e-01 -5.47456145e-01
4.50061291e-01 1.15988719e+00 3.61539602e-01 -8.01143229e-01
7.05084085e-01 -2.30684623e-01 -4.27256495e-01 -8.00246835e-01
-1.02234520e-01 -2.14657724e-01 -9.98321116e-01 1.29760608e-01
5.34507453e-01 -1.01532698e+00 4.09155399e-01 5.07658124e-01
-1.28711402e+00 -3.64475161e-01 3.32762152e-02 9.49140191e-01
-4.54455495e-01 3.31546932e-01 -8.74150038e-01 -3.36959779e-01
-8.26414049e-01 -1.13211775e+00 7.54798174e-01 -5.96352071e-02
-3.78047138e-01 -1.19678330e+00 5.95405519e-01 3.42698544e-01
5.84189236e-01 -1.88759014e-01 1.12338150e+00 -8.59472752e-01
-5.88620082e-03 -5.32110706e-02 -6.16602078e-02 5.68918824e-01
2.92234004e-01 -1.08393252e-01 -5.51064968e-01 -3.19501042e-01
-2.45197564e-01 -4.11815733e-01 2.79822439e-01 1.17631573e-02
-2.00820759e-01 -2.90315628e-01 -4.08535488e-02 5.81542440e-02
1.38409317e+00 4.43213254e-01 6.19608402e-01 7.27763355e-01
3.64842206e-01 2.80321687e-01 8.15934658e-01 -2.23765671e-01
3.72214824e-01 9.27145541e-01 -2.97578901e-01 -1.11861885e-01
-1.00354262e-01 -1.94964428e-02 7.79350877e-01 1.33304799e+00
-3.84490848e-01 -4.24409918e-02 -1.05797136e+00 7.41713703e-01
-1.60523641e+00 -4.04304832e-01 -3.07236850e-01 2.42487264e+00
1.29738975e+00 1.19252235e-01 2.76554912e-01 -1.94099858e-01
5.75041354e-01 -5.87735176e-01 2.05819622e-01 -1.11760926e+00
-1.36290625e-01 5.71317255e-01 6.05900645e-01 9.09482419e-01
-8.32598031e-01 1.58813560e+00 5.95251083e+00 8.51559699e-01
-1.30287910e+00 5.77780485e-01 4.01987672e-01 3.20523798e-01
-1.56633761e-02 1.47822782e-01 -8.52677703e-01 2.46635929e-01
1.58888721e+00 -2.01240614e-01 5.53320765e-01 4.94033307e-01
2.05601320e-01 5.43607138e-02 -8.54184031e-01 4.71761674e-01
8.20687041e-02 -9.18115377e-01 -1.07788322e-02 -1.69699773e-01
1.22009480e+00 6.99134886e-01 -2.80601531e-01 5.75924814e-01
3.76888931e-01 -6.67739987e-01 5.09068668e-01 -1.71372928e-02
7.22138166e-01 -8.81707370e-01 8.57937872e-01 4.42533225e-01
-5.17005563e-01 3.99446607e-01 -6.41567826e-01 1.35425932e-03
7.54331797e-02 2.17614025e-01 -1.27554142e+00 9.19645786e-01
3.10671896e-01 2.85988629e-01 -4.31524962e-01 6.57994449e-01
-3.22687238e-01 9.55142260e-01 -5.17262936e-01 -5.31681515e-02
4.85349029e-01 -3.45431209e-01 4.13454145e-01 1.63969827e+00
5.30256629e-01 -4.67515051e-01 -5.63656464e-02 7.43725747e-02
1.25470296e-01 8.50945354e-01 -4.28118736e-01 -1.35871410e-01
6.23578355e-02 1.05298626e+00 -6.75716996e-01 -5.58764577e-01
-2.24359646e-01 1.29648459e+00 4.00971882e-02 3.04855078e-01
-6.10243022e-01 -4.21204656e-01 2.71394163e-01 -5.94421774e-02
4.60678004e-02 -5.86017013e-01 -1.47183180e-01 -1.22915995e+00
-8.26515630e-02 -1.43534684e+00 2.38018900e-01 -4.22046781e-01
-8.56013536e-01 1.18412268e+00 5.16045913e-02 -1.16773343e+00
-6.93204939e-01 -5.75495660e-01 -3.42500210e-01 1.49572754e+00
-1.06184566e+00 -1.38812983e+00 5.57246447e-01 3.23188782e-01
8.64055574e-01 -6.47379398e-01 1.17246437e+00 6.58653319e-01
-2.23627955e-01 6.10360980e-01 6.05677426e-01 1.10754948e-02
1.13607585e+00 -1.31894183e+00 7.85595000e-01 9.17043030e-01
4.33260620e-01 7.33534873e-01 8.00838888e-01 -6.23246312e-01
-1.10777032e+00 -9.11827266e-01 1.78614998e+00 -3.87019694e-01
7.71174610e-01 -3.13381821e-01 -5.95681667e-01 7.84224808e-01
9.14369762e-01 -7.40127861e-01 5.83730578e-01 2.91212559e-01
1.10192988e-02 4.10688445e-02 -9.76075470e-01 5.24027646e-01
3.50823820e-01 -3.38466585e-01 -7.77034342e-01 7.37280786e-01
3.32165837e-01 -5.04409313e-01 -9.11042094e-01 8.17834884e-02
5.13677120e-01 -3.56346786e-01 7.38386393e-01 -7.04979539e-01
3.91424179e-01 -2.71007389e-01 -2.13739052e-01 -1.68454361e+00
-1.22999325e-02 -7.54876792e-01 4.27050471e-01 1.33646047e+00
1.04074836e+00 -6.40829325e-01 2.57910371e-01 -2.81496812e-02
2.56689433e-02 -1.93318665e-01 -8.62534761e-01 -8.06719661e-01
6.17686629e-01 -8.33496526e-02 2.57017493e-01 9.53868151e-01
1.64935157e-01 9.32106674e-01 -6.23532653e-01 -1.66730434e-01
2.71048129e-01 1.40387136e-02 8.30046177e-01 -7.35525668e-01
-4.75750089e-01 -2.52031744e-01 -2.28572935e-01 -8.00556302e-01
-1.21304736e-01 -1.05263734e+00 -1.21632695e-01 -1.60308731e+00
-4.77309003e-02 -3.85844290e-01 -2.55857054e-02 4.58632290e-01
-1.70026183e-01 6.21609032e-01 3.82778607e-02 3.54376167e-01
-1.67537019e-01 -5.67654818e-02 1.20616722e+00 -3.15563241e-03
-3.40856165e-01 1.43121988e-01 -3.74484271e-01 1.57071158e-01
9.92779315e-01 -5.67748010e-01 -2.99473464e-01 -8.98780942e-01
9.04790834e-02 1.11298606e-01 -4.08166200e-01 -7.08139241e-01
-1.97392985e-01 -8.36946890e-02 3.03794444e-01 -4.02571589e-01
3.12032383e-02 -5.81484139e-01 3.09444547e-01 6.58955514e-01
-9.76685584e-02 5.24345160e-01 4.42617923e-01 -2.51690418e-01
-2.02433363e-01 -2.60586798e-01 7.97189593e-01 -3.16813946e-01
-6.51749730e-01 -3.69815677e-01 -5.20866454e-01 -2.19062775e-01
8.24898660e-01 6.85456842e-02 -9.91585851e-02 -1.85565323e-01
-7.78396547e-01 -4.66757305e-02 3.59798670e-01 5.18730700e-01
-8.60326067e-02 -1.21849096e+00 -1.35040283e+00 2.65906185e-01
-8.17884281e-02 -7.82883763e-01 -3.84396076e-01 9.92256403e-01
-9.31788683e-01 7.26469278e-01 -6.58875644e-01 -5.18728495e-01
-1.44600213e+00 1.26975328e-01 2.37832695e-01 -5.98391831e-01
-1.45197853e-01 5.72319090e-01 -5.65773845e-01 -1.05186892e+00
-9.13323313e-02 -3.55269648e-02 -8.22395086e-02 -1.33906037e-01
2.15845257e-01 2.68261105e-01 5.55353701e-01 -8.84081781e-01
-3.07906896e-01 5.94877481e-01 -4.04875845e-01 -7.79351711e-01
9.30873692e-01 -3.59960347e-01 -3.55992198e-01 4.27743137e-01
9.50786054e-01 1.45742923e-01 -3.00276399e-01 -2.36321375e-01
3.28584552e-01 -1.34712383e-01 -3.36008251e-01 -1.48555672e+00
-2.35126391e-01 8.66806388e-01 6.43668592e-01 -1.58388987e-01
1.03080344e+00 -4.24524754e-01 6.70127928e-01 6.07931733e-01
6.90899670e-01 -1.30867279e+00 -6.37216806e-01 9.75206792e-01
7.09320724e-01 -1.23332345e+00 -1.81211531e-01 8.01449120e-02
-7.13021219e-01 1.41685712e+00 1.57522336e-01 5.37002757e-02
2.41328612e-01 1.35437146e-01 7.34506845e-01 4.20340925e-01
-6.96521103e-01 -3.17900516e-02 3.53792459e-01 5.31829000e-01
1.01940846e+00 2.99496263e-01 -1.35639572e+00 4.44017462e-02
-4.46191877e-01 -2.16880832e-02 4.61861819e-01 7.24464655e-01
-2.69396394e-01 -1.74357951e+00 -5.67660630e-01 -6.58585830e-03
-8.39959860e-01 -3.49944830e-01 -6.58976614e-01 1.06086636e+00
8.96276012e-02 9.27528203e-01 -3.42145205e-01 -3.25251430e-01
1.92071363e-01 3.82245213e-01 7.63022423e-01 -6.00216925e-01
-1.00684261e+00 5.89567661e-01 5.14349341e-01 2.03867242e-01
-1.48623616e-01 -7.96551108e-01 -8.63095880e-01 -2.03931853e-01
-3.51659894e-01 8.00398886e-01 1.17550027e+00 1.20347297e+00
-2.01191697e-02 4.33112159e-02 5.49876869e-01 -4.83225465e-01
-5.92421412e-01 -1.68863392e+00 -3.43410224e-02 -9.54232365e-03
-1.28771052e-01 5.63032106e-02 1.05424136e-01 3.71834189e-01]
|
[11.443184852600098, 10.46853256225586]
|
f0114910-f1e4-4755-b07f-04444cb6e987
|
polyu-cbs-comp-at-semeval-2021-task-1-lexical
| null | null |
https://aclanthology.org/2021.semeval-1.70
|
https://aclanthology.org/2021.semeval-1.70.pdf
|
PolyU CBS-Comp at SemEval-2021 Task 1: Lexical Complexity Prediction (LCP)
|
In this contribution, we describe the system presented by the PolyU CBS-Comp Team at the Task 1 of SemEval 2021, where the goal was the estimation of the complexity of words in a given sentence context. Our top system, based on a combination of lexical, syntactic, word embeddings and Transformers-derived features and on a Gradient Boosting Regressor, achieves a top correlation score of 0.754 on the subtask 1 for single words and 0.659 on the subtask 2 for multiword expressions.
|
['Chu-Ren Huang', 'Qin Lu', 'Wenjie Li', 'Emmanuele Chersoni', 'Jinghang Gu', 'Rong Xiang']
|
2021-08-01
| null | null | null |
semeval-2021
|
['lexical-complexity-prediction']
|
['natural-language-processing']
|
[-2.94366956e-01 -4.10916768e-02 -5.82193397e-02 -5.12488008e-01
-7.07317770e-01 -4.17633325e-01 7.37680316e-01 6.43794179e-01
-1.04072917e+00 5.46846449e-01 4.86713797e-01 -4.62998927e-01
1.52106553e-01 -3.06671649e-01 -3.38523626e-01 -1.46952942e-01
-1.37743264e-01 2.30821207e-01 1.52649134e-01 -5.66985071e-01
2.39125818e-01 1.51092848e-02 -1.07224166e+00 6.83630884e-01
2.23337278e-01 1.20692444e+00 3.01852554e-01 8.93145859e-01
-3.25740457e-01 8.00661862e-01 -5.47014892e-01 -8.01685333e-01
-1.66901216e-01 1.78272687e-02 -1.01842558e+00 -4.59911287e-01
3.88603091e-01 2.05323040e-01 -2.57859170e-01 8.20328593e-01
5.07858217e-01 4.13575411e-01 6.26944602e-01 -7.98548937e-01
-6.20703220e-01 1.08754408e+00 -1.18295718e-02 8.22875321e-01
6.06309533e-01 -1.05514430e-01 1.78906906e+00 -1.56803584e+00
7.01282740e-01 1.24581552e+00 6.07027054e-01 5.66789269e-01
-1.11825049e+00 -2.33958110e-01 3.25703830e-01 6.34121954e-01
-1.32042670e+00 -5.07622778e-01 5.15096784e-01 -5.04512429e-01
2.15843439e+00 1.28231451e-01 4.97700900e-01 1.27753222e+00
4.65106547e-01 7.79674828e-01 1.16563165e+00 -8.61651957e-01
2.46729925e-01 7.24425018e-02 8.12134504e-01 4.95424986e-01
-5.81651274e-03 -1.90575942e-01 -6.08863115e-01 -3.35704386e-02
-1.64704353e-01 -5.75487077e-01 9.37761888e-02 2.51798183e-01
-7.19211102e-01 1.04199624e+00 -4.28789817e-02 8.98015082e-01
-3.49646211e-01 1.81569591e-01 7.96013415e-01 3.35311323e-01
8.61535132e-01 5.75592756e-01 -1.05479991e+00 -5.58773279e-01
-6.29311740e-01 4.33749944e-01 7.54423201e-01 7.66709924e-01
2.97770113e-01 -1.40356466e-01 -4.35363024e-01 1.18383646e+00
1.62709519e-01 1.33387506e-01 7.85993874e-01 -2.90519774e-01
5.38993597e-01 4.05273706e-01 -1.70435309e-01 -7.59141028e-01
-7.84617245e-01 -2.53053874e-01 -1.02230728e-01 -2.87664562e-01
2.04115734e-01 -4.73688126e-01 -6.89880788e-01 1.85241747e+00
-5.26042329e-03 -1.57923192e-01 -5.70436418e-02 6.11681461e-01
1.08428323e+00 7.03798175e-01 5.53460419e-01 -1.98195145e-01
1.54504287e+00 -1.09520757e+00 -9.39434588e-01 -5.64616621e-01
9.51833606e-01 -8.98221910e-01 1.16294897e+00 4.97825056e-01
-1.21786630e+00 -6.86692894e-01 -1.09261990e+00 -3.22377592e-01
-7.71798313e-01 2.23715082e-02 2.88626701e-01 5.44742167e-01
-1.00316560e+00 1.80891037e-01 -4.24031794e-01 -2.78467655e-01
-2.88343638e-01 -2.63830461e-02 -3.84346217e-01 6.69424841e-03
-1.66550910e+00 1.44107068e+00 5.91284215e-01 -1.59983188e-01
-3.21779877e-01 -6.70733869e-01 -1.00901580e+00 -5.66680618e-02
1.15291514e-01 -2.73969710e-01 1.27907956e+00 -5.24311602e-01
-1.35674071e+00 1.04592824e+00 -2.16629878e-01 -6.37637675e-01
3.23633909e-01 -5.40837407e-01 -6.71362460e-01 -3.60772371e-01
-6.24978356e-02 2.66827732e-01 3.68180037e-01 -4.41451102e-01
-6.91207647e-01 -3.89337927e-01 2.39161029e-01 1.19973235e-01
-4.71553653e-01 5.60205758e-01 -1.76399395e-01 -5.82155228e-01
-5.40835619e-01 -7.12046862e-01 -2.25983202e-01 -9.32409942e-01
-9.37026665e-02 -9.03498828e-01 2.99586028e-01 -1.04788029e+00
1.71479881e+00 -2.03974319e+00 4.21435386e-01 -1.08120970e-01
-1.16421049e-02 2.25585282e-01 -3.11658859e-01 6.44406736e-01
-4.56501961e-01 1.29154176e-01 7.38473907e-02 -6.27675653e-01
1.89671919e-01 -7.79300183e-02 -1.02378137e-01 1.85287893e-01
4.17001516e-01 9.63563800e-01 -1.01045525e+00 -3.22554201e-01
7.74656460e-02 9.96902436e-02 -4.20514703e-01 3.17566663e-01
-1.66198730e-01 -3.73962671e-01 2.52357051e-02 4.11197469e-02
3.19041967e-01 2.69470423e-01 4.39484477e-01 4.99177948e-02
-2.46827722e-01 8.54608417e-01 -8.04395556e-01 1.61155295e+00
-1.10008967e+00 7.89526105e-01 -1.66212752e-01 -7.77529120e-01
9.00569379e-01 3.07964593e-01 2.44968131e-01 -6.46009147e-01
2.60783523e-01 -7.31841289e-03 1.47137001e-01 -6.80165946e-01
8.19906116e-01 -1.26872852e-01 -4.50272143e-01 1.90017726e-02
4.95887339e-01 -1.15538515e-01 5.83655059e-01 2.85100713e-02
1.45325041e+00 -1.00951143e-01 6.68847799e-01 -6.10106647e-01
9.30564046e-01 -3.38591903e-01 1.91989511e-01 2.90578544e-01
-2.35657766e-01 3.00356627e-01 7.03437567e-01 -5.90206683e-01
-1.09748459e+00 -6.57589495e-01 -2.98355877e-01 1.65966511e+00
-3.59813154e-01 -1.03558779e+00 -7.23104894e-01 -7.38213599e-01
-1.16553679e-02 1.23342741e+00 -6.26311183e-01 -7.76173547e-02
-6.93269670e-01 -6.33039236e-01 5.48125625e-01 6.36155665e-01
-3.08249533e-01 -1.17805946e+00 -2.39800811e-01 5.04054546e-01
-2.45611936e-01 -1.57583141e+00 -5.42962670e-01 4.23283815e-01
-4.88370568e-01 -7.39060879e-01 -1.75361037e-01 -8.65069687e-01
-3.37948650e-02 -3.91829878e-01 1.41721141e+00 -9.52681825e-02
-3.34712714e-01 5.73319234e-02 -7.41679609e-01 -4.77566123e-01
-3.12730074e-01 2.77955294e-01 1.75425634e-01 -1.69766605e-01
8.13110232e-01 -1.51533410e-01 -4.53760475e-02 -1.40207514e-01
-5.78650951e-01 -3.02564204e-01 2.07570285e-01 9.71096873e-01
1.99049130e-01 -5.51166058e-01 5.30684054e-01 -8.45488012e-01
1.07029629e+00 -6.14818990e-01 -3.65493596e-01 2.74993062e-01
-6.19296134e-01 -2.24644020e-02 6.64888322e-01 -3.97540927e-01
-7.12871969e-01 -8.18678513e-02 -7.57059515e-01 5.62855266e-02
-5.45828678e-02 6.74486279e-01 -1.65063292e-01 4.75278795e-01
8.23237896e-01 2.13467851e-02 -5.46520889e-01 -4.99866843e-01
6.40574872e-01 8.31434131e-01 1.50990307e-01 -7.28557706e-01
2.06184074e-01 -4.25171882e-01 -2.81568259e-01 -1.29033530e+00
-9.86927330e-01 -6.33884430e-01 -5.35354853e-01 -1.41168013e-01
9.63866711e-01 -8.65949690e-01 -4.23619032e-01 3.76405358e-01
-1.68405068e+00 -2.07586542e-01 -1.95443347e-01 4.27170634e-01
-3.21251988e-01 4.44891721e-01 -5.67190289e-01 -8.76872122e-01
-5.45710027e-01 -8.08160841e-01 9.91470456e-01 -2.75355548e-01
-7.97949195e-01 -1.20016396e+00 3.13927412e-01 3.01598728e-01
4.35599953e-01 -9.83275846e-02 1.10233939e+00 -1.14870703e+00
3.66565764e-01 -3.37299287e-01 -1.05685517e-01 6.64641201e-01
-1.91579238e-01 -1.15249299e-01 -9.64541256e-01 -2.02725112e-01
-1.16467439e-01 -5.02071738e-01 8.59783053e-01 2.94606358e-01
9.93907630e-01 -1.31841023e-02 1.35394379e-01 1.89985946e-01
1.24662507e+00 -3.81097384e-02 2.97974616e-01 1.76274106e-01
3.81046295e-01 5.57072043e-01 4.43045855e-01 5.04770577e-01
3.50092292e-01 8.68105412e-01 1.60588846e-01 5.42941988e-01
-1.79952443e-01 -1.10247247e-02 6.74112022e-01 1.33389199e+00
2.26116460e-02 -3.40537339e-01 -1.18850911e+00 7.86054850e-01
-1.71820974e+00 -4.97360080e-01 -4.65258420e-01 1.68667459e+00
7.91182578e-01 4.34617370e-01 6.01351000e-02 -2.93700323e-02
5.92870712e-01 5.47028363e-01 1.23435505e-01 -1.23377132e+00
-2.11545065e-01 5.62799573e-01 2.29992092e-01 8.98385644e-01
-1.13565278e+00 1.33813643e+00 7.24216843e+00 1.04392695e+00
-6.45314395e-01 5.07547140e-01 4.13561434e-01 -8.04969966e-02
-1.46999191e-02 -2.96377689e-01 -1.19083834e+00 5.45483530e-01
1.77069521e+00 -3.64395559e-01 2.46602193e-01 9.32415128e-01
-8.48154575e-02 -7.06042647e-02 -1.25171673e+00 7.97178328e-01
4.75636929e-01 -1.06721210e+00 -2.19556391e-01 -2.28432447e-01
4.41271454e-01 3.76622230e-01 -2.90716261e-01 8.64212453e-01
1.87487155e-01 -1.00209916e+00 7.97493219e-01 3.92470300e-01
5.87236702e-01 -8.37739289e-01 1.03122079e+00 3.85197908e-01
-1.01316988e+00 1.69917718e-02 -3.15627068e-01 -5.20808160e-01
1.91379860e-01 5.88550210e-01 -8.01909745e-01 3.25233072e-01
3.46329927e-01 5.06935239e-01 -4.89634573e-01 4.46008325e-01
-5.89002967e-01 8.27618062e-01 -7.89706334e-02 -7.38600552e-01
3.13747823e-01 2.14408457e-01 4.84131694e-01 2.16434383e+00
-1.17418468e-01 1.85357071e-02 1.79212779e-01 2.52763510e-01
-2.08386153e-01 8.48143101e-01 -4.30293471e-01 5.65459058e-02
3.07418764e-01 1.42254055e+00 -1.34894446e-01 -3.35500211e-01
-4.87216532e-01 6.68440223e-01 9.21375692e-01 -7.48293474e-02
-6.72441244e-01 -4.67263550e-01 7.41833389e-01 -1.92181721e-01
3.48432362e-01 -5.11411607e-01 -3.90137732e-01 -1.12435544e+00
1.29262239e-01 -5.91831684e-01 3.34363222e-01 -5.26241779e-01
-1.56197500e+00 9.75407839e-01 -1.78502575e-02 -4.16097581e-01
-3.61786872e-01 -1.22184706e+00 -5.13321459e-01 1.14907658e+00
-1.41676319e+00 -8.34407449e-01 1.87428191e-01 1.97317854e-01
9.22528982e-01 -4.03513461e-01 1.06702518e+00 2.08251312e-01
-5.09630084e-01 7.45847046e-01 -6.84588328e-02 1.35908797e-01
5.19774914e-01 -1.41039288e+00 9.25697029e-01 6.74645722e-01
3.74690980e-01 3.95350069e-01 8.39434147e-01 -4.10355568e-01
-1.13418806e+00 -8.82350266e-01 1.95802140e+00 -8.23911726e-01
1.26672506e+00 -6.54503167e-01 -6.47708118e-01 4.53755170e-01
3.24577004e-01 1.03214726e-01 6.75049901e-01 5.98926246e-01
-4.53682959e-01 1.65911302e-01 -7.17162371e-01 3.30372542e-01
9.15811062e-01 -7.89109409e-01 -1.24100161e+00 5.76783419e-01
9.10944760e-01 -4.26066339e-01 -1.03100407e+00 2.01464966e-01
5.47314942e-01 -4.51776057e-01 8.22777510e-01 -1.24290013e+00
9.25983846e-01 3.86996418e-01 -5.92397153e-01 -1.47570705e+00
-6.15476131e-01 -1.80453584e-01 -3.24597806e-01 9.63866532e-01
7.81433523e-01 -1.65433601e-01 3.51649612e-01 4.39991981e-01
-2.13164866e-01 -9.45996344e-01 -1.45402586e+00 -1.02492058e+00
4.62642252e-01 -1.07266438e+00 1.70333862e-01 6.65401340e-01
3.96473140e-01 9.67727780e-01 -1.73531786e-01 -2.62965441e-01
3.19313928e-02 -4.65108365e-01 2.70770758e-01 -8.96129727e-01
-3.04616541e-01 -4.71717954e-01 -6.59041822e-01 -6.47959411e-01
6.94987953e-01 -1.12224734e+00 3.87133062e-02 -1.25646102e+00
-3.83765716e-03 3.14897895e-01 -6.28097117e-01 3.06340337e-01
-3.43694061e-01 -9.59627330e-02 3.26161504e-01 -5.28409183e-01
-6.20800018e-01 4.78129208e-01 4.83254433e-01 -8.63305628e-02
1.45251647e-01 -3.49771291e-01 -4.77972388e-01 6.69744849e-01
7.01784611e-01 -3.97528827e-01 2.36008823e-01 -4.99098033e-01
5.60287237e-01 -3.24290663e-01 -1.37949526e-01 -6.40945375e-01
-8.17338154e-02 -4.13704803e-03 6.56968728e-02 -4.48785424e-01
5.00624776e-01 -5.92725992e-01 -3.82436544e-01 3.91921490e-01
-5.76860547e-01 5.09291947e-01 4.79936808e-01 2.26750284e-01
-2.75455505e-01 -6.70051992e-01 5.19658148e-01 7.57753551e-02
-9.06764150e-01 -2.40790546e-01 -6.39017165e-01 5.30445755e-01
9.21833038e-01 2.59595424e-01 -1.40958652e-01 -5.25930896e-02
-8.23455513e-01 1.47390231e-01 -2.46596158e-01 9.23125863e-01
4.31446493e-01 -1.24742556e+00 -1.03261518e+00 -2.18568370e-02
4.36970294e-01 -7.75304556e-01 7.61993900e-02 5.00074387e-01
-2.64188945e-01 5.68256676e-01 -6.48593083e-02 -1.26977429e-01
-1.54463601e+00 3.56676370e-01 2.01496631e-01 -7.43160665e-01
-2.33593374e-01 1.17454612e+00 -3.82974476e-01 -3.66722852e-01
4.81541734e-03 -4.85354394e-01 -4.11848426e-01 4.80751663e-01
7.76847601e-01 4.95150566e-01 6.99212730e-01 -7.30566025e-01
-6.45452261e-01 3.71602654e-01 -3.25499147e-01 -2.41109103e-01
1.44890082e+00 1.27435505e-01 -2.56792516e-01 7.88627088e-01
1.46683753e+00 -1.07292086e-01 -4.22179550e-01 -4.66666251e-01
6.78591430e-01 -4.07737400e-03 3.77502292e-01 -8.83736193e-01
-4.40288723e-01 9.37055826e-01 3.20078731e-01 3.27193320e-01
5.29777586e-01 3.41745913e-02 8.41548800e-01 6.01469398e-01
9.25108567e-02 -1.59396076e+00 -2.33576596e-01 1.45431459e+00
1.09998858e+00 -1.04229188e+00 -9.53284465e-03 -1.51324168e-01
-5.90296984e-01 1.37399662e+00 4.56412911e-01 -3.33149850e-01
8.14668119e-01 2.81709313e-01 -1.89499632e-01 -4.40480970e-02
-1.33636534e+00 -1.23931080e-01 5.59371471e-01 1.98902488e-01
9.78352070e-01 3.93394887e-01 -1.15474498e+00 1.03251290e+00
-3.99791300e-01 -2.46068031e-01 3.89397174e-01 6.84231341e-01
-4.75714564e-01 -1.17742765e+00 9.22462791e-02 2.04326913e-01
-7.38245308e-01 -5.27427375e-01 -4.98431385e-01 5.65276444e-01
1.34871602e-01 9.93470073e-01 7.67452642e-02 -4.99816120e-01
7.83382297e-01 6.65061653e-01 4.31198061e-01 -8.53067458e-01
-9.47945654e-01 -3.21964324e-01 8.09151411e-01 -5.57717383e-01
-5.14265113e-02 -7.59296834e-01 -9.12338793e-01 6.20332025e-02
-4.76575643e-02 1.17664136e-01 1.06847548e+00 1.12642086e+00
-3.18441391e-02 6.52330041e-01 5.16216457e-01 -5.25818527e-01
-7.83902705e-01 -1.44731700e+00 -4.28266674e-01 3.77899230e-01
-2.42348872e-02 -3.07915300e-01 -3.37752581e-01 -1.92285553e-01]
|
[10.56201171875, 10.34835147857666]
|
41f0b08e-5d03-4591-8467-aeffecf677d9
|
exploiting-the-intrinsic-neighborhood
|
2110.04202
| null |
https://arxiv.org/abs/2110.04202v3
|
https://arxiv.org/pdf/2110.04202v3.pdf
|
Exploiting the Intrinsic Neighborhood Structure for Source-free Domain Adaptation
|
Domain adaptation (DA) aims to alleviate the domain shift between source domain and target domain. Most DA methods require access to the source data, but often that is not possible (e.g. due to data privacy or intellectual property). In this paper, we address the challenging source-free domain adaptation (SFDA) problem, where the source pretrained model is adapted to the target domain in the absence of source data. Our method is based on the observation that target data, which might no longer align with the source domain classifier, still forms clear clusters. We capture this intrinsic structure by defining local affinity of the target data, and encourage label consistency among data with high local affinity. We observe that higher affinity should be assigned to reciprocal neighbors, and propose a self regularization loss to decrease the negative impact of noisy neighbors. Furthermore, to aggregate information with more context, we consider expanded neighborhoods with small affinity values. In the experimental results we verify that the inherent structure of the target features is an important source of information for domain adaptation. We demonstrate that this local structure can be efficiently captured by considering the local neighbors, the reciprocal neighbors, and the expanded neighborhood. Finally, we achieve state-of-the-art performance on several 2D image and 3D point cloud recognition datasets. Code is available in https://github.com/Albert0147/SFDA_neighbors.
|
['Shangling Jui', 'Luis Herranz', 'Joost Van de Weijer', 'Yaxing Wang', 'Shiqi Yang']
|
2021-10-08
| null |
http://proceedings.neurips.cc/paper/2021/hash/f5deaeeae1538fb6c45901d524ee2f98-Abstract.html
|
http://proceedings.neurips.cc/paper/2021/file/f5deaeeae1538fb6c45901d524ee2f98-Paper.pdf
|
neurips-2021-12
|
['source-free-domain-adaptation']
|
['computer-vision']
|
[ 4.67398949e-02 -1.74541265e-01 -4.63314831e-01 -5.14913738e-01
-6.88831925e-01 -7.42250264e-01 4.03845131e-01 1.24871977e-01
-2.75161296e-01 7.80384660e-01 2.77912647e-01 1.83784127e-01
-1.09964453e-01 -7.87235677e-01 -8.25291038e-01 -9.22669411e-01
2.80291140e-01 3.91515255e-01 1.98886558e-01 -3.94317973e-03
5.03241085e-02 5.98676085e-01 -1.32053351e+00 2.95017093e-01
1.04155850e+00 1.14209437e+00 7.71148354e-02 -1.08825073e-01
-1.94975287e-01 3.99928451e-01 -6.40039325e-01 2.64700688e-02
5.36547661e-01 -2.40227610e-01 -5.96255302e-01 1.58969551e-01
4.16300386e-01 -6.31848052e-02 -8.66888985e-02 1.18296611e+00
4.23381925e-01 2.63857186e-01 7.85975218e-01 -1.38365352e+00
-9.27636027e-01 3.59118767e-02 -6.34717345e-01 8.13307688e-02
6.97448105e-02 -3.93170305e-02 8.20708573e-01 -1.09380567e+00
7.89111018e-01 8.97982538e-01 4.67952549e-01 5.21588266e-01
-1.34377015e+00 -1.08372712e+00 4.51305479e-01 1.75500184e-01
-1.64859855e+00 -3.43390346e-01 1.12694263e+00 -3.94600630e-01
5.02641797e-01 3.28742340e-02 3.23212385e-01 1.30639780e+00
-1.95363969e-01 7.03770041e-01 9.22738492e-01 -3.44061583e-01
4.04727429e-01 4.30020720e-01 2.87271053e-01 1.31238908e-01
2.44472697e-01 1.38229728e-01 -5.38060963e-01 -4.18674022e-01
6.74211562e-01 1.76960394e-01 -4.08476382e-01 -9.34626997e-01
-1.07908297e+00 7.96358645e-01 7.03827918e-01 3.07796299e-01
-2.96502620e-01 -4.67686862e-01 1.44297957e-01 2.92343736e-01
4.75009143e-01 1.89764410e-01 -5.85835338e-01 1.96803495e-01
-4.29499418e-01 -1.33298233e-01 5.63604891e-01 1.26135397e+00
1.16887379e+00 -3.14361632e-01 1.85537651e-01 1.16998684e+00
1.95372850e-01 6.17760003e-01 5.37585378e-01 -8.28029633e-01
4.19221908e-01 8.36552620e-01 2.09051013e-01 -9.32386279e-01
-1.66242048e-01 -4.98858631e-01 -1.01142740e+00 2.14498311e-01
5.64734519e-01 5.26805315e-03 -8.43063235e-01 1.94653535e+00
6.88722551e-01 4.92164135e-01 1.97261140e-01 1.02703393e+00
6.60138309e-01 5.98932505e-01 -5.69403498e-03 -1.05809398e-01
1.05701792e+00 -6.26252234e-01 -4.37611818e-01 -3.08180988e-01
5.85955501e-01 -6.13662004e-01 1.24860311e+00 6.01595081e-02
-3.15031201e-01 -4.92204159e-01 -1.00264204e+00 4.55950871e-02
-4.64225382e-01 1.47628546e-01 1.92903042e-01 3.43212038e-01
-5.19702792e-01 3.36967736e-01 -5.93357146e-01 -5.05440891e-01
6.02197289e-01 4.30086732e-01 -6.21709526e-01 -2.75690556e-01
-1.26002550e+00 5.01788735e-01 3.04067731e-01 -3.39783698e-01
-5.04312873e-01 -8.86244655e-01 -7.05394208e-01 -1.81717649e-01
4.07567948e-01 -5.55987120e-01 8.21231723e-01 -1.08814323e+00
-1.13371825e+00 9.61132646e-01 -3.15516800e-01 -1.24958307e-01
2.52955228e-01 -1.38179407e-01 -6.44209266e-01 3.44007066e-03
2.88682759e-01 4.94178295e-01 9.69344139e-01 -1.50732648e+00
-7.67193079e-01 -6.05565608e-01 -2.73615181e-01 1.20850109e-01
-7.02104270e-01 -4.24920470e-01 -5.05071998e-01 -7.20027924e-01
1.52256310e-01 -1.08277619e+00 -8.88257939e-03 1.92558214e-01
-2.18351379e-01 -2.01559722e-01 9.69996989e-01 -2.85091758e-01
1.12056339e+00 -2.61405945e+00 -5.80284968e-02 6.15143597e-01
1.45605609e-01 2.23322868e-01 -2.41705090e-01 1.00267194e-01
-1.77062497e-01 -1.02757201e-01 -4.57353860e-01 -1.24542765e-01
-1.09277338e-01 4.15977210e-01 -4.19591665e-01 5.09722352e-01
1.10727459e-01 5.83694220e-01 -7.84752786e-01 -2.98432410e-01
6.54886514e-02 4.45980579e-01 -4.58651274e-01 9.61484909e-02
3.13827656e-02 6.12172127e-01 -7.15043366e-01 6.57450199e-01
1.06592011e+00 -5.34464359e-01 -3.39507945e-02 -3.20040941e-01
1.81022912e-01 8.29851106e-02 -1.38654852e+00 1.69020760e+00
-3.19664061e-01 2.40569666e-01 2.03543261e-01 -9.43042815e-01
1.29479647e+00 9.86470748e-03 7.54218459e-01 -6.24078691e-01
-1.86153520e-02 4.57186908e-01 -2.54640281e-01 -9.44208447e-03
2.29938343e-01 -1.86250005e-02 -4.21288572e-02 2.74005443e-01
-1.27572790e-01 2.34814197e-01 -2.89928049e-01 6.09512180e-02
8.14438820e-01 -1.15655169e-01 4.46432441e-01 -2.62584269e-01
5.18446922e-01 2.37602666e-01 9.39500928e-01 4.56591010e-01
-4.02989149e-01 7.49788761e-01 1.49470240e-01 -1.97557539e-01
-1.03630614e+00 -1.23649681e+00 -3.26469034e-01 9.15962160e-01
5.60493946e-01 -1.01087965e-01 -5.47492087e-01 -8.78349245e-01
3.38112801e-01 5.69880009e-01 -6.70751989e-01 -5.53329766e-01
-4.92610961e-01 -5.17696202e-01 3.80622357e-01 4.80241716e-01
5.30987918e-01 -7.03133643e-01 -4.93836999e-02 1.35663718e-01
-1.73139974e-01 -9.35379803e-01 -7.86175370e-01 1.62802055e-01
-8.46893013e-01 -9.92006540e-01 -7.47313499e-01 -8.70711982e-01
8.30626965e-01 4.22557533e-01 9.34196532e-01 -2.21468717e-01
2.74327427e-01 3.49384248e-01 -4.11691993e-01 -1.89636379e-01
-2.89641291e-01 2.83789068e-01 2.47678533e-01 2.92202175e-01
9.47509110e-01 -7.95312643e-01 -5.09231389e-01 6.97579443e-01
-8.34982216e-01 -3.67716968e-01 5.52521050e-01 9.75218117e-01
1.01954508e+00 -4.68258299e-02 6.73012853e-01 -1.01580739e+00
4.94751960e-01 -7.05356658e-01 -5.44070184e-01 1.06204726e-01
-6.48016393e-01 -1.08110994e-01 6.67803049e-01 -7.67868340e-01
-9.85426903e-01 4.44267541e-01 1.22395702e-01 -8.90497923e-01
-5.43485820e-01 2.57284999e-01 -6.01513803e-01 -1.28864229e-01
9.63145792e-01 1.68393299e-01 1.57214150e-01 -6.41872108e-01
2.50264943e-01 8.66478682e-01 4.37337935e-01 -7.56436825e-01
8.67053628e-01 6.74268603e-01 -1.72127441e-01 -7.82882571e-01
-8.57486725e-01 -8.22172165e-01 -7.95005798e-01 2.18210593e-01
3.95173430e-01 -1.07621205e+00 -1.62628993e-01 2.82355815e-01
-9.50988114e-01 -1.27168939e-01 -4.87523556e-01 3.96791488e-01
-3.32951695e-01 4.14518952e-01 -1.38713703e-01 -3.49666893e-01
-7.90619850e-02 -8.74222577e-01 8.87221158e-01 1.86505020e-01
-1.60298198e-01 -9.41421807e-01 8.86005089e-02 8.03652033e-02
1.44117132e-01 2.87228465e-01 7.04406977e-01 -9.26011741e-01
-4.08846498e-01 -9.79013965e-02 -3.37040305e-01 5.39842665e-01
6.82051718e-01 -3.51108849e-01 -9.21649277e-01 -3.73557895e-01
8.72720629e-02 2.67367642e-02 7.30345786e-01 3.28803807e-01
9.90380049e-01 -2.61340618e-01 -5.44959664e-01 6.53511941e-01
1.29120851e+00 9.62175429e-02 4.24960405e-01 3.36508691e-01
8.61211061e-01 5.36349237e-01 8.34043324e-01 5.31157613e-01
2.68632054e-01 7.10863888e-01 2.08949089e-01 -7.09199980e-02
-1.01155110e-01 -3.70271504e-01 2.76715100e-01 4.99096811e-01
3.96827638e-01 -1.27801985e-01 -1.04143763e+00 8.16894352e-01
-1.95492530e+00 -6.71334147e-01 -1.28144607e-01 2.41933799e+00
8.38832259e-01 -5.97791336e-02 1.70160145e-01 -2.18009397e-01
9.68345046e-01 -1.47282749e-01 -9.12579060e-01 1.47159368e-01
-3.77081752e-01 -1.41780414e-02 6.25930130e-01 4.21793789e-01
-1.24110496e+00 8.41394305e-01 5.19955730e+00 1.00474524e+00
-1.10558236e+00 8.09543952e-02 3.99642855e-01 -1.95527807e-01
-2.07305923e-01 -1.06038846e-01 -8.57749641e-01 6.52809024e-01
5.50425649e-01 -3.36715847e-01 3.12468529e-01 1.20406115e+00
2.00499836e-02 2.35305086e-01 -1.15397799e+00 1.01082945e+00
-1.65630117e-01 -1.20829761e+00 3.25693563e-02 3.38251144e-01
8.04793775e-01 1.29415959e-01 5.72716519e-02 8.35671946e-02
3.12844962e-01 -6.46357894e-01 4.59883064e-01 2.15019822e-01
8.11384618e-01 -7.95335710e-01 5.87276399e-01 4.23068225e-01
-1.18074930e+00 -1.93541929e-01 -6.55128837e-01 2.56944895e-01
-1.72943860e-01 8.85213494e-01 -6.69235766e-01 5.39902806e-01
9.55387294e-01 1.16591823e+00 -4.16262567e-01 8.88873160e-01
-9.85850766e-02 4.62093711e-01 -6.66212559e-01 2.41307333e-01
-4.23453785e-02 -2.80610353e-01 7.62609601e-01 8.93224180e-01
3.39094430e-01 2.14934558e-01 2.28091687e-01 8.79678071e-01
-2.49904335e-01 2.52540559e-01 -8.32116902e-01 3.18858922e-01
9.26458895e-01 8.45632970e-01 -2.31326684e-01 -1.97925046e-01
-4.90002245e-01 1.08789361e+00 3.67456704e-01 5.27274370e-01
-5.78247488e-01 -2.67270595e-01 1.19676256e+00 1.72003657e-01
3.81621838e-01 -3.65027152e-02 -5.79995394e-01 -1.30950367e+00
3.01877767e-01 -8.41810346e-01 7.54788756e-01 -4.81336236e-01
-1.92807543e+00 4.04255718e-01 -1.22753426e-01 -1.74225760e+00
7.47439489e-02 -4.55069572e-01 -3.47103447e-01 8.79258573e-01
-1.53902936e+00 -1.08040512e+00 -4.31499273e-01 1.07452047e+00
6.80576637e-02 -2.19165877e-01 7.03160822e-01 6.09072149e-01
-3.56278390e-01 8.86735082e-01 5.62030792e-01 1.73632085e-01
1.21529067e+00 -9.19487715e-01 2.10827533e-02 6.76217914e-01
5.62331043e-02 7.94290662e-01 3.10911834e-01 -6.64133668e-01
-9.50852871e-01 -1.42437065e+00 6.70570374e-01 -5.91050804e-01
5.36390066e-01 -3.83116096e-01 -1.42238152e+00 5.58652163e-01
-3.30803633e-01 3.29986453e-01 8.97643805e-01 1.08757660e-01
-6.14831924e-01 -4.42027539e-01 -1.42007494e+00 3.78060400e-01
1.09906244e+00 -4.77845311e-01 -5.67151904e-01 2.57302672e-01
6.93763137e-01 -2.61963099e-01 -1.00107288e+00 3.57116282e-01
2.09375679e-01 -6.94986045e-01 1.14242959e+00 -4.22251612e-01
7.80027658e-02 -7.33462274e-01 -3.41337353e-01 -1.37691700e+00
-5.40694714e-01 -1.46410808e-01 -4.58919397e-03 1.49025929e+00
4.01232302e-01 -7.76948512e-01 1.07048082e+00 7.19350278e-01
-2.04587318e-02 -3.33500803e-01 -1.28149736e+00 -1.14024103e+00
4.49340373e-01 -1.79077834e-01 9.77164090e-01 1.34380460e+00
-1.52758762e-01 2.71729022e-01 -3.64639759e-01 5.49159825e-01
7.35631466e-01 3.90075564e-01 8.66607487e-01 -1.36730695e+00
3.08925416e-02 -2.39283457e-01 -3.03301603e-01 -1.22411752e+00
3.11222672e-01 -8.66765201e-01 -1.07114054e-01 -1.18746567e+00
-1.54732876e-02 -8.18585813e-01 -6.45484209e-01 6.17918432e-01
-1.25082389e-01 2.46231809e-01 5.61569184e-02 7.20007718e-01
-4.59195942e-01 7.41685092e-01 1.09875691e+00 -2.15541825e-01
-4.54787999e-01 -3.44655663e-02 -8.32804859e-01 4.44157988e-01
9.83509362e-01 -7.05114245e-01 -3.71285647e-01 -3.76893610e-01
-3.41977149e-01 -5.37098110e-01 3.29288721e-01 -9.73878980e-01
3.19031507e-01 -4.01068032e-01 4.15726066e-01 -3.36562455e-01
2.73355752e-01 -1.33963537e+00 2.40305319e-01 6.37539998e-02
-2.00718597e-01 -5.01123428e-01 1.65479407e-01 7.83547282e-01
-4.02354509e-01 -4.75997711e-03 1.11683130e+00 1.87236428e-01
-9.70021546e-01 4.78052229e-01 1.06152065e-01 5.16252443e-02
1.08507991e+00 -4.05070990e-01 -2.32312769e-01 -1.70703217e-01
-6.84058249e-01 3.24479759e-01 9.50931311e-01 5.25728226e-01
5.20105898e-01 -1.72057617e+00 -5.60457289e-01 4.00601059e-01
6.70281947e-01 2.07410172e-01 1.64384291e-01 5.04506886e-01
9.52061564e-02 5.90624996e-02 -2.13721052e-01 -7.72781491e-01
-1.13081610e+00 7.28631318e-01 3.22234750e-01 4.94588204e-02
-6.62435889e-01 7.96634972e-01 5.50778747e-01 -7.39138782e-01
1.81307614e-01 -1.61018118e-01 -4.23978306e-02 -5.76823838e-02
4.44367647e-01 2.19376773e-01 2.99530700e-02 -7.02677131e-01
-6.87065899e-01 7.53780544e-01 -1.99595705e-01 2.98539937e-01
1.21688259e+00 -3.70178223e-01 8.52387547e-02 2.14158505e-01
1.39329529e+00 1.75104275e-01 -1.49579132e+00 -7.39625812e-01
3.38971466e-02 -7.81814992e-01 -1.61909312e-01 -7.29841411e-01
-1.14428413e+00 6.41953290e-01 8.18173230e-01 -1.50845602e-01
1.37114549e+00 2.17080429e-01 8.03094208e-01 2.80188531e-01
3.49647045e-01 -1.27502596e+00 -8.01324397e-02 4.52528000e-01
7.53908634e-01 -1.45861363e+00 -1.11199789e-01 -6.02297187e-01
-7.38686442e-01 8.11456442e-01 7.79603362e-01 -1.28792122e-01
8.55074942e-01 3.83339357e-03 1.89606771e-01 4.33649160e-02
-4.59799856e-01 -6.99007437e-02 2.73180425e-01 1.02686560e+00
-6.51985854e-02 2.76677981e-02 3.18857282e-02 8.31954062e-01
7.70846009e-02 -3.82886976e-02 2.49902546e-01 8.14412773e-01
-3.45352441e-01 -1.39121783e+00 -5.62296271e-01 3.26000184e-01
-1.10163942e-01 1.38171613e-01 -6.42353594e-01 6.77008212e-01
2.82317966e-01 9.01679277e-01 1.12619951e-01 -3.17297548e-01
6.69093788e-01 -2.66152676e-02 -1.63743235e-02 -5.79533160e-01
-2.11334929e-01 1.75591111e-02 -2.71009058e-01 -4.20052588e-01
-2.78865397e-01 -6.58347905e-01 -1.34142005e+00 -3.05963546e-01
-1.94310054e-01 2.16939852e-01 3.19379002e-01 5.72982669e-01
9.56694245e-01 4.85258289e-02 7.54669726e-01 -3.87331635e-01
-4.13449079e-01 -6.13350630e-01 -7.56529868e-01 8.34955335e-01
4.07744646e-01 -9.01233435e-01 -5.03214419e-01 1.11045673e-01]
|
[10.335724830627441, 3.0685224533081055]
|
302caf28-1b2d-4916-9330-a68621bdb587
|
iterative-proposal-refinement-for-weakly
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Cao_Iterative_Proposal_Refinement_for_Weakly-Supervised_Video_Grounding_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Cao_Iterative_Proposal_Refinement_for_Weakly-Supervised_Video_Grounding_CVPR_2023_paper.pdf
|
Iterative Proposal Refinement for Weakly-Supervised Video Grounding
|
Weakly-Supervised Video Grounding (WSVG) aims to localize events of interest in untrimmed videos with only video-level annotations. To date, most of the state-of-the-art WSVG methods follow a two-stage pipeline, i.e., firstly generating potential temporal proposals and then grounding with these proposal candidates. Despite the recent progress, existing proposal generation methods suffer from two drawbacks: 1) lack of explicit correspondence modeling; and 2) partial coverage of complex events. To this end, we propose a novel IteRative prOposal refiNement network (dubbed as IRON) to gradually distill the prior knowledge into each proposal and encourage proposals with more complete coverage. Specifically, we set up two lightweight distillation branches to uncover the cross-modal correspondence on both the semantic and conceptual levels. Then, an iterative Label Propagation (LP) strategy is devised to prevent the network from focusing excessively on the most discriminative events instead of the whole sentence content. Precisely, during each iteration, the proposal with the minimal distillation loss and its adjacent ones are regarded as the positive samples, which refines proposal confidence scores in a cascaded manner. Extensive experiments and ablation studies on two challenging WSVG datasets have attested to the effectiveness of our IRON.
|
['Daxin Jiang', 'Tao Shen', 'Yuexian Zou', 'Can Zhang', 'Long Chen', 'Xiubo Geng', 'Can Xu', 'Fangyun Wei', 'Meng Cao']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['video-grounding']
|
['computer-vision']
|
[ 3.84436250e-01 1.44044846e-01 -3.20501238e-01 -4.39389467e-01
-6.81937993e-01 -2.39328623e-01 5.87513149e-01 3.61091733e-01
-4.75189984e-01 5.71000934e-01 2.93033242e-01 7.07507282e-02
1.41190290e-01 -5.32340765e-01 -5.92113018e-01 -5.96491873e-01
1.74318731e-01 2.13157862e-01 8.04125071e-01 2.19783410e-02
1.35497540e-01 -6.65985942e-02 -1.37271881e+00 2.47117966e-01
9.39154744e-01 8.53791118e-01 3.13114882e-01 -3.43558937e-02
-1.57677472e-01 7.99332857e-01 -2.70997941e-01 -4.32921410e-01
-8.42900202e-02 -7.33131886e-01 -7.21293092e-01 3.99565756e-01
2.68026829e-01 -2.69478858e-01 -1.93999171e-01 1.12060034e+00
2.75267482e-01 2.96695739e-01 1.90315187e-01 -1.09089744e+00
-2.74339557e-01 7.89121091e-01 -7.30856538e-01 2.25844800e-01
3.21043164e-01 2.49734238e-01 1.28693390e+00 -1.22575951e+00
7.56745458e-01 1.01189518e+00 3.97066623e-01 4.84825075e-01
-1.29449534e+00 -6.59061074e-01 8.71992230e-01 2.51460671e-01
-1.46719718e+00 -2.90600777e-01 1.13408387e+00 -2.25260198e-01
4.03815269e-01 8.04238319e-02 8.34615469e-01 1.23257458e+00
-4.10107136e-01 1.04272997e+00 7.65297234e-01 -1.87672898e-01
2.89713681e-01 8.01289231e-02 3.20951976e-02 9.42365229e-01
3.12501639e-02 -3.85178149e-01 -7.78140426e-01 1.28362747e-02
6.45450354e-01 -8.75431374e-02 -2.88741767e-01 -4.28596318e-01
-1.46703184e+00 7.07821369e-01 3.36612940e-01 3.19577277e-01
-5.53892255e-01 -1.40508143e-02 4.35920626e-01 -2.03954399e-01
6.63887024e-01 2.93071210e-01 -2.45823786e-01 1.37512118e-01
-1.21656024e+00 2.15189427e-01 2.30252206e-01 7.85496354e-01
8.65672231e-01 -3.28168452e-01 -6.88469887e-01 7.35563040e-01
5.02015293e-01 -2.88139313e-01 2.82492846e-01 -6.10051632e-01
6.62042022e-01 8.20164323e-01 9.80721265e-02 -1.08670139e+00
-2.98546791e-01 -6.49541616e-01 -4.42362666e-01 -3.50511789e-01
1.64775372e-01 -2.59348154e-02 -7.35639930e-01 1.78227019e+00
6.50926530e-01 7.82832503e-01 -2.44245991e-01 1.11993265e+00
8.20834756e-01 7.37554193e-01 4.04070824e-01 -3.14402878e-01
1.33607090e+00 -1.13537002e+00 -5.95729530e-01 -2.48378634e-01
4.44868863e-01 -5.87134480e-01 1.14536285e+00 2.67819643e-01
-1.05336499e+00 -5.98914146e-01 -9.68437791e-01 6.33389726e-02
2.11537600e-01 4.04105246e-01 4.61720616e-01 1.15564600e-01
-7.52156079e-01 4.17918026e-01 -1.05586827e+00 -2.73118287e-01
5.94783306e-01 8.24928284e-03 -6.20518923e-02 -2.02718258e-01
-1.11315954e+00 4.49254990e-01 6.35478914e-01 3.26804399e-01
-9.50569868e-01 -5.20629406e-01 -8.86940658e-01 5.82678914e-02
8.82028699e-01 -6.16953313e-01 9.44083929e-01 -1.06745756e+00
-1.36996400e+00 6.79916680e-01 -3.74663830e-01 -4.32958275e-01
6.80052161e-01 -3.29150379e-01 -3.45850527e-01 3.02847058e-01
2.11601049e-01 9.70103979e-01 9.63684380e-01 -1.11993814e+00
-9.03968930e-01 2.48993654e-02 2.49082685e-01 3.27825367e-01
-4.48910654e-01 7.30947256e-02 -1.08946693e+00 -8.67307484e-01
5.20558238e-01 -8.20004225e-01 -2.38090351e-01 1.38789229e-03
-5.39942265e-01 -5.63828588e-01 5.33195496e-01 -4.79192406e-01
1.51687467e+00 -2.19249487e+00 3.98889303e-01 1.93453029e-01
2.46973693e-01 1.56407759e-01 -1.54280365e-01 1.49892539e-01
7.14286566e-02 -3.94003540e-02 -9.86628458e-02 -7.38528848e-01
-1.14411294e-01 1.27153406e-02 -2.40547046e-01 3.79431248e-01
6.26394689e-01 7.28587747e-01 -1.40998006e+00 -7.15872169e-01
2.57920057e-01 2.22688749e-01 -7.31751263e-01 1.08404249e-01
-4.79001582e-01 5.92738748e-01 -5.78197360e-01 5.94826221e-01
4.45898473e-01 -5.30868530e-01 2.05577835e-01 -6.46950424e-01
-1.67810619e-01 4.42732066e-01 -1.21894228e+00 1.94712329e+00
-1.73249468e-03 2.19096318e-01 -1.88055947e-01 -1.05206156e+00
8.57198715e-01 2.87582964e-01 5.29623568e-01 -3.71929497e-01
-1.38347715e-01 9.81277525e-02 -1.33887544e-01 -4.87606287e-01
5.53303123e-01 -1.00580633e-01 3.53224501e-02 2.96361446e-01
1.26578435e-01 4.31571066e-01 2.85686493e-01 4.53147054e-01
8.88589382e-01 6.51822567e-01 1.07316487e-01 5.66821955e-02
6.52499199e-01 -1.08344533e-01 1.03527355e+00 6.44413412e-01
-4.09399092e-01 7.93595254e-01 5.93022108e-01 -2.37781197e-01
-6.91748261e-01 -9.39008832e-01 3.39697689e-01 1.09230161e+00
4.98463154e-01 -8.06502342e-01 -6.84330225e-01 -1.11280870e+00
-4.77380991e-01 6.60596967e-01 -5.98506927e-01 -2.33477965e-01
-6.79857612e-01 -6.47892475e-01 3.22594792e-01 4.49393988e-01
5.74561596e-01 -1.16543365e+00 -5.36512792e-01 3.79340351e-01
-5.46027541e-01 -1.21857846e+00 -6.31421089e-01 -1.90217450e-01
-7.18693614e-01 -9.94074941e-01 -6.31950974e-01 -6.96227968e-01
9.60848033e-01 3.77668679e-01 8.45739365e-01 1.91379949e-01
9.46593806e-02 -6.17677569e-02 -6.29587829e-01 9.21901017e-02
-9.72434729e-02 -2.64596269e-02 -1.68202475e-01 4.94122267e-01
3.10181379e-01 -2.35877752e-01 -8.42204690e-01 4.35062140e-01
-7.24715829e-01 5.23168445e-01 6.42024517e-01 7.59652972e-01
9.54111636e-01 2.58188061e-02 6.55232370e-01 -8.45673323e-01
2.72328675e-01 -5.29069304e-01 -4.75832313e-01 2.78871477e-01
-3.50061625e-01 -2.04528049e-02 3.91811043e-01 -4.81809556e-01
-1.15463698e+00 2.06853211e-01 -7.32333437e-02 -5.78738809e-01
-1.27452016e-01 6.33025885e-01 -7.14736953e-02 4.86274421e-01
2.25375175e-01 3.98542434e-01 -4.17293727e-01 -2.73589969e-01
2.43587315e-01 1.44592300e-02 5.42868376e-01 -5.17631531e-01
7.37020910e-01 5.57450175e-01 -4.62415695e-01 -4.29421067e-01
-1.09789693e+00 -5.47011733e-01 -3.99405718e-01 -5.10292411e-01
1.02518392e+00 -9.02047873e-01 -2.60228574e-01 2.01953173e-01
-1.04540956e+00 -1.35403484e-01 -2.45773211e-01 6.13597512e-01
-2.08002359e-01 4.26076740e-01 -4.80034828e-01 -7.19985008e-01
-1.38105869e-01 -1.15476775e+00 1.07018471e+00 2.24026024e-01
-3.39581192e-01 -4.83202636e-01 -1.07640825e-01 4.25005317e-01
-1.31742293e-02 3.64354774e-02 6.72261953e-01 -5.29787838e-01
-6.34214222e-01 -1.49822488e-01 -3.56477201e-01 8.19570869e-02
7.31461495e-02 -2.44138949e-03 -8.02399397e-01 -1.51365757e-01
-2.15455920e-01 -2.61761844e-01 1.08461845e+00 2.88265765e-01
1.25146270e+00 -2.33235464e-01 -4.21528935e-01 3.04956853e-01
9.87848341e-01 -9.31459516e-02 3.62038046e-01 2.11338937e-01
7.02595592e-01 5.62307358e-01 1.04245758e+00 4.90562081e-01
3.29205513e-01 7.87161231e-01 5.48617363e-01 -9.29397792e-02
1.15048820e-02 -6.67748272e-01 4.73963410e-01 7.58967042e-01
-1.08435027e-01 -3.35735440e-01 -5.33340812e-01 6.38188064e-01
-2.12915277e+00 -9.03995931e-01 1.12270251e-01 2.13995695e+00
8.11739326e-01 6.55571222e-01 1.54139161e-01 1.47662582e-02
9.79772210e-01 4.66571182e-01 -5.35342216e-01 4.85929638e-01
-4.88105603e-03 -2.03377157e-01 -1.20941021e-01 1.89427748e-01
-1.26356256e+00 1.13982332e+00 4.49107552e+00 8.13777924e-01
-9.88996923e-01 2.10689738e-01 6.57969236e-01 -2.71561235e-01
-3.88839424e-01 3.08250457e-01 -8.32135379e-01 6.41827822e-01
2.38246843e-01 2.52378374e-01 1.30079851e-01 4.96664286e-01
5.88504910e-01 -9.59592611e-02 -8.97622347e-01 7.24883676e-01
1.26198083e-01 -1.42248154e+00 1.49167418e-01 -3.77489448e-01
5.81225574e-01 -3.06391474e-02 -3.20220530e-01 3.36715966e-01
1.08102649e-01 -2.85468966e-01 1.25898290e+00 3.67212564e-01
4.31324869e-01 -7.00010002e-01 6.84292138e-01 4.55185592e-01
-1.46473908e+00 8.13235343e-02 -1.29160523e-01 2.90704846e-01
4.96383965e-01 5.63932240e-01 -7.85768747e-01 6.22984588e-01
5.00778317e-01 1.05204821e+00 -4.36554432e-01 1.04205918e+00
-7.22820759e-01 1.00944352e+00 -2.25159988e-01 7.70914257e-02
5.95429182e-01 -1.51016712e-01 7.46099472e-01 1.18416119e+00
2.16412067e-01 2.52670318e-01 5.40397286e-01 1.02965271e+00
-3.54259498e-02 1.35456666e-01 -1.24841742e-03 1.36927783e-01
5.76751351e-01 1.33570743e+00 -1.18311453e+00 -3.44311774e-01
-6.04395390e-01 1.00867438e+00 3.89874697e-01 3.59555900e-01
-1.09983921e+00 4.63414080e-02 1.39595419e-01 1.19262971e-01
4.68474776e-01 -3.93946245e-02 5.30546159e-02 -1.38502741e+00
1.43245608e-01 -4.48731661e-01 6.45231247e-01 -6.41067386e-01
-1.10432887e+00 5.71178019e-01 -1.05380148e-01 -1.42996085e+00
1.14132211e-01 1.89488351e-01 -7.59581685e-01 5.77024341e-01
-1.47563636e+00 -1.23071110e+00 -4.28144455e-01 4.13369417e-01
9.78217959e-01 2.62459904e-01 3.79838943e-01 3.45652938e-01
-1.01319730e+00 4.94359910e-01 -6.10712647e-01 4.86107096e-02
6.95509851e-01 -1.00874472e+00 9.98869613e-02 1.22342062e+00
2.65470982e-01 6.20914996e-01 6.39791250e-01 -9.01406646e-01
-8.48780751e-01 -1.41982257e+00 8.87888908e-01 -9.22427699e-03
5.88772476e-01 -3.87700737e-01 -1.07759440e+00 5.62609792e-01
-1.19983539e-01 1.65256903e-01 4.00806069e-01 1.73383616e-02
-2.40461886e-01 -5.80862015e-02 -7.08086610e-01 7.52232313e-01
1.10972655e+00 -3.21532696e-01 -5.42281985e-01 2.26327851e-01
7.50138521e-01 -1.98323309e-01 -4.46377218e-01 4.23054427e-01
4.55334246e-01 -8.05745959e-01 8.04159343e-01 -3.56141269e-01
5.23665369e-01 -7.39041686e-01 1.20753102e-01 -9.03680623e-01
-3.12752753e-01 -7.46943891e-01 -3.55799079e-01 1.57966733e+00
4.33731496e-01 -4.74039242e-02 8.00090790e-01 2.72971332e-01
-3.44350994e-01 -8.93706083e-01 -7.06713438e-01 -5.28651476e-01
-7.79822350e-01 -5.90312958e-01 2.56362587e-01 9.62289989e-01
2.28664532e-01 3.60539913e-01 -5.37946045e-01 2.63676822e-01
5.45825660e-01 2.10202396e-01 4.99279737e-01 -9.74626362e-01
-3.29589695e-01 -4.11237597e-01 -9.62335095e-02 -1.31509542e+00
6.03824332e-02 -7.49262929e-01 4.51071680e-01 -1.47837853e+00
4.86550003e-01 -4.81785744e-01 -6.65931165e-01 6.29759252e-01
-7.28717327e-01 2.66806751e-01 1.92339737e-02 4.63086993e-01
-1.29448915e+00 7.14160800e-01 1.05679727e+00 -6.15290999e-02
-3.86900783e-01 -2.82133706e-02 -5.18971920e-01 9.59154606e-01
4.56190467e-01 -5.60334086e-01 -6.42797232e-01 -3.40990454e-01
2.94968307e-01 -1.32509232e-01 5.01572251e-01 -7.80800641e-01
2.36025408e-01 -2.89634198e-01 1.58117503e-01 -9.68261659e-01
1.69855461e-01 -4.34462100e-01 -3.57780829e-02 3.99502486e-01
-5.32481432e-01 -2.84638047e-01 -7.49722645e-02 9.01103854e-01
-3.38796884e-01 -1.21187583e-01 6.82869077e-01 1.61323268e-02
-9.37466741e-01 4.45827514e-01 -1.92949355e-01 -4.74699363e-02
1.11209643e+00 -8.94559026e-02 -1.25139117e-01 -9.73392129e-02
-7.51507521e-01 3.70281100e-01 3.36489290e-01 5.00952482e-01
6.98510647e-01 -1.20969915e+00 -7.94736385e-01 6.00119457e-02
3.52054954e-01 2.38573000e-01 2.92040944e-01 1.16056347e+00
-4.34996933e-03 1.86147675e-01 3.58820260e-01 -6.57160819e-01
-1.11686206e+00 4.93025690e-01 -2.60792691e-02 -3.50451350e-01
-9.69775677e-01 1.00646782e+00 3.70997131e-01 1.89908683e-01
3.06085378e-01 -3.26485693e-01 -4.59998518e-01 2.48479322e-01
4.75809366e-01 1.10377401e-01 -1.11085460e-01 -6.29508317e-01
-5.43376386e-01 3.19927573e-01 -3.61235648e-01 -7.79678002e-02
1.21502328e+00 -2.63312936e-01 1.13299906e-01 2.36794159e-01
7.65525579e-01 -6.32986575e-02 -1.66663194e+00 -4.88474578e-01
3.71590778e-02 -2.96792597e-01 8.82900432e-02 -4.44727898e-01
-1.20694113e+00 5.56060553e-01 1.60812467e-01 -2.34778523e-02
1.14068294e+00 2.80210078e-01 7.39898324e-01 1.40041253e-02
1.84557498e-01 -9.42040622e-01 5.03613591e-01 1.70192659e-01
6.19335771e-01 -1.01865327e+00 4.36785407e-02 -7.18302250e-01
-6.98973298e-01 8.59316289e-01 8.61846566e-01 2.87904534e-02
3.14530253e-01 -2.12959990e-01 -4.27220583e-01 -2.60652721e-01
-8.03696454e-01 -1.69680953e-01 4.21823263e-01 -3.37689780e-02
3.15467834e-01 -3.67603779e-01 -5.96645772e-01 8.49724591e-01
5.10125279e-01 1.96040645e-01 2.59754598e-01 9.09096658e-01
-4.57231522e-01 -7.84970939e-01 -7.17776045e-02 2.68759698e-01
-3.00483674e-01 -4.51962799e-02 -2.30227515e-01 4.32790637e-01
4.55459476e-01 8.67746055e-01 -5.05939312e-02 -3.91416609e-01
2.51262844e-01 1.04214931e-02 1.51646316e-01 -7.28797615e-01
-4.36789513e-01 5.09988308e-01 9.42840725e-02 -7.22206593e-01
-7.94101655e-01 -9.79151785e-01 -1.42042732e+00 3.33201140e-01
-6.08837485e-01 2.15048909e-01 1.36200532e-01 1.15997338e+00
2.83248007e-01 6.99885666e-01 4.30242717e-01 -8.64984512e-01
-1.75754532e-01 -9.56870675e-01 -1.59806430e-01 4.48530495e-01
6.94854259e-02 -7.97100186e-01 -1.19795360e-01 2.07505018e-01]
|
[9.423538208007812, 0.729713499546051]
|
61564f6c-74ca-4e32-a711-044086bab52e
|
spatial-separated-curve-rendering-network-for
|
2109.05750
| null |
https://arxiv.org/abs/2109.05750v4
|
https://arxiv.org/pdf/2109.05750v4.pdf
|
Spatial-Separated Curve Rendering Network for Efficient and High-Resolution Image Harmonization
|
Image harmonization aims to modify the color of the composited region with respect to the specific background. Previous works model this task as a pixel-wise image-to-image translation using UNet family structures. However, the model size and computational cost limit the ability of their models on edge devices and higher-resolution images. To this end, we propose a novel spatial-separated curve rendering network(S$^2$CRNet) for efficient and high-resolution image harmonization for the first time. In S$^2$CRNet, we firstly extract the spatial-separated embeddings from the thumbnails of the masked foreground and background individually. Then, we design a curve rendering module(CRM), which learns and combines the spatial-specific knowledge using linear layers to generate the parameters of the piece-wise curve mapping in the foreground region. Finally, we directly render the original high-resolution images using the learned color curve. Besides, we also make two extensions of the proposed framework via the Cascaded-CRM and Semantic-CRM for cascaded refinement and semantic guidance, respectively. Experiments show that the proposed method reduces more than 90% parameters compared with previous methods but still achieves the state-of-the-art performance on both synthesized iHarmony4 and real-world DIH test sets. Moreover, our method can work smoothly on higher resolution images(eg., $2048\times2048$) in 0.1 seconds with much lower GPU computational resources than all existing methods. The code will be made available at \url{http://github.com/stefanLeong/S2CRNet}.
|
['Jue Wang', 'Chi-Man Pun', 'Xiaodong Cun', 'Jingtang Liang']
|
2021-09-13
| null | null | null | null |
['image-harmonization', '2048']
|
['computer-vision', 'playing-games']
|
[ 3.25159132e-01 -5.09079061e-02 2.07912534e-01 -2.11045191e-01
-8.02155614e-01 -4.42548901e-01 2.55359709e-01 -2.72501200e-01
-2.92988539e-01 4.50420916e-01 -2.55715132e-01 -2.67236292e-01
9.16059613e-02 -1.03256464e+00 -8.32802713e-01 -6.69554651e-01
3.66763830e-01 3.64557430e-02 6.55956566e-01 -3.69335949e-01
1.76437840e-01 5.45351923e-01 -1.46608973e+00 2.11008191e-01
1.04652166e+00 1.04915237e+00 4.07412112e-01 6.38115764e-01
-4.69691679e-02 5.39242268e-01 -2.16996923e-01 -4.14623260e-01
4.12222177e-01 -4.41371232e-01 -3.99220020e-01 1.80180505e-01
4.81039733e-01 -3.62960279e-01 -3.40753019e-01 1.16696525e+00
5.93420684e-01 5.01987934e-02 3.56984675e-01 -1.01228333e+00
-7.89802074e-01 2.40039796e-01 -1.16271734e+00 1.79445609e-01
-8.47419798e-02 1.15355037e-01 5.99469244e-01 -1.09541321e+00
5.53310096e-01 1.01894701e+00 5.21331906e-01 4.97254223e-01
-1.30564153e+00 -1.12089765e+00 1.87875718e-01 2.16678575e-01
-1.65147269e+00 -2.29530379e-01 9.66095805e-01 -1.38710946e-01
5.69265544e-01 2.73717731e-01 6.05702460e-01 6.79787040e-01
-1.04066417e-01 5.15458584e-01 1.21067297e+00 -2.12400526e-01
-4.27075736e-02 1.37433544e-01 -2.54429281e-01 8.54415894e-01
1.84391901e-01 1.06501468e-01 -3.84302050e-01 2.75339812e-01
1.36077750e+00 -2.14256883e-01 -4.44006145e-01 -2.39528045e-01
-1.04187763e+00 5.42482495e-01 6.76510930e-01 1.97741434e-01
-1.22755013e-01 2.50212938e-01 5.23260757e-02 -5.49338832e-02
4.86924708e-01 1.50952131e-01 -2.98962981e-01 2.56290525e-01
-9.66944396e-01 9.27415937e-02 2.29970768e-01 1.12679696e+00
9.26113486e-01 1.72346666e-01 2.35787779e-02 9.14303184e-01
2.02345140e-02 4.20929760e-01 4.12405282e-02 -9.77800846e-01
4.81196344e-01 5.23984611e-01 2.32466966e-01 -1.09237826e+00
-2.74776876e-01 -5.11054516e-01 -1.15019286e+00 3.39316010e-01
2.79705733e-01 -3.71495187e-02 -9.08128858e-01 1.46262300e+00
5.23503900e-01 5.71596146e-01 -7.36759529e-02 1.06860757e+00
7.66790867e-01 9.61940408e-01 -7.74436146e-02 -5.01084998e-02
1.50479913e+00 -1.26519179e+00 -5.52903652e-01 -7.62744248e-02
1.47690490e-01 -9.60188746e-01 1.27513289e+00 3.36085916e-01
-1.39777064e+00 -7.90831864e-01 -1.15175176e+00 -4.68066990e-01
-1.49260297e-01 3.43195409e-01 3.92483383e-01 5.30506015e-01
-1.15480602e+00 3.69103044e-01 -6.39873981e-01 1.20479286e-01
5.51833570e-01 2.57884473e-01 -6.62614852e-02 -1.83546841e-01
-1.05856907e+00 4.97851878e-01 3.78347635e-01 2.52302349e-01
-5.80145597e-01 -9.89249289e-01 -7.43844151e-01 -1.01532958e-01
4.27429885e-01 -5.56348860e-01 8.42507124e-01 -9.68514204e-01
-1.57331121e+00 9.17064309e-01 -8.53578001e-02 -1.46510303e-01
6.53877139e-01 -7.00999722e-02 -4.56460536e-01 2.13510647e-01
-9.28515568e-02 8.27250361e-01 9.80251670e-01 -1.41452456e+00
-9.83707607e-01 -1.23748757e-01 -2.09647678e-02 3.75560045e-01
-1.68491498e-01 6.50707185e-02 -1.19148493e+00 -9.51329887e-01
7.30242878e-02 -7.33000338e-01 -1.58864796e-01 3.21130127e-01
-2.65721798e-01 8.88557285e-02 9.80344653e-01 -9.04346585e-01
1.19831550e+00 -2.35043836e+00 1.12483412e-01 1.48386657e-01
2.03164458e-01 3.43504757e-01 -2.14871004e-01 -2.14471981e-01
-8.74349177e-02 4.42710444e-02 -4.43235219e-01 -2.44320184e-01
-2.62915939e-01 -5.41909076e-02 7.10867695e-04 4.13702190e-01
3.43456030e-01 8.95718515e-01 -6.20454788e-01 -5.13725162e-01
4.03814137e-01 9.84633029e-01 -5.66788614e-01 1.49730563e-01
-9.46326032e-02 5.75446665e-01 -2.30360389e-01 4.54510063e-01
1.04893911e+00 -2.37690270e-01 1.20966822e-01 -5.49097419e-01
-3.66371810e-01 -3.08466852e-01 -1.33668113e+00 1.69662130e+00
-6.04963303e-01 5.31754017e-01 1.44096121e-01 -6.95025384e-01
1.11034489e+00 -6.81951195e-02 3.66797090e-01 -9.88947213e-01
1.72817990e-01 3.16509008e-01 -2.33256310e-01 -7.19964951e-02
4.73073602e-01 1.57510221e-01 -2.77826563e-02 1.90635756e-01
-4.72262144e-01 -2.46023178e-01 -6.95707276e-03 -4.91652265e-02
5.17058074e-01 2.20282108e-01 -8.06418061e-02 -2.21192002e-01
7.31995344e-01 -1.84541449e-01 5.86717725e-01 2.64141887e-01
7.67542869e-02 9.40851569e-01 3.34124178e-01 -4.40563112e-01
-1.37707424e+00 -1.14129114e+00 -1.25655010e-01 9.90581155e-01
7.13097632e-01 -1.64559916e-01 -1.00214565e+00 -2.52184868e-01
-3.97313058e-01 6.49122417e-01 -5.10073602e-01 4.43970971e-02
-9.86817837e-01 -8.04763973e-01 3.82528543e-01 6.31413877e-01
1.08547616e+00 -9.00576830e-01 -5.92465758e-01 1.42865822e-01
-1.91842213e-01 -1.24845088e+00 -8.81616890e-01 -3.07552963e-01
-6.20294750e-01 -9.09995675e-01 -8.53520453e-01 -1.10582149e+00
7.53619492e-01 3.84097248e-01 8.65743339e-01 2.51068532e-01
-5.01837790e-01 -8.37087259e-02 -2.14762181e-01 -4.65013180e-03
-3.08184847e-02 -3.69416401e-02 -3.83914769e-01 2.53208697e-01
-1.21248499e-01 -5.55026650e-01 -1.07842076e+00 5.40810943e-01
-1.07033980e+00 6.81642354e-01 6.12581134e-01 7.07792938e-01
9.86620247e-01 4.82753754e-01 2.01858655e-01 -9.01662350e-01
3.47808689e-01 1.50583070e-02 -9.13552582e-01 1.57093018e-01
-6.36874139e-01 -5.21040261e-02 6.43235981e-01 -4.27503943e-01
-1.31949782e+00 9.79632065e-02 -1.40814587e-01 -5.20652413e-01
7.58086592e-02 -1.77264050e-01 -3.45087290e-01 -1.61400288e-01
3.51023674e-01 4.02894169e-01 -2.95066684e-01 -4.86188531e-01
6.60340190e-01 4.37186927e-01 8.49076927e-01 -6.27623677e-01
1.07003844e+00 7.16832936e-01 -1.72588274e-01 -5.97744167e-01
-5.75076461e-01 -1.30217478e-01 -5.24703383e-01 -2.59235799e-01
1.11537039e+00 -9.65292275e-01 -5.77242374e-01 6.45748556e-01
-1.03316629e+00 -6.09782100e-01 -9.80939046e-02 8.94868374e-02
-5.25100112e-01 1.86350361e-01 -7.22709298e-01 -4.71918792e-01
-4.77342069e-01 -1.26686442e+00 1.15892804e+00 5.81850767e-01
3.75207812e-01 -6.37233317e-01 -3.81764919e-01 4.05342877e-01
4.96754497e-01 3.03514898e-01 9.60745931e-01 2.61408865e-01
-9.57449794e-01 1.98432088e-01 -9.46432233e-01 3.28929961e-01
5.36860153e-02 6.08270802e-03 -7.99553037e-01 -2.92763025e-01
-2.67658174e-01 6.54870123e-02 7.77693093e-01 3.04549217e-01
1.49390924e+00 -1.25495866e-01 -8.97293836e-02 1.09784651e+00
1.68743312e+00 3.80315542e-01 9.95990038e-01 3.45714658e-01
1.02204752e+00 4.85817671e-01 5.84218442e-01 3.31256509e-01
4.10975873e-01 9.47428286e-01 2.09408626e-01 -7.63165414e-01
-6.30057752e-01 -3.06127667e-01 1.55078918e-01 6.27267718e-01
-2.35065207e-01 -3.33644450e-02 -7.19503462e-01 3.23561162e-01
-1.57954729e+00 -6.73113585e-01 -3.11745703e-01 2.08688712e+00
8.83314610e-01 9.78079513e-02 -1.65545642e-02 4.71840091e-02
9.84748483e-01 2.04573989e-01 -7.53891885e-01 -2.01725841e-01
-3.17059189e-01 4.56407279e-01 6.14557445e-01 6.46478772e-01
-9.84190106e-01 1.17447686e+00 4.53059149e+00 1.20067358e+00
-1.16451120e+00 2.88551986e-01 9.98630702e-01 -6.05794489e-02
-3.38597924e-01 -1.12768389e-01 -7.76524842e-01 4.53510016e-01
4.99846965e-01 -3.69256139e-02 6.33825898e-01 4.82706517e-01
2.96678096e-01 9.20168534e-02 -5.69878876e-01 1.25326359e+00
8.49838182e-02 -1.47130752e+00 6.52700067e-02 -2.36774489e-01
8.56724143e-01 -3.05580676e-01 3.41566384e-01 -1.12224974e-01
2.61925887e-02 -9.14263427e-01 8.36372852e-01 3.31228733e-01
1.37883496e+00 -9.44872916e-01 3.34111273e-01 1.01899728e-02
-1.62312496e+00 3.45790088e-02 -3.35691690e-01 4.28500116e-01
1.79957047e-01 3.91053617e-01 -2.68896252e-01 5.31511426e-01
1.05551124e+00 5.87393701e-01 -6.16892338e-01 6.45148695e-01
-3.30672830e-01 2.36679286e-01 -2.19361573e-01 4.40114170e-01
6.05122410e-02 -4.07371700e-01 2.73576796e-01 1.14386714e+00
4.17778552e-01 4.28057581e-01 -1.14415713e-01 9.92137611e-01
-1.85038090e-01 1.84989572e-01 -7.79170170e-02 4.04709399e-01
4.58094180e-01 1.38188303e+00 -9.52627361e-01 -3.19193661e-01
-4.63366926e-01 1.33024085e+00 2.52341121e-01 3.88721645e-01
-1.32943964e+00 -4.64384168e-01 4.55074310e-01 4.09991771e-01
5.26996911e-01 -1.26375437e-01 -3.56395453e-01 -1.05699992e+00
1.44929796e-01 -8.21450472e-01 1.98063284e-01 -9.80012059e-01
-9.93268490e-01 8.38540375e-01 -1.45471409e-01 -1.21122205e+00
3.96050036e-01 -4.47687060e-01 -4.12754565e-01 8.97147059e-01
-1.80450225e+00 -1.26178002e+00 -7.02814817e-01 8.17537010e-01
6.87072754e-01 1.79766536e-01 4.88307148e-01 6.08977258e-01
-7.43125439e-01 6.88671112e-01 4.90569845e-02 2.32199356e-01
7.34677553e-01 -9.16744053e-01 5.27514637e-01 9.79023278e-01
-1.52794197e-01 1.97148874e-01 3.98636490e-01 -5.60250700e-01
-1.09128821e+00 -1.39837635e+00 2.58964032e-01 -6.18274398e-02
2.98020810e-01 -5.16987026e-01 -1.05573404e+00 2.81875759e-01
2.32954770e-01 2.61583745e-01 1.83465853e-01 -6.95060551e-01
-4.15622026e-01 -3.75371993e-01 -1.11147368e+00 7.99869359e-01
1.22686589e+00 -3.26796979e-01 6.61773458e-02 -4.37106565e-02
8.84450257e-01 -7.67962873e-01 -9.42658007e-01 3.91446710e-01
5.23845494e-01 -1.07669640e+00 1.19052136e+00 7.72390664e-02
4.66737479e-01 -8.22445273e-01 -1.19158834e-01 -8.93564701e-01
-4.10702199e-01 -6.20838404e-01 2.12594390e-01 1.14761078e+00
4.28998441e-01 -5.77827692e-01 7.30582595e-01 4.92481053e-01
-1.28324330e-01 -8.25760782e-01 -7.67473102e-01 -3.91365051e-01
9.24887806e-02 -3.16328317e-01 7.31790423e-01 8.97202790e-01
-7.69653857e-01 1.68993086e-01 -3.77783149e-01 4.64494854e-01
8.39542568e-01 3.02038431e-01 7.03903258e-01 -7.75161028e-01
-2.61200190e-01 -4.79117930e-01 -1.82277322e-01 -1.06366205e+00
-1.80699572e-01 -6.44451797e-01 -1.33088797e-01 -1.30862916e+00
1.68319970e-01 -7.17102885e-01 -2.08488151e-01 3.26484352e-01
-2.73927420e-01 7.44705319e-01 3.56459767e-01 1.24658868e-01
-3.07124436e-01 4.40157115e-01 1.62985647e+00 -8.98427963e-02
-3.39075357e-01 -3.50727826e-01 -7.38510132e-01 6.96627378e-01
1.05279469e+00 -2.10494265e-01 -4.33456689e-01 -6.12962902e-01
7.46546453e-03 5.67612872e-02 5.17927468e-01 -9.47842002e-01
1.93267077e-01 -1.56561315e-01 6.63768888e-01 -5.62512100e-01
3.44351828e-01 -8.01024735e-01 4.69807267e-01 2.92630017e-01
-1.70863271e-01 7.55155236e-02 3.66249144e-01 4.63316768e-01
-7.87363425e-02 1.81385770e-01 1.28384769e+00 -3.84937450e-02
-8.08469296e-01 5.33630311e-01 1.40162826e-01 -3.24192792e-02
1.19206667e+00 -3.52321088e-01 -5.02208829e-01 -1.78709254e-01
-4.78967607e-01 5.97348437e-02 7.58774936e-01 4.89156812e-01
8.37305546e-01 -1.25049245e+00 -7.47811913e-01 4.76310819e-01
-2.10752971e-02 4.21380639e-01 6.93384230e-01 8.30372334e-01
-8.50527644e-01 4.73300293e-02 -3.08814436e-01 -4.86392438e-01
-1.10392690e+00 4.82873678e-01 5.38561523e-01 -4.42034639e-02
-9.66111302e-01 8.72927785e-01 5.15677750e-01 -6.67120367e-02
6.38903007e-02 -1.97345182e-01 -7.15744346e-02 -1.99453697e-01
5.47247112e-01 3.43943089e-01 -1.62313253e-01 -6.57831907e-01
-1.87560812e-01 1.10637105e+00 -1.59519643e-01 -1.12807639e-01
1.30175924e+00 -1.96261138e-01 -1.72771424e-01 -4.59484719e-02
1.19664764e+00 4.79772165e-02 -1.57894957e+00 -2.18770474e-01
-4.45730746e-01 -6.93497062e-01 1.83799103e-01 -5.85364580e-01
-1.60018444e+00 8.18413675e-01 9.31555271e-01 -2.47145727e-01
1.55936909e+00 -1.53676197e-01 1.01862156e+00 -3.80077124e-01
1.95816651e-01 -1.04268515e+00 1.31940454e-01 -5.25505356e-02
7.89360583e-01 -9.26852822e-01 -7.82505572e-02 -7.87938178e-01
-6.28088295e-01 9.30441976e-01 9.09051776e-01 -2.64803618e-01
4.83087808e-01 3.94267768e-01 1.45704627e-01 -2.96628382e-02
-4.34665233e-01 -1.24148592e-01 2.92880923e-01 5.86063862e-01
3.14557076e-01 2.75480095e-02 -2.84462422e-01 4.39023525e-01
-1.39298216e-01 -9.48343128e-02 4.75455821e-01 4.69486773e-01
-2.35152557e-01 -1.05479884e+00 -3.38924766e-01 1.36556461e-01
-1.68727607e-01 -2.40977690e-01 2.37040836e-02 7.42851913e-01
4.10727441e-01 7.43034720e-01 1.72468588e-01 -3.81636024e-01
5.71736157e-01 -3.81498009e-01 4.45542008e-01 -2.52203286e-01
-3.56973827e-01 3.27213049e-01 -2.23576516e-01 -7.89234638e-01
-3.41504574e-01 -3.37093771e-01 -1.45495474e+00 -4.30665910e-01
-1.30762458e-01 -2.68744290e-01 5.04691780e-01 2.66501278e-01
4.02378082e-01 6.71806216e-01 6.34818196e-01 -9.04604077e-01
1.23280458e-01 -4.94894296e-01 -5.31685591e-01 3.63353521e-01
5.47030531e-02 -5.13572752e-01 -1.74776733e-01 2.48503864e-01]
|
[11.072680473327637, -1.5495492219924927]
|
7537d350-fa95-4ff3-ac12-41fd944eecca
|
towards-deep-attention-in-graph-neural
|
2306.02376
| null |
https://arxiv.org/abs/2306.02376v1
|
https://arxiv.org/pdf/2306.02376v1.pdf
|
Towards Deep Attention in Graph Neural Networks: Problems and Remedies
|
Graph neural networks (GNNs) learn the representation of graph-structured data, and their expressiveness can be further enhanced by inferring node relations for propagation. Attention-based GNNs infer neighbor importance to manipulate the weight of its propagation. Despite their popularity, the discussion on deep graph attention and its unique challenges has been limited. In this work, we investigate some problematic phenomena related to deep graph attention, including vulnerability to over-smoothed features and smooth cumulative attention. Through theoretical and empirical analyses, we show that various attention-based GNNs suffer from these problems. Motivated by our findings, we propose AEROGNN, a novel GNN architecture designed for deep graph attention. AERO-GNN provably mitigates the proposed problems of deep graph attention, which is further empirically demonstrated with (a) its adaptive and less smooth attention functions and (b) higher performance at deep layers (up to 64). On 9 out of 12 node classification benchmarks, AERO-GNN outperforms the baseline GNNs, highlighting the advantages of deep graph attention. Our code is available at https://github.com/syleeheal/AERO-GNN.
|
['Kijung Shin', 'Jaemin Yoo', 'Fanchen Bu', 'Soo Yong Lee']
|
2023-06-04
| null | null | null | null |
['deep-attention', 'graph-attention', 'deep-attention']
|
['computer-vision', 'graphs', 'natural-language-processing']
|
[-2.09837437e-01 5.30619204e-01 -2.10862398e-01 2.39265244e-02
-7.22317845e-02 -2.95125246e-01 5.41936696e-01 3.50213021e-01
-1.37837842e-01 5.86873949e-01 2.70889640e-01 -6.78086281e-01
-2.57174999e-01 -1.06673014e+00 -7.25486755e-01 -5.46362460e-01
-5.29335976e-01 3.17408413e-01 1.80678591e-01 -5.02425194e-01
1.32136628e-01 6.32575095e-01 -1.01129258e+00 -2.03885853e-01
7.58986473e-01 7.85455525e-01 -5.61458208e-02 1.08958709e+00
-1.26354486e-01 1.12898421e+00 -6.75935149e-01 -5.70960104e-01
3.89785580e-02 -3.03906173e-01 -9.27075446e-01 -5.10438144e-01
3.74619931e-01 -1.49954662e-01 -9.36056852e-01 1.09579313e+00
5.15509129e-01 3.43586713e-01 4.72719669e-01 -1.66423666e+00
-1.43605137e+00 1.04704988e+00 -4.75724965e-01 5.67453265e-01
-1.96856931e-02 2.72311032e-01 1.62651539e+00 -6.50083959e-01
3.82554770e-01 1.32912755e+00 8.93414438e-01 4.61569160e-01
-9.70525265e-01 -5.30915797e-01 4.79012698e-01 1.14968158e-01
-1.36405337e+00 -3.09903137e-02 7.82466531e-01 -2.13925883e-01
1.09117937e+00 3.57382506e-01 7.46880531e-01 1.08548713e+00
3.60830635e-01 7.53115833e-01 2.95811713e-01 -1.10432036e-01
-1.19782262e-01 -3.20003480e-01 4.90871310e-01 8.34914625e-01
5.77106297e-01 1.83122773e-02 -1.06756710e-01 5.21171279e-02
8.35840881e-01 1.91701129e-01 -3.71730834e-01 -3.28669399e-02
-7.67611742e-01 9.25016284e-01 1.46394300e+00 2.86292732e-01
-4.91653174e-01 9.01087761e-01 4.09190089e-01 3.01794976e-01
6.56103671e-01 7.15353906e-01 -3.78557354e-01 1.56802028e-01
-1.65243715e-01 8.32083076e-02 7.69239128e-01 1.10929203e+00
8.07981551e-01 3.09546590e-01 -4.17070955e-01 5.28721094e-01
2.78630674e-01 2.91456133e-01 7.71503225e-02 -4.53808159e-01
3.55378658e-01 7.28772342e-01 -5.75960457e-01 -1.38767958e+00
-7.14066327e-01 -8.02307010e-01 -1.24008310e+00 -2.74889693e-02
2.38897532e-01 -2.47208878e-01 -8.79236937e-01 1.73060095e+00
7.95407221e-02 1.82875276e-01 -1.07217953e-01 8.90040636e-01
1.23397040e+00 5.86502969e-01 3.05078596e-01 4.79054362e-01
1.06925392e+00 -1.31754541e+00 -6.48983419e-01 -3.18428993e-01
6.36545777e-01 -1.73610970e-01 1.17197812e+00 -1.59997091e-01
-9.76636052e-01 -3.93226922e-01 -8.96338761e-01 -4.95472461e-01
-5.97595751e-01 -4.73661751e-01 1.17703068e+00 4.06154394e-01
-1.53489101e+00 8.64134073e-01 -7.64661610e-01 -5.49620450e-01
6.83484375e-01 5.04812837e-01 -1.80609167e-01 3.02136727e-02
-1.41803849e+00 4.63301331e-01 1.92614079e-01 3.35839301e-01
-7.58717656e-01 -7.36876547e-01 -1.04460180e+00 6.75431430e-01
3.29476357e-01 -8.28716218e-01 9.74467039e-01 -8.29859138e-01
-1.15450299e+00 4.38091308e-01 2.65024006e-01 -6.82464004e-01
1.58263147e-01 -2.78124630e-01 -3.47458929e-01 1.57979522e-02
-1.33547813e-01 5.28493524e-01 5.35912037e-01 -8.49556565e-01
-1.53177634e-01 -1.33525878e-01 6.23909831e-01 1.06982239e-01
-4.90337491e-01 -2.51646876e-01 -5.52042365e-01 -8.12560260e-01
-2.77285129e-01 -7.67841518e-01 -5.46231747e-01 -1.27011791e-01
-6.85590446e-01 -3.88177663e-01 6.62782431e-01 -2.50620067e-01
1.34859335e+00 -1.94435871e+00 1.68402165e-01 9.75396633e-02
1.00897348e+00 3.65169972e-01 -4.44170445e-01 6.99153364e-01
-1.08085744e-01 5.53441048e-01 -8.57747197e-02 -2.25139096e-01
6.89659864e-02 2.31678039e-01 -1.52430013e-01 4.61152911e-01
5.04551589e-01 1.51219463e+00 -1.07147205e+00 -1.64035887e-01
5.30832782e-02 7.69339561e-01 -7.61185467e-01 2.18638644e-01
-3.02992076e-01 2.33600229e-01 -5.20103514e-01 6.49631381e-01
5.09252548e-01 -8.79533410e-01 2.23781019e-01 5.40122762e-03
3.13775718e-01 3.36050838e-01 -4.44272667e-01 1.25138068e+00
-2.50195771e-01 1.02010310e+00 1.30343825e-01 -9.68796909e-01
7.53085971e-01 -4.56561632e-02 3.46186399e-01 -6.42780602e-01
2.89012015e-01 -1.52251974e-01 2.56845683e-01 -2.69502968e-01
7.13195443e-01 3.38915527e-01 3.15156095e-02 4.73116994e-01
9.11166817e-02 2.74468124e-01 -2.29753703e-02 6.96143270e-01
1.48012590e+00 -4.17080849e-01 2.84775227e-01 -6.34626389e-01
2.92699367e-01 -3.18127573e-01 2.40645662e-01 9.95340288e-01
-2.67232865e-01 2.79333502e-01 1.02412426e+00 -6.25021279e-01
-7.96233475e-01 -6.97333217e-01 2.17763022e-01 1.48937297e+00
2.10109025e-01 -7.86236584e-01 -6.37135744e-01 -9.32603896e-01
3.17503691e-01 5.18718898e-01 -9.66308594e-01 -3.48997295e-01
-5.53400695e-01 -6.68401778e-01 6.91076994e-01 7.20812500e-01
1.03751525e-01 -1.26155114e+00 -1.30131105e-02 5.41755334e-02
2.14590088e-01 -9.16935086e-01 -5.75980425e-01 -1.18276095e-02
-6.76358998e-01 -1.23331821e+00 -6.39488935e-01 -6.27850771e-01
6.88515782e-01 3.85619462e-01 1.45754266e+00 8.58906686e-01
-1.21310674e-01 3.99416506e-01 -3.99125606e-01 -3.91103536e-01
-3.31344485e-01 7.80357242e-01 -2.48532727e-01 -2.10311696e-01
2.67742336e-01 -7.84961104e-01 -6.85042799e-01 3.09990402e-02
-7.79723346e-01 -2.94875819e-02 6.01480305e-01 8.14363897e-01
1.43602327e-01 -1.63801640e-01 6.43708825e-01 -1.27379858e+00
9.06416893e-01 -8.77032638e-01 -5.29030144e-01 -1.22523218e-01
-7.89439499e-01 2.66956061e-01 6.95881188e-01 -1.48663402e-01
-3.78673464e-01 -6.39815629e-01 -3.42643887e-01 -5.08908629e-01
3.47718209e-01 8.05085659e-01 1.31218255e-01 -2.97472715e-01
6.46090567e-01 -2.78104216e-01 7.79031292e-02 -3.11274230e-01
3.69049072e-01 1.70036048e-01 3.04318815e-01 -4.71825480e-01
8.33831906e-01 2.13323385e-01 1.19116977e-01 -8.99629235e-01
-8.83688331e-01 -1.20792218e-01 -1.57816350e-01 -1.43906057e-01
7.31079996e-01 -7.37171888e-01 -8.94465983e-01 3.62169355e-01
-1.11585939e+00 -8.36465061e-01 -2.66628563e-01 1.68413222e-01
-6.50314838e-02 3.16026598e-01 -1.07350194e+00 -6.21226072e-01
-7.03503668e-01 -1.01018584e+00 7.66625524e-01 1.39823914e-01
-1.20812349e-01 -1.50451076e+00 -1.45428672e-01 -1.33530229e-01
8.67061794e-01 4.56171006e-01 9.44738626e-01 -7.73583114e-01
-8.92576814e-01 2.97360942e-02 -7.07322896e-01 4.86797802e-02
-2.39633191e-02 1.93058208e-01 -8.96309733e-01 -5.20759940e-01
-6.88899219e-01 -6.31578863e-02 1.03980350e+00 6.11489654e-01
1.49364543e+00 -4.70895320e-01 -2.69418418e-01 9.86420870e-01
1.33035898e+00 -3.37265670e-01 6.21372640e-01 8.92791674e-02
1.27283585e+00 2.51254141e-01 2.37804465e-02 2.19874769e-01
5.40683031e-01 2.89355874e-01 1.05039942e+00 -4.32153434e-01
-4.16238248e-01 -3.09054077e-01 8.50549415e-02 9.42977786e-01
-3.31061453e-01 -9.15968716e-01 -9.46387172e-01 6.07150853e-01
-1.84886348e+00 -6.36770487e-01 -4.99631971e-01 1.56484544e+00
2.12844238e-01 1.91206768e-01 -2.30962969e-02 -1.63460270e-01
7.72364914e-01 6.60538018e-01 -6.55974388e-01 -4.99672443e-01
-2.01645307e-02 2.97394186e-01 7.04904377e-01 6.71715081e-01
-1.01756549e+00 1.15035236e+00 6.24981833e+00 5.55190921e-01
-9.77077246e-01 -1.58446804e-02 6.18054509e-01 1.12998575e-01
-6.63507402e-01 -1.47026509e-01 -5.38711905e-01 2.40859494e-01
1.00967968e+00 -4.30860698e-01 6.48974359e-01 7.69820511e-01
-2.73138225e-01 6.84953153e-01 -1.00086486e+00 5.69110751e-01
-1.69644326e-01 -1.58383465e+00 1.65455878e-01 2.59539425e-01
6.35716558e-01 5.93744040e-01 6.09355979e-02 7.22126961e-01
6.67867720e-01 -1.36023211e+00 4.17135209e-01 2.67443210e-01
8.17132354e-01 -7.81718135e-01 7.86191642e-01 -1.07007124e-01
-1.41286612e+00 -5.32131307e-02 -5.04541874e-01 -3.95836264e-01
-4.40424569e-02 5.59036493e-01 -7.24836528e-01 6.88272297e-01
7.02316940e-01 1.09130478e+00 -6.24211788e-01 6.47364497e-01
-4.68825281e-01 7.38636672e-01 -1.81808919e-01 -3.59512806e-01
5.83663881e-01 -3.67972255e-02 6.64959133e-01 1.20251930e+00
1.83022246e-01 1.65525563e-02 -9.33914483e-02 1.20375574e+00
-6.39147460e-01 -9.16423574e-02 -7.67258763e-01 -4.52815264e-01
3.61073315e-01 1.26924551e+00 -6.44802213e-01 -2.22077996e-01
-3.96993488e-01 8.28860700e-01 9.58081484e-01 7.11786985e-01
-1.07888055e+00 -6.92724943e-01 9.96838629e-01 2.68528521e-01
4.00156498e-01 -1.41618669e-01 -1.00486517e-01 -9.31654572e-01
-3.23133141e-01 -5.70671678e-01 6.67922735e-01 -6.40335321e-01
-1.44992602e+00 9.68656123e-01 -3.50729674e-01 -4.77395952e-01
2.19117656e-01 -6.34281635e-01 -9.24702764e-01 8.27286422e-01
-1.55088508e+00 -1.04954255e+00 -6.36692524e-01 3.93523842e-01
1.55142322e-01 9.79597494e-03 5.35381794e-01 3.97267520e-01
-7.89501011e-01 8.90305221e-01 -2.91616231e-01 3.68682384e-01
2.00283810e-01 -1.47918367e+00 1.34655428e+00 7.52566040e-01
6.81634173e-02 6.17076397e-01 5.74767232e-01 -5.30548275e-01
-1.55485475e+00 -1.47720611e+00 8.20393920e-01 -4.57937449e-01
9.51851130e-01 -4.26971316e-01 -1.11185384e+00 1.13789928e+00
3.04031104e-01 4.28065300e-01 3.42527002e-01 5.01101494e-01
-4.05568302e-01 2.55345609e-02 -7.73097217e-01 7.14151382e-01
1.41036975e+00 -4.48861271e-01 -2.61070542e-02 5.25542676e-01
1.44255280e+00 -4.39356029e-01 -1.03948414e+00 3.07946414e-01
2.58662373e-01 -9.79416668e-01 9.22608554e-01 -9.09878671e-01
5.01173019e-01 7.19614625e-02 1.62113890e-01 -1.42568481e+00
-8.47975373e-01 -8.66509974e-01 -4.01204854e-01 9.01401818e-01
4.38881695e-01 -1.09763801e+00 8.10124993e-01 3.46230835e-01
-3.96046638e-01 -8.43771696e-01 -4.67340678e-01 -5.62330365e-01
1.48971692e-01 -2.74733722e-01 9.59910989e-01 1.07580602e+00
-1.67892545e-01 5.50867558e-01 -2.44261086e-01 4.36594784e-01
3.94668609e-01 -1.12689845e-01 9.37076807e-01 -1.20640087e+00
-3.21174145e-01 -7.21762598e-01 -5.92831314e-01 -1.16863406e+00
2.95082331e-01 -1.22036266e+00 -2.91841626e-01 -1.74797249e+00
-7.23741436e-03 -4.56593990e-01 -6.08715475e-01 5.39673388e-01
-4.55048203e-01 7.43076429e-02 2.10837588e-01 -1.21657774e-01
-6.98141634e-01 6.78502321e-01 1.38670480e+00 -2.00666517e-01
-4.69224676e-02 -1.13835596e-01 -1.12624943e+00 3.87283683e-01
9.44819689e-01 -4.83131409e-01 -4.73751992e-01 -6.89024985e-01
4.07715648e-01 -1.25507265e-01 4.78519529e-01 -8.89916897e-01
3.13189715e-01 1.55781835e-01 -1.67516153e-02 -3.95269692e-01
1.51250931e-02 -6.35379612e-01 4.17888612e-02 5.54393113e-01
-3.85497361e-01 4.01418358e-01 3.00632119e-01 8.27062249e-01
-1.22506609e-02 2.49810934e-01 5.38724482e-01 -3.57050970e-02
-5.97499788e-01 9.33262169e-01 -1.54162556e-01 3.67292613e-01
7.38599479e-01 1.94475546e-01 -7.30273247e-01 -7.21598387e-01
-4.38545376e-01 4.75762516e-01 2.59192020e-01 3.75269026e-01
4.55306232e-01 -1.39188039e+00 -7.75823832e-01 3.81808549e-01
7.74453580e-02 2.59257108e-01 3.44133526e-01 9.34799969e-01
-8.51883888e-01 4.11675900e-01 5.11247925e-02 -4.30224568e-01
-8.23692143e-01 7.67908156e-01 4.06621665e-01 -4.85772073e-01
-9.18792486e-01 1.13834059e+00 5.21117926e-01 -4.56179380e-01
2.91917950e-01 -6.77824795e-01 -1.05037764e-01 -4.38020587e-01
2.74524778e-01 3.62549961e-01 -7.93912541e-03 -2.75445789e-01
-3.21347982e-01 1.61915272e-01 -1.06225394e-01 6.93256438e-01
1.38439357e+00 2.22970825e-02 -2.43533090e-01 -3.11017726e-02
1.16989231e+00 -5.82871400e-02 -1.33485317e+00 -1.27740279e-01
-1.79199100e-01 -1.45131886e-01 1.13092288e-01 -3.95372450e-01
-1.65024376e+00 8.70153666e-01 -1.97200969e-01 8.74541223e-01
9.66831923e-01 9.53362659e-02 9.32681918e-01 2.97235966e-01
-3.55310775e-02 -3.69545430e-01 -4.87470776e-02 7.98134983e-01
1.04848099e+00 -1.16126883e+00 -4.11387049e-02 -3.71480763e-01
-3.60319108e-01 8.82762611e-01 7.88010418e-01 -3.34622383e-01
9.22168136e-01 2.05131561e-01 -2.22263396e-01 -8.33535016e-01
-9.02624249e-01 -2.92722166e-01 3.18122059e-01 5.72746456e-01
3.85106385e-01 1.77330866e-01 1.63388327e-01 5.47775745e-01
-1.85901552e-01 -4.79507804e-01 4.49351907e-01 3.75819564e-01
-2.56907821e-01 -6.01593554e-01 1.70860693e-01 4.39410925e-01
-5.56874752e-01 -5.37527204e-01 -5.31747222e-01 1.08866668e+00
-6.21908247e-01 6.50423169e-01 1.76743999e-01 -4.43249106e-01
2.35334367e-01 -5.87373734e-01 2.18483880e-02 -5.22410035e-01
-8.03910851e-01 -4.55176741e-01 2.19107330e-01 -6.88086629e-01
4.31434400e-02 -2.66311094e-02 -1.18496525e+00 -8.97495806e-01
-2.62049526e-01 1.89033002e-01 2.50868589e-01 3.29860926e-01
8.10843050e-01 1.23555362e+00 2.93385446e-01 -8.51041377e-01
-4.32082593e-01 -1.05137229e+00 -5.57971895e-01 4.60937381e-01
5.72069108e-01 -5.04817307e-01 -6.29067540e-01 -7.01570272e-01]
|
[6.980536460876465, 6.231955051422119]
|
82798c3b-8c70-43a5-a2e6-a39a6f57d322
|
an-integrated-platform-for-live-3d-human
|
1712.03084
| null |
http://arxiv.org/abs/1712.03084v1
|
http://arxiv.org/pdf/1712.03084v1.pdf
|
An Integrated Platform for Live 3D Human Reconstruction and Motion Capturing
|
The latest developments in 3D capturing, processing, and rendering provide
means to unlock novel 3D application pathways. The main elements of an
integrated platform, which target tele-immersion and future 3D applications,
are described in this paper, addressing the tasks of real-time capturing,
robust 3D human shape/appearance reconstruction, and skeleton-based motion
tracking. More specifically, initially, the details of a multiple RGB-depth
(RGB-D) capturing system are given, along with a novel sensors' calibration
method. A robust, fast reconstruction method from multiple RGB-D streams is
then proposed, based on an enhanced variation of the volumetric Fourier
transform-based method, parallelized on the Graphics Processing Unit, and
accompanied with an appropriate texture-mapping algorithm. On top of that,
given the lack of relevant objective evaluation methods, a novel framework is
proposed for the quantitative evaluation of real-time 3D reconstruction
systems. Finally, a generic, multiple depth stream-based method for accurate
real-time human skeleton tracking is proposed. Detailed experimental results
with multi-Kinect2 data sets verify the validity of our arguments and the
effectiveness of the proposed system and methodologies.
|
['IEEE', 'Senior Member', 'Georgios Louizis', 'Dimitrios Zarpalas', 'Petros Daras', 'Olga Zoidi', 'Dimitrios S. Alexiadis', 'Nikolaos Zioulis', 'Anargyros Chatzitofis']
|
2017-12-08
| null | null | null | null |
['3d-human-reconstruction']
|
['computer-vision']
|
[ 3.34071279e-01 -2.01055676e-01 2.23549664e-01 -1.92394704e-01
-5.17172039e-01 -1.01414233e-01 2.11796433e-01 8.57990459e-02
-5.44433832e-01 3.33758831e-01 -2.13634241e-02 1.92678332e-01
-1.77375555e-01 -6.48015320e-01 -2.75313526e-01 -4.69802886e-01
1.86280627e-02 6.58674598e-01 5.13722360e-01 -3.58713120e-01
1.07095070e-01 9.41366434e-01 -1.91900146e+00 -6.08625896e-02
3.24394494e-01 1.28054786e+00 -1.80867344e-01 7.32209623e-01
-4.20077629e-02 -3.06539927e-02 -2.90742129e-01 -2.07667947e-01
3.95951360e-01 -3.84240955e-01 -3.14194441e-01 4.21722203e-01
3.16862851e-01 -6.69594586e-01 -8.91807526e-02 4.20513809e-01
1.04677618e+00 1.42242208e-01 5.33798277e-01 -9.18479204e-01
4.51910555e-01 -4.37003344e-01 -3.54775250e-01 -2.64590800e-01
1.10644448e+00 6.54267296e-02 6.97736666e-02 -8.89078438e-01
6.21200979e-01 1.13133872e+00 7.83612370e-01 8.11966181e-01
-8.21861923e-01 -2.17632100e-01 -2.85510570e-01 -1.58824325e-01
-1.22231138e+00 -2.83010513e-01 9.70241070e-01 -4.31958228e-01
7.83300757e-01 6.23999774e-01 1.44508934e+00 9.16287422e-01
2.15204790e-01 4.28400785e-01 1.11852264e+00 -6.31052196e-01
2.73144513e-01 1.38634101e-01 1.26760621e-02 7.65988290e-01
2.30575264e-01 1.71803266e-01 -6.90165699e-01 -4.42569852e-02
1.32708681e+00 1.55739456e-01 -2.73271292e-01 -9.60880160e-01
-8.98400426e-01 1.54721618e-01 -6.57019466e-02 2.00464115e-01
-5.33598840e-01 8.03961679e-02 4.97642010e-01 -1.88990712e-01
6.32929504e-01 -5.19530952e-01 -1.63536757e-01 -5.06974757e-01
-8.01511884e-01 3.38491946e-01 5.88331163e-01 8.81855607e-01
3.91602933e-01 -1.47968441e-01 2.77716458e-01 3.18039924e-01
7.23277748e-01 7.14646697e-01 4.18983489e-01 -8.94620240e-01
3.41988623e-01 8.15060914e-01 2.41053347e-02 -9.94404256e-01
-5.58498383e-01 1.29578814e-01 -5.85933089e-01 6.94416106e-01
2.56075382e-01 2.27748290e-01 -4.45787936e-01 8.96494746e-01
1.07132113e+00 -7.49827698e-02 -2.27995247e-01 1.10510814e+00
7.43490160e-01 1.11240901e-01 -3.25931638e-01 -2.19587848e-01
1.25970674e+00 -2.76500344e-01 -7.30729401e-01 3.34648281e-01
1.48942143e-01 -4.90789205e-01 1.14988983e+00 5.49750626e-01
-1.37435889e+00 -6.52361095e-01 -8.81496012e-01 -9.97225717e-02
-6.65726811e-02 -2.10405648e-01 2.45685026e-01 1.20405209e+00
-8.46247137e-01 4.87679124e-01 -1.05484748e+00 -6.61035299e-01
5.23952208e-02 4.81944323e-01 -4.33361113e-01 -2.58212574e-02
-6.13209784e-01 8.94997060e-01 -1.88157416e-03 4.19453114e-01
-3.99381548e-01 -4.53727514e-01 -7.33073235e-01 -5.52090764e-01
-1.21899016e-01 -1.07682288e+00 9.45078075e-01 -5.29319882e-01
-2.11619592e+00 1.33545673e+00 -1.29547924e-01 1.25446077e-02
1.02281559e+00 -4.71355796e-01 -2.01602742e-01 6.09752476e-01
-1.69431165e-01 4.59057689e-02 8.34047198e-01 -1.16755569e+00
-3.21682513e-01 -1.01958156e+00 -2.30743274e-01 4.16488945e-01
-2.19468087e-01 1.40609052e-02 -6.91485524e-01 -2.54875064e-01
5.45033932e-01 -6.19327426e-01 -2.63438761e-01 6.20724440e-01
1.22082181e-01 3.08004141e-01 6.18338645e-01 -6.56755865e-01
9.49234545e-01 -2.22320819e+00 2.74991840e-01 4.49964881e-01
2.15193629e-01 1.13610461e-01 5.94640791e-01 2.40382269e-01
4.43549231e-02 -5.06635070e-01 -3.62547785e-01 -7.65119672e-01
-1.57419834e-02 2.28867516e-01 1.75934196e-01 9.05969262e-01
-3.11979145e-01 4.07176316e-01 -5.90541661e-01 -6.21573985e-01
9.11292732e-01 8.74591172e-01 -1.05308749e-01 2.57551044e-01
2.73723811e-01 7.05323875e-01 -3.92637193e-01 1.02504134e+00
6.69317722e-01 3.07616979e-01 2.48272554e-04 -1.81320608e-01
-4.12719011e-01 -1.10951260e-01 -1.65150118e+00 2.12338924e+00
-3.30958635e-01 6.91678971e-02 3.72108668e-01 -6.10264063e-01
1.08460617e+00 5.51626027e-01 9.45896387e-01 -8.61116290e-01
3.02057028e-01 5.66011071e-01 -9.49186563e-01 -8.63334417e-01
5.10543644e-01 -3.28222007e-01 1.72378883e-01 4.12744671e-01
-3.58609676e-01 -2.33936474e-01 -3.19614470e-01 -1.59251362e-01
6.49716437e-01 8.63508165e-01 2.80056417e-01 -9.22844745e-03
5.99803984e-01 6.06501997e-02 1.21205397e-01 1.10559992e-01
-3.42701226e-01 9.08799767e-01 -1.22398280e-01 -4.92886573e-01
-9.53872681e-01 -1.05613840e+00 -1.22926228e-01 4.29839343e-01
3.44895124e-01 -3.33995044e-01 -7.54186392e-01 -1.38872787e-01
1.96124151e-01 -6.13285489e-02 -4.85417038e-01 2.92159975e-01
-7.23181725e-01 -6.12654984e-01 3.76953512e-01 3.40072870e-01
3.47268641e-01 -5.57134449e-01 -1.66481793e+00 3.51726562e-01
6.04842678e-02 -9.93402541e-01 7.32713342e-02 -1.20783225e-01
-1.60469019e+00 -1.14914334e+00 -9.33818102e-01 -2.64502794e-01
5.05577683e-01 2.34455854e-01 8.49997938e-01 8.68534818e-02
-5.72499990e-01 1.24171579e+00 -3.87565583e-01 -1.85879722e-01
-2.07682271e-02 -5.71061373e-01 2.96047717e-01 1.39034158e-02
-6.14990033e-02 -6.18773520e-01 -7.23031282e-01 3.66068542e-01
-8.18423927e-01 -8.67222995e-03 2.07934454e-01 1.67942926e-01
8.32922101e-01 -3.29332441e-01 -2.24972501e-01 -2.94954509e-01
3.28799427e-01 1.01160258e-01 -5.54245532e-01 3.81705947e-02
-5.38064241e-01 -4.41038489e-01 -2.51164604e-02 -3.26687515e-01
-8.98798108e-01 4.41227823e-01 -4.33107734e-01 -4.56036896e-01
-2.69260108e-01 -1.53257465e-02 -3.79710376e-01 -2.53142834e-01
5.98415971e-01 3.72576892e-01 5.56588888e-01 -7.45457649e-01
2.32598096e-01 6.51903391e-01 7.20225275e-01 -6.70309424e-01
5.64011574e-01 1.14575279e+00 1.82541028e-01 -1.04859066e+00
2.65216585e-02 -7.45304942e-01 -1.21044719e+00 -8.83621991e-01
7.48540342e-01 -7.45596945e-01 -7.24143207e-01 7.47682214e-01
-1.25465381e+00 -2.00305611e-01 -6.78617716e-01 6.76170468e-01
-8.84089053e-01 5.90622365e-01 -3.62040102e-01 -1.26901448e+00
-3.32711041e-01 -1.04210651e+00 1.38713157e+00 6.76799193e-03
-2.17862770e-01 -1.01243401e+00 4.03777361e-01 5.96886873e-01
1.64120317e-01 9.77131009e-01 3.08359772e-01 2.19678164e-01
-3.12245160e-01 -5.36183476e-01 2.73703873e-01 1.36503667e-01
8.49753469e-02 1.48883834e-01 -1.04487503e+00 -9.23216417e-02
3.91862124e-01 3.34645212e-02 1.82821434e-02 3.74169081e-01
5.38322210e-01 2.80137300e-01 -3.98551464e-01 7.28411436e-01
1.50816035e+00 -3.16852331e-03 6.66185439e-01 6.04678392e-01
6.19963825e-01 7.28351891e-01 7.95180500e-01 8.76062214e-01
3.72213691e-01 9.65162635e-01 6.47142172e-01 -1.84312880e-01
-2.31347039e-01 1.03465788e-01 4.37650532e-01 1.10064697e+00
-7.29888022e-01 4.49025363e-01 -7.08410442e-01 -1.68408621e-02
-1.47178853e+00 -6.29465282e-01 -6.95500374e-01 2.57806015e+00
3.67365956e-01 1.41386196e-01 6.00051463e-01 8.20277572e-01
4.57134217e-01 -3.70606750e-01 -2.78107464e-01 -2.54290491e-01
1.35519877e-01 5.76165855e-01 3.39839011e-01 5.58861315e-01
-6.66802227e-01 2.62297958e-01 6.27365828e+00 2.68414408e-01
-1.04376817e+00 2.79076010e-01 -1.20208368e-01 -1.83411643e-01
-1.89109802e-01 -3.40304554e-01 -7.43660092e-01 2.77818710e-01
6.96915984e-01 1.09597869e-01 -1.05632804e-01 6.82946861e-01
6.55302703e-01 -4.31409150e-01 -7.84712851e-01 1.22359562e+00
3.35243851e-01 -8.08129787e-01 -2.52108276e-02 1.91692784e-01
2.83888020e-02 -3.77570331e-01 -2.46600226e-01 -2.64080793e-01
-7.26516306e-01 -3.89111221e-01 1.15336156e+00 8.26606333e-01
9.10569668e-01 -6.83504164e-01 3.93461525e-01 3.92063498e-01
-1.30207801e+00 4.37747121e-01 -1.48874059e-01 -2.15230301e-01
5.05056560e-01 7.09722459e-01 -5.27651370e-01 9.29240704e-01
8.54833663e-01 5.79130411e-01 -3.01258981e-01 1.05569232e+00
-9.87328365e-02 -1.71246737e-01 -5.82003057e-01 -1.19532831e-02
-3.56926292e-01 -2.62490988e-01 5.23816884e-01 1.14825153e+00
3.62621367e-01 4.43904996e-01 -1.69232905e-01 5.22486925e-01
7.66725004e-01 2.91626424e-01 -5.64211965e-01 5.45115352e-01
-1.10386364e-01 1.08761716e+00 -8.28771293e-01 -1.57828823e-01
-3.09488088e-01 1.20819223e+00 -3.31716150e-01 -3.66592743e-02
-8.63064528e-01 -1.98627919e-01 3.43577862e-01 7.95068145e-01
-2.22415954e-01 -7.14669645e-01 -4.99348432e-01 -1.08332789e+00
2.88230568e-01 -4.13079023e-01 3.17012817e-01 -8.71295094e-01
-8.14173698e-01 3.70972157e-01 2.37301752e-01 -1.69039118e+00
-1.17136851e-01 -8.11577260e-01 -1.03672564e-01 7.70220637e-01
-1.37846529e+00 -1.11657667e+00 -6.82123363e-01 1.07669473e+00
3.83982599e-01 1.91785842e-01 1.01968300e+00 5.18098116e-01
-4.40219492e-01 1.73717365e-01 4.18873243e-02 -3.82572919e-01
3.99597824e-01 -9.40169096e-01 1.17014483e-01 6.74233675e-01
-3.59864414e-01 3.86557668e-01 5.12824535e-01 -7.73057044e-01
-1.91956830e+00 -4.24381047e-01 5.15574694e-01 -6.43823326e-01
1.10748127e-01 -3.25673252e-01 -5.58318317e-01 2.92884141e-01
-5.24348259e-01 4.87033501e-02 6.32821023e-01 -4.64907616e-01
1.69774473e-01 -1.61234125e-01 -1.56481218e+00 2.22638756e-01
1.12487972e+00 -3.71497989e-01 -6.49547637e-01 1.29880935e-01
2.38223761e-01 -1.02521992e+00 -1.09820068e+00 2.39027470e-01
1.17049503e+00 -1.54082835e+00 1.20736885e+00 1.57383412e-01
-3.96622233e-02 -2.93691546e-01 -2.94770300e-01 -4.09293234e-01
2.07257673e-01 -5.58844745e-01 -3.65501016e-01 7.20326304e-01
-4.23243403e-01 -4.68028635e-01 9.77960885e-01 7.79868960e-01
-1.74148455e-01 -5.51600158e-01 -1.42014277e+00 -6.59615576e-01
-6.75514042e-01 -8.01615596e-01 2.13830560e-01 4.41699922e-01
5.57270795e-02 -2.02624157e-01 -7.04326034e-02 1.66858345e-01
1.06688726e+00 -8.36512223e-02 1.12974060e+00 -1.35203040e+00
-1.20076194e-01 -2.30459362e-01 -7.82668114e-01 -1.00011230e+00
-5.86291790e-01 -3.17992240e-01 -1.77281708e-01 -1.55582571e+00
-3.11566979e-01 -2.61371553e-01 1.77368805e-01 -8.43112916e-02
2.64632314e-01 4.30198580e-01 -3.63531485e-02 4.24118400e-01
-2.76798338e-01 5.29152095e-01 1.21527648e+00 4.56800818e-01
-2.20001474e-01 3.18617344e-01 2.02551320e-01 6.80873156e-01
2.39388362e-01 -3.81042808e-01 -2.33034909e-01 -3.31329465e-01
7.14371502e-02 2.41604671e-01 6.36613786e-01 -1.28562331e+00
1.73803359e-01 2.44874239e-01 4.44501579e-01 -8.79519820e-01
8.90774846e-01 -1.31881690e+00 4.81279582e-01 8.99499536e-01
3.65196228e-01 2.54530042e-01 1.02684885e-01 3.94203305e-01
8.85170251e-02 -3.95124443e-02 7.78690040e-01 -3.12900215e-01
-5.50573528e-01 1.93776339e-01 -2.08571211e-01 -3.61608207e-01
1.20726717e+00 -1.23950458e+00 5.31836808e-01 -2.82888394e-02
-1.03143895e+00 -3.87083948e-01 6.95199311e-01 -6.66032508e-02
1.11298895e+00 -1.41388178e+00 -4.53251392e-01 5.55025697e-01
2.32072435e-02 2.36080974e-01 3.38051736e-01 8.30324233e-01
-1.04491198e+00 1.85553446e-01 -5.69644272e-01 -9.39210474e-01
-1.44922435e+00 1.41279772e-01 4.83728886e-01 1.02500081e-01
-9.26556230e-01 4.11043912e-01 -4.39538211e-01 -3.11360985e-01
2.39680439e-01 -4.51938927e-01 -2.77379882e-02 -8.93224031e-02
3.83730322e-01 8.81127059e-01 4.14681226e-01 -8.20917785e-01
-5.00913858e-01 1.25045872e+00 8.89922261e-01 -6.59452617e-01
1.25640833e+00 -5.46790957e-01 -7.06491321e-02 7.86418736e-01
7.91510880e-01 -6.75923051e-03 -1.23418725e+00 1.16276428e-01
-1.88213795e-01 -7.46367514e-01 -2.23741278e-01 -4.07283664e-01
-8.50861251e-01 1.02985740e+00 1.02775168e+00 -4.93941046e-02
1.25672662e+00 -3.67613137e-01 7.40254819e-01 -3.69207412e-01
8.79972398e-01 -8.54323685e-01 -3.66345909e-03 2.70406064e-02
6.88404322e-01 -8.96298230e-01 3.73293161e-01 -6.17924631e-01
-8.59763399e-02 1.39758563e+00 3.16613227e-01 8.48837867e-02
5.51991105e-01 3.36448789e-01 8.53001997e-02 -4.72590655e-01
8.34835395e-02 -7.18458146e-02 8.68562087e-02 8.82420659e-01
3.59065592e-01 -9.02894437e-02 -4.10825819e-01 3.47020388e-01
7.46911243e-02 2.83859909e-01 2.51236409e-01 1.32651699e+00
-4.35907990e-01 -1.05772972e+00 -1.00819516e+00 -2.69794941e-01
-1.80171862e-01 5.53437948e-01 -1.71485931e-01 1.06634963e+00
1.79264203e-01 4.54784036e-01 -1.30237848e-01 -2.93921739e-01
1.07080305e+00 -5.46121178e-03 1.09226692e+00 -3.94511133e-01
-6.79859996e-01 2.96147436e-01 -2.62128383e-01 -8.77307713e-01
-8.14052284e-01 -8.52394938e-01 -1.19714177e+00 -2.33931229e-01
-8.43976587e-02 -2.05544233e-01 1.28395855e+00 6.50022209e-01
6.88342899e-02 1.07149512e-01 4.35352117e-01 -1.42607796e+00
-2.33710915e-01 -4.54048216e-01 -7.70081341e-01 4.01640415e-01
1.64145067e-01 -7.82417893e-01 -2.65541583e-01 1.60656661e-01]
|
[7.2633490562438965, -1.042111873626709]
|
b06c38bd-b4a0-49c9-b5b8-97a39e19b6db
|
dialoguernn-an-attentive-rnn-for-emotion
|
1811.00405
| null |
https://arxiv.org/abs/1811.00405v4
|
https://arxiv.org/pdf/1811.00405v4.pdf
|
DialogueRNN: An Attentive RNN for Emotion Detection in Conversations
|
Emotion detection in conversations is a necessary step for a number of applications, including opinion mining over chat history, social media threads, debates, argumentation mining, understanding consumer feedback in live conversations, etc. Currently, systems do not treat the parties in the conversation individually by adapting to the speaker of each utterance. In this paper, we describe a new method based on recurrent neural networks that keeps track of the individual party states throughout the conversation and uses this information for emotion classification. Our model outperforms the state of the art by a significant margin on two different datasets.
|
['Soujanya Poria', 'Navonil Majumder', 'Erik Cambria', 'Devamanyu Hazarika', 'Rada Mihalcea', 'Alexander Gelbukh']
|
2018-11-01
| null | null | null | null |
['multimodal-emotion-recognition', 'emotion-recognition-in-conversation', 'multimodal-emotion-recognition']
|
['computer-vision', 'natural-language-processing', 'speech']
|
[ 4.71084751e-02 1.48062631e-01 -3.00459713e-01 -6.49563670e-01
-4.53309745e-01 -5.51209092e-01 7.39044785e-01 4.45477426e-01
-2.06547156e-01 6.44340098e-01 6.44051611e-01 -3.14495414e-01
5.49897730e-01 -5.83882511e-01 -1.59573015e-02 -6.30403936e-01
2.83219386e-02 3.75080317e-01 8.47032145e-02 -7.34058499e-01
3.30425262e-01 1.64494753e-01 -1.60912192e+00 8.78148198e-01
4.56063807e-01 1.27824080e+00 -2.23574638e-01 8.95597994e-01
-6.93850815e-01 1.24968636e+00 -9.39252317e-01 -6.56058788e-01
-4.82888937e-01 -8.55892181e-01 -1.10173857e+00 2.73160845e-01
-5.18088460e-01 1.25216976e-01 2.16275215e-01 9.06630874e-01
3.48664820e-01 2.39075735e-01 3.80617559e-01 -1.12723756e+00
1.28922343e-01 1.09304893e+00 -3.67205292e-01 2.97793239e-01
6.25706017e-01 -6.75752938e-01 1.08531940e+00 -4.34389830e-01
6.71300769e-01 1.21552491e+00 4.65914249e-01 5.54050565e-01
-6.77796900e-01 -5.82382321e-01 5.58356643e-01 3.65588933e-01
-5.48147380e-01 -5.19076347e-01 1.07009268e+00 -1.96118146e-01
9.92442966e-01 4.54191178e-01 9.12895560e-01 1.19387984e+00
2.72966176e-01 1.11090028e+00 1.08683467e+00 -4.69091207e-01
3.86687070e-01 5.45437455e-01 6.34078681e-01 4.51476961e-01
-7.65925705e-01 -6.14864945e-01 -7.82766700e-01 -6.18425667e-01
-1.23948917e-01 -1.41344994e-01 5.21790981e-02 2.25737885e-01
-6.47558868e-01 1.25430024e+00 -1.24188893e-01 4.70998347e-01
-7.30506063e-01 -3.51781785e-01 9.33433890e-01 8.37385654e-01
1.14100552e+00 1.98243886e-01 -7.13879704e-01 -7.98611999e-01
-4.99038994e-01 8.92525390e-02 1.67666447e+00 2.33867154e-01
5.53108990e-01 -3.26894850e-01 6.50762464e-04 9.73012805e-01
1.44885764e-01 -1.38999596e-01 8.94285083e-01 -8.45271230e-01
1.06185958e-01 8.97828043e-01 1.09387703e-01 -9.04301643e-01
-5.70648253e-01 -8.23505744e-02 -7.91909277e-01 -2.31509045e-01
1.11637652e-01 -8.89543355e-01 -4.70250636e-01 1.47562623e+00
6.56401575e-01 -2.15073511e-01 1.60187855e-01 4.79378134e-01
9.03816223e-01 9.73916709e-01 -1.62652358e-01 -7.66992927e-01
1.43962646e+00 -1.07233155e+00 -1.18689573e+00 -2.20919639e-01
5.19197464e-01 -7.77716041e-01 3.58290017e-01 7.89295495e-01
-1.00211108e+00 -7.84170628e-02 -7.68833160e-01 2.69520044e-01
-4.17362183e-01 -1.27621323e-01 9.53561485e-01 6.41862869e-01
-7.39686847e-01 3.72216851e-01 -5.88721693e-01 -4.01592672e-01
-1.80910118e-02 3.69894654e-01 -4.35567684e-02 7.09027410e-01
-1.39021873e+00 8.83944213e-01 -8.40362757e-02 2.58285135e-01
-1.89765677e-01 2.49550685e-01 -7.89681435e-01 1.13188820e-02
2.51790226e-01 -2.87570301e-02 1.74393082e+00 -1.51909554e+00
-2.38470459e+00 7.34747350e-01 -7.12674975e-01 -4.96588171e-01
2.76859730e-01 -2.46894613e-01 -6.28873050e-01 -2.50084251e-01
-4.01016772e-01 1.18690655e-01 7.97445714e-01 -7.53460586e-01
-8.01227450e-01 -3.74116033e-01 2.10154846e-01 3.66983712e-01
-2.14174405e-01 5.04604399e-01 -3.23510379e-01 1.19530717e-02
-2.32492648e-02 -9.97933388e-01 -3.16648692e-01 -6.73247755e-01
-2.70597786e-01 -8.70866179e-01 1.02482820e+00 -4.01787192e-01
1.09626412e+00 -2.00360870e+00 2.11067691e-01 2.59506673e-01
2.72451602e-02 1.22504286e-01 2.03563318e-01 5.34994304e-01
2.43354272e-02 -3.95434089e-02 3.28638554e-01 -4.81243074e-01
-9.53881163e-03 1.59520075e-01 -3.56064320e-01 3.20493460e-01
-6.12004735e-02 6.54140472e-01 -7.28886068e-01 -3.31006855e-01
-5.96821532e-02 3.94468993e-01 -2.27392465e-01 4.15932029e-01
-3.16157669e-01 4.08865988e-01 -4.91264135e-01 1.03817426e-01
8.71788859e-02 -3.68242413e-01 6.55246735e-01 1.47375733e-01
-2.36344188e-01 7.70912707e-01 -9.54455256e-01 1.37230897e+00
-6.73525870e-01 8.98756027e-01 3.69701385e-01 -1.07641113e+00
1.02318323e+00 8.11192632e-01 4.82146680e-01 -3.20821434e-01
6.53332710e-01 -2.67851762e-02 1.78363115e-01 -5.23093581e-01
7.27110684e-01 -1.62155598e-01 -3.77167910e-01 1.00719321e+00
-1.53109565e-01 2.54370719e-01 2.14830264e-01 5.32316975e-02
8.47117245e-01 -5.97975612e-01 5.13913274e-01 2.33286723e-01
7.33357012e-01 -3.01908076e-01 5.09750247e-01 4.19496298e-01
-4.04620916e-01 1.82799965e-01 9.96316791e-01 -6.44211471e-01
-4.44451481e-01 -2.50779569e-01 2.59177297e-01 1.53830504e+00
4.36562225e-02 -4.37495917e-01 -5.79545557e-01 -7.64782310e-01
-4.67917204e-01 4.65148300e-01 -9.14077699e-01 7.30218962e-02
-3.76760483e-01 -7.27508664e-01 1.56077251e-01 2.86144197e-01
4.15714651e-01 -1.54652238e+00 -5.65473258e-01 4.73432809e-01
-5.86868167e-01 -9.23307538e-01 -3.28236789e-01 5.31219184e-01
-5.86014867e-01 -7.28289247e-01 -2.65900403e-01 -7.33423114e-01
8.76462981e-02 -7.28488192e-02 1.13109159e+00 5.42259030e-02
2.39600942e-01 3.40582803e-02 -6.42879605e-01 -7.91719496e-01
-8.15205574e-01 5.99802315e-01 -1.71696499e-01 4.22935158e-01
4.57384586e-01 -4.36078787e-01 -2.93870658e-01 3.20586473e-01
-5.62071741e-01 -1.48775801e-01 1.64070398e-01 7.66857445e-01
-5.48406653e-02 -1.17262729e-01 8.58129203e-01 -1.34678781e+00
1.30182755e+00 -7.39110112e-01 -9.21814814e-02 2.55098879e-01
-2.07351848e-01 -4.40964438e-02 4.04378951e-01 -6.55901432e-01
-1.40031922e+00 -2.88745183e-02 -4.63500679e-01 3.12865347e-01
-2.81509966e-01 7.09280610e-01 7.43141696e-02 4.68498021e-01
3.85267258e-01 -1.60319015e-01 7.59770796e-02 -3.65628034e-01
1.78573400e-01 1.18701708e+00 4.93056839e-03 -1.52043715e-01
-2.97141552e-01 4.13674980e-01 -7.20945776e-01 -8.93497825e-01
-1.03427780e+00 -8.44873011e-01 -2.80278295e-01 -4.69374120e-01
7.81776428e-01 -5.00332713e-01 -1.20899415e+00 5.97362399e-01
-1.39644706e+00 -9.14657339e-02 -4.80940975e-02 3.50846142e-01
-2.80638844e-01 1.51862875e-01 -1.03826833e+00 -1.33844185e+00
-6.58694863e-01 -9.55947876e-01 6.53357208e-01 5.49687326e-01
-8.79624486e-01 -1.05195177e+00 4.02814627e-01 3.34175646e-01
4.47594017e-01 5.89718148e-02 4.01624054e-01 -1.12293470e+00
3.90256256e-01 -5.14777839e-01 3.69323432e-01 3.09425622e-01
3.26994568e-01 1.38771027e-01 -1.10266531e+00 3.10438663e-01
2.48148307e-01 -6.52278244e-01 7.65512168e-01 2.83666551e-01
8.13903511e-01 -3.41224939e-01 -1.41249403e-01 -1.49103984e-01
4.76932228e-01 4.50677812e-01 5.34839809e-01 9.83570740e-02
8.29196125e-02 1.03587854e+00 2.96343595e-01 6.69365704e-01
4.03743684e-01 3.43426436e-01 3.61285716e-01 -6.42255996e-04
7.50281036e-01 3.57074231e-01 6.81813776e-01 1.13274407e+00
-7.47290477e-02 -6.02201402e-01 -4.56990808e-01 5.90007722e-01
-2.23066449e+00 -1.15881431e+00 -1.28028721e-01 1.48399675e+00
8.49353254e-01 2.77361810e-01 3.03081989e-01 1.73569307e-01
8.38392317e-01 5.10992587e-01 -5.79009056e-01 -1.44645798e+00
5.75338230e-02 2.09239274e-01 -1.05978355e-01 5.05509913e-01
-1.08825719e+00 6.57589078e-01 6.48295593e+00 2.63474792e-01
-1.33088672e+00 1.72979131e-01 9.34401631e-01 4.71325591e-02
1.67726781e-02 -1.54993221e-01 -5.34898818e-01 1.24300621e-01
1.34055626e+00 -1.91803128e-01 3.12439442e-01 7.92677522e-01
1.94383249e-01 -3.37797612e-01 -8.52207780e-01 7.40204215e-01
3.65572929e-01 -1.31245148e+00 -5.48890412e-01 -4.68020625e-02
7.93436527e-01 2.14746937e-01 -2.24944800e-01 3.23341101e-01
4.71814007e-01 -6.95388675e-01 3.91691417e-01 1.98224649e-01
-1.48969322e-01 -1.03460419e+00 1.15044427e+00 5.20114720e-01
-8.94877553e-01 9.49675068e-02 1.42477602e-01 -5.30838966e-01
6.05948150e-01 6.30329669e-01 -6.42092168e-01 3.15343440e-01
7.20367730e-01 7.06152499e-01 9.83383507e-02 3.17971975e-01
-1.75416455e-01 9.25165474e-01 -3.35886925e-01 -6.01262212e-01
2.76907831e-01 -1.70778275e-01 3.64108980e-01 1.17040265e+00
-7.14331493e-02 2.19987661e-01 1.71946555e-01 1.34330496e-01
-3.70315552e-01 4.41944659e-01 -3.03230017e-01 -1.00784063e-01
1.17734987e-02 1.44245303e+00 -8.18620026e-01 -5.61797082e-01
-3.49649012e-01 1.08871484e+00 2.70952135e-01 6.06750622e-02
-5.24134338e-01 -3.00279468e-01 7.79759049e-01 -5.00869811e-01
3.94999653e-01 1.86439469e-01 9.86727625e-02 -1.17374694e+00
-6.37922212e-02 -1.01853538e+00 4.39076602e-01 -3.71041059e-01
-1.30884826e+00 9.23631608e-01 -5.02092004e-01 -8.36199462e-01
-8.66243005e-01 -3.58923227e-01 -1.23257470e+00 6.53250694e-01
-1.33134460e+00 -5.00754476e-01 3.86889279e-03 4.35618192e-01
7.43532300e-01 4.31574658e-02 1.15156579e+00 -4.12335657e-02
-6.78724408e-01 1.71597496e-01 -6.05005361e-02 4.14285392e-01
6.74988806e-01 -1.12481296e+00 3.32685947e-01 1.37392536e-01
2.02717438e-01 4.01247233e-01 9.57813859e-01 -2.45197847e-01
-1.03623354e+00 -3.28349084e-01 1.30178654e+00 -1.06657363e-01
9.16017175e-01 -3.77569735e-01 -8.60033631e-01 5.48641443e-01
9.21695650e-01 -4.13500875e-01 1.14368165e+00 7.78159261e-01
3.44004110e-02 -2.78034620e-02 -9.06231761e-01 4.22724038e-01
2.55826682e-01 -5.94266772e-01 -5.17217755e-01 2.92913616e-01
4.64014620e-01 -5.78641534e-01 -7.67111838e-01 -1.18467160e-01
9.74326134e-01 -1.05507076e+00 2.12670118e-01 -7.50134468e-01
4.33997929e-01 2.85196543e-01 2.14501828e-01 -1.63269389e+00
3.48209202e-01 -9.05169129e-01 -3.28563839e-01 1.21519315e+00
7.63469517e-01 -6.50412977e-01 9.50547993e-01 7.15774834e-01
1.15413293e-01 -6.50635839e-01 -8.43317568e-01 1.69264421e-01
-2.54533976e-01 -4.98135746e-01 4.88875240e-01 9.73046422e-01
8.90278816e-01 9.46105838e-01 -6.31805599e-01 -3.05119604e-01
-1.17160305e-01 4.08395290e-01 6.39901578e-01 -1.42149293e+00
-2.50449628e-01 -5.17865896e-01 -1.50257140e-01 -1.03433526e+00
3.84695053e-01 -2.44396955e-01 2.96428233e-01 -1.38152623e+00
-2.44799946e-02 6.92792386e-02 -2.14816019e-01 1.29574463e-01
-4.08560038e-02 3.03520299e-02 -8.47390071e-02 -5.35420962e-02
-8.91656995e-01 5.10315001e-01 6.59628928e-01 -2.30684653e-01
-5.56623280e-01 4.88876343e-01 -7.80593097e-01 1.09920084e+00
1.24390960e+00 -3.40162277e-01 -1.67011768e-01 1.25482529e-01
6.72594190e-01 2.84783393e-01 -4.76324856e-01 -3.16603184e-01
2.91634709e-01 -1.35715142e-01 -5.52294366e-02 -6.57459915e-01
4.67628330e-01 -6.16737783e-01 -1.36637628e-01 1.96571320e-01
-8.19278300e-01 2.84661025e-01 -1.21629285e-02 4.88465697e-01
-5.37450194e-01 -2.12917730e-01 4.22843277e-01 -1.33071542e-01
-4.09298688e-01 -2.29957715e-01 -1.09445286e+00 -1.20610833e-01
8.22949529e-01 2.65847445e-01 -1.65905818e-01 -1.23020089e+00
-8.65212560e-01 3.55550230e-01 -5.57888560e-02 6.04884207e-01
2.52805889e-01 -8.28164518e-01 -6.31263554e-01 -9.62824300e-02
-9.10792202e-02 -2.39636853e-01 1.44600287e-01 8.25181425e-01
1.16725452e-01 1.07806074e-02 1.04831390e-01 -4.48607028e-01
-1.86005604e+00 -1.99188292e-02 3.79635751e-01 -8.38945210e-01
-1.59249574e-01 7.51913369e-01 -1.65614203e-01 -4.21284884e-01
3.82060677e-01 -2.70264089e-01 -8.69502604e-01 6.72047079e-01
7.79640734e-01 1.03781343e-01 2.85495192e-01 -6.87785327e-01
-1.48390964e-01 -2.84759086e-02 -4.48061317e-01 -4.04212981e-01
1.40402389e+00 -3.04982930e-01 -3.23268801e-01 1.25686300e+00
1.18464732e+00 -7.75071746e-03 -7.78374791e-01 -2.71663398e-01
1.13861568e-01 2.27456301e-01 -4.81035113e-02 -7.33293414e-01
-1.02341759e+00 6.38989866e-01 1.44646853e-01 1.00828969e+00
8.73804212e-01 3.29936564e-01 9.31337059e-01 7.16015160e-01
1.24149667e-02 -1.51169479e+00 -2.91339424e-03 8.57663274e-01
6.50178730e-01 -1.37886536e+00 -1.71301112e-01 -1.04398862e-01
-1.13914764e+00 1.10706663e+00 1.47008732e-01 9.76674408e-02
1.01920092e+00 2.40252122e-01 6.17587984e-01 -3.40382904e-01
-1.48384011e+00 -1.18563175e-01 5.16875200e-02 6.73754290e-02
9.10543263e-01 4.99793375e-03 -4.91173416e-01 4.78193223e-01
-2.39882082e-01 -1.41701356e-01 5.88295817e-01 1.11096060e+00
-5.22835791e-01 -1.37835801e+00 -1.85738117e-01 6.53339028e-01
-7.99949706e-01 2.13382229e-01 -1.05800796e+00 2.34822199e-01
-1.48640290e-01 1.54738200e+00 2.46172085e-01 -4.53714103e-01
2.31860682e-01 6.07375324e-01 -1.36144549e-01 -4.58733529e-01
-1.23324263e+00 1.13909617e-01 7.37900734e-01 -3.88820708e-01
-1.22179329e+00 -9.18397546e-01 -1.07595468e+00 -3.80744129e-01
-6.53873563e-01 7.94991195e-01 9.40100670e-01 1.27856529e+00
2.02722281e-01 4.29251641e-01 1.12941933e+00 -7.73539960e-01
-1.68610618e-01 -1.34690642e+00 -4.45614338e-01 5.25066197e-01
4.03593093e-01 -3.09831560e-01 -4.95708138e-01 1.05858594e-02]
|
[12.982868194580078, 6.226597309112549]
|
ddf3e7c2-e020-4dcb-8223-94c1247dac13
|
fast-vid2vid-spatial-temporal-compression-for
|
2207.05049
| null |
https://arxiv.org/abs/2207.05049v1
|
https://arxiv.org/pdf/2207.05049v1.pdf
|
Fast-Vid2Vid: Spatial-Temporal Compression for Video-to-Video Synthesis
|
Video-to-Video synthesis (Vid2Vid) has achieved remarkable results in generating a photo-realistic video from a sequence of semantic maps. However, this pipeline suffers from high computational cost and long inference latency, which largely depends on two essential factors: 1) network architecture parameters, 2) sequential data stream. Recently, the parameters of image-based generative models have been significantly compressed via more efficient network architectures. Nevertheless, existing methods mainly focus on slimming network architectures and ignore the size of the sequential data stream. Moreover, due to the lack of temporal coherence, image-based compression is not sufficient for the compression of the video task. In this paper, we present a spatial-temporal compression framework, \textbf{Fast-Vid2Vid}, which focuses on data aspects of generative models. It makes the first attempt at time dimension to reduce computational resources and accelerate inference. Specifically, we compress the input data stream spatially and reduce the temporal redundancy. After the proposed spatial-temporal knowledge distillation, our model can synthesize key-frames using the low-resolution data stream. Finally, Fast-Vid2Vid interpolates intermediate frames by motion compensation with slight latency. On standard benchmarks, Fast-Vid2Vid achieves around real-time performance as 20 FPS and saves around 8x computational cost on a single V100 GPU.
|
['Ziwei Liu', 'Wayne Wu', 'Shikai Li', 'Guangcong Wang', 'Long Zhuo']
|
2022-07-11
| null | null | null | null |
['video-to-video-synthesis', 'motion-compensation']
|
['computer-vision', 'computer-vision']
|
[ 2.39041865e-01 -1.02955863e-01 -9.10629928e-02 -2.08447918e-01
-5.37001371e-01 -1.44202322e-01 6.97253704e-01 -4.03315604e-01
-3.78200799e-01 7.27961838e-01 1.76381052e-01 -3.19961727e-01
7.34462216e-02 -1.04490995e+00 -9.07239199e-01 -5.76206863e-01
1.98978364e-01 2.95964450e-01 4.84852105e-01 1.52623415e-01
6.16545044e-02 2.71709234e-01 -1.88763273e+00 3.94335747e-01
5.28328300e-01 1.20377409e+00 6.37264609e-01 9.65258598e-01
-2.53370970e-01 1.09922409e+00 -5.40353656e-01 -3.82683396e-01
2.84445852e-01 -6.49682462e-01 -6.10917449e-01 -7.95978904e-02
4.62073833e-01 -9.39308941e-01 -9.14711416e-01 9.81953740e-01
5.56571186e-01 7.27574378e-02 1.98648289e-01 -1.41575933e+00
-3.05544913e-01 5.64573407e-01 -3.29912484e-01 1.13348596e-01
8.12121928e-02 4.01521623e-01 5.41879952e-01 -8.43085110e-01
8.94955635e-01 1.24013066e+00 3.48101974e-01 5.24881601e-01
-1.01288819e+00 -6.50779009e-01 -2.07643863e-02 5.95522225e-01
-1.59403205e+00 -6.41040385e-01 4.38756734e-01 -1.60035715e-01
1.04301012e+00 2.65569687e-01 9.23540890e-01 1.07647800e+00
-2.15116981e-02 7.07737446e-01 7.54332662e-01 6.10832274e-02
4.30674583e-01 -4.01937425e-01 -5.01090586e-01 5.77362180e-01
-8.11259001e-02 1.88393027e-01 -9.19683814e-01 1.66968137e-01
1.16278720e+00 -7.04020038e-02 -2.48166487e-01 2.51720399e-01
-1.18883741e+00 4.85951304e-01 3.64002436e-01 -8.45356565e-03
-5.68999767e-01 7.05406070e-01 4.79186237e-01 1.20343693e-01
3.00891340e-01 -2.09205598e-01 2.12821644e-02 -4.80958372e-01
-1.44898570e+00 4.51742411e-01 5.62644482e-01 1.28651917e+00
7.10536957e-01 3.54089648e-01 -2.64563620e-01 4.23041850e-01
1.43195629e-01 6.98044181e-01 3.16164345e-01 -1.45020497e+00
5.84593594e-01 5.16469292e-02 -8.65258798e-02 -1.18531132e+00
-5.48234358e-02 -1.09365709e-01 -1.17568266e+00 9.53563228e-02
1.53908297e-01 -1.40022859e-01 -8.48446250e-01 1.49970698e+00
2.52258450e-01 5.96195400e-01 5.90831228e-02 1.11935961e+00
9.23213601e-01 1.02247810e+00 -8.02154616e-02 -3.45128566e-01
1.14134109e+00 -1.19989681e+00 -7.46159375e-01 -1.41463533e-01
2.80951828e-01 -7.53975511e-01 8.12899172e-01 2.28985131e-01
-1.62887251e+00 -8.39616776e-01 -1.06762981e+00 -3.76409948e-01
9.15858597e-02 -1.87442731e-02 5.31435907e-01 6.88477978e-02
-1.31728685e+00 5.72861493e-01 -9.94839549e-01 1.74417009e-03
3.42490643e-01 1.36747047e-01 -7.77269453e-02 -3.05531204e-01
-1.01252103e+00 2.60732830e-01 5.96816480e-01 1.53567657e-01
-1.00951517e+00 -8.00307810e-01 -7.46604860e-01 2.19385624e-01
4.51726168e-01 -1.03003430e+00 1.11733627e+00 -7.77759612e-01
-1.63847232e+00 2.02223152e-01 -4.73688155e-01 -7.72479594e-01
8.48563254e-01 -1.37076005e-01 -2.39279270e-01 4.74482536e-01
-1.64072528e-01 1.03739870e+00 9.48728919e-01 -8.60505044e-01
-6.94077313e-01 7.88974576e-04 -1.23443328e-01 2.30767533e-01
-1.95148483e-01 -1.94377482e-01 -1.14313293e+00 -6.48867905e-01
-5.19383177e-02 -9.12623048e-01 -1.48346260e-01 1.30611122e-01
-2.49273852e-01 2.47715749e-02 9.66579080e-01 -6.25515342e-01
1.28728151e+00 -2.27524877e+00 8.47817957e-02 -1.05223604e-01
2.73584515e-01 5.79980195e-01 -8.38882774e-02 2.50214875e-01
3.14743429e-01 -7.00404048e-02 -2.11475752e-02 -5.93935788e-01
-1.36172608e-01 3.56110781e-01 -5.45373678e-01 1.33534009e-02
6.01647906e-02 1.10125816e+00 -8.00701380e-01 -7.77492464e-01
3.25942636e-01 8.09869766e-01 -8.27329040e-01 3.01916003e-01
-4.51176941e-01 2.22758338e-01 -2.33542517e-01 3.79042745e-01
7.75573969e-01 -3.80744934e-01 2.24762633e-01 -5.89965463e-01
-2.54248679e-01 2.77037591e-01 -9.79181945e-01 1.99226844e+00
-3.95499736e-01 8.37730169e-01 -1.10752322e-01 -6.05628014e-01
7.36492872e-01 2.97165751e-01 4.39157516e-01 -8.92032266e-01
1.14674181e-01 1.59093276e-01 -3.74312073e-01 -4.40917879e-01
8.68436038e-01 4.17679399e-01 2.91082799e-01 3.15567613e-01
-5.21981828e-02 -1.45932540e-01 4.63720113e-01 2.54045248e-01
1.09058332e+00 6.07852638e-01 -1.69911861e-01 2.42139727e-01
2.26954505e-01 4.88396436e-02 5.43012977e-01 6.01955771e-01
5.96190728e-02 7.41163373e-01 4.69170719e-01 -4.90285724e-01
-1.41323483e+00 -1.03454018e+00 2.58259416e-01 6.73918903e-01
2.38195136e-01 -7.90109396e-01 -8.96709085e-01 -7.51620159e-02
-2.83601820e-01 6.40722215e-01 -1.61654443e-01 -4.64741290e-02
-8.76559734e-01 -6.37266159e-01 6.69345617e-01 5.70142865e-01
9.01532233e-01 -9.36726689e-01 -8.97819757e-01 3.72798920e-01
-4.25314814e-01 -1.54156733e+00 -3.57224464e-01 -3.65927100e-01
-9.26034927e-01 -7.76746094e-01 -5.86731553e-01 -5.87962925e-01
4.47919488e-01 5.30931234e-01 1.13021803e+00 5.09286523e-02
-1.09815694e-01 -2.41694361e-01 -1.16333216e-01 1.10562388e-02
-3.76519322e-01 -7.13667125e-02 -2.67949998e-01 -1.72807679e-01
1.30949035e-01 -8.46078932e-01 -8.79846275e-01 2.21817598e-01
-1.04315317e+00 9.62382436e-01 5.14042735e-01 5.90117157e-01
1.01807153e+00 8.16603005e-02 3.23356688e-01 -4.54485297e-01
2.21634045e-01 -3.86585951e-01 -7.47092903e-01 6.19401671e-02
-4.40918237e-01 -6.12092344e-03 7.82821000e-01 -2.60238439e-01
-1.16562796e+00 1.21878851e-02 -2.91407794e-01 -9.24389005e-01
1.07070729e-01 3.48721445e-01 2.45699659e-02 2.52164871e-01
3.45104188e-01 5.96195519e-01 5.44777773e-02 -2.01275766e-01
3.72655392e-01 4.91790265e-01 9.66608584e-01 -3.66539687e-01
4.95260924e-01 6.28148615e-01 1.72375306e-01 -8.24835181e-01
-4.49159324e-01 1.23946428e-01 -1.65290341e-01 -3.41701239e-01
9.26076055e-01 -1.29501498e+00 -9.16742563e-01 5.75989962e-01
-1.29917276e+00 -6.41829550e-01 -1.69455200e-01 5.72831511e-01
-7.64046133e-01 2.87377954e-01 -8.77550662e-01 -4.62234557e-01
-5.63035429e-01 -1.23657477e+00 1.06581557e+00 1.48410261e-01
1.38687804e-01 -5.38500428e-01 -3.38100702e-01 2.66002268e-01
6.18714690e-01 1.44330546e-01 6.33366764e-01 2.29768097e-01
-1.26401627e+00 1.90191075e-01 -6.24592841e-01 2.25927830e-01
-2.91837275e-01 1.24200657e-01 -7.95858741e-01 -7.96855316e-02
-1.43019050e-01 -1.80634513e-01 8.05941045e-01 4.16354299e-01
1.55620599e+00 -4.07822251e-01 -2.24940367e-02 1.21512663e+00
1.54344463e+00 2.36106455e-01 9.40692186e-01 4.69901115e-02
8.88079762e-01 1.13734111e-01 5.19696116e-01 6.15748286e-01
6.12938404e-01 6.78195417e-01 6.35720372e-01 1.00400865e-01
-5.82220376e-01 -5.54778695e-01 3.62581283e-01 1.17840326e+00
-4.17358875e-01 -5.18728435e-01 -5.37128747e-01 4.53959942e-01
-2.01969481e+00 -1.05345380e+00 -1.43199116e-01 2.06663322e+00
7.15901494e-01 1.78694233e-01 -3.49337794e-02 -1.76474284e-02
5.88252306e-01 3.67160112e-01 -6.21470273e-01 -1.63252503e-01
-2.52980620e-01 1.49633080e-01 5.99571705e-01 4.33118194e-01
-7.02232897e-01 1.12644649e+00 5.77132702e+00 1.04654038e+00
-1.24378610e+00 9.54241231e-02 5.77798009e-01 -5.97755194e-01
-2.27220446e-01 -9.58463326e-02 -8.99610221e-01 7.72132218e-01
1.22517622e+00 -2.15512708e-01 6.44594610e-01 8.27337861e-01
4.11081165e-01 -1.75610468e-01 -9.66121316e-01 1.33771002e+00
2.02440079e-02 -1.89077055e+00 2.79485822e-01 6.20176606e-02
7.58152366e-01 1.79008827e-01 -1.13155946e-01 5.86697683e-02
8.79243761e-02 -8.83697629e-01 9.04174268e-01 5.90632081e-01
1.15110648e+00 -8.53494525e-01 5.43451071e-01 2.98550248e-01
-1.30536306e+00 2.85528749e-01 -6.09900475e-01 -7.72738010e-02
7.03786850e-01 6.77429736e-01 -6.71773970e-01 4.14824337e-01
7.61339068e-01 7.03216910e-01 -2.20739052e-01 7.18470395e-01
-1.81116655e-01 5.54542482e-01 -3.91033500e-01 2.07412213e-01
3.55350435e-01 -9.68712196e-02 2.42279083e-01 1.13777733e+00
7.43135631e-01 2.02586770e-01 -3.72201651e-02 8.38829279e-01
-6.53507710e-02 -3.63405108e-01 -3.87965053e-01 -3.61464685e-03
6.34070277e-01 9.83949661e-01 -5.66324532e-01 -7.07528472e-01
-3.88994396e-01 1.15863979e+00 6.45103976e-02 3.42371017e-01
-1.17702830e+00 -1.73612595e-01 7.83384800e-01 1.17680229e-01
4.27029043e-01 -4.91880894e-01 -1.50937110e-01 -1.10137784e+00
7.97838792e-02 -6.84308350e-01 -1.17793813e-01 -1.03491497e+00
-5.02805829e-01 7.50720918e-01 -3.42754573e-02 -1.03404784e+00
-5.98039031e-01 -9.13429186e-02 -1.43418744e-01 7.06196129e-01
-1.56227291e+00 -1.01059222e+00 -8.71986747e-01 8.31597388e-01
7.76637554e-01 -2.99911220e-02 5.80736160e-01 6.36235416e-01
-4.20829117e-01 4.41160440e-01 -3.33860405e-02 -7.80633315e-02
5.22625268e-01 -6.81635320e-01 9.68043029e-01 1.00939894e+00
-1.20550640e-01 2.39456311e-01 5.21396577e-01 -6.74867630e-01
-1.64876533e+00 -1.32772434e+00 8.68990779e-01 2.27656037e-01
3.50271434e-01 -2.07154483e-01 -8.05671692e-01 4.94392276e-01
2.82987475e-01 2.39687100e-01 1.88628942e-01 -8.65261614e-01
-1.64372727e-01 -1.86191276e-01 -8.12699854e-01 7.90481448e-01
1.41842651e+00 -5.43902934e-01 9.48929340e-02 2.71421105e-01
1.01173055e+00 -6.60891771e-01 -8.06462348e-01 1.47944376e-01
4.81926411e-01 -1.26187468e+00 1.05605245e+00 -1.46079734e-01
9.37825620e-01 -5.19718230e-01 -1.94666550e-01 -8.77524972e-01
-3.01114470e-01 -7.43938804e-01 -5.21371841e-01 9.82250035e-01
-1.68101177e-01 -1.57117501e-01 8.64122570e-01 3.73270482e-01
-1.83519006e-01 -7.29094684e-01 -6.77250683e-01 -5.96994817e-01
-6.35900617e-01 -7.40473986e-01 7.65656769e-01 4.20838624e-01
-5.57634056e-01 8.53389278e-02 -7.39667654e-01 -3.16533819e-02
6.57471478e-01 2.38246307e-01 1.00650573e+00 -5.84222078e-01
-4.32213098e-01 -3.01010311e-01 -4.64096963e-01 -1.53058243e+00
-1.48982018e-01 -6.16092205e-01 4.91945520e-02 -1.44510221e+00
-2.36925092e-02 -3.83679599e-01 1.90298244e-01 2.83075958e-01
-3.52804847e-02 4.85696167e-01 5.22583902e-01 3.44277710e-01
-5.07233441e-01 5.74361622e-01 1.37554753e+00 1.16637245e-01
1.05898855e-02 -4.60794419e-01 -1.10366546e-01 6.09614253e-01
6.13237023e-01 -1.97874486e-01 -8.23343575e-01 -9.37419713e-01
3.37883085e-01 5.51527977e-01 6.46050811e-01 -1.19538128e+00
4.58486468e-01 -3.04975808e-01 2.75804937e-01 -8.80052984e-01
6.85190916e-01 -5.27166128e-01 7.55350947e-01 4.45515454e-01
-1.10884637e-01 2.19169021e-01 -2.91576423e-02 4.38500226e-01
-4.24192995e-01 8.51466432e-02 5.74680030e-01 -1.47396222e-01
-9.80885684e-01 5.19686162e-01 -2.23799750e-01 -4.41340171e-03
8.32789242e-01 -1.43178910e-01 -5.02980828e-01 -3.96309018e-01
-4.96738940e-01 -7.10646138e-02 5.31593621e-01 4.18709248e-01
8.25784326e-01 -1.47524107e+00 -6.64942205e-01 2.71521181e-01
-3.59762818e-01 4.58078951e-01 6.77275836e-01 7.54137456e-01
-1.08685637e+00 4.02358234e-01 -2.76171297e-01 -7.56673038e-01
-1.09673309e+00 5.71008623e-01 -6.96454644e-02 -2.62767542e-02
-9.16212976e-01 7.60598242e-01 2.91268349e-01 4.86361295e-01
1.34768009e-01 -2.44603887e-01 3.33718389e-01 -1.03104413e-01
8.24927866e-01 5.95247805e-01 -4.76487875e-02 -5.08074403e-01
-6.15588464e-02 3.20672423e-01 1.87041223e-01 -1.59279555e-01
1.26109183e+00 -2.12114468e-01 1.95383299e-02 2.89471820e-02
1.25021911e+00 -4.57280099e-01 -1.74274075e+00 -2.57113129e-01
-5.28813004e-01 -6.24890447e-01 2.41487920e-01 -2.27513731e-01
-1.49429286e+00 8.17559421e-01 2.82289594e-01 -2.12642357e-01
1.26505578e+00 -3.72070462e-01 1.39559853e+00 1.69939429e-01
5.81890404e-01 -1.16036940e+00 -8.63498915e-03 3.98843348e-01
6.65604830e-01 -7.46420622e-01 1.53461676e-02 -5.10372758e-01
-4.36098278e-01 1.05631447e+00 5.31948388e-01 -7.76516274e-02
4.11845773e-01 5.47405839e-01 -2.58952707e-01 1.87003151e-01
-1.02294433e+00 1.47482799e-03 -2.16735322e-02 4.26468730e-01
2.05696635e-02 -2.30401903e-02 -2.47327387e-01 1.53082535e-01
-3.87523144e-01 5.62947631e-01 4.15113986e-01 6.86077535e-01
-2.12329388e-01 -7.43069768e-01 -1.14373043e-01 1.62019506e-01
-2.48499557e-01 -2.06748977e-01 2.83827752e-01 5.33518374e-01
1.81323647e-01 7.71961570e-01 4.39296216e-01 -4.67412025e-01
2.54681651e-02 -2.72941917e-01 3.89796048e-01 4.54057269e-02
-2.22477630e-01 2.63527036e-01 -1.19828373e-01 -1.02465379e+00
-4.42644268e-01 -3.62941384e-01 -1.30530536e+00 -8.66347909e-01
2.89233834e-01 -1.37711450e-01 9.13900137e-01 7.41180599e-01
8.18464577e-01 7.22780108e-01 2.85390437e-01 -1.06813729e+00
-7.59350434e-02 -6.39225602e-01 -8.48016292e-02 1.49071097e-01
6.61589131e-02 -2.19048887e-01 -1.32250460e-02 4.05055344e-01]
|
[10.768903732299805, -0.9170828461647034]
|
772f851a-9ef5-4931-84e2-0e073b465f07
|
190600050
|
1906.00050
| null |
https://arxiv.org/abs/1906.00050v1
|
https://arxiv.org/pdf/1906.00050v1.pdf
|
DISCO: Depth Inference from Stereo using Context
|
Recent deep learning based approaches have outperformed classical stereo matching methods. However, current deep learning based end-to-end stereo matching methods adopt a generic encoder-decoder style network with skip connections. To limit computational requirement, many networks perform excessive down sampling, which results in significant loss of useful low-level information. Additionally, many network designs do not exploit the rich multi-scale contextual information. In this work, we address these aforementioned problems by carefully designing the network architecture to preserve required spatial information throughout the network, while at the same time achieve large effective receptive field to extract multiscale contextual information. For the first time, we create a synthetic disparity dataset reflecting real life images captured using a smartphone; this enables us to obtain state-of-the-art results on common real life images. The proposed model DISCO is pre-trained on the synthetic Scene Flow dataset and evaluated on popular benchmarks and our in-house dataset of challenging real life images. The proposed model outperforms existing state-of-the-art methods in terms of quality as well as quantitative metrics.
|
['Kaushik Raghavan', 'Kunal Swami', 'Rituparna Sarkar', 'Pankaj Bajpai', 'Nikhilanj Pelluri']
|
2019-05-31
| null | null | null | null |
['stereo-matching']
|
['computer-vision']
|
[ 3.33727211e-01 -3.28330547e-01 -8.60619992e-02 -4.40505624e-01
-4.89675820e-01 -8.13881606e-02 5.38904250e-01 -1.67732254e-01
-5.69646001e-01 6.91767633e-01 4.67372209e-01 -4.88079600e-02
1.77718937e-01 -9.49555457e-01 -6.71947658e-01 -3.73869866e-01
1.80956602e-01 -6.52593225e-02 4.98819381e-01 -3.46810132e-01
4.72821057e-01 3.48927081e-01 -1.74757254e+00 3.56675982e-01
8.85793567e-01 1.09162068e+00 3.59067380e-01 2.75099427e-01
7.41035715e-02 1.09084332e+00 -1.73666805e-01 -4.19054538e-01
6.06478930e-01 -2.72080570e-01 -5.70348442e-01 -2.32873216e-01
1.14907587e+00 -7.55456686e-01 -9.44584191e-01 1.14455044e+00
8.02083194e-01 1.96679756e-01 2.46793211e-01 -8.51974249e-01
-4.25798178e-01 9.96515676e-02 -6.54295862e-01 3.48209828e-01
2.62032986e-01 4.07419324e-01 9.67258036e-01 -9.33808088e-01
7.10942388e-01 1.22541726e+00 7.08535552e-01 4.70737845e-01
-1.14954317e+00 -9.16290641e-01 -8.57817009e-02 3.80251437e-01
-1.26781952e+00 -7.82823563e-01 1.17688882e+00 -4.06622380e-01
9.35967684e-01 -2.22093835e-01 7.22036242e-01 1.05235207e+00
3.58552009e-01 6.92211092e-01 1.05205333e+00 -8.91205817e-02
6.98340535e-02 -4.72477555e-01 -2.40548074e-01 8.01857054e-01
8.70176256e-02 5.28252721e-01 -8.39847803e-01 2.84572363e-01
1.00486422e+00 2.49290094e-01 -4.40471202e-01 -5.77761173e-01
-1.25972497e+00 5.95388710e-01 7.43401527e-01 7.84997195e-02
-2.80229151e-01 2.93128073e-01 6.03832424e-01 2.17284903e-01
2.73649007e-01 -1.12159383e-02 -1.03343002e-01 -1.72766685e-01
-1.08492029e+00 2.48558775e-01 3.36871475e-01 8.37087631e-01
1.03411913e+00 3.27802867e-01 -8.36239234e-02 9.94584262e-01
1.85454011e-01 2.01343253e-01 5.05795419e-01 -1.15776944e+00
8.70527685e-01 4.10509497e-01 -2.42335852e-02 -1.23892128e+00
-2.46604294e-01 -6.46345139e-01 -1.25541341e+00 4.73498493e-01
4.11819816e-01 7.71893337e-02 -8.17461014e-01 1.80269265e+00
-6.62502795e-02 4.51554984e-01 3.48849371e-02 1.22008991e+00
1.00643480e+00 5.76225817e-01 -1.85710922e-01 2.15100273e-01
8.95322800e-01 -1.28594148e+00 -5.60634792e-01 -5.48745215e-01
3.04593354e-01 -8.38000059e-01 1.08659530e+00 3.12099129e-01
-1.15230286e+00 -1.04596150e+00 -1.27511346e+00 -3.13518494e-01
-6.48976117e-02 -3.23780961e-02 7.68635571e-01 3.47850561e-01
-1.15325725e+00 7.09026873e-01 -5.80384731e-01 -1.62735254e-01
6.47243083e-01 3.40641975e-01 -4.38547969e-01 -4.16825473e-01
-1.20517349e+00 5.33618569e-01 2.03519404e-01 1.91322237e-01
-9.03232753e-01 -8.21972013e-01 -1.18101108e+00 1.11697372e-02
1.17673978e-01 -9.81958389e-01 1.08187640e+00 -9.47719753e-01
-1.51118934e+00 9.32463884e-01 -1.83824405e-01 -5.08952677e-01
6.92748129e-01 -2.76088566e-01 -2.36700624e-01 1.27257705e-01
1.87187225e-01 1.11413324e+00 6.57046318e-01 -1.07267046e+00
-7.94409752e-01 -2.65131235e-01 2.35721439e-01 2.35453516e-01
-3.03146929e-01 -4.98762012e-01 -3.71364385e-01 -8.81214023e-01
1.36937231e-01 -6.35660946e-01 -2.16358811e-01 4.07504767e-01
-1.03587203e-01 2.81875640e-01 7.65378654e-01 -4.43260223e-01
9.30224538e-01 -2.10068178e+00 -1.38157114e-01 -2.81108707e-01
3.09679061e-01 4.01632905e-01 -1.12147026e-01 2.16084704e-01
-5.82874333e-03 -4.88403082e-01 -2.15234160e-01 -5.35861850e-01
-2.45347351e-01 3.35650556e-02 -3.26676607e-01 6.20419502e-01
-4.89984453e-02 7.27121592e-01 -9.82840061e-01 -6.79152071e-01
7.95382559e-01 5.12016177e-01 -9.26702142e-01 2.73931056e-01
1.82584882e-01 6.04127169e-01 -2.07292333e-01 4.49804485e-01
9.58548844e-01 -6.28385022e-02 -2.08699226e-01 -4.58068490e-01
-9.70812589e-02 3.56020838e-01 -1.04219043e+00 2.44383907e+00
-8.11364889e-01 1.10019493e+00 -3.28332596e-02 -1.08366334e+00
9.48280811e-01 1.84026659e-02 5.19914627e-01 -1.15219569e+00
-3.73343080e-02 4.27475959e-01 -3.63502912e-02 -2.42612392e-01
5.21848023e-01 5.84087893e-02 1.96159557e-01 1.32110938e-02
6.92715496e-02 5.16863838e-02 8.34541395e-02 -7.32708797e-02
9.71964121e-01 2.08483383e-01 7.20624551e-02 -3.59372944e-01
8.72079253e-01 -1.84246048e-01 9.18499470e-01 5.46940386e-01
-4.18597192e-01 1.03287876e+00 2.89969053e-02 -8.09975266e-01
-1.15488493e+00 -1.06431317e+00 -1.49320453e-01 6.82918370e-01
6.38390124e-01 -2.16880158e-01 -5.27377486e-01 -2.89489448e-01
-2.41411194e-01 2.25177646e-01 -5.10413587e-01 -3.23527828e-02
-8.55488420e-01 -1.77987710e-01 5.56860983e-01 5.10800302e-01
1.36906815e+00 -1.13430893e+00 -9.15166080e-01 4.79264021e-01
-3.66229236e-01 -1.45876038e+00 -6.35363638e-01 -3.42524469e-01
-9.27235782e-01 -1.14617562e+00 -8.85097682e-01 -1.02857804e+00
4.24805731e-01 5.24749160e-01 1.33115566e+00 -4.09936048e-02
-4.03110832e-01 -3.36423665e-01 -1.25061283e-02 1.80360273e-01
-4.86806557e-02 1.11456096e-01 -3.75795901e-01 1.04573429e-01
2.37355247e-01 -1.01633334e+00 -1.28223479e+00 3.45089823e-01
-1.02058971e+00 5.27345359e-01 4.82726097e-01 1.14111400e+00
4.21649158e-01 -1.69945657e-01 3.80190045e-01 -4.50993925e-01
2.69284993e-01 5.66854961e-02 -8.47116292e-01 -1.22098684e-01
-4.62230206e-01 1.39444292e-01 9.07834053e-01 -4.36463431e-02
-1.22343552e+00 1.59767177e-02 -4.03376460e-01 -4.83385533e-01
-1.99623369e-02 1.79587305e-01 3.07843182e-02 -3.13104600e-01
5.56943834e-01 3.79359901e-01 -2.18522489e-01 -3.77895772e-01
9.36297467e-04 5.50589323e-01 8.46165419e-01 -3.60883206e-01
5.56636333e-01 8.96770477e-01 3.10426503e-01 -5.65557539e-01
-7.33554006e-01 -4.57467169e-01 -5.84749639e-01 -2.77587861e-01
6.31617010e-01 -1.26109028e+00 -6.71000361e-01 7.01880455e-01
-1.17167377e+00 -2.66924083e-01 1.45046990e-02 6.40797436e-01
-7.54505396e-01 3.81245255e-01 -6.79616690e-01 -2.36589223e-01
-4.67280477e-01 -1.37407506e+00 1.13682878e+00 2.60380805e-01
1.09677762e-01 -9.12861228e-01 1.78735346e-01 4.38130796e-01
6.54834390e-01 3.25181931e-01 5.86178482e-01 2.46992409e-01
-7.23118603e-01 1.20989256e-01 -5.71821928e-01 2.90229231e-01
1.69921026e-01 -3.92509699e-01 -1.05898213e+00 -4.13009435e-01
-1.83058649e-01 -3.80650431e-01 1.25015426e+00 4.48133469e-01
1.27399445e+00 1.20839961e-01 2.20860690e-02 1.15704882e+00
1.76794255e+00 3.83305401e-02 8.76826823e-01 4.90170866e-01
9.42151189e-01 5.58930993e-01 4.40380156e-01 3.08393836e-01
4.48459715e-01 7.82931566e-01 5.66684306e-01 -3.08329135e-01
-5.40300012e-01 -5.17757773e-01 1.38863280e-01 5.49871147e-01
4.25640121e-02 -9.40147671e-04 -9.75229800e-01 7.54893303e-01
-1.92788601e+00 -1.10346270e+00 1.34863466e-01 2.14751887e+00
6.77964151e-01 3.31580311e-01 -2.28623614e-01 1.03243023e-01
6.64513111e-01 7.58373857e-01 -5.39925277e-01 -1.75295696e-01
-2.51231074e-01 2.39755854e-01 6.82166934e-01 5.63280225e-01
-1.09877551e+00 1.02102959e+00 5.48773623e+00 9.09164131e-01
-1.47197807e+00 2.58995760e-02 7.26737320e-01 -1.18564993e-01
-2.03247964e-01 -8.78010225e-03 -4.88417715e-01 3.55816334e-01
3.72606993e-01 -5.56017868e-02 3.40972275e-01 7.75312245e-01
4.29681957e-01 -1.36099651e-01 -1.12161624e+00 1.47635722e+00
-2.73712743e-02 -1.73414719e+00 -1.55186672e-02 6.28526062e-02
9.93258297e-01 3.78031254e-01 -2.46658791e-02 7.29435831e-02
-6.90798536e-02 -9.82970119e-01 6.32486463e-01 4.23785746e-01
9.39571321e-01 -7.54580200e-01 7.90067196e-01 1.00640036e-01
-1.35475135e+00 2.56996267e-02 -4.79849160e-01 -2.71870285e-01
2.62992978e-01 6.62225068e-01 -5.74096404e-02 4.76485789e-01
6.96099222e-01 1.28149176e+00 -5.00647843e-01 1.27133763e+00
7.78304413e-02 2.50159919e-01 -1.21495903e-01 4.50278908e-01
5.62952220e-01 -1.52292579e-01 3.76104891e-01 1.21235979e+00
2.47192934e-01 -2.68588901e-01 -4.16235253e-02 7.67634928e-01
-2.73906440e-01 -4.70729917e-02 -9.18547034e-01 5.93833685e-01
2.81439215e-01 8.54130685e-01 -4.21012312e-01 -3.82850409e-01
-6.26772106e-01 9.92463589e-01 2.87691504e-01 3.49828094e-01
-6.16730630e-01 -4.89333302e-01 9.51085150e-01 1.43280670e-01
2.20463052e-01 -1.90128222e-01 -3.35314542e-01 -1.39736879e+00
1.83260739e-01 -9.07810926e-01 6.93591386e-02 -6.59467757e-01
-1.04852569e+00 5.72804570e-01 -2.29762450e-01 -1.66043150e+00
-4.19552863e-01 -3.80005866e-01 -6.45427644e-01 8.53596687e-01
-1.90339947e+00 -8.78141046e-01 -7.92429805e-01 8.21792603e-01
8.68216693e-01 -2.96800613e-01 4.10333604e-01 8.06120396e-01
-4.10931230e-01 5.43404818e-01 6.32148087e-02 3.32747698e-01
8.55142415e-01 -7.89778829e-01 7.34327674e-01 1.01551282e+00
-3.44018219e-03 3.43628645e-01 5.16849697e-01 -2.53983080e-01
-1.17539442e+00 -1.15323138e+00 6.49057925e-01 2.34487042e-01
2.24321887e-01 -3.37499619e-01 -7.13710785e-01 2.12146655e-01
3.77276242e-01 4.81674403e-01 7.17919096e-02 -3.13008785e-01
-3.92432749e-01 -5.59559762e-01 -1.11939204e+00 8.26044619e-01
1.52647555e+00 -6.49166822e-01 -3.92122388e-01 -8.60087201e-02
4.39694226e-01 -5.33296824e-01 -5.59713602e-01 7.35373855e-01
7.70947099e-01 -1.78342056e+00 1.07856858e+00 -9.13638100e-02
9.35498595e-01 -3.90135080e-01 -3.14632326e-01 -1.15964782e+00
-1.41459391e-01 -7.21782625e-01 3.86207364e-02 9.52110827e-01
5.56402421e-03 -5.71454048e-01 1.08993816e+00 1.76296219e-01
-2.58457959e-01 -8.78094077e-01 -9.91449833e-01 -6.21828794e-01
-8.94666016e-02 -3.72698128e-01 5.99780977e-01 8.33544433e-01
-3.39255124e-01 1.94930166e-01 -5.93210042e-01 -1.70831323e-01
1.02987838e+00 3.15969348e-01 8.29848528e-01 -8.27684999e-01
-1.76229432e-01 -5.14574289e-01 -8.88881266e-01 -1.49018705e+00
2.72003531e-01 -4.52757925e-01 -2.84277722e-02 -1.34647453e+00
8.37114006e-02 -3.14710319e-01 -1.49176657e-01 -4.75251675e-03
4.82016578e-02 6.32997096e-01 1.42076150e-01 2.36885995e-01
-3.64501953e-01 8.06359172e-01 1.43958855e+00 -2.23059118e-01
-5.46928532e-02 -3.19845974e-01 -3.03797036e-01 8.33396733e-01
8.74276519e-01 -2.12712139e-01 -6.93123043e-01 -9.33748484e-01
2.27793101e-02 1.41407296e-01 6.06619477e-01 -1.43638396e+00
3.83652031e-01 -7.66594857e-02 4.15515214e-01 -7.74838030e-01
4.25798297e-01 -7.65847147e-01 -5.65436482e-02 5.52618384e-01
-3.48923832e-01 1.72378823e-01 1.77943349e-01 4.33098286e-01
-8.24959278e-01 1.65180296e-01 1.06426990e+00 -7.41836615e-03
-1.16820240e+00 5.07570922e-01 9.17580277e-02 3.03406268e-01
7.24668562e-01 -5.12604177e-01 -3.35301280e-01 -5.90145350e-01
-8.01707655e-02 7.22693801e-02 6.57619715e-01 5.77475011e-01
8.30350935e-01 -1.47424603e+00 -6.49612904e-01 5.15495956e-01
2.06129029e-01 1.80092245e-01 4.71648604e-01 5.22593796e-01
-1.06136060e+00 5.64022839e-01 -7.67593861e-01 -6.97344005e-01
-9.25581694e-01 3.09076786e-01 5.01589417e-01 -1.43483385e-01
-8.81615102e-01 6.00058258e-01 4.87205863e-01 -2.13275507e-01
3.49147677e-01 -2.46221215e-01 -8.10792670e-02 -3.49078298e-01
3.03613037e-01 2.45661467e-01 1.91043746e-02 -7.59074152e-01
-3.15237910e-01 9.58598316e-01 6.27293661e-02 -1.37739092e-01
1.21262443e+00 -1.95524022e-01 2.78042674e-01 -2.19450742e-02
1.61481106e+00 -2.57292658e-01 -1.73975086e+00 -4.77758259e-01
-4.86314505e-01 -1.05621517e+00 3.47148240e-01 -1.71345770e-01
-1.55785167e+00 1.28519201e+00 9.03743029e-01 -5.08707523e-01
1.32258594e+00 -6.21495843e-01 1.23379302e+00 4.14067268e-01
5.63503265e-01 -1.07946503e+00 8.00653473e-02 3.69157881e-01
7.04597354e-01 -1.59820902e+00 -4.97011654e-02 -3.96798909e-01
-3.50175679e-01 1.13082635e+00 8.42563033e-01 -4.25679624e-01
6.03881955e-01 1.41521364e-01 1.02719046e-01 -8.98401998e-03
-5.83237767e-01 -2.54515886e-01 1.76001742e-01 5.48711538e-01
5.06603360e-01 -3.84989977e-01 -1.97426200e-01 -2.26059601e-01
-2.23718196e-01 3.12383264e-01 2.86508650e-01 8.21352780e-01
-2.03469157e-01 -9.93576109e-01 -1.25565305e-01 2.17268899e-01
-5.27507484e-01 -2.43382543e-01 1.33962855e-01 6.78866804e-01
1.00746557e-01 8.30812454e-01 1.88816532e-01 -4.09248799e-01
4.65775788e-01 -6.10329568e-01 4.76314783e-01 -9.25287083e-02
-3.38193119e-01 -2.01569870e-01 4.36077751e-02 -9.69355106e-01
-6.52982414e-01 -3.44403923e-01 -8.70921791e-01 -5.61461329e-01
2.03496516e-01 -4.09239948e-01 3.27943355e-01 7.62962639e-01
3.20457667e-01 4.56352562e-01 7.21056163e-01 -1.18198025e+00
-2.14863867e-01 -7.66539156e-01 -2.41766185e-01 6.23315454e-01
5.86766183e-01 -7.07925081e-01 -1.69555143e-01 6.28469437e-02]
|
[8.856545448303223, -2.309432029724121]
|
8e5ad6da-8ff4-4577-8222-8311113d90d4
|
graph-learning-with-1d-convolutions-on-random
|
2102.08786
| null |
https://arxiv.org/abs/2102.08786v2
|
https://arxiv.org/pdf/2102.08786v2.pdf
|
Graph Learning with 1D Convolutions on Random Walks
|
We propose CRaWl (CNNs for Random Walks), a novel neural network architecture for graph learning. It is based on processing sequences of small subgraphs induced by random walks with standard 1D CNNs. Thus, CRaWl is fundamentally different from typical message passing graph neural network architectures. It is inspired by techniques counting small subgraphs, such as the graphlet kernel and motif counting, and combines them with random walk based techniques in a highly efficient and scalable neural architecture. We demonstrate empirically that CRaWl matches or outperforms state-of-the-art GNN architectures across a multitude of benchmark datasets for classification and regression on graphs.
|
['Martin Grohe', 'Hinrikus Wolf', 'Martin Ritzert', 'Jan Toenshoff']
|
2021-02-17
| null | null | null | null |
['graph-regression']
|
['graphs']
|
[ 2.41915341e-02 1.95757732e-01 -3.27194124e-01 -1.75221592e-01
-1.28952498e-02 -5.66236496e-01 8.40547621e-01 4.26886767e-01
-3.68002623e-01 4.63628232e-01 6.70662522e-02 -9.56394374e-01
-1.20550610e-01 -1.46280169e+00 -1.08119774e+00 -3.38195741e-01
-9.38525438e-01 7.01043069e-01 5.83627999e-01 -1.16848223e-01
-3.14163305e-02 4.31936920e-01 -8.30977738e-01 2.71826506e-01
7.30208084e-02 7.42983222e-01 -2.81797558e-01 1.14085627e+00
-3.61164838e-01 1.36787283e+00 -2.97981620e-01 -3.95548373e-01
1.99787706e-01 -3.05594325e-01 -8.98097217e-01 -3.65055859e-01
8.80812109e-01 -3.12094927e-01 -1.20754457e+00 1.00162148e+00
2.97656626e-01 1.02348357e-01 4.52422440e-01 -1.14747941e+00
-1.04869640e+00 1.18337119e+00 -3.66874367e-01 6.96837485e-01
5.18393576e-01 4.24645901e-01 1.13874733e+00 -5.91369271e-01
8.43507648e-01 1.49478793e+00 1.38136661e+00 1.60661012e-01
-1.46892858e+00 -3.42821628e-01 1.17782608e-01 -3.92647721e-02
-1.02227545e+00 -3.90355894e-03 4.95271176e-01 -1.78495735e-01
1.66284788e+00 1.33167068e-02 1.08341193e+00 1.27630293e+00
4.93103385e-01 5.43595254e-01 4.56092417e-01 -1.53636277e-01
-1.10585606e-02 -1.03409469e+00 6.23648942e-01 1.63462245e+00
7.35944510e-01 2.07585990e-01 -2.58764237e-01 -1.40014708e-01
8.79063427e-01 1.35077432e-01 1.83022022e-02 -3.32265735e-01
-1.14449275e+00 1.10256481e+00 1.10320950e+00 1.23625264e-01
2.12181397e-02 1.28098035e+00 6.28057063e-01 7.68349767e-01
6.01387084e-01 3.39266449e-01 -2.65916586e-01 2.37061575e-01
-5.81133187e-01 2.38125339e-01 1.29915905e+00 1.00574648e+00
1.13116670e+00 1.75257578e-01 -1.00304924e-01 2.61679888e-01
2.92097777e-01 2.55125344e-01 1.01251818e-01 -4.46488023e-01
3.14425856e-01 9.50068295e-01 -8.20392013e-01 -1.36630034e+00
-8.19853961e-01 -3.70139301e-01 -1.15910065e+00 -1.99346021e-01
3.41440231e-01 -8.75056069e-03 -1.33554149e+00 1.33668745e+00
7.79700000e-03 6.71414614e-01 -5.07337749e-01 5.25148392e-01
1.41683364e+00 4.30285037e-01 -1.03219980e-02 5.92939258e-01
7.88717926e-01 -1.16449642e+00 -4.23028842e-02 -4.38495070e-01
8.94916117e-01 -2.46876344e-01 9.19344425e-01 -7.05114380e-02
-8.85429859e-01 -3.58609527e-01 -1.18238699e+00 -2.76709288e-01
-8.98561656e-01 -5.55949867e-01 1.33505547e+00 6.27162755e-01
-1.69136274e+00 1.17580342e+00 -1.14849353e+00 -6.71843112e-01
6.30415082e-01 4.73872095e-01 -2.84635216e-01 -2.29979992e-01
-9.09491122e-01 4.97298181e-01 4.52095896e-01 3.84417586e-02
-1.13593674e+00 -4.29989785e-01 -1.34832978e+00 2.08409533e-01
2.45041266e-01 -8.51287901e-01 1.16002834e+00 -5.13514578e-01
-9.97101903e-01 9.57561374e-01 3.58016253e-01 -1.15051162e+00
1.06191516e-01 4.60648127e-02 -2.25115150e-01 2.31069922e-02
2.75342073e-02 3.80242348e-01 6.07032120e-01 -4.09735620e-01
-1.15524018e-02 1.19453475e-01 1.62164375e-01 -5.28528988e-01
-1.14788987e-01 -1.52896181e-01 -4.31380272e-01 -4.95624751e-01
-1.73196599e-01 -7.89706171e-01 -5.95566869e-01 -9.05620158e-02
-8.30906689e-01 -4.23614860e-01 7.62072802e-01 3.88210602e-02
9.43627596e-01 -1.43685615e+00 -3.11262071e-01 5.22443295e-01
1.31271005e+00 2.10491911e-01 -6.41244948e-01 7.09458947e-01
-2.12466806e-01 2.81238377e-01 -6.25419989e-03 -3.48515995e-02
-1.79868040e-03 2.13206097e-01 1.57210715e-02 7.18556821e-01
4.36382324e-01 1.68119359e+00 -1.30232942e+00 -3.82371545e-01
1.50083810e-01 2.82063693e-01 -6.21586382e-01 9.72792432e-02
-6.79622352e-01 -2.50624627e-01 -2.39547551e-01 5.27810752e-01
6.37812436e-01 -1.04302669e+00 4.26620781e-01 2.06429034e-01
2.67811805e-01 6.33245766e-01 -6.89843178e-01 1.57265830e+00
-7.81649202e-02 8.52221489e-01 -4.41668481e-01 -1.10241461e+00
7.84651577e-01 -1.63823098e-01 1.31567329e-01 -6.38734937e-01
3.62630993e-01 -1.02345571e-01 -5.78892529e-02 -1.09957621e-01
3.91774587e-02 4.88847554e-01 -5.68777360e-02 7.27897406e-01
7.68791497e-01 1.22859970e-01 6.84587181e-01 5.11207283e-01
2.26972079e+00 -1.83318034e-01 3.81704122e-01 -4.81455714e-01
-3.05769295e-02 -3.92325269e-03 -8.99660140e-02 1.35373378e+00
-5.01959622e-02 2.50188142e-01 1.02131236e+00 -1.21880209e+00
-7.72727311e-01 -1.42777061e+00 6.32425010e-01 1.31890094e+00
-1.18918262e-01 -9.05690372e-01 -5.57907104e-01 -8.74389470e-01
1.58335552e-01 -2.13103966e-04 -6.97252333e-01 -1.75157383e-01
-7.77951479e-01 -7.40662217e-01 7.83443570e-01 4.95200574e-01
4.39269572e-01 -1.35323358e+00 1.75693661e-01 5.93847215e-01
4.60132867e-01 -1.29061580e+00 -6.75305903e-01 4.95910585e-01
-8.99110913e-01 -1.86798680e+00 -1.66355237e-01 -1.35445547e+00
5.41092396e-01 4.46699172e-01 2.14617395e+00 7.71424949e-01
-4.27411228e-01 3.32972676e-01 -1.86546713e-01 -1.13253027e-01
-5.32593429e-01 8.29263985e-01 -4.05795276e-01 -4.15336221e-01
5.44781506e-01 -1.04701388e+00 -5.07716298e-01 -1.25195608e-01
-6.31958723e-01 -1.57141984e-01 5.44433653e-01 6.81467593e-01
3.57382447e-01 -5.15905395e-02 5.99286593e-02 -1.48822761e+00
1.12838185e+00 -6.48351967e-01 -7.45994747e-01 -4.34217639e-02
-4.33800459e-01 3.23286921e-01 9.31518734e-01 -5.53546786e-01
1.78285643e-01 -2.54398078e-01 -3.27225864e-01 -3.36823076e-01
3.15640755e-02 6.45581484e-01 5.73255837e-01 -8.71873498e-01
8.78738880e-01 1.28275990e-01 -2.56552752e-02 -5.19790426e-02
6.24788523e-01 -2.97722101e-01 4.45588410e-01 -2.58239567e-01
1.01491380e+00 4.42391276e-01 4.38203722e-01 -7.37068832e-01
-8.28726709e-01 -3.41144651e-01 -5.07052779e-01 -2.06035316e-01
7.49183834e-01 -6.14205778e-01 -1.13402808e+00 6.68351650e-01
-1.12915874e+00 -9.36109900e-01 -2.28407338e-01 1.59646884e-01
-4.47012782e-01 3.45697910e-01 -1.32394505e+00 -1.71464011e-01
-6.06487453e-01 -5.63838303e-01 8.44487190e-01 4.59189498e-04
-1.44777894e-01 -1.39248967e+00 4.80015278e-01 -7.99369037e-01
7.61831701e-01 7.33005881e-01 1.18923020e+00 -9.32473123e-01
-9.82944369e-01 -3.58556926e-01 -5.71840942e-01 -5.82777821e-02
-1.86352998e-01 -9.46566537e-02 -3.92313272e-01 -4.52157140e-01
-6.59725428e-01 -4.69800055e-01 1.40437341e+00 5.19136250e-01
1.14072037e+00 -4.86930400e-01 -5.68710804e-01 1.09992492e+00
1.65673709e+00 -5.28100193e-01 5.55508018e-01 1.05698511e-01
1.09939885e+00 -3.92803133e-01 -5.69294035e-01 -1.63948148e-01
4.88614619e-01 -2.97174722e-01 8.55335116e-01 -3.52828592e-01
-3.98543000e-01 -5.61616898e-01 1.75352663e-01 1.05547571e+00
-7.52475187e-02 -5.02603292e-01 -1.02942884e+00 3.59573781e-01
-1.86473989e+00 -9.26150799e-01 -5.72418272e-01 1.49477351e+00
1.28591314e-01 7.40550458e-01 3.45964789e-01 -2.55153775e-01
7.90216863e-01 8.25865448e-01 -4.94946361e-01 -5.61160922e-01
-1.32213801e-01 9.50788558e-01 1.02782845e+00 3.61423671e-01
-1.32062972e+00 1.12319374e+00 7.83113241e+00 5.26083946e-01
-7.01440990e-01 -1.89314857e-02 5.46536922e-01 2.77551562e-01
-1.51968643e-01 -3.20057534e-02 -4.99390960e-01 1.23860292e-01
1.25951552e+00 -5.44184297e-02 9.61089134e-01 9.06434596e-01
-4.04549807e-01 3.95909041e-01 -1.29722345e+00 6.66819513e-01
-1.52845215e-02 -2.16954470e+00 3.62150371e-01 1.41903926e-02
6.32170439e-01 1.02324426e+00 -2.34304026e-01 6.08841062e-01
1.44503522e+00 -1.42780995e+00 1.96628720e-01 3.26360375e-01
8.03043723e-01 -5.48186660e-01 6.74571931e-01 -1.48231499e-02
-1.82788813e+00 1.18841454e-01 -5.83383083e-01 -3.78680438e-01
-1.85743615e-01 6.14767969e-01 -9.62635219e-01 2.02009186e-01
5.84622622e-01 1.10599506e+00 -8.73507380e-01 1.01525164e+00
-4.88989085e-01 9.59599912e-01 -3.36141199e-01 -6.64964259e-01
8.15139472e-01 -1.16137847e-01 3.65454793e-01 1.59980810e+00
-4.82995100e-02 -4.33005631e-01 4.35249776e-01 1.05766606e+00
-7.76425600e-01 -1.81908488e-01 -1.39102411e+00 -5.07140338e-01
7.94743970e-02 1.06154895e+00 -1.15325058e+00 -3.05052161e-01
-7.07177937e-01 6.97037995e-01 1.06852961e+00 4.15817887e-01
-8.11928451e-01 -7.39646256e-01 4.64936197e-01 3.08102936e-01
6.25029862e-01 -7.56367862e-01 2.36267298e-01 -9.53298926e-01
-3.45861644e-01 -6.63392544e-01 8.01457226e-01 -3.95210236e-01
-1.63686383e+00 7.45452106e-01 -6.01034105e-01 -5.56104302e-01
4.26366776e-02 -1.11782205e+00 -1.24069786e+00 3.42910528e-01
-1.39739931e+00 -1.25234389e+00 -5.15800416e-01 8.66565168e-01
3.80301140e-02 -8.81526172e-02 9.18016791e-01 5.43649048e-02
-3.28647047e-01 5.20482481e-01 -1.60132915e-01 9.03540909e-01
5.12614958e-02 -1.51497233e+00 2.10269380e+00 8.53249848e-01
5.85691750e-01 5.80249608e-01 1.78799301e-01 -7.75998473e-01
-1.87925327e+00 -1.66050994e+00 6.41898096e-01 -2.28727415e-01
1.27312624e+00 -7.54524946e-01 -6.30410135e-01 1.23293042e+00
2.12437391e-01 4.52404767e-01 2.35734642e-01 4.07785505e-01
-5.82766891e-01 1.92685515e-01 -8.13335657e-01 7.63804853e-01
1.83399785e+00 -6.63963974e-01 -1.67816013e-01 6.55891359e-01
1.04244077e+00 -3.23646992e-01 -5.82703114e-01 6.78716525e-02
2.17864603e-01 -8.64571393e-01 1.05502701e+00 -1.06532896e+00
1.81745782e-01 1.59864813e-01 1.77138388e-01 -1.24212480e+00
-7.48152554e-01 -1.11044836e+00 -6.81011796e-01 3.40247452e-01
4.03946549e-01 -8.52459252e-01 1.35784543e+00 -3.16837102e-01
-6.22302964e-02 -8.28237295e-01 -7.78462946e-01 -9.12006676e-01
-6.64362535e-02 -4.17329013e-01 7.17820346e-01 7.44150221e-01
-3.19265425e-01 6.47379518e-01 8.96648839e-02 9.25967023e-02
8.81517053e-01 1.51588559e-01 1.03723311e+00 -1.29209602e+00
-2.65852660e-01 -6.34400249e-01 -9.01621401e-01 -1.31217825e+00
3.23053241e-01 -1.71481693e+00 -3.22746396e-01 -1.87624693e+00
7.43524283e-02 5.70417941e-02 -2.35495165e-01 5.43057382e-01
1.83359712e-01 5.47588706e-01 -2.74779737e-01 -3.44645798e-01
-1.15786338e+00 1.70014575e-01 1.04525995e+00 -5.26674628e-01
-2.02287268e-02 -3.06794699e-02 -5.19175231e-01 7.53486454e-01
6.36240482e-01 -7.97910035e-01 -1.77361190e-01 -4.98585492e-01
6.31482363e-01 -2.58092821e-01 5.87585986e-01 -1.25013542e+00
3.54403108e-01 2.78150350e-01 3.89666647e-01 -7.19065726e-01
-2.14270279e-01 -1.05048172e-01 -2.63465732e-01 8.68630707e-01
2.69931508e-03 5.01872838e-01 2.63086036e-02 1.02915001e+00
1.03996739e-01 3.95818770e-01 5.55732489e-01 -6.95128858e-01
-6.31120265e-01 9.16648686e-01 -6.22685969e-01 4.88327205e-01
5.70923686e-01 -1.33647025e-01 -7.78347373e-01 -4.85574633e-01
-6.93006158e-01 3.44039619e-01 1.58876255e-01 1.77656770e-01
7.04901278e-01 -1.42833209e+00 -6.27071917e-01 2.72431731e-01
2.58129071e-02 -1.42784743e-02 -3.46265286e-01 4.44379061e-01
-1.23444986e+00 3.77541184e-01 -2.48942319e-02 -5.55746853e-01
-7.99671471e-01 8.17005336e-01 5.61132371e-01 -9.28650916e-01
-9.46236372e-01 1.05077064e+00 -2.05771863e-01 -8.37786794e-01
2.74249822e-01 -8.29470992e-01 3.07322741e-01 -5.74506879e-01
2.06314921e-01 3.26869935e-01 1.03431880e-01 1.33849278e-01
-3.14833403e-01 2.46981367e-01 -9.29941013e-02 7.53097415e-01
1.53096521e+00 4.50316310e-01 -4.46930707e-01 2.93251395e-01
1.48906481e+00 -5.58778346e-01 -8.53569388e-01 -3.41249704e-01
3.56184125e-01 6.74979910e-02 -3.06676596e-01 -2.38671720e-01
-1.09574270e+00 5.59718847e-01 4.98551764e-02 7.78697133e-01
6.71377003e-01 2.76376158e-01 1.17328918e+00 1.10725987e+00
4.38049614e-01 -5.41256309e-01 3.11924160e-01 1.18425333e+00
4.35379922e-01 -1.08981729e+00 9.83790532e-02 -4.32986826e-01
5.14836848e-01 1.38591588e+00 5.40439546e-01 -1.09601390e+00
1.24230361e+00 5.16283631e-01 -3.92803758e-01 -1.05125010e+00
-8.42184901e-01 -5.54131448e-01 1.35215268e-01 7.38570094e-01
3.16082209e-01 9.80330855e-02 9.73089337e-02 3.65119949e-02
-2.65556037e-01 -1.66109070e-01 5.82737029e-01 7.86281824e-01
-3.89050990e-01 -6.44354224e-01 4.43382740e-01 1.11352539e+00
-3.21974307e-01 -6.48864627e-01 -5.64671457e-01 1.01562715e+00
-3.45270514e-01 6.21321440e-01 2.64606386e-01 -7.09860861e-01
9.83178914e-02 -2.14612797e-01 5.02110124e-01 -8.79118025e-01
-7.39655554e-01 -7.25508451e-01 3.98953646e-01 -1.04428351e+00
-1.57135814e-01 2.41794109e-01 -9.29047048e-01 -9.46690917e-01
-3.39495599e-01 -1.09126598e-01 2.79778183e-01 5.27984381e-01
2.77869999e-01 6.77504778e-01 3.63013119e-01 -9.51575458e-01
-1.86527610e-01 -8.83778870e-01 -6.98042333e-01 2.17029393e-01
4.66587275e-01 -2.30689615e-01 -5.48403740e-01 -5.68780959e-01]
|
[6.895805358886719, 6.264211654663086]
|
867c018a-300e-44a6-a326-af31a2a2a444
|
fine-grained-software-vulnerability-detection
| null | null |
https://openreview.net/forum?id=sKiAuHhc3w
|
https://openreview.net/pdf?id=sKiAuHhc3w
|
Fine-grained Software Vulnerability Detection via Information Theory and Contrastive Learning
|
Software vulnerabilities existing in a program or function of computer systems have been becoming a serious and crucial concern. In a program or function consisting of hundreds or thousands of source code statements, there are only few statements causing the corresponding vulnerabilities. Vulnerability labeling on a function or program level is usually done by experts with the assistance of machine learning tools; however, it will be much more costly and time-consuming to do that on a statement level. In this paper, to tackle this challenging problem, we propose a novel end-to-end deep learning-based approach to obtain the vulnerability-relevant code statements of a specific function. Inspired from previous approaches, we first leverage the mutual information theory for learning a set of latent variables that can represent the relevance of the source code statements to the corresponding function's vulnerability. We then propose a novel clustered spatial contrastive learning in order to further improve the representation learning and robust the selection process of vulnerability-relevant code statements. The experimental results on real-world datasets show the superiority of our proposed method over other state-of-the-art baselines.
|
['Dinh Phung', 'John C. Grundy', 'Trung Le', 'Van Nguyen']
|
2021-09-29
| null | null | null | null |
['vulnerability-detection']
|
['miscellaneous']
|
[-3.89578901e-02 -4.01286930e-01 -3.00500039e-02 -4.85515773e-01
-1.01842749e+00 -7.91369855e-01 7.01647922e-02 6.48966730e-01
8.98473933e-02 1.60274774e-01 9.91708040e-02 -7.45673060e-01
-1.14370540e-01 -9.58719492e-01 -6.00046515e-01 -5.06600142e-01
-1.65037826e-01 -2.30155423e-01 5.00378072e-01 1.75479834e-03
6.32887185e-01 4.15371656e-02 -1.30535913e+00 2.95108974e-01
1.06965935e+00 7.13959992e-01 3.67936224e-01 2.56844610e-01
-4.71098453e-01 9.89511549e-01 -5.84135294e-01 -3.23488057e-01
6.12231232e-02 -2.66015887e-01 -8.25653315e-01 -4.06233013e-01
1.33629084e-01 -4.93749343e-02 8.33123997e-02 1.69608462e+00
2.55036354e-01 1.11896805e-02 3.62857282e-01 -1.07051098e+00
-6.49519384e-01 7.41883576e-01 -9.61924434e-01 2.85504669e-01
2.49990836e-01 -7.41840377e-02 1.20373106e+00 -6.82500303e-01
1.88077241e-01 1.13424134e+00 4.87878323e-01 2.22501665e-01
-9.03127432e-01 -7.15719998e-01 4.21407819e-01 3.03987056e-01
-1.19143891e+00 -8.71977285e-02 1.20190263e+00 -7.88099289e-01
9.10423517e-01 1.73862219e-01 -5.63030876e-03 6.73774958e-01
3.75535637e-01 3.46634567e-01 6.48856938e-01 -3.49119157e-01
3.73875231e-01 9.64291021e-02 4.80076641e-01 7.91724861e-01
1.29980102e-01 -2.61233121e-01 2.80771498e-02 -6.51362181e-01
2.53474802e-01 4.06149358e-01 -2.39342958e-01 -3.62413943e-01
-9.45908248e-01 9.66196477e-01 7.14052677e-01 4.53002542e-01
-2.56861895e-01 1.39797762e-01 7.31441617e-01 2.03250885e-01
4.84328181e-01 2.53311992e-01 -4.61480409e-01 -1.45072654e-01
-9.58499908e-01 5.09282015e-02 5.32825112e-01 5.72178364e-01
1.04760993e+00 -2.04639927e-01 9.50518101e-02 4.75879967e-01
5.20178616e-01 -1.43292040e-01 3.05550694e-01 -2.58315206e-01
8.80760908e-01 1.30822647e+00 -1.16223790e-01 -1.39608383e+00
-6.39722273e-02 -3.77293825e-01 -5.48385739e-01 4.20357347e-01
-6.74580932e-02 -6.61680996e-02 -4.99974042e-01 1.45603740e+00
2.88528293e-01 1.87165484e-01 3.48843560e-02 6.03017747e-01
4.98310626e-01 7.83133030e-01 9.54139903e-02 4.25338047e-03
1.13897014e+00 -9.12708163e-01 -3.14621806e-01 -2.73502707e-01
7.38178194e-01 -7.66834319e-01 1.15935111e+00 9.28746089e-02
-4.83867913e-01 -3.51127744e-01 -9.35453653e-01 2.00962916e-01
-3.18549901e-01 1.06189452e-01 6.84087038e-01 6.75021172e-01
-7.36395657e-01 3.87012362e-01 -9.61535215e-01 3.91370431e-02
3.14288497e-01 5.03951423e-02 -3.14599603e-01 8.73655677e-02
-1.01536489e+00 4.46899414e-01 4.29267019e-01 7.12164417e-02
-1.09688485e+00 -5.02932727e-01 -9.18571293e-01 3.78710806e-01
5.55051029e-01 2.77516861e-02 8.33405852e-01 -9.88941789e-01
-7.91395366e-01 4.17184502e-01 -2.30632141e-01 -8.77204444e-03
5.13548404e-02 -3.73457670e-01 -4.01001334e-01 -1.71143621e-01
2.67831981e-01 -2.37372130e-01 6.56047940e-01 -1.07641280e+00
-4.96550232e-01 -3.71438265e-01 4.72075939e-01 -1.87254280e-01
-9.30788159e-01 5.44268787e-01 -4.04081643e-01 -4.78731990e-01
-6.20951727e-02 -5.02994716e-01 -2.84967214e-01 -2.61077225e-01
-6.26632631e-01 -3.17120999e-01 7.60851979e-01 -1.00563419e+00
1.71325934e+00 -2.34547520e+00 1.91877604e-01 2.19085813e-01
3.20932746e-01 2.51112223e-01 -9.13728625e-02 5.02298057e-01
-5.04717290e-01 2.46315315e-01 -6.55243039e-01 5.56331091e-02
-7.76718333e-02 -4.93151039e-01 -5.47833323e-01 4.36435044e-01
1.82252213e-01 3.97715569e-01 -9.53753114e-01 -3.32227528e-01
-7.42760226e-02 3.97018790e-01 -5.09197891e-01 3.55526716e-01
-2.13786691e-01 2.18423471e-01 -7.61993825e-01 7.75197208e-01
6.39827967e-01 -3.21819335e-01 2.28473358e-03 8.71741399e-02
-1.37190267e-01 3.43125492e-01 -9.46218967e-01 1.46647191e+00
-6.77252173e-01 4.69294846e-01 -2.07250416e-01 -1.08234203e+00
1.21016204e+00 1.86603487e-01 1.01276621e-01 -3.36716652e-01
-2.68425673e-01 1.12401865e-01 8.03924203e-02 -7.06540585e-01
9.65093225e-02 2.21244752e-01 -4.71039057e-01 4.91300285e-01
-3.79516214e-01 3.60131830e-01 -5.87727316e-02 2.01912969e-01
1.51124918e+00 -1.49975410e-02 3.78821552e-01 -3.41651142e-01
9.92548883e-01 -1.36141866e-01 9.72823679e-01 1.74057499e-01
-2.78814316e-01 2.67001390e-01 1.07896376e+00 -5.06793916e-01
-7.23248899e-01 -7.42186427e-01 2.00493172e-01 9.97210383e-01
7.89330378e-02 -5.28867304e-01 -9.32592392e-01 -1.19814742e+00
-2.57697672e-01 6.52188957e-01 -7.17525482e-01 -3.28538448e-01
-5.78781128e-01 -6.01236224e-01 3.21242303e-01 6.78952515e-01
4.62136745e-01 -9.78566349e-01 -6.44864738e-01 6.48555085e-02
-1.40434235e-01 -5.18076003e-01 -6.17030740e-01 3.92107405e-02
-5.41149914e-01 -1.12361681e+00 -4.88125175e-01 -8.43103290e-01
8.75493169e-01 1.44257382e-01 1.02551019e+00 3.59306306e-01
-2.11958334e-01 -2.85658747e-01 -5.47741294e-01 4.93891127e-02
-2.34807506e-01 1.35942698e-01 -4.55566466e-01 6.02067783e-02
4.27474409e-01 -5.51514506e-01 -4.90371019e-01 1.17353037e-01
-9.49799657e-01 -4.46417481e-01 5.52723229e-01 4.88396347e-01
3.81934106e-01 7.68426895e-01 4.46667582e-01 -9.06671584e-01
7.34045386e-01 -9.42626953e-01 -9.96708274e-01 4.56359923e-01
-3.15584093e-01 1.78804874e-01 8.63735199e-01 -1.99645072e-01
-1.06280577e+00 2.87661970e-01 -2.35861652e-02 -3.51794213e-01
-1.17402812e-02 1.02921748e+00 -5.53815484e-01 -3.73081975e-02
5.88439882e-01 2.35824794e-01 -6.07827783e-01 -6.26268685e-01
1.57541782e-01 5.43655872e-01 2.56702840e-01 -5.62123060e-01
1.01418185e+00 8.93565640e-02 -2.31376350e-01 -1.60859734e-01
-4.73480105e-01 -4.19601500e-01 -4.64287907e-01 -2.29075900e-03
8.12561452e-01 -6.03838086e-01 -3.08742553e-01 2.84998953e-01
-1.28359497e+00 6.25922605e-02 4.96075690e-01 -3.99800986e-02
1.23764440e-01 7.35400200e-01 -1.59983441e-01 -7.55134463e-01
-3.69973183e-01 -1.73390067e+00 9.14477348e-01 1.07886247e-01
-8.30179304e-02 -1.05021405e+00 3.76801372e-01 9.67430547e-02
3.56096148e-01 4.51554567e-01 1.33840299e+00 -5.91843128e-01
-6.54304683e-01 -3.32984269e-01 -3.68637919e-01 2.33923927e-01
4.28689063e-01 2.11471930e-01 -6.72547102e-01 -3.45745832e-01
1.96272388e-01 -2.04760842e-02 8.33441198e-01 -8.62085670e-02
1.34290636e+00 -3.90213907e-01 -5.64625263e-01 5.87899745e-01
1.65017796e+00 3.10804158e-01 4.62590963e-01 5.14646590e-01
8.69694650e-01 9.70989227e-01 7.75822580e-01 3.17764848e-01
2.87660688e-01 5.85599005e-01 7.43642867e-01 1.79938152e-01
4.47698802e-01 -8.60168934e-02 5.48714161e-01 6.59705579e-01
4.19226736e-01 4.03642170e-02 -1.46471131e+00 8.88583958e-01
-1.75541294e+00 -7.50717640e-01 -1.55382901e-01 2.21285200e+00
6.88930511e-01 8.67618471e-02 -9.39892083e-02 6.66161403e-02
9.85529780e-01 2.68749803e-01 -6.03915811e-01 -1.37185320e-01
4.55296218e-01 -2.38381073e-01 1.59804240e-01 2.88894296e-01
-1.27406549e+00 7.15469122e-01 5.21435928e+00 7.90075004e-01
-1.25672328e+00 1.26533806e-01 6.27393544e-01 1.84367910e-01
-7.11143851e-01 4.29145426e-01 -6.14526868e-01 7.33749807e-01
8.82291913e-01 -2.65344858e-01 2.26800755e-01 1.19225597e+00
-1.39909923e-01 3.65372486e-02 -1.01515472e+00 6.45437956e-01
5.75949885e-02 -9.79729772e-01 -3.19422513e-01 -1.44200280e-01
6.80794656e-01 -2.61664033e-01 2.04444155e-01 2.52759814e-01
2.65261263e-01 -8.55730534e-01 6.16026282e-01 3.64753574e-01
4.30097669e-01 -1.05314434e+00 8.14472795e-01 2.65224069e-01
-1.64072788e+00 -3.75736475e-01 -4.08786923e-01 1.66852146e-01
-2.70975947e-01 9.07533467e-01 -5.38708925e-01 5.20322144e-01
9.77167010e-01 5.08940697e-01 -9.20395315e-01 9.61679280e-01
-4.83578116e-01 6.09885633e-01 2.63797402e-01 -1.05090015e-01
2.11226583e-01 2.10594773e-01 3.16478431e-01 1.05758905e+00
4.17018771e-01 -3.21313262e-01 1.77717268e-01 1.27372789e+00
-1.47841247e-02 2.35966623e-01 -5.65524220e-01 -1.99367851e-01
4.93418753e-01 1.41146421e+00 -8.55903804e-01 -9.37564000e-02
-7.68753111e-01 7.69083261e-01 5.21481037e-01 2.69638628e-01
-9.09380376e-01 -9.58566666e-01 5.88842034e-01 -2.09276080e-01
3.99192125e-01 -8.18824768e-02 -2.53262430e-01 -1.12294447e+00
6.46153212e-01 -8.54821444e-01 2.95743704e-01 -1.93186030e-01
-1.22864330e+00 9.80912447e-01 -1.17827915e-01 -1.34745312e+00
-1.58162072e-01 -1.90268770e-01 -1.18718314e+00 1.21225810e+00
-1.48495042e+00 -9.83624578e-01 -2.94220775e-01 4.43077326e-01
3.29794824e-01 -4.95168120e-01 6.93029046e-01 3.28446805e-01
-8.73463690e-01 5.89828551e-01 1.82717592e-01 3.58117640e-01
5.17949820e-01 -1.23530722e+00 9.36007142e-01 1.40632081e+00
-1.36471838e-01 1.21590698e+00 4.15083110e-01 -9.58072484e-01
-1.10211849e+00 -1.21430588e+00 7.36117125e-01 -3.80692542e-01
8.50237012e-01 -3.93116027e-01 -1.27562642e+00 6.74143255e-01
1.05598100e-01 1.35376200e-01 7.01623917e-01 1.11717008e-01
-6.80552840e-01 -1.19760431e-01 -9.90471900e-01 3.45474660e-01
4.61502284e-01 -9.14017916e-01 -4.99126554e-01 2.60559589e-01
7.08849370e-01 1.93763673e-02 -5.24390101e-01 2.60867625e-01
5.24169058e-02 -1.10326684e+00 8.27273011e-01 -2.44865030e-01
6.58578813e-01 -5.79093814e-01 -1.48345083e-01 -1.12588274e+00
-2.90223956e-01 -3.30784291e-01 1.47513047e-01 1.61176002e+00
2.56464809e-01 -6.21488869e-01 6.40911579e-01 5.69629669e-01
-5.10818213e-02 -6.25528276e-01 -7.84767449e-01 -5.63005149e-01
1.53055891e-01 -2.35239536e-01 9.04890060e-01 1.09373260e+00
2.29192272e-01 -2.21141130e-02 6.21810881e-03 5.32970250e-01
6.46581411e-01 4.91451025e-01 4.80319440e-01 -1.17543566e+00
-5.34559071e-01 -5.56540787e-01 -5.66891789e-01 -4.34428602e-01
5.73954642e-01 -7.87168324e-01 2.12583154e-01 -1.30392408e+00
4.25285310e-01 -2.77467906e-01 -5.78979015e-01 5.76344311e-01
-6.76743627e-01 -4.23098624e-01 -1.57452002e-01 2.60045648e-01
-5.61148942e-01 2.63410956e-01 4.35361475e-01 -3.99011821e-01
-3.12780738e-02 6.18635416e-02 -8.34046662e-01 8.47869039e-01
7.06434548e-01 -6.96986377e-01 -5.00259697e-01 -5.53274155e-01
4.60962445e-01 2.24265859e-01 2.95345992e-01 -9.13982570e-01
3.28229904e-01 -1.54148743e-01 -1.27779394e-01 -4.47810143e-01
-2.95110345e-01 -8.06044161e-01 -1.87654030e-02 4.59501386e-01
-3.90091836e-01 1.94453895e-01 2.89344817e-01 5.30607402e-01
-4.47824895e-01 -6.96347713e-01 6.25939488e-01 -1.49050564e-01
-9.29918945e-01 1.84242904e-01 -9.94930565e-02 -2.20042855e-01
1.30241013e+00 3.74048173e-01 -3.47650051e-01 2.75666099e-02
-1.61136687e-01 4.69980715e-03 6.23471379e-01 5.96759140e-01
7.63167262e-01 -1.17068994e+00 -5.34708142e-01 9.49604809e-02
3.94215554e-01 -2.05875903e-01 3.45602155e-01 3.93180400e-01
-4.71384555e-01 3.26512307e-01 -1.01225004e-01 -1.48098841e-01
-1.36350906e+00 8.34047377e-01 2.36893848e-01 -3.82964343e-01
-4.12454009e-01 1.05842292e+00 5.26717186e-01 -5.38225889e-01
1.74326792e-01 -5.92801794e-02 -5.20406604e-01 -2.30152264e-01
7.38787711e-01 2.71314353e-01 8.01848471e-02 -5.84832251e-01
-7.53775835e-01 6.33660972e-01 -3.23382527e-01 1.44937426e-01
1.34936631e+00 2.65684754e-01 -6.58251226e-01 2.92274028e-01
1.39059699e+00 7.68273026e-02 -1.12083244e+00 -1.51751310e-01
3.50676656e-01 -7.34802663e-01 2.50770181e-01 -6.18619680e-01
-1.37581766e+00 1.28137052e+00 6.27762556e-01 2.90534854e-01
1.22954154e+00 -8.13994706e-02 7.64616191e-01 2.97176629e-01
4.86251980e-01 -3.76979321e-01 3.22733596e-02 2.51474291e-01
6.59819722e-01 -1.30136585e+00 -2.82302290e-01 -3.87062937e-01
-2.76242495e-01 1.21135736e+00 7.85066009e-01 -1.44451529e-01
6.04283273e-01 3.78187627e-01 -2.04715468e-02 -2.88939416e-01
-5.73181927e-01 2.52785087e-01 3.83760065e-01 5.06112337e-01
8.50521326e-01 -1.18333893e-03 -7.55055994e-02 6.03268683e-01
3.11524481e-01 -5.16702116e-01 3.29184204e-01 1.12408388e+00
-4.96201873e-01 -1.41191030e+00 -4.70006108e-01 3.04784030e-01
-6.74012363e-01 -3.51546824e-01 -3.58579099e-01 8.82548615e-02
-2.90992558e-02 9.45245326e-01 -3.77051920e-01 -5.83362341e-01
1.34513736e-01 -1.12516358e-01 -1.75155014e-01 -8.93997550e-01
-7.61005640e-01 -9.59574208e-02 -5.24211049e-01 -5.28723180e-01
3.60945426e-02 -7.92173743e-01 -1.33256221e+00 2.25640112e-03
-3.60888958e-01 2.67042816e-01 4.75430071e-01 8.04903746e-01
3.75413805e-01 8.43582273e-01 8.08805704e-01 -4.58583295e-01
-5.69710314e-01 -6.90094650e-01 -2.10668311e-01 3.59286666e-01
4.27194536e-01 -5.64737141e-01 -3.05693597e-01 6.22462705e-02]
|
[7.117496967315674, 7.763134956359863]
|
85f684ce-102e-4add-a1b1-c021a6717839
|
a-new-expert-questioning-approach-to-more
|
1904.00317
| null |
https://arxiv.org/abs/1904.00317v2
|
https://arxiv.org/pdf/1904.00317v2.pdf
|
A New Expert Questioning Approach to More Efficient Fault Localization in Ontologies
|
When ontologies reach a certain size and complexity, faults such as inconsistencies, unsatisfiable classes or wrong entailments are hardly avoidable. Locating the incorrect axioms that cause these faults is a hard and time-consuming task. Addressing this issue, several techniques for semi-automatic fault localization in ontologies have been proposed. Often, these approaches involve a human expert who provides answers to system-generated questions about the intended (correct) ontology in order to reduce the possible fault locations. To suggest as informative questions as possible, existing methods draw on various algorithmic optimizations as well as heuristics. However, these computations are often based on certain assumptions about the interacting user. In this work, we characterize and discuss different user types and show that existing approaches do not achieve optimal efficiency for all of them. As a remedy, we suggest a new type of expert question which aims at fitting the answering behavior of all analyzed experts. Moreover, we present an algorithm to optimize this new query type which is fully compatible with the (tried and tested) heuristics used in the field. Experiments on faulty real-world ontologies show the potential of the new querying method for minimizing the expert consultation time, independent of the expert type. Besides, the gained insights can inform the design of interactive debugging tools towards better meeting their users' needs.
|
['Patrick Rodler', 'Michael Eichholzer']
|
2019-03-31
| null | null | null | null |
['fault-localization']
|
['computer-code']
|
[ 8.32901802e-03 5.59200108e-01 1.12937354e-01 -4.60215300e-01
-4.60213929e-01 -5.82704842e-01 1.76928908e-01 5.91880083e-01
2.49912962e-02 6.36863291e-01 -4.62826878e-01 -3.65058601e-01
-7.88511515e-01 -9.27418172e-01 -4.53202605e-01 -8.06294233e-02
2.73990870e-01 7.97586620e-01 7.80288100e-01 -3.70539874e-01
4.33848858e-01 3.43140185e-01 -2.21227384e+00 3.66977125e-01
1.53469300e+00 9.49683070e-01 2.02725217e-01 2.43932471e-01
-3.60700727e-01 1.06253481e+00 -8.13155413e-01 -4.70748574e-01
-4.84415330e-03 -4.63361055e-01 -1.19357395e+00 2.56222010e-01
1.19802266e-01 -1.70998976e-01 4.11027282e-01 1.49945366e+00
1.22935839e-01 -1.84202373e-01 1.91699997e-01 -1.53908241e+00
-1.82416022e-01 7.50919163e-01 2.14115262e-01 -9.69155356e-02
7.90157378e-01 -1.53219536e-01 9.90267158e-01 -5.70105314e-01
4.61430162e-01 1.06734920e+00 5.21306276e-01 4.24968928e-01
-8.15799475e-01 -1.74907133e-01 2.95416359e-02 7.03258514e-01
-1.66415310e+00 -4.31954145e-01 7.14671910e-01 -4.18608576e-01
8.27920258e-01 8.05510402e-01 4.31654036e-01 6.66416228e-01
-2.10188124e-02 2.25938335e-01 8.80873561e-01 -8.06659102e-01
5.61850131e-01 7.99817502e-01 2.77378559e-01 8.34207058e-01
5.90102613e-01 -7.16220200e-01 -2.73729324e-01 -5.02017796e-01
2.44302273e-01 -2.38506496e-01 -4.90555048e-01 -4.12269503e-01
-4.79996026e-01 4.62335944e-01 -2.69352823e-01 7.29041457e-01
-3.30842584e-01 -3.19234431e-01 1.03371128e-01 4.74835545e-01
1.16756849e-01 8.69297028e-01 -6.53615952e-01 -6.02385812e-02
-6.17227972e-01 3.37142706e-01 1.32317257e+00 1.07717907e+00
8.52403045e-01 -3.52488250e-01 1.84523344e-01 6.50838971e-01
4.02811557e-01 -8.78055394e-03 3.21483284e-01 -9.07390594e-01
1.44149184e-01 1.42793739e+00 5.57416737e-01 -1.27554262e+00
-3.22726250e-01 -3.59241724e-01 -8.62498060e-02 2.26100221e-01
5.80888271e-01 1.87172443e-01 -1.76952794e-01 1.36255598e+00
5.39492786e-01 -1.80086330e-01 -8.44924226e-02 8.28403533e-01
2.29641512e-01 5.80499507e-02 -1.87532082e-01 -2.79718578e-01
1.66992116e+00 -6.52456045e-01 -9.00769770e-01 -1.67109028e-01
7.72167921e-01 -6.45013392e-01 1.00237584e+00 7.77348697e-01
-1.13845909e+00 -2.24406704e-01 -9.13233817e-01 2.03734636e-01
-4.37895924e-01 2.66435921e-01 3.21701676e-01 8.44988108e-01
-9.32612717e-01 6.40248477e-01 -5.02797365e-01 -6.58989131e-01
-1.49309650e-01 3.77189547e-01 -6.02023900e-02 -1.38840839e-01
-1.27736712e+00 1.14738262e+00 4.16078776e-01 2.48525873e-01
-4.30983990e-01 -2.52995640e-01 -5.34628272e-01 2.72754818e-01
1.14069641e+00 -6.00437999e-01 1.32371628e+00 -1.16421103e+00
-1.33289385e+00 5.97886086e-01 -9.99707952e-02 -3.04718971e-01
6.41732275e-01 -1.41913489e-01 -5.34651577e-01 2.62478918e-01
1.75259799e-01 -7.14688823e-02 4.50200766e-01 -1.28931987e+00
-7.69361138e-01 -3.87916446e-01 7.34467626e-01 -8.75653327e-02
-5.16027033e-01 6.81969374e-02 -3.50719750e-01 -3.39742862e-02
3.31569284e-01 -6.02807462e-01 -1.59066543e-01 3.47783379e-02
-4.04557079e-01 -5.43033183e-01 3.86578947e-01 -6.33319020e-01
1.55738354e+00 -1.84269106e+00 1.36432052e-02 5.27438343e-01
1.69961154e-01 2.09943205e-01 3.00652802e-01 7.39642143e-01
-1.15968427e-02 2.35172912e-01 -1.91899881e-01 3.74450274e-02
2.85679281e-01 3.95310849e-01 -1.19643427e-01 2.47725770e-01
2.02307418e-01 1.58266127e-01 -7.50896096e-01 -6.69747114e-01
1.47142708e-01 -2.78508179e-02 -7.55239189e-01 3.11890483e-01
-6.27840996e-01 2.42884517e-01 -6.68521404e-01 7.70039380e-01
5.80122828e-01 -2.61897802e-01 6.74783766e-01 -2.18788892e-01
-3.01182438e-02 3.64009142e-01 -1.59286165e+00 1.03951454e+00
-3.69392484e-01 -1.25116602e-01 8.65106359e-02 -1.19968522e+00
8.71150374e-01 4.93255138e-01 5.09065509e-01 -5.34332395e-01
4.26809192e-02 6.84277296e-01 -3.43992971e-02 -1.25442958e+00
2.77765244e-01 1.20581873e-01 1.75806388e-01 1.64781332e-01
-1.38697952e-01 1.65404320e-01 4.76136237e-01 -1.15858011e-01
1.26445806e+00 5.10244742e-02 6.62884355e-01 -4.67233211e-01
9.89144266e-01 2.52162039e-01 6.74504340e-01 8.05844724e-01
-1.95280761e-02 1.64838359e-01 9.77089107e-01 -4.96017069e-01
-6.78916335e-01 -2.89005011e-01 -2.01854855e-02 6.32019162e-01
2.71468312e-01 -8.63381147e-01 -1.03654301e+00 -7.86267698e-01
-2.87037402e-01 8.47725809e-01 -2.40324229e-01 -8.56843311e-03
-3.60854864e-01 -1.45270795e-01 4.99829322e-01 -7.54837394e-02
8.97374228e-02 -8.68747413e-01 -1.05824006e+00 4.14946139e-01
-4.94259268e-01 -1.22549272e+00 3.25530440e-01 5.85134327e-02
-7.88050830e-01 -1.78101480e+00 3.32213640e-02 -4.16214973e-01
1.00056589e+00 -3.76621890e-03 1.09615803e+00 9.81708229e-01
-6.83252662e-02 4.30740058e-01 -8.07714403e-01 -1.76592499e-01
-6.23354733e-01 -7.11857975e-02 -1.54323712e-01 5.98315820e-02
3.59992325e-01 -5.01288235e-01 -1.42350972e-01 5.87592125e-01
-1.14959168e+00 -2.69127816e-01 3.23643178e-01 4.73170608e-01
2.44195148e-01 7.78488815e-01 4.89609390e-01 -1.13695443e+00
7.49538839e-01 -5.52235067e-01 -8.74011159e-01 7.92927325e-01
-1.04486883e+00 3.59030247e-01 9.51567531e-01 -1.98540121e-01
-9.30762708e-01 -1.03139922e-01 -3.05926204e-01 -7.86195770e-02
-5.09825587e-01 6.56049848e-01 -4.57748652e-01 -1.04596712e-01
7.66102850e-01 -6.65168241e-02 -2.90324569e-01 -4.88929272e-01
-1.45955220e-01 6.71171069e-01 1.45055577e-01 -8.74088049e-01
6.32395208e-01 1.99430272e-01 -2.03299075e-01 -6.98992968e-01
-6.25802338e-01 -2.90478200e-01 -1.73149258e-01 -2.82488674e-01
4.50536579e-01 -2.80493140e-01 -9.38889027e-01 7.90131744e-03
-1.34469783e+00 3.78744155e-02 -2.15820447e-01 6.66512991e-04
-4.09459442e-01 6.90010726e-01 -1.26934588e-01 -1.29974270e+00
1.14354379e-01 -1.43418276e+00 8.71846795e-01 4.19637822e-02
-4.81688321e-01 -7.02649057e-01 -2.04704136e-01 3.88603210e-01
3.43874723e-01 -2.85932403e-02 1.16159976e+00 -9.11786914e-01
-5.93977630e-01 -2.71527886e-01 1.05668381e-01 1.66751891e-01
2.13736564e-01 1.15798883e-01 -6.99891627e-01 9.35458764e-02
2.65738398e-01 1.23478070e-01 -7.59912804e-02 -2.72763908e-01
1.10407293e+00 -6.88877642e-01 -2.70080805e-01 -1.04930240e-03
1.51379645e+00 2.26668850e-01 7.15111434e-01 5.44996858e-01
2.50196010e-01 1.01508558e+00 7.33452559e-01 6.30291343e-01
5.06096125e-01 8.09048474e-01 7.08281040e-01 4.09746915e-01
4.98287827e-01 3.67692895e-02 2.94094030e-02 3.51954013e-01
-1.52982190e-01 -4.63362724e-01 -1.03974009e+00 4.73017305e-01
-1.99294353e+00 -7.90727973e-01 -3.69191557e-01 2.20515323e+00
7.61802793e-01 2.49756336e-01 -6.82730675e-02 7.24727571e-01
6.26100838e-01 -4.90927964e-01 -9.81933102e-02 -1.71795949e-01
1.34281397e-01 9.26583558e-02 1.17700556e-02 5.63790321e-01
-6.40474916e-01 5.43474257e-01 5.00045061e+00 4.02454406e-01
-6.98289633e-01 7.37147108e-02 -8.31372589e-02 4.24815059e-01
-5.55859089e-01 5.15928030e-01 -6.03727341e-01 3.65469426e-01
8.39639723e-01 -6.12873696e-02 5.80014765e-01 1.04883051e+00
1.16307944e-01 -4.56777304e-01 -1.19761097e+00 6.55177176e-01
5.10857105e-02 -9.32061315e-01 -2.46661752e-01 -4.73197736e-02
1.51533082e-01 -7.50534773e-01 -6.83735788e-01 7.32828975e-02
-1.39627859e-01 -5.52424252e-01 9.82790291e-01 8.66402447e-01
1.51447564e-01 -7.70557582e-01 1.06522012e+00 7.77874053e-01
-9.06471670e-01 -4.58139658e-01 -2.23212063e-01 -2.45214909e-01
-8.49554613e-02 7.98991978e-01 -8.57975185e-01 8.73395324e-01
8.10674846e-01 1.19841537e-02 -8.99776042e-01 1.14071810e+00
-3.84593040e-01 2.00331926e-01 -4.66015846e-01 -1.12195231e-01
-3.35509479e-01 -9.51015390e-03 4.87508804e-01 8.14267457e-01
5.78338921e-01 1.38303146e-01 9.91144255e-02 9.16824341e-01
3.91656131e-01 2.79720962e-01 -4.55795318e-01 2.49676257e-02
6.31648123e-01 1.04973137e+00 -8.24659228e-01 -3.56928140e-01
-2.65768528e-01 8.36685777e-01 4.26080585e-01 2.14532763e-01
-7.52280295e-01 -4.03892756e-01 4.50824022e-01 5.06205559e-01
-6.39293119e-02 1.12487331e-01 -1.63984224e-01 -1.03140235e+00
6.24070525e-01 -1.36203623e+00 4.56052750e-01 -7.50641406e-01
-9.87381101e-01 6.97997212e-01 7.79422373e-02 -1.17207122e+00
-3.42943758e-01 -4.68281657e-01 -1.73537910e-01 5.01071632e-01
-1.38359058e+00 -5.15294075e-01 -3.92906427e-01 4.82218295e-01
2.51374602e-01 3.74194562e-01 9.55017149e-01 6.22667432e-01
-5.19841015e-01 2.86762476e-01 -7.06410348e-01 -4.75146979e-01
6.30015314e-01 -1.18152988e+00 -3.87261897e-01 1.00339890e+00
-9.60877985e-02 7.47551858e-01 1.18458331e+00 -5.90743959e-01
-1.42061579e+00 -6.23759210e-01 1.40829539e+00 -2.03754991e-01
5.90842545e-01 4.13889671e-03 -1.24676490e+00 5.74962735e-01
1.18981503e-01 -2.32676432e-01 3.14772695e-01 -7.96418637e-02
2.55710278e-02 -1.91839412e-01 -1.34778678e+00 5.46654165e-01
9.12725270e-01 -4.72801119e-01 -6.33921444e-01 6.43688977e-01
4.62948173e-01 -3.29659790e-01 -8.36419880e-01 4.26179320e-01
1.93533525e-01 -1.44475353e+00 4.45375890e-01 -4.13949966e-01
2.64056958e-02 -6.44430697e-01 1.27534289e-02 -7.87767172e-01
3.00363395e-02 -5.28514981e-01 2.84935031e-02 1.25621426e+00
3.98384690e-01 -8.86362135e-01 3.69128078e-01 1.04658234e+00
-3.30510020e-01 -7.37950981e-01 -7.50559270e-01 -5.31906307e-01
-8.36103082e-01 -5.71432889e-01 7.32211709e-01 1.13116908e+00
3.61820996e-01 -1.58239320e-01 -1.05659261e-01 8.72954071e-01
3.19000065e-01 -8.02663714e-02 6.07055366e-01 -1.71955788e+00
-3.67077380e-01 -2.92485625e-01 -3.50857705e-01 -5.45557916e-01
7.26326508e-03 -4.42058057e-01 1.43564001e-01 -1.63296664e+00
-2.86015689e-01 -5.86053967e-01 5.39798513e-02 6.53000593e-01
-7.84335136e-02 -4.21597689e-01 -2.49134630e-01 9.65819135e-02
-7.16174066e-01 -5.93644232e-02 7.31922150e-01 1.70622945e-01
-2.89960764e-02 8.76580253e-02 -6.60968840e-01 9.34091032e-01
6.75222337e-01 -7.64879704e-01 -4.77029741e-01 -2.97264248e-01
9.76943135e-01 2.69110709e-01 4.23166782e-01 -1.17368305e+00
6.72234118e-01 -3.66527200e-01 -4.66036171e-01 -2.23192036e-01
-1.47378892e-02 -1.49683702e+00 5.11512697e-01 4.07408655e-01
-9.75644737e-02 1.66847110e-01 -2.43082240e-01 2.41540119e-01
-2.92103857e-01 -1.09301829e+00 5.43249846e-01 -2.12758854e-01
-5.11703253e-01 -3.49720389e-01 -5.19527018e-01 -2.28561088e-01
1.19483316e+00 -1.23520136e-01 -1.77721664e-01 -2.04955727e-01
-7.15797126e-01 3.25320035e-01 5.89919448e-01 8.57421160e-02
4.75190789e-01 -6.91376686e-01 -1.75295308e-01 4.91914712e-02
3.17333102e-01 -1.43665612e-01 8.55611339e-02 9.73296046e-01
-5.99565387e-01 3.50766152e-01 -1.52062075e-02 -3.87885094e-01
-1.13055754e+00 5.46878994e-01 6.15482807e-01 -2.40192413e-01
-3.14050585e-01 3.66445541e-01 -5.10260522e-01 -4.75359470e-01
2.97933310e-01 -5.42501986e-01 -4.19114977e-01 -5.57101965e-02
3.84703934e-01 4.23173219e-01 4.83612120e-01 -1.80527583e-01
-5.89842618e-01 3.82146567e-01 3.74585658e-01 1.74197480e-01
1.17530119e+00 -2.62215316e-01 -4.93993640e-01 2.69990116e-01
2.30028167e-01 3.49515408e-01 -5.24657190e-01 -7.13075250e-02
6.43686354e-01 -2.90799379e-01 -4.78201956e-01 -6.55974209e-01
-6.96444988e-01 5.39287388e-01 2.61692196e-01 1.15437901e+00
1.27718449e+00 3.40153091e-02 3.12964290e-01 6.60378873e-01
7.28879750e-01 -1.15774632e+00 -2.41804287e-01 1.73534513e-01
7.88548887e-01 -9.85006332e-01 -1.12927452e-01 -8.44372392e-01
-3.00350636e-01 1.44974399e+00 8.18815708e-01 2.02215046e-01
2.81899512e-01 2.97478974e-01 -7.22356960e-02 -3.85322988e-01
-8.02546561e-01 -3.24391305e-01 5.66780344e-02 2.68520147e-01
3.22802395e-01 -2.16815263e-01 -8.26076210e-01 8.38447452e-01
-9.41395164e-02 2.15828776e-01 6.82179272e-01 1.18630779e+00
-7.37235665e-01 -1.42174768e+00 -6.46862984e-01 1.75768808e-01
-3.92818719e-01 2.45074004e-01 -4.65789616e-01 8.03164363e-01
5.65034091e-01 1.40831172e+00 -4.66736794e-01 -3.27094138e-01
9.10376728e-01 3.50128323e-01 4.25366610e-01 -7.00107515e-01
-5.88649154e-01 -1.37388334e-01 5.28808475e-01 -7.20446050e-01
-3.71670842e-01 -3.86334866e-01 -1.19962776e+00 2.53959298e-02
-6.27093434e-01 4.94890213e-01 4.90599513e-01 1.40073323e+00
2.76752859e-01 6.09972835e-01 3.52448881e-01 -9.22885090e-02
-7.79080391e-01 -6.76937938e-01 -3.75520706e-01 4.00435209e-01
-5.53055294e-02 -7.60223687e-01 -4.06516492e-01 -8.64088256e-03]
|
[5.496842861175537, 2.83546781539917]
|
6f1b7762-a586-494b-9ed5-642d3ad22a93
|
pedestrian-trajectory-forecasting-using-deep
|
2305.16620
| null |
https://arxiv.org/abs/2305.16620v1
|
https://arxiv.org/pdf/2305.16620v1.pdf
|
Pedestrian Trajectory Forecasting Using Deep Ensembles Under Sensing Uncertainty
|
One of the fundamental challenges in the prediction of dynamic agents is robustness. Usually, most predictions are deterministic estimates of future states which are over-confident and prone to error. Recently, few works have addressed capturing uncertainty during forecasting of future states. However, these probabilistic estimation methods fail to account for the upstream noise in perception data during tracking. Sensors always have noise and state estimation becomes even more difficult under adverse weather conditions and occlusion. Traditionally, Bayes filters have been used to fuse information from noisy sensors to update states with associated belief. But, they fail to address non-linearities and long-term predictions. Therefore, we propose an end-to-end estimator that can take noisy sensor measurements and make robust future state predictions with uncertainty bounds while simultaneously taking into consideration the upstream perceptual uncertainty. For the current research, we consider an encoder-decoder based deep ensemble network for capturing both perception and predictive uncertainty simultaneously. We compared the current model to other approximate Bayesian inference methods. Overall, deep ensembles provided more robust predictions and the consideration of upstream uncertainty further increased the estimation accuracy for the model.
|
['Prasenjit Ghorai', 'Zachary Doerzaph', 'Azim Eskandarian', 'Anshul Nayak']
|
2023-05-26
| null | null | null | null |
['trajectory-forecasting', 'bayesian-inference']
|
['computer-vision', 'methodology']
|
[-9.20189247e-02 -6.78239986e-02 -5.20522930e-02 -6.94289029e-01
-6.42478585e-01 -3.57151806e-01 6.77952766e-01 1.52281135e-01
-3.00538301e-01 1.06606507e+00 3.73626083e-01 1.31020367e-01
-2.86386311e-01 -8.05546463e-01 -7.92790949e-01 -5.87919116e-01
-1.72609240e-01 2.49952823e-01 3.32340986e-01 1.48098215e-01
-2.21712925e-02 7.21627623e-02 -1.74900234e+00 -1.48567826e-01
9.54129577e-01 1.39495504e+00 4.56116915e-01 8.12823355e-01
2.50318348e-01 7.86651969e-01 -6.84676409e-01 -3.04819614e-01
1.84067897e-02 6.28750073e-03 9.61159691e-02 -2.34141842e-01
1.90388635e-01 -7.02488303e-01 -4.47625488e-01 1.26438117e+00
5.31344473e-01 1.73586100e-01 5.41058660e-01 -1.33938015e+00
-4.41502005e-01 9.49713886e-01 -8.20330679e-02 1.35615826e-01
2.05373242e-01 2.96227545e-01 5.75804830e-01 -3.03850502e-01
1.38856787e-02 1.45777559e+00 7.89216340e-01 6.10590637e-01
-9.85016584e-01 -9.32351053e-01 8.13163519e-01 3.44734192e-01
-1.19931984e+00 -6.43473327e-01 5.72681963e-01 -3.68506461e-01
7.80834019e-01 2.65882164e-02 6.16849780e-01 1.53099215e+00
6.13784671e-01 5.84466100e-01 1.13111293e+00 3.11111331e-01
5.63345313e-01 1.27479494e-01 -1.02208508e-03 1.53277129e-01
4.07146692e-01 8.57409120e-01 -5.48882306e-01 -3.52815241e-02
3.78335267e-01 1.66888505e-01 -1.41916394e-01 1.61658958e-01
-1.18015170e+00 4.37605649e-01 4.57002133e-01 -2.58590013e-01
-5.88374794e-01 5.83112061e-01 9.47428867e-02 6.67730719e-03
4.12114471e-01 7.92144984e-02 -4.96668696e-01 -4.56950277e-01
-8.41276228e-01 2.99008816e-01 7.99836576e-01 7.85436094e-01
5.39529443e-01 3.07923704e-01 -2.59406269e-01 2.15707511e-01
9.06509459e-01 1.19560933e+00 1.38189405e-01 -1.20137846e+00
3.36112559e-01 9.74227488e-02 5.99028468e-01 -1.16488957e+00
-6.06875777e-01 -8.15233350e-01 -1.11535501e+00 2.03305766e-01
2.01721996e-01 -6.08856559e-01 -1.01891065e+00 2.23228073e+00
2.18096480e-01 6.55737936e-01 4.10720348e-01 9.89116430e-01
3.93353045e-01 8.19964826e-01 2.36853123e-01 -4.05550689e-01
1.17993164e+00 -5.42030096e-01 -1.31889522e+00 -6.14032388e-01
-1.93033338e-01 -5.94225705e-01 1.49264470e-01 5.18332303e-01
-7.06985533e-01 -6.99019074e-01 -1.19851696e+00 5.08936763e-01
-1.61312968e-01 7.76751339e-02 4.78723615e-01 6.30358398e-01
-8.20543885e-01 5.87235451e-01 -1.40619004e+00 -9.88848135e-02
5.30184433e-02 2.09212273e-01 1.92221910e-01 6.97430819e-02
-1.48032546e+00 1.19385433e+00 5.91650665e-01 4.85384762e-01
-1.33157122e+00 -3.24465841e-01 -8.79281998e-01 6.19670376e-02
5.70746720e-01 -8.20020437e-01 1.42395878e+00 -5.61840951e-01
-2.04592705e+00 -5.62018573e-01 -2.76713252e-01 -7.71084905e-01
3.47606003e-01 -4.23046499e-01 -7.02735245e-01 -4.99508440e-01
-2.80359298e-01 5.90496778e-01 7.21108139e-01 -1.20608282e+00
-7.44565547e-01 -4.42608505e-01 1.87714115e-01 2.61411697e-01
1.39202744e-01 -5.68887353e-01 -1.04030155e-01 -4.60728168e-01
2.71366894e-01 -1.03127158e+00 -4.80886579e-01 -6.73778877e-02
-4.69025463e-01 3.31746861e-02 6.56402290e-01 -5.94163120e-01
1.22540653e+00 -1.85062575e+00 -1.77912302e-02 6.03281520e-02
4.92729731e-02 2.93796603e-02 1.30065009e-01 3.59591216e-01
5.28729975e-01 -2.10961059e-01 -8.46357495e-02 -7.52802789e-01
4.05541837e-01 4.12762433e-01 -6.82341695e-01 4.19498056e-01
-8.69143084e-02 5.79614878e-01 -8.86681020e-01 -2.19478428e-01
4.35061753e-01 6.65665507e-01 -3.96390676e-01 3.45207423e-01
-7.28749990e-01 6.24163151e-01 -5.55352271e-01 3.89264315e-01
7.72218704e-01 1.93735007e-02 2.74649318e-02 -3.52959603e-01
-2.23856509e-01 2.78758466e-01 -1.48619628e+00 1.69418585e+00
-3.56189281e-01 4.40795273e-01 1.80829942e-01 -5.59129298e-01
8.71028781e-01 2.92101115e-01 2.93259859e-01 -2.75315225e-01
2.31801093e-01 3.17505710e-02 -5.85549809e-02 -3.21407557e-01
7.09746420e-01 -6.65677041e-02 -2.54407555e-01 1.41092733e-01
-2.32008517e-01 -1.55955985e-01 -1.69424251e-01 -4.66538686e-03
7.66178191e-01 3.63359690e-01 1.48492068e-01 2.24771127e-01
1.18993796e-01 -3.58031154e-01 1.15625989e+00 9.09953713e-01
-4.67755049e-01 2.28090405e-01 2.83062132e-03 -2.36402571e-01
-5.93260050e-01 -1.26808715e+00 -9.61544961e-02 7.12972581e-01
5.37236810e-01 -4.76582736e-01 -3.48480582e-01 -2.21574754e-01
1.32105544e-01 1.01948690e+00 -3.22584033e-01 -3.98812681e-01
6.92617968e-02 -8.27765286e-01 4.44664955e-01 5.71896911e-01
5.45243859e-01 -3.47104728e-01 -7.20067978e-01 5.43895602e-01
-1.92809924e-01 -1.14276922e+00 -9.45731774e-02 5.46889491e-02
-7.12514222e-01 -6.14709079e-01 -3.46066535e-01 1.89216375e-01
2.59170204e-01 -1.97655350e-01 7.62852967e-01 -5.64144850e-01
4.45352465e-01 4.13838863e-01 -1.93366081e-01 -8.77285302e-01
-3.71305168e-01 -3.69149297e-01 7.42486000e-01 -1.36082590e-01
1.66224450e-01 -5.90121925e-01 -5.34231782e-01 4.12097484e-01
-6.88448727e-01 3.81829264e-03 5.43016493e-01 5.84807813e-01
4.86142933e-01 2.54275858e-01 6.60702050e-01 -7.77841955e-02
5.75212300e-01 -5.55364549e-01 -9.86426234e-01 2.74094373e-01
-6.59867048e-01 4.21880454e-01 4.51552868e-01 -5.54845870e-01
-1.42843878e+00 7.64720365e-02 -1.72759667e-01 -3.66419286e-01
-1.62386760e-01 5.96254408e-01 -5.57979867e-02 4.03081954e-01
3.34809572e-01 1.80105537e-01 2.69443262e-02 -3.54974985e-01
2.50538707e-01 7.05171943e-01 4.51796442e-01 -6.04718387e-01
4.11573917e-01 2.83195823e-01 9.45576504e-02 -2.95642316e-01
-1.05995464e+00 1.90285712e-01 4.38455604e-02 -4.36457038e-01
6.73067331e-01 -1.31003082e+00 -1.03423142e+00 6.96777463e-01
-1.30089748e+00 -3.35474834e-02 -1.03538044e-01 1.01614571e+00
-5.48719883e-01 3.70120049e-01 -3.76498938e-01 -1.64548481e+00
-2.19841808e-01 -1.38578081e+00 8.14628541e-01 4.67626035e-01
5.34502529e-02 -7.84713745e-01 3.04903090e-01 -3.73432934e-02
8.02664220e-01 2.22739965e-01 7.96041787e-02 -2.95196950e-01
-7.91772544e-01 -3.47193927e-01 9.31610465e-02 1.45631358e-01
-6.84454963e-02 7.66485482e-02 -1.08178818e+00 -2.91380465e-01
3.60012264e-03 -1.54694170e-01 1.06551182e+00 7.04071581e-01
8.08992326e-01 -3.88701230e-01 -4.92873311e-01 2.78293163e-01
1.15899682e+00 2.25814193e-01 2.96773046e-01 -6.36492744e-02
2.60903090e-01 3.37900072e-01 6.44661903e-01 8.77382934e-01
9.72274780e-01 6.61531389e-01 9.22491133e-01 7.92154551e-01
1.86188906e-01 -3.87065500e-01 8.72012079e-01 7.95822978e-01
2.29691584e-02 -8.10084164e-01 -5.68037748e-01 3.23257089e-01
-2.11034322e+00 -1.08143449e+00 2.68705130e-01 2.31168365e+00
7.48902142e-01 2.95118064e-01 -3.10086429e-01 -2.24235117e-01
6.86325073e-01 1.70980573e-01 -9.76432145e-01 8.26087743e-02
1.31002560e-01 -6.98241889e-01 6.27793610e-01 5.96681952e-01
-1.12046802e+00 5.88663638e-01 6.36609936e+00 6.35595262e-01
-9.67472553e-01 6.82079718e-02 4.17143703e-01 -2.66989827e-01
-2.56197393e-01 5.42024523e-02 -1.18906736e+00 8.68761063e-01
1.33590448e+00 2.71034271e-01 3.45448196e-01 7.29628682e-01
3.16272408e-01 -5.35326123e-01 -9.80116665e-01 7.32562304e-01
-2.23641142e-01 -9.83163416e-01 -3.82307321e-01 -7.27967769e-02
9.28942144e-01 3.25524747e-01 2.00832292e-01 5.01286685e-01
9.38074529e-01 -8.55429530e-01 9.92026150e-01 1.48398781e+00
2.57530570e-01 -6.39347970e-01 9.71779227e-01 7.82377601e-01
-1.04596102e+00 -3.80325377e-01 -5.53316176e-01 -4.90714103e-01
7.00581431e-01 1.13669121e+00 -5.06783307e-01 3.51581961e-01
6.87727690e-01 8.38196456e-01 -1.10002987e-01 1.06823647e+00
-3.19562465e-01 4.66778815e-01 -8.91529441e-01 -3.08557868e-01
7.61866793e-02 2.89718304e-02 8.66459846e-01 6.87708557e-01
8.21033716e-01 4.50699404e-02 5.09425938e-01 9.12024379e-01
3.71095359e-01 -8.54827881e-01 -3.64599168e-01 6.75805733e-02
8.68549466e-01 7.30736315e-01 -1.66074619e-01 -4.16602552e-01
-5.12775071e-02 3.71200830e-01 1.74793661e-01 4.06818271e-01
-1.06246829e+00 1.00828834e-01 1.13632369e+00 -4.67013836e-01
2.79718786e-01 -5.17055631e-01 -1.54620886e-01 -1.28666782e+00
-9.89583731e-02 -4.40262109e-01 1.75275207e-01 -6.63355291e-01
-1.44420254e+00 5.97339034e-01 1.60792768e-01 -1.27819204e+00
-5.57922244e-01 -3.71910453e-01 -4.20290053e-01 7.60806561e-01
-1.50678861e+00 -1.01298511e+00 -1.20083608e-01 1.98029593e-01
3.59009355e-01 1.00208089e-01 7.24723041e-01 7.05584437e-02
-8.45429599e-01 1.95683002e-01 3.29518735e-01 -4.07118917e-01
7.58183122e-01 -1.03252780e+00 7.21466169e-02 1.10022807e+00
-2.51807809e-01 5.36981463e-01 1.23692870e+00 -9.52584982e-01
-1.34034061e+00 -1.15933514e+00 4.81206328e-01 -5.19033253e-01
5.38864851e-01 -5.33024035e-02 -6.06338739e-01 6.64579690e-01
1.34340286e-01 -1.03459559e-01 3.23881298e-01 2.45830920e-02
-2.45357499e-01 -3.82216781e-01 -1.08587301e+00 3.76398832e-01
6.61089301e-01 -2.98328459e-01 -4.96735543e-01 -6.04697131e-02
8.11217606e-01 -5.45459092e-01 -9.69449639e-01 5.83890021e-01
8.45891058e-01 -9.44809020e-01 8.24803412e-01 -1.56858325e-01
5.99427382e-03 -7.27491796e-01 -5.33609390e-01 -1.69161117e+00
-2.55358428e-01 -4.39575404e-01 -6.37961507e-01 1.24903357e+00
4.82275903e-01 -7.81873167e-01 3.50873679e-01 9.49093580e-01
-2.00021565e-01 -4.17459786e-01 -1.15232503e+00 -6.87083662e-01
-3.98797214e-01 -1.09006345e+00 8.43313813e-01 1.89055532e-01
-7.33022615e-02 -8.23729783e-02 -9.27383721e-01 8.79287601e-01
9.38523054e-01 1.08085647e-01 6.07649207e-01 -1.15083694e+00
-3.29663724e-01 -2.61283219e-01 -4.49128330e-01 -1.44363344e+00
-1.62340887e-02 -1.08806238e-01 6.90543294e-01 -1.53989446e+00
-8.88274461e-02 -3.06276172e-01 -3.78811002e-01 -6.35605364e-04
-3.07976991e-01 -1.89994678e-01 1.68542117e-01 -1.60017852e-02
-8.84802878e-01 1.08720946e+00 9.95801747e-01 -1.40735209e-01
4.56756614e-02 5.48287749e-01 -4.56588060e-01 6.61495507e-01
8.92084301e-01 -4.70399737e-01 -5.61545789e-01 -5.89174747e-01
5.00415325e-01 2.82113314e-01 3.63098264e-01 -1.38162172e+00
6.60636127e-01 -3.55089664e-01 5.18716156e-01 -9.21381414e-01
8.74668360e-01 -1.13432062e+00 6.01220369e-01 3.94622177e-01
-3.29475254e-01 -4.98818932e-03 2.33181387e-01 1.38741696e+00
-9.66478065e-02 2.77581736e-02 4.34006780e-01 2.81872973e-02
-8.44205260e-01 2.96131372e-01 -7.37838864e-01 -4.44563001e-01
9.86251950e-01 6.97366744e-02 -2.94361353e-01 -8.82857621e-01
-7.46479213e-01 6.01145029e-01 1.54679433e-01 2.83464789e-01
6.08218014e-01 -1.19038808e+00 -5.81959426e-01 3.38070355e-02
-1.15364738e-01 1.12902611e-01 5.51477611e-01 7.89353609e-01
2.09041938e-01 3.97086442e-01 9.37368125e-02 -7.14801490e-01
-5.16738594e-01 4.98519540e-01 5.08793950e-01 -1.51403859e-01
2.30413284e-02 8.48215640e-01 -2.03425273e-01 -4.14292008e-01
5.75873375e-01 -8.96445990e-01 -2.91369051e-01 -5.18568084e-02
6.61170602e-01 4.73016441e-01 -5.11155486e-01 -5.34615219e-01
-3.35178196e-01 5.07126987e-01 3.37093532e-01 -2.62342811e-01
1.02070415e+00 -8.11668456e-01 2.48283297e-01 7.38017917e-01
4.37016606e-01 -3.45531583e-01 -2.08973885e+00 -4.91710067e-01
-2.20461905e-01 -2.34043509e-01 4.70878869e-01 -1.07805860e+00
-1.00206614e+00 8.67134750e-01 8.69493961e-01 1.60306290e-01
9.78111625e-01 -3.97055954e-01 7.28321016e-01 5.20469844e-01
7.06898570e-01 -1.09080374e+00 -4.14562553e-01 8.04684222e-01
6.11823082e-01 -1.54991984e+00 1.58069849e-01 -3.63334604e-02
-6.63748622e-01 1.01912630e+00 7.38979101e-01 1.07822523e-01
9.49872732e-01 5.04041731e-01 -2.14505494e-02 2.79718667e-01
-1.27488863e+00 -2.68961042e-01 3.83960605e-01 7.01629877e-01
-9.55376700e-02 3.62802476e-01 2.94050515e-01 9.26718354e-01
-1.66623205e-01 5.21942973e-02 2.77646571e-01 6.53698564e-01
-6.64242923e-01 -5.94410777e-01 -5.86903334e-01 4.09201950e-01
-2.58670807e-01 7.95734972e-02 3.67756516e-01 1.15025695e-02
2.17782050e-01 1.51497519e+00 2.13715687e-01 -7.16729820e-01
2.08245724e-01 -2.16701925e-01 1.86563328e-01 -6.49885759e-02
-4.65194657e-02 -4.08148840e-02 1.95198432e-01 -6.47004783e-01
-5.92086792e-01 -5.94747722e-01 -9.21021879e-01 -4.17095631e-01
-5.13493121e-01 8.72008502e-02 1.05617619e+00 1.03514087e+00
6.75559461e-01 8.15425992e-01 3.28528225e-01 -1.12774694e+00
-1.14329982e+00 -1.31030345e+00 -3.43525410e-01 -2.19803378e-01
6.14746511e-01 -1.06276095e+00 -3.70796144e-01 -1.82503968e-01]
|
[6.859510898590088, 3.447544813156128]
|
c110b62a-7954-4bf8-97a4-41975e3a4aed
|
representing-and-reasoning-with-qualitative
|
1401.3899
| null |
http://arxiv.org/abs/1401.3899v1
|
http://arxiv.org/pdf/1401.3899v1.pdf
|
Representing and Reasoning with Qualitative Preferences for Compositional Systems
|
Many applications, e.g., Web service composition, complex system design, team
formation, etc., rely on methods for identifying collections of objects or
entities satisfying some functional requirement. Among the collections that
satisfy the functional requirement, it is often necessary to identify one or
more collections that are optimal with respect to user preferences over a set
of attributes that describe the non-functional properties of the collection.
We develop a formalism that lets users express the relative importance among
attributes and qualitative preferences over the valuations of each attribute.
We define a dominance relation that allows us to compare collections of objects
in terms of preferences over attributes of the objects that make up the
collection. We establish some key properties of the dominance relation. In
particular, we show that the dominance relation is a strict partial order when
the intra-attribute preference relations are strict partial orders and the
relative importance preference relation is an interval order.
We provide algorithms that use this dominance relation to identify the set of
most preferred collections. We show that under certain conditions, the
algorithms are guaranteed to return only (sound), all (complete), or at least
one (weakly complete) of the most preferred collections. We present results of
simulation experiments comparing the proposed algorithms with respect to (a)
the quality of solutions (number of most preferred solutions) produced by the
algorithms, and (b) their performance and efficiency. We also explore some
interesting conjectures suggested by the results of our experiments that relate
the properties of the user preferences, the dominance relation, and the
algorithms.
|
['Vasant Honavar', 'Samik Basu', 'Ganesh Ram Santhanam']
|
2014-01-16
| null | null | null | null |
['service-composition']
|
['miscellaneous']
|
[ 4.73752571e-03 -1.78771988e-01 -2.33974651e-01 -4.13507760e-01
-2.00868964e-01 -9.16479647e-01 2.87441283e-01 5.75668395e-01
-3.30874711e-01 6.93391383e-01 2.85640866e-01 -5.52201457e-02
-9.37959254e-01 -9.85024989e-01 -4.19070870e-01 -7.62388170e-01
-3.63794327e-01 8.49046350e-01 3.27509463e-01 -3.70963186e-01
3.71666402e-01 5.01099408e-01 -2.14825869e+00 2.48385370e-01
9.87491369e-01 1.17721343e+00 2.42514893e-01 3.35068882e-01
-1.93894073e-01 8.07784796e-02 -2.84026355e-01 -6.34216741e-02
3.39132696e-01 -2.80203551e-01 -1.05775583e+00 3.97480458e-01
-2.70920455e-01 1.91245630e-01 7.58986056e-01 1.27400851e+00
7.52519667e-02 1.92485005e-01 7.76616275e-01 -1.69760942e+00
-4.34471250e-01 6.01637542e-01 -2.82954395e-01 -1.45339310e-01
6.48590088e-01 -3.47396195e-01 1.26663578e+00 -6.19294465e-01
5.51716447e-01 8.99241209e-01 7.34992847e-02 4.38128889e-01
-1.33888543e+00 -3.02946299e-01 2.87968725e-01 2.07116473e-02
-1.56791699e+00 -3.41672182e-01 4.33335483e-01 -4.41787511e-01
4.37156439e-01 9.14150178e-01 5.02850592e-01 1.71460405e-01
-4.25493643e-02 3.90169233e-01 1.00984907e+00 -5.95969081e-01
6.02773368e-01 5.89978456e-01 5.12354732e-01 3.93914849e-01
6.92137182e-01 -3.23993683e-01 -3.32370758e-01 -6.08586431e-01
2.51945883e-01 -1.75580546e-01 -3.85961115e-01 -8.32077503e-01
-9.40206647e-01 7.03013659e-01 -1.86933335e-02 5.03563821e-01
-5.54014444e-01 -3.31808269e-01 4.45561782e-02 3.43741834e-01
5.13978638e-02 5.86072385e-01 -4.35097516e-01 3.13013345e-01
-2.09382102e-01 3.54357928e-01 1.22085047e+00 1.17458999e+00
8.96417558e-01 -5.18806279e-01 1.03769265e-01 6.14976645e-01
2.18712479e-01 1.99326560e-01 1.09040648e-01 -9.78229225e-01
1.25785604e-01 8.68344188e-01 6.64199591e-01 -8.58810782e-01
-3.25335383e-01 -1.80465966e-01 -5.23084581e-01 -3.84400189e-02
3.40181649e-01 1.97310627e-01 -1.30057618e-01 2.12126350e+00
3.30960244e-01 -3.84833008e-01 2.24216253e-01 8.17879379e-01
3.35894197e-01 5.21524727e-01 -4.39763069e-02 -9.70182359e-01
1.14517665e+00 -7.91190490e-02 -5.69593966e-01 4.46469486e-01
4.85702515e-01 -6.09810650e-01 1.22259486e+00 2.37115741e-01
-1.19519901e+00 6.81131892e-03 -7.90440559e-01 5.86688459e-01
-1.87328473e-01 -2.50451922e-01 5.50169289e-01 7.04336941e-01
-1.03036165e+00 4.14807260e-01 -2.32666716e-01 -6.34507239e-01
-5.56355834e-01 7.77062535e-01 -2.70553946e-01 2.77734876e-01
-9.05974150e-01 5.33564568e-01 4.25051570e-01 -3.57352309e-02
-4.37068194e-01 -3.43860447e-01 -4.77376223e-01 2.63614863e-01
5.78077257e-01 -4.59622949e-01 1.04458153e+00 -1.26312113e+00
-9.68880236e-01 6.63469017e-01 1.77248698e-02 -7.03173503e-03
2.82275110e-01 2.76172101e-01 -5.18305779e-01 -4.35267165e-02
2.50763625e-01 4.64680977e-02 -1.40918404e-01 -1.43791151e+00
-1.13849986e+00 -3.24297011e-01 4.00692433e-01 4.09508675e-01
-5.60048342e-01 2.42909119e-01 -3.45619619e-01 4.42988016e-02
3.58841091e-01 -8.23247075e-01 -3.76286596e-01 -2.82808691e-01
-3.01432222e-01 -5.66011310e-01 2.83794254e-01 9.68428254e-02
1.20369697e+00 -2.12981677e+00 2.35322565e-01 9.33450580e-01
-1.50450379e-01 -4.67256814e-01 3.49979685e-03 7.14555085e-01
1.13093898e-01 3.50787640e-01 -1.96584836e-01 2.55666018e-01
2.62479156e-01 3.02215368e-01 -7.41548538e-02 5.87153256e-01
-2.95030534e-01 -1.29761785e-01 -6.64331615e-01 -4.27378416e-01
-1.33632913e-01 -2.27125790e-02 -6.24487638e-01 2.01962575e-01
-2.83377588e-01 -6.57408834e-02 -8.23548853e-01 3.22306335e-01
4.71745491e-01 -6.53049722e-02 7.13518500e-01 -2.80107826e-01
-3.91866684e-01 2.37897202e-01 -1.85019433e+00 7.74883151e-01
-2.17470989e-01 -3.07180554e-01 1.37032047e-01 -7.62268484e-01
8.80073071e-01 4.28990066e-01 9.42553997e-01 -3.90914470e-01
2.63124585e-01 5.38097203e-01 1.85806111e-01 -4.05977845e-01
5.06536603e-01 -1.67440489e-01 -2.31537580e-01 9.31602478e-01
-3.41204971e-01 3.45625252e-01 6.58511281e-01 6.64949045e-02
6.26521766e-01 -3.62442911e-01 5.74925780e-01 -9.99207020e-01
7.82755494e-01 -2.33544692e-01 8.75002682e-01 5.53458214e-01
1.48997501e-01 2.60994405e-01 8.57511938e-01 -4.71977621e-01
-9.47793663e-01 -9.43548441e-01 -1.45742996e-02 1.06242442e+00
5.93705177e-01 -2.80611396e-01 -4.35185134e-01 -3.89419436e-01
-2.82922611e-02 5.78107119e-01 -4.72921312e-01 -1.00598596e-02
-2.04844937e-01 -6.79988325e-01 -1.25127733e-01 2.75053591e-01
2.35404924e-01 -8.11658800e-01 -8.65594625e-01 1.16856679e-01
-2.63696879e-01 -8.14950585e-01 -4.33056563e-01 6.83541223e-02
-6.92160785e-01 -1.13396764e+00 -2.04335392e-01 -7.61399090e-01
9.56385493e-01 1.65746137e-01 9.25890923e-01 2.55538434e-01
4.96457070e-01 3.97921145e-01 -4.31813627e-01 -1.42235577e-01
-1.97021887e-01 -1.26216412e-01 5.44132054e-01 3.89399439e-01
-2.37616268e-03 -6.36657119e-01 -2.32300133e-01 6.85684800e-01
-1.05798578e+00 -1.08627543e-01 2.53587842e-01 2.88387567e-01
7.17920482e-01 5.57609916e-01 4.12949353e-01 -8.11751127e-01
7.89542317e-01 -4.96181041e-01 -8.25825155e-01 7.43326485e-01
-9.13400471e-01 4.53021318e-01 4.70827073e-01 -3.10382247e-01
-9.72608209e-01 1.09933712e-01 2.95515776e-01 3.05203736e-01
-7.52567453e-03 6.14533901e-01 -8.35833907e-01 1.30578920e-01
3.73091251e-01 5.90202138e-02 -3.61449212e-01 -4.32781518e-01
8.66991580e-02 6.76611066e-01 4.00598675e-01 -1.12052798e+00
4.29645449e-01 4.30048943e-01 1.99136958e-01 -7.24300742e-01
-4.03178155e-01 -4.81155843e-01 -4.13856387e-01 -2.10678473e-01
4.83493626e-01 -9.44049135e-02 -1.22439444e+00 -1.48781732e-01
-7.79193461e-01 3.57830882e-01 -1.67013615e-01 3.97516310e-01
-6.93101645e-01 3.11281413e-01 -1.29141077e-01 -1.31957948e+00
-1.88679725e-01 -1.11868238e+00 3.33279788e-01 1.24079525e-01
-3.66847992e-01 -5.93530297e-01 -1.65429384e-01 -2.41889775e-01
1.48364425e-01 2.04341322e-01 1.33230960e+00 -9.69656408e-01
-5.01771271e-01 -6.52996823e-03 1.16431460e-01 -3.99739780e-02
3.07075262e-01 2.25623161e-01 -2.37289324e-01 -1.77089706e-01
-8.68401453e-02 2.51385957e-01 2.46675849e-01 2.83839762e-01
7.25028574e-01 -8.78229082e-01 -3.85734707e-01 2.64352709e-01
1.57574391e+00 7.27041721e-01 2.66646385e-01 3.82281035e-01
9.24996138e-02 1.15102863e+00 6.92703485e-01 6.36094511e-01
3.42939883e-01 7.99678564e-01 4.33101505e-01 3.75431001e-01
8.65499377e-01 1.76008761e-01 1.81469232e-01 6.94202006e-01
-2.57200271e-01 -3.97260040e-01 -7.65069902e-01 7.60805011e-01
-2.04370904e+00 -7.72004664e-01 -2.41368428e-01 2.79762387e+00
7.75687873e-01 2.75849774e-02 5.74159384e-01 3.92636627e-01
9.20120299e-01 -4.34948653e-01 -1.09735921e-01 -7.65140176e-01
-1.41025305e-01 -2.22828373e-01 4.55164015e-01 7.33734369e-01
-6.71108902e-01 1.60523057e-01 6.14815998e+00 2.57102847e-01
-6.87485456e-01 -1.57715142e-01 3.17176700e-01 1.28568783e-01
-8.91753256e-01 1.51056871e-01 -5.62667131e-01 3.65686834e-01
5.20920873e-01 -6.93713784e-01 3.68886381e-01 7.14889407e-01
-2.01920141e-02 -2.40356192e-01 -1.35253656e+00 3.77600521e-01
-2.58945882e-01 -1.06994593e+00 -2.93690320e-02 3.86818945e-01
7.50516474e-01 -6.84835434e-01 -2.90928539e-02 -4.59059536e-01
4.53088313e-01 -5.79592228e-01 8.82343888e-01 3.57719839e-01
4.38027352e-01 -1.15580761e+00 7.28429496e-01 6.08966351e-01
-1.25444794e+00 -2.40788966e-01 -2.09070608e-01 -1.78270340e-01
-2.21155286e-02 4.90827680e-01 -5.28787494e-01 6.79831922e-01
7.65897572e-01 -9.41351429e-02 9.72568616e-02 1.24145472e+00
8.69046971e-02 3.56104895e-02 -5.44443667e-01 -3.03806186e-01
-1.35800578e-02 -5.43346226e-01 4.15872782e-01 7.51399815e-01
4.50836778e-01 4.98478055e-01 2.62897193e-01 7.25700855e-01
1.13856278e-01 4.25701618e-01 -3.69298071e-01 1.29622638e-01
6.62587106e-01 9.72488880e-01 -9.10226524e-01 -2.11113542e-01
-2.61944801e-01 3.44715416e-01 -7.81587884e-02 4.36978608e-01
-4.22898382e-01 -2.54598558e-01 8.62892687e-01 2.08565384e-01
-1.56291381e-01 1.01841167e-01 -2.77500063e-01 -7.41710842e-01
2.54799604e-01 -6.52433634e-01 8.89250398e-01 -3.62833291e-01
-1.13815093e+00 6.56631649e-01 2.49547899e-01 -1.14451635e+00
-2.41435245e-01 -2.79236853e-01 -3.44224542e-01 8.46903324e-01
-1.01550245e+00 -4.82037485e-01 3.99897322e-02 6.11383379e-01
2.07791803e-03 1.43252630e-02 9.31062579e-01 7.68708363e-02
-2.10895732e-01 1.20776586e-01 9.90228914e-03 -3.30346942e-01
1.46011099e-01 -1.17503715e+00 -5.33350587e-01 7.45058835e-01
-1.33509576e-01 7.82369912e-01 1.11897397e+00 -4.43212450e-01
-1.12477434e+00 -5.91033101e-01 1.44531882e+00 -9.48120467e-03
2.42486119e-01 -1.50584534e-01 -6.62779570e-01 6.24404967e-01
7.92161748e-02 -2.90873528e-01 7.83855021e-01 3.41996372e-01
-1.97088510e-01 -5.91314197e-01 -1.35460937e+00 5.13127327e-01
8.23195338e-01 -1.35013714e-01 -3.14762503e-01 2.39091605e-01
5.22932291e-01 1.74958766e-01 -7.12781668e-01 5.29779971e-01
9.31340814e-01 -1.09423101e+00 6.31021798e-01 -6.93203628e-01
-3.13489698e-02 -5.49406648e-01 -6.80744886e-01 -9.62684929e-01
-5.62678516e-01 -4.41230744e-01 5.18065333e-01 1.34887719e+00
5.48453271e-01 -7.22315073e-01 5.15429378e-01 1.07243776e+00
5.92331029e-02 -5.82956195e-01 -7.68619776e-01 -1.03587151e+00
-3.87119651e-01 -8.50070938e-02 1.15456104e+00 8.70349228e-01
3.25364172e-01 2.43893102e-01 -1.74953654e-01 2.96205521e-01
5.36530852e-01 6.27779067e-01 1.66633829e-01 -1.67050779e+00
-3.15268159e-01 -4.91128892e-01 -1.48579210e-01 -3.32889616e-01
-1.70226358e-02 -6.82309568e-01 5.93862869e-02 -1.72977567e+00
3.41081053e-01 -9.65019286e-01 -6.22901142e-01 2.90965766e-01
3.91753793e-01 -2.66179979e-01 1.34407297e-01 3.39922637e-01
-6.80986941e-01 5.09846210e-02 7.18945026e-01 3.33686590e-01
-5.49850762e-01 3.84083092e-01 -9.64886606e-01 7.52220452e-01
7.77506173e-01 -4.19907570e-01 -4.53878999e-01 -2.06611291e-01
6.92156136e-01 3.17395180e-01 -2.32143089e-01 -6.45594001e-01
1.67505786e-01 -8.79097700e-01 -1.75917044e-01 -2.81566471e-01
1.14463083e-01 -1.23836267e+00 7.43837118e-01 5.13749838e-01
-6.19548500e-01 2.14004025e-01 -3.70662212e-01 -2.70963483e-03
4.68228012e-02 -4.42743421e-01 5.25720596e-01 5.56578860e-02
-6.06031001e-01 8.68265703e-02 -2.98912823e-01 -3.16868812e-01
1.21171081e+00 -2.58803129e-01 -6.45910874e-02 -5.28358161e-01
-5.50728023e-01 4.04197067e-01 6.45189047e-01 1.42835587e-01
2.11429030e-01 -1.38396180e+00 -5.38206339e-01 6.10214891e-03
2.94250876e-01 -4.27226871e-01 -2.47858703e-01 8.07973683e-01
-1.45477623e-01 4.05995309e-01 -5.00658154e-01 -2.88124233e-01
-1.44975495e+00 7.16297746e-01 1.49018049e-01 3.45117934e-02
5.76556241e-03 4.92554814e-01 4.22972351e-01 -1.96603864e-01
2.16877490e-01 -1.68460578e-01 -3.77150327e-01 7.41792396e-02
2.81340450e-01 5.46434522e-01 -4.83584329e-02 -7.26868689e-01
-7.40813196e-01 4.80134100e-01 3.82448196e-01 -4.64431465e-01
1.32291615e+00 -1.71787083e-01 -5.67488968e-01 2.28572309e-01
7.24395573e-01 2.19072819e-01 -6.23151064e-01 -2.38235474e-01
2.70417243e-01 -3.97213161e-01 -4.70610887e-01 -6.73587024e-01
-8.87046576e-01 2.04022706e-01 2.21065715e-01 7.88266361e-01
1.68317521e+00 1.57024503e-01 6.42111674e-02 3.80573392e-01
6.84687018e-01 -1.03114259e+00 -4.62848186e-01 1.95154011e-01
7.81182826e-01 -4.18282390e-01 -8.91204104e-02 -7.21558928e-01
-6.34662926e-01 9.92186308e-01 4.30640489e-01 8.16754065e-03
5.72112978e-01 3.71641576e-01 -3.53736848e-01 1.28444090e-01
-1.02043390e+00 -6.00978315e-01 2.67437905e-01 4.79719222e-01
3.81672204e-01 4.43130583e-01 -1.24240601e+00 8.84152114e-01
-1.49331689e-01 -2.31174901e-01 6.38139904e-01 7.48708606e-01
-8.81111562e-01 -1.29673409e+00 -4.90546644e-01 2.91470706e-01
-2.49043167e-01 3.21032137e-01 -6.03866398e-01 6.45023584e-01
3.05598408e-01 1.24251807e+00 9.71535295e-02 -3.08918506e-01
5.29851794e-01 -1.42967969e-01 4.14781392e-01 -4.04632956e-01
-3.89356494e-01 2.61877388e-01 3.73461366e-01 -3.14355105e-01
-4.83251959e-01 -7.15053916e-01 -1.39662504e+00 -3.52384537e-01
-4.14414078e-01 8.78404081e-01 3.85616839e-01 8.93186688e-01
7.66332820e-02 -2.42588110e-02 8.51250291e-01 7.77240377e-03
-5.29437363e-01 -4.78375822e-01 -1.04941523e+00 3.70311320e-01
7.05271028e-03 -5.49999714e-01 -1.72373638e-01 -8.24192762e-02]
|
[7.962146282196045, 5.124330997467041]
|
f8d7fa68-0c12-4326-b149-d68fb1a27efe
|
neural-representations-of-cryo-em-maps-and-a
|
2104.01468
| null |
https://arxiv.org/abs/2104.01468v1
|
https://arxiv.org/pdf/2104.01468v1.pdf
|
Neural Representations of Cryo-EM Maps and a Graph-Based Interpretation
|
Advances in imagery at atomic and near-atomic resolution, such as cryogenic electron microscopy (cryo-EM), have led to an influx of high resolution images of proteins and other macromolecular structures to data banks worldwide. Producing a protein structure from the discrete voxel grid data of cryo-EM maps involves interpolation into the continuous spatial domain. We present a novel data format called the neural cryo-EM map, which is formed from a set of neural networks that accurately parameterize cryo-EM maps and provide native, spatially continuous data for density and gradient. As a case study of this data format, we create graph-based interpretations of high resolution experimental cryo-EM maps. Normalized cryo-EM map values interpolated using the non-linear neural cryo-EM format are more accurate, consistently scoring less than 0.01 mean absolute error, than a conventional tri-linear interpolation, which scores up to 0.12 mean absolute error. Our graph-based interpretations of 115 experimental cryo-EM maps from 1.15 to 4.0 Angstrom resolution provide high coverage of the underlying amino acid residue locations, while accuracy of nodes is correlated with resolution. The nodes of graphs created from atomic resolution maps (higher than 1.6 Angstroms) provide greater than 99% residue coverage as well as 85% full atomic coverage with a mean of than 0.19 Angstrom root mean squared deviation (RMSD). Other graphs have a mean 84% residue coverage with less specificity of the nodes due to experimental noise and differences of density context at lower resolutions. This work may be generalized for transforming any 3D grid-based data format into non-linear, continuous, and differentiable format for the downstream geometric deep learning applications.
|
['Dong Si', 'Nathan Ranno']
|
2021-04-03
| null | null | null | null |
['cryogenic-electron-microscopy-cryo-em']
|
['computer-vision']
|
[ 2.36412540e-01 3.48057359e-01 2.40867466e-01 -5.31708658e-01
-8.80725920e-01 -2.52235681e-01 1.78146109e-01 4.62009817e-01
-6.79915428e-01 1.38258970e+00 -1.64683059e-01 -5.90524614e-01
7.22376630e-02 -7.69123137e-01 -1.12855649e+00 -9.66063857e-01
-4.51532006e-01 1.01536882e+00 1.91743094e-02 -9.41745192e-02
1.19307272e-01 6.90089047e-01 -9.22605217e-01 4.26090270e-01
8.96276891e-01 8.72493505e-01 6.69268191e-01 5.07834554e-01
-1.31591991e-01 2.79884458e-01 -5.02975881e-01 2.83983082e-01
-1.66952740e-02 -2.25573167e-01 -7.75758505e-01 -3.74761939e-01
8.24126482e-01 -1.64872408e-01 2.68291473e-01 1.02559221e+00
3.28381866e-01 -7.66181247e-03 8.65935028e-01 -4.84762758e-01
-1.06127679e+00 -6.99425163e-03 -6.62023127e-01 1.17928460e-01
3.20403039e-01 3.24763894e-01 7.20291734e-01 -8.96445811e-01
1.12193632e+00 1.02353513e+00 1.00249481e+00 3.53988886e-01
-1.90752423e+00 -3.47414941e-01 -8.19283500e-02 9.87358987e-02
-1.37557673e+00 1.30352914e-01 4.24826950e-01 -5.37801504e-01
1.68467212e+00 -3.68451290e-02 6.43731773e-01 7.05845773e-01
1.09644437e+00 -4.58546132e-01 1.25646901e+00 -2.20985323e-01
4.06541020e-01 -4.83387619e-01 3.92066874e-02 6.58957839e-01
1.91392437e-01 -2.60599524e-01 -1.44167230e-01 -4.63739097e-01
1.14659762e+00 -1.48435794e-02 -2.76043743e-01 -4.60657090e-01
-1.08258021e+00 8.50908875e-01 8.73312414e-01 -4.88745607e-02
-7.88545012e-01 5.40898647e-03 1.61557332e-01 5.25028147e-02
4.88781452e-01 8.45753491e-01 -4.98073190e-01 1.32462338e-01
-9.45097089e-01 4.79837328e-01 4.05449569e-01 7.38207638e-01
9.96954799e-01 -3.43782641e-02 6.81919694e-01 6.33275032e-01
1.18909635e-01 1.13595493e-01 2.38993019e-01 -1.01085150e+00
1.66146234e-01 3.88363957e-01 2.23366335e-01 -6.68395698e-01
-9.50679719e-01 1.56762168e-01 -1.15919614e+00 7.31567740e-01
5.95107257e-01 -1.74256340e-02 -1.20779192e+00 1.67004311e+00
3.69988084e-01 -3.21416467e-01 -2.53416538e-01 1.15133727e+00
7.17070997e-01 7.74971545e-01 2.82425463e-01 -4.22228128e-01
1.20577896e+00 -2.41142407e-01 -4.46303517e-01 2.77708083e-01
6.55529261e-01 -5.35508871e-01 1.12295818e+00 4.08255011e-01
-1.10976517e+00 -4.64725673e-01 -1.32131755e+00 -4.19946194e-01
-4.93808210e-01 -4.11761701e-01 4.11238283e-01 6.39759749e-02
-1.27025867e+00 1.08198738e+00 -1.05294168e+00 -1.17044978e-01
4.22685325e-01 6.07264698e-01 -7.46279001e-01 4.01155442e-01
-1.04439521e+00 1.15735078e+00 5.16864419e-01 -2.67548144e-01
-3.81921321e-01 -1.08547676e+00 -7.24144578e-01 -8.49352702e-02
-3.08695912e-01 -5.72038710e-01 6.27869248e-01 -6.50419235e-01
-8.98514330e-01 9.35213387e-01 -3.47980350e-01 -7.27475882e-01
1.37801066e-01 1.90037310e-01 -1.02004163e-01 2.81296164e-01
-6.65808022e-02 8.59175861e-01 8.56410265e-02 -1.19637370e+00
-4.38530892e-02 -6.41074240e-01 -1.95269734e-01 2.00733855e-01
4.35217112e-01 -2.23648161e-01 4.92502242e-01 6.83643520e-02
4.95468050e-01 -4.25146580e-01 -5.13325989e-01 7.46879727e-03
-2.76290122e-02 2.09078640e-01 8.23919415e-01 -9.48272705e-01
5.78012407e-01 -1.33213162e+00 2.59470969e-01 2.36984387e-01
7.67949820e-01 -6.19420633e-02 2.00244591e-01 3.36921573e-01
-3.97697926e-01 9.74007100e-02 -4.49677229e-01 4.39307988e-02
-4.94143426e-01 5.08955307e-02 8.48245174e-02 8.26291203e-01
3.65916222e-01 9.36942875e-01 -6.45282805e-01 -3.98328692e-01
3.10519785e-01 9.20349658e-01 -4.89022136e-01 2.68066544e-02
-3.84734571e-01 6.86759770e-01 -5.19823050e-03 4.74467307e-01
1.06685293e+00 -6.07632935e-01 5.26788294e-01 -4.05510217e-01
-4.00985539e-01 4.81405139e-01 -5.73262870e-01 1.66248548e+00
2.03381851e-03 3.90501410e-01 3.83827776e-01 -6.64047003e-01
1.22163320e+00 -3.26537155e-02 6.63983166e-01 -9.09711957e-01
-3.45108598e-01 -3.38243097e-02 2.77442068e-01 1.35075316e-01
6.88720584e-01 -6.46334350e-01 3.64489436e-01 2.92286396e-01
1.85905516e-01 -5.60926944e-02 -1.17405780e-01 1.53194621e-01
7.73727775e-01 4.63012785e-01 2.85340786e-01 -6.48552895e-01
1.17764585e-01 5.08592904e-01 3.80659819e-01 3.12679380e-01
-4.72595319e-02 9.13578451e-01 5.11506379e-01 -9.69946861e-01
-2.19200492e+00 -1.19531178e+00 -7.54130900e-01 6.03073657e-01
-1.43814432e-02 -4.08740193e-01 -9.21275496e-01 1.17315259e-02
7.68321604e-02 1.16606936e-01 -5.80031514e-01 1.47234306e-01
-6.77515626e-01 -1.11705172e+00 3.05622309e-01 3.93598199e-01
7.33628348e-02 -1.22160304e+00 -5.98776102e-01 3.84746015e-01
1.28161088e-01 -8.78632665e-01 -1.45879760e-01 8.06892157e-01
-1.10107613e+00 -9.37927783e-01 -7.13241994e-01 -5.67716181e-01
8.21435094e-01 -1.35299176e-01 1.27614427e+00 3.20290811e-02
-4.88715142e-01 -4.07075524e-01 7.50395879e-02 -3.91651802e-02
-3.86011869e-01 5.61520681e-02 4.85867172e-01 -7.55706251e-01
6.51982963e-01 -8.22029114e-01 -6.21916533e-01 1.48162603e-01
-6.28059268e-01 3.47748697e-01 1.77928567e-01 1.08801270e+00
1.41441500e+00 -5.50945163e-01 5.66798389e-01 -1.00163555e+00
6.99739516e-01 -3.79272014e-01 -6.76349282e-01 -6.50154054e-02
-8.01432967e-01 2.17961162e-01 1.09288836e+00 -2.31128380e-01
-8.10201347e-01 1.68520257e-01 -3.48612905e-01 -3.65121365e-01
-6.02200508e-01 4.29058373e-01 1.12718336e-01 -2.43167937e-01
9.80368674e-01 8.97720680e-02 3.01180452e-01 -3.50493371e-01
1.39534131e-01 3.25448304e-01 7.14134514e-01 -9.06786621e-01
2.03628764e-01 4.59606260e-01 3.77974302e-01 -9.84599531e-01
-1.60213202e-01 -7.02653006e-02 -1.22846723e+00 5.54245114e-02
1.16463840e+00 -5.83945930e-01 -9.43837345e-01 1.45356372e-01
-1.14954531e+00 -4.50020909e-01 6.23358637e-02 3.54907691e-01
-8.85668218e-01 7.15421081e-01 -9.01795864e-01 -5.67237616e-01
-6.18926942e-01 -1.26689327e+00 9.20687020e-01 3.86539213e-02
-5.31662226e-01 -1.01453483e+00 1.18564203e-01 3.96975726e-02
3.49427521e-01 6.45450652e-01 1.34069371e+00 -2.46881947e-01
-4.46066111e-01 1.93603963e-01 -2.93203235e-01 -8.69351998e-02
1.49231538e-01 -6.62431642e-02 -7.61019588e-01 -3.20566446e-01
-1.20388806e-01 -4.22849387e-01 7.03136027e-01 8.84482086e-01
1.08855689e+00 -1.13263294e-01 -2.28970245e-01 1.04517138e+00
1.55138731e+00 4.36058342e-01 8.78501356e-01 5.83160162e-01
8.10510516e-01 4.78408515e-01 4.89867806e-01 2.03327000e-01
-1.18666235e-02 6.46889091e-01 6.16042137e-01 -4.91921365e-01
3.36592436e-01 -1.80595636e-01 -1.60878852e-01 2.12485537e-01
-6.94816351e-01 4.20809716e-01 -1.00482941e+00 -9.42382030e-03
-1.51924121e+00 -1.11520231e+00 -4.41267043e-01 2.20973229e+00
1.08402693e+00 7.67395496e-02 1.89220190e-01 -4.00765538e-01
6.38933241e-01 1.61683299e-02 -9.86895382e-01 -6.42293632e-01
-3.00206929e-01 3.94254684e-01 8.09007883e-01 7.47341335e-01
-8.40240657e-01 8.13409925e-01 7.05561161e+00 3.32001001e-01
-1.16943157e+00 -2.33824179e-01 9.21481013e-01 -1.15911260e-01
-2.34468639e-01 -1.80433050e-01 -7.59029627e-01 5.60972154e-01
1.34678912e+00 -5.39028086e-02 5.16402125e-01 8.16111565e-01
3.71684968e-01 -1.21911734e-01 -9.32422459e-01 9.51367855e-01
-4.95175689e-01 -1.96477735e+00 5.77710941e-02 5.62697411e-01
5.04678249e-01 3.70655388e-01 -4.27357815e-02 -3.44476134e-01
3.90509576e-01 -1.62182140e+00 3.09032708e-01 5.22906125e-01
1.45756567e+00 -1.14355373e+00 5.87270796e-01 3.06958050e-01
-1.04136693e+00 7.15649724e-01 -1.11862206e+00 -1.68943465e-01
4.44420397e-01 5.12024105e-01 -1.08802867e+00 1.95079044e-01
8.46746743e-01 3.29812139e-01 6.07713312e-02 3.17374408e-01
4.42745298e-01 3.59002613e-02 -5.06324112e-01 7.94226602e-02
3.70602220e-01 -9.50962842e-01 -4.68799472e-02 1.18697584e+00
-8.44767392e-02 2.13453293e-01 -7.62005802e-03 1.48523283e+00
3.58746648e-02 -9.23518687e-02 -6.00362778e-01 1.93578213e-01
3.98854822e-01 1.21212816e+00 -5.59245348e-01 -1.92436486e-01
-2.54270524e-01 7.33013868e-01 8.09329569e-01 5.46199858e-01
-8.47300172e-01 -3.16950738e-01 1.13810575e+00 5.49257278e-01
1.08770438e-01 -6.52363062e-01 -4.10292059e-01 -6.78587377e-01
-1.58655807e-01 -6.52380049e-01 -6.86395243e-02 -1.08751500e+00
-1.27561808e+00 7.31283128e-01 -1.33708436e-02 -4.37857836e-01
-2.47112319e-01 -1.07244778e+00 -3.92668217e-01 1.68365264e+00
-1.09631550e+00 -6.85750127e-01 -1.02714248e-01 4.67226654e-02
4.84310873e-02 5.45534082e-02 1.24788785e+00 -6.48571402e-02
-4.02199998e-02 2.65636325e-01 4.62735593e-01 -2.52953142e-01
5.31457782e-01 -1.42161071e+00 9.53318059e-01 1.80453748e-01
-3.97745430e-01 1.01975298e+00 8.87957275e-01 -9.86542165e-01
-1.26758552e+00 -1.07347465e+00 5.38069069e-01 -3.90642375e-01
4.83449399e-01 -3.43885899e-01 -1.58658409e+00 7.50818729e-01
-3.05343587e-02 -6.76178038e-02 5.75907767e-01 3.65774371e-02
-2.49641374e-01 5.11015475e-01 -1.77042794e+00 3.48741502e-01
8.66672277e-01 -6.87414229e-01 -3.79213542e-01 3.84145826e-01
6.46914482e-01 -6.50102377e-01 -1.52591205e+00 3.09209973e-01
5.78690052e-01 -1.08556938e+00 1.23198771e+00 -1.01108110e+00
4.85836476e-01 -3.47791433e-01 -9.00135115e-02 -1.21843266e+00
-7.42573202e-01 -6.81355596e-02 1.17045335e-01 2.28926599e-01
5.60581267e-01 -8.31893921e-01 1.09508371e+00 7.40533352e-01
-3.22625786e-01 -1.18839037e+00 -1.24029458e+00 -3.96441072e-01
6.06105387e-01 9.56714228e-02 5.27570903e-01 1.02720487e+00
2.60036170e-01 2.52633482e-01 -5.23033272e-03 1.79342031e-02
8.65140676e-01 -9.33679789e-02 4.44390863e-01 -1.28130579e+00
-7.79947564e-02 3.95658426e-02 -5.79455793e-01 -1.09481764e+00
9.81602669e-02 -8.39553297e-01 -2.12453514e-01 -1.67016792e+00
2.94497728e-01 -2.50455648e-01 -6.90159723e-02 2.76525587e-01
3.50506186e-01 3.14909011e-01 -2.42269203e-01 3.41615587e-01
-1.59661517e-01 3.03489387e-01 1.33174777e+00 8.44631717e-02
-1.21011376e-01 -9.61833417e-01 -2.61916459e-01 7.09661067e-01
9.27438855e-01 -2.45872051e-01 -9.06185620e-03 -9.22921374e-02
1.28445476e-01 4.09950837e-02 2.67131358e-01 -8.97871673e-01
-6.05984777e-02 -3.38513583e-01 1.01128709e+00 -9.07395005e-01
5.89435339e-01 -6.88464105e-01 5.42363763e-01 2.50856429e-01
-1.27623454e-01 4.89141911e-01 2.17525393e-01 1.50111839e-01
1.91627100e-01 1.85383379e-01 1.27915573e+00 -6.02298975e-01
-3.83166820e-01 3.50372523e-01 -3.74728352e-01 -2.75562197e-01
7.03682899e-01 -6.63735032e-01 -6.66235149e-01 -2.45560274e-01
-1.03452873e+00 -1.28908366e-01 1.18835950e+00 -4.15874600e-01
8.45460236e-01 -1.06084764e+00 -3.50277692e-01 3.84247899e-01
-1.86445102e-01 3.28029364e-01 2.46220320e-01 7.27498889e-01
-1.21273446e+00 5.74285448e-01 -8.27696502e-01 -9.27562654e-01
-1.21358240e+00 4.21580970e-01 7.60260284e-01 -3.49523984e-02
-9.84779179e-01 3.68202269e-01 3.46162409e-01 -5.45224011e-01
-3.11597139e-01 -4.97366160e-01 1.34402722e-01 -6.30891740e-01
5.97398162e-01 1.56173125e-01 1.67767629e-01 -8.79586518e-01
-2.52854586e-01 5.48975468e-01 -3.08386922e-01 2.67459512e-01
1.68294740e+00 1.19864985e-01 -2.77007759e-01 2.67634451e-01
1.04733205e+00 -6.27377927e-01 -1.67458785e+00 2.95139015e-01
-3.58803689e-01 -9.24205221e-03 -1.64293692e-01 -6.67207062e-01
-4.73136276e-01 7.33014822e-01 5.95466018e-01 -6.51956629e-03
5.96677363e-01 1.37063742e-01 5.41198909e-01 5.60124755e-01
7.26942003e-01 -8.76808882e-01 -6.18824780e-01 7.38662779e-01
8.23736966e-01 -9.86376762e-01 2.82502115e-01 -1.67588085e-01
-2.51071185e-01 1.28342116e+00 6.98240399e-01 -2.68714845e-01
4.25596945e-02 6.67246103e-01 8.38323385e-02 -4.73525375e-01
-6.79088056e-01 4.39895838e-01 -1.29514113e-01 1.12376952e+00
9.87594962e-01 8.19844231e-02 -1.53199703e-01 3.15063238e-01
-4.45199847e-01 -2.41767213e-01 3.40839565e-01 6.94944859e-01
-8.71668994e-01 -8.73596549e-01 -5.92294872e-01 5.23865283e-01
-2.94863939e-01 -2.89295077e-01 -5.15349746e-01 7.27287889e-01
-2.38519266e-01 3.24290663e-01 5.57121336e-01 -2.26401180e-01
8.21333006e-02 3.44015837e-01 7.35707223e-01 -4.08087343e-01
-2.98608989e-01 -8.94129872e-02 -1.80155039e-02 -5.54614067e-01
-3.29280972e-01 -1.59329146e-01 -2.24159384e+00 -9.08341885e-01
-1.39880955e-01 3.04813534e-01 8.36666167e-01 4.71784383e-01
5.73729575e-01 3.69590461e-01 -1.34412497e-01 -1.33544481e+00
-2.05365598e-01 -1.18578982e+00 -8.28294277e-01 4.58579481e-01
3.09436917e-01 -4.74463075e-01 -3.13529164e-01 1.30084753e-01]
|
[13.325600624084473, -3.086984157562256]
|
802d391e-8a9a-4d21-8218-5ac11c3b5d9a
|
discrete-contrastive-diffusion-for-cross
|
2206.07771
| null |
https://arxiv.org/abs/2206.07771v2
|
https://arxiv.org/pdf/2206.07771v2.pdf
|
Discrete Contrastive Diffusion for Cross-Modal Music and Image Generation
|
Diffusion probabilistic models (DPMs) have become a popular approach to conditional generation, due to their promising results and support for cross-modal synthesis. A key desideratum in conditional synthesis is to achieve high correspondence between the conditioning input and generated output. Most existing methods learn such relationships implicitly, by incorporating the prior into the variational lower bound. In this work, we take a different route -- we explicitly enhance input-output connections by maximizing their mutual information. To this end, we introduce a Conditional Discrete Contrastive Diffusion (CDCD) loss and design two contrastive diffusion mechanisms to effectively incorporate it into the denoising process, combining the diffusion training and contrastive learning for the first time by connecting it with the conventional variational objectives. We demonstrate the efficacy of our approach in evaluations with diverse multimodal conditional synthesis tasks: dance-to-music generation, text-to-image synthesis, as well as class-conditioned image synthesis. On each, we enhance the input-output correspondence and achieve higher or competitive general synthesis quality. Furthermore, the proposed approach improves the convergence of diffusion models, reducing the number of required diffusion steps by more than 35% on two benchmarks, significantly increasing the inference speed.
|
['Yan Yan', 'Sergey Tulyakov', 'Jian Ren', 'Kyle Olszewski', 'Yu Wu', 'Ye Zhu']
|
2022-06-15
| null | null | null | null |
['music-generation', 'music-generation']
|
['audio', 'music']
|
[ 3.07171792e-01 -1.07796639e-01 -4.99342605e-02 -2.74644256e-01
-1.09289730e+00 -4.69501406e-01 9.24915135e-01 -1.36446252e-01
-3.08254510e-01 6.46363318e-01 3.46337646e-01 3.09940688e-02
-9.68218148e-02 -8.42790365e-01 -7.61055350e-01 -9.64538276e-01
4.22827214e-01 3.31172556e-01 2.27429777e-01 -4.62022908e-02
2.33445510e-01 3.28038901e-01 -1.21062279e+00 3.49488080e-01
1.18702948e+00 8.78079593e-01 3.27594101e-01 6.58958972e-01
-7.56068435e-03 8.05977225e-01 -5.42318225e-01 -7.00207651e-01
-2.70620827e-02 -9.04328048e-01 -3.95638406e-01 1.51332840e-01
3.39360774e-01 -2.37005264e-01 -1.85100645e-01 1.14469564e+00
6.51608109e-01 1.90790489e-01 1.08724344e+00 -1.09620082e+00
-8.02366138e-01 8.12099457e-01 -8.35574746e-01 -1.47813290e-01
1.34601086e-01 2.92260051e-01 1.27001786e+00 -9.73279893e-01
6.37619913e-01 1.44674075e+00 4.49800342e-01 5.66010058e-01
-1.60517621e+00 -5.84084690e-01 1.47945091e-01 -1.00089656e-02
-1.39202249e+00 -4.87258792e-01 8.50464225e-01 -4.34293389e-01
5.24876654e-01 1.42449319e-01 3.92261952e-01 1.23690438e+00
-1.04485219e-02 1.05413651e+00 1.09104276e+00 -4.10173088e-01
2.22635239e-01 5.41806966e-02 -3.59489083e-01 6.52661204e-01
-1.70061812e-01 3.09124496e-03 -7.18457937e-01 1.18702225e-01
8.14307511e-01 -3.30503196e-01 -3.89532447e-01 -2.62476690e-02
-1.23598695e+00 8.74194324e-01 4.51474696e-01 3.87051910e-01
-3.61202657e-01 4.47367996e-01 7.33985156e-02 6.05137907e-02
7.03944623e-01 2.62112439e-01 4.75451685e-02 -6.77792504e-02
-1.30104470e+00 3.41929615e-01 6.92737877e-01 7.51674116e-01
5.25907934e-01 1.19199134e-01 -7.23019898e-01 9.18491364e-01
5.38886607e-01 7.71352649e-01 1.86036408e-01 -1.17935061e+00
5.52531183e-01 1.13143042e-01 -3.34261134e-02 -1.06149459e+00
8.45482200e-02 -5.20933807e-01 -1.14679229e+00 2.42420390e-01
3.99857372e-01 -1.60297319e-01 -8.85114014e-01 2.02724290e+00
1.96585268e-01 2.72573858e-01 -5.83708808e-02 8.30245972e-01
4.16361541e-01 1.02790415e+00 8.78050029e-02 -2.51345485e-01
1.00797069e+00 -1.02935588e+00 -9.82577920e-01 1.43005282e-01
2.50095487e-01 -1.04589927e+00 1.14405882e+00 5.91410220e-01
-1.38669670e+00 -5.77426255e-01 -1.05628574e+00 -5.76158985e-02
1.45014063e-01 3.27895731e-01 3.50813240e-01 6.22086704e-01
-9.00346518e-01 9.35024977e-01 -1.04539657e+00 1.85348466e-01
3.97004932e-01 1.81169167e-01 2.16843095e-02 -1.99688286e-01
-9.97468650e-01 6.12804055e-01 6.05248846e-02 1.19889244e-01
-9.66923773e-01 -7.86902785e-01 -6.15839124e-01 1.03450775e-01
1.34502277e-01 -9.46100235e-01 9.33463335e-01 -8.50538969e-01
-1.95085454e+00 4.82947290e-01 -7.65964985e-02 -2.82326818e-01
8.62724841e-01 -3.39554995e-01 -4.91565131e-02 1.04264893e-01
-1.41602391e-02 9.54336047e-01 1.18992996e+00 -1.44218624e+00
-3.28695387e-01 -1.43686935e-01 -3.30525786e-02 2.85074502e-01
-4.35544848e-01 -3.51002067e-01 -7.16854692e-01 -1.13345230e+00
7.48961493e-02 -8.93216312e-01 -1.31415918e-01 1.74729139e-01
-4.87638921e-01 -6.44455552e-02 6.05648696e-01 -7.43676901e-01
1.15899158e+00 -2.30963182e+00 9.17676568e-01 2.18357816e-01
1.02159970e-01 -5.65466397e-02 -1.52779758e-01 2.83375233e-01
3.09349507e-01 -2.99697798e-02 -6.82434738e-01 -8.96867812e-01
2.83879220e-01 2.92764336e-01 -4.18197483e-01 3.22910100e-01
4.36650008e-01 8.96011055e-01 -7.51378775e-01 -5.43465316e-01
1.41561836e-01 8.36135626e-01 -9.41216409e-01 2.04961464e-01
-5.06373882e-01 6.70093536e-01 -1.72625586e-01 2.41591588e-01
6.23857677e-01 -2.08994135e-01 2.22155854e-01 -3.31575066e-01
9.93443429e-02 2.85056513e-02 -1.37243247e+00 2.01435399e+00
-6.60031378e-01 6.46096826e-01 1.82759300e-01 -7.39640892e-01
7.13468850e-01 2.92461455e-01 2.68565178e-01 -5.15748680e-01
-7.81534123e-04 3.31953466e-01 -2.42280453e-01 -1.67235225e-01
4.42715764e-01 -3.61903518e-01 2.37078533e-01 4.26592857e-01
2.19672069e-01 -5.34344077e-01 3.20016831e-01 2.60122061e-01
5.61691344e-01 3.96478742e-01 -2.99420714e-01 -2.26966470e-01
5.23535728e-01 -4.43100482e-01 2.97783703e-01 5.66205084e-01
3.68297100e-01 9.03781712e-01 6.14654899e-01 4.78287131e-01
-9.32290494e-01 -1.36729634e+00 9.06873196e-02 9.73267138e-01
6.24456927e-02 -3.00622255e-01 -1.04035699e+00 -5.43815374e-01
-2.51344860e-01 8.33267570e-01 -5.31912088e-01 -9.20633525e-02
-4.99294460e-01 -8.38180661e-01 6.37736261e-01 4.83600020e-01
4.34966892e-01 -7.89721310e-01 1.09680872e-02 9.75545123e-02
-4.57787842e-01 -9.40711200e-01 -6.82901084e-01 2.94611938e-02
-8.97037864e-01 -4.74787980e-01 -1.29783320e+00 -6.34699523e-01
6.08699143e-01 -2.26398885e-01 9.54040587e-01 -1.26547009e-01
-9.87190679e-02 2.55789936e-01 -1.17510214e-01 1.01353504e-01
-5.94971836e-01 1.00319020e-01 -1.61521867e-01 3.16021144e-01
-4.57193941e-01 -7.17652261e-01 -7.12009549e-01 1.85657471e-01
-1.18595302e+00 1.93431541e-01 7.28988111e-01 9.44790125e-01
6.17291272e-01 7.69341737e-02 4.33564365e-01 -7.31705308e-01
8.45030010e-01 -1.92412719e-01 -5.00439763e-01 1.71152309e-01
-5.59175074e-01 3.42991203e-01 3.80383998e-01 -6.48465335e-01
-1.34174848e+00 -7.65705928e-02 -4.37212855e-01 -5.88539541e-01
3.19392115e-01 4.88588661e-01 -2.43740618e-01 1.90850496e-01
6.70060337e-01 2.75285602e-01 -1.31034404e-01 -4.27796841e-01
8.61501038e-01 2.79994398e-01 5.16089559e-01 -8.88737559e-01
7.97971904e-01 5.29324949e-01 8.08678195e-02 -7.86585748e-01
-7.33578622e-01 1.03748618e-02 -4.34490949e-01 -2.21651167e-01
1.03710330e+00 -8.43909144e-01 -5.87718010e-01 5.44257224e-01
-1.20880234e+00 -5.90715289e-01 -3.40487659e-01 5.49728513e-01
-4.93096739e-01 4.53598887e-01 -9.65451062e-01 -7.13806510e-01
-2.02118978e-01 -1.37982047e+00 1.14013338e+00 -2.57494836e-03
-1.18249908e-01 -1.09871948e+00 2.72779688e-02 1.81983888e-01
5.52279413e-01 1.87771723e-01 9.60642695e-01 7.08282962e-02
-6.39380395e-01 1.26398951e-01 -2.59523839e-01 6.42327070e-01
2.88218656e-03 2.38678426e-01 -9.80800390e-01 -1.87458813e-01
1.51610968e-03 -1.88224450e-01 1.29956079e+00 6.06767654e-01
9.06593084e-01 -1.96063407e-02 -7.24593177e-02 5.51044524e-01
1.22971928e+00 -2.88272351e-01 6.38167977e-01 -3.03196400e-01
7.16374457e-01 6.26913846e-01 3.07215631e-01 5.73919535e-01
1.21231593e-01 6.96260571e-01 2.94498384e-01 -7.24997297e-02
-7.16486514e-01 -3.75720471e-01 5.31382978e-01 1.15529680e+00
-1.47781864e-01 -5.04028320e-01 -5.10933638e-01 4.33604181e-01
-1.77176178e+00 -8.30746889e-01 -3.26860487e-01 2.02912259e+00
1.18395853e+00 2.30087653e-01 -1.33764476e-01 1.11558847e-01
6.18787766e-01 3.04905981e-01 -2.84305751e-01 1.29451737e-01
-2.28898451e-01 4.07249123e-01 3.38974670e-02 8.93059969e-01
-8.78807604e-01 9.33521390e-01 5.78846836e+00 1.28739703e+00
-1.10638475e+00 2.34935090e-01 7.54586279e-01 -1.69125050e-01
-6.31770432e-01 -1.46142691e-01 -6.54260278e-01 3.07055414e-01
5.75070202e-01 2.83510417e-01 5.13393462e-01 4.42845941e-01
2.49241188e-01 1.00651188e-02 -1.06756163e+00 8.08589995e-01
-5.10653183e-02 -1.38090444e+00 2.71347195e-01 -5.75105511e-02
1.13769591e+00 -4.30107385e-01 4.92331773e-01 8.82596821e-02
2.42578119e-01 -9.11861420e-01 9.75988030e-01 7.91855812e-01
7.52561629e-01 -7.83228695e-01 3.62736702e-01 2.05629915e-01
-1.05724382e+00 2.34406665e-01 -9.66147035e-02 2.68580288e-01
4.40629095e-01 8.49788010e-01 -1.08934395e-01 5.31534314e-01
3.60917091e-01 6.58952713e-01 -2.25724936e-01 7.12992728e-01
-5.83539248e-01 7.29936838e-01 -2.78560013e-01 8.65034387e-02
2.24107206e-01 -5.32293439e-01 6.05836868e-01 1.30014729e+00
4.17390227e-01 -1.38046950e-01 -5.56902848e-02 1.33702695e+00
-3.76940697e-01 2.11430304e-02 -1.51268795e-01 -1.29387036e-01
1.44490331e-01 1.08140397e+00 -7.63648748e-01 -3.49820048e-01
-1.11702964e-01 1.50640988e+00 1.00571111e-01 5.64377248e-01
-1.20583868e+00 -2.88480490e-01 4.07492310e-01 -1.05317764e-01
4.58391190e-01 -3.21528077e-01 -4.92762506e-01 -1.02254856e+00
1.78208953e-04 -8.61096382e-01 2.97462977e-02 -6.85129583e-01
-1.38000369e+00 4.82029587e-01 -1.46165714e-01 -9.82823670e-01
-1.43541873e-01 -3.11340243e-01 -4.49549735e-01 1.03937423e+00
-1.46807694e+00 -1.30188930e+00 -2.43664712e-01 6.02638602e-01
5.95401227e-01 1.36692718e-01 5.29076159e-01 7.75547266e-01
-4.42446589e-01 5.63488662e-01 1.35825351e-01 -1.78102404e-01
9.14350510e-01 -1.32758820e+00 1.19925976e-01 8.33468080e-01
3.41694951e-01 5.23503542e-01 7.29186237e-01 -5.57857513e-01
-1.20546675e+00 -9.48375821e-01 5.84777713e-01 -2.37550706e-01
6.73149765e-01 -3.65527719e-01 -7.51020789e-01 3.29446405e-01
5.02113938e-01 -2.81858116e-01 3.32144558e-01 -1.27697021e-01
-2.92213231e-01 -9.18800756e-02 -8.35239172e-01 8.40758920e-01
8.76579523e-01 -4.92819637e-01 -1.66403860e-01 3.53039354e-01
8.37434173e-01 -4.43081588e-01 -9.03127670e-01 2.63337344e-01
3.94987851e-01 -9.01373148e-01 9.28890407e-01 -6.69923276e-02
1.02417481e+00 -3.66660684e-01 -2.63402164e-01 -1.32585955e+00
-1.27487272e-01 -8.01037192e-01 -2.03152984e-01 1.53752112e+00
7.04954565e-01 -1.52178228e-01 6.43495917e-01 2.07369089e-01
-5.85021563e-02 -7.37524807e-01 -6.66371584e-01 -4.51851040e-01
2.53214836e-01 -5.55735528e-01 1.38611302e-01 7.45169520e-01
-4.56540883e-01 5.36819875e-01 -7.37993360e-01 9.59381238e-02
7.19884276e-01 -1.81911029e-02 6.65142059e-01 -7.70856619e-01
-8.48258436e-01 -8.23652387e-01 1.58577308e-01 -1.68520737e+00
-1.60577483e-02 -9.09061074e-01 1.83139980e-01 -1.55297196e+00
9.19065028e-02 -3.65038931e-01 -5.93718290e-02 4.25417833e-02
-3.84772331e-01 4.48602736e-01 3.63445431e-01 1.91925943e-01
-2.71985084e-01 9.43161070e-01 1.55645680e+00 -3.33537430e-01
-2.79382706e-01 3.65196876e-02 -3.90277088e-01 5.24644971e-01
4.40241247e-01 -6.03906870e-01 -4.43071544e-01 -6.65665805e-01
3.00258756e-01 1.17229387e-01 1.94006845e-01 -1.01170170e+00
2.85984159e-01 3.27067375e-02 2.63420939e-01 -3.62818509e-01
6.82093561e-01 -5.70672691e-01 1.07477114e-01 4.03109282e-01
-5.35396695e-01 -2.31381640e-01 7.38172531e-02 7.89788663e-01
-4.22221243e-01 -2.58232772e-01 1.03380275e+00 2.06908092e-01
-2.06713647e-01 1.70162648e-01 -1.68423593e-01 3.12944576e-02
7.38092661e-01 2.00775489e-01 2.51166113e-02 -5.62534332e-01
-9.23737466e-01 -7.93650523e-02 1.92780435e-01 1.88235015e-01
6.81015968e-01 -1.36016488e+00 -9.19412971e-01 7.07920417e-02
-2.14263022e-01 -1.76556915e-01 2.25716993e-01 1.15214026e+00
-5.47239602e-01 -4.04577702e-02 1.46839932e-01 -7.56761789e-01
-1.00699353e+00 2.60790586e-01 2.69062519e-01 -4.21381950e-01
-5.53244650e-01 1.07300639e+00 3.66832912e-01 -2.75197715e-01
4.47214186e-01 -3.55549663e-01 1.00400344e-01 1.20185457e-01
3.15595239e-01 3.72334361e-01 -2.66308933e-01 -3.95483345e-01
7.74021223e-02 6.05223238e-01 1.69554800e-01 -8.12785208e-01
1.20242047e+00 -7.26906136e-02 -1.61911577e-01 4.35245425e-01
1.04515231e+00 3.44239503e-01 -1.59008050e+00 -2.45670885e-01
-1.26038045e-01 -3.77000690e-01 2.15087190e-01 -6.93114996e-01
-1.36926901e+00 1.14374328e+00 5.28338075e-01 9.37842205e-02
1.13749850e+00 -1.01978399e-01 8.07183564e-01 8.14527348e-02
-7.19913244e-02 -9.40691113e-01 5.43075681e-01 2.40223512e-01
9.61156726e-01 -1.03838134e+00 1.92376450e-02 -4.89367425e-01
-7.40511179e-01 9.72966015e-01 1.35160059e-01 -1.94940984e-01
5.63047171e-01 2.86567479e-01 -1.75504878e-01 5.13460562e-02
-6.17480159e-01 3.33599262e-02 5.37233233e-01 3.22694749e-01
6.27543986e-01 7.20838725e-04 -3.91770482e-01 3.95680934e-01
1.60935372e-01 -2.07405463e-01 5.41263111e-02 5.49410641e-01
-1.65903866e-01 -1.20452178e+00 -2.20021635e-01 4.61395308e-02
-4.82893974e-01 -3.98781836e-01 -2.28861824e-01 5.44673920e-01
-9.13015828e-02 9.66515124e-01 -5.77133745e-02 -1.54964834e-01
2.36985296e-01 -1.02429755e-01 7.19702661e-01 -2.79152751e-01
-5.04740655e-01 5.84707797e-01 -9.97564346e-02 -3.62959355e-01
-4.83776271e-01 -5.48642337e-01 -1.12471843e+00 -3.90669376e-01
-4.72253680e-01 1.93426833e-02 6.62652850e-01 8.82086098e-01
1.75007209e-01 9.11318779e-01 4.40649569e-01 -9.74912465e-01
-6.09418094e-01 -1.04154170e+00 -4.97830033e-01 4.77607846e-01
4.46433462e-02 -5.08501112e-01 -3.66820604e-01 3.24242532e-01]
|
[11.44488525390625, -0.3229948580265045]
|
115ee166-5ac6-4bce-b83a-ac6f97e091b0
|
layer-wise-regularized-adversarial-training
|
2202.02626
| null |
https://arxiv.org/abs/2202.02626v3
|
https://arxiv.org/pdf/2202.02626v3.pdf
|
Layer-wise Regularized Adversarial Training using Layers Sustainability Analysis (LSA) framework
|
Deep neural network models are used today in various applications of artificial intelligence, the strengthening of which, in the face of adversarial attacks is of particular importance. An appropriate solution to adversarial attacks is adversarial training, which reaches a trade-off between robustness and generalization. This paper introduces a novel framework (Layer Sustainability Analysis (LSA)) for the analysis of layer vulnerability in an arbitrary neural network in the scenario of adversarial attacks. LSA can be a helpful toolkit to assess deep neural networks and to extend the adversarial training approaches towards improving the sustainability of model layers via layer monitoring and analysis. The LSA framework identifies a list of Most Vulnerable Layers (MVL list) of the given network. The relative error, as a comparison measure, is used to evaluate representation sustainability of each layer against adversarial inputs. The proposed approach for obtaining robust neural networks to fend off adversarial attacks is based on a layer-wise regularization (LR) over LSA proposal(s) for adversarial training (AT); i.e. the AT-LR procedure. AT-LR could be used with any benchmark adversarial attack to reduce the vulnerability of network layers and to improve conventional adversarial training approaches. The proposed idea performs well theoretically and experimentally for state-of-the-art multilayer perceptron and convolutional neural network architectures. Compared with the AT-LR and its corresponding base adversarial training, the classification accuracy of more significant perturbations increased by 16.35%, 21.79%, and 10.730% on Moon, MNIST, and CIFAR-10 benchmark datasets, respectively. The LSA framework is available and published at https://github.com/khalooei/LSA.
|
['Maryam Amirmazlaghani', 'Mohammad Mehdi Homayounpour', 'Mohammad Khalooei']
|
2022-02-05
| null | null | null | null |
['adversarial-defense']
|
['adversarial']
|
[ 2.71041363e-01 2.31858119e-01 4.03001875e-01 -3.19103152e-02
-3.14759225e-01 -9.23055589e-01 5.62760115e-01 6.17601499e-02
-4.63577867e-01 5.87628424e-01 -8.95872712e-02 -7.05368876e-01
-2.46521562e-01 -9.88615394e-01 -1.08464825e+00 -9.04201090e-01
-1.61708698e-01 -1.38606839e-02 2.78724551e-01 -4.58749712e-01
6.64083660e-02 1.08044326e+00 -1.29714382e+00 5.11121929e-01
6.21364832e-01 1.07688332e+00 -3.87854457e-01 6.02590680e-01
4.47561741e-02 7.77705133e-01 -7.51298666e-01 -5.92932403e-01
4.22903687e-01 -1.12065688e-01 -8.71897221e-01 -5.53420901e-01
3.84619534e-01 2.68512941e-03 -2.28617817e-01 1.41678333e+00
5.46986461e-01 -1.76100843e-02 6.92735434e-01 -1.25368786e+00
-3.06586564e-01 9.12300766e-01 -1.39747128e-01 3.00459445e-01
-1.56085074e-01 1.97854057e-01 3.75979602e-01 -6.31132543e-01
2.06510052e-01 1.26834178e+00 7.00487733e-01 7.71447897e-01
-1.00445127e+00 -1.01653516e+00 2.00226977e-01 5.32823801e-02
-1.11413968e+00 -2.72579104e-01 1.01605904e+00 -5.35084188e-01
7.28863358e-01 4.86814499e-01 1.76277667e-01 1.23493266e+00
4.01949793e-01 2.79529095e-01 1.05079329e+00 -4.62192833e-01
3.40729147e-01 4.07576680e-01 1.53293625e-01 5.04623950e-01
3.69386494e-01 3.81239682e-01 -3.89802344e-02 -1.52590619e-02
4.26632106e-01 -2.05366403e-01 -6.87554255e-02 2.54630316e-02
-4.63018149e-01 6.81832910e-01 9.33243036e-01 4.60304916e-01
-3.78701329e-01 1.50074363e-01 7.38680482e-01 4.77899224e-01
3.73097003e-01 7.18825638e-01 -4.19071913e-01 5.20969212e-01
-4.52179044e-01 1.73147216e-01 6.68593824e-01 3.14726025e-01
4.64335591e-01 5.86176991e-01 -6.09554835e-02 6.10147595e-01
2.92022049e-01 2.82182157e-01 4.14295673e-01 -5.12515604e-01
5.58418274e-01 8.38646889e-01 -2.67234087e-01 -9.79296803e-01
-4.45620000e-01 -6.79064393e-01 -1.07219315e+00 8.56845260e-01
4.01317775e-01 -4.51642603e-01 -8.98831904e-01 1.67525697e+00
2.31200889e-01 3.89173388e-01 5.03412247e-01 4.95367348e-01
7.27385581e-01 6.28632307e-01 1.88672081e-01 7.13287368e-02
9.14541423e-01 -5.27962565e-01 -3.09180528e-01 -1.73244461e-01
4.63494837e-01 -6.26568675e-01 9.14596319e-01 3.90028566e-01
-9.60476637e-01 -5.32740355e-01 -1.20004082e+00 6.46651328e-01
-9.49615657e-01 -2.49094725e-01 2.50394642e-01 1.01569974e+00
-8.12164903e-01 8.19494724e-01 -6.30884469e-01 5.86944558e-02
4.68798727e-01 5.63767672e-01 -4.19299066e-01 2.01820388e-01
-1.64901590e+00 1.03823340e+00 8.33440244e-01 2.69270360e-01
-1.19810700e+00 -9.32715237e-01 -6.63209200e-01 5.44849746e-02
-1.26316443e-01 -9.44947377e-02 6.96337581e-01 -1.31326032e+00
-1.26965213e+00 7.12274253e-01 5.48429132e-01 -8.80754948e-01
6.74111843e-01 -3.58596623e-01 -5.47918737e-01 -6.18805401e-02
-6.38136089e-01 2.41192907e-01 8.74440908e-01 -1.49845803e+00
-1.23528898e-01 -2.35573053e-01 4.09226894e-01 -9.43311974e-02
-8.79818261e-01 2.75517821e-01 1.10471666e-01 -8.84779692e-01
-1.49372861e-01 -8.50797474e-01 -2.25447506e-01 -2.45558441e-01
-5.90717852e-01 1.37396023e-01 9.56878304e-01 -6.17849708e-01
1.20380068e+00 -2.09480166e+00 5.51947691e-02 5.47389448e-01
-6.99469596e-02 1.06651068e+00 -7.55704418e-02 2.37858802e-01
-7.10837364e-01 3.07603210e-01 -4.29843277e-01 -9.36224833e-02
-1.47166893e-01 3.85138541e-02 -4.84125942e-01 4.81001467e-01
4.05376673e-01 6.18636370e-01 -5.45748889e-01 -5.02995402e-02
3.31225514e-01 6.29640758e-01 -4.51518208e-01 2.65223950e-01
1.89716965e-02 4.03997660e-01 -2.31967807e-01 6.20502532e-01
7.20775902e-01 5.91839790e-01 -9.12594423e-02 -2.84532398e-01
1.01367407e-01 -2.88276076e-01 -1.27447271e+00 7.88947344e-01
-3.92987818e-01 5.34113586e-01 -9.01051983e-02 -1.11441123e+00
1.19251454e+00 4.98512626e-01 1.29945830e-01 -3.63074630e-01
4.91224378e-01 2.78527975e-01 2.14433923e-01 -3.98948073e-01
-3.78214270e-02 -1.02534287e-01 -2.28661343e-01 1.44288346e-01
1.80651218e-01 3.60759526e-01 -1.66242555e-01 5.66044450e-02
1.11918294e+00 -1.71878427e-01 8.92537832e-02 -4.50038314e-01
1.06065404e+00 -4.31404740e-01 4.06697750e-01 5.03839612e-01
-2.22218454e-01 1.60694450e-01 5.99231899e-01 -5.95613241e-01
-1.05125320e+00 -1.02111590e+00 -2.58855671e-01 8.46873701e-01
-1.44615561e-01 2.45634317e-01 -1.07953060e+00 -8.51518393e-01
1.80734247e-02 7.88197637e-01 -7.44373977e-01 -8.49457204e-01
-7.25003362e-01 -7.89695442e-01 1.19707942e+00 4.60187495e-01
7.25973666e-01 -1.45410991e+00 -2.14165151e-01 -2.63166837e-02
3.59221548e-01 -8.85449708e-01 2.29197472e-01 4.54935223e-01
-7.43261933e-01 -1.06782818e+00 -4.70025122e-01 -6.78798974e-01
7.60887444e-01 -4.30797189e-01 8.52609932e-01 2.01579541e-01
-4.69083488e-02 -1.32046208e-01 -3.33394557e-01 -7.02014804e-01
-1.04262567e+00 -6.65590540e-02 4.79563773e-01 2.16207638e-01
-2.76948325e-02 -7.60020316e-01 -2.88926929e-01 3.74619603e-01
-1.17175281e+00 -6.02471888e-01 5.52746296e-01 6.27544105e-01
3.25964987e-01 3.66215289e-01 7.91696966e-01 -1.08018279e+00
6.74591243e-01 -7.25377262e-01 -7.25149930e-01 2.33163640e-01
-6.55532539e-01 -2.06486974e-02 1.17809105e+00 -5.75592279e-01
-8.52321386e-01 -1.40868217e-01 -5.17047584e-01 -7.87051797e-01
-4.30380195e-01 4.61543530e-01 -5.92110991e-01 -6.40435457e-01
1.10511875e+00 -1.99858531e-01 -1.84941188e-01 -4.98413980e-01
6.52612448e-02 2.93390989e-01 5.39004624e-01 -3.36455405e-01
1.21811056e+00 -1.65447555e-02 2.81756252e-01 -4.61956412e-01
-4.86129731e-01 -3.02974926e-03 -6.84849381e-01 -5.55377841e-01
5.38528860e-01 -4.87189919e-01 -5.90781868e-01 9.34371293e-01
-9.94665205e-01 -3.85352194e-01 -1.24340661e-01 1.66853413e-01
-7.33678862e-02 1.38259038e-01 -3.96977454e-01 -8.75076830e-01
-5.16553581e-01 -1.18128490e+00 8.57961699e-02 2.31326595e-01
1.94531992e-01 -1.25250566e+00 -1.07144699e-01 1.50267258e-01
4.73572671e-01 1.01660860e+00 9.35725629e-01 -1.04068506e+00
-1.21189550e-01 -6.53390527e-01 7.40414336e-02 1.02777064e+00
-1.40439719e-01 1.13473594e-01 -1.29820466e+00 -4.68452394e-01
-4.02757637e-02 -3.80656570e-01 7.86841691e-01 1.07072942e-01
1.12711585e+00 -7.23262608e-01 6.87606782e-02 8.62559795e-01
1.73169017e+00 3.15325499e-01 8.89694095e-01 6.40042901e-01
7.98659265e-01 4.92121100e-01 3.00892413e-01 1.50773019e-01
-5.22564173e-01 5.31056404e-01 1.11931872e+00 -2.11185664e-01
1.44554805e-02 5.48535213e-02 6.46481872e-01 3.89500529e-01
-3.36390696e-02 -3.00654292e-01 -1.13408005e+00 2.71694273e-01
-1.42544758e+00 -9.16299343e-01 -5.98839223e-02 2.14067435e+00
6.20444119e-01 6.98391557e-01 -8.92364681e-02 8.55110765e-01
6.88623726e-01 1.21978350e-01 -5.72630525e-01 -9.96048629e-01
-1.67631954e-01 3.07755351e-01 7.36908555e-01 6.38475120e-01
-1.38195598e+00 8.59093010e-01 5.35615683e+00 6.77438438e-01
-1.22421002e+00 2.17736568e-02 5.85133910e-01 3.85982618e-02
-3.80953308e-03 -4.51738447e-01 -7.20502436e-01 3.92609715e-01
1.19278824e+00 2.55072210e-03 4.24983382e-01 8.72927427e-01
7.01134577e-02 5.57225347e-01 -7.48482943e-01 3.29941869e-01
-5.80527969e-02 -1.28913009e+00 2.40045384e-01 -9.54061821e-02
7.92778492e-01 -4.24281061e-02 3.10065180e-01 3.41263801e-01
3.24880660e-01 -1.07031858e+00 6.39164805e-01 6.40717328e-01
5.95776975e-01 -1.24198270e+00 1.15091038e+00 3.35538447e-01
-1.02936387e+00 -4.09112334e-01 -3.02450806e-01 1.77804083e-01
-2.32589945e-01 2.89392799e-01 -6.42140150e-01 5.65326810e-01
8.54606211e-01 2.49390468e-01 -7.95945287e-01 6.78516924e-01
-2.11366132e-01 8.71733129e-01 -1.64421409e-01 2.51821905e-01
3.89342904e-01 1.52450800e-01 8.30830574e-01 1.15180862e+00
-6.18388057e-02 -2.43200332e-01 -3.07902753e-01 5.39360046e-01
-3.30919951e-01 2.70955414e-02 -7.65298903e-01 1.91595972e-01
6.31415844e-01 1.10881770e+00 -5.04843712e-01 7.45939985e-02
1.53971121e-01 6.43262208e-01 3.11069697e-01 3.56100142e-01
-9.26990449e-01 -4.61555272e-01 7.02393174e-01 -1.93976499e-02
-9.93153527e-02 2.92786241e-01 -4.45714384e-01 -5.41319907e-01
1.23022109e-01 -1.05008388e+00 3.99025202e-01 -3.72548968e-01
-1.16571629e+00 1.06233048e+00 1.97891574e-02 -1.37689590e+00
-4.49544527e-02 -9.73551750e-01 -1.13279188e+00 1.07354212e+00
-1.31017303e+00 -1.26068747e+00 -2.76732355e-01 8.86157513e-01
2.22820550e-01 -7.75287688e-01 9.51441944e-01 3.98452282e-01
-9.49480534e-01 1.23591208e+00 9.13731530e-02 4.08107728e-01
3.26478690e-01 -1.12275815e+00 3.76241922e-01 1.44198322e+00
-2.00812876e-01 4.64687109e-01 8.25640202e-01 -4.37453657e-01
-9.44567084e-01 -1.65267575e+00 5.17781615e-01 -3.97285521e-01
7.09358513e-01 -2.67589331e-01 -1.17080963e+00 5.05169332e-01
1.33481445e-02 1.31187573e-01 5.69776654e-01 -2.93278277e-01
-4.88373846e-01 -4.30929571e-01 -1.53449261e+00 6.30797446e-01
4.75338429e-01 -4.25060928e-01 -3.70528489e-01 1.85229555e-01
8.50376546e-01 -2.81132609e-01 -1.05975568e+00 6.32738411e-01
3.98778737e-01 -1.01779950e+00 1.22631788e+00 -7.95560062e-01
3.99737298e-01 -2.78196126e-01 -2.62954682e-01 -1.21569216e+00
-2.72286385e-01 -4.48435485e-01 -1.92773014e-01 1.47751248e+00
5.91114223e-01 -8.33443880e-01 6.78232968e-01 4.86027181e-01
-2.96572387e-01 -9.62767005e-01 -9.71002936e-01 -7.71941245e-01
4.95201141e-01 -5.06640136e-01 5.92537880e-01 9.94179726e-01
-5.26271045e-01 -3.37522388e-01 -2.07050696e-01 6.93402231e-01
6.15637302e-01 -8.01975489e-01 5.33857465e-01 -1.08011329e+00
-1.28943279e-01 -6.20022774e-01 -7.18114555e-01 3.23808283e-01
3.05561155e-01 -9.65097249e-01 -3.16071898e-01 -9.93853629e-01
-5.68306923e-01 -5.51875293e-01 -9.85051215e-01 6.17125094e-01
-1.91747308e-01 4.72355038e-01 3.56157482e-01 1.25175387e-01
1.03554420e-01 1.13116972e-01 7.58783102e-01 -1.62049815e-01
-7.13337511e-02 3.81353617e-01 -5.76083660e-01 9.13085520e-01
1.27869606e+00 -6.77952468e-01 -3.38865072e-01 -1.50030673e-01
1.41313478e-01 -4.58768845e-01 5.83946764e-01 -1.40465426e+00
3.00467703e-02 -4.88814479e-03 3.96507233e-01 -2.06612542e-01
7.68930987e-02 -1.10966206e+00 3.97490203e-01 9.04238164e-01
-4.81920809e-01 7.12047145e-02 6.41621530e-01 2.93472499e-01
-1.80961281e-01 -4.84499454e-01 1.24007952e+00 2.70700734e-02
-6.12426221e-01 2.02686250e-01 -1.53123260e-01 -2.16883972e-01
1.32249892e+00 -1.10870361e-01 -4.11709607e-01 2.33801305e-01
-7.73459971e-01 -6.84913769e-02 1.27263606e-01 3.92176598e-01
6.89559519e-01 -1.32599735e+00 -6.76462352e-01 3.38848650e-01
-1.46313012e-01 -2.02268049e-01 2.67481029e-01 3.80601496e-01
-7.30739892e-01 7.55028576e-02 -6.29695833e-01 -5.76828569e-02
-1.43448782e+00 7.97728479e-01 8.60715687e-01 -4.41696852e-01
-3.01748097e-01 1.11046219e+00 -4.22045626e-02 -4.03521240e-01
6.42560422e-01 2.26637386e-02 -7.04307795e-01 -1.44652454e-02
4.90717173e-01 4.37726587e-01 3.48222524e-01 -6.07040524e-01
-4.15543944e-01 2.89524704e-01 2.75376141e-02 2.51032919e-01
1.17508650e+00 3.95751506e-01 -1.51543662e-01 2.81299710e-01
1.19127738e+00 -2.08675489e-01 -1.04285371e+00 -9.34981108e-02
-1.55791119e-01 4.99382764e-02 9.88843385e-04 -9.28246617e-01
-1.29838693e+00 9.24296975e-01 8.97775769e-01 3.80441993e-01
1.48469830e+00 -4.17255789e-01 3.12474161e-01 3.59068096e-01
1.36850942e-02 -6.34211481e-01 -5.95024833e-03 4.59932894e-01
1.25335240e+00 -9.00690079e-01 -1.85585111e-01 -3.87995839e-02
-5.21797657e-01 1.13145161e+00 7.19836891e-01 -4.72743034e-01
9.93706822e-01 4.10423726e-01 2.10399121e-01 1.00484230e-01
-5.01280069e-01 3.17676008e-01 4.79512691e-01 5.77084005e-01
5.73154315e-02 4.65793833e-02 6.45762868e-03 6.51766598e-01
-2.34456137e-01 -4.73902017e-01 2.27450252e-01 6.91103041e-01
-3.58707070e-01 -9.46484089e-01 -6.50583565e-01 2.75251180e-01
-7.46264637e-01 -6.64976165e-02 -4.53321338e-01 6.94376647e-01
5.32409787e-01 8.07252169e-01 -3.12752426e-01 -9.34712172e-01
5.71042478e-01 1.54412195e-01 -2.96679372e-03 -1.98142245e-01
-1.30931711e+00 -4.71973360e-01 2.88837310e-02 -4.56035703e-01
-2.29349151e-01 -4.37726259e-01 -9.05481279e-01 -4.48827296e-01
-1.95390582e-01 -1.59501974e-02 7.32797921e-01 8.27153802e-01
-8.78731906e-02 9.89055097e-01 1.04612255e+00 -8.35554063e-01
-5.85433960e-01 -9.78547454e-01 -3.23451579e-01 4.74374473e-01
3.24208856e-01 -5.57808757e-01 -7.29423821e-01 7.92409386e-03]
|
[5.519748687744141, 7.931460857391357]
|
5aba3512-e0cd-4e92-b6c0-c6bf9a1afa6b
|
valor-vision-audio-language-omni-perception
|
2304.08345
| null |
https://arxiv.org/abs/2304.08345v1
|
https://arxiv.org/pdf/2304.08345v1.pdf
|
VALOR: Vision-Audio-Language Omni-Perception Pretraining Model and Dataset
|
In this paper, we propose a Vision-Audio-Language Omni-peRception pretraining model (VALOR) for multi-modal understanding and generation. Different from widely-studied vision-language pretraining models, VALOR jointly models relationships of vision, audio and language in an end-to-end manner. It contains three separate encoders for single modality representations, and a decoder for multimodal conditional text generation. We design two pretext tasks to pretrain VALOR model, including Multimodal Grouping Alignment (MGA) and Multimodal Grouping Captioning (MGC). MGA projects vision, language and audio to the same common space, building vision-language, audio-language and audiovisual-language alignment simultaneously. MGC learns how to generate text tokens in conditions of vision, audio or their both. To promote vision-audio-language pretraining research, we construct a large-scale high-quality tri-modality dataset named VALOR-1M, which contains 1M audiable videos with human annotated audiovisual captions. Extensive experiments show that VALOR can learn strong multimodal correlations and be generalized to various downstream tasks (e.g., retrieval, captioning and question answering), with different input modalities (e.g., vision-language, audio-language and audiovisual-language). VALOR achieves new state-of-the-art performances on series of public cross-modality benchmarks. Code and data are available at project page https://casia-iva-group.github.io/projects/VALOR.
|
['Jing Liu', 'Jinhui Tang', 'Weining Wang', 'Xinxin Zhu', 'Longteng Guo', 'Xingjian He', 'Sihan Chen']
|
2023-04-17
| null | null | null | null |
['audio-captioning', 'video-captioning', 'video-question-answering', 'video-retrieval', 'conditional-text-generation']
|
['audio', 'computer-vision', 'computer-vision', 'computer-vision', 'natural-language-processing']
|
[ 2.22480491e-01 -2.46950760e-02 -1.17968880e-01 -3.12063247e-01
-1.48901784e+00 -5.87002754e-01 7.38671303e-01 -1.81453675e-01
-3.47648978e-01 2.96057910e-01 8.40602100e-01 -3.45667183e-01
4.44567591e-01 -3.27091366e-01 -1.25974882e+00 -4.39419091e-01
3.75227660e-01 4.11931455e-01 -2.17453718e-01 -8.02145526e-02
-2.54682034e-01 -2.58264571e-01 -1.79459822e+00 1.10501611e+00
4.18636620e-01 9.18316543e-01 4.92804229e-01 1.37697995e+00
2.46141758e-02 1.01220787e+00 -2.50204086e-01 -4.36301291e-01
-1.04689993e-01 -4.25872505e-01 -8.65364134e-01 1.71158671e-01
7.85359204e-01 -4.37723041e-01 -6.55255973e-01 7.04741955e-01
7.70660400e-01 6.26693889e-02 7.81322896e-01 -1.69329762e+00
-1.11079288e+00 8.49292040e-01 -4.15351748e-01 -3.16044331e-01
7.25228786e-01 5.92539668e-01 1.36740398e+00 -1.26267743e+00
3.83838594e-01 1.75148237e+00 2.20519066e-01 8.48129034e-01
-1.12701046e+00 -6.79486334e-01 1.17107950e-01 3.17602187e-01
-1.21763098e+00 -6.64344370e-01 3.92985821e-01 -6.27787769e-01
9.76390064e-01 2.67134905e-01 2.20969304e-01 1.70208645e+00
-2.16731966e-01 1.24920964e+00 6.53194845e-01 -3.14826280e-01
-1.93262428e-01 -1.07214751e-03 -8.14857427e-03 4.42161113e-01
-4.13030803e-01 1.21252850e-01 -8.71126831e-01 5.42605706e-02
5.99611640e-01 -3.44098121e-01 -3.53369474e-01 5.76436780e-02
-1.67731297e+00 7.77628303e-01 2.50955850e-01 -6.22433349e-02
-1.46517873e-01 6.23456180e-01 5.38821876e-01 3.05633456e-01
-2.01462701e-01 3.46891619e-02 -1.92675024e-01 -2.65885387e-02
-4.59219962e-01 1.24685913e-01 3.53128463e-01 1.18224788e+00
5.34538031e-01 1.83930293e-01 -7.77657509e-01 1.04644608e+00
7.90328622e-01 9.51607704e-01 5.22018075e-01 -9.53441978e-01
8.93821955e-01 3.31457071e-02 -1.17446519e-01 -4.13152874e-01
-2.61439800e-01 6.05280288e-02 -9.50537920e-01 -2.62294978e-01
7.02689365e-02 -3.10332119e-01 -1.32561195e+00 2.03632021e+00
-4.61896248e-02 3.36614817e-01 6.32713616e-01 1.14498937e+00
1.80958474e+00 1.13434780e+00 2.76044220e-01 9.05629545e-02
1.60792422e+00 -1.37345910e+00 -5.34942746e-01 -2.24306196e-01
3.03418279e-01 -1.19899774e+00 1.31023943e+00 2.47328803e-01
-1.20209324e+00 -1.03699279e+00 -4.53840584e-01 -5.17883122e-01
-8.13758671e-02 3.36578429e-01 4.76792485e-01 -6.12186529e-02
-1.35030150e+00 -3.89701217e-01 -5.39257050e-01 -3.73786718e-01
8.41218159e-02 -1.45090837e-02 -4.46850687e-01 -2.50122160e-01
-1.25549769e+00 4.20309186e-01 4.88317281e-01 1.06720909e-01
-1.80382288e+00 -5.64599514e-01 -1.29483163e+00 -7.01007769e-02
2.09941164e-01 -1.27139711e+00 1.57706821e+00 -9.20339406e-01
-1.27833474e+00 9.62198317e-01 -3.46476108e-01 -4.61410940e-01
2.40601435e-01 -2.85750568e-01 -5.19466341e-01 3.38657886e-01
1.22262239e-01 1.52595007e+00 1.08615196e+00 -1.54897833e+00
-6.06213391e-01 1.16038412e-01 1.50395334e-01 5.35357475e-01
-2.41593435e-01 7.94468224e-02 -9.82420743e-01 -5.85618556e-01
-3.04217070e-01 -8.15334618e-01 1.17115103e-01 -2.04181880e-01
-6.97106123e-01 -3.64903569e-01 4.43283021e-01 -7.79800415e-01
7.11938798e-01 -2.23830938e+00 4.49860036e-01 -2.45373666e-01
1.02945320e-01 5.11272904e-03 -9.72344518e-01 5.76779664e-01
-2.70432293e-01 -1.20357908e-01 -3.66329704e-03 -7.32979774e-01
2.97753215e-01 2.43291005e-01 -6.91820443e-01 -3.97500731e-02
3.81808549e-01 1.14181209e+00 -8.20497572e-01 -6.18904293e-01
1.98323369e-01 6.83679044e-01 -6.90931141e-01 6.13330841e-01
-5.73573053e-01 4.60224003e-01 -1.03701465e-01 9.52337623e-01
3.22955966e-01 -4.16583896e-01 -2.22990021e-01 -4.75804836e-01
1.94985166e-01 2.64755059e-02 -8.32622707e-01 2.08679533e+00
-6.42860651e-01 8.65961254e-01 1.20735951e-01 -7.76867032e-01
5.72417915e-01 7.77023196e-01 3.21518242e-01 -8.22848618e-01
1.02000952e-01 -4.01969552e-02 -3.78073603e-01 -9.15110946e-01
7.46581852e-01 1.20304830e-01 -3.22894514e-01 4.13006917e-02
7.57779598e-01 -1.67779833e-01 1.83455974e-01 5.34524858e-01
6.57550991e-01 -5.27016679e-03 -2.63485789e-01 6.28919363e-01
5.42508066e-01 -1.99535862e-01 1.74597334e-02 6.78932965e-01
-4.02436480e-02 9.62075770e-01 2.41656587e-01 4.58995789e-01
-8.04115772e-01 -1.56048954e+00 1.79479897e-01 1.71248448e+00
1.38307393e-01 -5.97303629e-01 -4.01696593e-01 -1.44344002e-01
-1.39637170e-02 8.47757995e-01 -3.48084480e-01 -2.52023190e-01
-1.52555406e-01 -1.78644165e-01 8.64786148e-01 6.06941640e-01
1.89795628e-01 -1.25113940e+00 -1.26593253e-02 -1.39641091e-01
-8.80439997e-01 -1.53639829e+00 -7.12498724e-01 -2.53057510e-01
-3.05171043e-01 -6.85019612e-01 -1.01578629e+00 -1.07294250e+00
3.14704388e-01 4.29695487e-01 1.29843533e+00 -4.08511072e-01
-2.76602298e-01 1.36714172e+00 -4.39872086e-01 -3.67701530e-01
-4.99035984e-01 -3.72434974e-01 -6.06100485e-02 2.44561449e-01
1.94116235e-01 -2.98538536e-01 -4.24979329e-01 2.60493189e-01
-1.05676949e+00 4.72849190e-01 7.73366094e-01 9.61608887e-01
8.29223990e-01 -6.13227367e-01 5.52812517e-01 -3.55271958e-02
5.81600666e-01 -6.95588470e-01 -3.98770899e-01 3.15886915e-01
1.35942444e-01 -2.54758596e-01 3.53269577e-01 -7.18747556e-01
-8.13652039e-01 1.60015851e-01 -1.60256982e-01 -1.07026553e+00
-5.45115888e-01 6.68755054e-01 -4.08286035e-01 4.85182822e-01
4.46199417e-01 4.47369486e-01 -1.06404088e-01 -2.78283805e-01
1.05747569e+00 8.70939374e-01 1.39219725e+00 -6.67243540e-01
7.51331627e-01 2.62807101e-01 -3.96594673e-01 -8.96565557e-01
-6.75756037e-01 -7.49941409e-01 -1.10602953e-01 -4.25522834e-01
1.42503989e+00 -1.65257156e+00 -8.83615196e-01 4.18200046e-01
-1.49306619e+00 -5.47396541e-01 -1.01907276e-01 6.55819356e-01
-8.31613719e-01 3.27714890e-01 -4.78353530e-01 -7.04804599e-01
-2.86047429e-01 -1.24777555e+00 1.54104245e+00 2.84550071e-01
1.13340646e-01 -7.50186980e-01 2.22724434e-02 1.03773963e+00
2.12069955e-02 -9.19947699e-02 5.80412090e-01 -3.47744465e-01
-6.63855910e-01 2.91756570e-01 -4.04674023e-01 4.79016542e-01
-4.04924810e-01 -2.45290622e-02 -1.26796162e+00 -3.17046463e-01
-7.15770602e-01 -1.07526457e+00 1.27810395e+00 4.16790247e-01
1.26552165e+00 -2.72218764e-01 7.83838779e-02 5.53996801e-01
1.08353174e+00 -1.54439524e-01 7.12716937e-01 -7.75308535e-02
9.82527196e-01 5.60517907e-01 6.65201128e-01 4.27603275e-01
9.76157784e-01 6.93461001e-01 8.33390594e-01 -2.81274229e-01
-5.29622078e-01 -5.08151829e-01 1.01620245e+00 1.02974319e+00
2.17440233e-01 -5.63628674e-01 -8.53982031e-01 9.17519093e-01
-1.97601390e+00 -9.86219347e-01 -1.59940705e-01 1.88717818e+00
9.59608912e-01 -4.38104749e-01 2.67653167e-01 -2.82919317e-01
6.08579338e-01 -6.73733011e-04 -4.40135837e-01 -1.50383607e-01
-3.32127422e-01 -1.85805395e-01 7.18371868e-02 6.52807236e-01
-1.21591127e+00 1.02637541e+00 5.11062670e+00 7.87658393e-01
-9.89451289e-01 1.95852309e-01 2.59399116e-01 -3.48471135e-01
-7.11595595e-01 -2.15661824e-01 -7.74912417e-01 2.12294728e-01
1.06912303e+00 1.34791225e-01 5.21415293e-01 5.12613654e-01
2.25854352e-01 1.41555622e-01 -1.37898660e+00 1.60539913e+00
4.42067951e-01 -1.20697415e+00 5.74745417e-01 -3.44861358e-01
4.96449918e-01 3.71234208e-01 3.51901650e-01 6.58218980e-01
2.87809908e-01 -1.41237211e+00 1.02813911e+00 6.33008599e-01
1.10073507e+00 -4.49841797e-01 4.64405179e-01 6.61648856e-03
-1.26980484e+00 1.01360725e-03 -1.58356786e-01 4.84550685e-01
5.10259271e-01 6.20820299e-02 -7.82803357e-01 6.51497364e-01
8.49733174e-01 7.31380403e-01 -5.56139231e-01 8.64723086e-01
-1.29552603e-01 7.34309673e-01 -5.67987263e-02 2.59998590e-01
2.94574171e-01 2.00939953e-01 6.79128289e-01 1.49479628e+00
2.39036351e-01 -2.34373838e-01 2.96083897e-01 7.88826227e-01
-1.85200378e-01 6.84643835e-02 -5.43407619e-01 -4.67341334e-01
3.84679288e-01 1.15974820e+00 1.15731165e-01 -3.79342645e-01
-7.82698095e-01 9.58241284e-01 -5.29186800e-02 7.19094932e-01
-1.12438428e+00 -8.94003212e-02 5.96362591e-01 -3.29390198e-01
2.28002131e-01 -1.90968663e-01 2.09451601e-01 -1.22038126e+00
-4.07739282e-02 -1.16554010e+00 6.82592392e-01 -1.47510493e+00
-1.49511373e+00 5.47608852e-01 6.79173097e-02 -1.34126484e+00
-4.77237642e-01 -7.01381505e-01 -3.90726000e-01 8.29419494e-01
-1.58100522e+00 -1.68700790e+00 -5.52566767e-01 1.27352035e+00
7.79204667e-01 -5.57982862e-01 7.83451080e-01 3.69810402e-01
-3.04046452e-01 7.89973676e-01 -3.13672692e-01 3.02893817e-01
1.16942430e+00 -1.02598119e+00 5.08315302e-02 7.11443543e-01
4.98665869e-01 2.08275676e-01 5.91817439e-01 -2.59596556e-01
-1.71123660e+00 -1.38192058e+00 7.30826855e-01 -5.03532648e-01
8.77506077e-01 -5.12237430e-01 -6.50302589e-01 7.95974612e-01
6.64857030e-01 -2.23868236e-01 8.05025458e-01 -1.47635445e-01
-7.51403630e-01 7.07120895e-02 -4.01784271e-01 7.25906968e-01
7.97241509e-01 -1.14151168e+00 -5.93914092e-01 6.30770028e-01
1.40797079e+00 -5.36339521e-01 -8.62475693e-01 3.18789363e-01
4.35411453e-01 -5.77591956e-01 1.27532697e+00 -7.97716260e-01
9.20407295e-01 -4.10507858e-01 -7.53597975e-01 -1.09177053e+00
-6.06151763e-03 -7.48114288e-01 -1.53237000e-01 1.52714670e+00
5.72337210e-01 -2.52448618e-01 3.65364663e-02 -1.00081615e-01
-4.52015251e-01 -3.42680722e-01 -8.09529364e-01 -5.71254432e-01
-9.63709727e-02 -8.74051273e-01 1.32330701e-01 8.19299459e-01
-1.98645994e-01 9.04756606e-01 -6.72031581e-01 4.73914117e-01
5.25803924e-01 5.67522421e-02 1.07525980e+00 -7.06827998e-01
-7.78087974e-01 -3.87074679e-01 -2.24882364e-01 -1.23806262e+00
2.88679510e-01 -1.16277182e+00 2.76801765e-01 -1.81862521e+00
2.93319374e-01 2.50971556e-01 -2.34000325e-01 7.86102772e-01
-8.23487490e-02 2.89017677e-01 4.81138080e-01 9.21785310e-02
-9.79621291e-01 8.01239848e-01 1.23975348e+00 -5.28630137e-01
-1.61409274e-01 -2.33783811e-01 -8.07859600e-01 4.64818269e-01
5.02155840e-01 6.35493621e-02 -6.16036892e-01 -9.74604428e-01
1.48016214e-01 5.17894208e-01 8.84811759e-01 -8.09919059e-01
2.00615481e-01 -1.17611729e-01 1.25116497e-01 -9.98260021e-01
8.42434883e-01 -5.09480536e-01 -1.62240937e-01 -3.94807309e-02
-7.71540046e-01 1.04095042e-02 4.05451685e-01 6.66771412e-01
-6.58667326e-01 1.14206746e-01 4.99503911e-01 1.50332049e-01
-9.33543265e-01 3.77455324e-01 -3.31657320e-01 2.26503074e-01
6.09694839e-01 3.59941572e-01 -7.36525357e-01 -9.87591028e-01
-9.13991749e-01 7.92045593e-01 -8.03361014e-02 1.00422740e+00
9.88186657e-01 -1.59132755e+00 -1.25464785e+00 -2.62789994e-01
6.55538678e-01 7.18048736e-02 5.45069337e-01 8.59416842e-01
-4.65064915e-03 5.60743988e-01 4.00585309e-02 -1.15928185e+00
-1.49793398e+00 5.92919111e-01 8.29309300e-02 1.54456183e-01
-3.65275204e-01 1.11153674e+00 6.59435213e-01 -3.55734080e-01
8.05109441e-01 -2.76272684e-01 -1.93872750e-01 1.39871575e-02
6.59514904e-01 -6.57452941e-02 -4.64552075e-01 -9.65110600e-01
-1.33941218e-01 5.23900568e-01 1.91966847e-01 -5.12846589e-01
9.17773962e-01 -2.69200176e-01 -9.23752487e-02 6.72655880e-01
1.25993490e+00 -2.66755909e-01 -1.04345345e+00 -2.96599478e-01
-5.51063716e-01 2.73203533e-02 -9.15396586e-02 -8.45430732e-01
-9.03355896e-01 1.22588313e+00 6.08598053e-01 -2.24750787e-01
1.25422490e+00 5.39084852e-01 7.67062664e-01 3.78161043e-01
-1.69542715e-01 -8.53524268e-01 7.43280172e-01 7.25784540e-01
1.50490582e+00 -1.45449686e+00 -6.36200428e-01 -1.08417213e-01
-1.20211720e+00 8.54370415e-01 8.59337151e-01 5.08154154e-01
2.24510163e-01 -7.05184191e-02 3.65742743e-01 9.76995155e-02
-1.17167544e+00 -6.24005497e-01 7.00872004e-01 9.69024837e-01
4.72357810e-01 1.95838630e-01 4.74895358e-01 7.26364732e-01
-3.54810327e-01 -2.29766816e-01 4.36414868e-01 3.72926086e-01
-2.10341379e-01 -6.83643103e-01 -7.16298640e-01 -1.20180864e-02
5.99653553e-03 -3.64079863e-01 -4.40114111e-01 4.80015934e-01
1.70529597e-02 1.36212206e+00 2.64371961e-01 -6.09768867e-01
1.66985750e-01 3.57896797e-02 3.80769163e-01 -5.05062580e-01
-1.86178893e-01 4.99549568e-01 2.84147680e-01 -5.76364160e-01
-4.66085166e-01 -4.09572750e-01 -1.34042656e+00 1.42851755e-01
1.07024722e-01 -1.36349097e-01 6.31546736e-01 7.17967927e-01
3.44105124e-01 7.57243097e-01 4.28034335e-01 -9.80633140e-01
-2.85808325e-01 -1.05979395e+00 -1.10817820e-01 5.13810456e-01
6.35854125e-01 -4.00262713e-01 -1.63372010e-01 4.73380506e-01]
|
[10.836864471435547, 1.245600938796997]
|
fedd5913-f39f-4e6b-9390-392c16a53666
|
mol-instructions-a-large-scale-biomolecular
|
2306.08018
| null |
https://arxiv.org/abs/2306.08018v1
|
https://arxiv.org/pdf/2306.08018v1.pdf
|
Mol-Instructions: A Large-Scale Biomolecular Instruction Dataset for Large Language Models
|
Large Language Models (LLMs), with their remarkable task-handling capabilities and innovative outputs, have catalyzed significant advancements across a spectrum of fields. However, their proficiency within specialized domains such as biomolecular studies remains limited. To address this challenge, we introduce Mol-Instructions, a meticulously curated, comprehensive instruction dataset expressly designed for the biomolecular realm. Mol-Instructions is composed of three pivotal components: molecule-oriented instructions, protein-oriented instructions, and biomolecular text instructions, each curated to enhance the understanding and prediction capabilities of LLMs concerning biomolecular features and behaviors. Through extensive instruction tuning experiments on the representative LLM, we underscore the potency of Mol-Instructions to enhance the adaptability and cognitive acuity of large models within the complex sphere of biomolecular studies, thereby promoting advancements in the biomolecular research community. Mol-Instructions is made publicly accessible for future research endeavors and will be subjected to continual updates for enhanced applicability.
|
['Huajun Chen', 'Xiaohui Fan', 'Zhuo Chen', 'Rui Huang', 'Kangwei Liu', 'Ningyu Zhang', 'Xiaozhuan Liang', 'Yin Fang']
|
2023-06-13
| null | null | null | null |
['domain-motif-prediction', 'protein-design', 'chemical-entity-recognition', 'forward-reaction-prediction', 'chemical-protein-interaction-extraction', 'catalytic-activity-prediction', 'retrosynthesis', 'functional-description-generation', 'property-prediction', 'reagent-prediction', 'protein-function-prediction', 'chemical-disease-interaction-extraction', 'true-or-false-question', 'molecular-description-generation', 'open-question', 'description-guided-molecule-generation']
|
['medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'medical', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 3.06632757e-01 -1.81545988e-01 -3.38774413e-01 -4.82307434e-01
-6.31213844e-01 -4.93448853e-01 4.31118190e-01 5.98782957e-01
-3.66309613e-01 9.80245590e-01 9.06550884e-02 -7.95909584e-01
-6.27101064e-02 -4.29853022e-01 -9.94460762e-01 -4.60411757e-01
-2.75346518e-01 2.86886781e-01 -3.92965786e-02 -1.82092220e-01
4.46279466e-01 5.81716120e-01 -1.49708235e+00 4.07834828e-01
1.24400735e+00 5.74146688e-01 4.98218030e-01 5.00758946e-01
-6.55462816e-02 6.03534281e-01 -3.68832111e-01 -3.11956912e-01
-3.35213184e-01 -1.62882879e-01 -4.98739958e-01 -3.00228626e-01
5.07804692e-01 2.29379639e-01 -1.98796377e-01 6.44598544e-01
4.59487528e-01 2.72875518e-01 5.93217373e-01 -6.24619007e-01
-7.17378795e-01 2.49632433e-01 7.63759539e-02 3.48961949e-01
3.44195634e-01 3.75278085e-01 1.00365174e+00 -9.60860848e-01
6.95732474e-01 1.15756083e+00 4.35834646e-01 5.63503087e-01
-1.39749336e+00 -7.91655242e-01 3.06923062e-01 -1.22188739e-01
-1.14673936e+00 -6.60342753e-01 1.35040209e-01 -7.02990770e-01
1.50994718e+00 2.36695603e-01 3.49839360e-01 1.25698471e+00
9.83691812e-01 5.00145078e-01 1.12430286e+00 -1.64143980e-01
3.31814706e-01 -1.64783038e-02 5.56087971e-01 8.63942146e-01
4.19050425e-01 1.70062497e-01 -9.44580853e-01 -1.23175249e-01
5.61200321e-01 6.27937866e-03 -1.54776528e-01 -3.55111867e-01
-1.45034516e+00 5.62105954e-01 1.61178306e-01 2.32552245e-01
-3.22943568e-01 -6.65515959e-02 1.54082224e-01 2.07803175e-01
3.21898729e-01 7.36562729e-01 -6.26567721e-01 -3.12724501e-01
-4.01848674e-01 4.08681005e-01 9.17242110e-01 8.35691035e-01
5.46917140e-01 -1.19251408e-01 2.23339256e-02 7.79686809e-01
1.54878378e-01 3.14989328e-01 4.83014315e-01 -5.31410515e-01
4.22369152e-01 5.99471509e-01 1.26439497e-01 -7.77748048e-01
-8.30047309e-01 -6.78335726e-01 -4.57278401e-01 -1.60024166e-01
3.32561642e-01 1.21533133e-01 -5.02741933e-01 1.89602256e+00
2.24408790e-01 -1.87645465e-01 3.13591324e-02 6.32767916e-01
1.02814233e+00 5.14627814e-01 8.09189200e-01 2.05312110e-03
1.38117731e+00 -7.42463112e-01 -3.74545246e-01 -2.20107421e-01
1.13062358e+00 -7.45153248e-01 1.27254152e+00 5.57616651e-01
-1.03122425e+00 -5.45529068e-01 -1.26245534e+00 -1.88914865e-01
-7.56505311e-01 -6.35654405e-02 1.14140952e+00 7.36196935e-01
-8.12000453e-01 5.01238942e-01 -7.96269298e-01 -3.46799821e-01
3.88586998e-01 7.48802245e-01 -4.84021336e-01 -2.50549614e-01
-1.11530209e+00 9.07362223e-01 3.54748040e-01 -3.56438905e-01
-1.00657558e+00 -9.49098825e-01 -9.88025308e-01 9.42411423e-02
1.52324125e-01 -5.64196825e-01 1.06543660e+00 -1.42983735e-01
-1.19173777e+00 8.08766246e-01 -4.71437275e-01 -2.53979623e-01
-8.27357545e-02 -1.27908681e-02 -4.25039649e-01 -2.98737347e-01
-1.33939683e-01 6.25893354e-01 2.73005933e-01 -8.31944644e-01
-2.46055350e-01 -3.69109541e-01 -1.69733420e-01 1.12837128e-01
-3.45715046e-01 -3.75475436e-02 -2.72942156e-01 -7.45588899e-01
-9.94346663e-02 -8.09053183e-01 -2.03240603e-01 -5.48445463e-01
-9.61095914e-02 -2.93903381e-01 7.06766248e-02 -3.72608870e-01
1.23992252e+00 -2.04746628e+00 2.12205842e-01 -3.55401710e-02
5.79685211e-01 2.99578726e-01 -4.04617846e-01 6.50627434e-01
-3.30219150e-01 1.22427866e-01 4.09219787e-02 -1.52341992e-01
-3.54147479e-02 -2.14737937e-01 -1.89321533e-01 3.60369861e-01
2.99760997e-01 1.22427309e+00 -1.01346493e+00 -1.15135029e-01
3.06101399e-03 3.08596075e-01 -6.40423000e-01 1.03186518e-01
-6.53530538e-01 5.92588961e-01 -6.65077448e-01 9.26460028e-01
3.54368538e-01 -5.33969700e-01 2.00911433e-01 -5.33886626e-02
-2.27689072e-01 5.29763520e-01 -4.41111565e-01 1.60981643e+00
-1.73553929e-01 3.62772286e-01 -1.01977438e-01 -6.69136226e-01
9.72874939e-01 8.44700189e-05 5.45622170e-01 -9.44397271e-01
-8.22678804e-02 1.80526748e-01 3.58796895e-01 -5.45244634e-01
5.38372695e-01 -1.43359691e-01 9.89236087e-02 5.05520761e-01
-1.79562084e-02 1.68443426e-01 4.71437931e-01 3.77447635e-01
1.07105231e+00 2.94695050e-01 2.83290207e-01 -4.33797419e-01
5.60615540e-01 9.04454738e-02 2.16682360e-01 6.75045311e-01
-3.20623428e-01 -3.24628502e-01 1.67858049e-01 -3.33467335e-01
-8.82268131e-01 -1.02832103e+00 -5.96783876e-01 1.70908022e+00
-1.27203465e-01 -8.10569823e-01 -5.78657925e-01 -3.32769066e-01
2.67590344e-01 4.44894344e-01 -4.40977901e-01 -2.73702532e-01
-4.25688207e-01 -1.00474215e+00 2.52700567e-01 3.49683046e-01
5.39793968e-02 -1.12171686e+00 -9.25812274e-02 4.31757540e-01
1.50042577e-02 -9.90121543e-01 -4.65077102e-01 4.20774043e-01
-8.96747470e-01 -1.14820027e+00 -4.87632006e-01 -7.34929860e-01
5.43899059e-01 3.94524395e-01 1.00117755e+00 4.67532203e-02
-4.18538034e-01 2.50962406e-01 -6.15277968e-05 -4.85486954e-01
-6.75955236e-01 3.24201286e-01 3.40125114e-01 -3.99671078e-01
6.57760143e-01 -3.70133609e-01 -5.19901216e-01 4.51967984e-01
-8.49088490e-01 5.87813966e-02 8.39056611e-01 6.74853981e-01
6.55444086e-01 -5.33021808e-01 1.00139654e+00 -1.13185000e+00
8.78537238e-01 -5.58740258e-01 -4.27076250e-01 2.81472683e-01
-6.82338536e-01 2.91104257e-01 5.80941617e-01 -5.20970225e-01
-7.44218767e-01 -2.40464613e-01 -2.62025386e-01 5.95661879e-01
-2.83712029e-01 8.29676151e-01 -2.87385106e-01 -4.88666892e-01
7.33992577e-01 4.92901653e-01 -1.18097670e-01 -4.90616381e-01
6.80845901e-02 5.59605896e-01 5.30441523e-01 -1.18431580e+00
2.30109051e-01 -2.48944074e-01 1.37528300e-01 -9.77798939e-01
-5.26769876e-01 -2.52252579e-01 -3.40616167e-01 -1.08588479e-01
6.86141133e-01 -9.46846485e-01 -1.21370566e+00 3.38311344e-01
-6.82162881e-01 -4.93472368e-01 5.16270280e-01 5.85665286e-01
-4.49964374e-01 5.34278393e-01 -7.88911283e-01 -3.56377602e-01
-1.31842703e-01 -1.57632852e+00 8.73165071e-01 1.61522940e-01
-6.48221791e-01 -1.00817668e+00 1.31838734e-03 6.70729756e-01
3.65951031e-01 -2.31427252e-02 1.56370902e+00 -8.70427191e-01
-6.85295463e-01 -1.78107575e-01 -6.26078248e-02 -2.51506031e-01
1.48551673e-01 -2.40002409e-01 -8.54799688e-01 -3.91836315e-01
-4.09142077e-01 -7.12784886e-01 7.31971979e-01 9.05341879e-02
1.22615540e+00 1.63369879e-01 -4.72136974e-01 5.58263898e-01
7.56585300e-01 4.12008077e-01 2.74459988e-01 4.12856162e-01
3.32312465e-01 5.38520455e-01 4.66248333e-01 2.14771926e-01
4.72970903e-01 5.53918183e-01 2.33776674e-01 1.13585591e-01
1.01505719e-01 -2.50708610e-01 4.56874669e-01 8.72084320e-01
1.05772458e-01 -2.65863925e-01 -1.01223338e+00 -2.35521533e-02
-1.43447673e+00 -6.44849539e-01 -3.88036147e-02 2.39369392e+00
1.20302558e+00 2.95014113e-01 -4.75071706e-02 -4.67456013e-01
1.17070727e-01 6.01762235e-02 -1.03591228e+00 -4.69164252e-01
-1.50142357e-01 1.34149939e-01 1.17551267e-01 5.26562154e-01
-8.59094024e-01 1.16831696e+00 7.14745474e+00 8.95503938e-01
-9.79942322e-01 -1.31500557e-01 7.66599715e-01 2.61826161e-02
-4.32186097e-01 -3.45133811e-01 -1.28614330e+00 5.24236917e-01
1.44174075e+00 -2.84322739e-01 4.64533955e-01 5.96083999e-01
3.26981068e-01 -7.10553825e-02 -1.47515082e+00 6.80822134e-01
-1.95271254e-01 -1.60849643e+00 1.82379678e-01 1.73381105e-01
4.33881998e-01 1.90767661e-01 3.98220628e-01 6.55859768e-01
1.91320628e-01 -1.34630513e+00 2.55488038e-01 5.50284803e-01
8.06794047e-01 -4.96668726e-01 2.54678100e-01 4.82407838e-01
-8.91304791e-01 1.94937229e-01 -3.41713756e-01 -2.49839440e-01
-2.85470366e-01 3.19284230e-01 -5.47215760e-01 1.87119111e-01
3.17210138e-01 6.74900115e-01 -6.88000977e-01 7.50341773e-01
3.47952366e-01 5.51429152e-01 1.57962054e-01 -3.11886668e-01
1.39081016e-01 -2.77983427e-01 6.07801862e-02 1.41476226e+00
-1.68727964e-01 2.44188026e-01 3.11614722e-01 8.25836897e-01
-1.74007803e-01 4.10920888e-01 -3.22412282e-01 -7.80943751e-01
4.35673863e-01 1.03713667e+00 -5.38187921e-01 -3.10404927e-01
-5.40476501e-01 4.71069574e-01 7.03794897e-01 4.55964208e-01
-5.92225611e-01 -2.97670513e-01 8.80244374e-01 -9.95315891e-03
-2.51740158e-01 -6.77631855e-01 -3.52628142e-01 -1.15370381e+00
-3.25550199e-01 -1.20580888e+00 1.36178449e-01 -6.33814156e-01
-1.04390430e+00 3.56592983e-01 -1.77184135e-01 -6.22053206e-01
1.00895844e-01 -1.12097704e+00 -1.99027315e-01 1.03298306e+00
-1.28974926e+00 -7.79655755e-01 -2.31378227e-01 2.03270629e-01
6.14512682e-01 -2.72046536e-01 1.05030501e+00 2.69063711e-01
-9.93130863e-01 6.73938632e-01 5.00028312e-01 -5.50712645e-01
1.06618381e+00 -7.80978680e-01 4.64461982e-01 3.49409521e-01
-1.52115420e-01 1.45913005e+00 6.97212517e-01 -7.84656584e-01
-1.85395145e+00 -1.10101128e+00 9.57905173e-01 -8.36849093e-01
8.42285633e-01 -6.84930444e-01 -1.00393271e+00 7.84550548e-01
-3.56811821e-01 -5.19344211e-01 1.39781296e+00 2.16590866e-01
-4.17277128e-01 1.90366983e-01 -7.63333261e-01 7.60642469e-01
1.09273458e+00 -5.87922812e-01 -2.79116303e-01 7.59878278e-01
8.65198076e-01 -4.56724852e-01 -1.32662582e+00 2.72034913e-01
6.86822772e-01 -6.94626927e-01 1.18800271e+00 -1.20863414e+00
4.86404508e-01 1.69517592e-01 -2.03039035e-01 -9.22594368e-01
-5.96855462e-01 -6.00107908e-01 -3.90810728e-01 7.50363588e-01
4.87405598e-01 -9.46826220e-01 6.60480082e-01 7.76369810e-01
-4.52473730e-01 -1.14147592e+00 -4.33123797e-01 -6.16404474e-01
4.52660829e-01 -3.82218301e-01 5.75846016e-01 5.86462319e-01
5.11145353e-01 4.23728943e-01 -1.40233606e-01 -5.54575287e-02
5.19555390e-01 7.77037367e-02 8.56536090e-01 -1.19573593e+00
-4.68585819e-01 -6.73574328e-01 -1.38028026e-01 -1.45438957e+00
2.45589226e-01 -1.19626248e+00 -2.75932103e-01 -8.60217452e-01
5.33505321e-01 -5.00726819e-01 -4.80906606e-01 3.93669605e-01
-4.65536028e-01 1.14570946e-01 -9.44334343e-02 2.57913440e-01
-8.61043394e-01 4.45596606e-01 1.33427286e+00 -1.98425829e-01
-1.45851970e-01 -2.34573901e-01 -1.03108990e+00 2.77588695e-01
7.79058754e-01 -1.94629624e-01 -2.31724128e-01 -3.00263446e-02
2.58149266e-01 1.50635848e-02 -2.30607718e-01 -9.52344179e-01
5.81196211e-02 -5.54692507e-01 6.10708296e-01 -3.93514931e-01
4.46246982e-01 -2.09885418e-01 -5.89165539e-02 6.45275593e-01
-6.64721549e-01 3.76143344e-02 5.20522356e-01 6.05158210e-01
9.64528322e-02 1.55603498e-01 5.11201739e-01 -1.07600369e-01
-6.81109250e-01 5.60295641e-01 -7.68957555e-01 -1.29093185e-01
1.00242674e+00 -1.47875696e-01 -7.41647303e-01 -1.11840583e-01
-7.89484441e-01 2.73509324e-01 5.80917001e-01 4.55095649e-01
4.77984905e-01 -7.71935880e-01 -3.37250382e-01 6.42359078e-01
5.55383265e-01 -5.38935244e-01 2.94991165e-01 8.29022706e-01
-4.23021108e-01 1.28852522e+00 -1.53738543e-01 -2.93162763e-01
-1.26959538e+00 7.75058448e-01 9.10938755e-02 -1.69583440e-01
-2.34553352e-01 6.58904314e-01 6.04004264e-01 -5.34405768e-01
4.07709211e-01 -3.82059187e-01 -5.89776710e-02 -4.34776485e-01
7.37969518e-01 1.57290399e-01 1.22681342e-01 -5.10308027e-01
-1.72121927e-01 2.39409402e-01 -4.97678071e-01 5.29312551e-01
1.32324147e+00 2.52261385e-02 -1.04863487e-01 5.00770926e-01
9.40521419e-01 2.89362948e-02 -1.05817175e+00 -1.49177134e-01
1.53813839e-01 1.32461503e-01 -2.90081263e-01 -1.02259326e+00
-1.38145983e-01 8.31942558e-01 3.16976756e-01 -2.44657442e-01
4.49459910e-01 -2.61463448e-02 7.33279467e-01 7.68772602e-01
5.49315393e-01 -5.08042693e-01 1.67693228e-01 7.97078013e-01
6.01444721e-01 -1.29682338e+00 -6.67379424e-02 -3.21433395e-01
-1.46043003e-01 1.03781080e+00 8.56517851e-01 5.62870383e-01
3.32622558e-01 2.99982876e-01 -1.61871433e-01 -2.53808230e-01
-1.10365999e+00 1.85382009e-01 3.03570300e-01 6.00319982e-01
1.25361621e+00 5.46716787e-02 -5.01428902e-01 8.30085814e-01
-1.15045719e-01 1.93484519e-02 1.23521149e-01 9.43748951e-01
-7.91776240e-01 -1.41424251e+00 -3.38919282e-01 5.82159877e-01
-3.59295160e-01 -4.47188407e-01 -5.14657915e-01 4.87491250e-01
-1.94884986e-01 9.21420455e-01 -3.24125588e-01 -3.76571178e-01
8.24845731e-02 3.86023372e-01 5.79082787e-01 -7.45074213e-01
-5.22555351e-01 -2.63703793e-01 1.79202989e-01 -6.46437764e-01
7.98800811e-02 -3.85708511e-01 -1.45871377e+00 -6.65907681e-01
1.48035930e-02 2.81975091e-01 7.80962110e-01 8.74845386e-01
8.50977480e-01 5.92344820e-01 -1.03613876e-01 -8.45014930e-01
-5.31591117e-01 -8.19122016e-01 -3.46173137e-01 2.85061985e-01
1.03823900e-01 -7.60187328e-01 5.67210428e-02 -3.25545743e-02]
|
[4.757711887359619, 5.735409736633301]
|
cf9e2dd5-eaf6-4e09-a322-6d6c90fe0729
|
residue-based-natural-language-adversarial
| null | null |
https://openreview.net/forum?id=eFGgjI4Wk-V
|
https://openreview.net/pdf?id=eFGgjI4Wk-V
|
Residue-Based Natural Language Adversarial Attack Detection
|
Deep learning based systems are susceptible to adversarial attacks, where a small, imperceptible change at the input alters the model prediction. However, to date the majority of the approaches to detect these attacks have been designed for image processing systems. Many popular image adversarial detection approaches are able to identify adversarial examples from embedding feature spaces, whilst in the NLP domain existing state of the art detection approaches solely focus on input text features, without consideration of model embedding spaces. This work examines what differences result when porting these image designed strategies to Natural Language Processing (NLP) tasks - these detectors are found to not port over well. This is expected as NLP systems have a very different form of input: discrete and sequential in nature, rather than the continuous and fixed size inputs for images. As an equivalent model-focused NLP detection approach, this work proposes a simple sentence-embedding "residue" based detector to identify adversarial examples. On many tasks, it out-performs ported image domain detectors and recent state of the art NLP specific detectors.
|
['Anonymous']
|
2022-01-16
| null | null | null |
acl-arr-january-2022-1
|
['adversarial-attack-detection', 'adversarial-attack-detection']
|
['computer-vision', 'knowledge-base']
|
[ 6.75067484e-01 1.26455277e-01 2.28491679e-01 -4.30107936e-02
-6.06089890e-01 -1.09874845e+00 1.17434466e+00 1.02922134e-01
-4.05682445e-01 1.89230144e-01 -3.92568037e-02 -5.14867067e-01
4.04846132e-01 -8.13461304e-01 -8.22892964e-01 -4.86767650e-01
4.95560430e-02 1.06608838e-01 4.49853212e-01 -4.82138008e-01
4.43607450e-01 9.18479860e-01 -1.03218853e+00 7.26188660e-01
1.23164527e-01 6.97400689e-01 -4.51037824e-01 1.25613630e+00
5.60060609e-03 9.27271485e-01 -9.21185553e-01 -8.17460537e-01
9.18022454e-01 -2.78706640e-01 -6.59248710e-01 -3.96618433e-02
9.34326470e-01 -4.31287259e-01 -1.03342474e+00 1.58180225e+00
4.99040633e-01 -4.02914852e-01 8.43380034e-01 -1.45919871e+00
-1.12074184e+00 3.06662858e-01 -3.65617186e-01 4.65072781e-01
5.97573757e-01 7.96795011e-01 5.31403065e-01 -7.08866179e-01
6.50762796e-01 1.66386294e+00 5.36942601e-01 7.82293856e-01
-1.34480667e+00 -7.36609578e-01 -1.31355748e-01 1.44919410e-01
-1.08641815e+00 -3.13797414e-01 7.92556345e-01 -6.14383280e-01
1.20070350e+00 3.95746320e-01 1.38674840e-01 1.60272479e+00
6.89008653e-01 8.17786813e-01 1.33247709e+00 -6.68911517e-01
5.56199700e-02 5.86373866e-01 -1.73394248e-01 5.34816206e-01
-1.83634628e-02 5.34389913e-01 -1.03814252e-01 -3.55403304e-01
5.68833947e-01 -1.73368812e-01 -1.69042394e-01 -1.53990343e-01
-9.28077042e-01 1.02528882e+00 3.86707842e-01 4.11434919e-01
-9.16077122e-02 1.95059031e-01 8.46038818e-01 8.77004445e-01
1.33252084e-01 7.71207273e-01 -2.55677938e-01 2.83880472e-01
-8.16756189e-01 2.30720058e-01 8.43752444e-01 9.96327043e-01
3.66121590e-01 2.17520460e-01 -3.39138776e-01 2.01009929e-01
2.17342570e-01 4.51954991e-01 5.29565215e-01 -2.78572530e-01
5.03891826e-01 4.02871460e-01 -2.80876964e-01 -1.45262992e+00
-3.79462466e-02 9.18807164e-02 -7.29149222e-01 8.87573957e-01
5.21031618e-01 -2.42985748e-02 -1.16372311e+00 1.28818357e+00
-3.40638980e-02 -2.32862309e-01 3.05468529e-01 6.90052927e-01
5.24488568e-01 8.66679370e-01 2.98307747e-01 2.17256740e-01
1.56034029e+00 -7.37272441e-01 -2.88054943e-01 -4.43736464e-01
3.81620973e-01 -1.08797395e+00 9.99007463e-01 3.46930236e-01
-8.14020813e-01 -5.57600498e-01 -1.25729883e+00 2.22189531e-01
-1.03084755e+00 -3.79620194e-01 -1.33992229e-02 1.13900805e+00
-1.00103629e+00 5.62321484e-01 -3.70302141e-01 -4.42341596e-01
5.42805254e-01 2.97862530e-01 -4.92190421e-01 -1.30553275e-01
-1.46889818e+00 1.17747533e+00 5.05650699e-01 -3.58716697e-01
-1.09431410e+00 -5.36817253e-01 -7.98739254e-01 -2.03213751e-01
3.11277155e-02 -4.40536916e-01 1.05057442e+00 -1.60658765e+00
-1.19532073e+00 1.36234999e+00 4.22425300e-01 -9.32383716e-01
9.94229734e-01 1.20513737e-01 -7.47167706e-01 4.65368122e-01
-3.29246372e-01 7.10419416e-01 1.83944988e+00 -1.14496708e+00
-1.55124962e-01 -1.98069915e-01 1.26485974e-01 -2.13493347e-01
-5.38789749e-01 6.66037738e-01 1.33835211e-01 -9.74345565e-01
-4.64886487e-01 -7.96153843e-01 -3.88359845e-01 4.30247217e-01
-6.23915195e-01 1.86883762e-01 1.25785041e+00 -4.69913304e-01
9.27214742e-01 -2.30246568e+00 -3.55026364e-01 1.33369312e-01
2.45316297e-01 8.96530092e-01 -3.30958009e-01 8.29142869e-01
-5.09675205e-01 5.27372301e-01 -2.38980725e-01 8.35494399e-02
3.34591806e-01 9.31500196e-02 -8.14578831e-01 8.01798165e-01
6.88086450e-01 1.01686478e+00 -6.80139840e-01 -5.39371371e-01
4.29772168e-01 3.08053672e-01 -2.53469437e-01 1.02284670e-01
-9.69292298e-02 -3.13133411e-02 -3.21069688e-01 5.83685577e-01
8.57946217e-01 3.23386222e-01 -3.91084403e-01 -1.09668970e-01
1.81990966e-01 -2.04519600e-01 -9.13371921e-01 1.06223488e+00
-9.22192931e-02 1.14427125e+00 -1.31903207e-02 -1.10802329e+00
7.97682583e-01 4.15276349e-01 -8.87216777e-02 -5.90229809e-01
9.71625969e-02 1.17422931e-01 1.63550183e-01 -5.81210077e-01
3.17492157e-01 -7.04043880e-02 -1.45586044e-01 6.57927766e-02
9.73262787e-02 -2.12690696e-01 -8.90239999e-02 3.50824952e-01
1.68958890e+00 -2.68505067e-01 6.37162626e-01 -9.32621881e-02
8.96438718e-01 2.31080115e-01 -9.70458090e-02 1.33759141e+00
-7.00044274e-01 7.41146207e-01 5.61286688e-01 -6.18413210e-01
-1.38048160e+00 -1.23898900e+00 -2.47151300e-01 7.53630400e-01
-6.39436990e-02 -1.77545354e-01 -8.75087202e-01 -9.51125503e-01
4.45335507e-02 5.71224630e-01 -6.07572794e-01 -4.80769426e-01
-5.65644979e-01 -4.02673692e-01 1.21760750e+00 2.52269328e-01
3.59068215e-01 -1.47253156e+00 -5.13709009e-01 1.61756396e-01
6.22142911e-01 -1.24323547e+00 -4.44613665e-01 1.05614193e-01
-5.20046115e-01 -9.72573817e-01 -6.10502422e-01 -1.01983976e+00
7.36974537e-01 -2.22373575e-01 1.21648669e+00 -1.21511683e-01
-8.89372587e-01 6.84515476e-01 -3.47128183e-01 -8.76772165e-01
-1.38681257e+00 -4.05697346e-01 -2.19588690e-02 2.33900221e-03
7.44335413e-01 -2.78215647e-01 -4.63999301e-01 -4.20672335e-02
-1.45426989e+00 -5.61974227e-01 8.14060330e-01 6.67369425e-01
1.00089118e-01 2.03625992e-01 2.58131623e-01 -8.77532899e-01
1.01299226e+00 -2.12259769e-01 -4.22614992e-01 1.77553803e-01
-4.56091344e-01 -6.93699941e-02 1.19406736e+00 -1.03648913e+00
-4.69702244e-01 3.49439889e-01 -2.59720117e-01 -8.43047678e-01
-8.43851268e-01 -2.47996636e-02 -2.24888936e-01 -4.43585962e-01
1.19867671e+00 5.66190898e-01 7.10966066e-02 3.61043885e-02
3.99210125e-01 6.00910366e-01 5.15754044e-01 -2.05869690e-01
1.52233016e+00 3.33558649e-01 -6.49133027e-02 -1.05061257e+00
-8.67902786e-02 -3.34470689e-01 -6.27145231e-01 -1.23667076e-01
8.25261474e-01 -6.26372159e-01 -3.47350955e-01 7.62619734e-01
-1.34998894e+00 8.54377225e-02 -4.82453316e-01 1.00413142e-02
-4.97961730e-01 8.13968480e-01 -8.25170338e-01 -8.22725654e-01
-3.20671231e-01 -1.24896443e+00 9.00280654e-01 -2.54817665e-01
-2.34873876e-01 -1.00029624e+00 4.70803194e-02 2.71305814e-02
3.89721096e-01 6.21411502e-01 1.00987220e+00 -1.04578781e+00
-3.46691906e-01 -9.27278578e-01 -2.10801974e-01 7.76428282e-01
-1.49442062e-01 2.05952123e-01 -1.18838441e+00 -3.80781740e-01
3.22467417e-01 -3.19582790e-01 7.40353048e-01 -2.17246413e-02
7.22637296e-01 -8.62821162e-01 -1.38694122e-01 2.41601318e-01
1.76903427e+00 1.07554972e-01 1.04843116e+00 5.53168952e-01
4.94196028e-01 5.06640077e-01 2.94702232e-01 1.34543970e-01
-6.20780945e-01 4.63105887e-01 7.38273323e-01 -2.97500253e-01
-1.68767769e-03 -2.12732658e-01 9.42365587e-01 1.60998814e-02
7.72302210e-01 -6.13386333e-01 -1.08655512e+00 4.81445163e-01
-1.55589664e+00 -1.23367643e+00 -1.68478042e-01 1.66207969e+00
8.31553102e-01 5.59806943e-01 -1.54874856e-02 1.47376806e-01
8.47886264e-01 3.36045027e-01 -5.04101098e-01 -1.11460066e+00
-1.81325570e-01 3.87896031e-01 9.40435410e-01 9.12272707e-02
-1.65886497e+00 1.00129795e+00 6.76670551e+00 9.73525763e-01
-1.05616856e+00 -2.63493508e-02 4.95594561e-01 2.58537263e-01
1.53709829e-01 -2.05839276e-01 -4.10332680e-01 5.23496270e-01
9.50020790e-01 -8.19965526e-02 1.27847850e-01 1.14038944e+00
1.90165155e-02 4.04335320e-01 -1.37087548e+00 8.62675548e-01
1.92973971e-01 -1.23099494e+00 4.68597054e-01 1.29710898e-01
4.40339118e-01 -2.08957382e-02 5.19288480e-01 1.55733496e-01
3.09163719e-01 -1.37661541e+00 7.38624990e-01 4.04388815e-01
5.18704116e-01 -6.60740376e-01 6.73153460e-01 4.37513411e-01
-7.32867181e-01 -1.62062779e-01 -6.04655564e-01 -1.66864600e-02
-1.84510276e-01 1.26616955e-01 -9.31751370e-01 -3.45749371e-02
5.53992629e-01 2.88616270e-01 -8.13556790e-01 5.93338668e-01
-1.74552172e-01 6.29558384e-01 -2.34439746e-01 -1.15748094e-02
5.90965807e-01 2.58305341e-01 9.35703158e-01 1.71774673e+00
-1.88811228e-01 -2.59481013e-01 2.51881540e-01 9.60857451e-01
-8.09355974e-02 1.04215994e-01 -1.29958284e+00 -3.43786865e-01
1.52558446e-01 1.08096576e+00 -6.04500234e-01 -1.96011215e-01
-4.89020735e-01 1.37459683e+00 -1.81694299e-01 1.05674796e-01
-9.16590869e-01 -4.74473268e-01 8.63918424e-01 2.12987423e-01
2.50096321e-01 -1.33696347e-01 -2.09706780e-02 -8.95992935e-01
1.42492473e-01 -1.31274903e+00 2.31273651e-01 -4.92479622e-01
-1.85575151e+00 5.70687890e-01 -1.85281411e-01 -1.33472681e+00
-2.65647054e-01 -1.24543488e+00 -9.32807267e-01 8.59640837e-01
-1.15112627e+00 -1.34050703e+00 2.58128434e-01 8.63973439e-01
6.75094306e-01 -6.53706193e-01 9.50473249e-01 -6.49410114e-03
-2.21433148e-01 9.49164629e-01 1.48875028e-01 7.87982464e-01
6.92792654e-01 -1.23966742e+00 9.04170096e-01 1.34898329e+00
3.63403559e-01 3.99317414e-01 9.82959449e-01 -3.39957058e-01
-1.47096014e+00 -1.24364090e+00 4.82431561e-01 -8.38860571e-01
1.00360334e+00 -5.39304435e-01 -8.80857408e-01 4.74659801e-01
4.86923993e-01 2.76996285e-01 3.13157350e-01 -8.18851590e-01
-7.65981078e-01 2.77491689e-01 -1.58876276e+00 7.97950625e-01
5.23712516e-01 -1.01163316e+00 -7.35524118e-01 6.90158188e-01
6.21278703e-01 -2.14933574e-01 -6.82479978e-01 1.03242010e-01
3.10535729e-01 -7.78972030e-01 1.43777633e+00 -8.79731536e-01
6.40355170e-01 -3.16259444e-01 -5.03574982e-02 -9.35546279e-01
-3.73542577e-01 -7.92788327e-01 -1.45142376e-01 1.20122731e+00
3.50662827e-01 -7.08473325e-01 7.09859252e-01 5.23878098e-01
2.42396846e-01 -2.89346606e-01 -8.95549834e-01 -9.39499736e-01
3.75461608e-01 -4.06436086e-01 9.25808102e-02 9.13258076e-01
-2.85292268e-01 3.59647423e-02 -3.55348319e-01 4.52939153e-01
5.55476665e-01 -4.46864873e-01 7.42758751e-01 -6.77270651e-01
-3.41575861e-01 -6.00146413e-01 -1.36815560e+00 -4.48818415e-01
1.64975181e-01 -7.27945626e-01 -1.05621666e-01 -8.03541541e-01
-1.62881330e-01 7.03884885e-02 -2.03648686e-01 3.86691570e-01
5.85225038e-02 5.77215016e-01 5.32244503e-01 2.66096413e-01
-1.08840548e-01 -1.58485502e-01 7.92862535e-01 -7.87247419e-01
2.89877146e-01 -4.95470278e-02 -3.80464792e-01 8.30890179e-01
9.23211455e-01 -8.48875403e-01 -2.85639405e-01 -1.57384932e-01
-2.27218438e-02 -5.34465075e-01 7.29936540e-01 -1.21045721e+00
2.23273158e-01 -9.96326581e-02 6.36831760e-01 -1.60650492e-01
3.70280445e-02 -1.06203401e+00 -1.55839950e-01 8.28648925e-01
-4.68257964e-01 1.26094073e-01 3.77435982e-01 7.50171423e-01
-2.53127486e-01 -5.99248290e-01 1.30657041e+00 -6.86661601e-01
-9.30222929e-01 2.49089926e-01 -9.23276126e-01 9.73139107e-02
1.46128142e+00 -5.56476116e-01 -3.30703557e-01 -2.64863878e-01
-6.32905066e-01 -3.38606149e-01 6.30982280e-01 7.22701907e-01
7.93160677e-01 -9.32010233e-01 -8.87247801e-01 2.21131921e-01
2.50562608e-01 -5.54259777e-01 -9.20861959e-02 2.27453768e-01
-1.01096082e+00 3.41601551e-01 -4.82938737e-01 -4.71197248e-01
-1.49785292e+00 1.35549128e+00 4.78854537e-01 -3.24387580e-01
-6.57604635e-01 7.93604493e-01 4.13488477e-01 -3.05915862e-01
1.00352690e-01 8.39611515e-02 6.32891282e-02 -1.46165848e-01
6.71371937e-01 -1.19970679e-01 -4.47543450e-02 -6.73587263e-01
-3.64901781e-01 5.34683131e-02 -4.62361455e-01 1.54353291e-01
8.41410220e-01 3.67840856e-01 9.12053064e-02 5.06823249e-02
1.48733234e+00 -5.26231416e-02 -9.41379964e-01 -1.46149009e-01
-1.58423092e-02 -5.11768520e-01 -2.38670275e-01 -7.08830059e-01
-7.10738659e-01 1.09292912e+00 9.60657001e-01 6.50400519e-01
1.05803311e+00 -7.17485249e-02 6.30936265e-01 4.19178575e-01
2.54103038e-02 -9.53362584e-01 2.35458434e-01 3.86142910e-01
1.09869397e+00 -1.25624669e+00 -3.61817144e-02 -2.96116561e-01
-4.77223188e-01 1.44678080e+00 5.90292573e-01 -5.51489830e-01
5.29488266e-01 5.50735652e-01 1.33684367e-01 -1.53258204e-01
-3.25415432e-01 2.08767548e-01 9.47811827e-02 1.09506679e+00
1.99830458e-02 -1.59438729e-01 3.80749702e-02 -6.49817735e-02
-1.29812524e-01 -5.47512472e-01 6.72060251e-01 1.01663506e+00
-1.64720878e-01 -1.22702610e+00 -7.49147475e-01 2.69722164e-01
-7.66203344e-01 -4.69069779e-01 -9.67469215e-01 9.97752130e-01
-1.95667036e-02 8.11877549e-01 -6.91197962e-02 -4.51082826e-01
3.74831676e-01 9.36711580e-02 4.03144896e-01 -5.73077917e-01
-1.20105839e+00 -6.56835258e-01 -3.67598265e-01 -4.88951921e-01
-1.31744102e-01 -5.10769546e-01 -6.48285985e-01 -3.68464589e-01
-1.64386496e-01 -3.58986259e-01 5.42800665e-01 7.55211949e-01
1.39769405e-01 2.11069196e-01 7.87665546e-01 -7.93755651e-01
-1.04801571e+00 -7.12830067e-01 -3.48254591e-01 7.71732509e-01
5.12655318e-01 5.83605170e-02 -6.09125376e-01 2.75299460e-01]
|
[5.880184650421143, 8.002958297729492]
|
b2968aa3-65ad-4cda-86b0-024be437044e
|
time-to-embrace-natural-language-processing
|
2302.10406
| null |
https://arxiv.org/abs/2302.10406v1
|
https://arxiv.org/pdf/2302.10406v1.pdf
|
Time to Embrace Natural Language Processing (NLP)-based Digital Pathology: Benchmarking NLP- and Convolutional Neural Network-based Deep Learning Pipelines
|
NLP-based computer vision models, particularly vision transformers, have been shown to outperform CNN models in many imaging tasks. However, most digital pathology artificial-intelligence models are based on CNN architectures, probably owing to a lack of data regarding NLP models for pathology images. In this study, we developed digital pathology pipelines to benchmark the five most recently proposed NLP models (vision transformer (ViT), Swin Transformer, MobileViT, CMT, and Sequencer2D) and four popular CNN models (ResNet18, ResNet50, MobileNetV2, and EfficientNet) to predict biomarkers in colorectal cancer (microsatellite instability, CpG island methylator phenotype, and BRAF mutation). Hematoxylin and eosin-stained whole-slide images from Molecular and Cellular Oncology and The Cancer Genome Atlas were used as training and external validation datasets, respectively. Cross-study external validations revealed that the NLP-based models significantly outperformed the CNN-based models in biomarker prediction tasks, improving the overall prediction and precision up to approximately 10% and 26%, respectively. Notably, compared with existing models in the current literature using large training datasets, our NLP models achieved state-of-the-art predictions for all three biomarkers using a relatively small training dataset, suggesting that large training datasets are not a prerequisite for NLP models or transformers, and NLP may be more suitable for clinical studies in which small training datasets are commonly collected. The superior performance of Sequencer2D suggests that further research and innovation on both transformer and bidirectional long short-term memory architectures are warranted in the field of digital pathology. NLP models can replace classic CNN architectures and become the new workhorse backbone in the field of digital pathology.
|
['Xu Steven Xu', 'Hong Zhang', 'Jitendra Jonnagaddala', 'Bangwei Guo', 'Xingyu Li', 'Min Cen']
|
2023-02-21
| null | null | null | null |
['whole-slide-images']
|
['computer-vision']
|
[ 1.22582115e-01 1.85470609e-03 -4.86001998e-01 1.36523366e-01
-7.72758126e-01 -3.78859520e-01 3.72906029e-01 4.97989178e-01
-6.42748356e-01 6.11831069e-01 2.48665288e-01 -6.26672566e-01
-1.26421958e-01 -6.98533475e-01 -4.19770569e-01 -6.97570324e-01
1.83278173e-02 4.81085539e-01 2.92472273e-01 7.84259960e-02
-1.46268323e-01 5.71782112e-01 -9.01969314e-01 3.93221349e-01
8.94748986e-01 9.28799510e-01 3.71103823e-01 8.61220777e-01
-2.45592922e-01 8.50593984e-01 -1.06300011e-01 -5.31418204e-01
-9.93089601e-02 4.90400754e-02 -7.27492154e-01 -5.14974773e-01
3.30383122e-01 -1.54598549e-01 -5.89170098e-01 9.89082098e-01
8.72538090e-01 -5.60491443e-01 3.64770681e-01 -9.09610987e-01
-9.20234084e-01 1.40310064e-01 -3.99917334e-01 5.17690480e-01
5.55472001e-02 8.83789241e-01 8.27802837e-01 -7.22648382e-01
9.19065475e-01 6.93962157e-01 1.28904712e+00 6.34168029e-01
-1.01058209e+00 -5.29497683e-01 -4.57537860e-01 4.14269686e-01
-1.17798483e+00 -4.87115309e-02 1.08721867e-01 -5.42824686e-01
1.40627658e+00 2.66843259e-01 1.09119344e+00 1.20010602e+00
8.50308537e-01 7.90771127e-01 1.07217813e+00 -1.21669091e-01
-3.30501683e-02 -1.68730438e-01 3.95748526e-01 1.17816937e+00
2.79175073e-01 2.39096612e-01 -5.13261318e-01 -3.71310532e-01
7.08021939e-01 1.31032720e-01 -5.63616097e-01 2.49801204e-01
-1.59762061e+00 6.51533604e-01 7.85096347e-01 3.76420081e-01
-3.23662549e-01 2.46417075e-01 7.00212717e-01 1.41186193e-01
4.00525749e-01 4.47634995e-01 -5.03579676e-01 -4.85566333e-02
-9.20675337e-01 -2.15334117e-01 6.91611469e-01 3.21542859e-01
3.33673805e-01 -2.59948373e-01 -4.82786953e-01 7.82830417e-01
3.10895383e-01 4.10805464e-01 1.06666386e+00 -5.24368107e-01
4.31595929e-02 9.46793914e-01 -3.19997996e-01 -9.44943905e-01
-1.12623179e+00 -6.15401685e-01 -1.07274687e+00 -1.70169964e-01
5.59417009e-01 3.34347710e-02 -1.26669788e+00 1.27929235e+00
-1.43201813e-01 4.81399089e-01 5.75010590e-02 8.74288261e-01
1.33422887e+00 2.49000117e-01 4.02290136e-01 9.38299298e-02
1.62220728e+00 -9.17148292e-01 -4.12775248e-01 -1.85178906e-01
9.76494789e-01 -5.39111733e-01 9.18907702e-01 7.38393292e-02
-7.05280185e-01 -2.63237178e-01 -8.64828110e-01 -2.74869889e-01
-6.61994696e-01 4.03758794e-01 9.37815845e-01 5.80543101e-01
-1.42713737e+00 3.23233038e-01 -9.76365864e-01 -8.83553982e-01
7.28207469e-01 4.35058117e-01 -5.52082419e-01 -2.89803684e-01
-1.17975712e+00 1.08981800e+00 3.74713689e-01 3.44597369e-01
-1.21133494e+00 -1.35622227e+00 -6.63372457e-01 -4.46164943e-02
-3.27026516e-01 -1.23255694e+00 1.14547050e+00 -6.15336955e-01
-1.17662454e+00 1.25801063e+00 -1.79397296e-02 -7.53475308e-01
3.75474960e-01 3.11096430e-01 -3.01147848e-01 4.31040794e-01
-1.26736984e-01 1.04917324e+00 1.10666625e-01 -4.62359548e-01
-4.14568901e-01 -3.43427986e-01 -1.35648876e-01 -2.69210320e-02
-5.09728432e-01 -2.57826626e-01 -5.95944285e-01 -2.69938916e-01
-3.50242138e-01 -8.12887788e-01 -5.30834973e-01 5.28460681e-01
-4.59469199e-01 -1.96210691e-03 4.89230990e-01 -8.07986379e-01
6.82627261e-01 -1.97242558e+00 -2.26175323e-01 2.65469775e-02
6.58352792e-01 6.97206736e-01 -3.90363663e-01 -4.31600697e-02
-1.06496550e-01 4.15078253e-01 3.16226423e-01 2.51079407e-02
-3.15972149e-01 -6.83686091e-03 3.03086936e-01 6.58836961e-01
2.02513680e-01 1.56882417e+00 -1.03320396e+00 -6.07209563e-01
2.49869868e-01 5.20082057e-01 -1.87818721e-01 -2.01117218e-01
-2.57001013e-01 1.16632134e-01 -2.71717548e-01 1.05686140e+00
4.32038307e-01 -6.84870601e-01 1.49126932e-01 -1.74835637e-01
2.03453019e-01 -2.15967968e-01 6.13049511e-03 1.61155820e+00
-2.78303444e-01 8.34946096e-01 -6.91352263e-02 -6.38247609e-01
5.94677091e-01 4.33995217e-01 4.33391690e-01 -8.36583614e-01
1.19008198e-01 1.93724796e-01 2.40081698e-01 -8.33044469e-01
-1.03011735e-01 -2.72815436e-01 3.63003701e-01 -1.86831921e-01
2.08956495e-01 1.99539497e-01 2.25286379e-01 -7.59268552e-02
1.71020365e+00 -3.44935536e-01 3.10651839e-01 2.94266343e-02
6.00872517e-01 3.93229336e-01 6.09527946e-01 8.20333540e-01
-7.32428968e-01 6.73343182e-01 8.02520335e-01 -5.94227970e-01
-7.42298186e-01 -1.03825784e+00 -1.98546603e-01 5.74702144e-01
-3.57315391e-01 -1.68584302e-01 -2.84186870e-01 -7.87066936e-01
2.84026843e-02 1.19812153e-01 -8.40556204e-01 -1.93434566e-01
-2.26269796e-01 -1.39723098e+00 1.29346776e+00 6.06325269e-01
6.43309712e-01 -9.96985257e-01 -2.87564546e-01 2.03717709e-01
-8.12802389e-02 -1.02356899e+00 -1.18434608e-01 2.30043098e-01
-9.01634276e-01 -1.60541511e+00 -1.14375341e+00 -9.99707878e-01
7.93865979e-01 -1.68215428e-02 8.72959912e-01 2.28124097e-01
-6.46214604e-01 3.03120047e-01 -1.04312517e-01 -6.40167952e-01
-4.05459583e-01 1.30089656e-01 -2.85462260e-01 -4.28047717e-01
6.43235385e-01 -2.16744915e-01 -8.21419895e-01 -2.16096073e-01
-6.50436759e-01 3.47236663e-01 1.26889968e+00 1.21839309e+00
8.10958683e-01 -2.61459738e-01 5.06258011e-01 -6.93126202e-01
5.63000500e-01 -5.75811565e-01 -2.52948374e-01 2.50239372e-01
-5.33942997e-01 -5.75791955e-01 7.06057668e-01 -2.41679832e-01
-8.23055923e-01 -1.48095429e-01 -4.88259941e-01 -4.67629969e-01
-1.57939970e-01 1.11355102e+00 3.69838983e-01 -4.10538673e-01
6.66113138e-01 4.21632797e-01 5.05908489e-01 -5.70177473e-02
-1.80628896e-01 3.36430639e-01 3.41048449e-01 8.68912488e-02
4.41235304e-01 6.17605388e-01 2.78753340e-01 -8.34318817e-01
-7.44504571e-01 -5.96922576e-01 -2.74267882e-01 -1.83036104e-01
9.21719015e-01 -9.10900056e-01 -9.05796945e-01 7.34949768e-01
-9.09167051e-01 -5.35745621e-01 -8.81841630e-02 4.49233919e-01
-2.02052593e-01 3.67953897e-01 -1.09635174e+00 -2.22825538e-02
-9.44454610e-01 -1.20226300e+00 9.98279810e-01 3.53131622e-01
-1.23516925e-01 -1.36747038e+00 3.54494691e-01 4.69569594e-01
5.75555444e-01 2.72944003e-01 1.37685776e+00 -7.57282317e-01
-3.74132276e-01 -5.62837780e-01 -6.75013125e-01 1.21946864e-01
-1.05089441e-01 4.46508192e-02 -7.42476165e-01 -4.06150371e-02
-5.00572145e-01 -3.66690159e-01 1.12354517e+00 6.08592212e-01
1.29369199e+00 8.97581950e-02 -8.88499618e-01 7.67978072e-01
1.55152142e+00 -6.36310433e-04 8.73649180e-01 4.80204344e-01
5.04018486e-01 1.28755793e-01 1.63832307e-01 -1.26294523e-01
4.49030727e-01 1.66364074e-01 5.52757800e-01 -5.42234778e-01
-4.69336599e-01 -1.55395031e-01 1.60264507e-01 5.74140191e-01
2.45751530e-01 -2.55716860e-01 -1.58819401e+00 8.24817598e-01
-1.60408688e+00 -6.21759534e-01 -3.56584668e-01 1.56164098e+00
8.28607321e-01 1.07356653e-01 -3.74871790e-01 -2.61658669e-01
4.75899369e-01 -4.23949137e-02 -6.82766378e-01 -1.04181617e-01
-2.23586753e-01 1.67824209e-01 4.77322072e-01 -1.37886941e-01
-9.46384132e-01 7.29002953e-01 6.90933418e+00 8.52660418e-01
-1.48975873e+00 3.11480463e-01 9.92420971e-01 -3.89391370e-02
1.83791965e-02 -3.49453032e-01 -6.18556380e-01 4.34220076e-01
1.26001620e+00 5.85772432e-02 -1.55326709e-01 4.94487494e-01
4.72044021e-01 -2.00208381e-01 -9.37283397e-01 8.01442385e-01
-1.00881733e-01 -2.03716779e+00 7.66870156e-02 1.82880387e-01
4.06459272e-01 7.34502316e-01 2.85712332e-01 3.94558132e-01
1.55658528e-01 -1.32434106e+00 -2.13569373e-01 7.37662017e-01
1.01769865e+00 -2.46048123e-01 1.48631203e+00 1.89402327e-01
-6.01095378e-01 1.94375459e-02 -4.38619226e-01 2.91937083e-01
-2.34508380e-01 8.09566259e-01 -1.34441626e+00 4.56054717e-01
7.37131476e-01 8.99206758e-01 -9.32745159e-01 1.42411494e+00
-1.43897831e-02 9.57135499e-01 -1.05275020e-01 -3.14771503e-01
4.81637806e-01 4.45511043e-01 2.79220045e-01 1.48919499e+00
2.17022032e-01 -2.33378246e-01 3.18683758e-02 6.63030863e-01
-1.73303381e-01 5.38638607e-02 -2.64361054e-01 -2.83856541e-01
7.87957534e-02 1.66089380e+00 -8.14971387e-01 -1.23468705e-01
-7.57813096e-01 4.82657909e-01 2.89842844e-01 3.18418086e-01
-8.08425367e-01 -1.23312652e-01 5.81464291e-01 -2.35400908e-02
-2.82126796e-02 1.53001472e-01 -3.77766937e-01 -8.82561684e-01
-4.68820781e-01 -8.04955840e-01 5.98697603e-01 -8.29496384e-01
-1.59014118e+00 1.36741832e-01 -7.34685540e-01 -1.03890622e+00
4.03269619e-01 -1.07865155e+00 -8.11041117e-01 8.27799022e-01
-2.04768419e+00 -1.49115276e+00 -4.55095321e-01 4.39293772e-01
2.61917174e-01 -2.65484452e-01 1.10892832e+00 -9.25438665e-03
-7.68501103e-01 5.07048011e-01 2.91833460e-01 4.15882587e-01
6.64137900e-01 -1.20546663e+00 4.13810648e-03 4.83504295e-01
-5.39783537e-01 5.73277950e-01 2.28397802e-01 -6.77163899e-01
-1.53483140e+00 -1.50773501e+00 7.82473862e-01 -2.95101106e-01
1.02806532e+00 1.33250132e-01 -6.08313799e-01 6.90180361e-01
1.28694773e-01 3.66972119e-01 1.33079994e+00 -1.62219226e-01
-1.72030851e-01 4.89270538e-02 -1.38748360e+00 6.29616559e-01
6.48673952e-01 -4.77874726e-01 -3.79822224e-01 6.33718610e-01
6.14256382e-01 -4.72662657e-01 -1.19133627e+00 4.48160976e-01
6.48265958e-01 -6.62548482e-01 9.82283831e-01 -3.99889201e-01
5.26637256e-01 -2.27570847e-01 2.75225222e-01 -1.24324977e+00
-5.97885311e-01 4.45637703e-02 -5.32488599e-02 7.86033273e-01
5.82210720e-01 -7.75616705e-01 1.01938236e+00 1.62157252e-01
-2.69305408e-01 -1.28375661e+00 -1.14724314e+00 -4.42259073e-01
2.01399356e-01 -4.22548831e-01 1.84810027e-01 7.04590917e-01
2.10678384e-01 -2.19578277e-02 2.08560556e-01 9.97808427e-02
2.82712489e-01 -4.00951087e-01 2.38539800e-01 -1.06219900e+00
-7.98530057e-02 -6.90906107e-01 -8.31065476e-01 -2.85236359e-01
3.29869449e-01 -1.50709152e+00 -3.94022524e-01 -1.81068909e+00
5.98405421e-01 -2.13749781e-01 -5.66112220e-01 8.41553450e-01
-1.42986313e-01 6.36503041e-01 -1.16027221e-01 1.99857622e-01
-4.59564656e-01 2.19435290e-01 1.32580411e+00 -6.51285231e-01
7.88963139e-02 -2.50900716e-01 -8.42379689e-01 8.38803768e-01
8.39614272e-01 -1.73964333e-02 -8.32494423e-02 -2.24023953e-01
2.32451379e-01 4.77798842e-02 7.20922410e-01 -1.12996233e+00
5.48751175e-01 4.93092164e-02 1.01404333e+00 -6.50336206e-01
2.58731395e-01 -5.46919525e-01 7.65261706e-03 1.04509950e+00
-2.19641984e-01 -2.47009501e-01 4.67186093e-01 5.88127136e-01
-1.35976061e-01 5.55844568e-02 6.41848505e-01 -4.35746759e-01
-7.97106147e-01 5.98519087e-01 -8.80772531e-01 -3.54453743e-01
9.81330812e-01 -4.96865541e-01 -1.02429581e+00 6.67764097e-02
-7.43421972e-01 5.35221338e-01 2.21732304e-01 1.98269159e-01
9.20492947e-01 -1.03543341e+00 -7.30424345e-01 -2.02228323e-01
3.53568971e-01 -2.12508053e-01 7.26869702e-01 1.52984691e+00
-1.01465034e+00 1.02817333e+00 -3.22576106e-01 -7.69048393e-01
-1.11716747e+00 4.27270293e-01 7.50985265e-01 -9.03191566e-01
-5.05067825e-01 8.75478208e-01 1.01674728e-01 -6.31972194e-01
9.00671929e-02 -4.52897638e-01 -3.42689812e-01 -5.71225323e-02
5.44747114e-01 2.63308108e-01 2.39414215e-01 -1.13444276e-01
-5.37110031e-01 1.90820366e-01 -4.29544210e-01 4.76338387e-01
1.38678074e+00 2.98707962e-01 -3.51838678e-01 1.17083155e-01
1.07199872e+00 -6.55972123e-01 -6.94119334e-01 -6.04748391e-02
-6.85950667e-02 9.98165384e-02 2.78024495e-01 -1.35253334e+00
-1.10264444e+00 7.43292272e-01 8.72282267e-01 -3.24013382e-01
1.15103114e+00 -1.31614596e-01 9.46655989e-01 2.45111525e-01
2.40160629e-01 -6.97276115e-01 1.98887903e-02 5.57445824e-01
5.78524351e-01 -1.03815699e+00 -2.04960525e-01 -2.17586711e-01
-1.92609429e-01 1.27944684e+00 8.34035158e-01 6.01246208e-02
6.74435258e-01 2.63423055e-01 1.27060995e-01 -4.92740512e-01
-1.00220549e+00 -1.44093826e-01 1.75227806e-01 9.19587076e-01
6.25907302e-01 5.39571568e-02 -3.42898875e-01 7.65784979e-01
-9.39687341e-02 5.29268920e-01 5.56266725e-01 5.49463689e-01
-1.91608533e-01 -7.34776735e-01 -1.04930758e-01 1.01709390e+00
-7.01680303e-01 -3.57675970e-01 -4.70250636e-01 9.14194763e-01
2.72445560e-01 5.00744104e-01 -8.21827203e-02 -2.50744730e-01
1.17656305e-01 1.51062384e-01 2.10897595e-01 -5.46539307e-01
-9.92920101e-01 -1.69406652e-01 1.22243695e-01 -4.55531359e-01
-2.04557478e-01 -4.93973970e-01 -1.21646595e+00 -2.87701041e-01
-2.42928982e-01 -4.25974280e-01 3.93957108e-01 8.96306992e-01
4.90700811e-01 7.75974333e-01 -2.09849700e-01 -4.73833621e-01
-1.10942028e-01 -1.03112066e+00 -4.21338975e-01 -2.90638596e-01
2.25303233e-01 -1.96986362e-01 -2.19180179e-03 -1.68626159e-01]
|
[15.139322280883789, -2.9261550903320312]
|
20982686-39ee-46df-bd93-e5dccc522c4e
|
attention-aware-deep-reinforcement-learning
| null | null |
http://openaccess.thecvf.com/content_iccv_2017/html/Rao_Attention-Aware_Deep_Reinforcement_ICCV_2017_paper.html
|
http://openaccess.thecvf.com/content_ICCV_2017/papers/Rao_Attention-Aware_Deep_Reinforcement_ICCV_2017_paper.pdf
|
Attention-Aware Deep Reinforcement Learning for Video Face Recognition
|
In this paper, we propose an attention-aware deep reinforcement learning (ADRL) method for video face recognition, which aims to discard the misleading and confounding frames and find the focuses of attention in face videos for person recognition. We formulate the process of finding the attentions of videos as a Markov decision process and train the attention model through a deep reinforcement learning framework without using extra labels. Unlike existing attention models, our method takes information from both the image space and the feature space as the input to make better use of face information that is discarded in the feature learning process. Besides, our approach is attention-aware, which seeks different attentions of videos for the verification of different pairs of videos. Our approach achieves very competitive video face recognition performance on three widely used video face datasets.
|
['Jie zhou', 'Jiwen Lu', 'Yongming Rao']
|
2017-10-01
| null | null | null |
iccv-2017-10
|
['person-recognition']
|
['computer-vision']
|
[ 4.85219061e-02 -2.53236949e-01 -1.38840646e-01 -4.12649810e-01
-5.89292288e-01 -1.35304317e-01 3.24436843e-01 -7.21118391e-01
-3.25080395e-01 4.80525523e-01 9.63590145e-02 -1.31870145e-02
-1.02260962e-01 -5.05193174e-01 -6.85193241e-01 -7.30723023e-01
1.41751051e-01 1.81808442e-01 -3.48669112e-01 2.83036768e-01
3.78570944e-01 5.98711252e-01 -1.51186073e+00 2.87916064e-01
2.79147416e-01 1.20671570e+00 3.95110697e-02 6.85435116e-01
5.39132506e-02 1.27301502e+00 -5.75904489e-01 -4.82501447e-01
1.84559613e-01 -6.05434954e-01 -9.36235368e-01 5.96167624e-01
6.50053978e-01 -8.20587754e-01 -7.25046873e-01 1.16614461e+00
4.48269755e-01 2.76973426e-01 5.67902982e-01 -1.55198598e+00
-1.08836389e+00 6.38323426e-02 -7.33180642e-01 7.22797215e-01
5.81833780e-01 4.17470485e-01 6.74294114e-01 -1.23863804e+00
3.08604509e-01 1.54707491e+00 3.76614451e-01 8.89055252e-01
-7.89213181e-01 -7.46212721e-01 4.01682585e-01 9.64661539e-01
-1.45818651e+00 -8.55474710e-01 8.88108969e-01 -5.28958440e-01
7.00128675e-01 -1.09853819e-02 6.44100428e-01 1.36504340e+00
1.27024623e-02 9.78571236e-01 6.22767627e-01 -3.90403271e-01
9.69513133e-02 -2.64901847e-01 -7.51989186e-02 7.36061096e-01
-1.78638518e-01 1.67885303e-01 -6.61011755e-01 -5.43870069e-02
8.56413722e-01 3.94568563e-01 -3.43645662e-01 5.30987466e-03
-8.08248103e-01 7.68017709e-01 2.09833801e-01 2.48977855e-01
-7.16444254e-01 1.92027032e-01 9.60846916e-02 2.65844136e-01
3.04208815e-01 -2.04653554e-02 -2.16907352e-01 5.82691506e-02
-8.66243362e-01 -8.16352218e-02 2.89877146e-01 7.73220003e-01
7.03720450e-01 2.72078186e-01 -7.68272340e-01 6.02917850e-01
7.87347734e-01 4.97821957e-01 3.41446400e-01 -1.09763634e+00
1.13047853e-01 4.45931882e-01 4.77777459e-02 -9.92005825e-01
2.03990012e-01 1.59276891e-02 -4.84150857e-01 8.20819736e-02
2.07412720e-01 -2.53556997e-01 -9.75138783e-01 1.55306709e+00
2.80155897e-01 7.23337352e-01 2.58620083e-02 9.96621907e-01
9.49963927e-01 4.67400134e-01 3.59128982e-01 -4.04266596e-01
1.27779472e+00 -1.06892967e+00 -8.51753354e-01 -1.24092706e-01
-1.77541934e-02 -5.06515861e-01 5.26777804e-01 2.94869542e-01
-1.16704810e+00 -8.30243945e-01 -7.35370040e-01 -6.21152706e-02
3.31384912e-02 3.09306145e-01 3.54601413e-01 3.97791147e-01
-1.22606075e+00 4.89239842e-01 -5.07190943e-01 -1.43580630e-01
9.93356824e-01 5.92009366e-01 -4.10280824e-01 -3.66259038e-01
-9.49172795e-01 6.01044118e-01 -1.13847502e-01 3.71635586e-01
-1.58257878e+00 -4.07536268e-01 -8.66516888e-01 3.39528620e-01
6.00370049e-01 -3.20777893e-01 1.11439717e+00 -1.76692414e+00
-1.48169649e+00 6.89718902e-01 -4.00893599e-01 -3.41624841e-02
1.92755252e-01 -3.00995201e-01 -4.26102459e-01 4.88694400e-01
2.19500307e-02 7.59010136e-01 1.63181436e+00 -9.19065237e-01
-4.69920039e-01 -3.84019524e-01 9.65144485e-02 -7.21943080e-02
-2.24464893e-01 4.45629954e-01 -7.30819225e-01 -5.57369769e-01
-3.75221431e-01 -6.31331086e-01 1.39964774e-01 7.36127198e-02
-4.10704501e-03 -6.55260623e-01 1.16996634e+00 -9.36122358e-01
8.59140635e-01 -2.32380033e+00 3.12139124e-01 -3.67802717e-02
1.45478740e-01 4.37577218e-01 -5.60832739e-01 -4.47217263e-02
-1.95556000e-01 6.30636513e-03 9.41698998e-02 -3.59384656e-01
-2.25376830e-01 1.30620077e-01 -2.66488999e-01 5.54040372e-01
7.10894167e-01 9.84399915e-01 -9.61080015e-01 -6.51068807e-01
1.55263588e-01 7.95957446e-01 -6.84358537e-01 6.24565721e-01
4.87543978e-02 5.81252813e-01 -5.14561713e-01 9.14722860e-01
6.87318623e-01 -1.17598325e-01 1.13830820e-01 -2.41684884e-01
2.48733714e-01 -1.63273051e-01 -8.92460287e-01 1.28103876e+00
9.29104984e-02 7.39031732e-01 -7.00322762e-02 -9.54884768e-01
7.32904673e-01 5.15236318e-01 6.22425735e-01 -6.71884954e-01
3.63883018e-01 -3.18642348e-01 6.27069697e-02 -1.00005066e+00
1.47805586e-01 3.39132436e-02 5.94711781e-01 5.27805209e-01
6.01419866e-01 6.34881914e-01 4.01482917e-02 -1.60451382e-01
8.36225271e-01 3.00978541e-01 3.63082476e-02 -1.11725010e-01
1.05517220e+00 -7.22859740e-01 9.06707942e-01 6.35652065e-01
-7.77601242e-01 3.69818002e-01 6.40235364e-01 -7.06811368e-01
-6.49103105e-01 -4.49294090e-01 2.42034480e-01 1.25114965e+00
-4.35918793e-02 -2.46373475e-01 -8.55592549e-01 -1.27836430e+00
1.43340072e-02 2.42046446e-01 -1.03532505e+00 -2.58011580e-01
-4.02101338e-01 -1.68943405e-01 2.53459662e-01 5.80075502e-01
5.42499602e-01 -1.65525568e+00 -3.84574145e-01 -2.57220387e-01
-1.37427419e-01 -8.70915830e-01 -7.37360299e-01 -2.52528489e-01
-4.00890797e-01 -1.47998106e+00 -7.33657360e-01 -9.69594419e-01
8.01605880e-01 3.90250325e-01 8.40328753e-01 5.19870043e-01
-2.14649677e-01 6.84725165e-01 -2.90589213e-01 -1.62392959e-01
2.18482409e-02 -4.24887747e-01 1.07150264e-02 7.87716568e-01
8.08457673e-01 -2.81386524e-02 -6.46357000e-01 2.85302222e-01
-5.47447503e-01 -5.84128976e-01 3.36267918e-01 9.62624371e-01
4.47119713e-01 -1.59066439e-01 7.11725771e-01 -3.15928698e-01
3.80187035e-01 -5.40355563e-01 -5.67514539e-01 4.97592300e-01
-5.39992005e-02 -1.05052032e-02 3.69325072e-01 -3.74236405e-01
-8.91573727e-01 3.67924981e-02 -6.87530860e-02 -1.17629170e+00
-2.12931290e-01 -3.24752368e-02 -3.74466032e-01 -3.21734697e-01
-5.15839048e-02 2.60796666e-01 1.43122375e-01 -2.65638411e-01
-9.25597996e-02 6.21301889e-01 2.85964906e-01 -1.99960828e-01
3.45297933e-01 2.88055658e-01 -3.02657902e-01 -7.06198037e-01
-7.24376678e-01 -2.64400125e-01 -6.61399841e-01 -5.91844797e-01
1.03909862e+00 -7.89077163e-01 -1.13568389e+00 4.77683693e-01
-1.15202177e+00 -5.92238158e-02 -9.14076641e-02 4.62504417e-01
-4.81219321e-01 4.38777000e-01 -4.91765767e-01 -1.18156636e+00
-2.29297400e-01 -1.53972805e+00 1.15267825e+00 4.18694794e-01
8.65792409e-02 -7.39418805e-01 -1.57484084e-01 3.19456279e-01
1.51989341e-01 -3.03474635e-01 4.61110711e-01 -5.58354318e-01
-8.12641382e-01 1.06802277e-01 -3.46800178e-01 3.78068477e-01
2.35577479e-01 2.77959585e-01 -1.32488477e+00 -5.20299137e-01
1.95961177e-01 -5.53150535e-01 1.01383793e+00 5.06045282e-01
1.53743494e+00 -3.56166214e-01 -1.53343439e-01 6.53474748e-01
1.03723454e+00 6.80197239e-01 9.94643033e-01 -3.21350656e-02
5.87130666e-01 6.23623550e-01 3.80804479e-01 3.67606431e-01
1.16265960e-01 5.33300996e-01 6.19186342e-01 -2.75476314e-02
-5.95144928e-02 -1.21192694e-01 5.75943887e-01 2.27186620e-01
-1.41313747e-01 -2.73310333e-01 -4.98318553e-01 5.49618006e-01
-1.73627925e+00 -1.33259928e+00 5.74464440e-01 2.05801940e+00
2.20168561e-01 -4.35408056e-01 1.26551047e-01 -4.42762673e-03
9.23859239e-01 1.33268565e-01 -7.14091897e-01 -2.88873494e-01
1.28565446e-01 1.42748863e-01 -7.18151033e-02 4.13273543e-01
-1.20785272e+00 8.80827487e-01 6.97374487e+00 3.62078965e-01
-1.14034677e+00 2.62361057e-02 8.94751608e-01 -2.06478193e-01
2.21751198e-01 -2.22977340e-01 -7.79784799e-01 6.25314951e-01
7.78422177e-01 1.50327012e-01 5.30575454e-01 8.11437011e-01
2.28770897e-02 5.75663634e-02 -1.31206536e+00 1.43126190e+00
5.51530480e-01 -1.03492665e+00 3.85621458e-01 2.93819550e-02
4.27634269e-01 -2.75858998e-01 2.06943750e-01 3.35207462e-01
1.46486372e-01 -1.25717902e+00 6.01214528e-01 9.32439089e-01
5.64211667e-01 -8.11165452e-01 7.85576046e-01 -1.29810363e-01
-8.34757388e-01 -4.34896439e-01 -4.46361929e-01 3.57680162e-03
-2.42491379e-01 -1.10368170e-01 -5.82726896e-01 2.78011233e-01
8.27668965e-01 1.13132906e+00 -5.64330697e-01 9.05085802e-01
-3.04646641e-01 5.87724447e-01 3.69434029e-01 2.00562224e-01
9.67727825e-02 -7.21870959e-02 2.81364322e-01 8.40861142e-01
2.17436448e-01 2.16359198e-01 2.66447634e-01 8.10052454e-01
-4.40461278e-01 -1.73029646e-01 -5.34177065e-01 -8.32758322e-02
2.07791492e-01 1.17311585e+00 -5.02020121e-01 -3.78044814e-01
-6.04375482e-01 1.00900757e+00 4.68142301e-01 5.00267982e-01
-8.47849429e-01 -1.05766609e-01 8.02588999e-01 -2.17898652e-01
8.50338876e-01 2.01130807e-01 5.29781759e-01 -1.09721005e+00
-1.44576520e-01 -9.87477541e-01 5.88941872e-01 -8.58011246e-01
-1.24161959e+00 7.97092497e-01 -3.29115808e-01 -1.02152550e+00
-4.76772189e-02 -6.65966868e-01 -7.11700439e-01 8.30102265e-01
-1.76361895e+00 -7.92194784e-01 -2.84006238e-01 9.46073711e-01
8.92894208e-01 -5.65569997e-01 3.96226555e-01 4.18513149e-01
-9.20235932e-01 6.89789951e-01 -3.57168764e-01 5.92182577e-01
6.51243627e-01 -6.84990823e-01 1.45827696e-01 9.10401762e-01
1.54772848e-01 3.54981124e-01 1.11908875e-01 -5.55930555e-01
-1.46871674e+00 -1.05973482e+00 8.74094248e-01 -3.57958764e-01
1.90156460e-01 3.65170501e-02 -9.48504686e-01 7.78795838e-01
5.84318221e-01 2.67230958e-01 7.79104769e-01 -9.31014940e-02
-2.11186424e-01 -1.80879876e-01 -1.13604903e+00 2.59553403e-01
8.84386003e-01 -7.29651451e-01 -5.83072960e-01 2.60849357e-01
2.73447216e-01 4.68007587e-02 -5.39308727e-01 2.39515066e-01
5.77615976e-01 -7.50539958e-01 9.73678768e-01 -1.08150959e+00
4.45601463e-01 -2.35504776e-01 -1.72021717e-01 -9.89512742e-01
-6.89544022e-01 -5.48849046e-01 -2.69587010e-01 1.23284483e+00
-8.77556652e-02 -2.11485177e-01 5.88378489e-01 4.54186320e-01
1.47361532e-01 -5.28400242e-01 -9.65665996e-01 -3.27679068e-01
-4.15260702e-01 -1.32582754e-01 7.00977862e-01 8.12095165e-01
-2.47671634e-01 2.35355034e-01 -7.24130034e-01 1.91996723e-01
5.99213362e-01 8.36964771e-02 4.26144898e-01 -1.04650307e+00
-1.66744232e-01 -3.43426198e-01 -5.70562720e-01 -8.09880197e-01
5.21427631e-01 -6.61024272e-01 1.18856467e-01 -1.11570680e+00
5.30979931e-01 2.24703580e-01 -6.68439567e-01 7.51397789e-01
-2.18648911e-01 4.33125645e-01 3.27793449e-01 1.61044955e-01
-1.06557333e+00 8.17696869e-01 1.17809212e+00 -4.30399537e-01
1.36545643e-01 -1.47822797e-01 -7.06560969e-01 6.75564647e-01
4.70012337e-01 -3.77896816e-01 -1.41236842e-01 -6.89114153e-01
-3.84369999e-01 2.45566145e-01 6.42363906e-01 -8.67543101e-01
3.20015550e-01 -1.70377240e-01 1.12221253e+00 -4.77192551e-01
2.99208552e-01 -8.50273609e-01 -1.84206307e-01 4.42892104e-01
-4.17723149e-01 1.64671361e-01 1.75641045e-01 5.18258929e-01
-2.12691173e-01 -2.35162735e-01 7.22491324e-01 -1.09962270e-01
-8.47736716e-01 7.67781198e-01 -4.99691963e-01 -1.51463807e-01
1.15160811e+00 -1.25199661e-01 -1.00643031e-01 -3.61515015e-01
-9.78952348e-01 3.60937059e-01 1.94365643e-02 6.45974100e-01
9.95987713e-01 -1.54739976e+00 -7.99070060e-01 7.65082955e-01
-1.73317850e-01 -5.40273190e-01 5.36863089e-01 6.37259960e-01
-5.61623983e-02 4.37604010e-01 -5.78869164e-01 -5.75761795e-01
-1.50063670e+00 1.02925038e+00 5.87368548e-01 2.03154966e-01
-2.52704680e-01 9.15755630e-01 3.07369590e-01 2.44352177e-01
4.95314091e-01 5.41076548e-02 -6.35289252e-01 1.14820421e-01
1.09903133e+00 2.74249256e-01 -1.51734337e-01 -9.84749138e-01
-6.43993139e-01 5.26531756e-01 -3.08012038e-01 3.13877940e-01
1.22972107e+00 -1.06666423e-01 -5.59461378e-02 -8.03437382e-02
1.18674982e+00 -3.45733762e-01 -1.89698887e+00 -1.26482308e-01
-1.55114964e-01 -8.11006784e-01 2.23974839e-01 -5.61581552e-01
-1.70496476e+00 8.99094760e-01 6.61686063e-01 -1.54661149e-01
1.20992661e+00 8.48941430e-02 3.35236818e-01 1.42745972e-01
1.81725360e-02 -9.66960251e-01 4.14715886e-01 3.70808631e-01
1.08539128e+00 -1.47015631e+00 -6.76218346e-02 -1.72530692e-02
-5.62563181e-01 1.16925848e+00 9.69412804e-01 -1.81249484e-01
6.56376243e-01 -9.83841121e-02 -1.45973429e-01 -2.35539258e-01
-8.34586799e-01 -3.10002714e-01 3.68778795e-01 7.82852471e-01
1.41755834e-01 -3.58719856e-01 1.55207038e-01 5.79539895e-01
5.75715721e-01 3.09560359e-01 1.76571742e-01 8.06896031e-01
-3.60062212e-01 -8.90556633e-01 -3.92006129e-01 2.38754779e-01
-5.67517757e-01 2.73892246e-02 -5.49070656e-01 3.77335042e-01
2.72649020e-01 9.96999025e-01 2.99579084e-01 -3.82552713e-01
4.75313589e-02 3.26911032e-01 6.17792547e-01 -3.09847534e-01
-4.53382164e-01 -7.29562491e-02 -5.24487615e-01 -6.95655346e-01
-7.21473336e-01 -8.08502913e-01 -6.57931209e-01 -2.07572475e-01
-2.20275760e-01 1.68434381e-01 1.42734110e-01 1.12844181e+00
6.11644328e-01 5.77005982e-01 9.01059210e-01 -9.11434710e-01
-2.80763984e-01 -8.13818038e-01 -3.72023046e-01 3.90078545e-01
7.13753343e-01 -1.02420568e+00 -1.48258775e-01 2.44179830e-01]
|
[13.395435333251953, 1.2484471797943115]
|
5adf713f-71a5-4fe4-a9ac-c51d1a7cb1f1
|
improving-extreme-weather-events-detection
|
2304.00176
| null |
https://arxiv.org/abs/2304.00176v1
|
https://arxiv.org/pdf/2304.00176v1.pdf
|
Improving extreme weather events detection with light-weight neural networks
|
To advance automated detection of extreme weather events, which are increasing in frequency and intensity with climate change, we explore modifications to a novel light-weight Context Guided convolutional neural network architecture trained for semantic segmentation of tropical cyclones and atmospheric rivers in climate data. Our primary focus is on tropical cyclones, the most destructive weather events, for which current models show limited performance. We investigate feature engineering, data augmentation, learning rate modifications, alternative loss functions, and architectural changes. In contrast to previous approaches optimizing for intersection over union, we specifically seek to improve recall to penalize under-counting and prioritize identification of tropical cyclones. We report success through the use of weighted loss functions to counter class imbalance for these rare events. We conclude with directions for future research on extreme weather events detection, a crucial task for prediction, mitigation, and equitable adaptation to the impacts of climate change.
|
['David Lüdeke', 'Lucas Hendren', 'Hannah Grossman', 'Romain Lacombe']
|
2023-03-31
| null | null | null | null |
['feature-engineering']
|
['methodology']
|
[ 2.21266076e-01 -3.50389302e-01 -3.16938758e-02 -8.42774808e-01
-3.72095525e-01 -7.45530784e-01 6.57960236e-01 5.55698931e-01
-7.71530151e-01 6.78890467e-01 4.64534551e-01 -8.72550189e-01
-2.57057697e-02 -1.04473162e+00 -3.58334273e-01 -3.13341886e-01
-3.73481482e-01 3.01023155e-01 -2.40920931e-01 -2.92768002e-01
4.12665904e-01 9.24996734e-01 -1.29211032e+00 1.90694444e-02
9.67129111e-01 3.82300228e-01 -2.12379366e-01 9.41195130e-01
-1.80741817e-01 2.89592266e-01 -5.43904483e-01 -3.51962857e-02
5.02353668e-01 -2.51942307e-01 -5.85974455e-01 -3.53101403e-01
8.58729839e-01 -3.88689995e-01 -1.07616015e-01 7.44263232e-01
7.20732629e-01 4.69395757e-01 6.06622100e-01 -1.02316058e+00
-2.25354448e-01 3.86580557e-01 -4.90154386e-01 1.19264209e+00
-2.99362868e-01 3.77865791e-01 9.76564467e-01 -7.54625499e-01
2.56489754e-01 1.15995705e+00 9.22229111e-01 2.37043023e-01
-9.51257169e-01 -8.73289287e-01 5.79638302e-01 -1.24627121e-01
-1.00592542e+00 -4.63534415e-01 1.67843953e-01 -5.04359186e-01
1.39865792e+00 4.60754991e-01 5.31164050e-01 6.23883784e-01
-2.92753577e-02 1.68844357e-01 8.42108548e-01 -1.19002908e-01
9.98287499e-02 -1.87441275e-01 2.74019212e-01 4.11589712e-01
4.92010474e-01 3.59149814e-01 -2.22303033e-01 -1.54175133e-01
2.71451682e-01 1.68404117e-01 -3.32453288e-02 3.25744271e-01
-8.20128560e-01 9.54413176e-01 6.36711061e-01 -2.91827042e-02
-2.69690454e-01 2.18263030e-01 3.74791592e-01 1.37381226e-01
1.24194908e+00 9.96495783e-01 -9.16570067e-01 -4.76427414e-02
-1.28437984e+00 5.92229784e-01 6.09816611e-01 3.04826528e-01
6.26876235e-01 2.78432190e-01 -1.65297449e-01 6.89664721e-01
7.37727880e-02 8.73165190e-01 -3.00207227e-01 -7.90255189e-01
6.06927276e-01 4.78843331e-01 1.72446534e-01 -9.89226341e-01
-1.06434441e+00 -7.66648531e-01 -4.58806336e-01 2.23950177e-01
4.19014364e-01 -8.51983547e-01 -1.20886827e+00 1.56749129e+00
3.73457581e-01 3.86464745e-01 -1.50869548e-01 1.02257359e+00
3.77243042e-01 7.90399432e-01 7.32365727e-01 4.10879180e-02
1.28161275e+00 -3.18758965e-01 -3.82169664e-01 -5.54028988e-01
5.66737890e-01 -7.11143911e-01 9.04784977e-01 -1.27791807e-01
-4.40318882e-01 -1.40176788e-01 -6.36853874e-01 1.22330785e-01
-9.55927789e-01 -3.64540666e-02 8.31469059e-01 6.25622928e-01
-6.54018819e-01 5.87898552e-01 -7.91953683e-01 -4.49780017e-01
5.13467848e-01 7.24874213e-02 1.53205320e-01 2.84807473e-01
-1.43571150e+00 1.19617486e+00 1.43957555e-01 3.55195165e-01
-6.29289925e-01 -1.22452438e+00 -9.69606876e-01 3.63532811e-01
1.56656206e-01 -3.08453500e-01 8.69631410e-01 -7.46355236e-01
-6.68596745e-01 7.50845492e-01 3.68689597e-02 -7.04365492e-01
2.10646719e-01 -6.03342533e-01 -6.12020850e-01 -1.84549958e-01
5.06509654e-02 7.40967333e-01 3.44080985e-01 -7.98145056e-01
-1.19350576e+00 -1.38791949e-01 -4.97641414e-02 3.34189802e-01
-4.54635531e-01 3.95971954e-01 1.96522966e-01 -7.65037715e-01
-2.51640141e-01 -7.30993211e-01 -8.12360287e-01 -7.35857934e-02
-1.47898883e-01 -1.76975891e-01 7.12468803e-01 -7.79740214e-01
1.24200642e+00 -1.89134574e+00 -5.07690966e-01 4.00173992e-01
5.45834526e-02 4.47066665e-01 -1.20779373e-01 9.70476791e-02
-5.79254963e-02 5.17418444e-01 -6.18416905e-01 -8.14996213e-02
-1.29008204e-01 1.97205707e-01 -9.45714056e-01 5.92628658e-01
7.10729182e-01 4.08081084e-01 -1.06974626e+00 -2.28020743e-01
4.72191900e-01 3.67518306e-01 -5.15724897e-01 1.50518447e-01
-3.21108907e-01 3.43395621e-01 -2.64070630e-01 6.78771377e-01
7.47859597e-01 1.94326341e-01 -1.27577811e-01 3.09634376e-02
-6.68947399e-01 6.51649535e-01 -8.85527849e-01 8.49688590e-01
-5.87324202e-01 8.18904757e-01 -5.54170497e-02 -7.96542883e-01
8.54023278e-01 -5.44758067e-02 2.73085117e-01 -7.30481088e-01
-1.64212137e-02 4.57291380e-02 -8.29390213e-02 -5.49563229e-01
8.88781846e-01 -2.40391605e-02 -2.30161920e-01 4.93475139e-01
-4.70612615e-01 -2.25670204e-01 2.69009143e-01 1.04061201e-01
5.38638532e-01 4.34749052e-02 1.78912565e-01 -6.16495192e-01
-2.13921964e-01 2.73814976e-01 8.66959572e-01 8.11787784e-01
-4.07071322e-01 6.87908590e-01 4.68501270e-01 -1.01619852e+00
-1.14302659e+00 -8.93407226e-01 -3.92777264e-01 1.49157870e+00
-2.13646799e-01 2.01878306e-02 -3.34542185e-01 -7.04460859e-01
2.26623237e-01 1.10561717e+00 -5.74748158e-01 1.21397143e-02
-9.38061893e-01 -1.87288058e+00 7.26694286e-01 5.42154849e-01
3.83981586e-01 -8.62536371e-01 -1.01251233e+00 2.91904867e-01
-7.96766356e-02 -8.71065319e-01 -3.61341834e-01 5.88870704e-01
-7.89603829e-01 -1.20511615e+00 -6.65738225e-01 -2.37200946e-01
5.18403113e-01 -4.11698744e-02 1.39225054e+00 -1.85432926e-01
-4.55787390e-01 2.97098368e-01 -9.60490182e-02 -8.07606399e-01
4.97970581e-02 4.04980958e-01 -4.99523580e-02 -4.13824618e-01
5.41809320e-01 -3.60719562e-01 -1.04175222e+00 -2.76775248e-02
-6.93677723e-01 -2.97666281e-01 1.40041515e-01 3.84557486e-01
2.64207125e-01 -2.14048177e-01 5.11738122e-01 -9.33563411e-01
4.72966284e-01 -9.69227850e-01 -6.82485342e-01 6.42451793e-02
-7.06297219e-01 -9.64607894e-02 4.18440491e-01 -5.08891456e-02
-1.18495989e+00 7.43275955e-02 -1.59321368e-01 -8.80974531e-02
-5.35359442e-01 4.65667158e-01 6.23426557e-01 2.36687481e-01
8.30764949e-01 -2.74052054e-01 -4.49602962e-01 -4.42482769e-01
3.63991678e-01 4.90893602e-01 4.38245267e-01 -4.26520884e-01
5.86483300e-01 5.31685472e-01 -1.31057560e-01 -9.14665639e-01
-1.10058475e+00 -5.37938058e-01 -2.41940275e-01 3.67364846e-03
8.87336671e-01 -1.16851854e+00 -1.64655149e-01 4.28496212e-01
-8.62047672e-01 -4.98030335e-01 -7.31140599e-02 5.96151054e-01
2.98329890e-01 -5.02258465e-02 -3.63620967e-01 -8.77034128e-01
-6.44650519e-01 -5.90203285e-01 7.49993145e-01 6.18040979e-01
-2.43398339e-01 -1.09141433e+00 5.09573698e-01 -5.52112050e-02
8.67575288e-01 6.41712487e-01 8.72983932e-01 -8.08058977e-01
-1.56015381e-01 1.12520196e-01 -3.87845665e-01 1.78744927e-01
2.39277095e-01 4.49759662e-01 -9.33552086e-01 -1.11577347e-01
-6.28580391e-01 -6.32538050e-02 1.38062692e+00 4.67083663e-01
1.01755118e+00 -1.16676986e-01 -2.80957848e-01 1.00941658e+00
1.13296890e+00 1.31333604e-01 2.22721145e-01 4.67488527e-01
5.30889273e-01 6.88322842e-01 4.34228271e-01 5.10763884e-01
4.77998972e-01 1.72581226e-01 5.16557574e-01 -6.68844163e-01
3.00526004e-02 3.61232251e-01 1.46663962e-02 3.67398374e-02
1.26429468e-01 -3.14477056e-01 -1.47048318e+00 1.15080917e+00
-1.23106647e+00 -8.04718256e-01 -1.73720196e-01 2.08607697e+00
6.72053576e-01 2.95748800e-01 1.81363672e-02 -4.43984717e-01
7.12455928e-01 7.35321760e-01 -5.94997168e-01 -8.09871018e-01
-2.45191380e-01 4.80509341e-01 1.09546053e+00 5.38762450e-01
-1.66596901e+00 1.15055907e+00 7.02797937e+00 1.95585772e-01
-1.46196818e+00 -2.31584951e-01 1.24109161e+00 -3.10984164e-01
-3.86986136e-01 -1.56505555e-02 -1.19073927e+00 2.97222495e-01
1.17885721e+00 1.19740553e-01 1.97132751e-01 5.49548388e-01
6.39232159e-01 -2.07835753e-02 -4.42679763e-01 9.34602767e-02
-2.42668748e-01 -1.36262369e+00 9.42009240e-02 -4.02885348e-01
7.62610197e-01 6.97903812e-01 1.74859643e-01 2.84235567e-01
7.69750357e-01 -1.01967418e+00 2.39735425e-01 4.53043222e-01
6.42438054e-01 -1.08031440e+00 5.62274396e-01 -3.88550073e-01
-1.14608586e+00 -2.17734903e-01 -2.19780564e-01 -2.21789286e-01
1.40896484e-01 1.03890336e+00 -7.96198726e-01 -5.76246865e-02
8.64291787e-01 5.25084615e-01 -3.37439835e-01 1.04053700e+00
-1.74476534e-01 1.19982314e+00 -8.76346529e-01 1.67446807e-01
7.02962577e-01 8.90833512e-02 6.75371766e-01 1.74485850e+00
2.53608108e-01 2.14415729e-01 2.69985080e-01 6.79224491e-01
-8.68738629e-03 6.22917116e-02 -6.36666119e-01 -3.26778367e-02
5.47111511e-01 1.24629974e+00 -1.07731771e+00 -3.21244389e-01
-2.04856962e-01 3.69251817e-01 1.02270178e-01 5.04162848e-01
-8.51437330e-01 -6.24406099e-01 1.21345758e+00 1.41836016e-03
2.13524215e-02 -2.15788633e-01 -6.97084844e-01 -1.04136860e+00
-3.17119658e-01 -5.89262009e-01 7.20000207e-01 -2.57722229e-01
-1.01998198e+00 3.39801520e-01 2.07496760e-03 -8.43169928e-01
-8.38283449e-02 -3.27218741e-01 -1.31582296e+00 1.07426453e+00
-2.21612167e+00 -8.61001670e-01 -1.07621752e-01 7.30016455e-02
5.75456619e-01 1.44539565e-01 6.77561879e-01 3.42336774e-01
-6.17659688e-01 3.75261933e-01 1.57393396e-01 1.77631930e-01
7.68252254e-01 -1.41587555e+00 1.08841097e+00 1.09459031e+00
-2.86570847e-01 5.67219675e-01 7.82514572e-01 -8.04828286e-01
-6.61970019e-01 -1.82736039e+00 1.08586049e+00 -1.26299068e-01
4.39762920e-01 -3.46295059e-01 -8.72192204e-01 6.08026564e-01
-2.12536007e-01 -6.49092421e-02 7.74464488e-01 3.06656986e-01
-4.38825488e-01 -1.93850398e-01 -1.29979205e+00 6.16565228e-01
6.17958903e-01 -3.65259588e-01 -4.18377727e-01 5.15614390e-01
7.01313734e-01 -3.58355761e-01 -6.40258312e-01 6.33084357e-01
3.76295120e-01 -5.30167699e-01 8.56794059e-01 -9.24290836e-01
4.92866397e-01 -2.39241868e-01 -3.09540667e-02 -1.62582254e+00
-2.02706888e-01 -2.75786102e-01 3.74370009e-01 1.01527417e+00
6.45330429e-01 -6.46082699e-01 7.79251099e-01 6.15858257e-01
-3.70452106e-01 -3.86213332e-01 -9.06033278e-01 -3.39831769e-01
4.79692191e-01 -3.27871978e-01 6.66967690e-01 1.34934247e+00
-5.81966639e-01 -2.41271719e-01 -4.45295006e-01 6.16251767e-01
5.25388896e-01 3.96907508e-01 3.30375552e-01 -1.16414404e+00
1.99304253e-01 -5.50100505e-01 2.98116773e-01 -3.34102511e-01
-1.09076962e-01 -6.49666667e-01 1.85662702e-01 -1.38039005e+00
-1.95111707e-01 -5.41227818e-01 -4.80458289e-01 7.51978695e-01
-6.87265217e-01 4.32163209e-01 -2.87403259e-02 -2.79347628e-01
-1.79312602e-01 4.05197054e-01 5.67980886e-01 -8.10798034e-02
-4.89034086e-01 6.35745898e-02 -7.27964878e-01 7.77732909e-01
1.11020792e+00 -7.46316969e-01 -6.72710910e-02 -7.07845688e-01
5.39514959e-01 -3.50842237e-01 2.23333210e-01 -9.46955502e-01
-6.72590956e-02 -7.94989765e-01 6.39500499e-01 -6.58922911e-01
-2.69866258e-01 -4.70229596e-01 -2.94652373e-01 5.94795406e-01
-4.75307673e-01 3.42628568e-01 6.73297822e-01 3.73294145e-01
9.05548707e-02 3.04726988e-01 8.83305311e-01 -2.63662159e-01
-8.62014532e-01 3.48614901e-01 -9.29163277e-01 3.11115384e-01
8.10959637e-01 3.14190716e-01 -5.74320436e-01 -3.86994444e-02
-6.22678339e-01 9.19151068e-01 1.44890510e-02 5.56914747e-01
3.32433581e-01 -5.74419320e-01 -1.24775434e+00 6.20679110e-02
-3.91849391e-02 -5.55027910e-02 1.82164252e-01 3.74601543e-01
-9.60479200e-01 1.98087975e-01 -1.08827785e-01 -1.42726079e-01
-9.23750460e-01 2.01362018e-02 6.92121983e-01 -2.95066178e-01
-4.88992423e-01 7.94578016e-01 4.85316180e-02 -8.67334068e-01
1.84895590e-01 -4.96472359e-01 -4.90881681e-01 3.90078843e-01
5.49556613e-01 6.13262415e-01 5.72772250e-02 -2.78601259e-01
-5.39002001e-01 1.52544841e-01 -6.38678670e-02 2.27072164e-01
1.60778117e+00 -2.18404755e-01 1.58213019e-01 1.36779964e-01
8.20996642e-01 -1.88546509e-01 -1.44593120e+00 2.58018803e-02
1.63090438e-01 -1.41025946e-01 3.76163960e-01 -1.08670306e+00
-1.33436382e+00 9.42880034e-01 1.03145421e+00 -4.60934006e-02
1.00197458e+00 -5.13768494e-01 8.63156676e-01 5.09429336e-01
-5.16933084e-01 -1.12492144e+00 -6.19132936e-01 8.66926551e-01
4.93963391e-01 -1.24696100e+00 3.28667939e-01 2.95816392e-01
-2.44923964e-01 1.14567614e+00 7.03327239e-01 -3.03296804e-01
8.72455716e-01 4.65431184e-01 2.95693308e-01 -3.50497335e-01
-5.77732027e-01 -4.20922786e-01 5.78095391e-03 4.20012712e-01
4.52278733e-01 2.76913553e-01 -1.84693694e-01 -1.00293450e-01
-4.54917215e-02 -3.27588409e-01 4.16813105e-01 7.97304332e-01
-7.63289690e-01 -3.39950293e-01 -4.13358510e-01 9.27278757e-01
-9.12459493e-01 -8.07770908e-01 -2.25310653e-01 5.03658473e-01
3.74503314e-01 8.32584143e-01 8.61851752e-01 4.26877086e-04
3.50606769e-01 1.92670882e-01 -4.62616146e-01 -5.31545877e-01
-1.11532307e+00 -3.40498239e-01 3.66284996e-01 -3.06202084e-01
-2.12748438e-01 -6.83084667e-01 -1.13227069e+00 -3.70599180e-01
-8.47777128e-02 1.65853351e-01 9.20274913e-01 9.56556141e-01
4.55447346e-01 6.18717432e-01 6.73419297e-01 -7.09198356e-01
-3.67292941e-01 -8.52670074e-01 -3.06205988e-01 2.78577447e-01
7.08427191e-01 -3.47348899e-01 -6.71755612e-01 -2.41928831e-01]
|
[6.719411373138428, 2.9784047603607178]
|
10f7d8ae-4b47-40cc-8327-669e2c1369d0
|
can-a-frozen-pretrained-language-model-be
|
2303.05153
| null |
https://arxiv.org/abs/2303.05153v1
|
https://arxiv.org/pdf/2303.05153v1.pdf
|
Can a Frozen Pretrained Language Model be used for Zero-shot Neural Retrieval on Entity-centric Questions?
|
Neural document retrievers, including dense passage retrieval (DPR), have outperformed classical lexical-matching retrievers, such as BM25, when fine-tuned and tested on specific question-answering datasets. However, it has been shown that the existing dense retrievers do not generalize well not only out of domain but even in domain such as Wikipedia, especially when a named entity in a question is a dominant clue for retrieval. In this paper, we propose an approach toward in-domain generalization using the embeddings generated by the frozen language model trained with the entities in the domain. By not fine-tuning, we explore the possibility that the rich knowledge contained in a pretrained language model can be used for retrieval tasks. The proposed method outperforms conventional DPRs on entity-centric questions in Wikipedia domain and achieves almost comparable performance to BM25 and state-of-the-art SPAR model. We also show that the contextualized keys lead to strong improvements compared to BM25 when the entity names consist of common words. Our results demonstrate the feasibility of the zero-shot retrieval method for entity-centric questions of Wikipedia domain, where DPR has struggled to perform.
|
['Jun Deguchi', 'Osamu Torii', 'Youyang Ng', 'Yasuhiro Morioka', 'Daisuke Miyashita', 'Yasuto Hoshi']
|
2023-03-09
| null | null | null | null |
['passage-retrieval']
|
['natural-language-processing']
|
[-4.59112793e-01 -6.82379827e-02 -2.23156229e-01 1.96804553e-01
-1.29506886e+00 -6.54537320e-01 7.93225586e-01 6.24632061e-01
-9.85559821e-01 8.64044547e-01 6.23374104e-01 -6.58220872e-02
-6.30169034e-01 -1.17431366e+00 -8.33664894e-01 -1.00133494e-01
1.17059136e-02 8.34350288e-01 7.18481362e-01 -9.23219204e-01
3.23719263e-01 2.22809508e-01 -1.43685174e+00 3.89467537e-01
1.09157491e+00 7.11167037e-01 3.90317529e-01 3.28597665e-01
-5.54422081e-01 6.25968933e-01 -6.86315000e-01 -4.68098879e-01
2.99657937e-02 1.34990225e-02 -1.20796478e+00 -7.91270792e-01
6.71571732e-01 -4.62651700e-01 -9.12257671e-01 7.34730244e-01
7.21077561e-01 8.39182794e-01 9.36856449e-01 -4.38320965e-01
-1.47484946e+00 5.45327902e-01 -7.14929327e-02 4.12066311e-01
6.69827521e-01 -5.76999724e-01 1.43043971e+00 -1.15932786e+00
1.14892006e+00 1.21911955e+00 4.90287751e-01 7.10141718e-01
-9.37587142e-01 -3.39448571e-01 -1.51768416e-01 5.05600274e-01
-1.76315105e+00 -4.27764580e-02 3.25250119e-01 -1.35624692e-01
1.55826271e+00 3.31452638e-02 -3.32300626e-02 1.03717017e+00
-1.14514180e-01 9.35663223e-01 4.10781711e-01 -7.12893426e-01
1.91998452e-01 2.96067238e-01 6.03118360e-01 2.18625620e-01
4.74807173e-01 -3.14401686e-01 -3.78992647e-01 -4.47961748e-01
2.26110667e-01 1.00539111e-01 -5.52795410e-01 -2.15455517e-01
-9.42042410e-01 9.50163007e-01 5.69265187e-01 6.88408017e-01
-5.70375860e-01 -1.33428201e-01 4.61573511e-01 5.75017214e-01
4.46660787e-01 1.16080403e+00 -5.01555681e-01 2.62392730e-01
-1.17517674e+00 6.30848229e-01 1.07805705e+00 1.19278193e+00
9.93240893e-01 -6.29626334e-01 -6.95759475e-01 1.28884375e+00
7.42967874e-02 4.36271101e-01 8.87891114e-01 -6.29291594e-01
5.57473361e-01 6.26724780e-01 4.00521159e-01 -8.11444879e-01
-1.82845905e-01 -4.20195222e-01 -4.08642590e-01 -7.22019136e-01
1.07050285e-01 2.11599972e-02 -9.41738904e-01 1.65764022e+00
1.48901520e-02 -3.73876765e-02 4.32725996e-01 7.69312501e-01
1.34050632e+00 9.24437940e-01 2.14657307e-01 4.37177978e-02
1.31986547e+00 -1.09814525e+00 -7.05057502e-01 -8.20369273e-02
7.77108312e-01 -6.37516379e-01 1.02176452e+00 -2.48030752e-01
-8.70237470e-01 -4.51878339e-01 -8.76942873e-01 -5.96803963e-01
-1.01699638e+00 -6.59302101e-02 2.28063941e-01 3.30897599e-01
-1.14591920e+00 6.92774296e-01 -1.55338988e-01 -1.06335104e+00
2.35268450e-03 5.08989021e-03 -4.88044381e-01 -6.92467928e-01
-1.99245584e+00 1.28793228e+00 6.37270808e-01 -3.77644181e-01
-9.75320995e-01 -8.70953202e-01 -5.71965814e-01 4.55914080e-01
3.50010008e-01 -9.44155395e-01 1.07420492e+00 -1.29497916e-01
-9.64581847e-01 8.91197681e-01 -1.17461242e-01 -6.79342866e-01
-1.17806889e-01 -5.43343723e-01 -5.69468021e-01 5.26081979e-01
2.37557769e-01 6.76031411e-01 4.37664777e-01 -9.09645915e-01
-3.95898193e-01 -1.50054991e-01 6.20884418e-01 3.23851585e-01
-6.28671825e-01 -1.08507298e-01 -6.02592945e-01 -4.64040309e-01
-3.72876406e-01 -6.10745609e-01 -6.13006167e-02 -4.06885743e-01
-1.93631705e-02 -9.13582563e-01 4.12454367e-01 -7.26662278e-01
1.70567954e+00 -1.83798873e+00 1.41250283e-01 -1.33403286e-01
-2.26829965e-02 7.42403805e-01 -5.00782788e-01 1.22845054e+00
3.85628104e-01 2.04500824e-01 1.76400900e-01 3.89828123e-02
2.66527444e-01 1.99058101e-01 -7.70026267e-01 -2.02169339e-03
1.14946654e-02 1.10797107e+00 -1.10053051e+00 -6.30418420e-01
-2.81271994e-01 2.71588117e-01 -6.04200244e-01 3.04799557e-01
-4.26364213e-01 -3.55407596e-01 -8.01230073e-01 5.33545136e-01
5.00606239e-01 -2.87381172e-01 -2.19828337e-01 4.80577573e-02
3.15440208e-01 6.13435507e-01 -7.01691329e-01 2.09787130e+00
-6.12593710e-01 4.80419725e-01 -3.46823931e-01 -7.75088251e-01
8.52259099e-01 4.72982079e-01 4.68215868e-02 -1.02229321e+00
-4.02403831e-01 4.85735804e-01 -4.34274763e-01 -6.58372700e-01
1.27552354e+00 -1.35799035e-01 -2.20605835e-01 4.67742264e-01
6.97018981e-01 -3.63068432e-02 6.42290533e-01 7.58008957e-01
1.49404228e+00 3.25204842e-02 2.40242869e-01 -3.36978167e-01
5.82969546e-01 2.43181229e-01 -3.43796760e-02 1.30201674e+00
5.05951867e-02 6.62685156e-01 -1.56095132e-01 5.26819453e-02
-1.00470054e+00 -1.22825384e+00 -4.30886894e-01 1.53482008e+00
2.72633463e-01 -6.00493252e-01 -5.09520471e-01 -5.27513981e-01
3.42153251e-01 9.39625144e-01 -5.70427477e-01 -3.94064039e-01
-5.88795960e-01 -3.88671160e-01 5.91863930e-01 3.83701682e-01
2.70082355e-01 -1.16202438e+00 1.72024239e-02 5.04935265e-01
-2.80638367e-01 -1.04755771e+00 -2.62049705e-01 -1.62137095e-02
-8.28984857e-01 -7.54843712e-01 -1.48399830e+00 -9.16472077e-01
2.52094716e-01 4.46307421e-01 1.61911190e+00 4.82978746e-02
-2.18333542e-01 1.10878468e+00 -9.30723310e-01 -1.30268246e-01
-5.40890545e-02 6.94617748e-01 2.76675564e-03 -6.36199176e-01
1.04359639e+00 -3.80087852e-01 -8.81818473e-01 1.87717095e-01
-1.28234887e+00 -9.66749191e-01 3.85435909e-01 9.44118559e-01
3.99928123e-01 -3.74551684e-01 1.28957641e+00 -8.52820039e-01
1.35424304e+00 -9.71144617e-01 -1.17834158e-01 9.61977422e-01
-5.85369170e-01 3.49174559e-01 4.98020709e-01 -4.25389588e-01
-1.08813739e+00 -7.79997349e-01 -1.81488410e-01 -2.37909108e-01
1.79627299e-01 7.27542579e-01 3.90425086e-01 1.00788683e-01
1.29423738e+00 2.42109254e-01 -6.08165085e-01 -8.73912513e-01
7.70859003e-01 8.49905431e-01 -1.90889686e-02 -8.78279626e-01
6.31642103e-01 1.23025298e-01 -4.65128332e-01 -8.77346516e-01
-9.19687629e-01 -1.43291080e+00 -3.73973161e-01 1.89506054e-01
7.28262007e-01 -1.14512753e+00 -5.25719523e-02 -1.45915762e-01
-1.27722967e+00 1.89903289e-01 -5.90719402e-01 4.74664062e-01
-3.21631074e-01 4.66939479e-01 -7.61472285e-01 -4.09787595e-01
-6.85764968e-01 -3.82030129e-01 1.11367941e+00 4.20147538e-01
-1.05156220e-01 -1.12112355e+00 6.55261576e-01 1.87058538e-01
8.95025432e-01 -5.07726669e-01 1.14026618e+00 -1.31748724e+00
-6.36667132e-01 -5.08600295e-01 -1.99116603e-01 3.29461515e-01
-7.06208199e-02 -6.67507768e-01 -8.12244952e-01 -4.42160904e-01
-3.56162429e-01 -7.89623499e-01 1.34046614e+00 -5.66029996e-02
5.41550159e-01 -6.59329891e-02 -4.77151483e-01 -9.19224229e-03
1.61923099e+00 -3.01038802e-01 7.76476800e-01 6.65039301e-01
1.81125030e-01 6.39547408e-01 5.84239542e-01 1.44167796e-01
3.21647048e-01 4.64862972e-01 -3.60083878e-02 4.16588098e-01
-3.28007042e-01 -6.20059252e-01 2.15852894e-02 1.03473103e+00
5.66625409e-02 -4.99804199e-01 -8.05582881e-01 1.08111560e+00
-1.65094185e+00 -8.50199580e-01 3.13564181e-01 2.15626574e+00
1.06465566e+00 -4.50789958e-01 -3.69250208e-01 -5.58749259e-01
5.85355997e-01 2.64990896e-01 -3.16365272e-01 -2.41417766e-01
-2.71507800e-01 8.16061735e-01 3.92509550e-01 4.35942411e-01
-7.71254063e-01 1.20783293e+00 6.05289459e+00 1.21525896e+00
-3.70940477e-01 2.89276481e-01 -3.49106163e-01 -7.93093070e-03
-5.16542256e-01 2.13192031e-02 -1.17804837e+00 2.95493603e-02
9.39480782e-01 -7.32541203e-01 2.34140307e-01 8.50833356e-01
-5.26789069e-01 1.23254359e-02 -1.09280062e+00 5.88695765e-01
4.50858504e-01 -1.29779577e+00 5.77605486e-01 -3.86711776e-01
8.91876876e-01 2.79179990e-01 -1.37814015e-01 1.23237371e+00
3.08032215e-01 -8.26573372e-01 -2.01566680e-03 6.49252474e-01
4.32749927e-01 -2.44109780e-01 7.28127718e-01 4.84527022e-01
-7.89300323e-01 -2.23828317e-03 -1.25764942e+00 4.17404592e-01
4.64378297e-02 2.21370786e-01 -8.86326969e-01 8.04925740e-01
7.19360828e-01 3.81495088e-01 -6.44059837e-01 1.21091759e+00
-2.66365737e-01 3.68107766e-01 -2.63204247e-01 -4.20436442e-01
5.23766577e-01 1.61556825e-01 6.03187263e-01 1.39052272e+00
4.97751266e-01 1.79552808e-01 -2.90446967e-01 6.52601242e-01
-6.37304723e-01 6.63681805e-01 -7.25964963e-01 -2.05797344e-01
4.92320657e-01 9.69419897e-01 6.38638362e-02 -4.66980368e-01
-4.92406577e-01 9.35955465e-01 6.65956080e-01 8.65305364e-01
-2.29365662e-01 -9.25605536e-01 2.90770769e-01 2.30220094e-01
6.12488031e-01 4.11349386e-02 7.48575091e-01 -1.49519897e+00
2.71201670e-01 -6.37874842e-01 7.93337405e-01 -8.60147119e-01
-1.69390380e+00 5.98110020e-01 1.24201074e-01 -1.03866887e+00
-4.63658690e-01 -5.56948900e-01 -2.50240445e-01 1.06340337e+00
-2.09821820e+00 -9.05195534e-01 3.20331985e-03 6.71063364e-01
4.06461000e-01 -2.99810648e-01 1.19502652e+00 6.29967809e-01
1.07143987e-02 5.54199755e-01 8.82939816e-01 1.60379365e-01
1.19589674e+00 -1.21848822e+00 4.50907238e-02 4.76971984e-01
3.46934378e-01 1.34439135e+00 5.23622572e-01 -4.83179122e-01
-1.41386199e+00 -9.74147201e-01 1.24755859e+00 -5.35336375e-01
8.89255762e-01 -1.52213916e-01 -1.40692317e+00 5.29097438e-01
6.37708306e-01 -6.43262193e-02 7.78642535e-01 5.38507640e-01
-7.73449898e-01 -1.26646027e-01 -1.00368857e+00 4.36251134e-01
8.10247838e-01 -1.02856433e+00 -1.65209782e+00 6.47065580e-01
1.12307763e+00 1.58537440e-02 -1.03315794e+00 2.96731859e-01
2.77305424e-01 -3.66610616e-01 1.24999738e+00 -1.22457111e+00
3.72095972e-01 -8.04732516e-02 -3.68789613e-01 -1.46373177e+00
-3.54719669e-01 -1.96901530e-01 -3.39093506e-01 1.30391002e+00
6.34024501e-01 -5.34869909e-01 4.03826386e-01 4.74579304e-01
-9.00274217e-02 -4.26879048e-01 -1.01951444e+00 -1.03666842e+00
8.49392891e-01 1.00588441e-01 3.89797688e-01 7.35361934e-01
2.71035463e-01 6.14906132e-01 2.68627871e-02 -3.76196057e-02
1.96919739e-01 1.87850118e-01 4.89964664e-01 -1.20175886e+00
-1.58968195e-01 -1.64703697e-01 -3.00990134e-01 -1.53632951e+00
4.95611042e-01 -1.21695018e+00 -1.58370927e-01 -1.99854875e+00
3.26385140e-01 -3.44010144e-01 -7.15734422e-01 -2.51925178e-02
-4.07318383e-01 -2.05875691e-02 -2.14637276e-02 3.06588262e-01
-1.24080729e+00 7.24247634e-01 1.24115062e+00 -3.87830585e-01
4.67492547e-03 -3.86490524e-01 -6.95946217e-01 1.23071857e-01
3.63627076e-01 -6.69450819e-01 -3.59963119e-01 -6.27223551e-01
5.92923105e-01 5.13527878e-02 -1.63586214e-02 -7.51390636e-01
7.49346316e-01 4.26054895e-01 -8.52714479e-02 -4.55868334e-01
3.48316193e-01 -6.74563706e-01 -5.10813892e-01 -3.69183309e-02
-6.34530008e-01 -1.86319709e-01 1.17448084e-01 8.47935975e-01
-7.74431109e-01 -9.35917914e-01 2.14614421e-01 -4.38340336e-01
-1.06286943e+00 2.96102613e-01 -2.41182208e-01 8.84524286e-01
5.35391510e-01 2.93893993e-01 -6.96441412e-01 -4.13921863e-01
-5.47068357e-01 4.98780191e-01 2.36523405e-01 7.41263747e-01
5.55095792e-01 -1.38326228e+00 -9.15415406e-01 -6.03240609e-01
7.45181680e-01 -2.97934473e-01 5.72404146e-01 2.72105575e-01
-4.47575599e-01 1.17019463e+00 1.03710897e-01 -1.38238981e-01
-6.30214393e-01 9.28705275e-01 1.16229504e-01 -7.30319083e-01
-5.20352423e-01 8.02712679e-01 1.76283091e-01 -4.92905378e-01
2.61266321e-01 -1.15732722e-01 -6.02821887e-01 5.35478473e-01
8.00207138e-01 3.84014159e-01 4.17006105e-01 -2.85832196e-01
-2.12017149e-01 6.43303633e-01 -6.11034095e-01 9.03618243e-03
1.39213049e+00 -2.98819721e-01 -2.52651483e-01 2.28205904e-01
1.57394946e+00 -5.75743429e-02 -1.88762113e-01 -8.63765657e-01
3.56210500e-01 -1.67516500e-01 -7.61975870e-02 -7.35297799e-01
-3.10035855e-01 1.01023734e+00 3.90593588e-01 2.68553257e-01
8.67953777e-01 2.69378573e-01 1.16729009e+00 1.27273929e+00
6.62685096e-01 -1.27204084e+00 2.94550527e-02 9.03575361e-01
9.33480263e-01 -1.10399771e+00 -1.02935538e-01 2.80381083e-01
-3.16786468e-01 1.01289773e+00 3.87270898e-01 -4.32055175e-01
7.04388916e-01 -5.65972567e-01 -1.09775372e-01 -2.28876710e-01
-7.90893495e-01 -6.49550676e-01 6.87628925e-01 4.95874166e-01
4.36394960e-01 -2.73259521e-01 -6.85759008e-01 5.28077126e-01
1.55942068e-01 9.92417559e-02 3.27478439e-01 9.43938315e-01
-8.70036006e-01 -1.08914161e+00 6.19389024e-03 4.38070953e-01
-7.32059538e-01 -6.47157788e-01 -2.59346783e-01 8.18007469e-01
-4.12114918e-01 8.37499022e-01 -1.82111919e-01 1.45981554e-02
6.00666583e-01 5.80876946e-01 4.49307144e-01 -9.96020734e-01
-7.26288736e-01 -6.32700682e-01 2.85001934e-01 -2.73401052e-01
-3.58177364e-01 -4.71172621e-03 -6.38853550e-01 6.55817166e-02
-6.97741747e-01 7.53014505e-01 3.19530249e-01 7.60928869e-01
5.36067784e-01 1.62006825e-01 2.59680718e-01 -1.47016659e-01
-9.59492505e-01 -1.36539614e+00 -7.35209286e-01 6.43733025e-01
2.14373499e-01 -6.31659985e-01 -5.59173524e-01 -4.47160095e-01]
|
[11.459321975708008, 7.76085090637207]
|
301258fe-3f7f-4f30-a8fe-5f7446159686
|
low-rank-tensor-function-representation-for
|
2212.00262
| null |
https://arxiv.org/abs/2212.00262v1
|
https://arxiv.org/pdf/2212.00262v1.pdf
|
Low-Rank Tensor Function Representation for Multi-Dimensional Data Recovery
|
Since higher-order tensors are naturally suitable for representing multi-dimensional data in real-world, e.g., color images and videos, low-rank tensor representation has become one of the emerging areas in machine learning and computer vision. However, classical low-rank tensor representations can only represent data on finite meshgrid due to their intrinsical discrete nature, which hinders their potential applicability in many scenarios beyond meshgrid. To break this barrier, we propose a low-rank tensor function representation (LRTFR), which can continuously represent data beyond meshgrid with infinite resolution. Specifically, the suggested tensor function, which maps an arbitrary coordinate to the corresponding value, can continuously represent data in an infinite real space. Parallel to discrete tensors, we develop two fundamental concepts for tensor functions, i.e., the tensor function rank and low-rank tensor function factorization. We theoretically justify that both low-rank and smooth regularizations are harmoniously unified in the LRTFR, which leads to high effectiveness and efficiency for data continuous representation. Extensive multi-dimensional data recovery applications arising from image processing (image inpainting and denoising), machine learning (hyperparameter optimization), and computer graphics (point cloud upsampling) substantiate the superiority and versatility of our method as compared with state-of-the-art methods. Especially, the experiments beyond the original meshgrid resolution (hyperparameter optimization) or even beyond meshgrid (point cloud upsampling) validate the favorable performances of our method for continuous representation.
|
['Deyu Meng', 'Michael K. Ng', 'Zhemin Li', 'XiLe Zhao', 'YiSi Luo']
|
2022-12-01
| null | null | null | null |
['image-inpainting']
|
['computer-vision']
|
[-1.18173234e-01 -4.12226945e-01 -8.29285011e-02 2.20733538e-01
-4.37778294e-01 -3.53128433e-01 3.88136029e-01 -8.68531242e-02
-3.77242006e-02 5.47757745e-01 1.07442565e-01 -6.23689033e-04
-5.22297740e-01 -7.14338899e-01 -5.72018623e-01 -9.39270437e-01
-8.75409842e-02 2.29776934e-01 -4.34506796e-02 -4.65837002e-01
1.70629889e-01 6.92707300e-01 -1.58759511e+00 2.35857978e-01
1.14020669e+00 1.06870961e+00 4.37566638e-02 3.30289125e-01
-2.43204996e-01 3.93650770e-01 -1.67187348e-01 -1.79540947e-01
2.47236490e-01 4.95420843e-02 -8.29736471e-01 5.43277919e-01
4.47571248e-01 -3.97742391e-01 -4.88213450e-01 1.17306173e+00
-3.74603160e-02 2.13662460e-01 4.39133435e-01 -1.05103648e+00
-9.10945535e-01 -1.25854611e-01 -8.65383267e-01 -1.15100428e-01
3.37338239e-01 -9.23180878e-02 8.59329700e-01 -1.25782597e+00
4.64607149e-01 1.31138611e+00 6.07628286e-01 1.08004816e-01
-1.49657416e+00 -1.72213748e-01 1.52538329e-01 3.58918160e-02
-1.44549334e+00 1.90774556e-02 1.16370487e+00 -7.16128588e-01
2.12137029e-01 7.13973880e-01 6.06131613e-01 5.15918434e-01
1.64989978e-01 5.60257852e-01 1.13966656e+00 -1.56049401e-01
1.14724524e-02 -3.11999947e-01 4.54456136e-02 5.61680496e-01
2.73202956e-01 2.39646044e-02 -4.59731966e-01 -4.52523708e-01
1.47190177e+00 4.24230218e-01 -5.68750918e-01 -2.37502426e-01
-1.84927416e+00 7.43406177e-01 3.18092525e-01 4.91169631e-01
-5.85724652e-01 1.45973280e-01 5.38029492e-01 2.21638292e-01
6.86192632e-01 2.49143273e-01 -6.25839308e-02 -2.34978516e-02
-7.06350148e-01 2.68824339e-01 1.56552687e-01 7.62174606e-01
9.52105522e-01 3.28604102e-01 -9.38368291e-02 1.06201625e+00
2.33225465e-01 4.54486698e-01 4.21070516e-01 -1.17058802e+00
3.04433703e-01 6.57222986e-01 3.40360224e-01 -1.58283031e+00
-2.03262359e-01 -3.51064920e-01 -1.59056747e+00 1.19473904e-01
1.72391325e-01 5.08962810e-01 -4.51913744e-01 1.26040697e+00
5.95236242e-01 5.42546809e-01 -6.27164468e-02 1.20285189e+00
5.79142690e-01 7.05220282e-01 -4.22553450e-01 -5.37693441e-01
1.49881351e+00 -5.20449340e-01 -7.80564904e-01 4.64567900e-01
6.02982998e-01 -1.04363561e+00 1.15145779e+00 7.58710444e-01
-8.97024155e-01 -5.74589312e-01 -8.81779373e-01 -2.26446122e-01
1.48457587e-01 2.49895528e-01 1.01458359e+00 1.04925908e-01
-7.79193819e-01 8.37009609e-01 -7.80025780e-01 5.56475371e-02
-2.99678110e-02 -2.00967910e-03 -6.81721926e-01 -3.23271722e-01
-9.49714482e-01 2.38090813e-01 -1.12136096e-01 6.03406370e-01
-3.13430727e-01 -8.56297374e-01 -4.56914753e-01 -2.17999414e-01
8.76910314e-02 -5.88187754e-01 6.47361934e-01 -4.53901350e-01
-1.38098049e+00 6.16532505e-01 -8.15008804e-02 1.26012236e-01
4.93000388e-01 -1.98136568e-01 -3.10102880e-01 4.26200747e-01
1.78295121e-01 -3.50688025e-02 1.24560714e+00 -1.38296342e+00
-9.12378430e-02 -3.35477322e-01 1.26840353e-01 7.52300546e-02
-5.41428804e-01 -3.86952043e-01 -3.14792335e-01 -1.07118988e+00
8.89541268e-01 -8.24683189e-01 -4.39610481e-01 3.38336825e-01
-2.65088499e-01 -1.74890101e-01 9.28243160e-01 -6.84979200e-01
1.21001911e+00 -2.38856173e+00 5.78489304e-01 1.14509270e-01
4.96729195e-01 -3.94786485e-02 -4.96170223e-02 3.82853091e-01
-3.42903912e-01 2.37526763e-02 -1.85835958e-01 -1.67498782e-01
-2.87991881e-01 4.77157950e-01 -4.46837395e-01 8.37488949e-01
2.14047015e-01 3.68113160e-01 -9.22548354e-01 -3.83752733e-01
2.89406657e-01 7.16735780e-01 -6.57734156e-01 5.01342937e-02
2.85478029e-02 9.31773901e-01 -9.36388373e-01 4.61780161e-01
1.13693225e+00 -2.68880367e-01 -2.71391600e-01 -1.05825663e+00
-5.09162128e-01 -2.94429302e-01 -1.51398754e+00 1.81662738e+00
-4.82762188e-01 7.40742683e-02 5.37060142e-01 -1.02378285e+00
9.67487752e-01 3.98037761e-01 9.17876422e-01 -4.32392031e-01
-1.93182364e-01 4.71896887e-01 -3.49107563e-01 -5.09058237e-01
7.20023632e-01 -1.62183434e-01 1.95326865e-01 1.62646636e-01
-5.06122231e-01 -1.64049253e-01 1.82455286e-01 1.27160728e-01
7.23437071e-01 -2.85856705e-02 -2.47988865e-01 -4.89567846e-01
9.61920142e-01 -1.87036514e-01 6.41107500e-01 6.88098967e-02
2.52143592e-01 7.36242592e-01 5.01192212e-01 -5.80643535e-01
-1.12120235e+00 -6.14297986e-01 -5.51882446e-01 7.10943639e-01
3.05192083e-01 -4.43662375e-01 -4.42791462e-01 -1.09044962e-01
7.29396492e-02 6.13586158e-02 -5.16900957e-01 2.70236880e-02
-8.35013926e-01 -6.73485696e-01 2.17880353e-01 2.79809862e-01
5.76163054e-01 -3.53364378e-01 -8.41626432e-03 2.34542653e-01
-3.18314850e-01 -1.15599179e+00 -5.20619512e-01 -5.59405923e-01
-1.37426794e+00 -1.01173162e+00 -9.15606260e-01 -4.82191563e-01
8.40216517e-01 7.95706213e-01 7.99791157e-01 4.82815117e-01
-1.38863385e-01 4.94696707e-01 -3.02257776e-01 6.98609769e-01
-1.40333429e-01 -4.75287467e-01 2.84475505e-01 6.49975002e-01
-4.64976817e-01 -8.14356327e-01 -7.43385017e-01 5.60193181e-01
-1.47274816e+00 2.96680987e-01 3.57248962e-01 9.65511978e-01
1.02794075e+00 2.01030388e-01 2.13002652e-01 -6.95193112e-01
5.59023857e-01 -3.33344042e-01 -6.52128935e-01 1.20756455e-01
-1.92758262e-01 1.10108614e-01 8.12877655e-01 -5.45504570e-01
-7.20191121e-01 -2.19655246e-01 -5.97850867e-02 -1.09446549e+00
3.62803489e-01 6.92080438e-01 3.65649015e-02 -4.48948354e-01
4.24634397e-01 3.77656221e-01 1.45605311e-01 -9.11739886e-01
3.93084586e-01 1.85724407e-01 4.23920572e-01 -1.18836045e+00
1.07256424e+00 8.52171361e-01 5.66856384e-01 -1.25936937e+00
-5.07110775e-01 -4.42823380e-01 -5.64032614e-01 -1.22284666e-01
5.91395557e-01 -7.68721402e-01 -1.04846704e+00 5.29065907e-01
-1.30181086e+00 2.70867825e-01 -2.89919317e-01 7.15574563e-01
-5.30827940e-01 8.33131969e-01 -7.59938419e-01 -6.58690810e-01
-1.76210538e-01 -1.18459773e+00 1.23470223e+00 -2.78168380e-01
3.83663207e-01 -9.84994411e-01 -2.11395606e-01 4.06457961e-01
4.12749141e-01 6.39660895e-01 1.12324750e+00 4.07996237e-01
-7.94492602e-01 -7.81370252e-02 -4.08029169e-01 4.13409352e-01
1.48999467e-01 9.57604796e-02 -5.70297897e-01 -3.85192484e-01
2.94133514e-01 -1.20812543e-01 4.86105144e-01 8.27196687e-02
1.50272381e+00 -4.51274961e-01 -6.76601604e-02 8.25881422e-01
1.49153137e+00 -4.34323937e-01 6.35418534e-01 7.98478127e-02
1.14312291e+00 6.19735360e-01 7.29036510e-01 5.65369248e-01
1.29796565e-01 9.06364202e-01 5.77040970e-01 -2.48376623e-01
-3.88439745e-02 -2.82601193e-02 5.18811271e-02 1.39327633e+00
-5.82538247e-01 4.22154188e-01 -5.45998514e-01 3.58448178e-01
-1.93032646e+00 -5.69034457e-01 -7.29236424e-01 2.45620418e+00
6.11319721e-01 -2.89471805e-01 -3.06788944e-02 4.57902282e-01
7.15446115e-01 3.34191889e-01 -4.92232740e-01 -1.31498307e-01
-1.81870878e-01 -1.53817832e-01 3.45269680e-01 5.22737741e-01
-8.15196753e-01 5.29144824e-01 5.29132128e+00 1.00762343e+00
-1.28927946e+00 1.28128380e-01 3.76043588e-01 3.42372358e-01
-7.11543679e-01 -3.57327312e-02 -2.52284706e-01 2.99490571e-01
3.27865601e-01 -1.14546709e-01 7.25327611e-01 6.70775473e-01
4.21158195e-01 3.11042398e-01 -8.38888168e-01 1.36193371e+00
-2.65010864e-01 -1.54405999e+00 3.87366176e-01 2.79635787e-01
6.72772825e-01 -3.13139975e-01 1.39538646e-01 -2.43156264e-03
-2.57327855e-01 -8.56052160e-01 4.68457043e-01 6.26564801e-01
1.13605726e+00 -6.17476583e-01 3.23211104e-01 1.91833898e-01
-1.48496437e+00 3.32373440e-01 -6.36683166e-01 5.38492166e-02
1.25113666e-01 1.08003235e+00 -2.62920670e-02 1.06806707e+00
5.37782073e-01 1.07102859e+00 -2.32946932e-01 8.15236330e-01
2.64539301e-01 2.75835574e-01 -3.50672930e-01 5.56241512e-01
2.69915044e-01 -1.01647913e+00 7.71463752e-01 5.52776098e-01
4.68302071e-01 3.96039099e-01 5.01728415e-01 6.77268088e-01
8.25884044e-02 3.34959060e-01 -4.39907759e-01 -1.54600795e-02
5.41029386e-02 1.29271340e+00 -5.44015408e-01 -7.67492950e-02
-4.89457756e-01 8.75310659e-01 7.10468180e-03 7.32186258e-01
-4.68940556e-01 -1.20776363e-01 1.02310586e+00 3.52085441e-01
5.93298078e-02 -6.91390336e-01 -3.48855704e-01 -1.62809956e+00
4.67972338e-01 -7.77301133e-01 8.86049792e-02 -5.55490315e-01
-1.44083583e+00 6.52758598e-01 3.57412659e-02 -1.82416189e+00
1.64428607e-01 -7.39374518e-01 -3.53768200e-01 8.16677809e-01
-1.49478424e+00 -1.15738356e+00 -4.16775495e-01 8.86648178e-01
1.94825709e-01 2.98129350e-01 7.60195374e-01 6.37064755e-01
-4.05919939e-01 2.00043812e-01 4.67585862e-01 -2.70793657e-03
2.22615317e-01 -9.48327184e-01 -3.90804745e-02 6.26301885e-01
-3.08210820e-01 8.30311835e-01 5.48573315e-01 -5.09575188e-01
-2.16904640e+00 -1.01688337e+00 1.60835773e-01 1.72684282e-01
8.99555027e-01 -1.75554454e-01 -1.43827534e+00 3.36146742e-01
-3.31391066e-01 5.38459778e-01 3.24640930e-01 -5.61058037e-02
-4.75297868e-01 -4.32988912e-01 -1.12249577e+00 5.93699098e-01
7.32968926e-01 -5.03202140e-01 -1.48521841e-01 8.17987621e-01
7.18783319e-01 -3.65480602e-01 -1.44839191e+00 5.07271111e-01
2.70786077e-01 -8.88146996e-01 1.17391562e+00 -4.42601115e-01
4.12191510e-01 -7.17752516e-01 -3.69054079e-01 -1.05149627e+00
-4.69349772e-01 -1.01136172e+00 -2.71682352e-01 1.13423085e+00
-3.65873665e-01 -8.19204211e-01 5.78691423e-01 4.40107018e-01
-3.46549273e-01 -9.45080876e-01 -1.19669175e+00 -8.16648304e-01
1.21326834e-01 -3.83847415e-01 5.96355319e-01 1.20706940e+00
-3.34284097e-01 -1.11011557e-01 -5.46854496e-01 3.34798425e-01
8.60618174e-01 3.03521693e-01 5.88211834e-01 -1.37812436e+00
-2.97149688e-01 -3.28430325e-01 -6.38777494e-01 -1.40665555e+00
3.10182236e-02 -7.16007471e-01 -2.60897130e-01 -1.43667936e+00
-1.75914973e-01 -9.49149191e-01 -6.97339624e-02 1.21491127e-01
1.38679639e-01 4.29255694e-01 -2.29993034e-02 7.35235035e-01
-5.42756990e-02 9.41937685e-01 2.00853896e+00 -3.95194702e-02
-3.09713539e-02 -2.24790499e-01 -3.80527645e-01 7.29894102e-01
1.79868057e-01 -4.49795462e-02 -3.36693525e-01 -6.75447524e-01
4.00846303e-01 4.46040660e-01 3.59988242e-01 -7.44310915e-01
-1.51922693e-02 -5.14837503e-01 -1.75576080e-02 -4.77976441e-01
5.90468824e-01 -8.54917824e-01 3.29105496e-01 1.65211707e-01
8.68961886e-02 1.37269557e-01 1.80172194e-02 5.87244332e-01
-4.89638746e-01 -7.44473487e-02 6.69450164e-01 2.90717203e-02
-3.17933798e-01 7.01591134e-01 -1.56600531e-02 -7.91707262e-02
7.70337224e-01 -2.17932984e-01 -2.51249582e-01 -1.42382905e-01
-7.00872838e-01 -1.40333936e-01 6.01484835e-01 2.35534638e-01
9.11422193e-01 -1.84377706e+00 -8.69956672e-01 5.26147425e-01
3.73867862e-02 3.52075338e-01 5.94314754e-01 1.19192493e+00
-7.32063413e-01 1.11382440e-01 3.56549360e-02 -9.41835105e-01
-8.00343156e-01 5.02457738e-01 1.23275645e-01 -2.32929736e-01
-8.94279838e-01 4.61319208e-01 4.62006420e-01 -1.62313953e-01
-1.40391558e-01 -5.88905036e-01 -7.40924552e-02 -1.79095164e-01
6.31212592e-01 5.70376337e-01 2.52676934e-01 -9.90230680e-01
-8.54395628e-02 1.09838367e+00 3.12677249e-02 1.89638764e-01
1.33629394e+00 6.89796060e-02 -7.72362530e-01 4.54315394e-01
1.37149537e+00 1.39552101e-01 -1.20089090e+00 -2.15105087e-01
-3.48286510e-01 -9.56712484e-01 3.10199469e-01 2.80326724e-01
-1.18246531e+00 7.88881183e-01 2.03010604e-01 5.38952172e-01
1.12980449e+00 -2.62510270e-01 9.68906879e-01 3.18072140e-01
5.42569160e-01 -5.30307353e-01 1.98478848e-01 3.62041056e-01
1.37243378e+00 -8.77024293e-01 3.56942475e-01 -8.87529135e-01
-2.87585974e-01 1.34856546e+00 2.19474077e-01 -4.35021132e-01
6.50344074e-01 -2.38765597e-01 -2.47691393e-01 -1.36301100e-01
-4.31495547e-01 2.68240720e-01 5.45953691e-01 2.81561345e-01
5.67014635e-01 6.92893565e-02 -4.24535245e-01 4.65050014e-03
4.13244143e-02 -1.98472947e-01 4.94917274e-01 5.83999991e-01
-7.16456547e-02 -1.17138636e+00 -9.45485890e-01 4.55270529e-01
-2.34454304e-01 1.14042588e-01 3.40557784e-01 4.00067300e-01
1.36270851e-03 7.63693571e-01 -1.55924425e-01 -2.82750785e-01
3.40305984e-01 -4.58255440e-01 4.06058699e-01 -3.40231270e-01
-3.06771904e-01 3.58709067e-01 -2.71798223e-01 -7.44903088e-01
-4.56815302e-01 -5.93774676e-01 -1.18768692e+00 -5.03503442e-01
-1.73158869e-01 4.17031050e-01 6.73855007e-01 6.70188844e-01
4.05562967e-01 2.78943151e-01 9.69084799e-01 -1.04382706e+00
-4.36792433e-01 -6.80729389e-01 -8.83985996e-01 7.19438195e-01
5.57679296e-01 -1.04112279e+00 -4.92332876e-01 -5.11545874e-02]
|
[7.426629066467285, 4.456050872802734]
|
2ef9ce88-67ca-4c2e-9211-4dc9651da3d9
|
unbiased-teacher-for-semi-supervised-object-1
|
2102.09480
| null |
https://arxiv.org/abs/2102.09480v1
|
https://arxiv.org/pdf/2102.09480v1.pdf
|
Unbiased Teacher for Semi-Supervised Object Detection
|
Semi-supervised learning, i.e., training networks with both labeled and unlabeled data, has made significant progress recently. However, existing works have primarily focused on image classification tasks and neglected object detection which requires more annotation effort. In this work, we revisit the Semi-Supervised Object Detection (SS-OD) and identify the pseudo-labeling bias issue in SS-OD. To address this, we introduce Unbiased Teacher, a simple yet effective approach that jointly trains a student and a gradually progressing teacher in a mutually-beneficial manner. Together with a class-balance loss to downweight overly confident pseudo-labels, Unbiased Teacher consistently improved state-of-the-art methods by significant margins on COCO-standard, COCO-additional, and VOC datasets. Specifically, Unbiased Teacher achieves 6.8 absolute mAP improvements against state-of-the-art method when using 1% of labeled data on MS-COCO, achieves around 10 mAP improvements against the supervised baseline when using only 0.5, 1, 2% of labeled data on MS-COCO.
|
['Peter Vajda', 'Zsolt Kira', 'Bichen Wu', 'Peizhao Zhang', 'Kan Chen', 'Chia-Wen Kuo', 'Zijian He', 'Chih-Yao Ma', 'Yen-Cheng Liu']
|
2021-02-18
|
unbiased-teacher-for-semi-supervised-object
|
https://openreview.net/forum?id=MJIve1zgR_
|
https://openreview.net/pdf?id=MJIve1zgR_
|
iclr-2021-1
|
['semi-supervised-object-detection', 'semi-supervised-person-bounding-box-detection']
|
['computer-vision', 'computer-vision']
|
[ 2.98492134e-01 3.58439595e-01 -4.22252387e-01 -6.03089154e-01
-1.06077766e+00 -5.74075758e-01 6.38267636e-01 5.98442741e-02
-7.52162576e-01 7.90976763e-01 -3.44228476e-01 -1.92819893e-01
4.30287808e-01 -3.62230599e-01 -9.48176742e-01 -6.98826551e-01
1.89146951e-01 3.94648671e-01 5.77955782e-01 3.51780862e-01
-1.48180678e-01 1.13518089e-01 -1.51399970e+00 2.93150753e-01
9.18451786e-01 1.11412406e+00 2.31262863e-01 4.68808740e-01
1.57631584e-03 1.07797587e+00 -5.75496614e-01 -3.34355712e-01
2.20994189e-01 -2.40858778e-01 -8.69672537e-01 3.16418529e-01
1.15733910e+00 -3.17298204e-01 -2.50474244e-01 1.25323236e+00
4.22245711e-01 -2.69329101e-01 8.47679734e-01 -1.33096433e+00
-5.52617788e-01 7.07303226e-01 -8.82251918e-01 2.03363717e-01
-4.61981207e-01 1.02160141e-01 8.66878510e-01 -1.25610793e+00
4.63519067e-01 1.02565610e+00 8.68179739e-01 6.19992316e-01
-1.43268466e+00 -1.02988338e+00 1.95905715e-01 -2.23093592e-02
-1.42382216e+00 -2.43449256e-01 2.89571315e-01 -5.14530897e-01
5.06436169e-01 4.18005064e-02 1.23492226e-01 9.52924967e-01
-2.87478358e-01 1.28176224e+00 1.32500660e+00 -5.57093859e-01
6.91928715e-02 5.33581257e-01 5.13609886e-01 6.70140505e-01
4.10029471e-01 1.80589169e-01 -3.38006049e-01 2.23246381e-01
3.51345748e-01 -1.69799894e-01 4.28313613e-02 -4.95115250e-01
-1.05074823e+00 6.35730267e-01 6.85515881e-01 -1.73294276e-01
-1.04977958e-01 1.55263394e-01 3.63281727e-01 1.08333607e-03
8.97244215e-01 4.28955078e-01 -6.71344638e-01 2.04840451e-01
-1.20095456e+00 -1.86225995e-01 4.32654679e-01 1.18022346e+00
8.23816061e-01 2.36619875e-01 -3.66967440e-01 9.14323330e-01
2.87942946e-01 6.25347853e-01 1.41302884e-01 -7.48573303e-01
3.14501435e-01 6.05226159e-01 9.72699448e-02 -3.24878424e-01
-2.95995951e-01 -9.23771620e-01 -5.94230056e-01 3.25997472e-01
5.74767232e-01 -2.78163314e-01 -1.46509874e+00 1.69973087e+00
3.87304008e-01 3.86887997e-01 1.04514316e-01 7.74675012e-01
1.10760140e+00 4.48407322e-01 3.60303432e-01 2.87541505e-02
1.17076623e+00 -1.59258497e+00 -4.97026056e-01 -4.39123660e-01
6.98514462e-01 -6.39295220e-01 1.12771368e+00 3.58494639e-01
-8.98232341e-01 -7.33426869e-01 -1.25818515e+00 8.21963772e-02
-4.19902176e-01 8.14094663e-01 3.57166320e-01 6.26137257e-01
-8.85067165e-01 3.93183887e-01 -8.42576742e-01 -2.62216657e-01
1.02179956e+00 2.61147887e-01 -2.71531463e-01 -2.93904245e-01
-8.01715255e-01 7.50329852e-01 4.75450456e-01 -1.38597995e-01
-1.53631055e+00 -9.08808291e-01 -7.88879395e-01 -4.14112248e-02
7.78595269e-01 5.16931573e-03 1.52229953e+00 -9.51506019e-01
-8.95968139e-01 1.08742273e+00 6.26477450e-02 -6.70457900e-01
7.24976957e-01 -5.69435060e-01 -2.38443330e-01 8.24551955e-02
4.91854787e-01 1.35280859e+00 6.72761023e-01 -1.47369754e+00
-1.00566781e+00 -2.62370616e-01 -1.62669301e-01 1.14535540e-01
-4.05744791e-01 2.69520413e-02 -6.18733644e-01 -5.71890533e-01
5.98244276e-03 -1.13618875e+00 -1.79528326e-01 1.38383299e-01
-6.61784768e-01 -4.89351571e-01 1.09198117e+00 -1.71136439e-01
8.42061043e-01 -2.15203786e+00 -4.24450994e-01 -3.20291251e-01
3.55242878e-01 7.73958981e-01 -2.19219074e-01 -1.98984444e-01
-9.38564315e-02 9.78180394e-02 -2.45545164e-01 -7.10596740e-01
-2.68663496e-01 3.97434793e-02 -2.51329720e-01 5.70663869e-01
5.11766076e-01 8.44758093e-01 -1.17635500e+00 -5.85657477e-01
7.61925802e-02 3.90373409e-01 -1.40012488e-01 2.41760150e-01
-2.91262686e-01 3.51259947e-01 -1.24882802e-01 8.09411347e-01
6.96260273e-01 -6.42564118e-01 -9.84709114e-02 -1.80663615e-01
-2.01846614e-01 1.74319759e-01 -9.43158805e-01 1.41937983e+00
-1.11378685e-01 1.06452191e+00 -1.75152756e-02 -9.00598288e-01
7.50976145e-01 1.97669744e-01 1.06839992e-01 -5.52025974e-01
5.84361665e-02 2.33260199e-01 -1.66935608e-01 -1.30203471e-01
2.60247260e-01 2.25057647e-01 3.34590048e-01 3.79260600e-01
3.86279702e-01 2.50494272e-01 1.85584947e-01 3.60570252e-01
8.74994159e-01 2.92403191e-01 8.37549940e-02 -3.76168787e-01
2.81324983e-01 3.15909624e-01 5.60231864e-01 9.40047324e-01
-4.87596422e-01 7.75550306e-01 2.37221435e-01 -2.62245357e-01
-8.75415027e-01 -9.49629962e-01 -4.30740923e-01 1.47085667e+00
2.18839407e-01 -1.23658232e-01 -9.47720408e-01 -1.26431084e+00
1.14704341e-01 6.25901818e-01 -8.09251130e-01 -6.03979453e-02
-3.10947776e-01 -1.02265084e+00 8.19999397e-01 8.89080048e-01
8.36231172e-01 -8.81493509e-01 -2.96119273e-01 -1.02853239e-01
1.85940012e-01 -1.31826723e+00 -3.80996913e-01 5.90107977e-01
-7.30502367e-01 -1.04337037e+00 -9.22042012e-01 -9.51889038e-01
1.06219757e+00 6.57114208e-01 1.21642983e+00 -1.29635245e-01
-3.34242374e-01 8.01320598e-02 -2.93191671e-01 -7.31622159e-01
-2.00841472e-01 2.67237812e-01 6.30738959e-03 4.95912470e-02
4.42269474e-01 -5.46615012e-02 -4.98158485e-01 6.89704001e-01
-6.20643318e-01 1.56305864e-01 7.57427037e-01 1.00586176e+00
7.75057375e-01 -3.21326166e-01 8.22799802e-01 -1.41090560e+00
-1.72568277e-01 -4.64904279e-01 -7.24548638e-01 3.51483017e-01
-1.16960633e+00 -3.33193764e-02 2.97669053e-01 -6.93302453e-01
-1.20532155e+00 4.56276357e-01 1.31492198e-01 -5.57336211e-01
-2.62731105e-01 6.68617561e-02 -5.38378917e-02 -1.74167231e-01
9.92233694e-01 1.38133690e-01 -2.81701058e-01 -5.00499666e-01
4.72622842e-01 9.22376335e-01 8.09637010e-01 -3.68231833e-01
7.91632056e-01 6.22021556e-01 -3.02293867e-01 -5.19836009e-01
-1.60414481e+00 -8.15010548e-01 -7.34917104e-01 -1.30572364e-01
8.59228492e-01 -1.35905218e+00 -1.19011827e-01 6.74013734e-01
-8.68479490e-01 -5.90690315e-01 -2.60269403e-01 4.80283231e-01
-4.36162390e-02 5.60140163e-02 -5.00813782e-01 -7.69800127e-01
-4.42297697e-01 -1.10261118e+00 1.02292287e+00 3.36932480e-01
1.52891934e-01 -6.02602601e-01 -1.85312822e-01 5.52460670e-01
3.23504508e-01 5.64741977e-02 2.95434564e-01 -9.71048057e-01
-4.08152401e-01 -2.89165735e-01 -8.64793718e-01 7.94824004e-01
2.26533022e-02 -1.67524934e-01 -1.31721330e+00 -5.45953512e-01
-5.08662164e-01 -1.03640592e+00 1.14482057e+00 1.54416546e-01
1.26235127e+00 4.72242273e-02 -5.89849412e-01 5.97049177e-01
1.21314979e+00 -6.27722666e-02 2.79858232e-01 2.62768507e-01
9.20299172e-01 6.39851511e-01 9.24866676e-01 -1.06048500e-02
3.71759325e-01 4.25374717e-01 6.47909105e-01 -4.81849194e-01
-6.19554818e-01 -3.06200415e-01 9.78001356e-02 3.69056553e-01
2.42208481e-01 -1.11800902e-01 -1.02626455e+00 7.13171661e-01
-1.62824059e+00 -4.53429192e-01 -4.26297277e-01 2.08583069e+00
1.03389573e+00 4.76883411e-01 1.67670816e-01 -7.06802495e-03
9.51292694e-01 -2.75381911e-03 -7.95884550e-01 5.16652882e-01
-9.23592821e-02 -2.02596411e-02 1.05580831e+00 1.19854562e-01
-1.61895525e+00 1.07544720e+00 6.48372984e+00 1.09144318e+00
-1.21511114e+00 4.21886086e-01 9.25143480e-01 -1.54089391e-01
3.70258898e-01 -2.42851183e-01 -1.25635207e+00 4.13559794e-01
7.65676320e-01 3.79482061e-01 -7.09072649e-02 1.37588727e+00
-1.53863728e-01 -9.52969566e-02 -1.10357010e+00 8.29455733e-01
1.32151425e-01 -1.16226053e+00 -2.57535160e-01 -1.02304578e-01
1.31140614e+00 6.29756510e-01 1.74811542e-01 6.39045000e-01
4.47311461e-01 -9.72959459e-01 8.57075989e-01 -1.14831597e-01
1.18423307e+00 -5.27596354e-01 9.69855249e-01 5.10933220e-01
-1.02089810e+00 1.18632376e-01 -3.29619527e-01 1.15952969e-01
-1.75330624e-01 5.22860169e-01 -8.98625791e-01 2.26508468e-01
9.09953117e-01 7.39531755e-01 -1.04926121e+00 1.20137250e+00
-5.84789693e-01 1.41505182e+00 -2.86129266e-01 1.58956423e-01
5.28249264e-01 3.63742560e-01 1.27645403e-01 1.49399829e+00
-1.97768539e-01 -2.23319277e-01 3.92691463e-01 7.79614091e-01
-4.25339729e-01 -1.98391616e-01 -1.34048954e-01 8.61487240e-02
6.13562584e-01 1.42242706e+00 -9.05463219e-01 -7.44680762e-01
-4.63725418e-01 7.48092055e-01 5.96024930e-01 3.23539734e-01
-8.60559762e-01 -3.80829304e-01 2.23056786e-02 1.38107896e-01
3.85094523e-01 2.13953018e-01 -3.22174907e-01 -9.14380729e-01
-2.33899862e-01 -6.00332379e-01 3.02312136e-01 -6.18392944e-01
-1.21698081e+00 6.32577240e-01 4.30649221e-02 -1.24921405e+00
1.56973496e-01 -6.45435929e-01 -6.03893638e-01 6.71028852e-01
-1.79032338e+00 -1.32590973e+00 -5.71697295e-01 1.40318617e-01
6.62477374e-01 -1.48234442e-01 5.04246771e-01 5.91978967e-01
-8.82710278e-01 9.08837080e-01 2.50662148e-01 4.12189960e-01
1.04421306e+00 -1.55825305e+00 4.79667962e-01 8.60560894e-01
3.16136926e-01 2.51839340e-01 4.87347364e-01 -4.44985211e-01
-7.28981316e-01 -1.59258723e+00 6.27746224e-01 -5.17088950e-01
4.18728769e-01 -3.67556602e-01 -9.66228127e-01 8.07914734e-01
1.37235761e-01 7.77444780e-01 5.31642497e-01 -3.37414294e-02
-6.56855345e-01 -1.32527351e-01 -1.17735445e+00 2.51892239e-01
9.77555871e-01 -3.68005514e-01 -4.42103416e-01 4.69861716e-01
9.11547720e-01 -5.59537113e-01 -4.36147720e-01 7.82579184e-01
4.42494571e-01 -6.29692018e-01 8.60164225e-01 -6.45313978e-01
2.02543125e-01 -5.81385553e-01 -1.75079536e-02 -1.12702155e+00
-5.78830279e-02 -1.97871700e-01 -1.55076489e-01 1.34102976e+00
5.78583956e-01 -3.32852691e-01 1.20852852e+00 1.43311322e-01
-2.88316816e-01 -8.86691272e-01 -5.58852792e-01 -1.00873923e+00
-1.87328253e-02 -3.87137115e-01 5.93829453e-02 1.07492745e+00
-4.69872504e-01 4.94534105e-01 -4.76024538e-01 2.08774149e-01
9.70714271e-01 -2.61019737e-01 8.29325438e-01 -1.16149545e+00
-1.19816557e-01 -1.41728863e-01 -1.39118016e-01 -1.39160037e+00
-1.27407964e-02 -8.53758454e-01 5.17645538e-01 -1.31419015e+00
4.48856324e-01 -7.24493384e-01 -4.78932887e-01 7.43700922e-01
-5.88537872e-01 8.85268748e-01 1.04482152e-01 3.93034220e-01
-1.13377738e+00 3.92131895e-01 1.02355695e+00 -3.09854031e-01
2.33837552e-02 -6.86650649e-02 -5.67170382e-01 8.25497448e-01
7.41392314e-01 -8.29332829e-01 -4.44809616e-01 -5.16256034e-01
-2.45866269e-01 -6.33238077e-01 2.82734305e-01 -1.09757125e+00
2.47875065e-01 1.36239663e-01 4.77902979e-01 -8.59215319e-01
4.07132469e-02 -5.59576392e-01 -3.78366560e-01 5.10283709e-01
-6.30085886e-01 -3.77181500e-01 2.95640439e-01 7.36428976e-01
-2.20398098e-01 -4.03171152e-01 1.08245850e+00 1.24763712e-01
-9.18413639e-01 2.03557581e-01 -4.87950481e-02 2.60484189e-01
1.12493587e+00 -1.64099392e-02 -5.10872781e-01 1.21281832e-01
-6.06396079e-01 5.90958118e-01 1.01202533e-01 2.89130896e-01
2.21230224e-01 -1.26232076e+00 -8.68490040e-01 -9.40325484e-02
4.90189254e-01 5.01701832e-01 -8.42068046e-02 7.30512142e-01
-3.87928218e-01 4.17573482e-01 7.56477714e-02 -1.03347242e+00
-1.40485346e+00 3.73846442e-01 2.10310265e-01 -1.96871534e-01
-3.06713909e-01 1.18105626e+00 3.35646957e-01 -7.34146833e-01
7.86730111e-01 -9.84878615e-02 -1.86547786e-01 1.11500412e-01
5.84434092e-01 5.29020965e-01 -2.92408448e-02 -4.58925545e-01
-3.27032685e-01 2.83410043e-01 -3.74847978e-01 1.60224840e-01
9.93099630e-01 1.00475624e-01 2.76820570e-01 3.48321915e-01
1.15329802e+00 -2.74838090e-01 -1.77543795e+00 -6.21265471e-01
1.36439994e-01 -2.27431640e-01 2.32950300e-01 -1.10396016e+00
-1.10996366e+00 9.19255197e-01 9.01248038e-01 -4.61675860e-02
7.36940563e-01 2.45332867e-01 3.87328416e-01 4.16826308e-01
2.38080636e-01 -9.62296128e-01 4.41426456e-01 3.56803000e-01
3.66609365e-01 -1.82811534e+00 5.22949062e-02 -5.64209342e-01
-5.96192300e-01 6.15426779e-01 1.13349855e+00 3.20070758e-02
5.00459850e-01 1.59959689e-01 2.64822811e-01 -6.42859489e-02
-5.71356058e-01 -3.81092846e-01 5.17863870e-01 4.95864719e-01
4.08823669e-01 4.83139306e-02 2.36754306e-02 3.99175614e-01
3.11989188e-01 -7.94075280e-02 1.89894080e-01 9.64181662e-01
-6.81014001e-01 -6.91746116e-01 -3.25355262e-01 5.99285901e-01
-6.67087376e-01 -2.17701957e-01 -3.45125705e-01 8.88603985e-01
4.13883120e-01 9.14111376e-01 1.22015215e-01 -2.79339612e-01
2.31066793e-01 -7.15922117e-02 9.56523493e-02 -1.00148034e+00
-3.83909255e-01 5.94081134e-02 1.98279340e-02 -2.93759882e-01
-6.18148625e-01 -4.71166015e-01 -1.25581837e+00 1.81883618e-01
-8.87619138e-01 1.20897122e-01 6.83974385e-01 8.08085918e-01
2.80032843e-01 5.86977899e-01 4.31066334e-01 -7.99414575e-01
-8.49758983e-01 -1.34984076e+00 -3.11131597e-01 1.65337548e-01
2.98022717e-01 -8.14771771e-01 -5.39356649e-01 6.68146014e-02]
|
[9.169909477233887, 1.2545608282089233]
|
5dd94511-fa36-42e2-81eb-7c89e654cf6f
|
spectral-variability-augmented-sparse
|
2110.09744
| null |
https://arxiv.org/abs/2110.09744v2
|
https://arxiv.org/pdf/2110.09744v2.pdf
|
Spectral Variability Augmented Sparse Unmixing of Hyperspectral Images
|
Spectral unmixing (SU) expresses the mixed pixels existed in hyperspectral images as the product of endmember and abundance, which has been widely used in hyperspectral imagery analysis. However, the influence of light, acquisition conditions and the inherent properties of materials, results in that the identified endmembers can vary spectrally within a given image (construed as spectral variability). To address this issue, recent methods usually use a priori obtained spectral library to represent multiple characteristic spectra of the same object, but few of them extracted the spectral variability explicitly. In this paper, a spectral variability augmented sparse unmixing model (SVASU) is proposed, in which the spectral variability is extracted for the first time. The variable spectra are divided into two parts of intrinsic spectrum and spectral variability for spectral reconstruction, and modeled synchronously in the SU model adding the regular terms restricting the sparsity of abundance and the generalization of the variability coefficient. It is noted that the spectral variability library and the intrinsic spectral library are all constructed from the In-situ observed image. Experimental results over both synthetic and real-world data sets demonstrate that the augmented decomposition by spectral variability significantly improves the unmixing performance than the decomposition only by spectral library, as well as compared to state-of-the-art algorithms.
|
['Qian Du', 'Yan Feng', 'Mingyang Ma', 'Shaohui Mei', 'Ge Zhang']
|
2021-10-19
| null | null | null | null |
['spectral-reconstruction']
|
['computer-vision']
|
[ 7.60542512e-01 -8.04897547e-01 -1.39620751e-01 1.56020239e-01
-1.88644648e-01 -6.74547315e-01 5.10887742e-01 -4.08310920e-01
9.84293222e-02 9.82198775e-01 1.32019460e-01 8.85593891e-02
-1.88396186e-01 -8.06243718e-01 -4.26872164e-01 -1.47853494e+00
2.43763775e-01 -2.69882791e-02 -3.05322170e-01 -1.12359367e-01
-3.11199188e-01 3.20492327e-01 -1.88323915e+00 1.83485091e-01
1.38246620e+00 9.94273424e-01 4.89484042e-01 -1.17908129e-02
-4.76333857e-01 4.82743621e-01 -2.86497563e-01 6.12329364e-01
6.36659265e-01 -6.24240160e-01 -1.69302270e-01 8.15070391e-01
3.30420941e-01 3.04652620e-02 -2.50822246e-01 1.70019901e+00
1.03967905e-01 2.20637828e-01 6.50006771e-01 -8.90031040e-01
-5.29511154e-01 5.93197405e-01 -1.15327811e+00 -1.19930506e-01
-2.42894143e-01 2.72215120e-02 5.68288147e-01 -7.84734011e-01
3.40350866e-01 1.02106941e+00 5.39150298e-01 -9.51740816e-02
-1.46395624e+00 -8.20624650e-01 6.80869306e-03 -1.24551319e-02
-1.63174617e+00 -2.72336185e-01 1.09882379e+00 -7.70948410e-01
3.60522181e-01 6.25599384e-01 9.59847152e-01 6.14625752e-01
-1.48887992e-01 3.69289488e-01 1.64970529e+00 -4.51124519e-01
-1.62167065e-02 1.79011766e-02 4.46339548e-01 3.40209901e-01
6.41184270e-01 5.38404584e-01 -3.93116623e-01 -3.30796927e-01
5.40003359e-01 3.00740480e-01 -9.73451734e-01 -5.50973833e-01
-1.21055675e+00 5.06872833e-01 3.57388914e-01 1.47455677e-01
-6.72017395e-01 -5.98524034e-01 6.86749890e-02 1.22777916e-01
6.02185607e-01 3.84497829e-03 -2.51885653e-01 5.87218523e-01
-1.18243778e+00 -4.98082601e-02 4.86636341e-01 6.69480324e-01
1.46231186e+00 5.60835540e-01 1.84826657e-01 1.12542403e+00
4.94800895e-01 1.25561786e+00 7.71465123e-01 -5.81297934e-01
1.08715557e-01 7.20990300e-01 1.89123616e-01 -8.38613093e-01
-2.50359088e-01 -7.84272492e-01 -1.28478265e+00 1.12885892e-01
-2.30956264e-03 -1.11544274e-01 -1.13844311e+00 1.59755373e+00
2.95964032e-01 4.51538682e-01 3.06606174e-01 9.42462206e-01
7.79574871e-01 1.10085833e+00 3.19036506e-02 -1.05822062e+00
1.02572644e+00 -8.29090416e-01 -1.10158062e+00 -3.56683552e-01
6.07905127e-02 -7.75232196e-01 5.96026957e-01 3.32645923e-01
-4.10545379e-01 -5.06357014e-01 -1.30161595e+00 5.75892448e-01
-3.14265221e-01 2.33854532e-01 6.13663375e-01 6.51671588e-01
-4.16471839e-01 3.47901791e-01 -7.16268241e-01 -1.81825504e-01
1.26882270e-01 -3.38518433e-02 -3.30028534e-01 1.46930700e-03
-1.08922505e+00 6.61582649e-01 8.23505640e-01 3.94394487e-01
-7.19663143e-01 -5.34708202e-01 -7.57103980e-01 -7.25009516e-02
3.67719740e-01 -4.62104052e-01 4.89595771e-01 -1.68777192e+00
-1.30508065e+00 5.64036787e-01 -3.11000496e-01 7.26488084e-02
-8.13781247e-02 1.66123152e-01 -9.07158315e-01 8.18146020e-02
2.36534998e-02 -2.92545855e-01 1.08985615e+00 -1.69442546e+00
-3.20698291e-01 -5.18872678e-01 -5.66450417e-01 2.40242630e-01
-2.99234360e-01 -3.60783488e-01 9.58133787e-02 -6.60694242e-01
8.69187057e-01 -9.44974184e-01 -8.16101730e-02 -2.62800127e-01
-1.08190209e-01 7.17420697e-01 1.06785107e+00 -8.39466572e-01
1.18907213e+00 -2.43389297e+00 2.46802166e-01 5.20025611e-01
-1.39357269e-01 4.00278449e-01 -2.26145729e-01 3.37231338e-01
-5.36040604e-01 -2.09574461e-01 -1.06692708e+00 2.79358566e-01
-5.52895606e-01 4.11411554e-01 -3.74315947e-01 8.39231312e-01
-2.15552628e-01 3.17195952e-01 -1.01239014e+00 -2.23939225e-01
3.48934293e-01 4.52425122e-01 1.97598279e-01 1.48699820e-01
-2.09068939e-01 5.15649974e-01 -2.75849789e-01 8.23887408e-01
1.25367868e+00 -1.44714594e-01 3.69084865e-01 -6.65583909e-01
-4.47217405e-01 -3.88646394e-01 -1.61123884e+00 1.53759861e+00
-9.79900658e-02 1.41248301e-01 5.69178462e-01 -9.73580599e-01
8.01601410e-01 3.72283071e-01 6.34825289e-01 -1.01613335e-01
-1.18360013e-01 5.24504364e-01 7.24096000e-02 -5.02611518e-01
3.29675227e-01 -4.93759215e-01 8.96547377e-01 1.51649401e-01
3.18824849e-03 -1.18681870e-01 2.12512568e-01 -2.82365113e-01
1.10496186e-01 2.72856504e-01 6.86721504e-01 -7.17063487e-01
9.82307315e-01 2.80212104e-01 6.91214204e-01 2.60816574e-01
7.25963563e-02 3.53038043e-01 -2.31061175e-01 -1.10278934e-01
-8.24282348e-01 -9.91490185e-01 -4.98581111e-01 5.57116687e-01
3.29812050e-01 1.88826516e-01 -4.02980715e-01 -2.64819618e-02
2.62398360e-04 4.60202903e-01 -3.91719878e-01 3.48475762e-02
2.97390874e-02 -1.59735739e+00 1.41314268e-01 -2.62274761e-02
8.95865798e-01 -5.81986964e-01 -3.28068063e-02 1.86861038e-01
-3.13003898e-01 -6.73472762e-01 -1.79509044e-01 7.81637207e-02
-1.03556573e+00 -1.18094158e+00 -4.55064088e-01 -4.70037073e-01
6.95666432e-01 8.85509610e-01 5.77619433e-01 -2.39689827e-01
-1.94549203e-01 1.51726723e-01 -4.83127534e-01 -3.32804590e-01
-4.72848654e-01 -4.75077599e-01 1.27957538e-01 8.35584760e-01
1.55920431e-01 -8.62628102e-01 -1.68377295e-01 1.38839126e-01
-1.30625403e+00 2.25710779e-01 6.40679717e-01 1.12303424e+00
8.98457944e-01 4.08568978e-01 1.54545963e-01 -8.57896626e-01
1.32563680e-01 -6.14123225e-01 -6.57352328e-01 4.07535493e-01
-7.49468803e-01 -4.00296226e-02 3.23595464e-01 -4.41536903e-01
-1.45661795e+00 3.70358467e-01 5.50294042e-01 -4.28511798e-01
-9.28921252e-02 1.16240478e+00 -5.76923013e-01 -3.27940255e-01
8.48733604e-01 9.76816952e-01 4.82324094e-01 -5.27947664e-01
2.18806475e-01 8.80362272e-01 5.71391523e-01 -6.10335648e-01
1.04817033e+00 8.68641794e-01 2.76893407e-01 -1.49095356e+00
-9.16927397e-01 -8.67508471e-01 -4.83004332e-01 -1.05363734e-01
5.09677827e-01 -1.37691057e+00 7.16321692e-02 7.38553464e-01
-7.95505881e-01 7.33263791e-02 -1.68886408e-01 7.00542510e-01
-1.66905567e-01 8.96643043e-01 -1.52834311e-01 -9.86627102e-01
-3.42054993e-01 -9.45324957e-01 4.56559241e-01 1.02985822e-01
4.17627960e-01 -7.57587790e-01 4.96051982e-02 1.98315620e-01
3.26247394e-01 4.57673639e-01 8.11117470e-01 -7.86576327e-03
-4.94391710e-01 1.02373160e-01 -3.84766340e-01 7.70573854e-01
7.37391055e-01 2.23921031e-01 -1.06886363e+00 -3.04977417e-01
5.00312865e-01 1.19222797e-01 1.06949794e+00 6.07144058e-01
8.71820927e-01 -1.71069309e-01 -1.85538381e-01 9.49576497e-01
1.92977941e+00 2.92926997e-01 5.39644659e-01 2.77397335e-01
8.28341007e-01 6.22161269e-01 3.41497183e-01 3.97079617e-01
-4.15095866e-01 7.84761161e-02 5.89999974e-01 -1.09316446e-01
9.91321504e-02 9.62658599e-02 3.35031897e-01 8.59586298e-01
-4.24876183e-01 -4.74134125e-02 -7.63099194e-01 3.78927171e-01
-2.05345798e+00 -1.40672195e+00 -6.87746465e-01 2.40961242e+00
7.55008876e-01 -7.20294118e-01 -2.86019385e-01 2.61106104e-01
1.02773452e+00 6.43755436e-01 -4.69065130e-01 5.48253179e-01
-9.03358042e-01 -1.24174682e-02 9.56562281e-01 5.82996011e-01
-1.09084749e+00 5.19047618e-01 5.93251848e+00 7.63705075e-01
-1.27891386e+00 2.02665657e-01 1.16986729e-01 1.91578254e-01
-4.88686055e-01 1.66006148e-01 -2.35700816e-01 4.41669971e-01
5.48324406e-01 -2.65224934e-01 9.45223451e-01 3.87747049e-01
3.74479324e-01 -1.32126123e-01 -3.53209555e-01 1.23092401e+00
1.06296554e-01 -8.90068948e-01 3.46614361e-01 3.23064998e-02
1.12932420e+00 1.29479617e-01 -1.07550785e-01 -1.67391166e-01
-1.06456034e-01 -6.45674706e-01 6.66910648e-01 6.56520545e-01
7.40216851e-01 -3.50421757e-01 5.26574969e-01 5.32644689e-01
-1.32151866e+00 -8.51962194e-02 -4.72882271e-01 -1.40707076e-01
-4.88827080e-02 1.11224711e+00 -7.04804584e-02 1.14535737e+00
3.54159743e-01 1.08653522e+00 -1.54886872e-01 8.84399712e-01
-1.45182788e-01 8.06272447e-01 -4.21454996e-01 5.27679503e-01
1.72802418e-01 -1.13289070e+00 7.79684007e-01 9.28851068e-01
5.77237546e-01 4.56245750e-01 3.30432624e-01 1.14273524e+00
4.05068666e-01 2.92121202e-01 -3.91384542e-01 -5.77073276e-01
3.59269530e-01 1.31175864e+00 -2.26012975e-01 -3.79208863e-01
-5.08452177e-01 6.90342486e-01 -4.69465524e-01 8.50658834e-01
-4.01367337e-01 2.36732185e-01 6.99657500e-01 6.95425719e-02
-5.28904283e-03 -2.87295789e-01 -3.38689208e-01 -1.52662790e+00
-1.02299511e-01 -9.96269584e-01 2.51464427e-01 -8.57197583e-01
-1.40129399e+00 4.34320331e-01 3.12554240e-02 -1.62975812e+00
2.89331734e-01 -7.06470668e-01 -2.08623841e-01 1.47126317e+00
-1.70907676e+00 -1.32417428e+00 -7.04929769e-01 6.50572002e-01
1.70712888e-01 -3.18828911e-01 9.09985721e-01 7.84494579e-02
-7.19261408e-01 -4.95582789e-01 9.46450174e-01 -2.90925056e-01
5.07141113e-01 -7.83692956e-01 -6.51903749e-01 1.09453380e+00
-3.14090215e-02 7.10211635e-01 7.47692466e-01 -7.99450755e-01
-1.31548846e+00 -1.19817734e+00 1.76003262e-01 3.75062257e-01
9.17354107e-01 3.48650366e-01 -1.06649733e+00 4.79905874e-01
1.18555516e-01 1.99451819e-01 1.08855140e+00 -1.30554840e-01
-4.20845985e-01 -3.74120474e-01 -1.05244863e+00 2.68129259e-01
6.02709293e-01 -5.50471604e-01 -3.79778981e-01 4.39167738e-01
4.24285948e-01 -1.41891390e-01 -7.51240492e-01 5.42023540e-01
5.17920971e-01 -8.15088391e-01 9.68995273e-01 -1.82981640e-01
2.93900907e-01 -9.45959270e-01 -5.39008498e-01 -1.39481068e+00
-7.21593916e-01 -1.59099892e-01 -9.52040404e-02 1.09744573e+00
2.32021824e-01 -8.58776927e-01 2.98970491e-01 1.15861505e-01
-1.23812795e-01 1.35833696e-01 -5.97930193e-01 -1.00083590e+00
-3.09666604e-01 7.54294693e-02 7.68831968e-01 1.43428648e+00
-2.14894935e-01 1.69463247e-01 -5.45420229e-01 8.11918914e-01
1.01966524e+00 4.84316677e-01 3.08211416e-01 -1.68215406e+00
-2.25023657e-01 -3.66980731e-01 9.21039376e-03 -5.14629900e-01
4.28012669e-01 -8.42152357e-01 -2.99205482e-02 -1.28022861e+00
5.09930015e-01 -8.77414346e-02 -4.62321252e-01 2.99585313e-01
-4.44115639e-01 1.36556877e-02 -1.06426932e-01 7.78451383e-01
5.55772126e-01 6.43942416e-01 1.22806430e+00 -6.73694372e-01
-4.04009640e-01 -1.71005607e-01 -5.25078177e-01 5.82592130e-01
6.25859261e-01 -1.97287843e-01 -5.02106011e-01 -2.52920270e-01
-7.29239685e-03 5.81767075e-02 3.21257293e-01 -1.05188704e+00
-1.34856790e-01 -7.12105334e-01 7.35439882e-02 -5.78738391e-01
3.09857190e-01 -1.31029320e+00 9.97682750e-01 3.36841404e-01
2.22239375e-01 -8.23905051e-01 3.02464336e-01 6.59118116e-01
-5.39316475e-01 -3.59835327e-01 8.54723454e-01 -3.41447532e-01
-9.33367372e-01 4.91468281e-01 -2.33899027e-01 -5.84651530e-01
6.71587288e-01 -3.83419663e-01 -3.73796940e-01 -1.20729450e-02
-6.17925525e-01 -1.74594596e-01 4.80518878e-01 -2.23396868e-01
2.70456254e-01 -1.25719655e+00 -8.81582558e-01 4.91332710e-01
2.68553883e-01 -9.57033597e-03 6.01271331e-01 7.23882556e-01
-5.80584466e-01 -3.19803469e-02 -3.15352589e-01 -5.55420637e-01
-1.06927168e+00 7.82552779e-01 6.44637287e-01 1.13635316e-01
-3.04282755e-02 3.98384959e-01 3.07770073e-01 -3.59767884e-01
-4.89231318e-01 4.94581461e-02 -2.71415085e-01 3.32774192e-01
5.54083645e-01 5.11408210e-01 -1.37401953e-01 -1.15747440e+00
-1.13121830e-02 6.18562281e-01 6.48989379e-01 2.91650668e-02
1.27365649e+00 -2.70496964e-01 -1.09501755e+00 7.23111331e-01
1.03998697e+00 1.18908703e-01 -9.98121023e-01 -5.72894335e-01
-5.52733183e-01 -5.77328324e-01 4.54346836e-01 -8.36193919e-01
-1.10037422e+00 5.58022380e-01 8.12598526e-01 2.18509898e-01
1.46214724e+00 -8.05990934e-01 1.41123682e-01 2.01817930e-01
3.97802860e-01 -8.49093854e-01 -7.71240234e-01 2.84040958e-01
7.80262232e-01 -1.21509326e+00 4.23947513e-01 -1.00061393e+00
-3.20880592e-01 1.07222986e+00 4.03906822e-01 1.53044283e-01
8.17892611e-01 -1.96972843e-02 2.21427351e-01 2.92665251e-02
6.73557147e-02 -4.27384973e-01 3.95028055e-01 5.56114256e-01
3.98878008e-01 4.10360754e-01 -5.04235685e-01 4.09624070e-01
1.91450492e-01 -1.88912675e-01 4.17566359e-01 6.40991986e-01
-7.39581466e-01 -9.29819345e-01 -1.10138142e+00 5.07884085e-01
-2.24108398e-02 -3.35000455e-01 -9.81682613e-02 2.97888726e-01
3.96334738e-01 9.64673698e-01 -2.68091887e-01 -2.94387281e-01
1.01843305e-01 3.10689747e-01 2.22890824e-01 -6.58102810e-01
-5.28952815e-02 5.13077915e-01 -2.70525932e-01 -1.69373408e-01
-1.24090207e+00 -7.20163405e-01 -9.13691878e-01 -1.42040837e-04
-7.17069745e-01 2.60843128e-01 5.30923903e-01 7.97379017e-01
-1.68148607e-01 3.45351964e-01 6.73478723e-01 -1.01914680e+00
-6.75795019e-01 -1.17046702e+00 -1.62008631e+00 3.83176893e-01
4.56245691e-01 -7.07898021e-01 -7.56352067e-01 4.26614612e-01]
|
[10.077312469482422, -2.045492172241211]
|
d09ec51c-e417-4bd3-9e23-359c4b732ee6
|
the-dirha-english-corpus-and-related-tasks
|
1710.02560
| null |
http://arxiv.org/abs/1710.02560v1
|
http://arxiv.org/pdf/1710.02560v1.pdf
|
The DIRHA-English corpus and related tasks for distant-speech recognition in domestic environments
|
This paper introduces the contents and the possible usage of the
DIRHA-ENGLISH multi-microphone corpus, recently realized under the EC DIRHA
project. The reference scenario is a domestic environment equipped with a large
number of microphones and microphone arrays distributed in space.
The corpus is composed of both real and simulated material, and it includes
12 US and 12 UK English native speakers. Each speaker uttered different sets of
phonetically-rich sentences, newspaper articles, conversational speech,
keywords, and commands. From this material, a large set of 1-minute sequences
was generated, which also includes typical domestic background noise as well as
inter/intra-room reverberation effects. Dev and test sets were derived, which
represent a very precious material for different studies on multi-microphone
speech processing and distant-speech recognition. Various tasks and
corresponding Kaldi recipes have already been developed.
The paper reports a first set of baseline results obtained using different
techniques, including Deep Neural Networks (DNN), aligned with the
state-of-the-art at international level.
|
['Mirco Ravanelli', 'Maurizio Omologo']
|
2017-10-06
| null | null | null | null |
['distant-speech-recognition']
|
['speech']
|
[ 1.90394018e-02 -3.71126145e-01 6.13011181e-01 -6.43283606e-01
-1.10016823e+00 -5.70180655e-01 6.39085114e-01 -2.87778050e-01
-6.69963181e-01 6.65206432e-01 5.08743465e-01 -3.85730922e-01
1.34032533e-01 -3.51014078e-01 -5.36748707e-01 -8.11115563e-01
-1.05209455e-01 4.28290486e-01 -1.25565737e-01 -2.19577238e-01
-1.12058163e-01 4.74632978e-01 -1.76584685e+00 5.48219383e-01
3.10443729e-01 7.25784600e-01 7.16153622e-01 1.27782500e+00
1.62283912e-01 7.39453912e-01 -1.12369609e+00 -3.36925715e-01
7.92944729e-02 -3.28152478e-01 -3.71910274e-01 -2.24467754e-01
5.90294302e-01 -2.34524131e-01 -2.06280351e-01 9.97038543e-01
1.44783926e+00 4.24360007e-01 3.11640710e-01 -6.57403886e-01
-3.94504815e-01 9.67346728e-01 8.12453926e-02 7.20702708e-01
3.27691972e-01 -1.24900937e-02 8.12527955e-01 -1.08071053e+00
3.08152586e-01 1.24459469e+00 6.06908023e-01 6.56327188e-01
-7.58477807e-01 -5.86809993e-01 -2.04770193e-01 3.25603694e-01
-1.34918725e+00 -1.13135147e+00 8.03275943e-01 -2.86312312e-01
1.41085899e+00 8.60113144e-01 3.87581617e-01 1.84089124e+00
-2.13528275e-01 7.30683148e-01 1.00570512e+00 -7.95065641e-01
2.89497733e-01 4.94439334e-01 7.26150349e-02 -2.56393105e-01
-5.72233498e-01 2.47333810e-01 -5.14559984e-01 -1.26683414e-01
2.94868737e-01 -5.78886151e-01 -5.94808638e-01 2.19481647e-01
-1.22165966e+00 4.81181711e-01 -2.46691599e-01 9.75401700e-01
-5.88025689e-01 -4.09521163e-01 6.44863904e-01 3.99753451e-01
6.00627542e-01 -1.11176580e-01 -7.23256588e-01 -6.18183076e-01
-8.12071621e-01 2.30918065e-01 1.06637323e+00 8.81067216e-01
1.80577487e-01 4.38536465e-01 2.33417422e-01 1.74796367e+00
2.97512829e-01 7.37048686e-01 7.75253594e-01 -8.95078003e-01
9.03869510e-01 -6.35348558e-01 -3.43356691e-02 -7.73304224e-01
-2.64721751e-01 -4.15580392e-01 -1.01594484e+00 -8.59256610e-02
2.29102999e-01 -5.90432584e-01 -6.90107703e-01 1.63959372e+00
2.26139188e-01 3.51116695e-02 2.86863506e-01 6.82871163e-01
9.87572849e-01 1.17805076e+00 -4.91703033e-01 -5.35277247e-01
1.03950667e+00 -1.26387131e+00 -1.15730608e+00 -2.21882507e-01
1.45030126e-01 -1.10897934e+00 1.09853888e+00 7.62561023e-01
-1.34868956e+00 -1.00350654e+00 -8.10460150e-01 2.21589640e-01
-5.00435412e-01 7.06282780e-02 -1.62055284e-01 1.18273664e+00
-1.49438369e+00 2.28954017e-01 -5.66577673e-01 -1.78450882e-01
-2.42432907e-01 -2.26363074e-02 -3.67799252e-01 2.23650951e-02
-1.48165464e+00 8.25411558e-01 1.94555208e-01 6.01355791e-01
-9.13365901e-01 -4.66111898e-01 -8.07764709e-01 -2.09594667e-01
-2.91562919e-02 -1.60091579e-01 1.89472306e+00 -8.79219472e-01
-2.00393319e+00 8.54370952e-01 -2.62681335e-01 -4.81520891e-01
4.26599890e-01 -4.51103717e-01 -1.18371522e+00 -1.36057764e-01
-2.11569980e-01 1.75600305e-01 7.19762266e-01 -1.04389465e+00
-5.77205658e-01 -3.13472420e-01 -3.90931517e-01 4.28155541e-01
-1.51072562e-01 8.07191014e-01 -2.54215509e-01 -8.89407456e-01
-4.31206733e-01 -6.35162890e-01 -1.82371259e-01 -9.45107222e-01
-6.27994835e-01 5.39122112e-02 4.59112376e-01 -1.18416560e+00
1.24383485e+00 -2.50535870e+00 1.22558765e-01 4.19537863e-03
-3.49254787e-01 5.10495961e-01 -2.26593718e-01 5.32763779e-01
-3.62840593e-01 -9.15895179e-02 -9.69198868e-02 -8.21652293e-01
2.06308842e-01 1.97549060e-01 -3.24156493e-01 4.04869229e-01
-4.81650740e-01 2.82285571e-01 -5.00419974e-01 -1.23422807e-02
6.58321798e-01 6.86111152e-01 -2.00836241e-01 3.28950852e-01
2.37363309e-01 4.99041587e-01 2.45853454e-01 3.91719580e-01
7.74281979e-01 7.24618077e-01 -2.19894692e-01 1.69661343e-01
-4.52814698e-01 8.64258945e-01 -1.44482636e+00 1.66564190e+00
-8.94967794e-01 1.01426446e+00 8.64797056e-01 -9.37906742e-01
8.02060187e-01 8.55721414e-01 -8.58824626e-02 -6.23197913e-01
1.61460310e-01 6.06922448e-01 7.46757388e-02 -6.16718709e-01
6.66643500e-01 -8.30039456e-02 2.35716283e-01 2.85002161e-02
1.05788335e-01 -2.05945581e-01 1.46578297e-01 -4.30965275e-01
9.94429350e-01 -5.24138153e-01 1.33375570e-01 -2.44500443e-01
7.96480477e-01 -5.15951693e-01 4.49649245e-01 7.79576421e-01
-3.55038673e-01 9.04806077e-01 -2.49823079e-01 -1.01966538e-01
-1.05773711e+00 -1.20901060e+00 -2.81448513e-01 1.19770360e+00
-6.19841635e-01 -2.22801521e-01 -1.05944586e+00 1.30933940e-01
-7.15690494e-01 1.02439964e+00 -5.97714446e-02 5.17776847e-01
-9.27605689e-01 -6.02421641e-01 7.43964314e-01 3.29746455e-01
3.93797874e-01 -1.54705656e+00 -4.00160328e-02 4.60584044e-01
-4.61811125e-01 -1.32777154e+00 -4.32987779e-01 3.65949005e-01
-2.81892061e-01 -4.66648549e-01 -9.72263277e-01 -9.87932324e-01
-1.81861669e-01 8.52806196e-02 1.35697615e+00 -7.85550416e-01
-5.67480549e-02 2.23100215e-01 -4.59829479e-01 -5.33142745e-01
-1.17045331e+00 -2.55676419e-01 4.44019169e-01 3.57366726e-02
4.58966225e-01 -7.90958822e-01 -1.74977973e-01 2.49760032e-01
-5.12704611e-01 -2.60493606e-01 3.09488446e-01 6.84278071e-01
4.22900409e-01 3.37659836e-01 4.78138238e-01 -5.56665003e-01
8.00577641e-01 -3.08006346e-01 -4.62248832e-01 -1.79223761e-01
2.60732293e-01 -8.15753996e-01 8.64531040e-01 -4.61126119e-01
-1.26080382e+00 -3.36793125e-01 -1.08747029e+00 -2.71059752e-01
-1.00945115e+00 2.77900487e-01 -8.22239637e-01 6.23702705e-01
7.12693334e-01 3.42162520e-01 -4.42890316e-01 -1.05436814e+00
4.00431305e-01 1.78727865e+00 8.79819930e-01 -3.19228441e-01
2.57970124e-01 -5.36876395e-02 -9.93191779e-01 -1.54682398e+00
-1.81365743e-01 -7.46989191e-01 -5.23966730e-01 -1.50123969e-01
7.87778199e-01 -1.25278270e+00 -1.29084602e-01 1.30915725e+00
-1.56365681e+00 -5.47573984e-01 -3.13762397e-01 7.22156525e-01
-4.47410434e-01 1.30212709e-01 -1.01322281e+00 -1.00730419e+00
-3.32176149e-01 -1.26386487e+00 9.42965627e-01 -1.62364706e-01
-1.98794618e-01 -1.00310564e+00 4.29174453e-01 5.62099040e-01
6.46221519e-01 -2.75126666e-01 3.29965591e-01 -8.94683063e-01
3.78901698e-02 -5.87149188e-02 5.35113156e-01 1.36793423e+00
2.46858522e-01 1.18090138e-02 -1.72495949e+00 -2.06188619e-01
6.25951290e-01 -8.14240053e-02 3.54295224e-01 6.67401254e-01
9.81663227e-01 -2.43238464e-01 2.21785888e-01 2.21816510e-01
9.49355662e-01 6.65732801e-01 6.64402723e-01 6.46028221e-02
6.37330532e-01 8.08592260e-01 2.62946576e-01 2.63089627e-01
2.39674851e-01 7.54282713e-01 1.17674261e-01 2.53611840e-02
-1.70947567e-01 3.68404537e-01 8.56695712e-01 1.95001173e+00
1.63973883e-01 -6.57671332e-01 -1.02358568e+00 8.38708103e-01
-1.09363544e+00 -1.17106843e+00 -2.80585706e-01 2.07737398e+00
8.47937107e-01 1.55175729e-02 2.15891704e-01 5.53375304e-01
1.09058726e+00 4.64690775e-01 -8.91192034e-02 -8.99496436e-01
-5.20642579e-01 4.13544774e-01 -6.52679577e-02 6.89617395e-01
-1.17693615e+00 4.81350154e-01 6.65118694e+00 1.24159527e+00
-1.06753099e+00 4.33057994e-01 5.69133461e-01 -2.26992145e-01
-2.23865733e-02 -8.24962735e-01 -8.16333771e-01 6.11009181e-01
2.06449008e+00 1.58673346e-01 6.78180397e-01 8.55947435e-01
6.72012031e-01 5.40638603e-02 -1.08246171e+00 1.18354011e+00
2.27410346e-01 -1.14631665e+00 -4.23271447e-01 -1.67054236e-02
6.04337037e-01 8.42395902e-01 -2.83661149e-02 2.81280279e-01
6.62558898e-02 -8.36896598e-01 9.94718790e-01 1.58699870e-01
7.08697200e-01 -6.68030620e-01 8.13987970e-01 6.58788741e-01
-9.82229590e-01 -1.85928084e-02 -2.38455355e-01 -1.89733788e-01
3.93235415e-01 9.24572289e-01 -5.69963038e-01 5.86466908e-01
9.44806337e-01 5.54798543e-01 -3.22298035e-02 8.30239594e-01
-4.07596678e-02 9.50100362e-01 -5.06363988e-01 -1.25806391e-01
1.27880394e-01 -1.94434240e-01 8.71493638e-01 1.83114910e+00
4.99270976e-01 -3.22262794e-01 -3.55529904e-01 4.25421953e-01
6.75266087e-02 3.40019584e-01 -7.55133748e-01 1.04770713e-01
6.99627936e-01 1.11742294e+00 -2.01688945e-01 -4.22039032e-01
-5.72201848e-01 9.59105790e-01 -1.84690475e-01 5.58827221e-01
-6.88079953e-01 -5.22091568e-01 9.21973407e-01 -3.48251849e-01
3.46377939e-01 -1.95789650e-01 -1.73000582e-02 -9.03888166e-01
4.11317974e-01 -1.29012895e+00 -1.72693044e-01 -9.30931926e-01
-1.36775601e+00 1.07238746e+00 -1.25800341e-01 -1.05345309e+00
-6.46072268e-01 -7.58003950e-01 -5.66350520e-01 1.41116309e+00
-1.09696925e+00 -5.17287672e-01 6.90716207e-02 4.09580171e-01
1.16180634e+00 -5.73987663e-01 1.14655733e+00 1.02165699e+00
-6.05930090e-01 4.49363530e-01 7.01463044e-01 1.60198286e-01
5.05203307e-01 -1.30121291e+00 8.82243812e-01 8.63908827e-01
4.74833429e-01 4.95743126e-01 8.78390551e-01 1.76236078e-01
-8.51828158e-01 -9.86301541e-01 1.22269952e+00 -5.06118000e-01
7.15395868e-01 -7.96016276e-01 -7.98640788e-01 5.90363979e-01
6.79732025e-01 -1.77325606e-01 8.14710259e-01 -6.67437911e-02
1.13105841e-01 -2.73428470e-01 -8.14639211e-01 4.68896061e-01
8.35000098e-01 -8.71786773e-01 -8.35144281e-01 3.38304937e-01
8.53697598e-01 -7.18270183e-01 -5.02471745e-01 1.74376369e-01
2.51148969e-01 -1.42447495e+00 9.29265738e-01 -8.27835407e-03
8.71049911e-02 -1.67329982e-02 -7.86164105e-01 -1.78237736e+00
4.38232161e-02 -8.07305932e-01 4.00406905e-02 1.62453377e+00
4.87199783e-01 -5.39728940e-01 2.20039964e-01 1.32045612e-01
-7.09401369e-01 -2.68665195e-01 -1.45906091e+00 -8.71882200e-01
2.49861628e-01 -1.12674761e+00 5.16114354e-01 7.85051286e-01
-5.21315694e-01 5.30132532e-01 -4.53124404e-01 2.47457713e-01
2.31334403e-01 -5.81120431e-01 7.71931827e-01 -8.08942437e-01
-5.56837201e-01 -5.13412178e-01 -1.70399919e-01 -1.21805859e+00
1.89100474e-01 -4.10901308e-01 3.54081362e-01 -1.21821702e+00
-5.33326089e-01 -1.93765983e-01 -1.88685134e-01 -7.04082102e-02
1.81425691e-01 -3.65584567e-02 -1.00335004e-02 -3.03874075e-01
-5.66188321e-02 4.18730170e-01 7.25918353e-01 -1.17575206e-01
-1.92555621e-01 4.52051729e-01 -1.44589260e-01 7.73537576e-01
8.83784711e-01 -2.45093688e-01 -3.95610422e-01 -6.08743787e-01
-1.63766459e-01 1.83717132e-01 1.39515504e-01 -1.21142519e+00
7.91463330e-02 3.49653751e-01 -2.04356328e-01 -9.02144551e-01
8.65234017e-01 -7.98298419e-01 5.72039723e-01 -3.11569460e-02
-2.55768061e-01 1.94429323e-01 3.92409861e-01 3.20461810e-01
-7.42878854e-01 -2.09464177e-01 7.10151255e-01 -1.96414188e-01
-6.71365499e-01 -2.87499189e-01 -7.06919789e-01 9.81361195e-02
5.37261546e-01 -7.82367587e-02 -7.70232975e-02 -4.51930374e-01
-9.49660897e-01 -3.48824054e-01 -1.07724190e-01 7.16984034e-01
4.20303494e-01 -1.26882172e+00 -9.71529663e-01 5.65176070e-01
-3.48942548e-01 -3.11208405e-02 5.96814573e-01 6.29763186e-01
-4.78776872e-01 5.51430941e-01 9.11606029e-02 -5.45588374e-01
-1.48176205e+00 1.88788414e-01 6.09427392e-01 1.21852838e-01
-5.05754948e-01 1.14898407e+00 -3.04221436e-02 -9.59712327e-01
6.36923313e-01 -4.48163599e-01 -2.73429751e-01 5.67732304e-02
8.09867561e-01 4.61715072e-01 5.12603700e-01 -1.00097561e+00
-4.22228813e-01 2.09600851e-01 2.53799021e-01 -6.75559282e-01
1.18323517e+00 -3.79029483e-01 -2.44039327e-01 1.24605191e+00
1.42763269e+00 5.90762019e-01 -5.40707350e-01 -9.70046073e-02
-2.04282060e-01 -2.96359748e-01 7.41005596e-03 -8.79655957e-01
-8.17172587e-01 1.24621916e+00 6.57350898e-01 5.63084483e-01
1.16180933e+00 -2.81687200e-01 8.42511892e-01 3.67292404e-01
3.97261173e-01 -1.25084376e+00 -4.21531290e-01 7.89740980e-01
1.10582304e+00 -1.06326497e+00 -9.45947409e-01 -1.05546042e-01
-4.42946017e-01 8.67781341e-01 2.56891906e-01 3.82461280e-01
8.83759797e-01 5.90951204e-01 8.04506361e-01 3.66218597e-01
-6.20865166e-01 1.50719732e-01 -1.32150471e-01 8.30141425e-01
6.16996884e-01 2.63528585e-01 1.27994299e-01 6.44791484e-01
-7.89017737e-01 -4.69654322e-01 4.58267272e-01 5.90718687e-01
-2.49276057e-01 -1.09989655e+00 -7.44977295e-01 2.47606784e-02
-8.87932122e-01 -4.25080746e-01 -1.57056272e-01 4.96932000e-01
1.81126386e-01 1.40516579e+00 9.97571275e-02 -4.51575816e-01
5.97469985e-01 1.08399212e-01 8.46945040e-04 -5.58640540e-01
-7.55253196e-01 3.35936219e-01 6.69006050e-01 -2.02276528e-01
-5.01328945e-01 -7.96749949e-01 -6.96728885e-01 -3.37343723e-01
-2.72642136e-01 4.14204627e-01 1.09265471e+00 7.56301582e-01
8.16430002e-02 8.81735921e-01 7.00413644e-01 -1.29486907e+00
-5.14439285e-01 -1.59877682e+00 -9.13534939e-01 2.13870868e-01
6.79166138e-01 -2.56446134e-02 -8.01273406e-01 5.95609099e-02]
|
[14.879437446594238, 6.086802959442139]
|
24e352d9-7d64-45ee-83fb-8ded43833acb
|
eranns-efficient-residual-audio-neural
|
2106.01621
| null |
https://arxiv.org/abs/2106.01621v7
|
https://arxiv.org/pdf/2106.01621v7.pdf
|
ERANNs: Efficient Residual Audio Neural Networks for Audio Pattern Recognition
|
Audio pattern recognition (APR) is an important research topic and can be applied to several fields related to our lives. Therefore, accurate and efficient APR systems need to be developed as they are useful in real applications. In this paper, we propose a new convolutional neural network (CNN) architecture and a method for improving the inference speed of CNN-based systems for APR tasks. Moreover, using the proposed method, we can improve the performance of our systems, as confirmed in experiments conducted on four audio datasets. In addition, we investigate the impact of data augmentation techniques and transfer learning on the performance of our systems. Our best system achieves a mean average precision (mAP) of 0.450 on the AudioSet dataset. Although this value is less than that of the state-of-the-art system, the proposed system is 7.1x faster and 9.7x smaller. On the ESC-50, UrbanSound8K, and RAVDESS datasets, we obtain state-of-the-art results with accuracies of 0.961, 0.908, and 0.748, respectively. Our system for the ESC-50 dataset is 1.7x faster and 2.3x smaller than the previous best system. For the RAVDESS dataset, our system is 3.3x smaller than the previous best system. We name our systems "Efficient Residual Audio Neural Networks".
|
[]
|
2021-06-03
|
eranns-efficient-residual-audio-neural-1
|
https://arxiv.org/abs/2106.01621
|
https://arxiv.org/abs/2106.01621
| null |
['audio-tagging']
|
['audio']
|
[-9.43420753e-02 -3.97245288e-01 1.05681933e-01 -2.69170761e-01
-8.55606437e-01 -1.34125218e-01 -1.38847485e-01 2.13713013e-02
-8.04867744e-01 5.68187475e-01 -2.06094146e-01 -2.74873465e-01
1.04814358e-01 -7.68125534e-01 -8.05081964e-01 -3.65657836e-01
-1.62281901e-01 -9.02793035e-02 4.43322808e-01 -1.68466151e-01
6.20541349e-02 3.77435088e-01 -1.66988480e+00 5.95162034e-01
5.34731150e-01 1.56529033e+00 -1.47847906e-01 8.19298148e-01
2.86229104e-01 6.23615086e-01 -8.44557941e-01 -3.28325778e-01
9.97329205e-02 -4.40083556e-02 -6.05591595e-01 -4.69978601e-01
3.80217016e-01 -2.67215878e-01 -4.90988135e-01 8.35251093e-01
8.23916852e-01 3.41758654e-02 2.83167779e-01 -1.32068896e+00
-4.28582668e-01 7.95072794e-01 -7.10978448e-01 3.22757810e-01
9.93378088e-03 -1.02677181e-01 8.34761739e-01 -8.18742633e-01
-1.67251661e-01 9.99361932e-01 8.96502852e-01 3.63639891e-01
-7.98310399e-01 -1.31776500e+00 -1.42004460e-01 5.07029414e-01
-1.97627652e+00 -6.35946393e-01 6.65994942e-01 -1.74320653e-01
8.33920121e-01 1.39149353e-01 5.35411716e-01 8.85540724e-01
1.33878678e-01 7.38840580e-01 6.29968643e-01 -4.36344385e-01
2.88657248e-01 -4.31114100e-02 1.20936893e-01 6.75525367e-01
1.69297144e-01 -1.32816985e-01 -7.86182880e-01 -1.57381982e-01
5.22854984e-01 -1.73164994e-01 -2.66462207e-01 5.15867829e-01
-9.96019900e-01 6.24248087e-01 4.61694777e-01 3.03045005e-01
-4.36034143e-01 4.58713174e-01 6.81281030e-01 2.76070297e-01
4.05537903e-01 3.34525853e-01 -5.35435200e-01 -4.74583060e-01
-9.89961207e-01 3.43006819e-01 5.80665290e-01 8.84348273e-01
2.50174075e-01 3.18051219e-01 1.05669480e-02 1.25924504e+00
4.41912301e-02 5.21207094e-01 4.94723886e-01 -9.75548267e-01
6.41273856e-01 3.94919008e-01 -7.84188882e-02 -1.00766003e+00
-3.90938222e-01 -4.39256549e-01 -1.30294764e+00 -2.64611214e-01
3.50719154e-01 -2.54125088e-01 -7.50923097e-01 1.50920463e+00
1.77588798e-02 2.62524009e-01 -4.06049704e-03 7.21284032e-01
9.86163378e-01 8.97310138e-01 -1.20967582e-01 8.59737545e-02
1.36660552e+00 -8.77528131e-01 -6.63436115e-01 -5.20706326e-02
5.37928343e-01 -9.32180107e-01 1.14773333e+00 9.26747441e-01
-1.00714087e+00 -9.33922529e-01 -1.23231184e+00 1.94165453e-01
-2.35384345e-01 7.28006124e-01 3.83783281e-01 5.92589378e-01
-9.87645090e-01 5.15492916e-01 -8.03716719e-01 -2.53066212e-01
4.96120572e-01 5.75725615e-01 -1.38719335e-01 2.42558986e-01
-1.28248751e+00 2.63921887e-01 4.61358249e-01 1.55454829e-01
-6.53400183e-01 -7.84504056e-01 -3.84012282e-01 1.99169442e-01
3.04878712e-01 -2.51584232e-01 1.63068080e+00 -4.11101311e-01
-1.52727211e+00 2.93366849e-01 7.91551638e-03 -8.46424460e-01
2.67923683e-01 -4.93878812e-01 -8.56202126e-01 1.70886680e-01
-2.17380330e-01 8.75670969e-01 4.99533802e-01 -5.23887515e-01
-8.99017096e-01 -2.10262701e-01 2.54735164e-02 -2.45911226e-01
-6.87430084e-01 1.13968216e-01 -6.93076551e-01 -7.38603413e-01
-5.20850644e-02 -1.21349072e+00 -3.40590402e-02 1.11723527e-01
-3.12596679e-01 -1.82807222e-01 6.81642711e-01 -6.49939954e-01
1.46792889e+00 -2.57427406e+00 -5.05899131e-01 4.96884137e-02
4.84420285e-02 6.81187451e-01 -1.75078005e-01 -1.06755821e-02
1.12907821e-02 9.21240151e-02 4.97914292e-02 -1.82413355e-01
-2.37812489e-01 -3.02154906e-02 -1.01425134e-01 2.06690714e-01
2.21711457e-01 4.60346609e-01 -3.85561317e-01 -3.33924711e-01
2.26719640e-02 5.87015510e-01 -7.00976133e-01 2.13229299e-01
8.78497288e-02 1.80102699e-02 -2.57135570e-01 6.49086416e-01
5.53082526e-01 -3.52265090e-02 -1.87177569e-01 -2.89210021e-01
-9.61506739e-03 2.08592564e-01 -1.39317000e+00 1.47423589e+00
-5.95535576e-01 8.81398797e-01 -2.89666116e-01 -9.20980513e-01
1.28213561e+00 4.79668677e-01 4.03250486e-01 -6.86966777e-01
2.76852638e-01 2.75376737e-01 2.41114855e-01 -3.58616710e-01
7.59425998e-01 3.47506613e-01 6.97723031e-02 2.72406161e-01
-1.90583523e-02 3.93639028e-01 1.43785983e-01 1.49090327e-02
1.10161281e+00 -5.98976851e-01 1.01795554e-01 -2.01024175e-01
5.65810561e-01 -5.15504837e-01 5.00706315e-01 6.36498630e-01
-1.29346162e-01 6.42711937e-01 3.04732203e-01 -6.20287299e-01
-8.56960356e-01 -7.59860039e-01 -2.22999260e-01 9.05962884e-01
-2.69242525e-01 -7.10748971e-01 -7.04661131e-01 -2.50740677e-01
-1.55500457e-01 2.86770642e-01 -3.51150781e-01 -2.06018269e-01
-5.46242535e-01 -8.42713296e-01 1.50330806e+00 1.04178011e+00
1.06926250e+00 -1.07572508e+00 -4.77166981e-01 3.73915553e-01
-4.69576746e-01 -1.49168277e+00 -1.86894044e-01 -1.12853922e-01
-7.52038479e-01 -9.02259350e-01 -6.59895122e-01 -5.48259199e-01
5.99033386e-02 7.74087310e-02 8.29867065e-01 8.42198059e-02
8.70544929e-03 -1.85604155e-01 -4.22064364e-01 -8.23028564e-01
-1.28320187e-01 4.45443243e-01 3.75067562e-01 6.64099306e-02
3.31917971e-01 -5.87962866e-01 -5.35908937e-01 2.67393112e-01
-6.90988123e-01 -2.17493415e-01 5.63616693e-01 6.80426300e-01
6.24395311e-01 1.92885265e-01 8.88373435e-01 -5.56165993e-01
6.13388658e-01 -1.85022280e-01 -6.70722246e-01 -1.06612751e-02
-6.57888114e-01 -6.17576949e-02 8.78139496e-01 -7.54294097e-01
-6.99972749e-01 1.48881733e-01 -6.10143602e-01 -5.15259624e-01
-2.47945771e-01 3.98220688e-01 8.43271166e-02 -1.23911900e-02
7.72261798e-01 -5.92030995e-02 -2.54392415e-01 -6.86585188e-01
-1.65901780e-02 1.18703139e+00 7.00348675e-01 -5.48570633e-01
1.97730690e-01 1.63544461e-01 -3.65759507e-02 -1.17803407e+00
-7.54239142e-01 -3.59370291e-01 -2.34195948e-01 -1.72394767e-01
5.91612816e-01 -1.14418435e+00 -1.00963163e+00 8.16059113e-01
-1.21361899e+00 -1.09665073e-01 -2.39908118e-02 6.83004558e-01
-2.58530289e-01 1.36117786e-01 -9.13439572e-01 -8.90614212e-01
-8.29575360e-01 -1.15515482e+00 9.74241912e-01 1.45963669e-01
-2.48019129e-01 -3.42138797e-01 -1.05736628e-01 1.09888852e-01
4.14654911e-01 -4.48858971e-03 5.45224249e-01 -6.33731246e-01
-1.74492329e-01 -2.11385339e-01 -4.60230142e-01 7.40094781e-01
-1.56927854e-01 1.95599377e-01 -1.34963930e+00 -9.84880850e-02
-3.40638220e-01 -2.22744063e-01 9.20868754e-01 2.78165519e-01
1.84857690e+00 -2.58793712e-01 -4.60268892e-02 5.13187110e-01
1.17212057e+00 6.43595636e-01 8.95588994e-01 3.32960874e-01
4.35228944e-01 1.65010154e-01 7.30687976e-01 5.93652427e-01
1.85715303e-01 7.57850289e-01 1.65631220e-01 -2.40830574e-02
-1.17579877e-01 -6.59760535e-02 3.47339600e-01 1.01173973e+00
-2.70334244e-01 -2.44643494e-01 -1.04295743e+00 5.36814749e-01
-1.71914136e+00 -7.11170375e-01 -3.24685872e-01 2.06719279e+00
7.41100311e-01 2.55847722e-01 2.05056384e-01 9.28875506e-01
6.70096219e-01 1.66526008e-02 -1.80997699e-01 -4.85136539e-01
2.82783836e-01 7.35906541e-01 2.46266484e-01 2.75256753e-04
-1.17545843e+00 6.87925816e-01 6.07627535e+00 7.94132471e-01
-1.29184163e+00 -8.12044647e-03 6.64883912e-01 -1.97871044e-01
6.89068019e-01 -6.37518048e-01 -9.95857418e-01 6.42300069e-01
1.45896733e+00 2.53620185e-02 3.18812430e-01 9.31724727e-01
-3.39229666e-02 6.51058629e-02 -1.06403327e+00 1.45343447e+00
2.21927185e-03 -1.18864894e+00 -1.48588419e-01 5.01865968e-02
5.18308938e-01 5.23308143e-02 5.73350079e-02 4.14257765e-01
-2.07721278e-01 -8.09098601e-01 5.86391211e-01 3.52208644e-01
9.82621431e-01 -1.27462280e+00 1.27457392e+00 6.13195039e-02
-1.43955469e+00 -1.26148254e-01 -4.89067614e-01 -8.67784992e-02
-1.43162340e-01 8.51679981e-01 -8.08060884e-01 3.81015390e-01
1.28134727e+00 4.26134646e-01 -5.93328595e-01 1.18266284e+00
5.12856664e-03 8.97701025e-01 -5.07315457e-01 -3.33848715e-01
7.06323236e-02 4.70954031e-01 9.27679315e-02 1.36933827e+00
5.77989280e-01 2.01015443e-01 -2.06283003e-01 3.32531214e-01
-4.96645421e-01 1.14293948e-01 -3.70415956e-01 2.85350084e-02
6.37421846e-01 1.07645929e+00 -2.81322002e-01 -4.34203774e-01
-1.67154834e-01 4.85323936e-01 1.44811720e-01 8.35182071e-02
-1.08133364e+00 -9.45177436e-01 6.82443261e-01 -5.56122651e-03
3.36100638e-01 -6.10315539e-02 -2.83476621e-01 -9.18759882e-01
3.00721616e-01 -9.95376050e-01 3.66340846e-01 -6.80725634e-01
-9.19760644e-01 8.35470259e-01 -1.70374840e-01 -1.32144523e+00
-2.16597363e-01 -5.56567609e-01 -3.87216091e-01 5.42905271e-01
-1.24299777e+00 -6.34950578e-01 -4.73506808e-01 5.91879904e-01
3.88689488e-01 -3.95708054e-01 9.15729880e-01 9.67828155e-01
-5.71878731e-01 1.15383101e+00 -9.26919878e-02 4.06524271e-01
7.23386943e-01 -6.33051515e-01 4.89269018e-01 7.80552447e-01
2.94224530e-01 3.19825202e-01 2.73954391e-01 -6.72791749e-02
-1.11762917e+00 -1.41521513e+00 8.95573974e-01 1.16593741e-01
6.44619763e-01 -1.55698210e-01 -1.06933033e+00 3.85663480e-01
-2.56969854e-02 2.82824561e-02 7.48841822e-01 4.66357142e-01
-4.72101390e-01 -7.46461570e-01 -1.03055942e+00 4.30293441e-01
8.53513360e-01 -4.62378114e-01 -2.58099020e-01 -2.05846940e-04
8.12062323e-01 -5.47815502e-01 -1.01411450e+00 6.55848265e-01
9.22322750e-01 -6.11578166e-01 8.60341012e-01 -2.87234128e-01
4.58748251e-01 -3.47259402e-01 -4.81471568e-01 -1.10507691e+00
-1.87185600e-01 -2.03351036e-01 -3.02218974e-01 1.17254651e+00
4.35791075e-01 -6.01112723e-01 6.89145684e-01 8.18254575e-02
-2.15835720e-01 -8.65665436e-01 -9.78724778e-01 -9.10828710e-01
-1.92587256e-01 -1.09874976e+00 8.02132487e-01 6.94153488e-01
-2.67510414e-01 4.30314392e-01 -5.07599533e-01 2.94873536e-01
2.35302642e-01 -2.31413111e-01 8.63784552e-01 -1.29650986e+00
-2.05737293e-01 -2.63483196e-01 -6.88993752e-01 -8.72120976e-01
-9.69048068e-02 -4.32928771e-01 -7.61232078e-02 -1.04254448e+00
-9.51518193e-02 -4.97283131e-01 -7.15193927e-01 7.98039675e-01
1.82211787e-01 6.76220179e-01 2.10191265e-01 -2.08422676e-01
-4.47212726e-01 4.47919726e-01 6.76659405e-01 -2.49098659e-01
-3.84003401e-01 4.97964509e-02 -6.67572260e-01 7.58643985e-01
1.48392510e+00 -4.76087987e-01 -1.84440926e-01 -5.00860691e-01
8.70211348e-02 -8.49648267e-02 2.27105632e-01 -1.59183931e+00
3.45312148e-01 3.49667549e-01 5.26098907e-01 -8.47043097e-01
6.42676950e-01 -5.98427474e-01 9.75780785e-02 4.53463674e-01
-3.42167825e-01 1.94448188e-01 6.43396258e-01 2.45896935e-01
-3.86542469e-01 -5.24646044e-02 8.26100290e-01 3.60503316e-01
-5.10441899e-01 3.17522854e-01 -4.33515042e-01 -8.89251605e-02
6.76211298e-01 2.04620495e-01 -2.33036146e-01 -3.45990479e-01
-4.99578953e-01 -9.14809573e-03 -3.17683548e-01 5.55004597e-01
9.01891351e-01 -1.54868340e+00 -7.66397119e-01 3.29411477e-01
2.11106077e-01 5.86499050e-02 2.54060715e-01 6.14371538e-01
-5.55797100e-01 5.99576712e-01 -1.91352397e-01 -7.28804290e-01
-1.55563140e+00 1.61456242e-01 1.64258346e-01 -8.68443865e-04
-4.09651428e-01 6.80656195e-01 -1.81438506e-01 -3.17298502e-01
6.93717599e-01 -7.21624196e-01 -1.80244759e-01 -1.53588831e-01
9.61502373e-01 7.72979736e-01 5.08875608e-01 -2.26996332e-01
-2.99643338e-01 3.76286656e-01 -1.13600738e-01 5.67997433e-03
1.34267080e+00 5.46500206e-01 7.81960264e-02 4.43062305e-01
1.20899773e+00 -1.69920459e-01 -4.68749195e-01 -1.38360798e-01
-3.89555484e-01 -3.96311581e-01 1.01735495e-01 -5.64926922e-01
-1.42933297e+00 1.14850605e+00 9.73460913e-01 1.32085383e-01
1.47169507e+00 -1.72420666e-01 9.91818190e-01 7.73899436e-01
3.06572378e-01 -1.07974815e+00 3.77182253e-02 5.63112319e-01
9.09804463e-01 -1.00927162e+00 -5.61811924e-02 -2.50570208e-01
-4.15719688e-01 1.07314491e+00 7.69144237e-01 -6.75545260e-02
7.34666705e-01 4.59351212e-01 -8.99709389e-02 1.22616813e-01
-9.44642365e-01 1.11843869e-01 4.15858030e-01 3.53636265e-01
5.52877665e-01 2.37497643e-01 -2.23923296e-01 8.83684337e-01
-5.60939550e-01 3.07167172e-01 5.29530346e-01 7.09426343e-01
-2.64646411e-01 -8.79809976e-01 -3.52439791e-01 5.60341537e-01
-8.58568311e-01 -1.29444376e-01 -2.47334912e-01 8.63622069e-01
1.53241172e-01 1.08159053e+00 2.48560980e-01 -1.01487339e+00
6.76084816e-01 -7.30530918e-02 1.64714098e-01 -1.71156496e-01
-5.84676445e-01 6.15356304e-02 2.12374926e-01 -5.17380297e-01
-3.72993350e-01 -2.88182676e-01 -1.14523673e+00 -6.31906271e-01
-4.51432228e-01 1.70566723e-01 6.79926038e-01 5.45514762e-01
5.58704317e-01 7.72121966e-01 5.39901257e-01 -3.02162349e-01
-2.63374209e-01 -1.09477091e+00 -6.35077178e-01 -2.15103045e-01
1.06396526e-01 -4.42852020e-01 -1.42797485e-01 -5.33431806e-02]
|
[15.021461486816406, 5.2648396492004395]
|
867f86e9-1cdd-4c54-b13b-31bcebef40e1
|
self-supervised-multi-modal-sequential
|
2304.13277
| null |
https://arxiv.org/abs/2304.13277v1
|
https://arxiv.org/pdf/2304.13277v1.pdf
|
Self-Supervised Multi-Modal Sequential Recommendation
|
With the increasing development of e-commerce and online services, personalized recommendation systems have become crucial for enhancing user satisfaction and driving business revenue. Traditional sequential recommendation methods that rely on explicit item IDs encounter challenges in handling item cold start and domain transfer problems. Recent approaches have attempted to use modal features associated with items as a replacement for item IDs, enabling the transfer of learned knowledge across different datasets. However, these methods typically calculate the correlation between the model's output and item embeddings, which may suffer from inconsistencies between high-level feature vectors and low-level feature embeddings, thereby hindering further model learning. To address this issue, we propose a dual-tower retrieval architecture for sequence recommendation. In this architecture, the predicted embedding from the user encoder is used to retrieve the generated embedding from the item encoder, thereby alleviating the issue of inconsistent feature levels. Moreover, in order to further improve the retrieval performance of the model, we also propose a self-supervised multi-modal pretraining method inspired by the consistency property of contrastive learning. This pretraining method enables the model to align various feature combinations of items, thereby effectively generalizing to diverse datasets with different item features. We evaluate the proposed method on five publicly available datasets and conduct extensive experiments. The results demonstrate significant performance improvement of our method.
|
['Yaming Yang', 'Kai Zheng', 'Can Xu', 'Qingfeng Sun', 'Kunzhe Song']
|
2023-04-26
| null | null | null | null |
['sequential-recommendation']
|
['miscellaneous']
|
[-9.79622640e-03 -5.68148017e-01 -4.22797203e-01 -4.91750896e-01
-5.48251808e-01 -5.70393980e-01 4.04285491e-01 3.06183100e-02
-6.11867964e-01 3.81173909e-01 4.17241752e-01 -1.92781910e-02
-2.80747235e-01 -6.73854709e-01 -5.05202353e-01 -5.75018942e-01
2.68437952e-01 2.49027327e-01 1.76339615e-02 -2.64631301e-01
3.11349511e-01 4.60066795e-02 -1.44120848e+00 5.05742013e-01
9.97875214e-01 1.17307520e+00 3.35915387e-01 1.12566873e-02
-3.62437218e-01 3.46997231e-01 -2.60705948e-01 -5.96676648e-01
3.82589400e-01 -3.39078665e-01 -5.10323822e-01 -1.25449374e-01
2.88287163e-01 -5.60737014e-01 -2.67646164e-01 9.73812640e-01
4.82119143e-01 4.31656152e-01 7.29093790e-01 -9.33729827e-01
-1.33910632e+00 4.89110231e-01 -3.76657128e-01 2.14755476e-01
2.41956428e-01 -4.87526804e-02 1.39994991e+00 -1.14516890e+00
3.34339797e-01 8.41966212e-01 4.40707803e-01 5.68701029e-01
-1.13610697e+00 -7.49330342e-01 4.12770927e-01 5.13221800e-01
-1.43443763e+00 -2.46346608e-01 7.29328632e-01 -2.34915867e-01
9.41906095e-01 3.09470706e-02 6.07380509e-01 1.08871150e+00
-1.06742784e-01 8.64044666e-01 6.02120459e-01 -1.90583318e-01
9.06522349e-02 5.74346960e-01 1.47013947e-01 2.80383289e-01
2.16470912e-01 4.89617139e-02 -5.61044753e-01 -8.92238691e-02
7.27062762e-01 6.36034846e-01 -2.79230863e-01 -5.63398480e-01
-1.09923017e+00 8.81215274e-01 6.26039922e-01 4.08544838e-01
-4.96860206e-01 -2.39601016e-01 3.43244225e-01 4.86504763e-01
3.21244955e-01 5.50290644e-01 -5.29954255e-01 1.87582552e-01
-5.47397792e-01 1.60349458e-01 5.25885165e-01 8.92792583e-01
7.46606946e-01 -2.69831747e-01 -1.59210429e-01 1.08285356e+00
4.88272458e-01 3.45973700e-01 9.73315597e-01 -2.09364295e-01
6.30089641e-01 6.12659872e-01 2.49641061e-01 -1.16079330e+00
-2.69706756e-01 -7.00114727e-01 -7.52791584e-01 -6.57344162e-01
1.08758591e-01 1.04284756e-01 -5.82187116e-01 1.74923623e+00
2.35582069e-01 2.45509535e-01 1.46654502e-01 1.17839646e+00
5.04768252e-01 7.74503052e-01 1.58034608e-01 -2.09243774e-01
1.25453401e+00 -1.09352779e+00 -5.91750205e-01 -1.10871591e-01
8.47445905e-01 -6.40745938e-01 1.22498262e+00 3.05544436e-01
-6.91579163e-01 -8.39806914e-01 -1.01012039e+00 1.81144960e-02
-3.96217167e-01 1.65760607e-01 5.51443875e-01 3.50759089e-01
-5.04854023e-01 4.95261133e-01 -4.26184982e-01 -1.97583571e-01
1.86438635e-01 4.41032946e-01 -2.64869064e-01 -2.70092070e-01
-1.47647440e+00 8.24055672e-01 5.59018672e-01 1.35475859e-01
-2.39612207e-01 -6.81238174e-01 -5.42811036e-01 4.20119166e-01
2.42390260e-01 -6.77189589e-01 9.85749602e-01 -1.19969201e+00
-1.55028498e+00 1.59436420e-01 7.13908002e-02 -1.47169247e-01
-1.11710370e-01 -6.32104754e-01 -8.51610184e-01 -1.58809617e-01
-1.58217698e-01 4.06136692e-01 9.01300669e-01 -1.01353395e+00
-7.88918555e-01 -4.21790302e-01 -7.54448771e-02 5.25775075e-01
-1.14529383e+00 -2.63422519e-01 -5.48174739e-01 -6.91510141e-01
-1.14877010e-03 -8.35736334e-01 -1.00317299e-01 -3.91927779e-01
1.27990842e-01 -3.30353528e-01 4.56740797e-01 -4.90571588e-01
1.43663299e+00 -2.32365370e+00 2.67932624e-01 3.96723598e-01
-1.36554897e-01 3.97899419e-01 -5.66023350e-01 4.53743309e-01
5.26733398e-02 -2.26449788e-01 1.74391717e-01 -6.24799840e-02
9.06321332e-02 7.45073631e-02 -5.25956929e-01 1.17724821e-01
1.61934868e-01 9.39705074e-01 -9.55637634e-01 -1.38618052e-01
3.75004075e-02 3.88927460e-01 -8.86142373e-01 4.58161265e-01
-8.98650568e-03 3.62259507e-01 -5.65176547e-01 2.93924928e-01
5.78339815e-01 -4.63570833e-01 3.26657355e-01 -4.37006176e-01
3.22923303e-01 5.18218040e-01 -1.25277126e+00 1.81569910e+00
-7.57291019e-01 -1.89885929e-01 -5.24069011e-01 -1.07473671e+00
9.57281113e-01 2.32314661e-01 4.59556252e-01 -1.30110180e+00
-1.24818370e-01 2.45335370e-01 1.22268647e-01 -4.55436796e-01
7.17360914e-01 -2.77048826e-01 -8.59282017e-02 4.73955721e-01
2.15059683e-01 4.79353696e-01 -1.82367265e-01 1.45972267e-01
8.24767172e-01 9.29191411e-02 1.33565828e-01 7.29200020e-02
6.31947279e-01 -4.15061474e-01 5.28006136e-01 5.12183547e-01
1.51745781e-01 2.51705676e-01 -1.49313822e-01 -4.35728371e-01
-9.12367642e-01 -1.05356467e+00 -1.18091390e-01 1.32385612e+00
4.31987882e-01 -5.95475912e-01 -2.53676534e-01 -9.95225489e-01
1.81752264e-01 6.44065201e-01 -3.40483040e-01 -6.23986602e-01
-5.17348945e-01 -5.20851910e-01 2.12232377e-02 7.27180898e-01
2.27421582e-01 -8.90271246e-01 -1.07568964e-01 4.64434445e-01
-2.92082012e-01 -7.84169257e-01 -7.21855164e-01 1.08045503e-01
-9.90967512e-01 -6.41284525e-01 -6.65186763e-01 -8.05438876e-01
7.40491748e-01 5.93697190e-01 9.18560743e-01 1.75526440e-01
1.95168972e-01 1.06639996e-01 -6.91915810e-01 1.35239780e-01
4.66597266e-02 3.97815198e-01 4.01853949e-01 1.88580409e-01
7.46318758e-01 -4.29791361e-01 -9.42243934e-01 5.08246303e-01
-1.25783694e+00 -1.33908167e-01 8.27660978e-01 1.18653369e+00
4.81586158e-01 -1.59669787e-01 1.17677367e+00 -8.51256609e-01
7.00502515e-01 -8.10948133e-01 -4.14667577e-01 4.07005936e-01
-8.44974518e-01 1.88372418e-01 1.12370408e+00 -6.25219584e-01
-1.02969074e+00 -2.72576600e-01 -2.93614626e-01 -3.52222800e-01
1.45476356e-01 7.32313514e-01 -1.97181284e-01 2.99985021e-01
3.82102251e-01 2.51110554e-01 -2.20403880e-01 -7.19887316e-01
6.03681564e-01 8.97909999e-01 1.14379063e-01 -3.23104829e-01
7.64018953e-01 3.28366761e-03 -5.21058321e-01 -3.75289112e-01
-1.03427267e+00 -8.04430723e-01 -4.72991824e-01 1.84790671e-01
3.81628007e-01 -8.91655505e-01 -3.79501283e-01 4.84999754e-02
-8.07729661e-01 2.18033046e-01 -1.40642062e-01 8.18125904e-01
-2.54744202e-01 3.10807884e-01 -5.73181629e-01 -4.51663941e-01
-3.84618372e-01 -9.23329651e-01 7.37498105e-01 2.38898531e-01
-1.41671315e-01 -9.51681316e-01 2.68373013e-01 2.22095564e-01
6.89450324e-01 -7.65269220e-01 1.14034295e+00 -1.07242668e+00
-4.00623083e-01 -2.55019665e-01 -2.44806975e-01 5.56511760e-01
4.32703882e-01 -6.38377249e-01 -7.54100204e-01 -5.11590362e-01
1.93856526e-02 -3.45502257e-01 7.57457793e-01 -2.15858400e-01
1.15672886e+00 -4.83886421e-01 -2.65515476e-01 4.52702135e-01
1.54880643e+00 1.40334964e-01 5.16447902e-01 3.66809487e-01
7.90302455e-01 4.80687916e-01 8.74472201e-01 3.95466834e-01
3.63200426e-01 9.14564133e-01 1.11973792e-01 2.11799875e-01
2.11271167e-01 -5.28358638e-01 3.46089244e-01 1.28629327e+00
2.91817486e-01 -1.70242682e-01 -3.87517214e-01 5.20636797e-01
-1.90814209e+00 -8.62994432e-01 3.86814117e-01 2.45244241e+00
8.65253150e-01 -5.89262284e-02 5.61766922e-02 -2.53493726e-01
4.04040515e-01 -9.21887606e-02 -6.00490153e-01 -3.23467195e-01
2.65449852e-01 2.20383570e-01 1.86520711e-01 6.88845739e-02
-9.35817659e-01 8.71556163e-01 5.08198929e+00 8.05795491e-01
-1.23224044e+00 3.11815981e-02 1.31158963e-01 -2.83155799e-01
-6.42934501e-01 -1.52736500e-01 -8.89087200e-01 6.92471683e-01
9.21382666e-01 -1.46740243e-01 4.62927222e-01 8.06618214e-01
-1.41556785e-01 3.42425346e-01 -1.23144925e+00 8.75696301e-01
2.20588386e-01 -9.94414687e-01 3.78235161e-01 6.01334721e-02
7.76983917e-01 -1.70563892e-01 3.68191510e-01 8.03663671e-01
-1.54708475e-01 -6.88562810e-01 2.90516376e-01 5.67558527e-01
4.43188250e-01 -8.99783552e-01 7.68259704e-01 3.24228227e-01
-9.85763371e-01 -4.02965188e-01 -7.11440504e-01 1.57291666e-01
5.53939007e-02 3.71554732e-01 -7.67785668e-01 7.32261121e-01
5.79984188e-01 8.81427944e-01 -5.03180861e-01 1.07971847e+00
7.61661455e-02 4.49008584e-01 -9.51860249e-02 -1.11477219e-01
1.03905693e-01 -3.74500275e-01 1.41591012e-01 1.01588738e+00
4.70709980e-01 -6.74133524e-02 7.50120729e-02 5.04220009e-01
-2.92362779e-01 6.15204871e-01 -5.83451807e-01 -8.60849991e-02
4.80075210e-01 1.23574162e+00 -1.92819163e-01 -3.25158447e-01
-7.03240633e-01 1.14027631e+00 6.91189945e-01 3.95038247e-01
-8.19093466e-01 -3.59850466e-01 8.07688773e-01 -5.84741794e-02
6.96697950e-01 -5.13282754e-02 2.10547820e-02 -1.40924168e+00
1.86592683e-01 -9.97499764e-01 4.46226269e-01 -2.72347838e-01
-1.73867345e+00 4.07411754e-01 -2.82535791e-01 -1.67538583e+00
-3.67216319e-01 -4.11264777e-01 -2.22374558e-01 8.07232141e-01
-1.65683293e+00 -9.80751336e-01 6.74088970e-02 5.45354486e-01
3.58412296e-01 -1.90055445e-01 9.29439366e-01 7.37430513e-01
-5.23828626e-01 1.03517544e+00 5.85054219e-01 -1.27191797e-01
1.14915848e+00 -8.81663501e-01 9.04649943e-02 5.00815451e-01
3.92047405e-01 1.17497647e+00 1.91960230e-01 -4.09044534e-01
-1.66053319e+00 -1.24309635e+00 7.16067851e-01 -3.49656910e-01
5.63702106e-01 -2.52893567e-01 -1.21338224e+00 5.40303886e-01
1.47087825e-03 -1.20679386e-01 1.25249398e+00 5.49683988e-01
-7.15301394e-01 -5.71598768e-01 -8.02563071e-01 5.34403503e-01
9.31290567e-01 -5.98679125e-01 -7.20460951e-01 8.13262984e-02
6.07613742e-01 9.68732834e-02 -1.03607249e+00 3.43124747e-01
8.37528408e-01 -6.81565940e-01 8.51518154e-01 -8.94166470e-01
3.92542869e-01 -3.11145604e-01 -2.31056944e-01 -1.49139988e+00
-7.43863404e-01 8.36661458e-02 -3.64904433e-01 1.26240420e+00
4.51988220e-01 -6.01667941e-01 5.06173253e-01 6.01374686e-01
-2.82660704e-02 -7.78192401e-01 -6.37445927e-01 -8.52387488e-01
2.80623157e-02 -1.05506159e-01 7.55560637e-01 9.18800533e-01
2.91866452e-01 6.54729486e-01 -5.24498940e-01 2.41303556e-02
1.75162867e-01 3.93682569e-01 7.15267718e-01 -1.04283357e+00
-6.81684196e-01 -1.94674060e-01 -1.82684034e-01 -1.55956602e+00
1.10530265e-01 -1.12739992e+00 8.38205870e-03 -1.23971200e+00
4.40439165e-01 -6.97035074e-01 -1.06748509e+00 3.08261752e-01
-4.72301722e-01 3.09202343e-01 1.66053340e-01 4.11721021e-01
-8.63388538e-01 7.07382977e-01 1.15961635e+00 3.43491100e-02
-2.26119116e-01 -8.50401670e-02 -8.52842033e-01 3.11209679e-01
7.64676929e-01 -4.41555381e-01 -6.72046900e-01 -7.31273592e-01
5.43233335e-01 -2.33242109e-01 7.82662630e-02 -7.17136145e-01
1.77237332e-01 -7.22596655e-03 5.59009314e-01 -3.63375455e-01
2.38651797e-01 -1.19246984e+00 -7.99305961e-02 2.17895865e-01
-4.84920710e-01 7.54553452e-02 -3.80861526e-03 7.82063901e-01
-3.60706896e-01 -3.02732319e-01 4.43073422e-01 7.45754316e-02
-8.17853689e-01 3.58919293e-01 8.88623483e-03 -2.11583897e-01
7.14436531e-01 1.51802041e-02 -2.12145582e-01 -1.47587091e-01
-3.80612940e-01 3.30600619e-01 5.40204108e-01 9.19228494e-01
7.54986227e-01 -1.65364993e+00 -4.19635981e-01 4.30658668e-01
5.64160109e-01 -3.43038499e-01 5.16578794e-01 7.48050153e-01
2.58894533e-01 6.01231813e-01 -2.15123728e-01 -3.02271247e-01
-7.23254383e-01 9.08033550e-01 -2.23147124e-02 -4.26888108e-01
-4.03365225e-01 6.79261148e-01 3.60781044e-01 -4.21315342e-01
1.26687065e-01 -3.80266681e-02 -3.27291042e-01 1.37005508e-01
7.03942060e-01 -3.84623511e-03 2.29204074e-01 -3.77765119e-01
-2.37028703e-01 3.75313252e-01 -8.00411820e-01 2.72470303e-02
1.29893243e+00 -3.64006788e-01 1.87002555e-01 4.58372682e-01
1.49767172e+00 -5.08322120e-02 -8.37962866e-01 -7.48040557e-01
1.37554616e-01 -6.27769411e-01 -3.94271538e-02 -7.14501619e-01
-9.68391061e-01 7.33805120e-01 7.28146136e-01 3.07669155e-02
1.17515337e+00 -1.76765800e-01 1.19038963e+00 5.73736191e-01
4.63304490e-01 -1.17041051e+00 2.94703543e-01 4.66571808e-01
5.67341268e-01 -1.26711619e+00 -2.40439877e-01 6.13875985e-02
-7.99147844e-01 9.88669693e-01 7.40462899e-01 -1.05832003e-01
4.99939770e-01 -3.39607835e-01 1.20022213e-02 1.01951681e-01
-7.69875288e-01 -5.93443140e-02 6.36312842e-01 3.11179638e-01
5.46124458e-01 -8.52976814e-02 -5.12241483e-01 1.05805445e+00
1.95506155e-01 1.14726886e-01 -9.38365161e-02 7.69941211e-01
-3.32258999e-01 -1.45441675e+00 1.78708509e-01 7.13628650e-01
-2.60767728e-01 -3.20903957e-01 -2.64458936e-02 3.08353186e-01
-1.97335519e-02 6.88758492e-01 8.25060531e-02 -5.98492146e-01
3.93387645e-01 2.60310948e-01 4.23854232e-01 -7.01111078e-01
-6.10144734e-01 -2.54009687e-03 -1.52829871e-01 -3.43287855e-01
-2.23527789e-01 -4.02793795e-01 -9.70283568e-01 1.90744400e-01
-7.72207379e-01 4.34789181e-01 5.98436713e-01 9.09072697e-01
6.65648103e-01 2.33687013e-01 1.11034572e+00 -5.09764314e-01
-1.13072944e+00 -9.87055540e-01 -5.68434656e-01 8.11406732e-01
9.51342806e-02 -7.05282509e-01 -8.74557719e-02 -2.57597923e-01]
|
[10.168173789978027, 5.531567096710205]
|
fe17f9da-ae85-4b00-a9e1-5931c66143b1
|
react-temporal-action-detection-with
|
2207.07097
| null |
https://arxiv.org/abs/2207.07097v1
|
https://arxiv.org/pdf/2207.07097v1.pdf
|
ReAct: Temporal Action Detection with Relational Queries
|
This work aims at advancing temporal action detection (TAD) using an encoder-decoder framework with action queries, similar to DETR, which has shown great success in object detection. However, the framework suffers from several problems if directly applied to TAD: the insufficient exploration of inter-query relation in the decoder, the inadequate classification training due to a limited number of training samples, and the unreliable classification scores at inference. To this end, we first propose a relational attention mechanism in the decoder, which guides the attention among queries based on their relations. Moreover, we propose two losses to facilitate and stabilize the training of action classification. Lastly, we propose to predict the localization quality of each action query at inference in order to distinguish high-quality queries. The proposed method, named ReAct, achieves the state-of-the-art performance on THUMOS14, with much lower computational costs than previous methods. Besides, extensive ablation studies are conducted to verify the effectiveness of each proposed component. The code is available at https://github.com/sssste/React.
|
['DaCheng Tao', 'Jia Li', 'Lin Ma', 'Jing Zhang', 'Qiong Cao', 'Yujie Zhong', 'Dingfeng Shi']
|
2022-07-14
| null | null | null | null |
['action-classification']
|
['computer-vision']
|
[ 2.37521589e-01 -4.44369055e-02 -5.83487093e-01 -1.71295315e-01
-1.13877988e+00 -7.73194656e-02 5.22845089e-01 -1.78852484e-01
-4.70761031e-01 4.13234740e-01 2.31426120e-01 -2.25195717e-02
2.61649974e-02 -5.75931251e-01 -5.53784847e-01 -4.54312205e-01
1.01671167e-01 2.69243747e-01 6.69586599e-01 1.12074785e-01
8.71915966e-02 1.27345160e-01 -1.56739771e+00 3.65687072e-01
8.83472383e-01 1.37875223e+00 2.70814657e-01 4.86611873e-01
2.52548397e-01 1.33270156e+00 -5.33533871e-01 -5.06936729e-01
-4.17441987e-02 -6.61651075e-01 -7.10108697e-01 3.70757617e-02
2.62408197e-01 -8.57111990e-01 -7.94543862e-01 1.00996804e+00
6.13157094e-01 8.99756029e-02 4.32126552e-01 -1.25854063e+00
-6.12597942e-01 5.31134784e-01 -5.31693935e-01 3.72893035e-01
3.09743285e-01 1.15475431e-01 1.34312916e+00 -1.10437226e+00
3.80146474e-01 1.30843091e+00 3.03651631e-01 6.65268481e-01
-9.25691903e-01 -5.63503802e-01 3.07710052e-01 7.82306254e-01
-1.46631515e+00 -6.42363667e-01 5.86712658e-01 -2.26061940e-01
1.00677013e+00 4.66524065e-02 6.57057047e-01 1.30880678e+00
-8.34922716e-02 1.50076485e+00 5.13760805e-01 -2.10419968e-01
1.60521016e-01 -3.18922192e-01 -2.02545207e-02 7.09262848e-01
-1.82959855e-01 -3.20518129e-02 -9.00810242e-01 2.10779533e-01
6.00418389e-01 -8.86132419e-02 -8.27554688e-02 -1.05302915e-01
-1.10857201e+00 4.83229697e-01 4.63646024e-01 1.90677837e-01
-4.12469089e-01 3.62212628e-01 4.80653673e-01 3.02773360e-02
4.65309948e-01 2.77748276e-02 -1.86682194e-01 -5.80324352e-01
-7.49463737e-01 1.15209311e-01 1.71602532e-01 1.00411224e+00
2.09345847e-01 -1.14507236e-01 -7.20200121e-01 7.42619395e-01
4.08598930e-01 3.03646296e-01 3.22696447e-01 -1.11454034e+00
8.40779185e-01 7.01567233e-01 1.50121972e-01 -7.23682106e-01
-1.34567395e-01 -3.67921114e-01 -5.97674549e-01 -1.08676888e-02
2.98049867e-01 1.18827544e-01 -7.62411892e-01 1.78835309e+00
2.74943382e-01 3.81432354e-01 -2.37561837e-02 9.68456984e-01
5.36982834e-01 3.68825883e-01 5.25381304e-02 -9.58125219e-02
1.27881610e+00 -1.16191304e+00 -9.04466927e-01 -3.28246385e-01
8.02180350e-01 -6.04940653e-01 1.13581049e+00 3.77121538e-01
-1.18011236e+00 -6.97530687e-01 -1.02117383e+00 -4.09174681e-01
-7.45161250e-03 8.20243001e-01 5.49439847e-01 2.77668595e-01
-7.70531058e-01 3.17734599e-01 -1.29276705e+00 -2.99606413e-01
7.80849814e-01 2.68249989e-01 3.44773866e-02 -9.98978410e-03
-1.35157943e+00 7.47999132e-01 3.11640441e-01 2.92154223e-01
-1.11338782e+00 -2.28733793e-01 -7.18680084e-01 1.13068193e-01
8.35092306e-01 -4.37227160e-01 1.58269298e+00 -6.17229462e-01
-1.48110712e+00 6.56461537e-01 -4.20052886e-01 -6.32299364e-01
7.49136508e-01 -7.05012918e-01 -2.86164671e-01 2.67172009e-01
3.31281900e-01 7.57117927e-01 6.17452860e-01 -5.75532913e-01
-8.35785747e-01 -3.57195318e-01 3.45617384e-01 3.21698785e-01
-3.24596971e-01 -4.15074266e-02 -1.12525260e+00 -6.58397198e-01
2.36274645e-01 -8.22187245e-01 1.55092543e-02 1.62864909e-01
-4.99602437e-01 -5.11128068e-01 5.76981485e-01 -4.77194279e-01
1.55209243e+00 -2.43770623e+00 1.66994691e-01 -2.42097676e-01
1.25397786e-01 2.91581750e-01 -1.09122984e-01 2.52702534e-01
1.88641563e-01 -3.25010903e-02 -1.00977845e-01 -5.14634371e-01
6.60020933e-02 1.48226649e-01 -2.84508139e-01 4.38511372e-01
3.61105949e-01 8.72004151e-01 -8.92386258e-01 -6.72861397e-01
2.25965783e-01 3.17549765e-01 -5.93903124e-01 2.04969838e-01
-3.62613082e-01 4.06461269e-01 -9.00274158e-01 8.31330657e-01
3.30403626e-01 -3.94251108e-01 1.28056973e-01 -1.72155365e-01
-3.68322879e-02 7.00528979e-01 -1.01638818e+00 1.77050781e+00
-9.08422694e-02 5.35702050e-01 -3.08803409e-01 -9.86695290e-01
4.98766601e-01 3.48264903e-01 6.38959587e-01 -1.06840467e+00
1.65180698e-01 7.34085590e-02 -6.31234571e-02 -6.34072602e-01
3.93612415e-01 4.08734411e-01 3.19314599e-02 3.03905010e-01
3.89364995e-02 4.52416658e-01 4.50967491e-01 2.87928879e-01
1.22152948e+00 4.34962213e-01 1.63469940e-01 3.79948556e-01
6.14006758e-01 -2.79516846e-01 7.77208805e-01 8.53108227e-01
-4.17762101e-01 5.11453092e-01 7.39654422e-01 4.53572273e-02
-6.75675154e-01 -8.95156980e-01 -1.12313546e-01 1.08320582e+00
4.15049285e-01 -5.94487131e-01 -5.74536622e-01 -8.63684058e-01
-2.52181917e-01 7.50865877e-01 -5.83398461e-01 -4.07701194e-01
-4.83077317e-01 -5.64513206e-01 8.25438559e-01 8.14553976e-01
7.45315969e-01 -1.04692614e+00 -6.07398450e-01 9.99352261e-02
-5.10491967e-01 -1.28972936e+00 -5.20186007e-01 -4.55976576e-02
-7.44597852e-01 -1.22014499e+00 -5.62998772e-01 -5.00058830e-01
5.02035618e-01 2.63436675e-01 8.22060764e-01 1.43804708e-02
3.28114815e-02 1.96822241e-01 -5.64176679e-01 -2.14846209e-01
-1.02339037e-01 1.67212918e-01 -1.21322282e-01 1.13015138e-01
5.34027457e-01 -2.59504676e-01 -6.79205894e-01 6.02975190e-01
-7.24637866e-01 7.45053887e-02 8.30260277e-01 7.82681525e-01
7.48324335e-01 -4.39408720e-02 4.28518295e-01 -6.23819649e-01
3.37742120e-01 -3.70088875e-01 -5.32511353e-01 2.47576416e-01
-5.16931891e-01 -7.38614723e-02 2.52151191e-01 -2.31047124e-01
-1.14525580e+00 1.00344932e-02 -4.04521137e-01 -4.37609017e-01
3.35615836e-02 3.82883787e-01 -2.73143530e-01 3.90965909e-01
4.01642919e-01 3.58413190e-01 -2.41353944e-01 -6.12999916e-01
3.05597126e-01 5.56937039e-01 3.80694360e-01 -2.72140890e-01
3.84907126e-01 4.82411951e-01 -2.18562260e-01 -3.49074662e-01
-1.18400991e+00 -3.60192984e-01 -6.10409498e-01 -2.86494285e-01
9.76297498e-01 -1.10797715e+00 -8.33822966e-01 6.70520961e-01
-1.07253766e+00 -4.15829331e-01 -1.13948487e-01 6.62705421e-01
-6.57359719e-01 3.62699956e-01 -7.48744786e-01 -9.57237244e-01
-6.92133829e-02 -1.21210980e+00 1.28228414e+00 9.13490281e-02
-2.96643935e-02 -3.41050178e-01 -2.84451276e-01 5.97066581e-01
-4.37133908e-02 -2.32069060e-01 5.70957541e-01 -4.80531782e-01
-8.82535636e-01 -2.05668166e-01 -3.04408550e-01 2.79471457e-01
3.54377031e-02 -1.48195565e-01 -9.65440452e-01 -1.37917504e-01
-6.82993084e-02 -6.46915913e-01 1.09662962e+00 3.15151781e-01
1.38896930e+00 -1.06381793e-02 -4.03521925e-01 2.83473700e-01
9.05283034e-01 3.91600907e-01 7.96637356e-01 3.08523804e-01
4.49669480e-01 2.15957403e-01 1.21025479e+00 6.31040812e-01
3.87432516e-01 1.04742277e+00 4.90102321e-01 1.88417092e-01
-2.64121979e-01 -4.26972270e-01 6.39292359e-01 6.07068539e-01
-3.07689141e-02 -5.97894430e-01 -7.18403459e-01 3.70412409e-01
-2.15984964e+00 -9.86406147e-01 -6.87431395e-02 2.17324591e+00
6.54885471e-01 6.09322071e-01 1.49504483e-01 2.97458142e-01
5.12596071e-01 3.49381804e-01 -6.78466499e-01 2.89278388e-01
1.88413856e-03 -1.86635345e-01 2.64507264e-01 3.83143961e-01
-1.24817038e+00 1.08886564e+00 5.62891054e+00 9.91376340e-01
-8.19182456e-01 2.80739665e-01 4.65565294e-01 -2.62585074e-01
2.29585364e-01 -1.08476013e-01 -9.50620413e-01 5.80693185e-01
6.13106906e-01 4.28141467e-02 2.62752354e-01 8.49654019e-01
4.26382452e-01 -3.34534794e-01 -1.13261151e+00 8.51549387e-01
9.92889032e-02 -1.07179999e+00 -6.85031116e-02 -2.69923955e-01
2.33777806e-01 1.20724797e-01 1.04923926e-01 4.30767864e-01
-3.21066678e-02 -4.64903265e-01 9.06804919e-01 5.71186662e-01
6.77782238e-01 -4.98549312e-01 5.59813082e-01 3.31763208e-01
-1.28621995e+00 -2.38553360e-01 -2.51238734e-01 -4.84235724e-03
1.38818532e-01 3.61742795e-01 -6.15710020e-01 5.62570274e-01
7.00569689e-01 1.09109795e+00 -6.96285307e-01 1.03841555e+00
-6.21635020e-01 7.40384459e-01 -1.19252838e-01 5.39579913e-02
2.12446883e-01 1.49698863e-02 5.34675479e-01 7.52050519e-01
3.08823705e-01 6.79578111e-02 1.57405019e-01 7.19663560e-01
-9.47042257e-02 -1.40051514e-01 -2.78413206e-01 -1.43987909e-01
4.77802843e-01 6.56062365e-01 -4.12599474e-01 -3.35822016e-01
-6.08306587e-01 1.08648181e+00 3.77236009e-01 4.45577174e-01
-1.20502615e+00 -1.39329866e-01 4.87777352e-01 2.11262610e-02
4.25636441e-01 -1.31663442e-01 -7.96344429e-02 -1.15939128e+00
4.05437052e-01 -7.27054954e-01 6.12222552e-01 -7.67700672e-01
-8.83692503e-01 3.62951308e-01 -1.79292783e-01 -1.67313147e+00
-1.35950655e-01 -3.06518346e-01 -3.31959754e-01 4.38530594e-01
-1.33796299e+00 -9.38879550e-01 -1.87602237e-01 4.17385489e-01
7.73201942e-01 -3.47854383e-02 4.64309067e-01 7.89925575e-01
-9.67095017e-01 7.92248964e-01 -1.31457850e-01 3.35055739e-01
6.91098869e-01 -9.88180101e-01 3.22617471e-01 1.00550079e+00
1.20303258e-01 2.67012894e-01 3.24705362e-01 -6.16318226e-01
-1.17799795e+00 -1.14467013e+00 9.70940948e-01 -3.78694713e-01
5.88995278e-01 -3.74202818e-01 -8.78339052e-01 7.62002170e-01
-1.35752589e-01 -3.85159142e-02 4.38214123e-01 1.50636986e-01
-1.02014266e-01 -2.04517603e-01 -6.04618967e-01 7.67540753e-01
1.35804188e+00 -6.69287801e-01 -4.15917426e-01 3.17544997e-01
6.14760101e-01 -5.00380576e-01 -6.56134725e-01 3.61396164e-01
5.31993330e-01 -1.16070008e+00 8.77476156e-01 -3.92620564e-01
4.85542923e-01 -5.07881582e-01 -1.22187115e-01 -6.30588055e-01
-3.57805490e-01 -3.71014327e-01 -4.95781362e-01 1.22978985e+00
4.90022480e-01 -3.39374483e-01 8.55657816e-01 3.14823210e-01
-2.24650636e-01 -1.03027797e+00 -1.22936285e+00 -8.00743282e-01
-5.20195127e-01 -7.42352128e-01 2.25358233e-01 4.05058116e-01
-1.56034365e-01 5.80713153e-01 -7.12592185e-01 3.30666423e-01
4.05384481e-01 1.77109279e-02 6.40254080e-01 -6.83517277e-01
-5.06389141e-01 -3.96931261e-01 -4.19712782e-01 -1.75792539e+00
-2.02810634e-02 -6.51727438e-01 1.51060551e-01 -1.43983543e+00
2.62470186e-01 -2.88373679e-01 -5.41192710e-01 6.94175601e-01
-3.31323653e-01 1.29462063e-01 2.22058803e-01 4.02308106e-01
-1.30420613e+00 1.07935333e+00 1.32264209e+00 -1.71009034e-01
-1.64377004e-01 2.45559677e-01 -4.82815832e-01 5.65393746e-01
6.15535736e-01 -6.11253858e-01 -4.76057559e-01 -6.07978523e-01
1.61318034e-01 1.84488520e-01 4.53964859e-01 -1.19756770e+00
2.51503050e-01 9.64700207e-02 2.55062938e-01 -9.75720763e-01
6.10033154e-01 -5.98219573e-01 -2.85010517e-01 5.22388339e-01
-6.13899529e-01 -2.69565821e-01 -1.06530309e-01 7.11269319e-01
-4.29203659e-01 -7.53764212e-02 7.38794446e-01 2.28750542e-01
-8.05091441e-01 3.38194698e-01 -3.38158578e-01 1.27419978e-01
8.85562539e-01 -8.31362978e-02 -2.84753501e-01 -3.94368947e-01
-7.19427645e-01 6.63081527e-01 6.78111464e-02 5.53011835e-01
6.13827527e-01 -1.58122325e+00 -4.90295559e-01 2.56071482e-02
2.87096888e-01 -5.51954098e-02 3.91822994e-01 1.37519979e+00
-2.52025947e-02 5.17445505e-01 9.54986662e-02 -5.55508375e-01
-1.25321817e+00 5.04203975e-01 3.20534050e-01 -2.25670591e-01
-6.72549784e-01 8.67064595e-01 2.36203864e-01 9.64564234e-02
6.54363990e-01 -4.14166778e-01 -2.29186386e-01 2.05320984e-01
5.40889025e-01 4.43695128e-01 -1.33610874e-01 -5.06852746e-01
-5.66921115e-01 2.98130959e-01 -1.66537672e-01 -4.07420509e-02
9.67251301e-01 -2.17073664e-01 2.62178838e-01 4.65936571e-01
9.26765382e-01 -3.10102075e-01 -1.49862027e+00 -3.76460224e-01
6.22004718e-02 -6.85309529e-01 8.89127553e-02 -5.93172133e-01
-9.53771412e-01 8.76813829e-01 6.44656360e-01 4.40134555e-02
1.21258318e+00 1.89773649e-01 8.54752123e-01 4.72454011e-01
8.61911103e-02 -1.24789786e+00 4.44553167e-01 4.60076421e-01
8.45333874e-01 -1.31858897e+00 -4.11598086e-02 -4.53012884e-01
-6.98994398e-01 7.81817615e-01 9.02858377e-01 2.39308462e-01
3.52467507e-01 -2.56790984e-02 -1.23667523e-01 -7.94940963e-02
-9.22001898e-01 -5.59955478e-01 2.91863739e-01 1.13259405e-01
3.75526696e-01 -2.08078533e-01 -3.48244488e-01 5.37446856e-01
2.62706995e-01 2.42861569e-01 1.68415070e-01 9.03645575e-01
-3.69164139e-01 -1.09668112e+00 7.80423582e-02 6.19504869e-01
-5.71730673e-01 -7.74437115e-02 -3.76449436e-01 5.97637713e-01
1.67442814e-01 1.03163135e+00 1.31584387e-02 -5.00832140e-01
5.67741334e-01 -6.75018802e-02 3.66553962e-01 -4.82530385e-01
-4.10387479e-02 3.19548428e-01 2.13050187e-01 -9.50810909e-01
-5.33399820e-01 -7.81349242e-01 -1.28389585e+00 1.48781419e-01
-3.48435730e-01 -8.93312171e-02 -1.32684251e-02 9.76455331e-01
5.41528821e-01 6.68290138e-01 5.75626969e-01 -4.03202534e-01
-5.78185976e-01 -9.91980076e-01 -4.20871913e-01 2.77678400e-01
2.32448429e-01 -1.00214350e+00 -8.33292603e-02 -1.39942750e-01]
|
[8.44940185546875, 0.5115834474563599]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.