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
|
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
575d827a-4a60-4461-8320-92d36e6424d4
|
neural-knowledge-extraction-from-cloud
|
2007.05505
| null |
https://arxiv.org/abs/2007.05505v4
|
https://arxiv.org/pdf/2007.05505v4.pdf
|
Neural Knowledge Extraction From Cloud Service Incidents
|
In the last decade, two paradigm shifts have reshaped the software industry - the move from boxed products to services and the widespread adoption of cloud computing. This has had a huge impact on the software development life cycle and the DevOps processes. Particularly, incident management has become critical for developing and operating large-scale services. Incidents are created to ensure timely communication of service issues and, also, their resolution. Prior work on incident management has been heavily focused on the challenges with incident triaging and de-duplication. In this work, we address the fundamental problem of structured knowledge extraction from service incidents. We have built SoftNER, a framework for unsupervised knowledge extraction from service incidents. We frame the knowledge extraction problem as a Named-entity Recognition task for extracting factual information. SoftNER leverages structural patterns like key,value pairs and tables for bootstrapping the training data. Further, we build a novel multi-task learning based BiLSTM-CRF model which leverages not just the semantic context but also the data-types for named-entity extraction. We have deployed SoftNER at Microsoft, a major cloud service provider and have evaluated it on more than 2 months of cloud incidents. We show that the unsupervised machine learning based approach has a high precision of 0.96. Our multi-task learning based deep learning model also outperforms the state of the art NER models. Lastly, using the knowledge extracted by SoftNER we are able to build significantly more accurate models for important downstream tasks like incident triaging.
|
['Sumit Kumar', 'Thomas Zimmermann', 'Manish Shetty', 'Chetan Bansal', 'Nikitha Rao', 'Nachiappan Nagappan']
|
2020-07-10
| null | null | null | null |
['entity-extraction']
|
['natural-language-processing']
|
[-3.70867155e-03 -7.73741072e-03 -1.02000557e-01 -4.87692982e-01
-1.06499231e+00 -4.68818158e-01 3.01604182e-01 4.62964863e-01
-5.32091677e-01 3.58913451e-01 4.92032409e-01 -5.90303600e-01
9.86452252e-02 -7.57266045e-01 -4.89943862e-01 -1.44489482e-01
-5.25742099e-02 7.01616108e-01 -7.78430374e-04 -1.19007811e-01
2.48278022e-01 3.49313796e-01 -1.06060219e+00 8.12889397e-01
2.68142164e-01 9.38469112e-01 -1.75221309e-01 5.10963857e-01
-5.16170204e-01 9.03282762e-01 -8.94523025e-01 -5.34343481e-01
1.38147607e-01 2.79286683e-01 -1.04247642e+00 -9.68930647e-02
7.38778338e-03 -1.64169133e-01 -1.67622924e-01 6.65598392e-01
4.97876614e-01 -9.97321010e-02 1.92244604e-01 -1.47604024e+00
-4.87690210e-01 9.88496482e-01 -5.49362183e-01 5.04063666e-01
3.09698194e-01 -2.13176221e-01 1.06558514e+00 -8.22652996e-01
9.14474070e-01 5.24494648e-01 9.20925140e-01 2.31848896e-01
-7.96656966e-01 -6.92908525e-01 1.05998479e-02 5.00510573e-01
-1.34446836e+00 -4.91860837e-01 3.61493617e-01 -5.55104792e-01
2.15285110e+00 3.46869268e-02 2.51058526e-02 8.94509256e-01
1.18646473e-01 4.07426447e-01 8.44219029e-01 -7.70589352e-01
-4.49369214e-02 2.72756964e-01 3.32167774e-01 5.54540813e-01
2.65362978e-01 -2.72039503e-01 -3.61943841e-01 -1.29372403e-01
3.56272936e-01 4.35595900e-01 5.99856116e-02 4.24548894e-01
-7.75203884e-01 4.79386389e-01 1.16942681e-01 9.08014178e-01
-7.36569822e-01 7.97651708e-03 7.07672179e-01 3.18372309e-01
6.60377622e-01 6.44114435e-01 -1.11189735e+00 -6.18267655e-01
-1.04706776e+00 1.67614948e-02 1.43281901e+00 9.12332118e-01
5.85706949e-01 -1.39637396e-01 2.00200770e-02 7.34430015e-01
-4.56236303e-02 2.58626699e-01 8.02380964e-02 -4.35692549e-01
7.56256580e-01 7.53791332e-01 7.53317624e-02 -1.03940523e+00
-6.74460769e-01 -3.04136366e-01 -4.04900402e-01 -3.53052795e-01
5.13951741e-02 -2.68043548e-01 -1.02077079e+00 1.34478581e+00
1.06313281e-01 4.98062730e-01 -1.06341183e-01 4.52177346e-01
7.04420209e-01 4.32143569e-01 2.22389579e-01 -1.00405678e-01
1.28972042e+00 -7.99635947e-01 -7.64736354e-01 -1.34633571e-01
7.97110915e-01 -9.82013285e-01 4.63999808e-01 3.74273509e-01
-8.14949453e-01 4.12637070e-02 -6.07105553e-01 6.35304973e-02
-6.96872234e-01 -1.88278973e-01 6.36280835e-01 2.53710806e-01
-8.41165960e-01 4.70937639e-01 -8.38921905e-01 -5.45376301e-01
4.70217139e-01 2.96372563e-01 -5.54473400e-01 -3.29911709e-01
-1.09789956e+00 1.08036518e+00 2.21359804e-01 -1.97661981e-01
-6.38021529e-01 -9.21988845e-01 -7.82622278e-01 2.83466637e-01
7.00552285e-01 -4.30441111e-01 1.50660217e+00 -3.55507702e-01
-7.00579524e-01 8.41311991e-01 -1.59668818e-01 -3.62498045e-01
-3.90699476e-01 -2.50090390e-01 -1.04923391e+00 -2.91936874e-01
3.76594514e-01 -2.70514548e-01 4.59484488e-01 -1.19453609e+00
-1.08302915e+00 -4.92691338e-01 -5.60934469e-02 -3.64345253e-01
-5.01673520e-01 9.75436568e-01 -3.55532825e-01 -3.82951826e-01
-4.05250609e-01 -6.66035831e-01 -3.54140937e-01 -1.08240926e+00
-6.72966421e-01 -1.91027671e-01 4.42704856e-01 -1.06406093e+00
1.65284514e+00 -1.88009906e+00 -2.80895233e-01 1.68730646e-01
3.97246957e-01 3.12122405e-01 -5.11225723e-02 8.95801723e-01
-2.16081068e-01 4.81822580e-01 1.15040556e-01 -6.48779154e-01
1.67660967e-01 2.75908381e-01 -5.88431835e-01 -1.23730123e-01
2.87718594e-01 8.65148365e-01 -7.03784466e-01 -4.27027255e-01
-2.33336046e-01 5.51324129e-01 -4.64596093e-01 3.84754896e-01
7.41581153e-03 7.41308704e-02 -1.36677682e-01 9.29372072e-01
3.31585616e-01 -4.21684653e-01 3.24613661e-01 -1.06653988e-01
-3.50389481e-01 6.81605816e-01 -7.55914330e-01 1.52596498e+00
-1.03008664e+00 3.60760212e-01 -3.65205556e-02 -9.28980768e-01
5.92251599e-01 6.70629144e-01 8.10993910e-01 -7.13406324e-01
2.61862818e-02 2.23709553e-01 -3.35296154e-01 -9.34964359e-01
5.76174140e-01 -2.99931735e-01 -5.16481221e-01 7.05697834e-01
1.79989740e-01 1.62171096e-01 2.66843401e-02 2.64533609e-01
1.73485804e+00 -1.83574572e-01 4.14620310e-01 3.91452879e-01
-1.44658163e-01 4.25878137e-01 8.78025055e-01 3.96987408e-01
3.73110659e-02 3.74931902e-01 2.51387119e-01 -6.05346441e-01
-9.82418060e-01 -7.23282158e-01 -4.04536426e-02 1.07197738e+00
-4.42001641e-01 -7.36535966e-01 -6.19841993e-01 -9.89134014e-01
4.26639877e-02 9.81884599e-01 -1.59813583e-01 1.55288298e-02
-8.63415480e-01 -7.64339328e-01 7.58361816e-01 8.17210197e-01
7.85503164e-02 -1.21743417e+00 -4.58788812e-01 5.46539485e-01
-3.65523338e-01 -1.83468366e+00 -3.06723803e-01 3.50443423e-01
-2.16774315e-01 -1.16146302e+00 1.56799138e-01 -5.70136309e-01
3.41096818e-01 1.14136161e-02 1.71383989e+00 1.68036357e-01
-6.28141284e-01 3.98399144e-01 -5.68530917e-01 -6.97916210e-01
-1.90772951e-01 1.85150921e-01 3.45026962e-02 -3.34623307e-01
1.16573679e+00 -8.47251713e-01 -1.29903421e-01 -3.07106506e-02
-8.87539923e-01 -1.83921441e-01 5.48119128e-01 4.56533819e-01
3.54483664e-01 3.74104619e-01 8.02679718e-01 -1.27443659e+00
5.62477410e-01 -9.67272401e-01 -1.50058836e-01 2.58742839e-01
-1.03051317e+00 -2.88954884e-01 6.04906440e-01 5.73500022e-02
-1.06262815e+00 -4.66460466e-01 -2.89525121e-01 -1.55970991e-01
-6.10626817e-01 9.92430210e-01 6.20498769e-02 3.91301215e-01
3.83261442e-01 -2.78818548e-01 -6.85213685e-01 -7.45551288e-01
2.48627469e-01 1.23964655e+00 5.49503028e-01 -6.04960501e-01
7.71779120e-01 3.03611249e-01 -3.72157186e-01 -4.81594026e-01
-1.05698574e+00 -1.01117539e+00 -6.06999159e-01 9.69827399e-02
9.00334537e-01 -6.20342255e-01 -5.61438918e-01 2.76130229e-01
-1.50632775e+00 -2.53142953e-01 -1.92552015e-01 1.62783012e-01
-8.43657851e-02 -5.35774529e-02 -1.03566015e+00 -5.90519547e-01
-6.27125263e-01 -8.25670421e-01 1.06994939e+00 7.18979836e-02
-2.40822539e-01 -1.09912527e+00 2.42176756e-01 5.82459509e-01
6.66603088e-01 4.18092817e-01 1.12294471e+00 -1.29639816e+00
-4.55432504e-01 -4.88818735e-01 -4.32258040e-01 2.25109294e-01
1.48712352e-01 -4.67462353e-02 -9.31673169e-01 -5.01217786e-03
-1.83920979e-01 -6.43654242e-02 7.33155787e-01 -5.99860214e-02
9.46405888e-01 -3.79989237e-01 -5.36978185e-01 5.56349516e-01
1.54268503e+00 3.74038428e-01 3.09832335e-01 7.34188497e-01
8.84449899e-01 5.93929112e-01 5.18353105e-01 5.97431481e-01
9.57391739e-01 6.60998464e-01 1.82335153e-01 -1.39917970e-01
3.39662768e-02 -8.83639306e-02 4.40017181e-03 1.30057311e+00
-1.52536184e-01 -1.50060549e-01 -1.67896032e+00 9.14585292e-01
-1.76019084e+00 -8.49180520e-01 -1.29565209e-01 1.83747125e+00
1.01955855e+00 2.68969506e-01 -3.27294260e-01 4.55745775e-03
5.90626657e-01 -1.82674184e-01 -7.86690488e-02 -6.79120600e-01
3.31431776e-01 7.50974298e-01 6.08316302e-01 2.21532822e-01
-1.10423970e+00 1.05318725e+00 5.58463812e+00 3.06433380e-01
-1.05602872e+00 5.39030790e-01 2.21150249e-01 -1.29580542e-01
-1.85352758e-01 1.55026793e-01 -1.08041883e+00 2.14360267e-01
1.52353287e+00 -3.06753397e-01 4.16760653e-01 9.28911328e-01
1.84112012e-01 2.39096314e-01 -1.09714866e+00 8.09026301e-01
1.76812902e-01 -1.55936003e+00 -3.65097344e-01 2.31933128e-02
6.59734249e-01 4.53353822e-01 -1.94000348e-01 6.64038479e-01
5.68548083e-01 -1.13260913e+00 3.24029058e-01 5.62266529e-01
8.95492077e-01 -8.37941527e-01 1.15197897e+00 2.15252101e-01
-9.18387949e-01 -2.79772669e-01 1.05796136e-01 -3.71092819e-02
6.44057810e-01 9.82196391e-01 -1.42293632e+00 8.17231655e-01
9.10364449e-01 5.51602900e-01 -9.72112268e-02 9.59815562e-01
-2.15352967e-01 8.11763048e-01 -7.86430016e-02 6.60582781e-01
7.98113830e-03 2.51048088e-01 3.59271079e-01 1.63820624e+00
2.57675558e-01 7.08527565e-02 6.31163180e-01 5.72035015e-01
-3.20167035e-01 -1.32345170e-01 -6.56892478e-01 -3.80030811e-01
5.92459202e-01 1.53693509e+00 -6.22148633e-01 -3.79732937e-01
-7.20452785e-01 6.44516468e-01 3.03311616e-01 3.83696198e-01
-8.22522461e-01 -6.38235688e-01 9.41751778e-01 1.97192252e-01
4.17797595e-01 -3.63201082e-01 -3.91298741e-01 -1.15572023e+00
1.12444341e-01 -9.85919774e-01 4.82868791e-01 -4.05897379e-01
-1.48116100e+00 1.17281318e+00 -1.13698684e-01 -6.42031014e-01
-4.45705205e-01 -4.05667514e-01 -6.12536073e-01 7.54047394e-01
-1.77616084e+00 -1.39776850e+00 -1.48815036e-01 4.19649452e-01
3.66385490e-01 -9.73698199e-02 1.09375036e+00 9.55680251e-01
-9.14325535e-01 5.20594537e-01 -4.20515150e-01 5.26583552e-01
9.37145650e-01 -1.24443507e+00 8.22162032e-01 1.04471970e+00
5.19383967e-01 9.05148745e-01 4.49837923e-01 -7.63252437e-01
-1.21167576e+00 -1.32533503e+00 1.57038438e+00 -8.97453785e-01
9.70106006e-01 -3.96445513e-01 -9.12880361e-01 1.08822167e+00
2.21861482e-01 -1.12199280e-02 1.18420124e+00 6.10369563e-01
-7.47083485e-01 -8.42348114e-02 -1.08882976e+00 1.41448788e-02
9.05439496e-01 -9.07967806e-01 -5.88570714e-01 4.89789724e-01
1.17867184e+00 -2.53356665e-01 -1.19804418e+00 4.45678502e-01
9.46614891e-02 -6.72632039e-01 5.59292257e-01 -1.24174428e+00
4.92415488e-01 6.06284402e-02 -2.90322900e-01 -1.22120130e+00
-2.41995916e-01 -4.67081785e-01 -2.58380502e-01 1.64545155e+00
7.56773293e-01 -6.21388376e-01 4.70941156e-01 9.27878916e-01
-4.41956818e-01 -7.78043330e-01 -5.42521656e-01 -4.63842392e-01
-3.47417593e-01 -1.02323878e+00 7.95191824e-01 1.32326317e+00
1.77024081e-01 2.15414047e-01 -6.74713403e-02 5.47805369e-01
3.07462692e-01 2.41474986e-01 4.74801004e-01 -1.18436491e+00
-3.91464859e-01 -1.03712380e-02 -3.06926310e-01 -2.44546562e-01
2.97792286e-01 -9.72718477e-01 -4.63157594e-02 -1.89941323e+00
2.52659231e-01 -7.04185307e-01 -6.81056261e-01 1.21057129e+00
1.67438149e-01 7.02577457e-02 4.82573584e-02 1.78275183e-01
-8.35391819e-01 -3.48375320e-01 2.97161818e-01 1.52083904e-01
-1.10640906e-01 -2.66940352e-02 -9.35056090e-01 6.55541182e-01
6.67753100e-01 -8.44740629e-01 1.74364120e-01 -7.81039417e-01
6.02251768e-01 7.95957297e-02 2.22550817e-02 -6.93654537e-01
5.27052104e-01 -1.26813412e-01 -8.93330574e-02 -8.26921314e-02
-8.66341889e-02 -8.87923002e-01 -1.50500655e-01 -1.00707307e-01
-9.28345621e-02 3.86850864e-01 2.32690856e-01 2.02977479e-01
-3.79564852e-01 -8.58353171e-03 1.73100457e-01 -1.20223649e-01
-9.47429359e-01 3.93095106e-01 -1.85638309e-01 2.58050203e-01
9.04728353e-01 2.50032037e-01 -5.88916183e-01 4.91647515e-03
-7.84259319e-01 4.98824008e-02 2.94474512e-01 7.16620326e-01
4.65104342e-01 -9.75628674e-01 -6.02449715e-01 1.67181045e-01
1.51959792e-01 -4.32186313e-02 -1.74430925e-02 9.66272950e-01
-1.42767653e-01 5.44107258e-01 6.25070855e-02 -1.74958736e-01
-1.25350022e+00 5.70594013e-01 -9.31637511e-02 -6.85688615e-01
-6.16563916e-01 6.98707283e-01 -5.54242611e-01 -4.39623773e-01
1.16345733e-01 -4.45471823e-01 -2.36076713e-01 -9.52232536e-03
7.38663614e-01 3.36863697e-01 6.79068625e-01 -4.09008592e-01
-6.74729109e-01 7.86162615e-02 -4.32252079e-01 5.01233637e-02
2.13234496e+00 1.26416534e-01 -6.24192059e-01 3.42897803e-01
1.04451215e+00 2.31131151e-01 -4.90459532e-01 -2.69288123e-01
7.61212528e-01 -4.41943258e-01 1.31992668e-01 -1.22319829e+00
-1.02837658e+00 6.06689870e-01 -7.69293085e-02 5.72218359e-01
1.21574593e+00 6.03529029e-02 1.27682531e+00 2.75286585e-01
8.44091892e-01 -8.99952114e-01 -2.17900246e-01 7.57453024e-01
3.77233297e-01 -1.17645299e+00 -3.66333634e-01 -3.56079072e-01
-5.33281446e-01 1.00144160e+00 5.56469738e-01 1.15560509e-01
6.79459512e-01 1.29322231e+00 2.53074408e-01 -5.26087821e-01
-1.07826757e+00 -2.62100130e-01 -3.54644135e-02 6.78408384e-01
5.70577443e-01 1.91022530e-01 2.67845809e-01 1.22535861e+00
3.10290139e-02 1.71965346e-01 5.75411975e-01 1.03279757e+00
-2.35648513e-01 -1.46947706e+00 -2.12635547e-02 6.23557091e-01
-1.11915910e+00 -6.36819065e-01 -1.99727729e-01 5.68546712e-01
3.35413694e-01 1.22811973e+00 -2.78258268e-02 -7.62717962e-01
6.58397496e-01 3.82093132e-01 1.10358939e-01 -1.17693269e+00
-1.08289933e+00 -3.55495214e-01 5.62426090e-01 -7.40612328e-01
-6.72688112e-02 -5.49460053e-01 -1.56790996e+00 -2.69413739e-01
-2.58521318e-01 3.17127496e-01 9.88196313e-01 1.34178257e+00
8.25554669e-01 8.76122415e-01 5.38116276e-01 -3.98768872e-01
-4.64615464e-01 -9.45501029e-01 -3.34094286e-01 2.99576938e-01
1.03599496e-01 -3.63756746e-01 -1.06864274e-01 9.58453193e-02]
|
[9.663893699645996, 9.371321678161621]
|
612e842a-abd3-4a40-8f96-c902a6c55258
|
context-aware-adaptive-and-scalable-android
|
1706.00947
| null |
http://arxiv.org/abs/1706.00947v2
|
http://arxiv.org/pdf/1706.00947v2.pdf
|
Context-aware, Adaptive and Scalable Android Malware Detection through Online Learning (extended version)
|
It is well-known that Android malware constantly evolves so as to evade
detection. This causes the entire malware population to be non-stationary.
Contrary to this fact, most of the prior works on Machine Learning based
Android malware detection have assumed that the distribution of the observed
malware characteristics (i.e., features) does not change over time. In this
work, we address the problem of malware population drift and propose a novel
online learning based framework to detect malware, named CASANDRA
(Contextaware, Adaptive and Scalable ANDRoid mAlware detector). In order to
perform accurate detection, a novel graph kernel that facilitates capturing
apps' security-sensitive behaviors along with their context information from
dependency graphs is proposed. Besides being accurate and scalable, CASANDRA
has specific advantages: i) being adaptive to the evolution in malware features
over time ii) explaining the significant features that led to an app's
classification as being malicious or benign. In a large-scale comparative
analysis, CASANDRA outperforms two state-of-the-art techniques on a benchmark
dataset achieving 99.23% F-measure. When evaluated with more than 87,000 apps
collected in-the-wild, CASANDRA achieves 89.92% accuracy, outperforming
existing techniques by more than 25% in their typical batch learning setting
and more than 7% when they are continuously retained, while maintaining
comparable efficiency.
|
['Mahinthan Chandramohan', 'Annamalai Narayanan', 'Yang Liu', 'Lihui Chen']
|
2017-06-03
| null | null | null | null |
['android-malware-detection']
|
['miscellaneous']
|
[ 1.28615618e-01 -4.42562103e-01 -7.12037563e-01 8.95344540e-02
-3.56886685e-01 -6.07195735e-01 5.18410146e-01 3.10177326e-01
-1.56494007e-01 4.92060781e-01 -4.62153912e-01 -6.38492942e-01
-3.19489948e-02 -6.25230074e-01 -7.03617156e-01 -5.59697628e-01
-6.57556117e-01 2.74088621e-01 6.76390707e-01 8.23291987e-02
1.96127251e-01 2.88756937e-01 -1.59028029e+00 1.47951767e-01
8.22485566e-01 9.02396500e-01 -1.25813603e-01 7.75297344e-01
7.00611770e-02 7.98970938e-01 -9.36896324e-01 -5.93185365e-01
5.55503629e-02 -2.81418055e-01 -4.88823354e-01 -1.09707922e-01
1.06475040e-01 -2.90587217e-01 -1.72802821e-01 1.13693607e+00
-8.88140351e-02 -3.64721388e-01 4.74502504e-01 -1.45998538e+00
-4.34749335e-01 3.78520221e-01 -7.72863090e-01 5.39476514e-01
3.21302980e-01 8.65335241e-02 7.96294510e-01 -2.48150438e-01
5.03974020e-01 8.69836926e-01 9.22568381e-01 5.27116179e-01
-1.07851577e+00 -6.06056690e-01 3.36631835e-01 4.10172075e-01
-1.31092989e+00 -2.44116932e-01 9.59095240e-01 -5.24407566e-01
8.18066180e-01 4.21991646e-01 9.40015733e-01 1.51467681e+00
6.84749007e-01 5.44734359e-01 1.02375424e+00 1.79949738e-02
4.68177497e-01 1.86071172e-01 3.38422835e-01 6.61520660e-01
7.06129491e-01 -1.01649620e-01 -2.75783271e-01 -8.94321322e-01
-3.89797129e-02 5.61672509e-01 1.38663575e-01 -3.87556762e-01
-4.77812946e-01 8.05349588e-01 1.05291739e-01 2.60471016e-01
-3.38676989e-01 -7.96396509e-02 8.31451535e-01 9.18920040e-02
5.73532283e-01 2.28191521e-02 -8.06181788e-01 -7.10060120e-01
-6.56114042e-01 -5.59694655e-02 9.34996903e-01 3.91326398e-01
7.23390222e-01 1.89284578e-01 2.43743345e-01 5.19610941e-01
3.70550036e-01 4.18052882e-01 1.06774187e+00 -4.23312366e-01
8.53749067e-02 9.98910248e-01 -3.61847967e-01 -1.28221679e+00
-1.44044785e-02 -4.89069611e-01 -5.02612054e-01 -1.48851588e-01
2.50474662e-01 -5.36892414e-02 -6.21117592e-01 1.57937860e+00
4.98570681e-01 7.59337008e-01 -1.42069981e-01 -1.25316039e-01
2.53266394e-01 5.80532491e-01 6.63667619e-02 -6.89548135e-01
1.33871460e+00 -4.57768977e-01 -6.01458013e-01 -2.30964601e-01
6.52385592e-01 -2.89745867e-01 1.28756654e+00 5.18156588e-01
-2.80362815e-01 -1.42177343e-01 -1.02112722e+00 8.43815267e-01
-6.08291507e-01 -2.47662961e-01 6.54827178e-01 1.09770727e+00
-9.46307063e-01 3.37434590e-01 -1.26468611e+00 -3.75926971e-01
7.79274106e-01 5.16262889e-01 7.67526999e-02 2.28044301e-01
-6.21612728e-01 1.36928335e-01 2.44455740e-01 -5.60233474e-01
-1.20201170e+00 -6.63844943e-01 -4.33838636e-01 -9.68213528e-02
8.77365530e-01 -7.40809217e-02 1.32304454e+00 -1.10372591e+00
-1.40433216e+00 4.48167682e-01 -4.13280696e-01 -7.38444269e-01
3.81119341e-01 -2.19190925e-01 -7.71239877e-01 -1.55734107e-01
-1.35922506e-01 -2.55462587e-01 1.23501480e+00 -1.27521384e+00
-5.69964290e-01 -7.47893155e-01 -1.50697187e-01 -4.38665062e-01
-9.75244224e-01 -9.68718603e-02 -4.06000316e-01 -4.03367311e-01
-5.14891386e-01 -1.24525738e+00 1.06792159e-01 -7.27396309e-01
-4.43348587e-01 -2.35104650e-01 1.92459917e+00 -6.86367452e-01
1.72184503e+00 -2.01705885e+00 -3.70289505e-01 1.43015310e-01
4.64943647e-01 8.20410311e-01 1.82240441e-01 3.60939920e-01
2.09863514e-01 1.97006747e-01 -4.74658966e-01 -1.71139479e-01
-5.45201540e-01 2.73603857e-01 -3.15970868e-01 4.61539537e-01
-8.48997757e-02 8.04923296e-01 -9.86142814e-01 -2.78712213e-01
-6.04817923e-03 7.31774449e-01 -5.32833338e-01 3.53245251e-02
-3.55504215e-01 3.79119873e-01 -4.24811691e-01 9.13866758e-01
4.53628838e-01 -3.98349643e-01 5.96690118e-01 4.21646982e-01
3.37249428e-01 1.57369915e-02 -6.27174497e-01 7.22056925e-01
-4.26984966e-01 7.00494170e-01 -1.08439326e-01 -8.98928106e-01
7.69689083e-01 4.16705050e-02 8.09746981e-01 -4.05703515e-01
2.40839541e-01 1.86033159e-01 3.01762819e-01 -4.71963674e-01
2.64955014e-01 8.06556404e-01 1.56177402e-01 4.84448135e-01
-2.91934133e-01 5.58366358e-01 8.64487588e-02 2.14529663e-01
1.66211319e+00 -3.86200756e-01 6.37140810e-01 -9.90518034e-02
8.66764367e-01 -1.64414465e-01 5.49699903e-01 5.02273321e-01
-5.23675203e-01 -3.75330269e-01 8.25181782e-01 -3.86131048e-01
-4.60608751e-01 -9.26027775e-01 2.54150569e-01 1.23348272e+00
8.36121067e-02 -9.11516488e-01 -1.08520794e+00 -1.40517366e+00
1.06703438e-01 3.78294438e-01 -8.47927272e-01 -4.47112530e-01
-6.18102252e-01 -1.03296161e+00 4.18573558e-01 6.83990195e-02
5.30541360e-01 -8.02867472e-01 -6.42011285e-01 7.45026767e-02
2.29376629e-01 -1.06798518e+00 -2.82157093e-01 -9.17130858e-02
-9.58511651e-01 -1.67832029e+00 5.77841653e-04 -3.61883253e-01
5.88582337e-01 4.58931327e-01 8.83364618e-01 3.60245794e-01
-3.45018685e-01 6.34268165e-01 -3.76974493e-01 -4.92997319e-01
-8.12215745e-01 3.37790847e-01 1.06604807e-01 2.60542542e-01
5.13518929e-01 -6.04450166e-01 -4.59807187e-01 1.93716317e-01
-8.20066571e-01 -9.37215447e-01 5.08864224e-01 5.70284426e-01
5.97308934e-01 5.64700067e-01 5.85477471e-01 -1.24858451e+00
8.78269255e-01 -9.70378280e-01 -6.55907512e-01 1.67260602e-01
-1.08874428e+00 -4.42440361e-01 1.04117882e+00 -1.10151768e+00
-7.66170681e-01 1.05444685e-01 5.72200343e-02 -2.56349385e-01
5.58084138e-02 1.01257376e-01 -2.76034623e-01 -1.81580007e-01
6.41175747e-01 4.95106101e-01 1.74111009e-01 -2.14867830e-01
3.66574898e-03 8.07958007e-01 2.25661010e-01 -3.93892638e-02
8.24511826e-01 5.16506612e-01 -8.12730491e-02 -1.08237457e+00
-3.72766435e-01 -4.05851275e-01 -4.42889124e-01 -1.00750469e-01
4.91801530e-01 -5.63866556e-01 -9.34104025e-01 7.51318038e-01
-5.90561092e-01 -4.08998460e-01 6.51169494e-02 3.46480496e-02
-9.91780013e-02 5.95631540e-01 -4.74721789e-01 -1.14533186e+00
-4.20377702e-01 -1.16240704e+00 8.61336768e-01 1.59620166e-01
-2.41425663e-01 -1.19796503e+00 2.29458585e-01 8.73051211e-02
4.65282589e-01 2.41398200e-01 7.87982941e-01 -1.07360017e+00
-2.03926310e-01 -5.17496943e-01 1.86087385e-01 3.31217706e-01
6.50566220e-01 5.14745951e-01 -9.46670473e-01 -5.96513689e-01
2.56106585e-01 3.37550730e-01 5.90092301e-01 2.14475900e-01
1.44197106e+00 -9.21423554e-01 -7.19502151e-01 3.51076156e-01
1.21205437e+00 5.89884698e-01 3.03434670e-01 3.13901544e-01
7.69773483e-01 2.97935247e-01 4.68524396e-01 4.38085258e-01
4.39526469e-01 6.75424695e-01 9.56214070e-01 4.02614862e-01
2.06517264e-01 -4.08129662e-01 9.40395176e-01 7.38430142e-01
2.24012896e-01 -1.12052493e-01 -9.92733836e-01 3.28115821e-01
-1.70001864e+00 -9.16139483e-01 -1.06406219e-01 2.49227667e+00
6.00453794e-01 4.03942376e-01 8.14104676e-01 2.42990226e-01
6.21494770e-01 2.37021208e-01 -8.42230141e-01 -5.65609336e-01
2.79762477e-01 -8.21072534e-02 5.49163580e-01 3.22070181e-01
-1.17832744e+00 8.79777133e-01 5.79659367e+00 8.88001919e-01
-1.48429835e+00 4.53574687e-01 6.91071987e-01 2.25442693e-01
1.12937495e-01 -6.07403293e-02 -7.31172085e-01 9.90389645e-01
1.36460721e+00 -1.96601391e-01 6.10459566e-01 1.29537237e+00
-2.09006313e-02 -2.31171343e-02 -5.95196009e-01 9.56700563e-01
3.00251126e-01 -9.85057592e-01 -1.87365577e-01 5.25224268e-01
7.36881196e-01 8.28574672e-02 3.85374457e-01 3.81214112e-01
2.89482802e-01 -6.26266420e-01 9.71231684e-02 4.00409773e-02
6.49524212e-01 -8.71551454e-01 5.43506801e-01 6.01196289e-01
-1.30771577e+00 -8.64635825e-01 7.18501285e-02 -4.48108949e-02
-3.36339623e-01 7.09750891e-01 -1.24786627e+00 2.58435786e-01
8.39655280e-01 9.46794629e-01 -1.20343578e+00 6.55986488e-01
-8.65551308e-02 1.34142983e+00 4.16335352e-02 -1.57782838e-01
-2.50687413e-02 3.66207510e-02 6.12648249e-01 9.87134039e-01
1.81524470e-01 -4.93689299e-01 1.19388789e-01 2.26771280e-01
-8.64386410e-02 1.91394702e-01 -7.99214244e-01 -3.27483505e-01
6.22409463e-01 1.27126288e+00 -8.50764632e-01 -5.13898730e-01
-3.27019989e-01 8.06870937e-01 9.72731560e-02 1.39073268e-01
-9.04954612e-01 -9.90843773e-02 8.55290294e-01 4.45845008e-01
3.83413881e-01 -3.64574939e-01 7.12673739e-02 -9.31377292e-01
1.17513919e-02 -1.14339244e+00 4.01547968e-01 1.69065937e-01
-1.21957457e+00 7.93835402e-01 -1.05645783e-01 -1.04867435e+00
-6.04276419e-01 -6.05938375e-01 -8.57942581e-01 -1.83183506e-01
-1.18754911e+00 -9.78780091e-01 -3.81550044e-01 6.89787626e-01
5.20360768e-01 -3.76439333e-01 5.78623176e-01 1.40531167e-01
-8.61545622e-01 7.43652999e-01 1.91157073e-01 -1.84890047e-01
3.77768010e-01 -1.05057955e+00 4.46811497e-01 8.69923055e-01
2.65707821e-01 7.42904723e-01 4.94655877e-01 -1.25768495e+00
-1.77890170e+00 -1.27060628e+00 3.13126683e-01 -6.22258544e-01
1.05303288e+00 -5.40162444e-01 -1.21226716e+00 6.40696943e-01
-1.55580521e-01 1.10330433e-01 6.49908423e-01 -1.39760524e-01
-6.02814734e-01 -4.43196982e-01 -1.25532055e+00 4.81654555e-01
1.01822186e+00 -3.86787146e-01 9.25448164e-02 3.24000865e-01
6.32014930e-01 -5.66268452e-02 -5.49722433e-01 5.66102505e-01
5.59516549e-01 -1.02416015e+00 8.02550018e-01 -5.45166969e-01
-2.06344783e-01 -7.97026977e-03 -2.19466612e-02 -7.41728842e-01
8.03725645e-02 -9.63739574e-01 -1.22984743e+00 1.13293099e+00
1.52611151e-01 -1.06973660e+00 9.12467718e-01 -1.88835844e-01
3.35969418e-01 -1.20197487e+00 -8.98709714e-01 -1.20015824e+00
-2.79041171e-01 -4.34916437e-01 7.69151151e-01 8.60240996e-01
-3.16231459e-01 1.32146239e-01 -3.85596395e-01 9.34009627e-02
5.71416557e-01 -4.14536566e-01 9.83851194e-01 -1.49292541e+00
-3.84922057e-01 -5.20173132e-01 -5.85540652e-01 -4.77449507e-01
4.20492500e-01 -6.14970803e-01 -4.63160664e-01 -7.05068409e-01
4.40254688e-01 -2.17668936e-01 -2.02720329e-01 4.13828790e-01
-2.80158639e-01 1.90595821e-01 -2.10122406e-01 4.57656652e-01
-6.89367592e-01 1.93397015e-01 4.45969522e-01 -4.15423512e-02
-6.92231119e-01 5.63874364e-01 -6.28527224e-01 8.17963302e-01
1.17127514e+00 -4.24996018e-01 -6.55632436e-01 1.92124918e-01
2.09198203e-02 -6.43115520e-01 2.98894972e-01 -9.15854752e-01
-1.33690819e-01 9.04358458e-03 1.34485438e-01 -4.28648889e-01
1.99080512e-01 -8.78270268e-01 1.37536779e-01 1.08653009e+00
1.82225361e-01 4.82607245e-01 1.81988835e-01 9.78399098e-01
1.68990478e-01 1.77456662e-01 5.57798266e-01 2.04523653e-01
-6.54737055e-01 5.49127758e-01 -5.00153482e-01 3.26184779e-02
1.58278811e+00 -3.17828715e-01 -5.10881424e-01 -3.14592063e-01
-1.00104332e-01 -3.35514069e-01 8.06017041e-01 6.00021780e-01
4.23495710e-01 -8.24024737e-01 -1.86175540e-01 2.60498047e-01
1.05858244e-01 -6.97353244e-01 1.44521177e-01 8.88513863e-01
-1.46064177e-01 2.57725060e-01 1.65386349e-01 -9.07713652e-01
-1.84516537e+00 8.41805696e-01 9.92326215e-02 -6.05582476e-01
-3.89648199e-01 3.12752962e-01 -4.50213581e-01 -1.52737141e-01
2.45427683e-01 1.85432628e-01 -2.40726709e-01 -1.33058932e-02
6.64975047e-01 7.36708760e-01 2.18601152e-01 -8.22369874e-01
-7.53716826e-01 3.54295135e-01 -2.06847638e-01 4.74235564e-01
1.16274786e+00 1.94638088e-01 -2.53494829e-01 5.39259017e-01
1.32787681e+00 6.66822016e-01 -1.14146459e+00 -3.71461436e-02
3.39384198e-01 -5.90638518e-01 -5.25801599e-01 -5.81714928e-01
-1.12159431e+00 4.69798356e-01 9.94837999e-01 8.75474811e-01
1.07580149e+00 2.83552241e-02 8.74927580e-01 2.04716399e-01
3.47999871e-01 -6.04800165e-01 4.38725799e-01 3.91807973e-01
1.90530732e-01 -1.32716930e+00 5.01883663e-02 -4.95580375e-01
-4.77541715e-01 8.26122582e-01 6.68755293e-01 -4.92451340e-02
8.18757415e-01 2.58550763e-01 -2.27922037e-01 -2.11110145e-01
-7.20948040e-01 3.09017837e-01 1.95997491e-01 9.45638895e-01
8.21529850e-02 3.07235956e-01 -1.87995523e-01 3.72061759e-01
-1.84426859e-01 -3.11734408e-01 4.35074419e-01 9.26526248e-01
-3.37118387e-01 -1.26922405e+00 -1.56160638e-01 8.05535316e-01
-7.80061305e-01 2.84064770e-01 -9.01316762e-01 7.53958344e-01
1.40017420e-01 1.03475165e+00 -1.34487420e-01 -8.62090826e-01
-9.69159901e-02 -6.81674406e-02 4.24475083e-03 -5.71102262e-01
-5.98908365e-01 -3.00367534e-01 -2.88800657e-01 -6.50568604e-01
3.41740921e-02 -7.41193056e-01 -1.06271791e+00 -3.09384257e-01
-2.39789367e-01 4.71415035e-02 7.87064552e-01 7.64610112e-01
7.88364232e-01 5.15613556e-01 1.01583564e+00 -4.36467677e-01
-5.31529784e-01 -6.72804594e-01 -1.10133871e-01 2.35659793e-01
3.96216094e-01 -7.24074125e-01 -5.39489985e-01 -1.60930172e-01]
|
[14.422598838806152, 9.680986404418945]
|
fb8846a2-502d-4e67-baf5-d0b72038c9b9
|
entropic-causal-inference-identifiability-and-1
|
2101.03501
| null |
https://arxiv.org/abs/2101.03501v1
|
https://arxiv.org/pdf/2101.03501v1.pdf
|
Entropic Causal Inference: Identifiability and Finite Sample Results
|
Entropic causal inference is a framework for inferring the causal direction between two categorical variables from observational data. The central assumption is that the amount of unobserved randomness in the system is not too large. This unobserved randomness is measured by the entropy of the exogenous variable in the underlying structural causal model, which governs the causal relation between the observed variables. Kocaoglu et al. conjectured that the causal direction is identifiable when the entropy of the exogenous variable is not too large. In this paper, we prove a variant of their conjecture. Namely, we show that for almost all causal models where the exogenous variable has entropy that does not scale with the number of states of the observed variables, the causal direction is identifiable from observational data. We also consider the minimum entropy coupling-based algorithmic approach presented by Kocaoglu et al., and for the first time demonstrate algorithmic identifiability guarantees using a finite number of samples. We conduct extensive experiments to evaluate the robustness of the method to relaxing some of the assumptions in our theory and demonstrate that both the constant-entropy exogenous variable and the no latent confounder assumptions can be relaxed in practice. We also empirically characterize the number of observational samples needed for causal identification. Finally, we apply the algorithm on Tuebingen cause-effect pairs dataset.
|
['Dmitriy Katz', 'Kristjan Greenewald', 'Murat Kocaoglu', 'Spencer Compton']
|
2021-01-10
|
entropic-causal-inference-identifiability-and
|
http://proceedings.neurips.cc/paper/2020/hash/a979ca2444b34449a2c80b012749e9cd-Abstract.html
|
http://proceedings.neurips.cc/paper/2020/file/a979ca2444b34449a2c80b012749e9cd-Paper.pdf
|
neurips-2020-12
|
['causal-identification']
|
['reasoning']
|
[ 2.17111096e-01 3.16084743e-01 -4.87397194e-01 2.89480891e-02
-3.23161125e-01 -7.44854569e-01 7.28353083e-01 1.18644997e-01
3.25860195e-02 1.06474078e+00 4.66823846e-01 -5.87880373e-01
-5.83223820e-01 -8.28059018e-01 -9.61872399e-01 -9.13289368e-01
-5.29232919e-01 4.77237940e-01 -1.95988610e-01 1.74559921e-01
-1.58657804e-02 3.16954553e-01 -1.24035871e+00 -3.31074327e-01
7.69301772e-01 3.30762982e-01 -9.01215151e-02 7.87629962e-01
4.79153007e-01 8.47962081e-01 -2.43443266e-01 2.32692640e-02
3.02137434e-01 -7.99414814e-01 -9.19017076e-01 -3.85480165e-01
4.36731018e-02 -1.84037358e-01 -5.56430995e-01 1.11124420e+00
1.16561115e-01 -4.97493893e-02 9.01319742e-01 -1.32998800e+00
-4.95938480e-01 9.96278763e-01 -4.27784622e-01 2.01395363e-01
1.72443658e-01 4.18863557e-02 1.33760607e+00 -2.65022725e-01
6.27033770e-01 1.29786217e+00 3.55076134e-01 2.19299778e-01
-1.77283847e+00 -7.80644953e-01 1.75382402e-02 9.13482234e-02
-1.33361018e+00 -2.77006149e-01 6.71995401e-01 -8.29131782e-01
4.70679909e-01 4.48288023e-01 4.62548971e-01 1.18009531e+00
2.40487531e-01 1.53022170e-01 1.09809792e+00 -5.58475137e-01
5.65785587e-01 -1.56981707e-01 3.15256119e-01 6.65684521e-01
6.78518295e-01 8.11506748e-01 -3.88681740e-01 -6.81438565e-01
6.61141932e-01 -1.42726183e-01 -2.69965112e-01 -2.79909253e-01
-1.31779683e+00 1.16831267e+00 1.03237167e-01 2.53277391e-01
-2.67915308e-01 5.30975282e-01 1.61234424e-01 3.25240910e-01
2.25402266e-01 5.07596552e-01 -5.23092687e-01 4.05192822e-02
-5.33788383e-01 2.06628636e-01 9.10402775e-01 6.17253721e-01
4.87612814e-01 -3.20727319e-01 7.17499778e-02 5.69324121e-02
7.46869668e-02 8.99809241e-01 -3.94611806e-02 -1.22886252e+00
3.18063289e-01 3.97615761e-01 3.84420067e-01 -9.60129499e-01
-4.43339586e-01 -8.69705379e-02 -1.08252907e+00 -2.50434905e-01
6.74768627e-01 -3.36745441e-01 -5.81693411e-01 2.37321830e+00
2.74609566e-01 3.99682224e-01 -1.16542086e-01 7.39341855e-01
2.51095206e-01 4.66242373e-01 2.26527695e-02 -7.81902671e-01
1.16542864e+00 -5.27744219e-02 -8.64729643e-01 2.74542689e-01
6.90390170e-01 -5.11382639e-01 8.34509909e-01 -3.61641333e-03
-8.57127249e-01 -7.05427155e-02 -5.88139832e-01 1.77992329e-01
-2.68097762e-02 -1.79465830e-01 8.99506092e-01 7.00263500e-01
-7.10021138e-01 4.64667588e-01 -9.71183240e-01 -2.24537820e-01
-1.95783511e-01 5.18306613e-01 -2.43619293e-01 4.19618458e-01
-1.45508194e+00 4.77415979e-01 2.66330510e-01 -9.18967202e-02
-1.00117159e+00 -8.23978901e-01 -4.40590918e-01 2.81345278e-01
4.68421400e-01 -8.02045703e-01 9.96811152e-01 -6.42712533e-01
-1.07224214e+00 2.50630528e-01 -5.44052482e-01 -3.82312328e-01
4.17508453e-01 2.20270693e-01 -2.63461143e-01 1.41792804e-01
6.35991469e-02 -1.35566806e-02 4.27380919e-01 -1.13117111e+00
-2.85563767e-01 -3.81625384e-01 8.72397348e-02 -2.52233833e-01
9.44425017e-02 -2.63703585e-01 2.67113507e-01 -3.55573326e-01
1.48170024e-01 -1.27646196e+00 -2.25610822e-01 -5.99923432e-01
-6.53060198e-01 -7.97755346e-02 5.25660753e-01 -2.31390923e-01
1.17231834e+00 -2.00688601e+00 4.02746379e-01 3.72983485e-01
4.79082555e-01 -5.86758018e-01 1.34736136e-01 6.41801357e-01
-5.08755803e-01 6.17628753e-01 -3.83286655e-01 3.65145504e-02
-6.85421377e-02 2.00095952e-01 -6.53758526e-01 8.68925035e-01
-7.56436288e-02 8.13941121e-01 -9.79956150e-01 -2.96321273e-01
1.24177843e-01 7.57213682e-02 -6.99787736e-01 1.70759082e-01
-1.55179605e-01 7.81041563e-01 -4.53204840e-01 3.46691646e-02
4.63958979e-01 -3.96613330e-01 5.92695832e-01 2.40674183e-01
-1.50027722e-01 6.42288506e-01 -1.45106518e+00 7.86207318e-01
-2.53656715e-01 5.71623802e-01 -2.15487078e-01 -1.02776814e+00
2.81735271e-01 6.88838661e-01 7.07517385e-01 -2.10991800e-01
1.24135055e-01 5.87831903e-03 3.78207564e-01 -4.92087066e-01
9.24448743e-02 -5.82843244e-01 -4.29970741e-01 5.25591016e-01
-2.47731507e-01 3.06779653e-01 3.66924740e-02 3.90236735e-01
1.25617194e+00 -4.67523098e-01 6.27157986e-01 -5.93534470e-01
9.74409208e-02 -1.72080308e-01 7.10876167e-01 1.08452117e+00
8.24739113e-02 1.56336054e-01 1.11693835e+00 -5.63635379e-02
-1.26153493e+00 -1.20321202e+00 -3.26852322e-01 4.50464427e-01
7.62926191e-02 -3.10004115e-01 -6.00430131e-01 -3.65807593e-01
1.35073110e-01 7.90848494e-01 -1.25063050e+00 -2.13935703e-01
-3.56278300e-01 -9.64275479e-01 4.84263629e-01 2.57333785e-01
1.90645501e-01 -5.00112414e-01 -5.43669522e-01 -5.73912896e-02
-4.96835053e-01 -7.75922656e-01 -3.55433971e-01 3.45495194e-01
-9.36802447e-01 -1.30418956e+00 1.86815277e-01 6.90091774e-02
6.20758176e-01 -8.20938498e-02 7.68673301e-01 -1.21784963e-01
3.73167656e-02 1.94227710e-01 2.29194202e-02 -1.57614008e-01
-6.69789433e-01 -9.45319310e-02 3.33280832e-01 -1.35259926e-01
2.67526716e-01 -7.01914847e-01 -4.60326165e-01 1.55827582e-01
-7.28357136e-01 -6.05343804e-02 1.44881517e-01 1.00176501e+00
3.31567794e-01 3.10219109e-01 3.45039606e-01 -7.75320232e-01
2.73989409e-01 -7.96469331e-01 -9.54215348e-01 9.64133441e-02
-5.40995419e-01 6.10165119e-01 8.10826778e-01 -4.58358169e-01
-8.33000898e-01 4.01696935e-02 4.86394495e-01 -1.59562126e-01
-1.00894548e-01 6.70050919e-01 -2.96794981e-01 5.05523145e-01
4.20559555e-01 -1.40408829e-01 -4.08883512e-01 -3.10050368e-01
3.86827677e-01 2.96574980e-01 2.26173192e-01 -6.70093417e-01
6.87589526e-01 7.62575030e-01 7.43267536e-01 -6.62483573e-01
-5.46475470e-01 -1.93293408e-01 -5.55186808e-01 5.24556227e-02
8.39451253e-01 -6.92586899e-01 -1.51159966e+00 -1.23622939e-01
-1.01834416e+00 -2.53267765e-01 -3.83889198e-01 9.22500253e-01
-6.69127345e-01 2.12126791e-01 -4.39510792e-01 -1.33561420e+00
3.95778209e-01 -1.10690260e+00 7.73069501e-01 -3.53056073e-01
-4.84759003e-01 -1.28631246e+00 3.55981618e-01 -1.03399515e-01
-1.99945465e-01 3.81173700e-01 1.05050230e+00 -3.12094986e-01
-6.04490519e-01 -7.56031722e-02 2.14301441e-02 -4.00214195e-01
2.87649989e-01 2.60194987e-01 -7.15793610e-01 -4.87664342e-02
1.87422380e-01 1.96834147e-01 9.57371771e-01 9.68429327e-01
1.01253796e+00 -7.06774414e-01 -4.82539922e-01 4.03867513e-01
1.35768139e+00 1.46154210e-01 3.41528118e-01 -1.55390456e-01
6.15256965e-01 7.04187751e-01 1.67871416e-01 4.03377950e-01
2.94412881e-01 6.58211768e-01 2.75204718e-01 4.25969362e-02
4.41690594e-01 -6.44354761e-01 3.03427130e-01 7.31377602e-01
-1.77704811e-01 -2.77051717e-01 -7.47480750e-01 6.30158305e-01
-1.82938397e+00 -1.38285065e+00 -7.38435745e-01 2.56980538e+00
1.10208595e+00 -1.98407754e-01 2.78383374e-01 -2.30695233e-02
6.85786068e-01 -3.30814540e-01 -4.75261837e-01 -2.85321504e-01
-2.50161737e-01 8.05865228e-02 1.21343529e+00 1.09514928e+00
-8.59341502e-01 3.51960063e-01 6.85548067e+00 2.68994212e-01
-7.49809563e-01 1.89776883e-01 5.11263847e-01 -1.67013511e-01
-5.67867935e-01 5.17147422e-01 -4.96798784e-01 6.60764694e-01
1.38202536e+00 -4.90547419e-01 5.31172037e-01 3.31687987e-01
7.69601107e-01 -3.19255441e-01 -1.35401356e+00 3.36801797e-01
-8.24981749e-01 -1.06648719e+00 -3.60300332e-01 6.92602336e-01
8.49662602e-01 -1.29574999e-01 1.63581092e-02 -3.50624800e-01
9.84250784e-01 -1.07128751e+00 5.24609804e-01 5.25526345e-01
6.96956575e-01 -6.07356906e-01 5.51102638e-01 3.88578624e-01
-1.06325889e+00 -2.42389172e-01 -3.16888273e-01 -4.82548207e-01
6.63136169e-02 1.03553462e+00 -5.82179725e-01 4.42463636e-01
4.10468668e-01 3.94146889e-01 1.35213742e-02 5.41958034e-01
-3.62476438e-01 1.36365521e+00 -7.32123196e-01 1.85928553e-01
-7.99220726e-02 -3.66817057e-01 7.12255061e-01 8.65384936e-01
2.04946920e-01 5.28396010e-01 -2.54292548e-01 1.12824976e+00
-6.42251596e-02 -2.40900353e-01 -7.72042453e-01 -2.05703631e-01
7.11131990e-01 4.71146256e-01 -5.48351288e-01 -3.04239392e-01
-1.08457081e-01 4.36933845e-01 2.71147341e-02 5.16772151e-01
-8.41388702e-01 2.85554886e-01 6.03791595e-01 -6.84269294e-02
1.46427080e-01 -1.56859532e-01 -5.63255966e-01 -1.24002671e+00
-1.72352374e-01 -5.94552219e-01 5.98599315e-01 -1.76793218e-01
-1.11045051e+00 -3.14674556e-01 2.77177393e-01 -7.20767021e-01
-3.64075363e-01 -1.39428154e-01 -3.75902236e-01 1.09391153e+00
-9.03640747e-01 -5.10412037e-01 2.68624634e-01 4.66973633e-01
-1.09299459e-01 3.13836366e-01 8.75183821e-01 -1.17642365e-01
-8.00838351e-01 2.61984706e-01 4.31474298e-01 -7.08136559e-02
3.42083246e-01 -1.50863755e+00 1.21117570e-01 8.51893604e-01
-1.52241394e-01 1.12360144e+00 1.28771412e+00 -9.63146269e-01
-1.54216218e+00 -8.95800948e-01 1.06033623e+00 -5.49366772e-01
1.08759737e+00 -6.31107092e-01 -6.56218052e-01 9.46375370e-01
1.29799724e-01 -1.66073531e-01 6.41382098e-01 5.02974391e-01
-4.26480323e-01 2.30737492e-01 -8.99212956e-01 5.76616287e-01
1.04365265e+00 -4.77305919e-01 -6.09814286e-01 4.02434051e-01
9.40546215e-01 5.21788821e-02 -9.18796837e-01 2.05798581e-01
5.41239798e-01 -7.43385494e-01 7.06044972e-01 -7.96735525e-01
6.08410239e-01 -3.54736924e-01 -3.38582724e-01 -1.09498489e+00
-3.91918391e-01 -6.85550690e-01 -2.17685521e-01 9.36042130e-01
4.40496087e-01 -9.86246645e-01 3.56792122e-01 7.43059695e-01
7.39774823e-01 -7.72359893e-02 -1.08055329e+00 -9.69600439e-01
3.90413672e-01 -3.51617754e-01 5.34570158e-01 1.27301788e+00
4.00253952e-01 5.56247890e-01 -5.57829618e-01 6.49413228e-01
9.16872859e-01 2.78296679e-01 4.99111027e-01 -1.46479273e+00
-7.26979434e-01 -2.81036466e-01 -2.43986875e-01 -6.19356871e-01
2.21730635e-01 -7.44790852e-01 -2.39165455e-01 -9.05621231e-01
7.34651566e-01 -4.66293991e-01 -1.88863978e-01 2.08511174e-01
-3.33860785e-01 -3.88329506e-01 -5.69298081e-02 3.65337253e-01
1.31665871e-01 4.77171272e-01 9.16825414e-01 5.08481525e-02
-3.69228512e-01 -4.82727438e-02 -6.18320763e-01 3.92063528e-01
8.34031522e-01 -8.85102034e-01 -4.89266217e-01 2.41827399e-01
4.35661137e-01 5.33072054e-01 8.30867708e-01 -1.28788173e-01
3.62801775e-02 -5.76073825e-01 -1.55630484e-01 -3.68940383e-01
-9.25096199e-02 -7.72576034e-01 6.85106635e-01 7.02511430e-01
-7.46349335e-01 -1.51188970e-01 -4.81131263e-02 7.34557748e-01
2.08843470e-01 3.50643806e-02 6.66166902e-01 2.32772171e-01
2.97935426e-01 -7.94943869e-02 -3.88926625e-01 2.77629755e-02
8.88846099e-01 3.31345767e-01 -3.53494346e-01 -5.72962224e-01
-6.74364746e-01 9.09780413e-02 3.70193243e-01 -1.95930768e-02
1.14533029e-01 -1.17787540e+00 -7.80756176e-01 -1.51219532e-01
-2.09031045e-01 -5.69192469e-01 1.37679145e-01 1.39133704e+00
1.66122258e-01 8.98052931e-01 4.84645396e-01 -4.61692810e-01
-1.16505218e+00 1.05655110e+00 3.71595681e-01 -1.15192637e-01
-4.69486743e-01 3.14354569e-01 6.44671082e-01 -2.18572572e-01
-3.06568116e-01 -3.49703521e-01 3.60586017e-01 -1.36554703e-01
3.67790431e-01 5.90779185e-01 -5.14131069e-01 -3.73124778e-01
-3.54225099e-01 2.19008222e-01 3.19236994e-01 -4.43519056e-01
1.22448301e+00 -3.75723034e-01 -4.02423233e-01 9.11963046e-01
1.28507912e+00 3.03072125e-01 -1.06929481e+00 8.71099010e-02
6.63407147e-02 -2.25563869e-01 1.28177494e-01 -5.97765148e-01
-8.34113061e-01 6.58170700e-01 4.21869218e-01 6.98287845e-01
1.02436340e+00 3.17706198e-01 4.80437167e-02 1.84235647e-01
2.79005408e-01 -7.58448899e-01 -6.69158697e-01 1.12227537e-01
6.64784849e-01 -1.01207197e+00 -4.64937240e-02 -4.60875392e-01
-5.89077100e-02 7.67426252e-01 -1.61199823e-01 -8.82270932e-03
6.80239439e-01 4.30119693e-01 -5.83124697e-01 -2.55252033e-01
-1.02660370e+00 -8.97825658e-02 4.85206768e-02 1.24011964e-01
5.24264216e-01 7.05247521e-01 -7.06904471e-01 2.90079772e-01
-4.22079951e-01 -8.86440054e-02 8.28455508e-01 4.10261214e-01
-1.55067325e-01 -8.87030542e-01 -5.76129615e-01 3.76597434e-01
-4.32723254e-01 -2.83594757e-01 -6.63895726e-01 9.70566094e-01
-1.08874731e-01 1.32952619e+00 3.82810868e-02 -9.47447047e-02
2.72323173e-02 -6.71408474e-02 3.61522824e-01 -2.30771214e-01
-7.33876303e-02 4.34314385e-02 -1.49734095e-01 -4.86781001e-01
-5.86638689e-01 -9.77465868e-01 -1.12057662e+00 -7.80386508e-01
-6.01585865e-01 4.33331668e-01 2.48389110e-01 1.14333510e+00
2.48235419e-01 4.30816948e-01 8.65578592e-01 9.84898116e-03
-5.22557080e-01 -9.34504509e-01 -7.10833907e-01 1.18984811e-01
7.39703476e-01 -8.67185295e-01 -9.50955153e-01 3.01505059e-01]
|
[7.892215728759766, 5.330418109893799]
|
0e238e2c-c0fc-4d39-84d3-5f4b62558335
|
gsv-cities-toward-appropriate-supervised
|
2210.10239
| null |
https://arxiv.org/abs/2210.10239v1
|
https://arxiv.org/pdf/2210.10239v1.pdf
|
GSV-Cities: Toward Appropriate Supervised Visual Place Recognition
|
This paper aims to investigate representation learning for large scale visual place recognition, which consists of determining the location depicted in a query image by referring to a database of reference images. This is a challenging task due to the large-scale environmental changes that can occur over time (i.e., weather, illumination, season, traffic, occlusion). Progress is currently challenged by the lack of large databases with accurate ground truth. To address this challenge, we introduce GSV-Cities, a new image dataset providing the widest geographic coverage to date with highly accurate ground truth, covering more than 40 cities across all continents over a 14-year period. We subsequently explore the full potential of recent advances in deep metric learning to train networks specifically for place recognition, and evaluate how different loss functions influence performance. In addition, we show that performance of existing methods substantially improves when trained on GSV-Cities. Finally, we introduce a new fully convolutional aggregation layer that outperforms existing techniques, including GeM, NetVLAD and CosPlace, and establish a new state-of-the-art on large-scale benchmarks, such as Pittsburgh, Mapillary-SLS, SPED and Nordland. The dataset and code are available for research purposes at https://github.com/amaralibey/gsv-cities.
|
['Philippe Giguère', 'Brahim Chaib-Draa', 'Amar Ali-bey']
|
2022-10-19
| null | null | null | null |
['visual-localization', 'visual-place-recognition']
|
['computer-vision', 'computer-vision']
|
[-3.46330434e-01 -6.41688168e-01 -1.50980338e-01 -4.07311797e-01
-1.17542922e+00 -6.89099550e-01 8.01321387e-01 1.75468788e-01
-5.39561510e-01 8.43840599e-01 3.94288123e-01 -2.36751050e-01
7.84551725e-02 -1.00396287e+00 -8.80656183e-01 -4.33896452e-01
-1.95546091e-01 3.55824113e-01 1.04316600e-01 -2.11282104e-01
1.44848436e-01 7.01633811e-01 -1.49938047e+00 -3.28301564e-02
6.74659550e-01 9.24871862e-01 2.01379925e-01 5.31003058e-01
1.89897805e-01 6.04653537e-01 -4.66578662e-01 -2.03839675e-01
2.88008630e-01 -1.78298861e-01 -6.97055817e-01 -1.57152578e-01
9.09779549e-01 -3.59732121e-01 -6.11018538e-01 8.81600380e-01
5.94871819e-01 3.78639638e-01 5.89973152e-01 -1.49556637e+00
-9.84128833e-01 -7.92906154e-03 -3.87173116e-01 3.65410417e-01
2.66610324e-01 2.54879713e-01 1.07247400e+00 -9.15192008e-01
7.99452841e-01 1.13008380e+00 6.96329415e-01 1.50043353e-01
-9.77243841e-01 -6.12198055e-01 3.19358438e-01 4.81587619e-01
-1.94650471e+00 -5.95680654e-01 5.09121299e-01 -5.08306801e-01
1.08636427e+00 2.61155367e-01 3.20873380e-01 1.12883842e+00
-8.66689384e-02 6.56048238e-01 9.75429833e-01 -2.06512138e-01
2.30354324e-01 -2.31402457e-01 -2.92125970e-01 7.87414610e-01
2.41731316e-01 1.03005432e-01 -3.08062285e-01 1.40938461e-01
5.68934262e-01 2.07237631e-01 -3.19009811e-01 -4.10016358e-01
-1.43140411e+00 7.40801334e-01 1.11955750e+00 2.94950008e-01
-2.39500821e-01 5.67458391e-01 1.24870159e-01 7.90417567e-02
4.89965260e-01 3.07386011e-01 -2.58294225e-01 -1.15903758e-01
-1.08887732e+00 4.32557285e-01 4.07193333e-01 1.00165236e+00
1.03012586e+00 -6.15481846e-02 -2.34162495e-01 8.03196192e-01
4.04281765e-01 9.12256241e-01 2.65858084e-01 -9.36641276e-01
8.13580632e-01 3.93379092e-01 3.60640407e-01 -1.26742661e+00
-4.84777778e-01 -4.30526465e-01 -8.21516812e-01 2.09571645e-01
4.45558041e-01 5.84296882e-02 -1.06943452e+00 1.80083251e+00
2.48889163e-01 5.95833957e-01 -8.84245709e-02 9.32627618e-01
9.47632730e-01 6.54096007e-01 3.10038924e-02 7.04752743e-01
1.16508782e+00 -1.08688319e+00 -2.13189021e-01 -5.77198207e-01
5.99758506e-01 -5.05562603e-01 9.27837610e-01 -1.39891952e-01
-5.52533746e-01 -4.04881299e-01 -1.10875702e+00 -3.07515979e-01
-1.13088214e+00 1.95629120e-01 5.23839772e-01 3.62317145e-01
-1.37058175e+00 3.06727022e-01 -7.92201996e-01 -8.98395538e-01
6.64571345e-01 1.35014793e-02 -5.56477606e-01 -3.78690034e-01
-1.21529245e+00 8.84064972e-01 1.21179961e-01 1.23396724e-01
-9.41620290e-01 -6.86702728e-01 -1.21704853e+00 -1.01152815e-01
-1.68841094e-01 -4.53464538e-01 1.13700819e+00 -4.84558553e-01
-6.88689530e-01 1.10383165e+00 -2.89965034e-01 -5.55826902e-01
5.91986358e-01 1.19387142e-01 -6.44842744e-01 -2.04586193e-01
5.24786294e-01 9.51051831e-01 2.49197930e-02 -1.13312900e+00
-7.71405160e-01 -4.09602314e-01 6.08623028e-02 1.89587280e-01
1.27500132e-01 -3.27924527e-02 -6.78810656e-01 -4.03381169e-01
-2.08510458e-02 -1.13244951e+00 -4.13777947e-01 1.87014028e-01
-4.88764107e-01 -2.36206502e-03 4.73727018e-01 -5.51835597e-01
9.46169257e-01 -2.30208302e+00 -3.28994691e-01 1.35366172e-01
1.06008612e-01 2.09103655e-02 -4.46961075e-01 5.44919133e-01
1.65284306e-01 2.09572881e-01 -2.14457899e-01 -5.34078419e-01
3.28890562e-01 1.76181927e-01 -2.16627643e-01 8.00975323e-01
1.40401376e-02 1.25330722e+00 -1.00984073e+00 -1.96118176e-01
4.13211733e-01 5.37307382e-01 -2.24765554e-01 -1.33165732e-01
1.57472361e-02 3.10509622e-01 -2.60005444e-01 9.56998467e-01
6.99719727e-01 -4.34884399e-01 -4.38685775e-01 1.55150011e-01
-4.02824253e-01 3.51616681e-01 -1.15412033e+00 1.95251620e+00
-4.82214570e-01 1.12978494e+00 -1.89534143e-01 -7.33439505e-01
8.56367648e-01 -1.14322547e-02 3.25737059e-01 -1.22458410e+00
-1.47146270e-01 2.94566453e-01 -5.29689729e-01 -1.19898707e-01
7.47484326e-01 4.11666006e-01 -2.45660961e-01 8.28303695e-02
-1.94418713e-01 3.42883468e-02 1.87180951e-01 -8.90880544e-03
1.02097821e+00 -1.10466331e-01 2.25991830e-01 -1.72683969e-01
1.33786485e-01 3.04149151e-01 5.12937546e-01 7.02053785e-01
-5.69229782e-01 1.10700881e+00 1.48788661e-01 -4.93612051e-01
-1.03358126e+00 -1.19570494e+00 -1.74552724e-01 8.93694699e-01
2.35706463e-01 -1.05984822e-01 -3.34333688e-01 -3.81653965e-01
2.70921946e-01 5.56106985e-01 -8.40685725e-01 1.05172113e-01
-2.85860032e-01 -7.87169516e-01 6.55348063e-01 7.12695777e-01
9.34976995e-01 -8.57121289e-01 -4.60194558e-01 8.03127065e-02
-3.50996524e-01 -1.23555183e+00 -4.81229216e-01 -7.72879496e-02
-4.05091971e-01 -1.09605920e+00 -8.15812707e-01 -7.08772779e-01
4.79490280e-01 5.41582584e-01 1.45495152e+00 -7.92352408e-02
-2.79554576e-01 3.57958704e-01 -7.66269490e-02 -3.46555084e-01
2.66679615e-01 3.66163194e-01 -3.99423897e-01 -1.82920590e-01
4.46845591e-01 -4.27916467e-01 -8.86776626e-01 3.44996631e-01
-4.79733378e-01 -1.11109912e-01 3.45610946e-01 4.25317109e-01
8.36789608e-01 -2.30609983e-01 3.36680919e-01 -4.39511776e-01
2.74915636e-01 -9.06621456e-01 -8.98591459e-01 2.78109580e-01
-3.70539248e-01 -1.63526967e-01 1.36624485e-01 1.03550434e-01
-6.63768888e-01 9.56766605e-02 -1.25341013e-01 -3.73426884e-01
-4.43249941e-01 4.94241387e-01 -9.65611264e-02 -1.04076162e-01
7.98708975e-01 2.07505852e-01 -5.39611578e-01 -2.55579710e-01
5.19213855e-01 6.95734739e-01 8.22695673e-01 -4.25314397e-01
9.17817950e-01 7.62693524e-01 -4.57548648e-02 -7.08053827e-01
-6.95654750e-01 -6.88039720e-01 -6.82413161e-01 -2.74104904e-02
1.02770197e+00 -1.38317513e+00 -2.76321679e-01 5.25424957e-01
-1.05253148e+00 -5.52670836e-01 7.42700621e-02 4.26761717e-01
-3.57900530e-01 -2.31063679e-01 -1.87244445e-01 -3.41143698e-01
-2.06197619e-01 -1.01441216e+00 1.36979914e+00 2.51436710e-01
5.92189794e-03 -1.13371491e+00 4.52801228e-01 2.63497859e-01
6.77732110e-01 7.74801850e-01 3.54950935e-01 -3.17943215e-01
-1.10288835e+00 -1.32860392e-01 -5.15074134e-01 -3.05853877e-02
1.02493256e-01 -1.25556678e-01 -1.01662862e+00 -3.89484286e-01
-8.62876296e-01 -2.71682471e-01 1.03329647e+00 2.96348572e-01
1.11220074e+00 -4.71137352e-02 -4.58857000e-01 1.00736082e+00
1.76588023e+00 1.35496631e-01 7.80221522e-01 9.05377507e-01
8.30609500e-01 1.13435701e-01 5.07411540e-01 3.77308071e-01
1.03369141e+00 7.75305271e-01 7.02653110e-01 -4.13937420e-01
-3.76778126e-01 -4.16891962e-01 -1.56115457e-01 1.88059255e-01
1.03561608e-02 -5.37794530e-01 -1.29729748e+00 1.19760704e+00
-1.91381562e+00 -1.08056438e+00 -1.19283184e-01 2.25616884e+00
1.91373214e-01 -2.76128531e-01 -6.19095899e-02 -2.63366371e-01
4.35283303e-01 6.59169912e-01 -6.25499725e-01 1.71675179e-02
-3.43095273e-01 -7.63745829e-02 9.47289586e-01 3.78875464e-01
-1.57668889e+00 1.04027665e+00 6.03099966e+00 4.60526705e-01
-1.34606218e+00 8.83090794e-02 9.02510583e-01 -2.00539857e-01
-2.07804635e-01 -3.71766865e-01 -6.97159827e-01 4.55403417e-01
1.01499522e+00 -8.34637135e-02 5.79306483e-01 9.06882703e-01
3.09362680e-01 -5.01929112e-02 -9.14337635e-01 1.24647462e+00
1.79372609e-01 -1.53599489e+00 -4.22536314e-01 1.31668434e-01
1.02131379e+00 1.09752572e+00 3.51350635e-01 3.77996445e-01
5.45413315e-01 -1.27827489e+00 8.72782409e-01 3.11328113e-01
9.81449366e-01 -6.10863328e-01 4.87627894e-01 -1.02750637e-01
-1.65289843e+00 2.29643714e-02 -4.26117569e-01 3.65441255e-02
1.35823777e-02 3.02459776e-01 -6.65316403e-01 4.48865205e-01
9.10478294e-01 1.06490612e+00 -9.27765369e-01 1.50574112e+00
-3.78605008e-01 3.74603689e-01 -4.65921134e-01 1.82569593e-01
6.89015090e-01 -8.88553541e-03 1.79921582e-01 1.19036531e+00
5.90111196e-01 -1.34044424e-01 2.07528666e-01 9.06733155e-01
-3.58806968e-01 -7.62532651e-02 -1.03560066e+00 2.89591312e-01
7.54445493e-01 1.13964415e+00 -5.36305547e-01 -1.74668491e-01
-6.33698463e-01 9.56245244e-01 6.89158857e-01 6.98500514e-01
-1.01260245e+00 -2.55534828e-01 1.25766516e+00 8.20258781e-02
3.67085427e-01 -5.19292891e-01 -7.95032084e-02 -1.20104277e+00
5.32754771e-02 -4.93742198e-01 3.81700754e-01 -1.02156413e+00
-1.00754344e+00 5.09383619e-01 -1.19260378e-01 -1.29310381e+00
-1.54522061e-01 -5.15980721e-01 -7.32261419e-01 8.59539449e-01
-1.97452283e+00 -1.31250322e+00 -7.78235853e-01 7.36916304e-01
3.66712511e-01 -3.34803574e-02 7.53617227e-01 6.97897673e-01
-5.77497542e-01 5.90810359e-01 6.07601821e-01 4.78667974e-01
7.54860878e-01 -1.29571378e+00 9.86974895e-01 9.33998704e-01
4.44249451e-01 2.37632811e-01 2.77355701e-01 -2.23920047e-01
-1.10435247e+00 -1.67476666e+00 1.10567582e+00 -7.08655894e-01
6.04197860e-01 -5.25223970e-01 -5.73535383e-01 9.69611287e-01
1.32359266e-01 4.92767096e-01 4.65850681e-01 1.13900483e-01
-5.29250979e-01 -2.70990759e-01 -1.22310472e+00 5.85782170e-01
1.20409262e+00 -7.90724576e-01 -1.87333584e-01 4.25172657e-01
5.19250393e-01 -6.54149592e-01 -5.79124629e-01 3.54024805e-02
3.28102201e-01 -8.69167507e-01 1.13148046e+00 -3.03515583e-01
3.24181974e-01 -5.48970759e-01 -5.93186080e-01 -1.53387773e+00
-5.97820699e-01 -5.90016693e-02 4.17495668e-01 1.16346145e+00
5.78815758e-01 -9.01874006e-01 7.09570110e-01 6.10845506e-01
-1.17475033e-01 -6.15559518e-01 -1.15821350e+00 -9.19506609e-01
2.14250356e-01 -4.69906658e-01 1.17514241e+00 1.09133387e+00
-5.28302550e-01 -1.39260247e-01 -2.62869716e-01 5.58270395e-01
5.23091257e-01 2.63728291e-01 8.52154672e-01 -9.98708308e-01
2.78613776e-01 -4.46728021e-01 -9.06712055e-01 -8.44534278e-01
1.08293295e-01 -9.70847785e-01 2.02497654e-02 -2.20965028e+00
2.11966410e-02 -4.53373432e-01 -5.94648480e-01 7.40001738e-01
3.87935564e-02 5.79742908e-01 1.75297186e-01 1.37739643e-01
-9.87482131e-01 5.51288545e-01 9.05108392e-01 -5.61156213e-01
2.95480862e-02 -2.39253268e-01 -5.75935543e-01 2.31879056e-01
8.76270413e-01 -3.06855768e-01 -3.56059760e-01 -9.49241221e-01
7.65614063e-02 -4.63050395e-01 6.54012859e-01 -1.34906006e+00
2.76742220e-01 -2.54316926e-01 5.96842885e-01 -7.34310865e-01
3.31004143e-01 -7.70550311e-01 3.40534151e-01 1.87444612e-01
-1.59900010e-01 4.10804033e-01 4.29570735e-01 5.23738921e-01
-4.38465714e-01 4.43088502e-01 6.13691092e-01 -1.55555099e-01
-1.31883705e+00 6.35419369e-01 2.50092968e-02 2.81293392e-01
9.59880590e-01 -8.99067521e-02 -7.12677360e-01 -3.35673004e-01
-2.95759797e-01 5.30645311e-01 6.86958075e-01 7.76462495e-01
4.57637101e-01 -1.76985192e+00 -8.83169353e-01 3.73472199e-02
5.84938288e-01 8.92890841e-02 2.31135368e-01 5.76869011e-01
-7.97740817e-01 5.53686976e-01 -2.00228363e-01 -4.72524881e-01
-1.01012003e+00 3.36477548e-01 6.64500296e-01 -9.83034745e-02
-4.71756816e-01 7.26180494e-01 1.76951334e-01 -6.59497738e-01
9.52651352e-02 -6.74420059e-01 1.12510026e-02 6.50679395e-02
5.69179535e-01 4.14471656e-01 2.91228622e-01 -1.06245112e+00
-7.98594415e-01 5.78664124e-01 3.81926656e-01 -6.55863509e-02
1.40664780e+00 -1.33499548e-01 1.55161858e-01 2.93503076e-01
1.56691802e+00 -3.01402658e-01 -1.30924654e+00 -1.94911331e-01
-7.56413788e-02 -7.14671493e-01 1.30008578e-01 -8.66823077e-01
-1.28503489e+00 7.41543472e-01 1.01665449e+00 -3.04432452e-01
8.51370573e-01 1.34605855e-01 7.11246312e-01 5.17015338e-01
6.02129221e-01 -8.95272613e-01 -3.09437156e-01 6.64396524e-01
1.02433097e+00 -1.64179862e+00 -1.60633877e-01 2.14354247e-01
-3.77139241e-01 5.43818891e-01 5.15155852e-01 -8.53842869e-02
7.12245762e-01 -1.34083509e-01 2.95763075e-01 -3.33354145e-01
-4.45619047e-01 -5.60413182e-01 4.24123943e-01 7.39972174e-01
2.31504381e-01 3.95192564e-01 4.15579855e-01 -2.96434294e-02
-3.63648772e-01 -9.13418978e-02 3.01628917e-01 7.73828328e-01
-3.02214116e-01 -6.34921849e-01 -4.94190931e-01 2.48147979e-01
-3.50335874e-02 -1.49903223e-01 -3.14306259e-01 9.29252446e-01
1.30893514e-01 9.32659745e-01 2.97440201e-01 -1.60621464e-01
5.11162162e-01 -2.41758168e-01 1.12885602e-01 -3.22253197e-01
-1.74231276e-01 -5.62278688e-01 1.64155662e-01 -8.67346644e-01
-1.71925321e-01 -8.51661563e-01 -1.15624595e+00 -5.46965420e-01
3.30731541e-01 -1.95164740e-01 8.17527533e-01 6.34474933e-01
6.64390862e-01 3.01991880e-01 4.19298649e-01 -9.94738519e-01
2.67523658e-02 -7.27667332e-01 -5.02496123e-01 4.94312197e-01
6.59305274e-01 -7.85076201e-01 -2.22219422e-01 -1.97053432e-01]
|
[7.748217582702637, -1.8542766571044922]
|
59a9cb7b-bc97-40f6-9343-1ad04c0b63c1
|
frequency-domain-multi-channel-acoustic
|
1903.05299
| null |
http://arxiv.org/abs/1903.05299v2
|
http://arxiv.org/pdf/1903.05299v2.pdf
|
Frequency Domain Multi-channel Acoustic Modeling for Distant Speech Recognition
|
Conventional far-field automatic speech recognition (ASR) systems typically
employ microphone array techniques for speech enhancement in order to improve
robustness against noise or reverberation. However, such speech enhancement
techniques do not always yield ASR accuracy improvement because the
optimization criterion for speech enhancement is not directly relevant to the
ASR objective. In this work, we develop new acoustic modeling techniques that
optimize spatial filtering and long short-term memory (LSTM) layers from
multi-channel (MC) input based on an ASR criterion directly. In contrast to
conventional methods, we incorporate array processing knowledge into the
acoustic model. Moreover, we initialize the network with beamformers'
coefficients. We investigate effects of such MC neural networks through ASR
experiments on the real-world far-field data where users are interacting with
an ASR system in uncontrolled acoustic environments. We show that our MC
acoustic model can reduce a word error rate (WER) by~16.5\% compared to a
single channel ASR system with the traditional log-mel filter bank energy
(LFBE) feature on average. Our result also shows that our network with the
spatial filtering layer on two-channel input achieves a relative WER reduction
of~9.5\% compared to conventional beamforming with seven microphones.
|
[]
|
2019-04-28
| null | null | null | null |
['distant-speech-recognition']
|
['speech']
|
[ 3.72330844e-01 -2.54678041e-01 7.69159794e-01 -3.17879170e-01
-1.15242279e+00 -2.73294836e-01 1.54411495e-01 -1.53790414e-01
-7.44363189e-01 4.45104301e-01 6.26078546e-01 -6.16190076e-01
-2.54681855e-02 -4.49933499e-01 -6.66665733e-01 -5.83277166e-01
-6.81596026e-02 -5.01568377e-01 -5.12565672e-02 -3.74892890e-01
-4.20496687e-02 4.34887707e-01 -1.36791766e+00 3.03951681e-01
6.70549273e-01 1.04027855e+00 7.13954568e-01 1.34222960e+00
2.34409809e-01 3.72341126e-01 -1.16053307e+00 4.73160744e-02
2.91554391e-01 -2.02614412e-01 -1.83105707e-01 -2.53201276e-01
2.43025526e-01 -3.76125753e-01 -7.11664200e-01 1.15968633e+00
1.36093974e+00 5.98223507e-01 2.51932949e-01 -3.12521458e-01
-5.63259125e-01 7.04892635e-01 -3.12755018e-01 4.48592693e-01
4.23367888e-01 1.53701873e-02 7.26748109e-01 -1.21882391e+00
-2.60118484e-01 1.26858163e+00 6.86971784e-01 5.31714559e-01
-8.16552401e-01 -7.69669414e-01 5.62940612e-02 1.17006823e-01
-1.51981187e+00 -1.17264724e+00 6.75972581e-01 7.97999650e-02
1.54447889e+00 5.26498079e-01 2.17197552e-01 7.96743810e-01
4.50590067e-02 4.68782187e-01 9.13257182e-01 -7.40599334e-01
1.89229414e-01 8.41049179e-02 1.66763365e-01 2.02893734e-01
-2.97451794e-01 5.26341617e-01 -7.79591441e-01 1.59661755e-01
5.45980811e-01 -4.57196027e-01 -6.04726136e-01 8.71241629e-01
-9.61777091e-01 3.99191856e-01 2.71561027e-01 7.60109425e-01
-3.08878541e-01 3.21790487e-01 -1.58294827e-01 5.49623728e-01
6.32094264e-01 5.69267094e-01 -5.18586874e-01 -2.91997433e-01
-9.31243956e-01 -2.81856358e-01 5.47098041e-01 5.36590934e-01
2.02477455e-01 7.39848733e-01 -1.56829774e-01 1.52070463e+00
7.55068839e-01 9.51931357e-01 6.51024938e-01 -8.03072631e-01
4.56538737e-01 -5.12084126e-01 7.33295009e-02 -7.40405321e-01
-2.31279731e-01 -1.07488179e+00 -8.24170470e-01 1.02423020e-01
6.02721497e-02 -6.25103176e-01 -9.32548344e-01 1.75657475e+00
-3.59351151e-02 3.97792995e-01 3.14260781e-01 8.71968865e-01
7.76668549e-01 1.01850605e+00 -4.10138160e-01 -4.28219020e-01
1.01717699e+00 -8.00871730e-01 -1.38312483e+00 -3.49907279e-01
4.34896499e-01 -1.13709748e+00 1.02331746e+00 6.26626790e-01
-1.35400701e+00 -7.26007879e-01 -1.19333112e+00 3.08744192e-01
-3.77190918e-01 1.00970432e-01 1.43065471e-02 1.42115200e+00
-1.60562205e+00 3.11215401e-01 -6.52476192e-01 6.35659546e-02
-2.67439425e-01 5.40945768e-01 -1.29623577e-01 2.84332395e-01
-1.40796232e+00 7.66601205e-01 -1.26413137e-01 5.60641885e-01
-5.68271995e-01 -6.89239860e-01 -8.36341619e-01 3.35726619e-01
-1.79087408e-02 -2.03273296e-01 1.61985195e+00 -6.61743283e-01
-1.98826432e+00 1.68716349e-02 -6.05018020e-01 -3.63761067e-01
-2.26350993e-01 -5.24765551e-01 -1.08337176e+00 -1.36944130e-01
-6.90502048e-01 3.06978554e-01 8.43240619e-01 -1.15271461e+00
-5.22228599e-01 -1.59759060e-01 -1.99086502e-01 4.99471307e-01
-7.93090820e-01 3.75288248e-01 -7.69685209e-02 -1.08872700e+00
1.37935445e-01 -4.07986045e-01 -3.23786259e-01 -6.37156367e-01
-1.73534319e-01 2.12161258e-01 6.06849194e-01 -1.10715437e+00
1.78857279e+00 -2.28578639e+00 -4.70966727e-01 4.43150222e-01
-1.59997106e-01 6.76020145e-01 -2.95493245e-01 4.04449329e-02
-2.56842434e-01 2.13211507e-01 1.49567705e-03 -4.52434123e-01
-6.61169440e-02 -2.80915290e-01 -2.37509638e-01 3.33258331e-01
-1.35435045e-01 3.48437339e-01 -5.02891004e-01 3.80512141e-03
4.80937481e-01 9.52024698e-01 -7.36666620e-01 4.25534070e-01
5.32178164e-01 2.74111390e-01 7.23476261e-02 2.91571796e-01
7.28939772e-01 2.80993849e-01 -2.22359389e-01 -1.40030667e-01
-2.10471421e-01 4.16397214e-01 -1.32824492e+00 1.54250014e+00
-1.08171976e+00 8.95457447e-01 7.61950731e-01 -7.38430619e-01
8.77200603e-01 7.64202476e-01 8.50653499e-02 -9.02104855e-01
2.46927887e-02 4.85594183e-01 4.63600665e-01 -1.18085906e-01
5.32871723e-01 -1.96615875e-01 3.96630049e-01 1.84286162e-01
4.10242975e-02 5.88540919e-02 -4.34261739e-01 -1.40086547e-01
1.25054526e+00 -6.69178724e-01 7.87520409e-02 -1.24570832e-01
7.64127254e-01 -1.03938079e+00 1.75864100e-01 1.01334107e+00
-1.44460768e-01 5.75419188e-01 -6.98348403e-01 2.85877973e-01
-7.18672276e-01 -1.11190188e+00 -1.84347808e-01 1.18091953e+00
-1.58262253e-01 -2.83005059e-01 -8.79181743e-01 8.30179155e-02
-5.49723566e-01 1.00344265e+00 5.71526997e-02 -8.78677890e-02
-7.89760530e-01 -8.52289438e-01 8.03121209e-01 4.97278482e-01
4.25621122e-01 -8.09010565e-01 -1.10964745e-01 4.50869083e-01
-1.26696691e-01 -1.04815769e+00 -7.12717950e-01 4.48656976e-01
-4.45610017e-01 4.11406495e-02 -8.60416651e-01 -6.81486487e-01
3.75302792e-01 3.77003372e-01 5.98642468e-01 -6.22251220e-02
1.07527941e-01 5.92205405e-01 -4.24860358e-01 -3.83613288e-01
-4.86852169e-01 -3.11077297e-01 4.47160482e-01 9.89554673e-02
1.74664557e-01 -6.69973195e-01 -7.19756782e-01 3.72600496e-01
-6.56488895e-01 -3.85447413e-01 5.53061008e-01 8.04973900e-01
1.56389505e-01 4.31233764e-01 1.01272154e+00 3.51032391e-02
1.07287502e+00 -9.63963792e-02 -3.77969831e-01 1.24522761e-01
-4.66324925e-01 -1.87607586e-01 5.98211169e-01 -4.65489686e-01
-1.60294724e+00 -2.31369853e-01 -1.21231699e+00 -1.37720093e-01
-2.90693700e-01 4.16482031e-01 -4.53455776e-01 -1.38767153e-01
7.11921096e-01 2.77455568e-01 -4.67840195e-01 -6.56260610e-01
2.48726681e-01 1.50044382e+00 4.38783944e-01 -2.22310558e-01
7.36778140e-01 -3.94233279e-02 -4.02464867e-01 -1.50768113e+00
-1.74666703e-01 -7.02983141e-01 -1.20866269e-01 -2.03998893e-01
6.38920784e-01 -9.37431574e-01 -5.77747762e-01 5.23551285e-01
-1.26590657e+00 -3.32712471e-01 4.14434224e-02 1.11160529e+00
-2.22656846e-01 2.09781438e-01 -7.47721672e-01 -1.42228699e+00
-5.20856082e-01 -1.20001793e+00 8.92259300e-01 8.88219476e-02
-1.30306333e-02 -9.27937567e-01 -1.03058308e-01 1.46458268e-01
1.01855564e+00 -7.50718892e-01 3.85874033e-01 -5.52440047e-01
-9.50348526e-02 -6.96241111e-02 2.17681020e-01 7.88590372e-01
1.91410318e-01 -5.54364681e-01 -1.58886027e+00 -4.14305359e-01
5.63310266e-01 2.63849854e-01 6.22652948e-01 9.87187922e-01
1.30556369e+00 -3.33127499e-01 -1.64114460e-02 6.20766878e-01
1.12776923e+00 9.45670128e-01 8.53177845e-01 -1.04816489e-01
3.69646043e-01 2.56229430e-01 3.02974790e-01 3.41715127e-01
-4.22783911e-01 6.49026632e-01 -2.95289010e-02 -3.26759279e-01
-4.98536319e-01 6.50084838e-02 7.47677684e-01 1.45265889e+00
2.62582183e-01 -9.01548684e-01 -6.84115767e-01 3.43366772e-01
-1.10156393e+00 -8.19473386e-01 -1.40792904e-02 2.39370847e+00
7.98937559e-01 1.54370368e-02 -3.93203020e-01 4.75561500e-01
8.50624204e-01 1.45392537e-01 -6.75828606e-02 -7.21209228e-01
-1.90209597e-01 6.79244876e-01 5.82514882e-01 1.24718261e+00
-8.61667752e-01 5.32107949e-01 6.29617357e+00 1.19756532e+00
-1.18028200e+00 4.35564637e-01 6.62502050e-01 -2.59961456e-01
-2.20057607e-01 -6.66180134e-01 -6.84020877e-01 1.81120232e-01
1.49171853e+00 1.66529775e-01 8.27261209e-01 3.30735624e-01
7.72027254e-01 6.71340153e-02 -9.22628582e-01 1.27915490e+00
2.70741135e-02 -7.00468659e-01 -4.07587171e-01 7.18623847e-02
3.11471850e-01 -1.60348993e-02 4.72881913e-01 2.03335196e-01
1.09479325e-02 -1.38762331e+00 5.75254083e-01 3.13153476e-01
9.40199733e-01 -7.60079741e-01 6.51309848e-01 3.90952051e-01
-1.30082178e+00 -2.48566002e-01 -1.81793392e-01 -2.04449922e-01
5.04608989e-01 7.88753629e-01 -9.62734461e-01 1.12932511e-01
6.80784643e-01 -1.33904189e-01 -1.69219643e-01 1.08617949e+00
5.66441044e-02 1.13137448e+00 -5.67716599e-01 -9.74532664e-02
5.55417426e-02 2.28120983e-01 9.57648814e-01 1.62847221e+00
8.20981622e-01 3.48417819e-01 -4.34545487e-01 4.30313557e-01
-1.38567507e-01 6.08558729e-02 -4.69911397e-01 6.72450587e-02
5.98785818e-01 8.45824897e-01 -2.93745846e-01 -9.36355144e-02
-3.98050070e-01 7.60618806e-01 -4.60122049e-01 8.16439927e-01
-7.07850695e-01 -8.56954694e-01 6.80302262e-01 -3.87871601e-02
1.97593182e-01 -3.60870361e-01 -3.46784770e-01 -7.70470262e-01
-9.73997265e-02 -1.01958013e+00 -3.19976687e-01 -8.75801682e-01
-9.82531786e-01 8.10772061e-01 -4.42885816e-01 -8.57553363e-01
-1.20236069e-01 -7.75489688e-01 -4.62221533e-01 1.39402819e+00
-1.36205494e+00 -5.60755193e-01 2.12931544e-01 5.31953096e-01
8.14310074e-01 -2.59938180e-01 8.92845094e-01 9.02383387e-01
-3.35552990e-01 1.06579065e+00 3.31923515e-01 1.59351416e-02
5.32768011e-01 -1.25387335e+00 5.01364827e-01 1.33321285e+00
1.00637212e-01 9.84912276e-01 8.81301463e-01 -4.99663621e-01
-1.25158370e+00 -9.64642465e-01 8.19488525e-01 -5.12995943e-02
4.27653372e-01 -6.29587770e-01 -8.60829830e-01 3.22217554e-01
6.35727346e-01 -1.57656610e-01 8.12406182e-01 2.75760088e-02
1.60050064e-01 -3.09571773e-01 -9.22355652e-01 6.24799967e-01
1.05528164e+00 -6.28151178e-01 -5.35405874e-01 1.39115572e-01
1.09467387e+00 -2.88320988e-01 -7.64967084e-01 5.32077670e-01
5.15015781e-01 -5.47097802e-01 1.17311776e+00 -4.73393500e-02
-2.08469301e-01 -4.67735946e-01 -8.04953098e-01 -1.75986695e+00
-2.20281973e-01 -1.05037546e+00 -7.82279223e-02 1.20141530e+00
7.80562103e-01 -7.37515748e-01 2.13920146e-01 3.23219597e-01
-5.75325429e-01 -3.27516764e-01 -8.98545265e-01 -7.97218144e-01
-4.72967550e-02 -1.06659448e+00 3.18339854e-01 2.64223814e-01
-1.38493270e-01 3.50331664e-01 -5.91523528e-01 6.74519181e-01
3.47524166e-01 -8.46521258e-01 2.10438788e-01 -5.21606922e-01
-7.25569427e-01 -4.06762958e-01 4.83685248e-02 -1.60401464e+00
-1.30715072e-01 -5.25677800e-01 6.34652853e-01 -1.45635223e+00
-5.35088181e-01 -1.91248089e-01 -6.98005676e-01 -2.74246987e-02
-2.42485344e-01 7.05676824e-02 4.46419865e-02 -6.32106304e-01
-1.93778038e-01 5.45404732e-01 8.57807636e-01 -6.50450215e-02
-4.06289548e-01 3.36970389e-01 -6.21104956e-01 8.10130715e-01
7.32436895e-01 -2.27403536e-01 -4.88591105e-01 -7.14173436e-01
-7.78231770e-05 3.12501162e-01 1.34914219e-02 -1.16211867e+00
4.82410729e-01 2.83535868e-01 4.27359581e-01 -5.35311460e-01
7.67418444e-01 -9.09341276e-01 6.58482015e-02 3.12897503e-01
-6.03043556e-01 -9.75423306e-02 4.37558651e-01 4.68384445e-01
-3.33236247e-01 -2.11196199e-01 7.51703858e-01 1.43957451e-01
-4.15428102e-01 -2.78006256e-01 -8.57634544e-01 -5.06533802e-01
1.97221100e-01 -9.62289572e-02 -6.96779341e-02 -9.17457998e-01
-8.44773889e-01 -2.51278698e-01 -3.75955403e-01 1.67973205e-01
9.73912597e-01 -1.14377415e+00 -7.07238197e-01 3.18287015e-01
-5.60855567e-01 -4.89574283e-01 4.07108039e-01 5.34672618e-01
-3.16904709e-02 6.33967340e-01 4.37734485e-01 -4.08703893e-01
-1.48606229e+00 1.94575354e-01 7.45766640e-01 8.04207288e-03
-2.06103206e-01 1.25187874e+00 3.44041288e-01 -3.77081752e-01
6.24663591e-01 -3.37534904e-01 -2.87965894e-01 -5.07964849e-01
7.85225093e-01 5.12178481e-01 6.28082991e-01 -6.12780452e-01
-2.65108049e-01 3.74163330e-01 3.95004749e-01 -9.41067934e-01
1.10069275e+00 -5.53858221e-01 3.39275628e-01 2.50998467e-01
1.33250213e+00 6.63017631e-01 -5.80569565e-01 -1.87836185e-01
-3.77102077e-01 -5.19863605e-01 6.55117631e-01 -1.14666867e+00
-8.60539913e-01 1.22322845e+00 1.26648653e+00 7.34727979e-02
1.63637078e+00 -4.37170386e-01 8.88532162e-01 7.27401912e-01
1.16767339e-01 -1.16381812e+00 1.13471657e-01 6.35732532e-01
1.04789948e+00 -7.93204725e-01 -7.04985499e-01 -4.52397317e-02
-9.86867025e-03 9.32311058e-01 3.63163263e-01 3.34191501e-01
1.03685331e+00 8.84842396e-01 3.24737728e-01 3.29016328e-01
-5.29897809e-01 -2.09337488e-01 3.34952176e-01 5.96763492e-01
6.90640926e-01 5.73264509e-02 -1.09592140e-01 6.98268712e-01
-5.79040706e-01 -5.44300199e-01 2.97575891e-01 6.33329630e-01
-8.95816803e-01 -9.59641933e-01 -9.80001569e-01 3.21944058e-01
-8.66933405e-01 -5.94659269e-01 4.67718467e-02 -6.41448498e-02
-1.66650757e-01 1.79713213e+00 -4.42609116e-02 -7.32441545e-01
4.92349416e-01 5.01427315e-02 1.48813367e-01 -5.57808340e-01
-7.25709677e-01 8.69005442e-01 2.03657329e-01 -1.92470595e-01
-1.72681957e-01 -4.03307915e-01 -1.12167215e+00 -9.31688547e-02
-7.59898722e-01 1.79390013e-01 1.10367322e+00 8.71828556e-01
2.08029777e-01 1.28757906e+00 7.55723834e-01 -6.87115550e-01
-4.88644779e-01 -1.31724262e+00 -6.09506547e-01 -1.64911345e-01
5.97375572e-01 -2.48718545e-01 -5.17142057e-01 -1.59030817e-02]
|
[14.948020935058594, 5.952859878540039]
|
a0dfb573-b5ca-4d45-bd2f-758c770fb381
|
recasnet-improving-consistency-within-the-two
|
2202.13912
| null |
https://arxiv.org/abs/2202.13912v1
|
https://arxiv.org/pdf/2202.13912v1.pdf
|
ReCasNet: Improving consistency within the two-stage mitosis detection framework
|
Mitotic count (MC) is an important histological parameter for cancer diagnosis and grading, but the manual process for obtaining MC from whole-slide histopathological images is very time-consuming and prone to error. Therefore, deep learning models have been proposed to facilitate this process. Existing approaches utilize a two-stage pipeline: the detection stage for identifying the locations of potential mitotic cells and the classification stage for refining prediction confidences. However, this pipeline formulation can lead to inconsistencies in the classification stage due to the poor prediction quality of the detection stage and the mismatches in training data distributions between the two stages. In this study, we propose a Refine Cascade Network (ReCasNet), an enhanced deep learning pipeline that mitigates the aforementioned problems with three improvements. First, window relocation was used to reduce the number of poor quality false positives generated during the detection stage. Second, object re-cropping was performed with another deep learning model to adjust poorly centered objects. Third, improved data selection strategies were introduced during the classification stage to reduce the mismatches in training data distributions. ReCasNet was evaluated on two large-scale mitotic figure recognition datasets, canine cutaneous mast cell tumor (CCMCT) and canine mammary carcinoma (CMC), which resulted in up to 4.8% percentage point improvements in the F1 scores for mitotic cell detection and 44.1% reductions in mean absolute percentage error (MAPE) for MC prediction. Techniques that underlie ReCasNet can be generalized to other two-stage object detection networks and should contribute to improving the performances of deep learning models in broad digital pathology applications.
|
['Ekapol Chuangsuwanich', 'Sira Sriswasdi', 'Qingyi Tao', 'Shanop Shuangshoti', 'Sakun Santisukwongchote', 'Chawan Piansaddhayanon']
|
2022-02-28
| null | null | null | null |
['cell-detection', 'mitosis-detection']
|
['computer-vision', 'medical']
|
[ 3.50055367e-01 -3.46954986e-02 -1.39934480e-01 -2.42256880e-01
-1.20031548e+00 -3.23938519e-01 4.45801198e-01 8.47267926e-01
-7.77670622e-01 4.62586969e-01 -2.13573486e-01 -4.20244873e-01
7.29003474e-02 -7.61250257e-01 -3.60364854e-01 -1.08704031e+00
1.81238919e-01 5.16255736e-01 5.61262131e-01 2.62660265e-01
4.85478938e-01 7.77953029e-01 -1.23059046e+00 7.61962831e-01
6.91447556e-01 9.32491601e-01 2.75390089e-01 9.69698131e-01
-2.47078806e-01 5.39937615e-01 -6.78211510e-01 -2.63225168e-01
-1.68103501e-01 -6.63601086e-02 -5.75134516e-01 -1.13124654e-01
2.05211490e-01 -3.44903946e-01 1.27503470e-01 9.17351842e-01
6.49545014e-01 -5.31716287e-01 8.58212948e-01 -1.12367189e+00
-2.88049541e-02 3.28826904e-01 -9.83876705e-01 3.67502779e-01
-3.84525150e-01 3.74296427e-01 6.28832698e-01 -8.54641259e-01
4.95090544e-01 7.19996989e-01 1.02292514e+00 5.57740211e-01
-1.30013692e+00 -7.44423509e-01 -4.53955561e-01 3.12099218e-01
-1.39740789e+00 -2.48109385e-01 2.50876904e-01 -5.90126514e-01
9.90217328e-01 1.99562699e-01 7.94432282e-01 5.20804524e-01
6.57024860e-01 6.51468515e-01 9.01491046e-01 -4.48972315e-01
3.05228710e-01 3.52318317e-01 1.53813930e-02 5.17057657e-01
5.80107629e-01 -1.57579705e-01 -1.88189626e-01 4.77210581e-02
8.34246814e-01 8.09462667e-02 1.79256648e-02 -1.17574856e-01
-9.13457990e-01 8.84266078e-01 5.31298757e-01 4.94747072e-01
-1.07556507e-01 -3.89209087e-03 5.76488674e-01 -2.28740528e-01
4.45509583e-01 4.68820184e-01 -1.50577381e-01 2.37245992e-01
-1.14446354e+00 -1.37934713e-02 3.84776294e-01 3.28913629e-01
3.78899664e-01 -3.38172376e-01 -3.49058717e-01 7.04082251e-01
2.46659204e-01 1.09369732e-01 6.93818212e-01 -3.98772657e-01
-1.14049460e-03 1.11169410e+00 -2.04788774e-01 -8.95373940e-01
-8.55051219e-01 -6.50077999e-01 -1.01401305e+00 4.73255068e-01
7.67838955e-01 2.31619209e-01 -1.42344880e+00 1.19798005e+00
3.93297523e-01 -4.79040854e-02 -1.99008256e-01 7.46071815e-01
8.57186675e-01 2.27222204e-01 4.27274287e-01 9.38442256e-03
1.59381723e+00 -6.82389557e-01 -4.29063231e-01 -1.67682216e-01
1.09864593e+00 -7.75641978e-01 7.69579470e-01 4.51054610e-02
-8.76111031e-01 -3.07524294e-01 -1.29836845e+00 -2.53740102e-01
-5.62484264e-01 5.75434089e-01 4.80358422e-01 5.10370851e-01
-1.07142520e+00 2.41953135e-01 -1.25163090e+00 -4.84940141e-01
8.36010635e-01 7.27247596e-01 -3.14586997e-01 1.00119591e-01
-5.43922722e-01 9.43594575e-01 2.68689692e-01 3.52880895e-01
-5.31023383e-01 -1.06288743e+00 -5.99554420e-01 9.43942368e-02
-7.54184872e-02 -5.53945541e-01 1.18241251e+00 -7.40367293e-01
-1.07549691e+00 1.21755219e+00 -2.00495243e-01 -5.13661683e-01
5.84443092e-01 3.85656178e-01 -3.17909010e-02 1.91091150e-01
1.88079312e-01 1.00659585e+00 4.00866807e-01 -7.81550705e-01
-1.07282162e+00 -3.58928323e-01 -4.31781292e-01 -6.72694072e-02
-1.44433320e-01 -3.33831728e-01 -3.79455388e-01 -5.80949128e-01
8.41538161e-02 -7.00622559e-01 -3.09756577e-01 4.12188053e-01
-1.78473480e-02 -1.60027117e-01 7.38260090e-01 -6.76761687e-01
9.81670201e-01 -2.19310236e+00 -4.48168576e-01 3.07512939e-01
3.58323753e-01 4.63541210e-01 -6.45155758e-02 -1.57795787e-01
1.95106454e-02 4.05639023e-01 8.11259598e-02 -3.65383267e-01
-3.33064139e-01 -2.58314908e-01 2.23077059e-01 7.58809566e-01
6.01151049e-01 9.03426826e-01 -6.12037301e-01 -6.62570953e-01
1.58246905e-01 5.17732978e-01 -5.16356647e-01 9.48987901e-02
-2.32547633e-02 1.78718805e-01 1.27889797e-01 9.16687787e-01
7.98888683e-01 -5.19632220e-01 3.02119672e-01 -4.55063164e-01
9.29377824e-02 1.04487278e-01 -9.70651627e-01 1.10418415e+00
-2.17901230e-01 8.21827710e-01 1.89381510e-01 -6.78679883e-01
6.73478603e-01 2.30177075e-01 5.63805163e-01 -6.49310470e-01
3.60189021e-01 3.75924081e-01 2.64141142e-01 -4.84014422e-01
4.82310534e-01 -3.36319983e-01 2.35013798e-01 1.59971774e-01
-1.56535774e-01 -1.20374896e-01 3.01303983e-01 4.89627980e-02
1.00354540e+00 -4.43641752e-01 5.48338711e-01 -2.34644130e-01
4.96155083e-01 3.29491556e-01 8.35391343e-01 4.08955246e-01
-4.62248176e-01 8.10200691e-01 7.50142574e-01 -5.97000241e-01
-1.01705921e+00 -7.36476839e-01 -2.91167855e-01 6.02830946e-01
5.85907586e-02 -7.37246498e-02 -5.12951732e-01 -6.48495138e-01
7.90799409e-02 2.58761078e-01 -9.30481136e-01 -1.34862661e-01
-4.38715607e-01 -1.20107245e+00 6.78433478e-01 9.06832039e-01
3.32852781e-01 -8.46372426e-01 -7.18041539e-01 3.15814912e-01
-2.99502760e-02 -8.69108975e-01 -2.97907710e-01 4.56850708e-01
-8.38747025e-01 -1.47519612e+00 -6.01442695e-01 -8.50557983e-01
1.12109649e+00 1.41962796e-01 7.03857780e-01 5.53814530e-01
-8.25988173e-01 -3.45304012e-01 -3.70203890e-02 -6.06632590e-01
-6.27228081e-01 2.30888054e-01 -3.47141504e-01 -2.66566634e-01
7.20047235e-01 7.33203888e-02 -8.54073226e-01 3.03296506e-01
-1.04301727e+00 3.46757978e-01 9.91131127e-01 1.27147877e+00
9.19788957e-01 1.21247098e-01 6.37563646e-01 -8.65304589e-01
9.04456377e-02 -3.13489974e-01 -6.56895280e-01 1.07684448e-01
-6.99609935e-01 -2.01558992e-01 3.09304893e-01 -4.99591500e-01
-8.42561245e-01 1.79425806e-01 -3.24286252e-01 1.23985382e-02
-5.06290011e-02 5.13130009e-01 1.66625559e-01 -2.35356525e-01
6.63449645e-01 -1.70619145e-01 4.68883842e-01 3.51486281e-02
-4.32811826e-01 6.84943914e-01 5.44493258e-01 1.95495203e-01
4.01743412e-01 6.09042168e-01 2.52647042e-01 -5.10567844e-01
-4.39151287e-01 -7.26349115e-01 -5.04911304e-01 -1.45576596e-01
8.37086678e-01 -9.03123736e-01 -5.37476480e-01 8.45901847e-01
-9.13057625e-01 -4.37946945e-01 -3.50209996e-02 4.32519495e-01
-1.33456737e-01 6.93108067e-02 -9.36779797e-01 -4.23499405e-01
-7.41641283e-01 -1.19099224e+00 1.15194452e+00 7.16168404e-01
-3.89286578e-01 -9.35659587e-01 -1.06323436e-01 5.02963185e-01
4.91912663e-01 3.01623285e-01 1.07618487e+00 -8.00675273e-01
-4.66704428e-01 -5.51907837e-01 -3.89952421e-01 -1.27380461e-01
1.49324045e-01 4.80801284e-01 -1.04897869e+00 -5.02441764e-01
-4.20729011e-01 -1.59881011e-01 8.96772206e-01 6.27674937e-01
1.13557315e+00 9.88473818e-02 -9.09150481e-01 7.74054110e-01
1.47921550e+00 2.65373170e-01 6.64024055e-01 6.48569465e-01
2.69947797e-01 3.94773483e-01 6.66832566e-01 7.82935619e-02
1.52292415e-01 4.12515968e-01 3.12871754e-01 -6.34005308e-01
-3.89382184e-01 -1.74026005e-02 -6.34192526e-02 2.09034398e-01
2.30051711e-01 -5.64058833e-02 -1.12426615e+00 8.34173858e-01
-1.41549051e+00 -6.58396125e-01 -2.01803207e-01 1.87705767e+00
8.94433022e-01 2.62738496e-01 5.72052458e-03 4.29473072e-01
7.35896111e-01 -3.98865700e-01 -6.88589215e-01 -7.70694390e-02
1.01455858e-04 2.05288619e-01 4.68939215e-01 2.60299027e-01
-1.14456582e+00 5.57293773e-01 5.78833532e+00 9.39536691e-01
-1.54576349e+00 -3.32994983e-02 1.09957480e+00 -2.32888415e-01
2.28632808e-01 -3.17878008e-01 -1.06056142e+00 3.41186702e-01
6.47459149e-01 7.70675316e-02 -3.38986576e-01 5.46639442e-01
8.36480036e-02 -5.86225867e-01 -1.09768772e+00 7.79541016e-01
-1.41797572e-01 -1.78198278e+00 -2.06334859e-01 2.08592460e-01
4.35327917e-01 -1.08055994e-01 -5.43725491e-02 1.37423441e-01
-5.55111878e-02 -9.73614991e-01 2.56568551e-01 2.47560948e-01
1.04470778e+00 -7.54575133e-01 1.51128483e+00 2.46806949e-01
-1.09247148e+00 6.64200028e-03 -2.82471806e-01 3.64934593e-01
-2.01671094e-01 6.67689323e-01 -1.71538758e+00 3.78277339e-02
6.02151990e-01 2.38646358e-01 -9.34244752e-01 1.43051505e+00
2.17817634e-01 3.81530344e-01 -3.69380534e-01 -2.47809380e-01
-9.58969370e-02 5.74061036e-01 1.61579221e-01 1.44902778e+00
3.64864409e-01 -2.19080731e-01 -1.74315482e-01 7.61744857e-01
9.80826169e-02 -1.70941297e-02 2.01680273e-01 3.00910417e-02
5.94865084e-01 1.70730686e+00 -1.53103983e+00 -1.26416340e-01
-8.14380646e-02 3.91635120e-01 2.51990855e-01 -1.93682499e-03
-8.24903488e-01 -4.36998516e-01 3.22556347e-01 5.27101696e-01
2.32704118e-01 2.06331536e-01 -8.09462905e-01 -6.30815625e-01
-3.04292142e-01 -6.89525723e-01 6.25892282e-01 -2.42433548e-01
-1.15786934e+00 2.67205477e-01 -4.36622530e-01 -1.09274626e+00
7.03131706e-02 -6.79902554e-01 -8.11947465e-01 7.48319447e-01
-1.59518623e+00 -1.29972231e+00 -5.10417461e-01 2.97408812e-02
5.98726630e-01 7.61065632e-02 7.32267201e-01 2.48168081e-01
-5.83205879e-01 9.42710578e-01 -9.54731554e-02 2.29448080e-01
8.74047935e-01 -1.37262905e+00 2.58646104e-02 7.07007289e-01
-5.18677354e-01 3.90667617e-01 3.69038403e-01 -5.91387570e-01
-9.17965770e-01 -1.23371768e+00 9.20511365e-01 -1.93879664e-01
4.75255877e-01 -2.96269178e-01 -8.94789994e-01 3.25048178e-01
-3.07965606e-01 8.83983597e-02 1.06740677e+00 -3.14329714e-01
3.38803716e-02 -1.77363500e-01 -1.43757594e+00 4.52777326e-01
1.15813933e-01 -2.30131477e-01 1.53556536e-03 8.26092809e-02
4.38665226e-02 -7.09605396e-01 -9.01347995e-01 6.30501866e-01
7.47690499e-01 -7.40687430e-01 6.59254193e-01 3.81172821e-02
3.75532150e-01 -4.97847080e-01 3.44861746e-01 -9.49978232e-01
-5.90171695e-01 2.08619669e-01 1.02541305e-01 1.19359803e+00
5.71806610e-01 -4.12730128e-01 1.15598416e+00 4.46432501e-01
-1.66082174e-01 -1.15110326e+00 -1.04065681e+00 -6.95457309e-02
1.97063863e-01 7.69239590e-02 3.64136368e-01 4.77013499e-01
-3.05747311e-03 9.64913741e-02 4.11575228e-01 2.55069375e-01
3.52004111e-01 -1.94670171e-01 6.34239495e-01 -1.07285500e+00
-1.01949975e-01 -8.45745265e-01 -7.24570453e-01 -4.84240919e-01
-3.37313831e-01 -8.15471828e-01 1.56605795e-01 -1.51708853e+00
5.37014842e-01 -6.03776753e-01 -2.55500346e-01 5.65054297e-01
-5.26856899e-01 5.54825068e-01 -1.06159493e-01 2.27213979e-01
-2.53749311e-01 -8.41715261e-02 9.89487469e-01 -3.25764805e-01
-1.44741043e-01 2.65896618e-02 -7.71656036e-01 6.45964622e-01
8.37114334e-01 -4.40051407e-01 -8.55490565e-02 -1.34347275e-01
1.24870896e-01 -2.62989372e-01 5.29164493e-01 -1.26039565e+00
6.62378252e-01 1.80748384e-02 1.10294497e+00 -1.02697074e+00
1.35021761e-01 -6.05176330e-01 1.63253963e-01 8.43881428e-01
-1.95795536e-01 -2.34608889e-01 6.32090449e-01 4.73215282e-01
-7.24276528e-02 -5.12711480e-02 1.27280307e+00 8.24914649e-02
-3.48355591e-01 1.02892041e-01 -7.27671027e-01 -5.71846604e-01
1.25091326e+00 -7.13609338e-01 -6.03520751e-01 1.16918370e-01
-5.92006981e-01 4.08404738e-01 4.74649757e-01 -2.42333803e-02
5.31971097e-01 -9.82971787e-01 -5.63933372e-01 3.41730833e-01
1.45095602e-01 4.17997688e-01 4.38264281e-01 1.32590318e+00
-8.76803756e-01 4.30226654e-01 -2.34527558e-01 -8.83577704e-01
-1.59652317e+00 1.27356216e-01 6.27481759e-01 -8.58688056e-01
-3.45720559e-01 1.09455621e+00 5.37380949e-02 -2.02581450e-01
3.45400155e-01 -5.42772293e-01 -4.27477211e-01 1.16284929e-01
5.96435666e-01 3.70986521e-01 4.99742359e-01 -2.99151510e-01
-4.35306162e-01 2.70221144e-01 -7.46988177e-01 3.01949501e-01
1.30183542e+00 1.13829061e-01 -1.01798132e-01 1.03439666e-01
1.00709593e+00 -2.16227964e-01 -1.22481191e+00 8.80297646e-02
1.45133406e-01 -9.86407995e-02 2.07633108e-01 -1.01491714e+00
-1.14416921e+00 7.70635188e-01 9.53333497e-01 -9.02199149e-02
1.21959901e+00 -1.37949601e-01 5.63695312e-01 -1.41443819e-01
-9.55058783e-02 -1.04892385e+00 -1.44654857e-02 8.62582251e-02
4.50717837e-01 -1.34405637e+00 2.03969151e-01 -3.97185624e-01
-2.41575867e-01 1.40818083e+00 9.94665504e-01 1.60777211e-01
3.51164043e-01 6.96729660e-01 2.34082505e-01 -2.68080324e-01
-8.95057976e-01 2.12756082e-01 7.88077116e-02 5.45622528e-01
7.48817265e-01 2.70376052e-03 -4.59420770e-01 7.78644145e-01
9.32857841e-02 2.49072909e-01 4.84497249e-01 9.94689107e-01
-4.56036329e-01 -7.74068356e-01 -3.73846740e-01 7.43818164e-01
-7.21099198e-01 2.05113262e-01 -3.32186759e-01 9.43980217e-01
2.07530305e-01 7.08720684e-01 4.55892384e-01 -2.83259809e-01
-1.49820149e-02 -2.64034688e-01 2.25512117e-01 -5.93091130e-01
-6.86453223e-01 3.73931885e-01 -1.27786562e-01 -2.70995080e-01
-2.81602323e-01 -5.08366942e-01 -1.52468467e+00 -2.75131613e-01
-6.49761975e-01 -1.34176478e-01 6.40596569e-01 6.80015206e-01
3.17562729e-01 7.80201793e-01 2.12001652e-01 -7.26450324e-01
-3.35443318e-01 -1.12194002e+00 -5.08593917e-01 -7.43109267e-03
3.88160795e-01 -5.19060671e-01 -3.79894316e-01 3.20700973e-01]
|
[15.069877624511719, -3.096856117248535]
|
03001b1e-4d62-407e-a903-2b6a91c5ae06
|
blockwise-stochastic-variance-reduced-methods
|
2305.18730
| null |
https://arxiv.org/abs/2305.18730v2
|
https://arxiv.org/pdf/2305.18730v2.pdf
|
Blockwise Stochastic Variance-Reduced Methods with Parallel Speedup for Multi-Block Bilevel Optimization
|
In this paper, we consider non-convex multi-block bilevel optimization (MBBO) problems, which involve $m\gg 1$ lower level problems and have important applications in machine learning. Designing a stochastic gradient and controlling its variance is more intricate due to the hierarchical sampling of blocks and data and the unique challenge of estimating hyper-gradient. We aim to achieve three nice properties for our algorithm: (a) matching the state-of-the-art complexity of standard BO problems with a single block; (b) achieving parallel speedup by sampling $I$ blocks and sampling $B$ samples for each sampled block per-iteration; (c) avoiding the computation of the inverse of a high-dimensional Hessian matrix estimator. However, it is non-trivial to achieve all of these by observing that existing works only achieve one or two of these properties. To address the involved challenges for achieving (a, b, c), we propose two stochastic algorithms by using advanced blockwise variance-reduction techniques for tracking the Hessian matrices (for low-dimensional problems) or the Hessian-vector products (for high-dimensional problems), and prove an iteration complexity of $O(\frac{m\epsilon^{-3}\mathbb{I}(I<m)}{I\sqrt{I}} + \frac{m\epsilon^{-3}}{I\sqrt{B}})$ for finding an $\epsilon$-stationary point under appropriate conditions. We also conduct experiments to verify the effectiveness of the proposed algorithms comparing with existing MBBO algorithms.
|
['Tianbao Yang', 'Lijun Zhang', 'Zhishuai Guo', 'Zi-Hao Qiu', 'Quanqi Hu']
|
2023-05-30
| null | null | null | null |
['bilevel-optimization']
|
['methodology']
|
[-2.78067142e-02 -2.43585765e-01 -5.97643405e-02 -7.96607360e-02
-9.57166374e-01 -3.38305414e-01 7.51859546e-02 2.75841653e-01
-6.57923162e-01 7.98761845e-01 -1.81862906e-01 -6.55000865e-01
-6.85651302e-01 -4.75600332e-01 -8.92073631e-01 -1.04111159e+00
-6.18944824e-01 2.58117765e-01 1.64993554e-01 -1.18845254e-01
3.80278558e-01 3.13105583e-01 -1.44931912e+00 -1.22424223e-01
1.07840335e+00 1.34071159e+00 1.92718759e-01 8.46898377e-01
6.90511838e-02 5.09887218e-01 -1.44441515e-01 -3.26583296e-01
5.82006514e-01 -7.52260923e-01 -6.73548698e-01 4.39640731e-02
3.70582521e-01 -1.33832559e-01 9.41902623e-02 1.36939704e+00
3.74093801e-01 2.64910519e-01 6.42648280e-01 -1.01448214e+00
9.46001709e-02 2.49258339e-01 -1.09217298e+00 2.92638004e-01
5.95052466e-02 1.02725744e-01 1.11315942e+00 -7.79073477e-01
1.16062902e-01 9.27101195e-01 7.14572012e-01 7.34969527e-02
-1.09242332e+00 -6.20968878e-01 2.63223112e-01 6.93358928e-02
-1.49053180e+00 -2.65413970e-01 6.10002577e-01 -5.13797998e-01
5.63791454e-01 6.33007586e-01 7.16960847e-01 1.28145218e-01
-6.29581185e-03 7.64723241e-01 1.20732737e+00 -3.66175085e-01
2.11264655e-01 2.00672179e-01 2.76747972e-01 1.02559471e+00
4.25141871e-01 4.46183644e-02 -3.50420207e-01 -1.75063699e-01
7.70020366e-01 9.21216235e-02 -3.34635377e-01 -5.00832021e-01
-1.18201756e+00 1.02627361e+00 3.38455707e-01 1.38218865e-01
-3.54100317e-01 4.86481607e-01 1.97976083e-01 2.09950924e-01
4.26182151e-01 7.84386173e-02 -5.08288860e-01 -3.98897439e-01
-1.05131340e+00 4.71095502e-01 7.78690875e-01 1.05767369e+00
8.69213641e-01 6.33070171e-02 -5.93226068e-02 7.02956855e-01
4.03310895e-01 6.91974819e-01 2.58919895e-01 -6.47714198e-01
8.08186054e-01 3.74749511e-01 4.04104173e-01 -1.08503795e+00
-4.64323133e-01 -4.75036085e-01 -1.19733751e+00 8.10446739e-02
7.37063825e-01 -2.61603117e-01 -4.66347814e-01 1.51827788e+00
5.87178767e-01 -1.32724360e-01 -4.03788477e-01 8.57789874e-01
1.54944226e-01 7.95927048e-01 -3.25499862e-01 -7.26812243e-01
1.36257088e+00 -9.46081400e-01 -3.70456517e-01 -2.77123392e-01
5.92775106e-01 -6.59217894e-01 1.24512529e+00 4.26379114e-01
-1.27305496e+00 -2.81907678e-01 -1.14259791e+00 1.45944864e-01
3.44788772e-03 1.40953317e-01 6.91941023e-01 8.05475175e-01
-8.04870427e-01 5.75368822e-01 -8.65923703e-01 3.40500891e-01
1.95397720e-01 6.03257656e-01 1.88146364e-02 3.11437324e-02
-7.26485014e-01 3.98936749e-01 -3.58557887e-02 4.15692419e-01
-8.19598556e-01 -8.57947290e-01 -7.27352440e-01 1.66036144e-01
4.34529752e-01 -2.99621344e-01 7.58220553e-01 -6.19901776e-01
-1.36571717e+00 5.17611742e-01 -3.03077191e-01 -3.36718410e-01
7.44385898e-01 -2.19725817e-01 2.74129659e-01 -2.84097847e-02
8.75217766e-02 1.79421112e-01 8.19926560e-01 -8.44238281e-01
-7.80724227e-01 -8.04637611e-01 8.31244960e-02 1.88274056e-01
-4.28703249e-01 -3.75968255e-02 -5.38012028e-01 -6.09139979e-01
3.36366862e-01 -9.40153301e-01 -5.14588952e-01 -1.59880340e-01
-2.22146824e-01 1.22675806e-01 2.93974787e-01 -6.96139455e-01
1.48891604e+00 -2.04491091e+00 2.71712035e-01 4.23938006e-01
1.82496727e-01 2.64262676e-01 5.01069546e-01 3.00821453e-01
1.47457659e-01 1.61962837e-01 -2.06020936e-01 -1.36616006e-01
1.09110519e-01 -2.37957984e-01 -4.48327549e-02 7.78060257e-01
-2.97147453e-01 3.62215638e-01 -8.61583471e-01 -4.45341557e-01
1.19619533e-01 3.01355660e-01 -7.41105914e-01 -2.29255453e-01
1.35824785e-01 2.40944773e-01 -4.92882252e-01 2.96297759e-01
8.32768679e-01 -4.34957087e-01 3.23625922e-01 -1.71747014e-01
-2.45576769e-01 1.21385932e-01 -2.12469816e+00 1.17202973e+00
-5.00112534e-01 2.03797922e-01 6.40711963e-01 -1.27563679e+00
6.01513863e-01 3.87138105e-03 6.57230973e-01 -4.62272406e-01
6.01619519e-02 3.84374738e-01 -2.48112261e-01 -1.68830097e-01
8.10773596e-02 -3.53908867e-01 1.23418763e-03 4.36860681e-01
-4.74049181e-01 -1.00854494e-01 4.81427461e-01 -1.00703798e-01
1.05091584e+00 -2.52632558e-01 3.19176883e-01 -7.80683100e-01
8.34735751e-01 -1.90513015e-01 7.01436102e-01 8.70216966e-01
-2.40583628e-01 1.52789041e-01 8.57931972e-01 -4.64312851e-01
-8.64580631e-01 -8.16888928e-01 -2.67091066e-01 1.08651388e+00
2.50529081e-01 -2.82838970e-01 -7.19430327e-01 -5.70407927e-01
4.46574353e-02 3.17095190e-01 -5.97903132e-01 2.20193565e-01
-6.51115060e-01 -1.19578516e+00 6.49340376e-02 4.30406392e-01
5.26351988e-01 -5.11386752e-01 -7.56498992e-01 1.18745923e-01
4.42105941e-02 -5.83228528e-01 -8.19877803e-01 4.11228359e-01
-9.45864737e-01 -9.60113049e-01 -8.63750815e-01 -6.32429421e-01
9.39701974e-01 3.99235666e-01 7.06961274e-01 3.53725031e-02
-2.24076241e-01 1.94670893e-02 -3.78408208e-02 -3.19947362e-01
7.09734261e-02 3.92287225e-02 5.49939945e-02 -3.81118245e-02
-5.16084908e-03 -4.43105757e-01 -9.89774346e-01 5.19793570e-01
-8.09656501e-01 9.36321840e-02 5.79682469e-01 1.02097249e+00
7.60242581e-01 7.65723065e-02 2.48781428e-01 -7.04894304e-01
6.13835275e-01 -1.91688359e-01 -1.37108004e+00 6.21515028e-02
-7.25132942e-01 2.68727630e-01 7.64443934e-01 -3.35064918e-01
-6.16170287e-01 6.99013621e-02 -9.53459889e-02 -2.74680078e-01
4.87069517e-01 4.52136725e-01 1.89315826e-01 -9.23783798e-03
3.91149461e-01 2.98677742e-01 9.76172276e-03 -4.37953591e-01
2.98435867e-01 4.82482404e-01 1.11178979e-01 -7.14216173e-01
5.88977098e-01 6.31363213e-01 2.99807936e-01 -6.80457294e-01
-9.63932812e-01 -6.48921967e-01 -3.63494515e-01 -1.81599800e-02
6.35227799e-01 -7.08233833e-01 -1.45651901e+00 7.81946406e-02
-5.87912381e-01 -2.50479758e-01 -1.43898189e-01 6.40739024e-01
-6.32474363e-01 4.13932234e-01 -4.47314560e-01 -1.13774300e+00
-5.42601943e-01 -1.46727943e+00 7.55981386e-01 5.29098287e-02
1.05714478e-01 -8.32417011e-01 -1.29612058e-01 4.63015676e-01
2.81817019e-01 1.99365895e-02 8.42010379e-01 -2.11555436e-01
-6.59919500e-01 -3.81193876e-01 -3.97137463e-01 5.12153268e-01
-1.31108016e-01 -4.14039046e-01 -2.85814196e-01 -6.16231501e-01
3.03390265e-01 -6.22516759e-02 6.83364391e-01 5.67048192e-01
1.12562072e+00 -6.36399865e-01 -2.25712597e-01 7.17207432e-01
1.58049941e+00 2.55957812e-01 4.18526351e-01 1.96241885e-01
4.70090330e-01 2.75891721e-01 6.69005930e-01 8.69093180e-01
2.08255082e-01 7.15587854e-01 3.89389038e-01 -7.50439689e-02
3.11481446e-01 1.34916872e-01 3.18444103e-01 8.91692460e-01
2.29762252e-02 2.29261130e-01 -8.65642428e-01 8.85915160e-01
-1.78512132e+00 -7.18337774e-01 -3.71554434e-01 2.63110352e+00
1.00659013e+00 1.70253694e-01 1.61881462e-01 2.06587926e-01
5.22171795e-01 1.13402411e-01 -4.85113859e-01 -3.82357985e-01
3.57888550e-01 4.67156202e-01 8.17737639e-01 5.93100905e-01
-1.02337432e+00 6.36721611e-01 5.31411457e+00 1.08075011e+00
-9.30539608e-01 5.67219518e-02 7.20300078e-01 -4.70774472e-01
-2.62780450e-02 1.78359210e-01 -1.11914647e+00 6.48423731e-01
5.75121582e-01 2.27808356e-02 6.57898962e-01 9.45267320e-01
3.06114048e-01 -4.64513183e-01 -9.20537114e-01 1.16612697e+00
-8.22561383e-02 -1.23517346e+00 -3.30971807e-01 3.09478581e-01
9.27693963e-01 -1.81085825e-01 4.88001332e-02 1.44644916e-01
4.26102996e-01 -8.27690303e-01 4.97734368e-01 -2.85850298e-02
5.06409585e-01 -8.58765543e-01 5.29495358e-01 4.42691118e-01
-1.23474050e+00 -2.43919060e-01 -3.60259682e-01 -6.69493005e-02
1.17685154e-01 9.80860054e-01 -3.20986778e-01 4.31196809e-01
1.11964786e+00 1.36033937e-01 3.14644608e-03 8.71918559e-01
-2.78474316e-02 3.39707077e-01 -7.11543262e-01 -4.20508504e-01
4.67151135e-01 -7.00576305e-01 4.91822273e-01 1.07186019e+00
2.51803190e-01 3.46566439e-02 1.57022640e-01 4.80362833e-01
5.20961620e-02 5.35609305e-01 -4.25996371e-02 2.20273450e-01
1.23835944e-01 1.02193463e+00 -7.48395860e-01 -2.56858408e-01
-4.84376460e-01 6.63410068e-01 3.71376693e-01 2.71172881e-01
-9.83474970e-01 -7.10689008e-01 8.10399532e-01 1.58056855e-01
8.68699610e-01 -4.70096588e-01 -5.24388790e-01 -1.08127153e+00
4.61169511e-01 -9.04676139e-01 3.76291960e-01 -1.41661733e-01
-8.13487709e-01 2.10167989e-01 7.02815950e-02 -1.01190805e+00
-1.37650877e-01 -4.58127171e-01 -6.33401200e-02 8.12414169e-01
-1.27633309e+00 -6.54246151e-01 -2.00681925e-01 6.23906553e-01
3.00582081e-01 1.24602638e-01 3.21604580e-01 5.06091356e-01
-6.44726694e-01 6.30612075e-01 6.28906608e-01 -7.18618333e-02
2.93218225e-01 -1.11544001e+00 -1.30371779e-01 8.05548489e-01
-7.27299973e-02 5.86551726e-01 8.33059072e-01 -3.41582060e-01
-1.67820811e+00 -6.25111401e-01 6.71859741e-01 -4.54860218e-02
6.78772688e-01 -3.62834603e-01 -4.58032131e-01 5.52242398e-01
-2.80276686e-01 3.65280025e-02 4.24096435e-01 3.07284668e-02
1.52670860e-01 -6.50684237e-01 -1.05110848e+00 6.48291171e-01
8.36431324e-01 -1.93011127e-02 5.03340177e-02 4.80670005e-01
2.06809610e-01 -4.80087131e-01 -9.90365148e-01 5.59361696e-01
5.88530183e-01 -1.09886456e+00 1.08262444e+00 -4.22018021e-01
1.48809999e-01 -2.82048941e-01 -1.69286221e-01 -8.54458392e-01
-1.32304924e-02 -9.89696085e-01 -8.04653093e-02 7.23355412e-01
4.23450321e-01 -5.92314601e-01 9.20554578e-01 5.19515872e-01
1.60056546e-01 -1.21683896e+00 -1.22486091e+00 -6.87807143e-01
1.97687387e-01 -4.03971881e-01 1.74561918e-01 6.13027751e-01
5.84311364e-03 1.66828185e-01 -6.09717667e-01 1.30583331e-01
6.47526860e-01 1.30704775e-01 9.44397509e-01 -6.26691699e-01
-6.97661579e-01 -5.27836740e-01 -1.17920816e-01 -1.57687116e+00
-4.71258819e-01 -5.20174563e-01 4.48999330e-02 -1.29092944e+00
3.12847346e-01 -8.41692567e-01 -2.02418298e-01 1.16896629e-01
-2.91948408e-01 -4.63342434e-03 1.51553407e-01 3.41332518e-02
-5.14075220e-01 5.13641179e-01 1.10211384e+00 4.40722741e-02
-3.32807183e-01 2.59156555e-01 -6.60648167e-01 7.09435463e-01
3.67520452e-01 -4.42988992e-01 -4.94732499e-01 -4.45400149e-01
6.02016389e-01 4.85632479e-01 1.26310021e-01 -8.23758900e-01
1.89415932e-01 -2.14407086e-01 3.28511149e-02 -7.36988783e-01
3.20901603e-01 -5.80529153e-01 -1.93644892e-02 6.78598285e-01
2.01931992e-03 2.72462070e-01 1.32555380e-01 5.85691631e-01
-7.17725381e-02 -5.52837610e-01 1.04792631e+00 -2.05932081e-01
-4.78286929e-02 1.83568060e-01 -3.06277126e-01 1.01621568e-01
1.06275153e+00 -7.04803094e-02 -6.59455284e-02 -5.33867359e-01
-4.86405104e-01 3.72721702e-01 8.92861933e-02 -1.67258576e-01
2.14123502e-01 -1.05560553e+00 -6.08126163e-01 2.02632323e-01
-4.52382267e-01 3.81098002e-01 3.67590874e-01 1.35380626e+00
-8.39791059e-01 4.12760556e-01 3.49717289e-01 -7.34606445e-01
-1.16359603e+00 5.93986928e-01 2.34037206e-01 -7.88424075e-01
-2.59328455e-01 9.88571405e-01 1.07797138e-01 -1.54270023e-01
3.66965443e-01 -2.99315959e-01 4.33430672e-01 3.65605764e-02
4.64787155e-01 9.20424998e-01 1.40739962e-01 -2.03440502e-01
-3.05237532e-01 5.43230891e-01 -1.82011992e-01 -2.22679526e-01
1.30333328e+00 -1.30197197e-01 -2.94212133e-01 1.76881656e-01
1.33948004e+00 8.07224959e-02 -1.29958510e+00 -1.61526389e-02
-1.25622049e-01 -4.24556792e-01 1.75615236e-01 -3.12989682e-01
-1.00180054e+00 9.88436937e-01 6.40711963e-01 1.65275484e-01
1.12276065e+00 -2.37708747e-01 7.12540329e-01 2.18798786e-01
3.93452317e-01 -1.24276125e+00 8.52397978e-02 4.20260370e-01
7.10803628e-01 -1.06618190e+00 5.29637456e-01 -3.95307809e-01
-4.57313418e-01 8.10944796e-01 3.02188307e-01 -1.72015324e-01
1.04996467e+00 -9.72221345e-02 -2.70743877e-01 -1.77486911e-01
-3.83116484e-01 -1.05260968e-01 3.02802473e-01 -2.46356234e-01
3.45636755e-01 -1.02255121e-01 -9.05686736e-01 3.94755155e-01
-1.85110658e-01 -1.70011282e-01 1.70539066e-01 1.01252139e+00
-3.45812529e-01 -9.24484313e-01 -3.34487498e-01 5.58052540e-01
-6.53641343e-01 -1.60977840e-01 2.16738075e-01 8.70226383e-01
-2.20573634e-01 9.20378089e-01 -1.44875407e-01 7.87751079e-02
1.34121910e-01 -2.07432415e-02 4.73529845e-01 -1.59608126e-01
-4.97758895e-01 3.66339028e-01 -1.77934796e-01 -3.60102981e-01
-2.14392528e-01 -8.89410615e-01 -1.02073169e+00 -3.69813591e-01
-5.13642788e-01 3.01191688e-01 9.11463618e-01 8.57551932e-01
2.38194808e-01 1.27189398e-01 7.71095097e-01 -7.48597443e-01
-1.03053451e+00 -8.81338418e-01 -6.64192438e-01 3.71458977e-01
2.36249343e-01 -5.67343056e-01 -5.44741452e-01 -1.60929292e-01]
|
[6.584084510803223, 4.501469135284424]
|
650b0ddb-0a11-47f0-968a-a07a31108390
|
submodular-minimax-optimization-finding
|
2305.16903
| null |
https://arxiv.org/abs/2305.16903v1
|
https://arxiv.org/pdf/2305.16903v1.pdf
|
Submodular Minimax Optimization: Finding Effective Sets
|
Despite the rich existing literature about minimax optimization in continuous settings, only very partial results of this kind have been obtained for combinatorial settings. In this paper, we fill this gap by providing a characterization of submodular minimax optimization, the problem of finding a set (for either the min or the max player) that is effective against every possible response. We show when and under what conditions we can find such sets. We also demonstrate how minimax submodular optimization provides robust solutions for downstream machine learning applications such as (i) efficient prompt engineering for question answering, (ii) prompt engineering for dialog state tracking, (iii) identifying robust waiting locations for ride-sharing, (iv) ride-share difficulty kernelization, and (v) finding adversarial images. Our experiments demonstrate that our proposed algorithms consistently outperform other baselines.
|
['Amin Karbasi', 'Moran Feldman', 'Ethan R. Elenberg', 'Loay Mualem']
|
2023-05-26
| null | null | null | null |
['prompt-engineering']
|
['natural-language-processing']
|
[ 2.04769686e-01 2.71777898e-01 -4.84599680e-01 -3.88032824e-01
-1.44308805e+00 -1.23526216e+00 3.63398582e-01 4.20583636e-01
-4.85550404e-01 7.70256519e-01 1.36568770e-01 -6.47050917e-01
-5.48209667e-01 -2.97597915e-01 -8.47003937e-01 -5.11566579e-01
1.89355090e-01 7.01332033e-01 6.34520724e-02 -3.53443831e-01
6.09517217e-01 4.56166297e-01 -9.15025830e-01 7.01209456e-02
6.29258275e-01 9.16107357e-01 9.65623334e-02 8.63166034e-01
1.09566107e-01 9.63282704e-01 -8.80772054e-01 -4.21880603e-01
5.08247137e-01 -3.76595646e-01 -1.05698323e+00 3.01897734e-01
7.35543847e-01 -5.69616079e-01 -5.85724115e-01 1.16637135e+00
5.33928514e-01 3.20663005e-01 2.66041487e-01 -1.81098509e+00
-6.44914031e-01 6.27647460e-01 -4.56764609e-01 2.63702631e-01
5.58762848e-01 7.41527200e-01 1.34294486e+00 -4.68126953e-01
5.69687128e-01 1.50088024e+00 9.78121459e-02 8.04807723e-01
-1.04365671e+00 -3.03523064e-01 3.07952255e-01 9.93865803e-02
-9.62740421e-01 -5.27078450e-01 5.20388007e-01 -9.58092734e-02
8.32973242e-01 6.94274068e-01 -7.73533136e-02 8.81310582e-01
-1.30646780e-01 9.83620107e-01 1.24700701e+00 -3.38655889e-01
1.76644117e-01 4.27511483e-01 5.67022800e-01 8.61931503e-01
-2.44085506e-01 -1.13715485e-01 -6.66031837e-01 -4.24307853e-01
3.00720662e-01 -2.08919525e-01 -4.12247002e-01 3.22672278e-02
-1.00982106e+00 1.14971399e+00 2.71372199e-01 -2.39861548e-01
-3.58349264e-01 4.09252495e-01 9.65020955e-02 6.19688570e-01
3.90263386e-02 8.38339925e-01 -5.13656795e-01 -6.89651743e-02
-3.45913619e-01 6.61102176e-01 1.15152335e+00 1.28296149e+00
5.51809669e-01 -9.45503116e-02 -7.51292646e-01 6.05565131e-01
7.31796846e-02 4.61881906e-01 -2.18641385e-01 -1.58180475e+00
1.04051566e+00 4.56177115e-01 5.88669419e-01 -8.48271728e-01
-3.99744451e-01 3.73448730e-01 -3.88968885e-01 1.11666247e-01
8.13099921e-01 -5.18611431e-01 -4.25583541e-01 2.07826900e+00
4.88167644e-01 1.52048352e-03 -2.34031588e-01 1.15929437e+00
5.57622015e-01 6.99102700e-01 -1.62239909e-01 -4.04992968e-01
1.34916663e+00 -1.06428742e+00 -8.32304418e-01 -3.51143509e-01
4.91610408e-01 -6.98677123e-01 1.23007429e+00 3.40417683e-01
-1.29875028e+00 9.15836692e-02 -7.12392569e-01 -2.95737654e-01
-2.22939685e-01 -3.73854101e-01 8.12904835e-01 8.36594939e-01
-1.05184793e+00 1.59163609e-01 -1.05681084e-01 -1.74622446e-01
1.90794572e-01 6.47369385e-01 -2.04811096e-01 -4.98984940e-02
-1.02129543e+00 7.57615983e-01 -2.89980710e-01 9.45523903e-02
-1.08515000e+00 -8.22970152e-01 -7.26695180e-01 1.27149388e-01
1.16326070e+00 -5.90348423e-01 1.48086154e+00 -7.14227736e-01
-1.51987696e+00 9.71454799e-01 -1.03350602e-01 -4.47080493e-01
4.54236150e-01 -2.77968884e-01 3.67288917e-01 5.95337860e-02
-2.32895538e-01 6.07305288e-01 7.20883071e-01 -1.15859973e+00
-5.60395718e-01 -4.77884293e-01 8.32166910e-01 3.67732614e-01
-4.40958768e-01 4.52571094e-01 -4.47287589e-01 -3.45124096e-01
-2.76159227e-01 -1.01208615e+00 -5.10446429e-01 1.88125923e-01
-1.08716452e+00 -4.69358951e-01 8.14184308e-01 -5.87364256e-01
1.23257113e+00 -1.72834909e+00 2.36855745e-01 2.26872087e-01
3.57555836e-01 7.06284344e-02 -4.40441430e-01 5.37693679e-01
2.80602694e-01 1.85180575e-01 1.09241130e-02 -4.61275607e-01
4.62915748e-01 1.44770399e-01 -7.10894108e-01 6.87062860e-01
-5.38042746e-02 1.01761091e+00 -6.52426481e-01 -4.29449171e-01
1.95397809e-02 -2.86948860e-01 -6.43160701e-01 4.69727695e-01
-5.92191160e-01 1.16321243e-01 -5.00311613e-01 7.80729353e-01
6.79194570e-01 -2.45044842e-01 2.89461404e-01 2.93904722e-01
5.58764227e-02 2.97622472e-01 -1.33890176e+00 1.19156218e+00
-2.58129090e-01 5.65315187e-01 8.31049383e-01 -1.06551921e+00
5.99660993e-01 1.06342584e-01 4.25251722e-01 -6.27304852e-01
1.74576312e-01 -1.40158594e-01 -4.36075389e-01 -6.43642247e-01
7.42075861e-01 -8.38855375e-03 -4.78350341e-01 7.87774742e-01
-2.21597925e-01 3.94937657e-02 -4.33701687e-02 6.06782794e-01
1.23169494e+00 -7.34954774e-01 7.78564513e-02 -1.50439724e-01
5.40215731e-01 7.30434060e-02 3.75248671e-01 1.29941726e+00
-5.24437726e-01 5.30270517e-01 1.11044014e+00 -1.10264532e-01
-7.27239370e-01 -1.00510883e+00 3.37600082e-01 1.53829396e+00
4.59102213e-01 -1.72391877e-01 -1.02324927e+00 -7.49536872e-01
2.80621529e-01 6.01365209e-01 -3.67079556e-01 7.76310861e-02
-8.92308593e-01 -3.41243267e-01 5.69227576e-01 2.54755288e-01
1.17326848e-01 -1.01456964e+00 -2.72152811e-01 -6.70419484e-02
-3.80259961e-01 -1.25767493e+00 -1.42734933e+00 1.50802061e-01
-6.84762239e-01 -1.23794055e+00 -4.13455665e-01 -7.04617977e-01
6.41259730e-01 6.18019044e-01 1.03669035e+00 4.74822879e-01
-2.42679060e-01 9.03702021e-01 -3.81154642e-02 -3.85389924e-01
-4.65074390e-01 -9.49178785e-02 6.05775565e-02 3.88220139e-02
2.45661382e-02 1.43537596e-01 -6.90174758e-01 5.15491307e-01
-9.37326252e-01 -5.96064687e-01 7.38334879e-02 7.00927138e-01
6.62740350e-01 -3.83213371e-01 7.28876173e-01 -1.14687753e+00
1.18245840e+00 -5.04499912e-01 -1.07155931e+00 5.69350064e-01
-5.18449306e-01 4.62962687e-02 6.13645017e-01 -4.61031467e-01
-8.29902351e-01 2.06076875e-02 9.39326212e-02 -3.76923829e-01
5.63034341e-02 1.26431109e-02 -2.43933573e-01 -9.54033881e-02
6.95702195e-01 1.44304723e-01 1.91048265e-01 -2.03072339e-01
5.77717483e-01 6.86574221e-01 5.05961239e-01 -7.17260838e-01
7.91832209e-01 2.64069051e-01 -9.65093076e-03 -7.16006875e-01
-9.99381900e-01 -6.40593708e-01 -9.61631015e-02 -1.59608081e-01
9.79906559e-01 -3.93822908e-01 -1.60830700e+00 3.27965826e-01
-1.31777322e+00 -6.19040370e-01 -1.97548732e-01 -1.00272380e-01
-7.06346214e-01 5.18327594e-01 -5.03505290e-01 -1.23195004e+00
-2.59975910e-01 -1.25114965e+00 1.08581734e+00 4.46853548e-01
-1.15561776e-01 -9.22473609e-01 -5.95651083e-02 1.11924827e+00
1.20876580e-01 8.17089975e-02 8.26846063e-01 -7.24198699e-01
-1.38149321e+00 -2.12576181e-01 -5.52185848e-02 1.64248928e-01
-3.86242419e-01 -4.36477453e-01 -7.79879332e-01 -4.32258517e-01
-1.09686546e-01 -7.26154506e-01 6.01319194e-01 5.24471104e-01
1.37651777e+00 -9.39166367e-01 -1.12857059e-01 5.88357985e-01
1.15168381e+00 1.14721902e-01 4.33670074e-01 3.29655021e-01
3.10350865e-01 8.50287318e-01 9.21630800e-01 6.43912196e-01
5.55573463e-01 8.18826497e-01 7.79215813e-01 1.07185848e-01
3.37733775e-01 -1.36830077e-01 4.96240526e-01 1.07561804e-01
5.95595121e-01 -8.45551491e-01 -3.38833988e-01 5.80563843e-01
-2.11721992e+00 -9.76975977e-01 -1.44079804e-01 2.14543366e+00
8.83901596e-01 -1.59560665e-01 4.02711093e-01 -2.81182617e-01
8.03543329e-01 3.71763468e-01 -7.98959792e-01 -7.11878359e-01
-2.28608057e-01 6.47130832e-02 6.85858905e-01 9.65543687e-01
-1.05748105e+00 1.13614964e+00 7.06946373e+00 7.48696387e-01
-3.87621313e-01 2.57367402e-01 7.83768654e-01 -2.94726521e-01
-3.85954529e-01 1.88028738e-01 -8.98882389e-01 3.12944502e-01
3.37107301e-01 -9.61286053e-02 9.65056300e-01 7.89913535e-01
2.06472799e-01 -1.10454276e-01 -1.07235694e+00 8.24924886e-01
-3.58285569e-02 -1.45691347e+00 -5.04409075e-01 2.89454907e-02
8.24391365e-01 -5.70189655e-01 3.28463316e-01 2.47455657e-01
6.26440823e-01 -1.23934209e+00 4.60255444e-01 1.92046002e-01
4.93098795e-01 -6.38577878e-01 1.50531247e-01 7.22274840e-01
-6.66898847e-01 -4.80383724e-01 -4.66335922e-01 2.71839052e-01
3.79610449e-01 1.26475930e-01 -8.49578261e-01 2.42630672e-02
2.22015291e-01 -2.19804183e-01 -1.00121930e-01 8.88805389e-01
-2.70123422e-01 7.08564520e-01 -4.63066280e-01 -4.59633172e-01
4.80987161e-01 -1.86290190e-01 8.00502360e-01 1.13232207e+00
-2.97430128e-01 3.27716529e-01 2.92581826e-01 9.81442332e-01
-4.54442859e-01 5.20150848e-02 -4.51644093e-01 6.21616514e-03
6.63856745e-01 1.27476966e+00 -4.74350899e-01 1.33659914e-01
-5.98036312e-02 9.96826172e-01 4.52387273e-01 4.45913374e-01
-1.13550067e+00 -3.72928917e-01 9.82814252e-01 -1.53970182e-01
2.57095490e-02 -2.20981359e-01 -5.59834182e-01 -7.61327147e-01
2.01467782e-01 -1.09943497e+00 9.49485123e-01 -6.08729124e-01
-1.22372448e+00 4.94149141e-02 -1.90007929e-02 -4.13664103e-01
-4.74938564e-02 -5.78229010e-01 -8.45931768e-01 8.64569783e-01
-1.37204599e+00 -8.52855504e-01 -2.99238533e-01 9.86310244e-01
7.74621487e-01 -2.09425583e-01 3.24522465e-01 1.03892706e-01
-6.40485466e-01 1.06717074e+00 -1.91299245e-02 -2.96860039e-01
7.22027183e-01 -1.58895719e+00 1.24566126e-04 7.82864988e-01
3.29831392e-02 5.12737095e-01 7.60441422e-01 -1.96039766e-01
-2.09579659e+00 -7.83511460e-01 7.95548737e-01 -8.59878600e-01
6.63224399e-01 -3.99013638e-01 -6.97511554e-01 7.34266222e-01
1.64044529e-01 -4.44827497e-01 2.58669406e-01 4.92231734e-02
-2.68564880e-01 -9.47023928e-02 -1.32216930e+00 7.68845558e-01
9.00912702e-01 -4.24728841e-01 -5.86245991e-02 1.13577461e+00
9.33749616e-01 -7.15745807e-01 -3.59409660e-01 -3.79869133e-01
1.58173397e-01 -6.59862161e-01 1.18285739e+00 -1.27867901e+00
2.91935295e-01 2.22668111e-01 -1.80366933e-01 -8.70306134e-01
1.10502094e-02 -1.74113142e+00 -3.51306766e-01 1.11582780e+00
3.94258320e-01 -5.92662394e-01 9.40166593e-01 1.09323239e+00
-8.78337100e-02 -9.54094410e-01 -1.09546113e+00 -7.68771112e-01
2.01397806e-01 -2.25791439e-01 3.30360532e-01 7.64203966e-01
-4.85808961e-02 3.82450908e-01 -9.31003273e-01 3.65239680e-01
5.22424579e-01 2.12994307e-01 1.06100404e+00 -5.34336507e-01
-6.49537683e-01 -4.11874443e-01 2.69129366e-01 -1.73416662e+00
5.33338189e-01 -8.09296429e-01 3.30133080e-01 -1.38712800e+00
4.30881292e-01 -4.99455452e-01 8.54531750e-02 4.64327067e-01
-5.84989667e-01 -3.60026628e-01 5.47198415e-01 -3.47882360e-02
-1.00307989e+00 1.92136824e-01 1.31549704e+00 -2.09957018e-01
-1.35702953e-01 2.87263244e-01 -1.28507507e+00 2.44402140e-01
8.12496126e-01 -6.37634337e-01 -4.13592577e-01 -4.49510217e-01
1.93767413e-01 6.38203204e-01 3.80305111e-01 1.08937614e-01
6.52326524e-01 -7.73656249e-01 -5.62119484e-01 -3.02076489e-01
5.01437128e-01 -6.44346714e-01 -6.02725804e-01 4.44000036e-01
-8.63516748e-01 2.02268213e-01 -3.38189453e-02 7.28019774e-01
1.84511110e-01 -6.82674766e-01 9.24541831e-01 1.26100093e-01
-5.31522274e-01 5.41505873e-01 -4.13567066e-01 6.81049228e-01
1.13063085e+00 4.66004200e-02 -8.88502419e-01 -1.22214901e+00
-1.99426919e-01 9.90062177e-01 -6.38371930e-02 4.44834024e-01
7.39796400e-01 -9.55375016e-01 -7.30247796e-01 -3.62229109e-01
-2.68473595e-01 -2.85647750e-01 2.27661088e-01 8.71835351e-01
-2.41529062e-01 5.63171506e-01 3.75997692e-01 -1.67696550e-01
-1.64775908e+00 8.18056822e-01 4.18781519e-01 -6.69297934e-01
-1.64848775e-01 1.11396670e+00 3.47557873e-01 -6.32771730e-01
7.38555193e-01 2.65043050e-01 2.07340226e-01 -1.83450729e-01
3.96316439e-01 6.65613353e-01 -4.12617058e-01 -2.30682820e-01
-3.94229084e-01 3.52151021e-02 -2.09831423e-03 -4.20772493e-01
1.13655162e+00 -1.67462900e-01 -1.12832949e-01 -1.99591935e-01
1.17793798e+00 -3.08988914e-02 -1.15030503e+00 -1.54605985e-01
-9.79168415e-02 -6.51419878e-01 -2.15412274e-01 -7.44627833e-01
-9.28127944e-01 7.76738524e-01 2.47062281e-01 4.98376161e-01
9.73032594e-01 1.81597456e-01 1.08137643e+00 9.03695524e-01
2.56362766e-01 -1.26388097e+00 5.42042971e-01 6.27066553e-01
9.62028801e-01 -1.40336001e+00 -1.14564367e-01 -3.90591562e-01
-8.21108937e-01 9.11769986e-01 7.96381950e-01 -2.32158564e-02
4.75376278e-01 4.83903885e-01 -2.57702377e-02 -3.86004686e-01
-1.01018631e+00 6.90557435e-03 2.58183144e-02 6.43586934e-01
-1.74156696e-01 -6.48167878e-02 -2.00577095e-01 6.06930971e-01
-1.02276511e-01 -6.07682705e-01 6.56433702e-01 1.02581251e+00
-6.77351475e-01 -1.04277396e+00 -5.80374599e-01 2.79535025e-01
-6.16122663e-01 2.89246030e-02 -1.04856515e+00 6.02956772e-01
-5.85057437e-01 1.47087991e+00 -2.80484647e-01 -2.08948210e-01
6.78322136e-01 -1.56369105e-01 5.03332436e-01 -4.88183469e-01
-1.05518460e+00 -3.38355750e-01 1.57005668e-01 -7.97279954e-01
1.71664447e-01 -4.42883611e-01 -1.01854873e+00 -6.13443911e-01
-5.79198837e-01 -1.28268182e-01 4.50442582e-01 7.52970278e-01
1.81697622e-01 4.55409475e-02 1.02962828e+00 -2.03106612e-01
-1.57788122e+00 -6.72149420e-01 -4.33017969e-01 4.24886525e-01
3.98385733e-01 -2.76716471e-01 -5.10904491e-01 -1.19344041e-01]
|
[6.453372478485107, 4.863454818725586]
|
4622f240-35b6-41f7-b630-bac7719e7130
|
dynamic-predictive-sampling-analog-to-digital
|
2211.09901
| null |
https://arxiv.org/abs/2211.09901v1
|
https://arxiv.org/pdf/2211.09901v1.pdf
|
Dynamic Predictive Sampling Analog to Digital Converter for Sparse Signal Sensing
|
This paper presents a dynamic predictive sampling (DPS) based analog-to-digital converter (ADC) that provides a non-uniform sampling of input analog continuous-time signals. The processing unit generates a dynamic prediction of the input signal using two prior-quantized samplings to compute digital values of an upper threshold and a lower threshold. The digital threshold values are converted to analog thresholds to form a tracking window. A comparator compares the input analog signal with the tracking window to determine if the prediction is successful. A counter records timestamps between the unsuccessful predictions, which are the selected sampling points for quantization. No quantization is performed for successfully predicted sampling points so that the data throughput and power can be saved. The proposed circuits were designed as a 10-bit ADC using 0.18 micro CMOS process sampling at 1 kHz. The results show that the proposed system can achieve a data compression factor of 6.17 and a power saving factor of 31% compared to a Nyquist rate SAR ADC for ECG monitoring.
|
['Wei Tang', 'Mario Renteria-Pinon', 'Xiaochen Tang']
|
2022-11-17
| null | null | null | null |
['data-compression']
|
['time-series']
|
[ 9.96649265e-01 -2.68117547e-01 -3.98477733e-01 -4.92068857e-01
-5.04544079e-01 -4.28217262e-01 -1.56648345e-02 8.24172080e-01
-5.75488985e-01 7.56623089e-01 -1.08511217e-01 -2.90200740e-01
1.30614802e-01 -6.98612690e-01 3.08035733e-03 -3.51306021e-01
-9.02249739e-02 5.73092066e-02 7.29031265e-01 3.08743060e-01
4.58495557e-01 5.62669754e-01 -1.15192723e+00 6.19623721e-01
6.49681151e-01 1.57254755e+00 5.88290840e-02 1.15270245e+00
3.34903598e-01 6.33784294e-01 -7.57103324e-01 2.85719603e-01
4.64166939e-01 -7.61399388e-01 -4.52328473e-01 -7.54489720e-01
7.30661601e-02 -9.63093519e-01 -2.10399076e-01 1.04184127e+00
4.35572326e-01 -2.01342896e-01 5.53562224e-01 -8.98000002e-01
9.10410564e-03 5.37516236e-01 -2.65812069e-01 5.09001315e-01
6.42147660e-01 -1.28611429e-02 3.44976306e-01 -3.45518976e-01
3.05355310e-01 9.46125448e-01 8.90641451e-01 -3.85001004e-02
-1.30921972e+00 -8.78025472e-01 -7.26238847e-01 3.72099787e-01
-1.43822038e+00 -2.46280938e-01 7.30453491e-01 -1.97576016e-01
9.84166086e-01 4.76496607e-01 1.10165954e+00 1.17158793e-01
1.12096596e+00 -1.73344642e-01 1.17824876e+00 -5.07790923e-01
6.82111979e-01 -2.96327353e-01 2.96930283e-01 2.01648161e-01
3.62890601e-01 -1.38593763e-01 -3.20204645e-01 -6.18523240e-01
1.22549009e+00 3.53755116e-01 -3.40767205e-01 5.40270030e-01
-1.15130556e+00 4.86940593e-01 3.87635499e-01 2.48888388e-01
-8.01229477e-01 3.94510329e-01 4.80195940e-01 3.54298443e-01
-5.67024052e-01 1.93809316e-01 -1.46511510e-01 -3.66373181e-01
-1.43442619e+00 -1.46535754e-01 6.96513236e-01 7.89144218e-01
1.58381268e-01 4.03935730e-01 -2.84681749e-02 -9.14583206e-02
3.84296834e-01 5.18811524e-01 9.85304415e-01 -8.53166580e-01
1.93779513e-01 5.09068787e-01 5.05080581e-01 -7.57356465e-01
-2.02736244e-01 1.95781097e-01 -8.84115815e-01 2.90673405e-01
4.49227780e-01 -1.49582967e-01 -7.44852364e-01 8.53303194e-01
-1.18291616e-01 4.88919728e-02 1.26025170e-01 8.98862481e-01
1.73407093e-01 1.03625226e+00 6.98754787e-02 -8.18676353e-01
1.57870972e+00 -1.18032232e-01 -1.27435243e+00 -2.14447021e-01
-3.93785149e-01 -9.72051203e-01 7.75779068e-01 6.61149442e-01
-1.20283520e+00 -9.28768337e-01 -1.77966523e+00 -9.47413892e-02
3.36869895e-01 4.91714329e-01 -4.60910201e-02 4.72217947e-01
-6.31041646e-01 1.01323056e+00 -1.22158659e+00 -3.67520973e-02
-6.39385879e-02 5.58562398e-01 4.25448000e-01 6.68876886e-01
-1.22536993e+00 5.12891114e-01 1.57139748e-01 -2.17553321e-02
-3.91050011e-01 -5.43418109e-01 -3.56709778e-01 1.26010090e-01
-5.54706752e-01 -2.17388198e-01 1.34505260e+00 -1.04723167e+00
-1.57569957e+00 2.85162419e-01 -3.90156358e-01 -9.65586007e-01
4.55240697e-01 8.53167772e-02 -7.30974376e-01 7.37681746e-01
-2.60094553e-01 -4.45958227e-02 9.45606112e-01 -1.31972477e-01
-4.71086323e-01 -1.89559862e-01 -8.08257878e-01 -9.74329859e-02
-1.17097534e-02 -2.55750835e-01 2.14172721e-01 -4.53362525e-01
6.66310549e-01 -5.32926857e-01 -2.77130276e-01 3.70037943e-01
4.63705435e-02 5.85606933e-01 6.28294826e-01 -5.73621929e-01
1.85218191e+00 -2.15292358e+00 -7.00875998e-01 4.57185119e-01
-3.55871230e-01 4.66407239e-01 7.14945495e-01 4.40079689e-01
2.77740151e-01 -4.14666384e-01 -2.07488924e-01 5.62714577e-01
-6.61321819e-01 -2.20966294e-01 -4.90175605e-01 5.22418082e-01
-2.81208809e-02 3.92915338e-01 -7.08159506e-01 -3.18441629e-01
1.62238672e-01 6.20471239e-01 -1.37473762e-01 1.08895689e-01
3.78345877e-01 8.23192373e-02 -3.35821331e-01 4.59570438e-01
4.96715367e-01 8.43165368e-02 7.05450058e-01 -8.06546152e-01
-3.32727551e-01 9.29847360e-01 -1.50010800e+00 9.79296446e-01
1.99554674e-02 7.16900051e-01 -7.19147176e-02 -2.00210169e-01
1.57078171e+00 5.67032516e-01 1.08195387e-01 -4.49490041e-01
5.86247921e-01 5.48879623e-01 2.29053617e-01 -2.62638450e-01
6.08001769e-01 -5.35283446e-01 -1.81624088e-02 8.39223936e-02
-4.58385885e-01 -3.50673079e-01 -2.56265193e-01 -1.17750563e-01
8.74858737e-01 -1.76686406e-01 1.01081049e+00 -3.36545199e-01
5.48538446e-01 1.10340647e-01 8.38828683e-01 2.69503236e-01
-5.02450049e-01 3.03629369e-01 3.29251349e-01 -5.16340971e-01
-1.16256547e+00 -1.21762156e+00 -2.48630479e-01 2.56939322e-01
2.60387540e-01 -3.47384572e-01 -4.56804246e-01 4.19533163e-01
-9.22392588e-03 4.96758342e-01 6.48641810e-02 -5.22904024e-02
-7.26669610e-01 1.11496001e-02 5.62160015e-01 7.14731812e-01
5.35159111e-01 -6.07689798e-01 -1.68669391e+00 8.94785821e-01
1.54640347e-01 -9.27386880e-01 -4.42919463e-01 2.84452766e-01
-1.66668606e+00 -7.47943401e-01 -2.35010624e-01 -9.67045248e-01
4.42811936e-01 -4.53176200e-01 3.15628439e-01 -2.82099932e-01
-4.95016634e-01 -2.81214774e-01 -2.02851817e-02 -5.56499541e-01
-5.33568561e-01 -6.32346034e-01 1.50173903e-01 -2.29779959e-01
5.62878668e-01 -6.73557401e-01 -1.25685859e+00 6.05775826e-02
-4.08809751e-01 -8.48021656e-02 6.21296406e-01 6.36779010e-01
9.94449496e-01 -1.88868150e-01 9.57745850e-01 -4.34139699e-01
7.35038996e-01 1.30090773e-01 -9.92730081e-01 -2.11664721e-01
-5.72110951e-01 -1.53205141e-01 1.42411816e+00 -7.81170368e-01
-5.33480763e-01 5.81041574e-01 -6.88924193e-02 -1.60027593e-01
3.34805250e-01 1.35741845e-01 3.29884022e-01 2.28460327e-01
8.56307089e-01 2.90210694e-01 4.41004694e-01 -1.72545180e-01
-1.66599885e-01 1.14516270e+00 1.15246117e+00 1.54987335e-01
-1.71574894e-02 1.84260815e-01 1.62513047e-01 -6.78725302e-01
4.26934808e-01 -3.42903525e-01 -4.92471308e-01 -7.69948736e-02
6.30006194e-01 -1.04485619e+00 -8.21402073e-01 1.12861231e-01
-8.97494018e-01 1.62445277e-01 -2.92482704e-01 8.72295737e-01
-4.79239136e-01 3.55821401e-01 -1.21264315e+00 -1.15109646e+00
-1.22207236e+00 -1.03732789e+00 3.54240179e-01 4.72010046e-01
-9.24137712e-01 -3.17702532e-01 -4.36706662e-01 -3.61023545e-01
4.22812849e-01 5.64050853e-01 7.11800933e-01 -2.11284712e-01
-4.77379739e-01 -6.24577045e-01 4.29776341e-01 3.89500186e-02
2.02071637e-01 1.45605981e-01 -7.11574614e-01 -4.11732018e-01
4.63993520e-01 1.13491654e-01 1.73110828e-01 3.10066521e-01
7.25697160e-01 -5.75830460e-01 -5.30199170e-01 2.67362654e-01
1.89691341e+00 1.05224574e+00 9.71385837e-01 -1.61877096e-01
-1.84197485e-01 -3.85790646e-01 8.61155152e-01 8.58198464e-01
-1.99798286e-01 1.11949921e-01 -2.86076050e-02 1.36842176e-01
1.78389788e-01 -3.72029305e-01 2.79277027e-01 8.83689582e-01
4.38373946e-02 2.03080475e-01 -6.72315717e-01 4.33265150e-01
-1.08196568e+00 -9.21790481e-01 -4.58073646e-01 2.69909477e+00
1.28300083e+00 3.82664442e-01 -5.18169701e-02 1.03483176e+00
6.88083410e-01 -3.06460738e-01 -6.96287394e-01 -1.12955856e+00
6.12829685e-01 7.60878980e-01 8.37254822e-01 4.33395714e-01
-5.73018312e-01 9.20130834e-02 5.80134344e+00 2.40606070e-01
-1.64073253e+00 -4.00244296e-01 3.89359206e-01 -7.74949789e-02
2.00504988e-01 -2.23230850e-02 -7.44008482e-01 6.40849948e-01
1.44940686e+00 -4.98774618e-01 9.86654013e-02 9.11818147e-01
4.91584092e-01 -5.40967226e-01 -1.09974909e+00 8.92203987e-01
-5.96300542e-01 -1.15448272e+00 1.84577301e-01 -2.31088966e-01
3.23781937e-01 -6.16750121e-01 -1.83563739e-01 -2.98097998e-01
-6.29640996e-01 -6.13620877e-01 7.25435019e-01 6.07692659e-01
1.48815954e+00 -7.70172119e-01 5.93220532e-01 1.96778551e-01
-1.57273126e+00 -3.46022338e-01 -6.21878803e-01 -5.55538893e-01
3.37541223e-01 6.14532173e-01 -1.43745804e+00 4.06507626e-02
3.12355757e-01 -2.53572389e-02 1.59580678e-01 9.71953452e-01
1.50537968e-01 8.89015734e-01 -6.27060533e-01 -3.00775141e-01
-1.62812069e-01 -2.17418879e-01 2.88566202e-01 1.18421912e+00
5.06915271e-01 8.38535726e-01 -1.90101579e-01 5.47392070e-01
2.41012305e-01 -5.23717627e-02 1.21623822e-01 1.10174254e-01
1.37880719e+00 7.70994425e-01 -5.47417045e-01 -8.50466132e-01
-1.27433255e-01 9.12986457e-01 -6.19458973e-01 -3.45098376e-01
-7.68017471e-01 -1.38929641e+00 1.72020167e-01 5.34590900e-01
4.80854474e-02 -3.40905964e-01 -6.75173461e-01 -1.54836327e-01
-1.49647263e-03 -6.13791347e-01 4.69803303e-01 -5.03012002e-01
-6.58980489e-01 5.65529704e-01 -3.58938724e-01 -1.74572265e+00
-5.01005292e-01 -2.72238324e-03 -6.69907212e-01 1.36391544e+00
-1.04860568e+00 4.22814824e-02 -4.99734074e-01 5.20342767e-01
2.10786834e-01 3.84406954e-01 1.00519753e+00 -3.20453271e-02
3.95913064e-01 5.24333477e-01 9.20368880e-02 9.46417898e-02
5.72757781e-01 -8.61031413e-01 3.42959881e-01 8.34324658e-01
-8.43208790e-01 8.11572015e-01 7.18219936e-01 -1.01190817e+00
-1.60354257e+00 -8.52934897e-01 1.24036491e+00 6.11724615e-01
1.73268646e-01 -6.62654638e-02 -1.13326228e+00 4.05366510e-01
-9.39902067e-02 5.57181519e-03 6.40947402e-01 -1.24423909e+00
1.79427162e-01 -6.71568513e-01 -1.80233705e+00 3.45164001e-01
-9.26645845e-02 -4.58457530e-01 -9.25570488e-01 -5.09421527e-01
3.30386996e-01 -5.89877725e-01 -1.36104989e+00 2.33448818e-01
1.01077795e+00 -7.14350939e-01 7.15957820e-01 4.54594523e-01
1.33112529e-02 -8.61324728e-01 -1.00214399e-01 -5.39894700e-01
-3.41686904e-01 -8.69535744e-01 2.43860438e-01 7.68026352e-01
2.80230224e-01 -5.94624937e-01 5.13258815e-01 7.27600336e-01
3.42036873e-01 -8.55228662e-01 -1.09031034e+00 -5.05905330e-01
-6.02104425e-01 1.38418883e-01 4.47382450e-01 4.55848455e-01
7.97735512e-01 1.31088868e-01 -1.15257442e-01 2.43186817e-01
4.92439300e-01 4.12099212e-02 -1.97353244e-01 -1.02084458e+00
-2.12004930e-01 6.10518456e-02 -8.83120120e-01 -1.05408156e+00
-1.28026009e+00 -3.63701016e-01 -1.95223063e-01 -1.41346788e+00
-5.45198441e-01 -1.10087171e-01 -2.79935151e-01 2.49817789e-01
3.08048464e-02 3.95819247e-01 -5.79126040e-03 2.90948808e-01
2.35983744e-01 -1.81740358e-01 8.20497811e-01 2.81742096e-01
-8.46387684e-01 5.38186915e-02 1.88634813e-01 5.81174970e-01
9.30464029e-01 -1.64558366e-01 -4.88447577e-01 3.27559859e-01
-2.81905442e-01 9.35531437e-01 1.44375907e-02 -1.66085553e+00
3.19988281e-01 1.86543856e-02 1.21202815e+00 -9.39044535e-01
3.40993226e-01 -9.90866184e-01 7.52866924e-01 1.38766468e+00
-3.87436599e-01 4.30413455e-01 1.50185257e-01 3.82596612e-01
-1.58880316e-02 -9.50364023e-02 1.26040065e+00 2.16409624e-01
-3.66488695e-01 -3.08885187e-01 -7.97918499e-01 -4.53264147e-01
9.24791038e-01 -8.43336105e-01 -2.74376683e-02 -1.94950730e-01
-7.50851095e-01 -1.71437442e-01 3.75169218e-01 -2.73267508e-01
1.05069923e+00 -1.28143406e+00 -1.20985918e-01 6.53736889e-01
-4.38547343e-01 -2.32887194e-01 1.32883891e-01 5.94434261e-01
-1.15837216e+00 3.35744947e-01 -6.40981674e-01 -6.32118464e-01
-1.60576355e+00 1.30536094e-01 4.05320495e-01 8.10549632e-02
-8.02175403e-01 3.11526358e-01 -1.25252104e+00 1.22498870e+00
7.65184164e-02 -1.10072446e+00 4.49721748e-03 -3.28430124e-02
1.11293066e+00 7.86552489e-01 -1.03613429e-01 7.00241327e-02
-5.52621603e-01 6.29660606e-01 4.56961453e-01 -4.95289415e-01
8.83272350e-01 1.73582342e-02 8.77544507e-02 7.53693819e-01
1.16233230e+00 -1.62486225e-01 -1.12261832e+00 8.67371485e-02
-1.76827669e-01 -4.38977838e-01 -2.16802090e-01 -9.68081474e-01
-3.75751376e-01 7.47866929e-01 9.82230961e-01 2.24827211e-02
1.70709383e+00 -9.44425046e-01 1.17696381e+00 -3.86221893e-02
5.80934167e-01 -1.10072446e+00 -3.42859864e-01 5.26853604e-03
5.69846511e-01 -5.77234663e-02 6.03047907e-01 -1.22914381e-01
-4.99785483e-01 1.77627504e+00 8.63320604e-02 -6.09065056e-01
4.12269980e-01 6.44803047e-01 3.03966813e-02 6.32468045e-01
-5.86196244e-01 5.83394349e-01 -1.09063685e-01 4.63907123e-01
5.46778202e-01 2.80107319e-01 -1.08040416e+00 8.07182074e-01
-3.83770555e-01 7.84153223e-01 1.03632355e+00 1.20992780e+00
-9.24538195e-01 -6.12853944e-01 -6.79758489e-01 7.17396975e-01
-8.04489672e-01 4.46588546e-02 9.39776003e-03 1.51586145e-01
-1.38370350e-01 8.78013015e-01 6.40274525e-01 -4.07464892e-01
3.09902102e-01 3.03679675e-01 4.57893282e-01 -3.77504863e-02
-9.02780652e-01 4.89667565e-01 -1.77970052e-01 -5.31614482e-01
2.95753092e-01 -3.44587803e-01 -2.09740829e+00 4.37186025e-02
-1.06559731e-01 1.58153757e-01 8.59385848e-01 1.20025702e-01
3.66985977e-01 6.56217396e-01 5.24085104e-01 -2.46036723e-01
-9.02477443e-01 -8.64151478e-01 -5.19574821e-01 -1.43920124e-01
6.51772976e-01 2.59855837e-01 -2.91046321e-01 3.37578297e-01]
|
[13.944103240966797, 3.157130241394043]
|
a48ba6e0-5deb-4607-999d-08a028154cb3
|
bert-got-a-date-introducing-transformers-to
|
2109.14927
| null |
https://arxiv.org/abs/2109.14927v3
|
https://arxiv.org/pdf/2109.14927v3.pdf
|
BERT got a Date: Introducing Transformers to Temporal Tagging
|
Temporal expressions in text play a significant role in language understanding and correctly identifying them is fundamental to various retrieval and natural language processing systems. Previous works have slowly shifted from rule-based to neural architectures, capable of tagging expressions with higher accuracy. However, neural models can not yet distinguish between different expression types at the same level as their rule-based counterparts. In this work, we aim to identify the most suitable transformer architecture for joint temporal tagging and type classification, as well as, investigating the effect of semi-supervised training on the performance of these systems. Based on our study of token classification variants and encoder-decoder architectures, we present a transformer encoder-decoder model using the RoBERTa language model as our best performing system. By supplementing training resources with weakly labeled data from rule-based systems, our model surpasses previous works in temporal tagging and type classification, especially on rare classes. Our code and pre-trained experiments are available at: https://github.com/satya77/Transformer_Temporal_Tagger
|
['Michael Gertz', 'Dennis Aumiller', 'Satya Almasian']
|
2021-09-30
| null | null | null | null |
['temporal-tagging']
|
['natural-language-processing']
|
[ 1.78024173e-02 -7.48073608e-02 -6.47748947e-01 -5.34800470e-01
-7.74477720e-01 -8.49717200e-01 7.92940140e-01 2.63369262e-01
-6.19950712e-01 7.15098500e-01 2.14538679e-01 -3.91940296e-01
6.88094348e-02 -6.46962702e-01 -4.55841422e-01 -4.50470775e-01
-2.84145385e-01 5.14297903e-01 4.44871873e-01 -2.74167895e-01
-1.40769348e-01 8.64288583e-02 -1.46490014e+00 7.84856975e-01
4.31537032e-01 1.40180981e+00 -1.61239550e-01 5.72163463e-01
-4.94435221e-01 1.38047564e+00 -3.37096721e-01 -5.92370510e-01
-1.93098143e-01 -2.63787776e-01 -1.16525567e+00 -5.20561159e-01
1.06904164e-01 8.01789686e-02 -2.98011392e-01 7.06122518e-01
3.02412868e-01 7.10506886e-02 5.24159372e-01 -1.00130105e+00
-7.10568666e-01 9.71209645e-01 1.81510560e-02 3.80233616e-01
3.30507338e-01 -3.53375554e-01 1.20096409e+00 -7.57951617e-01
7.23032176e-01 1.01300931e+00 9.33135986e-01 6.91138923e-01
-1.01034689e+00 -5.70354104e-01 2.07265526e-01 4.56515044e-01
-1.50343013e+00 -7.24705577e-01 5.47133446e-01 -4.65643883e-01
1.56275952e+00 -1.75174687e-03 6.79893732e-01 1.32278037e+00
2.23967373e-01 1.09290779e+00 1.15900934e+00 -5.55078030e-01
7.33882636e-02 -1.97882342e-04 2.15586439e-01 8.18391025e-01
-3.84237528e-01 1.55822858e-01 -8.03893626e-01 -2.13044100e-02
4.24046040e-01 -3.05188477e-01 4.49301861e-02 -1.05612211e-01
-1.23784471e+00 5.17368317e-01 1.69504493e-01 1.06070912e+00
-3.23771834e-01 3.13651681e-01 7.23239839e-01 4.50750411e-01
7.69132614e-01 2.57355332e-01 -1.02319014e+00 -5.86863935e-01
-9.50298369e-01 -1.12230569e-01 7.16215312e-01 9.20234084e-01
3.45623642e-01 1.16588995e-02 -4.92916882e-01 1.07238257e+00
1.36686951e-01 2.36402974e-01 7.74548888e-01 -7.15950489e-01
7.35715404e-02 6.30878925e-01 -8.46943632e-02 -2.28407174e-01
-3.63167971e-01 -3.57371062e-01 -4.47509974e-01 -2.65860766e-01
6.64464891e-01 -1.34116590e-01 -9.85430002e-01 1.80030739e+00
-3.65062468e-02 3.09359971e-02 -1.66672152e-02 4.35817599e-01
5.26157796e-01 5.64280152e-01 5.08702338e-01 -2.17294335e-01
1.67676854e+00 -6.58815086e-01 -8.64622295e-01 -1.61280006e-01
1.28283393e+00 -7.88606524e-01 6.28308892e-01 3.60908896e-01
-1.03337514e+00 -1.46655187e-01 -6.51076198e-01 -2.28408739e-01
-7.59072125e-01 2.92736501e-01 8.18093240e-01 4.35089439e-01
-1.20870435e+00 6.49570704e-01 -1.14925313e+00 -6.71372235e-01
3.00960511e-01 3.49529386e-01 -4.30173755e-01 3.12141538e-01
-1.72367120e+00 1.29026914e+00 4.12056863e-01 2.20723882e-01
-6.75915480e-01 -5.88529706e-01 -8.82580519e-01 -2.50690341e-01
1.66807532e-01 -3.77024323e-01 1.96271026e+00 -1.18048108e+00
-1.44697273e+00 1.14629090e+00 -6.43873751e-01 -7.98181951e-01
8.32622051e-02 -1.37674049e-01 -6.21814251e-01 -1.81381226e-01
-6.08840841e-04 5.43328404e-01 3.45423698e-01 -5.66527843e-01
-6.43401146e-01 -1.09457493e-01 2.47972552e-02 -1.57635659e-01
-3.56433272e-01 3.97249013e-01 -3.50456029e-01 -5.45478523e-01
-2.17356756e-01 -9.17545855e-01 -2.01774351e-02 -1.97638586e-01
2.06054132e-02 -7.41475761e-01 5.14903665e-01 -5.06156683e-01
1.39389026e+00 -2.10666513e+00 -2.48058692e-01 -1.41645908e-01
-1.13672778e-01 3.72057945e-01 -5.28177135e-02 5.18770754e-01
-2.50951618e-01 1.74232543e-01 -1.27533659e-01 -3.75111520e-01
1.74806163e-01 4.79543626e-01 -3.40504229e-01 2.86709815e-01
2.76833624e-01 9.76443172e-01 -1.06825507e+00 -6.14771426e-01
-2.23373026e-02 5.11162460e-01 -2.12825209e-01 1.56779960e-01
-4.69999194e-01 2.35664263e-01 -5.10515094e-01 7.10513413e-01
9.75727960e-02 -1.08104788e-01 5.55311859e-01 -1.32518783e-01
-3.33630234e-01 1.12010241e+00 -4.16637450e-01 1.75389731e+00
-9.09890294e-01 7.90485322e-01 -2.67294824e-01 -1.16335022e+00
8.35522115e-01 8.65463197e-01 7.45738745e-01 -1.13155425e+00
1.80847391e-01 3.87576133e-01 -1.68158039e-02 -5.18424094e-01
5.39199769e-01 -4.52993661e-01 -3.25979769e-01 4.21969026e-01
3.10877025e-01 3.52862328e-01 5.09841621e-01 -1.96486246e-02
1.25019681e+00 4.25600171e-01 3.75616729e-01 -1.59068212e-01
2.68180668e-01 2.05195546e-01 7.99208224e-01 4.50290561e-01
-2.77732223e-01 1.80139482e-01 5.23510873e-01 -6.61184728e-01
-7.92469919e-01 -9.18432415e-01 -3.33311319e-01 1.42885995e+00
-3.92420411e-01 -6.35422409e-01 -3.41279656e-01 -7.22847581e-01
-2.84852237e-01 8.30909193e-01 -7.74374664e-01 -1.24482431e-01
-5.44902503e-01 -6.34597361e-01 1.18585253e+00 7.61313498e-01
3.83919001e-01 -1.18609035e+00 -3.53237867e-01 3.32230508e-01
-4.80493456e-01 -1.24271190e+00 -1.58649534e-01 7.60450244e-01
-7.57487357e-01 -8.89548957e-01 -4.74015594e-01 -8.05359483e-01
1.58234060e-01 -3.86132777e-01 1.42294061e+00 -1.17488774e-02
1.85097784e-01 1.26367494e-01 -7.11246431e-01 -3.36089402e-01
-3.64738971e-01 3.77959639e-01 2.01032702e-02 -6.01832420e-02
8.33239377e-01 -5.23946643e-01 -2.05735981e-01 2.09111005e-01
-6.29374981e-01 -1.73908576e-01 7.28536904e-01 6.51499689e-01
2.99160779e-01 -1.10727809e-01 1.56781197e-01 -1.00875926e+00
3.43185425e-01 -4.96676534e-01 -4.80738848e-01 4.00756687e-01
-6.99632406e-01 4.77613330e-01 4.66461480e-01 -4.65954781e-01
-1.23928404e+00 1.11179911e-01 -5.28200507e-01 -1.71961099e-01
-2.34235451e-01 7.44349599e-01 2.50135839e-01 2.51176953e-01
6.02528930e-01 3.37997139e-01 -3.62363845e-01 -6.29829526e-01
2.58154124e-01 4.16146874e-01 3.36792290e-01 -7.19195187e-01
3.05512697e-01 2.19866022e-01 -2.60479063e-01 -5.49739659e-01
-1.22168422e+00 -6.60298407e-01 -8.94423485e-01 -1.04461797e-01
8.83834898e-01 -8.94308209e-01 -4.83853817e-01 6.16097689e-01
-1.24895108e+00 -7.56413460e-01 -1.03701055e-01 3.76073390e-01
-5.43511450e-01 6.00213818e-02 -9.23940241e-01 -8.71068418e-01
-8.27824399e-02 -6.59176946e-01 1.14103389e+00 -2.17398763e-01
-5.70727170e-01 -1.31353617e+00 4.00493741e-01 1.23907328e-01
4.87286031e-01 -8.77148211e-02 8.66956115e-01 -8.10931444e-01
-2.91168511e-01 -1.05630800e-01 1.71703786e-01 1.35274837e-02
-1.09252878e-01 -1.35027850e-02 -1.09307873e+00 1.90942019e-01
-3.40830982e-01 -5.16467333e-01 1.00751531e+00 2.42697820e-01
7.60286212e-01 -3.95007521e-01 -5.60902596e-01 4.71997231e-01
1.10918367e+00 4.21817571e-01 6.65314496e-01 3.24841142e-01
3.80820692e-01 5.49610794e-01 7.25924551e-01 2.24843264e-01
5.25510788e-01 9.67268109e-01 -2.94631869e-02 1.16335765e-01
2.04831641e-02 -2.64732927e-01 7.63957024e-01 8.12308371e-01
-8.05676952e-02 -3.67135227e-01 -1.18674839e+00 8.67337644e-01
-1.84261405e+00 -1.14453042e+00 -4.68717813e-02 1.90865958e+00
1.28686118e+00 1.54936329e-01 1.37440458e-01 9.82297435e-02
3.58981639e-01 2.21293122e-01 -3.41833793e-02 -6.39913559e-01
-4.66805287e-02 4.48241740e-01 5.26863039e-01 3.61545354e-01
-1.24577761e+00 1.32541108e+00 6.49415350e+00 8.68550837e-01
-1.41273999e+00 3.61180007e-01 2.85709351e-01 -8.35810229e-02
-2.34315678e-01 1.65391058e-01 -8.09721589e-01 3.21578413e-01
1.54623365e+00 -7.95056820e-02 2.19275057e-01 5.32488883e-01
1.70086354e-01 -4.71026357e-03 -1.21655798e+00 6.69321656e-01
-1.85253799e-01 -1.24827182e+00 -2.49093473e-01 -1.92164347e-01
3.25676650e-01 3.09169620e-01 -9.68300253e-02 6.46584451e-01
6.40509009e-01 -8.99947345e-01 1.02124441e+00 4.85546321e-01
6.49866879e-01 -1.62460729e-01 8.66680145e-01 4.42515276e-02
-1.49040258e+00 6.16845600e-02 1.08183756e-01 -1.93093076e-01
1.22750171e-01 7.22687900e-01 -9.12077069e-01 5.09319842e-01
6.97931349e-01 1.05721796e+00 -4.25362825e-01 6.41633689e-01
-4.55319673e-01 9.03644264e-01 -2.97128379e-01 -2.68783152e-01
2.20084950e-01 2.84113795e-01 1.79204106e-01 1.58733523e+00
3.75912100e-01 2.17442855e-01 4.80431644e-03 5.11341751e-01
5.96632715e-03 9.76951197e-02 -7.22742021e-01 -4.65488315e-01
5.77459097e-01 1.04056692e+00 -6.17971539e-01 -4.53902394e-01
-5.44102907e-01 7.41902113e-01 5.58833778e-01 1.86256886e-01
-8.67911518e-01 -3.48081663e-02 6.82909369e-01 1.79508373e-01
3.73381704e-01 -5.44831336e-01 -1.31261824e-02 -1.32326448e+00
-2.90214363e-02 -6.16415322e-01 7.64872491e-01 -7.77622342e-01
-1.07458818e+00 8.02130699e-01 6.57037124e-02 -1.18068457e+00
-6.87327743e-01 -8.31646681e-01 -3.05753052e-01 6.72944605e-01
-1.41888356e+00 -1.38547230e+00 2.89067447e-01 5.11275649e-01
2.89319307e-01 8.39467496e-02 1.09747970e+00 6.17614746e-01
-4.51843619e-01 7.08254576e-01 -6.69113249e-02 5.92228830e-01
7.98205853e-01 -1.19008482e+00 2.00267702e-01 9.61775362e-01
4.00469273e-01 7.58744955e-01 5.44729054e-01 -2.88846523e-01
-1.15043378e+00 -1.20606029e+00 1.79071641e+00 -6.45022333e-01
1.02756810e+00 -4.64160144e-01 -8.53118539e-01 9.76974010e-01
3.04188639e-01 -1.90370139e-02 6.73799038e-01 7.47633815e-01
-6.99030697e-01 -5.10893092e-02 -6.39560938e-01 3.22495759e-01
1.27726710e+00 -9.72046196e-01 -5.72540760e-01 2.30716720e-01
3.41465801e-01 -3.07969868e-01 -1.12371039e+00 3.01519662e-01
7.44063675e-01 -8.12626064e-01 5.94423890e-01 -5.83457351e-01
3.96958083e-01 -2.43232697e-01 -2.09721811e-02 -1.13074660e+00
-3.60895455e-01 -3.37051451e-01 -3.66556905e-02 1.36954939e+00
7.01464295e-01 -5.67604542e-01 6.10822797e-01 4.44589257e-01
-1.82251051e-01 -7.45472133e-01 -1.04269028e+00 -1.05806184e+00
2.31937528e-01 -8.01670909e-01 2.06847474e-01 1.13875985e+00
4.03647721e-01 3.92673641e-01 -3.11629027e-01 -9.79417562e-02
1.10111542e-01 1.24468312e-01 7.74356797e-02 -8.89764667e-01
-1.44752264e-01 -6.62698388e-01 -4.30720568e-01 -9.36111033e-01
6.16538167e-01 -1.22821403e+00 2.33652562e-01 -1.32449591e+00
-8.29711854e-02 -6.24771535e-01 -4.51249689e-01 1.39779246e+00
2.85237283e-01 4.02219772e-01 -2.73074508e-01 1.28923878e-01
-8.00952613e-01 4.79817599e-01 5.23875058e-01 -2.86469847e-01
1.79637298e-01 -2.48028547e-01 -3.50526303e-01 6.10051930e-01
7.62682021e-01 -7.22915590e-01 -1.73682183e-01 -6.13622844e-01
5.34211814e-01 3.79172303e-02 1.19448751e-01 -7.02275217e-01
2.47563750e-01 -1.04154080e-01 3.67224813e-02 -2.78995484e-01
3.85590672e-01 -5.44463694e-01 1.40459538e-01 4.49351609e-01
-5.72834015e-01 2.34093398e-01 3.65907580e-01 3.93179357e-02
-4.49609488e-01 -2.26136178e-01 6.04331911e-01 -1.72642276e-01
-1.31049275e+00 2.44283527e-01 -8.94496262e-01 1.13413572e-01
6.31969571e-01 6.74857423e-02 -3.13707918e-01 -2.33647466e-01
-8.45250547e-01 9.39520821e-02 2.75106579e-01 6.37219131e-01
2.27131382e-01 -1.33534145e+00 -4.84399080e-01 -1.49382263e-01
4.24203992e-01 -4.85924661e-01 -2.74271816e-02 1.20784724e+00
-1.62429392e-01 8.37306440e-01 -1.53642774e-01 -4.14624542e-01
-1.26072049e+00 5.05053401e-01 6.13237917e-01 -6.32133424e-01
-1.36401683e-01 8.83344054e-01 -1.59096092e-01 -3.47806007e-01
4.65921424e-02 -5.41952550e-01 -2.32516721e-01 2.23897383e-01
2.44479254e-01 -2.78641105e-01 3.56900305e-01 -6.08454704e-01
-7.80048251e-01 3.68864655e-01 -7.62534514e-02 -2.08074287e-01
1.20754075e+00 2.09900737e-01 -2.18860567e-01 8.38728547e-01
1.24328339e+00 -1.51520550e-01 -6.24502540e-01 -5.05229473e-01
4.55677271e-01 7.86293149e-02 1.09678529e-01 -7.75025725e-01
-9.39272761e-01 7.45764792e-01 4.10046548e-01 2.34574884e-01
1.11390257e+00 1.92718282e-01 7.63117015e-01 3.92780572e-01
4.77978826e-01 -9.97723758e-01 -3.24888438e-01 1.10250247e+00
4.03842866e-01 -1.00863051e+00 -4.37577516e-01 -1.61025599e-01
-4.84591842e-01 8.95893693e-01 3.51811707e-01 2.28872761e-01
7.47418165e-01 6.60550892e-01 4.20715362e-01 -2.24613816e-01
-1.56638110e+00 -5.29068053e-01 2.79957026e-01 4.90925968e-01
1.47940922e+00 5.18005565e-02 -3.70985091e-01 4.58289981e-01
-2.30716169e-01 3.70804578e-01 4.53272462e-02 1.03472435e+00
-6.25001788e-02 -1.67157364e+00 4.15724851e-02 3.54928464e-01
-8.84951770e-01 -4.77343082e-01 -5.23092449e-01 6.00916803e-01
1.59447476e-01 8.19438457e-01 2.24459574e-01 -5.05224466e-01
2.00037375e-01 6.55591786e-01 6.74287498e-01 -6.17718041e-01
-7.45216608e-01 -2.88994193e-01 8.51586819e-01 -5.36316514e-01
-7.53237963e-01 -9.35160100e-01 -1.19231665e+00 -1.79490119e-01
-2.61163078e-02 4.76088583e-01 2.31506497e-01 1.29144645e+00
2.45001227e-01 4.31563675e-01 1.21155038e-01 -2.84900069e-01
-2.25595847e-01 -1.12451601e+00 -3.52164477e-01 4.34823245e-01
8.69454816e-02 -7.45555580e-01 2.37938352e-02 4.28278565e-01]
|
[9.955606460571289, 9.28342342376709]
|
a7e62bd8-9441-4bb8-8cf9-dd2b83557764
|
combing-policy-evaluation-and-policy
|
2109.11867
| null |
https://arxiv.org/abs/2109.11867v2
|
https://arxiv.org/pdf/2109.11867v2.pdf
|
The $f$-Divergence Reinforcement Learning Framework
|
The framework of deep reinforcement learning (DRL) provides a powerful and widely applicable mathematical formalization for sequential decision-making. This paper present a novel DRL framework, termed \emph{$f$-Divergence Reinforcement Learning (FRL)}. In FRL, the policy evaluation and policy improvement phases are simultaneously performed by minimizing the $f$-divergence between the learning policy and sampling policy, which is distinct from conventional DRL algorithms that aim to maximize the expected cumulative rewards. We theoretically prove that minimizing such $f$-divergence can make the learning policy converge to the optimal policy. Besides, we convert the process of training agents in FRL framework to a saddle-point optimization problem with a specific $f$ function through Fenchel conjugate, which forms new methods for policy evaluation and policy improvement. Through mathematical proofs and empirical evaluation, we demonstrate that the FRL framework has two advantages: (1) policy evaluation and policy improvement processes are performed simultaneously and (2) the issues of overestimating value function are naturally alleviated. To evaluate the effectiveness of the FRL framework, we conduct experiments on Atari 2600 video games and show that agents trained in the FRL framework match or surpass the baseline DRL algorithms.
|
['Xianjie Zhang', 'Xinwen Hou', 'Xiaoyu Chen', 'Zhou Yang', 'Guoliang Fan', 'Yu Liu', 'Yunpeng Bai', 'Qiang He', 'Chen Gong']
|
2021-09-24
| null | null | null | null |
['mathematical-proofs']
|
['miscellaneous']
|
[-3.09908390e-01 -4.78659049e-02 -4.47408170e-01 -7.43443817e-02
-7.31634080e-01 -5.75057328e-01 4.91091371e-01 7.27930441e-02
-9.56502914e-01 1.24993408e+00 -1.52140006e-01 -6.11097991e-01
-4.84619260e-01 -6.18023574e-01 -9.29937005e-01 -9.05249834e-01
-3.16288769e-01 1.55201584e-01 -2.32608095e-02 -2.24102139e-01
3.31271917e-01 2.82353550e-01 -1.20115995e+00 -4.09200668e-01
1.11384082e+00 1.19156396e+00 3.27035785e-01 5.85502028e-01
1.60003960e-01 1.15772486e+00 -6.19802952e-01 -1.63240567e-01
6.61888480e-01 -4.38995242e-01 -7.16320455e-01 -6.28342107e-02
-5.30316979e-02 -8.41161132e-01 -3.57824266e-01 1.41865683e+00
5.17571628e-01 4.58124846e-01 4.52544242e-01 -1.43121326e+00
-4.16172177e-01 5.30402660e-01 -7.35653579e-01 9.18063745e-02
6.57769442e-02 2.95040160e-01 1.06971061e+00 -4.71441120e-01
1.51743665e-01 1.48818696e+00 4.04724836e-01 6.49041057e-01
-9.68699694e-01 -5.01843929e-01 5.95228374e-01 -5.76532707e-02
-9.09203172e-01 -6.47912594e-03 4.56571370e-01 -3.85208130e-01
7.80670702e-01 -2.21654847e-01 6.25986874e-01 6.00075066e-01
3.05834085e-01 1.16303611e+00 1.29024911e+00 -2.94005096e-01
4.91461843e-01 -1.30283222e-01 -2.46772259e-01 8.10636580e-01
1.84740737e-01 8.19557250e-01 2.11133119e-02 -1.60786346e-01
1.03026032e+00 1.38956821e-02 -4.67194021e-02 -5.29334128e-01
-1.04531050e+00 1.06686342e+00 2.23733827e-01 -1.55516163e-01
-6.81489170e-01 5.46813726e-01 3.45666766e-01 5.91257274e-01
1.57573700e-01 3.92869979e-01 -2.70081758e-01 -3.02181393e-01
-4.32868332e-01 7.97367275e-01 5.72681427e-01 6.55078292e-01
5.72018683e-01 4.42005128e-01 -3.89646828e-01 6.32274449e-01
5.72339654e-01 8.27843189e-01 3.20758969e-01 -1.67306674e+00
4.92071420e-01 1.60207912e-01 7.17272997e-01 -7.38423228e-01
-4.64213938e-02 -4.29500997e-01 -5.89835465e-01 7.99259245e-01
4.33662325e-01 -7.64959812e-01 -2.82382250e-01 2.10005546e+00
1.84484318e-01 6.43787906e-02 3.51271778e-01 8.06514263e-01
-1.00182310e-01 7.99634933e-01 1.94200799e-02 -7.11358786e-01
5.77598274e-01 -8.61410320e-01 -7.40768552e-01 3.57071832e-02
5.27607381e-01 -3.42954665e-01 1.14452171e+00 3.62420857e-01
-1.39406347e+00 -3.76588076e-01 -9.17447507e-01 5.87648630e-01
2.00136438e-01 -1.02429487e-01 5.56367397e-01 4.32067156e-01
-1.17842913e+00 8.23459446e-01 -7.44546592e-01 1.93793684e-01
4.40040052e-01 4.65305477e-01 2.59805262e-01 3.55096161e-01
-1.20827091e+00 8.84058118e-01 5.54259837e-01 -2.34944448e-02
-1.40427876e+00 -5.94482899e-01 -6.79594040e-01 4.00742330e-03
8.85384977e-01 -3.80240351e-01 1.97208393e+00 -1.05104780e+00
-2.08006573e+00 2.39359826e-01 2.00810522e-01 -7.77262688e-01
7.43427873e-01 -4.05757427e-01 4.76190671e-02 -3.62606049e-02
2.09761202e-01 4.75155056e-01 8.48463178e-01 -1.11787951e+00
-1.15652204e+00 -1.10855132e-01 5.41531682e-01 5.32619715e-01
-1.57074165e-02 -2.17677429e-01 2.41664439e-01 -6.16828859e-01
-7.78741837e-01 -6.45459592e-01 -5.54680347e-01 -2.15530261e-01
1.77705675e-01 -5.18983364e-01 6.43154442e-01 -3.54289263e-01
1.31033826e+00 -1.86442959e+00 4.50605489e-02 3.08226675e-01
5.92912585e-02 5.15768945e-01 -1.85473263e-01 3.35230052e-01
2.48027042e-01 6.95129018e-03 -2.43230775e-01 5.55522181e-02
3.72246414e-01 4.16746318e-01 -5.70226550e-01 4.73629326e-01
-8.47095922e-02 8.77925694e-01 -1.18577886e+00 -2.29112118e-01
1.61527559e-01 -1.26278818e-01 -8.32306981e-01 3.23085636e-01
-5.66271901e-01 3.62387210e-01 -7.62614489e-01 2.91756302e-01
3.72788042e-01 8.52712691e-02 1.16664805e-01 4.11831826e-01
-4.29677844e-01 7.19414949e-02 -1.21184540e+00 1.33475578e+00
-3.51169199e-01 1.16939552e-01 2.74925262e-01 -1.44717157e+00
8.12464118e-01 2.25649461e-01 8.17257226e-01 -8.55771124e-01
2.05515429e-01 1.20889299e-01 -1.65503457e-01 -2.92364806e-01
2.08613738e-01 -2.73913503e-01 -8.00896809e-02 6.76115274e-01
-5.39398417e-02 5.43520115e-02 4.20813620e-01 -2.58565005e-02
7.47484565e-01 4.02534544e-01 3.48156869e-01 -6.15429878e-01
5.42302489e-01 -2.06597641e-01 8.16959798e-01 1.08656871e+00
-6.24414682e-01 -5.97427905e-01 8.03401768e-01 -3.75329852e-01
-9.45101678e-01 -1.20665061e+00 2.21492484e-01 1.05743670e+00
2.72203296e-01 -9.07119885e-02 -7.09249914e-01 -9.48235571e-01
3.46113712e-01 7.66009986e-01 -5.21773398e-01 -1.13792583e-01
-5.95343530e-01 -5.98802447e-01 2.61614144e-01 4.37712997e-01
9.31119859e-01 -1.06993449e+00 -7.93582618e-01 3.91771585e-01
7.10795894e-02 -6.58080161e-01 -7.28170693e-01 -7.75094852e-02
-8.20728838e-01 -1.02767742e+00 -7.64539599e-01 -6.04935646e-01
4.09759969e-01 1.66838378e-01 8.94239485e-01 -2.15582430e-01
2.60513723e-01 6.49027050e-01 -1.39257252e-01 -4.29202080e-01
-4.16697085e-01 -3.25941473e-01 4.21243191e-01 -9.04504359e-02
1.85488798e-02 -3.98104787e-01 -7.66804338e-01 3.21238607e-01
-5.90055525e-01 -2.67032743e-01 4.43685949e-01 8.94741118e-01
6.96729302e-01 2.54865527e-01 9.41657543e-01 -3.68512064e-01
1.24294400e+00 -2.32139617e-01 -1.42117119e+00 2.82494664e-01
-8.65389287e-01 5.14276147e-01 8.22181284e-01 -3.22227895e-01
-1.01034522e+00 -3.12853336e-01 -1.50017425e-01 -4.33632404e-01
2.58302033e-01 2.24465474e-01 1.81395993e-01 3.48048620e-02
3.73114258e-01 2.54155695e-01 5.47003686e-01 -2.76571870e-01
3.35963428e-01 4.72064614e-01 3.73289853e-01 -1.09276867e+00
5.42404115e-01 1.35296509e-01 -1.33357435e-01 -3.31917524e-01
-8.72432292e-01 9.27773416e-02 7.84484446e-02 -3.49339575e-01
6.87997282e-01 -6.82339609e-01 -1.63569260e+00 3.28015327e-01
-7.64517844e-01 -7.77445257e-01 -7.60811388e-01 6.83389843e-01
-1.14341152e+00 3.17806602e-01 -5.29546916e-01 -1.20932233e+00
-2.31617600e-01 -1.22488070e+00 4.88152802e-01 4.49541569e-01
4.87328678e-01 -1.13185656e+00 3.29718232e-01 -1.86926737e-01
2.69096524e-01 2.25269943e-01 8.40799093e-01 -1.99584156e-01
-2.89960444e-01 2.40814701e-01 5.81908971e-02 7.15582669e-01
1.09783038e-02 -2.15285778e-01 -3.67883861e-01 -8.28177512e-01
2.20968053e-01 -4.97238070e-01 5.06482780e-01 6.99743629e-01
1.24049997e+00 -6.58244908e-01 1.06543452e-01 4.43931192e-01
1.46172321e+00 9.31791484e-01 2.80171007e-01 4.60493803e-01
1.19163655e-01 1.65720433e-01 9.05151963e-01 8.53232324e-01
2.52208173e-01 3.87677640e-01 5.54495811e-01 2.04208747e-01
5.67560315e-01 -5.11212051e-01 8.92070413e-01 4.92044806e-01
-2.09255338e-01 2.14824006e-02 -5.49428284e-01 1.07944839e-01
-2.24479795e+00 -1.21679270e+00 6.56811833e-01 2.41368604e+00
8.86938691e-01 1.31466866e-01 6.82579398e-01 -2.47803971e-01
6.87229753e-01 -2.27141809e-02 -1.02208161e+00 -6.01715922e-01
1.86431378e-01 1.70364738e-01 6.81921542e-01 9.28694665e-01
-1.03979123e+00 9.09491241e-01 7.07824278e+00 1.05077839e+00
-9.55309212e-01 -4.01453115e-02 6.78655982e-01 -1.27344429e-01
-1.64248958e-01 -2.59528428e-01 -8.03900599e-01 5.28698266e-01
5.82222402e-01 -4.49145436e-01 9.15557146e-01 9.52245355e-01
6.60486877e-01 -7.14526623e-02 -8.15616727e-01 1.00678051e+00
-6.64185166e-01 -1.32580507e+00 -1.75978526e-01 2.20233500e-01
8.16384673e-01 -6.73028305e-02 1.33967400e-01 8.36130202e-01
1.17744231e+00 -8.90640616e-01 7.58512557e-01 3.92364800e-01
5.83548367e-01 -1.27332437e+00 5.18212914e-01 5.02578318e-01
-1.02044082e+00 -5.93727648e-01 -5.18632710e-01 -3.02252650e-01
7.93289840e-02 1.13931812e-01 -4.18506324e-01 4.76931602e-01
4.84634548e-01 7.62962759e-01 9.56197456e-02 7.82308400e-01
-3.55699033e-01 4.09773022e-01 -1.26564085e-01 -2.90785879e-01
8.58945787e-01 -6.99061692e-01 5.13377249e-01 7.64580309e-01
2.35625952e-01 -1.44616887e-02 6.95907593e-01 8.80623817e-01
2.47861445e-02 1.94996614e-02 -5.22237837e-01 -2.75742680e-01
3.49176556e-01 7.02854991e-01 -3.16931725e-01 -1.21101014e-01
-2.36840263e-01 5.51564574e-01 4.59713399e-01 5.95088124e-01
-9.84263957e-01 -3.35366130e-01 1.05243027e+00 -2.89762795e-01
1.54959962e-01 -3.78017634e-01 1.69306740e-01 -1.02852023e+00
-1.01602292e-02 -1.03960979e+00 2.57833362e-01 -1.37929186e-01
-1.17945969e+00 1.99842796e-01 1.51730282e-02 -1.20524764e+00
-6.15698099e-01 -6.26893222e-01 -5.24491847e-01 6.44311309e-01
-1.55340397e+00 -2.67199665e-01 4.21206653e-01 8.07640553e-01
5.09818077e-01 -4.11733568e-01 4.49529469e-01 9.22389850e-02
-7.05283940e-01 5.32914639e-01 6.18241191e-01 -3.14830840e-02
2.25549296e-01 -1.48407209e+00 -1.18679680e-01 7.37306714e-01
-4.00641918e-01 2.32177407e-01 4.95878547e-01 -3.94249350e-01
-1.31021225e+00 -9.23335373e-01 5.61745875e-02 6.72295541e-02
7.57119715e-01 1.95927173e-01 -3.98504496e-01 5.86618662e-01
2.16424674e-01 -3.08144510e-01 2.44185686e-01 -2.70098865e-01
2.14449987e-01 -4.11076903e-01 -1.08214962e+00 7.81629443e-01
8.46418917e-01 -1.66714981e-01 -4.27256912e-01 2.41193697e-01
7.46736944e-01 -1.44935369e-01 -7.92518497e-01 5.66061437e-01
3.98241192e-01 -8.31833899e-01 8.53139341e-01 -9.66112614e-01
2.02495098e-01 -1.87166914e-01 -2.90838152e-01 -1.57129288e+00
-1.76819429e-01 -1.28241050e+00 -2.24278286e-01 6.34352148e-01
1.67526901e-01 -7.45252848e-01 3.97236049e-01 3.01772118e-01
-1.88113749e-02 -1.11502218e+00 -8.90887320e-01 -1.10083103e+00
6.22053802e-01 -3.13751012e-01 4.46030498e-01 5.36406696e-01
1.41374737e-01 -5.07342666e-02 -5.69795728e-01 -1.36262804e-01
8.25450182e-01 1.54326722e-01 4.93584961e-01 -6.36227429e-01
-8.04514468e-01 -9.22274470e-01 2.20227569e-01 -1.53054070e+00
3.38276654e-01 -6.86167181e-01 1.71308547e-01 -1.40012383e+00
-6.52583614e-02 -6.37828290e-01 -5.68377733e-01 2.82571852e-01
-1.61482066e-01 -6.65945351e-01 4.39482480e-01 5.63728483e-03
-8.18741024e-01 9.98277426e-01 1.61754417e+00 1.08033910e-01
-2.96960235e-01 3.70588928e-01 -6.28645539e-01 8.77034605e-01
9.44184959e-01 -2.27334037e-01 -7.66051888e-01 -3.72371167e-01
1.85131267e-01 5.30546606e-01 2.01916143e-01 -6.42242014e-01
-1.30709648e-01 -7.25248039e-01 -6.34570047e-02 -2.54180342e-01
-5.82128651e-02 -5.59879601e-01 -5.08487999e-01 9.00627732e-01
-5.60897708e-01 4.55491841e-01 -3.84264253e-02 6.35761976e-01
-9.74024683e-02 -3.61671448e-01 9.84462917e-01 -1.29486933e-01
-6.60360873e-01 6.54714882e-01 -6.17629826e-01 4.86745387e-01
1.22642756e+00 2.80167967e-01 -1.11316741e-01 -6.58131778e-01
-5.10014117e-01 7.24799871e-01 8.01748112e-02 1.27901271e-01
5.37790358e-01 -1.49782073e+00 -5.68376899e-01 2.60889884e-02
-6.09955132e-01 -1.21759117e-01 -1.86015576e-01 6.30969167e-01
-3.94953102e-01 3.69732201e-01 -2.25907132e-01 -2.94576108e-01
-5.27785540e-01 6.21467769e-01 8.16543877e-01 -7.18953192e-01
-3.66917074e-01 5.96551776e-01 1.43844023e-01 -4.46228921e-01
5.92845023e-01 -2.90500194e-01 7.32234539e-03 -2.71160483e-01
5.95732450e-01 5.00684500e-01 -5.31992853e-01 -5.28886355e-02
-3.24125141e-02 2.87772208e-01 -5.23509309e-02 -5.79423428e-01
1.23372316e+00 -4.58263978e-02 2.17693195e-01 6.81179836e-02
8.34909320e-01 -4.28922117e-01 -2.04075527e+00 -3.47125769e-01
5.91390766e-02 -3.65403086e-01 6.11731932e-02 -6.63301170e-01
-9.73594606e-01 7.75164485e-01 7.10078657e-01 2.32261300e-01
1.02767420e+00 -5.32045543e-01 5.59707224e-01 5.60007393e-01
5.68379581e-01 -1.52901506e+00 3.31483930e-01 8.57305408e-01
7.19244838e-01 -1.05677259e+00 -3.22563916e-01 4.56873119e-01
-7.89379597e-01 8.84732246e-01 6.48290217e-01 -5.85987449e-01
6.56468272e-01 3.11507255e-01 -1.66496173e-01 2.36020789e-01
-7.89182723e-01 -2.93953329e-01 -2.83150643e-01 4.18463409e-01
2.10676733e-02 1.60979122e-01 -6.37254953e-01 4.60140437e-01
-1.17531188e-01 1.85854331e-01 2.08680555e-01 1.09935236e+00
-8.07890236e-01 -1.26742196e+00 -1.60623729e-01 2.34378889e-01
-4.56704170e-01 2.96436191e-01 2.38330275e-01 7.30215132e-01
-2.05692619e-01 9.13009107e-01 1.06734671e-01 -9.49949697e-02
2.31448457e-01 -1.39965713e-01 6.55539691e-01 -1.57113895e-01
-3.20573568e-01 2.00438142e-01 -3.37203532e-01 -7.77929485e-01
-3.79906714e-01 -4.20473576e-01 -1.33568263e+00 -4.46943730e-01
1.07536577e-01 4.99658585e-01 2.53120601e-01 1.20781887e+00
1.47725776e-01 4.13378894e-01 1.05659747e+00 -5.20270586e-01
-1.62823677e+00 -5.68715572e-01 -6.76455557e-01 2.40943223e-01
4.44838822e-01 -9.07152832e-01 -1.17510214e-01 -4.69997913e-01]
|
[4.156148433685303, 2.4112861156463623]
|
ead2a45b-c55e-4405-894f-8b353a92653b
|
text2model-model-induction-for-zero-shot
|
2210.15182
| null |
https://arxiv.org/abs/2210.15182v1
|
https://arxiv.org/pdf/2210.15182v1.pdf
|
Text2Model: Model Induction for Zero-shot Generalization Using Task Descriptions
|
We study the problem of generating a training-free task-dependent visual classifier from text descriptions without visual samples. This \textit{Text-to-Model} (T2M) problem is closely related to zero-shot learning, but unlike previous work, a T2M model infers a model tailored to a task, taking into account all classes in the task. We analyze the symmetries of T2M, and characterize the equivariance and invariance properties of corresponding models. In light of these properties, we design an architecture based on hypernetworks that given a set of new class descriptions predicts the weights for an object recognition model which classifies images from those zero-shot classes. We demonstrate the benefits of our approach compared to zero-shot learning from text descriptions in image and point-cloud classification using various types of text descriptions: From single words to rich text descriptions.
|
['Gal Chechik', 'Roi Reichart', 'Eyal Ben-David', 'Tomer Volk', 'Ohad Amosy']
|
2022-10-27
| null | null | null | null |
['point-cloud-classification']
|
['computer-vision']
|
[ 4.59946603e-01 4.38465685e-01 -4.89010721e-01 -6.98278844e-01
-5.77334344e-01 -1.97152406e-01 1.24570632e+00 -1.47858575e-01
-7.19109401e-02 2.76841879e-01 2.71914214e-01 -8.67253318e-02
-2.59934515e-01 -6.55302465e-01 -6.55234933e-01 -6.06210768e-01
2.56666601e-01 9.50106442e-01 3.05626512e-01 -4.01030369e-02
3.84668589e-01 1.67234972e-01 -1.95909297e+00 6.12144828e-01
3.16350967e-01 9.55272973e-01 4.21326101e-01 7.60930836e-01
-6.49449229e-01 9.72254634e-01 -2.87333250e-01 -3.91642421e-01
2.84635514e-01 -2.34765619e-01 -7.46154845e-01 5.85341156e-01
1.02662909e+00 -2.53420174e-01 -4.25454497e-01 1.09677410e+00
1.98623195e-01 5.85552275e-01 1.52672207e+00 -1.68121064e+00
-1.15562177e+00 4.10028189e-01 -2.97089607e-01 -1.01963192e-01
4.38303500e-02 1.16023220e-01 1.14535642e+00 -1.35404861e+00
1.00331819e+00 1.40871727e+00 6.11243784e-01 9.20896351e-01
-1.51704633e+00 -4.02187407e-01 2.78458353e-02 3.62122178e-01
-1.27862823e+00 -7.87164450e-01 6.88646615e-01 -8.30864251e-01
1.11979938e+00 -2.65657320e-03 4.66851801e-01 1.53357029e+00
1.12164326e-01 7.61958003e-01 7.97769427e-01 -7.47055411e-01
5.42408407e-01 4.31957126e-01 5.00263631e-01 7.88638055e-01
1.79458514e-01 -1.85439475e-02 -7.83862889e-01 -7.92499557e-02
5.11577010e-01 2.60088891e-01 -8.01160261e-02 -1.20915270e+00
-9.79040563e-01 1.00932050e+00 3.19606632e-01 1.38552845e-01
2.33431570e-02 1.90904498e-01 2.75257945e-01 2.66048282e-01
6.22114778e-01 3.32006812e-01 -1.90117598e-01 3.57170492e-01
-9.79599655e-01 -4.48175296e-02 8.25213373e-01 1.61495781e+00
9.19301927e-01 3.75430882e-01 -3.20147842e-01 9.48784947e-01
2.05082968e-01 4.10192460e-01 8.32788885e-01 -7.83265352e-01
2.27211416e-01 3.56813550e-01 -2.11101428e-01 -4.96710896e-01
-2.22780347e-01 4.91200984e-02 -8.65624011e-01 4.28264230e-01
-6.54672133e-03 2.67542452e-01 -1.37477291e+00 1.43613040e+00
-2.53229439e-01 3.14669535e-02 2.75222540e-01 6.43999159e-01
9.75504816e-01 5.84979594e-01 4.67003845e-02 -1.12970918e-02
1.01059043e+00 -9.02580798e-01 -5.98476589e-01 -2.18041897e-01
5.08467972e-01 -1.91544741e-01 1.05273461e+00 8.25711787e-02
-9.59402263e-01 -7.19188571e-01 -1.08406365e+00 -2.14690909e-01
-8.93504262e-01 -3.00252289e-01 8.88831243e-02 3.03156495e-01
-1.21252167e+00 6.95833683e-01 -3.68451506e-01 -9.08652902e-01
4.71216828e-01 2.01588631e-01 -4.90010709e-01 -1.31316185e-01
-7.57730961e-01 1.13984382e+00 6.61375880e-01 -6.39857650e-01
-1.16621578e+00 -5.83250582e-01 -1.27983749e+00 2.75777072e-01
3.14146698e-01 -5.56703866e-01 1.34419966e+00 -1.08646166e+00
-1.08949995e+00 1.13173175e+00 -2.14924723e-01 -3.79314691e-01
3.07447314e-01 5.52569389e-01 -3.71865891e-02 3.00645053e-01
2.54049659e-01 1.03152049e+00 1.48198986e+00 -1.48564422e+00
-5.60269117e-01 -3.17794144e-01 -1.22616924e-01 7.84885436e-02
-6.42749786e-01 -2.95867383e-01 -2.70805120e-01 -4.52661842e-01
-4.10105772e-02 -6.69365823e-01 -6.30611368e-03 5.03178656e-01
-2.46887758e-01 -1.85315311e-01 1.03204024e+00 -1.13581136e-01
4.53581572e-01 -2.16231346e+00 1.71633214e-01 -2.94700847e-03
5.75117469e-01 -4.34484426e-03 -4.48407620e-01 4.48860079e-01
-1.94704324e-01 1.92062289e-01 -8.60970393e-02 -5.41295230e-01
3.77525628e-01 3.98960888e-01 -3.80075455e-01 2.66263574e-01
2.65824258e-01 1.07020652e+00 -7.08468258e-01 -7.89916754e-01
6.64698422e-01 2.33544022e-01 -1.64217040e-01 1.71961769e-01
-4.57661569e-01 -1.67606413e-01 -1.67226776e-01 3.45446438e-01
4.16269362e-01 -6.00667179e-01 5.36083877e-02 1.04496963e-02
1.25440881e-01 -5.19662201e-01 -8.00701499e-01 1.50834668e+00
-5.14732003e-01 8.79861712e-01 -3.50628227e-01 -1.15395248e+00
9.84183192e-01 4.78191227e-01 2.04864740e-01 -5.13570964e-01
1.39375851e-01 -1.71171308e-01 -4.14651930e-01 -8.02551925e-01
3.04154485e-01 -6.50601029e-01 -9.24641080e-03 4.77528661e-01
1.04598212e+00 -1.84359953e-01 8.91579241e-02 2.91093528e-01
8.38406205e-01 1.38408944e-01 6.38626516e-01 -3.02010179e-01
-6.53612465e-02 4.92811315e-02 1.36945188e-01 1.18477690e+00
-2.50086576e-01 8.34348321e-01 3.21653605e-01 -7.94883788e-01
-1.53259647e+00 -1.13082957e+00 -2.60105580e-01 1.36408830e+00
5.39243557e-02 -3.03705573e-01 -3.94904047e-01 -6.12056971e-01
-2.07201336e-02 1.13290238e+00 -1.05507398e+00 -3.60503942e-01
2.82772183e-01 -2.64617026e-01 1.53900519e-01 5.33732533e-01
3.07780325e-01 -1.03654432e+00 -6.70084774e-01 -1.91727817e-01
7.84875005e-02 -1.03401959e+00 -3.54041427e-01 5.28675675e-01
-6.99679434e-01 -1.01609468e+00 -9.84617472e-01 -1.21203732e+00
7.73420453e-01 5.05569398e-01 1.11161256e+00 -3.19395751e-01
-5.57913363e-01 7.56075621e-01 -4.30097669e-01 -5.36936462e-01
-4.85639662e-01 -2.17752650e-01 5.10118827e-02 3.27476203e-01
6.79872394e-01 -4.11075652e-01 1.01658270e-01 4.43663262e-02
-9.08799827e-01 3.44257832e-01 4.28223848e-01 1.01469815e+00
3.30330521e-01 -7.96468854e-02 1.87583148e-01 -9.31538761e-01
3.88340861e-01 -3.58774424e-01 -4.36436862e-01 5.80197394e-01
-6.74460948e-01 3.27537864e-01 6.82433724e-01 -8.23515892e-01
-1.14470351e+00 3.72688770e-01 6.44088328e-01 -1.11316335e+00
-3.99191320e-01 9.20336619e-02 -9.39505547e-03 -1.33285388e-01
9.30514276e-01 4.34873730e-01 1.09183013e-01 -2.14464605e-01
6.97221816e-01 7.57565081e-01 3.53879988e-01 -3.55940431e-01
8.89007092e-01 6.63465023e-01 1.56046703e-01 -1.37907875e+00
-1.08113194e+00 -9.09354031e-01 -1.11968708e+00 -2.62692600e-01
1.20450318e+00 -6.63065553e-01 -3.13906640e-01 1.23465315e-01
-1.32802200e+00 -2.76866227e-01 -6.35073662e-01 2.54849464e-01
-1.19364512e+00 2.13424891e-01 -2.79459268e-01 -8.47099364e-01
-1.17652111e-01 -7.31891274e-01 1.20763457e+00 -2.19185069e-01
-2.07273245e-01 -1.29345095e+00 -1.31101133e-02 -1.07748937e-02
2.28235066e-01 9.33200773e-03 1.38538539e+00 -9.53441560e-01
-3.56527746e-01 -2.71544218e-01 -2.58251727e-01 2.75322437e-01
-3.13458741e-01 -2.75584787e-01 -1.36745405e+00 -2.56636679e-01
1.26292437e-01 -7.65836537e-01 1.00381875e+00 5.53939581e-01
1.12200582e+00 -4.79555339e-01 -2.69960195e-01 7.57659554e-01
1.70189726e+00 2.69698258e-02 5.15467167e-01 1.78038076e-01
6.71173692e-01 8.14620137e-01 2.62726873e-01 4.26723510e-01
1.13301445e-03 4.67907310e-01 4.86858487e-01 1.17882282e-01
-3.16652060e-01 -4.28598136e-01 8.39442909e-02 3.63965452e-01
9.75275487e-02 -2.53401369e-01 -9.87769186e-01 5.14615536e-01
-1.93321502e+00 -1.18091857e+00 2.35602513e-01 2.13125086e+00
4.42941576e-01 7.81638771e-02 -2.05109000e-01 -2.53660053e-01
9.34621811e-01 2.70133644e-01 -6.19503200e-01 -3.09029281e-01
-3.89402285e-02 1.69047132e-01 2.80073017e-01 4.30229157e-01
-1.12142968e+00 1.13464975e+00 7.04278755e+00 8.20151627e-01
-6.76335990e-01 2.26391748e-01 2.24202827e-01 -4.36518192e-02
-7.94588327e-02 1.19285457e-01 -7.96241462e-01 -6.35122135e-02
9.94407773e-01 -5.64011872e-01 2.48109996e-01 1.16199958e+00
-2.02543855e-01 4.80685562e-01 -1.53860998e+00 1.03724802e+00
9.48572576e-01 -1.59960032e+00 8.72254610e-01 7.83385932e-02
8.51628959e-01 6.06869422e-02 8.32488388e-02 6.30643368e-01
4.90090042e-01 -9.38107312e-01 7.53557682e-01 7.48362482e-01
1.08240354e+00 -3.54443222e-01 2.21460834e-01 5.84558249e-01
-1.07018995e+00 -3.07212651e-01 -7.97011256e-01 5.42960577e-02
-3.06960374e-01 -1.52082771e-01 -9.65570509e-01 8.26607794e-02
5.88774860e-01 9.94279742e-01 -7.35746622e-01 7.95762479e-01
2.37714276e-01 4.81906720e-02 4.08912987e-01 -2.57063508e-01
2.08033741e-01 1.89975947e-02 4.60106701e-01 1.03074431e+00
2.72734553e-01 -3.17043327e-02 3.51045817e-01 1.09507608e+00
-3.32938991e-02 3.48316580e-02 -1.38862908e+00 -1.77040249e-02
6.28076121e-02 1.26716483e+00 -7.36873388e-01 -9.22081709e-01
-6.61788762e-01 1.05061316e+00 3.47897947e-01 5.13748586e-01
-3.16745073e-01 -5.94781518e-01 2.52675503e-01 1.16208218e-01
5.05463541e-01 1.11325651e-01 -1.09598543e-02 -1.42089021e+00
-2.73096651e-01 -5.27266383e-01 1.85427129e-01 -1.43191195e+00
-1.62922370e+00 4.46828842e-01 2.99028248e-01 -1.60393786e+00
-4.52962130e-01 -1.08706093e+00 -5.81544697e-01 5.70384562e-01
-1.19484639e+00 -1.59425616e+00 -3.16659898e-01 6.49118006e-01
1.21520960e+00 -6.46324098e-01 1.23633015e+00 -4.75427628e-01
6.12827651e-02 2.83965856e-01 3.15838158e-01 1.70734197e-01
5.71650147e-01 -1.32326066e+00 6.05160773e-01 4.10777897e-01
4.25308913e-01 4.04649138e-01 6.31620526e-01 -5.01843214e-01
-1.10368371e+00 -1.28222322e+00 9.73638713e-01 -7.30152965e-01
9.33926821e-01 -7.23638892e-01 -8.27981830e-01 8.84233356e-01
1.57350883e-01 1.93895265e-01 6.15117133e-01 6.46037683e-02
-8.75068963e-01 2.44648635e-01 -8.32667589e-01 7.17035115e-01
9.89434361e-01 -8.19301784e-01 -9.77123201e-01 5.96124828e-01
6.15512133e-01 2.63452142e-01 -5.53008974e-01 -1.31455585e-01
4.75417703e-01 -6.97201490e-01 8.61209750e-01 -1.14799953e+00
5.99064887e-01 2.77080625e-01 -4.54170942e-01 -1.27574277e+00
-6.33597076e-01 -2.14568838e-01 -7.25288913e-02 8.11269939e-01
2.31838286e-01 -2.32450619e-01 7.05791235e-01 5.62792480e-01
-1.62821501e-01 -1.26865029e-01 -9.07907248e-01 -1.14713299e+00
2.66508162e-01 -2.95539856e-01 2.93047335e-02 1.14612925e+00
1.92163527e-01 8.84601295e-01 -6.56512618e-01 -2.75973648e-01
1.12076938e+00 -2.42526624e-02 6.11349940e-01 -1.78282809e+00
-2.20705807e-01 -4.31180000e-01 -8.03580105e-01 -5.33920646e-01
6.87667012e-01 -1.36847484e+00 4.06787932e-01 -1.64257169e+00
8.41751337e-01 1.55986503e-01 -3.08570303e-02 7.40306616e-01
4.42529380e-01 2.96209723e-01 4.85300183e-01 4.48871285e-01
-8.42476010e-01 5.41507065e-01 9.95019674e-01 -6.58029437e-01
2.03078046e-01 -1.73140958e-01 -3.62729996e-01 9.46497142e-01
4.12720352e-01 -4.42997307e-01 -6.01820946e-01 -2.87696868e-01
-6.94800019e-02 1.16130672e-01 6.34372056e-01 -9.89563763e-01
2.89455175e-01 -2.99017668e-01 6.02242708e-01 -4.36927885e-01
6.82109416e-01 -8.71454060e-01 -3.01361322e-01 2.04073533e-01
-9.87839162e-01 -3.90654206e-01 -2.25445285e-01 8.99633527e-01
1.07754543e-01 -7.54298270e-01 8.71530414e-01 -5.92577457e-01
-1.30288136e+00 5.03037751e-01 -4.75017726e-01 -1.28326891e-03
9.61548507e-01 -5.80263317e-01 -5.54584384e-01 -5.66766500e-01
-1.10275519e+00 8.61841142e-02 5.75603545e-01 6.82482481e-01
1.03193617e+00 -1.38299716e+00 -5.03254950e-01 5.14921606e-01
1.05135632e+00 -4.47056144e-01 1.36297256e-01 2.86918342e-01
3.80479395e-02 5.03554821e-01 -5.20533621e-01 -6.49512529e-01
-1.17694664e+00 1.11406302e+00 3.35070938e-01 3.53355289e-01
-1.12127268e+00 3.52896631e-01 6.76600456e-01 -5.30836284e-01
5.02780437e-01 1.66121259e-01 -3.14590424e-01 5.50009683e-02
5.79726100e-01 1.66683942e-01 -1.37492672e-01 -7.16450036e-01
2.33631693e-02 7.18929291e-01 -5.92180230e-02 -4.13373083e-01
1.24468184e+00 -1.78344492e-02 1.95030823e-01 1.24669242e+00
1.41151452e+00 -9.80436146e-01 -1.36118972e+00 -4.20172244e-01
1.42969459e-01 -3.89983535e-01 -4.17585894e-02 -3.92937005e-01
-6.24094307e-01 1.31334841e+00 4.80425328e-01 1.52176648e-01
5.31664312e-01 2.42296159e-01 -1.20475171e-02 9.26667273e-01
2.21025795e-01 -1.26210701e+00 6.00616634e-01 6.55904830e-01
8.10523748e-01 -1.48853016e+00 -2.96206862e-01 4.27246764e-02
-9.10087466e-01 1.37999272e+00 5.54406881e-01 -2.19422057e-01
8.74982715e-01 9.92141105e-03 -7.69592226e-02 -3.75958443e-01
-1.12131786e+00 -5.39144278e-01 4.95058984e-01 1.15399516e+00
-8.58062059e-02 -1.85760796e-01 4.20863718e-01 5.71132042e-02
7.00231493e-02 -1.15029141e-02 6.17114842e-01 9.24185395e-01
-8.34804475e-01 -5.70901990e-01 -1.37330040e-01 7.23118544e-01
3.67212623e-01 -3.24833810e-01 -6.13735795e-01 8.66425216e-01
-4.77022827e-02 6.83863163e-01 4.23009723e-01 -4.68096584e-01
1.73945576e-02 7.61525035e-01 5.14121234e-01 -1.19441473e+00
7.37964883e-02 -1.65308878e-01 -1.46459132e-01 -1.23170823e-01
-2.76689649e-01 -4.08989340e-01 -9.97439146e-01 4.08496223e-02
-2.24410161e-01 -1.75314039e-01 7.18019783e-01 9.82298791e-01
3.31143439e-02 2.93981612e-01 4.95756119e-01 -1.17006242e+00
-8.01877022e-01 -1.08927131e+00 -9.62337434e-01 7.39915907e-01
4.26390171e-01 -8.22852373e-01 -6.05010629e-01 5.10531545e-01]
|
[10.104207038879395, 2.3220856189727783]
|
7e623293-f71b-4b15-8adf-2a39d0de4f7d
|
the-cacapo-dataset-a-multilingual-multi
| null | null |
https://aclanthology.org/2020.inlg-1.10
|
https://aclanthology.org/2020.inlg-1.10.pdf
|
The CACAPO Dataset: A Multilingual, Multi-Domain Dataset for Neural Pipeline and End-to-End Data-to-Text Generation
|
This paper describes the CACAPO dataset, built for training both neural pipeline and end-to-end data-to-text language generation systems. The dataset is multilingual (Dutch and English), and contains almost 10,000 sentences from human-written news texts in the sports, weather, stocks, and incidents domain, together with aligned attribute-value paired data. The dataset is unique in that the linguistic variation and indirect ways of expressing data in these texts reflect the challenges of real world NLG tasks.
|
['Emiel Krahmer', 'Sander Wubben', 'Chris Emmery', 'Chris van der Lee']
| null | null | null | null |
inlg-acl-2020-12
|
['data-to-text-generation']
|
['natural-language-processing']
|
[-6.13773242e-02 3.46478879e-01 -4.16452289e-01 -7.42388606e-01
-1.09034979e+00 -8.63264680e-01 1.09310579e+00 2.03803003e-01
-6.39360070e-01 1.47332287e+00 1.12558067e+00 -1.89109549e-01
3.10382783e-01 -8.26550126e-01 -6.86061263e-01 -3.78371067e-02
5.95849566e-02 1.12120843e+00 -4.43031520e-01 -9.30705309e-01
-5.35008945e-02 -9.56728086e-02 -9.98002827e-01 8.19405019e-01
6.76887274e-01 7.99057901e-01 -8.54531489e-03 6.38740778e-01
-5.45876980e-01 1.09640586e+00 -8.63010466e-01 -9.02110338e-01
2.12526262e-01 -6.43062830e-01 -8.26768577e-01 -3.13261598e-01
3.02415788e-01 6.42716512e-02 -2.77338117e-01 4.38910455e-01
9.53556955e-01 2.76359022e-02 6.98702812e-01 -1.09527612e+00
-1.09169877e+00 1.54218078e+00 -4.66721654e-02 2.11444050e-01
7.06137836e-01 8.66478160e-02 1.32691252e+00 -9.26804185e-01
1.36161387e+00 1.39368892e+00 5.61539829e-01 7.67340362e-01
-1.18483591e+00 -4.44533348e-01 -2.43968397e-01 -1.11469977e-01
-1.10420454e+00 -7.78928459e-01 2.68199414e-01 -5.10436893e-01
1.38422179e+00 -1.00414112e-01 4.31031644e-01 1.79350901e+00
2.36705601e-01 1.01472855e+00 9.46925521e-01 -5.27280509e-01
-1.23831838e-01 8.59596580e-02 -2.38575757e-01 1.55702859e-01
-6.49059266e-02 3.71015519e-01 -1.00814176e+00 -1.19076647e-01
1.03525892e-01 -9.30356443e-01 -3.19022238e-02 4.16909307e-01
-1.77005649e+00 8.90427649e-01 9.97850895e-02 2.20404655e-01
-4.00013268e-01 -1.08745560e-01 7.95398533e-01 5.34874737e-01
7.31300890e-01 6.56130314e-01 -8.53427112e-01 -4.22140807e-01
-7.06786215e-01 7.91519880e-01 1.03495574e+00 1.34799004e+00
3.14627379e-01 3.67317200e-01 -5.45631826e-01 1.06045592e+00
1.55493859e-02 5.93055427e-01 9.63237166e-01 -4.86593038e-01
1.18445861e+00 1.16598040e-01 1.00808427e-01 -7.02121139e-01
-4.39431041e-01 6.60096342e-03 -5.95928967e-01 -4.66869473e-01
3.95432115e-01 -7.98353791e-01 -7.13374138e-01 1.74490023e+00
1.68723479e-01 -5.67267835e-01 8.36032093e-01 5.45653343e-01
1.41369140e+00 9.15943265e-01 2.62023479e-01 2.83073839e-02
1.24582911e+00 -6.94084704e-01 -9.03660238e-01 -5.77504396e-01
6.58808708e-01 -8.23510170e-01 1.06660283e+00 2.85038501e-02
-1.27158523e+00 -5.27832448e-01 -7.78426826e-01 -6.00286067e-01
-8.53962183e-01 1.24601414e-03 2.63084769e-01 1.37973011e-01
-7.81886876e-01 1.08424403e-01 -1.29241139e-01 -2.71744579e-01
8.19334239e-02 -3.84404540e-01 -4.96789396e-01 1.15596980e-01
-1.98451161e+00 1.19137216e+00 1.06903887e+00 -1.13738343e-01
-4.88244474e-01 -6.67947650e-01 -1.10778105e+00 -3.83735210e-01
2.18165919e-01 -3.74970078e-01 1.50046122e+00 -6.59767926e-01
-1.40102482e+00 1.03846359e+00 -3.14215256e-04 -7.32327580e-01
6.18326366e-01 -2.11907193e-01 -8.78538072e-01 -4.19993043e-01
5.04563749e-01 8.30175281e-01 1.69047967e-01 -6.49705946e-01
-8.28359544e-01 -1.27650604e-01 -3.77108097e-01 3.92095208e-01
1.95818394e-01 3.54106784e-01 1.44256398e-01 -1.15961719e+00
-4.48033571e-01 -6.26530528e-01 -1.07643567e-01 -6.53486788e-01
-6.78748846e-01 -4.28252518e-01 1.52445853e-01 -9.96017039e-01
8.27327490e-01 -2.08662724e+00 1.03588760e-01 -1.72465667e-01
-4.27228004e-01 -1.77767828e-01 -3.87661874e-01 9.36067879e-01
1.02040820e-01 2.31325716e-01 -8.70952457e-02 -1.88221231e-01
3.39219362e-01 3.42616647e-01 -6.87494457e-01 -4.52770144e-02
5.22971213e-01 1.21031737e+00 -1.09215033e+00 -5.85654318e-01
-2.07629859e-01 1.65487453e-01 -8.72601494e-02 1.84650019e-01
-5.98285317e-01 4.58410531e-01 -2.20692694e-01 4.33949947e-01
1.27175311e-03 5.50526083e-01 1.52930185e-01 -5.85917160e-02
-3.02917600e-01 1.00346994e+00 -6.82852626e-01 1.71424508e+00
-5.49771369e-01 8.09036851e-01 -3.35594475e-01 -5.91031730e-01
9.11509335e-01 6.48013473e-01 1.14400834e-01 -9.80133116e-01
1.00576580e-01 3.43052596e-01 -1.76357597e-01 -5.60567021e-01
1.01453292e+00 -3.04106712e-01 -1.06031561e+00 3.78244251e-01
3.54703754e-01 -6.13609374e-01 8.71224999e-01 2.59118583e-02
6.11757994e-01 1.61883369e-01 5.79256356e-01 -2.44759947e-01
8.74869600e-02 3.96507382e-01 7.16733098e-01 5.64600706e-01
2.34258458e-01 6.37830615e-01 5.16085684e-01 -5.92831194e-01
-1.22852874e+00 -1.04246700e+00 -2.31666416e-01 1.28107893e+00
-7.12942839e-01 -3.64081472e-01 -5.06607115e-01 -3.93492877e-01
1.95604004e-02 1.19695067e+00 -6.79880977e-01 2.89805770e-01
-6.08705342e-01 -6.06347799e-01 1.20358086e+00 5.38210094e-01
2.80598700e-01 -1.62052214e+00 -7.81299248e-02 7.42182255e-01
-6.82606578e-01 -1.51470971e+00 -3.98947626e-01 -3.46960537e-02
-1.51347563e-01 -6.47496700e-01 -4.70155984e-01 -6.55234635e-01
-6.98329210e-02 -7.73942053e-01 1.79342341e+00 -7.01016903e-01
-2.34927833e-02 -2.07955897e-01 -4.88237590e-01 -8.92447472e-01
-1.01498163e+00 4.60446954e-01 -1.17390916e-01 -2.64803499e-01
4.49087620e-01 3.69546004e-02 3.11933219e-01 -2.93884814e-01
-6.38453662e-01 1.85540915e-01 4.52942878e-01 1.00463080e+00
6.75848603e-01 -5.22769392e-01 9.57614303e-01 -1.06833196e+00
1.19486022e+00 -7.98379779e-01 -3.69265020e-01 3.35733265e-01
-1.37189597e-01 6.99068084e-02 7.87262559e-01 -8.37375671e-02
-1.24833083e+00 -1.02163464e-01 -2.27519080e-01 6.46893144e-01
-4.17755157e-01 1.03257918e+00 -2.30321825e-01 1.01918185e+00
8.79026890e-01 2.10483477e-01 -2.00644508e-01 -3.43692631e-01
8.70182157e-01 1.03311563e+00 9.77213085e-01 -7.44739652e-01
3.13889563e-01 -1.89556777e-01 -4.28213328e-01 -7.46503532e-01
-1.02341366e+00 2.50380784e-01 -6.73294723e-01 -2.53592767e-02
9.09590602e-01 -1.32732272e+00 -9.07291099e-03 5.18249214e-01
-1.32528198e+00 -6.04540408e-01 -7.50303090e-01 3.84367347e-01
-6.04637563e-01 -4.78322238e-01 -8.94010782e-01 -3.59203041e-01
-6.09842360e-01 -7.55261362e-01 1.03245139e+00 -1.68028697e-01
-6.85810804e-01 -1.02927959e+00 3.37537199e-01 1.17263466e-01
3.00085574e-01 7.21536040e-01 9.09072995e-01 -9.25237179e-01
1.89403683e-01 -1.25205711e-01 1.19956754e-01 2.20494226e-01
1.15739904e-01 7.68897757e-02 -5.84704459e-01 -2.17908975e-02
-5.58479726e-01 -9.62391734e-01 4.63329583e-01 9.02324319e-02
4.86241370e-01 -5.49510181e-01 1.61418587e-01 2.35841304e-01
1.01022875e+00 3.64288827e-03 2.81562358e-01 3.52855355e-01
5.92954576e-01 9.51585293e-01 5.42847335e-01 4.46240425e-01
9.25719917e-01 5.96293449e-01 -3.50098610e-01 -1.62492812e-01
-2.71152049e-01 -4.99383628e-01 5.24738133e-01 9.89658475e-01
3.88972104e-01 -6.10270858e-01 -1.15380669e+00 9.60483491e-01
-1.66416979e+00 -1.09702361e+00 -2.72932440e-01 1.65323162e+00
1.58929169e+00 1.44080287e-02 1.34603187e-01 -2.87283808e-01
5.44343531e-01 2.76340455e-01 -3.76076281e-01 -7.94286668e-01
-8.30191016e-01 2.35116571e-01 3.30285579e-01 3.76808852e-01
-9.77733672e-01 1.17327797e+00 7.40155745e+00 8.04302990e-01
-9.13863420e-01 -4.96283844e-02 6.45853996e-01 -3.31170619e-01
-4.25921857e-01 -3.21019530e-01 -9.71145928e-01 6.32639587e-01
1.36219740e+00 -7.09697187e-01 3.28249753e-01 5.05087912e-01
1.77151486e-01 2.93476045e-01 -1.19915009e+00 3.10084820e-01
7.71238580e-02 -1.37079382e+00 2.90130973e-01 -3.25848669e-01
8.53284538e-01 6.35834038e-01 -8.69831815e-02 4.87931520e-01
9.28320110e-01 -1.02200961e+00 1.36871862e+00 4.21017200e-01
1.09640682e+00 -8.24357629e-01 7.77706742e-01 4.17066008e-01
-8.12872052e-01 2.73909003e-01 -1.16898909e-01 2.71558594e-02
6.34592414e-01 6.29358470e-01 -1.02261448e+00 5.24653554e-01
5.05284965e-01 9.20130551e-01 -4.74160284e-01 2.23829955e-01
-4.41506743e-01 6.35746717e-01 -1.05150543e-01 -2.05005065e-01
4.77678031e-01 4.83490527e-02 4.80790704e-01 1.71633244e+00
3.22805792e-01 -9.82649475e-02 1.82902932e-01 6.92785800e-01
-3.11442286e-01 3.85711968e-01 -9.44813967e-01 -4.48339552e-01
6.03841245e-01 9.54936564e-01 -1.48647100e-01 -4.40363735e-01
-4.49572474e-01 6.02882385e-01 4.92088169e-01 3.02827150e-01
-3.47737730e-01 -5.77649951e-01 5.04468322e-01 -6.81987181e-02
-2.15762146e-02 -1.33406714e-01 -1.71855673e-01 -1.00856233e+00
-5.55840842e-02 -1.24932599e+00 4.97347355e-01 -8.09026897e-01
-1.99416316e+00 9.18789983e-01 1.10367604e-01 -8.36152732e-01
-1.24110770e+00 -4.67711389e-01 -2.89088756e-01 1.05651450e+00
-1.21383190e+00 -1.38995433e+00 2.51500607e-01 4.31932002e-01
6.68880880e-01 -8.62115979e-01 8.79996300e-01 1.90658212e-01
-2.13596448e-01 5.77059746e-01 1.93111643e-01 6.45408392e-01
9.48322058e-01 -1.33620071e+00 1.29149318e+00 6.14818335e-01
2.45471299e-01 2.44307414e-01 7.36031234e-01 -7.67347932e-01
-9.08082366e-01 -1.33372748e+00 1.69992411e+00 -6.59586430e-01
8.81336391e-01 -5.88803053e-01 -5.70242405e-01 8.73626769e-01
8.68458271e-01 -3.43010008e-01 7.52599657e-01 7.93536678e-02
-1.21770494e-01 1.61186215e-02 -1.15800476e+00 7.53605425e-01
8.44190776e-01 -7.09611952e-01 -8.38577569e-01 5.18211424e-01
8.63074005e-01 -9.45469618e-01 -8.72766554e-01 2.39227653e-01
4.27905023e-01 -3.00037503e-01 6.54683053e-01 -1.15816808e+00
8.88993680e-01 8.47428739e-02 -2.53397465e-01 -2.00716186e+00
-1.43379241e-01 -5.29459774e-01 2.09900543e-01 1.66577554e+00
1.20974839e+00 -4.89627242e-01 9.95549485e-02 4.12116796e-01
-3.89258027e-01 -4.39449340e-01 -9.82939124e-01 -5.76077580e-01
6.38283730e-01 -4.73557234e-01 1.09021509e+00 1.17984223e+00
7.97609761e-02 6.97912693e-01 -2.73259640e-01 -5.19381166e-01
1.28713876e-01 -3.92034240e-02 6.50767744e-01 -8.70514214e-01
-1.70476511e-01 -4.04941380e-01 -2.46781539e-02 -5.03782988e-01
5.92807710e-01 -1.18317592e+00 3.14942926e-01 -1.57320380e+00
-2.74201632e-01 -3.49228084e-01 2.72004098e-01 7.70991504e-01
-9.21406969e-02 6.01487570e-02 1.64244935e-01 -2.30827644e-01
-7.32252300e-02 5.40906429e-01 9.86572385e-01 -2.09328786e-01
2.67537311e-03 -4.62242126e-01 -6.30118549e-01 3.07370871e-01
8.91826451e-01 -4.44819570e-01 -2.08030477e-01 -1.03056443e+00
6.13543272e-01 1.24124609e-01 -1.40204400e-01 -5.67514420e-01
-1.52218476e-01 -3.92524272e-01 4.13590133e-01 -6.81666195e-01
1.10832579e-01 -1.94717586e-01 1.08834572e-01 1.47175640e-01
-9.75462139e-01 5.54125667e-01 2.80239373e-01 1.09654218e-02
-5.26631176e-01 3.32544819e-02 4.54907596e-01 -3.21106851e-01
-6.84958637e-01 9.35521051e-02 -6.64755821e-01 9.98667240e-01
7.98253298e-01 1.24369256e-01 -7.29230762e-01 -5.82163572e-01
-3.72150570e-01 3.86859626e-01 -3.53059429e-03 8.72021019e-01
2.89447814e-01 -1.63431430e+00 -1.81789517e+00 2.70842165e-01
4.43396300e-01 -2.37123594e-02 -2.88114101e-01 3.93276960e-02
-3.94689649e-01 4.86038029e-01 -4.09432054e-01 4.19775993e-02
-6.05447292e-01 8.84355307e-02 7.47131631e-02 -3.14152390e-01
-2.09812433e-01 6.39587820e-01 -5.84490895e-01 -9.82811213e-01
-1.42048359e-01 -2.12840199e-01 -2.48963550e-01 4.14957404e-01
4.98276711e-01 2.61937499e-01 2.07024347e-02 -9.68145072e-01
-1.46751408e-03 -1.97467551e-01 -2.15037726e-02 -7.63549387e-01
1.19568753e+00 2.23695170e-02 -4.91409078e-02 9.26531971e-01
9.34679210e-01 1.09623551e-01 -5.51986396e-01 -3.33036661e-01
2.61320710e-01 9.06119123e-02 -3.02932948e-01 -1.18362832e+00
-7.82864869e-01 6.47712171e-01 -1.33061096e-01 1.33971140e-01
5.23207843e-01 6.13030605e-02 1.14846790e+00 5.47146142e-01
2.28314206e-01 -1.57849717e+00 -3.71459574e-01 1.22800815e+00
1.50023210e+00 -1.12270391e+00 -2.14066103e-01 1.69576511e-01
-1.14718807e+00 8.59247506e-01 6.21805370e-01 2.27299616e-01
3.05945992e-01 5.52320361e-01 4.77427125e-01 8.97491723e-02
-1.08026540e+00 9.94000304e-03 2.53888398e-01 7.29872108e-01
9.91406560e-01 3.21735263e-01 -6.05702579e-01 6.46470368e-01
-1.14440989e+00 -1.69540375e-01 4.88832355e-01 6.07208908e-01
1.69765890e-01 -1.26417410e+00 6.60033077e-02 4.94142950e-01
-6.53095722e-01 -5.13351381e-01 -9.12481427e-01 9.00022089e-01
1.41826227e-01 9.80181932e-01 2.90154874e-01 -7.96277821e-02
6.43414021e-01 2.76373446e-01 -1.29635120e-02 -6.90312803e-01
-8.26582313e-01 -4.13861305e-01 9.96706069e-01 -3.19730639e-01
-2.25419328e-01 -8.51045609e-01 -1.44399071e+00 -2.07760081e-01
2.94985563e-01 2.96602339e-01 6.31983936e-01 8.57043147e-01
4.01882976e-01 4.28710610e-01 4.62567568e-01 -6.27632797e-01
-4.65344489e-01 -1.41031349e+00 -4.22662765e-01 5.39516509e-01
2.25508705e-01 -1.28844976e-01 2.36810427e-02 2.58820027e-01]
|
[11.599140167236328, 9.542366027832031]
|
ebe845e5-aefc-4ae9-8972-a340f8ed2672
|
multi-view-improved-monitored-distillation
|
2303.15840
| null |
https://arxiv.org/abs/2303.15840v2
|
https://arxiv.org/pdf/2303.15840v2.pdf
|
Enhancing Depth Completion with Multi-View Monitored Distillation
|
This paper presents a novel method for depth completion, which leverages multi-view improved monitored distillation to generate more precise depth maps. Our approach builds upon the state-of-the-art ensemble distillation method, in which we introduce a stereo-based model as a teacher model to improve the accuracy of the student model for depth completion. By minimizing the reconstruction error for a given image during ensemble distillation, we can avoid learning inherent error modes of completion-based teachers. To provide self-supervised information, we also employ multi-view depth consistency and multi-scale minimum reprojection. These techniques utilize existing structural constraints to yield supervised signals for student model training, without requiring costly ground truth depth information. Our extensive experimental evaluation demonstrates that our proposed method significantly improves the accuracy of the baseline monitored distillation method.
|
['Hung-Chyun Chou', 'Ning Ding', 'Ming Ouyang', 'Chang-Zheng Zhang', 'Sen-Hua Zhu', 'Cong Li', 'Jia-Wei Guo']
|
2023-03-28
| null | null | null | null |
['depth-completion']
|
['computer-vision']
|
[ 2.23174348e-01 4.66402858e-01 -1.50566861e-01 -4.43758637e-01
-1.13836253e+00 -4.46650207e-01 6.14431918e-01 -6.53092861e-02
-2.30087101e-01 5.94194293e-01 3.96060884e-01 -1.88507438e-01
1.64763719e-01 -9.25385773e-01 -7.59935379e-01 -6.32007420e-01
5.99541187e-01 5.24105489e-01 1.80559874e-01 -5.71220331e-02
3.65270525e-01 2.92743742e-01 -1.64963341e+00 2.05309957e-01
1.47177696e+00 6.91178262e-01 2.88598359e-01 8.08673382e-01
3.06097455e-02 1.04236603e+00 -3.81075352e-01 -2.16184855e-01
3.20502788e-01 -2.01545000e-01 -6.59805357e-01 1.76841483e-01
1.12085497e+00 -9.92490709e-01 -4.33199465e-01 7.51223981e-01
6.30152583e-01 3.23495090e-01 6.66701019e-01 -7.44200110e-01
-1.93091437e-01 1.97517523e-03 -6.69110596e-01 -1.77232072e-01
4.53715682e-01 4.48860265e-02 8.23360443e-01 -1.04925323e+00
5.98463356e-01 1.09559560e+00 4.86342400e-01 6.90289736e-01
-1.48911977e+00 -8.29532266e-01 2.29622319e-01 2.99675502e-02
-1.15623713e+00 -3.82155776e-01 9.44305241e-01 -5.66286802e-01
8.52923691e-01 -1.57995239e-01 8.72290075e-01 8.55657816e-01
7.57055506e-02 1.01635671e+00 1.32887661e+00 -4.83883291e-01
2.56195098e-01 1.69227626e-02 -2.85333902e-01 9.84748423e-01
-4.74165864e-02 3.17113310e-01 -7.03555703e-01 2.29499504e-01
1.14041948e+00 -2.12640837e-01 -3.02585661e-01 -7.97028303e-01
-8.55521917e-01 6.47000790e-01 5.11435628e-01 -4.14709836e-01
5.03576100e-02 1.39466703e-01 -9.29763168e-02 -5.92819080e-02
9.77604330e-01 4.21350330e-01 -4.47482258e-01 -2.77440310e-01
-1.17841208e+00 3.54568958e-01 5.35712719e-01 7.72536039e-01
1.28600502e+00 1.00350538e-02 1.73477784e-01 6.24531746e-01
5.08298278e-01 3.94845456e-01 2.12733522e-01 -1.49926841e+00
7.14748025e-01 7.51809776e-01 -1.72631606e-01 -6.84827328e-01
-6.55084616e-03 -3.18805724e-01 -4.56040591e-01 6.93126678e-01
2.88290292e-01 -9.56794024e-02 -1.11938572e+00 1.46261358e+00
7.74004400e-01 6.74023390e-01 1.14584059e-01 6.98381305e-01
7.42074549e-01 3.43423754e-01 -3.83764952e-01 2.20081732e-01
6.54935837e-01 -1.26573229e+00 -3.13007981e-01 -2.43700966e-01
8.84121299e-01 -5.23239017e-01 9.24854457e-01 7.94397831e-01
-1.23464012e+00 -5.75105131e-01 -1.19868946e+00 -4.37247097e-01
2.31101736e-01 -4.76561040e-02 6.46551549e-01 4.33338225e-01
-1.08350515e+00 8.55407178e-01 -1.22234619e+00 8.55596438e-02
3.64159256e-01 2.19615772e-01 -3.16032827e-01 -2.81758279e-01
-4.89693910e-01 7.06554353e-01 1.34361610e-01 -3.25578272e-01
-1.35628569e+00 -1.11993861e+00 -1.11279404e+00 -3.53397131e-01
7.42579699e-02 -9.90378022e-01 1.47821116e+00 -5.24902761e-01
-1.81660271e+00 9.05071735e-01 -3.20635676e-01 -7.19585046e-02
6.68337226e-01 -5.36313295e-01 3.65122497e-01 4.04557705e-01
6.64509535e-02 1.15535188e+00 6.53771579e-01 -1.62817204e+00
-6.86292291e-01 -5.30386090e-01 6.72432855e-02 7.91841209e-01
-3.17415297e-01 -8.65243793e-01 -4.68311697e-01 -5.17572165e-01
7.57862151e-01 -8.23935449e-01 -4.99627382e-01 2.03915924e-01
-2.14688495e-01 1.90113544e-01 8.44901264e-01 -5.72234631e-01
9.78236973e-01 -1.74749029e+00 4.03057575e-01 9.92900208e-02
5.58881760e-01 -3.65597397e-01 -5.43786548e-02 2.13634446e-01
1.05213642e-01 -2.58532554e-01 -2.81157255e-01 -1.16333544e+00
-3.61307591e-01 4.71179634e-01 -2.85827577e-01 2.95998424e-01
-1.42653123e-01 5.58086872e-01 -1.05268776e+00 -5.98136902e-01
6.74558520e-01 6.00538790e-01 -1.13558495e+00 6.05048180e-01
-2.21597269e-01 1.15260506e+00 -2.08275184e-01 4.78320062e-01
8.06647658e-01 -7.74140954e-02 -2.28290614e-02 -6.95820302e-02
-1.62985623e-01 6.55366182e-01 -1.18165123e+00 2.53163075e+00
-7.98275650e-01 4.21874553e-01 1.29834279e-01 -7.65765190e-01
8.86179566e-01 1.87355086e-01 6.60169363e-01 -3.79279196e-01
-2.59717971e-01 1.84541076e-01 -4.67331260e-01 -3.29268605e-01
7.28432536e-01 -5.55951595e-02 3.53257120e-01 5.13530195e-01
5.47604822e-02 -1.06792188e+00 -2.88733065e-01 4.46417809e-01
9.11853492e-01 7.95640767e-01 -7.96091780e-02 -5.78733124e-02
2.37133548e-01 -1.82720333e-01 6.27212346e-01 3.92394215e-01
-2.38742549e-02 1.05427575e+00 4.05385718e-02 -1.64651632e-01
-9.64445651e-01 -1.33331275e+00 -1.75200224e-01 6.84682429e-01
2.13636190e-01 -4.28662241e-01 -8.56574237e-01 -7.84289300e-01
-1.83057878e-02 5.70060194e-01 -4.96408820e-01 1.86524898e-01
-4.64424729e-01 -2.62206852e-01 2.40726382e-01 7.02604949e-01
7.59224594e-01 -3.28637540e-01 -4.86041099e-01 2.74223071e-02
-4.04842675e-01 -1.08671415e+00 -3.55599314e-01 1.07622147e-01
-1.55817842e+00 -1.09710979e+00 -6.04127347e-01 -6.80475652e-01
9.52589929e-01 5.66180706e-01 1.23408425e+00 -8.04675370e-02
2.69284453e-02 7.07666039e-01 3.60807516e-02 -2.65271038e-01
-2.97994077e-01 7.42217153e-02 -2.43350938e-02 -4.48989868e-01
4.19289842e-02 -1.12756777e+00 -9.64303553e-01 1.96780011e-01
-6.62785828e-01 5.80395877e-01 2.50000626e-01 7.06756949e-01
8.26553345e-01 -2.12174207e-01 1.94692239e-02 -8.54869425e-01
1.32830590e-01 -9.62977111e-02 -6.68550074e-01 -1.44221321e-01
-1.00848842e+00 2.70485580e-01 2.95901835e-01 -2.95715928e-01
-1.31674731e+00 2.29239017e-01 -2.25517005e-01 -7.35516489e-01
-8.62744451e-02 2.22852513e-01 -7.37801045e-02 -4.44860160e-01
6.60327613e-01 2.29566127e-01 -1.08501323e-01 -5.13323426e-01
3.79297644e-01 2.94147909e-01 7.15563834e-01 -1.02418470e+00
8.51688385e-01 7.26564586e-01 1.19145624e-01 -5.26515603e-01
-1.14472389e+00 -4.89430755e-01 -1.03701353e+00 -2.84430891e-01
8.28082681e-01 -1.62303329e+00 -5.22119582e-01 6.25753164e-01
-1.08216929e+00 -7.51808643e-01 -1.90200150e-01 5.81650734e-01
-6.86951399e-01 5.15307367e-01 -6.93745732e-01 -7.25212157e-01
-8.44760835e-02 -1.12129259e+00 1.41104424e+00 1.68421835e-01
-1.99384205e-02 -1.44076002e+00 4.55037594e-01 8.28821659e-01
2.68573761e-02 2.43792534e-01 4.57562178e-01 1.28418878e-01
-1.02124548e+00 2.32284546e-01 1.22158015e-02 3.59724522e-01
5.42353466e-02 2.09498722e-02 -1.32932007e+00 -3.80939454e-01
7.63640739e-03 -6.06735051e-01 1.04074824e+00 4.14269358e-01
1.28290629e+00 -6.01748228e-02 -2.70349950e-01 1.19913006e+00
1.45652151e+00 -1.86447963e-01 6.34431541e-01 4.38896418e-01
1.12665474e+00 6.89540684e-01 6.45818055e-01 6.81751370e-01
1.05189347e+00 4.03299421e-01 7.47012377e-01 -1.72966197e-01
-1.20659210e-01 -8.39002550e-01 2.20260844e-01 1.01935649e+00
-1.06665120e-01 6.16579577e-02 -8.61723781e-01 6.44861460e-01
-1.58101964e+00 -7.32089281e-01 6.61974698e-02 2.17361498e+00
9.37245190e-01 8.60847346e-03 -2.01500818e-01 1.52833596e-01
2.25239303e-02 2.70116895e-01 -5.95772684e-01 -3.88308465e-02
3.54795396e-01 4.86961216e-01 4.14193153e-01 1.02268505e+00
-8.16573203e-01 1.03522646e+00 6.68875837e+00 3.97194326e-01
-8.58686626e-01 -2.49126628e-02 5.53509891e-01 -1.88918501e-01
-8.66634548e-01 1.39961243e-01 -8.71367216e-01 2.84861960e-02
4.28634912e-01 2.32260153e-01 4.58311319e-01 9.08991992e-01
4.36970562e-01 -5.80102503e-01 -1.38412285e+00 1.06632304e+00
2.11829364e-01 -1.32441926e+00 -1.15292780e-01 2.35485956e-01
1.51075578e+00 -5.68867363e-02 2.08617464e-01 2.36443356e-01
8.46366465e-01 -8.05149257e-01 4.58380759e-01 2.91321129e-01
7.30220079e-01 -7.47058988e-01 1.12887658e-01 5.77321708e-01
-1.23352373e+00 2.28888407e-01 -1.22639604e-01 -2.41345331e-01
1.21268325e-01 8.96912694e-01 -8.37548673e-01 6.06185794e-01
5.23710072e-01 1.08891594e+00 -4.34824646e-01 6.80834711e-01
-8.77550304e-01 4.66437072e-01 -3.50651860e-01 6.15483642e-01
3.83345634e-02 -3.78302336e-01 4.40002084e-01 6.75283968e-01
3.59363347e-01 1.97792456e-01 3.77566963e-01 8.24207544e-01
-5.14987446e-02 -2.57448137e-01 -7.75537908e-01 6.73164666e-01
5.81231952e-01 1.13402724e+00 -1.46512836e-01 -3.22952896e-01
-2.57807046e-01 1.20164466e+00 5.51300168e-01 3.59510154e-01
-5.90854168e-01 1.22558281e-01 8.50734711e-01 1.71481609e-01
2.14506350e-02 -5.34457088e-01 -7.96146631e-01 -1.52572608e+00
-1.00739442e-01 -6.61606014e-01 2.36761957e-01 -9.79957700e-01
-8.34324598e-01 1.36614844e-01 4.76285368e-02 -1.35964406e+00
-4.96435404e-01 -3.57651651e-01 -8.33347201e-01 6.87382698e-01
-1.80664861e+00 -1.26397347e+00 -9.35996532e-01 5.01770318e-01
5.70063651e-01 1.67042688e-01 6.96187556e-01 1.18944950e-01
-5.26885211e-01 5.64012349e-01 4.58551198e-02 -1.55674323e-01
9.77785110e-01 -1.60451174e+00 2.92213112e-01 8.61650646e-01
1.30849555e-01 3.58152837e-01 5.92426836e-01 -6.85031533e-01
-1.10433650e+00 -1.05876219e+00 4.71927762e-01 -7.09717989e-01
3.25250149e-01 -1.46885976e-01 -6.71067894e-01 9.19672430e-01
-6.17290288e-02 3.80921997e-02 5.26586115e-01 1.45496815e-01
-3.83483022e-01 -7.79642835e-02 -1.22065902e+00 6.29791379e-01
1.10688102e+00 -8.10319722e-01 -4.24245358e-01 6.98419809e-02
7.14977622e-01 -1.05775225e+00 -9.40002441e-01 4.91140932e-01
6.14296257e-01 -1.39317870e+00 9.91423547e-01 2.02670246e-02
1.02885723e+00 -2.00933784e-01 -9.45998430e-02 -1.62285089e+00
8.00692961e-02 -3.59715641e-01 -5.54480135e-01 9.97855663e-01
8.40707347e-02 -4.06586558e-01 1.62909663e+00 7.21615911e-01
-5.31783223e-01 -8.90117586e-01 -8.19637597e-01 -6.23431683e-01
2.83825964e-01 -5.43842971e-01 3.55155349e-01 9.78648245e-01
-7.88817331e-02 1.55126020e-01 -5.72730862e-02 3.09650272e-01
1.08960688e+00 3.17774266e-02 1.30927563e+00 -1.12359262e+00
-7.04512715e-01 -1.02160890e-02 -1.85152307e-01 -1.97537971e+00
2.77409434e-01 -7.67555177e-01 5.12781069e-02 -1.79604411e+00
1.53351083e-01 -6.45400524e-01 2.25013122e-01 3.08126271e-01
-2.72592336e-01 2.38041535e-01 -8.46573636e-02 1.35743126e-01
-2.79800862e-01 9.60355937e-01 1.78989148e+00 1.09978684e-03
-4.68732744e-01 -1.56468615e-01 -5.51135719e-01 8.66412640e-01
4.23253357e-01 -3.69189531e-01 -9.16323304e-01 -8.99978340e-01
-4.62073274e-02 2.37485856e-01 2.92351902e-01 -1.20676816e+00
2.49036118e-01 -3.32335085e-01 3.92355025e-01 -8.53482485e-01
7.45052516e-01 -6.69811249e-01 -3.24214906e-01 3.11755478e-01
-1.48831353e-01 -9.09980610e-02 1.48963392e-01 5.21514714e-01
-1.55826434e-01 -6.45369738e-02 8.35846126e-01 -1.05039746e-01
-5.93276680e-01 4.35800225e-01 6.40330687e-02 1.26295209e-01
9.59724545e-01 -4.70883757e-01 -1.73365802e-01 -6.88080311e-01
-5.79886258e-01 3.74358386e-01 8.40920806e-01 -5.28869703e-02
8.57476771e-01 -1.26871669e+00 -6.28251493e-01 3.14036071e-01
8.27534497e-02 1.06868875e+00 2.12225944e-01 6.16747797e-01
-6.38254642e-01 -3.04553248e-02 8.29149261e-02 -9.48092937e-01
-1.16511178e+00 1.30385812e-02 5.14705956e-01 -5.16884804e-01
-6.60122454e-01 1.17799580e+00 4.68551040e-01 -8.14856470e-01
2.10179836e-01 -4.18569535e-01 2.08646923e-01 -3.11612099e-01
4.23160940e-01 4.53753740e-01 6.24190345e-02 1.06438017e-02
6.14808649e-02 7.93160498e-01 -9.77799222e-02 -5.66535473e-01
1.35269880e+00 -2.94153363e-01 2.60128170e-01 3.67687196e-01
1.04539359e+00 1.54656723e-01 -2.04912686e+00 -9.09032971e-02
-5.53437889e-01 -7.56790221e-01 4.92742687e-01 -5.68047881e-01
-1.01293623e+00 1.15617132e+00 3.17906678e-01 -7.33983696e-01
1.03299761e+00 -3.22110206e-01 8.67833793e-01 2.39420965e-01
4.98170286e-01 -1.07623911e+00 7.16568232e-01 6.28744841e-01
4.23253208e-01 -1.59109986e+00 3.45620543e-01 -6.54381692e-01
-3.85211110e-01 9.97432888e-01 1.36144924e+00 -9.64751840e-02
4.71224219e-01 4.11542952e-01 6.61027804e-02 -2.74202526e-02
-8.66723418e-01 2.92420536e-02 3.33498001e-01 7.06909239e-01
3.90248328e-01 -3.42607856e-01 3.01318079e-01 -2.91165747e-02
-4.26640362e-01 -2.29830861e-01 5.65141499e-01 9.65485513e-01
-5.02131164e-01 -1.12757623e+00 -3.62207413e-01 1.31892666e-01
7.51355514e-02 -1.73345596e-01 -2.43313581e-01 5.35671711e-01
9.42523312e-03 8.16859126e-01 2.60883301e-01 -6.89031184e-01
2.04386652e-01 -1.43478587e-01 1.03334796e+00 -9.19932365e-01
-3.53914827e-01 1.61159951e-02 -1.22613005e-01 -6.83371902e-01
-4.23449099e-01 -5.15019417e-01 -1.40871274e+00 -4.43467051e-01
-2.80252606e-01 -1.60222635e-01 6.38724625e-01 9.14333165e-01
3.08455735e-01 4.51746702e-01 9.49733138e-01 -1.25067592e+00
-4.85172421e-01 -8.49136651e-01 -2.68879771e-01 2.78233945e-01
4.24657077e-01 -7.00064659e-01 -5.33419430e-01 3.32488604e-02]
|
[8.680563926696777, -2.821828842163086]
|
3515365f-9f2e-451c-8ee0-d8f23c058305
|
unsupervised-neural-aspect-search-with
|
2005.02771
| null |
https://arxiv.org/abs/2005.02771v1
|
https://arxiv.org/pdf/2005.02771v1.pdf
|
Unsupervised Neural Aspect Search with Related Terms Extraction
|
The tasks of aspect identification and term extraction remain challenging in natural language processing. While supervised methods achieve high scores, it is hard to use them in real-world applications due to the lack of labelled datasets. Unsupervised approaches outperform these methods on several tasks, but it is still a challenge to extract both an aspect and a corresponding term, particularly in the multi-aspect setting. In this work, we present a novel unsupervised neural network with convolutional multi-attention mechanism, that allows extracting pairs (aspect, term) simultaneously, and demonstrate the effectiveness on the real-world dataset. We apply a special loss aimed to improve the quality of multi-aspect extraction. The experimental study demonstrates, what with this loss we increase the precision not only on this joint setting but also on aspect prediction only.
|
['Maria Khodorchenko', 'Timur Sokhin', 'Nikolay Butakov']
|
2020-05-06
| null | null | null | null |
['aspect-extraction']
|
['natural-language-processing']
|
[ 2.88406372e-01 3.55561152e-02 -3.41339350e-01 -2.91514099e-01
-1.07122970e+00 -5.54658532e-01 7.10558832e-01 2.85922140e-01
-5.16891718e-01 3.62453222e-01 1.43688545e-01 -1.01281069e-01
-1.57510370e-01 -7.95308173e-01 -5.82155764e-01 -5.30981839e-01
1.21176355e-01 5.80911934e-01 2.41304152e-02 -2.04306513e-01
-3.35874408e-02 7.88265094e-02 -1.49089015e+00 2.40709811e-01
8.33268166e-01 9.50086594e-01 1.31421000e-01 2.07010195e-01
-3.08289826e-01 4.37121242e-01 -5.49071014e-01 -7.68566787e-01
2.57347554e-01 -1.40747696e-01 -6.06224537e-01 2.05168426e-01
1.35016248e-01 -1.57850370e-01 5.86377084e-02 9.76201296e-01
5.51825881e-01 -2.79680751e-02 7.65607297e-01 -1.14569902e+00
-5.89423656e-01 8.21960449e-01 -6.96179807e-01 -2.10747972e-01
1.84746265e-01 -4.38043661e-02 1.64841545e+00 -8.43600333e-01
3.92245412e-01 8.58041704e-01 5.97558737e-01 9.45875943e-02
-9.12221372e-01 -3.48217100e-01 3.66929144e-01 9.14982557e-02
-1.18331468e+00 -2.85751879e-01 8.50979209e-01 -1.71000063e-01
1.13876629e+00 1.40634760e-01 6.46395266e-01 9.69832420e-01
-7.25309402e-02 1.11331117e+00 9.66326535e-01 -4.84413236e-01
-1.13397457e-01 3.18232447e-01 3.42283726e-01 5.62787473e-01
2.92018265e-01 -1.49194270e-01 -4.78450924e-01 4.61287871e-02
2.27993712e-01 8.58288631e-02 -3.26828867e-01 -1.89566582e-01
-1.15496802e+00 8.10491025e-01 2.12699667e-01 6.13210797e-01
-5.09245932e-01 -1.01634644e-01 3.71162802e-01 2.53218740e-01
8.61335456e-01 5.97032785e-01 -7.45160639e-01 -1.39799461e-01
-9.92238700e-01 2.29677171e-01 9.14691389e-01 9.82668638e-01
7.71982491e-01 -2.30412319e-01 -4.05966014e-01 9.87550974e-01
7.29553699e-02 5.16503334e-01 3.81740868e-01 -2.65161574e-01
5.91412485e-01 1.08693564e+00 -1.72831491e-01 -8.70160162e-01
-5.74353456e-01 -1.00604498e+00 -1.00847197e+00 -1.01975640e-02
1.70631573e-01 -1.53371871e-01 -7.88645029e-01 1.65224838e+00
3.70725572e-01 2.25166732e-04 -2.00707152e-01 6.96026325e-01
8.88456225e-01 5.82310617e-01 -8.40846542e-03 -2.49315873e-01
1.73417163e+00 -1.35092497e+00 -8.71804833e-01 -5.84789634e-01
5.89333415e-01 -8.67693365e-01 1.06121206e+00 3.64462972e-01
-7.44912863e-01 -2.56897986e-01 -9.34126556e-01 -1.51547343e-01
-5.24393559e-01 5.85511982e-01 8.91439676e-01 3.86112124e-01
-7.13653922e-01 2.94916332e-01 -6.69373512e-01 -1.93082467e-01
2.90305406e-01 3.23803633e-01 -4.90928113e-01 -1.64153233e-01
-1.10351717e+00 7.63340831e-01 3.43984067e-01 1.31444201e-01
-4.81905043e-01 -7.08581567e-01 -1.06287611e+00 4.71285284e-01
6.91336930e-01 -8.49787831e-01 1.12288773e+00 -8.87902260e-01
-1.06869876e+00 8.41670871e-01 -2.35735521e-01 -4.63703692e-01
3.30565184e-01 -4.54845488e-01 -2.78782457e-01 -5.84303141e-02
1.81993067e-01 3.52347225e-01 8.17865133e-01 -1.23124635e+00
-6.78034544e-01 -4.40209776e-01 5.38126588e-01 1.70370966e-01
-7.58975089e-01 1.61404952e-01 -8.46813858e-01 -7.59553432e-01
-1.98807776e-01 -8.99879754e-01 -3.40895981e-01 -9.87502113e-02
-5.76942801e-01 -3.36918831e-01 6.06931448e-01 -5.70553720e-01
1.13188910e+00 -2.13100266e+00 1.82994641e-02 -5.00888526e-02
1.50413290e-01 4.65841472e-01 -2.00739026e-01 3.92848760e-01
6.92929104e-02 2.09499635e-02 -6.13715768e-01 -8.18774641e-01
1.56167015e-01 2.39009485e-02 -2.10428402e-01 6.94975927e-02
5.76519370e-01 9.79746580e-01 -6.84220493e-01 -3.85286927e-01
-1.16082542e-01 6.08848572e-01 -6.22263372e-01 2.41020203e-01
-3.67976457e-01 1.47021905e-01 -3.82288367e-01 5.68946719e-01
6.77374542e-01 -4.13632840e-01 2.99932193e-02 -2.37973005e-01
2.61630058e-01 3.92871380e-01 -1.06011343e+00 1.65702188e+00
-9.81570065e-01 5.56288421e-01 -6.44628406e-02 -9.56058919e-01
8.53731394e-01 4.12735164e-01 4.82315838e-01 -5.96484125e-01
5.58917671e-02 4.94036265e-02 -7.20987767e-02 -5.77819169e-01
6.63587511e-01 -2.89075524e-01 -1.40301764e-01 6.89929068e-01
2.48051122e-01 -1.68232378e-02 5.03661275e-01 6.19704984e-02
1.04512513e+00 -1.76540501e-02 4.61930186e-01 -8.10694471e-02
5.48320889e-01 -1.71112612e-01 5.06820321e-01 5.03840268e-01
1.04651317e-01 8.08220387e-01 7.26253450e-01 -2.23939002e-01
-8.18411350e-01 -4.69865829e-01 8.09423346e-03 1.11947048e+00
5.21312617e-02 -7.85619736e-01 -5.73967934e-01 -1.04608428e+00
-2.19859883e-01 7.80078053e-01 -5.55115461e-01 -2.72872478e-01
-4.74863261e-01 -1.10954785e+00 2.80368626e-01 4.27289546e-01
5.19594252e-01 -1.09292722e+00 -2.32420698e-01 2.21524760e-01
-4.17078435e-01 -1.57520962e+00 -5.02026975e-01 2.09771171e-01
-4.39267695e-01 -9.43493903e-01 -5.20896912e-01 -7.39553809e-01
4.28117156e-01 2.06888288e-01 1.37031865e+00 2.36988306e-01
-7.73117319e-02 1.06152631e-01 -4.54073906e-01 -4.19838786e-01
-1.01218008e-01 6.45664632e-01 -5.13627589e-01 2.75015205e-01
6.44042075e-01 -5.72560310e-01 -4.76838559e-01 1.32529825e-01
-1.10726285e+00 -7.22912177e-02 9.50024962e-01 8.32565486e-01
4.70385253e-01 1.51399091e-01 4.79987770e-01 -1.11449194e+00
4.68999267e-01 -3.51754934e-01 -5.11631966e-01 2.30327487e-01
-8.58159900e-01 2.28865147e-01 6.76686168e-01 -2.51741856e-01
-9.51880336e-01 -2.76431069e-02 -4.80687767e-01 3.28697450e-02
-1.30562708e-01 7.98972785e-01 -3.80539536e-01 4.27331477e-01
2.03098491e-01 4.12799448e-01 -2.70830840e-01 -6.87867403e-01
1.97084323e-01 6.95856392e-01 2.37058718e-02 -2.23517552e-01
8.16862464e-01 4.25477922e-01 -8.52376670e-02 -7.84867942e-01
-1.38541996e+00 -6.26577199e-01 -4.62589204e-01 2.64917612e-01
6.57977521e-01 -1.15374601e+00 -2.84009665e-01 2.29665756e-01
-1.23963797e+00 1.14577971e-01 -2.61722803e-01 3.76600623e-01
-2.86230117e-01 4.04619575e-01 -3.78971398e-01 -7.14313626e-01
-7.12919235e-01 -1.26987779e+00 1.51161325e+00 -2.76625007e-02
7.58929504e-03 -9.09830034e-01 8.33530053e-02 4.97957796e-01
4.19321865e-01 -9.93054882e-02 9.88582671e-01 -8.77993643e-01
-7.06254601e-01 -3.75919670e-01 -3.39448184e-01 3.92475367e-01
2.19808042e-01 -9.82734114e-02 -1.11714327e+00 -8.42600968e-03
1.43377837e-02 -7.40094855e-02 1.16087282e+00 2.19312701e-02
9.07456577e-01 -2.31973290e-01 -1.64949998e-01 4.52978879e-01
1.43471706e+00 -2.92750001e-01 4.10497338e-01 3.77734691e-01
7.57142961e-01 6.23722136e-01 6.70437038e-01 1.93355531e-01
5.59702218e-01 9.28170323e-01 3.10383230e-01 -1.42099231e-01
-2.23863751e-01 4.61959001e-03 3.85473847e-01 1.15958261e+00
8.37051868e-02 -4.30474401e-01 -7.94394314e-01 9.76217330e-01
-1.82883465e+00 -5.87943077e-01 -5.09428792e-02 2.13503909e+00
8.75112534e-01 3.13073039e-01 9.07893665e-03 2.18084142e-01
4.71656710e-01 3.20016801e-01 -8.72045010e-02 -1.05103411e-01
-2.74078280e-01 1.90603182e-01 -1.80531535e-02 4.19471294e-01
-1.35220027e+00 7.30063915e-01 5.26458836e+00 1.06872714e+00
-9.45003450e-01 2.86043584e-01 5.12505710e-01 3.32433358e-03
-5.44541955e-01 -1.83652360e-02 -9.85265493e-01 2.79314101e-01
8.05185258e-01 9.21428949e-03 6.46861568e-02 8.56618166e-01
-9.24660712e-02 9.79645997e-02 -1.08717418e+00 8.23154390e-01
3.20831120e-01 -8.15761149e-01 1.53391942e-01 -1.23113379e-01
5.64262033e-01 -9.28004161e-02 -1.05531834e-01 6.65561318e-01
-1.61595061e-01 -8.30594599e-01 3.44414324e-01 1.52614504e-01
4.66660619e-01 -8.12886298e-01 1.03428304e+00 4.49133724e-01
-1.24771166e+00 1.49737775e-01 -2.09699601e-01 1.90838143e-01
2.86824256e-01 1.26167357e+00 -6.15257144e-01 8.24552596e-01
3.82183999e-01 7.73897886e-01 -7.23333895e-01 1.22600234e+00
-5.62148690e-01 5.56040466e-01 -3.86174291e-01 -8.30514431e-02
2.19803318e-01 -2.78498799e-01 7.32548177e-01 1.26021838e+00
2.55150735e-01 -5.49050748e-01 1.84339672e-01 7.00485289e-01
-2.97022760e-01 4.56029922e-01 -7.24753857e-01 -1.76119164e-01
-1.19049944e-01 1.74488962e+00 -6.64549112e-01 -3.44792426e-01
-6.03379428e-01 7.71684766e-01 5.56546867e-01 3.32856923e-01
-7.84035563e-01 -3.87150884e-01 6.04120851e-01 -3.42620462e-02
6.31797373e-01 -1.65008515e-01 -3.36461574e-01 -1.50597656e+00
5.18009424e-01 -9.56886292e-01 3.67269844e-01 -3.20024133e-01
-1.27582276e+00 9.69003201e-01 -2.17435643e-01 -1.46971917e+00
-4.69527364e-01 -5.61830759e-01 -4.90373254e-01 4.45475578e-01
-1.91443551e+00 -1.54932630e+00 -9.99971107e-02 1.88049316e-01
5.23531497e-01 -4.75811958e-03 8.48303378e-01 7.47030795e-01
-5.10071874e-01 8.83525133e-01 -5.66883199e-02 1.44682452e-01
7.20209837e-01 -1.29578102e+00 2.69048303e-01 7.40407884e-01
6.21293187e-01 5.59573472e-01 6.18400514e-01 -2.11137488e-01
-1.35475719e+00 -1.25475609e+00 1.27180743e+00 -3.97983849e-01
5.64976573e-01 -6.88002229e-01 -9.11493838e-01 6.79143310e-01
5.02643347e-01 -1.07339703e-01 6.91909432e-01 6.53916180e-01
-5.21321654e-01 -2.17583373e-01 -9.37029123e-01 6.53309703e-01
9.17046070e-01 -5.23772240e-01 -4.44945574e-01 4.63844627e-01
1.02617466e+00 -5.42201772e-02 -8.36707771e-01 6.21715128e-01
3.72822195e-01 -8.13972175e-01 8.53033245e-01 -5.23143828e-01
7.74105787e-01 -2.83801168e-01 -4.51488569e-02 -1.45970559e+00
5.52130230e-02 -3.22828740e-01 -1.88616127e-01 1.77392805e+00
8.90580356e-01 -5.95220029e-01 5.60974479e-01 2.17645094e-01
1.54986069e-01 -1.04966927e+00 -8.00831079e-01 -6.97662950e-01
-1.31519794e-01 -6.41531527e-01 7.05646515e-01 6.26030326e-01
-2.19077900e-01 9.98143077e-01 -4.59699303e-01 1.22236446e-01
4.58034664e-01 8.03149283e-01 5.85097313e-01 -8.62076402e-01
-7.96501696e-01 -4.32859451e-01 -4.53847677e-01 -9.37901676e-01
2.84614325e-01 -8.48734617e-01 9.54060853e-02 -1.52855420e+00
4.72328007e-01 -3.23840708e-01 -3.47209245e-01 5.88566720e-01
-6.00968540e-01 3.27233374e-01 1.93102151e-01 -1.67054936e-01
-8.24296594e-01 1.03602314e+00 1.00179505e+00 -5.99379659e-01
5.80757223e-02 2.62140304e-01 -8.79809201e-01 7.51965761e-01
6.72275901e-01 -6.18718207e-01 -2.57106036e-01 -5.62357605e-01
5.37557721e-01 -3.43193978e-01 3.56298164e-02 -8.73199165e-01
-3.83100696e-02 3.70510846e-01 -5.11232987e-02 -5.53791165e-01
4.04829353e-01 -1.03657115e+00 -3.04574400e-01 1.07271813e-01
-1.78706557e-01 -1.03861965e-01 1.70335874e-01 4.88142967e-01
-5.32472253e-01 -4.30744618e-01 4.26898032e-01 1.30624473e-01
-3.43828082e-01 4.04707223e-01 6.70160800e-02 1.90008909e-01
6.72451556e-01 5.84583402e-01 -2.57851988e-01 -5.72268307e-01
-5.42497098e-01 9.08273906e-02 2.07279354e-01 5.12217343e-01
3.11650038e-01 -1.32925737e+00 -7.34642386e-01 2.06180066e-01
5.22127926e-01 1.40224546e-01 6.62099198e-02 1.00421274e+00
2.58706976e-02 3.24395865e-01 3.19979697e-01 -4.80710477e-01
-1.18495655e+00 6.77600443e-01 1.73976982e-03 -1.04607940e+00
-4.82675225e-01 5.55366635e-01 4.19904917e-01 -7.97006667e-01
1.24938361e-01 -2.82663018e-01 -4.94987011e-01 2.26510331e-01
6.49125993e-01 -2.39204884e-01 5.38356662e-01 -5.15342593e-01
-2.63060838e-01 7.31790185e-01 -3.22000057e-01 8.66563246e-02
1.48380816e+00 -9.17247832e-02 -2.62386203e-01 3.34177881e-01
1.37323487e+00 1.59105182e-01 -7.90327668e-01 -4.81101841e-01
1.57787934e-01 -2.37418711e-01 -2.52876077e-02 -6.37497246e-01
-1.35754454e+00 9.43715930e-01 2.89226711e-01 3.91133487e-01
1.21255684e+00 9.34134722e-02 1.03928018e+00 5.75107038e-01
1.40077814e-01 -6.91166580e-01 -2.68739581e-01 6.05439663e-01
8.46832156e-01 -1.49030256e+00 5.93344159e-02 -6.08064592e-01
-5.02980351e-01 6.91295326e-01 3.56810778e-01 -1.74082685e-02
8.14970255e-01 4.37041283e-01 4.60648835e-02 -3.27909112e-01
-9.52694356e-01 -6.54799998e-01 5.21886766e-01 2.53226221e-01
5.82776666e-01 -7.19404668e-02 -4.96777892e-01 9.00518477e-01
-1.40110299e-01 -1.66137889e-01 3.23750556e-01 7.84529567e-01
-1.00172184e-01 -1.36062062e+00 4.53192964e-02 7.56025791e-01
-8.50151360e-01 -3.98886949e-01 -1.73205331e-01 5.80386519e-01
7.40205927e-04 9.45564866e-01 -2.11503744e-01 -1.67903513e-01
4.16721106e-01 1.23458588e-02 1.48880243e-01 -6.89512372e-01
-9.05846536e-01 2.25400954e-01 4.01874751e-01 -3.82400841e-01
-4.71649468e-01 -4.23955321e-01 -7.70737052e-01 9.92611870e-02
-5.99566281e-01 2.05777466e-01 7.89021671e-01 1.26537335e+00
4.93311733e-01 6.85538232e-01 6.20171905e-01 -4.10298318e-01
-4.46050435e-01 -1.19725394e+00 -5.83134055e-01 7.13245630e-01
3.56878638e-01 -5.93999982e-01 -3.55737925e-01 -2.44095046e-02]
|
[11.476030349731445, 6.648472309112549]
|
2ef13e31-1cb9-46f6-80d1-a235f4bbc365
|
toward-neural-network-simulation-of
|
2211.02929
| null |
https://arxiv.org/abs/2211.02929v1
|
https://arxiv.org/pdf/2211.02929v1.pdf
|
Toward Neural Network Simulation of Variational Quantum Algorithms
|
Variational quantum algorithms (VQAs) utilize a hybrid quantum-classical architecture to recast problems of high-dimensional linear algebra as ones of stochastic optimization. Despite the promise of leveraging near- to intermediate-term quantum resources to accelerate this task, the computational advantage of VQAs over wholly classical algorithms has not been firmly established. For instance, while the variational quantum eigensolver (VQE) has been developed to approximate low-lying eigenmodes of high-dimensional sparse linear operators, analogous classical optimization algorithms exist in the variational Monte Carlo (VMC) literature, utilizing neural networks in place of quantum circuits to represent quantum states. In this paper we ask if classical stochastic optimization algorithms can be constructed paralleling other VQAs, focusing on the example of the variational quantum linear solver (VQLS). We find that such a construction can be applied to the VQLS, yielding a paradigm that could theoretically extend to other VQAs of similar form.
|
['Shravan Veerapaneni', 'James Stokes', 'Oliver Knitter']
|
2022-11-05
| null | null | null | null |
['neural-network-simulation', 'variational-monte-carlo']
|
['computer-code', 'miscellaneous']
|
[ 8.01972598e-02 1.00745797e-01 3.81322920e-01 -1.44800857e-01
-1.03290880e+00 -5.82773328e-01 5.47560751e-01 -1.71133175e-01
-4.16355789e-01 9.10081625e-01 -2.02807151e-02 -5.58906376e-01
-2.95561135e-01 -1.00742698e+00 -5.82207322e-01 -1.11113071e+00
1.29921120e-02 7.04656482e-01 -1.97961152e-01 -6.50397897e-01
5.91347933e-01 4.16029751e-01 -1.23747313e+00 2.32776217e-02
9.75018322e-01 5.57113349e-01 -2.69076020e-01 5.72426021e-01
9.96685866e-03 3.93800855e-01 2.04256997e-02 -4.43702787e-01
4.33847100e-01 -7.57579863e-01 -1.04436374e+00 -4.53063667e-01
3.03414881e-01 4.60000560e-02 -6.81842446e-01 1.32475328e+00
4.52906549e-01 5.39073586e-01 7.20380068e-01 -8.26628745e-01
-5.67202806e-01 4.32087809e-01 7.15511618e-03 1.01911470e-01
2.99331725e-01 4.63258713e-01 1.40270269e+00 -5.53206563e-01
8.58632982e-01 1.06685495e+00 6.00972950e-01 7.14817762e-01
-1.90439749e+00 -3.05202872e-01 -7.87041843e-01 3.94148409e-01
-1.83913338e+00 -4.42873746e-01 4.90210354e-01 -3.85726064e-01
1.23490334e+00 9.93161649e-02 7.78307676e-01 7.74354160e-01
4.26305562e-01 4.49305534e-01 1.17287290e+00 -6.01824462e-01
6.12083435e-01 9.67417210e-02 1.51038870e-01 9.98927474e-01
1.22978918e-01 6.60838604e-01 -8.38932931e-01 -3.81696880e-01
3.24270636e-01 -4.53464031e-01 -1.23728439e-01 -6.64605677e-01
-1.23516834e+00 1.33390796e+00 6.09931469e-01 2.08992526e-01
-4.49078143e-01 4.48883444e-01 4.57960159e-01 1.98686615e-01
-8.85197222e-02 9.30438340e-01 -8.13845247e-02 -2.68452168e-01
-1.21977615e+00 6.59566164e-01 1.04360497e+00 5.13877392e-01
1.28487825e+00 2.37594381e-01 -2.55174905e-01 -1.17195129e-01
2.11336687e-01 5.19730568e-01 -2.54427698e-02 -1.43693137e+00
-1.37923164e-02 3.01197954e-02 1.71830967e-01 -4.91320848e-01
-3.35321039e-01 -9.18246135e-02 -8.69068086e-01 2.21598536e-01
3.69850904e-01 -1.48305282e-01 -5.37233710e-01 1.63605368e+00
2.29201138e-01 1.50785401e-01 2.96374053e-01 1.10613573e+00
4.04308617e-01 8.41200411e-01 -2.45501980e-01 -2.95967996e-01
9.00744617e-01 -4.30419028e-01 -4.34018165e-01 1.99166283e-01
8.49365592e-01 -5.25771320e-01 7.37054050e-01 4.14897501e-01
-1.02288306e+00 -1.36144301e-02 -1.10879827e+00 -2.92128772e-01
-3.13091129e-01 -5.47582746e-01 1.42996430e+00 1.06816828e+00
-1.22928798e+00 1.19272149e+00 -1.07372701e+00 -2.29787454e-01
2.37523392e-01 5.43251336e-01 -1.85202420e-01 1.92713723e-01
-1.38769937e+00 1.12576735e+00 5.59821904e-01 3.75753880e-01
-1.00171471e+00 -6.12505913e-01 -4.87840056e-01 -4.93509546e-02
2.11227000e-01 -1.14467633e+00 1.08652401e+00 -6.21915758e-01
-2.03084707e+00 7.84163892e-01 -4.46769536e-01 -5.31075120e-01
-8.45159590e-02 5.32247782e-01 1.85527839e-02 9.73194018e-02
-1.17211249e-02 2.63739526e-01 5.90500057e-01 -5.82297444e-01
1.65753320e-01 -3.96490186e-01 3.40640724e-01 4.89548407e-02
2.66248465e-01 -2.38142431e-01 3.05041254e-01 2.30319977e-01
5.39155543e-01 -1.34926534e+00 -7.51335323e-01 -4.98125017e-01
-3.91606420e-01 -1.58214122e-01 1.09418549e-01 -2.58094445e-02
1.10946834e+00 -1.63596952e+00 9.90723789e-01 4.67767030e-01
1.90790012e-01 1.26831114e-01 1.36533499e-01 8.30498576e-01
-2.01994479e-02 -7.72691071e-02 -4.22432989e-01 -8.32980722e-02
4.96656507e-01 4.96117085e-01 -2.97468275e-01 6.64320946e-01
2.05024675e-01 1.26706171e+00 -8.48596692e-01 -4.34661001e-01
1.51621804e-01 4.16076660e-01 -1.40578723e+00 -2.57965058e-01
-2.51551062e-01 5.16392469e-01 -7.79571295e-01 3.06334853e-01
5.62823594e-01 -4.12771851e-01 2.51137793e-01 -3.92714858e-01
-4.02753383e-01 4.43481326e-01 -1.37438083e+00 2.12918639e+00
-3.36854309e-01 6.13068223e-01 3.17567199e-01 -1.28075421e+00
1.87121212e-01 1.44391119e-01 5.50146937e-01 -5.72754800e-01
1.17714308e-01 5.00386298e-01 3.02362680e-01 -3.14733058e-01
1.02338433e+00 -8.07347894e-01 -2.19425127e-01 4.89882946e-01
3.45029861e-01 -7.04796195e-01 4.23734546e-01 6.86839700e-01
1.03656757e+00 4.06593412e-01 1.69075847e-01 -4.81820881e-01
7.19256759e-01 4.52372521e-01 1.56294823e-01 1.14037657e+00
-2.99892098e-01 3.74031514e-01 4.41972256e-01 -2.15004727e-01
-1.27219260e+00 -1.32792604e+00 -6.72006905e-01 7.16511786e-01
7.46770203e-02 -9.57427382e-01 -9.25429702e-01 2.57179022e-01
-1.26865819e-01 9.13530111e-01 -2.39121199e-01 -1.75770164e-01
-3.13022703e-01 -1.13370144e+00 5.18076956e-01 -1.94491550e-01
7.93732926e-02 -7.37397611e-01 -4.00749952e-01 3.54379267e-01
2.14134842e-01 -7.67454624e-01 7.86832049e-02 5.31243205e-01
-8.94139946e-01 -7.56983101e-01 -2.58106202e-01 7.06176320e-03
5.98319955e-02 -2.41314843e-01 1.12122560e+00 -4.02933776e-01
-4.80068117e-01 4.52203572e-01 3.92852798e-02 6.31294996e-02
-7.61406839e-01 1.96711361e-01 5.21838665e-01 -2.74656326e-01
5.86978972e-01 -7.53996730e-01 -5.30730426e-01 -4.41696286e-01
-8.87855589e-01 -5.83244255e-03 4.45459992e-01 9.98393118e-01
5.07353365e-01 -1.39240295e-01 -5.21475486e-02 -7.76325822e-01
4.60753262e-01 -3.16509515e-01 -9.60902512e-01 -1.40185738e-02
-5.68180978e-01 8.32058370e-01 6.99578762e-01 1.93317905e-01
-7.75088549e-01 1.24380000e-01 -4.54532206e-01 -1.27312928e-01
2.41128430e-01 7.40739286e-01 4.43880826e-01 -6.19494081e-01
8.74186516e-01 2.86282688e-01 -7.29827806e-02 -1.50113357e-02
7.96187162e-01 4.41802591e-01 2.43773580e-01 -8.68533731e-01
1.12787831e+00 5.76914012e-01 7.44844794e-01 -9.15307045e-01
-9.72605050e-01 -2.74297744e-01 -5.58686912e-01 1.05629571e-01
1.37377644e+00 -7.95625865e-01 -1.26801431e+00 -8.66964161e-02
-1.12589085e+00 1.42261714e-01 -3.82977903e-01 7.77525663e-01
-1.00353396e+00 3.83891404e-01 -7.28514552e-01 -1.06489801e+00
-7.81656476e-04 -1.61344779e+00 1.05614316e+00 1.95775747e-01
8.00283253e-02 -9.84677255e-01 3.51837456e-01 3.42366874e-01
5.33179343e-01 8.49531405e-03 8.51818740e-01 -5.63495196e-02
-1.21151209e+00 -1.41166642e-01 1.59598179e-02 1.68502122e-01
-7.94519603e-01 -3.38382982e-02 -8.63486588e-01 -3.86435002e-01
9.67855379e-02 -5.43349862e-01 9.25438881e-01 3.12403351e-01
6.81752741e-01 1.50579989e-01 -6.36454225e-02 8.86487842e-01
1.76481640e+00 -2.44715780e-01 7.30803072e-01 8.40922911e-03
6.31689489e-01 4.70209792e-02 -1.41305432e-01 2.87671924e-01
1.25722393e-01 8.63332868e-01 2.94656873e-01 5.12863994e-01
3.88080925e-01 7.75978416e-02 5.19182503e-01 1.11282206e+00
-4.22131419e-01 4.36462551e-01 -9.41448808e-01 1.41367903e-02
-1.62175035e+00 -1.49973500e+00 -5.34503341e-01 2.10258889e+00
7.56328642e-01 -1.03348248e-01 -2.96778619e-01 -2.00351447e-01
1.59972131e-01 3.71103942e-01 -5.01835227e-01 -8.69456053e-01
-4.42122929e-02 9.73429143e-01 5.98312020e-01 6.86931670e-01
-8.87136281e-01 1.02878702e+00 6.71992159e+00 9.21070516e-01
-7.92357922e-01 4.42484647e-01 -6.54077763e-03 -2.68705010e-01
-5.70238352e-01 7.44009197e-01 -7.82955527e-01 9.15829390e-02
1.55818784e+00 -3.06565076e-01 1.35318542e+00 6.16706610e-01
1.96980491e-01 -2.26580709e-01 -1.16064835e+00 1.16105652e+00
-3.14077467e-01 -1.84868526e+00 -1.51502997e-01 1.77283019e-01
1.11670470e+00 4.87396240e-01 4.29866426e-02 5.92539191e-01
1.37478769e-01 -1.22217155e+00 4.47172552e-01 6.42170012e-01
5.44726968e-01 -7.01741219e-01 4.38708425e-01 2.80459583e-01
-6.66558623e-01 1.96674272e-01 -5.77002645e-01 -3.20010334e-01
4.60305154e-01 4.28111434e-01 -1.36886060e-01 6.13141179e-01
3.22875559e-01 3.88022274e-01 -2.37910420e-01 6.22153640e-01
6.29863739e-02 5.53442776e-01 -5.06480217e-01 -4.12814647e-01
7.56697714e-01 -1.26694930e+00 7.41298079e-01 6.66903496e-01
2.69774407e-01 4.26969022e-01 -2.10474342e-01 1.52640033e+00
2.74437577e-01 1.75655425e-01 -5.58864236e-01 -6.74179554e-01
-9.91652235e-02 1.16403413e+00 -2.16893002e-01 -1.47414759e-01
-3.88771892e-01 9.86899972e-01 3.33472818e-01 4.41938728e-01
-7.82368898e-01 -4.18423653e-01 8.16247165e-01 -2.33838171e-01
2.47312889e-01 -6.09436214e-01 -1.82093456e-01 -1.76607466e+00
-4.45509493e-01 -8.16466987e-01 -4.43469845e-02 -6.93105578e-01
-1.13134813e+00 1.59127682e-01 -3.21364909e-01 -7.24063396e-01
-4.36384976e-01 -9.46499586e-01 -3.99620503e-01 1.32588649e+00
-1.17602968e+00 -5.63844562e-01 2.66991019e-01 5.12725174e-01
-3.57128710e-01 -7.41619915e-02 1.29231787e+00 4.32848632e-02
-6.79332793e-01 4.25293036e-02 9.16708767e-01 -3.40298295e-01
1.57035574e-01 -1.38822865e+00 8.08272958e-02 8.06651652e-01
5.31407058e-01 1.16694915e+00 1.14975214e+00 -1.04008026e-01
-2.45648098e+00 -4.48653042e-01 6.16038799e-01 -6.95573866e-01
1.12798548e+00 -3.51017952e-01 -6.04612648e-01 5.29619753e-01
1.94801033e-01 -2.54139174e-02 5.59534669e-01 5.11635184e-01
-2.55853146e-01 1.74776003e-01 -7.27123737e-01 5.97668231e-01
8.25302720e-01 -1.36429048e+00 -4.09366995e-01 8.33140731e-01
2.06848904e-01 -4.86484081e-01 -9.66366887e-01 -1.17663942e-01
4.85455364e-01 -1.30889964e+00 1.08883035e+00 -8.53126764e-01
5.44793189e-01 -3.26627761e-01 -4.45039570e-01 -1.15588117e+00
-7.92846456e-02 -1.20364678e+00 3.32998112e-02 2.63670534e-01
3.64888400e-01 -6.38064086e-01 1.06307292e+00 5.78343689e-01
-3.99124384e-01 -4.94511813e-01 -1.51216698e+00 -5.67411423e-01
7.12942243e-01 -6.69861615e-01 6.41200840e-02 7.43888021e-01
2.48281643e-01 4.30425823e-01 -3.00212979e-01 1.94294393e-01
8.99337769e-01 3.26282620e-01 5.71617663e-01 -8.28577757e-01
-8.05890977e-01 -5.99687040e-01 -6.01308644e-01 -8.04494441e-01
4.95503783e-01 -1.67462957e+00 4.32509072e-02 -1.14467418e+00
3.68650824e-01 -1.90264687e-01 -2.21498549e-01 -1.65043771e-01
9.27256607e-03 4.11787421e-01 5.61415218e-02 1.38987780e-01
-7.09632814e-01 8.41473937e-01 1.09740090e+00 -1.01526547e-02
-1.36880010e-01 -3.16341519e-01 -2.96450287e-01 3.70777309e-01
2.94037193e-01 -6.58108234e-01 -4.87470403e-02 -8.27905610e-02
8.12732339e-01 3.57156724e-01 4.99878109e-01 -1.06067955e+00
4.61668313e-01 -1.65968552e-01 -1.36096850e-01 -2.89001048e-01
5.10208189e-01 1.39956428e-02 3.38875622e-01 5.16745448e-01
-6.76431730e-02 -3.21950108e-01 -8.11352804e-02 4.65601414e-01
-2.03154057e-01 -6.28274500e-01 8.78059745e-01 -4.21483487e-01
-5.65735340e-01 2.76117265e-01 -5.64847112e-01 6.62305728e-02
7.09492147e-01 -6.84438050e-02 1.00409295e-02 -1.47839293e-01
-8.54152560e-01 -1.44686580e-01 6.62984729e-01 -4.70275700e-01
2.32832640e-01 -1.19818842e+00 -5.48153043e-01 2.03924794e-02
2.69958898e-02 -3.84365976e-01 4.97699559e-01 1.18678582e+00
-8.24543238e-01 8.16181123e-01 2.52742320e-02 -6.11946166e-01
-3.67569834e-01 4.80560333e-01 6.80358648e-01 -2.26275742e-01
-3.59925300e-01 9.47176099e-01 -1.02131665e-01 -6.39893472e-01
-5.13419032e-01 2.40194306e-01 5.90163767e-01 -1.88191310e-01
1.31001741e-01 3.50519627e-01 1.62747856e-02 -8.29229534e-01
-4.24235821e-01 3.88301045e-01 2.27870375e-01 -3.05435836e-01
1.23599601e+00 6.78403527e-02 -6.32007539e-01 4.86394912e-01
1.50130987e+00 -3.19220871e-02 -6.83875740e-01 -3.93504016e-02
-3.91685106e-02 -1.16724428e-02 3.89905900e-01 -1.00703530e-01
-5.77882349e-01 1.18084216e+00 5.69978416e-01 7.20364153e-02
6.00952446e-01 -1.44049734e-01 6.86381876e-01 1.09748149e+00
7.94114351e-01 -1.10627437e+00 -3.87429863e-01 7.53802240e-01
2.50385970e-01 -1.36788273e+00 2.65342593e-01 9.31238234e-02
-3.87585104e-01 1.27907920e+00 -1.75066859e-01 -4.95902777e-01
4.87465143e-01 -1.62882462e-01 -4.79150087e-01 -5.01896501e-01
-5.22386372e-01 -4.39389020e-01 1.32436141e-01 2.02283949e-01
4.75989968e-01 1.59323066e-01 -4.40100342e-01 3.86572815e-02
-3.50898594e-01 -9.18299332e-02 8.89390767e-01 6.41743660e-01
-1.99132338e-01 -1.41977036e+00 -1.68160394e-01 4.59756523e-01
-1.82356820e-01 -4.96331573e-01 1.99767277e-01 5.04442394e-01
-2.32372470e-02 6.92929864e-01 -2.78433859e-01 -2.42993101e-01
1.17841721e-01 4.43828017e-01 9.93996620e-01 -6.71049237e-01
-6.20460212e-01 -4.59722310e-01 -2.60346439e-02 -1.05846894e+00
-4.89680976e-01 -9.13640678e-01 -1.26202154e+00 -7.40238607e-01
-3.96604091e-01 6.11832678e-01 8.80212784e-01 1.03852510e+00
2.00898081e-01 1.41017988e-01 2.53188699e-01 -1.25648952e+00
-1.16224682e+00 -5.14853656e-01 -9.13550377e-01 2.08559245e-01
1.86072394e-01 -6.27862394e-01 -5.24957478e-01 -5.23671985e-01]
|
[5.570132255554199, 4.967688083648682]
|
938fa5f1-a8fe-4d21-aa1e-7331dcf122fa
|
a-word-is-worth-a-thousand-dollars
| null | null |
https://openreview.net/forum?id=uFXjHTmvBph
|
https://openreview.net/pdf?id=uFXjHTmvBph
|
A Word is Worth A Thousand Dollars: Adversarial Attack on Tweets Fools Meme Stock Prediction
|
More and more investors and machine learning models rely on social media (e.g., Twitter and Reddit) to gather information and predict certain stocks' prices (meme stock). However, text-based models are known to be vulnerable to adversarial attacks, but whether stock prediction models have similar adversarial vulnerability is underexplored.
In this paper, we experiment with a variety of adversarial attack configurations to fool three stock prediction victim models (StockNet, FinGRU, FinLSTM). We address the task of adversarial generation by solving combinatorial optimization problems with semantics and budget constraints. Our results show that the proposed attack method can achieve consistent success rates, with capabilities of causing thousands of dollars loss (with Long-Only Buy-Hold-Sell investing strategy) by simply concatenating a perturbed but semantically similar tweet.
|
['Anonymous']
|
2021-10-16
| null | null | null |
acl-arr-october-2021-10
|
['stock-prediction']
|
['time-series']
|
[-3.66037011e-01 3.19407463e-01 4.98205656e-03 -1.11465275e-01
-5.06915689e-01 -1.24342716e+00 1.01945186e+00 6.07842579e-02
-2.43825480e-01 1.01216877e+00 2.10616663e-01 -5.14955819e-01
3.37518454e-01 -1.45788503e+00 -8.39641392e-01 -1.87922463e-01
-2.82670915e-01 5.63565493e-01 3.39146793e-01 -6.79858804e-01
5.05940855e-01 5.31198740e-01 -1.02631938e+00 2.07169145e-01
5.99470735e-01 1.01001048e+00 -5.55814207e-01 4.19181734e-01
-4.24760401e-01 1.46713400e+00 -1.03745675e+00 -1.53415573e+00
9.46507275e-01 -1.52732069e-02 -5.59349298e-01 -5.42911708e-01
4.45782244e-01 -7.04639196e-01 -4.93027359e-01 1.43329775e+00
4.58743513e-01 -1.90939903e-01 5.20099461e-01 -1.63212049e+00
-9.25562263e-01 1.53732860e+00 -3.15416247e-01 3.52716863e-01
1.02671057e-01 3.37761670e-01 1.17235255e+00 -4.94744539e-01
4.73900586e-01 1.09604681e+00 7.12586939e-01 4.43631470e-01
-9.58496869e-01 -1.32990158e+00 2.82151047e-02 -3.32467198e-01
-9.16316748e-01 -2.26159945e-01 5.25237262e-01 -4.25009668e-01
8.54936898e-01 5.42327523e-01 6.68303609e-01 1.42622030e+00
6.70556545e-01 6.85361743e-01 1.30679286e+00 3.31470728e-01
2.90529549e-01 5.51688433e-01 -2.16640234e-01 4.87736702e-01
7.19936609e-01 4.71533179e-01 -3.74456316e-01 -6.80860639e-01
3.50847542e-01 1.43360704e-01 2.68208355e-01 4.12541032e-01
-7.77581692e-01 1.34736931e+00 5.14347732e-01 5.83421327e-02
-4.34341967e-01 4.42267478e-01 3.47548544e-01 6.66346669e-01
8.36113930e-01 1.03342366e+00 -3.72386873e-01 1.91794023e-01
-9.42205608e-01 8.65166485e-01 1.03105688e+00 1.06752419e+00
2.99165308e-01 7.08492041e-01 -1.82593316e-02 9.80573669e-02
3.57778043e-01 6.87980413e-01 6.77098811e-01 -5.14285028e-01
7.88739860e-01 2.17372566e-01 3.13664615e-01 -1.43731701e+00
-1.31123796e-01 -4.93560702e-01 -6.35599673e-01 3.52498919e-01
2.61617452e-01 -6.64570391e-01 -6.59448266e-01 1.24644673e+00
-1.34382084e-01 6.13718033e-01 2.83008963e-01 5.04507601e-01
4.68993425e-01 7.98694432e-01 2.89782703e-01 1.16930462e-01
8.01623106e-01 -8.93342376e-01 -4.23891455e-01 -3.58576834e-01
3.80448282e-01 -6.53931081e-01 5.69663405e-01 1.57517999e-01
-1.45512223e+00 1.28506824e-01 -1.01819718e+00 6.33667350e-01
-9.42687750e-01 -1.12562311e+00 7.20812798e-01 9.97115374e-01
-6.39196515e-01 1.05243850e+00 -4.76949215e-01 3.56036067e-01
5.02654076e-01 2.25120232e-01 -2.85471473e-02 4.09448624e-01
-1.99297428e+00 1.20265412e+00 4.14584398e-01 -3.04361403e-01
-9.17371094e-01 -1.05393136e+00 -5.86236775e-01 -1.17245868e-01
1.24089323e-01 -6.27735138e-01 1.14595318e+00 -1.00528812e+00
-1.45153379e+00 6.03150368e-01 9.25666928e-01 -1.33705378e+00
1.10617590e+00 -2.32710779e-01 -5.26957393e-01 -9.05088931e-02
-5.26612401e-02 3.20342541e-01 9.42206919e-01 -9.74883139e-01
-4.51724112e-01 -1.26365945e-01 3.04011971e-01 -1.61246374e-01
-5.16368330e-01 5.86320400e-01 6.49700284e-01 -1.40437555e+00
-3.26017886e-01 -8.00402939e-01 -4.89408851e-01 -4.61174667e-01
-9.02725458e-01 3.77968132e-01 4.31992888e-01 -6.87267661e-01
1.23041463e+00 -1.56841409e+00 -3.26379389e-01 4.82792854e-01
8.04960430e-02 3.39097321e-01 6.65838122e-02 7.51024902e-01
-8.16029236e-02 9.19584572e-01 -1.50284529e-01 -4.11411971e-01
4.96853679e-01 -1.60546422e-01 -1.29544914e+00 3.19971740e-01
8.45456421e-02 1.29604208e+00 -7.35620081e-01 -1.81316838e-01
-1.69088051e-01 -2.69365489e-01 -5.13455331e-01 2.24167883e-01
-5.68318963e-01 -8.89028311e-02 -6.16424203e-01 7.31176138e-01
5.69042861e-01 6.08273558e-02 -3.03840160e-01 3.75970483e-01
6.36043921e-02 2.42702588e-01 -9.48958814e-01 5.98529994e-01
-1.15985520e-01 1.43918246e-01 -4.20496643e-01 -4.18593466e-01
9.07110691e-01 -2.13337578e-02 2.42006496e-01 -3.43198478e-01
2.81512558e-01 1.84132025e-01 -2.13782117e-01 7.48544186e-02
5.70220530e-01 -5.73054790e-01 -5.62215865e-01 9.78423357e-01
-3.21855694e-01 -1.15556918e-01 -1.86509937e-01 4.05674130e-01
1.18618989e+00 -4.96481836e-01 4.72788550e-02 -1.40998021e-01
-4.54078019e-02 1.34449556e-01 3.21782291e-01 9.25619841e-01
-1.67935461e-01 2.78376818e-01 5.30677736e-01 -5.39247036e-01
-1.16073394e+00 -1.25609589e+00 1.94343001e-01 7.58238792e-01
-1.41909093e-01 -6.63315877e-02 -6.90760374e-01 -8.75885546e-01
6.70178652e-01 1.26774764e+00 -5.06110430e-01 -3.08983564e-01
-3.74076456e-01 -8.67156982e-01 1.23339999e+00 3.09252858e-01
6.53304458e-01 -1.21663797e+00 -1.86441660e-01 4.43748116e-01
4.16212916e-01 -7.63334394e-01 -5.55270791e-01 -2.95178235e-01
-5.25103688e-01 -7.14274704e-01 -6.46970809e-01 -9.27498788e-02
4.08515811e-01 -3.05868238e-01 1.21479344e+00 7.50996917e-02
6.92667142e-02 -8.45399126e-03 -1.81409702e-01 -1.00370193e+00
-7.66662598e-01 2.32090317e-02 2.23835543e-01 3.98601480e-02
3.60816643e-02 -5.52028656e-01 -4.29090023e-01 2.47998625e-01
-1.22984743e+00 -2.96955913e-01 2.09800541e-01 3.74872386e-01
7.25629255e-02 4.35302943e-01 9.90832210e-01 -1.42608130e+00
9.18901742e-01 -1.05604124e+00 -1.00614703e+00 1.79735750e-01
-8.18340242e-01 -2.64079362e-01 8.33251715e-01 -5.56406558e-01
-7.32744277e-01 -3.03777516e-01 -3.95106487e-02 -6.66506350e-01
3.42478573e-01 6.75195873e-01 8.10652599e-02 -2.56420821e-01
6.43263459e-01 2.37172961e-01 -1.15725026e-02 -1.21910915e-01
3.62090349e-01 2.13171497e-01 2.00380713e-01 -2.21841007e-01
1.64336634e+00 2.67416418e-01 -2.29654461e-01 -4.40788753e-02
-8.49268913e-01 5.02667129e-01 2.48028245e-02 -1.70042306e-01
5.77069700e-01 -7.72796988e-01 -5.74257135e-01 9.77093220e-01
-8.16763282e-01 -1.90436214e-01 -2.18209416e-01 4.81198654e-02
-2.41977006e-01 6.25991896e-02 -1.06187677e+00 -7.54462898e-01
-7.03696609e-01 -6.60715818e-01 1.70916855e-01 -2.91663106e-04
-2.46523067e-01 -1.17088842e+00 1.40947729e-01 2.44651467e-01
9.23942983e-01 7.84392774e-01 7.33114958e-01 -1.62030029e+00
-8.97705078e-01 -6.92165256e-01 1.28671482e-01 4.13227141e-01
-3.25546302e-02 6.96943421e-03 -7.34821260e-01 -1.60624459e-01
3.49294730e-02 -3.75848144e-01 7.03137517e-01 -2.19734143e-02
7.10444868e-01 -1.30921471e+00 1.39238924e-01 4.99775141e-01
1.40365756e+00 6.95557743e-02 7.98830271e-01 6.21690929e-01
5.32258809e-01 6.72690690e-01 4.80013281e-01 6.02694690e-01
1.40886545e-01 1.01983892e-02 6.97277367e-01 3.64965856e-01
6.70627594e-01 -7.26123989e-01 7.67543733e-01 3.52208525e-01
2.01294512e-01 -4.99022901e-01 -8.45111310e-01 1.29898144e-02
-1.43759167e+00 -1.51473033e+00 3.16217244e-01 1.87140250e+00
9.32777882e-01 9.02126431e-01 4.37744319e-01 -1.37236804e-01
6.11645103e-01 4.44184065e-01 -7.11140871e-01 -3.37205738e-01
-3.56062949e-01 2.39998221e-01 1.39076960e+00 3.60581368e-01
-1.09684610e+00 1.25949490e+00 7.27279377e+00 6.69357240e-01
-9.79943871e-01 -1.03892289e-01 9.00377750e-01 -3.48045230e-01
-8.41808438e-01 -3.25755358e-01 -9.89487469e-01 9.78905916e-01
1.32581997e+00 -9.82831717e-01 4.50450420e-01 9.83812273e-01
-1.71744213e-01 5.93794703e-01 -7.88978338e-01 1.16013430e-01
-9.09115002e-02 -1.82618666e+00 4.51788455e-01 1.27808779e-01
8.21522951e-01 -5.83248120e-03 5.45967996e-01 4.83074069e-01
1.17913532e+00 -1.16361797e+00 1.23803687e+00 8.15158010e-01
2.47169361e-01 -1.12159264e+00 7.87263036e-01 3.69105726e-01
-6.76236331e-01 -2.51694083e-01 -4.05007511e-01 1.02061674e-01
2.79285222e-01 3.61562341e-01 -6.58393741e-01 2.39338741e-01
4.33400512e-01 5.44202507e-01 -5.53090274e-01 6.57239377e-01
-6.33399189e-02 6.79773569e-01 -2.78412402e-01 -2.89132684e-01
5.67248404e-01 -1.64870635e-01 6.18815184e-01 8.42221737e-01
4.59893674e-01 2.03948617e-01 9.48272925e-03 1.19500625e+00
-4.25558001e-01 -1.94905743e-01 -9.60835159e-01 -5.82845330e-01
6.58188105e-01 7.73284733e-01 -4.02258009e-01 -3.20128918e-01
-1.93465814e-01 6.84934735e-01 -1.08867571e-01 1.70074757e-02
-1.01618147e+00 -3.89844894e-01 9.00154233e-01 4.57153440e-01
-1.50307938e-01 1.25934303e-01 -4.34879601e-01 -1.06273079e+00
-4.29483324e-01 -1.03762162e+00 4.36727762e-01 -6.85536802e-01
-2.02125478e+00 5.68077803e-01 -8.24751258e-02 -1.03452313e+00
-4.15866464e-01 -4.58701611e-01 -1.17819846e+00 8.64118814e-01
-1.40139353e+00 -1.13601160e+00 5.48385620e-01 6.89272642e-01
1.68623984e-01 -1.07907295e+00 6.21167839e-01 1.38135299e-01
-6.26618326e-01 9.49387133e-01 2.08387543e-02 5.09016216e-01
2.64160872e-01 -1.38805485e+00 1.25202751e+00 8.24912190e-01
-1.17066391e-01 5.28302789e-01 9.45017278e-01 -1.22440338e+00
-1.23065901e+00 -1.63630593e+00 7.49298632e-01 -6.00447714e-01
1.57818460e+00 -2.19893306e-01 -6.89834177e-01 1.05181623e+00
2.07747221e-01 -2.94607490e-01 6.24562323e-01 -6.98294640e-01
-5.58428168e-01 1.34151787e-01 -1.76223528e+00 8.56208384e-01
7.50833213e-01 -3.98155779e-01 -5.90674281e-01 5.21253228e-01
1.23950505e+00 -1.72417805e-01 -1.00042498e+00 -7.01684952e-02
1.51557043e-01 -1.09443569e+00 1.19252026e+00 -1.26286876e+00
7.64580727e-01 2.44795293e-01 -1.13338336e-01 -1.37686062e+00
-1.59751743e-01 -1.03909266e+00 -1.04273364e-01 1.38933372e+00
8.11800361e-01 -1.30966556e+00 8.60332251e-01 1.09389853e+00
6.85681477e-02 -5.55500805e-01 -6.24915898e-01 -9.32948112e-01
6.34351075e-01 -2.55041897e-01 1.23739719e+00 1.20317161e+00
-1.83810681e-01 -2.17741594e-01 -5.77140331e-01 4.27894205e-01
7.91399896e-01 -4.22746241e-02 5.66571832e-01 -1.05202031e+00
-3.46864522e-01 -6.31219208e-01 -3.31605136e-01 2.54021455e-02
5.58997750e-01 -9.30234432e-01 -6.65854990e-01 -8.16958368e-01
-3.14178556e-01 -4.96905655e-01 -3.23003292e-01 4.26250577e-01
6.28036559e-02 2.92455912e-01 6.17434919e-01 2.60088652e-01
-1.94352791e-01 2.64469624e-01 7.54104018e-01 -4.30909008e-01
2.64911145e-01 4.22480464e-01 -8.77955079e-01 8.65031004e-01
1.09120703e+00 -6.78909361e-01 -1.27128661e-01 2.61563957e-02
8.28144133e-01 6.38314039e-02 4.19179142e-01 -5.39714754e-01
1.61089107e-01 -4.15434957e-01 1.45204842e-01 -4.40972120e-01
7.66791999e-02 -6.66913509e-01 5.57180762e-01 8.04248452e-01
-5.51782310e-01 5.05530417e-01 2.39200950e-01 4.97404009e-01
-8.28117877e-02 -5.25028229e-01 8.38010728e-01 -4.77464437e-01
-1.93466336e-01 6.43267930e-01 -5.35346746e-01 2.28455290e-01
1.29364538e+00 2.08227918e-01 -6.48880005e-01 -5.89062572e-01
-7.48635888e-01 2.25728363e-01 3.63578022e-01 4.12426382e-01
6.03074729e-01 -1.45835912e+00 -1.07385767e+00 2.07809787e-02
-2.71518230e-01 -4.38389957e-01 -1.62047103e-01 -1.33725733e-01
-9.10161197e-01 1.44040257e-01 -2.91882515e-01 5.98346531e-01
-6.69544578e-01 7.56448269e-01 5.72729945e-01 -6.44721746e-01
-3.29164088e-01 1.03276157e+00 -3.10159773e-01 -2.70415604e-01
2.42663428e-01 1.26236573e-01 8.03946052e-03 3.18684340e-01
6.78618491e-01 6.34006321e-01 -1.58269614e-01 -3.25685084e-01
4.07748185e-02 -9.93476734e-02 -2.90255547e-01 -1.50962338e-01
1.40540206e+00 3.11470062e-01 -1.26696974e-01 2.80226678e-01
9.74121809e-01 2.84737766e-01 -7.28223205e-01 -1.45411611e-01
3.37325186e-01 -6.23216867e-01 -1.61577985e-01 -8.46050620e-01
-1.37380624e+00 3.42853695e-01 -5.34066581e-04 8.45106959e-01
5.36552131e-01 -4.11019474e-01 1.34546745e+00 3.83916557e-01
5.15074253e-01 -7.74013817e-01 2.95142997e-02 3.43172491e-01
1.10520327e+00 -8.69322598e-01 -6.82746544e-02 6.59726141e-03
-9.42802489e-01 5.25587559e-01 4.41418588e-01 -6.22924447e-01
9.96869564e-01 5.34799159e-01 1.48763672e-01 -1.31348565e-01
-8.52004945e-01 3.99812907e-01 -2.21533999e-02 2.09606275e-01
-3.16252351e-01 1.57021001e-01 3.06058794e-01 9.79667962e-01
-9.67947900e-01 -6.04346216e-01 8.73026609e-01 8.38361084e-01
-5.49026728e-01 -8.93358886e-01 -4.51864332e-01 8.77752125e-01
-1.21860099e+00 -5.31149805e-01 -7.28515565e-01 6.17106557e-01
-3.82819563e-01 7.06274867e-01 1.76573038e-01 -6.41103208e-01
3.36220026e-01 -4.54910956e-02 -3.08904380e-01 -6.66263461e-01
-1.46505988e+00 -4.98698235e-01 2.23383769e-01 -3.47687840e-01
1.91663757e-01 -6.92078590e-01 -9.78512168e-01 -1.23867977e+00
-6.70178384e-02 -2.27081962e-02 3.78949106e-01 5.60115397e-01
1.78145394e-01 2.00613998e-02 1.23347700e+00 -5.44305503e-01
-1.48277700e+00 -7.38152862e-01 -1.07083690e+00 4.73671436e-01
4.07290980e-02 -3.14898401e-01 -6.82623267e-01 -1.86537549e-01]
|
[5.701230525970459, 7.6908650398254395]
|
9450b9af-cb81-467a-b5bf-5c5926cb4557
|
roca-robust-cad-model-retrieval-and-alignment
|
2112.01988
| null |
https://arxiv.org/abs/2112.01988v2
|
https://arxiv.org/pdf/2112.01988v2.pdf
|
ROCA: Robust CAD Model Retrieval and Alignment from a Single Image
|
We present ROCA, a novel end-to-end approach that retrieves and aligns 3D CAD models from a shape database to a single input image. This enables 3D perception of an observed scene from a 2D RGB observation, characterized as a lightweight, compact, clean CAD representation. Core to our approach is our differentiable alignment optimization based on dense 2D-3D object correspondences and Procrustes alignment. ROCA can thus provide a robust CAD alignment while simultaneously informing CAD retrieval by leveraging the 2D-3D correspondences to learn geometrically similar CAD models. Experiments on challenging, real-world imagery from ScanNet show that ROCA significantly improves on state of the art, from 9.5% to 17.6% in retrieval-aware CAD alignment accuracy.
|
['Matthias Nießner', 'Angela Dai', 'Can Gümeli']
|
2021-12-03
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Gumeli_ROCA_Robust_CAD_Model_Retrieval_and_Alignment_From_a_Single_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Gumeli_ROCA_Robust_CAD_Model_Retrieval_and_Alignment_From_a_Single_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['3d-dense-shape-correspondence', '3d-shape-reconstruction-from-a-single-2d', '3d-object-detection-from-monocular-images', '3d-object-retrieval']
|
['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.35953337e-01 -2.11263269e-01 1.15972934e-02 -5.90814888e-01
-1.43044353e+00 -9.15649891e-01 7.17216790e-01 -4.05215621e-02
-4.69888735e-04 -3.17415267e-01 -6.36983365e-02 2.89163347e-02
-8.24051052e-02 -7.38175154e-01 -9.38870311e-01 -6.25790209e-02
9.80528519e-02 1.07390153e+00 2.73196846e-02 -3.64204049e-02
3.80522966e-01 1.03668427e+00 -1.47574711e+00 -1.51733547e-01
4.90032107e-01 1.39970517e+00 3.35761786e-01 5.68152905e-01
-4.72334512e-02 7.77970552e-02 -9.35401097e-02 -2.81066358e-01
8.42155635e-01 7.69853815e-02 -6.46459818e-01 3.63948047e-01
1.27009654e+00 -4.66727883e-01 -1.41865045e-01 8.49026978e-01
3.94550145e-01 -2.11979151e-01 6.19875789e-01 -8.81061733e-01
-6.94385409e-01 -2.22038537e-01 -7.44739354e-01 -5.22313356e-01
7.28598237e-01 1.69330448e-01 1.07340455e+00 -1.41059482e+00
8.78383040e-01 1.34881115e+00 7.64080882e-01 7.19783455e-02
-1.49837422e+00 -3.93380016e-01 -3.80700305e-02 -1.49368227e-01
-1.86481547e+00 -3.68547261e-01 7.46028543e-01 -4.89561081e-01
1.06599104e+00 2.85512686e-01 8.70798886e-01 5.19000292e-01
3.73929217e-02 5.54780602e-01 8.48653793e-01 -2.58728325e-01
2.15840220e-01 -4.21964735e-01 -3.20968702e-02 7.26366818e-01
2.15481192e-01 1.57407209e-01 -6.77484810e-01 -2.25396872e-01
1.13356209e+00 2.03582004e-01 3.43776755e-02 -1.09011865e+00
-1.34763145e+00 4.56408799e-01 9.36852336e-01 -4.10435736e-01
-5.31188011e-01 4.51331437e-01 -2.45182842e-01 2.89975494e-01
4.50212479e-01 5.03947616e-01 -4.08673823e-01 -4.97398935e-02
-8.04745078e-01 1.57852441e-01 5.65990508e-01 1.54576719e+00
1.13999009e+00 -1.35426372e-01 5.17557800e-01 5.63948154e-01
6.54507518e-01 1.41346014e+00 -1.91502124e-02 -1.11202407e+00
3.89146835e-01 7.50993133e-01 1.89764183e-02 -1.14884949e+00
-8.18988383e-02 -3.24159861e-01 -6.23392880e-01 3.91469181e-01
-1.79601699e-01 5.23010552e-01 -9.63646889e-01 1.14037478e+00
3.10125768e-01 6.96084797e-02 -1.26074050e-02 1.24795175e+00
7.05788493e-01 3.06068122e-01 -4.42213625e-01 5.85154414e-01
1.10472977e+00 -8.17811191e-01 -2.52919551e-02 -6.62398636e-01
2.85888135e-01 -1.21528089e+00 9.41863179e-01 2.35939354e-01
-1.26114118e+00 -6.71513677e-01 -1.11513901e+00 -4.11497504e-01
-1.21218085e-01 2.84149408e-01 5.71507573e-01 1.80298537e-01
-1.11672330e+00 4.56537813e-01 -1.04989851e+00 -4.25771058e-01
4.81175184e-01 3.19427401e-01 -7.61969745e-01 -5.42388260e-01
1.50909647e-02 9.77231026e-01 1.25392929e-01 9.79238600e-02
-8.20056736e-01 -8.30750763e-01 -1.08296323e+00 -2.99054086e-01
5.27394474e-01 -1.16673160e+00 1.30251706e+00 -5.59402764e-01
-1.43026161e+00 1.37582850e+00 -3.46149296e-01 -2.89844900e-01
6.04272068e-01 -8.26781631e-01 1.07031621e-01 2.43248940e-01
2.47096151e-01 9.75967109e-01 8.54092419e-01 -1.29006505e+00
-1.40468895e-01 -6.35185122e-01 -3.47289890e-01 2.81010151e-01
6.52513921e-01 -4.57908720e-01 -7.25220859e-01 -5.37297726e-01
1.13490224e+00 -1.10167491e+00 -3.13436866e-01 7.58377910e-01
-2.44140431e-01 2.28694990e-01 8.21924567e-01 -5.28566539e-01
2.34338135e-01 -1.87469864e+00 4.05626625e-01 4.21254933e-01
4.16999198e-02 -2.31595710e-01 -3.65832686e-01 4.11749929e-01
5.34695312e-02 -8.10765773e-02 -2.52750456e-01 -6.60303712e-01
1.95592530e-02 2.36644596e-01 -5.91726601e-01 6.63828671e-01
7.05102384e-01 1.22684324e+00 -7.08109856e-01 -1.70130044e-01
6.53139889e-01 3.78316164e-01 -5.71515679e-01 5.64466655e-01
-3.29556167e-01 4.63379383e-01 -5.09301364e-01 1.25474608e+00
1.12863326e+00 -2.75393963e-01 -1.22889437e-01 -5.64848423e-01
7.55583961e-03 2.04999492e-01 -1.29248667e+00 2.56212568e+00
-4.81250584e-01 4.82310832e-01 -5.00796624e-02 -6.65370166e-01
1.30334449e+00 -2.26876646e-01 4.82986778e-01 -7.93634832e-01
-1.40841275e-01 4.74022955e-01 -6.35339975e-01 -1.20460596e-02
5.33862591e-01 3.73276174e-01 -3.16523373e-01 3.55074644e-01
7.85423890e-02 -1.26318967e+00 -4.96322334e-01 1.32007688e-01
8.20641696e-01 6.52746022e-01 4.92950737e-01 -1.67727068e-01
3.14122230e-01 3.50996822e-01 2.46336050e-02 5.54219663e-01
5.62878311e-01 1.08945012e+00 -2.92499125e-01 -6.12893224e-01
-1.21575630e+00 -1.53778493e+00 -7.64620316e-04 3.15027714e-01
4.40775454e-01 -3.78624707e-01 -1.57255262e-01 -2.47309297e-01
3.47622693e-01 2.57023752e-01 -2.12474257e-01 -1.45633016e-02
-7.89400101e-01 2.45376807e-02 7.21202642e-02 6.98446989e-01
5.22741675e-01 -3.24514776e-01 -9.24341023e-01 6.00593686e-02
9.50213894e-02 -1.29328501e+00 -6.06754780e-01 7.68273547e-02
-1.22673690e+00 -1.14721668e+00 -4.56875145e-01 -6.21812224e-01
8.64120662e-01 7.31985271e-01 1.57083011e+00 9.64340717e-02
-5.01879990e-01 9.05113399e-01 -1.55832499e-01 -2.94973314e-01
-2.44941905e-01 -1.74191557e-02 -5.18346056e-02 -1.81947172e-01
1.81870610e-01 -7.10194170e-01 -4.85469431e-01 3.47812802e-01
-6.21244967e-01 2.74953365e-01 8.42457414e-01 6.13743067e-01
1.36835217e+00 -8.89238358e-01 -2.01025575e-01 -1.08713463e-01
-2.25912005e-01 -1.80240870e-02 -1.21715379e+00 9.68617499e-02
-6.37746155e-01 2.01409668e-01 -1.32287620e-02 -1.29835248e-01
-5.71420014e-01 8.31353784e-01 1.95160568e-01 -8.46427917e-01
-3.54101300e-01 4.71056849e-01 -2.53707647e-01 -3.38040352e-01
4.50863898e-01 1.46835059e-01 2.71431148e-01 -8.68103802e-01
7.90814042e-01 5.14539778e-01 1.07924831e+00 -6.25177324e-01
1.33857358e+00 7.28703618e-01 3.25967073e-01 -5.75138927e-01
-8.38282764e-01 -8.55651975e-01 -1.37123966e+00 8.10753107e-02
7.38937497e-01 -1.42297208e+00 -4.39561754e-01 3.53117436e-01
-1.25764966e+00 -1.07015610e-01 -3.46376866e-01 4.71376091e-01
-1.05432487e+00 4.32681262e-01 -2.79782824e-02 -5.16027272e-01
-5.34286499e-01 -1.11636853e+00 2.02795434e+00 -4.27522473e-02
-1.91795990e-01 -4.28404808e-01 8.40367302e-02 3.63561779e-01
1.85058743e-01 5.28683841e-01 5.24866223e-01 -1.24958970e-01
-1.42837238e+00 -4.57703978e-01 -3.39208066e-01 1.83754042e-01
1.79731518e-01 -3.16409096e-02 -9.03036892e-01 -2.08702713e-01
-3.60387027e-01 -3.33002478e-01 5.32365680e-01 1.20656393e-01
7.01798916e-01 4.28123288e-02 -2.84372091e-01 9.56794918e-01
1.75582159e+00 -2.25899100e-01 4.34086323e-01 2.26489931e-01
7.19734848e-01 1.25100300e-01 1.00944960e+00 1.79636642e-01
3.92939687e-01 9.54604149e-01 1.01967704e+00 -1.98006794e-01
-3.29013884e-01 -3.91868860e-01 1.03194192e-01 8.42694461e-01
3.90549190e-02 3.69719625e-01 -1.05363441e+00 4.27065939e-01
-1.73949468e+00 -3.80694807e-01 -1.33787602e-01 2.26105499e+00
6.16106749e-01 -2.48046353e-01 -2.86507040e-01 -3.86047155e-01
2.59167045e-01 1.49043975e-02 -7.32083321e-01 2.30458215e-01
-3.19435686e-01 5.53064942e-01 6.35066867e-01 5.40113330e-01
-9.87964630e-01 1.04124188e+00 6.21397066e+00 3.72235119e-01
-1.10563445e+00 -2.35850185e-01 4.90295961e-02 -1.06548518e-02
-1.97997570e-01 2.09003374e-01 -5.23970723e-01 -1.57315940e-01
4.13037479e-01 -1.69828176e-01 3.98581833e-01 1.19324446e+00
-3.67580116e-01 -1.59156531e-01 -1.58851337e+00 1.46862411e+00
1.94017008e-01 -1.63676071e+00 1.39455557e-01 1.79017588e-01
7.82425165e-01 5.30052066e-01 -9.01766568e-02 -2.95700997e-01
4.51403528e-01 -9.69127476e-01 1.17526889e+00 7.05243051e-01
1.04846621e+00 -4.60548013e-01 6.41427279e-01 3.10805172e-01
-1.29644358e+00 5.30772686e-01 -4.09152210e-01 2.35055402e-01
3.54637235e-01 3.91651154e-01 -9.85305011e-01 1.03775978e+00
7.26008296e-01 1.15434158e+00 -8.79267871e-01 1.20500219e+00
-4.39907378e-03 -5.94848990e-02 -7.21695900e-01 4.73600388e-01
2.40579173e-01 -4.47022051e-01 4.65491861e-01 8.34720731e-01
6.68821931e-01 1.51959106e-01 3.23826849e-01 1.16286564e+00
-1.40937818e-02 -1.93816707e-01 -9.08701777e-01 1.75077528e-01
6.95055187e-01 1.02565587e+00 -5.44643581e-01 -2.03897923e-01
-1.13311514e-01 1.13327086e+00 6.65492788e-02 -4.69358638e-02
-4.84158128e-01 -1.54255927e-02 8.32089603e-01 -1.05319098e-01
5.28173506e-01 -7.98672080e-01 -3.81439656e-01 -1.04550743e+00
3.69567931e-01 -6.99575186e-01 6.53648749e-02 -1.28028703e+00
-1.33970451e+00 5.82696080e-01 1.08181044e-01 -1.74058914e+00
-4.79823291e-01 -6.56417966e-01 -1.98382184e-01 1.13578725e+00
-1.54453623e+00 -1.45462835e+00 -1.00197494e+00 3.63057137e-01
6.34652793e-01 -1.01354122e-01 1.02448714e+00 -1.44976333e-01
2.57099807e-01 1.06910476e-02 -7.67184943e-02 2.84387656e-02
6.78004920e-01 -1.14344585e+00 1.08850515e+00 7.12511361e-01
6.85623109e-01 3.66606534e-01 1.61931559e-01 -5.09118795e-01
-2.17611551e+00 -1.02083945e+00 4.68535513e-01 -8.83118033e-01
2.75523722e-01 -3.15549910e-01 -6.74277008e-01 7.84716070e-01
-5.12524843e-02 3.09156150e-01 1.80234477e-01 -3.88336658e-01
-7.44162858e-01 -2.31868342e-01 -1.11941206e+00 4.79376704e-01
1.47239184e+00 -7.51909614e-01 -7.39906788e-01 2.97646105e-01
7.11969256e-01 -1.00771976e+00 -1.14240420e+00 3.74872923e-01
6.51961803e-01 -6.58204257e-01 1.61187744e+00 -4.40265656e-01
2.08742782e-01 -6.49690330e-01 -9.79647815e-01 -1.14976299e+00
-1.96016729e-02 -6.35320008e-01 5.04626743e-02 7.25532830e-01
-1.50745317e-01 -3.82589668e-01 6.88750267e-01 5.40834308e-01
-2.73396164e-01 -5.06022394e-01 -1.00426006e+00 -9.91157174e-01
-2.13264659e-01 -7.04430044e-01 9.48603749e-01 7.36469865e-01
-8.94700348e-01 2.06011936e-01 1.01248004e-01 6.64229035e-01
8.87627006e-01 7.41169453e-01 1.50882912e+00 -1.60805893e+00
-1.28484637e-01 -3.63864988e-01 -9.09928203e-01 -1.57420897e+00
1.09870121e-01 -9.82957542e-01 2.79195756e-02 -1.38058186e+00
-1.89052969e-01 -6.50261760e-01 4.68677133e-01 2.73809195e-01
3.34179968e-01 5.80228686e-01 4.34294939e-01 4.80848014e-01
-4.07611758e-01 5.05824089e-01 9.00120378e-01 -2.92847902e-01
-4.05268837e-03 -2.06762880e-01 -4.04890448e-01 7.23398864e-01
2.81221479e-01 -3.81170899e-01 5.10450602e-02 -1.06868863e+00
4.37195785e-02 -4.83809486e-02 8.67221475e-01 -8.67832303e-01
2.59086370e-01 -1.35293409e-01 4.40477997e-01 -1.40740609e+00
7.90598512e-01 -1.26321352e+00 5.81518769e-01 2.27361396e-01
9.08045855e-04 3.11065018e-01 2.55635917e-01 6.33708358e-01
-1.59422278e-01 1.13167323e-01 4.38486755e-01 -2.52577156e-01
-8.86218429e-01 3.96955758e-01 4.63431060e-01 -1.44738704e-01
7.62506068e-01 -4.17884678e-01 -6.89969286e-02 -3.16803306e-01
-6.60280347e-01 1.92479044e-01 9.21175659e-01 5.28705001e-01
8.79628301e-01 -1.68684649e+00 -8.03852201e-01 5.79539716e-01
4.77986634e-01 9.25076663e-01 -1.16253890e-01 5.61524689e-01
-9.48761165e-01 5.36813617e-01 -2.46817023e-01 -1.50366473e+00
-1.42762506e+00 1.39895618e-01 3.63206983e-01 2.04249382e-01
-7.43332326e-01 5.67695796e-01 -1.36293799e-01 -8.43854308e-01
1.24593779e-01 -6.20746434e-01 8.04797053e-01 -3.76831412e-01
1.56509712e-01 1.25292256e-01 4.05970484e-01 -7.25179553e-01
-6.40992284e-01 1.50696111e+00 3.01311404e-01 -2.67838806e-01
1.57387757e+00 -1.13880493e-01 -5.55873103e-02 2.30550960e-01
1.42993188e+00 4.07034382e-02 -1.36726022e+00 -7.36020744e-01
1.16652153e-01 -9.66285229e-01 1.23031579e-01 -7.26513326e-01
-9.20270324e-01 8.39955151e-01 6.77058816e-01 -2.98851430e-01
1.02640665e+00 5.92272460e-01 4.71651465e-01 8.98792624e-01
8.13297629e-01 -4.39674795e-01 3.08756739e-01 2.78800666e-01
1.35003972e+00 -1.35593903e+00 4.12409872e-01 -5.05902112e-01
-4.34143871e-01 1.42210066e+00 3.76879454e-01 -5.03336787e-01
6.00025058e-01 1.92582667e-01 2.02750444e-01 -5.19063711e-01
-3.43630403e-01 -2.19977781e-01 6.95720196e-01 8.12723756e-01
-9.13184956e-02 2.39066370e-02 6.99130058e-01 -8.87545664e-03
-4.39178079e-01 -3.07532489e-01 1.60984308e-01 9.79372323e-01
-1.15438662e-01 -1.06341779e+00 -4.80836153e-01 -1.02908112e-01
3.27781916e-01 5.49068041e-02 -5.93461156e-01 9.46909308e-01
-4.31574672e-01 4.13356692e-01 6.27598882e-01 -2.47853234e-01
7.64282048e-01 -8.82024392e-02 7.40445733e-01 -6.89625382e-01
-1.46981373e-01 3.72001588e-01 -1.83409095e-01 -1.19219351e+00
-6.22593582e-01 -8.48806381e-01 -1.00964856e+00 -1.11236796e-01
-2.87688941e-01 -4.67439890e-01 9.49315250e-01 6.46151841e-01
8.40884566e-01 -1.76137283e-01 7.54428506e-01 -1.35055244e+00
-8.12542915e-01 -6.48311913e-01 -1.73287660e-01 2.07543805e-01
5.03975749e-01 -5.30437350e-01 -4.47757170e-02 2.51093417e-01]
|
[7.728906631469727, -2.718454122543335]
|
58c26b12-1395-4704-b606-a736c3de6512
|
attention-model-for-articulatory-features
|
1907.01914
| null |
https://arxiv.org/abs/1907.01914v1
|
https://arxiv.org/pdf/1907.01914v1.pdf
|
Attention model for articulatory features detection
|
Articulatory distinctive features, as well as phonetic transcription, play important role in speech-related tasks: computer-assisted pronunciation training, text-to-speech conversion (TTS), studying speech production mechanisms, speech recognition for low-resourced languages. End-to-end approaches to speech-related tasks got a lot of traction in recent years. We apply Listen, Attend and Spell~(LAS)~\cite{Chan-LAS2016} architecture to phones recognition on a small small training set, like TIMIT~\cite{TIMIT-1992}. Also, we introduce a novel decoding technique that allows to train manners and places of articulation detectors end-to-end using attention models. We also explore joint phones recognition and articulatory features detection in multitask learning setting.
|
['Ievgen Karaulov', 'Dmytro Tkanov']
|
2019-07-02
| null | null | null | null |
['manner-of-articulation-detection']
|
['speech']
|
[ 2.33319461e-01 -3.12472135e-01 -1.68313429e-01 -4.58247453e-01
-1.39325559e+00 -6.40646577e-01 6.97878957e-01 -3.23199481e-01
-5.37217677e-01 4.24283594e-01 6.14314556e-01 -7.04337895e-01
1.90136179e-01 1.55064687e-02 -6.72036231e-01 -3.46594214e-01
3.03676009e-01 7.37267017e-01 -1.18786938e-01 7.95469284e-02
3.03491969e-02 2.37472832e-01 -1.42644763e+00 5.79798162e-01
6.64614379e-01 7.88906991e-01 8.88224125e-01 1.22214854e+00
-1.44734532e-01 5.56534469e-01 -5.18283963e-01 -2.09775746e-01
-2.47241244e-01 -3.16661477e-01 -8.06350708e-01 -3.17199200e-01
4.50159729e-01 -2.79470403e-02 -1.53525904e-01 7.43532300e-01
9.40546453e-01 3.08467716e-01 8.75417888e-01 -7.29435921e-01
-9.66938317e-01 7.34843314e-01 -1.54976368e-01 7.39262700e-01
4.19190019e-01 2.42607251e-01 9.12789524e-01 -1.46235895e+00
6.92387968e-02 1.55105174e+00 4.67482597e-01 7.29464710e-01
-8.65615845e-01 -5.89945614e-01 1.24784261e-01 4.90136892e-01
-1.09828174e+00 -1.59197998e+00 6.22052252e-01 -2.82384306e-01
1.53913343e+00 5.06221831e-01 8.02829936e-02 1.53779399e+00
-3.24282408e-01 1.33257151e+00 1.14276958e+00 -8.37325335e-01
-1.89795896e-01 -1.34850979e-01 -8.65547359e-02 4.17378247e-01
-5.93882382e-01 2.27130055e-01 -9.32668030e-01 4.59736824e-01
5.50815761e-01 -2.25591823e-01 -1.59264237e-01 7.06368148e-01
-1.52017212e+00 5.13506591e-01 -2.23022833e-01 5.45887709e-01
-2.64907420e-01 2.68115014e-01 7.17497766e-01 5.79585791e-01
6.23365164e-01 1.67243585e-01 -8.24966252e-01 -8.18540633e-01
-8.10216904e-01 -3.17915440e-01 4.38399076e-01 1.10696614e+00
1.78850070e-01 5.68787813e-01 -6.29326761e-01 1.53244269e+00
4.19156671e-01 7.46449530e-01 9.96212244e-01 -4.91867214e-01
8.92507255e-01 -4.49190766e-01 -4.07696635e-01 1.61808193e-01
1.13792894e-02 -2.63862163e-01 -7.74152756e-01 -2.55567402e-01
3.06900382e-01 -9.60978344e-02 -1.05686271e+00 1.81665814e+00
1.72824692e-02 4.36258256e-01 1.17690556e-01 5.03899097e-01
7.57800758e-01 9.79978144e-01 2.20864236e-01 -2.78375596e-01
1.49574649e+00 -1.42571247e+00 -9.34823453e-01 -3.22930127e-01
5.15836358e-01 -1.19113100e+00 1.60419655e+00 3.39035779e-01
-1.66050315e+00 -1.07241654e+00 -4.18124974e-01 -4.39546227e-01
-4.41850126e-01 6.09369695e-01 2.62641400e-01 6.85867429e-01
-1.24646950e+00 2.88064063e-01 -8.43923926e-01 -2.24777728e-01
3.21201771e-01 2.56593049e-01 -1.73311338e-01 8.67940933e-02
-1.00983012e+00 8.42485726e-01 1.37949020e-01 2.03170720e-02
-1.33285761e+00 -4.96439636e-01 -8.06490958e-01 8.76851380e-02
1.38331339e-01 -5.40846944e-01 1.62534380e+00 -6.88654721e-01
-2.12816525e+00 1.15054333e+00 -7.41919577e-01 -4.12703633e-01
9.65548679e-02 -5.92881739e-01 -7.77470350e-01 -2.76410937e-01
-1.01462513e-01 4.77723002e-01 1.12129045e+00 -6.74316883e-01
-7.06361175e-01 -2.51405686e-01 -8.29050183e-01 3.92551422e-01
-4.14117515e-01 7.43727744e-01 -3.84624898e-01 -9.38821554e-01
-2.91585386e-01 -7.95602441e-01 3.17802191e-01 -6.94049597e-01
-5.73740840e-01 -8.36658001e-01 6.87738836e-01 -9.99103427e-01
1.09996653e+00 -2.27086902e+00 1.93088710e-01 -2.95774609e-01
-3.65911543e-01 8.13651919e-01 -3.04053277e-01 4.51609045e-01
1.15398049e-01 1.45149961e-01 -3.21993493e-02 -9.93063390e-01
1.31660968e-01 1.40089259e-01 -4.45503503e-01 1.38715371e-01
3.74201626e-01 1.17545927e+00 -5.84124267e-01 -7.25532025e-02
6.40226066e-01 4.25528646e-01 -4.01305556e-01 3.68006319e-01
-1.14508905e-01 6.41232431e-01 -6.30733296e-02 6.83266282e-01
2.27586612e-01 4.91742551e-01 -3.13932985e-01 3.58108312e-01
-3.94846171e-01 1.22981644e+00 -5.28113484e-01 1.99114132e+00
-1.16041422e+00 7.93270588e-01 3.67008924e-01 -1.00216377e+00
6.70139790e-01 9.37889755e-01 -2.67837048e-01 -7.41428018e-01
7.24308714e-02 5.02199650e-01 3.11585665e-01 -5.51939189e-01
1.21169753e-01 -1.29146099e-01 6.52963892e-02 5.30164093e-02
1.81220412e-01 -2.51390755e-01 -4.32802439e-01 -3.84019405e-01
9.80370462e-01 -6.36894926e-02 1.17338091e-01 -1.21884428e-01
5.88229120e-01 -5.14891922e-01 -1.09878201e-02 6.31529570e-01
-2.52413958e-01 7.75101542e-01 -2.97309369e-01 1.32967383e-01
-8.34300578e-01 -1.29494119e+00 -8.45872611e-02 1.92418146e+00
-8.10889900e-01 -2.18386725e-01 -8.22155893e-01 -4.08774316e-01
-1.78085536e-01 9.20020580e-01 -1.68573961e-01 -8.95542949e-02
-8.41736376e-01 1.64426655e-01 1.08131564e+00 7.39439249e-01
1.64567426e-01 -1.87137175e+00 1.85632795e-01 3.98033798e-01
-1.83889702e-01 -1.24551749e+00 -8.96857083e-01 4.81759667e-01
-4.02323723e-01 -2.44671300e-01 -1.10828829e+00 -1.32916737e+00
-1.26418814e-01 1.22074127e-01 9.66097832e-01 -3.37905139e-01
1.63852572e-01 2.13846207e-01 -2.34242678e-01 -5.52723825e-01
-7.17120051e-01 4.25955117e-01 5.08441985e-01 -4.48528193e-02
3.16309750e-01 -6.33412361e-01 -1.83799282e-01 2.22798064e-01
-1.38670549e-01 -2.30296291e-02 7.39068091e-01 6.41657889e-01
7.49235690e-01 -6.82974339e-01 1.07826102e+00 -6.56845510e-01
7.03409851e-01 -3.29952687e-01 -9.79383439e-02 2.53049076e-01
4.81631048e-02 -2.27441952e-01 9.84609604e-01 -5.58047295e-01
-1.00666606e+00 -5.26777469e-02 -9.55620468e-01 -9.01643932e-01
-6.91061616e-01 3.16763073e-01 -5.92893004e-01 6.30599678e-01
2.15610668e-01 6.07271791e-01 -1.75434187e-01 -1.07850790e+00
6.20169580e-01 1.40114594e+00 9.65297103e-01 -4.26232487e-01
4.59258676e-01 -2.41814643e-01 -8.07886124e-01 -1.37524378e+00
-5.69752514e-01 -7.54136562e-01 -6.12763941e-01 2.52396077e-01
8.41501832e-01 -1.11847878e+00 -9.23695326e-01 6.38688445e-01
-1.43627417e+00 -5.57481170e-01 -3.38473767e-01 8.51380110e-01
-8.92847359e-01 -1.22736365e-01 -6.07468963e-01 -1.04415226e+00
-4.68813419e-01 -1.22692323e+00 1.42024934e+00 -2.81565696e-01
-3.42727810e-01 -8.86137068e-01 1.27303854e-01 5.96478164e-01
6.96703970e-01 -6.75195456e-01 8.95356238e-01 -9.07423556e-01
-3.74615788e-01 3.44017267e-01 6.60246685e-02 8.26546907e-01
2.32952967e-01 -2.20069632e-01 -1.51301396e+00 -2.34695479e-01
-2.23846361e-01 -4.31748807e-01 9.86284375e-01 6.94233716e-01
1.40473628e+00 -4.69115704e-01 -1.75387919e-01 6.62441850e-01
4.81340349e-01 4.66056913e-01 5.52167714e-01 -4.47771549e-01
9.68303442e-01 4.66385633e-01 2.18993500e-01 -1.11530222e-01
3.03975910e-01 8.96877050e-01 -1.14836395e-01 1.53821465e-02
-1.03363669e+00 -4.24033791e-01 9.53446746e-01 2.02440047e+00
7.10460395e-02 -5.16653001e-01 -9.38273907e-01 9.36326563e-01
-1.46816552e+00 -8.24400842e-01 -1.40361100e-01 2.14016914e+00
9.68138635e-01 4.18308713e-02 1.54486895e-01 2.18596399e-01
8.10904443e-01 1.99179113e-01 -3.21893752e-01 -8.43450129e-01
-8.26970041e-02 7.13475704e-01 -3.04207858e-02 7.22793102e-01
-9.57275927e-01 1.64547503e+00 5.78537035e+00 1.64206076e+00
-1.20967901e+00 7.12855697e-01 4.55470473e-01 7.83225428e-03
-2.39936545e-01 -3.95216197e-01 -9.56537843e-01 5.99618614e-01
1.61074054e+00 2.18065381e-01 7.44437516e-01 8.14483345e-01
3.71355712e-01 3.96659076e-01 -1.32867134e+00 1.18106675e+00
1.26775682e-01 -1.17094851e+00 -3.66411246e-02 1.95687097e-02
4.08923060e-01 6.75560892e-01 4.29021269e-01 7.02193201e-01
1.47062644e-01 -1.31021142e+00 9.04738188e-01 3.34431976e-01
1.55942321e+00 -6.69014752e-01 9.89010930e-02 4.81030285e-01
-1.38982606e+00 3.14680040e-01 -1.72726601e-01 -7.17795417e-02
2.37758219e-01 1.04734421e-01 -1.23622823e+00 1.98074117e-01
2.93269694e-01 6.43753290e-01 -2.12847218e-01 8.75651181e-01
-3.68851900e-01 1.49957645e+00 -3.11354697e-01 -3.19869220e-01
1.99491888e-01 2.74504572e-01 6.66444957e-01 1.74220920e+00
4.22635376e-01 -2.19311908e-01 7.85382167e-02 3.34456414e-01
-4.22989994e-01 2.84541607e-01 -6.72629178e-01 -3.52941394e-01
7.79284656e-01 7.79826164e-01 -3.82550120e-01 -2.92574286e-01
-2.64014214e-01 1.25967813e+00 5.25098622e-01 3.43674660e-01
-4.25436646e-01 -2.63990521e-01 9.72717166e-01 -3.65559459e-02
4.32393968e-01 -5.20693123e-01 -8.92879739e-02 -8.77816916e-01
-2.44675763e-03 -9.49355543e-01 -2.38031268e-01 -6.69616580e-01
-1.34470189e+00 6.70629740e-01 -3.17074269e-01 -8.23857248e-01
-5.88112891e-01 -7.95348763e-01 -8.98698628e-01 1.23087728e+00
-1.70463693e+00 -1.26294744e+00 5.56151628e-01 7.70245612e-01
1.58477926e+00 -6.47510052e-01 1.06509697e+00 5.38130701e-01
-5.95056534e-01 7.99082577e-01 4.38515872e-01 1.96049750e-01
6.36205673e-01 -1.31936777e+00 1.19586766e+00 6.88000023e-01
7.68883824e-01 4.77062821e-01 1.30395383e-01 -2.47847885e-01
-1.21355820e+00 -1.20610309e+00 1.32087934e+00 -6.68626606e-01
4.32565153e-01 -9.09649014e-01 -8.08922172e-01 7.05109239e-01
5.56693196e-01 -3.44515964e-02 5.89820445e-01 4.30673540e-01
-1.94780417e-02 -2.76242904e-02 -6.05340779e-01 5.64228535e-01
1.45406604e+00 -1.32793725e+00 -6.43254578e-01 5.47864616e-01
1.00281239e+00 -1.43401995e-01 -3.26326996e-01 1.43653201e-02
5.01858056e-01 -4.93287593e-01 1.03307807e+00 -7.00832427e-01
1.42318150e-02 1.01899847e-01 -3.78098041e-01 -1.83126938e+00
-2.43105695e-01 -1.19441855e+00 -8.59996304e-02 1.63247061e+00
6.08294010e-01 -4.24269706e-01 4.51632924e-02 -5.09806812e-01
-1.09561336e+00 -4.46195751e-01 -1.50624180e+00 -9.45165157e-01
2.84883201e-01 -1.00300336e+00 4.27652001e-01 7.84556925e-01
-2.46042848e-01 6.98477387e-01 -4.05213416e-01 -1.21191209e-02
-3.66397575e-02 -5.21577954e-01 6.22217774e-01 -9.77111995e-01
-4.70616549e-01 -5.45435190e-01 9.46709327e-03 -1.39252532e+00
4.59767252e-01 -1.23662651e+00 2.64418602e-01 -1.40377712e+00
-4.84987736e-01 -4.65982884e-01 -4.58102524e-01 6.63139701e-01
-1.01856276e-01 6.11014925e-02 2.89492518e-01 2.21216232e-01
-4.43599194e-01 8.47721338e-01 1.12125492e+00 -5.61921485e-03
-4.45663780e-02 4.57014143e-01 -2.57494628e-01 7.38445282e-01
6.35422587e-01 -4.34055567e-01 -2.65341789e-01 -9.29090559e-01
-5.33960164e-01 3.63225222e-01 -5.35831833e-03 -1.05772233e+00
2.38825366e-01 1.06602862e-01 5.69320358e-02 -7.56198823e-01
7.74442375e-01 -3.78865540e-01 -5.31683981e-01 1.00490570e-01
-5.74504435e-01 1.17964916e-01 1.76313162e-01 1.04984753e-01
-3.06419939e-01 -2.08554387e-01 6.13529921e-01 -2.35162135e-02
-4.23252553e-01 1.95956632e-01 -8.67398262e-01 3.64562541e-01
6.44550443e-01 -1.48744002e-01 -1.37639284e-01 -1.87426552e-01
-7.70098746e-01 -1.39268130e-01 -3.98739904e-01 8.49982381e-01
5.92469811e-01 -1.23306882e+00 -1.01182961e+00 6.09421670e-01
1.60807837e-02 -1.82818711e-01 2.32528858e-02 7.43753433e-01
5.85341863e-02 7.70295322e-01 1.49124697e-01 -5.21373153e-01
-1.26817155e+00 3.02843750e-01 1.54067650e-01 1.02679715e-01
-3.01894307e-01 1.47281241e+00 2.62947679e-01 -6.34111464e-01
4.69602555e-01 -6.78332210e-01 2.70406716e-02 4.37121093e-03
4.61394817e-01 4.04268473e-01 4.00458038e-01 -8.30149889e-01
-6.65899873e-01 3.52232426e-01 -2.86030650e-01 -6.11232638e-01
1.04848099e+00 -3.86411011e-01 5.72607636e-01 8.64668310e-01
1.21015513e+00 5.41444421e-01 -1.05836511e+00 -2.82906681e-01
-2.03112252e-02 -2.18412369e-01 1.83126837e-01 -1.15344954e+00
-5.96291304e-01 1.66871095e+00 4.11185473e-01 1.00635551e-01
5.86002529e-01 1.97845280e-01 1.27756763e+00 4.64194238e-01
-1.27764165e-01 -1.24698460e+00 -1.50822988e-02 1.13512337e+00
1.13051450e+00 -1.28643286e+00 -1.07601953e+00 -1.18654504e-01
-7.94491887e-01 7.10382760e-01 2.39356712e-01 5.04333377e-02
5.69809616e-01 3.47239792e-01 1.14482880e-01 3.25163633e-01
-7.71179020e-01 -6.97566569e-01 4.73609746e-01 9.02088225e-01
8.86109293e-01 4.37975675e-01 2.30493516e-01 6.19042635e-01
-6.44043684e-01 -4.30460721e-01 -2.52964169e-01 3.87984782e-01
-5.35833180e-01 -1.03845596e+00 -3.10660660e-01 3.21037650e-01
-5.86381853e-01 -9.02012825e-01 -3.40649605e-01 3.02877009e-01
1.25303134e-01 1.14211941e+00 2.09963128e-01 -1.36698574e-01
4.54421908e-01 6.60682201e-01 2.61826813e-01 -1.09319210e+00
-8.94518018e-01 5.24765134e-01 2.85015881e-01 -1.31812552e-02
1.53591380e-01 -8.98352742e-01 -1.20110261e+00 1.84115738e-01
-3.26098830e-01 6.01317286e-02 8.66117835e-01 1.20655179e+00
3.76907021e-01 8.85242701e-01 6.89577520e-01 -9.41575408e-01
-4.21154886e-01 -1.54467678e+00 -2.68082708e-01 8.66206959e-02
7.05920577e-01 -3.43619466e-01 -1.29300594e-01 5.96097335e-02]
|
[14.530191421508789, 6.87267541885376]
|
c398fa13-d2c5-4ac2-be04-05ef2b963eac
|
discoman-dataset-of-indoor-scenes-for
|
1909.12146
| null |
https://arxiv.org/abs/1909.12146v1
|
https://arxiv.org/pdf/1909.12146v1.pdf
|
DISCOMAN: Dataset of Indoor SCenes for Odometry, Mapping And Navigation
|
We present a novel dataset for training and benchmarking semantic SLAM methods. The dataset consists of 200 long sequences, each one containing 3000-5000 data frames. We generate the sequences using realistic home layouts. For that we sample trajectories that simulate motions of a simple home robot, and then render the frames along the trajectories. Each data frame contains a) RGB images generated using physically-based rendering, b) simulated depth measurements, c) simulated IMU readings and d) ground truth occupancy grid of a house. Our dataset serves a wider range of purposes compared to existing datasets and is the first large-scale benchmark focused on the mapping component of SLAM. The dataset is split into train/validation/test parts sampled from different sets of virtual houses. We present benchmarking results forboth classical geometry-based and recent learning-based SLAM algorithms, a baseline mapping method, semantic segmentation and panoptic segmentation.
|
['Anton Konushin', 'Sergey Bykov', 'Konstantin Sofiiuk', 'Igor Slinko', 'Dmitry Zhukov', 'Pavel Kirsanov', 'Filipp Konokhov', 'Olga Barinova', 'Anna Vorontsova', 'Airat Gaskarov']
|
2019-09-26
| null | null | null | null |
['semantic-slam']
|
['computer-vision']
|
[ 3.58739287e-01 -1.27444714e-01 3.03774595e-01 -5.19993126e-01
-7.89459288e-01 -4.17335808e-01 9.06060994e-01 7.52507001e-02
-4.63051081e-01 1.09556234e+00 8.00129846e-02 -5.56569844e-02
1.14063077e-01 -1.14421201e+00 -8.11177433e-01 -6.04513884e-01
-3.46919537e-01 1.28397644e+00 5.15075624e-01 -3.38448882e-01
6.64228126e-02 5.36082268e-01 -1.74769354e+00 -1.07602790e-01
7.82959998e-01 8.53299081e-01 7.21620739e-01 7.42737591e-01
-4.23015729e-02 5.97392559e-01 -5.11202276e-01 1.00041166e-01
3.91290516e-01 -6.31474555e-01 -8.30425322e-01 3.89222860e-01
3.01333159e-01 -2.95764208e-01 -3.24561685e-01 8.71495128e-01
5.19037366e-01 4.18734699e-01 3.74765992e-01 -1.53117001e+00
8.09986442e-02 1.37718767e-01 -2.29273066e-01 -1.37991294e-01
1.05121100e+00 1.10499687e-01 7.22890735e-01 -5.22046387e-01
1.16334403e+00 1.25802696e+00 8.24522913e-01 1.60019010e-01
-1.30748606e+00 -2.90833056e-01 -1.17412545e-01 1.31504729e-01
-1.50386655e+00 -2.23212272e-01 5.06481528e-01 -4.15627897e-01
1.06350529e+00 3.40772212e-01 1.26803625e+00 1.06431293e+00
2.14564592e-01 6.85051322e-01 1.28615260e+00 -1.66404843e-01
8.09949338e-01 -2.14507788e-01 -4.04704183e-01 8.80520821e-01
-4.27847579e-02 2.66856700e-01 -4.16029304e-01 -3.18048626e-01
9.27061975e-01 5.39300703e-02 -4.23504949e-01 -8.37107837e-01
-1.64953482e+00 9.21095073e-01 7.25006282e-01 -1.91609159e-01
-3.63170117e-01 3.34414124e-01 2.16646329e-01 1.29378736e-01
1.62601799e-01 5.99433817e-02 -3.83612633e-01 -1.47088110e-01
-1.18268418e+00 6.37381196e-01 8.85742426e-01 1.32166684e+00
1.20050526e+00 -1.16691537e-01 3.74791831e-01 4.33035851e-01
4.45717812e-01 8.92066896e-01 4.78023082e-01 -1.31239283e+00
3.95140946e-01 2.13279992e-01 4.76312011e-01 -9.10108864e-01
-5.84726036e-01 4.13547695e-01 -6.39534950e-01 1.77624837e-01
1.93651050e-01 1.02949897e-02 -1.07041109e+00 1.26609099e+00
5.63835442e-01 5.80683529e-01 -5.87726757e-02 1.00060022e+00
7.56461561e-01 9.37056422e-01 -2.11186141e-01 3.78587954e-02
1.03558588e+00 -1.07374549e+00 -4.42009687e-01 -3.76544148e-01
6.08263373e-01 -6.14100635e-01 9.23854947e-01 2.28509918e-01
-6.73057914e-01 -4.50166047e-01 -1.17198634e+00 8.69839266e-02
-4.62802291e-01 -1.87693492e-01 7.80192375e-01 4.06415671e-01
-1.41899765e+00 7.64005065e-01 -1.06832457e+00 -1.01257753e+00
1.32458046e-01 1.78444058e-01 -4.95194495e-01 -1.82132289e-01
-9.38613117e-01 7.63300717e-01 4.72794950e-01 -1.15431361e-01
-1.16800666e+00 -2.52413273e-01 -1.36184883e+00 -6.31385446e-01
-6.23927731e-03 -7.90525198e-01 1.29454517e+00 -5.11042297e-01
-1.35466373e+00 1.18134415e+00 -1.65036798e-01 -6.49405479e-01
9.07683432e-01 -2.58893669e-01 -1.54673651e-01 2.74653304e-02
6.77078843e-01 1.14380479e+00 5.36591589e-01 -1.61896920e+00
-8.13003719e-01 -2.62094229e-01 -1.78896517e-01 3.05094749e-01
9.43346262e-01 -3.72163862e-01 -4.95417774e-01 -2.84601688e-01
5.51992714e-01 -1.09751379e+00 -6.67696118e-01 -1.58890083e-01
-6.57287598e-01 4.96989906e-01 6.37114286e-01 -5.29634237e-01
6.37933254e-01 -1.75225949e+00 1.33067891e-01 1.91937000e-01
-2.81191915e-01 -3.25251639e-01 7.54777864e-02 6.42301857e-01
1.43228501e-01 -1.80582136e-01 -5.51288545e-01 -7.79804826e-01
1.76332489e-01 7.00069964e-01 -4.47630703e-01 6.84220076e-01
-4.22195017e-01 9.44007099e-01 -1.13060617e+00 -6.74271405e-01
9.25689757e-01 2.26385772e-01 -2.55649030e-01 2.31836200e-01
-5.07657647e-01 8.06720674e-01 -3.07977080e-01 6.47399604e-01
7.91865587e-01 -6.91631809e-02 9.18106958e-02 2.67026484e-01
-1.76364377e-01 2.42881685e-01 -1.27878547e+00 2.42495751e+00
-3.60221952e-01 4.49899405e-01 -2.54949272e-01 -5.18727064e-01
9.74270999e-01 1.77209228e-02 6.36921763e-01 -7.49949992e-01
8.84770751e-02 3.01577955e-01 -6.79828525e-01 -1.89498365e-01
6.76292896e-01 1.21997938e-01 -4.08036858e-01 3.62637281e-01
1.10019911e-02 -8.96967173e-01 1.07336223e-01 2.98725870e-02
1.17068362e+00 7.36748755e-01 3.96309018e-01 -3.52460235e-01
3.23819071e-01 7.31739044e-01 3.79852265e-01 6.48584545e-01
-1.65894702e-01 1.00024962e+00 8.55968744e-02 -9.37110603e-01
-1.52304196e+00 -1.32393026e+00 -1.08096272e-01 7.03506231e-01
6.91468418e-01 -5.79012811e-01 -9.04464662e-01 -1.61010414e-01
1.45962864e-01 7.01710045e-01 -6.27114892e-01 3.76179427e-01
-5.92862308e-01 -8.24651420e-01 2.55067706e-01 3.13725412e-01
9.05909479e-01 -1.29226112e+00 -1.39725184e+00 2.93961406e-01
-3.05006355e-01 -1.26624608e+00 7.57110817e-03 3.25351328e-01
-7.83639014e-01 -1.24306643e+00 -3.43760133e-01 -7.61368394e-01
4.80564922e-01 1.34910271e-01 1.40142107e+00 -2.23665372e-01
-3.86514664e-01 5.22950113e-01 -3.95976096e-01 -1.72317073e-01
-2.77182698e-01 7.72896931e-02 -1.33355418e-02 -4.30715591e-01
1.26601711e-01 -7.10229099e-01 -4.28188503e-01 4.60662335e-01
-4.68566567e-01 4.61726964e-01 2.03029183e-03 4.73363400e-01
1.23994792e+00 -7.86818862e-02 -2.86797971e-01 -6.32537067e-01
1.54783010e-01 -5.06333232e-01 -9.65902388e-01 -6.19934164e-02
-1.61961187e-02 -3.34543020e-01 3.46023172e-01 9.16910768e-02
-6.44422770e-01 5.95165312e-01 -1.47322819e-01 -1.31397650e-01
-6.11890495e-01 -8.67575631e-02 -1.39068201e-01 7.11068138e-02
5.46699703e-01 1.88107416e-01 -3.44817251e-01 -2.32382774e-01
4.12103206e-01 1.97342426e-01 7.83587694e-01 -6.41325295e-01
8.22568953e-01 9.89703417e-01 -1.11426190e-01 -1.02880037e+00
-3.90956402e-01 -4.87319827e-01 -1.13353062e+00 -2.18464628e-01
1.06830204e+00 -8.98657799e-01 -3.15090150e-01 5.21032214e-01
-1.09610164e+00 -1.09009945e+00 -5.18481970e-01 3.40773731e-01
-1.31427073e+00 -1.20352492e-01 -4.11058992e-01 -5.67881823e-01
5.48858270e-02 -1.32959104e+00 1.70274091e+00 4.45933118e-02
-4.16837275e-01 -9.61735129e-01 4.70452935e-01 -1.25718871e-02
-1.76881123e-02 1.08475530e+00 2.74099708e-01 -1.69043303e-01
-9.12240088e-01 7.79962093e-02 9.11889449e-02 -2.72477657e-01
4.26086895e-02 -2.09430963e-01 -9.96853352e-01 -2.37318084e-01
-1.58683747e-01 -3.31541806e-01 5.72797537e-01 4.14381355e-01
5.48053682e-01 3.95284779e-02 -6.92000389e-01 9.71102774e-01
1.73911059e+00 3.68784636e-01 7.54957855e-01 8.49094570e-01
7.90836871e-01 4.00569677e-01 9.52958882e-01 5.96299648e-01
6.90064490e-01 7.30867565e-01 6.81350231e-01 -1.27104104e-01
2.59982914e-01 -4.72910523e-01 9.05311331e-02 4.12949443e-01
-3.93932424e-02 -1.56043962e-01 -1.33545995e+00 5.53553581e-01
-2.16962028e+00 -6.79663002e-01 -2.71973908e-01 2.27275062e+00
1.93977267e-01 -1.23856083e-01 1.58982977e-01 1.63203269e-01
4.47652429e-01 5.06524503e-01 -5.45317948e-01 1.78333651e-02
-1.18656971e-01 2.78672963e-01 6.97041690e-01 8.21658134e-01
-1.42625248e+00 1.38136280e+00 6.37750340e+00 2.25058421e-01
-7.41703033e-01 2.06374586e-01 3.50976169e-01 -3.26264650e-02
-1.86458945e-01 2.20712461e-02 -5.96396267e-01 3.74266118e-01
1.09058642e+00 1.62711143e-01 5.34628332e-01 1.04847789e+00
2.24527642e-01 -9.14576471e-01 -1.11441994e+00 1.27477956e+00
-1.66174322e-02 -1.33876693e+00 -2.41066247e-01 1.39731869e-01
1.07922399e+00 6.71663225e-01 -4.29175496e-01 6.71139210e-02
8.67605329e-01 -1.16721261e+00 1.24103284e+00 3.58707607e-01
6.46650732e-01 -7.85793006e-01 6.06345832e-01 3.62911791e-01
-1.50853670e+00 3.00003916e-01 -5.08111358e-01 -9.38953534e-02
6.88838840e-01 2.90088892e-01 -8.64535570e-01 5.81958115e-01
9.89711881e-01 8.31297815e-01 -5.78633070e-01 1.05993474e+00
-4.10331756e-01 2.63722297e-02 -6.56417191e-01 2.39936352e-01
5.94153345e-01 -7.44226336e-01 2.43459627e-01 1.09940505e+00
3.95369142e-01 -9.25842896e-02 6.27820790e-01 8.04965317e-01
3.37488323e-01 -1.65576786e-01 -1.10905993e+00 5.45497298e-01
5.86881936e-01 9.73555505e-01 -1.21153426e+00 -3.72133762e-01
2.05471311e-02 1.57718515e+00 -3.71857844e-02 3.64212424e-01
-1.06217873e+00 -6.15962334e-02 8.11952949e-01 7.45159164e-02
7.71523714e-02 -5.82904577e-01 -1.17085189e-01 -9.58704650e-01
-2.28091523e-01 -4.43268061e-01 1.12650990e-01 -1.09874475e+00
-5.48380613e-01 4.92349774e-01 9.51362178e-02 -1.06223106e+00
-4.91052687e-01 -2.57890761e-01 -2.32999146e-01 6.41624570e-01
-1.41263974e+00 -1.05988145e+00 -9.59428072e-01 5.01287997e-01
6.75652564e-01 1.79749876e-01 1.00342560e+00 8.83836374e-02
-1.57781154e-01 -3.97734791e-01 3.37246835e-01 6.79711401e-02
3.07112068e-01 -1.41574299e+00 1.19152844e+00 5.11299074e-01
2.39449337e-01 4.27307263e-02 8.20415139e-01 -8.69695425e-01
-1.14826107e+00 -1.40443826e+00 6.04924798e-01 -7.56291866e-01
4.99222636e-01 -6.78923070e-01 -4.84861374e-01 1.20205963e+00
-1.44508466e-01 1.65978432e-01 1.44660577e-01 -6.70040548e-01
3.83298516e-01 2.54892290e-01 -1.43118775e+00 4.00133133e-01
1.51635671e+00 -4.59254146e-01 -4.23330843e-01 5.36382198e-01
5.78418553e-01 -9.92388606e-01 -6.46284819e-01 2.59025961e-01
3.23529273e-01 -1.62366152e+00 1.12975776e+00 -1.85791589e-02
1.97939381e-01 -4.87186879e-01 -8.06597829e-01 -1.39184856e+00
4.26928215e-02 -5.30302107e-01 2.46105254e-01 8.31488431e-01
-6.71623424e-02 -6.38887346e-01 1.01577091e+00 5.21725416e-02
-9.81309488e-02 -6.03865504e-01 -1.03490865e+00 -7.57664680e-01
-3.26176584e-01 -5.60278416e-01 1.09852946e+00 8.10600162e-01
-6.17866337e-01 -4.36962917e-02 -5.93641661e-02 4.48997527e-01
7.74221778e-01 2.84593344e-01 1.16649866e+00 -1.20902479e+00
1.59470066e-01 -2.66404040e-02 -8.60577106e-01 -1.08769846e+00
1.30451679e-01 -6.29430652e-01 3.42302948e-01 -2.10107994e+00
-2.84137160e-01 -7.39009857e-01 5.03970623e-01 1.59037724e-01
3.89955252e-01 2.85643309e-01 4.20338102e-02 2.82372028e-01
-6.65544033e-01 6.44541323e-01 8.42875302e-01 3.30544040e-02
-2.27545172e-01 -2.65415549e-01 3.58420402e-01 9.13081110e-01
6.74679041e-01 -3.45938206e-01 -7.13136613e-01 -3.96168679e-01
-2.10931320e-02 2.16924652e-01 6.52854919e-01 -1.56689548e+00
-1.33443112e-02 -2.05144390e-01 4.87969667e-01 -1.01419330e+00
6.84322178e-01 -8.44935000e-01 7.69150019e-01 6.86143458e-01
3.59311134e-01 4.35068697e-01 3.79169057e-03 4.98593867e-01
1.51414294e-02 1.34867072e-01 7.75321186e-01 -6.92114174e-01
-1.42718375e+00 3.53854269e-01 -1.72080845e-01 4.17287424e-02
1.30002844e+00 -4.52606291e-01 5.75745711e-03 -4.56774026e-01
-8.88534963e-01 2.75991440e-01 1.29107249e+00 3.59478712e-01
7.91292012e-01 -1.49677348e+00 -3.54836136e-01 4.29424942e-01
-1.48784844e-02 6.61086857e-01 -1.14234582e-01 5.06715596e-01
-1.37361538e+00 2.81057239e-01 -2.69599915e-01 -1.12314415e+00
-7.56174803e-01 2.96421885e-01 6.67518795e-01 9.74460021e-02
-1.03136551e+00 8.21054816e-01 1.04094155e-01 -1.03360188e+00
4.83231526e-03 -5.82749426e-01 2.25717098e-01 1.84454378e-02
3.13440502e-01 5.39098620e-01 9.42631364e-02 -1.18467414e+00
-5.05743444e-01 6.17187500e-01 7.19472170e-01 -5.45982778e-01
1.37653601e+00 -3.39424670e-01 -7.58801252e-02 8.97516131e-01
9.87616777e-01 -5.37847042e-01 -1.43841887e+00 1.10677898e-01
2.23598316e-01 -5.87775767e-01 -4.37908679e-01 -3.83699358e-01
-6.62370503e-01 5.89772522e-01 7.32684374e-01 -1.28956616e-01
7.05646038e-01 1.70232020e-02 9.58971560e-01 3.73539060e-01
1.59054387e+00 -9.90399659e-01 -2.54946034e-02 5.90996742e-01
6.94745243e-01 -1.27296245e+00 6.11737778e-04 -3.32405090e-01
-6.59595072e-01 7.84742415e-01 3.96325797e-01 -4.60509419e-01
3.06283593e-01 2.90171921e-01 1.71689108e-01 -3.01273495e-01
-1.45643666e-01 -3.95085156e-01 -4.08118755e-01 9.54539716e-01
-1.14337318e-01 3.18401933e-01 5.11297286e-01 -1.18502609e-01
-1.09024417e+00 -1.42734796e-01 3.35935861e-01 1.26184559e+00
-6.31146550e-01 -7.86141634e-01 -7.07375109e-01 1.07915439e-01
5.25945842e-01 1.36598781e-01 -2.13461161e-01 1.17082393e+00
3.72655988e-01 6.94229066e-01 3.66134316e-01 -3.90826613e-01
3.90697002e-01 -5.78513034e-02 6.35196865e-01 -5.16640246e-01
-9.93066728e-02 -2.42878884e-01 1.14942625e-01 -1.04039371e+00
-3.93105924e-01 -9.81981874e-01 -1.69462657e+00 -5.08229196e-01
2.45194227e-01 5.59840687e-02 7.36479700e-01 8.15962493e-01
3.20268329e-03 3.36754799e-01 4.04508233e-01 -1.70792806e+00
1.99922800e-01 -6.16056919e-01 -8.36506426e-01 5.58377683e-01
4.36342746e-01 -7.67580211e-01 -7.32234046e-02 1.27381399e-01]
|
[7.367717266082764, -2.1476480960845947]
|
97bb9d52-e091-411b-aef7-35b3440b8704
|
toward-real-world-single-image-deraining-a
|
2206.05514
| null |
https://arxiv.org/abs/2206.05514v2
|
https://arxiv.org/pdf/2206.05514v2.pdf
|
Toward Real-world Single Image Deraining: A New Benchmark and Beyond
|
Single image deraining (SID) in real scenarios attracts increasing attention in recent years. Due to the difficulty in obtaining real-world rainy/clean image pairs, previous real datasets suffer from low-resolution images, homogeneous rain streaks, limited background variation, and even misalignment of image pairs, resulting in incomprehensive evaluation of SID methods. To address these issues, we establish a new high-quality dataset named RealRain-1k, consisting of $1,120$ high-resolution paired clean and rainy images with low- and high-density rain streaks, respectively. Images in RealRain-1k are automatically generated from a large number of real-world rainy video clips through a simple yet effective rain density-controllable filtering method, and have good properties of high image resolution, background diversity, rain streaks variety, and strict spatial alignment. RealRain-1k also provides abundant rain streak layers as a byproduct, enabling us to build a large-scale synthetic dataset named SynRain-13k by pasting the rain streak layers on abundant natural images. Based on them and existing datasets, we benchmark more than 10 representative SID methods on three tracks: (1) fully supervised learning on RealRain-1k, (2) domain generalization to real datasets, and (3) syn-to-real transfer learning. The experimental results (1) show the difference of representative methods in image restoration performance and model complexity, (2) validate the significance of the proposed datasets for model generalization, and (3) provide useful insights on the superiority of learning from diverse domains and shed lights on the future research on real-world SID. The datasets will be released at https://github.com/hiker-lw/RealRain-1k
|
['DaCheng Tao', 'Xinmei Tian', 'Zhen Huang', 'Jing Zhang', 'Qiming Zhang', 'Wei Li']
|
2022-06-11
| null | null | null | null |
['single-image-deraining']
|
['computer-vision']
|
[ 1.12001806e-01 -7.20730543e-01 3.05356652e-01 -5.29141963e-01
-8.34598839e-01 -4.25515652e-01 3.24483126e-01 -5.51691294e-01
-1.96494564e-01 1.13286757e+00 9.52200126e-03 -1.30451426e-01
-2.03636497e-01 -7.27823734e-01 -7.59631932e-01 -1.27242076e+00
-2.20932424e-01 2.29100093e-01 1.16722845e-01 -4.21878248e-01
-1.53574720e-01 5.14469862e-01 -1.63607860e+00 1.19146638e-01
1.43044484e+00 7.20239103e-01 6.33456469e-01 6.07944727e-01
2.81915247e-01 7.98308492e-01 -5.83904386e-01 -4.36330959e-02
4.62779015e-01 -6.81183338e-01 -1.82602197e-01 2.54963249e-01
9.41536188e-01 -4.76471186e-01 -3.04708540e-01 1.18227077e+00
6.20590210e-01 -9.16380249e-03 3.71512383e-01 -9.77402210e-01
-7.67446876e-01 7.69967139e-02 -7.44547009e-01 5.29787183e-01
2.83903554e-02 3.99670601e-01 7.16454029e-01 -1.05438006e+00
5.28919637e-01 1.11930752e+00 6.97509110e-01 2.72664338e-01
-1.07682335e+00 -1.03360844e+00 9.11889002e-02 2.98199713e-01
-1.24578679e+00 -4.93468225e-01 5.01845717e-01 -3.45578134e-01
2.06061825e-01 3.98279756e-01 8.03572118e-01 1.13242364e+00
-7.02885538e-02 5.54617107e-01 1.88901377e+00 -1.20996535e-01
1.46794215e-01 -1.01819590e-01 2.12146670e-01 4.52403903e-01
7.21120417e-01 4.27007288e-01 -2.46099457e-01 1.12300649e-01
8.83792043e-01 1.53082833e-01 -9.69071686e-01 -1.56686842e-01
-1.10349929e+00 7.88649678e-01 6.53200150e-01 -1.85651451e-01
-2.69301116e-01 -4.88497496e-01 -1.80951748e-02 5.89904964e-01
5.62172055e-01 3.83420855e-01 -3.44834238e-01 3.43610168e-01
-9.47076440e-01 5.66539884e-01 4.97665375e-01 7.72194803e-01
1.02679920e+00 5.02863884e-01 8.56860802e-02 1.15496051e+00
2.60454100e-02 1.48845291e+00 2.32961982e-01 -9.33831632e-01
5.79305112e-01 -3.56033109e-02 6.13802314e-01 -9.10300195e-01
-2.62547225e-01 -3.20272803e-01 -1.35459113e+00 3.05501908e-01
3.35245371e-01 -9.81461331e-02 -1.14822924e+00 1.41336143e+00
1.34668142e-01 4.90100682e-01 3.96287084e-01 1.39252532e+00
9.52767730e-01 9.25572395e-01 -3.08895528e-01 -5.20756900e-01
1.07481873e+00 -9.32276428e-01 -6.61481321e-01 -3.28003824e-01
-3.08515821e-02 -7.84585297e-01 1.22550929e+00 4.71608549e-01
-6.49451494e-01 -8.40366185e-01 -9.56727207e-01 4.82584596e-01
-1.05684355e-01 2.07346514e-01 5.73331773e-01 4.30510193e-01
-8.75944912e-01 3.19624871e-01 -4.57544386e-01 -4.12807584e-01
4.79928613e-01 -3.98337156e-01 -2.74543315e-01 -8.18256021e-01
-1.30472589e+00 6.56934142e-01 8.17766115e-02 6.94859803e-01
-1.25844622e+00 -8.91318679e-01 -8.50935340e-01 -4.26175982e-01
7.08880136e-03 -5.21244645e-01 7.03646898e-01 -1.00836766e+00
-9.54740107e-01 8.45288694e-01 1.69003271e-02 -2.14962587e-01
4.09467220e-01 -5.16132534e-01 -9.42931771e-01 3.00151199e-01
1.22674160e-01 2.33014971e-01 1.24740005e+00 -1.87136662e+00
-6.91166699e-01 -2.13974774e-01 -1.42790094e-01 2.53718823e-01
8.32989514e-02 -1.60693690e-01 -1.09064676e-01 -9.10791874e-01
-9.24968496e-02 -7.79638648e-01 -1.81977674e-01 -5.44573069e-02
-1.56798899e-01 6.35819912e-01 8.03928077e-01 -9.24476683e-01
8.33588123e-01 -2.14483380e+00 -1.48997650e-01 -1.85322389e-01
2.70931840e-01 8.20765197e-01 -4.92319047e-01 2.35293359e-01
-2.38002107e-01 -2.30981559e-01 -6.52619481e-01 3.39448638e-02
-5.88663816e-01 4.98747796e-01 -6.85332656e-01 6.28791451e-01
1.90599382e-01 5.51567793e-01 -9.73831832e-01 -3.08930755e-01
3.39349449e-01 3.53329331e-01 -1.96165890e-02 7.50565708e-01
-6.39521629e-02 5.99071443e-01 -3.00728261e-01 9.03720677e-01
1.42131257e+00 -3.67229842e-02 -1.01770081e-01 -4.29631114e-01
-2.72964332e-02 -2.51479208e-01 -1.16328096e+00 1.02561033e+00
-5.28622091e-01 5.82001925e-01 3.44040155e-01 -9.61422682e-01
1.12362540e+00 -2.66577378e-02 1.47012630e-02 -1.11842382e+00
-4.21265066e-01 3.17205936e-01 -1.74874112e-01 -8.79720688e-01
2.74582952e-01 -2.76795238e-01 3.40064496e-01 1.13820434e-01
-2.45700166e-01 -3.05387259e-01 2.34690249e-01 6.20046519e-02
6.82163715e-01 -7.06738904e-02 -9.40249395e-03 -1.96409181e-01
2.53339171e-01 -5.56696802e-02 8.22984755e-01 9.34449971e-01
-2.65222639e-01 1.22106838e+00 -1.51971772e-01 -6.02989614e-01
-1.08967018e+00 -1.47887266e+00 -4.34712023e-01 5.19594312e-01
5.25739789e-01 2.64663309e-01 -3.11056703e-01 -3.75982255e-01
4.59974743e-02 4.53132868e-01 -6.65518045e-01 2.63463795e-01
-7.08834887e-01 -1.63301551e+00 4.65088964e-01 2.30742544e-01
1.08187127e+00 -1.18204594e+00 -2.41442695e-01 -1.07354850e-01
-5.84610164e-01 -1.38162589e+00 -1.23145424e-01 -2.35392433e-02
-8.78910840e-01 -1.27331710e+00 -8.72222185e-01 -5.95289886e-01
5.42096198e-01 9.95610535e-01 1.48981464e+00 -2.20339727e-02
-4.83988762e-01 -7.52298683e-02 -7.51140654e-01 -1.15845487e-01
-2.28784293e-01 -6.68777108e-01 1.22358158e-01 1.03021458e-01
-1.26514018e-01 -7.50535727e-01 -7.48964846e-01 5.64346015e-01
-1.03213310e+00 1.93866435e-02 7.10630655e-01 1.09949422e+00
5.46661735e-01 1.65007785e-01 5.92231631e-01 -8.69010448e-01
1.84722945e-01 -5.52239597e-01 -8.33677530e-01 1.86468542e-01
-4.40946609e-01 -4.63708043e-01 6.47746444e-01 -1.40686586e-01
-1.45509386e+00 -3.48681211e-01 6.33983389e-02 -6.02160156e-01
-2.24662498e-01 3.18188876e-01 -2.61429012e-01 -7.67627824e-03
8.62939000e-01 6.57064915e-01 -9.21757594e-02 -6.15176439e-01
3.77561152e-01 6.31591678e-01 7.97134042e-01 -4.44528639e-01
1.40286553e+00 7.80305326e-01 -4.51795757e-01 -1.20948839e+00
-1.15044522e+00 -4.48400468e-01 -1.74264401e-01 -1.24688342e-01
5.98775864e-01 -1.60350657e+00 8.61464143e-02 1.15219307e+00
-5.70600450e-01 -6.54632032e-01 -1.99995309e-01 4.30160880e-01
-9.00660157e-02 5.46829820e-01 -6.56848490e-01 -7.33468413e-01
-5.05335748e-01 -8.15410554e-01 7.63907015e-01 5.48602879e-01
5.37500083e-01 -7.55373299e-01 1.48977324e-01 5.80821514e-01
5.02680123e-01 2.88729668e-01 5.65915406e-01 1.57022789e-01
-8.20681810e-01 1.50934562e-01 -6.71945870e-01 8.21434617e-01
4.66314673e-01 3.04821059e-02 -8.86298239e-01 -7.11095095e-01
6.29892126e-02 -6.27484500e-01 1.11447144e+00 5.12294292e-01
8.08464646e-01 -2.48912036e-01 7.78902695e-02 9.49821413e-01
1.71371627e+00 -7.57071450e-02 1.00116181e+00 4.21959281e-01
7.31508970e-01 3.14782709e-01 1.14688110e+00 3.94645154e-01
2.39927933e-01 5.13479292e-01 5.48900008e-01 -6.03963017e-01
-4.80385244e-01 3.20533253e-02 4.42754269e-01 8.18783581e-01
-3.06215256e-01 -2.46786997e-01 -4.99733180e-01 8.89019847e-01
-1.60808671e+00 -1.17695129e+00 -1.78627729e-01 2.23228574e+00
8.42487276e-01 -2.52513945e-01 -1.39155433e-01 -2.21424967e-01
5.99547684e-01 7.08661139e-01 -4.63671207e-01 3.27767074e-01
-8.18698406e-01 2.45233625e-01 6.01217210e-01 4.83625680e-01
-1.20457602e+00 8.48628163e-01 5.36625290e+00 7.48177409e-01
-1.22966671e+00 8.79657269e-02 5.42160392e-01 6.88617527e-02
-2.03578949e-01 -2.11151883e-01 -9.34640050e-01 7.32232213e-01
5.81550121e-01 1.30459756e-01 5.04363775e-01 5.23496687e-01
6.15345299e-01 -1.14946388e-01 -3.93969148e-01 9.57802415e-01
1.64251283e-01 -1.18896449e+00 1.01213187e-01 -3.41368020e-01
1.04075873e+00 4.61779594e-01 -4.62347157e-02 2.21207529e-01
6.22412384e-01 -1.09518552e+00 2.80790061e-01 5.98940551e-01
9.48903084e-01 -3.24626833e-01 8.09389174e-01 2.15274483e-01
-1.01105797e+00 3.97087308e-03 -7.65984833e-01 8.58163759e-02
6.94257244e-02 1.16407204e+00 -3.84920180e-01 1.04479492e+00
1.37074292e+00 1.12432373e+00 -4.87243563e-01 1.21837401e+00
-5.33061802e-01 1.01752627e+00 -2.34174073e-01 6.81019545e-01
-1.92699842e-02 -5.98185003e-01 5.54460168e-01 1.35125327e+00
3.82846594e-01 2.89636850e-01 1.32710576e-01 5.20956993e-01
9.71341506e-02 -3.54663730e-01 -5.21101534e-01 2.65422732e-01
5.25530815e-01 1.28340948e+00 -2.62686580e-01 -2.92014837e-01
-3.86616498e-01 8.86110187e-01 -8.84913951e-02 8.52168441e-01
-9.64140713e-01 -2.50291884e-01 1.26509714e+00 2.73711868e-02
5.31317770e-01 -1.23153597e-01 -1.82326150e-03 -1.50460255e+00
1.71606451e-01 -1.39109278e+00 2.49548942e-01 -1.14103866e+00
-1.63771892e+00 9.18842137e-01 -5.62663050e-03 -1.55850661e+00
2.25107193e-01 -3.86885822e-01 -5.91567695e-01 1.08302951e+00
-2.31002140e+00 -1.07340658e+00 -1.24751616e+00 6.62846208e-01
6.60342574e-01 -1.05882071e-01 6.69425070e-01 4.48746711e-01
-7.08735466e-01 2.67544478e-01 6.39933228e-01 3.37612107e-02
1.10794318e+00 -1.02752364e+00 3.34919840e-01 1.28326011e+00
1.17745558e-02 1.49126530e-01 8.45753849e-01 -4.10104603e-01
-1.22027981e+00 -1.61577058e+00 4.30733711e-01 -1.81719974e-01
4.01921183e-01 -2.14195535e-01 -1.36660230e+00 4.60580230e-01
-1.43802851e-01 5.64724267e-01 8.52837116e-02 -1.95673198e-01
-5.89126348e-01 -7.90263951e-01 -1.09661138e+00 4.34947312e-01
8.39899957e-01 -1.33022919e-01 -5.71826160e-01 5.31590700e-01
6.30407453e-01 -4.94695067e-01 -5.84595084e-01 7.55384982e-01
2.47188777e-01 -1.41406631e+00 1.25798631e+00 -1.92803949e-01
7.19996035e-01 -6.14299715e-01 -4.21757758e-01 -1.57613397e+00
-2.96297759e-01 -2.19223946e-01 1.18757486e-01 1.13808501e+00
7.17050657e-02 -8.50773633e-01 3.55664819e-01 -1.91681162e-01
-1.06963493e-01 -3.84398758e-01 -6.04026556e-01 -9.07976210e-01
-1.96883492e-02 -1.44348247e-02 4.23920989e-01 1.17297530e+00
-9.36389327e-01 -3.18357255e-03 -9.07521427e-01 8.28131557e-01
1.05573237e+00 8.06870759e-01 9.40398812e-01 -1.03630328e+00
-2.10337192e-01 8.84627253e-02 4.27014008e-02 -1.13881648e+00
-1.52864501e-01 -3.86846244e-01 3.47504169e-01 -1.49165332e+00
2.95612812e-01 -7.37969041e-01 -2.91561425e-01 3.87029052e-01
-6.03646815e-01 5.97600937e-01 1.02332368e-01 6.74119234e-01
-3.86878878e-01 8.75258923e-01 1.50058782e+00 -8.43183175e-02
5.74507043e-02 5.54224141e-02 -5.11459768e-01 6.97937727e-01
7.78869033e-01 -2.25960687e-01 -3.34092110e-01 -6.51729345e-01
-1.92492247e-01 1.95316136e-01 5.48125565e-01 -1.18716764e+00
-4.70183134e-01 -4.95231301e-01 4.31800932e-01 -2.82704622e-01
3.92005175e-01 -5.36205411e-01 2.00269267e-01 1.74553558e-01
2.27711052e-01 -2.98353583e-01 8.56387708e-03 5.82961023e-01
-5.44691324e-01 1.34108201e-01 1.28384674e+00 -1.52087361e-01
-1.11561239e+00 4.07730073e-01 -1.50163118e-02 3.95565391e-01
5.60481548e-01 -1.83171630e-01 -8.42583537e-01 -4.05275404e-01
-4.13182825e-01 2.36625463e-01 6.02199495e-01 4.02326941e-01
7.27640450e-01 -8.64898682e-01 -1.30346680e+00 3.84397596e-01
1.99547514e-01 1.71132535e-01 7.08418429e-01 6.91904366e-01
-7.23453462e-01 -1.85073346e-01 -4.28199977e-01 -4.85941499e-01
-1.31817698e+00 1.82658866e-01 4.50771034e-01 -1.84964910e-01
-1.10489166e+00 8.16179991e-01 6.16455913e-01 -4.20007139e-01
-1.91903368e-01 -1.11534372e-01 -5.96271381e-02 -7.31054619e-02
7.98209548e-01 1.63243085e-01 1.05348013e-01 -4.67411071e-01
-2.09788620e-01 5.81896305e-01 -6.53593242e-02 4.49262142e-01
1.55530167e+00 -3.29251766e-01 -1.21813968e-01 2.47130811e-01
7.54885375e-01 6.66576950e-03 -1.59351599e+00 -4.38420445e-01
-6.34636700e-01 -1.03230417e+00 -1.37649342e-01 -1.00866199e+00
-1.58444607e+00 7.11140156e-01 9.72391784e-01 -9.16654915e-02
1.61941993e+00 -2.85802782e-01 8.72915208e-01 3.41033578e-01
3.08884412e-01 -4.90921766e-01 1.78747654e-01 5.19146979e-01
9.77573693e-01 -1.55036056e+00 1.55190095e-01 -3.46228808e-01
-7.88952053e-01 6.74814641e-01 6.33029819e-01 -1.99613422e-01
4.80149865e-01 2.62451798e-01 7.73785353e-01 -7.64784589e-02
-4.06790197e-01 -3.29164118e-01 -4.19277489e-01 9.90229130e-01
-3.79418470e-02 1.84114844e-01 5.35884276e-02 3.68151844e-01
-4.15392220e-02 8.24135020e-02 7.41108716e-01 5.99058330e-01
-6.14016473e-01 -5.80293894e-01 -7.96816587e-01 4.05576229e-01
1.35867437e-03 -3.62107873e-01 3.44537705e-01 8.12169075e-01
1.30779281e-01 1.05677783e+00 -2.09592119e-01 -2.94744819e-02
4.00994092e-01 -5.70414662e-01 3.87868136e-01 -1.28449067e-01
7.54411370e-02 -2.20513254e-01 4.05079238e-02 -4.26037461e-01
-6.08681381e-01 -6.82454765e-01 -6.48977816e-01 -2.61986375e-01
-4.75322865e-02 2.53603160e-01 1.49347886e-01 7.30974317e-01
1.81460023e-01 3.36492479e-01 9.33017135e-01 -1.13560581e+00
-4.62269992e-01 -1.03792024e+00 -1.22767508e+00 5.89354575e-01
7.15316355e-01 -5.87784827e-01 -7.76790261e-01 3.06154102e-01]
|
[10.899620056152344, -3.2632813453674316]
|
d3ed3b1b-c900-432a-a511-2c0cd1461fa6
|
error-bounds-of-projection-models-in-weakly
|
2010.12317
| null |
https://arxiv.org/abs/2010.12317v1
|
https://arxiv.org/pdf/2010.12317v1.pdf
|
Error Bounds of Projection Models in Weakly Supervised 3D Human Pose Estimation
|
The current state-of-the-art in monocular 3D human pose estimation is heavily influenced by weakly supervised methods. These allow 2D labels to be used to learn effective 3D human pose recovery either directly from images or via 2D-to-3D pose uplifting. In this paper we present a detailed analysis of the most commonly used simplified projection models, which relate the estimated 3D pose representation to 2D labels: normalized perspective and weak perspective projections. Specifically, we derive theoretical lower bound errors for those projection models under the commonly used mean per-joint position error (MPJPE). Additionally, we show how the normalized perspective projection can be replaced to avoid this guaranteed minimal error. We evaluate the derived lower bounds on the most commonly used 3D human pose estimation benchmark datasets. Our results show that both projection models lead to an inherent minimal error between 19.3mm and 54.7mm, even after alignment in position and scale. This is a considerable share when comparing with recent state-of-the-art results. Our paper thus establishes a theoretical baseline that shows the importance of suitable projection models in weakly supervised 3D human pose estimation.
|
['Rainer Lienhart', 'Stephan Brehm', 'Moritz Einfalt', 'Nikolas Klug']
|
2020-10-23
| null | null | null | null |
['monocular-3d-human-pose-estimation', 'weakly-supervised-3d-human-pose-estimation']
|
['computer-vision', 'computer-vision']
|
[ 6.09917752e-02 3.51523757e-01 -3.05976301e-01 -1.47757605e-01
-6.34970009e-01 -5.07070124e-01 6.54837012e-01 -1.52858227e-01
-7.17506289e-01 7.53441334e-01 1.94866821e-01 2.91194618e-01
-3.35787199e-02 -2.27797121e-01 -8.35578024e-01 -4.91874993e-01
-1.00680925e-01 1.07372332e+00 2.48910218e-01 -3.33298415e-01
3.78326885e-02 7.29543269e-01 -1.55513728e+00 -2.60936439e-01
3.21611971e-01 6.84087396e-01 -2.46566370e-01 5.51978111e-01
4.33731675e-01 1.67848602e-01 -6.88176751e-01 -3.95685792e-01
6.42299473e-01 -2.42330939e-01 -6.40638351e-01 2.13378102e-01
7.04852343e-01 -4.55751389e-01 -1.68808494e-02 6.50351763e-01
9.61747348e-01 -1.36227384e-01 7.47369707e-01 -1.26643240e+00
2.48811588e-01 5.79077750e-02 -5.60449779e-01 -2.05375671e-01
1.14595842e+00 2.66586943e-03 5.75008869e-01 -9.73264575e-01
1.08338189e+00 1.20496678e+00 9.17078197e-01 6.01689696e-01
-1.10344744e+00 -3.59598756e-01 -4.79157381e-02 -1.31055877e-01
-1.52520597e+00 -1.43843353e-01 7.28690922e-01 -6.91085935e-01
9.24432814e-01 4.37691599e-01 8.87099683e-01 1.26195979e+00
2.49932855e-01 6.11168742e-01 1.48352873e+00 -7.22493172e-01
-9.82683972e-02 2.48454630e-01 5.64876273e-02 7.22306430e-01
5.71424365e-01 2.13584751e-01 -7.07915306e-01 -9.95267183e-02
9.27257299e-01 -3.17090690e-01 -2.47335196e-01 -9.44296300e-01
-1.38665926e+00 4.20539409e-01 3.38357538e-01 -1.32932186e-01
-1.55365959e-01 3.01853959e-02 2.96678036e-01 4.06898977e-03
5.61740220e-01 2.85227031e-01 -3.90554786e-01 -1.18996546e-01
-7.69698620e-01 7.96865463e-01 7.60232866e-01 1.11127150e+00
4.85274523e-01 -3.90379399e-01 -5.49189164e-04 5.46499431e-01
3.57572705e-01 8.02370071e-01 -7.66912326e-02 -9.60222840e-01
5.48269749e-01 5.44105351e-01 4.31075603e-01 -9.34392989e-01
-8.86331379e-01 -5.61282456e-01 -4.44487482e-01 3.20989311e-01
7.79683888e-01 8.09830725e-02 -6.72203004e-01 1.65176952e+00
5.46727777e-01 -4.81396496e-01 -3.65787208e-01 1.23239684e+00
5.85672736e-01 2.54475810e-02 -3.07630658e-01 -1.07560366e-01
1.31470060e+00 -8.38690102e-01 -5.91309369e-01 -3.48890603e-01
6.11570597e-01 -8.63025427e-01 1.04415059e+00 5.27154148e-01
-1.22409868e+00 -4.91377085e-01 -1.11187279e+00 -1.48149192e-01
-1.22507669e-01 4.16604251e-01 1.55386463e-01 8.07532787e-01
-7.88365662e-01 5.89298666e-01 -7.50569820e-01 -6.98054731e-01
-2.51531214e-01 5.42216837e-01 -7.48868287e-01 2.37227425e-01
-1.07901955e+00 1.51293766e+00 2.90866435e-01 2.66470313e-01
-4.91906315e-01 -4.93927807e-01 -7.05832243e-01 -6.73437476e-01
6.33968890e-01 -1.09933889e+00 9.90684807e-01 -5.13815135e-02
-1.65164542e+00 1.61128628e+00 1.05429180e-02 -4.84241009e-01
1.30362344e+00 -1.02482569e+00 2.17053160e-01 2.60310650e-01
-4.61917720e-04 6.50816441e-01 7.02422798e-01 -1.47517395e+00
-8.48027542e-02 -6.87199891e-01 -2.92963982e-02 6.28158867e-01
5.59835248e-02 -8.45473483e-02 -7.51512408e-01 -5.02035856e-01
4.24989641e-01 -1.47735834e+00 -1.23668410e-01 2.34633312e-01
-6.30770028e-01 2.87769223e-03 3.35168153e-01 -7.46704221e-01
9.84499633e-01 -1.54856718e+00 6.93393767e-01 2.10665986e-01
1.12271689e-01 -4.89370758e-03 4.95328873e-01 3.49260569e-01
2.03947946e-02 -3.61112207e-01 -1.92706481e-01 -7.91626632e-01
2.52599835e-01 1.01035655e-01 7.37155527e-02 9.87544298e-01
-2.06213355e-01 7.35911667e-01 -6.95062101e-01 -4.81473595e-01
5.87558150e-01 5.87751389e-01 -4.72746193e-01 3.03375632e-01
2.15788543e-01 6.46207035e-01 5.19187115e-02 5.06309867e-01
6.13256156e-01 5.59200160e-02 9.77882966e-02 -3.71466696e-01
-2.14396361e-02 2.20286623e-01 -1.32976770e+00 1.96533716e+00
-1.50148571e-01 1.94292217e-01 -6.73693940e-02 -5.29880464e-01
8.79542291e-01 2.08252534e-01 5.76104403e-01 -3.57484333e-02
3.38515460e-01 4.75410193e-01 -4.20446366e-01 -1.34752333e-01
4.44799155e-01 -2.73730993e-01 -3.53067279e-01 2.71252126e-01
1.07127495e-01 -3.57100338e-01 -3.53171974e-02 -1.82309732e-01
5.12029767e-01 8.31743181e-01 6.44146681e-01 -4.90659952e-01
6.94858670e-01 -1.43504500e-01 1.73953041e-01 4.03411835e-01
-3.76154214e-01 9.48351622e-01 4.95335549e-01 -3.79129052e-01
-1.23179245e+00 -1.32073760e+00 -1.77585274e-01 6.85794890e-01
2.55971193e-01 -4.49839413e-01 -9.65101540e-01 -6.79958403e-01
1.65188864e-01 1.13351129e-01 -6.57443225e-01 -9.05400962e-02
-9.20013428e-01 -5.99479318e-01 6.37400508e-01 4.87489313e-01
2.92399317e-01 -5.63345551e-01 -1.13933468e+00 -3.63146633e-01
-1.84933990e-01 -1.35191405e+00 -2.66901493e-01 1.50497824e-01
-8.03254783e-01 -1.02502131e+00 -1.32108569e+00 -4.78369445e-01
7.10168839e-01 -1.29230082e-01 1.15128481e+00 -3.20181549e-01
-1.78965402e-03 5.44269860e-01 -2.55315453e-01 -1.78883746e-01
-1.88615277e-01 1.95961908e-01 6.59263551e-01 -4.84840214e-01
2.58691400e-01 -3.88056457e-01 -5.35734892e-01 7.14272618e-01
-1.24128059e-01 1.17602117e-01 5.86554885e-01 6.03509068e-01
7.13293910e-01 -6.58054471e-01 -2.50026792e-01 -7.64344037e-01
1.17960699e-01 1.81418806e-01 -4.55083877e-01 -3.23241726e-02
-5.00418067e-01 1.16272099e-01 8.83972347e-02 -3.36864889e-01
-8.34458172e-01 4.73227054e-01 -1.74597204e-01 -3.92801434e-01
-4.06569421e-01 1.54831512e-02 -1.59467533e-01 -2.12596834e-01
8.58064950e-01 -5.22556677e-02 1.75772056e-01 -6.11129642e-01
3.20520729e-01 4.13550317e-01 7.11171627e-01 -6.51151240e-01
8.57715368e-01 7.55334318e-01 4.41125065e-01 -7.58526564e-01
-8.01352084e-01 -6.16991043e-01 -1.34649861e+00 -4.41601932e-01
8.24268520e-01 -9.32977378e-01 -6.92930698e-01 4.20700818e-01
-1.19822347e+00 -1.80324167e-01 -2.87554175e-01 5.87967694e-01
-1.04655123e+00 6.98215544e-01 -5.79290986e-01 -9.60776269e-01
-2.58818120e-01 -1.18022966e+00 1.68024218e+00 -3.93338382e-01
-6.41701341e-01 -7.22791374e-01 1.80385679e-01 4.92022336e-01
-1.95292279e-01 7.48653889e-01 3.03823262e-01 -1.64822221e-01
-4.27489989e-02 -4.26728487e-01 9.88107324e-02 1.82472870e-01
-3.55031312e-01 -4.95573938e-01 -8.90727818e-01 -4.38132554e-01
1.34850694e-02 -1.60306692e-01 5.64984798e-01 4.64046746e-01
3.12494844e-01 1.33297935e-01 -3.92370015e-01 5.03088593e-01
1.03911114e+00 -5.40685892e-01 3.97675455e-01 4.80623722e-01
7.97391832e-01 1.01311088e+00 8.18260074e-01 5.55191159e-01
2.82822609e-01 1.32854009e+00 3.88669908e-01 1.49502978e-01
-3.24637473e-01 -5.46277344e-01 3.53517711e-01 9.12348986e-01
-5.82414031e-01 2.61644900e-01 -1.00684094e+00 1.57826737e-01
-1.72975028e+00 -3.54882598e-01 -1.49336740e-01 2.61298585e+00
5.59461296e-01 5.16467869e-01 5.83910286e-01 5.24259090e-01
5.52366138e-01 9.47757065e-02 -1.52109221e-01 3.25371101e-02
-1.52152330e-02 1.02189988e-01 7.94066906e-01 6.95064843e-01
-1.19608676e+00 6.78869843e-01 6.81448793e+00 5.17569125e-01
-7.69546270e-01 1.36985525e-01 -1.29669845e-01 -2.88820416e-01
1.68332785e-01 -1.23577856e-01 -1.17084074e+00 1.70691416e-01
5.67416251e-01 2.16031909e-01 -1.79330885e-01 8.47056389e-01
-3.47619387e-03 -2.15141520e-01 -1.28333342e+00 1.32709515e+00
4.30959970e-01 -5.47260880e-01 7.67123923e-02 3.38749915e-01
7.39101291e-01 -4.66207862e-01 2.10274179e-02 1.21119283e-02
-4.64959323e-01 -8.64602625e-01 1.07075083e+00 6.12239063e-01
8.64860833e-01 -6.89294398e-01 8.35424125e-01 6.77330971e-01
-1.16354084e+00 3.65193337e-01 -2.18155876e-01 -2.94461280e-01
5.15665352e-01 5.77931225e-01 -7.28144288e-01 7.58179188e-01
5.30726433e-01 5.23017704e-01 -5.48126280e-01 8.32773566e-01
-4.79270041e-01 -8.53414610e-02 -7.01857388e-01 6.67708972e-03
-9.04742479e-02 -1.18052125e-01 8.79129469e-01 1.14203215e+00
1.81589052e-01 -3.25038761e-01 2.06060842e-01 5.47682643e-01
3.13499779e-01 7.12048635e-02 -5.35100460e-01 6.40961468e-01
1.36583239e-01 8.78059685e-01 -7.52249658e-01 -1.05786964e-01
1.37238475e-02 1.25465846e+00 1.43797740e-01 -2.39616875e-02
-9.14657712e-01 6.70025300e-04 3.77346635e-01 5.56987226e-01
-1.34578906e-02 -5.13643205e-01 -3.29098076e-01 -1.13080192e+00
4.10337895e-01 -6.17330611e-01 1.79053992e-01 -7.07870662e-01
-8.71553421e-01 4.93610084e-01 7.56214678e-01 -1.49376404e+00
-5.78728020e-01 -9.89655852e-01 2.35786274e-01 6.65221512e-01
-1.04948187e+00 -1.15772617e+00 -3.97760332e-01 3.61553103e-01
2.42565349e-01 2.10920319e-01 8.05965483e-01 3.28494340e-01
-7.78161809e-02 6.35332406e-01 -5.99624515e-01 -1.26382887e-01
9.80942786e-01 -1.39359522e+00 4.23541695e-01 6.33516967e-01
6.42533973e-02 6.55350029e-01 1.17451751e+00 -5.89665532e-01
-1.35587442e+00 -5.04505575e-01 1.05639148e+00 -1.25491297e+00
4.87037040e-02 -6.86410367e-01 -2.93548584e-01 6.89189494e-01
-3.44071329e-01 9.03649777e-02 2.92870730e-01 1.21087775e-01
-3.20149451e-01 1.15033463e-01 -1.09583187e+00 5.33039391e-01
1.50492179e+00 -4.20587987e-01 -8.07222486e-01 3.64292622e-01
4.17434841e-01 -9.77643847e-01 -8.74182761e-01 8.22222292e-01
9.90955114e-01 -1.23355949e+00 1.36411071e+00 -1.39122501e-01
1.14362445e-02 -3.72091085e-01 -2.13019475e-01 -9.61286366e-01
9.26265046e-02 -6.55500054e-01 -2.97272891e-01 3.97760928e-01
7.08674490e-02 -3.57421100e-01 1.20481575e+00 2.12210968e-01
1.15797602e-01 -8.42418253e-01 -1.17778766e+00 -9.97767866e-01
1.25031516e-01 -4.65028852e-01 2.57703308e-02 5.28407216e-01
3.34587805e-02 3.08681399e-01 -7.80821979e-01 2.62507573e-02
9.65027392e-01 -1.94340870e-01 1.36194742e+00 -1.36179996e+00
-3.19651186e-01 -3.44833255e-01 -8.27841759e-01 -1.46542549e+00
5.74770905e-02 -4.80483562e-01 -7.38631096e-03 -1.11951041e+00
2.36577168e-01 7.38985538e-02 1.67052567e-01 -6.48174509e-02
1.88124329e-02 5.84387779e-01 2.94522971e-01 3.07724059e-01
-4.52220947e-01 3.87725711e-01 1.22428441e+00 4.59181011e-01
6.22147843e-02 9.39005166e-02 -8.07765126e-03 1.10242283e+00
4.99254167e-01 -3.01409632e-01 -1.70175895e-01 -1.01478621e-01
4.34496313e-01 -1.50504738e-01 6.29581213e-01 -1.23659205e+00
1.28200814e-01 3.22605819e-01 3.58450085e-01 -1.02787638e+00
7.58853614e-01 -8.23829532e-01 2.75748760e-01 7.90062904e-01
-1.77364141e-01 2.44832248e-03 -5.38930334e-02 4.30229008e-01
9.83183533e-02 -4.24456596e-02 6.94967806e-01 -2.74134576e-01
-4.14893806e-01 -1.69800017e-02 1.18763737e-01 7.33880326e-02
1.00114274e+00 -5.29676855e-01 1.82063192e-01 -5.23895741e-01
-1.03615236e+00 -2.55900890e-01 7.03060567e-01 2.31634736e-01
4.89437759e-01 -1.44022715e+00 -6.85350955e-01 1.36616856e-01
2.50978023e-01 -7.64650181e-02 -3.30038927e-03 1.15795612e+00
-8.32800865e-01 7.72929013e-01 -3.15302044e-01 -1.05817866e+00
-1.47611785e+00 4.30978477e-01 2.51204610e-01 -3.83319825e-01
-5.89635015e-01 7.02475309e-01 -1.31077021e-01 -8.97079706e-01
2.99245268e-01 -2.80558407e-01 1.30125448e-01 -2.11846620e-01
6.47330582e-02 8.17694426e-01 1.13937266e-01 -1.13774812e+00
-6.44980073e-01 1.20760763e+00 3.67329895e-01 -4.24928039e-01
8.94601822e-01 -2.62527436e-01 2.03084037e-01 5.71267188e-01
1.18390405e+00 1.08748846e-01 -1.33937109e+00 -6.47808611e-02
-9.53386500e-02 -4.71347719e-01 -5.23063064e-01 -6.26605093e-01
-4.57209349e-01 9.08321798e-01 7.46051073e-01 -2.59397388e-01
6.46936119e-01 1.58218294e-01 6.28630579e-01 2.78845042e-01
8.88073683e-01 -1.07734489e+00 -7.56426230e-02 4.21682000e-01
1.19758213e+00 -1.07615626e+00 6.58464074e-01 -9.69014287e-01
-4.02630329e-01 8.54257941e-01 5.32779872e-01 -5.14876127e-01
5.57691693e-01 2.58947313e-01 4.71804626e-02 -1.10910729e-01
-1.31754190e-01 -2.23815754e-01 7.30768085e-01 6.43401861e-01
5.49990714e-01 1.45279288e-01 -7.26767480e-01 3.93589765e-01
-6.40485644e-01 -2.72172689e-01 1.66569084e-01 1.03500760e+00
-2.89791286e-01 -1.04970825e+00 -7.99708664e-01 -1.88291103e-01
-2.56351531e-01 4.22719032e-01 -6.65564954e-01 1.20016408e+00
1.33682162e-01 4.19418573e-01 -3.03433061e-01 -4.03387517e-01
8.22289526e-01 1.06422953e-01 1.28399813e+00 -5.35604060e-01
-4.38091874e-01 2.65978187e-01 1.17294811e-01 -7.80124605e-01
-8.11652839e-01 -6.41222000e-01 -9.45950806e-01 -1.42773762e-01
-3.86155337e-01 -1.72233447e-01 6.06489897e-01 9.26217020e-01
7.29079843e-02 -4.68502752e-02 -8.89196526e-03 -1.50878012e+00
-8.06239605e-01 -9.59384620e-01 -5.39087355e-01 5.94462335e-01
1.66657746e-01 -1.27955568e+00 -5.18894136e-01 -1.01449206e-01]
|
[6.953007698059082, -0.9486154913902283]
|
51d32580-2c21-414c-99ac-a236c6d3ff21
|
a-conformer-based-asr-frontend-for-joint
|
2111.09935
| null |
https://arxiv.org/abs/2111.09935v1
|
https://arxiv.org/pdf/2111.09935v1.pdf
|
A Conformer-based ASR Frontend for Joint Acoustic Echo Cancellation, Speech Enhancement and Speech Separation
|
We present a frontend for improving robustness of automatic speech recognition (ASR), that jointly implements three modules within a single model: acoustic echo cancellation, speech enhancement, and speech separation. This is achieved by using a contextual enhancement neural network that can optionally make use of different types of side inputs: (1) a reference signal of the playback audio, which is necessary for echo cancellation; (2) a noise context, which is useful for speech enhancement; and (3) an embedding vector representing the voice characteristic of the target speaker of interest, which is not only critical in speech separation, but also helpful for echo cancellation and speech enhancement. We present detailed evaluations to show that the joint model performs almost as well as the task-specific models, and significantly reduces word error rate in noisy conditions even when using a large-scale state-of-the-art ASR model. Compared to the noisy baseline, the joint model reduces the word error rate in low signal-to-noise ratio conditions by at least 71% on our echo cancellation dataset, 10% on our noisy dataset, and 26% on our multi-speaker dataset. Compared to task-specific models, the joint model performs within 10% on our echo cancellation dataset, 2% on the noisy dataset, and 3% on the multi-speaker dataset.
|
['Nathan Howard', 'James Walker', 'Alex Park', 'Quan Wang', 'Arun Narayanan', "Tom O'Malley"]
|
2021-11-18
| null | null | null | null |
['acoustic-echo-cancellation', 'acoustic-echo-cancellation']
|
['medical', 'speech']
|
[ 2.90039867e-01 -8.26415941e-02 5.28315783e-01 -7.77281970e-02
-1.30865729e+00 -3.30039680e-01 3.28307360e-01 -9.77879986e-02
-6.65414095e-01 1.03739955e-01 7.28880942e-01 -4.97873425e-01
2.72281528e-01 -1.82486370e-01 -5.66831470e-01 -7.62910545e-01
-3.05741560e-02 -1.89066425e-01 2.47356549e-01 -5.96827269e-01
-1.88146904e-01 3.68541479e-01 -1.57984936e+00 2.55389065e-01
6.59009635e-01 1.04065776e+00 3.60446304e-01 1.23064101e+00
6.52747825e-02 4.82476503e-01 -9.98182774e-01 -1.99202135e-01
1.81385428e-01 -2.60213524e-01 -2.82044023e-01 5.03003672e-02
4.62470651e-01 -3.37345153e-01 -7.63422966e-01 1.06473505e+00
1.34872806e+00 4.71461535e-01 2.53828257e-01 -6.04089797e-01
-4.18238699e-01 7.95043826e-01 -1.83221832e-01 3.28109741e-01
3.33289802e-01 3.33791196e-01 7.71506190e-01 -1.20345986e+00
1.57623231e-01 1.27572381e+00 7.18424261e-01 7.01675355e-01
-1.11436033e+00 -8.17447603e-01 9.67545882e-02 2.51331944e-02
-1.26076913e+00 -1.46119511e+00 6.69566453e-01 9.48235318e-02
1.32361650e+00 5.44867754e-01 3.03873837e-01 1.25024569e+00
-1.96146831e-01 8.52557063e-01 7.22161293e-01 -4.56572860e-01
2.14241028e-01 -8.15141946e-02 3.35399032e-01 -8.89378600e-03
-4.70389724e-01 5.03598809e-01 -7.12637961e-01 -6.72073066e-02
3.06390345e-01 -5.49920738e-01 -6.70585990e-01 5.90060413e-01
-9.04220819e-01 4.02997226e-01 4.64444701e-03 4.70227718e-01
-3.94496351e-01 1.19802825e-01 5.23752868e-01 5.30820727e-01
4.99024123e-01 2.71444231e-01 -4.25318301e-01 -3.37552994e-01
-1.05068076e+00 -4.33057547e-02 7.72315383e-01 7.94304669e-01
2.22670063e-01 8.67125869e-01 -4.29169953e-01 1.52655280e+00
2.51964360e-01 9.49034214e-01 6.39316738e-01 -6.78234875e-01
7.17854619e-01 -2.91649193e-01 -7.12808222e-02 -4.95191336e-01
-2.24467859e-01 -9.44260716e-01 -8.55621159e-01 3.81356403e-02
1.76409457e-03 -3.39785486e-01 -1.07778943e+00 1.88142145e+00
4.74091619e-02 4.82463747e-01 3.52052629e-01 8.96170020e-01
1.12955761e+00 9.35274720e-01 -1.96181938e-01 -3.42177659e-01
1.33878279e+00 -1.13610649e+00 -1.29870975e+00 -6.17415309e-01
3.29954803e-01 -1.20667577e+00 9.25960362e-01 4.20101166e-01
-1.36292064e+00 -7.50042617e-01 -1.12783039e+00 6.10090755e-02
-2.40538552e-01 2.07563385e-01 -7.05239773e-02 9.86883342e-01
-1.46648920e+00 2.05176547e-01 -4.67741460e-01 -6.01088954e-03
-2.42996439e-01 1.99074343e-01 -4.49795753e-01 -4.97461632e-02
-1.53598845e+00 8.15460801e-01 -6.22714497e-02 2.03927100e-01
-1.04249978e+00 -6.88951254e-01 -1.14418793e+00 3.53667915e-01
2.20076695e-01 -2.06767544e-01 1.49245727e+00 -6.65629745e-01
-1.90259492e+00 4.37025905e-01 -4.34812874e-01 -4.13420171e-01
3.59418333e-01 -4.41467643e-01 -1.14113331e+00 -2.92476509e-02
-4.55112845e-01 3.26073408e-01 1.23771191e+00 -1.12302017e+00
-3.74563277e-01 -1.23202696e-01 -4.66269672e-01 3.22689176e-01
-3.28811049e-01 3.20696414e-01 -8.61552656e-01 -1.06327760e+00
5.67029454e-02 -7.38665402e-01 -1.39977187e-01 -5.57969511e-01
-4.83738542e-01 2.08893701e-01 9.71128583e-01 -1.46604979e+00
1.46386135e+00 -2.71872759e+00 -1.30480304e-01 1.69953048e-01
-2.22532630e-01 7.82275438e-01 -6.51621640e-01 2.10783899e-01
-5.02748072e-01 2.49135680e-02 -2.82657057e-01 -8.01400244e-01
-3.97608578e-02 -2.51573503e-01 -4.52299595e-01 2.71781415e-01
2.67759949e-01 4.17933673e-01 -7.65306175e-01 7.12381750e-02
4.72305506e-01 9.69048023e-01 -4.10085559e-01 3.81081283e-01
4.72882599e-01 7.96755999e-02 2.41575077e-01 2.59110630e-01
8.68082583e-01 5.75951815e-01 -9.99354199e-02 -1.06427200e-01
-3.18841962e-03 8.23975742e-01 -1.69877541e+00 1.36368775e+00
-6.94897056e-01 8.20213437e-01 9.57404852e-01 -3.57369810e-01
9.90414321e-01 8.19502950e-01 3.01277302e-02 -7.81584918e-01
-1.10906728e-01 3.57287377e-01 1.16900563e-01 -2.05238819e-01
6.44606233e-01 7.57900672e-03 2.81788826e-01 2.75032490e-01
5.99212237e-02 -3.25642437e-01 -1.10158414e-01 2.04891294e-01
1.40130496e+00 -6.51104450e-01 -8.46871063e-02 7.26673454e-02
7.18819320e-01 -9.03031707e-01 4.82890755e-01 6.53432488e-01
-3.48619193e-01 8.53259325e-01 -1.28091976e-01 2.56925046e-01
-8.20225716e-01 -1.20471179e+00 -1.45490170e-01 1.21513963e+00
-1.25967219e-01 -5.36666155e-01 -8.90315711e-01 -1.61266282e-01
-1.80679694e-01 1.04691541e+00 -9.08510536e-02 -3.26546550e-01
-6.40831590e-01 -4.57401037e-01 9.51378107e-01 5.59632838e-01
3.35139334e-01 -9.65950787e-01 2.18540922e-01 3.08181077e-01
-4.77312267e-01 -1.20722866e+00 -1.07700372e+00 5.28032959e-01
-4.07422662e-01 -4.16353613e-01 -8.86244297e-01 -6.41776741e-01
2.60594636e-01 5.77372015e-01 8.77017915e-01 1.17287204e-01
8.74895751e-02 3.23416024e-01 -3.64176512e-01 -2.74209678e-01
-9.86871004e-01 -2.29228839e-01 3.81957471e-01 1.42629832e-01
-4.34125476e-02 -5.17459333e-01 -4.21368808e-01 5.91006935e-01
-9.16035712e-01 -3.95926952e-01 4.89870489e-01 8.95568848e-01
1.48658022e-01 9.46349651e-02 7.27020919e-01 -1.17504090e-01
7.87219405e-01 -3.56012546e-02 -3.44424903e-01 7.07246438e-02
-8.77397954e-02 -1.46499544e-01 5.98779023e-01 -6.04236126e-01
-1.05220926e+00 -1.86888009e-01 -1.00701618e+00 -4.11312222e-01
-1.65031001e-01 1.85957700e-01 -5.29812753e-01 2.02745184e-01
6.67479992e-01 4.24990028e-01 -1.01948567e-01 -7.23201215e-01
3.74711841e-01 1.44744873e+00 6.15914285e-01 -1.56219210e-03
7.77198076e-01 -8.01986679e-02 -5.45522213e-01 -1.42453361e+00
-1.88775212e-01 -8.07003021e-01 -1.80652335e-01 -6.14069402e-03
3.59051973e-01 -1.17402101e+00 -3.11594158e-01 7.81892061e-01
-1.26578391e+00 -3.93817574e-01 -2.26113170e-01 7.01085865e-01
-2.07489073e-01 4.70094979e-01 -8.55565488e-01 -1.21223712e+00
-6.47411048e-01 -1.45545518e+00 1.16160882e+00 -2.32820481e-01
-1.28500894e-01 -6.25760734e-01 -1.65845171e-01 4.27441895e-01
9.08926785e-01 -6.89129829e-01 4.48348999e-01 -7.30684102e-01
-4.93326634e-02 -2.10865796e-01 1.13472953e-01 9.91322458e-01
1.87488198e-01 -1.38918951e-01 -1.54660678e+00 -5.37234366e-01
1.65201187e-01 5.84737994e-02 1.08249664e+00 4.54599380e-01
8.04408193e-01 -1.25338078e-01 -3.77698131e-02 4.93934333e-01
6.67360544e-01 4.06267136e-01 1.05482757e+00 -2.73212850e-01
3.64274651e-01 4.37925369e-01 2.88015753e-01 1.20371044e-01
-8.14925060e-02 9.79521692e-01 1.36588216e-01 -4.04084921e-01
-8.14741671e-01 3.98902297e-02 9.54246521e-01 1.42924881e+00
4.09717441e-01 -3.54219973e-01 -6.88259304e-01 6.59657538e-01
-1.38074088e+00 -1.03314507e+00 -5.99971190e-02 2.38766909e+00
9.60076869e-01 2.18466613e-02 4.22958806e-02 6.02342725e-01
1.00188625e+00 4.17825669e-01 -3.60989243e-01 -6.47075772e-01
-3.61215770e-01 2.77080595e-01 3.05129230e-01 9.40968335e-01
-1.00534117e+00 8.94342363e-01 6.83523989e+00 1.10493147e+00
-1.23843944e+00 1.57800376e-01 3.32172036e-01 -2.03796521e-01
-2.09614903e-01 -6.30102873e-01 -6.89219415e-01 3.26657683e-01
1.33923912e+00 2.53500789e-01 8.07830572e-01 7.17392623e-01
4.01005030e-01 2.67074883e-01 -8.97578895e-01 1.11094487e+00
1.63569838e-01 -7.15729594e-01 -3.02719593e-01 -8.73986259e-02
2.67457515e-01 2.49313384e-01 2.01540917e-01 4.67691243e-01
1.05895378e-01 -9.61254954e-01 8.70325267e-01 -4.80192564e-02
1.00424075e+00 -7.37142146e-01 7.29660571e-01 2.03409255e-01
-1.29980719e+00 -4.28196751e-02 -6.39292300e-02 2.75778502e-01
2.54683912e-01 9.24869239e-01 -7.24000514e-01 3.45775753e-01
6.36685491e-01 1.74733847e-01 -1.69965982e-01 1.02064669e+00
-5.10373235e-01 1.04355919e+00 -4.01685178e-01 4.40289617e-01
-1.33118257e-01 3.75005960e-01 1.15703464e+00 1.78993464e+00
3.46058339e-01 -6.64845691e-04 -2.77512670e-01 3.31593215e-01
-2.97655731e-01 1.10359162e-01 -3.55275005e-01 6.42974153e-02
7.05382407e-01 1.10551798e+00 -9.40361321e-02 -2.13115022e-01
-9.55625474e-02 1.10164583e+00 -3.28236967e-02 6.96981668e-01
-7.17503905e-01 -8.97400022e-01 9.65854824e-01 -2.13268355e-01
4.72240239e-01 -1.49811551e-01 -3.56921926e-02 -9.28899050e-01
1.03406556e-01 -1.24136007e+00 -1.35645330e-01 -8.56113493e-01
-8.87448311e-01 9.94495392e-01 -7.51725495e-01 -9.72169578e-01
-3.52865785e-01 -5.32563984e-01 -7.67593980e-01 1.40114319e+00
-1.59387195e+00 -5.91182530e-01 4.61709388e-02 5.49800754e-01
7.11828768e-01 -2.30968893e-01 9.39250886e-01 8.17026496e-01
-6.28448725e-01 1.00364387e+00 1.46335483e-01 2.12197572e-01
9.11438763e-01 -1.04093730e+00 9.62626159e-01 1.26857805e+00
2.26703975e-02 6.12402380e-01 7.73748159e-01 -4.94362473e-01
-1.40498459e+00 -1.08412015e+00 1.00587749e+00 -1.22927316e-01
3.49604338e-01 -6.83764398e-01 -1.19474280e+00 3.09982538e-01
1.92624256e-01 -2.62583774e-02 6.34616673e-01 2.60789722e-01
-4.98870462e-01 -3.07226002e-01 -1.03808773e+00 7.14277864e-01
9.91371036e-01 -8.77256572e-01 -6.75386131e-01 -7.48989126e-03
1.28218484e+00 -5.57464480e-01 -5.63939691e-01 1.99099883e-01
4.21290666e-01 -7.63706386e-01 1.07599640e+00 -2.56399572e-01
-4.76509929e-02 -3.38499099e-01 -4.21164423e-01 -1.84179163e+00
-3.27010155e-01 -1.06992185e+00 -8.04626495e-02 1.49437094e+00
7.09493220e-01 -5.71346521e-01 9.37189981e-02 3.91638249e-01
-6.53243721e-01 -1.16167419e-01 -1.08093309e+00 -1.00856960e+00
-3.04918438e-01 -1.10438871e+00 4.54155505e-01 6.37076557e-01
-2.27008373e-01 3.13635945e-01 -4.38314319e-01 5.05437613e-01
2.16931999e-01 -7.57812619e-01 6.04353249e-01 -4.93840545e-01
-4.48551327e-01 -5.20688057e-01 -1.13625936e-01 -1.25331783e+00
9.66800377e-02 -6.79751456e-01 5.90275884e-01 -1.43086803e+00
-4.15718675e-01 -3.02117430e-02 -4.46363986e-01 2.93814212e-01
-4.19655263e-01 -3.61555181e-02 4.50802743e-01 -2.15796828e-01
-2.02024266e-01 8.13632786e-01 8.01782548e-01 -3.28816295e-01
-5.10949016e-01 1.10611521e-01 -7.70725429e-01 5.96287847e-01
4.59837317e-01 -3.37490201e-01 -9.60588977e-02 -5.75438797e-01
-3.75432521e-01 1.89778760e-01 4.28609643e-03 -1.15331745e+00
3.34258795e-01 4.03499663e-01 8.24196339e-02 -5.27727544e-01
7.45102584e-01 -7.55002618e-01 -6.38027489e-02 3.23035508e-01
-4.46256369e-01 -3.17102402e-01 5.89667082e-01 3.28976870e-01
-4.02215064e-01 2.23080460e-02 9.19705391e-01 4.70778078e-01
-4.18237954e-01 -1.29916117e-01 -6.55223310e-01 -1.82474237e-02
2.38156110e-01 -5.47045544e-02 -2.92082369e-01 -8.83686066e-01
-7.41852820e-01 1.06443018e-02 -1.03976786e-01 6.77927673e-01
7.72675157e-01 -1.16259134e+00 -1.08471406e+00 5.32379985e-01
-8.39365125e-02 -4.73858446e-01 4.33919400e-01 6.87171340e-01
2.26426870e-01 3.15334111e-01 4.26125675e-01 -3.86835128e-01
-1.48272073e+00 3.45162362e-01 5.37245691e-01 -4.72173020e-02
-4.31008220e-01 9.92604315e-01 2.37740755e-01 -5.12664974e-01
7.09659815e-01 -3.67167473e-01 -4.50480580e-02 -1.37993425e-01
1.08381331e+00 6.25519574e-01 7.03437269e-01 -9.70027447e-01
-5.16894400e-01 3.95227760e-01 -1.03955586e-02 -6.63421631e-01
1.16417527e+00 -3.25074941e-01 2.91935861e-01 2.09466442e-01
1.21907127e+00 6.05483830e-01 -9.40687120e-01 -4.53854173e-01
-3.70477676e-01 -1.73131734e-01 6.62373722e-01 -1.17430866e+00
-1.03904748e+00 1.04621267e+00 7.89709151e-01 2.68278241e-01
1.38385832e+00 -3.57701838e-01 9.77157414e-01 2.11729199e-01
-6.24070987e-02 -1.24257612e+00 -3.68307233e-02 9.58658099e-01
1.44152379e+00 -9.70769167e-01 -5.82308292e-01 -3.60679328e-01
-6.37989521e-01 8.36493075e-01 1.58880100e-01 4.23535198e-01
7.36514091e-01 8.30064952e-01 4.33307648e-01 4.30436850e-01
-6.75828338e-01 -4.23072875e-01 4.66596037e-01 7.64064074e-01
4.22114491e-01 1.49554163e-02 3.08138549e-01 8.68064106e-01
-3.47789228e-01 -4.76163507e-01 1.30034953e-01 5.58553517e-01
-5.41149020e-01 -9.71439540e-01 -8.11923087e-01 2.55843345e-03
-5.05708814e-01 -6.25150442e-01 -3.82273823e-01 2.01097727e-01
-3.24197471e-01 1.68207979e+00 -1.37352841e-02 -6.87192798e-01
8.08642685e-01 1.84488654e-01 -1.83154196e-01 -4.59670961e-01
-9.89513278e-01 5.66770613e-01 4.38793421e-01 -4.22938973e-01
2.80441046e-01 -3.57509166e-01 -1.18443847e+00 -1.49302199e-01
-6.64717615e-01 5.67712821e-02 1.12875199e+00 7.96942949e-01
4.60900337e-01 9.84571934e-01 8.37090731e-01 -8.49213004e-01
-6.64118528e-01 -1.37564707e+00 -5.86505473e-01 2.89161146e-01
7.88824677e-01 -1.59151107e-01 -8.58690500e-01 -1.75151363e-01]
|
[14.84353256225586, 6.040369033813477]
|
2fa40905-7a90-4fea-87f9-888cfa7cc691
|
ml-net-multi-label-classification-of
|
1811.05475
| null |
http://arxiv.org/abs/1811.05475v2
|
http://arxiv.org/pdf/1811.05475v2.pdf
|
ML-Net: multi-label classification of biomedical texts with deep neural networks
|
In multi-label text classification, each textual document can be assigned
with one or more labels. Due to this nature, the multi-label text
classification task is often considered to be more challenging compared to the
binary or multi-class text classification problems. As an important task with
broad applications in biomedicine such as assigning diagnosis codes, a number
of different computational methods (e.g. training and combining binary
classifiers for each label) have been proposed in recent years. However, many
suffered from modest accuracy and efficiency, with only limited success in
practical use. We propose ML-Net, a novel deep learning framework, for
multi-label classification of biomedical texts. As an end-to-end system, ML-Net
combines a label prediction network with an automated label count prediction
mechanism to output an optimal set of labels by leveraging both predicted
confidence score of each label and the contextual information in the target
document. We evaluate ML-Net on three independent, publicly-available corpora
in two kinds of text genres: biomedical literature and clinical notes. For
evaluation, example-based measures such as precision, recall and f-measure are
used. ML-Net is compared with several competitive machine learning baseline
models. Our benchmarking results show that ML-Net compares favorably to the
state-of-the-art methods in multi-label classification of biomedical texts.
ML-NET is also shown to be robust when evaluated on different text genres in
biomedicine. Unlike traditional machine learning methods, ML-Net does not
require human efforts in feature engineering and is highly efficient and
scalable approach to tasks with a large set of labels (no need to build
individual classifiers for each separate label). Finally, ML-NET is able to
dynamically estimate the label count based on the document context in a more
systematic and accurate manner.
|
['Jingcheng Du', 'Cui Tao', 'Qingyu Chen', 'Yifan Peng', 'Zhiyong Lu', 'Yang Xiang']
|
2018-11-13
| null | null | null | null |
['multi-label-classification-of-biomedical']
|
['medical']
|
[ 4.36095268e-01 -8.54728222e-02 -3.80882591e-01 -6.01074278e-01
-1.13800824e+00 -5.13494253e-01 4.52313364e-01 8.61868739e-01
-5.59918880e-01 7.71787524e-01 -1.58581138e-01 -3.49975199e-01
-1.52866662e-01 -5.59556663e-01 -1.89307973e-01 -8.16174030e-01
3.73497337e-01 9.60840762e-01 -7.76625797e-02 2.95211375e-01
2.42072180e-01 2.31709540e-01 -1.57053685e+00 6.66748405e-01
5.06172776e-01 1.16974282e+00 5.94234578e-02 5.86483061e-01
-1.32115707e-01 7.25312948e-01 -6.05334103e-01 -3.44832391e-01
-2.43607700e-01 -2.77561069e-01 -9.41894352e-01 3.33469212e-02
4.41061348e-01 2.04231635e-01 3.91491055e-01 8.29152882e-01
7.63287067e-01 -1.79359958e-01 1.11898017e+00 -9.08495545e-01
-2.34436497e-01 5.43640614e-01 -5.29034913e-01 -1.20075636e-01
2.18180433e-01 -2.75864452e-01 1.17152333e+00 -8.96719635e-01
5.71557879e-01 1.12351871e+00 1.08718526e+00 4.70313221e-01
-1.35186434e+00 -6.12935841e-01 -2.02007413e-01 4.77933995e-02
-1.29904842e+00 -2.15678036e-01 4.02168304e-01 -6.91903234e-01
8.58369768e-01 3.88315856e-01 1.69343457e-01 1.11900401e+00
5.93974173e-01 6.72174037e-01 1.38865411e+00 -6.72411680e-01
1.64972872e-01 4.33911592e-01 2.57083356e-01 8.67177129e-01
-3.25356387e-02 -3.20296913e-01 -2.60304213e-01 -5.38003862e-01
3.73833217e-02 4.91459146e-02 5.17058047e-03 -1.06577296e-02
-1.34837651e+00 9.02199686e-01 -8.72168243e-02 6.66054726e-01
-1.01986103e-01 1.00812674e-01 8.33826184e-01 1.60921216e-01
9.51904118e-01 6.19777143e-01 -8.16462815e-01 7.76636750e-02
-1.13805652e+00 2.91724145e-01 8.28358471e-01 6.43448353e-01
4.58182544e-01 -6.32325709e-01 -6.08534575e-01 1.31039250e+00
3.42891783e-01 1.25089049e-01 7.65741050e-01 -4.59429681e-01
2.92388320e-01 7.74618626e-01 -2.45158464e-01 -9.25879240e-01
-1.19856501e+00 -5.77018917e-01 -9.79913712e-01 8.53585079e-02
4.95471656e-01 -7.25938007e-02 -7.42248774e-01 1.46961153e+00
3.12635392e-01 6.40213341e-02 -8.77576247e-02 3.35982382e-01
1.13424289e+00 4.11741614e-01 2.83148795e-01 -3.63525182e-01
1.69913530e+00 -1.00796533e+00 -7.65635312e-01 -7.23359734e-02
1.25775957e+00 -8.25403929e-01 6.82347417e-01 5.11890173e-01
-7.20422626e-01 -3.62388939e-01 -8.59912813e-01 -1.60434306e-01
-5.98720670e-01 4.95306820e-01 3.69622529e-01 6.86510026e-01
-8.96013200e-01 7.08845377e-01 -6.36480391e-01 -3.00409198e-01
5.64823031e-01 5.16445935e-01 -3.93764883e-01 -2.47596484e-02
-1.09565151e+00 1.01968849e+00 5.07280767e-01 -1.72527283e-01
-5.27420282e-01 -7.32597113e-01 -7.64621854e-01 5.66702001e-02
1.30370826e-01 -6.18906379e-01 1.31480694e+00 -7.71983325e-01
-1.31173837e+00 1.29200816e+00 4.22246307e-02 -1.51100576e-01
6.70732081e-01 3.66726935e-01 -3.07092786e-01 1.23345137e-01
3.74518514e-01 5.74786663e-01 4.02070522e-01 -1.08694446e+00
-8.10132861e-01 -2.72792935e-01 -3.64756614e-01 8.16333443e-02
-5.01444221e-01 1.55489221e-01 -1.41583920e-01 -6.50209546e-01
-6.53662160e-02 -9.91043270e-01 -1.79872096e-01 -9.46411118e-02
-5.40466368e-01 -6.54962242e-01 6.50327325e-01 -2.62983531e-01
1.22142124e+00 -1.89563930e+00 -1.17644109e-02 3.15091945e-02
3.84584546e-01 1.88473046e-01 1.44820705e-01 3.02532852e-01
-1.09138601e-01 3.14065635e-01 -2.30688706e-01 -7.35261679e-01
-8.62438008e-02 -3.16617303e-02 1.47232696e-01 7.70460546e-01
2.84817722e-02 7.35861659e-01 -1.00113654e+00 -9.28619146e-01
1.65263623e-01 3.73161137e-01 -3.59474421e-01 7.67092407e-02
-3.39753568e-01 5.28166890e-01 -3.08848381e-01 8.02565873e-01
2.45191008e-01 -8.39183271e-01 4.62195188e-01 -2.03601375e-01
1.51136488e-01 3.69482115e-02 -8.78770590e-01 1.57591200e+00
-6.54364049e-01 4.78296608e-01 -4.27938133e-01 -1.14524233e+00
7.65776575e-01 7.58317888e-01 8.26214552e-01 -2.89055943e-01
5.62814057e-01 5.48785806e-01 -1.63765132e-01 -5.28520465e-01
1.61598101e-01 -3.99041146e-01 -3.17990452e-01 5.87385297e-01
2.97677457e-01 -3.70784812e-02 3.85869205e-01 -2.15743244e-01
1.18817055e+00 -1.41578093e-01 7.60842621e-01 -3.85111123e-01
7.08345950e-01 -2.03255072e-01 4.51699197e-01 6.97492242e-01
-7.46669322e-02 6.01701736e-01 4.96106923e-01 -6.10095084e-01
-7.52944112e-01 -2.78898120e-01 -8.15228641e-01 1.37984157e+00
-1.78243056e-01 -4.59101737e-01 -5.12525022e-01 -9.90705729e-01
1.05970636e-01 4.00459051e-01 -8.47170889e-01 1.14995524e-01
-3.20892304e-01 -1.23229933e+00 7.96300054e-01 2.19253287e-01
6.79426640e-02 -1.08317304e+00 -4.29450750e-01 4.85037655e-01
-1.66185200e-01 -1.09305620e+00 -2.92069644e-01 6.78146720e-01
-6.21388674e-01 -1.11438274e+00 -6.53002620e-01 -9.96354520e-01
6.20613039e-01 -2.40129158e-01 1.22132516e+00 2.56857485e-01
-4.14317757e-01 6.07801005e-02 -3.48409086e-01 -5.30183494e-01
-8.27285767e-01 5.20072997e-01 -1.36125609e-01 1.68657020e-01
4.39417243e-01 -1.88537478e-01 -4.84887481e-01 4.20202225e-01
-8.74094963e-01 1.87709361e-01 5.36069036e-01 1.30982125e+00
7.75241792e-01 -1.42040208e-01 8.42117727e-01 -1.37754774e+00
6.30768716e-01 -5.21843076e-01 -2.16990501e-01 4.76552099e-01
-9.47045743e-01 -2.12141871e-02 8.04216385e-01 -5.26125550e-01
-5.64154565e-01 1.32495999e-01 -3.93473119e-01 6.54010177e-02
-2.82879323e-01 7.36001313e-01 2.17053920e-01 -4.64345068e-02
7.75324702e-01 -7.35685080e-02 -8.80111009e-03 -3.76953155e-01
1.60546571e-01 1.11753190e+00 7.68348202e-02 -4.02985066e-01
-6.31803572e-02 2.12915733e-01 3.78167301e-01 -4.23008651e-01
-1.41176760e+00 -8.18694413e-01 -6.57888532e-01 -1.96481973e-01
8.78774941e-01 -6.77056193e-01 -9.48469698e-01 5.73995650e-01
-1.07871854e+00 -2.05150381e-01 2.12493107e-01 2.92965442e-01
-6.14771068e-01 2.46063277e-01 -8.96706402e-01 -4.67159241e-01
-6.30304694e-01 -1.37518227e+00 1.48745799e+00 -3.81103605e-02
-5.22633433e-01 -1.40854013e+00 8.11753646e-02 5.12883008e-01
1.09273955e-01 4.31778580e-01 1.21683323e+00 -1.19163346e+00
3.05496752e-01 -4.16596711e-01 -2.32138827e-01 1.17152266e-01
2.48539552e-01 -1.17937729e-01 -1.18345320e+00 -4.51938093e-01
-1.23204492e-01 -7.40445077e-01 7.96355844e-01 3.97373170e-01
1.40854990e+00 -1.92907333e-01 -8.64781499e-01 3.55193764e-01
1.44012654e+00 1.11245334e-01 -5.05476370e-02 3.92951280e-01
6.44915879e-01 5.57088315e-01 5.64759374e-01 4.77401823e-01
3.61370176e-01 7.88406551e-01 2.96476394e-01 -2.11909562e-01
7.69516230e-02 4.60626245e-01 -2.24603564e-01 7.63463616e-01
3.55423808e-01 -5.26854813e-01 -1.15917242e+00 2.27150515e-01
-1.85423326e+00 -6.84303999e-01 -1.63834423e-01 1.81727588e+00
1.33991730e+00 7.10040703e-02 1.86850745e-02 2.81171948e-01
7.32556522e-01 -2.00054541e-01 -5.24248660e-01 -2.57877439e-01
-6.73993081e-02 1.57668754e-01 4.44927752e-01 1.60904765e-01
-1.52053821e+00 5.16004860e-01 6.25982380e+00 1.09703434e+00
-1.18235540e+00 3.94067883e-01 1.11294281e+00 -1.09636551e-02
2.27966413e-01 -5.07948637e-01 -1.11144435e+00 5.03696084e-01
1.18744791e+00 1.80625796e-01 -1.70863494e-01 6.80712342e-01
-1.37822166e-01 -7.80387223e-02 -1.41372073e+00 1.06143427e+00
1.45671159e-01 -1.46522772e+00 -1.44949272e-01 7.13358223e-02
7.05761611e-01 -5.98146431e-02 -6.21534942e-04 2.85467178e-01
1.48377180e-01 -1.10790837e+00 7.62177765e-01 2.94420600e-01
1.19341707e+00 -5.67717254e-01 1.09821844e+00 4.82621580e-01
-9.47423398e-01 -3.02908719e-01 -1.06927685e-01 2.24072903e-01
-1.92885287e-02 7.81239748e-01 -9.57725465e-01 5.59096217e-01
3.70823205e-01 8.71571064e-01 -6.51706636e-01 9.83729839e-01
1.54449150e-01 5.73817134e-01 -3.42469849e-02 -2.19892308e-01
2.36229658e-01 3.15555602e-01 -9.49981734e-02 1.67462087e+00
3.06624323e-01 -2.30703101e-01 5.60767174e-01 4.40672904e-01
-2.39140809e-01 6.38305247e-01 -2.69986361e-01 7.55238980e-02
3.54222298e-01 1.64863169e+00 -1.20783532e+00 -5.10781109e-01
-4.19142157e-01 5.61002791e-01 4.38220590e-01 -2.35848725e-01
-7.88107574e-01 -3.61465126e-01 1.07811935e-01 -1.44591317e-01
-5.88584617e-02 3.66553336e-01 -5.16793489e-01 -8.36642087e-01
-3.81366104e-01 -9.41375077e-01 6.04938984e-01 -3.42558771e-01
-1.50525427e+00 7.71672845e-01 -2.66715705e-01 -1.37803364e+00
-2.98101038e-01 -9.17552590e-01 1.35004550e-01 6.99481964e-01
-1.40392613e+00 -1.28608692e+00 -7.18597248e-02 1.55356079e-01
5.21022260e-01 -2.56547749e-01 1.37687504e+00 5.81249833e-01
-4.71515208e-01 8.33290160e-01 4.61296916e-01 1.19739071e-01
1.18797576e+00 -1.42858684e+00 -6.49802759e-02 -8.37545618e-02
2.21235417e-02 1.80221334e-01 5.25146425e-01 -4.00081187e-01
-7.20600188e-01 -1.28185976e+00 1.05730736e+00 -4.91838098e-01
6.55608773e-01 -4.41775739e-01 -8.09528232e-01 3.86453301e-01
4.28951494e-02 1.55010208e-01 1.32060647e+00 1.16129316e-01
-3.08863580e-01 3.00265215e-02 -1.20335424e+00 1.96755812e-01
5.24427772e-01 -3.04660022e-01 -1.08266182e-01 1.03465593e+00
5.04599214e-01 -7.26426840e-01 -1.25283968e+00 5.00645816e-01
6.77071452e-01 -6.23697281e-01 8.56813312e-01 -3.81415486e-01
6.63143456e-01 -1.09853931e-02 1.25744157e-02 -1.27754092e+00
-4.78214145e-01 -1.99289601e-02 1.12888716e-01 1.05770469e+00
7.75665104e-01 -5.39381206e-01 4.60257739e-01 2.56798387e-01
-1.95583388e-01 -1.34605110e+00 -9.73193884e-01 -4.51493889e-01
3.13787848e-01 -2.75514632e-01 3.84393811e-01 1.15005279e+00
1.89702511e-01 5.10959089e-01 -4.00040567e-01 -2.43359745e-01
3.87318105e-01 2.04575032e-01 2.75989711e-01 -1.73955429e+00
-2.71509409e-01 -6.83441162e-01 -3.14276636e-01 -4.89033192e-01
4.57698107e-01 -1.47645128e+00 1.04189001e-01 -1.58048725e+00
4.60545927e-01 -7.35224783e-01 -5.89721024e-01 9.14028049e-01
-2.30237439e-01 5.54058015e-01 -2.38652289e-01 3.16643775e-01
-7.39336014e-01 -7.26414770e-02 1.09259832e+00 -5.27004480e-01
5.81711046e-02 2.24965531e-02 -6.67040765e-01 6.35173321e-01
6.52106404e-01 -9.06459689e-01 2.25554109e-02 -8.79173540e-03
4.86763328e-01 2.24276990e-01 -1.03470102e-01 -8.94697666e-01
1.49536908e-01 -4.95742597e-02 3.20730597e-01 -5.11353135e-01
1.01367302e-01 -6.38048887e-01 1.83541179e-02 5.10873914e-01
-7.52439916e-01 -8.37553814e-02 1.00921966e-01 5.25497794e-01
-1.24589249e-01 -6.01939201e-01 9.30137575e-01 -1.42622963e-01
-1.69015586e-01 2.08806276e-01 -4.68961775e-01 -1.79011375e-01
1.07847357e+00 5.23002967e-02 -3.11799943e-01 2.69931912e-01
-8.54969800e-01 1.77168876e-01 6.99575245e-02 3.31519902e-01
1.49790928e-01 -1.21151900e+00 -8.29898298e-01 -1.44252300e-01
4.83299851e-01 -7.37419277e-02 1.66570261e-01 9.49517906e-01
-4.70409423e-01 6.45598173e-01 1.67674869e-01 -9.06986117e-01
-1.61300075e+00 5.65509498e-01 4.54543382e-01 -9.39841151e-01
-3.24632972e-01 7.28257895e-01 1.22442625e-01 -7.56163359e-01
2.98834920e-01 -3.46901894e-01 -5.85599661e-01 4.55553204e-01
6.36294186e-01 1.82705849e-01 5.17048001e-01 -6.04060352e-01
-3.72760862e-01 5.84850311e-01 -2.94934660e-01 2.12176934e-01
1.37415242e+00 1.37145489e-01 -3.72096896e-01 8.75349760e-01
1.38200355e+00 -4.21176285e-01 -5.13418376e-01 -2.79725909e-01
3.11678499e-01 6.05880469e-02 9.65520963e-02 -1.16842985e+00
-8.73904705e-01 6.58123076e-01 6.59605443e-01 2.67204612e-01
9.28250968e-01 9.94384103e-03 4.79809195e-01 3.82221937e-01
2.70864427e-01 -1.10436964e+00 1.82620451e-01 3.91179502e-01
5.53102493e-01 -1.59471750e+00 1.65869385e-01 -2.48753220e-01
-3.60219300e-01 1.32720768e+00 3.47065508e-01 5.51345348e-01
7.91692853e-01 3.73993009e-01 3.90585333e-01 -3.03542465e-01
-8.51639569e-01 2.25777790e-01 4.55188662e-01 3.04624904e-02
1.00344467e+00 1.14313021e-01 -6.14108264e-01 4.60677922e-01
8.70910138e-02 5.81195951e-02 3.33182842e-01 8.61491680e-01
-2.38121614e-01 -1.33777177e+00 -3.79388630e-01 9.96689618e-01
-1.15202463e+00 -6.95311800e-02 -4.31176871e-02 3.91116560e-01
3.75160038e-01 1.15269959e+00 -2.37795696e-01 -2.93922096e-01
-5.06608784e-02 2.81593442e-01 3.02580416e-01 -9.62073684e-01
-1.05859232e+00 1.39035344e-01 6.09394014e-02 -1.63096920e-01
-7.23032415e-01 -7.95024157e-01 -1.22408712e+00 3.84568423e-02
-6.73598588e-01 4.81934585e-02 7.23645389e-01 1.18438065e+00
1.69766307e-01 6.69731200e-01 4.11707610e-01 -6.84819400e-01
-4.79342043e-01 -1.25486636e+00 -6.51011527e-01 4.87688810e-01
2.34156623e-01 -7.64172733e-01 -2.96235323e-01 6.50241151e-02]
|
[9.449809074401855, 4.523286819458008]
|
88efb93e-948e-4085-a289-624dd15ad3fe
|
hybrid-long-document-summarization-using-c2f
|
2306.01169
| null |
https://arxiv.org/abs/2306.01169v1
|
https://arxiv.org/pdf/2306.01169v1.pdf
|
Hybrid Long Document Summarization using C2F-FAR and ChatGPT: A Practical Study
|
Text summarization is a downstream natural language processing (NLP) task that challenges the understanding and generation capabilities of language models. Considerable progress has been made in automatically summarizing short texts, such as news articles, often leading to satisfactory results. However, summarizing long documents remains a major challenge. This is due to the complex contextual information in the text and the lack of open-source benchmarking datasets and evaluation frameworks that can be used to develop and test model performance. In this work, we use ChatGPT, the latest breakthrough in the field of large language models (LLMs), together with the extractive summarization model C2F-FAR (Coarse-to-Fine Facet-Aware Ranking) to propose a hybrid extraction and summarization pipeline for long documents such as business articles and books. We work with the world-renowned company getAbstract AG and leverage their expertise and experience in professional book summarization. A practical study has shown that machine-generated summaries can perform at least as well as human-written summaries when evaluated using current automated evaluation metrics. However, a closer examination of the texts generated by ChatGPT through human evaluations has shown that there are still critical issues in terms of text coherence, faithfulness, and style. Overall, our results show that the use of ChatGPT is a very promising but not yet mature approach for summarizing long documents and can at best serve as an inspiration for human editors. We anticipate that our work will inform NLP researchers about the extent to which ChatGPT's capabilities for summarizing long documents overlap with practitioners' needs. Further work is needed to test the proposed hybrid summarization pipeline, in particular involving GPT-4, and to propose a new evaluation framework tailored to the task of summarizing long documents.
|
['Tu Tran', 'Sylvia B. Larcher', 'Guang Lu']
|
2023-06-01
| null | null | null | null |
['text-summarization', 'extractive-summarization', 'document-summarization']
|
['natural-language-processing', 'natural-language-processing', 'natural-language-processing']
|
[ 1.58396780e-01 3.80741239e-01 -2.45855227e-01 -1.67416230e-01
-1.29156756e+00 -7.02310860e-01 9.61281836e-01 6.60463631e-01
-2.30081543e-01 8.39298010e-01 1.08339310e+00 -2.01617718e-01
-2.52411924e-02 -3.71113777e-01 -3.75154316e-01 -1.58374116e-01
1.68938920e-01 7.26339817e-01 3.07340138e-02 -4.73269016e-01
8.62078011e-01 1.51210546e-01 -1.28533661e+00 6.88322008e-01
1.34145677e+00 3.29862714e-01 3.24305952e-01 9.83823299e-01
-5.90434968e-01 1.04045331e+00 -9.68616188e-01 -5.40689945e-01
-1.44825026e-01 -6.79667175e-01 -1.14913070e+00 -6.92306012e-02
5.51772535e-01 -1.89373881e-01 2.45750174e-01 6.71202660e-01
7.00332284e-01 3.36850658e-02 5.00654519e-01 -7.18411803e-01
-4.20231342e-01 1.08161461e+00 -3.18209320e-01 1.73625186e-01
7.22554922e-01 3.94420512e-02 1.31054807e+00 -4.51939225e-01
8.77998590e-01 1.23776364e+00 6.64820790e-01 4.47572589e-01
-1.10584426e+00 -1.81266099e-01 5.16285263e-02 1.38263283e-02
-8.79976571e-01 -7.11176932e-01 4.80117530e-01 -3.47436041e-01
1.46501040e+00 5.78274906e-01 5.34572661e-01 1.04887784e+00
4.42960501e-01 9.88920689e-01 8.16363811e-01 -5.19465327e-01
1.00163341e-01 2.94447839e-01 4.70031053e-01 3.14465284e-01
4.36556786e-01 -6.59663320e-01 -6.35438442e-01 -1.52176112e-01
-1.37517853e-02 -4.68146861e-01 -2.33427390e-01 3.77262503e-01
-1.24780965e+00 8.48070145e-01 -9.90778804e-02 6.23508394e-01
-5.45799494e-01 -1.82574335e-02 8.60224426e-01 2.77920067e-01
8.20276141e-01 1.06533062e+00 -3.13177168e-01 -6.79762661e-01
-1.49945533e+00 6.75609946e-01 1.47762752e+00 9.16014850e-01
2.96815693e-01 -1.72763079e-01 -5.89384496e-01 9.36677217e-01
-2.54986100e-02 2.00729087e-01 6.26318574e-01 -1.01942730e+00
8.25561166e-01 6.11969531e-01 1.68272853e-01 -1.07187581e+00
-3.90325904e-01 -4.96616274e-01 -6.02665961e-01 -3.33461851e-01
4.26905155e-02 -1.56498343e-01 -2.91855127e-01 1.20445907e+00
-2.63792187e-01 -3.64651799e-01 3.25571418e-01 5.59109390e-01
1.24615014e+00 9.68681693e-01 -1.11420706e-01 -6.12156451e-01
1.29563832e+00 -1.08640242e+00 -8.46820891e-01 -3.53921771e-01
8.46120775e-01 -1.18986344e+00 1.13584435e+00 3.67386431e-01
-1.52969396e+00 -3.79639059e-01 -1.08221638e+00 -3.19143534e-01
-1.33552000e-01 1.92615330e-01 4.61923629e-01 4.25772518e-01
-9.76117074e-01 7.27290094e-01 -8.11109841e-01 -7.84236252e-01
2.57620037e-01 -3.99453007e-02 -1.35677740e-01 -8.79018828e-02
-9.34378147e-01 1.22399127e+00 3.65264624e-01 -1.29557669e-01
-3.25951517e-01 -6.08327806e-01 -6.70239568e-01 2.12867528e-01
4.64655042e-01 -9.09411848e-01 1.60072410e+00 -4.64900315e-01
-1.45309711e+00 6.05965495e-01 -3.64991993e-01 -7.15761364e-01
5.24403751e-01 -5.54909825e-01 -8.28454867e-02 8.55451077e-02
3.50724936e-01 3.12128723e-01 7.94609264e-02 -1.02453840e+00
-6.03438020e-01 -1.39022157e-01 -1.48014143e-01 4.73537564e-01
-3.57766300e-01 4.08468395e-01 -3.52312565e-01 -5.31056762e-01
-2.95487463e-01 -6.84315324e-01 -2.07561061e-01 -1.04016364e+00
-7.81793475e-01 -3.72165859e-01 4.31341976e-01 -9.75263178e-01
1.70333755e+00 -1.56014669e+00 6.72438666e-02 -2.74029464e-01
7.40913302e-02 3.42517346e-01 -2.24979490e-01 1.25887978e+00
5.04237473e-01 5.45513391e-01 -9.86403003e-02 -5.87442815e-01
1.46165594e-01 -5.94856665e-02 -6.34170711e-01 -8.47254544e-02
1.03983715e-01 1.09366810e+00 -9.40082908e-01 -6.61768556e-01
-1.07360005e-01 1.70379654e-01 -3.90842319e-01 2.97625232e-02
-4.79312211e-01 2.14262411e-01 -5.14399350e-01 2.31908172e-01
2.03234389e-01 -1.21965535e-01 2.36506891e-02 1.31605640e-01
-5.48609614e-01 8.55280221e-01 -8.36531579e-01 1.77833712e+00
-3.98654968e-01 8.89864206e-01 -1.79487854e-01 -6.67197406e-01
7.39470661e-01 3.63515437e-01 1.57659218e-01 -5.33440113e-01
3.88382166e-03 3.40400040e-01 -3.99822369e-02 -6.53198063e-01
1.26341593e+00 -1.91531163e-02 -2.67825574e-01 8.33045125e-01
2.64987759e-02 -7.61866808e-01 9.68063354e-01 7.33564556e-01
1.13647604e+00 -4.62002717e-02 4.81930166e-01 -3.03180844e-01
3.75954151e-01 4.64623094e-01 2.03141615e-01 9.75925207e-01
2.63195097e-01 8.21332037e-01 8.30232441e-01 3.86365503e-02
-1.06997693e+00 -3.79840255e-01 1.25586122e-01 9.67442393e-01
-4.08573270e-01 -1.14140439e+00 -8.90905082e-01 -4.30831224e-01
-2.55379558e-01 1.31462801e+00 -2.09867492e-01 1.50097519e-01
-7.01120973e-01 -5.58406472e-01 6.90194666e-01 2.94684708e-01
2.58430094e-01 -1.18988967e+00 -6.05694115e-01 4.82138187e-01
-5.95305800e-01 -1.18871450e+00 -4.41313148e-01 -7.02254772e-02
-9.31954563e-01 -8.59867215e-01 -7.18480110e-01 -6.33559763e-01
1.58784881e-01 2.28230849e-01 1.26987720e+00 1.06431115e-02
1.73895866e-01 3.32765073e-01 -6.22312188e-01 -7.35000432e-01
-1.07221437e+00 6.58287048e-01 -3.00330132e-01 -6.47184968e-01
1.42544180e-01 -3.85740548e-01 -3.00091505e-01 -9.83475670e-02
-9.21579838e-01 3.49369437e-01 7.14093745e-01 5.38558781e-01
1.62012801e-01 -2.36336403e-02 9.28023577e-01 -1.21793425e+00
1.65478003e+00 -3.35787416e-01 -1.28561765e-01 4.39703524e-01
-7.12599635e-01 2.95218341e-02 6.68446124e-01 -4.77881953e-02
-1.23189569e+00 -7.36249030e-01 -2.73588151e-01 5.32207549e-01
-1.48573965e-02 1.16306996e+00 2.42159173e-01 5.92712402e-01
7.90410101e-01 1.70008734e-01 -3.72008868e-02 -5.09071052e-01
3.83825630e-01 7.89347470e-01 4.41448689e-01 -4.11582947e-01
4.15431380e-01 7.73802474e-02 -3.78534943e-01 -1.02345288e+00
-9.64797020e-01 -5.66750407e-01 -2.78940380e-01 3.01609896e-02
5.37098885e-01 -7.77860343e-01 -2.44237706e-01 1.37634814e-01
-1.34968734e+00 -3.39235693e-01 -4.82546091e-01 2.04395890e-01
-4.41037804e-01 6.71397269e-01 -8.45838547e-01 -6.23467803e-01
-1.06741929e+00 -8.44588876e-01 1.12134027e+00 2.68932462e-01
-1.01861525e+00 -1.10204244e+00 2.71024823e-01 7.08327293e-01
5.46731651e-01 1.71072215e-01 8.83237243e-01 -1.10247409e+00
-2.30500981e-01 -4.67498511e-01 5.42853884e-02 4.94638115e-01
-7.65052214e-02 3.53584886e-01 -5.79666257e-01 -1.48592040e-01
3.78516912e-02 -4.38482404e-01 9.04427946e-01 4.69529301e-01
4.86429542e-01 -6.81312859e-01 -1.05234683e-01 -1.02451064e-01
1.04599011e+00 -1.80535138e-01 5.75977445e-01 4.47587401e-01
5.33386409e-01 8.51027906e-01 6.80686295e-01 4.30112660e-01
5.84616601e-01 3.81990671e-01 -2.13814855e-01 3.04845899e-01
-2.65486389e-01 -3.52201104e-01 5.92279732e-01 1.43864012e+00
-4.71052416e-02 -5.91783524e-01 -8.54133010e-01 4.63270754e-01
-1.94743574e+00 -1.27455652e+00 -4.25998122e-01 1.85632002e+00
1.01363277e+00 4.25173521e-01 -4.02194932e-02 -1.44730797e-02
3.00407439e-01 3.77302051e-01 -1.30085405e-02 -1.00792909e+00
-2.00359344e-01 6.12752289e-02 3.50615272e-04 4.79277372e-01
-5.38887501e-01 9.28371787e-01 5.98530769e+00 8.81549895e-01
-9.43349600e-01 -1.49556831e-01 4.26523626e-01 -1.90692261e-01
-4.86173809e-01 2.94288188e-01 -9.75314021e-01 3.50079954e-01
1.26868308e+00 -7.96884060e-01 1.73781723e-01 5.63767910e-01
7.12888122e-01 -3.71726602e-01 -1.18215108e+00 4.63551760e-01
4.29156095e-01 -1.64712322e+00 1.56472191e-01 -7.78220519e-02
9.99349415e-01 1.77371368e-01 -3.69463980e-01 6.55644596e-01
2.39754736e-01 -9.32559133e-01 6.48766875e-01 4.80421424e-01
3.66210848e-01 -5.17181456e-01 1.04523122e+00 8.20347846e-01
-7.98016071e-01 1.49424970e-01 -4.16042149e-01 -1.98806569e-01
4.57187176e-01 6.31245792e-01 -1.25614989e+00 7.62054086e-01
3.13359052e-01 8.73152673e-01 -7.51192391e-01 9.65755701e-01
-3.19017082e-01 7.69256055e-01 -5.88554107e-02 -3.46830100e-01
3.03815097e-01 -1.20724402e-01 8.44481707e-01 1.74586439e+00
3.14434707e-01 4.46270928e-02 1.60626590e-01 7.34486759e-01
-1.17036380e-01 5.69040120e-01 -4.41583484e-01 -5.13947189e-01
3.41441691e-01 1.34041214e+00 -7.46699929e-01 -6.51538253e-01
-2.02966407e-01 7.47450233e-01 2.05799490e-01 2.33375013e-01
-3.55454832e-01 -3.73157710e-01 3.76347527e-02 1.25136390e-01
-2.87091602e-02 -1.65649891e-01 -5.86701632e-01 -1.15867603e+00
1.43731341e-01 -1.16076553e+00 2.45766327e-01 -8.06033611e-01
-1.04640949e+00 7.02378690e-01 2.11564347e-01 -8.83078158e-01
-5.78102946e-01 -9.91705805e-03 -1.06132007e+00 7.85344183e-01
-1.33053589e+00 -1.05610192e+00 -1.65195063e-01 -1.28619373e-01
1.08854914e+00 -1.14825048e-01 7.84877658e-01 -1.73682436e-01
-5.45291543e-01 2.56675363e-01 1.70580581e-01 -2.75854677e-01
8.07255805e-01 -1.26830864e+00 5.52877545e-01 8.71032298e-01
1.89161539e-01 8.30324411e-01 1.19228804e+00 -8.30422759e-01
-1.24513745e+00 -8.18423867e-01 1.49113071e+00 -5.99540651e-01
7.03337014e-01 -1.01457916e-01 -8.82393122e-01 5.84465265e-01
8.85315478e-01 -1.07110405e+00 7.05891371e-01 3.00049186e-01
1.38331622e-01 1.07686430e-01 -7.19478130e-01 6.72572851e-01
5.75449646e-01 -2.63865024e-01 -1.03880715e+00 5.18467426e-01
8.20414424e-01 -4.21909451e-01 -8.28207731e-01 2.25153297e-01
4.46931124e-01 -9.23306525e-01 3.71970564e-01 -3.66755784e-01
9.81428862e-01 8.90314206e-02 2.13595793e-01 -1.56236184e+00
2.84526367e-02 -8.61916602e-01 3.16664167e-02 1.79907978e+00
5.80043674e-01 -5.04359961e-01 5.30973375e-01 6.55267537e-01
-5.73403537e-01 -7.95682728e-01 -4.71510202e-01 -5.54402888e-01
3.34319025e-01 -4.84497041e-01 1.87543333e-01 6.16020381e-01
4.65529740e-01 8.80593121e-01 -1.57543242e-01 -5.21798730e-01
1.66376323e-01 1.62313163e-01 1.12193823e+00 -1.27546763e+00
-1.14235848e-01 -8.71502399e-01 1.20419629e-01 -8.99764001e-01
1.61607102e-01 -1.00083506e+00 5.73185794e-02 -2.36190224e+00
6.48078740e-01 5.88563867e-02 2.63112247e-01 1.43807441e-01
-2.23895237e-01 -1.47652715e-01 3.49684864e-01 4.16748464e-01
-9.28476036e-01 4.03266996e-01 1.28124106e+00 -1.66043062e-02
-5.04588366e-01 4.72497940e-02 -1.33885741e+00 6.15549266e-01
8.39142561e-01 -3.57580632e-01 -2.73313552e-01 -3.07292610e-01
4.68912959e-01 1.37310997e-01 -1.64825156e-01 -7.98574388e-01
4.87487376e-01 2.09011864e-02 -1.01800747e-01 -8.07738543e-01
5.16627543e-03 -1.60289243e-01 -2.23014429e-02 1.92617223e-01
-7.18208492e-01 1.68251470e-01 2.49354795e-01 1.49953142e-01
-4.16584313e-01 -5.74227989e-01 2.22298637e-01 -3.11534733e-01
-2.79488027e-01 -3.43681902e-01 -6.05675340e-01 4.37834829e-01
5.08757889e-01 -1.62648678e-01 -7.68193305e-01 -7.08697557e-01
-1.58549473e-01 4.37405437e-01 5.21358490e-01 3.38456750e-01
4.09284621e-01 -7.01475859e-01 -1.34286547e+00 -3.76442820e-01
5.15338071e-02 -6.66415226e-03 1.37492627e-01 9.63360727e-01
-6.29322290e-01 8.97652388e-01 8.10110047e-02 -4.08096611e-01
-1.31813800e+00 5.03605194e-02 -3.00427139e-01 -9.42753971e-01
-7.36442506e-01 5.01201093e-01 -2.58873552e-01 -2.07403809e-01
8.10769722e-02 -4.39052373e-01 -4.75607842e-01 4.78250384e-01
6.97377026e-01 6.54967666e-01 2.57111698e-01 -4.70128357e-01
2.63800323e-02 2.12004736e-01 -4.29263979e-01 -3.34997714e-01
1.51218486e+00 -1.92006826e-01 -3.44154269e-01 6.06792331e-01
9.73903954e-01 4.53807950e-01 -5.69399178e-01 7.16214702e-02
4.96870846e-01 2.46530436e-02 -2.06894696e-01 -1.03618300e+00
-2.67393053e-01 6.69604599e-01 -4.04267550e-01 5.11482000e-01
7.58968592e-01 9.78030711e-02 9.20486629e-01 6.19981050e-01
-3.34317773e-03 -1.26144326e+00 8.65617394e-02 8.48702013e-01
1.33660269e+00 -1.03160381e+00 4.79298294e-01 -5.29156066e-02
-1.05212748e+00 1.07360554e+00 2.79856831e-01 5.01260422e-02
-9.96879488e-02 1.51799858e-01 -5.34976879e-03 -2.82328963e-01
-1.18249440e+00 7.25303963e-02 4.17366892e-01 1.05541229e-01
1.04695034e+00 -6.61657676e-02 -9.13973153e-01 6.59456730e-01
-7.72466183e-01 -6.63716421e-02 9.13869798e-01 8.28511775e-01
-6.40378952e-01 -1.17255723e+00 -2.58852124e-01 7.81584501e-01
-7.01927006e-01 -3.18192214e-01 -7.01788723e-01 7.64186800e-01
-5.16428292e-01 1.32521474e+00 -2.36778781e-01 -1.13681545e-02
5.18462837e-01 1.57496735e-01 3.33932608e-01 -1.19719803e+00
-1.08629346e+00 2.53592193e-01 7.95476854e-01 -2.01966986e-01
-3.50617170e-01 -7.74712741e-01 -1.19878983e+00 -5.15882015e-01
-3.76774728e-01 6.40059948e-01 7.51884282e-01 9.23473299e-01
4.51078176e-01 5.93577087e-01 1.19179584e-01 -8.93882036e-01
-6.59920394e-01 -1.45872772e+00 -2.25016609e-01 2.74160892e-01
7.38103762e-02 9.68441516e-02 -3.29063773e-01 6.46744901e-03]
|
[12.337801933288574, 9.459553718566895]
|
e93613e6-e8f8-4a4a-a157-64f7df26e158
|
large-language-models-enable-few-shot
|
2307.00524
| null |
https://arxiv.org/abs/2307.00524v1
|
https://arxiv.org/pdf/2307.00524v1.pdf
|
Large Language Models Enable Few-Shot Clustering
|
Unlike traditional unsupervised clustering, semi-supervised clustering allows users to provide meaningful structure to the data, which helps the clustering algorithm to match the user's intent. Existing approaches to semi-supervised clustering require a significant amount of feedback from an expert to improve the clusters. In this paper, we ask whether a large language model can amplify an expert's guidance to enable query-efficient, few-shot semi-supervised text clustering. We show that LLMs are surprisingly effective at improving clustering. We explore three stages where LLMs can be incorporated into clustering: before clustering (improving input features), during clustering (by providing constraints to the clusterer), and after clustering (using LLMs post-correction). We find incorporating LLMs in the first two stages can routinely provide significant improvements in cluster quality, and that LLMs enable a user to make trade-offs between cost and accuracy to produce desired clusters. We release our code and LLM prompts for the public to use.
|
['Graham Neubig', 'Tongshuang Wu', 'Carolin Lawrence', 'Kiril Gashteovski', 'Vijay Viswanathan']
|
2023-07-02
| null | null | null | null |
['clustering', 'text-clustering']
|
['methodology', 'natural-language-processing']
|
[ 8.56470615e-02 9.67195034e-02 -1.44381091e-01 -8.15959394e-01
-1.15594101e+00 -9.98090208e-01 4.74745005e-01 9.17900860e-01
-4.87421155e-01 4.26338650e-02 3.90181005e-01 -5.44088960e-01
-7.27656856e-02 -4.52466965e-01 -2.35181496e-01 -4.09054816e-01
-3.70952412e-02 7.71574020e-01 2.64243275e-01 1.51031718e-01
6.19557858e-01 2.50229269e-01 -1.47070968e+00 5.67623913e-01
1.11030436e+00 4.61253338e-03 4.88094270e-01 7.25534141e-01
-4.90531921e-01 8.46359909e-01 -6.48206294e-01 3.23665291e-02
1.54938102e-01 -6.14661992e-01 -9.59230542e-01 4.08158988e-01
-2.36788675e-01 7.50285536e-02 2.73166835e-01 7.54954696e-01
3.29968095e-01 4.45695490e-01 8.19189429e-01 -1.20100009e+00
-2.97462821e-01 9.67933536e-01 -4.31554586e-01 -1.66756690e-01
5.72631359e-01 1.12254985e-01 1.00414002e+00 -7.95515358e-01
5.04489303e-01 1.11909223e+00 5.81637979e-01 1.23894326e-01
-1.22924924e+00 -4.59401876e-01 -2.75118556e-02 -1.61000118e-01
-1.59361303e+00 -7.02465892e-01 5.49448967e-01 -5.71970284e-01
7.43029594e-01 4.48445231e-01 2.68582225e-01 2.11139798e-01
-4.09266919e-01 5.78140438e-01 1.01627100e+00 -8.13029289e-01
6.72885358e-01 6.15312099e-01 4.26079363e-01 5.08145630e-01
4.24071811e-02 -4.30951387e-01 -3.47928703e-01 -5.30281305e-01
2.88717002e-01 6.14704676e-02 7.68072680e-02 -2.71923274e-01
-8.99716794e-01 8.09378564e-01 2.42044464e-01 5.31054914e-01
-2.30912864e-01 -5.25735356e-02 1.35205626e-01 1.80590898e-01
3.73924434e-01 9.27443981e-01 -2.98396319e-01 -4.11941558e-01
-1.39526665e+00 -2.10373685e-01 8.03338051e-01 9.81646180e-01
1.07189643e+00 -6.16209447e-01 -5.95750064e-02 9.07967210e-01
4.71299350e-01 1.33346751e-01 3.69560361e-01 -1.15735853e+00
1.75440133e-01 8.31213951e-01 2.84664750e-01 -8.00691068e-01
-3.55721682e-01 -4.74244393e-02 -4.39660341e-01 -5.36035523e-02
3.61204475e-01 -4.09774125e-01 -8.63458872e-01 1.27568340e+00
2.07042679e-01 -2.47032985e-01 -5.35700694e-02 5.74936569e-01
3.79714280e-01 7.46632576e-01 1.68759912e-01 -6.37820363e-01
1.04375041e+00 -7.36421645e-01 -5.93550026e-01 -2.84659117e-01
1.36085498e+00 -1.00253618e+00 1.30248094e+00 2.62508839e-01
-9.94236588e-01 -4.00443465e-01 -6.46873891e-01 1.16635829e-01
-2.52062589e-01 2.41189115e-02 3.92230690e-01 6.51267111e-01
-1.28859425e+00 6.37247443e-01 -9.89468932e-01 -9.03908491e-01
2.33516425e-01 3.77551377e-01 8.59863907e-02 -1.94418192e-01
-8.33355784e-01 4.37999219e-01 3.03658932e-01 -4.26954269e-01
-3.18895787e-01 -4.79476184e-01 -5.92144966e-01 2.68743504e-02
5.50499618e-01 -2.76242882e-01 1.37981391e+00 -1.10535264e+00
-9.45807815e-01 7.13843584e-01 -7.30874717e-01 -2.17883196e-02
2.89405316e-01 1.10407703e-01 9.24360473e-03 2.32356653e-01
3.83230329e-01 7.58442521e-01 6.10359013e-01 -1.50630903e+00
-7.15429187e-01 -2.57869512e-01 -3.63903672e-01 6.26658618e-01
-5.91590941e-01 4.13488358e-01 -8.49620759e-01 -3.01625669e-01
8.01071078e-02 -9.78965580e-01 -8.23120475e-01 -4.33900297e-01
-4.68267381e-01 -5.07068574e-01 7.55784571e-01 -4.83539283e-01
1.40160501e+00 -2.27980995e+00 -1.66822463e-01 6.66141212e-01
2.77039289e-01 9.40437093e-02 -3.11451387e-02 7.53428459e-01
6.06205799e-02 5.71459234e-01 -2.89476186e-01 -5.98488212e-01
-7.02896416e-02 6.53968826e-02 1.54599130e-01 3.12995106e-01
-1.45355567e-01 6.97458804e-01 -9.67397511e-01 -8.62371802e-01
1.72349498e-01 1.07080907e-01 -6.17839694e-01 5.41753173e-01
-2.19618857e-01 1.58945307e-01 -1.91376835e-01 4.83659089e-01
3.14637721e-01 -5.36195278e-01 3.22979093e-01 2.81051397e-01
-2.11429954e-01 3.29062611e-01 -1.32182992e+00 1.37526727e+00
-1.23639181e-01 5.41351199e-01 3.16432655e-01 -6.33250773e-01
7.47328281e-01 1.69847324e-01 5.66166341e-01 6.47383183e-03
-7.92249516e-02 -1.38657838e-01 -1.61842480e-01 -1.90653861e-01
5.79348683e-01 -6.55335113e-02 -3.61854248e-02 1.26655579e+00
-3.28410804e-01 -2.68037736e-01 3.51355851e-01 9.73502100e-01
9.96362507e-01 -5.95750153e-01 2.37850174e-01 -4.55645651e-01
-1.97376348e-02 5.75818181e-01 6.54263273e-02 1.06995285e+00
-2.03641169e-02 4.43556458e-01 2.45915651e-01 2.40661070e-01
-9.52357173e-01 -5.88819563e-01 2.57347077e-01 1.45410836e+00
-2.08446100e-01 -8.47633421e-01 -9.87633467e-01 -6.53270125e-01
-2.45971724e-01 7.62966871e-01 -3.27884197e-01 -2.99748480e-02
-5.18758483e-02 -5.76955140e-01 2.95473307e-01 5.12958109e-01
1.76176578e-01 -7.97259867e-01 -4.45828706e-01 1.02257438e-01
-1.98584035e-01 -6.12004459e-01 -8.41920972e-01 6.97670460e-01
-9.05158579e-01 -8.60580683e-01 -3.04660738e-01 -8.51178586e-01
1.15153849e+00 7.70766318e-01 8.27088296e-01 5.44028461e-01
-1.20742977e-01 5.14795542e-01 -7.14025855e-01 -2.29598075e-01
-7.20176458e-01 3.80903929e-01 8.52261856e-02 -4.94016051e-01
7.26297736e-01 -3.97151321e-01 -2.19208062e-01 4.60848540e-01
-8.71627331e-01 -3.87031212e-02 4.45629776e-01 2.12751821e-01
2.76148558e-01 6.54867589e-01 2.79937446e-01 -1.31642807e+00
7.82620430e-01 -4.60022390e-01 -2.15992376e-01 2.06445947e-01
-9.58778501e-01 7.50576183e-02 7.25852907e-01 -2.87990689e-01
-8.74259591e-01 6.54671788e-01 2.39486583e-02 -1.33008495e-01
-4.58031565e-01 7.57434905e-01 -6.78950408e-03 1.24428287e-01
9.53443706e-01 3.48293260e-02 -1.09447641e-02 -5.06682515e-01
6.86691880e-01 1.09952748e+00 3.72549981e-01 -3.90878558e-01
9.50221956e-01 2.76968330e-01 -7.69576848e-01 -7.57319272e-01
-7.53237665e-01 -1.15717936e+00 -1.15105093e+00 -2.08308682e-01
5.14755547e-01 -9.68146026e-01 -2.62755960e-01 -1.60556018e-01
-5.85803032e-01 -6.26673937e-01 -2.40757704e-01 2.14695171e-01
-2.37891823e-01 5.13018250e-01 -5.29181004e-01 -8.36062670e-01
-2.93626171e-02 -8.79332602e-01 5.92129290e-01 3.09180975e-01
-8.31913888e-01 -1.02687025e+00 -8.63678083e-02 4.24928099e-01
9.50180367e-02 -2.44042426e-01 7.72803843e-01 -1.07702732e+00
-2.61681885e-01 -2.60260671e-01 1.97115936e-03 -1.42981097e-01
4.20659721e-01 3.46865386e-01 -7.65603125e-01 -4.46194649e-01
-2.23296702e-01 -2.66669273e-01 6.30531669e-01 4.11145180e-01
9.23094511e-01 -4.23641860e-01 -7.32179999e-01 2.91091740e-01
1.10146654e+00 3.21733475e-01 2.62836069e-01 1.26468055e-02
7.22974122e-01 7.13021815e-01 7.54721284e-01 6.10943437e-01
3.46179336e-01 6.90376461e-02 -1.92277223e-01 -3.63423884e-01
4.50563192e-01 -2.60253400e-01 3.30706477e-01 1.10842443e+00
4.09319460e-01 -9.68479738e-02 -1.27761281e+00 6.95011735e-01
-1.95891273e+00 -7.65638828e-01 -4.15340662e-01 2.13600135e+00
9.21093524e-01 4.00826901e-01 5.17614365e-01 1.55883253e-01
7.87078381e-01 -1.79803967e-01 -4.46056992e-01 -2.65071481e-01
3.04305911e-01 -2.16325954e-01 4.16233838e-01 8.79188359e-01
-9.11095142e-01 1.13195240e+00 7.08218813e+00 7.69642413e-01
-5.95611572e-01 -1.00415640e-01 7.27626979e-01 -5.07074408e-02
-4.89540905e-01 4.38931584e-01 -6.01823926e-01 3.18898290e-01
1.19162822e+00 -3.63329768e-01 6.18917644e-01 8.74335825e-01
7.03939557e-01 -4.26604420e-01 -1.21793139e+00 6.83920026e-01
-8.07602331e-02 -1.12931967e+00 -2.06998304e-01 1.43118143e-01
7.14961290e-01 -1.05855212e-01 -4.44028109e-01 1.64353326e-01
1.20076394e+00 -1.09067321e+00 1.60343006e-01 2.72681355e-01
6.45053029e-01 -1.15787637e+00 4.59120691e-01 8.31824660e-01
-1.28524494e+00 -1.54300164e-02 -1.98561981e-01 -2.54206747e-01
7.53098279e-02 6.13577425e-01 -1.29155612e+00 1.43660545e-01
4.88194406e-01 6.06500626e-01 -9.37338471e-01 7.76535392e-01
6.44719824e-02 1.04519522e+00 -3.90323728e-01 -3.30121845e-01
2.75184005e-01 -1.27477601e-01 1.63486883e-01 1.61475778e+00
-5.61910458e-02 5.15745640e-01 8.22998822e-01 6.58771873e-01
1.28701493e-01 3.74152839e-01 -4.03873742e-01 -4.91594911e-01
1.04121733e+00 1.36930430e+00 -1.33456290e+00 -7.27338374e-01
-4.52000126e-02 1.01193261e+00 3.04271251e-01 3.29594284e-01
-2.20228523e-01 -7.55588114e-01 1.82349592e-01 5.09666502e-01
2.49084067e-02 -4.03763682e-01 -5.90877652e-01 -7.77901053e-01
-5.99486411e-01 -8.81442785e-01 5.79288900e-01 -9.13233221e-01
-1.00415134e+00 2.18353495e-01 -7.79525191e-02 -6.62211716e-01
-3.95286679e-01 1.83265712e-02 -6.69334471e-01 6.48578942e-01
-6.61130488e-01 -6.02586746e-01 -1.58479989e-01 5.59159338e-01
5.35084367e-01 1.65199652e-01 6.56979740e-01 4.86708879e-02
-4.03129995e-01 4.73645270e-01 2.54458636e-01 2.36025721e-01
1.16672122e+00 -1.56008005e+00 2.14034989e-01 8.86438489e-01
3.59476805e-01 1.13905430e+00 9.28812742e-01 -9.05667424e-01
-1.08505487e+00 -1.16564620e+00 1.12368059e+00 -7.34820306e-01
4.47256297e-01 -4.41927493e-01 -1.04645586e+00 6.17700517e-01
2.63257056e-01 -7.25695491e-01 1.25415552e+00 3.76488477e-01
-1.28194720e-01 2.67001390e-01 -9.91658330e-01 5.80319822e-01
6.56665564e-01 -5.94272256e-01 -5.33332884e-01 5.69719732e-01
7.75766194e-01 2.19404116e-01 -7.13984847e-01 -1.87070161e-01
-2.51738746e-02 -6.56823933e-01 4.42419589e-01 -3.80856276e-01
2.88595051e-01 -4.69334692e-01 3.81654873e-02 -1.39880443e+00
-6.87053263e-01 -8.95627439e-01 2.75514364e-01 1.32866085e+00
5.56380212e-01 -5.45706265e-02 8.50965798e-01 9.86942530e-01
3.90057638e-02 -2.56580263e-01 -1.24541000e-01 -2.83616960e-01
-2.85570938e-02 -5.77428997e-01 2.44213447e-01 1.34158647e+00
8.30796897e-01 4.89776611e-01 -2.47213021e-02 2.66011506e-01
5.55879951e-01 3.97708565e-02 8.89682055e-01 -1.37606263e+00
-1.35720417e-01 -3.27129960e-01 3.20224017e-01 -1.17219663e+00
-2.89691418e-01 -9.66791630e-01 4.11758035e-01 -1.87726986e+00
4.97428805e-01 -7.20175982e-01 5.25456630e-02 6.93236649e-01
-4.61884767e-01 1.72340497e-02 1.59438550e-01 7.22751737e-01
-1.22877002e+00 -1.77942380e-01 4.73855764e-01 3.01346928e-01
-8.30724299e-01 4.75704819e-02 -1.28715110e+00 7.08425164e-01
7.73567617e-01 -7.54541636e-01 -4.70462292e-01 -5.53552955e-02
-2.23864745e-02 -1.28466800e-01 -2.56021529e-01 -8.30099523e-01
1.01455283e+00 -3.23831677e-01 3.93390536e-01 -7.87528336e-01
-1.51087046e-01 -5.79849124e-01 4.00023680e-04 3.74994218e-01
-6.51492536e-01 1.35385999e-02 1.12848124e-02 4.60331619e-01
-2.64859647e-02 -3.47071886e-01 8.11524212e-01 -4.51720268e-01
-4.19860870e-01 4.26679384e-03 -1.10147786e+00 1.07523210e-01
8.72618616e-01 -1.46171004e-01 2.65346974e-01 -7.58684993e-01
-7.73149788e-01 6.51818275e-01 1.03770125e+00 -3.54449898e-02
2.00763345e-01 -9.09521580e-01 -4.58177984e-01 -7.57731423e-02
1.08534820e-01 1.17154911e-01 -2.64662683e-01 5.21554530e-01
-3.57001603e-01 4.09070969e-01 5.35862327e-01 -7.50951827e-01
-1.55559874e+00 6.65343046e-01 -6.24494143e-02 -1.76901203e-02
-2.27030247e-01 7.79946268e-01 -3.60653698e-01 -6.58684254e-01
6.93189323e-01 2.27857921e-02 -5.45584131e-03 2.97015071e-01
6.08407974e-01 1.80423334e-01 -1.66082889e-01 -3.60118359e-01
-4.37413216e-01 2.04155192e-01 -3.25691909e-01 -4.82043147e-01
1.29154849e+00 -4.05631632e-01 -1.43168077e-01 7.65279949e-01
1.08874774e+00 2.43011072e-01 -1.04443324e+00 -3.76381099e-01
4.30599391e-01 -2.77003467e-01 3.43304425e-02 -7.77488351e-01
-4.04189140e-01 6.31300151e-01 1.37720019e-01 4.78324562e-01
1.00920355e+00 3.67467344e-01 4.04071927e-01 7.33032525e-01
1.51069492e-01 -1.44500554e+00 3.15934449e-01 2.15052724e-01
7.48150349e-02 -1.42418206e+00 1.86046883e-01 -3.03967744e-01
-1.02785456e+00 8.16902518e-01 4.88551706e-01 2.16387853e-01
8.42681289e-01 4.46180820e-01 4.26443219e-01 -3.55764359e-01
-9.12049115e-01 -1.85289755e-01 1.01611957e-01 4.43443179e-01
6.86538041e-01 1.22355126e-01 -8.50707293e-02 3.83652151e-01
-1.66947708e-01 -2.85780370e-01 6.10833108e-01 1.19534481e+00
-1.11214304e+00 -1.12877190e+00 -6.63715243e-01 7.15701580e-01
-1.08089179e-01 -2.20645845e-01 -1.09788001e+00 3.01911324e-01
-7.46946111e-02 1.56054175e+00 1.36835352e-01 -4.84599352e-01
-2.78069582e-02 2.38759622e-01 -3.28886509e-01 -1.23681366e+00
-6.97747290e-01 6.24643147e-01 9.91628692e-02 -2.94580758e-01
-4.98851001e-01 -8.86003554e-01 -1.56647336e+00 -3.68413448e-01
-5.34995258e-01 9.41285312e-01 5.00687718e-01 9.82063234e-01
5.12071133e-01 1.26454625e-02 9.18800771e-01 -4.97107804e-01
-2.78840005e-01 -9.48881567e-01 -4.79404777e-01 3.83983403e-01
1.79609451e-02 8.38707611e-02 -4.47351098e-01 4.77220714e-01]
|
[10.55150032043457, 7.004774570465088]
|
1acb22e2-395c-44c4-a433-9aab0e34c427
|
actigraphy-based-sleepwake-pattern-detection
|
1802.07945
| null |
http://arxiv.org/abs/1802.07945v1
|
http://arxiv.org/pdf/1802.07945v1.pdf
|
Actigraphy-based Sleep/Wake Pattern Detection using Convolutional Neural Networks
|
Common medical conditions are often associated with sleep abnormalities.
Patients with medical disorders often suffer from poor sleep quality compared
to healthy individuals, which in turn may worsen the symptoms of the disorder.
Accurate detection of sleep/wake patterns is important in developing
personalized digital markers, which can be used for objective measurements and
efficient disease management. Big Data technologies and advanced analytics
methods hold the promise to revolutionize clinical research processes, enabling
the effective blending of digital data into clinical trials. Actigraphy, a
non-invasive activity monitoring method is heavily used to detect and evaluate
activities and movement disorders, and assess sleep/wake behavior. In order to
study the connection between sleep/wake patterns and a cluster headache
disorder, activity data was collected using a wearable device in the course of
a clinical trial. This study presents two novel modeling schemes that utilize
Deep Convolutional Neural Networks (CNN) to identify sleep/wake states. The
proposed methods are a sequential CNN, reminiscent of the bi-directional CNN
for slot filling, and a Multi-Task Learning (MTL) based model. Furthermore, we
expand standard "Sleep" and "Wake" activity states space by adding the "Falling
asleep" and "Siesta" states. We show that the proposed methods provide
promising results in accurate detection of the expanded sleep/wake states.
Finally, we explore the relations between the detected sleep/wake patterns and
onset of cluster headache attacks, and present preliminary observations.
|
['Shai Fine', 'Nancy Yacovzada', 'Gabi Shalev', 'Yotam Frank', 'Lena Granovsky']
|
2018-02-22
| null | null | null | null |
['sleep-quality-prediction']
|
['medical']
|
[-6.29569069e-02 -5.11508048e-01 -6.86372936e-01 -2.81204939e-01
-1.26348555e-01 -3.19868661e-02 1.46947354e-01 4.16475296e-01
-3.41263384e-01 7.18379736e-01 3.70637983e-01 -2.55974561e-01
-3.29241425e-01 -5.20141721e-01 1.38218477e-01 -9.01505649e-01
-2.88533658e-01 3.75364572e-01 -2.97735900e-01 8.40497029e-04
-1.07021250e-01 4.49867189e-01 -1.44984996e+00 5.92275225e-02
9.47001994e-01 1.24783278e+00 1.92546144e-01 3.46321046e-01
-1.47562221e-01 3.66996765e-01 -6.88601077e-01 3.60540450e-01
-1.25933185e-01 -4.44671690e-01 -3.52031946e-01 5.44193536e-02
-2.21182570e-01 -1.01809837e-01 6.02970272e-03 8.39340568e-01
6.53426766e-01 2.54388213e-01 -1.98110729e-01 -1.10547614e+00
-2.51467943e-01 -1.65638372e-01 -4.35109735e-01 8.62581968e-01
3.64620298e-01 1.90235481e-01 3.41499031e-01 -3.58300209e-01
6.01486079e-02 5.90499878e-01 4.92982000e-01 6.21496737e-01
-1.07546413e+00 -7.99847782e-01 -3.53678226e-01 5.09717822e-01
-1.29991913e+00 -6.49235785e-01 4.89661634e-01 -3.41866761e-01
8.34883988e-01 4.73304123e-01 1.31158519e+00 1.13283753e+00
6.67135715e-01 5.48407555e-01 1.13033366e+00 -5.60338236e-03
5.21680653e-01 -3.77841234e-01 2.59885281e-01 5.19545376e-01
3.79495203e-01 -7.09537491e-02 -7.54086733e-01 -1.72936603e-01
4.22085255e-01 8.84847999e-01 -8.20506141e-02 6.27000630e-02
-1.02805340e+00 4.27232444e-01 2.49932021e-01 7.96003699e-01
-4.28541422e-01 -7.72960037e-02 2.14242965e-01 -4.91487421e-02
7.21432090e-01 2.70532846e-01 -3.48865688e-01 -5.31253874e-01
-1.37652802e+00 4.53073494e-02 2.71049887e-01 4.20860872e-02
6.40696228e-01 -5.43600842e-02 -5.93007326e-01 7.87687302e-01
2.23856807e-01 4.57647800e-01 1.24726307e+00 -8.07201743e-01
9.85971279e-03 8.55328560e-01 2.95784950e-01 -9.32067037e-01
-1.19639969e+00 -5.89021385e-01 -1.08551681e+00 -3.08206797e-01
6.42435476e-02 1.05351821e-01 -7.22662270e-01 1.52310967e+00
3.65195841e-01 4.64239120e-01 -3.92039090e-01 8.21526349e-01
6.73600495e-01 7.80563056e-02 1.75465792e-01 -6.27048492e-01
1.74225461e+00 -6.37840450e-01 -1.12203157e+00 -5.16168952e-01
6.13484144e-01 -3.80080134e-01 1.19554865e+00 4.69896555e-01
-1.01868558e+00 -4.52079922e-01 -1.02614367e+00 6.64112493e-02
-4.97008920e-01 1.96817994e-01 8.80908608e-01 6.63564622e-01
-9.51131284e-01 9.18377697e-01 -1.62002945e+00 -4.10585016e-01
7.39151418e-01 6.29196167e-01 -9.01898965e-02 1.53633535e-01
-9.95951355e-01 7.59412169e-01 -9.99757648e-03 2.89515138e-01
-6.89921439e-01 -6.38780355e-01 -7.07096219e-01 2.03348443e-01
4.02217023e-02 -6.95930839e-01 9.90992904e-01 -4.10169333e-01
-1.41500080e+00 9.52825189e-01 -5.83080828e-01 -5.21508634e-01
-3.02451521e-01 -4.35743600e-01 -9.42429543e-01 4.20636088e-02
2.85260648e-01 -2.90732291e-02 5.86248517e-01 -9.97479632e-02
-3.58843416e-01 -1.06441391e+00 -3.96051019e-01 -4.38229330e-02
-2.91854143e-01 9.79232416e-03 -2.42868885e-01 -3.33729953e-01
1.42523989e-01 -8.68373930e-01 -3.80479842e-01 1.46617852e-02
-3.37409854e-01 -1.80006191e-01 7.40247667e-01 -3.19982111e-01
1.45065904e+00 -2.10794806e+00 -2.86686301e-01 -5.85277490e-02
7.90594995e-01 4.79797155e-01 2.18281999e-01 1.69457540e-01
7.26481229e-02 8.29547085e-03 9.51405149e-03 -5.15667081e-01
-4.18383300e-01 5.21372020e-01 1.66351214e-01 8.06164145e-01
-1.71406478e-01 9.60823715e-01 -8.73162448e-01 -2.42053926e-01
3.76035005e-01 3.75741780e-01 -2.95299858e-01 2.44432747e-01
-2.15072818e-02 8.29643309e-01 -4.82806861e-01 7.47985601e-01
3.41392100e-01 -6.69878960e-01 7.93478563e-02 -1.72321796e-02
-3.50963265e-01 4.91645873e-01 -6.47047162e-01 1.87051499e+00
-4.54538971e-01 4.68553483e-01 1.51333697e-02 -1.06867087e+00
6.86864018e-01 3.48209828e-01 1.00579286e+00 -1.26050556e+00
3.32866490e-01 -3.19846123e-02 1.56153560e-01 -8.24078679e-01
4.34427783e-02 -2.95755446e-01 1.21762149e-01 4.19260919e-01
-2.33345613e-01 4.89991605e-01 3.15317929e-01 -2.24014968e-01
1.35388315e+00 -3.48437130e-01 7.46357262e-01 -1.82197958e-01
1.03152506e-01 -3.83766949e-01 9.88263667e-01 2.69834131e-01
-5.62277913e-01 1.63800508e-01 3.45750481e-01 -7.40903318e-01
-3.80649805e-01 -1.02208066e+00 -1.56185329e-01 6.84286058e-01
1.15552247e-01 -7.00812578e-01 -3.36344361e-01 1.29357865e-02
-3.24648947e-01 3.47501755e-01 -7.72409916e-01 -5.00442326e-01
-2.82857895e-01 -1.29643309e+00 3.25314075e-01 5.23878932e-01
7.18292892e-01 -8.95544112e-01 -9.47449148e-01 3.50888431e-01
-3.99657160e-01 -9.42705035e-01 -1.55766129e-01 2.92114675e-01
-9.96181607e-01 -1.30144644e+00 -5.04170537e-01 -2.90391564e-01
3.59248042e-01 2.56247967e-01 7.17391312e-01 1.74483523e-01
-7.08246946e-01 9.31553729e-03 -1.80608809e-01 -3.62327605e-01
2.21954480e-01 -8.89679193e-02 4.14267242e-01 1.35259688e-01
7.91032970e-01 -1.09109592e+00 -1.48001289e+00 2.36153856e-01
-7.30798364e-01 -2.14006789e-02 6.94889426e-01 3.30013156e-01
7.80827045e-01 -2.15801135e-01 3.69065195e-01 -3.87376428e-01
8.02380145e-01 -7.75224805e-01 -3.82228047e-01 -1.72935516e-01
-1.07168365e+00 -4.94118705e-02 5.50601065e-01 -2.26233989e-01
-5.56635976e-01 -2.35140681e-01 -2.76330650e-01 -4.16706532e-01
-3.66831660e-01 3.12046587e-01 1.91646963e-02 1.91857576e-01
5.09271383e-01 2.06976444e-01 1.06870562e-01 -6.51728272e-01
-1.22840881e-01 7.11021423e-01 3.98968428e-01 1.52298715e-02
4.15284783e-02 8.12976956e-01 1.13137253e-01 -1.10444438e+00
-1.17071426e+00 -9.86173451e-01 -3.16931248e-01 1.07050568e-01
1.17149031e+00 -7.78224647e-01 -1.04327762e+00 1.97275445e-01
-5.85609674e-01 -4.70348924e-01 -3.18337172e-01 5.80670059e-01
-4.30156201e-01 2.19646350e-01 -4.90237951e-01 -4.86786097e-01
-6.84160531e-01 -9.94125247e-01 1.11242843e+00 5.72069108e-01
-5.10134220e-01 -1.01660120e+00 6.73938632e-01 4.67088908e-01
6.28623247e-01 2.88549364e-01 8.14984024e-01 -6.04415536e-01
-1.71353459e-01 -1.59044296e-01 1.35842055e-01 9.90711078e-02
4.46514100e-01 -4.09329921e-01 -7.71339655e-01 -1.74163207e-01
4.00507838e-01 2.60525104e-02 4.44745302e-01 9.01649535e-01
1.43716645e+00 -3.10827315e-01 -6.50126934e-01 8.53931010e-01
9.90549982e-01 5.79719424e-01 7.86563456e-01 2.81482399e-01
2.74300098e-01 -2.12769315e-01 1.32162049e-01 7.67242730e-01
4.19207454e-01 8.46981049e-01 3.02381963e-01 -2.10806563e-01
-8.17367136e-02 2.66359240e-01 9.33923125e-02 8.03357482e-01
-1.61201656e-01 -1.49101228e-03 -8.29284847e-01 5.03184378e-01
-1.61398721e+00 -9.39926982e-01 -1.41176164e-01 2.20214415e+00
7.01099813e-01 -7.01561794e-02 2.95996189e-01 3.33454832e-02
3.70491982e-01 2.54210502e-01 -6.59399927e-01 -2.88827091e-01
9.22395512e-02 8.51292968e-01 1.88530818e-01 -1.31792545e-01
-7.73204267e-01 5.20085216e-01 6.46337986e+00 4.55447137e-01
-1.44459128e+00 5.72758615e-01 3.72858733e-01 -7.07507908e-01
2.02366248e-01 -2.03479037e-01 -6.69785380e-01 9.76477683e-01
1.40141332e+00 -1.22176245e-01 5.32649338e-01 6.71936035e-01
1.13360727e+00 -5.17540634e-01 -7.33763397e-01 1.30943596e+00
-1.64482906e-01 -1.39757311e+00 -6.40984833e-01 2.89212406e-01
3.36553544e-01 2.36662656e-01 2.90272608e-02 7.38212764e-02
-1.45041108e-01 -8.38045359e-01 -1.49974838e-01 7.30770171e-01
8.72563362e-01 -3.20599794e-01 4.97630060e-01 2.04761863e-01
-1.20628762e+00 -3.37328523e-01 -3.14385146e-02 -3.41319531e-01
2.15906531e-01 1.02501535e+00 -7.12088883e-01 -7.22511709e-02
9.45080757e-01 8.98980319e-01 -5.99103510e-01 1.36443734e+00
-2.20177770e-01 6.44935846e-01 -2.37426922e-01 2.61846930e-02
-3.21197249e-02 -2.58528054e-01 3.92272741e-01 4.60210234e-01
4.58751976e-01 1.77195922e-01 1.34119317e-01 9.86213923e-01
7.41301775e-02 -1.16255395e-01 -3.97551030e-01 -2.47051954e-01
1.07067421e-01 1.46782923e+00 -9.52164412e-01 -2.40763828e-01
-3.52923006e-01 6.90782666e-01 2.75746710e-03 5.95842302e-02
-5.80032468e-01 -9.45514068e-02 1.03602111e+00 4.72199172e-01
-3.04317504e-01 -3.83411169e-01 -2.79417157e-01 -1.31669414e+00
-8.29508007e-02 -5.29463232e-01 3.11206222e-01 -7.14064002e-01
-7.15592802e-01 2.54363239e-01 -2.37890854e-01 -1.34024978e+00
-1.44750312e-01 -8.53501856e-02 -1.07458031e+00 4.64697689e-01
-1.31060028e+00 -5.45857847e-01 -6.42511666e-01 7.99888194e-01
5.55025637e-01 1.96460456e-01 1.00550258e+00 7.35035121e-01
-1.12879670e+00 1.36261493e-01 7.77131841e-02 -1.39061555e-01
4.11201358e-01 -1.03851926e+00 -7.67501146e-02 6.15231276e-01
2.43948493e-02 8.40647757e-01 4.97340858e-01 -6.58851802e-01
-1.07429159e+00 -1.04553640e+00 8.31998646e-01 -1.64842293e-01
5.18206060e-01 -1.54103905e-01 -6.31837070e-01 4.97536182e-01
-1.51673540e-01 1.04887426e-01 1.33046818e+00 3.04490119e-01
5.36896825e-01 -4.71496314e-01 -9.84444022e-01 4.27096635e-01
7.36759365e-01 -3.57940763e-01 -5.11619091e-01 6.06486857e-01
3.72655869e-01 -1.87897876e-01 -7.09295452e-01 3.16385567e-01
5.11902392e-01 -1.08952129e+00 7.73836553e-01 -5.81762373e-01
1.19450577e-01 1.39921438e-02 2.30904937e-01 -1.33069146e+00
-2.06273288e-01 -7.52012014e-01 -6.21070266e-01 5.33997416e-01
-1.46879375e-01 -4.26522076e-01 1.04412568e+00 5.63833177e-01
-4.65437382e-01 -9.35434282e-01 -1.13151610e+00 -7.42430091e-01
-5.48054576e-01 -5.59197247e-01 5.51636875e-01 5.92754006e-01
3.49181533e-01 2.50837684e-01 -2.63965398e-01 -1.05855808e-01
2.19657764e-01 2.52557516e-01 3.29541832e-01 -1.30358303e+00
-7.09633157e-02 -2.98910320e-01 -5.14644980e-01 -6.14467621e-01
-2.61470079e-01 -7.84889817e-01 -3.30702335e-01 -1.69117367e+00
1.66098297e-01 -2.47428209e-01 -5.75564265e-01 6.85851514e-01
1.80247873e-01 3.33670586e-01 -4.57856536e-01 2.20278561e-01
-8.26189280e-01 6.42915189e-01 1.09444821e+00 7.65672326e-02
-7.69293249e-01 5.37985086e-01 -6.97209775e-01 7.88947046e-01
1.09035134e+00 -4.48144853e-01 -3.55203182e-01 -1.26083538e-01
4.41636562e-01 2.07029015e-01 3.87665808e-01 -1.37991083e+00
2.48658597e-01 -6.32891655e-02 5.24025559e-01 -5.00751853e-01
5.98068118e-01 -5.13194561e-01 3.89127899e-03 7.29930878e-01
-2.46130973e-02 6.03775159e-02 2.38787949e-01 5.19788086e-01
-1.28950849e-01 4.73676741e-01 7.75394797e-01 -2.05328569e-01
-4.24480766e-01 4.23721582e-01 -7.70027220e-01 -4.03209478e-02
9.71793950e-01 -2.52677590e-01 -3.29610616e-01 -2.22994581e-01
-1.45472014e+00 1.32513016e-01 -2.96331998e-02 3.95735025e-01
5.85296333e-01 -1.29567003e+00 1.21382572e-01 5.96992314e-01
1.30110532e-01 -2.34304175e-01 6.41080141e-01 1.68353105e+00
-2.76194304e-01 6.20873153e-01 -4.03596699e-01 -5.93943000e-01
-1.03932595e+00 3.95887077e-01 3.91756266e-01 -4.13049579e-01
-5.65236688e-01 4.30790067e-01 -1.88719362e-01 2.88516670e-01
1.37085438e-01 -7.05585599e-01 -2.38378808e-01 6.82443008e-02
8.21397841e-01 6.06199145e-01 5.12726784e-01 -1.74200118e-01
-4.64440018e-01 2.07233369e-01 1.55738011e-01 4.07522261e-01
1.38533223e+00 -2.72658020e-01 -2.01403961e-01 4.42905813e-01
9.63496566e-01 -2.82308340e-01 -5.82358420e-01 5.96185103e-02
1.71519276e-02 -1.65358648e-01 3.51814955e-01 -7.40331948e-01
-1.11706293e+00 1.11057615e+00 1.39793777e+00 3.70334923e-01
1.32848811e+00 1.58745766e-01 1.37914848e+00 2.38351315e-01
2.70922691e-01 -9.88312602e-01 1.60580799e-01 -1.37262404e-01
2.86359221e-01 -1.05714989e+00 5.57431914e-02 2.62991935e-01
-2.20858663e-01 8.71315241e-01 4.45034355e-01 7.92365223e-02
9.13716733e-01 2.11113691e-01 -1.15855075e-01 -7.86933362e-01
-5.50701499e-01 -4.57608730e-01 1.66570827e-01 4.53219771e-01
2.39817858e-01 1.74549684e-01 -4.16353673e-01 9.32657719e-01
-1.55563742e-01 6.17056608e-01 2.10428625e-01 8.98879051e-01
-4.86917734e-01 -1.07680237e+00 -1.93135485e-01 1.09666300e+00
-7.11482525e-01 -8.23091418e-02 -7.23394006e-02 4.93603081e-01
5.72865009e-01 1.08091009e+00 3.69939297e-01 -4.01899099e-01
2.16260523e-01 2.33481199e-01 8.37950557e-02 -9.40651715e-01
-3.85135740e-01 2.46022880e-01 -2.46194512e-01 -1.17794168e+00
-6.23897552e-01 -4.77949798e-01 -1.18308616e+00 -1.20973520e-01
-5.58418594e-02 3.97232547e-02 6.16529346e-01 1.32674873e+00
8.24925601e-01 6.48734808e-01 3.82847428e-01 -6.34772003e-01
2.01361343e-01 -8.79945099e-01 -1.03825521e+00 2.46032655e-01
5.41925132e-01 -9.27712917e-01 -1.88649967e-01 -1.62988782e-01]
|
[13.595562934875488, 3.44077467918396]
|
6d845e92-e938-4e82-af65-25f4ec1d2790
|
soft-attention-improves-skin-cancer
|
2105.03358
| null |
https://arxiv.org/abs/2105.03358v3
|
https://arxiv.org/pdf/2105.03358v3.pdf
|
Soft-Attention Improves Skin Cancer Classification Performance
|
In clinical applications, neural networks must focus on and highlight the most important parts of an input image. Soft-Attention mechanism enables a neural network toachieve this goal. This paper investigates the effectiveness of Soft-Attention in deep neural architectures. The central aim of Soft-Attention is to boost the value of important features and suppress the noise-inducing features. We compare the performance of VGG, ResNet, InceptionResNetv2 and DenseNet architectures with and without the Soft-Attention mechanism, while classifying skin lesions. The original network when coupled with Soft-Attention outperforms the baseline[16] by 4.7% while achieving a precision of 93.7% on HAM10000 dataset [25]. Additionally, Soft-Attention coupling improves the sensitivity score by 3.8% compared to baseline[31] and achieves 91.6% on ISIC-2017 dataset [2]. The code is publicly available at github.
|
['Sargur N. Srihari', 'Mingchen Gao', 'Mohammad Abuzar Shaikh', 'Soumyya Kanti Datta']
|
2021-05-05
| null | null | null | null |
['skin-cancer-classification']
|
['medical']
|
[ 2.76934177e-01 4.70896900e-01 -3.21961105e-01 -2.02617154e-01
-8.05305660e-01 -1.20342433e-01 5.31164229e-01 1.02665238e-01
-7.32388079e-01 6.60358548e-01 5.18866837e-01 7.84732029e-02
-1.73767470e-03 -4.27391946e-01 -5.22592545e-01 -8.58439684e-01
5.42129166e-02 -1.21017449e-01 2.71644473e-01 -1.17272012e-01
-2.03953777e-02 1.84915513e-01 -9.42870021e-01 6.89774752e-01
7.97964692e-01 1.12811840e+00 2.79785872e-01 6.86932862e-01
1.56366020e-01 1.05789626e+00 -6.08059645e-01 -4.26956296e-01
-4.01675068e-02 -1.81108192e-01 -8.92149806e-01 -4.88929629e-01
5.85077763e-01 -2.08619341e-01 -5.42102814e-01 1.27478349e+00
9.15822566e-01 -1.39988065e-01 5.99174798e-01 -7.20531464e-01
-9.06200469e-01 5.72664559e-01 -5.46555877e-01 7.35427141e-01
-3.44332665e-01 2.78930753e-01 8.42324018e-01 -8.62245739e-01
4.88359183e-01 8.48453999e-01 7.67044842e-01 8.93659055e-01
-9.76156354e-01 -4.63868231e-01 1.25274301e-01 3.74592155e-01
-1.24180913e+00 -3.10209930e-01 3.99548113e-01 -2.64261127e-01
1.08323145e+00 4.23786104e-01 3.80643278e-01 1.33361816e+00
6.95384324e-01 7.63949990e-01 9.33458924e-01 -2.32549280e-01
2.85016727e-02 -1.01977751e-01 5.40474355e-01 3.27094555e-01
1.43450543e-01 -1.06319357e-02 -1.03328697e-01 7.49020576e-02
6.69688702e-01 2.99067259e-01 -3.41389924e-01 5.21159232e-01
-7.88394570e-01 5.78792453e-01 1.13193583e+00 5.23020029e-01
-9.04263198e-01 1.80381209e-01 5.79559147e-01 9.90907550e-02
4.26840335e-01 5.53286552e-01 -2.15085819e-01 5.69294803e-02
-7.13553071e-01 -2.14752316e-01 1.43896759e-01 4.41834837e-01
5.97567903e-03 2.19240904e-01 -7.56627500e-01 1.10705888e+00
-4.69310842e-02 4.18111235e-01 6.69422030e-01 -5.08428156e-01
1.41424790e-01 6.39888465e-01 -4.33320969e-01 -6.96534514e-01
-5.64683437e-01 -1.01985729e+00 -1.43290675e+00 2.08858743e-01
9.64761525e-02 -3.36299360e-01 -1.66495490e+00 1.51683736e+00
-2.03175604e-01 9.48628262e-02 -6.85131550e-02 9.23616946e-01
1.41306484e+00 3.03151846e-01 6.18843138e-01 1.61372140e-01
1.38569105e+00 -1.10596895e+00 -8.06949556e-01 -4.02167052e-01
3.92557740e-01 -5.85550368e-01 9.12233174e-01 2.15808973e-01
-1.17458618e+00 -6.36927128e-01 -7.98619211e-01 -1.90040827e-01
-2.01105401e-01 1.57270357e-01 3.40010345e-01 2.21533597e-01
-1.36727667e+00 4.44813818e-01 -9.08266187e-01 -4.24294055e-01
7.66679823e-01 5.50554752e-01 -4.41667378e-01 4.28264849e-02
-1.25054169e+00 1.07275581e+00 2.33439714e-01 1.22753195e-01
-1.05107784e+00 -9.19543326e-01 -4.97198015e-01 2.45904565e-01
1.09545171e-01 -6.73725367e-01 1.25509202e+00 -1.21489573e+00
-1.12354851e+00 9.51600194e-01 1.32661864e-01 -7.71722674e-01
4.60480690e-01 -4.55268145e-01 -2.52459645e-01 4.57444876e-01
3.99414003e-02 7.78153956e-01 4.43181187e-01 -5.21582544e-01
-3.70284140e-01 -3.77170771e-01 -2.95695234e-02 9.77773219e-02
-2.56796598e-01 1.25451446e-01 -6.24527276e-01 -8.04515600e-01
-2.66101390e-01 -8.27568948e-01 -5.51677465e-01 -1.35888696e-01
-7.86768377e-01 -9.75307152e-02 5.64996183e-01 -7.69700408e-01
9.39060450e-01 -2.22807455e+00 -1.16383599e-03 -1.65749639e-02
5.67951918e-01 8.09337378e-01 -4.83386844e-01 -1.55769959e-01
-6.53898716e-01 3.01582962e-01 -1.79968923e-02 -1.85970411e-01
-5.48221231e-01 -8.95211250e-02 1.74347401e-01 4.70102251e-01
5.37476063e-01 1.30360794e+00 -6.58533514e-01 -3.13093573e-01
4.42853868e-01 9.57415640e-01 -5.19927621e-01 1.29723521e-02
2.81746060e-01 2.43752033e-01 -2.56947696e-01 7.10425258e-01
4.95817333e-01 -6.39340222e-01 -3.51312459e-02 -2.86679268e-01
2.22113863e-01 1.10901661e-01 -5.17021775e-01 1.57398856e+00
-2.73078680e-01 7.27322698e-01 1.93856865e-01 -8.06484520e-01
5.80203056e-01 3.18273246e-01 4.74103719e-01 -9.98142898e-01
6.41715050e-01 -7.95496181e-02 3.14513683e-01 -3.83709610e-01
1.40572608e-01 6.03754371e-02 2.82962501e-01 -2.91974157e-01
2.32935712e-01 6.27879620e-01 -2.03276694e-01 2.25311756e-01
1.41223836e+00 -4.93230641e-01 4.53201175e-01 -5.39847434e-01
4.02924210e-01 -2.78238922e-01 4.15482908e-01 1.01089084e+00
-3.77729177e-01 9.44200695e-01 5.33580542e-01 -2.96266168e-01
-6.83104992e-01 -1.01424682e+00 -1.95112213e-01 1.11623073e+00
-2.53058314e-01 -2.55823016e-01 -8.05149972e-01 -8.75431299e-01
-1.35370910e-01 5.48940718e-01 -1.37645173e+00 -4.42466378e-01
-4.16698426e-01 -8.96530449e-01 5.85093319e-01 8.69934559e-01
5.13917983e-01 -1.18096268e+00 -5.20605206e-01 1.42265884e-02
-1.60050631e-01 -8.87922466e-01 -3.93163502e-01 4.46441025e-01
-6.15455806e-01 -9.74048078e-01 -1.25944865e+00 -8.17375004e-01
7.73391664e-01 7.80723244e-02 9.58130836e-01 2.31241867e-01
-6.37377501e-01 -4.87821959e-02 -2.10172817e-01 -5.27008593e-01
5.08122006e-03 4.17117089e-01 -4.93918002e-01 -2.56786104e-02
4.21159416e-01 -2.17414901e-01 -9.44557905e-01 -1.02574714e-02
-8.08315575e-01 5.73007278e-02 1.07607424e+00 1.01273847e+00
7.88102150e-01 -5.16172409e-01 6.68723106e-01 -9.32237387e-01
3.71914625e-01 -5.11557639e-01 -4.26445790e-02 8.70577693e-02
-3.91723245e-01 -1.68471098e-01 5.20349026e-01 -4.46210563e-01
-8.00053596e-01 -3.82223986e-02 -4.42605525e-01 -5.40615380e-01
-5.76443151e-02 3.34657103e-01 1.42300278e-01 -1.62023567e-02
9.12705481e-01 1.65662125e-01 -2.69270353e-02 -4.67537522e-01
8.71283934e-02 5.97073734e-01 6.75631702e-01 1.70118406e-01
8.36078152e-02 3.91302139e-01 -1.31673589e-01 -6.71822131e-01
-1.02282333e+00 -5.86012363e-01 -2.66738862e-01 1.81015171e-02
1.06852341e+00 -9.72711623e-01 -4.96061832e-01 6.24489844e-01
-9.04065192e-01 -4.16819811e-01 -4.51505154e-01 4.30406690e-01
-9.10096392e-02 -2.55689863e-02 -8.13496113e-01 -3.53235573e-01
-1.08514118e+00 -1.18354928e+00 8.34433079e-01 3.99899304e-01
-3.86352628e-01 -9.47074711e-01 -8.86960551e-02 -1.02937378e-01
7.86361456e-01 3.94309610e-01 5.81110239e-01 -9.19753969e-01
-1.03777245e-01 -4.33297724e-01 -5.97932518e-01 5.66964328e-01
2.08999947e-01 -1.45085305e-01 -1.22023535e+00 -2.62652069e-01
-3.25696021e-01 -3.02541107e-01 1.44084227e+00 1.06958604e+00
1.50299120e+00 -1.03478909e-01 -4.05631810e-01 7.34321058e-01
1.44823980e+00 1.69375494e-01 9.62091625e-01 5.78188486e-02
5.30471861e-01 2.56652087e-01 3.16911712e-02 1.71702299e-02
1.46816298e-01 4.11439151e-01 6.62212133e-01 -7.29304671e-01
-7.87164986e-01 2.04219401e-01 -3.85826491e-02 4.66941625e-01
-2.51656115e-01 -1.84820354e-01 -1.09088790e+00 7.70652592e-01
-1.68168306e+00 -7.59899616e-01 -1.85580879e-01 1.91288376e+00
8.24007869e-01 3.36813897e-01 -1.70135230e-01 -1.71099469e-01
8.93022180e-01 7.24385679e-02 -5.92933595e-01 -2.98681796e-01
-2.01696545e-01 3.96425754e-01 6.31233752e-01 4.64405864e-01
-1.29451346e+00 7.47440755e-01 6.43352413e+00 8.30103874e-01
-1.47595561e+00 3.23621631e-01 1.12473226e+00 -4.55786139e-01
1.95560664e-01 -7.15247989e-01 -7.50145853e-01 5.61476111e-01
1.01717126e+00 1.47462562e-01 -1.04799949e-01 8.44793916e-01
-4.69396543e-03 2.35971855e-03 -7.24907815e-01 9.09630954e-01
2.05442578e-01 -1.52340281e+00 -1.97793722e-01 5.58672547e-02
7.70847440e-01 6.58389807e-01 4.46110487e-01 3.90044451e-01
2.83240855e-01 -1.42153668e+00 1.37875944e-01 6.94229126e-01
1.11858463e+00 -7.48092234e-01 1.28629744e+00 -1.71071932e-01
-5.98595083e-01 3.04730833e-02 -1.53217256e-01 2.09259346e-01
-1.52754247e-01 6.82662725e-01 -9.80743647e-01 1.63421497e-01
9.62579668e-01 6.83494508e-01 -7.66700685e-01 1.22296417e+00
-4.65770274e-01 8.74556959e-01 -1.46229595e-01 -5.24770580e-02
5.29467762e-01 5.71634710e-01 4.38976824e-01 1.55040312e+00
-1.88117146e-01 8.79829973e-02 -1.73033774e-01 6.25799179e-01
-3.10306758e-01 1.50985032e-01 -3.02786320e-01 3.74599993e-02
-1.13161527e-01 1.51752341e+00 -5.55564463e-01 -4.22289640e-01
-3.11302722e-01 9.53525901e-01 1.98397845e-01 3.73753637e-01
-8.37921441e-01 -8.25004041e-01 2.97777444e-01 2.54098624e-01
2.23237380e-01 5.13711929e-01 -5.70148110e-01 -7.23373592e-01
-3.32657605e-01 -8.09953451e-01 6.03577137e-01 -7.32971489e-01
-1.18542325e+00 1.09184134e+00 -5.15831113e-01 -6.83916867e-01
-4.84798513e-02 -6.38742268e-01 -7.50437021e-01 1.13626301e+00
-1.63480031e+00 -1.03715158e+00 -6.32723451e-01 5.86244404e-01
4.97788876e-01 -7.71607235e-02 9.22341704e-01 2.82880604e-01
-7.20993280e-01 9.20021594e-01 4.91065998e-03 4.67043668e-01
8.01776588e-01 -1.27407253e+00 4.42493349e-01 8.04897070e-01
-5.47488928e-01 6.43148124e-01 2.93648392e-01 -6.86250806e-01
-6.33893907e-01 -1.36533689e+00 7.55614042e-01 -2.62350023e-01
5.04413128e-01 -8.66736472e-02 -1.09329641e+00 5.02217352e-01
6.26302719e-01 3.92900348e-01 6.46845400e-01 1.62314296e-01
-3.61318588e-01 -1.28587514e-01 -1.23453295e+00 4.80982900e-01
6.86333597e-01 -3.22264969e-01 -5.73551834e-01 2.69796520e-01
6.32412314e-01 -4.36378747e-01 -8.94512951e-01 6.45851195e-01
3.59995067e-01 -6.13623738e-01 9.95408416e-01 -8.68122995e-01
7.11348951e-01 3.16590071e-01 -1.51877096e-02 -1.43023789e+00
-9.56329525e-01 -3.91141772e-01 -7.98330754e-02 6.17191195e-01
4.27487075e-01 -6.08461440e-01 7.75616527e-01 2.61574149e-01
-4.52133715e-01 -1.07307899e+00 -7.59477735e-01 -2.68023849e-01
3.55963379e-01 -4.25086394e-02 1.73035592e-01 8.22068214e-01
-1.56195834e-01 2.73879349e-01 -2.85257638e-01 3.30617428e-02
3.35259765e-01 -5.51726460e-01 1.08461887e-01 -9.56382215e-01
-1.62677750e-01 -7.23185837e-01 -4.91966367e-01 -3.66754651e-01
-1.41188860e-01 -1.03724301e+00 -1.83340773e-01 -1.82863462e+00
6.11710429e-01 1.96699858e-01 -1.23062086e+00 1.10364974e+00
-4.94840801e-01 6.67740285e-01 1.27687812e-01 -2.03004420e-01
-7.28206992e-01 2.23371178e-01 1.35694742e+00 -2.51271695e-01
-3.96994874e-02 -2.61964679e-01 -1.15644228e+00 6.60001159e-01
1.08922446e+00 -2.62369663e-01 -9.80946645e-02 -4.20514077e-01
-3.07122260e-01 -3.52033436e-01 5.31400979e-01 -9.14912760e-01
3.04108053e-01 7.87338167e-02 8.16140592e-01 -5.59834540e-01
2.06658855e-01 -4.06196862e-01 -2.76811808e-01 9.03636813e-01
-5.34817338e-01 -1.57874823e-01 6.81444764e-01 1.88849464e-01
-2.21690252e-01 2.05039419e-02 1.22688162e+00 -9.58678350e-02
-7.04254627e-01 3.61984342e-01 -3.18751901e-01 3.58623415e-01
9.75215018e-01 6.75145388e-02 -7.23116219e-01 -8.55228305e-02
-1.11050570e+00 1.58748612e-01 -7.35774487e-02 4.57476735e-01
6.80261374e-01 -1.29067957e+00 -9.73566771e-01 1.05648637e-01
-4.46528159e-02 -1.76050276e-01 7.19137490e-01 1.22017479e+00
-3.17798138e-01 5.93713403e-01 -4.69521046e-01 -6.85525596e-01
-1.35034788e+00 2.72474766e-01 6.93626225e-01 -3.34728062e-01
-6.61175549e-01 1.19900417e+00 6.43939853e-01 1.62587967e-02
4.96083826e-01 -4.05439496e-01 -3.05960834e-01 -1.98284701e-01
1.05721986e+00 2.20037133e-01 4.03576881e-01 -2.57279396e-01
-7.87929773e-01 3.08905721e-01 -7.89653778e-01 5.41281521e-01
1.42398715e+00 4.17502522e-01 2.12449223e-01 -1.87828422e-01
1.34214652e+00 -3.83711904e-01 -1.10988379e+00 -3.60396922e-01
-3.98248643e-01 -1.14515331e-02 7.18874395e-01 -1.52152085e+00
-1.55893278e+00 1.05178726e+00 1.08700073e+00 -1.44781858e-01
1.39604926e+00 -5.07979048e-03 5.11500418e-01 -5.11257583e-03
-3.78938645e-01 -8.95766020e-01 2.56918073e-01 4.96816009e-01
1.17012405e+00 -1.24168146e+00 -2.82977879e-01 -1.35066703e-01
-7.74127662e-01 7.58600473e-01 8.13761711e-01 -4.23972040e-01
5.66041350e-01 4.55266893e-01 3.04038413e-02 -1.90245688e-01
-8.49325657e-01 -2.84007877e-01 7.03679323e-01 5.68527281e-01
5.15637994e-01 6.42078370e-02 -2.82612406e-02 8.41939211e-01
2.60131180e-01 9.09936205e-02 2.79807687e-01 5.48690379e-01
-3.88442129e-01 -4.53585118e-01 9.18707922e-02 6.78229392e-01
-1.27904010e+00 -5.41084230e-01 -5.17612875e-01 7.04780042e-01
7.04024956e-02 6.17203236e-01 1.21935681e-01 -3.95916909e-01
4.51306075e-01 -2.97831148e-01 6.26613796e-02 -6.78986907e-01
-9.01418030e-01 3.79438430e-01 -2.34885290e-02 -6.25927210e-01
-7.48569798e-03 -2.78433084e-01 -1.20216870e+00 2.08720681e-03
-1.80706382e-01 -1.29957184e-01 4.15828526e-01 4.74767685e-01
6.65853739e-01 1.20645607e+00 2.02532649e-01 -2.46384308e-01
-4.63687360e-01 -1.21754277e+00 -1.65314287e-01 2.84091264e-01
5.70935249e-01 -4.01241899e-01 -1.75351515e-01 -2.59294689e-01]
|
[14.750205993652344, -2.6078667640686035]
|
3ea6e276-3071-410b-9f64-d87b9056a693
|
ordered-counterfactual-explanation-by-mixed
|
2012.11782
| null |
https://arxiv.org/abs/2012.11782v2
|
https://arxiv.org/pdf/2012.11782v2.pdf
|
Ordered Counterfactual Explanation by Mixed-Integer Linear Optimization
|
Post-hoc explanation methods for machine learning models have been widely used to support decision-making. One of the popular methods is Counterfactual Explanation (CE), also known as Actionable Recourse, which provides a user with a perturbation vector of features that alters the prediction result. Given a perturbation vector, a user can interpret it as an "action" for obtaining one's desired decision result. In practice, however, showing only a perturbation vector is often insufficient for users to execute the action. The reason is that if there is an asymmetric interaction among features, such as causality, the total cost of the action is expected to depend on the order of changing features. Therefore, practical CE methods are required to provide an appropriate order of changing features in addition to a perturbation vector. For this purpose, we propose a new framework called Ordered Counterfactual Explanation (OrdCE). We introduce a new objective function that evaluates a pair of an action and an order based on feature interaction. To extract an optimal pair, we propose a mixed-integer linear optimization approach with our objective function. Numerical experiments on real datasets demonstrated the effectiveness of our OrdCE in comparison with unordered CE methods.
|
['Hiroki Arimura', 'Kento Uemura', 'Yuichi Ike', 'Ken Kobayashi', 'Takuya Takagi', 'Kentaro Kanamori']
|
2020-12-22
| null | null | null | null |
['counterfactual-explanation']
|
['miscellaneous']
|
[ 4.33590561e-01 1.60070345e-01 -4.01443362e-01 -4.18601543e-01
-2.96422452e-01 -4.26359743e-01 4.06503677e-01 2.38047704e-01
-3.20439696e-01 1.05394912e+00 1.33869633e-01 -6.16091728e-01
-4.89317268e-01 -7.40395665e-01 -7.96528101e-01 -6.93495572e-01
6.87270463e-02 1.52228296e-01 -2.88073033e-01 -6.10752776e-02
6.56350255e-01 2.31997028e-01 -1.58275640e+00 2.44881332e-01
1.36313784e+00 7.86278009e-01 1.37667358e-01 1.04470350e-01
-2.00375974e-01 3.10389876e-01 -4.62034523e-01 -2.94832289e-01
4.49482143e-01 -8.25890243e-01 -6.22605920e-01 1.97380453e-01
-3.58826697e-01 -2.61107504e-01 1.24004096e-01 1.15559947e+00
1.83488533e-01 3.96665215e-01 7.61477888e-01 -1.54082513e+00
-3.82671356e-01 7.65374899e-01 -5.42753041e-01 -4.50348631e-02
3.57117265e-01 1.82290301e-01 1.09479010e+00 -5.89205086e-01
2.54853964e-01 1.14721107e+00 1.23704404e-01 4.52187002e-01
-1.47509813e+00 -7.66150177e-01 5.94378889e-01 4.31947380e-01
-9.78292406e-01 -9.40140560e-02 1.18280733e+00 -2.98563093e-01
4.66083407e-01 6.29645407e-01 7.03144848e-01 5.61019540e-01
3.65167677e-01 7.35470712e-01 1.21843851e+00 -5.09633124e-01
5.60375214e-01 3.51663768e-01 -7.24835172e-02 4.02273983e-01
5.47372818e-01 3.22872490e-01 -2.94570774e-01 -5.76195955e-01
2.41579160e-01 2.73174912e-01 -5.77007830e-01 -5.76350510e-01
-9.60668206e-01 9.39410925e-01 4.61070329e-01 3.38395648e-02
-6.45978808e-01 -9.33928192e-02 1.63335413e-01 3.24041694e-01
3.86286557e-01 8.52480590e-01 -5.56906104e-01 5.03632873e-02
-3.30002397e-01 5.30557334e-01 6.13130927e-01 1.84429809e-01
6.43316269e-01 -9.06206593e-02 -3.07966858e-01 5.84532022e-01
1.87189773e-01 2.21876785e-01 5.52640021e-01 -6.85356259e-01
5.31223476e-01 8.45659018e-01 6.54220343e-01 -1.14086556e+00
-1.07164904e-01 -3.66914749e-01 -6.08621478e-01 3.34509522e-01
3.06139290e-01 -3.77053887e-01 -5.14055014e-01 1.92085445e+00
4.42917377e-01 3.89585018e-01 1.11568168e-01 1.07893622e+00
-4.60741073e-02 7.07445920e-01 1.80257037e-02 -8.33214223e-01
9.82805550e-01 -3.18753183e-01 -7.73986995e-01 -1.14958778e-01
6.29722059e-01 -4.50960428e-01 1.08454549e+00 3.00561100e-01
-6.09082639e-01 2.01606322e-02 -1.08610737e+00 7.85720944e-01
-9.81216133e-02 -1.85599908e-01 6.95522547e-01 4.26177412e-01
-3.40061694e-01 9.67821777e-01 -4.34420496e-01 1.57854229e-01
2.10203126e-01 4.69772696e-01 -8.85652080e-02 1.34193748e-01
-1.35581887e+00 6.90956235e-01 5.33684611e-01 1.52263194e-01
-5.30224502e-01 -7.89846539e-01 -7.53772795e-01 3.74870926e-01
8.95294070e-01 -5.44038117e-01 1.17576158e+00 -1.16436636e+00
-1.50258946e+00 1.30182594e-01 -9.71354768e-02 -5.67744970e-01
7.42220879e-01 -9.19527188e-02 -3.73496652e-01 -2.60901839e-01
2.00391680e-01 1.74740553e-01 8.89443934e-01 -1.25783944e+00
-7.87101924e-01 -4.91586685e-01 2.19418049e-01 4.62391853e-01
-2.14230612e-01 -2.51765251e-01 3.27458054e-01 -4.57040846e-01
6.40991926e-02 -1.01972890e+00 -5.55363417e-01 -3.05997878e-01
-7.99673080e-01 -1.39545247e-01 6.66004300e-01 -2.99020618e-01
1.38329256e+00 -1.96081805e+00 -1.95876621e-02 4.70911860e-01
-6.98262155e-02 -2.93989796e-02 1.57753125e-01 1.75669372e-01
-3.91223073e-01 3.50856334e-01 -5.42215228e-01 2.06957623e-01
-5.74780554e-02 -1.94386933e-02 -5.33468783e-01 4.20884252e-01
1.15561388e-01 3.24279249e-01 -7.36824811e-01 -1.26436487e-01
1.52997285e-01 -4.73378524e-02 -8.13614726e-01 3.03281039e-01
-3.10142249e-01 2.61797994e-01 -8.08804631e-01 1.98105365e-01
6.69649482e-01 -7.31420815e-02 4.26160038e-01 5.45845740e-02
-3.32489699e-01 1.80920765e-01 -1.46442688e+00 7.91237712e-01
-3.98604959e-01 2.05930918e-01 -3.96942735e-01 -1.34259951e+00
6.97053969e-01 3.36021036e-01 3.77213866e-01 -2.05433831e-01
1.79110318e-01 4.34750438e-01 2.87022591e-01 -3.79571855e-01
6.34113103e-02 -3.33669841e-01 -6.18887953e-02 6.29814029e-01
-7.43425012e-01 1.65256634e-01 -2.53136586e-02 -9.93531868e-02
8.43909204e-01 -1.80557519e-01 8.32181811e-01 -2.41566822e-01
7.41292417e-01 -5.38022593e-02 9.17785406e-01 6.89866066e-01
9.50742140e-02 2.30980366e-01 9.38190460e-01 -5.25403619e-01
-7.31935143e-01 -4.99413669e-01 6.38076216e-02 4.60931212e-01
1.16035208e-01 -1.52218994e-02 -4.30465311e-01 -1.04364991e+00
3.14035356e-01 1.23940361e+00 -7.17213809e-01 -4.61683959e-01
-3.07353020e-01 -7.78914630e-01 -2.19546869e-01 1.02204658e-01
7.05902040e-01 -9.54587936e-01 -9.33594048e-01 2.63492376e-01
-1.74371108e-01 -3.58073652e-01 -6.95520937e-01 8.73451680e-02
-7.93040037e-01 -1.20028663e+00 -4.25464720e-01 9.45263356e-02
9.69503462e-01 3.05283666e-01 5.21348417e-01 1.24334551e-01
1.74797103e-01 -1.93189412e-01 -2.33124644e-01 -4.43148494e-01
-2.21426994e-01 -3.02393287e-01 2.95855701e-01 4.94064152e-01
7.36815184e-02 -4.63629097e-01 -7.10559249e-01 1.97553366e-01
-9.83196139e-01 2.28721276e-01 6.98683798e-01 1.10581350e+00
4.54718709e-01 3.75241399e-01 7.26740599e-01 -1.07280576e+00
8.56317163e-01 -6.72364950e-01 -7.71701574e-01 4.17604268e-01
-1.15132177e+00 6.46948099e-01 1.00641453e+00 -5.66067636e-01
-1.27385283e+00 1.55760190e-02 3.86933267e-01 -3.06346327e-01
2.11392604e-02 8.14759135e-01 -6.82978034e-01 3.59667897e-01
3.11653495e-01 2.07570642e-01 -2.03147933e-01 -3.83417398e-01
2.06375852e-01 6.58778846e-01 1.23915359e-01 -5.71532607e-01
5.61556220e-01 1.78377628e-01 1.79231957e-01 -2.63580382e-01
-6.18285239e-01 -1.21729493e-01 -1.24673016e-01 -1.26532584e-01
4.62435722e-01 -1.91628590e-01 -8.34218085e-01 4.35349420e-02
-1.00594032e+00 1.84876412e-01 -1.54314011e-01 6.60007119e-01
-5.53200126e-01 1.16537414e-01 5.08109927e-01 -1.11559045e+00
1.37852151e-02 -1.14452517e+00 2.96804935e-01 2.50655740e-01
-2.27740541e-01 -6.42688513e-01 -1.55005455e-01 3.76978740e-02
5.64545915e-02 5.76045096e-01 1.21972895e+00 -9.71358240e-01
-4.60672647e-01 -2.58164823e-01 8.87782946e-02 4.46337648e-03
3.51008892e-01 -7.06201270e-02 -3.76232564e-01 -4.16237749e-02
8.68288651e-02 2.82036692e-01 5.94501853e-01 5.30335844e-01
1.32990420e+00 -1.14713287e+00 -5.73197007e-01 3.46799165e-01
1.38516498e+00 9.34404075e-01 3.30736339e-01 5.66433966e-01
3.23694348e-01 6.05627120e-01 8.83850217e-01 8.00649524e-01
-5.09725735e-02 8.75412941e-01 5.04181147e-01 1.70389563e-01
7.73712993e-01 -4.04500455e-01 1.81645483e-01 -2.35107839e-01
-9.73290950e-02 -1.49955615e-01 -5.74835598e-01 3.68936986e-01
-2.02240491e+00 -9.66132224e-01 5.43624423e-02 2.75588083e+00
6.91360831e-01 1.42563209e-01 -3.90614979e-02 3.95622939e-01
8.91514361e-01 -1.60457745e-01 -9.83448565e-01 -6.62379086e-01
2.33611792e-01 -4.21625197e-01 3.70857418e-01 5.32455862e-01
-7.49400437e-01 2.48574898e-01 4.74588919e+00 6.08410418e-01
-1.12786973e+00 -3.56361121e-01 7.46197104e-01 -1.28776491e-01
-8.61480951e-01 3.46825540e-01 -4.86220300e-01 7.32346714e-01
6.06080353e-01 -9.35531735e-01 4.63116974e-01 8.12421679e-01
7.37675011e-01 -1.40677184e-01 -1.15698922e+00 5.46239555e-01
-3.64817947e-01 -1.25312221e+00 3.06333750e-01 2.27025956e-01
6.99947238e-01 -8.83320451e-01 -2.89007220e-02 2.01611996e-01
1.42937183e-01 -7.20049679e-01 5.97731948e-01 3.38939548e-01
4.75401342e-01 -1.20341027e+00 7.97096610e-01 6.47755802e-01
-6.58517838e-01 -4.05632645e-01 -1.10501893e-01 -2.57815301e-01
-4.14495803e-02 6.71374619e-01 -9.40488219e-01 6.57675445e-01
2.10743114e-01 3.16059232e-01 -3.02650984e-02 9.19684231e-01
-4.63876754e-01 4.53947961e-01 -1.48679391e-01 -2.62323111e-01
7.38861114e-02 -3.44265074e-01 9.07380760e-01 5.78723788e-01
4.53087807e-01 5.07621884e-01 1.51197195e-01 9.46005166e-01
7.58980960e-02 2.97564059e-01 -7.63404489e-01 -1.92712676e-02
7.14662969e-01 6.98058009e-01 -5.76084375e-01 -2.89596140e-01
-6.13204390e-02 8.68396997e-01 8.00257698e-02 3.91185760e-01
-8.28735352e-01 -3.33759278e-01 7.73714066e-01 7.47310892e-02
1.19801359e-02 4.68311310e-01 -5.48433125e-01 -1.07903087e+00
3.38000692e-02 -8.53467047e-01 5.54441452e-01 -4.18538660e-01
-9.96037841e-01 2.08762899e-01 2.04334766e-01 -1.47135794e+00
-4.38146919e-01 2.37744879e-02 -8.43925834e-01 8.15499604e-01
-1.12331069e+00 -3.66480380e-01 1.44549683e-01 3.47235262e-01
5.28218865e-01 6.07232051e-03 6.28823400e-01 -2.75814742e-01
-7.46037602e-01 4.32293355e-01 1.02894157e-01 -4.33616638e-01
2.35204607e-01 -1.26869118e+00 -3.44163328e-01 8.83301139e-01
-1.18380181e-01 5.87382495e-01 1.30807543e+00 -7.04680204e-01
-1.12922716e+00 -8.76178205e-01 1.03651345e+00 2.17049330e-01
3.01844269e-01 1.23068728e-01 -9.10721958e-01 6.35613739e-01
-9.96915177e-02 -2.91924924e-01 5.75984299e-01 7.15787485e-02
6.29685149e-02 -3.24487150e-01 -1.43432128e+00 1.08121312e+00
6.90606117e-01 1.40335158e-01 -6.95474267e-01 8.70142430e-02
8.24644327e-01 -8.20355564e-02 -3.61785054e-01 6.28398001e-01
5.14033914e-01 -9.06268239e-01 6.33636415e-01 -8.47729862e-01
6.10174596e-01 -3.84734899e-01 -5.61659001e-02 -1.83038688e+00
-1.97062522e-01 -6.58530831e-01 -3.05291172e-02 1.08965755e+00
5.15508115e-01 -1.12507260e+00 6.48587406e-01 1.24623263e+00
1.86532229e-01 -1.07898676e+00 -8.72545779e-01 -7.98497379e-01
-3.54279518e-01 -2.80151308e-01 1.12333107e+00 9.95274305e-01
3.77518594e-01 1.68555349e-01 -4.25351977e-01 1.35759920e-01
6.66566730e-01 5.20590305e-01 7.10103810e-01 -1.02202988e+00
-3.56561393e-01 -4.30703998e-01 -1.55072853e-01 -5.44870973e-01
3.39696050e-01 -6.30859196e-01 -1.16425147e-02 -1.25089705e+00
3.03228855e-01 -4.11652476e-01 -5.35243452e-01 6.11623228e-01
-6.58448160e-01 -6.27959073e-01 2.39357859e-01 1.64312124e-01
4.25699959e-03 8.52553666e-01 1.02591932e+00 6.70527592e-02
-7.48199403e-01 4.90774095e-01 -1.03994679e+00 8.53221118e-01
9.12855268e-01 -6.85886800e-01 -4.93554741e-01 2.18448222e-01
4.50499132e-02 4.23539013e-01 2.74875224e-01 -3.94977242e-01
-1.04021944e-01 -7.35456347e-01 1.65545657e-01 -4.30276573e-01
-1.18870422e-01 -1.07596767e+00 4.71341878e-01 8.55513334e-01
-6.38208747e-01 -7.87611678e-02 -3.12887393e-02 7.40431905e-01
-1.07834145e-01 -2.44027108e-01 5.82441449e-01 4.52615842e-02
-5.78080893e-01 6.69200206e-03 -2.07768887e-01 -4.09193605e-01
1.23622406e+00 -2.65587382e-02 -1.00012377e-01 -5.00310242e-01
-5.59676588e-01 3.24467212e-01 1.93990633e-01 3.36322725e-01
7.25937665e-01 -1.44448841e+00 -7.82198846e-01 1.01202009e-02
7.05990940e-02 -3.35664749e-01 1.10993482e-01 6.59581363e-01
1.00852266e-01 6.92211509e-01 1.32092778e-02 -4.91414927e-02
-1.22974956e+00 5.93407691e-01 3.15493643e-01 -4.12021905e-01
-3.80356401e-01 2.79963225e-01 4.12521988e-01 -1.91812411e-01
-2.63080567e-01 -1.49619102e-01 -4.56910044e-01 -3.00042890e-02
4.93799269e-01 4.47528094e-01 -1.41715020e-01 -2.06795763e-02
-3.54690969e-01 -5.14818840e-02 -3.73850949e-02 -3.07840705e-01
1.27471590e+00 -3.66568901e-02 2.34590676e-02 3.76036078e-01
9.19775665e-01 -8.06177482e-02 -1.24579585e+00 4.12590895e-03
1.38574123e-01 -1.06917238e+00 -2.68980581e-02 -8.13568890e-01
-8.26117933e-01 2.81687140e-01 3.45429718e-01 4.11004603e-01
1.36538804e+00 -3.83840203e-01 3.10965002e-01 2.64661998e-01
3.07910591e-01 -9.16149020e-01 -3.79999161e-01 -1.26252070e-01
1.15466297e+00 -1.21754587e+00 -8.59264657e-02 -3.93594414e-01
-8.16982508e-01 7.93627441e-01 6.73663437e-01 8.86053070e-02
5.77626288e-01 -2.30085239e-01 -2.82694012e-01 1.04308240e-01
-9.23497796e-01 1.15789682e-01 4.16047782e-01 7.57271126e-02
2.80269027e-01 5.06156981e-01 -1.17006111e+00 8.70580554e-01
-1.38944283e-01 1.59657240e-01 5.24982750e-01 7.01282740e-01
-2.74784595e-01 -1.05287242e+00 -6.10175729e-01 7.69102395e-01
-4.90968168e-01 8.74057487e-02 -3.37007791e-01 7.69302011e-01
9.73385796e-02 1.00617278e+00 -7.24732429e-02 -3.60730559e-01
5.73359191e-01 7.86859617e-02 -2.64859628e-02 -5.67838669e-01
-1.79307699e-01 -1.37263358e-01 3.13614979e-02 -4.92918521e-01
-3.02500334e-02 -8.72455597e-01 -1.42699921e+00 -2.52069235e-01
-4.46502239e-01 6.41449630e-01 6.68029904e-01 1.32692099e+00
2.89142698e-01 3.91511142e-01 1.27150798e+00 -4.50731456e-01
-9.84024167e-01 -6.75569296e-01 -6.12849653e-01 4.63529706e-01
2.07007095e-01 -8.85960698e-01 -7.37018406e-01 -2.53395557e-01]
|
[8.61865520477295, 5.538671016693115]
|
78d510fd-2843-4b25-ab75-b65a2d6aeee9
|
salad-part-level-latent-diffusion-for-3d
|
2303.12236
| null |
https://arxiv.org/abs/2303.12236v1
|
https://arxiv.org/pdf/2303.12236v1.pdf
|
SALAD: Part-Level Latent Diffusion for 3D Shape Generation and Manipulation
|
We present a cascaded diffusion model based on a part-level implicit 3D representation. Our model achieves state-of-the-art generation quality and also enables part-level shape editing and manipulation without any additional training in conditional setup. Diffusion models have demonstrated impressive capabilities in data generation as well as zero-shot completion and editing via a guided reverse process. Recent research on 3D diffusion models has focused on improving their generation capabilities with various data representations, while the absence of structural information has limited their capability in completion and editing tasks. We thus propose our novel diffusion model using a part-level implicit representation. To effectively learn diffusion with high-dimensional embedding vectors of parts, we propose a cascaded framework, learning diffusion first on a low-dimensional subspace encoding extrinsic parameters of parts and then on the other high-dimensional subspace encoding intrinsic attributes. In the experiments, we demonstrate the outperformance of our method compared with the previous ones both in generation and part-level completion and manipulation tasks.
|
['Minhyuk Sung', 'Minh Hieu Nguyen', 'Seungwoo Yoo', 'Juil Koo']
|
2023-03-21
| null | null | null | null |
['3d-shape-generation']
|
['computer-vision']
|
[ 9.41651314e-02 2.08710253e-01 -6.80445433e-02 -1.02110077e-02
-4.28474069e-01 -5.61861217e-01 1.10205007e+00 -1.23925298e-01
-1.72177806e-01 3.75072449e-01 5.04210055e-01 1.82973325e-01
-6.99053109e-02 -1.00928116e+00 -6.81533635e-01 -6.22166753e-01
1.32563233e-01 8.10433209e-01 -3.30644846e-02 -4.19955164e-01
1.87877253e-01 6.96630478e-01 -1.48304975e+00 1.21778890e-01
9.50840890e-01 7.19713032e-01 2.70115197e-01 7.23082423e-01
-2.70427942e-01 3.98505896e-01 -4.28814530e-01 -4.35511887e-01
4.91713345e-01 -4.02711093e-01 -3.21359992e-01 4.11542356e-01
4.86524284e-01 -5.27192354e-01 -6.07110739e-01 5.89271545e-01
8.08245003e-01 2.29794770e-01 1.28852749e+00 -1.03292596e+00
-1.62095678e+00 2.40258679e-01 -4.89563257e-01 -4.82818305e-01
4.28263098e-01 2.11186871e-01 8.81891847e-01 -1.31666720e+00
1.24503529e+00 1.21008229e+00 4.21164930e-01 1.01446819e+00
-1.57094407e+00 -3.64353716e-01 1.09625511e-01 -2.75366127e-01
-1.21418345e+00 -3.29975635e-01 1.11497617e+00 -6.78542018e-01
8.07378054e-01 -1.42477155e-01 9.59638238e-01 1.48983443e+00
8.70533958e-02 9.98615742e-01 9.48156178e-01 -2.66532481e-01
2.89650947e-01 -5.77853695e-02 -3.93403590e-01 9.65637028e-01
9.75339487e-02 1.89154401e-01 -6.97669089e-01 -2.26974059e-02
1.47602808e+00 -6.90591559e-02 -1.78623006e-01 -9.48117197e-01
-1.51889706e+00 9.19845939e-01 4.77944374e-01 2.31024697e-01
-5.46815336e-01 3.10071230e-01 4.86810841e-02 3.73625219e-01
8.26294839e-01 4.42912608e-01 -3.09268862e-01 6.41417280e-02
-1.03314555e+00 3.62769634e-01 7.20859528e-01 1.34139204e+00
7.11339772e-01 2.47326165e-01 -8.12270522e-01 9.38748479e-01
1.66122783e-02 6.67155027e-01 4.87375289e-01 -9.04567719e-01
4.18544173e-01 6.45388007e-01 1.46851286e-01 -7.03703940e-01
-1.02620333e-01 -2.63774484e-01 -1.30014086e+00 5.51896453e-01
1.05414413e-01 -1.23144038e-01 -1.33196986e+00 1.86951506e+00
4.08984005e-01 1.01627201e-01 5.74747771e-02 7.29086936e-01
6.22929931e-01 4.87395406e-01 -1.07193232e-01 -3.51370089e-02
8.34602833e-01 -1.14134371e+00 -7.48234689e-01 4.82474625e-01
6.02102757e-01 -8.31495762e-01 1.02724993e+00 2.71843314e-01
-1.37776792e+00 -7.35219121e-01 -8.88564587e-01 -5.63173413e-01
-3.31131667e-01 4.72565621e-01 8.35408330e-01 5.06342471e-01
-1.28894472e+00 8.28244567e-01 -7.31665015e-01 -1.70662701e-01
6.30482256e-01 2.08152071e-01 -7.09796071e-01 -1.42107680e-01
-8.03634942e-01 7.11169839e-01 -1.08837545e-01 -9.56061333e-02
-1.19821131e+00 -8.72735977e-01 -9.41357493e-01 -2.41532862e-01
1.59492180e-01 -1.51571703e+00 7.86604404e-01 -3.59041303e-01
-1.89163733e+00 6.81668520e-01 1.35827631e-01 -1.68223992e-01
9.33549821e-01 -5.30815423e-01 1.35535941e-01 -6.46465272e-02
2.30008028e-02 1.03898227e+00 1.32415318e+00 -1.28772414e+00
-1.47896716e-02 -4.51210409e-01 1.20319396e-01 3.70137721e-01
-6.42008126e-01 -6.92484736e-01 -4.93488491e-01 -1.34588349e+00
1.39509887e-01 -9.76899087e-01 -3.08957189e-01 7.95758665e-01
-2.61864424e-01 -1.77064404e-01 8.92725527e-01 -5.93457162e-01
1.02422428e+00 -1.83158100e+00 8.83492708e-01 -2.84602288e-02
2.68297702e-01 2.39370406e-01 -5.59975147e-01 7.09562242e-01
1.96407616e-01 1.34912267e-01 -3.90755922e-01 -8.77178133e-01
1.44566447e-01 1.49601370e-01 -3.01304072e-01 2.05736294e-01
5.07304430e-01 1.10447800e+00 -9.59781647e-01 -2.11676955e-01
2.14210093e-01 9.41403329e-01 -8.12193930e-01 3.89391869e-01
-2.27316856e-01 6.35062039e-01 -3.65618885e-01 6.75492108e-01
6.09580636e-01 5.00671715e-02 -3.84203345e-01 -2.99510598e-01
1.95716172e-02 -3.40670854e-01 -1.23273504e+00 2.74438405e+00
-6.79830790e-01 2.02374458e-01 -1.81659462e-03 -6.13745928e-01
1.10774112e+00 3.22423398e-01 4.16901082e-01 -3.59676749e-01
-2.81810731e-01 3.27602535e-01 -3.13277096e-01 -1.66394621e-01
7.70390153e-01 -1.83882996e-01 -6.31854236e-02 7.35424399e-01
4.34780955e-01 -8.76211107e-01 7.46017247e-02 6.46947801e-01
9.16661918e-01 8.26885939e-01 -2.07983106e-01 -1.91376001e-01
4.69941109e-01 -3.80887985e-01 1.64286196e-01 6.26300335e-01
3.15622360e-01 1.05905211e+00 2.11511001e-01 -9.69695970e-02
-1.36209524e+00 -1.15853179e+00 1.29235253e-01 6.27056360e-01
3.91004905e-02 -4.83244479e-01 -7.36129999e-01 -7.75011420e-01
3.05942029e-01 8.53811324e-01 -9.53744352e-01 -2.92074621e-01
-4.87494111e-01 -3.96006763e-01 4.85257596e-01 7.75547624e-01
3.61686796e-01 -8.79034579e-01 -1.16249081e-02 2.57534713e-01
2.11396292e-01 -8.72981310e-01 -7.58256555e-01 -2.50754982e-01
-1.16596568e+00 -6.56393230e-01 -1.47715521e+00 -6.80143595e-01
1.01200557e+00 2.23741710e-01 7.57401109e-01 1.34925451e-02
-5.22114038e-01 6.52514875e-01 -3.61934483e-01 -3.88385057e-01
-4.48780656e-01 1.36398792e-01 2.24382669e-01 1.37482896e-01
-4.74155426e-01 -9.32111204e-01 -8.42651069e-01 8.88811499e-02
-1.10209537e+00 1.02193892e-01 8.55673552e-01 9.72590625e-01
5.65389037e-01 -4.10350114e-01 6.04847312e-01 -8.87803555e-01
9.79167938e-01 -5.92533126e-02 -2.21698478e-01 9.26419050e-02
-6.74414694e-01 5.30199051e-01 4.34695601e-01 -5.71952224e-01
-1.32625020e+00 9.82300937e-02 -1.88910574e-01 -8.77641320e-01
1.60757676e-02 -8.60762224e-02 -1.01958273e-03 1.73502509e-03
7.03095376e-01 2.39394143e-01 7.68857971e-02 -8.83387744e-01
1.03242123e+00 1.95284799e-01 1.21359825e-01 -8.18481982e-01
1.16464496e+00 6.69687092e-01 2.15679929e-01 -8.10561061e-01
-4.13650095e-01 -1.05369322e-01 -1.21794844e+00 -3.39518599e-02
7.66325533e-01 -9.75126207e-01 -1.31949127e-01 8.14401984e-01
-1.16830027e+00 -5.62103271e-01 -9.20972228e-01 2.73398101e-01
-9.74738359e-01 4.16017026e-01 -8.21099639e-01 -6.97524130e-01
-5.46974182e-01 -9.23187554e-01 1.64728546e+00 -2.00940415e-01
-7.26133138e-02 -1.00240362e+00 2.25855947e-01 1.92093719e-02
5.05528927e-01 2.90271163e-01 8.88727129e-01 -5.43577969e-02
-7.31388032e-01 -2.47140452e-01 7.22341612e-02 3.39822322e-01
1.50540873e-01 -8.85916501e-02 -7.44210184e-01 -3.46511453e-01
-2.16138959e-01 -4.19163734e-01 1.22686100e+00 1.89335823e-01
1.05052400e+00 1.26846328e-01 -3.70896190e-01 7.31426179e-01
1.22270107e+00 -4.23980862e-01 4.97124255e-01 -3.04663420e-01
9.90002990e-01 6.12437665e-01 5.01462996e-01 5.48653841e-01
1.57840997e-01 5.95490813e-01 2.18743399e-01 -1.54429868e-01
-8.44469130e-01 -7.27861762e-01 3.25111628e-01 9.83440578e-01
-6.31785989e-01 -2.79165506e-01 -3.67594302e-01 6.95835650e-01
-1.73308837e+00 -8.35877061e-01 6.16878308e-02 2.23715782e+00
6.54865801e-01 -1.78667814e-01 -9.60100144e-02 6.41630031e-03
4.00619507e-01 2.93085605e-01 -6.28437400e-01 -1.70785412e-01
-3.00795168e-01 3.11910748e-01 9.05308872e-03 4.87774730e-01
-8.38565886e-01 1.21859837e+00 6.23723793e+00 9.23162162e-01
-7.66846061e-01 1.22030623e-01 3.32271934e-01 -1.75723150e-01
-7.29048908e-01 -1.37037471e-01 -5.69053173e-01 -7.60234669e-02
1.35526776e-01 -1.12483330e-01 3.61775428e-01 6.58337355e-01
-1.78489499e-02 1.74454391e-01 -1.24918437e+00 9.56033766e-01
3.37536305e-01 -1.53922963e+00 7.06824064e-01 2.85663486e-01
1.21421564e+00 -4.77752000e-01 2.65618443e-01 3.17061812e-01
3.26704562e-01 -7.99605548e-01 9.30943131e-01 8.99962485e-01
1.15857542e+00 -5.86440802e-01 1.19345769e-01 6.49358392e-01
-1.04766929e+00 7.75643885e-02 -4.63516384e-01 1.97016597e-01
5.00436008e-01 6.01357937e-01 -4.91881788e-01 7.28306830e-01
1.41065404e-01 8.82911205e-01 -2.66131014e-01 6.25999689e-01
-4.98292327e-01 8.00855905e-02 -1.16544954e-01 1.74484745e-01
-4.24945168e-02 -6.01286530e-01 6.69503689e-01 1.06437349e+00
8.12512338e-01 2.26249397e-01 -4.17401269e-02 1.13516188e+00
-3.56173873e-01 8.51887017e-02 -8.93792391e-01 -1.73119426e-01
5.09564355e-02 1.34273326e+00 -4.56932843e-01 -4.52633977e-01
-8.64473358e-02 1.80495882e+00 5.45381546e-01 3.71206313e-01
-5.67881107e-01 -3.46542656e-01 7.13253021e-01 1.08925343e-01
6.16107106e-01 -8.26405644e-01 -1.40197083e-01 -1.44523358e+00
5.64116100e-03 -3.41994792e-01 -1.46075249e-01 -8.00646424e-01
-1.52450013e+00 4.94188130e-01 -1.70515403e-01 -1.07042634e+00
-1.44846439e-01 -6.14011526e-01 -3.86947125e-01 1.05153561e+00
-1.39592159e+00 -1.72973251e+00 -1.90697864e-01 8.24542820e-01
7.58253634e-01 -1.51558593e-01 1.10434175e+00 1.67326972e-01
-1.79856077e-01 5.52210152e-01 7.64941871e-02 -1.54362330e-02
9.14995015e-01 -1.39196873e+00 6.42496884e-01 4.11484748e-01
3.23266804e-01 6.54830456e-01 2.07592353e-01 -9.96081233e-01
-1.70782351e+00 -1.07850003e+00 6.13815010e-01 -6.94910884e-01
3.08402687e-01 -7.06095457e-01 -5.38793802e-01 4.63985354e-01
1.00717761e-01 -1.08714201e-01 4.59990233e-01 -2.83695124e-02
-3.45860094e-01 1.91159785e-01 -1.12672830e+00 7.92816341e-01
1.84397388e+00 -5.76151729e-01 -4.33477640e-01 3.97010207e-01
8.41648042e-01 -4.01553988e-01 -1.11129391e+00 3.75454038e-01
5.34924448e-01 -7.61938274e-01 1.04153979e+00 -5.22746205e-01
5.92549264e-01 -9.64019075e-02 6.69525191e-02 -1.61186147e+00
-5.56816459e-01 -8.72239888e-01 -6.07258976e-01 1.27542186e+00
2.66975880e-01 -1.88180730e-01 7.88139164e-01 3.80065948e-01
-2.44603217e-01 -7.70544052e-01 -5.13664544e-01 -8.01158428e-01
2.45801702e-01 -2.98611611e-01 5.75273573e-01 7.66917109e-01
-4.76828694e-01 5.59763432e-01 -6.43759549e-01 -1.95666298e-01
6.47540689e-01 2.43749470e-01 1.02895308e+00 -1.19765913e+00
-2.63996184e-01 -4.58440930e-01 -3.12687039e-01 -1.58647799e+00
1.50891960e-01 -1.01311040e+00 -2.01717913e-01 -1.85564327e+00
-2.13533211e-02 -5.46878278e-01 2.30038837e-01 7.29319677e-02
-1.26352385e-01 3.25228691e-01 3.37470502e-01 4.33196098e-01
2.71231942e-02 1.34820354e+00 2.08373690e+00 -2.70800561e-01
-2.67654151e-01 -1.61563307e-01 -4.98854756e-01 3.83714765e-01
2.78316468e-01 -1.82803065e-01 -6.55587018e-01 -8.07572424e-01
-3.16951983e-02 -1.45106055e-02 7.86755309e-02 -8.78182828e-01
1.07622452e-01 7.55652264e-02 6.72236264e-01 -4.32651639e-01
8.96310747e-01 -5.47887385e-01 6.38305694e-02 3.34495097e-01
-4.58836585e-01 -6.18617833e-02 -1.05509236e-01 1.11642611e+00
5.68938022e-03 4.24039848e-02 3.69856566e-01 -2.53762841e-01
-4.54610854e-01 8.98837507e-01 -8.51231664e-02 -3.01158309e-01
1.13331747e+00 -2.67640412e-01 4.01594415e-02 -4.34490502e-01
-1.14367509e+00 -2.08885282e-01 8.83449376e-01 6.96459353e-01
7.93260813e-01 -1.73301709e+00 -9.12368238e-01 6.74572766e-01
1.46665558e-01 7.72392824e-02 4.08586264e-01 5.23883581e-01
-3.43809396e-01 1.23891324e-01 -3.01675528e-01 -3.74189824e-01
-8.86222363e-01 7.37838268e-01 -5.42875491e-02 -2.79376417e-01
-7.72944212e-01 1.04819763e+00 4.31036443e-01 -6.05847061e-01
-1.24443043e-02 -2.63523668e-01 5.26952036e-02 6.65906742e-02
2.55998850e-01 4.65642005e-01 -1.00858726e-01 -7.06153870e-01
2.11916968e-01 9.66420293e-01 -6.11874498e-02 -5.78748941e-01
1.30799961e+00 1.35246634e-01 2.21213102e-02 3.36342573e-01
9.43705916e-01 9.76660177e-02 -1.56061924e+00 -7.40335286e-02
-5.92036366e-01 -5.57350814e-01 7.57113770e-02 -6.41676784e-01
-9.98459637e-01 1.32216501e+00 3.00282180e-01 -2.20286250e-01
8.52189839e-01 -3.57740074e-02 9.06958342e-01 2.60183573e-01
6.01358354e-01 -1.19713390e+00 6.71626866e-01 4.77759391e-01
1.54088652e+00 -9.38550889e-01 -4.13613068e-03 -5.65490484e-01
-7.72550642e-01 1.04825437e+00 5.13392329e-01 -4.38309282e-01
8.46154034e-01 -5.37229106e-02 -2.37974077e-01 -1.58241481e-01
-8.09888005e-01 -2.53036112e-01 4.66349632e-01 1.00471580e+00
4.24594194e-01 5.58963977e-02 -3.87312353e-01 1.70636177e-01
4.62436341e-02 -1.60076991e-02 3.08979601e-01 8.77171636e-01
-1.39496494e-02 -1.62124026e+00 -1.12613169e-02 2.41113946e-01
1.40043646e-01 -6.90809935e-02 -5.87908804e-01 6.70348823e-01
7.96586499e-02 4.87056077e-01 -4.37141545e-02 -3.42729688e-01
5.33306837e-01 8.39194357e-02 9.79210079e-01 -7.37011433e-01
-4.57977027e-01 6.12569861e-02 -5.90111166e-02 -5.92706740e-01
-4.07073528e-01 -7.00755477e-01 -8.78681898e-01 -5.21754585e-02
-3.93696904e-01 -2.45544240e-01 7.96801329e-01 4.36341733e-01
8.21405053e-01 3.80894005e-01 5.15432477e-01 -1.31773770e+00
-8.34432960e-01 -1.06798160e+00 -8.42807531e-01 8.16359758e-01
1.30282313e-01 -9.11661088e-01 -1.03511751e-01 1.84450075e-01]
|
[9.223625183105469, -3.3666768074035645]
|
9d3b646d-5d34-4b2f-af58-84a2f141e972
|
adaptive-dereverberation-noise-and-interferer
|
2303.07027
| null |
https://arxiv.org/abs/2303.07027v1
|
https://arxiv.org/pdf/2303.07027v1.pdf
|
Adaptive Dereverberation, Noise and Interferer Reduction Using Sparse Weighted Linearly Constrained Minimum Power Beamforming
|
Interfering sources, background noise and reverberation degrade speech quality and intelligibility in hearing aid applications. In this paper, we present an adaptive algorithm aiming at dereverberation, noise and interferer reduction and preservation of binaural cues based on the wBLCMP beamformer. The wBLCMP beamformer unifies the multi-channel weighted prediction error method performing dereverberation and the linearly constrained minimum power beamformer performing noise and interferer reduction into a single convolutional beamformer. We propose to adaptively compute the optimal filter by incorporating an exponential window into a sparsity-promoting lp-norm cost function, which enables to track a moving target speaker. Simulation results with successive target speakers at different positions show that the proposed adaptive version of the wBLCMP beamformer outperforms a non-adaptive version in terms of objective speech enhancement performance measures.
|
['Simon Doclo', 'Henri Gode']
|
2023-03-13
| null | null | null | null |
['speech-enhancement']
|
['speech']
|
[ 1.40241832e-01 -2.43083060e-01 4.10838455e-01 1.55303702e-01
-8.56210232e-01 -1.45635143e-01 3.11239436e-02 -2.14916915e-01
-5.78722596e-01 6.70003891e-01 9.82654810e-01 -3.15157115e-01
-4.89058375e-01 -2.41042227e-01 -3.73263329e-01 -1.12106740e+00
-1.04075387e-01 -4.64310527e-01 5.80293089e-02 -8.45385119e-02
-6.02052696e-02 2.54859954e-01 -1.69264734e+00 2.18274400e-01
1.05656898e+00 9.77592647e-01 7.33115375e-01 1.09863663e+00
4.80133533e-01 3.13388944e-01 -8.23448956e-01 2.15691030e-02
2.82490760e-01 -5.73463321e-01 2.46099070e-01 8.89882818e-03
5.45242488e-01 -1.39363557e-01 -1.62088022e-01 1.10395455e+00
1.52755022e+00 5.73752403e-01 3.06856275e-01 -6.30695343e-01
-1.68604389e-01 3.93468887e-01 -2.55075991e-01 7.13307500e-01
3.68042231e-01 -1.70450825e-02 7.40418017e-01 -1.17406964e+00
-6.16068579e-02 1.09740257e+00 1.07585144e+00 2.50821143e-01
-1.05673504e+00 -7.95558870e-01 -1.51811883e-01 7.19095469e-01
-1.51072514e+00 -9.44758236e-01 8.92435312e-01 -1.31251141e-01
9.22525525e-01 5.83270490e-01 6.59473360e-01 5.44093668e-01
4.25749719e-02 6.46182656e-01 1.08944845e+00 -9.26970720e-01
2.44302616e-01 -2.40536928e-01 -8.35038349e-03 4.06538129e-01
2.19346195e-01 6.51833534e-01 -9.29619253e-01 -2.16070578e-01
3.45266551e-01 -5.82764566e-01 -1.06040418e+00 1.94427848e-01
-7.93633223e-01 2.96591192e-01 2.11391211e-01 7.55079210e-01
-4.81591374e-01 9.07580554e-03 -1.56288490e-01 3.04853618e-01
6.23714387e-01 2.94446856e-01 -3.84131908e-01 2.15442657e-01
-9.69031751e-01 2.05391422e-01 7.34310269e-01 5.36960065e-01
1.36087567e-01 5.95155239e-01 -5.43316483e-01 1.38383627e+00
7.30151057e-01 9.48222160e-01 5.62902331e-01 -6.26028717e-01
1.40992880e-01 -6.38917863e-01 3.04609239e-01 -8.55226755e-01
-4.55839962e-01 -1.31751215e+00 -6.89786077e-01 5.32793880e-01
2.24703878e-01 -5.24967551e-01 -8.43271136e-01 1.69249570e+00
5.12325108e-01 7.57754505e-01 -5.12828939e-02 1.12047899e+00
4.49787438e-01 8.58024538e-01 -1.73807055e-01 -8.85128975e-01
1.12993634e+00 -1.06722724e+00 -1.37883151e+00 1.90596767e-02
-1.39112279e-01 -1.31324661e+00 5.32802224e-01 9.01929259e-01
-1.37049651e+00 -9.17621613e-01 -1.00044465e+00 3.76410812e-01
9.00219679e-02 2.43871268e-02 -1.55785799e-01 1.22953355e+00
-1.28991997e+00 1.20487303e-01 -6.18970692e-01 3.29643041e-01
9.15107504e-02 5.65189421e-02 -1.00283049e-01 -1.40258018e-02
-8.17371309e-01 1.00243914e+00 -7.97873884e-02 2.90857285e-01
-5.98807633e-01 -9.91921782e-01 -7.46110082e-01 1.32454157e-01
-6.90676644e-02 -5.75690806e-01 1.35973001e+00 -8.60770404e-01
-1.81280518e+00 2.02936642e-02 -7.29981601e-01 -5.99751651e-01
1.85938880e-01 -5.72334290e-01 -1.09768891e+00 -1.46906659e-01
-4.13052708e-01 -3.51298392e-01 1.30173123e+00 -1.20584834e+00
-8.32324207e-01 -6.95670620e-02 -8.65179300e-01 5.58095992e-01
-3.34720701e-01 -5.37380129e-02 -1.92385465e-02 -1.26488411e+00
3.69109362e-01 -1.82276532e-01 -2.14116246e-01 -1.59196675e-01
-1.08315766e-01 9.42588151e-02 5.98500371e-01 -1.21321034e+00
1.64379740e+00 -2.36265850e+00 -2.57834829e-02 1.05157964e-01
-2.83660758e-02 6.85657382e-01 -3.74418825e-01 1.37041584e-01
-2.16670349e-01 -6.53011978e-01 1.78808533e-03 -4.93243247e-01
-1.62336364e-01 -1.53403431e-01 -1.08495668e-01 7.46808767e-01
-3.29174042e-01 5.73931672e-02 -8.79485190e-01 -2.57871538e-01
4.95021790e-01 1.01169193e+00 -7.56292641e-01 4.74871069e-01
4.13718790e-01 3.77500355e-01 3.04355085e-01 4.31888849e-01
9.85603869e-01 7.14986980e-01 -2.97080427e-01 -3.92718643e-01
-4.82999772e-01 2.01162070e-01 -1.51063824e+00 1.45682776e+00
-8.67227435e-01 7.45176971e-01 9.66953874e-01 -6.24908686e-01
8.25990200e-01 7.24113524e-01 1.46708310e-01 -5.69414079e-01
-3.14902514e-02 6.77703321e-01 2.80198038e-01 -5.84219217e-01
2.57077720e-02 -6.15068734e-01 7.59368896e-01 -7.59199113e-02
4.12356436e-01 -2.99846916e-03 -1.71123177e-01 -5.81354797e-01
1.05640721e+00 -4.94166344e-01 4.96788472e-01 -3.51664037e-01
1.02186656e+00 -1.04418182e+00 5.35515964e-01 7.32199192e-01
-3.13822538e-01 6.20839000e-01 -6.74936712e-01 1.26027420e-01
-5.89670479e-01 -1.36117649e+00 -3.91363025e-01 1.27005303e+00
-2.51615971e-01 1.14076436e-01 -6.69985056e-01 1.15407020e-01
-1.92984313e-01 1.08065069e+00 4.42921296e-02 8.95462930e-02
-7.14935839e-01 -4.79394794e-01 2.34882727e-01 3.97639237e-02
3.07151228e-01 -6.75757408e-01 -2.72119075e-01 6.26553535e-01
-3.09838951e-01 -4.82357770e-01 -8.02947104e-01 3.59886706e-01
-3.63495439e-01 -6.02566838e-01 -1.05224168e+00 -1.02767444e+00
4.51858968e-01 2.74206609e-01 5.78158438e-01 -3.38304996e-01
-1.45081103e-01 6.51835799e-01 -4.69901055e-01 -7.21821725e-01
-1.94530725e-01 -7.09136069e-01 2.75809795e-01 4.32679832e-01
-6.55229390e-02 -9.71681654e-01 -9.63136554e-01 2.69998163e-01
-5.50618708e-01 -3.27011764e-01 4.73192602e-01 1.07328081e+00
4.78626072e-01 5.70360065e-01 6.90955222e-01 1.05159029e-01
1.06062496e+00 -1.39115050e-01 -5.06934881e-01 -1.19186826e-01
-6.38237000e-01 -4.07109290e-01 4.73300815e-01 -6.17173851e-01
-1.62508905e+00 -7.89184272e-02 -7.37202108e-01 -1.44071236e-01
-1.38624474e-01 2.11707562e-01 -4.48039919e-01 -2.71861911e-01
7.37694860e-01 4.63141233e-01 -2.44279638e-01 -8.65325749e-01
2.86809385e-01 9.16368365e-01 7.88688481e-01 1.08606391e-01
7.47195959e-01 1.61149397e-01 -2.49577373e-01 -1.07333946e+00
-4.00200158e-01 -9.50048029e-01 -1.29052333e-03 -5.69345057e-01
5.02328515e-01 -7.87015855e-01 -5.20610273e-01 6.00320578e-01
-1.23404193e+00 -1.75006777e-01 -2.20183790e-01 1.05775988e+00
-5.51946342e-01 2.62786478e-01 -3.53540331e-01 -1.45232391e+00
-6.43055022e-01 -8.25155854e-01 4.68437821e-01 1.48730084e-01
-6.25464544e-02 -6.52697146e-01 3.41823399e-01 8.07797611e-02
9.37715650e-01 -2.46532857e-01 4.41316634e-01 -4.03170377e-01
8.46856683e-02 -2.01789483e-01 5.68070233e-01 6.99313879e-01
4.06226307e-01 -9.07313049e-01 -1.19028473e+00 -4.67187911e-01
7.22724438e-01 3.98601085e-01 7.04450905e-01 1.16317284e+00
6.11656547e-01 -6.98231757e-01 -1.00898355e-01 7.59765625e-01
1.35251176e+00 7.46339738e-01 6.38353109e-01 -1.32880986e-01
1.01646543e-01 3.23333830e-01 2.96390742e-01 5.71040213e-01
-2.53866076e-01 5.76816797e-01 2.85991967e-01 -1.52005747e-01
-9.11341488e-01 2.56671980e-02 9.97169539e-02 1.27756059e+00
-3.98586504e-02 -5.64048409e-01 -5.99945366e-01 6.42198205e-01
-1.41099191e+00 -9.54761863e-01 -1.04638278e-01 2.33937216e+00
8.52491856e-01 -3.05815995e-01 -2.00035274e-01 6.73636675e-01
7.55741060e-01 2.28344455e-01 -2.77889371e-01 -2.02285051e-01
-3.65127623e-01 8.74348998e-01 5.28257966e-01 1.24910021e+00
-7.73848534e-01 3.60018164e-01 6.24334145e+00 1.11892116e+00
-9.14787650e-01 6.22733593e-01 2.37608626e-01 -4.18545157e-01
-1.33081095e-03 -7.29026377e-01 -5.78950942e-01 2.93042243e-01
7.49281228e-01 -3.39098155e-01 6.58044398e-01 5.11643231e-01
7.84132302e-01 -1.64677687e-02 -5.75730503e-01 1.30313230e+00
5.00440180e-01 -8.50849211e-01 -6.08352840e-01 -3.46622765e-01
5.04090130e-01 -1.44796640e-01 2.96506882e-01 -1.53402597e-01
-1.21439867e-01 -8.41735482e-01 9.22283590e-01 7.71551371e-01
2.74855018e-01 -6.53980851e-01 5.45669913e-01 2.25758538e-01
-1.12287855e+00 -4.73142505e-01 -1.33888066e-01 -3.91339362e-02
4.29988503e-01 7.95602739e-01 -8.02139521e-01 2.08099589e-01
6.49355888e-01 7.34162866e-04 3.63178939e-01 2.08418536e+00
-4.26846564e-01 8.99895608e-01 -2.76061416e-01 1.18791573e-02
-3.12014610e-01 9.34245214e-02 1.32015288e+00 1.43878353e+00
7.20804691e-01 4.94963080e-01 -2.72814602e-01 2.33074427e-01
3.55697542e-01 1.23697899e-01 -4.68402281e-02 8.19655478e-01
6.14827931e-01 8.87287557e-01 -8.97436887e-02 1.31483123e-01
-2.66033381e-01 7.31384933e-01 -5.95833778e-01 8.21661115e-01
-3.88054252e-01 -5.99157751e-01 8.93134832e-01 1.28412142e-01
7.29039848e-01 -2.41390184e-01 -2.57436782e-01 -4.84238774e-01
-4.86898869e-02 -9.24834669e-01 1.80515915e-01 -1.05146170e+00
-9.14661646e-01 7.84098506e-01 -4.49599564e-01 -1.26766288e+00
7.49652907e-02 -6.78684413e-01 -6.96920991e-01 1.42221475e+00
-1.62035668e+00 -6.19091034e-01 1.76388249e-01 8.98075461e-01
7.91031957e-01 -1.55844837e-01 8.15916181e-01 6.67573988e-01
-2.75368392e-01 9.25953686e-01 3.07273746e-01 -5.11054039e-01
6.87837005e-01 -1.00283098e+00 -5.82483947e-01 1.16279531e+00
-1.77540202e-02 3.93840164e-01 1.49329257e+00 -3.60822439e-01
-1.11448586e+00 -9.78008449e-01 1.03199720e+00 3.72938782e-01
4.48983580e-01 -1.70221731e-01 -7.00990140e-01 1.12355486e-01
4.97263044e-01 -1.07456028e-01 1.03436983e+00 -7.75648579e-02
-7.69509049e-03 -6.25034750e-01 -1.12752855e+00 4.79414910e-01
8.49309325e-01 -2.93536842e-01 -7.67145097e-01 3.95772219e-01
6.58720851e-01 -4.08342570e-01 -5.14426768e-01 3.09103221e-01
5.58462024e-01 -9.06067133e-01 1.09097862e+00 4.23422307e-02
-3.19950998e-01 -5.26657045e-01 -3.44749838e-01 -1.81951857e+00
-5.90665579e-01 -1.22440600e+00 -3.09525102e-01 1.08844578e+00
3.55885148e-01 -7.87942767e-01 3.54243159e-01 -7.95463547e-02
-7.45881736e-01 -3.05534750e-01 -1.40028965e+00 -8.12171936e-01
-3.16186339e-01 -8.01024616e-01 1.27085894e-01 2.49755010e-01
8.28282908e-02 5.11386711e-03 -6.17935300e-01 7.62002170e-01
8.55508029e-01 -3.71785820e-01 1.90351218e-01 -6.76068127e-01
-8.85654211e-01 -4.72237021e-01 -2.81722635e-01 -1.15187991e+00
-1.37143001e-01 -2.73034155e-01 5.80538630e-01 -1.51679587e+00
-7.23542392e-01 1.17154434e-01 -7.17546225e-01 -1.78351086e-02
-4.32571054e-01 8.32671896e-02 -1.22757778e-01 -4.59286660e-01
-1.12189315e-01 6.12844169e-01 9.87283766e-01 -1.95103094e-01
-4.93601501e-01 5.96034944e-01 -4.94648784e-01 8.18187654e-01
4.64857906e-01 -3.37767333e-01 -4.03529286e-01 -4.79651570e-01
-4.17445868e-01 3.91040713e-01 1.59529924e-01 -1.34672809e+00
6.11402035e-01 2.97902077e-01 1.56968161e-01 -4.86450940e-01
6.28994524e-01 -1.02241874e+00 2.44040221e-01 7.73564160e-01
-1.49399668e-01 -5.01796484e-01 3.95959973e-01 7.68879831e-01
-4.14460927e-01 -1.96058497e-01 1.01216316e+00 2.61956096e-01
-3.04441363e-01 -2.08671674e-01 -9.14808333e-01 -6.29830718e-01
4.30932879e-01 -1.09978147e-01 -1.22234421e-02 -6.25953317e-01
-1.12100482e+00 -3.43372226e-01 -6.52536035e-01 6.66878372e-02
8.89563620e-01 -1.13370669e+00 -9.60479140e-01 3.84262532e-01
-4.41067457e-01 -6.02170646e-01 5.20826697e-01 8.08658361e-01
-4.85008359e-02 4.67805982e-01 -1.68003459e-02 -2.27198496e-01
-1.64097953e+00 2.61995494e-01 7.85554647e-01 1.24724522e-01
-3.34329724e-01 1.55560458e+00 -4.07468677e-02 2.15836674e-01
7.21145689e-01 -2.13388443e-01 -3.18066806e-01 -3.43296587e-01
1.10442507e+00 7.61621177e-01 3.59471112e-01 -6.43487692e-01
-9.13009569e-02 3.60529423e-01 3.90812188e-01 -5.03723502e-01
1.30389178e+00 -4.81409729e-01 2.47156862e-02 1.50865927e-01
1.02120352e+00 7.39465237e-01 -9.54429388e-01 -3.99260044e-01
-3.36885810e-01 -7.44794786e-01 7.29697824e-01 -1.23585594e+00
-7.95541108e-01 6.12262487e-01 1.44774556e+00 1.80764422e-01
2.05469251e+00 -4.91292387e-01 6.69606626e-01 3.88991088e-02
2.87575662e-01 -7.59857357e-01 1.11382253e-01 2.89872974e-01
1.31569684e+00 -6.20647848e-01 -4.58456755e-01 -2.38702059e-01
1.84550196e-01 9.93096471e-01 2.18882695e-01 3.59695330e-02
1.15751195e+00 4.84164178e-01 3.35534394e-01 4.78619695e-01
-3.54612440e-01 -4.25484955e-01 5.01220763e-01 9.49177563e-01
4.14392471e-01 6.03241213e-02 -8.17624271e-01 7.15229571e-01
-5.63735902e-01 -2.72026420e-01 -4.98144366e-02 5.28295815e-01
-1.04829109e+00 -8.81979287e-01 -1.06150448e+00 2.95625716e-01
-4.71478224e-01 -5.64920902e-01 3.44735175e-01 -8.95754546e-02
5.77457845e-01 1.81036925e+00 2.47814823e-02 -4.51800644e-01
8.52371454e-01 2.04668105e-01 3.44516009e-01 -2.56708890e-01
-6.76320791e-01 1.01670170e+00 4.12889346e-02 -2.10197732e-01
-2.22271413e-01 -5.75366855e-01 -5.97878933e-01 1.56697631e-01
-7.05348372e-01 1.39540881e-01 8.02434146e-01 7.00567722e-01
1.13110036e-01 1.00031447e+00 6.26942515e-01 -8.26118648e-01
-4.22711045e-01 -1.44212162e+00 -8.44289899e-01 -8.32828060e-02
9.65416849e-01 -3.91193002e-01 -7.71289766e-01 4.24824841e-02]
|
[15.09001636505127, 5.827152729034424]
|
72a417a5-4f64-41e5-83b3-99b3ca31a38f
|
difference-in-differences-with-time-varying
|
2202.02903
| null |
https://arxiv.org/abs/2202.02903v2
|
https://arxiv.org/pdf/2202.02903v2.pdf
|
Difference-in-Differences with Time-Varying Covariates in the Parallel Trends Assumption
|
In this paper, we study difference-in-differences identification and estimation strategies where the parallel trends assumption holds after conditioning on time-varying covariates and/or time-invariant covariates. Our first main contribution is to point out a number of weaknesses of commonly used two-way fixed effects (TWFE) regressions in this context. In addition to issues related to multiple periods and variation in treatment timing that have been emphasized in the literature, we show that, even in the case with only two time periods, TWFE regressions are not generally robust to (i) paths of untreated potential outcomes depending on the level of time-varying covariates (as opposed to only the change in the covariates over time), (ii) paths of untreated potential outcomes depending on time-invariant covariates, and (iii) violations of linearity conditions for outcomes over time and/or the propensity score. Even in cases where none of the previous three issues hold, we show that TWFE regressions can suffer from negative weighting and weight-reversal issues. Thus, TWFE regressions can deliver misleading estimates of causal effect parameters in a number of empirically relevant cases. Second, we extend these arguments to the case of multiple periods and variation in treatment timing. Third, we provide simple diagnostics for assessing the extent of misspecification bias arising due to TWFE regressions. Finally, we propose alternative (and simple) estimation strategies that can circumvent these issues with two-way fixed regressions.
|
['Brantly Callaway', 'Carolina Caetano']
|
2022-02-07
| null | null | null | null |
['econometrics']
|
['miscellaneous']
|
[ 2.20193371e-01 -2.10197866e-01 -9.41444814e-01 -1.52559072e-01
-6.99944675e-01 -5.47317207e-01 7.17251897e-01 3.87903959e-01
-6.78863287e-01 9.98179972e-01 6.12416387e-01 -8.35469842e-01
-8.80907714e-01 -6.47900641e-01 -5.62689543e-01 -5.26912987e-01
-2.66011655e-01 5.38627096e-02 -8.51740986e-02 2.30446398e-01
3.36488038e-01 4.32205737e-01 -1.02195561e+00 -5.23094594e-01
1.05770981e+00 2.46418744e-01 -2.21707255e-01 1.49012864e-01
3.80365103e-01 5.58828831e-01 -2.77070910e-01 -4.24524039e-01
2.78530449e-01 -1.54007494e-01 -4.04404759e-01 7.14924634e-02
2.43979514e-01 -3.25842351e-01 -2.31911600e-01 8.18855762e-01
3.14899921e-01 1.48142666e-01 1.14383721e+00 -1.38407385e+00
-6.30438089e-01 4.28909093e-01 -8.60265791e-01 2.08000079e-01
1.95379317e-01 3.26954216e-01 7.49652743e-01 -4.82386589e-01
5.73960304e-01 1.46464372e+00 7.44836092e-01 1.17606536e-01
-1.41708994e+00 -6.68672919e-01 4.32102054e-01 -2.20192850e-01
-1.01701725e+00 -4.82361168e-01 3.63591254e-01 -9.78124082e-01
5.90599000e-01 3.16905379e-01 1.16588555e-01 9.79127765e-01
5.86830437e-01 1.53027043e-01 1.40800691e+00 -5.73356986e-01
-4.08823565e-02 4.13881801e-02 6.60393476e-01 6.01080991e-02
6.48905635e-01 6.46105886e-01 1.66432455e-01 -6.59041941e-01
1.00443721e+00 5.54570071e-02 -1.31116256e-01 -2.80124545e-01
-1.32156396e+00 1.16665316e+00 1.03597037e-01 3.90841842e-01
-7.28892326e-01 -4.54176851e-02 5.42284787e-01 2.16990933e-01
4.47855800e-01 3.87757748e-01 -6.96015120e-01 2.31855899e-01
-7.79417217e-01 4.69743103e-01 3.88320088e-01 3.57853770e-01
4.29793149e-01 1.52127519e-01 -5.49029768e-01 5.91081619e-01
-7.17650056e-02 7.10950911e-01 5.75252116e-01 -8.49815249e-01
5.44430733e-01 1.36716500e-01 4.75261092e-01 -9.38438296e-01
-6.12225831e-01 -4.87169057e-01 -8.40540767e-01 1.35887548e-01
8.08193028e-01 -6.06949151e-01 -6.79811716e-01 2.21186543e+00
1.11045159e-01 -6.18022420e-02 -1.67238310e-01 5.95207155e-01
3.36022526e-02 1.95353210e-01 5.27694523e-01 -9.65819359e-01
1.55836761e+00 -6.08232439e-01 -7.32582331e-01 -1.89152598e-01
8.56904924e-01 -3.64806592e-01 7.99904048e-01 3.70508283e-02
-1.22518158e+00 -2.92348742e-01 -1.61232412e-01 1.49057120e-01
-1.71789408e-01 -2.32633173e-01 7.73780584e-01 6.20031893e-01
-7.62341917e-01 5.99460840e-01 -6.25914156e-01 -5.10247231e-01
-8.00929964e-02 5.00861108e-01 -1.81371793e-01 1.68139771e-01
-1.07293332e+00 9.11756516e-01 -3.47353876e-01 -1.02145158e-01
-3.15444827e-01 -1.25771475e+00 -6.60934508e-01 4.02368516e-01
5.03543556e-01 -9.32182550e-01 9.45689619e-01 -1.20815158e+00
-7.41725266e-01 3.84490281e-01 -3.43967915e-01 -1.79517165e-01
8.26250255e-01 7.94165283e-02 -4.95403826e-01 -5.04292071e-01
6.74285352e-01 -1.04138985e-01 2.78408706e-01 -7.63191223e-01
-5.43138206e-01 -7.84819484e-01 -1.44816060e-02 8.50230753e-02
-7.53995404e-02 5.92751265e-01 1.50149927e-01 -8.37582052e-01
-2.87707955e-01 -8.07019770e-01 -5.64818501e-01 -5.77918291e-01
-3.28889519e-01 -1.68062374e-01 2.12622173e-02 -4.84575063e-01
1.59004509e+00 -2.11759686e+00 -2.92151310e-02 1.60029083e-01
-1.50851995e-01 -2.59392083e-01 -4.81752940e-02 5.98226845e-01
-8.34682345e-01 2.70377874e-01 -2.45518863e-01 -7.75297731e-02
6.77216351e-02 -1.36410803e-01 -2.71629304e-01 7.56452501e-01
3.47212218e-02 7.88475335e-01 -6.63900256e-01 -1.40790761e-01
2.35314161e-01 1.37408704e-01 -3.77340615e-01 -4.86744553e-01
6.61072612e-01 4.47088242e-01 -2.83138156e-01 2.00148836e-01
6.83756411e-01 6.60316944e-02 2.94822276e-01 2.16849700e-01
-7.22186804e-01 5.90218127e-01 -1.27578306e+00 6.95438385e-01
-3.26930970e-01 3.73973429e-01 -1.34513438e-01 -1.42443407e+00
1.14299744e-01 5.26068270e-01 5.22234261e-01 -6.69237912e-01
-3.98358740e-02 3.76730561e-01 3.28594416e-01 -4.02346760e-01
1.78468287e-01 -7.51571953e-01 -2.96656877e-01 2.74763465e-01
-2.17214763e-01 3.57516408e-01 1.72921970e-01 -2.11165771e-01
1.09777498e+00 -4.29686338e-01 5.50780535e-01 -6.25490904e-01
1.66322410e-01 -9.01631564e-02 1.07160103e+00 9.22499537e-01
8.68386626e-02 2.47517690e-01 8.54557037e-01 4.27605920e-02
-9.73065495e-01 -9.22342956e-01 -6.98859274e-01 6.45982862e-01
-4.59693104e-01 3.13236177e-01 -8.51597637e-02 -4.10714269e-01
5.40765047e-01 9.20429528e-01 -9.56157565e-01 -4.52469476e-03
-3.56582910e-01 -1.31540334e+00 2.96356857e-01 7.11709142e-01
1.50012807e-03 -4.78849083e-01 -5.25627673e-01 3.12544286e-01
3.07512105e-01 -5.09579659e-01 -3.06138068e-01 -9.33171734e-02
-1.16131544e+00 -1.15397584e+00 -8.66088212e-01 -1.24171644e-01
5.14688075e-01 4.71273094e-01 8.34007740e-01 -1.03507012e-01
3.83211255e-01 4.86361831e-01 2.06646230e-02 -5.99677205e-01
-9.40199792e-02 -2.32059509e-01 6.34350106e-02 -2.32685193e-01
4.88078743e-01 -4.58395123e-01 -6.07578576e-01 -5.13905212e-02
-9.08749580e-01 -2.46162966e-01 2.00371534e-01 8.94263148e-01
-6.22426793e-02 1.16296085e-02 1.06726217e+00 -1.19124568e+00
5.40632904e-01 -7.52005279e-01 -8.58459890e-01 3.02257180e-01
-8.17410111e-01 -3.19188923e-01 2.57380724e-01 -8.43860328e-01
-1.05907512e+00 -6.17842555e-01 4.16914344e-01 -3.19850631e-02
-1.51558053e-02 8.72856140e-01 2.57066861e-02 3.93051684e-01
3.85640830e-01 -4.73102421e-01 7.12261945e-02 -4.25361693e-01
-1.72229901e-01 3.72357488e-01 7.51321316e-02 -4.31787342e-01
9.08049822e-01 4.05886054e-01 1.39927998e-01 -2.95316100e-01
-4.40087259e-01 -3.37597340e-01 -4.42907393e-01 2.90932238e-01
9.49662745e-01 -1.24360788e+00 -4.87732828e-01 4.05561507e-01
-6.04724586e-01 -5.31479537e-01 -3.48115772e-01 1.21750259e+00
-5.14998674e-01 2.65567899e-01 -6.47929549e-01 -1.13651967e+00
1.84569687e-01 -1.20146894e+00 4.68131602e-01 9.52297002e-02
-4.38192397e-01 -1.52263033e+00 -6.06933385e-02 -2.55651474e-02
2.67818362e-01 4.65226859e-01 1.18239915e+00 -4.38277692e-01
-1.92154482e-01 -2.55029887e-01 -8.14109296e-02 -2.79748976e-01
5.00947714e-01 3.53504449e-01 -4.44066703e-01 -4.14170206e-01
1.51289269e-01 3.40596497e-01 5.78043044e-01 1.33261693e+00
8.56767356e-01 -5.50392926e-01 -5.50179422e-01 2.73719996e-01
1.44295537e+00 7.46716261e-01 4.50191617e-01 4.73376393e-01
3.86304289e-01 1.12852144e+00 5.03989339e-01 5.07950068e-01
4.47383255e-01 8.33428502e-01 -1.46391928e-01 -5.83417237e-01
5.98644018e-01 -9.62148756e-02 2.51706392e-01 1.03594288e-01
-7.14948997e-02 -1.09615728e-01 -6.88113809e-01 1.10065472e+00
-1.74798083e+00 -1.11841917e+00 -9.50903475e-01 2.83334303e+00
6.42846823e-01 -7.83173814e-02 7.45535612e-01 6.99882582e-02
7.58337617e-01 -2.89769992e-02 -2.36303970e-01 -6.89521492e-01
-1.77233279e-01 4.66915444e-02 1.01299202e+00 6.56150103e-01
-7.77407765e-01 3.04348826e-01 6.89384079e+00 4.09702390e-01
-9.42802846e-01 2.10500956e-01 7.23430991e-01 -3.19461226e-02
-5.42590976e-01 5.19287646e-01 -4.45936739e-01 4.75148082e-01
9.10322428e-01 -5.78522623e-01 -1.01017013e-01 1.33170068e-01
6.48736954e-01 -3.92415583e-01 -1.05998707e+00 1.90836713e-02
-3.96977544e-01 -6.08254731e-01 -5.99138558e-01 6.15364075e-01
9.69232142e-01 -4.70680982e-01 3.97100776e-01 3.95309448e-01
5.84212422e-01 -8.99349868e-01 7.06046939e-01 1.00308090e-01
8.82306814e-01 -7.21678853e-01 6.44916952e-01 3.77493262e-01
-5.52971900e-01 -5.44537306e-01 -1.42652348e-01 -6.05619192e-01
1.45544797e-01 9.52923238e-01 -1.05930716e-01 7.44348586e-01
2.78466344e-01 4.07707572e-01 -1.11248463e-01 1.09719765e+00
5.28233778e-03 8.21160614e-01 -1.22451462e-01 6.47822797e-01
3.51550579e-01 -3.56906891e-01 3.70044231e-01 8.89318466e-01
5.62373042e-01 4.22796607e-01 -5.26015818e-01 7.17362702e-01
3.91358167e-01 2.61084214e-02 -8.25960159e-01 1.14760846e-01
3.41005534e-01 6.82104111e-01 -4.97932315e-01 -2.82266259e-01
-1.11846304e+00 3.36765856e-01 9.63831414e-03 6.83487177e-01
-5.89749634e-01 1.51849315e-01 6.34008706e-01 5.40066838e-01
-1.85501143e-01 -5.12540340e-02 -3.59179199e-01 -1.39029551e+00
-1.67624783e-02 -8.21023285e-01 8.30859900e-01 -2.69871175e-01
-1.34878814e+00 -5.28088093e-01 6.02417946e-01 -7.47659564e-01
-2.20785260e-01 -2.90782660e-01 -6.53700769e-01 1.17677510e+00
-1.43083763e+00 -7.45575190e-01 5.53626359e-01 3.74367774e-01
3.94073546e-01 4.33802575e-01 6.17177606e-01 3.58126521e-01
-9.10557091e-01 3.69444460e-01 4.72742617e-01 -1.78280219e-01
9.98269379e-01 -1.12329459e+00 6.12716824e-02 7.90869832e-01
-4.83908981e-01 1.02784073e+00 7.31796324e-01 -9.20712173e-01
-7.81552076e-01 -7.41066992e-01 1.32652068e+00 -3.80177945e-01
8.70491862e-01 -1.80818871e-01 -9.15583730e-01 1.29876077e+00
-1.03780814e-01 -5.03943682e-01 5.66312432e-01 7.08466113e-01
-1.54357925e-01 3.55048105e-02 -1.08529282e+00 7.12392628e-01
7.57608354e-01 -2.35906705e-01 -4.73239481e-01 4.18530673e-01
4.41066802e-01 1.43200224e-02 -1.11018419e+00 7.17302918e-01
7.27881312e-01 -9.19529617e-01 8.84865940e-01 -9.55066383e-01
4.42262203e-01 2.92133689e-01 1.33820236e-01 -1.18683636e+00
-8.69632065e-01 -2.51265049e-01 8.97894859e-01 1.48311722e+00
4.12561804e-01 -1.11940765e+00 8.71556252e-02 1.12638819e+00
1.49339616e-01 -4.47201669e-01 -9.81364787e-01 -8.72866035e-01
9.18320775e-01 -1.37428254e-01 7.10697234e-01 1.51466918e+00
3.69031429e-02 1.62477508e-01 -3.84149045e-01 2.52002534e-02
2.91297972e-01 9.55857411e-02 5.45034170e-01 -1.46846569e+00
-3.27066422e-01 -4.15003419e-01 7.38668144e-02 -3.48794699e-01
3.62972379e-01 -8.38086233e-02 -4.57670212e-01 -1.33169174e+00
7.28926003e-01 -6.53019130e-01 -3.15709829e-01 2.60266125e-01
-5.88125110e-01 -3.45222205e-01 1.32831737e-01 1.25671148e-01
3.10075760e-01 1.24182276e-01 7.74368048e-01 2.16789752e-01
-3.52642208e-01 3.89079899e-01 -9.01309967e-01 6.88554347e-01
5.83626211e-01 -5.95593631e-01 -3.47106665e-01 -2.21337751e-01
8.26862007e-02 7.64231980e-01 7.56374359e-01 -3.06960642e-01
-2.22501174e-01 -1.01714742e+00 2.00167760e-01 -2.60843784e-01
-8.98978338e-02 -9.33729589e-01 7.18504488e-01 6.45909786e-01
-3.07866782e-01 4.31456208e-01 3.46608192e-01 3.42306018e-01
1.31538138e-01 -2.94322968e-01 3.67491871e-01 1.00258905e-02
1.22801274e-01 -1.42569989e-01 -8.05626214e-01 -1.03709459e-01
1.06588602e+00 -2.05525190e-01 -3.94233823e-01 -3.80830437e-01
-5.49548507e-01 3.35053146e-01 7.28538990e-01 1.55082673e-01
-2.68574476e-01 -1.29178870e+00 -7.86626041e-01 -4.84699979e-02
5.96473366e-03 -7.54511833e-01 4.73257810e-01 1.43811548e+00
3.59337956e-01 8.32134128e-01 6.38510520e-03 1.17865190e-01
-1.29642439e+00 1.10171175e+00 2.75194108e-01 -4.26780969e-01
-2.91159868e-01 2.21712172e-01 1.10160577e+00 -1.19831324e-01
-2.07073003e-01 -1.67898759e-01 -5.89400679e-02 3.77077162e-01
2.18882143e-01 6.90135062e-01 -1.68126956e-01 -6.98858082e-01
-3.88143659e-01 6.26203895e-01 2.40679324e-01 -3.80472749e-01
1.26489878e+00 -3.98531914e-01 -7.15462118e-02 8.38811994e-01
8.58921468e-01 3.96561265e-01 -1.02147806e+00 -1.46071061e-01
6.99807629e-02 -3.63352627e-01 -1.35575533e-01 -5.46909988e-01
-8.38081479e-01 5.36154926e-01 3.57239068e-01 1.65582567e-01
1.27837706e+00 -5.39481163e-01 -2.97447518e-02 -5.68703115e-01
1.75183713e-01 -7.43785083e-01 -6.56164885e-01 -1.12611875e-01
5.71325660e-01 -1.09890938e+00 2.77688771e-01 -3.44088972e-01
-4.24086034e-01 5.38303912e-01 1.14619523e-01 -1.95957959e-01
4.86513734e-01 -1.09972000e-01 -2.42389575e-01 1.27160810e-02
-8.82930338e-01 -1.73898891e-01 3.47573757e-01 4.74355161e-01
6.78042710e-01 2.26460859e-01 -1.54968488e+00 4.79406476e-01
-5.16380630e-02 -2.56705403e-01 7.72932947e-01 6.12982452e-01
2.48495132e-01 -1.12487423e+00 -7.14345813e-01 6.55797243e-01
-1.20790589e+00 -6.72121197e-02 -8.31658840e-02 1.47472262e+00
-4.17150930e-02 1.03906512e+00 5.07177651e-01 4.02648360e-01
5.89552045e-01 2.45877028e-01 1.35903180e-01 -5.63357472e-01
-8.85474026e-01 5.09273112e-01 1.42089292e-01 -1.15165979e-01
-7.51693308e-01 -1.25628257e+00 -5.86179495e-01 -5.81939042e-01
-7.72605658e-01 1.66498497e-01 6.33706227e-02 9.77932692e-01
8.52972940e-02 4.46353555e-01 7.39762843e-01 -4.03857231e-01
-7.58817434e-01 -9.96491194e-01 -8.67009342e-01 3.67728531e-01
5.27906656e-01 -9.60462451e-01 -7.46180773e-01 -3.68348807e-01]
|
[7.916110038757324, 5.205323219299316]
|
32068467-aa13-464b-a31e-c6ee1a98edfb
|
automatic-instrument-recognition-in
|
1511.05520
| null |
http://arxiv.org/abs/1511.05520v1
|
http://arxiv.org/pdf/1511.05520v1.pdf
|
Automatic Instrument Recognition in Polyphonic Music Using Convolutional Neural Networks
|
Traditional methods to tackle many music information retrieval tasks
typically follow a two-step architecture: feature engineering followed by a
simple learning algorithm. In these "shallow" architectures, feature
engineering and learning are typically disjoint and unrelated. Additionally,
feature engineering is difficult, and typically depends on extensive domain
expertise.
In this paper, we present an application of convolutional neural networks for
the task of automatic musical instrument identification. In this model, feature
extraction and learning algorithms are trained together in an end-to-end
fashion. We show that a convolutional neural network trained on raw audio can
achieve performance surpassing traditional methods that rely on hand-crafted
features.
|
['Peter Li', 'Tian Wang', 'Jiyuan Qian']
|
2015-11-17
| null | null | null | null |
['instrument-recognition']
|
['audio']
|
[ 2.95146137e-01 -3.77058536e-01 2.39452615e-01 -3.85203123e-01
-9.94199455e-01 -9.59192753e-01 4.18051392e-01 1.37520730e-01
-5.07614493e-01 2.29955927e-01 -9.27632600e-02 -1.37889087e-01
-4.49831277e-01 -4.52745229e-01 -6.34365797e-01 -1.42058626e-01
-1.73821002e-01 3.16006035e-01 -2.03337669e-01 -3.59993666e-01
3.41839522e-01 3.22491348e-01 -1.62362230e+00 3.30746800e-01
2.07228988e-01 1.27245307e+00 -1.17584504e-01 1.02911437e+00
-1.79275021e-01 7.28538752e-01 -5.51454008e-01 -2.86986500e-01
4.22834039e-01 -4.15390700e-01 -1.05100119e+00 -1.48480073e-01
5.44710636e-01 -1.79414570e-01 -2.36940250e-01 7.08370328e-01
6.27196431e-01 3.31738710e-01 6.72159612e-01 -1.04012156e+00
-2.56408066e-01 8.85217190e-01 -2.31842712e-01 -8.84294435e-02
4.46265310e-01 -8.40112939e-02 1.54365945e+00 -1.00087130e+00
1.65326625e-01 8.06311250e-01 1.23285306e+00 4.27152514e-01
-1.40625262e+00 -7.70444930e-01 -9.96840075e-02 1.04646899e-01
-1.54780602e+00 -5.37276745e-01 9.89503801e-01 -6.71679258e-01
8.60020220e-01 7.32964426e-02 7.50370026e-01 7.56971419e-01
-3.22585374e-01 8.53411138e-01 5.39386213e-01 -4.92161691e-01
1.11262068e-01 -2.41573974e-01 2.00236455e-01 5.96388996e-01
-8.61069784e-02 1.87376305e-01 -8.45364511e-01 -2.01168835e-01
7.30081081e-01 -1.08937822e-01 -1.43605441e-01 -4.88967270e-01
-1.06817794e+00 7.94286251e-01 3.08740675e-01 3.56965721e-01
-1.98802352e-01 4.93928164e-01 7.47702837e-01 8.29202712e-01
7.63398735e-03 9.98167992e-01 -6.49513304e-01 -3.23379129e-01
-1.35848463e+00 4.11209345e-01 9.97457623e-01 7.45272338e-01
7.00985491e-01 2.23643575e-02 1.00164481e-01 8.31571102e-01
1.01655260e-01 -3.29640545e-02 5.64505637e-01 -8.21906686e-01
7.38882869e-02 3.23447585e-01 3.44669558e-02 -7.59394050e-01
-4.90256339e-01 -7.77393222e-01 -6.71838403e-01 3.94823551e-01
6.44615233e-01 -3.40759873e-01 -7.21197367e-01 1.63123953e+00
-9.91344675e-02 1.27251640e-01 -6.85531721e-02 8.11255753e-01
7.97138453e-01 1.40923887e-01 -2.42097601e-01 2.14513838e-01
1.10627127e+00 -1.01642036e+00 -3.72262746e-01 -5.35723455e-02
3.56228888e-01 -1.01803899e+00 1.06237495e+00 8.78906488e-01
-1.16696262e+00 -7.40649164e-01 -1.27798998e+00 -2.08129674e-01
-2.15383753e-01 2.02556670e-01 6.01944447e-01 4.50661987e-01
-8.85386348e-01 1.20625603e+00 -4.23444539e-01 -4.41790223e-02
4.98487979e-01 7.08480477e-01 -3.36120486e-01 2.95266628e-01
-8.56429040e-01 4.12481666e-01 3.80111694e-01 1.97910219e-01
-9.07691061e-01 -6.92752182e-01 -6.31906807e-01 2.58087963e-01
4.92140561e-01 -8.84141803e-01 1.88175678e+00 -1.26986408e+00
-1.74556589e+00 5.81909180e-01 2.27481306e-01 -4.24124092e-01
2.76105911e-01 -8.50216031e-01 -2.86587141e-02 6.09382987e-03
-1.35237366e-01 1.19489543e-01 1.31282246e+00 -8.87688339e-01
-5.32956719e-01 -1.16316825e-01 4.25453484e-02 -2.13969704e-02
-4.72460926e-01 2.89029956e-01 -3.93989235e-01 -8.97680402e-01
-1.09238168e-02 -8.98618281e-01 -2.80588746e-01 -8.30601677e-02
-3.14338267e-01 -2.78380901e-01 5.04012108e-01 -3.40601265e-01
1.15438688e+00 -2.37048626e+00 2.03007117e-01 3.95285457e-01
2.54356831e-01 2.46169090e-01 -2.16153443e-01 3.94196153e-01
-2.05189168e-01 -8.65532458e-03 -1.01688311e-01 -9.69615951e-02
4.27136183e-01 -3.57061327e-01 -3.46409440e-01 1.23997375e-01
2.17809781e-01 9.19642568e-01 -9.55727816e-01 -1.69696882e-01
-1.11597115e-02 3.21455508e-01 -7.11629033e-01 2.95740485e-01
-2.80772269e-01 1.98138863e-01 -2.02585295e-01 6.26602709e-01
-2.40717921e-02 -2.23643154e-01 1.52338192e-01 -2.57910997e-01
-7.83381462e-02 4.52598184e-01 -1.30498755e+00 2.17905068e+00
-6.61710680e-01 8.78566265e-01 8.41351599e-02 -9.62623894e-01
8.47789526e-01 5.79520583e-01 5.47966361e-01 -5.73588274e-02
4.27840829e-01 4.49269503e-01 1.52044460e-01 -2.19125792e-01
4.35958266e-01 -2.22903416e-01 -3.91082585e-01 7.69207180e-01
5.22641242e-01 -3.44298393e-01 1.20603241e-01 -1.09492101e-01
1.27102792e+00 3.18048596e-01 3.22922349e-01 6.90983608e-02
4.50537354e-01 -6.37424365e-02 5.05294681e-01 9.88001704e-01
2.51178052e-02 7.36110270e-01 2.82955974e-01 -7.10831046e-01
-9.72890675e-01 -8.57357085e-01 5.58534451e-02 1.50377190e+00
-4.61025327e-01 -9.54970360e-01 -5.98934531e-01 -7.11727798e-01
9.92660522e-02 5.99497445e-02 -5.25214136e-01 -1.35495275e-01
-5.88196278e-01 -6.41796067e-02 8.22954714e-01 6.94090545e-01
1.52644590e-01 -1.28063571e+00 -5.82356989e-01 6.73187077e-01
3.12867582e-01 -7.70233274e-01 -3.68349731e-01 6.42169952e-01
-7.12667763e-01 -1.14419055e+00 -5.86938679e-01 -9.85875249e-01
3.34714018e-02 -4.98048104e-02 1.43505120e+00 2.59864181e-01
-5.23733139e-01 2.94161111e-01 -5.31898439e-01 -7.76453853e-01
6.99808523e-02 6.38261378e-01 2.98286863e-02 2.88078189e-02
4.66952622e-01 -8.84808540e-01 -4.46874559e-01 2.81129275e-02
-7.84720957e-01 -3.66971135e-01 8.02597404e-01 1.13584518e+00
3.85684580e-01 -3.31525132e-02 5.03934562e-01 -6.77052617e-01
7.62252390e-01 -9.87903103e-02 -3.02840948e-01 1.92537904e-01
-3.69159639e-01 1.78532138e-01 6.48943841e-01 -6.47142589e-01
-3.31352830e-01 7.45729148e-01 -2.81675905e-01 -6.59083605e-01
-3.19291860e-01 6.44374132e-01 5.24716936e-02 -2.24202096e-01
7.86058128e-01 1.46719173e-01 -3.40083838e-02 -8.88522863e-01
4.21882331e-01 7.16936827e-01 7.49911249e-01 -6.37688100e-01
9.50541437e-01 2.12791450e-02 -1.62483186e-01 -7.15765297e-01
-1.10945654e+00 -3.43698740e-01 -8.85044992e-01 -1.70289367e-01
5.41488349e-01 -7.11022437e-01 -8.39254379e-01 4.33421016e-01
-9.06691730e-01 -3.13664854e-01 -6.19604349e-01 6.57272279e-01
-8.74023855e-01 1.40278175e-01 -4.27328229e-01 -5.99949002e-01
-5.47843397e-01 -7.21833229e-01 9.95219231e-01 7.93395638e-02
-6.06683493e-01 -6.41618788e-01 2.24605381e-01 5.47966473e-02
4.58586782e-01 -6.66318834e-02 8.50654960e-01 -9.62527156e-01
-2.34434769e-01 -4.32019770e-01 2.01574787e-02 3.29697967e-01
2.02542573e-01 1.17006358e-02 -1.24520063e+00 -1.59703612e-01
-2.16161743e-01 -7.29976296e-01 9.79047000e-01 1.34292096e-01
1.39729536e+00 4.43578884e-02 6.39137551e-02 8.13401520e-01
1.20893610e+00 -1.56390622e-01 1.83854431e-01 4.25229728e-01
7.47491181e-01 4.47073311e-01 3.05219352e-01 5.67952096e-01
-6.19857721e-02 7.05807328e-01 2.40838438e-01 -2.33526379e-02
4.45677191e-02 -1.04206681e-01 1.68686494e-01 6.73189223e-01
-1.55499727e-01 3.06825459e-01 -9.09131169e-01 6.00658059e-01
-1.97615457e+00 -1.04111803e+00 1.88727111e-01 2.07441616e+00
9.52780068e-01 2.23571479e-01 4.55922663e-01 6.59978390e-01
3.58176857e-01 -1.26933649e-01 -5.39367199e-01 -3.90304327e-01
1.73194885e-01 8.35379064e-01 -7.18345717e-02 1.03206225e-01
-1.47513282e+00 8.77762973e-01 7.38015842e+00 6.62261546e-01
-1.04757476e+00 -2.04207748e-01 -1.47170424e-01 -5.02032816e-01
1.06212711e-02 -6.34141117e-02 -2.85120070e-01 1.99865978e-02
8.20786119e-01 2.89818756e-02 5.87861300e-01 7.94561207e-01
-3.43635738e-01 4.79286343e-01 -1.48381448e+00 1.31187379e+00
-1.59429416e-01 -1.33445656e+00 -2.04594731e-02 -1.52922228e-01
2.56469429e-01 -1.21455882e-02 5.70770092e-02 4.36373770e-01
4.67588574e-01 -1.28770387e+00 8.26122344e-01 6.56150937e-01
6.36765778e-01 -9.21976686e-01 6.50164485e-01 5.68620525e-02
-1.42316270e+00 -3.42630953e-01 -5.31043559e-02 -2.78432816e-01
3.73507664e-02 4.00079846e-01 -5.15372992e-01 5.12013078e-01
7.79006600e-01 8.05854321e-01 -3.84194642e-01 1.33786380e+00
-2.03235224e-01 5.60482800e-01 -2.13982463e-01 4.20901030e-02
2.41927668e-01 1.40030071e-01 3.60298842e-01 1.27427387e+00
2.61939913e-01 -1.62029609e-01 2.63749599e-01 6.36398315e-01
-2.07285345e-01 2.02907249e-01 -6.22565448e-01 -3.88625056e-01
2.80633420e-01 1.31803417e+00 -4.41677392e-01 -2.75484324e-01
-4.50619280e-01 8.03784072e-01 4.97327149e-01 1.33309945e-01
-3.86183411e-01 -1.02557409e+00 6.99323535e-01 -2.92080250e-02
7.40905523e-01 -2.59219170e-01 -3.41731310e-01 -1.01831746e+00
-6.76800236e-02 -1.05962896e+00 4.12570119e-01 -4.04329985e-01
-1.38474977e+00 7.19301164e-01 -5.92181623e-01 -1.29246604e+00
-7.01189578e-01 -6.31892383e-01 -8.19691241e-01 7.03338861e-01
-1.28156912e+00 -9.32530522e-01 -1.34199277e-01 8.31699610e-01
4.65980291e-01 -5.48211575e-01 1.20226824e+00 3.79824460e-01
-3.81529808e-01 8.68179977e-01 4.88660112e-02 5.76627791e-01
6.94311261e-01 -1.43629420e+00 4.09203440e-01 3.53296638e-01
8.39733422e-01 6.03128970e-01 5.95456779e-01 -9.62034091e-02
-1.42109191e+00 -8.45026076e-01 8.22141945e-01 -2.17495725e-01
7.43339002e-01 -3.45721483e-01 -6.85097277e-01 5.82200289e-01
2.47182518e-01 4.27379087e-02 1.17558849e+00 7.97325075e-01
-6.63276374e-01 -1.61385462e-01 -6.66395605e-01 2.66295105e-01
9.03365970e-01 -9.95725691e-01 -7.14582145e-01 6.72009811e-02
2.89178222e-01 -8.83380622e-02 -8.86406779e-01 2.92903394e-01
1.14085686e+00 -6.77098393e-01 1.06757069e+00 -9.84949708e-01
2.77903825e-01 -3.33425730e-01 -2.02948004e-01 -1.16037500e+00
-4.44382221e-01 -1.07330370e+00 -2.32934445e-01 9.05864239e-01
4.73704964e-01 2.18517825e-01 7.28883207e-01 3.09052050e-01
-4.61970456e-02 -5.57457030e-01 -4.84254450e-01 -8.99311006e-01
7.43555184e-03 -6.98789239e-01 7.00584710e-01 8.90459955e-01
7.83977583e-02 7.69556999e-01 -2.78931528e-01 -1.61998034e-01
5.09930074e-01 4.52774554e-01 9.72938716e-01 -1.91176403e+00
-6.54247403e-01 -8.09832513e-01 -5.48902094e-01 -7.89883256e-01
1.38898462e-01 -8.53527546e-01 2.79696643e-01 -1.01322639e+00
-4.59679030e-03 -4.42867786e-01 -7.49987721e-01 6.66106761e-01
3.87728363e-02 6.78320587e-01 1.55567899e-01 1.18695177e-01
-6.18463933e-01 2.89520919e-01 5.59352160e-01 -2.71028757e-01
-4.19076174e-01 1.63370088e-01 -7.50055730e-01 1.00118291e+00
1.06195593e+00 -5.68153501e-01 -2.55605340e-01 -4.24023956e-01
5.39032340e-01 -1.80670395e-01 4.24434900e-01 -1.28873348e+00
3.88161510e-01 2.18586057e-01 4.45749938e-01 -4.74364847e-01
4.44691718e-01 -7.48327255e-01 -2.94929203e-02 1.65750384e-01
-5.30560970e-01 -7.79242143e-02 1.19959079e-01 3.61399472e-01
-5.24212360e-01 -5.63760221e-01 2.91824281e-01 -2.84542918e-01
-5.42672992e-01 1.69166446e-01 -3.01731586e-01 -4.80065197e-02
5.48688471e-01 -9.76930261e-02 4.01445687e-01 -3.70763451e-01
-1.14210129e+00 -2.60265619e-01 5.03191240e-02 2.68815905e-01
5.22384703e-01 -1.44767237e+00 -6.96018815e-01 2.85968333e-01
9.71287712e-02 -1.23835452e-01 -2.47270197e-01 5.52555740e-01
-2.40185484e-01 1.12468705e-01 -2.16639429e-01 -3.80907983e-01
-1.33842552e+00 4.74934518e-01 3.60506326e-01 -1.17057696e-01
-6.19751155e-01 1.17394900e+00 -2.56632775e-01 -4.43039179e-01
5.91562271e-01 -9.48595256e-02 -1.60230309e-01 3.64896148e-01
6.42738938e-01 -6.33562133e-02 1.25766218e-01 -3.28453660e-01
-2.64893711e-01 7.99475849e-01 -2.03115568e-01 -2.81989366e-01
1.59526765e+00 3.49616975e-01 1.95880771e-01 6.13837123e-01
9.86910403e-01 -8.48373175e-02 -9.79106903e-01 -5.58303118e-01
3.20859909e-01 -2.65308648e-01 2.61947334e-01 -7.08938479e-01
-9.52845573e-01 9.77154434e-01 3.50253791e-01 2.54234910e-01
1.18612289e+00 -1.64167508e-01 8.09308946e-01 9.23655510e-01
7.13642240e-02 -1.34898853e+00 1.98921800e-01 7.39300430e-01
8.68146479e-01 -1.07607782e+00 -1.19638644e-01 2.83687189e-02
-1.91606849e-01 1.55417252e+00 2.71208882e-01 -4.44074094e-01
9.77898955e-01 4.59804833e-01 1.71939805e-01 -3.37085158e-01
-7.90896952e-01 -5.00953674e-01 7.73521364e-01 3.35632324e-01
7.45485425e-01 -2.83497810e-01 8.77933130e-02 1.12920237e+00
-6.58399284e-01 2.45574251e-01 9.81010720e-02 1.13730228e+00
-4.45414186e-01 -1.35543227e+00 -1.90229550e-01 4.91274893e-01
-6.98918760e-01 -2.47787043e-01 -8.09613347e-01 5.71993172e-01
6.83214068e-02 7.85240412e-01 4.03443053e-02 -8.43767703e-01
3.72180045e-01 4.04400200e-01 6.71193123e-01 -6.68521881e-01
-1.21603763e+00 2.20002040e-01 1.40948281e-01 -5.58140516e-01
-3.67601901e-01 -5.93198478e-01 -9.27366495e-01 -9.35871080e-02
-3.80691230e-01 1.87992319e-01 6.56717181e-01 1.04114699e+00
3.12258005e-01 7.26154804e-01 7.01556146e-01 -1.05734837e+00
-7.02267945e-01 -9.95230973e-01 -6.46444559e-01 2.91400701e-01
5.73479235e-01 -5.43925941e-01 -3.22902203e-02 -1.02954535e-02]
|
[15.782386779785156, 5.238184452056885]
|
c5be3985-bea3-4866-93a6-085833f6334b
|
ultrasonic-image-s-annotation-removal-a-self
|
2307.04133
| null |
https://arxiv.org/abs/2307.04133v1
|
https://arxiv.org/pdf/2307.04133v1.pdf
|
Ultrasonic Image's Annotation Removal: A Self-supervised Noise2Noise Approach
|
Accurately annotated ultrasonic images are vital components of a high-quality medical report. Hospitals often have strict guidelines on the types of annotations that should appear on imaging results. However, manually inspecting these images can be a cumbersome task. While a neural network could potentially automate the process, training such a model typically requires a dataset of paired input and target images, which in turn involves significant human labour. This study introduces an automated approach for detecting annotations in images. This is achieved by treating the annotations as noise, creating a self-supervised pretext task and using a model trained under the Noise2Noise scheme to restore the image to a clean state. We tested a variety of model structures on the denoising task against different types of annotation, including body marker annotation, radial line annotation, etc. Our results demonstrate that most models trained under the Noise2Noise scheme outperformed their counterparts trained with noisy-clean data pairs. The costumed U-Net yielded the most optimal outcome on the body marker annotation dataset, with high scores on segmentation precision and reconstruction similarity. We released our code at https://github.com/GrandArth/UltrasonicImage-N2N-Approach.
|
['Yueyang Teng', 'Junying Cao', 'Zhaoheng Xie', 'Nan Jiang', 'Yuanheng Zhang']
|
2023-07-09
| null | null | null | null |
['denoising']
|
['computer-vision']
|
[ 4.88266438e-01 3.81552398e-01 2.65810311e-01 -6.27527237e-01
-1.44721639e+00 -5.75474679e-01 1.13613367e-01 2.60223389e-01
-6.19747937e-01 4.32265729e-01 4.30729706e-03 -2.54523396e-01
7.23106414e-02 -4.54045832e-01 -7.91122794e-01 -8.65404427e-01
8.76755267e-02 4.08608437e-01 3.73437941e-01 9.28699970e-02
3.07692811e-02 8.35576504e-02 -1.11540949e+00 2.47123018e-01
6.44899368e-01 1.05972540e+00 4.30237263e-01 9.43355322e-01
1.64639503e-01 8.11032653e-01 -6.19492352e-01 -7.04039872e-01
5.35504401e-01 -6.19645298e-01 -8.36322546e-01 1.03555471e-01
4.95231122e-01 -1.75793797e-01 -1.93569884e-02 1.22853398e+00
9.65071142e-01 -6.29703999e-02 4.41046655e-01 -5.65247536e-01
-4.11435843e-01 7.34289289e-01 -4.65869784e-01 4.34262037e-01
4.23651099e-01 1.67813510e-01 6.44429147e-01 -6.70577586e-01
6.65903747e-01 7.59490192e-01 1.25014532e+00 4.21974510e-01
-1.36905491e+00 -4.74799216e-01 -5.64541876e-01 -1.31224200e-01
-1.15766788e+00 -5.88258564e-01 4.42983806e-01 -5.42999625e-01
4.09962058e-01 4.06701326e-01 4.38341767e-01 9.26600099e-01
1.98401064e-01 4.91125703e-01 1.02906358e+00 -5.90788782e-01
5.70304468e-02 5.12583666e-02 9.48324129e-02 7.42644131e-01
2.75010437e-01 -1.89644426e-01 -1.66304439e-01 1.94140840e-02
7.69323468e-01 -2.79297233e-01 -6.29009426e-01 -2.28581294e-01
-1.18175960e+00 4.65405554e-01 2.89737105e-01 5.05958915e-01
-4.69340146e-01 1.78120017e-01 6.26605272e-01 2.25805804e-01
4.85968918e-01 5.89105666e-01 -2.97650546e-01 -1.79720838e-02
-1.12087607e+00 -1.11895680e-01 7.97858894e-01 6.66091204e-01
4.47502971e-01 -1.86210468e-01 -2.23449752e-01 9.59292173e-01
1.17811516e-01 2.57068127e-01 6.08473003e-01 -1.22933173e+00
1.74566165e-01 1.14278100e-01 -1.63076483e-02 -1.15036488e+00
-4.97762114e-01 -6.20589435e-01 -9.15098608e-01 1.87414438e-01
7.26619482e-01 -3.60470414e-01 -1.15990674e+00 1.26980591e+00
3.36750627e-01 1.42090604e-01 -8.48986432e-02 1.05350471e+00
1.03778195e+00 1.48026675e-01 8.46625566e-02 -1.98588133e-01
1.35442686e+00 -1.00232375e+00 -1.05250585e+00 -1.30025223e-01
7.91296363e-01 -1.09447539e+00 7.84235060e-01 5.12610435e-01
-1.33179343e+00 -6.64414465e-01 -8.75624597e-01 1.28214121e-01
5.17426245e-02 2.62444288e-01 7.85384998e-02 9.03785765e-01
-1.19297981e+00 1.04009199e+00 -1.13443208e+00 -2.07451493e-01
4.70246285e-01 4.47275400e-01 -6.38709068e-01 -1.79979935e-01
-8.03429365e-01 1.09916651e+00 4.58116204e-01 5.07066131e-01
-7.32751846e-01 -5.05723953e-01 -9.50725853e-01 -2.02205747e-01
5.92455149e-01 -4.83812451e-01 1.60288656e+00 -9.28723633e-01
-1.06245697e+00 1.13716698e+00 7.14448094e-02 -4.98101324e-01
1.02108514e+00 -2.64656454e-01 -1.63724601e-01 3.29263061e-01
3.19711596e-01 4.45699424e-01 7.84207344e-01 -1.46397316e+00
-3.48346114e-01 -1.07474223e-01 -3.99511307e-01 -5.15452325e-02
2.50585616e-01 1.26685485e-01 -7.56661534e-01 -7.77284265e-01
5.66233337e-01 -8.82813513e-01 -4.09241349e-01 -4.67323996e-02
-4.23819572e-01 1.57110542e-01 4.28991348e-01 -1.25313282e+00
1.08398664e+00 -2.12747788e+00 -1.61868691e-01 4.45544124e-01
2.34586060e-01 1.58472806e-01 1.40809774e-01 -1.14383191e-01
-3.22570503e-01 2.97582507e-01 -4.71572548e-01 -5.85652769e-01
-2.82575995e-01 3.61590922e-01 3.41992438e-01 6.00553989e-01
-9.39121693e-02 7.83815861e-01 -1.04638708e+00 -8.10826719e-01
1.60234943e-01 4.75127280e-01 -4.54795659e-01 3.58265966e-01
2.24298701e-01 6.98487759e-01 -6.45441860e-02 7.00474799e-01
4.81091470e-01 -2.58558780e-01 3.17490131e-01 -6.09171689e-01
2.13225439e-01 -1.00638621e-01 -1.30426037e+00 1.81389701e+00
-3.23275030e-01 5.04433930e-01 4.68629181e-01 -1.12757051e+00
6.82112336e-01 7.75742054e-01 6.62264764e-01 -4.93893385e-01
3.54928672e-01 6.30043745e-01 2.22889453e-01 -1.02924550e+00
2.80572057e-01 -2.20879301e-01 1.93961903e-01 2.78322101e-01
2.60652155e-01 -3.27039242e-01 3.44972402e-01 -4.90619577e-02
1.26529741e+00 2.23698646e-01 2.49590024e-01 -1.85602859e-01
2.60805368e-01 -7.73814172e-02 4.94542837e-01 1.07872880e+00
-4.40183759e-01 1.33261836e+00 4.60941732e-01 -4.98658866e-01
-9.89354789e-01 -6.61348462e-01 -2.47636542e-01 7.00959027e-01
-1.41599178e-01 -1.12457119e-01 -1.08071387e+00 -8.14271271e-01
-5.21773279e-01 4.91650969e-01 -7.76075602e-01 1.29025042e-01
-6.11772001e-01 -7.90683866e-01 6.69062614e-01 4.75416213e-01
3.38674486e-01 -1.01515770e+00 -8.17383826e-01 1.89623088e-01
-4.91811275e-01 -9.88289356e-01 -5.05072534e-01 5.91997147e-01
-7.34245121e-01 -1.21644390e+00 -9.26976979e-01 -8.57525766e-01
9.52868462e-01 -1.68565035e-01 1.30213058e+00 3.93668294e-01
-3.85196388e-01 4.03048515e-01 -4.54758137e-01 -4.24816549e-01
-7.49799132e-01 -1.63452715e-01 -2.15073645e-01 -1.57762766e-01
-7.05876797e-02 -3.51412535e-01 -8.29213023e-01 3.14731091e-01
-1.12100363e+00 -2.01592326e-01 6.45686507e-01 1.01971710e+00
7.64538288e-01 -1.00151472e-01 2.67203897e-01 -1.02352059e+00
6.29451811e-01 -3.15767467e-01 -2.79774725e-01 1.81490391e-01
-2.63283014e-01 -9.08447951e-02 1.78135037e-01 -2.50341445e-01
-9.02898312e-01 5.44240296e-01 -6.34372115e-01 -4.51560497e-01
-3.96927208e-01 4.66247112e-01 1.05799675e-01 -1.57000825e-01
9.04917181e-01 -3.31557505e-02 1.42239660e-01 -4.27586943e-01
5.25275432e-02 5.19386530e-01 1.02231300e+00 -3.99806768e-01
5.33513725e-01 3.10559899e-01 -2.02127576e-01 -6.25105083e-01
-9.56147015e-01 -6.20553195e-01 -9.33238745e-01 -4.06061739e-01
1.20204508e+00 -5.10691464e-01 -1.52570650e-01 3.49424660e-01
-1.21524131e+00 -4.17264581e-01 -2.86475509e-01 4.80883121e-01
-4.25719976e-01 6.20197117e-01 -5.92626393e-01 -5.14031291e-01
-3.16321313e-01 -1.40876484e+00 1.00955188e+00 4.08914983e-02
-4.50907320e-01 -8.71263623e-01 -7.03351051e-02 6.19493842e-01
3.44682038e-01 4.86237645e-01 5.11750042e-01 -8.82661283e-01
-2.10695833e-01 -5.34753323e-01 -8.25048760e-02 6.82461560e-01
2.56045997e-01 -2.67127305e-01 -1.07878530e+00 2.77872607e-02
3.81162524e-01 -3.18285346e-01 7.29859173e-01 7.03392088e-01
1.29306448e+00 -1.35793135e-01 -8.07870105e-02 5.67163348e-01
1.25251484e+00 2.37775981e-01 6.58186555e-01 3.63157868e-01
6.26992583e-01 6.27653480e-01 4.83082861e-01 6.88965339e-03
9.94130969e-03 4.32991236e-01 5.80084205e-01 -4.92205530e-01
-1.70340076e-01 2.95744777e-01 -2.90794998e-01 6.81899667e-01
-3.24585706e-01 -1.98766440e-01 -1.27652764e+00 6.40743732e-01
-1.65865898e+00 -7.50024736e-01 -3.64614218e-01 1.95949244e+00
9.91982698e-01 1.86741948e-01 -1.99989110e-01 3.80636305e-01
6.73866987e-01 -2.19521478e-01 -6.46455139e-02 -4.88176160e-02
1.50628373e-01 1.73686042e-01 5.48018277e-01 3.97033602e-01
-1.43139875e+00 3.69519711e-01 6.36884642e+00 6.87143326e-01
-1.06120586e+00 4.73497480e-01 9.86110508e-01 7.55301565e-02
1.89785168e-01 -5.86640656e-01 -2.11567581e-01 5.85026741e-01
9.67914283e-01 4.92174357e-01 -1.72361419e-01 7.58354366e-01
3.43689919e-01 -5.08215368e-01 -1.00315082e+00 9.50698912e-01
1.77857742e-01 -1.11806095e+00 -4.89538133e-01 -3.21656853e-01
6.76398993e-01 1.00472581e-03 -2.53391445e-01 -1.15606070e-01
8.79293159e-02 -1.04677331e+00 7.38354385e-01 5.89107275e-01
7.17014432e-01 -2.40016907e-01 1.33041835e+00 7.14409724e-02
-7.31163859e-01 2.54009157e-01 -9.33473706e-02 3.93834680e-01
4.00201738e-01 6.15034521e-01 -8.56386364e-01 6.01954162e-01
1.01369417e+00 3.52034271e-01 -5.87421775e-01 1.39836776e+00
-1.99432135e-01 7.86757946e-01 -2.61588693e-01 6.12086833e-01
7.49971718e-02 -8.72006193e-02 3.56548816e-01 1.47271705e+00
4.07836288e-01 1.93333566e-01 2.21454591e-01 4.20308441e-01
-2.58947127e-02 2.53718626e-02 -4.57165182e-01 3.78677517e-01
1.56898238e-02 1.39255655e+00 -1.32708263e+00 -4.83828098e-01
-2.60749847e-01 9.35388088e-01 -1.99930698e-01 1.26006112e-01
-8.50935757e-01 -2.03388661e-01 -1.15885727e-01 2.98426479e-01
1.88081250e-01 1.26262322e-01 -5.35609722e-01 -7.09030390e-01
1.76823229e-01 -9.85079587e-01 4.22831953e-01 -9.20217335e-01
-1.09225154e+00 8.06327164e-01 -1.83507577e-01 -1.24684966e+00
-2.09171638e-01 -3.99665296e-01 -4.60388750e-01 7.11917937e-01
-1.10310423e+00 -8.04627895e-01 -5.99174440e-01 1.98504090e-01
5.01705766e-01 3.75087440e-01 8.07443917e-01 7.59127557e-01
-4.30616587e-01 4.58454221e-01 -5.04650399e-02 4.70041573e-01
8.85274231e-01 -1.41125894e+00 -8.07441946e-04 9.26583529e-01
9.02170539e-02 4.42815930e-01 9.83412445e-01 -6.60638392e-01
-8.43270659e-01 -9.39230323e-01 5.62809587e-01 -4.13247198e-01
4.68304843e-01 1.15242638e-01 -1.13997853e+00 6.96791172e-01
4.66089994e-01 3.08125615e-01 5.97409308e-01 -4.30186689e-01
2.88562238e-01 1.79989755e-01 -1.24899030e+00 2.48621881e-01
8.02076101e-01 -1.93667024e-01 -5.86989343e-01 4.97454375e-01
4.21854049e-01 -1.04212797e+00 -1.13468671e+00 4.19294715e-01
4.58238453e-01 -9.93131816e-01 8.22882533e-01 -1.55877009e-01
5.62782943e-01 -2.74224460e-01 1.17729895e-01 -1.18357253e+00
-1.76688489e-02 -4.24393386e-01 2.76970029e-01 9.03512537e-01
5.68428755e-01 -3.27184558e-01 7.27771282e-01 7.32953429e-01
-4.10895616e-01 -6.95578337e-01 -8.30358088e-01 -4.76640224e-01
-3.22614819e-01 -6.31755114e-01 -9.99169797e-02 8.99427295e-01
-4.47774917e-01 -1.35472476e-01 -2.07191080e-01 2.59729952e-01
5.77081740e-01 -5.01523495e-01 4.62140262e-01 -1.02280474e+00
-2.58406073e-01 -3.55512857e-01 -3.33623260e-01 -4.23259914e-01
-1.71423674e-01 -8.42607677e-01 5.41659653e-01 -1.75724483e+00
-7.30956392e-03 -4.15421933e-01 -1.15007654e-01 6.97199643e-01
-2.77345538e-01 9.21937048e-01 -3.53295356e-02 2.58412927e-01
-5.76633394e-01 -1.42860070e-01 1.21494710e+00 -4.78045866e-02
-5.77239469e-02 2.72540271e-01 -4.34021354e-01 9.90794420e-01
8.40244889e-01 -7.23596871e-01 -7.97253773e-02 -5.70498168e-01
1.35292381e-01 1.49342999e-01 4.67514217e-01 -1.17915344e+00
2.68418223e-01 4.80515748e-01 5.03904998e-01 -4.15732145e-01
2.43908748e-01 -9.86761153e-01 3.13548058e-01 6.18943036e-01
-3.34845781e-01 5.16745746e-02 2.40519047e-01 3.19973707e-01
-3.20881099e-01 -7.73921132e-01 7.94811308e-01 -5.76827049e-01
-3.85906070e-01 -2.08043948e-01 -3.05739522e-01 -4.81411479e-02
8.58371139e-01 -4.60274100e-01 1.19366474e-01 -4.50590789e-01
-1.37431121e+00 4.37901877e-02 3.00355703e-01 8.06963071e-03
4.49813396e-01 -8.32653105e-01 -6.20604336e-01 8.70141461e-02
-3.95585626e-01 4.26715255e-01 3.68346691e-01 1.37187386e+00
-8.61146331e-01 -1.02452472e-01 6.55607972e-03 -9.67648864e-01
-1.51328754e+00 1.44666061e-01 7.65186727e-01 -4.01687413e-01
-6.88862443e-01 9.67586815e-01 -2.58575022e-01 -5.83638191e-01
3.71547848e-01 -4.56642538e-01 -1.82827696e-01 2.10286602e-01
3.31329912e-01 3.09665978e-01 5.59840500e-01 -6.88003242e-01
-5.38941696e-02 5.42131007e-01 4.27398384e-02 -6.77705258e-02
1.35375273e+00 -1.44142658e-01 -1.91945046e-01 1.15766495e-01
9.57831502e-01 -2.09070623e-01 -1.03847313e+00 -1.23563157e-02
2.23538607e-01 -2.87945658e-01 1.21992648e-01 -8.89684141e-01
-1.33257401e+00 6.09731376e-01 8.81877542e-01 4.19620872e-01
1.16339302e+00 -7.97140822e-02 5.13110936e-01 2.02481315e-01
8.10294524e-02 -9.82844770e-01 4.32817005e-02 7.48896673e-02
8.21305096e-01 -1.57822359e+00 4.38520871e-02 -5.08972287e-01
-6.04652822e-01 1.04804313e+00 3.82162482e-01 2.70012040e-02
5.34414291e-01 3.82582515e-01 6.59028590e-01 -3.77553701e-01
3.38005051e-02 -4.10562828e-02 3.79749835e-01 5.34757495e-01
6.38025045e-01 -2.49419913e-01 -4.69815344e-01 4.75201070e-01
-1.55266628e-01 9.48466547e-03 6.75988197e-01 1.20014918e+00
-1.30792007e-01 -9.46671546e-01 -7.99582779e-01 6.36413455e-01
-1.15078485e+00 5.25374711e-03 1.96528379e-02 7.08761930e-01
3.61537725e-01 8.87007535e-01 -1.69681758e-01 -1.00452431e-01
3.81257862e-01 8.16836730e-02 2.10395530e-01 -6.69066012e-01
-9.19108748e-01 4.43797469e-01 1.99215725e-01 -6.24547541e-01
-7.19169319e-01 -6.16487086e-01 -1.15882480e+00 2.47609988e-01
-3.73183280e-01 2.11418271e-01 7.22794950e-01 8.53633940e-01
-4.60412353e-02 9.58876073e-01 6.83164224e-02 -1.22178817e+00
-5.33407211e-01 -1.11939919e+00 -2.76053190e-01 6.89456284e-01
3.12725604e-01 -3.58564109e-01 -2.12181732e-01 6.62300527e-01]
|
[14.50753402709961, -2.307673454284668]
|
2bfabb1e-6f3e-4ced-8260-9d3645a4bfa1
|
autooptlib-a-library-of-automatically
|
2303.06536
| null |
https://arxiv.org/abs/2303.06536v1
|
https://arxiv.org/pdf/2303.06536v1.pdf
|
AutoOptLib: A Library of Automatically Designing Metaheuristic Optimization Algorithms in MATLAB
|
Metaheuristic algorithms are widely-recognized solvers for challenging optimization problems with multi-modality, discretization, large-scale, multi-objectivity, etc. Automatically designing metaheuristic algorithms leverages today's increasing computing resources to conceive, build up, and verify the design choices of algorithms. It requires much less expertise, labor resources, and time cost than the traditional manual design. Furthermore, by fully exploring the design choices with computing power, automated design is potential to reach or even surpass human-level design, subsequently gaining enhanced performance compared with human problem-solving. These significant advantages have attracted increasing interest and development in the automated design techniques. Open source software is indispensable in response to the increasing interest and development of the techniques. To this end, we have developed a MATLAB library, AutoOptLib, to automatically design metaheuristic algorithms. AutoOptLib, for the first time, provides throughout support to the whole design process, including: 1) plenty of algorithmic components for continuous, discrete, and permutation problems, 2) flexible algorithm representation for evolving diverse algorithm structures, 3) various design objectives and design techniques for different experimentation and application scenarios, and 4) useful experimental tools and graphic user interface (GUI) for practicability and accessibility. In this paper, we first introduce the key features and architecture of the AutoOptLib library. We then illustrate how to use the library by either command or GUI. We further describe additional uses and experimental tools, including parameter importance analysis and benchmark comparison. Finally, we present academic and piratical applications of AutoOptLib, which verifies its efficiency and practicability.
|
['Yuhui Shi', 'Xianglong Chen', 'Taiwei Hu', 'Bai Yan', 'Qi Zhao']
|
2023-03-12
| null | null | null | null |
['metaheuristic-optimization']
|
['methodology']
|
[-1.46162659e-01 -5.06296992e-01 -4.18128707e-02 7.20569193e-02
-4.46762174e-01 -9.47640240e-01 1.33502424e-01 -2.13655129e-01
1.03962332e-01 1.09899259e+00 -3.20036262e-01 -4.71135288e-01
-7.61003911e-01 -8.05482626e-01 -3.51978749e-01 -8.91450763e-01
-1.89778179e-01 6.03446841e-01 -1.54730305e-01 -3.15022707e-01
7.97991037e-01 6.20656610e-01 -2.12982392e+00 -1.68459252e-01
1.32105243e+00 7.98507810e-01 2.57477880e-01 5.28442264e-01
-2.33770549e-01 2.58199330e-02 -5.25390327e-01 -8.97925347e-02
3.32355767e-01 -5.00458300e-01 -7.61124432e-01 -3.53410579e-02
-3.95883292e-01 3.24145824e-01 1.88964769e-01 9.19611752e-01
7.37354040e-01 2.35270590e-01 4.35121983e-01 -1.51840341e+00
-5.23322761e-01 4.96679962e-01 -6.99623644e-01 -3.00052445e-02
5.75588644e-01 6.07186079e-01 8.02897632e-01 -5.32520115e-01
4.68315631e-01 1.04238439e+00 4.30328280e-01 2.11967275e-01
-8.69336188e-01 -7.08904266e-01 -4.70666476e-02 2.98980325e-01
-1.55623281e+00 -1.29603878e-01 7.77834535e-01 -1.19869396e-01
1.09649920e+00 9.50168550e-01 1.07440865e+00 7.15702951e-01
1.47888586e-01 7.10307598e-01 1.01612079e+00 -5.09210885e-01
4.75741982e-01 2.05764323e-01 4.75899503e-02 7.76996911e-01
4.74879295e-01 2.95220971e-01 -1.80731416e-01 -2.66524434e-01
5.48380077e-01 -2.86658019e-01 -3.49162787e-01 -5.72565198e-01
-8.93375874e-01 7.94376612e-01 6.95263967e-02 3.45747650e-01
-3.08249176e-01 5.93999028e-02 2.15163156e-01 2.32331961e-01
-2.01499686e-01 1.05855894e+00 -5.72253466e-01 -5.56137323e-01
-8.63621652e-01 4.12923187e-01 7.90436745e-01 1.09335268e+00
4.48640049e-01 3.12039405e-01 2.11294249e-01 7.21321881e-01
7.88061321e-02 5.20976543e-01 6.99201822e-01 -8.34006667e-01
1.58615902e-01 8.66155624e-01 3.01206917e-01 -1.02999198e+00
-6.42983675e-01 -7.04947352e-01 -6.07535899e-01 2.65655667e-01
-9.07795876e-02 -1.91864297e-01 -7.41326511e-01 1.25191188e+00
5.63875020e-01 -3.45097095e-01 -1.03789851e-01 8.82738948e-01
5.45849264e-01 9.56778884e-01 -4.23127443e-01 -6.92374468e-01
1.31436884e+00 -1.10060930e+00 -6.11036122e-01 8.94196779e-02
4.72290516e-01 -1.01410198e+00 1.06157041e+00 6.61702573e-01
-1.33042407e+00 -2.67186016e-01 -1.09281933e+00 5.50879240e-01
-6.43471777e-01 3.54502760e-02 9.86863375e-01 1.12289846e+00
-7.84505129e-01 4.77529138e-01 -7.44483411e-01 -3.78203213e-01
1.25421733e-01 6.80518031e-01 2.31928468e-01 1.71540640e-02
-9.15792525e-01 8.69320810e-01 5.41449070e-01 1.94805115e-01
-2.35673383e-01 -8.92859340e-01 -6.91152990e-01 2.44909480e-01
4.87384826e-01 -8.26776981e-01 8.81515205e-01 -9.42987561e-01
-1.81477213e+00 4.69023913e-01 1.03420056e-01 9.01555195e-02
3.35516363e-01 2.63148695e-01 -5.76171994e-01 -2.79770434e-01
-3.14666837e-01 1.94725722e-01 4.44484740e-01 -1.18168235e+00
-6.60824537e-01 -8.49462077e-02 -5.99075370e-02 1.96091607e-01
-2.71579742e-01 6.33739382e-02 -3.97794813e-01 -5.95708489e-01
-7.50089139e-02 -8.05773199e-01 -3.99673074e-01 -3.44684303e-01
-1.78449705e-01 6.79140165e-02 8.63122106e-01 -2.83642292e-01
1.73509169e+00 -1.91111672e+00 3.30247790e-01 6.94908857e-01
-2.78257102e-01 4.38181221e-01 -8.99737328e-02 7.91851044e-01
-6.99733645e-02 1.04106061e-01 -3.17178637e-01 5.61465919e-01
3.79467100e-01 1.47165060e-01 1.46858916e-01 3.76968414e-01
-1.89244792e-01 8.04211020e-01 -7.97256768e-01 -3.32936794e-01
4.45181906e-01 7.40605779e-03 -8.09400320e-01 -6.59284219e-02
-2.11866245e-01 1.22831659e-02 -6.93100750e-01 1.18032074e+00
7.69518733e-01 -1.91380575e-01 4.58568722e-01 -3.62663209e-01
-7.17189133e-01 -2.15476796e-01 -1.82803369e+00 1.41592860e+00
-5.74479103e-01 2.97243744e-01 3.72401595e-01 -1.19780684e+00
9.42207396e-01 7.93540850e-02 4.88691688e-01 -8.26052070e-01
3.59496474e-01 5.39965749e-01 7.86812156e-02 -7.40029335e-01
5.37307918e-01 3.06961298e-01 -4.42359261e-02 5.99588871e-01
-2.79289752e-01 -4.24305081e-01 8.17526996e-01 -3.50693226e-01
7.97313213e-01 1.29331589e-01 4.91890550e-01 -6.50044441e-01
7.99065590e-01 3.49131286e-01 5.24090648e-01 3.78301144e-01
1.18008047e-01 1.92267194e-01 2.21252158e-01 -6.50215566e-01
-8.36856306e-01 -8.27901661e-01 -1.70783758e-01 6.66413426e-01
3.06056887e-01 -3.82816166e-01 -3.59457046e-01 -2.33239666e-01
5.30994199e-02 9.19799626e-01 -3.85903418e-01 5.91537356e-03
-6.68911874e-01 -1.01316929e+00 1.65104538e-01 1.43640622e-01
3.79554152e-01 -1.07370126e+00 -9.91177320e-01 2.85087734e-01
9.96956155e-02 -4.19346541e-01 -1.89011306e-01 1.10272340e-01
-7.93682635e-01 -1.20325494e+00 -3.46896648e-01 -9.92647707e-01
6.40777946e-01 1.48614749e-01 1.03992403e+00 2.33629659e-01
-9.84283805e-01 4.76058304e-01 -3.09583724e-01 -1.30273685e-01
8.05537328e-02 6.93210363e-02 -6.66254992e-03 -5.71962595e-01
-1.82119131e-01 -7.13531852e-01 -4.63800669e-01 6.91758871e-01
-8.90198946e-01 7.74045587e-02 6.18406236e-01 1.04357159e+00
4.16789979e-01 6.83779478e-01 5.12982607e-01 -4.33089942e-01
7.67220795e-01 -3.73647392e-01 -1.19654655e+00 6.20567381e-01
-7.69681990e-01 2.63211399e-01 6.96066499e-01 -2.58115232e-01
-8.68422806e-01 -1.15143314e-01 3.17192040e-02 -9.50814262e-02
2.13493541e-01 6.96812272e-01 -3.60593081e-01 -4.10597205e-01
4.37975734e-01 2.27034122e-01 5.96177727e-02 -2.43709534e-01
2.84486413e-01 6.65129960e-01 1.78189635e-01 -1.04925323e+00
7.49306679e-01 -2.76638214e-02 1.51855096e-01 -6.24659836e-01
-1.15366518e-01 -2.91734524e-02 1.97405554e-02 -2.12912902e-01
2.51933664e-01 -5.19938096e-02 -1.15934026e+00 2.71634758e-01
-7.50742137e-01 8.67934674e-02 -8.33071172e-02 1.93308830e-01
-3.89255404e-01 5.38311526e-02 -6.58529066e-03 -7.93607354e-01
-4.76254612e-01 -1.59337461e+00 4.98555481e-01 7.70199358e-01
-2.27794737e-01 -1.04647338e+00 -2.01300785e-01 2.94885099e-01
6.34995162e-01 5.29267132e-01 9.82104719e-01 -1.45738244e-01
-6.53717160e-01 -8.17234069e-02 3.62914167e-02 -3.03965479e-01
2.36295879e-01 6.74430966e-01 -4.23839957e-01 -4.00020152e-01
-2.18104333e-01 -9.26894471e-02 4.68913186e-03 5.05667925e-01
1.35979390e+00 -4.02804464e-01 -6.39399171e-01 8.60669017e-01
1.58634901e+00 8.30391169e-01 6.69500470e-01 7.87054777e-01
1.41295746e-01 6.05679393e-01 9.06733215e-01 8.61035705e-01
-2.25999132e-02 7.33452082e-01 3.57742369e-01 -2.32874434e-02
4.27726656e-01 2.09959552e-01 8.62164721e-02 8.13384712e-01
-1.59092173e-01 -4.14532244e-01 -9.49226975e-01 3.10658604e-01
-1.70288193e+00 -9.60779130e-01 -6.80723861e-02 2.05216646e+00
6.38127029e-01 -1.39907785e-02 1.62609428e-01 2.17021167e-01
8.09871376e-01 -3.19090962e-01 -5.41082203e-01 -1.00674760e+00
7.56576061e-02 4.08243328e-01 4.21146333e-01 2.11584389e-01
-6.87719524e-01 6.48383677e-01 6.48085499e+00 1.12149966e+00
-1.18728471e+00 -5.24907708e-01 2.81954914e-01 -3.32074344e-01
-4.98789161e-01 4.96617518e-02 -4.71650720e-01 5.81721187e-01
6.29137635e-01 -8.90886724e-01 8.61681044e-01 7.73402035e-01
3.08857352e-01 -9.92144495e-02 -7.95072377e-01 1.03538001e+00
-1.80181637e-01 -1.79485452e+00 -7.14205056e-02 -2.87696719e-02
9.18535829e-01 -7.41764307e-01 1.44756794e-01 2.39820912e-01
2.02724338e-01 -1.10497499e+00 6.09309494e-01 3.76194790e-02
5.07649362e-01 -1.21094549e+00 6.93907917e-01 -9.80006158e-03
-1.32219517e+00 -4.19444680e-01 -5.96558973e-02 -2.27713231e-02
1.90336660e-01 2.44782761e-01 -3.66509050e-01 1.17402589e+00
7.23172665e-01 3.45845640e-01 -3.36932659e-01 1.58035672e+00
2.36452624e-01 3.91910851e-01 -4.33321059e-01 -5.13319969e-01
3.62825394e-01 -4.83416051e-01 6.70237899e-01 8.54547977e-01
5.42038918e-01 2.39175484e-01 9.73062590e-03 8.40942025e-01
6.22622550e-01 2.17948169e-01 -1.67913318e-01 -3.87992799e-01
7.48978972e-01 1.29682016e+00 -9.29820418e-01 1.77582204e-01
-8.52116868e-02 3.48215818e-01 -2.67069072e-01 3.41127932e-01
-1.26257169e+00 -9.13210988e-01 6.93673611e-01 -2.40569279e-01
3.08913291e-01 -1.89282686e-01 -6.41982615e-01 -5.83365142e-01
-1.71400055e-01 -1.38538802e+00 5.57956696e-01 -7.08341479e-01
-9.29322958e-01 4.57139939e-01 -5.62062999e-03 -1.25805628e+00
-2.27366254e-01 -7.90212035e-01 -7.27778316e-01 7.38503754e-01
-1.01945519e+00 -8.15080166e-01 -2.75948614e-01 2.53820300e-01
5.64741075e-01 -4.58063781e-01 7.01606333e-01 3.13975155e-01
-8.78035665e-01 7.43063807e-01 4.77259904e-01 -7.59892106e-01
2.38916367e-01 -8.52160335e-01 -6.66339844e-02 6.06593430e-01
-5.62727213e-01 8.84849310e-01 9.71826851e-01 -3.25932920e-01
-2.29665661e+00 -4.29437757e-01 2.03721493e-01 -4.50886413e-03
8.50961506e-01 -4.80146073e-02 -3.96782249e-01 6.73115179e-02
1.77676335e-01 -6.58492029e-01 7.09732831e-01 1.95816755e-01
3.91466379e-01 -2.83351570e-01 -1.34787667e+00 7.79157579e-01
8.69206607e-01 4.48404104e-01 -3.14084470e-01 1.76729977e-01
3.85366678e-01 -6.31481409e-01 -7.06142962e-01 3.17449749e-01
6.97929084e-01 -8.51586342e-01 1.11153805e+00 -3.50310236e-01
1.75224006e-01 -5.66301942e-01 4.42734472e-02 -1.10255837e+00
-4.30085450e-01 -1.07652497e+00 1.39824813e-02 1.32252526e+00
4.89520252e-01 -8.12030911e-01 6.08926594e-01 6.89686894e-01
-3.88909519e-01 -1.11194396e+00 -7.67704248e-01 -8.58407140e-01
-1.61613211e-01 -1.68932468e-01 1.14418471e+00 9.72913504e-01
2.14305192e-01 -1.74812123e-01 -1.75099656e-01 1.87917069e-01
4.26885605e-01 8.09813142e-01 6.28938735e-01 -8.92049015e-01
-5.00805199e-01 -8.57583344e-01 -3.43094677e-01 -5.03001928e-01
-2.90127188e-01 -5.83618462e-01 -4.16664541e-01 -1.47831321e+00
-7.73501396e-02 -5.60456514e-01 -7.88197666e-02 3.04342449e-01
7.81571642e-02 -2.56362613e-02 -6.39516041e-02 -1.70619056e-01
-3.24979544e-01 6.52415872e-01 1.40617263e+00 -1.08589239e-01
-5.60067713e-01 5.67846373e-02 -8.64683449e-01 4.80143458e-01
9.23430920e-01 -2.51038641e-01 -4.54835951e-01 -1.75116226e-01
3.85447562e-01 4.04122286e-02 -3.67921889e-02 -1.00653100e+00
9.79722813e-02 -7.13740826e-01 2.85859168e-01 -5.92570782e-01
-6.94625080e-03 -1.04912925e+00 6.92114890e-01 8.15878272e-01
1.78354070e-01 5.25699258e-01 5.48147678e-01 -5.81688154e-03
-9.56441984e-02 -4.40852165e-01 6.67908669e-01 -3.77401933e-02
-6.69390678e-01 -1.09396204e-01 -4.49822366e-01 -2.13712379e-02
1.46332407e+00 -7.24974453e-01 -2.51456141e-01 1.08626455e-01
-3.52219790e-01 7.91469634e-01 6.51467502e-01 3.43003958e-01
3.46655250e-01 -1.13999033e+00 -1.90697446e-01 3.84131640e-01
-2.44416952e-01 -4.19768363e-01 2.99181998e-01 7.01223373e-01
-1.07524431e+00 4.35625225e-01 -5.46228588e-01 -3.36635619e-01
-1.31590712e+00 7.59870589e-01 3.14271063e-01 -1.40473053e-01
-1.85218811e-01 6.44639194e-01 -3.19813311e-01 -4.34876323e-01
1.53615149e-02 -1.35407627e-01 5.86698912e-02 -3.71914618e-02
3.16051900e-01 9.02033508e-01 3.45276222e-02 5.38793094e-02
-7.28433847e-01 7.60093987e-01 1.57265022e-01 3.56468260e-02
1.59440339e+00 4.14194278e-02 -3.20201993e-01 -2.05221251e-01
7.00974047e-01 6.16386756e-02 -6.06330395e-01 6.75500512e-01
-1.95816010e-01 -6.36755645e-01 5.09741204e-03 -1.14591408e+00
-1.12116623e+00 2.93608427e-01 5.60350060e-01 2.66243696e-01
1.53158987e+00 -5.72293818e-01 6.23892486e-01 3.30279261e-01
6.99840248e-01 -1.34178209e+00 -2.93330044e-01 3.61470252e-01
8.68683040e-01 -5.67467988e-01 4.42104071e-01 -2.95542270e-01
-4.96384501e-01 1.42831647e+00 6.77845597e-01 2.09196270e-01
3.20926040e-01 5.50495863e-01 -2.05468461e-01 -2.03445107e-01
-6.60459101e-01 3.85922305e-02 2.81483680e-01 6.08804047e-01
3.57594699e-01 -1.11765712e-01 -8.23470056e-01 7.01100349e-01
-5.10580599e-01 2.67442260e-02 3.22031230e-01 1.35437524e+00
-4.77408111e-01 -1.37674522e+00 -8.42416704e-01 2.34281525e-01
-2.72525400e-02 1.24921761e-01 -2.01551005e-01 9.35763121e-01
1.18878625e-01 9.73612189e-01 -2.59938240e-01 -2.16691419e-01
2.98171282e-01 -3.10919851e-01 6.40099287e-01 7.83919692e-02
-8.21419299e-01 -1.95733517e-01 2.49359921e-01 -5.65876782e-01
3.18155512e-02 -5.86017549e-01 -1.17850709e+00 -6.56543076e-01
-4.41269726e-01 8.02895367e-01 7.80489028e-01 7.11238086e-01
5.13904154e-01 6.36143684e-01 6.20100319e-01 -8.18381071e-01
-3.28267485e-01 -2.37212971e-01 -4.10458714e-01 -3.67671430e-01
-2.63311833e-01 -1.01322985e+00 -2.37374470e-01 -3.40677261e-01]
|
[5.736851692199707, 3.6100149154663086]
|
a311f197-97cc-42e6-bbc8-89a6fe33d36b
|
plug-and-play-recipe-generation-with-content
|
2212.05093
| null |
https://arxiv.org/abs/2212.05093v1
|
https://arxiv.org/pdf/2212.05093v1.pdf
|
Plug-and-Play Recipe Generation with Content Planning
|
Recent pre-trained language models have shown promising capabilities in generating fluent and realistic natural language text. However, generating multi-sentence text with global content planning has been a long-existing research question. Current approaches for controlled text generation can hardly address this issue, as they usually condition on single known control attributes. In this study, we propose a low-cost yet effective framework which explicitly models the global content plan of the generated text. Specifically, it optimizes the joint distribution of the natural language sequence and the global content plan in a plug-and-play manner. We conduct extensive experiments on the well-established Recipe1M+ benchmark. Both automatic and human evaluations verify that our model achieves the state-of-the-art performance on the task of recipe generation
|
['Nigel Collier', 'Ehsan Shareghi', 'Yixuan Su', 'Yinhong Liu']
|
2022-12-09
| null | null | null | null |
['recipe-generation']
|
['miscellaneous']
|
[ 1.29078180e-01 1.55284837e-01 -8.17672387e-02 -2.46619821e-01
-1.00694251e+00 -5.70491135e-01 1.09814215e+00 2.24162161e-01
-2.38285527e-01 8.85188699e-01 6.11790419e-01 -1.78659678e-01
4.72487301e-01 -1.01951742e+00 -8.98871899e-01 -3.82782310e-01
2.40420520e-01 7.75797427e-01 4.81360778e-03 -4.77901518e-01
2.84135044e-01 -2.12959334e-01 -1.25169396e+00 3.92543107e-01
1.09687734e+00 3.21327120e-01 5.37012935e-01 8.63713145e-01
-3.33088368e-01 1.13546216e+00 -6.65608585e-01 -3.14730376e-01
6.61696196e-02 -8.82091463e-01 -8.07285964e-01 7.89704844e-02
1.38592333e-01 -2.43487298e-01 5.22819720e-02 8.10259640e-01
5.95674396e-01 2.72045523e-01 7.12364674e-01 -1.02841449e+00
-4.87403780e-01 1.03618574e+00 -1.94336981e-01 -1.84158579e-01
6.69212162e-01 4.58132386e-01 1.15453815e+00 -4.73548025e-01
7.85855234e-01 1.23135364e+00 3.23356003e-01 8.20908546e-01
-1.44638193e+00 -1.34698808e-01 1.80287585e-01 -3.68237257e-01
-1.10894406e+00 -3.47398072e-01 5.06268263e-01 -3.82690817e-01
1.12412965e+00 8.23983327e-02 5.22914767e-01 1.17716193e+00
3.81673872e-01 1.05707419e+00 1.08218634e+00 -4.79032397e-01
3.12278301e-01 -1.45130828e-01 -4.02424634e-01 7.48638451e-01
-9.56985205e-02 -1.39142890e-02 -6.07155621e-01 -2.11795494e-02
5.71471453e-01 -6.28403187e-01 -1.85305685e-01 -1.38694504e-02
-1.73488188e+00 1.00696862e+00 -1.14019349e-01 2.33873591e-01
-4.89439964e-01 4.48361605e-01 4.40800637e-01 7.75862560e-02
7.32573569e-01 9.83535111e-01 -5.87728500e-01 -3.54079366e-01
-9.55619156e-01 1.02301371e+00 1.09892917e+00 1.19343400e+00
3.64948422e-01 2.69692782e-02 -7.77264178e-01 5.66377103e-01
1.77530229e-01 4.93585914e-01 4.74678129e-01 -8.02209795e-01
7.39467680e-01 3.57988238e-01 4.75724339e-01 -6.07798398e-01
-3.01837564e-01 2.59618573e-02 -6.91723228e-01 -7.05451518e-02
4.66556489e-01 -5.61225057e-01 -8.79579186e-01 1.92584229e+00
3.65556389e-01 7.67603097e-03 1.82626322e-01 6.46410644e-01
5.78663707e-01 1.05667043e+00 3.69610012e-01 -1.74431503e-01
1.06791723e+00 -1.38150215e+00 -8.02793324e-01 -3.86739403e-01
6.51671469e-01 -9.53085959e-01 1.33761632e+00 2.15362549e-01
-1.41442585e+00 -4.90741730e-01 -8.59447122e-01 -1.16990902e-01
-1.67250887e-01 2.51847953e-01 6.74023807e-01 2.76332825e-01
-9.61706996e-01 7.82065928e-01 -6.50511146e-01 -1.74566343e-01
-1.20223761e-01 -6.71288893e-02 -8.84045213e-02 -1.04072668e-01
-1.17279434e+00 8.29200089e-01 6.31519437e-01 -1.86340913e-01
-1.15068519e+00 -8.49720240e-01 -1.00562823e+00 -9.25881509e-03
5.76571286e-01 -1.08165681e+00 1.82138348e+00 -4.45792288e-01
-2.08729172e+00 4.74285066e-01 6.14638557e-04 -4.09732491e-01
6.61678433e-01 -3.52177858e-01 -1.14071928e-01 1.98442722e-03
3.44011724e-01 9.91308451e-01 5.65995693e-01 -9.90966201e-01
-5.23310959e-01 3.28862429e-01 1.34522334e-01 3.37866485e-01
6.23164773e-02 3.21460366e-02 -1.44354090e-01 -8.10877919e-01
-4.14940834e-01 -8.48437369e-01 -6.44520581e-01 -3.66616637e-01
-6.20496809e-01 -3.68880451e-01 6.24435879e-02 -5.35477817e-01
1.18807399e+00 -1.75846922e+00 2.52798885e-01 -4.12797570e-01
-3.39138478e-01 1.19295143e-01 -4.28233564e-01 8.16306412e-01
3.20331931e-01 4.19374883e-01 -1.97109014e-01 -6.08972490e-01
3.50759476e-01 -1.77945480e-01 -4.73472983e-01 -2.72634476e-02
3.46303850e-01 9.90899205e-01 -1.23077428e+00 -6.62076175e-01
1.80349827e-01 1.63682967e-01 -7.69931555e-01 7.69056082e-01
-1.37848270e+00 5.29491067e-01 -6.12359166e-01 -1.12100570e-02
2.30933756e-01 -3.02062362e-01 1.13147154e-01 2.52045959e-01
-3.34375978e-01 7.62129605e-01 -7.82959878e-01 2.20193863e+00
-6.48178756e-01 8.83271918e-02 -2.02539712e-01 -3.71217221e-01
7.45737851e-01 5.61784685e-01 2.48149529e-01 -4.43007350e-01
-5.45455189e-03 4.07051441e-04 -1.39227197e-01 -4.32290077e-01
8.68300319e-01 -3.00969034e-01 -5.46239316e-01 7.32914150e-01
6.86454922e-02 -8.41340661e-01 8.30305457e-01 3.89497399e-01
9.47380543e-01 7.67671585e-01 4.18471366e-01 -3.64649713e-01
3.33995521e-01 2.71347135e-01 1.46395132e-01 8.92957687e-01
1.63076743e-01 8.52842569e-01 6.80276155e-01 -1.32654771e-01
-1.23511040e+00 -7.62614071e-01 5.80660939e-01 1.10139370e+00
-2.32981905e-01 -8.75178099e-01 -1.06663799e+00 -7.68707454e-01
-4.26801115e-01 1.40678513e+00 -4.92137522e-01 1.33276656e-01
-6.56849921e-01 -7.20876694e-01 4.65118408e-01 2.41670564e-01
3.35571587e-01 -1.54001749e+00 -4.88519073e-01 6.38006151e-01
-6.33708119e-01 -1.23070514e+00 -8.80464494e-01 -2.51461029e-01
-6.01264954e-01 -6.38695002e-01 -6.26067042e-01 -7.44879007e-01
4.73959535e-01 -1.08413759e-03 1.64124596e+00 8.64202008e-02
-1.28063515e-01 1.76363647e-01 -5.05277097e-01 -4.55018163e-01
-1.13476634e+00 3.35706770e-01 -4.25412476e-01 -2.31205806e-01
-1.17810480e-01 -2.67132759e-01 -3.99455339e-01 3.06913760e-02
-9.42993164e-01 8.13794672e-01 3.53738278e-01 7.86602676e-01
5.01536489e-01 -1.36292219e-01 7.31548429e-01 -8.02412391e-01
9.41133738e-01 -3.15403551e-01 -6.39534593e-01 3.91017765e-01
-4.41474080e-01 4.00094569e-01 9.96950090e-01 -3.22030246e-01
-1.32388818e+00 1.71921119e-01 -3.74896079e-01 3.06699127e-01
-4.24276918e-01 6.60590053e-01 -1.34887308e-01 7.23947585e-01
4.90594804e-01 3.22040081e-01 -2.60128886e-01 -2.85484165e-01
8.19214225e-01 3.25611979e-01 3.63759041e-01 -1.13987553e+00
6.60460532e-01 2.85550971e-02 -6.40778467e-02 -5.96018076e-01
-1.06881511e+00 -2.63508886e-01 -3.41210097e-01 -1.05443135e-01
1.20890713e+00 -9.64522719e-01 -4.55472052e-01 5.88237405e-01
-1.47400534e+00 -1.00765789e+00 -2.73905128e-01 2.67099887e-01
-1.06906223e+00 1.77968130e-01 -7.95611501e-01 -6.60881639e-01
-6.49645388e-01 -1.04767215e+00 1.34655631e+00 4.47773263e-02
-4.45106655e-01 -9.71609890e-01 4.16010886e-01 1.49225652e-01
5.43410718e-01 3.22537333e-01 9.90012050e-01 -4.34025675e-01
-8.57768953e-01 -8.72125551e-02 1.35535851e-01 -3.72518157e-03
-6.58819778e-03 6.19261712e-02 -5.86725354e-01 5.23523875e-02
-2.02802226e-01 -6.94973171e-01 6.29695296e-01 1.64503649e-01
7.19947219e-01 -4.31889981e-01 8.95750970e-02 1.82534695e-01
1.32388997e+00 -1.44604430e-01 6.81117117e-01 4.19050716e-02
6.30611062e-01 6.42247736e-01 7.83675551e-01 6.60021603e-01
4.66687799e-01 6.28149807e-01 2.37728044e-01 1.08869456e-01
-1.65588453e-01 -8.89517665e-01 5.68547606e-01 9.26642597e-01
1.86612457e-01 -8.60055685e-01 -6.31360114e-01 6.38449192e-01
-1.98937058e+00 -1.13799763e+00 -1.75726816e-01 1.94892454e+00
1.33190989e+00 1.00016847e-01 2.62272358e-03 -2.73129374e-01
4.85769451e-01 4.20650870e-01 -8.63430202e-02 -3.05013150e-01
1.17759585e-01 1.80264682e-01 1.19147442e-01 5.54559231e-01
-1.10552049e+00 1.32260752e+00 6.16609335e+00 8.83144915e-01
-8.57644022e-01 -1.70018319e-02 6.64077878e-01 -1.02062151e-01
-4.73177463e-01 7.26405438e-03 -9.24629390e-01 4.24613178e-01
1.14416277e+00 -4.61321294e-01 5.75666428e-01 8.35527241e-01
6.52108014e-01 -2.03131139e-01 -1.29054189e+00 3.60652864e-01
-2.82838922e-02 -1.51655304e+00 2.33979568e-01 -1.11462101e-01
1.02120876e+00 -1.55031592e-01 -2.68626988e-01 6.09769821e-01
7.24517703e-01 -9.74649966e-01 1.22079432e+00 4.94278133e-01
7.42223978e-01 -5.25796652e-01 4.55061316e-01 7.76805282e-01
-1.06289661e+00 4.33913350e-01 -1.46060422e-01 -2.07027346e-01
8.45769763e-01 5.67278028e-01 -9.89392936e-01 5.41162014e-01
1.20330535e-01 4.98466641e-01 -3.97667885e-01 7.19326794e-01
-6.64767981e-01 6.98111534e-01 -1.14767991e-01 -4.64846939e-01
3.93406928e-01 -1.88612461e-01 4.00473893e-01 1.33062518e+00
3.93225253e-01 -3.05560902e-02 6.44925594e-01 1.30246150e+00
-9.99248847e-02 4.32348371e-01 -5.98006427e-01 -6.32432461e-01
-4.05171365e-02 1.22536016e+00 -6.91761553e-01 -4.93729949e-01
-2.13299274e-01 9.87751126e-01 3.80236357e-01 9.82252136e-02
-8.86514604e-01 -1.46533266e-01 3.36816132e-01 2.53532417e-02
3.31436634e-01 -3.71174634e-01 -3.36500913e-01 -1.21142769e+00
-2.20425710e-01 -1.12017858e+00 1.15223169e-01 -8.18109274e-01
-1.26189888e+00 5.34348130e-01 5.47287427e-02 -8.26786458e-01
-7.31994510e-01 -2.77250081e-01 -6.96617961e-01 8.39804471e-01
-1.29880822e+00 -1.14125884e+00 -1.18340254e-02 2.33813509e-01
8.73935759e-01 -5.99359572e-02 1.15902519e+00 -2.61545237e-02
-4.91090596e-01 1.27035260e-01 -1.88788190e-01 -2.65253186e-02
6.72467530e-01 -1.49920118e+00 1.03072512e+00 9.40265715e-01
1.37623727e-01 5.39718926e-01 1.14330220e+00 -9.03266430e-01
-1.23511839e+00 -1.25067127e+00 1.27795625e+00 -3.81033599e-01
6.35885775e-01 -6.96810663e-01 -4.64357942e-01 4.35602218e-01
6.16888881e-01 -5.75517297e-01 2.71196544e-01 -1.94686919e-01
-6.65990487e-02 4.52169299e-01 -9.19998467e-01 8.85607779e-01
8.57833862e-01 -1.94535166e-01 -3.40827793e-01 7.04684019e-01
1.24540806e+00 -6.65483773e-01 -6.82300150e-01 1.33048475e-01
7.02025890e-02 -8.24478149e-01 6.10238552e-01 -5.63537717e-01
1.22841644e+00 -3.27607363e-01 4.72066738e-02 -1.61193311e+00
-1.28040373e-01 -1.09416449e+00 1.48161083e-01 1.35257852e+00
7.44524539e-01 -8.25516433e-02 6.18133187e-01 6.03692770e-01
-3.60493034e-01 -6.37300074e-01 -3.27340543e-01 -6.50955379e-01
2.25837618e-01 -2.67111003e-01 6.99318707e-01 5.10598481e-01
1.58150762e-01 8.65853786e-01 -6.96777701e-01 -4.28477168e-01
2.55494505e-01 2.44078666e-01 9.65420663e-01 -6.11926258e-01
-5.66653609e-01 -3.66122335e-01 3.92526537e-01 -1.33670533e+00
4.28912014e-01 -8.57074976e-01 7.59813547e-01 -2.02369142e+00
2.06110850e-01 -3.33818465e-01 4.86340284e-01 1.74277797e-01
-5.31339824e-01 -3.43284369e-01 4.18443948e-01 -2.86271960e-01
-7.56321073e-01 8.36137116e-01 1.55242586e+00 -8.74580070e-02
-1.56743839e-01 -6.70692623e-02 -5.91948211e-01 3.59299779e-01
1.03848612e+00 -5.39720356e-01 -5.69865763e-01 -4.46228325e-01
4.26699728e-01 2.53305942e-01 -3.93355079e-02 -7.96664834e-01
-5.17752506e-02 -6.53108180e-01 -3.03732604e-01 -4.80351716e-01
1.59362748e-01 -1.43863887e-01 1.15937851e-02 1.79106548e-01
-8.07892740e-01 2.26390064e-01 1.36760280e-01 5.16263843e-01
-3.14094387e-02 -4.42023069e-01 5.11915386e-01 -6.50725067e-01
-2.80264169e-01 2.65645921e-01 -4.98166442e-01 5.79660237e-01
7.65672445e-01 7.07847416e-01 -4.16443169e-01 -6.61228180e-01
-6.53370470e-02 2.19678462e-01 4.47322607e-01 4.13632393e-01
3.34934294e-01 -1.15856862e+00 -1.11715364e+00 -2.06939355e-01
2.71290280e-02 1.79448426e-01 1.32141113e-01 3.63199413e-01
-6.55196249e-01 5.50473392e-01 2.20604375e-01 -2.08058566e-01
-6.69782043e-01 7.66930401e-01 1.72003463e-01 -8.99116874e-01
-5.33504546e-01 6.08137369e-01 1.88903451e-01 -6.02621078e-01
-1.23324960e-01 -3.99955273e-01 2.20684242e-02 -3.01156700e-01
7.33685076e-01 9.82120410e-02 -2.21382558e-01 -4.09482300e-01
2.29824618e-01 5.00177480e-02 4.10628878e-02 -5.45234621e-01
1.11639774e+00 1.86556712e-01 -1.56010062e-01 4.61903006e-01
8.27653885e-01 1.91851929e-02 -1.29207551e+00 1.37355179e-01
4.26562391e-02 -4.40419465e-02 -2.47761399e-01 -1.02856064e+00
-6.20602012e-01 7.59465873e-01 -3.54010999e-01 2.62299538e-01
6.06980145e-01 -1.00287832e-01 1.06047559e+00 2.98897505e-01
4.92071331e-01 -1.12599313e+00 4.04358387e-01 8.36731553e-01
1.09973085e+00 -1.06349897e+00 -1.43416911e-01 -3.95035177e-01
-8.47371936e-01 9.11327124e-01 7.44069517e-01 -6.14506342e-02
1.46688655e-01 2.84367561e-01 1.49795502e-01 1.41644090e-01
-1.24176919e+00 -1.52291968e-01 1.40113235e-01 4.04911369e-01
8.42409551e-01 2.43404239e-01 -5.18651068e-01 4.27820414e-01
-3.59479427e-01 1.57826826e-01 7.51367033e-01 8.44076812e-01
-3.98794711e-01 -1.33823788e+00 -3.21083440e-04 2.09248126e-01
-5.70216477e-01 -4.73556697e-01 -3.01381409e-01 4.67828631e-01
-2.38818929e-01 1.21781695e+00 -2.23614663e-01 -9.12901387e-02
2.71275520e-01 2.61411220e-01 5.89654684e-01 -9.74831283e-01
-8.96678090e-01 1.43323302e-01 5.88686287e-01 -4.45389420e-01
-2.79756218e-01 -6.91819131e-01 -1.39242315e+00 -2.95328736e-01
-1.55149341e-01 2.94453323e-01 7.53471315e-01 9.97593343e-01
2.81918526e-01 6.19484067e-01 4.37451810e-01 -9.83203173e-01
-1.00622785e+00 -1.17969620e+00 -2.29446217e-01 6.28836036e-01
1.39754936e-02 -8.36436823e-02 -1.29987895e-01 3.17520857e-01]
|
[11.752083778381348, 8.957332611083984]
|
0ac42fcc-7375-4e49-9b41-807ce83f5aa5
|
refsr-nerf-towards-high-fidelity-and-super
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Huang_RefSR-NeRF_Towards_High_Fidelity_and_Super_Resolution_View_Synthesis_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Huang_RefSR-NeRF_Towards_High_Fidelity_and_Super_Resolution_View_Synthesis_CVPR_2023_paper.pdf
|
RefSR-NeRF: Towards High Fidelity and Super Resolution View Synthesis
|
We present Reference-guided Super-Resolution Neural Radiance Field (RefSR-NeRF) that extends NeRF to super resolution and photorealistic novel view synthesis. Despite NeRF's extraordinary success in the neural rendering field, it suffers from blur in high resolution rendering because its inherent multilayer perceptron struggles to learn high frequency details and incurs a computational explosion as resolution increases. Therefore, we propose RefSR-NeRF, an end-to-end framework that first learns a low resolution NeRF representation, and then reconstructs the high frequency details with the help of a high resolution reference image. We observe that simply introducing the pre-trained models from the literature tends to produce unsatisfied artifacts due to the divergence in the degradation model. To this end, we design a novel lightweight RefSR model to learn the inverse degradation process from NeRF renderings to target HR ones. Extensive experiments on multiple benchmarks demonstrate that our method exhibits an impressive trade-off among rendering quality, speed, and memory usage, outperforming or on par with NeRF and its variants while being 52x speedup with minor extra memory usage.
|
['Yunhe Wang', 'Hanting Chen', 'Jie Hu', 'Wei Li', 'Xudong Huang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['neural-rendering', 'novel-view-synthesis']
|
['computer-vision', 'computer-vision']
|
[ 5.40489733e-01 -5.32397479e-02 2.12657154e-01 -3.16872180e-01
-9.44996655e-01 -1.41138941e-01 6.84202313e-01 -6.32557392e-01
1.04633346e-02 7.16373861e-01 5.88189960e-01 -1.08359708e-02
-2.38869917e-02 -7.80532598e-01 -9.60207582e-01 -6.22324109e-01
2.03821346e-01 -4.08341140e-02 7.79998153e-02 -5.76519072e-01
2.95261979e-01 5.54401338e-01 -1.80087411e+00 5.31563163e-01
8.37130547e-01 8.94027650e-01 3.67414266e-01 1.15261590e+00
-2.16985494e-02 1.29707742e+00 -6.89682007e-01 -4.54913348e-01
4.68325645e-01 -3.13012242e-01 -5.03319442e-01 -3.30963790e-01
1.13477254e+00 -7.22355783e-01 -6.81492031e-01 7.82449841e-01
7.11377740e-01 3.88607830e-01 4.25160915e-01 -5.83562553e-01
-1.29531991e+00 2.42745921e-01 -9.41202819e-01 3.25015962e-01
4.29368079e-01 4.52341996e-02 9.05088246e-01 -9.64729786e-01
7.31565058e-01 1.43972802e+00 8.33284318e-01 7.13803768e-01
-1.44640565e+00 -5.33350945e-01 4.56832014e-02 -2.20787019e-01
-1.23210096e+00 -4.83036250e-01 5.51867485e-01 -1.29855096e-01
1.00925958e+00 5.89209020e-01 2.86738992e-01 1.39901781e+00
4.18840170e-01 2.96836734e-01 1.22769356e+00 -5.76403849e-02
1.12508096e-01 -1.06309533e-01 -4.17858474e-02 4.16454077e-01
-8.27661157e-02 5.92610300e-01 -7.38413334e-01 3.40240598e-02
1.49130678e+00 -1.16406485e-01 -7.97306657e-01 1.49375752e-01
-1.03191853e+00 3.82375270e-01 7.83119082e-01 -5.33702150e-02
-4.29297477e-01 2.79064417e-01 6.61919340e-02 3.44184339e-01
6.78501666e-01 6.03503287e-01 -3.99508119e-01 -1.73455924e-02
-7.97818899e-01 1.77648976e-01 3.73709321e-01 7.17447042e-01
5.34894347e-01 3.75064254e-01 -3.87123585e-01 1.16016591e+00
-3.42193544e-01 1.93847165e-01 2.12874502e-01 -1.43276715e+00
3.31745178e-01 1.16927698e-01 3.72751862e-01 -8.29255462e-01
-1.93041980e-01 -7.46526062e-01 -1.41259778e+00 6.40839756e-01
-1.20795235e-01 2.50394225e-01 -1.04447806e+00 1.67909610e+00
4.38769124e-02 4.80188936e-01 3.11546206e-01 1.19000041e+00
1.10025871e+00 1.02386236e+00 -1.41948938e-01 -6.76595196e-02
1.10751605e+00 -1.16418600e+00 -6.80085838e-01 -1.22313812e-01
-3.79755944e-01 -7.21254289e-01 1.43810475e+00 3.89568657e-01
-1.33968568e+00 -1.05465901e+00 -1.19978011e+00 -6.14627898e-01
6.08637184e-02 -9.61530805e-02 6.73896611e-01 3.32152039e-01
-1.38124418e+00 1.10662913e+00 -4.16070908e-01 2.16986910e-01
4.75142330e-01 -6.21663481e-02 -9.50380713e-02 -2.67708540e-01
-9.46691155e-01 6.90347314e-01 2.24251617e-02 4.79940735e-02
-5.72096586e-01 -1.43711102e+00 -7.31591225e-01 1.44017443e-01
1.89742357e-01 -1.16724300e+00 1.16085804e+00 -8.61221850e-01
-1.93945944e+00 4.35142487e-01 -9.57626551e-02 -2.62440741e-01
6.90584123e-01 -5.03908634e-01 -5.88688135e-01 -5.04094139e-02
-1.29728198e-01 7.61319101e-01 1.14428687e+00 -1.63279593e+00
-8.11442912e-01 -1.36125445e-01 3.04663420e-01 2.47406945e-01
1.03208892e-01 -2.78131545e-01 -3.41370702e-01 -7.69031167e-01
-1.27651710e-02 -3.57730746e-01 -2.75722563e-01 -8.28987081e-03
-9.84525010e-02 1.19840778e-01 5.48186302e-01 -8.26503694e-01
9.39720392e-01 -2.25935411e+00 2.11331785e-01 -3.07409227e-01
6.73163891e-01 5.40172420e-02 -3.42497647e-01 -6.55100346e-02
-2.70878434e-01 6.32727444e-02 -4.97243814e-02 -3.05686235e-01
-2.79173702e-01 -1.00277178e-02 -7.36125350e-01 1.35835186e-01
3.18461835e-01 7.32222855e-01 -9.50572371e-01 5.65866381e-02
3.96820188e-01 1.49831367e+00 -5.20160735e-01 4.84295100e-01
-1.38630331e-01 5.73530793e-01 2.51414981e-02 3.87821078e-01
9.90471900e-01 -4.60151374e-01 -2.47344434e-01 -6.41708612e-01
-1.51752532e-01 1.88934177e-01 -1.03494835e+00 1.75543201e+00
-9.31751847e-01 7.77868330e-01 -8.57140571e-02 -1.87549621e-01
8.91065121e-01 1.97298247e-02 1.30035430e-01 -1.22157872e+00
-2.57759392e-01 7.69295590e-03 -5.70764661e-01 -1.83750287e-01
1.06425512e+00 -1.48886323e-01 3.20564032e-01 3.18865143e-02
-1.41111553e-01 2.99205966e-02 -2.40149915e-01 4.97689694e-02
1.11357653e+00 5.96747279e-01 6.63647428e-02 5.25881862e-03
3.39136332e-01 -4.01999563e-01 4.36411887e-01 8.14144194e-01
4.40421551e-01 1.14098430e+00 1.99928507e-01 -7.11656153e-01
-1.34437370e+00 -1.39576936e+00 -1.90637916e-01 1.13564861e+00
1.62673101e-01 -4.39180046e-01 -6.23131931e-01 -6.11071177e-02
-3.69341135e-01 7.64713228e-01 -6.91771388e-01 -6.57343566e-02
-7.79231131e-01 -6.85492396e-01 2.35115647e-01 5.84800839e-01
8.07844460e-01 -7.63140082e-01 -9.31192040e-01 2.69513428e-02
-1.49221748e-01 -1.05584216e+00 -4.28513378e-01 -1.36971846e-01
-8.24395299e-01 -6.87080920e-01 -1.00279534e+00 -2.45959729e-01
3.60556841e-01 5.34176648e-01 1.81057847e+00 -1.12954117e-01
-3.73895437e-01 5.39211854e-02 5.04770949e-02 7.11041465e-02
-5.31271279e-01 -1.05001613e-01 -3.30043852e-01 -2.55286515e-01
-3.13399464e-01 -9.27171767e-01 -1.00668192e+00 6.32275781e-03
-8.89362693e-01 5.62687635e-01 7.15592206e-01 9.37462986e-01
8.84953499e-01 1.36467978e-01 2.28346661e-01 -8.83661389e-01
6.30571485e-01 -6.64663240e-02 -6.38537288e-01 1.38390183e-01
-6.33665383e-01 2.20316306e-01 9.80000019e-01 -4.21770096e-01
-1.56525922e+00 -3.69821340e-01 -1.79976016e-01 -7.17780948e-01
1.42539497e-02 -1.94453463e-01 1.22818045e-01 1.80087294e-02
8.98016214e-01 1.12228006e-01 -5.08691907e-01 -7.72300899e-01
5.25453746e-01 3.01150352e-01 1.22088861e+00 -4.34739262e-01
9.13547993e-01 4.17623729e-01 5.87529726e-02 -6.79575622e-01
-1.09304452e+00 4.28389162e-02 -1.22633211e-01 -9.71197635e-02
7.45604157e-01 -1.29209685e+00 -7.32377112e-01 1.87587798e-01
-1.14424574e+00 -3.94786566e-01 -6.77124143e-01 3.18064839e-02
-6.06179953e-01 3.30169164e-02 -9.09684896e-01 -8.06235313e-01
-6.00515544e-01 -9.33010161e-01 1.39511800e+00 4.06676173e-01
1.34021536e-01 -6.05826199e-01 1.89163506e-01 3.26965123e-01
8.85011375e-01 5.68355620e-01 8.04700017e-01 5.13512313e-01
-1.02387059e+00 2.18369111e-01 -7.89437413e-01 3.69937271e-01
-4.31412943e-02 -8.18178356e-02 -1.34126031e+00 -3.50684375e-01
4.48662527e-02 -1.23463020e-01 1.07796848e+00 2.86506295e-01
1.29548538e+00 -3.81136924e-01 3.25963885e-01 1.22068095e+00
1.96435249e+00 -9.31192636e-02 1.09107161e+00 2.60807782e-01
1.00915086e+00 3.65243703e-01 3.87076318e-01 3.26000035e-01
3.11902434e-01 6.44152343e-01 3.91940027e-01 -4.85700786e-01
-7.31042027e-01 -3.62877816e-01 2.75896400e-01 5.49457610e-01
-6.96192503e-01 -1.11445218e-01 -2.91067392e-01 7.51715293e-03
-1.59512305e+00 -9.51505601e-01 2.26647295e-02 2.24061704e+00
8.03994894e-01 4.62368540e-02 -2.09445789e-01 -1.74267575e-01
5.09739578e-01 5.43179035e-01 -7.20365405e-01 -4.92387205e-01
-2.43917689e-01 5.15948832e-01 4.08811569e-01 7.32669294e-01
-6.91851437e-01 7.20316708e-01 6.38115406e+00 7.92446315e-01
-1.13882422e+00 1.32715791e-01 8.79132330e-01 -2.74789333e-01
-5.94332039e-01 -4.22988772e-01 -5.54553390e-01 9.31763947e-02
1.13663745e+00 1.12629987e-01 1.04870236e+00 6.00071609e-01
3.45524512e-02 2.16625914e-01 -1.01959860e+00 1.38486457e+00
7.00836852e-02 -1.69907701e+00 4.34244812e-01 -1.34767652e-01
8.04253578e-01 5.73529303e-02 3.49799633e-01 3.13093483e-01
3.99679214e-01 -1.44720221e+00 5.70644557e-01 8.75590920e-01
1.34567022e+00 -8.52291167e-01 3.65050942e-01 7.27864262e-03
-1.15875912e+00 2.83711161e-02 -7.48054147e-01 6.30396511e-03
-6.25071023e-03 6.21832490e-01 -2.11559400e-01 7.04648733e-01
8.83454263e-01 6.69847131e-01 -4.53376830e-01 4.42858517e-01
4.96830232e-03 5.84576763e-02 8.81483033e-02 6.96883917e-01
-9.94799957e-02 -2.07361549e-01 5.18591523e-01 1.10335732e+00
3.66452247e-01 2.45682806e-01 -1.86294600e-01 1.09365559e+00
-3.87349784e-01 -1.55605257e-01 -3.45273167e-01 4.14782941e-01
1.19659215e-01 1.24181294e+00 -3.49402457e-01 -2.15715945e-01
-3.38445485e-01 1.49165154e+00 4.63943839e-01 6.80376172e-01
-9.36883509e-01 -2.03465074e-01 8.11290920e-01 3.69947463e-01
2.11117432e-01 1.59859851e-01 -3.35532278e-01 -1.22635198e+00
7.81873092e-02 -8.26758206e-01 3.21047790e-02 -1.41841757e+00
-1.21141958e+00 1.26655209e+00 -3.02862108e-01 -1.14646375e+00
-3.04210544e-01 -4.37999696e-01 -2.13754922e-01 1.12853420e+00
-1.97655392e+00 -8.72407138e-01 -6.93938971e-01 4.57871020e-01
9.21440005e-01 1.51386246e-01 7.32511640e-01 2.57874459e-01
-3.70243818e-01 5.85899949e-01 9.05997306e-02 -3.35405231e-01
6.13146603e-01 -1.40703785e+00 8.65164459e-01 7.80420542e-01
-8.16274881e-02 3.68991375e-01 9.24520373e-01 -3.97643924e-01
-1.44817400e+00 -1.20498252e+00 3.61914307e-01 -4.68272507e-01
1.68537855e-01 -2.01600626e-01 -1.09906697e+00 4.07157630e-01
1.97029173e-01 2.42646083e-01 1.83386251e-01 2.72932574e-02
-8.63647103e-01 -3.56502265e-01 -1.26584387e+00 9.58132088e-01
1.37993395e+00 -6.16434753e-01 -2.54343212e-01 5.24557196e-02
1.18293679e+00 -7.87066936e-01 -1.29334164e+00 5.25683463e-01
5.09973109e-01 -1.50487280e+00 1.45179987e+00 -2.24468336e-01
1.13075030e+00 -2.58556366e-01 -4.57778096e-01 -1.32776999e+00
-7.00195730e-01 -8.03837597e-01 -4.15348262e-01 1.12298739e+00
5.81482165e-02 -4.18600291e-01 3.96859258e-01 5.08043528e-01
-2.17161532e-02 -7.52536416e-01 -6.98185682e-01 -4.62740719e-01
-1.85873941e-01 -1.36295900e-01 6.89031482e-01 7.76269674e-01
-9.86150146e-01 7.47015715e-01 -7.58399725e-01 3.95521671e-01
9.76619899e-01 2.69345224e-01 7.31263995e-01 -1.10434294e+00
-6.65253162e-01 -3.36397469e-01 8.03292245e-02 -1.17185283e+00
-1.41204461e-01 -3.14619869e-01 4.68330644e-02 -1.51492953e+00
3.08359981e-01 -2.34503120e-01 -3.11450630e-01 -8.44395086e-02
-2.87603378e-01 6.04180515e-01 1.86306283e-01 3.13018173e-01
-4.98112798e-01 4.83073592e-01 1.55332804e+00 8.51052254e-02
-3.45398962e-01 -2.48011768e-01 -1.00153625e+00 5.92671692e-01
4.77238804e-01 4.11701798e-02 -5.28063238e-01 -6.58455968e-01
4.42251384e-01 4.46729660e-01 5.17933786e-01 -1.02144694e+00
-2.47190341e-01 8.43608901e-02 8.41929495e-01 -4.44917113e-01
6.85211718e-01 -5.89900076e-01 5.97860515e-01 -9.83185247e-02
-5.07110775e-01 1.52935326e-01 1.11502632e-01 6.06857479e-01
6.95410743e-03 3.46913993e-01 1.09729636e+00 -2.64500529e-01
-6.64718449e-01 3.73181194e-01 2.85036296e-01 1.37392385e-02
2.32557356e-01 -2.65065640e-01 -6.56818390e-01 -3.91672760e-01
-3.59056741e-01 -3.34640115e-01 7.39889741e-01 5.87096095e-01
8.52765739e-01 -1.28141165e+00 -1.00611079e+00 2.23434776e-01
-1.14129260e-01 2.66256630e-01 7.33665049e-01 9.59251449e-02
-4.24187869e-01 -1.17736690e-01 -2.58938402e-01 -5.17126739e-01
-9.68067527e-01 7.14456379e-01 4.49750692e-01 -3.40940833e-01
-1.40140712e+00 6.92614615e-01 5.15747607e-01 -1.51515663e-01
3.25841814e-01 -2.61459857e-01 -1.23503208e-01 -5.25516510e-01
1.00972795e+00 5.34262538e-01 2.98142247e-02 -4.12449270e-01
2.29754284e-01 7.16509044e-01 -1.89872801e-01 6.64330870e-02
1.56150949e+00 -3.23649824e-01 1.21675409e-01 3.61196131e-01
1.15661967e+00 -2.74769254e-02 -1.90984154e+00 -2.16893256e-01
-5.16277313e-01 -9.09122944e-01 5.22017241e-01 -9.62934792e-01
-1.29073441e+00 5.81208706e-01 8.37893307e-01 -5.58824465e-02
1.55204320e+00 -2.63319910e-01 9.97001231e-01 1.68660313e-01
3.33723128e-01 -7.00573146e-01 -3.23249735e-02 2.85157174e-01
1.25091577e+00 -9.78159010e-01 6.47248793e-03 -4.51637059e-01
-5.14956594e-01 9.62648809e-01 5.79251945e-01 -3.99876654e-01
1.68884560e-01 5.80370426e-01 8.49592164e-02 -1.15508698e-02
-1.01357996e+00 1.44501731e-01 2.58849114e-01 6.93164229e-01
4.43695396e-01 -1.18526272e-01 2.31375575e-01 2.41767570e-01
-5.11578202e-01 1.67719960e-01 5.82660496e-01 2.85290539e-01
-1.63200676e-01 -5.40636182e-01 -3.63276005e-01 3.63205135e-01
-4.30501372e-01 -3.84264827e-01 2.17469037e-01 5.40312648e-01
1.86968185e-02 5.29892147e-01 2.49976188e-01 -4.18027580e-01
6.37825727e-01 -4.92965996e-01 6.62528336e-01 -8.84662792e-02
-6.32080257e-01 1.20650589e-01 -1.38838850e-02 -1.06398296e+00
-3.36076975e-01 -8.56833011e-02 -7.41741538e-01 -5.65711379e-01
1.90092385e-01 -3.10488373e-01 4.31825280e-01 3.29515755e-01
4.98909980e-01 1.21251929e+00 7.17803240e-01 -1.20687783e+00
-5.32550097e-01 -7.01339543e-01 -3.23183715e-01 4.17761743e-01
7.62563109e-01 -3.67040277e-01 -3.82816523e-01 -1.33146986e-01]
|
[10.939817428588867, -2.083038568496704]
|
f55502dc-1b4a-45a8-b056-985b84ed6188
|
post-hoc-selection-of-pareto-optimal
|
2306.12165
| null |
https://arxiv.org/abs/2306.12165v1
|
https://arxiv.org/pdf/2306.12165v1.pdf
|
Post-hoc Selection of Pareto-Optimal Solutions in Search and Recommendation
|
Information Retrieval (IR) and Recommender Systems (RS) tasks are moving from computing a ranking of final results based on a single metric to multi-objective problems. Solving these problems leads to a set of Pareto-optimal solutions, known as Pareto frontier, in which no objective can be further improved without hurting the others. In principle, all the points on the Pareto frontier are potential candidates to represent the best model selected with respect to the combination of two, or more, metrics. To our knowledge, there are no well-recognized strategies to decide which point should be selected on the frontier. In this paper, we propose a novel, post-hoc, theoretically-justified technique, named "Population Distance from Utopia" (PDU), to identify and select the one-best Pareto-optimal solution from the frontier. In detail, PDU analyzes the distribution of the points by investigating how far each point is from its utopia point (the ideal performance for the objectives). The possibility of considering fine-grained utopia points allows PDU to select solutions tailored to individual user preferences, a novel feature we call "calibration". We compare PDU against existing state-of-the-art strategies through extensive experiments on tasks from both IR and RS. Experimental results show that PDU and combined with calibration notably impact the solution selection. Furthermore, the results show that the proposed framework selects a solution in a principled way, irrespective of its position on the frontier, thus overcoming the limits of other strategies.
|
['Tommaso Di Noia', 'Raffaele Perego', 'Franco Maria Nardini', 'Vito Walter Anelli', 'Vincenzo Paparella']
|
2023-06-21
| null | null | null | null |
['information-retrieval']
|
['natural-language-processing']
|
[ 6.01233691e-02 -2.05076292e-01 -2.12186173e-01 -2.10007615e-02
-7.76071191e-01 -6.98263109e-01 2.49113441e-01 3.64159763e-01
-3.87764812e-01 5.76298773e-01 1.52955875e-01 -9.75613073e-02
-1.21730554e+00 -7.40787089e-01 -4.47031796e-01 -8.62283170e-01
7.05416203e-02 7.62044549e-01 -5.31183509e-03 -3.65076751e-01
6.74627483e-01 5.09036243e-01 -2.20999455e+00 3.79478604e-01
1.32501960e+00 9.92909968e-01 5.07104456e-01 1.40320674e-01
-2.94623584e-01 7.43149167e-06 -6.43280029e-01 -2.72185951e-01
3.44739318e-01 -2.54799247e-01 -7.50977576e-01 -4.35654134e-01
1.12068504e-01 9.98650566e-02 6.40457869e-01 1.06513321e+00
5.57509542e-01 3.58442247e-01 7.74146438e-01 -9.81548905e-01
-5.65584362e-01 7.12424636e-01 -3.27255815e-01 -8.84455964e-02
5.29048443e-01 -4.47804213e-01 1.41338408e+00 -6.33027315e-01
6.17041886e-01 1.16118109e+00 2.75370419e-01 3.39392275e-01
-1.24318039e+00 -1.76779285e-01 1.60061508e-01 1.08589150e-01
-1.57697475e+00 -1.01178847e-01 7.44802117e-01 -4.17848825e-01
3.96180987e-01 8.33824456e-01 3.49102676e-01 5.46181500e-01
-2.69833114e-02 7.13846624e-01 1.03123665e+00 -6.56100571e-01
5.67855299e-01 3.73189360e-01 2.19160467e-01 1.87888071e-01
4.24353004e-01 2.82176584e-02 -3.90750110e-01 -4.12629932e-01
1.27230346e-01 -1.35814145e-01 -5.59192181e-01 -7.46265113e-01
-9.05525565e-01 6.95634186e-01 4.30375129e-01 6.55409336e-01
-5.93953848e-01 -4.54737753e-01 -9.94506776e-02 2.13019073e-01
2.53806710e-01 9.78345573e-01 -5.56950510e-01 1.50919527e-01
-7.38253355e-01 2.27843970e-01 6.80607319e-01 5.40190101e-01
9.16524529e-01 -5.10583937e-01 -7.63880491e-01 8.42472672e-01
3.56448114e-01 2.26419792e-01 5.91137648e-01 -8.44444335e-01
3.21341336e-01 1.04650939e+00 5.63265443e-01 -9.97179031e-01
-4.09167200e-01 -1.03204989e+00 -2.70808071e-01 2.34672353e-01
5.11778414e-01 -1.47584826e-01 -3.64931375e-01 1.62829745e+00
2.97879428e-01 -2.33962119e-01 9.08301324e-02 1.23458958e+00
6.29411578e-01 5.90770662e-01 -1.69256419e-01 -4.39955682e-01
1.07017350e+00 -4.29950088e-01 -3.11229944e-01 9.23209339e-02
6.30435169e-01 -6.86342597e-01 1.09254420e+00 7.17057765e-01
-8.21041703e-01 -3.29795927e-01 -8.17632496e-01 4.38067764e-01
-2.37840459e-01 3.05523574e-01 2.79841930e-01 8.68643165e-01
-1.10385406e+00 7.99074054e-01 -6.81623966e-02 -3.23697925e-01
-1.75331995e-01 5.11274278e-01 7.41249248e-02 1.14793941e-01
-1.16483939e+00 9.30873513e-01 4.66371328e-01 -5.09806387e-02
-4.20672655e-01 -6.98005319e-01 -1.15244471e-01 5.49972832e-01
6.41692698e-01 -7.82323062e-01 7.85874784e-01 -1.29631615e+00
-1.62190831e+00 4.67617422e-01 -1.84979346e-02 -4.18551564e-01
6.36874616e-01 -2.01810077e-01 -4.51164782e-01 -9.51729044e-02
-1.84814110e-01 2.85413802e-01 6.66004181e-01 -1.57706964e+00
-9.58217621e-01 -3.43866140e-01 3.47808480e-01 5.10441184e-01
-6.00776494e-01 2.05971807e-01 -4.36790466e-01 -6.98866099e-02
-1.92006270e-03 -8.31213117e-01 -1.10781051e-01 -4.62891817e-01
-2.48996630e-01 -5.22531688e-01 1.72745302e-01 -1.87289149e-01
1.79338837e+00 -1.79242945e+00 4.54382509e-01 7.17777193e-01
-2.60584466e-02 4.08742905e-01 -1.36586905e-01 5.58493793e-01
2.42752030e-01 1.24970056e-01 1.54666945e-01 8.46539065e-02
2.52548009e-01 -1.13588393e-01 -2.66425125e-02 2.37784863e-01
-3.79996657e-01 4.53293949e-01 -1.03090417e+00 -2.19247699e-01
1.07369855e-01 3.35850596e-01 -5.41567981e-01 -2.07217652e-02
-2.66778022e-01 4.42580223e-01 -7.24181354e-01 4.54137266e-01
5.60205817e-01 -1.98317677e-01 3.56123894e-01 -1.17435254e-01
-4.11547899e-01 1.24885626e-02 -1.51132572e+00 1.22673547e+00
-5.17471015e-01 1.99349169e-02 -2.94572473e-01 -8.87587905e-01
1.33405185e+00 8.97632539e-03 6.75746620e-01 -7.94024706e-01
3.84569466e-01 5.43894947e-01 1.29231557e-01 -2.63471305e-01
6.45484090e-01 2.00141102e-01 6.74739182e-02 2.73274690e-01
-2.88341492e-01 5.58523953e-01 4.53046024e-01 -3.73475641e-01
8.25061321e-01 1.52625188e-01 4.03834432e-01 -6.95649505e-01
9.06211495e-01 -1.11445472e-01 4.87187803e-01 8.70802164e-01
-7.25307362e-03 6.41676843e-01 4.52961385e-01 -1.74138829e-01
-4.48836327e-01 -9.02714372e-01 -4.58537824e-02 1.20635545e+00
4.50395167e-01 -2.83423692e-01 -7.09087670e-01 -5.81755221e-01
5.35070747e-02 9.42083836e-01 -5.57363331e-01 1.41840670e-02
-2.59465814e-01 -5.35034657e-01 -1.14947200e-01 -2.91499078e-01
-2.50335075e-02 -9.65815723e-01 -8.96808386e-01 3.15153040e-02
-3.70943159e-01 -3.91562760e-01 -1.44394875e-01 -2.61645745e-02
-8.09860706e-01 -1.01340318e+00 -1.00392914e+00 -1.70340538e-01
6.09145343e-01 4.89615649e-01 1.12472880e+00 1.46566510e-01
2.78185010e-01 3.78232956e-01 -7.65152991e-01 -3.72155905e-01
-2.84723282e-01 2.37829477e-01 -1.14502441e-02 5.46491563e-01
2.85934806e-01 -1.73424572e-01 -6.43948197e-01 6.40542805e-01
-7.20855832e-01 -2.71636903e-01 6.69405043e-01 2.63486952e-01
8.16125631e-01 4.41702187e-01 6.03131533e-01 -7.95388937e-01
9.18698192e-01 -6.16524279e-01 -6.86172783e-01 8.92937541e-01
-9.59052980e-01 5.12493551e-01 7.73318708e-01 -2.00920507e-01
-1.00984561e+00 -1.12364255e-01 2.51506686e-01 -3.26633424e-01
-7.92167112e-02 7.79772699e-01 -2.74447978e-01 -2.43029948e-02
9.11546886e-01 3.10323574e-02 -2.33754992e-01 -7.20560253e-01
2.85965115e-01 6.11629963e-01 7.71275684e-02 -7.83075273e-01
4.98068362e-01 3.89195055e-01 -7.86230341e-03 -4.96531665e-01
-8.68393421e-01 -7.18341589e-01 -5.18540621e-01 -4.02053386e-01
4.26122338e-01 -2.94911623e-01 -8.64107609e-01 -2.22447067e-01
-8.22566688e-01 2.87728161e-01 -1.28694773e-01 4.56202477e-01
-5.24303377e-01 3.66905123e-01 4.23648208e-01 -1.10714459e+00
-5.35653830e-01 -1.13694668e+00 6.87157512e-01 4.90701914e-01
-3.37203681e-01 -7.78311014e-01 1.14092961e-01 2.86548853e-01
2.80585468e-01 1.33241519e-01 9.61588919e-01 -7.08209217e-01
-4.30892587e-01 -1.15676597e-01 7.35627860e-02 6.56983927e-02
-2.57182065e-02 9.54950228e-02 -5.88194728e-01 -3.82475346e-01
-1.55977413e-01 4.16199267e-01 7.50535548e-01 6.23401403e-01
8.02515507e-01 -2.54468620e-01 -4.09743458e-01 4.35539007e-01
1.60245371e+00 3.40083927e-01 6.17984474e-01 7.21067548e-01
1.06968522e-01 8.72667015e-01 9.96862054e-01 6.50172949e-01
1.99250862e-01 1.06937182e+00 5.94503582e-01 2.71335065e-01
9.32838619e-02 2.72113979e-02 2.65751272e-01 1.24768436e-01
-3.06444347e-01 -5.48672616e-01 -8.90641749e-01 3.79729837e-01
-2.06404781e+00 -7.59856462e-01 -1.58641860e-01 3.00940824e+00
2.41858169e-01 -5.49354181e-02 2.96927720e-01 2.59405732e-01
9.02366161e-01 -3.60087961e-01 -3.22653830e-01 -4.54923034e-01
-1.51805177e-01 -3.26865464e-02 4.38276887e-01 3.75735164e-01
-9.18318927e-01 4.33142006e-01 5.11409998e+00 9.24621403e-01
-1.10759103e+00 -1.82686076e-01 3.10755670e-01 -2.09766388e-01
-6.86679125e-01 7.18597099e-02 -7.74932444e-01 5.13037503e-01
7.31690764e-01 -5.42639434e-01 6.29288614e-01 6.42941177e-01
3.28888178e-01 -1.11030489e-02 -9.17868137e-01 7.93448627e-01
4.89451224e-03 -1.06135583e+00 2.29421228e-01 2.32747197e-01
8.41473818e-01 -2.82109439e-01 2.07818866e-01 1.04808686e-02
2.46275634e-01 -7.31627762e-01 8.15714240e-01 6.95178986e-01
3.49249601e-01 -1.04356432e+00 7.10911274e-01 6.15164638e-01
-9.08789158e-01 -5.82583666e-01 -3.74444187e-01 2.71799773e-01
-2.14619502e-01 6.48383737e-01 -6.31647527e-01 1.24710667e+00
6.92438900e-01 2.09793612e-01 -4.75974977e-01 1.42796981e+00
-3.54574174e-02 1.00910917e-01 -2.30417341e-01 -3.85422498e-01
2.67920762e-01 -4.92072403e-01 8.38202357e-01 8.29590619e-01
1.07610142e+00 -8.96914750e-02 -2.01052688e-02 7.59656489e-01
3.17282438e-01 7.91697860e-01 -2.82390207e-01 2.79551744e-01
6.01445019e-01 1.16782868e+00 -6.93019986e-01 6.43768981e-02
-1.09308062e-03 3.74902368e-01 2.42036283e-01 3.22183460e-01
-5.41030288e-01 -1.80973068e-01 4.86682355e-01 1.14041001e-01
2.37733632e-01 2.23596215e-01 -4.00236636e-01 -8.90557408e-01
-7.52070397e-02 -6.87865615e-01 7.13972688e-01 -4.91155386e-01
-1.10748529e+00 7.42197096e-01 7.82700032e-02 -1.72130275e+00
-1.19574204e-01 -4.15539533e-01 -3.77392828e-01 8.57694864e-01
-1.54745293e+00 -5.47750831e-01 -2.50887811e-01 3.48293692e-01
-2.21534487e-04 -1.99971050e-01 6.72432244e-01 2.70704627e-01
-2.71261871e-01 4.33836550e-01 7.79509783e-01 -7.24370897e-01
6.26671433e-01 -1.11544275e+00 -6.93562210e-01 6.98771119e-01
2.13471711e-01 7.00451434e-01 9.93902266e-01 -3.97611171e-01
-1.20830321e+00 -5.26859581e-01 1.03149807e+00 -2.64680058e-01
2.59254068e-01 1.63330406e-01 -8.06685030e-01 -7.46280178e-02
-1.86134905e-01 -6.51841283e-01 7.36903608e-01 5.17879486e-01
-1.28044188e-01 -4.42703754e-01 -1.12023282e+00 6.17097020e-01
7.91659474e-01 2.12338120e-01 -3.70997250e-01 1.93975940e-01
2.47792110e-01 -8.47900286e-02 -7.13731289e-01 4.66211647e-01
8.56816530e-01 -1.31497633e+00 1.16579425e+00 -2.80362159e-01
2.90583730e-01 -4.18828517e-01 -2.13572279e-01 -1.58069539e+00
-5.47225654e-01 -6.22732997e-01 -3.76590826e-02 1.22110581e+00
4.63144034e-01 -6.99590862e-01 5.66619635e-01 5.45236826e-01
-8.39712098e-02 -8.50329220e-01 -8.83193076e-01 -9.62441623e-01
-2.66773161e-02 -9.49511081e-02 1.02416635e+00 6.78923547e-01
-2.15931669e-01 1.22541182e-01 -3.06642562e-01 1.46943659e-01
4.92946595e-01 7.41429090e-01 5.78425884e-01 -1.79379773e+00
-3.67964327e-01 -1.08846629e+00 9.88833234e-03 -7.74412215e-01
-6.01691715e-02 -9.55402732e-01 -4.40915786e-02 -1.73616099e+00
-1.07624128e-01 -6.81257188e-01 -9.50685978e-01 -2.30413098e-02
-1.54938698e-01 -3.08125526e-01 4.98233974e-01 5.17301857e-01
-7.11457193e-01 2.53011078e-01 1.12715554e+00 1.64838225e-01
-7.58954823e-01 3.65189582e-01 -1.02172089e+00 5.62187254e-01
7.26141155e-01 -4.89279568e-01 -5.03615260e-01 -3.24652225e-01
5.60013235e-01 7.21255541e-02 -2.42869839e-01 -8.92925262e-01
2.29524344e-01 -4.42044228e-01 1.41975088e-02 -4.61561859e-01
8.23559761e-02 -1.02789152e+00 4.76666629e-01 2.79838473e-01
-4.74791974e-01 -4.27486241e-01 -2.36143917e-01 3.63837659e-01
4.27931733e-02 -7.60107100e-01 5.72777271e-01 5.56186289e-02
-6.57703161e-01 -9.63745192e-02 2.37135775e-02 -2.99113691e-01
1.12371743e+00 -3.03068042e-01 -3.70616049e-01 -2.35113353e-01
-4.49488431e-01 3.28789949e-01 3.81796598e-01 5.51118612e-01
2.82269359e-01 -1.01935935e+00 -8.23255062e-01 -2.15031147e-01
2.58364797e-01 -5.09442985e-01 3.45885545e-01 7.59993136e-01
-2.37443388e-01 7.01761007e-01 -2.44910225e-01 -3.65384430e-01
-1.33530557e+00 6.97880983e-01 2.61663944e-01 -5.35330296e-01
-2.21377984e-01 4.94287223e-01 1.79735988e-01 -3.58467042e-01
1.50634542e-01 -6.53124824e-02 -9.45822120e-01 4.91903573e-01
4.02208775e-01 7.05059171e-01 3.34532052e-01 -6.61389947e-01
-4.03007954e-01 6.84243023e-01 3.25619578e-01 1.35039940e-01
1.20012128e+00 -2.22017482e-01 -7.62967765e-02 1.27929062e-01
8.13087285e-01 2.33201474e-01 -8.25901985e-01 -3.00161004e-01
3.70646834e-01 -7.39453495e-01 9.83574092e-02 -1.25774133e+00
-8.95910919e-01 4.42152739e-01 6.20060027e-01 5.89898705e-01
1.44913960e+00 -2.77708769e-01 3.72985303e-01 4.16247964e-01
6.25032604e-01 -1.15344560e+00 -2.45293319e-01 3.22178870e-01
1.00760269e+00 -8.77190292e-01 -1.46936014e-01 -3.15789521e-01
-6.76615775e-01 1.11350083e+00 3.45603108e-01 -5.60123026e-02
3.80258054e-01 -3.55045736e-01 -2.83216927e-02 1.28606230e-01
-6.11082196e-01 -6.03167176e-01 7.96016932e-01 4.22661126e-01
3.33445668e-01 3.41392308e-01 -9.75389779e-01 7.22752094e-01
-1.38860106e-01 -1.78663898e-02 1.91293806e-01 4.86200571e-01
-8.43323767e-01 -1.30224967e+00 -7.51929224e-01 4.28075641e-01
-3.22660208e-01 1.02345496e-01 -4.19200063e-01 5.12245595e-01
1.71289772e-01 1.21798968e+00 -1.86537951e-01 -3.28392327e-01
4.92259055e-01 -5.61488420e-02 3.52879405e-01 -3.77106637e-01
-8.39953363e-01 1.36727333e-01 1.21361479e-01 -3.22807044e-01
-2.21787900e-01 -6.17173016e-01 -1.20913553e+00 -1.04964249e-01
-4.13950562e-01 4.52742040e-01 6.85910225e-01 9.36720729e-01
6.75402999e-01 3.76411587e-01 8.63376021e-01 -6.17751241e-01
-6.94540381e-01 -4.96663511e-01 -4.74788576e-01 4.61742640e-01
-7.80323744e-02 -9.14745867e-01 -2.51738667e-01 -6.79544389e-01]
|
[9.609007835388184, 5.586344242095947]
|
b6749c34-c596-41ec-a8e8-4c97d6d7c39f
|
aenet-learning-deep-audio-features-for-video
|
1701.00599
| null |
http://arxiv.org/abs/1701.00599v2
|
http://arxiv.org/pdf/1701.00599v2.pdf
|
AENet: Learning Deep Audio Features for Video Analysis
|
We propose a new deep network for audio event recognition, called AENet. In
contrast to speech, sounds coming from audio events may be produced by a wide
variety of sources. Furthermore, distinguishing them often requires analyzing
an extended time period due to the lack of clear sub-word units that are
present in speech. In order to incorporate this long-time frequency structure
of audio events, we introduce a convolutional neural network (CNN) operating on
a large temporal input. In contrast to previous works this allows us to train
an audio event detection system end-to-end. The combination of our network
architecture and a novel data augmentation outperforms previous methods for
audio event detection by 16%. Furthermore, we perform transfer learning and
show that our model learnt generic audio features, similar to the way CNNs
learn generic features on vision tasks. In video analysis, combining visual
features and traditional audio features such as MFCC typically only leads to
marginal improvements. Instead, combining visual features with our AENet
features, which can be computed efficiently on a GPU, leads to significant
performance improvements on action recognition and video highlight detection.
In video highlight detection, our audio features improve the performance by
more than 8% over visual features alone.
|
['Luc van Gool', 'Naoya Takahashi', 'Michael Gygli']
|
2017-01-03
| null | null | null | null |
['highlight-detection']
|
['computer-vision']
|
[ 2.49994427e-01 -3.15617383e-01 2.83330292e-01 -2.58020945e-02
-8.78501594e-01 -5.00674129e-01 4.12266165e-01 4.42066103e-01
-6.10517204e-01 3.90076071e-01 3.93029809e-01 1.12028509e-01
2.29747251e-01 -6.49930060e-01 -8.50321174e-01 -4.46396887e-01
-3.55817080e-01 -3.03322762e-01 5.05217373e-01 1.03896305e-01
5.75127341e-02 6.51113212e-01 -1.93124807e+00 6.65951550e-01
-1.26185431e-03 1.47440994e+00 1.26980767e-01 1.19376230e+00
-2.44431105e-02 8.73812079e-01 -8.01696420e-01 1.95427418e-01
9.17680189e-02 -2.71376401e-01 -6.45735979e-01 1.88059941e-01
6.42009735e-01 -6.30561113e-01 -6.71787858e-01 6.69159234e-01
7.62390971e-01 3.46833616e-01 3.60757202e-01 -1.42022777e+00
-1.45789057e-01 5.77156126e-01 -3.10257822e-01 5.12720168e-01
5.00453770e-01 2.55253553e-01 1.02053380e+00 -8.48822653e-01
4.71304238e-01 1.13453782e+00 9.32402849e-01 1.79192379e-01
-9.90577996e-01 -5.77793658e-01 2.08198130e-01 6.89757407e-01
-1.25400460e+00 -4.61932391e-01 8.68873179e-01 -2.61946827e-01
1.24896014e+00 1.85122132e-01 8.79344702e-01 1.20844924e+00
-1.41282946e-01 9.18632746e-01 5.96550584e-01 -3.94910157e-01
8.24796259e-02 -3.16141695e-01 -1.94674447e-01 3.87116641e-01
-4.00030911e-01 1.67280249e-02 -8.57268929e-01 -7.71568666e-05
6.06119633e-01 2.08602548e-01 -4.04766321e-01 8.63163099e-02
-1.31108379e+00 6.27394497e-01 2.93233514e-01 2.12232456e-01
-4.09107178e-01 7.39346743e-01 1.04511452e+00 5.45373380e-01
3.19153786e-01 2.07467392e-01 -5.69212496e-01 -5.37524462e-01
-1.11893439e+00 1.23161726e-01 6.16777122e-01 4.76957977e-01
5.64643621e-01 4.83825266e-01 -2.73915410e-01 7.56270289e-01
-1.21364407e-01 4.04634774e-01 4.36490923e-01 -9.77313221e-01
1.56873539e-01 1.87275991e-01 -1.88861430e-01 -8.38293374e-01
-5.25521040e-01 -4.29959267e-01 -6.46664619e-01 3.57423455e-01
4.96413946e-01 -1.80073023e-01 -8.96475673e-01 1.64627099e+00
1.43010110e-01 5.92510581e-01 -1.82618603e-01 8.79487693e-01
8.88434827e-01 7.56640255e-01 -5.40475957e-02 -5.55304214e-02
1.52332294e+00 -7.08585382e-01 -7.46430755e-01 1.85599878e-01
3.46593946e-01 -9.55796480e-01 1.02802253e+00 6.99420154e-01
-1.07799137e+00 -7.36410081e-01 -1.06271541e+00 -2.05130681e-01
-5.94663799e-01 1.85543075e-01 5.37037790e-01 4.29048240e-01
-1.05242169e+00 6.94852531e-01 -8.62619936e-01 -4.30908769e-01
4.17179346e-01 3.65759254e-01 -2.47040614e-01 3.78772199e-01
-1.14603424e+00 4.70181614e-01 3.66238207e-01 -1.82128042e-01
-1.21660161e+00 -8.09873581e-01 -9.62146401e-01 4.27585632e-01
5.56446850e-01 -3.73131484e-01 1.54037869e+00 -1.08992553e+00
-1.53676951e+00 3.97957385e-01 -5.03491536e-02 -9.35016274e-01
3.74924511e-01 -3.16107571e-01 -5.01949787e-01 7.19498634e-01
-3.18868458e-01 8.59434724e-01 1.14368975e+00 -4.78027314e-01
-8.45463693e-01 3.53099614e-01 2.52900362e-01 -9.95159373e-02
-5.32262087e-01 3.89568657e-01 -4.85910982e-01 -1.04710114e+00
-4.52412844e-01 -7.97615647e-01 1.08898997e-01 3.53924274e-01
-1.09891713e-01 -1.15806818e-01 1.18748188e+00 -6.07774854e-01
1.05374599e+00 -2.48076725e+00 -1.12350516e-01 -8.11931584e-03
2.92091638e-01 2.77536571e-01 -3.10462028e-01 3.69347215e-01
-3.30498010e-01 -2.24737987e-01 -6.94981813e-02 -3.06079566e-01
1.04852103e-01 3.36911646e-03 -5.39771318e-01 2.12809071e-01
5.40336728e-01 7.10930109e-01 -7.93259561e-01 -2.86746860e-01
4.00640666e-01 8.91744494e-01 -6.41000569e-01 -2.00567581e-02
-1.19533509e-01 2.17649624e-01 5.88140488e-02 5.71139634e-01
3.10078889e-01 -1.36623710e-01 -1.78816050e-01 -2.24626213e-01
-4.98558730e-02 5.82248628e-01 -1.40539932e+00 1.74646306e+00
-7.11159289e-01 1.22618544e+00 3.33289318e-02 -1.07127810e+00
5.11000037e-01 8.01095903e-01 5.73244691e-01 -4.94920105e-01
6.66290969e-02 2.31069159e-02 -5.83415627e-02 -3.77866387e-01
4.46592301e-01 1.87401205e-01 8.70071445e-03 4.35795754e-01
4.02188599e-01 5.04558440e-03 2.62984812e-01 1.77919835e-01
1.49390221e+00 -1.62140459e-01 1.30849153e-01 2.93546468e-01
4.50908840e-01 -3.24616909e-01 4.75392848e-01 6.14772379e-01
-1.22605972e-01 8.15857470e-01 5.08081973e-01 -5.55557668e-01
-8.43191922e-01 -1.09894407e+00 5.39757460e-02 1.47991920e+00
-3.77889574e-01 -9.30221260e-01 -5.74628294e-01 -5.93759298e-01
-1.13777786e-01 1.14120968e-01 -4.58489388e-01 -1.89304724e-01
-5.82822323e-01 -2.95773327e-01 9.33367848e-01 9.24972713e-01
5.12305081e-01 -1.29401529e+00 -9.60455358e-01 5.77767551e-01
-4.07155417e-02 -1.35416913e+00 -5.68773746e-01 4.94535625e-01
-6.60628080e-01 -1.03152788e+00 -6.33400798e-01 -5.63932121e-01
-8.77951756e-02 1.47450760e-01 1.13769257e+00 -6.01667166e-02
-8.09994400e-01 7.83844650e-01 -4.81049269e-01 -7.06433177e-01
-1.71239525e-01 -1.64009213e-01 2.52715014e-02 1.25890687e-01
2.79288411e-01 -8.98394167e-01 -7.39400983e-01 -2.06806824e-01
-1.10010421e+00 -2.72372574e-01 4.12841618e-01 7.74401903e-01
4.84146744e-01 1.50287198e-02 6.10637426e-01 -2.67351568e-01
3.44116569e-01 -2.81183124e-01 -3.47950608e-01 -2.23709345e-01
3.61479521e-02 -2.66321957e-01 5.37578821e-01 -9.29450631e-01
-6.04298234e-01 3.72037888e-01 -3.27769846e-01 -8.08579803e-01
-5.20803273e-01 3.82780671e-01 1.81314185e-01 4.96359542e-02
4.86684948e-01 8.77069831e-02 -1.45672709e-01 -5.41078687e-01
2.21934006e-01 4.92199212e-01 6.03780687e-01 -1.83912247e-01
4.99049634e-01 6.05028689e-01 5.64822331e-02 -1.08933485e+00
-5.96836686e-01 -4.34732884e-01 -4.35393631e-01 -3.09397638e-01
9.50563431e-01 -1.15890658e+00 -9.41117764e-01 5.17400146e-01
-1.22544217e+00 -3.05183679e-01 -5.94336987e-01 7.43811071e-01
-5.79256296e-01 3.46135646e-01 -8.94367993e-01 -7.85973966e-01
-2.14547768e-01 -9.54554498e-01 1.29435658e+00 -9.93870050e-02
-3.81285548e-01 -6.78139746e-01 -1.17659897e-01 -3.39585632e-01
4.98793870e-01 3.84512454e-01 5.31501234e-01 -5.11493683e-01
-4.86860245e-01 -7.91220739e-02 -1.66061148e-01 4.82745588e-01
-2.38790307e-02 1.95286959e-01 -1.37648857e+00 -1.59710109e-01
-1.55214757e-01 -3.81427646e-01 1.40145230e+00 5.00417888e-01
1.53963828e+00 3.37598938e-03 -4.75643203e-02 5.91217041e-01
9.88144100e-01 2.06093356e-01 4.75861073e-01 2.15453446e-01
6.17083192e-01 2.37025961e-01 3.89756620e-01 8.05760384e-01
3.67433997e-03 8.18440557e-01 6.51760757e-01 -3.80228311e-01
-5.74291825e-01 1.00503124e-01 6.84840560e-01 6.43547177e-01
-9.43085924e-02 -3.22926641e-02 -7.88967311e-01 7.51261294e-01
-1.68390071e+00 -1.20561552e+00 3.26745957e-02 1.95967972e+00
7.41821170e-01 1.42817453e-01 4.64144915e-01 8.37476969e-01
6.52113259e-01 1.46536469e-01 -3.61177206e-01 -5.08139551e-01
-2.28096858e-01 7.78445482e-01 2.37642527e-01 1.17832124e-01
-1.39560127e+00 5.38263023e-01 6.51043463e+00 8.87439668e-01
-1.65891802e+00 2.45331153e-01 1.65417701e-01 -7.14331985e-01
1.95784092e-01 -3.66018534e-01 -3.97993803e-01 3.37593615e-01
1.32541466e+00 1.36620581e-01 2.36205012e-01 7.25655258e-01
1.69729665e-01 4.56284583e-02 -1.23150063e+00 1.40237868e+00
5.53184450e-02 -1.39448583e+00 -8.65262225e-02 -8.02094191e-02
1.86956629e-01 1.48964778e-01 1.73654616e-01 3.43362570e-01
-4.97244075e-02 -9.24173355e-01 8.44759047e-01 2.57129341e-01
7.26427972e-01 -1.05205381e+00 4.26542014e-01 -2.05815315e-01
-1.64714396e+00 -1.59489825e-01 -1.87809825e-01 -1.94554925e-01
1.52535766e-01 6.58963025e-01 -8.26267779e-01 2.98297912e-01
1.19564712e+00 7.91733682e-01 -4.31787670e-01 1.28167546e+00
-1.14346839e-01 9.03100014e-01 -5.02937675e-01 2.21582010e-01
3.45518917e-01 5.68246365e-01 7.41843402e-01 1.73774779e+00
4.42900032e-01 -3.53354007e-01 1.47627845e-01 4.52200264e-01
-1.91068217e-01 -7.45837018e-02 -6.03648901e-01 -7.94612393e-02
1.15238734e-01 1.31293809e+00 -7.45747328e-01 -4.60807651e-01
-6.36266232e-01 1.07057452e+00 -4.77406234e-02 1.95905522e-01
-1.05623770e+00 -9.36871171e-01 8.61858249e-01 -8.37425962e-02
8.75814319e-01 -3.04580241e-01 3.33172560e-01 -9.50338840e-01
7.38308802e-02 -8.61715376e-01 4.23917383e-01 -7.70952880e-01
-1.00133777e+00 6.37864649e-01 -2.96918571e-01 -1.54359269e+00
-5.02480090e-01 -7.08326936e-01 -7.69930363e-01 5.04999936e-01
-1.62091827e+00 -9.81441498e-01 -2.87109703e-01 9.18546379e-01
7.50041068e-01 -9.27966554e-03 8.67366791e-01 5.88490248e-01
-1.55749127e-01 5.41589379e-01 -2.12100238e-01 4.10784632e-01
9.98924017e-01 -1.29851580e+00 4.27800685e-01 7.05655158e-01
6.87356591e-01 1.14891969e-01 5.10576963e-01 -2.42546171e-01
-1.24305570e+00 -1.21240008e+00 6.11440599e-01 -1.32123798e-01
8.30493450e-01 -4.44849283e-01 -9.79269624e-01 5.23017585e-01
5.10519028e-01 3.66095722e-01 7.85371304e-01 1.80841953e-01
-7.00785398e-01 -2.02802151e-01 -7.20147669e-01 3.75710875e-01
9.52499807e-01 -1.04348683e+00 -4.81010079e-01 2.61014044e-01
7.82963157e-01 -3.96029383e-01 -6.79259717e-01 2.81812459e-01
4.85877782e-01 -6.71162605e-01 1.12976098e+00 -6.05758131e-01
3.47973973e-01 -4.38210070e-01 -1.09930731e-01 -1.17709827e+00
-9.00973938e-03 -7.48004675e-01 -4.52634722e-01 1.17186117e+00
9.61855426e-02 -1.84746042e-01 3.97709548e-01 -2.02443123e-01
-2.01323092e-01 -3.09909374e-01 -9.89076793e-01 -9.62512612e-01
-4.42785680e-01 -1.02910340e+00 3.98894608e-01 8.36808205e-01
-5.31113967e-02 1.78477570e-01 -4.39336330e-01 2.36775652e-01
2.69501597e-01 1.72016397e-01 4.83073175e-01 -1.36653364e+00
-5.09799898e-01 -4.23226893e-01 -7.78641462e-01 -8.14903677e-01
1.61179721e-01 -7.52321064e-01 -3.76854278e-02 -1.12880373e+00
-9.47541967e-02 1.52521297e-01 -7.58610368e-01 6.12598836e-01
1.43388107e-01 8.22278082e-01 3.60836715e-01 -3.38419676e-01
-7.34589458e-01 3.50273907e-01 7.05777228e-01 -2.70609021e-01
-2.55509079e-01 -2.96668321e-01 -9.74486545e-02 1.02677023e+00
7.03399181e-01 -4.53606606e-01 -8.10565278e-02 -2.95126617e-01
2.78738856e-01 4.44397107e-02 8.25900376e-01 -1.42040873e+00
3.00982445e-01 3.99503142e-01 7.12992072e-01 -7.64781713e-01
7.69420862e-01 -7.90067494e-01 -1.41137898e-01 3.71772557e-01
-3.13130558e-01 3.45580541e-02 7.32283711e-01 6.77535534e-01
-7.58042514e-01 9.63401720e-02 4.81732398e-01 -3.06348549e-03
-7.75703728e-01 1.16601340e-01 -9.69658315e-01 -1.48242459e-01
6.39555395e-01 -2.05438808e-01 -1.06827371e-01 -7.08351552e-01
-1.06613481e+00 -1.84103772e-01 -8.53634849e-02 4.77953345e-01
6.78645074e-01 -1.43089640e+00 -5.36633372e-01 -4.18787618e-04
-5.84584773e-02 -3.59431893e-01 4.03147072e-01 9.28815067e-01
-3.65932882e-01 1.71835154e-01 -4.03738678e-01 -7.31654286e-01
-1.65464520e+00 5.60032606e-01 1.23811536e-01 -1.28018344e-02
-8.24973226e-01 7.88990140e-01 2.44612277e-01 2.63186008e-01
6.87384427e-01 -7.80743480e-01 -2.04720601e-01 4.39091682e-01
8.34724128e-01 3.90887171e-01 2.25636393e-01 -2.84919739e-01
-3.85669440e-01 4.71152276e-01 7.00562596e-02 -2.76860535e-01
1.53685975e+00 9.79623348e-02 2.91569769e-01 6.26602352e-01
1.21654546e+00 1.92874655e-01 -1.32224154e+00 -3.18609476e-01
-2.20493555e-01 -1.19302906e-01 3.41174304e-01 -5.94350934e-01
-1.36383486e+00 1.40060115e+00 8.46051037e-01 3.71880382e-01
1.62208509e+00 -1.87603518e-01 9.46968436e-01 5.41598976e-01
-1.37461171e-01 -1.13140321e+00 4.52539474e-01 6.69546127e-01
9.85764086e-01 -1.10894048e+00 -2.75829822e-01 -2.55551308e-01
-4.55714881e-01 1.50892699e+00 1.81977659e-01 -3.21331322e-01
6.95373297e-01 6.59650266e-01 -3.59032390e-04 2.86961216e-02
-9.11665380e-01 -6.81055665e-01 4.44234937e-01 5.20643950e-01
3.48895550e-01 -2.99051255e-01 3.03392619e-01 4.44766819e-01
-6.57435954e-02 -1.39161512e-01 5.31893373e-01 9.93497252e-01
-3.97744209e-01 -9.93451178e-01 -3.76183540e-01 1.42085046e-01
-1.04777157e+00 -2.62096882e-01 -2.60066122e-01 7.67847598e-01
1.64573357e-01 9.84421551e-01 5.82459867e-01 -3.50026071e-01
3.35066497e-01 3.65067899e-01 4.07305419e-01 -4.45672244e-01
-1.02452648e+00 4.24868613e-01 1.19948424e-02 -8.10567915e-01
-5.10661125e-01 -6.12970531e-01 -1.35566998e+00 -7.22951666e-02
-6.24667741e-02 -1.85434654e-01 6.92981362e-01 7.46176898e-01
4.36700255e-01 1.16644561e+00 3.34759027e-01 -1.03543055e+00
-3.25869620e-02 -9.32459712e-01 -5.59277833e-01 3.49289566e-01
7.77561545e-01 -6.08729720e-01 -3.93692523e-01 3.38152289e-01]
|
[15.192208290100098, 5.234137535095215]
|
e90b1c75-3cd4-480f-96e3-c2e999b677ab
|
making-a-case-for-3d-convolutions-for-object
|
2008.11516
| null |
https://arxiv.org/abs/2008.11516v1
|
https://arxiv.org/pdf/2008.11516v1.pdf
|
Making a Case for 3D Convolutions for Object Segmentation in Videos
|
The task of object segmentation in videos is usually accomplished by processing appearance and motion information separately using standard 2D convolutional networks, followed by a learned fusion of the two sources of information. On the other hand, 3D convolutional networks have been successfully applied for video classification tasks, but have not been leveraged as effectively to problems involving dense per-pixel interpretation of videos compared to their 2D convolutional counterparts and lag behind the aforementioned networks in terms of performance. In this work, we show that 3D CNNs can be effectively applied to dense video prediction tasks such as salient object segmentation. We propose a simple yet effective encoder-decoder network architecture consisting entirely of 3D convolutions that can be trained end-to-end using a standard cross-entropy loss. To this end, we leverage an efficient 3D encoder, and propose a 3D decoder architecture, that comprises novel 3D Global Convolution layers and 3D Refinement modules. Our approach outperforms existing state-of-the-arts by a large margin on the DAVIS'16 Unsupervised, FBMS and ViSal dataset benchmarks in addition to being faster, thus showing that our architecture can efficiently learn expressive spatio-temporal features and produce high quality video segmentation masks. Our code and models will be made publicly available.
|
['Laura Leal-Taixé', 'Aljoša Ošep', 'Sabarinath Mahadevan', 'Bastian Leibe', 'Sebastian Hennen', 'Ali Athar']
|
2020-08-26
| null | null | null | null |
['unsupervised-video-object-segmentation']
|
['computer-vision']
|
[ 4.33662266e-01 2.62395173e-01 -2.87452370e-01 -3.26984465e-01
-6.07237816e-01 -3.85978997e-01 6.17311358e-01 -1.65232748e-01
-5.28429210e-01 2.67021745e-01 9.42086503e-02 -3.01425457e-01
3.12535346e-01 -4.44420308e-01 -1.16589618e+00 -5.04306257e-01
-3.09998274e-01 3.82451624e-01 5.94996870e-01 8.64454582e-02
-2.38843122e-03 4.25469071e-01 -1.51374030e+00 1.77404195e-01
4.36556578e-01 1.46302748e+00 8.89881104e-02 7.49983132e-01
2.19220370e-02 9.31927919e-01 -8.54370371e-02 -2.69530922e-01
7.01773942e-01 -2.10629255e-01 -9.15121317e-01 4.81778890e-01
6.99326694e-01 -8.34404051e-01 -5.20210981e-01 7.58044660e-01
4.09618258e-01 4.31340747e-02 5.95523298e-01 -9.25645709e-01
-4.17194903e-01 3.68758589e-01 -7.67699301e-01 2.77126759e-01
1.91466510e-01 2.15828866e-01 9.08766985e-01 -7.84860253e-01
6.79453254e-01 1.29607594e+00 8.10337842e-01 6.35895312e-01
-1.37034929e+00 -2.63559371e-01 3.91947001e-01 2.38404702e-02
-1.00109065e+00 -3.65442872e-01 8.21935594e-01 -6.14170372e-01
1.26776659e+00 -1.11673817e-01 7.98236012e-01 1.18334568e+00
-1.16238063e-02 1.27115679e+00 7.23084867e-01 -1.44234002e-01
1.57097131e-01 -2.74373114e-01 -6.00683391e-02 9.00840461e-01
5.21967374e-02 -1.42785804e-02 -3.11038554e-01 4.47482079e-01
1.00013316e+00 8.35470259e-02 -1.93083256e-01 -7.05795765e-01
-1.11135590e+00 7.28765845e-01 5.47927082e-01 -2.18468290e-02
-3.78416032e-01 5.69052100e-01 5.30696273e-01 1.18296057e-01
9.02108908e-01 4.79487702e-02 -6.21452630e-01 -5.56919053e-02
-1.22118306e+00 4.13576812e-01 4.06921238e-01 9.31211710e-01
7.65936613e-01 4.80168778e-03 -1.52964935e-01 5.22679985e-01
3.56628060e-01 2.32844472e-01 1.25129551e-01 -1.11961031e+00
3.43236804e-01 5.94895542e-01 3.02833859e-02 -7.25909472e-01
-5.02975643e-01 -3.32969785e-01 -8.29669595e-01 4.37012345e-01
3.52892369e-01 -4.95756306e-02 -1.23505819e+00 1.64986467e+00
4.50063348e-01 5.11424780e-01 -2.54548126e-04 1.23506200e+00
6.53780222e-01 5.24767876e-01 5.28458469e-02 1.49149537e-01
1.18406498e+00 -1.20964944e+00 -3.53940457e-01 -3.90797615e-01
5.82140923e-01 -4.93228287e-01 6.60406411e-01 1.44625619e-01
-1.50128996e+00 -6.23786807e-01 -9.80325460e-01 -5.93428254e-01
-1.36377156e-01 -9.45691168e-02 6.90169632e-01 4.04783636e-01
-1.21546376e+00 7.42496073e-01 -1.29891956e+00 -3.56107891e-01
1.13998210e+00 5.62530100e-01 -3.33273590e-01 7.09046721e-02
-8.82512629e-01 8.50647688e-01 5.24937451e-01 2.11558700e-01
-1.13792896e+00 -8.91879141e-01 -1.14144325e+00 -5.08915149e-02
4.79271322e-01 -1.00022864e+00 1.39609325e+00 -1.27721810e+00
-1.44954491e+00 1.13394582e+00 5.11755496e-02 -9.86841559e-01
6.79928839e-01 -3.50774467e-01 2.80800849e-01 5.21803916e-01
5.54708652e-02 1.13078558e+00 9.47455108e-01 -9.31235969e-01
-7.55822122e-01 -2.37560987e-01 3.51876408e-01 1.49745330e-01
-1.18484341e-01 -4.44038436e-02 -9.09430802e-01 -7.45486319e-01
-9.57662165e-02 -8.34348202e-01 -4.26536411e-01 4.50399429e-01
-2.93381333e-01 -3.89090069e-02 9.78534222e-01 -8.19862902e-01
8.17115247e-01 -2.05803657e+00 4.41134304e-01 -4.56906892e-02
3.88981551e-01 4.16907758e-01 -6.38922602e-02 -1.90573320e-01
7.45136887e-02 7.19693601e-02 -6.30849659e-01 -7.42224693e-01
1.97259620e-01 1.04784355e-01 -1.01240620e-01 5.41331530e-01
8.60451937e-01 1.20331013e+00 -8.26279998e-01 -3.84012789e-01
5.09780407e-01 6.61921918e-01 -9.57187116e-01 3.52007657e-01
-5.33120215e-01 3.75376701e-01 -3.82251203e-01 5.41443110e-01
5.89751959e-01 -4.79171336e-01 -4.94858287e-02 -1.22511648e-01
5.93261421e-02 3.56832206e-01 -7.92832851e-01 2.10555887e+00
-3.04532409e-01 8.12703490e-01 1.69608802e-01 -1.37919569e+00
5.91823995e-01 2.46670201e-01 7.20340788e-01 -7.03549325e-01
2.32475102e-01 1.99890733e-01 -4.70607728e-01 -5.04637301e-01
2.91971028e-01 4.33644615e-02 -3.04212142e-02 1.38411880e-01
3.84562016e-01 -7.87839442e-02 1.73946783e-01 1.10762693e-01
1.04305542e+00 6.43525243e-01 1.27257794e-01 -1.51322648e-01
2.77339578e-01 4.82322462e-02 4.93470639e-01 5.58142900e-01
-1.20631523e-01 8.10737431e-01 7.05374002e-01 -4.98527735e-01
-1.30537629e+00 -9.35225189e-01 4.04561982e-02 7.65739202e-01
3.81230682e-01 -2.62044549e-01 -8.72398674e-01 -9.21341598e-01
9.40034464e-02 3.13145101e-01 -6.90078795e-01 6.31332174e-02
-7.78730631e-01 -3.52315456e-01 5.02404690e-01 8.03484082e-01
6.56797290e-01 -7.20215380e-01 -8.75391304e-01 3.13295841e-01
6.95140138e-02 -1.51937020e+00 -4.79997486e-01 4.02256668e-01
-9.57288146e-01 -1.11531377e+00 -8.99156749e-01 -8.61613750e-01
4.59310949e-01 3.63753170e-01 1.16002166e+00 -4.24368158e-02
-3.47865582e-01 4.70058531e-01 -2.36929014e-01 -2.15555340e-01
-1.68150708e-01 1.95543528e-01 -2.43998528e-01 -7.61978999e-02
2.87673533e-01 -7.01195061e-01 -9.00816679e-01 1.10166691e-01
-1.02509141e+00 5.35257638e-01 6.66105151e-01 6.68944955e-01
7.08689630e-01 -4.27229017e-01 2.62525648e-01 -7.65740573e-01
-2.38884240e-01 -4.69686270e-01 -7.69854307e-01 -1.00359254e-01
-1.39305204e-01 9.61929262e-02 5.16176999e-01 -8.10074806e-02
-8.70132864e-01 4.28668290e-01 -5.62328458e-01 -9.35254633e-01
-4.05544311e-01 1.14863582e-01 -4.82816882e-02 -1.06391139e-01
2.46996894e-01 4.20574844e-02 1.92302912e-01 -7.16614902e-01
3.64731073e-01 4.33779627e-01 5.29539227e-01 -2.10198760e-01
8.34235132e-01 5.67075312e-01 1.40257403e-01 -6.20091915e-01
-1.04934859e+00 -5.28508246e-01 -8.42264950e-01 -1.25319064e-01
1.45317757e+00 -1.33859825e+00 -4.02107567e-01 5.37865043e-01
-1.18078971e+00 -7.23301411e-01 -3.93155903e-01 3.24319541e-01
-9.15261567e-01 4.38253164e-01 -7.12542832e-01 -5.12306035e-01
-1.77872822e-01 -1.28581727e+00 1.50712466e+00 1.66577458e-01
-3.28995325e-02 -1.07916915e+00 -3.79386693e-01 4.49064940e-01
2.84841657e-01 5.71484447e-01 7.85011411e-01 -4.86386508e-01
-9.70206022e-01 3.26594822e-02 -3.74103189e-01 6.34289145e-01
-2.54149944e-01 -1.80861458e-01 -1.05415750e+00 -2.13862613e-01
-2.14913592e-01 -4.85186219e-01 1.35942268e+00 5.92676938e-01
1.34042370e+00 -2.72542596e-01 -2.58897483e-01 9.37743723e-01
1.30233729e+00 -2.07552880e-01 5.07042527e-01 1.43399432e-01
9.28126216e-01 3.46746564e-01 3.67041290e-01 3.14544171e-01
3.52428377e-01 8.12789679e-01 7.60168195e-01 -3.55765313e-01
-4.44453359e-01 -1.73748523e-01 3.80912691e-01 3.31446528e-01
-1.56575501e-01 -2.09749669e-01 -7.25518107e-01 5.90765893e-01
-1.96556795e+00 -1.03913677e+00 5.77090308e-02 1.77247572e+00
7.01346338e-01 4.37400550e-01 3.13518554e-01 1.06308311e-02
4.86780703e-01 3.46408635e-01 -6.00401640e-01 -2.80239373e-01
-1.42436028e-01 3.50191951e-01 6.71084166e-01 4.01689321e-01
-1.73335409e+00 9.91770029e-01 5.85670376e+00 7.52173841e-01
-1.05473495e+00 4.73900959e-02 8.92630339e-01 -2.95305282e-01
-1.75880075e-01 -1.03001408e-01 -6.98918104e-01 2.87508011e-01
6.96328700e-01 3.81073445e-01 1.32466644e-01 9.01444554e-01
2.82846063e-01 -9.72123537e-03 -1.36225235e+00 1.02392435e+00
-3.59572168e-03 -1.64765370e+00 9.95241851e-03 2.92215142e-02
8.70753586e-01 2.09743872e-01 3.24988291e-02 1.31574199e-01
-5.56186549e-02 -9.84394193e-01 1.06332242e+00 2.50410557e-01
6.14992499e-01 -5.57409227e-01 4.95700270e-01 1.36388078e-01
-1.10182285e+00 -6.01062514e-02 -1.70136034e-01 -7.82498568e-02
3.14896762e-01 5.61245859e-01 -6.20993257e-01 5.31075895e-01
8.15622509e-01 1.17868364e+00 -3.84500146e-01 1.14169776e+00
4.56864655e-04 5.95477462e-01 -2.95222193e-01 3.37390393e-01
7.96019793e-01 3.55000384e-02 5.28008699e-01 1.36804855e+00
2.72766352e-01 -5.89856021e-02 2.46465728e-01 8.68951321e-01
-3.79407942e-01 -2.68054307e-01 -4.98458415e-01 1.92493349e-02
-2.41296217e-01 9.59010065e-01 -8.22871387e-01 -4.03133094e-01
-6.93011284e-01 1.18716025e+00 2.60120332e-01 2.94665098e-01
-1.03056347e+00 -1.81957468e-01 9.91905987e-01 1.70192257e-01
1.03690338e+00 -3.55739743e-01 -2.24039316e-01 -1.21457601e+00
8.75570253e-02 -5.05506039e-01 1.16567127e-01 -5.39669812e-01
-1.23939502e+00 2.72685349e-01 -1.18657500e-01 -1.11007249e+00
-2.45951489e-01 -9.36760426e-01 -2.30520919e-01 5.61036468e-01
-1.84920275e+00 -1.09794998e+00 -1.86432630e-01 5.12698174e-01
7.68867850e-01 1.16833173e-01 4.26933169e-01 4.64651704e-01
-5.61784923e-01 3.26911032e-01 -2.45277192e-02 4.34092581e-01
3.68921936e-01 -1.26414132e+00 7.11776912e-01 9.84844744e-01
6.55999035e-02 1.52180213e-02 3.72039557e-01 -4.49809611e-01
-1.48977864e+00 -1.50168419e+00 6.36204422e-01 -3.71803433e-01
4.13222522e-01 -6.11991942e-01 -7.22336352e-01 7.91670740e-01
2.43204340e-01 3.80009085e-01 3.37375015e-01 -2.09207833e-01
-3.17626029e-01 5.15788719e-02 -8.61191034e-01 5.60763955e-01
1.48669052e+00 -2.75571287e-01 -4.32880253e-01 2.73967892e-01
1.01355135e+00 -6.63522243e-01 -7.98282921e-01 5.63798845e-01
3.95993978e-01 -1.10490799e+00 1.18138528e+00 -6.15891278e-01
8.59908879e-01 -3.06544662e-01 -1.25023946e-01 -6.80898547e-01
-2.19973058e-01 -6.50097013e-01 -4.62159365e-01 8.46038461e-01
2.43361041e-01 2.16622446e-02 9.20756340e-01 4.75809008e-01
-5.91188431e-01 -1.00749207e+00 -1.08947277e+00 -5.52522063e-01
-1.41273588e-01 -6.92548215e-01 2.09494919e-01 5.06788135e-01
-5.01398325e-01 2.69775510e-01 -5.36302269e-01 4.53416146e-02
6.73492670e-01 1.62069112e-01 7.31602013e-01 -9.52117562e-01
-3.45392048e-01 -8.04556489e-01 -8.27228069e-01 -1.78887403e+00
2.72148520e-01 -9.02793765e-01 9.79162827e-02 -1.55984986e+00
2.08568275e-01 -2.78784037e-01 -1.18915029e-01 4.92860138e-01
-1.42511919e-01 7.29922235e-01 1.92779660e-01 -4.55544293e-02
-9.72897589e-01 5.73900700e-01 1.24291122e+00 -2.73798227e-01
-9.99404639e-02 -1.24082454e-01 -4.42391247e-01 7.57151484e-01
4.90206361e-01 -1.68080419e-01 -3.85256380e-01 -7.21940100e-01
-8.58108550e-02 -1.39004663e-01 8.91930580e-01 -1.07470298e+00
-1.04802527e-01 1.58461466e-01 7.36609340e-01 -5.89829266e-01
5.06807089e-01 -9.07271445e-01 -2.34516874e-01 3.15305203e-01
-4.02784258e-01 -3.04857880e-01 3.36675167e-01 7.27943182e-01
-1.87803760e-01 1.56692743e-01 7.90844083e-01 -1.78751573e-02
-1.05586135e+00 7.28150606e-01 -3.03614736e-01 2.11682841e-01
1.20860565e+00 -4.55989063e-01 1.46229625e-01 -5.97248077e-02
-6.35105550e-01 2.78238177e-01 7.06957996e-01 3.93120199e-01
6.21659577e-01 -1.21297038e+00 -6.62454784e-01 2.17160612e-01
-2.49319747e-01 4.82143968e-01 4.12419111e-01 1.08365190e+00
-8.20133746e-01 4.59138602e-01 -3.64396095e-01 -1.07415211e+00
-1.01092422e+00 6.08823240e-01 4.29434925e-01 -1.13575704e-01
-9.38836396e-01 8.80946815e-01 4.50787753e-01 -1.90945044e-02
5.36316752e-01 -5.85427046e-01 3.24914120e-02 -3.35207470e-02
3.60612184e-01 -5.85348830e-02 6.44663209e-03 -5.45584977e-01
-3.80807608e-01 7.04944193e-01 -2.07482278e-03 2.13256747e-01
1.63556731e+00 -1.63521990e-01 2.19418913e-01 2.13699937e-01
1.57455575e+00 -6.18376374e-01 -2.10180759e+00 -2.76926816e-01
-3.85269523e-02 -4.63916600e-01 2.84886137e-02 -3.80909979e-01
-1.37129486e+00 1.13785839e+00 4.98968214e-01 5.37938252e-02
1.43691003e+00 1.94015205e-01 1.03831804e+00 1.24340437e-01
-1.07087856e-02 -9.93477941e-01 1.28273025e-01 5.28087854e-01
4.46058512e-01 -1.40698814e+00 -1.97153300e-01 -3.29450548e-01
-5.01940250e-01 1.14971483e+00 5.11088371e-01 -4.05326098e-01
6.09913170e-01 3.19585383e-01 -2.55184203e-01 7.16000423e-02
-7.52387583e-01 -4.51888859e-01 4.84263450e-01 4.95344281e-01
3.65690380e-01 -3.97593915e-01 7.74616376e-02 4.00786400e-01
4.34369624e-01 7.88477287e-02 2.31830075e-01 8.92663419e-01
-3.32161456e-01 -8.33551645e-01 5.71943074e-02 3.78062606e-01
-5.76283813e-01 -1.67170074e-02 -1.32105663e-01 7.93487787e-01
3.11734527e-01 4.76818532e-01 3.77064019e-01 -3.13522875e-01
2.06785306e-01 2.51920875e-02 6.10181808e-01 -5.20114481e-01
-4.70131069e-01 2.08631366e-01 7.99025223e-02 -9.10601795e-01
-8.94095242e-01 -7.17013538e-01 -1.21903837e+00 -1.82814509e-01
1.29708275e-01 -4.08630490e-01 4.69402999e-01 9.35377419e-01
4.33101296e-01 5.48843980e-01 4.31294203e-01 -1.49163592e+00
-4.09906477e-01 -4.62043136e-01 -2.92490602e-01 4.96532679e-01
5.56769073e-01 -6.96758986e-01 -1.10512748e-01 2.93580949e-01]
|
[9.220084190368652, 0.03214976191520691]
|
09844273-82d2-4d37-97d9-a141d6411ccb
|
tackling-universal-properties-of-minimal-trap
|
2305.02442
| null |
https://arxiv.org/abs/2305.02442v1
|
https://arxiv.org/pdf/2305.02442v1.pdf
|
Tackling Universal Properties of Minimal Trap Spaces of Boolean Networks
|
Minimal trap spaces (MTSs) capture subspaces in which the Boolean dynamics is trapped, whatever the update mode. They correspond to the attractors of the most permissive mode. Due to their versatility, the computation of MTSs has recently gained traction, essentially by focusing on their enumeration. In this paper, we address the logical reasoning on universal properties of MTSs in the scope of two problems: the reprogramming of Boolean networks for identifying the permanent freeze of Boolean variables that enforce a given property on all the MTSs, and the synthesis of Boolean networks from universal properties on their MTSs. Both problems reduce to solving the satisfiability of quantified propositional logic formula with 3 levels of quantifiers ($\exists\forall\exists$). In this paper, we introduce a Counter-Example Guided Refinement Abstraction (CEGAR) to efficiently solve these problems by coupling the resolution of two simpler formulas. We provide a prototype relying on Answer-Set Programming for each formula and show its tractability on a wide range of Boolean models of biological networks.
|
['Loïc Paulevé', 'Gustavo Magaña López', 'Jean-Marie Lagniez', 'Sara Riva']
|
2023-05-03
| null | null | null | null |
['logical-reasoning']
|
['reasoning']
|
[ 5.01540482e-01 5.86865187e-01 -2.21540079e-01 -6.14490546e-03
-1.87031716e-01 -8.84951055e-01 3.36274892e-01 2.19699502e-01
-1.15081044e-02 1.25463259e+00 -2.77877599e-01 -7.38565087e-01
-8.60761583e-01 -1.02715099e+00 -7.44758070e-01 -8.19054902e-01
-5.56766152e-01 6.75935805e-01 5.61953187e-01 -5.23433506e-01
1.43563017e-01 5.18195808e-01 -1.70930791e+00 3.28961998e-01
4.04103994e-01 9.04236615e-01 -1.84293121e-01 5.71913719e-01
4.93460484e-02 7.71055698e-01 -3.76267940e-01 -2.99680457e-02
1.26352638e-01 -4.58844095e-01 -1.19456267e+00 -3.56408358e-01
3.08291465e-01 1.21843494e-01 -9.46561620e-02 1.23016703e+00
-1.05908118e-01 -7.59336650e-02 3.94373775e-01 -1.78141141e+00
5.49867637e-02 6.35167897e-01 6.37297556e-02 2.40527362e-01
5.66393137e-01 1.67628065e-01 1.51264644e+00 -5.73684052e-02
9.87619221e-01 1.09002769e+00 4.03700083e-01 7.70607471e-01
-1.91856289e+00 -2.96249449e-01 -6.92299902e-02 1.71171561e-01
-1.44596887e+00 -5.39497137e-01 4.87314820e-01 -3.16251338e-01
1.20399821e+00 9.76436615e-01 9.67322886e-01 5.32496929e-01
5.02147555e-01 1.34614021e-01 1.31816387e+00 -3.48695815e-01
5.81043780e-01 -5.22221290e-02 3.93583864e-01 9.18242097e-01
5.33227086e-01 -2.12709568e-02 -6.51481807e-01 -5.40037811e-01
6.73297584e-01 -3.16458464e-01 -3.54713887e-01 -4.38579023e-01
-9.18914497e-01 5.42332172e-01 4.57029603e-03 4.54401881e-01
2.16437712e-01 4.91178691e-01 1.51796997e-01 6.24105930e-01
-1.76914200e-01 8.61109018e-01 -6.24746263e-01 1.26725748e-01
-5.35455048e-01 4.71200287e-01 1.28500319e+00 9.69584107e-01
8.67318034e-01 -5.35405993e-01 -5.71530312e-02 -2.71833271e-01
1.09616868e-01 3.92309874e-01 -2.74440020e-01 -1.09548247e+00
8.84392709e-02 1.00298488e+00 2.30544209e-01 -1.05976355e+00
-5.04426956e-01 -1.89140439e-01 -8.39914978e-01 -1.02536462e-01
6.04105651e-01 1.15955815e-01 -3.52411717e-01 2.16899514e+00
2.78092563e-01 7.97491968e-02 -3.55541930e-02 4.23280269e-01
2.02430904e-01 9.09585953e-01 -4.61628318e-01 -8.26590598e-01
1.30100179e+00 -1.34866506e-01 -7.12891698e-01 5.36099821e-02
6.26823962e-01 3.19362938e-01 6.31751120e-01 5.52710474e-01
-1.28250885e+00 3.26353580e-01 -1.11040473e+00 1.98906571e-01
-4.42470908e-01 -5.05577207e-01 8.93821597e-01 3.00756097e-01
-1.43925619e+00 6.02652669e-01 -7.54503787e-01 -2.81756490e-01
1.18800275e-01 1.11185730e+00 -4.65386480e-01 1.20452374e-01
-1.39183962e+00 7.71130919e-01 2.80270249e-01 3.95146310e-01
-9.58189785e-01 -3.69042844e-01 -7.34739840e-01 2.76529104e-01
7.92901814e-01 -6.13247573e-01 7.86556661e-01 -7.88430989e-01
-1.24887502e+00 9.16693211e-01 -3.18238616e-01 -6.90395474e-01
1.41410381e-01 8.35664868e-01 8.20838660e-02 -2.35029217e-02
-3.44823711e-02 4.59033757e-01 4.25452381e-01 -8.61129165e-01
-3.18675250e-01 -3.33479583e-01 8.14825654e-01 -4.82387304e-01
-9.83540043e-02 2.43887585e-02 -5.96258566e-02 2.74259567e-01
3.86988848e-01 -9.27277505e-01 -3.47918034e-01 -7.55155832e-02
-6.45309150e-01 -3.09742451e-01 2.05052361e-01 2.17234656e-01
1.63374627e+00 -1.90810776e+00 8.63640249e-01 6.75473928e-01
4.17222738e-01 -2.62495488e-01 2.87087522e-02 5.50268829e-01
1.95827615e-02 4.97575432e-01 -2.87273318e-01 2.91613787e-01
1.10776678e-01 7.24187255e-01 -6.24447703e-01 5.91885030e-01
4.42771554e-01 8.11695933e-01 -8.63525510e-01 -7.31559873e-01
-2.63767958e-01 -2.11433515e-01 -8.64923716e-01 -2.00125992e-01
-9.05037999e-01 1.37070328e-01 -3.91017526e-01 5.76431990e-01
4.81689811e-01 -3.15802574e-01 8.89753580e-01 -1.65307038e-02
-2.39288136e-01 3.41722280e-01 -1.39876425e+00 1.06642962e+00
2.04074591e-01 3.55753690e-01 4.25904632e-01 -6.76431239e-01
5.07200420e-01 1.06055424e-01 3.79164904e-01 -3.72585535e-01
1.57987013e-01 2.71431804e-01 2.18497723e-01 -3.27032655e-01
1.93814352e-01 -4.11620080e-01 -3.85759026e-01 4.35739636e-01
1.56820670e-03 -2.73445696e-01 6.63577676e-01 1.92024514e-01
1.77580714e+00 -2.64505565e-01 2.26548404e-01 -8.13163638e-01
8.41283143e-01 1.61207929e-01 8.13036859e-01 9.14343834e-01
-1.22429155e-01 -1.39168113e-01 1.51886082e+00 -5.08645058e-01
-7.42395699e-01 -8.30321670e-01 -2.04975620e-01 4.92541909e-01
3.67122173e-01 -1.02795017e+00 -7.31194854e-01 -8.39681476e-02
-1.38983473e-01 3.08520347e-01 -7.82359779e-01 -2.94615835e-01
-6.35387182e-01 -5.90324521e-01 6.58507407e-01 7.70624951e-02
-5.67145422e-02 -8.15888822e-01 -1.02834117e+00 -2.07160432e-02
-2.38597080e-01 -7.44721055e-01 1.68505415e-01 7.54718602e-01
-9.72849190e-01 -1.25987279e+00 3.65877330e-01 -4.33586806e-01
8.85196209e-01 -3.09251398e-01 8.16624939e-01 4.91962969e-01
-3.01796943e-01 8.28391016e-02 3.19024831e-01 -5.75320348e-02
-3.71070117e-01 1.63192272e-01 2.50846505e-01 -2.07540452e-01
-1.25563025e-01 -6.80106521e-01 1.25886425e-01 4.18881595e-01
-1.17423201e+00 1.21606246e-01 1.03613327e-03 8.58363807e-01
9.45220649e-01 6.69681251e-01 1.37168905e-02 -7.32817888e-01
5.23466170e-01 -3.23612928e-01 -1.20164514e+00 4.13591355e-01
-3.59625787e-01 5.51137149e-01 5.77977359e-01 -1.04767151e-01
-3.52846831e-01 9.05445889e-02 4.78705227e-01 5.84797282e-03
2.95745641e-01 6.98715985e-01 -2.27978691e-01 9.94294696e-03
4.05276000e-01 1.55428961e-01 -1.97522994e-02 1.83095202e-01
-4.90377769e-02 4.43996303e-02 4.40058857e-01 -9.06571507e-01
6.71819508e-01 6.99150145e-01 1.10360742e+00 -5.91328859e-01
-4.02029037e-01 1.44602545e-02 -2.78251171e-01 -2.79089391e-01
4.59768504e-01 -1.09638415e-01 -1.28404415e+00 1.04197592e-01
-1.20698857e+00 -3.79912108e-01 -4.71939236e-01 -4.13936019e-01
-8.44963014e-01 1.75208747e-02 -4.15823698e-01 -1.12841368e+00
1.42797306e-01 -1.10934496e+00 8.83487165e-01 -2.03568190e-01
-3.99116695e-01 -6.11011505e-01 4.32938129e-01 -2.24754408e-01
2.57057577e-01 5.04255056e-01 1.50203097e+00 -3.62795651e-01
-1.02216327e+00 -1.23253874e-01 1.76599115e-01 -2.62161046e-01
-2.27594420e-01 3.45528692e-01 -6.46091044e-01 -2.73346722e-01
-5.33231124e-02 -4.12598476e-02 5.55756092e-01 2.75107771e-01
8.28845680e-01 -5.81403017e-01 -5.62246323e-01 2.63184845e-01
1.53523958e+00 -5.29252216e-02 7.57476389e-01 2.96437114e-01
-2.29844689e-01 6.32094324e-01 3.27115238e-01 1.83909193e-01
-1.00606799e-01 5.38728058e-01 8.19480181e-01 5.67653835e-01
5.21316230e-01 9.67977419e-02 3.79503042e-01 4.19195920e-01
-1.32312119e-01 -1.57857746e-01 -9.52408314e-01 3.94952625e-01
-1.99615180e+00 -1.06453335e+00 -1.58472940e-01 2.07099795e+00
1.21669924e+00 5.25120437e-01 2.50758808e-02 4.21433210e-01
6.77055955e-01 -8.64062905e-02 -2.03083605e-01 -5.50553739e-01
-3.40305388e-01 3.21861207e-01 4.31460947e-01 9.46096241e-01
-7.65936732e-01 6.69696987e-01 6.66422701e+00 5.38170040e-01
-7.58450389e-01 -7.95053989e-02 3.66960615e-01 -2.74116457e-01
-5.72732747e-01 5.36092877e-01 -8.51516664e-01 1.38207495e-01
1.13185072e+00 -2.55346801e-02 9.72801745e-01 3.14237118e-01
6.31930456e-02 -3.74919504e-01 -1.52645731e+00 3.75589788e-01
-4.28320527e-01 -1.55582869e+00 -5.86165749e-02 1.70428216e-01
7.02838480e-01 -4.92624521e-01 -5.06884232e-02 -1.39690295e-01
1.81658223e-01 -1.11467576e+00 9.44594204e-01 6.85199916e-01
8.12001646e-01 -4.95218933e-01 5.40797889e-01 3.73145431e-01
-8.75036418e-01 -2.76158512e-01 -1.88001588e-01 -4.43304509e-01
-5.18787615e-02 4.55333680e-01 -4.14194256e-01 2.69590884e-01
4.97254640e-01 2.45868161e-01 -2.87149608e-01 6.56672955e-01
-1.22281112e-01 1.70798853e-01 -9.87595558e-01 -3.24768931e-01
6.86220303e-02 -2.89767861e-01 7.96738923e-01 7.82882929e-01
-9.37093049e-02 2.87163168e-01 -4.52925831e-01 1.33278799e+00
1.53312773e-01 -5.27633011e-01 -5.98563254e-01 -3.03332299e-01
3.03168178e-01 1.10923064e+00 -9.16919887e-01 -4.94263880e-02
2.31193647e-01 2.21718565e-01 4.03044164e-01 1.20556563e-01
-7.89093614e-01 -3.28829795e-01 6.53388143e-01 3.31243098e-01
1.18131779e-01 -6.01789914e-02 -1.75694093e-01 -1.00798476e+00
3.48340720e-02 -9.06560600e-01 3.66129518e-01 -6.17205262e-01
-6.15463555e-01 3.98870170e-01 2.46569291e-01 -6.00304842e-01
-9.07134488e-02 -7.73008883e-01 -3.44332010e-01 4.36237574e-01
-9.93899882e-01 -5.44618309e-01 1.49776146e-01 5.94547927e-01
-3.34328443e-01 5.70606589e-01 9.30974901e-01 2.31859982e-02
-7.06347406e-01 -4.27067839e-02 -2.63310641e-01 -5.77623546e-01
-4.44306433e-02 -1.19251931e+00 -2.29512587e-01 8.86480987e-01
-2.96783000e-01 1.11349821e+00 1.00587368e+00 -4.43452299e-01
-2.26219940e+00 -7.59993911e-01 1.21545863e+00 -4.71560866e-01
9.14183199e-01 -6.14594817e-01 -3.36717069e-01 9.29861665e-01
-2.22705901e-01 -2.06872702e-01 2.60023803e-01 7.29084536e-02
-2.73534030e-01 -3.98479372e-01 -1.08796465e+00 7.82587409e-01
1.22188127e+00 -6.04264259e-01 -1.94993660e-01 2.78185755e-01
5.47682643e-01 -4.04908687e-01 -6.58416867e-01 3.66223514e-01
4.58349288e-01 -9.59388435e-01 5.21905720e-01 -1.10578775e+00
4.38939273e-01 -8.13015342e-01 -3.21741223e-01 -4.97687310e-01
-2.67959625e-01 -1.14391255e+00 -4.11957949e-01 7.69568861e-01
4.93551105e-01 -8.20430875e-01 5.50375819e-01 6.46440089e-01
2.19383508e-01 -1.07215858e+00 -1.75353849e+00 -4.95170057e-01
-2.62474567e-01 -2.88589120e-01 7.77161956e-01 6.36683285e-01
7.59795547e-01 3.82960737e-01 2.26407796e-02 1.25989273e-01
2.84628361e-01 3.01790714e-01 3.71099740e-01 -1.41717398e+00
-6.17765844e-01 -5.76606512e-01 -7.24179447e-01 -4.77098018e-01
1.91849455e-01 -1.04326355e+00 3.39557640e-02 -1.17733431e+00
3.43692422e-01 -3.43895763e-01 -9.97731462e-02 8.34768414e-01
7.34649062e-01 -2.76466440e-02 -1.90925613e-01 1.17865786e-01
-1.14018655e+00 1.18421875e-02 9.73960757e-01 -4.11519498e-01
-2.18965188e-02 -2.17268631e-01 -6.50629699e-01 4.28483129e-01
5.53098857e-01 -5.80788910e-01 -2.71568239e-01 -2.51620309e-04
1.46125889e+00 4.05027598e-01 4.91369665e-01 -8.10781896e-01
4.87022281e-01 -5.58964491e-01 -5.60906887e-01 -1.86889753e-01
2.90059566e-01 -1.01889098e+00 8.66430879e-01 9.33614552e-01
-4.90281373e-01 -3.51249836e-02 2.70508498e-01 5.52936196e-01
5.75224087e-02 -3.57061267e-01 4.54061061e-01 -3.84135060e-02
-2.26781726e-01 -7.80855864e-02 -7.42448568e-01 -1.65169522e-01
1.19333518e+00 -1.22350916e-01 -6.29885137e-01 -1.84067965e-01
-9.62298214e-01 3.81570637e-01 4.67753708e-01 -3.87932807e-01
4.15223330e-01 -1.00555134e+00 -1.12139061e-01 3.11862200e-01
4.87032114e-03 9.17551219e-02 -1.55140415e-01 1.21668410e+00
-5.13064623e-01 8.26898515e-01 -2.64648229e-01 -4.67538565e-01
-1.21608758e+00 5.48317850e-01 8.92782032e-01 -5.11694908e-01
2.25590169e-01 5.90615690e-01 -9.91686136e-02 -1.48124084e-01
6.15607724e-02 -9.12639678e-01 1.34562507e-01 -2.23697782e-01
3.60620230e-01 4.85304564e-01 -6.42339885e-02 -2.87819654e-01
-7.58072555e-01 2.06194475e-01 3.59874934e-01 -1.06364772e-01
1.30860555e+00 1.90055013e-01 -1.34085488e+00 4.15141493e-01
6.47481740e-01 -6.35700300e-03 -4.75762933e-01 2.59462655e-01
-2.48485338e-02 1.65594310e-01 -2.40872294e-01 -5.18737376e-01
-6.66283906e-01 3.03056061e-01 -1.33017480e-01 7.96831131e-01
1.18880594e+00 3.57607812e-01 1.23955801e-01 9.35932338e-01
9.42716479e-01 -7.90181935e-01 -6.14270687e-01 7.28624880e-01
6.38866901e-01 -2.52834707e-01 7.18262866e-02 -6.23265803e-01
4.57063496e-01 1.15499318e+00 3.41550648e-01 -6.76812306e-02
4.03552741e-01 7.13787496e-01 -9.29014802e-01 -4.70106333e-01
-1.39873779e+00 -5.74439857e-03 -2.60640919e-01 2.15849400e-01
-1.10012397e-01 8.77462104e-02 -5.58378160e-01 7.21626043e-01
-4.51902971e-02 1.76240280e-01 6.55621469e-01 1.14547980e+00
-4.78675514e-01 -1.06352782e+00 -2.93931216e-01 5.26417077e-01
-2.20505238e-01 -7.91393034e-03 -8.25842500e-01 5.96740067e-01
3.17271709e-01 8.63664389e-01 -6.45987019e-02 -4.34818923e-01
1.28173605e-01 4.30300534e-02 1.09681797e+00 -4.07992601e-01
-4.72507417e-01 -3.38063091e-01 3.06085229e-01 -7.35514283e-01
-3.92275840e-01 -8.25958490e-01 -1.32111907e+00 -5.30027211e-01
-6.62561715e-01 3.75994533e-01 1.11250803e-01 9.41312313e-01
2.63908535e-01 3.97963613e-01 9.05361772e-02 -3.20825249e-01
-8.08132350e-01 -3.00146878e-01 -8.13686252e-01 -3.34311813e-01
4.39543009e-01 -5.98906159e-01 -6.26803219e-01 -2.06331819e-01]
|
[8.647235870361328, 6.792140960693359]
|
6998eed3-8631-4565-92a2-250caa3fd6ff
|
novice-type-error-diagnosis-with-natural
|
2210.03682
| null |
https://arxiv.org/abs/2210.03682v1
|
https://arxiv.org/pdf/2210.03682v1.pdf
|
Novice Type Error Diagnosis with Natural Language Models
|
Strong static type systems help programmers eliminate many errors without much burden of supplying type annotations. However, this flexibility makes it highly non-trivial to diagnose ill-typed programs, especially for novice programmers. Compared to classic constraint solving and optimization-based approaches, the data-driven approach has shown great promise in identifying the root causes of type errors with higher accuracy. Instead of relying on hand-engineered features, this work explores natural language models for type error localization, which can be trained in an end-to-end fashion without requiring any features. We demonstrate that, for novice type error diagnosis, the language model-based approach significantly outperforms the previous state-of-the-art data-driven approach. Specifically, our model could predict type errors correctly 62% of the time, outperforming the state-of-the-art Nate's data-driven model by 11%, in a more rigorous accuracy metric. Furthermore, we also apply structural probes to explain the performance difference between different language models.
|
['Xujie Si', 'Brigitte Pientka', 'Tianyu Han', 'Yixuan Li', 'Haolin Ye', 'Chuqin Geng']
|
2022-10-07
| null | null | null | null |
['type']
|
['speech']
|
[-3.77641618e-02 1.37651339e-01 -2.71301746e-01 -4.75783199e-01
-8.34835947e-01 -5.44324279e-01 1.01986423e-01 6.75195098e-01
-3.54416937e-01 3.41819048e-01 -2.60536522e-01 -8.28335226e-01
-4.78376774e-03 -7.55809367e-01 -8.10598850e-01 1.78120166e-01
-5.33076860e-02 2.86772043e-01 2.99028784e-01 -2.22386241e-01
8.00453067e-01 -1.62063897e-01 -1.78828049e+00 5.20600796e-01
1.44177222e+00 6.95659637e-01 -1.35201188e-02 8.28302085e-01
-5.85565567e-01 8.87116194e-01 -5.35994709e-01 -4.75415796e-01
-2.40633935e-01 -1.37865230e-01 -9.76650536e-01 -5.56324244e-01
6.09153032e-01 -6.74096271e-02 3.23772520e-01 1.19551039e+00
3.00120354e-01 -2.43696004e-01 1.07917286e-01 -1.15577674e+00
-7.70750463e-01 7.25916624e-01 -5.11095166e-01 2.92218709e-03
9.26432192e-01 7.67092183e-02 1.12466800e+00 -9.66721714e-01
6.44061327e-01 1.10336220e+00 1.00055897e+00 7.76592493e-01
-1.45247829e+00 -2.56839246e-01 2.61354208e-01 2.47080013e-01
-1.06194150e+00 -1.98178142e-01 5.01346827e-01 -8.99874926e-01
1.62550700e+00 3.09175164e-01 4.74504530e-01 7.13165224e-01
2.43766233e-02 7.15212584e-01 1.27054203e+00 -7.84798503e-01
2.37158686e-01 2.46725425e-01 7.14008152e-01 1.14446735e+00
4.17341948e-01 1.18245378e-01 -5.12111485e-01 -5.15439391e-01
1.01719983e-01 -1.81220874e-01 -1.67826489e-01 -3.89831215e-01
-1.08841300e+00 5.26904047e-01 1.97814688e-01 3.91815335e-01
1.45182639e-01 3.18617821e-01 5.49689829e-01 4.59009767e-01
4.21362311e-01 9.35099900e-01 -1.05646908e+00 -8.38027120e-01
-8.39931965e-01 4.93157834e-01 1.09181845e+00 1.14650667e+00
6.48162842e-01 3.29454131e-02 -1.52540402e-02 7.04003990e-01
3.79434973e-01 1.73581645e-01 3.41533542e-01 -6.25213861e-01
3.90535086e-01 1.26950300e+00 -5.95476292e-02 -5.49803972e-01
-3.37728411e-01 -5.15250683e-01 -1.03735059e-01 6.36223793e-01
7.81499922e-01 -8.61530378e-02 -6.57143712e-01 1.63427842e+00
6.29541054e-02 -5.23263961e-02 -7.43012652e-02 6.88004375e-01
5.59834599e-01 1.60699412e-01 -1.62557542e-01 1.07172340e-01
1.18340778e+00 -8.65827024e-01 -2.63048828e-01 -3.55900586e-01
1.31557047e+00 -7.51204610e-01 1.50847256e+00 6.73344314e-01
-1.14352155e+00 -1.27184123e-01 -9.39958096e-01 -2.00539842e-01
-3.71289104e-01 -1.09947689e-01 7.91522384e-01 9.33413088e-01
-1.00084674e+00 6.85748816e-01 -9.81316447e-01 -3.15615982e-01
1.97110698e-01 2.35004798e-01 -1.76578775e-01 -1.89120799e-01
-4.89509761e-01 8.52333188e-01 8.83365721e-02 -1.49699509e-01
-6.82835579e-01 -1.33561027e+00 -8.49214554e-01 1.45179838e-01
6.49141908e-01 -6.33575439e-01 1.45840502e+00 -8.40017021e-01
-1.12193131e+00 7.69788325e-01 -6.08291507e-01 -1.83576927e-01
5.19071519e-01 -4.03346956e-01 -9.68563557e-02 -6.14521503e-01
-9.47109684e-02 -1.28741831e-01 2.73425251e-01 -1.22472382e+00
-7.20946252e-01 -1.89462274e-01 4.15951014e-01 -5.06776810e-01
-2.99400061e-01 3.62750918e-01 -1.23088390e-01 -3.92583638e-01
-1.11515582e-01 -7.39036024e-01 -1.53632835e-01 7.23027885e-02
-2.83445895e-01 -4.52255219e-01 5.54672301e-01 -6.52319789e-01
1.53701222e+00 -2.10565114e+00 2.67643481e-01 1.63962722e-01
5.18467546e-01 3.62052739e-01 -5.73341958e-02 2.21759021e-01
-2.05617547e-01 6.55465484e-01 -4.53357309e-01 -2.88710743e-01
4.35511440e-01 1.45328147e-02 9.78445858e-02 1.52595907e-01
3.86375099e-01 6.89811051e-01 -1.17635846e+00 -1.50586337e-01
-8.02463368e-02 9.56120566e-02 -1.43620825e+00 5.07944345e-01
-7.29982555e-01 1.57215800e-02 -1.86186522e-01 8.60024929e-01
3.99296910e-01 -3.86979468e-02 2.81865478e-01 4.65170205e-01
-2.77099401e-01 7.56942093e-01 -1.03071129e+00 1.77712750e+00
-8.88665915e-01 5.37601411e-01 8.48693773e-03 -8.96822810e-01
8.60466361e-01 1.14874005e-01 -1.06052374e-02 -7.00867355e-01
-3.22830051e-01 7.65800834e-01 3.65155697e-01 -8.01623464e-01
4.61023867e-01 1.43754721e-01 -4.05153960e-01 3.21033537e-01
-2.16597214e-01 1.23801664e-01 3.08472097e-01 -5.16612455e-02
1.49679494e+00 4.13747638e-01 1.09706730e-01 -3.49696636e-01
5.51685214e-01 2.72370875e-01 9.34359670e-01 8.73349309e-01
1.52359456e-01 4.11775827e-01 8.32176387e-01 -5.27483821e-01
-1.01792145e+00 -5.96192360e-01 1.57073233e-02 1.18541646e+00
-2.16685146e-01 -9.79784548e-01 -8.22740197e-01 -8.19299459e-01
1.40891328e-01 8.82329047e-01 -3.61437529e-01 -2.28104904e-01
-7.51257002e-01 -4.54820782e-01 6.20770276e-01 6.32003069e-01
1.06416412e-01 -6.76168919e-01 -5.13812900e-01 1.98429659e-01
-1.17854774e-01 -7.02550232e-01 -1.37042493e-01 1.53102070e-01
-6.59482479e-01 -1.12770164e+00 -4.60986085e-02 -7.07668006e-01
8.21348310e-01 -2.59620249e-01 1.44467986e+00 9.88178372e-01
-5.43492913e-01 2.02683523e-01 -5.38262963e-01 -2.44989112e-01
-6.45533025e-01 2.58124858e-01 -3.00867707e-01 -6.31202757e-01
5.89014232e-01 -4.59371537e-01 -9.13366079e-02 3.21002528e-02
-4.76305187e-01 -1.80278882e-01 4.17585790e-01 1.04031205e+00
5.81202544e-02 -2.06314713e-01 3.00723940e-01 -1.48919678e+00
4.57865864e-01 -5.09197950e-01 -7.58817255e-01 4.38910693e-01
-1.16851687e+00 4.04943734e-01 8.69394422e-01 -2.58478105e-01
-9.19728518e-01 -2.76277035e-01 -4.17489707e-01 9.40122828e-02
-1.76664904e-01 9.27348197e-01 3.62857319e-02 -3.32396895e-01
8.71653497e-01 1.34515315e-01 -2.30830565e-01 -7.99489021e-01
-4.37347358e-03 5.11366904e-01 4.68228161e-01 -1.16646850e+00
7.25658715e-01 -4.77964669e-01 -2.25541756e-01 -5.44405580e-01
-5.50933421e-01 -3.93783420e-01 -5.53101122e-01 1.50306970e-01
5.42344093e-01 -5.05916417e-01 -8.35089564e-01 6.42905116e-01
-1.27516973e+00 -5.50862491e-01 2.81970706e-02 -9.24463421e-02
-4.66430724e-01 4.01189357e-01 -5.61135769e-01 -8.95256758e-01
1.09464131e-01 -1.32852852e+00 9.59918082e-01 -4.37928922e-02
-5.64041257e-01 -1.07545650e+00 -2.30551641e-02 4.54487771e-01
7.17628896e-01 2.90368736e-01 1.75654948e+00 -7.21888244e-01
-8.69401872e-01 -2.83221781e-01 -1.69032589e-01 1.02635920e-01
-2.82237589e-01 1.63713291e-01 -8.16899717e-01 -2.10983828e-01
-3.35070729e-01 -2.52307415e-01 4.50935483e-01 -2.60533303e-01
1.31356168e+00 -3.89443368e-01 -1.70447037e-01 7.06039965e-01
1.63681042e+00 -1.47832140e-01 4.27039057e-01 4.99362290e-01
8.61869276e-01 3.79514843e-01 5.39375722e-01 4.42720890e-01
7.77344465e-01 6.55405641e-01 5.29906988e-01 -1.11246547e-02
8.08731839e-02 -1.78636879e-01 2.73056895e-01 5.46798408e-01
-7.53747299e-03 -4.39302474e-02 -1.47141349e+00 7.95062721e-01
-2.01709628e+00 -6.09287202e-01 -5.72763383e-01 2.35104990e+00
1.19926572e+00 2.85261810e-01 1.50795847e-01 3.09043109e-01
1.89297050e-01 -3.30776006e-01 -2.34520957e-01 -1.02578354e+00
5.39580345e-01 2.86221117e-01 2.44450450e-01 6.43824458e-01
-5.69538951e-01 8.04508567e-01 6.41216660e+00 4.09082174e-01
-1.06543875e+00 9.37926695e-02 -2.36824136e-02 1.55232966e-01
-7.48565376e-01 4.88535017e-01 -8.37531507e-01 4.29350913e-01
1.02533340e+00 -2.43466631e-01 6.45871043e-01 1.29813862e+00
-7.84777850e-02 -2.10906759e-01 -1.63892078e+00 5.08262098e-01
-5.04968055e-02 -1.22779477e+00 -5.49637139e-01 -3.09305470e-02
4.94857341e-01 -2.58552670e-01 -2.49585077e-01 6.08853161e-01
3.28526795e-01 -1.03949738e+00 8.88939023e-01 5.78342259e-01
5.84630251e-01 -5.36949337e-01 8.12610149e-01 6.48118019e-01
-8.37002039e-01 -3.93535584e-01 -8.23132396e-02 -5.81026495e-01
-3.41077417e-01 8.33453059e-01 -7.77401209e-01 3.47642809e-01
7.01519549e-01 5.32174289e-01 -1.00791478e+00 1.25702679e+00
-2.03117684e-01 7.37271190e-01 -7.78579339e-02 -3.51826429e-01
-2.38390282e-01 4.09211159e-01 4.80872959e-01 1.51422060e+00
2.49917194e-01 -2.42027208e-01 4.57011610e-01 1.29787016e+00
9.83933955e-02 -1.62782684e-01 -3.94583136e-01 -1.45920351e-01
6.04017675e-01 8.84331465e-01 -6.44105524e-02 -3.10738415e-01
-5.92610061e-01 5.57994068e-01 7.99011230e-01 1.09478831e-02
-4.54203248e-01 -7.24465072e-01 6.63091123e-01 3.19160521e-01
9.29262862e-02 -3.91891509e-01 -7.71640241e-01 -1.29053891e+00
4.70453709e-01 -1.29966950e+00 9.13431272e-02 -2.98643082e-01
-1.09862483e+00 3.08359683e-01 -7.90728703e-02 -6.61550462e-01
-5.01785517e-01 -8.15347075e-01 -6.60385966e-01 1.06392431e+00
-1.52524471e+00 -8.67493033e-01 -3.11021835e-01 -1.49282292e-02
3.19640160e-01 -2.01490987e-02 1.03388214e+00 4.56883132e-01
-7.60077715e-01 1.12047064e+00 -1.94307461e-01 2.02523202e-01
5.31063259e-01 -1.94707310e+00 6.45208657e-01 1.22531164e+00
-4.40003097e-01 1.19936693e+00 9.03798461e-01 -5.43949723e-01
-1.92054021e+00 -8.53317976e-01 1.33212268e+00 -6.97613835e-01
7.92661190e-01 -5.68246305e-01 -1.30052221e+00 6.01800263e-01
-2.10676447e-01 1.58484042e-01 5.31457543e-01 8.19566667e-01
-9.22312498e-01 1.30475938e-01 -1.08491063e+00 3.74153912e-01
1.29365337e+00 -6.03149414e-01 -7.18225181e-01 2.80071467e-01
5.85042715e-01 -6.44432008e-01 -1.10448599e+00 2.25915894e-01
5.52212238e-01 -1.11903691e+00 6.38117135e-01 -6.54876709e-01
9.59025145e-01 -3.58251929e-01 -1.71314672e-01 -1.25749171e+00
-2.97412694e-01 -5.29924273e-01 -9.39402822e-03 1.49239469e+00
8.53658438e-01 -7.96458185e-01 6.56260848e-01 1.09845150e+00
-5.83434820e-01 -8.02004457e-01 -5.10852933e-01 -7.74916947e-01
2.95196176e-01 -5.92051685e-01 5.77784717e-01 9.79382753e-01
6.35305285e-01 -8.04830864e-02 -5.65170459e-02 1.30215794e-01
3.79915625e-01 1.92527294e-01 8.31002355e-01 -1.18584192e+00
-8.36711466e-01 -7.40323305e-01 -4.58422393e-01 -6.93435371e-01
4.18676347e-01 -1.05215514e+00 3.94088119e-01 -1.18641376e+00
1.16992168e-01 -1.00783801e+00 1.13157623e-01 8.64509761e-01
-4.99887675e-01 -9.72688049e-02 4.47791219e-02 -2.89768934e-01
-3.80156875e-01 -5.64854592e-02 4.92102653e-01 -2.73582358e-02
7.21331686e-02 -6.40182793e-02 -8.56483877e-01 9.16539013e-01
4.43389624e-01 -6.43837929e-01 -4.30580825e-02 -7.87683547e-01
8.57741177e-01 1.11756004e-01 3.03534687e-01 -7.91263759e-01
2.28434816e-01 -2.38635257e-01 -2.57141560e-01 -3.53007801e-02
-3.26365858e-01 -5.96860945e-01 -1.34562597e-01 6.59996092e-01
-5.25097013e-01 2.96988934e-01 2.98644096e-01 2.41075128e-01
-1.01574019e-01 -7.72900999e-01 4.83603954e-01 -1.28027514e-01
-8.85019004e-01 -1.72187880e-01 -4.02028441e-01 3.79983306e-01
7.48057842e-01 -3.76205780e-02 -7.03459382e-01 3.72251213e-01
-3.62337708e-01 1.59425542e-01 1.04245675e+00 4.10573930e-01
3.76042783e-01 -7.59136081e-01 -4.60293442e-01 4.28880483e-01
4.76717681e-01 6.76731095e-02 -2.18384668e-01 8.07568789e-01
-5.73060036e-01 1.60772488e-01 1.01071671e-01 -7.18032598e-01
-1.35696292e+00 6.09700143e-01 3.17879945e-01 -3.67519200e-01
-4.62866336e-01 9.59330738e-01 -3.36156696e-01 -8.59571755e-01
1.66711167e-01 -7.37557173e-01 3.29273343e-01 -4.86074269e-01
4.63590145e-01 2.14240953e-01 5.03478467e-01 8.82245004e-02
-7.57919669e-01 5.32678902e-01 -2.29732975e-01 2.40552798e-01
1.39787483e+00 5.25976419e-01 -5.65750778e-01 5.99412799e-01
8.90101790e-01 2.58560181e-01 -6.42289102e-01 -2.32220426e-01
6.15652978e-01 -8.51533711e-01 -1.34052441e-01 -1.33804727e+00
-5.71022868e-01 9.67536569e-01 2.97992378e-01 2.36287668e-01
8.54682684e-01 -2.96987474e-01 5.12108743e-01 4.37622547e-01
7.10309148e-01 -6.93987131e-01 -1.58265203e-01 6.84750021e-01
5.06762683e-01 -1.38233042e+00 -1.90789565e-01 -7.19099343e-01
-1.05423585e-01 1.13493657e+00 9.69617724e-01 -1.56831339e-01
2.85143942e-01 7.45992601e-01 -1.70887351e-01 1.48908822e-02
-1.14936495e+00 1.81747407e-01 2.33757764e-01 8.52150798e-01
1.07568848e+00 -9.81589034e-03 -3.72715652e-01 6.29229307e-01
-3.56504887e-01 1.47531182e-01 8.41781974e-01 1.46823347e+00
-3.17097694e-01 -1.70825815e+00 -2.85575211e-01 5.23374677e-01
-4.92887467e-01 -4.49428678e-01 -4.16587651e-01 7.34917939e-01
8.21440741e-02 8.83674443e-01 -2.22806647e-01 -4.64469880e-01
5.87905943e-01 3.30283701e-01 6.68857098e-01 -1.20797324e+00
-1.17032421e+00 -4.59370673e-01 5.77781558e-01 -6.94379926e-01
1.41214386e-01 -6.56827450e-01 -1.27557814e+00 -6.33721054e-01
-3.54958445e-01 -7.04467148e-02 6.44222677e-01 1.11081624e+00
5.40276110e-01 6.22326493e-01 2.93350548e-01 -4.56382722e-01
-7.19200015e-01 -6.71539426e-01 6.73990473e-02 4.46462452e-01
5.71171224e-01 -6.00123107e-01 -2.89725274e-01 -6.52202070e-02]
|
[7.762411117553711, 7.725215911865234]
|
faaeeeae-2983-4386-85ef-5d4b4524c70a
|
sound-to-visual-scene-generation-by-audio-to
|
2303.17490
| null |
https://arxiv.org/abs/2303.17490v1
|
https://arxiv.org/pdf/2303.17490v1.pdf
|
Sound to Visual Scene Generation by Audio-to-Visual Latent Alignment
|
How does audio describe the world around us? In this paper, we propose a method for generating an image of a scene from sound. Our method addresses the challenges of dealing with the large gaps that often exist between sight and sound. We design a model that works by scheduling the learning procedure of each model component to associate audio-visual modalities despite their information gaps. The key idea is to enrich the audio features with visual information by learning to align audio to visual latent space. We translate the input audio to visual features, then use a pre-trained generator to produce an image. To further improve the quality of our generated images, we use sound source localization to select the audio-visual pairs that have strong cross-modal correlations. We obtain substantially better results on the VEGAS and VGGSound datasets than prior approaches. We also show that we can control our model's predictions by applying simple manipulations to the input waveform, or to the latent space.
|
['Tae-Hyun Oh', 'Andrew Owens', 'Hyunwoo Ha', 'Arda Senocak', 'Kim Sung-Bin']
|
2023-03-30
| null |
http://openaccess.thecvf.com//content/CVPR2023/html/Sung-Bin_Sound_to_Visual_Scene_Generation_by_Audio-to-Visual_Latent_Alignment_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Sung-Bin_Sound_to_Visual_Scene_Generation_by_Audio-to-Visual_Latent_Alignment_CVPR_2023_paper.pdf
|
cvpr-2023-1
|
['scene-generation']
|
['computer-vision']
|
[ 4.13476706e-01 -8.17531124e-02 2.89373338e-01 -2.77594239e-01
-1.22995222e+00 -9.26433086e-01 6.68817043e-01 -2.61647940e-01
1.48143545e-01 2.40988150e-01 5.25993288e-01 -4.85904552e-02
1.83859348e-01 -6.45516574e-01 -7.55391121e-01 -5.55995047e-01
-1.42271295e-01 1.00551181e-01 1.80492923e-01 1.24174170e-01
2.43735895e-01 1.85759738e-01 -2.02589297e+00 7.74360597e-01
4.65470552e-01 8.72999966e-01 3.86575133e-01 1.22430813e+00
-1.50396422e-01 8.51795614e-01 -5.81977785e-01 6.93442896e-02
2.14171112e-01 -8.54551494e-01 -7.09439993e-01 2.49508664e-01
6.37629688e-01 -1.87053844e-01 -1.41642675e-01 7.80308068e-01
6.05316699e-01 1.45857213e-02 7.43400276e-01 -1.59272277e+00
-3.57959837e-01 6.11625612e-01 -4.45705920e-01 1.09193949e-02
4.73325163e-01 4.24660444e-02 1.13551366e+00 -8.35392296e-01
4.89606768e-01 1.29770088e+00 5.05899727e-01 4.32829976e-01
-1.43466198e+00 -7.76549578e-01 -5.25822751e-02 1.53334484e-01
-1.42716539e+00 -9.19860363e-01 9.02623117e-01 -7.75011718e-01
6.52184725e-01 3.43906879e-01 5.74045062e-01 9.40066516e-01
-1.87656477e-01 3.82766247e-01 7.10745811e-01 -7.99905002e-01
1.72239423e-01 9.10117254e-02 -4.15237576e-01 5.21741331e-01
-5.20304084e-01 -1.05996244e-02 -1.03327727e+00 -4.52349424e-01
7.85542250e-01 -4.72035736e-01 -2.15107203e-01 -3.16592306e-01
-1.12726939e+00 7.37618625e-01 1.96582794e-01 1.27544686e-01
-4.48001735e-02 5.93208492e-01 -2.32341345e-02 1.91586420e-01
2.84817636e-01 4.10412073e-01 -7.34863207e-02 -3.49041969e-02
-1.04242635e+00 1.70317128e-01 4.61594701e-01 8.80236328e-01
6.77624166e-01 1.47910804e-01 1.75329670e-03 8.96776855e-01
4.74952787e-01 5.15598536e-01 1.53913602e-01 -1.46204507e+00
3.52318525e-01 -2.02052519e-01 2.37639740e-01 -8.61267865e-01
-1.25175893e-01 -2.61220902e-01 -3.37550879e-01 1.89299881e-01
2.37129912e-01 -1.36188447e-01 -8.60287428e-01 1.91033554e+00
9.31343138e-02 5.05792320e-01 4.10374366e-02 6.57790005e-01
6.73340738e-01 1.00055528e+00 2.67549977e-02 -1.45534515e-01
1.17596233e+00 -7.60230064e-01 -4.98917252e-01 -2.25745499e-01
1.21669404e-01 -1.08950901e+00 1.20139134e+00 2.18108580e-01
-1.12081850e+00 -7.43102074e-01 -9.73421037e-01 3.75695825e-02
1.11461915e-01 2.89962381e-01 4.68567610e-01 3.28776777e-01
-1.18962085e+00 2.38744184e-01 -7.57255256e-01 -2.39661053e-01
-2.63399258e-03 -7.47727603e-02 -1.97205737e-01 3.56351972e-01
-9.46110785e-01 3.17978650e-01 1.61037128e-02 -3.45937580e-01
-1.26952207e+00 -5.87014496e-01 -8.82150352e-01 9.28231105e-02
1.24735422e-02 -7.65520990e-01 1.50035048e+00 -1.12675297e+00
-1.34154570e+00 6.00024760e-01 -4.63500887e-01 -1.22410834e-01
1.09478116e-01 9.22608748e-02 -3.07769418e-01 5.06563187e-01
2.81520933e-01 9.63533103e-01 1.14447391e+00 -1.71374989e+00
-7.45710552e-01 6.59652725e-02 -9.38109234e-02 1.82912633e-01
-1.28983185e-01 1.52035251e-01 -5.88304996e-01 -5.92179477e-01
2.24004075e-01 -9.73176003e-01 -1.03363255e-02 -4.26820815e-02
-4.13176537e-01 1.71122983e-01 4.66254383e-01 -3.81496429e-01
9.67963994e-01 -2.49441147e+00 -7.69063225e-03 3.19964349e-01
3.36435661e-02 -3.86533886e-01 -5.46474278e-01 4.51747626e-01
-1.23563744e-01 2.47070435e-02 -8.79757479e-02 -6.08575761e-01
2.79471669e-02 -4.31681462e-02 -1.00697947e+00 2.59107143e-01
1.13916278e-01 4.75856215e-01 -8.90473127e-01 -6.32908642e-01
-1.63921967e-01 5.74079573e-01 -8.85075867e-01 4.24313307e-01
-3.16167414e-01 4.48024392e-01 -1.82099342e-01 4.20258135e-01
4.65719670e-01 -1.82352290e-01 1.72691107e-01 -1.45119876e-01
-2.69885898e-01 4.90215153e-01 -1.26741445e+00 1.81260002e+00
-6.87354386e-01 1.00418961e+00 2.05522582e-01 -5.37117302e-01
8.14440966e-01 4.94398743e-01 3.84375930e-01 -4.33333635e-01
-2.48695672e-01 4.21951478e-03 -2.75303572e-01 -6.34999812e-01
3.05407912e-01 -4.13938135e-01 -1.53158024e-01 6.86579287e-01
3.26149195e-01 -5.28621376e-01 -8.07647482e-02 4.06503111e-01
8.52875769e-01 3.75327580e-02 -1.68856084e-01 2.63902664e-01
-9.22519714e-04 -2.38281175e-01 2.41944373e-01 7.38903224e-01
2.70226747e-01 1.11114681e+00 6.24076068e-01 4.32558022e-02
-1.13682735e+00 -1.59025824e+00 -1.39056647e-03 1.30972588e+00
-1.02837831e-01 -6.97991908e-01 -6.81691468e-01 -2.40582600e-01
-2.92678833e-01 7.12285221e-01 -4.34563875e-01 -3.65322568e-02
-2.10940480e-01 -2.26126671e-01 7.46545792e-01 4.93790150e-01
-6.42433390e-03 -1.00925636e+00 -5.84102392e-01 3.15617658e-02
-6.53115869e-01 -9.76591587e-01 -4.93215561e-01 1.79253086e-01
-3.69094461e-01 -6.36402190e-01 -5.45759976e-01 -7.94178545e-01
5.65567434e-01 3.05339485e-01 1.09987450e+00 -2.48089507e-01
-3.39412451e-01 5.55899441e-01 -2.95903265e-01 -4.29510742e-01
-4.89169121e-01 -2.08165094e-01 -6.76135644e-02 2.11971834e-01
-2.17382669e-01 -8.51106226e-01 -3.23627204e-01 1.08528167e-01
-8.33385646e-01 3.01271468e-01 1.32390723e-01 6.26368463e-01
5.93622148e-01 1.54101029e-01 4.37239736e-01 -4.10230666e-01
4.55432653e-01 -3.88711184e-01 -6.33999228e-01 6.71353489e-02
9.65806022e-02 8.72841105e-02 4.04047251e-01 -4.96691138e-01
-8.76452148e-01 4.91116852e-01 7.68226311e-02 -4.46676880e-01
-2.56816417e-01 2.42834210e-01 -2.52656847e-01 2.94435740e-01
7.44092643e-01 4.74984832e-02 -2.59343356e-01 -5.05715013e-01
6.67951763e-01 9.30600345e-01 8.27995121e-01 -5.58462918e-01
8.37701380e-01 5.92937171e-01 -1.59687370e-01 -9.57862854e-01
-8.26867342e-01 -2.83199817e-01 -4.07980323e-01 -2.31923744e-01
8.72937977e-01 -1.09217811e+00 -4.82134342e-01 2.09548950e-01
-1.34489775e+00 -4.83377546e-01 -2.98980951e-01 5.38029194e-01
-8.28660965e-01 5.21879382e-02 -3.58157486e-01 -1.01259243e+00
1.88891932e-01 -9.94693339e-01 1.25732112e+00 -4.64262031e-02
-3.33613247e-01 -6.06715441e-01 5.16397834e-01 -9.11702495e-03
2.58887649e-01 -2.49208976e-02 9.74117756e-01 -7.67791197e-02
-6.52015507e-01 2.60651022e-01 -2.05779031e-01 1.71308085e-01
1.27709582e-01 3.38480502e-01 -1.53849089e+00 -5.36193363e-02
-1.48554817e-01 -4.29058433e-01 9.86653864e-01 4.99668986e-01
1.13798928e+00 -2.17138484e-01 -2.44441763e-01 7.35463798e-01
1.22566879e+00 2.80844361e-01 4.65814263e-01 4.65441160e-02
5.18773019e-01 8.54527771e-01 3.08314443e-01 5.42927265e-01
3.95115227e-01 8.31667364e-01 2.72460133e-01 -1.06048815e-01
-3.99976403e-01 -7.49515355e-01 4.75250721e-01 6.79142177e-01
3.22454065e-01 -2.22028926e-01 -9.03590977e-01 7.21225202e-01
-1.59588337e+00 -1.25144207e+00 1.67338789e-01 2.02014470e+00
8.92880261e-01 -1.10920548e-01 1.17032029e-01 1.71959415e-01
6.21923923e-01 9.77400243e-02 -1.10097351e-02 -1.77254558e-01
-7.87834749e-02 2.78481394e-01 1.24585181e-02 8.53434622e-01
-1.15388811e+00 7.87406921e-01 7.63758326e+00 5.73172629e-01
-1.29512691e+00 4.78954148e-03 3.13400298e-01 -4.75281090e-01
-7.56493747e-01 2.17950225e-01 -4.42444474e-01 2.29607120e-01
9.36537087e-01 -5.95883280e-03 6.96088254e-01 4.31098908e-01
3.89108777e-01 -9.17536300e-03 -1.22287500e+00 1.15183258e+00
2.02823088e-01 -1.35877800e+00 9.13622752e-02 -1.68423668e-01
6.43121243e-01 -1.90398082e-01 3.86926532e-01 -1.45441249e-01
3.51618975e-01 -1.17742741e+00 1.18079400e+00 5.98313332e-01
1.00646567e+00 -7.03308046e-01 -3.26371230e-02 1.77964017e-01
-1.30690956e+00 -7.84792230e-02 -2.61794835e-01 -8.66921805e-03
2.85348088e-01 3.22190523e-01 -8.99260819e-01 1.25892684e-01
7.89375722e-01 3.31549972e-01 -5.40808797e-01 1.18257809e+00
-3.32375377e-01 7.53172576e-01 -3.60728860e-01 4.33951885e-01
-7.14697167e-02 1.82701066e-01 4.81025130e-01 1.18430555e+00
8.36912453e-01 -2.16758534e-01 1.91874698e-01 1.01827443e+00
1.13631181e-01 4.96936776e-02 -9.30066586e-01 -1.22182406e-01
6.93011522e-01 1.21342373e+00 -5.29810250e-01 -2.73403138e-01
-2.32614711e-01 8.42853248e-01 -1.99613094e-01 4.36495453e-01
-8.66114795e-01 -6.48485363e-01 4.22085255e-01 1.46025226e-01
2.79191017e-01 -1.86092779e-01 -1.24116682e-01 -9.53136623e-01
-9.17924866e-02 -7.47866154e-01 1.11766838e-01 -1.45918274e+00
-1.02199602e+00 7.88141787e-01 1.04173303e-01 -1.45428562e+00
-6.05645180e-01 -1.53527530e-02 -6.87806785e-01 8.90120327e-01
-1.20746386e+00 -1.34906518e+00 -2.90930927e-01 6.09962940e-01
3.84010702e-01 -1.09421365e-01 8.67259800e-01 2.01757237e-01
3.46592106e-02 4.01000381e-01 -1.65106699e-01 8.29864740e-02
9.59149182e-01 -1.19174206e+00 4.28416342e-01 7.00097859e-01
7.84297466e-01 4.45030361e-01 9.72925365e-01 -3.47691655e-01
-8.50208223e-01 -9.22568381e-01 9.10858452e-01 -3.86944920e-01
6.83043778e-01 -6.03583872e-01 -5.65562308e-01 8.44322383e-01
4.08699602e-01 -6.16993718e-02 9.48752165e-01 1.38726383e-02
-7.41503298e-01 -1.61464855e-01 -5.99202573e-01 5.24950325e-01
7.82845318e-01 -1.11959791e+00 -5.55483580e-01 9.43373442e-02
8.58537376e-01 -1.69769987e-01 -1.89998269e-01 -1.58464704e-02
6.53786004e-01 -8.24784458e-01 1.00589323e+00 -4.25118685e-01
4.78773981e-01 -6.71238959e-01 -4.28074569e-01 -1.44962597e+00
-2.80538321e-01 -8.80850375e-01 3.18772435e-01 1.55395734e+00
5.18521070e-01 -1.31498456e-01 4.39162850e-01 5.87435812e-02
-8.03868398e-02 9.06879604e-02 -7.37390220e-01 -5.19810736e-01
-1.48917601e-01 -7.69138992e-01 6.34153426e-01 7.33593285e-01
1.85135856e-01 4.94048029e-01 -4.61582273e-01 4.67941046e-01
6.46578193e-01 4.07364994e-01 8.55332434e-01 -1.05109346e+00
-7.37055600e-01 -1.45214021e-01 -2.44457975e-01 -9.20731008e-01
1.49158254e-01 -7.33357131e-01 4.70632672e-01 -1.41602945e+00
2.57216603e-01 -4.40507174e-01 -2.28455260e-01 5.72845519e-01
2.05077931e-01 6.78777516e-01 3.61809880e-01 1.22519173e-01
-6.05077326e-01 2.84607857e-01 1.02748382e+00 -1.33087978e-01
-3.57478231e-01 -1.27881706e-01 -7.63334930e-01 8.07498932e-01
7.08300650e-01 -5.16948700e-01 -6.36134148e-01 -5.90314746e-01
3.79413396e-01 3.47932011e-01 7.39144504e-01 -1.11702013e+00
2.47194931e-01 -1.70275345e-01 4.27727431e-01 -5.80296397e-01
6.12091064e-01 -5.54269731e-01 3.47384065e-01 1.62795056e-02
-8.01156938e-01 -4.54380699e-02 1.48618981e-01 2.75614291e-01
-4.63635653e-01 -1.30797803e-01 8.24204922e-01 -4.05154424e-03
-1.99244455e-01 -4.85608876e-02 -6.67525053e-01 3.13312723e-03
6.30049586e-01 2.38233373e-01 -2.45818689e-01 -9.36020136e-01
-9.92082894e-01 -2.34800968e-02 4.18930888e-01 4.06745762e-01
5.73056519e-01 -1.62092650e+00 -5.81464052e-01 4.06871557e-01
1.95707858e-01 -3.31105471e-01 2.70134479e-01 5.01388848e-01
-3.49456429e-01 2.41144732e-01 -2.33799174e-01 -6.98961079e-01
-1.44804907e+00 3.26023161e-01 2.96439886e-01 3.79248470e-01
-3.90484810e-01 9.22123611e-01 5.34394562e-01 2.10887138e-02
5.28867066e-01 -1.76967308e-01 -1.99647263e-01 3.00627440e-01
6.53233945e-01 7.39277601e-02 -2.78856963e-01 -7.58726120e-01
-3.39571089e-01 6.52819932e-01 4.93859529e-01 -9.43578482e-01
1.27924550e+00 -2.53237754e-01 -1.17513752e-02 8.32681656e-01
1.26479733e+00 5.98801851e-01 -1.39635432e+00 -3.73181850e-02
-5.01844704e-01 -4.83506590e-01 -4.89049144e-02 -5.42603374e-01
-8.68826568e-01 1.32607973e+00 6.89435184e-01 5.24870336e-01
1.16823947e+00 4.75858539e-01 4.41212654e-01 4.06281501e-02
1.99109033e-01 -8.05582225e-01 5.49880683e-01 3.91972095e-01
8.79443407e-01 -6.86004817e-01 -3.88714671e-01 -3.04801881e-01
-6.85655117e-01 1.07927978e+00 3.09218973e-01 8.25063586e-02
5.56498468e-01 6.47588611e-01 4.72687632e-01 -7.22124055e-02
-1.03422534e+00 -3.10166329e-01 2.29545400e-01 7.94444144e-01
4.27517176e-01 -5.68805635e-02 5.82445920e-01 5.39467812e-01
-7.04304039e-01 -1.66635498e-01 5.86747885e-01 6.63536489e-01
-7.43111372e-01 -1.09784687e+00 -6.52197301e-01 -5.80158830e-02
-4.18536097e-01 -1.23908006e-01 -4.92403299e-01 2.41683364e-01
2.94807404e-01 1.21597528e+00 4.19599384e-01 -5.42014599e-01
7.36669153e-02 1.86856985e-01 6.64720774e-01 -7.37109840e-01
-6.28339425e-02 8.63422394e-01 2.30442975e-02 -4.88404155e-01
-4.16845709e-01 -6.33829296e-01 -1.27690327e+00 2.08485901e-01
-1.98872805e-01 3.88535351e-01 6.32419050e-01 6.13006055e-01
2.57824808e-01 5.38002670e-01 9.11734104e-01 -1.12360585e+00
-2.05991879e-01 -6.73778117e-01 -6.29231274e-01 2.67230481e-01
7.20232427e-01 -5.46811879e-01 -6.15756392e-01 7.13549793e-01]
|
[15.203411102294922, 5.129062652587891]
|
5b1734ce-920f-4a6e-ac9a-a0fccfe9c656
|
bsrt-improving-burst-super-resolution-with
|
2204.08332
| null |
https://arxiv.org/abs/2204.08332v2
|
https://arxiv.org/pdf/2204.08332v2.pdf
|
BSRT: Improving Burst Super-Resolution with Swin Transformer and Flow-Guided Deformable Alignment
|
This work addresses the Burst Super-Resolution (BurstSR) task using a new architecture, which requires restoring a high-quality image from a sequence of noisy, misaligned, and low-resolution RAW bursts. To overcome the challenges in BurstSR, we propose a Burst Super-Resolution Transformer (BSRT), which can significantly improve the capability of extracting inter-frame information and reconstruction. To achieve this goal, we propose a Pyramid Flow-Guided Deformable Convolution Network (Pyramid FG-DCN) and incorporate Swin Transformer Blocks and Groups as our main backbone. More specifically, we combine optical flows and deformable convolutions, hence our BSRT can handle misalignment and aggregate the potential texture information in multi-frames more efficiently. In addition, our Transformer-based structure can capture long-range dependency to further improve the performance. The evaluation on both synthetic and real-world tracks demonstrates that our approach achieves a new state-of-the-art in BurstSR task. Further, our BSRT wins the championship in the NTIRE2022 Burst Super-Resolution Challenge.
|
['Shuaicheng Liu', 'Jian Sun', 'Haoqiang Fan', 'Zhihong Wen', 'Qi Wu', 'Lei Yu', 'Shen Cheng', 'Youwei Li', 'Ziwei Luo']
|
2022-04-18
| null | null | null | null |
['multi-frame-super-resolution', 'burst-image-super-resolution']
|
['computer-vision', 'computer-vision']
|
[ 1.21700831e-01 -6.03643894e-01 4.73757610e-02 -2.96744436e-01
-7.28202760e-01 -3.13618660e-01 4.23488528e-01 -7.33827710e-01
-1.03921510e-01 8.77837598e-01 6.92923307e-01 2.77430952e-01
-6.08869568e-02 -7.91714847e-01 -8.55210960e-01 -4.55031395e-01
-6.84769675e-02 -1.44087419e-01 7.43360519e-01 -5.02182186e-01
1.48779042e-02 7.22153246e-01 -1.66051197e+00 8.25027406e-01
8.72677028e-01 1.11134875e+00 4.96464968e-01 7.94009686e-01
8.40178803e-02 1.37017858e+00 -2.86076218e-01 -1.82755351e-01
3.96897316e-01 -3.10709178e-01 -7.30210960e-01 7.70735964e-02
8.88159573e-01 -8.01704168e-01 -8.63173246e-01 9.25729871e-01
4.80075687e-01 3.43640745e-01 -1.00777015e-01 -7.78553486e-01
-8.59301448e-01 5.17367423e-01 -5.36939383e-01 9.62444246e-01
2.39895120e-01 4.50207949e-01 8.01691651e-01 -1.02183461e+00
6.50992632e-01 1.58902133e+00 8.00199330e-01 4.78143036e-01
-1.05139863e+00 -8.55077982e-01 1.66996896e-01 3.74012053e-01
-1.08990026e+00 -7.23971546e-01 3.42396468e-01 -1.83798194e-01
7.60836542e-01 2.52897352e-01 5.02430081e-01 1.20516503e+00
-5.17315194e-02 5.25175273e-01 1.03632021e+00 3.02328259e-01
-2.39383772e-01 -7.02542365e-01 -1.05029456e-01 4.80298072e-01
4.56335619e-02 3.58629942e-01 -1.02713883e+00 2.09842667e-01
1.47198653e+00 7.89815187e-02 -5.73964536e-01 5.14997602e-01
-1.42896795e+00 3.57335925e-01 5.29334188e-01 2.65599102e-01
-2.70667404e-01 3.68137926e-01 2.37463623e-01 1.85726240e-01
7.05980420e-01 6.82137236e-02 -2.56088935e-02 -2.94106215e-01
-1.13109016e+00 3.71703863e-01 1.87364042e-01 7.97445297e-01
6.87637508e-01 3.81359041e-01 -5.27000844e-01 8.33216667e-01
-8.71423259e-02 4.17946011e-01 2.72879839e-01 -1.46253610e+00
4.65498775e-01 1.21397957e-01 4.56914693e-01 -1.09975529e+00
-2.47882590e-01 -6.12883270e-01 -1.07279611e+00 -5.00361323e-02
1.62399694e-01 1.17110275e-01 -7.59181261e-01 1.61345029e+00
2.78805375e-01 9.79971230e-01 -3.75159606e-02 1.50823951e+00
1.09382915e+00 8.09635341e-01 -3.23902935e-01 -4.01198238e-01
1.26061022e+00 -1.14371419e+00 -9.65677142e-01 -9.03653875e-02
-1.63461968e-01 -8.81249189e-01 8.18932593e-01 3.18456411e-01
-1.35622239e+00 -1.05228746e+00 -9.06095982e-01 -3.21713895e-01
4.13410008e-01 -1.52767226e-01 4.17880774e-01 -5.20759970e-02
-1.01423085e+00 9.21755552e-01 -9.55984950e-01 -2.23705452e-02
4.89734381e-01 -1.85133711e-01 -2.65579104e-01 -4.26055402e-01
-1.21593094e+00 4.45177436e-01 1.24963261e-01 2.19185084e-01
-9.04741585e-01 -1.10276210e+00 -6.53001726e-01 -7.96421245e-02
3.89150500e-01 -7.15912640e-01 9.73600090e-01 -6.61594868e-01
-1.51062429e+00 3.41199249e-01 -2.43339255e-01 -5.82842052e-01
5.86255491e-01 -4.69458789e-01 -7.29873300e-01 3.72935086e-01
7.79318139e-02 3.29192281e-01 8.58992100e-01 -1.02086818e+00
-8.85910630e-01 -1.26035372e-03 9.37891901e-02 9.00961757e-02
-1.32363051e-01 3.06821704e-01 -5.88096380e-01 -1.14949811e+00
-1.06776468e-01 -6.19332135e-01 -7.54075721e-02 -1.64317712e-02
1.53170861e-02 1.14529915e-01 1.01324785e+00 -1.00731826e+00
1.33713996e+00 -2.24523449e+00 2.08347306e-01 -4.73537803e-01
5.88738561e-01 4.08794731e-01 -2.41255537e-01 2.05660816e-02
1.25573114e-01 -1.03994116e-01 -1.26328647e-01 -5.58038652e-01
-3.20447087e-01 4.72049356e-01 -6.74592435e-01 3.53954554e-01
2.95856118e-01 8.51177156e-01 -9.45753157e-01 -3.07219625e-01
1.85132697e-01 8.44882607e-01 -5.78758538e-01 4.51361477e-01
-9.62907076e-02 7.93806255e-01 -3.01176429e-01 5.29791117e-01
1.01733041e+00 -5.79774857e-01 -1.82567224e-01 -8.00537944e-01
-4.89878535e-01 2.41762251e-01 -1.34952307e+00 1.83860099e+00
-4.01567698e-01 7.82567203e-01 2.52884865e-01 -5.47102630e-01
9.44148779e-01 1.08352723e-03 5.57865918e-01 -1.09461868e+00
-3.03126693e-01 1.59742713e-01 -3.58702034e-01 -4.71339017e-01
1.01709211e+00 1.08641438e-01 4.43001270e-01 2.58771777e-01
-1.25899732e-01 4.54916388e-01 2.11817622e-01 3.29437107e-01
1.24307442e+00 3.99483800e-01 -2.96724826e-01 -1.13577917e-01
4.45957661e-01 -4.02774394e-01 1.18516505e+00 6.02239966e-01
-1.60845280e-01 1.01798105e+00 7.16705248e-02 -7.71542668e-01
-1.21265197e+00 -1.21313202e+00 1.87996149e-01 8.47150385e-01
4.26078975e-01 -5.13222873e-01 -4.09920186e-01 -3.23160589e-01
-1.68593034e-01 6.80444837e-02 -3.79810870e-01 1.64055109e-01
-1.13391328e+00 -7.67762840e-01 4.69178408e-01 6.42955422e-01
9.91386712e-01 -7.66756594e-01 -5.87121904e-01 4.98712450e-01
-9.77387309e-01 -1.88770306e+00 -8.42610896e-01 -6.13595843e-01
-6.28811181e-01 -9.02962565e-01 -5.66612601e-01 -3.29186887e-01
2.73320102e-03 5.88822305e-01 1.31879294e+00 1.31902307e-01
-3.81190300e-01 3.03381830e-02 -5.39301991e-01 4.00117427e-01
-2.21189559e-01 -3.46445978e-01 3.87758366e-03 6.40572250e-01
-2.79232174e-01 -8.84418905e-01 -9.10084426e-01 5.37131429e-01
-1.00941980e+00 3.35269839e-01 3.23626250e-01 7.83767343e-01
9.45996165e-01 -4.62795086e-02 4.88801420e-01 -4.22735840e-01
3.78499746e-01 -2.36243337e-01 -5.41567504e-01 7.05884323e-02
-1.69576913e-01 -8.31049457e-02 7.27201879e-01 -5.46277225e-01
-1.40130556e+00 -3.75781268e-01 8.37800838e-03 -8.38287652e-01
2.08178937e-01 -1.17164496e-02 1.00462727e-01 -2.11069271e-01
5.77489018e-01 3.48712832e-01 -1.08740255e-01 -8.11871707e-01
2.80426741e-01 4.97586012e-01 1.06265163e+00 -5.31391203e-01
8.60193014e-01 1.02959514e+00 2.26457175e-02 -5.42145312e-01
-1.17150795e+00 -3.64142090e-01 -2.94686526e-01 -4.11782026e-01
9.17467535e-01 -1.45967305e+00 -7.44273722e-01 6.52344465e-01
-1.17367256e+00 -5.89051664e-01 -2.86039263e-01 4.56189185e-01
-5.93219459e-01 4.77898389e-01 -1.03765690e+00 -4.76815075e-01
-4.37779844e-01 -1.14831269e+00 1.04039097e+00 4.54112679e-01
3.66995603e-01 -4.87884104e-01 -1.46733541e-02 3.92568529e-01
9.94143069e-01 4.31241691e-01 2.46819873e-05 2.08154291e-01
-1.28657472e+00 6.05511189e-01 -8.65412951e-01 4.85269576e-01
2.53195852e-01 -1.69432983e-01 -7.25098372e-01 -5.18247247e-01
-3.31073739e-02 -2.37785771e-01 1.06829369e+00 1.43086776e-01
1.23495770e+00 -2.52599090e-01 8.54013339e-02 1.32324123e+00
1.54703104e+00 -2.84729064e-01 9.98133242e-01 2.83837318e-01
9.56277370e-01 1.16782777e-01 7.01520383e-01 6.79098248e-01
5.93697667e-01 9.75574553e-01 4.78623331e-01 -1.45798251e-01
-9.27203834e-01 3.48683610e-03 6.06128752e-01 8.02420437e-01
-5.83137751e-01 -6.83494210e-02 -3.61827284e-01 4.75518405e-01
-2.10330534e+00 -1.40675914e+00 -2.18237653e-01 1.69592178e+00
8.39913368e-01 -8.20418894e-02 1.36399910e-01 -2.87039101e-01
7.33696163e-01 6.46513224e-01 -4.60105866e-01 1.25674918e-01
-5.31026840e-01 2.23861739e-01 4.67303991e-01 5.50574243e-01
-9.05896008e-01 9.52379346e-01 5.74818230e+00 1.14032936e+00
-1.20287943e+00 2.83772141e-01 4.66361701e-01 -2.40021959e-01
-2.99732506e-01 -2.28719860e-01 -1.03428435e+00 7.50642776e-01
8.55174482e-01 -7.55828097e-02 9.30469096e-01 2.35844135e-01
5.85179389e-01 1.61347926e-01 -6.29021287e-01 1.20736372e+00
5.36816530e-02 -2.03738618e+00 7.06300065e-02 -1.12302706e-01
7.91842282e-01 2.68416077e-01 1.13511115e-01 -1.11304577e-02
3.91494036e-01 -1.06332064e+00 7.00750411e-01 9.54146564e-01
1.10476255e+00 -7.83643365e-01 4.27064717e-01 9.17926058e-02
-1.81888521e+00 -1.77441821e-01 -4.40875977e-01 2.43095346e-02
5.83043635e-01 7.77557313e-01 -1.28504159e-02 1.03908312e+00
1.30011129e+00 1.35068989e+00 -4.18406546e-01 9.01199400e-01
1.83144674e-01 4.55415815e-01 -2.99997181e-01 8.71917844e-01
6.06077090e-02 -4.86526154e-02 8.20872784e-01 1.39919364e+00
3.62897962e-01 3.53316367e-01 4.23497409e-01 1.06689000e+00
-7.96084180e-02 -5.14699697e-01 1.34401157e-01 2.17348516e-01
4.69042569e-01 1.43596613e+00 -4.47292030e-01 -3.53176385e-01
-5.23005247e-01 1.08800173e+00 3.61487687e-01 3.34628075e-01
-1.16610754e+00 1.92171931e-02 1.19587326e+00 3.04755658e-01
6.42397702e-01 -3.88438165e-01 -3.75727937e-03 -1.53819168e+00
1.80908754e-01 -1.02408564e+00 2.13422239e-01 -8.67248416e-01
-1.44219458e+00 8.75207365e-01 -2.82963157e-01 -1.44679368e+00
1.52486742e-01 -1.07152291e-01 -3.46325189e-01 8.95572484e-01
-2.12110472e+00 -1.20178521e+00 -9.38830614e-01 1.02889144e+00
7.50160038e-01 7.82861635e-02 1.72774792e-01 8.06033015e-01
-4.42459762e-01 4.46129054e-01 -1.06025577e-01 1.41945779e-01
9.84937310e-01 -7.98738718e-01 6.23925090e-01 1.34676170e+00
-1.54472038e-01 2.09231526e-01 5.45472503e-01 -7.25423157e-01
-1.27836835e+00 -1.64355850e+00 3.82050693e-01 -1.01535395e-01
6.26886845e-01 -4.68346961e-02 -1.11951900e+00 6.50184870e-01
-2.17742268e-02 8.44885409e-01 6.67777061e-02 -5.88960946e-01
-6.44520938e-01 -3.41723472e-01 -9.52991962e-01 3.57776493e-01
1.58813632e+00 -4.91435677e-01 -1.99569881e-01 1.76848188e-01
1.21986926e+00 -7.87261426e-01 -1.14256668e+00 6.46395028e-01
3.21482211e-01 -1.47673678e+00 1.32936490e+00 -3.11834127e-01
6.75756335e-01 -6.71259940e-01 -4.04733360e-01 -1.03731775e+00
-6.45045161e-01 -1.14232731e+00 -5.29238999e-01 1.27041399e+00
-3.94644886e-01 -3.59486192e-01 5.40980816e-01 7.51615092e-02
-3.60453010e-01 -5.46654403e-01 -9.05418575e-01 -8.79922569e-01
-4.26111519e-01 -1.94234788e-01 8.23779643e-01 8.60947549e-01
-7.40227878e-01 -2.84321122e-02 -9.72086430e-01 3.64980251e-01
1.07070971e+00 2.43412480e-01 7.43879855e-01 -8.19415212e-01
-5.49939573e-01 -5.08662537e-02 -2.45397940e-01 -1.43080783e+00
-1.94062442e-01 -5.50755501e-01 -1.36284441e-01 -1.41425896e+00
4.63729016e-02 -4.06393826e-01 -1.41534269e-01 1.43034592e-01
-2.89744765e-01 6.41497552e-01 6.00009024e-01 4.05267596e-01
-8.47134709e-01 4.92026240e-01 1.62406909e+00 2.69886702e-01
-9.87923741e-02 -3.42632920e-01 -5.35191715e-01 5.76284647e-01
4.18418825e-01 -1.35046929e-01 -1.11379042e-01 -9.05372262e-01
6.27214015e-02 4.04983610e-01 6.95473433e-01 -1.00616431e+00
2.82780588e-01 -3.17041814e-01 3.90327960e-01 -6.41232550e-01
4.34275687e-01 -4.92190182e-01 5.24214864e-01 3.23347747e-02
-1.36535943e-01 2.27562398e-01 1.11045815e-01 7.17839181e-01
-3.41709048e-01 6.12011731e-01 1.07128739e+00 -2.12099418e-01
-8.76754940e-01 8.12390506e-01 8.65388587e-02 3.11544746e-01
5.70695996e-01 -8.58538374e-02 -7.82874703e-01 -2.17295900e-01
-6.81384742e-01 2.55074769e-01 4.09857303e-01 4.69816178e-01
7.66301334e-01 -1.58216882e+00 -1.17125583e+00 2.00056985e-01
-3.02852511e-01 2.48664543e-01 8.72034967e-01 8.95220041e-01
-5.04238546e-01 -2.40762427e-01 -4.97787088e-01 -6.51703656e-01
-1.10739660e+00 3.89821768e-01 4.38409775e-01 -5.31725049e-01
-1.25073910e+00 7.83473730e-01 2.36394465e-01 2.68163204e-01
-8.95260181e-03 -1.78428113e-01 -1.98728636e-01 -1.38699546e-01
1.18448555e+00 7.23418713e-01 -1.11763686e-01 -7.46048152e-01
-2.67761238e-02 6.76344097e-01 -1.16595946e-01 1.25293747e-01
1.60325766e+00 -6.06557846e-01 -1.07216381e-01 6.86398819e-02
9.20632422e-01 1.51837766e-01 -1.96153712e+00 -6.45457804e-01
-4.84311402e-01 -9.33630347e-01 1.58535749e-01 -5.81684291e-01
-1.50413084e+00 5.43855786e-01 4.44332838e-01 3.69346817e-03
1.37805927e+00 -2.89766401e-01 1.49747252e+00 -2.00217471e-01
2.48147979e-01 -8.45097184e-01 4.26411629e-01 6.33837044e-01
9.13961470e-01 -9.87788975e-01 -6.19373769e-02 -6.32133126e-01
-4.33566630e-01 1.33675194e+00 6.26915574e-01 -5.13323724e-01
4.40912664e-01 5.77296853e-01 -2.08488643e-01 1.14493802e-01
-9.51222599e-01 -3.39047045e-01 2.07733467e-01 5.74470580e-01
6.79398477e-02 -2.40998492e-01 -4.24208008e-02 5.67890704e-01
4.07086983e-02 4.10159826e-01 7.72458255e-01 5.26672840e-01
-4.25453186e-01 -7.78349161e-01 -3.75074267e-01 1.50555139e-02
-5.57563007e-01 -2.42421255e-01 3.57830614e-01 2.88806319e-01
1.59046963e-01 8.26811552e-01 2.62943089e-01 -5.54129779e-01
3.49466026e-01 -5.67890108e-01 4.58294749e-01 -1.39935702e-01
-5.33185363e-01 2.46428326e-01 4.95644361e-02 -1.40352416e+00
-8.30245197e-01 -3.34084183e-01 -1.11203897e+00 -7.67967582e-01
1.50971845e-01 -3.00198108e-01 1.82526316e-02 8.73424768e-01
6.99080646e-01 9.66534376e-01 6.66295528e-01 -1.07476377e+00
-2.82248706e-01 -7.02800035e-01 -3.80518168e-01 5.59653461e-01
8.11961770e-01 -5.47091782e-01 -3.29879969e-01 3.55136037e-01]
|
[11.02558708190918, -1.9052196741104126]
|
40fa00fc-e0da-4120-802e-f7f030a7b4bc
|
negation-detection-in-dutch-spoken-human
| null | null |
https://aclanthology.org/2022.lrec-1.56
|
https://aclanthology.org/2022.lrec-1.56.pdf
|
Negation Detection in Dutch Spoken Human-Computer Conversations
|
Proper recognition and interpretation of negation signals in text or communication is crucial for any form of full natural language understanding. It is also essential for computational approaches to natural language processing. In this study we focus on negation detection in Dutch spoken human-computer conversations. Since there exists no Dutch (dialogue) corpus annotated for negation we have annotated a Dutch corpus sample to evaluate our method for automatic negation detection. We use transfer learning and trained NegBERT (an existing BERT implementation used for negation detection) on English data with multilingual BERT to detect negation in Dutch dialogues. Our results show that adding in-domain training material improves the results. We show that we can detect both negation cues and scope in Dutch dialogues with high precision and recall. We provide a detailed error analysis and discuss the effects of cross-lingual and cross-domain transfer learning on automatic negation detection.
|
['Helmer Strik', 'Iris Hendrickx', 'Tom Sweers']
| null | null | null | null |
lrec-2022-6
|
['negation-detection']
|
['natural-language-processing']
|
[ 2.15974897e-01 2.67326146e-01 8.71450230e-02 -7.53167927e-01
-9.32508945e-01 -8.49597633e-01 7.48831213e-01 3.43316197e-01
-1.06537783e+00 1.12092447e+00 4.66863751e-01 -5.00041187e-01
4.68454808e-01 -6.71840072e-01 -5.13035059e-01 -4.52311113e-02
1.29443750e-01 6.37996078e-01 4.51114208e-01 -9.30670559e-01
-4.32426259e-02 2.18880221e-01 -9.14973199e-01 1.03449762e+00
7.12551236e-01 2.30489537e-01 -6.62570074e-02 9.75145340e-01
-2.27071583e-01 8.50545585e-01 -1.07157540e+00 -6.92188680e-01
-1.98928230e-02 -8.31456959e-01 -9.73553002e-01 -2.13248283e-01
2.51616597e-01 -3.05864453e-01 2.78933764e-01 1.06626034e+00
4.63605076e-01 -1.13337144e-01 7.52421916e-01 -9.96702671e-01
-9.48770493e-02 7.90465415e-01 -7.97173232e-02 3.78272355e-01
8.80711019e-01 1.07825398e-01 1.06362236e+00 -7.93666363e-01
9.29744840e-01 1.33186877e+00 7.32115686e-01 7.83538640e-01
-1.11117280e+00 -5.34390748e-01 -1.81460500e-01 1.03318095e-01
-1.10068262e+00 -6.28570735e-01 5.47699749e-01 -2.44340435e-01
1.58846867e+00 6.28617313e-03 6.12944543e-01 1.04106009e+00
8.14912096e-02 7.80339360e-01 9.13351297e-01 -1.03592074e+00
-7.07538947e-02 3.81371081e-01 -1.66779999e-02 5.05461037e-01
-2.67930955e-01 -1.10273011e-01 -6.38330936e-01 -1.72947362e-01
2.44323343e-01 -9.50034022e-01 -3.14032316e-01 9.96581763e-02
-1.30785215e+00 1.00303125e+00 -7.20655173e-02 7.05185652e-01
-1.34432703e-01 -2.20428154e-01 1.15572572e+00 1.02191484e+00
4.79947001e-01 4.08042699e-01 -9.93376613e-01 -6.35050476e-01
-4.47722107e-01 3.01490068e-01 1.55169499e+00 8.16819966e-01
5.41885734e-01 -3.08758691e-02 -1.42888334e-02 1.05883706e+00
-1.16475008e-01 7.06662774e-01 3.10530484e-01 -8.23130369e-01
6.04365528e-01 5.31530559e-01 2.63156235e-01 -7.61750042e-01
-5.99291146e-01 3.75400662e-01 -3.36760521e-01 -4.35010195e-02
9.04063106e-01 -7.43968308e-01 -3.62740457e-01 1.75596368e+00
2.17663005e-01 -5.60735345e-01 7.19830155e-01 6.84467793e-01
6.64605856e-01 6.45948231e-01 8.00217167e-02 -4.15121943e-01
1.48918450e+00 -4.15779024e-01 -1.20352793e+00 -1.42420292e-01
1.31124449e+00 -1.13369966e+00 1.14020622e+00 5.96063972e-01
-9.42774832e-01 -1.35665327e-01 -8.19895327e-01 -1.31566435e-01
-4.52540249e-01 1.21855669e-01 4.76939887e-01 8.63300204e-01
-7.80415177e-01 1.81157053e-01 -6.18523240e-01 -6.39584064e-01
-2.48620614e-01 4.24589485e-01 -5.86428165e-01 -1.43609479e-01
-1.66935611e+00 1.17878175e+00 5.05334139e-01 1.05489291e-01
-1.65502563e-01 -3.05972230e-02 -1.27425325e+00 -2.90934205e-01
4.36392784e-01 -6.80438355e-02 1.95830631e+00 -1.38050807e+00
-1.35011399e+00 1.27600908e+00 -3.45943749e-01 -5.14006853e-01
6.04553640e-01 -2.16679782e-01 -5.78582048e-01 1.30702313e-02
1.28975093e-01 6.42093003e-01 3.66020232e-01 -7.35726297e-01
-8.08330774e-01 -6.52507842e-02 1.16451405e-01 2.93630630e-01
-1.40355572e-01 5.85808754e-01 -6.48649335e-02 -3.91247988e-01
-1.58877999e-01 -6.43818319e-01 4.07075375e-01 -2.50851065e-01
-6.02136031e-02 -6.10181689e-01 7.46117592e-01 -6.01352096e-01
9.24293280e-01 -1.94857228e+00 -9.67960358e-02 6.69149980e-02
-2.06296772e-01 5.33504725e-01 -1.21811278e-01 6.76859260e-01
1.14893399e-01 -1.47487484e-02 -1.56737819e-01 -2.07751527e-01
7.81046003e-02 8.83328855e-01 -4.74246517e-02 4.80547160e-01
5.67940295e-01 8.95777047e-01 -1.02328753e+00 -4.47777241e-01
1.26673535e-01 4.33061630e-01 -6.59267306e-01 1.06183633e-01
-5.48863471e-01 2.91650981e-01 3.70076559e-02 5.88097811e-01
2.38532737e-01 5.77200115e-01 5.22545397e-01 -1.40482232e-01
-4.27427739e-02 8.82158041e-01 -9.57953453e-01 1.20675540e+00
-4.73872125e-01 7.57674277e-01 5.74063241e-01 -9.54115689e-01
7.39932954e-01 7.05162764e-01 -2.89237082e-01 -7.85648763e-01
2.03330025e-01 6.63231134e-01 5.63121021e-01 -5.58023453e-01
4.31400776e-01 -6.16981149e-01 -3.88982743e-01 3.94441396e-01
2.83362389e-01 -5.14842391e-01 5.46744883e-01 1.24554724e-01
1.06394696e+00 -1.67419583e-01 7.23591447e-01 -2.99164444e-01
1.00277162e+00 5.68372726e-01 5.78063190e-01 5.44615924e-01
-4.66745079e-01 9.28843245e-02 1.13820708e+00 -2.66223609e-01
-7.39817202e-01 -5.24749756e-01 -1.39831402e-03 1.30453479e+00
-5.19423127e-01 -4.65765983e-01 -7.26860821e-01 -8.77294958e-01
-1.62099376e-01 9.95162606e-01 -4.34273720e-01 1.88865840e-01
-1.02040577e+00 -3.23103964e-01 9.79002357e-01 3.42836410e-01
2.07125515e-01 -1.63491642e+00 -2.66776174e-01 3.29784185e-01
-4.08241093e-01 -1.41603088e+00 -2.80106395e-01 6.18886948e-01
-2.97014564e-01 -1.37339950e+00 -2.12358370e-01 -1.15248740e+00
2.07899138e-01 -4.50301200e-01 1.12658441e+00 5.40207922e-02
5.67819597e-03 4.70669061e-01 -4.95135039e-01 -6.38973355e-01
-1.23862839e+00 1.22139350e-01 -6.96906671e-02 -7.54392087e-01
1.05861080e+00 2.07387824e-02 2.30745465e-01 1.98570535e-01
-7.53495991e-01 -5.60804546e-01 2.80917346e-01 1.19739175e+00
-6.85127303e-02 -3.09089363e-01 5.85111856e-01 -1.25811815e+00
1.22346377e+00 9.17895325e-03 -6.97002769e-01 -1.62870631e-01
-5.45785055e-02 -3.65691446e-02 5.27998567e-01 -2.10312068e-01
-1.03192520e+00 2.95718238e-02 -6.57209337e-01 3.54845077e-01
-4.86765862e-01 7.16114879e-01 -2.17999339e-01 1.06057100e-01
8.17147076e-01 -1.70032635e-01 1.08479246e-01 -3.51597965e-02
3.45711887e-01 8.82452965e-01 4.37468529e-01 -5.66777170e-01
3.17263961e-01 -3.70145515e-02 -4.03496504e-01 -1.35845935e+00
-7.20638037e-01 -6.27964616e-01 -8.43202889e-01 -3.76367420e-02
9.25814688e-01 -9.44865584e-01 -7.40076005e-01 3.36458981e-01
-1.65536952e+00 -5.05956531e-01 -2.28403240e-01 4.83023107e-01
-3.66925359e-01 4.33534920e-01 -8.76873612e-01 -8.26220870e-01
-1.38660267e-01 -7.82972991e-01 9.16411638e-01 -5.01754403e-01
-8.84983897e-01 -1.18869460e+00 1.94432974e-01 2.46442273e-01
1.74527109e-01 -1.90529838e-01 9.13763821e-01 -1.27025187e+00
1.40753493e-01 -1.85133636e-01 -1.52662979e-03 7.63821483e-01
2.26251826e-01 -2.62366951e-01 -9.06189144e-01 4.26136926e-02
6.54075593e-02 -8.21471274e-01 5.40188313e-01 -1.70971245e-01
-1.35263294e-01 -3.66699338e-01 3.47979739e-02 -1.69043347e-01
8.91853571e-01 6.42029643e-02 4.29732502e-01 4.08780664e-01
2.88146466e-01 9.47422266e-01 8.64912391e-01 1.96033344e-01
3.46351713e-01 2.62079537e-01 -2.98591666e-02 -4.39081108e-03
2.44550809e-01 1.93166643e-01 7.37188816e-01 7.92513549e-01
4.91573721e-01 -5.73805988e-01 -1.17802715e+00 8.39044690e-01
-1.67396390e+00 -9.25989449e-01 -2.99717039e-01 1.69169843e+00
1.40566099e+00 4.96477902e-01 9.48815718e-02 3.44943076e-01
5.47777534e-01 -2.49552757e-01 1.75464839e-01 -1.08908963e+00
-2.91277677e-01 2.36713186e-01 1.05284266e-01 1.05332124e+00
-1.24566412e+00 1.35068917e+00 6.38470316e+00 4.34528232e-01
-9.00053680e-01 1.78756848e-01 1.82198733e-01 2.55284309e-01
-1.82536110e-01 -3.64068955e-01 -8.26780200e-01 -2.59145528e-01
9.30053353e-01 2.54807360e-02 2.23708138e-01 5.13758540e-01
2.71183878e-01 -5.39089203e-01 -1.42125356e+00 6.05351269e-01
1.76509842e-01 -8.88382137e-01 -3.75669807e-01 -3.86062890e-01
4.67254728e-01 3.07210714e-01 -8.27191651e-01 7.14139223e-01
9.78107452e-02 -8.65550339e-01 3.41937661e-01 -6.52171019e-03
5.23883343e-01 -1.01191628e+00 1.43997574e+00 6.89078867e-01
-7.04486549e-01 2.92158633e-01 -9.21813697e-02 -4.84397173e-01
1.89192206e-01 3.17137390e-01 -1.28975761e+00 4.70664911e-02
2.34155461e-01 4.44270998e-01 -1.65161952e-01 3.23154807e-01
-7.10005879e-01 7.49937534e-01 -6.33467317e-01 -5.65971732e-01
2.66278952e-01 -8.65532979e-02 6.02576077e-01 1.85225248e+00
-1.94227397e-01 9.70670879e-02 4.74548340e-01 4.31210667e-01
-2.46089280e-01 4.79718924e-01 -6.95017517e-01 -1.12457298e-01
7.19825327e-02 1.05152035e+00 -6.64244056e-01 -4.91623759e-01
-7.43077219e-01 1.11514211e+00 3.02930862e-01 6.61286041e-02
-3.59493375e-01 -6.54693663e-01 6.07120514e-01 2.89072804e-02
2.86729038e-01 -2.88836151e-01 6.82132468e-02 -9.94751394e-01
1.58670202e-01 -1.27416539e+00 4.09296095e-01 -6.01750016e-01
-1.50444674e+00 5.36030114e-01 -2.06698216e-02 -6.28269792e-01
-8.54624271e-01 -9.95990157e-01 -5.34144461e-01 1.00424683e+00
-1.65568066e+00 -8.88251364e-01 1.45949379e-01 5.51272035e-01
3.59140098e-01 8.74894205e-03 1.21985900e+00 1.94102541e-01
8.74445438e-02 4.05309826e-01 -3.78293544e-01 7.55576670e-01
1.24586153e+00 -1.43357909e+00 1.75388694e-01 5.89925289e-01
-6.57081679e-02 6.05628729e-01 8.68582726e-01 -8.18621755e-01
-1.19735575e+00 -6.59188271e-01 1.72673225e+00 -3.78813952e-01
1.02409601e+00 -6.02380514e-01 -1.06165075e+00 7.33974934e-01
8.46076906e-01 -3.44307244e-01 7.55210280e-01 3.67883176e-01
-2.28699416e-01 1.81786150e-01 -1.38179696e+00 4.90115762e-01
6.82614982e-01 -8.42853487e-01 -1.20592451e+00 5.48228204e-01
6.79795325e-01 -5.98302245e-01 -6.02451384e-01 3.00919861e-01
3.04178417e-01 -9.51005757e-01 1.77269042e-01 -4.75403011e-01
-1.11248214e-02 -2.15342075e-01 -2.79707581e-01 -1.14464331e+00
5.33129871e-01 -5.55111349e-01 5.67221701e-01 1.21877599e+00
7.60076404e-01 -7.18958557e-01 6.62598014e-01 5.85981965e-01
-1.10107129e-02 -7.75038525e-02 -1.00560153e+00 -4.76796418e-01
3.50566179e-01 -8.58164489e-01 -9.23971385e-02 1.01931810e+00
7.31297314e-01 9.66338396e-01 1.18856452e-01 -4.71019261e-02
3.56171168e-02 -2.97186434e-01 7.34108031e-01 -1.04504108e+00
-5.41317947e-02 -1.39401466e-01 -4.47393030e-01 -8.72872293e-01
5.96178174e-01 -7.46242464e-01 3.74311835e-01 -1.30590701e+00
-3.20170879e-01 2.57080704e-01 2.88982242e-01 8.18460226e-01
1.94222108e-01 2.46533424e-01 -1.36894733e-01 -3.49450618e-01
-5.89910686e-01 5.36036611e-01 7.78748810e-01 -2.97857895e-02
-2.95259863e-01 -1.91471100e-01 -1.98434815e-01 9.36511695e-01
8.89510214e-01 -6.07189357e-01 3.26280557e-02 -2.83042252e-01
4.55094337e-01 -5.44391274e-02 1.14862837e-01 -5.78895152e-01
1.32661298e-01 1.60763949e-01 2.64174819e-01 -6.22384369e-01
2.64605701e-01 -7.37223208e-01 -1.02732217e+00 5.34652472e-01
-2.89090574e-01 2.74182409e-01 6.13813758e-01 1.64824441e-01
-5.55776834e-01 -5.23149729e-01 5.79474807e-01 -3.66924912e-01
-6.56980813e-01 -7.90590107e-01 -1.14355445e+00 5.65928221e-01
7.61609733e-01 4.75842990e-02 -1.12865083e-01 -6.90657616e-01
-8.54557455e-01 4.39436197e-01 4.42068428e-01 1.70147687e-01
4.73307043e-01 -6.58935487e-01 -7.28994966e-01 4.41110462e-01
3.00669849e-01 -3.47296059e-01 -5.47791064e-01 8.24082851e-01
-6.62582457e-01 5.11767745e-01 6.33592531e-02 -6.00248039e-01
-1.62654972e+00 -1.29793240e-02 5.07060111e-01 -5.85515425e-02
-6.96823150e-02 8.81371260e-01 -5.43904185e-01 -1.22881067e+00
4.43046689e-01 -6.29566014e-01 -2.57946014e-01 2.26777792e-01
3.57031405e-01 -1.12474933e-01 2.92781383e-01 -8.85982215e-01
-6.04457796e-01 8.40907693e-02 -1.34647921e-01 -5.07929802e-01
1.07318890e+00 1.06175214e-01 -5.02167881e-01 6.84406698e-01
1.06148231e+00 5.45321763e-01 -3.44185650e-01 -1.33503541e-01
4.29087728e-01 5.88678047e-02 -3.13253492e-01 -9.32559669e-01
-1.40659705e-01 6.66743636e-01 1.47080109e-01 2.63251483e-01
8.68749082e-01 -2.56478433e-02 5.77318013e-01 1.29391122e+00
1.95428163e-01 -1.26263857e+00 -2.11067885e-01 1.43037224e+00
7.44134307e-01 -1.73686445e+00 -2.11628139e-01 -5.74532568e-01
-8.57956350e-01 1.24896455e+00 6.09260798e-01 8.82756338e-02
5.57504177e-01 5.63004494e-01 7.88152754e-01 -2.97694892e-01
-8.85782361e-01 -3.23625475e-01 -2.86148954e-03 7.21351564e-01
1.14525926e+00 -2.60145396e-01 -7.52837181e-01 3.78371805e-01
-2.83398777e-01 -1.10560767e-01 7.20000327e-01 1.21800578e+00
-3.63112211e-01 -1.64407396e+00 -3.24298143e-01 1.27224505e-01
-7.33892918e-01 -4.49897498e-01 -1.04272842e+00 1.17265356e+00
1.46186262e-01 1.21072388e+00 -8.77069458e-02 2.81033907e-02
7.43081093e-01 5.10055900e-01 5.58305085e-01 -1.01961839e+00
-1.08573389e+00 3.86749446e-01 9.22966599e-01 -2.58074522e-01
-6.63294315e-01 -8.14608872e-01 -1.60497904e+00 -1.33341640e-01
-4.66770560e-01 5.32327294e-01 5.43667495e-01 1.17163527e+00
-2.68000156e-01 2.97332406e-01 -1.25742078e-01 -4.18331116e-01
-5.55174410e-01 -1.15935183e+00 -3.05587560e-01 1.69445828e-01
4.32545543e-01 -7.01905265e-02 -4.63109583e-01 9.89474580e-02]
|
[10.5242919921875, 9.297021865844727]
|
04e58659-7568-45ba-a22c-6d5cfa2df59f
|
artificial-life-using-the-book-and-bookmarker
|
2210.12854
| null |
https://arxiv.org/abs/2210.12854v2
|
https://arxiv.org/pdf/2210.12854v2.pdf
|
Artificial Life using a Book and Bookmarker
|
Reproduction, development, and individual interactions are essential topics in artificial life. The cellular automata, which can handle these in a composite way, is highly restricted in its form and behavior because it represents life as a pattern of cells. In contrast, the virtual creatures proposed by Karl Sims have a very high degree of freedom in terms of morphology and behavior. However, they have limited expressive capacity in terms of those viewpoints. This study carefully extracts the characteristics of the cellular automata and Sims models to propose a new artificial life model that can simulate reproduction, development, and individual interactions while exhibiting high expressive power for morphology and behavior. The simulation was performed by sequentially reading a book with genetic information and repeatedly executing four actions: expansion, connection, disconnection, and transition. The virtual creatures in the proposed model exhibit unique survival strategies and lifestyles and acquire interesting properties in reproduction, development, and individual interactions while having freedom in morphology and behavior.
|
['Keishu Utimula']
|
2022-10-07
| null | null | null | null |
['artificial-life']
|
['miscellaneous']
|
[-2.59873480e-01 -4.88644466e-02 2.68830597e-01 4.07587916e-01
1.08682013e+00 -7.51079857e-01 1.04821873e+00 1.05753615e-01
-2.19124362e-01 9.99553144e-01 -3.16300392e-01 -7.57953525e-02
-1.38826936e-01 -1.30362117e+00 -2.61369765e-01 -9.45265770e-01
-1.97111726e-01 5.99991024e-01 3.33886951e-01 -5.77315629e-01
1.29739523e-01 6.63055718e-01 -1.94495010e+00 -6.40277207e-01
1.08987045e+00 2.74561584e-01 4.02227603e-02 7.90810347e-01
-1.90540239e-01 5.18796921e-01 -5.18203735e-01 -3.04478228e-01
-1.69269040e-01 -9.19169068e-01 -3.23474646e-01 1.95580468e-01
-7.55778968e-01 1.52231693e-01 -2.45040253e-01 6.35422170e-01
4.77980554e-01 1.52375726e-02 1.05572402e+00 -1.30241430e+00
-7.30991781e-01 3.87937099e-01 -3.46009582e-01 -2.89853513e-01
6.47516727e-01 6.09071217e-02 2.84838825e-01 -3.16246063e-01
5.34287214e-01 1.01361263e+00 3.62497836e-01 8.23179007e-01
-9.39939797e-01 -1.49888903e-01 -3.43050659e-01 -3.47787142e-01
-1.57160771e+00 -2.32493207e-01 3.99993569e-01 -4.36055481e-01
6.84154093e-01 5.78466773e-01 1.51402783e+00 5.54765344e-01
6.89169049e-01 4.76478487e-01 8.32556009e-01 -5.20842731e-01
5.62469721e-01 3.97144482e-02 5.48981316e-02 9.00001287e-01
6.63650453e-01 5.77962361e-02 -9.61888582e-02 -1.00812517e-01
1.24023020e+00 4.86107208e-02 -4.79951017e-02 -1.53262228e-01
-1.25103819e+00 1.53160021e-01 -2.80784130e-01 7.49892056e-01
-1.88836917e-01 3.27238292e-01 -2.15156913e-01 2.07841188e-01
-1.61371246e-01 4.41349864e-01 -9.70176049e-03 -2.52780885e-01
-3.97365600e-01 2.08397999e-01 1.29605067e+00 8.49638760e-01
4.35397625e-01 3.10973644e-01 9.61644351e-02 5.59162676e-01
2.81590462e-01 7.52352417e-01 5.75346947e-01 -9.52469766e-01
-6.78907871e-01 1.01908612e+00 1.44470744e-02 -1.11787331e+00
-5.64752758e-01 -5.33634067e-01 -1.30667460e+00 3.74015838e-01
2.75248945e-01 -5.84759451e-02 -5.90304017e-01 1.63487947e+00
5.17919362e-01 1.47600368e-01 3.58599037e-01 3.30625772e-01
1.00625479e+00 9.03403342e-01 3.02974656e-02 -6.24240816e-01
1.18540347e+00 -8.59111607e-01 -8.72273505e-01 4.57720667e-01
4.59150583e-01 -2.88172185e-01 7.92817533e-01 2.18417794e-01
-1.51002562e+00 -3.34385514e-01 -1.14667988e+00 3.67311418e-01
-5.59492946e-01 -1.85656533e-01 8.24801326e-01 8.75614762e-01
-1.20265651e+00 4.59460527e-01 -8.62535417e-01 -7.05436945e-01
-1.80944711e-01 4.89653170e-01 -2.02649608e-01 7.83137977e-01
-1.06955242e+00 6.55625224e-01 -4.25643921e-02 -1.35049284e-01
-6.76285446e-01 -3.34132798e-02 -4.31927651e-01 2.17138946e-01
7.00312154e-03 -1.14850557e+00 5.63863158e-01 -6.25770986e-01
-1.80467403e+00 9.52449739e-01 5.59047610e-02 -1.59359097e-01
7.16052115e-01 6.23265564e-01 -4.14696068e-01 -2.02245817e-01
-3.20520252e-01 3.80195051e-01 1.13513023e-01 -1.37998044e+00
-3.71621192e-01 -3.17823738e-01 2.07787409e-01 2.63320088e-01
-3.28615725e-01 -4.16361809e-01 -5.84519088e-01 -6.23514116e-01
1.48303658e-01 -9.33008432e-01 -2.86294401e-01 -1.46259610e-02
-8.35998952e-02 -1.63378622e-02 4.76732045e-01 -1.47631057e-02
1.46218646e+00 -1.99468303e+00 4.79267478e-01 1.56741157e-01
2.54164457e-01 5.61803728e-02 1.71293825e-01 8.03181469e-01
8.37879300e-01 4.38288480e-01 -4.12525207e-01 3.51071986e-03
1.06728952e-02 5.02738297e-01 4.56109554e-01 3.51836115e-01
-4.08981770e-01 8.40182066e-01 -8.78347933e-01 -9.62497890e-01
-1.53457746e-02 4.91468579e-01 -5.10670483e-01 3.94344665e-02
-8.96012038e-02 6.25246167e-01 -5.73790193e-01 9.45521057e-01
4.04715359e-01 -1.58009484e-01 2.57518321e-01 6.01758301e-01
-3.88731509e-01 -7.65706778e-01 -1.02772975e+00 9.58375573e-01
-4.08918113e-01 2.11793974e-01 1.96298301e-01 -5.78864515e-01
1.11924648e+00 3.38818520e-01 4.91145253e-01 -3.97061527e-01
4.48821455e-01 2.55399764e-01 2.05725521e-01 -5.11324644e-01
3.74587774e-01 8.70434195e-02 -8.65454450e-02 8.26795578e-01
-2.78965652e-01 -4.49770153e-01 5.93996167e-01 4.41047885e-02
1.00614786e+00 1.14274450e-01 6.63193405e-01 -5.46188295e-01
9.30842936e-01 -3.48026663e-01 6.32316232e-01 5.79730570e-01
-1.01411320e-01 1.75678208e-01 4.11257684e-01 -4.16376144e-01
-1.04318666e+00 -1.25481248e+00 -5.71780801e-02 5.83765805e-01
8.76113713e-01 -5.63099748e-03 -9.62468207e-01 1.50496081e-01
-1.04456432e-01 4.81320679e-01 -7.73993969e-01 -2.38428295e-01
-4.47939575e-01 -8.48841786e-01 6.52427673e-01 -3.53834666e-02
8.98879051e-01 -1.34787178e+00 -8.00844550e-01 2.18143225e-01
9.51024741e-02 -5.93628168e-01 -6.45661056e-02 -1.73359841e-01
-8.65877330e-01 -8.20062101e-01 -6.40789628e-01 -1.02404451e+00
9.62905228e-01 -1.51159644e-01 8.89942229e-01 8.25986147e-01
-8.95897523e-02 3.25098127e-01 -4.17669266e-01 -2.49868736e-01
-9.39217031e-01 6.18090928e-02 3.29843551e-01 -2.31619790e-01
-2.74922937e-01 -1.06296515e+00 -4.70476359e-01 5.79510152e-01
-1.07005084e+00 2.30115950e-01 3.16118032e-01 6.91637039e-01
4.81639415e-01 3.67204040e-01 6.20519102e-01 -5.74130356e-01
6.58978164e-01 -4.47333634e-01 -3.67128223e-01 5.35151482e-01
-5.79336047e-01 -2.19474673e-01 9.15105402e-01 -4.04339224e-01
-1.06600177e+00 -9.98995155e-02 4.79039401e-02 4.66078579e-01
-1.62400976e-01 2.02997714e-01 -3.24941099e-01 -2.31466234e-01
1.89478546e-01 7.91233420e-01 2.71128327e-01 -3.06759566e-01
-1.44940510e-01 5.87793529e-01 4.66226310e-01 -5.65606952e-01
7.42551863e-01 4.71311241e-01 5.18720210e-01 -1.14418650e+00
4.79793578e-01 4.19434160e-01 -7.00040877e-01 -6.07630134e-01
5.71621478e-01 -2.26127788e-01 -1.26194429e+00 1.17127430e+00
-7.52228498e-01 -1.55810773e-01 -5.11714637e-01 1.75160572e-01
-7.54055321e-01 3.93486559e-01 -6.66874588e-01 -1.12478328e+00
-3.02939415e-01 -6.30170584e-01 4.24608171e-01 7.63265848e-01
-1.72898382e-01 -1.23041821e+00 1.64424241e-01 -1.79601729e-01
6.02761865e-01 6.83445871e-01 1.12968588e+00 -2.25157604e-01
-4.56576705e-01 -2.81898111e-01 5.14544606e-01 -5.28917968e-01
1.26463220e-01 7.63990462e-01 -2.66782790e-01 -2.06353754e-01
-2.03072116e-01 3.33554804e-01 3.73118967e-01 2.55042940e-01
5.62257528e-01 -2.65832186e-01 -6.29337072e-01 6.28744721e-01
1.36498988e+00 1.04360282e+00 7.40884244e-01 3.13086003e-01
1.88295897e-02 7.82138050e-01 2.69756347e-01 7.24105835e-01
3.47745508e-01 3.52201521e-01 5.40774107e-01 -9.09484252e-02
3.84531543e-02 -1.87068820e-01 1.96879029e-01 1.27714360e+00
-7.69118786e-01 -7.81179547e-01 -8.81238759e-01 4.62096512e-01
-1.73601043e+00 -9.89269197e-01 -4.32812236e-02 1.89407575e+00
5.70666730e-01 -8.36134329e-02 2.75460333e-01 4.02542442e-01
9.51232135e-01 -2.27701485e-01 -4.34745580e-01 -5.99920154e-01
-6.57521904e-01 -2.62565404e-01 1.53608158e-01 2.62270242e-01
-3.38875622e-01 7.83815086e-01 7.68418503e+00 7.34640419e-01
-7.50333309e-01 -2.48116225e-01 6.21367037e-01 6.89021945e-02
-6.20337009e-01 6.20355224e-03 -4.59542274e-01 6.77575350e-01
4.30725813e-01 -5.81206203e-01 6.13470614e-01 1.70629159e-01
1.68012872e-01 -3.93931091e-01 -6.37089610e-01 7.84961343e-01
-1.89215571e-01 -1.27974594e+00 2.72897422e-01 1.58306018e-01
8.18997860e-01 -8.56385171e-01 -1.03735261e-01 -4.47444618e-02
3.11474383e-01 -1.12427223e+00 6.51906490e-01 8.49348068e-01
5.97204864e-01 -6.00246787e-01 6.09018266e-01 7.89589405e-01
-1.24113238e+00 -3.83679532e-02 -1.54551864e-01 -5.37910759e-01
3.33877295e-01 1.62880942e-01 -1.58346314e-02 4.57534939e-01
3.36669892e-01 3.10786039e-01 -4.25490201e-01 1.03626800e+00
1.65879905e-01 4.38534796e-01 -4.67588514e-01 -9.98599708e-01
-2.27655351e-01 -1.05697274e+00 6.60370529e-01 8.66009533e-01
7.38136590e-01 5.17353714e-01 -2.94009775e-01 1.04765630e+00
3.64692122e-01 2.88121045e-01 -7.27685213e-01 -2.63822258e-01
6.52927279e-01 1.01962149e+00 -1.18266678e+00 -2.99444973e-01
-6.24485426e-02 6.38109803e-01 -4.37811881e-01 2.87636340e-01
-8.65659237e-01 -4.74123389e-01 3.59600723e-01 6.01040781e-01
-3.62199008e-01 -4.67942566e-01 -5.23722768e-01 -9.54048693e-01
-5.04640937e-01 -4.42198515e-01 -1.86283067e-01 -5.05269766e-01
-6.89413607e-01 6.22322857e-01 -9.48578492e-02 -1.07859707e+00
-4.91847508e-02 -1.40994951e-01 -9.82262969e-01 2.91250199e-01
-4.23275530e-01 -1.28586662e+00 -5.01100302e-01 4.51197326e-01
8.23359638e-02 -5.30902565e-01 7.51285911e-01 -5.51420413e-02
-7.63040781e-01 4.61051375e-01 4.81643677e-01 -3.24619919e-01
-1.88861921e-01 -9.44662452e-01 1.33029312e-01 4.85368639e-01
-4.34598655e-01 5.91788709e-01 9.94463325e-01 -6.72771394e-01
-1.65476012e+00 -3.56257826e-01 8.88988316e-01 1.17663005e-02
2.49781132e-01 -3.34748149e-01 -3.49608600e-01 1.67570457e-01
2.46522397e-01 -5.46981156e-01 5.37700295e-01 -3.75141591e-01
6.43484414e-01 5.32504953e-02 -1.45394993e+00 1.26928997e+00
1.45413661e+00 1.31456882e-01 -2.97871530e-01 -9.11875591e-02
5.52305758e-01 1.61397979e-01 -8.96769822e-01 3.33299696e-01
1.15468574e+00 -1.09082210e+00 7.14760065e-01 4.92313281e-02
1.80329293e-01 -4.58890647e-01 1.79031894e-01 -9.43872094e-01
-5.03565371e-01 -9.39776063e-01 3.06361746e-02 1.49580741e+00
2.63863564e-01 -1.20189416e+00 6.14235461e-01 2.94196069e-01
2.36632049e-01 -9.58916903e-01 -6.68046653e-01 -1.03405261e+00
1.82964280e-01 5.21093130e-01 1.14343166e+00 8.60748351e-01
4.88879502e-01 2.57159062e-02 -1.94311872e-01 -4.50627685e-01
5.74179828e-01 2.41946548e-01 8.72958064e-01 -1.62490749e+00
-2.36602664e-01 -4.84229892e-01 -5.07040441e-01 -6.95943654e-01
-9.60282087e-02 -5.39093316e-01 -4.14914489e-01 -1.58091676e+00
2.28189930e-01 -6.24670804e-01 6.33388236e-02 4.23573740e-02
4.90120411e-01 8.04273337e-02 -6.64243195e-03 5.11360705e-01
-2.90713191e-01 5.56751490e-01 1.73786414e+00 1.02715716e-01
-5.47854602e-01 -9.10610408e-02 -3.11097592e-01 7.30546892e-01
7.99671590e-01 1.00995354e-01 -3.45963806e-01 -4.46797721e-02
3.61953974e-01 5.88929594e-01 -1.30080253e-01 -1.21340060e+00
4.55406904e-01 -5.24887800e-01 1.29756078e-01 -3.62887740e-01
2.79589325e-01 -7.60431230e-01 1.11330533e+00 1.12285697e+00
8.84383023e-02 1.82071924e-01 -2.57583559e-01 4.00092065e-01
-1.55969813e-01 -1.15462713e-01 8.38872254e-01 -2.99441904e-01
-2.93595672e-01 2.22115725e-01 -1.23124611e+00 -3.09116572e-01
1.69048393e+00 -1.10834813e+00 -4.02053028e-01 -4.08885688e-01
-1.02781558e+00 2.22555220e-01 1.45788538e+00 -1.67766064e-02
3.41759622e-01 -1.17444742e+00 -4.51959401e-01 3.42515439e-01
-2.48719707e-01 -2.58161932e-01 1.07073806e-01 6.34412885e-01
-1.37487245e+00 9.60130915e-02 -9.38406348e-01 -1.63101390e-01
-1.08368599e+00 4.60156560e-01 4.14308399e-01 -9.18845311e-02
-9.65259969e-02 3.71907771e-01 3.61621737e-01 -3.64431977e-01
-1.21760622e-01 1.77118883e-01 -7.35326767e-01 -1.09308384e-01
1.39421299e-02 8.93006563e-01 -6.03192210e-01 -8.26250792e-01
-3.83452356e-01 1.09760296e+00 7.86635876e-01 -1.79779172e-01
1.09036779e+00 -5.46678364e-01 -6.64130509e-01 4.16939437e-01
3.17609966e-01 2.69083142e-01 -6.12788975e-01 4.49582785e-01
-4.22062695e-01 -2.39048526e-01 -6.71327710e-01 -6.80282176e-01
-8.17646682e-01 5.54568410e-01 1.15180239e-02 7.42154837e-01
1.24758160e+00 -1.83835477e-01 7.33634710e-01 1.78124264e-01
8.34547520e-01 -1.07981074e+00 -1.01529151e-01 5.17227232e-01
6.83588743e-01 -4.05233592e-01 -1.94803432e-01 -5.61494470e-01
-2.48392642e-01 9.81952846e-01 7.66964376e-01 -2.30508417e-01
9.44066107e-01 8.62504661e-01 -3.22426975e-01 1.98275838e-02
-1.15321398e+00 -6.10871576e-02 -4.76807684e-01 8.65416169e-01
3.81479561e-01 1.54386103e-01 -1.28713000e+00 6.40529931e-01
-2.29473799e-01 -1.10691215e-03 8.41317296e-01 1.01824498e+00
-9.12537932e-01 -1.01942098e+00 -3.01338315e-01 6.97183460e-02
-3.12639207e-01 3.15775812e-01 -7.25879729e-01 7.96185434e-01
3.73722851e-01 9.19872463e-01 1.90359220e-01 -5.07325292e-01
2.31174771e-02 -3.51446420e-01 5.38876176e-01 3.96553352e-02
-6.45125985e-01 -2.00881943e-01 -6.16095476e-02 7.59870037e-02
-3.31714034e-01 -5.30047417e-01 -1.59622419e+00 -9.66782570e-01
-2.65880793e-01 4.87622440e-01 4.70593393e-01 5.92523336e-01
2.16427177e-01 3.89304966e-01 5.79884231e-01 -4.66137201e-01
8.85347873e-02 -5.25826693e-01 -1.26501632e+00 5.78270778e-02
-1.58880979e-01 -6.33140385e-01 -3.66764307e-01 1.70123488e-01]
|
[5.6089348793029785, 4.150115966796875]
|
d60947d1-4e78-4796-a96f-f2ef80001667
|
formalisation-of-action-with-durations-in
|
2109.08305
| null |
https://arxiv.org/abs/2109.08305v1
|
https://arxiv.org/pdf/2109.08305v1.pdf
|
Formalisation of Action with Durations in Answer Set Programming
|
In this paper, I will discuss the work I am currently doing as a Ph.D. student at the University of Potsdam, under the tutoring of T. Schaub. I'm currently looking into action description in ASP. More precisely, my goal is to explore how to represent actions with durations in ASP, in different contexts. Right now, I'm focused on Multi-Agent Path Finding (MAPF), looking at how to represent speeds for different agents and contexts. Before tackling duration, I wanted to explore and compare different representations of action taking in ASP. For this, I started comparing different simple encodings tackling the MAPF problem. Even in simple code, choices and assumptions have been made in their creations. The objective of my work is to present the consequences of those design decisions in terms of performance and knowledge representation. As far as I know, there is no current research on this topic. Besides that, I'm also exploring different ways to represent duration and to solve related problems. I planed to compare them the same way I described before. I also want this to help me find innovative and effective ways to solve problems with duration.
|
['Etienne Tignon']
|
2021-09-17
| null | null | null | null |
['multi-agent-path-finding']
|
['playing-games']
|
[-2.85743922e-02 3.08464259e-01 -1.18862972e-01 -2.28666827e-01
-1.96777154e-02 -7.76222944e-01 3.78626466e-01 4.44519877e-01
-3.51102591e-01 9.33816850e-01 2.85088301e-01 -5.65670848e-01
-5.67486525e-01 -1.03666329e+00 -1.93314865e-01 -2.12844372e-01
-7.60825351e-02 6.00767672e-01 5.98303795e-01 -4.90038186e-01
7.64056206e-01 5.48074663e-01 -1.66942883e+00 2.51339734e-01
6.99066162e-01 1.96192861e-01 2.33993039e-01 6.98623836e-01
-3.32126886e-01 9.75152552e-01 -8.82788002e-01 -3.52930903e-01
1.61691627e-03 -4.26196098e-01 -1.50410986e+00 1.57110050e-01
2.95114927e-02 -2.43606344e-01 -2.26153970e-01 6.44400179e-01
2.06350237e-01 5.68857610e-01 4.52238441e-01 -1.52258337e+00
4.13751341e-02 7.97911644e-01 -8.91205966e-02 2.88898587e-01
9.79701400e-01 1.96725324e-01 7.51913309e-01 2.28216872e-01
7.68783748e-01 1.18676925e+00 3.86431903e-01 5.35886168e-01
-8.78174186e-01 6.02488313e-03 4.80977803e-01 6.58135355e-01
-1.22856963e+00 -2.79177576e-01 4.04919773e-01 -2.72092581e-01
1.33414745e+00 5.76784730e-01 9.87259984e-01 7.11919487e-01
1.20169774e-01 9.93622184e-01 1.02765584e+00 -6.72761679e-01
2.55670160e-01 4.13611650e-01 1.83949396e-01 4.98452336e-01
3.01776946e-01 -1.05094694e-01 -1.83794603e-01 -1.73918083e-01
8.34677458e-01 -4.07467484e-01 -2.42996261e-01 -4.42035079e-01
-1.14815807e+00 8.71484101e-01 -2.50069618e-01 5.62869847e-01
-1.62730902e-01 1.70858413e-01 2.96381593e-01 4.33639973e-01
-1.61628440e-01 8.07771742e-01 -4.66639519e-01 -8.91923189e-01
-5.14359295e-01 6.92674339e-01 1.43723381e+00 9.30488944e-01
4.11176473e-01 -1.32533401e-01 1.73253924e-01 5.19188881e-01
4.03000116e-01 -3.01487714e-01 3.27141732e-01 -1.50021482e+00
4.38238829e-01 6.64774001e-01 4.32676077e-01 -8.22324514e-01
-6.01074934e-01 1.29564404e-01 7.72969127e-02 6.88919365e-01
8.74180436e-01 -2.89086431e-01 -3.01094711e-01 1.65842974e+00
1.18545242e-01 -1.74846489e-03 3.85854006e-01 6.89559042e-01
4.71628726e-01 8.48998904e-01 1.41241729e-01 -2.47327864e-01
1.32317436e+00 -8.60194027e-01 -6.57920301e-01 -1.92857727e-01
1.20446610e+00 -6.53992057e-01 6.85659945e-01 6.42881513e-01
-1.55493212e+00 -9.09258723e-02 -9.53895509e-01 2.72165507e-01
-5.14239132e-01 -3.30571890e-01 1.10048223e+00 8.67199540e-01
-8.36220384e-01 7.59355068e-01 -9.18393195e-01 -7.57806420e-01
-1.34346634e-01 2.35540107e-01 -9.54136476e-02 -1.13469437e-01
-1.25905526e+00 1.27472222e+00 5.24085522e-01 -2.85373926e-01
-4.52366620e-01 -4.78959382e-01 -8.03376555e-01 4.18469645e-02
6.93967044e-01 -8.00156415e-01 1.51767993e+00 -7.62538433e-01
-1.59612560e+00 5.42645335e-01 -4.59508300e-02 -3.56641769e-01
4.81010377e-01 9.03096423e-02 -2.97878861e-01 -9.75152627e-02
1.17023371e-01 4.67088819e-01 -1.91315293e-01 -9.40428197e-01
-8.89077485e-01 -2.91891396e-01 1.02782345e+00 6.70494497e-01
4.53882180e-02 1.57625809e-01 -8.79482403e-02 -6.09689951e-01
5.17366678e-02 -9.50635731e-01 -2.50093728e-01 -2.90155530e-01
5.16101010e-02 -3.83469015e-01 3.63264531e-01 -2.14833155e-01
1.44879425e+00 -1.90618753e+00 3.49360257e-01 1.97433218e-01
-2.88020313e-01 1.76254451e-01 -6.42636567e-02 1.08294117e+00
-3.16579975e-02 1.69675142e-01 -1.26889035e-01 2.63482481e-01
3.41567010e-01 4.05829370e-01 -1.15715317e-01 2.13530257e-01
-1.66556895e-01 4.85334456e-01 -8.34346294e-01 -5.13353169e-01
2.50459552e-01 2.36432359e-01 -5.51674604e-01 -1.52407825e-01
-4.45899546e-01 8.28359351e-02 -5.14331400e-01 4.81333643e-01
4.72710341e-01 2.54939646e-01 3.79212976e-01 3.76765966e-01
-4.83174741e-01 6.22934401e-01 -1.74179053e+00 1.85700738e+00
-3.62220794e-01 4.98827636e-01 -1.24906741e-01 -1.11927056e+00
6.95019245e-01 3.87856454e-01 5.66543698e-01 -6.07139349e-01
-4.53261239e-03 2.51377374e-01 3.81580770e-01 -6.82680547e-01
6.32826269e-01 -9.50463042e-02 -1.94580123e-01 7.69431114e-01
-6.48520708e-01 -1.87619299e-01 9.82504189e-01 1.42357677e-01
1.39714885e+00 5.26371479e-01 3.60796541e-01 -1.73052609e-01
6.14285827e-01 4.63487774e-01 2.77781308e-01 6.70442581e-01
-9.30711403e-02 1.64254665e-01 8.18782568e-01 -6.30511880e-01
-6.24170601e-01 -6.44571543e-01 2.75575817e-02 1.00573862e+00
2.26452202e-01 -7.84630775e-01 -6.37341917e-01 -3.28428596e-01
-2.59638727e-01 1.12470114e+00 -1.77327722e-01 2.03821927e-01
-8.96636903e-01 -2.89874047e-01 5.95854700e-01 4.05704141e-01
4.62528467e-01 -1.21687257e+00 -1.36676502e+00 6.65782690e-01
-2.63404638e-01 -7.84956276e-01 1.02123559e-01 1.56950012e-01
-1.10170305e+00 -1.10380459e+00 -5.54064631e-01 -4.94173110e-01
6.93959296e-02 1.86886191e-01 1.09210742e+00 5.15366048e-02
-2.37044871e-01 7.26448774e-01 -7.68680573e-01 -6.19727671e-01
-4.15684640e-01 1.11446276e-01 -3.68516892e-01 -9.72539008e-01
3.39623600e-01 -5.35265088e-01 -2.88649082e-01 5.57875752e-01
-1.13873231e+00 -1.04035385e-01 5.30304790e-01 1.46525070e-01
1.51102901e-01 5.49150944e-01 9.39969048e-02 -1.01912880e+00
8.40704501e-01 -3.15169632e-01 -5.51199913e-01 4.05934095e-01
-5.25247633e-01 2.04214007e-01 1.80636272e-01 -2.75991976e-01
-1.03951097e+00 -1.98222756e-01 -1.51660964e-01 5.14566302e-01
-7.58510053e-01 5.37287951e-01 -3.08835417e-01 1.64895311e-01
5.11277258e-01 -1.61044836e-01 -9.46607813e-02 -1.13319503e-02
1.63572922e-01 4.29230452e-01 -9.39236302e-03 -1.01976132e+00
-4.99638766e-02 5.66946715e-02 8.40447620e-02 -8.18682909e-01
-2.72580385e-01 -2.44854897e-01 -1.67845249e-01 -3.54516394e-02
6.99306130e-01 -3.66062850e-01 -7.53800392e-01 4.27160472e-01
-1.15161562e+00 -7.81894028e-01 -3.37225616e-01 4.95935738e-01
-1.05860829e+00 5.47244489e-01 -5.13784468e-01 -7.45931685e-01
5.81627309e-01 -1.28700638e+00 3.54631484e-01 2.94474453e-01
-6.58560216e-01 -1.15330458e+00 3.39503199e-01 4.46518451e-01
2.89359957e-01 1.88290954e-01 9.05908585e-01 -4.96727794e-01
-6.41440451e-01 1.06697135e-01 1.95583805e-01 -2.55493045e-01
1.02808312e-01 3.14443596e-02 -4.13597584e-01 -1.69378638e-01
-1.19647771e-01 2.45021626e-01 1.00407109e-01 3.78409028e-01
1.04773629e+00 -5.56865513e-01 -5.66948712e-01 1.30920500e-01
1.48751867e+00 8.83112848e-01 1.13150978e+00 1.27529621e+00
-8.80144238e-02 8.60534191e-01 9.70777571e-01 5.47753274e-01
5.51334500e-01 1.00428689e+00 3.35575342e-01 5.31270027e-01
8.77682418e-02 2.01934695e-01 2.81226754e-01 -1.32716760e-01
-2.78836399e-01 -6.40958726e-01 -9.79767084e-01 5.92859626e-01
-2.18386579e+00 -1.31448281e+00 -2.40934402e-01 1.97480965e+00
4.73834783e-01 2.04192191e-01 5.16821384e-01 3.18044096e-01
5.30164838e-01 6.38443977e-02 -1.18452162e-02 -1.18262303e+00
3.22837234e-01 9.01451334e-02 3.95435661e-01 7.85535634e-01
-8.13001931e-01 7.20538616e-01 5.74899769e+00 3.02023500e-01
-7.08106518e-01 -3.55948806e-01 1.11689888e-01 8.20544064e-02
-3.46726298e-01 5.82479537e-01 -7.93958127e-01 3.48934412e-01
1.08490694e+00 -3.41652453e-01 5.55892050e-01 6.60617292e-01
2.17656568e-01 -6.36753619e-01 -1.09289932e+00 4.86892521e-01
-1.21116601e-01 -1.15616238e+00 -3.35456431e-01 1.92675576e-01
3.26035649e-01 -6.04024470e-01 -3.89298201e-01 3.60119045e-01
4.44839090e-01 -1.09788358e+00 5.76789975e-01 5.40624738e-01
-1.97275028e-01 -9.67513084e-01 6.80894971e-01 5.41271925e-01
-9.58300114e-01 -3.78011495e-01 -2.01218307e-01 -8.24024856e-01
3.29040527e-01 8.16127732e-02 -9.14917469e-01 7.30002463e-01
4.49287087e-01 3.47446620e-01 5.91357425e-03 1.47314918e+00
-2.27981642e-01 -4.21313830e-02 -4.56491977e-01 -2.81227022e-01
4.23058301e-01 -1.75298706e-01 4.67397600e-01 1.19732833e+00
3.10778975e-01 5.42496324e-01 1.25380039e-01 4.26442057e-01
7.27520823e-01 1.04669621e-02 -7.13042378e-01 -9.72144902e-02
3.61267418e-01 7.52245128e-01 -8.02558243e-01 -2.60103971e-01
-4.72363353e-01 6.59033656e-01 -9.61384401e-02 2.91961908e-01
-1.02615154e+00 -6.60333276e-01 9.00042951e-01 5.04517496e-01
1.08548336e-01 -3.56882125e-01 -1.72507569e-01 -7.15480983e-01
-7.30647594e-02 -9.92164850e-01 6.23146296e-01 -9.61432159e-01
-6.49862230e-01 3.73365223e-01 8.42850089e-01 -8.99000347e-01
-3.31181318e-01 -4.50210452e-01 -6.68620884e-01 8.30501437e-01
-1.20557487e+00 -4.78929758e-01 -3.53056528e-02 2.82640070e-01
5.94947934e-01 1.00446045e-01 9.26403403e-01 1.74123675e-01
-3.83223474e-01 2.36159742e-01 -3.83618683e-01 -3.06869626e-01
5.26116490e-01 -1.19516182e+00 4.68368679e-02 7.10679412e-01
9.27364156e-02 7.66292691e-01 1.00143516e+00 -4.41078842e-01
-1.18736076e+00 -4.00654733e-01 9.48390841e-01 -1.90230623e-01
6.18469179e-01 6.30613983e-01 -7.56335258e-01 9.75895464e-01
2.78303742e-01 -7.57265449e-01 6.24431431e-01 -1.70770451e-01
1.24480218e-01 3.03909451e-01 -1.20517790e+00 5.94091117e-01
9.39931870e-01 4.31345738e-02 -8.85315180e-01 3.47517192e-01
3.35229784e-01 -7.41335332e-01 -9.13583219e-01 1.63045272e-01
3.55319828e-01 -1.27798092e+00 1.00482666e+00 -6.45619750e-01
2.03076780e-01 -6.15618706e-01 -2.95123365e-02 -1.34229171e+00
-2.94166595e-01 -5.59199154e-01 1.75352752e-01 1.07550859e+00
4.08825010e-01 -8.48772824e-01 1.05887854e+00 8.98275018e-01
-4.60551411e-01 -7.47929215e-01 -7.33405232e-01 -9.61746275e-01
6.62280619e-02 -5.73007643e-01 7.07049251e-01 8.09680402e-01
6.19746923e-01 2.66750064e-02 7.59397149e-02 1.79035693e-01
1.42580241e-01 2.04012081e-01 8.38309586e-01 -1.01140058e+00
-4.74907219e-01 -7.17424095e-01 -4.56204563e-01 -1.02140713e+00
-6.37042001e-02 -5.63094139e-01 -3.27972859e-01 -2.24140191e+00
-3.38344008e-01 -6.78508937e-01 2.15378255e-01 6.21495128e-01
4.54138786e-01 -3.59433383e-01 3.31650525e-01 5.11814132e-02
-4.26387042e-01 -1.25519291e-01 1.24205399e+00 4.17826623e-02
-4.21762109e-01 2.09177986e-01 -7.63123453e-01 6.34078026e-01
1.22754657e+00 -4.54745799e-01 -7.84411967e-01 -4.86466974e-01
5.08983135e-01 5.81270099e-01 2.00216576e-01 -1.20981514e+00
3.30982596e-01 -8.79394829e-01 -1.87823698e-01 -3.80654305e-01
5.08961260e-01 -8.55901957e-01 5.82965493e-01 5.09911895e-01
-2.85148472e-01 3.58662814e-01 5.47786891e-01 4.69253995e-02
-2.37504095e-01 -1.18701422e+00 2.95042694e-01 -9.31105435e-01
-1.15132010e+00 -2.00825483e-01 -9.88331795e-01 -5.26998052e-03
1.63837063e+00 -6.41454875e-01 -4.37092394e-01 -4.05376375e-01
-1.03262794e+00 3.23982537e-01 6.32401943e-01 8.58854577e-02
4.28273052e-01 -8.24804664e-01 -2.56341338e-01 -4.75760192e-01
-3.93875577e-02 -1.26753256e-01 1.60977349e-01 6.87771618e-01
-1.03917623e+00 6.41884148e-01 -5.98837376e-01 -1.12634130e-01
-1.41359806e+00 6.11411273e-01 2.00442150e-01 -4.06435817e-01
-7.11173892e-01 5.66030204e-01 -3.77121419e-01 -1.55936092e-01
3.43259484e-01 -3.05698186e-01 -4.15253818e-01 1.66772544e-01
5.22936046e-01 6.74751401e-01 -1.75499991e-02 5.26761524e-02
-4.87824410e-01 5.58588445e-01 1.01595283e-01 -4.15929705e-01
1.54879344e+00 -2.08850577e-01 -1.51959807e-01 2.52819091e-01
5.96200585e-01 -1.13258190e-01 -8.67997050e-01 4.02367830e-01
1.83522791e-01 -6.16260469e-01 -4.87242907e-01 -8.10260773e-01
-5.16095698e-01 6.71639740e-01 2.86024094e-01 6.87219262e-01
1.00501013e+00 -1.30606547e-01 4.66239363e-01 7.79747665e-01
6.68307245e-01 -1.18366957e+00 -3.11571769e-02 7.85729229e-01
7.14877546e-01 -5.86629033e-01 5.34987986e-01 -4.16957885e-01
-8.02422762e-01 1.48425591e+00 6.33273900e-01 2.54110485e-01
3.75077724e-02 3.48727703e-01 -1.47145852e-01 -2.92821437e-01
-6.09768867e-01 -2.71761239e-01 -5.95790029e-01 9.17206109e-01
4.72455114e-01 -1.08559653e-01 -7.73346663e-01 -2.13353023e-01
-2.78367847e-01 2.34886110e-01 1.40341616e+00 1.56586099e+00
-7.74528027e-01 -1.82674325e+00 -6.39712512e-01 3.28659974e-02
-3.61326993e-01 3.59958708e-01 -3.95080805e-01 1.41184080e+00
-4.11220491e-02 1.01228833e+00 -1.04500368e-01 2.84648389e-02
5.21947980e-01 1.28992379e-01 9.70666289e-01 -7.78628170e-01
-4.52688634e-01 -4.77983415e-01 6.96339726e-01 -4.58849430e-01
-5.89997709e-01 -8.04295480e-01 -1.46978307e+00 -6.59917831e-01
6.53467402e-02 5.83796084e-01 5.77050149e-01 6.97699249e-01
-9.61806551e-02 5.89267731e-01 -1.38768163e-02 -4.24606502e-01
-3.65030318e-01 -5.69863737e-01 -2.98471212e-01 -9.48819797e-03
-3.33227456e-01 -7.85234928e-01 4.10002004e-03 -4.30247366e-01]
|
[3.576200008392334, 1.3976904153823853]
|
0310e954-166c-4b2a-9ab3-dddfd652c8b9
|
diffusion-models-a-comprehensive-survey-of
|
2209.00796
| null |
https://arxiv.org/abs/2209.00796v9
|
https://arxiv.org/pdf/2209.00796v9.pdf
|
Diffusion Models: A Comprehensive Survey of Methods and Applications
|
Diffusion models have emerged as a powerful new family of deep generative models with record-breaking performance in many applications, including image synthesis, video generation, and molecule design. In this survey, we provide an overview of the rapidly expanding body of work on diffusion models, categorizing the research into three key areas: efficient sampling, improved likelihood estimation, and handling data with special structures. We also discuss the potential for combining diffusion models with other generative models for enhanced results. We further review the wide-ranging applications of diffusion models in fields spanning from computer vision, natural language processing, temporal data modeling, to interdisciplinary applications in other scientific disciplines. This survey aims to provide a contextualized, in-depth look at the state of diffusion models, identifying the key areas of focus and pointing to potential areas for further exploration. Github: https://github.com/YangLing0818/Diffusion-Models-Papers-Survey-Taxonomy.
|
['Ming-Hsuan Yang', 'Yingxia Shao', 'Yue Zhao', 'Runsheng Xu', 'Shenda Hong', 'Yang song', 'Bin Cui', 'Wentao Zhang', 'Zhilong Zhang', 'Ling Yang']
|
2022-09-02
| null | null | null | null |
['video-generation']
|
['computer-vision']
|
[ 1.34983808e-02 -2.26787940e-01 -5.70573151e-01 -4.72005159e-02
-4.59154129e-01 -6.17172301e-01 1.02286863e+00 -3.07768553e-01
-1.55367032e-01 6.37796223e-01 4.12289709e-01 -3.63609701e-01
-2.23627195e-01 -8.24209511e-01 -3.81534189e-01 -1.03337228e+00
-2.73703068e-01 5.00850320e-01 -3.38922963e-02 1.96686760e-02
3.30764234e-01 4.59466338e-01 -1.20147133e+00 2.17637792e-01
6.69511735e-01 4.63542134e-01 3.09137106e-01 9.45362926e-01
-1.00003406e-01 7.64544725e-01 -7.68644214e-01 -3.43935311e-01
-7.67928883e-02 -8.26362550e-01 -4.41733629e-01 1.49691388e-01
-7.42956996e-02 -2.45504275e-01 -8.84830117e-01 9.20367599e-01
7.04059482e-01 2.60329130e-03 8.63945544e-01 -1.18988669e+00
-1.38313413e+00 3.02015781e-01 -7.04641342e-01 7.46762753e-01
1.96713045e-01 1.72845408e-01 7.99949706e-01 -8.89693260e-01
9.95816410e-01 1.23901713e+00 4.72294062e-01 9.01248038e-01
-9.40838516e-01 -7.73257554e-01 3.97467107e-01 3.94946039e-01
-1.29167295e+00 -3.19938689e-01 7.04291224e-01 -6.56472504e-01
1.11880934e+00 2.04362303e-01 8.45230997e-01 1.57201326e+00
7.47671545e-01 1.14868939e+00 1.08608711e+00 -1.71811134e-01
2.06593364e-01 -2.12493658e-01 -8.69044363e-02 5.75631499e-01
2.66012818e-01 3.39085788e-01 -9.22090828e-01 -2.42396101e-01
1.15250015e+00 1.37938112e-01 -4.22286130e-02 -1.28407314e-01
-1.25837409e+00 1.21247411e+00 8.72991309e-02 4.27342266e-01
-5.36893189e-01 3.53877872e-01 8.27897191e-02 3.66747767e-01
8.04034770e-01 6.29420206e-02 -1.30853251e-01 -2.97217429e-01
-1.08017814e+00 5.71302116e-01 6.09564602e-01 1.18801641e+00
3.64190698e-01 4.46926653e-01 -2.35039681e-01 8.10740650e-01
5.62875211e-01 5.54412067e-01 4.62489307e-01 -9.35321569e-01
-5.61844707e-02 -1.32672518e-01 -1.89871550e-01 -6.14681959e-01
-2.91468263e-01 -3.55347067e-01 -1.18948126e+00 -5.76592498e-02
1.01115689e-01 -3.69426817e-01 -8.43286395e-01 1.64415526e+00
2.71120727e-01 1.95867404e-01 -1.70241907e-01 5.77224612e-01
1.13096535e+00 8.89636874e-01 9.63818282e-02 -4.41748351e-01
1.17148209e+00 -1.01570702e+00 -8.69866431e-01 -1.41687654e-02
2.06787512e-01 -1.09872687e+00 2.94127852e-01 5.42298079e-01
-1.18868351e+00 -3.05341691e-01 -6.70855284e-01 -2.00333148e-01
-3.09951574e-01 -2.97936164e-02 1.18434644e+00 5.50122678e-01
-1.29973996e+00 4.62489396e-01 -1.17912340e+00 -6.36049509e-01
6.56099975e-01 1.86286673e-01 2.29070500e-01 -2.44659543e-01
-1.23514020e+00 5.15507281e-01 -3.59454602e-01 -3.05034041e-01
-1.39131761e+00 -7.34472156e-01 -4.75360543e-01 -4.08247650e-01
4.05015163e-02 -1.18568575e+00 1.30681396e+00 -4.55983758e-01
-1.52550161e+00 7.25083768e-01 -4.96759206e-01 -4.59170908e-01
5.15965223e-01 -7.04167560e-02 -3.91944617e-01 -1.44729868e-01
6.47469051e-03 7.78999627e-01 6.34769559e-01 -8.77316117e-01
-5.44329286e-01 -3.24291438e-01 -2.64818192e-01 1.40181899e-01
-2.88928688e-01 1.93406835e-01 -6.35502934e-01 -1.13290071e+00
-3.29483360e-01 -1.15219808e+00 -4.75510120e-01 -3.93611044e-02
-4.85623538e-01 -4.98881280e-01 8.51401210e-01 -3.13702077e-01
1.57716095e+00 -1.78587914e+00 2.90468931e-01 -2.01939657e-01
6.32547021e-01 -1.00384034e-01 -9.44513604e-02 1.05935335e+00
1.10822611e-01 1.21968739e-01 -1.05797000e-01 -5.27403116e-01
-2.58428156e-01 -5.09992167e-02 -4.46600944e-01 5.83397150e-01
4.43372689e-02 1.35999584e+00 -9.37314987e-01 -4.74267267e-02
1.01940870e-01 8.84445429e-01 -3.20523888e-01 -1.34749904e-01
-3.36159259e-01 6.80843055e-01 -5.14147103e-01 7.16560185e-01
4.74110842e-01 -5.18166840e-01 1.23563923e-01 3.02328646e-01
-2.62872010e-01 1.31987512e-01 -8.74144375e-01 1.79062462e+00
-4.73951474e-02 1.10839272e+00 -1.99885760e-02 -6.13410234e-01
7.72754908e-01 2.64607221e-01 8.10000420e-01 -5.30390263e-01
8.53571743e-02 1.48081750e-01 1.89862445e-01 -1.03961289e-01
3.88805419e-01 -4.83800359e-02 1.44721523e-01 7.06356645e-01
1.13694422e-01 -1.70037895e-01 3.81905913e-01 4.53928351e-01
9.83059287e-01 -4.85756695e-02 2.97111034e-01 -2.86956459e-01
-2.08081529e-02 -7.77904987e-02 1.29743263e-01 1.00212109e+00
-2.13283911e-01 2.84617305e-01 2.23335460e-01 -4.34396446e-01
-1.04586792e+00 -9.64335382e-01 -6.95745945e-02 8.17394614e-01
-2.26125959e-02 -8.97181213e-01 -6.44913554e-01 -1.63766131e-01
-2.36119047e-01 3.88577789e-01 -9.63878512e-01 -4.21443395e-02
-2.97340661e-01 -1.66475892e+00 5.41870534e-01 5.47545850e-01
2.28042096e-01 -9.08090711e-01 -2.27539718e-01 1.01303771e-01
5.33911549e-02 -6.21333182e-01 -4.03124481e-01 -1.11446254e-01
-1.13091958e+00 -7.02418685e-01 -1.39798570e+00 -5.76281309e-01
3.58863205e-01 5.27039707e-01 1.03055859e+00 -1.09698497e-01
-3.87248814e-01 5.68117738e-01 -1.11077897e-01 -7.16727436e-01
-4.16386575e-01 3.18911485e-02 5.69355525e-02 -4.04185176e-01
6.11100972e-01 -4.18989062e-01 -8.66009474e-01 1.07671648e-01
-1.00835848e+00 -1.41461685e-01 5.19566119e-01 7.64508665e-01
6.78216815e-01 -6.96720481e-02 5.61038315e-01 -8.37109745e-01
1.17866457e+00 -7.96295822e-01 -3.93192738e-01 -1.29409805e-01
-9.59776878e-01 -2.81476676e-01 6.12310693e-02 -6.61009490e-01
-1.00328994e+00 -4.66573685e-01 -3.87838185e-01 -3.02843869e-01
5.22665791e-02 4.80506212e-01 2.34233513e-01 1.10452056e-01
6.76851392e-01 5.83607972e-01 1.76966593e-01 -2.59000778e-01
6.83830142e-01 3.83609474e-01 -1.03079721e-01 -4.47033823e-01
3.76251698e-01 7.82991886e-01 7.49619827e-02 -1.30588460e+00
-4.47979540e-01 -3.87999177e-01 -3.45890969e-01 -1.88081458e-01
6.78221643e-01 -7.83253968e-01 -3.69877398e-01 8.85682762e-01
-1.17306066e+00 -6.11216426e-01 -3.59291911e-01 4.87597644e-01
-7.47423708e-01 3.64373386e-01 -1.19407332e+00 -7.31383383e-01
-3.56382549e-01 -1.35212791e+00 1.03731227e+00 3.57826680e-01
-4.42294657e-01 -1.54505444e+00 5.61716676e-01 1.02773570e-01
6.11612260e-01 -6.34515136e-02 5.58379829e-01 -4.20897722e-01
-8.47799897e-01 2.22877916e-02 2.72417456e-01 -1.09361194e-01
8.24378580e-02 5.02099991e-01 -8.97916436e-01 -1.49567857e-01
-4.92757410e-02 1.81030948e-02 1.21292150e+00 1.26599872e+00
9.37203944e-01 2.44614109e-02 -9.03941870e-01 6.73679471e-01
9.21596169e-01 5.75955391e-01 6.21594548e-01 2.24350333e-01
4.31389689e-01 3.33219051e-01 3.16000551e-01 6.02226138e-01
4.34601694e-01 4.74184811e-01 -2.97701452e-02 -2.19579622e-01
-5.51198721e-01 -1.50761485e-01 2.87553817e-01 1.03147995e+00
-2.89051026e-01 -9.81171429e-01 -6.87062144e-01 3.70678604e-01
-1.77588081e+00 -1.32172406e+00 -4.60419327e-01 1.95374584e+00
6.34889603e-01 -1.81288406e-01 4.88179654e-01 -4.66253877e-01
6.57422543e-01 4.07525659e-01 -9.50489640e-01 -9.36812386e-02
-4.47952271e-01 1.10339299e-02 4.26531702e-01 6.58923090e-01
-8.19535911e-01 1.10996962e+00 8.00032616e+00 1.15119672e+00
-1.13928771e+00 3.69553745e-01 9.12803054e-01 -3.35055292e-01
-6.02873266e-01 -8.48492235e-02 -1.09116173e+00 3.75112891e-01
9.35087264e-01 -4.88198668e-01 3.50193322e-01 5.81110597e-01
3.54133248e-01 2.64247833e-03 -7.02117920e-01 1.05048454e+00
1.88313991e-01 -1.87271893e+00 1.19774856e-01 6.47939086e-01
1.10497844e+00 4.76624221e-01 6.51409388e-01 -9.51920152e-02
5.63096285e-01 -8.98593426e-01 4.73341554e-01 6.91756666e-01
6.50251269e-01 -4.95893329e-01 1.54588625e-01 2.51154214e-01
-9.17469501e-01 8.99626762e-02 -4.52420712e-01 -1.00738622e-01
5.72555363e-01 8.22751880e-01 -5.00684261e-01 1.66428089e-01
6.84911728e-01 1.31667423e+00 -4.00355943e-02 1.08281088e+00
-1.47053942e-01 7.07594872e-01 -8.04685056e-02 -2.94924110e-01
9.85866189e-02 -3.68157417e-01 7.76400149e-01 1.45385587e+00
6.44268632e-01 2.49611605e-02 -1.87259614e-01 9.60184991e-01
-1.33761521e-02 -1.59355402e-01 -9.29097533e-01 -4.91216391e-01
4.96095121e-01 9.76141274e-01 -9.26789463e-01 -5.38037956e-01
-5.60495198e-01 9.99351382e-01 -2.79758386e-02 6.22156322e-01
-9.22261596e-01 7.73713440e-02 8.90988231e-01 1.08966790e-01
1.69250444e-01 -7.38990426e-01 -1.97260216e-01 -1.09304881e+00
-6.77784920e-01 -8.17383111e-01 3.98941249e-01 -6.69030249e-01
-1.41559458e+00 4.98933464e-01 1.46097494e-02 -7.67636657e-01
-2.29817718e-01 -5.56859493e-01 -4.16406900e-01 8.31556380e-01
-1.23713040e+00 -8.86024892e-01 -1.10744223e-01 2.64633477e-01
1.04743540e+00 -1.94564626e-01 5.61680555e-01 3.34572554e-01
-6.43891454e-01 1.34080067e-01 4.67509151e-01 -3.24958354e-01
7.52727628e-01 -1.03821397e+00 1.02204490e+00 6.02855682e-01
2.82515556e-01 1.03865123e+00 6.49600029e-01 -9.56465304e-01
-1.39935780e+00 -8.58755946e-01 7.70084739e-01 -6.93089783e-01
7.76313961e-01 -3.92497718e-01 -6.93131030e-01 6.91451728e-01
5.83796680e-01 -6.78899586e-01 1.09842288e+00 -2.91723348e-02
8.63615870e-02 3.77774298e-01 -5.91936111e-01 8.06621194e-01
1.15226281e+00 -2.70315468e-01 2.13843137e-01 6.47693276e-01
3.34652841e-01 -4.50416118e-01 -9.14293051e-01 -6.87140226e-02
8.29416990e-01 -9.67107773e-01 1.12318218e+00 -5.12059093e-01
5.31191945e-01 2.45204438e-02 1.33856446e-01 -1.25493014e+00
-8.96778822e-01 -1.12807178e+00 -5.56348205e-01 9.51261759e-01
2.56630301e-01 -4.56968367e-01 8.40252161e-01 4.26139504e-01
-2.12444663e-02 -1.13769436e+00 -5.46005547e-01 -6.65009737e-01
4.93948430e-01 -4.47359622e-01 3.59579057e-01 7.87875295e-01
-1.43874332e-01 5.10993540e-01 -5.74997127e-01 -5.15544415e-01
5.66405594e-01 -5.88545613e-02 7.23086357e-01 -9.72283304e-01
-2.45884478e-01 -1.08337784e+00 -2.91725427e-01 -1.83500147e+00
-2.05444530e-01 -7.19524622e-01 -3.63314301e-01 -2.03411651e+00
3.99120420e-01 -2.01679170e-01 4.90791388e-02 -1.27122045e-01
3.71806845e-02 3.46377283e-01 -7.61720091e-02 4.39997554e-01
-5.40245712e-01 6.00962579e-01 1.73273325e+00 -2.27556452e-01
-6.73893020e-02 1.83608100e-01 -9.87323046e-01 4.62159067e-01
9.51495528e-01 -4.55092192e-01 -7.96195447e-01 -5.73310435e-01
4.24994111e-01 -9.72968787e-02 4.66893017e-02 -4.15661991e-01
4.65952903e-01 -4.67713624e-01 4.16959763e-01 -4.91514385e-01
4.56423759e-01 -1.61983147e-01 3.57118189e-01 6.21606410e-01
-2.45693624e-01 5.71826220e-01 1.69574261e-01 7.90622830e-01
-1.24080470e-02 2.65488997e-02 6.02409124e-01 -3.47772390e-01
-7.18462229e-01 7.96738684e-01 -9.29755509e-01 -1.05856389e-01
1.14987040e+00 -2.14551225e-01 -4.53127503e-01 -9.44117546e-01
-7.35203624e-01 -1.30692035e-01 4.93865341e-01 7.81562626e-01
5.84972143e-01 -1.18194175e+00 -7.13036418e-01 5.58603220e-02
-2.24988945e-02 -2.06095561e-01 6.19153738e-01 9.40073252e-01
-1.93781719e-01 6.61693692e-01 1.88890487e-01 -5.68167865e-01
-1.01308191e+00 7.11615682e-01 2.00339377e-01 -1.64234221e-01
-2.89841801e-01 1.16327524e+00 5.90466976e-01 1.52742311e-01
-2.37082131e-02 -2.32366472e-01 1.84219815e-02 -8.35480988e-02
6.08925045e-01 6.78828657e-01 -2.06904709e-01 -4.69823480e-01
-3.25199693e-01 4.77328509e-01 -1.97950080e-01 -1.97776124e-01
1.29884362e+00 -2.42552072e-01 -2.28056479e-02 5.70643067e-01
8.12341154e-01 2.05506682e-02 -1.24149132e+00 -1.31541178e-01
-5.04525363e-01 -6.04655147e-02 1.59824137e-02 -6.14772260e-01
-9.99571085e-01 1.08460474e+00 4.28972155e-01 3.05828601e-01
8.91515732e-01 4.94315416e-01 6.49590790e-01 -1.40843391e-01
1.15004487e-01 -8.15968275e-01 4.88330573e-01 6.41243577e-01
9.27558541e-01 -8.76737893e-01 9.94694456e-02 -2.37576261e-01
-7.05657005e-01 9.16789234e-01 2.98758209e-01 -1.86206505e-01
1.11803639e+00 5.37613869e-01 -3.32314819e-02 -5.27943373e-01
-1.10356891e+00 1.47819901e-02 2.81839073e-01 9.59816098e-01
9.43658710e-01 -3.60047594e-02 -3.04164559e-01 2.57036775e-01
-2.21158758e-01 1.99370477e-02 4.99103576e-01 7.88297832e-01
-3.63112926e-01 -1.49979377e+00 -1.22877665e-01 5.95420420e-01
-2.75059402e-01 -4.79896545e-01 -5.30174971e-01 5.20095706e-01
-1.42937720e-01 9.00628269e-01 1.68978110e-01 -1.29622832e-01
-1.57048583e-01 -2.26579979e-01 7.34332204e-01 -4.84339476e-01
-1.24372847e-01 7.09291935e-01 -3.23879957e-01 -2.46749684e-01
-6.55374646e-01 -1.09387267e+00 -7.11293042e-01 -7.93905556e-01
-2.48149738e-01 -2.89893523e-02 6.82998121e-01 6.91920996e-01
8.02976131e-01 5.96078038e-01 8.54813159e-02 -6.23882353e-01
-9.58388746e-02 -9.66657937e-01 -5.47579885e-01 -2.32484862e-01
3.70230168e-01 -7.31004059e-01 3.11653893e-02 2.05347896e-01]
|
[11.272176742553711, 0.00010863344505196437]
|
8d02e6fd-834f-416d-b18d-3def3ccf8e90
|
pttr-relational-3d-point-cloud-object
|
2112.02857
| null |
https://arxiv.org/abs/2112.02857v5
|
https://arxiv.org/pdf/2112.02857v5.pdf
|
PTTR: Relational 3D Point Cloud Object Tracking with Transformer
|
In a point cloud sequence, 3D object tracking aims to predict the location and orientation of an object in the current search point cloud given a template point cloud. Motivated by the success of transformers, we propose Point Tracking TRansformer (PTTR), which efficiently predicts high-quality 3D tracking results in a coarse-to-fine manner with the help of transformer operations. PTTR consists of three novel designs. 1) Instead of random sampling, we design Relation-Aware Sampling to preserve relevant points to given templates during subsampling. 2) Furthermore, we propose a Point Relation Transformer (PRT) consisting of a self-attention and a cross-attention module. The global self-attention operation captures long-range dependencies to enhance encoded point features for the search area and the template, respectively. Subsequently, we generate the coarse tracking results by matching the two sets of point features via cross-attention. 3) Based on the coarse tracking results, we employ a novel Prediction Refinement Module to obtain the final refined prediction. In addition, we create a large-scale point cloud single object tracking benchmark based on the Waymo Open Dataset. Extensive experiments show that PTTR achieves superior point cloud tracking in both accuracy and efficiency.
|
['Shijian Lu', 'Haiyu Zhao', 'Zhongang Cai', 'Liang Pan', 'Tianrui Liu', 'Yueru Luo', 'Zhipeng Luo', 'Changqing Zhou']
|
2021-12-06
| null |
http://openaccess.thecvf.com//content/CVPR2022/html/Zhou_PTTR_Relational_3D_Point_Cloud_Object_Tracking_With_Transformer_CVPR_2022_paper.html
|
http://openaccess.thecvf.com//content/CVPR2022/papers/Zhou_PTTR_Relational_3D_Point_Cloud_Object_Tracking_With_Transformer_CVPR_2022_paper.pdf
|
cvpr-2022-1
|
['3d-object-tracking']
|
['computer-vision']
|
[-5.84557056e-02 -3.01318616e-01 -8.21006894e-02 -1.93645835e-01
-7.60622263e-01 -2.54876971e-01 5.71032107e-01 7.57252276e-02
-6.74601719e-02 9.90965217e-02 5.73450215e-02 -3.13213356e-02
-1.54069677e-01 -8.84686410e-01 -1.13396299e+00 -4.90980625e-01
2.21293136e-01 6.06148243e-01 7.69783854e-01 -6.49745762e-02
2.77880490e-01 1.04642189e+00 -1.58865464e+00 -6.21642591e-03
9.26496565e-01 1.38067722e+00 5.00496387e-01 2.08066612e-01
-2.56294966e-01 3.30731511e-01 -3.10982913e-01 -2.65836269e-01
5.44084370e-01 2.05943942e-01 -3.51546854e-01 2.18575120e-01
7.49654174e-01 -3.37356180e-01 -4.38562691e-01 1.06583023e+00
4.35381919e-01 1.59469143e-01 3.02131265e-01 -1.19580173e+00
-6.48585975e-01 3.68985608e-02 -7.06126213e-01 1.42484412e-01
1.28211766e-01 3.68565708e-01 8.68738055e-01 -1.05058992e+00
5.06351888e-01 1.38417757e+00 8.57423842e-01 3.90603185e-01
-1.05947793e+00 -9.21616673e-01 4.54171896e-01 1.24272607e-01
-1.41128576e+00 -2.26081759e-01 9.75112021e-01 -3.50310385e-01
9.84739304e-01 1.74843818e-01 9.72561598e-01 5.36746740e-01
2.40163371e-01 7.04994142e-01 5.54636598e-01 -5.78698069e-02
1.72972232e-02 -2.57973999e-01 -2.25168876e-02 6.60053432e-01
7.05515891e-02 4.21045572e-01 -2.84791976e-01 -2.34304622e-01
9.51270521e-01 4.43946064e-01 -2.53098905e-01 -7.34384954e-01
-1.34967160e+00 5.36438882e-01 1.05516684e+00 9.21615362e-02
-7.34437764e-01 2.38219157e-01 1.23857878e-01 -5.22858240e-02
5.18494248e-01 1.55541990e-02 -4.62388515e-01 3.15792412e-01
-8.03575337e-01 3.95243764e-01 2.00097978e-01 1.49073160e+00
8.58420610e-01 -1.90926373e-01 -5.46029627e-01 3.61664712e-01
5.72380841e-01 8.68264139e-01 1.81007475e-01 -6.54193640e-01
4.39384490e-01 9.40807402e-01 2.47996673e-01 -1.01896489e+00
-1.56304210e-01 -7.56405950e-01 -6.46321058e-01 3.40470225e-01
-5.22610098e-02 5.33255816e-01 -9.97298121e-01 1.46488011e+00
8.16911101e-01 6.46326482e-01 -4.58764464e-01 1.07764375e+00
7.11021721e-01 6.99142396e-01 1.71063915e-01 -6.92194551e-02
1.30435216e+00 -8.01655233e-01 -3.69163007e-01 6.35497868e-02
3.26957434e-01 -5.88063896e-01 7.67287791e-01 -1.34932339e-01
-1.15568137e+00 -9.91916060e-01 -8.95524442e-01 -2.52349347e-01
-9.71592963e-02 2.78843008e-02 3.25675428e-01 1.39886454e-01
-7.18430698e-01 6.82277203e-01 -1.12323523e+00 -1.65659562e-02
8.18049967e-01 6.02682292e-01 -9.58384797e-02 -2.24932097e-02
-5.95787942e-01 6.27530217e-01 1.77847818e-01 9.02292505e-02
-5.92133403e-01 -1.21515989e+00 -8.27008605e-01 1.50445461e-01
2.88014650e-01 -1.02459574e+00 1.05220473e+00 -4.30832893e-01
-1.22110426e+00 7.94242740e-01 -5.28042555e-01 -4.21941757e-01
4.19765174e-01 -4.72263157e-01 -1.96705624e-01 -1.07481390e-01
4.03945923e-01 8.65499794e-01 9.39715922e-01 -1.18801546e+00
-1.00928509e+00 -7.13378549e-01 -3.47671658e-01 2.06159368e-01
2.18038231e-01 -1.09527044e-01 -9.28539336e-01 -6.11327648e-01
5.16213119e-01 -9.34600472e-01 -2.28514224e-01 3.31766725e-01
-3.08698714e-01 -6.16235733e-01 1.15023077e+00 -3.68568927e-01
8.65197897e-01 -2.25710893e+00 1.15035541e-01 2.10586742e-01
4.06960875e-01 1.88090801e-01 -5.07655255e-02 -2.35182062e-01
-8.39305967e-02 -2.73436636e-01 1.55231148e-01 -4.35418963e-01
1.03601731e-01 -2.92024445e-02 -6.30565464e-01 4.67229098e-01
5.22656679e-01 1.33357537e+00 -8.55052471e-01 -5.49315929e-01
5.78073919e-01 6.01506054e-01 -6.93327010e-01 1.31572142e-01
-5.08844495e-01 3.67539078e-01 -9.52530265e-01 8.27564478e-01
9.26063776e-01 -5.70351481e-01 -5.42941570e-01 -6.23355925e-01
-4.81647253e-01 3.26201141e-01 -8.93240809e-01 1.83183551e+00
-1.85222834e-01 1.76329166e-01 -2.28895843e-01 -3.42547148e-01
1.18769670e+00 4.91509354e-03 7.58701801e-01 -6.43621206e-01
1.55353099e-01 1.30210534e-01 -2.95523703e-01 1.39257967e-01
5.70541084e-01 -1.18880291e-02 1.18363470e-01 9.14759710e-02
-1.33476362e-01 3.84679902e-03 -3.27902287e-01 -2.79535092e-02
9.71360803e-01 4.42570627e-01 7.19915479e-02 -6.60951659e-02
6.83158159e-01 2.83207715e-01 8.83931160e-01 5.15867889e-01
-3.76601934e-01 5.83357215e-01 -1.27358630e-01 -6.15394950e-01
-1.01729906e+00 -1.00992894e+00 -1.29134819e-01 6.69623673e-01
6.43429220e-01 -2.70229131e-01 -2.17432007e-01 -6.99175417e-01
3.55723798e-01 4.05438691e-01 -5.45435786e-01 -2.43589357e-01
-8.10725510e-01 -6.91071525e-02 -4.87656705e-02 8.12698781e-01
5.68207026e-01 -1.07681715e+00 -8.34204376e-01 2.36531153e-01
1.59291908e-01 -1.08084297e+00 -9.57682490e-01 -2.33861078e-02
-1.21037626e+00 -8.31996799e-01 -5.72458804e-01 -8.37423444e-01
6.07324600e-01 5.92963874e-01 9.40628231e-01 2.41288260e-01
1.81270301e-01 2.42110521e-01 -2.57061452e-01 -3.45411658e-01
6.42677471e-02 -8.04491714e-03 1.00044288e-01 -2.22239207e-04
6.74748898e-01 -4.40503865e-01 -5.58032751e-01 5.19157052e-01
-3.75676632e-01 7.32752681e-02 7.34238923e-01 6.43603742e-01
1.31803071e+00 2.34683119e-02 5.20484149e-02 -2.04862013e-01
7.91967586e-02 -1.64958760e-01 -1.10800827e+00 1.46427169e-01
-2.21745700e-01 6.34411126e-02 2.33340636e-01 -5.80040276e-01
-7.15612471e-01 4.66931581e-01 -1.91892043e-01 -1.50624418e+00
-8.32954124e-02 -7.71578103e-02 -4.32372332e-01 -3.61009121e-01
1.62331849e-01 4.10569787e-01 -8.86003450e-02 -6.66428208e-01
1.70331329e-01 2.00454488e-01 6.57066524e-01 -4.45306361e-01
1.28350675e+00 5.91678441e-01 7.00987875e-03 -4.42430884e-01
-8.12057316e-01 -4.60962355e-01 -8.22607458e-01 -1.90439269e-01
1.03210402e+00 -1.01350558e+00 -1.09182549e+00 1.94644347e-01
-1.23080885e+00 -2.76447013e-02 -4.71249998e-01 4.72322106e-01
-4.94190812e-01 1.38945445e-01 -2.84666300e-01 -6.14465952e-01
-6.43573880e-01 -1.33942389e+00 1.72958410e+00 3.15534025e-01
2.26464540e-01 -4.97956872e-01 6.74268557e-03 6.90676551e-03
1.05996937e-01 1.20125704e-01 5.29547572e-01 -4.06086624e-01
-1.42596996e+00 -3.13761115e-01 -5.79276681e-01 -2.08063215e-01
1.34220019e-01 -2.56750017e-01 -6.74083889e-01 -3.70341986e-01
5.31480974e-03 2.39714503e-01 5.53630829e-01 5.91325879e-01
1.31061804e+00 6.35849014e-02 -8.22409272e-01 9.47650909e-01
1.26916480e+00 2.34945148e-01 5.08109987e-01 3.37590337e-01
9.80485022e-01 6.22856505e-02 1.01259661e+00 3.19947779e-01
2.84412801e-01 1.05317140e+00 5.92351615e-01 2.92996049e-01
-2.42053792e-01 -7.25699306e-01 -2.06619371e-02 6.36825562e-01
-4.43572290e-02 3.34530711e-01 -7.67129779e-01 4.70133543e-01
-1.76855016e+00 -9.47774112e-01 -9.11945254e-02 2.31091356e+00
3.84223729e-01 3.83691579e-01 8.44009742e-02 -1.87361941e-01
8.38928580e-01 3.05414237e-02 -8.66689742e-01 5.41447163e-01
2.27907106e-01 9.78148282e-02 4.70934719e-01 2.54240245e-01
-1.08635628e+00 9.35568452e-01 4.69001913e+00 8.56610954e-01
-1.14141417e+00 1.91957899e-03 7.78773725e-02 -8.86913687e-02
-2.61964738e-01 -5.39930398e-03 -1.33320034e+00 7.15470552e-01
3.45673889e-01 -1.93645254e-01 4.33065929e-02 9.89624321e-01
-5.01077473e-02 6.79321408e-01 -9.87220824e-01 9.71847951e-01
-3.07774872e-01 -1.58598781e+00 1.38369009e-01 2.52705395e-01
3.65380287e-01 3.54273736e-01 -6.95763975e-02 3.69337380e-01
9.11301747e-02 -4.41698790e-01 1.06231475e+00 5.92264473e-01
5.26759505e-01 -7.70284474e-01 3.86875004e-01 4.71934736e-01
-1.82938707e+00 1.75839793e-02 -5.42984426e-01 4.51810569e-01
1.54690742e-01 3.89166445e-01 -6.21431291e-01 7.58484304e-01
1.03699207e+00 1.05246234e+00 -4.53103364e-01 1.48548687e+00
8.57293755e-02 6.26148805e-02 -6.24595523e-01 6.48688301e-02
1.23208947e-01 -1.07385442e-01 9.40248311e-01 6.48494482e-01
4.50390369e-01 2.83218920e-01 2.90926516e-01 1.17949939e+00
-9.21076350e-03 -1.81888506e-01 -3.62571627e-01 4.77345973e-01
8.61180186e-01 9.99022663e-01 -5.58532596e-01 -2.47419804e-01
-3.03823411e-01 6.87546432e-01 3.71638477e-01 -5.29340617e-02
-9.01145160e-01 -1.77464262e-01 8.59309494e-01 2.67175972e-01
9.69695032e-01 -1.12351753e-01 -3.13692838e-01 -9.90371287e-01
1.92027092e-01 -4.95238125e-01 2.01059312e-01 -9.75978911e-01
-1.25728810e+00 8.01879227e-01 -1.90319806e-01 -1.82306242e+00
2.72818208e-01 -3.78560483e-01 -5.54646015e-01 1.05415726e+00
-1.53408241e+00 -1.37915683e+00 -6.15946531e-01 5.62318444e-01
6.20929182e-01 8.86528045e-02 3.56398195e-01 4.15096670e-01
-4.25876677e-01 4.47048962e-01 -5.15410841e-01 5.19313738e-02
2.28200823e-01 -9.55906272e-01 9.47601438e-01 6.11128926e-01
7.66839599e-03 7.47908950e-01 3.13833296e-01 -1.15023839e+00
-1.54984558e+00 -1.56491351e+00 7.80629694e-01 -6.76669836e-01
4.51756120e-01 -2.70439059e-01 -1.19066036e+00 8.42696846e-01
-3.11481327e-01 4.28092062e-01 4.30352353e-02 -1.33225322e-01
-2.78114766e-01 -1.90517530e-01 -9.90711391e-01 4.87517387e-01
1.30237567e+00 -3.19156468e-01 -8.37120235e-01 1.93014264e-01
1.19551778e+00 -9.29104507e-01 -1.06276178e+00 7.20656157e-01
3.84535074e-01 -6.33522809e-01 1.40649962e+00 -3.03204298e-01
-1.57634709e-02 -8.44068527e-01 -1.23474091e-01 -9.49252009e-01
-8.85454595e-01 -4.87583697e-01 -3.96283418e-01 1.02480102e+00
-1.04535937e-01 -4.26087558e-01 1.07562256e+00 5.01552701e-01
-5.37148297e-01 -8.70057285e-01 -1.04638553e+00 -8.36671770e-01
-2.25013703e-01 -4.49283093e-01 1.23272383e+00 5.64151943e-01
-6.22729123e-01 2.47520924e-01 -1.29341930e-01 6.55991256e-01
9.99780059e-01 7.64044940e-01 9.19816375e-01 -1.50287116e+00
-2.12450624e-02 -4.98027176e-01 -5.55109978e-01 -1.75518048e+00
1.51301846e-01 -7.44162083e-01 1.97109178e-01 -1.19251835e+00
-9.09925848e-02 -8.83478880e-01 -1.99090838e-01 3.71436477e-01
-3.02246839e-01 1.40857577e-01 3.82594556e-01 5.04105210e-01
-6.46458864e-01 9.73654985e-01 1.42896056e+00 -1.73310041e-01
-3.79606754e-01 4.26572382e-01 -4.15948361e-01 4.98180687e-01
3.76492709e-01 -5.38250685e-01 -2.98482925e-03 -5.36303699e-01
-3.56833011e-01 9.48794857e-02 6.89225674e-01 -1.18949938e+00
5.01555860e-01 4.02277783e-02 6.81664944e-01 -1.64553475e+00
5.37952006e-01 -1.24237216e+00 2.76628822e-01 5.69396615e-01
1.15796430e-02 2.05226511e-01 3.21466804e-01 7.73742616e-01
9.77715552e-02 1.84905246e-01 7.71344602e-01 1.48104221e-01
-7.72206187e-01 1.07045352e+00 5.84284842e-01 -4.01932418e-01
1.07095861e+00 -3.91228110e-01 -5.20183779e-02 2.50777274e-01
-4.13258821e-01 5.77493846e-01 7.60462284e-01 6.40773594e-01
7.75299191e-01 -1.69367576e+00 -4.49378371e-01 4.76227224e-01
1.87745422e-01 5.75959682e-01 2.37031609e-01 8.12451184e-01
-2.07490236e-01 5.69760203e-01 -1.34059176e-01 -1.17980707e+00
-1.17172861e+00 8.54265749e-01 3.50034267e-01 -6.41788766e-02
-1.26571167e+00 7.25310624e-01 3.79898638e-01 -4.02921647e-01
2.27355212e-01 -6.45365059e-01 8.27659518e-02 -4.54484344e-01
5.43221235e-01 1.28216995e-02 1.05138011e-01 -8.22757959e-01
-5.70809305e-01 1.07525456e+00 -1.36952475e-01 4.18887943e-01
1.28298759e+00 4.88843173e-02 1.70007184e-01 2.56240536e-02
1.01619613e+00 5.84937120e-03 -1.70667207e+00 -5.03911316e-01
-1.13503434e-01 -9.86376107e-01 2.78039634e-01 -3.19590688e-01
-1.25615990e+00 5.66371560e-01 6.41279578e-01 3.39239389e-02
1.08905232e+00 2.03656346e-01 9.77056086e-01 1.05149597e-01
5.05822301e-01 -2.97331601e-01 -2.79644191e-01 4.50269133e-01
9.03634369e-01 -1.07168746e+00 -6.15961254e-02 -5.65207958e-01
-3.09690952e-01 6.91955209e-01 8.40240240e-01 -3.57113957e-01
6.55237198e-01 1.32348239e-01 -3.56138319e-01 -3.90924454e-01
-8.24608326e-01 -3.51786107e-01 7.19754338e-01 6.45638466e-01
-1.23311974e-01 -3.13472569e-01 3.19743007e-01 5.23793519e-01
-1.19020350e-01 2.31396481e-01 -4.10788476e-01 8.68772030e-01
-5.27997732e-01 -7.55790412e-01 -5.18043518e-01 3.48919362e-01
1.26389712e-01 1.38046205e-01 -1.47179931e-01 7.30621934e-01
1.46629825e-01 2.45848149e-01 4.11182076e-01 -6.52949035e-01
8.19194436e-01 -2.65855551e-01 4.86698747e-01 -4.31073248e-01
-6.76447272e-01 2.30246127e-01 -6.54064476e-01 -7.16385543e-01
-2.31991082e-01 -8.23319793e-01 -1.30021954e+00 -2.47289076e-01
-5.45515597e-01 1.93245083e-01 4.47321415e-01 7.85410464e-01
8.07369828e-01 6.64643943e-01 5.24606705e-01 -1.35124350e+00
-4.49096054e-01 -7.63012171e-01 -2.00857610e-01 3.96460354e-01
4.58823830e-01 -9.85420167e-01 4.93674912e-02 -2.72705972e-01]
|
[6.641216278076172, -2.388139009475708]
|
16144c61-c89c-492a-ba15-8ec78ad6431a
|
enhanced-characterness-for-text-detection-in
|
1712.04927
| null |
http://arxiv.org/abs/1712.04927v1
|
http://arxiv.org/pdf/1712.04927v1.pdf
|
Enhanced Characterness for Text Detection in the Wild
|
Text spotting is an interesting research problem as text may appear at any
random place and may occur in various forms. Moreover, ability to detect text
opens the horizons for improving many advanced computer vision problems. In
this paper, we propose a novel language agnostic text detection method
utilizing edge enhanced Maximally Stable Extremal Regions in natural scenes by
defining strong characterness measures. We show that a simple combination of
characterness cues help in rejecting the non text regions. These regions are
further fine-tuned for rejecting the non-textual neighbor regions.
Comprehensive evaluation of the proposed scheme shows that it provides
comparative to better generalization performance to the traditional methods for
this task.
|
['Brejesh lall', 'Siddharth Srivastava', 'Prerana Mukherjee', 'Aarushi Agrawal']
|
2017-12-04
| null | null | null | null |
['text-spotting']
|
['computer-vision']
|
[ 6.37937784e-01 -4.56158280e-01 -2.31987461e-01 -1.06108077e-01
-5.27002692e-01 -6.52546704e-01 8.14983726e-01 3.52631122e-01
-5.00922203e-01 7.74419785e-01 6.30347356e-02 -1.89858481e-01
4.14978974e-02 -5.83036840e-01 -8.09964910e-02 -7.74892449e-01
2.76037186e-01 4.20104831e-01 6.39097035e-01 -1.62002668e-01
9.65685189e-01 6.32331073e-01 -1.54018068e+00 4.83272135e-01
9.40782845e-01 6.97273016e-01 2.41892502e-01 8.07002962e-01
-4.83691782e-01 4.23010707e-01 -4.76402551e-01 -1.56655625e-01
3.18336189e-01 -3.45133662e-01 -3.75041068e-01 5.10492325e-01
7.32030213e-01 -1.03700548e-01 -1.12272725e-01 1.28364670e+00
4.20393586e-01 4.12692428e-01 9.96646881e-01 -8.25501502e-01
-6.32426322e-01 3.32762212e-01 -1.15517867e+00 6.69344902e-01
3.81387979e-01 -2.06043869e-01 1.06325054e+00 -1.30105031e+00
5.60235500e-01 1.06115699e+00 6.45504475e-01 2.92885065e-01
-8.06015611e-01 -2.27116123e-01 1.07418276e-01 2.98238844e-01
-1.52539468e+00 -3.24778587e-01 7.11268127e-01 -3.37556064e-01
7.49279499e-01 6.74172223e-01 2.19531640e-01 7.37886608e-01
4.91913348e-01 1.08386075e+00 1.16056991e+00 -8.77953768e-01
2.19917353e-02 4.25552398e-01 3.71408492e-01 9.36972797e-01
4.66065437e-01 -4.13571358e-01 -4.87800479e-01 -1.14808045e-01
4.20854539e-01 -1.01801150e-01 -2.35616639e-01 -5.40864430e-02
-1.14420664e+00 7.78442323e-01 -1.43642858e-01 5.45717955e-01
-8.15663487e-02 -3.43258113e-01 5.16848743e-01 4.17148136e-02
5.41009545e-01 2.16337577e-01 -6.82248846e-02 1.86662212e-01
-1.29296076e+00 -8.06040391e-02 5.83945692e-01 9.12511051e-01
6.26914680e-01 1.00209802e-01 -5.35984755e-01 9.97860193e-01
-1.08165117e-02 3.73485267e-01 6.89390719e-01 -1.59526870e-01
5.19110799e-01 7.98463941e-01 -1.29784092e-01 -1.15686977e+00
-3.11328858e-01 -3.70638698e-01 -8.30446124e-01 1.90878808e-01
2.52954543e-01 1.60235595e-02 -1.24518061e+00 7.39911079e-01
3.97504658e-01 -2.68048495e-01 -2.18029067e-01 6.83269858e-01
5.64125001e-01 7.46657610e-01 -2.42408410e-01 -2.30695918e-01
1.35588908e+00 -9.84808683e-01 -8.60012054e-01 -3.20125103e-01
3.44229341e-01 -1.33868980e+00 9.93173003e-01 5.86476922e-01
-6.86066926e-01 -1.77640587e-01 -1.01310825e+00 -2.53212359e-03
-7.27062047e-01 6.16615057e-01 4.17486906e-01 8.77143502e-01
-9.62789893e-01 3.06386888e-01 -3.79349619e-01 -8.59042883e-01
2.33036488e-01 2.62633055e-01 -1.92309678e-01 1.58886001e-01
-7.74206161e-01 6.64020538e-01 7.61322737e-01 2.16961861e-01
-2.08186477e-01 3.03438634e-01 -6.91071928e-01 -1.73087180e-01
5.27786434e-01 -3.18281144e-01 7.35832393e-01 -1.17696238e+00
-1.10868359e+00 1.11787117e+00 -3.70636344e-01 -3.30815256e-01
9.73624349e-01 2.12972276e-02 -6.28790736e-01 5.49906135e-01
2.60260850e-01 3.91861767e-01 1.41420710e+00 -8.58955801e-01
-9.25799191e-01 -3.71776044e-01 -6.41015887e-01 5.45333743e-01
-7.12315500e-01 1.63890108e-01 -7.86444962e-01 -1.18806970e+00
5.92871189e-01 -7.40575850e-01 -5.39529920e-02 2.43383452e-01
-8.33143294e-01 -3.75779986e-01 1.27934432e+00 -4.74450767e-01
1.28604543e+00 -1.81519544e+00 -4.45049763e-01 4.16088998e-01
3.04660290e-01 8.06662515e-02 1.85775980e-01 5.73872387e-01
1.93470955e-01 3.26713562e-01 -1.73824489e-01 3.76483165e-02
-1.28230050e-01 -2.02670738e-01 -1.56784564e-01 7.51175821e-01
5.75145967e-02 5.27572215e-01 -5.20164847e-01 -1.10840929e+00
5.02941549e-01 2.57917881e-01 6.37177378e-02 -4.32300776e-01
3.07934621e-04 -1.45957366e-01 -8.03830028e-01 8.17365587e-01
7.25897729e-01 -1.18475899e-01 -8.53719637e-02 5.15156016e-02
-2.30133891e-01 -3.34465891e-01 -1.34082127e+00 8.47932994e-01
2.03173250e-01 1.37574160e+00 1.37118725e-02 -1.01123393e+00
9.13218796e-01 7.87548870e-02 2.35893726e-01 -4.36232388e-01
3.57303351e-01 4.53752512e-03 -2.54517734e-01 -6.21086061e-01
1.09988880e+00 -4.61515859e-02 2.31990114e-01 2.42495343e-01
-5.20799100e-01 1.42469168e-01 4.46551651e-01 2.45673180e-01
7.03129232e-01 -1.98389471e-01 4.73885626e-01 -5.61662376e-01
9.93543327e-01 1.62897080e-01 2.49185085e-01 1.12072980e+00
-5.41527033e-01 6.52796447e-01 1.44563675e-01 -2.75076330e-01
-9.27230775e-01 -9.45332468e-01 -4.31305975e-01 1.22549593e+00
3.73873025e-01 -3.25949639e-01 -7.75509000e-01 -7.56934941e-01
-9.91234854e-02 5.56217492e-01 -7.75285363e-01 1.49097219e-01
-5.89008629e-01 -7.88802266e-01 5.82949281e-01 3.28900039e-01
7.15943217e-01 -9.01029170e-01 -3.06672245e-01 -1.76580384e-01
-1.59352675e-01 -1.04622769e+00 -8.53011310e-01 3.05238813e-01
-1.06333613e+00 -7.65396237e-01 -1.09100068e+00 -1.21023405e+00
1.00424349e+00 6.47455752e-01 6.48534715e-01 1.45338967e-01
-7.12469339e-01 3.57360810e-01 -4.12586421e-01 -2.66385257e-01
-3.55947793e-01 -5.18923402e-02 -1.81796506e-01 2.11869210e-01
5.60646296e-01 2.29587123e-01 -4.87481594e-01 3.70498598e-01
-9.23312187e-01 -1.21150658e-01 6.14668906e-01 8.00986469e-01
6.29915416e-01 5.74764311e-01 3.69589329e-01 -9.73041713e-01
9.20234501e-01 -4.89300340e-02 -4.07730848e-01 4.39142734e-01
-6.66727304e-01 1.80780888e-01 6.31706297e-01 -5.17380476e-01
-1.15272284e+00 3.25347781e-01 3.12469870e-01 3.45048010e-02
-3.63928169e-01 3.17779243e-01 8.42576101e-02 -2.88939953e-01
6.31635189e-01 7.01158464e-01 -5.90083539e-01 -3.52499008e-01
1.12738043e-01 9.29329872e-01 3.92763764e-01 -3.90698820e-01
9.24325943e-01 7.10389793e-01 3.82936001e-02 -1.58392453e+00
-5.20697296e-01 -9.96432185e-01 -7.47783720e-01 -3.75204235e-01
6.67653024e-01 -3.60026062e-01 -2.08530411e-01 3.64965975e-01
-7.64002979e-01 2.90488213e-01 2.84912318e-01 2.40926757e-01
-1.68561116e-01 1.13066828e+00 -5.87307274e-01 -1.17529714e+00
-5.53526402e-01 -6.50932848e-01 1.19504654e+00 3.11073273e-01
-2.08786540e-02 -1.18571126e+00 -1.34657204e-01 3.11449200e-01
4.71942201e-02 -2.45264899e-02 7.78754652e-01 -8.62807155e-01
-2.91391969e-01 -6.23579204e-01 -3.27515960e-01 -1.34808766e-02
1.73549011e-01 3.82671684e-01 -8.25790405e-01 -2.42432967e-01
-8.74810573e-03 1.06281862e-01 1.15601063e+00 4.11927313e-01
8.16246867e-01 -2.40490273e-01 -5.98457754e-01 2.89328843e-01
1.56064713e+00 1.93137005e-01 4.88735408e-01 6.00879610e-01
6.24763489e-01 4.80627179e-01 5.88999748e-01 4.90584075e-01
-3.30150634e-01 3.43910873e-01 -3.96165624e-02 -3.26091766e-01
4.99597080e-02 5.18405586e-02 9.67362523e-02 4.34854835e-01
2.25640297e-01 -7.30735242e-01 -9.03295100e-01 6.00239456e-01
-1.88221478e+00 -1.00035250e+00 -7.27101862e-01 2.07650757e+00
5.51561058e-01 4.66373205e-01 -5.33344820e-02 3.95198435e-01
1.45970976e+00 1.04086623e-01 -4.79532123e-01 -4.29064989e-01
-7.13829994e-01 -7.00640008e-02 6.82921886e-01 3.32297653e-01
-1.49281669e+00 1.22612476e+00 6.80985689e+00 1.16465306e+00
-1.16496122e+00 -3.48229021e-01 6.35389507e-01 2.92952865e-01
8.64450410e-02 -1.31653473e-01 -8.73762429e-01 2.22986132e-01
1.58627421e-01 -2.70417750e-01 -2.06268713e-01 7.14913011e-01
3.31089228e-01 -6.07838452e-01 -6.32274628e-01 1.07089996e+00
6.28294885e-01 -1.12185478e+00 4.67278510e-01 -2.54615366e-01
9.33541179e-01 -1.99665174e-01 9.13396105e-02 -2.49466911e-01
-1.10313863e-01 -8.71436000e-01 6.75970376e-01 2.54641116e-01
4.85286623e-01 -7.24863291e-01 5.07428765e-01 3.67825925e-01
-1.28855121e+00 1.84067234e-01 -4.28909421e-01 2.10619256e-01
-2.46201023e-01 6.18781328e-01 -1.20118141e+00 3.05344224e-01
3.78368735e-01 4.85446841e-01 -8.98629904e-01 1.41157687e+00
5.09984791e-02 5.06646097e-01 -4.64167088e-01 -5.27327597e-01
4.17072892e-01 -1.91986114e-01 9.01411593e-01 1.70889127e+00
1.41552225e-01 -1.95245296e-01 2.14263067e-01 5.31394839e-01
8.10041502e-02 8.81234288e-01 -6.57268882e-01 -6.08710432e-03
3.35382298e-02 1.31482852e+00 -1.82886016e+00 -5.10137260e-01
-3.22333366e-01 1.55894256e+00 -1.68272957e-01 5.16417801e-01
-5.02499402e-01 -8.56166780e-01 -1.94766119e-01 5.20069674e-02
3.59368354e-01 -2.82760203e-01 -7.09654868e-01 -1.36396980e+00
1.74772948e-01 -6.00172222e-01 6.74308300e-01 -7.36363828e-01
-1.20869684e+00 6.57414615e-01 -4.06338960e-01 -1.34720850e+00
1.92409456e-01 -8.11211944e-01 -8.28222275e-01 6.83383346e-01
-1.34900808e+00 -9.37419772e-01 -3.52160394e-01 7.90068984e-01
1.17324710e+00 -1.77629977e-01 2.63613731e-01 -8.48978292e-03
-7.10297346e-01 7.18155026e-01 6.15846932e-01 3.59956443e-01
8.80324543e-01 -1.22887552e+00 1.92342222e-01 1.33972609e+00
3.33288014e-01 4.96315926e-01 1.05126858e+00 -1.05934298e+00
-1.04552007e+00 -9.01108563e-01 8.33973944e-01 -1.69329569e-01
6.16725504e-01 -2.82221228e-01 -8.76725614e-01 2.74548709e-01
3.32988858e-01 -3.85866135e-01 2.13649303e-01 -3.77570361e-01
-7.79579803e-02 2.99754769e-01 -1.18978167e+00 9.43002462e-01
3.56548011e-01 -3.37182224e-01 -6.93243802e-01 4.74406779e-01
-1.32010549e-01 5.96666783e-02 6.61869422e-02 1.44621849e-01
3.54377806e-01 -9.07001734e-01 6.16316736e-01 -7.08717778e-02
6.36568367e-02 -3.90057564e-01 4.00045440e-02 -5.88293433e-01
-1.70051441e-01 -7.47689009e-01 2.88580686e-01 1.08108270e+00
5.81095457e-01 -5.23097456e-01 1.36011064e+00 9.77737755e-02
-1.60193238e-02 -3.36476713e-01 -8.44068825e-01 -8.03316951e-01
-4.66166474e-02 -2.01298371e-01 -2.87865847e-01 1.07891107e+00
6.83967248e-02 4.18153971e-01 -3.77960116e-01 -5.77509515e-02
7.97856510e-01 6.27454296e-02 2.68041015e-01 -1.34774184e+00
8.96815285e-02 -7.70036817e-01 -5.24058998e-01 -1.12696934e+00
-1.13011055e-01 -7.49029040e-01 3.53860348e-01 -1.52117968e+00
3.31756592e-01 -8.36847872e-02 -1.65076688e-01 2.05486700e-01
-4.91882682e-01 4.69293773e-01 -1.66164577e-01 3.09731513e-01
-5.02025962e-01 3.23019177e-01 1.10600936e+00 -2.90304661e-01
-2.44837478e-01 1.56753764e-01 -3.16468805e-01 9.96537268e-01
1.05078542e+00 -3.08364064e-01 -2.29810745e-01 -1.10395402e-01
1.21456921e-01 -2.90858895e-01 -9.31812972e-02 -1.02550745e+00
4.97106850e-01 -3.88293535e-01 6.86914444e-01 -1.05762815e+00
1.18945226e-01 -6.70779109e-01 -6.02914929e-01 1.77170426e-01
-3.97407919e-01 1.35412082e-01 2.40892082e-01 8.85385036e-01
-2.86883339e-02 -8.35354447e-01 9.21868205e-01 -1.13394931e-01
-9.25638735e-01 -1.08679689e-01 -8.77964914e-01 -8.88204128e-02
1.08522820e+00 -1.01898432e+00 -1.33852869e-01 -2.31636524e-01
-4.61618930e-01 1.97750796e-02 5.05966961e-01 2.41258785e-01
8.13728988e-01 -6.25273883e-01 -7.29419768e-01 1.41935050e-01
3.05091172e-01 -7.22284436e-01 -7.88585842e-02 7.41227210e-01
-8.59043777e-01 3.86656582e-01 -3.41939479e-02 -6.25626504e-01
-1.81511629e+00 5.63094497e-01 1.89128190e-01 3.16591598e-02
-8.87625575e-01 6.99991941e-01 3.13812822e-01 1.97697952e-01
3.39537501e-01 -1.64631352e-01 -4.32447731e-01 7.11977705e-02
5.58236480e-01 6.05822921e-01 1.38848126e-01 -7.41663337e-01
-3.64472151e-01 7.61073053e-01 -4.30694014e-01 -1.78838149e-01
7.44382739e-01 -4.45968479e-01 -1.21848203e-01 5.17278254e-01
8.72187316e-01 4.01463270e-01 -7.44078279e-01 -2.01424792e-01
5.63906372e-01 -5.40775418e-01 3.08908880e-01 -8.10035169e-01
-6.76237464e-01 7.49479115e-01 7.57559478e-01 4.90060747e-01
1.21831942e+00 -2.77486056e-01 6.00678205e-01 7.12176085e-01
-1.20773748e-01 -1.74788392e+00 9.00601372e-02 5.30185997e-01
5.34239352e-01 -1.37509441e+00 3.24712545e-01 -6.98955119e-01
-5.49736202e-01 1.67647481e+00 3.93523633e-01 -9.77704898e-02
3.31270128e-01 3.77647072e-01 3.41588892e-02 -7.69045949e-02
-5.07147551e-01 -3.70564729e-01 5.07471144e-01 4.68843907e-01
5.07154703e-01 -2.02537492e-01 -6.70918465e-01 -2.66150594e-01
3.42126846e-01 -5.14155030e-01 7.18741834e-01 1.02752566e+00
-1.13587391e+00 -7.90612757e-01 -8.66891205e-01 7.58657157e-01
-8.85882676e-01 -5.39940357e-01 -1.03633583e+00 7.10873842e-01
-2.16243267e-01 9.72300470e-01 -1.15819395e-01 -6.01484925e-02
-7.83933699e-02 1.45432696e-01 1.57208949e-01 -4.41582263e-01
-5.19373715e-01 6.26293004e-01 6.85290843e-02 2.56503612e-01
-1.41666591e-01 -6.78248823e-01 -1.24982727e+00 -1.67099670e-01
-9.34809983e-01 1.12281583e-01 5.87693632e-01 8.79716814e-01
-1.14543080e-01 8.09293799e-03 5.25964022e-01 -4.22165990e-01
-2.77047396e-01 -7.69749224e-01 -7.45770574e-01 2.85637826e-01
4.48297411e-01 -3.21637273e-01 -5.64212203e-01 5.10921776e-01]
|
[11.987567901611328, 2.3229868412017822]
|
f8c37545-a718-4a07-9709-4f29afc1946b
|
selective-frequency-network-for-image
| null | null |
https://openreview.net/forum?id=tyZ1ChGZIKO
|
https://openreview.net/forum?id=tyZ1ChGZIKO
|
Selective Frequency Network for Image Restoration
|
Image restoration aims to reconstruct the latent sharp image from its corrupted counterpart. Besides dealing with this long-standing task in the spatial domain, a few approaches seek solutions in the frequency domain in consideration of the large discrepancy between spectra of sharp/degraded image pairs. However, these works commonly utilize transformation tools, e.g., wavelet transform, to split features into several frequency parts, which is not flexible enough to select the most informative frequency component to recover. In this paper, we exploit a multi-branch and content-aware module to decompose features into separate frequency subbands dynamically and locally, and then accentuate the useful ones via channel-wise attention weights. In addition, to handle large-scale degradation blurs, we propose an extremely simple decoupling and modulation module to enlarge the receptive field via global and window-based average pooling. Integrating two developed modules into a U-Net backbone, the proposed Selective Frequency Network (SFNet) performs favorably against state-of-the-art algorithms on five image restoration tasks, including single-image defocus deblurring, image dehazing, image motion deblurring, image desnowing, and image deraining.
|
['Alois Knoll', 'Kai Huang', 'Xiaochun Cao', 'Xinwei Gao', 'Wenqi Ren', 'Zhenshan Bing', 'Yi Tao', 'Yuning Cui']
|
2023-04-13
| null | null | null |
conference-2023-4
|
['image-dehazing', 'deblurring']
|
['computer-vision', 'computer-vision']
|
[ 6.36953473e-01 -3.91758710e-01 -3.00559644e-02 -2.04203874e-01
-7.44953454e-01 -4.27836150e-01 3.46854866e-01 -2.85387456e-01
-3.43487382e-01 8.07131052e-01 6.19935274e-01 1.63916081e-01
-4.45810944e-01 -6.64242327e-01 -7.32999384e-01 -1.23956370e+00
1.29001737e-01 -5.95116854e-01 2.51898885e-01 -2.05951110e-01
6.15241826e-01 5.15679598e-01 -1.69789505e+00 2.87135333e-01
1.38678479e+00 1.03471076e+00 4.66785610e-01 5.19845128e-01
1.37443200e-01 6.37258828e-01 -5.66486776e-01 4.33520973e-02
1.01447769e-01 -5.35213530e-01 -6.63022697e-01 2.22179487e-01
5.61185122e-01 -4.08143729e-01 -6.67601049e-01 1.51022375e+00
7.28220940e-01 2.08132625e-01 4.57576305e-01 -7.20220804e-01
-1.08394814e+00 4.40228581e-01 -6.24608696e-01 5.87671161e-01
3.16510797e-01 2.70670950e-01 5.96260667e-01 -7.89478183e-01
3.44238162e-01 1.14721942e+00 5.79790235e-01 3.19840729e-01
-1.30851877e+00 -5.13677001e-01 2.88862977e-02 6.95774078e-01
-1.12083125e+00 -7.55093157e-01 1.07367802e+00 -1.68780595e-01
8.67998362e-01 2.71745116e-01 3.35287958e-01 9.40287352e-01
4.90037739e-01 4.13304985e-01 1.24602842e+00 -2.84283489e-01
-1.34988930e-02 -3.59564543e-01 1.57158569e-01 2.09946364e-01
1.35287240e-01 1.95861056e-01 -5.45567572e-01 1.11524709e-01
8.57023478e-01 4.97726090e-02 -1.36729312e+00 -7.64261484e-02
-1.36552417e+00 4.56556559e-01 6.01228952e-01 5.66059768e-01
-4.73035544e-01 1.90335568e-02 1.23678833e-01 4.01518315e-01
6.04953647e-01 5.38894415e-01 -2.90613204e-01 1.83154315e-01
-9.32024896e-01 1.40095413e-01 2.03654811e-01 3.48225713e-01
7.46306539e-01 1.37282163e-01 -3.66843164e-01 9.84512389e-01
1.60964951e-02 2.13162586e-01 8.37549090e-01 -8.28216136e-01
1.65756896e-01 -1.59442294e-02 2.48087555e-01 -1.15400600e+00
-3.09343576e-01 -5.85636616e-01 -1.29135072e+00 1.05496474e-01
2.44568110e-01 1.19946897e-01 -9.05266881e-01 1.82696652e+00
8.12996030e-02 4.86788362e-01 2.03491412e-02 1.32769561e+00
7.02662408e-01 7.47220933e-01 -2.28133395e-01 -6.20785415e-01
1.34612954e+00 -9.89081562e-01 -1.08697271e+00 -2.14804754e-01
-2.76268542e-01 -9.54742491e-01 8.82399082e-01 4.10080284e-01
-1.20567310e+00 -9.27093029e-01 -1.23095047e+00 -3.48869413e-01
-1.18144758e-01 -2.93471459e-02 4.33859289e-01 4.27248001e-01
-1.18470109e+00 8.39002669e-01 -4.44050699e-01 -1.02523685e-01
3.21403146e-01 1.11574374e-01 -3.79315704e-01 -4.52067763e-01
-1.36653662e+00 9.87128437e-01 2.69327819e-01 3.33629251e-01
-7.46479332e-01 -6.10182583e-01 -7.61617422e-01 2.23323300e-01
2.30393916e-01 -8.28916430e-01 7.24417329e-01 -9.27406490e-01
-1.57222879e+00 5.21761715e-01 -1.61352992e-01 -3.90675664e-01
1.51256159e-01 -2.25786865e-01 -7.36544073e-01 4.57708389e-01
-1.52696997e-01 2.98556387e-01 1.65276957e+00 -1.11038530e+00
-3.67566884e-01 -3.16262424e-01 -3.24507281e-02 2.50995427e-01
-3.56500894e-01 -1.29303396e-01 -1.07102990e-01 -1.12159538e+00
3.83718967e-01 -3.16516936e-01 1.08080491e-01 -2.16625571e-01
-1.74687088e-01 1.35245591e-01 7.80627012e-01 -1.10410118e+00
1.30669761e+00 -2.34274507e+00 4.98025417e-01 -2.98271984e-01
2.31230900e-01 2.39697993e-01 -3.28224868e-01 1.33968920e-01
-4.64890867e-01 -2.12895259e-01 -4.29206133e-01 -9.91296675e-03
-2.51775295e-01 -1.07787594e-01 -4.89033282e-01 8.50730956e-01
1.99241951e-01 5.68399966e-01 -7.75384367e-01 -1.15344025e-01
2.66663313e-01 7.49185205e-01 -5.92178464e-01 1.94488406e-01
1.55812994e-01 6.18008196e-01 -1.59686804e-01 5.41112602e-01
1.20486033e+00 -1.56391226e-02 -1.54598445e-01 -8.40512156e-01
-3.47202688e-01 -3.17666270e-02 -9.53688622e-01 1.89729118e+00
-5.00782132e-01 4.68277007e-01 4.29096967e-01 -1.17412746e+00
7.60161340e-01 3.32153946e-01 3.77197504e-01 -5.57557285e-01
1.75026935e-02 4.28189665e-01 -3.70241068e-02 -8.01653862e-01
5.13143599e-01 -2.97006190e-01 4.09384191e-01 9.20705721e-02
2.33429492e-01 -1.60818473e-01 -1.30623281e-01 -4.18404788e-01
1.01703560e+00 1.05000302e-01 2.19599605e-01 -3.19433630e-01
8.23853195e-01 -5.76691747e-01 3.46876353e-01 6.05352104e-01
-2.91714638e-01 1.18144917e+00 1.24989539e-01 -3.13202232e-01
-7.49642193e-01 -1.03392482e+00 -3.75510961e-01 8.61977994e-01
5.58437884e-01 7.56397396e-02 -8.78897488e-01 -3.55202377e-01
-1.95172474e-01 4.83572036e-01 -3.63732845e-01 -5.75132132e-01
-5.86926520e-01 -1.07351601e+00 1.50460199e-01 -6.53418377e-02
9.51635480e-01 -1.07534873e+00 -4.81872916e-01 5.01150966e-01
-7.32941806e-01 -7.99853683e-01 -9.19188559e-01 2.05497801e-01
-6.77390933e-01 -9.17365193e-01 -1.15970325e+00 -9.84649599e-01
4.55877274e-01 8.25929880e-01 7.61267006e-01 -2.42937326e-01
-3.34837556e-01 1.12742223e-01 -4.37037468e-01 3.66935492e-01
-6.63229972e-02 -1.19704664e-01 -7.44345710e-02 4.60918307e-01
-5.64027429e-02 -1.06851399e+00 -9.62250412e-01 3.24256688e-01
-1.31809354e+00 -1.50972396e-01 7.75319815e-01 1.07446492e+00
3.87186527e-01 5.37506759e-01 7.29412019e-01 -2.14142159e-01
8.74940038e-01 -3.06946635e-01 -1.73654541e-01 1.99540988e-01
-3.63155514e-01 1.88991930e-02 7.58513987e-01 -7.06446111e-01
-1.23047745e+00 -3.33695322e-01 -1.57519072e-01 -5.47518134e-01
-1.52588740e-01 3.74823034e-01 -3.37898016e-01 -3.96945477e-01
7.04505026e-01 8.06846201e-01 1.03173584e-01 -6.11849725e-01
3.54335219e-01 7.53583431e-01 1.03832459e+00 -2.25903809e-01
8.84870470e-01 3.93366516e-01 -2.77061433e-01 -6.84226036e-01
-7.40090787e-01 -2.49904841e-01 -3.28693449e-01 -1.93368942e-01
8.96118820e-01 -9.98032093e-01 -5.64599931e-01 9.61701691e-01
-1.21530151e+00 1.16873331e-01 -2.09884733e-01 4.63106275e-01
-4.71911788e-01 6.52274966e-01 -7.23538220e-01 -2.96255141e-01
-2.51537949e-01 -1.19224322e+00 8.69937301e-01 4.55295622e-01
2.99716175e-01 -4.99755710e-01 -1.65964603e-01 2.62712419e-01
9.82993484e-01 8.74178410e-02 9.68902707e-01 4.21915427e-02
-5.93638182e-01 3.59242670e-02 -3.53040040e-01 6.97134376e-01
5.17129838e-01 -5.32882929e-01 -1.19272614e+00 -6.12022698e-01
6.02418244e-01 -1.47893921e-01 1.42372632e+00 7.72821963e-01
1.35099602e+00 -4.18780565e-01 -2.66064610e-02 1.00555980e+00
1.23995721e+00 1.41508207e-01 1.13660741e+00 3.17226499e-01
5.04596233e-01 5.07868886e-01 2.75865883e-01 2.39650786e-01
2.22518276e-02 6.75462425e-01 4.29342151e-01 -1.71405390e-01
-5.58257163e-01 1.14401646e-01 3.87854934e-01 7.49125361e-01
-1.31033808e-01 -1.99898764e-01 -3.17243576e-01 4.59474355e-01
-1.60469961e+00 -1.08785391e+00 3.46433967e-01 2.06517148e+00
1.07390702e+00 -3.93634364e-02 -4.49642360e-01 1.16065793e-01
1.01279140e+00 6.19742215e-01 -6.37562633e-01 1.32443786e-01
-5.19344330e-01 2.05052629e-01 3.68219286e-01 6.08787000e-01
-1.23747993e+00 6.17244303e-01 5.39592028e+00 1.22210813e+00
-1.28656995e+00 1.45837337e-01 6.84600830e-01 6.26257136e-02
-2.45737821e-01 -1.49865717e-01 -1.87468380e-01 6.35712028e-01
5.72358072e-01 -2.06732333e-01 9.05392706e-01 3.34447652e-01
2.64812648e-01 -1.20238885e-01 -6.96773291e-01 1.17552662e+00
1.74717098e-01 -1.18477011e+00 -1.16214186e-01 -1.84341371e-01
6.36233389e-01 -3.00812095e-01 2.81614870e-01 -8.77906084e-02
-3.56360823e-01 -1.09179056e+00 5.82885623e-01 7.56911159e-01
8.58165443e-01 -7.77950048e-01 7.01683104e-01 1.43842563e-01
-1.00595391e+00 -1.49807498e-01 -4.99590248e-01 1.85524702e-01
1.28482971e-02 9.21704710e-01 4.65330668e-02 8.25995564e-01
9.80020583e-01 9.15520251e-01 -2.95140475e-01 1.22770441e+00
-6.47722855e-02 3.78564328e-01 1.38003841e-01 7.01052904e-01
-2.03068823e-01 -3.22214067e-01 7.88043618e-01 1.00904703e+00
5.18781185e-01 1.63920909e-01 -1.94620430e-01 7.82776117e-01
-1.01668108e-02 -3.79898578e-01 -3.13316494e-01 2.23657385e-01
1.91412807e-01 1.28429449e+00 -4.63720828e-01 -8.11843872e-02
-2.06901580e-01 1.30234218e+00 -1.03575207e-01 6.67859614e-01
-7.82899320e-01 -7.81092107e-01 8.06940734e-01 -1.31115541e-02
3.66346598e-01 -1.17520757e-01 2.04970948e-02 -1.39837623e+00
1.63521007e-01 -9.89144981e-01 -2.91079450e-02 -9.59103346e-01
-1.42806053e+00 7.56536841e-01 -2.62852579e-01 -1.52571559e+00
1.07116513e-01 -2.11009994e-01 -5.14814019e-01 1.17617631e+00
-2.19748569e+00 -7.96624124e-01 -5.25235772e-01 7.78022349e-01
5.43921232e-01 9.16411057e-02 4.80413109e-01 5.36113262e-01
-4.09264088e-01 2.39426538e-01 2.27334619e-01 -3.30812186e-01
1.15521407e+00 -8.98223221e-01 6.05654009e-02 1.08518124e+00
-4.98939037e-01 6.97400749e-01 8.44111562e-01 -4.49135423e-01
-1.35779667e+00 -1.02136803e+00 6.12523258e-01 3.37160677e-01
4.84747618e-01 -5.93229644e-02 -1.39351809e+00 1.47494048e-01
3.75646055e-01 3.80190969e-01 6.32751361e-02 -4.87678140e-01
-3.44323009e-01 -3.85428280e-01 -1.24086022e+00 4.40393180e-01
9.70594347e-01 -6.24218047e-01 -6.46468937e-01 2.83670604e-01
9.40694273e-01 -4.04693902e-01 -7.64869273e-01 4.57495034e-01
2.81443417e-01 -1.28752625e+00 1.30618262e+00 -8.12001973e-02
6.24485493e-01 -6.40109539e-01 -1.37428597e-01 -1.62382007e+00
-7.08593786e-01 -9.40732837e-01 -2.16091471e-03 1.17404759e+00
-1.50687441e-01 -9.09687877e-01 1.92404687e-01 -1.45997852e-01
-4.43137020e-01 -3.44047844e-01 -1.07121003e+00 -4.95377779e-01
-2.47138724e-01 7.71896020e-02 5.05949080e-01 9.72276986e-01
-2.24953175e-01 4.01960127e-02 -7.37141550e-01 3.66778105e-01
6.81827724e-01 2.95104355e-01 1.13198958e-01 -8.40543032e-01
-3.95614028e-01 -5.20425677e-01 -2.07973629e-01 -1.21631432e+00
1.60804585e-01 -6.35370433e-01 2.86862195e-01 -1.33993411e+00
9.21926573e-02 2.08159328e-01 -5.02650142e-01 2.24361613e-01
-3.49361628e-01 3.76502961e-01 2.01726006e-03 4.54083383e-01
-1.37606561e-01 8.87046218e-01 1.64124739e+00 -2.52923250e-01
-6.56870455e-02 -5.90218566e-02 -8.47027957e-01 5.79439044e-01
5.68465948e-01 -2.09270492e-01 -3.29929233e-01 -5.33842325e-01
-2.56435007e-01 4.08933997e-01 5.11205256e-01 -1.13472438e+00
3.07064325e-01 -1.31197914e-01 5.02412319e-01 -2.37232313e-01
2.61661947e-01 -6.65543199e-01 7.63903186e-02 3.30752939e-01
-2.64802843e-01 -2.49571294e-01 9.17373449e-02 7.86518395e-01
-7.51287758e-01 -1.44190386e-01 1.25188601e+00 -1.06469311e-01
-6.59938097e-01 2.09829271e-01 -3.22077572e-01 -2.43894264e-01
5.93974531e-01 -2.05060080e-01 -7.72393823e-01 -3.76037657e-01
-5.92965126e-01 -2.06385985e-01 3.05347711e-01 4.05659050e-01
8.34538221e-01 -1.12081242e+00 -7.77025938e-01 4.05502230e-01
-2.70058900e-01 -2.63759047e-01 9.88318622e-01 9.68255639e-01
-2.60873914e-01 1.85201302e-01 -5.83492875e-01 -3.32851112e-01
-8.87391627e-01 6.56722963e-01 5.57854474e-01 -3.45701575e-02
-7.82950461e-01 9.26585495e-01 4.17654723e-01 1.39717013e-01
1.45283923e-01 -4.16190743e-01 -4.79903609e-01 2.06122518e-01
8.05903077e-01 2.71301210e-01 1.99086234e-01 -5.99735439e-01
-2.12120801e-01 9.38111424e-01 1.91177633e-02 1.90489843e-01
1.43813312e+00 -4.81811851e-01 -4.88302499e-01 -2.25968063e-02
1.35471380e+00 4.51462939e-02 -1.53954148e+00 -3.73220205e-01
-3.69958878e-01 -6.39386177e-01 5.13452172e-01 -7.44557977e-01
-1.26779044e+00 7.10342765e-01 8.10486555e-01 4.33030784e-01
1.97812188e+00 -2.78763890e-01 7.20677793e-01 -1.10072307e-01
1.17676049e-01 -7.19609201e-01 1.47274241e-01 2.56491482e-01
1.35188568e+00 -9.27839160e-01 2.63119712e-02 -4.66074258e-01
-9.24910232e-02 1.34487522e+00 3.14546078e-01 -1.44403696e-01
5.76151609e-01 4.09060577e-03 -1.70828462e-01 2.39070415e-01
-4.05609280e-01 -5.36866635e-02 3.75985622e-01 5.88421226e-01
1.88663781e-01 -3.06742787e-01 -2.31553286e-01 5.63944519e-01
2.47431379e-02 6.90158978e-02 6.74697459e-01 5.00976145e-01
-7.44810998e-01 -7.18632817e-01 -7.09115028e-01 1.69217885e-01
-5.51226914e-01 -3.53316069e-01 3.07984918e-01 3.93042326e-01
2.28784353e-01 1.14847958e+00 -7.53253475e-02 -3.06381643e-01
3.00542146e-01 -1.84477583e-01 6.13297462e-01 -1.53985336e-01
-5.18488944e-01 5.28261483e-01 -3.63787264e-01 -5.87737620e-01
-7.01516986e-01 -4.08907384e-01 -6.94484413e-01 -1.20440617e-01
-3.63853753e-01 -1.84533343e-01 3.31169099e-01 6.18656516e-01
2.96544254e-01 8.75703037e-01 7.49050915e-01 -1.32874620e+00
-6.13609016e-01 -1.12203825e+00 -9.01787996e-01 4.54518795e-01
9.92785931e-01 -4.95169640e-01 -8.14904332e-01 8.52314681e-02]
|
[11.312270164489746, -2.336535930633545]
|
69ff2f73-4218-4621-b053-aab7c02ba2d4
|
dave-a-unified-framework-for-fast-vehicle
|
1607.04564
| null |
http://arxiv.org/abs/1607.04564v3
|
http://arxiv.org/pdf/1607.04564v3.pdf
|
DAVE: A Unified Framework for Fast Vehicle Detection and Annotation
|
Vehicle detection and annotation for streaming video data with complex scenes
is an interesting but challenging task for urban traffic surveillance. In this
paper, we present a fast framework of Detection and Annotation for Vehicles
(DAVE), which effectively combines vehicle detection and attributes annotation.
DAVE consists of two convolutional neural networks (CNNs): a fast vehicle
proposal network (FVPN) for vehicle-like objects extraction and an attributes
learning network (ALN) aiming to verify each proposal and infer each vehicle's
pose, color and type simultaneously. These two nets are jointly optimized so
that abundant latent knowledge learned from the ALN can be exploited to guide
FVPN training. Once the system is trained, it can achieve efficient vehicle
detection and annotation for real-world traffic surveillance data. We evaluate
DAVE on a new self-collected UTS dataset and the public PASCAL VOC2007 car and
LISA 2010 datasets, with consistent improvements over existing algorithms.
|
['Matt Mellor', 'Li Liu', 'Yi Zhou', 'Ling Shao']
|
2016-07-15
| null | null | null | null |
['fast-vehicle-detection']
|
['computer-vision']
|
[ 6.61923885e-02 -8.23396668e-02 -2.00886056e-01 -6.89117134e-01
-8.25339317e-01 -4.78785306e-01 7.51429319e-01 -1.86844930e-01
-5.21039546e-01 3.95096540e-01 -2.48375684e-01 -3.54691923e-01
6.40541732e-01 -6.64861262e-01 -1.04476547e+00 -7.52731025e-01
-2.09070399e-01 5.55500805e-01 9.40566957e-01 8.40748623e-02
-3.56009662e-01 7.65453994e-01 -1.96513653e+00 2.47396186e-01
4.74851012e-01 1.40740418e+00 2.58029222e-01 9.51498568e-01
1.53053001e-01 1.18457806e+00 -2.19260722e-01 -5.80452859e-01
3.70585918e-01 4.90527093e-01 -4.94185865e-01 1.83093056e-01
1.04302335e+00 -9.06203270e-01 -6.29106402e-01 1.01278222e+00
4.71464247e-02 -2.46477872e-02 6.59731507e-01 -1.70262897e+00
1.29108280e-01 2.43525133e-01 -2.98590809e-01 2.67362684e-01
-2.79512137e-01 5.85776389e-01 7.68041253e-01 -8.34173203e-01
6.55806303e-01 1.34557998e+00 7.54281759e-01 5.84666312e-01
-8.30195844e-01 -9.13524032e-01 1.43511176e-01 6.55365944e-01
-1.42084002e+00 -8.26849699e-01 7.31397033e-01 -7.29415178e-01
8.06397617e-01 1.40803844e-01 6.85058296e-01 1.16931462e+00
-2.89301753e-01 1.33436692e+00 2.58745283e-01 3.16705108e-01
1.93352196e-02 3.54770660e-01 1.47658914e-01 9.09151077e-01
3.46941590e-01 4.11618322e-01 -1.24948844e-01 -1.62129030e-02
2.15096787e-01 1.89723238e-01 2.23862290e-01 -6.07937694e-01
-1.02460372e+00 7.91210711e-01 2.97502995e-01 -4.46699172e-01
-5.27731001e-01 4.38666254e-01 6.48158073e-01 -7.28702918e-02
3.92252803e-01 -1.66754097e-01 -6.26430392e-01 1.37182131e-01
-8.80153477e-01 3.70515227e-01 4.17433500e-01 1.17497420e+00
1.03028512e+00 1.82952642e-01 -4.47654039e-01 6.59735680e-01
5.18203497e-01 9.75662231e-01 -4.92747724e-01 -1.18345678e+00
4.44926262e-01 5.68574786e-01 9.19168591e-02 -8.55242848e-01
-2.25341931e-01 -2.00101748e-01 -6.07201934e-01 1.96296573e-01
2.81676710e-01 -2.81204671e-01 -9.58070457e-01 1.55613852e+00
7.26176798e-01 4.55428600e-01 1.48260947e-02 7.34081209e-01
1.16875076e+00 8.67792547e-01 4.77089465e-01 6.42736405e-02
1.36202419e+00 -1.04382312e+00 -5.34650743e-01 -4.18066800e-01
7.29614198e-01 -4.80599374e-01 1.99052155e-01 -1.37912910e-02
-6.91576421e-01 -8.24081719e-01 -7.63204157e-01 -3.34044322e-02
-5.28300047e-01 6.32854342e-01 5.63605368e-01 4.91796225e-01
-9.61897373e-01 -1.09560214e-01 -7.99209118e-01 -2.76419789e-01
1.03409362e+00 4.22436655e-01 -3.82129431e-01 -1.98768839e-01
-1.10883856e+00 5.74364960e-01 5.77190578e-01 3.51348609e-01
-1.73393357e+00 -6.75938845e-01 -1.25099790e+00 -5.39655499e-02
7.71676779e-01 -5.05270481e-01 1.27135444e+00 -7.52409220e-01
-1.10126436e+00 9.79989707e-01 -1.49811774e-01 -8.32871139e-01
6.03330970e-01 -2.05269143e-01 -5.38302600e-01 1.88818172e-01
2.44301185e-01 1.35426545e+00 9.99392688e-01 -1.25350070e+00
-1.30456865e+00 4.67035808e-02 -8.66585076e-02 -3.37800831e-01
-2.05242753e-01 2.85104185e-01 -9.69387174e-01 -2.29789898e-01
-5.30524731e-01 -9.90472674e-01 -1.52467623e-01 4.27860618e-01
-5.22177756e-01 -6.82510793e-01 1.49805486e+00 -6.75497174e-01
7.21890271e-01 -2.12465715e+00 -3.93203020e-01 6.41242564e-02
4.78786200e-01 9.58788931e-01 -4.51295882e-01 -1.41967759e-01
1.86871335e-01 -4.77052659e-01 5.05608507e-02 -4.50704306e-01
-3.98487039e-03 3.97473067e-01 -2.29105696e-01 5.86841524e-01
5.97835422e-01 1.06919241e+00 -8.10816944e-01 -8.31768155e-01
5.65082490e-01 5.74127734e-01 -5.45244753e-01 4.26873893e-01
-4.92373109e-01 1.43035203e-01 -4.40195590e-01 1.03257990e+00
9.21196699e-01 1.18104786e-01 -2.79173046e-01 -4.64493036e-01
-3.18831682e-01 -1.10873573e-01 -9.75689054e-01 7.35466897e-01
2.24119034e-02 1.20933723e+00 3.45460057e-01 -9.78393793e-01
8.62852931e-01 1.94023490e-01 6.02425516e-01 -5.85255086e-01
3.18855882e-01 -6.14694878e-02 -3.81913751e-01 -9.10278082e-01
5.52954495e-01 6.50239944e-01 1.13040082e-01 -1.65640935e-01
1.70951322e-01 3.47386420e-01 5.53403080e-01 3.86224598e-01
1.03188872e+00 -1.06557362e-01 -7.40562081e-02 -2.09738333e-02
9.13433671e-01 1.31455809e-01 9.40018356e-01 6.81059003e-01
-6.19808257e-01 2.63231620e-02 5.07278502e-01 -1.04979801e+00
-1.28910685e+00 -1.01691651e+00 -1.24646559e-01 1.17262852e+00
1.27927050e-01 -2.74146110e-01 -5.75869501e-01 -1.05116951e+00
3.62049788e-02 4.56817240e-01 -6.42065406e-01 3.83747704e-02
-7.58263767e-01 -2.98113614e-01 5.71338713e-01 8.04050267e-01
6.36354685e-01 -1.04054236e+00 -5.82724988e-01 1.16178736e-01
-1.01696804e-01 -1.91120219e+00 -4.08322781e-01 -2.34500900e-01
-9.98093486e-02 -1.37451434e+00 -2.64752835e-01 -7.69592106e-01
5.37183583e-01 3.19567710e-01 1.05896866e+00 1.79147422e-01
-3.90861750e-01 3.13464105e-01 -1.63113803e-01 -6.83767140e-01
-5.90280056e-01 -5.40654026e-02 3.56655866e-02 4.66032803e-01
6.11410141e-01 1.08679444e-01 -5.60911417e-01 5.13446808e-01
-5.12493730e-01 1.56261832e-01 6.96540117e-01 3.38906407e-01
6.00162446e-01 -9.38900560e-02 2.63326347e-01 -7.27653086e-01
-3.43943536e-01 -4.62272853e-01 -1.29820013e+00 1.45028653e-02
-4.11890745e-02 -3.48132968e-01 3.91457856e-01 -2.01265737e-01
-9.89360750e-01 7.35609651e-01 -5.04685163e-01 -5.79215229e-01
-6.45459235e-01 -3.24449599e-01 -3.76535624e-01 -9.58416685e-02
1.34200335e-01 2.00412229e-01 -1.27299473e-01 -1.34432048e-01
2.36634910e-01 6.73595667e-01 9.47525024e-01 -1.39077827e-01
9.40021813e-01 5.98917127e-01 -2.02006940e-02 -1.00235927e+00
-9.59149957e-01 -8.30871582e-01 -6.88705921e-01 -6.50422037e-01
1.28915441e+00 -1.30972064e+00 -1.01875198e+00 5.81810594e-01
-1.53992403e+00 -3.27128053e-01 -5.64508438e-02 5.49575210e-01
-4.70560372e-01 1.61532573e-02 -4.85510409e-01 -1.01267374e+00
-1.16234988e-01 -1.34437275e+00 1.55548680e+00 8.16467032e-02
4.22897995e-01 -6.71768844e-01 -1.60339609e-01 4.51744437e-01
2.76693076e-01 2.85328686e-01 2.36920163e-01 -6.69973850e-01
-1.14583576e+00 -5.39011240e-01 -7.07619667e-01 5.59899151e-01
-5.42781174e-01 4.27491367e-01 -1.03282619e+00 -1.76245540e-01
-6.04709327e-01 -2.89193034e-01 1.41168892e+00 4.07135725e-01
1.17997003e+00 -3.67767006e-01 -6.49495006e-01 1.00468016e+00
1.13641632e+00 2.13787466e-01 5.98873019e-01 1.46033674e-01
1.06285131e+00 6.77853227e-01 6.86198175e-01 1.33911088e-01
5.88933945e-01 6.83380365e-01 8.15763712e-01 -1.97364807e-01
-3.17720294e-01 -1.28199803e-02 6.07092857e-01 2.45209426e-01
5.03365025e-02 -2.30876625e-01 -9.00356591e-01 7.17687845e-01
-1.94856715e+00 -1.22962201e+00 -4.18018073e-01 1.60021460e+00
1.67925909e-01 1.71950012e-01 3.75762850e-01 -2.54737765e-01
8.27454090e-01 2.51118630e-01 -4.66280997e-01 9.00534391e-02
-1.32689029e-01 -6.66201890e-01 9.07301247e-01 2.80332774e-01
-1.77219725e+00 1.10399234e+00 5.80016708e+00 8.40045631e-01
-8.60878468e-01 1.30009606e-01 6.68978095e-01 2.15476707e-01
1.02296650e-01 -3.81852150e-01 -1.41468358e+00 4.30057257e-01
8.38131070e-01 4.55306768e-01 -2.00768169e-02 1.20708454e+00
1.44133434e-01 1.01244614e-01 -1.01169538e+00 8.81418943e-01
1.94575936e-02 -1.56532240e+00 3.34933214e-02 -3.83406654e-02
5.73347211e-01 7.08001077e-01 -1.12476572e-02 5.55618346e-01
4.18415070e-01 -5.98877132e-01 7.62163162e-01 4.58684415e-01
6.94110274e-01 -8.08146238e-01 1.04026854e+00 2.30291352e-01
-1.59182262e+00 -2.24090815e-01 -3.46786052e-01 5.40882707e-01
3.72642368e-01 1.91698998e-01 -9.74000931e-01 6.37733191e-02
9.24423873e-01 9.58191156e-01 -7.70627558e-01 1.14001310e+00
-1.35367960e-01 9.29646313e-01 -3.48149031e-01 1.30886212e-01
5.38124382e-01 2.02353179e-01 8.01861584e-01 1.48652935e+00
9.90618840e-02 -8.42840970e-02 5.22179604e-01 7.94603288e-01
-1.49054721e-01 -3.53923798e-01 -5.83044231e-01 2.33898550e-01
4.64286834e-01 1.73244047e+00 -6.26890421e-01 -7.44582593e-01
-5.15491188e-01 3.93565118e-01 1.89885557e-01 4.19594765e-01
-1.29224503e+00 -1.51502155e-02 1.02026510e+00 2.24593356e-01
8.68062496e-01 -1.74499899e-01 5.24544060e-01 -8.21132481e-01
-2.05174372e-01 -4.21055496e-01 3.51937622e-01 -6.25428855e-01
-1.07933331e+00 6.59629345e-01 1.26668304e-01 -1.31384945e+00
-6.90403804e-02 -9.80475307e-01 -6.69273078e-01 1.66210517e-01
-1.92561626e+00 -1.61676168e+00 -4.15243685e-01 7.25395024e-01
6.63488507e-01 -4.34414804e-01 1.52276769e-01 7.86498725e-01
-8.91277909e-01 5.32485902e-01 -1.57242477e-01 6.27531588e-01
3.34109038e-01 -8.94279242e-01 4.56398010e-01 9.14556801e-01
6.23264872e-02 -1.96981937e-01 6.21919036e-01 -6.71514273e-01
-1.39943457e+00 -2.09206629e+00 7.20792115e-01 -6.62284195e-01
6.28538907e-01 -5.58608830e-01 -6.81865394e-01 8.30135643e-01
-1.54823055e-02 6.10696256e-01 6.52553961e-02 -3.30214679e-01
-3.47348332e-01 -4.98996049e-01 -8.62195134e-01 1.88660592e-01
8.92902911e-01 -3.34242016e-01 -8.61553475e-02 5.76771677e-01
8.17269027e-01 -2.74862438e-01 -3.83781582e-01 5.89753389e-01
4.33793068e-01 -7.38682449e-01 1.21019351e+00 -5.50570846e-01
-3.80110890e-02 -6.47548497e-01 -2.66582042e-01 -7.27751553e-01
-1.52251497e-01 -3.66326571e-01 -3.22912514e-01 1.39919150e+00
2.40298986e-01 -3.27407867e-01 7.17036307e-01 3.33991408e-01
-4.54964757e-01 -4.43215698e-01 -8.41130793e-01 -7.27365255e-01
-6.67715847e-01 -9.33243275e-01 6.79728925e-01 4.74624306e-01
-1.00748348e+00 3.95577401e-01 -7.35929728e-01 5.92848480e-01
1.12223732e+00 -2.75735974e-01 1.24475944e+00 -1.33748233e+00
2.73141176e-01 -2.85248369e-01 -9.01281595e-01 -1.04768264e+00
4.68734354e-01 -6.36973739e-01 3.29834372e-01 -1.19488192e+00
4.33533609e-01 -2.85449952e-01 1.82199907e-02 4.67719942e-01
-1.05839334e-01 5.38330078e-01 1.81609660e-01 -5.55886468e-03
-1.46262932e+00 5.80257833e-01 7.93028474e-01 -5.33854187e-01
2.20951274e-01 2.69820273e-01 2.47320328e-02 1.00511158e+00
4.87734169e-01 -5.94857812e-01 -1.33327901e-01 -3.12308103e-01
-2.09125236e-01 -1.85815003e-02 1.17317355e+00 -1.05745757e+00
5.35961807e-01 -3.02933045e-02 4.08139527e-01 -1.50003970e+00
3.35341662e-01 -1.03332996e+00 -4.64238971e-02 4.87165898e-01
-1.36781275e-01 -1.33376777e-01 2.13180259e-01 7.25476861e-01
-1.68056250e-01 3.22125375e-01 7.66077220e-01 3.32554728e-01
-1.46470749e+00 1.10511279e+00 -4.99972016e-01 -1.27183571e-01
1.47299552e+00 -9.25457776e-02 -3.34116966e-01 -1.11920826e-01
-3.62459213e-01 8.48192453e-01 -5.35454368e-03 7.09887564e-01
6.26586676e-01 -1.47208846e+00 -1.02891386e+00 4.62127298e-01
5.09101570e-01 3.13320830e-02 4.75295633e-01 7.15144575e-01
-4.98291910e-01 5.29085040e-01 -4.74420339e-02 -1.22235131e+00
-1.54108560e+00 8.76456439e-01 2.38722473e-01 1.00296862e-01
-6.32782638e-01 8.30735207e-01 4.59757686e-01 -3.83469284e-01
4.67110038e-01 -3.92719284e-02 -4.82452005e-01 6.78474605e-02
8.33436012e-01 3.29975843e-01 -2.35632002e-01 -1.31886876e+00
-5.83322763e-01 3.57849926e-01 -1.11398518e-01 5.25533020e-01
1.29630864e+00 -1.77734941e-01 1.05984926e-01 -1.68458879e-01
1.34988141e+00 -4.58979368e-01 -1.87479734e+00 -4.16742861e-01
-1.90574199e-01 -3.32023621e-01 6.12631664e-02 -1.46656409e-01
-1.51928103e+00 8.64083529e-01 6.08332396e-01 4.18750495e-02
7.52957165e-01 3.57139349e-01 1.11159396e+00 7.25561857e-01
-3.07316650e-02 -1.21881413e+00 -1.16076700e-01 5.50913930e-01
5.06753206e-01 -1.60786951e+00 -3.26191664e-01 -4.93054330e-01
-6.60722256e-01 9.49514985e-01 9.82035279e-01 3.78877893e-02
6.00301266e-01 3.37959439e-01 6.55479431e-02 -3.54795903e-01
-8.73770118e-01 -6.32875204e-01 5.29120326e-01 7.49283969e-01
-3.65908325e-01 2.40038969e-02 4.69474256e-01 2.20132321e-01
2.01381490e-01 -1.77824825e-01 3.56196649e-02 3.65145713e-01
-6.78959787e-01 -4.56475228e-01 -3.23462009e-01 4.63386208e-01
-1.04269050e-01 1.87052295e-01 -1.03803568e-01 8.15466940e-01
7.51759231e-01 8.58998656e-01 4.25923169e-01 -3.84347826e-01
2.07943261e-01 -3.00819129e-01 -1.08265266e-01 -2.19874948e-01
-1.79733261e-01 -7.97143430e-02 3.77800703e-01 -8.75101328e-01
-5.60012817e-01 -9.10733759e-01 -9.07333195e-01 -3.13652754e-01
-2.53089815e-01 -1.21289948e-02 6.05547667e-01 1.04253781e+00
9.46460590e-02 5.38249075e-01 6.57880127e-01 -9.88099337e-01
-5.62490523e-02 -5.61512947e-01 -3.05477768e-01 4.49515343e-01
7.80127943e-01 -7.21711338e-01 1.98812205e-02 3.65112990e-01]
|
[8.089118957519531, -1.4612313508987427]
|
0bd88a5e-8b7e-49ad-bb02-9cff24af2037
|
dual-semantic-knowledge-composed-multimodal
|
2305.09990
| null |
https://arxiv.org/abs/2305.09990v1
|
https://arxiv.org/pdf/2305.09990v1.pdf
|
Dual Semantic Knowledge Composed Multimodal Dialog Systems
|
Textual response generation is an essential task for multimodal task-oriented dialog systems.Although existing studies have achieved fruitful progress, they still suffer from two critical limitations: 1) focusing on the attribute knowledge but ignoring the relation knowledge that can reveal the correlations between different entities and hence promote the response generation}, and 2) only conducting the cross-entropy loss based output-level supervision but lacking the representation-level regularization. To address these limitations, we devise a novel multimodal task-oriented dialog system (named MDS-S2). Specifically, MDS-S2 first simultaneously acquires the context related attribute and relation knowledge from the knowledge base, whereby the non-intuitive relation knowledge is extracted by the n-hop graph walk. Thereafter, considering that the attribute knowledge and relation knowledge can benefit the responding to different levels of questions, we design a multi-level knowledge composition module in MDS-S2 to obtain the latent composed response representation. Moreover, we devise a set of latent query variables to distill the semantic information from the composed response representation and the ground truth response representation, respectively, and thus conduct the representation-level semantic regularization. Extensive experiments on a public dataset have verified the superiority of our proposed MDS-S2. We have released the codes and parameters to facilitate the research community.
|
['Tat-Seng Chua', 'Liqiang Nie', 'Yinwei Wei', 'Xuemeng Song', 'Xiaolin Chen']
|
2023-05-17
| null | null | null | null |
['response-generation']
|
['natural-language-processing']
|
[ 2.35150024e-01 2.69528240e-01 -1.94970414e-01 -4.91583228e-01
-7.69055367e-01 -1.86245084e-01 5.22125006e-01 -1.04802661e-02
-3.73144746e-01 7.51715302e-01 5.58020353e-01 -8.97831097e-02
-3.25113297e-01 -8.46257865e-01 -1.74548715e-01 -6.18029356e-01
5.14332831e-01 6.03914618e-01 2.33653247e-01 -5.92864215e-01
-2.08816938e-02 -2.31830001e-01 -9.09732342e-01 4.71334338e-01
1.34763026e+00 1.10459566e+00 2.09414527e-01 8.10426846e-02
-3.94459248e-01 9.03247118e-01 -3.94999802e-01 -7.28163302e-01
-2.72551954e-01 -6.81701779e-01 -1.08751345e+00 1.54991448e-01
-3.36590350e-01 -2.37823784e-01 -4.75095242e-01 1.03327847e+00
5.43364406e-01 4.77473587e-01 5.75477540e-01 -1.15370059e+00
-9.52479780e-01 7.37018704e-01 -3.40789467e-01 -2.12605000e-01
6.53319299e-01 2.24520490e-01 1.20087659e+00 -9.36391115e-01
5.84084809e-01 1.72459483e+00 1.93809167e-01 7.61733115e-01
-1.26079786e+00 -6.21816695e-01 3.54927570e-01 2.10690051e-01
-1.45614755e+00 -4.37633991e-01 1.14435756e+00 -1.62713736e-01
6.20652795e-01 3.31016809e-01 3.02622952e-02 1.23056138e+00
-5.50380588e-01 1.06346846e+00 1.07289076e+00 -1.99146196e-01
-7.54475668e-02 5.38328290e-01 2.71676064e-01 7.45020032e-01
-3.73524934e-01 -2.41404206e-01 -4.73975509e-01 -2.43880346e-01
3.97770137e-01 -1.20942913e-01 -5.02460659e-01 -6.53202236e-02
-1.28457785e+00 9.97597039e-01 5.50538182e-01 2.41611362e-01
-2.93869823e-01 -3.48444968e-01 4.44344163e-01 3.21427494e-01
2.82838255e-01 7.77474493e-02 -3.65818530e-01 2.46957943e-01
-2.46331871e-01 9.55627933e-02 8.34092617e-01 1.23490262e+00
9.75771248e-01 -2.81756699e-01 -5.69931507e-01 1.03467870e+00
6.46533370e-01 3.47166508e-01 5.13719022e-01 -7.14225471e-01
1.06709540e+00 1.27900708e+00 6.17969260e-02 -1.51286340e+00
-3.28585386e-01 -8.72068182e-02 -1.09216547e+00 -7.62583733e-01
1.24429099e-01 -4.72037375e-01 -3.25690359e-01 2.00483799e+00
4.25599426e-01 -2.95961857e-01 4.18183833e-01 1.14012468e+00
1.33524811e+00 5.78112364e-01 3.14879596e-01 -3.59518886e-01
1.56301689e+00 -9.74772811e-01 -1.03092074e+00 -2.21326128e-01
7.52461672e-01 -5.77486992e-01 1.46395767e+00 -1.43130049e-01
-7.66203761e-01 -4.80815589e-01 -8.22792232e-01 -2.33013660e-01
-2.18338653e-01 4.33536679e-01 5.35386682e-01 2.86249459e-01
-6.23972714e-01 -6.24947660e-02 -1.85852140e-01 -2.54929572e-01
1.64885104e-01 3.98164719e-01 -2.85558611e-01 -1.10883862e-02
-2.05934191e+00 8.54943037e-01 5.78417778e-01 3.41048092e-01
-4.50047165e-01 -1.92202508e-01 -8.64578187e-01 6.08026348e-02
8.87642145e-01 -8.91497731e-01 1.06314099e+00 -7.05504358e-01
-1.53147149e+00 5.91868579e-01 -3.22928101e-01 1.54033557e-01
5.03292501e-01 2.02467918e-01 -4.50995505e-01 7.21210008e-03
-3.88818532e-02 5.36668420e-01 4.85659689e-01 -1.38624382e+00
-5.04041553e-01 -3.66075069e-01 4.90883201e-01 7.51055598e-01
-6.12176299e-01 -1.79917991e-01 -5.84162474e-01 -4.46793556e-01
2.67878801e-01 -7.61396825e-01 -1.04411334e-01 -6.13370717e-01
-7.74410069e-01 -6.23451293e-01 5.28033197e-01 -6.12691343e-01
1.40114772e+00 -1.85498941e+00 4.77672637e-01 5.86115383e-02
2.89901346e-01 1.23773813e-01 -1.21575616e-01 5.92853725e-01
2.87222326e-01 9.64396745e-02 -3.78454357e-01 -3.75404388e-01
2.18649611e-01 9.99062657e-02 -4.44424927e-01 -1.69077873e-01
1.62781283e-01 1.03121567e+00 -8.82163882e-01 -6.50721312e-01
-4.34484892e-02 1.59423143e-01 -4.22574222e-01 4.52015370e-01
-4.80821162e-01 6.15276098e-01 -1.16639960e+00 4.74913746e-01
5.44069827e-01 -5.20995259e-01 3.52716029e-01 -6.46136522e-01
4.12781060e-01 4.07956988e-01 -9.92022634e-01 1.77752018e+00
-5.44984519e-01 3.34185623e-02 1.41965106e-01 -8.42742205e-01
1.15040553e+00 3.43299985e-01 3.56115520e-01 -7.66128600e-01
1.63530946e-01 -3.83933708e-02 -1.37795925e-01 -1.06690359e+00
5.27232230e-01 -3.68808508e-01 -3.79396290e-01 3.87149781e-01
-2.56197974e-02 1.55796692e-01 -2.19883844e-01 6.24397039e-01
6.43245280e-01 -1.46506324e-01 -5.02230376e-02 3.02442871e-02
1.14086986e+00 -9.66099091e-03 4.90435123e-01 2.58452952e-01
-4.76126187e-02 1.54341280e-01 7.40737140e-01 1.92521974e-01
-4.75096613e-01 -7.11593926e-01 1.61812991e-01 1.10023463e+00
6.21979833e-01 -3.60856861e-01 -6.01530433e-01 -9.54157770e-01
-1.04298957e-01 8.35900784e-01 -5.20613194e-01 -5.78416228e-01
-4.07005519e-01 -8.62267315e-01 5.88499665e-01 4.20580626e-01
8.90116274e-01 -1.17824030e+00 2.93006867e-01 1.26982674e-01
-9.88404274e-01 -1.18434930e+00 -5.86966932e-01 -2.01803163e-01
-5.65027893e-01 -9.31298971e-01 -4.73899573e-01 -8.50991309e-01
7.47804701e-01 2.39292905e-01 9.53385949e-01 1.87654778e-01
3.53382111e-01 3.84490192e-01 -4.84133959e-01 2.00393975e-01
-2.31900573e-01 2.13410720e-01 -1.23603225e-01 3.79403949e-01
7.07804024e-01 -3.86533886e-01 -5.88840187e-01 4.70515490e-01
-9.11706269e-01 1.71424732e-01 6.56026423e-01 1.01629758e+00
3.31489772e-01 -8.03759322e-02 1.11416662e+00 -1.05888879e+00
1.35710013e+00 -7.05726564e-01 -2.00316578e-01 6.17150903e-01
-6.64404511e-01 7.51164258e-02 6.41200244e-01 -3.42933208e-01
-1.58619702e+00 -1.46807849e-01 -9.68050584e-02 -1.57737762e-01
-7.72479130e-03 8.85362267e-01 -7.59427488e-01 3.07382494e-01
2.92163819e-01 4.48343754e-01 -5.48146218e-02 -4.73143309e-01
6.55137062e-01 9.26042259e-01 1.77666038e-01 -7.79476404e-01
6.54934168e-01 1.60136580e-01 -2.57469267e-01 -4.36208427e-01
-9.96315897e-01 -3.80107284e-01 -3.62444490e-01 -1.59569876e-03
1.05998528e+00 -7.71260083e-01 -1.29420757e+00 1.70895696e-01
-1.37755477e+00 1.50425375e-01 1.76251993e-01 3.57279092e-01
-3.46882373e-01 5.18559396e-01 -7.20090806e-01 -9.98664379e-01
-2.75974780e-01 -1.15495908e+00 8.45181286e-01 4.35063571e-01
-2.92302072e-02 -1.15060580e+00 -1.26623407e-01 9.00914490e-01
1.86582461e-01 -1.09739326e-01 1.36942542e+00 -1.00864971e+00
-7.30096459e-01 2.09796187e-02 -5.50845444e-01 1.57288745e-01
2.65379280e-01 -5.09696364e-01 -8.91009510e-01 -1.62662417e-02
2.88720548e-01 -7.75272548e-01 7.78668404e-01 -2.54133224e-01
9.25581753e-01 -5.75098872e-01 -2.49695852e-01 6.72811717e-02
1.02460086e+00 1.15440898e-01 2.73719281e-01 6.91616163e-02
8.24830174e-01 1.27519560e+00 7.84908235e-01 3.06967914e-01
9.43245471e-01 7.81659484e-01 1.01694599e-01 -4.48269546e-02
1.62295982e-01 -7.54033923e-01 2.08510771e-01 1.36726058e+00
1.99984699e-01 -3.87444973e-01 -5.25684178e-01 2.37889782e-01
-2.11556745e+00 -5.81595480e-01 -2.34141618e-01 1.91804051e+00
1.12685347e+00 -2.80223548e-01 -1.92231480e-02 -4.15537328e-01
7.18935966e-01 2.00645030e-01 -6.72365963e-01 1.78704068e-01
-3.19421470e-01 -4.14239556e-01 -9.98445824e-02 7.18025982e-01
-8.14318061e-01 1.13969707e+00 4.74045706e+00 1.11130893e+00
-5.92649817e-01 1.52508793e-02 3.81935269e-01 3.45750988e-01
-8.86875808e-01 2.39382327e-01 -8.46446872e-01 4.60994452e-01
3.11067790e-01 -4.11861926e-01 3.52698743e-01 5.17312229e-01
2.17207342e-01 1.87346503e-01 -9.45962131e-01 8.85434449e-01
3.89554165e-02 -8.83596539e-01 4.41459477e-01 -2.00163543e-01
3.55529636e-01 -7.03806639e-01 -2.75281034e-02 6.55283093e-01
3.38696033e-01 -9.44782495e-01 5.69552332e-02 7.81912267e-01
7.28567719e-01 -5.86916029e-01 8.19738150e-01 6.56835079e-01
-1.20580029e+00 -3.21066082e-02 -4.58120167e-01 3.49235982e-01
2.05635339e-01 2.03897417e-01 -6.96808279e-01 1.17524719e+00
8.41572657e-02 6.22842491e-01 -4.17762607e-01 2.86588520e-01
-3.97294611e-01 3.50045234e-01 2.97294743e-02 -2.47841731e-01
2.75740862e-01 -5.35155773e-01 4.17026222e-01 9.59248483e-01
-1.97624452e-02 5.56333244e-01 4.41692919e-01 1.10100865e+00
-3.63181710e-01 4.57238942e-01 -3.23036700e-01 -9.86201838e-02
7.28101134e-01 1.15973604e+00 -1.26666874e-01 -2.74001211e-01
-5.24354875e-01 8.94311547e-01 3.98343235e-01 6.55962050e-01
-6.08054221e-01 -4.75546330e-01 3.00650179e-01 -3.48624259e-01
-1.02315120e-01 1.24675632e-01 -7.90063143e-02 -1.46226871e+00
1.71541229e-01 -1.01848853e+00 6.23958886e-01 -6.31801605e-01
-1.62455523e+00 6.76078022e-01 1.89573206e-02 -1.10343957e+00
-1.24523297e-01 -1.32191181e-01 -3.79420638e-01 1.30904269e+00
-1.51543629e+00 -1.26600361e+00 -3.82063776e-01 8.27659547e-01
4.81958300e-01 -2.54289806e-01 8.90245199e-01 3.60417217e-01
-8.07804465e-01 7.63836384e-01 -2.46014938e-01 2.37831131e-01
7.06830323e-01 -8.94000351e-01 -2.60246366e-01 1.86400056e-01
-3.10093820e-01 9.46109295e-01 3.65347147e-01 -6.84106112e-01
-1.41028190e+00 -7.90081978e-01 1.01355612e+00 -3.85529339e-01
6.89853787e-01 -3.36127043e-01 -1.23095024e+00 5.63864052e-01
2.05982804e-01 -6.60591602e-01 8.94855022e-01 2.48965263e-01
-9.97747108e-02 -1.04332373e-01 -9.32655215e-01 7.37399578e-01
9.82518196e-01 -8.11993718e-01 -9.55820501e-01 4.60310608e-01
1.24857867e+00 -2.02631310e-01 -7.70123601e-01 5.50615966e-01
5.34497760e-02 -7.14786351e-01 8.72890294e-01 -8.67034554e-01
4.99422282e-01 -2.63781130e-01 -2.14912534e-01 -1.12954044e+00
-4.67430241e-02 -5.20174682e-01 -3.26318108e-02 1.76349366e+00
4.96380687e-01 -5.97220480e-01 7.38180459e-01 1.01258004e+00
7.64022246e-02 -8.33077490e-01 -7.24469543e-01 -4.20412540e-01
1.09499926e-02 -4.38930169e-02 5.87529659e-01 1.15569675e+00
3.01214725e-01 1.03216970e+00 -6.20784461e-01 4.95086145e-03
3.27485293e-01 4.14196372e-01 7.31296718e-01 -1.04412568e+00
-2.53182836e-02 -3.53680193e-01 1.95336953e-01 -1.78780723e+00
4.72802579e-01 -9.80206370e-01 -1.73579734e-02 -1.51614487e+00
2.79730350e-01 -7.06739962e-01 -2.57665873e-01 3.68081897e-01
-7.21277356e-01 -5.41540980e-01 -1.41879812e-01 4.12378043e-01
-7.51884818e-01 1.26236188e+00 1.66036487e+00 -2.19353527e-01
-2.42836818e-01 -1.19312955e-02 -1.04668117e+00 5.44899285e-01
4.23801750e-01 -4.11367893e-01 -9.44721818e-01 -3.64560664e-01
2.74809211e-01 5.81123173e-01 3.74703437e-01 -2.20896587e-01
4.92637068e-01 -4.44818020e-01 -6.66998848e-02 -4.59897548e-01
4.97173280e-01 -8.77046466e-01 -2.86950290e-01 1.58123583e-01
-9.45789278e-01 -2.29917422e-01 -1.68313012e-01 9.08771276e-01
-4.74601358e-01 -2.56028652e-01 2.62535632e-01 -4.57424186e-02
-6.87863588e-01 4.36782628e-01 5.34158535e-02 4.05514777e-01
6.62698925e-01 1.65858209e-01 -4.85513896e-01 -5.87337434e-01
-6.07971847e-01 8.02506924e-01 1.69195861e-01 5.71027398e-01
7.11789668e-01 -1.59930158e+00 -7.37714052e-01 -7.44500160e-02
3.16645592e-01 9.53940488e-03 6.13990784e-01 8.76168072e-01
1.45002618e-01 5.64611673e-01 1.50012523e-01 -2.75444716e-01
-1.04942000e+00 5.85737348e-01 2.62027085e-01 -3.59312743e-01
-3.39867204e-01 9.27709401e-01 3.96904349e-01 -8.52526188e-01
1.78874820e-01 2.61317700e-01 -6.90143704e-01 2.37887233e-01
1.34243473e-01 1.36125237e-01 -2.93301255e-01 -6.68395221e-01
-4.03550476e-01 2.91078776e-01 -8.86647180e-02 -2.91088223e-01
8.50649595e-01 -6.40989363e-01 -3.29889923e-01 4.14521813e-01
1.17546344e+00 2.74738688e-02 -8.87256384e-01 -8.56339097e-01
1.19838566e-01 -3.60343516e-01 -2.44550124e-01 -9.19619858e-01
-9.13545430e-01 1.12165427e+00 1.64400768e-02 2.31234804e-01
1.15774560e+00 4.77957465e-02 1.10088849e+00 6.63149655e-01
1.13521740e-01 -1.00026214e+00 3.24281365e-01 6.13171399e-01
9.54157829e-01 -1.27655888e+00 -2.84337759e-01 -7.44246244e-01
-1.27956712e+00 8.06566656e-01 1.03018081e+00 4.56410140e-01
3.56865674e-01 -3.04862648e-01 2.02120647e-01 -2.26737231e-01
-8.41586351e-01 -3.36485654e-01 4.99859273e-01 4.81701046e-01
4.56298709e-01 -6.45875484e-02 -4.98318195e-01 1.24992597e+00
-1.20995669e-02 -2.23287180e-01 -1.88075975e-02 4.41265523e-01
-3.15951049e-01 -1.26851499e+00 2.10992128e-01 2.13990971e-01
5.04675228e-03 -2.49075815e-01 -8.34325731e-01 3.73626590e-01
-9.68105569e-02 1.36901677e+00 -7.04820037e-01 -8.14189315e-01
5.72181880e-01 3.00683975e-01 -6.95967376e-02 -4.52254206e-01
-5.68087161e-01 -1.21638797e-01 3.55814338e-01 -1.88171327e-01
-3.81315500e-01 -1.96674824e-01 -1.33480585e+00 -1.50258258e-01
-5.22606015e-01 5.49931049e-01 2.16490179e-01 1.14727521e+00
4.07661110e-01 6.13036633e-01 8.56885135e-01 2.59462260e-02
-8.76252532e-01 -1.15370429e+00 -3.79859447e-01 7.60048330e-01
-3.58184129e-02 -6.53600097e-01 -3.26676786e-01 -8.23596865e-02]
|
[12.34627914428711, 7.934515476226807]
|
6e174b2b-8778-43d6-b471-2c25a8ff7ee6
|
viewrefer-grasp-the-multi-view-knowledge-for
|
2303.16894
| null |
https://arxiv.org/abs/2303.16894v2
|
https://arxiv.org/pdf/2303.16894v2.pdf
|
ViewRefer: Grasp the Multi-view Knowledge for 3D Visual Grounding with GPT and Prototype Guidance
|
Understanding 3D scenes from multi-view inputs has been proven to alleviate the view discrepancy issue in 3D visual grounding. However, existing methods normally neglect the view cues embedded in the text modality and fail to weigh the relative importance of different views. In this paper, we propose ViewRefer, a multi-view framework for 3D visual grounding exploring how to grasp the view knowledge from both text and 3D modalities. For the text branch, ViewRefer leverages the diverse linguistic knowledge of large-scale language models, e.g., GPT, to expand a single grounding text to multiple geometry-consistent descriptions. Meanwhile, in the 3D modality, a transformer fusion module with inter-view attention is introduced to boost the interaction of objects across views. On top of that, we further present a set of learnable multi-view prototypes, which memorize scene-agnostic knowledge for different views, and enhance the framework from two perspectives: a view-guided attention module for more robust text features, and a view-guided scoring strategy during the final prediction. With our designed paradigm, ViewRefer achieves superior performance on three benchmarks and surpasses the second-best by +2.8%, +1.2%, and +0.73% on Sr3D, Nr3D, and ScanRefer. Code will be released at https://github.com/ZiyuGuo99/ViewRefer3D.
|
['Xuelong Li', 'Bin Zhao', 'Zhigang Wang', 'Dong Wang', 'Renrui Zhang', 'Yiwen Tang', 'Ziyu Guo']
|
2023-03-29
| null | null | null | null |
['visual-grounding']
|
['computer-vision']
|
[-1.84670329e-01 -1.38344929e-01 -1.63161635e-01 -5.90413153e-01
-7.54862666e-01 -7.21229613e-01 7.28358448e-01 -1.58179030e-01
1.88415170e-01 -3.39128971e-02 6.20569050e-01 -1.59477547e-01
1.34687796e-01 -6.51505232e-01 -7.61756003e-01 -3.67794722e-01
6.08309269e-01 3.99163365e-01 1.50626808e-01 -2.73990303e-01
1.59282044e-01 3.10559630e-01 -1.58723855e+00 7.46214092e-01
6.18617892e-01 1.09133732e+00 5.32289624e-01 3.66966873e-01
-2.28187531e-01 5.55277884e-01 2.57766582e-02 -4.26407427e-01
3.17194074e-01 -8.70951712e-02 -5.86342514e-01 3.69790465e-01
9.06586111e-01 -4.09546196e-01 -4.53591049e-01 8.99289310e-01
5.44182360e-01 7.25438967e-02 4.26557064e-01 -1.11240590e+00
-1.04785228e+00 4.31017071e-01 -9.94532764e-01 2.32666731e-02
5.47811210e-01 1.36486918e-01 1.19197416e+00 -1.56254256e+00
5.81204176e-01 1.50458241e+00 3.22948128e-01 4.29116845e-01
-9.56920564e-01 -5.09517431e-01 7.71260798e-01 1.84016436e-01
-1.14210188e+00 -5.02895594e-01 8.73490095e-01 -6.22721612e-01
1.10530090e+00 1.75646693e-01 7.55699933e-01 1.27437556e+00
2.03987718e-01 1.00804245e+00 1.31589866e+00 -2.00552851e-01
-2.29775622e-01 1.09014884e-01 8.49342048e-02 8.31651926e-01
1.47133455e-01 1.31690592e-01 -8.92573237e-01 2.07099512e-01
7.50285745e-01 4.43154752e-01 -3.86015058e-01 -7.00182438e-01
-1.38172531e+00 6.37422621e-01 6.86571419e-01 -4.98065799e-02
-2.79098511e-01 -1.07385382e-01 2.83830166e-01 -8.03080127e-02
7.72166729e-01 1.50018007e-01 -4.12358075e-01 1.21518500e-01
-5.59839070e-01 -4.56252694e-03 3.68074954e-01 1.36842966e+00
7.12148428e-01 -1.04912102e-01 -2.61350989e-01 8.01817536e-01
7.07550824e-01 9.04634178e-01 1.77685043e-03 -6.81876063e-01
1.03819799e+00 8.95576239e-01 -1.25977054e-01 -1.03416789e+00
-4.72463787e-01 -4.87835884e-01 -8.28749478e-01 1.00574285e-01
-3.30016427e-02 4.12686676e-01 -1.03091764e+00 1.63589561e+00
4.23444569e-01 -1.82631671e-01 -7.26938024e-02 1.10659754e+00
1.31478119e+00 4.21466202e-01 -1.81406885e-01 2.63196796e-01
1.41415501e+00 -1.39919412e+00 -3.98173332e-01 -5.64600110e-01
4.49281365e-01 -8.55380297e-01 1.34540176e+00 3.85063797e-01
-1.17290807e+00 -6.97602332e-01 -9.58601594e-01 -4.84982789e-01
-4.17401910e-01 2.88905561e-01 4.48684275e-01 2.43485168e-01
-1.05663943e+00 6.53017834e-02 -8.96535158e-01 -4.17621583e-01
4.34413075e-01 1.05462838e-02 -5.27022660e-01 -5.59780300e-01
-9.10151660e-01 8.74281585e-01 1.89956665e-01 1.70439318e-01
-7.33402729e-01 -6.36132300e-01 -1.05670178e+00 -1.92844689e-01
5.89135587e-01 -1.19909716e+00 9.11338627e-01 -4.83376414e-01
-1.11542666e+00 1.22525764e+00 -3.27213258e-01 1.20700374e-01
5.10759592e-01 -5.52699685e-01 -2.77684718e-01 1.97265640e-01
2.41657123e-01 5.84441781e-01 6.27665401e-01 -1.69212532e+00
-3.98463100e-01 -7.78002083e-01 4.43216830e-01 6.91428959e-01
4.26975489e-02 -3.53863388e-01 -9.17110801e-01 -6.58675134e-01
6.00736439e-01 -8.51945639e-01 1.42136022e-01 2.27025211e-01
-5.97849369e-01 5.57351261e-02 6.28642499e-01 -7.07167506e-01
8.36584687e-01 -2.07602000e+00 5.73738456e-01 -1.60023049e-01
5.34574687e-01 -1.95801288e-01 -7.77789354e-02 5.69441736e-01
-7.49593675e-02 3.28548849e-02 1.71856284e-01 -5.84990025e-01
2.33100932e-02 1.65448099e-01 -2.85930306e-01 3.02248687e-01
1.56736925e-01 1.04046810e+00 -8.38737130e-01 -2.81888723e-01
4.96353865e-01 5.31347275e-01 -6.99508131e-01 1.74119711e-01
-9.87939313e-02 4.64199305e-01 -5.99010944e-01 7.59400964e-01
7.91978240e-01 -8.25848639e-01 -4.17569913e-02 -5.83421528e-01
-5.91770932e-02 2.53741026e-01 -1.02862060e+00 2.27353454e+00
-7.32147753e-01 3.15812796e-01 -5.01240278e-03 -6.76427007e-01
8.63510430e-01 7.53459036e-02 2.16065675e-01 -6.96377516e-01
-2.76897894e-03 -5.28678522e-02 -4.40064400e-01 -5.17026663e-01
6.00320041e-01 -3.16977724e-02 -4.08302210e-02 4.70280290e-01
1.97264120e-01 -2.22582281e-01 -2.92153060e-01 3.50834340e-01
5.95423222e-01 4.79335397e-01 3.77019197e-01 4.83545661e-02
3.86365414e-01 -2.78074741e-01 2.55479485e-01 5.88188350e-01
1.78770460e-02 9.33584929e-01 1.52771354e-01 -4.62638497e-01
-7.73573279e-01 -1.19154406e+00 1.28891855e-01 1.15432024e+00
6.91155732e-01 -5.61223149e-01 -1.02312408e-01 -7.87340641e-01
6.50062710e-02 6.74197018e-01 -7.09945977e-01 -1.96250066e-01
-3.23386878e-01 -2.76810467e-01 -1.06187657e-01 8.21437120e-01
5.25093019e-01 -5.51016808e-01 -4.61154193e-01 -2.51573622e-01
-4.16069895e-01 -1.32725334e+00 -6.75040722e-01 2.40904316e-02
-8.63162875e-01 -1.01168048e+00 -6.22088313e-01 -4.86244470e-01
5.94427407e-01 1.05132687e+00 1.39045179e+00 3.13298702e-02
1.47115231e-01 7.53288031e-01 -6.25059426e-01 -2.88181633e-01
-3.04048788e-02 -7.79525191e-02 1.47501603e-02 1.61485467e-03
2.24183977e-01 -6.03756189e-01 -6.51099086e-01 4.14218813e-01
-5.22100866e-01 8.29693377e-01 7.40771055e-01 9.04350340e-01
8.56714964e-01 -6.22230709e-01 3.94989178e-02 -7.12949097e-01
-5.47809973e-02 -4.59014803e-01 -3.50069761e-01 4.52419311e-01
-4.43006575e-01 -8.95651728e-02 3.30802500e-01 -7.85467848e-02
-9.67437625e-01 -1.19166791e-01 5.43582961e-02 -1.01294327e+00
-1.61853164e-01 4.90589768e-01 -6.64848328e-01 1.26847222e-01
3.14024866e-01 3.45925599e-01 -1.92589715e-01 -5.79451978e-01
6.53812408e-01 3.49346042e-01 1.76460132e-01 -5.76315522e-01
8.56195271e-01 7.91029334e-01 -2.15614244e-01 -4.56048548e-01
-1.43861759e+00 -6.32135510e-01 -8.19776952e-01 -2.42643118e-01
9.81793523e-01 -1.38853645e+00 -3.73946518e-01 3.46071273e-01
-1.16694117e+00 -9.17202607e-02 -3.78196947e-02 4.54449028e-01
-6.29727840e-01 3.85943949e-01 -2.94975698e-01 -4.68839824e-01
-3.39685500e-01 -1.17009807e+00 1.65919471e+00 6.53868839e-02
1.82271168e-01 -8.95405412e-01 -3.24441701e-01 7.61763453e-01
1.01643533e-01 1.35842860e-01 8.94636929e-01 -5.73797405e-01
-7.69766092e-01 6.32957891e-02 -5.55355072e-01 1.97385371e-01
1.51351422e-01 -3.02455813e-01 -1.17726028e+00 -3.60293299e-01
-3.93882096e-02 -4.37176853e-01 9.52109039e-01 2.03403622e-01
1.08283162e+00 -2.49173846e-02 -3.89217854e-01 8.21388483e-01
1.35698807e+00 -4.08926327e-03 2.25196376e-01 1.80237919e-01
1.26220953e+00 5.81072211e-01 6.88832939e-01 4.29369152e-01
8.54415298e-01 8.76694620e-01 6.66177213e-01 -7.68903717e-02
-2.71935165e-01 -5.23456216e-01 2.50056624e-01 1.12435794e+00
-1.83545705e-03 -5.37593186e-01 -1.09664595e+00 2.97550023e-01
-1.79034793e+00 -9.38194096e-01 -2.23240107e-02 1.98523116e+00
2.78201818e-01 1.73013210e-01 -1.45978734e-01 -2.54507780e-01
6.02431238e-01 5.14930129e-01 -7.34463394e-01 -4.18721735e-02
-2.28643253e-01 -3.80233675e-01 1.36215687e-01 4.14687455e-01
-1.14614689e+00 9.50837016e-01 4.58820677e+00 5.27272344e-01
-1.12635756e+00 7.34648854e-02 4.84727085e-01 -2.35098302e-01
-6.76574111e-01 -2.64778212e-02 -8.41527939e-01 9.84266475e-02
1.81925118e-01 9.80756730e-02 2.62489468e-01 7.44067430e-01
1.52900055e-01 1.12721413e-01 -1.31624627e+00 1.20064175e+00
5.72711825e-01 -1.29119372e+00 4.93217826e-01 1.71559870e-01
7.00533748e-01 3.19375962e-01 2.38059491e-01 2.45894715e-01
4.13049348e-02 -9.03785467e-01 1.03578746e+00 7.08775938e-01
7.09777713e-01 -4.81919050e-01 5.12633979e-01 3.38226050e-01
-1.50699914e+00 6.79236576e-02 -1.27328366e-01 1.17529251e-01
3.11008900e-01 3.85455132e-01 -3.43885094e-01 1.15631843e+00
8.45014393e-01 1.27903187e+00 -7.00422168e-01 6.71471655e-01
-2.30403826e-01 9.41872597e-02 -1.41748205e-01 2.65687108e-01
3.53291571e-01 -8.08359161e-02 6.39370799e-01 7.27455318e-01
3.38403165e-01 1.79108620e-01 4.16246772e-01 9.14427459e-01
1.01808079e-01 -5.72994612e-02 -7.63275146e-01 1.99053004e-01
3.46484691e-01 1.15532851e+00 -4.27373976e-01 -3.37425500e-01
-8.33747685e-01 9.89199221e-01 4.54736888e-01 3.79587293e-01
-7.37791717e-01 1.54919803e-01 4.46498603e-01 1.12403437e-01
5.35667598e-01 -3.62161487e-01 -4.30281729e-01 -1.64025223e+00
3.03913295e-01 -7.86757112e-01 3.58709604e-01 -1.34510708e+00
-1.55905259e+00 7.21251547e-01 1.46645010e-01 -1.55459189e+00
7.88741037e-02 -8.54380071e-01 -3.59174579e-01 1.07371509e+00
-1.50398219e+00 -1.86810756e+00 -6.52935982e-01 5.67162097e-01
8.30288112e-01 -5.13316877e-02 5.71003735e-01 1.54590204e-01
-3.46210659e-01 4.44515347e-01 -1.28563106e-01 -1.17352232e-01
8.10870111e-01 -1.25348580e+00 6.03180945e-01 6.23325169e-01
4.50167984e-01 6.58298671e-01 2.76254952e-01 -5.77672184e-01
-1.80728841e+00 -1.11384106e+00 7.37651825e-01 -8.27375174e-01
5.05884469e-01 -5.88990152e-01 -8.66122365e-01 7.48057246e-01
8.84357169e-02 1.76571816e-01 5.88915706e-01 3.07534248e-01
-9.10088837e-01 1.00018673e-01 -6.55311465e-01 6.52933538e-01
1.54818475e+00 -7.86162138e-01 -6.08944595e-01 1.72917992e-01
8.85434449e-01 -8.70533347e-01 -8.70023370e-01 3.55704486e-01
6.75241709e-01 -1.17727196e+00 1.17376530e+00 -5.35863280e-01
6.96387827e-01 -2.69168735e-01 -7.55035222e-01 -1.15464127e+00
-4.04396951e-01 -9.47667435e-02 -1.07020430e-01 1.10253966e+00
3.14801127e-01 -4.81685102e-01 5.17437339e-01 1.69367537e-01
-5.14244080e-01 -1.09392989e+00 -8.30016673e-01 -5.59429049e-01
6.96349293e-02 -5.65531373e-01 4.89950299e-01 9.48111117e-01
-2.31739014e-01 7.05821633e-01 -4.89671707e-01 4.76278722e-01
4.35574025e-01 8.35827231e-01 8.64741564e-01 -1.06324518e+00
-4.26811576e-01 -3.52804810e-01 -3.34699631e-01 -1.52261066e+00
4.59720381e-02 -1.21194565e+00 -2.34466836e-01 -1.85734367e+00
5.10335803e-01 -2.35928535e-01 -2.50043869e-01 4.57479805e-01
-3.04314762e-01 8.14552158e-02 5.70836186e-01 1.50584683e-01
-7.44712651e-01 7.53374755e-01 1.69489574e+00 -1.64424151e-01
6.32343292e-02 -1.82750657e-01 -9.01573002e-01 9.09647822e-01
6.17081583e-01 -1.00528724e-01 -5.82820892e-01 -9.92608786e-01
4.34166819e-01 1.95456818e-01 7.33606696e-01 -6.73245370e-01
1.10854998e-01 -2.08158061e-01 5.04264057e-01 -1.19161177e+00
6.98362768e-01 -8.15921843e-01 -1.14855565e-01 -6.85926750e-02
-2.54692852e-01 3.24606568e-01 2.48031929e-01 9.48808730e-01
-8.83203447e-02 3.98322433e-01 4.18795615e-01 -3.28096777e-01
-8.53171051e-01 6.12213671e-01 2.77230710e-01 2.96442270e-01
6.64803565e-01 -4.07760531e-01 -4.75899011e-01 -3.33280921e-01
-7.93430209e-01 5.12433052e-01 7.10644901e-01 8.75307202e-01
8.89945209e-01 -1.49862301e+00 -5.61920762e-01 2.43547678e-01
7.16707051e-01 1.97610244e-01 7.49637783e-01 9.62527275e-01
-4.55775969e-02 3.76202226e-01 -8.49105120e-02 -9.59763885e-01
-1.19110513e+00 6.34232104e-01 3.30717534e-01 -1.97170317e-01
-8.74972582e-01 9.38983977e-01 9.17730153e-01 -5.62256694e-01
1.99439917e-02 -4.23405707e-01 -1.23492375e-01 6.38320073e-02
3.84144157e-01 5.42974360e-02 7.40877986e-02 -9.61138248e-01
-4.25082833e-01 1.13380539e+00 -2.05733716e-01 -6.48457408e-02
1.23213518e+00 -5.55398107e-01 2.30454877e-01 8.69518399e-01
1.12121415e+00 -1.40036419e-02 -1.34752941e+00 -4.93876934e-01
-4.02308434e-01 -6.36044860e-01 3.60330865e-02 -9.73704517e-01
-1.08969331e+00 1.30106509e+00 4.37883526e-01 -1.40469164e-01
1.04141784e+00 3.87503386e-01 6.01046443e-01 2.60875046e-01
4.51206475e-01 -5.16027153e-01 3.75594527e-01 7.01019466e-01
1.41143656e+00 -1.60292518e+00 1.60360709e-01 -4.77617294e-01
-9.17055666e-01 9.73842621e-01 9.33667779e-01 1.67369023e-01
6.69545233e-01 -1.09363273e-01 2.99134962e-02 -5.43332100e-01
-1.09175849e+00 -1.59522548e-01 7.99062133e-01 5.56290448e-01
3.23659718e-01 1.45537406e-02 4.47919488e-01 6.37703240e-01
-3.79967466e-02 -4.54915106e-01 2.03156292e-01 7.67442584e-01
-2.15968087e-01 -6.76193416e-01 -1.52958244e-01 3.65944266e-01
-1.76269729e-02 -2.68729627e-01 -4.58825648e-01 8.63651156e-01
9.64409411e-02 8.04842889e-01 -8.81736651e-02 -5.40679336e-01
5.96916676e-01 -7.99933970e-02 4.86378729e-01 -8.01302075e-01
-3.58718574e-01 2.55059034e-01 8.30084756e-02 -7.82226861e-01
-6.18980765e-01 -5.35632014e-01 -9.36144948e-01 -2.15337619e-01
-3.29482973e-01 -3.55871379e-01 3.53884757e-01 8.28359604e-01
6.01334512e-01 4.26518589e-01 6.21161461e-01 -1.07291424e+00
-3.45455527e-01 -7.01260865e-01 -4.48318124e-01 2.90929914e-01
2.47639447e-01 -1.02238297e+00 -3.26050013e-01 -8.52783993e-02]
|
[8.219754219055176, -3.4956963062286377]
|
f7676539-6846-458b-8f44-6be0e53ed535
|
using-deep-convolutional-neural-networks-for
| null | null |
https://www.frontiersin.org/articles/10.3389/fnins.2020.00207/full?report=reader
|
https://www.frontiersin.org/articles/10.3389/fnins.2020.00207/full?report=reader
|
Using deep convolutional neural networks for neonatal brain image segmentation
|
Introduction: Deep learning neural networks are especially potent at dealing with structured data, such as images and volumes. Both modified LiviaNET and HyperDense-Net performed well at a prior competition segmenting 6-month-old infant magnetic resonance images, but neonatal cerebral tissue type identification is challenging given its uniquely inverted tissue contrasts. The current study aims to evaluate the two architectures to segment neonatal brain tissue types at term equivalent age.
Methods: Both networks were retrained over 24 pairs of neonatal T1 and T2 data from the Developing Human Connectome Project public data set and validated on another eight pairs against ground truth. We then reported the best-performing model from training and its performance by computing the Dice similarity coefficient (DSC) for each tissue type against eight test subjects.
Results: During the testing phase, among the segmentation approaches tested, the dual-modality HyperDense-Net achieved the best statistically significantly test mean DSC values, obtaining 0.94/0.95/0.92 for the tissue types and took 80 h to train and 10 min to segment, including preprocessing. The single-modality LiviaNET was better at processing T2-weighted images than processing T1-weighted images across all tissue types, achieving mean DSC values of 0.90/0.90/0.88 for gray matter, white matter, and cerebrospinal fluid, respectively, while requiring 30 h to train and 8 min to segment each brain, including preprocessing.
Discussion: Our evaluation demonstrates that both neural networks can segment neonatal brains, achieving previously reported performance. Both networks will be continuously retrained over an increasingly larger repertoire of neonatal brain data and be made available through the Canadian Neonatal Brain Platform to better serve the neonatal brain imaging research community.
|
['Lodygensky GA.', 'Dolz J', 'Luck D', 'Ortmann J', 'Suffren S', 'Enguix V', 'Acosta R', 'Ding Y']
|
2020-03-26
| null | null | null | null |
['brain-image-segmentation']
|
['medical']
|
[ 2.05123857e-01 4.18465108e-01 -2.35281400e-02 -6.29564166e-01
-8.24316204e-01 -6.65439606e-01 1.40051305e-01 2.66828209e-01
-8.46793473e-01 6.21121585e-01 -8.80055279e-02 -4.26962197e-01
-1.21541992e-01 -3.23825866e-01 -7.39320934e-01 -5.66189289e-01
-5.94686031e-01 9.06772733e-01 2.67149508e-01 3.57116133e-01
4.25042864e-03 4.66724157e-01 -9.54078496e-01 2.84346998e-01
9.54980075e-01 1.03502822e+00 1.49490118e-01 5.97714603e-01
3.57301310e-02 5.20542204e-01 -1.18463211e-01 -4.30628836e-01
4.05861974e-01 -3.22549909e-01 -1.08756721e+00 -4.87151265e-01
1.00700271e+00 -6.18257403e-01 -1.32038772e-01 7.97538042e-01
9.93128538e-01 1.58744492e-02 8.45095515e-01 -7.68513262e-01
-3.76649797e-01 9.74019587e-01 -6.10626161e-01 1.02177215e+00
-1.63135931e-01 1.12873264e-01 3.69185448e-01 -7.82662511e-01
6.94277585e-01 6.31918192e-01 1.04282904e+00 7.80532002e-01
-1.20717740e+00 -9.88459527e-01 -1.31475419e-01 1.27639994e-01
-1.04514146e+00 -4.53280747e-01 7.66632482e-02 -7.56765246e-01
1.28621829e+00 -8.00645724e-02 9.73456323e-01 8.56197059e-01
2.65143037e-01 5.30390143e-01 1.10536194e+00 1.47260636e-01
2.89574657e-02 -5.73390126e-01 1.11611106e-01 7.30418503e-01
4.01636392e-01 -8.94613117e-02 7.56792054e-02 6.50100186e-02
7.96221793e-01 -3.21066469e-01 -2.75574207e-01 -2.17820555e-01
-1.48758757e+00 5.35784781e-01 3.38127434e-01 5.75034976e-01
-5.24749994e-01 8.42766371e-03 7.33893037e-01 1.00358136e-01
6.22837543e-01 4.25056875e-01 -3.79006147e-01 -2.18583480e-01
-1.41949046e+00 -9.99407843e-02 5.43631017e-01 8.57762039e-01
1.66603506e-01 2.49470817e-03 3.22100930e-02 1.27622318e+00
-1.02951139e-01 3.73328716e-01 6.85647130e-01 -1.14125621e+00
7.14947283e-01 1.18391834e-01 -6.69572353e-01 -4.65938598e-01
-1.18823612e+00 -3.69179755e-01 -1.04161382e+00 6.76764324e-02
6.69496298e-01 -6.57606900e-01 -1.41909444e+00 1.83824241e+00
1.35137722e-01 -5.13871610e-02 -2.80994684e-01 9.42147493e-01
1.03271925e+00 7.53583089e-02 2.56185889e-01 -2.05518171e-01
1.09812939e+00 -6.92254245e-01 -1.25366852e-01 -1.25850737e-01
8.21214318e-01 -4.38222319e-01 2.61200160e-01 2.67182618e-01
-1.78274405e+00 -1.30958319e-01 -9.96743500e-01 1.86215758e-01
-1.10694535e-01 -3.22001338e-01 4.11911666e-01 6.83318615e-01
-1.56801951e+00 7.62090385e-01 -1.28566360e+00 -2.95431733e-01
9.23364758e-01 7.67138064e-01 -7.44043529e-01 -2.43660629e-01
-9.56043601e-01 1.44704151e+00 5.12047589e-01 1.89292803e-02
-1.01156259e+00 -1.28806162e+00 -5.61230242e-01 -1.22008264e-01
-2.65003711e-01 -5.06251156e-01 1.17661345e+00 -8.15007448e-01
-7.14798748e-01 1.33725631e+00 3.57407033e-01 -5.56359351e-01
6.94099367e-01 2.68898398e-01 -3.61229718e-01 7.85349190e-01
2.88030893e-01 1.16197324e+00 2.84960032e-01 -8.94909978e-01
-3.13089132e-01 -5.97209334e-01 -4.42612082e-01 1.24379620e-01
1.71657294e-01 4.20174271e-01 -1.00563772e-01 -5.18378735e-01
4.80667472e-01 -7.88636863e-01 -1.21207178e-01 -5.12818657e-02
-9.67735127e-02 1.24144845e-01 2.79359281e-01 -1.24227381e+00
4.41146135e-01 -1.67049086e+00 -2.98399597e-01 3.88046801e-01
9.11210656e-01 2.47915059e-01 -4.12965715e-01 -3.39430958e-01
-6.48092985e-01 3.43725681e-01 -5.30226052e-01 -1.06218144e-01
-4.59525287e-01 -1.22362792e-01 7.38438845e-01 8.77357662e-01
-4.02350463e-02 9.69227731e-01 -8.25185478e-01 -6.41991138e-01
-1.88407391e-01 5.48156083e-01 -4.79963154e-01 -4.33400162e-02
5.42353451e-01 5.14894307e-01 6.77705407e-02 4.26042795e-01
8.03060472e-01 1.22847222e-01 2.37850219e-01 -1.63286433e-01
-9.71272066e-02 -8.83317590e-02 -3.12368602e-01 1.77873540e+00
-1.22519597e-01 7.01993942e-01 5.04723430e-01 -1.32962859e+00
7.93692529e-01 7.24762082e-01 1.05438316e+00 -1.05657625e+00
3.50881934e-01 4.86610711e-01 8.39558959e-01 -6.72377110e-01
-3.79599988e-01 -5.86129785e-01 6.01651967e-01 7.61973619e-01
3.61778438e-01 -3.31826031e-01 5.08140147e-01 1.92374215e-02
1.17661548e+00 -2.89029330e-01 -4.57859457e-01 -6.58837140e-01
1.03546277e-01 -2.14344129e-01 5.55279493e-01 5.58509946e-01
-5.85367978e-01 1.15610766e+00 5.27005136e-01 -5.70055366e-01
-1.47790480e+00 -1.34024060e+00 -4.58731890e-01 8.69736195e-01
-5.03302693e-01 5.02846420e-01 -1.26130080e+00 -7.02702522e-01
-2.53917366e-01 4.88651246e-01 -8.31956148e-01 -3.45782712e-02
-8.76710653e-01 -9.44681406e-01 1.04924047e+00 9.28417087e-01
3.05762947e-01 -1.00140333e+00 -6.74081087e-01 2.47168615e-01
-1.42434567e-01 -9.13920641e-01 -6.00376427e-01 3.67521554e-01
-1.05392301e+00 -1.13329351e+00 -1.56759667e+00 -1.13136089e+00
8.24273109e-01 -3.34203869e-01 1.08492625e+00 2.04876214e-01
-3.68080586e-01 3.37580889e-01 -1.47118777e-01 -2.99384594e-01
-4.29665774e-01 2.26782426e-01 2.43494481e-01 -5.65729558e-01
-1.49030879e-01 -9.24822450e-01 -1.05651045e+00 3.27389717e-01
-7.80555904e-01 2.21760914e-01 5.28639019e-01 7.15140939e-01
4.70067531e-01 -5.83004057e-01 7.70251036e-01 -5.20137310e-01
4.58537698e-01 -8.23804140e-01 -2.13503256e-01 4.01570886e-01
-6.65499210e-01 -2.88507372e-01 3.94320697e-01 -6.29156113e-01
-7.74943590e-01 -4.20123369e-01 -2.19550565e-01 -2.35469803e-01
-1.75046250e-01 3.82017076e-01 4.74640250e-01 -1.69489220e-01
6.07310116e-01 -8.03754926e-02 3.80035758e-01 -4.33257431e-01
-1.02822177e-01 1.24391504e-01 8.48999977e-01 -8.40776443e-01
2.03014433e-01 2.65271544e-01 6.50820904e-04 -2.50661790e-01
-4.47959691e-01 -1.56270117e-01 -1.10884643e+00 -3.16271722e-01
1.36757505e+00 -4.13817048e-01 -2.22135708e-01 4.56798106e-01
-9.15910721e-01 -7.73444712e-01 5.68694621e-02 1.01831746e+00
-5.13664424e-01 2.25046739e-01 -6.78368568e-01 4.65866700e-02
-1.02397060e+00 -1.43993950e+00 4.02495116e-01 -1.93483829e-02
-1.96629807e-01 -1.14565730e+00 3.27190459e-01 5.46357453e-01
7.48620927e-01 4.16244626e-01 1.27025914e+00 -1.15855145e+00
2.47828901e-01 -1.69801280e-01 -6.03347301e-01 5.22877514e-01
2.53350791e-02 -1.94776863e-01 -6.40563607e-01 -4.95162398e-01
-1.09932475e-01 -3.79171222e-01 7.48072267e-01 7.29862452e-01
1.14028013e+00 1.71687186e-01 2.94794198e-02 8.70995581e-01
9.80070591e-01 6.65260196e-01 2.73311079e-01 1.16279468e-01
6.70642972e-01 7.38110185e-01 -1.80658698e-01 -1.18798755e-01
4.30654645e-01 -9.64576844e-03 2.53720552e-01 -4.44069207e-01
-3.57663661e-01 3.23740423e-01 -2.81038463e-01 1.00111127e+00
-5.82343996e-01 3.07001650e-01 -1.70393133e+00 8.23281765e-01
-1.13833630e+00 -8.41204047e-01 1.28256055e-02 2.00403428e+00
8.76613677e-01 5.90169169e-02 3.12168479e-01 -2.47218505e-01
8.56421769e-01 -2.08136618e-01 -8.61765027e-01 -5.98632872e-01
-1.82769045e-01 6.53515756e-01 6.21248543e-01 1.60531066e-02
-7.96563327e-01 3.23018640e-01 6.99326038e+00 3.26653183e-01
-1.32877576e+00 4.64412719e-01 1.15009511e+00 -5.04375339e-01
-5.03172092e-02 -5.21293879e-01 2.24310607e-02 3.44010532e-01
1.27366424e+00 -1.02149323e-01 4.63184029e-01 3.44708234e-01
1.19860142e-01 -1.17717735e-01 -1.27766454e+00 7.22833276e-01
5.20805269e-02 -1.23005474e+00 -3.62682402e-01 -2.79243916e-01
9.21185136e-01 9.93040144e-01 -4.19080667e-02 1.47399202e-01
-6.40635937e-02 -1.48017287e+00 7.18898475e-01 5.55425525e-01
1.42575431e+00 -6.62372172e-01 7.86134064e-01 7.61022642e-02
-7.77122736e-01 2.45571077e-01 -5.84175140e-02 3.74177158e-01
-7.24624610e-03 2.36442745e-01 -1.10022044e+00 3.98484021e-02
1.02214825e+00 2.44704232e-01 -4.89892095e-01 1.41264307e+00
4.48010802e-01 6.95214331e-01 -4.89986330e-01 3.16384405e-01
3.16299230e-01 -6.04172759e-02 2.81442970e-01 1.39348328e+00
3.16885859e-01 4.74699885e-01 -2.65185475e-01 8.41421247e-01
-1.86140999e-01 2.80584008e-01 -2.69595355e-01 2.34215334e-01
1.58952966e-01 1.24564064e+00 -1.19631040e+00 -4.28162545e-01
-5.82654297e-01 2.78258562e-01 3.53099138e-01 3.12398523e-01
-7.88088679e-01 -2.99870670e-01 1.32053494e-01 3.10884506e-01
3.80799137e-02 -4.32266146e-02 -6.84768617e-01 -8.33477437e-01
-2.00888529e-01 -6.79081202e-01 3.08306813e-01 -7.75477886e-01
-1.20996487e+00 7.42835641e-01 3.29342991e-01 -5.62049270e-01
5.61103560e-02 -5.33351183e-01 -8.31265152e-01 1.04068720e+00
-9.84572291e-01 -7.07738936e-01 -1.79931790e-01 4.02592301e-01
2.19632417e-01 -1.41782507e-01 5.13718843e-01 6.37854695e-01
-6.84736073e-01 8.96040797e-01 -7.03411922e-02 5.14034808e-01
4.44091141e-01 -1.15410721e+00 3.64151955e-01 6.73647225e-01
-4.59627390e-01 7.99962640e-01 1.32006958e-01 -7.25380778e-01
-8.13265145e-01 -9.59851205e-01 3.17658961e-01 -3.31852282e-03
5.59229493e-01 1.81876063e-01 -1.06006038e+00 7.46180117e-01
3.45584571e-01 5.97599111e-02 7.37001538e-01 -3.45870763e-01
-3.15936804e-01 7.14086592e-02 -1.65330446e+00 2.43593156e-01
1.07054555e+00 -2.27183804e-01 -6.84182227e-01 2.47710913e-01
3.99586320e-01 -6.10659182e-01 -1.63954675e+00 8.23920250e-01
7.32816517e-01 -8.51165771e-01 7.83287704e-01 -5.10295093e-01
4.70943749e-01 3.09780061e-01 2.25775689e-01 -1.19449091e+00
-1.69900507e-01 -2.45982558e-01 2.87646294e-01 8.75188291e-01
9.02303278e-01 -9.19189811e-01 7.85813212e-01 1.05601370e+00
-5.97171783e-01 -1.20735073e+00 -1.24884737e+00 -4.18001652e-01
9.94871914e-01 -5.72090268e-01 5.14228940e-01 9.42110956e-01
-3.28697786e-02 -1.62427247e-01 4.51572686e-01 -3.04168195e-01
7.28209019e-01 -4.27791685e-01 -2.39619896e-01 -9.56571460e-01
3.44106108e-01 -1.00848675e+00 -3.53908181e-01 -2.97097504e-01
1.89781740e-01 -1.56889355e+00 1.13178752e-02 -1.77524853e+00
5.94481826e-01 -7.05528498e-01 -5.33319294e-01 7.23018289e-01
1.35536015e-01 5.11885822e-01 -7.39709809e-02 -1.78705394e-01
-1.55513138e-01 -3.38619314e-02 1.56666446e+00 -1.78639129e-01
1.43478200e-01 -1.08415313e-01 -5.07275522e-01 7.05922008e-01
1.02451921e+00 -2.56437063e-01 -4.54898357e-01 -8.17623436e-01
-2.07940295e-01 2.95682102e-01 1.68948010e-01 -1.09243488e+00
4.20743898e-02 1.45946980e-01 8.65810931e-01 -6.00506186e-01
-2.01804899e-02 -4.36103016e-01 4.74482141e-02 4.95890886e-01
-2.73133516e-01 4.58339244e-01 3.46006423e-01 -5.68076074e-01
2.31095850e-01 -1.39616251e-01 1.19005859e+00 -2.66386896e-01
-1.34670466e-01 8.17497730e-01 -5.46440303e-01 6.03859067e-01
7.96777904e-01 -4.58008617e-01 -2.95221597e-01 -6.96469098e-02
-1.12725747e+00 4.10850227e-01 3.56130928e-01 2.86540896e-01
7.05318689e-01 -9.24246728e-01 -9.38654065e-01 7.30624869e-02
-4.93989170e-01 1.24767810e-01 4.92057085e-01 1.69010830e+00
-1.02059388e+00 2.80546069e-01 -6.46555126e-01 -7.22811103e-01
-1.08228469e+00 1.94322884e-01 8.26478899e-01 1.32807568e-01
-7.90499926e-01 1.09508216e+00 1.74519241e-01 -4.22533154e-01
8.90505016e-02 -6.56696081e-01 -1.22581735e-01 1.60311088e-01
3.84906352e-01 6.64556086e-01 2.82380879e-01 -8.16868663e-01
-4.36777353e-01 6.39230013e-01 -1.40781701e-01 -1.29398197e-01
1.63697278e+00 5.97353242e-02 -5.22297084e-01 -7.38013014e-02
1.60086238e+00 -6.72572196e-01 -9.65323687e-01 1.30004331e-01
-1.35702193e-02 3.39031726e-01 1.07664894e-02 -1.03587615e+00
-1.70783389e+00 8.42487097e-01 9.42883611e-01 7.59123415e-02
1.03032970e+00 9.51843895e-03 1.00644374e+00 -1.56463176e-01
2.31223240e-01 -9.22454596e-01 -4.36398149e-01 4.56701815e-01
8.88013303e-01 -9.41912413e-01 -8.46993029e-02 2.88461566e-01
-5.76555789e-01 1.12504768e+00 7.80067027e-01 5.09950146e-03
5.67393899e-01 4.85299140e-01 3.08541745e-01 -4.40992028e-01
-4.78593677e-01 3.28536421e-01 5.75536966e-01 7.92232096e-01
6.22557163e-01 1.39664486e-01 -4.12104726e-01 4.66066927e-01
-3.73941571e-01 -2.88856715e-01 2.40337625e-01 6.60426140e-01
-2.82107055e-01 -5.87952435e-01 -1.91652566e-01 1.21684599e+00
-9.56631422e-01 -2.75743157e-01 -9.24971327e-02 5.92503726e-01
4.45823282e-01 5.55516899e-01 1.97541863e-01 -3.17112282e-02
2.12417424e-01 2.02986836e-01 8.15053463e-01 -5.52082419e-01
-8.21958542e-01 -1.70859650e-01 2.43421331e-01 -5.77813625e-01
-3.41971129e-01 -1.03241003e+00 -1.66176414e+00 -2.63051182e-01
-8.01276937e-02 8.76312852e-02 8.88467729e-01 9.44429576e-01
-9.01460201e-02 4.44195837e-01 3.23264271e-01 -1.16474617e+00
4.34248820e-02 -1.19728005e+00 -4.51055259e-01 1.21416427e-01
3.41393709e-01 -7.03862607e-01 -1.47191420e-01 -2.01089144e-01]
|
[14.159113883972168, -2.3087477684020996]
|
141c2ac6-78f7-4f88-96ce-fed4723ca0b5
|
revisiting-the-stack-based-inverse-tone
| null | null |
http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_Revisiting_the_Stack-Based_Inverse_Tone_Mapping_CVPR_2023_paper.html
|
http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_Revisiting_the_Stack-Based_Inverse_Tone_Mapping_CVPR_2023_paper.pdf
|
Revisiting the Stack-Based Inverse Tone Mapping
|
Current stack-based inverse tone mapping (ITM) methods can recover high dynamic range (HDR) radiance by predicting a set of multi-exposure images from a single low dynamic range image. However, there are still some limitations. On the one hand, these methods estimate a fixed number of images (e.g., three exposure-up and three exposure-down), which may introduce unnecessary computational cost or reconstruct incorrect results. On the other hand, they neglect the connections between the up-exposure and down-exposure models and thus fail to fully excavate effective features. In this paper, we revisit the stack-based ITM approaches and propose a novel method to reconstruct HDR radiance from a single image, which only needs to estimate two exposure images. At first, we design the exposure adaptive block that can adaptively adjust the exposure based on the luminance distribution of the input image. Secondly, we devise the cross-model attention block to connect the exposure adjustment models. Thirdly, we propose an end-to-end ITM pipeline by incorporating the multi-exposure fusion model. Furthermore, we propose and open a multi-exposure dataset that indicates the optimal exposure-up/down levels. Experimental results show that the proposed method outperforms some state-of-the-art methods.
|
['Ronggang Wang', 'Yang Zhao', 'Yuyao Ye', 'Ning Zhang']
|
2023-01-01
| null | null | null |
cvpr-2023-1
|
['tone-mapping', 'inverse-tone-mapping']
|
['computer-vision', 'computer-vision']
|
[ 6.52369559e-01 -5.43597043e-01 3.02277267e-01 -6.27085626e-01
-7.45155573e-01 -1.31811798e-01 3.75056952e-01 -5.31633735e-01
-2.44941249e-01 3.32459539e-01 -3.76994610e-02 -1.51558012e-01
-2.16731325e-01 -9.92710233e-01 -7.12922096e-01 -6.66335344e-01
4.81930614e-01 -1.64109379e-01 5.61554193e-01 -3.77990603e-01
3.13873172e-01 1.59170926e-01 -1.74647307e+00 1.77907079e-01
1.09982765e+00 1.25312853e+00 8.01676750e-01 6.89785182e-01
-7.89791197e-02 7.85081923e-01 -3.90388191e-01 -1.10944763e-01
3.48903477e-01 -6.55288935e-01 -3.90428782e-01 1.11175634e-01
5.87946534e-01 -9.07679021e-01 -3.90833765e-01 1.04097569e+00
6.53351247e-01 1.32222146e-01 1.93337962e-01 -9.77072954e-01
-8.75439167e-01 2.26358041e-01 -1.00150561e+00 2.19781980e-01
1.73220113e-01 1.94429502e-01 4.74000961e-01 -9.04479861e-01
2.36288428e-01 1.04858708e+00 6.75575614e-01 1.92840427e-01
-1.18492448e+00 -7.64864981e-01 1.35444224e-01 4.27247941e-01
-1.24331522e+00 -4.58532900e-01 9.53094482e-01 -9.42787752e-02
6.13908589e-01 5.38184047e-01 7.21500039e-01 4.79060382e-01
2.38705114e-01 3.65305960e-01 1.72554076e+00 -5.29671848e-01
-6.43913522e-02 -2.03604355e-01 -4.38322015e-02 5.87252617e-01
-2.96914160e-01 3.28725398e-01 -5.38726091e-01 3.06843221e-01
8.39147925e-01 -1.45909442e-02 -6.20507121e-01 1.18513629e-01
-1.17135465e+00 4.77602929e-01 5.15251398e-01 6.64186776e-02
-1.83972359e-01 1.21511169e-01 -1.19869635e-01 3.07945818e-01
5.78773379e-01 1.38588948e-02 -4.60874617e-01 2.01163009e-01
-9.76960957e-01 -8.90557691e-02 1.34627551e-01 8.40624392e-01
1.13887429e+00 -1.94753468e-01 1.36684449e-02 9.38604176e-01
5.30288398e-01 6.27597690e-01 2.26681948e-01 -8.94402504e-01
5.22999525e-01 1.85233861e-01 6.06843904e-02 -9.44423318e-01
-3.25986952e-01 -1.87370270e-01 -8.33630562e-01 3.32896948e-01
4.75450270e-02 1.93792716e-01 -1.00804305e+00 1.60362697e+00
3.61425996e-01 4.58384037e-01 -2.10749164e-01 1.26553094e+00
7.74088502e-01 8.99734259e-01 -1.47665799e-01 -5.04494011e-01
1.40432930e+00 -1.18920743e+00 -1.00752687e+00 -2.48743430e-01
-1.35949641e-01 -1.03146935e+00 1.41408980e+00 5.47719002e-01
-1.32134545e+00 -1.01711392e+00 -1.33578861e+00 -4.47120667e-01
-1.64472267e-01 -5.16695194e-02 4.75657761e-01 6.07803822e-01
-1.03136075e+00 5.43506861e-01 -4.86060053e-01 9.49777216e-02
6.57970980e-02 3.49504687e-02 2.32554197e-01 -3.36155951e-01
-1.40790367e+00 8.72023582e-01 5.70022091e-02 4.38511372e-01
-5.30246854e-01 -9.70758259e-01 -6.91653848e-01 -1.70122206e-01
3.03474605e-01 -6.35175109e-01 1.02595377e+00 -9.42870438e-01
-1.71273661e+00 5.58465004e-01 -1.01667725e-01 -1.59727875e-02
4.14631069e-01 -1.80760935e-01 -6.40038669e-01 2.31760353e-01
-1.26536056e-01 5.22350073e-01 8.05224061e-01 -1.38451803e+00
-6.39547527e-01 -1.43927574e-01 6.73169419e-02 3.69484037e-01
-2.54586428e-01 -9.03158262e-03 -8.53282809e-01 -6.88206732e-01
5.59968114e-01 -6.07560575e-01 -1.23062983e-01 3.02642345e-01
-2.40527630e-01 4.82529551e-01 8.96678805e-01 -7.39593446e-01
1.42900550e+00 -2.10312963e+00 -8.68524089e-02 -4.76148389e-02
2.33671553e-02 6.26913682e-02 -1.45881891e-01 8.24533328e-02
-1.12892501e-01 -8.15948024e-02 -4.24476564e-01 -3.88344437e-01
-1.13647990e-01 -1.22305349e-01 -5.54150462e-01 5.26396692e-01
-1.49920091e-01 6.52259529e-01 -6.55970156e-01 -5.41234970e-01
8.50346446e-01 7.73997426e-01 -1.75102085e-01 2.96698272e-01
-1.41083166e-01 6.59961998e-01 -1.29675195e-01 4.18473899e-01
1.31265032e+00 -1.93406656e-01 -7.43173510e-02 -6.85510933e-01
-5.39641678e-01 9.34873819e-02 -1.31278968e+00 1.87322807e+00
-8.16671550e-01 6.87668741e-01 -3.04580629e-01 -2.28744581e-01
9.77410316e-01 -7.03799427e-02 4.14914519e-01 -1.28412974e+00
3.06712389e-02 2.19938755e-01 -2.84152985e-01 -5.15776217e-01
6.26982868e-01 -3.33344102e-01 1.13408387e-01 5.30490756e-01
-4.26460743e-01 -1.93214014e-01 -1.69967473e-01 -2.32806042e-01
4.92868364e-01 4.04794872e-01 2.53623515e-01 -3.10693239e-03
8.11101615e-01 -4.56316054e-01 5.99458218e-01 4.80802834e-01
-7.28770122e-02 1.12637770e+00 -3.34176049e-02 -3.88464212e-01
-9.45760846e-01 -1.10161734e+00 -3.41640294e-01 9.93576527e-01
8.55417490e-01 -1.91496491e-01 -7.87440896e-01 -5.98979779e-02
-5.30001819e-01 5.76119542e-01 -3.89433414e-01 -4.39567864e-02
-7.15613306e-01 -9.52054143e-01 2.06994385e-01 3.23716015e-01
1.21396148e+00 -8.82603288e-01 -9.30551887e-01 3.33979279e-02
-5.05326331e-01 -9.74545777e-01 -8.58390033e-01 -2.19750732e-01
-5.50440907e-01 -8.67986858e-01 -4.69759971e-01 -6.25415266e-01
4.36122030e-01 6.36595428e-01 1.00320613e+00 2.63278395e-01
-2.88482189e-01 -4.54907380e-02 -2.52495289e-01 -1.46808609e-01
-1.06979921e-01 -3.07535768e-01 -4.13316399e-01 1.85097948e-01
9.63628292e-03 -6.38678432e-01 -1.15352488e+00 6.22335017e-01
-1.05605960e+00 7.71245480e-01 5.41670144e-01 4.89212841e-01
1.08060527e+00 3.06409836e-01 4.13097799e-01 -6.52724504e-01
1.53152615e-01 -1.78216800e-01 -7.49701619e-01 5.18326938e-01
-7.81005681e-01 -2.59953409e-01 5.85549533e-01 -5.29184043e-01
-1.53096771e+00 9.94019210e-02 -1.51553586e-01 -4.47513461e-01
9.67251062e-02 1.35213837e-01 -5.71635187e-01 -2.58767039e-01
3.25059354e-01 3.68424505e-01 -3.39436501e-01 -5.48851132e-01
5.51818192e-01 6.97327495e-01 7.44152784e-01 -1.35119960e-01
8.10581803e-01 6.08747065e-01 2.66145132e-02 -5.74768007e-01
-8.39479208e-01 -3.61647159e-02 -3.87645125e-01 -5.79591691e-01
8.17315400e-01 -9.97641385e-01 -5.33731699e-01 8.51526320e-01
-8.39866519e-01 -5.64665139e-01 -5.83834089e-02 2.99272031e-01
-6.41116142e-01 4.18310732e-01 -6.36340737e-01 -5.44611394e-01
-3.58106822e-01 -1.22188592e+00 1.12525499e+00 6.03743136e-01
2.81412303e-01 -5.97699225e-01 -8.48796591e-02 2.68079996e-01
7.59310186e-01 1.40274048e-01 7.16837823e-01 5.71960032e-01
-9.95577812e-01 3.82954955e-01 -4.86881137e-01 1.14958724e-02
1.98217928e-01 -1.59127757e-01 -1.21657646e+00 -1.99612319e-01
2.99279988e-01 5.84753277e-03 9.07160342e-01 5.60214639e-01
1.48927081e+00 8.43810961e-02 -3.27289738e-02 1.18247652e+00
1.82241416e+00 1.72952250e-01 1.28092146e+00 6.35289133e-01
5.97609222e-01 5.27521670e-01 1.02712524e+00 2.79056191e-01
5.62615275e-01 8.36182237e-01 3.11309874e-01 -4.92907077e-01
-6.37007713e-01 -3.55947375e-01 1.46577626e-01 6.20319843e-01
2.12899491e-01 -3.80621076e-01 -3.10197771e-01 1.43833086e-01
-1.46887648e+00 -7.04973817e-01 -5.04117608e-01 2.24654341e+00
9.04585660e-01 -2.63263863e-02 -2.95634389e-01 2.27526009e-01
7.83358455e-01 5.45788944e-01 -5.93427420e-01 -1.41912326e-01
-3.33960742e-01 2.79840082e-01 5.18749952e-01 7.63057411e-01
-8.99211407e-01 6.27366066e-01 5.80468845e+00 7.63963997e-01
-1.35115159e+00 2.33957663e-01 6.79505408e-01 -7.79063031e-02
-6.75614715e-01 2.43023932e-01 -6.57119811e-01 8.34299982e-01
7.43148029e-01 1.36615545e-01 7.40336895e-01 1.93823844e-01
2.85952538e-01 -4.62511510e-01 -7.60560334e-01 1.13848984e+00
2.76013285e-01 -8.65304112e-01 -2.91476458e-01 -6.84590191e-02
7.49498665e-01 -3.05054277e-01 2.69089788e-01 -5.07298671e-02
-3.46919507e-01 -6.33347750e-01 9.56099570e-01 9.70628083e-01
1.24574506e+00 -7.23397255e-01 2.11918280e-01 2.51003742e-01
-1.40437388e+00 2.18683854e-02 -3.63059521e-01 2.16659337e-01
5.61674833e-01 7.82411814e-01 3.38834189e-02 5.67842484e-01
9.18520212e-01 4.18488890e-01 -6.54079676e-01 8.32551539e-01
-2.94929326e-01 1.70127243e-01 -2.07453728e-01 5.19823849e-01
-2.75266707e-01 -3.80455434e-01 1.13613501e-01 8.47016871e-01
4.88337070e-01 5.75378180e-01 -6.31045699e-02 9.37187791e-01
4.69419993e-02 -2.76771665e-01 -3.30212973e-02 6.91196680e-01
4.58960384e-01 1.24618137e+00 -5.61811984e-01 -9.25275460e-02
-5.21255076e-01 1.20855629e+00 -1.35705143e-01 3.42264235e-01
-1.22771251e+00 -5.42563200e-01 2.88447708e-01 1.64673716e-01
3.16782236e-01 -3.37905362e-02 -4.04091418e-01 -9.55346227e-01
2.28817776e-01 -6.34622574e-01 2.43625231e-02 -1.40214288e+00
-1.08539307e+00 5.81546783e-01 -5.20453453e-02 -1.45103550e+00
2.13272706e-01 -9.68839601e-02 -4.60896224e-01 1.10390067e+00
-2.11151814e+00 -1.11789382e+00 -8.29214394e-01 3.65740240e-01
7.69161344e-01 4.28353906e-01 4.11002189e-01 5.88183939e-01
-5.31047583e-01 3.31909478e-01 -1.76863167e-02 -3.57430905e-01
8.67916584e-01 -9.58435297e-01 4.23024207e-01 1.03753865e+00
-2.11515352e-01 4.06952471e-01 6.71240389e-01 -4.84346867e-01
-1.30441582e+00 -1.05683100e+00 5.90971470e-01 -1.01807207e-01
1.23209476e-01 -2.70342380e-01 -1.11381102e+00 3.77907395e-01
7.76692480e-02 1.57705560e-01 3.93420696e-01 -4.97788221e-01
-2.00427443e-01 -5.17274439e-01 -1.21076965e+00 4.83693779e-01
9.85923529e-01 -4.86680120e-01 -1.64337620e-01 -1.51711613e-01
9.25505638e-01 -6.42926574e-01 -9.71309304e-01 4.64034528e-01
6.87812090e-01 -1.52089810e+00 1.27349293e+00 5.56767225e-01
5.19710779e-01 -8.54197323e-01 -3.99970084e-01 -9.40895259e-01
-3.24357480e-01 -4.26252156e-01 -2.00207159e-01 1.48270690e+00
6.42694831e-02 -6.96557224e-01 2.17259333e-01 5.66193640e-01
-1.26185730e-01 -8.78951609e-01 -5.41034460e-01 -3.49516541e-01
-3.47362995e-01 -4.33284551e-01 1.06807482e+00 7.93638289e-01
-6.80387318e-01 1.91742197e-01 -7.48867512e-01 4.28277344e-01
9.15094018e-01 6.39846087e-01 5.32264292e-01 -7.67267823e-01
-4.54716831e-01 -1.62751287e-01 2.07049549e-01 -1.32106054e+00
-2.01496184e-01 -2.76742518e-01 2.98340350e-01 -1.48910344e+00
3.00138235e-01 -6.39470100e-01 -3.64506394e-01 2.33822167e-02
-4.77735072e-01 5.90490162e-01 7.05098808e-02 3.25541884e-01
-4.20271844e-01 5.36355257e-01 1.34573662e+00 1.08122341e-01
-2.31105462e-01 -3.17192614e-01 -8.45653594e-01 6.43305123e-01
7.64587581e-01 -2.71034509e-01 -6.59018517e-01 -8.65752578e-01
2.02054456e-01 2.39202932e-01 4.98450428e-01 -9.41053748e-01
2.47382194e-01 -3.78747791e-01 7.43685067e-01 -9.96524692e-01
3.74812454e-01 -7.40536630e-01 4.65770781e-01 1.44307718e-01
-2.12715372e-01 -1.04841128e-01 -1.65541284e-02 5.47020793e-01
3.92192677e-02 -1.38978381e-02 1.16232240e+00 -3.58170383e-02
-1.04023468e+00 3.35962713e-01 1.31483391e-01 -3.49929005e-01
8.94933879e-01 -3.33011001e-01 -4.84598666e-01 -1.34539843e-01
-1.15007661e-01 1.76534891e-01 7.91291416e-01 3.36831838e-01
8.59393120e-01 -1.31956708e+00 -5.62393665e-01 4.40299213e-01
-9.81391817e-02 1.33339807e-01 1.00554240e+00 7.34221995e-01
-5.35986185e-01 -1.46984279e-01 -1.49289995e-01 -5.95597923e-01
-1.19112885e+00 7.01938033e-01 5.24086356e-01 9.36109722e-02
-8.13360155e-01 5.86100519e-01 4.02780831e-01 -1.36367083e-01
-1.43470451e-01 -1.13732524e-01 -3.47245187e-02 -2.67365277e-01
8.63552332e-01 4.23159212e-01 -4.63933088e-02 -5.66448569e-01
-1.37414217e-01 1.23771238e+00 1.57696396e-01 -2.47478083e-01
1.23625064e+00 -8.48888695e-01 -2.38477588e-01 4.78849083e-01
1.30292761e+00 -1.93808809e-01 -1.58864403e+00 -2.69482166e-01
-5.37784696e-01 -1.15094662e+00 4.80312318e-01 -8.88041258e-01
-1.25034821e+00 1.00360668e+00 1.09752882e+00 -1.50298864e-01
2.09811735e+00 -3.44959229e-01 1.10727596e+00 -1.27795339e-01
3.37031454e-01 -1.06549072e+00 4.12055142e-02 -4.45960043e-03
8.02472770e-01 -1.10249531e+00 3.97718549e-01 -5.81530690e-01
-3.30043912e-01 1.04938865e+00 7.68115401e-01 2.77534902e-01
5.20061016e-01 3.12786311e-01 1.12558402e-01 -5.55345044e-03
-6.48629665e-01 -1.25351340e-01 1.60303369e-01 4.76167738e-01
3.53268892e-01 -2.14858115e-01 -2.68930197e-01 3.36861998e-01
-3.22657116e-02 1.02874152e-01 5.38577437e-01 6.06503904e-01
-5.82934678e-01 -8.62790644e-01 -5.60720861e-01 1.52757093e-01
-2.41051823e-01 -2.06571028e-01 2.22417980e-01 5.40671349e-01
3.95749152e-01 1.05187416e+00 1.27721921e-01 -4.89825100e-01
4.87995386e-01 -4.53019202e-01 6.02748454e-01 -2.59034336e-02
-2.76260257e-01 3.04836333e-01 -3.46386015e-01 -6.01721764e-01
-6.28376842e-01 -4.23864573e-01 -1.09077477e+00 -3.23817968e-01
-5.20679355e-01 -4.75822508e-01 7.26424396e-01 6.09380424e-01
2.29519606e-01 7.76207924e-01 1.08584368e+00 -7.99556375e-01
-1.22477822e-01 -9.27222848e-01 -5.43351114e-01 1.12255305e-01
4.99346524e-01 -4.41068143e-01 -3.33730519e-01 1.58938140e-01]
|
[10.92519760131836, -2.15541672706604]
|
8cc1478e-f7cd-4d52-bbd5-28d4f1921317
|
higher-order-generalization-bounds-learning
|
2203.15972
| null |
https://arxiv.org/abs/2203.15972v1
|
https://arxiv.org/pdf/2203.15972v1.pdf
|
Higher-Order Generalization Bounds: Learning Deep Probabilistic Programs via PAC-Bayes Objectives
|
Deep Probabilistic Programming (DPP) allows powerful models based on recursive computation to be learned using efficient deep-learning optimization techniques. Additionally, DPP offers a unified perspective, where inference and learning algorithms are treated on a par with models as stochastic programs. Here, we offer a framework for representing and learning flexible PAC-Bayes bounds as stochastic programs using DPP-based methods. In particular, we show that DPP techniques may be leveraged to derive generalization bounds that draw on the compositionality of DPP representations. In turn, the bounds we introduce offer principled training objectives for higher-order probabilistic programs. We offer a definition of a higher-order generalization bound, which naturally encompasses single- and multi-task generalization perspectives (including transfer- and meta-learning) and a novel class of bound based on a learned measure of model complexity. Further, we show how modified forms of all higher-order bounds can be efficiently optimized as objectives for DPP training, using variational techniques. We test our framework using single- and multi-task generalization settings on synthetic and biological data, showing improved performance and generalization prediction using flexible DPP model representations and learned complexity measures.
|
['Mark Gerstein', 'Jonathan Warrell']
|
2022-03-30
| null | null | null | null |
['probabilistic-programming']
|
['methodology']
|
[ 1.41491309e-01 3.46956044e-01 -5.95752180e-01 -4.25357372e-01
-1.49593222e+00 -7.81575799e-01 6.67878747e-01 -4.91224276e-03
-7.71030113e-02 8.46368670e-01 1.18424684e-01 -1.15769327e-01
-5.16816735e-01 -7.39945769e-01 -1.16673529e+00 -9.99677420e-01
-2.43283972e-01 7.88014352e-01 1.65325031e-01 1.15498729e-01
2.07438290e-01 3.93511534e-01 -1.54882050e+00 4.72411454e-01
8.30025256e-01 8.60124171e-01 -8.83312225e-02 8.62996817e-01
9.35016107e-03 2.10461125e-01 -2.06883043e-01 -6.99571908e-01
1.64769426e-01 -1.42418981e-01 -7.84736753e-01 -3.56948376e-01
4.32132781e-01 -6.58882633e-02 -1.22527875e-01 9.25834775e-01
3.83534849e-01 2.98254997e-01 1.04974759e+00 -1.30011868e+00
-7.87226915e-01 7.96043575e-01 -1.67458579e-01 1.07532516e-01
1.55755997e-01 6.69937953e-02 1.48147452e+00 -5.87558389e-01
3.81615222e-01 1.50811005e+00 9.59113359e-01 8.22950423e-01
-1.96812332e+00 -2.55249321e-01 1.95729971e-01 -1.33433953e-01
-9.78486001e-01 -1.48394257e-01 6.04439497e-01 -6.81611896e-01
7.13225901e-01 3.62214863e-01 4.48985577e-01 1.38794780e+00
2.85116404e-01 1.14415741e+00 9.47476447e-01 -3.58168006e-01
4.94585276e-01 1.44450158e-01 4.49596465e-01 9.19327319e-01
2.91107386e-01 2.91349888e-01 -6.59448445e-01 -6.70932293e-01
4.92923856e-01 -2.10251555e-01 -2.56761938e-01 -6.42493010e-01
-9.00744677e-01 1.28069413e+00 3.83681688e-03 -3.02559078e-01
-2.81841513e-02 6.11258268e-01 6.34878457e-01 -1.96651265e-01
5.63028574e-01 3.72378379e-01 -8.14873159e-01 -2.68973708e-01
-9.99957144e-01 7.46442556e-01 1.15588880e+00 7.93663025e-01
7.31150270e-01 2.31188722e-02 -4.66169178e-01 8.53201807e-01
5.03584802e-01 5.17311037e-01 1.41557232e-01 -1.41121471e+00
1.56140774e-01 -7.43032098e-02 1.59000069e-01 -5.46982467e-01
-2.31872156e-01 -5.17781556e-01 -4.32802975e-01 1.55897692e-01
5.50528944e-01 3.95237422e-03 -6.68386161e-01 2.22330523e+00
3.03052008e-01 9.91285145e-02 -3.10055595e-02 4.01854068e-01
3.09412241e-01 8.98831606e-01 1.30454287e-01 -2.26559415e-01
1.07589841e+00 -7.22942233e-01 -1.78162858e-01 2.86813110e-01
7.17857182e-01 -1.03433654e-02 1.32171977e+00 5.01885593e-01
-1.27729177e+00 -2.27439284e-01 -9.59788740e-01 -1.47924066e-01
-3.32581043e-01 -2.19606563e-01 8.36116016e-01 9.67500091e-01
-9.53903854e-01 1.13916647e+00 -1.12171507e+00 -5.82577214e-02
7.48417795e-01 2.85167903e-01 2.54777312e-01 2.66482204e-01
-8.88687789e-01 6.50047719e-01 4.57035005e-01 -7.24350363e-02
-1.30864692e+00 -1.32567942e+00 -7.99304783e-01 2.99942136e-01
6.27355874e-02 -1.00622809e+00 1.33972943e+00 -4.68707442e-01
-1.93996871e+00 9.06931520e-01 -1.66899279e-01 -6.70177341e-01
6.61867499e-01 -1.70822337e-01 2.40491033e-01 8.03853646e-02
-2.39450887e-01 5.14623284e-01 6.49122834e-01 -1.13261616e+00
-2.83996910e-01 -3.86901766e-01 4.30528104e-01 -7.26176873e-02
-7.82936215e-02 -1.37636244e-01 9.42696258e-02 -3.03360790e-01
-1.50994077e-01 -9.97985601e-01 -9.41073298e-02 2.49813482e-01
-3.96573991e-01 -5.73632419e-01 3.76855284e-01 -3.49368155e-01
9.92111385e-01 -1.77576411e+00 6.79535866e-01 5.81585467e-02
1.03478402e-01 -9.85081047e-02 6.88884333e-02 3.01269919e-01
3.10822666e-01 5.28938711e-01 -6.42534494e-01 -4.97734547e-01
8.30132484e-01 5.48851669e-01 -6.39467180e-01 4.04676139e-01
2.40041167e-01 9.27997231e-01 -7.46094406e-01 -5.03298402e-01
-1.20857365e-01 3.80897254e-01 -1.16505468e+00 -4.97317985e-02
-9.33547378e-01 2.62161523e-01 -5.01369834e-01 4.34632152e-01
5.75443387e-01 -2.00697273e-01 4.06093039e-02 4.37067114e-02
1.17142260e-01 2.84544438e-01 -9.43008006e-01 1.70635724e+00
-6.25317216e-01 5.03835797e-01 -3.62356612e-03 -1.36656177e+00
4.32956249e-01 -6.13011234e-02 1.83793604e-01 1.92167744e-01
-1.24986261e-01 2.21291691e-01 -4.49881613e-01 -3.17177981e-01
2.00202182e-01 -3.28050882e-01 -2.32117906e-01 3.66164237e-01
3.75445813e-01 -4.13551450e-01 2.50657946e-01 -3.94213535e-02
5.31988800e-01 6.70970857e-01 -1.90530971e-01 -8.47917199e-01
3.44491869e-01 -1.71023354e-01 6.13179684e-01 1.04418576e+00
-2.04169661e-01 3.17193538e-01 1.01835072e+00 -4.58137542e-01
-1.22140265e+00 -1.56958878e+00 -6.79968476e-01 1.64672732e+00
-4.15620565e-01 -3.31704319e-01 -8.17477584e-01 -3.63326371e-01
2.81002134e-01 8.35874319e-01 -9.91796076e-01 -1.16434708e-01
-3.62480402e-01 -1.09448993e+00 7.67050624e-01 5.42920828e-01
-6.90120757e-02 -5.26676297e-01 -5.87829173e-01 -4.68571223e-02
2.28082761e-01 -7.73551881e-01 -1.85015693e-01 3.78055066e-01
-1.03004766e+00 -8.15780818e-01 -8.78035426e-01 -4.19839770e-01
-2.03357413e-02 -4.61231142e-01 1.03895819e+00 -3.18162084e-01
-1.25067383e-01 6.65939093e-01 2.59208918e-01 -4.37267184e-01
-6.21811271e-01 2.46355653e-01 2.87908822e-01 -3.35027665e-01
-1.27864316e-01 -9.44058597e-01 -2.84748316e-01 7.13595655e-03
-8.00720394e-01 -6.89296573e-02 1.59788117e-01 1.17187870e+00
7.90211618e-01 -3.86845767e-01 3.92867565e-01 -9.33965981e-01
6.15425706e-01 -5.62883735e-01 -1.00196385e+00 4.39943075e-01
-5.03995478e-01 7.53843606e-01 5.57433188e-01 -5.93927324e-01
-1.07079160e+00 -2.55243748e-01 -1.12816393e-01 -3.29839796e-01
2.88607359e-01 5.88591337e-01 -1.37860298e-01 4.13319692e-02
8.70261610e-01 5.46187043e-01 -2.09721819e-01 -2.78498590e-01
6.88118696e-01 2.18467012e-01 2.64272630e-01 -1.73294759e+00
4.70174402e-01 5.47098100e-01 3.64395916e-01 -7.37835944e-01
-1.21727610e+00 2.68717855e-01 -4.99118805e-01 1.83707684e-01
8.11382532e-01 -6.14189267e-01 -1.02191448e+00 1.12527981e-01
-1.09759092e+00 -7.31507540e-01 -3.91392767e-01 3.15014958e-01
-1.14557874e+00 3.13731611e-01 -7.50684857e-01 -1.18249106e+00
-1.65282354e-01 -1.27275956e+00 1.27448130e+00 1.86433256e-01
3.73335481e-02 -1.28232670e+00 3.60995919e-01 4.15574163e-02
2.73241699e-01 4.33842301e-01 1.35085201e+00 -7.87466764e-01
-5.33736408e-01 1.80560231e-01 -4.88523990e-02 3.43091100e-01
-6.66145563e-01 5.45627356e-01 -1.16532242e+00 1.41363675e-02
-1.49947062e-01 -5.58787405e-01 1.13540113e+00 8.81703436e-01
1.51950979e+00 -4.45589781e-01 -3.78386915e-01 1.07932973e+00
1.37271750e+00 -3.06119382e-01 3.21308136e-01 5.60748503e-02
5.23944914e-01 4.80284095e-01 -2.29931787e-01 4.48413372e-01
2.80031055e-01 5.28784275e-01 8.46768767e-02 6.88192070e-01
3.92736405e-01 -3.80430460e-01 3.34483087e-01 4.40778762e-01
-8.40466991e-02 2.49961808e-01 -9.04910922e-01 2.37080574e-01
-1.84634686e+00 -1.22997487e+00 1.26879692e-01 2.23738384e+00
1.16458797e+00 3.38114560e-01 4.27488238e-01 -1.88949302e-01
5.64699948e-01 1.35430664e-01 -9.18137789e-01 -6.97537065e-01
-1.43243894e-01 3.70827854e-01 3.19044232e-01 7.28969932e-01
-9.85564709e-01 6.65021122e-01 7.52406931e+00 1.05380726e+00
-7.50822246e-01 2.93298900e-01 8.83766949e-01 -3.07218313e-01
-8.32094967e-01 -6.44270778e-02 -1.06095123e+00 3.88122082e-01
1.35748470e+00 -4.65007156e-01 6.02793574e-01 1.29275250e+00
-2.48934090e-01 2.27117643e-01 -1.68342960e+00 9.09430385e-01
-3.87536615e-01 -1.74053550e+00 8.85839909e-02 2.70445228e-01
8.94202650e-01 1.57440305e-01 3.92670482e-01 7.40780056e-01
6.01651907e-01 -9.41885829e-01 9.10378575e-01 5.57962656e-01
3.81974936e-01 -8.64912748e-01 2.34050110e-01 4.13955659e-01
-7.30301440e-01 -2.56430089e-01 -6.69255733e-01 9.78861824e-02
-1.48989871e-01 7.00194240e-01 -8.73712078e-02 2.66658992e-01
5.09899020e-01 4.32842493e-01 1.08538002e-01 8.55160892e-01
-1.10636696e-01 6.67335510e-01 -6.65651977e-01 -4.65709180e-01
6.10954203e-02 -2.71004051e-01 6.86929941e-01 1.57246304e+00
1.02685042e-01 -1.00231558e-01 -3.14563923e-02 1.63219821e+00
-1.91672727e-01 -3.23026091e-01 -4.62528825e-01 -6.13593720e-02
2.75078028e-01 8.42554271e-01 -3.78675818e-01 -2.36463711e-01
9.38208401e-02 4.86647815e-01 6.67407930e-01 5.41632831e-01
-1.17107379e+00 -4.80997637e-02 9.92671609e-01 -4.30940002e-01
4.98094916e-01 -4.79932159e-01 -5.17915606e-01 -1.37912250e+00
-3.21844742e-02 -5.24376690e-01 4.57138538e-01 -4.64389503e-01
-1.40154445e+00 9.25732851e-02 5.24701715e-01 -4.83198464e-01
-3.80396128e-01 -1.12036479e+00 -6.09868884e-01 7.45664477e-01
-1.29118264e+00 -9.81729209e-01 4.07297730e-01 1.40804142e-01
3.92358810e-01 2.48932809e-01 9.82780576e-01 -2.07604155e-01
-8.36956263e-01 8.89051080e-01 5.30582607e-01 -1.83563456e-01
-8.71990919e-02 -1.45637214e+00 -1.24888569e-01 5.72360516e-01
3.16464543e-01 6.03228271e-01 8.88274074e-01 -7.42098019e-02
-1.65506625e+00 -8.02452981e-01 1.83105722e-01 -7.60662675e-01
9.83779788e-01 -6.41791821e-01 -7.80702710e-01 7.36572921e-01
-3.20501685e-01 4.85245064e-02 9.23355758e-01 6.35406375e-01
-7.09849954e-01 -1.71492733e-02 -1.25996983e+00 5.81770539e-01
1.06700683e+00 -8.50293994e-01 -5.00852764e-01 7.78131187e-01
1.05771768e+00 -5.13640642e-01 -1.04577947e+00 2.14162186e-01
9.29283023e-01 -9.42588031e-01 1.00348616e+00 -1.07566583e+00
7.89060652e-01 2.32162416e-01 -7.70767212e-01 -1.08196509e+00
-1.04757331e-01 -8.61164808e-01 -6.87167406e-01 9.58209932e-01
7.10589767e-01 -7.44266629e-01 8.19188237e-01 9.92560148e-01
-2.91708082e-01 -1.36834109e+00 -1.14787316e+00 -9.27874863e-01
1.02526057e+00 -8.27673256e-01 4.05579358e-01 5.58936298e-01
1.26126245e-01 -8.90560001e-02 5.00219353e-02 2.92013884e-01
9.39050376e-01 2.72315174e-01 4.86179978e-01 -1.28344202e+00
-8.19083989e-01 -8.84392560e-01 -3.91308069e-01 -1.16001427e+00
6.55388772e-01 -1.15062714e+00 6.45149648e-02 -1.14083004e+00
4.91343975e-01 -3.05489868e-01 -2.81772643e-01 1.97678819e-01
6.75775036e-02 -2.76599795e-01 -2.28311401e-02 2.62791544e-01
-4.33921069e-01 9.19614136e-01 7.88543642e-01 -9.82911438e-02
-1.54375866e-01 2.14172825e-01 -6.64535761e-01 8.16766381e-01
6.26446009e-01 -5.12454391e-01 -3.96296442e-01 -4.53174263e-01
4.75038499e-01 -8.14264826e-03 5.84299684e-01 -8.71162951e-01
-3.84782697e-03 -3.88838351e-01 2.44226575e-01 -3.08630288e-01
4.71665710e-01 -9.57281813e-02 -1.75695032e-01 4.31772530e-01
-8.18303287e-01 -5.24161279e-01 4.27108258e-01 8.06362808e-01
3.49314183e-01 -2.94002026e-01 9.95751619e-01 -4.87017371e-02
-6.81133866e-02 4.09754515e-01 -1.65654451e-01 4.61587787e-01
6.74752474e-01 -4.48114844e-03 -2.04426780e-01 -6.79545626e-02
-1.04779983e+00 1.97876513e-01 2.09346622e-01 -2.63946176e-01
3.01258683e-01 -1.06582320e+00 -8.78582597e-01 -3.04213196e-01
-9.00475383e-02 1.04063019e-01 2.38884628e-01 6.96697295e-01
-4.23033148e-01 4.51949596e-01 7.75743797e-02 -1.01293492e+00
-5.32738447e-01 5.00044525e-01 6.48717105e-01 -3.86707067e-01
-3.28165948e-01 1.14877856e+00 3.57066929e-01 -6.50867105e-01
2.46606201e-01 -7.98602879e-01 5.58898866e-01 -1.68203756e-01
4.69704807e-01 3.55537683e-01 -3.59455734e-01 -6.07132092e-02
-1.92636251e-01 4.61762905e-01 1.46847904e-01 -3.26084793e-01
1.57415986e+00 1.47093862e-01 1.50804715e-02 8.19814026e-01
1.38979590e+00 -1.77602246e-01 -1.70403767e+00 -8.28801021e-02
8.25939253e-02 -7.13983737e-03 -1.42207935e-01 -5.93564510e-01
-5.37362337e-01 1.18898094e+00 3.90426904e-01 2.13836625e-01
6.31649613e-01 3.04915190e-01 4.60161835e-01 8.25956941e-01
3.87159467e-01 -9.11189139e-01 -1.48576558e-01 4.97188509e-01
8.91192675e-01 -1.05191207e+00 -8.53524059e-02 -1.74449563e-01
-2.42859378e-01 1.32725811e+00 9.20906365e-02 -3.70485216e-01
9.14169967e-01 2.30547652e-01 -7.58196831e-01 1.29567102e-01
-9.07974184e-01 1.79187834e-01 3.47454667e-01 8.68016779e-01
2.47395799e-01 2.08296448e-01 2.28796229e-02 9.37792182e-01
-3.77356648e-01 -6.47591054e-02 3.85460794e-01 4.40673918e-01
-5.86109221e-01 -7.88286626e-01 1.43609671e-02 3.59880716e-01
-6.00678444e-01 -2.00413316e-01 2.02076003e-01 6.22307241e-01
-2.64614224e-01 3.44787925e-01 -1.63272038e-01 -1.91979721e-01
-1.84738278e-01 4.25921887e-01 8.67741227e-01 -5.71098983e-01
4.38228026e-02 -4.75931793e-01 1.35526173e-02 -5.56884825e-01
-3.20061862e-01 -7.09759831e-01 -1.01572978e+00 -3.00925434e-01
-1.26871675e-01 1.72848403e-01 7.42116690e-01 1.12773585e+00
3.33999097e-01 1.91964135e-01 3.15020680e-01 -1.24102783e+00
-1.37527478e+00 -6.37342215e-01 -6.52055562e-01 -3.53028788e-03
2.56705612e-01 -8.61190796e-01 -6.90251112e-01 -5.69300577e-02]
|
[7.013063907623291, 4.068179607391357]
|
96952dc0-21d8-427f-aefd-aa6697fa7f15
|
what-makes-a-question-inquisitive-a-study-on
|
2205.08056
| null |
https://arxiv.org/abs/2205.08056v3
|
https://arxiv.org/pdf/2205.08056v3.pdf
|
"What makes a question inquisitive?" A Study on Type-Controlled Inquisitive Question Generation
|
We propose a type-controlled framework for inquisitive question generation. We annotate an inquisitive question dataset with question types, train question type classifiers, and finetune models for type-controlled question generation. Empirical results demonstrate that we can generate a variety of questions that adhere to specific types while drawing from the source texts. We also investigate strategies for selecting a single question from a generated set, considering both an informative vs.~inquisitive question classifier and a pairwise ranker trained from a small set of expert annotations. Question selection using the pairwise ranker yields strong results in automatic and manual evaluation. Our human evaluation assesses multiple aspects of the generated questions, finding that the ranker chooses questions with the best syntax (4.59), semantics (4.37), and inquisitiveness (3.92) on a scale of 1-5, even rivaling the performance of human-written questions.
|
['Kevin Gimpel', 'Debanjan Ghosh', 'Lingyu Gao']
|
2022-05-17
| null | null | null | null |
['question-selection']
|
['natural-language-processing']
|
[ 1.92220315e-01 6.54462934e-01 2.30762973e-01 -5.72868824e-01
-1.60988569e+00 -1.08201790e+00 7.36175239e-01 2.75471300e-01
-4.42940533e-01 8.03687990e-01 5.25562108e-01 -5.37071407e-01
-2.70294070e-01 -5.98810315e-01 -3.07727784e-01 6.64425781e-03
5.06667376e-01 8.00117552e-01 5.83710670e-01 -5.68263948e-01
5.81874788e-01 -1.44717515e-01 -1.51467931e+00 7.03051805e-01
1.39871538e+00 1.02060723e+00 1.43463582e-01 1.09520268e+00
-6.48505270e-01 1.16953743e+00 -1.24796128e+00 -7.27614880e-01
-1.41940221e-01 -7.93797910e-01 -1.49725437e+00 -8.41609985e-02
7.80805230e-01 -4.42276113e-02 3.67215663e-01 6.75262153e-01
3.50619376e-01 2.18324512e-01 9.80677485e-01 -1.02518022e+00
-9.31812465e-01 7.15930104e-01 2.71349430e-01 3.89719099e-01
1.14604199e+00 2.11417750e-01 1.48082495e+00 -7.69011378e-01
6.65640950e-01 1.26976585e+00 4.59576905e-01 6.36561155e-01
-1.09967852e+00 -3.16941559e-01 -1.25925735e-01 6.43525720e-02
-9.09502089e-01 -5.38147748e-01 3.58978182e-01 -4.40167695e-01
6.92441821e-01 7.52543330e-01 1.87628627e-01 8.34408939e-01
-7.73498416e-02 5.53891420e-01 1.36820483e+00 -7.31209755e-01
3.22519958e-01 6.26026452e-01 3.46220046e-01 4.95396644e-01
1.14092827e-01 -5.55494666e-01 -3.48740190e-01 -6.16783857e-01
2.28668943e-01 -8.51443350e-01 -4.02858704e-01 3.08686346e-01
-9.48751867e-01 9.55152333e-01 2.23390281e-01 4.40041602e-01
-3.00361246e-01 -5.40556759e-02 -4.01626807e-03 6.73594296e-01
1.41989648e-01 1.62522042e+00 -8.55323613e-01 -2.11753517e-01
-5.01345158e-01 7.80678213e-01 1.39801645e+00 1.17693067e+00
7.61814356e-01 -3.29172164e-01 -8.46053600e-01 1.11351717e+00
3.12488973e-01 4.39005136e-01 4.19193029e-01 -1.37726474e+00
3.75139564e-01 7.51271844e-01 5.56595027e-01 -6.89455092e-01
-1.40709981e-01 -8.84643421e-02 7.72617087e-02 -2.01927796e-01
7.47566462e-01 -4.33660626e-01 -3.36186707e-01 1.65549421e+00
4.11557704e-01 -9.30738568e-01 4.20470536e-02 4.36181575e-01
1.17309546e+00 6.86859131e-01 3.91403973e-01 -2.60094982e-02
1.83479798e+00 -5.45309663e-01 -6.25588596e-01 -2.04773381e-01
8.18825662e-01 -8.75898182e-01 1.56291592e+00 2.30749503e-01
-1.21471643e+00 -4.38369274e-01 -5.05751729e-01 -1.84148788e-01
-1.08425222e-01 1.75401241e-01 1.68193415e-01 6.32841349e-01
-1.19460630e+00 7.90033638e-02 9.11438316e-02 -4.44330633e-01
-1.08060665e-01 -1.31584883e-01 5.57469502e-02 4.70086969e-02
-1.25923502e+00 1.06185615e+00 9.01399404e-02 -6.52893662e-01
-5.16937912e-01 -6.13905668e-01 -6.79889560e-01 1.12650841e-02
3.93591046e-01 -8.71200264e-01 1.94394326e+00 -8.33702505e-01
-1.38426328e+00 7.46466279e-01 -1.94346532e-01 -1.51025802e-01
2.02845916e-01 -1.48643836e-01 -3.15316141e-01 7.46917903e-01
5.78171730e-01 7.01848447e-01 6.24711633e-01 -1.39127994e+00
-8.33119631e-01 5.42808510e-02 6.57673597e-01 3.26173902e-01
-1.35017186e-01 3.50507319e-01 1.72076210e-01 -5.80163836e-01
-1.21743143e-01 -4.97270465e-01 -1.32670239e-01 -9.61607397e-02
-3.95067990e-01 -9.58000362e-01 4.32736129e-01 -7.49009967e-01
1.37307250e+00 -1.42153215e+00 -3.87166053e-01 -1.01462286e-02
1.89791098e-01 -1.95894167e-02 -1.53690577e-01 5.57049394e-01
1.80014163e-01 5.09711266e-01 -8.33441243e-02 6.54491261e-02
3.72534126e-01 -6.10084683e-02 -2.72998810e-01 -4.11032826e-01
2.88454950e-01 9.26067054e-01 -1.02750015e+00 -8.70082200e-01
-3.94183457e-01 -3.84559214e-01 -8.17120194e-01 8.27153981e-01
-9.81825292e-01 -1.08033933e-01 -7.56002009e-01 4.55525577e-01
1.54773975e-02 -4.07622248e-01 1.20914392e-02 -1.84162017e-02
2.74932891e-01 1.06655407e+00 -1.01934052e+00 9.54866946e-01
-6.33298337e-01 4.19100285e-01 1.16007447e-01 -2.27440894e-01
1.09234881e+00 5.07176161e-01 -2.50541329e-01 -6.04219615e-01
3.82404700e-02 3.71293992e-01 -9.26895440e-02 -1.05278194e+00
7.37613022e-01 -3.57256830e-01 -4.12341714e-01 9.75814104e-01
2.58606434e-01 -6.77692890e-01 4.36682522e-01 5.54022253e-01
1.45755720e+00 -4.95028906e-02 1.99774787e-01 -5.04780054e-01
4.96763915e-01 3.78406852e-01 -1.08519830e-02 1.26353323e+00
-1.38769627e-01 6.73570931e-01 6.71106935e-01 1.05610967e-01
-6.26667082e-01 -9.53405440e-01 -4.30620089e-03 1.57194293e+00
-2.58930981e-01 -2.14461029e-01 -9.34775829e-01 -1.01566327e+00
-2.34240234e-01 1.43928945e+00 -4.84814435e-01 3.95567045e-02
-5.85492134e-01 -1.72173232e-01 6.10894859e-01 2.48491734e-01
1.71825364e-01 -1.32545042e+00 -4.84993190e-01 2.07959518e-01
-7.67320931e-01 -8.57970238e-01 -6.07280433e-01 4.61397916e-02
-4.01573837e-01 -1.14584744e+00 -1.95335835e-01 -8.86214614e-01
4.21127319e-01 -2.19400902e-03 1.74398971e+00 4.00581568e-01
2.45308042e-01 8.18441749e-01 -8.09509337e-01 -2.13224009e-01
-9.28071856e-01 4.32275414e-01 -4.92663115e-01 -2.93073863e-01
2.92235315e-01 -1.52546540e-01 -5.60035765e-01 3.55255693e-01
-1.04211533e+00 -4.48205024e-01 3.59473825e-01 6.17173195e-01
-1.32614359e-01 -4.52863991e-01 1.22410953e+00 -9.22991812e-01
1.35757911e+00 -6.69082880e-01 -2.66295403e-01 6.23323441e-01
-5.45810163e-01 2.79155850e-01 6.80833697e-01 -2.64567971e-01
-1.47986650e+00 -5.85507929e-01 -3.97830158e-01 4.58367527e-01
-3.24183166e-01 2.62559354e-01 -2.25647941e-01 3.30502540e-02
1.37644613e+00 -7.15489089e-02 -3.03149316e-02 -3.88062894e-01
6.12785518e-01 6.92289293e-01 2.97738940e-01 -1.01183903e+00
8.97289217e-01 -2.55776882e-01 -8.62847507e-01 -6.20805740e-01
-1.22891045e+00 -4.34790671e-01 -5.53228594e-02 -2.47811988e-01
7.69987822e-01 -5.11741042e-01 -6.41677320e-01 -1.75624564e-01
-1.16267812e+00 -3.22189838e-01 -5.74606657e-01 4.89136530e-03
-5.19447744e-01 2.76967913e-01 -6.37848854e-01 -7.96041787e-01
-4.38087821e-01 -8.36963952e-01 9.88708198e-01 2.96976686e-01
-1.07612765e+00 -1.04848409e+00 9.46070924e-02 9.06812608e-01
5.55714250e-01 -2.69190799e-02 1.43159485e+00 -1.36763763e+00
-5.29646099e-01 -1.12179495e-01 -1.16435535e-01 1.49516165e-01
8.52080211e-02 1.70058981e-02 -7.38755405e-01 2.90440142e-01
2.53242701e-01 -7.95516908e-01 2.25454494e-01 -1.55066267e-01
6.12887263e-01 -8.92618477e-01 8.72060731e-02 -1.94664285e-01
1.01460028e+00 2.24100798e-01 4.76502806e-01 3.38522166e-01
2.03365490e-01 9.95530784e-01 6.41466498e-01 1.75011262e-01
8.60410631e-01 3.70563030e-01 1.40730456e-01 6.34258926e-01
-4.98999245e-02 -4.48987037e-01 1.17141925e-01 5.24814188e-01
7.80483961e-01 -4.85356838e-01 -8.30202818e-01 9.03742790e-01
-1.17179728e+00 -9.00661290e-01 -3.70471150e-01 1.80023360e+00
1.50942993e+00 1.08735837e-01 1.33908644e-01 -4.96396869e-02
4.58309531e-01 5.19706234e-02 -1.31999739e-02 -5.07315397e-01
9.03239548e-02 4.82643545e-01 -1.74943849e-01 6.79910898e-01
-5.23790836e-01 7.52612591e-01 7.06131554e+00 5.46695530e-01
-3.62425596e-01 -1.35557711e-01 6.99381590e-01 3.41924548e-01
-1.15887022e+00 3.05088490e-01 -8.82134974e-01 3.07773888e-01
1.09922445e+00 -3.80915105e-01 3.10762972e-02 7.85303175e-01
-9.18781832e-02 -2.56353706e-01 -1.07828939e+00 1.95203960e-01
3.26308966e-01 -1.31532335e+00 3.74649048e-01 -3.63120884e-01
5.45394897e-01 -5.60053468e-01 -3.74154925e-01 6.14033103e-01
6.12672746e-01 -9.50044930e-01 8.50348294e-01 5.46091199e-01
2.85656512e-01 -3.16212386e-01 5.62744439e-01 5.22374153e-01
-7.04132915e-01 -5.98378181e-02 -4.02473845e-02 2.65358537e-02
8.93012211e-02 2.69988030e-01 -1.25894713e+00 1.49089009e-01
5.93002737e-01 -3.18345487e-01 -1.01978469e+00 7.95615971e-01
-4.03847098e-01 8.98954451e-01 -3.06843221e-01 -8.15271080e-01
1.44552231e-01 1.32047534e-01 5.16959846e-01 1.25027311e+00
8.70837569e-02 5.44388354e-01 -9.49194934e-03 9.36510623e-01
-1.99724481e-01 3.36058110e-01 -2.27628142e-01 6.39433637e-02
8.06260288e-01 1.34451830e+00 -2.33262733e-01 -5.86199701e-01
2.78940015e-02 3.81255358e-01 3.64357412e-01 2.44571790e-01
-3.76984954e-01 -7.03539610e-01 6.69367760e-02 2.81752706e-01
1.79402620e-01 2.42234856e-01 -4.40476686e-01 -1.00469148e+00
2.32432827e-01 -1.50082886e+00 6.33399725e-01 -1.17158461e+00
-1.33531749e+00 7.94350624e-01 2.59743541e-01 -7.32281685e-01
-6.79200828e-01 -3.93834323e-01 -8.95971239e-01 9.94625986e-01
-1.14767849e+00 -4.94940817e-01 -3.07131976e-01 1.45263344e-01
5.91547549e-01 1.75658807e-01 6.66167617e-01 -8.32019672e-02
-1.50576755e-01 7.40240932e-01 -6.51799202e-01 9.40862447e-02
6.72069371e-01 -1.73458326e+00 3.27977151e-01 5.25251985e-01
-9.98766050e-02 8.71303141e-01 8.33875954e-01 -4.66610849e-01
-1.15166342e+00 -9.45165098e-01 1.56064904e+00 -1.06387961e+00
6.74923897e-01 -1.22484351e-02 -1.38085973e+00 3.91820133e-01
6.77239358e-01 -5.95549643e-01 8.53984594e-01 6.78781420e-02
-5.16843975e-01 1.03944294e-01 -1.52776325e+00 6.54516339e-01
5.38135827e-01 -6.05703533e-01 -1.30098581e+00 5.30438304e-01
8.95855784e-01 -1.22037038e-01 -1.08427215e+00 -2.70309243e-02
1.81527764e-01 -9.35766757e-01 5.32122612e-01 -6.43209517e-01
4.80813295e-01 -2.97658890e-01 -1.91110089e-01 -1.11532831e+00
-2.89783776e-01 -7.87997484e-01 2.04333097e-01 1.56837547e+00
8.95328820e-01 -5.89727342e-01 4.85103935e-01 1.10545075e+00
-1.67126164e-01 -6.75863743e-01 -7.24243164e-01 -4.67079461e-01
1.91769138e-01 -9.19730589e-02 5.47061861e-01 7.86385775e-01
2.78104663e-01 1.05505300e+00 3.36218655e-01 -1.63754091e-01
3.36630344e-01 1.50390893e-01 6.91991091e-01 -9.94787514e-01
-3.73348385e-01 -4.73703682e-01 3.82419944e-01 -1.32552576e+00
-5.87185547e-02 -7.32993007e-01 5.35320461e-01 -1.81536865e+00
-2.25388139e-01 -6.20752811e-01 4.02612984e-01 3.20821732e-01
-8.68253589e-01 -1.51120707e-01 9.08842459e-02 4.47452925e-02
-6.30462766e-01 4.78141636e-01 1.09953666e+00 1.04191504e-01
-1.51627632e-02 7.54510835e-02 -1.44750535e+00 5.71770668e-01
8.65913153e-01 -5.33588052e-01 -5.13213277e-01 -3.16182554e-01
4.33638602e-01 4.84105855e-01 1.76012725e-01 -7.50478148e-01
6.50175959e-02 -3.19477290e-01 3.25049832e-02 -2.19081923e-01
7.80188963e-02 -2.15264723e-01 -3.36098194e-01 1.44461706e-01
-1.11745036e+00 3.19836229e-01 -1.48259237e-01 1.71781152e-01
-7.26764724e-02 -9.66427445e-01 8.19365382e-01 -3.91654551e-01
-1.15119569e-01 -1.65826783e-01 -6.08477652e-01 1.05644321e+00
4.08586621e-01 -1.63027883e-01 -4.66237664e-01 -7.78750479e-01
-3.49962085e-01 3.29046041e-01 2.90522039e-01 3.60870928e-01
4.72340107e-01 -1.05713975e+00 -9.19993460e-01 -2.62817740e-01
4.75909948e-01 -2.37735927e-01 -3.85519952e-01 1.14842169e-01
-3.74372333e-01 3.32701594e-01 3.31961602e-01 -2.23522976e-01
-7.41085768e-01 -3.51700187e-02 4.34557736e-01 -2.90944844e-01
9.04206245e-04 1.02839351e+00 -4.66651879e-02 -8.02458584e-01
4.80442345e-02 -4.74267513e-01 -6.21126592e-01 2.46540740e-01
4.69865590e-01 3.34682912e-01 6.40496463e-02 -2.49641657e-01
-3.52968089e-02 1.04382843e-01 -1.30289018e-01 -5.38667858e-01
6.25530660e-01 -1.73252955e-01 -2.35613361e-01 3.25747699e-01
1.15754664e+00 1.03127003e-01 -6.54732168e-01 -3.04914862e-01
3.87636930e-01 -3.07303488e-01 -5.57647884e-01 -1.15937054e+00
-1.73382610e-01 1.89435229e-01 -1.19836420e-01 9.43963230e-01
8.27239811e-01 2.88535476e-01 8.25821102e-01 7.72103667e-01
2.81819791e-01 -9.77193773e-01 5.98369658e-01 6.87152028e-01
1.26915169e+00 -9.77529168e-01 -2.56698012e-01 -2.76459277e-01
-7.74690032e-01 8.43266129e-01 1.00106370e+00 7.06591606e-02
2.91665435e-01 -2.04056218e-01 3.85473996e-01 -4.33691233e-01
-1.32246661e+00 -9.38217118e-02 2.79608339e-01 5.35813987e-01
6.34699345e-01 -2.09570780e-01 -4.87816989e-01 5.13762832e-01
-7.50557125e-01 -4.19240832e-01 8.00772667e-01 9.90634978e-01
-9.53124762e-01 -9.53122914e-01 -4.51949507e-01 9.22288954e-01
-5.09011149e-01 -2.68420368e-01 -7.86398530e-01 4.92398620e-01
-3.15937281e-01 1.64660680e+00 -1.93990305e-01 9.86542553e-03
5.03123164e-01 3.69425476e-01 1.81425333e-01 -1.01451516e+00
-1.36672306e+00 -4.10373151e-01 8.83423686e-01 5.74284680e-02
2.76355371e-02 -6.46340311e-01 -8.78212154e-01 1.63509265e-01
-5.74505270e-01 9.33950007e-01 2.86564469e-01 1.09532523e+00
3.43852103e-01 3.24874297e-02 8.22081864e-01 2.15446249e-01
-1.09756398e+00 -1.23760700e+00 -3.10005080e-02 5.70938051e-01
2.84577668e-01 -2.13183627e-01 -6.92803383e-01 1.12520464e-01]
|
[11.595683097839355, 8.159745216369629]
|
a9d9cdaf-005f-42b2-9331-f03013f1cf91
|
coseg-cognitively-inspired-unsupervised
|
2109.15170
| null |
https://arxiv.org/abs/2109.15170v1
|
https://arxiv.org/pdf/2109.15170v1.pdf
|
CoSeg: Cognitively Inspired Unsupervised Generic Event Segmentation
|
Some cognitive research has discovered that humans accomplish event segmentation as a side effect of event anticipation. Inspired by this discovery, we propose a simple yet effective end-to-end self-supervised learning framework for event segmentation/boundary detection. Unlike the mainstream clustering-based methods, our framework exploits a transformer-based feature reconstruction scheme to detect event boundary by reconstruction errors. This is consistent with the fact that humans spot new events by leveraging the deviation between their prediction and what is actually perceived. Thanks to their heterogeneity in semantics, the frames at boundaries are difficult to be reconstructed (generally with large reconstruction errors), which is favorable for event boundary detection. Additionally, since the reconstruction occurs on the semantic feature level instead of pixel level, we develop a temporal contrastive feature embedding module to learn the semantic visual representation for frame feature reconstruction. This procedure is like humans building up experiences with "long-term memory". The goal of our work is to segment generic events rather than localize some specific ones. We focus on achieving accurate event boundaries. As a result, we adopt F1 score (Precision/Recall) as our primary evaluation metric for a fair comparison with previous approaches. Meanwhile, we also calculate the conventional frame-based MoF and IoU metric. We thoroughly benchmark our work on four publicly available datasets and demonstrate much better results.
|
['Jiebo Luo', 'Tao Mei', 'Jingen Liu', 'Xiao Wang']
|
2021-09-30
| null | null | null | null |
['boundary-detection']
|
['computer-vision']
|
[ 2.19065100e-01 -4.77172099e-02 -1.01404749e-01 -5.56798697e-01
-7.21514642e-01 -3.80444795e-01 7.04185605e-01 5.26250482e-01
-6.31034732e-01 4.01871175e-01 6.62188947e-01 3.59375834e-01
1.13102384e-01 -8.28833640e-01 -6.51651323e-01 -4.07818794e-01
-1.07792698e-01 -7.88558275e-02 4.51227158e-01 -1.37844011e-02
3.70049506e-01 1.35864198e-01 -1.64060640e+00 5.80884874e-01
5.81368208e-01 1.11558914e+00 3.71406347e-01 2.87183672e-01
3.04047558e-02 8.61320972e-01 -4.48366195e-01 -2.26173282e-01
3.18223536e-02 -6.81808710e-01 -8.77888024e-01 2.29372218e-01
-1.78971469e-01 -2.50457674e-01 -4.93269384e-01 9.38433528e-01
4.63576943e-01 5.52451849e-01 4.77420658e-01 -1.11642885e+00
-3.87834907e-01 5.03002346e-01 -3.05407524e-01 4.77977067e-01
7.51341045e-01 1.15330644e-01 1.15335786e+00 -9.87917066e-01
8.90411079e-01 1.08255637e+00 6.27599776e-01 1.87539130e-01
-9.72259521e-01 -2.29442194e-01 5.11367559e-01 6.69777572e-01
-1.47619152e+00 -4.85757291e-01 1.06401491e+00 -3.79353851e-01
8.59711289e-01 2.18725532e-01 9.10679519e-01 1.34616816e+00
2.15105623e-01 1.02443087e+00 8.87679636e-01 -3.05653989e-01
5.11161745e-01 -1.90325692e-01 -3.18895467e-02 4.54730988e-01
-2.97508538e-01 1.68484703e-01 -8.18608105e-01 2.34346032e-01
7.63521314e-01 3.99133891e-01 -2.99432993e-01 -3.56507599e-02
-1.73856032e+00 5.84711194e-01 6.35544479e-01 3.59845817e-01
-5.70908427e-01 -2.60930639e-02 5.47886431e-01 9.54495147e-02
3.14306736e-01 4.17291336e-02 -1.99618787e-01 -2.93117434e-01
-1.08513761e+00 1.97223723e-01 4.65238661e-01 4.57029819e-01
6.89636886e-01 -1.79695681e-01 -5.85696995e-01 6.20465040e-01
3.84762943e-01 -1.66394338e-01 7.51401901e-01 -9.84247923e-01
8.94472525e-02 6.84714675e-01 2.54755288e-01 -1.22295880e+00
-5.45064926e-01 -3.84081244e-01 -5.51252782e-01 -4.91884053e-02
4.22097534e-01 1.28972143e-01 -6.36562765e-01 1.80027020e+00
4.53936100e-01 5.73545694e-01 -2.60131825e-02 1.24214470e+00
6.72838092e-01 6.87663555e-01 3.23753059e-01 -4.83704865e-01
1.55823791e+00 -8.00915241e-01 -8.97039294e-01 -2.61045247e-01
4.45151061e-01 -6.25927389e-01 1.08149731e+00 2.23835275e-01
-9.61532712e-01 -7.03535736e-01 -9.68871474e-01 -2.49022126e-01
-2.78806180e-01 -3.25759836e-02 8.03718209e-01 6.86688349e-02
-7.77398884e-01 5.51999211e-01 -9.08516169e-01 -5.37266016e-01
4.31273460e-01 -3.23131382e-01 -1.95728377e-01 2.61503577e-01
-1.25459993e+00 6.92359686e-01 6.60394669e-01 5.17180711e-02
-9.94404376e-01 -6.19457424e-01 -8.86807144e-01 5.76118231e-02
4.41295415e-01 -5.14022350e-01 1.17171609e+00 -1.02467632e+00
-1.30661058e+00 9.28968370e-01 -4.81562644e-01 -7.00829089e-01
5.27513444e-01 -1.33109674e-01 -4.99719948e-01 4.75266993e-01
3.24057460e-01 8.48060369e-01 7.80000627e-01 -1.16477239e+00
-7.54981518e-01 -2.01873213e-01 1.20727457e-01 3.91916424e-01
-1.45708412e-01 1.21039174e-01 -3.98548961e-01 -8.71221721e-01
5.56452155e-01 -4.29798216e-01 -2.38583356e-01 7.13255331e-02
-2.34575510e-01 -4.34736699e-01 4.75331932e-01 -6.18168056e-01
1.44527829e+00 -2.38105607e+00 -1.49840072e-01 -2.54642367e-02
3.45804811e-01 -2.98670918e-01 2.04866573e-01 3.95903856e-01
-1.49409413e-01 -1.59666538e-01 -1.42808124e-01 -3.53726208e-01
1.43528521e-01 -8.51616785e-02 -5.77872515e-01 4.71225113e-01
1.71882331e-01 9.07397687e-01 -1.15583360e+00 -7.00265884e-01
4.89617527e-01 3.88561904e-01 -4.25814718e-01 1.95816070e-01
-1.20916225e-01 5.55894077e-01 -4.88336027e-01 6.03882492e-01
2.92180419e-01 -2.99398810e-01 -7.07626343e-04 -4.45465505e-01
-3.10257584e-01 3.60714287e-01 -1.14549148e+00 2.18841839e+00
-1.20495185e-01 5.64874113e-01 -6.36318326e-01 -1.26535165e+00
7.74673462e-01 5.05441606e-01 7.39729702e-01 -8.26787233e-01
2.59368390e-01 -2.36007839e-01 -3.38829994e-01 -4.64711189e-01
3.34501684e-01 -9.86249000e-02 -1.65515572e-01 3.66677821e-01
2.11344585e-02 3.82173568e-01 2.32888088e-01 2.78437495e-01
1.19869006e+00 3.73981953e-01 4.19207782e-01 7.77630508e-03
2.44978204e-01 -8.33002501e-04 9.64796841e-01 7.13570535e-01
-6.48569047e-01 8.87531102e-01 3.91308963e-01 -5.87429464e-01
-7.05749810e-01 -1.44513464e+00 -9.71929729e-02 1.06687152e+00
5.81423879e-01 -7.01714635e-01 -8.80252123e-01 -6.40439749e-01
-5.08290291e-01 8.30949008e-01 -5.30972719e-01 -2.55374670e-01
-5.04751027e-01 -7.68001616e-01 2.99443215e-01 6.60702825e-01
8.32310617e-01 -1.34856308e+00 -1.06564760e+00 5.19566834e-01
-6.58935308e-01 -1.10404873e+00 -4.93827999e-01 1.55072855e-02
-6.55634880e-01 -1.02165234e+00 -3.95231009e-01 -8.36702228e-01
5.00406623e-01 3.05796295e-01 9.73156691e-01 -2.42744222e-01
-3.24267328e-01 3.88889015e-01 -6.21084690e-01 -2.01884374e-01
2.34110162e-01 -5.04247844e-01 -1.44028217e-01 2.62745857e-01
5.66112101e-01 -5.68785369e-01 -1.08421385e+00 2.67713785e-01
-9.16722775e-01 2.88024545e-01 2.60437399e-01 5.61864674e-01
8.37089777e-01 2.54196167e-01 6.24160409e-01 -3.30661327e-01
4.40248787e-01 -5.36042035e-01 -2.85283953e-01 2.47750208e-01
-3.15398604e-01 -3.13384384e-01 4.15088505e-01 -4.11224335e-01
-1.29513276e+00 1.69140220e-01 -1.80904099e-04 -3.08497131e-01
-3.53903353e-01 4.16317612e-01 5.01130568e-03 4.62052286e-01
6.50904179e-01 3.84417683e-01 -4.24351007e-01 -1.91836759e-01
2.65350282e-01 2.71879286e-01 6.01889431e-01 -2.81682074e-01
3.67210925e-01 8.34616423e-01 -4.78159219e-01 -4.70370620e-01
-1.09979069e+00 -4.44498956e-01 -5.41948676e-01 -6.57939494e-01
1.05625343e+00 -1.04824352e+00 -8.04490626e-01 2.82607913e-01
-1.25956595e+00 -1.42816737e-01 -5.72502494e-01 7.35925019e-01
-7.51317620e-01 4.06301141e-01 -6.91383004e-01 -6.66802406e-01
-3.87947448e-02 -8.91819060e-01 1.09695208e+00 3.12407494e-01
-4.50675666e-01 -9.37959731e-01 3.79270725e-02 1.97650284e-01
4.51357886e-02 5.53987503e-01 4.06743228e-01 -4.78275239e-01
-5.96357167e-01 -2.15455424e-02 -2.76058197e-01 -9.25022662e-02
-5.12207523e-02 -3.08008701e-01 -9.68989372e-01 -2.45351810e-02
4.18039680e-01 2.76088458e-03 1.02249014e+00 5.60334861e-01
1.27576244e+00 4.36961371e-03 -4.08608675e-01 4.48872268e-01
1.27646101e+00 2.29864761e-01 8.51292729e-01 3.99804860e-01
5.41951418e-01 6.07262075e-01 8.18052292e-01 8.17804933e-01
7.82336116e-01 4.92022157e-01 1.87922001e-01 7.43923858e-02
-1.69882402e-01 -4.34641212e-01 6.04062796e-01 5.18849850e-01
3.29054594e-02 -2.22427219e-01 -8.24770689e-01 6.27648532e-01
-2.21585822e+00 -1.23306406e+00 9.35706496e-02 2.15560174e+00
8.41350377e-01 2.50214487e-01 1.22616952e-02 2.77253598e-01
9.26190794e-01 3.46248955e-01 -4.45241004e-01 -7.03732809e-03
-1.67331412e-01 3.34344059e-02 1.99043211e-02 1.48025349e-01
-1.30461550e+00 1.04805827e+00 6.10248518e+00 8.48083496e-01
-9.17919397e-01 3.15388232e-01 6.30238354e-01 -1.50745586e-01
-1.19410090e-01 1.55395225e-01 -4.93871778e-01 6.60965681e-01
7.93934762e-01 -6.62309602e-02 3.00512582e-01 5.07393837e-01
6.75377786e-01 -3.90635520e-01 -1.26807570e+00 1.23178685e+00
-2.76191700e-02 -1.35147762e+00 -1.22468777e-01 -3.89087170e-01
4.04118538e-01 -3.47807914e-01 -1.82079047e-01 3.08721989e-01
-1.08857721e-01 -8.19482684e-01 1.15064597e+00 7.49663651e-01
4.13909674e-01 -5.75095892e-01 2.56450355e-01 1.96902081e-01
-1.49126434e+00 -3.29179596e-03 -2.71963447e-01 -3.14917684e-01
5.08083880e-01 9.31595445e-01 -5.61959088e-01 4.35093880e-01
8.09969664e-01 1.00487542e+00 -4.43810135e-01 1.11989462e+00
-3.86224955e-01 6.04392529e-01 -2.09191337e-01 2.72032768e-01
1.09450482e-01 7.39320219e-02 6.48581862e-01 1.18658018e+00
3.25954258e-01 1.43405572e-01 4.15983349e-01 1.03023016e+00
1.37650788e-01 8.21962208e-02 -3.49339098e-01 1.04092315e-01
5.27715087e-01 1.15062332e+00 -1.27194667e+00 -4.12546247e-01
-3.89018357e-01 1.30589163e+00 1.91294461e-01 3.64710718e-01
-1.26711190e+00 -3.05369407e-01 3.52787405e-01 -5.19274659e-02
2.79087633e-01 -1.11160137e-01 -3.56557220e-01 -1.39463031e+00
2.25765824e-01 -3.21965903e-01 5.08561730e-01 -7.90885270e-01
-1.20037842e+00 3.63274187e-01 -1.70076881e-02 -1.43586779e+00
-1.17288753e-02 2.45864522e-02 -8.30382586e-01 2.70396113e-01
-1.45470893e+00 -8.61720085e-01 -4.29024905e-01 6.89470410e-01
7.85707295e-01 2.76037276e-01 6.19436860e-01 3.05429250e-01
-5.62293530e-01 4.21353340e-01 -3.38837534e-01 1.65714338e-01
6.92256391e-01 -1.08687258e+00 1.40874162e-01 1.04991996e+00
2.90136635e-01 3.68394583e-01 7.01332211e-01 -8.30607712e-01
-1.03152514e+00 -1.08656931e+00 1.01501584e+00 -3.40197206e-01
5.42205095e-01 -2.78778940e-01 -8.07606161e-01 7.01019466e-01
6.16529025e-02 1.41635537e-01 6.13453448e-01 -4.32620272e-02
-2.48385593e-01 -1.66728683e-02 -1.14308012e+00 7.85586655e-01
1.48941088e+00 -5.28746665e-01 -1.01205015e+00 3.56131017e-01
7.45538652e-01 -1.93501905e-01 -7.70338476e-01 4.06793833e-01
4.28502560e-01 -1.30716574e+00 1.04254270e+00 -2.17998594e-01
3.90532494e-01 -5.36255121e-01 -1.41334802e-01 -9.83484507e-01
-3.94244283e-01 -5.47392011e-01 -1.14892133e-01 1.31782401e+00
-6.40097111e-02 -3.23146850e-01 5.11740029e-01 4.04636770e-01
-2.12084696e-01 -5.04904568e-01 -9.91342783e-01 -6.04527354e-01
-5.40491879e-01 -7.26043046e-01 5.20126581e-01 9.45160508e-01
1.22738913e-01 1.27685592e-01 -1.37642190e-01 1.60506427e-01
5.79870403e-01 3.63647133e-01 2.07568660e-01 -1.05043972e+00
-5.38343098e-03 -3.64714950e-01 -5.37540376e-01 -1.03839934e+00
1.98285043e-01 -9.33485091e-01 2.57949322e-01 -1.47127354e+00
2.86784679e-01 -1.41079575e-01 -7.57278323e-01 3.41232836e-01
-1.82334766e-01 1.89202249e-01 1.08224012e-01 2.44071618e-01
-1.10919356e+00 6.41981065e-01 1.13645566e+00 1.92524865e-01
-3.08862776e-01 -3.00336421e-01 -5.12643516e-01 1.00469148e+00
7.07389295e-01 -4.08884048e-01 -3.46516222e-01 -2.55803913e-01
1.49266720e-01 2.00693369e-01 7.36738563e-01 -1.21814764e+00
2.94933081e-01 -8.55852664e-02 6.59290850e-01 -6.75493062e-01
3.19616497e-01 -6.13437831e-01 3.31896022e-02 1.82501778e-01
-5.12759209e-01 -1.29044399e-01 -1.70086876e-01 7.70637393e-01
-4.43799436e-01 -6.42200187e-02 5.09226799e-01 -3.15402001e-01
-1.34314215e+00 2.17341274e-01 -5.58561862e-01 8.19065124e-02
1.23587453e+00 -3.79185677e-01 -1.03986643e-01 -2.97565103e-01
-9.13189650e-01 1.37777418e-01 2.80395746e-01 4.16502208e-01
9.05729651e-01 -1.43715680e+00 -4.69076097e-01 5.00782914e-02
1.88448355e-01 -7.72936940e-02 4.25382078e-01 1.05396903e+00
-2.47005969e-01 5.02382293e-02 -1.70466110e-01 -6.19951248e-01
-7.21348524e-01 5.61160207e-01 -2.88156904e-02 -2.89643090e-02
-9.82933700e-01 7.94587195e-01 4.73735690e-01 9.61682722e-02
2.86048889e-01 -3.96050036e-01 -3.23308200e-01 3.70979369e-01
7.51777291e-01 2.91681647e-01 -1.83529511e-01 -4.73073363e-01
-4.33526039e-01 1.62324980e-01 8.90167654e-02 -1.78753063e-01
1.18749273e+00 -4.76312727e-01 1.09640203e-01 7.83115447e-01
1.13664687e+00 -2.82561392e-01 -1.51723921e+00 -3.16738099e-01
9.31957737e-02 -4.77086216e-01 1.53470203e-01 -6.06855929e-01
-7.64048159e-01 7.29740083e-01 7.96347320e-01 2.17813089e-01
1.36592615e+00 1.08977675e-01 1.13117385e+00 9.43773761e-02
3.00344110e-01 -1.37269962e+00 3.16861451e-01 3.68649483e-01
6.48072362e-01 -1.40030074e+00 -1.26282647e-01 -2.99900085e-01
-8.64552081e-01 9.51247692e-01 4.23869401e-01 -2.28338033e-01
5.23765683e-01 -7.72235021e-02 -1.60413727e-01 -1.97388694e-01
-6.26195073e-01 -5.40616393e-01 2.30831891e-01 3.78164619e-01
3.23770612e-01 1.04548531e-02 -5.55981100e-01 9.12439644e-01
-8.46274942e-02 8.94625336e-02 1.72295228e-01 1.01530993e+00
-7.05241978e-01 -7.82195926e-01 -2.68334389e-01 2.73918480e-01
-4.17310178e-01 9.75161418e-02 3.74095812e-02 3.85178238e-01
2.63520718e-01 1.05939436e+00 2.97260731e-01 -2.95007974e-01
1.10461161e-01 2.31922343e-01 3.82680476e-01 -4.74083662e-01
-4.92760837e-01 2.55369186e-01 -2.39190340e-01 -1.00615680e+00
-6.59263253e-01 -9.01501477e-01 -1.78598571e+00 5.85242026e-02
9.88673344e-02 -1.39319494e-01 3.55062693e-01 1.05564213e+00
3.82428765e-01 6.38241649e-01 7.15858459e-01 -8.03715289e-01
-3.94731984e-02 -5.16834795e-01 -3.86445254e-01 8.01302314e-01
4.32808809e-02 -9.00017738e-01 -3.25264663e-01 4.16537046e-01]
|
[8.550943374633789, 0.5661830902099609]
|
c2ca5fef-a816-4df8-841a-9f35d2a87f7a
|
a-semantic-network-based-evolutionary
|
1404.7765
| null |
http://arxiv.org/abs/1404.7765v2
|
http://arxiv.org/pdf/1404.7765v2.pdf
|
A semantic network-based evolutionary algorithm for computational creativity
|
We introduce a novel evolutionary algorithm (EA) with a semantic
network-based representation. For enabling this, we establish new formulations
of EA variation operators, crossover and mutation, that we adapt to work on
semantic networks. The algorithm employs commonsense reasoning to ensure all
operations preserve the meaningfulness of the networks, using ConceptNet and
WordNet knowledge bases. The algorithm can be interpreted as a novel memetic
algorithm (MA), given that (1) individuals represent pieces of information that
undergo evolution, as in the original sense of memetics as it was introduced by
Dawkins; and (2) this is different from existing MA, where the word "memetic"
has been used as a synonym for local refinement after global optimization. For
evaluating the approach, we introduce an analogical similarity-based fitness
measure that is computed through structure mapping. This setup enables the
open-ended generation of networks analogous to a given base network.
|
['Atilim Gunes Baydin', 'Santiago Ontanon', 'Ramon Lopez de Mantaras']
|
2014-04-30
| null | null | null | null |
['analogical-similarity']
|
['reasoning']
|
[ 5.03687799e-01 5.26384473e-01 1.27585232e-01 9.02107917e-03
7.14006782e-01 -4.54979420e-01 9.17021394e-01 4.19420689e-01
-5.99299431e-01 9.08351839e-01 7.03490674e-02 -1.04697034e-01
-8.65609527e-01 -1.34570956e+00 -4.34281379e-01 -5.90433121e-01
-4.42292839e-02 4.72232401e-01 3.16027373e-01 -8.88286531e-01
5.17170191e-01 5.28030396e-01 -1.96889269e+00 1.98008604e-02
1.08044291e+00 6.06495500e-01 1.76521704e-01 2.40017444e-01
-5.41934490e-01 5.90100944e-01 -9.47413981e-01 -7.05743670e-01
7.01203421e-02 -9.98045564e-01 -1.27643037e+00 -2.61578292e-01
-2.71226108e-01 7.28036284e-01 6.80651702e-03 1.12424922e+00
2.27815256e-01 4.03497696e-01 5.50927341e-01 -1.31163657e+00
-7.26067305e-01 1.05758417e+00 1.82674959e-01 3.49851251e-02
5.16381383e-01 -1.88971952e-01 8.82366121e-01 -7.42206573e-02
9.01997447e-01 1.36097908e+00 7.42595732e-01 7.90484965e-01
-1.49681664e+00 -1.46639332e-01 -1.99047461e-01 4.20467854e-01
-1.43963790e+00 3.64589170e-02 7.30447173e-01 -1.44990787e-01
6.05921924e-01 5.64042091e-01 1.24782526e+00 1.14238358e+00
3.68001647e-02 2.53814906e-01 8.80280852e-01 -9.56274152e-01
9.13445175e-01 4.75530982e-01 -8.30891654e-02 3.62383336e-01
5.95583439e-01 2.85949796e-01 -5.25476336e-01 -1.49991840e-01
4.14214611e-01 -4.70198214e-01 -1.81595430e-01 -7.39467382e-01
-1.03271890e+00 1.05530548e+00 6.08547568e-01 1.10146093e+00
-3.39913219e-01 7.28185475e-02 3.70354593e-01 6.64194226e-01
2.13727932e-02 1.21957779e+00 -1.00002840e-01 4.20125248e-03
-7.24751890e-01 3.81853402e-01 1.12651324e+00 5.79891682e-01
6.12858534e-01 7.01854452e-02 -4.32266062e-03 5.19627929e-01
1.50639579e-01 -5.35650691e-03 9.69615936e-01 -8.80919993e-01
-3.21498394e-01 9.99835551e-01 -1.81124702e-01 -1.19213879e+00
-3.50324750e-01 -7.14192986e-01 -6.52354896e-01 3.95484179e-01
-6.30082339e-02 1.45985246e-01 -5.21109939e-01 2.13801312e+00
3.27755511e-01 3.33744258e-01 3.51712257e-01 4.20311123e-01
6.04898989e-01 2.13503093e-01 4.65672389e-02 -3.30864370e-01
1.00086069e+00 -6.74729228e-01 -7.16440260e-01 1.63710952e-01
3.40610802e-01 -1.96223423e-01 8.08690488e-01 2.78893232e-01
-1.11491454e+00 -3.96946043e-01 -1.35155928e+00 4.75846231e-01
-1.24005616e+00 -6.76242352e-01 5.41117072e-01 1.23810267e+00
-1.61681509e+00 9.92923498e-01 -3.16832781e-01 -9.68500912e-01
2.13723168e-01 3.89687121e-01 -1.02854051e-01 4.87121016e-01
-1.54474008e+00 1.38270271e+00 1.17968869e+00 -8.55079740e-02
-4.36414033e-01 -4.66771722e-01 -6.94106400e-01 2.42001414e-01
4.40844715e-01 -1.24349105e+00 6.22195303e-01 -1.50524604e+00
-1.74637556e+00 1.12012458e+00 1.87906995e-01 -6.51772618e-01
4.87821817e-01 6.38938606e-01 -6.13804162e-01 1.83121413e-01
-1.74432635e-01 6.20164573e-01 8.57050955e-01 -1.57336915e+00
-3.48741829e-01 -2.11465031e-01 3.31407696e-01 1.22716911e-02
-7.42848396e-01 1.85102634e-02 2.01591939e-01 -9.22597468e-01
1.67170614e-01 -7.30107248e-01 -1.37569979e-01 -8.78389776e-02
-3.02486420e-01 -2.72817127e-02 4.17655349e-01 -1.40909269e-01
1.41133201e+00 -1.94217288e+00 7.72714198e-01 8.95288348e-01
3.54263633e-01 3.59343112e-01 -2.45705053e-01 5.92990398e-01
-3.16472292e-01 4.93617862e-01 -6.66734755e-01 4.36571352e-02
1.81573316e-01 4.65768248e-01 2.11602245e-02 -4.49179336e-02
1.49457559e-01 9.09169376e-01 -1.02245915e+00 -3.29912066e-01
1.14202030e-01 3.17452282e-01 -4.52761501e-01 -3.36216211e-01
-1.96219489e-01 9.34375897e-02 -4.20781940e-01 3.16093385e-01
4.25904900e-01 4.84063253e-02 5.08963823e-01 -2.60515157e-02
-1.55130941e-02 -1.75668344e-01 -1.16671848e+00 1.88329685e+00
-4.10934865e-01 3.45930874e-01 -1.31218016e-01 -1.33346272e+00
1.22507310e+00 1.39888331e-01 2.37294823e-01 -7.70935535e-01
5.01020133e-01 3.38401705e-01 1.98246062e-01 -3.97944182e-01
5.59294701e-01 -2.28485674e-01 2.18621552e-01 4.06278968e-01
2.40562111e-01 -3.18030000e-01 7.06843138e-01 4.58721537e-03
1.11502934e+00 1.94800153e-01 5.15301466e-01 -7.45207787e-01
9.34350669e-01 1.48964807e-01 3.36476266e-01 8.75726104e-01
1.97599009e-01 1.41987562e-01 4.81710315e-01 -3.07647318e-01
-9.81429100e-01 -1.24360693e+00 -1.32757768e-01 8.58413994e-01
1.57546565e-01 -4.87187177e-01 -1.04832399e+00 -2.95533985e-01
-1.05026126e-01 9.93355453e-01 -9.99473751e-01 -7.02291906e-01
-4.02835488e-01 -6.90366566e-01 7.04826832e-01 5.51977865e-02
3.98405999e-01 -1.34516025e+00 -9.57556546e-01 2.18020678e-01
3.21831077e-01 -4.03973371e-01 4.46809113e-01 2.11699441e-01
-1.00671744e+00 -9.30017173e-01 -3.52266341e-01 -5.69997072e-01
4.28635240e-01 -4.71571118e-01 1.42204285e+00 6.61237299e-01
-3.99708956e-01 3.07397366e-01 -8.24238718e-01 -4.43220556e-01
-9.96772110e-01 1.96466848e-01 7.68624470e-02 -7.40673207e-03
2.43990436e-01 -1.18494427e+00 -1.22031093e-01 -4.59191203e-02
-1.44768250e+00 -2.68996269e-01 3.54842335e-01 7.92171717e-01
2.71428451e-02 4.63573903e-01 7.04712749e-01 -6.73327982e-01
1.05304956e+00 -4.15554792e-01 -2.70270318e-01 5.63983262e-01
-9.25775647e-01 3.43719333e-01 5.17580688e-01 -3.64033222e-01
-8.24067950e-01 -4.89921659e-01 8.38455744e-03 -9.11513492e-02
-2.58758999e-02 5.28396904e-01 -1.98890164e-01 -4.09746379e-01
8.83367956e-01 2.96632975e-01 2.22371325e-01 -3.32041770e-01
4.64999259e-01 3.00206125e-01 4.67776358e-01 -8.15649986e-01
6.81399822e-01 4.62201327e-01 3.63150567e-01 -4.91574258e-01
-2.00810477e-01 2.12648675e-01 -5.90530694e-01 -1.48958769e-02
8.51350725e-01 -1.54016577e-02 -9.66115654e-01 1.58178374e-01
-1.15598762e+00 1.21463008e-01 -1.09460962e+00 1.71652853e-01
-6.90281332e-01 2.05543321e-02 -6.06718995e-02 -8.39869320e-01
-6.95166364e-02 -6.92028940e-01 1.93071201e-01 3.60378176e-01
-4.70127344e-01 -1.36227226e+00 2.24706247e-01 -2.78294146e-01
7.23276913e-01 4.72705364e-01 1.30672967e+00 -8.93589973e-01
2.82110251e-03 5.66693917e-02 1.60060838e-01 3.12522084e-01
7.08598718e-02 1.53423488e-01 -7.82063246e-01 -1.20200746e-01
1.86977908e-01 2.16190666e-01 6.65346324e-01 2.56388672e-02
6.94932163e-01 -1.29489183e-01 -3.96265000e-01 4.45162565e-01
1.56183982e+00 5.88094771e-01 9.12234366e-01 9.02825058e-01
1.45409957e-01 9.98521626e-01 4.33836095e-02 3.50715280e-01
2.99884621e-02 7.42150247e-01 3.39850307e-01 1.64832249e-01
-2.48220805e-02 -2.26107091e-01 8.04106891e-03 6.68579400e-01
-1.54876411e-01 -3.53593022e-01 -8.08137059e-01 4.99636501e-01
-1.95632350e+00 -1.17392588e+00 2.87690699e-01 2.10819817e+00
6.92609608e-01 5.33601232e-02 2.48363942e-01 3.68420780e-01
1.05314851e+00 -3.62365991e-02 -1.98855102e-01 -8.52833986e-01
-5.51865935e-01 8.16334665e-01 -8.27146135e-03 4.60371315e-01
-4.44153249e-01 7.59221196e-01 6.41211510e+00 9.55183327e-01
-7.40449786e-01 2.71770954e-01 4.49565426e-02 5.80173843e-02
-5.90693653e-01 3.24408889e-01 -9.23279896e-02 5.96654534e-01
8.95884395e-01 -6.19880497e-01 7.34475672e-01 2.86199450e-01
3.11595332e-02 -5.09685054e-02 -1.01671636e+00 7.19916642e-01
2.64050186e-01 -1.58084869e+00 4.98487592e-01 -1.59010828e-01
6.61603689e-01 -7.11086035e-01 -1.59401521e-01 -5.31484149e-02
2.66844660e-01 -1.10695648e+00 8.96257579e-01 8.34708869e-01
1.43883988e-01 -8.04007590e-01 6.16453707e-01 1.14998007e-02
-9.44309831e-01 -3.91798228e-01 -1.20460875e-01 -8.34514126e-02
1.39309302e-01 2.98088193e-01 -1.92297548e-01 1.03279173e+00
5.66276491e-01 4.30934101e-01 -7.86786675e-01 1.08153343e+00
-2.80903518e-01 4.35187146e-02 -3.21803808e-01 -4.04941529e-01
1.81270897e-01 -5.26230693e-01 1.01233733e+00 8.49098861e-01
4.86504495e-01 -3.79336059e-01 -3.63035738e-01 1.56438327e+00
6.55322820e-02 1.62368849e-01 -4.21526760e-01 -3.40644456e-02
7.86402345e-01 9.76881564e-01 -1.19796336e+00 -2.24710807e-01
3.49112421e-01 1.06887209e+00 -1.92945331e-01 1.47501931e-01
-6.04014754e-01 -6.09879732e-01 3.56755912e-01 -3.06662340e-02
8.04838911e-02 3.90437990e-01 -2.84987360e-01 -7.36904740e-01
-1.21712439e-01 -7.37570405e-01 2.19483212e-01 -6.81285620e-01
-1.23634362e+00 8.21225405e-01 1.30106926e-01 -7.82552600e-01
-2.21682653e-01 -6.77593052e-01 -7.46964991e-01 7.06756830e-01
-1.07048929e+00 -8.78571510e-01 -2.68665910e-01 5.87255895e-01
-1.60176784e-01 -3.83361131e-01 1.02105677e+00 2.60002203e-02
-4.31971729e-01 4.83432442e-01 -5.83822206e-02 -3.94766867e-01
6.59143701e-02 -1.22278380e+00 1.87743172e-01 6.77478492e-01
-2.93510780e-03 7.48215854e-01 9.78863239e-01 -5.32391846e-01
-8.96596849e-01 -6.55659020e-01 9.88800466e-01 -3.22992384e-01
6.48992479e-01 -1.38764143e-01 -6.84349895e-01 2.01117858e-01
4.23196703e-01 -5.43193221e-01 4.12620544e-01 -3.76017317e-02
-7.12016970e-02 6.03108294e-02 -1.54117942e+00 7.04706907e-01
1.54416609e+00 -2.69547731e-01 -1.00601995e+00 1.09221213e-01
8.87753248e-01 2.04103649e-01 -9.19701278e-01 2.76430935e-01
3.08259934e-01 -1.23184395e+00 1.01959729e+00 -6.76571429e-01
1.23227008e-01 -5.53120673e-01 -3.61133069e-02 -1.47816205e+00
-3.50959688e-01 -7.13450789e-01 1.93049107e-02 1.25107932e+00
3.63979250e-01 -1.15301836e+00 4.13215637e-01 7.38123208e-02
-9.63546187e-02 -3.77000272e-01 -1.08769262e+00 -1.21517491e+00
4.19737622e-02 1.66603737e-02 1.23809803e+00 1.36927259e+00
3.04107457e-01 1.77549809e-01 2.00051263e-01 -4.95583683e-01
3.32909286e-01 -2.13297680e-01 -7.07837269e-02 -1.71851432e+00
-4.62539673e-01 -1.26634920e+00 -9.72497821e-01 8.54970664e-02
5.48800766e-01 -1.12245059e+00 -4.14372623e-01 -1.21130872e+00
-2.67835222e-02 -4.43614632e-01 -4.16073829e-01 2.15649828e-01
2.62996763e-01 2.36793682e-01 3.75586987e-01 -1.71323255e-01
-2.24493474e-01 6.94021165e-01 7.50974298e-01 3.72746736e-02
-3.71529520e-01 -3.20926696e-01 -8.28910112e-01 4.88745272e-01
7.97221005e-01 -5.28759181e-01 -5.63031197e-01 1.56716369e-02
8.94496858e-01 -4.94585276e-01 5.49345315e-01 -1.17269850e+00
2.45770216e-01 -2.49589630e-03 1.72153246e-02 3.78058821e-01
1.32652462e-01 -8.98739278e-01 7.26172447e-01 9.49139357e-01
-4.77358460e-01 3.15710962e-01 1.63719609e-01 3.28764439e-01
-2.88451940e-01 -9.22127068e-01 6.85603619e-01 -2.75992513e-01
-8.77169490e-01 -3.69918674e-01 -2.92143792e-01 -8.64649098e-03
1.14621186e+00 -8.16823184e-01 -3.39946687e-01 -3.68594415e-02
-9.53402102e-01 -2.19091654e-01 9.19018090e-01 4.34175193e-01
3.58343333e-01 -1.20425272e+00 -5.89979291e-01 -9.56707001e-02
8.66266191e-02 -6.48450792e-01 1.56770963e-02 7.24446535e-01
-6.76004350e-01 -6.07130490e-02 -6.44781351e-01 -9.75487083e-02
-8.31086278e-01 7.26741552e-01 5.93308568e-01 1.07855171e-01
-3.96684289e-01 6.83582366e-01 -4.05891180e-01 -4.78016436e-01
-2.17471913e-01 -4.26697731e-03 -4.66867238e-01 1.75326750e-01
2.88000047e-01 6.04387820e-01 1.74881712e-01 -3.54193658e-01
-4.33515966e-01 4.58932996e-01 4.96317923e-01 -2.62025088e-01
1.26407576e+00 -1.20504843e-02 -8.90793502e-01 3.92257661e-01
7.34531999e-01 -9.73101705e-02 -2.58136213e-01 1.02981053e-01
3.23857069e-01 -2.15739146e-01 -1.14689685e-01 -9.54465091e-01
-7.92985737e-01 2.87144721e-01 7.55365729e-01 6.64628744e-01
1.24059486e+00 -1.55154914e-01 2.11050436e-01 4.38566178e-01
3.74232203e-01 -1.21746647e+00 -2.25495517e-01 3.07523161e-01
8.64767790e-01 -3.56728703e-01 -1.76377878e-01 -2.44246259e-01
-1.48480132e-01 1.28067315e+00 1.46802470e-01 -1.02218889e-01
3.97098780e-01 1.76528156e-01 -4.39512223e-01 -3.01630825e-01
-4.81004417e-01 -4.25702453e-01 4.31953147e-02 1.01676631e+00
1.27935797e-01 -1.28514796e-01 -1.05971622e+00 4.70510691e-01
-5.09207129e-01 4.27661352e-02 4.96113867e-01 1.13453782e+00
-3.88705611e-01 -1.38832319e+00 -3.66937995e-01 4.38651331e-02
1.06207341e-01 -2.12879866e-01 -8.26107085e-01 1.01032615e+00
6.57762647e-01 8.83336484e-01 1.87390313e-01 -4.96933877e-01
3.49753439e-01 6.27971292e-02 7.43658602e-01 -3.65131259e-01
-1.03270960e+00 -9.07364488e-01 5.04869893e-02 -4.68668193e-01
-8.91330481e-01 -2.92636782e-01 -1.14366162e+00 -6.81576252e-01
-1.69234708e-01 6.17293954e-01 7.85743475e-01 1.15763676e+00
3.29408944e-01 7.94044673e-01 3.59024704e-01 -5.97990751e-01
-2.97157437e-01 -7.22177684e-01 -6.96694553e-01 5.61148643e-01
-2.71657079e-01 -8.69317651e-01 -3.01923186e-01 -1.10395707e-01]
|
[5.834269046783447, 3.807438850402832]
|
b516335e-575b-4960-9db6-dc3d2fcb07ce
|
ecgbert-understanding-hidden-language-of-ecgs
|
2306.06340
| null |
https://arxiv.org/abs/2306.06340v1
|
https://arxiv.org/pdf/2306.06340v1.pdf
|
ECGBERT: Understanding Hidden Language of ECGs with Self-Supervised Representation Learning
|
In the medical field, current ECG signal analysis approaches rely on supervised deep neural networks trained for specific tasks that require substantial amounts of labeled data. However, our paper introduces ECGBERT, a self-supervised representation learning approach that unlocks the underlying language of ECGs. By unsupervised pre-training of the model, we mitigate challenges posed by the lack of well-labeled and curated medical data. ECGBERT, inspired by advances in the area of natural language processing and large language models, can be fine-tuned with minimal additional layers for various ECG-based problems. Through four tasks, including Atrial Fibrillation arrhythmia detection, heartbeat classification, sleep apnea detection, and user authentication, we demonstrate ECGBERT's potential to achieve state-of-the-art results on a wide variety of tasks.
|
['Fatemeh Afghah', 'Fatemeh Khadem', 'Haben G. Yhdego', 'Phillip Si', 'Sajad Mousavi', 'Seokmin Choi']
|
2023-06-10
| null | null | null | null |
['arrhythmia-detection', 'sleep-apnea-detection', 'heartbeat-classification', 'unsupervised-pre-training']
|
['medical', 'medical', 'medical', 'methodology']
|
[ 4.40605819e-01 3.28817219e-01 -1.82125792e-01 -6.51861012e-01
-9.79867458e-01 -4.15473133e-01 -6.56199604e-02 6.84689462e-01
-5.04938304e-01 5.77772439e-01 3.47799510e-01 -6.06939971e-01
-4.05786224e-02 -3.96362394e-01 -1.87797919e-01 -2.09189594e-01
-4.90081936e-01 4.01124835e-01 -4.04640108e-01 -8.59515667e-02
-1.33263841e-01 3.64987075e-01 -6.98917389e-01 4.84864235e-01
3.87741685e-01 9.29922223e-01 -5.85228324e-01 1.08411193e+00
1.33010045e-01 1.20505762e+00 -7.99045682e-01 -4.56574336e-02
-4.87526767e-02 -5.47868788e-01 -9.56433117e-01 -1.41668975e-01
6.25611171e-02 -1.60291284e-01 -3.54578257e-01 4.81714129e-01
1.12649310e+00 -4.47250932e-01 3.08753699e-01 -8.55797827e-01
-4.46871340e-01 8.33747029e-01 -2.18577161e-01 5.17206311e-01
8.18030015e-02 1.18355930e-01 7.15071023e-01 -6.26653016e-01
3.45748067e-01 6.40279770e-01 1.36649799e+00 7.50385344e-01
-1.28655839e+00 -5.52841842e-01 -4.01919037e-01 -2.37501502e-01
-1.59050643e+00 -6.11130834e-01 6.38480365e-01 -4.99469906e-01
7.51888156e-01 2.01943144e-01 5.16591132e-01 1.12914073e+00
2.08619341e-01 4.83345687e-01 8.66498947e-01 -4.78583813e-01
2.12262541e-01 -1.48565665e-01 3.65558535e-01 7.42049634e-01
1.51848614e-01 -1.77089125e-02 -5.55949152e-01 -5.63515663e-01
5.89304686e-01 -9.73668322e-03 -2.29519203e-01 -1.62448123e-01
-1.36346936e+00 7.11660981e-01 5.46052307e-02 4.46666092e-01
-4.30436581e-01 1.23086408e-01 9.70763683e-01 4.17500854e-01
3.86761516e-01 8.33746076e-01 -7.91403234e-01 -1.88748866e-01
-1.26663053e+00 -1.26911938e-01 8.87791872e-01 6.35788441e-01
2.58388102e-01 2.92835712e-01 -4.50877577e-01 6.77041888e-01
1.77504033e-01 2.83560157e-01 6.20759845e-01 -6.69684112e-01
1.15799136e-01 5.44577360e-01 -1.77093089e-01 -6.74607813e-01
-9.44479465e-01 -9.56399024e-01 -1.19724226e+00 -6.13771826e-02
1.36294320e-01 -5.00980496e-01 -8.90223920e-01 1.63899815e+00
-1.28298849e-01 2.64343023e-01 3.32572788e-01 5.51540256e-01
1.17409897e+00 3.61559987e-02 3.40881079e-01 -2.82763720e-01
1.31562805e+00 -4.52284306e-01 -6.46454871e-01 -2.37076700e-01
6.35933042e-01 -3.42846423e-01 8.86668146e-01 4.19278264e-01
-1.00578082e+00 -5.92687726e-01 -1.13704753e+00 -8.49084407e-02
-1.58238620e-01 1.23251364e-01 4.63902593e-01 9.73530054e-01
-1.13005269e+00 6.78928733e-01 -9.86366928e-01 -1.78737506e-01
9.03479397e-01 4.93953317e-01 -2.91208506e-01 2.39935860e-01
-1.33053505e+00 7.19900966e-01 1.48568138e-01 1.57810479e-01
-1.03792858e+00 -6.88754320e-01 -8.92511904e-01 1.17827905e-02
8.63982141e-02 -7.91576028e-01 1.21186852e+00 -6.47421539e-01
-1.05178988e+00 1.34785390e+00 1.34629399e-01 -1.02171314e+00
4.44474667e-01 -2.96951920e-01 -5.94019413e-01 3.49142522e-01
-1.22765237e-02 2.38656744e-01 9.03557301e-01 -6.95684433e-01
-1.42003477e-01 -2.19894275e-01 -2.73925126e-01 -2.24953592e-01
-5.24944663e-01 1.50328994e-01 -8.95147696e-02 -8.72517526e-01
-3.17455560e-01 -7.06421554e-01 -6.59559429e-01 -2.89581064e-02
-4.03170407e-01 1.35043273e-02 3.81350428e-01 -5.82263827e-01
1.32259572e+00 -2.36167884e+00 -2.46164739e-01 2.66024351e-01
9.01846886e-01 5.14077783e-01 -9.39399675e-02 2.80692160e-01
-3.37195009e-01 3.89787644e-01 -3.83822441e-01 -4.36455160e-01
-3.39749604e-01 2.21358553e-01 -2.55300850e-01 5.33431888e-01
2.21181989e-01 1.20041370e+00 -9.82471585e-01 -4.81794327e-01
6.70064986e-03 5.54347396e-01 -3.33918333e-01 4.31992710e-01
2.53761441e-01 7.32205451e-01 -2.89197534e-01 4.17049736e-01
9.31631252e-02 -6.05550408e-01 3.91894728e-01 -1.92180678e-01
3.03089291e-01 4.00897264e-01 -7.91205883e-01 2.12742615e+00
-2.35014156e-01 5.35075784e-01 -1.26496643e-01 -1.18054938e+00
7.66997099e-01 7.68239617e-01 9.90518808e-01 -2.17631727e-01
2.75813550e-01 1.21554546e-03 1.57649722e-02 -7.66094446e-01
-1.25740588e-01 -3.82280916e-01 -1.96606040e-01 7.65506625e-01
2.07894430e-01 1.68070838e-01 -1.19514972e-01 2.82166183e-01
1.48148966e+00 -1.31280854e-01 6.10446811e-01 -3.50241244e-01
4.12916809e-01 -2.27179423e-01 5.52059531e-01 1.06522036e+00
-4.27459806e-01 8.80601287e-01 2.42801145e-01 -8.62407923e-01
-5.33317566e-01 -9.77565169e-01 -3.22482407e-01 8.16971362e-01
-4.81356055e-01 -9.86075342e-01 -7.18203545e-01 -7.38750994e-01
-2.34580904e-01 1.57609120e-01 -7.71071613e-01 -5.12519598e-01
-6.00505650e-01 -1.17411184e+00 1.45073628e+00 8.56295943e-01
2.16672927e-01 -1.28301501e+00 -1.19044268e+00 5.92476487e-01
-3.41589481e-01 -1.22775292e+00 -2.80052364e-01 5.23968637e-01
-1.00023627e+00 -1.28376937e+00 -6.02723002e-01 -8.38036060e-01
4.40877974e-01 -4.57786292e-01 1.71481192e+00 2.52763808e-01
-8.65008295e-01 5.92949450e-01 -2.87639797e-01 -8.25067222e-01
-6.55232012e-01 3.08985323e-01 7.87025765e-02 -2.06315480e-02
4.61520433e-01 -7.04241037e-01 -7.04560637e-01 -3.16652358e-01
-6.15220249e-01 -2.78111130e-01 4.01651084e-01 8.39770496e-01
4.66536075e-01 -2.38143921e-01 1.14320993e+00 -1.33343506e+00
8.71047020e-01 -2.59368509e-01 1.59177125e-01 1.27098843e-01
-8.34894061e-01 -6.85981065e-02 4.97629881e-01 -4.23543483e-01
-9.20511037e-02 -2.59782411e-02 -3.71652991e-01 -3.12529355e-01
-3.35074246e-01 8.97948265e-01 2.97429353e-01 9.23050418e-02
1.31153798e+00 1.73109949e-01 1.07400008e-01 -3.93953770e-01
1.81910098e-01 8.57494473e-01 9.38831270e-01 -3.31790298e-01
5.91774762e-01 3.30317736e-01 -1.77035660e-01 -6.73917234e-01
-1.04161561e+00 -4.48329419e-01 -7.32746840e-01 2.37361208e-01
9.80017126e-01 -1.11594772e+00 -5.04770815e-01 3.17172438e-01
-8.32898378e-01 -5.05697608e-01 -8.71685088e-01 1.96936652e-01
-4.16375577e-01 5.61938941e-01 -8.41780961e-01 -8.23168039e-01
-1.09709525e+00 -3.80316019e-01 1.05040777e+00 -3.08625102e-01
-8.21344137e-01 -9.75205958e-01 1.98218256e-01 1.41131312e-01
5.50906420e-01 4.69044715e-01 1.07213187e+00 -9.04784977e-01
1.65115193e-01 -3.77510458e-01 1.66779742e-01 5.95290005e-01
5.18815219e-01 -6.39972568e-01 -1.23274565e+00 -4.10776675e-01
1.91354230e-01 -6.02315485e-01 5.79382539e-01 3.10359865e-01
1.35813868e+00 -2.01682642e-01 -5.88936396e-02 6.89579010e-01
8.48458290e-01 -8.26487839e-02 6.65994346e-01 -1.15104839e-01
7.12089181e-01 8.01860467e-02 -6.35498986e-02 5.71633339e-01
4.48386669e-01 -5.83952926e-02 -7.26833194e-02 -9.04231966e-01
-6.22553900e-02 5.71193472e-02 -1.10926427e-01 7.23853528e-01
-1.67009234e-02 7.85004646e-02 -1.30850697e+00 6.01487696e-01
-1.71457398e+00 -5.78851581e-01 2.01348200e-01 2.00693202e+00
1.27537870e+00 2.15610594e-01 3.11799377e-01 6.51478171e-01
2.62592942e-01 -6.32642433e-02 -6.34670019e-01 -1.33602500e-01
-1.57255344e-02 7.80288339e-01 1.05612233e-01 7.59919956e-02
-1.45808339e+00 5.96599758e-01 7.84858322e+00 1.53036034e-02
-1.20490158e+00 1.96334183e-01 8.35318387e-01 1.47191599e-01
2.56062508e-01 -5.57202816e-01 -1.03516364e-02 1.19273566e-01
1.30069995e+00 -4.82901093e-03 3.11603785e-01 5.11004329e-01
2.23345295e-01 6.41297221e-01 -1.39717042e+00 1.51334226e+00
2.13478059e-01 -1.48495376e+00 -3.37285846e-01 -3.63672763e-01
1.99568391e-01 4.30727780e-01 -1.71072319e-01 3.19320768e-01
9.92683601e-03 -1.31234550e+00 1.95591688e-01 4.18587685e-01
1.39627409e+00 -4.11316425e-01 8.89298737e-01 2.14332983e-01
-9.12823498e-01 -1.88746706e-01 7.05277100e-02 -5.43256812e-02
-7.63794556e-02 6.19372904e-01 -8.46921265e-01 4.10871714e-01
7.61331677e-01 1.10738170e+00 -7.19196260e-01 8.37144196e-01
-9.00857970e-02 1.12066758e+00 4.82937396e-02 4.62500781e-01
-3.09817612e-01 5.27826846e-01 2.21321508e-01 1.40033650e+00
-2.11569920e-01 1.53115660e-01 6.48010910e-01 6.67710781e-01
-2.42264315e-01 2.78590411e-01 -5.84475219e-01 -3.41622770e-01
3.27882380e-03 1.16312551e+00 -5.98700345e-01 -5.27966559e-01
-2.38825083e-01 7.47939944e-01 8.96818042e-02 3.07841420e-01
-5.91515601e-01 -3.73228103e-01 1.76282555e-01 3.42668533e-01
-2.19962716e-01 -2.99577950e-03 -7.15536714e-01 -1.31540251e+00
-3.01069736e-01 -1.32578254e+00 7.48042285e-01 -5.19821942e-01
-1.48771918e+00 9.37820733e-01 -4.45052862e-01 -1.20136130e+00
-4.79802489e-01 -3.82553607e-01 -4.26031321e-01 7.04318881e-01
-1.34456384e+00 -1.10680568e+00 -7.38733783e-02 7.88287342e-01
2.09597155e-01 -3.60774606e-01 1.79296267e+00 6.97948694e-01
-2.65299261e-01 7.97403991e-01 -5.79275548e-01 8.54835689e-01
8.08210790e-01 -1.23497403e+00 5.62983453e-01 6.14627957e-01
4.33744162e-01 8.00915360e-01 2.93184727e-01 -3.91600162e-01
-1.05604446e+00 -1.15285635e+00 1.08284652e+00 -7.19319880e-01
3.90944064e-01 -4.25477803e-01 -8.17655206e-01 6.42712414e-01
5.56381531e-02 5.42445242e-01 1.43780303e+00 5.24066985e-01
-4.62248027e-01 -1.59563288e-01 -9.94753242e-01 1.66544124e-01
8.83074701e-01 -1.04668009e+00 -7.41049707e-01 4.23109978e-01
2.84693331e-01 -3.88290077e-01 -1.07383704e+00 7.02774584e-01
5.29700935e-01 -3.82514119e-01 1.07134354e+00 -1.20671725e+00
1.15111738e-01 -5.01870997e-02 1.96132749e-01 -9.12097692e-01
-2.72145152e-01 -1.04136765e+00 -3.84708017e-01 8.10341656e-01
4.75364536e-01 -3.55509400e-01 7.81326771e-01 5.12454450e-01
-8.09136406e-02 -7.06353068e-01 -7.14169741e-01 -1.97172269e-01
-2.10677925e-02 -5.96233249e-01 2.45239303e-01 1.23133707e+00
3.78831774e-01 8.14300895e-01 -6.42202556e-01 3.38951200e-02
4.68426079e-01 6.79366365e-02 5.84286153e-01 -1.57972562e+00
-4.23785448e-01 1.45835623e-01 -6.56764448e-01 -4.20648694e-01
-9.78196971e-03 -1.11700070e+00 2.74670739e-02 -1.55372274e+00
9.44178328e-02 -5.17107487e-01 -1.05576503e+00 1.07766676e+00
-2.41689041e-01 7.21501410e-01 -5.15183806e-02 3.43772359e-02
-7.37752676e-01 1.44191727e-03 5.87985635e-01 -3.04370165e-01
-4.04725641e-01 8.52288827e-02 -1.23420143e+00 8.46754491e-01
9.90067303e-01 -6.28990114e-01 -5.18689752e-01 -4.32258517e-01
2.57877082e-01 1.19214304e-01 2.77511626e-01 -1.22976553e+00
2.85288673e-02 3.59966099e-01 5.38866043e-01 -1.05257317e-01
8.61783549e-02 -6.60079837e-01 -3.35246027e-01 7.26864398e-01
-8.31368148e-01 1.34521067e-01 3.19242120e-01 4.06677634e-01
-6.74555600e-02 3.19593966e-01 7.29611039e-01 -2.55341858e-01
-2.05705643e-01 3.53625506e-01 -7.23424435e-01 5.67715704e-01
4.03695375e-01 2.01420680e-01 -1.76474247e-02 -4.37734902e-01
-1.24290657e+00 1.45218387e-01 -2.66310543e-01 4.51545537e-01
7.71847010e-01 -8.89104128e-01 -1.02247751e+00 5.09124100e-01
3.10425371e-01 -5.89067973e-02 8.25336874e-02 7.19032288e-01
-4.55608994e-01 -7.48487888e-03 -4.20265608e-02 -7.14320421e-01
-1.15946472e+00 4.93853718e-01 6.15921140e-01 -4.92938668e-01
-1.16461170e+00 4.71160620e-01 -2.24330664e-01 -4.64886785e-01
4.47176039e-01 -5.02240419e-01 -3.21181148e-01 -2.02791750e-01
7.91244686e-01 -2.37443391e-02 2.40929246e-01 -2.57237554e-01
-5.54377854e-01 2.45170712e-01 3.40314433e-02 1.32974938e-01
1.34067297e+00 7.74784535e-02 9.05276649e-03 6.41918242e-01
8.59164417e-01 -9.27134380e-02 -6.48898721e-01 -8.58956948e-02
1.89609647e-01 2.15540096e-01 -5.10917325e-03 -1.21364415e+00
-9.03671622e-01 1.12534666e+00 1.10730600e+00 -1.12013323e-02
1.26055610e+00 -1.61878377e-01 8.40756536e-01 6.20287418e-01
2.66833514e-01 -7.22543716e-01 -2.69156452e-02 3.07095766e-01
4.82122779e-01 -1.16500425e+00 2.56524198e-02 -7.25432113e-02
-6.17406070e-01 9.88870323e-01 9.98906866e-02 -8.07887129e-03
1.05236304e+00 6.30065262e-01 8.55330765e-01 -3.72296661e-01
-4.35708165e-01 2.52798554e-02 2.38048509e-01 6.46293759e-01
7.15692520e-01 3.01720500e-02 -6.33521378e-02 1.11918902e+00
-5.78757450e-02 6.22496724e-01 3.92661005e-01 1.08090603e+00
1.67033389e-01 -1.24303591e+00 9.54064131e-02 6.35312617e-01
-1.16854084e+00 -4.97546107e-01 -3.35424155e-01 2.11416200e-01
1.75651908e-01 1.11605299e+00 -3.10189217e-01 -3.71016800e-01
2.20535412e-01 4.03165936e-01 2.59496242e-01 -1.12637842e+00
-1.04597580e+00 1.81301877e-01 1.28320798e-01 -2.30071083e-01
-5.32129169e-01 -2.19571546e-01 -1.44169331e+00 4.22873795e-01
1.40454203e-01 3.04451615e-01 2.01862827e-01 1.01809847e+00
7.18808591e-01 8.89418066e-01 2.50277579e-01 -2.71161765e-01
-5.04040003e-01 -9.83797729e-01 -6.31250381e-01 5.74531496e-01
8.64703059e-01 7.30388090e-02 1.32806972e-01 4.25954759e-01]
|
[14.372573852539062, 3.3714871406555176]
|
dfd6311d-471d-4260-90e7-e4092415f714
|
evaluation-and-comparison-of-eight-popular
|
2208.02063
| null |
https://arxiv.org/abs/2208.02063v1
|
https://arxiv.org/pdf/2208.02063v1.pdf
|
Evaluation and comparison of eight popular Lidar and Visual SLAM algorithms
|
In this paper, we evaluate eight popular and open-source 3D Lidar and visual SLAM (Simultaneous Localization and Mapping) algorithms, namely LOAM, Lego LOAM, LIO SAM, HDL Graph, ORB SLAM3, Basalt VIO, and SVO2. We have devised experiments both indoor and outdoor to investigate the effect of the following items: i) effect of mounting positions of the sensors, ii) effect of terrain type and vibration, iii) effect of motion (variation in linear and angular speed). We compare their performance in terms of relative and absolute pose error. We also provide comparison on their required computational resources. We thoroughly analyse and discuss the results and identify the best performing system for the environment cases with our multi-camera and multi-Lidar indoor and outdoor datasets. We hope our findings help one to choose a sensor and the corresponding SLAM algorithm combination suiting their needs, based on their target environment.
|
['Reza Ghabcheloo', 'Nataliya Strokina', 'Bharath Garigipati']
|
2022-08-03
| null | null | null | null |
['simultaneous-localization-and-mapping']
|
['computer-vision']
|
[-2.22948849e-01 -6.37910664e-01 -2.49461476e-02 -2.91324943e-01
-2.60961622e-01 -6.03579462e-01 6.85700595e-01 2.52202511e-01
-6.11801207e-01 1.15115857e+00 -1.41248301e-01 -3.21034908e-01
-4.14644569e-01 -8.19328189e-01 -4.56373841e-01 -2.73194969e-01
-3.53992313e-01 8.50665271e-01 5.38288057e-01 -3.67141843e-01
5.73120356e-01 9.57359374e-01 -1.81341219e+00 -6.85621321e-01
1.10908473e+00 6.65876746e-01 4.89030570e-01 6.49205565e-01
-9.60098356e-02 1.04031250e-01 -3.22205722e-01 2.18403116e-01
3.87457520e-01 -8.08523074e-02 -3.39823127e-01 -1.99551910e-01
4.37674969e-01 1.21380612e-01 2.12665543e-01 7.92536616e-01
8.24651778e-01 -4.18401696e-02 6.81809366e-01 -1.37838244e+00
1.61388710e-01 -3.77440942e-03 -6.53420985e-01 4.95579615e-02
8.72867942e-01 8.28010496e-03 3.16018462e-01 -8.05064082e-01
1.01696932e+00 1.05869710e+00 1.06731606e+00 -1.16737127e-01
-1.30886269e+00 -6.60262048e-01 -3.31889570e-01 1.56886309e-01
-1.91056037e+00 -4.27110642e-01 4.98407722e-01 -3.54985178e-01
8.01818669e-01 2.98899680e-01 5.60131729e-01 8.55891287e-01
8.06645393e-01 -1.23557433e-01 1.59318769e+00 -5.26692748e-01
3.50071222e-01 3.77328128e-01 -1.86087564e-01 7.55739391e-01
8.16625059e-01 1.08721152e-01 -5.56407511e-01 -1.79246932e-01
4.64749724e-01 -3.67369980e-01 -1.36993930e-01 -9.06255245e-01
-1.10672712e+00 6.43790305e-01 4.47418183e-01 3.21771443e-01
-1.57425061e-01 1.68794706e-01 1.39895052e-01 2.07399935e-01
1.40070841e-01 3.26139033e-01 -4.72184271e-01 -1.56269208e-01
-1.00233912e+00 2.40555614e-01 7.43740082e-01 1.16360652e+00
1.20717442e+00 -1.50042802e-01 5.45459151e-01 5.32624722e-01
5.53469121e-01 8.64917517e-01 3.20452094e-01 -6.29885793e-01
4.42831516e-01 3.61929923e-01 4.61339265e-01 -9.49085832e-01
-7.06432581e-01 -6.22573011e-02 -2.64934272e-01 3.29767197e-01
1.13312423e-01 -2.06237629e-01 -8.12922597e-01 1.12332141e+00
3.61221403e-01 -1.19991183e-01 -1.65770203e-02 7.35812187e-01
6.33179665e-01 2.29880631e-01 -2.86612272e-01 -1.18567474e-01
1.00905299e+00 -3.84747475e-01 -6.70031667e-01 -4.61655915e-01
6.62415087e-01 -1.08185971e+00 8.17553997e-01 5.12158871e-02
-4.00146335e-01 -5.34945667e-01 -1.18647242e+00 1.23571210e-01
-6.67997181e-01 2.32786134e-01 4.49020058e-01 5.56088090e-01
-1.07625389e+00 5.36350369e-01 -8.03079665e-01 -1.08942461e+00
-2.80656934e-01 4.49437737e-01 -3.07737768e-01 7.37161189e-02
-8.35871696e-01 1.47728050e+00 3.21530849e-01 2.87077930e-02
-5.41718185e-01 -1.98575050e-01 -7.72135019e-01 -6.19404554e-01
1.84533328e-01 -6.25593483e-01 6.19526207e-01 -1.22214653e-01
-1.23527944e+00 8.38401377e-01 -3.68663758e-01 -4.15903777e-01
8.09859753e-01 -3.24881524e-01 -4.51154739e-01 -3.41919184e-01
5.84267199e-01 5.59688330e-01 1.50810510e-01 -1.47858405e+00
-5.46815455e-01 -5.82262158e-01 -4.56021547e-01 2.57974356e-01
5.72571754e-01 -2.86031663e-01 5.13875373e-02 1.28229931e-01
6.20207667e-01 -1.08056092e+00 -2.18984753e-01 -5.69967814e-02
-3.02450091e-01 1.58361033e-01 1.05532002e+00 -1.44931525e-01
6.86469436e-01 -2.10183907e+00 -1.41702935e-01 3.20394337e-01
-4.21782404e-01 -1.97177410e-01 2.21565887e-01 9.35346484e-01
5.04457653e-01 6.79030363e-03 6.76600337e-02 -3.20655733e-01
-1.49023503e-01 6.20351255e-01 1.92328990e-02 8.79207671e-01
-4.15352941e-01 4.78785783e-01 -6.21837854e-01 -8.66090596e-01
7.59208143e-01 4.33943093e-01 -9.68781486e-02 -1.11263335e-01
2.58885741e-01 6.36932850e-01 -4.99885529e-01 9.30858195e-01
8.83291304e-01 4.36603963e-01 -1.28978083e-03 -2.50960998e-02
-9.32331085e-01 2.73600101e-01 -1.74476552e+00 1.81177688e+00
-7.41372645e-01 6.68328345e-01 1.01202369e-01 -9.01624560e-02
1.42369485e+00 1.61606185e-02 3.73437166e-01 -5.93731761e-01
1.30214855e-01 8.88822198e-01 -2.60012716e-01 -3.44656378e-01
9.13592935e-01 -2.71573127e-03 -8.21828321e-02 -7.41650164e-02
-2.62951907e-02 -5.03603697e-01 7.76857585e-02 -1.56039000e-01
8.09909105e-01 4.66791600e-01 7.10609019e-01 -7.64729738e-01
5.58267474e-01 1.35737926e-01 1.39206931e-01 6.48041308e-01
-3.86368364e-01 2.85748273e-01 6.56949580e-02 -3.69629472e-01
-9.13862288e-01 -1.09302783e+00 -5.36665499e-01 5.12656391e-01
8.69482577e-01 -3.82107705e-01 9.86410379e-02 2.14950852e-02
5.45894861e-01 5.32942533e-01 -2.76766539e-01 3.35421711e-01
-4.06851053e-01 -5.18118441e-01 3.85703802e-01 8.50288644e-02
5.26365340e-01 -6.83473766e-01 -1.22653377e+00 2.39315093e-01
1.60284489e-02 -1.15939140e+00 2.61005193e-01 6.02690935e-01
-8.50367665e-01 -9.51519728e-01 -2.03638580e-02 -3.62736642e-01
3.89088690e-01 3.26608002e-01 7.92458177e-01 -1.04682304e-01
-3.02951723e-01 4.58339512e-01 -3.90007377e-01 -4.95192230e-01
1.78553872e-02 8.21076706e-02 4.46190953e-01 -4.15637106e-01
1.07680134e-01 -8.16444576e-01 -1.58500269e-01 4.95204419e-01
-2.44562835e-01 -2.98537493e-01 4.47969437e-01 1.86490357e-01
6.64502740e-01 -1.33085385e-01 -3.09172302e-01 -4.72720653e-01
4.72454518e-01 -3.53763849e-01 -1.03979993e+00 -7.25771412e-02
-8.66082788e-01 4.47152629e-02 7.29506090e-02 7.00440630e-02
-4.31474090e-01 3.34844857e-01 1.98995862e-02 3.85243609e-03
-4.22977298e-01 3.26945364e-01 -7.12733641e-02 -8.67000103e-01
6.81388021e-01 1.28821284e-01 -2.83979148e-01 -5.05397201e-01
2.82774955e-01 7.54350126e-01 4.98608500e-01 -5.71651399e-01
1.08609891e+00 6.78721905e-01 5.16352534e-01 -1.21552980e+00
-2.34694973e-01 -5.15689373e-01 -1.11080921e+00 -3.63705695e-01
8.18003774e-01 -8.29540730e-01 -5.66582322e-01 2.36126527e-01
-1.05943215e+00 -2.84042978e-03 -2.16989499e-02 6.33970976e-01
-4.93378043e-01 3.51049304e-01 -7.55738020e-02 -1.06585312e+00
4.31133397e-02 -1.27747381e+00 1.21774960e+00 2.78287888e-01
-3.41435373e-01 -9.81818259e-01 3.67196202e-01 -1.00764511e-02
5.17814577e-01 8.13523591e-01 2.80646026e-01 -1.01283580e-01
-7.23696470e-01 -1.37586087e-01 -1.14764810e-01 -4.25575137e-01
1.43690646e-01 1.15600169e-01 -6.95825875e-01 -4.31280881e-01
-3.64713311e-01 -1.18118003e-01 4.92674649e-01 3.31275731e-01
2.53197402e-01 7.51596987e-02 -6.74521089e-01 8.49858463e-01
2.11708546e+00 2.81101894e-02 5.31714201e-01 9.77439821e-01
6.12988532e-01 4.33481544e-01 1.04270637e+00 2.53040701e-01
4.19554532e-01 9.93327975e-01 8.35645080e-01 1.01001233e-01
1.00339472e-01 -2.89468229e-01 2.18937337e-01 4.68000323e-01
-4.11122978e-01 -1.30643100e-01 -1.16937733e+00 4.29670811e-01
-1.59462619e+00 -4.52832758e-01 -7.31296897e-01 2.49367666e+00
-2.18421649e-02 1.81773722e-01 -8.75180811e-02 2.14722782e-01
4.67978835e-01 2.95136243e-01 6.73847795e-02 -5.74801326e-01
-1.02700971e-01 1.20897211e-01 1.30793273e+00 8.86438072e-01
-9.01526034e-01 9.56426501e-01 6.48619270e+00 2.68649429e-01
-1.30090714e+00 2.01168835e-01 -5.96579075e-01 2.19365627e-01
-1.62571985e-02 3.96070153e-01 -1.04058945e+00 5.35395980e-01
1.00545716e+00 -5.54439239e-02 1.54506341e-01 7.73495018e-01
1.11113206e-01 -9.30835009e-01 -7.08356440e-01 7.63283610e-01
-1.18589014e-01 -8.19682479e-01 -3.23451817e-01 4.56032068e-01
4.58339661e-01 6.31175280e-01 -5.25728524e-01 -4.75447737e-02
-1.13625728e-01 -7.17944384e-01 8.92334938e-01 3.53832543e-01
6.13666594e-01 -5.71125627e-01 9.59561646e-01 4.33079422e-01
-1.63243735e+00 2.76430130e-01 -2.85044044e-01 -4.20811921e-01
3.32271039e-01 3.92108381e-01 -9.81356323e-01 1.14355278e+00
6.05915964e-01 4.04439628e-01 -7.60738969e-01 1.16866732e+00
-3.86529744e-01 -9.10018086e-02 -7.41455555e-01 -3.53052139e-01
2.98156261e-01 -4.88466561e-01 7.57446587e-01 1.11383188e+00
6.69872940e-01 -5.52896023e-01 2.88371474e-01 6.14640772e-01
7.19795167e-01 1.38038620e-01 -1.01875782e+00 5.40471673e-01
8.71241570e-01 1.00535262e+00 -7.94773757e-01 -1.22681586e-02
-1.02574274e-01 7.33200490e-01 -2.48944581e-01 4.59995717e-02
-7.02455223e-01 -3.99392575e-01 6.65532768e-01 4.75235552e-01
1.19453669e-01 -8.54048193e-01 -3.16569656e-01 -8.99782658e-01
3.02566774e-02 -1.84279636e-01 -9.31342132e-03 -1.01609933e+00
-5.66107929e-01 3.20882291e-01 2.05795005e-01 -1.42390811e+00
-1.77655965e-01 -5.26089966e-01 -3.81377667e-01 8.44857037e-01
-1.47497141e+00 -1.12400222e+00 -5.71645498e-01 3.64475757e-01
1.44383982e-01 1.21726140e-01 7.05410004e-01 4.47917998e-01
-6.11698590e-02 -1.39723256e-01 1.43027797e-01 -4.06596392e-01
7.17337668e-01 -1.09436095e+00 1.65080860e-01 8.98032546e-01
1.45516828e-01 4.16038066e-01 1.25692976e+00 -9.43166256e-01
-1.61482179e+00 -7.25817323e-01 1.25354409e+00 -6.20900273e-01
6.50446296e-01 -6.19772911e-01 -4.56624687e-01 7.72794366e-01
-1.27041817e-01 -4.45647016e-02 6.00159690e-02 2.32596248e-01
1.86128110e-01 -2.64586389e-01 -1.32454419e+00 2.70158768e-01
1.09992528e+00 -4.00942534e-01 -4.35806602e-01 3.08751911e-01
3.36184472e-01 -8.49371254e-01 -7.62529790e-01 5.83247781e-01
6.69164062e-01 -1.48014295e+00 8.59640539e-01 2.77803600e-01
-4.31366771e-01 -6.60183966e-01 -5.42400599e-01 -1.08243692e+00
-1.19994141e-01 -2.16688588e-01 5.20537734e-01 1.22747087e+00
2.12712228e-01 -1.12959933e+00 5.37453115e-01 -1.58647388e-01
-1.20517820e-01 -4.49468493e-01 -1.32202184e+00 -1.04774249e+00
-5.79967141e-01 -3.06377470e-01 5.46738803e-01 8.04835439e-01
-4.55570400e-01 2.29412451e-01 -4.25006658e-01 4.94427800e-01
6.41031384e-01 2.22063959e-01 1.40021706e+00 -1.59762073e+00
2.68622875e-01 3.30170505e-02 -1.02225137e+00 -4.06896055e-01
-2.38423035e-01 -4.78745997e-01 3.62414978e-02 -1.69936299e+00
-2.97293544e-01 -6.87506616e-01 1.95353866e-01 1.20538153e-01
3.99563491e-01 2.01416314e-01 4.59303632e-02 3.38681370e-01
-3.75633925e-01 2.66586810e-01 7.85722017e-01 4.14061219e-01
-2.86690682e-01 -1.12025060e-01 1.88026745e-02 6.09099329e-01
4.67063189e-01 -6.32009149e-01 1.70135498e-02 -4.41058457e-01
1.71491250e-01 -1.05684631e-01 4.52197701e-01 -1.65730417e+00
2.14274928e-01 -2.46227667e-01 3.73437524e-01 -9.55795705e-01
5.84057808e-01 -1.07796311e+00 6.36334300e-01 9.37630832e-01
5.84063053e-01 4.86408025e-01 2.52488077e-01 2.73352534e-01
-4.50343825e-02 -1.97593182e-01 8.21024716e-01 -4.86849435e-02
-1.15189958e+00 -3.03786732e-02 -1.77256674e-01 -4.59800750e-01
1.12056935e+00 -6.92464292e-01 -3.05564910e-01 -3.65585953e-01
-4.00021225e-01 3.64043385e-01 1.04926836e+00 3.36173087e-01
2.37056360e-01 -1.28312111e+00 -4.15709823e-01 3.97745728e-01
2.98628360e-01 -2.90757269e-01 -2.38871589e-01 1.03633416e+00
-1.03832996e+00 5.75984716e-01 -5.97052515e-01 -9.48417008e-01
-1.39160407e+00 2.77904212e-01 3.24938178e-01 3.68897170e-02
-2.98338294e-01 4.88510102e-01 -7.29762077e-01 -7.33676612e-01
-6.83212793e-03 -3.63997728e-01 2.12420791e-01 1.84786245e-02
-2.61105597e-01 7.41463006e-01 8.26237127e-02 -9.15369630e-01
-1.01186466e+00 1.34220302e+00 8.07497144e-01 -2.73479402e-01
9.72266138e-01 -4.46087271e-01 -1.54477820e-01 7.66891241e-01
8.26341748e-01 6.69146419e-01 -6.48308635e-01 3.49958926e-01
3.05848807e-01 -6.42854154e-01 -2.14762419e-01 -3.95639390e-01
-1.40855074e-01 5.68946898e-01 1.01540160e+00 7.97782093e-02
8.18150997e-01 -2.00428426e-01 2.05948472e-01 1.63523346e-01
1.33445942e+00 -8.81367624e-01 -5.52690327e-01 6.69184268e-01
6.03007913e-01 -1.10415602e+00 6.46175325e-01 -4.09279078e-01
-2.22653806e-01 9.70431983e-01 4.31028187e-01 -3.65194350e-01
4.23161924e-01 3.51314962e-01 3.48403424e-01 -1.91409320e-01
-1.58509612e-01 -5.08611858e-01 -6.75817728e-02 7.86021113e-01
3.42792302e-01 1.28883019e-01 -8.21583033e-01 -4.43563342e-01
-6.29403293e-01 -4.58056778e-02 4.71222550e-01 1.33764148e+00
-8.27128708e-01 -1.29715860e+00 -7.68065989e-01 5.67012392e-02
2.29838818e-01 4.99383748e-01 -5.75803280e-01 1.58619285e+00
7.15764523e-01 7.53340721e-01 -8.72064754e-02 -4.92838204e-01
5.83043098e-01 -1.67417619e-02 6.63294673e-01 -4.28755879e-01
-2.31970266e-01 -6.91422001e-02 2.94694126e-01 -7.04232633e-01
-3.72213304e-01 -8.50427449e-01 -1.14607739e+00 -4.12801921e-01
-5.25785923e-01 2.85645366e-01 1.39377868e+00 7.93038249e-01
3.00571740e-01 7.16230646e-02 3.10182214e-01 -1.22215772e+00
-2.62006581e-01 -1.05945563e+00 -9.08192873e-01 -1.77175969e-01
3.31069201e-01 -1.11098588e+00 -4.67943311e-01 -5.93851864e-01]
|
[7.356385231018066, -2.0439932346343994]
|
368e528c-7e36-4b1d-9965-a13ebf205a35
|
federated-few-shot-learning
|
2306.10234
| null |
https://arxiv.org/abs/2306.10234v3
|
https://arxiv.org/pdf/2306.10234v3.pdf
|
Federated Few-shot Learning
|
Federated Learning (FL) enables multiple clients to collaboratively learn a machine learning model without exchanging their own local data. In this way, the server can exploit the computational power of all clients and train the model on a larger set of data samples among all clients. Although such a mechanism is proven to be effective in various fields, existing works generally assume that each client preserves sufficient data for training. In practice, however, certain clients may only contain a limited number of samples (i.e., few-shot samples). For example, the available photo data taken by a specific user with a new mobile device is relatively rare. In this scenario, existing FL efforts typically encounter a significant performance drop on these clients. Therefore, it is urgent to develop a few-shot model that can generalize to clients with limited data under the FL scenario. In this paper, we refer to this novel problem as federated few-shot learning. Nevertheless, the problem remains challenging due to two major reasons: the global data variance among clients (i.e., the difference in data distributions among clients) and the local data insufficiency in each client (i.e., the lack of adequate local data for training). To overcome these two challenges, we propose a novel federated few-shot learning framework with two separately updated models and dedicated training strategies to reduce the adverse impact of global data variance and local data insufficiency. Extensive experiments on four prevalent datasets that cover news articles and images validate the effectiveness of our framework compared with the state-of-the-art baselines. Our code is provided at https://github.com/SongW-SW/F2L.
|
['Jundong Li', 'Huiyuan Chen', 'Chen Chen', 'Kaize Ding', 'Xingbo Fu', 'Song Wang']
|
2023-06-17
| null | null | null | null |
['few-shot-learning']
|
['methodology']
|
[-1.67940319e-01 -3.53221029e-01 -4.87457871e-01 -3.21962863e-01
-9.20005500e-01 -3.10666829e-01 4.90569532e-01 -2.55748779e-01
-9.57427323e-02 6.15747988e-01 2.45420009e-01 -2.62860693e-02
-5.43545000e-02 -8.17951620e-01 -6.86480343e-01 -8.36629689e-01
3.06323946e-01 3.14492017e-01 3.44318986e-01 8.37717298e-03
-1.14532327e-02 6.92755058e-02 -1.80638719e+00 4.11643028e-01
8.41698766e-01 1.01985264e+00 3.46274376e-01 1.78301319e-01
-4.79008168e-01 1.00221598e+00 -4.91002351e-01 -3.06639850e-01
4.05477285e-01 -5.18430650e-01 -6.30531609e-01 2.01225281e-01
1.84669286e-01 -8.41307521e-01 -5.02592266e-01 8.99787426e-01
5.79154968e-01 4.00611252e-01 2.19318420e-01 -1.51364172e+00
-5.66186428e-01 5.82453191e-01 -4.74720001e-01 2.32646555e-01
-1.50034484e-02 2.73394346e-01 9.41674650e-01 -8.76786530e-01
6.86572492e-01 6.39353514e-01 3.79517108e-01 6.20823324e-01
-9.71551716e-01 -7.01820612e-01 1.38125986e-01 5.20325422e-01
-1.35622513e+00 -6.24697506e-01 7.23336577e-01 -1.41843736e-01
5.49728632e-01 3.10010582e-01 2.78062671e-01 1.10897589e+00
-3.44720244e-01 9.29214776e-01 8.55750382e-01 -2.40334451e-01
6.30487263e-01 2.94758171e-01 6.35356456e-02 1.77930579e-01
3.30284797e-02 -2.78986007e-01 -7.82768428e-01 -3.46022159e-01
3.94059479e-01 7.56257176e-01 -3.90896171e-01 -5.16541898e-01
-7.22547650e-01 7.19476700e-01 2.34775588e-01 3.53050292e-01
-4.39901054e-01 -2.78850645e-01 4.44971919e-01 3.81432652e-01
4.86164272e-01 -2.86400527e-01 -3.69525641e-01 -3.50130320e-01
-8.98674488e-01 -5.95600791e-02 1.00066614e+00 9.70875621e-01
1.07849300e+00 -1.68156981e-01 -1.51721701e-01 9.75643396e-01
3.65582388e-03 8.96515772e-02 5.84363818e-01 -9.42980230e-01
6.14229679e-01 5.96984923e-01 5.99240959e-02 -7.74959087e-01
2.73799241e-01 -2.55390555e-01 -7.73362756e-01 -3.78054045e-02
3.17484289e-01 -1.83022276e-01 -4.68924671e-01 1.70180881e+00
6.07269287e-01 6.25073969e-01 -4.29554209e-02 9.45910990e-01
5.99947631e-01 6.01676464e-01 -2.44294286e-01 -4.45437074e-01
8.53922427e-01 -1.18378937e+00 -6.41779184e-01 2.41958397e-03
5.48221648e-01 -4.49127465e-01 1.24097109e+00 7.44591728e-02
-7.49014080e-01 -1.69522703e-01 -7.06184208e-01 2.96481609e-01
-1.76607296e-01 -3.47514719e-01 4.46389228e-01 4.87385660e-01
-5.87987244e-01 5.92825353e-01 -6.46380901e-01 -5.04632831e-01
7.03602314e-01 7.07396790e-02 -2.62501001e-01 -6.91293538e-01
-9.78562653e-01 2.85484135e-01 -6.73028827e-02 -5.02194107e-01
-8.39647353e-01 -9.23646271e-01 -3.00777584e-01 3.28372091e-01
8.82748246e-01 -4.18217510e-01 1.53630781e+00 -9.43058848e-01
-1.27571380e+00 2.82069296e-01 -1.40429467e-01 -1.75970912e-01
8.54161501e-01 -4.63442877e-02 -2.61116713e-01 -6.24135248e-02
-4.10156585e-02 -1.00802571e-01 7.35244811e-01 -1.20158052e+00
-8.06218028e-01 -5.30216992e-01 1.04685768e-01 6.45473152e-02
-9.35981095e-01 -1.07670628e-01 -8.59300852e-01 -2.97675401e-01
-1.19853504e-01 -6.47168398e-01 4.56423685e-02 1.48280948e-01
4.37563993e-02 -4.11450386e-01 1.11765945e+00 -2.34048843e-01
1.24381757e+00 -2.39509773e+00 -3.51807177e-01 2.57237721e-02
1.66191846e-01 4.14123446e-01 -2.50450939e-01 8.03388536e-01
3.19280297e-01 7.65723363e-02 6.91772923e-02 -5.65290987e-01
-9.92733911e-02 4.85034525e-01 -4.01304722e-01 3.75964403e-01
-3.49788606e-01 7.72841334e-01 -8.95775676e-01 -4.79271472e-01
1.49976343e-01 3.92236501e-01 -4.97949362e-01 4.28898215e-01
-2.50826895e-01 3.66833478e-01 -5.60300589e-01 7.63527751e-01
7.17871726e-01 -5.77441394e-01 2.76062071e-01 1.57265109e-04
-3.22785303e-02 3.40611525e-02 -1.37734342e+00 1.69979322e+00
-5.79071641e-01 1.92946464e-01 1.54179543e-01 -1.04044032e+00
6.03561521e-01 4.89916533e-01 6.48322344e-01 -6.03138447e-01
1.14306822e-01 2.47267261e-01 -3.83560002e-01 -5.99921286e-01
7.93241635e-02 -9.23929587e-02 3.43572646e-01 9.75208759e-01
1.99781382e-03 6.87803388e-01 6.48087412e-02 2.35114396e-01
1.31772375e+00 -2.43464798e-01 1.67006880e-01 3.95618290e-01
6.42286688e-02 -2.28714570e-01 9.55660284e-01 8.75369072e-01
-4.36835557e-01 8.16898823e-01 2.26431340e-01 -3.41122061e-01
-9.74947870e-01 -7.05181062e-01 1.41699359e-01 1.55940270e+00
2.86656111e-01 -4.86682177e-01 -4.77449119e-01 -7.76684582e-01
-3.93959172e-02 6.87834442e-01 -4.55982536e-01 -2.31914967e-01
-2.36727551e-01 -4.75405782e-01 1.86920196e-01 4.68731552e-01
5.40595710e-01 -8.53365719e-01 -6.55634105e-01 2.77360708e-01
-3.43510538e-01 -9.69788313e-01 -6.37358427e-01 -1.23154111e-01
-7.55277812e-01 -1.23467600e+00 -6.86181366e-01 -3.53907049e-01
4.76544738e-01 1.19943905e+00 7.79280722e-01 7.36340657e-02
-1.98062360e-01 3.19710374e-01 -8.04742336e-01 -2.19108611e-01
-1.15824781e-01 -3.29406895e-02 -1.82788097e-03 3.55608314e-01
5.75101137e-01 -7.49642611e-01 -7.41522849e-01 5.99693298e-01
-1.12944591e+00 -1.59366831e-01 3.19881052e-01 8.81752670e-01
4.11468595e-01 1.68556422e-01 7.60637164e-01 -9.18259144e-01
4.93284255e-01 -1.07443690e+00 -1.38137370e-01 4.10379648e-01
-6.40776575e-01 -3.66279006e-01 8.42686296e-01 -5.97153664e-01
-1.22996724e+00 -2.05358759e-01 2.23964870e-01 -8.25433016e-01
-1.09195061e-01 4.64824229e-01 -2.75481403e-01 1.93804324e-01
5.45294106e-01 3.86228174e-01 1.31039828e-01 -8.94658267e-01
3.08290929e-01 1.19502020e+00 1.96021512e-01 -3.65500808e-01
6.24940932e-01 6.73765540e-01 -5.45253992e-01 -6.48391068e-01
-6.82106018e-01 -7.30081201e-01 -1.43268377e-01 -2.84787476e-01
1.43068895e-01 -8.61937582e-01 -4.29987282e-01 5.02275288e-01
-7.47247517e-01 -2.57922947e-01 -6.32179141e-01 4.04799968e-01
-3.60632122e-01 4.70056415e-01 -3.75898480e-01 -8.63379776e-01
-5.75241387e-01 -8.85390878e-01 5.74419737e-01 3.99799436e-01
8.23506042e-02 -6.05561674e-01 1.88821539e-01 5.14119923e-01
6.33032858e-01 -1.90763384e-01 6.13047898e-01 -9.44387972e-01
-5.56980312e-01 -4.77858424e-01 -2.77127206e-01 2.89379716e-01
3.69225025e-01 -4.12428342e-02 -1.01302469e+00 -4.46698844e-01
1.06938452e-01 -6.12284243e-01 6.54831588e-01 -7.43464231e-02
1.16275501e+00 -4.71680015e-01 -2.21180946e-01 4.13814366e-01
1.46725357e+00 1.32962856e-02 3.88778090e-01 2.19255969e-01
4.31253821e-01 5.20402074e-01 5.73865891e-01 1.12291479e+00
4.35857087e-01 6.40587389e-01 5.50392568e-01 4.05544877e-01
-9.11541134e-02 -2.66784549e-01 2.26201221e-01 7.70634711e-01
2.78661959e-02 -2.40130305e-01 -7.82953143e-01 6.89113319e-01
-2.13169837e+00 -1.12444699e+00 2.39487141e-01 2.37393141e+00
7.54473448e-01 -3.42345983e-01 1.28617764e-01 -3.74271646e-02
8.80718648e-01 2.62503922e-01 -1.00713134e+00 1.88704595e-01
4.78213318e-02 -1.47782132e-01 2.31002137e-01 -2.30104014e-01
-6.61923885e-01 5.02220750e-01 4.70120621e+00 1.01602590e+00
-1.02250099e+00 6.19223535e-01 4.34002250e-01 -6.20773077e-01
-2.70452380e-01 7.43215233e-02 -6.55672729e-01 8.83854210e-01
7.87731588e-01 -4.41935718e-01 6.35857105e-01 1.13675845e+00
1.22282811e-01 6.68549910e-02 -9.08247113e-01 1.07686746e+00
1.50222465e-01 -1.39117408e+00 -1.32728145e-01 1.83435678e-01
9.41611946e-01 5.46023250e-01 -9.38644186e-02 5.78663528e-01
2.31924087e-01 -4.59653527e-01 3.25994998e-01 4.60767001e-01
7.20312536e-01 -6.61221147e-01 6.78628027e-01 9.42559898e-01
-1.03707385e+00 -4.23875868e-01 -4.26104099e-01 8.61824583e-03
7.50852451e-02 5.67921162e-01 -3.78462762e-01 5.84500372e-01
8.29674959e-01 4.82941657e-01 -2.26938859e-01 1.21683764e+00
2.59563476e-01 6.69493735e-01 -4.19039279e-01 2.08343059e-01
-2.15388252e-03 -1.48831427e-01 3.36639553e-01 6.55850470e-01
5.04710853e-01 2.66061306e-01 1.37925074e-01 5.81556439e-01
-4.42112923e-01 4.00329173e-01 -5.40457308e-01 1.78870652e-02
8.43394458e-01 1.20039666e+00 -2.31786862e-01 -3.35170984e-01
-1.06549454e+00 8.94333422e-01 5.37072241e-01 4.16921318e-01
-6.46654308e-01 -1.50967523e-01 7.98394620e-01 2.00643927e-01
4.42256063e-01 1.98583692e-01 8.92537162e-02 -1.50844586e+00
3.19307715e-01 -9.44451094e-01 6.08972430e-01 -4.13395166e-01
-1.73418474e+00 2.78601199e-01 -3.00208539e-01 -1.36273015e+00
-2.27870420e-01 1.74198151e-01 -1.00427771e+00 4.88153517e-01
-1.40301490e+00 -1.01756525e+00 -6.11653566e-01 8.75699103e-01
7.22217858e-01 -2.66906828e-01 6.24522507e-01 4.95046854e-01
-7.27340400e-01 7.40136027e-01 7.24486887e-01 -7.84075856e-02
9.46494699e-01 -6.28242314e-01 3.29078101e-02 5.58216989e-01
1.32514089e-01 5.06683469e-01 4.72606122e-01 -4.35994774e-01
-1.61241090e+00 -1.15352261e+00 6.26734138e-01 -6.11987785e-02
6.27864003e-01 -1.19846992e-01 -1.38644862e+00 6.01515830e-01
-2.96117589e-02 5.35538495e-01 1.06170559e+00 -1.06940810e-02
-4.76027369e-01 -3.80176455e-01 -1.16690946e+00 5.60520232e-01
9.75400865e-01 -5.11571705e-01 -2.30085030e-01 3.36681217e-01
5.82549930e-01 7.60872290e-02 -6.96266532e-01 1.94553733e-01
5.06566465e-01 -1.20108390e+00 5.59299886e-01 -7.38267601e-01
1.63517922e-01 2.05017589e-02 -4.59361702e-01 -1.08741784e+00
-8.51325989e-02 -6.12289608e-01 -7.33727336e-01 1.60615921e+00
-1.62154973e-01 -7.15475261e-01 9.23399687e-01 1.08084142e+00
2.66539633e-01 -7.12564170e-01 -1.08811975e+00 -1.01503706e+00
-1.81242481e-01 -3.98987591e-01 8.44296694e-01 1.00935149e+00
1.55136317e-01 2.79841982e-02 -6.43349349e-01 -1.36002257e-01
6.40220642e-01 4.22338158e-01 1.11148643e+00 -1.11355960e+00
-3.75921786e-01 -1.36911169e-01 -1.08363125e-02 -9.22635972e-01
1.07679479e-01 -6.69568062e-01 -1.20849714e-01 -1.46219182e+00
7.19703972e-01 -5.07874012e-01 -4.77433145e-01 7.04761982e-01
-2.26448685e-01 3.57749537e-02 4.55568999e-01 8.64191651e-01
-8.76520157e-01 7.44868100e-01 8.65545571e-01 1.55858338e-01
-2.81487882e-01 2.55976140e-01 -7.14287877e-01 4.07878816e-01
9.22004998e-01 -5.16861558e-01 -5.74700594e-01 -2.67940700e-01
-1.18530832e-01 -6.87048510e-02 2.46449262e-01 -8.12655985e-01
5.57452202e-01 -4.18790936e-01 -8.80510211e-02 -3.33439499e-01
2.28442758e-01 -9.77579117e-01 1.66221187e-01 1.48468584e-01
1.51048312e-02 -5.78463376e-01 -3.72219086e-01 8.56572211e-01
-2.49888614e-01 -2.19184250e-01 6.61708534e-01 -1.22224294e-01
-7.37269700e-01 6.70486867e-01 8.94867908e-03 1.42407849e-01
1.35589385e+00 -8.31675008e-02 -5.74391186e-01 -7.09996998e-01
-2.80197918e-01 3.68045658e-01 7.54015505e-01 5.18632889e-01
4.45368946e-01 -1.25403035e+00 -5.67492247e-01 2.06593141e-01
4.15949017e-01 -4.01572809e-02 7.37241387e-01 1.00904918e+00
7.63077214e-02 3.99336293e-02 2.01731245e-03 -2.11778387e-01
-1.27971089e+00 6.82618916e-01 7.60689285e-03 -1.61565505e-02
-8.27269673e-01 4.97788191e-01 6.07642792e-02 -3.48792672e-01
3.91515255e-01 3.57450813e-01 1.19901299e-01 2.80010730e-01
9.04688239e-01 6.78410411e-01 4.34901305e-02 -4.54561442e-01
-2.25140765e-01 4.37224209e-02 -2.05744714e-01 2.57151634e-01
1.54757118e+00 -3.44725609e-01 1.71043962e-01 5.59774578e-01
1.35415041e+00 -1.75611109e-01 -1.57645714e+00 -9.41163182e-01
-4.82627153e-01 -9.01955128e-01 -1.56948529e-02 -4.34842020e-01
-1.30812132e+00 7.05340683e-01 5.20132422e-01 1.66244090e-01
1.08567715e+00 1.50711253e-01 1.06776738e+00 2.85887748e-01
5.70995450e-01 -1.20478976e+00 1.38497010e-01 1.58804283e-01
4.16497946e-01 -1.42414641e+00 -2.13854134e-01 -7.90963024e-02
-8.06364894e-01 8.54996026e-01 6.34927928e-01 2.47904181e-01
5.77195585e-01 2.65027490e-03 9.43375751e-02 1.59048244e-01
-1.14416122e+00 -1.94206595e-01 -2.29167759e-01 4.00031626e-01
-4.84612174e-02 -1.63298026e-01 -1.43014714e-01 8.26555610e-01
4.06665742e-01 4.54624712e-01 4.54452425e-01 1.33448684e+00
-4.28434402e-01 -1.05534577e+00 -2.27031171e-01 5.50836265e-01
-4.60568517e-01 2.21712068e-01 -4.52286117e-02 3.79252881e-01
3.21847349e-02 1.22738481e+00 6.30106330e-02 -3.68014216e-01
2.91873991e-01 2.21607924e-01 -8.09303969e-02 -7.04971850e-01
-3.74800116e-01 8.94814283e-02 -3.49072754e-01 -7.20667481e-01
-3.15137029e-01 -6.67851508e-01 -1.14490056e+00 -6.44574881e-01
-3.56388003e-01 2.44237613e-02 5.74619651e-01 8.65603924e-01
6.74036920e-01 1.82011902e-01 1.07917261e+00 -6.29909515e-01
-1.07110059e+00 -8.46279204e-01 -8.96724761e-01 5.18976331e-01
8.48741904e-02 -5.20490825e-01 -5.93292117e-01 -8.01634938e-02]
|
[5.817436218261719, 6.291619300842285]
|
af96146a-edd6-4900-a28d-00e3f8708127
|
s-dccrn-super-wide-band-dccrn-with-learnable
|
2111.08387
| null |
https://arxiv.org/abs/2111.08387v1
|
https://arxiv.org/pdf/2111.08387v1.pdf
|
S-DCCRN: Super Wide Band DCCRN with learnable complex feature for speech enhancement
|
In speech enhancement, complex neural network has shown promising performance due to their effectiveness in processing complex-valued spectrum. Most of the recent speech enhancement approaches mainly focus on wide-band signal with a sampling rate of 16K Hz. However, research on super wide band (e.g., 32K Hz) or even full-band (48K) denoising is still lacked due to the difficulty of modeling more frequency bands and particularly high frequency components. In this paper, we extend our previous deep complex convolution recurrent neural network (DCCRN) substantially to a super wide band version -- S-DCCRN, to perform speech denoising on speech of 32K Hz sampling rate. We first employ a cascaded sub-band and full-band processing module, which consists of two small-footprint DCCRNs -- one operates on sub-band signal and one operates on full-band signal, aiming at benefiting from both local and global frequency information. Moreover, instead of simply adopting the STFT feature as input, we use a complex feature encoder trained in an end-to-end manner to refine the information of different frequency bands. We also use a complex feature decoder to revert the feature to time-frequency domain. Finally, a learnable spectrum compression method is adopted to adjust the energy of different frequency bands, which is beneficial for neural network learning. The proposed model, S-DCCRN, has surpassed PercepNet as well as several competitive models and achieves state-of-the-art performance in terms of speech quality and intelligibility. Ablation studies also demonstrate the effectiveness of different contributions.
|
['Tao Yu', 'Yannan Wang', 'Jun Huang', 'Lei Xie', 'Jiayao Sun', 'Mengtao Xing', 'Yihui Fu', 'Shubo Lv']
|
2021-11-16
| null | null | null | null |
['speech-denoising']
|
['speech']
|
[ 3.17867219e-01 -6.18197843e-02 1.38962120e-01 -2.07045525e-01
-8.48527491e-01 -5.06273881e-02 2.39503458e-01 -1.30001247e-01
-5.85102856e-01 4.20937985e-01 5.82274199e-01 -2.62584955e-01
-1.03399895e-01 -5.59114158e-01 -4.77833629e-01 -7.07184792e-01
-3.43416780e-02 -5.58316767e-01 -7.08995536e-02 -5.33369839e-01
-2.87002236e-01 1.49308085e-01 -1.57433820e+00 3.45047653e-01
9.12789702e-01 1.03724575e+00 4.90992486e-01 8.98176432e-01
3.79065543e-01 2.91178912e-01 -8.78574491e-01 -2.61715740e-01
2.23459378e-01 -5.31443834e-01 -3.58556151e-01 -8.61672014e-02
-4.34656395e-03 -4.44794685e-01 -5.42786360e-01 1.31051397e+00
9.85262513e-01 4.45778251e-01 1.63218871e-01 -5.87612867e-01
-6.00577950e-01 7.32106566e-01 -3.25010210e-01 3.47773820e-01
2.20278993e-01 2.82922201e-02 7.51829028e-01 -8.11436534e-01
7.21262470e-02 1.23711586e+00 9.90951240e-01 4.32339966e-01
-1.07289815e+00 -7.73359179e-01 1.06830284e-01 4.39465076e-01
-1.13037980e+00 -7.81505108e-01 1.11228585e+00 1.56153858e-01
1.35946262e+00 1.38060346e-01 7.38121152e-01 1.30986047e+00
-6.90405890e-02 5.07244766e-01 1.16645718e+00 -6.14909053e-01
9.44546834e-02 -2.41281435e-01 5.29696094e-03 1.61683217e-01
-3.57725948e-01 5.87081254e-01 -7.74173617e-01 2.29050189e-01
6.72896504e-01 -1.54876500e-01 -8.04433048e-01 5.78256190e-01
-9.95717764e-01 6.88817680e-01 4.72740352e-01 6.19916260e-01
-5.53665340e-01 1.08468570e-01 5.35439372e-01 6.17737234e-01
7.74382293e-01 2.62959361e-01 -5.66610396e-01 -2.96746671e-01
-9.92852569e-01 1.09417982e-01 4.57880497e-01 4.97116923e-01
3.98337007e-01 6.52592599e-01 -2.02347592e-01 1.33712125e+00
1.69363320e-01 2.49203816e-01 8.36295247e-01 -7.67269373e-01
5.14631331e-01 -3.10601324e-01 -1.47054255e-01 -6.90069139e-01
-4.84870583e-01 -9.08397496e-01 -1.38128090e+00 1.97530285e-01
-1.17323108e-01 -3.55816633e-01 -1.08799255e+00 1.88643217e+00
1.97144583e-01 4.06997919e-01 3.39297026e-01 1.12279916e+00
6.38328016e-01 9.53592181e-01 -4.18903120e-02 -4.53138411e-01
1.44461083e+00 -9.79441464e-01 -1.33916068e+00 -3.04246116e-02
3.57240066e-02 -8.83129597e-01 1.04403126e+00 6.56413853e-01
-1.30103254e+00 -1.04033303e+00 -1.32035935e+00 -2.42072552e-01
-3.82037312e-01 2.13065401e-01 3.55300635e-01 9.08111751e-01
-1.24538398e+00 8.70113015e-01 -7.13527560e-01 9.12053064e-02
2.02276558e-01 1.75382614e-01 -1.81206360e-01 1.20196089e-01
-1.68964684e+00 8.16168427e-01 4.37161237e-01 3.23360950e-01
-6.08300507e-01 -9.62032020e-01 -9.76030767e-01 3.87835890e-01
1.25397101e-01 -6.14745677e-01 1.30426836e+00 -7.69804239e-01
-2.05535269e+00 6.15922809e-02 -1.50132731e-01 -7.70922601e-01
3.68970037e-02 -5.22268951e-01 -1.08853412e+00 3.42411280e-01
-3.71881127e-01 3.43004167e-01 1.45730770e+00 -6.81476295e-01
-5.53638935e-01 -4.02663574e-02 -2.01817065e-01 2.45692000e-01
-7.25777030e-01 -1.32189179e-02 -1.14235461e-01 -1.44644022e+00
2.24973395e-01 -4.19043660e-01 -1.20550029e-01 -4.67943996e-01
-1.17682993e-01 2.28383075e-02 1.00401795e+00 -1.24063599e+00
1.44029272e+00 -2.51176882e+00 6.75195605e-02 -6.18865676e-02
-2.42168270e-02 7.43651330e-01 -2.84525096e-01 3.67871195e-01
-4.48182642e-01 -4.66015972e-02 -1.32033512e-01 -5.79976976e-01
3.04278191e-02 -2.08013445e-01 -3.76573920e-01 4.22300398e-01
5.13180912e-01 3.52299511e-01 -6.34188652e-01 1.14360183e-01
4.51877683e-01 1.18724155e+00 -7.40322709e-01 1.19668869e-02
2.41136044e-01 1.99686006e-01 3.47607844e-02 2.91513294e-01
9.38391328e-01 3.04096818e-01 -1.53133929e-01 -6.20717227e-01
-2.06233099e-01 5.55614471e-01 -1.14544654e+00 1.91918314e+00
-8.51625383e-01 5.35768747e-01 4.83578146e-01 -9.06200528e-01
7.79033959e-01 8.16673398e-01 2.58901119e-01 -7.78429151e-01
-7.09062815e-02 2.14750990e-01 1.20464757e-01 -3.19192320e-01
4.62350368e-01 -3.33583146e-01 4.41807151e-01 8.93545300e-02
3.91651064e-01 -2.42577255e-01 -1.10780865e-01 -2.22360164e-01
1.03360808e+00 -5.63275106e-02 2.28049308e-01 1.01054974e-01
5.86198568e-01 -8.25855911e-01 4.14963245e-01 5.10069191e-01
-2.38843173e-01 7.58807480e-01 4.15961929e-02 1.00693211e-01
-1.01553667e+00 -1.04952550e+00 -2.30374455e-01 1.27472711e+00
-1.66703761e-02 -5.51417410e-01 -9.54092979e-01 -1.66450500e-01
-3.50447029e-01 5.65371990e-01 -3.29798192e-01 -4.37433094e-01
-6.30418360e-01 -6.30780399e-01 7.37268388e-01 3.37020785e-01
8.39116931e-01 -9.95226622e-01 -3.51624519e-01 5.99977255e-01
-4.78976101e-01 -1.03394294e+00 -7.41820216e-01 4.71327484e-01
-5.86250842e-01 -3.63231242e-01 -9.05443430e-01 -7.44598389e-01
7.17687830e-02 2.99618900e-01 5.86612999e-01 -1.53798521e-01
-2.20628921e-02 9.35011283e-02 -6.99780822e-01 -3.59476298e-01
-3.84863406e-01 7.68958032e-02 1.09814517e-01 1.67375639e-01
-6.23588376e-02 -9.52940941e-01 -7.93554723e-01 1.43542606e-02
-1.06571198e+00 -6.97482377e-02 7.20887423e-01 1.11427879e+00
3.34221661e-01 6.57934785e-01 1.14213324e+00 -2.43105486e-01
1.12825310e+00 -9.37165692e-03 -2.04254568e-01 -1.92567170e-01
-4.07071590e-01 -2.39458218e-01 8.70664358e-01 -7.48016834e-01
-1.39739251e+00 -2.96316981e-01 -8.76462400e-01 -4.77874100e-01
-2.28649881e-02 6.06687844e-01 -9.96922627e-02 1.48198277e-01
6.49032772e-01 2.41910324e-01 -1.87141169e-02 -7.79344857e-01
2.18897045e-01 1.02051353e+00 7.62852967e-01 -2.14225277e-01
8.08809221e-01 1.13005079e-01 -4.02132928e-01 -1.07549441e+00
-6.82194710e-01 -5.27523160e-01 -6.67414218e-02 -3.46921012e-02
6.86140358e-01 -1.27213776e+00 -4.57233906e-01 7.96327353e-01
-1.11741567e+00 -2.40523517e-01 -4.08591658e-01 9.57782388e-01
-3.58273268e-01 4.32409793e-01 -1.10593390e+00 -9.01834607e-01
-7.59465635e-01 -9.64610934e-01 9.56125379e-01 2.19738334e-01
8.00716728e-02 -6.47336543e-01 -2.73863345e-01 7.46956021e-02
1.04553449e+00 -2.15238243e-01 6.59164548e-01 -2.47277886e-01
6.62132204e-02 -1.74512282e-01 1.13614388e-01 9.44367111e-01
2.22169086e-01 -4.77157235e-01 -1.44592428e+00 -5.72355568e-01
5.87103128e-01 -1.40053615e-01 1.08304060e+00 6.27600372e-01
1.25375700e+00 -2.84085989e-01 2.10714474e-01 8.23734760e-01
1.08524978e+00 2.10182846e-01 9.14821327e-01 7.42887659e-03
2.43912831e-01 3.45012933e-01 4.23502326e-01 3.43556881e-01
-5.30682206e-02 6.24701738e-01 1.82297945e-01 -4.13748294e-01
-6.91508949e-01 -8.03420693e-02 5.48933625e-01 1.33490026e+00
-2.37590373e-01 -5.23847900e-02 -3.69540423e-01 3.94118249e-01
-1.43514419e+00 -1.06322932e+00 3.81086022e-01 2.01017380e+00
1.18095875e+00 1.60940140e-01 -4.37619872e-02 7.41314650e-01
7.85715044e-01 3.65999401e-01 -5.03756106e-01 -3.84673774e-01
-2.55492419e-01 7.83190846e-01 1.17943659e-01 5.24296522e-01
-1.12032640e+00 6.78508461e-01 5.42224455e+00 1.34447515e+00
-1.37149465e+00 3.10215890e-01 5.42676032e-01 -8.51682872e-02
6.49866164e-02 -4.71512526e-01 -4.83273029e-01 2.59477019e-01
1.22930443e+00 3.86842676e-02 9.98527646e-01 6.59314096e-01
5.54367661e-01 2.74541706e-01 -5.95529974e-01 1.09078455e+00
-1.65469185e-01 -9.85568702e-01 -2.87340879e-01 -1.09009594e-01
4.22709584e-01 -2.82700211e-02 2.96444952e-01 7.17801690e-01
-2.27208257e-01 -1.04843366e+00 8.06021452e-01 2.30443791e-01
1.09237742e+00 -9.55455899e-01 7.09152400e-01 2.91267216e-01
-1.42001045e+00 -2.43077397e-01 -3.17584693e-01 -9.82544124e-02
9.42372829e-02 8.91453624e-01 -7.29848325e-01 7.10732043e-01
9.32528734e-01 5.89103520e-01 5.92893362e-03 7.36470878e-01
-1.61028072e-01 8.50337029e-01 -3.95037860e-01 3.87100846e-01
7.67394081e-02 -3.96062285e-02 5.33768356e-01 1.50141203e+00
4.40605730e-01 1.95953190e-01 -2.32546315e-01 3.73860836e-01
-1.29455850e-01 -2.27173463e-01 -8.09587985e-02 1.29847214e-01
4.74979401e-01 1.18873656e+00 -1.60486400e-01 -2.10869163e-01
-4.33096766e-01 9.41191912e-01 4.01203558e-02 7.49073148e-01
-8.75225723e-01 -9.27328825e-01 9.00215626e-01 -1.74503878e-01
6.12070024e-01 -1.55838802e-01 -1.03710525e-01 -8.94380450e-01
1.58424765e-01 -1.11088288e+00 5.15947156e-02 -8.12998354e-01
-1.05868006e+00 8.73742998e-01 -3.42240632e-01 -1.23165441e+00
-3.34684014e-01 -3.81886631e-01 -4.51653093e-01 1.42110515e+00
-1.91444719e+00 -9.77870822e-01 -5.20091355e-02 8.04941595e-01
7.01329112e-01 -1.96132511e-02 9.07142937e-01 5.68585277e-01
-4.18881863e-01 7.18018472e-01 -1.87176756e-05 -2.65631266e-02
6.13265693e-01 -1.05663061e+00 5.01431882e-01 1.04056489e+00
-2.54903376e-01 6.60064816e-01 6.32374763e-01 -3.75846595e-01
-1.15010643e+00 -1.34364915e+00 6.77677274e-01 4.71854478e-01
3.86136323e-01 -4.82446343e-01 -1.05873549e+00 1.66981235e-01
4.10207808e-01 1.76565349e-01 3.91719669e-01 8.55129287e-02
-3.15262169e-01 -4.24115896e-01 -1.11685801e+00 7.47793078e-01
9.96438265e-01 -9.14689302e-01 -5.04917443e-01 -4.12745290e-02
1.31834269e+00 -5.30133486e-01 -9.98723090e-01 6.21631682e-01
3.34983736e-01 -1.00874865e+00 1.20636129e+00 -1.22724488e-01
2.94028908e-01 -3.19737524e-01 -2.52261221e-01 -1.81939399e+00
-3.79352659e-01 -1.06740677e+00 -3.21578890e-01 1.27136528e+00
2.71063358e-01 -7.55583227e-01 2.79479325e-01 -2.67090499e-01
-7.46970832e-01 -6.15631461e-01 -1.19359088e+00 -6.38575554e-01
-3.08611602e-01 -6.91433489e-01 6.21048093e-01 7.19634533e-01
6.98097646e-02 2.95253247e-01 -5.86422145e-01 5.47030687e-01
3.19751412e-01 -2.58701056e-01 2.07201838e-01 -6.76176190e-01
-6.03017926e-01 -3.82127345e-01 -1.05952159e-01 -1.14656222e+00
8.22261814e-03 -4.74674135e-01 2.26934031e-01 -1.20709193e+00
-4.68369931e-01 1.51834920e-01 -4.79148924e-01 2.01874018e-01
-3.46908480e-01 2.33661562e-01 2.33890563e-01 -3.41282994e-01
1.28203481e-02 8.90743732e-01 1.34891605e+00 -1.76758736e-01
-4.62990731e-01 9.64054745e-03 -5.64216673e-01 7.32437134e-01
8.02182555e-01 -1.78252921e-01 -4.80803281e-01 -2.71610200e-01
-3.46366793e-01 3.61232311e-01 3.33278298e-01 -1.31378794e+00
1.81931868e-01 4.07363951e-01 4.55472022e-01 -4.77281332e-01
8.92856300e-01 -7.17167497e-01 9.62786674e-02 4.63099778e-01
-3.54514420e-01 -2.87972599e-01 3.41262728e-01 6.20821297e-01
-6.52411461e-01 -5.46673089e-02 9.99398768e-01 1.15990989e-01
-3.67983013e-01 9.36871096e-02 -4.25332338e-01 -2.65358120e-01
3.40717316e-01 -1.46766335e-01 -3.47369492e-01 -5.89340746e-01
-8.18020821e-01 -3.83499831e-01 -3.15283149e-01 3.20639133e-01
8.32560301e-01 -1.15566754e+00 -8.68209541e-01 4.43546474e-01
-4.51653928e-01 -2.48864949e-01 7.58791566e-01 7.04289436e-01
-7.33215287e-02 2.37317055e-01 5.01258560e-02 -3.26785803e-01
-1.01520097e+00 4.04435962e-01 5.05201936e-01 -2.69938797e-01
-9.48331892e-01 7.43941963e-01 1.15075625e-01 -7.52158985e-02
4.94217485e-01 -6.63788378e-01 -3.17402065e-01 -3.76628041e-02
7.76584029e-01 2.32979149e-01 5.95611274e-01 -4.56241041e-01
-5.15101589e-02 4.78626400e-01 -4.97748666e-02 -3.48417491e-01
1.52928698e+00 -2.38966689e-01 1.43716827e-01 4.32350822e-02
1.29992843e+00 1.76526278e-01 -1.31185877e+00 -2.77020812e-01
-4.05633420e-01 -1.94003791e-01 5.67162454e-01 -1.02000642e+00
-1.05458772e+00 8.63961637e-01 9.18806970e-01 3.46734405e-01
2.01841521e+00 -5.32418728e-01 1.04332292e+00 5.41358702e-02
4.06371662e-03 -1.24555695e+00 1.90402210e-01 6.48102343e-01
1.18151569e+00 -8.28040361e-01 -2.64295638e-01 -3.79970372e-01
-1.83766276e-01 1.20130897e+00 1.65861934e-01 3.53938551e-03
8.20541203e-01 4.01035964e-01 2.02671602e-01 2.88636297e-01
-5.81967771e-01 -3.04725587e-01 1.28446609e-01 7.88176537e-01
4.99451786e-01 -7.90521204e-02 -1.00208350e-01 8.70626509e-01
-6.46311641e-01 -1.60681009e-01 3.53307337e-01 4.86249179e-01
-3.92244190e-01 -8.68902922e-01 -6.91137433e-01 2.92218268e-01
-7.10304260e-01 -4.61139858e-01 3.58864427e-01 3.88849676e-01
1.54676482e-01 1.50714481e+00 -1.80035248e-01 -5.13675809e-01
5.82214296e-01 -1.42876143e-02 1.49684250e-01 -2.46653304e-01
-9.35710549e-01 6.76606297e-01 1.32661775e-01 -4.10846442e-01
-3.29451919e-01 -3.51139456e-01 -9.74684000e-01 -1.85993060e-01
-5.86428702e-01 4.05721515e-02 9.52908874e-01 6.26265287e-01
2.51852244e-01 1.21755075e+00 7.22133219e-01 -1.15138626e+00
-7.42898285e-01 -1.51270247e+00 -7.74492383e-01 3.21273744e-01
1.06603301e+00 -2.49428168e-01 -4.33557749e-01 1.09531172e-01]
|
[14.969880104064941, 5.971793174743652]
|
f94b9c69-0f93-44ec-9268-5a920c5d57dc
|
neuralrecon-real-time-coherent-3d
|
2104.00681
| null |
https://arxiv.org/abs/2104.00681v1
|
https://arxiv.org/pdf/2104.00681v1.pdf
|
NeuralRecon: Real-Time Coherent 3D Reconstruction from Monocular Video
|
We present a novel framework named NeuralRecon for real-time 3D scene reconstruction from a monocular video. Unlike previous methods that estimate single-view depth maps separately on each key-frame and fuse them later, we propose to directly reconstruct local surfaces represented as sparse TSDF volumes for each video fragment sequentially by a neural network. A learning-based TSDF fusion module based on gated recurrent units is used to guide the network to fuse features from previous fragments. This design allows the network to capture local smoothness prior and global shape prior of 3D surfaces when sequentially reconstructing the surfaces, resulting in accurate, coherent, and real-time surface reconstruction. The experiments on ScanNet and 7-Scenes datasets show that our system outperforms state-of-the-art methods in terms of both accuracy and speed. To the best of our knowledge, this is the first learning-based system that is able to reconstruct dense coherent 3D geometry in real-time.
|
['Hujun Bao', 'Xiaowei Zhou', 'Linghao Chen', 'Yiming Xie', 'Jiaming Sun']
|
2021-04-01
| null |
http://openaccess.thecvf.com//content/CVPR2021/html/Sun_NeuralRecon_Real-Time_Coherent_3D_Reconstruction_From_Monocular_Video_CVPR_2021_paper.html
|
http://openaccess.thecvf.com//content/CVPR2021/papers/Sun_NeuralRecon_Real-Time_Coherent_3D_Reconstruction_From_Monocular_Video_CVPR_2021_paper.pdf
|
cvpr-2021-1
|
['3d-scene-reconstruction']
|
['computer-vision']
|
[ 2.59176552e-01 -1.13497920e-01 1.80412024e-01 -5.34776270e-01
-5.74832320e-01 -1.89684764e-01 3.27071518e-01 -4.72041011e-01
-1.12916373e-01 3.90199482e-01 3.40206288e-02 1.18696857e-02
6.55482709e-02 -9.84589577e-01 -1.01389563e+00 -2.81715184e-01
3.29856598e-03 5.38279057e-01 4.78025287e-01 7.59992450e-02
2.02148601e-01 9.41599429e-01 -2.01353717e+00 4.67614740e-01
5.38071454e-01 1.13125944e+00 6.59156561e-01 5.87822914e-01
-5.21647334e-02 8.32028806e-01 -1.58938970e-02 2.81805694e-01
4.22901511e-01 -1.85455397e-01 -5.29208183e-01 2.43442208e-01
9.32947516e-01 -9.28395987e-01 -6.96004272e-01 7.10742593e-01
4.80138093e-01 7.02200234e-02 6.13837779e-01 -3.62205356e-01
-2.04572976e-01 -4.21751142e-02 -5.07216513e-01 -1.30168647e-01
8.07740688e-01 -1.88603103e-01 7.02268600e-01 -1.24770451e+00
1.11271739e+00 1.31552386e+00 4.76870418e-01 3.58829468e-01
-1.05207050e+00 -4.10886109e-01 3.14283311e-01 1.59826145e-01
-1.16642678e+00 -5.49439669e-01 1.04282665e+00 -2.75079459e-01
1.37749600e+00 -1.95469782e-02 9.44170773e-01 7.60762691e-01
3.57357353e-01 9.10033762e-01 8.21051359e-01 -3.54579180e-01
1.56175166e-01 -5.01676559e-01 -3.51525009e-01 1.05164516e+00
-1.00763952e-02 2.74721980e-01 -7.88460970e-01 1.46785378e-01
1.54465032e+00 2.41963580e-01 -5.31616271e-01 -6.28424346e-01
-1.21290922e+00 5.49365103e-01 4.24841255e-01 1.61316648e-01
-7.75098979e-01 5.15709400e-01 8.86724442e-02 3.58040065e-01
9.92260754e-01 3.74561548e-02 -3.75929922e-01 1.51986331e-01
-9.80791152e-01 1.21812731e-01 7.62711525e-01 7.35596716e-01
1.02963758e+00 3.52118254e-01 3.35817993e-01 5.89833975e-01
3.78524750e-01 6.69319093e-01 -7.44873099e-03 -1.43499351e+00
2.47666582e-01 5.11254847e-01 1.12690926e-02 -8.92597258e-01
-2.97951639e-01 -8.62832293e-02 -6.97140574e-01 5.14948726e-01
-8.83160010e-02 1.87585384e-01 -1.14379311e+00 1.32529330e+00
4.36147392e-01 6.25155330e-01 -2.02026635e-01 1.02674437e+00
1.26780856e+00 6.86009586e-01 -7.35433102e-01 -1.86273322e-01
6.87251747e-01 -7.24652171e-01 -5.67483008e-01 -1.46234781e-01
1.08154580e-01 -6.77060604e-01 4.58746433e-01 6.11648262e-01
-1.46774447e+00 -6.44423783e-01 -9.80966210e-01 -3.96514535e-01
1.68329954e-01 6.34250194e-02 6.09643936e-01 -2.69258227e-02
-1.31570005e+00 6.66141808e-01 -1.16712117e+00 -2.41768047e-01
4.22756672e-01 3.22287858e-01 -6.29509747e-01 -4.48328733e-01
-5.90664208e-01 6.08035743e-01 -1.11859683e-02 1.12882808e-01
-1.16054893e+00 -6.52973056e-01 -1.02799070e+00 -2.18758792e-01
4.00031954e-01 -1.17267716e+00 1.20018280e+00 -8.13407898e-01
-1.77499676e+00 9.33806837e-01 -6.31220043e-01 -1.89084932e-01
1.99375033e-01 -4.72152859e-01 1.50307000e-01 4.83673930e-01
4.47917245e-02 7.97322214e-01 8.97016525e-01 -1.56731236e+00
-5.64209759e-01 -3.88197899e-01 4.29071449e-02 4.68058944e-01
3.80346984e-01 -4.22605455e-01 -5.84444165e-01 -2.90094525e-01
8.58086646e-01 -5.31309366e-01 -1.84979424e-01 3.82515490e-01
-2.34464973e-01 -5.31122759e-02 7.84174025e-01 -7.02172041e-01
5.13654828e-01 -1.98639667e+00 5.83706081e-01 -1.08822044e-02
3.63061190e-01 -4.84142974e-02 -5.19305468e-02 1.77799150e-01
7.25612119e-02 -4.11171198e-01 -7.38314614e-02 -9.21979904e-01
-5.04747629e-01 3.09911072e-01 -2.57786989e-01 7.03748703e-01
-5.79825230e-02 8.21692348e-01 -7.75252998e-01 -4.39734571e-02
7.23749042e-01 7.98905373e-01 -6.71341419e-01 6.47078574e-01
-3.67323369e-01 5.42590320e-01 -3.54744345e-01 7.77207851e-01
8.51817667e-01 -2.31114343e-01 -6.92725694e-03 -4.23981100e-01
-3.87985885e-01 3.23850393e-01 -1.06393909e+00 2.45453691e+00
-5.68460524e-01 5.64957440e-01 3.68553281e-01 -8.34465563e-01
9.66691315e-01 3.40141833e-01 5.64058721e-01 -6.46879673e-01
1.23052344e-01 3.80294651e-01 -7.61974871e-01 -3.65227520e-01
5.72484851e-01 -9.51167643e-02 3.94755214e-01 4.28246260e-01
2.47929528e-01 -6.27954304e-01 -3.19913089e-01 -3.73846889e-02
9.50406909e-01 7.29450524e-01 9.51960757e-02 1.60167277e-01
4.87959474e-01 -4.65293646e-01 4.15378720e-01 3.84202749e-01
4.58516300e-01 1.13194752e+00 -4.50219028e-02 -8.38226855e-01
-1.04505265e+00 -1.23049247e+00 1.85607180e-01 4.49598342e-01
3.97308439e-01 -1.29703224e-01 -3.55001539e-01 -3.07699382e-01
-8.60369503e-02 4.50385064e-01 -5.07080793e-01 3.43033344e-01
-6.98836625e-01 1.68485224e-01 -1.06305778e-01 3.89543563e-01
5.77673614e-01 -1.07042539e+00 -9.02726173e-01 2.98288107e-01
-2.06827179e-01 -1.25307381e+00 -1.82151675e-01 1.59721985e-01
-1.34634316e+00 -1.11920357e+00 -6.34996772e-01 -8.26957285e-01
6.12241507e-01 7.97312558e-01 1.15931809e+00 -2.68891528e-02
-9.58254337e-02 4.91727412e-01 -2.56174058e-02 9.86620113e-02
-1.00702122e-01 -1.95969284e-01 -1.61869556e-01 -1.73492748e-02
-6.31108060e-02 -1.11921144e+00 -6.32919967e-01 1.36000425e-01
-8.12036395e-01 5.51286757e-01 3.71255666e-01 6.73909128e-01
1.07094312e+00 -2.66413838e-01 -1.22270780e-02 -6.94673836e-01
-1.11672305e-01 -3.80917609e-01 -7.21708417e-01 -2.26588380e-02
-4.23900820e-02 -9.95166600e-02 3.86555195e-01 2.01116055e-02
-1.25868464e+00 3.78157824e-01 -2.86514312e-01 -1.01612508e+00
-4.90485191e-01 3.02425295e-01 8.81951116e-03 -1.77164629e-01
4.80086535e-01 3.70643884e-01 1.92493312e-02 -7.08470762e-01
3.01560104e-01 2.62880772e-01 6.66331470e-01 -2.52288043e-01
4.75026160e-01 1.07270861e+00 9.70145389e-02 -9.64098752e-01
-8.07403564e-01 -5.21126211e-01 -8.99243712e-01 -3.94892335e-01
9.31263864e-01 -1.32152438e+00 -6.07392669e-01 7.95469165e-01
-1.57477021e+00 -3.68787736e-01 -2.91180730e-01 6.14581168e-01
-1.04271424e+00 3.82836282e-01 -7.34515309e-01 -5.80407500e-01
-3.85029972e-01 -1.00798321e+00 1.47862387e+00 -1.17864944e-01
1.35218143e-01 -8.95253897e-01 1.21283807e-01 1.15225017e-01
1.71960145e-01 4.28758532e-01 4.08688039e-01 3.23720276e-01
-1.32198906e+00 1.82713404e-01 -2.53675193e-01 3.22920650e-01
3.22337836e-01 -1.44845769e-01 -1.22464526e+00 -2.13897213e-01
3.33232433e-01 -1.38351023e-01 1.13708973e+00 9.42356229e-01
8.51348579e-01 5.88452145e-02 -3.11338782e-01 1.24672484e+00
1.63317859e+00 1.56874701e-01 6.97839797e-01 9.94274095e-02
1.14100409e+00 4.76707369e-01 3.10224891e-01 4.25984144e-01
5.08433104e-01 4.92817760e-01 7.23820686e-01 -2.06933990e-01
-4.06064779e-01 -3.08306903e-01 4.23648387e-01 9.06589329e-01
-2.64953464e-01 -3.14764678e-01 -6.11693680e-01 4.68019217e-01
-1.78391480e+00 -8.53311539e-01 -4.43298407e-02 2.08782697e+00
3.09204072e-01 2.45615188e-02 -3.06978941e-01 -7.89703950e-02
3.73571485e-01 3.96663636e-01 -6.36940300e-01 -1.23923153e-01
-1.41291454e-01 2.67762929e-01 2.75242388e-01 7.95304656e-01
-8.53421807e-01 1.14154744e+00 6.47844505e+00 4.94979352e-01
-1.25835252e+00 4.68512550e-02 3.43895823e-01 -3.50128442e-01
-6.40324533e-01 -1.66972443e-01 -3.97958905e-01 -1.08042553e-01
5.02769828e-01 3.63226503e-01 5.95056653e-01 5.57106912e-01
1.35033339e-01 -3.39945048e-01 -1.10675633e+00 1.25436950e+00
2.42614701e-01 -1.60075903e+00 7.78314322e-02 -6.03805110e-03
9.50407207e-01 7.03488708e-01 -3.22448254e-01 -4.22516108e-01
2.77364969e-01 -9.44402933e-01 9.34683859e-01 1.00204587e+00
1.10737157e+00 -7.04397440e-01 5.51466048e-01 3.95620137e-01
-1.28207457e+00 3.83898377e-01 -4.09570366e-01 -2.16895983e-01
6.14868701e-01 9.20752764e-01 -5.07403135e-01 8.42810929e-01
7.69103289e-01 1.46598518e+00 3.17433774e-02 8.38017404e-01
-2.67025381e-01 1.39200538e-01 -5.39140701e-01 4.30841655e-01
4.83791903e-02 -1.85646504e-01 5.87122977e-01 6.88886821e-01
7.14936733e-01 2.36322701e-01 2.38462031e-01 1.06686115e+00
2.02211086e-02 -2.49619335e-01 -1.15461147e+00 4.16251183e-01
2.43717328e-01 9.58056450e-01 -6.10055566e-01 -3.59035999e-01
-5.97774386e-01 1.19266999e+00 4.68018442e-01 4.69819307e-01
-2.90098011e-01 1.23102702e-01 3.98029536e-01 3.42881113e-01
6.92491114e-01 -6.30016565e-01 -3.49481970e-01 -1.36419189e+00
8.34826455e-02 -3.00538927e-01 -7.24139065e-02 -1.18660820e+00
-1.17766309e+00 7.39227831e-01 -1.01226315e-01 -1.30034375e+00
-3.55218083e-01 -3.78443837e-01 -2.84692079e-01 7.72241831e-01
-1.68311870e+00 -1.23279643e+00 -5.62102973e-01 7.42043495e-01
6.00095510e-01 -1.81278422e-01 7.46588707e-01 -3.35062912e-04
2.99023017e-02 -2.13863447e-01 -9.17807519e-02 -3.42737794e-01
4.44983006e-01 -8.23726237e-01 7.60293722e-01 6.84487402e-01
1.77425504e-01 2.12697968e-01 2.53790379e-01 -9.34479237e-01
-1.62802470e+00 -8.22399557e-01 5.92294931e-01 -1.58474281e-01
-3.61222215e-02 -3.27950448e-01 -9.85875070e-01 6.69206917e-01
5.05178869e-02 1.96765512e-01 8.37932453e-02 -2.71548629e-01
-2.05944076e-01 -5.68124913e-02 -1.02088642e+00 1.28985405e-01
1.44517267e+00 -7.34427392e-01 -4.86005008e-01 5.41674905e-02
7.58374095e-01 -9.76610959e-01 -7.30918467e-01 5.30503988e-01
6.33270681e-01 -1.57065451e+00 1.10446620e+00 2.78681964e-01
5.58639765e-01 -3.27983439e-01 -4.82043833e-01 -1.19798803e+00
-3.13047081e-01 -5.10888040e-01 -3.90024722e-01 4.91747469e-01
-1.95784494e-01 -5.67736804e-01 8.85404706e-01 1.88265461e-02
-5.22704303e-01 -8.01363587e-01 -1.12892067e+00 -3.58698726e-01
-4.44792747e-01 -6.48264766e-01 5.00013053e-01 5.30663431e-01
-7.26416230e-01 1.18454590e-01 -4.62745905e-01 2.66499281e-01
8.26840401e-01 3.86047840e-01 8.24330151e-01 -1.32972479e+00
-1.78233743e-01 -1.32908568e-01 -4.82915848e-01 -1.63597786e+00
3.70047450e-01 -7.94177353e-01 1.94889635e-01 -1.84913492e+00
-1.39327571e-01 -2.66973794e-01 1.62021801e-01 1.90055460e-01
3.45212162e-01 3.00587833e-01 8.99027288e-03 1.56765029e-01
-4.39158171e-01 8.05712044e-01 1.50514317e+00 2.75990665e-01
-3.14714223e-01 -2.12779135e-01 -1.93593487e-01 9.02673006e-01
3.52807581e-01 -2.85941809e-01 -4.57826585e-01 -7.95142591e-01
2.16690421e-01 5.01243412e-01 5.40241659e-01 -9.63274002e-01
3.18358302e-01 -7.89652839e-02 4.98061985e-01 -1.18657422e+00
8.04453135e-01 -7.38985658e-01 3.69417638e-01 1.20797977e-01
2.16425657e-01 -1.78810924e-01 1.61203757e-01 6.21106625e-01
-3.09261173e-01 1.24679074e-01 5.63118279e-01 -4.13242847e-01
-8.34078789e-01 7.08878219e-01 -2.40107089e-01 -3.06053132e-01
6.33893132e-01 -4.88098860e-01 1.57749042e-01 -5.83460331e-01
-5.42972863e-01 -1.61759093e-01 8.32385123e-01 2.33556375e-01
1.30479741e+00 -1.35506570e+00 -7.59931445e-01 8.03020358e-01
-3.35489064e-01 7.36494422e-01 4.84672010e-01 3.50194961e-01
-8.83261681e-01 3.34340483e-01 -2.88063765e-01 -1.03418326e+00
-1.01733971e+00 1.02892749e-01 5.48259854e-01 1.32491700e-02
-1.13417351e+00 1.04293513e+00 6.06205583e-01 -4.88293529e-01
1.03969671e-01 -5.38374305e-01 2.72681937e-02 -3.07788223e-01
3.61084461e-01 2.76864141e-01 2.11828887e-01 -5.44157147e-01
-2.53447980e-01 1.17407703e+00 1.18497826e-01 -2.53598750e-01
1.85978436e+00 -1.81322813e-01 -2.52188176e-01 6.26886606e-01
1.29326117e+00 -2.54760176e-01 -1.75272584e+00 -4.03239697e-01
-7.06857741e-01 -8.10015500e-01 4.98504192e-01 -3.45548928e-01
-1.35739923e+00 9.15816963e-01 5.05534768e-01 -3.41069341e-01
1.15570891e+00 3.43533829e-02 8.35081816e-01 3.92824322e-01
8.18981528e-01 -5.60596943e-01 1.69176549e-01 7.94130623e-01
1.05642557e+00 -9.13370669e-01 2.25770146e-01 -5.43817461e-01
-2.60093696e-02 1.33836496e+00 5.08146524e-01 -6.36878908e-01
7.86987066e-01 2.15344965e-01 -1.24311581e-01 -5.32181978e-01
-8.34389508e-01 3.97330476e-03 5.10770917e-01 4.95395690e-01
3.40869993e-01 -1.94647402e-01 3.60878438e-01 -1.62177622e-01
9.33334604e-03 2.04809412e-01 4.48490411e-01 8.47008765e-01
-4.48974162e-01 -9.10492182e-01 -2.18818501e-01 3.18019450e-01
-1.96139235e-02 8.75666887e-02 -1.10775977e-01 5.47095060e-01
2.89522368e-03 5.57552457e-01 3.69874060e-01 -3.90855998e-01
4.41485882e-01 -2.65839309e-01 1.00439239e+00 -7.52544940e-01
-1.68398127e-01 4.61996973e-01 8.73979032e-02 -1.18501854e+00
-7.29909837e-01 -6.97620034e-01 -1.31850076e+00 -2.81727016e-01
-1.74184471e-01 -4.83067304e-01 5.67874074e-01 7.82844007e-01
5.18093348e-01 4.56217378e-01 7.50759423e-01 -1.77186131e+00
1.81251958e-01 -7.08955884e-01 -6.53083384e-01 8.34986195e-02
6.76091552e-01 -8.34922850e-01 -3.25175047e-01 -2.55999956e-02]
|
[8.69855785369873, -2.8062498569488525]
|
4ce4e89e-4bf0-4356-861a-030ef129e1a0
|
revisiting-the-adversarial-robustness
|
2204.07373
| null |
https://arxiv.org/abs/2204.07373v2
|
https://arxiv.org/pdf/2204.07373v2.pdf
|
Revisiting the Adversarial Robustness-Accuracy Tradeoff in Robot Learning
|
Adversarial training (i.e., training on adversarially perturbed input data) is a well-studied method for making neural networks robust to potential adversarial attacks during inference. However, the improved robustness does not come for free but rather is accompanied by a decrease in overall model accuracy and performance. Recent work has shown that, in practical robot learning applications, the effects of adversarial training do not pose a fair trade-off but inflict a net loss when measured in holistic robot performance. This work revisits the robustness-accuracy trade-off in robot learning by systematically analyzing if recent advances in robust training methods and theory in conjunction with adversarial robot learning, are capable of making adversarial training suitable for real-world robot applications. We evaluate three different robot learning tasks ranging from autonomous driving in a high-fidelity environment amenable to sim-to-real deployment to mobile robot navigation and gesture recognition. Our results demonstrate that, while these techniques make incremental improvements on the trade-off on a relative scale, the negative impact on the nominal accuracy caused by adversarial training still outweighs the improved robustness by an order of magnitude. We conclude that although progress is happening, further advances in robust learning methods are necessary before they can benefit robot learning tasks in practice.
|
['Thomas A. Henzinger', 'Daniela Rus', 'Alexander Amini', 'Mathias Lechner']
|
2022-04-15
| null | null | null | null |
['gesture-recognition']
|
['computer-vision']
|
[ 5.46196103e-01 5.41924119e-01 7.53961056e-02 -3.46158177e-01
-1.08521986e+00 -7.64006972e-01 8.29198539e-01 -2.70078838e-01
-5.70777893e-01 8.50809932e-01 -2.76499778e-01 -5.07458568e-01
2.44625919e-02 -6.46656752e-01 -1.28670084e+00 -7.66264558e-01
-5.25611043e-01 2.67494202e-01 2.12280914e-01 -4.62985516e-01
-5.33982776e-02 9.21644807e-01 -1.24451137e+00 -2.76912391e-01
6.14238977e-01 6.83483005e-01 -4.47063029e-01 8.08284044e-01
5.56569695e-01 1.11751533e+00 -7.58791268e-01 -3.04194957e-01
6.09195590e-01 -5.49700856e-02 -7.11742878e-01 -3.81955743e-01
5.91815233e-01 -5.08578062e-01 -6.25115335e-01 1.22009408e+00
6.03531718e-01 1.42188698e-01 8.22149515e-01 -1.66659999e+00
-2.76482940e-01 5.76117992e-01 -1.05811074e-01 -2.53355980e-01
4.60414067e-02 6.78901196e-01 5.18990219e-01 -3.08237135e-01
3.08609724e-01 1.45376170e+00 1.10303915e+00 5.87014377e-01
-1.09863603e+00 -9.87709880e-01 2.66358908e-02 -2.22113714e-01
-1.10196114e+00 -6.64490104e-01 5.01872301e-01 -4.02507544e-01
1.09251308e+00 8.41504931e-02 3.12506184e-02 1.31356633e+00
8.43601704e-01 3.79446059e-01 7.45933294e-01 1.11866742e-03
2.91050613e-01 -2.31291410e-02 -3.89764041e-01 6.50282860e-01
3.21022272e-01 8.67412686e-01 8.15524906e-02 -5.80770373e-02
4.97913837e-01 -3.97744209e-01 1.02400649e-02 -5.62832475e-01
-1.00630534e+00 7.58059263e-01 7.98435986e-01 -1.50975943e-01
-1.04511373e-01 7.68663168e-01 9.39273179e-01 7.10428715e-01
2.29077041e-02 8.81075978e-01 -5.62042654e-01 -8.93529654e-02
-4.31186259e-01 5.79265714e-01 8.03313017e-01 8.98637891e-01
6.56653106e-01 6.33905113e-01 3.39107096e-01 4.27427649e-01
1.19772196e-01 7.97791243e-01 3.97583246e-01 -1.37561524e+00
4.58866894e-01 2.95969974e-02 2.32068419e-01 -1.08512163e+00
-6.70256615e-01 -7.50144348e-02 -5.91115534e-01 1.14276719e+00
4.72830147e-01 -6.59840703e-01 -9.53473508e-01 2.06059766e+00
1.00273654e-01 7.28875101e-02 5.62313318e-01 6.46442890e-01
3.04128140e-01 4.59507465e-01 1.27220422e-01 2.14317262e-01
8.20090055e-01 -7.52129853e-01 -2.68326074e-01 -7.65035272e-01
5.36024094e-01 -7.78812826e-01 9.26466584e-01 3.66539598e-01
-8.82477522e-01 -5.03196418e-01 -1.62011433e+00 2.30405867e-01
-4.22458738e-01 -6.84180737e-01 4.80472714e-01 8.96908045e-01
-6.41179919e-01 8.78696084e-01 -1.02991581e+00 -2.47362137e-01
2.20335796e-01 6.14602566e-01 -6.58012390e-01 -4.88523804e-02
-1.31242085e+00 1.54397392e+00 4.12248224e-01 -2.06147581e-02
-1.30915916e+00 -6.88095570e-01 -9.85481858e-01 -2.71348268e-01
2.83618778e-01 -3.27961564e-01 1.54338837e+00 -1.01565742e+00
-1.57925463e+00 4.25424993e-01 4.98955041e-01 -8.29549193e-01
8.13920796e-01 -3.80371392e-01 -4.49870527e-01 -6.91525489e-02
-9.94622484e-02 8.65157545e-01 1.14099681e+00 -1.47091222e+00
-3.83244902e-01 -1.05907112e-01 3.79802287e-01 2.23554343e-01
-4.95726541e-02 -9.51211005e-02 1.41908526e-01 -5.61905801e-01
-2.30236411e-01 -1.48404455e+00 -4.86923993e-01 1.36079699e-01
-7.67381862e-02 2.92626739e-01 9.09010649e-01 -3.89480203e-01
2.32216164e-01 -2.06111813e+00 -4.71607335e-02 1.12720206e-01
-1.66268274e-01 3.69324356e-01 -3.35663617e-01 4.38922286e-01
-1.81745529e-01 5.98430261e-02 -4.99755442e-01 5.15386462e-02
1.33315399e-01 4.26083237e-01 -6.44454598e-01 7.85141826e-01
5.71746945e-01 7.72276402e-01 -1.03707755e+00 -1.06998876e-01
3.69839281e-01 3.79906863e-01 -4.97875601e-01 1.61047414e-01
-5.80092892e-02 4.50459123e-01 -2.11645901e-01 4.71112579e-01
4.91494834e-01 6.39108121e-01 -1.20553903e-01 2.35050857e-01
3.85420203e-01 6.43482730e-02 -1.02973735e+00 1.26087153e+00
-6.32556319e-01 8.07477951e-01 3.59383821e-01 -9.64367509e-01
7.98978686e-01 1.73792034e-01 2.76996404e-01 -5.80832720e-01
2.15949953e-01 3.17156702e-01 3.35060626e-01 -4.44015473e-01
5.78832388e-01 -4.91685331e-01 -4.85916823e-01 2.00789243e-01
1.90087594e-02 -8.20903063e-01 -3.13878864e-01 -1.15550742e-01
1.37514520e+00 3.58826458e-01 9.34133008e-02 -2.09362030e-01
1.67882144e-01 3.48096073e-01 4.65079248e-01 8.01270962e-01
-4.70454931e-01 4.07885283e-01 4.47071701e-01 -1.00808755e-01
-1.33241034e+00 -1.08532000e+00 -1.42813608e-01 9.50256109e-01
9.09541324e-02 2.44281977e-01 -8.57853174e-01 -8.16211343e-01
3.47446084e-01 7.95284569e-01 -4.23451960e-01 -8.95703852e-01
-7.57903814e-01 -6.85230374e-01 1.46518517e+00 6.19346380e-01
3.26152742e-01 -1.18352270e+00 -7.66902447e-01 2.33977381e-02
3.68241727e-01 -1.15772939e+00 -1.48167908e-01 6.86314225e-01
-7.41887510e-01 -9.98940825e-01 -4.50924605e-01 -6.75270557e-01
5.93589067e-01 1.24832422e-01 9.60374355e-01 -2.09884524e-01
-8.08145255e-02 3.32480788e-01 -1.71192095e-01 -6.43047452e-01
-1.00667048e+00 -6.80904686e-02 6.34278893e-01 -8.09436917e-01
-1.01279289e-01 -6.65676355e-01 -1.16618589e-01 4.25853372e-01
-9.33820605e-01 -7.47855544e-01 7.79777884e-01 1.01909101e+00
9.66905877e-02 3.73549014e-01 8.39459360e-01 -8.77613187e-01
5.71640491e-01 -6.23650193e-01 -6.45593226e-01 -2.29975671e-01
-8.38476181e-01 2.01021910e-01 1.05390239e+00 -6.12280130e-01
-7.81150520e-01 2.20967710e-01 -4.62301105e-01 -4.94261563e-01
-3.04691136e-01 3.63807291e-01 -2.21197695e-01 -6.64753914e-01
1.27861583e+00 -3.13555121e-01 3.60729307e-01 1.46687463e-01
5.64917505e-01 4.47676241e-01 9.43200171e-01 -5.31476140e-01
1.35188413e+00 2.74492145e-01 2.19147086e-01 -5.57239532e-01
-2.06767336e-01 1.58876374e-01 -4.33183908e-01 -6.37001023e-02
5.06069481e-01 -1.03830957e+00 -6.46650314e-01 5.25128603e-01
-6.95693016e-01 -8.39522719e-01 -3.30158174e-01 3.17380875e-01
-9.94968295e-01 3.82891059e-01 -5.25448322e-01 -7.44717956e-01
-3.56814340e-02 -1.37432599e+00 7.30576873e-01 -8.94656703e-02
-2.54086256e-01 -8.20667446e-01 -6.47821948e-02 -3.95276137e-02
4.72428292e-01 7.40970135e-01 6.96597099e-01 -7.79647231e-01
-3.44962686e-01 -6.51502371e-01 8.06306116e-03 6.38708889e-01
-4.41749245e-02 -3.80867384e-02 -1.17842746e+00 -6.07882142e-01
4.13428955e-02 -8.39155257e-01 5.20338535e-01 1.18527189e-03
5.15429020e-01 -4.87900138e-01 -3.56620699e-02 5.59839487e-01
1.24438262e+00 2.23879084e-01 7.04660892e-01 5.41377366e-01
7.79286385e-01 5.77005982e-01 7.61880398e-01 -2.90966123e-01
-7.51877427e-02 5.85114002e-01 8.68949533e-01 4.01889980e-02
-3.95408049e-02 -3.35998476e-01 9.33253706e-01 1.39692336e-01
4.19087887e-01 1.18915206e-02 -1.06841433e+00 3.46832305e-01
-1.75207961e+00 -9.07730520e-01 2.17151135e-01 2.23360944e+00
6.08094931e-01 7.73875833e-01 2.99375281e-02 2.77784705e-01
3.92928153e-01 4.98612151e-02 -9.67669129e-01 -7.90505946e-01
7.41121769e-02 -1.28937900e-01 1.15667987e+00 6.40588820e-01
-1.17461884e+00 9.79917526e-01 7.31422377e+00 7.04027176e-01
-1.35310364e+00 -1.03946805e-01 4.41633195e-01 7.41304085e-02
2.32491195e-02 -2.26492792e-01 -3.57204676e-01 1.41862437e-01
1.30752540e+00 -1.56252775e-02 6.22319579e-01 1.30325150e+00
-2.04935670e-02 1.44430161e-01 -1.14477289e+00 4.93860185e-01
-4.85782139e-02 -7.79448152e-01 -3.47888768e-01 4.29430529e-02
7.28974938e-01 4.39693451e-01 4.18946326e-01 7.62762249e-01
8.51900637e-01 -1.42707038e+00 9.35886979e-01 2.04386100e-01
7.93932915e-01 -1.12917912e+00 9.51516271e-01 5.11412561e-01
-6.25403047e-01 -2.24746197e-01 -4.63796526e-01 -1.85994536e-01
-1.77449603e-02 8.64733458e-02 -1.11293185e+00 2.82790303e-01
4.34461594e-01 3.11001658e-01 -4.68153000e-01 5.83653569e-01
-3.03118765e-01 4.69082117e-01 -2.85424590e-01 1.82549909e-01
4.02954042e-01 2.86596775e-01 7.13546515e-01 1.18946481e+00
-6.53343052e-02 -2.69928873e-01 -6.30235821e-02 4.64424163e-01
-1.31833255e-02 -4.69654232e-01 -1.22202849e+00 4.24132571e-02
6.27241254e-01 1.00346184e+00 -5.27910888e-01 9.41283535e-03
-9.41251740e-02 7.96330214e-01 2.75354326e-01 2.50641048e-01
-1.01266885e+00 -6.42763376e-01 1.07945132e+00 -3.80731612e-01
1.21465057e-01 -3.65504503e-01 -6.07408166e-01 -5.81992865e-01
-3.95192616e-02 -1.34822798e+00 -2.09054410e-01 -4.58991498e-01
-1.18805659e+00 4.26904738e-01 -1.15905106e-01 -1.25894809e+00
-8.40783477e-01 -8.34262013e-01 -5.19705415e-01 6.81438327e-01
-1.21410966e+00 -1.12978518e+00 4.94652353e-02 3.47163141e-01
3.58747602e-01 -3.79447967e-01 1.00860822e+00 5.29494658e-02
-4.71928656e-01 1.07135415e+00 3.01358700e-01 5.62350936e-02
8.93759131e-01 -1.15226531e+00 8.96917403e-01 1.15792656e+00
-1.29928991e-01 5.09885311e-01 1.29038525e+00 -4.89411354e-01
-1.52246249e+00 -1.41709542e+00 7.74711892e-02 -1.01502299e+00
7.58502662e-01 -2.99882054e-01 -8.07042301e-01 8.76493216e-01
-1.15291037e-01 -1.06195778e-01 1.07412830e-01 1.32052880e-02
-6.05695069e-01 -8.53007287e-02 -1.57727242e+00 8.26451421e-01
7.82635510e-01 -6.60201550e-01 -5.67637384e-01 2.09378451e-01
1.04016042e+00 -6.46929026e-01 -1.02751780e+00 7.55293489e-01
6.12916410e-01 -7.15179980e-01 1.20922697e+00 -5.96375406e-01
4.86197084e-01 -2.74738789e-01 -3.60942960e-01 -1.52477980e+00
-4.37552258e-02 -7.11834967e-01 1.57282889e-01 1.04877746e+00
5.30928731e-01 -6.86043859e-01 7.46859789e-01 8.57859671e-01
-4.68817323e-01 -5.08211434e-01 -1.03227663e+00 -1.11764884e+00
8.02231491e-01 -7.79436707e-01 4.09485400e-01 7.06034958e-01
1.55604146e-02 1.36186764e-01 -4.73045975e-01 4.59089816e-01
5.19512653e-01 -8.05370390e-01 1.15822625e+00 -6.51465774e-01
-1.69374913e-01 -4.45915043e-01 -7.94109583e-01 -4.96433198e-01
4.51392919e-01 -6.43832326e-01 7.18607008e-01 -8.96004498e-01
-5.03973722e-01 -6.80979192e-01 -2.21905008e-01 5.29613554e-01
-1.72634259e-01 3.88446808e-01 3.81428897e-01 1.42669275e-01
-2.52255857e-01 2.35907048e-01 7.10195005e-01 -3.60476643e-01
2.77966075e-03 9.49216075e-03 -6.58257365e-01 8.66311133e-01
1.03594637e+00 -5.90855896e-01 -4.89930928e-01 -4.20105457e-01
9.00150537e-02 -2.62800187e-01 4.07011569e-01 -1.51187134e+00
1.86111361e-01 -1.15285225e-01 2.57687777e-01 3.15195978e-01
4.04696763e-01 -9.45196867e-01 2.30047256e-02 8.07058752e-01
-2.54027367e-01 3.84636596e-02 7.27206588e-01 6.48841560e-01
1.02715485e-01 -2.94054329e-01 1.15380478e+00 6.16478287e-02
-7.41621554e-01 -4.50066514e-02 -6.04903698e-01 7.09605217e-02
9.89899158e-01 2.37566382e-02 -3.12747836e-01 -4.03703153e-01
-3.28749627e-01 1.54173478e-01 7.59651423e-01 7.04579651e-01
2.32456088e-01 -1.17651916e+00 -5.94838917e-01 1.00341074e-01
6.94381073e-02 -4.66424935e-02 -6.36322126e-02 2.39490911e-01
-5.93766570e-01 3.13548476e-01 -5.35319507e-01 -4.02898192e-01
-9.76126075e-01 8.25096667e-01 5.27835250e-01 -1.29117863e-02
-5.43925345e-01 9.35784936e-01 -7.27642421e-03 -9.31303859e-01
5.65068781e-01 -2.29294032e-01 2.28120327e-01 -3.58829230e-01
2.42719352e-01 3.08829308e-01 2.55613297e-01 -7.10110664e-01
-2.15281457e-01 3.21996510e-01 8.45968723e-02 -2.28667870e-01
9.03709531e-01 1.39108643e-01 4.02856678e-01 4.23353434e-01
1.10437298e+00 -5.76236323e-02 -1.67226958e+00 2.42375910e-01
-1.60152450e-01 -1.30288154e-01 -1.85198113e-01 -7.72030354e-01
-8.35867643e-01 6.69099689e-01 7.23802030e-01 1.23066656e-01
7.64478207e-01 -4.14133310e-01 7.60719001e-01 9.76543546e-01
6.17742956e-01 -9.83244538e-01 -2.08986811e-02 7.94054687e-01
9.62248683e-01 -1.41834593e+00 -4.88517620e-02 -6.96137697e-02
-8.02794218e-01 7.44359255e-01 6.88178599e-01 -6.81213021e-01
4.26362365e-01 6.22632742e-01 3.80015552e-01 1.78274423e-01
-4.48113620e-01 1.57107651e-01 -2.13356942e-01 1.14692080e+00
2.01580822e-02 6.71244860e-02 2.83350825e-01 2.16316834e-01
-6.92417264e-01 -5.08738518e-01 5.15357614e-01 1.00253677e+00
-5.42628646e-01 -8.13496828e-01 -6.95478916e-01 2.53681421e-01
-4.37729895e-01 1.99603394e-01 -3.53201032e-01 1.12274790e+00
1.00515798e-01 1.03275299e+00 -1.95462987e-01 -7.61753082e-01
5.33296764e-01 6.52819723e-02 3.25621307e-01 -3.65777671e-01
-5.92111707e-01 -5.93467176e-01 3.32716554e-01 -6.07764661e-01
1.02286816e-01 -5.71185827e-01 -1.39862013e+00 -5.46796322e-01
-1.56822741e-01 -1.05173759e-01 7.26978660e-01 8.82191658e-01
6.85769171e-02 8.18670928e-01 5.41863322e-01 -1.38534045e+00
-1.06126070e+00 -8.24939430e-01 -1.93167746e-01 1.24055423e-01
7.49790370e-01 -7.09844232e-01 -5.44738173e-01 -1.46528378e-01]
|
[5.535090446472168, 7.852194786071777]
|
16d39a24-6e94-4812-a324-56368262c35c
|
sulcal-pattern-matching-with-the-wasserstein
|
2307.00385
| null |
https://arxiv.org/abs/2307.00385v1
|
https://arxiv.org/pdf/2307.00385v1.pdf
|
Sulcal Pattern Matching with the Wasserstein Distance
|
We present the unified computational framework for modeling the sulcal patterns of human brain obtained from the magnetic resonance images. The Wasserstein distance is used to align the sulcal patterns nonlinearly. These patterns are topologically different across subjects making the pattern matching a challenge. We work out the mathematical details and develop the gradient descent algorithms for estimating the deformation field. We further quantify the image registration performance. This method is applied in identifying the differences between male and female sulcal patterns.
|
['Moo K. Chung', 'Soumya Das', 'Zijian Chen']
|
2023-07-01
| null | null | null | null |
['image-registration']
|
['computer-vision']
|
[-1.09962426e-01 1.67801782e-01 7.81154446e-03 -6.93889558e-01
-3.11091002e-02 -5.01370251e-01 6.20909870e-01 -2.48313501e-01
-4.89287704e-01 3.48201901e-01 3.31153959e-01 2.61771884e-02
-2.65854806e-01 -3.30400914e-01 -3.96509975e-01 -5.47321022e-01
-7.55534232e-01 4.44086462e-01 7.27027208e-02 -3.02599967e-01
5.83754599e-01 6.46291912e-01 -7.63421357e-01 -2.73895651e-01
7.97103703e-01 5.30014217e-01 3.40373129e-01 2.18858227e-01
3.52463335e-01 7.61841014e-02 6.75819218e-02 -3.03011894e-01
5.61358035e-01 -2.99600631e-01 -1.07780087e+00 -5.85432202e-02
9.74156857e-01 2.27655452e-02 -5.28194368e-01 1.16910827e+00
4.41665262e-01 -5.38284667e-02 8.40772808e-01 -9.80941951e-01
-7.52703190e-01 5.16978800e-01 -6.78831756e-01 7.74305642e-01
4.45739836e-01 -4.44727570e-01 6.69207811e-01 -1.08817804e+00
1.17057002e+00 1.13403642e+00 7.92643726e-01 5.48844755e-01
-9.86710727e-01 -3.27841938e-01 -2.82802939e-01 4.70543385e-01
-1.41476429e+00 -3.32292885e-01 9.11414862e-01 -7.87402928e-01
5.39525092e-01 3.49235296e-01 8.92026961e-01 7.05805302e-01
1.02472317e+00 3.98749352e-01 1.57138836e+00 -2.30318725e-01
-1.16653599e-01 -5.54901183e-01 4.34073120e-01 9.87018406e-01
3.24145317e-01 1.17737003e-01 -5.59320927e-01 -1.85441256e-01
1.03946388e+00 -1.84574440e-01 -1.06871128e-01 -3.86572033e-01
-1.36026180e+00 5.25698781e-01 5.99438727e-01 9.83733416e-01
-9.30587277e-02 5.36214299e-02 2.92685684e-02 1.77334860e-01
1.86706260e-01 4.56993252e-01 -5.76992184e-02 3.13461423e-01
-1.21866357e+00 5.62436640e-01 2.80365884e-01 5.55879533e-01
5.13276935e-01 6.28583655e-02 6.72213435e-02 7.68916786e-01
6.00813985e-01 5.10932982e-01 7.92107880e-01 -1.00953043e+00
3.02168787e-01 3.32467169e-01 -4.84714419e-01 -1.64437020e+00
-7.83058822e-01 -2.83373613e-03 -1.08324289e+00 3.37759554e-02
5.72824657e-01 1.09659523e-01 -8.37079227e-01 1.50491011e+00
2.51769423e-01 7.01788887e-02 -6.22025013e-01 8.72039557e-01
5.04980981e-01 9.44057852e-03 -2.06489280e-01 -8.82124230e-02
1.41142929e+00 -4.92875189e-01 -8.41739953e-01 -2.72790045e-01
3.26983690e-01 -7.19433784e-01 6.05625331e-01 -4.08293277e-01
-1.40332353e+00 -1.09353304e-01 -1.08721817e+00 -7.68158957e-02
-1.69845596e-01 -1.11603297e-01 4.87621188e-01 4.64211792e-01
-1.38877904e+00 9.73153174e-01 -1.42827260e+00 -1.84612647e-01
4.87666249e-01 2.17672423e-01 -8.07085514e-01 6.12479210e-01
-8.76683712e-01 1.32580042e+00 -6.04821974e-03 5.13355196e-01
-2.80871779e-01 -7.52283096e-01 -9.32115972e-01 -4.49041039e-01
-5.78536272e-01 -3.54763657e-01 5.87351263e-01 -4.59479898e-01
-1.16925132e+00 1.58843935e+00 -4.04130459e-01 -3.42059851e-01
6.81364357e-01 4.63025957e-01 -2.19678149e-01 -5.03172725e-03
1.54803857e-01 5.00297606e-01 6.66113198e-01 -8.45856667e-01
2.31483594e-01 -7.36415446e-01 -6.81677759e-01 1.40498534e-01
1.37997851e-01 3.56311411e-01 8.20666179e-02 -9.92193758e-01
1.09759986e+00 -9.63727415e-01 -2.24237084e-01 -2.35601962e-02
-4.54373002e-01 -1.37835041e-01 4.71234709e-01 -1.25452578e+00
8.85968506e-01 -1.79550612e+00 3.54233980e-01 7.61682749e-01
6.18104398e-01 -5.42050123e-01 2.41594315e-01 -1.22740634e-01
-3.45388085e-01 7.74895325e-02 -6.01843596e-01 -1.73309639e-01
-1.23366453e-01 -3.96849476e-02 2.74625551e-02 1.18709624e+00
-4.17894602e-01 1.05848587e+00 -6.80269539e-01 -5.87495744e-01
-3.29118043e-01 2.33565524e-01 -3.95291835e-01 1.71518087e-01
9.57372665e-01 8.19602191e-01 -3.19259524e-01 5.39374888e-01
9.91900623e-01 5.11293225e-02 3.99074465e-01 -4.92681652e-01
-1.54764920e-01 6.52970374e-02 -1.03326154e+00 1.63055861e+00
2.03988925e-01 6.82680666e-01 8.69236290e-02 -1.26112688e+00
8.74934196e-01 1.13180019e-01 6.85270727e-01 -7.88436234e-01
1.67285100e-01 4.77541298e-01 3.48752171e-01 -3.90357107e-01
1.00506373e-01 -3.63376945e-01 9.14243832e-02 5.83431184e-01
2.60641962e-01 -3.84545952e-01 2.92170614e-01 -1.72613993e-01
6.22239769e-01 4.91213845e-03 1.75665423e-01 -1.32892120e+00
6.43398583e-01 -3.85044098e-01 6.61518157e-01 1.85269311e-01
-3.80775541e-01 5.98361492e-01 3.12895507e-01 -1.10784972e+00
-1.09911561e+00 -1.49953163e+00 -8.86658192e-01 2.31730223e-01
3.23286384e-01 -1.38164356e-01 -1.13548613e+00 -2.27344885e-01
1.39066158e-02 8.18299223e-03 -9.17240620e-01 1.92515552e-03
-1.18438578e+00 -1.01474905e+00 5.81977546e-01 3.99634212e-01
4.96554047e-01 -8.66911352e-01 -6.14841402e-01 -2.90314347e-01
-6.10046871e-02 -9.13913965e-01 -1.05512226e+00 -4.25523907e-01
-1.13428116e+00 -1.25670230e+00 -7.24474251e-01 -1.12711966e+00
1.12564564e+00 -4.96598631e-01 1.05022538e+00 3.81844372e-01
-5.45838177e-01 1.72329307e-01 4.51287717e-01 2.03138068e-01
-2.28964195e-01 -2.43643999e-01 5.85908175e-01 -1.45894006e-01
3.78081724e-02 -8.84467065e-01 -7.49907732e-01 5.47393024e-01
-5.54942429e-01 2.78135557e-02 -8.17155167e-02 4.97594953e-01
6.01684034e-01 -6.00048363e-01 -7.96627626e-02 -4.51485723e-01
8.17057908e-01 -4.30858135e-01 -5.00535071e-01 2.99749076e-01
-6.53246820e-01 4.62774515e-01 3.68278176e-02 -3.66117865e-01
-6.91429138e-01 -2.11569831e-01 9.39093530e-02 3.06753427e-01
1.64385974e-01 3.64228487e-01 3.50801200e-01 -6.86671317e-01
5.43255329e-01 1.36275947e-01 2.30817869e-01 -4.73058224e-01
6.74655214e-02 9.21998918e-02 6.69761658e-01 -5.86603642e-01
9.32785034e-01 6.89760983e-01 2.66996294e-01 -6.97492301e-01
-2.72271663e-01 8.83810744e-02 -1.30784786e+00 -3.80662441e-01
1.02853572e+00 -1.59640089e-01 -6.60531819e-01 4.44705635e-01
-1.11560678e+00 -2.92755246e-01 1.48310944e-01 5.38106859e-01
-6.93224251e-01 5.91075599e-01 -6.70081437e-01 1.93623453e-02
-5.13095081e-01 -1.40334737e+00 7.04334736e-01 7.36639872e-02
-4.30967897e-01 -1.48287809e+00 7.26242840e-01 -2.24010479e-02
4.90591019e-01 4.73439336e-01 8.35400760e-01 -2.56698221e-01
-1.94937751e-01 6.09006099e-02 7.65460730e-02 -2.93420583e-01
1.91883400e-01 -1.46599084e-01 -3.52489620e-01 -3.11538666e-01
3.56207043e-01 2.79097468e-01 4.00602847e-01 5.84927499e-01
1.05386627e+00 -1.03776872e-01 -4.72673267e-01 1.01418948e+00
1.06878412e+00 1.38209656e-01 4.96005625e-01 3.64632457e-01
7.93083966e-01 9.78703022e-01 -3.15045506e-01 -2.38880470e-01
5.77224255e-01 8.27847302e-01 -1.79325625e-01 1.04506202e-02
-1.48752198e-01 2.85952687e-02 1.63109943e-01 1.33213139e+00
-5.93567610e-01 8.92906010e-01 -1.36773145e+00 5.45799315e-01
-1.53829896e+00 -9.67142403e-01 -1.42876610e-01 1.89533174e+00
9.82295334e-01 -2.58392453e-01 -1.54139057e-01 -2.85413057e-01
7.79951751e-01 1.22865818e-01 -3.41234416e-01 -2.89733529e-01
-2.97437876e-01 3.14011544e-01 6.66408002e-01 8.84849310e-01
-7.63304532e-01 5.84464610e-01 9.10127544e+00 3.76410097e-01
-1.28418243e+00 2.91630030e-01 6.65970504e-01 3.64790112e-01
-3.62535268e-01 -2.17686947e-02 -4.62865859e-01 4.83765155e-01
6.13444686e-01 -5.21262348e-01 6.48525178e-01 2.71352023e-01
1.04576118e-01 1.87180284e-02 -1.09510767e+00 1.17668390e+00
2.70031065e-01 -1.57050908e+00 -2.46492982e-01 1.49411753e-01
7.78085411e-01 2.75682896e-01 3.01561859e-02 -5.33868134e-01
-9.87702608e-02 -1.20059240e+00 8.13098669e-01 9.93851960e-01
5.98727465e-01 -1.78511098e-01 3.30353200e-01 -8.91931131e-02
-1.24311161e+00 3.23382854e-01 -4.71830398e-01 1.52487621e-01
4.12963957e-01 4.02143508e-01 -4.92388487e-01 3.15079361e-01
9.36207712e-01 6.86957598e-01 -7.70107806e-01 9.53329861e-01
-2.77087148e-02 4.52551097e-02 -4.80071902e-01 3.71287614e-01
-2.82546729e-01 -9.63070333e-01 6.16472185e-01 8.40135098e-01
4.14084285e-01 -1.35139629e-01 -1.77070215e-01 1.42093122e+00
-3.72861959e-02 2.65224189e-01 -6.16775274e-01 3.79796863e-01
2.86466837e-01 1.33209920e+00 -1.24599683e+00 -7.09177228e-03
-7.23518431e-02 7.37828255e-01 5.30948937e-01 1.16769515e-01
-3.35523993e-01 2.54898760e-02 6.17147148e-01 4.20041233e-01
-4.15418357e-01 -7.66538620e-01 -6.85692787e-01 -1.41947782e+00
3.04425120e-01 -3.29993725e-01 5.12545519e-02 -5.33609927e-01
-1.35731781e+00 8.85330141e-01 5.43281019e-01 -6.62566900e-01
-2.23730326e-01 -8.21567535e-01 -1.13669145e+00 8.60271633e-01
-9.17923152e-01 -8.07520211e-01 -8.60102568e-03 6.53055370e-01
-1.26868844e-01 -2.89066285e-01 6.52456462e-01 2.03477204e-01
-2.62993008e-01 4.94913667e-01 -1.19842507e-01 3.13756973e-01
3.97512197e-01 -1.45244968e+00 7.20703661e-01 9.09962356e-01
8.83565024e-02 1.15705669e+00 7.73064911e-01 -7.76160717e-01
-1.25790203e+00 -4.33753282e-01 1.31089437e+00 -5.90675294e-01
8.77914429e-01 -5.27289093e-01 -7.55114853e-01 8.86790752e-01
4.37087208e-01 1.60535425e-01 4.32509780e-01 -1.37432411e-01
-6.86206296e-02 8.73355195e-02 -1.32818747e+00 7.15573251e-01
1.18373382e+00 -4.08076048e-01 -1.14897096e+00 4.58421856e-01
-1.55588657e-01 -5.45446217e-01 -1.24162042e+00 3.35450262e-01
6.35329306e-01 -6.30928874e-01 1.25696528e+00 -3.84000778e-01
1.13214009e-01 -1.71474591e-01 3.12227935e-01 -1.38453960e+00
-2.63755679e-01 -7.30487823e-01 2.77019858e-01 5.48203528e-01
2.43647531e-01 -8.32939625e-01 6.98897362e-01 8.29024732e-01
1.36477053e-01 -7.87814915e-01 -1.43999815e+00 -6.31029606e-01
7.50243187e-01 1.49106294e-01 5.03564417e-01 1.25389206e+00
4.91572142e-01 -2.39552632e-01 -6.80049434e-02 -2.84701318e-01
9.41655397e-01 9.42298397e-02 -5.08378306e-03 -1.21383142e+00
3.01002651e-01 -8.33885729e-01 -7.40986168e-01 -4.83430862e-01
4.71685797e-01 -1.56982172e+00 -2.73095518e-01 -1.11047328e+00
1.99446917e-01 -3.93937498e-01 -1.68319792e-04 4.02399927e-01
1.58043131e-01 2.98718244e-01 4.18637991e-02 5.28212070e-01
1.14464402e-01 3.37119102e-01 1.47512007e+00 -3.16740647e-02
1.02997102e-01 -5.54533482e-01 -2.63597667e-01 9.02792156e-01
8.55120957e-01 -3.70116383e-01 3.28562222e-02 -4.97196764e-01
3.33883911e-01 -2.34244704e-01 2.08721027e-01 -7.23848343e-01
2.94249594e-01 8.65892507e-03 4.56855297e-01 -3.02385688e-01
-1.86517850e-01 -6.46470010e-01 1.15100592e-01 6.78676426e-01
-2.59969145e-01 8.36275280e-01 -2.04441980e-01 -1.32224858e-01
-1.46120682e-01 -1.46242678e-01 1.01909626e+00 -9.92168188e-02
-6.36377707e-02 6.41479552e-01 -4.13484186e-01 1.94595754e-01
7.56831467e-01 -2.71232516e-01 1.32030204e-01 1.13911003e-01
-1.10511553e+00 2.73011643e-02 4.36208934e-01 2.82318980e-01
6.50576651e-01 -1.90452969e+00 -7.66547978e-01 5.86495638e-01
-3.84588867e-01 -5.91952384e-01 1.20157816e-01 1.48483849e+00
-1.07182705e+00 1.82638451e-01 -9.46549892e-01 -5.84431350e-01
-1.28893173e+00 2.81017069e-02 1.03997660e+00 -2.13719010e-01
-7.33239472e-01 7.38458097e-01 -6.75817132e-02 -8.64762127e-01
-4.27690417e-01 -3.65290642e-01 -5.20121038e-01 -1.71571851e-01
4.89803702e-01 5.82304418e-01 9.62938368e-03 -1.22172832e+00
-6.76306784e-01 1.17489088e+00 2.90288121e-01 -7.47467205e-02
1.54371130e+00 -2.20647350e-01 -9.40098226e-01 2.20066741e-01
1.36177838e+00 1.29181370e-01 -8.45228612e-01 -1.87228158e-01
3.93497139e-01 -3.26244920e-01 -1.74892619e-01 -1.65317059e-01
-1.36086750e+00 5.52275121e-01 1.02763462e+00 1.41108036e-01
4.84102279e-01 1.67271331e-01 5.42728662e-01 5.52467406e-02
4.89139706e-01 -1.14176118e+00 -4.10105795e-01 5.92731118e-01
1.30398071e+00 -8.19122374e-01 8.27172399e-02 -3.12199712e-01
-1.32474437e-01 1.30946159e+00 3.10014218e-01 -8.91650975e-01
1.16446722e+00 4.71846908e-01 1.58290453e-02 -6.95516407e-01
1.63070887e-01 5.16804159e-01 1.05212784e+00 3.96614790e-01
7.66800165e-01 1.24545656e-01 -1.04230773e+00 1.39870346e-01
-8.74483287e-01 -5.81581533e-01 2.72873282e-01 8.52129519e-01
-9.72885713e-02 -1.32862198e+00 -4.98316765e-01 3.39556068e-01
-5.56886375e-01 -3.28253061e-02 -2.94865370e-01 4.23773766e-01
5.85524691e-03 3.72624427e-01 3.55084807e-01 -1.11200484e-02
3.09091061e-01 7.07919672e-02 1.17025983e+00 -7.20795467e-02
-3.85628015e-01 -1.74909398e-01 -4.42766398e-01 -8.72735322e-01
-6.04118407e-01 -9.30273771e-01 -1.43672228e+00 -3.82532120e-01
2.99895555e-01 4.41503488e-02 6.77363694e-01 1.04956329e+00
1.94220975e-01 -1.19046263e-01 5.46661079e-01 -1.11588609e+00
-1.87779412e-01 -1.00179040e+00 -7.52672851e-01 8.26555550e-01
1.31445900e-01 -1.03628612e+00 -3.48938227e-01 2.47668266e-01]
|
[13.970860481262207, -2.5228018760681152]
|
f96d3b8e-7777-49be-ab4c-2ec32c06e18b
|
towards-the-universal-defense-for-query-based
|
2304.10088
| null |
https://arxiv.org/abs/2304.10088v1
|
https://arxiv.org/pdf/2304.10088v1.pdf
|
Towards the Universal Defense for Query-Based Audio Adversarial Attacks
|
Recently, studies show that deep learning-based automatic speech recognition (ASR) systems are vulnerable to adversarial examples (AEs), which add a small amount of noise to the original audio examples. These AE attacks pose new challenges to deep learning security and have raised significant concerns about deploying ASR systems and devices. The existing defense methods are either limited in application or only defend on results, but not on process. In this work, we propose a novel method to infer the adversary intent and discover audio adversarial examples based on the AEs generation process. The insight of this method is based on the observation: many existing audio AE attacks utilize query-based methods, which means the adversary must send continuous and similar queries to target ASR models during the audio AE generation process. Inspired by this observation, We propose a memory mechanism by adopting audio fingerprint technology to analyze the similarity of the current query with a certain length of memory query. Thus, we can identify when a sequence of queries appears to be suspectable to generate audio AEs. Through extensive evaluation on four state-of-the-art audio AE attacks, we demonstrate that on average our defense identify the adversary intent with over 90% accuracy. With careful regard for robustness evaluations, we also analyze our proposed defense and its strength to withstand two adaptive attacks. Finally, our scheme is available out-of-the-box and directly compatible with any ensemble of ASR defense models to uncover audio AE attacks effectively without model retraining.
|
['Lei Ju', 'Yuxuan Chen', 'Zheng Sun', 'Feng Guo']
|
2023-04-20
| null | null | null | null |
['audio-fingerprint']
|
['audio']
|
[ 3.63662362e-01 -2.50819743e-01 9.42481980e-02 1.49033172e-02
-1.24868286e+00 -1.09948647e+00 4.22433108e-01 -2.46436484e-02
-1.69133678e-01 2.58048326e-01 -2.19329298e-02 -6.95023894e-01
-5.93443699e-02 -7.07135677e-01 -7.13344514e-01 -6.18372560e-01
-3.36790562e-01 -3.41615118e-02 1.58906281e-01 -3.19492996e-01
1.08315602e-01 7.90903628e-01 -1.23309004e+00 5.73201478e-01
1.55736297e-01 1.21890593e+00 -2.38178268e-01 1.04672146e+00
1.72313645e-01 8.04683745e-01 -1.56500161e+00 -4.46789950e-01
3.99930388e-01 -1.20261326e-01 -6.25146866e-01 -4.73669767e-01
1.99200615e-01 -6.43098772e-01 -1.11590719e+00 1.00818419e+00
9.33837473e-01 -1.19543836e-01 1.47369787e-01 -1.36283505e+00
-3.83160591e-01 1.03812349e+00 -1.82287708e-01 4.69418287e-01
6.01194918e-01 3.47890854e-01 6.73671067e-01 -5.20016849e-01
-1.05038442e-01 1.15235782e+00 6.33351684e-01 8.76796961e-01
-8.54217172e-01 -1.48769188e+00 5.56147844e-02 3.34886789e-01
-1.55683661e+00 -8.87793422e-01 1.16979408e+00 1.02439955e-01
7.71976650e-01 7.13137269e-01 2.62581676e-01 1.65426993e+00
-4.96094450e-02 7.74236798e-01 8.18985760e-01 -2.01344907e-01
3.83135110e-01 6.74766153e-02 1.01500571e-01 3.27451676e-01
-4.57916670e-02 4.98255342e-01 -9.48661566e-01 -8.24531496e-01
1.18564300e-01 -2.31728166e-01 -3.79654914e-01 4.62825209e-01
-5.93927860e-01 7.30644226e-01 -3.63273337e-03 1.58004686e-01
-2.44114548e-01 4.56896305e-01 4.52953488e-01 5.96227765e-01
1.32152624e-02 3.49236310e-01 -3.54154736e-01 -3.30222040e-01
-8.37432981e-01 6.96266964e-02 7.84842610e-01 6.74153686e-01
5.00220478e-01 7.69507647e-01 -9.32890773e-02 4.70411777e-01
3.83427620e-01 8.01504791e-01 6.34593248e-01 -5.45787334e-01
4.11232978e-01 8.18880461e-03 -4.51750338e-01 -9.81490850e-01
1.59630001e-01 -5.98580420e-01 -7.12246060e-01 1.22959532e-01
3.67228985e-02 -2.05205679e-01 -6.85978353e-01 1.76490545e+00
7.39022642e-02 7.25561976e-01 4.66742724e-01 4.59601879e-01
5.23534775e-01 6.77888572e-01 -1.74898550e-01 -3.36620063e-02
1.38745952e+00 -3.49840820e-01 -7.06794262e-01 -4.35860753e-02
4.40898895e-01 -8.61813426e-01 9.87621188e-01 7.02649713e-01
-8.79013836e-01 -5.67823708e-01 -1.52027857e+00 6.81376100e-01
-3.44825149e-01 -5.05265892e-02 5.48965812e-01 1.65062737e+00
-8.96068037e-01 1.55881062e-01 -6.73882008e-01 2.50063002e-01
2.19613910e-01 6.39516890e-01 -1.78220838e-01 2.65843153e-01
-1.61020470e+00 8.42604265e-02 2.16828242e-01 4.61119376e-02
-1.56096721e+00 -7.80162096e-01 -5.99803150e-01 1.28543109e-01
3.54655087e-01 -2.77484149e-01 1.48326612e+00 -7.01218486e-01
-1.68929267e+00 5.08183420e-01 4.22421470e-02 -1.02242994e+00
4.02017444e-01 -1.44003779e-01 -1.18018067e+00 3.74957919e-01
-5.28940797e-01 5.14007136e-02 1.36436975e+00 -1.15642858e+00
-4.57784891e-01 -2.89482236e-01 2.11635694e-01 -4.30350423e-01
-8.49715352e-01 2.56966889e-01 -1.14683576e-01 -1.06469977e+00
-1.82744294e-01 -1.03296125e+00 -9.18845311e-02 -5.07675827e-01
-5.41780114e-01 1.80294827e-01 1.40541220e+00 -4.81114119e-01
1.67006922e+00 -2.42582035e+00 -6.71551049e-01 6.57512307e-01
1.46541372e-01 6.18831754e-01 -1.19861670e-01 5.11743903e-01
-2.03466147e-01 3.56473535e-01 3.97776850e-02 -7.52326325e-02
1.61387950e-01 -1.91077307e-01 -1.42834878e+00 4.10286397e-01
-6.66645542e-02 4.09693301e-01 -5.06360650e-01 1.90834771e-03
-4.19213204e-03 5.26714623e-01 -6.00558698e-01 5.27879298e-01
-5.30248992e-02 1.08178526e-01 -6.06335282e-01 6.55280948e-01
6.10189676e-01 3.42028379e-01 7.87735656e-02 -3.02668810e-01
3.62848073e-01 6.46853268e-01 -1.19120049e+00 1.28474283e+00
-4.41466510e-01 6.09789014e-01 9.18369964e-02 -6.75931633e-01
1.02098382e+00 7.93882906e-01 1.92597732e-01 -4.44491893e-01
1.13931946e-01 2.49248266e-01 3.05092216e-01 -3.33469883e-02
3.07630002e-01 4.47465122e-01 -3.22894335e-01 8.84683132e-01
-5.71639538e-02 9.45018902e-02 -5.45092404e-01 2.90431976e-01
1.52160692e+00 -6.61391675e-01 -5.45425490e-02 2.48163000e-01
8.68241429e-01 -6.02224171e-01 2.78764158e-01 1.33043242e+00
-2.01460421e-01 1.95379630e-01 5.65757491e-02 -2.87685990e-01
-5.25163829e-01 -1.27820051e+00 1.64431289e-01 9.13117588e-01
-1.01573281e-01 -6.35286093e-01 -8.66248310e-01 -7.33946621e-01
-2.50406533e-01 7.08933353e-01 -3.14595670e-01 -6.18518353e-01
-7.48416841e-01 -4.26505029e-01 1.89824724e+00 5.16896725e-01
5.33159614e-01 -1.04806960e+00 -4.23884273e-01 1.97932839e-01
5.74565306e-02 -1.22918403e+00 -5.46752274e-01 1.76678613e-01
-4.40114826e-01 -7.13732183e-01 -1.90786988e-01 -4.00526494e-01
1.17799602e-01 1.87881291e-01 6.06283247e-01 9.70816985e-02
-2.46660575e-01 7.23684072e-01 -2.74995923e-01 -6.19348109e-01
-1.13087332e+00 1.95471644e-01 5.65085948e-01 2.45445400e-01
4.24936354e-01 -7.99264669e-01 -4.77825612e-01 3.55686754e-01
-1.10591626e+00 -8.79931986e-01 5.41503727e-01 4.21432823e-01
1.97736502e-01 2.93277353e-01 8.74366760e-01 -4.00586426e-01
9.61428523e-01 -2.58398354e-01 -4.73377407e-01 3.39525253e-01
-3.01120609e-01 8.92047361e-02 8.34041655e-01 -1.16382217e+00
-6.08626544e-01 -7.33597651e-02 -6.50366008e-01 -7.91628301e-01
-1.30161628e-01 1.46151513e-01 -6.35110378e-01 -4.34988379e-01
6.81878865e-01 7.12588608e-01 -2.73962945e-01 -2.54736692e-01
1.73121601e-01 1.13223827e+00 6.97857201e-01 -8.14906836e-01
1.25274456e+00 3.29165488e-01 -1.98385715e-01 -9.74728167e-01
-2.72825569e-01 -1.45488590e-01 8.90954807e-02 -2.71757692e-01
2.74500459e-01 -8.20329845e-01 -1.26286423e+00 8.30471218e-01
-1.10359812e+00 -1.67523734e-02 2.20929042e-01 3.91663104e-01
-3.50159824e-01 5.90867221e-01 -7.55254447e-01 -1.18758428e+00
-7.85645008e-01 -1.29796958e+00 8.53202343e-01 -1.68950304e-01
-2.70543218e-01 -5.12121618e-01 -1.64655194e-01 3.13748896e-01
6.21818244e-01 -2.57961869e-01 9.55645204e-01 -1.38118255e+00
-6.20114863e-01 -5.07366240e-01 5.04275441e-01 3.31489027e-01
1.25549629e-01 -1.81530043e-01 -1.57116878e+00 -4.96523857e-01
4.79385793e-01 -3.04001123e-01 4.34584707e-01 -4.39270288e-02
1.58460057e+00 -6.81844532e-01 -6.00439236e-02 7.42618382e-01
8.59554112e-01 7.05013514e-01 8.70320082e-01 2.11365849e-01
4.43165243e-01 1.51072875e-01 3.21644843e-01 5.13027012e-01
-4.38897073e-01 8.38933527e-01 2.15084627e-01 2.03857496e-01
1.17313273e-01 -5.22894800e-01 8.44151556e-01 8.29514384e-01
4.94825333e-01 -4.38658237e-01 -6.66255116e-01 1.11927249e-01
-1.02846074e+00 -1.11600840e+00 4.11362469e-01 2.28388953e+00
9.34433758e-01 5.39087355e-01 8.97636265e-02 8.07291269e-01
6.54021680e-01 3.28750819e-01 -5.25765538e-01 -5.48292577e-01
-1.50961906e-01 7.39582002e-01 5.17500281e-01 3.17087740e-01
-9.49427783e-01 9.23093438e-01 6.35415173e+00 1.33962762e+00
-1.43900192e+00 7.75993615e-02 4.56385672e-01 4.70068865e-03
-3.02494884e-01 -2.50342786e-01 -8.10037374e-01 4.01815295e-01
1.37471509e+00 -3.31252992e-01 5.50804496e-01 7.68231809e-01
2.04218216e-02 7.54528642e-01 -1.15753698e+00 1.00392342e+00
3.92145216e-02 -1.16324604e+00 4.20911312e-01 1.58227086e-01
-8.81452560e-02 -4.60345119e-01 7.04210043e-01 3.06645513e-01
2.35658884e-01 -1.00256956e+00 6.96588874e-01 -4.27526832e-02
8.86847973e-01 -1.15705514e+00 3.71776819e-01 1.20197363e-01
-1.33586776e+00 -4.22861218e-01 -1.25640661e-01 3.16972971e-01
-5.04640043e-02 1.22744173e-01 -1.10765409e+00 3.06064308e-01
6.18336976e-01 -2.75356591e-01 -4.75365072e-01 5.26624739e-01
-1.67135790e-01 1.25278020e+00 -2.90879130e-01 -4.94164973e-02
1.43087953e-01 5.49325883e-01 8.21663439e-01 9.96988237e-01
4.53695029e-01 2.36173300e-03 -5.07555120e-02 6.61895216e-01
-2.31481314e-01 -2.09963545e-01 -7.30613530e-01 -2.08094269e-01
1.15040815e+00 7.56733894e-01 -2.11936548e-01 3.27939875e-02
1.09498881e-01 9.12309647e-01 -3.04061115e-01 2.66774654e-01
-9.81396437e-01 -6.39741123e-01 8.93446922e-01 -8.38630944e-02
5.54635152e-02 -2.19108969e-01 2.52018660e-01 -6.96854591e-01
-6.16001077e-02 -1.69041169e+00 4.26604211e-01 -4.03737158e-01
-1.19387567e+00 9.30067003e-01 -3.09572011e-01 -1.24750566e+00
-4.52590883e-01 -3.18579525e-01 -6.61441565e-01 5.80056489e-01
-1.01654458e+00 -1.05452108e+00 1.82581618e-01 1.03352714e+00
4.13817912e-01 -7.71767735e-01 1.16751802e+00 5.10589182e-01
-2.76254743e-01 1.63523602e+00 -4.18309301e-01 4.40798819e-01
4.63453203e-01 -7.10548937e-01 8.26836646e-01 1.03492582e+00
5.85598826e-01 9.75671470e-01 6.80902719e-01 -4.99226391e-01
-1.71183336e+00 -9.02930439e-01 1.40309900e-01 -3.82587194e-01
7.86539793e-01 -7.68926382e-01 -9.60640490e-01 5.05721867e-01
3.32235768e-02 -2.42601529e-01 1.16841221e+00 -2.93407232e-01
-7.73260534e-01 -3.88736635e-01 -1.11059570e+00 6.81018233e-01
6.90353036e-01 -1.32452393e+00 -4.77584779e-01 -7.62436688e-02
1.22241426e+00 -1.42042860e-01 -5.84255576e-01 5.12215734e-01
6.79775476e-01 -6.89691067e-01 1.17820954e+00 -6.40804291e-01
-3.85977000e-01 -3.36705834e-01 -5.65941751e-01 -7.53379285e-01
3.22534710e-01 -1.35798728e+00 -6.97194099e-01 1.52134764e+00
4.66737181e-01 -6.20017290e-01 8.92363608e-01 1.02670036e-01
-7.22140148e-02 -3.62072110e-01 -1.07679629e+00 -8.92019212e-01
-3.48586291e-01 -9.15933251e-01 1.07941389e+00 7.94296086e-01
-4.19942021e-01 -8.64840597e-02 -5.54522276e-01 1.05351901e+00
6.02318168e-01 -5.01638234e-01 9.12686884e-01 -7.55939484e-01
-7.48039782e-01 -1.93901837e-01 -6.30232155e-01 -9.00350630e-01
3.17947060e-01 -6.24276876e-01 -3.06061953e-01 -2.30969280e-01
-3.47020805e-01 -3.57929587e-01 -6.46856904e-01 2.78415024e-01
1.56338230e-01 1.56734809e-01 1.32322788e-01 3.04206926e-02
-1.73602328e-01 3.33792925e-01 1.37949944e-01 -5.35926402e-01
-3.04156959e-01 6.24960423e-01 -5.61346471e-01 5.45337915e-01
9.06036794e-01 -7.87010193e-01 -6.96216524e-01 -1.44633546e-01
-6.45909235e-02 1.61600932e-01 2.58490711e-01 -1.42076981e+00
3.54464918e-01 3.36983263e-01 6.17804900e-02 -5.93914747e-01
4.09149110e-01 -9.81570244e-01 2.40581021e-01 6.42499208e-01
-6.35520518e-01 2.73881137e-01 3.85761976e-01 6.33306682e-01
-1.24315165e-01 -1.75837100e-01 5.11808455e-01 3.81107092e-01
-4.28151429e-01 3.45000446e-01 -6.20237529e-01 -7.24855140e-02
7.93742299e-01 -1.56743035e-01 -3.17853957e-01 -7.02927887e-01
-3.94764930e-01 -4.16850865e-01 -1.18088968e-01 5.39873123e-01
8.90270829e-01 -1.13417590e+00 -4.83556986e-01 4.78019834e-01
-5.84866991e-03 -6.99986398e-01 3.31011295e-01 4.07793522e-02
-2.12781191e-01 3.70553583e-01 1.23924367e-01 -4.13842052e-01
-1.70871270e+00 7.66886950e-01 4.16165262e-01 -7.65034854e-02
-3.43883663e-01 6.60643816e-01 6.01758212e-02 -1.23586200e-01
8.99100542e-01 8.43198001e-02 1.00105055e-01 -3.66196096e-01
9.01894867e-01 1.83866084e-01 3.79348099e-01 -4.37292486e-01
-4.34844047e-01 2.14700446e-01 -1.32981777e-01 -4.45522189e-01
7.13581085e-01 2.52647370e-01 3.83565694e-01 1.36458784e-01
1.30945969e+00 6.89051807e-01 -5.91114640e-01 -1.17408268e-01
-1.25205681e-01 -3.88864368e-01 -4.11904976e-02 -6.31922722e-01
-1.09512115e+00 1.08980131e+00 1.03024054e+00 3.50823134e-01
1.31341386e+00 -2.65366733e-01 1.26869702e+00 7.32548833e-01
4.76933807e-01 -6.60902977e-01 4.42074627e-01 2.34770879e-01
5.09593844e-01 -4.60528016e-01 -4.36457813e-01 -5.69711700e-02
-3.19184273e-01 1.08243299e+00 4.54085231e-01 1.23490654e-01
7.84491539e-01 8.72660875e-01 2.90190130e-01 8.32521617e-02
-8.06681275e-01 3.44318062e-01 9.46993381e-02 8.33275199e-01
-1.62569448e-01 -1.37653127e-01 2.13927984e-01 1.01665616e+00
-5.68196952e-01 -4.78548884e-01 4.86074656e-01 8.72096360e-01
-2.62847632e-01 -1.21867752e+00 -6.31874621e-01 -2.29598954e-02
-1.15573347e+00 -2.01755255e-01 -6.25557423e-01 3.73043954e-01
-4.26828533e-01 1.32304811e+00 -2.14988708e-01 -1.15726280e+00
1.65487111e-01 -4.60973568e-02 3.38213183e-02 -1.23087674e-01
-8.38967502e-01 -5.93126416e-02 -9.67162009e-03 -3.49306703e-01
1.09981969e-01 -1.77091405e-01 -1.08276892e+00 -3.69656205e-01
-4.26296622e-01 2.86916763e-01 7.15434253e-01 6.93613589e-01
4.33741242e-01 5.45191944e-01 1.26045775e+00 -2.50025094e-01
-1.15696669e+00 -6.80260837e-01 -4.93692666e-01 1.37866680e-02
4.94281381e-01 -1.22365460e-01 -7.01691270e-01 -2.90398866e-01]
|
[13.979853630065918, 5.817153453826904]
|
305d7217-32e3-4f8e-b574-7c0d70c1c136
|
image-retrieval-with-a-bayesian-model-of
|
1603.09522
| null |
http://arxiv.org/abs/1603.09522v1
|
http://arxiv.org/pdf/1603.09522v1.pdf
|
Image Retrieval with a Bayesian Model of Relevance Feedback
|
A content-based image retrieval system based on multinomial relevance
feedback is proposed. The system relies on an interactive search paradigm where
at each round a user is presented with k images and selects the one closest to
their ideal target. Two approaches, one based on the Dirichlet distribution and
one based the Beta distribution, are used to model the problem motivating an
algorithm that trades exploration and exploitation in presenting the images in
each round. Experimental results show that the new approach compares favourably
with previous work.
|
['Shawe-Taylor John', 'Teh Yee Whye', 'Glowacka Dorota']
|
2016-03-31
| null | null | null | null |
['content-based-image-retrieval']
|
['computer-vision']
|
[ 1.57936081e-01 -1.51186541e-01 -2.69364893e-01 -3.85151088e-01
-7.23832190e-01 -5.05918741e-01 7.02380836e-01 2.84934729e-01
-1.05649388e+00 5.17430067e-01 -7.32151493e-02 -2.52941579e-01
-5.00164926e-01 -4.97491986e-01 -2.91641615e-02 -8.82347643e-01
2.21739300e-02 8.54622066e-01 5.49113035e-01 -1.34696737e-01
1.07219648e+00 3.72027099e-01 -1.88114691e+00 3.59839320e-01
4.22628254e-01 1.06551182e+00 6.62258983e-01 8.49211752e-01
-2.14084819e-01 5.88688850e-01 -5.86168528e-01 -3.08974385e-01
4.98645544e-01 -3.18274111e-01 -9.55887377e-01 -4.46265452e-02
1.01885502e-03 -4.31087077e-01 6.53566346e-02 1.10488307e+00
6.62989020e-01 3.61823916e-01 1.14774370e+00 -1.11180139e+00
-6.47700787e-01 1.10598132e-01 -5.93493402e-01 6.47187650e-01
7.14806378e-01 -1.38967127e-01 9.02664185e-01 -9.59247887e-01
7.38448739e-01 1.47779810e+00 -3.03617865e-02 2.75311947e-01
-1.24567235e+00 -3.19711208e-01 3.15713361e-02 8.59965265e-01
-1.63552213e+00 -1.71733320e-01 5.24541020e-01 -4.10291910e-01
8.04244697e-01 4.07409370e-01 6.81225717e-01 2.90023595e-01
2.30033800e-01 8.06180596e-01 1.30555439e+00 -1.09420991e+00
5.57775021e-01 6.64341867e-01 -1.40479222e-01 2.73844719e-01
-1.41443297e-01 3.22136492e-01 -5.13926744e-01 -5.47858655e-01
6.70817494e-01 -1.29356995e-01 -6.59429431e-02 -6.17083251e-01
-7.89605737e-01 1.21708310e+00 5.27707458e-01 2.97288030e-01
-6.17771506e-01 6.70239627e-02 1.41164839e-01 4.33649987e-01
4.37500298e-01 3.22826743e-01 -6.39402121e-02 3.13493907e-01
-9.77762520e-01 2.91357070e-01 7.55352557e-01 3.77265841e-01
8.19130123e-01 -8.45779061e-01 -3.24004829e-01 6.38492346e-01
9.76503432e-01 4.98583823e-01 7.72865653e-01 -7.23688662e-01
-3.68430525e-01 4.48604345e-01 3.70063871e-01 -1.18221247e+00
2.63702512e-01 3.41666907e-01 -9.18032899e-02 6.70510113e-01
-1.02705151e-01 2.18255386e-01 -1.13718534e+00 9.98222530e-01
4.58679467e-01 -3.37559521e-01 -1.27242044e-01 1.04918742e+00
6.60878420e-01 6.32106066e-01 2.00411305e-01 -3.87512594e-01
1.19967222e+00 -7.91579723e-01 -6.28286958e-01 4.46865112e-01
-2.28118894e-04 -1.09903884e+00 6.51474297e-01 7.37025917e-01
-1.22888303e+00 -3.38295072e-01 -7.06128538e-01 2.41061136e-01
-6.61285400e-01 -8.08447078e-02 1.79081723e-01 8.64433587e-01
-1.68009794e+00 4.27045971e-01 -2.64794886e-01 -4.72948194e-01
-2.90128570e-02 4.56180423e-01 -9.16004628e-02 3.98560055e-02
-1.17037511e+00 8.89479399e-01 6.19260252e-01 2.19731573e-02
-8.20889294e-01 5.02885729e-02 -3.39852691e-01 -1.39070049e-01
1.23163283e-01 -5.47298491e-01 1.39652956e+00 -1.27528775e+00
-1.71571076e+00 1.05692244e+00 4.44528535e-02 -4.95296031e-01
5.39129198e-01 6.65408149e-02 1.09411433e-01 7.49337912e-01
6.90870509e-02 1.07082570e+00 1.28352809e+00 -1.44101548e+00
-1.06986237e+00 -3.47285002e-01 5.31299487e-02 7.50717700e-01
-2.38735884e-01 3.47667545e-01 -8.10325921e-01 -2.77454555e-01
-3.42001989e-02 -8.16206157e-01 -5.52182436e-01 -5.98299503e-02
8.19743518e-03 -5.28589547e-01 5.96105516e-01 -2.30599657e-01
1.31350768e+00 -1.84716308e+00 5.49395420e-02 9.38818991e-01
-9.41021070e-02 2.32807428e-01 3.01046204e-02 7.18743563e-01
3.35683227e-02 3.98172349e-01 4.76663977e-01 -1.35103405e-01
-1.38866767e-01 1.22970268e-01 -7.40704164e-02 3.89344633e-01
-2.36875549e-01 4.77815270e-01 -8.89477849e-01 -9.75582659e-01
2.38793030e-01 2.96273023e-01 -4.06712711e-01 2.22777039e-01
-1.35531694e-01 -8.42548814e-03 -4.88550097e-01 3.93204391e-01
5.69635570e-01 -3.23781520e-01 1.81564257e-01 1.42496720e-01
-1.47419214e-01 -3.93330753e-01 -1.35458684e+00 1.22453141e+00
-7.04657426e-03 5.09175420e-01 -4.33024392e-02 -8.36876273e-01
6.95543528e-01 3.34953129e-01 2.97237486e-01 -8.62709641e-01
1.01616614e-01 1.57173693e-01 -3.83331150e-01 -5.83279133e-01
6.77958071e-01 8.09178781e-03 2.85640240e-01 7.67014682e-01
-1.14511393e-01 -8.97378176e-02 3.82762372e-01 5.62972665e-01
7.37950146e-01 1.05528392e-01 4.19535071e-01 -3.66096735e-01
5.01155257e-01 5.55123240e-02 8.95167291e-02 1.17138970e+00
-2.70249277e-01 4.88656998e-01 3.08532506e-01 -4.27989125e-01
-8.66040051e-01 -8.50021303e-01 1.33621216e-01 1.35553527e+00
6.58710480e-01 -1.52175650e-01 -5.74338675e-01 -6.06883705e-01
-5.63877113e-02 2.76360929e-01 -8.99992049e-01 1.22018009e-01
7.00290687e-03 -2.69417167e-01 -2.29357094e-01 -1.63176358e-01
2.18896031e-01 -1.48863757e+00 -1.13748062e+00 -1.17483899e-01
-5.05913198e-02 -1.45289809e-01 -3.97991031e-01 8.03278983e-02
-7.58108079e-01 -8.05621922e-01 -1.08180010e+00 -8.03324580e-01
8.75800908e-01 4.50331807e-01 1.20975888e+00 3.16399902e-01
-6.48046255e-01 9.68240261e-01 -9.73433852e-01 -5.44161201e-01
-3.67043376e-01 -9.62907970e-02 -1.62393346e-01 6.83647171e-02
5.64059913e-01 3.84021029e-02 -1.17887723e+00 4.55120683e-01
-1.37265348e+00 -3.70345235e-01 1.01811481e+00 8.50094318e-01
7.51690209e-01 1.97107449e-01 2.27027670e-01 -7.52331078e-01
8.48062038e-01 -5.32189429e-01 -7.03181624e-01 5.95410883e-01
-1.14304078e+00 1.56815544e-01 -2.50773519e-01 -3.36593658e-01
-9.76956844e-01 3.17067951e-01 4.32571560e-01 -3.26136589e-01
-7.96615705e-02 3.77249449e-01 3.59924972e-01 -4.26558673e-01
8.55601549e-01 1.52417645e-01 9.60877687e-02 -3.61105770e-01
4.17217553e-01 7.03941345e-01 2.61838317e-01 6.40956461e-02
3.71375948e-01 3.69063109e-01 -3.79274487e-01 -8.17629218e-01
2.92878132e-02 -1.17538869e+00 -3.95894766e-01 -6.79082811e-01
7.91824877e-01 -5.95250428e-01 -6.73678160e-01 3.87581170e-01
-8.84754181e-01 9.81347114e-02 -3.26144457e-01 5.70820570e-01
-5.69065154e-01 4.87765878e-01 -1.88055769e-01 -1.24649668e+00
-2.95849442e-01 -1.08907413e+00 1.01704907e+00 6.31761670e-01
1.35188982e-01 -7.80017674e-01 3.18635046e-01 2.69992948e-01
5.06752849e-01 -5.12088299e-01 7.11014092e-01 -8.64740372e-01
-7.37644434e-01 -4.63160366e-01 -2.69372106e-01 3.39160673e-02
-5.78214414e-02 -5.74469790e-02 -7.37411499e-01 -3.15417409e-01
-1.44560933e-01 -4.67217565e-01 8.81930232e-01 5.01684368e-01
7.25993872e-01 -2.18132451e-01 -5.87191761e-01 -2.38997191e-01
1.74377441e+00 4.61869061e-01 8.84435296e-01 6.29412234e-01
-2.83470660e-01 5.05053699e-01 1.08369398e+00 5.82972050e-01
2.68106461e-01 5.33171296e-01 6.68097675e-01 -2.04946399e-01
3.86683315e-01 2.31375158e-01 -6.98138252e-02 3.21956575e-02
2.12397665e-01 -5.30144453e-01 -7.91298807e-01 8.07708561e-01
-1.94827044e+00 -8.65405202e-01 5.06224334e-01 2.44194889e+00
8.90123844e-01 2.29732636e-02 3.14480484e-01 4.55362983e-02
7.74627507e-01 -2.70868957e-01 -3.39649379e-01 -4.89864439e-01
1.77760854e-01 1.00055866e-01 5.27128458e-01 6.77802265e-01
-1.10522997e+00 7.33843327e-01 7.60538673e+00 1.10145462e+00
-9.37329233e-01 -9.71700773e-02 9.35624063e-01 2.18261359e-03
-2.71247447e-01 1.69693142e-01 -7.64438629e-01 3.37667912e-01
6.87961102e-01 -3.98893893e-01 1.73995346e-01 7.36450791e-01
1.46476015e-01 -1.26608920e+00 -6.64042234e-01 1.06829894e+00
4.08338547e-01 -9.09782887e-01 3.18683207e-01 9.31582004e-02
6.91336691e-01 -3.02651167e-01 2.59602159e-01 -2.53984332e-01
5.02865374e-01 -1.00900614e+00 6.56207681e-01 9.27275896e-01
4.39406902e-01 -7.42914438e-01 8.22851121e-01 3.84959579e-01
-7.12891638e-01 -5.12563512e-02 -2.50276327e-01 1.73169747e-01
5.27295582e-02 1.29237607e-01 -1.16682613e+00 3.60450596e-01
1.06928325e+00 -1.68234900e-01 -6.72458649e-01 1.71476316e+00
-6.88752905e-02 1.65828645e-01 -5.51145852e-01 -5.51959813e-01
2.49308124e-01 -1.64791241e-01 5.66999793e-01 1.25052559e+00
6.48306310e-02 1.73086628e-01 4.05533075e-01 2.23267749e-01
5.16269207e-01 6.50782943e-01 -6.80656195e-01 3.79387438e-01
4.28848505e-01 1.28647077e+00 -1.26597357e+00 -5.09215236e-01
1.46737576e-01 1.30669069e+00 -1.00156881e-01 2.37227514e-01
-1.16823442e-01 -5.44382930e-01 -2.90023029e-01 2.55017132e-02
8.37670803e-01 2.65488058e-01 3.55236709e-01 -3.67245018e-01
-2.12075830e-01 -7.35571086e-01 7.75654197e-01 -9.63235795e-01
-1.17197239e+00 7.01677799e-01 5.88453054e-01 -1.02071202e+00
-6.89156175e-01 -3.22527379e-01 -2.84058124e-01 1.12839615e+00
-1.63371766e+00 -7.78843403e-01 -6.42491318e-03 5.69925070e-01
4.57242310e-01 -1.13494761e-01 7.26580799e-01 1.31534375e-02
1.55994564e-01 2.21442655e-01 3.20894361e-01 -5.54452538e-01
9.08665895e-01 -1.42321396e+00 -3.48141819e-01 3.72702777e-01
1.55899733e-01 4.68024552e-01 8.72561932e-01 -5.67666829e-01
-8.96663070e-01 -2.27630705e-01 1.02975965e+00 -7.17461035e-02
2.47574195e-01 2.71566033e-01 -5.84953368e-01 -5.40336361e-04
6.51605606e-01 -2.71773040e-01 9.36028719e-01 -3.46234918e-01
1.72060609e-01 1.67370532e-02 -1.40757298e+00 3.77992362e-01
8.04182813e-02 -2.80745387e-01 -4.95377809e-01 5.12775183e-01
8.07681084e-02 2.27512624e-02 -4.28786933e-01 9.39771384e-02
6.85633719e-01 -1.21618569e+00 9.16717172e-01 -1.77332580e-01
-5.81314005e-02 -3.52044135e-01 -9.34584066e-02 -1.26305306e+00
-5.04674017e-01 -6.14535987e-01 4.52124715e-01 5.91819942e-01
4.25326169e-01 -3.87664348e-01 8.76103044e-01 5.73835373e-01
9.57943261e-01 -6.76796734e-01 -9.70400572e-01 -4.58991736e-01
-2.83785999e-01 1.87514246e-01 5.44869788e-02 7.33534172e-02
-4.64793406e-02 1.99859276e-01 -1.42717376e-01 -1.59829214e-01
5.53894222e-01 3.92805040e-02 3.42632979e-01 -1.30794001e+00
-2.09506467e-01 -6.48663461e-01 -5.84476829e-01 -9.17513192e-01
-4.33086038e-01 -3.55976671e-01 4.41906273e-01 -1.43314159e+00
5.42475462e-01 -4.15808916e-01 -6.24386489e-01 1.28037781e-01
-3.21711242e-01 7.09930360e-01 8.31046700e-02 5.29413342e-01
-1.09785092e+00 1.25961736e-01 6.10724330e-01 1.38988674e-01
-2.54309058e-01 1.09321043e-01 -4.82055962e-01 6.64392412e-01
5.09222269e-01 -4.87612903e-01 -5.46811581e-01 1.71301130e-03
5.06691635e-01 1.19295329e-01 2.52504319e-01 -4.60446477e-01
7.76764929e-01 -3.19576566e-03 3.66760254e-01 -1.02864468e+00
2.25995585e-01 -8.85986507e-01 9.34809223e-02 5.70290923e-01
-5.99802613e-01 3.75531554e-01 -1.33112982e-01 1.04682124e+00
-3.73827279e-01 -9.37413156e-01 6.11238003e-01 -4.69840616e-01
-9.81520295e-01 8.09977651e-02 -6.50432110e-01 -7.54466772e-01
1.23171413e+00 -4.32443708e-01 4.78487253e-01 -8.83434832e-01
-9.58429277e-01 3.78111750e-01 4.61197197e-01 1.51248410e-01
8.09533238e-01 -1.38438535e+00 -6.07980132e-01 1.07474431e-01
9.94959623e-02 -5.55809617e-01 1.84660763e-01 4.77839351e-01
-6.70820057e-01 4.56658632e-01 -1.08330183e-01 -6.36443555e-01
-1.78659368e+00 7.97865629e-01 1.72372386e-02 -6.57816768e-01
1.15642548e-01 1.07454824e+00 7.19440654e-02 1.17113322e-01
4.56649780e-01 5.54587543e-01 -9.18315172e-01 4.11842585e-01
7.40816712e-01 3.09397966e-01 4.44564372e-02 -5.68781495e-01
-3.10987651e-01 4.53811109e-01 -4.88876909e-01 -9.07125413e-01
8.76608431e-01 -5.39241076e-01 -3.82025540e-01 5.14258802e-01
9.07934785e-01 -4.21272814e-01 -8.35344136e-01 -4.47666436e-01
1.43607706e-01 -1.05109477e+00 4.23543334e-01 -1.03856158e+00
-6.67999268e-01 3.97659391e-01 1.31361055e+00 7.85527587e-01
1.43209040e+00 -1.97016895e-01 -5.02433665e-02 5.76064706e-01
3.12770307e-01 -1.46978414e+00 3.00408512e-01 1.39968887e-01
8.58465254e-01 -1.45659053e+00 3.53153527e-01 2.44601425e-02
-8.02934110e-01 1.03303623e+00 1.76382959e-01 -2.50062197e-01
9.72084045e-01 -2.43618399e-01 2.47833997e-01 -3.71717989e-01
-9.69694495e-01 -2.65267342e-01 2.87331551e-01 5.78461170e-01
1.57508597e-01 -2.42648467e-01 -8.32948506e-01 -3.21191639e-01
3.84271324e-01 1.79288253e-01 2.04402558e-03 1.41468501e+00
-7.31052697e-01 -1.41426480e+00 -7.97651768e-01 4.45144117e-01
-6.64026260e-01 -1.76779762e-01 -6.58610463e-01 4.46266174e-01
-1.02723509e-01 1.12346470e+00 -1.07079512e-03 -3.07506137e-02
-1.11551739e-01 -1.88439906e-01 4.57224488e-01 -4.55115050e-01
-7.21402645e-01 4.40981507e-01 -4.63277370e-01 -5.37162840e-01
-6.94631636e-01 -7.43010581e-01 -6.32468820e-01 1.49378106e-01
-9.01413858e-01 6.55641019e-01 1.03411710e+00 5.03042281e-01
4.04894561e-01 -2.49659672e-01 1.04406595e+00 -1.28307903e+00
-7.47380435e-01 -8.34666133e-01 -5.84119439e-01 1.58328936e-01
1.70365334e-01 -5.95508635e-01 -3.86661351e-01 2.77587593e-01]
|
[10.752792358398438, 0.11138852685689926]
|
658c0fc0-7756-43e6-b5d6-1cce6e5717d8
|
generate-to-understand-for-representation
|
2306.10056
| null |
https://arxiv.org/abs/2306.10056v1
|
https://arxiv.org/pdf/2306.10056v1.pdf
|
Generate to Understand for Representation
|
In recent years, a significant number of high-quality pretrained models have emerged, greatly impacting Natural Language Understanding (NLU), Natural Language Generation (NLG), and Text Representation tasks. Traditionally, these models are pretrained on custom domain corpora and finetuned for specific tasks, resulting in high costs related to GPU usage and labor. Unfortunately, recent trends in language modeling have shifted towards enhancing performance through scaling, further exacerbating the associated costs. Introducing GUR: a pretraining framework that combines language modeling and contrastive learning objectives in a single training step. We select similar text pairs based on their Longest Common Substring (LCS) from raw unlabeled documents and train the model using masked language modeling and unsupervised contrastive learning. The resulting model, GUR, achieves impressive results without any labeled training data, outperforming all other pretrained baselines as a retriever at the recall benchmark in a zero-shot setting. Additionally, GUR maintains its language modeling ability, as demonstrated in our ablation experiment. Our code is available at \url{https://github.com/laohur/GUR}.
|
['Xiaoqing Liu', 'Xiande Zhong', 'Changshang Xue']
|
2023-06-14
| null | null | null | null |
['contrastive-learning', 'contrastive-learning', 'text-generation']
|
['computer-vision', 'methodology', 'natural-language-processing']
|
[ 2.87475199e-01 -1.70995191e-01 -6.22006357e-01 -3.37372541e-01
-1.38552701e+00 -6.43706679e-01 1.01543224e+00 3.33709747e-01
-7.16367364e-01 7.02514112e-01 4.52236950e-01 -3.57223392e-01
4.05675948e-01 -7.09847152e-01 -5.89069486e-01 -2.66913533e-01
2.42913023e-01 8.16915512e-01 3.31717916e-02 -2.39518836e-01
2.55490541e-01 1.11519076e-01 -1.30762458e+00 5.99851131e-01
9.26874697e-01 6.51378989e-01 3.31504047e-01 6.17321372e-01
-5.82308054e-01 6.76461399e-01 -4.88603860e-01 -4.30906743e-01
9.07538906e-02 -3.38286400e-01 -9.92405891e-01 -2.48517707e-01
4.23988014e-01 -3.21246535e-01 -4.62375939e-01 7.22443938e-01
6.40498102e-01 1.96082339e-01 6.59269452e-01 -8.67315114e-01
-9.07158732e-01 9.15055156e-01 -6.40484929e-01 1.16879597e-01
1.31807551e-01 -5.80988750e-02 1.27907729e+00 -1.27480149e+00
6.22676015e-01 1.35780907e+00 3.09258848e-01 6.55564964e-01
-1.40492463e+00 -7.83463717e-01 1.45631462e-01 -1.18373828e-02
-1.48641872e+00 -5.83871484e-01 4.39237654e-01 -2.70256847e-01
1.30556715e+00 -1.12704402e-02 -4.24254388e-02 1.46377194e+00
-7.66583979e-02 1.29244256e+00 7.59102702e-01 -7.97989249e-01
2.59596817e-02 1.94541693e-01 3.28326404e-01 6.26993835e-01
2.40056545e-01 -1.72584593e-01 -6.09212339e-01 -3.05463582e-01
4.33109522e-01 1.91660058e-02 -1.99294597e-01 -9.25869495e-02
-1.15925395e+00 9.83765066e-01 3.03795546e-01 3.65622103e-01
-1.21718571e-01 5.12425043e-02 4.97661650e-01 2.65951604e-01
8.19181085e-01 6.37160122e-01 -3.67106199e-01 -1.26620218e-01
-1.16594803e+00 2.67161697e-01 7.87940204e-01 1.11258507e+00
7.13401675e-01 -1.01652436e-01 -4.08417106e-01 1.19015765e+00
2.01539621e-02 5.16126335e-01 8.12487900e-01 -4.92198706e-01
6.23512745e-01 6.00220144e-01 -7.84854591e-02 -5.62576056e-01
-2.95335650e-01 -5.21255493e-01 -8.13046813e-01 -4.07806396e-01
9.21282396e-02 -7.86362868e-03 -1.23078907e+00 1.75949144e+00
-1.45412490e-01 1.24166854e-01 8.52697939e-02 5.88664174e-01
8.71569157e-01 1.00354493e+00 3.11700433e-01 1.04173593e-01
1.12586665e+00 -1.31079602e+00 -4.06863391e-01 -6.70370102e-01
1.14989436e+00 -9.70349252e-01 1.61245966e+00 3.58082741e-01
-1.09043288e+00 -3.19045246e-01 -9.68134224e-01 -6.21830940e-01
-4.69750285e-01 3.66071939e-01 6.02622271e-01 3.73556703e-01
-1.07727742e+00 3.23047608e-01 -8.17578256e-01 -4.58109587e-01
3.40383381e-01 1.35004610e-01 -1.12980120e-01 -3.14304590e-01
-1.29821086e+00 7.04432607e-01 4.82608020e-01 -3.06880713e-01
-8.97342563e-01 -6.89713299e-01 -7.19197273e-01 1.88554153e-01
4.52538788e-01 -7.63329566e-01 1.49477541e+00 -7.01870978e-01
-1.22241938e+00 1.05262339e+00 -3.32669288e-01 -7.07988083e-01
3.84529591e-01 -6.35346115e-01 -4.18188602e-01 -1.76312968e-01
1.19709387e-01 9.24869299e-01 6.23162985e-01 -1.08549881e+00
-4.43100095e-01 -1.79624885e-01 -1.09086722e-01 2.96795577e-01
-5.06624699e-01 1.18616320e-01 -8.15205514e-01 -9.86997783e-01
-2.00917453e-01 -8.30481112e-01 -1.43443361e-01 -3.65787089e-01
-3.91542673e-01 -4.11292315e-01 5.88807225e-01 -6.25480235e-01
1.54222310e+00 -1.96988904e+00 6.06174069e-03 -5.11500239e-02
6.39578775e-02 4.19663459e-01 -5.26154935e-01 7.67672181e-01
1.85118899e-01 4.44644570e-01 -4.28959966e-01 -6.57007396e-01
2.94712167e-02 -9.72546265e-02 -8.92179549e-01 -4.59379703e-02
1.57395557e-01 1.20939004e+00 -9.80239689e-01 -4.64376569e-01
-1.03751868e-01 3.74540955e-01 -5.75951278e-01 2.89632082e-01
-7.41786540e-01 3.12423147e-02 -3.36181521e-01 5.37595570e-01
3.17446768e-01 -5.88865936e-01 2.26782456e-01 2.13925928e-01
3.06704849e-01 7.91213393e-01 -6.59301221e-01 2.00903726e+00
-7.70108342e-01 5.85562706e-01 -2.31557071e-01 -8.88399780e-01
9.78926182e-01 1.96546644e-01 1.42472148e-01 -9.19363976e-01
3.77571546e-02 3.44619960e-01 -3.23415399e-01 -1.55100256e-01
9.26667571e-01 8.71224925e-02 -2.30634451e-01 8.99713814e-01
2.44347647e-01 -1.89019844e-01 3.27823997e-01 7.40011275e-01
1.05921078e+00 7.56892487e-02 3.47811490e-01 -1.13832466e-01
3.19680661e-01 9.73599702e-02 7.63100609e-02 9.72752035e-01
3.48215282e-01 6.88251555e-01 1.65561706e-01 -1.64940670e-01
-9.91851926e-01 -1.14597297e+00 1.31796272e-02 1.52715075e+00
-4.38002385e-02 -7.52090037e-01 -5.12421966e-01 -6.32382154e-01
-4.88175936e-02 1.01925969e+00 -3.04167986e-01 -4.49813575e-01
-7.35730171e-01 -7.22746611e-01 7.64706910e-01 4.57980156e-01
2.61054009e-01 -1.10597289e+00 -5.73294424e-02 3.13623250e-01
-3.30271631e-01 -1.11661303e+00 -7.11108744e-01 1.98736504e-01
-8.84073257e-01 -5.33798099e-01 -7.46342957e-01 -9.39976156e-01
5.70808351e-01 4.97576505e-01 1.60469329e+00 2.09345788e-01
-4.01252687e-01 2.12860078e-01 -3.68492723e-01 -4.76072460e-01
-3.47111225e-01 5.44511199e-01 -7.76722431e-02 -3.92486811e-01
6.67388320e-01 -3.93611372e-01 -4.64617044e-01 -2.90193409e-02
-1.12230492e+00 1.64939031e-01 7.40537763e-01 1.09375334e+00
5.47215939e-01 -4.38103765e-01 7.04448283e-01 -1.08826840e+00
9.66677189e-01 -6.35433137e-01 -4.95821804e-01 5.21882534e-01
-7.78770268e-01 2.14331746e-01 6.12362742e-01 -3.94406408e-01
-1.08726335e+00 -3.92232627e-01 1.28630400e-01 -1.79960638e-01
-6.51866244e-03 6.58379972e-01 4.03346680e-02 4.45301861e-01
9.32597697e-01 4.88163799e-01 -3.14664066e-01 -6.37332022e-01
6.07561886e-01 8.87977421e-01 4.07976896e-01 -6.97491467e-01
5.88197649e-01 2.46562898e-01 -5.87632000e-01 -8.26971233e-01
-1.08934462e+00 -5.41673839e-01 -4.75424081e-01 3.30981642e-01
4.59908098e-01 -1.24873936e+00 6.42275363e-02 1.97124779e-01
-1.24869311e+00 -5.43286204e-01 -8.35210923e-03 4.23930615e-01
-1.68773636e-01 2.45191723e-01 -6.12945974e-01 -6.02087736e-01
-7.27641702e-01 -9.45421040e-01 1.32345164e+00 -4.32625227e-02
-3.69837284e-01 -1.01716721e+00 9.13286656e-02 3.86259407e-01
4.17614341e-01 -3.83606941e-01 1.17890227e+00 -9.60232556e-01
-5.39658129e-01 -2.98439145e-01 -2.62078941e-01 2.55785376e-01
-5.72984144e-02 -3.81627172e-01 -9.28499520e-01 -4.86325502e-01
-3.93865764e-01 -8.84829700e-01 1.17338896e+00 1.56654894e-01
1.16697562e+00 -2.34605312e-01 -5.86738646e-01 4.42648262e-01
1.20680046e+00 9.07227248e-02 2.17933521e-01 2.78179675e-01
5.53176999e-01 4.37353969e-01 4.52829093e-01 2.01863959e-01
3.04592699e-01 6.75608099e-01 -1.72853157e-01 -1.26508921e-01
-2.24627525e-01 -6.20843887e-01 3.37716818e-01 7.78219461e-01
4.49007213e-01 -7.86967397e-01 -1.25947821e+00 6.49037480e-01
-1.83970177e+00 -7.58804202e-01 2.91386425e-01 2.21820378e+00
1.14451003e+00 1.98954791e-01 -1.33096352e-01 -4.09294635e-01
5.75672328e-01 2.02104539e-01 -5.93858778e-01 -2.98001617e-01
-1.89876750e-01 4.81906116e-01 2.60652721e-01 7.18004882e-01
-9.86465335e-01 1.42225182e+00 6.17296696e+00 9.50363815e-01
-1.29856086e+00 9.55057368e-02 8.95528257e-01 -5.44540942e-01
-5.56488872e-01 1.13013862e-02 -1.04833746e+00 3.08733076e-01
9.96216357e-01 -6.63157523e-01 3.80648524e-01 8.57360780e-01
3.04453932e-02 8.89750477e-03 -1.05992746e+00 1.02127051e+00
3.03130001e-01 -1.32915759e+00 5.10673642e-01 -2.18042418e-01
7.94454575e-01 4.76418942e-01 9.86675769e-02 7.82194853e-01
5.68293691e-01 -1.15261829e+00 3.46029252e-01 1.75680414e-01
7.73747325e-01 -6.04267418e-01 4.69527930e-01 4.68619317e-01
-8.67140770e-01 2.47995898e-01 -4.30133909e-01 1.77003175e-01
2.89736748e-01 6.04139984e-01 -1.21055734e+00 3.87552649e-01
3.88828069e-01 4.90504861e-01 -6.56302989e-01 6.63397372e-01
-1.98551640e-01 8.98549080e-01 -2.10530207e-01 -1.93656161e-01
4.52112734e-01 -4.07626852e-02 5.02905548e-01 1.48172367e+00
2.55438983e-01 -2.09035818e-02 3.23014796e-01 9.51971292e-01
-6.73032820e-01 3.51353168e-01 -5.59015691e-01 -4.28243399e-01
7.30746567e-01 1.13699770e+00 -5.39109349e-01 -6.35120809e-01
-4.25276339e-01 9.56371963e-01 5.78324616e-01 5.00213444e-01
-6.00132287e-01 -5.20073712e-01 4.06938732e-01 5.77506945e-02
3.47354673e-02 -3.54541808e-01 -3.44957829e-01 -1.38580644e+00
1.48996204e-01 -1.00703633e+00 5.11942863e-01 -5.48759103e-01
-1.17407274e+00 9.29491341e-01 6.38662726e-02 -1.04640567e+00
-6.03021860e-01 -4.00413126e-01 -3.85082006e-01 1.05012178e+00
-1.47316527e+00 -1.05565667e+00 -8.26190785e-02 3.19766372e-01
8.69197309e-01 -3.91056687e-01 8.88768852e-01 3.08053911e-01
-4.38385814e-01 7.95700848e-01 2.61345685e-01 9.87113491e-02
1.08370864e+00 -1.10077345e+00 8.61860931e-01 9.19419050e-01
5.16796947e-01 9.15535212e-01 3.24101359e-01 -6.32253110e-01
-1.20850694e+00 -1.18329048e+00 1.28623509e+00 -5.00904143e-01
7.61500478e-01 -7.92694509e-01 -1.05393326e+00 8.21113229e-01
2.88718969e-01 -2.83352822e-01 7.62227118e-01 2.68935204e-01
-6.18724525e-01 1.47326663e-01 -6.45577908e-01 8.64169717e-01
1.02654886e+00 -7.93747723e-01 -6.13142490e-01 6.05316341e-01
9.75991130e-01 -2.21964374e-01 -3.80072206e-01 2.88400918e-01
3.92572761e-01 -4.96185631e-01 9.07200873e-01 -8.74158621e-01
6.23799205e-01 1.12838589e-01 3.25867087e-02 -1.10750103e+00
-1.64670229e-01 -5.75781405e-01 -3.05605292e-01 1.20083010e+00
7.86253810e-01 -4.48848426e-01 7.42432952e-01 5.05559325e-01
-1.95128191e-02 -7.30725229e-01 -3.57065707e-01 -9.06354368e-01
4.08762693e-01 -4.32384312e-01 4.40190792e-01 9.01558757e-01
3.69907767e-02 9.38216090e-01 -5.00511527e-01 -2.94713616e-01
3.32766324e-01 2.10107893e-01 7.90273130e-01 -9.39059436e-01
-4.33938920e-01 -7.00521648e-01 2.73800552e-01 -1.59814775e+00
3.34020793e-01 -1.42288685e+00 6.54895306e-02 -1.56775808e+00
3.51230323e-01 -3.09417248e-01 -1.16360761e-01 6.00394249e-01
-4.51028436e-01 1.90860257e-01 9.33326259e-02 4.02765483e-01
-6.73824787e-01 6.04873002e-01 8.34813893e-01 -2.16363594e-01
-3.14306170e-01 -2.46483579e-01 -8.34647000e-01 4.59445924e-01
9.00406122e-01 -4.09660548e-01 -5.61918437e-01 -1.07606947e+00
1.00594640e-01 -1.55439287e-01 2.20566839e-02 -7.60294795e-01
2.91525722e-01 9.13390666e-02 1.22243896e-01 -5.04109919e-01
3.37297231e-01 -3.28685135e-01 -4.09826934e-01 2.70673484e-01
-8.57898235e-01 2.29193613e-01 4.71592188e-01 3.89260054e-01
-3.93182099e-01 -2.57984728e-01 6.26015842e-01 -9.35419649e-02
-7.27765501e-01 4.20329154e-01 -2.63255805e-01 3.63258988e-01
5.62112987e-01 2.92712629e-01 -4.80880111e-01 -5.59707642e-01
-3.53724539e-01 2.94656008e-01 3.54467034e-01 8.51625979e-01
4.17610496e-01 -1.16638553e+00 -8.54692698e-01 9.94896442e-02
4.45179820e-01 8.28579143e-02 6.32081255e-02 4.50698853e-01
-3.38750094e-01 8.52234006e-01 3.31486642e-01 -4.09202904e-01
-1.01451480e+00 4.99395639e-01 -7.14131445e-02 -6.93161726e-01
-4.61499542e-01 1.00296330e+00 2.80902028e-01 -6.65231645e-01
3.75097454e-01 -6.37842789e-02 1.08815782e-01 -1.08426034e-01
7.40110457e-01 1.33454904e-01 2.18347430e-01 -2.11317629e-01
-1.77830055e-01 1.30448267e-01 -7.61241376e-01 -2.87668437e-01
1.14935386e+00 -6.44486472e-02 -1.85363218e-02 3.19759309e-01
1.40234542e+00 -1.19978059e-02 -5.91881275e-01 -7.50114620e-01
4.93523151e-01 -1.75181046e-01 2.62179941e-01 -1.03910291e+00
-6.12352073e-01 9.63869274e-01 2.18321383e-01 -1.42359495e-01
9.79622960e-01 1.06183939e-01 1.02382004e+00 7.80343831e-01
2.66843945e-01 -9.40708101e-01 2.50528157e-01 9.01768506e-01
8.55229855e-01 -1.37801433e+00 -1.35830790e-01 -7.64989182e-02
-6.67305171e-01 9.41414714e-01 6.57568336e-01 -3.41521204e-02
4.53523666e-01 3.75323296e-01 1.51500136e-01 4.25390452e-02
-1.07608390e+00 -7.19133988e-02 3.81848633e-01 2.09735185e-01
1.03045571e+00 -7.53302127e-02 -4.98629093e-01 6.06998324e-01
-3.11641514e-01 2.39954367e-02 1.57237977e-01 9.80293095e-01
-4.84482288e-01 -1.47588444e+00 4.43633310e-02 6.21289551e-01
-3.82290810e-01 -8.05232286e-01 -5.81764400e-01 6.05489671e-01
-4.60871786e-01 7.80303895e-01 1.48326233e-01 -1.76754788e-01
9.77105647e-02 3.30943167e-01 2.27735847e-01 -1.00864851e+00
-4.63384718e-01 9.75881666e-02 1.83796570e-01 -3.09866399e-01
8.73315334e-02 -3.11210841e-01 -1.18614984e+00 -2.60174513e-01
-2.09383219e-01 3.12065512e-01 3.84036362e-01 7.78787374e-01
7.24381685e-01 1.69939712e-01 4.07097518e-01 -6.17723942e-01
-6.90385759e-01 -1.20442486e+00 -1.92557573e-01 3.34881991e-01
2.37505045e-02 -3.24140042e-01 -1.56009316e-01 7.26387575e-02]
|
[11.190160751342773, 8.113241195678711]
|
447b1a6a-c3ec-49b9-b939-d472e2ace985
|
persistence-curves-a-canonical-framework-for
|
1904.07768
| null |
https://arxiv.org/abs/1904.07768v4
|
https://arxiv.org/pdf/1904.07768v4.pdf
|
Persistence Curves: A canonical framework for summarizing persistence diagrams
|
Persistence diagrams are one of the main tools in the field of Topological Data Analysis (TDA). They contain fruitful information about the shape of data. The use of machine learning algorithms on the space of persistence diagrams proves to be challenging as the space lacks an inner product. For that reason, transforming these diagrams in a way that is compatible with machine learning is an important topic currently researched in TDA. In this paper, our main contribution consists of three components. First, we develop a general and unifying framework of vectorizing diagrams that we call the \textit{Persistence Curves} (PCs), and show that several well-known summaries, such as Persistence Landscapes, fall under the PC framework. Second, we propose several new summaries based on PC framework and provide a theoretical foundation for their stability analysis. Finally, we apply proposed PCs to two applications---texture classification and determining the parameters of a discrete dynamical system; their performances are competitive with other TDA methods.
|
['Yu-Min Chung', 'Austin Lawson']
|
2019-04-16
| null | null | null | null |
['texture-classification']
|
['computer-vision']
|
[ 1.57654971e-01 -5.02188385e-01 -4.39341873e-01 1.16883136e-01
-1.06480472e-01 -7.03408122e-01 9.15191114e-01 5.71036875e-01
8.67307484e-02 6.14313126e-01 -1.56284362e-01 -6.23290181e-01
-6.42978311e-01 -7.74499774e-01 -6.26935124e-01 -1.12520027e+00
-4.23538506e-01 2.91989625e-01 4.62006599e-01 -6.00404382e-01
4.41988766e-01 6.72749579e-01 -1.87686062e+00 7.41847605e-02
8.92066538e-01 9.07644153e-01 -1.51199654e-01 6.22520626e-01
-2.00857922e-01 3.48014265e-01 -1.58447415e-01 1.31400628e-02
-5.67215867e-02 -2.52374977e-01 -7.39211500e-01 -1.10939600e-01
2.61038035e-01 3.63581866e-01 -4.22727644e-01 8.98680806e-01
-4.84400429e-02 -5.71038760e-02 1.09700990e+00 -1.55297089e+00
-2.47257292e-01 3.91597390e-01 -3.84771615e-01 2.12762460e-01
-1.14924861e-02 -3.24912190e-01 1.03820348e+00 -6.00613773e-01
7.96814740e-01 1.09000242e+00 6.72011971e-01 8.03080723e-02
-1.63939190e+00 -1.04679063e-01 -8.91316608e-02 3.33257884e-01
-9.40988183e-01 -1.89979762e-01 1.16488791e+00 -1.03786838e+00
2.97323972e-01 6.92584872e-01 9.70252931e-01 8.63905728e-01
6.96683884e-01 6.90828383e-01 1.68542898e+00 -6.38438761e-01
2.93761104e-01 1.19947232e-01 8.82436037e-01 6.18222177e-01
5.14507055e-01 -4.96383896e-03 -6.25300854e-02 -3.53537351e-01
6.29078209e-01 -3.75633806e-01 -3.24334241e-02 -1.23599303e+00
-1.16125882e+00 8.65429699e-01 1.50528565e-01 5.18685102e-01
1.48704380e-01 -5.10487966e-02 5.42411089e-01 5.68777084e-01
4.01872486e-01 3.56945813e-01 -3.80481333e-02 4.30036783e-02
-6.14867270e-01 5.04716635e-01 8.79966080e-01 5.11332929e-01
7.40461648e-01 -2.57462233e-01 9.54362378e-02 6.75100684e-01
7.30108768e-02 6.35327756e-01 1.82507008e-01 -6.72538698e-01
9.16207209e-02 7.30005443e-01 1.34293362e-01 -1.52903497e+00
-5.91153383e-01 -3.90540183e-01 -1.13110828e+00 1.22115292e-01
5.81071854e-01 4.46948916e-01 -3.59042078e-01 1.56536770e+00
2.73223788e-01 -1.70189857e-01 -6.03368431e-02 2.60706753e-01
3.73054266e-01 7.05742180e-01 -3.60440999e-01 -5.69035292e-01
1.01365697e+00 -6.60892963e-01 -7.51865864e-01 4.48465973e-01
5.83606124e-01 -4.54773784e-01 9.98587012e-01 3.64916861e-01
-5.14899492e-01 -2.31407225e-01 -1.17038727e+00 3.39344591e-01
-6.12417459e-01 3.86261344e-01 7.53270268e-01 3.86122048e-01
-1.01153398e+00 8.52251768e-01 -9.77726281e-01 -7.79509902e-01
-3.28423157e-02 2.50200629e-01 -4.09878671e-01 5.56250334e-01
-8.00616682e-01 7.09200323e-01 3.01820803e-02 -7.30327144e-02
-3.32997084e-01 -3.33354652e-01 -6.42330706e-01 -1.20139331e-01
3.05891186e-01 -2.11668238e-01 7.35651731e-01 -4.64647353e-01
-1.14513946e+00 8.27471554e-01 -3.76343355e-02 -4.12291169e-01
6.59818709e-01 8.29995796e-02 -2.58041471e-01 -1.88089952e-01
1.69975590e-02 1.19012743e-01 9.82233763e-01 -1.06873322e+00
-2.01272309e-01 -4.14314151e-01 -1.58994168e-01 -3.98419112e-01
-5.11326969e-01 -4.55465436e-01 -1.88406613e-02 -7.11573243e-01
1.95455983e-01 -1.10914361e+00 1.82416048e-02 -1.22750327e-01
-4.79153514e-01 -6.15812659e-01 1.00080526e+00 -3.16594809e-01
1.75789475e+00 -1.97799957e+00 6.91797316e-01 5.48990965e-01
4.53813404e-01 3.34829062e-01 1.02921091e-01 9.01042700e-01
-1.00864142e-01 7.04715354e-03 -4.62119907e-01 2.49216795e-01
5.73146492e-02 -2.94686630e-02 -6.04803979e-01 7.16004193e-01
3.00482009e-02 5.14136255e-01 -5.79647601e-01 -5.08978367e-01
3.83210570e-01 -3.31648216e-02 -3.56888801e-01 -2.74716854e-01
-3.43964726e-01 4.81731743e-01 -3.96311849e-01 4.65659946e-01
7.13767052e-01 -2.24532083e-01 2.33903125e-01 -1.83815151e-01
-5.43679416e-01 -2.29312345e-01 -1.17751825e+00 1.09923410e+00
-4.17266600e-02 9.16502833e-01 -3.33220623e-02 -1.54012787e+00
1.15357244e+00 -4.19305056e-01 8.69402468e-01 -2.77717769e-01
2.07198918e-01 2.83086240e-01 -2.57839747e-02 -6.20135069e-01
2.59838164e-01 3.02615941e-01 -2.20358193e-01 5.97980082e-01
-1.79300129e-01 1.47206783e-01 5.18703163e-01 -1.05784163e-02
1.05577612e+00 -1.56338319e-01 3.30793709e-01 -8.23591352e-01
9.33950067e-01 6.67683557e-02 2.93944627e-01 6.99104667e-01
-2.24330977e-01 -2.13358566e-01 1.22282851e+00 -6.30646706e-01
-1.31142056e+00 -9.55283642e-01 -6.46923125e-01 5.88287413e-01
2.42697254e-01 -5.67020655e-01 -6.37098432e-01 -3.43811035e-01
2.06823066e-01 -1.36934385e-01 -8.50329518e-01 3.96662764e-02
-5.71185231e-01 -8.55305731e-01 3.34471822e-01 1.61960423e-01
3.78275454e-01 -7.18502462e-01 -3.55083644e-01 -1.53096216e-02
-4.24349308e-02 -7.63107598e-01 3.33754420e-02 4.70843259e-03
-1.01041508e+00 -1.40736330e+00 -5.60153842e-01 -7.16829836e-01
2.89106369e-01 2.32606366e-01 6.40899241e-01 2.97523402e-02
-4.06176418e-01 2.50083447e-01 -3.64586294e-01 -9.90841258e-03
-6.85787797e-01 3.25624347e-01 2.54298061e-01 2.16071874e-01
-2.28323769e-02 -7.02874601e-01 -2.24072620e-01 4.64658409e-01
-8.38211417e-01 5.86148314e-02 5.93908668e-01 8.38527024e-01
5.06053030e-01 2.31645286e-01 6.03873849e-01 -6.92302585e-01
7.52983749e-01 -3.41689914e-01 -8.83158028e-01 4.77945387e-01
-7.36490607e-01 4.71719414e-01 4.65364873e-01 -3.43063205e-01
-4.55410898e-01 -1.18351197e-02 1.77504420e-01 -1.34088472e-01
1.28066674e-01 4.11596954e-01 -1.16114154e-01 -3.26970160e-01
4.82023001e-01 3.65965694e-01 4.36604112e-01 -5.35069644e-01
2.93525130e-01 6.02654517e-01 3.11296612e-01 -6.56666875e-01
6.39764667e-01 5.27268052e-01 6.33829415e-01 -1.34074366e+00
-2.58492768e-01 -5.32747090e-01 -8.35189760e-01 -2.72494763e-01
3.51528019e-01 -7.07047433e-02 -7.68199742e-01 7.69311368e-01
-9.41169620e-01 -2.33258113e-01 6.65764362e-02 2.54581153e-01
-1.09693027e+00 5.04431248e-01 -3.35659117e-01 -8.48845124e-01
-9.27486643e-03 -1.14112782e+00 7.25016057e-01 -1.49541363e-01
-7.80461356e-02 -1.15840006e+00 8.08408320e-01 -4.86788712e-02
2.62724727e-01 5.79138696e-01 1.52986491e+00 -4.19638157e-01
-4.00164276e-01 -1.01167537e-01 -1.40240639e-01 1.93829790e-01
1.11672282e-01 4.45742309e-01 -4.78442311e-01 -3.24888796e-01
-1.67156503e-01 2.69417405e-01 1.08079219e+00 5.13371289e-01
9.59534943e-01 -2.15176448e-01 -7.80399740e-01 4.23215538e-01
1.37913716e+00 7.66893402e-02 3.32714379e-01 4.18934435e-01
3.38275164e-01 8.00775409e-01 6.29450798e-01 1.65793583e-01
1.47468552e-01 9.64605451e-01 1.14254110e-01 1.96333364e-01
-6.65013120e-02 -1.38256088e-01 1.75227940e-01 1.08777106e+00
-9.55868140e-02 1.84912235e-01 -1.02488828e+00 3.34802985e-01
-2.25225282e+00 -9.76415575e-01 -5.13320684e-01 2.20758581e+00
4.94859904e-01 -1.66759789e-02 6.69747531e-01 6.77009881e-01
9.00116861e-01 1.58309951e-01 -3.28813434e-01 -4.58822161e-01
-3.71924639e-01 -1.99412294e-02 3.07649076e-01 6.44342721e-01
-1.53085649e+00 6.58114791e-01 6.29774380e+00 8.10506284e-01
-1.29175270e+00 -2.33797505e-01 8.69926959e-02 6.60726249e-01
-2.15191424e-01 1.19160056e-01 -5.66948056e-01 5.11216104e-01
6.21574581e-01 -1.64575189e-01 2.36406982e-01 6.16137326e-01
2.32476100e-01 -1.69830203e-01 -9.24521744e-01 9.62280214e-01
-9.64026228e-02 -1.51503813e+00 6.67527318e-02 3.67585540e-01
5.66281319e-01 -4.38243836e-01 2.79911876e-01 6.13802020e-03
-1.51992485e-01 -7.70436168e-01 3.64494085e-01 7.49923348e-01
5.89123785e-01 -5.23908854e-01 5.59823930e-01 2.61641085e-01
-1.23402464e+00 -2.36168519e-01 -3.74244988e-01 -1.19597144e-01
-2.08566546e-01 7.72228479e-01 -3.56316715e-01 6.09344721e-01
4.19039309e-01 1.00322342e+00 -9.63585496e-01 1.37765884e+00
3.65086615e-01 4.47947145e-01 -2.14842856e-01 -3.79883528e-01
-6.01239912e-02 -4.78522062e-01 9.36408520e-01 9.82125580e-01
2.86136657e-01 -2.28330761e-01 -1.32835926e-02 6.52015567e-01
3.29391330e-01 4.61512953e-01 -1.16429317e+00 -2.96425909e-01
2.65597045e-01 1.09972143e+00 -1.06774831e+00 1.53979789e-02
3.85183357e-02 1.97600290e-01 2.62725592e-01 1.31492484e-02
-4.75160450e-01 -5.52201748e-01 3.36900532e-01 3.86935502e-01
9.50759351e-02 -4.27570760e-01 -3.38217497e-01 -1.31044102e+00
1.43404931e-01 -6.60090029e-01 2.17991352e-01 -2.50639439e-01
-1.16737676e+00 4.58389252e-01 3.63717169e-01 -1.29790390e+00
-1.06431721e-02 -9.05262470e-01 -5.51227033e-01 1.14145137e-01
-1.03942263e+00 -1.17556274e+00 -1.53933927e-01 4.43489432e-01
1.92015722e-01 -1.81691974e-01 6.67214394e-01 1.48165897e-01
-5.95955372e-01 3.03257763e-01 8.16413164e-01 -1.56281263e-01
5.46767175e-01 -1.22292674e+00 9.38191637e-02 5.21708071e-01
5.58095947e-02 4.53471184e-01 9.55760121e-01 -6.49696052e-01
-1.63538563e+00 -7.43662238e-01 7.78142393e-01 -5.41159987e-01
1.13590240e+00 -5.12465000e-01 -8.86605144e-01 2.97393888e-01
-1.97355568e-01 -2.63518095e-01 3.54726583e-01 3.49689126e-01
-1.68942511e-01 -4.34384763e-01 -6.24585748e-01 5.19489944e-01
8.92241180e-01 -3.83342892e-01 -4.82857943e-01 2.53103107e-01
2.88258106e-01 2.28391439e-01 -7.29706883e-01 5.48621595e-01
9.08386171e-01 -9.85021412e-01 8.67493153e-01 -5.69215834e-01
8.44164267e-02 -2.93485314e-01 -1.47892445e-01 -1.18512762e+00
-1.57926306e-01 -6.84866130e-01 -1.72248080e-01 1.08067155e+00
1.71368420e-02 -7.61102855e-01 4.21079308e-01 -2.79799610e-01
2.23249540e-01 -8.49744618e-01 -8.94669652e-01 -1.03165948e+00
4.20168459e-01 -2.05715507e-01 3.61909419e-01 8.56696367e-01
4.33598429e-01 3.58816475e-01 -2.79719591e-01 -3.11287701e-01
8.46709311e-01 4.14538711e-01 8.76762986e-01 -1.89528704e+00
5.69002479e-02 -8.80619586e-01 -7.97176182e-01 -6.21508956e-01
3.56989413e-01 -1.01625156e+00 -2.78917015e-01 -1.06797707e+00
1.82344273e-01 -8.97958755e-01 -2.19246864e-01 1.96063779e-02
4.05791402e-01 -9.53926966e-02 -8.14226922e-03 5.80835998e-01
-3.64212602e-01 5.03401756e-01 1.04568660e+00 -2.79892921e-01
-3.89130205e-01 3.58976930e-01 -1.99218139e-01 6.37511909e-01
8.20603728e-01 -8.64885002e-02 -3.86019379e-01 2.95749575e-01
2.65254200e-01 1.95003152e-01 5.00324965e-01 -9.87018347e-01
1.57054693e-01 -2.55795926e-01 -1.94221109e-01 -9.16685104e-01
7.78499842e-02 -4.81873602e-01 1.27131835e-01 5.63914001e-01
-3.59275937e-01 1.78955093e-01 1.00075372e-01 8.07275891e-01
-7.91014507e-02 2.08705198e-02 8.23046327e-01 4.12866443e-01
-4.31941122e-01 8.94666538e-02 -5.30831337e-01 -1.95173010e-01
1.12853980e+00 -1.63881294e-02 -6.63351476e-01 -2.04085588e-01
-9.54146504e-01 -7.87619594e-03 6.90179050e-01 3.75938565e-01
3.48431855e-01 -1.38657796e+00 -3.62220585e-01 3.06043625e-01
2.14506447e-01 -6.77497983e-01 2.49349952e-01 1.30663157e+00
-6.53583288e-01 9.76733506e-01 -5.42264342e-01 -1.04686320e+00
-1.39001799e+00 7.86389530e-01 9.17425677e-02 -2.52210647e-01
-7.40514755e-01 2.26752032e-02 1.24603160e-01 -2.81892926e-01
1.54273063e-01 -3.22039574e-01 -4.76566792e-01 2.65479296e-01
2.27401108e-01 6.93989813e-01 9.50021595e-02 -6.45877719e-01
-3.98563802e-01 9.12880063e-01 2.43288204e-01 -5.11674769e-02
1.27270830e+00 6.50792755e-03 -6.34427428e-01 7.93777287e-01
1.07324171e+00 -1.01248965e-01 -7.88814008e-01 -8.05357248e-02
3.49628001e-01 -2.29696035e-01 -3.77976298e-01 -1.95673481e-01
-3.67761433e-01 1.06903541e+00 6.89367771e-01 8.59620690e-01
8.07403803e-01 1.78699881e-01 2.14623526e-01 6.09895706e-01
2.52036631e-01 -9.28562999e-01 9.62519422e-02 4.96904850e-01
1.00524199e+00 -8.41847956e-01 -2.35418975e-01 -5.19815385e-01
-1.07855253e-01 1.31448317e+00 1.73550472e-01 -2.66980112e-01
9.64900553e-01 -1.23861665e-02 -3.62596661e-01 -2.12266833e-01
-7.53603578e-01 -2.16820896e-01 2.19595805e-01 4.85040933e-01
6.05256744e-02 3.43395382e-01 -8.54169250e-01 2.76481211e-01
-2.97308922e-01 -2.09739685e-01 7.02983975e-01 8.55201900e-01
-7.79121578e-01 -1.33580291e+00 -3.38811010e-01 4.86861646e-01
1.28205284e-01 4.07204479e-01 -5.20795643e-01 8.98904145e-01
-7.52899796e-02 6.77311182e-01 -7.50279427e-02 -5.58924615e-01
8.68464354e-03 9.07238871e-02 6.41224921e-01 -1.31401822e-01
3.50467592e-01 7.71759264e-03 -3.68571617e-02 -2.14674369e-01
-4.61578131e-01 -9.19900119e-01 -5.25888801e-01 -5.98298550e-01
-3.37855428e-01 4.88307238e-01 6.69943035e-01 8.97079110e-01
1.40017673e-01 1.67782858e-01 9.30371284e-01 -4.47458744e-01
-3.61215949e-01 -7.64072001e-01 -7.07452297e-01 -4.75612283e-02
6.75842285e-01 -1.25267434e+00 -4.25282121e-01 -7.57990256e-02]
|
[7.472250461578369, 4.182532787322998]
|
2b159917-cdae-4d5b-b315-03d430d902ac
|
iterative-deep-graph-learning-for-graph
| null | null |
https://openreview.net/forum?id=Bkl2UlrFwr
|
https://openreview.net/pdf?id=Bkl2UlrFwr
|
Iterative Deep Graph Learning for Graph Neural Networks
|
In this paper, we propose an end-to-end graph learning framework, namely Iterative Deep Graph Learning (IDGL), for jointly learning graph structure and graph embedding simultaneously. We first cast graph structure learning problem as similarity metric learning problem and leverage an adapted graph regularization for controlling smoothness, connectivity and sparsity of the generated graph. We further propose a novel iterative method for searching for hidden graph structure that augments the initial graph structure. Our iterative method dynamically stops when learning graph structure approaches close enough to the ground truth graph. Our extensive experiments demonstrate that the proposed IDGL model can consistently outperform or match state-of-the-art baselines in terms of both classification accuracy and computational time. The proposed approach can cope with both transductive training and inductive training.
|
['Mohammed J. Zaki', 'Lingfei Wu', 'Yu Chen']
|
2019-09-25
| null | null | null | null |
['graph-structure-learning']
|
['graphs']
|
[ 6.39925450e-02 4.52426344e-01 -4.64297593e-01 -3.66371036e-01
-6.78766787e-01 -5.41992068e-01 6.29630864e-01 3.96152765e-01
-1.58010483e-01 4.36382085e-01 3.07588708e-02 -4.59030986e-01
-7.02076852e-02 -9.13795710e-01 -7.98139930e-01 -4.00466383e-01
-3.27550441e-01 4.34345752e-01 -1.05390977e-02 1.82113916e-01
4.75100335e-03 3.06533784e-01 -9.27731633e-01 -1.00961179e-01
8.95803034e-01 7.32710004e-01 4.86742072e-02 6.82179809e-01
-7.26919025e-02 1.05950522e+00 9.68449842e-03 -3.86461347e-01
2.68268794e-01 -4.35254246e-01 -8.66206408e-01 4.77423698e-01
6.11927569e-01 -7.48906732e-02 -6.59282386e-01 1.06974649e+00
3.24635029e-01 2.30294630e-01 6.74911201e-01 -1.33856940e+00
-8.27144086e-01 6.05317473e-01 -6.41980886e-01 -6.68842858e-03
3.34467918e-01 -3.49794403e-02 1.55888963e+00 -9.88267481e-01
6.31365538e-01 1.18812311e+00 6.90661311e-01 3.19825023e-01
-1.39222491e+00 -3.95166129e-01 5.07598579e-01 -5.26253916e-02
-1.40393102e+00 -1.59882113e-01 1.23172092e+00 -5.60853064e-01
6.65929377e-01 -1.79611921e-01 6.01751983e-01 8.22948694e-01
-5.02071269e-02 6.67672634e-01 8.30690265e-01 -5.89574099e-01
3.23214442e-01 -2.39095718e-01 3.76097292e-01 1.58786702e+00
4.44787890e-01 4.48490344e-02 -3.54553759e-01 -2.01401487e-01
7.17686951e-01 -1.49655029e-01 -1.21305130e-01 -8.33186388e-01
-1.03116798e+00 1.06572008e+00 9.01312947e-01 1.28980994e-01
-3.03866476e-01 5.49744368e-01 2.78225869e-01 4.29021239e-01
5.81754923e-01 2.55854547e-01 -1.70069292e-01 3.64316970e-01
-5.30352294e-01 -5.92676848e-02 7.95627594e-01 1.07498646e+00
1.23061156e+00 4.01171111e-02 -1.27684712e-01 5.97122669e-01
7.91202545e-01 1.47869974e-01 -3.75879854e-02 -5.37855983e-01
6.42024875e-01 1.08325410e+00 -4.34915602e-01 -1.21364510e+00
-2.52750784e-01 -4.45424080e-01 -8.38013291e-01 -1.97783753e-01
2.55251825e-01 -9.48892385e-02 -1.07994246e+00 1.74006093e+00
5.07513285e-01 7.65718520e-01 -4.88038808e-02 7.19737053e-01
1.07491660e+00 3.80001724e-01 6.39280910e-03 1.16267689e-01
9.55095828e-01 -1.19371772e+00 -5.53110898e-01 -4.02711868e-01
1.18682325e+00 -2.60513991e-01 1.32159829e+00 -4.52057421e-02
-8.48864257e-01 -3.34681004e-01 -1.06385207e+00 -1.68680370e-01
-8.16949755e-02 9.68299732e-02 7.90173471e-01 3.27938020e-01
-1.16644442e+00 6.50386214e-01 -9.20131743e-01 -3.12036455e-01
4.96610045e-01 2.81023026e-01 -5.29308677e-01 -4.39868808e-01
-7.79542387e-01 2.24234715e-01 5.34586310e-01 8.59935069e-04
-8.67687106e-01 -5.77761531e-01 -1.34931171e+00 -2.89447233e-02
5.81439912e-01 -8.12610745e-01 7.26639032e-01 -6.11901820e-01
-1.33073878e+00 1.11299574e+00 3.12818624e-02 -5.06271362e-01
9.40770656e-02 -3.11932601e-02 -1.17674820e-01 1.00549906e-01
-2.70914380e-02 2.26867378e-01 7.85974503e-01 -1.24821520e+00
-7.28539973e-02 -4.20497477e-01 8.10454860e-02 1.68199897e-01
-3.60369593e-01 -3.38886440e-01 -6.98535562e-01 -6.50966108e-01
2.73577929e-01 -1.01213324e+00 -3.96218598e-01 5.28022693e-03
-5.89739203e-01 -2.98783123e-01 7.39046931e-01 -4.26471233e-01
1.30172980e+00 -1.90280962e+00 3.46239477e-01 4.85062182e-01
8.10510993e-01 1.30784169e-01 -4.32636470e-01 6.21898830e-01
-3.57101448e-02 8.98471847e-02 -2.58564502e-01 -5.10983348e-01
8.90477449e-02 3.40552241e-01 -1.20038213e-03 6.96845233e-01
1.40029520e-01 1.27914345e+00 -1.21375775e+00 -4.46952671e-01
1.73502624e-01 4.13603753e-01 -7.08945453e-01 5.54945409e-01
-2.77608901e-01 4.37956095e-01 -4.79785562e-01 5.45740545e-01
6.33830726e-01 -7.85328627e-01 6.22166097e-01 -2.69520491e-01
4.65207934e-01 3.55593532e-01 -1.18276417e+00 1.99216032e+00
-3.61068875e-01 3.68745774e-01 5.63256778e-02 -1.27925432e+00
1.11427879e+00 -7.84045905e-02 2.58302689e-01 -3.72242481e-01
1.29479736e-01 4.29612175e-02 -7.38586932e-02 -1.49023309e-01
-3.76889370e-02 1.47060558e-01 8.56153574e-03 5.21137774e-01
3.64449024e-01 5.49090840e-02 6.69062734e-02 5.79481602e-01
1.32062733e+00 7.20903948e-02 3.64763916e-01 -4.21663344e-01
5.50224364e-01 -4.29815680e-01 3.89854491e-01 6.27247691e-01
5.61427288e-02 1.95702702e-01 6.47291481e-01 -3.95652384e-01
-7.08992898e-01 -1.15363944e+00 4.64496434e-01 1.07946444e+00
2.09981039e-01 -8.03539753e-01 -6.76553369e-01 -1.19319868e+00
4.18876410e-02 2.94814557e-01 -5.71925700e-01 -3.82976592e-01
-5.63505054e-01 -4.07901973e-01 2.20177189e-01 4.10027057e-01
3.81618142e-01 -6.92758679e-01 2.87446558e-01 8.12788531e-02
1.82577953e-01 -1.42545891e+00 -7.63814986e-01 -1.50021121e-01
-1.05616188e+00 -1.30524707e+00 -4.65898402e-02 -1.26022661e+00
1.15323937e+00 3.42754930e-01 1.30739152e+00 5.97950995e-01
-9.17025059e-02 5.30851722e-01 -3.70042562e-01 1.23759672e-01
-3.92241746e-01 3.28662634e-01 -3.96179497e-01 2.65648276e-01
-6.93917423e-02 -6.80342913e-01 -5.60357153e-01 5.05826809e-02
-7.48702705e-01 1.06587611e-01 3.54799986e-01 8.17106605e-01
7.56444037e-01 -1.94304153e-01 4.30251032e-01 -1.30012441e+00
8.09012830e-01 -4.63540494e-01 -6.83380306e-01 3.98630559e-01
-1.00232446e+00 5.33110738e-01 5.70283055e-01 -4.89121042e-02
-3.70766759e-01 1.00154482e-01 1.14806756e-01 -7.53237009e-01
1.31712392e-01 9.97033596e-01 -2.10897192e-01 -4.40587491e-01
5.11140466e-01 -7.55153224e-02 6.93117678e-02 -4.84177679e-01
7.93755233e-01 2.01038972e-01 4.82545465e-01 -5.87960243e-01
1.12537956e+00 3.45159531e-01 3.56133014e-01 -5.00121593e-01
-1.02682316e+00 -5.75431943e-01 -8.37842405e-01 -3.44348103e-02
6.41116560e-01 -1.04223418e+00 -4.63396430e-01 3.23610634e-01
-9.16154027e-01 -5.36590397e-01 -3.79730351e-02 4.68067527e-01
-3.59374940e-01 7.06868827e-01 -6.62673235e-01 -4.39522713e-01
-5.63146651e-01 -7.76374102e-01 1.24536741e+00 -2.76276976e-01
2.42594019e-01 -1.81292486e+00 3.90292615e-01 1.81765795e-01
-5.67029417e-02 6.63154483e-01 8.95809054e-01 -5.94866872e-01
-7.76140153e-01 -3.82635504e-01 -3.95281523e-01 2.30325803e-01
3.93414348e-01 -3.79041970e-01 -6.81731522e-01 -7.12328911e-01
-4.64473009e-01 -4.54146951e-01 1.01940084e+00 5.00729419e-02
9.83343124e-01 -6.06066883e-01 -3.46951634e-01 9.81699109e-01
1.59008133e+00 -4.29030538e-01 4.52113897e-01 -5.38329333e-02
1.59551549e+00 3.94728988e-01 4.40709770e-01 2.99209237e-01
7.26420820e-01 5.32875955e-01 4.61099744e-01 -3.93698424e-01
-2.38621444e-01 -7.51674712e-01 2.74590462e-01 1.07788837e+00
2.07816631e-01 -3.66077900e-01 -9.46342826e-01 4.45514321e-01
-2.14452362e+00 -5.39214373e-01 -1.66976824e-01 2.22994828e+00
4.44330633e-01 8.52151439e-02 1.46584436e-01 2.79151322e-03
7.97675550e-01 3.48100483e-01 -3.71764839e-01 -1.62739366e-01
2.22836792e-01 2.50597745e-01 4.83612746e-01 9.90199149e-01
-1.01608133e+00 1.39813626e+00 6.09587383e+00 4.43097919e-01
-9.75136042e-01 -1.09528176e-01 4.66755420e-01 3.54774624e-01
-5.30560374e-01 3.03062111e-01 -3.67245376e-01 1.79853648e-01
7.21874237e-01 -2.01390982e-01 7.25800633e-01 7.61874020e-01
-2.97367461e-02 5.75413465e-01 -1.28143919e+00 1.12346590e+00
1.05427690e-01 -1.51866436e+00 1.97738737e-01 2.26936087e-01
7.68832803e-01 1.31215528e-01 -2.12532640e-01 3.66580456e-01
7.18786716e-01 -9.94628191e-01 2.15791076e-01 1.96428359e-01
7.71083534e-01 -6.22145832e-01 3.88143957e-01 2.01809809e-01
-1.69519985e+00 2.29305312e-01 -1.74187243e-01 -7.63641372e-02
-5.74589409e-02 6.26143634e-01 -9.91875529e-01 7.87024438e-01
1.12894587e-01 1.11962867e+00 -7.48091102e-01 6.54886365e-01
-7.16396511e-01 8.76679480e-01 -1.21805139e-01 1.80920884e-01
4.58751321e-01 -5.56330562e-01 4.63629633e-01 1.01707208e+00
-1.67838763e-03 -9.80171859e-02 8.63010347e-01 9.17883694e-01
-5.08244157e-01 1.99121237e-01 -9.77502465e-01 -3.78101528e-01
4.84867036e-01 1.35278535e+00 -7.84607470e-01 -2.78581321e-01
-5.84447443e-01 1.00919175e+00 1.05393243e+00 4.73251134e-01
-7.43178070e-01 -3.50167215e-01 3.91345680e-01 1.16697446e-01
5.37467301e-01 -4.19830918e-01 9.14823189e-02 -1.40932691e+00
1.41784757e-01 -5.95085561e-01 7.72451699e-01 -2.81979233e-01
-1.36304367e+00 5.20932198e-01 -1.48230851e-01 -8.37047756e-01
-2.01267555e-01 -6.10647082e-01 -8.86248767e-01 4.40928608e-01
-1.57059729e+00 -1.45929766e+00 -5.95380545e-01 7.97548175e-01
-4.63317223e-02 -4.54657264e-02 6.22980535e-01 9.96561199e-02
-5.26320040e-01 7.53122091e-01 -1.37471810e-01 3.80076349e-01
3.81469667e-01 -1.54703844e+00 7.76339352e-01 9.39134359e-01
5.77621996e-01 4.47821945e-01 2.08233148e-01 -6.98818207e-01
-1.80876756e+00 -1.58640015e+00 6.92621171e-01 -1.97601706e-01
9.18046117e-01 -8.43646824e-01 -1.02031291e+00 8.67995918e-01
-9.68626663e-02 5.16229928e-01 5.89101970e-01 2.61930257e-01
-7.13978529e-01 -2.37354115e-02 -1.02682436e+00 5.06581187e-01
1.41352832e+00 -7.26064205e-01 -2.18985766e-01 5.64211905e-01
9.86290634e-01 -4.16023105e-01 -9.54723120e-01 3.47798020e-01
6.01194091e-02 -5.57749867e-01 7.96959639e-01 -7.67736018e-01
-4.00282107e-02 -2.54821420e-01 -6.31560460e-02 -1.34577703e+00
-5.40040553e-01 -1.10872388e+00 -4.89533067e-01 1.07307124e+00
3.87322515e-01 -6.76882863e-01 1.04335761e+00 3.01354676e-01
-6.61022216e-02 -9.44569826e-01 -6.04809821e-01 -8.13548625e-01
-1.86106384e-01 -3.10142815e-01 4.64418471e-01 1.24626684e+00
-7.42895007e-02 8.55818212e-01 -2.80713260e-01 3.23973864e-01
1.03903413e+00 3.60320807e-01 9.80611324e-01 -1.37421536e+00
-5.69840848e-01 -1.27857313e-01 -8.86922717e-01 -9.64841485e-01
6.01361990e-01 -1.64033592e+00 -2.65545398e-01 -1.91834891e+00
1.54710934e-01 -2.68342495e-01 -4.50805217e-01 5.28190255e-01
-4.12510931e-01 7.77726471e-02 7.64505565e-02 -5.84931895e-02
-7.94755816e-01 6.55756593e-01 1.28131592e+00 -9.78794470e-02
-3.47595990e-01 -2.44111344e-01 -8.40434194e-01 5.03934145e-01
6.27524614e-01 -4.27873075e-01 -7.61703312e-01 -5.18597662e-01
2.12241784e-01 -7.76339648e-03 4.35906202e-01 -6.64336741e-01
1.16880082e-01 -3.91497910e-02 -2.87195832e-01 -1.95527032e-01
-1.36653647e-01 -5.65888524e-01 -1.93026900e-01 3.33701491e-01
-5.24335802e-01 -3.96837108e-02 1.73225682e-02 9.99345541e-01
-1.05734263e-02 7.28786513e-02 7.72903740e-01 1.44141883e-01
-5.30635536e-01 9.26169813e-01 2.09985346e-01 5.00539899e-01
8.64240885e-01 -2.40826458e-02 -1.66336432e-01 -3.60689700e-01
-8.05948973e-01 4.00612354e-01 4.54785943e-01 3.06563735e-01
7.83977449e-01 -1.59470248e+00 -7.85137773e-01 1.02974862e-01
3.70741606e-01 1.22741967e-01 -3.10191721e-01 7.25317180e-01
-4.47949916e-01 -2.24586185e-02 3.93497944e-01 -5.15289426e-01
-1.21806049e+00 7.21699774e-01 2.47782007e-01 -6.60696626e-01
-9.21677887e-01 8.82639945e-01 3.63726854e-01 -7.20946848e-01
3.61092955e-01 -4.93930608e-01 1.92656800e-01 -4.41627860e-01
5.71310893e-02 2.34376207e-01 -5.24185114e-02 -4.46098864e-01
-3.25825661e-01 5.87261438e-01 -2.08726555e-01 2.40913928e-01
1.23758686e+00 -1.40828025e-02 -1.95030525e-01 3.04378152e-01
1.62791073e+00 -1.43558964e-01 -1.12149298e+00 -6.03714228e-01
2.17539012e-01 -4.62515146e-01 3.34353715e-01 -2.75448322e-01
-1.14122653e+00 6.52542293e-01 2.53993779e-01 -8.81124008e-03
8.72749984e-01 3.22661519e-01 7.16514885e-01 7.13279843e-01
2.50200003e-01 -6.71656430e-01 4.88113642e-01 3.99205327e-01
7.25899398e-01 -1.31855404e+00 2.32489064e-01 -7.95550644e-01
-3.96240085e-01 7.87121713e-01 5.12112141e-01 -6.63098752e-01
7.78211951e-01 -8.84853452e-02 -2.98686177e-01 -6.00864291e-01
-8.65017116e-01 -3.45360607e-01 7.95295179e-01 6.26320124e-01
4.38232809e-01 6.20868206e-02 -6.02743328e-02 1.89261124e-01
-3.17806890e-03 -2.44962335e-01 1.93065390e-01 7.80964494e-01
-2.72045940e-01 -1.11343777e+00 1.82417721e-01 3.73154759e-01
-5.17019592e-02 -2.54288018e-01 -8.43819499e-01 7.46116519e-01
-3.82976890e-01 8.90934527e-01 -3.14900577e-02 -5.02839446e-01
2.21715674e-01 -3.21139991e-01 7.50522137e-01 -9.03595746e-01
-2.03414083e-01 -3.89887579e-02 1.87187374e-01 -7.22988904e-01
-2.33445242e-01 -2.20636383e-01 -1.42615473e+00 -5.19928224e-02
-5.78818738e-01 1.81991279e-01 3.14826965e-01 8.02119911e-01
6.42061174e-01 3.17923218e-01 8.80540788e-01 -5.93067229e-01
-3.79970104e-01 -7.27605522e-01 -5.52780747e-01 6.00608230e-01
3.00786763e-01 -4.95361924e-01 -4.70623195e-01 -2.87835207e-02]
|
[7.142280578613281, 6.259690761566162]
|
6dc404ab-0153-48c0-a861-a7a87f52cc07
|
counting-motifs-with-graph-sampling
|
1802.07773
| null |
http://arxiv.org/abs/1802.07773v1
|
http://arxiv.org/pdf/1802.07773v1.pdf
|
Counting Motifs with Graph Sampling
|
Applied researchers often construct a network from a random sample of nodes
in order to infer properties of the parent network. Two of the most widely used
sampling schemes are subgraph sampling, where we sample each vertex
independently with probability $p$ and observe the subgraph induced by the
sampled vertices, and neighborhood sampling, where we additionally observe the
edges between the sampled vertices and their neighbors.
In this paper, we study the problem of estimating the number of motifs as
induced subgraphs under both models from a statistical perspective. We show
that: for any connected $h$ on $k$ vertices, to estimate $s=\mathsf{s}(h,G)$,
the number of copies of $h$ in the parent graph $G$ of maximum degree $d$, with
a multiplicative error of $\epsilon$, (a) For subgraph sampling, the optimal
sampling ratio $p$ is $\Theta_{k}(\max\{ (s\epsilon^2)^{-\frac{1}{k}}, \;
\frac{d^{k-1}}{s\epsilon^{2}} \})$, achieved by Horvitz-Thompson type of
estimators. (b) For neighborhood sampling, we propose a family of estimators,
encompassing and outperforming the Horvitz-Thompson estimator and achieving the
sampling ratio $O_{k}(\min\{ (\frac{d}{s\epsilon^2})^{\frac{1}{k-1}}, \;
\sqrt{\frac{d^{k-2}}{s\epsilon^2}}\})$. This is shown to be optimal for all
motifs with at most $4$ vertices and cliques of all sizes.
The matching minimax lower bounds are established using certain algebraic
properties of subgraph counts. These results quantify how much more informative
neighborhood sampling is than subgraph sampling, as empirically verified by
experiments on both synthetic and real-world data. We also address the issue of
adaptation to the unknown maximum degree, and study specific problems for
parent graphs with additional structures, e.g., trees or planar graphs.
|
['Jason M. Klusowski', 'Yihong Wu']
|
2018-02-21
| null | null | null | null |
['graph-sampling']
|
['graphs']
|
[ 4.37662393e-01 5.52928507e-01 -2.05763340e-01 8.69509671e-03
-6.01143897e-01 -6.73769355e-01 -4.69772667e-02 2.58701414e-01
-3.68972331e-01 9.61815417e-01 -7.04562485e-01 -5.24466872e-01
-6.74424350e-01 -1.29713511e+00 -9.42342877e-01 -9.10489500e-01
-8.68616700e-01 6.29305363e-01 4.36804980e-01 9.06146988e-02
1.76114991e-01 3.45478743e-01 -1.65669405e+00 -4.63049263e-01
9.78253365e-01 8.63330543e-01 4.25333679e-02 7.33841360e-01
1.86941270e-02 2.71433860e-01 -5.29737711e-01 -4.64797914e-01
2.06074417e-01 -7.09784031e-01 -7.45330215e-01 2.25937366e-01
1.44780651e-01 1.99392870e-01 -1.84966907e-01 1.48347127e+00
7.56536648e-02 -7.01410919e-02 8.23624134e-01 -1.19212747e+00
-3.87491845e-02 7.82723069e-01 -1.11653912e+00 8.85840952e-02
4.18443948e-01 -1.95021793e-01 1.03336370e+00 -3.24370146e-01
5.07758915e-01 7.36121237e-01 6.10976517e-01 9.73833650e-02
-1.52693236e+00 -9.09021735e-01 2.05141217e-01 -3.68306309e-01
-1.73217356e+00 3.08310594e-02 7.10140944e-01 -2.32000291e-01
4.82534260e-01 6.95723891e-01 5.41985810e-01 3.40402812e-01
-1.05852097e-01 3.03748280e-01 1.22856319e+00 -6.43305719e-01
3.90119702e-01 2.42111422e-02 1.91607207e-01 9.15363550e-01
8.10354948e-01 -7.00352043e-02 -9.49117690e-02 -3.64100277e-01
9.95509505e-01 -4.04809862e-02 -2.24408627e-01 -4.11402173e-02
-7.71991670e-01 7.74370730e-01 9.89181399e-02 3.33688825e-01
-2.02821270e-01 4.46389914e-01 -1.24503613e-01 2.82903045e-01
3.53111893e-01 1.52355686e-01 -5.11475265e-01 4.17059846e-02
-9.48064208e-01 1.02722928e-01 1.03752816e+00 1.43607855e+00
1.07444990e+00 -1.35799229e-01 3.07620972e-01 6.30311310e-01
1.76543340e-01 7.72278309e-01 -4.30780619e-01 -9.70849931e-01
4.85735267e-01 6.86717629e-01 1.06034786e-01 -9.11580205e-01
-3.24259609e-01 -3.32335174e-01 -1.04006386e+00 7.23072467e-03
8.31250370e-01 -2.31977761e-01 -6.06839120e-01 2.11405659e+00
2.43459284e-01 6.17493354e-02 -5.08625805e-01 3.43038589e-01
1.55119568e-01 5.43310285e-01 -1.03670694e-01 -8.81107748e-01
1.38096416e+00 -6.92788661e-02 2.64376476e-02 -6.24825843e-02
4.45333123e-01 -6.41630828e-01 8.99135649e-01 4.08054978e-01
-1.29491568e+00 -2.66388565e-01 -8.59683275e-01 7.23168612e-01
8.42917431e-03 -1.16935432e-01 4.58448410e-01 1.00226235e+00
-1.05133569e+00 7.00603306e-01 -7.28597641e-01 -1.15593985e-01
5.02923802e-02 5.43919861e-01 -2.16780275e-01 -2.13509396e-01
-8.05062234e-01 4.24828082e-02 1.70670107e-01 -1.02982074e-01
-6.03382289e-01 -3.46236944e-01 -5.20474851e-01 7.58106858e-02
6.21189237e-01 -2.10727319e-01 5.35318732e-01 -7.08391368e-01
-7.77981579e-01 7.78446496e-01 -3.13804716e-01 -1.56850010e-01
1.25789583e-01 6.59382403e-01 -2.79106170e-01 3.32297385e-01
3.56083155e-01 1.50015801e-01 4.08973932e-01 -1.23914957e+00
-4.94639784e-01 -8.79829228e-01 1.21911861e-01 -3.54097635e-01
-1.08377941e-01 -3.20248655e-03 -4.84197199e-01 -3.25753033e-01
5.97712100e-01 -1.11105335e+00 -3.72658670e-01 -4.09723699e-01
-6.72394276e-01 -1.48684099e-01 2.51791626e-01 -1.98174074e-01
1.49541402e+00 -1.99329817e+00 -2.93264375e-03 1.13177705e+00
5.55773020e-01 -8.27998761e-03 3.69887441e-01 7.06788719e-01
-8.21643136e-03 3.86254579e-01 -4.04735416e-01 1.86348855e-01
-7.32868686e-02 3.37872319e-02 2.60325700e-01 6.92308486e-01
-3.38580012e-01 7.21018165e-02 -8.43002021e-01 -2.40853965e-01
-2.03404874e-01 2.00853273e-01 -4.55428869e-01 -1.79522693e-01
-2.53497392e-01 1.17617667e-01 -8.07054877e-01 3.96779001e-01
1.02540004e+00 -7.69850850e-01 8.73972535e-01 3.00738871e-01
-1.29296184e-01 -1.28832571e-02 -1.96667051e+00 9.57192302e-01
4.02873456e-02 3.04836291e-03 4.44524139e-01 -1.35887134e+00
9.46380377e-01 1.71823755e-01 5.97560883e-01 -2.86501110e-01
1.10852361e-01 4.21123445e-01 -1.56659231e-01 -9.71955210e-02
-2.73806453e-02 -3.11839014e-01 -2.05458716e-01 6.96090400e-01
-1.75943509e-01 4.92421761e-02 3.68350565e-01 3.36399645e-01
1.63702154e+00 -3.78866315e-01 2.57549495e-01 -7.27727950e-01
3.21843088e-01 -2.84172714e-01 7.66115785e-01 8.61670971e-01
-7.04490989e-02 2.05678523e-01 1.43838930e+00 2.16711240e-04
-9.98282790e-01 -1.10384810e+00 -3.27403188e-01 9.23680961e-01
3.31906378e-01 -3.50382924e-01 -1.12034798e+00 -4.06078637e-01
-2.76312809e-02 4.87820476e-01 -7.18118548e-01 4.56430092e-02
-3.65825146e-01 -8.26166272e-01 4.83943045e-01 4.63951588e-01
4.22278404e-01 -8.65956724e-01 -5.57328224e-01 4.72238809e-02
1.11068878e-03 -7.86604345e-01 -4.07831222e-01 4.56688881e-01
-9.12421346e-01 -1.32296622e+00 -5.78845561e-01 -6.85413420e-01
1.20911527e+00 1.84132725e-01 9.55275178e-01 1.78943813e-01
-3.32652986e-01 2.59398401e-01 -2.47883007e-01 -2.21602824e-02
-1.38800979e-01 -7.94190466e-02 -2.09895931e-02 -3.09262842e-01
3.67498755e-01 -1.06915545e+00 -7.55352616e-01 4.33392406e-01
-8.73688400e-01 -3.17132890e-01 4.13329124e-01 6.10437155e-01
8.20216179e-01 3.81504685e-01 3.60529631e-01 -1.16705775e+00
1.15662128e-01 -4.44566905e-01 -1.14194250e+00 9.51971486e-02
-6.09302282e-01 1.17629334e-01 8.23663950e-01 -2.76297152e-01
-3.08218569e-01 -1.05040587e-01 9.89947841e-02 -3.60446155e-01
-1.33033663e-01 3.90866607e-01 -1.70943916e-01 9.70299914e-02
5.20327628e-01 3.77304941e-01 -3.29917520e-02 -3.98756295e-01
4.90355343e-02 2.25718781e-01 2.30478749e-01 -8.44542563e-01
6.48141503e-01 5.16585350e-01 5.57444930e-01 -9.39416528e-01
-3.16810459e-01 -7.84931108e-02 -2.47082323e-01 -4.36806083e-02
5.50512493e-01 -5.32768190e-01 -1.40256798e+00 5.85164316e-03
-6.32539809e-01 -6.43825755e-02 -1.81288198e-01 5.21725178e-01
-5.73263168e-01 5.62527061e-01 -4.26219434e-01 -1.31716406e+00
3.94662395e-02 -1.06989944e+00 7.07259417e-01 2.11562261e-01
-1.11699246e-01 -7.37320423e-01 -2.47720376e-01 7.63406083e-02
-4.80144694e-02 5.75989664e-01 1.33288825e+00 -5.84876299e-01
-7.52346754e-01 -3.84454876e-01 -3.34568471e-01 2.53152281e-01
-1.51529384e-03 6.03150390e-02 -3.64419222e-01 -4.91399080e-01
-3.89014542e-01 1.83558583e-01 6.02646589e-01 5.41329980e-01
1.18145347e+00 -5.09160280e-01 -7.39866614e-01 2.75050402e-01
1.65443408e+00 2.45906323e-01 7.59928882e-01 -2.94999152e-01
1.32721990e-01 3.64400715e-01 2.20478356e-01 6.88758433e-01
8.90902057e-02 3.08429241e-01 4.64061499e-01 1.45051390e-01
2.85885334e-01 -2.48150066e-01 -1.32516017e-02 5.27934730e-01
-3.22719485e-01 -3.12813222e-01 -7.70966709e-01 7.96388566e-01
-1.37304342e+00 -7.77715385e-01 -4.15175170e-01 2.80056214e+00
8.65824938e-01 3.99326771e-01 4.23616588e-01 3.10568869e-01
1.18909323e+00 -4.56985757e-02 -3.58424306e-01 -2.14963332e-01
6.05924055e-02 1.09603906e+00 7.50533462e-01 4.44765359e-01
-5.01455247e-01 5.79632282e-01 4.79147243e+00 1.05060196e+00
-4.42078382e-01 -2.11338848e-01 7.52564967e-01 -6.87718838e-02
-5.33649623e-01 6.56048894e-01 -7.68140733e-01 8.80055666e-01
8.19651544e-01 -1.00892425e-01 4.33620989e-01 7.59660780e-01
5.07041859e-03 -6.44880474e-01 -9.64464366e-01 4.60311502e-01
-3.23246628e-01 -1.03429365e+00 -3.16828668e-01 6.12801373e-01
8.12515974e-01 -5.17208159e-01 -7.83506185e-02 -5.59876375e-02
7.62112141e-01 -9.73404229e-01 2.44521901e-01 1.13944381e-01
1.08256829e+00 -1.03007078e+00 2.93439478e-01 5.73111892e-01
-1.62532461e+00 -5.76633262e-04 -2.50025004e-01 -8.69612321e-02
-7.50677660e-02 1.11848593e+00 -4.27769154e-01 6.49833083e-01
8.64833474e-01 -8.42925310e-02 1.71605423e-01 6.30228043e-01
-6.28167763e-02 6.52747333e-01 -7.07939446e-01 -2.24988848e-01
5.24223037e-02 -6.03846073e-01 6.02429569e-01 7.38387406e-01
3.40177596e-01 7.41107583e-01 3.09338748e-01 8.03775012e-01
-1.89433545e-01 1.19005084e-01 -5.25714219e-01 8.22083745e-03
1.00297916e+00 8.14833522e-01 -1.24325287e+00 -1.33616909e-01
-2.26517349e-01 4.30949122e-01 4.30638045e-01 4.08008426e-01
-9.31868017e-01 -1.00137484e+00 6.34877861e-01 7.78380930e-01
7.33830690e-01 -2.87440211e-01 -1.49348214e-01 -5.88029087e-01
2.04479292e-01 -5.86721122e-01 3.45996559e-01 -1.64867565e-01
-1.00304747e+00 2.72566110e-01 2.51409203e-01 -1.04306936e+00
-4.82133850e-02 -4.08546746e-01 -5.09636164e-01 9.32298899e-01
-6.67122662e-01 -4.08967435e-01 1.03787696e-02 6.15073323e-01
-3.82855922e-01 1.21069565e-01 6.89076066e-01 2.36073643e-01
-3.15434217e-01 6.45407617e-01 3.00958931e-01 1.74099430e-01
-7.72637799e-02 -9.83623981e-01 -8.99335742e-02 6.74692690e-01
-1.37725964e-01 6.52420282e-01 9.00119781e-01 -6.50245607e-01
-1.40165925e+00 -8.08117568e-01 8.56483757e-01 1.37353882e-01
4.09818649e-01 -3.25231671e-01 -6.03950799e-01 8.02843392e-01
-2.13942289e-01 5.52844033e-02 6.94657862e-01 2.70185024e-02
-1.76973268e-01 -1.86369821e-01 -1.50556970e+00 5.22147059e-01
1.28762913e+00 -1.70914859e-01 2.10036382e-01 2.69808799e-01
3.95530373e-01 -1.17911831e-01 -1.30425489e+00 5.76386571e-01
4.36951607e-01 -1.39034677e+00 6.85805678e-01 -2.46342406e-01
1.11117415e-01 -2.68286586e-01 -4.39436436e-01 -6.84979379e-01
-1.72556803e-01 -8.01987529e-01 2.87907392e-01 1.09837449e+00
7.18870342e-01 -8.52884889e-01 1.10538411e+00 4.36591238e-01
3.74074399e-01 -7.48953879e-01 -1.46815741e+00 -7.09318876e-01
2.42777735e-01 -2.67770439e-01 4.80743378e-01 7.51471519e-01
8.64013135e-02 -1.62921138e-02 -9.15079415e-02 3.34086329e-01
9.32884037e-01 6.89312667e-02 6.02908432e-01 -1.26378560e+00
-7.86160469e-01 -2.94903338e-01 -3.29006582e-01 -9.59946454e-01
-1.82546884e-01 -6.45738065e-01 -2.63844788e-01 -1.16929388e+00
5.15600443e-01 -8.99737358e-01 -5.28959297e-02 2.73468703e-01
6.74595088e-02 -3.89824659e-02 -3.26167017e-01 -1.87609598e-01
-3.40355486e-01 -3.69134452e-03 1.10433447e+00 2.67617106e-01
-1.56730026e-01 2.58863121e-01 -7.23189712e-01 7.11155713e-01
4.58567798e-01 -5.50511718e-01 -4.69786674e-01 -3.78840305e-02
5.32735586e-01 8.98817241e-01 2.39823669e-01 -7.04341769e-01
-6.08560145e-02 -2.66232431e-01 2.01676145e-01 -5.63153923e-01
2.26040989e-01 -5.22237301e-01 5.70274174e-01 5.22527397e-01
-6.81933984e-02 1.24024622e-01 -7.82827958e-02 8.05180669e-01
2.88007408e-01 -2.76708394e-01 8.99977803e-01 -3.97073060e-01
1.57955721e-01 2.02499896e-01 -3.21904629e-01 1.52505502e-01
1.26203334e+00 -5.21000147e-01 -2.82657534e-01 -5.40828705e-01
-7.50009954e-01 1.67044118e-01 4.64494407e-01 -5.15053630e-01
1.77166641e-01 -9.14380968e-01 -5.72267592e-01 2.86174804e-01
-1.45882405e-02 3.15665841e-01 2.78040886e-01 9.83835340e-01
-3.52437437e-01 -3.76539826e-02 3.14386487e-01 -7.12473810e-01
-1.03973615e+00 3.97234082e-01 8.64700750e-02 -3.32589477e-01
-1.60604715e-01 1.09238625e+00 -1.67313442e-01 -1.10274501e-01
5.86747676e-02 -2.62266338e-01 4.70253557e-01 -2.76248187e-01
5.04472069e-02 8.45995665e-01 -2.08141759e-01 -4.01839674e-01
-4.19899255e-01 5.45783281e-01 -1.43310770e-01 -1.52771324e-01
1.06188786e+00 4.03764918e-02 -5.60366392e-01 2.08745867e-01
1.15531242e+00 2.34857678e-01 -1.04882121e+00 -6.30346090e-02
-1.17365018e-01 -4.73341912e-01 -6.36328161e-01 -3.93041641e-01
-1.03306544e+00 5.16005695e-01 2.22496331e-01 8.70103955e-01
1.06861746e+00 4.27138478e-01 2.86876559e-01 -1.97901577e-01
9.06610429e-01 -1.13914323e+00 -1.41019300e-01 1.44515112e-01
1.24291122e-01 -5.66337287e-01 7.70023242e-02 -7.34037220e-01
-1.29403010e-01 8.27940822e-01 2.95768499e-01 -4.68428344e-01
1.01739562e+00 2.19301507e-01 -6.86704993e-01 -3.31644267e-01
-4.43802536e-01 -1.05062164e-01 -2.84063220e-01 2.39381656e-01
5.28827786e-01 3.83408576e-01 -6.13234520e-01 6.34724617e-01
-2.91469187e-01 -1.56697929e-01 5.10090709e-01 9.98423517e-01
-6.91835225e-01 -1.22497737e+00 -2.10098043e-01 8.55817795e-01
-4.83955294e-01 4.39148657e-02 -2.62688577e-01 9.65442300e-01
2.27464333e-01 1.05147088e+00 9.11543518e-02 -1.33954808e-01
9.12885293e-02 -8.61923024e-02 7.87088335e-01 -4.91667807e-01
-1.26678631e-01 3.28048646e-01 1.59149945e-01 -1.48707226e-01
-1.47415489e-01 -6.81654572e-01 -1.41434407e+00 -7.40891993e-01
-4.30441886e-01 5.42997122e-01 3.40986401e-01 7.51240969e-01
2.33834967e-01 1.10119261e-01 9.32937384e-01 -3.13603520e-01
-4.67054933e-01 -7.60464251e-01 -1.52600884e+00 4.38663783e-03
-2.08412006e-01 -5.76237738e-01 -7.74360895e-01 -2.91725516e-01]
|
[6.630346775054932, 4.889187812805176]
|
576dab8d-7806-44b1-bfd7-b687631fa374
|
high-accuracy-malware-classification-with-a
|
2004.05258
| null |
https://arxiv.org/abs/2004.05258v2
|
https://arxiv.org/pdf/2004.05258v2.pdf
|
Exploring Optimal Deep Learning Models for Image-based Malware Variant Classification
|
Analyzing a huge amount of malware is a major burden for security analysts. Since emerging malware is often a variant of existing malware, automatically classifying malware into known families greatly reduces a part of their burden. Image-based malware classification with deep learning is an attractive approach for its simplicity, versatility, and affinity with the latest technologies. However, the impact of differences in deep learning models and the degree of transfer learning on the classification accuracy of malware variants has not been fully studied. In this paper, we conducted an exhaustive survey of deep learning models using 24 ImageNet pre-trained models and five fine-tuning parameters, totaling 120 combinations, on two platforms. As a result, we found that the highest classification accuracy was obtained by fine-tuning one of the latest deep learning models with a relatively low degree of transfer learning, and we achieved the highest classification accuracy ever in cross-validation on the Malimg and Drebin datasets. We also confirmed that this trend holds true for the recent malware variants using the VirusTotal 2020 Windows and Android datasets. The experimental results suggest that it is effective to periodically explore optimal deep learning models with the latest models and malware datasets by gradually reducing the degree of transfer learning from half.
|
['Takahiro Shinagawa', 'Rikima Mitsuhashi']
|
2020-04-10
| null | null | null | null |
['computer-security']
|
['miscellaneous']
|
[-1.89877614e-01 -7.20556319e-01 -4.17958170e-01 -8.85690302e-02
-1.69084325e-01 -6.59273684e-01 8.09129357e-01 -2.67934829e-01
-5.47091246e-01 6.16286278e-01 -4.78891999e-01 -7.11857617e-01
7.52771571e-02 -6.47535682e-01 -8.20815623e-01 -5.76720834e-01
-4.98610288e-01 4.71066028e-01 2.30355456e-01 -3.18199873e-01
2.89453119e-01 6.39282465e-01 -1.40511274e+00 4.69069004e-01
7.16878831e-01 1.12121487e+00 -6.28834544e-03 7.38494635e-01
-2.50925839e-01 6.94320500e-01 -8.25285137e-01 -6.98388934e-01
3.31838429e-01 1.67775780e-01 -7.00664461e-01 -2.40747035e-01
2.82881498e-01 -6.87219858e-01 -2.29023471e-01 1.04239857e+00
1.66505784e-01 -5.43763340e-01 6.94914043e-01 -1.46992815e+00
-7.12254822e-01 4.74530131e-01 -7.17996299e-01 5.98633289e-01
-6.04967512e-02 3.12086582e-01 5.83641946e-01 -4.99257803e-01
4.67606723e-01 1.25148666e+00 1.17026961e+00 6.21017873e-01
-1.00166655e+00 -1.11001372e+00 -7.61477575e-02 4.78127241e-01
-1.29187632e+00 -1.93538904e-01 5.78231990e-01 -7.23225176e-01
1.34683537e+00 -5.49590290e-02 7.09785104e-01 1.67051780e+00
7.78817177e-01 2.78364331e-01 8.91828656e-01 2.17024256e-02
-1.97541481e-03 4.31969762e-01 3.18580598e-01 7.58803666e-01
4.83223915e-01 1.95936888e-01 2.76689053e-01 -3.78618091e-01
1.57414153e-01 3.57460350e-01 4.69133146e-02 -2.20091358e-01
-7.46301532e-01 1.14417815e+00 3.55609089e-01 6.29863560e-01
-1.23292916e-01 3.09630726e-02 1.04977489e+00 5.66504419e-01
6.55920386e-01 4.99637693e-01 -8.24693084e-01 -3.44990999e-01
-7.77909696e-01 -4.84596342e-02 7.40012407e-01 2.62672275e-01
7.42084086e-01 2.97877967e-01 2.89927691e-01 6.29537106e-01
1.97783019e-02 3.59691024e-01 7.33965218e-01 -5.23953438e-01
1.86354622e-01 6.41644180e-01 -4.20621783e-01 -1.06893444e+00
-3.48038465e-01 -4.86946255e-01 -8.29081416e-01 1.71115607e-01
2.33798698e-01 -2.84204781e-01 -8.66985381e-01 1.54914963e+00
-3.65495309e-02 2.52091438e-01 -2.01142892e-01 7.82780498e-02
5.43458045e-01 8.89452577e-01 -3.38951801e-03 -7.91963711e-02
1.17270732e+00 -7.59701490e-01 -3.38869005e-01 -8.59707519e-02
6.47826254e-01 -3.82822365e-01 1.02505410e+00 4.90412682e-01
-2.22927213e-01 -6.68111980e-01 -1.10812891e+00 3.53910834e-01
-7.57308900e-01 -1.18824601e-01 6.43022478e-01 1.09303570e+00
-1.17208850e+00 5.45979917e-01 -9.37696874e-01 -5.71436048e-01
6.92340195e-01 5.72804213e-01 -3.12727302e-01 1.22870892e-01
-1.06373072e+00 7.06279874e-01 5.29447734e-01 -3.74290854e-01
-1.22977293e+00 -6.58032715e-01 -3.21760207e-01 1.60686880e-01
5.62967919e-02 -2.12818012e-01 1.00888348e+00 -1.12187994e+00
-1.26725435e+00 7.81290472e-01 2.62603641e-01 -8.00729454e-01
2.52404094e-01 1.14698019e-02 -5.48672318e-01 -1.45508200e-01
-1.45588592e-01 7.19173908e-01 1.27478158e+00 -1.15585792e+00
-5.51280141e-01 -4.91908491e-01 1.73907936e-01 -5.58765650e-01
-1.06340706e+00 8.60198066e-02 -6.07554317e-02 -1.96523786e-01
-9.28666770e-01 -1.02849030e+00 2.93525815e-01 -4.46178794e-01
2.98973620e-02 -2.74263322e-01 1.55663347e+00 -7.94306815e-01
1.01466537e+00 -2.22561908e+00 -1.06520250e-01 -8.25343505e-02
4.11719173e-01 8.85649204e-01 -3.48968923e-01 1.54140145e-01
-4.27321233e-02 4.86186475e-01 2.77513843e-02 -8.04649368e-02
-3.26831549e-01 1.79087922e-01 -3.86373937e-01 4.18694824e-01
3.36643636e-01 1.00077128e+00 -7.12904811e-01 -2.40030900e-01
2.97092706e-01 8.82170975e-01 -4.94087130e-01 1.11129217e-01
-2.14584887e-01 3.13455850e-01 -2.79599965e-01 7.10114837e-01
6.93718374e-01 -3.29351306e-01 1.93927623e-03 -8.81534442e-02
1.39856026e-01 3.19153517e-02 -2.37587914e-01 9.61044908e-01
-6.05030894e-01 9.72319067e-01 1.69372126e-01 -1.28168905e+00
7.52726376e-01 9.97128263e-02 6.06553078e-01 -7.64385343e-01
3.88626784e-01 1.58652186e-01 5.20315647e-01 -5.40512562e-01
1.72725573e-01 1.06177554e-01 7.50883371e-02 3.48236203e-01
1.36943415e-01 9.94710103e-02 1.24845229e-01 -1.29287571e-01
1.25660527e+00 -3.96387190e-01 2.46785194e-01 -1.98568076e-01
6.44299150e-01 1.36478161e-02 2.30820999e-01 5.69105148e-01
-6.11932278e-01 -1.20934084e-01 7.08518863e-01 -8.62401724e-01
-1.09168613e+00 -8.42063963e-01 -2.27288947e-01 1.26870811e+00
-3.43138367e-01 -4.77789879e-01 -1.03343689e+00 -1.02419782e+00
4.72462503e-03 4.79154736e-01 -8.58953416e-01 -3.04851592e-01
-5.98394692e-01 -1.00856650e+00 9.29499149e-01 1.26491681e-01
6.66809857e-01 -1.07247698e+00 -5.15647829e-01 -2.42961749e-01
1.59297153e-01 -1.10261989e+00 -2.45368294e-02 2.24152252e-01
-7.83829629e-01 -1.34602773e+00 -4.67099071e-01 -7.30277419e-01
3.22622627e-01 2.61866182e-01 9.39144611e-01 2.06442073e-01
-1.68512076e-01 1.74359649e-01 -4.59274352e-01 -4.68722790e-01
-8.73452306e-01 5.64810932e-01 2.76526809e-01 -1.48384228e-01
5.13956189e-01 -4.08424765e-01 -1.14583790e-01 1.01169102e-01
-7.64306903e-01 -4.66292322e-01 5.76415420e-01 8.71460199e-01
6.82732910e-02 5.04489183e-01 5.22478580e-01 -8.31033349e-01
8.15282702e-01 -8.02797616e-01 -6.09700978e-01 7.10860789e-02
-6.99280441e-01 -1.48729563e-01 1.13454437e+00 -7.31926501e-01
-7.53868878e-01 -1.75260112e-01 -3.27845186e-01 -7.37438262e-01
4.62291390e-02 1.49881169e-01 1.64677829e-01 -3.11315536e-01
8.88684928e-01 1.65848210e-01 3.23509395e-01 -1.96059555e-01
8.88136625e-02 7.94815242e-01 -3.48141231e-02 -2.07884386e-01
7.40060568e-01 3.75616461e-01 -3.12310010e-01 -1.03579497e+00
-2.70890951e-01 -4.54644859e-02 -4.46215391e-01 5.36363535e-02
8.75945926e-01 -7.01670647e-01 -8.00739646e-01 8.44343305e-01
-1.07404757e+00 -3.32536876e-01 1.65679902e-01 1.39375016e-01
-3.52798626e-02 4.88219529e-01 -7.62094080e-01 -3.76721382e-01
-5.86221695e-01 -1.74950421e+00 7.39117265e-01 -1.58323929e-01
-1.49872482e-01 -1.10278869e+00 1.12909541e-01 1.91113159e-01
7.69854844e-01 1.40234306e-01 1.07639980e+00 -9.79493916e-01
-1.12212822e-01 -3.75152767e-01 -4.43595469e-01 6.63472593e-01
5.69684982e-01 5.07432938e-01 -9.72649038e-01 -7.26189077e-01
2.18889460e-01 -4.01664138e-01 1.09116530e+00 3.04549694e-01
1.68968642e+00 -3.84899646e-01 -4.78034288e-01 7.34798849e-01
1.16711879e+00 4.50707763e-01 4.14891154e-01 3.98395687e-01
9.24359620e-01 3.12279671e-01 7.73333311e-02 1.43018082e-01
5.37813157e-02 4.90043193e-01 7.68965065e-01 3.33986878e-01
2.13349059e-01 1.65550888e-01 6.83292866e-01 7.00781524e-01
9.90531128e-03 -2.05625355e-01 -1.17309546e+00 1.04651205e-01
-1.26389170e+00 -1.10555518e+00 3.35169226e-01 1.84353828e+00
4.96715367e-01 3.56723279e-01 3.51224005e-01 -2.95685511e-02
6.91756010e-01 2.95469195e-01 -5.83766818e-01 -7.94578910e-01
1.43530160e-01 2.95702159e-01 4.12969857e-01 2.01665044e-01
-1.46684480e+00 8.53693187e-01 7.02996159e+00 1.17882729e+00
-1.84190404e+00 4.31710303e-01 9.23941374e-01 3.93058313e-03
2.24793568e-01 -5.13343573e-01 -8.21209192e-01 8.42019439e-01
1.61420834e+00 -3.61235552e-02 7.28715658e-01 1.21118784e+00
-1.61879376e-01 3.59569758e-01 -8.00483286e-01 9.13761556e-01
1.67825729e-01 -1.44235516e+00 1.16319545e-01 5.23673594e-01
7.49597311e-01 5.72499812e-01 3.31321061e-01 7.65498281e-01
-5.78817613e-02 -1.09787476e+00 2.23865002e-01 -9.27987099e-02
9.07911241e-01 -9.26653802e-01 8.00192952e-01 4.60134327e-01
-1.02002931e+00 -7.81111896e-01 -3.24174106e-01 -1.60615426e-02
-2.86861241e-01 3.58671039e-01 -1.14765203e+00 5.62433824e-02
9.53916669e-01 7.86966980e-01 -7.81087399e-01 4.82676387e-01
3.14317167e-01 8.47175777e-01 -3.77837345e-02 -1.68547571e-01
2.07622737e-01 1.01311415e-01 1.79191664e-01 1.39567518e+00
2.33798996e-01 -5.81765175e-01 1.49585500e-01 6.18078351e-01
-2.09326893e-01 -1.98693350e-01 -9.04243052e-01 -5.32095611e-01
3.14371407e-01 1.44005799e+00 -7.67636418e-01 -3.93751621e-01
-3.94314170e-01 6.48156703e-01 3.76432657e-01 -9.22779962e-02
-1.18904102e+00 -8.93621743e-02 9.07110512e-01 2.98675299e-01
3.49130243e-01 -2.49596760e-01 -1.09157689e-01 -9.46274996e-01
-4.22681719e-01 -1.31254053e+00 2.41786540e-01 -1.41634986e-01
-1.41848350e+00 8.72693300e-01 4.32471856e-02 -9.66538072e-01
-5.38149118e-01 -1.24892628e+00 -7.39838541e-01 3.28770965e-01
-1.01221561e+00 -1.09788537e+00 -2.53720194e-01 4.31353420e-01
5.32680690e-01 -6.84315979e-01 7.12710381e-01 5.07240653e-01
-7.30868697e-01 7.55291879e-01 3.15064669e-01 1.47139177e-01
4.08473879e-01 -7.04730034e-01 5.68396807e-01 6.56845450e-01
-9.95782018e-02 5.71975887e-01 3.77698183e-01 -7.14398801e-01
-1.47463655e+00 -1.28024006e+00 1.79853454e-01 -5.07919312e-01
9.27792430e-01 -4.71712977e-01 -1.11894059e+00 6.57549858e-01
2.90152013e-01 -1.99152306e-01 4.97291595e-01 -1.12736121e-01
-6.62308037e-01 -3.10766995e-01 -1.26795971e+00 2.67473936e-01
7.52253830e-01 -6.28634095e-01 -1.19557112e-01 2.72783369e-01
9.37657177e-01 -2.01457534e-02 -6.35909736e-01 7.36935616e-01
6.64163411e-01 -1.13578153e+00 1.03508949e+00 -5.08467078e-01
3.79689008e-01 2.23743543e-01 -1.19429812e-01 -1.24710548e+00
-3.26045066e-01 1.83637943e-02 -5.18521488e-01 9.87500250e-01
2.66442329e-01 -8.23670864e-01 7.64758825e-01 -1.42912447e-01
6.96810484e-02 -9.88288701e-01 -8.75799417e-01 -9.15197849e-01
3.53145450e-01 -3.45572263e-01 5.69749057e-01 1.06034756e+00
-5.24816930e-01 2.39301592e-01 -3.90027761e-01 -3.20983231e-01
3.62486035e-01 -3.67938340e-01 8.16285491e-01 -1.27093244e+00
-2.86278427e-01 -7.04481661e-01 -5.16314924e-01 -2.99717158e-01
7.44739771e-01 -8.36035848e-01 -5.36097109e-01 -9.31026638e-01
4.88488853e-01 -2.58267313e-01 -2.84224093e-01 4.73845601e-01
9.96927097e-02 4.05552655e-01 -6.37622997e-02 2.69938558e-01
-2.79207855e-01 2.48226747e-01 8.16586733e-01 -4.89888906e-01
-5.49155548e-02 5.74745573e-02 -4.86958355e-01 7.22579122e-01
1.11770642e+00 -1.64458603e-01 -5.68337500e-01 -4.08269137e-01
-1.28309742e-01 -5.72065234e-01 2.46282339e-01 -1.07911849e+00
-3.22982043e-01 1.18341129e-02 6.33522272e-01 -5.61065674e-01
4.86860245e-01 -7.29323030e-01 2.22406179e-01 1.05661750e+00
1.67793602e-01 4.25353527e-01 5.82887888e-01 2.91632146e-01
1.46454260e-01 -5.77735491e-02 9.24171686e-01 -1.29821494e-01
-9.83443677e-01 3.98536891e-01 -4.99231339e-01 -1.46901354e-01
1.17862248e+00 -2.91425675e-01 -4.45946842e-01 -5.23037724e-02
-1.08497970e-01 -2.06219137e-01 5.89168608e-01 7.85008371e-01
3.90363097e-01 -9.11579132e-01 -4.90125418e-01 4.67042774e-01
-3.67812291e-02 -6.84416890e-01 1.19664997e-01 6.47053540e-01
-7.10105658e-01 7.01719761e-01 -7.21111178e-01 -9.46662605e-01
-1.40978754e+00 9.62963164e-01 4.52448606e-01 -5.05505085e-01
-1.57711178e-01 7.09351718e-01 9.18738544e-02 -6.07918501e-01
2.32638180e-01 -1.95598099e-02 -3.22590232e-01 1.01470590e-01
5.99667192e-01 5.10740221e-01 2.38417059e-01 -7.68411994e-01
-5.21521389e-01 3.31255972e-01 -4.71252382e-01 5.69509685e-01
1.18999290e+00 3.43685359e-01 -3.29455763e-01 2.55169958e-01
1.70161283e+00 -1.03572235e-01 -9.65620041e-01 3.12970161e-01
-1.58114627e-01 -3.81894678e-01 -1.82170317e-01 -5.05113006e-01
-1.35021341e+00 1.00577760e+00 9.04224455e-01 5.41848958e-01
1.14659631e+00 -1.12983696e-01 7.77955890e-01 5.61973214e-01
4.89525020e-01 -5.36551356e-01 3.38231802e-01 8.34751904e-01
5.82312107e-01 -1.40937853e+00 -2.06109777e-01 1.98721159e-02
-2.03841254e-01 8.99580121e-01 1.07984102e+00 -1.69266671e-01
9.07911658e-01 4.10479337e-01 -1.84973791e-01 -1.31576687e-01
-7.63795078e-01 4.10911947e-01 1.02014560e-02 8.03613603e-01
4.32086557e-01 2.18479991e-01 -1.15473852e-01 2.43795648e-01
-2.12609679e-01 -1.92308754e-01 2.55278349e-01 5.75254738e-01
-4.08998817e-01 -1.09497583e+00 -3.47530216e-01 8.36851895e-01
-6.56389534e-01 -6.84053227e-02 -5.75956464e-01 9.20222938e-01
3.85510057e-01 7.02984631e-01 2.26687089e-01 -1.09828579e+00
-2.27633446e-01 -7.41699757e-03 4.71873641e-01 -4.02439713e-01
-7.26812243e-01 -6.79334641e-01 -2.03252703e-01 -3.84722352e-01
-4.72178943e-02 -2.02989385e-01 -8.89843643e-01 -8.85836184e-01
-1.78137302e-01 -7.65587315e-02 8.14482868e-01 8.57873797e-01
6.45722270e-01 4.81638551e-01 8.00037801e-01 -1.14083827e+00
-8.67278576e-01 -1.05048728e+00 -6.56569973e-02 3.11002135e-01
3.45889777e-01 -8.35767627e-01 -4.83406484e-01 -2.20138282e-01]
|
[14.412548065185547, 9.673664093017578]
|
01b24dca-f721-403e-821d-993dd9826e45
|
improvement-of-computational-performance-of
|
2301.05102
| null |
https://arxiv.org/abs/2301.05102v1
|
https://arxiv.org/pdf/2301.05102v1.pdf
|
Improvement of Computational Performance of Evolutionary AutoML in a Heterogeneous Environment
|
Resource-intensive computations are a major factor that limits the effectiveness of automated machine learning solutions. In the paper, we propose a modular approach that can be used to increase the quality of evolutionary optimization for modelling pipelines with a graph-based structure. It consists of several stages - parallelization, caching and evaluation. Heterogeneous and remote resources can be involved in the evaluation stage. The conducted experiments confirm the correctness and effectiveness of the proposed approach. The implemented algorithms are available as a part of the open-source framework FEDOT.
|
['Denis Nasonov', 'Sergey Pakulin', 'Valerii Pokrovskii', 'Sergey Teryoshkin', 'Nikolay O. Nikitin']
|
2023-01-12
| null | null | null | null |
['automl']
|
['methodology']
|
[-1.08931974e-01 -3.65817212e-02 3.18833321e-01 -2.00911552e-01
8.42541605e-02 -2.95425951e-01 5.32197118e-01 5.10053039e-01
-5.25390327e-01 6.09491885e-01 -1.88453868e-01 -2.63235509e-01
-4.37804312e-01 -9.85274613e-01 -3.12809020e-01 -4.52016085e-01
-4.46880087e-02 6.39357448e-01 6.37800097e-01 -2.54489392e-01
6.13785923e-01 8.41276228e-01 -2.06847191e+00 1.81456298e-01
7.98615515e-01 7.38688827e-01 3.46888155e-01 7.50305295e-01
-3.31687391e-01 6.50523901e-01 -3.44385982e-01 -1.01714320e-01
1.78202480e-01 -4.55044299e-01 -7.62416184e-01 -1.20020851e-01
-6.17330968e-01 2.04622954e-01 4.44298089e-01 7.71597087e-01
6.51827812e-01 1.61928713e-01 4.96942192e-01 -1.28211343e+00
3.96802932e-01 3.87447119e-01 -1.13243803e-01 1.11470833e-01
2.62784481e-01 1.94072142e-01 6.24680519e-01 -7.41919458e-01
7.22192585e-01 9.28132594e-01 3.97457778e-01 -9.39230472e-02
-6.89255059e-01 -2.17991635e-01 -2.34078065e-01 6.12504184e-01
-1.58531797e+00 -4.81887966e-01 5.85839629e-01 -5.32665491e-01
1.32921278e+00 6.96750939e-01 9.51225877e-01 2.25918829e-01
3.70428056e-01 7.00527430e-02 9.93920505e-01 -6.63750529e-01
7.53000498e-01 3.36191714e-01 -2.83566397e-02 7.85366297e-01
4.41105604e-01 1.65629193e-01 -5.97799420e-01 -3.03345799e-01
1.62478387e-01 -3.42587799e-01 1.29628167e-01 -3.65236402e-01
-4.89976496e-01 8.22738647e-01 1.64910406e-01 6.66454673e-01
-5.45238853e-01 -5.37909120e-02 5.51902294e-01 9.24061537e-02
5.36398590e-01 4.48148400e-01 -3.56987059e-01 -4.73601937e-01
-1.19796038e+00 3.50573927e-01 1.03264558e+00 9.13396299e-01
6.39281213e-01 5.40472232e-02 1.60531893e-01 2.71809697e-01
5.25109053e-01 -1.55189196e-02 5.68204820e-01 -2.55669713e-01
-1.22567408e-01 1.21350253e+00 1.26235157e-01 -9.31476951e-01
-7.63119817e-01 -5.47537088e-01 -3.93375367e-01 3.54471326e-01
-2.28262216e-01 -3.25899452e-01 -3.16174209e-01 1.00208497e+00
9.67728734e-01 2.13349462e-01 3.02073825e-02 8.67608130e-01
8.08421135e-01 5.45757532e-01 1.16883151e-01 -2.55000979e-01
1.17525649e+00 -1.11664295e+00 -6.77283823e-01 5.71693003e-01
5.89820504e-01 -1.19360375e+00 6.26350284e-01 2.48500183e-01
-1.21201169e+00 -4.70239282e-01 -1.17607248e+00 3.75063330e-01
-7.09872425e-01 3.30558151e-01 7.71600306e-01 8.44816685e-01
-1.09977639e+00 8.00096929e-01 -1.13211489e+00 -5.50743103e-01
-2.35963643e-01 3.18131924e-01 -2.33308468e-02 3.47347558e-01
-6.54009283e-01 1.00813806e+00 9.65820193e-01 3.94717991e-01
-6.42075181e-01 -4.44756269e-01 -3.66614401e-01 2.58904964e-01
3.91862839e-01 -7.03450024e-01 8.32014740e-01 -7.80123591e-01
-1.71167827e+00 5.31682372e-01 -1.72073543e-02 -2.62769729e-01
9.35209334e-01 -1.08100533e-01 -3.30235183e-01 -6.36139512e-02
-4.15487051e-01 -5.05440831e-02 4.34145123e-01 -7.10244060e-01
-5.39771557e-01 -3.22879940e-01 -2.37004220e-01 2.67943799e-01
-3.00849020e-01 4.70234811e-01 -3.77399236e-01 -2.42981285e-01
-1.03005201e-01 -8.50692868e-01 -2.89936274e-01 -3.36253911e-01
-1.92419700e-02 1.30816381e-02 4.78850037e-01 -6.27636433e-01
1.41143286e+00 -1.67908204e+00 3.32719624e-01 3.14658076e-01
-2.22541019e-01 3.88387650e-01 3.32071692e-01 9.18759763e-01
2.04000682e-01 9.54687670e-02 -2.59712189e-01 -4.14132811e-02
1.56930998e-01 -8.94886628e-02 5.24227381e-01 3.06856215e-01
1.96332693e-01 4.41413313e-01 -4.35223043e-01 -6.27104402e-01
5.64261138e-01 3.69259119e-01 -3.47645342e-01 3.87668103e-01
-5.26522756e-01 3.17665428e-01 -6.54394031e-01 4.56464291e-01
7.25142121e-01 1.15485534e-01 3.94656718e-01 -1.50162891e-01
-8.55761409e-01 -4.79469523e-02 -1.76477861e+00 1.58818316e+00
-5.20420611e-01 3.08177732e-02 1.57814264e-01 -1.06514549e+00
9.58826482e-01 3.66755784e-01 4.57611829e-01 -2.80076504e-01
4.98700410e-01 3.46005768e-01 1.43429413e-01 -6.52305007e-01
8.06026876e-01 4.30223942e-01 4.04173642e-01 4.28025156e-01
1.01021461e-01 -1.27613276e-01 6.68150246e-01 -2.50937909e-01
9.34195757e-01 8.97649229e-01 6.12403452e-01 -5.86628437e-01
1.08595610e+00 3.87256265e-01 4.62333620e-01 2.42942795e-01
2.14402869e-01 -1.59571588e-01 4.44572449e-01 -4.98943746e-01
-1.27980018e+00 -4.09595251e-01 -7.69393966e-02 8.16682696e-01
-1.37232691e-01 -6.01866722e-01 -8.36020172e-01 -8.62253234e-02
-3.98269892e-01 5.27255595e-01 -1.63807556e-01 1.76949427e-01
-2.87843376e-01 -1.08165276e+00 3.12555850e-01 -1.79039445e-02
3.04709762e-01 -1.04330051e+00 -1.36311483e+00 4.48420048e-01
3.40692490e-01 -9.31293964e-01 4.51386750e-01 4.15698774e-02
-1.00889468e+00 -1.06747758e+00 -1.97649077e-01 -2.60949194e-01
4.82694805e-01 -3.81754756e-01 1.24090934e+00 6.90546513e-01
-7.30832219e-01 -3.75057273e-02 -6.58517301e-01 -5.43550253e-01
-5.05855083e-01 3.38380963e-01 -5.00156581e-01 5.71548268e-02
9.85728055e-02 -6.20424688e-01 -4.63050842e-01 2.81163305e-01
-8.55940402e-01 2.09921136e-01 4.59789813e-01 4.48459983e-01
5.26462615e-01 3.67642075e-01 2.58252084e-01 -9.24653411e-01
6.61969304e-01 -4.70597923e-01 -1.44803977e+00 3.43563110e-01
-8.37807715e-01 3.15476984e-01 6.11239314e-01 1.19254403e-01
-1.10699654e+00 3.88429791e-01 -4.44394082e-01 4.41590436e-02
-2.78600365e-01 6.87057495e-01 -2.20379382e-01 -4.57625985e-01
3.62210982e-02 1.77421629e-01 -2.40433410e-01 -6.25655293e-01
-5.93829155e-03 5.46170771e-01 -1.35338694e-01 -6.69266343e-01
4.03802544e-01 -3.35302986e-02 5.35638392e-01 -8.09883416e-01
2.14709222e-01 -4.47589159e-01 -4.59557265e-01 -5.64792514e-01
7.04343140e-01 -5.75040102e-01 -7.59650767e-01 4.19670999e-01
-1.13859653e+00 7.78858364e-02 1.05595946e-01 3.34846616e-01
-1.78474575e-01 1.82292223e-01 -1.30201459e-01 -1.08197641e+00
-6.62867248e-01 -1.10180509e+00 7.23913491e-01 3.59026462e-01
2.18536016e-02 -9.48714793e-01 3.27965498e-01 1.82942837e-01
7.40110695e-01 3.45865369e-01 7.17292368e-01 -6.87032878e-01
-6.21658921e-01 1.27068488e-02 2.08507866e-01 -6.60333559e-02
-4.37935174e-01 7.36802578e-01 -7.47023046e-01 5.04223024e-03
-2.33429372e-01 1.36231119e-02 9.38165709e-02 1.43238418e-02
7.87998438e-01 -6.68040663e-02 -2.57867217e-01 5.55847943e-01
1.95474076e+00 1.69490561e-01 5.48715949e-01 4.79325056e-01
1.11907616e-01 6.99243665e-01 8.80901694e-01 1.01811242e+00
1.49802836e-02 7.63121724e-01 4.66438055e-01 -8.05137008e-02
1.71755016e-01 1.97705388e-01 -3.07681654e-02 1.10965502e+00
-3.51423651e-01 -2.98177391e-01 -1.29402304e+00 4.19004321e-01
-1.97842526e+00 -7.35657156e-01 -5.86254597e-01 2.05961871e+00
2.71089107e-01 6.69855950e-03 1.84724350e-02 2.94694841e-01
6.22639954e-01 -9.11729708e-02 -1.42211601e-01 -9.11417127e-01
2.68752187e-01 6.31511331e-01 2.74131179e-01 5.27128816e-01
-7.36733735e-01 8.37408125e-01 5.72355890e+00 4.77959394e-01
-1.18756711e+00 4.37929213e-01 6.91087497e-03 -3.90769579e-02
-6.05424494e-02 1.12380877e-01 -5.71634412e-01 5.16449392e-01
1.38974059e+00 -3.40257227e-01 3.14722568e-01 1.02902317e+00
4.60401297e-01 -4.75858688e-01 -6.44086421e-01 4.49152559e-01
-1.86584502e-01 -1.44198680e+00 -3.06124926e-01 -2.30026707e-01
5.64081967e-01 1.22749463e-01 -4.55631107e-01 -1.59868643e-01
8.86414647e-02 -7.06047654e-01 6.90888941e-01 5.53887844e-01
-3.28831747e-03 -1.07822931e+00 1.12478518e+00 3.47073406e-01
-1.46294415e+00 1.56142116e-01 -5.10176063e-01 -2.90355057e-01
1.21409796e-01 5.68310618e-01 -9.70781267e-01 1.29705656e+00
7.29968786e-01 -5.64961322e-03 -7.32764602e-01 1.28678489e+00
1.00367172e-02 4.01125789e-01 -4.50432748e-01 -5.94785511e-01
-4.59128767e-02 -5.37793756e-01 4.72063094e-01 1.15145278e+00
5.03840566e-01 -1.74178541e-01 2.47507319e-02 8.77106547e-01
4.26066101e-01 9.61464524e-01 -4.45315093e-01 -1.92159683e-01
4.19312686e-01 1.47782838e+00 -9.88059163e-01 -2.47224376e-01
-3.02891403e-01 6.97326958e-01 1.59099996e-01 -2.08761632e-01
-8.20006907e-01 -4.03949887e-01 -8.84864181e-02 6.14608340e-02
2.87397832e-01 -1.26479581e-01 -2.78930843e-01 -8.23277593e-01
2.05764640e-02 -7.96776235e-01 3.46574783e-01 -6.32853031e-01
-3.74422580e-01 1.02442336e+00 6.44069239e-02 -8.62664998e-01
-4.23982054e-01 -4.85867918e-01 -6.08827710e-01 9.32644248e-01
-1.10596669e+00 -1.19330978e+00 -5.85234642e-01 4.06762332e-01
2.91196674e-01 -4.05177355e-01 1.12633288e+00 3.43944222e-01
-6.57677352e-01 3.12381666e-02 1.99757382e-01 -6.97865605e-01
-5.60465381e-02 -9.24296916e-01 1.31182037e-02 1.34062052e+00
-1.37491822e-01 2.92293966e-01 7.93013215e-01 -6.46454692e-01
-1.50460076e+00 -6.81509316e-01 9.47382808e-01 3.32420230e-01
6.51796997e-01 -2.58271396e-01 -4.62225765e-01 1.68018162e-01
5.35449743e-01 -2.29644105e-01 5.88133156e-01 2.83377934e-02
4.91773814e-01 -6.08011223e-02 -1.31842458e+00 2.67320067e-01
5.71996212e-01 1.17414162e-01 -3.44526410e-01 2.43881166e-01
1.59054086e-01 -3.57556313e-01 -1.10764313e+00 2.78649271e-01
2.56085426e-01 -1.21615088e+00 4.85212356e-01 -2.68748939e-01
1.32752955e-01 -4.27973032e-01 2.62815237e-01 -9.50315654e-01
1.82274446e-01 -5.80449760e-01 -2.97743585e-02 1.21708238e+00
3.96655381e-01 -5.77318370e-01 6.63851023e-01 5.80262959e-01
-4.38400134e-02 -6.55343950e-01 -9.04231250e-01 -3.01277757e-01
-4.90587622e-01 -1.81895778e-01 8.68536890e-01 5.15271664e-01
-1.74496859e-01 1.67154565e-01 -8.00861120e-02 3.04115653e-01
3.76596808e-01 2.56620616e-01 6.40550137e-01 -1.22831559e+00
-6.07887566e-01 -3.62427264e-01 -6.57802701e-01 3.31148833e-01
-2.48330370e-01 -6.76325798e-01 -3.89362544e-01 -1.42806721e+00
-5.27129211e-02 -4.01352793e-01 -1.45823807e-01 1.24281421e-01
1.79615170e-01 -1.62085935e-01 1.14848353e-01 -1.09472997e-01
-4.12637472e-01 4.74292040e-01 7.00035393e-01 4.14264321e-01
-2.15168580e-01 -7.97712654e-02 1.04914881e-01 7.18380928e-01
9.47029233e-01 -7.42152452e-01 -4.95702058e-01 -2.05720365e-01
3.13624322e-01 1.33096718e-03 -8.66087526e-03 -1.22517824e+00
2.92482585e-01 -2.23804191e-01 -1.90307982e-02 -6.37522995e-01
2.20095694e-01 -1.20899343e+00 1.08497548e+00 9.54120159e-01
2.17364188e-02 6.68130577e-01 3.99523109e-01 6.56195059e-02
-2.76665092e-01 -8.08083177e-01 5.75361311e-01 -1.19516186e-01
-7.05010533e-01 -1.10432141e-01 -3.41973245e-01 -3.59699041e-01
1.46245480e+00 1.29264593e-01 1.50061743e-02 3.00317258e-01
-6.08607888e-01 -8.62924382e-02 5.01873136e-01 4.53574806e-02
3.63994896e-01 -6.85047090e-01 -7.27327883e-01 6.97898567e-02
-1.16975084e-01 -4.13492680e-01 1.55667156e-01 7.54898787e-01
-1.49045479e+00 4.58097219e-01 -7.15029597e-01 -2.68927097e-01
-1.48128867e+00 5.09861231e-01 4.64607030e-01 -5.56602478e-01
-2.84101754e-01 4.20908213e-01 -7.69077063e-01 -4.44037735e-01
-9.30212662e-02 2.07648873e-01 -3.17668796e-01 -1.96927771e-01
2.92368174e-01 8.74003112e-01 6.06969297e-01 -6.55970812e-01
-6.53457284e-01 4.78277504e-01 4.86041665e-01 -3.09975833e-01
1.72678816e+00 1.32012725e-01 -6.49311781e-01 2.07768962e-01
5.56492865e-01 1.31438952e-02 -4.01213348e-01 4.36157525e-01
3.36121619e-01 -3.83320421e-01 4.10386860e-01 -8.20211589e-01
-9.07828152e-01 5.43680131e-01 8.67434800e-01 -8.33620057e-02
1.28079379e+00 -5.99194407e-01 7.72579759e-02 3.45123798e-01
3.39487582e-01 -1.24801517e+00 -6.67898953e-01 3.06393355e-02
7.01370239e-01 -6.74576938e-01 4.02961075e-01 -5.83314180e-01
-3.16503137e-01 1.24447536e+00 6.10620797e-01 4.42326888e-02
7.31602192e-01 6.49372041e-01 -3.05525452e-01 -2.83667535e-01
-9.52517092e-01 -3.79804850e-01 2.14183062e-01 2.44976312e-01
7.20993638e-01 -4.24111374e-02 -1.50665200e+00 5.15520871e-01
1.10037383e-02 3.21101278e-01 3.97712320e-01 1.25304055e+00
-3.87991607e-01 -1.45280588e+00 -4.66396868e-01 1.43632600e-02
-4.24978971e-01 -2.92634610e-02 -2.87755162e-01 8.74143660e-01
4.91192222e-01 1.08157337e+00 -2.94633508e-01 -2.64160573e-01
2.21584499e-01 8.96905512e-02 4.98126924e-01 -3.25469315e-01
-1.21642280e+00 -8.73548794e-04 3.68131250e-01 -6.91018283e-01
-4.50160861e-01 -6.28139019e-01 -9.92186368e-01 -4.11359161e-01
-3.10085267e-01 6.61117136e-01 1.45115566e+00 6.65678799e-01
6.57805264e-01 8.02668452e-01 4.71105516e-01 -6.39640808e-01
-2.63754278e-01 -9.43028212e-01 -1.87052280e-01 -5.89947449e-03
-6.64776921e-01 -6.93862796e-01 -6.43972009e-02 -3.03591918e-02]
|
[6.052398204803467, 3.6042537689208984]
|
9f4d3b94-bed4-47ae-864b-8cb9ad0e6791
|
diffusion-models-for-constrained-domains
|
2304.05364
| null |
https://arxiv.org/abs/2304.05364v1
|
https://arxiv.org/pdf/2304.05364v1.pdf
|
Diffusion Models for Constrained Domains
|
Denoising diffusion models are a recent class of generative models which achieve state-of-the-art results in many domains such as unconditional image generation and text-to-speech tasks. They consist of a noising process destroying the data and a backward stage defined as the time-reversal of the noising diffusion. Building on their success, diffusion models have recently been extended to the Riemannian manifold setting. Yet, these Riemannian diffusion models require geodesics to be defined for all times. While this setting encompasses many important applications, it does not include manifolds defined via a set of inequality constraints, which are ubiquitous in many scientific domains such as robotics and protein design. In this work, we introduce two methods to bridge this gap. First, we design a noising process based on the logarithmic barrier metric induced by the inequality constraints. Second, we introduce a noising process based on the reflected Brownian motion. As existing diffusion model techniques cannot be applied in this setting, we derive new tools to define such models in our framework. We empirically demonstrate the applicability of our methods to a number of synthetic and real-world tasks, including the constrained conformational modelling of protein backbones and robotic arms.
|
['Michael Hutchinson', 'Emile Mathieu', 'Valentin De Bortoli', 'Leo Klarner', 'Nic Fishman']
|
2023-04-11
| null | null | null | null |
['protein-design']
|
['medical']
|
[ 4.04954463e-01 2.44370863e-01 2.22325251e-01 -2.22463936e-01
-5.37416488e-02 -6.72876298e-01 1.00149417e+00 -1.96006820e-01
-4.87029940e-01 3.93477410e-01 6.43342808e-02 -5.20010531e-01
-3.67817611e-01 -7.67780781e-01 -6.72130585e-01 -1.16634631e+00
-5.92077784e-02 2.23367780e-01 2.67884284e-01 -5.03352821e-01
4.15906072e-01 4.90324706e-01 -1.08023548e+00 -2.07772627e-01
8.35725963e-01 4.31698948e-01 2.52958983e-01 7.25579977e-01
9.37043205e-02 4.54162210e-01 -2.37520278e-01 -3.03415865e-01
3.15675825e-01 -8.02613318e-01 -6.41316712e-01 2.70950764e-01
-8.98799822e-02 1.21224239e-01 -4.49392766e-01 1.16384566e+00
5.06760597e-01 3.92701149e-01 8.27920914e-01 -1.05200791e+00
-9.66702223e-01 1.78615585e-01 -5.33055544e-01 -3.88998613e-02
7.78117850e-02 -7.42979720e-02 8.72575581e-01 -8.57296109e-01
8.91558528e-01 1.12190938e+00 4.45001125e-01 8.23680401e-01
-1.47932053e+00 -1.06002778e-01 2.35564495e-03 1.02069206e-03
-1.10143435e+00 -2.78173000e-01 9.22670364e-01 -6.33287728e-01
4.90496278e-01 2.59459794e-01 5.01985073e-01 1.18308043e+00
3.54554892e-01 4.70271796e-01 1.02924192e+00 -3.57764065e-01
5.07689893e-01 -2.33800933e-01 1.67168658e-02 6.62386239e-01
6.78933710e-02 -7.59311859e-03 -3.81047308e-01 -1.07451349e-01
7.90580392e-01 -6.47997558e-02 -3.68315369e-01 -8.52941334e-01
-1.49684000e+00 1.26331055e+00 1.51974395e-01 4.26059306e-01
-2.69711971e-01 8.46542343e-02 3.20100710e-02 3.90202880e-01
6.51494205e-01 3.43526304e-01 9.18244496e-02 -4.88852561e-02
-6.92457616e-01 1.76877588e-01 9.98092473e-01 8.84933650e-01
5.78548551e-01 -2.28288293e-01 9.47276279e-02 5.82466066e-01
5.24813414e-01 2.76989847e-01 2.30048001e-01 -1.00124645e+00
2.96029896e-01 1.79780573e-01 2.11794510e-01 -1.07052708e+00
-4.05252218e-01 -1.80681899e-01 -9.93505418e-01 2.99874544e-01
6.01243317e-01 -1.50554210e-01 -8.21234941e-01 2.06800008e+00
5.26589930e-01 1.39164299e-01 -4.18510586e-02 1.05026782e+00
8.68012086e-02 5.58359265e-01 -3.01461786e-01 -2.60171473e-01
8.40808153e-01 -8.05824637e-01 -6.21522486e-01 1.28912032e-01
7.96880960e-01 -7.52135932e-01 7.28693962e-01 4.73230153e-01
-1.15282452e+00 -1.90372065e-01 -1.17641115e+00 -1.85546756e-01
-3.53074729e-01 -2.59370267e-01 4.66335714e-01 6.70306563e-01
-1.25369048e+00 9.95307565e-01 -1.08875799e+00 -6.66942656e-01
1.00919090e-01 2.20125645e-01 -2.71963805e-01 -1.89114586e-02
-1.00849211e+00 8.84929359e-01 -2.04351977e-01 3.61851603e-01
-8.14571977e-01 -3.06569874e-01 -7.48134732e-01 -3.98824483e-01
3.05795342e-01 -9.78110373e-01 8.72449458e-01 -6.07487559e-01
-1.83338380e+00 8.20927978e-01 1.76909044e-02 -3.74479264e-01
9.80705023e-01 -1.29849926e-01 2.98924912e-02 8.61444790e-03
-1.13335922e-01 4.69378352e-01 1.05895615e+00 -9.06065881e-01
-1.02757476e-01 -4.88192469e-01 2.55115569e-01 1.02347024e-01
-2.40146369e-01 -9.59575325e-02 -1.52849734e-01 -8.16510201e-01
2.43386239e-01 -1.28914022e+00 -4.58954901e-01 1.27469689e-01
-4.99386311e-01 -1.37184501e-01 9.16365087e-01 -4.04444575e-01
8.62717748e-01 -2.13305950e+00 9.99925077e-01 2.02416986e-01
3.47232610e-01 1.39363214e-01 -1.75925761e-01 5.53398311e-01
-5.34378327e-02 1.91006079e-01 -7.86091387e-01 -6.57823026e-01
2.68045664e-01 2.32966781e-01 -3.14017266e-01 9.19986129e-01
3.50956380e-01 8.55947375e-01 -9.52324510e-01 -1.11302011e-01
1.09438188e-01 6.95692956e-01 -6.11620963e-01 -8.33004527e-03
-2.83119768e-01 7.65274704e-01 -4.92171943e-01 -9.40628946e-02
6.41304910e-01 -7.93168973e-03 2.04347707e-02 2.75812477e-01
-1.96781024e-01 6.90437481e-02 -1.13197887e+00 2.12006450e+00
-1.75663769e-01 3.08349788e-01 2.85172552e-01 -1.25885653e+00
7.41309762e-01 -2.83147534e-03 6.45137072e-01 -2.51406074e-01
-2.07518060e-02 2.48400345e-01 1.63176030e-01 -4.32019144e-01
3.86526555e-01 -3.37643981e-01 1.23818614e-01 6.47964954e-01
-2.51764894e-01 -4.04080808e-01 2.91346818e-01 2.90648907e-01
1.17890739e+00 3.30079764e-01 -2.23535761e-01 -6.21129274e-01
4.95533586e-01 -2.49831945e-01 3.48537296e-01 5.58643818e-01
4.74517792e-02 9.34284925e-01 5.63269019e-01 -1.86811343e-01
-1.21773756e+00 -9.84514117e-01 -4.07589152e-02 7.55555212e-01
1.22172013e-01 -3.52471262e-01 -1.18829858e+00 -6.26206279e-01
-3.11720073e-01 3.92579287e-01 -7.62923121e-01 -4.07754302e-01
-5.17235935e-01 -1.15772223e+00 3.91697168e-01 8.89480338e-02
2.98451304e-01 -8.66833091e-01 -6.18516743e-01 2.86851108e-01
1.22918122e-01 -1.01436627e+00 -8.16421986e-01 6.21715151e-02
-8.19574893e-01 -1.14282274e+00 -1.14673865e+00 -7.03578711e-01
7.98069239e-01 3.07769835e-01 6.81603968e-01 1.85666382e-02
-2.46235073e-01 5.76264203e-01 -2.30721593e-01 -1.30788296e-01
-5.96169651e-01 1.35602802e-01 9.94995534e-02 4.91366267e-01
1.11205511e-01 -6.42527521e-01 -7.00082600e-01 5.39020956e-01
-1.46194065e+00 1.31330848e-01 3.21596563e-01 6.99686289e-01
4.79937732e-01 -1.25487149e-01 4.08649623e-01 -5.98313451e-01
8.73296976e-01 -5.89451671e-01 -4.71855789e-01 -3.23455371e-02
-5.59890926e-01 4.20087010e-01 3.93400103e-01 -6.79031849e-01
-7.42765725e-01 -2.63654981e-02 -1.68064788e-01 -1.86952353e-01
2.15796560e-01 4.80744958e-01 -1.56981930e-01 -1.39260918e-01
3.88171881e-01 1.42327756e-01 2.21849367e-01 -5.31565547e-01
6.58954024e-01 4.00996298e-01 2.89885342e-01 -5.76958001e-01
9.57438648e-01 9.57705617e-01 4.17954147e-01 -1.09556007e+00
-4.84900326e-01 -2.45977089e-01 -8.52839649e-01 -2.35810094e-02
1.27406001e+00 -1.39482364e-01 -5.22155762e-01 6.17327869e-01
-1.22101486e+00 -5.34061015e-01 -3.68615776e-01 4.05258864e-01
-8.91391814e-01 7.57552385e-01 -4.68177766e-01 -7.89469600e-01
7.70450663e-03 -1.24132395e+00 9.32404399e-01 -1.51738944e-02
-2.90669072e-02 -1.18988383e+00 3.57346833e-01 -8.16925615e-02
5.36528230e-01 4.32734191e-01 9.63975072e-01 -2.96234876e-01
-4.77245629e-01 1.77766114e-01 2.52124459e-01 5.20618021e-01
1.60965651e-01 -1.50072798e-01 -4.77303356e-01 -2.32222855e-01
5.34746945e-01 1.39092207e-01 1.03716397e+00 2.08940014e-01
7.37930000e-01 -7.56626809e-03 -2.93981433e-01 6.45394862e-01
1.05441904e+00 1.91927284e-01 6.74712837e-01 2.18650877e-01
8.69841158e-01 9.49427307e-01 1.24730118e-01 1.05391592e-01
2.01358363e-01 8.11726511e-01 5.58652163e-01 -1.90348160e-02
-1.46036251e-02 -2.47604866e-02 4.81960922e-01 1.12073779e+00
-2.33802527e-01 -1.70636430e-01 -6.48078263e-01 4.41174001e-01
-2.11463356e+00 -7.92971313e-01 -3.39068592e-01 2.45316100e+00
6.46648049e-01 1.48316927e-03 5.83262406e-02 2.36313045e-02
7.70487070e-01 1.42748594e-01 -6.98303878e-01 -3.19912970e-01
-3.34075987e-01 6.54318556e-02 3.27828497e-01 7.61884153e-01
-9.59114671e-01 7.14419901e-01 6.04509783e+00 4.84085649e-01
-1.01919830e+00 3.30951869e-01 4.07686710e-01 8.90014693e-02
-4.84314531e-01 1.45489663e-01 -4.13529813e-01 4.45952803e-01
7.20622599e-01 7.51065016e-02 5.12237191e-01 5.19993722e-01
3.89967412e-01 1.59134239e-01 -1.18980420e+00 7.95587957e-01
5.03102429e-02 -1.05920780e+00 -4.95298430e-02 5.48051417e-01
7.11913824e-01 1.84320081e-02 2.53535897e-01 -2.85760731e-01
3.06711257e-01 -9.51784730e-01 7.26373672e-01 7.55041540e-01
3.62845957e-01 -6.10515356e-01 2.85152107e-01 4.89927769e-01
-8.51662397e-01 3.15256238e-01 -2.74951875e-01 -2.63978145e-04
4.53611612e-01 7.92293370e-01 -1.81355819e-01 5.49870551e-01
1.70040175e-01 7.82341063e-01 -2.26997793e-01 8.12334001e-01
-2.08755165e-01 4.30478901e-01 -4.16965276e-01 4.13261056e-02
2.68564552e-01 -1.12595701e+00 9.75195646e-01 1.08770871e+00
4.34451193e-01 -8.10867026e-02 -2.44004324e-01 1.07527375e+00
4.03561443e-02 -7.57712126e-02 -8.28223646e-01 -5.10905534e-02
-2.25638181e-01 1.04935849e+00 -9.21132922e-01 1.69677705e-01
-4.05523926e-01 1.32343709e+00 1.19009830e-01 6.03929758e-01
-9.80995357e-01 -6.22222662e-01 8.07158649e-01 1.06649362e-01
3.94649595e-01 -9.47159111e-01 2.32645944e-02 -1.30338073e+00
1.98758438e-01 -5.55999875e-01 -9.57760289e-02 -4.87032354e-01
-1.23948002e+00 2.28717849e-01 -1.36712253e-01 -8.06598008e-01
-1.50958896e-01 -6.90215409e-01 -5.97879648e-01 7.10932195e-01
-1.42561603e+00 -8.30899715e-01 2.65330281e-02 5.52785337e-01
4.33224857e-01 2.67972916e-01 5.22500813e-01 2.49491051e-01
-5.01123726e-01 -1.02783509e-01 4.07087266e-01 -8.05038735e-02
5.90655088e-01 -1.39659345e+00 6.78801060e-01 1.04067457e+00
2.55340934e-01 8.70463073e-01 8.32061648e-01 -4.88896251e-01
-1.71937430e+00 -9.14089799e-01 6.84872270e-01 -6.72838032e-01
8.79027307e-01 -7.40185261e-01 -8.34076822e-01 5.81458509e-01
3.27566662e-03 -1.93786956e-02 1.87853560e-01 -3.18660975e-01
-3.32621858e-02 2.82204807e-01 -7.74597883e-01 8.11589479e-01
1.47141516e+00 -3.91909838e-01 -3.06424826e-01 4.90545958e-01
5.88897407e-01 -2.14085385e-01 -8.14872861e-01 2.52447784e-01
3.23096007e-01 -8.91355336e-01 8.62036705e-01 -7.02859342e-01
3.50478411e-01 -3.87310684e-01 -7.76302069e-02 -1.42778063e+00
-1.73901170e-01 -1.48561990e+00 -2.67913103e-01 1.02706122e+00
3.17800343e-01 -8.32746804e-01 4.99284089e-01 5.19806206e-01
-1.47097751e-01 -7.94102192e-01 -1.08280957e+00 -9.16817546e-01
5.67058563e-01 -3.75735998e-01 2.24381268e-01 8.12819958e-01
-1.57059520e-01 2.92328715e-01 -2.89214224e-01 -1.24263488e-01
6.87327683e-01 -1.94617063e-01 6.15248919e-01 -1.10282147e+00
-2.91001320e-01 -6.18985474e-01 -3.26331526e-01 -1.47768557e+00
1.00430697e-01 -8.45753372e-01 9.27921906e-02 -1.49150169e+00
-8.77227448e-03 -4.31887120e-01 2.67856717e-02 -1.74344599e-01
9.77764130e-02 7.03079924e-02 1.25961125e-01 3.94547820e-01
-3.80401105e-01 8.93595815e-01 1.33872175e+00 -6.99808821e-02
-1.87017709e-01 4.76955576e-03 -4.94963080e-01 6.93165779e-01
7.90586531e-01 -4.81373966e-01 -5.08558095e-01 -5.24761915e-01
5.01808286e-01 -4.44606721e-01 3.65189880e-01 -6.66416168e-01
2.21727803e-01 -1.15124144e-01 -2.89748073e-01 1.79666027e-01
2.36862794e-01 -4.80231166e-01 6.15072437e-02 4.30464655e-01
-1.96126178e-01 2.09252045e-01 -3.87614608e-01 8.85290980e-01
1.48396522e-01 -2.92355299e-01 9.19466972e-01 -6.73643500e-02
-1.56621516e-01 3.10251415e-01 -4.53026652e-01 5.96918203e-02
1.17052603e+00 -8.66781250e-02 -2.13526115e-01 -4.47996050e-01
-8.63041639e-01 -9.12962407e-02 7.46719658e-01 5.51642716e-01
5.72400272e-01 -1.09558153e+00 -6.53164148e-01 1.80896565e-01
-2.30564490e-01 1.26365408e-01 -1.16803840e-01 1.35755026e+00
-4.67304558e-01 3.94101620e-01 1.50486812e-01 -6.05921388e-01
-8.63203704e-01 7.84901857e-01 4.20452625e-01 -1.37581542e-01
-7.66477048e-01 4.67972547e-01 6.24527276e-01 -3.96436423e-01
-1.37832776e-01 -4.60517198e-01 2.03478727e-02 -1.93197638e-01
3.70877951e-01 3.97723079e-01 3.56411636e-02 -7.90633678e-01
-1.53540477e-01 7.94265211e-01 1.52170718e-01 -5.34164548e-01
1.37619090e+00 -3.77517462e-01 -8.78023803e-02 3.83929133e-01
1.18969440e+00 3.72242741e-02 -1.50975907e+00 3.52721624e-02
2.30927035e-01 2.82399040e-02 -1.55022725e-01 -1.24197289e-01
-9.20583487e-01 1.03165162e+00 3.18662256e-01 6.73835278e-01
7.83689976e-01 3.67829688e-02 7.60776281e-01 2.38014445e-01
3.68898988e-01 -1.14234293e+00 1.98898450e-01 6.42698526e-01
9.83889341e-01 -9.42906141e-01 -4.59784776e-01 -2.89881736e-01
-3.30985397e-01 1.01185298e+00 -1.28196895e-01 -2.70578146e-01
8.80645633e-01 1.23846456e-01 -1.66312024e-01 -1.74019739e-01
-3.70719045e-01 -2.91988790e-01 9.08788815e-02 6.92169309e-01
2.85803318e-01 -2.34117657e-01 -6.70107961e-01 1.27977759e-01
-1.08545505e-01 -2.08020180e-01 6.98489606e-01 1.12556708e+00
-2.05845833e-01 -1.41751385e+00 -1.97260216e-01 -1.15739420e-01
-4.48939830e-01 -7.26053938e-02 -4.86398250e-01 5.71399271e-01
-1.35525823e-01 1.03436911e+00 -1.52564913e-01 -2.26055965e-01
2.16694832e-01 1.55283794e-01 6.23176396e-01 -5.66607177e-01
-8.49777311e-02 -2.00822186e-02 -2.87765801e-01 -4.61168230e-01
-7.30884910e-01 -7.64910698e-01 -9.97597158e-01 -3.53275418e-01
-3.67732942e-01 2.38245085e-01 1.05637550e+00 1.04678237e+00
5.31391621e-01 3.39173347e-01 4.89680886e-01 -9.20374870e-01
-7.40891576e-01 -1.00636864e+00 -6.93959653e-01 6.26277030e-01
4.86078531e-01 -7.90004551e-01 -5.78754842e-01 2.76473343e-01]
|
[7.391412734985352, 3.8652255535125732]
|
0304d1cb-74b6-4ca9-83c5-a0f474e0e4d6
|
layered-embeddings-for-amodal-instance
|
2002.06264
| null |
https://arxiv.org/abs/2002.06264v1
|
https://arxiv.org/pdf/2002.06264v1.pdf
|
Layered Embeddings for Amodal Instance Segmentation
|
The proposed method extends upon the representational output of semantic instance segmentation by explicitly including both visible and occluded parts. A fully convolutional network is trained to produce consistent pixel-level embedding across two layers such that, when clustered, the results convey the full spatial extent and depth ordering of each instance. Results demonstrate that the network can accurately estimate complete masks in the presence of occlusion and outperform leading top-down bounding-box approaches. Source code available at https://github.com/yanfengliu/layered_embeddings
|
['Lance Pérez', 'Yanfeng Liu', 'Eric Psota']
|
2020-02-14
| null | null | null | null |
['amodal-instance-segmentation']
|
['computer-vision']
|
[-2.65803952e-02 4.26446140e-01 -3.30991000e-01 -6.74300790e-01
-7.58707464e-01 -4.38112974e-01 3.38736713e-01 9.16274861e-02
-1.85035795e-01 4.44328725e-01 2.04035968e-01 -8.86293873e-02
2.79813796e-01 -8.24353397e-01 -8.27592909e-01 -2.60498405e-01
2.12160870e-02 9.11210626e-02 3.20336789e-01 2.68953383e-01
1.50920764e-01 7.38864064e-01 -1.42001832e+00 4.67274994e-01
6.29616201e-01 1.17419231e+00 2.69063026e-01 7.09569573e-01
-3.02739143e-01 4.77083415e-01 -4.87220138e-01 -1.23757489e-01
3.20278347e-01 8.54757428e-02 -6.35162294e-01 4.07042325e-01
9.49852705e-01 -7.32725263e-01 -6.67703509e-01 8.39154482e-01
9.02048424e-02 -1.95996985e-02 6.20116651e-01 -9.38256979e-01
-9.25831616e-01 1.07200287e-01 -6.79768443e-01 1.07138030e-01
1.09058201e-01 2.50708789e-01 1.14403427e+00 -1.20893633e+00
8.13298106e-01 1.17760944e+00 5.00845611e-01 4.79209185e-01
-1.25734854e+00 -5.96925259e-01 6.47142708e-01 -7.40065947e-02
-1.29135537e+00 -2.63264507e-01 8.39905262e-01 -5.15344262e-01
9.25860107e-01 8.10666084e-02 6.42629385e-01 8.17564487e-01
1.51872328e-02 1.00881970e+00 8.81194174e-01 -1.11529119e-01
9.95657742e-02 1.20434321e-01 3.58758032e-01 8.45264196e-01
4.73815024e-01 -1.31152645e-01 -3.68104041e-01 3.23541724e-04
1.05582297e+00 2.50276744e-01 -1.85709476e-01 -5.83011866e-01
-8.43016148e-01 5.91252804e-01 9.67666924e-01 1.87619656e-01
-2.77218908e-01 5.92218816e-01 6.28830791e-02 -4.12958682e-01
8.45966399e-01 2.57657260e-01 -4.66759086e-01 3.55357438e-01
-1.18084681e+00 2.40076244e-01 3.33000332e-01 1.01573873e+00
1.00397718e+00 7.13341832e-02 -1.90564856e-01 6.49551511e-01
3.14699173e-01 2.19238698e-01 2.89420821e-02 -1.43477714e+00
4.18185383e-01 8.96645248e-01 1.35584325e-01 -8.32199097e-01
-2.64283895e-01 -4.06005859e-01 -3.11391115e-01 5.95056474e-01
3.22460741e-01 1.07015766e-01 -1.35828841e+00 1.29849446e+00
4.77402866e-01 3.29232275e-01 -3.64616036e-01 8.79473507e-01
1.03087378e+00 6.20299876e-01 1.49005532e-01 5.64009249e-01
1.31288695e+00 -1.20843303e+00 -6.51486456e-01 -6.53853953e-01
3.41595292e-01 -4.70011324e-01 9.58698273e-01 1.84297845e-01
-1.20170963e+00 -6.70193851e-01 -1.12917626e+00 -7.48379886e-01
-5.45335650e-01 5.21693528e-01 7.11463213e-01 2.85606623e-01
-1.26878989e+00 4.44569349e-01 -1.00513184e+00 6.69858307e-02
1.22091055e+00 1.31107196e-01 -5.49187601e-01 -4.31909621e-01
-6.77015185e-01 5.33380270e-01 2.33748794e-01 3.96935433e-01
-9.80571330e-01 -9.89333093e-01 -1.18112826e+00 1.59454554e-01
-9.28183272e-02 -4.42176819e-01 1.03592217e+00 -6.86423719e-01
-9.61484253e-01 9.62993920e-01 -6.02375925e-01 -1.92005575e-01
5.47364891e-01 -4.41533774e-01 2.72113591e-01 5.22876680e-01
1.63020834e-01 1.20488346e+00 6.69993877e-01 -1.61601853e+00
-6.04272544e-01 -6.28190458e-01 1.92239359e-01 1.56302571e-01
-2.75355726e-01 -3.54140162e-01 -7.49376833e-01 -5.58352947e-01
3.93812209e-01 -4.87220705e-01 -3.95002544e-01 8.86554956e-01
-5.87236404e-01 -5.55081032e-02 1.01213801e+00 -8.36529672e-01
1.00927281e+00 -2.25975180e+00 -6.78992793e-02 -9.07561108e-02
3.40240031e-01 3.87327410e-02 -1.34682700e-01 2.82559574e-01
-5.92649169e-02 3.23122144e-01 -6.55375898e-01 -8.02915514e-01
-1.03884218e-02 2.91929781e-01 -3.47949147e-01 6.52074277e-01
4.63164330e-01 1.15452111e+00 -7.58679569e-01 -4.98342305e-01
7.36459374e-01 8.87088418e-01 -4.53666627e-01 1.49986625e-01
-4.00746793e-01 3.50629032e-01 -5.19122243e-01 9.03717458e-01
8.32322299e-01 -4.35930640e-01 -2.62034144e-02 -1.66485295e-01
-7.35285431e-02 3.31879944e-01 -1.03448224e+00 1.99018860e+00
-4.97043610e-01 1.11286712e+00 2.84280181e-01 -7.89132595e-01
6.19528115e-01 8.91977176e-02 2.59206831e-01 -5.46000957e-01
5.92357926e-02 -9.91469622e-03 -4.58522171e-01 -1.76754594e-01
3.72941434e-01 3.94055814e-01 6.33986806e-03 1.72947198e-01
8.76229405e-02 -2.83735782e-01 -8.21554363e-02 3.76707047e-01
8.00394952e-01 4.15979624e-01 6.80184513e-02 8.48350152e-02
1.50448129e-01 1.02493197e-01 4.67188776e-01 5.30511975e-01
-1.89730853e-01 1.06178963e+00 6.25364363e-01 -4.65382934e-01
-1.06457591e+00 -1.17334723e+00 -5.23081303e-01 7.90702283e-01
3.73949200e-01 -1.03098147e-01 -6.65453196e-01 -5.31482339e-01
4.22646135e-01 5.22709131e-01 -1.04544926e+00 3.39301080e-01
-5.21913826e-01 -1.58957988e-01 3.08097571e-01 9.15679455e-01
4.05326724e-01 -9.54740286e-01 -5.90938210e-01 4.00710292e-02
1.28995821e-01 -1.27413344e+00 -2.26218507e-01 1.35250077e-01
-9.89013970e-01 -1.18284523e+00 -6.19317234e-01 -7.76704311e-01
1.06567299e+00 1.97612822e-01 8.18381548e-01 3.16761047e-01
-7.92992473e-01 2.31796131e-01 -1.40647829e-01 -1.23580828e-01
3.00347567e-01 -1.59273833e-01 -7.20012367e-01 -2.26438105e-01
1.85249299e-01 -3.62498224e-01 -9.36692417e-01 -1.68496203e-02
-8.79967034e-01 3.53868127e-01 3.91361028e-01 5.52154243e-01
8.06569755e-01 -3.76810133e-01 -5.18249758e-02 -9.76518750e-01
1.16496705e-01 -2.07601100e-01 -6.10794008e-01 3.36406007e-02
-1.74952850e-01 -2.53801852e-01 1.95947751e-01 -3.48548889e-02
-8.82103384e-01 2.22713813e-01 -1.49455205e-01 -6.65203393e-01
-5.20017207e-01 -1.40650585e-01 -1.16158433e-01 1.02953926e-01
2.13011488e-01 6.90350458e-02 -2.94872493e-01 -4.89368767e-01
7.50103772e-01 5.68801999e-01 4.18643683e-01 -3.68452281e-01
6.77230000e-01 1.15089166e+00 -2.98634022e-01 -7.22602963e-01
-9.47551191e-01 -4.88628238e-01 -1.03681290e+00 -1.87409684e-01
1.16298163e+00 -9.30969357e-01 -2.30989382e-01 2.96503901e-01
-1.41282904e+00 -6.77388787e-01 -3.81314516e-01 2.26670370e-01
-3.40765357e-01 -2.93472167e-02 -8.03391874e-01 -6.67523563e-01
-4.97648455e-02 -1.13872099e+00 1.45215464e+00 3.29583675e-01
-1.91768900e-01 -1.02790010e+00 -3.10293257e-01 5.71172833e-01
1.67499296e-02 5.05572736e-01 5.74163854e-01 -2.00983465e-01
-1.01134408e+00 -2.75717676e-01 -6.42832458e-01 3.93110693e-01
1.63093448e-01 2.90950030e-01 -1.47801435e+00 3.16500990e-03
-5.00116289e-01 -1.14543967e-01 1.10078812e+00 8.35794032e-01
1.50359881e+00 -1.12976260e-01 -5.96126974e-01 9.26838636e-01
1.62535560e+00 -1.73000544e-01 5.83327770e-01 2.53923684e-01
1.00877428e+00 6.54360175e-01 3.55062425e-01 1.75016657e-01
4.21757817e-01 3.27743113e-01 7.39853323e-01 -6.08183801e-01
-5.02331018e-01 -3.50182861e-01 -1.65632457e-01 5.56807481e-02
4.01891351e-01 -2.94769108e-01 -7.72377908e-01 8.51619005e-01
-1.62869823e+00 -7.14375675e-01 -1.80354118e-01 1.85701442e+00
5.42541742e-01 1.24037363e-01 -4.09315079e-01 -4.43533920e-02
6.47749782e-01 7.35825002e-01 -8.29017162e-01 -6.50939167e-01
8.45236778e-02 3.01595569e-01 7.45421231e-01 8.88306558e-01
-1.21933591e+00 1.14018607e+00 6.61512470e+00 5.92839837e-01
-9.49785352e-01 3.08995783e-01 1.02256203e+00 -2.35713482e-01
-5.81427395e-01 -1.59153685e-01 -7.16636240e-01 3.50750238e-01
4.65851814e-01 1.48503721e-01 -8.55093971e-02 8.96217227e-01
2.13045254e-01 -2.89787292e-01 -8.75756681e-01 5.23701906e-01
7.73272011e-04 -1.54932678e+00 -3.17035876e-02 1.48107886e-01
1.07082069e+00 1.28119349e-01 3.20141077e-01 1.05221802e-02
2.11205825e-01 -1.25674498e+00 7.84478486e-01 2.17509910e-01
9.54399526e-01 -3.62709969e-01 3.64943266e-01 4.76217223e-03
-1.39873075e+00 -1.91412717e-01 -3.27296942e-01 -1.09306514e-01
2.62994707e-01 5.87553561e-01 -6.46364927e-01 3.10624301e-01
6.86289489e-01 8.89845550e-01 -5.45708597e-01 8.58143568e-01
-6.96960449e-01 4.22220796e-01 -2.03525260e-01 4.68673915e-01
4.41965729e-01 -1.24446660e-01 1.54306993e-01 1.17816627e+00
2.79733986e-02 1.25301793e-01 5.88090764e-03 1.26294219e+00
-2.47389138e-01 -2.38702461e-01 -6.69396281e-01 7.16180727e-02
5.90577543e-01 1.28755403e+00 -8.74415219e-01 -4.35769200e-01
-5.09622514e-01 1.16647112e+00 7.13930070e-01 7.89749086e-01
-8.72463346e-01 -1.86992049e-01 1.10553634e+00 2.59226322e-01
6.60066366e-01 -4.40915495e-01 -1.01253414e+00 -7.72470891e-01
2.27420315e-01 7.55630061e-02 2.22257257e-01 -1.05972373e+00
-8.26903939e-01 4.34235036e-01 -1.81393236e-01 -1.02596021e+00
3.49671751e-01 -9.86400187e-01 -7.45254159e-01 9.98300910e-01
-1.86544645e+00 -1.22147989e+00 -5.16459227e-01 1.14323549e-01
7.74187982e-01 5.13323367e-01 7.50783443e-01 1.57596231e-01
-6.50793374e-01 2.55470365e-01 -5.35270534e-02 3.99011195e-01
2.77286410e-01 -1.31561661e+00 4.02697176e-01 8.29627991e-01
2.19127744e-01 5.58517814e-01 3.04887712e-01 -5.04210591e-01
-8.73360515e-01 -1.28493726e+00 6.74195051e-01 -5.73331356e-01
3.80319446e-01 -6.33795023e-01 -9.96735752e-01 8.80118191e-01
1.23315193e-01 5.78567207e-01 2.47776568e-01 -1.85472980e-01
-4.73656863e-01 1.59750599e-02 -1.18095481e+00 4.89757985e-01
9.23587859e-01 -6.81232452e-01 -4.80315655e-01 3.82095963e-01
8.31848562e-01 -7.26577878e-01 -6.61878049e-01 1.88515380e-01
4.94375080e-01 -1.05183351e+00 1.23330224e+00 -3.97731036e-01
7.13302374e-01 -2.94817090e-01 -8.29598233e-02 -8.06649923e-01
-1.50015041e-01 1.28439605e-01 -2.29272008e-01 7.79749155e-01
6.21027052e-01 -5.92005253e-01 1.13183999e+00 8.28586638e-01
-3.55225444e-01 -1.28023362e+00 -8.82505357e-01 -4.10754055e-01
1.90405145e-01 -5.48182070e-01 4.14987832e-01 6.64476275e-01
-3.93326730e-01 -4.30981278e-01 1.95957556e-01 6.80797696e-01
6.06829047e-01 2.85496831e-01 5.20172596e-01 -8.77160251e-01
1.85549706e-01 -3.56278181e-01 -6.12824857e-01 -1.31077886e+00
2.69908965e-01 -8.17977965e-01 -1.51786339e-02 -2.30641723e+00
-6.78868368e-02 -3.22300673e-01 -4.29550171e-01 5.60743988e-01
-2.41900831e-01 7.47293472e-01 3.88262719e-02 5.22047579e-02
-4.34552878e-01 4.51231122e-01 1.39442980e+00 -2.57288665e-01
1.06758647e-01 -4.44696486e-01 -6.01552606e-01 6.92055166e-01
9.06377554e-01 -4.24647599e-01 -3.49596292e-01 -8.38482559e-01
-2.51557201e-01 -2.89946347e-01 7.50679791e-01 -8.57284009e-01
6.65846169e-02 -1.37844339e-01 9.25109386e-01 -6.46771193e-01
7.91831970e-01 -8.15365672e-01 -1.20593801e-01 2.91893452e-01
-4.67151225e-01 -2.23830193e-01 4.64117467e-01 5.35593510e-01
-1.69960663e-01 -1.24983005e-02 6.72822714e-01 -1.46870986e-01
-8.83544028e-01 6.21719778e-01 3.49212587e-02 -1.69356540e-02
1.24112797e+00 -6.61920488e-01 -3.55548233e-01 -1.06054075e-01
-8.10528517e-01 3.66558969e-01 7.15202391e-01 3.64179045e-01
8.35580170e-01 -1.01689649e+00 -3.37289125e-01 3.11311662e-01
3.83168757e-02 4.84252572e-01 3.79186660e-01 4.96379644e-01
-9.49710131e-01 4.16907817e-01 -1.84003972e-02 -6.67532027e-01
-1.01979542e+00 1.12241462e-01 4.66727525e-01 2.15548858e-01
-1.02131128e+00 1.29570568e+00 6.84167802e-01 -4.40073788e-01
2.94261873e-01 -5.60138404e-01 4.00186256e-02 -1.37856856e-01
4.80249763e-01 7.32320175e-02 -2.16584846e-01 -6.28428280e-01
-3.76231790e-01 7.09734619e-01 4.38056923e-02 -1.49830684e-01
1.39788413e+00 -7.76055306e-02 -1.12614017e-02 4.57143068e-01
1.42883754e+00 -1.90558925e-01 -1.94057631e+00 -6.74326671e-03
-3.39649051e-01 -8.64468813e-01 3.75694096e-01 -4.74633962e-01
-1.28967714e+00 1.31854141e+00 4.95720536e-01 -2.04318643e-01
6.77082777e-01 2.79754609e-01 7.03450739e-01 -1.84632480e-01
1.03635401e-01 -8.69172812e-01 -2.76745204e-02 1.24066874e-01
7.76880920e-01 -1.24087811e+00 3.22625875e-01 -6.30828023e-01
-5.18402755e-01 9.71158922e-01 8.43362451e-01 -5.83215356e-01
7.10484803e-01 3.60620767e-01 2.17982903e-01 -4.38962668e-01
-6.51058674e-01 -2.50843227e-01 3.22744220e-01 6.77520752e-01
5.15148938e-01 1.32708281e-01 1.65442321e-02 9.15110260e-02
1.11261792e-01 -3.30381602e-01 2.58631378e-01 8.20048392e-01
-4.73858505e-01 -6.83165252e-01 -8.88728499e-02 4.81875390e-01
-1.73659980e-01 -1.33745493e-02 -2.76112825e-01 1.01955807e+00
3.21561396e-01 6.16535246e-01 5.84083974e-01 1.28916383e-01
2.49623775e-01 -4.15967554e-02 3.83728385e-01 -1.00316918e+00
-1.65737018e-01 -1.76333532e-01 -2.16856509e-01 -9.72273231e-01
-1.75116047e-01 -4.71598953e-01 -1.59354508e+00 1.53914168e-01
-1.19526707e-01 -2.57369101e-01 7.87476659e-01 5.24784505e-01
3.74609947e-01 7.59135187e-01 4.05672103e-01 -1.33325255e+00
1.94734663e-01 -8.12669277e-01 -5.10401011e-01 2.97200173e-01
6.38403475e-01 -5.88004112e-01 -5.53035676e-01 1.50846377e-01]
|
[8.073039054870605, -3.1100800037384033]
|
61cd3f77-6a54-45e1-a4b3-2790416ea5ee
|
open-world-semi-supervised-novel-class
|
2305.13095
| null |
https://arxiv.org/abs/2305.13095v1
|
https://arxiv.org/pdf/2305.13095v1.pdf
|
Open-world Semi-supervised Novel Class Discovery
|
Traditional semi-supervised learning tasks assume that both labeled and unlabeled data follow the same class distribution, but the realistic open-world scenarios are of more complexity with unknown novel classes mixed in the unlabeled set. Therefore, it is of great challenge to not only recognize samples from known classes but also discover the unknown number of novel classes within the unlabeled data. In this paper, we introduce a new open-world semi-supervised novel class discovery approach named OpenNCD, a progressive bi-level contrastive learning method over multiple prototypes. The proposed method is composed of two reciprocally enhanced parts. First, a bi-level contrastive learning method is introduced, which maintains the pair-wise similarity of the prototypes and the prototype group levels for better representation learning. Then, a reliable prototype similarity metric is proposed based on the common representing instances. Prototypes with high similarities will be grouped progressively for known class recognition and novel class discovery. Extensive experiments on three image datasets are conducted and the results show the effectiveness of the proposed method in open-world scenarios, especially with scarce known classes and labels.
|
['Junming Shao', 'Qinli Yang', 'Yulu Fan', 'Tongze Zhang', 'Yangqiming Wang', 'Jiaming Liu']
|
2023-05-22
| null | null | null | null |
['novel-class-discovery', 'novel-class-discovery']
|
['computer-vision', 'methodology']
|
[ 2.81618655e-01 -1.11781597e-01 -3.20866972e-01 -5.70811629e-01
-6.88448429e-01 -5.71802139e-01 5.78827977e-01 3.29074144e-01
-1.59321800e-01 6.91667855e-01 -1.38690367e-01 2.12270111e-01
-2.60887802e-01 -5.89544773e-01 -3.16309363e-01 -9.24992204e-01
-1.41424136e-02 6.21119738e-01 3.65887552e-01 2.50866354e-01
2.91299254e-01 4.21331406e-01 -2.00266886e+00 3.40551555e-01
9.59863007e-01 1.12795341e+00 1.12897970e-01 6.21002465e-02
-4.99809802e-01 6.18162394e-01 -3.86597186e-01 7.02284202e-02
5.03409505e-01 -4.15024549e-01 -8.13508749e-01 5.78066468e-01
5.09535074e-01 9.38844904e-02 6.08171895e-02 1.15311909e+00
4.45966810e-01 3.22627962e-01 8.30666661e-01 -1.24890161e+00
-5.54071546e-01 5.23834884e-01 -8.25785220e-01 2.76845545e-01
7.42230043e-02 -1.11923233e-01 7.66315281e-01 -1.12949562e+00
6.43855631e-01 1.33376527e+00 4.38562274e-01 3.66866350e-01
-9.72198486e-01 -9.39223588e-01 4.15858954e-01 5.92778504e-01
-1.87188256e+00 -3.10366839e-01 9.30843353e-01 -4.35362995e-01
1.89799890e-01 7.30546191e-02 3.76469284e-01 4.97682422e-01
-4.17623490e-01 8.79587352e-01 1.35991728e+00 -5.27846754e-01
4.01271522e-01 7.94473827e-01 4.46654558e-01 4.63892341e-01
2.89867818e-01 9.33077037e-02 -1.35790795e-01 -6.48096129e-02
2.37736955e-01 4.45410877e-01 -3.32879901e-01 -7.87005305e-01
-1.21879196e+00 6.98224664e-01 6.42301202e-01 5.11190593e-01
-1.88628346e-01 -8.75094175e-01 3.78703892e-01 2.54617959e-01
4.86859262e-01 8.97681341e-02 -4.30573434e-01 3.27606708e-01
-1.05424178e+00 -1.33875728e-01 5.39388180e-01 1.08036590e+00
1.16668379e+00 -1.40278414e-01 5.60852624e-02 1.05970824e+00
4.29011554e-01 3.54899734e-01 9.74542141e-01 -2.62728453e-01
3.08688194e-01 9.50597405e-01 -4.62850481e-02 -1.06625783e+00
-2.47307673e-01 -7.76337326e-01 -6.34476542e-01 2.36391071e-02
1.24232404e-01 1.03415497e-01 -9.72247303e-01 1.32012498e+00
7.86498487e-01 7.07568467e-01 4.59505647e-01 7.33713627e-01
1.03280044e+00 8.59073043e-01 -1.34409696e-01 -7.23963559e-01
8.89157772e-01 -1.16926491e+00 -6.10228360e-01 3.70690823e-02
4.66294289e-01 -7.63935447e-01 8.25036883e-01 2.23429680e-01
-5.84545553e-01 -9.15296197e-01 -1.40321887e+00 4.17993635e-01
-5.62502265e-01 1.01721101e-01 2.87276208e-01 5.64727962e-01
-3.52207720e-01 4.45643753e-01 -3.62497419e-01 -3.60999137e-01
5.38175285e-01 2.09519297e-01 -2.93267161e-01 -5.31813860e-01
-8.58004928e-01 5.54916620e-01 1.02337754e+00 9.24605355e-02
-7.64359832e-01 -5.06200969e-01 -8.16231430e-01 -8.49810690e-02
4.34254348e-01 1.17068261e-01 9.88189042e-01 -1.18546796e+00
-9.38282788e-01 9.25531089e-01 2.04030052e-02 -3.27308051e-04
3.22947979e-01 3.58166546e-01 -9.38657820e-01 7.94964805e-02
1.92084908e-01 3.89467061e-01 8.18650484e-01 -1.67675531e+00
-1.06905723e+00 -6.19021773e-01 -2.87933946e-01 4.19189036e-01
-5.61170876e-01 -3.34045947e-01 -2.97349654e-02 -3.81883979e-01
6.59225702e-01 -6.84478879e-01 -1.63788591e-02 -6.13785274e-02
-1.43547922e-01 -3.91610056e-01 1.28735995e+00 -9.57749262e-02
1.04350972e+00 -2.31098747e+00 -2.61235952e-01 4.50270265e-01
2.24588558e-01 6.64497554e-01 -1.79455280e-01 1.69211775e-01
-3.30029994e-01 -3.19219708e-01 -3.67232233e-01 1.23351276e-01
-2.63103127e-01 1.97635427e-01 -2.25145474e-01 3.80818486e-01
1.45871386e-01 4.53029722e-01 -1.12251627e+00 -8.24905634e-01
1.83960319e-01 8.68398044e-03 1.88315734e-01 2.52470106e-01
7.04888478e-02 4.23494458e-01 -4.38894629e-01 9.04681623e-01
1.02652705e+00 -2.56732225e-01 1.34526730e-01 -2.10290864e-01
2.74983421e-02 -4.95986402e-01 -1.84331632e+00 1.23885465e+00
-1.23912424e-01 8.68722796e-02 -2.14060992e-01 -1.33288872e+00
1.37924886e+00 1.71639726e-01 2.69032866e-01 -4.29100275e-01
2.10211575e-01 5.41386247e-01 1.14260882e-01 -5.15524387e-01
1.82147279e-01 -3.06085438e-01 3.61190438e-01 4.88397628e-01
1.83792040e-01 5.91638908e-02 3.25996935e-01 8.92959610e-02
4.97610390e-01 -5.07119969e-02 5.05618691e-01 -3.05781186e-01
8.38117480e-01 9.73187909e-02 8.09729517e-01 4.05812919e-01
-4.00585473e-01 6.11229122e-01 -1.40821517e-01 -5.64227521e-01
-6.59930885e-01 -1.23751426e+00 -5.08346796e-01 6.63590670e-01
5.70799172e-01 4.07138504e-02 -3.55009347e-01 -1.03354633e+00
-8.99633765e-02 3.26670349e-01 -4.40806389e-01 -3.09438884e-01
-1.57537088e-01 -5.77310681e-01 1.44881040e-01 3.07142138e-01
6.23733163e-01 -1.04808116e+00 -1.74582303e-01 1.70609459e-01
-4.36893851e-02 -7.99063802e-01 -3.21794808e-01 8.44775960e-02
-9.18644369e-01 -1.36033750e+00 -9.17209387e-01 -1.61099219e+00
1.13289130e+00 6.27312541e-01 6.56818867e-01 7.53387213e-02
-3.00045639e-01 9.11393613e-02 -7.42997944e-01 -2.55321652e-01
-3.72307926e-01 -2.41658270e-01 2.31172487e-01 6.41678214e-01
6.63803756e-01 -5.14137566e-01 -2.58521438e-01 6.10294759e-01
-9.21346784e-01 -2.04422817e-01 7.08984911e-01 1.11370540e+00
9.45520997e-01 4.66554165e-01 8.71411324e-01 -9.62245762e-01
2.72381842e-01 -7.20092475e-01 -4.50377792e-01 6.18845284e-01
-8.00661922e-01 -2.02144742e-01 7.23939002e-01 -8.84388149e-01
-1.10984766e+00 1.86708108e-01 4.05508727e-01 -5.17274976e-01
-3.39865029e-01 5.67803562e-01 -4.67495501e-01 -1.37894914e-01
7.46107459e-01 3.21736217e-01 -9.89204720e-02 -3.83643001e-01
4.04757351e-01 1.16901982e+00 5.31840265e-01 -5.12425125e-01
9.98662770e-01 4.21682537e-01 -2.82814056e-01 -7.23644197e-01
-8.35922360e-01 -8.91962409e-01 -8.50559354e-01 -1.22516915e-01
2.71582782e-01 -9.88871217e-01 -8.53027403e-02 5.09024143e-01
-5.63012898e-01 2.32757568e-01 -6.30562782e-01 7.49639988e-01
-1.95388541e-01 6.95749819e-01 -1.69775948e-01 -7.55187511e-01
-2.72522807e-01 -1.03675759e+00 7.02838361e-01 7.57943928e-01
2.80324012e-01 -7.46362090e-01 8.48328620e-02 2.57010818e-01
1.74328089e-02 1.55012280e-01 8.16760480e-01 -1.38962030e+00
-3.87541950e-01 -3.85677576e-01 -4.29209262e-01 5.01357138e-01
5.44472873e-01 -1.33347318e-01 -9.16833222e-01 -4.31005120e-01
4.71975282e-02 -6.16817355e-01 6.48872137e-01 -1.05598785e-01
9.41747427e-01 -1.16460279e-01 -5.27929187e-01 3.52014065e-01
1.31776154e+00 5.41618526e-01 2.28131875e-01 3.71783003e-02
7.31516123e-01 7.47795820e-01 1.13614357e+00 4.59307283e-01
-9.15687997e-03 3.60475808e-01 1.20712966e-01 -3.31044309e-02
4.36908305e-02 -1.69356197e-01 -1.41831577e-01 1.22094750e+00
4.04483050e-01 2.68563349e-02 -8.20379734e-01 6.19573116e-01
-1.75622070e+00 -9.55906034e-01 -1.05465921e-02 2.28207541e+00
7.79681563e-01 1.84093311e-01 -7.62804896e-02 5.26967704e-01
1.32541811e+00 -1.51849732e-01 -9.00240898e-01 -2.16017701e-02
-2.51971692e-01 1.18898191e-01 -1.02314629e-01 7.60892853e-02
-1.18647170e+00 5.28893352e-01 5.33277321e+00 1.19030869e+00
-1.09648037e+00 -3.08420919e-02 6.15146339e-01 2.32888475e-01
3.73106562e-02 1.09035768e-01 -8.31787050e-01 4.90102738e-01
4.76585716e-01 -3.68882984e-01 -1.29317492e-02 1.00764358e+00
-1.51722014e-01 -1.58895273e-02 -1.12060916e+00 1.36234212e+00
3.65159571e-01 -9.96223629e-01 1.99819386e-01 -2.43474558e-01
1.12940311e+00 -3.15239578e-01 -2.92358249e-02 3.92240345e-01
6.17405139e-02 -4.10475820e-01 4.64192837e-01 3.14096898e-01
6.89313173e-01 -9.15195584e-01 7.76148379e-01 5.74360430e-01
-1.51199973e+00 -3.19276899e-01 -4.80615109e-01 8.53989050e-02
-1.70146719e-01 5.44010103e-01 -7.87362754e-01 7.03193903e-01
4.93615985e-01 1.19576228e+00 -7.62419939e-01 1.43546891e+00
-2.94966623e-04 5.01339018e-01 -1.38466164e-01 5.10979854e-02
2.65654802e-01 -2.74562120e-01 1.36443555e-01 7.91657388e-01
1.14405796e-01 2.98106700e-01 7.25369275e-01 5.26563525e-01
-3.96356173e-03 5.30071080e-01 -4.02337462e-01 2.32315753e-02
7.28231430e-01 1.35523725e+00 -1.06875467e+00 -6.89135373e-01
-4.65779871e-01 6.23029172e-01 1.57110780e-01 1.55979261e-01
-6.72065675e-01 -6.54600382e-01 -2.03995425e-02 -1.33297190e-01
3.17808956e-01 2.71756470e-01 8.56309533e-02 -1.19360268e+00
1.34105995e-01 -7.97572613e-01 8.46478701e-01 -5.50229907e-01
-1.71431613e+00 6.03320658e-01 6.27209544e-02 -1.87205100e+00
7.60009885e-03 -3.72075856e-01 -6.52551770e-01 6.37146771e-01
-1.61322784e+00 -1.15475810e+00 -5.04001737e-01 8.72663558e-01
7.62469828e-01 -4.07157093e-01 6.53171301e-01 4.67914551e-01
-5.13565958e-01 7.20883429e-01 7.14701593e-01 7.46925622e-02
7.02957392e-01 -9.86909091e-01 -1.89342827e-01 7.90838420e-01
2.50899106e-01 5.21003962e-01 2.37439051e-01 -6.12913787e-01
-8.40338647e-01 -1.25919461e+00 5.85287690e-01 1.85538843e-01
4.86119241e-01 -1.31589234e-01 -1.21269405e+00 4.63068575e-01
-2.47030228e-01 5.52818477e-01 8.57906640e-01 -1.26975715e-01
-4.63445991e-01 -4.28178638e-01 -1.41590905e+00 1.63182586e-01
8.40862572e-01 -3.18383485e-01 -9.39209402e-01 5.30646265e-01
6.60321653e-01 -7.19262585e-02 -6.87746406e-01 6.77179694e-01
4.33832049e-01 -6.02994978e-01 6.98978841e-01 -4.72597629e-01
-6.69352785e-02 -7.89255321e-01 -2.14309484e-01 -1.31489611e+00
-2.10665286e-01 3.95294502e-02 -2.64424877e-03 1.53419673e+00
2.59443283e-01 -7.21756339e-01 8.50862443e-01 1.40914440e-01
-5.63831180e-02 -6.66237950e-01 -7.34268188e-01 -1.20516586e+00
-1.29189342e-01 2.63930321e-01 4.92704988e-01 1.47251844e+00
8.03832859e-02 4.61815298e-01 -6.57123849e-02 9.49308500e-02
6.82500958e-01 6.42208815e-01 4.98164415e-01 -1.60468137e+00
-8.02339539e-02 -1.37528121e-01 -8.27416480e-01 -8.11437130e-01
6.99098110e-02 -1.00485277e+00 2.04915062e-01 -1.15230858e+00
5.69037080e-01 -8.60393405e-01 -6.24109924e-01 3.28865260e-01
-2.61486858e-01 1.78753465e-01 -5.98860122e-02 5.87555945e-01
-8.36691916e-01 6.58094347e-01 1.03171241e+00 -4.47635412e-01
-4.04096633e-01 1.39992028e-01 -5.80546618e-01 7.38084197e-01
6.28741384e-01 -4.25341547e-01 -8.10392618e-01 -4.20240201e-02
-5.90323091e-01 -3.43736529e-01 -1.19988225e-01 -1.42889154e+00
2.60042757e-01 -1.47385433e-01 4.18005377e-01 -8.51898968e-01
-1.57796834e-02 -1.07281244e+00 1.44442469e-01 4.93315607e-01
-3.62741917e-01 -4.04713362e-01 -5.04198745e-02 7.28793144e-01
-5.14528871e-01 -3.23329866e-01 1.11995256e+00 -9.04532894e-02
-1.22755146e+00 5.11459053e-01 1.72339499e-01 1.76544756e-01
1.50441146e+00 -6.58005595e-01 -2.93751419e-01 1.30449504e-01
-9.16261733e-01 3.42726499e-01 2.84009576e-01 5.58138192e-01
8.00714254e-01 -1.45870042e+00 -6.85735822e-01 2.00258136e-01
8.30528796e-01 9.48972255e-02 3.94191593e-01 4.35870439e-01
-2.90939987e-01 2.32963040e-02 -2.55695373e-01 -7.43722260e-01
-1.36755347e+00 1.13756990e+00 1.63371280e-01 -7.42137758e-03
-3.28095108e-01 6.59515798e-01 2.17976585e-01 -7.47129261e-01
4.01115298e-01 2.08090469e-01 -5.76747894e-01 2.90749699e-01
8.54334831e-01 3.24507445e-01 9.82700437e-02 -1.07123923e+00
-1.99807063e-01 7.61478126e-01 -4.26123440e-01 4.22281653e-01
1.12358809e+00 -2.72276580e-01 -7.37412274e-02 7.70852089e-01
1.43098581e+00 -3.41641188e-01 -1.01958573e+00 -9.20965493e-01
5.24831638e-02 -5.96601784e-01 -3.40128779e-01 -7.51190305e-01
-9.94161308e-01 8.49040389e-01 1.21790707e+00 -8.39129686e-02
1.16561127e+00 9.78282839e-03 6.34116888e-01 4.69012290e-01
5.47153413e-01 -1.07847166e+00 1.06857426e-01 1.78292260e-01
4.05600399e-01 -1.45690262e+00 1.03470027e-01 -6.07463241e-01
-4.66301560e-01 9.69100654e-01 8.95934343e-01 6.74238876e-02
8.99259984e-01 -2.68058628e-01 2.62934297e-01 -1.00483522e-01
-3.22303414e-01 -3.19282353e-01 2.40533978e-01 7.68411100e-01
2.56853960e-02 4.49004211e-02 -4.48077708e-01 4.68759269e-01
2.65826166e-01 -2.32473701e-01 4.26424921e-01 1.28185749e+00
-8.31129074e-01 -9.56787050e-01 -5.87980509e-01 5.92605174e-01
1.33447602e-01 1.53246537e-01 -2.14990661e-01 5.00603974e-01
6.07887328e-01 1.13878500e+00 2.96527334e-02 -5.64233184e-01
2.28917167e-01 8.47157687e-02 1.92846775e-01 -9.08157110e-01
-6.43854588e-02 -1.88206211e-01 -4.58682060e-01 -3.07860747e-02
-7.22927392e-01 -5.99265516e-01 -1.30596948e+00 1.84276566e-01
-9.09167886e-01 6.15313470e-01 4.28018570e-01 9.08148110e-01
1.26577646e-01 1.19887903e-01 1.27781689e+00 -6.33902907e-01
-6.65633082e-01 -7.74805009e-01 -1.07168424e+00 6.24598861e-01
8.07083473e-02 -9.81362760e-01 -4.76820141e-01 3.48186374e-01]
|
[9.731759071350098, 2.9180076122283936]
|
ab4a210d-1f12-4129-b47b-d4679c077549
|
low-rankness-of-complex-valued-spectrogram
|
1903.05603
| null |
http://arxiv.org/abs/1903.05603v1
|
http://arxiv.org/pdf/1903.05603v1.pdf
|
Low-rankness of Complex-valued Spectrogram and Its Application to Phase-aware Audio Processing
|
Low-rankness of amplitude spectrograms has been effectively utilized in audio
signal processing methods including non-negative matrix factorization. However,
such methods have a fundamental limitation owing to their amplitude-only
treatment where the phase of the observed signal is utilized for resynthesizing
the estimated signal. In order to address this limitation, we directly treat a
complex-valued spectrogram and show a complex-valued spectrogram of a sum of
sinusoids can be approximately low-rank by modifying its phase. For evaluating
the applicability of the proposed low-rank representation, we further propose a
convex prior emphasizing harmonic signals, and it is applied to audio
denoising.
|
[]
|
2019-03-13
| null | null | null | null |
['audio-signal-processing', 'audio-denoising']
|
['audio', 'audio']
|
[ 5.89773417e-01 -6.87038302e-02 1.78675070e-01 1.31510451e-01
-8.79119992e-01 -6.02118254e-01 5.25160059e-02 -1.25462160e-01
-2.04828292e-01 6.99221432e-01 4.75026071e-01 5.52972332e-02
-4.25792158e-01 -2.33688191e-01 -4.27429616e-01 -8.97307575e-01
-2.67389596e-01 -4.17500943e-01 -3.22689384e-01 -2.55679309e-01
-4.42216806e-02 1.94746912e-01 -1.60215998e+00 3.40749115e-01
1.00635016e+00 7.48811364e-01 1.34962395e-01 7.21085668e-01
4.50957596e-01 2.80876845e-01 -8.98413718e-01 -8.39920193e-02
3.12773585e-01 -5.35698295e-01 -1.93379715e-01 2.29125828e-01
1.84543550e-01 -7.31760412e-02 -2.85209298e-01 1.41809976e+00
3.92677009e-01 2.96233416e-01 6.15688264e-01 -1.20407367e+00
-1.91939905e-01 6.39801443e-01 -7.01212585e-01 1.04008473e-01
6.78023577e-01 -4.47609484e-01 1.17929602e+00 -1.18213475e+00
1.25346005e-01 1.25445116e+00 8.54968548e-01 3.59479547e-03
-1.26365757e+00 -7.29406238e-01 1.13249898e-01 2.10403323e-01
-1.51496005e+00 -6.11166120e-01 1.26110017e+00 -3.30638468e-01
5.61750531e-01 7.01855958e-01 7.29463577e-01 5.77684402e-01
1.77938372e-01 6.97602391e-01 1.12271285e+00 -5.93677163e-01
-3.84918340e-02 -4.77502733e-01 -1.32284490e-02 2.83991665e-01
3.65889668e-02 -3.17803361e-02 -8.75049174e-01 -4.26525295e-01
5.77999651e-01 -2.52059758e-01 -8.10638428e-01 7.16857091e-02
-1.32537794e+00 5.56358814e-01 -2.15768851e-02 4.37494755e-01
-6.37226999e-01 1.76065698e-01 2.46106014e-01 2.83932090e-01
5.51029682e-01 5.08307219e-01 -1.40598744e-01 -3.66409957e-01
-1.11693096e+00 1.94916412e-01 5.20686448e-01 4.22752768e-01
5.51194370e-01 7.53989816e-01 1.16750719e-02 9.32628095e-01
3.73441190e-01 6.62947416e-01 5.71256995e-01 -9.01403844e-01
3.99380922e-01 7.62258470e-02 2.51110494e-01 -1.41289258e+00
-4.28395301e-01 -3.93940359e-01 -1.22934747e+00 -8.87747258e-02
2.50510514e-01 -1.49130762e-01 -4.64147866e-01 1.72413278e+00
2.67493129e-01 6.74536943e-01 1.02991581e-01 1.02411461e+00
6.01560831e-01 9.29695785e-01 -2.34485120e-01 -9.79191959e-01
1.23398376e+00 -5.16866863e-01 -1.27797592e+00 1.63167641e-02
-5.01330048e-02 -1.17894566e+00 1.03139091e+00 1.02686679e+00
-1.16220093e+00 -5.00420332e-01 -1.31629789e+00 1.00440495e-01
2.64152616e-01 4.24538523e-01 4.05998230e-01 6.89338088e-01
-8.42115223e-01 6.15994751e-01 -5.73221743e-01 4.32141393e-01
-2.00365707e-01 1.73676088e-01 -4.70295489e-01 3.79972816e-01
-1.41167438e+00 3.76472890e-01 5.44140376e-02 4.91069347e-01
-4.83490825e-01 -9.51531768e-01 -8.36500525e-01 2.79204965e-01
1.81836486e-01 -3.82761687e-01 1.01395142e+00 -7.11610079e-01
-1.60693383e+00 2.17854574e-01 -4.13342983e-01 -2.03179762e-01
1.91642582e-01 -4.00311559e-01 -7.76196480e-01 3.42376024e-01
-5.04099540e-02 -2.25344390e-01 1.90730762e+00 -1.10684979e+00
-2.82204032e-01 -1.93483874e-01 -1.05777942e-01 1.78212002e-01
-5.94499230e-01 -2.40171567e-01 -4.03021723e-02 -1.37550354e+00
8.34059536e-01 -9.66479421e-01 -2.59147942e-01 -3.88232201e-01
-4.13659424e-01 2.95006633e-01 7.16186225e-01 -9.04280543e-01
1.56211269e+00 -2.68920994e+00 2.52263814e-01 3.45689327e-01
1.17042802e-01 3.93857658e-02 -1.98742807e-01 6.23572707e-01
-4.89946395e-01 -7.04943612e-02 -2.15112239e-01 -2.88421631e-01
-1.43117547e-01 -6.57660291e-02 -8.62248838e-01 6.57679498e-01
3.97412926e-01 9.69805419e-02 -9.34421182e-01 -2.19408169e-01
1.72725916e-01 8.44995260e-01 -7.29151905e-01 7.85294920e-02
3.88885140e-01 4.68648344e-01 -2.11514995e-01 2.48756051e-01
8.04975569e-01 1.89257681e-01 3.16676348e-01 -1.11154246e+00
-2.87816882e-01 2.59075254e-01 -1.82691634e+00 1.48283470e+00
-6.31737173e-01 5.75302243e-01 6.20856345e-01 -8.47617090e-01
6.83830202e-01 6.44001961e-01 9.59291637e-01 -1.82988763e-01
-2.32679117e-03 9.13756266e-02 1.28719121e-01 -3.55650693e-01
7.18728065e-01 -3.62536550e-01 2.99131244e-01 2.98317313e-01
-3.01596552e-01 -4.36343402e-01 3.06609988e-01 1.27102748e-01
9.12566781e-01 -4.79204841e-02 4.27445441e-01 -2.10300952e-01
9.01227534e-01 -4.99318928e-01 7.23292291e-01 2.09898263e-01
4.06814590e-02 8.23146999e-01 3.21649700e-01 2.42299870e-01
-7.00775504e-01 -9.16027367e-01 -2.33377680e-01 1.12667167e+00
-2.18335450e-01 -7.60801911e-01 -6.11163199e-01 -1.82086183e-03
-1.21616296e-01 3.77819866e-01 -3.03758264e-01 -1.16831928e-01
-6.30825698e-01 -7.35001147e-01 4.95429873e-01 2.31272846e-01
1.99713603e-01 -3.39070797e-01 -3.46513182e-01 3.26788753e-01
-6.30828619e-01 -9.51496899e-01 -7.39932597e-01 1.64184332e-01
-7.82728314e-01 -9.13426697e-01 -7.00632215e-01 -4.95434165e-01
5.77246070e-01 5.27056277e-01 7.17561603e-01 -1.11847803e-01
-6.38021305e-02 6.01907313e-01 -3.81685764e-01 -2.90164083e-01
-2.18039364e-01 -2.77500361e-01 5.57466090e-01 5.99668741e-01
-4.38014567e-01 -1.07598877e+00 -5.22947550e-01 2.62709349e-01
-1.09946322e+00 -1.23236850e-01 5.36367416e-01 9.53771949e-01
6.19229376e-01 5.55902243e-01 9.23336267e-01 -6.21712923e-01
1.10596824e+00 -1.11718193e-01 -3.46535265e-01 1.23422638e-01
-2.44101077e-01 -1.81827188e-01 8.21619928e-01 -8.71617377e-01
-1.06338263e+00 2.07271859e-01 -1.13742061e-01 -4.94594097e-01
5.57247460e-01 1.01105297e+00 -1.98786750e-01 -6.75217435e-02
5.43530285e-01 1.65132225e-01 -8.94973874e-02 -4.37873930e-01
5.90623140e-01 5.47687948e-01 6.28157735e-01 -5.03797770e-01
1.11910939e+00 6.00102127e-01 1.95747331e-01 -1.40446281e+00
-7.37357199e-01 -5.86565077e-01 -4.05338228e-01 -2.98420429e-01
3.09085071e-01 -9.10041153e-01 -9.85076904e-01 3.75120968e-01
-1.16493809e+00 3.77302259e-01 -3.88180941e-01 9.58404660e-01
-3.71085733e-01 7.66992450e-01 -5.46855271e-01 -1.13829553e+00
-3.16134185e-01 -8.73977780e-01 8.72564316e-01 -2.26933047e-01
-3.15955043e-01 -5.76310217e-01 3.26128006e-02 4.53117080e-02
6.18626215e-02 3.93836852e-03 9.32822108e-01 -8.75971764e-02
-3.87624651e-02 -1.56056076e-01 1.66695073e-01 5.97031772e-01
3.82188708e-01 8.87278616e-02 -1.20491505e+00 -5.23401082e-01
4.87040311e-01 -2.51841806e-02 5.96033156e-01 4.12131369e-01
9.26164269e-01 -3.41683507e-01 2.22208966e-02 4.80336577e-01
9.65098560e-01 3.80240567e-02 4.41411555e-01 -8.18835422e-02
5.20616949e-01 5.34214139e-01 8.72217715e-01 9.16406333e-01
-1.89736232e-01 5.01792908e-01 1.50888175e-01 -6.54053017e-02
2.83755753e-02 -1.33543581e-01 4.73983884e-01 1.59407282e+00
-1.23193651e-01 -5.13335578e-02 -3.84138227e-01 4.65168715e-01
-1.73121107e+00 -9.10088301e-01 -2.41440997e-01 2.26666522e+00
8.66890073e-01 -1.87524527e-01 3.91209833e-02 1.12508547e+00
5.45173883e-01 4.11475599e-01 -4.33465779e-01 -6.30541071e-02
-1.69516519e-01 3.88282210e-01 6.09304719e-02 5.67791402e-01
-1.03120780e+00 3.41836900e-01 6.89854002e+00 9.89026606e-01
-1.18275094e+00 -2.47347966e-01 -1.13899268e-01 1.36748716e-01
-5.35297751e-01 -2.74909586e-02 -1.60947531e-01 1.91148058e-01
7.11304009e-01 -5.23577988e-01 7.57904112e-01 3.68690550e-01
6.23546302e-01 5.92346676e-02 -8.74435186e-01 1.24772501e+00
-1.26772225e-02 -7.62451768e-01 2.81426068e-02 -2.56775796e-01
5.73536932e-01 -8.00942957e-01 3.90383333e-01 1.60104841e-01
-5.74843645e-01 -9.10002828e-01 5.87163687e-01 4.83972609e-01
7.56969452e-01 -8.16380084e-01 1.98054537e-01 1.74899071e-01
-1.42824280e+00 -5.44812866e-02 -3.25682312e-01 -3.74749690e-01
3.59190315e-01 1.07786322e+00 -9.88058209e-01 5.69574773e-01
4.37298119e-01 8.03566992e-01 -7.36420006e-02 8.40196848e-01
-2.97501504e-01 1.00150883e+00 -4.21226025e-01 5.51150858e-01
-2.40381718e-01 -6.24352217e-01 9.17987347e-01 1.05684698e+00
5.35580218e-01 4.72564459e-01 1.48124784e-01 4.34839517e-01
3.99865098e-02 4.22021985e-01 -2.65713632e-01 -3.59110236e-01
3.89422029e-01 1.26070166e+00 -4.58872616e-01 1.09725676e-01
-3.85455340e-01 7.01623976e-01 -4.43883985e-01 6.12313986e-01
-6.68734491e-01 -4.78357047e-01 8.13371420e-01 1.13327973e-01
-2.10111067e-02 -4.02149916e-01 -1.77041113e-01 -1.28039563e+00
9.81141031e-02 -1.32540429e+00 1.06138162e-01 -8.27856839e-01
-1.06321335e+00 5.23304403e-01 -1.52792677e-01 -1.88124180e+00
-5.57191491e-01 -1.72348320e-01 -2.69538701e-01 8.35665286e-01
-1.16781950e+00 -9.79800582e-01 -9.31574730e-04 7.47688830e-01
3.41549307e-01 -6.14173897e-02 8.74939620e-01 7.67963886e-01
-3.00052226e-01 4.87793922e-01 1.41360134e-01 -3.28133613e-01
7.41497040e-01 -1.08923948e+00 4.69371006e-02 1.02314949e+00
2.95696139e-01 1.01311255e+00 1.17873132e+00 -5.94502330e-01
-1.66392684e+00 -7.46045411e-01 4.43358034e-01 1.30595177e-01
9.00007248e-01 -1.95715860e-01 -1.02744770e+00 3.09143931e-01
4.97344658e-02 -1.81139857e-01 8.46744716e-01 1.11260839e-01
-1.47462219e-01 -2.95362741e-01 -8.50938022e-01 6.35808468e-01
6.34758890e-01 -7.80994952e-01 -6.90753937e-01 1.94699094e-01
6.04895473e-01 -3.96227151e-01 -1.02740777e+00 5.35879135e-01
5.98669648e-01 -7.43616998e-01 1.08883643e+00 -3.39308530e-01
1.64922863e-01 -7.46168792e-01 -2.49114200e-01 -1.52393019e+00
-4.74773854e-01 -1.23344374e+00 -1.00612156e-01 1.26561391e+00
1.51991278e-01 -5.98796785e-01 6.88326418e-01 6.84994012e-02
-3.47398296e-02 -4.83092129e-01 -1.03180265e+00 -5.95735252e-01
-4.55719650e-01 -6.52793825e-01 1.92108706e-01 1.04671550e+00
2.23897204e-01 4.54611450e-01 -9.49012756e-01 6.48580432e-01
4.65976119e-01 1.33573681e-01 6.35520160e-01 -1.02179193e+00
-4.83668715e-01 7.00929714e-03 -3.26749235e-01 -1.30779469e+00
1.35925174e-01 -4.84549105e-01 2.54004180e-01 -8.93516183e-01
-1.95607588e-01 -1.71559200e-01 -2.97161609e-01 6.71688691e-02
-1.54286161e-01 4.54486221e-01 3.53906453e-01 1.45360500e-01
8.16069543e-02 8.29594791e-01 1.30834424e+00 -2.01880723e-01
-4.45891351e-01 2.79369175e-01 -5.00005305e-01 9.72455323e-01
2.58673042e-01 -3.93865556e-01 -7.74815381e-01 5.68955801e-02
7.20722020e-01 4.71891314e-01 -2.99743395e-02 -9.20989037e-01
1.03320487e-01 1.20331943e-01 5.44481762e-02 -8.10323894e-01
8.34436238e-01 -9.84541416e-01 4.16884333e-01 3.38991016e-01
-1.83549002e-01 5.01931049e-02 2.94497102e-01 8.04784775e-01
-7.98031151e-01 -3.04546416e-01 5.21497071e-01 3.86383712e-01
-1.23312235e-01 -2.52303988e-01 -6.52662098e-01 -4.71232384e-02
5.00975013e-01 -1.58454835e-01 3.64857204e-02 -9.15760636e-01
-7.98995733e-01 -3.98568660e-01 -7.85688311e-02 5.99143505e-02
9.11833882e-01 -1.28500021e+00 -7.16316462e-01 4.05330747e-01
-3.52156997e-01 -3.83695692e-01 7.11934328e-01 9.83166993e-01
-2.86830604e-01 1.84416875e-01 -4.13569547e-02 -4.93166655e-01
-1.42441738e+00 3.07178527e-01 -7.12981969e-02 -5.15357172e-03
-2.97332883e-01 6.03542387e-01 3.40251267e-01 -4.22118604e-03
1.41702563e-01 -6.30008519e-01 -4.40678537e-01 3.95488292e-01
6.65611446e-01 6.42429590e-01 1.86993524e-01 -8.84231806e-01
-3.94129813e-01 7.65142262e-01 3.87913525e-01 -4.87041980e-01
1.26676250e+00 -1.60511568e-01 -3.99854630e-01 7.28228033e-01
1.10538030e+00 8.25380862e-01 -9.85605299e-01 1.34980038e-01
-2.84263581e-01 -5.43985903e-01 2.29050159e-01 -2.46545091e-01
-8.85009050e-01 7.15322196e-01 5.06840527e-01 3.98152351e-01
1.55194747e+00 -8.52687061e-01 6.87788367e-01 4.48033333e-01
2.74006099e-01 -8.12430620e-01 2.29811102e-01 4.16934341e-01
1.30822921e+00 -6.28611922e-01 5.80433786e-01 -9.86406505e-01
-4.54262912e-01 1.15635705e+00 -1.51925951e-01 -1.93540543e-01
9.88801479e-01 2.01746196e-01 2.05874488e-01 2.69447803e-01
-3.25869411e-01 -3.48606613e-03 7.46671975e-01 7.10704863e-01
6.17251694e-01 -9.07167643e-02 -6.23515487e-01 7.92821825e-01
-5.13145506e-01 -3.51934522e-01 7.72866428e-01 6.06030405e-01
-1.16385326e-01 -1.02738333e+00 -8.40093613e-01 1.02936499e-01
-6.82274759e-01 -3.31968457e-01 -1.37456417e-01 3.37970078e-01
-2.41176799e-01 1.35145772e+00 -3.14543694e-01 -4.92845982e-01
5.32120824e-01 -1.27822846e-01 3.44040036e-01 -5.13554096e-01
-6.40536845e-01 7.13778317e-01 -7.94530436e-02 -3.62398177e-01
-5.18638372e-01 -6.09908342e-01 -1.10380423e+00 -1.35484235e-02
-4.56205606e-01 4.03765202e-01 6.40720248e-01 4.67929810e-01
7.62312859e-02 7.45269418e-01 9.01654363e-01 -9.51697588e-01
-7.31201887e-01 -1.04596770e+00 -9.11475778e-01 4.81736451e-01
6.17016137e-01 -6.37743831e-01 -7.88043499e-01 1.93315342e-01]
|
[15.432947158813477, 5.585911750793457]
|
ff77f46e-be16-4e18-8974-63fdb7ea7dc3
|
temporally-coherent-embeddings-for-self
|
2004.02753
| null |
https://arxiv.org/abs/2004.02753v5
|
https://arxiv.org/pdf/2004.02753v5.pdf
|
Temporally Coherent Embeddings for Self-Supervised Video Representation Learning
|
This paper presents TCE: Temporally Coherent Embeddings for self-supervised video representation learning. The proposed method exploits inherent structure of unlabeled video data to explicitly enforce temporal coherency in the embedding space, rather than indirectly learning it through ranking or predictive proxy tasks. In the same way that high-level visual information in the world changes smoothly, we believe that nearby frames in learned representations will benefit from demonstrating similar properties. Using this assumption, we train our TCE model to encode videos such that adjacent frames exist close to each other and videos are separated from one another. Using TCE we learn robust representations from large quantities of unlabeled video data. We thoroughly analyse and evaluate our self-supervised learned TCE models on a downstream task of video action recognition using multiple challenging benchmarks (Kinetics400, UCF101, HMDB51). With a simple but effective 2D-CNN backbone and only RGB stream inputs, TCE pre-trained representations outperform all previous selfsupervised 2D-CNN and 3D-CNN pre-trained on UCF101. The code and pre-trained models for this paper can be downloaded at: https://github.com/csiro-robotics/TCE
|
['Olivia Mackenzie-Ross', 'Peyman Moghadam', 'Joshua Knights', 'Daniel Ward', 'Ben Harwood', 'Anthony Vanderkop']
|
2020-03-21
| null | null | null | null |
['self-supervised-action-recognition']
|
['computer-vision']
|
[-1.40875429e-01 -4.47701216e-02 -5.38661838e-01 -3.06606919e-01
-3.77366722e-01 -6.18614018e-01 6.88027620e-01 -1.46826245e-02
-3.69877458e-01 5.62904954e-01 5.41681290e-01 1.77366495e-01
3.18213254e-02 -5.02059281e-01 -1.08599293e+00 -6.47381604e-01
-5.53224325e-01 1.60960108e-01 4.13578004e-01 -7.74357244e-02
1.45400437e-02 3.89172316e-01 -1.71609342e+00 4.46640313e-01
1.16915725e-01 9.12740886e-01 1.61940902e-01 8.56194854e-01
3.20257753e-01 1.29470778e+00 -2.20092803e-01 2.49569133e-01
3.72553051e-01 -3.80755275e-01 -7.71083891e-01 3.10486346e-01
4.94280487e-01 -4.31619138e-01 -1.15713906e+00 6.22567356e-01
1.78998008e-01 3.44433039e-01 6.85758471e-01 -1.38404560e+00
-8.39370489e-01 2.42499739e-01 -1.70604676e-01 4.40699428e-01
3.61335099e-01 2.70637006e-01 9.11932349e-01 -9.17347729e-01
1.07796943e+00 8.98375988e-01 4.56939131e-01 6.46651626e-01
-1.03099847e+00 -3.26791435e-01 2.13591188e-01 4.86422330e-01
-1.30099535e+00 -4.85404074e-01 7.37047732e-01 -5.85140646e-01
1.11238503e+00 -1.11756325e-01 8.74015033e-01 1.58618307e+00
7.84230977e-02 8.69752109e-01 6.94617033e-01 -2.07742989e-01
2.29285538e-01 -9.54497308e-02 6.44429177e-02 8.49128425e-01
2.59239990e-02 3.75741988e-01 -8.45410347e-01 2.92923480e-01
9.25021112e-01 3.41281146e-01 -3.85377556e-01 -9.24606800e-01
-1.60443652e+00 6.89557254e-01 5.44125199e-01 3.45296204e-01
-4.03092988e-02 6.15810037e-01 6.74682200e-01 4.93418425e-01
4.51921016e-01 1.93445176e-01 -5.30051470e-01 -4.13059771e-01
-7.29680359e-01 -1.20522082e-01 3.60220790e-01 9.94171143e-01
8.06663454e-01 2.55215198e-01 -2.85016149e-02 4.22193110e-01
2.27470517e-01 2.76303977e-01 8.08874071e-01 -1.14879143e+00
3.41557711e-01 3.34380597e-01 5.58763370e-02 -1.06775331e+00
-1.07890651e-01 -2.63442528e-02 -6.03935301e-01 1.49390399e-01
1.54154032e-01 7.46967196e-02 -9.10147190e-01 1.69767082e+00
-3.02586500e-02 7.44712889e-01 3.35273713e-01 9.80166495e-01
8.47898006e-01 9.79661703e-01 -1.07134096e-01 -1.36268660e-01
7.13642240e-01 -1.24020720e+00 -5.58235168e-01 1.47812646e-02
7.27752566e-01 -3.70900065e-01 8.18577111e-01 2.06955329e-01
-7.57770658e-01 -8.78131568e-01 -1.08280456e+00 -1.70684785e-01
-4.40196246e-01 3.22763294e-01 5.44549465e-01 -4.49751923e-03
-1.26573610e+00 8.10753465e-01 -1.14189124e+00 -6.67561531e-01
6.17198706e-01 1.50112092e-01 -9.51765776e-01 -1.67501599e-01
-9.74459052e-01 7.71445513e-01 5.86394250e-01 7.83013403e-02
-1.44529307e+00 -4.94709849e-01 -1.13250279e+00 -3.89969558e-01
2.75284857e-01 -2.27804676e-01 9.62696970e-01 -1.16021502e+00
-1.52011621e+00 9.37325597e-01 -3.49806771e-02 -6.45731330e-01
3.57190162e-01 -5.09936512e-01 -4.19328988e-01 6.11395240e-01
1.66433677e-02 1.05952275e+00 9.53364611e-01 -1.12617052e+00
-4.29584295e-01 -1.77933395e-01 1.51486039e-01 1.93164229e-01
-3.79733831e-01 -3.56669664e-01 -4.60252702e-01 -5.83875000e-01
-1.88155532e-01 -1.02147198e+00 -1.11309364e-01 4.15170878e-01
6.01497442e-02 -1.90366223e-01 1.36051571e+00 -3.35959882e-01
8.93664658e-01 -2.28378630e+00 4.52052772e-01 -2.88114667e-01
1.45510331e-01 3.27436835e-01 -3.73355389e-01 4.13678080e-01
-3.67397219e-01 -1.30757138e-01 4.97931503e-02 -2.00774625e-01
-1.05394386e-01 6.56457782e-01 -3.18114102e-01 8.60894084e-01
4.78586614e-01 1.09877157e+00 -1.31728923e+00 -4.16678756e-01
6.45502508e-01 5.27819157e-01 -6.29630268e-01 4.07957166e-01
-3.73865396e-01 5.18773675e-01 -2.68970102e-01 6.20899796e-01
1.92930952e-01 -3.35849911e-01 2.03654200e-01 -2.91978180e-01
-1.28795236e-01 1.46428883e-01 -8.81019771e-01 2.23894143e+00
-1.51871592e-01 1.18600225e+00 -5.34136772e-01 -1.61294401e+00
7.39385664e-01 3.67852151e-01 7.80309916e-01 -7.31115460e-01
-6.20893016e-02 -5.80016077e-02 -4.05509830e-01 -7.62379408e-01
4.13561612e-01 3.66062045e-01 6.31449372e-02 3.54229271e-01
7.15486646e-01 9.87315327e-02 2.68244594e-01 4.26794678e-01
1.23591328e+00 7.77881444e-01 2.82702982e-01 -1.65553272e-01
1.70367390e-01 1.57185242e-01 5.20945966e-01 2.91394562e-01
-5.09041548e-01 6.97355807e-01 3.54508072e-01 -7.18887210e-01
-1.12659872e+00 -1.22784829e+00 5.99408634e-02 1.06840205e+00
2.34241262e-01 -7.60556638e-01 -1.72602624e-01 -7.89677262e-01
-7.33232424e-02 1.89282879e-01 -1.03550684e+00 -2.20946386e-01
-5.29948354e-01 4.83532734e-02 3.49415690e-01 7.39116609e-01
3.91791105e-01 -9.95657921e-01 -8.55240107e-01 1.40821919e-01
-3.66882011e-02 -1.07092166e+00 -3.04448038e-01 5.31394243e-01
-9.75383043e-01 -1.27480495e+00 -6.56746089e-01 -8.80248427e-01
6.14347637e-01 5.77554226e-01 1.17721808e+00 -8.33736286e-02
-4.52737391e-01 9.74155009e-01 -8.34414124e-01 1.37951419e-01
-5.60728796e-02 -3.56229216e-01 2.09987178e-01 -1.55224949e-01
3.63833338e-01 -5.44572711e-01 -5.88988602e-01 2.92133749e-01
-8.41318786e-01 -9.05945376e-02 2.04939470e-01 9.10740733e-01
7.17996895e-01 -2.37612575e-01 1.79420233e-01 -3.55713189e-01
-1.77735284e-01 -7.06205249e-01 -3.57989907e-01 1.83211684e-01
-1.00846753e-01 3.93109210e-02 5.39160609e-01 -6.57722294e-01
-5.37370026e-01 4.57528830e-01 3.21760893e-01 -1.26226544e+00
-2.98571497e-01 2.57613927e-01 1.56346839e-02 2.25901157e-01
6.04043782e-01 3.59029055e-01 1.12215549e-01 -1.28840476e-01
6.18419945e-01 3.65368098e-01 6.69595242e-01 -5.18249094e-01
7.52408981e-01 6.71893299e-01 -2.42962480e-01 -7.89486229e-01
-7.68726647e-01 -5.86237490e-01 -1.01760828e+00 -6.04665458e-01
1.12383938e+00 -1.25403166e+00 -3.66209477e-01 2.21334726e-01
-9.08516526e-01 -7.39041388e-01 -7.01643765e-01 7.65135825e-01
-9.20503020e-01 3.66088331e-01 -4.91739988e-01 -3.94665480e-01
3.64109814e-01 -8.57438982e-01 1.07144463e+00 5.74749485e-02
-2.21489504e-01 -1.19655740e+00 4.27745342e-01 1.64404020e-01
1.50655195e-01 5.63906312e-01 3.44298452e-01 -4.28889006e-01
-6.19117558e-01 6.12875409e-02 8.76240879e-02 5.84396362e-01
3.80496114e-01 4.33366835e-01 -9.07065034e-01 -4.55464333e-01
-3.08378160e-01 -8.65325809e-01 1.04137087e+00 4.05509830e-01
1.12156498e+00 -2.27333665e-01 -2.20380440e-01 7.35579669e-01
1.44664168e+00 7.37651139e-02 7.43385613e-01 4.84764218e-01
7.85516798e-01 5.66879094e-01 7.22387910e-01 4.94886965e-01
2.81602591e-01 5.58027804e-01 6.72564745e-01 3.35241742e-02
-1.35657951e-01 -3.74694407e-01 8.65937769e-01 9.44166839e-01
-3.11021715e-01 -2.93317884e-01 -6.34765267e-01 6.66075587e-01
-2.19847631e+00 -1.25190043e+00 1.80343658e-01 1.89928901e+00
5.80623925e-01 -4.44941744e-02 1.81490704e-01 7.48144984e-02
5.22781610e-01 5.68378150e-01 -6.05658114e-01 -1.12576634e-01
-3.19213629e-01 4.40621339e-02 4.92286652e-01 9.15189013e-02
-1.56001878e+00 9.58542347e-01 5.91447163e+00 4.16913927e-01
-1.32237434e+00 1.76560760e-01 4.83382523e-01 -3.71049523e-01
9.82242599e-02 -9.92335752e-02 -2.32206091e-01 2.95556962e-01
1.00848055e+00 -1.09524369e-01 1.85113519e-01 8.97881925e-01
3.02907199e-01 9.35945809e-02 -1.56060767e+00 1.10913897e+00
1.99755192e-01 -1.76889813e+00 6.75840303e-02 -1.47615224e-01
8.74783576e-01 3.66657346e-01 9.96367708e-02 2.70646542e-01
3.67495954e-01 -1.07675838e+00 8.14397454e-01 5.39424419e-01
8.64167035e-01 -3.46401870e-01 4.99800384e-01 -2.00533226e-01
-1.45435452e+00 -1.16643682e-01 -4.69418228e-01 -1.37011424e-01
-5.08172810e-02 3.11969090e-02 -3.20954442e-01 6.31984890e-01
9.79062617e-01 1.98258281e+00 -6.50646269e-01 7.62403667e-01
-2.10409269e-01 6.07211590e-01 2.89404728e-02 2.77432829e-01
5.49126744e-01 -1.63179129e-01 3.26181471e-01 1.24713159e+00
3.22300255e-01 1.07658714e-01 1.93144679e-01 4.35417324e-01
2.64981762e-02 -2.35613614e-01 -1.09485936e+00 -2.91892558e-01
8.74306038e-02 9.04343784e-01 -6.19660556e-01 -5.46745598e-01
-5.97074270e-01 1.23013353e+00 4.13604915e-01 6.05163932e-01
-1.01585412e+00 8.61280039e-02 8.73329580e-01 -1.24263791e-02
6.83019698e-01 -5.48671782e-01 6.23492599e-01 -1.54318750e+00
-4.29532677e-02 -7.47106135e-01 3.83023351e-01 -9.44508255e-01
-1.08088040e+00 5.81410229e-01 2.57368316e-03 -1.98905981e+00
-3.15938473e-01 -9.25637662e-01 -4.89492297e-01 -6.55653626e-02
-1.45531929e+00 -9.45206761e-01 -3.48529816e-01 1.06902993e+00
8.42281222e-01 -4.01218891e-01 8.58370185e-01 1.46877214e-01
-5.12786925e-01 2.23561034e-01 4.69729781e-01 4.01079386e-01
7.46448159e-01 -1.09777701e+00 4.04437371e-02 7.80464888e-01
6.45664334e-01 4.79362816e-01 3.99861515e-01 -3.14384162e-01
-1.62302101e+00 -1.40679789e+00 3.24324548e-01 -5.72892606e-01
9.83845592e-01 -3.79464716e-01 -7.85379767e-01 9.74974751e-01
3.78581703e-01 8.01439047e-01 6.17749751e-01 -3.56596440e-01
-6.59491897e-01 -1.32115334e-01 -7.23448217e-01 3.56608391e-01
1.44363713e+00 -8.35022748e-01 -7.80800283e-01 5.50157428e-01
7.30983973e-01 -1.86547741e-01 -9.33709264e-01 1.82336956e-01
5.26855886e-01 -1.01316655e+00 1.01715994e+00 -9.45197046e-01
7.02992260e-01 -5.09108067e-01 -5.31324267e-01 -1.28220880e+00
-2.29815528e-01 -5.50164461e-01 -4.05833572e-01 8.10088396e-01
-4.43238653e-02 -2.70287097e-01 8.64543319e-01 8.08911473e-02
-2.54066557e-01 -6.43756688e-01 -9.80012000e-01 -1.18060029e+00
-5.09617478e-02 -3.49882454e-01 -5.87273985e-02 9.73191857e-01
7.84376636e-02 1.02157453e-02 -5.53209126e-01 1.39273122e-01
5.42870402e-01 -1.82729512e-02 7.70512164e-01 -7.80230522e-01
-2.28744537e-01 -1.02036230e-01 -1.08426261e+00 -1.26362729e+00
4.67245489e-01 -7.71522403e-01 9.59393531e-02 -1.31939471e+00
9.61489230e-02 -1.79857433e-01 -5.75868368e-01 6.22724235e-01
4.16124851e-01 4.73597139e-01 2.16453627e-01 4.80469435e-01
-1.22362828e+00 9.11746860e-01 1.05057001e+00 -4.36948299e-01
1.38253316e-01 -5.84099591e-01 -1.24448864e-02 5.06989062e-01
7.21999943e-01 -3.32107097e-01 -7.93425143e-01 -7.01398492e-01
-2.87683547e-01 -2.94673537e-05 6.03443027e-01 -1.07986879e+00
1.06123142e-01 -1.33752361e-01 5.46158969e-01 -3.41812044e-01
5.16283154e-01 -8.59603703e-01 4.38779332e-02 5.08601189e-01
-5.20115554e-01 1.86161757e-01 1.55009523e-01 8.72223139e-01
-6.17458880e-01 3.89447585e-02 5.48209846e-01 -2.02968225e-01
-1.44191897e+00 4.75689858e-01 -5.76942742e-01 2.95085777e-02
1.31679237e+00 -4.21382278e-01 -4.11898345e-01 -4.58017528e-01
-8.93589914e-01 1.00356705e-01 6.93605423e-01 7.78383017e-01
9.90253747e-01 -1.58534598e+00 -3.90995950e-01 1.37435406e-01
5.35640478e-01 -1.53486371e-01 1.88637123e-01 5.49993992e-01
-6.63738489e-01 4.51825410e-01 -7.43877709e-01 -8.85327935e-01
-1.06965673e+00 7.32635856e-01 2.30658948e-01 1.12533011e-01
-8.45474601e-01 8.03321779e-01 1.03665672e-01 -2.45809600e-01
4.78639156e-01 -4.99150455e-01 -1.60215586e-01 7.44007230e-02
4.00645614e-01 7.49068484e-02 -3.89957249e-01 -8.69913578e-01
-5.81206918e-01 6.07796192e-01 1.94937363e-01 1.06814772e-01
1.73267078e+00 -8.36842731e-02 1.54047653e-01 7.35050797e-01
1.79553878e+00 -5.53533435e-01 -1.78787446e+00 -2.40049824e-01
9.44120716e-03 -5.88343322e-01 -1.11853406e-02 -8.54474977e-02
-1.27622402e+00 8.75718057e-01 7.85363078e-01 -2.21334100e-01
9.77126837e-01 2.00914711e-01 3.79924119e-01 5.89054286e-01
3.54507357e-01 -1.14592600e+00 9.43621039e-01 4.10758793e-01
9.08452213e-01 -1.46663344e+00 -2.81648077e-02 1.89396828e-01
-8.71277213e-01 1.35812902e+00 6.60287738e-01 -6.43035710e-01
7.33249128e-01 4.24294434e-02 -1.35568269e-02 -2.80720651e-01
-1.06368995e+00 -2.40822136e-01 1.65300027e-01 8.46315145e-01
3.95339400e-01 -2.14326680e-01 3.01020086e-01 -8.29709694e-03
3.42396080e-01 1.96412966e-01 5.93433738e-01 1.17668664e+00
-1.63293332e-01 -7.89317667e-01 1.38026193e-01 1.97301015e-01
8.73268303e-03 1.86113939e-01 -1.42925769e-01 9.48330700e-01
1.21550016e-01 6.91477537e-01 4.33895588e-01 -6.86492622e-01
1.46994498e-02 -1.49737746e-01 5.02045393e-01 -6.80485189e-01
1.95137095e-02 -4.45908122e-02 3.86520065e-02 -1.08536124e+00
-1.21063733e+00 -6.85082078e-01 -1.17960107e+00 3.29135582e-02
6.10719472e-02 8.54226053e-02 3.25749159e-01 6.12394154e-01
4.04705465e-01 4.20676231e-01 7.16776252e-01 -1.31539190e+00
-1.18535921e-01 -7.17809260e-01 -5.02423167e-01 6.29098654e-01
5.07496297e-01 -1.01786065e+00 -3.72310460e-01 7.76179671e-01]
|
[8.675714492797852, 0.727519154548645]
|
4edf22ea-c80d-4394-b375-302067c7ad12
|
solution-of-debertav3-on-commonsenseqa
|
2206.05033
| null |
https://arxiv.org/abs/2206.05033v2
|
https://arxiv.org/pdf/2206.05033v2.pdf
|
Solution of DeBERTaV3 on CommonsenseQA
|
We report the performance of DeBERTaV3 on CommonsenseQA in this report. We simply formalize the answer selection as a text classification for DeBERTaV3. The strong natural language inference ability of DeBERTaV3 helps its single and ensemble model set the new (w/o external knowledge) state-of-the-art on CommonsenseQA.
|
['Hai Zhao', 'Zuchao Li', 'Letian Peng']
|
2022-04-30
| null | null | null | null |
['answer-selection']
|
['natural-language-processing']
|
[-3.18715721e-01 4.64467525e-01 -1.75941423e-01 -4.51974392e-01
-6.39252424e-01 -8.99547756e-01 6.13856494e-01 3.06619108e-01
-2.17584893e-01 1.14023280e+00 4.20327902e-01 -8.71844292e-01
-2.42181849e-02 -1.31601954e+00 -6.36874616e-01 6.64979815e-02
5.40127695e-01 9.61592197e-01 6.17397130e-01 -1.17150402e+00
1.51215598e-01 3.72857265e-02 -1.14431095e+00 6.10394180e-01
1.35813940e+00 9.15981233e-01 -5.50921440e-01 9.18895245e-01
-5.76128542e-01 2.08534718e+00 -9.00911272e-01 -1.33631635e+00
-1.69109300e-01 -4.94502455e-01 -1.74994218e+00 -8.32031429e-01
6.93086028e-01 -3.81716311e-01 -4.68097240e-01 1.02137780e+00
2.95028418e-01 3.31054389e-01 8.43654811e-01 -1.49771345e+00
-1.53293824e+00 1.46550608e+00 2.95540363e-01 5.55984795e-01
7.79356182e-01 5.53896964e-01 1.75013626e+00 -6.74664795e-01
7.15002537e-01 1.52662551e+00 6.75433218e-01 7.98440456e-01
-1.18848085e+00 -5.89921117e-01 -2.05008924e-01 9.39797759e-01
-1.06497872e+00 -5.42933226e-01 5.57779431e-01 -1.48431242e-01
1.73173845e+00 5.25577068e-01 4.57224846e-01 1.17215717e+00
3.35391492e-01 1.11127758e+00 1.16585362e+00 -4.21237797e-01
3.41638476e-01 -5.86701073e-02 1.06766427e+00 9.81712759e-01
2.59675682e-01 -1.70129523e-01 -6.75352335e-01 -5.21772742e-01
1.72042444e-01 -5.06347358e-01 -1.92235425e-01 1.38007119e-01
-8.29499185e-01 1.28613722e+00 5.49846470e-01 9.06726867e-02
-1.59212779e-02 7.47961044e-01 7.64282346e-01 7.50499725e-01
2.59445727e-01 1.18127418e+00 -7.36420989e-01 -2.58271188e-01
-6.46948993e-01 5.90858042e-01 1.27842593e+00 1.15323448e+00
3.24225873e-01 4.89394367e-03 -4.29352731e-01 1.99878335e-01
1.70760706e-01 1.05472016e+00 6.39522016e-01 -1.25459599e+00
4.86515313e-01 7.34676421e-01 -5.89575432e-02 -9.82004404e-01
-1.60215229e-01 -2.98558354e-01 -4.61892307e-01 -3.30237031e-01
2.04149649e-01 -5.56208491e-02 -6.68650448e-01 1.33676040e+00
-1.15176648e-01 -2.41451263e-01 4.01821077e-01 6.59888625e-01
1.37398481e+00 5.68385541e-01 2.28404388e-01 3.82654630e-02
1.05011868e+00 -9.65578079e-01 -7.73213506e-01 -3.59427065e-01
8.85954797e-01 -5.35635874e-02 1.42796445e+00 5.50044060e-01
-9.97233272e-01 -6.76092282e-02 -8.52657139e-01 -6.68260217e-01
-6.25855386e-01 -4.17088091e-01 8.44455063e-01 3.21749568e-01
-1.00294495e+00 3.90280247e-01 -3.53008658e-01 4.10593022e-03
4.59640145e-01 -7.95986205e-02 -1.65298525e-02 -8.95140022e-02
-2.10310197e+00 1.52271080e+00 5.84629714e-01 -3.12637717e-01
-9.38673735e-01 -5.29188216e-01 -9.22970355e-01 -1.64764687e-01
8.08391213e-01 -1.22761643e+00 1.53272033e+00 -5.55598497e-01
-1.57213748e+00 1.06783462e+00 -3.05091202e-01 -9.86273587e-01
3.19641113e-01 -5.46393931e-01 -6.52689159e-01 2.78188616e-01
2.58110821e-01 9.99003723e-02 4.61742610e-01 -9.01169002e-01
-4.42115404e-02 -3.31347793e-01 6.17474020e-01 -2.23589167e-01
2.93669440e-02 -3.36312056e-02 2.63829947e-01 -3.63485783e-01
-4.41577762e-01 -5.63620627e-01 3.23568970e-01 -1.57422066e-01
-6.07552290e-01 -8.90630901e-01 4.73313242e-01 -3.42158198e-01
1.38025331e+00 -1.45442080e+00 2.94434488e-01 -1.40070645e-02
6.44903183e-01 7.18759000e-02 -3.90808731e-02 3.58865291e-01
7.65899047e-02 4.51692492e-01 -1.78083852e-01 -7.97570869e-03
4.48542386e-01 5.98851800e-01 -1.14573789e+00 4.90774103e-02
3.11897218e-01 1.80086362e+00 -1.44408906e+00 -1.09901226e+00
-4.95395027e-02 -4.11405355e-01 -4.40119267e-01 1.98552653e-01
-9.06500876e-01 -4.92484331e-01 -6.57085359e-01 6.70872927e-01
-1.08448626e-03 -6.24715090e-01 -3.92929256e-01 -1.94498282e-02
5.84616423e-01 7.77911901e-01 -3.73185068e-01 1.59342384e+00
-3.80395800e-01 7.28434503e-01 -4.41782236e-01 -4.07082409e-01
5.29469609e-01 9.34114903e-02 -4.79256392e-01 -4.32917774e-01
1.19750254e-01 1.83145314e-01 1.20517135e-01 -7.87177444e-01
6.60857618e-01 -8.05831611e-01 -3.13819796e-01 6.66580498e-01
4.15472567e-01 -1.05347455e+00 2.32951894e-01 1.01877213e+00
1.23564720e+00 1.28802001e-01 6.39595389e-01 -5.70288360e-01
5.84936738e-01 5.51360071e-01 3.46917147e-03 1.06063402e+00
-2.36853555e-01 -2.20116023e-02 5.30153990e-01 -3.73313487e-01
-6.15696728e-01 -1.01637387e+00 1.54078682e-03 1.57438898e+00
6.29024059e-02 -6.42682016e-01 -3.72754753e-01 -1.13370156e+00
4.07129198e-01 1.89220583e+00 -9.68037903e-01 -4.50463057e-01
-2.79717326e-01 -1.19383544e-01 1.40984643e+00 6.75605476e-01
5.05510986e-01 -1.38034582e+00 -6.55691803e-01 8.26167390e-02
-5.57420790e-01 -8.04134667e-01 -6.47696033e-02 5.24474323e-01
-7.75451243e-01 -1.13746679e+00 2.53437072e-01 -2.40825154e-02
-2.35908508e-01 -3.06271046e-01 2.18884611e+00 2.38503516e-01
1.90997288e-01 3.47460300e-01 -7.52597332e-01 -6.84057951e-01
-7.25188315e-01 1.95259541e-01 -1.97256118e-01 -9.43741858e-01
7.72423685e-01 -4.81179148e-01 -1.27454609e-01 -5.12384735e-02
-8.36333334e-01 -3.32412153e-01 -1.58517063e-01 8.93609703e-01
6.29850104e-02 -1.28229618e-01 6.27098501e-01 -1.20657670e+00
1.08973026e+00 -7.42260993e-01 -5.85667416e-02 5.30783236e-01
-6.19683504e-01 3.12028974e-01 9.81064737e-01 -4.22270894e-02
-8.00550818e-01 -9.46722209e-01 -9.82374921e-02 -2.28127345e-01
1.86172649e-01 6.55933142e-01 4.12355587e-02 3.21373403e-01
1.44576371e+00 3.43387008e-01 -3.19925904e-01 2.38209531e-01
1.11315024e+00 5.24798453e-01 7.38550901e-01 -1.19991255e+00
8.10702264e-01 2.12510183e-01 -2.70419270e-01 -3.06129128e-01
-1.58542395e+00 -1.56762794e-01 -3.23809832e-01 1.89167529e-01
6.94598436e-01 -4.54308152e-01 -9.16864216e-01 1.68283328e-01
-1.48805988e+00 -6.67425632e-01 -6.05494499e-01 -4.67881829e-01
-6.45304561e-01 1.92994341e-01 -6.94935977e-01 -7.95598149e-01
-9.88019824e-01 -4.98740643e-01 8.36100996e-01 -3.38537663e-01
-9.16589856e-01 -1.25893867e+00 2.32079923e-01 5.68956494e-01
5.92871726e-01 5.55859767e-02 1.32287073e+00 -1.21867192e+00
-1.70993045e-01 1.38930064e-02 -1.56684250e-01 5.69171309e-01
-1.77546978e-01 1.75341424e-02 -1.16419065e+00 3.98834854e-01
4.26662266e-02 -1.31057560e+00 9.11663234e-01 -1.98619604e-01
1.21879065e+00 -6.17553174e-01 1.78638339e-01 3.40329446e-02
1.17736292e+00 -3.48138332e-01 5.88797987e-01 3.61089677e-01
4.32072282e-01 9.45148319e-02 3.89064431e-01 3.35497499e-01
6.98734462e-01 7.13674799e-02 5.77095509e-01 6.40270770e-01
-1.77944839e-01 -3.20119232e-01 3.95688385e-01 7.49269724e-01
-1.83113560e-01 -3.87307853e-01 -1.33322263e+00 6.56137168e-01
-1.89140785e+00 -1.23363578e+00 -2.85634667e-01 1.05309701e+00
1.70405209e+00 2.01066643e-01 -1.27330914e-01 -1.30925914e-02
7.03368187e-02 1.78617150e-01 -7.80400395e-01 -7.30404854e-01
-6.14404678e-01 6.32653594e-01 -9.93747562e-02 8.29723358e-01
-6.81809187e-01 1.38837135e+00 7.59520006e+00 1.10904169e+00
-4.88520563e-01 1.29935026e-01 3.66650783e-02 -1.19013980e-01
-9.59896624e-01 -8.55429322e-02 -4.39300478e-01 1.83550999e-01
9.40437138e-01 -5.87262690e-01 5.96169710e-01 9.60344851e-01
-7.39105523e-01 3.41419093e-02 -1.49462998e+00 6.71234190e-01
3.24050039e-01 -1.51232445e+00 6.66097224e-01 -5.48992217e-01
4.77370083e-01 2.15791374e-01 -1.97728887e-01 9.34526801e-01
1.23504269e+00 -1.52483368e+00 8.36319804e-01 6.41832829e-01
4.76912886e-01 -7.79103577e-01 1.11850882e+00 5.29805541e-01
-4.35173959e-01 -2.10490301e-01 -2.76300341e-01 -2.29441762e-01
-6.82153851e-02 6.02178037e-01 -4.73050594e-01 4.50720936e-01
5.38402736e-01 6.19671822e-01 -7.88708150e-01 -6.64745942e-02
-9.80494797e-01 9.28584576e-01 -6.97770342e-02 -6.56163096e-01
2.18791202e-01 3.92516106e-01 6.22164011e-01 1.44220805e+00
-4.45463359e-01 8.50652575e-01 -1.45529002e-01 1.38335192e+00
-5.16627371e-01 -2.10448503e-01 -6.19353056e-01 -1.29145578e-01
5.88226438e-01 7.56446660e-01 2.41825402e-01 -1.00871146e+00
9.90909040e-02 9.64437485e-01 7.23625302e-01 1.64695457e-01
-7.87166238e-01 -6.98588073e-01 3.46963465e-01 -1.92397922e-01
3.70848514e-02 7.78605863e-02 -5.63228011e-01 -1.40448809e+00
-1.61813781e-01 -1.25617588e+00 7.38876045e-01 -1.41068053e+00
-2.09659719e+00 6.32928491e-01 1.30608365e-01 -1.32836878e-01
-5.50542891e-01 -9.17054057e-01 -6.85154140e-01 6.54697835e-01
-1.27101374e+00 -1.10927308e+00 -1.51044488e-01 9.65087295e-01
3.68060648e-01 -2.46378556e-01 1.29856789e+00 -6.04471147e-01
6.64443988e-03 5.92744112e-01 -4.05721545e-01 3.70069683e-01
5.80955803e-01 -1.76303554e+00 4.96576756e-01 5.19104898e-01
4.72980812e-02 1.14108682e+00 1.03578842e+00 -7.38851190e-01
-1.52481246e+00 -9.03853059e-01 9.89380002e-01 -1.59842443e+00
1.71055663e+00 2.12581083e-01 -9.67079699e-01 1.20669019e+00
6.06285989e-01 2.88237096e-03 8.92010927e-01 4.95171756e-01
-1.06693411e+00 1.03659898e-01 -1.20946801e+00 5.74618936e-01
1.12911665e+00 -1.11926234e+00 -1.86676395e+00 7.76149035e-02
1.26866806e+00 -3.87405127e-01 -8.48234355e-01 6.39937147e-02
2.18509629e-01 -5.81457019e-01 6.13345981e-01 -1.53143382e+00
1.37121177e+00 -2.03199163e-01 -5.31833649e-01 -1.62496686e+00
-1.97236553e-01 -5.84936023e-01 -1.00766027e+00 8.39835227e-01
5.53252995e-01 -7.49612927e-01 2.62616634e-01 1.04103267e+00
9.14211497e-02 -6.28388643e-01 -1.10438025e+00 -8.47124338e-01
7.88684249e-01 -8.56056809e-01 6.39878333e-01 1.17642558e+00
7.22127914e-01 9.19220686e-01 3.24817747e-01 -4.48174328e-01
5.52438915e-01 5.57440877e-01 5.63951433e-01 -1.12621188e+00
-5.99345505e-01 -4.30858761e-01 -2.10788324e-01 -9.82097805e-01
7.12039888e-01 -1.35838628e+00 4.52384390e-02 -1.61175346e+00
4.66302395e-01 -6.37808591e-02 -2.01840356e-01 9.66697872e-01
-4.09712046e-01 -2.91417632e-02 1.30678087e-01 -2.09166229e-01
-1.08179438e+00 7.98897684e-01 1.21920192e+00 -6.16698146e-01
3.16985607e-01 -4.92197216e-01 -1.17780006e+00 6.64665222e-01
8.23694289e-01 -2.60767639e-01 -6.51546419e-01 -7.01487422e-01
1.03223336e+00 -1.99990869e-01 7.24520326e-01 -3.74368638e-01
3.90419453e-01 -5.08079827e-01 -9.11856666e-02 -3.01600456e-01
3.59347127e-02 -4.69471246e-01 -7.39452302e-01 2.39504188e-01
-8.15917432e-01 1.38665423e-01 3.75752270e-01 2.50055999e-01
-7.78318420e-02 -4.31835443e-01 3.45283926e-01 -4.00727630e-01
-5.58092117e-01 -3.33386540e-01 -3.02824348e-01 1.15066218e+00
4.28816319e-01 2.64350802e-01 -1.23075199e+00 -6.13322854e-01
-5.43514550e-01 3.56190532e-01 3.19158733e-01 1.37062995e-02
9.07740653e-01 -9.28419173e-01 -1.07015812e+00 -6.96011364e-01
4.25076842e-01 9.08896029e-02 -2.62602746e-01 3.73022050e-01
-4.32219535e-01 5.73909402e-01 1.70889907e-02 -6.76731765e-03
-1.03049767e+00 7.29148567e-01 6.64350986e-01 -3.49865884e-01
-3.34701240e-01 1.11936367e+00 -3.88167471e-01 -5.44465601e-01
-2.63233513e-01 -4.83287454e-01 -4.27821651e-02 -2.33432204e-01
5.15048206e-01 5.97013116e-01 1.34577593e-02 -2.87569873e-02
-6.70005441e-01 5.11675775e-02 -6.74573109e-02 -1.94730133e-01
1.11236715e+00 2.59389639e-01 -9.00110245e-01 8.18527043e-01
7.51247764e-01 2.35745490e-01 7.45074973e-02 -3.23127300e-01
-6.99518174e-02 -4.53710323e-03 1.53782457e-01 -1.61750484e+00
-3.73430312e-01 9.10344243e-01 -4.89114791e-01 3.67100269e-01
6.31645620e-01 5.57437122e-01 8.14094007e-01 1.28982472e+00
6.88801706e-01 -9.02143121e-01 5.99296652e-02 1.28582394e+00
1.38492823e+00 -1.10773277e+00 3.08947682e-01 -1.96933106e-01
-8.56094539e-01 1.17997110e+00 8.07580888e-01 -2.38057986e-01
3.68182719e-01 1.70680046e-01 -5.36427386e-02 -6.54015124e-01
-1.18755102e+00 -1.52373254e-01 4.07383263e-01 4.16970074e-01
4.82800245e-01 2.30300903e-01 1.20579883e-01 1.02953362e+00
-9.19148684e-01 1.66663349e-01 5.22825122e-01 7.65474916e-01
-4.90815341e-01 -4.57737923e-01 -1.93740241e-02 5.47084928e-01
-3.15387487e-01 -9.00686860e-01 -1.21890986e+00 8.27778578e-01
-6.20841607e-02 1.28979743e+00 -4.62215066e-01 -4.30257887e-01
2.86572039e-01 3.67710203e-01 7.94699073e-01 -7.04262078e-01
-9.32219863e-01 -1.30610049e+00 7.12485850e-01 -6.71588063e-01
-1.18856966e-01 -5.54601029e-02 -1.69808269e+00 -8.51596355e-01
-3.98512661e-01 4.31999922e-01 -2.52094492e-02 1.18520880e+00
2.08860233e-01 4.37683254e-01 -2.05688784e-03 2.96643227e-01
-1.01746297e+00 -1.05352962e+00 -2.36300126e-01 5.71646094e-01
4.81432915e-01 -3.33599687e-01 -6.08595252e-01 1.92962158e-02]
|
[10.08289623260498, 8.03609848022461]
|
cdc34961-cf50-4758-9609-f6bc43a1f03e
|
event-based-moving-object-detection-and
|
1803.04523
| null |
https://arxiv.org/abs/1803.04523v3
|
https://arxiv.org/pdf/1803.04523v3.pdf
|
Event-based Moving Object Detection and Tracking
|
Event-based vision sensors, such as the Dynamic Vision Sensor (DVS), are ideally suited for real-time motion analysis. The unique properties encompassed in the readings of such sensors provide high temporal resolution, superior sensitivity to light and low latency. These properties provide the grounds to estimate motion extremely reliably in the most sophisticated scenarios but they come at a price - modern event-based vision sensors have extremely low resolution and produce a lot of noise. Moreover, the asynchronous nature of the event stream calls for novel algorithms. This paper presents a new, efficient approach to object tracking with asynchronous cameras. We present a novel event stream representation which enables us to utilize information about the dynamic (temporal) component of the event stream, and not only the spatial component, at every moment of time. This is done by approximating the 3D geometry of the event stream with a parametric model; as a result, the algorithm is capable of producing the motion-compensated event stream (effectively approximating egomotion), and without using any form of external sensors in extremely low-light and noisy conditions without any form of feature tracking or explicit optical flow computation. We demonstrate our framework on the task of independent motion detection and tracking, where we use the temporal model inconsistencies to locate differently moving objects in challenging situations of very fast motion.
|
['Yiannis Aloimonos', 'Chethan Parameshwara', 'Cornelia Fermuller', 'Anton Mitrokhin']
|
2018-03-12
| null | null | null | null |
['motion-detection', 'moving-object-detection', 'event-based-vision']
|
['computer-vision', 'computer-vision', 'computer-vision']
|
[ 1.41512439e-01 -6.90965295e-01 2.19176844e-01 9.84684527e-02
-6.14569709e-02 -6.09876096e-01 7.30185688e-01 -1.27798580e-02
-6.66459322e-01 5.50763011e-01 -2.49810845e-01 9.25179794e-02
2.71557886e-02 -6.04829252e-01 -4.26143080e-01 -7.19444811e-01
1.18490942e-02 1.97311312e-01 8.85483503e-01 1.10379800e-01
1.38276875e-01 9.68008935e-01 -1.97129512e+00 -1.65626839e-01
3.79787862e-01 1.08162749e+00 3.67482722e-01 1.19115138e+00
-1.28007025e-01 9.41372573e-01 -3.77762109e-01 5.11141978e-02
3.43781739e-01 -3.82077187e-01 -1.73359871e-01 3.02774549e-01
5.11363387e-01 -7.34880626e-01 -5.37030160e-01 8.30066562e-01
3.69393796e-01 2.52325356e-01 3.75920475e-01 -1.21358860e+00
1.84822395e-01 -2.88990825e-01 -4.88419533e-01 4.94304150e-01
8.00146699e-01 4.95854437e-01 4.26720142e-01 -8.88613284e-01
1.02070212e+00 9.54012871e-01 6.79010034e-01 5.27957499e-01
-1.08800006e+00 -4.20138612e-02 -2.12338954e-01 4.91316646e-01
-1.21386039e+00 -7.34049797e-01 7.59244025e-01 -5.72492301e-01
8.16114724e-01 2.99082786e-01 9.67003405e-01 7.69856453e-01
3.05416942e-01 4.92520750e-01 7.97616959e-01 -2.98025817e-01
3.28157306e-01 -9.49772373e-02 -1.17018018e-02 7.38036633e-01
2.54486561e-01 4.49398786e-01 -6.98635340e-01 -1.19315356e-01
1.02505791e+00 2.72809505e-01 -5.94110489e-01 -5.66936910e-01
-1.51127934e+00 3.34767103e-01 4.52883616e-02 1.61012396e-01
-5.68399847e-01 4.99579906e-01 3.16367060e-01 1.71758592e-01
5.90608045e-02 -1.99625239e-01 -1.17742829e-01 -3.92729133e-01
-8.14519584e-01 1.71621531e-01 8.41791749e-01 8.07313442e-01
6.92418814e-01 2.40840837e-01 7.39323124e-02 2.51264852e-02
4.38062936e-01 7.63346851e-01 6.15565717e-01 -1.27840531e+00
1.61662757e-01 2.12606788e-01 5.24097860e-01 -9.70511019e-01
-4.47172016e-01 5.32284379e-02 -6.13973498e-01 7.94251978e-01
8.22853744e-01 -4.13458794e-02 -5.66891491e-01 1.41483569e+00
7.17787445e-01 4.98669684e-01 -5.59361419e-03 9.48483527e-01
5.11718512e-01 7.28977919e-01 -2.74413556e-01 -8.46142650e-01
1.20768809e+00 -3.69565368e-01 -9.24922049e-01 1.20203659e-01
1.23314999e-01 -8.20103109e-01 6.48150742e-01 4.05859292e-01
-1.21740842e+00 -5.82528472e-01 -9.24573839e-01 8.77327695e-02
-2.02427372e-01 -2.28874519e-01 4.15476918e-01 4.58132058e-01
-9.75512922e-01 4.73749816e-01 -1.15623224e+00 -3.92222524e-01
-2.09834844e-01 2.64699757e-01 -3.28645259e-01 3.70941162e-02
-6.44768238e-01 8.24674368e-01 1.46747217e-01 2.16941953e-01
-6.57646835e-01 -4.61764038e-01 -7.66308069e-01 -1.07318193e-01
2.97552347e-01 -8.91783416e-01 1.19094527e+00 -8.38162065e-01
-1.63724434e+00 5.85135698e-01 -5.37771463e-01 -4.75791782e-01
8.49583745e-01 -1.03071928e-01 -1.39999658e-01 5.88319302e-01
-1.91419080e-01 2.84486771e-01 1.09970284e+00 -8.17758679e-01
-6.79853797e-01 -3.16970527e-01 -1.32542938e-01 3.38858813e-01
-5.85835874e-02 7.21036270e-02 -2.82666743e-01 -2.96944827e-01
7.34618902e-02 -7.60269284e-01 -2.26080745e-01 6.61580563e-01
2.39745617e-01 4.98976931e-02 1.13783717e+00 -8.90657380e-02
9.26105499e-01 -2.30787897e+00 -1.66060492e-01 -1.22320205e-01
1.43863440e-01 4.15600955e-01 2.62951910e-01 2.93906629e-01
2.09506422e-01 -6.15908921e-01 -3.79237346e-02 -2.43270636e-01
-4.88431275e-01 2.79456347e-01 -3.69199634e-01 7.98943818e-01
-1.77491046e-02 5.85469961e-01 -1.15848100e+00 -6.47536278e-01
9.09740031e-01 5.66738129e-01 -1.84343517e-01 1.07055828e-01
-1.99645996e-01 6.48344100e-01 -4.15095419e-01 4.43448961e-01
4.22832549e-01 -4.57025468e-02 -2.78315008e-01 -2.05576986e-01
-6.82771683e-01 -2.73880869e-01 -1.74866307e+00 1.46133697e+00
-2.86794752e-01 1.13903916e+00 1.48334622e-01 -5.43173492e-01
7.68607914e-01 4.98189956e-01 9.42908823e-01 -5.36204457e-01
1.51150435e-01 1.15684584e-01 -3.08347344e-01 -7.09285200e-01
5.15138626e-01 -1.06348796e-02 5.10699332e-01 3.57960641e-01
-3.02339137e-01 -1.57898162e-02 1.89247161e-01 2.93971300e-02
1.19706035e+00 3.24410349e-01 5.29932737e-01 2.33855352e-01
5.97088516e-01 8.31895396e-02 5.34375429e-01 6.47981942e-01
-4.48306590e-01 6.62559330e-01 -9.95818004e-02 -5.68590105e-01
-9.62834418e-01 -1.13611603e+00 -6.16675653e-02 2.93991387e-01
5.06903291e-01 -2.10415557e-01 -2.99015552e-01 -7.77032599e-02
-9.18934345e-02 1.72235355e-01 -1.56568542e-01 2.11795405e-01
-7.38403559e-01 -4.39296305e-01 6.93996027e-02 4.88043159e-01
3.34114522e-01 -7.26863623e-01 -1.56337225e+00 5.40294409e-01
-6.57317713e-02 -1.52112377e+00 -4.02797759e-01 -1.15924060e-01
-1.08323538e+00 -1.18023264e+00 -5.79687774e-01 -3.36797476e-01
3.79905820e-01 6.15059018e-01 7.95751572e-01 -2.02267393e-01
-5.82279861e-01 9.80888069e-01 -1.44336328e-01 -3.13368380e-01
-1.92909002e-01 -1.01958835e+00 1.36279449e-01 5.23287952e-01
2.18717724e-01 -4.66829956e-01 -7.46547759e-01 2.77464539e-01
-8.79386544e-01 -1.22623309e-01 -8.45761225e-03 5.33084273e-01
6.19567454e-01 -1.91625915e-02 -2.64100637e-02 -1.57363430e-01
1.17414994e-02 -1.12741940e-01 -1.11864889e+00 -1.17299825e-01
-1.87590972e-01 -2.34122038e-01 7.07315505e-01 -7.33527124e-01
-1.12137496e+00 6.92332983e-01 2.40924537e-01 -8.01887989e-01
-1.97123736e-01 -2.30004750e-02 1.93451196e-01 -2.99308985e-01
6.94519341e-01 3.62263292e-01 1.63906619e-01 -2.28499606e-01
3.55336666e-01 3.87736678e-01 8.60370874e-01 -4.61505726e-02
6.38364911e-01 1.25528646e+00 5.38047791e-01 -1.28073418e+00
-1.43031687e-01 -9.19194818e-01 -7.34907627e-01 -6.84098005e-01
8.28166902e-01 -8.39399159e-01 -9.35498059e-01 7.96207786e-01
-1.52999926e+00 -9.78612378e-02 -6.67344153e-01 9.94758308e-01
-9.04240727e-01 7.30754197e-01 -5.20587742e-01 -1.17067027e+00
-5.89768440e-02 -9.43721771e-01 8.64851058e-01 3.75951231e-01
-4.86761667e-02 -9.86164570e-01 1.29856318e-01 -6.23530261e-02
2.60146230e-01 5.73819935e-01 8.64605084e-02 1.24844993e-02
-1.07355392e+00 -3.55393082e-01 -8.80758241e-02 -3.97328883e-02
-3.96137219e-03 4.25092638e-01 -1.04366875e+00 -8.81759152e-02
4.13777441e-01 3.58631253e-01 6.08518183e-01 8.40180933e-01
5.84039986e-01 9.69296396e-02 -2.72538871e-01 7.98939109e-01
1.71425974e+00 3.27016830e-01 4.54125285e-01 1.40971273e-01
6.95378184e-01 5.21828532e-01 6.56728983e-01 7.33584464e-01
1.72761530e-01 9.76792216e-01 6.00273550e-01 2.76319593e-01
-3.71882021e-01 2.07706809e-01 5.33845603e-01 4.43195999e-01
-4.26728457e-01 -1.89347774e-01 -6.67471290e-01 5.80309212e-01
-1.94843328e+00 -1.25252497e+00 -7.78091311e-01 2.49029994e+00
3.96508485e-01 -6.43599778e-02 2.18955755e-01 2.87657261e-01
6.81146026e-01 1.11736758e-02 -4.26079601e-01 -2.10919559e-01
-1.21826030e-01 -9.43040475e-02 6.14015877e-01 4.92499202e-01
-8.75353634e-01 3.52243602e-01 6.60878181e+00 1.64997667e-01
-1.23008347e+00 1.65886879e-02 -2.55342066e-01 -3.20328146e-01
1.27388850e-01 7.17615634e-02 -7.92115867e-01 7.05229461e-01
9.45552886e-01 -2.23317593e-01 1.02647729e-01 5.12241542e-01
5.94084442e-01 -5.43773234e-01 -1.14326298e+00 1.28363431e+00
-3.06736175e-02 -1.19919586e+00 -1.85111821e-01 4.24046069e-02
3.44692856e-01 6.17002547e-02 -3.17927152e-01 -6.60145402e-01
-1.89735502e-01 -3.94338459e-01 6.85665607e-01 8.13009679e-01
6.54245138e-01 -3.35707366e-01 3.90507281e-01 4.34911758e-01
-1.36966038e+00 -5.21087125e-02 -3.24977547e-01 -1.65609583e-01
7.25195825e-01 6.47200465e-01 -5.76616228e-01 3.18606704e-01
5.40980518e-01 8.33848417e-01 -7.51362592e-02 1.45904398e+00
3.04088593e-01 4.17622514e-02 -7.98218966e-01 4.73254127e-03
-3.50813791e-02 -2.57105410e-01 1.07138431e+00 1.05189276e+00
5.22493064e-01 1.99274451e-01 5.14035784e-02 5.49793124e-01
5.97841799e-01 -2.91576445e-01 -9.09949839e-01 3.92313629e-01
3.03497791e-01 9.93984640e-01 -7.03746796e-01 -3.08209538e-01
-7.24337876e-01 1.02624464e+00 -2.88997084e-01 3.49040270e-01
-6.04704618e-01 -3.41427207e-01 6.34960055e-01 1.73704118e-01
4.08497423e-01 -6.90897346e-01 2.36673355e-02 -1.29290557e+00
2.42624581e-01 -1.97446316e-01 3.39711547e-01 -9.64438260e-01
-9.00831521e-01 2.30123937e-01 2.74812561e-02 -1.67355943e+00
-7.12266982e-01 -6.77978098e-01 -6.22466147e-01 7.19153106e-01
-1.59795702e+00 -7.94150651e-01 -5.79678595e-01 1.12387776e+00
5.19513309e-01 1.24530926e-01 4.55606908e-01 2.56631106e-01
-1.19868271e-01 -1.89475894e-01 2.36244038e-01 -5.97332083e-02
6.18196070e-01 -1.13101745e+00 2.30766274e-02 1.15934896e+00
1.49582714e-01 1.95734009e-01 9.51876879e-01 -4.46755677e-01
-1.88715506e+00 -6.88316405e-01 7.91957021e-01 -6.55600965e-01
6.70942605e-01 7.99351260e-02 -7.47196019e-01 3.82052094e-01
-2.90457100e-01 5.26270628e-01 1.03892505e-01 -6.97529912e-01
7.30744749e-02 -2.97956556e-01 -1.00107670e+00 3.07438403e-01
8.79880011e-01 -5.36207736e-01 -5.61919868e-01 5.49298190e-02
4.14059192e-01 -4.28867638e-01 -5.07893682e-01 6.75304383e-02
7.35848010e-01 -1.29824042e+00 1.19616604e+00 1.54010085e-02
-1.73550531e-01 -7.52947867e-01 1.01194955e-01 -7.34937429e-01
-7.62630329e-02 -9.35227573e-01 -6.84079111e-01 8.91010523e-01
-3.47063601e-01 -6.80082381e-01 7.28682995e-01 5.75454772e-01
1.29789889e-01 -1.49432480e-01 -1.17243326e+00 -9.03401613e-01
-9.29150581e-01 -4.57581609e-01 -7.69925714e-02 6.19635403e-01
-2.92562068e-01 -1.98690388e-02 -3.08156312e-01 1.59589723e-01
1.10739768e+00 1.74209908e-01 7.79284894e-01 -1.24973249e+00
-3.35839629e-01 -2.52848446e-01 -1.06069672e+00 -1.08006513e+00
-3.12440485e-01 -1.52299628e-01 8.99955332e-02 -1.22777712e+00
-2.21876070e-01 -2.35166904e-02 1.81715578e-01 -2.57751375e-01
6.55258149e-02 2.90338814e-01 2.56314844e-01 4.91149783e-01
-3.02656800e-01 1.91664711e-01 1.02267599e+00 2.84248352e-01
-3.15493047e-01 1.38252661e-01 4.52064544e-01 9.01202559e-01
2.71133214e-01 -2.08027095e-01 -3.78581226e-01 -3.79149079e-01
5.67870885e-02 4.40157950e-01 7.73625374e-01 -1.16990137e+00
8.42238128e-01 -1.93221807e-01 4.77306306e-01 -7.60833800e-01
7.29128540e-01 -1.21563923e+00 5.70471048e-01 6.32725954e-01
1.65203020e-01 2.51341820e-01 -5.60530052e-02 9.68021572e-01
-2.49967471e-01 -3.11458260e-01 8.98660004e-01 -3.31094921e-01
-1.16914165e+00 3.53144318e-01 -6.22561634e-01 -3.64932626e-01
1.28768027e+00 -9.00497556e-01 -1.99733526e-01 -3.79923820e-01
-7.56401002e-01 -1.18964508e-01 7.13624775e-01 1.79648167e-04
6.71580315e-01 -1.10886252e+00 -4.96111214e-01 3.61775517e-01
-6.47982806e-02 3.29730706e-03 1.74607277e-01 9.59777713e-01
-8.85613203e-01 1.96259812e-01 -3.50198179e-01 -1.04218233e+00
-1.34990466e+00 6.92317307e-01 3.64699364e-01 1.84440553e-01
-1.11010373e+00 3.68122637e-01 4.93083298e-02 6.08183980e-01
1.59237370e-01 -2.35431910e-01 -1.95751622e-01 2.36894190e-01
8.85164142e-01 7.67213881e-01 -3.25504169e-02 -6.78424835e-01
-3.79755139e-01 9.64702129e-01 5.36410928e-01 -2.79943913e-01
8.78246963e-01 -5.44090807e-01 6.58865422e-02 6.34754002e-01
1.04095924e+00 4.34757918e-02 -1.58306384e+00 -1.43653110e-01
-6.00887872e-02 -7.34025896e-01 1.21962115e-01 -5.19918837e-02
-7.83140481e-01 8.62450302e-01 7.13836551e-01 4.62214798e-01
1.29886568e+00 -3.06312144e-01 7.89133072e-01 2.21660182e-01
4.41508621e-01 -8.79474878e-01 -1.26111850e-01 4.53833967e-01
3.15099329e-01 -1.06839991e+00 -8.48566592e-02 -3.96528363e-01
-1.00051902e-01 1.45669115e+00 2.18725517e-01 -1.60264343e-01
5.25125504e-01 5.60881853e-01 9.48965549e-02 6.17880002e-02
-7.88380086e-01 -5.25480926e-01 -3.89347523e-02 9.80600834e-01
-1.62905402e-04 -3.06074411e-01 -2.14170098e-01 -5.38600504e-01
3.65062803e-01 2.17166021e-01 8.57661366e-01 1.07029462e+00
-5.18043458e-01 -7.58634269e-01 -7.54318714e-01 -3.50178480e-02
-3.48590195e-01 3.42753679e-01 4.07413989e-02 6.28999531e-01
-6.49261028e-02 8.00479233e-01 3.04313123e-01 9.53587815e-02
3.71685892e-01 -1.12251803e-01 7.27003038e-01 -1.68442503e-01
-3.01851213e-01 2.04223990e-01 -9.63253677e-02 -1.01882279e+00
-8.28426301e-01 -9.58536565e-01 -1.18803632e+00 -3.23465198e-01
-1.86422169e-01 -2.65674084e-01 9.12239611e-01 6.96734130e-01
1.82444215e-01 1.22492611e-01 6.55082345e-01 -1.14797652e+00
-3.32249582e-01 -3.51041019e-01 -4.88051683e-01 5.33289611e-01
9.20858204e-01 -5.60992658e-01 -4.98581290e-01 5.17893195e-01]
|
[8.575807571411133, -1.3782639503479004]
|
fec7425a-4b16-4708-9a8a-db21fec76aca
|
gfnet-geometric-flow-network-for-3d-point
|
2207.02605
| null |
https://arxiv.org/abs/2207.02605v2
|
https://arxiv.org/pdf/2207.02605v2.pdf
|
GFNet: Geometric Flow Network for 3D Point Cloud Semantic Segmentation
|
Point cloud semantic segmentation from projected views, such as range-view (RV) and bird's-eye-view (BEV), has been intensively investigated. Different views capture different information of point clouds and thus are complementary to each other. However, recent projection-based methods for point cloud semantic segmentation usually utilize a vanilla late fusion strategy for the predictions of different views, failing to explore the complementary information from a geometric perspective during the representation learning. In this paper, we introduce a geometric flow network (GFNet) to explore the geometric correspondence between different views in an align-before-fuse manner. Specifically, we devise a novel geometric flow module (GFM) to bidirectionally align and propagate the complementary information across different views according to geometric relationships under the end-to-end learning scheme. We perform extensive experiments on two widely used benchmark datasets, SemanticKITTI and nuScenes, to demonstrate the effectiveness of our GFNet for project-based point cloud semantic segmentation. Concretely, GFNet not only significantly boosts the performance of each individual view but also achieves state-of-the-art results over all existing projection-based models. Code is available at \url{https://github.com/haibo-qiu/GFNet}.
|
['DaCheng Tao', 'Baosheng Yu', 'Haibo Qiu']
|
2022-07-06
| null | null | null | null |
['robust-3d-semantic-segmentation', 'lidar-semantic-segmentation']
|
['computer-vision', 'computer-vision']
|
[-5.27251624e-02 -1.69062793e-01 5.17312139e-02 -5.88655114e-01
-5.00439346e-01 -6.81841671e-01 4.98551220e-01 -2.11452767e-01
1.14673346e-01 3.09500583e-02 -7.42811561e-02 -2.58593589e-01
2.48249415e-02 -1.04239988e+00 -8.96600664e-01 -3.71831745e-01
4.65248227e-01 5.49645543e-01 5.20283759e-01 -2.16023296e-01
3.37070704e-01 5.87997913e-01 -1.44705939e+00 2.95796335e-01
9.64727938e-01 8.59904647e-01 4.73230153e-01 3.02987456e-01
-4.17915702e-01 2.91957170e-01 2.38052122e-02 -5.16680121e-01
5.33305407e-01 9.33810249e-02 -7.58051097e-01 2.52218515e-01
7.40955234e-01 -1.71098426e-01 -2.45882988e-01 1.04898643e+00
2.18023777e-01 2.99442038e-02 2.80400962e-01 -1.36366105e+00
-4.28396106e-01 1.23635538e-01 -1.03993785e+00 8.25770795e-02
3.32844436e-01 2.64122695e-01 9.96589899e-01 -1.10574150e+00
8.76890242e-01 1.30401421e+00 4.36454147e-01 4.31556523e-01
-9.74366248e-01 -8.56823266e-01 7.02511489e-01 1.69530258e-01
-1.17891967e+00 5.12067229e-03 1.19660997e+00 -5.01562893e-01
6.84923053e-01 2.79272050e-01 9.27618742e-01 7.32393086e-01
-2.09984221e-02 9.97087181e-01 1.03678608e+00 7.18977228e-02
-1.19605795e-01 -8.80154371e-02 9.24305338e-03 6.56393290e-01
-6.30492065e-03 2.58060813e-01 -5.93083441e-01 6.47179857e-02
6.90857530e-01 4.55795079e-01 -4.83940572e-01 -8.29092324e-01
-1.17801416e+00 6.91568434e-01 7.78259873e-01 6.33965852e-03
-3.12463075e-01 2.05296546e-01 -3.58875096e-02 -1.11393332e-01
8.23405445e-01 -2.77368650e-02 -4.78862882e-01 2.58312404e-01
-8.07811379e-01 1.99446008e-01 5.84401608e-01 1.13395000e+00
9.37303066e-01 -1.63273931e-01 1.70607522e-01 6.93263352e-01
7.58023560e-01 5.57959795e-01 -2.14044631e-01 -1.07107008e+00
9.75534141e-01 8.75340700e-01 -2.54203618e-01 -1.10734069e+00
-2.59906709e-01 -4.33439553e-01 -5.68102181e-01 2.99749553e-01
1.97228953e-01 8.33435357e-02 -1.06214535e+00 1.34076345e+00
7.70555496e-01 4.91244704e-01 -1.18181586e-01 1.11467123e+00
1.01323080e+00 6.99853122e-01 -1.03444234e-01 3.25199693e-01
1.23296237e+00 -1.26365638e+00 -8.28492045e-02 -4.59011823e-01
2.51925796e-01 -8.75007033e-01 8.16772044e-01 3.29470068e-01
-1.17562175e+00 -5.19777894e-01 -9.12967324e-01 -2.76147842e-01
-2.76419103e-01 -6.50008768e-02 6.07741535e-01 3.03637683e-01
-9.35622454e-01 5.81645846e-01 -1.06621873e+00 -2.86785930e-01
7.57736683e-01 1.64190799e-01 -3.97242010e-01 -5.63170433e-01
-6.29807115e-01 4.12319571e-01 2.70086527e-01 2.63490707e-01
-5.70627630e-01 -1.05577385e+00 -7.42157340e-01 9.59053710e-02
4.84097958e-01 -1.17966580e+00 9.10190523e-01 -6.56027138e-01
-1.09649754e+00 9.20638680e-01 -2.74166048e-01 6.13543205e-02
6.85690284e-01 -3.20475489e-01 -1.29819110e-01 3.71697217e-01
2.14257762e-01 1.00132203e+00 4.51803207e-01 -1.78892195e+00
-9.33616519e-01 -7.84642339e-01 1.13808371e-01 5.56836784e-01
1.24492615e-01 -2.01836780e-01 -9.89032030e-01 -3.93666208e-01
7.00176120e-01 -9.67250586e-01 -1.64684996e-01 2.86407053e-01
-5.34781873e-01 -1.84293166e-01 1.00223887e+00 -6.80679202e-01
7.66279399e-01 -1.93261504e+00 2.51836419e-01 1.07369550e-01
3.86884511e-01 4.35239412e-02 5.97958341e-02 3.33632469e-01
-4.75346856e-02 8.79937597e-03 -5.10707021e-01 -5.56554973e-01
-2.36877009e-01 1.57660767e-01 -1.98336571e-01 5.55716872e-01
-3.26086618e-02 1.04614139e+00 -8.98128510e-01 -4.28232223e-01
7.92851031e-01 5.67274153e-01 -5.72446406e-01 8.19614753e-02
-2.53022581e-01 8.52026284e-01 -5.64039469e-01 8.63693476e-01
1.24342179e+00 -3.65144849e-01 -2.37833574e-01 -3.12848359e-01
-1.77768439e-01 -1.26509396e-02 -1.01339328e+00 2.15842557e+00
-5.18121421e-01 3.97546083e-01 6.38206780e-04 -6.20279253e-01
9.02559221e-01 2.07069293e-02 7.28431821e-01 -5.51993310e-01
-3.41771767e-02 1.61830232e-01 -3.27418685e-01 -2.02404723e-01
2.36506000e-01 -9.27477553e-02 1.70264453e-01 1.03601985e-01
-5.35662957e-02 -4.26963776e-01 3.80856320e-02 2.52891064e-01
4.64300901e-01 4.55108762e-01 -2.45976709e-02 2.62997057e-02
7.67809093e-01 5.32257333e-02 6.13146722e-01 2.34242752e-01
9.74551290e-02 1.17885506e+00 2.76732922e-01 -4.07232672e-01
-8.31137240e-01 -1.29926991e+00 4.56188992e-03 6.46000862e-01
8.79559100e-01 -2.69386917e-01 -5.81738830e-01 -9.35636818e-01
1.83085307e-01 9.35324907e-01 -2.41376117e-01 2.11658940e-01
-6.91463828e-01 -4.55089569e-01 -1.94193479e-02 7.30841637e-01
5.00206292e-01 -7.53250837e-01 -5.35689473e-01 -3.59712280e-02
-3.54862243e-01 -1.39659810e+00 -6.03447914e-01 -2.52922744e-01
-1.22325146e+00 -1.20876276e+00 -6.17565274e-01 -4.38430667e-01
7.22163975e-01 8.63194346e-01 1.24497962e+00 4.56171595e-02
8.36193487e-02 5.49251080e-01 -2.74998188e-01 -2.37854123e-01
1.00552261e-01 -8.58985074e-03 -5.55449247e-01 -3.70361917e-02
4.32060659e-01 -8.01885724e-01 -8.94366384e-01 6.14538372e-01
-7.27600574e-01 5.48861265e-01 3.95280719e-01 4.38012719e-01
9.34010506e-01 -3.40562612e-01 -1.32554397e-01 -9.95196164e-01
-1.21466264e-01 -6.26583040e-01 -8.17876637e-01 2.42818490e-01
-5.13148665e-01 -3.67753804e-01 2.78975070e-01 2.92836577e-01
-1.14233148e+00 2.90999889e-01 -3.21215898e-01 -1.07134044e+00
-3.03296238e-01 1.70513213e-01 -4.64211136e-01 -1.95602909e-01
-2.38888953e-02 2.93410897e-01 -1.65290803e-01 -5.57234645e-01
6.46955431e-01 3.05825830e-01 5.08916318e-01 -5.07043362e-01
9.37108457e-01 1.04087269e+00 -7.35387579e-02 -5.32978773e-01
-8.50044847e-01 -1.00274074e+00 -8.71131659e-01 -4.24800545e-01
1.07886982e+00 -1.08850884e+00 -5.94604492e-01 3.79423857e-01
-1.34385037e+00 9.86204743e-02 -7.40415901e-02 3.71604741e-01
-5.86432576e-01 4.84203160e-01 -2.44598731e-01 -3.75016689e-01
-4.54645425e-01 -1.38910651e+00 1.42624128e+00 4.44467247e-01
2.93749064e-01 -9.69211698e-01 -1.24877296e-01 8.77174616e-01
-1.04853936e-01 3.49555790e-01 5.84720433e-01 -2.45558858e-01
-1.17710221e+00 1.16402976e-01 -5.24281085e-01 3.49285543e-01
-9.81277227e-02 -6.85745943e-03 -1.10938036e+00 -1.92129821e-01
1.25284895e-01 3.26623291e-01 9.93752241e-01 4.06766117e-01
1.29313338e+00 2.04778388e-01 -6.12995565e-01 1.19208157e+00
1.77137172e+00 1.48315609e-01 6.00507319e-01 2.42971703e-01
1.43849957e+00 7.68986583e-01 6.27745390e-01 1.66126683e-01
7.21121788e-01 7.12893486e-01 8.18123758e-01 -1.63539082e-01
-1.77042574e-01 -5.12345850e-01 -8.10014531e-02 8.31267715e-01
-1.76427037e-01 -4.52921212e-01 -1.04360485e+00 4.42759365e-01
-1.79960036e+00 -6.76857769e-01 -5.32803416e-01 1.98133290e+00
3.32671255e-02 1.01895686e-02 -1.34705350e-01 -3.19246829e-01
7.37491369e-01 3.62761915e-01 -6.65329039e-01 -5.62862027e-04
3.24966535e-02 5.21849245e-02 6.92169428e-01 3.74882936e-01
-1.06391847e+00 1.11209142e+00 4.28401232e+00 6.92437410e-01
-1.12264228e+00 3.66511196e-01 6.37436867e-01 -9.25365388e-02
-6.05456710e-01 2.83141375e-01 -7.64493287e-01 4.85349834e-01
1.24141432e-01 9.44334567e-02 2.50292361e-01 9.00121987e-01
4.58000749e-02 -1.22525096e-01 -9.55736220e-01 1.11836362e+00
5.85864447e-02 -1.46579230e+00 3.47499996e-02 1.27001300e-01
8.92268240e-01 5.02797663e-01 1.74755715e-02 -1.25168562e-01
1.61781713e-01 -6.87130451e-01 8.72966945e-01 5.45693040e-01
4.53910828e-01 -7.37626731e-01 5.29308796e-01 3.62081528e-01
-1.57683086e+00 9.48127136e-02 -2.28654772e-01 3.93069953e-01
6.36630297e-01 5.92367649e-01 -4.02611405e-01 1.26854551e+00
8.14683914e-01 1.13510692e+00 -5.87709129e-01 1.10630786e+00
-2.85481602e-01 2.77174860e-01 -3.82642686e-01 5.66648543e-01
4.78158891e-01 -6.39573514e-01 7.83067405e-01 7.39967346e-01
4.52775925e-01 1.16769917e-01 2.84439564e-01 1.24086976e+00
3.77617292e-02 -3.86461243e-02 -5.69553733e-01 2.96154261e-01
3.17851663e-01 1.34228671e+00 -1.06056821e+00 -2.42385268e-01
-7.04522610e-01 8.24935436e-01 1.55084431e-01 3.26886922e-01
-8.63613129e-01 1.32028043e-01 6.78703010e-01 3.37110162e-01
5.53719103e-01 -1.87467411e-01 -6.31862640e-01 -1.25706494e+00
2.70459265e-01 -3.66530210e-01 2.94474691e-01 -9.88004267e-01
-1.33214653e+00 5.23506045e-01 2.06879735e-01 -1.67772543e+00
1.66377485e-01 -3.99512678e-01 -7.18446732e-01 1.12438500e+00
-1.65495336e+00 -1.45756102e+00 -6.19235218e-01 4.97147083e-01
8.27821791e-01 6.53659701e-02 1.09566495e-01 2.86966383e-01
-2.62188524e-01 3.27139199e-02 -5.68461418e-02 -1.00244984e-01
3.21463585e-01 -1.23870397e+00 7.24005342e-01 9.52134490e-01
4.07862782e-01 3.79933804e-01 2.90064305e-01 -7.58744121e-01
-1.24306691e+00 -1.23245168e+00 4.17246342e-01 -6.28813863e-01
1.50747880e-01 -2.95174122e-01 -9.96296942e-01 5.91934443e-01
1.08412169e-01 2.70249337e-01 2.88661957e-01 -1.49347395e-01
-3.27729166e-01 -7.91318044e-02 -9.63124037e-01 2.84724534e-01
1.29072034e+00 -3.20639253e-01 -3.64006758e-01 3.61921370e-01
9.22822595e-01 -7.79888988e-01 -7.10023105e-01 6.18601561e-01
4.10327375e-01 -1.35881484e+00 1.37819719e+00 -2.15452835e-01
6.35701716e-01 -5.95828235e-01 -2.47420684e-01 -1.21671128e+00
-4.62671258e-02 -1.24749273e-01 -2.94544883e-02 1.13105607e+00
1.54758379e-01 -7.03878224e-01 1.03667724e+00 3.61302167e-01
-5.09308457e-01 -1.16413164e+00 -7.26410627e-01 -5.60573578e-01
7.81541616e-02 -6.65148854e-01 8.10016632e-01 8.41208816e-01
-6.45075500e-01 2.31052443e-01 6.89472957e-03 4.82138157e-01
6.75541818e-01 5.73330402e-01 8.77067447e-01 -1.36442983e+00
-2.21702144e-01 -5.44615388e-01 -5.49659431e-01 -1.24605584e+00
3.99615653e-02 -1.22810113e+00 -1.90419823e-01 -1.94170952e+00
9.27893594e-02 -5.10507286e-01 -8.79694894e-02 1.63268939e-01
-2.80441999e-01 1.33507967e-01 6.72662854e-01 2.80748427e-01
-4.47869569e-01 6.06102705e-01 1.55957019e+00 -1.05159193e-01
-9.13425311e-02 2.47453690e-01 -6.85306609e-01 9.72343087e-01
5.99962711e-01 -5.25635362e-01 -5.13371468e-01 -7.76731789e-01
9.10113826e-02 2.02638909e-01 6.85173392e-01 -1.06365192e+00
3.22449267e-01 -1.67598084e-01 4.22910631e-01 -1.32013261e+00
4.91831124e-01 -8.98322284e-01 3.82771641e-01 1.45301148e-01
2.13390231e-01 1.65219650e-01 1.20944962e-01 9.43884492e-01
-3.35547239e-01 3.67164314e-02 6.64924860e-01 -3.76544356e-01
-8.68239105e-01 7.52790749e-01 5.65297067e-01 6.00727685e-02
1.11695290e+00 -4.63093311e-01 -2.49681935e-01 -1.81486066e-02
-6.17146552e-01 7.11370170e-01 8.41278732e-01 6.81930780e-01
8.88134837e-01 -1.05679846e+00 -5.24559975e-01 2.18512625e-01
2.64413565e-01 7.39840448e-01 7.99796641e-01 9.05282319e-01
-8.05843771e-01 3.71349514e-01 -1.29773080e-01 -1.21367681e+00
-1.29518950e+00 4.91980940e-01 3.41442198e-01 -6.30424097e-02
-9.80744421e-01 1.09491146e+00 8.87208462e-01 -8.80078733e-01
-1.56544194e-01 -2.67918527e-01 -1.54696867e-01 -1.24105282e-01
1.94176752e-02 4.01344866e-01 1.34403080e-01 -8.28718007e-01
-4.03042316e-01 1.11876512e+00 -8.99349228e-02 -6.36392906e-02
1.31919634e+00 -2.28040442e-01 -3.40865999e-02 2.76489586e-01
1.25633061e+00 -1.56158447e-01 -1.49276590e+00 -2.77080119e-01
-3.11340511e-01 -9.66854453e-01 1.71982437e-01 -5.13369739e-01
-1.79510295e+00 1.24084747e+00 5.82225800e-01 -2.65525937e-01
1.02057147e+00 1.57253876e-01 9.96065795e-01 -2.85363168e-01
5.37174761e-01 -7.58586168e-01 -1.31046042e-01 1.99485466e-01
7.23107994e-01 -1.25872993e+00 2.23594084e-01 -1.07539558e+00
-7.53149807e-01 1.16081297e+00 8.75340760e-01 -1.63681865e-01
7.34663069e-01 -1.95065469e-01 -9.84212477e-03 -6.77033544e-01
-5.31707287e-01 -1.93920374e-01 5.53131163e-01 5.07423937e-01
7.17219114e-02 1.55504093e-01 -4.71235206e-03 1.14224851e-01
-1.56759888e-01 -1.23015299e-01 1.47658587e-01 8.79004121e-01
-8.06510821e-02 -9.65359747e-01 -3.68872434e-01 3.48169744e-01
2.17690319e-02 1.14188828e-01 -2.80618310e-01 7.75816858e-01
3.36243600e-01 7.80128717e-01 3.15665007e-01 -3.73970717e-01
4.50392067e-01 -2.02334583e-01 4.23156381e-01 -6.69627130e-01
-4.64515746e-01 2.18903616e-01 -3.28280658e-01 -8.96189630e-01
-5.88623405e-01 -7.93438971e-01 -1.23583066e+00 -2.94560969e-01
-2.55613118e-01 -2.69166470e-01 7.46033192e-01 8.38483036e-01
4.81653512e-01 5.38720965e-01 7.29760945e-01 -1.08503830e+00
-2.50206608e-02 -5.21976411e-01 -3.71855706e-01 3.79025310e-01
-1.07352011e-01 -8.32770586e-01 -2.34555349e-01 -1.26198709e-01]
|
[8.200126647949219, -3.021498203277588]
|
07849698-9add-4bd1-be93-b202e62b2dfe
|
feddct-a-dynamic-cross-tier-federated
|
2307.04420
| null |
https://arxiv.org/abs/2307.04420v1
|
https://arxiv.org/pdf/2307.04420v1.pdf
|
FedDCT: A Dynamic Cross-Tier Federated Learning Scheme in Wireless Communication Networks
|
With the rapid proliferation of Internet of Things (IoT) devices and the growing concern for data privacy among the public, Federated Learning (FL) has gained significant attention as a privacy-preserving machine learning paradigm. FL enables the training of a global model among clients without exposing local data. However, when a federated learning system runs on wireless communication networks, limited wireless resources, heterogeneity of clients, and network transmission failures affect its performance and accuracy. In this study, we propose a novel dynamic cross-tier FL scheme, named FedDCT to increase training accuracy and performance in wireless communication networks. We utilize a tiering algorithm that dynamically divides clients into different tiers according to specific indicators and assigns specific timeout thresholds to each tier to reduce the training time required. To improve the accuracy of the model without increasing the training time, we introduce a cross-tier client selection algorithm that can effectively select the tiers and participants. Simulation experiments show that our scheme can make the model converge faster and achieve a higher accuracy in wireless communication networks.
|
['Dongcheng Li', 'Jianyong Jiang', 'Lianghaojie Zhou', 'Xiaoyun Gan', 'Chuanjian Yao', 'Youquan Xian', 'Peng Liu']
|
2023-07-10
| null | null | null | null |
['federated-learning']
|
['methodology']
|
[-2.14485019e-01 -4.88897562e-02 -6.29577041e-01 -7.06321836e-01
-4.34069246e-01 -7.39040852e-01 4.35112268e-02 -9.56451371e-02
-2.97070265e-01 6.95113540e-01 3.83999236e-02 -5.55092394e-01
-4.19131935e-01 -9.49404538e-01 -4.41949010e-01 -8.94153893e-01
-1.92094687e-02 1.82987452e-01 2.34087899e-01 5.98630130e-01
-2.56670892e-01 3.37004036e-01 -9.96749520e-01 4.08143342e-01
9.01484609e-01 1.42197371e+00 -2.59149134e-01 4.02885042e-02
-3.94830883e-01 6.55026317e-01 -4.20234054e-01 -6.62187815e-01
6.56934798e-01 -9.58211720e-02 -7.57660925e-01 -1.23972818e-01
-2.72164553e-01 -4.80842769e-01 -3.59052569e-01 1.08125770e+00
5.15034199e-01 -3.13560367e-01 -3.40707786e-02 -1.75983059e+00
-1.82953835e-01 1.00136268e+00 -3.56027961e-01 -2.41074994e-01
-1.21956170e-02 -2.04136282e-01 5.19008458e-01 -1.85473740e-01
3.57061028e-01 9.31487739e-01 6.91693604e-01 7.99974501e-01
-1.03492367e+00 -1.46229517e+00 4.09986466e-01 2.87843883e-01
-1.30009437e+00 -4.38121438e-01 7.58120298e-01 1.72821924e-01
-5.48716821e-03 5.31827211e-01 4.12290037e-01 7.66960919e-01
-1.00193247e-02 6.58795953e-01 8.52048516e-01 -1.54386178e-01
7.27753997e-01 3.82238477e-01 -2.27138326e-01 3.69580388e-01
5.38978100e-01 -5.69113046e-02 -4.46536779e-01 -8.12651813e-01
5.21820366e-01 6.30797148e-01 -2.49716744e-01 -9.06888664e-01
-8.26384544e-01 4.52972859e-01 4.06017274e-01 1.75762564e-01
-4.56707388e-01 1.15104549e-01 4.00849849e-01 2.97381520e-01
2.16071501e-01 -4.23027068e-01 -8.53736281e-01 1.49578914e-01
-7.36739457e-01 -3.36797863e-01 1.05233741e+00 1.05981398e+00
7.03899026e-01 -4.05504435e-01 -6.95607439e-03 4.17265087e-01
3.69971722e-01 2.92858630e-01 3.49701792e-01 -1.08495712e+00
5.69403887e-01 9.06287014e-01 4.28118184e-02 -6.80528402e-01
-1.00775160e-01 -4.57350075e-01 -1.13990629e+00 -7.01250583e-02
2.03093573e-01 -6.49569511e-01 -3.41825426e-01 1.89471328e+00
7.91336834e-01 1.58203095e-01 1.50358155e-01 7.62968242e-01
2.11977381e-02 5.87219954e-01 3.51091087e-01 -5.54459095e-01
1.02406728e+00 -5.76687992e-01 -6.46167219e-01 2.26436034e-01
6.67918205e-01 -2.41349071e-01 4.83513892e-01 2.27665663e-01
-6.30274892e-01 2.06332579e-01 -7.49028325e-01 4.92783070e-01
-2.03662738e-01 -1.70450151e-01 9.23952818e-01 1.05441928e+00
-6.58792317e-01 1.50942951e-01 -9.25884604e-01 -4.24162090e-01
9.09155905e-01 8.68492484e-01 -3.01046759e-01 -2.28778124e-01
-9.84310150e-01 -2.89204661e-02 1.27572253e-01 -2.68330216e-01
-6.33126736e-01 -7.65893579e-01 -3.38850409e-01 1.45592853e-01
2.73084790e-01 -5.95323563e-01 1.20857930e+00 -8.23143780e-01
-1.30769598e+00 3.33193541e-01 -1.45984516e-02 -5.23971796e-01
7.40420997e-01 3.65014106e-01 -7.04478264e-01 -8.39640796e-02
-2.58742005e-01 1.48683805e-02 4.82818604e-01 -1.32125568e+00
-1.02635288e+00 -6.80076718e-01 1.10713057e-01 -1.97160900e-01
-9.67432141e-01 2.07186192e-02 -3.45657051e-01 -1.60861403e-01
2.65812755e-01 -7.13667274e-01 -4.00402814e-01 5.87048054e-01
-2.58012265e-01 -1.82749882e-01 1.43688846e+00 -1.95062742e-01
1.37104678e+00 -2.20648980e+00 -4.94422227e-01 6.41436040e-01
3.60927641e-01 1.55013859e-01 2.03404069e-01 2.26486906e-01
4.13739800e-01 2.68677711e-01 -1.03082985e-01 -1.33506402e-01
-4.48240153e-02 5.11660159e-01 2.79691052e-02 3.77651811e-01
-6.76502883e-01 4.13302451e-01 -5.71273267e-01 -6.71733379e-01
-1.57473609e-02 3.54342014e-01 -6.78313315e-01 2.24807546e-01
-9.35667455e-02 5.53624630e-01 -1.07898152e+00 5.93236148e-01
1.06667244e+00 -4.60612327e-01 6.83659434e-01 -2.18933135e-01
-1.01074651e-02 -3.07203736e-02 -1.25568032e+00 1.23070848e+00
-6.26893699e-01 -1.72455966e-01 8.51130128e-01 -8.07322681e-01
8.80654514e-01 6.62301481e-01 9.17098641e-01 -5.46901703e-01
2.98872411e-01 1.97119072e-01 -3.64124984e-01 -4.37796265e-01
-4.59425092e-01 1.92405522e-01 -7.75026381e-02 7.05522180e-01
-3.61281693e-01 8.70762706e-01 -5.64906061e-01 3.41922194e-02
1.24471152e+00 -5.32873333e-01 8.54539350e-02 -2.24021837e-01
5.35050035e-01 -3.76865357e-01 1.05283570e+00 6.58568919e-01
-4.35923100e-01 -5.14931651e-03 1.78016409e-01 -7.80256629e-01
-5.26844978e-01 -8.67304444e-01 4.79660220e-02 1.06579971e+00
3.63398492e-01 -2.46260971e-01 -8.30115378e-01 -1.01735497e+00
2.27308691e-01 4.25154954e-01 -2.93549150e-01 -3.35777551e-01
-3.12783957e-01 -6.61771953e-01 5.00442505e-01 2.20204815e-02
1.06097388e+00 -6.39825344e-01 -3.79854709e-01 2.25545526e-01
-2.71804184e-01 -1.01639664e+00 -5.76955914e-01 3.18360329e-01
-8.67366016e-01 -1.16470408e+00 -1.84999451e-01 -7.01875329e-01
8.72760534e-01 3.44983250e-01 4.43595499e-01 1.28087942e-02
9.49211121e-02 3.65670085e-01 -2.05556408e-01 -4.21252578e-01
-2.94041932e-01 4.15005326e-01 1.11146411e-02 5.51750600e-01
3.53766769e-01 -1.01575756e+00 -9.10932779e-01 7.10056603e-01
-1.05726826e+00 -1.97070614e-01 4.92291391e-01 2.57674098e-01
3.12917292e-01 3.31973672e-01 5.92879832e-01 -9.99671519e-01
4.41263765e-01 -6.89293802e-01 -7.77731001e-01 6.13087177e-01
-1.05244148e+00 5.70952669e-02 8.88876677e-01 -6.05316043e-01
-1.01987028e+00 2.77785093e-01 3.70724231e-01 -2.62795269e-01
1.03061691e-01 2.68154573e-02 -8.34432602e-01 -5.19603968e-01
3.50274205e-01 7.53227696e-02 2.22014990e-02 -7.57411659e-01
1.46190673e-01 1.18603134e+00 1.50207967e-01 -4.95567948e-01
9.01448369e-01 5.74442744e-01 -7.12640360e-02 -1.35441810e-01
-5.22926390e-01 -1.42556682e-01 -8.53107497e-02 -8.71478394e-02
3.36302608e-01 -8.83413136e-01 -1.24387228e+00 2.68125951e-01
-9.09577727e-01 -4.52327468e-02 -4.48895916e-02 4.31368023e-01
1.12457395e-01 5.28221577e-02 -3.48707855e-01 -8.82943392e-01
-8.70206118e-01 -7.56125152e-01 1.92883670e-01 4.34790134e-01
1.05572440e-01 -7.67192185e-01 -3.42319727e-01 2.98563570e-01
8.86044323e-01 1.69830173e-01 8.51452231e-01 -7.14771748e-01
-8.61815035e-01 -3.66081119e-01 -2.42212549e-01 4.18240614e-02
3.68253946e-01 -4.88252252e-01 -8.17248046e-01 -4.42815900e-01
1.72152892e-02 8.87953565e-02 1.96685001e-01 2.48215981e-02
1.92132044e+00 -1.26516140e+00 -6.90724432e-01 9.27160680e-01
1.44746578e+00 3.78257066e-01 2.61754572e-01 2.44428858e-01
5.05169392e-01 4.29599524e-01 1.50759578e-01 8.97734344e-01
5.74297726e-01 2.88922936e-01 7.34673142e-01 5.84844351e-02
4.27246034e-01 -3.50748241e-01 1.57983720e-01 3.41367036e-01
4.36796039e-01 -3.07915717e-01 -4.40198272e-01 4.07798737e-01
-1.92750168e+00 -7.45849013e-01 4.41045582e-01 2.37801433e+00
6.86293304e-01 -9.29601863e-02 1.69571027e-01 2.02261403e-01
8.77766013e-01 -3.65499072e-02 -9.60646212e-01 -1.56511009e-01
2.24245831e-01 -2.51876026e-01 1.00976777e+00 -4.80813757e-02
-7.16228962e-01 3.26772749e-01 5.37568760e+00 7.33073354e-01
-1.32919753e+00 4.05819833e-01 7.50557065e-01 -5.79047054e-02
-3.33699346e-01 8.17468837e-02 -3.78819585e-01 5.74017704e-01
7.93945193e-01 -4.10414934e-01 7.50466526e-01 1.23932552e+00
2.61831790e-01 5.22316575e-01 -9.41661358e-01 1.01781881e+00
-7.49611378e-01 -1.32138956e+00 1.26157654e-02 3.55942875e-01
5.99320114e-01 1.29687220e-01 -2.10941315e-01 -4.33366410e-02
5.08536935e-01 -5.25256574e-01 4.87769961e-01 1.78919375e-01
6.34116709e-01 -1.07749343e+00 4.97210711e-01 5.90669453e-01
-1.16089725e+00 -7.09755063e-01 -2.60622948e-01 2.68300682e-01
-2.12920263e-01 6.45191252e-01 -2.59054095e-01 5.41521430e-01
1.00288379e+00 1.95560992e-01 -2.17423141e-01 1.19625127e+00
4.48746860e-01 8.08023274e-01 -6.74498558e-01 -1.32041216e-01
-1.64862037e-01 -9.47915837e-02 2.51445830e-01 6.61837339e-01
2.08467111e-01 3.34154487e-01 2.99987525e-01 4.97833580e-01
-5.53251624e-01 6.54980615e-02 -2.39056930e-01 3.07910889e-01
1.14113986e+00 1.29818082e+00 -7.43855089e-02 1.21138521e-01
-4.45319504e-01 7.79122770e-01 1.17423102e-01 3.59838933e-01
-5.86194158e-01 -3.52479696e-01 1.00052989e+00 3.27411562e-01
2.21222013e-01 1.16473302e-01 -3.79982471e-01 -8.79905641e-01
2.18319520e-01 -6.26612484e-01 8.06402922e-01 -1.06258795e-01
-1.26360559e+00 4.64572966e-01 -3.44950557e-01 -1.23715639e+00
3.12222540e-01 1.39147788e-01 -6.99406087e-01 3.37802947e-01
-1.27192926e+00 -1.12227750e+00 -2.61158943e-01 1.22321248e+00
-2.80844778e-01 1.68759506e-02 9.45934415e-01 5.18037498e-01
-6.98439121e-01 1.21188962e+00 4.18294817e-01 2.13237301e-01
3.59686166e-01 -4.55711842e-01 -2.30261281e-01 7.00741291e-01
-2.36988395e-01 6.54693604e-01 2.23379001e-01 -2.88326919e-01
-1.48273265e+00 -1.55852866e+00 7.33938873e-01 2.39379138e-01
1.53367713e-01 -5.39170384e-01 -7.80567944e-01 7.18515337e-01
-2.85992503e-01 6.21173739e-01 8.99996102e-01 -8.41066837e-02
-5.24496377e-01 -1.24172843e+00 -2.03635788e+00 3.84788454e-01
9.64359999e-01 -4.28947777e-01 5.35484910e-01 3.02013040e-01
9.75562990e-01 2.03367528e-02 -9.92151141e-01 3.04129571e-01
7.23281503e-01 -7.64226139e-01 5.84255815e-01 -4.30742353e-01
-7.48755872e-01 -2.90123224e-01 -2.02899829e-01 -7.05599546e-01
-2.68538296e-01 -1.05498314e+00 -3.54962647e-01 1.62493491e+00
4.24888492e-01 -1.12159038e+00 1.34904075e+00 1.29695964e+00
6.55355632e-01 -7.77588904e-01 -1.32491350e+00 -6.71113372e-01
-3.46803576e-01 -3.13871205e-01 1.40198779e+00 8.49933684e-01
1.13484249e-01 -1.07310832e-01 -4.23320591e-01 4.12697375e-01
9.88590360e-01 1.11138411e-01 7.06576943e-01 -1.52418113e+00
1.21040754e-01 1.32585242e-01 -2.43212014e-01 -8.06594253e-01
-6.85250014e-02 -6.82864368e-01 -4.51460958e-01 -1.35151494e+00
3.96847837e-02 -1.18797386e+00 -8.65656853e-01 9.34223711e-01
4.80247110e-01 -2.90301532e-01 7.16720009e-03 4.19264615e-01
-9.48242247e-01 3.04718554e-01 7.74628460e-01 -7.44608045e-03
-2.57170439e-01 8.08042407e-01 -9.97102559e-01 3.34378153e-01
1.22173178e+00 -8.17601740e-01 -5.58211267e-01 -5.37303627e-01
-2.39216089e-01 2.38614604e-01 1.90302238e-01 -1.07080281e+00
6.01496875e-01 -4.21976924e-01 3.89176339e-01 -2.11458340e-01
-2.21839830e-01 -1.81512272e+00 7.21360207e-01 7.46058285e-01
-3.81621182e-01 -3.13270867e-01 -2.75438547e-01 7.52726436e-01
2.50653863e-01 5.37520826e-01 7.71334291e-01 1.53673857e-01
-1.12030283e-01 1.00163245e+00 -1.80631489e-01 -4.31130528e-01
1.19207180e+00 -7.17912838e-02 -1.50502965e-01 -5.48777282e-01
-2.55503863e-01 6.20436370e-01 5.41754246e-01 4.09750164e-01
2.34594047e-01 -1.22063696e+00 -8.48446786e-02 2.85708040e-01
1.83780655e-01 6.24730587e-02 5.12199029e-02 5.57708979e-01
8.44969675e-02 3.00520897e-01 1.37642384e-01 -1.69747934e-01
-1.26167524e+00 6.56020463e-01 5.04036844e-01 -1.56145930e-01
-4.69017297e-01 4.52879876e-01 -8.64675269e-02 -6.71541870e-01
8.37506592e-01 -1.03112031e-02 2.95523077e-01 -6.42660379e-01
5.69717407e-01 5.03628254e-01 2.64127622e-03 -2.82607377e-01
-6.67294860e-01 9.02775824e-02 -4.08692780e-04 2.59508640e-01
1.17849457e+00 -6.06818020e-01 -2.73591489e-01 -1.47138819e-01
1.46955049e+00 3.89717668e-02 -1.25305748e+00 -5.97219050e-01
-7.97266290e-02 -5.14987350e-01 2.42429689e-01 -8.69970739e-01
-1.71264064e+00 1.17976539e-01 8.42356265e-01 3.07282031e-01
1.49711823e+00 -1.46744162e-01 1.21816146e+00 2.88235277e-01
1.03379023e+00 -7.06941903e-01 -5.34855962e-01 -2.13934064e-01
-9.85141918e-02 -9.23598588e-01 -2.73914367e-01 -5.48277915e-01
-1.51753068e-01 1.00408411e+00 6.99287534e-01 4.91101712e-01
1.04795539e+00 3.28583956e-01 1.46787584e-01 8.95664841e-02
-7.58879185e-01 5.47390759e-01 -4.74166930e-01 7.54283845e-01
-2.66697526e-01 1.61776781e-01 -3.28630835e-01 1.09432507e+00
1.76553607e-01 2.75552511e-01 1.44190080e-02 1.10696089e+00
-4.52725857e-01 -1.62629342e+00 -2.09479988e-01 5.65399051e-01
-6.85570776e-01 4.92569834e-01 -9.38879922e-02 2.01652050e-01
3.09124202e-01 1.25970125e+00 -2.29475871e-01 -6.08596683e-01
1.48700684e-01 -4.61343601e-02 -1.34682447e-01 2.67431047e-03
-5.83807588e-01 -3.76860470e-01 -2.80412048e-01 -7.57160246e-01
-2.43905485e-01 -3.64521801e-01 -1.26802409e+00 -6.31322086e-01
-4.46233749e-01 6.78827584e-01 9.42223310e-01 8.28759730e-01
8.69600773e-01 5.99717759e-02 1.52433789e+00 1.05825905e-03
-5.77967882e-01 -2.66815960e-01 -6.34885728e-01 -7.58090615e-02
3.55814517e-01 -8.49839449e-02 -3.12856734e-01 -2.71443129e-01]
|
[5.838403224945068, 6.242898941040039]
|
b51d2cd4-67ab-48c6-a327-e0ff562c90a3
|
occformer-dual-path-transformer-for-vision
|
2304.05316
| null |
https://arxiv.org/abs/2304.05316v1
|
https://arxiv.org/pdf/2304.05316v1.pdf
|
OccFormer: Dual-path Transformer for Vision-based 3D Semantic Occupancy Prediction
|
The vision-based perception for autonomous driving has undergone a transformation from the bird-eye-view (BEV) representations to the 3D semantic occupancy. Compared with the BEV planes, the 3D semantic occupancy further provides structural information along the vertical direction. This paper presents OccFormer, a dual-path transformer network to effectively process the 3D volume for semantic occupancy prediction. OccFormer achieves a long-range, dynamic, and efficient encoding of the camera-generated 3D voxel features. It is obtained by decomposing the heavy 3D processing into the local and global transformer pathways along the horizontal plane. For the occupancy decoder, we adapt the vanilla Mask2Former for 3D semantic occupancy by proposing preserve-pooling and class-guided sampling, which notably mitigate the sparsity and class imbalance. Experimental results demonstrate that OccFormer significantly outperforms existing methods for semantic scene completion on SemanticKITTI dataset and for LiDAR semantic segmentation on nuScenes dataset. Code is available at \url{https://github.com/zhangyp15/OccFormer}.
|
['Dalong Du', 'Zheng Zhu', 'Yunpeng Zhang']
|
2023-04-11
| null | null | null | null |
['lidar-semantic-segmentation']
|
['computer-vision']
|
[ 2.76735008e-01 1.82579741e-01 -1.24410771e-01 -7.64473617e-01
-6.54482484e-01 -3.42861235e-01 5.02216756e-01 -1.94860116e-01
-1.52089223e-01 6.30381703e-02 3.46421272e-01 -1.99555531e-01
1.03584724e-02 -1.00585485e+00 -9.37717259e-01 -5.74244022e-01
4.63245243e-01 5.87198496e-01 4.44370508e-01 5.09069152e-02
4.28784907e-01 4.67244655e-01 -2.16668439e+00 5.48775673e-01
6.41314685e-01 1.41222143e+00 7.69636512e-01 4.45316076e-01
-3.53436023e-01 4.80156988e-01 1.09810561e-01 2.76608020e-01
5.51839709e-01 1.25552058e-01 -4.82660532e-01 3.04952949e-01
8.65393519e-01 -4.37149554e-01 -4.79894012e-01 9.80039239e-01
2.43901044e-01 1.81955054e-01 6.14914417e-01 -1.35247302e+00
-1.93215922e-01 2.82831609e-01 -6.80953264e-01 2.02206314e-01
1.25221893e-01 1.73095599e-01 1.11652219e+00 -1.53749335e+00
5.99481702e-01 1.57795262e+00 4.84434485e-01 1.91164225e-01
-9.62918818e-01 -7.93326259e-01 4.17855471e-01 4.26843673e-01
-1.60918403e+00 -4.09477383e-01 6.92545950e-01 -6.76273525e-01
1.27103710e+00 2.79225826e-01 9.43927407e-01 8.20080221e-01
2.54534870e-01 8.48175704e-01 1.23702550e+00 9.43471491e-02
2.22203508e-01 -2.63278037e-01 3.12528610e-01 7.19798684e-01
1.09728925e-01 2.33259141e-01 -1.06136239e+00 2.57196903e-01
8.13696384e-01 -1.49243111e-02 -1.30368853e-02 -6.95878804e-01
-1.12171435e+00 7.12225735e-01 5.02227247e-01 -4.26002651e-01
-1.14009529e-01 5.36761880e-01 1.34925172e-01 -3.22210312e-01
6.23350620e-01 3.10176495e-03 -5.27735889e-01 2.38700405e-01
-9.37397540e-01 3.87998283e-01 3.24777275e-01 1.28581619e+00
1.25639689e+00 7.38492236e-02 -3.04288238e-01 6.40008628e-01
8.53951812e-01 8.43297184e-01 3.55949216e-02 -1.30330169e+00
4.66713488e-01 5.86438060e-01 -9.73049626e-02 -5.01679718e-01
-3.78857672e-01 -4.65244412e-01 -6.11304402e-01 2.23825306e-01
1.22786999e-01 5.18084109e-01 -1.41949844e+00 1.34082663e+00
5.49086273e-01 2.89896339e-01 -1.45397797e-01 1.24802291e+00
1.14546096e+00 6.63473725e-01 -1.19583488e-01 3.58851492e-01
1.55889606e+00 -1.07502949e+00 -5.40572226e-01 -7.03758061e-01
2.26937294e-01 -5.86711884e-01 8.86901140e-01 2.48979449e-01
-9.44425285e-01 -7.40829706e-01 -1.25004196e+00 -9.45525110e-01
-3.95891964e-01 -1.28229573e-01 5.89950681e-01 3.24741900e-01
-1.05659604e+00 1.16342446e-02 -9.40255642e-01 -1.21328488e-01
8.79477501e-01 5.35074063e-02 -1.23336703e-01 -4.28908914e-01
-1.00899220e+00 7.13089883e-01 3.85114551e-01 2.62426555e-01
-1.21110594e+00 -8.82375717e-01 -1.25613189e+00 3.81947272e-02
3.04319084e-01 -7.91245282e-01 1.08457136e+00 -2.41539851e-01
-9.27344978e-01 9.64099169e-01 -6.88716471e-01 -3.40580314e-01
1.81903362e-01 -1.73234046e-01 3.47797304e-01 -7.01512035e-04
4.09783214e-01 1.35528255e+00 8.61673832e-01 -1.33084202e+00
-8.67171764e-01 -7.75831878e-01 -9.04784501e-02 6.05633318e-01
2.96034366e-01 -7.58519828e-01 -6.62156224e-01 -2.80703068e-01
1.01787066e+00 -7.15040088e-01 -3.08409989e-01 1.80073291e-01
-3.80708665e-01 -2.09483936e-01 6.33707523e-01 -6.02709353e-01
6.87069476e-01 -2.30850816e+00 5.32373041e-02 2.32172355e-01
4.17708308e-01 -4.50553924e-01 3.41470949e-02 8.25197399e-02
1.57497540e-01 -1.59745008e-01 -3.49543780e-01 -5.73179424e-01
2.82732755e-01 3.81728619e-01 -6.47296846e-01 6.15595639e-01
1.18654005e-01 1.03374600e+00 -7.41034746e-01 -3.69515061e-01
7.28806555e-01 6.08821690e-01 -9.51788723e-01 4.87235840e-03
-3.83466929e-01 4.76183951e-01 -6.25156224e-01 7.20224559e-01
1.17784691e+00 1.41964527e-04 -1.90656722e-01 -3.63398999e-01
-4.32913691e-01 5.00132620e-01 -9.25992906e-01 2.24892592e+00
-1.81253776e-01 7.25531995e-01 1.12267181e-01 -7.59291828e-01
9.26325917e-01 -1.90978125e-01 4.30244684e-01 -1.09035373e+00
2.11392328e-01 2.74920732e-01 -5.14255345e-01 -1.08662464e-01
7.48404324e-01 2.88992673e-02 -3.73136431e-01 -6.13127574e-02
9.60900914e-03 -8.06907594e-01 -1.21621683e-01 2.43268702e-02
6.70803666e-01 4.20272321e-01 9.94595066e-02 -5.40223062e-01
2.94299513e-01 2.21078530e-01 6.23531580e-01 6.39484048e-01
-2.59743243e-01 7.77044177e-01 2.00943321e-01 -3.13855171e-01
-1.02196693e+00 -1.50035918e+00 -4.50432420e-01 8.19722116e-01
6.67699039e-01 -3.29016536e-01 -7.53537536e-01 -1.25651821e-01
2.88340271e-01 8.60610604e-01 -4.68155563e-01 -7.27270618e-02
-2.87163526e-01 -2.99755812e-01 2.19504818e-01 5.59605360e-01
7.12283790e-01 -6.47008121e-01 -8.96184564e-01 -2.52290517e-02
-4.40020233e-01 -1.36282635e+00 -4.13275868e-01 4.95235324e-01
-7.70302474e-01 -9.62876439e-01 -2.39728540e-02 -6.00169778e-01
5.20238698e-01 7.53100812e-01 8.33729506e-01 -3.17629069e-01
-5.14851451e-01 3.69839072e-01 -1.46566540e-01 -5.08918941e-01
4.02866244e-01 -1.28813133e-01 -8.80788490e-02 -2.00052977e-01
7.42444158e-01 -6.49340808e-01 -7.25769043e-01 2.79211223e-01
-6.13800585e-01 6.81203365e-01 1.18579343e-01 5.02221882e-01
1.30641341e+00 -9.43406522e-02 -3.36885303e-02 -4.38491970e-01
-2.38692388e-01 -4.35241550e-01 -7.89465904e-01 -4.16629344e-01
-4.79337782e-01 -7.95676559e-02 -8.15770552e-02 3.93427074e-01
-1.05073476e+00 4.15517420e-01 -2.73974061e-01 -6.18986011e-01
-2.33623326e-01 -4.04172316e-02 -2.70618856e-01 1.73707813e-01
3.44154328e-01 2.83187658e-01 -2.63174772e-01 -4.67692018e-01
6.24034166e-01 4.66798842e-01 4.52927470e-01 -3.58371079e-01
7.57905185e-01 1.01829231e+00 -3.94643247e-02 -8.68556082e-01
-1.26937854e+00 -8.29473555e-01 -8.86879504e-01 -4.21433240e-01
1.34643292e+00 -1.60617554e+00 -3.81354690e-01 5.50417304e-01
-1.13796222e+00 -3.88391674e-01 -5.60876489e-01 3.64807069e-01
-9.19721782e-01 5.37661500e-02 -1.72468275e-01 -6.87764347e-01
-1.64299622e-01 -1.26751864e+00 1.65435445e+00 4.31616418e-02
-1.37146302e-02 -3.89953107e-01 -3.17710012e-01 8.98179412e-01
-8.66477843e-03 5.01006693e-02 6.74444318e-01 4.16056477e-02
-1.38979852e+00 3.41865331e-01 -5.25600255e-01 1.99305534e-01
-5.29513538e-01 -4.06533301e-01 -1.48932254e+00 1.09285498e-02
7.94522688e-02 -2.00298369e-01 1.47004688e+00 7.59787083e-01
1.44839549e+00 3.08663458e-01 -3.49607021e-01 1.22139311e+00
1.39868331e+00 -5.23675606e-02 7.38429010e-01 1.00292698e-01
1.14986587e+00 7.08694518e-01 7.87935615e-01 5.83066344e-01
1.12640941e+00 4.68751878e-01 9.38275456e-01 -2.17968225e-02
-4.83585328e-01 -6.06199920e-01 2.09275633e-01 6.91477299e-01
3.21033001e-01 -1.01289272e-01 -9.75747406e-01 4.93115187e-01
-1.80265903e+00 -6.01006269e-01 -4.83739078e-01 1.81877708e+00
2.96530694e-01 4.65826578e-02 -5.95470369e-01 -7.95520917e-02
3.15135479e-01 5.33277929e-01 -9.20541346e-01 -3.13506663e-01
-2.19826981e-01 2.09150553e-01 9.27203000e-01 8.24847341e-01
-1.05472755e+00 1.33002162e+00 5.27283430e+00 8.78814280e-01
-7.26242542e-01 4.45677787e-01 6.20873809e-01 -4.16945875e-01
-6.74522281e-01 3.61089148e-02 -1.20988274e+00 9.70792994e-02
4.65969563e-01 2.67787457e-01 4.40051496e-01 7.33913302e-01
3.06436121e-01 -5.37451386e-01 -9.41029370e-01 1.25701284e+00
1.22762926e-01 -1.27830124e+00 3.72989178e-02 2.55865455e-01
7.00463474e-01 6.11562014e-01 2.20749110e-01 8.36270005e-02
1.44963577e-01 -1.12875736e+00 1.49606717e+00 6.11547768e-01
9.35485721e-01 -5.25564432e-01 2.27391750e-01 4.19291168e-01
-1.65833402e+00 -1.91457272e-01 -5.92709005e-01 -1.12403132e-01
2.67747283e-01 7.21823394e-01 -5.78788280e-01 3.26786458e-01
1.12651098e+00 9.36844528e-01 -4.09108490e-01 7.06395388e-01
-1.24539763e-01 1.37883618e-01 -3.79750699e-01 3.82074207e-01
3.22642475e-01 -2.60511667e-01 6.00301802e-01 8.31585288e-01
4.89509046e-01 2.08133325e-01 2.42703363e-01 1.18661177e+00
2.15813205e-01 -3.74922335e-01 -6.24251425e-01 3.31938505e-01
5.99326730e-01 1.08870518e+00 -7.64918447e-01 -2.88780421e-01
-1.54796332e-01 7.88648069e-01 1.75490484e-01 4.35711712e-01
-9.36706305e-01 4.71921191e-02 1.03167653e+00 4.06429529e-01
7.05809712e-01 -4.94149834e-01 -9.21848059e-01 -9.28421378e-01
1.00705713e-01 -1.18649572e-01 -2.20847260e-02 -1.15906477e+00
-7.39147723e-01 2.04184234e-01 2.44098172e-01 -1.10316479e+00
2.41787016e-01 -7.58147180e-01 9.50579122e-02 9.59472120e-01
-1.92735159e+00 -1.11561012e+00 -6.86449826e-01 4.79029328e-01
9.35700178e-01 3.00449878e-01 4.48559612e-01 2.17949584e-01
-2.49121353e-01 -1.80307720e-02 -3.21105957e-01 -5.02959430e-01
2.68913150e-01 -9.21034098e-01 5.58667064e-01 7.42818892e-01
-6.59871399e-02 2.00159643e-02 3.95613045e-01 -5.58443248e-01
-1.60286200e+00 -1.48810101e+00 7.10669518e-01 -6.90763116e-01
2.63659209e-01 -8.39022636e-01 -5.74500799e-01 6.51911139e-01
-1.49022147e-01 -5.15629277e-02 3.89254123e-01 -4.11754847e-01
-5.80442131e-01 -1.09112747e-01 -9.73119974e-01 3.28940362e-01
1.71322453e+00 -7.31068790e-01 -3.62154990e-01 2.99748868e-01
9.36172545e-01 -7.83012688e-01 -5.47151089e-01 5.29201090e-01
5.78301966e-01 -1.12540114e+00 1.23757780e+00 2.46581435e-01
4.66050446e-01 -6.98977709e-01 -8.39131951e-01 -7.28816390e-01
-4.21392173e-01 6.00939430e-02 -7.61813521e-02 6.08795285e-01
2.17458025e-01 -5.36077678e-01 7.75214016e-01 1.58949375e-01
-7.88866520e-01 -7.50029087e-01 -1.31295216e+00 -2.16621593e-01
-2.54022032e-01 -1.06874847e+00 6.60856605e-01 5.44525146e-01
-4.34309661e-01 3.07913154e-01 -1.08945640e-02 5.05143344e-01
9.73584950e-01 1.83313414e-01 6.55345559e-01 -1.11914992e+00
2.11407453e-01 -2.43572637e-01 -4.44874942e-01 -1.75583577e+00
6.64522126e-02 -1.35012770e+00 5.17154932e-01 -1.96686840e+00
2.25235343e-01 -3.90737295e-01 -1.00841932e-01 3.61180663e-01
2.18213215e-01 4.57631677e-01 3.55496816e-02 7.44411647e-02
-5.03141344e-01 9.20697033e-01 1.62504470e+00 -2.04441622e-01
1.15572579e-01 -5.89390993e-01 -6.09597504e-01 7.15693057e-01
8.04527760e-01 -2.13707522e-01 -6.67342424e-01 -7.59392023e-01
1.16396718e-01 -1.83566719e-01 6.60139680e-01 -1.03965628e+00
1.51949584e-01 -1.41329318e-01 4.94157910e-01 -1.54023814e+00
9.45347607e-01 -6.93829536e-01 4.57963869e-02 1.79273471e-01
8.62681419e-02 -3.14656138e-01 2.27622703e-01 6.84673369e-01
-9.68445018e-02 2.25034446e-01 6.93989277e-01 -1.93268925e-01
-1.27908874e+00 6.78106487e-01 -3.81837100e-01 -5.02721071e-02
8.73960674e-01 -6.79110646e-01 -3.75783652e-01 9.30749476e-02
-5.69115996e-01 6.82080805e-01 5.00765026e-01 5.67772210e-01
1.00605428e+00 -1.20914030e+00 -4.76481438e-01 7.43380487e-01
2.39981428e-01 8.44714165e-01 6.86542690e-01 8.30869913e-01
-5.92326283e-01 6.72605991e-01 -1.29948691e-01 -1.19289005e+00
-8.86171818e-01 1.07831530e-01 4.39682156e-01 3.02172899e-01
-9.07005847e-01 1.28882289e+00 9.57575142e-01 -7.00031698e-01
2.24031970e-01 -6.97996080e-01 -1.41947374e-01 8.38321745e-02
2.89673179e-01 2.04452246e-01 -4.26254719e-02 -9.82821703e-01
-3.78523290e-01 7.76440620e-01 2.82557487e-01 -2.30423823e-01
1.17345083e+00 -5.11446714e-01 -2.52391875e-01 5.97797096e-01
1.00559437e+00 -4.26438779e-01 -1.78187990e+00 -1.17044300e-01
-4.56361413e-01 -5.92899501e-01 5.12686431e-01 -3.54793370e-01
-1.09141636e+00 1.21254635e+00 7.07270741e-01 -2.91329890e-01
8.94981980e-01 2.35068306e-01 7.56480813e-01 6.96369335e-02
6.05751097e-01 -1.12804627e+00 -1.50923967e-01 1.02786648e+00
9.42325413e-01 -1.10725570e+00 1.70079935e-02 -9.46197450e-01
-4.30456132e-01 7.04533875e-01 8.17986846e-01 -2.07236230e-01
9.74090099e-01 1.02392226e-01 -1.64355606e-01 -6.57067358e-01
-7.32089579e-01 -2.76378959e-01 2.62017757e-01 5.56351244e-01
-7.25568160e-02 4.07507986e-01 2.19703510e-01 4.63564903e-01
-5.80295682e-01 -4.39655453e-01 3.00187498e-01 5.86293697e-01
-8.33116889e-01 -5.04064381e-01 -4.62179095e-01 6.05022013e-01
3.76934171e-01 -3.41459751e-01 8.23096707e-02 4.61728066e-01
6.55950427e-01 8.49040747e-01 6.86324418e-01 -4.57935154e-01
4.18249041e-01 9.56411213e-02 6.18444741e-01 -8.93358409e-01
8.00426528e-02 3.17718238e-01 1.00187697e-01 -1.11290157e+00
-2.39935175e-01 -8.26149642e-01 -1.60331619e+00 -4.26172391e-02
-5.33660352e-02 -5.04409373e-01 9.21592176e-01 7.92393744e-01
4.04764622e-01 8.44299853e-01 5.28044343e-01 -1.28164375e+00
-1.31666800e-02 -7.42713034e-01 -7.06737578e-01 -1.54695928e-01
3.33203256e-01 -9.96208072e-01 -4.08760726e-01 -6.44995347e-02]
|
[8.308125495910645, -2.700193166732788]
|
84529901-7709-48ba-bda7-adc5424363cc
|
rediscovery-of-the-effectiveness-of-standard
|
2204.01209
| null |
https://arxiv.org/abs/2204.01209v2
|
https://arxiv.org/pdf/2204.01209v2.pdf
|
EResFD: Rediscovery of the Effectiveness of Standard Convolution for Lightweight Face Detection
|
This paper analyses the design choices of face detection architecture that improve efficiency between computation cost and accuracy. Specifically, we re-examine the effectiveness of the standard convolutional block as a lightweight backbone architecture on face detection. Unlike the current tendency of lightweight architecture design, which heavily utilizes depthwise separable convolution layers, we show that heavily channel-pruned standard convolution layer can achieve better accuracy and inference speed when using a similar parameter size. This observation is supported by the analyses concerning the characteristics of the target data domain, face. Based on our observation, we propose to employ ResNet with a highly reduced channel, which surprisingly allows high efficiency compared to other mobile-friendly networks (e.g., MobileNet-V1,-V2,-V3). From the extensive experiments, we show that the proposed backbone can replace that of the state-of-the-art face detector with a faster inference speed. Also, we further propose a new feature aggregation method maximizing the detection performance. Our proposed detector EResFD obtained 80.4% mAP on WIDER FACE Hard subset which only takes 37.7 ms for VGA image inference in on CPU. Code will be available at https://github.com/clovaai/EResFD.
|
['Youngjoon Yoo', 'Joonsang Yu', 'Beomyoung Kim', 'JoonHyun Jeong']
|
2022-04-04
| null | null | null | null |
['face-detection']
|
['computer-vision']
|
[-2.71463066e-01 1.45274043e-01 6.88481703e-03 -3.95191759e-01
-4.55231220e-03 -1.99870452e-01 2.57381111e-01 -5.73605597e-01
-5.39386392e-01 3.31327260e-01 -2.66225487e-01 -4.60818797e-01
1.30595505e-01 -9.78625417e-01 -5.44711769e-01 -5.23119330e-01
-1.59932822e-01 -1.46268636e-01 2.71065533e-01 -9.87330526e-02
-5.94420210e-02 8.43347430e-01 -1.69647074e+00 1.35104418e-01
3.40622544e-01 1.27731025e+00 1.20912746e-01 6.44031525e-01
1.17859095e-01 3.14302683e-01 -5.53146958e-01 -7.64855266e-01
5.18890440e-01 2.54076142e-02 -4.89285588e-01 -3.35315347e-01
7.12874711e-01 -9.41580057e-01 -6.80207133e-01 1.03235030e+00
9.27978158e-01 -2.14634180e-01 3.13032866e-01 -1.28272569e+00
-3.77024174e-01 5.30164242e-01 -7.32722223e-01 5.41328132e-01
5.61280176e-02 1.39604449e-01 5.36467552e-01 -1.17319369e+00
4.33134556e-01 1.59730363e+00 8.12892675e-01 8.35035801e-01
-7.93642282e-01 -1.22885513e+00 1.06520429e-01 2.00913608e-01
-1.88448930e+00 -9.45071161e-01 3.25214863e-01 1.54269021e-03
1.15334582e+00 1.24063820e-01 4.53015625e-01 9.37048793e-01
1.25379823e-02 5.89482009e-01 6.75132990e-01 -3.64018053e-01
7.03929067e-02 2.51868248e-01 8.07877705e-02 1.24857736e+00
6.76972091e-01 3.74238379e-02 -4.56688672e-01 -7.23457187e-02
9.01482821e-01 -7.11696148e-02 -5.32020777e-02 1.49801165e-01
-4.00645912e-01 9.19464648e-01 5.51977873e-01 1.77473366e-01
-2.65565336e-01 4.67809975e-01 4.58207637e-01 2.38818958e-01
3.40590894e-01 -3.28933090e-01 -2.68489420e-01 2.56597459e-01
-1.07810712e+00 6.87952340e-02 7.45878875e-01 1.16744375e+00
5.58172584e-01 3.29221278e-01 -2.02909470e-01 5.02526224e-01
7.10495532e-01 7.98789024e-01 4.77254242e-02 -7.52635658e-01
1.40157580e-01 5.39924204e-01 -4.42318618e-01 -8.75107586e-01
-5.47755003e-01 -6.64505541e-01 -7.16337323e-01 4.66171026e-01
4.27837282e-01 -4.42555785e-01 -9.88881946e-01 1.67465806e+00
4.77935493e-01 2.70353138e-01 -9.75279063e-02 8.16245198e-01
1.17564011e+00 1.67868733e-01 1.43564582e-01 1.01080537e-01
1.89155507e+00 -6.11199379e-01 -4.50946093e-01 1.00746736e-01
5.56629598e-01 -7.67921209e-01 5.95059395e-01 1.58008263e-01
-1.06877732e+00 -7.65068173e-01 -1.20476294e+00 -3.18763703e-02
-4.43446308e-01 6.99009418e-01 9.51595426e-01 1.40229607e+00
-1.52007890e+00 3.34516823e-01 -7.74754584e-01 -7.00553894e-01
9.02570844e-01 8.42152953e-01 -1.68691292e-01 1.93148643e-01
-9.80309606e-01 6.35115027e-01 2.41193280e-01 2.36039221e-01
-9.46959913e-01 -5.82895219e-01 -5.17284632e-01 2.36080170e-01
2.76767373e-01 -7.14504302e-01 1.24922240e+00 -7.78077424e-01
-1.39208055e+00 8.68270874e-01 -2.21269026e-01 -5.67658603e-01
3.32365036e-01 -1.00931615e-01 -5.63461065e-01 3.31223458e-01
-2.30097026e-01 1.01306367e+00 1.01379704e+00 -6.86198115e-01
-8.85229468e-01 -4.86325383e-01 3.14735681e-01 -1.26155451e-01
-6.55195236e-01 3.12444359e-01 -6.72070444e-01 -2.06549257e-01
-6.82464540e-02 -8.22291970e-01 5.85094169e-02 4.77628767e-01
-4.26362574e-01 -3.50640357e-01 1.06450367e+00 -4.00415719e-01
1.23653293e+00 -2.17871690e+00 -5.34293592e-01 4.36432719e-01
4.92214948e-01 7.34239340e-01 2.69528795e-02 -4.23565842e-02
6.12307414e-02 -1.39600681e-02 1.90511525e-01 -3.88740033e-01
-9.29864496e-02 -1.49803549e-01 1.18478090e-02 7.79511631e-01
2.85268463e-02 7.83919990e-01 -3.64688814e-01 -6.66789949e-01
1.71515450e-01 7.97509849e-01 -7.40485787e-01 -3.13950814e-02
3.15369397e-01 -1.32911414e-01 -5.20018160e-01 1.00124371e+00
1.26597416e+00 -1.71050206e-01 1.55859187e-01 -5.02373457e-01
-2.15724483e-01 -3.73297408e-02 -1.31158900e+00 1.40399957e+00
-2.77885795e-01 7.26837754e-01 5.49104869e-01 -5.91245651e-01
8.51827383e-01 2.63825655e-01 9.52086449e-02 -6.36731505e-01
6.10557556e-01 1.74875557e-01 2.12295771e-01 -2.36955911e-01
3.43416631e-01 3.83938551e-01 5.49501061e-01 1.98672414e-01
2.87995726e-01 6.91419601e-01 -2.80533414e-02 2.05636308e-01
1.05590725e+00 -1.35203302e-01 1.56808913e-01 -5.74851036e-01
7.13807106e-01 -4.89531398e-01 3.19370449e-01 8.44801664e-01
-4.32073891e-01 1.73396111e-01 3.58874321e-01 -4.25316334e-01
-7.60000825e-01 -9.59525824e-01 -6.00569725e-01 1.21798050e+00
3.19116935e-02 -4.68039095e-01 -1.19309020e+00 -6.46082997e-01
1.62783772e-01 5.76372929e-02 -4.87945855e-01 1.56318158e-01
-6.42890692e-01 -9.83857036e-01 1.23384643e+00 6.70449734e-01
1.11536562e+00 -7.97512054e-01 -8.30415070e-01 -7.28484169e-02
4.42821175e-01 -1.18640947e+00 -1.33835837e-01 -2.30665710e-02
-6.32122755e-01 -1.08578229e+00 -5.62695861e-01 -8.44309390e-01
6.77209139e-01 3.79233539e-01 8.28070283e-01 5.64267874e-01
-5.74938536e-01 3.32012624e-01 -8.95674974e-02 -5.69666803e-01
6.05640598e-02 2.03754872e-01 2.36554384e-01 -9.82784480e-02
7.36117065e-01 -4.22481358e-01 -1.02948892e+00 9.16088521e-02
-4.36020792e-01 -2.21717268e-01 6.14397824e-01 4.23997551e-01
1.48045003e-01 -2.22275168e-01 4.99808967e-01 -6.95132732e-01
4.17663634e-01 -4.36025083e-01 -7.14460492e-01 1.29491791e-01
-6.38110578e-01 -9.12381634e-02 3.61682177e-01 -1.51838884e-01
-1.15381241e+00 2.16960266e-01 -4.46311414e-01 -4.24004346e-01
-6.60576075e-02 -2.97690004e-01 -6.93662912e-02 -5.67552984e-01
6.15245223e-01 6.08598217e-02 1.82591677e-01 -4.75017488e-01
2.23862648e-01 9.50568795e-01 3.96287858e-01 -2.69237429e-01
5.80780268e-01 8.46602201e-01 1.66903019e-01 -1.23623657e+00
-1.36371538e-01 -2.78434187e-01 -4.76812631e-01 -2.90585816e-01
7.16081619e-01 -1.21728170e+00 -1.29582310e+00 5.29553354e-01
-1.21714950e+00 -4.15553823e-02 3.40042531e-01 3.71050775e-01
-2.91764550e-03 2.25694269e-01 -6.98581040e-01 -9.98315394e-01
-8.11609745e-01 -1.01583362e+00 1.09503019e+00 4.55657065e-01
3.15424427e-02 -6.28830671e-01 -4.93738830e-01 -8.71193409e-02
5.94771981e-01 2.25180387e-03 3.09905052e-01 -4.18153018e-01
-5.70781112e-01 6.46216273e-02 -7.33133733e-01 1.65206850e-01
-2.12360054e-01 1.68003753e-01 -1.47140098e+00 -6.30767703e-01
-1.76255405e-01 -8.95297676e-02 1.07385743e+00 4.80615497e-01
1.28298497e+00 -6.75023422e-02 -6.06087208e-01 1.02971816e+00
1.45887733e+00 7.96191096e-02 6.33639336e-01 4.23778892e-02
5.59089899e-01 3.28134358e-01 3.05619776e-01 6.06409192e-01
1.16237730e-01 6.98739767e-01 5.82201838e-01 -2.77831823e-01
-5.47912836e-01 -6.67167976e-02 3.69525403e-01 2.33061239e-01
-2.69612789e-01 -3.41230810e-01 -6.72648787e-01 2.65426844e-01
-1.52670312e+00 -8.58063221e-01 -8.02737027e-02 1.91805732e+00
2.94469714e-01 1.51110768e-01 3.28322500e-01 -7.95902610e-02
8.41524661e-01 3.50308903e-02 -4.08988357e-01 -5.10067284e-01
7.15828538e-02 5.94212294e-01 9.84075844e-01 2.58478820e-01
-1.14657867e+00 1.01166809e+00 6.46262932e+00 1.06109929e+00
-1.13796532e+00 4.10140485e-01 5.27630210e-01 -3.65250736e-01
2.64138341e-01 -4.16673481e-01 -1.56606746e+00 5.16123176e-01
1.05921590e+00 1.70398623e-01 2.66198516e-01 1.07810664e+00
2.80915238e-02 2.10279264e-02 -1.04279435e+00 1.30176497e+00
7.63851628e-02 -1.33151364e+00 3.23958136e-02 2.16058135e-01
1.33031219e-01 4.40940894e-02 1.82630554e-01 1.90867931e-01
-5.79677261e-02 -1.11732543e+00 6.24287486e-01 1.09532565e-01
1.10967624e+00 -9.13819969e-01 7.26517797e-01 -8.24775547e-02
-1.51757073e+00 -2.28048503e-01 -6.72184169e-01 -7.79694766e-02
-2.07909703e-01 4.04002964e-01 -7.94201136e-01 9.73969996e-02
9.31432605e-01 1.70891911e-01 -7.01397657e-01 8.25479388e-01
-5.96608035e-04 4.45913672e-01 -6.11579657e-01 -1.00836776e-01
2.09419549e-01 2.99967438e-01 3.16866040e-01 1.60986423e+00
3.59908044e-01 1.45331889e-01 -2.39265591e-01 7.44564772e-01
-3.27983975e-01 -1.09400600e-01 -6.14125311e-01 4.83264089e-01
8.31081033e-01 1.53908134e+00 -9.39331770e-01 -3.14819276e-01
-7.24835932e-01 8.14401627e-01 3.15630227e-01 1.46016598e-01
-1.09354186e+00 -3.91810119e-01 1.01949430e+00 2.27730483e-01
6.61100090e-01 7.61692375e-02 -1.01724423e-01 -5.98331928e-01
-6.65788352e-02 -5.48327208e-01 3.43759537e-01 -1.93939000e-01
-7.04546094e-01 7.71174192e-01 -1.45700067e-01 -8.58732104e-01
1.16545796e-01 -1.01523471e+00 -6.45898581e-01 6.77150428e-01
-1.48553717e+00 -1.14362073e+00 -4.63452876e-01 9.04685974e-01
3.68439198e-01 -4.18344438e-01 6.28078818e-01 7.30894148e-01
-8.91403496e-01 1.38125622e+00 -2.16193259e-01 3.98884237e-01
3.66372257e-01 -7.34632373e-01 6.19781375e-01 9.74955320e-01
-9.47945863e-02 7.44781613e-01 3.20618987e-01 -5.08635521e-01
-1.58881545e+00 -1.18615246e+00 3.92294168e-01 -7.33188689e-02
1.18838772e-01 -5.92386842e-01 -5.51568389e-01 5.49055696e-01
1.39488071e-01 2.76594222e-01 3.59105438e-01 5.55861788e-03
-3.46160233e-01 -3.68666589e-01 -1.59355283e+00 6.24737024e-01
1.49029922e+00 -4.63453263e-01 1.28467947e-01 1.51807427e-01
3.84809554e-01 -3.50369960e-01 -5.01245499e-01 2.78081834e-01
8.52293074e-01 -1.04110599e+00 1.12842584e+00 -2.46131927e-01
-6.67856038e-02 -1.51512921e-01 -2.24214643e-01 -5.55933714e-01
-5.04799247e-01 -5.47816992e-01 -5.18564999e-01 1.07379985e+00
2.37951279e-01 -8.30818117e-01 1.08042932e+00 2.86118209e-01
1.01198822e-01 -8.22432876e-01 -1.16683280e+00 -8.47391427e-01
-2.75548995e-01 -3.08241844e-01 6.79958045e-01 3.63292217e-01
-3.68918955e-01 1.30100232e-02 -1.67254448e-01 4.87558901e-01
7.93854713e-01 -2.95012236e-01 6.59909010e-01 -1.04724944e+00
-2.04620212e-02 -3.88731122e-01 -7.81550765e-01 -1.06727648e+00
2.06109919e-02 -8.23583186e-01 -3.25640261e-01 -9.60717559e-01
1.72394902e-01 -3.95051748e-01 -2.05374315e-01 5.22583544e-01
1.42075032e-01 7.23221600e-01 1.58667192e-01 -1.20624840e-01
-5.47165811e-01 1.07838646e-01 9.49540019e-01 2.55058020e-01
6.00867048e-02 -1.89050451e-01 -7.29823112e-01 9.03555155e-01
9.83681798e-01 -3.30443501e-01 -3.91188771e-01 -5.17899513e-01
-1.73949778e-01 -3.70124400e-01 4.19948220e-01 -1.19055843e+00
4.49790597e-01 4.76284027e-01 8.08744192e-01 -3.83554995e-01
4.89133209e-01 -7.56784678e-01 -4.66312207e-02 8.20396066e-01
2.24224955e-01 -4.80436422e-02 4.61557418e-01 2.47497395e-01
2.85377830e-01 -1.02010496e-01 9.09718156e-01 -3.80706862e-02
-1.08652186e+00 4.62562650e-01 -2.14331344e-01 -3.52481008e-01
1.15310550e+00 -5.81139088e-01 -5.58866978e-01 -2.71682050e-02
-3.98734629e-01 1.23580381e-01 8.50407034e-02 3.28337103e-01
7.24557996e-01 -1.18791842e+00 -8.50282848e-01 5.49987614e-01
-2.93388367e-01 -4.52381074e-01 2.46918708e-01 7.65211821e-01
-7.40088046e-01 6.24476612e-01 -3.03605884e-01 -5.76544642e-01
-1.71861279e+00 2.60074526e-01 4.40137237e-01 2.79109389e-01
-6.92347407e-01 1.14865065e+00 2.87174582e-01 7.70280659e-02
4.66325492e-01 -7.33664185e-02 -1.25884563e-01 1.19194034e-02
9.22534227e-01 7.64234662e-01 2.39730895e-01 -6.07887030e-01
-7.80674577e-01 3.92556876e-01 -1.86672300e-01 3.20777416e-01
1.06093884e+00 -1.33945525e-01 7.95287862e-02 -5.79547286e-01
1.19867456e+00 -9.02968273e-02 -1.20842528e+00 4.18787077e-02
-4.11729693e-01 -4.76457596e-01 4.02859092e-01 -4.12825853e-01
-1.57093179e+00 6.46399677e-01 1.27802491e+00 -2.45657153e-02
1.21753418e+00 2.45523043e-02 5.77783108e-01 3.81820261e-01
4.29067194e-01 -9.59868908e-01 -2.35762149e-01 2.67724454e-01
6.64926052e-01 -1.15371490e+00 1.16520517e-01 -7.53111541e-01
-2.73241140e-02 1.19604445e+00 1.03594720e+00 -1.68872267e-01
8.67075264e-01 6.99438214e-01 -3.00743818e-01 -5.36020756e-01
-4.57422554e-01 -5.66169143e-01 -1.17025696e-01 5.72833538e-01
3.94128442e-01 1.29049748e-01 -1.96411014e-01 3.17043245e-01
-2.06344098e-01 -8.09839070e-02 1.43115580e-01 7.92181432e-01
-5.82916915e-01 -8.23567510e-01 -4.70452547e-01 6.41072869e-01
-6.99890196e-01 -2.43115783e-01 -1.58140123e-01 1.05118668e+00
5.26672721e-01 9.48193908e-01 4.10290003e-01 -4.27791417e-01
1.65178329e-01 -1.15506053e-01 6.14404380e-01 -4.05089676e-01
-6.13543510e-01 -1.63511738e-01 8.05022418e-02 -7.97245026e-01
-2.00024635e-01 -2.53040016e-01 -1.23016143e+00 -1.04216599e+00
-3.76810849e-01 -3.08764815e-01 7.35200286e-01 5.06751180e-01
5.68766594e-01 4.37352419e-01 3.91778111e-01 -6.42870426e-01
-4.59668368e-01 -8.77970338e-01 -6.70145214e-01 -2.23857686e-01
2.82350481e-01 -7.49065995e-01 -1.82705820e-01 -3.32985878e-01]
|
[13.29586410522461, 0.6910514831542969]
|
57dea243-0114-4ae6-96cf-6595da0b87f9
|
approximate-bisimulation-relations-for-neural
|
2202.01214
| null |
https://arxiv.org/abs/2202.01214v1
|
https://arxiv.org/pdf/2202.01214v1.pdf
|
Approximate Bisimulation Relations for Neural Networks and Application to Assured Neural Network Compression
|
In this paper, we propose a concept of approximate bisimulation relation for feedforward neural networks. In the framework of approximate bisimulation relation, a novel neural network merging method is developed to compute the approximate bisimulation error between two neural networks based on reachability analysis of neural networks. The developed method is able to quantitatively measure the distance between the outputs of two neural networks with the same inputs. Then, we apply the approximate bisimulation relation results to perform neural networks model reduction and compute the compression precision, i.e., assured neural networks compression. At last, using the assured neural network compression, we accelerate the verification processes of ACAS Xu neural networks to illustrate the effectiveness and advantages of our proposed approximate bisimulation approach.
|
['Zhongzhu Shao', 'Weiming Xiang']
|
2022-02-02
| null | null | null | null |
['neural-network-compression', 'neural-network-compression']
|
['methodology', 'miscellaneous']
|
[ 2.55918652e-01 6.91304728e-02 1.61892418e-02 -2.65113562e-01
2.60352582e-01 -3.76333535e-01 2.52204031e-01 2.16846019e-01
-1.71013907e-01 4.21355188e-01 -6.41098738e-01 -5.60076535e-01
-4.94886845e-01 -9.11447108e-01 -1.20703316e+00 -1.63303867e-01
2.40617678e-01 3.02582830e-01 7.14969710e-02 -9.53480378e-02
3.91942821e-02 7.94538975e-01 -1.71851575e+00 -1.38929233e-01
9.65168417e-01 1.39626884e+00 -3.20074439e-01 3.84643942e-01
-3.64019573e-02 5.78830123e-01 -4.47853386e-01 -3.05651993e-01
2.91771024e-01 -5.94328701e-01 -6.99417353e-01 -4.81643111e-01
2.74557114e-01 -5.15365779e-01 -5.28756738e-01 1.59422243e+00
-7.03660846e-02 1.41269907e-01 5.77721834e-01 -1.42623866e+00
-1.51978612e-01 1.02801931e+00 1.49569228e-01 -3.00891668e-01
-1.32533282e-01 -1.93939805e-01 5.81612647e-01 -4.37556118e-01
2.55943686e-01 1.03549290e+00 8.22285056e-01 3.33310932e-01
-1.31410825e+00 -8.93861890e-01 -9.37776417e-02 1.96569160e-01
-1.68509793e+00 -1.77254796e-01 6.81384861e-01 -2.56080134e-03
8.03902090e-01 2.81433910e-01 9.62320447e-01 1.04764976e-01
4.58333224e-01 2.69826323e-01 8.72448742e-01 -2.98042297e-01
5.98341882e-01 -7.83172175e-02 5.72405696e-01 1.06664801e+00
7.77240336e-01 2.55171686e-01 -1.35272697e-01 -4.27777916e-02
7.17015743e-01 2.81660467e-01 -3.09217811e-01 -1.95283648e-02
-4.66023237e-01 4.63264614e-01 5.72255969e-01 4.13640529e-01
-5.88057302e-02 5.73334873e-01 6.22437418e-01 3.32580805e-01
1.34910628e-01 2.48799220e-01 -3.10574137e-02 6.81074187e-02
-1.05801046e+00 2.95176923e-01 9.13663745e-01 9.74570215e-01
7.09701180e-01 4.85839486e-01 4.43129599e-01 8.54117274e-02
2.84375042e-01 4.88494217e-01 6.31125629e-01 -1.01431429e+00
1.52053058e-01 8.40722263e-01 -4.09986556e-01 -1.18721378e+00
-8.51818621e-02 -4.97012943e-01 -1.19974804e+00 3.71073514e-01
1.03612870e-01 -3.67884748e-02 -6.20626867e-01 1.77029133e+00
5.50535098e-02 5.77397108e-01 1.72857389e-01 4.81647193e-01
1.42941073e-01 8.27700555e-01 -4.17869449e-01 -4.04009342e-01
8.29100072e-01 -5.37404358e-01 -8.90331268e-01 3.26462179e-01
7.14673996e-01 1.66417956e-02 5.54923475e-01 3.66550088e-01
-1.39586854e+00 -4.54124898e-01 -1.71371365e+00 2.42734060e-01
-5.21515310e-01 1.77352235e-01 2.01258287e-01 4.08115089e-01
-9.27524149e-01 1.08858371e+00 -1.01067042e+00 5.64868934e-02
7.44918883e-02 4.90311414e-01 -1.16353214e-01 2.24818930e-01
-1.22184479e+00 1.02050257e+00 1.08074999e+00 5.98693848e-01
-6.76428974e-01 -8.57207835e-01 -7.69657433e-01 5.77089489e-01
-2.26432651e-01 -5.70287108e-01 1.09397423e+00 -1.04686952e+00
-1.33004105e+00 7.84608945e-02 1.30072892e-01 -1.14384544e+00
1.85442477e-01 3.33573908e-01 -3.64599586e-01 1.31278202e-01
-6.36658251e-01 3.32463652e-01 3.63182366e-01 -7.25413978e-01
-4.75011021e-01 -3.56897712e-01 -1.62011027e-01 -1.86894849e-01
-6.76079929e-01 -4.23787475e-01 1.97842065e-02 -2.68818498e-01
3.96865398e-01 -5.64398587e-01 -8.23608860e-02 3.69850963e-01
-2.50720829e-01 9.59330052e-03 7.67427444e-01 -4.07280207e-01
1.27956092e+00 -2.01914644e+00 6.10926412e-02 9.05405641e-01
1.84801430e-01 5.34277081e-01 2.12843537e-01 -1.69256050e-02
-1.53343081e-01 2.11052131e-02 -3.02851707e-01 1.36343673e-01
4.59438078e-02 3.03538471e-01 -3.59247446e-01 3.06591600e-01
9.09098908e-02 5.79508424e-01 -4.46156293e-01 -4.25772309e-01
2.38157421e-01 3.84577036e-01 -3.45784307e-01 1.30145147e-01
-2.92382121e-01 -6.10559285e-01 -1.44001380e-01 1.92079663e-01
8.17951202e-01 3.64688009e-01 4.14747149e-01 -5.43897688e-01
-9.49951410e-02 5.56967556e-02 -1.19961071e+00 1.30143893e+00
-5.69907725e-01 6.06733024e-01 -1.14927508e-01 -1.24032629e+00
1.27158916e+00 3.26427370e-01 -3.36065590e-02 -4.48988557e-01
8.01354229e-01 5.67321956e-01 -2.69244581e-01 2.28049874e-01
1.96712449e-01 -6.39367625e-02 1.05008140e-01 3.99520367e-01
1.35240331e-01 -2.87192553e-01 2.72710949e-01 -3.10543776e-02
5.77797353e-01 -5.01660109e-02 1.45485520e-01 -4.58732396e-01
8.30429316e-01 -4.34261896e-02 3.59682828e-01 4.36205715e-01
-6.55097328e-03 8.64102617e-02 6.96290135e-01 -4.94646132e-01
-1.15215015e+00 -1.00969398e+00 -5.78301819e-03 -1.29431799e-01
1.30091077e-02 -1.46426871e-01 -1.20915711e+00 -3.65270525e-01
-1.56384379e-01 6.57135308e-01 -3.32604259e-01 -9.24700260e-01
-4.01050001e-01 2.29186758e-01 1.28302693e+00 7.55813241e-01
7.17166185e-01 -3.91383767e-01 -5.80538213e-01 1.22959390e-02
1.89946249e-01 -7.12441266e-01 -9.20119137e-02 3.29460591e-01
-1.32614803e+00 -8.90600204e-01 -5.20858467e-01 -9.32700515e-01
7.66980350e-01 -3.14117223e-01 4.09777462e-01 2.29086593e-01
5.34875505e-02 -2.50234604e-01 4.35274631e-01 -9.34457928e-02
-7.44167447e-01 -8.13745409e-02 2.92115271e-01 -2.91035652e-01
5.31064533e-03 -8.81118357e-01 -4.83681709e-02 1.84608489e-01
-1.03407311e+00 3.21983904e-01 3.10985118e-01 6.41499817e-01
8.77042592e-01 6.06967330e-01 4.50706542e-01 -3.90063524e-01
7.73230433e-01 1.14678396e-02 -1.29146826e+00 4.85279590e-01
-1.21444201e+00 6.44111991e-01 1.08065450e+00 -5.00348747e-01
-6.94804490e-01 1.53143048e-01 -1.77372307e-01 -8.99829090e-01
3.21663231e-01 8.30004990e-01 -1.78881779e-01 -2.98758894e-01
5.20550966e-01 1.26898021e-01 4.48263079e-01 -1.39217809e-01
2.23154843e-01 4.01169986e-01 9.55015361e-01 -5.41377902e-01
4.72842276e-01 -9.29278042e-03 7.05723286e-01 -1.31042227e-01
1.71679258e-03 4.72687781e-01 7.22524300e-02 -3.54939729e-01
2.38110244e-01 -4.21266764e-01 -1.25571215e+00 4.87114847e-01
-1.33240700e+00 1.55218437e-01 -3.67454976e-01 5.69013596e-01
-5.38359165e-01 1.51169106e-01 -8.81492198e-01 -9.55139577e-01
-6.84389889e-01 -1.09123600e+00 4.50267494e-02 2.37024814e-01
-6.82755634e-02 -6.23922825e-01 -1.28434181e-01 -6.53524399e-01
2.93943197e-01 3.78589720e-01 1.07373083e+00 -8.03010166e-01
-3.27621818e-01 -6.15418255e-01 -3.01612139e-01 7.78889477e-01
-3.32536966e-01 4.57045496e-01 -5.74391425e-01 6.68897107e-02
2.53098160e-01 9.26204324e-02 5.22051811e-01 3.03395540e-01
7.82478750e-01 -6.86210215e-01 -2.29902163e-01 4.20047402e-01
1.59402740e+00 5.17856359e-01 6.54496253e-01 -2.92472541e-01
3.60061854e-01 4.76519883e-01 3.96134555e-01 1.53454199e-01
-2.36168310e-01 2.05914944e-01 6.20874941e-01 3.75223190e-01
2.98507065e-01 -3.30481440e-01 3.01035523e-01 9.90694165e-01
6.57036081e-02 2.23307505e-01 -7.19829679e-01 2.59104341e-01
-1.67171955e+00 -5.84415197e-01 -9.36181925e-04 2.11575079e+00
7.54884839e-01 5.09340405e-01 -2.05331996e-01 7.35819161e-01
1.09452140e+00 -5.17570734e-01 -2.93377876e-01 -9.30194855e-01
3.18262964e-01 4.32033628e-01 3.02094460e-01 7.49548197e-01
-1.50286630e-01 4.28079456e-01 5.19955015e+00 9.65587497e-01
-1.15518272e+00 -2.95413971e-01 4.12477702e-01 1.21735036e-01
-1.15861267e-01 -6.49196431e-02 -4.38268572e-01 4.07378554e-01
1.41838336e+00 -5.64119816e-01 6.17220223e-01 9.25674260e-01
-9.78735983e-02 2.48492435e-01 -1.28443706e+00 5.15437663e-01
-1.85066521e-01 -1.38174284e+00 2.93767214e-01 -3.89816284e-01
4.35780376e-01 -6.05327129e-01 5.99905327e-02 1.04098998e-01
-2.88563579e-01 -8.57249439e-01 9.62671280e-01 6.60794079e-01
6.71436906e-01 -1.54154623e+00 9.96207297e-01 3.48894358e-01
-1.27705908e+00 6.93830475e-02 -3.47591847e-01 -3.34145606e-01
-1.35845110e-01 4.53231514e-01 -6.32408857e-01 6.32997096e-01
2.33361199e-01 3.67863029e-01 -2.50236869e-01 9.52493191e-01
-2.91094407e-02 1.56485647e-01 -3.58894289e-01 -4.77669269e-01
3.09549104e-02 -4.92683619e-01 1.38327241e-01 6.42991483e-01
6.00732923e-01 -1.43284023e-01 -7.75313914e-01 1.55949402e+00
-8.97179395e-02 -4.86333102e-01 -6.21832907e-01 -3.41743171e-01
8.29205394e-01 8.06354821e-01 -5.70030093e-01 -3.51268172e-01
4.36837047e-01 4.52752411e-01 1.48338854e-01 2.24559829e-01
-9.23296094e-01 -9.44408894e-01 4.21382993e-01 -3.43092591e-01
-2.08509713e-01 -1.25392631e-01 -7.49904871e-01 -5.92354655e-01
1.47230923e-02 -6.43668711e-01 -1.07456133e-01 -8.28547180e-01
-2.86497355e-01 6.52286053e-01 1.58128455e-01 -1.18897140e+00
-7.72870123e-01 -5.78417718e-01 -6.94874704e-01 9.38614368e-01
-6.46373749e-01 -7.58851528e-01 -1.41315028e-01 4.26237702e-01
-2.29726538e-01 8.54950547e-02 9.63400185e-01 3.38518709e-01
-6.18478596e-01 7.10600376e-01 -9.23948810e-02 -9.99229029e-02
-1.19456388e-01 -5.04404485e-01 2.80971289e-01 1.12999701e+00
-2.45573565e-01 8.72194409e-01 8.85299504e-01 -4.01251435e-01
-1.34917593e+00 -1.07256114e+00 9.12898123e-01 6.50136471e-01
5.17610133e-01 -5.13760597e-02 -8.40953648e-01 7.58310378e-01
-5.27466834e-02 -2.42355224e-02 -2.08602712e-01 -4.82412845e-01
-5.38736165e-01 -7.15189576e-01 -1.44669056e+00 7.94325888e-01
5.74825585e-01 -6.01473749e-01 -4.42221284e-01 -3.59790146e-01
1.22641063e+00 -4.76115942e-01 -1.00205112e+00 5.55936098e-01
9.47978795e-01 -6.93870723e-01 7.46829689e-01 -2.66961575e-01
6.19900227e-01 -4.90265191e-01 -5.13467863e-02 -1.00797474e+00
1.51761577e-01 -2.74848789e-01 -6.63192570e-01 1.27385092e+00
4.01731253e-01 -8.91449690e-01 6.15955651e-01 6.01192594e-01
-1.90418825e-01 -1.01533031e+00 -1.00848186e+00 -9.71004486e-01
3.91106717e-02 -3.28194261e-01 9.17223096e-01 4.28078562e-01
3.46065134e-01 -1.86188042e-01 1.37995273e-01 1.86560914e-01
3.22252482e-01 -9.15287435e-02 -1.05214976e-01 -1.11521971e+00
-2.44151920e-01 -8.22961986e-01 -6.67967737e-01 -4.32579100e-01
4.49777007e-01 -7.63248980e-01 -1.63415018e-02 -1.02528906e+00
-4.33534622e-01 -1.16108455e-01 -5.54753363e-01 1.76374108e-01
6.94392145e-01 1.26282647e-01 3.64534296e-02 -2.21659705e-01
6.18153252e-03 4.05727297e-01 6.24646664e-01 -2.60199159e-01
-6.33538067e-02 1.38015568e-01 -1.40942827e-01 4.86436486e-01
8.61169934e-01 -4.34627235e-01 -6.46850586e-01 -1.24525145e-01
2.31739044e-01 2.73621857e-01 4.14884955e-01 -1.56639981e+00
5.61980844e-01 -5.29529676e-02 -2.88132671e-02 -6.98066771e-01
1.45864502e-01 -1.35671747e+00 5.80786943e-01 1.28437853e+00
-4.93969530e-01 2.48235196e-01 3.15019041e-01 3.05812031e-01
-3.29620779e-01 -7.17949450e-01 6.43442333e-01 4.72933769e-01
-1.04978479e-01 9.12932232e-02 -3.85783792e-01 -4.15563017e-01
8.12830389e-01 -1.45328045e-02 -3.10316175e-01 -1.47591503e-02
-3.83692652e-01 -9.68954787e-02 1.31802499e-01 -2.46644601e-01
9.33459461e-01 -1.52481747e+00 -1.93033218e-02 5.60656309e-01
-3.40981334e-01 7.09127784e-02 3.96436006e-02 6.59841359e-01
-8.44426990e-01 3.95287663e-01 -4.96796906e-01 -2.32991889e-01
-1.12557566e+00 6.22309923e-01 6.95762694e-01 -6.03862181e-02
-2.75197681e-02 1.71285361e-01 -5.68276942e-01 -3.15012038e-01
4.35241699e-01 -1.08117700e+00 2.69160509e-01 -4.22655851e-01
3.22101772e-01 7.17551410e-01 2.89175153e-01 1.25791892e-01
-2.88488626e-01 2.92903990e-01 1.60285234e-01 -3.54933769e-01
8.83137524e-01 2.52222806e-01 -5.08029521e-01 3.08294624e-01
1.51310408e+00 -8.74267161e-01 -7.99270570e-01 9.92560945e-03
-2.59764493e-01 2.48906940e-01 5.72359562e-03 -3.22133482e-01
-1.23053157e+00 8.65245044e-01 6.81422114e-01 1.38703674e-01
1.50985944e+00 -6.46672010e-01 7.02898026e-01 9.91307080e-01
2.02951357e-01 -9.80639100e-01 -5.00868440e-01 5.73135853e-01
4.89123046e-01 6.87328950e-02 -1.79667443e-01 -1.05483755e-01
1.57325417e-01 1.46714556e+00 6.16401494e-01 -5.21720886e-01
5.79032421e-01 8.20339322e-01 -8.69886279e-01 3.02610189e-01
-6.11478090e-01 3.85388762e-01 1.60744697e-01 3.47911239e-01
-6.63358867e-02 -1.36591747e-01 -6.19823754e-01 9.35305357e-01
-1.48351476e-01 4.55424398e-01 6.05214179e-01 7.25378335e-01
-5.01135409e-01 -4.68657941e-01 5.30756405e-03 1.91058069e-01
1.06384255e-01 7.39776790e-02 -1.69007257e-01 7.95875669e-01
-7.87578598e-02 5.42287827e-01 2.45724425e-01 -9.78608072e-01
3.38315994e-01 1.73653379e-01 6.10337615e-01 1.98040798e-01
-4.65814292e-01 -6.34153903e-01 2.02370249e-02 -3.87406230e-01
2.27386709e-02 1.75161019e-01 -1.61551523e+00 -7.72421420e-01
-6.83999240e-01 2.26734757e-01 1.04309201e+00 1.03293765e+00
2.14132860e-01 6.75085247e-01 5.38803756e-01 -3.18820059e-01
-8.17033052e-01 -4.90372509e-01 -5.13506532e-01 -2.71856010e-01
1.24648005e-01 -9.33667868e-02 -3.82315487e-01 -2.37194166e-01]
|
[8.30261516571045, 3.1552932262420654]
|
d07def8e-1faf-4a76-80ea-1d172fcf0e72
|
gnowee-a-hybrid-metaheuristic-optimization
|
1804.05429
| null |
http://arxiv.org/abs/1804.05429v1
|
http://arxiv.org/pdf/1804.05429v1.pdf
|
Gnowee: A Hybrid Metaheuristic Optimization Algorithm for Constrained, Black Box, Combinatorial Mixed-Integer Design
|
This paper introduces Gnowee, a modular, Python-based, open-source hybrid
metaheuristic optimization algorithm (Available from
https://github.com/SlaybaughLab/Gnowee). Gnowee is designed for rapid
convergence to nearly globally optimum solutions for complex, constrained
nuclear engineering problems with mixed-integer and combinatorial design
vectors and high-cost, noisy, discontinuous, black box objective function
evaluations. Gnowee's hybrid metaheuristic framework is a new combination of a
set of diverse, robust heuristics that appropriately balance diversification
and intensification strategies across a wide range of optimization problems.
This novel algorithm was specifically developed to optimize complex nuclear
design problems; the motivating research problem was the design of material
stack-ups to modify neutron energy spectra to specific targeted spectra for
applications in nuclear medicine, technical nuclear forensics, nuclear physics,
etc. However, there are a wider range of potential applications for this
algorithm both within the nuclear community and beyond. To demonstrate Gnowee's
behavior for a variety of problem types, comparisons between Gnowee and several
well-established metaheuristic algorithms are made for a set of eighteen
continuous, mixed-integer, and combinatorial benchmarks. These results
demonstrate Gnoweee to have superior flexibility and convergence
characteristics over a wide range of design spaces. We anticipate this wide
range of applicability will make this algorithm desirable for many complex
engineering applications.
|
['James Bevins', 'Rachel Slaybaugh']
|
2018-04-15
| null | null | null | null |
['metaheuristic-optimization']
|
['methodology']
|
[ 1.60992920e-01 -3.21256995e-01 -2.15294629e-01 -1.63781960e-02
-7.12231159e-01 -5.63222349e-01 -1.56203723e-02 6.67039528e-02
-1.99646950e-01 1.10321152e+00 -8.11546221e-02 -3.47060353e-01
-8.92788947e-01 -6.76842272e-01 -1.76345587e-01 -1.16778207e+00
5.02607226e-02 8.39652658e-01 -1.08621135e-01 -4.63626474e-01
4.88785386e-01 9.69760835e-01 -1.37317371e+00 8.54679719e-02
8.08760822e-01 6.59359872e-01 3.47985327e-01 6.54289126e-01
1.33828118e-01 1.33770749e-01 -4.85329598e-01 9.00649000e-03
2.09851444e-01 -3.02022994e-01 -8.20610583e-01 -1.99064776e-01
-6.48714423e-01 6.27656341e-01 2.27521122e-01 9.31940258e-01
1.16665065e+00 4.09572244e-01 5.14248788e-01 -1.20425570e+00
-1.71653941e-01 5.32406271e-01 -7.63363242e-01 3.44612688e-01
5.00216007e-01 6.61152303e-01 8.38990211e-01 -6.40542448e-01
7.28547692e-01 9.20688868e-01 8.65297258e-01 5.05989909e-01
-1.00072682e+00 -4.98650491e-01 -4.02219266e-01 1.47844061e-01
-1.30153549e+00 1.61726810e-02 6.82594061e-01 -1.68006256e-01
1.20091915e+00 1.00529504e+00 7.51778185e-01 6.45021379e-01
7.12752402e-01 3.83412153e-01 1.03043616e+00 -4.39679921e-01
6.64286971e-01 -2.14066789e-01 3.33795249e-02 5.20330489e-01
3.38097811e-01 4.54610020e-01 -2.16938779e-01 -6.33771896e-01
2.76358783e-01 -1.10147879e-01 -3.87020379e-01 -4.65637594e-01
-1.37496579e+00 9.71015036e-01 3.53092402e-01 4.95923668e-01
-6.57937884e-01 4.46145348e-02 4.85270619e-01 6.36932850e-02
-2.03320123e-02 1.04532361e+00 -4.22363490e-01 -2.67394483e-01
-7.76224852e-01 6.40716314e-01 8.34260404e-01 5.68473101e-01
1.95014507e-01 3.37688439e-02 -3.93293232e-01 7.97757149e-01
2.73464650e-01 5.08406579e-01 2.31318697e-01 -6.95913970e-01
2.53954172e-01 2.72726268e-01 -6.60117418e-02 -7.69132793e-01
-9.29491520e-01 -7.73422658e-01 -7.04916835e-01 3.52830440e-01
-3.46190743e-02 -2.39343151e-01 -8.86133790e-01 1.16542220e+00
7.01301098e-01 -5.36206186e-01 -9.33678914e-03 1.01438463e+00
6.11687005e-01 8.16526711e-01 5.95663488e-02 -7.08073258e-01
1.15359378e+00 -9.86249208e-01 -5.82706928e-01 8.45955592e-03
3.74749273e-01 -1.02424932e+00 6.39564395e-01 4.72839147e-01
-1.48658359e+00 3.28214690e-02 -1.10654056e+00 5.18602312e-01
-4.24390882e-01 -4.58377302e-01 9.21785295e-01 1.19078422e+00
-7.63744771e-01 5.65584958e-01 -6.47655845e-01 -2.05871969e-01
2.55217046e-01 7.72040904e-01 2.80767322e-01 -2.81938046e-01
-9.56709266e-01 7.46456087e-01 6.88406348e-01 3.65333706e-01
-3.08540702e-01 -9.02391076e-01 -5.17625809e-01 -5.24560399e-02
6.01589322e-01 -9.86516356e-01 1.02959549e+00 -4.57177192e-01
-1.41353190e+00 4.48448092e-01 2.22438186e-01 3.42307910e-02
4.22712147e-01 6.73704207e-01 -4.69671190e-01 -1.79971442e-01
-3.99238467e-02 7.95681700e-02 1.17122889e-01 -1.01748931e+00
-3.08650911e-01 -6.65280670e-02 -2.83442676e-01 2.10828826e-01
1.33390054e-01 2.70896643e-01 -1.15703149e-02 -7.59703398e-01
8.12640414e-02 -8.51109982e-01 -8.15239966e-01 -6.50198102e-01
-3.53403568e-01 2.58081425e-02 5.71140349e-01 -2.77892590e-01
1.32997966e+00 -1.59233189e+00 6.03411734e-01 9.18155730e-01
-7.44989514e-02 -1.38476053e-02 8.04988518e-02 8.14710379e-01
-4.55419987e-01 3.14207189e-02 -6.60945594e-01 6.01178825e-01
1.15307018e-01 1.24058254e-01 5.45877874e-01 4.92612809e-01
-3.27785552e-01 9.60493326e-01 -1.21855938e+00 -2.32838407e-01
2.44678184e-01 7.69875944e-02 -6.15249038e-01 -4.10122842e-01
-2.70881414e-01 2.63155699e-01 -7.54160166e-01 1.39997303e+00
8.92429829e-01 4.41780537e-02 3.12534422e-01 -3.20389032e-01
-2.62253016e-01 -6.04885697e-01 -1.28373265e+00 1.47389531e+00
-2.28115678e-01 -1.84722245e-02 3.63002181e-01 -9.53955829e-01
5.77978969e-01 2.25633353e-01 1.21135879e+00 -6.43365562e-01
5.93392313e-01 5.12540698e-01 2.67291754e-01 -5.23025572e-01
5.81895888e-01 -5.74695170e-01 -1.36996597e-01 5.28345823e-01
-2.90212601e-01 -4.90988851e-01 6.23029768e-01 -2.92104930e-01
1.40276325e+00 -2.79508322e-01 4.29362565e-01 -7.73196638e-01
7.42941976e-01 3.80920798e-01 8.34619045e-01 4.81322914e-01
-1.44239053e-01 4.58441794e-01 -1.86375380e-02 -4.17519540e-01
-9.94813383e-01 -8.25700223e-01 -2.88229913e-01 7.48720229e-01
6.42939135e-02 -2.04624116e-01 -4.40765262e-01 -4.06092882e-01
1.95052288e-02 8.22970569e-01 -1.24243848e-01 -1.35323048e-01
-6.96120620e-01 -1.52945840e+00 2.02337980e-01 1.55204013e-01
9.99087915e-02 -1.20936656e+00 -7.55362391e-01 6.76627755e-01
-7.20742643e-02 -7.08888888e-01 -4.63422060e-01 3.49105388e-01
-8.24362993e-01 -1.22228622e+00 -6.05725348e-01 -6.16434216e-01
6.01975799e-01 -1.30665973e-01 1.37385559e+00 3.10037225e-01
-1.02187073e+00 5.30662239e-01 -3.71231407e-01 -2.98976898e-01
-5.84102929e-01 2.13644304e-03 -6.85100034e-02 -5.00494123e-01
-1.49419948e-01 -3.24438423e-01 -4.70675647e-01 7.07001269e-01
-1.01051807e+00 -2.72186011e-01 5.10218382e-01 1.13551509e+00
7.57682502e-01 6.24571681e-01 4.00403857e-01 -7.74037242e-01
7.58717060e-01 -6.98730826e-01 -5.89545667e-01 4.83070076e-01
-5.01475751e-01 1.65168300e-01 3.89567345e-01 -3.18032682e-01
-8.01691294e-01 -1.05534419e-01 -5.31942725e-01 2.54367776e-02
3.71318907e-01 7.25271881e-01 -2.02469200e-01 -5.79070091e-01
6.22571766e-01 -3.11516039e-02 -1.42311811e-01 -1.35020941e-01
-8.64202902e-02 3.93104702e-01 3.12280327e-01 -1.01043546e+00
7.58029342e-01 1.00792155e-01 3.86117488e-01 -6.33478642e-01
-2.04346508e-01 -5.05156636e-01 1.45540074e-01 -4.02637780e-01
7.14103162e-01 -1.12224810e-01 -9.38139260e-01 2.72435367e-01
-6.57188773e-01 -6.76059425e-02 -4.82816905e-01 4.50409144e-01
-5.65499008e-01 2.71870017e-01 -1.93197176e-01 -8.02634418e-01
-5.87431014e-01 -1.57710290e+00 5.35564959e-01 4.67267543e-01
-2.73749381e-01 -1.11153066e+00 7.68344849e-02 5.08057237e-01
5.39714515e-01 7.93336332e-01 1.06742072e+00 -2.99039096e-01
-4.59028214e-01 -2.68222298e-02 4.15249377e-01 -2.00172126e-01
1.15152851e-01 2.03006953e-01 -4.16474253e-01 -6.20177507e-01
2.56958485e-01 -8.29000697e-02 2.37198576e-01 7.68023551e-01
1.18761730e+00 8.58287364e-02 -7.20287204e-01 6.82391167e-01
1.67584467e+00 7.80649960e-01 8.04647803e-01 8.93711030e-01
1.40988484e-01 1.70981556e-01 9.28287566e-01 6.43820643e-01
-2.77396470e-01 6.69561267e-01 5.07705212e-01 -1.82437003e-01
2.69544989e-01 6.47169054e-01 -7.49806315e-02 6.71717823e-01
-3.00070077e-01 -4.07883495e-01 -1.11069870e+00 3.39742452e-01
-1.62538159e+00 -1.16713822e+00 -9.28012282e-02 2.01640224e+00
6.38715446e-01 4.51316349e-02 1.52135655e-01 3.35618824e-01
1.07143235e+00 -1.72723934e-01 -6.26842201e-01 -9.29241300e-01
-1.30217582e-01 5.87231576e-01 8.51676404e-01 3.82519513e-02
-6.74323976e-01 2.17807353e-01 6.61036348e+00 1.25936937e+00
-1.07710779e+00 1.08396746e-02 6.66075408e-01 -4.66519266e-01
-6.49970114e-01 -1.69221789e-01 -4.20873642e-01 4.42840427e-01
8.26134264e-01 -4.60184097e-01 5.04859924e-01 3.25094402e-01
2.87157327e-01 -7.47116804e-02 -6.12064481e-01 9.11485076e-01
-2.10106447e-01 -1.72410476e+00 -6.82264626e-01 1.58408269e-01
1.06555319e+00 -1.63044497e-01 -1.33455381e-01 1.30208239e-01
1.80179372e-01 -1.03324449e+00 6.77528381e-01 2.59637892e-01
3.74457210e-01 -1.29555404e+00 8.03979099e-01 -1.68491639e-02
-1.04986846e+00 -5.31853616e-01 -2.59885881e-02 4.29458499e-01
6.33294463e-01 8.47454131e-01 -5.97970843e-01 1.26313555e+00
6.45969093e-01 2.34911330e-02 -4.27569486e-02 1.75803626e+00
5.07189751e-01 1.48934394e-01 -4.83391464e-01 -3.12597394e-01
5.21418273e-01 -2.93548763e-01 9.52520847e-01 9.45463777e-01
4.89315540e-01 3.76831621e-01 2.46962115e-01 6.93363488e-01
4.38674122e-01 1.81896120e-01 -1.38190851e-01 -2.30972663e-01
4.53873336e-01 1.28510976e+00 -1.25679743e+00 3.34347963e-01
7.44794980e-02 3.30468625e-01 -4.69809651e-01 2.18243912e-01
-1.39243317e+00 -5.31311035e-01 5.68650782e-01 1.46747958e-02
2.08662763e-01 -1.84919104e-01 -6.43661439e-01 -3.99674475e-01
-3.58162791e-01 -1.15433502e+00 8.22855413e-01 -5.77777982e-01
-1.14762139e+00 4.92038995e-01 2.62282252e-01 -1.14335692e+00
-1.32121116e-01 -6.48599386e-01 -7.61892974e-01 6.39339447e-01
-1.20481014e+00 -7.58029222e-01 -6.07219487e-02 2.86349297e-01
5.09944439e-01 -4.44349051e-01 3.70234698e-01 3.76042694e-01
-6.86521590e-01 4.12541211e-01 5.69614351e-01 -7.46449590e-01
1.38476849e-01 -9.73489761e-01 8.78564864e-02 3.78037930e-01
-7.41053343e-01 3.90568346e-01 1.18776560e+00 -6.54455602e-01
-2.16257691e+00 -5.52891433e-01 2.12787271e-01 1.75127640e-01
4.98860300e-01 1.38433859e-01 -2.93233573e-01 -6.22666022e-03
2.53417552e-01 -2.46102393e-01 7.01742053e-01 -6.52878284e-02
5.89816749e-01 7.15283677e-02 -1.82011938e+00 5.07035017e-01
8.74974847e-01 3.35229397e-01 -8.15097243e-02 7.66055346e-01
2.31552973e-01 -9.84365463e-01 -1.28321743e+00 7.84626722e-01
3.79155010e-01 -9.80495095e-01 1.46683204e+00 -1.81252450e-01
1.83420718e-01 -3.77499521e-01 -6.28651306e-02 -1.28594184e+00
-5.03682613e-01 -8.56763363e-01 2.14817822e-01 9.37549055e-01
5.70099711e-01 -8.29915166e-01 8.88285100e-01 6.14434898e-01
-5.54824054e-01 -1.32658172e+00 -1.10801840e+00 -9.93503809e-01
6.80387095e-02 -2.18520135e-01 9.39860046e-01 8.48144531e-01
-9.88895744e-02 -2.26354018e-01 1.24134785e-02 5.15789650e-02
7.65611112e-01 1.71448231e-01 1.20911151e-01 -7.98876822e-01
-5.59301317e-01 -6.73820078e-01 -4.46267039e-01 -1.66802872e-02
-2.81520426e-01 -9.31702971e-01 -3.94751914e-02 -1.43216729e+00
1.15975901e-01 -7.98278809e-01 -2.93043733e-01 2.81570703e-01
6.09391667e-02 1.74179554e-01 1.84600368e-01 -2.75382340e-01
-6.62121326e-02 4.75186318e-01 1.34445655e+00 -5.24223745e-01
-3.36100489e-01 2.71270871e-01 -7.10346937e-01 3.62448215e-01
8.32203984e-01 -6.03469908e-01 -2.41186142e-01 5.85368238e-02
4.22231078e-01 3.63752991e-01 1.84462927e-02 -1.18902230e+00
1.69330955e-01 -8.54567170e-01 2.25255787e-01 -7.34859645e-01
1.21352121e-01 -9.19119000e-01 1.02007473e+00 8.75428021e-01
2.55076319e-01 4.44629669e-01 3.49468619e-01 3.00793082e-01
-1.44128613e-02 -7.32123792e-01 9.56530213e-01 -3.26682270e-01
-6.56720519e-01 1.53121248e-01 -2.84828305e-01 5.85641665e-03
1.57859385e+00 -7.10204780e-01 -1.51219338e-01 3.54771227e-01
-8.71387243e-01 4.66916800e-01 4.81109619e-01 -4.13900428e-03
4.39499825e-01 -1.43597639e+00 -4.52364624e-01 -1.15092672e-01
-1.95526719e-01 -2.40724847e-01 4.94626671e-01 9.04870391e-01
-9.10383165e-01 2.77810752e-01 -3.31783324e-01 -6.93000972e-01
-1.28584445e+00 3.94054145e-01 6.43683434e-01 -5.68218231e-01
-4.91355866e-01 8.33683491e-01 -5.22080243e-01 -6.69577181e-01
-3.52544159e-01 -2.44833916e-01 2.99648851e-01 -7.12462738e-02
2.10473001e-01 9.29919004e-01 5.26839793e-01 -2.28229880e-01
-5.80578744e-01 6.51643038e-01 2.30806157e-01 1.20274514e-01
1.75197148e+00 2.31871113e-01 -3.47493649e-01 -1.46235704e-01
9.63329315e-01 -2.53277235e-02 -5.14718056e-01 4.50363427e-01
1.07370064e-01 -5.14793396e-01 2.00180858e-01 -1.03936064e+00
-1.26748729e+00 -8.16882476e-02 6.73352718e-01 1.11582719e-01
1.60361993e+00 -2.85588920e-01 7.75103092e-01 1.93191767e-01
4.16774154e-01 -1.29679167e+00 -2.71830946e-01 5.00635087e-01
9.19977963e-01 -6.28297806e-01 7.34430909e-01 -4.36640859e-01
-6.14489496e-01 1.35017860e+00 2.82436281e-01 4.15628493e-01
3.83029431e-01 5.23468733e-01 -3.00214797e-01 -5.67704022e-01
-4.82850939e-01 3.93814631e-02 5.77676436e-03 4.87939388e-01
3.48755479e-01 -4.35288846e-02 -8.61542761e-01 2.34998956e-01
-1.35836065e-01 -2.24265512e-02 4.12069440e-01 1.54434991e+00
-4.44399804e-01 -1.42243052e+00 -9.41707134e-01 4.25766438e-01
-3.05932790e-01 9.42877084e-02 2.04772756e-01 8.22200477e-01
1.02231711e-01 7.37829864e-01 -5.88886380e-01 1.20430022e-01
3.93208683e-01 -2.03339443e-01 6.46839440e-01 -2.28304997e-01
-1.28968084e+00 -9.88946669e-03 1.95530325e-01 -6.29448831e-01
-4.11999732e-01 -8.89248312e-01 -1.43960094e+00 -5.49391270e-01
-4.34673220e-01 5.64322889e-01 9.11396563e-01 9.90375280e-01
-2.44042519e-02 1.03321171e+00 6.65009439e-01 -9.87546980e-01
-5.56655765e-01 -1.73960835e-01 -6.64847195e-01 -1.53560147e-01
-1.98196426e-01 -9.53682005e-01 -3.68837975e-02 -6.98326051e-01]
|
[5.754472732543945, 3.6067540645599365]
|
a3b640d4-4d25-46cc-ac00-b0346ddca207
|
robust-speech-recognition-using-generative
|
1711.01567
| null |
http://arxiv.org/abs/1711.01567v1
|
http://arxiv.org/pdf/1711.01567v1.pdf
|
Robust Speech Recognition Using Generative Adversarial Networks
|
This paper describes a general, scalable, end-to-end framework that uses the
generative adversarial network (GAN) objective to enable robust speech
recognition. Encoders trained with the proposed approach enjoy improved
invariance by learning to map noisy audio to the same embedding space as that
of clean audio. Unlike previous methods, the new framework does not rely on
domain expertise or simplifying assumptions as are often needed in signal
processing, and directly encourages robustness in a data-driven way. We show
the new approach improves simulated far-field speech recognition of vanilla
sequence-to-sequence models without specialized front-ends or preprocessing.
|
['Anuroop Sriram', 'Yashesh Gaur', 'Heewoo Jun', 'Sanjeev Satheesh']
|
2017-11-05
| null | null | null | null |
['robust-speech-recognition']
|
['speech']
|
[ 4.40081328e-01 1.34874657e-01 3.82595688e-01 -4.53431100e-01
-1.36232364e+00 -5.95843911e-01 7.86435425e-01 -6.79980874e-01
-4.28694546e-01 8.00572336e-01 4.35734779e-01 -3.56774867e-01
4.10633758e-02 -4.28770453e-01 -9.09316301e-01 -7.73880243e-01
-8.94793421e-02 -1.10259302e-01 4.22958881e-02 -2.62995809e-01
-1.52712286e-01 4.59359229e-01 -1.13289523e+00 1.82487026e-01
4.29281741e-01 9.98098075e-01 1.87585771e-01 1.24554694e+00
3.39768082e-01 7.67161727e-01 -9.22094584e-01 -2.66617239e-01
5.40198624e-01 -5.63223243e-01 -4.07736033e-01 -5.96649274e-02
4.45901811e-01 -3.14755887e-01 -9.21267092e-01 1.15780401e+00
9.73909736e-01 3.46384972e-01 5.81817865e-01 -9.17917669e-01
-7.06188917e-01 3.84799421e-01 -9.80036147e-03 2.49865279e-01
5.67707658e-01 3.37261498e-01 6.08698964e-01 -8.90211821e-01
4.29061770e-01 1.24470389e+00 7.04230249e-01 6.30889177e-01
-1.17527974e+00 -7.42225647e-01 -6.03705309e-02 9.16084498e-02
-1.35781443e+00 -9.95195866e-01 8.89653444e-01 -1.92203373e-01
9.34207797e-01 3.30719799e-01 2.93241084e-01 1.85004151e+00
-2.49479283e-02 5.47093511e-01 1.00600278e+00 -4.48705614e-01
4.45347846e-01 6.86896518e-02 -4.23261434e-01 2.46881098e-01
-2.82515347e-01 5.46995044e-01 -5.34408212e-01 -9.06492695e-02
6.90076172e-01 -2.34343961e-01 -5.26790798e-01 -2.96036929e-01
-1.25284612e+00 7.01349556e-01 3.85398060e-01 2.97319978e-01
-4.90353793e-01 3.76832932e-01 4.83969331e-01 8.15817714e-01
3.54954541e-01 3.86990011e-01 -2.11539581e-01 -3.29670608e-01
-1.35884535e+00 -2.01512929e-02 6.03105485e-01 9.37865555e-01
1.85720101e-01 1.03502142e+00 -5.05983979e-02 7.95672596e-01
3.84392112e-01 8.78811479e-01 7.60390878e-01 -8.91425014e-01
4.16418493e-01 -4.58476722e-01 -1.26515239e-01 -5.94274342e-01
2.34200373e-01 -6.59890950e-01 -9.18701947e-01 5.37061453e-01
5.99616878e-02 -3.75318825e-01 -1.12656391e+00 1.69101572e+00
3.85987237e-02 5.17637014e-01 3.72950017e-01 7.76943147e-01
4.84547615e-01 9.02408123e-01 -2.51488537e-01 1.08142821e-02
8.51647437e-01 -9.91041362e-01 -8.78746688e-01 -3.27257037e-01
-9.99366269e-02 -9.85780180e-01 9.60579932e-01 4.93841559e-01
-1.18170488e+00 -5.95790684e-01 -1.35164106e+00 1.28297895e-01
-2.69688517e-01 -1.70683980e-01 1.71824172e-01 1.02314556e+00
-1.30244279e+00 4.39404696e-01 -7.28947997e-01 -1.35845751e-01
2.87068188e-01 1.70464799e-01 -5.82196951e-01 2.30705347e-02
-1.20128262e+00 8.63890827e-01 1.31094962e-01 6.06852211e-02
-1.54860520e+00 -5.09888291e-01 -1.07620692e+00 -6.64663874e-03
1.97376847e-01 -6.56862676e-01 1.43100369e+00 -1.01371729e+00
-2.27514935e+00 2.39847258e-01 -1.91481575e-01 -7.29125738e-01
5.84149718e-01 -3.77485633e-01 -7.66605735e-01 2.91989207e-01
-3.35945845e-01 4.11129057e-01 1.63059628e+00 -1.08207905e+00
8.94652754e-02 1.17599443e-01 -1.76287979e-01 -8.73498172e-02
-1.85166046e-01 2.19832852e-01 -6.87140301e-02 -1.13359904e+00
-1.11192465e-01 -5.34079194e-01 -2.18365461e-01 -7.12785125e-02
-4.39310640e-01 4.41246390e-01 9.73649561e-01 -8.47592890e-01
8.30230236e-01 -2.33978391e+00 1.16887137e-01 3.95504087e-01
-1.43369094e-01 6.61345661e-01 -3.81182611e-01 6.64476812e-01
-3.88338029e-01 -1.25754997e-01 -3.05042475e-01 -4.62978870e-01
2.55388707e-01 1.91117331e-01 -7.57900894e-01 5.76446354e-01
2.94556081e-01 6.52613342e-01 -8.96164656e-01 -3.32112163e-02
5.22835016e-01 8.60757470e-01 -6.48777485e-01 4.94851500e-01
1.43312946e-01 6.16130471e-01 4.39441986e-02 3.53542060e-01
6.29959226e-01 3.76320869e-01 -2.11160675e-01 8.49574506e-02
2.77212471e-01 6.27577484e-01 -1.16005492e+00 1.84334707e+00
-9.09993827e-01 9.14139926e-01 4.89032775e-01 -1.23300564e+00
9.21841800e-01 8.45884264e-01 -8.79445523e-02 -4.40861166e-01
-4.85576577e-02 3.42313737e-01 -7.28710145e-02 -3.17811131e-01
2.71853097e-02 -2.74312824e-01 1.07824601e-01 1.53102398e-01
4.95199323e-01 -4.59375530e-01 -4.49124008e-01 7.43728578e-02
1.31756496e+00 -6.67465804e-03 3.40581179e-01 -2.65395530e-02
5.20837665e-01 -6.75083399e-01 3.04557115e-01 9.07228172e-01
-5.26464060e-02 8.46729040e-01 -3.97928022e-02 9.32985768e-02
-1.23561764e+00 -1.39656854e+00 -1.02559723e-01 9.05041099e-01
-4.77820456e-01 -1.29420429e-01 -8.58966112e-01 -6.37263894e-01
-3.62265080e-01 8.55515718e-01 -4.04278010e-01 -2.83118427e-01
-3.23460609e-01 3.71296778e-02 1.17855787e+00 5.99530041e-01
3.43155712e-01 -9.73844111e-01 -8.85839202e-03 3.31927299e-01
6.78725690e-02 -1.21335602e+00 -7.07640886e-01 3.95772785e-01
-4.58803803e-01 -4.33006883e-01 -1.13054967e+00 -7.78062761e-01
4.86677289e-01 -2.24074442e-03 7.75466859e-01 -4.90073860e-01
-2.15995107e-02 5.16960025e-01 -3.16022456e-01 -2.91039735e-01
-7.01014757e-01 -1.95525229e-01 3.32627267e-01 1.80692494e-01
5.00203632e-02 -8.98629427e-01 -2.99567103e-01 2.15167135e-01
-9.09947097e-01 -5.13929129e-01 6.34147346e-01 1.22011292e+00
2.35390753e-01 -1.01824127e-01 8.39996219e-01 -4.15021330e-01
7.04718292e-01 -3.33654344e-01 -4.20811504e-01 -1.16089381e-01
-2.55841583e-01 2.89793722e-02 1.05506647e+00 -6.08266592e-01
-8.88922811e-01 7.72622973e-02 -8.15799236e-01 -8.69830966e-01
-5.31754434e-01 1.02776721e-01 -4.03405249e-01 -2.36525536e-01
7.81827211e-01 5.77833176e-01 6.58696331e-03 -5.36390364e-01
5.82597077e-01 9.59771037e-01 1.04353595e+00 -2.44205877e-01
1.40560877e+00 2.98012644e-01 -3.56804341e-01 -1.18192458e+00
-1.69268087e-01 -3.30677122e-01 -5.28075933e-01 4.75205295e-02
6.19206548e-01 -1.07193160e+00 -2.73276985e-01 3.43756825e-01
-1.20372903e+00 -2.91450143e-01 -5.26479423e-01 8.20566714e-01
-8.83694768e-01 4.72022176e-01 -5.26959062e-01 -9.38461125e-01
-4.21398103e-01 -1.04227197e+00 1.18276727e+00 -2.57866263e-01
-2.21120715e-01 -9.43317533e-01 1.74529910e-01 1.21513456e-01
7.46570468e-01 3.35593373e-02 2.88142234e-01 -7.73773909e-01
-4.78895396e-01 -4.82959211e-01 3.02601546e-01 1.07289958e+00
3.96170765e-02 -1.70100391e-01 -1.50954258e+00 -5.77203691e-01
2.21752152e-01 -3.93116117e-01 7.09891677e-01 2.84691095e-01
9.04537916e-01 -6.99759364e-01 1.45364866e-01 9.37150300e-01
1.06121600e+00 3.50722522e-01 9.42132771e-01 9.77413356e-02
3.26946586e-01 1.41180038e-01 1.15440913e-01 1.27237737e-01
-3.30269903e-01 7.32290089e-01 2.65063047e-01 -2.22734898e-01
-3.91382188e-01 -4.48710829e-01 8.95228088e-01 9.86664057e-01
2.46835366e-01 -2.80399024e-01 -5.61095715e-01 6.24482691e-01
-1.30444169e+00 -1.43487322e+00 5.77095747e-01 2.14754415e+00
8.21175337e-01 2.31659375e-02 -8.72090980e-02 5.41376472e-01
6.08640611e-01 3.37854892e-01 -2.97179490e-01 -5.22034168e-01
7.26234121e-03 7.51100600e-01 3.22131574e-01 8.70449066e-01
-9.37143981e-01 7.06434786e-01 7.67515564e+00 1.07410574e+00
-1.33266664e+00 2.86771089e-01 5.73616549e-02 -1.78809226e-01
-2.60543138e-01 -2.58892179e-01 -1.42398626e-01 2.98318982e-01
1.16946459e+00 -8.90400484e-02 8.35174024e-01 9.32396591e-01
2.30594084e-01 5.92108488e-01 -1.11921561e+00 9.70949113e-01
1.73296794e-01 -1.02162838e+00 -1.99592158e-01 -2.95842420e-02
4.67245102e-01 3.72138023e-02 4.26479280e-01 2.89496154e-01
4.24734980e-01 -1.31478441e+00 8.32869411e-01 4.32417482e-01
1.00381184e+00 -7.13216126e-01 6.60682559e-01 3.16093683e-01
-9.35521901e-01 2.81530619e-02 -3.52074027e-01 -4.85906787e-02
3.25567812e-01 5.24275482e-01 -1.18847096e+00 4.66827393e-01
4.20171529e-01 1.63188457e-01 -1.73694924e-01 9.49291587e-01
-3.63826334e-01 1.08740604e+00 -3.20277631e-01 1.55449331e-01
3.42409939e-01 1.56164482e-01 9.86175179e-01 1.36596406e+00
5.17194033e-01 -3.91744375e-01 -1.57105356e-01 6.11276805e-01
-8.42649862e-02 -1.47817090e-01 -1.10258162e+00 -1.41762331e-01
4.35000211e-01 9.17520940e-01 -1.19135506e-01 -2.82215357e-01
-3.10078442e-01 1.11971354e+00 -1.13234758e-01 7.24509180e-01
-7.62257636e-01 -8.36264014e-01 7.45274246e-01 -5.23866666e-03
6.86456501e-01 -3.45204681e-01 -3.23086791e-02 -1.09523571e+00
3.33620347e-02 -1.28481436e+00 -2.13369489e-01 -8.05953264e-01
-1.34406757e+00 9.52453375e-01 -2.82462060e-01 -1.28127623e+00
-8.03110182e-01 -4.12679434e-01 -8.02067280e-01 1.14230144e+00
-1.53213847e+00 -1.00916111e+00 6.60116524e-02 9.65963364e-01
5.54107785e-01 -5.89361787e-01 1.10154402e+00 4.37599182e-01
-1.61239102e-01 9.84651804e-01 4.56394434e-01 2.09298223e-01
7.08062708e-01 -1.20702684e+00 7.13937342e-01 1.32677805e+00
4.23273444e-01 6.77044928e-01 9.72828805e-01 -1.74301788e-01
-1.15304279e+00 -1.10105562e+00 6.24626994e-01 -3.19051385e-01
5.77560902e-01 -6.41134143e-01 -8.37012827e-01 7.54558921e-01
7.20547557e-01 2.47307479e-01 5.98201275e-01 -1.77909330e-01
-5.84623694e-01 -2.16989174e-01 -1.19149113e+00 3.72816950e-01
9.59794462e-01 -1.04748631e+00 -9.66647387e-01 2.05007210e-01
6.48439229e-01 -3.13479006e-01 -5.62487960e-01 1.06159166e-01
2.59519339e-01 -6.83661580e-01 1.28261995e+00 -5.58348894e-01
3.24904695e-02 -3.64759713e-01 -5.20630121e-01 -1.78354156e+00
-2.82245755e-01 -1.30857635e+00 -1.42172709e-01 1.13993633e+00
3.33163321e-01 -7.88912058e-01 3.28754038e-01 -1.63320661e-01
-5.13893485e-01 -1.98992848e-01 -1.30927885e+00 -1.19524240e+00
4.76712920e-02 -7.51120269e-01 5.74242353e-01 6.58108413e-01
-1.21835232e-01 1.39551014e-01 -6.94673240e-01 5.61556220e-01
5.66295624e-01 -5.61870575e-01 9.97833967e-01 -6.74765527e-01
-7.86767781e-01 -2.51078695e-01 -8.07643533e-01 -1.16371143e+00
1.64988786e-01 -7.73771822e-01 3.15095276e-01 -1.14026344e+00
-6.95846140e-01 7.85529763e-02 -3.17115337e-01 3.53954762e-01
4.38731909e-02 3.67449194e-01 1.65232286e-01 -2.99127400e-01
-2.82568574e-01 8.50200951e-01 8.19860518e-01 -3.32441002e-01
1.87819496e-01 1.03008471e-01 -5.02700627e-01 5.64625144e-01
6.23354733e-01 -4.60417807e-01 -6.89570546e-01 -3.52521896e-01
-3.96258295e-01 4.95191664e-02 4.49418843e-01 -1.29950941e+00
1.64462060e-01 1.66513696e-01 3.73076826e-01 7.04425722e-02
7.68811285e-01 -1.01168299e+00 1.37764111e-01 2.10527286e-01
-4.02779877e-01 -1.67431101e-01 2.07477450e-01 6.88135803e-01
-5.78695714e-01 -2.73997873e-01 9.89120722e-01 1.44629389e-01
-2.93441117e-01 -4.85031754e-02 -5.00537515e-01 -1.56121585e-03
9.53146696e-01 5.89870615e-04 6.82993233e-02 -1.08062780e+00
-8.22118402e-01 -3.66849959e-01 2.54111260e-01 3.20499837e-01
6.59304261e-01 -1.42304528e+00 -8.17085862e-01 6.58234119e-01
-2.26177186e-01 -2.65840620e-01 1.91836059e-01 5.82566202e-01
-3.23862910e-01 4.91764635e-01 -9.76682380e-02 -4.47534800e-01
-1.06950986e+00 6.82665408e-01 4.50000793e-01 6.49476796e-02
-7.95017362e-01 9.86370385e-01 -1.78548861e-02 -5.38881600e-01
4.69302267e-01 -2.06646651e-01 2.59666085e-01 -4.48155820e-01
7.37783909e-01 1.96433678e-01 3.76781642e-01 -6.15300298e-01
-3.54360461e-01 1.77884504e-01 1.20286569e-01 -6.70531452e-01
1.28892648e+00 6.88083470e-02 6.18553281e-01 3.40654433e-01
1.39438570e+00 3.90474081e-01 -1.30570996e+00 -2.33291045e-01
-4.38307405e-01 -6.49279833e-01 4.30089593e-01 -6.60791099e-01
-9.38419938e-01 1.19987071e+00 6.46103442e-01 2.25199223e-01
1.12826133e+00 -3.41446042e-01 8.33520949e-01 5.15141666e-01
2.67002434e-01 -9.15770650e-01 5.59212305e-02 4.34195280e-01
1.18098676e+00 -8.13550055e-01 -4.64764625e-01 -1.84014644e-02
-5.58816135e-01 9.78631794e-01 4.36493307e-02 -2.10683703e-01
6.19795263e-01 6.75691187e-01 4.20074344e-01 3.18789810e-01
-7.96896636e-01 -1.49687231e-01 2.79287577e-01 1.30534160e+00
1.99462980e-01 -2.63658345e-01 2.89371103e-01 3.12882930e-01
-4.72909093e-01 -1.14599608e-01 2.94361144e-01 7.92388737e-01
-1.84939384e-01 -1.05981863e+00 -7.09690154e-01 1.08396344e-01
-5.56624532e-01 -3.49583566e-01 -6.02103993e-02 3.69303942e-01
-2.42413893e-01 1.08973014e+00 -1.18898824e-01 -5.11399150e-01
3.24296832e-01 4.65377122e-01 3.62419933e-01 -4.17916298e-01
-4.08679426e-01 2.36197188e-01 4.53728400e-02 -3.94935101e-01
-1.26461074e-01 -5.59110940e-01 -7.56129384e-01 -2.26278454e-01
-3.03597242e-01 1.69433400e-01 7.73556352e-01 7.74701416e-01
4.58942771e-01 7.48650730e-01 9.90119636e-01 -1.10627389e+00
-9.57661927e-01 -1.21915543e+00 -4.60316449e-01 3.76678407e-01
9.56486106e-01 -3.29837680e-01 -5.60584962e-01 1.05050839e-01]
|
[15.035205841064453, 6.229610443115234]
|
73f22ecb-bd13-41bf-8ee6-e5171053ea06
|
cross-lingual-alignment-vs-joint-training-a
|
1910.04708
| null |
https://arxiv.org/abs/1910.04708v4
|
https://arxiv.org/pdf/1910.04708v4.pdf
|
Cross-lingual Alignment vs Joint Training: A Comparative Study and A Simple Unified Framework
|
Learning multilingual representations of text has proven a successful method for many cross-lingual transfer learning tasks. There are two main paradigms for learning such representations: (1) alignment, which maps different independently trained monolingual representations into a shared space, and (2) joint training, which directly learns unified multilingual representations using monolingual and cross-lingual objectives jointly. In this paper, we first conduct direct comparisons of representations learned using both of these methods across diverse cross-lingual tasks. Our empirical results reveal a set of pros and cons for both methods, and show that the relative performance of alignment versus joint training is task-dependent. Stemming from this analysis, we propose a simple and novel framework that combines these two previously mutually-exclusive approaches. Extensive experiments demonstrate that our proposed framework alleviates limitations of both approaches, and outperforms existing methods on the MUSE bilingual lexicon induction (BLI) benchmark. We further show that this framework can generalize to contextualized representations such as Multilingual BERT, and produces state-of-the-art results on the CoNLL cross-lingual NER benchmark.
|
['Zirui Wang', 'Jiateng Xie', 'Jaime Carbonell', 'Yiming Yang', 'Ruochen Xu', 'Graham Neubig']
|
2019-10-10
|
cross-lingual-alignment-vs-joint-training-a-1
|
https://openreview.net/forum?id=S1l-C0NtwS
|
https://openreview.net/pdf?id=S1l-C0NtwS
|
iclr-2020-1
|
['cross-lingual-ner']
|
['natural-language-processing']
|
[-3.94235440e-02 -1.60950705e-01 -7.19496965e-01 -4.81716543e-01
-1.52133012e+00 -8.79321218e-01 1.04032779e+00 -6.37519136e-02
-4.89617527e-01 9.51181412e-01 4.92699146e-01 -6.03756607e-01
1.31894037e-01 -4.07955438e-01 -9.20227468e-01 -5.09612322e-01
2.28077799e-01 6.06221199e-01 -1.66948751e-01 -4.18020993e-01
-1.30857885e-01 1.94245175e-01 -1.23088980e+00 2.16086626e-01
9.40222681e-01 4.80559409e-01 3.23155448e-02 1.09193988e-01
-2.69803405e-01 7.34273732e-01 -2.94742078e-01 -4.64958340e-01
1.45304367e-01 -4.97109711e-01 -1.10218132e+00 -3.04837674e-01
4.49101001e-01 1.51425838e-01 -2.11435676e-01 1.02353871e+00
4.19686109e-01 9.96603630e-03 8.73391271e-01 -8.43562007e-01
-1.05493557e+00 1.03198946e+00 -7.09734559e-01 2.42786363e-01
2.73216635e-01 -3.74628901e-01 1.33343124e+00 -9.61634755e-01
7.42970645e-01 1.33816969e+00 7.22185731e-01 3.50329638e-01
-1.23381245e+00 -9.58263338e-01 4.86661196e-01 1.82488978e-01
-1.33558571e+00 -5.90349674e-01 5.26146293e-01 -4.38329518e-01
1.02271390e+00 -1.72239706e-01 -4.83634463e-03 1.19495261e+00
7.04326183e-02 1.03827238e+00 1.58895886e+00 -8.94824147e-01
-3.43929321e-01 2.93661207e-01 1.85271755e-01 4.79414314e-01
2.77531534e-01 8.19673911e-02 -4.81055468e-01 -3.37661291e-03
4.69020933e-01 -3.16848755e-01 -3.21321636e-01 -5.17363250e-01
-1.48641312e+00 1.02030563e+00 2.74640054e-01 8.10298562e-01
8.61603692e-02 -1.04681358e-01 6.26938701e-01 4.87956136e-01
6.41320288e-01 3.10540140e-01 -6.99376106e-01 1.97464079e-01
-7.83749878e-01 -4.03336547e-02 7.25048363e-01 1.03626847e+00
1.06829619e+00 2.57452410e-02 7.87303746e-02 1.10801804e+00
2.61732161e-01 5.95494509e-01 8.44180405e-01 -2.45664492e-01
7.06444860e-01 2.11044312e-01 -2.23423272e-01 -5.13270020e-01
-2.72789150e-01 -3.24001968e-01 -6.06024802e-01 -1.91877797e-01
3.02538246e-01 -1.68747410e-01 -5.25862217e-01 2.26686144e+00
-4.34868373e-02 9.50438902e-03 5.04109919e-01 4.60525870e-01
8.07203054e-01 5.89310110e-01 2.81613231e-01 -1.65170074e-01
1.28195739e+00 -1.23392808e+00 -8.55204582e-01 -3.54261309e-01
8.24565470e-01 -1.02451813e+00 9.05988097e-01 -9.32908133e-02
-8.52031946e-01 -6.26323044e-01 -1.23667991e+00 -2.76070893e-01
-8.17410469e-01 3.47711951e-01 8.14462364e-01 7.03666747e-01
-1.01388574e+00 3.64992648e-01 -5.73290050e-01 -6.08582914e-01
1.57505870e-02 2.10998580e-01 -4.44102794e-01 -3.28334719e-01
-1.38323081e+00 1.22777259e+00 4.38963890e-01 -1.70853361e-01
-7.02490747e-01 -4.20284629e-01 -1.12800050e+00 -1.86590850e-01
2.20723152e-02 -3.97513717e-01 1.17087996e+00 -1.00316942e+00
-1.44331503e+00 1.24543881e+00 -2.25309938e-01 -2.65411466e-01
1.43163338e-01 -5.41249394e-01 -5.45459330e-01 -3.01900119e-01
5.18430710e-01 6.30058110e-01 3.82631272e-01 -1.35779762e+00
-5.84095240e-01 -2.82012552e-01 2.67196037e-02 4.50977087e-01
-3.68630767e-01 3.72662872e-01 -5.12196243e-01 -8.10546875e-01
-9.43484530e-02 -9.91576612e-01 4.32264106e-03 -7.79074192e-01
-2.53794432e-01 -4.51451719e-01 3.97051007e-01 -7.71514058e-01
1.04029751e+00 -1.77093232e+00 3.62146050e-01 -1.53664917e-01
-2.82811463e-01 1.79837391e-01 -3.17302406e-01 5.18818080e-01
-3.06921542e-01 6.69992790e-02 -1.36869401e-01 -4.90168720e-01
3.67551930e-02 3.34750056e-01 -4.03282851e-01 4.43705678e-01
1.39910594e-01 1.11170924e+00 -1.09724808e+00 -5.17943442e-01
6.06805310e-02 6.19094551e-01 -3.07248563e-01 1.90616608e-01
1.29281789e-01 8.18790734e-01 -2.85464078e-01 6.76776350e-01
4.25035149e-01 -3.06631904e-02 8.38415444e-01 -2.60036856e-01
-2.05356032e-01 8.24387670e-01 -8.29375148e-01 2.14726281e+00
-9.19377506e-01 5.91978669e-01 -2.38502920e-01 -1.21254218e+00
9.03422177e-01 5.92253804e-01 1.90084994e-01 -7.28577256e-01
1.08464979e-01 5.87892830e-01 -1.57254636e-01 -2.04112485e-01
3.93509209e-01 -2.79839069e-01 -5.02602518e-01 8.60002816e-01
7.03008711e-01 1.78640708e-01 2.93419600e-01 -1.02895632e-01
5.46078146e-01 7.43988156e-01 9.03062284e-01 -7.01092005e-01
6.54984772e-01 -2.89105713e-01 5.30786395e-01 6.47725105e-01
-1.15758277e-01 3.26910675e-01 1.23256721e-01 -4.05299425e-01
-6.75701857e-01 -1.11925495e+00 -4.01049942e-01 1.62187815e+00
3.70272994e-02 -3.91173482e-01 -5.23327947e-01 -1.06008172e+00
-2.12253124e-01 5.93399465e-01 -6.34611785e-01 2.66699493e-01
-8.71572554e-01 -9.21361864e-01 5.86728275e-01 6.95641339e-01
1.28902256e-01 -1.05346143e+00 5.23543358e-02 1.75315768e-01
-4.43095177e-01 -1.33620679e+00 -4.03243184e-01 5.04648566e-01
-7.36325920e-01 -9.65868711e-01 -6.92093730e-01 -1.27580333e+00
4.99211431e-01 4.06502813e-01 1.53065455e+00 -2.93653160e-01
1.29541442e-01 4.84960735e-01 -4.69703048e-01 -1.60078943e-01
-5.16032398e-01 5.14129519e-01 2.01550633e-01 -5.70336320e-02
6.59561694e-01 -6.23279333e-01 -1.24472519e-02 1.23449005e-01
-7.06263065e-01 -2.00347379e-01 8.66676688e-01 9.33885276e-01
8.44734132e-01 -5.69896102e-01 9.02666926e-01 -1.19876754e+00
6.99279964e-01 -6.90617919e-01 -3.06808323e-01 5.89616537e-01
-5.19149840e-01 4.73744243e-01 5.34828246e-01 -3.27006936e-01
-1.03287947e+00 -2.99988128e-02 -7.04348320e-03 -1.34626582e-01
-2.08459646e-01 7.33380616e-01 -3.34153354e-01 1.63154099e-02
4.57061470e-01 2.45832235e-01 -3.45940530e-01 -7.31012821e-01
8.67366433e-01 6.63135290e-01 6.46406710e-01 -1.12228942e+00
6.81964278e-01 2.28439987e-01 -5.36111891e-01 -5.42522609e-01
-1.13632548e+00 -4.45348144e-01 -1.11049175e+00 2.58249253e-01
8.61751676e-01 -1.29181874e+00 6.01479635e-02 3.64850499e-02
-1.17104959e+00 -1.19491316e-01 -2.68385988e-02 6.21070504e-01
-5.89827120e-01 3.23637486e-01 -5.72173893e-01 -2.91384727e-01
-2.21557692e-01 -1.41138935e+00 9.74504888e-01 1.06714116e-02
-2.12438881e-01 -1.47663915e+00 6.36689425e-01 3.21869105e-01
2.48536035e-01 8.99854749e-02 1.07472479e+00 -1.03856647e+00
-2.90594131e-01 -2.23242329e-03 -2.28382424e-01 5.18421710e-01
4.33145106e-01 -4.98337388e-01 -1.07784843e+00 -5.75071275e-01
-2.08664805e-01 -8.54986191e-01 8.10489893e-01 7.05453753e-02
6.29688680e-01 3.10672726e-02 -3.61769676e-01 6.83102608e-01
1.50972521e+00 -1.01412870e-01 3.57069224e-01 5.57350039e-01
7.12617040e-01 5.98678946e-01 4.17285472e-01 -3.03051382e-01
8.17429125e-01 8.54046226e-01 -4.71532196e-02 -3.43297929e-01
-3.57295781e-01 -2.61656612e-01 6.68292761e-01 1.75719750e+00
-2.75009900e-01 1.12605326e-01 -8.56735170e-01 8.36471617e-01
-1.89064658e+00 -6.66275918e-01 3.57165337e-01 2.24453640e+00
1.17112124e+00 -1.95380569e-01 -3.91133055e-02 -2.90667236e-01
7.27225184e-01 2.04577744e-01 -2.05943406e-01 -5.35878658e-01
-4.74902451e-01 6.97879851e-01 4.38235044e-01 5.64674854e-01
-1.40511620e+00 1.42523587e+00 6.70122194e+00 7.52003729e-01
-9.91594195e-01 6.69563591e-01 3.56197506e-01 3.23447824e-01
-3.27903390e-01 5.81116602e-02 -1.01775742e+00 -8.92555118e-02
9.21119094e-01 -4.21734780e-01 4.75056618e-01 7.72460938e-01
-5.55854619e-01 4.55770284e-01 -1.35205948e+00 8.28411043e-01
4.88094032e-01 -9.96299922e-01 3.10274690e-01 1.34181958e-02
1.22252369e+00 4.93222862e-01 4.19971123e-02 8.10050547e-01
7.22154200e-01 -1.12385118e+00 5.69378674e-01 8.07347149e-02
9.68872190e-01 -8.36756527e-01 7.93079555e-01 5.65262400e-02
-1.50974798e+00 3.33358228e-01 -3.33787829e-01 1.09536968e-01
7.38523081e-02 9.20206681e-02 -4.90780860e-01 1.08146381e+00
5.56963563e-01 9.60090578e-01 -6.12883270e-01 4.46162105e-01
-6.42737865e-01 5.29276907e-01 -2.77149789e-02 3.45850915e-01
3.59330028e-01 -2.75291443e-01 2.42329061e-01 1.67957890e+00
1.63756236e-01 -6.90458059e-01 4.06870604e-01 6.87836111e-01
-3.53785068e-01 6.55073881e-01 -9.72933471e-01 -3.34209315e-02
4.83545244e-01 1.12409711e+00 -3.94138038e-01 -3.18050563e-01
-9.53910351e-01 8.14948201e-01 9.32338834e-01 4.82945621e-01
-7.03886092e-01 -3.33356678e-01 7.01778948e-01 -4.83590424e-01
3.82379562e-01 -2.79484242e-01 -1.14866428e-01 -1.54312980e+00
-7.16120005e-02 -1.13301730e+00 5.12922525e-01 -2.38438278e-01
-1.52139175e+00 1.04374945e+00 1.15965933e-01 -1.32447338e+00
-6.47196054e-01 -7.56377280e-01 -3.35706234e-01 1.10616744e+00
-2.22616673e+00 -1.68934298e+00 2.30337888e-01 7.44401097e-01
6.34388149e-01 -4.53966439e-01 1.30835617e+00 3.67782414e-01
-4.73548681e-01 9.27741289e-01 4.19511229e-01 3.59395832e-01
1.14942503e+00 -1.21157920e+00 4.12575036e-01 8.39073956e-01
7.76988804e-01 8.85116994e-01 1.83542714e-01 -3.24673563e-01
-1.09427464e+00 -1.11236751e+00 1.46791875e+00 -6.93411291e-01
8.66801083e-01 -3.59992564e-01 -7.85783291e-01 1.35870337e+00
6.67486846e-01 -1.96852073e-01 1.04594874e+00 7.86426604e-01
-8.92962575e-01 9.73499119e-02 -5.12140751e-01 6.09362423e-01
7.90396750e-01 -8.11032832e-01 -1.03141701e+00 4.95838851e-01
5.60287416e-01 -1.19906351e-01 -1.03699636e+00 5.44550180e-01
4.65057462e-01 -9.07962441e-01 9.67354417e-01 -7.69805133e-01
2.45643571e-01 -7.67232552e-02 -5.30799508e-01 -1.53232169e+00
-1.94242135e-01 -4.68703181e-01 2.48707846e-01 1.48932481e+00
6.61443233e-01 -7.52623141e-01 -1.00835197e-01 -4.29360956e-01
-2.45362684e-01 -6.62968397e-01 -9.21821773e-01 -1.01183903e+00
9.29846883e-01 -3.54522496e-01 4.92856234e-01 1.40632641e+00
1.30841404e-01 8.25214744e-01 -4.31303501e-01 7.48472139e-02
5.33945978e-01 3.71784568e-01 6.96317255e-01 -1.12845850e+00
-2.93388218e-01 -5.30314863e-01 -1.29818276e-01 -1.25227034e+00
8.47254694e-01 -1.64854395e+00 4.81175371e-02 -1.25759149e+00
3.64874065e-01 -4.72503304e-01 -7.33495116e-01 6.55044138e-01
-3.18257213e-01 2.24069506e-01 7.61541352e-02 4.54580128e-01
-6.41610444e-01 4.90250051e-01 8.00760925e-01 -1.23105399e-01
8.53157043e-02 -4.14585352e-01 -9.51876938e-01 7.21241534e-01
6.98659241e-01 -5.36229193e-01 -3.13277304e-01 -8.02072227e-01
9.77130905e-02 -2.90672153e-01 -3.37225735e-01 -8.15063894e-01
-1.34541705e-01 8.71535391e-02 1.14572987e-01 -2.58407980e-01
3.00301617e-04 -3.35061103e-01 -4.09492552e-01 1.55263888e-02
-3.04887176e-01 2.56051183e-01 3.35231990e-01 3.97708207e-01
-6.17820382e-01 -9.70123187e-02 7.74326205e-01 -1.57662019e-01
-7.37141788e-01 -4.52719219e-02 -7.95441791e-02 4.67401206e-01
6.40944600e-01 2.93702811e-01 -2.99754471e-01 -1.98602811e-01
-5.43225050e-01 1.30042061e-01 2.44143009e-01 9.01910722e-01
-4.90398854e-02 -1.66616464e+00 -9.78477299e-01 2.20189199e-01
4.56079364e-01 -5.55937946e-01 -1.82156980e-01 8.31553280e-01
-1.15918562e-01 8.46857429e-01 -2.91382998e-01 -6.46157742e-01
-8.25146675e-01 5.86682677e-01 1.97039589e-01 -8.52358043e-01
-3.43654454e-01 5.61135769e-01 5.77573597e-01 -1.18289208e+00
7.80112296e-02 -8.77633393e-02 -4.14824009e-01 6.97773471e-02
3.79685283e-01 -1.23058029e-01 1.93982616e-01 -1.29456997e+00
-3.88011426e-01 9.08045232e-01 -4.35301095e-01 -1.24337316e-01
1.19583762e+00 -2.84088701e-01 -1.01798154e-01 8.80644381e-01
1.19364417e+00 2.30736837e-01 -7.80465841e-01 -7.19395518e-01
1.27678901e-01 7.90293962e-02 -1.38755426e-01 -7.71737456e-01
-9.36594725e-01 1.05167627e+00 3.67701828e-01 -2.32143864e-01
9.22988594e-01 2.75444567e-01 6.33783162e-01 4.19967860e-01
6.05693102e-01 -9.21300590e-01 -2.14910999e-01 8.32539856e-01
8.31326365e-01 -1.29994440e+00 -3.92586105e-02 -2.82136887e-01
-6.49735570e-01 1.05634832e+00 4.47372019e-01 -6.64679930e-02
4.37550575e-01 1.41214579e-01 4.26106483e-01 3.16770114e-02
-5.56646943e-01 -5.12736559e-01 4.60256636e-01 7.16587842e-01
1.16367459e+00 2.35311091e-01 -5.04947543e-01 7.56631434e-01
-2.57444263e-01 -3.69904101e-01 4.91062962e-02 9.80167985e-01
-5.24335466e-02 -1.64677167e+00 -1.99955612e-01 -2.60799646e-01
-5.88656247e-01 -4.29660887e-01 -3.27293634e-01 1.19834554e+00
1.73516512e-01 7.38419116e-01 -7.68053532e-02 -1.29029959e-01
1.52950659e-01 5.66148102e-01 7.05699742e-01 -6.70667589e-01
-4.96311814e-01 2.11836547e-01 2.93537136e-02 -3.08501899e-01
-8.18373263e-01 -7.50935674e-01 -7.95422614e-01 2.30516478e-01
-2.50572413e-01 2.07196698e-01 6.93663597e-01 1.27324700e+00
2.02498525e-01 4.03647900e-01 5.23716092e-01 -9.53161240e-01
-5.30072927e-01 -1.16888809e+00 -3.04941684e-01 4.84775484e-01
2.17492394e-02 -8.18039715e-01 -3.16490769e-01 2.22351477e-01]
|
[11.040057182312012, 9.933225631713867]
|
69deff87-10f5-41f0-a09c-7d4e471fca3a
|
odoviz-a-3d-odometry-visualization-and
|
2107.07557
| null |
https://arxiv.org/abs/2107.07557v1
|
https://arxiv.org/pdf/2107.07557v1.pdf
|
OdoViz: A 3D Odometry Visualization and Processing Tool
|
OdoViz is a reactive web-based tool for 3D visualization and processing of autonomous vehicle datasets designed to support common tasks in visual place recognition research. The system includes functionality for loading, inspecting, visualizing, and processing GPS/INS poses, point clouds and camera images. It supports a number of commonly used driving datasets and can be adapted to load custom datasets with minimal effort. OdoViz's design consists of a slim server to serve the datasets coupled with a rich client frontend. This design supports multiple deployment configurations including single user stand-alone installations, research group installations serving datasets internally across a lab, or publicly accessible web-frontends for providing online interfaces for exploring and interacting with datasets. The tool allows viewing complete vehicle trajectories traversed at multiple different time periods simultaneously, facilitating tasks such as sub-sampling, comparing and finding pose correspondences both across and within sequences. This significantly reduces the effort required in creating subsets of data from existing datasets for machine learning tasks. Further to the above, the system also supports adding custom extensions and plugins to extend the capabilities of the software for other potential data management, visualization and processing tasks. The platform has been open-sourced to promote its use and encourage further contributions from the research community.
|
['John McDonald', 'Saravanabalagi Ramachandran']
|
2021-07-15
| null | null | null | null |
['visual-place-recognition']
|
['computer-vision']
|
[-4.02630001e-01 -4.08338249e-01 7.64494389e-02 -6.27425730e-01
-5.59358954e-01 -8.84227455e-01 5.59095442e-01 3.63552481e-01
-3.23739260e-01 7.74172023e-02 -4.49488573e-02 -8.87582541e-01
1.71451852e-01 -7.94537246e-01 -4.14635986e-01 -4.70777333e-01
-1.50807396e-01 5.20199239e-01 7.93201208e-01 -4.07565117e-01
4.35173124e-01 1.26346731e+00 -2.42680478e+00 -3.87434438e-02
2.54745752e-01 6.13699853e-01 5.65559804e-01 7.20280588e-01
-1.23089865e-01 1.83564395e-01 -5.49945533e-01 2.26464540e-01
5.67642093e-01 3.86214763e-01 -1.41189694e-01 -1.26321048e-01
5.92888474e-01 -3.71871084e-01 -1.18392073e-01 5.54395974e-01
6.49858296e-01 3.18194568e-01 -5.41822100e-03 -1.95194185e+00
3.48915011e-02 -5.89840591e-01 -3.69791597e-01 4.21582162e-01
5.72908640e-01 5.79628050e-01 3.07623684e-01 -8.41113210e-01
6.20356560e-01 9.76864457e-01 7.07804859e-01 2.21708901e-02
-1.02356136e+00 -9.71508861e-01 -3.82405743e-02 4.25651520e-01
-1.33921862e+00 -7.30075359e-01 6.30526781e-01 -7.75760949e-01
1.34259701e+00 6.77017868e-01 5.92252374e-01 7.00584710e-01
-1.80446908e-01 5.15236139e-01 6.06630981e-01 6.04880862e-02
3.39962721e-01 4.50000882e-01 4.20087546e-01 5.45939803e-01
3.12988400e-01 2.78235879e-02 -4.68862981e-01 -4.03430432e-01
4.42301273e-01 2.22799197e-01 1.28179297e-01 -7.99477041e-01
-1.09545040e+00 5.69369197e-01 1.06528930e-01 -1.04913786e-01
-2.56698310e-01 -2.47563228e-01 5.29807508e-01 3.19205940e-01
3.28261286e-01 6.52090609e-02 -5.60956895e-01 -5.62331259e-01
-7.19970345e-01 6.52130663e-01 5.06611884e-01 1.40097296e+00
1.10347295e+00 -5.14094047e-02 4.05703038e-01 5.63389778e-01
5.58603942e-01 4.93704259e-01 2.83257037e-01 -1.04573858e+00
4.26307708e-01 7.44841576e-01 3.36223513e-01 -1.05203736e+00
-8.08731616e-01 2.58246660e-01 4.93326448e-02 9.75563645e-01
2.88691640e-01 -4.20469306e-02 -6.65186465e-01 1.11768281e+00
8.38491380e-01 -6.27018064e-02 -2.23332763e-01 8.21686983e-01
1.05268049e+00 6.22048676e-01 -7.88262289e-04 3.32799017e-01
1.31973827e+00 -3.59245569e-01 -4.81075019e-01 -3.99894565e-01
8.08685184e-01 -8.46033692e-01 1.06146038e+00 -4.48552892e-02
-5.93527496e-01 -5.62032640e-01 -1.36930430e+00 -3.12995464e-01
-1.11044133e+00 -4.32850234e-02 3.62441510e-01 5.70386827e-01
-1.30720580e+00 3.64314228e-01 -1.06829023e+00 -8.33691061e-01
2.36838862e-01 2.21134797e-01 -5.73418140e-01 1.68425236e-02
-6.62282705e-01 1.18292689e+00 1.55222610e-01 -1.77814141e-01
-3.95260572e-01 -7.83493519e-01 -1.05893195e+00 -3.66911948e-01
3.69011685e-02 -2.92079300e-01 1.10860753e+00 -9.38455537e-02
-1.07449961e+00 7.54385293e-01 -4.34757441e-01 -1.10668555e-01
5.56541681e-01 -4.82190177e-02 -5.55383801e-01 -2.44327933e-01
5.60271859e-01 7.24485397e-01 2.26481095e-01 -1.01855850e+00
-8.22067142e-01 -6.85697317e-01 -3.10941875e-01 2.53881425e-01
1.69556364e-01 2.38173679e-01 -7.98717320e-01 2.45645232e-02
4.41885330e-02 -1.02110732e+00 -1.04344308e-01 1.73866197e-01
5.62546626e-02 -8.68321434e-02 1.70221996e+00 -7.44936466e-01
6.90699160e-01 -2.50803041e+00 -7.79051244e-01 4.66590315e-01
-9.64653865e-02 1.88982263e-01 1.75547615e-01 8.80750835e-01
-1.74559176e-01 -1.45771489e-01 4.11692232e-01 -2.91183650e-01
1.99190810e-01 1.89104989e-01 -1.02393469e-02 6.88638926e-01
-9.00963843e-02 4.00527060e-01 -7.59236395e-01 -2.89493680e-01
9.10222411e-01 3.50731432e-01 -1.08580574e-01 -7.68831968e-02
9.99680310e-02 4.60120887e-01 -7.68565685e-02 4.98643517e-01
1.03412640e+00 6.86232448e-02 2.90060453e-02 2.16092005e-01
-8.09254587e-01 4.90517437e-01 -1.56795919e+00 1.44171548e+00
-2.68190742e-01 1.37028182e+00 1.44011632e-01 -5.50100982e-01
1.09009528e+00 -3.97532508e-02 3.70498896e-01 -5.81757426e-01
-8.35358799e-02 9.13336948e-02 -6.26362979e-01 -8.77758741e-01
9.73563135e-01 6.33101761e-01 1.91238612e-01 3.73319983e-01
-3.92264575e-01 2.90659424e-02 2.69769520e-01 3.15807685e-02
1.08567822e+00 2.76499301e-01 8.53004381e-02 -1.78382576e-01
1.12296604e-01 5.41401565e-01 2.24227428e-01 1.89492434e-01
-2.39238560e-01 2.02728808e-01 1.30183518e-01 -5.09159684e-01
-1.39894879e+00 -9.82463658e-01 -3.13529849e-01 1.32717311e+00
3.99281859e-01 -5.52323401e-01 -4.68724929e-02 -1.69693679e-01
4.92318600e-01 9.04315293e-01 -2.52833426e-01 3.46498638e-01
-3.20515603e-01 -8.36320892e-02 3.49397212e-01 6.24452710e-01
3.65789324e-01 -7.40610659e-01 -1.15462518e+00 -1.31219223e-01
8.22658092e-02 -8.33999932e-01 -1.14002585e-01 -2.80789007e-02
-7.18567252e-01 -9.57514703e-01 -2.37024367e-01 -6.63652122e-01
5.85236669e-01 9.14734900e-01 6.56037092e-01 -1.31882578e-01
-4.03481036e-01 5.41519821e-01 -1.07752681e-01 -7.97237277e-01
-9.17025879e-02 -2.26647049e-01 -3.41822542e-02 -4.44513857e-01
7.30987489e-01 -5.38794816e-01 -3.37206066e-01 6.95921659e-01
-5.08129001e-01 2.44693562e-01 7.24815484e-03 -5.58693707e-02
4.34404194e-01 -2.54578352e-01 3.21182609e-01 -1.98455900e-01
4.42835182e-01 -7.99082100e-01 -1.10048211e+00 -3.85044038e-01
-6.37991488e-01 -3.68365645e-01 1.55725509e-01 -5.61406575e-02
-6.90353453e-01 1.90923408e-01 -1.52250350e-01 -5.69430590e-01
-7.82192528e-01 3.58529001e-01 -2.69273669e-01 -1.30346209e-01
5.63443303e-01 3.96943539e-02 5.67290246e-01 -6.01298511e-01
3.46133113e-01 1.05528641e+00 5.62408864e-01 8.41714665e-02
7.74795830e-01 5.34657300e-01 -2.55978525e-01 -1.20271909e+00
1.59804344e-01 -8.89631689e-01 -6.60151303e-01 -5.32351136e-01
7.27961898e-01 -8.33720624e-01 -9.84613717e-01 2.65007585e-01
-7.29776919e-01 -3.45964313e-01 -5.29086404e-02 4.49364811e-01
-3.46138000e-01 8.30442011e-02 1.74605012e-01 -6.87381089e-01
2.16136202e-01 -1.17342651e+00 1.10062921e+00 4.43466783e-01
-3.35102320e-01 -8.01331103e-01 1.63785785e-01 3.40985596e-01
6.32413328e-01 6.07210994e-01 3.19837302e-01 -6.53788030e-01
-7.53072262e-01 -6.87154949e-01 -2.10848227e-01 -2.68009633e-01
-7.43130222e-02 2.02148378e-01 -9.74045932e-01 -3.58936459e-01
-6.98464990e-01 1.34989351e-01 1.55341625e-01 -2.73313299e-02
7.59377778e-01 -5.84271215e-02 -8.46346200e-01 4.01642293e-01
1.12950563e+00 5.84588647e-01 5.80087006e-01 9.29758608e-01
5.71417451e-01 8.46905410e-01 7.98227251e-01 4.17194903e-01
8.35574865e-01 9.87820804e-01 5.76178551e-01 -2.60716647e-01
3.07623088e-01 -1.35418490e-01 1.00211747e-01 1.75243333e-01
9.58159193e-02 2.69491225e-01 -1.12471235e+00 6.28222048e-01
-2.02299142e+00 -1.24943733e+00 -6.37496054e-01 2.44738412e+00
-1.12141278e-02 7.08637433e-03 4.48179066e-01 -3.34543549e-02
5.15186548e-01 1.48458824e-01 -6.87228024e-01 -4.58417863e-01
4.79996234e-01 -3.62689734e-01 1.01560009e+00 4.41840440e-01
-1.00967717e+00 5.10861039e-01 6.26338148e+00 2.28809312e-01
-1.39949846e+00 1.03463300e-01 1.56868575e-03 -3.34605873e-01
-1.27117550e-02 -6.15849905e-02 -8.78512084e-01 3.73763502e-01
1.05515230e+00 -5.41305900e-01 2.99211383e-01 1.29803598e+00
8.10010731e-01 -2.74255425e-01 -9.83209729e-01 9.80502605e-01
-9.38207582e-02 -1.59336925e+00 -6.63178861e-01 4.55043316e-01
8.66371840e-02 9.55340087e-01 -2.30890825e-01 5.21517061e-02
3.77137661e-01 -5.68540215e-01 9.26386535e-01 3.98871392e-01
7.17018306e-01 -8.11575413e-01 4.74849254e-01 5.17677605e-01
-1.34727395e+00 9.32562277e-02 -4.11524028e-02 -2.00071827e-01
2.50846088e-01 1.57269165e-02 -1.21006715e+00 4.15134996e-01
1.13159728e+00 5.29968023e-01 -9.60698843e-01 1.40407789e+00
5.44111907e-01 7.23214671e-02 -7.17847288e-01 -1.19967684e-01
-6.94733858e-02 -1.95595115e-01 6.07830107e-01 1.45332992e+00
3.39912444e-01 -4.15513247e-01 2.80306429e-01 4.58803594e-01
8.27811956e-01 -7.58290291e-02 -1.25170910e+00 4.24148619e-01
9.40885901e-01 1.45590580e+00 -7.94541419e-01 -1.90348700e-01
-3.72448087e-01 3.97608191e-01 -1.33342296e-01 3.64070058e-01
-7.85350323e-01 -1.00543749e+00 1.36007988e+00 4.68253940e-01
2.85047889e-01 -1.03563786e+00 -4.16715801e-01 -4.01436538e-01
7.85495713e-02 -5.37096679e-01 1.29871115e-01 -1.16913557e+00
-5.63054085e-01 2.10285261e-01 5.12905836e-01 -1.66700077e+00
-4.17873353e-01 -5.28981805e-01 -6.69878721e-01 1.21199048e+00
-1.14749849e+00 -7.79804647e-01 -8.67798209e-01 6.91583157e-01
3.41481566e-01 -1.33310929e-01 7.08965540e-01 6.43403411e-01
-6.01332068e-01 9.89938453e-02 1.79844797e-01 -7.51765668e-02
5.29814661e-01 -1.02004206e+00 8.72025430e-01 8.43991756e-01
-1.70926347e-01 7.83366740e-01 8.74159038e-01 -7.02719986e-01
-1.69156659e+00 -1.05691838e+00 7.48681128e-01 -5.31830072e-01
5.41915953e-01 -4.26523417e-01 -7.63292730e-01 1.03545177e+00
1.90023601e-01 -1.07787408e-01 8.54853272e-01 -1.12009428e-01
-8.79704058e-02 -2.06742749e-01 -1.23268878e+00 6.08307064e-01
7.66825020e-01 -4.19930071e-01 -1.69131830e-01 3.60057443e-01
1.89332664e-01 -7.76140213e-01 -5.72177052e-01 -9.60005261e-03
4.91471261e-01 -9.98764932e-01 8.61341417e-01 -1.14954360e-01
-2.47595519e-01 -9.58975613e-01 8.50347616e-03 -1.00480711e+00
-3.73891562e-01 -5.70288777e-01 4.27015841e-01 1.15691686e+00
4.33708012e-01 -1.06134343e+00 6.41186774e-01 8.97655606e-01
-6.10583365e-01 -6.53459430e-02 -9.37898755e-01 -6.35133564e-01
-7.89158940e-01 -1.15009952e+00 9.60570157e-01 7.80871689e-01
2.69424558e-01 3.32492180e-02 2.15616524e-01 7.28272080e-01
3.90674114e-01 -6.36263192e-02 1.64543939e+00 -1.32349396e+00
2.56533772e-01 -4.26877201e-01 -9.70102727e-01 -8.61194968e-01
-3.90739679e-01 -9.32280898e-01 -1.93200499e-01 -1.90263832e+00
-6.27912581e-01 -7.51277685e-01 5.17639697e-01 7.55793393e-01
4.31003153e-01 1.92899466e-01 2.37328127e-01 5.13991475e-01
-2.83642560e-01 -1.40971392e-01 4.74401921e-01 1.71168551e-01
-3.50762755e-01 1.56608805e-01 -5.16824305e-01 5.58132172e-01
8.35045099e-01 -3.04145604e-01 -5.35312891e-01 -2.74975896e-01
3.90834659e-02 -3.45162272e-01 6.70303762e-01 -1.27591109e+00
3.20083618e-01 -2.95097947e-01 5.99514723e-01 -1.37403536e+00
3.93232703e-01 -1.13583875e+00 6.90068364e-01 7.62282088e-02
1.33236483e-01 7.23832846e-01 8.27685118e-01 2.49105081e-01
1.79989293e-01 1.40546694e-01 3.60771656e-01 -4.59812917e-02
-1.09644985e+00 6.41792640e-02 -7.44718134e-01 -3.69178295e-01
1.71545243e+00 -7.79028058e-01 -6.20461285e-01 -3.12183708e-01
-5.77821612e-01 6.82952523e-01 9.91231382e-01 9.67950225e-01
3.36176217e-01 -1.32830000e+00 -9.50054675e-02 7.09473908e-01
6.03546262e-01 -1.68170093e-03 1.20341055e-01 5.92347503e-01
-1.05644870e+00 4.46663409e-01 -5.39249659e-01 -8.65458310e-01
-1.67591667e+00 5.57732046e-01 1.13637894e-01 5.47339678e-01
-1.20665896e+00 3.56515139e-01 -4.53145355e-01 -7.04014421e-01
2.18964636e-01 -2.81050831e-01 -2.96785891e-01 1.57825753e-01
9.62072611e-01 7.69581914e-01 5.97970665e-01 -8.73140097e-01
-6.96643412e-01 1.56826988e-01 3.38612825e-01 -3.31302226e-01
1.32291937e+00 -3.74453187e-01 3.66935104e-01 8.12334478e-01
1.25480664e+00 -7.55841145e-04 -1.44353962e+00 3.23336273e-01
1.04349051e-02 -5.32725990e-01 -7.21447635e-04 -4.14703071e-01
-6.93550289e-01 5.71305513e-01 1.03832996e+00 5.99916697e-01
6.67974770e-01 1.16752349e-01 3.20635736e-01 2.17466265e-01
6.07007921e-01 -9.91726577e-01 -6.79521203e-01 3.61380339e-01
7.83345222e-01 -9.82172370e-01 -2.16197968e-02 -5.49691170e-02
-7.75753379e-01 1.08677888e+00 4.85391647e-01 1.80939972e-01
7.16576934e-01 6.15571499e-01 6.18741095e-01 -4.22839642e-01
-7.49888718e-01 -3.42315584e-01 7.91630894e-03 1.04428566e+00
2.50187337e-01 2.39105392e-02 -4.54875566e-02 -2.90771574e-01
-5.76502562e-01 1.14920162e-01 4.71906662e-01 1.42501724e+00
-5.37865400e-01 -8.19885373e-01 -8.07713985e-01 4.03923541e-01
1.66248068e-01 3.90511334e-01 -1.12604029e-01 7.45992005e-01
1.59689665e-01 9.76540446e-01 5.69556117e-01 -4.93670374e-01
4.69675303e-01 1.31741926e-01 -1.74071833e-01 -2.56484061e-01
-4.14827406e-01 -2.55054086e-01 3.85737032e-01 -8.29640448e-01
2.26183366e-02 -1.12655222e+00 -1.32122660e+00 -6.28547132e-01
1.06910385e-01 -1.02583475e-01 1.48336196e+00 5.48419237e-01
1.15704679e+00 1.33697420e-01 4.70533282e-01 -1.61472869e+00
7.66658559e-02 -7.50417531e-01 -4.02257562e-01 1.86206833e-01
3.77947748e-01 -8.31944585e-01 -8.81286897e-03 2.79591177e-02]
|
[7.677425384521484, -1.5377899408340454]
|
d4d45b6c-760e-4472-aa3f-2e3fb34263ad
|
unsupervised-text-style-transfer-via
|
1901.11333
| null |
https://arxiv.org/abs/1901.11333v4
|
https://arxiv.org/pdf/1901.11333v4.pdf
|
IMaT: Unsupervised Text Attribute Transfer via Iterative Matching and Translation
|
Text attribute transfer aims to automatically rewrite sentences such that they possess certain linguistic attributes, while simultaneously preserving their semantic content. This task remains challenging due to a lack of supervised parallel data. Existing approaches try to explicitly disentangle content and attribute information, but this is difficult and often results in poor content-preservation and ungrammaticality. In contrast, we propose a simpler approach, Iterative Matching and Translation (IMaT), which: (1) constructs a pseudo-parallel corpus by aligning a subset of semantically similar sentences from the source and the target corpora; (2) applies a standard sequence-to-sequence model to learn the attribute transfer; (3) iteratively improves the learned transfer function by refining imperfections in the alignment. In sentiment modification and formality transfer tasks, our method outperforms complex state-of-the-art systems by a large margin. As an auxiliary contribution, we produce a publicly-available test set with human-generated transfer references.
|
['Jonas Mueller', 'Enrico Santus', 'Zhijing Jin', 'Nicholas Matthews', 'Di Jin']
|
2019-01-31
|
imat-unsupervised-text-attribute-transfer-via
|
https://aclanthology.org/D19-1306
|
https://aclanthology.org/D19-1306.pdf
|
ijcnlp-2019-11
|
['text-attribute-transfer']
|
['natural-language-processing']
|
[ 7.93587208e-01 2.92328417e-01 -2.06236780e-01 -5.96224010e-01
-1.21939862e+00 -9.82647955e-01 7.69812107e-01 2.73700923e-01
-4.81554389e-01 9.79077339e-01 4.37561959e-01 -1.68833986e-01
2.53151417e-01 -6.17487788e-01 -8.47010553e-01 -2.73938507e-01
5.97673714e-01 1.00109339e+00 4.00857106e-02 -4.94656205e-01
2.74416268e-01 1.29687656e-02 -1.39769685e+00 5.11470616e-01
1.29250586e+00 5.90672135e-01 2.36378103e-01 3.20453912e-01
-3.51023763e-01 6.12254918e-01 -3.88730258e-01 -1.05117834e+00
2.53279328e-01 -1.09373295e+00 -1.17593479e+00 -1.78915873e-01
6.34959579e-01 2.27654446e-02 2.66930640e-01 1.31581831e+00
4.04247910e-01 4.51946296e-02 7.27609456e-01 -1.06353056e+00
-1.00309181e+00 8.18852246e-01 -2.55722672e-01 -1.24323770e-01
3.57298315e-01 -1.99441358e-01 1.40194607e+00 -9.29691315e-01
8.41614723e-01 1.15074039e+00 5.20369947e-01 8.93958449e-01
-1.60565794e+00 -6.13833427e-01 2.17158850e-02 1.54222220e-01
-1.07878840e+00 -5.52532375e-01 8.88349712e-01 -1.66340008e-01
1.21678507e+00 2.77990013e-01 4.45346743e-01 1.17722166e+00
1.66207496e-02 6.18981838e-01 1.18997335e+00 -7.71647990e-01
-5.29612266e-02 1.82093769e-01 -1.92657545e-01 5.81671476e-01
-1.34154528e-01 -5.48066609e-02 -5.90142071e-01 -6.93223439e-03
8.61581266e-02 -3.88586462e-01 -2.79712439e-01 -3.07132810e-01
-1.50191915e+00 7.55866230e-01 1.81082398e-01 3.68188828e-01
-2.08595231e-01 -1.54413313e-01 5.88625371e-01 8.63543689e-01
5.91844380e-01 8.75374675e-01 -7.18762457e-01 -1.81818455e-01
-7.31055617e-01 3.58931959e-01 9.65728462e-01 1.18102610e+00
1.05663610e+00 -3.56403440e-01 -1.68901488e-01 9.20278251e-01
-7.27509484e-02 4.75319773e-01 6.76669657e-01 -7.95816600e-01
7.61644185e-01 5.61923265e-01 3.98616400e-03 -5.90535104e-01
3.13561969e-02 -1.99071378e-01 -5.51261485e-01 -2.02047639e-02
3.29301238e-01 1.20978728e-01 -6.32090747e-01 2.18450737e+00
2.33403698e-01 -2.22851500e-01 3.27547729e-01 6.66605473e-01
6.11258864e-01 4.94684249e-01 2.64379621e-01 -9.37348604e-02
1.27822351e+00 -1.12267089e+00 -6.64525688e-01 -5.22348940e-01
8.50590289e-01 -9.84341443e-01 1.64733875e+00 -2.45047510e-02
-1.25414062e+00 -5.15041053e-01 -1.01227474e+00 -4.19068098e-01
-3.34230274e-01 -3.38776112e-02 2.41455585e-01 5.39715052e-01
-9.53714192e-01 7.67301202e-01 -5.46427786e-01 -3.97775799e-01
2.16220304e-01 3.21026415e-01 -6.51048362e-01 -7.84307569e-02
-1.29136491e+00 1.26412916e+00 4.46274012e-01 -4.14605230e-01
-3.73995095e-01 -1.00504029e+00 -9.90742445e-01 -1.64663233e-02
1.75472707e-01 -1.13002515e+00 1.36000717e+00 -1.69113064e+00
-1.81428123e+00 1.35449779e+00 -1.29473940e-01 -1.35812864e-01
4.44259256e-01 -4.99137729e-01 -1.23011492e-01 -2.51355380e-01
2.59455472e-01 8.24737370e-01 6.91464424e-01 -9.88889873e-01
-6.16845727e-01 -3.20246041e-01 7.91898817e-02 4.88308936e-01
-5.61112404e-01 3.12818110e-01 -1.82372198e-01 -9.10031617e-01
-2.53416926e-01 -9.72227752e-01 -2.29688883e-02 -2.73130506e-01
-3.24414670e-01 -1.20152563e-01 4.12544340e-01 -7.18452454e-01
9.87053216e-01 -1.96674860e+00 8.14990699e-01 -1.34076416e-01
-1.72356918e-01 2.48112783e-01 -4.11337614e-01 6.32114410e-01
-1.21115841e-01 9.09424424e-02 -6.12040877e-01 -5.84616005e-01
1.39129311e-01 2.75598109e-01 -5.49070656e-01 1.50195360e-01
2.88247317e-01 1.05166912e+00 -1.04505396e+00 -5.95754921e-01
-6.06672466e-02 1.44187823e-01 -8.41473341e-01 2.84076452e-01
-4.21838403e-01 6.10501826e-01 -1.99670911e-01 3.54715466e-01
2.47737050e-01 -3.76781262e-02 1.97282270e-01 -1.44257545e-01
1.63430735e-01 8.95489156e-01 -6.68910265e-01 2.01499152e+00
-6.76481068e-01 3.41128290e-01 -3.89701456e-01 -1.04705083e+00
9.40718830e-01 2.86250532e-01 2.22179264e-01 -8.47121418e-01
-1.99078713e-02 5.30545473e-01 -1.74374834e-01 -3.44051629e-01
6.68313384e-01 -7.41064668e-01 -4.54964995e-01 6.33029282e-01
2.77474046e-01 -3.43872458e-01 1.21499158e-01 2.84133274e-02
9.79348302e-01 6.07008934e-01 5.26932776e-01 -2.63457417e-01
7.63983130e-01 8.40113685e-02 6.79585814e-01 3.09292257e-01
7.74374679e-02 5.80201924e-01 4.34977114e-01 -2.36270159e-01
-1.32386100e+00 -1.05877233e+00 2.55775988e-01 1.31409097e+00
3.92209738e-02 -3.75739276e-01 -9.85287666e-01 -1.04944468e+00
-1.45368978e-01 1.18646669e+00 -7.42214501e-01 -3.69453013e-01
-8.78288269e-01 -3.91447961e-01 7.40416288e-01 5.27206898e-01
2.51468301e-01 -1.23439622e+00 -6.18945733e-02 2.06016779e-01
-7.37296224e-01 -1.00267088e+00 -7.87509501e-01 6.17806055e-03
-6.41975582e-01 -7.18865514e-01 -2.80421853e-01 -9.56336021e-01
8.16868305e-01 -2.05614224e-01 1.58717668e+00 1.72846392e-01
3.36175174e-01 -1.97157174e-01 -4.25160855e-01 -3.70390296e-01
-1.02625012e+00 4.86511528e-01 -1.18968159e-01 -1.03899635e-01
4.80204135e-01 -6.49799824e-01 -9.22302455e-02 2.06396788e-01
-9.03498411e-01 2.96268314e-01 5.95522702e-01 9.87754226e-01
8.01414311e-01 -5.70504785e-01 8.15338373e-01 -1.44369197e+00
5.63574135e-01 -3.80165316e-02 -5.38850605e-01 4.42923725e-01
-7.99696922e-01 3.33133221e-01 9.55826521e-01 -4.01291162e-01
-1.23880470e+00 1.79445386e-01 -2.29079261e-01 -1.40697002e-01
-1.22760475e-01 4.29740429e-01 -5.99074304e-01 3.16108823e-01
7.30346978e-01 1.69869065e-01 5.30426204e-02 -4.38830465e-01
7.65073359e-01 6.28545344e-01 7.36387312e-01 -8.41437459e-01
9.07102942e-01 2.35665843e-01 -2.62978971e-01 -3.41356874e-01
-1.25873113e+00 -1.82491735e-01 -9.56804633e-01 1.03992887e-01
7.30166674e-01 -6.73219562e-01 -7.09864199e-02 2.71082848e-01
-1.26360953e+00 -4.30093288e-01 -4.82513696e-01 3.40120703e-01
-8.92540276e-01 2.63333440e-01 -5.19698679e-01 1.35687618e-02
-6.35404527e-01 -8.71132970e-01 1.14781249e+00 -2.86094129e-01
-7.32117116e-01 -9.89593744e-01 4.08987075e-01 4.33007926e-01
3.82776678e-01 7.48687834e-02 1.35192072e+00 -9.24371004e-01
-1.91034138e-01 -4.39849682e-02 9.29841679e-03 6.18063092e-01
3.64713013e-01 -2.60064840e-01 -7.20860541e-01 -2.51834035e-01
-1.20902717e-01 -6.16273642e-01 6.41219139e-01 -2.39769697e-01
6.74590886e-01 -5.50850451e-01 -1.41314462e-01 6.35043681e-01
1.15758359e+00 -2.77542174e-01 7.52286971e-01 5.38028300e-01
6.88148201e-01 9.69129264e-01 6.58513427e-01 -1.06965199e-01
5.22937179e-01 7.74104655e-01 9.46703181e-02 -7.73583353e-02
-3.30619782e-01 -6.49208248e-01 5.42232990e-01 1.20840323e+00
1.77724481e-01 -1.71313688e-01 -5.92869699e-01 6.46348894e-01
-1.80343568e+00 -9.08773601e-01 1.73615471e-01 2.12463522e+00
1.45474041e+00 -1.05797369e-02 -1.62243135e-02 -8.81310701e-02
7.01945245e-01 -1.25913739e-01 -2.74637461e-01 -2.76522815e-01
-2.11861089e-01 3.80597681e-01 2.08934978e-01 6.80477560e-01
-9.21974778e-01 1.45072162e+00 5.68946457e+00 7.18008339e-01
-9.32331920e-01 2.53859609e-01 2.11165801e-01 4.94445860e-02
-9.11065578e-01 3.85758817e-01 -7.59975970e-01 4.31785971e-01
8.57306480e-01 -3.26119006e-01 4.93110657e-01 6.29575431e-01
-2.72627413e-01 5.46661139e-01 -1.56304371e+00 4.07807946e-01
4.49591339e-01 -1.13414383e+00 3.82997811e-01 -1.67526886e-01
7.59370744e-01 -7.14288503e-02 -7.14404359e-02 3.99119020e-01
4.29796696e-01 -9.49728131e-01 9.84007478e-01 4.61830676e-01
1.08628678e+00 -7.64392138e-01 6.55330241e-01 2.59851992e-01
-8.19954872e-01 2.86909312e-01 -2.26212129e-01 -3.66226882e-02
2.34268814e-01 3.27009678e-01 -6.10245228e-01 5.58979630e-01
4.51994032e-01 7.35294104e-01 -4.74458903e-01 4.24766183e-01
-8.74851942e-01 4.88095969e-01 5.42203933e-02 -1.39659852e-01
-1.75031535e-02 -3.17045093e-01 5.32421529e-01 1.23092711e+00
2.68177509e-01 -1.09378949e-01 4.10983898e-02 9.16750848e-01
-4.93597299e-01 4.65932041e-01 -5.91725588e-01 -6.41735345e-02
5.60029268e-01 1.19558632e+00 -3.62045765e-01 -4.11996067e-01
-6.48426116e-01 1.43571627e+00 8.73130500e-01 1.12662852e-01
-5.84793627e-01 -5.38584650e-01 6.98034942e-01 -4.13503647e-02
2.69203931e-01 1.09745219e-01 -2.68285215e-01 -1.32492304e+00
1.47376359e-01 -1.15582657e+00 4.42884833e-01 -6.72874272e-01
-1.53936100e+00 7.41498828e-01 -1.82567805e-01 -1.29104626e+00
-6.00240231e-01 -3.00308824e-01 -3.01173270e-01 9.93223846e-01
-1.56634641e+00 -1.58157313e+00 1.82059668e-02 4.89783943e-01
6.22673512e-01 -2.66017556e-01 1.16368139e+00 3.46536547e-01
-1.01403266e-01 8.63509536e-01 -1.09160520e-01 4.35924418e-02
1.18816924e+00 -1.38303590e+00 6.61902606e-01 8.13241661e-01
2.34314278e-01 6.36108637e-01 8.51408482e-01 -6.16855145e-01
-9.18471813e-01 -1.28686368e+00 1.64982915e+00 -7.62377679e-01
7.50909507e-01 -4.82967257e-01 -1.20650852e+00 9.97024357e-01
3.85834783e-01 -2.53305912e-01 8.78147185e-01 5.84886409e-02
-7.34839082e-01 3.70259881e-02 -1.07437146e+00 6.90251112e-01
1.18780303e+00 -6.90507174e-01 -1.15675128e+00 3.17183256e-01
9.50587928e-01 -3.78651381e-01 -8.22394133e-01 2.65702665e-01
3.57152730e-01 -5.94360769e-01 6.47368371e-01 -9.93565023e-01
8.08515191e-01 -2.40453854e-01 -2.49311090e-01 -1.75821805e+00
-3.06298584e-01 -7.79940307e-01 3.60467404e-01 1.60695386e+00
8.08599412e-01 -6.43326640e-01 4.98666346e-01 3.78352106e-01
-4.02769297e-01 -5.13226867e-01 -9.73703444e-01 -8.57090235e-01
5.01090825e-01 -1.24926910e-01 8.28210890e-01 1.17527819e+00
1.91540703e-01 9.74522650e-01 -2.28269428e-01 -2.25619465e-01
4.05240744e-01 3.55488867e-01 7.89803326e-01 -1.14752555e+00
-3.81413519e-01 -5.30201554e-01 -7.94329047e-02 -8.44653308e-01
8.52814674e-01 -1.44296908e+00 1.91438854e-01 -1.35507143e+00
3.62309009e-01 -3.80757779e-01 -8.37281346e-02 7.18865037e-01
-3.80094260e-01 2.93557793e-01 4.58938591e-02 3.44603866e-01
-4.97540683e-01 8.44567835e-01 1.08610749e+00 5.38432796e-04
-7.29637891e-02 -2.05582127e-01 -9.12406325e-01 6.21447504e-01
8.15738618e-01 -7.00559199e-01 -3.40555400e-01 -6.90422654e-01
5.03542781e-01 -1.19034864e-01 2.18058135e-02 -4.70400661e-01
-1.29583543e-02 -1.84372097e-01 3.69480103e-02 -2.09566817e-01
1.18086249e-01 -6.39644682e-01 -4.58637699e-02 2.70888031e-01
-6.86544001e-01 4.13706869e-01 1.34052992e-01 3.04693699e-01
-2.79494375e-01 -4.60239917e-01 7.79626131e-01 -6.15341999e-02
-3.72843176e-01 4.24650721e-02 -7.74814710e-02 5.10021508e-01
7.52500474e-01 5.27936779e-02 -5.09835124e-01 -1.47327989e-01
-2.79015988e-01 -7.71263316e-02 9.98999298e-01 6.48425221e-01
2.79729873e-01 -1.44278252e+00 -1.11472487e+00 1.55764148e-01
4.63408262e-01 -2.88923562e-01 -1.80647731e-01 5.92597187e-01
-2.21147895e-01 2.28324980e-02 -4.17236984e-01 -2.40551054e-01
-1.35235536e+00 5.05254924e-01 2.93376684e-01 -3.11872333e-01
-5.32179475e-01 6.91281915e-01 1.58337966e-01 -1.11963594e+00
-2.08748415e-01 1.53523088e-01 -3.21418121e-02 -8.02777410e-02
3.31723660e-01 8.15336853e-02 3.49732846e-01 -9.03650999e-01
-3.30915481e-01 5.96878886e-01 -3.57861608e-01 -5.31950176e-01
1.29318464e+00 -1.73663214e-01 -5.08253157e-01 1.89576149e-01
1.26842797e+00 2.22767636e-01 -9.26964641e-01 -6.32824540e-01
1.73062220e-01 -4.64191437e-01 -3.92257899e-01 -8.26109290e-01
-6.56340241e-01 7.00051069e-01 -5.07220551e-02 -1.94996625e-01
1.10921705e+00 2.38569841e-01 9.89426136e-01 4.76166248e-01
8.48899484e-02 -1.18492937e+00 2.34766901e-01 8.12091053e-01
9.99014974e-01 -9.23910022e-01 -3.14491570e-01 -5.54164469e-01
-7.49086440e-01 7.83992767e-01 6.87143505e-01 8.56340751e-02
1.62350595e-01 2.08215594e-01 1.87177256e-01 7.14658350e-02
-8.04499030e-01 -9.13336352e-02 3.69318634e-01 6.25012934e-01
7.19904840e-01 -3.37439403e-02 -4.66615260e-01 6.18909001e-01
-7.70694971e-01 -2.60172814e-01 2.79909581e-01 8.31766427e-01
-1.92281708e-01 -1.68996799e+00 -1.54290032e-02 2.62985021e-01
-5.22008181e-01 -4.38151002e-01 -7.08054662e-01 6.16004348e-01
-1.30120635e-01 6.62066400e-01 -2.75868662e-02 -2.82661080e-01
7.04773247e-01 2.47717291e-01 6.76521897e-01 -8.72500181e-01
-7.18216240e-01 -1.38687611e-01 3.65666032e-01 -3.82222623e-01
-3.18342090e-01 -8.79647076e-01 -1.23103464e+00 -8.85734186e-02
-1.49157643e-01 1.80147827e-01 5.98330021e-01 1.12921846e+00
3.71122599e-01 4.55391556e-01 3.64046067e-01 -5.59234202e-01
-8.41982603e-01 -1.08098567e+00 -4.97431457e-02 1.02159965e+00
8.24701637e-02 -3.71507436e-01 -2.52105802e-01 5.04991949e-01]
|
[11.575116157531738, 9.531254768371582]
|
2de1e46c-f346-46f8-8b2f-448b29670841
|
context-generation-improves-open-domain
|
2210.06349
| null |
https://arxiv.org/abs/2210.06349v2
|
https://arxiv.org/pdf/2210.06349v2.pdf
|
Context Generation Improves Open Domain Question Answering
|
Closed-book question answering (QA) requires a model to directly answer an open-domain question without access to any external knowledge. Prior work on closed-book QA either directly finetunes or prompts a pretrained language model (LM) to leverage the stored knowledge. However, they do not fully exploit the parameterized knowledge. To address this issue, we propose a two-stage, closed-book QA framework which employs a coarse-to-fine approach to extract relevant knowledge and answer a question. Our approach first generates a related context for a given question by prompting a pretrained LM. We then prompt the same LM for answer prediction using the generated context and the question. Additionally, to eliminate failure caused by context uncertainty, we marginalize over generated contexts. Experimental results on three QA benchmarks show that our method significantly outperforms previous closed-book QA methods (e.g. exact matching 68.6% vs. 55.3%), and is on par with open-book methods that exploit external knowledge sources (e.g. 68.6% vs. 68.0%). Our method is able to better exploit the stored knowledge in pretrained LMs without adding extra learnable parameters or needing finetuning, and paves the way for hybrid models that integrate pretrained LMs with external knowledge.
|
['Bryan Catanzaro', 'Anima Anandkumar', 'Pascale Fung', 'Mohammad Shoeybi', 'Ryan Prenger', 'Peng Xu', 'Shrimai Prabhumoye', 'Mostofa Patwary', 'Dan Su']
|
2022-10-12
| null | null | null | null |
['open-domain-question-answering']
|
['natural-language-processing']
|
[ 2.79862404e-01 5.07437408e-01 4.60489132e-02 -5.39882302e-01
-1.77547598e+00 -1.11930251e+00 5.59378982e-01 1.47444710e-01
-5.50875187e-01 7.72120357e-01 2.12374717e-01 -5.73931694e-01
-1.04004800e-01 -1.04254746e+00 -9.19451833e-01 -7.63967857e-02
6.32099748e-01 8.91122937e-01 8.77390206e-01 -5.51372111e-01
2.50165910e-01 -1.58796743e-01 -1.37974727e+00 6.73291922e-01
1.27750134e+00 1.10809970e+00 3.18540961e-01 7.81879961e-01
-5.71424663e-01 1.09677207e+00 -7.07639158e-01 -7.63289213e-01
-3.44540109e-03 -3.52114946e-01 -1.36911106e+00 -6.11069918e-01
7.72661209e-01 -2.03782037e-01 -2.83335727e-02 6.60348654e-01
2.73794532e-01 3.71496588e-01 4.43254113e-01 -6.80748343e-01
-8.58254433e-01 6.25145078e-01 1.26002833e-01 1.64192349e-01
7.79022694e-01 1.42915726e-01 1.27423549e+00 -8.46813262e-01
6.51398599e-01 1.25031924e+00 5.29477239e-01 6.56374991e-01
-1.09810269e+00 -2.76701182e-01 1.16898082e-01 3.98784846e-01
-1.13663399e+00 -4.68779504e-01 4.82263088e-01 -1.08095013e-01
1.18814361e+00 4.10471976e-01 -6.35089502e-02 7.46084392e-01
5.27867638e-02 6.33713782e-01 1.30479741e+00 -7.37904906e-01
1.20530851e-01 2.61824936e-01 5.37057221e-01 6.86580598e-01
-2.92128026e-01 -9.28113535e-02 -4.00049806e-01 -4.10654217e-01
6.44512698e-02 -5.31270564e-01 -3.00579339e-01 -9.40393060e-02
-9.32407558e-01 8.07196975e-01 3.17565680e-01 6.72587827e-02
-1.27641618e-01 -2.25652028e-02 2.56378859e-01 5.33726633e-01
-2.04837788e-02 8.96396577e-01 -8.06737006e-01 -2.12431774e-01
-9.78216171e-01 3.69509906e-01 1.30226481e+00 9.51952338e-01
1.13799024e+00 -5.21937132e-01 -5.83196044e-01 8.32938313e-01
2.28825837e-01 7.42501795e-01 3.25014830e-01 -1.50105023e+00
7.94434190e-01 8.62374961e-01 2.58732527e-01 -6.89150929e-01
-1.67581305e-01 -2.56793469e-01 -9.88611653e-02 -4.42801327e-01
6.18122935e-01 -1.20528489e-01 -8.07208300e-01 1.73009622e+00
3.55759054e-01 7.88286701e-02 3.43091995e-01 7.02473879e-01
1.11820662e+00 7.22212374e-01 6.47026896e-02 1.13840662e-01
1.54579365e+00 -1.32481432e+00 -6.34981394e-01 -4.85928535e-01
7.11411536e-01 -7.87252843e-01 1.67629385e+00 2.52729833e-01
-1.16819298e+00 -6.47847652e-01 -7.59214580e-01 -4.86503154e-01
-4.45183426e-01 -5.02003133e-02 9.63300467e-02 6.56409323e-01
-1.09319997e+00 1.57098234e-01 -2.27311715e-01 -1.68545887e-01
-9.20048391e-04 3.42613786e-01 -7.77897611e-02 -4.41266000e-01
-1.68094778e+00 1.15691352e+00 4.00115460e-01 -2.39537135e-01
-6.93867922e-01 -9.71404552e-01 -9.60479081e-01 7.64753297e-02
1.03111720e+00 -1.08479452e+00 1.79823923e+00 -7.61003375e-01
-1.81899953e+00 6.85038865e-01 -5.45916498e-01 -5.09548724e-01
1.39636457e-01 -5.07195473e-01 -3.18474442e-01 3.78349543e-01
1.37330577e-01 6.54083133e-01 7.89428651e-01 -1.20040214e+00
-5.59716463e-01 -7.67817423e-02 8.48888040e-01 1.29065365e-01
-8.36531967e-02 3.63862030e-02 -6.86468482e-01 -2.54741222e-01
-5.54094017e-02 -6.88300967e-01 -9.36408155e-03 -4.98598009e-01
-1.83416605e-02 -5.25154889e-01 5.58901310e-01 -7.41315007e-01
1.46778238e+00 -1.62692392e+00 -1.06517322e-01 2.45549992e-01
2.12053075e-01 4.19588238e-01 -4.55619961e-01 6.45679295e-01
4.40000743e-01 -1.13937639e-01 -4.08099055e-01 -1.94876283e-01
2.71317542e-01 4.62919503e-01 -7.40523219e-01 -4.62019503e-01
2.56896108e-01 1.28291726e+00 -1.08524454e+00 -6.45035446e-01
-1.72236279e-01 1.54698074e-01 -8.45857322e-01 4.83768761e-01
-9.69947577e-01 1.43813297e-01 -3.58724952e-01 4.65660572e-01
3.79698485e-01 -5.51555693e-01 1.88807592e-01 -1.15864828e-01
4.51232791e-01 8.47759664e-01 -1.11621559e+00 1.63091838e+00
-9.74933922e-01 1.52021721e-01 -1.96794849e-02 -6.41472340e-01
9.83684957e-01 8.60700682e-02 -3.55652750e-01 -9.45670068e-01
-1.27557099e-01 5.62272608e-01 -2.26368740e-01 -6.57304764e-01
6.23084962e-01 -1.04975864e-01 -2.11488917e-01 5.58409572e-01
2.75431216e-01 -3.39259952e-01 2.46987239e-01 5.57285309e-01
1.18477964e+00 3.28427583e-01 1.02269605e-01 -2.65659969e-02
1.10826564e+00 2.34050497e-01 2.62051046e-01 1.04332304e+00
5.43851405e-02 4.30325150e-01 3.52185935e-01 5.34286583e-03
-2.20422521e-01 -1.07949376e+00 1.99408263e-01 1.51372433e+00
1.16090067e-01 -7.96253979e-01 -8.47020149e-01 -1.17912829e+00
-8.76910314e-02 1.02501297e+00 -3.75050902e-01 -3.11942279e-01
-9.23414528e-01 -4.63763281e-04 7.21993268e-01 6.17525101e-01
5.43249547e-01 -9.89075363e-01 -6.23166680e-01 3.19267780e-01
-7.60313213e-01 -1.22322392e+00 -5.01136363e-01 4.34327610e-02
-7.72850156e-01 -1.15369129e+00 -2.64691710e-01 -5.89458883e-01
3.68498385e-01 1.21004619e-02 1.70770919e+00 1.37972444e-01
4.15551186e-01 9.02836144e-01 -4.63126510e-01 -5.54525070e-02
-5.79600990e-01 4.45034027e-01 -3.98729384e-01 -2.08787963e-01
6.39450073e-01 -2.33535826e-01 -4.57934856e-01 4.31326956e-01
-8.83883417e-01 -2.73794770e-01 4.83664840e-01 9.64488328e-01
7.19013870e-01 -3.84665132e-01 1.02618706e+00 -1.15821183e+00
7.52323925e-01 -5.55750370e-01 -7.29866266e-01 9.14063811e-01
-7.24068820e-01 5.05608618e-01 7.39427507e-01 -2.79168040e-01
-1.43255460e+00 -2.51516491e-01 -3.54506701e-01 -3.41850482e-02
-1.28717765e-01 7.56948829e-01 -2.57999271e-01 -4.10400219e-02
8.80079508e-01 9.44858417e-02 -3.67085308e-01 -4.80162382e-01
8.59164536e-01 6.47215188e-01 7.27423370e-01 -1.12589872e+00
8.85004163e-01 5.40569089e-02 -3.72031212e-01 -1.51701123e-01
-1.38767350e+00 -5.03836334e-01 -3.37101877e-01 1.86643556e-01
7.09258795e-01 -6.41490757e-01 -6.25573277e-01 -2.06469037e-02
-1.09779656e+00 -6.60039783e-01 -4.62025017e-01 -2.26950366e-02
-5.88502109e-01 5.01574814e-01 -5.44814169e-01 -6.35282934e-01
-5.52835047e-01 -9.02535737e-01 1.09738779e+00 2.61220455e-01
-4.28346038e-01 -1.01803815e+00 1.07213043e-01 1.04944718e+00
7.21078932e-01 -3.75591338e-01 1.14975798e+00 -9.04520988e-01
-8.00859392e-01 -7.45858103e-02 -2.51598239e-01 4.52859879e-01
-9.66873020e-03 -4.84940588e-01 -1.17942846e+00 8.13255534e-02
1.88843325e-01 -8.32091630e-01 8.57888520e-01 -3.16074550e-01
1.06785512e+00 -4.77734476e-01 2.48844409e-03 2.36012936e-01
1.14556932e+00 -1.00666128e-01 5.09684861e-01 4.24781054e-01
3.19162846e-01 8.62504482e-01 7.40435958e-01 -1.05998643e-01
1.09128487e+00 6.40546083e-01 1.36922032e-01 4.70074415e-01
-3.06751937e-01 -6.21498764e-01 3.48365009e-01 7.59607315e-01
4.12458897e-01 -9.63994190e-02 -1.13205206e+00 5.94664752e-01
-1.65477264e+00 -5.61335206e-01 -3.09458748e-02 2.09318757e+00
1.39561975e+00 1.20265476e-01 -2.12445587e-01 -1.63672119e-01
1.51015669e-01 -1.23515368e-01 -5.36377251e-01 -5.16947389e-01
-1.23659872e-01 8.16248655e-01 -6.39560958e-03 1.06885326e+00
-6.69094384e-01 1.25688720e+00 5.80181789e+00 1.00012398e+00
-6.60441518e-01 2.55734295e-01 1.34505004e-01 1.68959379e-01
-7.60875165e-01 3.98661345e-01 -1.08269715e+00 1.35819882e-01
1.38396037e+00 -1.54823968e-02 3.61626238e-01 5.66969573e-01
-4.09153193e-01 -4.15290892e-01 -1.11716068e+00 6.13987863e-01
2.22972542e-01 -1.37141120e+00 1.45524383e-01 -4.96128261e-01
6.26519263e-01 -1.73745558e-01 -1.98408701e-02 9.43254113e-01
3.54099154e-01 -1.00231278e+00 5.01241803e-01 8.42643619e-01
6.13959610e-01 -5.26180446e-01 8.48670125e-01 7.82557130e-01
-9.45983291e-01 -1.96851447e-01 -2.07648411e-01 -1.41486809e-01
2.88261652e-01 3.20485473e-01 -8.41070354e-01 7.34157622e-01
5.98469615e-01 -2.26815324e-02 -9.62760329e-01 5.95475554e-01
-6.50806248e-01 9.70097125e-01 -4.79308844e-01 -1.99929741e-03
2.02475563e-01 6.16563037e-02 2.82783687e-01 9.83378291e-01
-1.44001730e-02 2.34110594e-01 1.36636356e-02 9.38178241e-01
-2.53570557e-01 1.40414223e-01 1.01011749e-02 1.87379405e-01
5.60232699e-01 9.88895714e-01 5.32787256e-02 -7.08270609e-01
-6.02981329e-01 9.63647425e-01 6.53095901e-01 4.78912354e-01
-5.86310744e-01 -5.49084723e-01 4.18972602e-04 4.52607088e-02
3.21335375e-01 7.13708103e-02 -1.27164498e-01 -1.24645317e+00
3.08841884e-01 -1.38680661e+00 8.69684756e-01 -9.77710664e-01
-1.36834240e+00 6.44559383e-01 1.02257617e-01 -5.89020908e-01
-6.99329555e-01 -4.84672427e-01 -3.67871404e-01 1.16710782e+00
-2.08482337e+00 -1.07225406e+00 -1.63683638e-01 7.51212239e-01
3.97462487e-01 8.28772411e-02 1.04501700e+00 9.62913334e-02
-3.94461527e-02 8.40573788e-01 -3.58968407e-01 -1.62474051e-01
1.16363347e+00 -1.48913813e+00 2.68855125e-01 5.98388433e-01
2.69748867e-01 1.10120249e+00 4.55949455e-01 -5.29559970e-01
-1.50484467e+00 -8.71164978e-01 1.59013748e+00 -1.30011165e+00
7.98670948e-01 -2.32065454e-01 -1.33786893e+00 6.27028823e-01
3.69346499e-01 -2.33417768e-02 7.66963243e-01 2.37843588e-01
-7.29778349e-01 -2.75068343e-01 -1.09439695e+00 4.94969040e-01
7.36662984e-01 -9.05979633e-01 -1.42886829e+00 1.20194450e-01
1.20491660e+00 -7.30034471e-01 -8.20641041e-01 5.03996015e-01
2.24394098e-01 -6.86632991e-01 9.02685285e-01 -6.00749493e-01
1.69424906e-01 -4.18961883e-01 -4.33253735e-01 -9.48788583e-01
1.37248650e-01 -6.82438433e-01 -7.89035380e-01 1.30038249e+00
7.28950262e-01 -8.62916529e-01 5.20678282e-01 1.12584853e+00
-2.53496438e-01 -8.17247391e-01 -8.16084385e-01 -7.47643411e-01
3.53115678e-01 -5.35766602e-01 7.74285853e-01 5.51212966e-01
-1.40432090e-01 6.41436756e-01 1.74442276e-01 2.88756460e-01
2.35439256e-01 4.69404012e-01 9.33436930e-01 -8.63309026e-01
-5.12942970e-01 -2.21359525e-02 3.83006811e-01 -1.71561515e+00
3.64068806e-01 -8.69733453e-01 -5.02335615e-02 -1.64645815e+00
-5.09157293e-02 -6.30340338e-01 -1.73349440e-01 4.52122241e-01
-6.08687818e-01 8.54999572e-02 2.20663533e-01 -4.56085838e-02
-1.04705930e+00 4.42487061e-01 1.17358458e+00 6.88097440e-03
-6.20762706e-02 -7.56940171e-02 -9.54411209e-01 6.04761958e-01
8.34673882e-01 -3.17780286e-01 -7.30629981e-01 -5.42319775e-01
6.03364766e-01 1.10007316e-01 5.20639539e-01 -7.95012057e-01
6.16723716e-01 -8.80358145e-02 -1.23376988e-01 -4.98938829e-01
3.94898355e-01 -6.91005826e-01 -5.01169801e-01 2.51179487e-02
-4.81593221e-01 -1.46589279e-02 3.09682876e-01 5.13232768e-01
-4.77833092e-01 -6.57217622e-01 3.72489363e-01 -1.80592090e-01
-7.85275817e-01 -2.69916713e-01 -1.49441928e-01 8.59886885e-01
4.95739162e-01 -1.54368449e-02 -6.39064133e-01 -5.07249475e-01
-5.47345936e-01 6.75580442e-01 7.14525953e-02 4.11714792e-01
5.89341342e-01 -9.61216033e-01 -4.65287000e-01 4.02137544e-03
2.78032929e-01 2.31698323e-02 2.36263067e-01 5.82855940e-01
-6.68769255e-02 8.13420773e-01 5.69845855e-01 -4.40999031e-01
-9.12897527e-01 4.47630376e-01 5.98666310e-01 -7.44825125e-01
-1.64741114e-01 8.58103991e-01 -4.35488932e-02 -1.09129786e+00
1.76273823e-01 -5.51876247e-01 -1.78957045e-01 3.56731899e-02
5.62000692e-01 2.84719139e-01 3.78562540e-01 -3.07436764e-01
-2.31562525e-01 7.21366763e-01 -5.10511138e-02 -2.98009545e-01
8.32245767e-01 -2.29370624e-01 -2.44478479e-01 2.75402993e-01
9.23763096e-01 3.19584221e-01 -8.58650208e-01 -8.37363064e-01
3.83687615e-01 -3.61431956e-01 -2.34470159e-01 -1.56961262e+00
-4.70761091e-01 8.87399256e-01 9.44053754e-02 -2.78475713e-02
1.15615523e+00 2.83864200e-01 1.30511904e+00 1.06658089e+00
4.76732016e-01 -1.03539419e+00 3.49432766e-01 1.03760874e+00
9.11358356e-01 -1.20632589e+00 -5.26329935e-01 -3.63557100e-01
-5.66043615e-01 9.28128421e-01 1.05166483e+00 1.35470122e-01
3.15538049e-01 -2.16624979e-02 4.79588300e-01 -1.97454453e-01
-1.05611873e+00 -3.25991899e-01 6.19348168e-01 5.13975680e-01
2.35146135e-01 -2.57105678e-01 -7.03317150e-02 1.12174845e+00
-4.38056439e-01 1.18913792e-01 2.57514000e-01 8.74353230e-01
-6.10086441e-01 -1.50816047e+00 -4.79073018e-01 4.63159025e-01
-3.51377875e-01 -5.02285302e-01 -5.27666032e-01 5.10382473e-01
7.51072317e-02 1.34443963e+00 -4.14311200e-01 -7.93592110e-02
6.44829392e-01 6.27404451e-01 4.57743227e-01 -9.66218293e-01
-9.21288013e-01 -4.60290372e-01 5.03861129e-01 -7.02704251e-01
-3.07077467e-01 -2.78459460e-01 -1.33653080e+00 5.89427389e-02
-4.27776158e-01 4.78014141e-01 8.07348490e-02 1.16975462e+00
5.65020859e-01 2.39971817e-01 8.31662640e-02 1.64404526e-01
-1.06228018e+00 -7.29170978e-01 2.40918949e-01 1.76825285e-01
3.99821371e-01 -3.68344307e-01 -3.15268099e-01 -1.44431693e-02]
|
[11.10256290435791, 7.942512512207031]
|
e014c029-6ead-4726-8dbf-93ee20f8af5f
|
dink-net-neural-clustering-on-large-graphs
|
2305.18405
| null |
https://arxiv.org/abs/2305.18405v2
|
https://arxiv.org/pdf/2305.18405v2.pdf
|
Dink-Net: Neural Clustering on Large Graphs
|
Deep graph clustering, which aims to group the nodes of a graph into disjoint clusters with deep neural networks, has achieved promising progress in recent years. However, the existing methods fail to scale to the large graph with million nodes. To solve this problem, a scalable deep graph clustering method (Dink-Net) is proposed with the idea of dilation and shrink. Firstly, by discriminating nodes, whether being corrupted by augmentations, representations are learned in a self-supervised manner. Meanwhile, the cluster centres are initialized as learnable neural parameters. Subsequently, the clustering distribution is optimized by minimizing the proposed cluster dilation loss and cluster shrink loss in an adversarial manner. By these settings, we unify the two-step clustering, i.e., representation learning and clustering optimization, into an end-to-end framework, guiding the network to learn clustering-friendly features. Besides, Dink-Net scales well to large graphs since the designed loss functions adopt the mini-batch data to optimize the clustering distribution even without performance drops. Both experimental results and theoretical analyses demonstrate the superiority of our method. Compared to the runner-up, Dink-Net achieves 9.62% NMI improvement on the ogbn-papers100M dataset with 111 million nodes and 1.6 billion edges. The source code is released at https://github.com/yueliu1999/Dink-Net. Besides, a collection (papers, codes, and datasets) of deep graph clustering is shared at https://github.com/yueliu1999/Awesome-Deep-Graph-Clustering.
|
['Stan Z. Li', 'Xinwang Liu', 'Xihong Yang', 'Sihang Zhou', 'Jun Xia', 'Ke Liang', 'Yue Liu']
|
2023-05-28
| null | null | null | null |
['graph-clustering']
|
['graphs']
|
[-4.43895638e-01 1.27887085e-01 -1.20525248e-01 -4.75603610e-01
-6.97525442e-01 -6.48099184e-01 1.58718407e-01 9.52517241e-02
-2.42938787e-01 3.96199107e-01 -1.62031069e-01 -2.68199414e-01
-1.12401411e-01 -8.77488554e-01 -7.78595507e-01 -8.71781647e-01
-2.16772795e-01 5.99294782e-01 -2.83382982e-01 1.82738945e-01
-4.73838151e-02 1.85967743e-01 -7.56403685e-01 -1.31219447e-01
8.68879676e-01 7.77675986e-01 2.52123266e-01 5.00738740e-01
2.09445193e-01 6.38913155e-01 -3.30457300e-01 -4.47781831e-01
3.78766239e-01 -2.46348381e-01 -8.59690249e-01 1.87654138e-01
2.59484589e-01 -8.00801218e-02 -9.80384350e-01 1.25909460e+00
6.56454086e-01 1.28505141e-01 4.07296091e-01 -1.50762963e+00
-1.09450710e+00 1.04346275e+00 -8.39662671e-01 -1.30062193e-01
-3.05811971e-01 2.98311532e-01 1.03998411e+00 -5.32451212e-01
3.48998904e-01 1.13461912e+00 7.04874992e-01 6.26846313e-01
-1.22816455e+00 -1.11493993e+00 3.00833225e-01 4.73271497e-02
-1.76632214e+00 -1.11547962e-01 9.38767314e-01 -3.18241656e-01
4.45390016e-01 -4.72587720e-02 3.72698039e-01 9.39632952e-01
-2.20131472e-01 7.73981512e-01 4.28419113e-01 -3.99397798e-02
1.32320747e-01 -2.80386269e-01 1.67110577e-01 7.55887032e-01
4.12773728e-01 -2.04943016e-01 2.31120214e-01 2.18575388e-01
6.44464314e-01 2.98963308e-01 -1.46983042e-01 -4.20509189e-01
-1.03037965e+00 9.21086133e-01 1.03726280e+00 1.69677958e-01
-1.18099257e-01 6.38217390e-01 6.16832733e-01 2.27056086e-01
4.30962473e-01 2.47200012e-01 -3.49194020e-01 1.98116139e-01
-7.37454832e-01 9.34117194e-03 5.77450752e-01 1.25450814e+00
8.63814414e-01 -3.59553620e-02 -4.29661125e-02 7.80804634e-01
3.61298621e-01 3.11577201e-01 2.85907894e-01 -1.01020062e+00
6.26159191e-01 7.98893571e-01 -5.39333403e-01 -1.37601054e+00
-5.52944958e-01 -6.54616594e-01 -1.58496535e+00 -2.58870244e-01
1.52999997e-01 -4.44336861e-01 -1.02579856e+00 1.93358314e+00
3.09247315e-01 3.26389998e-01 -1.08814083e-01 9.25521135e-01
9.07698750e-01 7.87875235e-01 -1.57975838e-01 1.17218874e-01
9.92133439e-01 -1.31793976e+00 -4.70773786e-01 -5.70550002e-02
7.85918653e-01 -2.69511253e-01 1.06697512e+00 1.91653177e-01
-9.80118990e-01 -5.24557352e-01 -9.00974035e-01 -8.81509036e-02
-3.51132214e-01 1.23152182e-01 8.70972335e-01 3.43308628e-01
-1.22894168e+00 5.41122913e-01 -1.03477097e+00 -3.19117278e-01
8.37810755e-01 6.98982298e-01 -2.13110715e-01 -3.09162349e-01
-9.84205782e-01 -4.24855128e-02 6.38781428e-01 2.87594289e-01
-1.01504898e+00 -5.39725959e-01 -8.76674056e-01 3.13592613e-01
4.74215239e-01 -6.80458963e-01 8.01948249e-01 -8.35999727e-01
-1.17519248e+00 7.95330107e-01 3.01442742e-01 -3.77197176e-01
4.06075388e-01 1.18440129e-01 -2.90005445e-01 1.68643788e-01
1.20361172e-01 6.56877995e-01 4.34913635e-01 -1.24257445e+00
-3.45177710e-01 -3.98694783e-01 -8.00738484e-02 -3.00390478e-02
-7.32677698e-01 -2.23474130e-01 -1.28976059e+00 -7.24480748e-01
-2.47910060e-02 -1.08107388e+00 -5.45949876e-01 -3.10835183e-01
-9.96535182e-01 -1.86717480e-01 7.58345068e-01 -4.35298592e-01
1.40261626e+00 -2.29755688e+00 2.85220385e-01 5.33280432e-01
8.34341884e-01 1.43481702e-01 -3.68974984e-01 3.30669522e-01
-1.78070307e-01 4.48347092e-01 -3.69170964e-01 -6.48275673e-01
1.49262473e-01 -1.82658006e-02 1.37079552e-01 7.28033006e-01
1.72073524e-02 1.12490225e+00 -7.30360806e-01 -4.01572347e-01
2.64921218e-01 4.60604697e-01 -6.50218189e-01 2.77111709e-01
1.27453143e-02 3.64181012e-01 -5.12567222e-01 5.72977424e-01
9.69929397e-01 -7.69627452e-01 5.01215398e-01 -2.30250470e-02
3.43768835e-01 -1.20159149e-01 -1.07113671e+00 1.81962430e+00
-2.27901578e-01 5.09738326e-01 3.47987115e-01 -1.39187944e+00
8.74169886e-01 -1.53014176e-02 4.87175167e-01 -4.89993304e-01
3.50971013e-01 -6.76970091e-03 -1.60517520e-03 -8.14781785e-02
1.17554098e-01 3.04006487e-01 -3.32008451e-01 2.60640413e-01
-8.40412627e-04 3.40632439e-01 4.38990146e-01 7.06160784e-01
1.38781321e+00 -3.32872778e-01 -2.49642879e-01 -2.29314685e-01
2.75465012e-01 -2.00999573e-01 7.31754959e-01 5.26716471e-01
-8.14363733e-02 7.34512806e-01 8.27462673e-01 -3.46010089e-01
-9.23274338e-01 -8.82876813e-01 1.64128989e-01 7.46095419e-01
3.62351984e-01 -6.83281839e-01 -9.42332745e-01 -7.86062717e-01
1.92266449e-01 2.02543274e-01 -7.73428023e-01 -4.05341566e-01
-5.24624467e-01 -8.83842885e-01 5.43245554e-01 4.29975033e-01
4.42551583e-01 -1.10375726e+00 3.58388245e-01 1.80850118e-01
-7.61090126e-03 -1.04931068e+00 -8.05688202e-01 1.47213563e-01
-6.92335784e-01 -1.20345116e+00 -5.02491832e-01 -1.15110052e+00
1.08099914e+00 3.12797785e-01 1.14002573e+00 6.22763634e-01
-3.77712786e-01 1.47608206e-01 -4.21584815e-01 1.72532022e-01
-1.31312519e-01 6.65861189e-01 -1.47701979e-01 -4.77534607e-02
2.18049645e-01 -9.33562338e-01 -8.44209015e-01 2.17686474e-01
-8.60250533e-01 -2.28127576e-02 6.94241345e-01 6.96399927e-01
7.59813070e-01 4.36061293e-01 5.95049441e-01 -1.13830471e+00
4.49337900e-01 -8.07249248e-01 -6.89828813e-01 -4.06058243e-04
-7.48253405e-01 -2.11803779e-01 1.06965590e+00 -3.46975774e-01
-3.17721069e-01 1.12564906e-01 -3.83820795e-02 -8.26145589e-01
-1.23016380e-01 6.22146726e-01 -6.48548782e-01 9.06373635e-02
2.04784423e-01 2.82319095e-02 1.12825587e-01 -4.26437736e-01
5.68642974e-01 4.28982764e-01 5.92355251e-01 -4.89585340e-01
1.25361717e+00 3.15139532e-01 -1.32346451e-01 -3.28111947e-01
-6.98028922e-01 -5.19901097e-01 -4.61925268e-01 -4.74471003e-02
7.71995425e-01 -1.05822694e+00 -9.75838363e-01 6.44531906e-01
-7.25563645e-01 -7.82510281e-01 -1.72870010e-02 2.79757321e-01
-2.16879725e-01 5.40228724e-01 -1.00682163e+00 -2.98157156e-01
-6.01542056e-01 -1.08856821e+00 8.16780388e-01 2.69788742e-01
3.19085866e-01 -1.09791160e+00 -2.64909148e-01 3.96913916e-01
1.21344574e-01 4.52958882e-01 7.87286282e-01 -6.96717978e-01
-6.46300197e-01 -3.00105572e-01 -5.16381681e-01 4.13575262e-01
1.67763233e-01 1.43619657e-01 -5.11419773e-01 -7.82157779e-01
-6.00528538e-01 -4.75937724e-01 1.04125464e+00 4.86731797e-01
1.95775855e+00 -4.60461229e-01 -4.72335875e-01 1.27465248e+00
1.59315753e+00 5.32260612e-02 5.74509203e-01 2.00814918e-01
1.33236051e+00 2.01912522e-01 1.69986993e-01 2.52564281e-01
6.94464147e-01 3.66121441e-01 7.72197306e-01 -4.27459300e-01
-9.36005935e-02 -2.43791416e-01 3.92881334e-02 1.12277257e+00
2.62319088e-01 -5.96467257e-01 -1.03816092e+00 4.42071885e-01
-1.97267473e+00 -6.94693863e-01 -1.27875939e-01 1.89457142e+00
7.14825273e-01 2.34918311e-01 2.19181776e-01 -1.77650869e-01
1.18054628e+00 2.44111091e-01 -8.86650980e-01 7.59499744e-02
2.42906436e-02 -1.15954071e-01 6.11035287e-01 4.45767522e-01
-1.27183723e+00 1.14005673e+00 4.04795885e+00 1.00854623e+00
-1.01201057e+00 6.96278736e-02 1.06559145e+00 -1.00185581e-01
-2.91912347e-01 1.57088079e-02 -5.63465297e-01 8.38554502e-01
8.62530529e-01 -2.11746871e-01 7.53821731e-01 8.02675545e-01
1.29214093e-01 6.38370514e-01 -8.35738778e-01 9.09044981e-01
-1.30805805e-01 -1.39578128e+00 -1.50085956e-01 3.30442816e-01
8.45990181e-01 2.95242041e-01 9.54991505e-02 5.95340014e-01
7.09796727e-01 -1.19545579e+00 3.87809604e-01 2.00895339e-01
8.62059832e-01 -1.15255439e+00 6.58821583e-01 2.47062430e-01
-1.45386314e+00 -1.39955014e-01 -5.47187746e-01 1.45348147e-01
-7.87017122e-02 8.62282753e-01 -5.07762611e-01 6.99857950e-01
8.66391599e-01 1.01223207e+00 -6.43206120e-01 9.21982288e-01
-8.17407295e-02 9.28532064e-01 -2.95025080e-01 1.34665981e-01
3.93483430e-01 -5.32945871e-01 1.55879125e-01 1.18971610e+00
4.98431213e-02 -9.68084410e-02 6.08651519e-01 1.06770551e+00
-8.31042647e-01 4.82099205e-02 -3.59631151e-01 -2.53094018e-01
8.34615767e-01 1.67399478e+00 -1.00123727e+00 -1.48481846e-01
-1.92601025e-01 9.94678617e-01 7.58584619e-01 4.62568730e-01
-1.10068917e+00 -7.78664947e-01 5.59881449e-01 -3.54538709e-02
4.65973049e-01 -1.39549911e-01 -1.08838670e-01 -9.89439011e-01
6.35993704e-02 -7.57336736e-01 4.98179376e-01 -3.49558860e-01
-1.59185624e+00 6.34394586e-01 -3.87986690e-01 -9.69336212e-01
3.75146091e-01 -2.55751401e-01 -1.03800416e+00 6.01164758e-01
-1.20380831e+00 -1.36638606e+00 -6.01759613e-01 5.63510299e-01
1.04071818e-01 -1.50420338e-01 4.31269020e-01 6.30563736e-01
-1.20060360e+00 1.08169532e+00 4.86335158e-01 7.44110882e-01
5.75041950e-01 -1.51431406e+00 7.54407823e-01 8.98428023e-01
-2.26324186e-01 5.81620276e-01 1.30833358e-01 -5.03009439e-01
-1.31949973e+00 -1.75371706e+00 3.19151521e-01 -5.02821244e-02
6.87386215e-01 -9.13321376e-01 -9.44552779e-01 6.78572059e-01
1.27064422e-01 4.65219587e-01 4.93412167e-01 1.93509832e-01
-4.03643817e-01 -4.18953776e-01 -9.16133463e-01 5.42753935e-01
1.14322317e+00 -1.99791819e-01 2.37421691e-01 7.47431517e-01
1.30109751e+00 -3.85502875e-01 -1.08073759e+00 2.36279339e-01
-1.17490971e-02 -5.10975063e-01 8.37813497e-01 -5.71309030e-01
5.13544977e-01 -3.45347166e-01 6.86387122e-02 -1.23737466e+00
-6.69070303e-01 -7.80259252e-01 -8.15860182e-02 1.58616722e+00
5.35611749e-01 -6.39689922e-01 9.94625986e-01 3.79247844e-01
-3.57793033e-01 -1.00375056e+00 -5.44203639e-01 -6.78760529e-01
4.99379873e-01 -8.63320157e-02 7.24644959e-01 1.31076503e+00
-3.37548614e-01 5.11329710e-01 -2.57343620e-01 3.42073262e-01
8.48040760e-01 2.42200091e-01 9.86502647e-01 -1.16601276e+00
-1.34220645e-01 -5.79762101e-01 -3.69108379e-01 -9.24475193e-01
5.19207418e-01 -1.41282964e+00 -6.24254234e-02 -1.75416791e+00
4.44810510e-01 -7.48501956e-01 -4.87715393e-01 7.13904560e-01
-3.78984839e-01 1.31509721e-01 2.14198291e-01 1.85704187e-01
-1.13366008e+00 6.53472900e-01 1.16111112e+00 -3.29753846e-01
-1.62329644e-01 -6.27090111e-02 -1.03219402e+00 3.91428143e-01
1.12117028e+00 -5.73914647e-01 -4.22012955e-01 -5.08237302e-01
1.76815107e-01 -1.54212460e-01 3.51471394e-01 -9.68810916e-01
3.81932527e-01 1.30491003e-01 3.06562692e-01 -5.56183398e-01
-1.47783995e-01 -7.29912281e-01 7.83202797e-02 4.41562712e-01
-4.42727685e-01 1.36050284e-01 -2.90282816e-02 7.92348146e-01
-9.66010317e-02 2.18271032e-01 8.68875921e-01 3.56128998e-02
-4.91451591e-01 1.08215749e+00 7.40140826e-02 4.50787485e-01
1.09308708e+00 8.39099735e-02 -4.43870366e-01 -3.58871967e-01
-7.69982815e-01 9.65021908e-01 5.75053692e-01 3.33120793e-01
4.19932336e-01 -1.38316190e+00 -8.53557825e-01 -1.89807173e-02
-3.28827202e-02 7.96783984e-01 5.51128447e-01 7.44097471e-01
-5.89611471e-01 2.31837761e-02 1.45626307e-01 -4.00413007e-01
-8.74637246e-01 9.23439264e-01 3.60105097e-01 -3.56891781e-01
-8.53097260e-01 1.03083956e+00 3.36337805e-01 -8.77920210e-01
5.84658325e-01 -1.68665703e-02 9.72848535e-02 -3.18774223e-01
1.08713664e-01 1.98467866e-01 -7.86497518e-02 -3.23377311e-01
-3.55510175e-01 2.27498084e-01 -3.58098596e-01 6.60002410e-01
1.53698087e+00 -1.48579419e-01 -3.81040931e-01 1.02980668e-02
1.73116219e+00 -1.79380611e-01 -1.19127190e+00 -1.61663964e-01
-2.24527463e-01 -1.49154648e-01 -3.56257171e-03 -4.68626738e-01
-2.04053640e+00 7.84928381e-01 4.06605422e-01 2.22464517e-01
1.21510780e+00 2.40635082e-01 1.03009999e+00 3.84023964e-01
-3.61487493e-02 -9.40595984e-01 1.09223761e-01 3.19896162e-01
6.12689018e-01 -1.34400260e+00 -1.16378762e-01 -3.56088012e-01
-4.98700202e-01 7.65882909e-01 9.49170947e-01 -4.17586923e-01
8.23899508e-01 2.08243906e-01 -7.70146027e-02 -4.29003060e-01
-6.86375856e-01 6.03406765e-02 -7.95779452e-02 4.98327851e-01
1.72804236e-01 3.42887789e-01 6.15040548e-02 8.68520796e-01
-1.38522476e-01 -4.70436424e-01 3.93723905e-01 4.02607262e-01
2.69248411e-02 -8.84739220e-01 9.26873684e-02 6.14335358e-01
-3.47792149e-01 -2.59276837e-01 -5.13797343e-01 9.19365406e-01
-2.74731498e-02 8.33474159e-01 1.37122452e-01 -7.89440155e-01
1.39669403e-01 -6.82521224e-01 -1.11517720e-01 -5.76736867e-01
-5.89463770e-01 3.87049727e-02 -3.52120042e-01 -5.82729757e-01
-6.41322583e-02 -4.02631968e-01 -1.58125257e+00 -6.99046373e-01
-3.93216699e-01 3.99444640e-01 3.42100620e-01 3.97678435e-01
6.11312866e-01 7.86046207e-01 1.12282801e+00 -7.09631622e-01
-4.00647283e-01 -8.66330266e-01 -7.43684769e-01 5.50918341e-01
1.89813450e-01 -2.59483397e-01 -6.14967465e-01 -2.37001657e-01]
|
[7.337675094604492, 6.028116226196289]
|
6f8b211c-3bb6-4a23-8b26-c5bfefdf5a47
|
synctalkface-talking-face-generation-with
|
2211.00924
| null |
https://arxiv.org/abs/2211.00924v2
|
https://arxiv.org/pdf/2211.00924v2.pdf
|
SyncTalkFace: Talking Face Generation with Precise Lip-Syncing via Audio-Lip Memory
|
The challenge of talking face generation from speech lies in aligning two different modal information, audio and video, such that the mouth region corresponds to input audio. Previous methods either exploit audio-visual representation learning or leverage intermediate structural information such as landmarks and 3D models. However, they struggle to synthesize fine details of the lips varying at the phoneme level as they do not sufficiently provide visual information of the lips at the video synthesis step. To overcome this limitation, our work proposes Audio-Lip Memory that brings in visual information of the mouth region corresponding to input audio and enforces fine-grained audio-visual coherence. It stores lip motion features from sequential ground truth images in the value memory and aligns them with corresponding audio features so that they can be retrieved using audio input at inference time. Therefore, using the retrieved lip motion features as visual hints, it can easily correlate audio with visual dynamics in the synthesis step. By analyzing the memory, we demonstrate that unique lip features are stored in each memory slot at the phoneme level, capturing subtle lip motion based on memory addressing. In addition, we introduce visual-visual synchronization loss which can enhance lip-syncing performance when used along with audio-visual synchronization loss in our model. Extensive experiments are performed to verify that our method generates high-quality video with mouth shapes that best align with the input audio, outperforming previous state-of-the-art methods.
|
['Yong Man Ro', 'Jeongsoo Choi', 'Joanna Hong', 'Minsu Kim', 'Se Jin Park']
|
2022-11-02
| null | null | null | null |
['audio-visual-synchronization', 'audio-visual-synchronization', 'talking-face-generation', 'face-generation']
|
['audio', 'computer-vision', 'computer-vision', 'computer-vision']
|
[-3.20110954e-02 -3.12668160e-02 -5.62712431e-01 4.59182113e-02
-1.02091169e+00 -5.41559339e-01 5.14147043e-01 -2.08468974e-01
2.53342092e-01 3.68284822e-01 6.64798737e-01 3.16914439e-01
3.28676552e-01 -4.82807368e-01 -7.28536367e-01 -6.50407791e-01
2.26074859e-01 1.13649398e-01 2.42293626e-01 9.57755893e-02
3.76282960e-01 4.79947835e-01 -2.23296309e+00 8.17085743e-01
2.92082429e-01 1.37670314e+00 3.51633221e-01 9.21357155e-01
-2.46329203e-01 5.05743861e-01 -5.75627446e-01 -5.23431972e-02
9.81099457e-02 -3.30308318e-01 -4.47099477e-01 1.87651277e-01
8.62953067e-01 -5.42486370e-01 -4.48365897e-01 6.90713882e-01
7.66184092e-01 3.19404341e-02 4.90816861e-01 -1.48252451e+00
-2.99316347e-01 1.50413305e-01 -4.47436750e-01 -7.81993046e-02
8.75791013e-01 4.17573512e-01 9.31852758e-01 -1.34966195e+00
9.25337493e-01 1.64341080e+00 6.57961011e-01 6.82873428e-01
-1.21347129e+00 -7.97285795e-01 3.46288122e-02 4.60339338e-01
-1.59408998e+00 -1.30791879e+00 1.15035868e+00 -3.24260652e-01
7.54602432e-01 1.55818462e-01 7.95809329e-01 1.14074731e+00
1.37640983e-01 5.42480767e-01 5.90221167e-01 -3.07196677e-01
4.02131677e-02 -1.18321419e-01 -4.92880195e-01 7.28057623e-01
-4.10179585e-01 3.68627399e-01 -1.62925744e+00 -5.42020053e-02
8.33621562e-01 -3.03662002e-01 -5.50711334e-01 -2.35454857e-01
-1.21439683e+00 5.37828326e-01 2.11755902e-01 2.20647410e-01
-3.16727102e-01 3.79991412e-01 2.60436982e-01 1.38633311e-01
1.84292316e-01 -2.29173928e-01 -1.02249250e-01 -1.17925078e-01
-1.43367326e+00 -8.99223387e-02 5.22646248e-01 8.24535131e-01
8.76278996e-01 2.55575538e-01 -3.13232213e-01 6.20439649e-01
7.72341847e-01 8.42284620e-01 6.37271047e-01 -1.58754373e+00
4.52145368e-01 1.41379952e-01 -8.36658478e-02 -1.26518083e+00
-1.44814193e-01 2.52031118e-01 -6.53061926e-01 1.84983790e-01
2.55731225e-01 2.46138915e-01 -7.39826500e-01 1.96954346e+00
4.13803577e-01 8.47878277e-01 -5.99144548e-02 8.59459937e-01
1.12723446e+00 6.88410640e-01 -2.50036180e-01 -5.21882117e-01
1.37446117e+00 -7.66810894e-01 -9.93241429e-01 -1.52783334e-01
-3.04328725e-02 -1.01428425e+00 1.03022873e+00 1.89080492e-01
-1.47355509e+00 -1.17742872e+00 -9.12484169e-01 -7.17642084e-02
1.49306757e-02 1.34171784e-01 6.88562691e-02 5.28750420e-01
-1.53516126e+00 4.97975975e-01 -7.96517372e-01 -9.91736054e-02
9.93956998e-02 3.30632299e-01 -4.52241361e-01 1.86119795e-01
-9.75112200e-01 2.72661090e-01 -1.58508062e-01 -1.62121784e-02
-1.02751696e+00 -8.73232067e-01 -1.12229896e+00 -1.39513060e-01
6.18888997e-02 -8.90708625e-01 1.23507380e+00 -8.64023089e-01
-1.86212015e+00 7.74144411e-01 -9.85703349e-01 -2.06103474e-01
2.75507420e-01 1.50847971e-01 -4.62521225e-01 9.10814822e-01
-2.53839884e-03 1.27270174e+00 1.62286496e+00 -1.50896943e+00
-4.58460748e-01 -2.09842324e-01 -3.31889004e-01 5.55081777e-02
-2.90075988e-01 -3.37662548e-01 -8.18418741e-01 -8.01049471e-01
3.06853265e-01 -7.02640772e-01 4.61316317e-01 5.54560244e-01
-2.88981080e-01 5.28901815e-03 1.08866930e+00 -6.23905897e-01
1.29539335e+00 -2.32773447e+00 -1.50943417e-02 1.26632050e-01
1.68541268e-01 7.58999512e-02 -3.95225704e-01 1.76945150e-01
2.30108202e-03 3.77610363e-02 3.30796659e-01 -8.94424200e-01
-6.71082512e-02 8.20690095e-02 -7.47943223e-01 4.54668045e-01
2.53958404e-01 9.76002574e-01 -8.03954542e-01 -9.99620736e-01
3.29757959e-01 1.09651744e+00 -8.54079545e-01 3.37932318e-01
-2.04430819e-01 5.96596360e-01 1.43532259e-02 8.93568337e-01
5.58335960e-01 -1.74891248e-01 1.84555739e-01 -7.22542226e-01
2.38617122e-01 4.37735885e-01 -1.28390741e+00 1.96164274e+00
-5.40090501e-01 8.57890844e-01 3.67669076e-01 -4.80095506e-01
8.41372073e-01 7.10285842e-01 5.50121009e-01 -8.62448454e-01
-1.15231976e-01 8.54922235e-02 -5.33269167e-01 -3.44977647e-01
4.81533378e-01 2.04944368e-02 1.68052778e-01 4.36268300e-01
2.15488642e-01 -4.26872611e-01 -1.30339816e-01 -6.77987412e-02
5.76049149e-01 1.56086788e-01 -7.17105344e-02 2.47191399e-01
7.34342754e-01 -6.56822205e-01 3.56848598e-01 3.25728267e-01
-2.50006735e-01 1.05737603e+00 1.94657236e-01 1.76393241e-01
-9.21024740e-01 -1.39480078e+00 -4.00990061e-02 1.02011883e+00
1.75466672e-01 -8.74886632e-01 -8.14548075e-01 -1.61964908e-01
-1.84659794e-01 9.51139033e-02 -5.29266953e-01 -1.88474700e-01
-6.66273177e-01 3.01752120e-01 5.20052016e-01 4.59688455e-01
3.21671009e-01 -1.02927041e+00 -5.40583849e-01 1.87752396e-01
-5.86391211e-01 -1.16426492e+00 -9.53160286e-01 -5.14811814e-01
-6.90191984e-01 -9.12368894e-01 -7.64399290e-01 -7.27656186e-01
4.64367926e-01 4.66241300e-01 1.08478951e+00 1.51196480e-01
-3.41557622e-01 7.94550657e-01 1.38417616e-01 2.65868992e-01
-7.41601348e-01 -2.42133304e-01 2.99576461e-01 4.03659701e-01
-2.95448095e-01 -9.38217700e-01 -8.17277491e-01 4.61231858e-01
-7.39271164e-01 7.61725679e-02 1.58739328e-01 7.93572009e-01
8.51229429e-01 -2.64320165e-01 5.13980746e-01 1.10908419e-01
2.86065698e-01 -2.68118054e-01 -3.72581750e-01 9.29300264e-02
-2.29304731e-01 1.31399319e-01 4.03310239e-01 -7.62105107e-01
-8.89967799e-01 1.79131776e-01 -8.07934180e-02 -1.02686596e+00
1.87531915e-02 3.06423046e-02 -3.69547427e-01 1.36205912e-01
3.39025378e-01 2.28908569e-01 4.09011900e-01 -3.94861549e-01
5.23605883e-01 6.79692268e-01 9.71625030e-01 -4.08158988e-01
6.71403050e-01 7.11474955e-01 1.04841113e-01 -9.72302854e-01
-3.50180745e-01 -2.74598241e-01 -5.86618721e-01 -4.85793740e-01
6.38800621e-01 -1.11121643e+00 -1.23565686e+00 4.14203584e-01
-1.34208739e+00 -2.51615912e-01 -2.80510694e-01 1.98572233e-01
-9.93023753e-01 4.21310574e-01 -5.46426654e-01 -8.55189025e-01
-1.03694558e-01 -1.30485284e+00 1.65053129e+00 -1.90485977e-02
-4.85780716e-01 -7.95916677e-01 6.71524147e-04 3.80951464e-01
2.04245910e-01 -1.04133151e-02 6.09151602e-01 2.09172532e-01
-9.25193787e-01 1.74739316e-01 5.43284751e-02 -7.11844862e-02
1.82391569e-01 2.78588891e-01 -1.44251096e+00 -2.58322835e-01
-2.46637076e-01 -2.55167633e-01 8.51708651e-01 5.20832300e-01
7.91578114e-01 -6.09807253e-01 -4.23263550e-01 6.09275162e-01
8.36759150e-01 -2.19087899e-01 5.38013041e-01 -3.36131036e-01
5.00059843e-01 8.10353160e-01 6.22264624e-01 6.13867342e-01
4.75934863e-01 1.16312456e+00 4.54857409e-01 1.19358368e-01
-1.05790639e+00 -6.48965776e-01 8.80250216e-01 1.01878417e+00
2.93700010e-01 7.32171312e-02 -5.87840676e-01 4.77426201e-01
-1.66107070e+00 -1.24396074e+00 4.56619591e-01 2.27900529e+00
1.18486953e+00 -7.01609701e-02 2.03384206e-01 3.41882408e-01
7.37562716e-01 3.86053294e-01 -5.11258125e-01 6.59940541e-02
-2.60143697e-01 2.10947022e-01 -2.49700546e-01 9.54729795e-01
-6.94744587e-01 8.75561953e-01 6.35196161e+00 8.43671560e-01
-1.52517152e+00 1.57361582e-01 3.89543921e-01 -4.63148445e-01
-5.54498076e-01 -2.62597740e-01 -9.22052622e-01 5.45390010e-01
1.14558518e+00 -4.14452553e-02 4.11300063e-01 4.19692189e-01
5.18151224e-01 -3.93438600e-02 -1.27691746e+00 1.37040675e+00
2.74758190e-01 -1.75120986e+00 3.10406536e-01 -6.30008336e-03
4.47569430e-01 -3.82379383e-01 5.52668095e-01 -3.21142137e-01
-3.24575484e-01 -1.17016852e+00 1.10981250e+00 8.19952965e-01
1.26851439e+00 -6.33534074e-01 -1.57707604e-03 1.51827512e-03
-1.80424750e+00 -6.76510064e-03 7.35810176e-02 4.52072889e-01
2.61875421e-01 1.50316268e-01 -1.08565986e+00 1.62169427e-01
8.40127230e-01 6.98221684e-01 -4.38733518e-01 6.52756453e-01
-5.19537292e-02 3.08452725e-01 -3.73959541e-01 5.40472567e-01
-3.51397395e-01 5.05739868e-01 5.97923517e-01 8.50840449e-01
5.71024895e-01 -3.19480449e-01 2.80985106e-02 9.54792857e-01
-6.54935837e-02 -5.88737130e-02 -7.07841814e-01 1.11076556e-01
9.24613297e-01 9.07501876e-01 -3.30883771e-01 -1.30462706e-01
-1.66854218e-01 8.53221476e-01 -1.34252384e-01 3.66376460e-01
-7.28721082e-01 -1.18289515e-02 1.10050476e+00 4.34147477e-01
4.03264940e-01 -2.46627837e-01 6.28426298e-02 -9.04869318e-01
9.91018862e-02 -7.84119964e-01 -1.37810027e-02 -1.10287142e+00
-8.27245712e-01 5.33695340e-01 -3.22175413e-01 -1.42087233e+00
-9.87766504e-01 -3.56024861e-01 -3.65735680e-01 6.47449791e-01
-1.65869164e+00 -1.24764633e+00 -4.08270270e-01 1.14423764e+00
5.63063204e-01 -7.68687725e-02 9.17954981e-01 1.69350937e-01
-1.08302407e-01 1.02386963e+00 -3.34148318e-01 -7.55820647e-02
1.11448038e+00 -3.92741501e-01 1.54985547e-01 4.04250413e-01
5.10175467e-01 4.43529814e-01 5.74899435e-01 -4.94013250e-01
-1.56805634e+00 -8.17005396e-01 9.71336067e-01 -3.62734437e-01
4.06204104e-01 -3.89207959e-01 -7.85505593e-01 2.10283682e-01
1.59915522e-01 2.90267974e-01 5.70577025e-01 -5.61399221e-01
-6.29287004e-01 -3.83573383e-01 -9.63458359e-01 6.80064142e-01
9.82256770e-01 -1.30011225e+00 -3.66826385e-01 -9.45893377e-02
9.31192517e-01 -4.62397367e-01 -8.32760572e-01 1.91314310e-01
7.07463562e-01 -1.11213136e+00 1.45709550e+00 -9.31915417e-02
2.73675978e-01 -5.12371778e-01 -3.83418798e-01 -8.65122437e-01
3.19986224e-01 -1.16831315e+00 -6.40687704e-01 1.64437866e+00
1.09753720e-01 -1.94555983e-01 6.83034182e-01 1.43833026e-01
1.21866375e-01 -4.57946986e-01 -1.39077783e+00 -4.70294803e-01
-4.00033832e-01 -6.49322927e-01 6.16874695e-01 5.22395015e-01
6.41004741e-02 1.72008485e-01 -3.87427151e-01 2.39373744e-01
5.26472270e-01 3.50842476e-01 8.47951889e-01 -1.14565039e+00
-2.66086429e-01 -4.52424586e-01 -3.27960968e-01 -1.29432678e+00
5.75065136e-01 -6.92005336e-01 1.10443793e-01 -1.05263090e+00
-1.54376581e-01 -1.40324429e-01 9.86898616e-02 4.67048764e-01
2.93795258e-01 7.04251409e-01 4.91598010e-01 3.50378186e-01
-3.95711392e-01 5.63814342e-01 1.26581085e+00 -2.99444884e-01
-3.09020996e-01 -1.59186155e-01 -2.54761010e-01 6.99512899e-01
4.08100426e-01 -1.97932273e-01 -3.71578306e-01 -1.72594979e-01
-6.98066130e-02 7.16062367e-01 7.23054945e-01 -1.12111104e+00
6.32364571e-01 6.93680197e-02 4.97858047e-01 -8.35204005e-01
1.15078151e+00 -7.46177435e-01 1.82825461e-01 1.41569108e-01
-2.73441941e-01 7.96508323e-03 3.41598481e-01 5.85012317e-01
-4.49063629e-01 2.09375009e-01 7.18964696e-01 2.02280417e-01
-5.34565389e-01 3.87278467e-01 -2.58961678e-01 4.12204862e-02
6.82434320e-01 -5.91389775e-01 -1.01643100e-01 -8.21922421e-01
-9.03391302e-01 -2.30491742e-01 5.88994741e-01 5.83750427e-01
9.23976064e-01 -1.72663629e+00 -3.93103331e-01 7.49742985e-01
-7.55545571e-02 -2.94523239e-01 3.95316243e-01 8.71242940e-01
7.47824367e-03 4.49889332e-01 -1.96476489e-01 -1.16736615e+00
-1.41103435e+00 5.46462178e-01 2.27134705e-01 2.04222023e-01
-5.06523848e-01 6.67625964e-01 2.30279282e-01 2.30304509e-01
6.49618685e-01 -1.88235462e-01 -2.81637833e-02 4.98150826e-01
6.86458170e-01 1.27350599e-01 -2.47412790e-02 -1.07436562e+00
-3.68538171e-01 1.22070181e+00 5.87069750e-01 -5.92202246e-01
9.30656195e-01 -6.54497266e-01 1.45386562e-01 5.29560328e-01
1.44061172e+00 5.20544291e-01 -1.67532098e+00 -3.12143207e-01
-3.84689569e-01 -5.97449541e-01 -4.96582277e-02 -2.88738400e-01
-1.27381408e+00 1.36878335e+00 6.64389193e-01 -9.73290876e-02
1.24451613e+00 1.14695057e-01 8.90912771e-01 -6.44696236e-04
2.33365476e-01 -1.00582027e+00 8.59730363e-01 2.58687705e-01
1.00308275e+00 -1.00111318e+00 -3.70407790e-01 -3.97152007e-01
-4.31558669e-01 1.26920950e+00 3.20880085e-01 3.54428411e-01
8.58595729e-01 5.97596586e-01 1.97349787e-01 2.25057662e-01
-9.28223193e-01 -1.85267746e-01 5.41555703e-01 1.00649559e+00
2.10672200e-01 -3.95705074e-01 6.60951138e-01 5.19595146e-01
-4.94328082e-01 -1.31012678e-01 7.60532320e-02 4.75554883e-01
-4.36148584e-01 -1.03645658e+00 -6.49027407e-01 -3.76422286e-01
-1.95420682e-01 -1.89594701e-01 -2.65351415e-01 4.71120417e-01
-5.14979511e-02 1.21266115e+00 4.45554197e-01 -4.28113818e-01
-1.60716604e-02 2.56738037e-01 5.17651141e-01 -2.77830482e-01
-2.54058033e-01 4.76877630e-01 -2.39870235e-01 -1.13644671e+00
-4.36277300e-01 -5.28892338e-01 -1.46889997e+00 -5.23422956e-01
-9.21286717e-02 -6.54285327e-02 4.78274494e-01 5.37675560e-01
8.59900892e-01 3.49624038e-01 7.26289988e-01 -1.63256896e+00
-1.99219584e-01 -4.92273062e-01 -2.41794690e-01 2.92669386e-01
9.29825485e-01 -7.49242485e-01 -5.39713085e-01 4.91888642e-01]
|
[13.236777305603027, -0.40611204504966736]
|
55d87a84-1785-4d8d-bd1d-71feedbd978b
|
octave-deep-plane-sweeping-network-reducing
| null | null |
https://ieeexplore.ieee.org/document/8867874
|
https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8867874
|
Octave Deep Plane-Sweeping Network: Reducing Spatial Redundancy for Learning-Based Plane-Sweeping Stereo
|
In this paper, we propose the octave deep plane-sweeping network (OctDPSNet). OctDPSNet is a novel learning-based plane-sweeping stereo, which drastically reduces the required GPU memory and computation time while achieving a state-of-the-art depth estimation accuracy. Inspired by octave convolution, we divide image features into high and low spatial frequency features, and two cost volumes are generated from these using our proposed plane-sweeping module. To reduce spatial redundancy, the resolution of the cost volume from the low spatial frequency features is set to half that of the high spatial frequency features, which enables the memory consumption and computational cost to be reduced. After refinement, the two cost volumes are integrated into a final cost volume through our proposed pixel-wise “squeeze-and-excitation” based attention mechanism, and the depth maps are estimated from the final cost volume. We evaluate the proposed model on five datasets: SUN3D, RGB-D SLAM, MVS, Scenes11, and ETH3D. Our model outperforms previous methods on five datasets while drastically reducing the memory consumption and computational cost. Our source code is available at https://github.com/matsuren/octDPSNet.
|
['R. Komatsu', 'H. Asama', 'Y. Tamura', 'H. Fujii', 'A. Yamashita']
|
2019-10-14
| null | null | null |
ieee-access-2019-10
|
['stereo-depth-estimation']
|
['computer-vision']
|
[-6.62205219e-02 -2.15045467e-01 1.23177961e-01 -4.32508379e-01
-5.39270163e-01 -1.25209987e-01 5.00130296e-01 1.18087418e-01
-6.43131256e-01 4.83216435e-01 7.28357881e-02 1.85395237e-02
2.56150663e-02 -1.30454051e+00 -8.72039139e-01 -6.13105059e-01
1.97048366e-01 1.58294424e-01 5.91326296e-01 1.08408127e-02
3.93732131e-01 3.39948356e-01 -1.72087717e+00 2.47089654e-01
1.03444219e+00 1.39797163e+00 3.47448617e-01 2.98498690e-01
-7.32616857e-02 3.41771662e-01 -1.23995580e-01 4.78831604e-02
4.30456847e-01 1.75669752e-02 -4.92013216e-01 -1.83305979e-01
6.09822392e-01 -6.27164245e-01 -6.36824369e-01 1.03183055e+00
6.72433376e-01 1.80107743e-01 2.68317968e-01 -9.82787609e-01
-2.60479987e-01 -2.14838386e-02 -9.40429807e-01 1.15909152e-01
1.59705162e-01 2.90006936e-01 6.87826455e-01 -1.05716014e+00
6.28099203e-01 1.23917711e+00 4.85743016e-01 3.14707279e-01
-1.12138116e+00 -7.43407786e-01 1.44237190e-01 1.87438250e-01
-1.37578797e+00 -1.24422863e-01 8.34892273e-01 -4.10919428e-01
1.09269369e+00 9.01576579e-02 1.06957555e+00 6.21397376e-01
2.42097974e-01 7.02766776e-01 8.73581707e-01 -4.28196006e-02
2.34690815e-01 -2.94176817e-01 9.74465013e-02 8.66446137e-01
1.69864163e-01 9.90427285e-02 -8.20510805e-01 6.83647692e-02
1.24911213e+00 1.98864400e-01 -5.21362662e-01 -4.31539446e-01
-1.28852928e+00 8.29294026e-01 9.20922458e-01 -1.64758004e-02
-5.08744538e-01 3.04051220e-01 3.43330950e-01 -2.03910738e-01
6.21045113e-01 1.15573525e-01 -4.04602587e-01 -3.61595452e-02
-8.39320362e-01 4.48447019e-01 2.57586688e-01 7.97502160e-01
1.27097726e+00 -1.75944328e-01 -3.92922722e-02 7.92182148e-01
3.18322629e-01 4.39421624e-01 3.11135411e-01 -1.12940252e+00
5.63869178e-01 8.85232270e-01 -9.63665396e-02 -1.09692764e+00
-6.38410330e-01 -3.52150977e-01 -9.72705066e-01 2.49123469e-01
2.24229351e-01 -2.51954626e-02 -9.66894984e-01 1.44158411e+00
5.53964436e-01 1.31772429e-01 -3.11512828e-01 1.30407953e+00
9.34804380e-01 8.28788638e-01 -3.07209551e-01 1.34252563e-01
1.39826012e+00 -1.07906413e+00 -2.68812180e-01 -4.39065307e-01
5.65730691e-01 -4.25599992e-01 1.13839447e+00 2.77838230e-01
-8.51777971e-01 -4.68461484e-01 -1.15975380e+00 -6.53954208e-01
-1.09708264e-01 8.85696486e-02 8.86513710e-01 1.82533115e-02
-9.55013096e-01 6.58974588e-01 -8.98707271e-01 -2.61788905e-01
5.35405517e-01 3.92911494e-01 -3.14179569e-01 -2.68536031e-01
-8.52215528e-01 2.98546255e-01 2.92630553e-01 1.73235521e-01
-6.37294829e-01 -8.55509639e-01 -9.68283474e-01 1.90788507e-02
1.91752553e-01 -1.00810754e+00 1.00839138e+00 -3.16639423e-01
-1.23027921e+00 8.26840699e-01 -4.17865038e-01 -2.86943614e-01
4.73060429e-01 -3.69053006e-01 2.13959664e-01 1.31706834e-01
3.51785898e-01 9.65877652e-01 5.75159252e-01 -1.03691745e+00
-7.21164286e-01 -7.57969379e-01 8.26082975e-02 4.57818389e-01
-1.88088536e-01 -6.42497897e-01 -8.02495360e-01 -3.50469887e-01
7.20079064e-01 -7.23222911e-01 -1.94388643e-01 2.58489549e-01
-2.75226533e-01 -3.64639759e-02 6.66321754e-01 -4.28337008e-01
1.00190842e+00 -2.27368808e+00 2.28699028e-01 3.70665789e-02
6.05831027e-01 -1.65819645e-01 3.54726464e-02 1.29751503e-01
3.39764833e-01 -1.63813323e-01 -3.15956652e-01 -4.43219364e-01
-2.51424253e-01 8.12507868e-02 -1.37139997e-02 5.32904387e-01
1.01984628e-01 6.23365879e-01 -8.45522761e-01 -3.22529674e-01
5.43733001e-01 7.07526684e-01 -8.77232909e-01 5.95590100e-03
-2.33594060e-01 5.58188081e-01 -4.59060878e-01 6.43586993e-01
1.20232069e+00 -1.16304792e-01 -3.59316260e-01 -3.85741085e-01
-6.06422186e-01 5.13440490e-01 -9.86961365e-01 1.98379564e+00
-6.08212948e-01 6.02067232e-01 -2.21351981e-02 -5.21353066e-01
1.01482511e+00 -3.43287885e-01 4.33089584e-01 -1.09752321e+00
2.72341818e-01 3.29293430e-01 -3.25212359e-01 -2.77631760e-01
5.71235538e-01 1.78430021e-01 2.01063696e-03 7.32615739e-02
1.06562003e-01 -3.75004888e-01 -6.21505938e-02 4.58885245e-02
8.84621024e-01 2.79592633e-01 -6.02468289e-03 -2.80950516e-01
5.03694415e-01 9.94709358e-02 7.78155625e-01 2.84669071e-01
-5.05345277e-02 6.68586075e-01 4.99612778e-01 -7.15971351e-01
-9.65260386e-01 -8.91028404e-01 -2.38155842e-01 6.81449950e-01
6.08570218e-01 -4.80272979e-01 -8.34659874e-01 -2.24133298e-01
2.04496682e-01 2.92667747e-01 -6.18159413e-01 4.69372869e-02
-5.02048016e-01 -4.88440216e-01 1.45527005e-01 3.52725118e-01
1.11739755e+00 -8.71212482e-01 -1.03736949e+00 1.87910140e-01
-2.77488589e-01 -9.90470827e-01 -4.40344512e-01 1.38033688e-01
-9.95520651e-01 -9.15363610e-01 -6.87376976e-01 -6.74917042e-01
4.83003527e-01 5.48736811e-01 9.03460383e-01 -1.77156210e-01
-3.27392519e-01 -1.20913193e-01 -9.87650901e-02 -2.49430925e-01
6.33524060e-01 4.65389155e-03 -8.75497237e-02 -1.26383513e-01
4.47248220e-01 -8.77053499e-01 -1.09223437e+00 9.51542184e-02
-6.34046614e-01 5.31625748e-01 4.53890383e-01 7.98444629e-01
1.11344242e+00 -1.06824622e-01 -1.91300213e-01 -5.16588986e-01
1.17989480e-01 -3.00172627e-01 -9.22761202e-01 -5.52173734e-01
-1.97212934e-01 -1.23908482e-02 4.88937289e-01 -1.24806181e-01
-7.87617147e-01 1.34767622e-01 -2.26942003e-01 -6.10552728e-01
5.39230965e-02 2.81567693e-01 -8.69622529e-02 -2.07608834e-01
4.16066289e-01 4.16203916e-01 -2.63904959e-01 -5.67002535e-01
8.42571259e-02 3.25946927e-01 5.74600279e-01 -3.06247473e-01
5.66326261e-01 8.24152112e-01 1.00400992e-01 -8.23675990e-01
-7.45230019e-01 -3.10770959e-01 -5.56710958e-01 -1.48342386e-01
9.05126095e-01 -1.11536980e+00 -9.05317128e-01 8.81082118e-01
-1.40094209e+00 -3.34310800e-01 -1.15221322e-01 6.18454158e-01
-4.70501900e-01 1.11923084e-01 -4.93936718e-01 -4.94215697e-01
-4.95546341e-01 -1.16484571e+00 1.40851641e+00 4.07787889e-01
7.88540021e-02 -4.75534618e-01 -3.69529799e-02 2.21525937e-01
1.82689816e-01 4.38085198e-01 6.83921635e-01 2.88429767e-01
-1.05673194e+00 1.55791149e-01 -6.30926192e-01 8.37245397e-03
-1.84872866e-01 -2.31723458e-01 -1.02521718e+00 -2.03765303e-01
1.09193563e-01 -1.79606676e-01 1.24744630e+00 6.54582441e-01
1.43124866e+00 -9.66166705e-02 -1.87514424e-01 1.39401388e+00
1.66624594e+00 6.77545220e-02 5.90046942e-01 6.36903048e-01
1.03336072e+00 5.51336050e-01 6.16267920e-01 6.82373226e-01
7.58900046e-01 5.75661421e-01 8.37503314e-01 -1.30630434e-01
-3.28352563e-02 -4.31761563e-01 -5.39889038e-02 6.39680743e-01
-1.49360985e-01 -5.22887111e-02 -9.45693672e-01 6.11822486e-01
-1.77836215e+00 -6.38665080e-01 -3.12226176e-01 2.23627782e+00
4.59708691e-01 1.72591537e-01 -7.06402883e-02 4.52737138e-02
3.90469968e-01 4.60011542e-01 -8.24980021e-01 -1.84965432e-01
-1.22374021e-01 1.93500638e-01 5.30929923e-01 5.87740421e-01
-1.08542848e+00 8.72269690e-01 4.31747532e+00 6.96272254e-01
-1.30615520e+00 1.51464073e-02 5.91203928e-01 -5.30134618e-01
-3.77269149e-01 -2.57258769e-02 -8.09375644e-01 5.12675881e-01
3.16480011e-01 -1.40400678e-01 3.11374605e-01 9.19622898e-01
3.27764690e-01 -5.59195280e-01 -1.03118503e+00 1.40090132e+00
-2.10750595e-01 -1.45505571e+00 -2.79011987e-02 1.95583552e-01
5.39369285e-01 4.82067883e-01 6.04977235e-02 4.24957797e-02
-1.23318672e-01 -8.08593869e-01 8.54842663e-01 2.10822091e-01
8.58874440e-01 -9.72987056e-01 6.73894227e-01 2.88285494e-01
-1.42007697e+00 -5.70188090e-02 -5.44802129e-01 -2.13450015e-01
5.24159558e-02 1.02894473e+00 -2.18544245e-01 4.04345512e-01
1.17445254e+00 8.22631896e-01 -2.27213472e-01 9.74528432e-01
-4.74937633e-02 9.45570394e-02 -6.22171640e-01 -1.10353660e-02
4.37219381e-01 -3.54824215e-01 4.48383301e-01 8.38267565e-01
5.11157155e-01 2.82305181e-01 -3.02886516e-02 1.06133747e+00
-1.87149167e-01 -1.62401468e-01 -4.75463450e-01 4.45846975e-01
6.06883168e-01 1.11986327e+00 -5.99455953e-01 -2.70505428e-01
-3.33842814e-01 1.11799312e+00 3.94670486e-01 8.51431340e-02
-7.49187350e-01 -6.97189689e-01 8.11562777e-01 3.85346115e-01
2.40814939e-01 -2.86610842e-01 -6.13350511e-01 -1.23406744e+00
3.90592754e-01 -3.72521907e-01 1.82273313e-02 -7.92588711e-01
-7.55784154e-01 7.30832338e-01 -2.93741167e-01 -1.25926900e+00
2.06508443e-01 -3.47498089e-01 -4.54451203e-01 9.95581150e-01
-1.69616246e+00 -8.76524091e-01 -1.09599030e+00 5.76017499e-01
4.52664942e-01 3.38035345e-01 4.54513669e-01 3.81286502e-01
-5.22511959e-01 3.35542798e-01 -1.46072969e-01 -1.28237858e-01
4.63470459e-01 -8.67678821e-01 6.12632334e-01 5.61543584e-01
-3.05605352e-01 4.16139364e-01 3.18466783e-01 -5.18103600e-01
-1.39636767e+00 -1.20712078e+00 7.10358858e-01 4.92051393e-02
2.29497492e-01 -7.07248330e-01 -9.21721697e-01 3.22983176e-01
-2.51504987e-01 2.97135353e-01 -1.20131597e-02 -3.24247628e-01
-4.43824738e-01 -2.67234206e-01 -1.19323242e+00 5.32872319e-01
1.34530294e+00 -4.94581074e-01 -2.20316514e-01 1.79243833e-01
9.62689638e-01 -6.69494152e-01 -6.66051149e-01 5.14231443e-01
7.36805916e-01 -1.49287236e+00 7.91906059e-01 3.77430916e-01
7.58160114e-01 -4.31380391e-01 -1.64896145e-01 -1.17423689e+00
-4.40213203e-01 -3.18047374e-01 -2.10863538e-02 8.07396829e-01
7.61352628e-02 -7.89841950e-01 9.97409225e-01 1.53806508e-01
-3.72827232e-01 -1.19686294e+00 -1.13553762e+00 -4.39012110e-01
-1.45163551e-01 -4.36771601e-01 6.83246970e-01 7.42140889e-01
-2.95316279e-01 1.96666092e-01 -8.36543143e-02 1.92424878e-01
8.71124804e-01 1.75870940e-01 7.78726101e-01 -1.10643995e+00
-1.66636109e-01 -3.17990422e-01 -5.36637187e-01 -1.63605022e+00
-3.32559347e-01 -5.58629632e-01 -8.21733922e-02 -1.65997958e+00
1.29927009e-01 -5.11107445e-01 1.35548532e-01 4.30116147e-01
7.05708265e-02 4.48395878e-01 1.36385471e-01 1.32180423e-01
-1.98076755e-01 8.82064104e-01 1.44473720e+00 -3.80354375e-02
-3.47707570e-01 -4.01281416e-01 -3.28113735e-01 1.02650213e+00
7.20926046e-01 -2.91231155e-01 -2.86279678e-01 -9.50507045e-01
4.14870903e-02 1.34674340e-01 5.71917772e-01 -1.35526645e+00
3.03882062e-01 -1.19581781e-02 5.84661961e-01 -1.03678489e+00
7.31110871e-01 -6.31077170e-01 -6.91740289e-02 6.06280565e-01
1.96389377e-01 -1.98689923e-01 4.08176035e-01 3.16415817e-01
-2.96811789e-01 4.13001150e-01 9.94410813e-01 -1.24181837e-01
-8.73859227e-01 6.98039770e-01 2.29182746e-02 -2.19118640e-01
9.54879761e-01 -4.05796051e-01 -3.88401806e-01 -1.16141364e-01
-3.08554232e-01 3.81601810e-01 6.82449341e-01 2.46313572e-01
8.98907661e-01 -1.37441218e+00 -5.48811674e-01 6.72969460e-01
1.22669071e-01 8.77846122e-01 7.93433547e-01 7.93138325e-01
-9.56133485e-01 5.25161266e-01 -4.43060338e-01 -9.70390618e-01
-9.27840054e-01 1.81226313e-01 3.70809942e-01 -1.09587938e-01
-8.63865793e-01 1.31835127e+00 6.67335629e-01 -5.14015019e-01
8.10675398e-02 -4.37826395e-01 4.98432443e-02 -1.55831426e-01
3.85935456e-01 4.57466334e-01 8.27658735e-03 -5.10808110e-01
-5.13666987e-01 1.13120985e+00 9.64379534e-02 -4.98976633e-02
1.59478223e+00 -8.21026489e-02 -2.68019915e-01 2.85177290e-01
1.32558453e+00 -1.59132928e-01 -1.57153845e+00 -1.66400015e-01
-5.71316361e-01 -7.77551770e-01 3.82551193e-01 -2.44177833e-01
-1.23808277e+00 1.12616634e+00 6.15874529e-01 -3.15246850e-01
1.43072057e+00 -6.90092444e-02 1.09044671e+00 2.37837620e-02
6.75573349e-01 -7.74547338e-01 -3.00200321e-02 7.54330993e-01
8.20352077e-01 -1.13240170e+00 4.21609953e-02 -6.68505788e-01
-3.15403163e-01 9.22191978e-01 9.85819697e-01 -5.29405296e-01
5.31679690e-01 3.23464841e-01 -2.82023847e-01 -3.07816148e-01
-5.79113424e-01 -1.56846344e-02 6.98451549e-02 5.03950655e-01
1.94420353e-01 -5.89242540e-02 -3.52207661e-01 5.77520370e-01
-4.07111943e-01 3.07175815e-02 2.94994563e-01 7.26536155e-01
-4.73761022e-01 -7.04628110e-01 -2.44390190e-01 4.58929479e-01
-7.29722232e-02 -2.47096315e-01 -7.11944252e-02 6.19651556e-01
5.44796228e-01 6.07428074e-01 5.72512388e-01 -8.04796517e-01
4.72435415e-01 -4.42173421e-01 4.95645404e-01 -5.09575009e-01
-2.41500810e-01 1.51064157e-01 -9.29660872e-02 -1.11784196e+00
-1.23346612e-01 -3.74773413e-01 -1.43596292e+00 -5.21239281e-01
-4.76076864e-02 -1.56407520e-01 7.22160161e-01 4.90463078e-01
6.38073087e-01 5.81796944e-01 6.55815959e-01 -1.38662469e+00
4.19037417e-02 -9.28107440e-01 -5.23671627e-01 1.62460431e-02
4.89143789e-01 -8.08412194e-01 -2.94004858e-01 -4.09558535e-01]
|
[8.84377384185791, -2.4800033569335938]
|
c9450109-2b8a-4b71-af5a-3a6c4ada0bb5
|
biphasic-learning-of-gans-for-high-resolution
|
1904.06624
| null |
http://arxiv.org/abs/1904.06624v1
|
http://arxiv.org/pdf/1904.06624v1.pdf
|
Biphasic Learning of GANs for High-Resolution Image-to-Image Translation
|
Despite that the performance of image-to-image translation has been
significantly improved by recent progress in generative models, current methods
still suffer from severe degradation in training stability and sample quality
when applied to the high-resolution situation. In this work, we present a novel
training framework for GANs, namely biphasic learning, to achieve
image-to-image translation in multiple visual domains at $1024^2$ resolution.
Our core idea is to design an adjustable objective function that varies across
training phases. Within the biphasic learning framework, we propose a novel
inherited adversarial loss to achieve the enhancement of model capacity and
stabilize the training phase transition. Furthermore, we introduce a
perceptual-level consistency loss through mutual information estimation and
maximization. To verify the superiority of the proposed method, we apply it to
a wide range of face-related synthesis tasks and conduct experiments on
multiple large-scale datasets. Through comprehensive quantitative analyses, we
demonstrate that our method significantly outperforms existing methods.
|
['Huaibo Huang', 'Yi Li', 'Jingtuo Liu', 'Zhenan Sun', 'Jie Cao', 'Ran He']
|
2019-04-14
| null | null | null | null |
['mutual-information-estimation']
|
['methodology']
|
[ 6.85471535e-01 -1.42815694e-01 -1.08258761e-01 -2.77223229e-01
-1.10216010e+00 -4.27083254e-01 6.16631746e-01 -6.81269169e-01
-2.92072892e-02 9.06949341e-01 -3.72082628e-02 3.85181941e-02
6.73957616e-02 -6.31669521e-01 -7.64038861e-01 -7.89350450e-01
3.41508150e-01 -8.38594213e-02 -1.25248030e-01 8.75466689e-02
-3.15673761e-02 3.77876997e-01 -1.19098508e+00 1.71443418e-01
1.01217568e+00 1.06296968e+00 2.10801974e-01 3.52148622e-01
3.62449050e-01 8.35317373e-01 -6.12172902e-01 -5.60981095e-01
5.32424033e-01 -9.24641907e-01 -5.15718460e-01 3.09749991e-01
5.60284972e-01 -3.45433921e-01 -3.58162403e-01 1.05520260e+00
8.69621754e-01 -3.60226370e-02 5.84174931e-01 -1.19220066e+00
-7.40141571e-01 2.45673016e-01 -7.41435289e-01 1.02149785e-01
1.40326932e-01 2.89666086e-01 7.79252529e-01 -1.04119003e+00
5.95959246e-01 1.16856229e+00 5.47261894e-01 7.65716612e-01
-1.53960478e+00 -9.55959976e-01 -6.00215793e-02 -5.23062460e-02
-1.47240579e+00 -7.37114251e-01 9.92477477e-01 -2.78069824e-01
4.56549168e-01 1.09987371e-01 3.06557000e-01 1.14712024e+00
1.36491284e-01 4.86063212e-01 1.53913176e+00 -3.70605469e-01
1.88995358e-02 4.98055816e-02 -8.36785138e-01 7.89176464e-01
2.89218072e-02 2.72742271e-01 -7.02702940e-01 1.98588625e-01
1.26706123e+00 -3.36720973e-01 -3.99547368e-01 -2.79013753e-01
-9.70796883e-01 6.45667493e-01 4.75350738e-01 1.60022587e-01
-2.32855439e-01 3.18024337e-01 1.61441285e-02 2.84482092e-01
5.12234509e-01 3.54975194e-01 4.84236181e-02 5.61284162e-02
-1.21275294e+00 1.18154384e-01 2.32679084e-01 8.93897533e-01
6.05887234e-01 5.48176467e-01 -4.86017078e-01 1.05838442e+00
3.17850500e-01 6.23953879e-01 3.32445800e-01 -1.15218151e+00
4.85582858e-01 1.97623968e-01 1.20976567e-02 -9.81226921e-01
3.82887907e-02 -7.75867164e-01 -1.22831345e+00 2.36016080e-01
1.43735215e-01 -1.38868794e-01 -9.33546245e-01 2.13849878e+00
3.16238612e-01 3.07496458e-01 -6.35634288e-02 8.63216698e-01
6.78307533e-01 5.54161787e-01 8.39844048e-02 -5.21026909e-01
1.10903704e+00 -9.77082551e-01 -8.93692493e-01 -2.45974839e-01
-7.02959970e-02 -9.92576778e-01 1.12709486e+00 2.71484405e-01
-1.46540010e+00 -8.62316608e-01 -1.17168260e+00 3.42392959e-02
3.41145247e-01 2.51866847e-01 4.26816970e-01 6.57689393e-01
-1.05065262e+00 4.75513220e-01 -6.78113759e-01 -5.41672595e-02
7.10007668e-01 3.19568306e-01 -1.52253091e-01 -1.10187449e-01
-1.02598453e+00 6.72724545e-01 1.80734415e-02 7.60188922e-02
-1.10725737e+00 -7.85263002e-01 -7.23562896e-01 -1.24425590e-02
5.64599149e-02 -1.03178942e+00 1.02103567e+00 -1.12057865e+00
-1.92523718e+00 8.40594590e-01 -1.87793657e-01 -2.63941735e-01
6.79864645e-01 -7.13873729e-02 -3.22781771e-01 1.65072545e-01
2.49484126e-02 8.06584656e-01 1.28433537e+00 -1.60856056e+00
-3.38912100e-01 -1.87106252e-01 -2.31349543e-01 2.31838182e-01
-4.82516438e-01 -4.29558270e-02 -5.43222606e-01 -1.05638015e+00
-8.64652277e-04 -9.51868236e-01 -1.74900368e-01 1.52936333e-03
-1.75015643e-01 4.10881758e-01 6.49625540e-01 -6.08826220e-01
1.20249844e+00 -2.02461123e+00 2.94067293e-01 -1.12984702e-02
2.02913642e-01 3.10435236e-01 -2.33284578e-01 1.96441010e-01
-6.62414997e-04 2.72493213e-02 -4.81958658e-01 -5.68019390e-01
-2.08989158e-01 8.70829001e-02 -4.00452375e-01 4.23662663e-01
4.76269335e-01 1.09358823e+00 -5.72592020e-01 -6.54734910e-01
1.12306260e-01 8.05521548e-01 -7.73324549e-01 4.69568968e-01
5.60917072e-02 9.11880493e-01 -3.41311604e-01 6.18123114e-01
8.69830906e-01 -4.65516418e-01 1.68178901e-01 -2.38218069e-01
2.33424306e-01 -1.92534313e-01 -8.20101082e-01 1.76716244e+00
-6.37348354e-01 6.34947896e-01 1.62284330e-01 -9.83373940e-01
9.10410881e-01 3.49278092e-01 5.32020032e-01 -9.51449573e-01
7.33908489e-02 1.70042321e-01 -8.50164071e-02 -1.67071342e-01
2.83521682e-01 -4.51183110e-01 1.87048554e-01 1.21252917e-01
8.28654990e-02 -4.03119355e-01 -1.01788975e-01 -5.01302816e-02
8.63745451e-01 1.49108753e-01 1.35015801e-01 7.83596467e-03
5.85747302e-01 -4.85719770e-01 6.26851976e-01 4.41375494e-01
-2.00005233e-01 8.58350337e-01 2.71218330e-01 -2.14709658e-02
-1.26956904e+00 -1.21765149e+00 -2.75292933e-01 6.89482868e-01
8.44248757e-02 -2.15616480e-01 -9.74290371e-01 -5.08628249e-01
-2.32292697e-01 3.95629019e-01 -3.50978851e-01 -2.23675713e-01
-6.82514489e-01 -9.20848370e-01 5.41752756e-01 3.33589941e-01
9.41923857e-01 -8.60458076e-01 -2.73004770e-01 4.52116467e-02
-3.26289624e-01 -1.41515338e+00 -6.46378577e-01 -3.37824911e-01
-8.21461618e-01 -6.17574215e-01 -9.44578350e-01 -8.59220922e-01
7.36261070e-01 8.08939412e-02 1.20530260e+00 -1.53402820e-01
-2.68027723e-01 2.43033960e-01 -1.24477141e-01 -2.08711475e-02
-6.09140456e-01 -8.18623602e-03 -4.40772250e-03 8.40148628e-02
-4.33644533e-01 -8.05125713e-01 -9.30959761e-01 5.35767972e-01
-9.50089455e-01 2.24768281e-01 7.76218414e-01 1.10877085e+00
7.62177706e-01 1.38311818e-01 8.68634224e-01 -6.13001585e-01
6.33051574e-01 -1.71797678e-01 -6.51183486e-01 1.01584159e-01
-8.94294918e-01 -6.27067452e-03 6.88090503e-01 -4.13714260e-01
-1.23878205e+00 -1.51153104e-02 -1.81465551e-01 -5.91067791e-01
3.24266762e-01 2.57232249e-01 -4.04850572e-01 -3.90964597e-01
5.38899362e-01 4.58838671e-01 5.66988289e-02 -1.71808913e-01
4.40561265e-01 4.42475080e-01 6.31985724e-01 -6.85033262e-01
1.10999703e+00 4.12017912e-01 1.23869918e-01 -5.00004709e-01
-7.37757742e-01 1.83688506e-01 -2.16572687e-01 -3.26779544e-01
6.67765439e-01 -1.22251260e+00 -6.93819344e-01 5.90667427e-01
-8.92765820e-01 -5.00379145e-01 -2.45949581e-01 3.03568572e-01
-8.99602592e-01 2.84791321e-01 -5.89030087e-01 -6.67542994e-01
-4.83839631e-01 -1.22491491e+00 1.13805676e+00 1.97318375e-01
2.40180254e-01 -8.33815157e-01 9.01731923e-02 5.21181762e-01
5.38626373e-01 3.22834969e-01 5.50046742e-01 1.57868057e-01
-7.99805939e-01 1.52927414e-01 -3.04323643e-01 6.53609693e-01
3.31354171e-01 -1.38132408e-01 -8.48693073e-01 -6.97547734e-01
1.76616803e-01 -4.68156099e-01 8.81116569e-01 3.78079802e-01
1.40850961e+00 -3.85703236e-01 -1.66425467e-01 1.01192451e+00
1.45702422e+00 1.70673549e-01 9.54756916e-01 2.82547846e-02
6.44950807e-01 2.50572205e-01 5.37114263e-01 4.07544702e-01
1.31653160e-01 1.01427889e+00 2.88033575e-01 -3.60270888e-01
-6.03900790e-01 -4.91994977e-01 4.92675066e-01 8.66972625e-01
-1.50217667e-01 -2.58166283e-01 -4.88468736e-01 3.04088861e-01
-1.55634534e+00 -1.09296465e+00 5.25407970e-01 2.07392669e+00
1.22976601e+00 3.89607809e-02 -7.69905979e-03 -9.70625058e-02
7.02573299e-01 2.95385242e-01 -5.85402966e-01 -8.38643909e-02
-1.79597676e-01 5.28129518e-01 3.56264263e-01 3.99189800e-01
-9.38565016e-01 1.03457570e+00 6.99902344e+00 1.25218093e+00
-1.35272849e+00 1.40378579e-01 9.93538141e-01 -1.25307515e-01
-3.52440983e-01 -2.34639242e-01 -5.96661031e-01 4.81199950e-01
7.44754076e-01 -1.57542348e-01 6.02328598e-01 6.48229420e-01
2.42994487e-01 2.25659773e-01 -9.01576519e-01 1.10157549e+00
4.11839038e-02 -1.39525568e+00 1.05659813e-01 2.75132302e-02
1.19126558e+00 -4.03145403e-01 6.63659036e-01 7.63378367e-02
1.56306043e-01 -1.34432220e+00 5.35152972e-01 2.75426060e-01
1.39645863e+00 -7.26627827e-01 1.93184018e-01 1.73026100e-02
-1.11639416e+00 3.06343939e-02 -1.85446605e-01 2.85048991e-01
1.67576790e-01 4.82424110e-01 -7.05587447e-01 6.74547672e-01
4.71772790e-01 6.52666271e-01 -4.42781299e-01 6.69514954e-01
-2.93035626e-01 5.36931217e-01 1.00962855e-01 5.05969584e-01
-1.08141996e-01 -2.88250595e-01 3.86704445e-01 8.98914158e-01
6.04760408e-01 1.66971460e-02 -1.74007371e-01 1.13036549e+00
-4.03120607e-01 3.66527550e-02 -5.52305043e-01 2.63508037e-03
5.57705104e-01 1.22652471e+00 -3.93880695e-01 -5.62614612e-02
-2.52483577e-01 1.22105968e+00 2.39069536e-01 4.92746562e-01
-1.21565855e+00 -1.07095756e-01 6.25236571e-01 1.88787758e-01
4.42041487e-01 -2.59478211e-01 -2.61018097e-01 -1.12689316e+00
2.20374271e-01 -1.20717669e+00 7.39182672e-03 -6.77086055e-01
-1.19692147e+00 6.83132470e-01 -1.48660496e-01 -1.31853843e+00
-2.74330020e-01 -2.34181523e-01 -4.22849029e-01 8.51348221e-01
-1.66355765e+00 -1.41200686e+00 -4.36868995e-01 7.57045925e-01
4.08138752e-01 -3.20154697e-01 6.09008789e-01 6.32107794e-01
-5.89813292e-01 1.10042393e+00 1.76108718e-01 -9.92952008e-03
8.89428616e-01 -7.88840592e-01 2.83241600e-01 9.76216793e-01
1.28785849e-01 4.60935116e-01 6.01005793e-01 -4.57396477e-01
-1.17627418e+00 -1.25373650e+00 3.22667450e-01 -2.02108622e-01
3.04199368e-01 -3.60092789e-01 -6.84099674e-01 6.16719961e-01
4.40841168e-01 1.61044657e-01 4.94627267e-01 -4.89025205e-01
-4.04865056e-01 -4.66083676e-01 -1.33850574e+00 6.66232049e-01
1.24393737e+00 -4.76133823e-01 5.96694946e-02 1.64188877e-01
7.32342303e-01 -5.88636816e-01 -1.07390749e+00 8.07302296e-01
4.94513065e-01 -9.21152771e-01 1.18325830e+00 -2.73437619e-01
8.36347759e-01 -3.42810541e-01 -2.49616995e-01 -1.35442626e+00
-3.18120480e-01 -9.52583790e-01 -7.00776204e-02 1.38111913e+00
2.04626396e-01 -5.08106232e-01 7.85732806e-01 1.52011052e-01
-9.15155187e-02 -8.32717419e-01 -1.11306369e+00 -7.42255747e-01
2.15414882e-01 -1.12643624e-02 4.47149009e-01 7.49404013e-01
-3.83613557e-01 3.56252640e-01 -8.98048222e-01 4.69881035e-02
9.09458160e-01 1.56220913e-01 8.80328178e-01 -5.54010093e-01
-4.64658856e-01 -4.26831871e-01 -2.38224387e-01 -1.24526417e+00
1.72574833e-01 -6.21564090e-01 6.74527064e-02 -1.19364405e+00
3.63696426e-01 -3.78915071e-01 -2.09716260e-01 1.52762249e-01
-3.15740168e-01 6.80823088e-01 2.60702282e-01 3.48353624e-01
-2.95445472e-01 9.64321911e-01 1.87721288e+00 -8.19152370e-02
8.14410001e-02 -1.50015920e-01 -7.21799374e-01 3.36870223e-01
7.35165834e-01 -2.98649579e-01 -6.79712772e-01 -4.26505297e-01
5.52001595e-03 1.70761168e-01 3.78801495e-01 -1.20633781e+00
-1.31429732e-01 -2.53024578e-01 5.60952365e-01 -8.70526209e-02
5.03872633e-01 -7.20487297e-01 3.78046364e-01 3.73680919e-01
-3.47653359e-01 -5.90119548e-02 1.04109176e-01 5.02512991e-01
-3.95866871e-01 3.66522342e-01 1.34697187e+00 1.85336292e-01
-2.97583491e-01 5.80597818e-01 2.14178801e-01 2.18116716e-01
1.00280964e+00 -3.97628583e-02 -3.34105581e-01 -5.38233221e-01
-4.60251838e-01 -7.37097934e-02 6.33942187e-01 3.25499147e-01
6.67716205e-01 -1.73355258e+00 -8.15253258e-01 3.86488050e-01
-2.20910981e-02 -1.04602240e-01 3.15355390e-01 8.15507233e-01
-3.51075739e-01 2.58836538e-01 -4.15388703e-01 -6.14274204e-01
-1.16598344e+00 4.60624814e-01 3.97751421e-01 -3.82096976e-01
-4.11505133e-01 7.84633934e-01 3.73474926e-01 -1.01485439e-01
-5.15026413e-02 5.18693328e-02 1.38171747e-01 -3.40786129e-01
3.08785290e-01 4.21747603e-02 -9.74577591e-02 -6.62801266e-01
-2.02586651e-01 8.09619963e-01 9.65110287e-02 -2.45776653e-01
1.15647829e+00 -2.59220511e-01 6.61090389e-02 -2.56437138e-02
1.15587056e+00 9.12839398e-02 -1.72220290e+00 -2.43125319e-01
-5.68689406e-01 -8.52169514e-01 -5.23242950e-02 -6.32974625e-01
-1.51085675e+00 7.37935960e-01 7.66945660e-01 -2.68424582e-02
1.53684843e+00 -1.63738161e-01 9.05501068e-01 -2.23759547e-01
2.62179703e-01 -9.27107871e-01 6.13032997e-01 4.09823507e-02
9.45229352e-01 -1.34640396e+00 6.74781129e-02 -4.50025558e-01
-6.81854546e-01 6.00332439e-01 6.88118160e-01 1.54556623e-02
3.84410381e-01 3.04305524e-01 1.57408044e-03 1.61186486e-01
-6.38969958e-01 6.84372149e-03 3.51332396e-01 6.38936698e-01
5.46634674e-01 -1.63571313e-01 -3.39624524e-01 2.56999403e-01
-2.03178406e-01 2.02495977e-01 2.10903719e-01 4.67088461e-01
-8.42732713e-02 -1.27233231e+00 -2.16635510e-01 8.16306844e-02
-7.01252282e-01 -9.17115584e-02 -2.23608688e-02 6.82926357e-01
-5.88026876e-03 9.26014125e-01 -2.16876920e-02 -3.64486814e-01
2.05280855e-01 -2.23025575e-01 9.51295853e-01 -3.16249728e-01
-3.29974830e-01 2.28038654e-01 -2.48798057e-01 -5.01694739e-01
-7.47637749e-01 -4.65636998e-01 -9.28670883e-01 -4.81663555e-01
-1.79015890e-01 -1.45341605e-01 4.40171450e-01 5.91151416e-01
5.25630176e-01 7.45124996e-01 9.98623550e-01 -6.48597419e-01
-5.67039430e-01 -7.95861900e-01 -3.68117452e-01 5.38676798e-01
2.28998944e-01 -6.42126977e-01 -2.44265690e-01 3.09874684e-01]
|
[11.731344223022461, -0.5497217774391174]
|
0fef911e-d3c1-4f25-875b-3baeb5fc5c5b
|
rethinking-two-consensuses-of-the
|
2212.00399
| null |
https://arxiv.org/abs/2212.00399v1
|
https://arxiv.org/pdf/2212.00399v1.pdf
|
Rethinking Two Consensuses of the Transferability in Deep Learning
|
Deep transfer learning (DTL) has formed a long-term quest toward enabling deep neural networks (DNNs) to reuse historical experiences as efficiently as humans. This ability is named knowledge transferability. A commonly used paradigm for DTL is firstly learning general knowledge (pre-training) and then reusing (fine-tuning) them for a specific target task. There are two consensuses of transferability of pre-trained DNNs: (1) a larger domain gap between pre-training and downstream data brings lower transferability; (2) the transferability gradually decreases from lower layers (near input) to higher layers (near output). However, these consensuses were basically drawn from the experiments based on natural images, which limits their scope of application. This work aims to study and complement them from a broader perspective by proposing a method to measure the transferability of pre-trained DNN parameters. Our experiments on twelve diverse image classification datasets get similar conclusions to the previous consensuses. More importantly, two new findings are presented, i.e., (1) in addition to the domain gap, a larger data amount and huge dataset diversity of downstream target task also prohibit the transferability; (2) although the lower layers learn basic image features, they are usually not the most transferable layers due to their domain sensitivity.
|
['Li Liu', 'Chris Ding', 'Jingxian Li', 'Yixiong Chen']
|
2022-12-01
| null | null | null | null |
['general-knowledge']
|
['miscellaneous']
|
[ 7.24684820e-02 3.26568969e-02 -8.86818171e-02 -3.98356527e-01
-1.11723086e-02 -6.80413604e-01 7.81572163e-01 -1.07717961e-01
-7.81427324e-01 1.12201309e+00 -1.10586286e-01 -1.30955443e-01
-3.54750425e-01 -9.98176038e-01 -7.87819564e-01 -7.92569458e-01
1.00013964e-01 6.37973174e-02 5.60066402e-01 -4.87129629e-01
2.12880239e-01 6.80242896e-01 -1.55779994e+00 2.58617729e-01
1.05604827e+00 9.53781307e-01 5.39887130e-01 4.95865047e-02
-1.06106043e-01 6.12042189e-01 -6.78676844e-01 -3.81945103e-01
2.28163660e-01 -5.56470454e-01 -1.06817830e+00 -3.66211414e-01
2.13877007e-01 -2.83044338e-01 -3.29949558e-01 1.20707500e+00
4.19627100e-01 2.00787231e-01 7.38362432e-01 -1.24095154e+00
-1.27079570e+00 6.00076199e-01 -6.74126372e-02 4.04668301e-01
-2.18745008e-01 2.03031093e-01 7.43871152e-01 -7.09060609e-01
5.84947646e-01 1.05539632e+00 7.28169262e-01 7.66495049e-01
-8.55266392e-01 -7.77215958e-01 2.73290306e-01 5.06176472e-01
-1.37066031e+00 -1.43387988e-01 7.05747247e-01 -5.18141448e-01
7.73089528e-01 -8.52111876e-02 6.43809319e-01 1.38033104e+00
1.20342821e-01 6.88275099e-01 1.32884002e+00 -3.27394783e-01
1.93585858e-01 5.98726928e-01 -9.50265303e-02 2.26198688e-01
3.77978712e-01 2.06461892e-01 -4.86691624e-01 3.60558778e-01
1.05733609e+00 3.31994072e-02 -4.85698342e-01 -3.94340217e-01
-1.20096147e+00 7.96583354e-01 8.78232300e-01 9.56211329e-01
-2.51507819e-01 -2.68252313e-01 5.75068712e-01 7.41538465e-01
1.81989998e-01 4.89101499e-01 -7.32161939e-01 2.52248198e-02
-4.86340016e-01 -1.31179675e-01 5.86696446e-01 1.00578010e+00
1.11629879e+00 1.32605925e-01 -1.60092369e-01 8.98407936e-01
-3.84750664e-02 3.09716135e-01 9.89767015e-01 -4.46410239e-01
5.13595223e-01 6.04323566e-01 -1.85861140e-01 -8.64467740e-01
-1.73495024e-01 -5.86968541e-01 -1.05591416e+00 3.30493480e-01
5.42757392e-01 -2.23152474e-01 -7.63553202e-01 2.05967975e+00
-1.84242159e-01 -1.71411521e-04 3.74330193e-01 8.46367061e-01
8.53697240e-01 6.96344495e-01 3.56867582e-01 1.52011355e-02
1.03975523e+00 -9.09263492e-01 -4.08371717e-01 -2.87815273e-01
6.09660864e-01 -4.54721689e-01 1.11543775e+00 2.48159170e-01
-7.49182105e-01 -1.01715720e+00 -1.18192112e+00 7.36205652e-02
-8.36956918e-01 3.06035355e-02 4.39083636e-01 4.57146794e-01
-1.25441027e+00 9.10702169e-01 -2.77822703e-01 -6.74053669e-01
4.87414658e-01 2.82421052e-01 -4.28230971e-01 -4.96985801e-02
-1.64622736e+00 1.24738610e+00 1.02938247e+00 -1.72234736e-02
-8.11516106e-01 -8.76600504e-01 -4.43015188e-01 2.15786129e-01
1.24177605e-01 -7.45050192e-01 1.05295622e+00 -1.58330643e+00
-1.72847629e+00 8.95420611e-01 3.98562789e-01 -2.52017885e-01
5.34532666e-01 -2.99815208e-01 -5.24103940e-01 -9.27695557e-02
-5.58255054e-02 7.44213045e-01 7.73265183e-01 -1.34627926e+00
-7.70968318e-01 -2.51780093e-01 1.34118661e-01 2.64589489e-01
-9.14898634e-01 -3.89874041e-01 -1.18128844e-01 -8.52049232e-01
-3.10456324e-02 -7.88247526e-01 2.33978152e-01 4.85452674e-02
8.30294639e-02 -3.77939790e-01 9.06679869e-01 -4.47630852e-01
1.09196556e+00 -2.23243523e+00 2.82673746e-01 5.61662428e-02
1.15846828e-01 8.87121439e-01 -2.89991945e-01 4.77043301e-01
-2.48079807e-01 1.99177284e-02 -9.70052034e-02 3.87645602e-01
-1.40407324e-01 3.14678788e-01 -3.92818421e-01 -1.71880685e-02
1.99284911e-01 1.11351371e+00 -1.11095071e+00 -1.59008130e-01
1.03329591e-01 2.92534679e-01 -1.69813737e-01 7.26628602e-02
4.04570587e-02 6.22724831e-01 -3.96559834e-01 1.59576505e-01
6.63517714e-01 -1.08864710e-01 7.41771013e-02 -5.14535844e-01
-1.30442679e-01 -1.95094496e-02 -8.87723625e-01 1.55607224e+00
-4.43146706e-01 8.34925950e-01 -3.58424842e-01 -1.40874457e+00
1.13592172e+00 2.44411141e-01 1.76106766e-01 -9.96489584e-01
1.50346950e-01 3.35197449e-01 2.85828918e-01 -5.48210502e-01
1.88474178e-01 -3.88058841e-01 1.29518092e-01 2.95961946e-01
4.99285370e-01 9.57878903e-02 -1.07157819e-01 -3.05440903e-01
7.25478709e-01 1.53890833e-01 4.60879385e-01 -5.75290680e-01
6.66990697e-01 -1.78435177e-01 7.16262043e-01 5.71783602e-01
-3.49361181e-01 2.51055449e-01 2.60962218e-01 -5.35099804e-01
-9.88407135e-01 -1.31792951e+00 -1.36844859e-01 1.31221855e+00
2.15246707e-01 9.00226384e-02 -6.86641216e-01 -7.63838112e-01
-5.07827550e-02 5.13748825e-01 -7.63306558e-01 -4.98432577e-01
-5.45021057e-01 -5.01358569e-01 6.85881734e-01 8.24898064e-01
1.20052803e+00 -1.33692980e+00 -6.06801689e-01 1.09331153e-01
-5.02107330e-02 -9.77496028e-01 -1.19476907e-01 1.37895033e-01
-1.29574156e+00 -8.17966402e-01 -1.18443727e+00 -1.13805342e+00
7.64715672e-01 2.47397274e-01 9.99562144e-01 6.33446872e-02
1.08359404e-01 2.73513138e-01 -4.58231568e-01 -5.37225544e-01
-4.91809696e-01 3.40304017e-01 1.97064355e-01 -1.37432113e-01
5.10302305e-01 -7.58730769e-01 -6.51246667e-01 5.93859375e-01
-1.02708852e+00 3.95319005e-03 1.07466936e+00 8.26193154e-01
3.03260177e-01 1.81046054e-01 1.00822985e+00 -7.29106128e-01
7.34435916e-01 -5.56887090e-01 -3.39177608e-01 5.19578993e-01
-6.43356204e-01 5.23931906e-02 8.04282188e-01 -6.92230046e-01
-1.33069527e+00 -4.14056927e-01 1.63605958e-01 -2.69529700e-01
-3.18902105e-01 5.32618999e-01 -2.28857204e-01 -1.07420072e-01
9.25537169e-01 3.74136597e-01 -3.21692564e-02 -4.67855334e-01
2.44238481e-01 5.21681130e-01 5.32650650e-01 -7.48328149e-01
9.24298346e-01 2.13745743e-01 -3.43397766e-01 -7.34774530e-01
-8.52395117e-01 -2.53554378e-02 -9.65196252e-01 -1.30248681e-01
7.88110852e-01 -7.10366428e-01 -5.50177097e-01 7.71491170e-01
-9.54617977e-01 -5.66826940e-01 -3.27004254e-01 6.51607752e-01
-3.78004670e-01 1.69992983e-01 -4.84432995e-01 -2.91612059e-01
-1.04327276e-01 -8.44409883e-01 2.07038790e-01 4.36719596e-01
-3.60544445e-03 -1.38633144e+00 -1.06720507e-01 -2.46343855e-02
6.70059979e-01 1.52856722e-01 1.18138063e+00 -6.77403092e-01
-3.51282954e-01 3.39858234e-01 -4.91883725e-01 8.78832400e-01
4.84979779e-01 -2.75125325e-01 -1.18577516e+00 -4.33469981e-01
5.22588268e-02 -4.22030658e-01 8.13234925e-01 1.29842147e-01
1.10909486e+00 -2.13631555e-01 -3.61559868e-01 3.91070098e-01
1.51349890e+00 4.50532258e-01 8.38168442e-01 7.71794617e-01
4.81303185e-01 7.01322079e-01 4.91425902e-01 4.62937960e-03
7.23847896e-02 4.54405218e-01 9.23528448e-02 -1.14704706e-01
-3.08695942e-01 -2.98717856e-01 4.02941883e-01 1.06541443e+00
-6.07531965e-01 -6.34117499e-02 -9.63546932e-01 6.50734723e-01
-1.87505591e+00 -7.64288545e-01 2.85468847e-01 2.22198033e+00
1.01299238e+00 2.26415128e-01 -3.85959670e-02 3.22047137e-02
7.57683992e-01 -2.16667488e-01 -9.00100887e-01 -2.77559161e-01
-2.68699378e-01 1.60426959e-01 3.72518331e-01 3.56714912e-02
-8.04656208e-01 1.05826735e+00 5.87809038e+00 9.83719587e-01
-1.43710434e+00 1.03666157e-01 4.59996790e-01 3.29562187e-01
-1.29731357e-01 -2.66782612e-01 -7.57657528e-01 3.77611965e-01
7.33727813e-01 -5.30785799e-01 2.31253311e-01 8.29788744e-01
-4.27946419e-01 1.39782384e-01 -1.40925968e+00 1.04834843e+00
-1.33777291e-01 -1.11223042e+00 4.60113257e-01 -5.91592975e-02
9.45564747e-01 3.17333266e-02 2.27517888e-01 6.69992030e-01
8.05790797e-02 -1.05969417e+00 6.59347355e-01 6.19327366e-01
6.84919655e-01 -7.41264343e-01 8.41196597e-01 4.24770862e-01
-8.96279037e-01 -2.71444798e-01 -6.23601198e-01 -1.78850487e-01
-2.39426196e-01 4.11880940e-01 -7.50550210e-01 6.74091399e-01
1.05914295e+00 8.00216615e-01 -6.01428628e-01 8.49338114e-01
-5.11946797e-01 4.48147595e-01 6.72006607e-02 -6.11925647e-02
2.85188049e-01 -1.78570420e-01 1.61537603e-01 1.20346057e+00
4.77842718e-01 1.93579718e-02 -4.64026093e-01 9.71751392e-01
-2.23551258e-01 -4.41610329e-02 -8.25061440e-01 1.26196846e-01
5.51622033e-01 9.24297392e-01 -6.01539195e-01 -3.01778376e-01
-5.36908090e-01 1.05163705e+00 5.00455022e-01 6.54330969e-01
-6.80256426e-01 -4.83380318e-01 6.63732767e-01 -1.13459893e-01
4.81012493e-01 -9.97795984e-02 -3.59224886e-01 -9.63424861e-01
8.64640437e-03 -6.48262024e-01 2.90543735e-01 -6.71273530e-01
-1.73465812e+00 7.34030902e-01 2.56421238e-01 -1.34620380e+00
-5.90596907e-02 -9.30036545e-01 -5.01251996e-01 9.37927902e-01
-1.74976468e+00 -9.55708981e-01 -5.33271134e-01 8.89425755e-01
3.30750585e-01 -4.73764688e-01 8.72135401e-01 3.79981697e-01
-3.08288485e-01 8.08536887e-01 2.27941051e-01 3.52480441e-01
8.14170122e-01 -8.58647525e-01 8.51979107e-02 4.61017609e-01
-1.28072754e-01 6.92293465e-01 2.47536182e-01 -3.09404314e-01
-8.25394392e-01 -1.03684366e+00 6.76331639e-01 -4.02040958e-01
5.74630141e-01 -2.28946134e-01 -1.32735920e+00 7.24665403e-01
2.95924425e-01 -3.41238260e-01 6.88409746e-01 1.23766504e-01
-4.12368298e-01 -4.84016120e-01 -1.10473883e+00 6.74836457e-01
1.12584066e+00 -5.04602432e-01 -1.07047379e+00 -8.17495361e-02
5.77713311e-01 9.18119308e-03 -1.13665950e+00 5.31833649e-01
4.86642867e-01 -1.16095448e+00 8.16590309e-01 -5.17484665e-01
3.52057606e-01 -1.37985066e-01 4.59047519e-02 -1.64331841e+00
-5.24054825e-01 -2.02149466e-01 2.21692026e-01 1.49712408e+00
3.04693103e-01 -1.06813240e+00 4.70526665e-01 4.53696251e-01
-3.26894484e-02 -5.71831942e-01 -8.25849116e-01 -1.31489027e+00
6.21612430e-01 -6.33905549e-03 6.97409272e-01 1.41967118e+00
-2.15219989e-01 4.06562090e-01 2.64077764e-02 -1.06065899e-01
1.84973314e-01 -1.35209858e-01 4.72040892e-01 -1.52456975e+00
2.73922160e-02 -5.97135365e-01 -5.00803113e-01 -1.02962029e+00
8.23690817e-02 -9.54142570e-01 -1.62298903e-01 -1.33470678e+00
1.58011720e-01 -7.46510446e-01 -8.00896049e-01 7.53389955e-01
-5.49128912e-02 1.14569897e-02 2.21078798e-01 4.21200901e-01
-2.99092382e-01 6.32700205e-01 1.58501387e+00 1.55532565e-02
-2.12333024e-01 -1.51453093e-02 -8.01373303e-01 7.85692573e-01
1.05787218e+00 -3.63477856e-01 -8.78311336e-01 -7.29720235e-01
2.37264261e-01 -4.89046395e-01 3.86583745e-01 -1.23192120e+00
2.22385466e-01 -2.43406266e-01 5.63985944e-01 -1.69392116e-02
3.60727943e-02 -8.91921461e-01 -1.31142482e-01 6.71496093e-01
-7.58356154e-02 -1.16984181e-01 5.60419679e-01 3.73240560e-01
-3.88468325e-01 -5.04533410e-01 9.28865731e-01 -2.62204766e-01
-1.43130946e+00 2.32107997e-01 -2.83604175e-01 1.47793055e-01
1.16734922e+00 -5.21874487e-01 -4.30605531e-01 -2.02296972e-01
-5.98413765e-01 -8.04136619e-02 2.66554207e-01 7.85546660e-01
4.40310270e-01 -1.51494527e+00 -6.72076225e-01 1.60934135e-01
2.15214267e-01 -3.08333542e-02 3.19211453e-01 5.37230909e-01
-3.48749489e-01 5.24966061e-01 -9.48619664e-01 -5.19334435e-01
-7.28968799e-01 6.15895391e-01 4.24781084e-01 -2.24400274e-02
-5.42599976e-01 8.80560994e-01 6.42482996e-01 -4.95282322e-01
1.51747584e-01 -3.87364507e-01 -2.77252644e-01 2.04377800e-01
4.11365598e-01 3.80129546e-01 1.59153536e-01 -3.45665812e-01
-2.69036978e-01 6.81864679e-01 -2.23471642e-01 1.64753705e-01
1.49757671e+00 5.86142801e-02 8.45112896e-04 5.18422544e-01
1.22975004e+00 -6.01825237e-01 -1.39427340e+00 -4.87338066e-01
4.71472442e-02 -1.67697951e-01 -2.25648433e-01 -8.86214852e-01
-1.09007120e+00 9.44711804e-01 6.58996463e-01 1.01364940e-01
1.34297049e+00 -1.22154772e-04 6.19772494e-01 7.31850684e-01
4.49724704e-01 -1.45452523e+00 3.31857651e-01 7.02436507e-01
1.09465873e+00 -1.12589562e+00 -2.30582789e-01 -1.06524929e-01
-7.06317961e-01 1.29217172e+00 9.37602341e-01 -9.05965827e-03
6.34792805e-01 -2.70034969e-01 9.85110775e-02 -9.67416819e-03
-3.77330333e-01 -4.65051644e-02 3.66228819e-01 9.25263107e-01
2.99864262e-01 -2.58777589e-01 -1.64615244e-01 6.78572536e-01
-2.49480635e-01 2.96531409e-01 4.17966582e-02 8.23786855e-01
-5.14036775e-01 -1.08788753e+00 -1.50251552e-01 1.15599059e-01
-6.72541931e-03 7.63018429e-02 -3.00397128e-01 1.17021286e+00
5.19981802e-01 6.93902731e-01 -8.59823078e-02 -4.10993934e-01
4.50762480e-01 -2.51904726e-02 5.77667117e-01 -5.59375644e-01
-5.10164440e-01 -7.86919355e-01 -3.16239744e-01 -2.28722557e-01
-6.11636996e-01 -2.22320244e-01 -1.07599199e+00 -2.45743930e-01
-9.88506749e-02 1.60404891e-01 5.56660771e-01 1.04383063e+00
3.55774760e-01 6.58435762e-01 4.40592110e-01 -6.80038631e-01
-6.62017047e-01 -1.08648455e+00 -5.52831233e-01 5.76363742e-01
1.34721816e-01 -7.91565359e-01 -3.20953995e-01 2.35140622e-01]
|
[9.86319637298584, 2.93998646736145]
|
187d88c6-3fc9-47bc-a060-31eabd5ff118
|
step-by-step-loss-goes-very-far-multi-step
|
2302.05120
| null |
https://arxiv.org/abs/2302.05120v1
|
https://arxiv.org/pdf/2302.05120v1.pdf
|
Step by Step Loss Goes Very Far: Multi-Step Quantization for Adversarial Text Attacks
|
We propose a novel gradient-based attack against transformer-based language models that searches for an adversarial example in a continuous space of token probabilities. Our algorithm mitigates the gap between adversarial loss for continuous and discrete text representations by performing multi-step quantization in a quantization-compensation loop. Experiments show that our method significantly outperforms other approaches on various natural language processing (NLP) tasks.
|
['Klaudia Bałazy', 'Piotr Gaiński']
|
2023-02-10
| null | null | null | null |
['adversarial-text']
|
['adversarial']
|
[ 2.74004996e-01 -4.34108265e-02 -2.59523749e-01 -3.22672427e-01
-1.69276524e+00 -7.81729698e-01 8.99076283e-01 3.70767742e-01
-7.24499106e-01 5.63533127e-01 1.29354283e-01 -7.37107873e-01
4.83900845e-01 -8.78577232e-01 -7.86366820e-01 -3.31303120e-01
3.80690396e-02 4.23195451e-01 3.16977680e-01 -3.04257780e-01
1.69253796e-01 6.56674325e-01 -5.09707570e-01 4.46997643e-01
6.93076491e-01 8.19818139e-01 -4.01891321e-01 7.57156014e-01
-2.96616018e-01 9.40424919e-01 -9.92475390e-01 -1.02345634e+00
5.30951440e-01 -1.47613928e-01 -7.07974434e-01 -4.51114476e-01
5.35846055e-01 -3.30524862e-01 -8.50221157e-01 1.53038919e+00
4.49591160e-01 -1.66144557e-02 5.57554364e-01 -1.35965943e+00
-7.61139691e-01 9.16889966e-01 -3.44025761e-01 3.42113495e-01
4.79478657e-01 2.48331591e-01 1.17405248e+00 -9.25327003e-01
2.70231783e-01 1.96908152e+00 6.37203217e-01 5.93100429e-01
-1.12011743e+00 -9.33788896e-01 1.19200304e-01 2.57549345e-01
-1.44163203e+00 -2.24958465e-01 7.85731494e-01 3.21298055e-02
1.28584611e+00 1.23007119e-01 2.95589026e-02 1.19391418e+00
8.36632371e-01 1.03542554e+00 9.40123737e-01 -5.84801972e-01
2.59739280e-01 9.51719359e-02 -4.52272892e-01 9.71201241e-01
-2.31111012e-02 7.36817345e-02 -5.56787848e-01 -7.71257401e-01
5.58711171e-01 -1.45979792e-01 8.84961933e-02 -8.88358951e-02
-9.27238822e-01 1.43283582e+00 3.14732522e-01 1.84203893e-01
-2.50382274e-01 5.86078465e-01 8.80737603e-01 4.86625761e-01
4.72793579e-01 6.10219598e-01 -4.35227931e-01 -1.07042834e-01
-9.97028351e-01 4.17652786e-01 7.20475912e-01 7.94861972e-01
4.28470224e-01 4.54094112e-01 -3.52073848e-01 4.64871317e-01
4.29219574e-01 8.67610157e-01 6.95528984e-01 -6.35185361e-01
8.61926675e-01 -3.43235731e-02 3.81809240e-03 -9.51157570e-01
3.31279039e-01 6.84314668e-02 -6.74433589e-01 2.33144611e-01
3.24060977e-01 -2.81589270e-01 -8.95483375e-01 1.60453606e+00
1.02205537e-01 1.93809241e-01 1.46670967e-01 6.07729793e-01
7.56913424e-02 8.66867661e-01 4.57309335e-01 6.33079410e-02
1.27547824e+00 -7.17960358e-01 -7.37869918e-01 -3.63644361e-01
4.03717756e-01 -7.33607113e-01 1.08508718e+00 3.04663241e-01
-1.18680215e+00 -3.53620797e-02 -9.01105285e-01 -1.79209769e-01
-5.64231873e-01 -2.96456158e-01 4.15694326e-01 9.39575374e-01
-7.89036870e-01 7.04307914e-01 -8.73263061e-01 3.06610107e-01
4.92808431e-01 1.90939039e-01 -1.71623260e-01 1.29463643e-01
-1.95595968e+00 9.37277794e-01 3.79597992e-01 -1.57839939e-01
-9.54638958e-01 -6.51283443e-01 -1.26315570e+00 1.91318527e-01
1.03195049e-01 -1.63864449e-01 1.29192376e+00 -6.55807197e-01
-1.71442342e+00 7.59413719e-01 -1.86783075e-01 -1.38095021e+00
6.85077786e-01 -4.34089035e-01 -5.40076256e-01 2.20656753e-01
-1.43459290e-01 3.40560228e-01 1.34968054e+00 -8.33202660e-01
-4.53081220e-01 -1.46640629e-01 -1.14753693e-02 3.55228446e-02
-4.76476282e-01 3.67887497e-01 4.89528067e-02 -1.09395003e+00
-3.14138085e-01 -5.64710915e-01 -4.90728050e-01 3.24328005e-01
-5.49728036e-01 -4.08999264e-01 9.35655236e-01 -7.34139919e-01
1.15115941e+00 -1.93901956e+00 -3.52223545e-01 3.99759799e-01
3.97865362e-02 7.59260893e-01 -1.50195643e-01 5.71946979e-01
2.08192706e-01 3.98977339e-01 -2.59830207e-01 -4.96359348e-01
4.08156246e-01 1.56592473e-01 -1.18466341e+00 5.55055022e-01
3.56226385e-01 8.83494079e-01 -1.14894748e+00 -7.66281605e-01
3.42386425e-01 5.81365764e-01 -5.11832476e-01 2.01956794e-01
-2.82181025e-01 -3.25588405e-01 -3.43186438e-01 5.55137157e-01
6.72600746e-01 3.92810255e-01 1.40158590e-02 2.11775213e-01
4.38761145e-01 8.10333073e-01 -8.89173329e-01 1.58541310e+00
-7.12926030e-01 6.90245032e-01 5.18469922e-02 -9.42041397e-01
9.93168831e-01 5.14477909e-01 4.09716889e-02 -4.27422941e-01
1.14824727e-01 3.56624871e-02 -4.23575789e-01 8.12395513e-02
6.32489860e-01 -4.83511269e-01 -6.53332353e-01 2.63570189e-01
5.63781932e-02 -6.63422763e-01 -2.73827821e-01 4.97200012e-01
1.15878749e+00 -3.31302792e-01 1.75823525e-01 1.24436595e-01
6.73766017e-01 -2.65476525e-01 5.03300428e-01 1.20379698e+00
-6.88523412e-01 1.62108168e-01 5.69890857e-01 -3.45280588e-01
-9.92194355e-01 -1.50549722e+00 7.41788745e-02 1.10620415e+00
-2.71089792e-01 -6.78557515e-01 -5.46693504e-01 -1.15149140e+00
3.13243061e-01 8.53394806e-01 -3.66548240e-01 -4.83741105e-01
-5.50881505e-01 -2.37733871e-01 1.36427355e+00 4.30769205e-01
4.50914860e-01 -1.05795479e+00 -1.11213863e-01 3.16851705e-01
2.63885371e-02 -1.26448321e+00 -8.00606549e-01 2.82629997e-01
-6.06828511e-01 -6.28470600e-01 -1.13944493e-01 -8.69750202e-01
2.51853466e-01 -2.11895853e-01 1.02604043e+00 -3.40141565e-01
-2.36721128e-01 9.61045027e-02 3.05326488e-02 -6.34224534e-01
-9.54884171e-01 -1.18942201e-01 1.49134040e-01 -3.10717434e-01
5.97417951e-01 -2.48675585e-01 -1.45017236e-01 -2.93395549e-01
-8.52414310e-01 -7.50009418e-01 2.81436652e-01 1.11544704e+00
5.58543801e-01 2.61576414e-01 5.01440823e-01 -7.30292559e-01
1.09396565e+00 -3.64464581e-01 -9.38395679e-01 3.69074047e-01
-4.96177614e-01 4.89499271e-01 1.27622783e+00 -5.81682622e-01
-7.03965068e-01 9.86333564e-02 -3.76526922e-01 -9.65311885e-01
6.08665831e-02 1.65484592e-01 -2.27520078e-01 -2.12598056e-01
7.72663295e-01 4.54656988e-01 -2.88398117e-01 -1.69979468e-01
6.94095969e-01 6.73022568e-01 7.97905266e-01 -7.33493209e-01
1.34616148e+00 2.28190750e-01 -1.50057539e-01 -3.86117220e-01
-6.54282093e-01 -1.27579808e-01 -3.87085110e-01 2.74866253e-01
4.04312462e-01 -7.79360592e-01 -6.84742093e-01 4.13747072e-01
-1.20312011e+00 -1.32771701e-01 -4.27056879e-01 3.03323567e-01
-4.49434459e-01 6.78060234e-01 -1.10963857e+00 -1.03912473e+00
-8.56733918e-01 -1.06906152e+00 1.08415258e+00 -1.67415574e-01
9.58599616e-03 -1.24677336e+00 1.12080790e-01 -1.09757699e-01
3.32710326e-01 4.98849265e-02 7.40993679e-01 -1.07702541e+00
-1.74696460e-01 -8.10145378e-01 7.97683001e-02 5.38650692e-01
-2.44621566e-04 -3.61642241e-02 -8.43479514e-01 -3.81463856e-01
4.59785052e-02 -8.21803451e-01 9.48321521e-01 -1.07303046e-01
1.38801813e+00 -9.87138569e-01 4.76768911e-02 5.11108518e-01
1.35820270e+00 -6.70495182e-02 6.13046765e-01 2.45793581e-01
6.77774727e-01 -8.70847628e-02 5.99058986e-01 4.69583273e-01
1.00866303e-01 3.12628299e-01 4.04969692e-01 3.53192613e-02
3.26471984e-01 -7.65940189e-01 8.07456791e-01 2.71152616e-01
9.88203049e-01 -2.92275965e-01 -8.53163123e-01 6.42431676e-01
-1.40722585e+00 -1.25382447e+00 6.53257191e-01 2.00672388e+00
1.41586721e+00 6.83082819e-01 -9.19314176e-02 2.68603563e-01
7.51699328e-01 4.88288224e-01 -6.25858188e-01 -1.21803045e+00
3.81477475e-02 7.71230578e-01 8.78081203e-01 7.74996817e-01
-1.56428814e+00 1.48768914e+00 8.02272034e+00 1.22154069e+00
-1.18077934e+00 -8.27375799e-03 3.83333117e-01 1.30379111e-01
-3.44641685e-01 -3.01986873e-01 -6.99544787e-01 3.35565329e-01
1.30093884e+00 -8.56058717e-01 3.98971975e-01 1.08761871e+00
6.87313848e-04 5.54172873e-01 -9.36124623e-01 6.97452545e-01
-1.85513683e-02 -1.20107496e+00 5.45106411e-01 -3.01000655e-01
3.88891280e-01 1.14052080e-01 4.51742351e-01 3.54812443e-01
1.13533878e+00 -1.13986003e+00 7.60956764e-01 6.74988925e-02
8.13527107e-01 -1.06247973e+00 4.60243613e-01 1.57772332e-01
-1.16318977e+00 -1.53112207e-02 -5.20843506e-01 2.74528921e-01
-1.09481081e-01 3.68443847e-01 -1.32252288e+00 1.81187183e-01
1.76136851e-01 2.97731102e-01 -2.48722166e-01 5.00740469e-01
-5.53797245e-01 8.98319840e-01 -2.98393726e-01 -7.43716434e-02
5.13636470e-01 3.65396976e-01 6.21224046e-01 1.73126757e+00
-2.34114692e-01 -1.90650091e-01 3.78892839e-01 9.49345887e-01
-5.54961324e-01 -9.27499384e-02 -8.38442445e-01 -2.17558905e-01
1.03940785e+00 9.43165779e-01 -1.66129231e-01 -4.24885452e-01
-9.26444083e-02 1.17883754e+00 2.81325638e-01 6.81790486e-02
-1.01941419e+00 -1.01114213e+00 8.08810711e-01 -3.76250118e-01
4.06407356e-01 -3.25378329e-01 -1.62970692e-01 -1.12539172e+00
1.20495316e-02 -9.88680363e-01 6.10596716e-01 -3.33527207e-01
-1.49333060e+00 3.45790297e-01 -2.30307668e-01 -1.02676451e+00
-6.28938794e-01 -4.48862374e-01 -6.45165265e-01 1.07168555e+00
-1.59289098e+00 -8.48521233e-01 7.82948196e-01 1.10948920e+00
5.15322983e-01 -3.71608078e-01 1.09180999e+00 -8.03092122e-03
-2.29161739e-01 1.51812935e+00 1.82382315e-01 7.65595555e-01
6.55736923e-01 -1.50123382e+00 1.15579844e+00 1.22427297e+00
4.36610401e-01 5.81534564e-01 7.18216062e-01 -4.29517508e-01
-1.40946472e+00 -1.36239457e+00 1.11998343e+00 -2.84647316e-01
1.15435791e+00 -6.87486291e-01 -9.82262611e-01 8.77494991e-01
1.18502796e-01 2.06545919e-01 4.29182380e-01 -4.74130243e-01
-1.10274339e+00 5.25179990e-02 -1.82013702e+00 5.60959935e-01
3.51545841e-01 -1.28019893e+00 -8.10952783e-01 7.12135732e-01
9.50314403e-01 -7.30682492e-01 -9.82838988e-01 -1.38613805e-01
2.67687410e-01 -2.49339655e-01 1.24254978e+00 -8.13208222e-01
3.07105809e-01 1.30749434e-01 -6.40897751e-02 -1.32204187e+00
2.17724778e-02 -1.09746242e+00 -2.29681924e-01 9.27749276e-01
3.21181953e-01 -8.19529057e-01 5.98490596e-01 6.02878869e-01
2.88384914e-01 -6.57559454e-01 -1.28278327e+00 -8.60182941e-01
7.80044436e-01 -5.76871037e-01 5.02425015e-01 7.54381061e-01
3.16841453e-01 -1.23880364e-01 -4.23934847e-01 3.70379269e-01
8.07065189e-01 -4.16443914e-01 3.12441409e-01 -5.30920744e-01
-1.62246034e-01 -3.44762415e-01 -7.06039071e-01 -8.58469844e-01
8.36508155e-01 -8.99940372e-01 2.03733966e-01 -1.05628204e+00
-3.97162229e-01 -2.56819665e-01 -4.74854618e-01 7.21572995e-01
-3.43422353e-01 1.02236196e-01 2.32053950e-01 -1.05325863e-01
-5.81984520e-01 5.83588123e-01 5.76391280e-01 -6.70979440e-01
2.38628820e-01 -8.62225443e-02 -4.48227406e-01 6.13902032e-01
9.05040264e-01 -8.25250924e-01 -1.60206184e-01 -4.34018373e-01
-5.06649986e-02 9.13448781e-02 3.01178366e-01 -8.13553691e-01
2.92594016e-01 -2.88338840e-01 2.21573681e-01 -4.70444053e-01
3.09470475e-01 -6.54079556e-01 -9.59094703e-01 7.57495999e-01
-1.02390063e+00 2.68239886e-01 3.07946533e-01 5.61353087e-01
-2.48063326e-01 -5.45380637e-02 9.76479352e-01 -1.06020689e-01
-2.41146639e-01 3.05719912e-01 -7.03257084e-01 4.58686888e-01
5.12674987e-01 5.59903741e-01 -9.97342840e-02 -4.61771399e-01
-5.69905639e-01 1.42095134e-01 1.50618717e-01 4.99302447e-01
9.12286580e-01 -1.35436130e+00 -9.24032032e-01 2.86187381e-01
-2.51762092e-01 -3.89161199e-01 -4.55720603e-01 -1.35953501e-01
-4.49483871e-01 5.66825330e-01 8.51631463e-02 -1.14171997e-01
-1.14374697e+00 7.01314807e-01 6.37280285e-01 -8.30764949e-01
-5.26050210e-01 1.11827922e+00 -3.67805600e-01 -3.01351011e-01
5.18628716e-01 -2.29869246e-01 3.28135431e-01 -6.42619967e-01
7.77154624e-01 3.00126821e-02 8.21094494e-03 -4.84333277e-01
-4.95641530e-01 -1.69364005e-01 -4.86658096e-01 -5.53195655e-01
6.50422573e-01 1.33841366e-01 5.36199883e-02 4.22537982e-01
1.52549934e+00 2.40154117e-01 -9.73521411e-01 -6.22928262e-01
1.65851712e-02 -6.51312113e-01 -7.50722364e-02 -5.22098064e-01
-8.23996186e-01 9.49816048e-01 5.28649926e-01 2.24568754e-01
9.02796626e-01 -3.56618822e-01 1.34723198e+00 8.02513599e-01
2.85737723e-01 -1.03968215e+00 4.67418768e-02 8.05729389e-01
7.09728837e-01 -9.02570903e-01 -8.47805515e-02 -9.07355770e-02
-6.78917527e-01 1.12907696e+00 2.93218642e-01 -6.12990320e-01
6.15370095e-01 5.87801337e-01 1.48368210e-01 2.82858133e-01
-8.52081478e-01 4.17691529e-01 -3.77799496e-02 8.01721931e-01
2.17717245e-01 1.74073175e-01 -1.01236656e-01 3.49203110e-01
-5.29019713e-01 -3.16959918e-01 4.04166698e-01 1.00573373e+00
-3.96972179e-01 -1.35806513e+00 -6.37389004e-01 1.40242040e-01
-1.15238094e+00 -5.97874165e-01 -3.65331799e-01 3.07537258e-01
-4.71108913e-01 1.03915524e+00 -3.77628207e-02 -4.52633560e-01
1.35766476e-01 3.54927510e-01 3.19854945e-01 -3.68257314e-01
-8.67495716e-01 -3.09991032e-01 -1.78290918e-01 -6.87901855e-01
1.97525471e-01 -4.80920970e-01 -1.36747527e+00 -5.43400586e-01
-2.34862834e-01 4.64206398e-01 6.48421884e-01 8.01046193e-01
1.59821257e-01 1.53885409e-01 1.21054804e+00 -5.34565091e-01
-1.46917129e+00 -6.26549423e-01 -4.41099703e-01 4.12776560e-01
5.16476989e-01 -7.72922486e-02 -6.90723777e-01 -1.76998660e-01]
|
[6.0290327072143555, 8.088109970092773]
|
5ef682dd-af47-465d-9596-11c8471b5f85
|
learning-the-regularization-in-dce-mr-image
|
2109.07548
| null |
https://arxiv.org/abs/2109.07548v2
|
https://arxiv.org/pdf/2109.07548v2.pdf
|
Learning the Regularization in DCE-MR Image Reconstruction for Functional Imaging of Kidneys
|
Kidney DCE-MRI aims at both qualitative assessment of kidney anatomy and quantitative assessment of kidney function by estimating the tracer kinetic (TK) model parameters. Accurate estimation of TK model parameters requires an accurate measurement of the arterial input function (AIF) with high temporal resolution. Accelerated imaging is used to achieve high temporal resolution, which yields under-sampling artifacts in the reconstructed images. Compressed sensing (CS) methods offer a variety of reconstruction options. Most commonly, sparsity of temporal differences is encouraged for regularization to reduce artifacts. Increasing regularization in CS methods removes the ambient artifacts but also over-smooths the signal temporally which reduces the parameter estimation accuracy. In this work, we propose a single image trained deep neural network to reduce MRI under-sampling artifacts without reducing the accuracy of functional imaging markers. Instead of regularizing with a penalty term in optimization, we promote regularization by generating images from a lower dimensional representation. In this manuscript we motivate and explain the lower dimensional input design. We compare our approach to CS reconstructions with multiple regularization weights. Proposed approach results in kidney biomarkers that are highly correlated with the ground truth markers estimated using the CS reconstruction which was optimized for functional analysis. At the same time, the proposed approach reduces the artifacts in the reconstructed images.
|
['Sila Kurugol', 'Onur Afacan', 'Cemre Ariyurek', 'Aziz Koçanaoğulları']
|
2021-09-15
| null | null | null | null |
['kidney-function']
|
['medical']
|
[ 4.35042650e-01 -2.04913169e-02 -3.39531898e-02 -5.29330194e-01
-5.90735912e-01 -3.06650609e-01 1.17416747e-01 1.66817322e-01
-3.91546994e-01 7.76896000e-01 3.44823658e-01 9.47269723e-02
-2.98438728e-01 -5.65441191e-01 -6.51830435e-01 -8.38482857e-01
-2.30194330e-01 2.75423646e-01 2.97167189e-02 2.92135835e-01
2.24108770e-01 6.67232156e-01 -7.61070073e-01 3.98100838e-02
1.29408801e+00 9.16433096e-01 4.95512605e-01 4.12392676e-01
-3.49960506e-01 1.09487724e+00 -3.62967104e-02 1.45510986e-01
4.70177114e-01 -6.29514515e-01 -4.85580772e-01 6.22180477e-02
4.73076314e-01 -3.91811758e-01 -3.07462662e-01 1.21324444e+00
6.12238288e-01 1.27382353e-02 6.88060820e-01 -5.73633909e-01
-5.21085799e-01 7.82364309e-01 -5.94558656e-01 3.64506781e-01
-1.87044844e-01 1.86349005e-01 -2.48552170e-02 -6.99033320e-01
5.84191263e-01 6.86921656e-01 8.01471233e-01 4.42526579e-01
-1.49242187e+00 -5.24156213e-01 -4.27937329e-01 -4.33040485e-02
-1.26062596e+00 -4.05032665e-01 8.50360096e-01 -6.70525372e-01
3.01506102e-01 3.66941959e-01 7.51746058e-01 3.20551038e-01
4.12094593e-01 3.92440051e-01 1.45471823e+00 -3.49632740e-01
-3.09062041e-02 7.09621608e-02 1.58115298e-01 7.81336069e-01
3.10053527e-01 1.22843616e-01 -7.90788047e-03 1.06574167e-02
1.15752339e+00 2.24466607e-01 -6.81532741e-01 -4.62585688e-01
-1.26919389e+00 5.62494040e-01 4.88325685e-01 5.12124300e-01
-8.59253407e-01 3.61144543e-01 4.69518125e-01 -8.49170536e-02
4.54514399e-02 2.98006833e-01 -1.72277987e-02 3.33229423e-01
-9.49830174e-01 -7.95363188e-02 3.21230710e-01 5.68782032e-01
4.98372525e-01 5.30015647e-01 -4.73018408e-01 7.93923497e-01
3.95889282e-01 4.62535471e-01 7.16116548e-01 -1.31830978e+00
1.33451045e-01 4.63781387e-01 4.29008119e-02 -6.58332527e-01
-3.73276770e-01 -7.31010735e-01 -1.21588528e+00 2.90672481e-01
5.49199402e-01 -7.42939860e-03 -1.26495302e+00 1.55405104e+00
2.90311337e-01 4.85469878e-01 -1.76838502e-01 1.29596925e+00
5.50439835e-01 4.36691582e-01 3.71981025e-01 -4.72453237e-01
1.28365922e+00 -7.29094088e-01 -1.00900352e+00 5.24605848e-02
4.22863662e-01 -6.89519286e-01 7.29314089e-01 8.70028809e-02
-1.20301247e+00 -3.90173525e-01 -1.04005277e+00 3.11195880e-01
2.53710061e-01 2.55956322e-01 5.65236688e-01 5.41321814e-01
-7.34389722e-01 9.07509625e-01 -1.21501708e+00 8.68605599e-02
4.06086206e-01 3.82792473e-01 -3.58150095e-01 -1.42880172e-01
-1.03461063e+00 1.21383274e+00 -5.22446446e-02 3.97477239e-01
-6.75235510e-01 -1.27871239e+00 -7.06204951e-01 -3.80848460e-02
-2.17738077e-01 -5.43253064e-01 7.42942035e-01 -7.93524921e-01
-1.69800293e+00 6.83662236e-01 6.51479512e-03 -4.90095377e-01
7.96227157e-01 1.37582555e-01 -2.19416931e-01 4.46221560e-01
7.76035935e-02 2.47540653e-01 6.93248153e-01 -1.09125125e+00
2.32786432e-01 -4.74824846e-01 -6.78907573e-01 9.14105400e-02
2.30952665e-01 -2.79461294e-01 -1.18299380e-01 -5.50085366e-01
6.54704988e-01 -9.60520327e-01 -7.01110959e-01 4.28694099e-01
-3.46919410e-02 6.92258358e-01 1.87962875e-01 -9.72793162e-01
9.64584053e-01 -1.95482337e+00 -2.58633971e-01 3.88521612e-01
3.18614811e-01 8.90120566e-02 2.77098328e-01 -5.39579839e-02
-2.61445701e-01 -1.53116554e-01 -6.86026990e-01 7.42489025e-02
-6.18855119e-01 2.48078242e-01 2.20085323e-01 1.12804663e+00
-1.86317474e-01 7.39377439e-01 -7.44748592e-01 -7.43484259e-01
5.59635878e-01 6.60684526e-01 -4.05296773e-01 1.86620355e-01
2.83318311e-01 1.10514295e+00 -4.88110662e-01 2.51378775e-01
1.11570632e+00 -2.46249065e-01 2.73241878e-01 -6.56654179e-01
-4.29365963e-01 -5.39152846e-02 -1.29486978e+00 1.87931228e+00
-4.20807123e-01 2.34283179e-01 4.46882933e-01 -1.30536449e+00
8.43853116e-01 4.53972548e-01 1.15876389e+00 -9.26397204e-01
3.77509301e-03 4.42913324e-01 2.37479493e-01 -8.66766870e-01
-2.89386481e-01 -5.36081672e-01 5.74111104e-01 1.35378599e-01
-1.74969211e-01 1.56318486e-01 -1.25979409e-01 -5.80219645e-03
8.51826847e-01 1.16879769e-01 1.65692687e-01 -8.12223196e-01
7.85660744e-01 -3.62923592e-02 6.12994850e-01 5.73022544e-01
-4.43531185e-01 5.91310263e-01 1.15568131e-01 -4.89863425e-01
-1.18063521e+00 -1.03332150e+00 -7.45688736e-01 7.80843645e-02
2.81584561e-01 2.59453624e-01 -4.33784723e-01 -2.28240639e-01
2.40578040e-01 3.15436840e-01 -6.89007580e-01 1.45916075e-01
-1.02407932e+00 -8.60538602e-01 4.54222083e-01 3.42065662e-01
4.60662931e-01 -6.32797122e-01 -7.76549578e-01 5.07772148e-01
-1.21308111e-01 -7.60638714e-01 -4.26923364e-01 2.58655161e-01
-1.39640880e+00 -8.22677910e-01 -1.08294356e+00 -5.99600315e-01
8.66615474e-01 -2.60804575e-02 7.90967286e-01 1.05924778e-01
-5.37256062e-01 -7.11015016e-02 1.55526489e-01 -1.20926075e-01
-4.18672264e-01 -6.42438591e-01 -2.08062440e-01 2.76338905e-02
-2.35741705e-01 -6.67186022e-01 -1.15996110e+00 -1.13018332e-02
-8.37291479e-01 6.79568350e-02 6.76403165e-01 7.60748088e-01
7.26338863e-01 -5.14250100e-01 5.45718372e-01 -9.92013037e-01
2.50582874e-01 -4.18760389e-01 -7.43391871e-01 2.46170878e-01
-8.99768770e-01 5.94117939e-01 5.48482537e-01 -4.74777341e-01
-1.18739426e+00 2.29982018e-01 3.75612751e-02 -5.45970023e-01
9.39276069e-02 5.71457624e-01 7.02698350e-01 -4.42812383e-01
8.36185336e-01 4.34335619e-01 3.78190786e-01 -4.10649627e-01
5.94274700e-02 1.62037358e-01 4.84577209e-01 -6.86933696e-01
2.08610967e-01 5.61856568e-01 4.69627440e-01 -6.94833040e-01
-3.79163206e-01 -3.29205573e-01 -4.74211961e-01 -4.50955927e-01
7.94244170e-01 -7.60372221e-01 -6.74561083e-01 1.90833211e-01
-8.37538481e-01 -1.39056593e-01 -2.88787603e-01 1.14033389e+00
-4.11933362e-01 6.05124831e-01 -9.00253534e-01 -8.36425900e-01
-5.99738300e-01 -1.42186916e+00 3.90192866e-01 1.84398189e-01
1.68837368e-01 -8.23706925e-01 8.67714807e-02 1.89676285e-01
1.04809177e+00 6.39660299e-01 6.77931130e-01 3.89930271e-02
-7.87040651e-01 -2.45724525e-03 -2.34281391e-01 3.48504394e-01
3.94675210e-02 -4.03711051e-01 -5.94007790e-01 3.43085565e-02
4.16946054e-01 1.32414013e-01 7.61514544e-01 1.25968063e+00
1.25392377e+00 -1.12274811e-01 -3.46243847e-03 8.66822898e-01
2.04480743e+00 3.67834002e-01 7.08543479e-01 7.35760853e-02
5.41252434e-01 3.78387511e-01 2.88137525e-01 4.55018610e-01
-1.78386476e-02 3.43778998e-01 2.75190085e-01 -2.36551210e-01
-3.78749281e-01 4.23376672e-02 -1.10399537e-01 7.90424883e-01
-2.77573287e-01 3.31267387e-01 -9.54878867e-01 5.98405659e-01
-1.62469649e+00 -8.58364701e-01 -6.16467357e-01 2.26763320e+00
8.78290772e-01 -3.21606755e-01 -2.61849374e-01 -3.01020533e-01
7.97186255e-01 -8.30774754e-02 -4.78083372e-01 -1.16099797e-01
1.96860015e-01 3.92354608e-01 1.03642201e+00 8.69272947e-01
-6.11241758e-01 1.98095456e-01 6.01207304e+00 2.16706991e-01
-1.45203960e+00 2.23255008e-01 4.53215003e-01 1.02257460e-01
-3.49266559e-01 1.38104096e-01 -1.25406653e-01 6.40757203e-01
6.29106104e-01 -6.06000721e-02 4.68024671e-01 5.04431844e-01
5.52608609e-01 -3.48997474e-01 -7.35560298e-01 9.37494397e-01
-3.31090689e-01 -1.61240697e+00 -3.51790130e-01 -1.11315601e-01
5.02659500e-01 1.55214056e-01 -2.96268433e-01 -2.14115590e-01
-2.56191641e-01 -1.02929366e+00 3.66452217e-01 9.45508301e-01
9.46470320e-01 -3.38689983e-01 7.20784307e-01 1.90960199e-01
-1.01946437e+00 2.25416332e-01 -2.21850649e-01 1.97987109e-01
3.60904187e-01 8.93460572e-01 -7.23504424e-01 2.76583552e-01
2.77759284e-01 4.67712790e-01 -1.71543732e-01 1.20659578e+00
2.10376501e-01 4.94368970e-01 -3.05916369e-01 3.95764053e-01
1.74655337e-02 -8.03796291e-01 4.79127675e-01 1.17869532e+00
2.63526887e-01 4.99716878e-01 1.52704865e-01 1.22318387e+00
1.84797212e-01 2.05493405e-01 -4.48810905e-01 2.22793356e-01
3.02035272e-01 1.05721045e+00 -6.91745937e-01 -4.31926966e-01
-2.76540458e-01 7.91054428e-01 -4.59993742e-02 3.90125543e-01
-7.10079670e-01 4.65036333e-02 1.41363531e-01 5.72356462e-01
-3.09226871e-01 -1.19711176e-01 -4.84103829e-01 -1.13283479e+00
-6.47796467e-02 -4.01543051e-01 1.74847662e-01 -4.62625355e-01
-1.06888974e+00 3.76587898e-01 -1.32181764e-01 -9.94437039e-01
8.48936290e-02 -4.20507640e-02 -4.95825112e-01 1.42557406e+00
-1.62655056e+00 -8.15197051e-01 -5.44967294e-01 4.66777146e-01
3.28869194e-01 4.28338200e-01 5.81146777e-01 7.70111561e-01
-3.74902576e-01 4.79273163e-02 6.69950247e-02 7.39099607e-02
7.42850900e-01 -1.28610861e+00 -5.49726546e-01 8.38946998e-01
-8.25134635e-01 8.32062900e-01 7.77757347e-01 -9.04137731e-01
-1.43266916e+00 -8.30645502e-01 4.38749492e-01 2.66570061e-01
3.74970496e-01 2.96510875e-01 -9.23484147e-01 5.45245826e-01
5.02189063e-02 7.25362301e-01 4.94846880e-01 -6.73824549e-01
3.12950194e-01 -1.96464926e-01 -1.62405324e+00 1.32603213e-01
4.55195516e-01 -1.83683261e-01 -1.83509171e-01 1.44658789e-01
1.65585086e-01 -6.02795124e-01 -1.34701860e+00 5.00251293e-01
8.02321672e-01 -7.56177545e-01 1.00592279e+00 -3.59521061e-01
3.98026377e-01 -5.56415498e-01 -6.58970773e-02 -9.40001488e-01
-5.30951500e-01 -2.15374082e-01 2.34695971e-02 7.01932907e-01
5.34473695e-02 -4.54158604e-01 1.02579272e+00 8.47479641e-01
-2.87615865e-01 -6.46530271e-01 -7.30496943e-01 -5.35636842e-01
-9.14517641e-02 1.47844143e-02 -6.92562084e-04 1.29926836e+00
-1.77833661e-01 -7.71822110e-02 -3.27809095e-01 3.98190886e-01
1.42050242e+00 9.49505419e-02 -1.40373688e-02 -1.08521235e+00
-1.19487420e-02 -6.71583042e-02 -2.62216628e-01 -6.59775257e-01
-2.61925101e-01 -1.10889637e+00 4.09946442e-02 -1.50288153e+00
1.49391770e-01 -7.89682209e-01 -3.44085008e-01 -3.04481797e-02
3.66265955e-03 1.59678832e-01 5.58536910e-02 4.49308425e-01
1.74876288e-01 2.55653828e-01 1.74117947e+00 3.87651473e-02
-3.00575644e-01 -1.78637937e-01 -1.56756505e-01 4.23870474e-01
7.45919704e-01 -6.90126657e-01 -3.14543486e-01 -3.58003974e-01
-1.51868597e-01 8.39455783e-01 3.80325258e-01 -1.05772579e+00
2.62461454e-01 -1.76590323e-01 7.36778319e-01 -3.83295238e-01
1.99581049e-02 -1.05001247e+00 7.34337866e-01 1.04702377e+00
-5.32243013e-01 -2.00783595e-01 -3.93874235e-02 3.06361824e-01
-1.66644156e-01 -5.11103213e-01 1.52399051e+00 -6.86974466e-01
-3.27796966e-01 2.80444503e-01 -3.09737444e-01 -2.72323936e-01
6.40136361e-01 -2.92121947e-01 9.89086181e-02 -9.03877020e-02
-1.03107166e+00 1.45242333e-01 1.66585147e-01 -3.27254206e-01
6.60205662e-01 -1.27903366e+00 -7.60015726e-01 2.07037628e-01
-3.70757312e-01 -3.87268141e-02 7.92878866e-01 1.57997847e+00
-1.20097733e+00 2.62831956e-01 -4.27723080e-01 -8.36886108e-01
-7.58759499e-01 4.71129596e-01 8.56198013e-01 -3.76623362e-01
-9.85460818e-01 5.63463151e-01 -4.18541953e-02 -2.31726378e-01
5.22443242e-02 -4.44451183e-01 -1.59615293e-01 -3.45783442e-01
3.09537470e-01 4.07677233e-01 -1.31384870e-02 -3.50944042e-01
-4.86954927e-01 6.99175119e-01 2.32607126e-01 5.20108594e-03
1.56791162e+00 -3.16428244e-01 -2.05607176e-01 1.43300116e-01
1.18734348e+00 9.69315022e-02 -1.19322121e+00 -1.33001313e-01
-5.97430654e-02 -4.63754952e-01 6.78001821e-01 -9.86374795e-01
-1.34033644e+00 6.57561064e-01 1.21711934e+00 -2.55943507e-01
9.75385070e-01 -4.96000558e-01 7.21573234e-01 -3.02805185e-01
6.43873289e-02 -8.25787008e-01 -2.78357565e-01 -1.57743812e-01
8.23139429e-01 -1.21094942e+00 2.80333370e-01 -5.09871244e-01
-7.70367682e-01 1.18655896e+00 4.05695319e-01 -5.28638601e-01
6.86079741e-01 5.40219486e-01 2.50292599e-01 -3.33310634e-01
-3.46062407e-02 1.45110354e-01 1.12393290e-01 2.98101544e-01
7.00760901e-01 1.04505479e-01 -8.34920764e-01 1.72005072e-01
5.58142185e-01 4.44424927e-01 5.82200170e-01 8.19893181e-01
-4.98644859e-01 -8.41023088e-01 -4.43875819e-01 5.42727292e-01
-7.38251328e-01 -2.23102234e-02 5.10702550e-01 4.25197750e-01
9.23888683e-02 2.96957821e-01 -2.92216569e-01 3.66318733e-01
2.54745543e-01 2.56770141e-02 8.93703997e-01 -2.28037909e-01
-5.83850741e-01 3.80113125e-01 -1.34129584e-01 -5.46333969e-01
-5.70231557e-01 -5.82188964e-01 -1.57655561e+00 -3.63096967e-02
-2.68786043e-01 9.46837142e-02 1.05748129e+00 7.60089993e-01
1.04760133e-01 6.10126197e-01 5.83339989e-01 -5.77958167e-01
-6.26240075e-01 -8.48105013e-01 -8.00460160e-01 5.16482413e-01
4.75066394e-01 -5.00147760e-01 -2.80568659e-01 9.36834142e-02]
|
[13.596352577209473, -2.512986898422241]
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.