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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
221ddaee-0da7-4a10-834f-adba333ef5ab | sad-saliency-based-defenses-against | 2003.04820 | null | https://arxiv.org/abs/2003.04820v1 | https://arxiv.org/pdf/2003.04820v1.pdf | SAD: Saliency-based Defenses Against Adversarial Examples | With the rise in popularity of machine and deep learning models, there is an increased focus on their vulnerability to malicious inputs. These adversarial examples drift model predictions away from the original intent of the network and are a growing concern in practical security. In order to combat these attacks, neural networks can leverage traditional image processing approaches or state-of-the-art defensive models to reduce perturbations in the data. Defensive approaches that take a global approach to noise reduction are effective against adversarial attacks, however their lossy approach often distorts important data within the image. In this work, we propose a visual saliency based approach to cleaning data affected by an adversarial attack. Our model leverages the salient regions of an adversarial image in order to provide a targeted countermeasure while comparatively reducing loss within the cleaned images. We measure the accuracy of our model by evaluating the effectiveness of state-of-the-art saliency methods prior to attack, under attack, and after application of cleaning methods. We demonstrate the effectiveness of our proposed approach in comparison with related defenses and against established adversarial attack methods, across two saliency datasets. Our targeted approach shows significant improvements in a range of standard statistical and distance saliency metrics, in comparison with both traditional and state-of-the-art approaches. | ['Michael Geyer', 'David Patrick', 'Richard Tran', 'Amanda Fernandez'] | 2020-03-10 | null | null | null | null | ['music-genre-recognition'] | ['music'] | [ 5.41049421e-01 1.86408043e-01 1.14213616e-01 -2.22959980e-01
-6.44237161e-01 -8.31832647e-01 7.16842353e-01 3.54916066e-01
-4.05747503e-01 4.20604348e-01 1.48364410e-01 -2.65955895e-01
-1.01479053e-01 -7.41230905e-01 -1.02328503e+00 -5.93655884e-01
-7.38132894e-02 -1.92014709e-01 6.07517183e-01 -6.37440622e-01
5.85796535e-01 7.59536922e-01 -1.19799626e+00 2.21325323e-01
7.91671336e-01 1.02190852e+00 -3.61303777e-01 6.74608529e-01
3.00840974e-01 8.84635627e-01 -9.90777910e-01 -7.22302973e-01
6.72125518e-01 -1.15696028e-01 -6.80312216e-01 -5.30869365e-01
8.20684373e-01 -1.87035754e-01 -6.26382232e-01 1.67847121e+00
5.28083861e-01 -2.39705481e-02 3.17405194e-01 -1.51616526e+00
-9.12710011e-01 5.69769144e-01 -8.55083942e-01 6.64266646e-01
-5.54128550e-02 2.43194088e-01 6.54988110e-01 -5.91548026e-01
4.63022649e-01 1.33138883e+00 6.53894067e-01 8.32878470e-01
-1.06250346e+00 -8.38208795e-01 3.63958329e-01 2.93740630e-01
-8.29962015e-01 -4.38786089e-01 1.23824143e+00 -1.03348583e-01
4.57668394e-01 5.11109352e-01 1.08267605e-01 1.27439594e+00
3.63226771e-01 6.38900995e-01 9.83091474e-01 -3.20231259e-01
3.72770429e-01 2.01832876e-01 -9.99451801e-03 5.47954738e-01
4.37091947e-01 4.34061617e-01 -4.88237381e-01 -3.31634998e-01
2.27174640e-01 1.80804893e-01 -9.12839100e-02 -4.59025592e-01
-6.64491892e-01 1.01896000e+00 1.13477266e+00 2.11409549e-03
-5.07877707e-01 2.44464502e-01 4.33358878e-01 3.48401815e-02
5.76553166e-01 8.10706496e-01 -3.26655596e-01 3.14334810e-01
-1.05973995e+00 4.15850013e-01 5.03487110e-01 4.37373072e-01
3.78073394e-01 5.55176616e-01 -3.18075687e-01 3.74462903e-01
-1.00409547e-02 4.99986529e-01 2.12143958e-01 -5.61482131e-01
2.58902818e-01 5.47046721e-01 -2.09944382e-01 -1.38151813e+00
1.16935670e-02 -6.31034672e-01 -6.60814762e-01 8.13804150e-01
9.19546857e-02 -2.74666637e-01 -1.29280865e+00 1.96512532e+00
3.58252555e-01 3.03374857e-01 1.61310866e-01 7.73500323e-01
2.80406862e-01 2.84302890e-01 2.84340203e-01 2.96241105e-01
8.92802775e-01 -8.72502923e-01 -3.94265980e-01 -5.38336635e-01
-2.68296730e-02 -6.43970490e-01 8.39875758e-01 2.76000202e-01
-9.20286834e-01 -2.07204849e-01 -1.13778138e+00 2.61412024e-01
-7.98497438e-01 -5.93876660e-01 2.73387402e-01 9.09539938e-01
-9.07816708e-01 7.48192847e-01 -6.16613686e-01 -1.57923192e-01
1.08301353e+00 4.13382381e-01 -2.14604959e-01 2.77590096e-01
-1.26364350e+00 1.15227079e+00 6.75807819e-02 -2.36511260e-01
-1.54589987e+00 -1.08101892e+00 -7.33846605e-01 1.51791438e-01
4.80587959e-01 -1.66332275e-01 1.09129095e+00 -1.18112671e+00
-7.49900818e-01 6.47520065e-01 3.42933089e-01 -1.32386363e+00
5.92046082e-01 -6.65757537e-01 -4.28196371e-01 3.39945108e-01
-1.84160993e-02 8.02843630e-01 1.52408993e+00 -1.46268713e+00
-5.54764569e-01 -2.17875242e-01 3.48958343e-01 5.54216988e-02
-5.62965155e-01 2.81489372e-01 2.64910102e-01 -1.06639433e+00
-5.30153751e-01 -8.84196579e-01 -3.28914553e-01 7.27522597e-02
-7.40013719e-01 6.20822906e-01 1.76786959e+00 -6.30691111e-01
1.01205134e+00 -2.12496281e+00 3.25362831e-02 1.95161581e-01
4.04069543e-01 9.32052970e-01 -5.41389957e-02 2.17367470e-01
-3.71613920e-01 4.73375440e-01 -5.35407245e-01 -1.88530505e-01
-1.45105124e-01 -1.10176831e-01 -1.04395080e+00 5.74046075e-01
4.39417332e-01 8.18715870e-01 -7.81091809e-01 -2.35151142e-01
1.60511136e-01 4.82592762e-01 -4.83831137e-01 1.09441124e-01
-8.70203376e-02 1.06434807e-01 -2.73400754e-01 8.19892883e-01
6.62265897e-01 3.38457584e-01 -5.46358585e-01 -1.10989921e-01
2.31303588e-01 -7.28958771e-02 -5.95783949e-01 1.09613562e+00
6.88977689e-02 6.76475763e-01 1.80879593e-01 -1.07447648e+00
7.67451882e-01 3.34992036e-02 1.17003746e-01 -6.71025991e-01
3.51701140e-01 -8.95729065e-02 1.44287452e-01 1.46332635e-02
3.80098224e-01 2.36474380e-01 -2.51935959e-01 2.73056000e-01
-8.14912245e-02 6.50979951e-02 -2.79570818e-01 5.33196449e-01
1.38991201e+00 -2.83670425e-01 1.43448323e-01 -4.30974871e-01
4.03620988e-01 1.32980794e-01 4.91957694e-01 9.48708117e-01
-7.05963373e-01 4.46704537e-01 4.71451938e-01 -6.82242930e-01
-1.09349155e+00 -1.16387725e+00 2.53784359e-01 1.10991669e+00
3.44550371e-01 9.80698690e-02 -1.47405362e+00 -1.30013478e+00
1.15657240e-01 7.63555646e-01 -1.10329819e+00 -9.17468905e-01
-6.08565152e-01 -5.88154256e-01 1.02725637e+00 3.24672014e-01
8.48456740e-01 -1.24696767e+00 -9.86463368e-01 -1.88421577e-01
2.81288117e-01 -1.02238750e+00 -3.40838850e-01 1.09187312e-01
-6.48358464e-01 -1.22880459e+00 -2.87327558e-01 -4.64292288e-01
7.88536787e-01 5.22146404e-01 8.86785984e-01 3.58285189e-01
-5.39741516e-01 2.67473221e-01 -2.88185388e-01 -1.27602863e+00
-4.53743279e-01 -1.25038950e-03 2.23075837e-01 1.17658608e-01
2.49994412e-01 -4.51388121e-01 -6.34473979e-01 9.04551968e-02
-1.31970882e+00 -4.34228897e-01 6.86333537e-01 5.64997077e-01
3.33451897e-01 9.73332003e-02 5.82834482e-01 -9.42622900e-01
7.88526356e-01 -5.72035730e-01 -5.43893993e-01 1.29915208e-01
-4.69644696e-01 1.49567351e-01 8.36724520e-01 -6.39235198e-01
-7.43382096e-01 -6.61388785e-02 1.20268889e-01 -9.62010264e-01
-2.44036913e-01 -1.10819386e-02 -1.63296923e-01 -7.85633445e-01
1.10474706e+00 7.72799924e-03 -1.21719494e-01 -1.90641761e-01
4.27145302e-01 1.62525102e-01 8.67563307e-01 -2.49541402e-01
1.57696486e+00 6.76045716e-01 2.34763339e-01 -5.85231423e-01
-1.16728711e+00 -1.09215844e-02 -3.63639951e-01 -1.61293477e-01
6.73849046e-01 -4.05095249e-01 -2.15710655e-01 6.72957122e-01
-9.17621553e-01 -3.17129344e-02 -3.16667080e-01 -3.14910173e-01
-1.67258620e-01 2.61136472e-01 -2.44607419e-01 -8.00099134e-01
-7.69054353e-01 -1.16816378e+00 7.97904432e-01 4.77145493e-01
3.58343422e-02 -8.84584069e-01 -7.22060725e-02 1.19790561e-01
8.27874839e-01 9.25324798e-01 6.50744379e-01 -1.21987474e+00
-4.78931248e-01 -5.04624009e-01 -2.36489415e-01 5.42056441e-01
1.85893938e-01 -6.29561171e-02 -1.12495232e+00 -3.14652920e-01
3.19977373e-01 -2.27585733e-01 9.15044546e-01 3.55847836e-01
1.15824986e+00 -6.89337730e-01 -1.54352084e-01 5.54278910e-01
1.53005159e+00 1.47880554e-01 7.70490170e-01 6.25517845e-01
7.66461134e-01 6.21865571e-01 7.10099041e-01 4.48413612e-03
-2.36637026e-01 3.42017293e-01 1.35597575e+00 -3.40839893e-01
6.40047640e-02 -2.53330618e-01 2.44721711e-01 -3.52532893e-01
4.22441214e-01 -6.43085390e-02 -8.46568048e-01 7.62263000e-01
-1.69435549e+00 -1.17862308e+00 3.24567825e-01 2.15632486e+00
7.12939858e-01 6.98421717e-01 6.95698783e-02 2.78671473e-01
9.66981411e-01 5.73086023e-01 -8.11490357e-01 -8.84472907e-01
-1.32010207e-01 3.16817880e-01 9.69885468e-01 4.05623257e-01
-1.70469296e+00 1.20741689e+00 6.31372929e+00 8.08105230e-01
-1.27874351e+00 7.44038299e-02 8.79887700e-01 -3.95484060e-01
1.27135485e-01 -7.50786886e-02 -5.30058742e-01 4.13883835e-01
9.97848451e-01 -2.82050729e-01 4.90320534e-01 1.25620949e+00
1.96027588e-02 1.40157595e-01 -7.37356544e-01 4.36291307e-01
4.09568518e-01 -1.54701936e+00 2.16091901e-01 1.20789111e-01
7.57879913e-01 -9.86116454e-02 6.68587625e-01 7.92057291e-02
5.25908887e-01 -1.11298025e+00 9.25233901e-01 3.92512888e-01
2.79552639e-01 -1.17609942e+00 6.67237520e-01 1.04945093e-01
-8.03362548e-01 -3.11962724e-01 -2.17786252e-01 7.20578358e-02
-7.81116113e-02 3.96342725e-01 -6.22388721e-01 1.41279191e-01
9.24255908e-01 1.72214121e-01 -1.03568947e+00 8.87849510e-01
-3.30861896e-01 7.10424304e-01 6.97857440e-02 1.93700209e-01
5.40041983e-01 5.16089320e-01 1.04340863e+00 1.00445962e+00
-2.07355633e-01 -3.64115268e-01 -7.82972351e-02 7.71046460e-01
-3.73278886e-01 -3.23998779e-01 -9.55972016e-01 9.40542966e-02
6.10417843e-01 1.18092978e+00 -6.41270757e-01 -2.00640038e-01
-6.96329251e-02 8.31222951e-01 1.92445219e-01 1.41396582e-01
-1.31052184e+00 -7.19940603e-01 1.13537121e+00 9.68721043e-03
3.81886631e-01 1.39915109e-01 -6.91936672e-01 -6.50741875e-01
-2.78188176e-02 -1.27347410e+00 2.59615630e-01 -3.91429275e-01
-1.09633672e+00 8.11070800e-01 9.17913094e-02 -1.02540863e+00
1.16174303e-01 -3.20470482e-01 -1.08575404e+00 7.13637710e-01
-1.45562518e+00 -1.24921346e+00 -1.40670434e-01 6.90947354e-01
5.92540085e-01 -5.52145660e-01 5.18272400e-01 -4.33398597e-02
-4.74240005e-01 7.33182907e-01 -3.23970467e-01 3.98372501e-01
7.30883539e-01 -1.20443654e+00 8.94878328e-01 1.77787566e+00
2.57107109e-01 5.40805817e-01 1.13604140e+00 -8.31791759e-01
-9.39220965e-01 -1.46583390e+00 1.19633257e-01 -5.80884874e-01
7.69040227e-01 -5.10927916e-01 -1.10923934e+00 4.60375786e-01
4.30358589e-01 1.22498564e-01 3.50867152e-01 -5.11119425e-01
-6.84361994e-01 -1.56068176e-01 -1.75630009e+00 8.67150486e-01
7.72148907e-01 -5.43587267e-01 -7.28480637e-01 6.20081797e-02
1.11545360e+00 -3.66733253e-01 -7.86994919e-02 5.61243296e-01
1.90800875e-01 -1.07728529e+00 1.33503926e+00 -8.68814528e-01
4.51259941e-01 -2.60527581e-01 7.55543448e-03 -1.41483462e+00
-2.95072973e-01 -8.00910115e-01 -2.80534089e-01 1.06844795e+00
3.22394460e-01 -4.24775273e-01 8.95249069e-01 5.45742631e-01
-2.04976216e-01 -6.84365094e-01 -8.46838474e-01 -5.74006379e-01
6.45071939e-02 -2.13865474e-01 5.95423520e-01 7.96035469e-01
-5.26531160e-01 -2.01684371e-01 -5.74043214e-01 6.84347212e-01
9.93491948e-01 -4.17744070e-01 6.23803675e-01 -1.04757702e+00
1.21658161e-01 -4.94757652e-01 -8.54292333e-01 1.84833452e-01
9.31079015e-02 -3.12393606e-01 -9.26939771e-02 -1.01228678e+00
-4.28685434e-02 4.47683930e-02 -5.97112477e-01 7.07041562e-01
-4.00434792e-01 5.71476638e-01 3.43382001e-01 -2.82001588e-02
-5.04721999e-01 2.71271110e-01 6.01395190e-01 -3.27732861e-01
7.07708448e-02 -2.18043521e-01 -1.27467108e+00 9.59268570e-01
1.12203515e+00 -9.34098303e-01 -4.31472182e-01 -3.80169928e-01
-1.31046504e-01 -8.73565435e-01 7.97877848e-01 -1.40888989e+00
2.27222905e-01 -1.49534672e-01 4.17619258e-01 -1.66049793e-01
1.89128637e-01 -8.49666715e-01 -3.92466873e-01 7.98079431e-01
-4.28556621e-01 2.29583323e-01 5.74074090e-01 6.75692677e-01
-1.95049882e-01 -3.79608907e-02 1.36624324e+00 4.09675315e-02
-6.38554275e-01 2.91906536e-01 6.34916499e-02 2.43058786e-01
1.41699302e+00 -8.63815099e-02 -8.18361044e-01 -4.04554129e-01
-4.27563488e-01 -9.94078070e-02 5.98110080e-01 7.80538261e-01
8.73994946e-01 -1.09741330e+00 -4.70266730e-01 1.99290752e-01
-1.81314290e-01 -3.56580019e-01 5.85846975e-02 4.24359858e-01
-5.03158212e-01 6.48580715e-02 -5.87724209e-01 -1.57242984e-01
-1.59667230e+00 1.16425920e+00 3.26846719e-01 -1.72899649e-01
-3.85102808e-01 8.51367295e-01 4.15404469e-01 7.00094923e-02
5.04295349e-01 1.18733183e-01 -1.58840612e-01 -3.40219945e-01
7.55110800e-01 3.07049632e-01 1.61949228e-02 -7.51412928e-01
-5.35411477e-01 7.25926235e-02 -4.30025548e-01 6.62890300e-02
1.19352806e+00 2.38984734e-01 9.90839899e-02 -2.66134888e-01
9.09259915e-01 2.54063636e-01 -1.35916626e+00 -7.58315325e-02
4.72016037e-02 -6.96418166e-01 9.37733799e-02 -9.92365956e-01
-1.32737255e+00 8.59848559e-01 8.55096102e-01 5.06827831e-01
1.38928258e+00 -2.93814927e-01 8.26780081e-01 1.55274093e-01
1.10371843e-01 -9.74853039e-01 2.44344100e-01 3.31726670e-01
9.61323082e-01 -1.29662347e+00 -5.29443175e-02 -2.69402176e-01
-9.41814482e-01 5.33020973e-01 6.92600965e-01 -7.97385156e-01
5.83057761e-01 3.44509423e-01 2.65182763e-01 -1.61455125e-01
-5.26616156e-01 1.31048918e-01 4.74855274e-01 9.16269600e-01
-3.85706395e-01 -2.05675676e-01 3.00802141e-01 2.96264708e-01
-2.77513623e-01 -6.97189152e-01 5.82693815e-01 1.24513543e+00
-6.46172285e-01 -8.32074702e-01 -6.34388208e-01 3.22020501e-01
-1.05415845e+00 -2.66567230e-01 -1.02295434e+00 6.48391545e-01
1.21056475e-03 9.61916983e-01 -3.99952859e-01 -6.01175725e-01
4.62893516e-01 -1.54515967e-01 -1.54326722e-01 -4.48014468e-01
-9.44769859e-01 -4.46537167e-01 -2.40752414e-01 -8.05290699e-01
-1.25940219e-01 -5.44848621e-01 -9.24643993e-01 -2.95558631e-01
-1.81133881e-01 3.28932405e-02 8.85779977e-01 7.86496520e-01
6.13881409e-01 6.29098237e-01 7.43825793e-01 -9.30859387e-01
-7.42129743e-01 -6.53751016e-01 -9.73054208e-03 6.66098177e-01
6.85197234e-01 -7.63156950e-01 -5.97319305e-01 -8.49254802e-02] | [5.530428409576416, 7.932652473449707] |
fdf6660c-0f57-4fd7-9e6e-693fb2e3e6d3 | chartqa-a-benchmark-for-question-answering | 2203.10244 | null | https://arxiv.org/abs/2203.10244v1 | https://arxiv.org/pdf/2203.10244v1.pdf | ChartQA: A Benchmark for Question Answering about Charts with Visual and Logical Reasoning | Charts are very popular for analyzing data. When exploring charts, people often ask a variety of complex reasoning questions that involve several logical and arithmetic operations. They also commonly refer to visual features of a chart in their questions. However, most existing datasets do not focus on such complex reasoning questions as their questions are template-based and answers come from a fixed-vocabulary. In this work, we present a large-scale benchmark covering 9.6K human-written questions as well as 23.1K questions generated from human-written chart summaries. To address the unique challenges in our benchmark involving visual and logical reasoning over charts, we present two transformer-based models that combine visual features and the data table of the chart in a unified way to answer questions. While our models achieve the state-of-the-art results on the previous datasets as well as on our benchmark, the evaluation also reveals several challenges in answering complex reasoning questions. | ['Enamul Hoque', 'Shafiq Joty', 'Jia Qing Tan', 'Do Xuan Long', 'Ahmed Masry'] | 2022-03-19 | null | https://aclanthology.org/2022.findings-acl.177 | https://aclanthology.org/2022.findings-acl.177.pdf | findings-acl-2022-5 | ['chart-question-answering', 'chart-question-answering'] | ['computer-code', 'computer-vision'] | [-2.94424176e-01 -9.73526447e-04 -7.36735482e-03 -3.51259649e-01
-1.03332353e+00 -1.02715743e+00 7.02640533e-01 5.90717554e-01
3.35635781e-01 3.39867741e-01 7.16829479e-01 -9.26093876e-01
-2.43047222e-01 -8.57020855e-01 -6.27160072e-01 2.50093371e-01
2.17048123e-01 4.99452382e-01 4.79445606e-01 -4.11060929e-01
6.40004098e-01 4.23256665e-01 -1.48819387e+00 1.02055228e+00
1.11029637e+00 1.07331526e+00 -4.10367757e-01 9.40479040e-01
-1.02764797e+00 1.75371182e+00 -9.94118214e-01 -8.61872077e-01
-6.25670478e-02 -2.98973948e-01 -1.06533575e+00 -1.54618919e-01
1.00812626e+00 -3.53256673e-01 -2.79853284e-01 9.46554601e-01
2.44700208e-01 -2.36626659e-02 7.05715239e-01 -1.61036587e+00
-1.27423060e+00 3.88440728e-01 -4.95860487e-01 1.89278901e-01
1.23525083e+00 1.21019788e-01 1.21629369e+00 -1.01693738e+00
7.39780664e-01 1.52533674e+00 4.08756942e-01 1.13026842e-01
-1.00809896e+00 -3.54302943e-01 3.34426880e-01 7.41598725e-01
-1.04479373e+00 -2.02597454e-01 7.81957269e-01 -4.54713494e-01
1.01138747e+00 9.49829400e-01 6.75581276e-01 4.41815883e-01
1.20118059e-01 9.25149739e-01 9.39497828e-01 -2.86843657e-01
2.46552736e-01 -4.69176769e-02 3.24923486e-01 8.75970244e-01
1.52574554e-01 -6.89858437e-01 -5.49573302e-01 -1.12501144e-01
6.52873278e-01 1.11677952e-01 -5.58224246e-02 -5.77298522e-01
-1.50067019e+00 7.40301192e-01 6.23322010e-01 -1.43037075e-02
3.41892689e-02 6.20226972e-02 5.56473851e-01 4.96280760e-01
-2.64826596e-01 6.85802698e-01 -2.05037862e-01 -1.37019202e-01
-5.15902340e-01 5.97658515e-01 1.16517091e+00 1.48674107e+00
7.93760717e-01 -4.41406459e-01 -1.01939464e+00 4.36335981e-01
2.48635843e-01 4.75225329e-01 -1.27234891e-01 -1.04566824e+00
1.06144679e+00 1.60073435e+00 4.18918133e-02 -1.32743704e+00
-3.61447036e-01 1.29941136e-01 -6.27909541e-01 3.34099978e-01
6.41577482e-01 2.80203968e-01 -8.91780019e-01 9.62608159e-01
3.25636894e-01 -6.86302245e-01 -1.04670979e-01 6.24164343e-01
1.53427362e+00 6.28548265e-01 -1.44652367e-01 3.75844091e-01
1.93054879e+00 -1.20986819e+00 -1.06464088e+00 -5.50871454e-02
4.90328431e-01 -8.88688505e-01 1.76732183e+00 3.88884276e-01
-1.12647474e+00 -4.80579942e-01 -8.16840291e-01 -9.68479872e-01
-8.89685929e-01 1.66324377e-02 4.72616762e-01 4.74210382e-01
-1.05075848e+00 -3.00430983e-01 -1.89457819e-01 -4.23250169e-01
4.34831470e-01 -4.25420344e-01 -1.34896904e-01 -2.45461851e-01
-9.33784902e-01 9.87109482e-01 -7.21329004e-02 -1.93815187e-01
-3.75432163e-01 -1.09168839e+00 -8.66839230e-01 4.07694340e-01
9.13614154e-01 -6.66403472e-01 1.48425472e+00 1.96867481e-01
-9.93920207e-01 6.31833851e-01 -4.19224709e-01 -1.48438632e-01
8.03811610e-01 -3.80250603e-01 -5.25227427e-01 3.28877032e-01
1.25214949e-01 3.99568975e-01 4.05679017e-01 -9.87927377e-01
-3.43239069e-01 -2.24270821e-01 7.09951341e-01 -1.00419305e-01
-1.40462920e-01 1.21765196e-01 -9.31487858e-01 -6.41529620e-01
-1.68747559e-01 -2.85651803e-01 4.26858626e-02 6.26593053e-01
-6.74019277e-01 -4.29359853e-01 1.08381188e+00 -6.88601851e-01
1.87613499e+00 -1.90618658e+00 -1.66030303e-01 1.83862105e-01
6.50992572e-01 -2.96788868e-02 3.75740565e-02 8.83381844e-01
1.12900004e-01 4.50764447e-01 2.52737780e-03 1.59186184e-01
5.77766716e-01 -7.33837336e-02 -8.12273383e-01 -2.91862965e-01
3.04611504e-01 1.39524174e+00 -7.06067860e-01 -1.02409244e+00
2.99129069e-01 2.16489546e-02 -4.84807223e-01 3.80090266e-01
-7.10485101e-01 -5.10922708e-02 -4.58728462e-01 1.02984834e+00
6.90391541e-01 -7.06300616e-01 6.86038705e-03 -2.91076958e-01
-1.34122297e-01 2.15385929e-01 -1.14864266e+00 1.76531661e+00
-1.89926803e-01 7.77808487e-01 -4.40615237e-01 -4.05102670e-01
1.19061363e+00 -2.25570910e-02 -1.25885189e-01 -1.13447106e+00
-3.91362846e-01 -2.08476022e-01 -4.41700965e-01 -8.94500017e-01
6.20995283e-01 4.86540258e-01 -3.02271456e-01 4.21501011e-01
-3.15307975e-01 -6.01923585e-01 8.76134574e-01 6.56211138e-01
1.38709092e+00 -6.05730414e-02 6.33352339e-01 5.12971543e-04
8.40676844e-01 4.46675658e-01 -5.44150546e-02 8.38874876e-01
9.75319073e-02 7.68950760e-01 1.32669652e+00 -8.56981814e-01
-8.47532868e-01 -1.37660837e+00 2.86224067e-01 1.01988590e+00
-6.20618388e-02 -1.28458500e+00 -3.99661869e-01 -6.49068713e-01
2.10191518e-01 8.32092702e-01 -8.55719447e-01 3.52354974e-01
-4.14800406e-01 2.36238763e-01 3.86253566e-01 8.52248490e-01
5.02786696e-01 -1.00524998e+00 -8.12776744e-01 -8.32067654e-02
-6.75843135e-02 -1.04892790e+00 -4.28163081e-01 -3.87149006e-01
-5.65869868e-01 -1.59449089e+00 -5.00886619e-01 -4.75009114e-01
6.56892002e-01 2.24355698e-01 1.67151833e+00 2.71376908e-01
-3.62247050e-01 5.62701225e-01 -3.68406326e-01 -8.50188315e-01
-1.66363448e-01 1.75779629e-02 -1.10275078e+00 -4.55175728e-01
4.87025559e-01 -1.02629915e-01 -4.02826279e-01 2.25268662e-01
-1.17954433e+00 1.94621712e-01 4.15661603e-01 1.54253885e-01
4.06501532e-01 -2.93217868e-01 1.88570082e-01 -8.18443537e-01
1.04409051e+00 -2.77567893e-01 -6.57367945e-01 1.21363842e+00
-3.23648781e-01 5.73446155e-01 8.54351640e-01 -1.25269279e-01
-1.02210701e+00 -3.94623339e-01 2.57012129e-01 -3.58391047e-01
9.32820316e-04 6.02422595e-01 -9.45731103e-02 3.55077624e-01
6.71052992e-01 -1.08084984e-01 -3.71655315e-01 -4.90827262e-01
7.80059397e-01 1.11194186e-01 6.21772468e-01 -8.33630204e-01
1.12559140e+00 5.04541159e-01 1.12699643e-01 -3.33166212e-01
-7.48227954e-01 -2.76368499e-01 -4.09442902e-01 -4.33000118e-01
8.91607463e-01 -5.73094904e-01 -1.39915693e+00 -1.38707086e-01
-1.33659017e+00 1.32185638e-01 -4.46515024e-01 -2.04428405e-01
-3.85119498e-01 2.52502799e-01 -1.96419939e-01 -8.47857714e-01
-2.97951043e-01 -1.16416836e+00 1.07007420e+00 2.25701004e-01
-4.40355331e-01 -8.99622500e-01 4.54301089e-02 4.21152890e-01
5.44317961e-01 6.98510349e-01 1.50286508e+00 -3.03634048e-01
-1.05324411e+00 -4.61330935e-02 -7.71172702e-01 -3.77592176e-01
1.11992471e-01 4.44208145e-01 -7.20948756e-01 3.28946024e-01
-4.55113798e-01 -6.25177264e-01 6.16498888e-01 -2.62402445e-01
1.53834569e+00 -2.94936597e-01 1.15342690e-02 3.36126804e-01
1.32144427e+00 2.90506721e-01 7.39392817e-01 2.93438226e-01
8.01155090e-01 6.33012950e-01 3.86814445e-01 4.99322146e-01
1.10511839e+00 1.34064123e-01 5.63751936e-01 5.14412895e-02
-2.66253296e-02 -7.26282656e-01 -9.07956213e-02 6.90590858e-01
2.45072767e-01 -1.24849238e-01 -1.47665167e+00 4.78569150e-01
-1.95078087e+00 -9.61899281e-01 -6.52536094e-01 1.75260246e+00
6.52506232e-01 1.07989959e-01 1.05408333e-01 2.74073362e-01
3.72291237e-01 5.58613658e-01 -6.71771646e-01 -7.59993553e-01
-5.40763661e-02 1.32444724e-01 -1.13148727e-01 1.49570763e-01
-7.15633452e-01 2.39536613e-01 6.87751818e+00 4.96441990e-01
-6.02065086e-01 -5.47199786e-01 3.53367895e-01 -6.60976246e-02
-9.65084791e-01 1.20888315e-01 -5.19869566e-01 -1.18612312e-01
5.88608384e-01 -5.86350799e-01 3.72578323e-01 6.94491923e-01
-3.44372720e-01 -2.71027237e-01 -1.31807804e+00 1.19498253e+00
2.73107886e-01 -1.83635676e+00 5.76439798e-01 -3.67028654e-01
5.96262336e-01 -9.29464459e-01 1.57244638e-01 5.56945026e-01
3.45424801e-01 -1.40408361e+00 1.01351893e+00 1.03806400e+00
7.90946960e-01 -5.79833269e-01 1.63250938e-01 3.98374535e-02
-1.67382085e+00 -1.65289253e-01 -1.24332719e-01 -2.45640486e-01
-1.41144633e-01 4.38113600e-01 -6.01150811e-01 8.24967563e-01
9.91911232e-01 5.62656581e-01 -1.44250190e+00 1.01175344e+00
-3.54044378e-01 2.46045496e-02 8.90229121e-02 -4.94938821e-01
1.41663164e-01 9.66321900e-02 -9.68294367e-02 1.06130528e+00
3.54206562e-01 -1.45809278e-01 -8.53315890e-02 1.08777452e+00
-2.05848128e-01 2.71511674e-01 -5.63626051e-01 -2.08763927e-01
3.04557711e-01 1.05211198e+00 -5.88605702e-01 -6.10654533e-01
-9.18872654e-01 2.25104660e-01 6.69789195e-01 4.29867566e-01
-9.19206321e-01 -6.96462333e-01 4.37216252e-01 1.92461640e-01
2.86700726e-01 -1.19517468e-01 -6.60907328e-01 -1.26422727e+00
5.63441157e-01 -1.23419249e+00 8.75686646e-01 -1.44772875e+00
-1.22944009e+00 3.82207960e-01 2.29462966e-01 -1.36998379e+00
-1.33201256e-01 -7.13599205e-01 -7.82902002e-01 7.64421165e-01
-1.45970023e+00 -9.51170802e-01 -8.43896925e-01 6.58239782e-01
4.29563165e-01 1.67841703e-01 7.66363442e-01 5.49968258e-02
-2.75708705e-01 2.32756674e-01 -2.08122358e-01 4.74175602e-01
8.88358772e-01 -1.70754683e+00 8.75831008e-01 6.13674223e-01
8.88421759e-02 7.65422344e-01 5.45739412e-01 -4.32320118e-01
-1.85497022e+00 -8.15487504e-01 8.47600102e-01 -9.43432987e-01
8.48225117e-01 -6.21513665e-01 -1.21062410e+00 7.83417702e-01
5.87306917e-01 5.56236766e-02 6.02969110e-01 1.97651479e-02
-9.43577886e-01 -1.43399775e-01 -8.95603418e-01 9.31155980e-01
1.02639687e+00 -8.65071714e-01 -1.20759952e+00 1.12139985e-01
8.02963078e-01 -7.33349502e-01 -8.44019294e-01 1.37682632e-01
6.76043630e-01 -1.21480703e+00 1.02175963e+00 -9.50425088e-01
8.20492208e-01 -7.34478235e-01 -6.38246536e-02 -9.04435098e-01
-2.39945635e-01 -5.33654809e-01 -5.63411534e-01 1.14867234e+00
4.51203287e-01 -2.35545665e-01 6.35903180e-01 9.38349962e-01
2.22833455e-01 -6.73793852e-01 -4.19992566e-01 -4.75219697e-01
-2.26053163e-01 -4.56036359e-01 1.25001287e+00 7.91850626e-01
3.35939974e-01 1.83347404e-01 -1.91951618e-02 -2.13206112e-01
3.01472604e-01 5.95960796e-01 1.22514617e+00 -1.31735587e+00
3.37295234e-01 -5.51210105e-01 -2.16443062e-01 -9.01292682e-01
-2.73452818e-01 -7.33698845e-01 -5.35431147e-01 -2.46442246e+00
7.13228285e-02 3.62270564e-01 7.27154985e-02 2.59282529e-01
-3.99812669e-01 -3.40327770e-01 7.26801932e-01 -1.75713882e-01
-1.00752556e+00 1.48916051e-01 1.70342827e+00 -4.90593702e-01
7.50250071e-02 -4.89486605e-01 -9.88911808e-01 4.82709408e-01
5.41610897e-01 -5.66079766e-02 -7.24060953e-01 -7.00215280e-01
1.02397346e+00 3.86610150e-01 4.42146838e-01 -1.02200615e+00
7.66002357e-01 -4.35024261e-01 6.64584517e-01 -1.26928246e+00
7.61830136e-02 -7.51883805e-01 -1.17196918e-01 2.67334074e-01
-5.55615544e-01 8.62747014e-01 2.63202459e-01 4.19215113e-01
-5.26698887e-01 1.08518638e-01 1.72121108e-01 -2.75549978e-01
-7.10629582e-01 2.38765068e-02 -1.14018574e-01 6.13827944e-01
9.56612825e-01 -3.59148867e-02 -1.25014269e+00 -7.85088420e-01
-3.00420046e-01 7.28529811e-01 3.06422830e-01 6.01403773e-01
7.50779390e-01 -1.49700713e+00 -5.67788005e-01 2.36454774e-02
7.28758276e-01 2.54785895e-01 2.67842829e-01 4.39803213e-01
-1.00656211e+00 7.20765650e-01 -4.40638185e-01 -2.54670441e-01
-9.84785259e-01 1.08116913e+00 -1.04308821e-01 -5.58659196e-01
-3.87519956e-01 4.02012289e-01 1.33847639e-01 -4.95806992e-01
3.62637699e-01 -1.19914913e+00 -3.94397080e-01 3.40097338e-01
9.37951982e-01 7.15237319e-01 1.26951501e-01 3.65354754e-02
-4.47956234e-01 5.93759239e-01 7.58127123e-02 9.21668783e-02
9.39000547e-01 2.05538809e-01 -1.84831277e-01 6.21785522e-01
1.00902259e+00 2.60665089e-01 -8.53156388e-01 -2.41582662e-01
3.02218467e-01 -5.84818184e-01 -7.99626946e-01 -8.91530633e-01
-5.93090653e-01 1.17003131e+00 -1.42570555e-01 7.38114953e-01
1.09246767e+00 1.30792901e-01 3.82925004e-01 9.22215700e-01
-1.81972012e-02 -9.60859895e-01 4.56448197e-01 6.11012816e-01
1.68974042e+00 -1.13532782e+00 2.81070739e-01 -5.02261460e-01
-8.50968957e-01 1.56119776e+00 8.50812137e-01 1.43080726e-01
3.79412711e-01 3.59832346e-01 3.50026548e-01 -4.93636221e-01
-1.25320375e+00 -1.32867470e-01 6.97840929e-01 5.23680568e-01
5.45824170e-01 -2.26346016e-01 1.21651366e-01 1.56039447e-01
-5.42955458e-01 1.17543481e-01 6.55130208e-01 1.18850648e+00
-3.21630269e-01 -6.90349638e-01 -9.80266094e-01 7.28015900e-01
-4.07734625e-02 3.42039317e-02 -8.33998024e-01 9.93427455e-01
-6.69174910e-01 1.05668449e+00 1.86717749e-01 -5.19022122e-02
9.17597651e-01 3.15495580e-01 2.36332104e-01 -3.02087694e-01
-7.49647737e-01 -6.40031993e-01 1.13293622e-02 -7.31050193e-01
-1.96208835e-01 -2.81963229e-01 -1.27590549e+00 -5.82071066e-01
5.94068825e-01 1.05247840e-01 2.63494343e-01 5.03282607e-01
3.28147322e-01 8.40211809e-01 -4.16409522e-02 1.92151874e-01
-4.53864306e-01 -6.40381932e-01 -1.58282176e-01 5.89603841e-01
5.35555899e-01 -3.10744762e-01 8.37635715e-03 8.41837078e-02] | [11.19690227508545, 2.0398027896881104] |
7b8ea859-b36a-44e6-9175-1e1dcf3a6b42 | faster-video-moment-retrieval-with-point | 2305.14017 | null | https://arxiv.org/abs/2305.14017v1 | https://arxiv.org/pdf/2305.14017v1.pdf | Faster Video Moment Retrieval with Point-Level Supervision | Video Moment Retrieval (VMR) aims at retrieving the most relevant events from an untrimmed video with natural language queries. Existing VMR methods suffer from two defects: (1) massive expensive temporal annotations are required to obtain satisfying performance; (2) complicated cross-modal interaction modules are deployed, which lead to high computational cost and low efficiency for the retrieval process. To address these issues, we propose a novel method termed Cheaper and Faster Moment Retrieval (CFMR), which well balances the retrieval accuracy, efficiency, and annotation cost for VMR. Specifically, our proposed CFMR method learns from point-level supervision where each annotation is a single frame randomly located within the target moment. It is 6 times cheaper than the conventional annotations of event boundaries. Furthermore, we also design a concept-based multimodal alignment mechanism to bypass the usage of cross-modal interaction modules during the inference process, remarkably improving retrieval efficiency. The experimental results on three widely used VMR benchmarks demonstrate the proposed CFMR method establishes new state-of-the-art with point-level supervision. Moreover, it significantly accelerates the retrieval speed with more than 100 times FLOPs compared to existing approaches with point-level supervision. | ['Heng Tao Shen', 'Guoqing Wang', 'Yang Yang', 'Xing Xu', 'Zailei Zhou', 'Xun Jiang'] | 2023-05-23 | null | null | null | null | ['moment-retrieval'] | ['computer-vision'] | [ 9.09939259e-02 -5.63430667e-01 -4.68447238e-01 -1.27915770e-01
-1.29614604e+00 -3.24983448e-01 5.69803119e-01 2.31386244e-01
-6.90138757e-01 2.97325432e-01 1.37948558e-01 7.73883760e-02
-3.91483009e-01 -5.00006735e-01 -6.92079842e-01 -6.44523740e-01
1.96932517e-02 4.05424386e-01 5.76382637e-01 -1.90976605e-01
3.04605514e-01 2.34603941e-01 -1.67921722e+00 1.51162744e-01
6.63455546e-01 1.23343074e+00 3.96503538e-01 3.37268800e-01
1.19792655e-01 8.44999492e-01 -4.92019802e-01 -4.77252871e-01
-3.58665884e-02 -8.56153741e-02 -7.22844481e-01 -6.99083209e-02
3.14012945e-01 -5.87061107e-01 -6.78090274e-01 8.37914348e-01
6.28705502e-01 5.65031171e-01 4.06094074e-01 -1.07988679e+00
-4.17141318e-01 3.87630552e-01 -9.62654710e-01 3.43761951e-01
7.00345695e-01 -1.57569975e-01 1.14860022e+00 -9.23678100e-01
7.12402761e-01 1.18103540e+00 2.16017663e-01 4.20471072e-01
-7.26097822e-01 -5.14790595e-01 3.37903678e-01 5.09132743e-01
-1.88805676e+00 -4.91368264e-01 7.40846217e-01 -1.03530519e-01
8.05729091e-01 4.09890562e-01 4.58662480e-01 9.25790846e-01
-2.05127597e-01 1.00511503e+00 3.77558559e-01 -1.19377188e-01
4.90735844e-02 -2.51518846e-01 -6.31079823e-02 6.86603665e-01
-3.18861783e-01 -3.07526469e-01 -8.29956830e-01 -1.87196881e-01
8.18910956e-01 3.79516661e-01 -4.66769755e-01 -1.35255039e-01
-1.77382207e+00 5.84553361e-01 1.33132845e-01 2.65076041e-01
-1.63940594e-01 1.32077798e-01 8.64969015e-01 1.45914495e-01
3.50636959e-01 -1.50744498e-01 -2.34417319e-01 -3.09332401e-01
-1.00401986e+00 1.25539422e-01 3.06914270e-01 1.09788847e+00
6.26950741e-01 -4.33172196e-01 -3.08155626e-01 9.36858296e-01
3.89770746e-01 6.40897393e-01 5.68389595e-01 -9.69328940e-01
8.30129743e-01 5.18777788e-01 2.70955175e-01 -1.42680693e+00
-1.03774674e-01 2.20953375e-02 -7.64028370e-01 -7.94776857e-01
7.48473555e-02 4.32817310e-01 -6.25623107e-01 1.52635801e+00
4.03642237e-01 2.95147419e-01 2.85357283e-03 1.27558887e+00
9.13607895e-01 9.16124940e-01 -2.63590063e-03 -4.14984077e-01
1.48192930e+00 -1.10451567e+00 -9.18816447e-01 8.15009847e-02
6.08688831e-01 -8.57071638e-01 1.08958817e+00 2.19975054e-01
-1.11794353e+00 -4.56547827e-01 -8.72763157e-01 -3.45477909e-01
-2.01776564e-01 4.29427534e-01 6.53034031e-01 -1.00870751e-01
-7.97143817e-01 2.23528579e-01 -7.64649570e-01 -1.66056305e-01
-6.80627907e-03 3.25223625e-01 -5.08212566e-01 -2.40401283e-01
-1.27378178e+00 4.94756520e-01 4.22983736e-01 3.21936935e-01
-8.06917071e-01 -4.86964643e-01 -9.03530002e-01 1.16861992e-01
6.74826741e-01 -4.46649522e-01 1.14003360e+00 -5.30003309e-01
-1.42693949e+00 7.50527024e-01 -5.28518319e-01 -1.03446342e-01
4.99542624e-01 -4.48432237e-01 -4.53599960e-01 7.46326685e-01
1.60814747e-01 6.20573997e-01 7.85534382e-01 -8.89929056e-01
-6.74979270e-01 -2.88937211e-01 3.12703341e-01 4.62521315e-01
-3.38289678e-01 9.82786715e-02 -1.22748315e+00 -8.00033689e-01
2.88714290e-01 -7.60869741e-01 6.69022352e-02 -1.30087450e-01
-7.88688064e-02 -5.56964338e-01 8.01962435e-01 -5.96526861e-01
1.46540344e+00 -2.31693792e+00 3.75646412e-01 1.09944511e-02
8.51408541e-02 8.51440132e-02 -1.43056199e-01 5.45144677e-01
1.44436479e-01 -2.79768854e-01 1.32874891e-01 -3.71167094e-01
-5.84639004e-03 4.42691743e-02 -6.10453188e-01 3.95778686e-01
5.62214702e-02 8.45988691e-01 -1.05782855e+00 -1.06723034e+00
3.10996532e-01 6.37212932e-01 -4.72379476e-01 1.97303832e-01
-2.13123098e-01 3.44714701e-01 -5.15124798e-01 9.40702379e-01
3.59993815e-01 -4.27205175e-01 -5.65752909e-02 -5.92927158e-01
1.04553578e-02 1.14127055e-01 -1.08015716e+00 2.26059055e+00
-3.33589733e-01 4.16688532e-01 -3.28162700e-01 -9.79386449e-01
7.19326138e-01 5.79957962e-01 6.60881698e-01 -7.97567904e-01
1.56106679e-02 1.81045085e-01 -6.44585252e-01 -7.05291867e-01
7.59044290e-01 2.75749683e-01 -1.43040940e-01 3.02075565e-01
4.58302237e-02 5.15860021e-01 2.84561038e-01 4.24658716e-01
1.05687702e+00 2.83065706e-01 8.59597921e-02 2.73795336e-01
7.61355758e-01 -1.44501731e-01 8.51138771e-01 5.84242284e-01
-2.59418368e-01 4.93695229e-01 1.76733226e-01 -3.32828492e-01
-6.38307154e-01 -8.88476670e-01 5.43159544e-02 1.31977630e+00
8.39499116e-01 -7.64317572e-01 -4.25820857e-01 -5.37397742e-01
-4.22755361e-01 1.75159037e-01 -2.93319046e-01 -4.53569293e-02
-7.39053190e-01 -5.88270664e-01 4.95740205e-01 3.96734983e-01
6.16460502e-01 -8.14046860e-01 -5.28872132e-01 -7.56116025e-03
-9.97731805e-01 -1.37754095e+00 -9.38014030e-01 -5.73303401e-01
-9.27879274e-01 -9.45154011e-01 -9.22182977e-01 -7.64005780e-01
6.98108971e-01 8.44624400e-01 9.69881654e-01 2.35960379e-01
-1.70996681e-01 5.51560402e-01 -7.33283639e-01 1.65003091e-01
4.56836194e-01 6.45641163e-02 -3.05826608e-02 1.38944477e-01
5.02730310e-01 -4.45970327e-01 -1.03960228e+00 6.34496331e-01
-1.22198546e+00 -2.76196413e-02 5.98442256e-01 7.77715027e-01
9.52367127e-01 1.64148346e-01 5.34374774e-01 -3.74065846e-01
8.02008584e-02 -4.39164609e-01 -5.84730864e-01 5.62035024e-01
-3.08008641e-01 -1.57676324e-01 2.60603428e-01 -7.17471778e-01
-1.01618242e+00 -4.42681201e-02 1.37550667e-01 -7.93153346e-01
2.68235683e-01 4.79616106e-01 -4.27123308e-02 7.23512694e-02
6.64841011e-02 5.25610924e-01 -3.42715323e-01 -4.60837573e-01
3.10934812e-01 6.44325912e-01 6.24280930e-01 -7.40687788e-01
6.38607681e-01 7.51188338e-01 -4.67044115e-02 -5.58484554e-01
-8.85744095e-01 -8.72778594e-01 -3.60743314e-01 -3.91348660e-01
8.08238447e-01 -1.24158323e+00 -1.21340537e+00 1.82574749e-01
-1.17255032e+00 1.59335554e-01 1.84959635e-01 6.21680379e-01
-4.96398181e-01 6.36819482e-01 -7.44820476e-01 -8.10332835e-01
-5.58948755e-01 -1.11049879e+00 1.68131793e+00 1.71713695e-01
1.33008063e-01 -6.16143703e-01 -8.66337270e-02 7.40498722e-01
6.02679290e-02 1.04430295e-03 6.34274483e-01 -3.24405462e-01
-9.97607648e-01 -3.36355090e-01 -3.69242489e-01 -2.03294903e-01
-1.42704323e-01 4.05906551e-02 -5.13552904e-01 -2.72234261e-01
-2.36480281e-01 -2.85968035e-01 6.41903102e-01 4.20758128e-02
1.24056709e+00 -1.46227852e-01 -4.82603788e-01 3.02833587e-01
1.30607891e+00 2.62314081e-01 6.36341155e-01 3.56786370e-01
6.61946893e-01 4.66192067e-01 1.39817894e+00 6.24143422e-01
6.41482413e-01 9.78247046e-01 2.48130649e-01 1.75557882e-01
2.85172760e-01 -3.93895417e-01 4.28291559e-01 1.27869511e+00
-2.57009089e-01 -2.50674218e-01 -6.43252432e-01 6.85544372e-01
-2.39944816e+00 -1.10886860e+00 2.40562439e-01 2.23481083e+00
7.58082211e-01 -9.26997066e-02 4.57649752e-02 2.28823274e-01
7.15888262e-01 3.55801642e-01 -3.82206500e-01 3.12131733e-01
7.15577900e-02 -1.47478297e-01 1.72006071e-01 2.51368880e-01
-1.07759535e+00 8.21365237e-01 5.44236851e+00 1.26861644e+00
-9.11814511e-01 3.46399426e-01 2.57292002e-01 -5.04516184e-01
-8.89420584e-02 -1.10242970e-01 -8.75129402e-01 5.67604899e-01
6.10360920e-01 1.20971493e-01 4.83788371e-01 7.15595067e-01
3.92558932e-01 -2.26253852e-01 -1.13146889e+00 1.57852554e+00
3.84850770e-01 -1.18082309e+00 3.20378304e-01 -2.15295121e-01
4.12451059e-01 -1.31464213e-01 -1.22575223e-01 2.21932724e-01
-4.28050816e-01 -4.94791389e-01 7.08086967e-01 7.20183611e-01
7.15413392e-01 -9.96594846e-01 7.40869462e-01 2.80158073e-01
-1.61586332e+00 8.22741464e-02 -4.02732462e-01 3.94250900e-01
3.94786298e-01 4.78684306e-01 -1.46068230e-01 8.19795370e-01
8.74052584e-01 7.78931916e-01 -4.07466412e-01 7.81422377e-01
2.06133071e-02 1.19866580e-01 -3.65596890e-01 2.07080081e-01
1.85891896e-01 -2.02188283e-01 6.35090590e-01 1.07144332e+00
3.46712083e-01 2.57751286e-01 2.78494954e-01 2.68681109e-01
-1.34393930e-01 2.00606301e-01 -3.30297053e-01 -2.49349233e-02
7.11639047e-01 1.19748449e+00 -7.07775772e-01 -3.66529763e-01
-5.01024127e-01 1.18740666e+00 2.56016761e-01 3.47363502e-01
-1.22353745e+00 -6.66027248e-01 2.99382299e-01 -6.43174425e-02
3.11898589e-01 -1.58380270e-01 5.06591737e-01 -1.48696220e+00
4.40190226e-01 -7.90272057e-01 6.26502395e-01 -1.00835991e+00
-1.09624588e+00 4.04974312e-01 9.99978185e-02 -1.53644395e+00
-6.25136420e-02 -5.21013476e-02 -1.53241754e-01 3.04684341e-01
-1.57221496e+00 -1.15686059e+00 -4.54123199e-01 9.18726623e-01
5.78437805e-01 1.06243387e-01 6.54464364e-01 9.90019619e-01
-6.93405151e-01 6.19940758e-01 -1.18661053e-01 2.05858976e-01
9.31710005e-01 -7.81116128e-01 -1.51968554e-01 5.29209554e-01
1.96457610e-01 8.53528380e-01 2.67974764e-01 -4.15092707e-01
-1.75284016e+00 -9.38896000e-01 9.23772037e-01 -3.61160517e-01
6.10065818e-01 -6.80650249e-02 -7.77074516e-01 6.15860343e-01
-1.55201867e-01 2.28905585e-02 6.97707832e-01 -6.84142858e-02
-5.18361330e-01 -5.16520619e-01 -7.91684389e-01 7.63589859e-01
1.04424047e+00 -8.38577628e-01 -7.14712679e-01 4.84022886e-01
1.03229332e+00 -5.63964427e-01 -1.19076157e+00 6.34634376e-01
5.80835164e-01 -6.50759816e-01 1.10938430e+00 -5.62420003e-02
4.33696419e-01 -7.25197792e-01 -2.83553958e-01 -4.31152105e-01
1.68441966e-01 -7.39860237e-01 -5.58930516e-01 1.49832201e+00
-5.10232523e-02 -3.53113741e-01 3.90548468e-01 4.31393147e-01
2.03336790e-01 -8.49810004e-01 -9.85983849e-01 -5.97591519e-01
-7.72067487e-01 -3.05987895e-01 4.73710626e-01 8.71542633e-01
3.31748612e-02 3.55353475e-01 -5.26578903e-01 3.84515971e-01
6.41860902e-01 3.89892191e-01 6.09476507e-01 -9.15809035e-01
-3.41809034e-01 -7.37677664e-02 -5.49465477e-01 -1.55013621e+00
1.67719759e-02 -4.79555875e-01 8.95384774e-02 -1.24154449e+00
5.33095002e-01 -3.14124942e-01 -4.51332331e-01 3.41320634e-01
-1.37738630e-01 3.77030492e-01 1.27258822e-01 7.10379958e-01
-1.46938181e+00 6.38719857e-01 1.11064744e+00 -2.36246046e-02
-6.57115057e-02 -2.94656157e-01 -1.99251696e-01 6.29468858e-01
3.18176478e-01 -3.97193730e-01 -5.84667981e-01 -7.79619217e-01
4.78964925e-01 5.02715409e-01 4.02818084e-01 -8.22692037e-01
4.74000156e-01 -5.01632653e-02 1.31968126e-01 -1.16168582e+00
6.19463086e-01 -8.62916410e-01 1.99435487e-01 -1.52788147e-01
-3.02960753e-01 2.55588293e-01 -9.53217968e-02 9.45900202e-01
-6.68034852e-01 -1.74591571e-01 3.12904686e-01 3.25004868e-02
-7.29336977e-01 4.55223471e-01 -2.64901575e-02 1.52873462e-02
1.08522856e+00 1.37826994e-01 -2.12262526e-01 -2.93506414e-01
-4.35922056e-01 4.26813960e-01 3.27538669e-01 5.74533284e-01
6.94946170e-01 -1.50661743e+00 -1.69063896e-01 -3.96491438e-01
3.60448450e-01 3.36671680e-01 5.20572841e-01 9.66397226e-01
-4.45188552e-01 6.57673120e-01 4.02533501e-01 -8.65523338e-01
-1.29928339e+00 7.24165916e-01 -1.04299575e-01 -2.23121598e-01
-7.49476552e-01 5.53997219e-01 1.12832859e-01 -6.89009652e-02
5.16867936e-01 1.40407398e-01 -1.54861107e-01 2.07014233e-01
7.74738789e-01 5.03648639e-01 -1.08191863e-01 -7.87483156e-01
-4.54608172e-01 9.11290824e-01 -2.99312472e-01 -1.43465221e-01
1.12412333e+00 -4.04736042e-01 -2.67032087e-01 4.77596968e-01
1.23073781e+00 -1.69821233e-01 -9.97749507e-01 -4.54733878e-01
4.66984101e-02 -6.70826256e-01 1.18762121e-01 -3.43971431e-01
-1.15226114e+00 6.93661988e-01 4.37133074e-01 -1.67662442e-01
1.37496495e+00 -3.00266966e-02 1.32104397e+00 6.41269028e-01
6.30909741e-01 -1.19184422e+00 2.12953925e-01 1.60538569e-01
6.15157366e-01 -1.31086361e+00 1.27407238e-01 -3.89419526e-01
-5.70761681e-01 8.37298453e-01 4.73408580e-01 4.61854823e-02
4.87393707e-01 -3.26744318e-01 -1.68978050e-01 -2.77321756e-01
-8.41666996e-01 -1.36904463e-01 3.91877830e-01 8.76613483e-02
1.74367383e-01 -3.15379769e-01 -4.77231681e-01 4.90709901e-01
4.22023743e-01 8.01730826e-02 -1.09696992e-01 1.05497861e+00
-1.86182544e-01 -8.66899967e-01 -2.08678544e-01 1.12736963e-01
-5.99138498e-01 -2.79232673e-02 3.62526923e-02 6.44615233e-01
-1.57406375e-01 1.01360309e+00 -7.81913474e-03 -2.50729710e-01
2.28038594e-01 -1.32193953e-01 3.42683136e-01 -2.46008307e-01
-2.78906077e-01 2.94449478e-01 -2.55293310e-01 -8.57865751e-01
-9.37795699e-01 -4.38157439e-01 -1.26993477e+00 -2.59649426e-01
-5.80289721e-01 3.14090252e-01 4.17439431e-01 1.00565720e+00
5.63308060e-01 2.74077982e-01 6.60497010e-01 -9.01890635e-01
-2.84407079e-01 -8.03614020e-01 -2.46771500e-01 3.99840564e-01
2.08468005e-01 -7.90435016e-01 -2.73593485e-01 2.96018362e-01] | [10.125072479248047, 0.7333472371101379] |
1dea319c-3c77-40a6-8753-4674b3ec730e | event-based-monocular-dense-depth-estimation | 2212.02791 | null | https://arxiv.org/abs/2212.02791v1 | https://arxiv.org/pdf/2212.02791v1.pdf | Event-based Monocular Dense Depth Estimation with Recurrent Transformers | Event cameras, offering high temporal resolutions and high dynamic ranges, have brought a new perspective to address common challenges (e.g., motion blur and low light) in monocular depth estimation. However, how to effectively exploit the sparse spatial information and rich temporal cues from asynchronous events remains a challenging endeavor. To this end, we propose a novel event-based monocular depth estimator with recurrent transformers, namely EReFormer, which is the first pure transformer with a recursive mechanism to process continuous event streams. Technically, for spatial modeling, a novel transformer-based encoder-decoder with a spatial transformer fusion module is presented, having better global context information modeling capabilities than CNN-based methods. For temporal modeling, we design a gate recurrent vision transformer unit that introduces a recursive mechanism into transformers, improving temporal modeling capabilities while alleviating the expensive GPU memory cost. The experimental results show that our EReFormer outperforms state-of-the-art methods by a margin on both synthetic and real-world datasets. We hope that our work will attract further research to develop stunning transformers in the event-based vision community. Our open-source code can be found in the supplemental material. | ['Yonghong Tian', 'Xiaopeng Fan', 'Jianing Li', 'Xu Liu'] | 2022-12-06 | null | null | null | null | ['event-based-vision'] | ['computer-vision'] | [ 2.19732106e-01 -4.55738962e-01 -3.88487987e-02 -2.96601862e-01
-4.51639414e-01 -2.59910256e-01 6.39174759e-01 -3.62971753e-01
-3.09567422e-01 5.67737937e-01 2.38123670e-01 -6.45035282e-02
1.99644327e-01 -7.52019644e-01 -6.62727654e-01 -7.86564648e-01
1.98365256e-01 -2.59168893e-01 8.03971291e-01 3.16222697e-01
2.81371564e-01 3.68157238e-01 -1.78473008e+00 4.14218664e-01
9.58774328e-01 1.27012289e+00 6.70930624e-01 7.38566577e-01
7.73733631e-02 1.52061903e+00 -1.27330348e-01 -2.79667765e-01
1.92524418e-02 -4.19923544e-01 -1.65916905e-01 -5.30050173e-02
4.88705814e-01 -7.81878591e-01 -9.06597078e-01 7.79629707e-01
6.67825162e-01 -1.60315856e-01 2.16420203e-01 -1.00540519e+00
-2.39841297e-01 9.23394337e-02 -5.29659152e-01 3.66708100e-01
4.20792162e-01 4.24391091e-01 7.02516973e-01 -1.07365286e+00
6.27050996e-01 1.01012754e+00 3.67442459e-01 4.52553600e-01
-9.08787847e-01 -7.64388978e-01 3.99861276e-01 7.73808181e-01
-1.17001283e+00 -6.02754593e-01 7.06149697e-01 -3.48583370e-01
1.20756483e+00 7.69931497e-03 9.55792725e-01 1.28764772e+00
4.15518582e-01 1.07782149e+00 9.63922322e-01 -1.26837194e-02
3.92733485e-01 -2.80734390e-01 -1.58404410e-01 7.49589980e-01
-7.55382609e-03 5.02730489e-01 -1.21796501e+00 2.99452454e-01
1.05169713e+00 2.84202188e-01 -6.12226248e-01 -4.22106326e-01
-1.35307300e+00 4.98723924e-01 5.20584226e-01 1.97149105e-02
-4.16414678e-01 5.64508200e-01 1.96204618e-01 4.41446528e-03
5.82612097e-01 -1.79196894e-02 -3.26794386e-01 -5.23128033e-01
-9.61772382e-01 5.54202422e-02 5.86576939e-01 9.83820677e-01
5.93554318e-01 9.62637886e-02 -1.63215086e-01 5.11769414e-01
3.78178239e-01 4.88318771e-01 3.46166730e-01 -1.14006817e+00
2.89158553e-01 3.85807455e-01 5.08021973e-02 -6.57828987e-01
-2.37489313e-01 -4.34109479e-01 -8.24935973e-01 2.29657486e-01
2.01201588e-01 -1.75599456e-02 -7.32407153e-01 1.62955058e+00
3.62042129e-01 9.13193822e-01 -1.61272630e-01 1.01766396e+00
9.48369086e-01 7.69614458e-01 -3.19704801e-01 -3.98541033e-01
1.38433421e+00 -9.81423557e-01 -8.37218881e-01 -3.72684240e-01
1.85463391e-02 -9.05252218e-01 7.40805030e-01 5.25994301e-01
-1.23991740e+00 -4.42163944e-01 -1.08518374e+00 -4.17164892e-01
9.00809914e-02 1.42359734e-01 9.23197567e-01 2.36447409e-01
-1.07128978e+00 3.45909029e-01 -1.38284278e+00 -4.27093625e-01
3.04585367e-01 4.90944795e-02 -9.88371223e-02 -3.23396266e-01
-9.42130446e-01 5.63975215e-01 5.31321876e-02 6.86466098e-02
-1.14471638e+00 -7.39857435e-01 -1.07100916e+00 9.05470550e-03
3.47404480e-01 -1.09418905e+00 1.56680870e+00 -3.30110520e-01
-1.78028691e+00 5.72363913e-01 -7.62383044e-01 -5.63788474e-01
5.79024434e-01 -2.54083693e-01 1.96722504e-02 3.01084578e-01
2.35618483e-02 6.37710512e-01 9.07145441e-01 -8.79147649e-01
-8.38907897e-01 -2.83069968e-01 2.24458784e-01 3.07182640e-01
-2.93695837e-01 -9.50204507e-02 -7.17862368e-01 -6.39619350e-01
4.27926481e-01 -6.60393417e-01 -1.22313738e-01 4.28890824e-01
7.05532134e-02 1.53564066e-01 7.82671750e-01 -2.50437289e-01
1.26454771e+00 -2.14513159e+00 2.39968479e-01 -5.87206244e-01
4.07210261e-01 7.60550424e-02 2.62824088e-01 5.06073058e-01
2.34605759e-01 -5.63098133e-01 -1.10484645e-01 -7.51053154e-01
-2.71160990e-01 1.27463698e-01 -5.23796618e-01 3.44497621e-01
5.37481084e-02 9.02374387e-01 -1.12963712e+00 -3.88357133e-01
7.80985892e-01 6.10735416e-01 -5.83837688e-01 2.49718547e-01
-2.52184123e-01 4.93667901e-01 -2.00849921e-01 7.75589943e-01
7.59641111e-01 -5.82931995e-01 -1.24692082e-01 -3.70827556e-01
-4.72912878e-01 4.78543162e-01 -1.13058329e+00 1.99617171e+00
-6.11766696e-01 9.45913553e-01 -7.92071670e-02 -3.41689557e-01
6.20010316e-01 2.49988124e-01 4.17113274e-01 -8.67761433e-01
-1.51261000e-03 2.15321437e-01 -3.87639761e-01 -4.72072303e-01
7.38778949e-01 -4.75759134e-02 3.56902570e-01 1.34408027e-01
-7.09516332e-02 -1.71775654e-01 -1.75959673e-02 3.47550094e-01
1.35845292e+00 3.51713449e-01 1.33237988e-01 2.71483809e-01
3.52458775e-01 -2.83758730e-01 8.37467492e-01 5.55324256e-01
-3.56880605e-01 9.13518369e-01 2.53591835e-01 -3.51451039e-01
-7.96743810e-01 -1.32364964e+00 -9.97562185e-02 4.64229345e-01
5.11881173e-01 -7.03761101e-01 -2.37501413e-01 -1.04930133e-01
-2.88026839e-01 3.20151478e-01 -3.79741222e-01 1.83083825e-02
-5.49297631e-01 -6.47331655e-01 2.69067794e-01 7.98646450e-01
9.75019872e-01 -6.24698043e-01 -1.06120765e+00 4.71665144e-01
-5.65630972e-01 -1.57436049e+00 -4.89289433e-01 2.47660384e-01
-1.02061605e+00 -9.48290229e-01 -7.80615330e-01 -7.02030122e-01
2.18437061e-01 7.81281412e-01 9.77245808e-01 -4.19127643e-01
-2.35833883e-01 3.58993500e-01 -2.64880329e-01 -2.44997576e-01
3.63812566e-01 -1.79027706e-01 -3.85340780e-01 1.03624687e-01
2.83640593e-01 -8.65044236e-01 -1.20384347e+00 3.80956322e-01
-9.10255611e-01 7.28028059e-01 4.16008234e-01 8.60882342e-01
5.94025195e-01 -1.09296709e-01 1.13735631e-01 -3.75270277e-01
-6.42617419e-02 -3.33835393e-01 -9.06877160e-01 -1.72623232e-01
-3.26547295e-01 -1.55881405e-01 5.06781578e-01 -3.70239884e-01
-1.51186240e+00 1.76133558e-01 -1.51773700e-02 -7.28293896e-01
1.69844314e-01 3.19937974e-01 -1.18302880e-02 -2.68378202e-02
4.43489492e-01 6.45428896e-01 -3.36423844e-01 -3.16424370e-01
1.82359457e-01 4.06060576e-01 7.48878419e-01 -3.12291652e-01
4.81914997e-01 1.01359951e+00 8.44443142e-02 -6.38430595e-01
-6.95396900e-01 -6.31057203e-01 -1.46658182e-01 -4.78869081e-01
8.37463915e-01 -1.61220169e+00 -7.73858666e-01 1.07534158e+00
-1.47985864e+00 -4.64615047e-01 -1.37222543e-01 7.38936424e-01
-7.10705042e-01 4.16951239e-01 -9.61972535e-01 -7.17622519e-01
-1.13595486e-01 -1.12305772e+00 1.40280724e+00 4.40548420e-01
1.99098751e-01 -7.00857878e-01 6.30550236e-02 3.33010972e-01
4.08737034e-01 5.23144044e-02 2.71998376e-01 4.57941502e-01
-1.65952992e+00 2.96333402e-01 -5.29151082e-01 9.14727300e-02
4.92478833e-02 -1.50924593e-01 -1.33968282e+00 -1.10444210e-01
4.17561978e-01 -1.80423856e-02 1.18550134e+00 5.86573362e-01
1.01775789e+00 2.21659854e-01 -3.41395438e-01 1.14547932e+00
1.54877067e+00 4.51493531e-01 1.02505612e+00 2.91462451e-01
8.48284245e-01 2.18238056e-01 7.04373538e-01 9.29400206e-01
7.32767880e-01 7.43168890e-01 5.68026125e-01 4.30938862e-02
-4.16118562e-01 -4.04834539e-01 5.75764894e-01 8.06004643e-01
8.11032131e-02 -4.35236216e-01 -6.37495339e-01 6.72695994e-01
-2.24999928e+00 -1.14404500e+00 -2.40242258e-01 1.93768764e+00
6.88526928e-01 7.09874332e-02 -4.26014066e-01 6.83266446e-02
3.45066756e-01 6.10736132e-01 -7.53559470e-01 1.26945615e-01
-2.55244315e-01 1.90357029e-01 4.28857595e-01 4.83410299e-01
-1.01001596e+00 9.66821730e-01 5.15288305e+00 7.23825037e-01
-1.31425214e+00 3.02729923e-02 3.35076123e-01 -4.41873014e-01
-2.25955814e-01 1.15534060e-01 -9.65000987e-01 6.70667291e-01
7.34813213e-01 -1.44512698e-01 3.60843778e-01 5.78726768e-01
6.66163683e-01 -5.79981685e-01 -1.24029434e+00 1.53606570e+00
9.58479941e-02 -1.51773667e+00 -2.51178950e-01 4.65456359e-02
7.76347399e-01 2.28244513e-01 -2.98787951e-02 -1.13099992e-01
1.24512292e-01 -6.70497119e-01 8.12491834e-01 7.46663928e-01
7.06468403e-01 -4.39001530e-01 4.30214673e-01 2.90475786e-01
-1.71044481e+00 6.48228303e-02 -2.28430420e-01 -3.59324068e-01
5.65001786e-01 1.01393902e+00 -4.65279162e-01 5.68433464e-01
8.90632987e-01 1.50652385e+00 -3.30292851e-01 1.38115692e+00
-3.36531848e-01 3.68615568e-01 -3.39502096e-01 1.69320241e-01
-3.44703794e-02 3.32264751e-02 5.65217435e-01 8.79907846e-01
5.22842348e-01 2.23291770e-01 -1.20921113e-01 7.96495140e-01
1.40388846e-01 -5.28974116e-01 -6.71381593e-01 2.72113681e-01
5.27025580e-01 1.01477396e+00 -4.83192980e-01 -2.90233880e-01
-7.37887621e-01 1.30920792e+00 5.32138608e-02 3.15768898e-01
-1.06449628e+00 -2.52899438e-01 9.23121929e-01 8.93490613e-02
5.87075531e-01 -5.59907913e-01 -5.38024783e-01 -1.80649889e+00
3.37479979e-01 -5.74173510e-01 6.57233298e-02 -1.27782214e+00
-9.90332246e-01 4.72772539e-01 -2.17763200e-01 -1.56113791e+00
-3.00008744e-01 -4.59491372e-01 -3.59969437e-01 6.59675419e-01
-2.06419802e+00 -1.02908587e+00 -9.68623757e-01 8.47373486e-01
8.16574693e-01 1.89954445e-01 2.83485770e-01 5.84454000e-01
-3.67139488e-01 1.63873121e-01 -4.07388918e-02 -2.42403358e-01
7.87056029e-01 -1.07953250e+00 3.94109517e-01 1.30125129e+00
2.12801956e-02 3.10790479e-01 4.73409802e-01 -6.00438237e-01
-1.75424826e+00 -1.16838241e+00 9.75697875e-01 -3.03882062e-01
5.75087667e-01 -3.38825792e-01 -6.55162871e-01 7.10349143e-01
2.16180429e-01 7.97368288e-02 2.35412121e-01 -3.82246673e-01
-1.47487029e-01 -3.90026808e-01 -7.36361325e-01 6.92223430e-01
1.45370436e+00 -7.19806254e-01 -2.17639267e-01 1.64192915e-01
8.51942897e-01 -7.82035351e-01 -6.45577133e-01 4.82348442e-01
5.10007560e-01 -1.47002876e+00 9.64451015e-01 5.42428732e-01
7.12442636e-01 -6.74091816e-01 -1.76649809e-01 -8.46299350e-01
-8.37195218e-02 -7.16842651e-01 -6.20782852e-01 1.02278364e+00
-1.99946284e-01 -7.46627927e-01 8.92523885e-01 2.75497615e-01
-3.22863609e-01 -7.87052095e-01 -1.02161765e+00 -7.59771943e-01
-6.63549483e-01 -7.55379617e-01 3.15859675e-01 4.26132619e-01
-2.86483914e-01 3.01458031e-01 -5.38785875e-01 2.70443976e-01
8.25790286e-01 1.44665360e-01 7.35202670e-01 -8.66538167e-01
-5.01511574e-01 -4.48765308e-02 -4.70598012e-01 -2.00344276e+00
-2.67144531e-01 -2.83307374e-01 3.18969488e-02 -1.52102530e+00
1.66193515e-01 -1.11595556e-01 3.54414172e-02 9.05883908e-02
-5.03407419e-02 2.46446639e-01 2.34687597e-01 1.22590862e-01
-6.82845235e-01 1.06648958e+00 1.39867997e+00 5.46206869e-02
9.23745614e-03 -2.94718802e-01 -3.16942304e-01 7.20162690e-01
3.08159620e-01 -2.41923764e-01 -7.17008412e-01 -7.29903460e-01
1.93473011e-01 5.12993872e-01 7.26049185e-01 -1.36965442e+00
8.96396041e-01 -8.29715952e-02 3.52717459e-01 -9.84853625e-01
7.98011541e-01 -8.12499464e-01 3.91949773e-01 3.48588377e-01
1.80729643e-01 1.03425331e-01 1.92441687e-01 7.47311175e-01
-5.77485561e-01 2.10198253e-01 7.88063645e-01 -1.16121560e-01
-1.07816446e+00 6.04778826e-01 -3.41707468e-01 -7.70847723e-02
1.04634762e+00 -3.08236897e-01 -5.43127477e-01 -5.60767889e-01
-3.88498664e-01 1.81473196e-01 7.26611316e-01 2.23867744e-01
9.96663511e-01 -1.19754672e+00 -4.72348690e-01 3.19210321e-01
1.33256778e-01 2.16964006e-01 4.94660318e-01 8.41242552e-01
-5.36363482e-01 4.79583621e-01 -6.84513450e-02 -9.96123314e-01
-1.21083117e+00 3.11298400e-01 4.33476061e-01 -2.02798426e-01
-7.92912900e-01 9.37495768e-01 7.06790686e-01 3.04806083e-01
2.73771048e-01 -6.36346936e-01 1.62453696e-01 -9.37759131e-02
7.02816367e-01 4.03379202e-01 5.25141843e-02 -9.15648714e-02
-3.31845611e-01 7.37471044e-01 -2.03693733e-02 -2.53904343e-01
1.25397551e+00 -4.55473900e-01 1.01508453e-01 4.87952262e-01
9.09587204e-01 -1.49232537e-01 -1.84210515e+00 -3.93707097e-01
-5.10280550e-01 -8.79720688e-01 3.33305329e-01 -6.19583428e-01
-1.12973738e+00 1.05186963e+00 6.66501641e-01 -2.26465926e-01
1.65456963e+00 -1.92656800e-01 1.02689540e+00 1.04158223e-02
7.52661645e-01 -5.45081019e-01 2.94180214e-01 7.39551008e-01
7.02505350e-01 -1.31820560e+00 -2.15005115e-01 -6.96558058e-01
-3.71349126e-01 1.06337166e+00 7.82033086e-01 3.85714583e-02
6.64745688e-01 4.44727093e-01 -1.74627542e-01 -1.17037445e-01
-1.31031334e+00 -3.20155531e-01 -3.50268595e-02 4.81907845e-01
3.41168880e-01 -3.09042007e-01 -1.68759376e-01 3.03836942e-01
3.35336067e-02 6.11682117e-01 5.32183707e-01 8.74088645e-01
-2.12987751e-01 -1.05392492e+00 -1.17010213e-01 1.15074128e-01
-1.53539628e-01 -4.76058692e-01 1.93820372e-01 4.11739618e-01
1.25911161e-01 1.02304673e+00 2.18610063e-01 -3.11075985e-01
2.60626167e-01 -4.10109311e-01 7.15599120e-01 -4.42728937e-01
-2.73516923e-01 1.85456514e-01 2.83024833e-02 -9.59981918e-01
-5.12425125e-01 -7.55828977e-01 -1.00619960e+00 -4.81510639e-01
-2.24059671e-02 -5.67090154e-01 4.89072978e-01 6.99429154e-01
5.71728885e-01 6.91081345e-01 5.17407835e-01 -9.89129841e-01
9.36105624e-02 -6.07458115e-01 -2.84939378e-01 -1.06126070e-01
4.56453025e-01 -7.55095124e-01 -3.86020988e-01 1.42390281e-01] | [9.109579086303711, -1.7504867315292358] |
9f7e3c1e-cd2b-44b1-8b50-ebe62c7fe390 | slideflow-deep-learning-for-digital | 2304.04142 | null | https://arxiv.org/abs/2304.04142v1 | https://arxiv.org/pdf/2304.04142v1.pdf | Slideflow: Deep Learning for Digital Histopathology with Real-Time Whole-Slide Visualization | Deep learning methods have emerged as powerful tools for analyzing histopathological images, but current methods are often specialized for specific domains and software environments, and few open-source options exist for deploying models in an interactive interface. Experimenting with different deep learning approaches typically requires switching software libraries and reprocessing data, reducing the feasibility and practicality of experimenting with new architectures. We developed a flexible deep learning library for histopathology called Slideflow, a package which supports a broad array of deep learning methods for digital pathology and includes a fast whole-slide interface for deploying trained models. Slideflow includes unique tools for whole-slide image data processing, efficient stain normalization and augmentation, weakly-supervised whole-slide classification, uncertainty quantification, feature generation, feature space analysis, and explainability. Whole-slide image processing is highly optimized, enabling whole-slide tile extraction at 40X magnification in 2.5 seconds per slide. The framework-agnostic data processing pipeline enables rapid experimentation with new methods built with either Tensorflow or PyTorch, and the graphical user interface supports real-time visualization of slides, predictions, heatmaps, and feature space characteristics on a variety of hardware devices, including ARM-based devices such as the Raspberry Pi. | ['Alexander T. Pearson', 'Prajval Mohan', 'Anran Li', 'Frederick M. Howard', 'Matteo Sacco', 'Andrew Srisuwananukorn', 'Emma Dyer', 'Sara Kochanny', 'James M. Dolezal'] | 2023-04-09 | null | null | null | null | ['whole-slide-images', 'histopathological-segmentation', 'histopathological-image-classification', 'multiple-instance-learning'] | ['computer-vision', 'computer-vision', 'medical', 'methodology'] | [-3.13122690e-01 -3.52887243e-01 2.86421441e-02 -4.60080266e-01
-1.20290279e+00 -5.94920456e-01 2.32860614e-02 5.86005390e-01
-4.83272374e-01 2.67018080e-01 -2.11749569e-01 -6.68878138e-01
-1.42552435e-01 -6.34541988e-01 -1.04869872e-01 -1.08276796e+00
-3.11406970e-01 6.20575190e-01 1.53009072e-01 2.98799723e-01
9.42349732e-02 8.80838990e-01 -1.28784156e+00 5.32830715e-01
1.83797926e-01 9.54645097e-01 -1.66014940e-01 1.43722284e+00
-3.22877973e-01 2.97101468e-01 -6.17759347e-01 -1.30607873e-01
3.97568978e-02 2.47867212e-01 -6.02536678e-01 -2.26967528e-01
2.69662648e-01 -4.52282608e-01 -5.35943499e-03 7.70609915e-01
8.88954759e-01 -5.32831132e-01 3.79784048e-01 -1.24908483e+00
-1.02347575e-01 2.11823195e-01 -5.51916063e-01 5.64426959e-01
-9.89698619e-02 7.72174001e-01 6.74739480e-01 -5.53779900e-01
6.22597754e-01 8.81805003e-01 1.19467092e+00 3.44061613e-01
-1.36327267e+00 -5.92472374e-01 -7.82275498e-01 -5.34177795e-02
-1.58842957e+00 -2.56451041e-01 1.58618674e-01 -8.20192277e-01
1.15877771e+00 6.38975918e-01 6.28762960e-01 7.31300950e-01
1.01247072e+00 5.97484112e-01 1.03199792e+00 -1.27131075e-01
3.01293939e-01 2.42527306e-01 2.30415404e-01 8.75551462e-01
2.08684415e-01 -1.66445285e-01 -1.99396551e-01 -5.38549960e-01
8.40007901e-01 3.07997257e-01 8.15712512e-02 -9.39954221e-02
-1.31450915e+00 6.61990106e-01 2.02292383e-01 3.08501810e-01
6.13754466e-02 1.52191907e-01 8.29602897e-01 1.74497575e-01
5.59640884e-01 5.04733801e-01 -4.55426633e-01 1.53172913e-03
-9.12658334e-01 1.21590853e-01 5.22265613e-01 3.82717669e-01
6.59434736e-01 -4.02974695e-01 -1.13181300e-01 4.78667676e-01
1.81219861e-01 2.86311448e-01 9.24371243e-01 -6.97008371e-01
-3.66743982e-01 8.44209433e-01 -2.73853689e-01 -8.50201488e-01
-1.30980706e+00 -4.76267844e-01 -8.75781894e-01 4.79357749e-01
4.95971262e-01 -1.49263844e-01 -8.87933969e-01 8.80930066e-01
7.00867534e-01 -4.52965684e-02 -3.52512330e-01 6.27219677e-01
9.60037649e-01 2.12721869e-01 6.63866475e-02 2.08711132e-01
1.54426563e+00 -4.40723091e-01 -5.69183946e-01 3.02099168e-01
1.38715732e+00 -7.03778803e-01 1.30960679e+00 3.64388645e-01
-6.82955384e-01 5.18734828e-02 -8.36883783e-01 -5.38587868e-01
-7.46125102e-01 1.87675029e-01 8.32586050e-01 6.93903208e-01
-1.38792694e+00 6.66925669e-01 -1.58780181e+00 -5.56246758e-01
9.65048790e-01 8.03415835e-01 -7.59175301e-01 1.59051850e-01
-4.11925822e-01 7.64178455e-01 1.34031111e-02 -1.23041883e-01
-5.17382205e-01 -1.19646454e+00 -8.45716417e-01 1.41979799e-01
-4.91863966e-01 -6.97572470e-01 1.11976469e+00 -2.58760333e-01
-1.49419940e+00 1.18890333e+00 1.69645727e-01 -1.73381150e-01
3.14279020e-01 4.43244427e-01 -1.95168242e-01 1.65774766e-02
-1.07004941e-01 6.50263309e-01 3.51749331e-01 -4.28379267e-01
-4.82928932e-01 -5.04134417e-01 -4.68868822e-01 -2.31205061e-01
-5.29487252e-01 1.14790015e-01 -2.01326430e-01 -1.25241160e-01
-2.58426279e-01 -7.18824804e-01 -4.04921055e-01 5.99616647e-01
-4.61992890e-01 5.93345053e-02 9.51145351e-01 -4.93955225e-01
9.45798278e-01 -2.24099422e+00 -5.98329306e-01 4.21189994e-01
4.07422334e-01 2.14869440e-01 -1.19759060e-01 1.48353502e-01
-1.38706550e-01 3.34924012e-01 2.31792584e-01 -3.17935318e-01
5.68202548e-02 -2.11680487e-01 2.67957479e-01 7.91872263e-01
9.36116278e-02 7.82718658e-01 -6.66310608e-01 -8.75413477e-01
3.38951349e-01 6.97592139e-01 -5.34224212e-01 2.36208797e-01
1.26079619e-01 2.94238389e-01 -1.97867364e-01 1.01329148e+00
8.53584528e-01 -7.65996397e-01 3.05729091e-01 -1.61950573e-01
-1.25721768e-01 1.17010181e-03 -1.00782669e+00 1.66586578e+00
-4.90465492e-01 8.94338250e-01 2.83755124e-01 -4.01362330e-01
6.49336755e-01 -5.60641438e-02 3.96188736e-01 5.08359298e-02
5.16167581e-01 9.55643877e-02 -1.21156193e-01 -8.25139701e-01
7.25749731e-02 -1.27001069e-02 2.68441111e-01 8.34856570e-01
1.73633024e-01 -2.82976657e-01 1.20220944e-01 -6.01067021e-02
1.69923031e+00 -3.39028150e-01 3.62852007e-01 -2.61218399e-01
-3.20376758e-03 2.88200289e-01 9.42106172e-02 3.25112134e-01
-3.47795635e-01 5.08351088e-01 5.53308785e-01 -9.77422774e-01
-1.00497532e+00 -9.44946826e-01 -6.48217857e-01 1.31081879e+00
-5.03008664e-01 -2.02801809e-01 -5.51518500e-01 -4.99070823e-01
2.23266572e-01 1.45727977e-01 -8.65779459e-01 2.70876765e-01
-1.26869082e-01 -1.24682570e+00 7.39882171e-01 4.73197550e-01
-1.52014107e-01 -6.20437562e-01 -8.40532601e-01 3.74361537e-02
3.75172108e-01 -4.78622437e-01 -2.49877542e-01 4.46310163e-01
-1.02264416e+00 -1.20090461e+00 -2.37915367e-01 -6.73152804e-01
9.02875364e-01 2.57390030e-02 7.93402553e-01 3.62240821e-01
-1.30053651e+00 1.11727864e-01 2.85641164e-01 -5.35694659e-01
-3.97514105e-01 8.79579335e-02 -1.87852636e-01 -4.77121949e-01
6.05060518e-01 -3.59871179e-01 -7.51104414e-01 5.72424149e-04
-1.16583860e+00 6.72607720e-02 7.39786029e-01 1.14013696e+00
9.92350340e-01 9.40701086e-03 1.36425152e-01 -7.93914676e-01
4.09838825e-01 -4.35083210e-01 -7.10721970e-01 2.32785642e-01
-4.14766759e-01 -2.21898239e-02 5.69973052e-01 -1.65413231e-01
-6.70822263e-01 1.32671311e-01 -4.16549444e-01 -1.67396069e-01
-2.14053005e-01 6.14534080e-01 1.62689641e-01 -3.50949436e-01
8.89869273e-01 -1.73545107e-01 7.38850474e-01 -2.75958210e-01
-8.36440846e-02 8.21168303e-01 2.44231120e-01 6.28249273e-02
3.63782734e-01 8.29641223e-01 7.81190917e-02 -5.47416568e-01
-3.13696504e-01 -4.76987720e-01 -4.65330422e-01 2.29295380e-02
6.04113936e-01 -6.10257924e-01 -1.23513198e+00 5.21856070e-01
-6.71304345e-01 -5.79416335e-01 3.85776497e-02 4.04386014e-01
-3.41189615e-02 9.19857249e-02 -9.57082748e-01 -9.04230550e-02
-9.68064249e-01 -1.45146382e+00 1.20681047e+00 3.25878292e-01
-6.69688582e-01 -1.39568543e+00 5.62703550e-01 1.20734505e-01
7.42769480e-01 4.70100075e-01 9.62412417e-01 -9.32121933e-01
-2.96109200e-01 -9.47803497e-01 -3.63932490e-01 -2.88795996e-02
1.32854417e-01 7.80670404e-01 -1.19386506e+00 -4.28350389e-01
-4.00170535e-01 -4.52845037e-01 4.71668303e-01 5.99740148e-01
1.66209662e+00 -2.41789743e-01 -9.44592714e-01 1.01071441e+00
1.39220810e+00 -3.51528794e-01 3.66861343e-01 5.15908480e-01
3.57514828e-01 3.97684336e-01 1.77171946e-01 5.42425752e-01
2.75176764e-01 2.29865611e-01 3.76707762e-01 -4.17279512e-01
1.42983720e-01 4.60123956e-01 -2.22021192e-01 2.70822525e-01
5.03594816e-01 1.58091277e-01 -1.32158172e+00 3.87249887e-01
-1.28034282e+00 -5.89319646e-01 -1.21228531e-01 1.86314404e+00
1.01583874e+00 -1.40211239e-01 -1.25903010e-01 -1.47634670e-01
4.60710943e-01 -4.53117967e-01 -6.56152904e-01 -2.21711963e-01
1.54826850e-01 3.33365172e-01 4.34165329e-01 3.02078009e-01
-1.11635566e+00 3.15186262e-01 7.06501389e+00 9.17256415e-01
-1.92050278e+00 -1.33453812e-02 1.20563781e+00 -5.19940555e-01
-9.26285163e-02 -5.15058100e-01 -6.52973413e-01 1.09553926e-01
1.00001132e+00 -2.43883878e-01 -1.38746575e-01 1.07827413e+00
1.78071201e-01 -2.10306346e-01 -1.33689380e+00 9.59025502e-01
-3.51818353e-01 -2.05168104e+00 -2.97528982e-01 2.68495411e-01
1.59380019e-01 4.21074390e-01 2.68953979e-01 4.74459305e-03
2.77855039e-01 -1.07805276e+00 -1.51547477e-01 2.50810981e-01
1.33326137e+00 -5.99567235e-01 1.21539116e+00 -2.96086669e-01
-4.85420138e-01 1.25261739e-01 -2.17039049e-01 1.39406607e-01
-4.73907679e-01 9.21659589e-01 -1.64617360e+00 8.44745114e-02
9.43504095e-01 2.65255868e-01 -9.46446955e-01 1.07652855e+00
6.72591448e-01 4.60627437e-01 -5.49807489e-01 -1.13122612e-01
-2.30150014e-01 5.12259424e-01 1.46175757e-01 1.81912851e+00
4.40619916e-01 -2.40987614e-01 -2.17943415e-01 4.20302749e-01
2.11852595e-01 1.67465076e-01 -3.40043530e-02 -8.94474518e-03
4.42060679e-01 2.08862424e+00 -1.28935587e+00 -1.20196797e-01
-1.47776932e-01 1.85795054e-01 1.48496047e-01 -1.85800418e-01
-6.35450602e-01 -7.62848556e-01 8.17290187e-01 2.93557793e-01
-1.10759743e-01 1.22836083e-01 -7.69995630e-01 -7.98950791e-01
-5.05434155e-01 -8.28534186e-01 6.25422180e-01 -6.40804708e-01
-1.36202168e+00 4.76343155e-01 -5.02253830e-01 -1.04040039e+00
-7.69094974e-02 -1.13455367e+00 -9.14526045e-01 6.40149176e-01
-1.13624227e+00 -9.44910944e-01 -6.45005524e-01 5.24496436e-01
-1.74258932e-01 -1.19220234e-01 1.46296036e+00 1.79402709e-01
-8.61691058e-01 7.35655606e-01 4.96331811e-01 2.78315365e-01
1.02119100e+00 -1.30314064e+00 1.50429964e-01 2.81502157e-01
-4.02082324e-01 5.58514774e-01 5.99747419e-01 -2.15777487e-01
-1.63440502e+00 -1.15550768e+00 4.37256545e-01 -4.73119199e-01
1.00246418e+00 -3.73770624e-01 -9.04344678e-01 6.56486690e-01
6.69904752e-03 6.72909319e-01 1.78401172e+00 1.87016025e-01
-2.55447268e-01 -3.46598953e-01 -1.54924774e+00 5.76671124e-01
6.66223243e-02 -3.45074415e-01 3.34903151e-01 7.36209810e-01
2.17194051e-01 -8.65079880e-01 -1.02847362e+00 4.73437570e-02
7.72403240e-01 -8.26936960e-01 5.57191372e-01 -3.86258602e-01
1.34590507e-01 -3.28682393e-01 3.24895978e-01 -1.10661244e+00
-3.96333128e-01 -4.27996308e-01 2.53384739e-01 7.32494414e-01
3.86277080e-01 -6.13541782e-01 9.95364249e-01 6.82067037e-01
-2.88507938e-01 -1.10897458e+00 -9.63123798e-01 -2.24395424e-01
-5.11223525e-02 -3.50888044e-01 7.65479982e-01 8.40523005e-01
6.29637361e-01 -2.62230694e-01 6.90149009e-01 1.99516356e-01
4.16636586e-01 -3.84758264e-02 8.74447525e-01 -9.57383335e-01
-3.32774907e-01 -7.27024317e-01 -9.39999640e-01 5.66171817e-02
-1.45247325e-01 -9.84807670e-01 -2.63074607e-01 -1.42785013e+00
3.77232313e-01 -5.53848088e-01 -2.94137061e-01 9.22749937e-01
-4.63737287e-02 5.26752770e-01 -3.71678293e-01 2.42686257e-01
-4.80884373e-01 -3.80209565e-01 8.66565108e-01 -2.96889842e-01
-1.37980014e-01 -2.73995370e-01 -6.65746629e-01 7.17528105e-01
7.00948000e-01 -5.89991391e-01 5.83479442e-02 -4.38763201e-01
2.71425903e-01 -2.77548671e-01 5.41700006e-01 -1.31009686e+00
3.40328068e-01 -3.60862985e-02 1.01456809e+00 -5.20328343e-01
-3.01304236e-02 -5.39236188e-01 4.44571853e-01 6.25648856e-01
-3.79972130e-01 -1.94027215e-01 7.57415414e-01 8.45590830e-02
9.52907801e-02 1.78568751e-01 8.91872346e-01 -3.00833806e-02
-3.28110382e-02 3.47349077e-01 -6.87498927e-01 -4.75433499e-01
1.16671479e+00 -2.92858988e-01 -6.90444827e-01 9.76019427e-02
-8.51774812e-01 3.30605745e-01 5.82645833e-01 -1.11142546e-01
3.18439990e-01 -1.06535006e+00 -6.53682709e-01 2.84550965e-01
3.10506195e-01 1.01740249e-01 9.31071877e-01 1.08973897e+00
-1.10722959e+00 3.08281213e-01 -4.55168903e-01 -1.05551970e+00
-1.55374885e+00 3.15546870e-01 6.00129724e-01 -7.31083810e-01
-4.76781815e-01 1.20048046e+00 1.82973325e-01 -4.40141469e-01
3.75445299e-02 -4.98304218e-01 1.16822481e-01 -9.94645283e-02
1.05641639e+00 3.35881144e-01 6.75947487e-01 2.24793509e-01
-6.72887504e-01 1.35551900e-01 -5.05710363e-01 2.30223626e-01
1.42316949e+00 1.88766956e-01 -3.62615615e-01 4.37963367e-01
1.40881884e+00 -1.77060097e-01 -1.10257506e+00 2.51062721e-01
-2.62893289e-01 -2.03020915e-01 2.31328830e-01 -8.47671270e-01
-9.28341866e-01 8.89132321e-01 9.60411310e-01 9.34784953e-03
1.04427314e+00 -1.57992736e-01 5.46277285e-01 3.74008238e-01
1.04034357e-01 -7.85523713e-01 -7.94693083e-02 1.30417839e-01
4.78649408e-01 -1.30450749e+00 3.26762766e-01 1.18172392e-01
-1.98657528e-01 1.59009397e+00 5.14050066e-01 1.48240641e-01
9.34879124e-01 1.08198500e+00 5.99423230e-01 -2.98864067e-01
-1.05852616e+00 4.30404961e-01 -1.25267133e-01 6.38760686e-01
8.91876101e-01 1.40424952e-01 3.30524743e-01 6.50528610e-01
-2.01806590e-01 3.10709119e-01 7.46425569e-01 9.54888821e-01
-2.40754515e-01 -7.52550662e-01 -5.97317576e-01 9.06471491e-01
-7.97285259e-01 3.84772047e-02 1.77554600e-02 6.33036613e-01
-9.11829621e-02 3.98119003e-01 5.97236574e-01 -2.49527067e-01
-3.01784337e-01 -3.32463766e-03 2.56361961e-01 -7.14220107e-01
-7.62032628e-01 1.65003747e-01 -2.33552992e-01 -5.85345984e-01
2.24851653e-01 -4.74238873e-01 -1.37644494e+00 -6.32046640e-01
-3.72430831e-01 -7.77493939e-02 1.00511360e+00 7.21056342e-01
8.76134098e-01 4.18260247e-01 2.62219459e-01 -1.07807791e+00
-2.32528389e-01 -9.72945869e-01 -6.84246182e-01 -2.15950236e-01
5.21983266e-01 -2.82206178e-01 -3.98636281e-01 1.95096835e-01] | [15.096793174743652, -3.0741498470306396] |
3414ac75-3685-4411-ae4c-ee3a523f2555 | fooling-a-real-car-with-adversarial-traffic | 1907.00374 | null | https://arxiv.org/abs/1907.00374v1 | https://arxiv.org/pdf/1907.00374v1.pdf | Fooling a Real Car with Adversarial Traffic Signs | The attacks on the neural-network-based classifiers using adversarial images have gained a lot of attention recently. An adversary can purposely generate an image that is indistinguishable from a innocent image for a human being but is incorrectly classified by the neural networks. The adversarial images do not need to be tuned to a particular architecture of the classifier - an image that fools one network can fool another one with a certain success rate.The published works mostly concentrate on the use of modified image files for attacks against the classifiers trained on the model databases. Although there exists a general understanding that such attacks can be carried in the real world as well, the works considering the real-world attacks are scarce. Moreover, to the best of our knowledge, there have been no reports on the attacks against real production-grade image classification systems.In our work we present a robust pipeline for reproducible production of adversarial traffic signs that can fool a wide range of classifiers, both open-source and production-grade in the real world. The efficiency of the attacks was checked both with the neural-network-based classifiers and legacy computer vision systems. Most of the attacks have been performed in the black-box mode, e.g. the adversarial signs produced for a particular classifier were used to attack a variety of other classifiers. The efficiency was confirmed in drive-by experiments with a production-grade traffic sign recognition systems of a real car. | ['Alexander Kreines', 'Shachar Mendelowitz', 'Yuval Weisglass', 'Nir Morgulis'] | 2019-06-30 | null | null | null | null | ['traffic-sign-recognition'] | ['computer-vision'] | [ 2.80569464e-01 2.81495064e-01 5.56418121e-01 -3.18717271e-01
-4.53497469e-01 -9.99245882e-01 6.16427183e-01 -4.71470803e-01
-4.43188518e-01 4.59247977e-01 -8.31327498e-01 -5.82740545e-01
2.94783384e-01 -8.90132070e-01 -1.27741945e+00 -7.82772303e-01
3.35517675e-02 3.88180315e-01 9.86274898e-01 -4.75661397e-01
1.59044147e-01 9.64757860e-01 -1.98675835e+00 7.08164632e-01
3.54002953e-01 9.56808686e-01 -3.50860536e-01 1.19136274e+00
3.49320233e-01 9.10968423e-01 -1.27211905e+00 -8.94658387e-01
8.58015418e-01 -2.16599032e-01 -4.79134619e-01 -3.44262421e-01
9.01516616e-01 -1.95306957e-01 -4.55027044e-01 1.32484865e+00
2.49715641e-01 -5.42723775e-01 3.53961825e-01 -1.92733812e+00
-3.00868154e-01 3.16167384e-01 3.10427487e-01 -4.87508252e-02
-1.25336528e-01 8.66615653e-01 3.01093578e-01 -2.02740312e-01
6.01214230e-01 1.25177622e+00 3.95819366e-01 6.47311866e-01
-8.57166290e-01 -1.08848560e+00 -3.43723238e-01 3.97764176e-01
-1.02737868e+00 -1.69483051e-01 5.31017661e-01 -5.08016109e-01
6.05489671e-01 5.22597611e-01 3.22690278e-01 1.46603560e+00
4.03002441e-01 1.79749191e-01 1.41240323e+00 -4.36286658e-01
1.39110282e-01 6.10841632e-01 1.42507941e-01 6.65525317e-01
4.66287434e-01 7.55659878e-01 4.94690873e-02 -1.38140634e-01
3.75222296e-01 -4.16744471e-01 -2.31974870e-02 -3.26571256e-01
-9.09260213e-01 8.47945333e-01 5.46687067e-01 2.81607807e-01
-1.13446705e-01 4.37609762e-01 7.65540838e-01 5.91462970e-01
-2.13525608e-01 5.72516084e-01 -4.52869236e-01 4.84135374e-02
-6.75773859e-01 2.39013329e-01 1.02171338e+00 7.33113885e-01
6.08791053e-01 3.77963752e-01 5.07592447e-02 2.06748322e-01
1.62542388e-01 8.86019588e-01 3.10725331e-01 -6.78180397e-01
2.06654653e-01 3.19565624e-01 -5.73719293e-02 -1.08667958e+00
-3.06133330e-02 -8.65618363e-02 -3.11078668e-01 1.34624863e+00
8.66374075e-01 -3.13081950e-01 -1.00589657e+00 1.30867589e+00
2.07287550e-01 3.53821427e-01 3.70578170e-01 6.85905695e-01
7.82921672e-01 4.75700229e-01 -8.40041041e-02 5.31474531e-01
1.27275991e+00 -7.60841548e-01 -3.83504748e-01 -5.47030717e-02
5.12881517e-01 -8.83358538e-01 7.90374994e-01 6.55159593e-01
-6.21879935e-01 -8.78748775e-01 -1.41876936e+00 6.53764904e-01
-1.19480526e+00 -1.85744628e-01 2.55395532e-01 1.55278516e+00
-7.63896048e-01 4.28161114e-01 -5.38212657e-01 -3.04685265e-01
5.08213699e-01 4.36139077e-01 -5.31380415e-01 8.21829140e-02
-1.54406893e+00 1.27448833e+00 3.87493432e-01 7.03826593e-03
-1.32429874e+00 -3.59646767e-01 -4.72731501e-01 -3.34329635e-01
1.42341793e-01 -2.33056530e-01 1.08018398e+00 -1.49460173e+00
-1.35161781e+00 1.15218711e+00 5.45162797e-01 -1.03332043e+00
1.09526706e+00 1.31024331e-01 -9.61793661e-01 1.18055016e-01
-2.68470287e-01 5.56558251e-01 1.17770612e+00 -1.35619402e+00
-7.08442092e-01 7.60584697e-02 4.06193137e-01 -7.45774686e-01
6.15968555e-03 3.50338370e-01 7.62274936e-02 -5.54061592e-01
-7.09482670e-01 -1.23709774e+00 1.03681587e-01 1.69086516e-01
-3.79511416e-01 4.53110695e-01 1.44850838e+00 -2.88696170e-01
5.81921577e-01 -2.08882594e+00 -7.20930040e-01 4.84774530e-01
-2.93024272e-01 1.12090373e+00 -1.92682222e-01 2.12550551e-01
-3.31169546e-01 1.95597544e-01 -4.44358103e-02 4.83641416e-01
-1.20276315e-02 4.17423278e-01 -7.88856924e-01 6.40625119e-01
3.49777013e-01 8.25353920e-01 -5.74413955e-01 -1.79199487e-01
4.40970868e-01 2.97163248e-01 -2.61356533e-01 2.08254933e-01
-7.05634989e-03 1.48514330e-01 -1.98530287e-01 2.85392404e-01
8.47194076e-01 4.75123644e-01 -1.83158934e-01 -2.89390832e-01
1.32033601e-01 -2.79582322e-01 -1.14431107e+00 6.32999897e-01
-4.01313782e-01 9.11002934e-01 7.34774768e-02 -8.04060340e-01
9.38739777e-01 2.95085102e-01 -2.54529536e-01 -3.44970912e-01
3.11478585e-01 4.78509009e-01 6.03464663e-01 -5.84320605e-01
2.18243361e-01 7.56087452e-02 -1.87683076e-01 3.44389051e-01
6.02299385e-02 -4.16028768e-01 2.36630246e-01 -3.35845090e-02
1.38430667e+00 7.12287873e-02 -2.58307397e-01 1.46288559e-01
9.51264143e-01 2.61088818e-01 1.42141178e-01 1.08234107e+00
-3.27107698e-01 5.18434405e-01 4.62499470e-01 -3.90261739e-01
-1.13625216e+00 -1.02256465e+00 -8.89019370e-02 6.95071101e-01
-5.24774753e-02 -2.94667277e-02 -1.06588233e+00 -1.24658287e+00
1.06396802e-01 7.91846037e-01 -5.35993099e-01 -5.56574285e-01
-6.05185628e-01 -2.86250889e-01 1.61097765e+00 2.39570454e-01
8.11160743e-01 -1.32524467e+00 -7.96190202e-01 -9.24386233e-02
4.40783322e-01 -1.42717648e+00 2.00214148e-01 1.84623189e-02
-2.78880984e-01 -1.59833336e+00 -1.11679971e-01 -6.07283413e-01
7.15035021e-01 -1.60471812e-01 9.42090273e-01 3.45054299e-01
-3.83363396e-01 2.03333154e-01 -4.82215613e-01 -9.20538962e-01
-1.60668993e+00 -2.92262882e-01 -1.32587418e-01 3.75206500e-01
3.48576248e-01 -2.03952849e-01 -7.35007152e-02 8.76521647e-01
-1.25914395e+00 -4.64738309e-01 5.35582960e-01 6.47926807e-01
-1.74877346e-01 3.02971274e-01 3.72696787e-01 -1.14727366e+00
3.93449157e-01 -1.07612349e-01 -1.10315812e+00 1.24035329e-01
-2.82563210e-01 -5.44495694e-02 9.52450275e-01 -7.05481529e-01
-6.21495187e-01 2.48339102e-01 -3.18085611e-01 -5.25053501e-01
-7.75442004e-01 -1.29722610e-01 -3.63254219e-01 -6.86893523e-01
1.07875371e+00 3.60728018e-02 8.56726617e-02 3.91027853e-02
3.33325207e-01 7.94734597e-01 7.83760548e-01 -2.46837899e-01
1.64808583e+00 5.45418978e-01 3.15134436e-01 -5.75894535e-01
-3.62065852e-01 1.65129632e-01 -4.06600118e-01 -6.02335036e-01
7.01255202e-01 -4.33431178e-01 -8.37274671e-01 1.12214971e+00
-1.24960268e+00 -3.05522799e-01 -2.17652261e-01 1.18388250e-01
-3.05045992e-01 3.70296359e-01 -3.00652444e-01 -5.57774484e-01
-8.62893928e-03 -1.54010069e+00 4.35792863e-01 1.41412783e-02
1.03198223e-01 -5.76701224e-01 -2.03881264e-01 4.31444138e-01
7.31767297e-01 5.09200811e-01 5.15502214e-01 -1.05285645e+00
-6.44796431e-01 -9.39019799e-01 -2.75644520e-03 1.04764438e+00
-1.82549387e-01 6.47204876e-01 -1.27945316e+00 -7.07530975e-02
-1.26178056e-01 -3.95814478e-01 4.61052835e-01 -2.29431942e-01
8.60550344e-01 -3.48429114e-01 -3.06610577e-02 3.93409818e-01
1.26439714e+00 3.59096706e-01 1.17817199e+00 5.35280824e-01
4.88729596e-01 4.87382889e-01 7.58104920e-01 -3.16786379e-01
-4.45868224e-01 7.35349000e-01 1.00177109e+00 -1.20425940e-01
-5.64617291e-03 1.05744392e-01 7.27373481e-01 -1.87653691e-01
1.35117441e-01 -2.27752373e-01 -8.73267651e-01 1.35073557e-01
-1.45137954e+00 -1.22015941e+00 -7.77508974e-01 2.15096879e+00
4.27881300e-01 7.90796638e-01 4.23060693e-02 3.94886404e-01
9.30004358e-01 -5.21855205e-02 -2.88568944e-01 -9.20280278e-01
-2.98650742e-01 4.76382196e-01 1.02077091e+00 2.88765579e-01
-1.35229397e+00 1.06693375e+00 6.34639502e+00 8.20848703e-01
-1.48938286e+00 1.79628596e-01 3.09928656e-01 9.83995199e-02
4.00735557e-01 2.56153613e-01 -7.78544605e-01 6.66622043e-01
1.29815090e+00 1.93527669e-01 1.52581468e-01 1.18547070e+00
6.89057633e-02 1.01529090e-02 -9.58033085e-01 6.15388632e-01
1.89286143e-01 -1.31447804e+00 -1.04520313e-01 2.27702912e-02
4.50030148e-01 1.90169632e-01 -1.10716168e-02 5.26467621e-01
5.79287589e-01 -1.07963097e+00 9.93920267e-01 4.54457134e-01
5.90219736e-01 -8.75401974e-01 1.04532087e+00 3.26200902e-01
-6.44772112e-01 -8.62785876e-02 -2.36303106e-01 1.17096581e-01
-5.85543923e-02 1.98918164e-01 -8.39901984e-01 2.67278731e-01
6.37655795e-01 -4.72119376e-02 -1.07434940e+00 8.58760834e-01
-5.32421589e-01 8.29737067e-01 -3.71075332e-01 1.36040211e-01
3.37480962e-01 2.09005773e-01 5.98681509e-01 1.34327114e+00
-1.29395872e-01 -6.57158256e-01 -3.09775501e-01 8.24962199e-01
2.03602880e-01 -2.95536548e-01 -1.01424909e+00 5.03607951e-02
1.43497884e-01 1.46919179e+00 -4.71473008e-01 -3.09391081e-01
-2.12363645e-01 8.70767415e-01 -3.12505662e-01 1.74188931e-02
-1.46660161e+00 -6.03765011e-01 6.49136484e-01 1.80211470e-01
4.96712953e-01 7.78934062e-02 2.33407274e-01 -7.65407383e-01
3.66561227e-02 -1.30045712e+00 2.55470634e-01 -9.57679093e-01
-1.23157597e+00 9.23687756e-01 -1.11497985e-02 -1.36705351e+00
-2.02930182e-01 -1.09559953e+00 -8.00268114e-01 7.65158176e-01
-1.09654248e+00 -1.26582325e+00 -3.64214778e-01 6.04403079e-01
3.28617804e-02 -7.77850330e-01 7.88502872e-01 2.50024945e-01
-5.06409824e-01 7.75232852e-01 -1.67785063e-01 5.73122919e-01
8.25612366e-01 -8.55409205e-01 6.38438284e-01 1.24721706e+00
4.40997258e-03 1.47272468e-01 7.92727232e-01 -4.69252527e-01
-1.03716373e+00 -1.23977423e+00 4.33934689e-01 -6.79599404e-01
8.06507707e-01 -4.52705234e-01 -8.66405904e-01 5.57256639e-01
2.97518671e-01 4.65941966e-01 2.24517345e-01 -7.05378592e-01
-7.53610253e-01 -3.88979107e-01 -1.53812230e+00 5.71175218e-01
4.31418478e-01 -5.67000806e-01 -4.77807730e-01 2.96465546e-01
2.24451572e-01 -3.65238994e-01 -6.30382538e-01 2.76233971e-01
6.93677545e-01 -1.21678865e+00 9.80075598e-01 -7.66147316e-01
2.16959268e-01 -8.89754534e-01 9.33128875e-03 -1.08578038e+00
2.35859573e-01 -4.83958274e-01 1.57253504e-01 1.35480618e+00
3.88454258e-01 -1.18313909e+00 5.68119526e-01 6.38187468e-01
3.26951519e-02 -5.81287928e-02 -9.53368068e-01 -1.08324456e+00
1.45848051e-01 -6.87950194e-01 6.78849757e-01 7.90950596e-01
-5.76473713e-01 -1.42006889e-01 -2.63500243e-01 4.86191422e-01
3.21582884e-01 -4.16685402e-01 1.48437285e+00 -9.84052658e-01
-4.08124626e-01 -3.35495800e-01 -1.31359839e+00 -1.01052254e-01
2.24144980e-01 -8.70847702e-01 -8.08165222e-02 -6.31893933e-01
-5.33553183e-01 -3.89099032e-01 -8.13534856e-02 5.25924981e-01
1.83110535e-01 5.92341423e-01 6.15675032e-01 -3.24189365e-02
-1.15276627e-01 -3.49524409e-01 7.30881631e-01 -3.37795377e-01
5.10297775e-01 7.64345527e-02 -2.19123840e-01 8.68559837e-01
9.02443409e-01 -1.09239769e+00 -5.03803231e-02 1.92605808e-01
1.94538593e-01 -6.40493274e-01 9.74291563e-01 -1.61805081e+00
1.13174520e-01 6.89255744e-02 1.64449111e-01 -3.94207507e-01
1.09295025e-01 -1.42218804e+00 4.83714819e-01 7.18590021e-01
2.28978246e-02 -7.95903206e-02 3.67622018e-01 3.28647137e-01
-2.26960853e-01 -5.18771052e-01 1.22329450e+00 -1.14809550e-01
-7.65616238e-01 -1.71113342e-01 -6.67332053e-01 -3.00884068e-01
1.48234630e+00 -4.90948588e-01 -6.52092338e-01 -3.07619542e-01
-4.69785184e-01 -2.31953278e-01 6.70405447e-01 5.96849978e-01
3.43224764e-01 -1.07145131e+00 -7.48255193e-01 4.70446348e-01
1.97662979e-01 -5.97218037e-01 -1.26040250e-01 3.84940028e-01
-1.07184005e+00 1.93859756e-01 -6.97380245e-01 -5.90314507e-01
-1.66017413e+00 9.38367963e-01 9.20929670e-01 -1.91629529e-01
-2.13345796e-01 3.40714782e-01 -2.88532879e-02 -5.17163217e-01
9.74085927e-03 -1.41229510e-01 4.20351438e-02 -1.91197604e-01
4.67732817e-01 2.84484446e-01 4.12821501e-01 -9.59699512e-01
-4.08305347e-01 3.21327180e-01 2.18580857e-01 -3.67281996e-02
9.41385627e-01 6.56796455e-01 5.89758717e-02 1.10831738e-01
1.20322406e+00 6.28600121e-02 -1.03151941e+00 3.22456092e-01
-2.77607828e-01 -6.05867505e-01 -2.24874794e-01 -9.65394855e-01
-1.49624932e+00 7.68727362e-01 1.07525015e+00 5.54633141e-01
8.47361147e-01 -4.17004675e-01 4.56969261e-01 5.90662718e-01
5.73358953e-01 -8.55901361e-01 -1.65208906e-01 5.47796547e-01
9.90273118e-01 -1.07138717e+00 -3.95591915e-01 -3.68829817e-01
-6.82240903e-01 1.43255115e+00 5.58451593e-01 -5.83078682e-01
6.71743274e-01 5.79439342e-01 7.09357679e-01 -6.10785708e-02
-4.47262883e-01 -5.43696135e-02 5.71929552e-02 9.81544316e-01
-1.98868483e-01 -1.00145839e-01 1.58801302e-02 9.25586820e-02
-2.36373037e-01 1.09797671e-01 8.88012052e-01 1.01092052e+00
-1.38398811e-01 -1.35340238e+00 -1.08045733e+00 1.90334365e-01
-5.52547693e-01 1.48712412e-01 -7.65500903e-01 1.00836456e+00
6.28770292e-01 1.19560444e+00 -2.20355824e-01 -5.32600045e-01
6.51810050e-01 2.67787308e-01 2.25660235e-01 -2.44773552e-01
-1.38992774e+00 -7.21195877e-01 4.89387035e-01 -6.69550180e-01
-3.35938752e-01 -5.13840914e-01 -9.16448116e-01 -4.16420102e-01
-3.27875137e-01 -2.37654373e-01 8.32000434e-01 6.84499264e-01
1.79693162e-01 5.01795709e-01 6.90854907e-01 -7.96550155e-01
-5.59207380e-01 -8.19305599e-01 -3.51556897e-01 7.98366368e-01
5.52240387e-02 -4.72512126e-01 -7.61363924e-01 2.55635679e-01] | [5.480008602142334, 7.816496849060059] |
a161baed-6ccd-44b9-95da-380edf05b31d | comparison-between-voting-classifier-and-deep | null | null | https://aclanthology.org/2020.wanlp-1.23 | https://aclanthology.org/2020.wanlp-1.23.pdf | Comparison between Voting Classifier and Deep Learning methods for Arabic Dialect Identification | In this paper, we present three methods developed for the NADI shared task on Arabic Dialect Identification for tweets. The first and the second method use respectively a machine learning model based on a Voting Classifier with words and character level features and a deep learning model at word level. The third method uses only character-level features. We explored different text representation such as Tf-idf (first model) and word embeddings (second model). The Voting Classifier was the most powerful prediction model, achieving the best macro-average F1 score of 18.8% and an accuracy of 36.54% on the official test. Our model ranked 9 on the challenge and in conclusion we propose some ideas to improve its results. | ['Gaël Lejeune', 'Ghoul Dhaou'] | null | null | null | null | coling-wanlp-2020-12 | ['dialect-identification'] | ['natural-language-processing'] | [-3.74185234e-01 1.90846361e-02 -1.23757191e-01 -4.60947692e-01
-5.75007081e-01 -4.66659218e-01 1.09417284e+00 4.38228279e-01
-8.72587323e-01 5.65743744e-01 3.97880971e-01 -4.78947461e-01
-6.77170455e-02 -1.10886347e+00 3.51741686e-02 -6.16279185e-01
-1.32746369e-01 8.01936030e-01 1.70811027e-01 -8.34307134e-01
5.67283988e-01 4.60442066e-01 -1.48271739e+00 4.85048056e-01
9.00813878e-01 9.08323109e-01 -2.88651884e-01 6.80332422e-01
-3.87122542e-01 7.01167405e-01 -5.52938879e-01 -6.47940516e-01
1.25272170e-01 -2.24556960e-02 -1.15226865e+00 -6.57434881e-01
3.56527001e-01 -3.15094739e-01 -3.08278561e-01 7.12490022e-01
6.40630662e-01 6.23343736e-02 1.00172472e+00 -9.32479024e-01
-6.18390560e-01 8.51917744e-01 -4.48629200e-01 3.21397111e-02
5.34488499e-01 -3.43185782e-01 1.04279435e+00 -1.20700622e+00
4.50408041e-01 1.10894644e+00 9.23682213e-01 2.70182520e-01
-9.06566262e-01 -4.96166795e-01 -8.66378844e-02 4.77209955e-01
-1.57842207e+00 -1.63053930e-01 1.97910473e-01 -5.22033811e-01
1.19190013e+00 2.82689273e-01 3.66588563e-01 7.37103701e-01
3.39524932e-02 6.43490791e-01 1.44091189e+00 -8.64072382e-01
-1.65211409e-01 4.95997757e-01 7.99503803e-01 7.66560137e-01
4.11554761e-02 -1.93097040e-01 -2.88204342e-01 -4.82405245e-01
7.80448988e-02 -3.48834038e-01 1.65150046e-01 4.02908474e-01
-9.44007993e-01 1.40497541e+00 8.77115652e-02 8.86417210e-01
-2.06904992e-01 -3.26493353e-01 4.95306969e-01 4.28702325e-01
7.67132103e-01 3.01825583e-01 -7.22189367e-01 -1.14769064e-01
-9.74432945e-01 7.12445617e-01 1.03728771e+00 1.90497652e-01
8.07071626e-01 7.68450694e-03 -3.58204663e-01 1.21655154e+00
4.33518678e-01 6.05848134e-01 9.69014823e-01 -5.62704764e-02
1.99359655e-01 4.91992265e-01 4.25288044e-02 -9.76475120e-01
-7.67159641e-01 -1.72456264e-01 -4.64448392e-01 1.13251649e-01
8.92590880e-01 -6.03511691e-01 -9.11142111e-01 1.44046879e+00
-8.58280510e-02 -1.67097837e-01 -2.90487051e-01 6.39109612e-01
1.00992119e+00 8.58709872e-01 -1.04813389e-01 1.38600856e-01
1.49789917e+00 -9.68961954e-01 -5.86977005e-01 3.53658229e-01
9.74041879e-01 -1.05786455e+00 7.36068606e-01 5.59609294e-01
-8.47211897e-01 -4.96601522e-01 -1.00455773e+00 2.21335784e-01
-1.13120162e+00 1.09356679e-01 4.31835920e-01 1.34796250e+00
-1.35874629e+00 4.71638203e-01 -3.27782780e-02 -7.85379410e-01
-1.21312939e-01 6.06695354e-01 -4.28521007e-01 1.09173819e-01
-1.50275052e+00 1.21803594e+00 1.72301263e-01 -1.46659851e-01
-3.57666463e-01 -3.42490792e-01 -7.55910277e-01 -1.68514624e-01
-4.26385373e-01 6.86301813e-02 8.23582649e-01 -9.40382838e-01
-1.75452292e+00 1.25139952e+00 -1.03757508e-01 -6.05645478e-01
3.88457239e-01 -2.16913968e-01 -6.97682917e-01 -2.61544049e-01
-1.65481970e-01 4.30144280e-01 6.24949336e-01 -9.32768703e-01
-8.21051955e-01 -3.95368248e-01 -1.39562890e-01 -8.26156363e-02
-5.43308914e-01 2.81983763e-01 1.83761567e-01 -6.47665799e-01
-5.91188446e-02 -9.52539742e-01 2.32415926e-02 -8.71546268e-01
7.87837803e-02 -7.26622462e-01 6.87143207e-01 -1.14179778e+00
1.49703276e+00 -1.59631813e+00 -1.37997791e-01 5.28840899e-01
-4.63001579e-02 7.51117468e-01 -2.56010294e-01 8.22353244e-01
-1.30230337e-01 1.16413742e-01 1.02626175e-01 -1.43393055e-01
1.20824136e-01 -6.99344501e-02 -3.05472165e-01 4.75671649e-01
3.03433686e-02 5.93770504e-01 -6.44410014e-01 -1.68751851e-01
1.04334742e-01 5.27258813e-01 -3.77487272e-01 4.54848371e-02
1.55330911e-01 -7.98874274e-02 1.26205921e-01 4.66442883e-01
1.12980187e+00 5.40663540e-01 2.87593722e-01 -1.63394719e-01
-5.26506960e-01 4.43176746e-01 -1.26229358e+00 1.13234508e+00
-5.27121127e-01 7.37291157e-01 -1.09712839e-01 -8.70486438e-01
1.29632127e+00 2.27038786e-01 3.13032985e-01 -6.62390769e-01
2.67963082e-01 4.20043826e-01 -7.46142417e-02 -4.36976701e-01
8.60420763e-01 -3.41561139e-02 -1.34819835e-01 7.14294076e-01
2.53135324e-01 2.99065799e-01 9.18874666e-02 9.02211145e-02
5.30442655e-01 -5.85439354e-02 1.87096044e-01 -7.45207727e-01
1.04131103e+00 -1.73626229e-01 8.24365839e-02 8.44488680e-01
-3.51318568e-01 6.53077543e-01 4.94162261e-01 -8.56033385e-01
-5.94378769e-01 -6.70936227e-01 -4.29667473e-01 1.59690416e+00
-4.19980496e-01 -5.90078592e-01 -8.82813632e-01 -9.55287158e-01
3.09985280e-02 6.00621819e-01 -8.48804116e-01 2.65846401e-01
-7.48099208e-01 -1.10594440e+00 1.03146100e+00 1.51438817e-01
3.15271825e-01 -8.15912545e-01 5.96826822e-02 4.66035083e-02
-4.07543391e-01 -5.14153779e-01 -1.76035091e-01 7.79443979e-02
-3.22725534e-01 -7.72076011e-01 -9.94904637e-01 -1.24556863e+00
1.19700782e-01 -1.11938074e-01 9.41103697e-01 3.11537802e-01
-1.99347697e-02 2.51273252e-02 -8.48328471e-01 -4.64917541e-01
-2.76698917e-01 4.86408591e-01 9.81101170e-02 4.15305436e-01
7.58198023e-01 -5.87498248e-02 -3.21094304e-01 1.07461058e-01
-5.44453323e-01 -5.36698878e-01 1.93282694e-01 9.35599744e-01
-2.39705443e-01 -4.72041100e-01 7.01811492e-01 -9.29725349e-01
7.42805719e-01 -4.55005854e-01 -2.62749702e-01 3.30679715e-02
-7.24853218e-01 2.30009947e-02 3.10715705e-01 -1.96317568e-01
-8.60235870e-01 -3.64584220e-03 -9.43294764e-01 7.84664989e-01
-1.87794179e-01 5.93641758e-01 2.58111298e-01 -1.49711356e-01
6.39705002e-01 1.85265526e-01 2.16099378e-02 -6.64724469e-01
2.50815809e-01 1.18854845e+00 -1.47630006e-01 -3.30994248e-01
3.66452992e-01 1.40560508e-01 -3.89602780e-01 -1.20126903e+00
-4.30957556e-01 -5.90112865e-01 -9.36803758e-01 -1.03459604e-01
9.40523326e-01 -7.57630646e-01 -6.47952676e-01 1.12377012e+00
-1.08121312e+00 -1.85087144e-01 2.88976341e-01 4.95739877e-01
-1.58974528e-01 4.45797026e-01 -7.56791711e-01 -8.37292433e-01
-4.49899584e-01 -8.29137921e-01 6.69602931e-01 -3.98727171e-02
-4.57616627e-01 -1.21732175e+00 5.84843934e-01 1.26586556e-01
7.74990916e-01 8.01524296e-02 1.07443893e+00 -1.28512263e+00
5.02706945e-01 -1.92109376e-01 -3.29586655e-01 1.74915344e-01
-3.07037402e-02 1.82506531e-01 -1.32368493e+00 -2.49632671e-01
-3.91925752e-01 -3.02065611e-01 1.07640243e+00 3.84117633e-01
7.41685092e-01 -9.27127302e-02 -4.61848117e-02 1.83293328e-01
1.54056895e+00 3.49146836e-02 8.11201990e-01 5.25862992e-01
4.41285849e-01 8.13399136e-01 4.03107256e-01 5.68022847e-01
7.19016373e-01 9.08603609e-01 2.05597788e-01 1.12668544e-01
-6.38836436e-03 -5.29473461e-03 4.08122599e-01 8.74182284e-01
-1.94862649e-01 -1.05156429e-01 -1.31954658e+00 6.73750997e-01
-1.80209899e+00 -9.54249263e-01 -6.79666936e-01 2.27989912e+00
4.97311950e-01 2.67618820e-02 7.89312899e-01 5.30717731e-01
6.35804057e-01 2.35038131e-01 5.56168556e-01 -1.29322207e+00
-3.03941280e-01 6.67795956e-01 5.42094290e-01 1.09517658e+00
-1.30673850e+00 1.13643873e+00 6.87586689e+00 8.88018072e-01
-1.07905722e+00 2.75116235e-01 4.19579804e-01 4.26934302e-01
-1.28452331e-01 -2.59989172e-01 -1.06501389e+00 4.53874260e-01
1.37123716e+00 1.50903026e-02 1.60083964e-01 3.96363258e-01
-1.40839696e-01 -1.16899401e-01 -4.60069776e-01 8.31983685e-01
4.30006266e-01 -1.16392219e+00 3.30561221e-01 3.06213111e-01
5.68587422e-01 3.37666869e-01 1.03355289e-01 6.48100376e-01
9.39538851e-02 -1.25436234e+00 6.74726903e-01 4.35384572e-01
6.08478069e-01 -1.14060366e+00 1.25288630e+00 1.01343110e-01
-1.08218682e+00 -7.21368417e-02 -3.21261704e-01 -4.54740196e-01
-2.73694098e-01 3.42918277e-01 -7.32595861e-01 5.55379152e-01
4.72572893e-01 5.55841029e-01 -7.50673532e-01 7.75297046e-01
2.07261190e-01 9.11442995e-01 -2.66519219e-01 -5.28825760e-01
7.42255747e-01 -2.76493698e-01 2.18890354e-01 1.98805594e+00
2.63460398e-01 -2.89021492e-01 5.01620620e-02 -7.71962404e-02
2.92100787e-01 7.18798280e-01 -5.89653909e-01 4.65182632e-01
1.66010380e-01 1.43359435e+00 -5.56851864e-01 -3.11843723e-01
-4.75885004e-01 9.47971523e-01 3.37796211e-01 -6.21745884e-02
-7.50936925e-01 -8.10649633e-01 7.09794104e-01 -6.08128197e-02
3.59562248e-01 -2.36556038e-01 -3.69154423e-01 -9.77890253e-01
-5.15003622e-01 -8.54918599e-01 6.38116598e-01 -3.86142023e-02
-1.20386720e+00 9.86868322e-01 -2.68492341e-01 -9.18585181e-01
-4.05218661e-01 -1.30186760e+00 -5.73324621e-01 1.11269093e+00
-1.29200745e+00 -1.36014521e+00 4.92217503e-02 6.11241698e-01
2.90351242e-01 -7.46508896e-01 1.36622572e+00 5.19346356e-01
-2.96793938e-01 1.06697369e+00 4.62176174e-01 3.36977571e-01
9.28143620e-01 -1.38084400e+00 1.67504996e-01 3.38464290e-01
2.11082876e-01 3.67588013e-01 5.06903529e-01 -1.71061024e-01
-1.04598582e+00 -7.07189918e-01 1.92825019e+00 -5.34037530e-01
6.92446589e-01 -4.06519622e-01 -6.65538847e-01 4.03616935e-01
5.89711368e-01 -5.56463540e-01 8.99005413e-01 4.64788258e-01
-4.47285175e-01 -4.99192514e-02 -1.45324123e+00 2.98909694e-01
2.59226024e-01 -4.37798560e-01 -5.75965047e-01 3.53499889e-01
2.24632710e-01 6.38080463e-02 -9.14683402e-01 1.61844026e-02
9.89235878e-01 -1.01180148e+00 6.66090190e-01 -5.38778305e-01
-4.48718145e-02 1.02685161e-01 -4.24840152e-01 -1.41111803e+00
-4.39989895e-01 -2.97121942e-01 3.29413623e-01 1.34103346e+00
7.60445535e-01 -9.87519801e-01 5.33749700e-01 1.93707928e-01
3.33730757e-01 -6.66871846e-01 -7.28544474e-01 -5.78516841e-01
8.09611976e-01 -5.15882075e-01 5.78081131e-01 1.17154491e+00
2.87214249e-01 3.39099258e-01 -4.35242206e-01 -1.98494166e-01
5.44713289e-02 -2.02944249e-01 4.81504530e-01 -1.41953242e+00
1.53802961e-01 -7.12479591e-01 -5.81183970e-01 -5.71504772e-01
3.44846964e-01 -1.03407919e+00 -3.42170924e-01 -1.28633010e+00
1.02385702e-02 -3.23067605e-01 -3.15254301e-01 2.44495779e-01
9.58261192e-02 6.20455682e-01 2.46018037e-01 -2.63813645e-01
-1.46389529e-01 6.26084581e-03 3.46109778e-01 -2.13590309e-01
-1.23849712e-01 6.31220192e-02 -5.74466348e-01 5.99981129e-01
1.11939251e+00 -2.92259067e-01 1.82983652e-01 -4.31614637e-01
3.18600267e-01 -5.68861187e-01 -1.76993623e-01 -7.83262610e-01
-8.26316029e-02 1.39075667e-01 2.63193995e-01 -7.32766688e-01
1.79659069e-01 -2.82610834e-01 -6.77322507e-01 8.19768667e-01
-1.47246987e-01 3.16952258e-01 1.33986443e-01 -6.29223064e-02
-2.31861487e-01 -5.40333986e-01 8.67164493e-01 2.02431008e-01
-7.76053309e-01 8.83859172e-02 -1.09614980e+00 -2.26951987e-01
5.71199715e-01 -1.63651034e-02 -1.72097653e-01 -4.60231096e-01
-8.04250240e-01 -5.98002523e-02 8.41642451e-03 6.44147635e-01
4.62858707e-01 -1.31912673e+00 -1.35623276e+00 5.16027570e-01
1.51438519e-01 -1.26845157e+00 -1.97564319e-01 6.59828782e-01
-8.45167518e-01 6.83627188e-01 -4.12774473e-01 -2.34801874e-01
-1.45519495e+00 6.04122095e-02 3.34337533e-01 -6.64915502e-01
1.00185357e-01 8.19923639e-01 -4.47464973e-01 -1.14897680e+00
1.56101137e-01 1.04607277e-01 -9.74335432e-01 6.70664907e-01
9.23036873e-01 8.69399905e-01 4.61094677e-01 -1.34576881e+00
-7.16244400e-01 7.92906761e-01 -2.81154782e-01 -5.72614253e-01
1.20571876e+00 8.53254348e-02 -5.25512695e-01 4.49816972e-01
1.46819127e+00 3.35672528e-01 -1.73399240e-01 -1.86652422e-01
3.03552032e-01 -2.94291735e-01 1.75141901e-01 -1.03282809e+00
-6.63133502e-01 8.63718450e-01 1.01932490e+00 6.54256821e-01
6.27102554e-01 -6.14005923e-01 7.26849258e-01 2.72855520e-01
2.64258713e-01 -1.27726352e+00 -4.28971112e-01 1.39294016e+00
6.56242073e-01 -1.21479154e+00 -2.29204044e-01 -1.60052609e-02
-5.86673677e-01 1.50215757e+00 2.44324490e-01 -2.47655839e-01
1.30352259e+00 5.43753896e-03 3.52143168e-01 5.85609078e-02
-4.29860890e-01 -5.30323803e-01 3.93620491e-01 7.80303299e-01
1.02611673e+00 3.25118184e-01 -1.06178892e+00 7.18025625e-01
-1.39706939e-01 -2.85169333e-01 3.16794395e-01 5.63626230e-01
-6.25086010e-01 -1.39970779e+00 -3.95658821e-01 5.53490639e-01
-7.04005837e-01 -1.46679312e-01 -6.45733893e-01 6.89588785e-01
3.33451301e-01 1.47021520e+00 1.71052620e-01 -9.79428649e-01
2.00541511e-01 4.78626668e-01 5.03849506e-01 -3.65978867e-01
-1.19558573e+00 -3.42647731e-01 4.17217255e-01 -1.17355347e-01
-2.08141223e-01 -6.48000240e-01 -9.44381833e-01 -8.70208740e-01
-3.22644740e-01 2.79810131e-01 9.90558803e-01 9.85022783e-01
-8.43725167e-03 -3.43343139e-01 9.82532263e-01 -7.99462378e-01
-5.02890229e-01 -1.42477667e+00 -8.09899509e-01 3.16585839e-01
1.70988649e-01 -1.89125136e-01 -2.17089832e-01 -2.89509356e-01] | [10.1629638671875, 10.670921325683594] |
86f946b8-5a37-4738-8c8f-770198e6de9e | the-voice-and-eye-gaze-behavior-of-an | null | null | https://aclanthology.org/W12-0408 | https://aclanthology.org/W12-0408.pdf | The Voice and Eye Gaze Behavior of an Imposter: Automated Interviewing and Detection for Rapid Screening at the Border | null | ['Monica Gariup', 'Aaron Elkins', 'Douglas Derrick'] | 2012-04-01 | null | null | null | ws-2012-4 | ['deception-detection'] | ['miscellaneous'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.402584552764893, 3.623408555984497] |
6faec35b-e5b4-4d01-8ad4-55a0d3fa32ac | self-supervised-learning-by-cross-modal-audio | 1911.12667 | null | https://arxiv.org/abs/1911.12667v3 | https://arxiv.org/pdf/1911.12667v3.pdf | Self-Supervised Learning by Cross-Modal Audio-Video Clustering | Visual and audio modalities are highly correlated, yet they contain different information. Their strong correlation makes it possible to predict the semantics of one from the other with good accuracy. Their intrinsic differences make cross-modal prediction a potentially more rewarding pretext task for self-supervised learning of video and audio representations compared to within-modality learning. Based on this intuition, we propose Cross-Modal Deep Clustering (XDC), a novel self-supervised method that leverages unsupervised clustering in one modality (e.g., audio) as a supervisory signal for the other modality (e.g., video). This cross-modal supervision helps XDC utilize the semantic correlation and the differences between the two modalities. Our experiments show that XDC outperforms single-modality clustering and other multi-modal variants. XDC achieves state-of-the-art accuracy among self-supervised methods on multiple video and audio benchmarks. Most importantly, our video model pretrained on large-scale unlabeled data significantly outperforms the same model pretrained with full-supervision on ImageNet and Kinetics for action recognition on HMDB51 and UCF101. To the best of our knowledge, XDC is the first self-supervised learning method that outperforms large-scale fully-supervised pretraining for action recognition on the same architecture. | ['Bruno Korbar', 'Humam Alwassel', 'Du Tran', 'Dhruv Mahajan', 'Lorenzo Torresani', 'Bernard Ghanem'] | 2019-11-28 | null | http://proceedings.neurips.cc/paper/2020/hash/6f2268bd1d3d3ebaabb04d6b5d099425-Abstract.html | http://proceedings.neurips.cc/paper/2020/file/6f2268bd1d3d3ebaabb04d6b5d099425-Paper.pdf | neurips-2020-12 | ['self-supervised-action-recognition'] | ['computer-vision'] | [ 2.41547331e-01 6.64162543e-03 -7.57528841e-01 -3.95852268e-01
-1.04215002e+00 -4.70034808e-01 7.94270456e-01 -9.32534412e-02
-3.53465229e-01 3.35722536e-01 6.92610681e-01 2.68266678e-01
5.85197918e-02 -2.17218444e-01 -1.05550826e+00 -5.97774386e-01
-1.23517096e-01 3.87326568e-01 1.67756081e-01 2.29787663e-01
-1.46318614e-01 -1.59196600e-01 -1.87545669e+00 1.06519580e+00
3.05800259e-01 1.21595693e+00 4.36967053e-02 4.66560751e-01
1.06362119e-01 1.36773932e+00 -4.93225940e-02 -2.01197520e-01
4.43870835e-02 -4.74770546e-01 -1.07380033e+00 3.88318866e-01
6.77878261e-01 -2.39859641e-01 -5.47507882e-01 7.52833366e-01
2.73358345e-01 1.48446783e-01 8.05240273e-01 -1.52526534e+00
-6.53762102e-01 8.89754593e-01 -5.47924638e-01 1.84526443e-01
3.87634963e-01 3.24310094e-01 1.27201796e+00 -7.43769407e-01
7.14782178e-01 1.29150093e+00 5.60309231e-01 7.42556989e-01
-1.40090132e+00 -5.23725390e-01 2.63641775e-01 5.07561266e-01
-1.24385154e+00 -6.58535004e-01 7.54602134e-01 -5.71004808e-01
9.99493182e-01 -1.00201741e-01 5.67832172e-01 1.71273208e+00
-2.37449825e-01 1.42863882e+00 1.09139884e+00 -3.01223606e-01
1.93151295e-01 8.21173936e-02 -9.76757109e-02 5.57634652e-01
-4.63885009e-01 8.71142372e-02 -1.19145787e+00 7.22436532e-02
3.62370312e-01 2.30520010e-01 -2.11909503e-01 -6.18660331e-01
-1.53564417e+00 6.41907096e-01 2.24339753e-01 4.74710554e-01
-2.18832359e-01 4.45543259e-01 7.60092556e-01 3.31717581e-01
2.69942164e-01 2.18970701e-01 -5.68926334e-01 -6.41891420e-01
-8.34184170e-01 -3.63914669e-01 3.69840056e-01 6.97232962e-01
6.90150559e-01 2.24852428e-01 -7.21537098e-02 7.75572717e-01
4.36077833e-01 5.11858463e-01 9.59193826e-01 -1.23812926e+00
4.15967792e-01 4.63595897e-01 -3.57983410e-01 -6.31106377e-01
-2.73990631e-01 -3.02426577e-01 -7.40737736e-01 2.28144713e-02
3.91912818e-01 5.11286594e-02 -8.02737534e-01 1.94147861e+00
3.75191160e-02 6.22516930e-01 3.91644835e-01 8.00900102e-01
9.86492217e-01 5.33559501e-01 2.93913007e-01 -6.77362010e-02
9.56771970e-01 -1.23219693e+00 -4.24752980e-01 -1.24863863e-01
6.72393203e-01 -4.53388691e-01 9.60144520e-01 4.13786262e-01
-7.30388582e-01 -7.68230915e-01 -8.13197136e-01 2.15120822e-01
-1.13693267e-01 2.85371929e-01 6.52487576e-01 2.72043556e-01
-9.58080828e-01 6.27293766e-01 -1.09504437e+00 -5.84618032e-01
5.64808786e-01 8.04573018e-03 -7.36972988e-01 -5.41827567e-02
-1.04813206e+00 5.58077991e-01 5.44321060e-01 -4.78111744e-01
-1.34318841e+00 -7.72688866e-01 -8.39369357e-01 -1.07290827e-01
4.51778352e-01 -3.06608737e-01 1.22319996e+00 -1.68614352e+00
-1.27293849e+00 1.06165087e+00 -1.19361736e-01 -8.77261877e-01
2.11103261e-01 -3.11274707e-01 -6.58172965e-01 6.98181808e-01
3.42463672e-01 1.30034101e+00 1.38839221e+00 -1.39000630e+00
-8.21404755e-01 -3.96628439e-01 -1.54362515e-01 3.03041816e-01
-5.17889798e-01 -2.98717499e-01 -5.68403542e-01 -5.59106171e-01
-9.25764367e-02 -9.94564414e-01 4.02007490e-01 -1.65234074e-01
-4.81875926e-01 -3.03540021e-01 1.06688988e+00 -1.31551594e-01
9.76066351e-01 -2.55845499e+00 2.78618455e-01 -4.99272421e-02
9.19804797e-02 4.68878858e-02 -3.26537877e-01 4.29243684e-01
-3.97098303e-01 -2.22055942e-01 1.50700435e-01 -6.16793692e-01
5.82774691e-02 3.59314024e-01 -3.46455634e-01 4.99626815e-01
2.00195774e-01 8.85662794e-01 -9.80717659e-01 -6.60385907e-01
1.79765686e-01 4.74881262e-01 -4.40000117e-01 1.53024226e-01
-2.31827512e-01 6.39599383e-01 -3.02561056e-02 7.58157194e-01
-2.22223252e-02 -6.67741776e-01 1.91489622e-01 -4.85831708e-01
3.31638813e-01 2.22409114e-01 -8.95283043e-01 2.08682346e+00
-2.54568130e-01 7.71975100e-01 -2.15427250e-01 -1.38326228e+00
4.78989005e-01 5.94140768e-01 9.89188790e-01 -5.00642657e-01
-7.32677728e-02 1.02116875e-01 -1.73514739e-01 -4.86093253e-01
1.69616967e-01 -2.20618680e-01 4.87171710e-02 6.63604259e-01
8.31735969e-01 3.52273285e-01 8.83902460e-02 6.70744419e-01
1.15915561e+00 1.97573677e-01 1.18568219e-01 1.91977963e-01
3.25511128e-01 1.85940228e-02 4.97361809e-01 6.52442932e-01
-4.24494922e-01 5.53671598e-01 2.62562215e-01 -8.43279511e-02
-5.68833292e-01 -1.04684043e+00 2.13295165e-02 1.42558908e+00
-4.46613692e-02 -7.81422317e-01 -4.54739869e-01 -9.23973739e-01
2.59594530e-01 3.29687953e-01 -8.66664171e-01 -3.67586136e-01
-1.50855118e-02 -4.24434468e-02 6.98656440e-01 9.28784728e-01
4.94679034e-01 -9.74167943e-01 -2.81768292e-01 -8.90443381e-03
-3.51571023e-01 -1.41917717e+00 -3.47384721e-01 4.45081115e-01
-8.75756860e-01 -1.17161572e+00 -4.27266955e-01 -8.01068604e-01
2.49550268e-01 4.96929914e-01 1.20215380e+00 -1.75609112e-01
3.35155837e-02 1.38003552e+00 -6.61294639e-01 2.81301122e-02
-3.83384317e-01 -8.78117159e-02 3.10793400e-01 5.00953794e-01
5.70362389e-01 -6.94640040e-01 -3.35557312e-01 4.05808389e-01
-8.22736263e-01 -1.62336940e-03 4.50899988e-01 9.26296949e-01
8.19498718e-01 2.13120207e-01 5.27592063e-01 -6.56498194e-01
-9.60734487e-02 -6.71746969e-01 1.60521399e-02 1.99058741e-01
-2.88486332e-01 1.37620280e-03 6.16171598e-01 -5.75618625e-01
-9.50783014e-01 5.56193590e-01 3.27451468e-01 -1.03984833e+00
-6.87561631e-01 5.70889950e-01 -1.39852822e-01 2.86397308e-01
6.38604641e-01 3.88924688e-01 2.32026130e-01 -4.10631180e-01
5.13297915e-01 5.69407701e-01 8.41647625e-01 -4.85798597e-01
5.92026889e-01 8.12186897e-01 -2.44774967e-01 -7.27812052e-01
-1.20337164e+00 -7.88250566e-01 -8.84550095e-01 -5.20342231e-01
1.27413738e+00 -1.44969773e+00 -6.69772744e-01 5.97218454e-01
-7.04916716e-01 -5.99074244e-01 -4.18564081e-01 7.50061274e-01
-8.39484453e-01 2.81941086e-01 -4.58763093e-01 -5.89882851e-01
2.33192950e-01 -9.27184999e-01 1.29011214e+00 -2.46952936e-01
-3.44739735e-01 -1.08078241e+00 1.05951363e-02 9.08975899e-01
-8.80778804e-02 1.74456108e-02 3.73249412e-01 -9.15044427e-01
-4.87615675e-01 8.08066726e-02 6.52117878e-02 6.10310912e-01
1.96558878e-01 -1.60922818e-02 -1.18847489e+00 -2.88434893e-01
-2.16584727e-01 -1.15321374e+00 1.34510767e+00 2.27488413e-01
1.08474422e+00 -6.32043704e-02 -1.94250405e-01 4.42037940e-01
1.08820057e+00 -9.17206109e-02 4.19891417e-01 4.01958972e-01
9.18313503e-01 4.10441786e-01 7.33187258e-01 5.33452868e-01
5.08739412e-01 7.23870754e-01 4.61805850e-01 1.48200303e-01
-1.88258275e-01 -4.65609699e-01 8.39840651e-01 6.99371517e-01
9.08320174e-02 7.88906217e-02 -8.62573326e-01 5.01870155e-01
-2.07979870e+00 -1.49968541e+00 1.16987534e-01 1.91654193e+00
9.94332552e-01 8.41260701e-02 5.45834303e-01 2.17914313e-01
5.89872122e-01 3.25707525e-01 -6.35692894e-01 3.37791651e-01
-3.60837311e-01 -1.69946879e-01 2.05693871e-01 1.21547237e-01
-1.71650040e+00 7.87397265e-01 5.50814676e+00 1.04152012e+00
-1.12888145e+00 3.06876600e-01 5.71548760e-01 -3.71178538e-01
-5.44848815e-02 -1.24902323e-01 -6.88867986e-01 5.42857111e-01
1.14064825e+00 2.62927085e-01 3.35928053e-01 9.61766481e-01
-3.34949903e-02 -1.96367905e-01 -1.60908115e+00 1.28118598e+00
4.18850034e-01 -1.40635860e+00 -1.97867509e-02 -2.08541393e-01
8.09743881e-01 3.73538226e-01 2.21164405e-01 4.96368885e-01
3.23372841e-01 -9.10377204e-01 9.32004929e-01 4.57812339e-01
7.17976511e-01 -4.76370007e-01 2.99649388e-01 3.12193453e-01
-1.17515206e+00 -2.09910378e-01 3.41499620e-03 1.61586434e-01
1.28775746e-01 1.98008284e-01 -3.79525930e-01 2.71578819e-01
9.58483279e-01 1.65393651e+00 -6.75964594e-01 4.96723473e-01
-1.40312076e-01 1.01076388e+00 -6.19231611e-02 5.18850803e-01
4.37528908e-01 2.38459051e-01 4.14499760e-01 1.26493871e+00
-8.53622109e-02 -6.42523542e-02 4.21951056e-01 4.23822403e-01
-1.63002446e-01 -6.36725174e-03 -6.68572247e-01 -4.06182468e-01
3.74984443e-01 8.71307492e-01 -4.25672531e-01 -6.38880670e-01
-7.05221295e-01 9.82239008e-01 3.20467919e-01 2.84930795e-01
-9.42693591e-01 3.37621838e-01 5.44960499e-01 -5.68379015e-02
5.94922543e-01 7.45343789e-02 -7.80485570e-02 -1.45366967e+00
-2.01654285e-01 -1.09257388e+00 8.02569628e-01 -8.82679522e-01
-1.49294984e+00 3.84342432e-01 -1.17720366e-01 -1.84360468e+00
-3.78179342e-01 -4.90948826e-01 -2.77705371e-01 4.65842262e-02
-1.36781502e+00 -1.33980322e+00 -1.51860699e-01 1.31506574e+00
5.70030212e-01 -7.71475673e-01 8.54769588e-01 1.93904281e-01
-3.79305542e-01 6.55649483e-01 2.86563069e-01 3.44637305e-01
1.10088897e+00 -1.26528025e+00 -4.59519595e-01 5.91265738e-01
7.72813678e-01 3.20433289e-01 2.14503780e-01 -3.99669737e-01
-1.50079930e+00 -1.23493552e+00 3.98364097e-01 -5.50150812e-01
1.11107397e+00 -2.83713900e-02 -8.61402154e-01 7.74617374e-01
3.34419340e-01 1.26019791e-01 1.11223590e+00 2.62379199e-01
-9.94230092e-01 -3.01887691e-01 -5.51697612e-01 1.55188158e-01
9.54419017e-01 -1.20155525e+00 -6.74234271e-01 5.24051905e-01
8.33720028e-01 -8.25100094e-02 -1.00662887e+00 2.52526343e-01
5.57882965e-01 -1.15728104e+00 9.79093969e-01 -8.74685526e-01
9.08534765e-01 -1.90345854e-01 -6.09730959e-01 -1.19938302e+00
-5.91689944e-02 -4.63231832e-01 -4.26310241e-01 1.34604239e+00
4.70835537e-01 -1.47729039e-01 7.59254575e-01 2.18870595e-01
-2.45226845e-01 -5.40257573e-01 -8.70181501e-01 -1.20377016e+00
-1.99271485e-01 -8.47751856e-01 4.36461009e-02 1.24349844e+00
3.94399941e-01 5.73864937e-01 -6.98839784e-01 -4.52193357e-02
6.99179173e-01 2.73069620e-01 7.93749928e-01 -1.07579482e+00
-7.82715738e-01 -2.36008450e-01 -7.25412011e-01 -9.84360158e-01
7.25409091e-01 -1.14744067e+00 -8.29392076e-02 -9.90989566e-01
5.01442373e-01 -3.09359916e-02 -4.51279581e-01 8.51088703e-01
7.99524561e-02 4.53081548e-01 2.63738930e-01 5.46127617e-01
-1.42414284e+00 7.38965809e-01 7.59749711e-01 -5.18061876e-01
-1.06620096e-01 -1.06581599e-01 -6.21960104e-01 8.17269325e-01
6.78667188e-01 -2.86653012e-01 -5.80697417e-01 -4.55513865e-01
-1.84572279e-01 -2.31915414e-02 5.54065585e-01 -1.16574073e+00
2.81054318e-01 4.40425351e-02 2.90553510e-01 -5.29983103e-01
5.21408975e-01 -9.36813295e-01 -1.19202055e-01 1.91812024e-01
-8.36143434e-01 -4.66032773e-01 3.50816399e-02 1.10137093e+00
-7.22205341e-01 9.75914896e-02 7.39317060e-01 -2.54585985e-02
-1.13708198e+00 1.06536508e-01 -4.82613683e-01 3.26511174e-01
9.01107013e-01 -1.88502848e-01 -3.08946341e-01 -8.21032882e-01
-1.16117871e+00 1.46535754e-01 2.84170806e-01 6.19807482e-01
6.03816509e-01 -1.69006073e+00 -4.25786823e-01 -4.42986563e-02
5.33456802e-01 -5.19063413e-01 4.56420600e-01 1.24465263e+00
3.56252939e-01 5.33645689e-01 -2.53149629e-01 -1.16470742e+00
-1.50990415e+00 6.63611293e-01 5.30106016e-02 -1.13707580e-01
-3.73893946e-01 8.05870473e-01 1.74052536e-01 -2.81700104e-01
7.12301910e-01 -4.94500063e-02 -9.05506238e-02 2.38063708e-01
5.38888693e-01 1.61993682e-01 -1.60332590e-01 -1.11341095e+00
-4.37302083e-01 3.10162991e-01 2.58090273e-02 -2.19793946e-01
1.44073379e+00 -8.80496651e-02 9.40240845e-02 9.52262938e-01
1.48260939e+00 -4.91492629e-01 -1.54758096e+00 -7.08069980e-01
-7.90507272e-02 -4.25199896e-01 4.25535962e-02 -7.95504093e-01
-1.30341208e+00 8.60287011e-01 5.15034020e-01 -9.38039422e-02
1.19387782e+00 4.45071280e-01 5.31690121e-01 6.04614317e-01
2.80886441e-01 -1.43227577e+00 8.68318737e-01 4.93415385e-01
6.67815506e-01 -1.68232822e+00 -2.25995108e-01 -4.17569740e-04
-1.39198685e+00 8.73579264e-01 8.54607999e-01 1.24509133e-01
8.06625724e-01 -1.38184782e-02 9.65437815e-02 -2.76748121e-01
-1.17045629e+00 -4.89426404e-01 4.69780773e-01 7.74591804e-01
4.58958238e-01 2.17076913e-02 4.55682099e-01 5.66095233e-01
2.77293444e-01 -2.72720940e-02 2.00116113e-01 9.38340008e-01
-3.03209454e-01 -9.75521445e-01 -2.38586485e-01 4.64944035e-01
-3.23166251e-01 -2.77934484e-02 -5.48404753e-01 6.44671261e-01
8.07750225e-02 1.12083960e+00 1.17450900e-01 -7.76114523e-01
-6.53356388e-02 2.58886039e-01 3.19187045e-01 -6.23193145e-01
-3.55348796e-01 3.12114954e-01 4.54551838e-02 -9.52802777e-01
-1.23693466e+00 -9.55934227e-01 -1.24158788e+00 -3.32614481e-02
-7.86066428e-03 -1.62504315e-02 2.36278206e-01 1.12586820e+00
4.75142300e-01 2.24015042e-01 6.96517944e-01 -8.52841079e-01
-4.45533216e-01 -8.33991289e-01 -7.74363697e-01 9.09298718e-01
2.34062687e-01 -9.15499270e-01 -4.10510212e-01 5.26432991e-01] | [9.630102157592773, 1.034551978111267] |
fcb9fa38-11ab-4ad6-9da8-c778bf02cf1c | stack-based-multi-layer-attention-for | null | null | https://aclanthology.org/D17-1175 | https://aclanthology.org/D17-1175.pdf | Stack-based Multi-layer Attention for Transition-based Dependency Parsing | Although sequence-to-sequence (seq2seq) network has achieved significant success in many NLP tasks such as machine translation and text summarization, simply applying this approach to transition-based dependency parsing cannot yield a comparable performance gain as in other state-of-the-art methods, such as stack-LSTM and head selection. In this paper, we propose a stack-based multi-layer attention model for seq2seq learning to better leverage structural linguistics information. In our method, two binary vectors are used to track the decoding stack in transition-based parsing, and multi-layer attention is introduced to capture multiple word dependencies in partial trees. We conduct experiments on PTB and CTB datasets, and the results show that our proposed model achieves state-of-the-art accuracy and significant improvement in labeled precision with respect to the baseline seq2seq model. | ['Ming Zhou', 'Shujie Liu', 'Zhirui Zhang', 'Mu Li', 'Enhong Chen'] | 2017-09-01 | null | null | null | emnlp-2017-9 | ['transition-based-dependency-parsing'] | ['natural-language-processing'] | [ 3.56946021e-01 2.71501541e-01 -5.80011189e-01 -6.58927739e-01
-1.43591201e+00 -5.85004747e-01 2.93068916e-01 3.59778225e-01
-6.50890946e-01 8.07355106e-01 7.17987895e-01 -7.95152128e-01
4.11555737e-01 -4.26053077e-01 -8.81910086e-01 -3.61313343e-01
1.45918399e-01 4.39553767e-01 3.68096590e-01 -2.05812812e-01
2.82901555e-01 1.28051275e-02 -9.35754478e-01 5.96123219e-01
1.13365328e+00 5.96107662e-01 4.07889187e-01 8.15344691e-01
-6.55471802e-01 9.10646617e-01 -6.49787128e-01 -5.78164220e-01
-2.66062498e-01 -3.97846282e-01 -1.06707191e+00 -4.59466100e-01
4.96884823e-01 -1.95838451e-01 -1.37797639e-01 1.04381526e+00
5.70953369e-01 -6.72470629e-02 2.16863737e-01 -3.93499643e-01
-4.18454349e-01 1.19777632e+00 -6.87985778e-01 3.84283721e-01
3.56530398e-01 -4.35385369e-02 1.45123363e+00 -6.21425569e-01
5.15968263e-01 1.40809250e+00 6.11265123e-01 6.47866189e-01
-1.05146420e+00 -6.07945144e-01 5.61600685e-01 1.38602927e-01
-5.35080731e-01 -2.92996377e-01 4.07588989e-01 -6.37947470e-02
1.52876198e+00 8.91433284e-02 2.80005872e-01 1.24147296e+00
4.82679814e-01 1.35482705e+00 8.37374747e-01 -5.20513058e-01
-6.19385615e-02 -3.76048684e-01 5.84634840e-01 7.06455410e-01
1.96867004e-01 -3.48016292e-01 -6.37677908e-01 1.66774586e-01
4.09107059e-01 -4.31275964e-01 -3.94997261e-02 2.75316477e-01
-1.20003259e+00 9.52673614e-01 3.68894726e-01 3.11653674e-01
-4.29515332e-01 2.42244124e-01 7.80286431e-01 -3.63364182e-02
6.83526278e-01 3.73640895e-01 -7.67407060e-01 -4.29541856e-01
-9.17673469e-01 1.20214731e-01 9.50089455e-01 9.41748857e-01
3.64010125e-01 -1.35225337e-02 -6.01692438e-01 8.65567327e-01
1.36982411e-01 4.44627613e-01 6.19770050e-01 -9.28823173e-01
1.11724317e+00 5.26541591e-01 -1.68294892e-01 -2.91794926e-01
-5.43843746e-01 -2.42697090e-01 -6.88034415e-01 -4.78003025e-01
2.53921479e-01 -5.08310735e-01 -1.18471491e+00 1.73207891e+00
1.76908791e-01 4.82294336e-02 2.07488760e-01 6.50482118e-01
1.04775345e+00 9.41533506e-01 3.94095957e-01 -2.52836347e-01
1.47087359e+00 -1.34111929e+00 -9.59776103e-01 -6.46653593e-01
1.07331240e+00 -5.45420229e-01 9.90856886e-01 5.57659268e-02
-1.27454162e+00 -3.51235092e-01 -7.81258166e-01 -3.71990561e-01
3.98511253e-02 -6.72179535e-02 6.08913243e-01 4.54160541e-01
-8.26101243e-01 5.24014235e-01 -1.21917403e+00 -4.01978910e-01
3.28959703e-01 3.01560223e-01 -3.39083731e-01 -4.42676097e-02
-1.22235250e+00 9.28796709e-01 5.28113127e-01 2.42347226e-01
-6.92465305e-01 -5.15149295e-01 -1.05829251e+00 3.35075110e-01
4.00487363e-01 -5.09386480e-01 1.76703298e+00 -4.88949627e-01
-1.68882418e+00 6.02225423e-01 -6.24769807e-01 -5.50342202e-01
9.45816189e-03 -7.04180181e-01 -9.87529103e-03 -1.56863764e-01
1.10349745e-01 5.73280752e-01 2.14069813e-01 -7.33632565e-01
-9.12203670e-01 -3.71723145e-01 8.72887895e-02 2.01323494e-01
-2.23347396e-01 5.38254857e-01 -6.72876656e-01 -4.24858510e-01
-5.39213382e-02 -8.96742046e-01 -5.58029115e-01 -8.52026105e-01
-7.02434778e-01 -4.49200898e-01 3.65405709e-01 -1.03519571e+00
1.58373487e+00 -1.79526401e+00 3.36983204e-01 -4.70281988e-01
-2.90414065e-01 7.50026286e-01 -4.44265068e-01 5.68301380e-01
1.56129241e-01 2.96504945e-01 -4.59933698e-01 -5.95920384e-01
-1.88076645e-01 3.45168978e-01 -1.79438189e-01 -2.24171299e-02
5.43880522e-01 1.12084651e+00 -1.02962184e+00 -5.91315269e-01
-2.08331913e-01 3.45537901e-01 -4.23381656e-01 3.44742566e-01
-5.71549296e-01 4.97554064e-01 -6.91516280e-01 5.36122441e-01
4.72588450e-01 -1.30463094e-01 6.10835016e-01 2.01368824e-01
-1.46990120e-01 1.14808381e+00 -3.44655991e-01 2.00883245e+00
-6.71914458e-01 4.30900276e-01 -6.00525877e-03 -9.30560291e-01
6.57830477e-01 3.27122480e-01 -1.47113413e-01 -8.31463397e-01
2.86746435e-02 8.56103078e-02 2.35526130e-01 -4.58452821e-01
7.34292269e-01 3.99778560e-02 -4.65486854e-01 2.34994620e-01
1.50349036e-01 1.32785127e-01 2.22911030e-01 1.50960520e-01
1.31961751e+00 3.23845178e-01 3.31011623e-01 -1.50247797e-01
5.58301151e-01 7.48970211e-02 9.92545605e-01 7.04607666e-01
8.38318001e-03 4.85006005e-01 8.24508071e-01 -1.33014381e-01
-8.52935970e-01 -5.36353052e-01 1.96426332e-01 1.45389020e+00
-3.20249677e-01 -3.90204906e-01 -1.07392669e+00 -1.09782302e+00
-2.69676447e-01 9.39509511e-01 -3.83205861e-01 2.09768653e-01
-1.38364851e+00 -9.09338117e-01 7.14840174e-01 9.82417107e-01
5.41780412e-01 -1.42493749e+00 -3.37415606e-01 6.41310930e-01
-3.63635480e-01 -1.41361821e+00 -6.12886488e-01 4.93275523e-01
-1.18339860e+00 -7.32046843e-01 -5.85004032e-01 -9.90818679e-01
1.69358552e-01 -1.21282257e-01 1.20627308e+00 -1.58451319e-01
2.13927299e-01 -3.57189447e-01 -6.28642797e-01 -4.42173272e-01
-5.00025928e-01 7.69419193e-01 -4.15730953e-01 -3.71846080e-01
4.14943457e-01 -1.83293417e-01 -4.12773997e-01 -2.39059061e-01
-6.35057807e-01 -9.46889520e-02 8.15001070e-01 1.04509664e+00
4.68019217e-01 -7.10446000e-01 6.14170730e-01 -1.43487382e+00
7.01273143e-01 -2.33648092e-01 -4.80307460e-01 3.78655374e-01
-3.28859270e-01 2.47131735e-01 6.93145812e-01 -1.97434723e-01
-1.39245772e+00 8.29715002e-03 -5.83643019e-01 4.39743735e-02
-2.79733725e-02 9.98205066e-01 -3.31696391e-01 4.88211244e-01
1.01754829e-01 1.74284235e-01 -3.88434798e-01 -7.20770597e-01
4.08947259e-01 7.43389606e-01 4.70738113e-01 -4.34843510e-01
1.88751593e-01 -1.10445425e-01 -1.79024369e-01 -6.14834130e-01
-1.37021828e+00 -3.85584176e-01 -7.07793653e-01 4.87724274e-01
1.26897776e+00 -8.21060181e-01 -5.53512633e-01 5.63011467e-01
-1.55710757e+00 -5.50240874e-01 1.20882951e-01 3.77254754e-01
-3.20659578e-01 3.71513158e-01 -1.13885951e+00 -8.26341331e-01
-7.46154845e-01 -1.18794167e+00 1.14632154e+00 3.91310185e-01
-1.54891506e-01 -9.18763757e-01 1.95938215e-01 4.79356021e-01
2.01980725e-01 -1.10085502e-01 1.17692029e+00 -1.07392478e+00
-5.74358642e-01 -6.00784533e-02 -8.95993635e-02 3.38047475e-01
-1.84193924e-01 -2.13445455e-01 -8.96204829e-01 -1.34738103e-01
-3.44507217e-01 -4.18771178e-01 1.22377312e+00 5.31055093e-01
1.05440664e+00 -2.52160281e-01 -4.97350663e-01 5.65524518e-01
1.14131224e+00 1.64703935e-01 6.21043742e-01 3.86284173e-01
1.01612639e+00 6.36631787e-01 7.82393277e-01 -2.55204458e-02
6.20101035e-01 4.36101586e-01 2.81510502e-01 4.56308797e-02
-5.85370697e-02 -4.70251471e-01 5.51100850e-01 1.25098073e+00
2.11280584e-01 -5.85325122e-01 -9.45171297e-01 7.35337257e-01
-1.95698369e+00 -5.48671663e-01 -2.45794520e-01 2.01539111e+00
9.61896122e-01 4.14871097e-01 -9.18317884e-02 -3.46888602e-01
8.11697483e-01 5.01304865e-01 -4.75382447e-01 -1.00669527e+00
-2.76577733e-02 4.41949993e-01 5.87426841e-01 5.52992821e-01
-1.07749462e+00 1.58786666e+00 6.43239880e+00 6.53675795e-01
-1.14095068e+00 3.05905700e-01 5.94066620e-01 1.78924501e-02
-3.18179041e-01 -9.51760076e-03 -1.28087735e+00 3.28254580e-01
1.37689912e+00 2.59869069e-01 -5.31463400e-02 5.73154271e-01
5.34111517e-04 6.72750995e-02 -1.07561707e+00 2.93417543e-01
-4.27550450e-02 -1.18271947e+00 -5.16747311e-02 -1.82396943e-05
6.03705347e-01 4.70247567e-01 -1.49073869e-01 6.17263258e-01
6.10869944e-01 -9.79890466e-01 4.29389060e-01 2.29675621e-02
7.67864168e-01 -7.71496296e-01 1.05665851e+00 5.06552935e-01
-1.02730227e+00 -1.60259888e-01 -3.93452793e-01 1.34916161e-03
7.17432499e-01 4.69970345e-01 -7.80754507e-01 7.30320454e-01
5.82308412e-01 4.94726956e-01 -1.74512520e-01 7.16935039e-01
-7.65883267e-01 1.29165173e+00 1.70814805e-02 -4.18094218e-01
5.87206244e-01 3.82524692e-02 4.64726329e-01 1.71204424e+00
6.01384006e-02 1.51250765e-01 1.43318519e-01 3.49956870e-01
-4.61382926e-01 2.47070834e-01 -2.05856577e-01 -2.55057424e-01
5.49666762e-01 1.12383711e+00 -4.34193164e-01 -6.01109028e-01
-7.10310340e-01 8.98143888e-01 8.30503583e-01 1.98909998e-01
-7.08899736e-01 -4.90021944e-01 6.41128778e-01 -2.69279122e-01
8.29502583e-01 -3.05837095e-01 -3.27302188e-01 -1.10921240e+00
5.30305505e-03 -9.04560268e-01 4.26506549e-01 -3.40834826e-01
-1.03915012e+00 8.08772564e-01 -1.96400061e-01 -5.09877980e-01
-3.49773198e-01 -5.79963148e-01 -7.68744528e-01 1.05618143e+00
-1.72618890e+00 -1.30479109e+00 2.47865230e-01 -1.33553192e-01
8.77875805e-01 9.71533358e-02 9.19065833e-01 -2.58101765e-02
-8.33529830e-01 7.62649596e-01 1.99346378e-01 3.83937478e-01
6.87500894e-01 -1.32835412e+00 1.23683941e+00 9.79812086e-01
1.89927872e-02 5.96294701e-01 3.88019770e-01 -8.36190462e-01
-1.37293088e+00 -1.21423817e+00 1.52355385e+00 -3.65399510e-01
6.02624953e-01 -4.92833018e-01 -1.07925534e+00 1.06281829e+00
4.51537609e-01 -1.77409247e-01 4.86739904e-01 5.84834635e-01
-3.46998930e-01 1.26508892e-01 -6.57448888e-01 4.42624927e-01
1.05140829e+00 -3.94133836e-01 -8.86010468e-01 1.51476264e-01
1.29079497e+00 -7.30246782e-01 -7.69914865e-01 5.51635861e-01
4.95276064e-01 -6.98289335e-01 5.35448849e-01 -9.01326835e-01
7.36252964e-01 2.18808189e-01 -6.68952614e-02 -1.48974073e+00
-4.65691090e-01 -4.66390073e-01 -1.70655251e-02 1.41363323e+00
9.39370096e-01 -5.57061851e-01 7.96550870e-01 2.44053021e-01
-5.77652514e-01 -1.01132631e+00 -9.51107144e-01 -6.10069394e-01
4.50428337e-01 -1.96874097e-01 5.45980930e-01 5.89223742e-01
1.41092330e-01 8.94044280e-01 -3.17929029e-01 1.19403280e-01
2.71288186e-01 7.93111175e-02 4.88736719e-01 -9.40646052e-01
-3.68500143e-01 -2.64995456e-01 3.18134911e-02 -1.45123672e+00
6.98380113e-01 -9.26011264e-01 4.58711267e-01 -1.95952988e+00
4.15657341e-01 -2.13572040e-01 -3.30560416e-01 4.88559365e-01
-7.50968635e-01 -1.63807392e-01 2.56464779e-01 -1.55419618e-01
-7.09779561e-01 5.56450188e-01 1.20467186e+00 -1.04555488e-01
-2.16734111e-01 -1.69837713e-01 -7.70735860e-01 5.60748756e-01
6.70392811e-01 -6.26024306e-01 1.05226979e-01 -1.01619911e+00
-2.10740846e-02 5.77793658e-01 -4.12275761e-01 -5.06892562e-01
1.94789395e-01 1.05750434e-01 4.18135040e-02 -7.38653481e-01
1.95944294e-01 -1.43495932e-01 -6.75919652e-01 4.49047536e-01
-5.94568312e-01 1.56844854e-01 3.05606782e-01 4.65220302e-01
-3.04554224e-01 -4.75557566e-01 3.55116129e-01 -2.86412090e-01
-4.81738240e-01 1.50563821e-01 -4.09237832e-01 2.85176605e-01
3.40731382e-01 2.60817736e-01 -5.01691043e-01 -1.64481863e-01
-2.94120580e-01 5.21311164e-01 7.77935609e-02 4.26356524e-01
2.32295647e-01 -8.15346003e-01 -9.63004947e-01 -3.22301686e-02
-3.51671912e-02 4.15645391e-01 1.06071718e-01 6.54834747e-01
-3.58352154e-01 8.44776571e-01 1.36604652e-01 -5.53977013e-01
-1.35495329e+00 3.06433886e-01 -4.08824719e-02 -8.92264485e-01
-5.71866632e-01 1.24282444e+00 3.52643132e-01 -8.33682835e-01
2.58336037e-01 -6.48042560e-01 -3.63470048e-01 -7.17949048e-02
5.31281292e-01 7.48707801e-02 2.22623363e-01 -4.85160023e-01
-5.47704041e-01 4.30064321e-01 -6.26266539e-01 -1.32361889e-01
1.31025517e+00 1.10020243e-01 -7.80405477e-02 4.45293784e-01
9.75647748e-01 -7.26261660e-02 -1.13909924e+00 -1.65916741e-01
5.03876448e-01 9.39529296e-03 4.57369499e-02 -1.00630510e+00
-8.01641703e-01 1.26680672e+00 -2.03945115e-02 -2.02469781e-01
8.44152093e-01 7.28142709e-02 1.42776132e+00 4.99931544e-01
1.74962744e-01 -8.76946330e-01 -2.91858017e-01 1.15038240e+00
4.94181842e-01 -1.31166315e+00 -3.55664194e-01 -4.42980826e-01
-7.66905904e-01 8.07929695e-01 8.30144823e-01 5.66639891e-03
1.75465941e-01 4.54027981e-01 2.16125935e-01 1.28219485e-01
-1.04232204e+00 -2.34223247e-01 6.32881746e-02 3.34695965e-01
1.07832170e+00 1.33397311e-01 -6.49537086e-01 7.91137099e-01
-2.64128018e-02 -2.03674227e-01 4.51403201e-01 1.01798260e+00
-6.88129604e-01 -1.49799383e+00 -6.88578486e-02 3.10123026e-01
-1.07820833e+00 -5.69210768e-01 -2.55153060e-01 6.77933753e-01
-5.18242002e-01 8.76513898e-01 2.05466837e-01 -2.95636922e-01
5.22200763e-01 4.10407364e-01 5.09043574e-01 -9.80242193e-01
-1.08372664e+00 2.76184559e-01 6.86283708e-01 -6.63880169e-01
-2.64401287e-01 -6.61997378e-01 -1.48837352e+00 -1.15388045e-02
-4.93561774e-01 2.43843287e-01 8.37461829e-01 9.47300136e-01
3.89543384e-01 8.35119188e-01 3.10511351e-01 -7.52525032e-01
-6.94191277e-01 -1.40549624e+00 -2.28811234e-01 8.96557644e-02
2.72734702e-01 -9.73902345e-02 4.84800227e-02 -3.05921078e-01] | [10.414997100830078, 9.559991836547852] |
36d89537-e7cc-44df-97c8-7e2ad46e6bb1 | prompt-pre-training-with-twenty-thousand | 2304.04704 | null | https://arxiv.org/abs/2304.04704v1 | https://arxiv.org/pdf/2304.04704v1.pdf | Prompt Pre-Training with Twenty-Thousand Classes for Open-Vocabulary Visual Recognition | This work proposes POMP, a prompt pre-training method for vision-language models. Being memory and computation efficient, POMP enables the learned prompt to condense semantic information for a rich set of visual concepts with over twenty-thousand classes. Once pre-trained, the prompt with a strong transferable ability can be directly plugged into a variety of visual recognition tasks including image classification, semantic segmentation, and object detection, to boost recognition performances in a zero-shot manner. Empirical evaluation shows that POMP achieves state-of-the-art performances on 21 downstream datasets, e.g., 67.0% average accuracy on 10 classification dataset (+3.1% compared to CoOp) and 84.4 hIoU on open-vocabulary Pascal VOC segmentation (+6.9 compared to ZSSeg). | ['Xu sun', 'Alex Smola', 'Mu Li', 'Shuai Zheng', 'Shuai Zhang', 'Yi Zhu', 'Aston Zhang', 'Shuhuai Ren'] | 2023-04-10 | null | null | null | null | ['open-vocabulary-object-detection'] | ['computer-vision'] | [ 1.91837698e-01 2.08187446e-01 -5.90354979e-01 -2.39699274e-01
-1.18868852e+00 -8.87875259e-01 9.52210665e-01 1.43173143e-01
-6.41678870e-01 4.25350934e-01 -1.83691651e-01 -1.54808134e-01
4.87415701e-01 -7.33833790e-01 -9.87463057e-01 -6.76638424e-01
3.50111604e-01 7.39080489e-01 6.80711031e-01 1.34811968e-01
2.88402528e-01 1.59269363e-01 -1.93376350e+00 5.09496093e-01
4.88635361e-01 1.27011704e+00 6.97917759e-01 9.90937471e-01
-4.52281296e-01 9.77897763e-01 -4.63719845e-01 -4.10589784e-01
-6.00252822e-02 9.51273963e-02 -9.24729049e-01 -4.28749584e-02
6.41347408e-01 -3.82997245e-01 -2.46687889e-01 1.07910776e+00
2.74028629e-01 3.41507971e-01 7.03553319e-01 -1.23929822e+00
-7.10666537e-01 3.13489318e-01 -2.80212581e-01 1.99452057e-01
1.96903273e-01 5.73319852e-01 1.15332985e+00 -1.12676334e+00
8.70091558e-01 1.21661782e+00 1.39931917e-01 9.37428653e-01
-9.36556041e-01 -4.48612720e-01 1.23637848e-01 3.30435902e-01
-9.96357918e-01 -4.48749572e-01 2.56455421e-01 -4.05223966e-01
1.51177382e+00 2.77814299e-01 8.87761056e-01 1.23813748e+00
1.15486622e-01 1.36708379e+00 1.16237950e+00 -2.77965605e-01
6.08406544e-01 3.59667582e-03 4.82872874e-01 8.52138340e-01
-1.84019685e-01 2.04435974e-01 -1.04182959e+00 1.95882171e-01
3.34306657e-01 -3.61490965e-01 2.48305753e-01 -1.17894910e-01
-8.86864126e-01 7.82892704e-01 5.67261100e-01 -3.21088225e-01
-3.31661969e-01 4.33796078e-01 3.38166893e-01 -1.70322552e-01
3.63584965e-01 5.26912093e-01 -4.85644192e-01 -3.50043476e-01
-8.05623412e-01 3.31009060e-01 6.70625567e-01 1.12010062e+00
5.44407606e-01 1.02748998e-01 -5.86893499e-01 5.98143101e-01
4.93192405e-01 9.51416016e-01 3.92404854e-01 -1.17153811e+00
2.70030469e-01 2.53398985e-01 4.11340315e-03 -2.30781242e-01
-1.96386084e-01 -1.20220289e-01 -2.93940187e-01 2.04267055e-01
3.32130075e-01 1.21168733e-01 -1.66272950e+00 1.45833063e+00
2.32504398e-01 2.47563645e-01 2.85212219e-01 8.27524781e-01
1.53155339e+00 1.18970716e+00 6.53927028e-01 -1.51250765e-01
1.56697869e+00 -1.41957927e+00 -2.37803191e-01 -5.62708318e-01
2.18933508e-01 -5.21864057e-01 1.19698143e+00 5.04610658e-01
-1.04866183e+00 -5.43159127e-01 -8.95741463e-01 -3.14369798e-01
-4.22134757e-01 -1.77671269e-01 8.87211204e-01 3.99644852e-01
-1.26606417e+00 2.72452891e-01 -9.79540944e-01 -3.73522639e-01
9.46040213e-01 1.92288265e-01 -2.92824563e-02 -2.61859953e-01
-6.97255850e-01 6.31196618e-01 6.76684558e-01 -6.70239806e-01
-1.71507800e+00 -7.87367940e-01 -6.84673309e-01 1.58373386e-01
4.21888441e-01 -7.07118809e-01 1.73804665e+00 -8.26865196e-01
-1.60866952e+00 1.20344174e+00 -2.44716257e-01 -8.13629150e-01
3.48425925e-01 -2.49177337e-01 -1.23806454e-01 5.43300807e-01
2.43265599e-01 1.68089449e+00 1.10074914e+00 -8.86505961e-01
-8.98542166e-01 -2.04391971e-01 1.53109327e-01 5.18261909e-01
-1.56337157e-01 2.26022929e-01 -1.01578891e+00 -2.13738263e-01
-6.42425343e-02 -8.86284649e-01 -1.46850422e-01 1.36039242e-01
-1.32650569e-01 -5.00770092e-01 7.94751346e-01 -5.84171534e-01
3.06741059e-01 -1.97031343e+00 -9.29401489e-04 -4.01897728e-01
2.24839970e-01 5.53668261e-01 -4.04030830e-01 -1.35269966e-02
3.46435308e-01 -1.41307220e-01 -3.02774496e-02 -4.72760260e-01
1.59838516e-02 3.35291922e-01 -5.97972095e-01 5.92865013e-02
5.02063096e-01 1.52561593e+00 -1.11225104e+00 -5.22887349e-01
5.21412313e-01 1.46642283e-01 -4.01093334e-01 2.20809326e-01
-9.21653628e-01 2.80194134e-01 -3.37911427e-01 8.92549038e-01
2.54713714e-01 -4.01357204e-01 -1.71904013e-01 4.33230430e-01
4.78402860e-02 2.20262378e-01 -5.58967710e-01 1.96464980e+00
-2.47845845e-03 8.59138727e-01 -1.46393031e-01 -8.37120533e-01
7.54933119e-01 1.09762579e-01 8.36139470e-02 -1.01011562e+00
1.98597357e-01 -2.62396127e-01 -5.06618142e-01 -2.41184399e-01
7.52959728e-01 3.18612233e-02 -1.07721902e-01 1.28897369e-01
5.97272635e-01 -4.76264358e-01 1.50727853e-01 3.28489989e-01
9.55238819e-01 -7.96747804e-02 7.26093873e-02 -3.24772179e-01
1.69162840e-01 5.21372557e-01 2.56458610e-01 1.05645108e+00
-6.21147215e-01 6.26211643e-01 3.47899407e-01 -2.78757751e-01
-7.27291226e-01 -1.40670705e+00 2.51929834e-02 1.58355892e+00
2.86641866e-01 -5.01066864e-01 -8.13073516e-01 -5.13385236e-01
-1.78867251e-01 8.91811907e-01 -2.08122686e-01 -1.63316533e-01
-1.44235730e-01 -5.83499968e-01 5.37289619e-01 9.43935215e-01
6.98781192e-01 -1.37565970e+00 -8.48195612e-01 -6.13803864e-02
-1.73008308e-01 -1.28283226e+00 2.29464229e-02 3.06375384e-01
-8.19174528e-01 -8.93147409e-01 -5.10589361e-01 -9.10730481e-01
3.03029597e-01 3.79739612e-01 1.31031394e+00 -2.92041838e-01
-3.37185800e-01 7.34427214e-01 -3.18594456e-01 -8.23172688e-01
-1.26522183e-01 -1.04220487e-01 -1.24344796e-01 -3.98895085e-01
4.72986162e-01 1.27656654e-01 -3.63101155e-01 -2.36140057e-01
-4.77454156e-01 4.88187164e-01 2.70401895e-01 7.88832068e-01
1.00183833e+00 -5.10454357e-01 1.64524123e-01 -2.96173930e-01
1.83015287e-01 -1.64873585e-01 -9.64279830e-01 3.31526399e-01
-5.57934940e-01 -2.57738237e-03 2.19552130e-01 -4.03108120e-01
-1.11467791e+00 3.26502830e-01 -1.25007182e-01 -5.63561320e-01
-3.59821498e-01 2.76801467e-01 1.84595156e-02 -4.16409448e-02
6.22580230e-01 4.92337108e-01 -3.66871893e-01 -1.49056569e-01
1.05016065e+00 5.12014627e-01 1.00065672e+00 -5.81800759e-01
3.10945064e-01 4.94803250e-01 -2.31488556e-01 -9.99103069e-01
-9.80273664e-01 -7.11510837e-01 -4.11799550e-01 -2.76503325e-01
1.34722924e+00 -1.23460126e+00 -6.97618127e-01 7.83837020e-01
-1.19730151e+00 -7.30552256e-01 -4.58947450e-01 1.75708354e-01
-7.92476475e-01 -1.27247572e-01 -7.82281280e-01 -7.88695931e-01
-7.16447949e-01 -1.00747442e+00 1.27839231e+00 5.85675001e-01
-9.10005867e-02 -4.68900591e-01 -1.34331152e-01 6.70623302e-01
-6.55041859e-02 -3.10952365e-01 6.67014599e-01 -5.88935375e-01
-1.10082138e+00 1.28375560e-01 -5.59611201e-01 2.10691825e-01
-7.11449623e-01 1.08434178e-03 -1.43054032e+00 -1.48694992e-01
-3.24381679e-01 -9.89668369e-01 1.28020442e+00 5.83800256e-01
1.28849220e+00 -2.09972821e-02 -3.91533226e-01 6.00587010e-01
1.29115629e+00 3.15607995e-01 5.27856529e-01 8.88921469e-02
6.51033163e-01 3.61475110e-01 9.17745709e-01 2.36398146e-01
4.31514323e-01 4.36950356e-01 6.52665019e-01 4.01337475e-01
-3.64995241e-01 -4.02159572e-01 3.26193601e-01 5.80768049e-01
1.55886188e-01 -3.17797422e-01 -1.38097835e+00 6.13631666e-01
-1.88785243e+00 -6.83904707e-01 5.75818904e-02 1.79661095e+00
6.33061945e-01 2.78338253e-01 -7.46302307e-02 -5.35389483e-01
3.18484962e-01 4.05702412e-01 -7.19359338e-01 -3.04057389e-01
-1.15484111e-01 4.55423743e-01 5.99321902e-01 3.93298090e-01
-1.36574101e+00 1.74553645e+00 6.89596844e+00 1.03779209e+00
-1.10870218e+00 2.24447444e-01 7.43804395e-01 -3.41283590e-01
1.42433783e-02 -1.66567177e-01 -1.02123654e+00 1.69151589e-01
1.23762679e+00 -4.98354658e-02 5.11123121e-01 1.08165562e+00
-4.38514858e-01 -3.71108562e-01 -9.84669864e-01 1.51649940e+00
1.54866144e-01 -1.82780719e+00 9.66429710e-02 -4.56289530e-01
7.94902623e-01 6.66412830e-01 3.34379703e-01 7.89619565e-01
6.63282156e-01 -1.22259974e+00 1.13481152e+00 3.72287989e-01
9.49843168e-01 -7.05856800e-01 2.61525363e-01 6.09599829e-01
-1.06831467e+00 4.25834842e-02 -4.52122897e-01 -3.43071848e-01
-8.87386035e-03 4.28818911e-02 -9.75608051e-01 1.69948369e-01
9.27483618e-01 6.16335988e-01 -6.50618315e-01 8.37626457e-01
-6.19026423e-01 8.23825479e-01 -2.02825308e-01 -4.59243864e-01
6.39199138e-01 2.37573743e-01 4.65752959e-01 1.10640931e+00
-1.43250644e-01 4.45572138e-01 2.98809290e-01 7.93746114e-01
-2.30214402e-01 -1.45720556e-01 -4.26914006e-01 -3.12339306e-01
4.97335762e-01 1.18240154e+00 -1.08520603e+00 -9.08833683e-01
-2.23089665e-01 1.07223141e+00 4.04759139e-01 2.76903927e-01
-8.94923270e-01 1.28008440e-01 6.48392022e-01 -5.58835208e-01
4.59036738e-01 -5.85256107e-02 -4.46244359e-01 -9.87385392e-01
-1.79910585e-01 -6.95616424e-01 3.44862849e-01 -1.21505308e+00
-9.55621421e-01 2.47228861e-01 -6.03506565e-02 -5.17726362e-01
-3.66076320e-01 -1.01569092e+00 -4.72148091e-01 5.24525404e-01
-1.22634327e+00 -1.26186049e+00 -2.88374782e-01 5.08175552e-01
9.74702895e-01 -3.01323831e-01 1.02015185e+00 -3.04024905e-01
-3.37476671e-01 1.94897801e-01 -2.58914735e-02 6.48766235e-02
2.62919009e-01 -1.26747000e+00 7.50082910e-01 9.46404815e-01
6.28860950e-01 -1.99122801e-02 5.76288640e-01 -5.46322823e-01
-1.58235705e+00 -1.33534110e+00 7.77514398e-01 -1.00400496e+00
5.15407562e-01 -4.60020632e-01 -5.98796487e-01 5.41775525e-01
3.02325785e-01 1.61553547e-01 3.00468594e-01 1.30060781e-02
-6.26792192e-01 6.24214560e-02 -9.42719221e-01 7.15851545e-01
1.02857149e+00 -9.23916578e-01 -8.92016530e-01 5.11254489e-01
1.15511990e+00 -7.00644910e-01 -1.55365482e-01 1.99008808e-01
4.33034688e-01 -5.88479996e-01 1.23324084e+00 -8.31304312e-01
4.33638275e-01 -3.39467078e-02 -4.87202346e-01 -8.58169436e-01
-2.47542068e-01 -2.39622653e-01 -3.79989713e-01 9.32632744e-01
1.32385164e-01 -3.03784430e-01 9.77494121e-01 7.07396746e-01
-1.87746227e-01 -6.09749436e-01 -9.21380043e-01 -7.26236939e-01
-8.15527737e-02 -1.04601955e+00 1.83388427e-01 5.91303468e-01
-1.81262404e-01 5.98902166e-01 -2.35528685e-02 8.74322206e-02
5.74983478e-01 1.62000626e-01 6.34344161e-01 -1.10790265e+00
-3.34093094e-01 -4.10773546e-01 -4.27904755e-01 -1.17772269e+00
3.82274926e-01 -1.15838170e+00 3.22342992e-01 -1.60770380e+00
5.43638408e-01 -1.73358887e-01 -5.60975730e-01 7.37091899e-01
4.80086990e-02 2.47140780e-01 6.20364606e-01 1.33981243e-01
-1.13660526e+00 6.01080835e-01 1.03434694e+00 -5.86639225e-01
-1.30538136e-01 -2.51003951e-01 -4.97198194e-01 7.95113623e-01
7.24236727e-01 -4.65527147e-01 -5.39607048e-01 -4.44881678e-01
-2.48920098e-01 1.03018358e-01 6.90070570e-01 -1.04122615e+00
3.30457479e-01 -2.72124559e-01 2.57124782e-01 -9.78368282e-01
7.72556484e-01 -2.34285340e-01 -4.40605938e-01 4.87555146e-01
-2.49201819e-01 -5.24623275e-01 5.85844338e-01 6.85521603e-01
-1.24605983e-01 -9.02308822e-02 7.41700053e-01 -2.92390168e-01
-1.69452679e+00 2.95215666e-01 -3.88038456e-01 2.15044394e-01
1.10948575e+00 6.58678487e-02 -7.66349912e-01 -7.23309815e-02
-6.73076808e-01 4.59072322e-01 2.62936860e-01 7.28665590e-01
8.05276155e-01 -9.59363937e-01 -4.13934797e-01 1.16119497e-01
3.94169450e-01 9.40808281e-02 1.10636048e-01 3.77081871e-01
-4.69636351e-01 8.25778842e-01 -4.49111044e-01 -1.16012967e+00
-1.42093384e+00 5.89408576e-01 -5.46290614e-02 -1.20008796e-01
-7.28334963e-01 1.47899163e+00 4.02337253e-01 -2.05308631e-01
4.94624853e-01 -5.00718772e-01 1.97161827e-02 2.31308118e-01
7.71136045e-01 1.47467420e-01 -1.16222583e-01 -2.87212431e-01
-5.71007252e-01 2.26246133e-01 -1.81347709e-02 -4.58304644e-01
1.01947367e+00 2.08074912e-01 8.50973278e-02 5.34927905e-01
8.95254672e-01 -8.09473336e-01 -1.66761506e+00 -1.51732862e-01
-2.20798664e-02 -1.45997077e-01 2.63957590e-01 -9.81965542e-01
-7.03196704e-01 1.09236431e+00 7.90532649e-01 -4.04163957e-01
8.05990756e-01 5.05320430e-01 6.09530389e-01 7.54914165e-01
6.76278830e-01 -1.25836802e+00 4.19252068e-01 8.87438893e-01
5.69445610e-01 -1.57664573e+00 -4.01591182e-01 -3.36156726e-01
-9.37523544e-01 7.85703301e-01 6.59628153e-01 -7.94653744e-02
3.11737657e-01 1.06994607e-01 -3.80118899e-02 -2.13440076e-01
-1.09614086e+00 -5.02485752e-01 4.89079416e-01 7.56067038e-01
-1.51709974e-01 5.81390798e-01 1.97958887e-01 6.64718568e-01
-7.41240978e-02 -5.49584962e-02 1.25549898e-01 8.50472808e-01
-8.93114448e-01 -5.40199876e-01 -5.55741228e-02 4.49829966e-01
-6.04791306e-02 -3.41244191e-01 -3.91790032e-01 4.96219426e-01
-2.91398428e-02 9.77512598e-01 3.80810529e-01 -2.45924011e-01
-1.00063540e-01 4.92701322e-01 5.77165067e-01 -8.13343644e-01
-2.56428689e-01 -3.86733711e-02 2.25931928e-01 -9.30144727e-01
-2.31622681e-01 -6.09565318e-01 -1.57122386e+00 -4.94053066e-02
2.72825453e-02 -3.75632793e-01 5.62750936e-01 1.04719031e+00
2.29326531e-01 4.84839261e-01 -1.29404739e-01 -6.88491464e-01
-4.43910182e-01 -6.46978557e-01 -3.34088415e-01 6.14649802e-02
1.64394438e-01 -5.10834813e-01 2.26015061e-01 2.03394696e-01] | [10.273396492004395, 1.5546144247055054] |
8cae14a9-90c0-4e73-bce3-991cf69dedb4 | dp-fp-differentially-private-forward | 2112.14430 | null | https://arxiv.org/abs/2112.14430v1 | https://arxiv.org/pdf/2112.14430v1.pdf | DP-FP: Differentially Private Forward Propagation for Large Models | When applied to large-scale learning problems, the conventional wisdom on privacy-preserving deep learning, known as Differential Private Stochastic Gradient Descent (DP-SGD), has met with limited success due to significant performance degradation and high memory overhead when compared to the non-privacy counterpart. We show how to mitigate the performance drop by replacing the DP-SGD with a novel DP Forward-Propagation (DP-FP) followed by an off-the-shelf non-DP optimizer. Our DP-FP employs novel (1) representation clipping followed by noise addition in the forward propagation stage, as well as (2) micro-batch construction via subsampling to achieve DP amplification and reduce noise power to $1/M$, where $M$ is the number of micro-batch in a step. When training a classification model, our DP-FP with all of the privacy-preserving operations on the representation is innately free of gradient bias, total noise proportionally to model size, and memory issues in DP-SGD. As a result, our DP-FP outperforms cutting-edge DP-SGD while retaining the same level of privacy, and it approaches non-private baselines and significantly outperforms state-of-the-art DP-SGD variants. When applied to RoBERTa-large on four downstream tasks, for example, DP-FP achieves an average accuracy of 91.34\% with privacy budgets less than 3, representing a 3.81\% performance improvement over the state-of-the-art DP-SGD and only a 0.9\% loss compared to the non-private baseline but with a significantly lower privacy leakage risk. | ['Haitao Mi', 'Jian Du'] | 2021-12-29 | null | null | null | null | ['privacy-preserving-deep-learning', 'privacy-preserving-deep-learning'] | ['methodology', 'natural-language-processing'] | [ 5.83560951e-02 2.49929994e-01 1.32242795e-02 -6.36978805e-01
-1.40563297e+00 -5.98402023e-01 3.65418553e-01 2.39791244e-01
-8.83234799e-01 7.73614526e-01 1.72005087e-01 -7.15454400e-01
4.89645422e-01 -6.44225240e-01 -1.03650177e+00 -7.31834590e-01
-5.79037368e-02 -1.55161902e-01 -7.07438514e-02 2.78553158e-01
-2.07620278e-01 3.12564164e-01 -1.00435543e+00 2.90357947e-01
4.79944021e-01 1.23881328e+00 -4.01969552e-01 4.12042499e-01
1.51993766e-01 5.52781224e-01 -3.28963369e-01 -9.75784779e-01
9.93845522e-01 -1.18084490e-01 -5.31421185e-01 -5.57847798e-01
7.24581122e-01 -9.40838456e-01 -5.86004078e-01 1.00864506e+00
7.80548275e-01 1.44307613e-01 1.66905999e-01 -1.14865565e+00
-5.17936885e-01 4.33259130e-01 -7.88665891e-01 -1.96744993e-01
-3.79812643e-02 2.05157384e-01 1.02628839e+00 -6.46408379e-01
5.57956040e-01 9.60308194e-01 1.08683240e+00 7.81270802e-01
-1.67051351e+00 -9.43472028e-01 8.22186917e-02 -4.70709056e-01
-1.39442694e+00 -7.08406627e-01 3.44387531e-01 -2.17473015e-01
1.14137173e+00 4.40734416e-01 2.15994373e-01 9.45271015e-01
6.69880062e-02 1.00391340e+00 1.12184954e+00 -1.35700360e-01
5.73385775e-01 4.36007231e-01 1.81530103e-01 7.48934627e-01
3.40656281e-01 5.56052141e-02 -5.46463013e-01 -8.30596924e-01
2.54429489e-01 1.58615232e-01 -3.06287974e-01 -5.56903183e-01
-4.47470576e-01 8.24861407e-01 3.42443377e-01 -1.84359953e-01
-8.62153247e-02 3.90169740e-01 7.62146354e-01 4.32172537e-01
7.83065915e-01 2.79785007e-01 -7.99268484e-01 -1.21232927e-01
-1.19987488e+00 7.66490281e-01 1.10227621e+00 9.03833389e-01
7.40546584e-01 -1.31202534e-01 -4.28404719e-01 5.18631339e-01
4.11177017e-02 1.83167547e-01 4.14485127e-01 -9.60750341e-01
8.49991381e-01 2.91556686e-01 8.19438100e-02 -8.04693222e-01
-4.38528582e-02 -6.37700319e-01 -7.46295273e-01 3.16934258e-01
5.40588558e-01 -6.75320089e-01 -6.64684594e-01 2.15813231e+00
3.51669490e-01 -3.22237164e-02 1.73643321e-01 5.43024480e-01
2.88601965e-01 5.51251531e-01 1.11045659e-01 -6.54206797e-02
1.22041118e+00 -9.22762811e-01 -2.56375760e-01 -3.66099268e-01
1.18203318e+00 -1.75440401e-01 1.16715038e+00 8.78837034e-02
-1.07365656e+00 1.28984973e-01 -1.06427491e+00 -5.56087792e-01
-3.19725901e-01 5.03887683e-02 6.58578873e-01 1.05819547e+00
-1.23581529e+00 7.93995202e-01 -8.78363907e-01 3.87897119e-02
1.09152579e+00 5.01533389e-01 -8.08452487e-01 -1.47046432e-01
-9.42692459e-01 3.59316915e-01 -1.10962830e-01 -1.91974416e-01
-8.71393085e-01 -1.35459661e+00 -8.22882771e-01 2.92158574e-01
2.89106965e-01 -9.04307008e-01 1.21661627e+00 -4.79087561e-01
-1.55446362e+00 1.16266942e+00 -3.10997635e-01 -1.12891865e+00
1.25892746e+00 -4.55969900e-01 1.18920021e-01 -2.14260101e-01
-6.17481135e-02 3.88027698e-01 8.19974363e-01 -8.66176665e-01
-7.47503936e-01 -7.91017771e-01 -1.28146773e-02 2.74956375e-01
-4.97141987e-01 -5.00156693e-02 -4.06249195e-01 -5.47256947e-01
-2.04485103e-01 -9.86537993e-01 -4.52909350e-01 5.41298151e-01
-4.31204855e-01 3.10667604e-01 9.65903521e-01 -8.07562590e-01
1.24657202e+00 -2.59881544e+00 -4.50064898e-01 1.31679446e-01
6.16720855e-01 5.51643252e-01 1.70218185e-01 1.43264383e-01
2.46830210e-01 2.67646641e-01 -6.96122587e-01 -1.19984627e+00
2.97741622e-01 -5.26103638e-02 -3.60296398e-01 8.19272041e-01
-2.94214755e-01 7.68176198e-01 -7.41668999e-01 3.53902802e-02
-1.99341893e-01 4.63645846e-01 -8.53478670e-01 2.52414960e-02
-4.99659888e-02 6.13152795e-02 -4.02651668e-01 3.38385165e-01
1.16679633e+00 -1.22363798e-01 8.37515146e-02 2.46632248e-01
9.10923332e-02 3.32754135e-01 -1.00239217e+00 1.69369650e+00
-3.98172319e-01 4.89876062e-01 6.60271704e-01 -6.66823983e-01
6.40983820e-01 1.15037918e-01 1.39783189e-01 -3.22799623e-01
-8.26911777e-02 3.63620132e-01 -5.20138621e-01 -4.81049940e-02
3.61046195e-01 -5.32745868e-02 -1.02275256e-02 6.35672092e-01
-1.39384404e-01 1.85290545e-01 -5.67280948e-01 1.87694535e-01
1.42610455e+00 -2.04769835e-01 8.97600278e-02 -3.67690444e-01
3.86316270e-01 -4.74316329e-01 8.75264227e-01 1.03988755e+00
-4.74825114e-01 5.67930698e-01 5.58911741e-01 -4.10893887e-01
-7.77121127e-01 -8.04702938e-01 -1.28395677e-01 1.15608776e+00
-4.25853223e-01 -5.85907578e-01 -1.05800807e+00 -9.90444064e-01
6.05581105e-01 8.65732372e-01 -6.30430877e-01 -2.71068990e-01
-3.53178531e-01 -1.01585531e+00 9.64941919e-01 2.52497673e-01
9.70017195e-01 -2.81572670e-01 -4.90084112e-01 3.32536362e-02
3.85138482e-01 -7.83566535e-01 -8.02453995e-01 3.03118497e-01
-8.73998463e-01 -4.72406983e-01 -7.11519241e-01 -4.10325706e-01
6.72023714e-01 1.91533968e-01 7.48773575e-01 -4.09917772e-01
-1.07808419e-01 3.21693830e-02 1.87119678e-01 -4.20171678e-01
-6.51771426e-02 2.23410904e-01 -1.14831351e-01 2.09796458e-01
6.19835854e-01 -7.08395302e-01 -9.36806619e-01 -1.84319615e-01
-6.86067283e-01 -2.62487948e-01 4.36702728e-01 8.60089839e-01
7.78849602e-01 -1.86871558e-01 1.47998288e-01 -1.35332465e+00
7.66762972e-01 -4.78792876e-01 -8.44680607e-01 6.58037374e-03
-1.10641229e+00 2.39333715e-02 8.28386724e-01 -1.63591906e-01
-1.13033807e+00 2.10249215e-01 -1.34573326e-01 -5.83565593e-01
3.15581530e-01 1.43822908e-01 -3.62387180e-01 -3.65500003e-01
7.16480315e-01 2.80256420e-01 2.66516358e-01 -7.83104241e-01
5.16767859e-01 6.62742913e-01 3.09041291e-01 -3.93062323e-01
5.68374932e-01 5.50408363e-01 1.59510434e-01 -6.03133261e-01
-6.37109816e-01 -1.21378861e-01 -9.91968736e-02 7.77800620e-01
4.55138803e-01 -1.37731838e+00 -6.82484925e-01 7.89694905e-01
-7.09785283e-01 -3.91987354e-01 -7.14404941e-01 3.05455953e-01
-3.75908941e-01 2.45461270e-01 -8.41946125e-01 -8.11104119e-01
-9.55352128e-01 -1.01216447e+00 8.70457888e-01 -8.62766802e-02
-6.23787940e-02 -7.23445952e-01 -1.35342538e-01 4.40793127e-01
5.83940327e-01 3.11300904e-01 6.70693636e-01 -1.02266550e+00
-4.82514501e-01 -6.23893619e-01 -2.39828452e-01 7.98818707e-01
-3.10274214e-01 -6.52445972e-01 -1.41192126e+00 -7.55294323e-01
4.72331762e-01 -2.52878904e-01 8.77919972e-01 3.52506220e-01
1.47756243e+00 -9.41172183e-01 -3.04929048e-01 1.40927684e+00
1.50434482e+00 -7.40553662e-02 3.39912295e-01 1.75285131e-01
6.12751484e-01 2.59193510e-01 1.44669622e-01 7.13796735e-01
3.29862416e-01 3.61644745e-01 1.80786163e-01 -1.36181504e-01
2.72911459e-01 -6.47376657e-01 3.88364822e-01 1.15096450e-01
6.39148116e-01 5.23168445e-02 -6.20750725e-01 4.13734943e-01
-1.84981453e+00 -6.18163586e-01 1.41631901e-01 2.63922739e+00
1.04157019e+00 4.25064266e-02 3.33667547e-02 -3.07906240e-01
1.95748627e-01 5.27173579e-01 -9.66323972e-01 -6.38811111e-01
-1.82965547e-01 1.37081116e-01 1.24707699e+00 5.64706445e-01
-1.19058394e+00 8.62986028e-01 6.01119041e+00 9.23925877e-01
-9.41709995e-01 3.75970900e-01 1.16238356e+00 -7.61429727e-01
-2.76794523e-01 5.34719639e-02 -1.05600548e+00 4.86446738e-01
1.00335944e+00 -4.89112526e-01 5.76540053e-01 1.33809960e+00
-1.88071921e-01 9.19772312e-02 -1.31735718e+00 1.16547251e+00
-3.04810703e-01 -1.38565958e+00 -2.45912462e-01 4.94142324e-01
6.27445698e-01 3.42190832e-01 3.21612000e-01 2.74980038e-01
6.29468083e-01 -8.49481285e-01 4.88523722e-01 3.64648029e-02
8.59815717e-01 -8.27675760e-01 6.40848160e-01 4.55627292e-01
-6.75746620e-01 -2.31301904e-01 -4.72865820e-01 -7.42153153e-02
-1.32535890e-01 9.00792181e-01 -6.44503891e-01 2.78379172e-01
9.60249782e-01 3.17607343e-01 -3.53660166e-01 5.35664797e-01
-8.99760351e-02 8.31566751e-01 -7.88847625e-01 9.50811580e-02
2.55622089e-01 -1.97492808e-01 6.31525159e-01 1.32500005e+00
1.40970394e-01 3.21220048e-02 -3.58616233e-01 8.66033733e-01
-8.29404235e-01 1.41811535e-01 -6.32700562e-01 1.36568829e-01
7.03069866e-01 8.92005622e-01 2.03783616e-01 -2.88430542e-01
-5.10224938e-01 1.21886313e+00 6.31453335e-01 2.74076283e-01
-5.25516987e-01 -5.42007506e-01 1.37084472e+00 2.57624358e-01
3.63013864e-01 1.02410302e-01 -4.66444880e-01 -1.19508910e+00
5.13985872e-01 -8.06651831e-01 4.88995314e-01 2.65159877e-03
-1.34901845e+00 2.65133083e-01 -3.57550055e-01 -6.84568703e-01
-1.32473400e-02 -1.47343799e-01 -3.57070595e-01 1.29286540e+00
-1.35168171e+00 -9.76533771e-01 2.75465429e-01 6.76775336e-01
-1.99729875e-01 -1.00673631e-01 1.08222902e+00 3.31045628e-01
-6.78641260e-01 1.68644810e+00 8.29056859e-01 2.35167757e-01
6.26349330e-01 -1.18491852e+00 8.38609934e-01 1.06270397e+00
-2.67526984e-01 9.22637939e-01 4.74639654e-01 -3.75886351e-01
-1.60591471e+00 -1.34324956e+00 1.01032281e+00 -3.37556869e-01
3.26952994e-01 -9.73175168e-01 -9.24083054e-01 9.94715095e-01
-3.20971131e-01 5.59916794e-01 9.46310043e-01 -4.46432177e-03
-6.76118612e-01 -6.11467540e-01 -1.98953640e+00 7.25427508e-01
1.15154862e+00 -8.71786475e-01 -1.80751279e-01 3.87355357e-01
1.07163036e+00 -5.49803317e-01 -8.45745146e-01 1.78540200e-02
6.59913719e-01 -1.04109216e+00 8.81244957e-01 -6.00531518e-01
9.59603116e-02 -7.28110969e-03 -4.16189879e-01 -8.39237630e-01
-1.52440101e-01 -1.36734283e+00 -2.54099309e-01 1.32193351e+00
5.63495994e-01 -1.18662632e+00 1.33906102e+00 1.63248253e+00
2.68773943e-01 -1.00027454e+00 -1.17416453e+00 -6.81639373e-01
3.54332775e-01 -2.19181806e-01 7.87717223e-01 8.90283644e-01
-3.41566652e-01 -1.35720819e-01 -5.12902856e-01 2.42685050e-01
9.35637057e-01 -2.25961208e-01 1.10528600e+00 -5.32152236e-01
-7.40210593e-01 -1.84337959e-01 -1.99960411e-01 -1.12414122e+00
1.26721412e-01 -9.21484172e-01 -3.32034290e-01 -9.00456250e-01
1.82376914e-02 -4.16448802e-01 -3.87886405e-01 7.94867456e-01
-2.03699082e-01 -7.41502270e-02 3.30541372e-01 1.42361790e-01
-1.81503177e-01 6.00689411e-01 6.31012619e-01 7.56240264e-02
-3.99857253e-01 1.91851199e-01 -1.13768744e+00 4.67282236e-01
6.57717228e-01 -6.77255869e-01 -4.50069577e-01 -5.09713292e-01
-6.13159835e-02 -1.66919529e-01 1.70043260e-01 -8.67678046e-01
2.60244250e-01 3.60496193e-01 2.44652867e-01 -1.44582674e-01
3.31879973e-01 -8.31523955e-01 2.53925681e-01 4.88651514e-01
-5.31198919e-01 -1.79505005e-01 3.24432552e-01 7.76967287e-01
3.84454168e-02 1.57935079e-02 8.38407636e-01 -1.70196667e-01
-8.37382600e-02 5.87427676e-01 -2.54701450e-02 2.83043087e-01
1.05838609e+00 -3.75315100e-02 -3.43874037e-01 -2.73378968e-01
-3.29478383e-01 2.66059816e-01 7.03309834e-01 -1.06086865e-01
1.91704601e-01 -7.75269210e-01 -5.97095013e-01 4.90422159e-01
-1.30167395e-01 2.57332891e-01 3.66099268e-01 4.90810156e-01
-5.16637266e-01 2.12595731e-01 3.17841560e-01 -1.29762024e-01
-1.33020985e+00 4.37965453e-01 3.58529985e-01 -5.75382948e-01
-9.39871550e-01 1.55823791e+00 3.56984437e-01 -5.32507300e-01
5.72595835e-01 -1.69408098e-01 9.28212047e-01 -3.19043875e-01
6.27114236e-01 4.30860013e-01 2.07832605e-01 -1.85673192e-01
-3.88752788e-01 2.75811162e-02 -6.00489259e-01 -1.96698740e-01
1.42674994e+00 1.20796887e-02 -3.31499539e-02 6.64839055e-03
1.94893122e+00 1.47469193e-01 -1.82115841e+00 -4.32525992e-01
-4.19753581e-01 -8.04760039e-01 2.54998744e-01 -8.39661658e-01
-1.27689636e+00 8.44291568e-01 7.24702120e-01 -4.26331013e-01
1.09746492e+00 -4.51299995e-01 1.13277531e+00 4.22213823e-01
5.31781912e-01 -9.14340854e-01 -8.82252932e-01 1.48839116e-01
5.29656172e-01 -1.10535812e+00 3.64845604e-01 4.17362992e-03
-5.69135487e-01 4.11022693e-01 3.23981792e-01 -1.04061574e-01
1.03193665e+00 4.13132250e-01 -1.01639487e-01 2.45232746e-01
-7.45025158e-01 7.16983855e-01 -2.69860893e-01 3.41136783e-01
-1.80421602e-02 5.74700460e-02 -2.62801766e-01 8.75074744e-01
-2.78276920e-01 1.94177419e-01 -8.93978029e-02 1.28657734e+00
2.13596597e-01 -1.02806902e+00 1.46779537e-01 7.96399474e-01
-1.03011143e+00 -2.26549283e-01 -2.49674097e-01 5.19935787e-01
-1.52017206e-01 5.58420122e-01 -1.17129788e-01 -4.38645989e-01
2.95830280e-01 1.36613309e-01 -4.54728864e-02 -3.01778376e-01
-1.17590201e+00 -4.41953987e-01 1.04217462e-01 -1.01172531e+00
3.79818022e-01 -9.19862986e-01 -1.00972497e+00 -8.10658157e-01
-4.70095314e-02 -4.45378684e-02 7.34530389e-01 3.16405624e-01
1.21656394e+00 -2.45683149e-01 6.57100320e-01 -3.90364885e-01
-1.25983500e+00 -4.67609078e-01 -8.11549902e-01 4.02363271e-01
6.23062611e-01 5.02916574e-02 -7.77216136e-01 -5.15668273e-01] | [5.903469562530518, 6.848443031311035] |
475708fa-97b4-44d3-b011-17e5d2e1d44f | word-translation-without-parallel-data | 1710.04087 | null | http://arxiv.org/abs/1710.04087v3 | http://arxiv.org/pdf/1710.04087v3.pdf | Word Translation Without Parallel Data | State-of-the-art methods for learning cross-lingual word embeddings have
relied on bilingual dictionaries or parallel corpora. Recent studies showed
that the need for parallel data supervision can be alleviated with
character-level information. While these methods showed encouraging results,
they are not on par with their supervised counterparts and are limited to pairs
of languages sharing a common alphabet. In this work, we show that we can build
a bilingual dictionary between two languages without using any parallel
corpora, by aligning monolingual word embedding spaces in an unsupervised way.
Without using any character information, our model even outperforms existing
supervised methods on cross-lingual tasks for some language pairs. Our
experiments demonstrate that our method works very well also for distant
language pairs, like English-Russian or English-Chinese. We finally describe
experiments on the English-Esperanto low-resource language pair, on which there
only exists a limited amount of parallel data, to show the potential impact of
our method in fully unsupervised machine translation. Our code, embeddings and
dictionaries are publicly available. | ['Hervé Jégou', "Marc'Aurelio Ranzato", 'Ludovic Denoyer', 'Guillaume Lample', 'Alexis Conneau'] | 2017-10-11 | word-translation-without-parallel-data-1 | https://openreview.net/forum?id=H196sainb | https://openreview.net/pdf?id=H196sainb | iclr-2018-1 | ['unsupervised-machine-translation'] | ['natural-language-processing'] | [-2.65309304e-01 -2.42003292e-01 -7.51556039e-01 -3.72780353e-01
-8.26963246e-01 -8.56780708e-01 9.16732490e-01 2.40625307e-01
-8.82707059e-01 8.05095732e-01 3.52168351e-01 -7.48626649e-01
3.03642780e-01 -5.96774697e-01 -5.93760610e-01 -3.73977065e-01
1.74183145e-01 7.95530736e-01 -4.38353838e-03 -6.83358431e-01
-6.70445524e-03 3.49920034e-01 -9.51014459e-01 5.79516962e-02
8.44564974e-01 1.45974504e-02 2.45598048e-01 1.68173581e-01
-4.00283247e-01 1.38374135e-01 -8.27725790e-03 -7.25939035e-01
5.13582408e-01 -2.92374313e-01 -8.76128793e-01 -2.50844896e-01
5.20556271e-01 -9.51477662e-02 -3.06362659e-01 1.06381583e+00
5.71225345e-01 -3.05985481e-01 5.42767346e-01 -8.87615740e-01
-1.07067633e+00 8.57776761e-01 -3.18336636e-01 1.44583747e-01
3.71629655e-01 -1.11000523e-01 1.40857160e+00 -1.27308011e+00
9.32744205e-01 9.50634122e-01 8.55619073e-01 3.35212916e-01
-1.36037874e+00 -6.40366435e-01 1.02635901e-02 1.30969375e-01
-1.50809026e+00 -4.33907866e-01 6.40961289e-01 -4.72264200e-01
1.41021585e+00 -1.27387330e-01 4.87692922e-01 1.19176090e+00
6.89555109e-02 4.97717828e-01 1.39174557e+00 -1.05473602e+00
-3.09860379e-01 6.13146722e-01 1.88339487e-01 7.57586420e-01
2.40380794e-01 2.22843915e-01 -5.05977929e-01 -1.78239733e-01
4.89457041e-01 -2.13295728e-01 -1.12103745e-01 -5.49295187e-01
-1.59979272e+00 1.05069602e+00 3.45899910e-02 1.01336396e+00
2.67240312e-02 -4.15159523e-01 6.28649116e-01 7.21297085e-01
6.92529678e-01 4.95677829e-01 -6.57390773e-01 -1.04546182e-01
-6.94412112e-01 -2.71284312e-01 9.41604316e-01 1.07125771e+00
9.85318482e-01 -1.05988294e-01 3.77035767e-01 1.14277434e+00
2.87665147e-02 5.61984122e-01 8.07580411e-01 -2.93698490e-01
6.37665927e-01 2.79736668e-01 -5.67717403e-02 -7.34481812e-01
-2.43593663e-01 -5.05671985e-02 -5.02834737e-01 -1.91837922e-01
4.56853122e-01 -1.74827829e-01 -4.71650988e-01 1.84011543e+00
6.07932396e-02 -2.84303218e-01 4.74549830e-01 7.30953634e-01
3.62597764e-01 5.54977000e-01 -2.52957880e-01 -1.06903948e-01
1.42237186e+00 -1.21616566e+00 -7.38787234e-01 -3.85357678e-01
9.77624953e-01 -1.23256874e+00 1.27204537e+00 1.00690924e-01
-7.59848595e-01 -6.08010888e-01 -1.15092933e+00 -3.71880919e-01
-7.78698504e-01 3.23228925e-01 7.17203557e-01 7.37184107e-01
-9.99779761e-01 5.92268586e-01 -7.95173764e-01 -9.38356698e-01
-2.32499808e-01 3.47256064e-01 -8.62936139e-01 -2.63906926e-01
-1.45601356e+00 1.39239407e+00 4.08926636e-01 -2.87440389e-01
-4.27846462e-01 -4.16676462e-01 -1.09775150e+00 -4.13031965e-01
-1.80085197e-01 -2.79712498e-01 7.52951980e-01 -1.07931566e+00
-1.40936136e+00 1.32421458e+00 -1.19657837e-01 -3.52078408e-01
3.26580465e-01 -2.43696019e-01 -6.77181780e-01 -2.26225525e-01
3.35057348e-01 6.50432050e-01 3.22843999e-01 -9.51046824e-01
-4.61152345e-01 -2.82636255e-01 1.53849190e-02 2.39130720e-01
-8.85926127e-01 4.10486013e-01 -5.87907672e-01 -7.23328769e-01
-2.33850852e-01 -1.13357449e+00 -2.15156287e-01 -2.58998394e-01
-1.05339810e-02 -3.95737857e-01 3.17344248e-01 -7.91892946e-01
1.10114479e+00 -2.11029840e+00 3.08310777e-01 -7.95502439e-02
-3.45043927e-01 3.84488672e-01 -4.61068124e-01 9.37657893e-01
-1.91601694e-01 1.37608454e-01 -2.65217304e-01 -3.96645367e-01
7.62843341e-02 6.48153841e-01 -2.70279139e-01 7.41795778e-01
3.02387357e-01 7.03221858e-01 -1.04751241e+00 -4.52219516e-01
8.24357942e-02 4.40836757e-01 -5.22933185e-01 1.58385679e-01
2.28249669e-01 2.90840000e-01 8.41562301e-02 4.30130601e-01
4.68629867e-01 2.93805689e-01 7.66025007e-01 -1.31650403e-01
-5.35190940e-01 8.32653284e-01 -9.19995129e-01 2.19322014e+00
-9.47182536e-01 7.14449227e-01 -1.72301754e-01 -1.19277608e+00
1.05031538e+00 4.00505215e-01 2.96414018e-01 -6.69938862e-01
-7.01599121e-02 8.59762609e-01 2.22046182e-01 -3.11246127e-01
5.93607903e-01 -3.71790767e-01 -1.97493404e-01 6.25519633e-01
5.24103343e-01 -1.95745707e-01 4.46521997e-01 -1.28559396e-01
6.76133633e-01 2.59108216e-01 5.24608076e-01 -7.38837004e-01
3.94470632e-01 2.75383830e-01 5.27358055e-01 2.89889723e-01
2.98956502e-02 4.34460342e-01 1.86107174e-01 -5.43849051e-01
-1.38373220e+00 -9.93808329e-01 -6.43710434e-01 1.05659997e+00
6.17720261e-02 -7.23783970e-01 -4.27585483e-01 -7.88047850e-01
-4.69625853e-02 4.93649185e-01 -4.34566438e-01 2.46438920e-01
-8.67440283e-01 -9.08186316e-01 8.12889159e-01 4.08559561e-01
-9.05546397e-02 -7.57655203e-01 1.93103403e-01 2.89605856e-01
-6.89877644e-02 -1.36569941e+00 -7.05894768e-01 5.40195882e-01
-8.02773476e-01 -8.83354664e-01 -7.32903481e-01 -1.44505012e+00
6.84506416e-01 9.25729945e-02 1.28267193e+00 -2.46797726e-01
4.82393652e-02 1.36555821e-01 -3.60319108e-01 -4.72892337e-02
-5.39741635e-01 5.04471183e-01 7.84372211e-01 -2.43812576e-01
9.33331788e-01 -6.53489709e-01 6.65647388e-02 3.28928202e-01
-5.84268928e-01 -1.12965487e-01 4.75619018e-01 1.18172467e+00
5.48625469e-01 -5.82448423e-01 3.48995745e-01 -1.07284844e+00
5.15142620e-01 -4.41363394e-01 -5.83534479e-01 2.74260551e-01
-7.29508877e-01 4.38533664e-01 9.81009364e-01 -5.71986079e-01
-5.74204147e-01 -2.82770228e-02 -3.97920758e-01 -7.13899881e-02
-1.21126086e-01 4.18695927e-01 -7.23671243e-02 2.54186871e-03
7.15316117e-01 2.11262077e-01 -1.92145497e-01 -8.20551634e-01
6.18765593e-01 9.74156916e-01 2.42656603e-01 -7.65773892e-01
8.49639833e-01 2.44563714e-01 -4.86637592e-01 -7.57636905e-01
-3.93786848e-01 -6.00824952e-01 -1.18940651e+00 5.76563835e-01
8.44607472e-01 -1.25062108e+00 1.67343244e-01 -2.71115336e-04
-1.38937724e+00 -1.71071738e-01 -1.71520308e-01 9.85196710e-01
-2.72481352e-01 4.16547328e-01 -7.29994416e-01 -9.61325467e-02
-4.29149084e-02 -1.14910388e+00 8.49113226e-01 -3.88887554e-01
-2.55305290e-01 -1.53770638e+00 7.13399470e-01 1.90454215e-01
2.71608561e-01 -3.67249310e-01 1.05803525e+00 -1.08508599e+00
-1.27579104e-02 -4.74884808e-02 1.24451900e-02 6.28760099e-01
5.58994234e-01 -2.46506006e-01 -6.16991580e-01 -5.16655803e-01
-2.02505857e-01 -3.95431340e-01 5.52588999e-01 -3.42669845e-01
3.03778946e-01 -7.26171434e-02 -2.55974919e-01 7.27940977e-01
1.61134386e+00 -1.49474934e-01 2.61094689e-01 3.89546663e-01
8.23901236e-01 5.58385670e-01 4.31480885e-01 -7.52016827e-02
6.08264387e-01 8.84096205e-01 -2.30547324e-01 -3.62614900e-01
-2.91463614e-01 -3.41886282e-01 8.28459680e-01 2.01890635e+00
2.78537888e-02 1.40646189e-01 -1.10647202e+00 1.24634075e+00
-1.50713217e+00 -5.16338229e-01 -2.30796754e-01 2.11603618e+00
1.32297683e+00 -6.66280761e-02 -3.47888172e-02 -2.03894958e-01
4.92042392e-01 1.00759111e-01 3.73612158e-03 -8.79771709e-01
-4.35451001e-01 5.74793518e-01 7.78374314e-01 9.29259121e-01
-1.03474975e+00 1.36434078e+00 6.59125900e+00 6.94301844e-01
-1.15001619e+00 6.60953462e-01 -1.42300818e-02 2.31281996e-01
-5.83037734e-01 3.39291334e-01 -1.01680064e+00 1.90518677e-01
9.40241396e-01 -7.52730519e-02 4.90244210e-01 6.85026646e-01
-2.63381034e-01 2.81120777e-01 -1.44058526e+00 9.59442258e-01
3.03380728e-01 -1.16555655e+00 -1.00125864e-01 5.90692610e-02
9.10983920e-01 7.64995515e-01 -2.28617907e-01 3.46929312e-01
5.59939742e-01 -8.99565518e-01 3.13548654e-01 -1.57751366e-01
1.03231251e+00 -6.95483983e-01 8.13321114e-01 1.55959517e-01
-1.04166365e+00 5.15532374e-01 -7.34438181e-01 -8.57671201e-02
1.36464983e-01 4.66542393e-01 -6.09780312e-01 7.73617625e-01
3.80641282e-01 9.26611543e-01 -5.66739917e-01 3.73875767e-01
-4.36657846e-01 4.97914195e-01 -3.11587393e-01 2.25096364e-02
3.92821968e-01 -5.95363855e-01 3.02777439e-01 1.60304701e+00
4.05497253e-01 -3.77584875e-01 3.19428056e-01 3.80732596e-01
-1.01556614e-01 8.14631581e-01 -1.16885090e+00 -3.16164494e-01
1.46658808e-01 1.01569402e+00 -3.53445083e-01 -2.23750770e-01
-1.02732229e+00 1.35018384e+00 7.63003647e-01 4.35956009e-02
-5.84554076e-01 -5.50737858e-01 8.74922931e-01 -1.52083598e-02
2.69453198e-01 -6.56598628e-01 -1.31915718e-01 -1.70422196e+00
1.29509196e-01 -1.15459454e+00 7.53229633e-02 -1.88166723e-01
-1.66405225e+00 9.27666426e-01 -1.85756788e-01 -1.26532137e+00
-3.52055639e-01 -1.09458423e+00 -2.69397467e-01 1.10351002e+00
-1.65131867e+00 -1.28281605e+00 5.72613478e-01 7.02390313e-01
4.25723255e-01 -6.78658664e-01 1.39932740e+00 7.39745975e-01
-3.85931283e-01 7.10600674e-01 5.47450244e-01 4.18040752e-01
1.28088403e+00 -1.22981501e+00 6.49128854e-01 9.06093001e-01
8.94600332e-01 8.61038268e-01 5.07316709e-01 -3.70939374e-01
-1.50714636e+00 -6.99294448e-01 1.79710972e+00 -5.16226590e-01
1.30694556e+00 -8.97714078e-01 -7.78474450e-01 8.52398038e-01
8.57470989e-01 1.70334190e-01 1.04913044e+00 7.68858016e-01
-6.84734285e-01 -7.40997586e-03 -6.10813797e-01 7.94909120e-01
1.16399682e+00 -1.00945079e+00 -9.06984568e-01 7.59269059e-01
6.09089136e-01 -7.97679648e-02 -1.03122556e+00 1.34359837e-01
4.60997403e-01 -6.69654250e-01 7.31910229e-01 -8.00304532e-01
3.14838588e-01 -3.20948303e-01 -4.40884531e-01 -1.68162632e+00
-1.45967856e-01 -5.09271026e-01 5.03844440e-01 1.34006226e+00
7.16849923e-01 -7.98258424e-01 1.45180359e-01 -1.14967205e-01
-8.81441608e-02 -4.10956264e-01 -9.59634304e-01 -1.20669115e+00
7.64580131e-01 -3.70461941e-01 3.49818856e-01 1.51322019e+00
3.74471933e-01 7.04611599e-01 -4.98009771e-01 8.13859925e-02
2.36398876e-01 2.25689724e-01 7.67227411e-01 -9.17957962e-01
-4.98000622e-01 -2.53917515e-01 -4.01089817e-01 -1.09740472e+00
8.27791989e-01 -1.52504647e+00 -9.80045274e-02 -1.17609024e+00
1.15779735e-01 -7.12131321e-01 -2.97654182e-01 5.68605423e-01
-6.79253638e-02 5.34919679e-01 -4.60186489e-02 2.31203794e-01
-1.14158899e-01 4.55847353e-01 7.73796082e-01 -2.84959733e-01
7.21582770e-02 -6.80177927e-01 -4.87486571e-01 5.57180524e-01
7.46247053e-01 -8.13089311e-01 -6.79401979e-02 -1.02828610e+00
1.60796046e-01 -4.56669569e-01 -3.35017025e-01 -6.17808998e-01
8.42209384e-02 -3.64811271e-02 -1.40552700e-01 1.76785234e-02
1.10424072e-01 -8.25994074e-01 -7.70747736e-02 3.57728034e-01
-4.23711911e-02 6.49723947e-01 2.88601577e-01 7.03964755e-02
-5.59844434e-01 -3.57009858e-01 5.92991054e-01 1.12094410e-01
-5.82015216e-01 1.21656701e-01 -4.26856011e-01 2.05331966e-01
7.53824413e-01 2.48439580e-01 5.42470776e-02 1.37557596e-01
-4.07573044e-01 -5.69594465e-02 8.79411101e-01 8.35987270e-01
-8.08908418e-03 -1.76414299e+00 -9.30439293e-01 5.62768102e-01
5.67751110e-01 -7.75279880e-01 -5.23931324e-01 8.34449768e-01
-6.24047637e-01 6.16768479e-01 -3.97598922e-01 -3.80867213e-01
-1.03097332e+00 7.06469834e-01 2.45181564e-02 -5.29805541e-01
-4.02740538e-01 4.84875977e-01 -1.05257496e-01 -1.14057946e+00
-8.27651843e-02 -1.67799234e-01 1.35629728e-01 1.71763390e-01
1.68317780e-01 -1.72496900e-01 2.69250244e-01 -1.02691352e+00
-6.61103427e-01 8.03256810e-01 -2.12538347e-01 -3.72266531e-01
1.39569151e+00 -7.86626115e-02 -3.90602708e-01 7.47947931e-01
1.52683604e+00 7.24454582e-01 -4.91928041e-01 -4.35802072e-01
1.54007703e-01 -3.32584023e-01 -2.93306440e-01 -3.69032413e-01
-8.19147050e-01 1.07567847e+00 4.90165502e-01 -1.35506615e-01
7.54484415e-01 -3.01429536e-02 8.43748927e-01 4.88656431e-01
5.80747426e-01 -1.19580281e+00 -4.56899375e-01 9.71716344e-01
4.97843415e-01 -1.54958153e+00 -7.46338591e-02 -1.26528904e-01
-5.26188195e-01 1.16923594e+00 1.93846017e-01 -3.29266280e-01
6.84402823e-01 3.82270992e-01 4.73677814e-01 2.35506892e-01
-4.89656955e-01 -4.08493757e-01 2.42095843e-01 6.70489073e-01
9.68722284e-01 2.10142329e-01 -9.32477057e-01 4.11955953e-01
-3.67234111e-01 -3.38769436e-01 3.46816033e-01 7.52162755e-01
1.12069890e-01 -2.21962523e+00 -1.50612429e-01 -7.82155469e-02
-6.20943010e-01 -7.62820959e-01 -3.80923837e-01 1.07688093e+00
2.23141134e-01 7.11039305e-01 1.55846372e-01 -2.83004165e-01
1.93289459e-01 3.68220419e-01 7.15381145e-01 -8.19525003e-01
-6.42487168e-01 -1.45970836e-01 3.63583952e-01 -2.67670333e-01
-5.21456540e-01 -6.98870718e-01 -8.18817973e-01 -3.34522277e-01
-2.37220094e-01 3.52699339e-01 7.02153563e-01 9.15920794e-01
1.74842626e-01 -3.43451090e-02 5.48921168e-01 -6.31619155e-01
-4.01995301e-01 -9.71676528e-01 -4.53781426e-01 4.40829188e-01
1.26705915e-01 -3.67659241e-01 -1.89247385e-01 8.56618062e-02] | [11.018477439880371, 10.027776718139648] |
36b3f43c-f3e7-4a37-bc21-b796ed57ef4c | search-and-learn-improving-semantic-coverage-1 | 2112.02770 | null | https://arxiv.org/abs/2112.02770v1 | https://arxiv.org/pdf/2112.02770v1.pdf | Search and Learn: Improving Semantic Coverage for Data-to-Text Generation | Data-to-text generation systems aim to generate text descriptions based on input data (often represented in the tabular form). A typical system uses huge training samples for learning the correspondence between tables and texts. However, large training sets are expensive to obtain, limiting the applicability of these approaches in real-world scenarios. In this work, we focus on few-shot data-to-text generation. We observe that, while fine-tuned pretrained language models may generate plausible sentences, they suffer from the low semantic coverage problem in the few-shot setting. In other words, important input slots tend to be missing in the generated text. To this end, we propose a search-and-learning approach that leverages pretrained language models but inserts the missing slots to improve the semantic coverage. We further fine-tune our system based on the search results to smooth out the search noise, yielding better-quality text and improving inference efficiency to a large extent. Experiments show that our model achieves high performance on E2E and WikiBio datasets. Especially, we cover 98.35% of input slots on E2E, largely alleviating the low coverage problem. | ['Lili Mou', 'Andreas Dengel', 'Zi Xuan Zhang', 'Shailza Jolly'] | 2021-12-06 | search-and-learn-improving-semantic-coverage | https://openreview.net/forum?id=brOPxR6Bav | https://openreview.net/pdf?id=brOPxR6Bav | null | ['data-to-text-generation'] | ['natural-language-processing'] | [ 2.05787733e-01 4.54588264e-01 -5.37699103e-01 -3.95342350e-01
-1.31024611e+00 -4.17874426e-01 6.35107756e-01 1.71591699e-01
-1.05341323e-01 1.13851535e+00 6.60257876e-01 -1.94152743e-01
5.82841523e-02 -1.30580342e+00 -8.58815551e-01 -9.79081839e-02
7.09474564e-01 8.54579866e-01 8.37115869e-02 -4.83864665e-01
1.46765947e-01 -3.58991951e-01 -1.59473813e+00 6.15620196e-01
1.30806506e+00 6.33545697e-01 3.92286152e-01 3.58314276e-01
-8.31030071e-01 7.26850271e-01 -7.20382214e-01 -6.45312488e-01
5.78563698e-02 -5.77889085e-01 -8.01203609e-01 4.91036698e-02
5.51905781e-02 -2.16596588e-01 -2.99759895e-01 8.67468297e-01
4.55871254e-01 8.90514329e-02 6.33336425e-01 -1.09844637e+00
-5.96750498e-01 1.27242947e+00 -1.96847796e-01 -2.69179821e-01
4.36554015e-01 3.10164317e-02 1.38243783e+00 -1.07307529e+00
8.04576397e-01 1.31297994e+00 3.98370177e-01 8.19381177e-01
-1.29132080e+00 -5.30073106e-01 7.39922673e-02 -3.09699774e-02
-1.26461792e+00 -7.11339176e-01 5.15141726e-01 -5.00299148e-02
9.00424123e-01 3.41758043e-01 3.58838916e-01 1.36441207e+00
-1.31103665e-01 1.14734054e+00 5.27238131e-01 -5.64581156e-01
3.59324872e-01 3.49790424e-01 -2.43346319e-01 3.79115582e-01
5.34105897e-01 -3.83889407e-01 -7.32168078e-01 -1.10218510e-01
5.07643282e-01 -5.06427102e-02 1.40144369e-02 -1.56258911e-01
-1.17249334e+00 1.00281072e+00 1.98653400e-01 7.05834404e-02
-3.43822926e-01 -1.40737565e-02 5.64442992e-01 8.08842108e-02
5.20588636e-01 8.10218513e-01 -4.44848686e-01 -3.02680463e-01
-9.85543966e-01 6.14296317e-01 9.07966495e-01 1.52116048e+00
6.14744246e-01 -1.68038443e-01 -8.24979603e-01 1.01673210e+00
-2.78827660e-02 5.42038202e-01 4.75452334e-01 -5.25522113e-01
1.02149212e+00 5.72141886e-01 3.66504908e-01 -6.32103145e-01
-1.12216949e-01 -2.46302888e-01 -8.75994146e-01 -5.25008559e-01
3.71613890e-01 -3.32852125e-01 -8.85075808e-01 1.71044850e+00
1.63874850e-01 -3.41691941e-01 3.49773824e-01 6.65789127e-01
7.06916451e-01 9.56114054e-01 1.33371726e-01 -5.80824912e-02
1.26426482e+00 -1.08421922e+00 -9.08019185e-01 -6.26010239e-01
9.32927966e-01 -6.99135840e-01 1.50667465e+00 2.40430385e-02
-1.04230881e+00 -3.49572182e-01 -7.50681996e-01 -1.19102493e-01
-3.10139835e-01 3.25877905e-01 5.72923064e-01 3.69253486e-01
-3.93811911e-01 4.87766623e-01 -5.80087304e-01 -2.76949257e-01
5.02051592e-01 -2.23498628e-01 7.62767717e-02 -3.31667095e-01
-1.57264674e+00 5.05244374e-01 6.96703315e-01 -4.29504544e-01
-5.64130187e-01 -8.29081476e-01 -1.00578678e+00 3.27050507e-01
9.22187746e-01 -8.54748547e-01 1.45649254e+00 -3.64758551e-01
-1.27385056e+00 2.84226269e-01 -4.41419184e-01 -5.63884020e-01
4.89917159e-01 -1.32920027e-01 -2.54307330e-01 -2.37279028e-01
3.90387863e-01 8.63119721e-01 4.10531610e-01 -1.10834455e+00
-7.62848556e-01 4.34917659e-02 1.67670369e-01 2.69970626e-01
-4.44889516e-01 -3.49371105e-01 -6.87516332e-01 -8.73588443e-01
-9.07171592e-02 -6.33804500e-01 -3.62103164e-01 -2.70814061e-01
-6.95880890e-01 -4.35071975e-01 2.82449573e-01 -4.57892835e-01
1.65598381e+00 -1.73102033e+00 -2.60488272e-01 -4.05750088e-02
-5.32197058e-02 1.04444996e-01 -2.39868656e-01 7.14427948e-01
4.26118731e-01 3.20219189e-01 -9.51356143e-02 -3.66844267e-01
3.59141529e-01 1.92685813e-01 -7.34093666e-01 -4.84764278e-01
2.68406719e-01 1.20641160e+00 -1.04218566e+00 -6.51974857e-01
-6.56781271e-02 -3.79163362e-02 -7.41499066e-01 1.92107141e-01
-9.12952065e-01 -4.53887247e-02 -6.75487161e-01 4.68286693e-01
2.39148200e-01 -3.55175525e-01 5.34515753e-02 5.64559735e-02
3.15908432e-01 9.02268291e-01 -1.06574929e+00 1.85320294e+00
-9.51051772e-01 3.37813556e-01 -4.60117191e-01 -7.81387031e-01
1.05273438e+00 2.55426764e-01 1.71473160e-01 -8.10382962e-01
-1.56748310e-01 3.93019795e-01 -2.95173556e-01 -3.99970263e-01
9.07443941e-01 -2.30571106e-01 -4.63337302e-01 6.95894837e-01
-6.67056367e-02 -3.64303976e-01 6.98865414e-01 4.31480438e-01
8.48510385e-01 1.24929048e-01 2.71979839e-01 -4.23886813e-02
1.06631808e-01 3.19584459e-01 4.95368510e-01 9.75286543e-01
4.94716287e-01 7.03305185e-01 5.97479641e-01 -1.22415550e-01
-1.38440537e+00 -6.61334574e-01 3.23615186e-02 9.97192383e-01
1.38347030e-01 -9.46237147e-01 -9.14197743e-01 -6.68440998e-01
-7.49043748e-02 1.45509255e+00 -4.35897470e-01 -4.05887932e-01
-4.34398949e-01 -6.95167303e-01 3.91334176e-01 5.42468250e-01
4.62585017e-02 -1.18329287e+00 -3.74730229e-01 5.33774197e-01
-7.34567761e-01 -9.76573944e-01 -6.63686097e-01 2.82517560e-02
-7.65045822e-01 -6.71671808e-01 -5.64473569e-01 -5.34605503e-01
6.46696329e-01 1.65582478e-01 1.52055562e+00 -2.51143038e-01
-1.76282853e-01 -2.37704664e-01 -5.59557438e-01 -5.78640401e-01
-6.93704486e-01 5.21408021e-01 -3.06728870e-01 -3.80795836e-01
6.61758125e-01 -2.36304685e-01 -2.94476807e-01 2.78796434e-01
-1.04660940e+00 5.75177431e-01 6.48246050e-01 1.17069495e+00
6.32052600e-01 1.54340804e-01 9.31421280e-01 -1.07044137e+00
8.30725133e-01 -6.00796938e-01 -4.85522509e-01 5.03062904e-01
-7.91077912e-01 5.89656651e-01 1.02818871e+00 -3.94522041e-01
-1.09740365e+00 -2.65087157e-01 -8.22531208e-02 -1.24924913e-01
1.40170649e-01 7.33542144e-01 -3.53503317e-01 8.86417866e-01
7.87316322e-01 3.76328319e-01 -2.81821162e-01 -4.01116937e-01
4.26953942e-01 7.85025179e-01 2.70517111e-01 -7.71422267e-01
7.98036635e-01 1.03263564e-01 -3.38847995e-01 -3.75939786e-01
-1.36622047e+00 -2.43977144e-01 -1.26222312e-01 1.35250449e-01
3.83502871e-01 -9.90314186e-01 -8.13784450e-02 -8.23731571e-02
-1.06943095e+00 -4.37780410e-01 -6.41141653e-01 1.54959813e-01
-6.00711465e-01 -7.33168200e-02 -3.63330156e-01 -8.03403616e-01
-6.89673424e-01 -8.60788107e-01 1.30642128e+00 1.86195284e-01
-4.23687190e-01 -7.91330457e-01 -1.05855122e-01 3.08711439e-01
3.33257228e-01 -2.40230992e-01 1.08078980e+00 -7.96774805e-01
-6.63074374e-01 -1.83414117e-01 -1.24248728e-01 -2.04002410e-01
2.23373130e-01 -1.67724252e-01 -6.31988645e-01 1.36078924e-01
-4.40434843e-01 -6.66352451e-01 8.27138543e-01 1.86642572e-01
1.49641609e+00 -5.48918784e-01 -3.19678187e-01 2.31754258e-01
1.19239914e+00 -1.46095101e-02 5.56433380e-01 2.78196812e-01
4.73214447e-01 8.21242213e-01 1.17753327e+00 8.02190781e-01
5.25413275e-01 7.71929204e-01 -1.08831460e-02 4.54108194e-02
-7.91687518e-02 -1.09787261e+00 1.39166236e-01 5.93365252e-01
6.46298826e-01 -5.51266193e-01 -8.21592927e-01 8.20079446e-01
-1.98040712e+00 -9.12620246e-01 2.64146060e-01 2.19405746e+00
1.54833269e+00 5.44953287e-01 -2.63431184e-02 -2.07083132e-02
6.99504793e-01 7.09483922e-02 -5.74743330e-01 -7.52093568e-02
-1.55015901e-01 1.52858302e-01 3.09121639e-01 3.53025913e-01
-7.73109078e-01 1.13114703e+00 5.26306820e+00 1.39984643e+00
-7.25690901e-01 -1.72979563e-01 8.38063061e-01 -4.18561399e-01
-9.39880252e-01 8.21180921e-03 -1.22388959e+00 6.20818257e-01
1.06345153e+00 -9.09410715e-01 3.50789875e-01 1.08325446e+00
3.33567470e-01 9.70150083e-02 -1.12437189e+00 9.14766073e-01
-1.02746539e-01 -1.68649721e+00 4.69506890e-01 -1.49287611e-01
8.52333546e-01 -4.80330765e-01 -2.71277457e-01 7.06803679e-01
5.11967003e-01 -1.12509739e+00 9.53949332e-01 3.24666291e-01
1.03927267e+00 -8.69973242e-01 6.17040455e-01 6.27806187e-01
-9.94869530e-01 8.37580785e-02 -7.48010576e-01 1.46430939e-01
3.68750006e-01 8.35200965e-01 -1.02057076e+00 4.09246504e-01
1.89800262e-01 4.56502229e-01 -4.09494191e-01 6.95559740e-01
-2.09832519e-01 4.06058550e-01 -2.36623153e-01 -4.77046847e-01
3.05857241e-01 3.17330100e-02 3.08030009e-01 1.08954465e+00
5.91266215e-01 -2.94616371e-02 2.20959067e-01 1.41857529e+00
-4.96521235e-01 1.49086654e-01 -5.73207319e-01 -3.06546509e-01
8.99797678e-01 1.11817491e+00 -4.49316084e-01 -6.54181361e-01
-3.57283503e-01 7.64467895e-01 2.69764751e-01 1.20472603e-01
-7.10997641e-01 -7.08968759e-01 5.54143012e-01 1.38863742e-01
3.08401763e-01 2.54263043e-01 -6.61718190e-01 -1.34132469e+00
3.61569345e-01 -9.97490108e-01 1.91361114e-01 -7.17839181e-01
-1.25441861e+00 4.61632997e-01 3.30179036e-02 -1.20656252e+00
-7.87330866e-01 -1.60510227e-01 -5.13993323e-01 9.03392315e-01
-1.25417900e+00 -7.36126482e-01 -2.18120515e-01 1.11187376e-01
1.19145000e+00 -8.35542530e-02 7.92662561e-01 7.28867278e-02
-5.43223023e-01 8.54222834e-01 2.21880838e-01 6.31823242e-02
7.92000055e-01 -1.24398482e+00 9.39774632e-01 7.96752095e-01
2.39452735e-01 7.62996912e-01 7.42764890e-01 -8.41648281e-01
-1.32739604e+00 -1.43815494e+00 1.35315013e+00 -2.65323937e-01
5.93204260e-01 -7.40361452e-01 -8.84666562e-01 3.97544146e-01
-1.12399846e-01 -3.76103938e-01 5.90259492e-01 1.62496448e-01
-2.31298551e-01 -4.77878973e-02 -8.52471590e-01 1.02158010e+00
9.98158395e-01 -4.55840945e-01 -6.43442452e-01 5.77031732e-01
9.85982120e-01 -5.19435763e-01 -6.46108806e-01 3.39329988e-02
2.79658526e-01 -5.14380455e-01 7.45439053e-01 -6.59294784e-01
1.09727383e+00 2.82410672e-03 3.04072984e-02 -1.52594483e+00
2.61402689e-02 -7.65944302e-01 -3.29756528e-01 1.40044904e+00
1.03997111e+00 -1.29438624e-01 9.66158330e-01 8.87137353e-01
-1.77063718e-02 -9.41578388e-01 -5.59424698e-01 -7.48726726e-01
1.09609179e-01 -3.43701929e-01 1.05541933e+00 5.56945026e-01
2.88076609e-01 7.72597432e-01 -4.45453554e-01 -5.02829432e-01
5.08333087e-01 3.42212260e-01 9.07185733e-01 -8.98583770e-01
-2.37460330e-01 -2.43904144e-01 4.59152073e-01 -1.31133187e+00
1.50789082e-01 -8.84598315e-01 4.06163931e-01 -1.68671894e+00
3.66781533e-01 -8.53937030e-01 1.94128007e-01 4.53770518e-01
-7.26404607e-01 -1.81605607e-01 4.45919558e-02 1.13186605e-01
-6.70403361e-01 9.25762415e-01 1.26604843e+00 -1.36371151e-01
-2.20605001e-01 5.98707842e-03 -1.10574460e+00 2.70782977e-01
8.70592535e-01 -5.15214741e-01 -6.16424799e-01 -4.36653167e-01
3.43839377e-01 2.25111112e-01 -1.00803815e-01 -7.70777345e-01
1.38045281e-01 -5.31244934e-01 1.49245366e-01 -6.37003660e-01
1.02470398e-01 -3.64343375e-01 -3.18863131e-02 2.48391718e-01
-9.58864868e-01 -3.07748556e-01 1.20991822e-02 4.64369059e-01
-2.53091067e-01 -5.47243893e-01 5.12120485e-01 -2.13921398e-01
-4.26981598e-01 3.39585066e-01 -1.65809914e-01 5.86371541e-01
6.23791039e-01 6.64236993e-02 -3.20287228e-01 -5.63160598e-01
-9.93879139e-02 4.55749840e-01 4.08176988e-01 5.63297808e-01
3.92489284e-01 -1.54034173e+00 -7.84466505e-01 1.47203729e-01
6.70789003e-01 4.45196092e-01 5.47994189e-02 2.96627641e-01
-1.14278860e-01 9.10685599e-01 1.15939662e-01 -1.67021513e-01
-7.20740318e-01 5.02053797e-01 -8.62200037e-02 -6.28492475e-01
-6.42575502e-01 6.56362414e-01 1.31850109e-01 -5.19905448e-01
2.92589784e-01 -2.68518627e-01 1.19528593e-03 -2.97293570e-02
7.67202735e-01 1.39095888e-01 1.93688758e-02 5.80792055e-02
1.08781628e-01 -1.49825647e-01 -3.18600416e-01 -2.17076942e-01
1.25162292e+00 -2.09991738e-01 2.40990326e-01 3.51852864e-01
8.64573121e-01 8.26903805e-02 -1.06592429e+00 -6.00800455e-01
2.69849509e-01 -5.18377900e-01 -1.31375432e-01 -8.44878018e-01
-6.85175538e-01 7.40109980e-01 -3.64886969e-01 4.70273569e-02
8.68122756e-01 1.33622065e-01 1.34742117e+00 7.42397666e-01
4.01627153e-01 -1.36269176e+00 1.06966197e-01 5.39015055e-01
6.23835564e-01 -1.21718597e+00 -3.26662660e-01 -4.50455368e-01
-9.30072665e-01 9.97385561e-01 7.65093625e-01 4.07338589e-01
-1.03887260e-01 2.98107266e-01 -2.03434914e-01 9.60351154e-02
-1.31643772e+00 -1.37787834e-01 2.23250017e-01 4.77685958e-01
5.45275331e-01 3.33537124e-02 -3.30927074e-01 1.02231121e+00
-5.29312432e-01 1.32353306e-01 6.91110253e-01 7.65333951e-01
-6.12768054e-01 -1.31680775e+00 -1.81895018e-01 9.08878863e-01
-2.52966583e-01 -5.29572964e-01 -2.81485438e-01 4.23314482e-01
-2.88911939e-01 9.54351068e-01 3.51339802e-02 -1.62552610e-01
3.60040367e-01 2.15234831e-01 1.23531386e-01 -9.00195479e-01
-1.98501632e-01 5.03795967e-02 4.98593420e-01 -3.86954337e-01
2.54238307e-01 -4.96063948e-01 -1.15696847e+00 -4.05434191e-01
-3.04705590e-01 4.20489281e-01 4.82425719e-01 8.24133754e-01
4.97893512e-01 6.02998495e-01 5.49740672e-01 -2.70785809e-01
-9.99847651e-01 -1.12971389e+00 -4.94579643e-01 5.69684803e-01
-1.05035208e-01 -5.06463170e-01 -2.66299173e-02 3.02944630e-02] | [11.660916328430176, 8.811751365661621] |
3159835b-ed29-4d08-ba3e-3faeb9ff033d | adaptive-hierarchical-similarity-metric | 2111.00006 | null | https://arxiv.org/abs/2111.00006v1 | https://arxiv.org/pdf/2111.00006v1.pdf | Adaptive Hierarchical Similarity Metric Learning with Noisy Labels | Deep Metric Learning (DML) plays a critical role in various machine learning tasks. However, most existing deep metric learning methods with binary similarity are sensitive to noisy labels, which are widely present in real-world data. Since these noisy labels often cause severe performance degradation, it is crucial to enhance the robustness and generalization ability of DML. In this paper, we propose an Adaptive Hierarchical Similarity Metric Learning method. It considers two noise-insensitive information, \textit{i.e.}, class-wise divergence and sample-wise consistency. Specifically, class-wise divergence can effectively excavate richer similarity information beyond binary in modeling by taking advantage of Hyperbolic metric learning, while sample-wise consistency can further improve the generalization ability of the model using contrastive augmentation. More importantly, we design an adaptive strategy to integrate this information in a unified view. It is noteworthy that the new method can be extended to any pair-based metric loss. Extensive experimental results on benchmark datasets demonstrate that our method achieves state-of-the-art performance compared with current deep metric learning approaches. | ['Heng Huang', 'Cheng Deng', 'Lei Luo', 'Jiexi Yan'] | 2021-10-29 | null | null | null | null | ['learning-with-noisy-labels', 'learning-with-noisy-labels'] | ['computer-vision', 'natural-language-processing'] | [-0.09870818 -0.5802549 0.0411524 -0.8830792 -0.87624 -0.34917134
0.24438809 0.3142202 -0.48337328 0.64628476 0.13838565 -0.03420069
-0.673165 -0.8137064 -0.21633571 -0.89420724 0.16939202 0.2985468
0.2398484 -0.23420365 0.25370023 0.36049047 -1.3312131 -0.09389382
1.176731 1.4264355 0.08944861 -0.09943433 -0.03617208 0.75446385
-0.5485265 -0.31407765 0.36516455 -0.40800825 -0.5619576 -0.26065668
0.26905176 -0.5586853 -0.6418617 1.2664747 0.7614847 0.34050715
0.63641614 -1.4745357 -0.8092944 0.5988003 -0.57607895 0.34938538
-0.01315182 0.19142635 1.2587274 -0.9916431 0.08487887 1.2516801
0.8510303 0.26214966 -1.0129951 -0.9828056 0.18496035 0.7393413
-1.6667744 -0.09643393 1.1223625 -0.15683474 0.26209205 0.17412819
0.0894943 0.9229783 -0.02833911 0.7866615 1.1241102 0.04797552
0.27352935 -0.11859238 0.02983012 0.5472993 0.17072113 0.0223388
-0.25044388 -0.03802048 0.3192316 0.3543216 -0.24613999 -0.53550595
-1.220876 0.88839257 0.82335216 0.37994504 0.09949216 0.09213964
0.628084 0.39534956 0.5592325 0.33011374 -0.3315036 -0.2314938
-0.7815411 0.24735735 0.3804805 0.80095917 0.8854783 -0.30373114
-0.5025359 1.1506401 0.23110326 0.26280028 0.6828606 -0.978015
0.51334494 0.8028593 -0.10848098 -1.3174509 -0.62989277 -0.8041249
-1.1547737 -0.08634753 0.29330623 0.06471055 -0.46150506 1.9663783
0.3603241 0.18527824 -0.34225738 0.98979545 0.7674761 0.41793868
-0.14308195 -0.06515543 0.86354643 -0.7470315 -0.67785835 0.10107554
0.8675992 -0.6572219 1.1570377 0.2831236 -0.7239689 -0.49880904
-1.372953 -0.29936928 -0.30131996 -0.10186643 0.49563217 0.4220949
-0.7410956 0.8149069 -0.6264583 -0.1030343 0.6721289 0.3246786
-0.17077388 -0.52290326 -1.4630408 0.7221486 0.4131756 0.06220651
-0.6281987 -0.6460726 -0.93863964 0.10911834 0.34852704 -0.3646792
1.2674248 -0.16021179 -1.0879511 0.65198034 0.04951129 -0.23955323
0.7623802 0.09711833 -0.40337074 -0.10697037 0.23173124 0.24179472
0.20971537 -1.0727912 -0.6847036 -0.56155074 0.10501078 0.20994452
-0.9728677 -0.18416445 -0.07341534 -0.85861194 0.4701315 -0.48925474
-0.13411923 0.511849 -0.2113056 -0.6103331 0.8260162 -0.46059874
1.6386901 -2.237209 -0.08475436 0.22591552 0.52620304 0.27977756
-0.23135222 0.23942132 0.2114814 0.08747211 -0.6185858 -0.40872097
0.3375073 0.23183218 0.09233701 0.61976624 0.13498817 0.7001788
-1.0795647 -0.6384437 0.19925648 0.26893252 -0.33710292 0.06945426
0.09109551 0.2581141 -0.56788564 0.6837686 1.0001202 -0.10463922
-0.31634277 -0.2897447 0.22294278 0.30789173 -1.1930511 1.9792159
-0.4469063 0.4400695 -0.23027627 -1.4508216 1.3197354 -0.08413894
0.6269988 -0.93226415 0.30231392 0.37861377 0.15808003 -0.26600367
0.25919464 -0.22866748 -0.09567222 0.50968987 -0.24406467 0.11072374
0.25406447 0.10446412 1.0568974 -0.29020655 0.16317648 -0.3172804
0.6431832 -0.6137649 1.0725416 0.5882178 -0.6763442 0.6842093
0.41049653 -0.26432914 -0.8745719 -1.1185598 -0.48209494 1.01781
0.5658612 -0.26375347 -0.5700469 -0.8312682 0.14926563 0.45078117
-0.64914894 -0.89386463 -0.5112675 -1.0817283 0.5485481 0.71214867
0.86874866 -0.6409763 0.18159646 0.33818918 -0.26269916 -0.8251572
-0.74701923 0.01863316 -0.71368605 -1.0783104 -0.54222685 -0.8336658
0.35128146 0.5191327 0.86013514 0.15254803 -0.29406658 -0.15835644
-0.4941808 -0.10311722 -0.15293978 0.20378155 0.13254519 0.20034951
0.7281811 -0.76323193 -0.8974553 0.66998404 -0.9290242 -0.46170554
0.50486773 1.0261971 0.5084523 0.28164706 0.89691156 -0.47374633
0.7027211 -0.65299416 -0.22368856 0.14463393 -0.74228 0.08288058
0.8567827 -0.3591194 -0.81515265 -0.59594154 -0.06507844 -0.19672671
0.21983029 0.51258725 -0.714543 -0.057361 0.54480696 0.18619537
-0.05982281 -0.6655507 0.26697493 0.941699 0.3473755 -0.7083918
0.7351192 0.49258146 0.2139665 -0.07364452 -1.087582 -0.49899963
-0.46011472 0.03778485 0.43438295 -0.70540935 -0.6347127 0.7585086
-0.839434 -0.0309283 -0.16827595 0.55621403 -0.29951164 0.5224002
-0.5993726 -0.43160942 -0.23533604 -1.141187 0.79924804 0.2635676
0.09511386 -1.0290519 -0.00860132 0.253141 0.46571112 0.26012495
0.8569223 -1.0180988 -0.27148905 -0.33070797 -0.7586002 0.8012845
0.42846778 -0.26914045 -0.8151113 -0.37731093 0.15515095 -0.48652065
0.9869888 -0.05450894 1.7660971 -0.02871514 -0.13175981 0.80285954
1.1250492 0.1650157 0.35118088 0.3585367 0.94172704 0.42594582
0.97066164 0.76806414 0.678652 0.68986946 0.6379152 -0.0408293
-0.03328182 -0.14595293 -0.10155963 0.9464125 0.5538673 -0.03235597
-0.84789705 0.40099254 -1.938272 -0.80202734 0.05224607 2.228702
1.0440068 0.14258423 -0.07285819 0.68767273 0.91345817 0.2590983
-0.9432784 -0.07503503 -0.21857543 -0.06151079 0.31023875 0.14407757
-1.206352 0.49247026 4.68751 1.2548008 -0.9047756 0.18024446
0.7346575 -0.2970486 -0.4882254 -0.32675922 -0.48961583 0.9469963
0.39828676 -0.4385492 0.35207793 0.72977644 0.25900573 0.25186104
-1.275763 1.1943527 0.12466259 -0.83118284 -0.08416516 0.04094423
0.7252405 -0.12344071 0.21210085 0.54877937 0.305787 -0.9098193
0.48352295 0.43937942 0.7719449 -1.1775756 0.98796535 0.30403858
-1.2674764 -0.33907762 -0.71044374 -0.18316887 -0.04556902 0.99882
-0.30159992 0.52062625 0.70871854 1.1518096 -0.7615975 1.361903
0.00562268 0.45712298 -0.23997389 -0.10189433 0.37725374 -0.26547992
0.30811357 0.9218007 0.44991162 0.0931697 0.32654047 0.68905413
-0.38344833 0.18142028 -0.40831476 0.28582847 0.96025527 1.0290937
-0.33224368 -0.0445001 -0.3520993 0.8654175 0.537281 0.10540887
-0.87647617 -0.8025938 0.9178672 -0.12919928 0.03143679 -0.1476576
-0.59870565 -0.92631716 0.46393192 -0.628291 0.5657914 -0.327612
-1.8033292 0.2985346 -0.29958853 -1.5427219 0.18074481 -0.42844027
-0.6278778 0.7210521 -1.467272 -1.0249245 -0.6436444 0.5413649
0.22732992 -0.04564674 0.41973603 0.7137174 -0.66386867 1.1440754
0.6988685 0.2561642 0.9743624 -1.3340712 0.26090842 0.6856437
-0.09233122 0.34986028 0.24217375 -0.16947103 -0.8815614 -1.3463607
0.6291672 -0.24630685 0.7785808 -0.42643443 -1.1273288 0.18588029
-0.4663165 0.28205875 0.90565354 -0.0979592 -0.680893 -0.841915
-1.4331068 0.5239086 1.3071043 -0.57031435 -0.4128261 0.42336097
0.96584004 -0.14855847 -0.90196216 0.7581462 0.35535023 -1.0896634
0.8120872 -0.44946724 0.3503217 -0.43554962 -0.40993893 -1.2833233
-0.34774983 -0.3660552 -0.09187211 1.5148017 -0.02598321 -0.8092128
0.45686662 0.54559606 -0.3131975 -1.130627 -1.2652427 -1.1077228
0.44236523 -0.3301749 0.998125 1.1438191 -0.0911511 0.05528521
-0.25869194 -0.11601323 0.7100668 -0.03970925 0.44101217 -1.4841319
-0.00842224 -0.73754954 -0.8057384 -1.00741 0.0981357 -1.0582281
0.0083333 -1.6016057 0.34772912 -0.853735 -1.0173856 0.3943249
-0.37881488 0.3532672 0.02325895 0.135958 -0.85456795 1.1719322
1.3014168 -0.2405264 0.16726443 -0.04973236 -0.94249034 0.5469645
0.84951013 -0.60000193 -0.4818755 -0.49037284 0.067603 -0.4851896
0.14753872 -1.0625099 0.24596122 -0.12697412 0.15413073 -0.24495533
0.09564454 -0.77474946 -0.52250993 0.4034319 -0.4078521 0.02730385
-0.23028277 0.64611804 -0.3769159 -0.12003392 0.98973966 0.22343925
-0.5524055 0.66457903 0.27801177 0.45039222 0.9110841 0.03031258
-0.34853214 -0.20840672 -0.23848304 0.55964154 0.44668692 0.51289177
0.5516952 -1.8349471 -0.86483026 -0.02021399 0.46857697 0.2334992
0.3154528 0.76791084 -0.32722166 0.0234679 0.02535859 -0.5360638
-0.8366921 0.55459684 0.3850743 -0.23989168 -0.2761035 0.90396875
0.13176599 -0.5702081 0.5142183 -0.33023492 0.07846663 0.07758719
0.6296503 0.7739927 0.22618254 -0.34849164 -0.51326734 0.54023606
-0.39915332 0.18701422 0.9961729 -0.25069493 -0.22700392 0.54375935
1.6545404 -0.46699286 -1.3058052 -0.7691072 0.04051956 -0.5582297
0.179403 -0.78192705 -1.140305 1.057059 0.67852545 0.02524291
1.2272422 -0.24150565 1.1698248 0.5272443 0.38343132 -1.1490842
0.2214274 0.25962344 0.7120148 -1.6056006 0.02478032 -0.10677402
-0.10961717 0.6997658 0.7600613 -0.01173187 0.7467705 -0.06474026
-0.00768637 0.16921364 -0.30174187 -0.08111084 0.3000862 0.5125806
0.2787366 0.06450622 -0.6893261 0.9074525 -0.04475654 -0.3226271
0.2322812 0.8109917 -0.51359904 -1.0566552 -0.01966214 0.58058673
-0.20109838 -0.12030556 -0.23533116 0.5927839 0.201391 1.1480517
0.04255766 -0.63867956 0.18583879 -0.3232861 0.26229835 -0.35744855
-0.222184 -0.57109493 -0.3577759 -0.55610085 -0.03416872 -0.63596433
-1.2238431 -0.4526387 -0.2617977 0.06152908 0.5961322 1.1463419
0.23983926 0.33503786 1.1719532 -0.34697536 -1.0963418 -1.1693691
-0.77356154 0.6307821 0.2972937 -0.94428205 -0.69801295 -0.5418331 ] | [9.441994667053223, 3.2249670028686523] |
84c45e86-4e73-4bbf-a727-57fe122c6052 | weakly-and-semi-supervised-panoptic | 1808.03575 | null | http://arxiv.org/abs/1808.03575v3 | http://arxiv.org/pdf/1808.03575v3.pdf | Weakly- and Semi-Supervised Panoptic Segmentation | We present a weakly supervised model that jointly performs both semantic- and
instance-segmentation -- a particularly relevant problem given the substantial
cost of obtaining pixel-perfect annotation for these tasks. In contrast to many
popular instance segmentation approaches based on object detectors, our method
does not predict any overlapping instances. Moreover, we are able to segment
both "thing" and "stuff" classes, and thus explain all the pixels in the image.
"Thing" classes are weakly-supervised with bounding boxes, and "stuff" with
image-level tags. We obtain state-of-the-art results on Pascal VOC, for both
full and weak supervision (which achieves about 95% of fully-supervised
performance). Furthermore, we present the first weakly-supervised results on
Cityscapes for both semantic- and instance-segmentation. Finally, we use our
weakly supervised framework to analyse the relationship between annotation
quality and predictive performance, which is of interest to dataset creators. | ['Philip H. S. Torr', 'Anurag Arnab', 'Qizhu Li'] | 2018-08-10 | weakly-and-semi-supervised-panoptic-1 | http://openaccess.thecvf.com/content_ECCV_2018/html/Anurag_Arnab_Weakly-_and_Semi-Supervised_ECCV_2018_paper.html | http://openaccess.thecvf.com/content_ECCV_2018/papers/Anurag_Arnab_Weakly-_and_Semi-Supervised_ECCV_2018_paper.pdf | eccv-2018-9 | ['weakly-supervised-panoptic-segmentation', 'weakly-supervised-instance-segmentation'] | ['computer-vision', 'computer-vision'] | [ 3.24871361e-01 6.98910952e-01 -3.97236884e-01 -5.68868101e-01
-1.29542077e+00 -6.86824322e-01 7.13902116e-01 2.44803295e-01
-5.22740245e-01 7.86012173e-01 -1.99199915e-01 -3.34655568e-02
9.06453133e-02 -7.95058966e-01 -9.56080675e-01 -7.08526552e-01
1.90825596e-01 7.71277964e-01 7.16182232e-01 8.23099986e-02
-4.52364311e-02 1.28451273e-01 -1.80718398e+00 7.13335335e-01
8.23919952e-01 9.77860570e-01 3.06210309e-01 5.53683519e-01
-3.19329232e-01 8.04840326e-01 -3.58345062e-01 -6.44413590e-01
1.66087851e-01 -2.54485190e-01 -1.28663754e+00 4.79273587e-01
9.48360503e-01 2.40829602e-01 3.03338975e-01 1.08752501e+00
-3.34085971e-02 -2.25810111e-01 7.16549635e-01 -8.16232979e-01
-2.40881085e-01 5.93629539e-01 -5.23959041e-01 -1.11134842e-01
1.07496671e-01 1.04385048e-01 1.35332966e+00 -7.19857454e-01
7.71755636e-01 8.57102275e-01 7.88061619e-01 5.47327518e-01
-1.29192972e+00 -1.91618860e-01 3.54451984e-01 2.09844448e-02
-1.27773392e+00 -2.21905395e-01 3.93087536e-01 -5.97969294e-01
9.22775447e-01 3.92090112e-01 5.28213203e-01 8.20074260e-01
-5.13835669e-01 1.15695810e+00 1.36540508e+00 -5.58070660e-01
1.31253198e-01 6.28631413e-01 2.69491911e-01 7.09506571e-01
8.21174607e-02 -1.51072457e-01 -2.19413444e-01 3.30271244e-01
5.86363375e-01 -2.12787032e-01 -1.14136688e-01 -4.17791903e-01
-1.17018414e+00 5.05398571e-01 4.14838910e-01 4.76049066e-01
-9.52429697e-02 2.82519042e-01 3.14720929e-01 -1.22644223e-01
9.03347194e-01 3.56668204e-01 -9.17501092e-01 2.28879992e-02
-1.41929948e+00 7.67185912e-02 8.50733995e-01 1.10247350e+00
1.06391704e+00 -1.71739116e-01 -7.03900978e-02 9.33081746e-01
6.39226884e-02 3.02945226e-01 1.12489738e-01 -9.63246942e-01
5.12700438e-01 6.96901619e-01 1.42967850e-01 -3.14953595e-01
-4.29270983e-01 -7.39919960e-01 -4.50866997e-01 1.41382933e-01
7.57696509e-01 6.45480528e-02 -1.23436534e+00 1.42273390e+00
2.48459354e-01 2.38869801e-01 8.19913968e-02 8.38883221e-01
9.38271642e-01 4.21821594e-01 4.94384855e-01 1.06097236e-02
1.61470115e+00 -1.31139982e+00 -3.56256992e-01 -6.33942962e-01
8.22394371e-01 -5.47808647e-01 1.20269561e+00 2.44659737e-01
-1.16049671e+00 -7.49495685e-01 -5.69052577e-01 -1.01007499e-01
-5.98945856e-01 4.12592441e-01 6.69205785e-01 5.84566116e-01
-1.01884735e+00 6.52922630e-01 -6.78771138e-01 -4.40234661e-01
7.10213661e-01 1.90289572e-01 -2.72601813e-01 8.01573787e-03
-9.10640121e-01 7.29393184e-01 6.18899763e-01 -2.34447524e-01
-9.50355947e-01 -7.93073714e-01 -8.23692620e-01 3.06295697e-02
3.82696122e-01 -3.63500208e-01 1.20501208e+00 -1.37309051e+00
-8.56696248e-01 1.78250492e+00 -1.93580315e-01 -7.36232221e-01
7.49539614e-01 -1.47095263e-01 -2.11793125e-01 1.93683475e-01
3.12364936e-01 1.23913002e+00 6.33176446e-01 -1.63091481e+00
-1.12260044e+00 -4.76046860e-01 1.49355233e-01 1.05940320e-01
-1.24299295e-01 -2.48736635e-01 -8.07793140e-01 -4.47567761e-01
2.14411795e-01 -7.96806157e-01 -3.77074033e-01 -1.68858275e-01
-6.05357528e-01 -4.35882270e-01 7.21880853e-01 -6.19682610e-01
7.57701278e-01 -2.28345132e+00 -8.75534117e-02 -3.82409841e-02
-1.20026536e-01 2.68010229e-01 -5.47464974e-02 -3.34532745e-02
-2.05019653e-01 1.92363977e-01 -6.74816370e-01 -5.98730445e-01
4.68617864e-02 4.45521116e-01 -2.58224249e-01 2.90181041e-01
3.88027489e-01 9.75024402e-01 -8.52511823e-01 -7.04521716e-01
3.86992007e-01 3.16072822e-01 -3.93916816e-01 1.68602169e-01
-7.21664131e-01 3.89716804e-01 -2.86127776e-01 6.66052580e-01
5.43151438e-01 -3.82807821e-01 1.72193665e-02 -7.60128573e-02
-2.38384947e-01 1.52199253e-01 -9.03085649e-01 1.71672499e+00
-4.58323836e-01 6.45908535e-01 9.69811976e-02 -1.19262993e+00
7.96999812e-01 1.92701831e-01 3.36566687e-01 -7.70548344e-01
-1.05312914e-01 4.36054379e-01 -4.73575503e-01 -4.53359842e-01
5.45188665e-01 -1.51022226e-01 -1.08528072e-02 -1.84171144e-02
3.33218724e-01 -4.38870400e-01 4.53778625e-01 6.54064715e-02
7.19327986e-01 6.96764827e-01 9.50028971e-02 -5.83654404e-01
5.66572368e-01 4.94055599e-01 4.01043236e-01 7.70832479e-01
-1.16744116e-01 1.03964746e+00 7.11745262e-01 -4.33058053e-01
-1.08111429e+00 -8.71574342e-01 -4.14318502e-01 1.22956371e+00
1.93880394e-01 -1.73572049e-01 -1.34508824e+00 -1.01661313e+00
-2.39017084e-01 8.29605103e-01 -7.69653857e-01 4.06108618e-01
-3.11871052e-01 -7.39952683e-01 4.33095962e-01 7.08802521e-01
6.72799885e-01 -1.13418543e+00 -3.28447253e-01 5.00048436e-02
-2.00491786e-01 -1.55319917e+00 1.61264658e-01 6.76263630e-01
-8.43563318e-01 -1.10454023e+00 -7.74207771e-01 -1.09899890e+00
7.04071045e-01 -1.32858053e-01 1.58356774e+00 1.56732216e-01
-3.74225765e-01 4.09818619e-01 -2.61632651e-01 -3.09750795e-01
-2.26911411e-01 3.16692680e-01 -7.43380070e-01 -1.15132645e-01
5.28367281e-01 6.46330416e-02 -4.16781455e-01 3.55680585e-01
-7.88940609e-01 8.68197680e-02 5.07643938e-01 6.02853835e-01
1.07770801e+00 6.42846525e-02 1.66125745e-01 -1.54343474e+00
-4.24789399e-01 -2.37253189e-01 -6.59223199e-01 2.63202161e-01
-3.51059496e-01 -1.03314012e-01 4.11595523e-01 8.16238746e-02
-1.25891364e+00 5.74848115e-01 -4.11631525e-01 1.21975601e-01
-1.03663850e+00 8.21755081e-02 -1.96993500e-01 2.15467200e-01
6.95153356e-01 2.64380760e-02 -5.03631592e-01 -5.67486227e-01
4.24655408e-01 6.09400630e-01 7.00190485e-01 -6.21367455e-01
5.22073627e-01 8.90892029e-01 -2.54423499e-01 -9.80844140e-01
-1.61455142e+00 -9.77710664e-01 -9.87930536e-01 -4.72538173e-02
1.36968029e+00 -1.20961702e+00 -2.71350801e-01 3.22314978e-01
-1.07414842e+00 -7.33422041e-01 -6.58661187e-01 6.37536943e-02
-7.58870661e-01 2.61224210e-01 -5.72258711e-01 -7.47849286e-01
1.01547711e-01 -1.07191527e+00 1.53173840e+00 8.29608068e-02
-1.35209352e-01 -1.19780517e+00 -1.68476939e-01 9.52185273e-01
-1.03020310e-01 1.86963528e-01 7.01952159e-01 -7.10933149e-01
-5.50363421e-01 -1.48376629e-01 -6.66069329e-01 4.88426119e-01
-5.31547129e-01 -1.39035359e-01 -1.47984934e+00 1.58837155e-01
-4.70789701e-01 -3.97913486e-01 1.29395378e+00 3.52385849e-01
1.27167153e+00 8.35424215e-02 -3.52356732e-01 4.53729749e-01
1.64659643e+00 -4.13225591e-01 6.77106023e-01 5.57956636e-01
8.43764484e-01 1.05514121e+00 8.08907866e-01 1.23886086e-01
3.47449481e-01 7.05893397e-01 6.67976379e-01 -6.58400774e-01
-3.56022477e-01 -2.00205654e-01 5.33473722e-05 7.68477796e-04
-2.40711480e-01 -1.64098442e-01 -9.29459095e-01 9.83937323e-01
-2.02182460e+00 -7.58842468e-01 -8.82334352e-01 1.94357705e+00
8.43759656e-01 3.99082959e-01 2.26962462e-01 2.50619799e-01
6.54376268e-01 1.69865400e-01 -9.53993872e-02 -2.41812915e-01
-2.01056808e-01 3.65871578e-01 9.09160197e-01 4.43601370e-01
-1.47768044e+00 1.30514431e+00 6.44282389e+00 1.10565472e+00
-6.85757220e-01 3.12150747e-01 1.05672157e+00 2.14046329e-01
-2.45046109e-01 1.15129717e-01 -1.06314337e+00 3.55160624e-01
8.31197321e-01 8.98162305e-01 -9.74046364e-02 1.24844420e+00
3.27697806e-02 -2.71218300e-01 -1.11598885e+00 6.95689380e-01
-3.63911502e-02 -1.24221635e+00 -1.71134517e-01 -6.09150529e-02
1.15738964e+00 2.06374481e-01 -8.39863196e-02 2.83172250e-01
2.89277822e-01 -1.02635491e+00 1.04227257e+00 1.51837841e-01
7.67258406e-01 -5.12881160e-01 9.51562822e-01 4.41214204e-01
-1.03178084e+00 5.71430475e-02 -4.81523722e-01 9.24703702e-02
4.02887948e-02 9.32205141e-01 -5.29057205e-01 2.88860559e-01
1.00005996e+00 7.57882893e-01 -6.88359320e-01 8.11105669e-01
-6.86510980e-01 7.97258317e-01 -2.24963173e-01 2.94645548e-01
6.54942274e-01 -1.49788171e-01 6.92477822e-02 1.77961385e+00
-1.35084137e-01 -5.19193672e-02 4.03266430e-01 9.01718438e-01
-5.75037114e-02 2.78567635e-02 -1.69887275e-01 2.78954089e-01
-1.21844001e-01 1.37990892e+00 -1.16103375e+00 -5.61491311e-01
-4.23816383e-01 1.02723098e+00 4.33622152e-01 3.27733457e-01
-7.71694779e-01 -8.22746530e-02 3.17434162e-01 4.18111593e-01
4.85892266e-01 8.62428546e-02 -6.82352960e-01 -1.03170896e+00
5.14619462e-02 -3.25963497e-01 4.77145880e-01 -7.73963153e-01
-9.99596953e-01 4.47151452e-01 -1.25908256e-01 -8.99929285e-01
-3.19380430e-03 -8.33205104e-01 -3.42440277e-01 5.63496470e-01
-2.06735420e+00 -1.62550318e+00 -3.93710196e-01 4.35480118e-01
7.80795336e-01 3.05771381e-01 7.61787415e-01 2.86307305e-01
-4.11499321e-01 1.05948918e-01 -6.17539063e-02 1.89740628e-01
4.15254951e-01 -1.82501161e+00 2.40033969e-01 7.41099775e-01
5.63987434e-01 2.27828175e-02 6.77751839e-01 -2.93408632e-01
-4.46014643e-01 -1.29480302e+00 1.10069358e+00 -7.43348241e-01
5.01231074e-01 -4.43004310e-01 -8.93614233e-01 6.48448765e-01
9.63700712e-02 4.36779171e-01 4.61501598e-01 2.82668531e-01
-4.81081486e-01 9.79414880e-02 -1.08527017e+00 2.09168613e-01
1.12162185e+00 -5.05738437e-01 -5.42198956e-01 7.36595631e-01
6.30557060e-01 -3.49127233e-01 -6.21644497e-01 4.38080698e-01
1.76032051e-01 -1.41485286e+00 1.04517543e+00 -3.41104537e-01
6.01993978e-01 -2.15217456e-01 7.61476308e-02 -9.09767389e-01
-1.13950104e-01 -2.10297126e-02 3.38694155e-01 1.35989189e+00
8.59720826e-01 -1.76627368e-01 1.35703075e+00 3.81096780e-01
-4.11806226e-01 -5.71281254e-01 -6.71105504e-01 -8.14271271e-01
2.92304382e-02 -7.58497834e-01 2.81859308e-01 9.28566277e-01
-2.70888299e-01 4.68792804e-02 -1.25516579e-01 1.35750696e-01
6.02876782e-01 3.12189639e-01 5.47658026e-01 -1.17584276e+00
-4.07028496e-01 -4.46674049e-01 -3.68188858e-01 -1.02442539e+00
3.77738208e-01 -8.28427374e-01 4.41688687e-01 -1.71907341e+00
2.75394738e-01 -5.18859982e-01 -1.48293868e-01 6.60027623e-01
-1.55993864e-01 9.02852654e-01 -2.51960784e-01 1.96302265e-01
-1.18962252e+00 -4.33412194e-02 1.05012131e+00 -1.34155393e-01
6.39123842e-02 -2.30669007e-02 -4.61803526e-01 1.12984669e+00
5.61261654e-01 -3.37231845e-01 -8.49936157e-02 -5.84670126e-01
6.12197444e-02 -2.96762675e-01 6.20755136e-01 -9.70700383e-01
-1.46629930e-01 1.93685293e-03 3.76245290e-01 -6.53383374e-01
2.03826442e-01 -1.02469611e+00 -3.38201880e-01 3.68865430e-01
-4.41418618e-01 -7.91512489e-01 1.28879726e-01 4.86294210e-01
-5.18176436e-01 -5.31719506e-01 1.06274068e+00 -5.66522300e-01
-1.13841546e+00 1.45259798e-01 8.50565657e-02 3.16139489e-01
1.08782768e+00 -3.69914770e-01 -1.48932278e-01 6.52079210e-02
-1.08256209e+00 1.56371966e-01 4.86161441e-01 5.95226176e-02
-9.63819120e-03 -7.85299182e-01 -7.28477299e-01 -3.01039610e-02
5.44001818e-01 4.87483978e-01 1.55213431e-01 6.36239767e-01
-6.78161144e-01 5.10416746e-01 2.25201681e-01 -1.00009775e+00
-1.17916656e+00 3.96381378e-01 2.26384759e-01 -3.12456578e-01
-3.68300110e-01 1.03967226e+00 3.42394859e-01 -5.73803484e-01
3.34478259e-01 -2.63633877e-01 -1.72788188e-01 1.54757559e-01
2.52230048e-01 1.07196264e-01 2.24757656e-01 -6.42842829e-01
-3.70724320e-01 6.22774720e-01 2.59264797e-01 6.19956814e-02
1.34435976e+00 -2.91700028e-02 -1.99271068e-01 2.77312994e-01
1.08824611e+00 -2.66453177e-01 -1.45251751e+00 -1.56688437e-01
3.96607727e-01 -2.97798663e-01 -4.39526513e-02 -9.59004641e-01
-1.01494360e+00 1.19168866e+00 5.09771466e-01 3.73221070e-01
8.06434035e-01 5.32588303e-01 5.08770049e-01 1.52805030e-01
3.73192698e-01 -1.36783838e+00 -3.40106525e-02 4.21030313e-01
2.27491930e-01 -1.55209291e+00 -1.43522635e-01 -8.85908306e-01
-6.15195394e-01 8.07370961e-01 6.93539977e-01 -1.94009781e-01
2.45840013e-01 3.13449711e-01 8.07100907e-02 -3.56442034e-01
-3.49871278e-01 -8.67873669e-01 5.40768206e-01 7.69126892e-01
6.90558612e-01 1.98972464e-01 -1.84882104e-01 4.75991070e-01
-2.96158455e-02 -2.52270490e-01 1.59468532e-01 5.47742248e-01
-7.64183462e-01 -1.00728858e+00 -3.25990826e-01 2.91944623e-01
-5.94951093e-01 -1.35535613e-01 -4.86322373e-01 9.22899961e-01
6.15601003e-01 8.42208922e-01 2.81536132e-01 7.55094737e-02
3.07412148e-01 1.84892461e-01 3.76853138e-01 -1.05040038e+00
-7.42950022e-01 3.02929372e-01 3.38628531e-01 -5.76449513e-01
-8.00805449e-01 -6.68362677e-01 -1.36414230e+00 2.47751862e-01
-3.99261713e-01 2.54795998e-01 7.36375868e-01 1.21230149e+00
-2.03223333e-01 4.86230344e-01 1.41050324e-01 -7.38660514e-01
4.46916849e-04 -8.15869987e-01 -7.30727077e-01 8.54717970e-01
2.99030766e-02 -1.48663253e-01 -4.71651733e-01 3.45676810e-01] | [9.492303848266602, 0.5123147368431091] |
7f6b62e2-3b5e-4106-97b9-4257904ea8f9 | entity-relative-position-representation-based | null | null | https://aclanthology.org/2020.ccl-1.89 | https://aclanthology.org/2020.ccl-1.89.pdf | Entity Relative Position Representation based Multi-head Selection for Joint Entity and Relation Extraction | Joint entity and relation extraction has received increasing interests recently, due to the capability of utilizing the interactions between both steps. Among existing studies, the Multi-Head Selection (MHS) framework is efficient in extracting entities and relations simultaneously. However, the method is weak for its limited performance. In this paper, we propose several effective insights to address this problem. First, we propose an entity-specific Relative Position Representation (eRPR) to allow the model to fully leverage the distance information between entities and context tokens. Second, we introduce an auxiliary Global Relation Classification (GRC) to enhance the learning of local contextual features. Moreover, we improve the semantic representation by adopting a pre-trained language model BERT as the feature encoder. Finally, these new keypoints are closely integrated with the multi-head selection framework and optimized jointly. Extensive experiments on two benchmark datasets demonstrate that our approach overwhelmingly outperforms previous works in terms of all evaluation metrics, achieving significant improvements for relation F1 by +2.40% on CoNLL04 and +1.90% on ACE05, respectively. | ['Zhoujun Li', 'Yunbo Cao', 'Zhao Yan', 'Tianyang Zhao'] | null | null | null | null | ccl-2020-10 | ['relation-classification', 'joint-entity-and-relation-extraction'] | ['natural-language-processing', 'natural-language-processing'] | [ 7.35907406e-02 1.78723693e-01 -5.74212849e-01 -5.82516372e-01
-9.54794705e-01 -3.95662606e-01 7.11319745e-01 4.79983002e-01
-4.84093755e-01 6.96410418e-01 3.57908607e-01 -1.91524044e-01
-1.36181846e-01 -8.54920506e-01 -6.42623961e-01 -5.17332733e-01
-6.39821813e-02 7.93068856e-02 3.21481258e-01 -2.20067471e-01
1.01283610e-01 3.22493494e-01 -1.19659698e+00 6.95593730e-02
1.11217403e+00 1.03438795e+00 2.82506645e-01 8.31453502e-02
-1.25558255e-02 8.37569833e-01 -3.25159490e-01 -5.74216187e-01
-1.24107450e-01 -1.59267247e-01 -9.26002383e-01 -2.59929329e-01
-1.25045031e-01 -1.30134165e-01 -3.57543379e-01 8.62358093e-01
4.47064072e-01 2.21657008e-01 3.73468369e-01 -9.99327779e-01
-6.49658740e-01 9.39555585e-01 -8.09143901e-01 7.91139305e-02
4.56972003e-01 -4.79152352e-01 1.67459345e+00 -1.24199450e+00
6.00283027e-01 1.09231150e+00 4.28517967e-01 -5.53141460e-02
-8.35750818e-01 -6.66169882e-01 3.94611359e-01 6.16861880e-01
-1.76444018e+00 -4.87079799e-01 7.44618416e-01 -3.47394720e-02
1.38862634e+00 2.72102863e-01 1.85017914e-01 7.46819079e-01
-1.50666922e-01 1.02977538e+00 7.32242823e-01 -6.81476116e-01
-1.70867145e-01 5.31187132e-02 2.71862686e-01 6.14927828e-01
3.29422385e-01 -1.37104854e-01 -5.35079837e-01 -9.22205374e-02
4.25442129e-01 -9.79486182e-02 -3.00279975e-01 -1.04957931e-01
-9.70470786e-01 6.98909461e-01 5.62615275e-01 4.37872171e-01
-4.77771938e-01 -8.90270621e-02 3.63944381e-01 -7.92413801e-02
6.25429571e-01 5.50481379e-01 -6.59553587e-01 -4.38639969e-02
-4.74596709e-01 1.76868320e-01 6.89881325e-01 1.20694470e+00
8.16903055e-01 -5.29545724e-01 -4.03625548e-01 1.18069482e+00
4.45920706e-01 2.80140072e-01 2.96008468e-01 -5.16054749e-01
9.30931091e-01 9.70535994e-01 -9.83179510e-02 -1.30449986e+00
-4.47347552e-01 -7.79789150e-01 -7.78528035e-01 -6.45772994e-01
-9.93300080e-02 -6.09487295e-02 -5.30355394e-01 1.77741039e+00
5.21238923e-01 4.64142114e-01 3.79193932e-01 6.22199178e-01
9.92255032e-01 6.27860963e-01 3.67111355e-01 -2.16322556e-01
1.52384949e+00 -1.30228555e+00 -9.12558734e-01 -3.48232716e-01
9.44659710e-01 -6.72719300e-01 6.19439065e-01 -1.49571300e-01
-8.72528434e-01 -4.40124243e-01 -1.11939907e+00 -2.60455728e-01
-5.32255590e-01 4.84349281e-01 8.69786561e-01 2.68883079e-01
-4.85194564e-01 4.31361824e-01 -8.86045098e-01 -2.44307145e-01
3.62132341e-01 2.46241018e-01 -4.39975351e-01 -2.05239266e-01
-1.61514306e+00 9.78888333e-01 7.29061842e-01 2.75812060e-01
-9.55535620e-02 -6.15831852e-01 -1.22740579e+00 3.30730259e-01
8.20175409e-01 -4.81887549e-01 9.16637719e-01 -7.93715492e-02
-1.20758224e+00 4.82539952e-01 -5.72053611e-01 -4.63539094e-01
8.94495845e-02 -5.71340203e-01 -5.59065521e-01 1.49907291e-01
3.00162733e-01 4.41426545e-01 8.15827176e-02 -1.25619650e+00
-9.74855423e-01 -3.42443228e-01 1.24402694e-01 5.06208777e-01
-4.73921984e-01 2.01771393e-01 -9.50621307e-01 -7.91113913e-01
1.36559471e-01 -8.37351382e-01 -1.99934974e-01 -7.00832665e-01
-7.32855916e-01 -6.21915698e-01 6.23208225e-01 -7.40297377e-01
1.66427755e+00 -2.01293302e+00 1.43673465e-01 3.00116211e-01
1.84238508e-01 4.75532174e-01 -7.16476887e-02 5.77922285e-01
-1.04034863e-01 2.13053286e-01 -1.03928939e-01 -5.87691545e-01
-1.98170674e-04 1.06098272e-01 -1.53197870e-01 2.22155806e-02
5.04597902e-01 1.37502658e+00 -7.96260655e-01 -6.52837634e-01
-2.16102600e-02 5.79615712e-01 -3.53975058e-01 1.97337404e-01
2.93731749e-01 9.03377607e-02 -7.35132873e-01 6.16012096e-01
6.73918426e-01 -4.90392655e-01 6.25800610e-01 -3.03098321e-01
8.11268315e-02 6.64424241e-01 -1.08335721e+00 1.54982817e+00
-5.80073178e-01 2.06571609e-01 -3.66681218e-01 -1.05803442e+00
8.98644805e-01 4.13331687e-01 4.37407285e-01 -6.99758708e-01
-6.23858757e-02 2.97172725e-01 -1.40708700e-01 -3.67970049e-01
6.34849846e-01 2.55979896e-01 -1.33505091e-01 6.77468702e-02
1.04528442e-01 4.21846330e-01 1.67165071e-01 3.82873356e-01
1.18420029e+00 1.50010586e-01 8.14280391e-01 3.17077264e-02
7.45794833e-01 -4.44213927e-01 8.75974298e-01 4.31836635e-01
1.32332677e-02 3.75104159e-01 4.76644754e-01 2.91730314e-02
-3.48961323e-01 -5.52799940e-01 3.06553463e-03 1.02467656e+00
4.51041222e-01 -9.56065297e-01 -4.22172278e-01 -1.02200437e+00
-5.57814501e-02 7.46878684e-01 -6.01547003e-01 -3.06817114e-01
-8.12215447e-01 -8.66002381e-01 4.23642814e-01 9.52156007e-01
5.53911448e-01 -8.94306898e-01 -2.44124532e-01 2.79900312e-01
-4.52071935e-01 -1.66141021e+00 -2.35936940e-01 2.71929026e-01
-5.66795707e-01 -9.25941944e-01 -3.76774490e-01 -8.45518351e-01
3.69829893e-01 3.80663902e-01 9.40421999e-01 1.64949015e-01
9.05302241e-02 -1.13497235e-01 -8.41221035e-01 -1.02257341e-01
2.56198943e-01 6.85177147e-01 -3.27032179e-01 -3.53304967e-02
6.19132936e-01 -5.49426913e-01 -3.46695065e-01 1.42729998e-01
-5.55179417e-01 1.79409117e-01 9.27063107e-01 8.91166627e-01
5.92849255e-01 2.41285667e-01 9.09112871e-01 -1.34316528e+00
3.80007416e-01 -6.67106688e-01 -1.03076302e-01 6.20649874e-01
-7.95657158e-01 1.08559718e-02 4.35045928e-01 -1.41703980e-02
-1.38638783e+00 -1.28137851e-02 -9.94845852e-02 3.46462950e-02
3.43922079e-02 9.79536414e-01 -6.75062478e-01 2.61155576e-01
4.60213386e-02 8.72752350e-03 -7.01066434e-01 -5.57891786e-01
4.08780605e-01 7.06810713e-01 4.46836770e-01 -6.59277141e-01
8.44250679e-01 1.79770336e-01 -1.57985687e-01 -5.23262441e-01
-1.22629571e+00 -7.06526518e-01 -7.46585250e-01 4.15985376e-01
6.74311638e-01 -1.19822562e+00 -4.76796567e-01 2.43478864e-01
-1.24041533e+00 1.18414335e-01 -4.49638441e-02 5.15372574e-01
-1.42315438e-03 3.18545938e-01 -6.45782590e-01 -6.60989821e-01
-3.06095302e-01 -1.03691697e+00 1.23361063e+00 4.40550923e-01
-6.99734539e-02 -9.88645911e-01 -2.49352708e-01 4.48541403e-01
2.61612862e-01 2.54448444e-01 8.74673307e-01 -9.50568676e-01
-6.42357111e-01 -3.63934100e-01 -6.33297563e-01 -1.25392321e-02
4.04025763e-01 -2.57887751e-01 -8.75853360e-01 -6.41770428e-03
-3.68532836e-01 -1.28668532e-01 1.03186917e+00 -7.95108080e-02
1.07528949e+00 -6.12925701e-02 -8.77689242e-01 4.88998055e-01
1.30077028e+00 2.42799476e-01 5.39005756e-01 3.31659883e-01
1.02907777e+00 6.69545531e-01 1.04631305e+00 3.99533391e-01
1.06810009e+00 1.00935864e+00 1.72956601e-01 -7.72779807e-02
-1.58045530e-01 -4.05931622e-01 -6.35933056e-02 9.85690713e-01
-2.51313627e-01 -2.38936156e-01 -8.30130160e-01 5.61079383e-01
-2.10360932e+00 -6.51502967e-01 -9.17661563e-03 1.78957665e+00
9.60750520e-01 1.33258134e-01 -1.77810594e-01 7.25255460e-02
6.75575018e-01 2.70121396e-01 -2.84511715e-01 1.80304110e-01
-2.40616724e-01 3.57356429e-01 3.81929725e-01 5.27496219e-01
-1.40904713e+00 1.24126816e+00 4.95511389e+00 1.01493979e+00
-6.60676897e-01 3.00888903e-02 4.75592792e-01 3.18900943e-01
-2.36986548e-01 2.85339117e-01 -1.23974526e+00 2.71177441e-01
7.54368424e-01 -1.65016145e-01 4.18250002e-02 6.54565096e-01
-1.23601094e-01 3.11194044e-02 -9.47113931e-01 8.02446544e-01
8.95399228e-02 -1.01347244e+00 -2.18962878e-01 1.78576022e-01
5.36645353e-01 -2.18147874e-01 -1.66922078e-01 6.39222443e-01
2.52034903e-01 -9.14809644e-01 4.02863234e-01 3.34977448e-01
6.33985877e-01 -1.09153557e+00 1.07773960e+00 1.72783569e-01
-1.74154854e+00 8.51749182e-02 -2.28478804e-01 9.32475254e-02
2.03822583e-01 7.55237639e-01 -6.61046147e-01 1.34136403e+00
4.44139749e-01 9.81237233e-01 -7.40785658e-01 6.65048301e-01
-5.76619267e-01 6.08369887e-01 -3.24543387e-01 7.88887218e-02
1.27004534e-01 8.77969787e-02 2.78142452e-01 1.40236723e+00
1.26023024e-01 3.20283860e-01 2.74054140e-01 4.98468161e-01
-2.62361437e-01 4.25888747e-01 -2.68427372e-01 5.67542426e-02
9.95379388e-01 1.49682045e+00 -5.09732783e-01 -3.44079196e-01
-7.00930119e-01 8.82310510e-01 8.04453135e-01 3.42412680e-01
-8.59644830e-01 -8.42375338e-01 4.48719054e-01 -4.91552085e-01
5.37650228e-01 -1.47588044e-01 -1.56929046e-01 -1.32146347e+00
2.79118836e-01 -5.52693486e-01 4.07924682e-01 -2.23983601e-01
-9.95842755e-01 6.87379301e-01 -1.62869915e-02 -8.84822905e-01
-2.79652923e-01 -2.79474109e-01 -3.40416372e-01 8.31554055e-01
-2.00760913e+00 -1.54144394e+00 -1.96522832e-01 1.13928556e-01
3.17658097e-01 -6.02156818e-02 8.63034844e-01 7.27451026e-01
-1.02637541e+00 1.03496945e+00 -2.10308418e-01 3.86814445e-01
7.28805244e-01 -1.33010221e+00 3.66022676e-01 8.90430152e-01
2.80165792e-01 8.57218027e-01 2.36232147e-01 -6.32432640e-01
-1.21240985e+00 -1.25249851e+00 1.47616887e+00 -2.97375530e-01
4.83052313e-01 -2.87205815e-01 -1.00353706e+00 8.44827712e-01
1.66282713e-01 2.17088431e-01 8.57280612e-01 6.04565799e-01
-5.83993256e-01 -9.04330164e-02 -7.75735259e-01 5.12479722e-01
1.11440945e+00 -4.49479818e-01 -5.52997768e-01 8.10740441e-02
1.04056418e+00 -4.75779951e-01 -1.26329386e+00 8.60963523e-01
3.15643072e-01 -5.85444868e-01 1.00607932e+00 -3.98062676e-01
4.85569954e-01 -2.58586079e-01 -5.13205588e-01 -1.15132928e+00
-3.53465199e-01 -3.83415163e-01 -6.32799745e-01 1.93168402e+00
6.52578115e-01 -7.04731226e-01 5.06067634e-01 4.16989177e-01
-4.63687778e-02 -1.24949050e+00 -4.95703191e-01 -7.19366491e-01
-1.95414484e-01 -3.04903924e-01 8.90360355e-01 1.07261920e+00
2.13378593e-01 9.68863666e-01 -3.05705398e-01 4.13616210e-01
1.70285001e-01 4.92307335e-01 5.64767361e-01 -1.01048541e+00
-3.55172753e-01 -2.06698596e-01 -2.55015224e-01 -1.31424177e+00
4.48156297e-01 -1.01105452e+00 2.58823838e-02 -1.56130111e+00
3.77821892e-01 -7.23465919e-01 -6.22628152e-01 6.65187478e-01
-9.30276871e-01 -9.79893431e-02 1.32493258e-01 2.30395079e-01
-9.35371339e-01 7.76827216e-01 8.87633562e-01 2.82903701e-01
-1.94552764e-01 -4.25038487e-02 -1.17186749e+00 5.66656291e-01
6.74106836e-01 -3.97090226e-01 -3.28182161e-01 -3.85614038e-01
1.78829014e-01 -8.34845454e-02 -1.90806966e-02 -6.60757780e-01
3.37510228e-01 2.31591379e-03 2.18178362e-01 -6.75865650e-01
4.07885909e-01 -7.69716680e-01 -3.72633308e-01 -7.26331398e-02
-4.14672852e-01 -2.50988841e-01 -9.84154791e-02 7.06552923e-01
-5.81982911e-01 -9.93015021e-02 3.10781717e-01 3.35354984e-01
-7.33500898e-01 2.39891797e-01 3.78241658e-01 1.14718236e-01
1.01852632e+00 3.26101065e-01 -2.77143657e-01 -1.89877942e-01
-4.09909934e-01 5.42043805e-01 1.72769129e-02 5.51180303e-01
4.15760040e-01 -1.52106333e+00 -6.55798793e-01 -4.38589007e-02
4.09391880e-01 2.85230756e-01 1.42571792e-01 8.63266885e-01
4.83337790e-02 6.94191813e-01 5.67190766e-01 -1.59903392e-01
-1.26313567e+00 4.73011136e-01 -1.29764035e-01 -9.09676790e-01
-6.09684885e-01 8.75327766e-01 1.31718174e-01 -4.28349882e-01
1.31438464e-01 -2.13941142e-01 -5.50433874e-01 1.17094643e-01
4.87761319e-01 1.98403567e-01 2.30181247e-01 -7.95585275e-01
-6.83749676e-01 4.79814440e-01 -5.46600461e-01 9.30405185e-02
1.43241620e+00 -3.23801965e-01 -8.81342143e-02 1.21054433e-01
1.43749893e+00 3.71386647e-01 -8.43456328e-01 -8.39671016e-01
6.38004065e-01 -2.84176648e-01 -1.32391468e-01 -7.48648882e-01
-1.04272091e+00 6.59399331e-01 -7.32835904e-02 1.65124968e-01
1.19439554e+00 2.48059392e-01 1.07091451e+00 4.05778944e-01
1.62609115e-01 -8.09287488e-01 -2.43141979e-01 6.74775839e-01
5.73882699e-01 -1.31012869e+00 1.52356088e-01 -1.13000321e+00
-7.25476980e-01 7.22225785e-01 6.17243052e-01 -2.49879286e-02
6.57532096e-01 2.01345176e-01 -1.95270374e-01 -2.23189428e-01
-8.70381474e-01 -4.36777651e-01 5.09109437e-01 3.35811675e-01
7.59771943e-01 1.94523260e-01 -5.43559968e-01 1.01228082e+00
-1.02773473e-01 -2.98414081e-01 -1.04624003e-01 1.00564790e+00
-2.34896198e-01 -1.40873468e+00 2.64213324e-01 2.67825782e-01
-6.38120949e-01 -3.49384993e-01 -2.78283626e-01 7.12541521e-01
1.02993332e-01 1.07331824e+00 -2.58359373e-01 -5.22423744e-01
3.99699956e-01 2.08411478e-02 2.26804197e-01 -8.09269488e-01
-3.69496673e-01 7.51160383e-02 3.59959394e-01 -5.20688236e-01
-5.01011729e-01 -6.73795938e-01 -1.39073718e+00 1.49497569e-01
-8.27985168e-01 3.20175320e-01 4.25522029e-01 1.22063828e+00
6.80417657e-01 7.49041021e-01 7.52904117e-01 -3.08818161e-01
-3.01601857e-01 -1.01321638e+00 -4.52651292e-01 2.24285513e-01
4.94653620e-02 -9.27250266e-01 2.89790006e-03 -3.01724404e-01] | [9.275659561157227, 8.686155319213867] |
b0996466-dd76-4142-8be2-be5826184146 | exploration-of-the-search-space-of-gaussian | 2303.05561 | null | https://arxiv.org/abs/2303.05561v1 | https://arxiv.org/pdf/2303.05561v1.pdf | Exploration of the search space of Gaussian graphical models for paired data | We consider the problem of learning a Gaussian graphical model in the case where the observations come from two dependent groups sharing the same variables. We focus on a family of coloured Gaussian graphical models specifically suited for the paired data problem. Commonly, graphical models are ordered by the submodel relationship so that the search space is a lattice, called the model inclusion lattice. We introduce a novel order between models, named the twin order. We show that, embedded with this order, the model space is a lattice that, unlike the model inclusion lattice, is distributive. Furthermore, we provide the relevant rules for the computation of the neighbours of a model. The latter are more efficient than the same operations in the model inclusion lattice, and are then exploited to achieve a more efficient exploration of the search space. These results can be applied to improve the efficiency of both greedy and Bayesian model search procedures. Here we implement a stepwise backward elimination procedure and evaluate its performance by means of simulations. Finally, the procedure is applied to learn a brain network from fMRI data where the two groups correspond to the left and right hemispheres, respectively. | ['Dung Ngoc Nguyen', 'Alberto Roverato'] | 2023-03-09 | null | null | null | null | ['efficient-exploration'] | ['methodology'] | [ 3.78639907e-01 3.08054745e-01 2.31960677e-02 -4.29685563e-01
-2.60432005e-01 -4.76376176e-01 7.64441669e-01 1.51352152e-01
-6.22334719e-01 7.05779910e-01 -6.34170994e-02 -2.90072799e-01
-5.38603961e-01 -7.14006960e-01 -6.22507572e-01 -1.03347683e+00
-4.02651668e-01 8.35166514e-01 1.17357217e-01 2.64539242e-01
3.06687087e-01 5.53292811e-01 -1.47874558e+00 6.30870014e-02
8.81819248e-01 6.80591106e-01 3.40398878e-01 3.68989050e-01
-1.35758044e-02 2.64890224e-01 -2.67299265e-02 -1.60863250e-01
3.44550312e-01 -6.05335593e-01 -7.93944240e-01 1.46199510e-01
-1.31164655e-01 2.89371580e-01 4.17053979e-03 1.13559484e+00
4.00767446e-01 1.42330542e-01 1.00072968e+00 -1.41486549e+00
1.51072040e-01 7.45678008e-01 -5.15068591e-01 -1.28213823e-01
1.13208346e-01 -4.47800994e-01 9.38799679e-01 -8.48679960e-01
7.85679221e-01 1.45497441e+00 3.39719772e-01 2.72924870e-01
-1.80762756e+00 -4.59868550e-01 5.02544641e-02 4.55183417e-01
-1.64752281e+00 -3.72889966e-01 6.44731581e-01 -7.11173296e-01
2.66624033e-01 3.73943955e-01 8.27353537e-01 7.44204402e-01
1.92656383e-01 5.32340825e-01 1.57198477e+00 -8.84337068e-01
7.43901491e-01 9.98844951e-02 4.20718610e-01 8.07898045e-01
2.15067327e-01 2.82810926e-01 -6.12785161e-01 -7.07137525e-01
7.53157139e-01 -2.40576699e-01 -2.89810926e-01 -1.18204272e+00
-9.33091760e-01 1.14687765e+00 4.78447348e-01 4.08444107e-01
-2.87781596e-01 -1.94893759e-02 -1.44944359e-02 1.42353192e-01
2.91884869e-01 2.22364724e-01 3.60470004e-02 3.77133548e-01
-1.05079567e+00 2.40940049e-01 1.14783514e+00 6.28431499e-01
8.26201141e-01 -5.24695396e-01 1.65834710e-01 5.89560151e-01
6.88121498e-01 1.66375399e-01 5.00508510e-02 -7.08191037e-01
1.12013899e-01 1.97679117e-01 1.10242993e-01 -8.14946473e-01
-7.06047118e-01 -5.00035226e-01 -1.03473508e+00 3.01540017e-01
7.03588247e-01 -1.38820270e-02 -7.30973721e-01 1.92867184e+00
3.01552743e-01 2.30348080e-01 -1.55056834e-01 4.91278827e-01
2.58134663e-01 2.49276474e-01 -9.67120528e-02 -5.05640686e-01
1.42547417e+00 -4.80809033e-01 -6.29678071e-01 -8.02746788e-02
6.25443518e-01 -3.66148382e-01 4.37835544e-01 6.26350701e-01
-1.08531988e+00 -2.69904703e-01 -8.53545368e-01 3.46230626e-01
-1.37028337e-01 4.16009836e-02 4.13745761e-01 6.74613357e-01
-1.24535155e+00 7.19611943e-01 -8.87574673e-01 -4.33923066e-01
1.18074529e-01 3.54223132e-01 -5.01918554e-01 -7.31338859e-02
-1.10121000e+00 9.83008683e-01 6.46506488e-01 4.72847641e-01
-7.94154823e-01 -1.83497235e-01 -8.04498851e-01 1.14435842e-02
3.04518461e-01 -6.45786166e-01 7.93761671e-01 -6.23878658e-01
-1.05439579e+00 1.11552572e+00 -3.98930430e-01 -4.62516397e-01
8.34131002e-01 3.49475265e-01 4.08515632e-02 2.41631351e-04
-1.79931834e-01 5.22002399e-01 1.02946591e+00 -1.37126482e+00
-2.75852114e-01 -6.53387249e-01 -3.79737437e-01 5.64842038e-02
-1.04224542e-02 2.33022854e-01 -5.00734031e-01 -4.21097010e-01
6.85769379e-01 -1.13662779e+00 -6.25002742e-01 -7.40115494e-02
-4.74688262e-01 -1.96555972e-01 1.61915317e-01 -6.53062463e-01
1.28365183e+00 -2.18778038e+00 6.16031110e-01 1.02585268e+00
4.46159393e-01 -4.68718529e-01 1.33706689e-01 5.26800334e-01
-3.90470147e-01 -9.81338788e-03 -5.72969079e-01 -3.78936470e-01
1.19200051e-01 3.62942815e-01 2.87007168e-02 8.43219161e-01
-2.69193381e-01 5.42583525e-01 -5.41203916e-01 -5.72816193e-01
-9.57269147e-02 1.78370282e-01 -4.59664553e-01 1.15693539e-01
-9.58122686e-03 4.83387023e-01 -4.14398491e-01 -1.63930029e-01
8.31088901e-01 -1.61765084e-01 5.00091612e-01 -8.80638808e-02
-1.30595341e-01 -3.77961472e-02 -1.75299513e+00 1.29277027e+00
-9.67258215e-02 1.63833618e-01 2.84465820e-01 -1.02182817e+00
1.08299029e+00 2.26908386e-01 2.20257640e-01 -2.76877433e-01
1.08115263e-01 2.33900398e-01 2.91207820e-01 -2.72716045e-01
-1.71889707e-01 -2.87543118e-01 9.70663056e-02 5.83090305e-01
-4.31940163e-04 2.78817639e-02 3.96239758e-01 9.67422947e-02
6.92924380e-01 -1.37076899e-01 6.41532183e-01 -6.83349490e-01
3.98352236e-01 -5.27891457e-01 4.67116684e-01 1.04690182e+00
2.70007879e-01 3.69343638e-01 9.99657452e-01 -2.12820441e-01
-7.36451685e-01 -1.15841484e+00 -4.30765241e-01 7.81714737e-01
-1.96527556e-01 -4.39306319e-01 -7.99508154e-01 -3.75580847e-01
-7.34430850e-02 6.64534628e-01 -9.32088494e-01 -1.13376804e-01
-3.93581212e-01 -8.81623507e-01 3.34906466e-02 2.55936682e-01
1.19387850e-01 -8.90926898e-01 -7.77420819e-01 -1.97962150e-02
-1.32390589e-01 -5.69539487e-01 -5.32247089e-02 6.13992751e-01
-1.05126989e+00 -1.14008248e+00 -6.41509473e-01 -6.67653382e-01
8.69609356e-01 -3.01855862e-01 9.70456123e-01 -1.42566651e-01
-1.97467562e-02 1.84606373e-01 -9.92192104e-02 -2.30314627e-01
-4.25150603e-01 -1.36072993e-01 8.06404799e-02 2.97475785e-01
5.31484671e-02 -5.98050535e-01 -1.10481657e-01 2.47243732e-01
-8.73230100e-01 3.20952207e-01 4.26731378e-01 1.01731277e+00
5.60387909e-01 1.18838005e-01 -1.49626732e-02 -8.78434241e-01
6.28394961e-01 -3.46379697e-01 -9.45439696e-01 3.66017848e-01
-5.24426401e-01 5.33213139e-01 1.69325873e-01 -1.33702964e-01
-7.99219489e-01 3.82239699e-01 2.58291394e-01 -1.74506113e-01
-1.71525016e-01 8.99878204e-01 -4.45484668e-01 -2.81354159e-01
4.57613409e-01 7.86603093e-02 1.18362062e-01 -7.73996532e-01
3.81972671e-01 3.28393221e-01 2.18574524e-01 -3.74165356e-01
5.11390030e-01 3.17831993e-01 5.14453232e-01 -8.06171715e-01
-3.76465991e-02 -2.14178741e-01 -9.57873106e-01 -2.32739672e-01
7.49434471e-01 -4.98052955e-01 -7.10045934e-01 3.32152814e-01
-1.10995042e+00 -1.20205350e-01 -3.56688388e-02 6.73246264e-01
-8.81772161e-01 5.20686150e-01 -4.03872311e-01 -1.01607013e+00
2.92769909e-01 -1.10828745e+00 5.85027874e-01 -1.99118227e-01
-2.82345176e-01 -1.10175419e+00 2.45006949e-01 -1.43017948e-01
-5.80536351e-02 1.29172713e-01 1.51219344e+00 -1.03390229e+00
-3.30938399e-01 -4.74442877e-02 6.19749837e-02 1.18230851e-02
-2.91440040e-01 -2.92851061e-01 -6.73465729e-01 -4.01115090e-01
2.47693732e-01 1.16373159e-01 1.03138232e+00 5.01865923e-01
8.98336828e-01 -1.17662683e-01 -6.47121131e-01 3.23901862e-01
1.24711084e+00 2.81000018e-01 3.04137737e-01 -5.25239296e-03
3.95622909e-01 9.64469612e-01 2.20835522e-01 4.77893174e-01
-3.22388075e-02 8.94587219e-01 4.06502157e-01 -4.18118797e-02
5.41010380e-01 -2.71981247e-02 -4.68458422e-02 7.06988811e-01
-4.46878597e-02 -9.48366076e-02 -1.20601857e+00 2.86332071e-01
-2.05098724e+00 -6.64903462e-01 -3.54547352e-01 2.65266609e+00
8.05390060e-01 5.79438396e-02 1.99981079e-01 8.28915387e-02
9.23773944e-01 -4.10794541e-02 -2.12884113e-01 -1.88216895e-01
-5.31414598e-02 2.58407652e-01 3.38410884e-01 9.75003898e-01
-9.72957075e-01 3.90442789e-01 7.05391455e+00 8.54691327e-01
-5.28348565e-01 3.46063189e-02 5.99057198e-01 2.00202495e-01
-2.57221907e-01 1.16353020e-01 -6.40489936e-01 3.81393969e-01
8.38714123e-01 -1.43861324e-01 3.83936375e-01 6.17752194e-01
3.07468921e-01 -3.99945587e-01 -1.34496272e+00 1.02447867e+00
-1.09582819e-01 -9.16864395e-01 -2.74613500e-01 3.87269050e-01
4.23591822e-01 -3.29390705e-01 -1.07413426e-01 -2.77823240e-01
3.59923005e-01 -1.07410407e+00 7.59486854e-01 7.27140963e-01
3.81007850e-01 -7.36188114e-01 4.15809900e-01 6.37668192e-01
-9.42379296e-01 2.50898935e-02 -2.59827137e-01 2.68637016e-02
3.51785302e-01 5.80140650e-01 -7.06487656e-01 6.25004709e-01
3.52317333e-01 4.77107227e-01 -5.31082988e-01 1.58193517e+00
-2.24815130e-01 4.33016151e-01 -4.67021495e-01 -2.81789247e-02
7.31269047e-02 -7.76937902e-01 5.93201935e-01 9.40451503e-01
2.67302364e-01 -9.70272720e-02 1.24690542e-02 9.17734444e-01
2.75512159e-01 9.69530120e-02 -5.77017903e-01 3.14582169e-01
2.71722794e-01 9.72691119e-01 -1.11233366e+00 -1.79644808e-01
-1.90775648e-01 6.81155622e-01 3.40296507e-01 3.79612178e-01
-4.51056838e-01 -1.99267473e-02 1.56211495e-01 -4.80153635e-02
3.53275597e-01 -1.52765185e-01 1.15644396e-03 -8.73483241e-01
-1.09370716e-01 -7.78128564e-01 4.57883418e-01 -5.35556853e-01
-1.32894421e+00 6.39725745e-01 5.82853317e-01 -7.44336724e-01
-5.74790776e-01 -6.64649069e-01 -2.66978651e-01 1.16453218e+00
-8.63259971e-01 -7.29306459e-01 4.73692752e-02 6.35854185e-01
-2.03061908e-01 1.47835925e-01 1.03051770e+00 -1.01594418e-01
-2.98159420e-01 6.25803843e-02 4.45578814e-01 -1.81317732e-01
3.37350994e-01 -1.34231019e+00 1.08387552e-01 7.75941312e-01
2.69685894e-01 6.99022412e-01 9.11915720e-01 -6.55141711e-01
-8.10695231e-01 -6.77008092e-01 1.11730659e+00 -8.12771693e-02
5.74823022e-01 -9.59487975e-01 -8.09148669e-01 6.63679540e-01
9.38072279e-02 -2.05382213e-01 7.32239842e-01 3.03366631e-01
-1.50554389e-01 1.18886039e-01 -9.41761792e-01 5.46894610e-01
9.98379886e-01 -3.71522248e-01 -6.15882993e-01 3.97595525e-01
5.78532964e-02 4.76450473e-02 -6.42189264e-01 2.67653823e-01
5.43506742e-01 -1.02113128e+00 6.22322977e-01 -7.99882770e-01
-2.05156028e-01 -3.88301790e-01 -5.93129769e-02 -1.49043989e+00
-5.22135258e-01 -6.01865053e-01 -3.23443040e-02 8.12749505e-01
3.28578740e-01 -6.91798031e-01 7.00379670e-01 5.57076693e-01
3.73891920e-01 -6.91413343e-01 -1.32947457e+00 -6.39025629e-01
-2.52335556e-02 -2.78149843e-01 3.09333980e-01 6.07415557e-01
-2.67253369e-02 3.09052140e-01 -4.46508676e-01 -1.04099855e-01
8.62630010e-01 1.40158996e-01 2.24109292e-01 -1.68428743e+00
-5.71323872e-01 -4.20913875e-01 -5.26165485e-01 -8.10043693e-01
2.60742515e-01 -8.81420612e-01 1.61369801e-01 -1.25005913e+00
4.05554414e-01 -4.60321695e-01 -2.42001653e-01 3.38493526e-01
1.42033100e-01 -5.70075251e-02 3.75412762e-01 9.28779393e-02
-3.60996664e-01 4.42661583e-01 8.35090399e-01 3.55695009e-01
-1.10873714e-01 3.96458715e-01 -4.38583851e-01 7.54806399e-01
5.52562475e-01 -5.58700502e-01 -2.23263860e-01 7.01469555e-02
3.77812743e-01 1.59202546e-01 5.04941285e-01 -7.87454307e-01
3.42462152e-01 1.46002680e-01 3.04442257e-01 -4.65620428e-01
2.97748536e-01 -9.47252333e-01 7.79850245e-01 6.70296013e-01
-5.09476483e-01 3.31427306e-02 -1.98126376e-01 6.28897548e-01
-1.81099456e-02 -7.34351516e-01 9.08503115e-01 -1.08051695e-01
-3.49437296e-01 8.48527774e-02 -3.76947433e-01 -2.82620370e-01
9.55390573e-01 -1.11776970e-01 3.71240556e-01 -4.05794710e-01
-1.64904213e+00 1.41662911e-01 2.99708247e-01 -1.64836515e-02
4.72755462e-01 -1.31759894e+00 -6.27514541e-01 4.56136942e-01
7.64362048e-03 -3.69832069e-01 1.02016978e-01 1.29789817e+00
-2.01595664e-01 3.93267095e-01 -2.36953273e-01 -6.35557652e-01
-1.59626281e+00 6.51711106e-01 4.42155510e-01 -4.24941748e-01
-2.34890074e-01 6.14943027e-01 5.23682177e-01 -4.09364253e-01
3.06804240e-01 -1.56513810e-01 -3.62759531e-01 3.43294978e-01
5.37669182e-01 5.65492928e-01 1.55226991e-01 -5.23204148e-01
-3.57242793e-01 3.85354906e-01 1.04740061e-01 -4.65415776e-01
1.24983287e+00 -4.87758555e-02 -7.83440530e-01 6.91593468e-01
1.03975213e+00 -1.87528253e-01 -1.11637342e+00 -3.66874039e-01
3.41785938e-01 -3.08120996e-01 -1.94608092e-01 -4.52490836e-01
-7.12490797e-01 7.40688145e-01 6.54072046e-01 4.59961087e-01
1.00548053e+00 2.00628221e-01 -5.89108646e-01 1.71354905e-01
4.46931034e-01 -6.70643985e-01 -5.40726721e-01 3.52512509e-01
9.26243544e-01 -5.94639897e-01 -2.25308284e-01 -5.97242415e-01
-2.75339693e-01 1.19124806e+00 -3.27725597e-02 -2.42571644e-02
9.23968554e-01 1.54993311e-01 -4.10646200e-01 -1.07917555e-01
-6.12490416e-01 -2.13536724e-01 3.22582275e-01 5.85794508e-01
2.53151447e-01 1.85349599e-01 -6.95664346e-01 5.88823795e-01
-1.07500598e-01 -9.37701389e-02 1.96201578e-01 7.41320014e-01
-3.54392707e-01 -1.29949152e+00 -6.46983445e-01 5.49425662e-01
-2.89987754e-02 -1.23977475e-01 -3.68885815e-01 8.37944269e-01
-1.53278727e-02 8.45505953e-01 5.71670197e-02 7.23519772e-02
1.17311753e-01 3.15754652e-01 7.70554125e-01 -4.31065857e-01
-1.09896734e-01 3.09265971e-01 -1.89458542e-02 -4.73897398e-01
-4.35392439e-01 -9.81556833e-01 -7.71762609e-01 -9.92187709e-02
-4.31118071e-01 6.75606370e-01 5.23464859e-01 1.08758879e+00
4.38470952e-02 8.17899182e-02 5.13508022e-01 -8.89860630e-01
-6.63326919e-01 -9.72826302e-01 -1.13068104e+00 9.92686022e-03
1.20857157e-01 -8.47105145e-01 -2.66958088e-01 -1.49566621e-01] | [7.091087818145752, 4.911638259887695] |
0c2a1735-8a51-4fe6-afd8-7b33963d5002 | discriminative-feature-alignment | 2006.12770 | null | https://arxiv.org/abs/2006.12770v5 | https://arxiv.org/pdf/2006.12770v5.pdf | Discriminative Feature Alignment: Improving Transferability of Unsupervised Domain Adaptation by Gaussian-guided Latent Alignment | In this study, we focus on the unsupervised domain adaptation problem where an approximate inference model is to be learned from a labeled data domain and expected to generalize well to an unlabeled data domain. The success of unsupervised domain adaptation largely relies on the cross-domain feature alignment. Previous work has attempted to directly align latent features by the classifier-induced discrepancies. Nevertheless, a common feature space cannot always be learned via this direct feature alignment especially when a large domain gap exists. To solve this problem, we introduce a Gaussian-guided latent alignment approach to align the latent feature distributions of the two domains under the guidance of the prior distribution. In such an indirect way, the distributions over the samples from the two domains will be constructed on a common feature space, i.e., the space of the prior, which promotes better feature alignment. To effectively align the target latent distribution with this prior distribution, we also propose a novel unpaired L1-distance by taking advantage of the formulation of the encoder-decoder. The extensive evaluations on nine benchmark datasets validate the superior knowledge transferability through outperforming state-of-the-art methods and the versatility of the proposed method by improving the existing work significantly. | ['Jing Wang', 'Jianzhe Lin', 'Clarence W. de Silva', 'Leonid Sigal', 'Jiahong Chen'] | 2020-06-23 | null | null | null | null | ['traffic-sign-recognition'] | ['computer-vision'] | [ 2.65767485e-01 -5.23499213e-02 -5.37055850e-01 -7.43720770e-01
-7.11213946e-01 -4.36622113e-01 5.96570671e-01 -1.68786839e-01
-1.00098372e-01 9.74184155e-01 2.92749733e-01 2.98073500e-01
-2.75690466e-01 -5.86178422e-01 -6.81340098e-01 -8.05119812e-01
6.27041399e-01 5.17098725e-01 3.46428938e-02 9.93276909e-02
1.34581283e-01 -1.88355982e-01 -1.18825841e+00 6.91073909e-02
1.30285728e+00 8.01161706e-01 3.93483013e-01 -2.68778801e-01
-2.88327485e-01 1.70847133e-01 -3.34729016e-01 -3.79488379e-01
2.92402446e-01 -6.56047463e-01 -6.21471405e-01 3.36702853e-01
2.04682782e-01 -1.58491805e-01 -7.82897696e-02 1.42921019e+00
2.51747787e-01 2.19432674e-02 1.03007412e+00 -1.27437866e+00
-8.23724449e-01 4.18461621e-01 -4.27555114e-01 -1.86563104e-01
2.39164054e-01 -1.66010991e-01 1.09896135e+00 -7.93263495e-01
5.90154529e-01 1.08947861e+00 4.07534063e-01 2.75782973e-01
-1.39224052e+00 -8.10008466e-01 3.98921490e-01 1.85759768e-01
-1.70058215e+00 -2.44462729e-01 9.41526413e-01 -5.63364387e-01
4.50918049e-01 -4.92218435e-01 2.36925855e-01 1.20608318e+00
7.30557442e-02 6.91880584e-01 1.12242150e+00 -5.11274576e-01
3.15760672e-01 6.98949158e-01 -1.71790659e-01 3.49730104e-01
2.65699744e-01 9.51508805e-02 -6.55905902e-01 -5.13175875e-02
6.81153417e-01 -5.26693743e-03 -3.15590531e-01 -1.00202930e+00
-1.27583766e+00 9.33070898e-01 1.63861945e-01 2.48561099e-01
-2.73758322e-01 -5.00140190e-01 3.13309342e-01 2.07864925e-01
4.48359400e-01 2.54999101e-01 -5.31044900e-01 1.68318469e-02
-8.12779725e-01 8.89269337e-02 6.43178046e-01 1.34119844e+00
1.07466829e+00 -1.12882838e-01 -1.68000013e-01 8.82988214e-01
6.99885309e-01 4.77848440e-01 5.95809817e-01 -5.34057975e-01
6.80961728e-01 6.50656521e-01 2.05470994e-01 -9.65721428e-01
1.73429519e-01 -6.07460558e-01 -5.97513139e-01 -1.67417228e-01
5.17253339e-01 -7.25522488e-02 -5.77188909e-01 2.15912366e+00
4.06978726e-01 3.30622166e-01 4.68442470e-01 7.28368461e-01
2.67684013e-01 5.40485919e-01 -3.44241457e-03 -2.12021559e-01
1.07785487e+00 -8.55194211e-01 -7.99452782e-01 -4.61623102e-01
4.58530843e-01 -7.78370261e-01 1.00858915e+00 1.96454734e-01
-4.87760156e-01 -7.76779473e-01 -1.29853249e+00 1.89142227e-01
-5.86680025e-02 3.69905263e-01 1.88747779e-01 5.04739583e-01
-3.83459091e-01 3.96438479e-01 -7.40950882e-01 -5.06734967e-01
3.17996919e-01 1.83726564e-01 -4.87615764e-01 -1.96176142e-01
-1.28130805e+00 8.40824246e-01 7.97316730e-01 -1.29853740e-01
-5.55104315e-01 -5.72234333e-01 -8.56147528e-01 -2.09545996e-03
1.83873877e-01 -5.95097840e-01 1.04367363e+00 -1.26648939e+00
-1.70933843e+00 6.90321088e-01 -2.97125041e-01 -2.71713704e-01
3.00312668e-01 -3.04933369e-01 -4.96734113e-01 -8.51896256e-02
3.74241143e-01 5.21084368e-01 8.44358742e-01 -1.21934259e+00
-7.72205055e-01 -3.71974766e-01 -2.81976193e-01 5.52113533e-01
-7.34980762e-01 -4.05251831e-01 -3.75056088e-01 -5.44954240e-01
1.24252148e-01 -8.39779556e-01 8.11279863e-02 -1.11984350e-01
-2.40200430e-01 -1.90526694e-01 6.50535345e-01 -5.73845565e-01
1.15200961e+00 -2.47915769e+00 2.93874383e-01 3.30930144e-01
-1.06353983e-01 3.91270407e-02 -5.50312959e-02 3.67870718e-01
-1.21937945e-01 -3.43476117e-01 -4.75104004e-01 -1.76039532e-01
1.61035702e-01 2.55585760e-01 -6.49700463e-01 6.24326587e-01
2.73680508e-01 4.80747283e-01 -9.14864838e-01 -5.08638740e-01
-4.60629240e-02 2.02424854e-01 -6.00040495e-01 5.40957689e-01
-1.37503371e-01 7.50850916e-01 -6.34393692e-01 2.43237466e-01
8.11089635e-01 -2.77104706e-01 4.55666423e-01 -2.72483438e-01
1.40753806e-01 4.14694875e-01 -1.30090404e+00 2.09469533e+00
-3.03233743e-01 3.32326323e-01 -3.58997554e-01 -1.43849528e+00
1.52721167e+00 3.70980203e-01 4.57511604e-01 -6.05162442e-01
-6.05116300e-02 4.52105016e-01 8.94927047e-03 -2.47811168e-01
1.61302030e-01 -3.70160073e-01 -1.14417985e-01 3.33563536e-01
4.71500009e-01 6.26648143e-02 -1.76976249e-01 -1.05927311e-01
3.33048552e-01 4.00323391e-01 6.29362285e-01 -3.79661053e-01
5.74892044e-01 -1.15872547e-01 8.54354739e-01 4.57171947e-01
-1.21418998e-01 5.21081209e-01 3.75314981e-01 2.26674322e-02
-1.04691887e+00 -1.28841388e+00 -4.73142713e-01 7.29671836e-01
3.94534141e-01 -3.21662068e-01 -7.38993227e-01 -9.21026170e-01
-9.81267616e-02 9.55529630e-01 -3.24215114e-01 -4.21290278e-01
-2.39370659e-01 -4.31122094e-01 2.38468736e-01 5.59985578e-01
6.48628831e-01 -5.08525074e-01 -1.04639925e-01 3.30608100e-01
-2.16937602e-01 -1.14072502e+00 -5.39781868e-01 2.58349717e-01
-8.38045359e-01 -7.23763406e-01 -6.50509477e-01 -1.01909029e+00
8.70724380e-01 4.11367640e-02 7.89379358e-01 -5.27064741e-01
3.92828345e-01 2.00422630e-01 -4.28734571e-01 -1.79384619e-01
-3.43924493e-01 1.98152959e-01 3.64352912e-01 2.67938316e-01
9.16715980e-01 -7.33892441e-01 -2.88392127e-01 5.69965601e-01
-8.25026572e-01 2.12081194e-01 7.19145477e-01 1.12353849e+00
6.50833666e-01 1.40802994e-01 8.26267123e-01 -8.51379514e-01
4.84620392e-01 -7.44253278e-01 -5.95527053e-01 4.79325414e-01
-7.19038844e-01 4.76451308e-01 8.07617843e-01 -6.10835135e-01
-1.37571967e+00 2.68659443e-01 2.65357882e-01 -3.05739701e-01
-3.22040617e-01 6.01138294e-01 -7.38429964e-01 3.87288213e-01
5.16631782e-01 5.21162570e-01 9.57621560e-02 -4.57464784e-01
5.83522260e-01 1.00626671e+00 4.77903664e-01 -8.82821858e-01
8.75351787e-01 3.25911134e-01 -3.96780998e-01 -3.97756487e-01
-8.43070865e-01 -3.94292712e-01 -9.54067230e-01 2.85446733e-01
7.23749340e-01 -1.00044584e+00 1.29385442e-01 2.60792524e-01
-1.15497899e+00 9.94628593e-02 -2.80763775e-01 9.58393991e-01
-6.84649587e-01 5.35175860e-01 1.31422222e-01 -5.03323853e-01
1.33254439e-01 -1.22685623e+00 8.90025258e-01 2.67799675e-01
-2.78506458e-01 -1.22715104e+00 2.78300494e-01 1.54339324e-03
2.43748903e-01 -1.78298950e-01 1.00741124e+00 -1.14050782e+00
-3.60133022e-01 2.99883280e-02 -2.58960605e-01 5.75077176e-01
6.37406290e-01 -3.02959114e-01 -9.04022098e-01 -3.83923709e-01
2.37206757e-01 -2.56398052e-01 4.70079035e-01 1.64106175e-01
7.05639780e-01 -1.28401026e-01 -5.09191275e-01 5.19485235e-01
1.35074115e+00 2.61587530e-01 4.06319350e-01 2.49008596e-01
4.21359122e-01 7.38065898e-01 8.68978739e-01 4.41879928e-01
3.27902108e-01 9.03679013e-01 -8.96823183e-02 1.41689911e-01
9.88624319e-02 -6.31756365e-01 4.56380457e-01 9.65837717e-01
4.50870305e-01 -1.13054730e-01 -7.87503779e-01 7.08738565e-01
-1.92349136e+00 -5.88970840e-01 3.65821332e-01 2.32205486e+00
1.21265686e+00 1.95424616e-01 -1.31940648e-01 -3.10907185e-01
9.48748648e-01 -1.67298481e-01 -6.86351895e-01 4.10540290e-02
7.18858689e-02 -6.40468523e-02 1.65123850e-01 3.98436695e-01
-1.06083524e+00 8.11818719e-01 5.85512447e+00 1.01095557e+00
-1.09625590e+00 -9.15950462e-02 2.52257288e-01 3.16536218e-01
-3.99109662e-01 3.05588692e-01 -1.13994622e+00 6.95207059e-01
6.76804602e-01 -5.10161638e-01 2.62851804e-01 1.04268980e+00
-9.86672714e-02 1.73021734e-01 -1.48956454e+00 9.48628664e-01
-4.63566277e-03 -8.48545730e-01 1.51352018e-01 1.78963378e-01
9.30777609e-01 -3.70143443e-01 1.94579154e-01 4.09258991e-01
1.85541973e-01 -8.30122888e-01 6.08319938e-01 5.97116768e-01
9.24899638e-01 -6.01878583e-01 7.35211074e-01 4.41905856e-01
-1.14060414e+00 1.44041687e-01 -6.33063734e-01 3.04847304e-02
4.15726677e-02 6.65271997e-01 -1.03799605e+00 7.81577826e-01
3.11637759e-01 1.08211458e+00 -3.05889010e-01 8.67549300e-01
-3.97546947e-01 5.99004030e-01 -1.87180966e-01 2.48714671e-01
-1.27704469e-02 -4.34367627e-01 4.53503460e-01 1.01522493e+00
5.47643542e-01 -2.95590878e-01 3.70373994e-01 1.15698314e+00
1.47180468e-01 2.74168342e-01 -7.99912035e-01 -1.45157024e-01
7.35768497e-01 8.56230557e-01 -3.13592643e-01 -3.02405983e-01
-8.36941063e-01 9.91670251e-01 3.87351543e-01 4.28313851e-01
-9.08665717e-01 -3.88222337e-01 6.51904941e-01 -1.88990816e-01
5.26352584e-01 -1.45574957e-01 -2.37288743e-01 -1.32656729e+00
1.74457967e-01 -8.41217995e-01 2.19560146e-01 -3.90083581e-01
-1.84379303e+00 5.81656814e-01 3.01046550e-01 -1.81854832e+00
-4.24979240e-01 -4.88075674e-01 -3.69768351e-01 1.07928157e+00
-1.58064306e+00 -9.42712128e-01 -1.05252996e-01 7.29940593e-01
3.53964865e-01 -6.25851393e-01 9.40413058e-01 2.15572983e-01
-2.91443825e-01 8.86958241e-01 6.58928454e-01 3.91894355e-02
1.29887605e+00 -1.08879995e+00 7.02443626e-03 7.50710547e-01
1.32970646e-01 8.05641711e-01 5.97143054e-01 -7.10429132e-01
-9.05792058e-01 -1.06596136e+00 6.48083389e-01 -3.00027788e-01
5.74732482e-01 -4.10813481e-01 -1.15680230e+00 7.25209236e-01
3.46441716e-01 -2.12448537e-01 9.64123666e-01 1.88135788e-01
-5.70188522e-01 -3.52707595e-01 -1.00152433e+00 3.70214283e-01
8.24292600e-01 -5.63275158e-01 -9.96739626e-01 1.16707139e-01
5.96407950e-01 -1.96555808e-01 -8.76691580e-01 2.55087316e-01
5.52940249e-01 -5.95117509e-01 6.25229418e-01 -5.61293602e-01
5.04741907e-01 -4.84221965e-01 -4.87647325e-01 -1.54666841e+00
-3.41223538e-01 -2.21165016e-01 1.18904360e-01 1.68458855e+00
2.98341662e-01 -7.44289041e-01 5.00096381e-01 3.80000889e-01
5.21764010e-02 -5.05298734e-01 -8.72823000e-01 -9.59201157e-01
2.05591172e-01 -1.22654289e-01 7.15040743e-01 9.89310682e-01
8.12329724e-02 5.50825953e-01 -3.20094436e-01 3.44667166e-01
5.67360461e-01 2.91582942e-01 8.54458869e-01 -1.27826297e+00
-4.22837824e-01 -1.67904064e-01 -4.60244209e-01 -1.54321027e+00
4.14040685e-01 -9.89053667e-01 2.77979016e-01 -1.16493273e+00
4.17851955e-01 -4.01631355e-01 -4.64865834e-01 2.94042319e-01
-1.98895290e-01 -3.62370789e-01 -8.75767171e-02 4.19669211e-01
-5.75194418e-01 1.00583422e+00 1.18645239e+00 1.72832012e-02
-1.98661387e-01 -1.36086047e-01 -9.27533209e-01 6.54064476e-01
7.30819285e-01 -6.72717810e-01 -7.91668713e-01 -4.02089030e-01
-1.41390577e-01 -1.38327762e-01 -5.56962080e-02 -9.12088752e-01
2.01744750e-01 -2.58862913e-01 2.82967448e-01 -3.99837375e-01
1.12560064e-01 -1.07906413e+00 -1.08964695e-03 -2.25594267e-02
-3.68675321e-01 -4.09705609e-01 -3.13216448e-03 8.58938158e-01
-5.96468866e-01 -2.63205707e-01 8.79337132e-01 2.31614530e-01
-7.14414954e-01 1.08316392e-01 3.04260217e-02 1.83550328e-01
1.11232960e+00 -3.96330833e-01 -7.40902722e-02 -2.75250226e-01
-4.17694241e-01 2.00322345e-01 5.15674651e-01 4.81437802e-01
4.39899474e-01 -1.72747505e+00 -7.80374587e-01 5.99209368e-01
4.62537766e-01 5.68263009e-02 5.09561114e-02 8.18752825e-01
2.01879710e-01 4.43850100e-01 -4.83030409e-01 -8.86966825e-01
-7.85079181e-01 4.77779567e-01 1.34099320e-01 -3.64676148e-01
-4.12657619e-01 5.10225594e-01 7.42771208e-01 -7.12925613e-01
1.46864548e-01 -8.50893632e-02 -1.92424610e-01 -8.88142362e-02
3.41656357e-01 -6.22873530e-02 -1.80432543e-01 -6.42722726e-01
-4.05826688e-01 6.56454742e-01 -2.90003508e-01 -1.50746599e-01
1.22719538e+00 -4.22241867e-01 1.22849591e-01 4.53663737e-01
1.31446266e+00 6.13357201e-02 -1.61572945e+00 -8.76631856e-01
7.64825270e-02 -6.01692379e-01 -3.44428182e-01 -5.19342124e-01
-7.66835034e-01 9.41110730e-01 6.82131410e-01 -1.51610583e-01
1.12725091e+00 3.81400511e-02 4.40177172e-01 1.74704865e-01
2.70966381e-01 -1.10422409e+00 2.11807400e-01 4.91950095e-01
6.21664107e-01 -1.17983627e+00 -9.62631404e-02 -4.04917389e-01
-9.19989824e-01 1.22405267e+00 6.93883777e-01 -7.54631534e-02
5.72027981e-01 -1.08671226e-01 -9.46275890e-02 2.03957424e-01
-4.55227673e-01 7.57534951e-02 4.86510426e-01 8.18296373e-01
4.50906128e-01 -1.82686350e-03 -3.31421584e-01 1.09083200e+00
-1.39753312e-01 2.88364962e-02 8.76988694e-02 5.83950162e-01
-3.54604274e-01 -1.47058499e+00 -2.50642508e-01 3.09872236e-02
-3.66661772e-02 7.62534440e-02 -1.63080603e-01 6.31637454e-01
1.51946723e-01 7.41695642e-01 -5.96462078e-02 -3.36842328e-01
1.89348668e-01 3.07604194e-01 3.62102181e-01 -8.38068664e-01
3.07927012e-01 2.31955662e-01 -2.93278247e-01 -2.23521605e-01
-4.57227647e-01 -7.44666398e-01 -1.05898750e+00 2.18411505e-01
-4.89244252e-01 3.15110981e-01 4.64724988e-01 1.25661588e+00
4.50772285e-01 2.72551596e-01 6.86599553e-01 -2.73143888e-01
-1.01787877e+00 -1.00178266e+00 -7.54145920e-01 5.96645534e-01
3.83627079e-02 -1.09482551e+00 -1.70367911e-01 2.58832127e-01] | [10.406590461730957, 3.170940637588501] |
a590099d-495f-4231-888b-538508173f64 | a-neural-state-space-model-approach-to | 2305.16932 | null | https://arxiv.org/abs/2305.16932v1 | https://arxiv.org/pdf/2305.16932v1.pdf | A Neural State-Space Model Approach to Efficient Speech Separation | In this work, we introduce S4M, a new efficient speech separation framework based on neural state-space models (SSM). Motivated by linear time-invariant systems for sequence modeling, our SSM-based approach can efficiently model input signals into a format of linear ordinary differential equations (ODEs) for representation learning. To extend the SSM technique into speech separation tasks, we first decompose the input mixture into multi-scale representations with different resolutions. This mechanism enables S4M to learn globally coherent separation and reconstruction. The experimental results show that S4M performs comparably to other separation backbones in terms of SI-SDRi, while having a much lower model complexity with significantly fewer trainable parameters. In addition, our S4M-tiny model (1.8M parameters) even surpasses attention-based Sepformer (26.0M parameters) in noisy conditions with only 9.2 of multiply-accumulate operation (MACs). | ['Eng Siong Chng', 'Pin-Jui Ku', 'Yuchen Hu', 'Kai Li', 'Chao-Han Huck Yang', 'Chen Chen'] | 2023-05-26 | null | null | null | null | ['speech-separation'] | ['speech'] | [ 1.68167483e-02 -1.39531597e-01 2.97746565e-02 -1.18444398e-01
-1.09039533e+00 -5.10401905e-01 6.20034993e-01 -3.85820657e-01
-3.40117484e-01 4.28270340e-01 1.51964396e-01 -6.99673593e-01
-1.39690816e-01 2.19736382e-01 -5.97079217e-01 -7.55706728e-01
-6.16574250e-02 3.87326509e-01 5.72353937e-02 -1.31689683e-01
-4.59699407e-02 6.15338922e-01 -1.12446463e+00 1.71904817e-01
1.02979422e+00 7.65705347e-01 5.17712533e-01 1.05943978e+00
-1.39246210e-01 7.65621603e-01 -8.28282475e-01 -6.96963118e-03
1.04187362e-01 -4.18455631e-01 -7.31063664e-01 -5.69068454e-02
2.81277955e-01 -9.33057144e-02 -8.85256410e-01 9.35304880e-01
6.07917964e-01 5.47654986e-01 8.75555217e-01 -1.13076520e+00
-7.14513779e-01 7.20058799e-01 -4.31673855e-01 5.51812828e-01
1.62411444e-02 -1.48606300e-02 6.50363982e-01 -1.05598378e+00
6.44769473e-03 1.51877236e+00 7.42030263e-01 7.17584908e-01
-1.45595062e+00 -7.43627250e-01 3.56486976e-01 1.35597438e-01
-1.27255523e+00 -1.06172609e+00 5.27534187e-01 -9.81102884e-02
1.52721977e+00 6.50728703e-01 2.03271434e-01 1.22202933e+00
8.20473582e-02 1.10315239e+00 9.42061245e-01 -3.44133765e-01
3.24631035e-01 -1.73398986e-01 6.77400947e-01 4.31715459e-01
1.71494573e-01 -2.17734426e-01 -8.42185020e-01 -1.15169138e-01
8.08430254e-01 -3.24681066e-02 -4.03255641e-01 1.32199347e-01
-1.18624711e+00 4.34121221e-01 5.11792190e-02 4.31590110e-01
-2.91355938e-01 2.70795017e-01 9.72440317e-02 4.28051919e-01
2.79672742e-01 4.27625149e-01 -5.28221011e-01 -1.97791994e-01
-1.48723471e+00 -2.70211279e-01 7.59725809e-01 9.48009193e-01
2.72416323e-01 9.86046970e-01 -1.04969725e-01 1.03172588e+00
5.34973979e-01 8.73020053e-01 1.02426362e+00 -1.14842057e+00
4.47719902e-01 -8.40663239e-02 -1.06392682e-01 -5.20463288e-01
-4.67051536e-01 -8.97161663e-01 -1.29516268e+00 -4.20533977e-02
1.45148456e-01 -2.24755839e-01 -1.23700118e+00 1.81843376e+00
-3.40751916e-01 8.38999331e-01 5.10463059e-01 7.86327302e-01
6.24751210e-01 1.18009472e+00 -3.30315053e-01 -4.30935144e-01
1.05829263e+00 -1.42637026e+00 -9.54752684e-01 -6.07163608e-01
1.44302532e-01 -6.72303379e-01 5.29860973e-01 5.11175632e-01
-1.55786538e+00 -7.00554729e-01 -1.13369977e+00 8.97143036e-02
-4.53004315e-02 3.38349462e-01 3.88611764e-01 5.30063212e-01
-1.46453702e+00 6.74777746e-01 -1.20382380e+00 -1.69880733e-01
-4.20438945e-02 5.54006159e-01 -1.69625238e-01 3.07161629e-01
-1.03072536e+00 8.50172341e-01 2.31416687e-01 2.11568564e-01
-1.11093175e+00 -6.61347568e-01 -1.01008546e+00 4.10440654e-01
2.68685427e-02 -6.72778726e-01 1.38957548e+00 -5.01895130e-01
-2.04605436e+00 2.97364146e-01 -6.38193011e-01 -7.77775645e-01
-1.97884619e-01 -3.50992382e-01 -6.85637951e-01 1.74424723e-01
-2.12695494e-01 3.83468449e-01 1.34347880e+00 -1.22469032e+00
-1.55489147e-01 4.35913093e-02 -5.08319497e-01 1.28684849e-01
-2.06638336e-01 3.11711758e-01 -5.29565752e-01 -9.03790772e-01
5.05652368e-01 -1.10328913e+00 -5.03507614e-01 -5.96673787e-01
-3.17416757e-01 2.57129312e-01 4.55572695e-01 -1.02171731e+00
1.35879147e+00 -2.21215749e+00 6.53049707e-01 -8.21036845e-03
2.10049823e-01 9.27441001e-01 -6.15538716e-01 4.97661740e-01
-1.73590258e-01 -1.31686836e-01 -4.39121991e-01 -1.15239286e+00
1.73693657e-01 2.81025112e-01 -5.99565208e-01 3.48828375e-01
2.03930363e-01 1.02334094e+00 -4.71197307e-01 -7.56196529e-02
1.91505030e-01 5.06526053e-01 -4.63022530e-01 2.48399645e-01
3.63142759e-01 3.99954557e-01 1.82774186e-01 2.22244963e-01
8.57623696e-01 -1.28401786e-01 3.17076325e-01 3.20756324e-02
-8.91006514e-02 6.37367666e-01 -1.34212053e+00 1.96156549e+00
-7.96722233e-01 9.91559207e-01 7.03680813e-01 -1.15589666e+00
7.72082150e-01 6.80328190e-01 2.95731843e-01 -2.69013405e-01
1.47606343e-01 2.91056514e-01 1.17976524e-01 -3.86866741e-02
3.37040633e-01 6.53743967e-02 5.98850362e-02 2.76499242e-01
6.49297357e-01 -6.63503557e-02 -2.57844757e-02 3.43158096e-01
9.76122320e-01 -3.42226744e-01 2.65969932e-02 -2.88636208e-01
6.26885235e-01 -7.11593270e-01 6.39725566e-01 9.34500039e-01
-1.37879178e-01 6.07000411e-01 4.39960742e-03 2.86504716e-01
-6.71279669e-01 -1.30987799e+00 7.12864771e-02 7.65742838e-01
8.51186663e-02 -5.06261051e-01 -9.26808596e-01 9.40382332e-02
-2.28250459e-01 6.67945027e-01 4.84371372e-02 -4.24591988e-01
-6.13484979e-01 -6.98378563e-01 9.09854412e-01 6.75837815e-01
2.05015153e-01 -7.62631416e-01 -1.15453303e-01 2.88136721e-01
-2.09263325e-01 -1.03039122e+00 -7.87318647e-01 6.79161727e-01
-9.33513045e-01 -1.71811894e-01 -1.04123557e+00 -9.29977715e-01
3.69014531e-01 6.77769721e-01 7.80674517e-01 -4.80154872e-01
1.49478734e-01 4.05297518e-01 3.45037095e-02 -5.81932403e-02
-5.66123486e-01 1.17410995e-01 7.01845765e-01 1.71777710e-01
1.38375551e-01 -9.34004247e-01 -1.37249857e-01 1.40657738e-01
-7.28172362e-01 3.35373282e-02 7.31137753e-01 8.28745246e-01
1.87901214e-01 1.15806065e-01 6.67837203e-01 -1.55998617e-01
7.15481699e-01 -4.39698994e-01 -3.24270427e-01 2.29907513e-01
-5.16157389e-01 4.79837179e-01 6.81602597e-01 -8.70546818e-01
-1.13634670e+00 -1.68090776e-01 -1.29031152e-01 -8.86169255e-01
-1.52993038e-01 3.40172887e-01 -2.14752987e-01 1.74034461e-01
2.23011658e-01 7.62205303e-01 1.57953411e-01 -9.99145508e-01
4.12022769e-01 7.99770594e-01 8.70916545e-01 -4.98073727e-01
9.05157149e-01 4.32769991e-02 -2.16995746e-01 -1.15973222e+00
-5.45742691e-01 -8.23629260e-01 -5.80397964e-01 2.10257709e-01
6.95768893e-01 -1.15489411e+00 -6.88909054e-01 8.42683792e-01
-1.35711980e+00 -3.68314356e-01 -3.69250216e-02 8.93447042e-01
-3.78555298e-01 6.44140661e-01 -1.21588314e+00 -1.22784221e+00
-3.65708083e-01 -1.17437208e+00 1.04267347e+00 1.42260939e-01
-3.15860331e-01 -8.60984504e-01 4.72197980e-02 5.40140092e-01
6.45178258e-01 -7.75268078e-01 5.11945307e-01 -6.93925560e-01
-4.02116805e-01 6.78252056e-02 5.51787391e-02 6.55463278e-01
-7.84759969e-02 -2.85158306e-01 -1.31903231e+00 -5.44161558e-01
4.81718421e-01 3.14498991e-01 1.15677726e+00 6.87088251e-01
7.59103477e-01 -2.41708487e-01 -1.81574762e-01 9.64736938e-01
8.79114211e-01 5.41103542e-01 4.30243522e-01 2.41971686e-02
6.39172018e-01 2.06365511e-01 -6.15817197e-02 1.74927846e-01
1.00653574e-01 6.46522462e-01 -2.96140020e-03 -2.11197138e-01
-4.21954542e-01 -1.92666668e-02 1.14912999e+00 1.93027294e+00
1.03561066e-01 -3.88579786e-01 -6.04889989e-01 2.52412289e-01
-1.85507369e+00 -9.91256416e-01 -1.01709954e-01 1.93471503e+00
6.80551469e-01 4.14164215e-02 -3.80997248e-02 3.20460171e-01
5.81714988e-01 3.41022640e-01 -4.91342306e-01 -5.53097904e-01
-2.43030936e-01 4.15082783e-01 4.48339552e-01 8.90290558e-01
-1.02178085e+00 8.63978326e-01 6.85156727e+00 1.04191697e+00
-1.14429212e+00 8.04231763e-02 -8.70697573e-02 -2.95742661e-01
-1.31222382e-01 -2.19915077e-01 -8.30646396e-01 5.34073949e-01
1.74608290e+00 6.25424907e-02 8.60324919e-01 3.87174040e-01
1.43243521e-01 3.58106971e-01 -8.94070864e-01 1.39426911e+00
1.53303534e-01 -1.04070592e+00 -4.19288203e-02 -1.28898531e-01
4.51725990e-01 1.54792026e-01 2.53774852e-01 5.28598666e-01
1.07227646e-01 -1.03545201e+00 7.75874317e-01 5.69781065e-01
5.41994691e-01 -5.85224748e-01 2.85259783e-01 6.82033777e-01
-1.28633678e+00 -1.72545686e-01 -3.23874682e-01 -1.48486525e-01
4.11258340e-01 5.25950134e-01 -5.35143912e-01 5.25457680e-01
4.08020645e-01 6.62813425e-01 -1.56072989e-01 6.67602539e-01
1.86588019e-02 1.12695992e+00 -2.91924745e-01 3.94419283e-01
2.00855464e-01 -4.11267936e-01 8.77011061e-01 1.71051455e+00
4.67224687e-01 -1.29774697e-02 -3.44404988e-02 6.32988572e-01
1.55475557e-01 -4.68568861e-01 -1.13614425e-01 -1.21557534e-01
4.82921928e-01 1.08834445e+00 -3.50181311e-01 -3.94247890e-01
-3.42037112e-01 1.47798288e+00 1.51683241e-01 7.69633949e-01
-1.02075088e+00 -2.72049904e-01 1.16279638e+00 -3.78146678e-01
4.40906316e-01 -7.80337572e-01 -2.12791543e-02 -1.63212550e+00
-1.90208241e-01 -1.09916317e+00 -1.48736611e-01 -7.50429511e-01
-9.91328835e-01 9.33605552e-01 -2.39656493e-01 -1.16294456e+00
-3.65331113e-01 -6.92832828e-01 -5.97970128e-01 1.07784593e+00
-1.49841988e+00 -9.49301600e-01 2.87778854e-01 3.91859621e-01
8.66442382e-01 -5.03230691e-01 9.58826423e-01 4.75581050e-01
-1.25830019e+00 6.58182085e-01 6.01993144e-01 -1.80943999e-02
5.21397114e-01 -1.32765532e+00 1.03100681e+00 1.28068709e+00
6.32717133e-01 1.05061626e+00 5.77931941e-01 -2.10419908e-01
-1.42923796e+00 -8.52031887e-01 7.68913746e-01 -3.96361232e-01
6.62806392e-01 -4.90573078e-01 -1.23836219e+00 6.39433742e-01
4.27766830e-01 -3.59320283e-01 7.21899807e-01 2.75666751e-02
-4.37058270e-01 3.96800786e-02 -5.23862660e-01 6.52790427e-01
1.01424563e+00 -7.21159041e-01 -8.41650784e-01 -1.49420410e-01
8.04053783e-01 -3.36292863e-01 -6.80347562e-01 -4.02707271e-02
2.34643504e-01 -6.89104497e-01 1.38555384e+00 -7.33506203e-01
-2.39280432e-01 -4.45598811e-01 -4.08563793e-01 -1.61603677e+00
-9.04652834e-01 -1.25400364e+00 -8.48191917e-01 1.31359518e+00
4.95626241e-01 -7.95190334e-01 1.36350974e-01 4.58539546e-01
-5.89349568e-01 -5.23767292e-01 -1.01605248e+00 -1.34256124e+00
1.39167711e-01 -4.90795165e-01 4.26662683e-01 6.74673617e-01
-8.30656588e-02 5.11853456e-01 -5.85775495e-01 6.77633464e-01
4.16736513e-01 1.89126611e-01 3.68925482e-01 -1.04592347e+00
-7.99280941e-01 -6.71156466e-01 -2.40550131e-01 -1.76520562e+00
5.91638684e-01 -1.10653949e+00 1.96062818e-01 -1.40964329e+00
2.16003768e-02 -3.30766141e-02 -7.75543451e-01 2.05987170e-01
-1.74100578e-01 -1.02988472e-02 4.61942852e-01 3.10705036e-01
-5.11276424e-01 5.36590755e-01 6.47456050e-01 -3.04851472e-01
-2.42473722e-01 2.12967157e-01 -5.64232886e-01 7.89363444e-01
7.98456252e-01 -3.07753772e-01 -6.30790889e-01 -6.21049047e-01
-6.05425835e-01 3.69389653e-01 1.68628544e-01 -1.34875679e+00
3.28859210e-01 -5.35987876e-02 1.68712497e-01 -5.37297070e-01
1.02214825e+00 -5.71219027e-01 1.09954990e-01 4.49748546e-01
-3.56209934e-01 -2.36031860e-01 4.76852179e-01 4.18967456e-01
-4.02350038e-01 -2.43753985e-01 7.17810392e-01 3.41741234e-01
-6.76894009e-01 7.93053135e-02 -9.39808726e-01 1.61724649e-02
5.26329219e-01 1.75760537e-02 -6.98137656e-02 -5.08925855e-01
-8.57578278e-01 2.33971365e-02 1.32315038e-02 3.73065650e-01
5.14461756e-01 -1.12683308e+00 -7.03827500e-01 4.56830144e-01
-6.41626537e-01 -1.84743643e-01 3.90329391e-01 8.83046746e-01
9.71726701e-03 7.50002801e-01 2.46947572e-01 -5.00447989e-01
-1.15820062e+00 6.44265413e-01 3.39704692e-01 -2.05363587e-01
-6.09579563e-01 1.03280377e+00 2.53449261e-01 -3.79807591e-01
3.70026946e-01 -6.20149255e-01 1.19487703e-01 -3.67281064e-02
6.30360782e-01 6.83216333e-01 1.09592183e-02 -7.12949038e-01
-6.62986338e-01 5.25658429e-01 9.85426158e-02 -5.93088806e-01
1.30044484e+00 -2.50188500e-01 2.99469121e-02 6.50911450e-01
1.17440164e+00 2.47076049e-01 -1.28848314e+00 -2.21184671e-01
-5.39418533e-02 1.29928023e-01 2.56072730e-01 -5.88490367e-01
-9.00793016e-01 1.11917686e+00 2.61899531e-01 4.21606824e-02
1.16196859e+00 -1.65382132e-01 1.13484740e+00 3.04747313e-01
9.42394808e-02 -8.33428860e-01 -4.20071743e-02 8.96766245e-01
8.82076502e-01 -7.73913085e-01 -4.21397686e-01 -1.39505059e-01
-7.27137864e-01 8.30828965e-01 3.02208304e-01 4.66952547e-02
6.24926329e-01 8.16332221e-01 2.03033105e-01 4.09005851e-01
-8.51241231e-01 -1.80474296e-01 5.75255692e-01 6.72926009e-01
3.66934538e-01 3.35936248e-02 3.06861132e-01 1.05204594e+00
-6.19335622e-02 -5.58662772e-01 4.50608581e-01 4.86725539e-01
-4.69135463e-01 -1.07276905e+00 -6.31271124e-01 4.68210988e-02
-1.62192121e-01 -3.01642150e-01 -2.79925726e-02 1.06111981e-01
-3.88349742e-01 1.26102602e+00 8.78187492e-02 -3.00269008e-01
1.06722817e-01 4.18354094e-01 3.53172153e-01 -4.13185418e-01
-4.12804455e-01 4.33246285e-01 -1.81403264e-01 -3.15567642e-01
-1.96027458e-01 -7.11149275e-01 -1.43530083e+00 -1.06458329e-01
-3.89758378e-01 2.35941932e-01 6.47275269e-01 8.29979122e-01
7.45771229e-01 9.30921376e-01 4.80700344e-01 -1.11812639e+00
-9.86980796e-01 -1.12649417e+00 -6.23445332e-01 -1.32701546e-02
8.25531304e-01 -4.11009341e-01 -4.58167851e-01 1.09419063e-01] | [14.876596450805664, 6.097800254821777] |
b6219249-79a4-4673-8edf-d98dd10b3a00 | towards-robust-aspect-based-sentiment | 2306.13971 | null | https://arxiv.org/abs/2306.13971v1 | https://arxiv.org/pdf/2306.13971v1.pdf | Towards Robust Aspect-based Sentiment Analysis through Non-counterfactual Augmentations | While state-of-the-art NLP models have demonstrated excellent performance for aspect based sentiment analysis (ABSA), substantial evidence has been presented on their lack of robustness. This is especially manifested as significant degradation in performance when faced with out-of-distribution data. Recent solutions that rely on counterfactually augmented datasets show promising results, but they are inherently limited because of the lack of access to explicit causal structure. In this paper, we present an alternative approach that relies on non-counterfactual data augmentation. Our proposal instead relies on using noisy, cost-efficient data augmentations that preserve semantics associated with the target aspect. Our approach then relies on modelling invariances between different versions of the data to improve robustness. A comprehensive suite of experiments shows that our proposal significantly improves upon strong pre-trained baselines on both standard and robustness-specific datasets. Our approach further establishes a new state-of-the-art on the ABSA robustness benchmark and transfers well across domains. | ['Jingbo Zhu', 'Tong Xiao', 'Pranava Madhyastha', 'Chunyang Xiao', 'Kaikai An', 'Yan Ding', 'Xinyu Liu'] | 2023-06-24 | null | null | null | null | ['sentiment-analysis', 'aspect-based-sentiment-analysis'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.44315112e-01 2.21950054e-01 -4.41613883e-01 -4.53430563e-01
-1.12423539e+00 -8.30312788e-01 1.32388759e+00 3.84169936e-01
-4.20113683e-01 6.60106361e-01 8.93043697e-01 -3.44700426e-01
-2.29824841e-01 -6.27602518e-01 -7.67285287e-01 -3.85452569e-01
2.10402668e-01 3.97970647e-01 -6.86898455e-02 -6.73741579e-01
3.29363853e-01 2.31591299e-01 -1.33661103e+00 3.36670905e-01
4.23338592e-01 6.89624190e-01 -8.37475419e-01 2.45998442e-01
8.37801173e-02 5.40724158e-01 -5.58681726e-01 -9.31634903e-01
4.28312212e-01 -1.78138062e-01 -7.28949487e-01 -1.13735981e-01
4.90404963e-01 -6.49409220e-02 -1.19587548e-01 8.59269917e-01
4.79222387e-01 9.00217593e-02 7.96857774e-01 -1.41322541e+00
-7.64680386e-01 7.43676782e-01 -9.05738056e-01 1.50767982e-01
4.50904757e-01 3.00298572e-01 1.45436049e+00 -4.62664098e-01
6.21266544e-01 1.26431608e+00 1.06584013e+00 5.25141299e-01
-1.57912517e+00 -6.74328208e-01 4.60558861e-01 -1.55105680e-01
-7.15746939e-01 -6.22462630e-01 8.59730542e-01 -1.82169840e-01
1.21023035e+00 2.43325561e-01 3.27632904e-01 1.62060118e+00
-4.78130812e-03 8.17390144e-01 1.42646611e+00 -3.52604151e-01
2.98791885e-01 3.21911462e-02 1.76574901e-01 1.20609850e-01
4.65830833e-01 2.30677456e-01 -7.46355295e-01 -3.73644173e-01
1.19309910e-01 -4.32049841e-01 -1.54459789e-01 -5.52672148e-01
-1.20338047e+00 9.34824228e-01 1.95416063e-01 2.64829874e-01
-3.70067596e-01 8.83617625e-02 6.39707327e-01 3.56970191e-01
8.24802101e-01 7.96194017e-01 -1.04929423e+00 -2.39884093e-01
-1.17994452e+00 6.66102827e-01 8.31205904e-01 4.71819192e-01
3.46069217e-01 -1.84453577e-02 -2.49099687e-01 5.95651627e-01
1.74348563e-01 2.80688047e-01 6.79531574e-01 -7.23253012e-01
6.21755779e-01 7.46453047e-01 2.05938160e-01 -1.00513351e+00
-5.84560156e-01 -3.84760439e-01 -6.20466530e-01 1.27044857e-01
6.23014510e-01 -6.00938648e-02 -1.03885686e+00 2.17258024e+00
3.37379754e-01 3.83083224e-02 2.18724728e-01 6.69635773e-01
2.87776470e-01 1.90414235e-01 1.97448269e-01 -5.60894497e-02
1.28771222e+00 -7.12066710e-01 -6.83609605e-01 -4.01811033e-01
6.74664676e-01 -7.34529555e-01 1.50309908e+00 3.80228490e-01
-8.11171174e-01 -3.95240411e-02 -1.19323659e+00 -1.43752620e-02
-7.74244785e-01 -2.42131278e-01 8.44969988e-01 9.97782350e-01
-8.45754147e-01 5.87815940e-01 -6.36954725e-01 -3.71471256e-01
5.65265656e-01 2.34516829e-01 -6.95802748e-01 1.71437934e-02
-1.20473409e+00 9.64703083e-01 3.34733874e-01 -3.22311997e-01
-6.75475001e-01 -1.34498143e+00 -1.13534009e+00 -1.09909162e-01
5.87683082e-01 -8.12725782e-01 1.21198297e+00 -9.62331295e-01
-1.45279491e+00 8.01927030e-01 1.42723382e-01 -6.58168793e-01
6.77357435e-01 -5.69058836e-01 -4.31228369e-01 -2.66712785e-01
2.59401649e-01 4.65884596e-01 9.59739208e-01 -1.25412047e+00
-3.28365088e-01 -4.88059402e-01 3.17923635e-01 -9.71356593e-03
-5.04471183e-01 2.42495731e-01 -2.30366513e-02 -1.10774171e+00
-3.87740493e-01 -9.04648304e-01 -2.45045617e-01 -4.88135904e-01
-5.92957139e-01 -1.23337075e-01 8.31159055e-01 -4.90888625e-01
1.29528153e+00 -1.94030821e+00 -7.53056332e-02 -3.67607400e-02
-3.21728587e-01 2.40292832e-01 -3.40550393e-01 4.64159191e-01
-5.41099608e-01 5.17467618e-01 -6.29575014e-01 -5.72145998e-01
2.92268813e-01 -2.47065779e-02 -7.99326479e-01 5.43251038e-01
4.59608912e-01 8.31878960e-01 -8.51033926e-01 -9.61407870e-02
1.67410597e-01 3.12791198e-01 -8.34166646e-01 -1.85448766e-01
-2.88661003e-01 2.34057695e-01 -7.00181127e-02 5.63537836e-01
5.58542848e-01 4.46583703e-02 2.17479780e-01 -2.57050902e-01
5.42070307e-02 7.07718492e-01 -1.10871840e+00 1.88303375e+00
-4.67486918e-01 2.93886095e-01 -2.52769321e-01 -9.66885686e-01
3.76684338e-01 2.48047814e-01 3.70846063e-01 -5.55770755e-01
7.04215169e-02 1.34002745e-01 -5.78642264e-02 -1.12640716e-01
6.83616459e-01 -5.38460851e-01 -4.86493975e-01 8.43947470e-01
1.04042053e-01 -4.26496893e-01 2.80020028e-01 2.41364077e-01
1.08804560e+00 5.12663603e-01 5.75432122e-01 -4.47829843e-01
4.80338752e-01 7.06679448e-02 5.59774160e-01 7.83519924e-01
-2.50912130e-01 7.23460019e-01 6.98181450e-01 -2.60600686e-01
-1.00510681e+00 -8.68697047e-01 -2.23503023e-01 9.91494954e-01
-4.44961816e-01 -5.70937634e-01 -6.42758012e-01 -1.31074953e+00
9.46805030e-02 1.17168331e+00 -1.07540751e+00 -2.13567108e-01
-5.14021277e-01 -1.41692531e+00 6.54196501e-01 4.28823769e-01
4.01096880e-01 -7.97714412e-01 -3.60563695e-01 1.10246316e-01
-1.53135255e-01 -1.26988256e+00 -1.92140192e-01 -1.18147202e-01
-7.84035623e-01 -1.16290438e+00 -3.46088558e-01 8.44812319e-02
2.43310884e-01 6.11551963e-02 1.28905177e+00 -3.57574373e-01
3.90332900e-02 4.32721853e-01 -1.79387316e-01 -8.83822203e-01
-4.57222551e-01 1.22896172e-01 3.21655959e-01 2.52349637e-02
4.53129500e-01 -5.47446191e-01 -5.19979119e-01 1.01286583e-01
-1.12796414e+00 -4.70634699e-01 5.11496067e-01 8.28214943e-01
2.88570315e-01 -7.70694241e-02 8.16996157e-01 -1.19197404e+00
8.80027533e-01 -5.52595377e-01 -4.98063207e-01 4.46968072e-04
-9.33935463e-01 2.61324793e-01 6.73674464e-01 -2.14233547e-01
-1.16181612e+00 -2.87075639e-01 3.30089331e-02 -4.56570610e-02
-3.03769886e-01 5.51667213e-01 -1.19756334e-01 4.23654199e-01
7.47326970e-01 -2.90013909e-01 -7.40780495e-03 -4.50320750e-01
7.79006600e-01 3.33095104e-01 4.60982710e-01 -5.39924443e-01
8.96066725e-01 8.10529172e-01 1.06440879e-01 -5.31498909e-01
-1.24238265e+00 -3.93484265e-01 -4.06184316e-01 3.37500155e-01
4.79322731e-01 -8.21112216e-01 -3.99967104e-01 5.08301675e-01
-7.75748849e-01 -2.30169907e-01 -5.63360810e-01 2.66878366e-01
-6.08280778e-01 4.63658065e-01 -3.56924981e-01 -7.28721082e-01
-2.40209803e-01 -9.20658648e-01 1.08561277e+00 -1.79163873e-01
-5.62701166e-01 -1.06835115e+00 4.26305741e-01 4.65855747e-01
4.52150941e-01 3.36645007e-01 8.56527269e-01 -9.76150513e-01
-1.00559816e-01 -5.05628049e-01 -4.61271107e-02 3.37891549e-01
2.51773566e-01 -1.59372836e-02 -1.25618434e+00 -2.72418112e-01
1.79822072e-02 -5.44097960e-01 1.00542200e+00 3.30225438e-01
7.52332091e-01 -4.66794938e-01 -4.27577160e-02 3.35951149e-01
1.31041443e+00 -6.24747455e-01 4.99237776e-01 8.53758931e-01
3.19600046e-01 8.66610467e-01 7.57570326e-01 1.90804154e-01
5.11098325e-01 7.58692324e-01 5.28968513e-01 -2.68185083e-02
-1.18111745e-01 -3.65352035e-01 5.62671721e-01 1.20488063e-01
4.63133082e-02 -2.36400813e-01 -7.87238538e-01 9.75223720e-01
-1.88860977e+00 -9.33698595e-01 -8.58064815e-02 2.18867779e+00
9.61346865e-01 3.34499508e-01 3.74359429e-01 4.74678963e-01
1.63589060e-01 5.50919294e-01 -3.48489523e-01 -6.08241022e-01
-3.53419811e-01 1.50414526e-01 4.70591784e-01 3.84905457e-01
-1.47936213e+00 9.33487236e-01 6.92186260e+00 7.15923131e-01
-7.84679234e-01 1.16603658e-01 5.93100548e-01 -2.85939068e-01
-4.93386865e-01 6.56579956e-02 -4.86678481e-01 2.28562161e-01
1.01335812e+00 -1.36871666e-01 1.49327755e-01 6.52269423e-01
2.37278551e-01 -1.32992864e-01 -1.09319103e+00 4.12887692e-01
2.02555954e-01 -1.23641515e+00 1.62743837e-01 6.96325973e-02
9.82935667e-01 2.35768910e-02 4.71616507e-01 4.40432310e-01
5.89132845e-01 -8.73193026e-01 6.36291206e-01 1.43613681e-01
3.98335308e-01 -7.25218296e-01 9.41943049e-01 2.35832348e-01
-5.92764676e-01 -9.26043689e-02 -1.18509121e-01 -3.22637439e-01
1.14660613e-01 6.37041152e-01 -7.21266031e-01 8.19201350e-01
5.89596152e-01 7.26965845e-01 -6.82513118e-01 4.80816394e-01
-5.78956544e-01 8.87550354e-01 -4.28663820e-01 3.22239161e-01
3.51720810e-01 1.30260319e-01 7.09398985e-01 1.25282609e+00
-8.64188597e-02 -2.26780057e-01 -2.62519062e-01 6.60796344e-01
-3.68387073e-01 2.32985228e-01 -9.92767036e-01 -3.83616681e-03
2.75446251e-02 1.15289998e+00 -4.78778750e-01 -3.15814257e-01
-7.05494642e-01 6.55608416e-01 4.16029841e-01 2.41223469e-01
-7.62673259e-01 5.55270202e-02 9.40349281e-01 6.87917694e-02
3.74809116e-01 5.94287962e-02 -6.06084704e-01 -1.35450590e+00
1.61204617e-02 -1.29904997e+00 8.72929692e-01 -4.23976511e-01
-1.52929425e+00 1.10872366e-01 2.70124406e-01 -9.91348326e-01
-4.71059412e-01 -5.45621395e-01 -4.67865527e-01 7.35418141e-01
-1.76595533e+00 -1.56101084e+00 3.73351753e-01 4.87758070e-01
3.78740430e-01 -1.28865853e-01 9.24621403e-01 -2.06674393e-02
-5.90749264e-01 7.37676620e-01 -1.67819649e-01 -1.11051120e-01
1.10460150e+00 -1.40472472e+00 6.98977351e-01 1.23469043e+00
3.04706126e-01 8.10427725e-01 1.07769191e+00 -6.33868575e-01
-1.17528307e+00 -9.01006758e-01 1.00733674e+00 -1.11488414e+00
1.18521261e+00 -2.83423603e-01 -7.63602734e-01 9.20727849e-01
3.73414129e-01 -4.00108546e-02 8.53038490e-01 7.34755993e-01
-9.56682146e-01 -7.45268119e-03 -1.30977869e+00 7.01329172e-01
7.94789195e-01 -4.72666144e-01 -1.12987900e+00 1.00977786e-01
7.08876014e-01 -7.13120103e-02 -5.72096586e-01 6.93846524e-01
5.00949502e-01 -1.10431850e+00 9.64757621e-01 -1.14989376e+00
6.83324456e-01 -2.22274214e-01 -2.74300307e-01 -1.38286042e+00
-2.72645019e-02 -6.76468134e-01 -3.18239778e-02 1.45227480e+00
6.73201382e-01 -6.71325922e-01 7.90552378e-01 8.56106281e-01
8.58547762e-02 -4.85782117e-01 -9.41981494e-01 -7.53736734e-01
2.81590223e-01 -8.94639432e-01 6.90051258e-01 1.35820591e+00
1.27240375e-01 3.73270392e-01 -3.59840631e-01 2.59122223e-01
5.95740199e-01 1.43982217e-01 9.67174888e-01 -9.88419950e-01
-3.25013280e-01 -6.19341969e-01 -3.92870784e-01 -3.19002777e-01
3.55605721e-01 -7.40014017e-01 -2.77579010e-01 -1.25862288e+00
1.96119159e-01 -6.22821078e-02 -5.26762903e-01 8.53629827e-01
-5.37594259e-01 6.09412670e-01 2.03993276e-01 -6.36352897e-02
-4.04732525e-01 4.70651478e-01 6.44239485e-01 -7.76655599e-02
-1.74728349e-01 -4.15791608e-02 -1.18829012e+00 1.00537956e+00
1.00985157e+00 -6.05795681e-01 -4.65665549e-01 -2.09707722e-01
5.15805781e-01 -6.32958472e-01 4.23461765e-01 -5.65277398e-01
-2.91636735e-01 -3.84700187e-02 9.05227065e-02 -3.42436135e-01
2.79993504e-01 -6.86721027e-01 -2.88275689e-01 3.19092810e-01
-4.24882799e-01 2.36992151e-01 6.84055805e-01 9.04486239e-01
-1.30620360e-01 -3.87404710e-02 6.02288902e-01 1.24347853e-02
-3.35361093e-01 -9.67556313e-02 -1.46914169e-01 4.26114589e-01
7.65245497e-01 2.77379304e-01 -4.89535987e-01 -3.55009228e-01
-3.14589649e-01 -6.04712144e-02 6.66473925e-01 6.47057474e-01
7.61656165e-02 -1.07103765e+00 -7.68153369e-01 8.65403637e-02
2.78017223e-01 -4.08580333e-01 1.37846291e-01 8.02176714e-01
1.38793588e-01 5.76049089e-01 1.05234966e-01 -1.10395700e-01
-1.05656016e+00 7.93260872e-01 7.26713911e-02 -8.17507625e-01
-3.56457114e-01 6.05602086e-01 1.98270038e-01 -6.75659359e-01
-1.33670926e-01 -3.40377808e-01 -1.23595484e-01 3.16703379e-01
4.32137638e-01 2.77089238e-01 3.09123427e-01 -6.60811484e-01
-5.19416392e-01 3.58975381e-01 -3.14349562e-01 -4.63563651e-01
1.43529570e+00 1.14321172e-01 -6.81823958e-03 3.29902381e-01
9.64138567e-01 4.74690288e-01 -1.03377163e+00 -6.85186163e-02
2.70777315e-01 -5.06666422e-01 9.36841220e-02 -1.29642212e+00
-6.90086365e-01 5.61899006e-01 9.41824242e-02 1.29027709e-01
1.10603833e+00 -1.88243464e-01 3.85074258e-01 2.17648461e-01
1.55074447e-01 -1.18363523e+00 -8.27161223e-02 3.38111430e-01
1.03518832e+00 -1.54770076e+00 5.11735678e-01 -2.37742692e-01
-7.07511365e-01 6.57895565e-01 3.03706288e-01 -1.61204875e-01
5.81571221e-01 2.19887063e-01 2.84723431e-01 -1.55491233e-01
-7.87225783e-01 -1.42457768e-01 2.89832413e-01 5.69681883e-01
5.23984551e-01 -1.07306346e-01 -5.30266404e-01 8.08197916e-01
-4.54203218e-01 -2.30917305e-01 6.92847610e-01 9.16961670e-01
3.24221104e-01 -1.20081830e+00 -2.58693516e-01 3.15538108e-01
-1.06849849e+00 -3.72393250e-01 -5.61808348e-01 1.35933566e+00
-3.38900000e-01 1.16655135e+00 -3.04776102e-01 -7.64848441e-02
6.10185981e-01 3.28963637e-01 4.01207209e-01 -4.94692624e-01
-8.20806503e-01 -1.36045516e-01 4.76731420e-01 -8.60849619e-01
-7.77210951e-01 -1.07041538e+00 -7.14478314e-01 -2.22706333e-01
-5.87184802e-02 -3.59534994e-02 9.17512417e-01 8.62921894e-01
4.17672664e-01 3.74720007e-01 3.37299764e-01 -5.71111500e-01
-9.00192976e-01 -1.06614745e+00 -2.18179628e-01 8.68818343e-01
5.03542125e-01 -5.40461063e-01 -5.45948088e-01 -1.77593455e-01] | [11.283243179321289, 6.89691686630249] |
588aa7de-428f-4301-81bf-113635031c76 | a-bayesian-3d-multi-view-multi-object | 2001.04118 | null | https://arxiv.org/abs/2001.04118v4 | https://arxiv.org/pdf/2001.04118v4.pdf | A Bayesian Filter for Multi-view 3D Multi-object Tracking with Occlusion Handling | This paper proposes an online multi-camera multi-object tracker that only requires monocular detector training, independent of the multi-camera configurations, allowing seamless extension/deletion of cameras without retraining effort. The proposed algorithm has a linear complexity in the total number of detections across the cameras, and hence scales gracefully with the number of cameras. It operates in the 3D world frame, and provides 3D trajectory estimates of the objects. The key innovation is a high fidelity yet tractable 3D occlusion model, amenable to optimal Bayesian multi-view multi-object filtering, which seamlessly integrates, into a single Bayesian recursion, the sub-tasks of track management, state estimation, clutter rejection, and occlusion/misdetection handling. The proposed algorithm is evaluated on the latest WILDTRACKS dataset, and demonstrated to work in very crowded scenes on a new dataset. | ['Sven Nordholm', 'Du Yong Kim', 'Ba Tuong Vo', 'Jonah Ong', 'Ba Ngu Vo'] | 2020-01-13 | null | null | null | null | ['occlusion-handling', '3d-multi-object-tracking'] | ['computer-vision', 'computer-vision'] | [-7.31090978e-02 -6.74264431e-01 6.60835579e-02 1.80342287e-01
-7.21651971e-01 -1.11509824e+00 5.10947227e-01 -1.32684186e-01
-4.48255479e-01 4.09147769e-01 -2.74999350e-01 -1.45180479e-01
-1.70112103e-01 -5.51178493e-02 -8.18294287e-01 -5.75996101e-01
6.50018528e-02 6.42107844e-01 1.01033604e+00 4.25499052e-01
-3.89094278e-02 9.98908222e-01 -1.74369597e+00 -4.57670987e-01
3.01651001e-01 9.89103317e-01 3.18556219e-01 1.38011587e+00
4.93442714e-01 7.05692708e-01 -4.40151513e-01 -3.25522393e-01
6.17534757e-01 4.40409124e-01 4.22524884e-02 7.72551417e-01
1.27046144e+00 -5.80592632e-01 -6.45228863e-01 8.74813199e-01
4.61650491e-01 4.49560657e-02 2.35090762e-01 -1.26644075e+00
-4.51821536e-02 -5.05112946e-01 -7.97896922e-01 5.71386695e-01
4.67708856e-01 3.45349401e-01 5.10105073e-01 -8.93318415e-01
5.84781408e-01 1.34948087e+00 1.07151341e+00 2.35273078e-01
-1.14322782e+00 -5.57324231e-01 3.20841968e-01 -1.19123952e-02
-1.48214650e+00 -6.89969838e-01 2.36088224e-02 -8.89768958e-01
7.92189419e-01 1.46994680e-01 8.34739625e-01 7.52878547e-01
1.89846098e-01 6.61739588e-01 1.08162344e+00 -1.04324125e-01
4.05895747e-02 -2.48125987e-03 2.35538393e-01 6.97576702e-01
1.00643408e+00 3.35077912e-01 -7.73735821e-01 -4.41278040e-01
7.94019878e-01 2.78298914e-01 -1.24795236e-01 -8.11668634e-01
-1.07819080e+00 5.32860637e-01 -3.23879868e-01 -3.51969898e-01
-2.14435011e-01 3.35778236e-01 2.42741689e-01 -1.03436895e-01
2.17154667e-01 -2.42390037e-01 -5.13993442e-01 7.32264714e-03
-9.51660693e-01 4.95105356e-01 5.02136946e-01 1.71572649e+00
5.39089620e-01 7.58533925e-02 -1.27494380e-01 2.81926244e-01
5.80165505e-01 1.45586610e+00 -2.79244095e-01 -1.11013675e+00
3.46193731e-01 2.18961537e-01 7.85863400e-01 -6.97667420e-01
-4.38457191e-01 -6.02263868e-01 -2.74049610e-01 4.08469290e-01
4.45623457e-01 -3.06975275e-01 -6.23911679e-01 1.27559376e+00
9.43676412e-01 3.31013918e-01 -1.42811298e-01 8.05711091e-01
4.86866027e-01 1.18050396e-01 -3.31458926e-01 -4.02218342e-01
1.59251809e+00 -8.55245292e-01 -7.45141566e-01 -7.22748399e-01
-6.98343813e-02 -1.14466584e+00 5.74536435e-02 5.30295908e-01
-1.02100289e+00 -6.13627136e-01 -1.06754684e+00 2.02432916e-01
-2.05676645e-01 3.74219447e-01 6.88165784e-01 9.33345497e-01
-9.95931327e-01 -6.78530261e-02 -9.09505725e-01 -5.49465239e-01
1.61461517e-01 4.31466192e-01 -2.05773994e-01 -2.63468146e-01
-4.12867546e-01 1.22823596e+00 3.53818536e-01 1.68058291e-01
-1.22455275e+00 -5.38298547e-01 -7.38048971e-01 -2.89784431e-01
6.25860870e-01 -9.68017995e-01 1.24595487e+00 -1.60607323e-01
-1.34724283e+00 7.40544796e-01 -3.64831537e-01 -4.76142108e-01
6.99690282e-01 -8.86138558e-01 -5.15890300e-01 1.59490913e-01
2.13534921e-01 3.46100479e-01 1.14746046e+00 -1.15250278e+00
-1.30189657e+00 -5.15773535e-01 -1.06766209e-01 3.62119824e-01
2.31447324e-01 3.04446965e-01 -8.86280000e-01 -3.07720900e-01
1.30467638e-01 -1.10090029e+00 -2.41888687e-01 3.66090268e-01
2.35639475e-02 3.25424373e-01 1.05674243e+00 -3.23804140e-01
7.30395854e-01 -2.03898406e+00 2.26142984e-02 -2.14975208e-01
1.44425884e-01 1.52517438e-01 1.51097849e-01 2.81972498e-01
4.82279986e-01 -8.85994017e-01 3.26263607e-01 -6.36049032e-01
-2.37692356e-01 2.41182566e-01 -2.75218934e-01 1.23876965e+00
-1.18311502e-01 3.00726235e-01 -9.23238575e-01 -5.46134770e-01
7.03178644e-01 3.40887845e-01 -5.22051156e-01 8.66758525e-02
9.56627056e-02 4.60362405e-01 -3.56931478e-01 8.09279501e-01
1.16593325e+00 -2.80744582e-02 -1.38684168e-01 -6.53490722e-02
-7.11511493e-01 -4.88314569e-01 -2.05259085e+00 1.65169919e+00
-1.57540113e-01 6.94623291e-01 5.87262392e-01 -1.81786180e-01
5.19175231e-01 2.72478372e-01 6.40057206e-01 1.48789138e-01
-8.08995683e-04 2.79469974e-02 -5.28385758e-01 -3.19859266e-01
8.21195424e-01 2.48438165e-01 -1.85825136e-02 4.60226350e-02
3.14090192e-01 -3.05630062e-02 4.33665216e-01 2.08159834e-01
1.14044261e+00 3.17042589e-01 5.44205904e-01 -1.20932676e-01
4.18322861e-01 2.16392979e-01 8.61170828e-01 1.20854068e+00
-3.71493250e-01 2.76325941e-01 -4.61561888e-01 -4.26207781e-01
-9.33835685e-01 -1.47194386e+00 -2.29496196e-01 8.65189135e-01
6.50506616e-01 -1.68756589e-01 -2.36573622e-01 -7.77476430e-02
3.29890072e-01 1.10323858e-02 -9.87484157e-02 3.09521198e-01
-6.10832632e-01 -8.27192545e-01 4.34142977e-01 3.73869240e-01
2.52893984e-01 -8.17075297e-02 -1.28335392e+00 4.14089501e-01
1.68456674e-01 -1.72287929e+00 -5.63243270e-01 1.28673747e-01
-8.91037345e-01 -1.48754466e+00 -3.05111200e-01 -2.36329794e-01
4.31145281e-01 1.08004212e+00 8.64800096e-01 -2.69520879e-01
-5.97404301e-01 1.14108014e+00 -6.91300780e-02 -5.80097258e-01
-1.03085907e-02 -6.18581116e-01 5.76318324e-01 9.64988843e-02
2.45615900e-01 -1.55381233e-01 -4.02512640e-01 4.22557622e-01
-3.32731456e-01 -1.84583992e-01 4.43715662e-01 7.16073811e-01
6.91674590e-01 -1.15560643e-01 -1.77855223e-01 -3.15176308e-01
-2.48674810e-01 -2.58241966e-02 -1.60812902e+00 1.80396140e-01
-9.31326896e-02 -4.88914371e-01 1.44387081e-01 -6.21777773e-01
-1.31065083e+00 8.27127755e-01 5.29547095e-01 -7.89523780e-01
-3.20425242e-01 -4.51402694e-01 -4.46928032e-02 -5.97424507e-01
3.91926825e-01 1.21898003e-01 -3.38813454e-01 -6.00168347e-01
5.56847394e-01 4.49451476e-01 9.89807189e-01 -2.26317897e-01
1.20756447e+00 1.09716785e+00 4.30680066e-01 -1.11291838e+00
-9.36536431e-01 -1.24462605e+00 -1.05845654e+00 -6.33642495e-01
7.98735678e-01 -1.71961236e+00 -9.22883034e-01 5.88842034e-01
-1.35830653e+00 2.88144529e-01 -2.60314774e-02 8.03530812e-01
-5.29166520e-01 5.47508001e-01 -2.57346541e-01 -1.33088076e+00
-4.10245210e-02 -9.74950969e-01 1.40970993e+00 3.34112942e-01
2.56703585e-01 -7.28487909e-01 8.28743502e-02 4.12431926e-01
1.08960494e-01 1.66406572e-01 -2.86969751e-01 -5.53848371e-02
-1.44013488e+00 -5.90475559e-01 -6.25783131e-02 -1.22607946e-01
-2.51938313e-01 5.25763705e-02 -1.17458916e+00 -7.97058165e-01
1.40283495e-01 8.81442577e-02 6.87393665e-01 6.77073419e-01
3.16893756e-01 7.55901560e-02 -7.22114563e-01 6.46570802e-01
1.64169681e+00 1.43532947e-01 -6.33752868e-02 2.26195365e-01
7.08467305e-01 -6.41146228e-02 9.80559945e-01 7.88564026e-01
4.34459984e-01 9.57094908e-01 5.72294652e-01 2.14597419e-01
-2.87489235e-01 2.89629493e-02 4.75453079e-01 4.37922150e-01
1.25143200e-01 -2.19354108e-01 -5.23929477e-01 5.56120515e-01
-2.04589415e+00 -1.15372252e+00 -5.17617583e-01 2.29177761e+00
-6.65762499e-02 9.19273421e-02 2.24826768e-01 -3.16633433e-01
1.02049303e+00 4.98875044e-02 -6.33790493e-01 4.24816161e-01
-2.31312066e-01 -3.84876966e-01 1.37249196e+00 5.55512965e-01
-1.58800483e+00 7.93340206e-01 7.18715763e+00 4.77248669e-01
-3.40411872e-01 5.44834375e-01 -4.84909773e-01 -4.67189670e-01
6.85007095e-01 1.84205249e-01 -1.76557839e+00 2.07970932e-01
7.49210835e-01 8.07643086e-02 3.42679620e-01 8.56763840e-01
1.64421484e-01 -6.20637536e-01 -9.94039953e-01 1.29220450e+00
2.86201596e-01 -1.24315393e+00 -2.87407488e-01 8.71729329e-02
5.31798780e-01 4.79283094e-01 -1.88439295e-01 -1.58266127e-01
5.39431155e-01 6.34532049e-02 1.24012375e+00 5.14294565e-01
3.76815796e-01 -4.79238391e-01 3.36874872e-01 6.08416617e-01
-1.60592210e+00 -4.76818740e-01 -4.31940049e-01 -7.47114196e-02
6.12404943e-01 4.09352392e-01 -5.36891937e-01 6.19763315e-01
9.99888182e-01 6.42141938e-01 -6.26595140e-01 1.80825877e+00
1.83619887e-01 -1.44371688e-01 -5.93056321e-01 3.93509299e-01
-1.38885319e-01 -1.17748544e-01 1.10328591e+00 1.21789837e+00
3.07460010e-01 6.09769598e-02 7.10404515e-01 1.28478393e-01
4.54683095e-01 -5.79198122e-01 -6.46717370e-01 7.52382994e-01
6.77834332e-01 1.22813606e+00 -6.47648633e-01 -4.21293318e-01
-8.94379377e-01 6.14785969e-01 -5.84217645e-02 1.72804400e-01
-9.37090456e-01 1.32348865e-01 8.80387604e-01 -1.76923767e-01
8.11630785e-01 -5.83510101e-01 5.03344759e-02 -1.30162871e+00
9.88756344e-02 -5.67616582e-01 5.62326729e-01 -6.78900599e-01
-1.00595474e+00 2.71980315e-01 3.20461184e-01 -1.55433249e+00
7.19987750e-02 -7.73354113e-01 -1.43231034e-01 6.33149266e-01
-1.36997342e+00 -1.35197437e+00 -2.81525731e-01 6.08126402e-01
5.54042339e-01 -2.26880148e-01 4.41898733e-01 7.34152377e-01
-4.62577373e-01 9.20192450e-02 4.01393920e-01 -3.15560400e-01
5.69960415e-01 -1.05434775e+00 3.19267899e-01 1.55020773e+00
4.06244583e-02 3.09922129e-01 1.00074041e+00 -7.58207560e-01
-1.93990147e+00 -1.19792187e+00 4.41937596e-01 -9.51940358e-01
7.47259378e-01 -6.37980759e-01 -2.80178428e-01 9.53467071e-01
-5.12555521e-03 1.75883725e-01 5.19413888e-01 -3.07088601e-04
-1.04014836e-01 -1.40415475e-01 -9.55877900e-01 1.26875117e-01
1.13182485e+00 -3.28062743e-01 -4.65966672e-01 6.29739583e-01
4.58658516e-01 -1.03269541e+00 -8.38572204e-01 2.11139068e-01
6.06602371e-01 -7.17781484e-01 1.35304046e+00 -8.09556469e-02
-1.08716404e+00 -1.02338493e+00 -2.40850508e-01 -7.15791047e-01
-4.43654597e-01 -8.89038622e-01 -7.06820428e-01 1.04482412e+00
-1.90937564e-01 -3.73977214e-01 5.84539771e-01 2.44366199e-01
-2.96914786e-01 1.91983446e-01 -1.34337890e+00 -1.14676476e+00
-8.50145698e-01 -4.27345812e-01 1.29609793e-01 2.73355603e-01
-7.60665476e-01 1.61610812e-01 -9.34835911e-01 1.14907730e+00
1.44327843e+00 3.05927992e-01 1.35218811e+00 -1.57179630e+00
-6.48409963e-01 6.42855093e-02 -4.29920763e-01 -1.41996074e+00
-3.42740804e-01 -1.34612650e-01 3.73714343e-02 -1.08624077e+00
3.83457929e-01 -1.75164759e-01 3.74209881e-01 -1.47663340e-01
-7.79736638e-02 8.24291185e-02 4.03074324e-01 2.72100002e-01
-1.15125072e+00 2.39409387e-01 8.09416831e-01 -7.63105005e-02
1.35792300e-01 4.27353382e-01 -1.10166810e-01 9.76083577e-01
2.57107858e-02 -6.25354528e-01 3.33712995e-02 -7.32809305e-01
-1.32945135e-01 2.63334602e-01 6.85405672e-01 -1.43457162e+00
6.93838954e-01 -1.19997039e-02 6.59475803e-01 -1.34772956e+00
8.60331059e-01 -1.23537028e+00 5.31354725e-01 4.68524635e-01
4.84268934e-01 3.56041491e-01 4.79612947e-01 1.08449650e+00
2.17236176e-01 -6.44426569e-02 9.30826366e-01 -1.33851752e-01
-8.67366552e-01 4.09046203e-01 -6.26277983e-01 -5.22617362e-02
1.25678313e+00 -4.89060402e-01 -2.23310575e-01 1.55743673e-01
-8.09332669e-01 4.25119787e-01 5.64753413e-01 5.60912549e-01
3.44966471e-01 -1.12683523e+00 -3.92573118e-01 1.45819798e-01
8.24020505e-02 1.61081459e-02 3.67138505e-01 6.87217653e-01
-4.60220873e-01 6.63143396e-01 1.45253297e-02 -1.13278544e+00
-1.60960913e+00 6.88066483e-01 3.96182805e-01 -2.92088743e-02
-7.08455563e-01 4.58841175e-01 -1.09557502e-01 7.33538568e-02
4.90490615e-01 -2.07854867e-01 1.55483961e-01 -8.38922784e-02
8.97509694e-01 8.18882763e-01 -1.73212096e-01 -8.75941932e-01
-5.08853674e-01 1.03142989e+00 1.25633121e-01 -9.52395517e-03
1.05228782e+00 -6.60172820e-01 2.43936822e-01 3.37647498e-01
3.64007056e-01 2.49485858e-03 -1.98929143e+00 -3.35588992e-01
8.05457979e-02 -8.81314158e-01 1.00369357e-01 -3.35009307e-01
-6.78565562e-01 4.85016644e-01 1.04530776e+00 -2.00432390e-01
7.43594646e-01 -4.60409448e-02 3.14682096e-01 6.66352034e-01
7.82963693e-01 -9.57561791e-01 -1.98667765e-01 6.49138987e-01
3.69376600e-01 -1.13660634e+00 5.14780998e-01 -4.61278319e-01
-2.14431033e-01 8.10369372e-01 6.14115596e-01 -2.32241731e-02
4.72624719e-01 6.14562809e-01 -1.19496487e-01 -2.92840928e-01
-7.15369642e-01 -4.71282184e-01 -8.38723704e-02 8.85968685e-01
-3.84790093e-01 -9.92944837e-02 4.34426367e-01 -2.25840375e-01
5.35359025e-01 -1.04370832e-01 5.99778116e-01 1.10024953e+00
-7.51527905e-01 -6.31476283e-01 -1.15876365e+00 1.92811117e-01
-2.20048070e-01 2.05238387e-01 1.25642940e-01 8.40193450e-01
4.29214507e-01 1.16401124e+00 1.53570607e-01 2.28095129e-01
6.82333529e-01 -2.98780352e-01 7.89496779e-01 -5.33662319e-01
-4.28888351e-01 4.49690074e-01 1.64326690e-02 -6.05621397e-01
-7.06142724e-01 -1.43187475e+00 -4.33889449e-01 -1.32734209e-01
-8.77422094e-01 -2.12510630e-01 5.74054480e-01 9.23898816e-01
3.72581393e-01 2.90952981e-01 4.12190318e-01 -1.20425308e+00
-4.55613405e-01 -7.93395817e-01 -5.85408866e-01 -1.13312215e-01
7.65511513e-01 -1.14988518e+00 -2.38707229e-01 2.05969617e-01] | [6.495640754699707, -2.072140693664551] |
c7dc3a44-331d-4f4e-bf91-1114b6e646fb | towards-open-vocabulary-object-detection | 2111.09452 | null | https://arxiv.org/abs/2111.09452v3 | https://arxiv.org/pdf/2111.09452v3.pdf | Open Vocabulary Object Detection with Pseudo Bounding-Box Labels | Despite great progress in object detection, most existing methods work only on a limited set of object categories, due to the tremendous human effort needed for bounding-box annotations of training data. To alleviate the problem, recent open vocabulary and zero-shot detection methods attempt to detect novel object categories beyond those seen during training. They achieve this goal by training on a pre-defined base categories to induce generalization to novel objects. However, their potential is still constrained by the small set of base categories available for training. To enlarge the set of base classes, we propose a method to automatically generate pseudo bounding-box annotations of diverse objects from large-scale image-caption pairs. Our method leverages the localization ability of pre-trained vision-language models to generate pseudo bounding-box labels and then directly uses them for training object detectors. Experimental results show that our method outperforms the state-of-the-art open vocabulary detector by 8% AP on COCO novel categories, by 6.3% AP on PASCAL VOC, by 2.3% AP on Objects365 and by 2.8% AP on LVIS. Code is available at https://github.com/salesforce/PB-OVD. | ['Caiming Xiong', 'Wenhao Liu', 'ran Xu', 'Junnan Li', 'Juan Carlos Niebles', 'Chen Xing', 'Mingfei Gao'] | 2021-11-18 | null | null | null | null | ['open-vocabulary-object-detection'] | ['computer-vision'] | [ 1.68999135e-01 2.09640935e-01 -1.75167695e-01 -4.56992298e-01
-1.16261053e+00 -9.19980884e-01 6.57477796e-01 1.93844303e-01
-5.32161474e-01 4.12741393e-01 -1.00455277e-01 -4.20705788e-03
5.32372117e-01 -5.64147353e-01 -9.35149074e-01 -3.04542750e-01
1.88084185e-01 5.95707536e-01 8.88033926e-01 -8.13622102e-02
1.26655102e-01 2.84014225e-01 -1.78698039e+00 4.41514760e-01
6.06934369e-01 8.12969744e-01 3.08157921e-01 6.65956140e-01
-1.69952914e-01 3.01505864e-01 -4.96077001e-01 -2.04037875e-01
5.61914504e-01 -1.72382563e-01 -5.60968041e-01 1.21283390e-01
8.69091272e-01 -4.92049575e-01 -3.19925368e-01 1.15571284e+00
3.81845087e-01 3.02063733e-01 8.00465226e-01 -1.15780103e+00
-8.47141504e-01 4.29564178e-01 -5.55151105e-01 3.54104847e-01
1.51561439e-01 3.32267404e-01 1.17302763e+00 -1.37779939e+00
7.59795487e-01 9.97253001e-01 5.59969246e-01 9.88666892e-01
-1.43615425e+00 -9.38321412e-01 1.48146182e-01 3.54281478e-02
-1.72946393e+00 -5.58960438e-01 4.19007331e-01 -6.45074010e-01
9.34239447e-01 1.41439244e-01 3.03250253e-01 9.55261350e-01
-4.66264248e-01 7.55463839e-01 8.37136865e-01 -4.66360301e-01
2.30467513e-01 4.81615961e-01 2.89772362e-01 5.29741585e-01
2.94047236e-01 5.79692833e-02 -1.48279652e-01 -5.89636713e-02
5.56979239e-01 -1.49056658e-01 -9.59029421e-02 -6.67723715e-01
-9.87177014e-01 9.48142409e-01 6.70534849e-01 2.07189783e-01
6.57799169e-02 1.63240567e-01 4.06621963e-01 -7.05613792e-02
5.07082164e-01 5.87783575e-01 -3.87013435e-01 2.14254856e-01
-9.08155024e-01 2.76314318e-01 5.91410995e-01 1.26867461e+00
7.83395886e-01 -1.59976766e-01 -3.84898573e-01 9.72894311e-01
2.68436939e-01 6.13270283e-01 3.88182312e-01 -6.77197635e-01
3.18129689e-01 6.14512742e-01 1.88765317e-01 -5.52939713e-01
-7.76993185e-02 -3.68126094e-01 -1.26752168e-01 1.00388639e-01
5.05192637e-01 1.29053861e-01 -1.46905863e+00 1.58919382e+00
4.79966134e-01 -4.86024395e-02 -8.79342780e-02 8.73649836e-01
8.52377832e-01 7.00063109e-01 1.22948781e-01 2.40999699e-01
1.44700325e+00 -1.08857763e+00 -1.36801943e-01 -4.89435464e-01
8.18852663e-01 -6.51614189e-01 1.23816121e+00 -6.78073289e-03
-6.64138913e-01 -6.87075496e-01 -9.99820948e-01 -6.30111992e-02
-5.23782790e-01 1.07139274e-01 2.26885512e-01 6.85224831e-01
-8.12724113e-01 8.61913785e-02 -5.15658200e-01 -5.67358971e-01
8.38716865e-01 2.60625064e-01 -1.93758652e-01 -2.36511230e-01
-7.18803465e-01 7.55628169e-01 7.20878005e-01 -3.72192323e-01
-1.18999803e+00 -6.96320057e-01 -8.40928853e-01 -6.06753603e-02
6.67096794e-01 -3.00388843e-01 1.48446059e+00 -1.01964033e+00
-7.58841574e-01 1.02633011e+00 9.11554992e-02 -5.53671360e-01
5.76843798e-01 -2.37054721e-01 -2.50639230e-01 2.30333537e-01
4.38643038e-01 1.32611966e+00 8.16334844e-01 -1.41723096e+00
-9.47805643e-01 -1.89390957e-01 1.69279307e-01 1.00325711e-01
-3.86139721e-01 9.82200503e-02 -6.23418808e-01 -5.39712667e-01
-2.00761352e-02 -1.06488931e+00 -2.73455024e-01 2.67446011e-01
-2.33820438e-01 -5.03125370e-01 9.42959726e-01 -3.78700286e-01
7.48662591e-01 -2.16534472e+00 -3.51585031e-01 -7.45496973e-02
2.63589919e-01 5.62961698e-01 -3.91918004e-01 2.76468452e-02
2.60825306e-01 1.37332767e-01 -5.62880561e-02 -3.36637944e-01
-1.27425179e-01 1.47038579e-01 -6.15632832e-01 3.22599620e-01
3.06167781e-01 8.10552359e-01 -1.09946167e+00 -5.57086766e-01
2.20160395e-01 3.27800602e-01 -6.16632342e-01 1.13870099e-01
-5.21000028e-01 2.82897800e-01 -2.64674067e-01 6.59824014e-01
5.27966619e-01 -2.56488323e-01 -1.09147564e-01 2.02112347e-02
-2.64431071e-02 1.19324371e-01 -1.19517875e+00 1.51202869e+00
-2.49898151e-01 7.97871053e-01 -3.05983901e-01 -7.19695866e-01
9.10136878e-01 8.18351135e-02 1.57649532e-01 -3.13112944e-01
2.88940012e-01 3.09515089e-01 1.43191248e-01 -2.56958991e-01
5.35520375e-01 1.75959710e-02 -1.69670716e-01 2.05277309e-01
2.69837767e-01 -2.92306155e-01 4.82830882e-01 3.59947681e-01
1.04394901e+00 5.89440167e-02 1.75858364e-01 -1.21644042e-01
2.80210942e-01 2.77128965e-01 5.16076803e-01 1.03207242e+00
-5.11968672e-01 8.75560820e-01 8.28666911e-02 -4.72162932e-01
-1.50100589e+00 -1.19529760e+00 -4.55061525e-01 1.40702474e+00
1.91513062e-01 -3.74045879e-01 -7.43455887e-01 -9.77068782e-01
1.41278476e-01 8.44650090e-01 -6.30026996e-01 -5.47678545e-02
-3.33097398e-01 -2.62122661e-01 6.49721861e-01 7.39607692e-01
1.98294207e-01 -9.87178206e-01 -6.41322970e-01 6.73165619e-02
-1.23640690e-02 -1.24295795e+00 -6.15970314e-01 8.53044465e-02
-5.52025080e-01 -9.73794520e-01 -7.33611345e-01 -9.84550059e-01
8.07065547e-01 5.43014288e-01 9.91552234e-01 -4.19204868e-02
-6.43313646e-01 3.94317269e-01 -5.20429671e-01 -7.51492679e-01
-5.82644999e-01 -7.34556094e-02 1.39375865e-01 -5.53242415e-02
7.18876362e-01 -6.05144650e-02 -5.30913711e-01 4.88794535e-01
-7.13378608e-01 4.95885648e-02 5.40585995e-01 7.79019058e-01
6.27828598e-01 -5.42808115e-01 6.11979246e-01 -6.56832039e-01
9.42726508e-02 -4.07174587e-01 -8.83701682e-01 7.61886388e-02
-4.45934296e-01 -1.10744890e-02 3.37972671e-01 -8.55008662e-01
-7.96358407e-01 3.68982464e-01 1.30331621e-01 -7.90979564e-01
-3.29596430e-01 -8.23853463e-02 1.45962089e-01 -1.87106244e-02
1.13797879e+00 1.47654176e-01 -2.91680425e-01 -3.55273306e-01
7.24984765e-01 9.23746288e-01 5.44528663e-01 -2.05034405e-01
9.71733868e-01 6.09666288e-01 -6.12968504e-01 -7.76565254e-01
-1.14196515e+00 -1.01936162e+00 -6.72529638e-01 -1.38292357e-01
9.08491194e-01 -1.14145064e+00 8.09156988e-03 2.32701227e-02
-1.13275003e+00 -2.61817396e-01 -4.61345077e-01 2.92865336e-01
-4.10771340e-01 1.84305042e-01 -2.94344902e-01 -7.42383599e-01
-4.73474324e-01 -8.95613253e-01 1.23693526e+00 1.73636571e-01
-3.08369040e-01 -4.47966129e-01 -4.78661433e-02 5.05516708e-01
1.83356628e-01 -1.09768054e-02 3.65114450e-01 -1.03846371e+00
-7.95250237e-01 -6.37894571e-01 -5.18019021e-01 5.10526359e-01
-8.68784562e-02 -2.77255088e-01 -1.10686946e+00 -3.58140677e-01
-3.53344560e-01 -5.94266653e-01 9.89794672e-01 2.72810906e-02
1.08781791e+00 2.04005148e-02 -7.89642692e-01 4.27701592e-01
1.27894378e+00 1.24211967e-01 3.07580650e-01 1.10561870e-01
6.99256718e-01 4.25943792e-01 7.75221705e-01 3.19266379e-01
1.57739654e-01 7.13275254e-01 2.70925224e-01 1.07042372e-01
-3.93057585e-01 -3.38885307e-01 7.32166469e-02 1.48790464e-01
2.27131933e-01 -1.79549739e-01 -1.27974117e+00 1.11704838e+00
-1.70131397e+00 -8.47937644e-01 -2.62942351e-02 2.24402642e+00
9.38988924e-01 5.40194869e-01 1.62086025e-01 -2.98693329e-01
9.42789912e-01 -1.71090931e-01 -6.10094726e-01 -1.18320920e-01
1.52315497e-01 3.31160449e-03 6.21617138e-01 2.78345346e-01
-1.36817968e+00 1.21987903e+00 5.52870035e+00 8.54998767e-01
-8.97484481e-01 3.96910250e-01 4.02854890e-01 -4.47491050e-01
1.43952638e-01 2.33959313e-02 -1.35662007e+00 3.00940484e-01
6.92684233e-01 -1.58377960e-01 2.48054974e-02 1.22605968e+00
-1.42650530e-01 -7.73991197e-02 -1.30690181e+00 1.04123926e+00
3.54893118e-01 -1.22120333e+00 4.31131907e-02 -7.33189238e-03
9.29704249e-01 4.80657101e-01 -1.21508956e-01 6.68944657e-01
2.87981391e-01 -7.48047709e-01 7.89656937e-01 -3.75291333e-02
1.03560269e+00 -3.82945806e-01 4.61458117e-01 3.50666076e-01
-1.15099871e+00 -2.08348542e-01 -5.93400836e-01 -3.90417241e-02
-6.71005175e-02 1.88784570e-01 -1.36690021e+00 -2.19522402e-01
7.75188208e-01 2.92965144e-01 -7.71004200e-01 1.26308692e+00
-3.95448118e-01 7.15983629e-01 -5.91413915e-01 -1.80464879e-01
3.07339996e-01 3.58609378e-01 6.16373777e-01 1.22361743e+00
4.72319610e-02 1.26474336e-01 5.06262600e-01 1.04709041e+00
-3.05817187e-01 1.21787205e-01 -7.37376511e-01 -4.40814979e-02
6.29839599e-01 1.32573998e+00 -8.60712051e-01 -6.28870726e-01
-5.95611453e-01 8.12846303e-01 4.40656245e-01 6.24625869e-02
-1.05138648e+00 -5.98805249e-01 4.57161695e-01 2.83900231e-01
6.32217050e-01 -1.88960314e-01 -1.52585939e-01 -1.04682171e+00
1.10054381e-01 -6.06707871e-01 3.88512164e-01 -7.19320536e-01
-1.24106491e+00 5.34969866e-01 8.60564411e-02 -1.13994825e+00
-6.74921274e-02 -5.93170106e-01 -4.06024486e-01 6.45627439e-01
-1.11735046e+00 -1.41302514e+00 -4.12910461e-01 2.52484530e-01
1.00468540e+00 -3.50247175e-02 7.03624964e-01 2.91274548e-01
-2.28742093e-01 6.79717779e-01 2.42957518e-01 4.61918324e-01
8.08282077e-01 -1.12788248e+00 7.85443008e-01 9.42219615e-01
5.58178842e-01 3.76032829e-01 7.74639726e-01 -7.34597743e-01
-9.51967180e-01 -1.36281908e+00 7.04396367e-01 -1.07188237e+00
6.47363663e-01 -9.81245577e-01 -1.04632139e+00 7.71370232e-01
-2.14546323e-01 5.35857618e-01 4.34565991e-01 9.72718075e-02
-8.60990942e-01 9.64784026e-02 -9.37748909e-01 6.88824952e-01
1.14331222e+00 -4.44192111e-01 -8.80079746e-01 6.95338488e-01
9.20242906e-01 -4.54040706e-01 -5.03726244e-01 3.90466422e-01
4.42964673e-01 -4.73141134e-01 9.51255441e-01 -6.33953750e-01
2.05883726e-01 -4.85721499e-01 -2.19025701e-01 -8.78964365e-01
-3.30698550e-01 -1.63781002e-01 -8.41816701e-03 1.14391387e+00
5.78116357e-01 -4.02151793e-01 7.41792798e-01 6.38160467e-01
-2.17353538e-01 -5.13819158e-01 -7.62708247e-01 -1.19691944e+00
4.45837565e-02 -5.53044438e-01 6.99584633e-02 6.90922379e-01
-2.11122230e-01 5.94106376e-01 1.24936718e-02 1.86372504e-01
6.58526003e-01 5.39391823e-02 1.11841357e+00 -1.04643822e+00
-2.76997745e-01 -3.13960850e-01 -6.37599707e-01 -9.68068838e-01
-3.16447541e-02 -9.64644969e-01 4.64744508e-01 -1.30749381e+00
4.96640980e-01 -7.32886851e-01 -2.71566987e-01 7.62973726e-01
-1.74668670e-01 8.93304229e-01 4.36577588e-01 4.04769897e-01
-9.46981609e-01 2.94773281e-01 8.06225002e-01 -2.78193384e-01
-3.80381286e-01 -2.32630819e-01 -5.39947808e-01 7.86056995e-01
8.14711094e-01 -7.44559944e-01 -3.29762995e-01 -3.55385244e-01
-1.75818935e-01 -6.76997840e-01 5.26822925e-01 -1.25459492e+00
5.16093932e-02 -2.57888599e-03 3.33384097e-01 -7.01266587e-01
4.01131570e-01 -4.85658944e-01 -3.50125909e-01 4.36745733e-01
-3.32739830e-01 -6.07492089e-01 4.26508635e-01 8.08455825e-01
8.91417712e-02 -3.85742038e-01 1.02634037e+00 -1.16607182e-01
-1.13813043e+00 2.55234838e-01 -1.85972288e-01 2.87466615e-01
1.43819356e+00 -4.08360720e-01 -2.82871693e-01 -6.67003021e-02
-6.51354492e-01 2.92236745e-01 5.92977405e-01 8.35584939e-01
4.67588514e-01 -1.05747902e+00 -6.86669350e-01 1.47924647e-01
7.39216566e-01 2.14053780e-01 -3.75288837e-02 4.01598632e-01
-3.57420832e-01 4.46968108e-01 -1.47933830e-02 -9.17474270e-01
-1.39400434e+00 9.01384890e-01 3.87510240e-01 1.84297442e-01
-6.65062129e-01 1.16908181e+00 7.40376472e-01 -2.71585703e-01
3.00905436e-01 -1.71948761e-01 1.28056884e-01 -3.33977267e-02
6.33447766e-01 8.26727748e-02 -2.04154298e-01 -6.82351172e-01
-4.40842271e-01 3.86953861e-01 -4.68725175e-01 -5.38797453e-02
1.01459718e+00 6.09841309e-02 4.42109674e-01 3.44131410e-01
1.10376024e+00 -1.35283172e-01 -1.47613049e+00 -4.70561564e-01
8.41030777e-02 -5.28103471e-01 -1.31108880e-01 -6.98028386e-01
-4.44405735e-01 7.69601583e-01 8.40806544e-01 -4.97261807e-03
6.13067031e-01 5.56412458e-01 6.21415198e-01 5.28902233e-01
5.27641237e-01 -1.09828031e+00 3.90163213e-01 5.34664571e-01
8.66273344e-01 -1.56679344e+00 -1.68876946e-01 -5.89121103e-01
-4.00302798e-01 8.60727668e-01 9.54673946e-01 -2.34896183e-01
4.22009081e-01 3.78508903e-02 5.40466346e-02 -5.55234924e-02
-5.16195655e-01 -4.77009922e-01 5.17097414e-01 6.22885346e-01
1.52011082e-01 1.08881943e-01 -9.00188833e-02 4.37742829e-01
3.17558949e-03 -1.86937720e-01 5.38735449e-01 1.00937974e+00
-8.94848049e-01 -8.13753605e-01 -4.22647715e-01 6.17438555e-01
-2.82426268e-01 -2.43322536e-01 -3.42991084e-01 7.47336447e-01
2.51573086e-01 9.27320838e-01 3.25713158e-01 -5.49476072e-02
2.77386874e-01 1.54781386e-01 3.72643501e-01 -1.32777452e+00
-2.28730127e-01 -6.84589744e-02 1.05041943e-01 -3.27922016e-01
-3.59618850e-02 -6.99210644e-01 -1.22928572e+00 3.26491833e-01
-8.42209160e-01 -1.56955034e-01 6.18321359e-01 6.84024215e-01
4.32726741e-01 2.34308958e-01 2.15726912e-01 -8.49350929e-01
-7.77611852e-01 -1.12463832e+00 -1.84305757e-01 6.61435544e-01
2.35479444e-01 -7.38431275e-01 -2.76061147e-01 2.43762761e-01] | [9.537271499633789, 1.4344860315322876] |
5386ae0a-4b70-4ae1-9ef1-9d73ccd17429 | will-it-blend-mixing-training-paradigms | 2209.08966 | null | https://arxiv.org/abs/2209.08966v2 | https://arxiv.org/pdf/2209.08966v2.pdf | Will It Blend? Mixing Training Paradigms & Prompting for Argument Quality Prediction | This paper describes our contributions to the Shared Task of the 9th Workshop on Argument Mining (2022). Our approach uses Large Language Models for the task of Argument Quality Prediction. We perform prompt engineering using GPT-3, and also investigate the training paradigms multi-task learning, contrastive learning, and intermediate-task training. We find that a mixed prediction setup outperforms single models. Prompting GPT-3 works best for predicting argument validity, and argument novelty is best estimated by a model trained using all three training paradigms. | ['Selene Báez Santamaría', 'Lea Krause', 'Urja Khurana', 'Myrthe Reuver', 'Michiel van der Meer'] | 2022-09-19 | null | https://aclanthology.org/2022.argmining-1.8 | https://aclanthology.org/2022.argmining-1.8.pdf | argmining-acl-2022-10 | ['argument-mining'] | ['natural-language-processing'] | [-1.28054559e-01 8.18747699e-01 -8.70877862e-01 -3.34967762e-01
-1.35611522e+00 -4.49666113e-01 8.92398596e-01 8.83190870e-01
-4.53313202e-01 9.46332157e-01 6.00977719e-01 -1.06514072e+00
-5.87404072e-01 -6.30967438e-01 -7.47765481e-01 8.68953317e-02
2.36062761e-02 1.17044199e+00 2.65408605e-01 -4.46071684e-01
7.94816017e-01 -2.73514062e-01 -1.33911562e+00 1.08253121e+00
1.13860309e+00 1.02786589e+00 -4.14501816e-01 8.86994123e-01
-6.87633753e-01 1.18057001e+00 -6.25558197e-01 -9.67813909e-01
-6.21027201e-02 2.85891518e-02 -1.54585564e+00 -9.91982639e-01
5.33278048e-01 1.19094536e-01 6.46842301e-01 5.76340377e-01
6.89941049e-01 -1.46209955e-01 4.98574704e-01 -1.25538969e+00
-6.68138862e-01 1.25912642e+00 -1.83777556e-01 4.89157259e-01
7.48354733e-01 -2.36445546e-01 1.52623582e+00 -8.35376978e-01
7.29530334e-01 1.70146596e+00 1.00319481e+00 3.21031749e-01
-1.34924662e+00 -3.60389560e-01 4.38150674e-01 1.98097840e-01
2.61932135e-01 -4.28802639e-01 9.23225760e-01 -5.61707497e-01
1.39364004e+00 1.24942392e-01 6.76019371e-01 1.47079158e+00
5.26872128e-02 9.87069249e-01 1.56824458e+00 -6.02303863e-01
1.12556487e-01 7.03423470e-02 8.33738327e-01 5.41169524e-01
4.24143136e-01 2.27522016e-01 -8.29083562e-01 -1.11703837e+00
-3.80763710e-02 -7.98585534e-01 4.46877927e-01 3.15791279e-01
-9.32802260e-01 1.17914522e+00 6.70587458e-03 4.04481858e-01
-3.66758019e-01 -6.26575947e-02 6.94281578e-01 9.79882658e-01
1.08905113e+00 8.26354861e-01 -1.27883995e+00 -6.33802354e-01
-6.84328854e-01 8.01269412e-01 1.24601197e+00 2.94876277e-01
1.57855347e-01 -2.86174089e-01 -4.74941254e-01 1.06688428e+00
3.52175266e-01 3.31436366e-01 2.74566799e-01 -1.06884491e+00
8.42493892e-01 6.18622303e-01 2.50803053e-01 -5.40881634e-01
-5.10721862e-01 -3.04358780e-01 2.10731089e-01 2.64739662e-01
6.65601075e-01 -3.19467247e-01 -3.26136619e-01 1.37095940e+00
2.92981356e-01 -3.99060667e-01 3.12864959e-01 1.32881299e-01
1.22602248e+00 2.34435469e-01 5.97176850e-01 -1.63872153e-01
1.22432327e+00 -1.13264799e+00 -5.96536815e-01 -4.06653941e-01
1.31228888e+00 -1.18551362e+00 1.21097898e+00 6.74922764e-01
-1.47415996e+00 -4.00162518e-01 -6.40075564e-01 -9.63812694e-02
-4.09925938e-01 -1.48034647e-01 7.13368714e-01 5.38519442e-01
-7.30583429e-01 8.30395639e-01 -3.22469547e-02 6.50428757e-02
5.04932642e-01 -3.24294269e-02 2.49598431e-03 3.47793251e-01
-1.36113453e+00 1.34318137e+00 2.49575824e-01 -4.54529434e-01
-2.10065648e-01 -6.51285648e-01 -6.29906952e-01 -3.40791255e-01
1.79731041e-01 -6.93243802e-01 1.70917845e+00 -8.31839204e-01
-1.24552274e+00 1.03809547e+00 -1.87524185e-01 -6.70918584e-01
5.14079213e-01 -8.19682896e-01 -4.19386357e-01 -4.58882779e-01
5.69317400e-01 1.69184625e-01 6.97550654e-01 -9.78251159e-01
-6.90860629e-01 -4.98226322e-02 2.36969233e-01 -6.64052665e-02
-1.12396553e-01 5.14513969e-01 7.56846547e-01 -3.96007717e-01
-3.64999354e-01 -5.48654258e-01 -1.76492959e-01 -4.76897627e-01
-2.36971125e-01 -1.22994125e+00 7.71030664e-01 -6.93714738e-01
1.29066622e+00 -1.62520301e+00 -3.10405761e-01 7.50517026e-02
1.82203695e-01 5.31532466e-01 -2.19953910e-01 3.21531266e-01
-1.44405410e-01 6.18126988e-01 4.91155565e-01 -1.58082083e-01
3.98637950e-02 2.19551086e-01 -2.99625546e-01 -3.59769255e-01
2.00018510e-01 1.19976103e+00 -1.06373847e+00 -9.21639264e-01
-2.47525483e-01 -1.42490223e-01 -7.18686700e-01 1.11866735e-01
-7.51675487e-01 8.60766694e-02 -7.58250535e-01 6.26375735e-01
3.76103044e-01 -5.58927357e-01 1.83719128e-01 1.12008996e-01
-2.50911176e-01 1.44154739e+00 -5.95016360e-01 1.46252549e+00
-4.50215876e-01 2.76046544e-01 -1.24873273e-01 -1.05921364e+00
9.87004280e-01 3.75962347e-01 4.81651872e-02 -1.16026318e+00
-8.66601244e-03 8.58036578e-01 2.98867702e-01 -6.46020889e-01
1.45049646e-01 -9.85111818e-02 -9.32174847e-02 8.85140181e-01
-1.66218504e-01 -1.57336425e-02 2.80887067e-01 2.97385603e-01
1.02690244e+00 2.61914074e-01 2.91018963e-01 -5.95151424e-01
1.97402284e-01 3.55164379e-01 3.93056333e-01 9.81840491e-01
4.14004996e-02 -1.82003409e-01 9.19507921e-01 -1.00018060e+00
-8.92676651e-01 -4.87313360e-01 -3.67714852e-01 1.46223724e+00
-3.85403067e-01 -6.56007528e-01 -1.81803361e-01 -1.42472970e+00
3.81902397e-01 1.02564061e+00 -8.97105515e-01 2.71341681e-01
-8.87229562e-01 -6.79375947e-01 5.10755777e-01 3.56182367e-01
2.21615463e-01 -1.39158618e+00 -8.16256344e-01 5.32622039e-01
-4.44110632e-01 -6.75421178e-01 2.97510177e-01 7.62499630e-01
-1.23622668e+00 -1.44616783e+00 -1.73026025e-02 -7.80056059e-01
-1.36554226e-01 -3.15705359e-01 1.83482742e+00 7.16634154e-01
2.03661710e-01 6.99804202e-02 -5.88704109e-01 -6.80486143e-01
-6.52644753e-01 6.33678958e-02 -4.11902308e-01 -9.33457375e-01
3.67661685e-01 -3.85659218e-01 -8.28253031e-02 -6.93613887e-02
-7.18687475e-02 5.38124368e-02 3.46941471e-01 1.25506032e+00
1.56061262e-01 -5.55970192e-01 1.02586913e+00 -1.33998919e+00
1.41915977e+00 -6.21827424e-01 -4.79505025e-03 4.93040860e-01
-1.39833236e+00 4.19769853e-01 2.82469124e-01 -4.48402196e-01
-1.30628312e+00 -1.17959929e+00 -5.85203707e-01 7.24621415e-01
2.01946363e-01 7.59233236e-01 2.95971304e-01 1.17577665e-01
1.07485652e+00 -8.13599586e-01 1.93184894e-02 -6.89244449e-01
4.33027506e-01 3.02745223e-01 -1.66345984e-01 -1.44531000e+00
4.07440662e-01 3.86511870e-02 -2.30837137e-01 -1.50517344e-01
-1.58414447e+00 -1.71975829e-02 -2.54593134e-01 -3.09679396e-02
3.43035668e-01 -4.55853134e-01 -5.58067739e-01 -2.88351090e-03
-1.46009636e+00 -6.78538978e-01 -5.64535558e-01 2.61890471e-01
-3.96402836e-01 2.67929584e-01 -7.08181977e-01 -7.67223239e-01
-9.30662990e-01 -6.74146473e-01 8.68850231e-01 -1.06720977e-01
-7.71093249e-01 -1.57347369e+00 6.65457964e-01 6.53055191e-01
4.74046290e-01 1.59093529e-01 1.52065897e+00 -1.40845287e+00
2.62111396e-01 1.41574636e-01 -1.62406147e-01 1.85220659e-01
-3.52354616e-01 -7.80766504e-03 -8.03902626e-01 2.47066006e-01
-8.56180862e-02 -1.05201781e+00 9.45771217e-01 5.65505683e-01
8.62834692e-01 -4.00905997e-01 -3.26131105e-01 -1.29672542e-01
7.23532438e-01 -1.05090640e-01 2.98617721e-01 1.25233138e+00
1.31702289e-01 8.28000605e-01 1.22686851e+00 1.36677265e-01
4.59651351e-01 3.71180326e-01 2.84289062e-01 -2.49473210e-02
-1.30745173e-01 -4.24435049e-01 2.11389303e-01 2.66836405e-01
-2.19108343e-01 -2.13255405e-01 -1.13516510e+00 4.54793423e-01
-2.20590663e+00 -1.12048483e+00 -8.53141129e-01 1.61464250e+00
8.75160277e-01 8.63421082e-01 2.25884393e-01 3.72446686e-01
1.47206753e-01 1.71799794e-01 -1.84475049e-01 -1.20611203e+00
-4.58276808e-01 6.26724899e-01 -1.39714405e-01 6.88213646e-01
-9.29472804e-01 8.27162027e-01 7.62427759e+00 7.12592006e-01
-2.86503971e-01 4.55416769e-01 6.93589449e-01 1.46145806e-01
-9.29975271e-01 7.13365734e-01 -6.78867459e-01 2.72403330e-01
1.09990621e+00 -1.69164285e-01 -2.93699801e-01 9.70222712e-01
3.99140343e-02 -1.89149961e-01 -9.30509448e-01 2.38991067e-01
-1.03008881e-01 -1.33633184e+00 1.28501996e-01 1.02680713e-01
6.84672177e-01 1.47445619e-01 9.84144285e-02 6.99175060e-01
7.26286292e-01 -8.70899916e-01 8.18984091e-01 2.76005387e-01
5.23090959e-02 -2.91332304e-01 8.88909221e-01 5.83471477e-01
-6.82522714e-01 -5.26781619e-01 -1.07004747e-01 -6.54822528e-01
3.62302810e-01 9.23226178e-01 -6.68434143e-01 4.43561614e-01
7.20418513e-01 8.14754188e-01 -5.78701138e-01 8.49355757e-01
-5.22931576e-01 1.01783299e+00 -1.59893528e-01 -2.59173930e-01
3.32576871e-01 2.72046089e-01 5.39522588e-01 1.00737631e+00
2.58721225e-02 -1.14794932e-01 3.62656415e-01 3.19259226e-01
-9.33301300e-02 3.53263706e-01 -4.12637830e-01 1.56140745e-01
3.93120795e-01 9.41811800e-01 -2.15065479e-01 -7.58316398e-01
-2.90189117e-01 1.03147253e-01 7.01275408e-01 -1.99696928e-01
-5.19051850e-01 3.11158419e-01 1.28570586e-01 2.32028335e-01
-2.00999901e-01 4.35996085e-01 -8.70367646e-01 -8.43260467e-01
5.56270182e-02 -1.24730372e+00 8.26007187e-01 -6.07913673e-01
-1.62537503e+00 5.51903009e-01 2.04696164e-01 -6.40589297e-01
-4.54304814e-01 -6.98187649e-01 -1.03471518e+00 8.04661632e-01
-1.95213187e+00 -1.33631849e+00 2.61801749e-01 1.46010622e-01
5.45060277e-01 -2.97981024e-01 9.84059334e-01 9.70649570e-02
3.75159010e-02 3.14243376e-01 -4.55164313e-01 -3.39718461e-01
1.00803590e+00 -1.55958390e+00 7.70194173e-01 9.55804139e-02
1.25588641e-01 4.27352846e-01 6.98904574e-01 -1.05115139e+00
-6.28817260e-01 -5.06202936e-01 1.83516443e+00 -1.14725065e+00
9.55217123e-01 3.36686432e-01 -8.68304193e-01 4.39335823e-01
3.90703380e-01 -3.76510471e-01 9.26458955e-01 1.24359524e+00
-7.89328158e-01 5.32789171e-01 -9.37235653e-01 1.11229047e-02
1.06777847e+00 -4.99935895e-01 -1.35585439e+00 5.90655327e-01
6.56129062e-01 -1.93409771e-01 -9.62805450e-01 6.19426787e-01
7.55530119e-01 -8.85091126e-01 9.91444886e-01 -1.22785854e+00
6.79849684e-01 5.25719345e-01 2.27902725e-01 -1.13704562e+00
-2.67763585e-01 -6.18322670e-01 -3.54694486e-01 1.00952256e+00
1.16597676e+00 -7.46238947e-01 8.08338523e-01 4.45816875e-01
-1.62583873e-01 -1.03227699e+00 -7.73962736e-01 -6.45066261e-01
8.65970790e-01 -5.23474514e-01 5.44767201e-01 8.73359919e-01
2.42420375e-01 8.38615656e-01 -1.67004973e-01 -6.06453836e-01
5.64770699e-01 5.13136923e-01 7.24050939e-01 -2.24728227e+00
-3.62796903e-01 -6.46780193e-01 6.23334825e-01 -8.77405822e-01
4.93318975e-01 -9.46937084e-01 -4.56805229e-01 -1.94655824e+00
2.01899946e-01 -8.21922004e-01 -9.04502571e-02 5.86212397e-01
-4.57153678e-01 -5.42763412e-01 -2.08414748e-01 1.79265678e-01
-7.14149415e-01 1.70294940e-01 1.05126393e+00 -5.82393073e-02
-1.28499374e-01 2.32553855e-01 -9.10080850e-01 8.88440728e-01
1.20740807e+00 -9.30573702e-01 -2.60982901e-01 -6.61302209e-01
1.02010000e+00 -5.33050485e-02 4.38380033e-01 -3.20325971e-01
6.10707141e-02 -1.34356216e-01 7.76179433e-02 -7.04638481e-01
5.62978014e-02 -1.80801466e-01 -4.47087377e-01 5.43213010e-01
-8.86540413e-01 4.00135189e-01 1.96054801e-01 3.46870095e-01
-1.23743691e-01 -5.60380518e-01 2.99992263e-01 -1.64176822e-01
1.12014273e-02 -2.44538024e-01 -5.96487075e-02 8.03641617e-01
3.31422985e-01 -1.12200901e-01 -1.06705582e+00 4.79161181e-02
-7.48705804e-01 4.47184950e-01 -1.15894768e-02 5.07986724e-01
5.79349816e-01 -1.04897368e+00 -1.16505587e+00 -2.19963282e-01
1.49133250e-01 -5.01071274e-01 -2.80785531e-01 1.06782246e+00
1.71469077e-01 4.48351622e-01 1.81352124e-01 -2.07912028e-01
-1.51813412e+00 2.10740477e-01 -6.72559887e-02 -1.05128574e+00
-4.53154147e-01 1.04850769e+00 -7.37199605e-01 -7.28236079e-01
-1.68581679e-02 -1.77348167e-01 -7.00452328e-01 2.79498637e-01
2.45324925e-01 4.42335367e-01 1.18670292e-01 -7.12431818e-02
-7.76397660e-02 3.22972566e-01 -3.07871759e-01 -2.29535550e-01
1.47267199e+00 5.06720960e-01 -2.02848941e-01 6.89741790e-01
7.68885672e-01 1.49901301e-01 -4.25429285e-01 -3.15414429e-01
8.71620893e-01 -2.17947692e-01 1.47459246e-02 -1.29220390e+00
-6.43359348e-02 7.99793482e-01 1.33750081e-01 6.73978627e-01
3.26680899e-01 2.84304082e-01 6.64359689e-01 6.28837466e-01
2.99094975e-01 -1.51471531e+00 5.10745823e-01 9.09895539e-01
8.77456546e-01 -1.43188989e+00 5.61325252e-02 -7.09379166e-02
-5.41488707e-01 1.17576039e+00 5.77178776e-01 -9.80032310e-02
9.41858113e-01 2.38773316e-01 1.82272648e-04 -8.85382473e-01
-1.52936172e+00 3.16676078e-03 3.17531943e-01 4.17175055e-01
9.85828340e-01 -8.64406526e-02 -9.51524675e-01 7.34026730e-01
-4.30219531e-01 -1.84732839e-01 2.71637104e-02 1.11909592e+00
-6.88627839e-01 -1.74797940e+00 -1.55541763e-01 8.73037577e-01
-8.03317070e-01 -5.33713937e-01 -8.73415172e-01 7.34010160e-01
1.53661743e-01 1.34156287e+00 -4.65066791e-01 -1.26357168e-01
4.95456427e-01 4.24449682e-01 2.44123295e-01 -6.87950432e-01
-1.35857093e+00 -1.46442607e-01 1.40579498e+00 -5.29407322e-01
-7.90472627e-01 -9.24737334e-01 -9.90040660e-01 -3.20975900e-01
-5.78565001e-01 1.03073299e+00 5.68586528e-01 1.00482845e+00
1.40017539e-01 2.89329767e-01 2.37495109e-01 -1.00163519e-01
-8.30814004e-01 -1.17432821e+00 9.58638042e-02 2.38712177e-01
1.80875659e-01 -8.63310635e-01 -6.07975960e-01 -2.89246589e-01] | [9.601283073425293, 9.616650581359863] |
3764058d-ffef-4130-a423-fb35198c264d | utilizing-bidirectional-encoder | 2011.07208 | null | https://arxiv.org/abs/2011.07208v1 | https://arxiv.org/pdf/2011.07208v1.pdf | Utilizing Bidirectional Encoder Representations from Transformers for Answer Selection | Pre-training a transformer-based model for the language modeling task in a large dataset and then fine-tuning it for downstream tasks has been found very useful in recent years. One major advantage of such pre-trained language models is that they can effectively absorb the context of each word in a sentence. However, for tasks such as the answer selection task, the pre-trained language models have not been extensively used yet. To investigate their effectiveness in such tasks, in this paper, we adopt the pre-trained Bidirectional Encoder Representations from Transformer (BERT) language model and fine-tune it on two Question Answering (QA) datasets and three Community Question Answering (CQA) datasets for the answer selection task. We find that fine-tuning the BERT model for the answer selection task is very effective and observe a maximum improvement of 13.1% in the QA datasets and 18.7% in the CQA datasets compared to the previous state-of-the-art. | ['Jimmy Xiangji Huang', 'Enamul Hoque', 'Md Tahmid Rahman Laskar'] | 2020-11-14 | null | null | null | null | ['answer-selection'] | ['natural-language-processing'] | [ 6.91169053e-02 1.73981100e-01 1.68891326e-02 -4.77021128e-01
-1.26108384e+00 -4.94221121e-01 6.06015027e-01 2.47181371e-01
-4.96985197e-01 6.23445630e-01 6.08535349e-01 -8.06550384e-01
5.74384518e-02 -9.24126208e-01 -5.61338127e-01 -2.50304580e-01
1.62630782e-01 6.58856034e-01 4.47115690e-01 -6.32523835e-01
1.87480420e-01 -8.03555697e-02 -1.25726819e+00 7.47234523e-01
1.15765667e+00 8.43205035e-01 3.60788763e-01 6.40705645e-01
-7.16336608e-01 1.27744102e+00 -5.19602120e-01 -7.52474427e-01
-7.78529048e-02 -6.70822859e-01 -1.27956283e+00 -3.13201010e-01
2.44568750e-01 -5.19214272e-02 -2.87200332e-01 4.10407215e-01
6.08441234e-01 6.18208200e-02 3.21140379e-01 -6.74409509e-01
-8.14013124e-01 7.15635717e-01 -1.57712340e-01 2.44904697e-01
3.17701072e-01 -1.95253447e-01 1.39117146e+00 -1.04440105e+00
6.25980556e-01 1.55241787e+00 4.71119672e-01 4.38760877e-01
-1.26499212e+00 -5.74174821e-01 5.19624539e-02 5.28887749e-01
-1.03591979e+00 -4.96290475e-01 6.06460989e-01 -2.81555831e-01
1.50243843e+00 1.76269233e-01 1.85832083e-01 7.92508006e-01
3.59139681e-01 9.19453204e-01 1.03326035e+00 -7.59486437e-01
4.18463051e-02 1.51139468e-01 3.67127717e-01 6.96893036e-01
-4.08126205e-01 -3.26335698e-01 -5.34195542e-01 -1.23901270e-01
1.18954912e-01 -1.68110549e-01 -1.28799453e-01 -2.41445284e-02
-9.86413002e-01 1.31427932e+00 6.08154893e-01 4.18171585e-01
-2.44372770e-01 8.73918906e-02 5.80440640e-01 8.22130144e-01
7.65888929e-01 5.55116057e-01 -6.48483276e-01 -1.23011470e-01
-7.66901314e-01 1.60301089e-01 6.12912536e-01 7.09035814e-01
7.14249849e-01 -3.01598221e-01 -6.77699208e-01 1.24186790e+00
2.59875089e-01 1.49671957e-01 5.04729748e-01 -8.86264324e-01
9.15751398e-01 8.99550557e-01 -9.41211134e-02 -3.93148690e-01
-2.43761018e-01 -6.42502844e-01 -7.80644774e-01 -4.46195990e-01
5.37538707e-01 -1.60730630e-01 -8.82586718e-01 1.75526547e+00
1.14540249e-01 -2.40229651e-01 1.91393986e-01 5.27589440e-01
9.49960828e-01 9.41785634e-01 1.31254330e-01 -5.07099144e-02
1.37915444e+00 -1.47445381e+00 -7.72297621e-01 -4.00031626e-01
1.04431379e+00 -8.06458712e-01 1.41469491e+00 8.45067352e-02
-9.80037928e-01 -7.48923302e-01 -6.87396467e-01 -5.50225854e-01
-3.14667076e-01 1.47539735e-01 5.18772125e-01 2.72832155e-01
-1.08814132e+00 3.35689969e-02 -4.79803056e-01 -4.38030750e-01
1.95721805e-01 2.35533118e-01 -2.67438531e-01 -4.19455469e-01
-1.57635128e+00 1.08798409e+00 -2.10611727e-02 -1.13384113e-01
-7.73798645e-01 -6.63286150e-01 -6.54056787e-01 3.03633660e-01
2.15990961e-01 -1.04890490e+00 1.31678772e+00 -6.94574475e-01
-1.63439035e+00 1.01822293e+00 -6.99683130e-01 -6.07246876e-01
1.48312584e-01 -3.19113702e-01 -6.98988363e-02 -3.61556187e-02
1.67413309e-01 7.12797284e-01 7.68078089e-01 -8.50016057e-01
-4.99902129e-01 -3.92648727e-01 5.69993019e-01 -3.42074037e-02
-3.71594101e-01 3.83076727e-01 -4.99635488e-01 -4.39001203e-01
-2.17460871e-01 -6.96467340e-01 -1.47010252e-01 -4.56310630e-01
-1.06017537e-01 -7.22635984e-01 5.65056801e-01 -8.90582025e-01
1.54064381e+00 -1.82829583e+00 2.33720765e-01 -3.85144144e-01
-1.25686005e-01 4.75243330e-01 -4.95450824e-01 7.51065254e-01
-6.88210353e-02 2.17815638e-01 -2.07465425e-01 -6.40263677e-01
-7.26175830e-02 2.39303008e-01 -6.74615026e-01 -3.61129120e-02
4.84575689e-01 1.12808168e+00 -7.67854989e-01 -3.23775053e-01
-3.41544189e-02 3.70583892e-01 -7.84245551e-01 5.90607285e-01
-4.72070098e-01 5.34333229e-01 -3.47018003e-01 2.55182236e-01
4.06611234e-01 -3.61740649e-01 1.13494076e-01 5.57306930e-02
1.08875297e-01 1.15099049e+00 -5.39384007e-01 1.79151440e+00
-1.11155927e+00 7.04865038e-01 6.14299579e-03 -1.06909287e+00
9.86747921e-01 4.61718976e-01 9.36904252e-02 -1.36567330e+00
-1.98020995e-01 2.42641777e-01 3.18642020e-01 -6.30781829e-01
4.45743471e-01 -3.37176740e-01 2.50424072e-02 5.43078363e-01
3.85571420e-01 -5.95972873e-02 2.46779263e-01 2.73507923e-01
1.30605292e+00 -1.46694750e-01 6.25087619e-02 -2.45296746e-01
1.07991529e+00 -9.89935845e-02 2.57129699e-01 5.90789974e-01
8.98528695e-02 4.86684531e-01 3.13580155e-01 -2.26149857e-01
-7.59469330e-01 -7.60006487e-01 -5.36147095e-02 1.53892469e+00
-4.73982513e-01 -7.12023437e-01 -6.84778869e-01 -8.40529203e-01
-1.90943271e-01 9.97143388e-01 -3.39118570e-01 -3.02004099e-01
-9.59756672e-01 -5.53465724e-01 4.71418500e-01 5.26596785e-01
6.48814857e-01 -1.09691226e+00 -2.72469848e-01 3.99417639e-01
-7.26451635e-01 -9.71622884e-01 -3.19209903e-01 2.87012219e-01
-9.44223225e-01 -7.46262372e-01 -5.69168568e-01 -8.12646210e-01
2.59734243e-01 7.46995881e-02 1.66998601e+00 3.32159281e-01
2.80024827e-01 1.82185665e-01 -5.57143211e-01 -1.74265653e-01
-4.64019150e-01 5.75188518e-01 -5.18760860e-01 1.46743089e-01
4.43804502e-01 -2.50391722e-01 -4.50500697e-01 1.86059117e-01
-8.17114711e-01 -3.98922600e-02 2.29166850e-01 1.08027661e+00
3.39207292e-01 -2.67087013e-01 1.20458937e+00 -1.00506496e+00
9.58788812e-01 -3.76023322e-01 -3.35648835e-01 5.65244496e-01
-4.52165723e-01 3.56294423e-01 6.83116972e-01 -4.15548459e-02
-1.19233036e+00 -4.37694699e-01 -7.84887493e-01 2.10255101e-01
2.12485448e-01 7.39746273e-01 -7.60574937e-02 3.33089828e-01
4.14426148e-01 -1.26181766e-01 -3.99878442e-01 -6.51113808e-01
5.68042815e-01 8.78044426e-01 1.26970679e-01 -4.92175758e-01
5.10276496e-01 1.68199748e-01 -3.68401617e-01 -5.52236199e-01
-1.33306670e+00 -7.83502519e-01 -5.56249261e-01 1.98012680e-01
9.16032195e-01 -9.47727203e-01 -4.27445263e-01 8.61586556e-02
-1.46881115e+00 -4.65903759e-01 -2.32891545e-01 1.47801608e-01
-3.79622489e-01 1.17070347e-01 -6.58321321e-01 -7.59631693e-01
-6.68825984e-01 -1.28543746e+00 1.08146250e+00 -2.77937520e-02
-2.47583508e-01 -1.21374321e+00 1.65746361e-01 9.75161314e-01
7.87927330e-01 -4.91573334e-01 1.46974695e+00 -5.96430659e-01
-6.90449476e-01 -1.10489056e-02 -2.13507161e-01 4.36389685e-01
5.05913310e-02 -4.74265635e-01 -1.16972315e+00 -2.55287141e-01
2.82704979e-01 -7.54725575e-01 1.31540072e+00 6.99677914e-02
1.06753778e+00 -6.18611760e-02 -8.87324288e-02 1.83839854e-02
1.11530769e+00 -5.07997386e-02 7.56882787e-01 3.33709508e-01
3.74735773e-01 7.59165406e-01 6.56757772e-01 -1.60528541e-01
7.89751410e-01 8.39775085e-01 2.96737462e-01 8.23158473e-02
-3.75405133e-01 -3.99188071e-01 5.48874319e-01 1.31171131e+00
3.74251604e-01 -4.46127802e-01 -8.57627809e-01 8.28784943e-01
-1.83136952e+00 -6.72874451e-01 -3.53036106e-01 2.21685815e+00
1.14422262e+00 8.35747123e-02 -2.18254134e-01 6.06525280e-02
3.32705587e-01 1.23103723e-01 -1.68819562e-01 -7.68302381e-01
-3.53303328e-02 7.35216498e-01 -4.37174514e-02 8.99096608e-01
-9.20587182e-01 1.10173130e+00 6.14091539e+00 9.54136908e-01
-8.55193794e-01 6.21172249e-01 6.92654848e-01 3.52337122e-01
-6.61163330e-01 3.55128467e-01 -8.93639565e-01 2.40070939e-01
1.43882000e+00 2.05480307e-01 2.66056478e-01 3.88720065e-01
1.89867616e-01 -1.04965329e-01 -1.19352973e+00 7.44971156e-01
1.03815213e-01 -1.38494122e+00 1.34683758e-01 -3.07126254e-01
5.94698012e-01 1.98343903e-01 -1.53833002e-01 1.07578063e+00
2.08394855e-01 -1.26816297e+00 4.48313296e-01 4.76358026e-01
4.73074555e-01 -6.65025651e-01 9.07352984e-01 6.99074268e-01
-9.34515417e-01 -1.85328811e-01 -5.67757130e-01 -3.19573432e-01
4.18392152e-01 8.55670750e-01 -7.41356373e-01 6.42894447e-01
6.48168683e-01 4.80432212e-01 -8.08805645e-01 7.98729241e-01
-6.10738277e-01 1.28040862e+00 -1.59910902e-01 1.51166722e-01
3.83779883e-01 -1.69454888e-01 1.27934575e-01 1.16673350e+00
1.42897531e-01 -2.43550941e-01 -1.20164521e-01 8.21862280e-01
-4.98270243e-01 2.46494755e-01 -3.03566009e-01 -9.91517119e-03
3.19419116e-01 1.10932505e+00 -1.57087818e-01 -4.11247283e-01
-5.91500819e-01 9.86184955e-01 6.78692877e-01 3.23744446e-01
-7.00710475e-01 -3.48488569e-01 3.52331370e-01 9.00872648e-02
4.42317575e-01 -1.54230937e-01 -1.77488759e-01 -1.32388926e+00
5.58056012e-02 -1.10457075e+00 4.99427557e-01 -6.77968681e-01
-1.34858263e+00 6.50474370e-01 -3.86715859e-01 -6.17336452e-01
-6.35931611e-01 -5.39957643e-01 -6.97732449e-01 1.18621385e+00
-1.88562965e+00 -1.41149795e+00 1.40862212e-01 4.72666383e-01
7.87580550e-01 -1.30505189e-01 1.05056930e+00 6.99173212e-01
-4.04076546e-01 6.06189907e-01 1.24017395e-01 8.79740417e-02
8.75392854e-01 -1.13124740e+00 3.75420660e-01 6.30537450e-01
3.75220865e-01 8.62025201e-01 3.62947255e-01 -1.08336382e-01
-1.46009994e+00 -1.08740449e+00 1.88731980e+00 -7.71297991e-01
6.81097388e-01 -7.13669360e-01 -1.11167860e+00 7.92278469e-01
7.14804649e-01 -2.83625901e-01 6.99195683e-01 4.71566856e-01
-4.67563778e-01 -2.54489183e-01 -6.76142216e-01 4.12289053e-01
7.41912365e-01 -1.02316201e+00 -9.83983934e-01 3.31175059e-01
1.04339683e+00 -1.31272525e-01 -8.02415729e-01 5.41572094e-01
6.10453635e-02 -9.72793281e-01 1.13682342e+00 -7.60382354e-01
5.20834684e-01 -1.45509571e-01 -1.21988781e-01 -1.34343112e+00
-2.76296824e-01 -3.17420751e-01 -1.48422062e-01 1.53106141e+00
6.20976627e-01 -5.84505916e-01 3.50136906e-01 2.57003307e-01
-1.64558798e-01 -9.07162726e-01 -1.01267588e+00 -4.07678753e-01
7.02940941e-01 -4.42976892e-01 4.69775140e-01 6.28184736e-01
-2.61130869e-01 1.16806126e+00 -1.73627198e-01 -3.43386322e-01
1.28774121e-01 4.31746036e-01 7.17599809e-01 -1.00082076e+00
-3.07993233e-01 -2.25282088e-01 2.31517807e-01 -1.45832014e+00
5.23688078e-01 -1.27283001e+00 -1.35522455e-01 -1.97839665e+00
2.00635672e-01 -5.21610081e-01 -2.25197420e-01 4.04145271e-01
-5.45878172e-01 -2.52354145e-02 2.05153212e-01 -5.89675382e-02
-6.82536066e-01 8.47880304e-01 1.09550917e+00 -3.56956750e-01
9.38143656e-02 3.62847261e-02 -8.05828273e-01 3.44054699e-01
7.49473095e-01 -5.29223800e-01 -4.87127453e-01 -9.74688649e-01
4.16641325e-01 -2.06106566e-02 1.98887229e-01 -4.26232636e-01
8.31416994e-02 3.11258346e-01 -8.74532014e-02 -6.32590055e-01
3.70864242e-01 -4.67977285e-01 -5.55582047e-01 4.53890234e-01
-6.11414135e-01 1.58369273e-01 2.24912941e-01 3.22201788e-01
-7.23158360e-01 -4.51980114e-01 5.08785605e-01 -1.83763444e-01
-5.01946092e-01 -1.20688781e-01 -4.88335848e-01 5.45518160e-01
3.20416629e-01 3.73901457e-01 -3.06742966e-01 -7.34125555e-01
-3.70373577e-01 4.79400665e-01 -2.07080066e-01 6.29509449e-01
4.73429918e-01 -1.19778728e+00 -9.75679696e-01 3.96329425e-02
2.06755087e-01 -9.82544795e-02 1.74341783e-01 8.28148842e-01
-1.40769467e-01 1.02859032e+00 2.73517638e-01 -5.11169016e-01
-1.06935275e+00 4.38497514e-01 4.36146796e-01 -1.06030154e+00
-2.62463301e-01 1.02687109e+00 2.52291083e-01 -1.02342057e+00
1.18643023e-01 -5.63934267e-01 -4.01928037e-01 6.55241758e-02
3.98371160e-01 1.34803444e-01 3.99604052e-01 -3.83258671e-01
-3.85890484e-01 5.08676410e-01 -4.80956212e-02 2.70791929e-02
1.27432585e+00 -2.91944772e-01 -4.83633518e-01 3.48641932e-01
1.35712552e+00 1.04681365e-01 -5.74996412e-01 -4.78945702e-01
3.67656589e-01 -1.30992711e-01 8.82984772e-02 -8.82885396e-01
-8.15763354e-01 1.49367523e+00 2.73279846e-01 1.37137339e-01
1.15475988e+00 2.09081888e-01 9.68738735e-01 6.11765921e-01
3.42621148e-01 -8.73691082e-01 2.34281987e-01 1.19896841e+00
1.14491785e+00 -1.28474665e+00 -5.71273506e-01 -3.12982082e-01
-4.22844529e-01 9.54118550e-01 3.98643911e-01 -9.65725034e-02
5.68267107e-01 5.50013781e-03 1.97315782e-01 -9.43018794e-02
-1.15981221e+00 -3.59256387e-01 3.38844538e-01 2.77756721e-01
1.00573528e+00 -1.19932450e-01 -4.43318903e-01 5.79978466e-01
-2.70436585e-01 3.35215367e-02 1.47912994e-01 7.05642760e-01
-3.00091773e-01 -1.60644162e+00 -1.75092295e-01 3.39776725e-01
-5.55089533e-01 -4.33478296e-01 -4.19338375e-01 3.01980376e-01
-2.39298921e-02 1.55374610e+00 -1.33672565e-01 -2.75411755e-01
3.75358790e-01 6.33639336e-01 3.87639642e-01 -8.24204445e-01
-1.04017746e+00 -3.00611913e-01 5.26925147e-01 -4.50447053e-01
-5.28781712e-01 -2.69862264e-01 -1.20503795e+00 -3.22140157e-02
-4.40890551e-01 3.76279861e-01 2.64537305e-01 1.20485425e+00
5.67170680e-01 6.72885358e-01 3.39926034e-01 -5.25921322e-02
-7.31545925e-01 -1.27860856e+00 1.46514490e-01 3.47895205e-01
3.12548846e-01 -3.82506579e-01 -1.18109964e-01 -8.77939314e-02] | [11.198731422424316, 8.309313774108887] |
a5557b4e-8496-42c5-b43b-7f8288b5a8b1 | ai4opt-ai-institute-for-advances-in | 2307.02671 | null | https://arxiv.org/abs/2307.02671v1 | https://arxiv.org/pdf/2307.02671v1.pdf | AI4OPT: AI Institute for Advances in Optimization | This article is a short introduction to AI4OPT, the NSF AI Institute for Advances in Optimization. AI4OPT fuses AI and Optimization, inspired by end-use cases in supply chains, energy systems, chip design and manufacturing, and sustainable food systems. AI4OPT also applies its "teaching the teachers" philosophy to provide longitudinal educational pathways in AI for engineering. | ['Kevin Dalmeijer', 'Pascal Van Hentenryck'] | 2023-07-05 | null | null | null | null | ['philosophy'] | ['miscellaneous'] | [-1.24649502e-01 1.72033846e-01 -6.17399633e-01 1.64954122e-02
1.85508117e-01 -3.58966619e-01 -3.97905223e-02 -2.78456528e-02
4.23248500e-01 4.91160452e-01 -4.19045389e-02 -3.85879457e-01
-5.75135052e-01 -6.29098654e-01 -9.10409749e-01 -3.87641758e-01
-1.49186566e-01 1.92350015e-01 -6.56052232e-01 -4.10492033e-01
3.03230762e-01 6.66553915e-01 -1.22454548e+00 -4.43284363e-02
9.99162197e-01 1.05068719e+00 1.38008147e-01 5.99762380e-01
-1.86048560e-02 6.09606206e-01 -2.87692010e-01 -2.30647892e-01
3.63288134e-01 -4.59978908e-01 -4.65763390e-01 1.20801941e-01
-1.58564463e-01 9.71121043e-02 -1.81955114e-01 1.12866771e+00
1.16045929e-01 1.13356531e-01 4.85837728e-01 -1.82657039e+00
-1.22699642e+00 9.92714942e-01 -3.89101297e-01 -8.15059915e-02
4.16772425e-01 6.90778613e-01 1.02313268e+00 -9.14684296e-01
3.97675902e-01 1.37695563e+00 4.90601987e-01 2.70828396e-01
-9.66751814e-01 -7.27817655e-01 3.56071562e-01 2.60539830e-01
-1.04041517e+00 -3.49572033e-01 8.92794371e-01 -2.41921619e-01
1.44092906e+00 4.42474931e-01 1.33249235e+00 5.09141147e-01
7.74029374e-01 1.29003811e+00 4.15104598e-01 -3.77218246e-01
5.30548513e-01 -4.48687710e-02 1.62005097e-01 4.85671937e-01
5.56691289e-01 3.42192948e-01 -5.78595400e-01 4.49296795e-02
6.29394889e-01 -1.12369359e-01 1.44857213e-01 -4.96913403e-01
-1.10406733e+00 7.62141049e-01 6.57828927e-01 2.11329356e-01
-5.79285204e-01 8.28620851e-01 -2.23705322e-01 6.12902403e-01
-7.84073174e-02 1.39419901e+00 -5.90167999e-01 -1.37052536e-01
-2.70594507e-01 4.61913973e-01 9.52302933e-01 1.41851628e+00
3.47144246e-01 8.18818271e-01 1.95883483e-01 3.49546969e-01
7.22397447e-01 6.87240779e-01 -3.28756645e-02 -1.77515328e+00
-1.86797038e-01 5.74609339e-01 2.20742524e-01 -8.90551507e-01
-5.00240743e-01 -6.49319172e-01 -4.70367134e-01 9.55240950e-02
-3.88010204e-01 -3.76094848e-01 -8.17584217e-01 1.12307477e+00
1.17773607e-01 -3.21988672e-01 6.77614138e-02 8.73095572e-01
4.56460297e-01 8.15818191e-01 -3.41954529e-02 -3.46696079e-01
8.08244109e-01 -1.57202685e+00 -1.03223705e+00 -2.31734946e-01
2.58688837e-01 -7.46200681e-01 5.60231388e-01 8.60984802e-01
-1.46307147e+00 -1.58511907e-01 -1.30799079e+00 3.12604100e-01
-4.80404496e-01 -2.58448988e-01 1.06700873e+00 6.99710071e-01
-9.41584945e-01 7.51459420e-01 -6.08741224e-01 -4.28833067e-01
3.94503981e-01 8.10432792e-01 5.11579096e-01 -1.30108282e-01
-7.42178738e-01 1.23891318e+00 5.33214994e-02 1.19085111e-01
-7.28465557e-01 -1.31277776e+00 -7.50970840e-01 2.69249827e-01
3.96333545e-01 -8.96305203e-01 1.35125792e+00 -9.01206195e-01
-2.16536355e+00 8.99766684e-02 1.82882130e-01 -2.96589315e-01
-2.56223261e-01 -5.66043437e-01 -4.71360147e-01 -2.95926452e-01
-2.81388789e-01 5.82249820e-01 5.08841872e-01 -1.02991009e+00
-5.14832497e-01 8.15204903e-03 -2.90585011e-01 1.25532672e-01
-2.64774352e-01 -9.57830623e-02 8.76095295e-02 -5.63741624e-01
-1.87363084e-02 -1.17259562e+00 -8.71019363e-01 -1.24581113e-01
-4.49386239e-01 -4.51391220e-01 9.22994614e-01 -5.10504544e-02
8.64552140e-01 -1.79930079e+00 5.81581295e-01 5.74469268e-01
9.55108628e-02 -2.77162701e-01 -1.05042882e-01 7.44676173e-01
6.53633326e-02 1.70720533e-01 2.45369315e-01 3.94891232e-01
5.44105768e-01 9.39652324e-02 -3.56196240e-02 2.22682178e-01
3.38910908e-01 1.00748670e+00 -1.11783946e+00 7.32232928e-02
3.83682311e-01 -2.23296568e-01 -7.40200877e-01 -1.22497655e-01
-6.32036746e-01 4.17683050e-02 -6.99048281e-01 1.33710194e+00
2.92833924e-01 -2.88139999e-01 3.59717280e-01 -3.01899672e-01
-5.44516087e-01 -2.80333787e-01 -8.08358908e-01 1.72387540e+00
-2.26993993e-01 5.67556560e-01 3.61318290e-01 -9.37263489e-01
7.90983438e-01 2.20803887e-01 1.28619158e+00 -6.11934185e-01
2.65312463e-01 5.94009578e-01 5.71098149e-01 -3.98764223e-01
4.36300427e-01 4.00925845e-01 -1.71819806e-01 2.17637837e-01
-5.71411736e-02 -1.16144025e+00 6.85847163e-01 5.24030440e-02
1.18825734e+00 2.51828343e-01 2.04568207e-02 -1.10588479e+00
3.20780218e-01 5.53415537e-01 5.83419502e-01 1.14279971e-01
-2.08155274e-01 -5.09842113e-02 -3.26230526e-02 -3.43757868e-01
-1.21853697e+00 -1.16884387e+00 -7.85497278e-02 1.00435114e+00
4.78821009e-01 -4.23004448e-01 -3.15889418e-01 -2.73463249e-01
5.10615826e-01 1.08684456e+00 -1.92842428e-02 -3.55879754e-01
-6.78879678e-01 -4.10815954e-01 -2.62742937e-01 4.61783648e-01
1.15221642e-01 -1.12018728e+00 -6.37252033e-01 4.66078520e-01
7.11995482e-01 -8.13401639e-01 -6.60994112e-01 4.13162231e-01
-7.05763638e-01 -7.74084270e-01 -3.33373427e-01 -9.45555747e-01
4.50035185e-01 8.16812739e-02 1.31525064e+00 6.26378730e-02
-7.44919837e-01 4.72197592e-01 -1.43907011e-01 -8.71261418e-01
-2.94784635e-01 -1.31366059e-01 4.81108665e-01 -9.46790814e-01
2.31432363e-01 -6.43464804e-01 -7.34763980e-01 2.72045881e-01
-2.03296781e-01 -2.20252685e-02 6.89275801e-01 6.55162156e-01
2.92248100e-01 3.09135288e-01 8.65886271e-01 -2.47833878e-01
4.63560998e-01 -7.06023037e-01 -8.05925131e-01 3.29239547e-01
-1.12976778e+00 8.50168243e-02 6.89143300e-01 -4.53837186e-01
-4.39989954e-01 2.08699524e-01 1.29515067e-01 -4.66773182e-01
4.33936775e-01 2.78632790e-01 -3.20630312e-01 -4.39205974e-01
5.83917461e-02 -2.18041450e-01 -9.59781632e-02 -1.29559398e-01
3.45676839e-01 1.84475303e-01 3.39545608e-01 -7.73162365e-01
8.10194433e-01 -2.39854962e-01 3.66234839e-01 -8.96338224e-01
-3.97669047e-01 -4.71838973e-02 1.57023251e-01 -4.44217324e-01
5.65105319e-01 -2.80223101e-01 -1.28238261e+00 1.33155301e-01
-8.25120032e-01 -2.50699610e-01 -8.36566865e-01 5.35227299e-01
-6.39892161e-01 -4.34389919e-01 -4.57900643e-01 -8.23375165e-01
-4.83438343e-01 -1.36987567e+00 5.34441650e-01 8.73996139e-01
-4.94817942e-01 -5.86289763e-01 -2.10319474e-01 -1.96166188e-02
7.34278500e-01 2.47844577e-01 1.05425441e+00 -1.10216521e-01
-1.00312221e+00 2.00630844e-01 3.39376450e-01 5.58608323e-02
3.59308091e-03 6.22343600e-01 -3.72707069e-01 9.73529741e-03
-4.66944247e-01 -3.46119016e-01 1.93427190e-01 6.87136054e-01
1.10720396e+00 -4.92726564e-01 -7.11243808e-01 5.09951234e-01
1.41722465e+00 1.12777567e+00 2.91320354e-01 3.44946295e-01
6.59500360e-01 6.40274525e-01 5.13839543e-01 3.88508230e-01
8.26916322e-02 4.01931256e-01 6.08325064e-01 -4.60739024e-02
1.01904415e-01 -9.00024995e-02 3.52309138e-01 1.12740481e+00
3.83460701e-01 -4.83166337e-01 -6.42600060e-01 7.25452185e-01
-1.81775260e+00 -4.48810518e-01 -2.00054318e-01 1.38026357e+00
5.33717215e-01 7.47520328e-02 -8.48659128e-03 6.20391369e-02
2.05090538e-01 -4.07580465e-01 -1.09192717e+00 -1.13201404e+00
9.21573043e-02 4.53660578e-01 9.34058130e-01 2.43397713e-01
-5.11174977e-01 7.55541384e-01 8.40352821e+00 7.25121439e-01
-6.94064379e-01 -3.77876520e-01 4.05257493e-01 -3.83826137e-01
-8.45244884e-01 -1.61630921e-02 -4.52840418e-01 2.40273774e-01
1.01083684e+00 -9.95677829e-01 8.09786439e-01 9.10618603e-01
1.39875382e-01 1.60796404e-01 -1.41103101e+00 4.94133502e-01
-6.70381546e-01 -1.67315686e+00 -2.56480217e-01 2.50665724e-01
1.24033546e+00 -4.22493875e-01 4.74741787e-01 3.00112277e-01
1.09255862e+00 -1.43836820e+00 7.94983506e-01 3.76810968e-01
8.95098969e-03 -1.05598938e+00 2.12894142e-01 -1.02571726e-01
-1.12626004e+00 -6.24659479e-01 -1.70738414e-01 -1.10144190e-01
1.37586877e-01 5.55157542e-01 -6.24059081e-01 3.89199704e-01
8.87718320e-01 7.42716312e-01 2.21998766e-01 1.02661991e+00
1.68222502e-01 4.18488801e-01 -4.18074846e-01 -6.79169416e-01
1.63008511e-01 -4.73541170e-01 8.93560708e-01 7.58479297e-01
3.35001200e-01 3.41749400e-01 4.82848227e-01 1.22279131e+00
-1.41911358e-01 -1.02273345e-01 -5.95807850e-01 -7.19867587e-01
7.04829693e-01 1.04435825e+00 -4.56676096e-01 1.71108302e-02
4.77677882e-02 2.75456518e-01 -6.19619846e-01 5.57797432e-01
-7.01545000e-01 -7.65175223e-01 1.06602252e+00 -2.70606697e-01
1.73195958e-01 -4.98790145e-01 -9.36384618e-01 -3.81700903e-01
-6.94484890e-01 -1.01186609e+00 -2.99934387e-01 -6.92062378e-01
-1.15431464e+00 -1.18583016e-01 1.31367490e-01 -7.90454149e-01
-5.57582557e-01 -7.93935895e-01 -6.16917551e-01 5.21825373e-01
-1.11813498e+00 -8.53809953e-01 -7.76876807e-02 -1.77571028e-01
9.61890459e-01 -2.86836922e-01 5.15595615e-01 1.62275717e-01
-8.97326827e-01 4.53676462e-01 3.75714749e-01 -6.41024470e-01
-8.43144134e-02 -1.16766059e+00 5.05035400e-01 7.16100395e-01
-5.10430515e-01 3.74803990e-01 1.32077467e+00 -5.83258390e-01
-2.77729535e+00 -6.84624434e-01 3.01267564e-01 2.78375503e-02
9.36767161e-01 8.84678289e-02 -2.46139646e-01 3.99872214e-01
8.39157164e-01 -6.66383445e-01 2.78106123e-01 -1.31185055e-01
7.59081125e-01 -4.76439714e-01 -1.18600786e+00 7.74902225e-01
1.16975868e+00 4.18413728e-01 1.58303827e-02 5.47442198e-01
1.35334396e+00 -3.06727201e-01 -1.39463747e+00 7.09564924e-01
8.54770660e-01 -1.36931181e-01 1.10321164e+00 -3.11976910e-01
5.20530105e-01 -6.22097552e-02 -1.87886268e-01 -1.52337897e+00
-6.86013401e-01 -1.17757130e+00 -6.41414642e-01 9.28916812e-01
4.20902759e-01 -4.05098975e-01 9.44733381e-01 7.48409450e-01
-9.01394486e-01 -1.06427276e+00 -3.54025930e-01 -1.07020390e+00
7.42472187e-02 -7.26075768e-02 1.02772546e+00 6.91697240e-01
2.68102646e-01 1.01400897e-01 -2.86132190e-02 -2.18692258e-01
7.86408246e-01 1.22718833e-01 5.05168736e-01 -9.11177695e-01
-3.18473041e-01 -1.11320519e+00 -5.65886870e-02 -7.71292090e-01
-3.54853421e-02 -4.80128735e-01 4.03833628e-01 -1.46007955e+00
-4.12601531e-01 -6.63486481e-01 -1.33909896e-01 3.66477966e-02
2.79661566e-01 -1.06321044e-01 4.27738279e-01 -1.04222938e-01
-2.19723403e-01 7.43143797e-01 1.46775687e+00 -4.25777078e-01
-4.67436850e-01 -2.00366452e-02 -9.00752842e-01 6.92346990e-01
7.28291869e-01 -1.06544770e-01 -4.58459646e-01 -4.62210000e-01
2.28597015e-01 -7.79629722e-02 -3.12928647e-01 -1.04131937e+00
2.19490707e-01 -1.21221721e+00 3.27355146e-01 -5.15008450e-01
4.18250024e-01 -1.28801930e+00 5.28652906e-01 1.08074224e+00
-2.68151432e-01 5.14103830e-01 3.44613045e-01 -2.45571788e-02
3.20079416e-01 -2.01251283e-02 7.75789678e-01 -9.54619125e-02
-5.41686356e-01 7.37484619e-02 -6.89997435e-01 -3.14112723e-01
1.28752315e+00 -2.70779997e-01 -7.54001975e-01 -1.15419149e-01
-4.47788537e-01 1.18543506e+00 4.09254134e-01 3.67760986e-01
7.97756195e-01 -1.33167827e+00 -4.95692998e-01 1.90204531e-02
-5.69852948e-01 -2.41459295e-01 -3.08818966e-01 7.25279987e-01
-8.40855062e-01 6.60331547e-01 -4.09955025e-01 -5.65452456e-01
-6.20747745e-01 5.64628780e-01 2.81529993e-01 -9.44249034e-02
-1.99406981e-01 9.88637924e-01 -1.14233367e-01 -8.99585560e-02
1.46521345e-01 -3.81037891e-01 3.32926512e-01 -7.03933537e-01
2.45445967e-03 9.65706468e-01 -3.61484170e-01 3.02315414e-01
-4.91614103e-01 5.90016246e-01 3.34342092e-01 -4.07086164e-02
1.44337535e+00 5.13061807e-02 6.81471601e-02 7.13628948e-01
7.15172946e-01 -1.26305208e-01 -8.11155021e-01 5.10473371e-01
2.31649220e-01 -1.28010437e-01 1.19006731e-01 -9.63675439e-01
-1.12432039e+00 3.40134561e-01 4.73799914e-01 3.93452257e-01
1.20277524e+00 -4.12241638e-01 1.28490365e+00 5.00383079e-01
1.45230472e-01 -1.36381233e+00 2.90668279e-01 4.64216888e-01
9.92246926e-01 -6.50196791e-01 5.40651083e-01 -4.52385247e-01
-4.27298993e-01 1.16938162e+00 1.02521014e+00 -5.21168590e-01
8.88628423e-01 5.85134506e-01 -3.18091393e-01 -1.88940123e-01
-1.08194602e+00 -1.95477512e-02 5.53807676e-01 6.19192302e-01
5.04944563e-01 3.92661020e-02 -5.37259758e-01 3.11645150e-01
-2.92982429e-01 1.43078208e-01 2.88978934e-01 1.28833807e+00
-6.35736883e-01 -1.11535549e+00 -3.21217924e-01 3.96133184e-01
-1.51028156e-01 9.03936625e-02 -5.45474291e-01 8.95732403e-01
1.42644688e-01 1.03096986e+00 1.49361178e-01 -6.23060226e-01
3.53943080e-01 -3.73129129e-01 7.33749688e-01 -2.08830237e-01
-1.09392786e+00 -7.01317042e-02 4.18693155e-01 -6.68538272e-01
-1.21890269e-01 -5.87499857e-01 -1.62029314e+00 -8.06944668e-01
-2.95090050e-01 5.46053112e-01 1.17703342e+00 7.90927112e-01
5.28613746e-01 1.05189562e+00 1.07916176e+00 -7.64792144e-01
-2.82312095e-01 -4.40950155e-01 -4.83168364e-01 -5.17519474e-01
9.63765681e-02 -4.51817900e-01 1.07897520e-01 -1.25998899e-01] | [5.754880428314209, 3.4158477783203125] |
fe40a396-b629-413a-a843-82fbcf1814af | deep-tiny-network-for-recognition-oriented | 2106.04852 | null | https://arxiv.org/abs/2106.04852v1 | https://arxiv.org/pdf/2106.04852v1.pdf | Deep Tiny Network for Recognition-Oriented Face Image Quality Assessment | Face recognition has made significant progress in recent years due to deep convolutional neural networks (CNN). In many face recognition (FR) scenarios, face images are acquired from a sequence with huge intra-variations. These intra-variations, which are mainly affected by the low-quality face images, cause instability of recognition performance. Previous works have focused on ad-hoc methods to select frames from a video or use face image quality assessment (FIQA) methods, which consider only a particular or combination of several distortions. In this work, we present an efficient non-reference image quality assessment for FR that directly links image quality assessment (IQA) and FR. More specifically, we propose a new measurement to evaluate image quality without any reference. Based on the proposed quality measurement, we propose a deep Tiny Face Quality network (tinyFQnet) to learn a quality prediction function from data. We evaluate the proposed method for different powerful FR models on two classical video-based (or template-based) benchmark: IJB-B and YTF. Extensive experiments show that, although the tinyFQnet is much smaller than the others, the proposed method outperforms state-of-the-art quality assessment methods in terms of effectiveness and efficiency. | ['Dongsheng Li', 'Zhaoning Zhang', 'Heng Yang', 'Min Liu', 'Baoyun Peng'] | 2021-06-09 | null | null | null | null | ['face-image-quality', 'face-image-quality-assessment'] | ['computer-vision', 'computer-vision'] | [ 9.72486734e-02 -6.87186182e-01 1.23159237e-01 -7.81348050e-01
-7.01320052e-01 -1.28168121e-01 3.65892500e-01 -6.31161809e-01
-1.71182677e-01 5.89279056e-01 -3.67431939e-02 1.75572172e-01
-4.55059618e-01 -7.75924206e-01 -5.85402071e-01 -6.73149526e-01
2.09513213e-02 1.58242285e-02 -7.12560266e-02 -2.05131382e-01
2.69005954e-01 6.86589062e-01 -1.84036648e+00 2.25398734e-01
8.51130724e-01 1.52803302e+00 -1.32517368e-01 4.86563146e-01
-7.09754750e-02 6.88981175e-01 -9.16970432e-01 -7.62568712e-01
3.58138859e-01 -5.60248911e-01 -6.27379298e-01 1.17496006e-01
8.37438464e-01 -7.14371383e-01 -5.24216533e-01 1.24483180e+00
8.58912706e-01 3.71224508e-02 4.64290887e-01 -1.50404620e+00
-7.23953366e-01 4.24929708e-02 -4.83232468e-01 4.49787527e-01
4.63705659e-01 4.11972523e-01 4.42105055e-01 -9.39587951e-01
3.08869451e-01 1.65512741e+00 7.59948015e-01 7.23266840e-01
-8.01027596e-01 -9.39977407e-01 -2.73931950e-01 7.63598442e-01
-1.42091143e+00 -8.45883667e-01 7.84464598e-01 -1.43087238e-01
6.02259874e-01 1.20383576e-01 3.73036742e-01 9.39604342e-01
1.84021294e-01 4.08833951e-01 1.10746658e+00 -1.20114870e-01
4.93634865e-02 -3.54610473e-01 -1.72836035e-01 7.96011269e-01
-1.29846856e-02 3.76898259e-01 -5.82387209e-01 1.01565972e-01
7.52349019e-01 -1.03541479e-01 -4.52454567e-01 1.01050198e-01
-8.15025449e-01 5.06721497e-01 3.99177045e-01 3.84520233e-01
-2.29926050e-01 -2.41738614e-02 3.22779715e-01 7.36104667e-01
2.52274454e-01 -1.36626318e-01 -1.35578766e-01 -2.32935861e-01
-1.17563725e+00 8.53999145e-03 5.71145892e-01 4.56143111e-01
6.97413206e-01 3.40616792e-01 -6.08055234e-01 1.07502735e+00
3.76614600e-01 8.07314217e-01 4.68288243e-01 -1.00496113e+00
3.83907825e-01 3.68195415e-01 6.33569434e-02 -1.39960814e+00
-2.71705598e-01 -3.03694397e-01 -1.00476861e+00 3.56552452e-01
4.18347150e-01 1.40591398e-01 -8.86566579e-01 1.66402149e+00
2.28114352e-01 4.09014285e-01 -4.88549769e-02 1.03537333e+00
1.18728459e+00 7.15159237e-01 -1.00621261e-01 -5.39662004e-01
1.08404541e+00 -8.09718370e-01 -9.16633666e-01 4.87279683e-01
9.77326930e-03 -1.00214410e+00 9.59363997e-01 7.00530410e-01
-1.13873219e+00 -1.11062002e+00 -1.17152417e+00 2.81677902e-01
2.62347646e-02 1.71328798e-01 2.08292052e-01 1.16138148e+00
-1.48171461e+00 8.56586695e-01 -4.25869793e-01 -8.17002580e-02
6.71180308e-01 4.61558640e-01 -7.03937471e-01 -5.53858519e-01
-1.21024644e+00 6.38894558e-01 3.70060578e-02 4.35416430e-01
-1.33541942e+00 -5.54948390e-01 -6.15281701e-01 1.16141222e-01
3.50578010e-01 -4.05985177e-01 9.59253728e-01 -1.42065346e+00
-1.83354366e+00 5.15981495e-01 -1.10488482e-01 1.78921353e-02
3.83119643e-01 -8.66861716e-02 -1.03145504e+00 4.75296170e-01
-3.22096437e-01 4.79134142e-01 1.23572636e+00 -1.13611865e+00
-2.82912731e-01 -5.63704252e-01 1.13554880e-01 4.99119386e-02
-5.03134608e-01 4.02378172e-01 -5.72695076e-01 -4.90660429e-01
-2.27370456e-01 -3.09496939e-01 4.54898298e-01 3.27581614e-01
-3.89975198e-02 -4.39895332e-01 9.58358467e-01 -7.56921291e-01
1.23197961e+00 -1.97511089e+00 2.40975879e-02 9.61686671e-02
3.23501080e-02 7.54102707e-01 -6.82467997e-01 8.11020583e-02
-2.04799464e-03 1.24037430e-01 -6.87160045e-02 -1.70891657e-01
-7.05765560e-02 1.05372466e-01 2.98855722e-01 5.58158636e-01
4.30197269e-01 6.80848002e-01 -6.78395808e-01 -6.53686285e-01
2.34436482e-01 8.83804977e-01 -4.43004370e-01 6.16812527e-01
1.77536875e-01 4.33004349e-01 -1.24408551e-01 9.06139553e-01
1.32710254e+00 1.06721342e-01 -6.25978112e-02 -6.77062988e-01
2.66782105e-01 -2.47044370e-01 -1.32031977e+00 1.56674993e+00
-4.47848767e-01 3.02932620e-01 3.82849351e-02 -9.29245472e-01
9.84226942e-01 4.16219741e-01 4.90413636e-01 -1.04226041e+00
2.60222465e-01 1.97380766e-01 7.86357000e-02 -7.37172544e-01
-6.28438070e-02 -1.52750447e-01 6.72972739e-01 2.52623737e-01
5.12316525e-01 2.14309871e-01 2.97782362e-01 -2.64303982e-01
1.22665966e+00 1.35459490e-02 1.28514990e-01 -6.05792999e-02
1.12062383e+00 -9.21293795e-01 7.45555282e-01 3.48092586e-01
-6.80854023e-01 6.26828790e-01 4.10394847e-01 -4.91523504e-01
-9.95389044e-01 -9.95709777e-01 -3.44467103e-01 6.34949744e-01
2.03180820e-01 -3.97793472e-01 -9.40833986e-01 -7.13261843e-01
-1.93073124e-01 -1.54877856e-01 -5.69846690e-01 -2.60020465e-01
-6.04325235e-01 -6.59015954e-01 7.04686463e-01 1.83134675e-01
1.08418834e+00 -1.30207694e+00 -1.61280066e-01 -1.17360568e-02
-1.38010129e-01 -1.02352428e+00 -5.98553360e-01 -6.81696892e-01
-6.09964132e-01 -1.16829717e+00 -9.61692154e-01 -5.44338405e-01
4.65166897e-01 2.68363029e-01 1.20520592e+00 6.08740628e-01
-1.88341528e-01 3.50036442e-01 -4.73469347e-01 2.32211336e-01
-2.72159666e-01 -4.88034397e-01 1.05462238e-01 5.28508306e-01
2.87843198e-01 -4.27169293e-01 -8.91761959e-01 6.72525883e-01
-1.10532951e+00 -5.00557899e-01 6.32896423e-01 9.62050021e-01
3.63857597e-01 4.15077746e-01 6.67237997e-01 -3.14975947e-01
5.33533812e-01 -2.67152071e-01 -5.11860192e-01 5.51083744e-01
-5.38528323e-01 -5.63194565e-02 6.79461837e-01 -1.70695320e-01
-1.21446586e+00 -4.81355846e-01 -5.75818062e-01 -7.30037570e-01
-1.65073916e-01 2.39389166e-01 -6.96218073e-01 -5.97111940e-01
4.01762515e-01 1.52652442e-01 1.24526098e-01 -3.78028065e-01
-9.00220796e-02 6.46875739e-01 5.13083994e-01 -4.99473512e-01
8.00825894e-01 2.07161814e-01 2.04117700e-01 -7.36943245e-01
-4.02640194e-01 -4.08749953e-02 -3.06232184e-01 -6.76467240e-01
5.49835622e-01 -7.29685068e-01 -1.14650810e+00 8.35120678e-01
-1.08799291e+00 -6.06397316e-02 2.89503008e-01 3.71002465e-01
-3.72960716e-01 7.02827930e-01 -6.77379191e-01 -7.85639167e-01
-5.06537557e-01 -1.60086000e+00 1.10799444e+00 3.96422178e-01
6.32502675e-01 -5.77618361e-01 -1.40099362e-01 4.13838774e-01
6.49541080e-01 2.12869048e-01 4.77624953e-01 1.30631998e-02
-4.68320787e-01 8.04949775e-02 -5.43162405e-01 8.18543315e-01
2.98081964e-01 1.88559264e-01 -1.02800977e+00 -5.57167888e-01
1.82985336e-01 -5.52896738e-01 7.11684465e-01 2.57481843e-01
1.51524627e+00 -3.40385407e-01 2.82475024e-01 8.99149776e-01
1.58293116e+00 3.69436234e-01 1.15964043e+00 6.13908805e-02
4.79431361e-01 4.64758486e-01 5.65982044e-01 4.27932918e-01
-1.62762105e-02 1.03999043e+00 5.17668009e-01 2.14956686e-01
-3.33627164e-01 1.11674391e-01 6.14788115e-01 7.67164767e-01
-2.68345535e-01 -2.78607160e-01 -4.98410523e-01 3.88685837e-02
-1.36850321e+00 -1.15299857e+00 2.83387929e-01 2.12180924e+00
7.28073895e-01 -1.20042667e-01 6.58502951e-02 4.97678906e-01
8.57440054e-01 4.98394743e-02 -5.13100982e-01 -2.61965007e-01
-2.48723179e-01 5.42198300e-01 1.83464622e-03 2.04125866e-01
-9.37141299e-01 5.19051909e-01 5.87888002e+00 1.26973999e+00
-1.29037571e+00 2.86155879e-01 9.24260139e-01 -1.47177149e-02
-9.28414986e-02 -6.25286162e-01 -6.62692130e-01 6.50052905e-01
1.02829981e+00 2.62594912e-02 6.97145343e-01 6.51632071e-01
1.68922424e-01 1.46662727e-01 -9.76970136e-01 1.59051836e+00
4.79603142e-01 -9.65234756e-01 3.17740500e-01 -9.77788195e-02
6.66976094e-01 -5.46581686e-01 4.47573304e-01 2.11644962e-01
-2.97827750e-01 -1.52963889e+00 4.28060621e-01 7.30413377e-01
1.18281734e+00 -1.05249953e+00 1.07067764e+00 -2.70906270e-01
-1.29596686e+00 -1.84707463e-01 -7.58275986e-01 3.27064812e-01
-1.16898842e-01 7.81708002e-01 1.01228748e-02 8.64663184e-01
9.20034230e-01 6.88601434e-01 -7.74062634e-01 1.03319609e+00
7.17073306e-02 4.74308819e-01 1.92403756e-02 3.25266361e-01
1.17232632e-02 -5.94384074e-02 2.88869172e-01 7.76524484e-01
7.60870039e-01 2.17157766e-01 -2.11038798e-01 6.85037911e-01
-4.94985849e-01 2.38556981e-01 -3.63551855e-01 1.60506442e-01
2.96752483e-01 1.42906082e+00 -4.07967091e-01 -1.93876669e-01
-4.45833027e-01 8.71182203e-01 5.62717207e-02 9.80203822e-02
-8.55406702e-01 -2.97779232e-01 6.80215597e-01 -1.70765042e-01
2.87701488e-01 1.65305167e-01 3.50369513e-01 -1.18198192e+00
3.42227191e-01 -1.36919475e+00 2.82601327e-01 -7.37203956e-01
-1.33819056e+00 1.11054862e+00 -3.27085435e-01 -1.28902054e+00
4.33991328e-02 -6.85051024e-01 -4.58078921e-01 7.25083113e-01
-1.81885874e+00 -9.87671912e-01 -7.08831072e-01 9.75424111e-01
5.30059576e-01 -4.58188266e-01 5.05719423e-01 8.53700757e-01
-6.34099841e-01 1.18201923e+00 -3.13478783e-02 1.73561424e-01
8.63391459e-01 -8.07565570e-01 -3.65501978e-02 8.39884162e-01
-6.79842606e-02 3.55551541e-01 4.00196642e-01 -4.00884449e-01
-1.63633418e+00 -9.74950016e-01 3.46978247e-01 -1.38620153e-01
-1.70513839e-02 1.59357876e-01 -9.15695548e-01 -3.96443754e-02
2.76514918e-01 5.01409233e-01 4.59407330e-01 -3.62172544e-01
-4.92903918e-01 -8.26327682e-01 -1.65775549e+00 6.45935088e-02
1.14936221e+00 -4.03938413e-01 -1.20059013e-01 -3.73036414e-02
4.47735399e-01 2.20747758e-02 -1.13529563e+00 8.51079643e-01
8.08898568e-01 -1.49691236e+00 1.01840699e+00 -1.11266278e-01
3.86899292e-01 -3.99509221e-01 -2.68175900e-01 -1.33504415e+00
-2.47770011e-01 -3.21548641e-01 -5.92979789e-02 1.42210102e+00
-2.16730773e-01 -3.05791408e-01 6.19908154e-01 1.34087980e-01
2.98831370e-02 -6.09369397e-01 -1.14163065e+00 -9.82241690e-01
-2.55786866e-01 -1.93832859e-01 1.17709208e+00 6.54829979e-01
-5.81868410e-01 -4.63832393e-02 -6.94850445e-01 2.65227426e-02
7.84083605e-01 -2.16000617e-01 5.96614361e-01 -1.18448484e+00
-1.77418187e-01 -4.93018806e-01 -9.01289880e-01 -5.91119230e-01
1.47068903e-01 -4.65351909e-01 2.93675270e-02 -1.15742326e+00
2.73324043e-01 -2.87398964e-01 -6.96797311e-01 1.70137033e-01
-2.27836877e-01 7.56934583e-01 2.17647746e-01 1.00418098e-01
-6.89472675e-01 8.41305137e-01 1.49838841e+00 -3.35029006e-01
2.97985405e-01 -2.60941744e-01 -2.11730197e-01 3.09047878e-01
6.13000512e-01 -1.37073025e-01 -3.46153021e-01 -5.05482733e-01
7.20375553e-02 3.50430965e-01 2.15482965e-01 -1.56682539e+00
-1.95646733e-02 -7.65612498e-02 8.00693214e-01 -3.70295167e-01
3.19882572e-01 -8.47275853e-01 3.07411075e-01 3.98925662e-01
-4.30330932e-02 1.71126127e-01 1.20533615e-01 2.54532337e-01
-5.92222571e-01 -2.10739568e-01 1.15471113e+00 4.44431305e-02
-7.36669540e-01 7.66063929e-01 1.95091829e-01 -2.64533460e-01
7.61684716e-01 -1.70138180e-01 -4.05440211e-01 -4.72290039e-01
-3.82873744e-01 -1.95663825e-01 2.65319526e-01 3.82847309e-01
1.09918094e+00 -1.80868602e+00 -8.80289674e-01 3.29246759e-01
5.52547239e-02 -5.54434597e-01 7.00280249e-01 7.26686776e-01
-6.03187978e-01 3.00459623e-01 -9.20262218e-01 -5.31432033e-01
-1.43047416e+00 6.40799165e-01 5.79457760e-01 -1.47969916e-01
-1.43631041e-01 7.16764748e-01 1.37128860e-01 -2.65616894e-01
2.64484704e-01 9.65540931e-02 -5.23160398e-01 -2.05465466e-01
1.11930203e+00 5.40230691e-01 3.88661981e-01 -1.02003145e+00
-3.11972082e-01 8.77991021e-01 2.71764696e-01 2.05963194e-01
1.21225452e+00 -7.72737861e-02 -2.53482431e-01 -8.38239864e-02
1.41594708e+00 -4.42348748e-01 -1.21137202e+00 -8.73395205e-02
-2.90420383e-01 -1.01826310e+00 1.52773529e-01 -8.67288351e-01
-1.77723217e+00 1.00594676e+00 1.37771189e+00 -3.98519747e-02
1.68088651e+00 -5.25618494e-01 8.59876215e-01 2.23741442e-01
4.96942014e-01 -8.68904054e-01 5.52041769e-01 1.13659196e-01
1.11998045e+00 -1.35580468e+00 -1.57514825e-01 -1.58155799e-01
-1.54755130e-01 1.38030005e+00 9.26171005e-01 1.12124279e-01
7.57057667e-01 -9.44560543e-02 -1.50105366e-02 1.58841107e-02
-5.90822279e-01 -8.06820095e-02 3.94007295e-01 7.07307756e-01
3.32001656e-01 -2.31531531e-01 -3.63973439e-01 2.83422530e-01
6.18831813e-02 4.96103138e-01 1.78607672e-01 4.51088727e-01
-3.74433845e-01 -1.24838865e+00 -6.92635238e-01 5.29128909e-01
-6.00047529e-01 1.20107695e-01 8.01160038e-02 3.81101131e-01
3.97150517e-01 1.36290038e+00 -9.80347767e-02 -6.87461555e-01
2.38687053e-01 -1.69163913e-01 8.60794425e-01 -1.10285962e-02
-3.92102301e-01 -1.72765970e-01 -2.20600620e-01 -9.46018517e-01
-7.92237759e-01 -2.45510980e-01 -5.62793195e-01 -8.52359533e-01
-2.04693213e-01 3.29282507e-02 5.77708006e-01 8.42785656e-01
1.09772302e-01 4.71163332e-01 1.03413260e+00 -7.73473620e-01
-4.49025422e-01 -1.09716916e+00 -6.46155417e-01 6.22197628e-01
2.85190195e-01 -1.00735986e+00 -2.37355262e-01 -7.45993033e-02] | [12.956868171691895, 0.5963693261146545] |
2f6841cc-7400-47e9-a723-6697427b48a0 | learning-with-labeling-induced-abstentions | null | null | http://proceedings.neurips.cc/paper/2021/hash/689041c2baed0f6d91050495d632d6e0-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/689041c2baed0f6d91050495d632d6e0-Paper.pdf | Learning with Labeling Induced Abstentions | Consider a setting where we wish to automate an expensive task with a machine learning algorithm using a limited labeling resource. In such settings, examples routed for labeling are often out of scope for the machine learning algorithm. For example, in a spam detection setting, human reviewers not only provide labeled data but are such high-quality detectors of spam that examples routed to them no longer require machine evaluation. As a consequence, the distribution of examples routed to the machine is intimately tied to the process generating labels. We introduce a formalization of this setting, and give an algorithm that simultaneously learns a model and decides when to request a label by leveraging ideas from both the abstention and active learning literatures. We prove an upper bound on the algorithm's label complexity and a matching lower bound for any algorithm in this setting. We conduct a thorough set of experiments including an ablation study to test different components of our algorithm. We demonstrate the effectiveness of an efficient version of our algorithm over margin sampling on a variety of datasets. | ['Afshin Rostamizadeh', 'Giulia Desalvo', 'Kareem Amin'] | 2021-12-01 | null | https://openreview.net/forum?id=-1OkHh56c2m | https://openreview.net/pdf?id=-1OkHh56c2m | neurips-2021-12 | ['spam-detection'] | ['natural-language-processing'] | [ 4.48616803e-01 5.19568384e-01 -5.48153937e-01 -5.80798209e-01
-1.18234110e+00 -9.96873200e-01 5.88138163e-01 3.90180081e-01
-6.18962348e-01 6.44915998e-01 -2.49704033e-01 -7.12580085e-01
-1.63226090e-02 -6.82956398e-01 -6.54397964e-01 -6.42827094e-01
2.10911244e-01 9.27442014e-01 2.00749233e-01 3.56001377e-01
3.05928856e-01 5.12640476e-01 -1.16659939e+00 4.09307107e-02
7.84290731e-01 5.71386456e-01 -3.66372854e-01 5.72182178e-01
-5.76157011e-02 6.55425727e-01 -5.44022024e-01 -6.13450706e-01
7.09620774e-01 -4.68122840e-01 -1.01602292e+00 6.76823020e-01
6.51477218e-01 -1.22920565e-01 -1.67351246e-01 1.21573067e+00
9.97339562e-02 1.89153641e-01 9.00071383e-01 -1.50933576e+00
-4.40281272e-01 8.37975204e-01 -6.85127497e-01 1.64004058e-01
-6.80925772e-02 2.30339900e-01 1.51370883e+00 -6.54205263e-01
5.27554572e-01 9.75276709e-01 3.15507084e-01 7.05694318e-01
-1.58592403e+00 -4.82363641e-01 5.13937891e-01 -5.99784814e-02
-8.74776900e-01 -3.99870068e-01 5.38231850e-01 -5.02284467e-01
2.69726455e-01 3.84914935e-01 3.91936809e-01 8.82654905e-01
-5.58590531e-01 1.03339326e+00 1.14329660e+00 -5.81163347e-01
6.35837793e-01 6.82620168e-01 7.15711474e-01 7.24189699e-01
6.30770683e-01 1.55246437e-01 -5.78688681e-01 -7.93077767e-01
2.75795400e-01 7.23139867e-02 -1.59906492e-01 -7.69423425e-01
-6.23840332e-01 1.24501240e+00 3.42999727e-01 -2.34158814e-01
-6.74612494e-03 1.14151590e-01 7.67825693e-02 4.81852740e-01
7.44807661e-01 7.20099986e-01 -3.46879810e-01 2.98474044e-01
-8.95593226e-01 1.86868042e-01 1.19234109e+00 9.33123946e-01
8.09266031e-01 -6.26173437e-01 -7.02847019e-02 5.29048026e-01
3.20276201e-01 3.39229077e-01 -1.83352694e-01 -9.57284033e-01
3.89226109e-01 6.44953012e-01 6.11403525e-01 -4.60884541e-01
-1.74993109e-02 -4.64926422e-01 -1.81161702e-01 3.55095237e-01
8.67908895e-01 -1.77268088e-01 -7.36739993e-01 1.75405478e+00
5.07458746e-01 -4.99136522e-02 -5.27737379e-01 1.00219166e+00
-1.62704095e-01 2.26280376e-01 1.67091399e-01 -4.87287015e-01
1.02621186e+00 -1.18322062e+00 -3.44263196e-01 -6.65665567e-01
9.81560290e-01 -5.69445550e-01 1.18302298e+00 4.75994289e-01
-8.81303072e-01 1.73453718e-01 -8.48363996e-01 -5.17456559e-03
-1.12901598e-01 -1.43475160e-01 8.05541754e-01 6.59985781e-01
-6.21217132e-01 5.98873734e-01 -6.59168661e-01 -3.67450655e-01
8.12487364e-01 1.92685932e-01 -1.46577656e-01 -1.73073083e-01
-7.96769202e-01 8.01605642e-01 3.69533151e-02 -1.56844556e-01
-9.56713140e-01 -5.71308970e-01 -4.45745677e-01 1.55349031e-01
8.20324838e-01 -5.22646666e-01 1.83341682e+00 -1.15699029e+00
-8.39422584e-01 1.17208433e+00 -2.78338939e-01 -6.00051463e-01
8.86440396e-01 -2.44357347e-01 7.62687027e-02 1.78128257e-01
2.35692397e-01 2.33973205e-01 9.51980710e-01 -1.36836922e+00
-1.07074034e+00 -3.81236523e-01 5.47902942e-01 9.40496475e-02
-4.21647221e-01 5.60786948e-02 9.88992397e-04 -2.92486638e-01
2.99510360e-02 -1.08673859e+00 -5.07148683e-01 4.50238287e-01
-5.60642779e-01 -5.37816882e-01 7.41471827e-01 1.02948919e-01
1.14628339e+00 -2.01330972e+00 -3.44247371e-01 4.47482884e-01
6.63166940e-01 2.26188332e-01 9.78418440e-02 2.88495898e-01
2.99307927e-02 4.01953489e-01 -5.22769094e-01 -3.64448547e-01
2.47474268e-01 6.81163929e-03 -6.89552665e-01 8.73250544e-01
5.76649867e-02 7.53550470e-01 -1.05414629e+00 -5.12179673e-01
-1.96362048e-01 -3.99900168e-01 -3.95060211e-01 3.13586742e-01
-4.21104401e-01 3.69035974e-02 -5.17237604e-01 4.18479651e-01
5.25893271e-01 -8.04469347e-01 1.06296614e-01 4.50459450e-01
3.26294214e-01 4.27825719e-01 -1.13262486e+00 1.13637578e+00
-3.95109177e-01 4.06483948e-01 4.11814392e-01 -1.09005105e+00
3.97668391e-01 1.49166644e-01 1.83320493e-01 5.39076179e-02
1.24528535e-01 3.63347173e-01 -1.29842073e-01 -2.93183357e-01
2.21156970e-01 -2.42899925e-01 -1.09148651e-01 1.34185803e+00
-3.35057110e-01 2.67650396e-01 2.26721823e-01 7.65699804e-01
1.42104542e+00 -3.33791226e-01 2.74297595e-01 -2.34167784e-01
2.33235419e-01 3.37615758e-01 4.14137989e-01 1.43733656e+00
-4.81763780e-01 1.40446916e-01 7.97167540e-01 -2.94630796e-01
-1.12657475e+00 -1.07419384e+00 -1.14042960e-01 1.17715096e+00
2.04423502e-01 -2.43696138e-01 -6.14111662e-01 -1.42380583e+00
1.53610408e-01 6.45813107e-01 -6.32557273e-01 -9.03572291e-02
-4.38584775e-01 -8.78795207e-01 -6.85729384e-02 1.29835308e-01
8.04301053e-02 -6.50697827e-01 -4.10679609e-01 5.97107876e-03
2.55495995e-01 -9.05584216e-01 -6.97160721e-01 3.53079468e-01
-9.13003981e-01 -1.44260323e+00 -1.53163463e-01 -7.36268699e-01
1.20285630e+00 4.05864984e-01 1.30077922e+00 3.56033981e-01
-2.04685718e-01 2.89720088e-01 -3.15041244e-01 -4.89519715e-01
-5.95807910e-01 5.02245545e-01 -2.04832703e-01 5.88235408e-02
6.88168108e-01 -3.97021353e-01 -5.65481126e-01 3.53413582e-01
-7.37953186e-01 -2.58365959e-01 6.54457092e-01 8.84218335e-01
2.89869845e-01 -1.00511082e-01 7.46677816e-01 -1.81013536e+00
6.58611894e-01 -6.57956958e-01 -1.05384171e+00 2.87900329e-01
-8.43125045e-01 8.57380033e-03 7.95934439e-01 -5.09689689e-01
-8.09924006e-01 5.84950566e-01 5.82296908e-01 -3.96426953e-02
-6.81276172e-02 -1.05346506e-02 -2.63696015e-01 9.36850235e-02
1.10712922e+00 -2.64295816e-01 -2.26768851e-02 -4.05645221e-01
5.52332819e-01 9.66661572e-01 1.96915522e-01 -6.84906304e-01
1.31898737e+00 6.27349734e-01 -1.63323060e-01 -2.57841319e-01
-1.53584874e+00 -6.21693730e-01 -3.55627060e-01 2.35345066e-02
1.59695208e-01 -5.16549230e-01 -7.25117445e-01 5.32764457e-02
-8.57480049e-01 -3.74761522e-01 -5.39951563e-01 2.77270257e-01
-5.11623740e-01 4.37148303e-01 -4.51989472e-01 -1.14896321e+00
-7.31049478e-02 -9.90495980e-01 7.01072276e-01 1.79337263e-01
-3.00492793e-01 -9.18189645e-01 -7.19055608e-02 4.29679394e-01
2.22257510e-01 -3.42959881e-01 1.05868697e+00 -1.37251687e+00
-1.04284847e+00 -6.31279588e-01 -1.19784072e-01 1.82621062e-01
-5.43241091e-02 -3.35841119e-01 -1.13246143e+00 -2.90576875e-01
2.78649889e-02 -5.77295125e-01 9.57591295e-01 2.68038418e-02
1.18904674e+00 -5.97480774e-01 -5.03916919e-01 9.06438529e-02
1.27573526e+00 -2.16000304e-01 7.21877068e-02 3.12327117e-01
3.23517978e-01 7.31766343e-01 7.68195212e-01 1.68862656e-01
-2.52957374e-01 4.02216941e-01 2.90959179e-01 -7.42080957e-02
4.15209830e-01 -2.70014375e-01 1.74040422e-02 1.21118538e-01
7.42360055e-01 -3.16236019e-01 -6.46372855e-01 6.05630577e-01
-1.96157026e+00 -6.99238420e-01 4.13491651e-02 2.58327961e+00
1.25494695e+00 4.90101755e-01 3.40755939e-01 1.63413420e-01
1.13093472e+00 -1.06491692e-01 -9.17551637e-01 -2.23610252e-01
2.34576002e-01 2.15510204e-01 7.16015935e-01 8.38127553e-01
-1.36145043e+00 8.51511061e-01 6.64937258e+00 8.84386122e-01
-4.64532346e-01 1.46173105e-01 8.72149110e-01 -3.31612229e-01
-3.96821469e-01 5.91737449e-01 -7.63401151e-01 2.90290177e-01
8.77407670e-01 -5.08973181e-01 3.93563211e-01 1.30680609e+00
2.26945169e-02 -3.20616096e-01 -1.64396584e+00 5.05602896e-01
-2.68029720e-01 -1.14826751e+00 -2.27625266e-01 3.14631462e-01
7.12724805e-01 -3.43750805e-01 -3.17957252e-02 6.20668903e-02
9.90960956e-01 -9.53554273e-01 5.70273817e-01 4.33625542e-02
4.81539279e-01 -5.98221004e-01 2.57077277e-01 7.21978188e-01
-4.92978781e-01 -3.87538254e-01 -2.60409176e-01 -1.93846505e-02
2.64389932e-01 9.94562507e-01 -1.07668090e+00 -5.89655004e-02
3.26799490e-02 2.35115930e-01 -5.57121515e-01 1.23335660e+00
-6.40428543e-01 1.00633192e+00 -4.00173128e-01 -1.60776466e-01
1.01257212e-01 -1.39811352e-01 4.22824651e-01 9.64351416e-01
-2.07051724e-01 8.38979781e-02 3.49695951e-01 1.06213832e+00
-5.35936475e-01 4.51276079e-02 -5.21434724e-01 -1.98059693e-01
1.00903153e+00 1.58324504e+00 -8.65126073e-01 -4.85696495e-01
-3.11339021e-01 6.66115046e-01 5.84039807e-01 2.76495904e-01
-6.93282247e-01 -2.99009979e-01 4.80044812e-01 3.26031834e-01
-2.00050045e-02 1.92889161e-02 -5.76717079e-01 -1.00431538e+00
1.19317517e-01 -7.96684146e-01 4.80402768e-01 -3.49209785e-01
-1.82876611e+00 1.21411555e-01 -2.44095907e-01 -1.06594670e+00
-1.70463383e-01 -5.99739730e-01 -5.14900923e-01 9.62173104e-01
-1.26130545e+00 -5.58252335e-01 -1.85890924e-02 1.46128193e-01
4.12469268e-01 7.49327838e-02 5.61911702e-01 5.79966046e-02
-5.74317873e-01 6.54498160e-01 -7.24193081e-02 1.58260524e-01
8.25398207e-01 -1.49485719e+00 4.56490815e-01 1.15282619e+00
3.95476550e-01 8.09349477e-01 7.55407751e-01 -4.76891816e-01
-1.15794396e+00 -1.12874401e+00 8.89687836e-01 -7.13177025e-01
8.78097296e-01 -6.10028982e-01 -7.42962956e-01 9.39029813e-01
-1.66748270e-01 1.49791166e-01 7.52779126e-01 3.96767586e-01
-5.72230577e-01 9.69855636e-02 -1.21921408e+00 5.12392581e-01
1.19182575e+00 -6.17205977e-01 -4.01584804e-01 7.42859125e-01
4.55223560e-01 -5.40410355e-02 -2.64476031e-01 -9.21906531e-02
1.90445244e-01 -6.90070212e-01 4.95079696e-01 -1.05955553e+00
3.52673471e-01 -1.71606615e-01 1.89236417e-01 -1.11077249e+00
-2.79201835e-01 -8.29839289e-01 -3.65168631e-01 9.97386336e-01
1.02196383e+00 -7.31565595e-01 1.14081466e+00 1.13598835e+00
2.22024694e-01 -7.35867262e-01 -6.72934115e-01 -7.16561973e-01
1.24157943e-01 -2.56360263e-01 1.46957770e-01 8.21097434e-01
9.00958702e-02 6.79571569e-01 -1.37463242e-01 1.83785290e-01
9.25662994e-01 4.04357016e-01 6.66703582e-01 -1.44856822e+00
-4.45447087e-01 -2.44421422e-01 7.69857094e-02 -1.32701051e+00
3.86133850e-01 -1.06646550e+00 1.33022204e-01 -1.29846835e+00
6.85943902e-01 -1.00193524e+00 1.28194699e-02 4.78679985e-01
-4.09131020e-01 3.29377018e-02 3.42190303e-02 4.85203087e-01
-8.19333732e-01 -3.83185819e-02 8.76669288e-01 -1.17874533e-01
2.80950144e-02 3.93739551e-01 -1.17297661e+00 7.78595984e-01
7.86568582e-01 -9.52597380e-01 -6.43714547e-01 -1.60618886e-01
5.37801504e-01 -2.32051075e-01 4.23860490e-01 -2.72925347e-01
4.46155578e-01 -3.11174601e-01 -6.98650815e-03 -1.11453727e-01
-9.96531621e-02 -8.18385661e-01 -3.50940913e-01 3.61182958e-01
-1.13191986e+00 -3.60698760e-01 -6.81838334e-01 9.16535914e-01
3.89230847e-01 -7.38745451e-01 1.14786220e+00 -2.37665579e-01
-1.78818598e-01 5.61486602e-01 -4.53083694e-01 5.62285006e-01
1.05898774e+00 -1.15042198e-02 -6.24645352e-01 -4.01618212e-01
-8.01243126e-01 3.15294474e-01 7.70557702e-01 -1.00185633e-01
5.39414063e-02 -9.64551330e-01 -5.37383795e-01 -5.06342575e-02
2.41808012e-01 1.78565606e-01 -3.20856512e-01 8.17789674e-01
-2.12111235e-01 1.91039711e-01 5.54609060e-01 -3.35854322e-01
-1.10902762e+00 7.85931349e-01 2.93745369e-01 -2.28149921e-01
-5.73201835e-01 8.89725447e-01 3.46592367e-01 -2.96208650e-01
4.96995032e-01 3.82540077e-01 3.51507843e-01 -4.79548052e-02
4.62482899e-01 3.26025307e-01 7.65984729e-02 4.81046271e-03
-1.24021403e-01 -2.94376791e-01 -5.91567099e-01 -3.53932738e-01
9.51801002e-01 -1.03296295e-01 -1.01795286e-01 3.00373942e-01
9.84223723e-01 1.80405647e-01 -1.15037000e+00 -6.97334707e-01
4.12245244e-01 -8.19219589e-01 -2.49920130e-01 -8.46265197e-01
-9.57794487e-01 6.06891692e-01 2.52208710e-01 6.16119504e-01
8.10511529e-01 3.62563640e-01 5.00452936e-01 7.37597466e-01
4.13539737e-01 -1.25347257e+00 9.25257206e-02 4.27507795e-02
9.51833203e-02 -1.26029563e+00 6.46024272e-02 -6.95684731e-01
-4.27051991e-01 7.89314985e-01 5.86092651e-01 -2.04628974e-01
4.53136146e-01 2.97449291e-01 1.45522729e-01 -1.70705616e-01
-1.11009359e+00 -3.10448315e-02 -2.82772958e-01 2.25176841e-01
2.00310335e-01 1.72329266e-02 -5.23159862e-01 4.27143842e-01
4.65082265e-02 -2.19546244e-01 8.41565669e-01 9.85560179e-01
-8.46577942e-01 -1.32367682e+00 -2.26232350e-01 7.32484818e-01
-4.19691920e-01 1.05192801e-02 -8.01776528e-01 4.28031802e-01
-1.65647268e-01 1.25412750e+00 -4.05924134e-02 1.14852756e-01
-5.36347665e-02 2.08149880e-01 4.27071065e-01 -1.00800824e+00
-3.92003000e-01 -2.24207252e-01 2.43947431e-01 -2.75584579e-01
-3.22547436e-01 -5.86618841e-01 -1.00638986e+00 -3.98742527e-01
-7.91912198e-01 6.33316159e-01 3.29585135e-01 9.28077161e-01
1.72173440e-01 -1.82456270e-01 1.23551118e+00 -1.11392096e-01
-1.32773733e+00 -7.45584905e-01 -9.02077138e-01 4.73985463e-01
1.70910388e-01 -4.47833210e-01 -9.62573767e-01 6.24592900e-02] | [9.13507080078125, 4.148754119873047] |
f604e248-ffc0-4f58-b3c7-b8e453fc5d91 | what-do-patients-say-about-their-disease | 2305.04905 | null | https://arxiv.org/abs/2305.04905v1 | https://arxiv.org/pdf/2305.04905v1.pdf | What Do Patients Say About Their Disease Symptoms? Deep Multilabel Text Classification With Human-in-the-Loop Curation for Automatic Labeling of Patient Self Reports of Problems | The USA Food and Drug Administration has accorded increasing importance to patient-reported problems in clinical and research settings. In this paper, we explore one of the largest online datasets comprising 170,141 open-ended self-reported responses (called "verbatims") from patients with Parkinson's (PwPs) to questions about what bothers them about their Parkinson's Disease and how it affects their daily functioning, also known as the Parkinson's Disease Patient Report of Problems. Classifying such verbatims into multiple clinically relevant symptom categories is an important problem and requires multiple steps - expert curation, a multi-label text classification (MLTC) approach and large amounts of labelled training data. Further, human annotation of such large datasets is tedious and expensive. We present a novel solution to this problem where we build a baseline dataset using 2,341 (of the 170,141) verbatims annotated by nine curators including clinical experts and PwPs. We develop a rules based linguistic-dictionary using NLP techniques and graph database-based expert phrase-query system to scale the annotation to the remaining cohort generating the machine annotated dataset, and finally build a Keras-Tensorflow based MLTC model for both datasets. The machine annotated model significantly outperforms the baseline model with a F1-score of 95% across 65 symptom categories on a held-out test set. | ['Ira Shoulson', 'Vikram Ramanarayanan', 'Abhishek Hosamath', 'Lakshmi Arbatti'] | 2023-05-08 | null | null | null | null | ['multi-label-text-classification', 'multi-label-text-classification'] | ['methodology', 'natural-language-processing'] | [ 9.83969569e-02 2.52837986e-01 -8.59507859e-01 -5.32509625e-01
-1.33773661e+00 -7.20989645e-01 -7.47128502e-02 6.53461158e-01
-6.32430494e-01 9.59337533e-01 8.04623187e-01 -1.90343797e-01
-2.95870215e-01 -3.62116814e-01 2.00004913e-02 1.16167432e-02
-1.74158111e-01 9.58216012e-01 2.53446311e-01 -1.59619823e-01
6.85047284e-02 5.95147945e-02 -5.58277071e-01 6.04042470e-01
1.13638020e+00 8.40986550e-01 2.66511559e-01 5.34823716e-01
6.45072311e-02 1.18070090e+00 -3.03963453e-01 -5.55236161e-01
-1.89745370e-02 -9.28521454e-02 -1.23685706e+00 1.31357357e-01
3.73920888e-01 -4.44130659e-01 -1.02825187e-01 1.03989649e+00
6.85187221e-01 -2.97905296e-01 4.06965524e-01 -1.03520143e+00
-7.19448388e-01 4.20678437e-01 -2.51412004e-01 3.92844230e-01
6.56495035e-01 4.18369740e-01 1.42770612e+00 -6.87352002e-01
1.04539227e+00 1.23465776e+00 9.18154240e-01 5.73196054e-01
-1.51145637e+00 -3.96036714e-01 -1.78211942e-01 4.55058604e-01
-1.20467484e+00 -1.61543369e-01 2.50440598e-01 -7.64642775e-01
1.13921106e+00 1.73961341e-01 7.45259523e-01 1.22345161e+00
2.04712391e-01 4.07791734e-01 1.19366324e+00 1.04778484e-01
4.53198671e-01 5.69876917e-02 4.66138065e-01 5.54786265e-01
1.50060564e-01 -3.24285954e-01 -1.78187147e-01 -8.83560419e-01
3.65284383e-01 -5.53547181e-02 -3.70487005e-01 8.24414492e-02
-1.41873515e+00 1.21828318e+00 4.64151919e-01 3.27314675e-01
-5.42995453e-01 -1.27513453e-01 7.21975386e-01 2.81130016e-01
4.10473675e-01 1.98899090e-01 -7.12432146e-01 -1.67416111e-01
-7.93727040e-01 4.01395649e-01 8.65919411e-01 7.56362021e-01
6.35102689e-02 -5.03447950e-01 -4.45626855e-01 1.26278865e+00
5.69352210e-01 4.83528733e-01 9.36750889e-01 -7.81094551e-01
9.06488836e-01 1.08842015e+00 -8.14780295e-02 -9.30999994e-01
-9.48906183e-01 -2.58847386e-01 -6.54176652e-01 -2.56819576e-01
2.16058865e-01 -1.86115131e-01 -6.96544766e-01 1.59443104e+00
3.73796582e-01 -6.09207392e-01 1.05178386e-01 9.73787129e-01
9.21576917e-01 1.17992386e-02 6.27026916e-01 -1.73490122e-01
1.86209297e+00 -7.11749732e-01 -8.59269083e-01 -6.23975813e-01
1.17761993e+00 -2.77635783e-01 1.07555330e+00 3.66055995e-01
-4.68112797e-01 5.85245192e-02 -4.54889596e-01 -2.13218749e-01
-2.78116703e-01 3.26309055e-01 5.35783768e-01 3.89375061e-01
-1.05790746e+00 5.01300514e-01 -5.67613780e-01 -8.42094421e-01
9.88352120e-01 5.09820163e-01 -6.45599067e-01 -1.12782709e-01
-1.19752693e+00 1.29845166e+00 5.56539178e-01 -3.99066716e-01
-6.14924371e-01 -6.49729311e-01 -5.08653939e-01 -5.53895116e-01
2.93403774e-01 -8.81884992e-01 1.21014667e+00 -3.90534967e-01
-7.62074709e-01 1.25217605e+00 8.71765148e-03 -6.13845348e-01
4.22931015e-01 1.21900722e-01 -7.08592415e-01 3.49698156e-01
6.92739666e-01 8.62466037e-01 4.69195604e-01 -4.31438297e-01
-6.81951463e-01 -8.74659061e-01 -2.17582881e-01 1.68095723e-01
-3.09425622e-01 2.79959261e-01 1.44112766e-01 -7.19413817e-01
-4.85913455e-02 -1.04795969e+00 -4.42276388e-01 4.05812830e-01
-5.09170830e-01 -6.48161948e-01 6.30583167e-01 -9.28608358e-01
1.31649923e+00 -1.93342614e+00 -5.61685376e-02 -1.38058558e-01
9.38145220e-01 1.52492583e-01 1.92160401e-02 4.82799649e-01
-3.87508810e-01 1.10114537e-01 -5.12671173e-01 -7.51429573e-02
-6.68299422e-02 4.74489808e-01 3.23143639e-02 4.20714676e-01
1.46633849e-01 1.06605351e+00 -1.14171684e+00 -6.81570232e-01
-1.48157903e-03 8.20400119e-02 -6.70402765e-01 -1.79697033e-02
-3.84048969e-01 7.01438010e-01 -6.25067174e-01 7.44363427e-01
9.23290774e-02 -6.25027120e-01 2.19590649e-01 -4.51876879e-01
2.26986453e-01 1.94342285e-01 -5.42080700e-01 1.62874162e+00
-2.11459473e-02 1.99835837e-01 -3.12050134e-01 -8.31385672e-01
6.34360671e-01 3.91976684e-01 7.88845599e-01 -6.42388701e-01
1.53549746e-01 5.61910927e-01 9.46757384e-03 -1.13286328e+00
-1.95168644e-01 -3.81785959e-01 -2.84001946e-01 2.59013325e-01
-1.73389435e-01 1.80416435e-01 2.57032156e-01 3.16347897e-01
1.75536370e+00 -4.46051627e-01 7.86732078e-01 -1.66197151e-01
5.03037095e-01 6.87597096e-01 6.56373560e-01 3.50626469e-01
-6.83986306e-01 3.59654963e-01 3.97410005e-01 -4.15434599e-01
-8.82418036e-01 -6.27936363e-01 -2.93974191e-01 7.52394199e-01
-7.08175361e-01 -6.81401372e-01 -4.80586559e-01 -8.44705999e-01
2.08378196e-01 8.58738601e-01 -5.23730397e-01 -1.84625164e-01
-1.16003715e-01 -8.37490499e-01 8.35165083e-01 5.69133341e-01
3.65191787e-01 -8.57544124e-01 -4.91227567e-01 5.49716830e-01
-5.61945617e-01 -1.42060506e+00 -7.27240562e-01 1.81841567e-01
-1.04412520e+00 -1.55578065e+00 -7.28675067e-01 -7.53286958e-01
2.21183270e-01 -4.18910503e-01 1.04727995e+00 -6.36779249e-01
-3.87233287e-01 4.44426775e-01 -4.93737251e-01 -1.59365505e-01
-4.61313635e-01 -3.02247792e-01 7.91322440e-02 -1.49287939e-01
8.11477184e-01 -3.91709566e-01 -7.45866954e-01 1.08103260e-01
-7.65618026e-01 -2.72777379e-01 6.87592804e-01 5.06898165e-01
7.41533279e-01 -3.23151290e-01 8.10244501e-01 -8.17592323e-01
1.16517198e+00 -8.06217730e-01 1.42861707e-02 2.21528467e-02
-7.52817631e-01 -5.77492900e-02 5.54752588e-01 -5.55963159e-01
-1.80519983e-01 4.25616443e-01 -2.33999640e-01 -2.95409143e-01
-2.36930877e-01 7.99739361e-01 1.42506242e-01 -1.51413679e-02
1.10452187e+00 -1.89698488e-02 1.45083480e-02 -5.62420905e-01
5.37237763e-01 1.22294116e+00 5.59816957e-01 2.81912778e-02
2.41943821e-01 3.29666525e-01 -2.84684420e-01 -5.23741901e-01
-1.08053029e+00 -8.53967369e-01 -3.79709363e-01 9.94676054e-02
1.22636127e+00 -1.01644957e+00 -7.23603427e-01 1.70573294e-01
-1.08854115e+00 -2.70820946e-01 -4.34630156e-01 5.60781181e-01
-5.83703160e-01 4.57104832e-01 -4.65901017e-01 -3.71140957e-01
-8.58650148e-01 -1.11172926e+00 1.19676292e+00 -5.16573310e-01
-1.09134841e+00 -1.06324816e+00 5.08384585e-01 8.11471939e-01
3.30180645e-01 3.21470976e-01 1.26688373e+00 -1.05452073e+00
3.23936254e-01 -2.36435100e-01 -5.27288139e-01 2.06449166e-01
4.74055916e-01 -1.03825569e+00 -3.68891567e-01 -1.49591610e-01
3.74390781e-01 -7.67869771e-01 4.95848507e-01 2.68675983e-01
6.94550574e-01 -5.91437340e-01 -3.50038528e-01 -1.45955563e-01
1.41109467e+00 1.75804198e-01 4.66617316e-01 3.18190068e-01
8.74426842e-01 3.92722934e-01 2.84234673e-01 6.33436084e-01
9.90122974e-01 7.80834436e-01 2.92779446e-01 2.74438083e-01
-2.09446594e-01 7.09970593e-02 5.22881329e-01 7.43145406e-01
3.33223969e-01 -1.20347470e-01 -1.19524360e+00 7.32980192e-01
-1.83056962e+00 -5.40123880e-01 -6.43097401e-01 1.81534767e+00
1.14252734e+00 -2.34458402e-01 5.34635723e-01 8.92556161e-02
7.20287681e-01 -2.85082012e-01 -9.87600505e-01 -1.03738226e-01
-9.70974844e-03 6.45486414e-02 7.44618237e-01 9.98508707e-02
-9.83775973e-01 8.55589330e-01 6.24209070e+00 6.12246394e-01
-7.11681306e-01 6.49860561e-01 3.56654197e-01 -4.47313562e-02
2.25858726e-02 -1.06246158e-01 -6.22830987e-01 6.60989821e-01
1.25633085e+00 -3.07413167e-03 3.92435133e-01 7.89418042e-01
8.97509336e-01 -1.66491568e-01 -1.22370207e+00 1.39339399e+00
4.73284237e-02 -9.13995385e-01 -3.63252945e-02 2.18841717e-01
2.65529782e-01 7.76405871e-01 -3.78313869e-01 6.65072026e-03
5.30883789e-01 -9.74316180e-01 3.92243236e-01 5.99879682e-01
1.09331203e+00 3.34952325e-02 4.71459538e-01 3.17416459e-01
-1.08127463e+00 -4.01785165e-01 -2.17851140e-02 4.36328113e-01
1.79625273e-01 7.24769592e-01 -1.12662721e+00 2.30187342e-01
6.53987825e-01 9.28963482e-01 -6.84580386e-01 1.04327679e+00
-8.73713419e-02 6.99681640e-01 -3.28834206e-01 2.82690972e-02
3.40432525e-01 9.38631222e-02 5.64477801e-01 1.03177035e+00
6.38791919e-02 3.92488152e-01 3.52925777e-01 9.94824648e-01
2.38606222e-02 3.80876958e-01 -5.13166189e-01 -5.98119795e-01
5.17784804e-02 1.21545911e+00 -4.40048695e-01 -4.99803066e-01
-5.72300255e-01 1.06436336e+00 3.94123405e-01 -1.18682683e-01
-5.88601112e-01 2.73031779e-02 1.63880736e-01 1.94924250e-01
-1.23192750e-01 1.64590597e-01 -2.94047445e-01 -1.16235065e+00
8.83276537e-02 -1.22485638e+00 8.59271705e-01 -7.47398257e-01
-1.91011131e+00 5.42250395e-01 -4.18497175e-01 -1.40677750e+00
-5.92380241e-02 -4.40747678e-01 -1.15911402e-01 6.76957548e-01
-1.10953176e+00 -1.07132101e+00 -2.64843464e-01 9.67490733e-01
5.33132970e-01 -1.07351333e-01 1.04463983e+00 7.11733341e-01
-3.83294523e-01 7.29001462e-02 -2.63333678e-01 1.24261854e-02
9.60774422e-01 -1.25521815e+00 6.23044968e-02 -2.15124518e-01
-3.10065746e-01 3.17835897e-01 3.30400676e-01 -1.24392855e+00
-1.23466277e+00 -1.67594099e+00 1.38260055e+00 -6.48416579e-01
1.17677498e+00 9.17651057e-02 -7.80934215e-01 7.43729055e-01
-1.75749511e-01 4.37741727e-03 9.80066597e-01 -4.70810980e-01
-1.23537593e-01 2.70119637e-01 -1.55188560e+00 2.48925343e-01
1.16322446e+00 -7.52853036e-01 -1.05615056e+00 1.31525409e+00
4.47202146e-01 -7.70558193e-02 -1.55327117e+00 2.06824228e-01
1.89706579e-01 -2.71686405e-01 6.91756964e-01 -1.00352347e+00
3.27561051e-01 -9.47086066e-02 -3.27397764e-01 -1.27246594e+00
-3.12555343e-01 -1.60044700e-01 2.05385640e-01 6.20012879e-01
5.21319389e-01 -8.36511910e-01 5.84223092e-01 8.78885925e-01
-2.61277825e-01 -8.33227158e-01 -9.57662761e-01 -5.55246413e-01
4.76526618e-02 -5.91798604e-01 -2.02227071e-01 1.25237191e+00
5.90594172e-01 7.49839783e-01 -2.51815692e-02 -7.42754936e-02
3.52020860e-01 -3.88145894e-01 1.47597671e-01 -1.55109620e+00
-8.34130868e-02 3.92856002e-02 -7.63758242e-01 -4.13002729e-01
-1.47950798e-01 -1.47070324e+00 -4.25369292e-01 -2.06540656e+00
5.70785642e-01 -3.09455484e-01 1.70533031e-01 1.14947820e+00
9.09864381e-02 1.61476493e-01 -2.16158807e-01 7.79112756e-01
-7.70297825e-01 3.64622027e-01 1.08425403e+00 -3.82300138e-01
-2.62620121e-01 -2.11332515e-01 -8.97319615e-01 6.89893544e-01
7.77898490e-01 -9.78991807e-01 -4.79237050e-01 -2.19945446e-01
3.82764488e-01 1.99603930e-01 6.78161800e-01 -8.41306806e-01
2.55933285e-01 5.21434024e-02 -1.06360838e-01 -5.54948330e-01
1.71989009e-01 -8.25073540e-01 1.95021108e-01 8.34117770e-01
-4.84615982e-01 4.96060867e-03 8.21040869e-02 8.01076412e-01
-1.10685369e-02 -2.22303942e-01 6.65646374e-01 -4.92432296e-01
-6.38919294e-01 3.92678946e-01 -4.88468289e-01 5.79438388e-01
1.09575152e+00 7.02678263e-02 -5.80223382e-01 -2.07095250e-01
-1.50699878e+00 3.98989201e-01 -2.69913096e-02 3.07812840e-01
5.34228325e-01 -1.38047421e+00 -8.36174250e-01 -5.12679279e-01
5.01903415e-01 -3.58796477e-01 7.65218362e-02 1.32598960e+00
-4.11824465e-01 7.16611803e-01 1.03560187e-01 -5.89036465e-01
-1.03363228e+00 3.61036360e-01 1.92926496e-01 -4.76920277e-01
-9.31359410e-01 3.29495281e-01 -3.28585744e-01 -4.57477629e-01
6.71454519e-02 -5.82670569e-01 -4.29245740e-01 3.39009523e-01
3.81927967e-01 2.78478771e-01 2.53445834e-01 -8.18412840e-01
-5.65927982e-01 4.59896892e-01 -1.82713971e-01 -3.24121639e-02
1.43617308e+00 -3.43588367e-02 -3.04387599e-01 3.94082218e-01
1.37730300e+00 -3.78316134e-01 -3.58347923e-01 -2.17980310e-01
4.42668796e-01 1.18252702e-01 3.13422620e-01 -1.17687070e+00
-9.78572726e-01 2.35229030e-01 1.14464819e+00 1.12591363e-01
6.86135650e-01 3.46994340e-01 1.09169960e+00 5.88627875e-01
3.04535329e-01 -1.19444990e+00 -4.28109765e-02 3.71151209e-01
1.05733013e+00 -1.43282723e+00 -2.64493197e-01 -5.14788628e-01
-1.05588925e+00 9.85014796e-01 2.66572814e-02 -6.13743365e-02
4.00329471e-01 -3.25767279e-01 8.69018510e-02 -7.98673272e-01
-6.99137211e-01 -1.73316061e-01 1.94301516e-01 6.04435146e-01
2.15441540e-01 2.25883558e-01 -8.66944313e-01 9.54908550e-01
-6.84884116e-02 5.96049309e-01 2.52003163e-01 8.84378970e-01
-2.74459630e-01 -1.22535861e+00 -1.67760193e-01 1.23632097e+00
-4.76109922e-01 -3.68367702e-01 -7.81313121e-01 2.91947216e-01
2.75356323e-01 1.09729922e+00 -3.94638717e-01 -4.65857923e-01
5.85461199e-01 2.04967394e-01 5.22756539e-02 -1.01288021e+00
-6.62380457e-01 -1.15483038e-01 6.68645918e-01 -7.55661309e-01
-6.59299910e-01 -7.50641108e-01 -1.46694100e+00 2.91127920e-01
-1.76502556e-01 -1.55176848e-01 5.04698098e-01 1.18771863e+00
6.00198865e-01 3.13230664e-01 1.61438227e-01 -8.36865394e-04
-7.50042319e-01 -1.03514624e+00 -4.53278869e-01 5.50089478e-01
7.77622536e-02 -5.85922360e-01 -1.47176415e-01 1.24304719e-01] | [8.543041229248047, 8.684863090515137] |
d69bdbf6-a1b7-46f5-92fa-05dff866c491 | sentence-incremental-neural-coreference | 2305.16947 | null | https://arxiv.org/abs/2305.16947v1 | https://arxiv.org/pdf/2305.16947v1.pdf | Sentence-Incremental Neural Coreference Resolution | We propose a sentence-incremental neural coreference resolution system which incrementally builds clusters after marking mention boundaries in a shift-reduce method. The system is aimed at bridging two recent approaches at coreference resolution: (1) state-of-the-art non-incremental models that incur quadratic complexity in document length with high computational cost, and (2) memory network-based models which operate incrementally but do not generalize beyond pronouns. For comparison, we simulate an incremental setting by constraining non-incremental systems to form partial coreference chains before observing new sentences. In this setting, our system outperforms comparable state-of-the-art methods by 2 F1 on OntoNotes and 7 F1 on the CODI-CRAC 2021 corpus. In a conventional coreference setup, our system achieves 76.3 F1 on OntoNotes and 45.8 F1 on CODI-CRAC 2021, which is comparable to state-of-the-art baselines. We also analyze variations of our system and show that the degree of incrementality in the encoder has a surprisingly large effect on the resulting performance. | ['Mark Steedman', 'Shay B. Cohen', 'Matt Grenander'] | 2023-05-26 | null | null | null | null | ['coreference-resolution'] | ['natural-language-processing'] | [ 3.05472106e-01 4.30934280e-01 -4.17916238e-01 -4.68706548e-01
-1.43765187e+00 -8.23934257e-01 7.10814476e-01 6.04258366e-02
-9.00355697e-01 7.52092421e-01 6.79906428e-01 -2.70338953e-01
-2.13582963e-01 -3.53922606e-01 -8.76377642e-01 -1.87717125e-01
-1.77560136e-01 1.25758255e+00 3.07651341e-01 -4.52026963e-01
2.77497321e-01 2.63762563e-01 -1.56398404e+00 7.51576483e-01
6.43804252e-01 2.56241053e-01 2.44054362e-01 6.75780118e-01
-3.45942974e-02 6.69389307e-01 -5.52264273e-01 -5.78036845e-01
9.50775966e-02 -2.94151336e-01 -1.35392046e+00 -7.79370487e-01
1.08895779e+00 5.09095304e-02 -6.89966440e-01 1.21267998e+00
6.20859861e-01 2.56695449e-01 2.42128611e-01 -7.36718774e-01
-4.83791143e-01 1.49969375e+00 -3.52784634e-01 5.76268256e-01
3.34251225e-01 -3.31967384e-01 1.41123986e+00 -6.79822028e-01
1.00594103e+00 1.64404714e+00 7.55145609e-01 9.03369308e-01
-1.17641211e+00 -8.15474987e-01 3.83821219e-01 6.85995996e-01
-1.15515184e+00 -8.56601059e-01 4.16970015e-01 1.21705584e-01
1.54678595e+00 3.48901033e-01 1.99883245e-02 1.03392065e+00
-1.42803863e-01 8.16510379e-01 5.68802357e-01 -6.78418458e-01
-8.12313631e-02 -5.15189171e-01 5.22264838e-01 2.42502481e-01
1.41135439e-01 4.02040362e-01 -8.53412867e-01 -1.55768581e-02
1.97040141e-01 -5.72748899e-01 -5.38926184e-01 3.53425220e-02
-9.76358235e-01 7.21753418e-01 2.99337417e-01 3.96459699e-01
-2.30043635e-01 1.79328788e-02 6.68473482e-01 4.19624299e-01
4.66136308e-03 7.62781143e-01 -5.55857480e-01 -3.47949177e-01
-1.03164589e+00 4.12855029e-01 1.01908612e+00 1.20141387e+00
2.23203763e-01 -3.85109514e-01 -1.05100967e-01 9.66621280e-01
-1.52059242e-01 2.52156645e-01 4.45356756e-01 -1.56908834e+00
8.73643100e-01 3.03817838e-01 -2.11096052e-02 -3.47347140e-01
-5.66347599e-01 -4.62914139e-01 -2.93300480e-01 -1.66865990e-01
5.05968630e-01 -1.79924905e-01 -5.87905109e-01 2.30094528e+00
-1.37526691e-01 8.95259157e-02 4.81849492e-01 7.62077689e-01
9.95339394e-01 3.64475340e-01 -4.64134291e-02 -5.36912799e-01
1.49537122e+00 -1.06711519e+00 -8.72664571e-01 -3.64357919e-01
6.31860733e-01 -7.29616463e-01 6.80568278e-01 3.72524917e-01
-1.57391858e+00 -4.65272725e-01 -1.09343195e+00 -1.89119935e-01
9.33490172e-02 -2.84351856e-01 4.46586996e-01 2.20757648e-01
-1.13736784e+00 8.58792543e-01 -7.70042002e-01 -3.90326619e-01
-1.83156095e-02 6.80111289e-01 -3.01340967e-01 -2.71824505e-02
-1.54477084e+00 1.16500700e+00 6.91274524e-01 -1.17910415e-01
-6.21648669e-01 -9.73973632e-01 -7.28541076e-01 3.57430100e-01
5.43613195e-01 -4.42844152e-01 1.84146953e+00 -5.94800949e-01
-1.26579142e+00 8.85558546e-01 -4.40250754e-01 -8.57734621e-01
1.83055446e-01 -8.11910927e-01 -5.86487353e-01 2.69103557e-01
6.19031228e-02 9.50779259e-01 -2.04895541e-01 -1.18214703e+00
-1.08176923e+00 -1.65264338e-01 2.41865799e-01 6.07850492e-01
-3.92804444e-02 3.57350528e-01 -6.26225770e-01 -2.37126186e-01
-6.50884435e-02 -1.20451415e+00 8.98545161e-02 -6.66034937e-01
-5.27652442e-01 -5.71295559e-01 5.25850713e-01 -3.57410491e-01
1.29399741e+00 -1.92915511e+00 2.21726522e-01 -3.07682425e-01
-2.34997302e-01 5.92854142e-01 -4.97722358e-01 5.20653248e-01
-3.79489720e-01 -9.08842683e-02 -3.14815938e-01 -5.27185977e-01
7.10500926e-02 1.46179408e-01 -3.82246614e-01 1.81165919e-01
-1.20805867e-01 7.06776977e-01 -9.93006945e-01 -5.39783180e-01
-1.79217815e-01 2.50608861e-01 -8.87098968e-01 3.61191332e-02
-1.86649293e-01 3.69275585e-02 2.49243230e-01 1.35151967e-01
5.21327257e-01 1.30742356e-01 8.78915787e-01 -2.30627373e-01
-4.39305723e-01 1.08631325e+00 -1.01193464e+00 2.23046303e+00
-3.16497564e-01 7.17349470e-01 1.99999452e-01 -9.17313218e-01
6.65375113e-01 6.08916998e-01 2.30964169e-01 -5.88658869e-01
-7.74726570e-02 2.01636776e-01 5.23774981e-01 -7.33238459e-02
9.41193163e-01 -1.91030994e-01 -2.91426748e-01 4.47603136e-01
2.57230729e-01 2.68355757e-01 4.95477647e-01 4.24359620e-01
1.30770063e+00 -7.69372433e-02 2.13362835e-02 -6.29532933e-01
5.32870710e-01 1.32998034e-01 9.53023434e-01 8.75606954e-01
-1.58746302e-01 6.76262021e-01 3.79482090e-01 -3.67548108e-01
-6.43642843e-01 -1.02323484e+00 -2.39479423e-01 1.25478506e+00
1.33219972e-01 -6.11865044e-01 -9.45838571e-01 -6.19329691e-01
-2.65060663e-01 1.12751698e+00 -4.05700564e-01 -1.10132702e-01
-1.43654907e+00 -4.46873814e-01 9.82370496e-01 8.15616965e-01
1.80094555e-01 -1.47231829e+00 -4.67441112e-01 3.79717976e-01
-7.52136111e-01 -1.16201949e+00 -8.93328905e-01 4.28667128e-01
-7.65006542e-01 -1.41113806e+00 -2.48892188e-01 -1.07864380e+00
8.28559324e-02 -2.81402264e-02 1.63667476e+00 1.41461957e-02
9.33231935e-02 5.37932571e-03 -1.84141815e-01 -1.96012989e-01
-3.81174654e-01 5.62939525e-01 2.75178015e-01 -7.46812940e-01
9.87616718e-01 -5.88144362e-01 -3.02692056e-01 -3.04863621e-02
-4.92388844e-01 -2.85714358e-01 5.63750267e-01 9.80132043e-01
3.79923522e-01 -3.70114267e-01 5.51218808e-01 -1.31574702e+00
5.75302362e-01 -1.15420066e-01 -4.09020364e-01 3.64560813e-01
-5.69680512e-01 2.67440885e-01 3.82729143e-01 -3.54657561e-01
-1.42912221e+00 1.62555240e-02 -1.28268018e-01 -4.11023825e-01
-2.12120324e-01 2.87985802e-01 -2.41883591e-01 5.30383170e-01
7.28199482e-01 -1.64472789e-01 -3.32271725e-01 -7.35646188e-01
6.63277030e-01 6.95802748e-01 1.60245717e+00 -9.11718071e-01
3.97093117e-01 1.24152355e-01 -2.47771859e-01 -4.65960920e-01
-1.13097739e+00 -5.61282337e-01 -9.46764648e-01 4.36092705e-01
7.23058581e-01 -8.55641544e-01 -1.09523356e+00 1.33234084e-01
-1.66835845e+00 -3.96167070e-01 -2.77650505e-01 5.98827660e-01
-5.80974340e-01 3.19815367e-01 -1.12907004e+00 -4.34229016e-01
-8.32946599e-01 -8.69182885e-01 7.80851305e-01 2.50185132e-01
-7.30139196e-01 -7.11250007e-01 4.97239739e-01 3.69387925e-01
7.04320073e-02 -2.31647819e-01 9.78212178e-01 -9.71906304e-01
-2.60511190e-01 3.39638323e-01 8.28813179e-04 -3.35326493e-02
-3.18918407e-01 -5.06631255e-01 -8.47141862e-01 -4.58163232e-01
-2.41050407e-01 -3.87597829e-01 9.86016154e-01 2.60825992e-01
5.92295051e-01 -2.92403698e-01 -6.00736737e-01 6.39737904e-01
1.22617555e+00 6.23199224e-01 7.47934163e-01 5.97807705e-01
3.89159888e-01 6.80136561e-01 5.34790695e-01 -1.69816539e-01
4.68130410e-01 9.55472946e-01 2.47576445e-01 3.31103086e-01
-4.48284477e-01 -2.34911069e-01 2.91966617e-01 1.06158233e+00
-2.01315910e-01 -2.29549691e-01 -8.96284759e-01 9.35891688e-01
-1.99583077e+00 -1.28901124e+00 -2.30609715e-01 2.01174784e+00
1.37153721e+00 3.49766850e-01 -1.03361636e-01 -3.25469896e-02
1.15468276e+00 8.86101797e-02 -3.16414386e-01 -5.95640600e-01
-1.86443880e-01 5.69404244e-01 3.01533848e-01 9.85676646e-01
-1.15345931e+00 1.37027812e+00 6.83334064e+00 3.46927822e-01
-7.17685819e-01 6.44504279e-02 -1.38628900e-01 -7.11252511e-01
-8.39067176e-02 8.50783437e-02 -1.29881966e+00 1.46458268e-01
1.32789159e+00 -1.60902768e-01 4.76053417e-01 4.83182073e-01
-4.83860552e-01 3.34758341e-01 -1.59926808e+00 7.20295191e-01
2.68884808e-01 -1.36477304e+00 -1.29514098e-01 -1.24679953e-01
5.80546021e-01 4.49539125e-01 -3.48074526e-01 6.16098523e-01
6.97090387e-01 -7.74357498e-01 5.97851157e-01 3.55513960e-01
8.59859467e-01 -1.03988695e+00 8.29941630e-01 1.99799567e-01
-1.13604414e+00 -6.19604811e-02 -2.79957354e-01 -2.15933304e-02
7.61429191e-01 -5.53494990e-02 -5.10044873e-01 5.66930592e-01
7.39325702e-01 3.51527900e-01 -1.77837655e-01 8.65601897e-01
-5.70608437e-01 5.43639183e-01 -2.58764595e-01 3.05474252e-01
1.99557126e-01 4.77231562e-01 8.87774587e-01 1.59600222e+00
-2.72844583e-02 4.24859107e-01 -1.11136623e-01 6.66982055e-01
-5.09608805e-01 -3.54944378e-01 -1.66811690e-01 4.84629095e-01
1.39612877e+00 8.92161012e-01 -3.11587192e-02 -3.84803027e-01
-3.48803967e-01 7.62292087e-01 6.81357503e-01 1.00351341e-01
-4.81282890e-01 -7.49429345e-01 6.79080904e-01 -2.73154378e-01
5.70414484e-01 2.18391880e-01 3.66983302e-02 -9.31498051e-01
-2.68833399e-01 -1.18597078e+00 8.40063632e-01 -3.96380246e-01
-1.15270543e+00 8.88657451e-01 1.79321691e-01 -6.77830338e-01
-6.59974337e-01 -4.58219647e-01 -7.46569276e-01 6.62915409e-01
-1.37389326e+00 -9.13622558e-01 1.79221615e-01 4.82310802e-01
7.12192416e-01 -3.16443980e-01 1.12290525e+00 4.94848192e-01
-3.85064125e-01 1.01082110e+00 -1.43957987e-01 3.36800724e-01
1.17485404e+00 -1.42538714e+00 8.43998253e-01 9.85461056e-01
4.31656092e-01 1.10209143e+00 7.48524547e-01 -5.54791152e-01
-1.02385306e+00 -8.94684076e-01 1.60639179e+00 -4.73857254e-01
3.67017329e-01 -2.45933816e-01 -1.13504553e+00 1.07085216e+00
6.98520720e-01 -4.43083018e-01 5.10031164e-01 9.93009269e-01
-5.61994076e-01 1.27775073e-02 -6.98278368e-01 4.90539640e-01
1.60501301e+00 -6.16721094e-01 -1.57712781e+00 2.80092787e-02
9.88986731e-01 -7.09186256e-01 -9.12585080e-01 6.37352765e-01
5.14606416e-01 -8.45063567e-01 9.69727218e-01 -9.19112980e-01
2.70632833e-01 4.47192267e-02 -3.13203514e-01 -1.18115568e+00
-7.44285882e-01 -8.88642430e-01 -3.09823126e-01 1.33607435e+00
5.95714986e-01 -2.12589085e-01 7.47500062e-01 4.85601515e-01
-7.94481337e-01 -2.96218574e-01 -1.09396160e+00 -7.70113170e-01
6.35581315e-01 -2.49002963e-01 5.71949065e-01 1.04993474e+00
5.81072271e-01 8.69741738e-01 -1.45214945e-01 1.05200887e-01
7.39189565e-01 2.63168216e-01 4.11105067e-01 -1.45506203e+00
-3.51137638e-01 -4.68128473e-01 2.15593874e-01 -1.23903394e+00
6.81067586e-01 -9.56740499e-01 3.03675741e-01 -1.08339310e+00
4.91791755e-01 -3.86443168e-01 -4.35964346e-01 6.27446413e-01
-1.46102622e-01 1.87651105e-02 4.29107398e-01 4.28814828e-01
-8.29814017e-01 1.77509293e-01 7.31716275e-01 -1.10791534e-01
-2.41467685e-01 -3.66502941e-01 -1.10731232e+00 7.62662947e-01
5.52198529e-01 -7.23316133e-01 -1.73058435e-01 -7.01493025e-01
-2.03088179e-01 2.10986033e-01 -2.13867947e-01 -8.34554970e-01
9.53614473e-01 1.76125333e-01 -2.99473494e-01 -1.00186718e+00
3.78645808e-01 -1.37135729e-01 -1.46662727e-01 5.83050430e-01
-7.25101590e-01 2.61264592e-01 4.59489167e-01 2.17938229e-01
-3.51318687e-01 -5.08118927e-01 5.75702131e-01 -2.30797306e-01
-7.67361462e-01 -1.96691185e-01 -2.05073997e-01 7.17786491e-01
3.92221570e-01 4.56699401e-01 -8.97629797e-01 -2.61878043e-01
-5.90958118e-01 3.11187029e-01 1.95954859e-01 7.68819094e-01
6.24942556e-02 -1.21469080e+00 -9.78123248e-01 -2.03041211e-01
-1.35555014e-01 4.96880412e-02 1.21037781e-01 5.19725680e-01
-1.55854747e-01 1.14665723e+00 -3.49461846e-02 -4.52763766e-01
-1.63028896e+00 4.68733549e-01 3.29960942e-01 -5.33452272e-01
-6.83463752e-01 9.44981933e-01 1.70157894e-01 -7.80745208e-01
5.85351825e-01 1.19163163e-01 -3.50978643e-01 9.84170958e-02
7.32765377e-01 3.30368459e-01 2.10878283e-01 -6.63272500e-01
-7.42852509e-01 3.46591264e-01 -6.79435313e-01 -4.02596414e-01
1.35454869e+00 1.22929871e-01 3.34574096e-02 1.85836196e-01
8.79506052e-01 -6.54726550e-02 -9.62019801e-01 -4.40672517e-01
4.43793058e-01 6.96982816e-02 -1.64693780e-02 -1.10292077e+00
-8.07860136e-01 7.44499266e-01 4.47038382e-01 -3.75766963e-01
7.72057295e-01 2.84191221e-01 9.73958969e-01 8.83225739e-01
3.13929945e-01 -1.17466688e+00 -2.84806401e-01 9.85901058e-01
8.82982612e-01 -8.08953702e-01 -7.29691088e-02 -3.11442286e-01
-4.87807572e-01 8.55177820e-01 7.46609509e-01 -5.50000481e-02
1.28591374e-01 7.03777254e-01 2.18167156e-01 -2.34718040e-01
-1.22693455e+00 -6.22338727e-02 2.08502561e-01 4.44708049e-01
7.79233992e-01 -1.50103774e-02 -5.07523417e-01 8.86077821e-01
-5.72754025e-01 -1.95481837e-01 4.36204344e-01 7.18640745e-01
-3.64092797e-01 -1.42518413e+00 -2.24895567e-01 -5.07807322e-02
-9.11358058e-01 -5.34206927e-01 -5.02194464e-01 1.16111720e+00
-1.88536458e-02 9.87194598e-01 5.65628707e-01 -2.08541065e-01
7.24285424e-01 2.34345406e-01 5.59805095e-01 -6.86120391e-01
-7.76406109e-01 -1.67865142e-01 8.12764227e-01 -5.87937772e-01
-4.56570059e-01 -1.15253401e+00 -1.43206882e+00 -2.51242340e-01
-5.29318869e-01 5.02053380e-01 2.09379703e-01 9.22840655e-01
3.96976709e-01 5.81773102e-01 7.11101517e-02 -7.04495132e-01
-7.38833249e-01 -1.39803016e+00 -1.48310751e-01 5.76626122e-01
1.20336628e-02 -5.89368403e-01 -3.00816208e-01 -9.04804394e-02] | [9.294904708862305, 9.545530319213867] |
e01d4186-687a-4326-a2d5-876a9b3fe532 | have-my-arguments-been-replied-to-argument | null | null | https://aclanthology.org/2022.acl-short.4 | https://aclanthology.org/2022.acl-short.4.pdf | Have my arguments been replied to? Argument Pair Extraction as Machine Reading Comprehension | Argument pair extraction (APE) aims to automatically mine argument pairs from two interrelated argumentative documents. Existing studies typically identify argument pairs indirectly by predicting sentence-level relations between two documents, neglecting the modeling of the holistic argument-level interactions. Towards this issue, we propose to address APE via a machine reading comprehension (MRC) framework with two phases. The first phase employs an argument mining (AM) query to identify all arguments in two documents. The second phase considers each identified argument as an APE query to extract its paired arguments from another document, allowing to better capture the argument-level interactions. Also, this framework enables these two phases to be jointly trained in a single MRC model, thereby maximizing the mutual benefits of them. Experimental results demonstrate that our approach achieves the best performance, outperforming the state-of-the-art method by 7.11% in F1 score. | ['Ruifeng Xu', 'Qinglin Zhu', 'Jingyi Sun', 'Jianzhu Bao'] | null | null | null | null | acl-2022-5 | ['argument-pair-extraction-ape', 'machine-reading-comprehension'] | ['natural-language-processing', 'natural-language-processing'] | [ 6.08971059e-01 8.07915092e-01 -2.85013735e-01 -2.53734648e-01
-1.10425389e+00 -5.73134720e-01 1.06518674e+00 9.52507496e-01
-3.43797922e-01 6.23632550e-01 4.73842084e-01 -8.66135836e-01
-4.73883748e-01 -8.80245984e-01 -7.52810955e-01 -2.99752831e-01
1.08619355e-01 7.55372107e-01 3.22651654e-01 -3.69874060e-01
6.01461530e-01 -2.11233497e-01 -1.69323456e+00 8.32554936e-01
1.39296854e+00 6.75255477e-01 1.70526773e-01 5.30513406e-01
-5.78674138e-01 1.16155028e+00 -5.40123999e-01 -8.10312033e-01
-2.34461173e-01 -6.01725340e-01 -1.38656664e+00 -4.67619509e-01
1.72086284e-01 -4.57856506e-02 5.95875442e-01 7.84237385e-01
1.95464119e-01 -1.79424226e-01 5.21150827e-01 -9.05191779e-01
-2.37295061e-01 1.15555787e+00 -5.41571021e-01 1.05559573e-01
6.12781107e-01 -4.56970960e-01 1.71265197e+00 -8.53442669e-01
4.33118790e-01 1.43294215e+00 2.05347836e-01 2.09125817e-01
-1.11706316e+00 -2.53844619e-01 3.60530049e-01 1.88969687e-01
-4.64541286e-01 -2.74158299e-01 1.11551738e+00 -1.95055097e-01
9.91923392e-01 4.83278126e-01 6.58714950e-01 1.00020075e+00
-2.80131787e-01 1.23165464e+00 1.21827912e+00 -9.13355649e-01
4.68903519e-02 1.56840961e-02 8.12127233e-01 2.86176413e-01
1.97290689e-01 -3.95631224e-01 -6.46248460e-01 -4.44659561e-01
-9.97222378e-04 -4.98962820e-01 -1.68819487e-01 6.75055459e-02
-1.06372273e+00 9.13960159e-01 1.85769096e-01 5.90451002e-01
-7.62512386e-01 -2.40637511e-01 3.03497195e-01 3.79381418e-01
3.28744829e-01 5.88694870e-01 -4.21080798e-01 -1.54947490e-01
-5.48856318e-01 6.29052103e-01 1.00725937e+00 4.58577007e-01
3.14739525e-01 -1.08266151e+00 -3.69764388e-01 9.12356675e-01
7.35729337e-01 2.01759651e-01 2.66408384e-01 -6.40839458e-01
1.03579473e+00 1.14564717e+00 1.01230755e-01 -9.09953356e-01
-2.01347515e-01 -4.75101590e-01 -1.64357379e-01 -2.90381372e-01
5.74667275e-01 -2.97999494e-02 -9.19881314e-02 1.68408644e+00
6.37720942e-01 -3.80172402e-01 3.84932131e-01 5.89134216e-01
9.65663850e-01 4.47806716e-01 2.82259166e-01 -3.43044281e-01
1.78968096e+00 -1.20498657e+00 -7.71542072e-01 -5.20913541e-01
9.71309781e-01 -1.05887544e+00 1.07336748e+00 2.95189351e-01
-1.39423645e+00 -2.06508160e-01 -9.70040560e-01 -2.65409708e-01
-1.08381249e-01 3.17341924e-01 4.02536243e-01 1.13469534e-01
-3.93849045e-01 3.48727584e-01 -4.59638268e-01 1.11741655e-01
3.49454522e-01 2.17404962e-01 4.27674130e-02 1.21732488e-01
-1.51315010e+00 8.64881754e-01 4.17772949e-01 -8.87622088e-02
-6.20028749e-02 -5.77591300e-01 -5.64380169e-01 3.08576643e-01
7.01929629e-01 -8.06171358e-01 1.30354023e+00 -8.83354247e-01
-1.24523258e+00 7.71728158e-01 -5.27705848e-01 -4.90793824e-01
5.44231057e-01 -9.15027678e-01 -9.03596357e-02 3.29721242e-01
3.43363434e-01 3.85815293e-01 5.64171910e-01 -1.38989079e+00
-6.26671612e-01 -5.16580582e-01 3.38196844e-01 6.20648563e-01
-3.60526085e-01 2.73368299e-01 -1.96658537e-01 -3.76497209e-01
2.00149015e-01 -5.51228881e-01 2.09311932e-01 -4.89033699e-01
-5.74442387e-01 -9.07165051e-01 7.93752491e-01 -6.43950224e-01
1.50154328e+00 -1.68762076e+00 2.72679448e-01 6.78065792e-02
2.92163610e-01 4.07658160e-01 -1.60977140e-01 7.69234300e-01
-9.42672864e-02 1.36035547e-01 -1.46449283e-01 -3.94823104e-01
3.92710418e-02 7.76735544e-02 -4.79148448e-01 -2.36307591e-01
3.52718264e-01 1.10243988e+00 -1.12740195e+00 -6.76868081e-01
-1.30057558e-01 1.63708329e-01 -3.67493302e-01 2.69533426e-01
-6.27827585e-01 3.76189440e-01 -7.69046009e-01 3.45442802e-01
6.66721821e-01 -3.74481022e-01 7.84345865e-01 -1.01281703e-01
-1.28971590e-02 1.13058972e+00 -7.17034996e-01 1.19588542e+00
-3.69847983e-01 3.69080096e-01 -2.11006224e-01 -1.21265054e+00
8.74178469e-01 3.92805755e-01 2.09800169e-01 -9.73387599e-01
7.14874864e-02 5.32272696e-01 4.18063223e-01 -6.59670413e-01
1.91248760e-01 9.71320942e-02 5.22692129e-02 9.46443915e-01
-4.36021954e-01 2.89983720e-01 4.25165117e-01 4.52079356e-01
9.84325528e-01 2.36983616e-02 4.44341838e-01 -1.52561232e-01
1.03383243e+00 -1.00126393e-01 1.29288793e-01 7.29451001e-01
2.43233815e-01 -6.97911577e-03 9.48821366e-01 -1.90329045e-01
-6.28380895e-01 -9.03553486e-01 1.08305089e-01 8.06354761e-01
2.70115107e-01 -5.77300847e-01 -8.89632344e-01 -1.21006465e+00
-2.73125648e-01 1.09385896e+00 -6.71099246e-01 2.26255432e-01
-1.07853949e+00 -6.08535230e-01 3.11802596e-01 4.31169450e-01
5.41314006e-01 -1.01591337e+00 -9.15683806e-01 3.49818975e-01
-9.81767952e-01 -9.65780735e-01 2.78670728e-01 2.42664173e-01
-8.28560531e-01 -1.56075060e+00 -2.29676217e-01 -6.16145134e-01
5.21770358e-01 2.00490445e-01 1.25058794e+00 5.40614605e-01
3.46825242e-01 2.90236790e-02 -7.31922865e-01 -6.47850633e-01
-6.54434383e-01 7.90725052e-02 -5.21214604e-01 -4.29191552e-02
6.30632281e-01 -5.79255998e-01 -5.42489469e-01 1.39414206e-01
-7.07865596e-01 4.10875976e-01 8.54882896e-01 1.04564106e+00
4.15719301e-01 -2.11565748e-01 7.36616850e-01 -9.53958035e-01
1.22486889e+00 -7.21070170e-01 -1.76870763e-01 8.07795167e-01
-7.40293622e-01 1.67416796e-01 5.02551734e-01 -3.86538029e-01
-1.52180886e+00 -5.71604371e-01 -2.00266823e-01 6.42560780e-01
-4.30895798e-02 8.10215890e-01 -1.58210531e-01 7.57267058e-01
4.54965115e-01 6.06279373e-02 2.51871850e-02 -3.91289800e-01
2.97717720e-01 7.41029501e-01 3.05108279e-01 -9.51070786e-01
5.87834179e-01 9.81129259e-02 -8.92104879e-02 -5.11866271e-01
-1.41937160e+00 -5.82824111e-01 -5.64312041e-01 -3.79028693e-02
7.09640741e-01 -5.03597021e-01 -7.52715707e-01 -3.09887491e-02
-1.55843699e+00 1.32531434e-01 -9.69442427e-02 4.21842426e-01
-4.06371981e-01 4.00673658e-01 -3.58877569e-01 -9.67752159e-01
-5.76095521e-01 -8.74545932e-01 1.20697272e+00 3.22233349e-01
-7.65378356e-01 -1.08105314e+00 2.25908220e-01 1.11323166e+00
3.46361008e-03 -3.43950503e-02 1.40554631e+00 -1.32763147e+00
-3.68814230e-01 1.70448758e-02 -2.28549495e-01 -4.35084291e-02
6.02937788e-02 -2.49759942e-01 -8.80035758e-01 2.41458729e-01
3.32763702e-01 -3.96551102e-01 6.79777801e-01 1.11570753e-01
7.30338633e-01 -4.69408214e-01 -4.32139546e-01 -4.31181073e-01
8.82902145e-01 1.07370406e-01 5.01385629e-01 6.03280842e-01
2.09549055e-01 1.28989255e+00 1.12895608e+00 9.31893662e-02
5.44687867e-01 6.00785434e-01 5.59109807e-01 -5.55417836e-02
6.62360489e-02 -4.69295532e-01 9.75626856e-02 6.75146520e-01
-7.07559586e-02 -5.84200919e-01 -1.01387060e+00 6.12600446e-01
-2.43098760e+00 -9.50024962e-01 -7.47474194e-01 1.82334983e+00
6.84086800e-01 3.90746683e-01 -1.26491427e-01 4.88215268e-01
4.41311687e-01 7.61180371e-02 -1.50277406e-01 -4.73504275e-01
-1.62738189e-01 1.68228835e-01 -4.28794324e-01 4.94817257e-01
-9.82644737e-01 7.51471698e-01 5.09144354e+00 6.44357979e-01
-3.46651793e-01 -2.18846619e-01 6.68840885e-01 3.47099513e-01
-7.88759589e-01 4.18526947e-01 -6.69059157e-01 1.87701970e-01
7.78457880e-01 -2.92930491e-02 -1.04545511e-01 5.28905571e-01
1.56077862e-01 -3.67272049e-01 -1.11173296e+00 1.62704840e-01
-9.85638127e-02 -1.15222859e+00 2.24851876e-01 2.12865680e-01
3.02435040e-01 -4.42170918e-01 -2.30279654e-01 1.06671512e-01
1.21244617e-01 -7.57361114e-01 5.72703123e-01 3.93378615e-01
-1.94355384e-01 -6.81354582e-01 9.71173406e-01 7.79467404e-01
-9.11597729e-01 -2.55123526e-01 2.71441251e-01 -3.68629962e-01
3.13256860e-01 5.65518439e-01 -8.15535069e-01 9.34148073e-01
4.75897640e-01 4.17060107e-01 -4.41636860e-01 5.06638944e-01
-7.67900467e-01 8.37444246e-01 -1.07552268e-01 -3.59223306e-01
2.67402649e-01 -2.70703673e-01 8.71222198e-01 7.62309313e-01
3.26590501e-02 2.40938544e-01 -1.35846436e-01 8.96276295e-01
-1.29842728e-01 5.07351160e-01 -3.30076039e-01 -2.01087013e-01
7.14030325e-01 1.23931563e+00 -6.74600422e-01 -4.41335410e-01
-4.16258007e-01 7.19114542e-01 6.86262369e-01 -1.91310301e-01
-6.84552193e-01 -1.68419555e-01 2.66669303e-01 7.60338604e-02
1.56247720e-01 2.32194379e-01 -3.12329233e-01 -8.31364632e-01
6.10561967e-01 -1.05840492e+00 4.84976202e-01 -5.70366800e-01
-1.26724231e+00 4.69465435e-01 2.19106451e-01 -1.00520849e+00
-4.83006269e-01 -3.24300230e-01 -9.47921574e-01 8.15903604e-01
-1.62530279e+00 -1.36574459e+00 -3.80447204e-03 -9.75426193e-03
6.31625414e-01 2.47106493e-01 7.80303955e-01 -9.24135968e-02
-3.17809373e-01 2.90939748e-01 -3.22329402e-01 4.58644629e-02
3.39109361e-01 -1.34688425e+00 3.19041073e-01 7.73085713e-01
2.21097603e-01 9.60495830e-01 5.93944490e-01 -6.60058856e-01
-1.00545526e+00 -4.15020883e-01 1.88312256e+00 -5.52460790e-01
6.11122966e-01 -1.15886051e-02 -1.08603108e+00 3.53530049e-01
5.34051359e-01 -6.53813064e-01 9.79671597e-01 4.37396169e-01
-4.11182940e-01 4.17869329e-01 -7.55419314e-01 8.41890931e-01
9.84093547e-01 -3.04839283e-01 -1.48324740e+00 2.07444072e-01
7.37882316e-01 -1.29624695e-01 -7.17613637e-01 4.87683296e-01
5.50612211e-01 -1.12337875e+00 1.07735515e+00 -5.98154306e-01
1.00310588e+00 -1.76976234e-01 -3.37125920e-02 -9.86867130e-01
2.07645893e-01 -4.41762805e-01 -4.99388129e-01 1.49258840e+00
6.79343820e-01 -6.69559717e-01 3.03976804e-01 5.53237140e-01
8.74108970e-02 -1.18594062e+00 -4.62231219e-01 -4.24608052e-01
1.39860973e-01 -2.19779924e-01 6.99648559e-01 7.48788774e-01
3.88362765e-01 1.04905868e+00 1.31064802e-01 -7.01396028e-03
5.04261196e-01 7.23501444e-01 7.86607444e-01 -1.61658573e+00
-5.31027496e-01 -6.32901847e-01 5.62694609e-01 -1.16330397e+00
3.70076537e-01 -8.19399416e-01 -2.26367444e-01 -1.73394525e+00
4.13665265e-01 -3.31292570e-01 -3.11348047e-02 3.00350517e-01
-7.70236194e-01 -4.20835644e-01 4.10974212e-02 3.03332239e-01
-5.49911559e-01 5.77225924e-01 1.26922512e+00 -8.48472491e-02
-2.87341326e-01 1.32061899e-01 -1.07138503e+00 1.01937962e+00
7.93070376e-01 -5.19840121e-01 -6.92872524e-01 -3.16867501e-01
3.64415020e-01 1.23413086e-01 2.44110361e-01 -2.78559536e-01
2.02917546e-01 -9.76963267e-02 -3.67943309e-02 -7.88882434e-01
9.14847180e-02 -4.77968872e-01 -3.71232003e-01 4.93060023e-01
-7.25665092e-01 1.52425662e-01 -1.26745403e-02 5.71257055e-01
-5.84618628e-01 -5.62334716e-01 1.41746640e-01 2.09865406e-01
1.61308479e-02 -5.96647322e-01 -1.21040389e-01 2.76779413e-01
9.02447104e-01 9.76384338e-03 -6.69733405e-01 -1.58023149e-01
-4.36950773e-01 4.09561366e-01 -1.52127609e-01 4.46245432e-01
6.11899614e-01 -8.91451836e-01 -8.61534595e-01 -1.29997090e-01
1.00095361e-01 2.27880850e-01 5.41473627e-02 1.06240642e+00
-4.52316925e-02 6.46463156e-01 2.62930900e-01 -5.09512544e-01
-1.89511812e+00 2.17485026e-01 -5.33239171e-02 -1.18904734e+00
-4.34973925e-01 7.44536579e-01 -5.39740641e-03 -4.87955421e-01
-4.72825915e-02 7.45545402e-02 -8.44588995e-01 2.61721224e-01
5.40911019e-01 2.68697232e-01 -2.05942933e-02 -4.43781227e-01
-2.40075126e-01 4.51979369e-01 -3.36837262e-01 -2.61673003e-01
1.34305346e+00 -1.05585769e-01 -4.32800114e-01 1.39888972e-01
9.36249435e-01 3.76616001e-01 -6.15567803e-01 -3.25918019e-01
6.24226987e-01 -1.22792013e-01 -2.90868491e-01 -1.05133414e+00
-9.02119130e-02 8.06470275e-01 -1.46211028e-01 4.55841303e-01
9.73425031e-01 3.51273805e-01 8.07711005e-01 3.90958190e-01
-1.56213745e-01 -9.51421857e-01 3.04014087e-01 5.55253744e-01
9.99997914e-01 -1.20249522e+00 -1.33918796e-03 -9.10181761e-01
-5.36132872e-01 1.24802554e+00 5.79890966e-01 3.85726243e-01
2.18082651e-01 3.08796987e-02 -3.94036807e-02 -6.78303897e-01
-1.06552160e+00 -1.14965670e-01 6.12839937e-01 3.50023322e-02
8.36442828e-01 -1.00650199e-01 -1.21156359e+00 7.49700010e-01
-2.34694138e-01 -4.36392993e-01 -3.33504491e-02 1.02706051e+00
-3.03722352e-01 -1.79449105e+00 -2.93843895e-01 3.68039906e-01
-4.41347271e-01 -3.11780602e-01 -9.86924052e-01 8.15401852e-01
-1.87066585e-01 1.31011569e+00 2.25201044e-02 -6.70783594e-02
1.76467523e-01 2.54378110e-01 4.27550763e-01 -3.36533546e-01
-9.70831931e-01 2.77912728e-02 6.98626041e-01 -2.77317166e-01
-7.31486738e-01 -8.19830418e-01 -1.47693527e+00 1.29398525e-01
-6.27906740e-01 6.31308854e-01 6.32757604e-01 1.35138762e+00
3.95948559e-01 5.89655757e-01 4.63597387e-01 -1.34246096e-01
-6.23462260e-01 -9.36409652e-01 3.10242087e-01 5.33539474e-01
-1.97561905e-02 -5.74865639e-01 -2.40072653e-01 -2.78510720e-01] | [9.798008918762207, 9.275716781616211] |
71470572-8ae1-4647-8a43-bdd5385f879d | self-supervised-learning-of-3d-human-pose | 1903.02330 | null | http://arxiv.org/abs/1903.02330v2 | http://arxiv.org/pdf/1903.02330v2.pdf | Self-Supervised Learning of 3D Human Pose using Multi-view Geometry | Training accurate 3D human pose estimators requires large amount of 3D
ground-truth data which is costly to collect. Various weakly or self supervised
pose estimation methods have been proposed due to lack of 3D data.
Nevertheless, these methods, in addition to 2D ground-truth poses, require
either additional supervision in various forms (e.g. unpaired 3D ground truth
data, a small subset of labels) or the camera parameters in multiview settings.
To address these problems, we present EpipolarPose, a self-supervised learning
method for 3D human pose estimation, which does not need any 3D ground-truth
data or camera extrinsics. During training, EpipolarPose estimates 2D poses
from multi-view images, and then, utilizes epipolar geometry to obtain a 3D
pose and camera geometry which are subsequently used to train a 3D pose
estimator. We demonstrate the effectiveness of our approach on standard
benchmark datasets i.e. Human3.6M and MPI-INF-3DHP where we set the new
state-of-the-art among weakly/self-supervised methods. Furthermore, we propose
a new performance measure Pose Structure Score (PSS) which is a scale
invariant, structure aware measure to evaluate the structural plausibility of a
pose with respect to its ground truth. Code and pretrained models are available
at https://github.com/mkocabas/EpipolarPose | ['Salih Karagoz', 'Muhammed Kocabas', 'Emre Akbas'] | 2019-03-06 | self-supervised-learning-of-3d-human-pose-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Kocabas_Self-Supervised_Learning_of_3D_Human_Pose_Using_Multi-View_Geometry_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Kocabas_Self-Supervised_Learning_of_3D_Human_Pose_Using_Multi-View_Geometry_CVPR_2019_paper.pdf | cvpr-2019-6 | ['weakly-supervised-3d-human-pose-estimation'] | ['computer-vision'] | [-1.72599658e-01 8.12691748e-02 -2.11542130e-01 -4.43880498e-01
-8.01600099e-01 -5.64429760e-01 4.16900933e-01 -9.73762497e-02
-3.76199633e-01 6.45015180e-01 1.85955271e-01 3.82402718e-01
1.94310278e-01 -3.98738265e-01 -1.00634742e+00 -4.84374583e-01
7.47587904e-02 9.51680541e-01 2.83124566e-01 -2.53496647e-01
4.18768078e-02 3.98401767e-01 -1.35037351e+00 -4.22425002e-01
4.42474484e-01 7.72761762e-01 9.00470465e-02 5.89406371e-01
5.28567076e-01 1.07004069e-01 -3.44361484e-01 -4.39091563e-01
4.49052125e-01 -2.62805283e-01 -5.95931649e-01 4.07585382e-01
7.71321595e-01 -5.09491026e-01 -2.71439344e-01 7.93778419e-01
7.89494097e-01 -4.03887080e-03 6.51828229e-01 -1.36869717e+00
5.26311919e-02 -1.08963311e-01 -7.91231334e-01 -9.00419056e-02
9.84565079e-01 1.49934888e-01 7.41839886e-01 -1.18617213e+00
8.69283378e-01 1.23104775e+00 9.00777400e-01 5.05649447e-01
-1.11753809e+00 -4.91903871e-01 -1.31770462e-01 -1.76195920e-01
-1.62637520e+00 -3.10786575e-01 1.03464413e+00 -4.80109930e-01
5.83720803e-01 1.08714439e-01 8.65959048e-01 1.55650818e+00
1.94945693e-01 7.51884282e-01 1.21604395e+00 -3.60212326e-01
-8.57911557e-02 8.59743729e-02 -8.91900882e-02 1.13452530e+00
2.51404732e-01 5.85012371e-03 -7.64280200e-01 -2.04043776e-01
1.06890547e+00 2.70218868e-02 -2.20821679e-01 -1.07674634e+00
-1.49189794e+00 3.88672292e-01 4.92922992e-01 -3.66161197e-01
-3.03406864e-01 1.16921254e-01 2.38638535e-01 -9.60397571e-02
4.57403749e-01 5.71339764e-02 -6.32590353e-01 -1.40581960e-02
-6.47062600e-01 3.60525757e-01 5.47615230e-01 1.28535700e+00
8.96941841e-01 -4.11956996e-01 2.73568660e-01 5.17623425e-01
5.44363797e-01 8.83426845e-01 1.39123797e-01 -9.65175986e-01
7.51560509e-01 6.66531861e-01 2.47876391e-01 -1.22570837e+00
-6.82791948e-01 -4.18239743e-01 -7.52615929e-01 -9.69663411e-02
4.14784640e-01 -3.12103815e-02 -6.32641554e-01 1.73820889e+00
8.24860990e-01 1.45667847e-02 -1.99348196e-01 1.29474485e+00
7.45290041e-01 2.84329057e-01 -3.84958327e-01 1.82977272e-03
1.25139689e+00 -9.60552692e-01 -3.12688231e-01 -3.79084319e-01
5.81232965e-01 -8.26134801e-01 1.02723849e+00 3.00808519e-01
-9.96050894e-01 -6.99674606e-01 -9.91662681e-01 -1.03615701e-01
-4.84691821e-02 4.66586620e-01 3.76378894e-01 5.78517437e-01
-5.82226455e-01 3.96348953e-01 -1.04467762e+00 -6.02675974e-01
8.30352828e-02 3.31606448e-01 -9.46003318e-01 -1.64396297e-02
-9.60562825e-01 9.76136088e-01 3.85106921e-01 2.03865349e-01
-9.67129290e-01 -2.67822146e-01 -1.06552708e+00 -6.54535413e-01
7.74466097e-01 -1.09386003e+00 9.55181599e-01 -3.20622981e-01
-1.33243239e+00 1.29104733e+00 4.31339666e-02 -1.57180846e-01
9.53702807e-01 -7.53523231e-01 1.29378811e-01 2.58501917e-01
2.96282977e-01 6.56202734e-01 7.41544425e-01 -1.51346517e+00
-1.83901086e-01 -8.26184988e-01 3.82459685e-02 6.75064266e-01
3.11120562e-02 -4.38247234e-01 -9.15606737e-01 -4.08574671e-01
6.70884311e-01 -1.39329648e+00 -1.03352278e-01 8.50454718e-02
-7.91602612e-01 -5.78395016e-02 6.15165651e-01 -7.05010951e-01
8.23833108e-01 -1.77455235e+00 5.55475652e-01 2.64543951e-01
1.46119848e-01 -1.46523133e-01 1.49507806e-01 4.00890678e-01
1.41502783e-01 -2.12726325e-01 -8.63698348e-02 -7.55934954e-01
6.83215559e-02 1.19825453e-01 3.06585878e-01 9.67329741e-01
2.49121077e-02 7.36453593e-01 -8.56739402e-01 -9.52944219e-01
4.82197493e-01 4.98846769e-01 -5.53246081e-01 4.33762550e-01
2.47253496e-02 7.94521213e-01 -4.27916408e-01 7.12793052e-01
5.75652719e-01 -3.24797302e-01 -2.93955356e-02 -4.96694714e-01
1.80933833e-01 -4.07538339e-02 -1.47228920e+00 2.28325915e+00
-1.56196535e-01 -8.01151395e-02 -1.44778684e-01 -7.02979088e-01
7.99827099e-01 3.57946366e-01 6.18866503e-01 1.42392619e-02
4.37075138e-01 2.13669822e-01 -5.76347172e-01 -4.35808599e-01
2.84217149e-01 4.55929711e-02 -3.41186255e-01 2.89139807e-01
2.78524488e-01 -2.26178840e-01 -1.33059844e-01 2.04370305e-01
7.99425185e-01 7.89478362e-01 5.70085704e-01 1.57511421e-02
5.80090940e-01 -8.85096714e-02 6.14417553e-01 2.64628023e-01
-1.74760684e-01 1.04110837e+00 2.06658617e-01 -3.76670539e-01
-1.31749606e+00 -1.42699683e+00 -3.74109671e-02 6.46890342e-01
4.57013667e-01 -6.28249824e-01 -7.91763484e-01 -8.16716611e-01
6.92982972e-03 -7.96190724e-02 -4.46922868e-01 1.50355871e-03
-6.30502820e-01 -3.54586393e-01 4.62455302e-01 4.89891827e-01
5.95529020e-01 -4.47515726e-01 -6.61194682e-01 -3.11339915e-01
-5.19502819e-01 -1.35434389e+00 -8.02059948e-01 -8.05539712e-02
-9.15030420e-01 -1.23639667e+00 -8.26244533e-01 -6.37277007e-01
1.03730929e+00 2.05237910e-01 1.04828334e+00 -7.47289658e-02
-1.57165080e-01 7.02612460e-01 -2.80043721e-01 -8.40341523e-02
1.02744967e-01 1.91125706e-01 6.60592437e-01 -1.25362426e-01
1.19230881e-01 -5.47087967e-01 -6.69993579e-01 7.58844256e-01
-2.21266404e-01 1.76126599e-01 5.94713449e-01 7.93448031e-01
1.06289530e+00 -2.41027147e-01 7.73537382e-02 -9.02901709e-01
5.56793362e-02 -1.39575288e-01 -4.46495980e-01 9.85316187e-02
-1.15353040e-01 -6.54559731e-02 2.60245293e-01 -2.86765754e-01
-9.86635208e-01 6.85355186e-01 -1.58228688e-02 -6.87335789e-01
-3.19464713e-01 3.49624991e-01 -4.83941734e-01 -8.19358751e-02
8.89864683e-01 1.05483467e-02 1.67972982e-01 -5.06885767e-01
2.27533102e-01 4.18387651e-01 7.22443640e-01 -7.25594163e-01
1.22145116e+00 5.66070855e-01 2.47812137e-01 -6.75328255e-01
-1.08674061e+00 -7.48771191e-01 -1.34322560e+00 -3.79542619e-01
9.05395091e-01 -1.33070493e+00 -7.00832069e-01 4.84634250e-01
-1.09316587e+00 8.61043334e-02 1.98562577e-01 7.84316957e-01
-9.67436135e-01 6.08162105e-01 -4.32357222e-01 -8.21889341e-01
-1.75362080e-01 -1.09483242e+00 1.58283377e+00 -1.62429094e-01
-4.51136172e-01 -7.66810954e-01 9.49157402e-02 8.45127702e-01
-3.70377839e-01 8.41594219e-01 1.96250647e-01 -1.96600705e-01
-5.51220834e-01 -5.46216071e-01 2.38356441e-01 2.42861599e-01
-6.23074034e-03 -4.30358201e-01 -7.05575705e-01 -5.22094905e-01
-6.20896295e-02 -6.98151410e-01 3.16107899e-01 2.47212946e-01
7.31844366e-01 -6.26316443e-02 -4.22045946e-01 6.95097744e-01
1.15406692e+00 -5.87848365e-01 2.13519529e-01 2.41527811e-01
1.09448850e+00 6.43629909e-01 1.06101239e+00 6.93519831e-01
6.63086474e-01 9.45973933e-01 4.57428396e-01 8.23931477e-04
5.59337921e-02 -8.17040980e-01 2.74049282e-01 1.07107496e+00
-5.34676909e-01 4.54224460e-02 -9.62394655e-01 9.85831916e-02
-1.79372203e+00 -5.74384987e-01 -1.15674026e-01 2.41108084e+00
7.58718193e-01 3.28492492e-01 3.69274259e-01 1.49878398e-01
6.82041645e-01 7.52956197e-02 -6.63959444e-01 5.74120045e-01
3.17392796e-02 -1.83792129e-01 6.24425530e-01 4.35817242e-01
-1.28427494e+00 9.42581356e-01 4.90166330e+00 4.79053915e-01
-6.40091181e-01 1.03493132e-01 2.15793729e-01 -9.60156918e-02
1.40695661e-01 3.63584720e-02 -9.69169021e-01 1.74164772e-01
3.28836262e-01 1.81725338e-01 1.63935512e-01 1.01741612e+00
1.51632249e-01 -2.03081265e-01 -1.30364668e+00 1.39787531e+00
4.40246165e-01 -7.29357243e-01 -1.89861000e-01 1.11600816e-01
6.62065804e-01 -1.95184961e-01 -2.52590358e-01 4.30616969e-03
-1.21983752e-01 -6.60137177e-01 8.20142567e-01 5.33569753e-01
7.77638137e-01 -7.16557562e-01 8.59616458e-01 8.03461969e-01
-1.28104794e+00 4.70091045e-01 -3.07229221e-01 2.92329267e-02
2.70716548e-01 4.41872030e-01 -6.81967437e-01 7.26720035e-01
8.29264104e-01 8.19111884e-01 -6.58033192e-01 7.97417045e-01
-4.94521558e-01 1.15660377e-01 -5.63048840e-01 1.69347554e-01
-1.91248104e-01 -3.02735507e-03 6.15993202e-01 6.79831564e-01
1.73098817e-01 4.88005020e-02 6.15629494e-01 4.57343400e-01
-3.48028354e-02 2.96797641e-02 -5.41078269e-01 4.57986593e-01
4.88325804e-01 1.26856101e+00 -7.46978462e-01 -4.01570722e-02
-1.57076582e-01 1.26022398e+00 1.83989838e-01 6.37628064e-02
-9.59305525e-01 7.26528838e-02 2.66646475e-01 3.82747352e-01
-1.53136790e-01 -5.73913693e-01 1.12322994e-01 -1.53903985e+00
3.84036362e-01 -8.99192929e-01 3.10416937e-01 -9.11500692e-01
-1.18924832e+00 3.69950712e-01 3.83055747e-01 -1.61123610e+00
-3.57592255e-01 -4.99322742e-01 2.54747160e-02 5.39113581e-01
-9.24945235e-01 -1.38308406e+00 -7.06840515e-01 6.79383099e-01
2.60485351e-01 -3.83556093e-04 5.77448905e-01 1.33051991e-01
-3.73302788e-01 6.36866987e-01 -4.09265578e-01 2.67877400e-01
1.15899074e+00 -1.29945838e+00 3.96580458e-01 4.49496299e-01
1.75835341e-01 6.71018898e-01 8.16792667e-01 -7.49508679e-01
-1.80757368e+00 -8.96479189e-01 6.73806190e-01 -1.15804529e+00
1.61986023e-01 -6.87022626e-01 -3.04457307e-01 8.99568439e-01
-3.81766528e-01 2.62388766e-01 5.20806313e-01 1.68744653e-01
-2.31867298e-01 -8.42670724e-02 -1.11711121e+00 3.97434801e-01
1.51651919e+00 -3.78800511e-01 -6.71750069e-01 4.18793082e-01
6.00090563e-01 -1.04554725e+00 -1.06488621e+00 5.42849541e-01
6.79478228e-01 -1.01928222e+00 1.27739489e+00 -2.41018638e-01
2.16541782e-01 -6.06027901e-01 -2.41387054e-01 -1.06462741e+00
-3.88347805e-02 -4.14973378e-01 -4.38535698e-02 9.17092800e-01
1.10196441e-01 -3.43218982e-01 1.32858574e+00 1.71622172e-01
-3.18154693e-02 -7.99950004e-01 -8.70459437e-01 -9.00525808e-01
-3.82047772e-01 -2.58287638e-01 2.49343708e-01 7.72384882e-01
-2.93678105e-01 5.18838227e-01 -7.36661255e-01 5.02493382e-01
1.12420988e+00 -2.62567066e-02 1.44793987e+00 -1.15715027e+00
-2.43543521e-01 3.07580292e-01 -7.88469970e-01 -1.23068011e+00
8.38093832e-02 -7.28929460e-01 1.07330628e-01 -1.25304079e+00
2.92670906e-01 -1.84109077e-01 2.44242132e-01 3.06390464e-01
-8.36507753e-02 4.17479724e-01 9.61574018e-02 4.56657648e-01
-8.40740263e-01 7.06891000e-01 1.26792228e+00 2.19730660e-01
-5.41923158e-02 4.26700078e-02 -1.39843091e-01 1.09269166e+00
5.61989069e-01 -4.28475231e-01 -5.04910886e-01 -2.66939580e-01
1.65167317e-01 2.55151629e-01 7.83097088e-01 -1.25829720e+00
2.44387567e-01 2.36321837e-02 8.07476878e-01 -1.04626501e+00
7.66065538e-01 -8.17202628e-01 3.60726178e-01 4.31670785e-01
-6.38834154e-03 2.32765988e-01 -2.88470298e-01 7.42595613e-01
-2.68126260e-02 -7.46449381e-02 6.56230688e-01 -5.01006365e-01
-5.29132187e-01 5.80300093e-01 3.72820735e-01 2.32323200e-01
1.06167769e+00 -3.64229023e-01 1.15148976e-01 -3.15245777e-01
-7.73381829e-01 2.53175229e-01 8.47759366e-01 3.84412408e-01
7.21413493e-01 -1.56738830e+00 -5.63302040e-01 4.70534936e-02
3.63233417e-01 4.59534645e-01 1.69552281e-01 8.93456697e-01
-8.01292896e-01 3.46403301e-01 -1.78511411e-01 -1.10330379e+00
-1.43654382e+00 4.02133852e-01 2.32253134e-01 2.92287786e-02
-4.04616773e-01 7.45025456e-01 1.61504634e-02 -9.75049496e-01
3.05250376e-01 -9.23912972e-02 2.17425585e-01 -1.30598411e-01
1.94706805e-02 3.65644276e-01 -7.59202391e-02 -1.07675910e+00
-6.75830543e-01 1.11811614e+00 3.03819537e-01 -1.12083003e-01
1.15579379e+00 -2.88860083e-01 1.82221159e-01 5.06188393e-01
1.34514284e+00 -6.28666952e-02 -1.41081607e+00 -3.60578775e-01
-2.73444325e-01 -6.70072377e-01 -3.29808205e-01 -4.44969296e-01
-8.02796483e-01 7.95577526e-01 5.62635839e-01 -5.23805916e-01
8.56761813e-01 2.82909781e-01 7.91178882e-01 5.86574912e-01
8.95038545e-01 -1.20563376e+00 5.05112708e-01 2.44735658e-01
1.01192737e+00 -1.60139203e+00 4.39911962e-01 -7.03355551e-01
-5.63743353e-01 8.36227477e-01 9.12230790e-01 -2.25710377e-01
6.11095250e-01 -3.02898977e-02 -1.35059999e-02 -2.00434476e-01
-3.12898576e-01 -1.69843942e-01 3.80832195e-01 5.67494333e-01
3.61052454e-01 3.76887284e-02 -2.00978309e-01 4.14805353e-01
-5.48318088e-01 -1.88994035e-01 1.02782644e-01 1.03806102e+00
-3.41432720e-01 -9.72303748e-01 -6.77340806e-01 5.23019731e-02
-1.41156465e-01 4.21022683e-01 -5.59051812e-01 8.77704680e-01
1.99874759e-01 5.74008286e-01 -4.07474458e-01 -6.39706194e-01
6.50639176e-01 -1.67213067e-01 8.45161796e-01 -7.40867317e-01
-1.94672361e-01 3.48071754e-01 9.26236585e-02 -6.54312372e-01
-6.25312448e-01 -8.08016479e-01 -1.16770077e+00 -2.76662976e-01
-4.61366057e-01 -1.10915318e-01 5.13974845e-01 8.98477018e-01
1.79901406e-01 -7.65939578e-02 5.05080044e-01 -1.27137625e+00
-5.67629576e-01 -8.88995230e-01 -4.44289893e-01 7.22213089e-01
1.24309845e-01 -1.12429762e+00 -3.61741066e-01 1.30811915e-01] | [7.011940956115723, -0.9501915574073792] |
0660adc1-5b66-4a80-8483-0ba7c57e1343 | temposum-evaluating-the-temporal | 2305.01951 | null | https://arxiv.org/abs/2305.01951v1 | https://arxiv.org/pdf/2305.01951v1.pdf | TempoSum: Evaluating the Temporal Generalization of Abstractive Summarization | Recent pre-trained language models (PLMs) achieve promising results in existing abstractive summarization datasets. However, existing summarization benchmarks overlap in time with the standard pre-training corpora and finetuning datasets. Hence, the strong performance of PLMs may rely on the parametric knowledge that is memorized during pre-training and fine-tuning. Moreover, the knowledge memorized by PLMs may quickly become outdated, which affects the generalization performance of PLMs on future data. In this work, we propose TempoSum, a novel benchmark that contains data samples from 2010 to 2022, to understand the temporal generalization ability of abstractive summarization models. Through extensive human evaluation, we show that parametric knowledge stored in summarization models significantly affects the faithfulness of the generated summaries on future data. Moreover, existing faithfulness enhancement methods cannot reliably improve the faithfulness of summarization models on future data. Finally, we discuss several recommendations to the research community on how to evaluate and improve the temporal generalization capability of text summarization models. | ['Lidia S. Chao', 'Shudong Liu', 'Yanming Sun', 'Zhaocong Li', 'Xuebo Liu', 'Derek F. Wong', 'Hou Pong Chan', 'Chi Seng Cheang'] | 2023-05-03 | null | null | null | null | ['abstractive-text-summarization', 'text-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.08631098e-01 1.89287990e-01 -5.52379012e-01 -2.40810633e-01
-7.97368824e-01 -5.02510607e-01 6.71960115e-01 5.66689789e-01
-3.93982530e-01 1.02043343e+00 8.88007760e-01 8.93433578e-03
8.16275179e-02 -7.46686578e-01 -7.42419302e-01 -2.34130248e-01
2.18488835e-03 2.96910614e-01 2.24567413e-01 -3.25571597e-01
6.99670851e-01 3.40496823e-02 -1.27711558e+00 7.12273836e-01
1.51212764e+00 4.85911995e-01 2.15627179e-01 8.45117927e-01
-6.82191998e-02 6.38578653e-01 -1.24039543e+00 -3.45107853e-01
-2.10609287e-01 -5.20991862e-01 -9.49623168e-01 -1.22198775e-01
8.03204358e-01 -4.47740585e-01 -4.73949403e-01 8.20744932e-01
5.99168539e-01 5.18055081e-01 8.55485260e-01 -7.49868870e-01
-9.72314894e-01 1.23636472e+00 -4.37114835e-01 5.99711955e-01
3.66736501e-01 1.31033733e-01 9.96870935e-01 -6.93878770e-01
5.23337901e-01 1.14703512e+00 7.07334816e-01 7.13457823e-01
-1.00154173e+00 -4.12663162e-01 3.10795665e-01 3.57407659e-01
-1.01461744e+00 -7.47746587e-01 6.78321540e-01 -1.27450913e-01
1.22399855e+00 4.68843341e-01 6.22512698e-01 1.20827103e+00
3.87785256e-01 1.14980865e+00 2.88102716e-01 -2.85629928e-01
1.67767555e-01 5.05993813e-02 5.46924174e-01 2.80078620e-01
6.67819679e-01 -2.13099495e-01 -9.21209216e-01 1.94244292e-02
1.55930296e-01 -4.88464981e-01 -3.37610602e-01 2.97022730e-01
-9.76503193e-01 6.34141624e-01 1.48018286e-01 2.76875019e-01
-4.78725702e-01 1.82814570e-03 8.12081337e-01 1.12877190e-01
7.85191834e-01 8.47700357e-01 -3.29739511e-01 -2.86322862e-01
-1.70327008e+00 2.38031477e-01 6.16497457e-01 1.00748754e+00
4.75265622e-01 3.44105572e-01 -8.78010511e-01 8.11164618e-01
-2.08122328e-01 4.46015745e-01 1.01289487e+00 -9.93366063e-01
8.04858088e-01 5.43839455e-01 -9.69902202e-02 -9.46179926e-01
-3.75740051e-01 -6.70748234e-01 -9.54688430e-01 -8.25040042e-01
-1.36839271e-01 -2.39818975e-01 -5.16842425e-01 1.56221950e+00
-2.42892891e-01 4.00219597e-02 3.66512150e-01 3.21776360e-01
1.09443724e+00 9.53816354e-01 -3.48256007e-02 -7.22132206e-01
8.10068011e-01 -1.25429654e+00 -9.57013786e-01 -5.20299792e-01
6.67179644e-01 -5.46732426e-01 1.25462449e+00 4.18542415e-01
-1.50556934e+00 -7.63846397e-01 -1.35557950e+00 -1.83958709e-01
7.48434663e-02 2.96667367e-01 4.32066411e-01 5.64684451e-01
-9.39077377e-01 1.05282879e+00 -9.60391402e-01 -5.27850926e-01
4.30542648e-01 -1.54409567e-02 7.68773556e-02 1.57680839e-01
-1.33199656e+00 8.26873899e-01 9.11761284e-01 -2.38907114e-01
-7.00775146e-01 -9.43190157e-01 -6.06552541e-01 2.80562907e-01
1.40414879e-01 -1.01720917e+00 1.42936528e+00 -5.11379659e-01
-1.53786647e+00 3.41193855e-01 -2.27409557e-01 -1.00036204e+00
5.06727695e-01 -7.26876378e-01 -3.61465961e-01 1.76330701e-01
2.06560388e-01 6.66493595e-01 6.58169985e-01 -1.15832996e+00
-6.18524611e-01 -1.35848716e-01 -2.66730666e-01 3.24832916e-01
-8.22682500e-01 -5.11177778e-01 -4.17793632e-01 -9.32434380e-01
-3.24265748e-01 -5.72503924e-01 6.68030456e-02 -9.86069739e-01
-8.07332218e-01 -3.82803500e-01 5.56506634e-01 -9.11474466e-01
2.22224450e+00 -1.97971356e+00 1.39083844e-02 -3.47755253e-01
-2.04637945e-01 5.32448292e-01 -3.83540571e-01 6.72727406e-01
3.08812380e-01 4.15342212e-01 -1.81655735e-01 -4.26922381e-01
-5.31695969e-02 3.34416404e-02 -9.24902618e-01 5.97181395e-02
8.38645697e-02 1.07550752e+00 -9.83122468e-01 -6.95356846e-01
-2.36098260e-01 -1.15127690e-01 -6.09205246e-01 9.37421247e-02
-5.04998863e-01 1.90375596e-01 -3.80924910e-01 1.32143706e-01
4.88529176e-01 -1.33286744e-01 -1.29869301e-02 -1.63853779e-01
-1.13514103e-01 5.83242118e-01 -5.22054136e-01 1.94846177e+00
-2.59083122e-01 7.64113009e-01 -7.84572899e-01 -9.46303904e-01
7.75018454e-01 1.22503228e-01 2.40725860e-01 -8.09062123e-01
-1.01976857e-01 8.59438777e-02 -2.68696368e-01 -4.22270387e-01
1.54503751e+00 5.02577573e-02 -8.85596871e-02 6.39973402e-01
1.86470244e-02 -3.51237118e-01 8.59350085e-01 5.71516991e-01
7.90851414e-01 -2.49727115e-01 2.94751495e-01 -3.14446837e-01
4.31212276e-01 1.02057472e-01 5.07215619e-01 1.04783523e+00
4.04092744e-02 5.86667180e-01 4.60066140e-01 9.61116627e-02
-9.99890327e-01 -1.04264247e+00 -1.31263956e-01 1.21022689e+00
-2.71388013e-02 -8.41683090e-01 -8.68537962e-01 -7.44158804e-01
-1.90632045e-01 1.61114740e+00 -5.00532985e-01 -8.28724384e-01
-6.91567063e-01 -7.82659233e-01 9.91585135e-01 6.90889895e-01
6.30661428e-01 -9.60338354e-01 -4.41847980e-01 3.50522280e-01
-6.44691527e-01 -9.14274216e-01 -7.70450532e-01 -3.48920405e-01
-1.45958829e+00 -6.40475452e-01 -7.03745842e-01 -5.00349700e-01
4.10078496e-01 3.41650814e-01 1.34736705e+00 6.33288994e-02
3.65099698e-01 4.25595313e-01 -4.65685040e-01 -6.79280460e-01
-7.95789659e-01 7.37681210e-01 2.99978197e-01 -6.30037427e-01
5.85948229e-02 -5.07025599e-01 -6.09139442e-01 -2.81747319e-02
-9.79410529e-01 1.88124344e-01 6.29412353e-01 6.83366537e-01
3.46798986e-01 1.23277500e-01 1.18500066e+00 -8.15741718e-01
1.25520563e+00 -3.73037070e-01 5.02307527e-02 5.53121269e-01
-8.55572343e-01 2.94735163e-01 6.73735678e-01 -5.42836785e-01
-1.37875164e+00 -8.36337984e-01 1.11986652e-01 7.87298828e-02
4.36644346e-01 9.17268157e-01 6.16324097e-02 7.17667162e-01
1.03881264e+00 5.33831000e-01 -2.29626119e-01 -4.69512165e-01
5.10970950e-01 4.90742952e-01 8.74025106e-01 -7.34756589e-01
3.88717115e-01 2.33368248e-01 -4.62927967e-01 -9.98814166e-01
-1.16512513e+00 -3.00265849e-01 -4.79725212e-01 1.62293459e-03
4.97058094e-01 -9.05013621e-01 -3.01266685e-02 3.51721078e-01
-1.23862243e+00 -4.20330316e-01 -6.33285165e-01 3.68327498e-01
-5.41634619e-01 9.69793677e-01 -6.49341166e-01 -4.90679026e-01
-1.06218016e+00 -4.24360991e-01 8.54034841e-01 3.75958025e-01
-7.11175323e-01 -1.11247742e+00 2.14951739e-01 4.48759109e-01
3.73383343e-01 -1.35182679e-01 1.00412810e+00 -8.05021226e-01
-1.84559934e-02 -2.27564275e-01 1.18245587e-01 7.42508531e-01
7.18827844e-02 3.02952200e-01 -6.32412016e-01 -4.82186824e-01
4.08273153e-02 -3.77825022e-01 1.38501096e+00 7.27799058e-01
1.09657860e+00 -8.46331358e-01 -2.23006979e-01 2.04814136e-01
7.15268254e-01 -1.71105236e-01 7.21535921e-01 3.21172953e-01
4.52287972e-01 5.06536484e-01 8.27285111e-01 7.22929358e-01
4.32461739e-01 2.73301631e-01 -1.24755479e-01 4.28615272e-01
-3.61064583e-01 -5.69850445e-01 8.12337756e-01 1.49146700e+00
-6.85394704e-02 -6.55511141e-01 -6.50035620e-01 5.97837806e-01
-1.92924881e+00 -1.17043817e+00 -3.01340837e-02 2.16081691e+00
1.33910728e+00 3.09264541e-01 -9.25056189e-02 8.37192908e-02
5.92020690e-01 3.54628980e-01 -7.14623928e-01 -5.13617575e-01
-4.70470309e-01 -7.10890219e-02 2.25303918e-01 4.04272884e-01
-7.62188137e-01 1.02085638e+00 6.61991119e+00 1.09530127e+00
-1.03744316e+00 -1.35416135e-01 4.48572129e-01 -4.77715075e-01
-5.10547459e-01 -1.46655008e-01 -7.65371561e-01 5.01291811e-01
1.19851005e+00 -1.09036231e+00 -4.39647026e-03 6.35151029e-01
3.48438621e-01 -2.17902586e-01 -1.39544189e+00 5.04753351e-01
3.76647294e-01 -1.58137035e+00 7.08795905e-01 -3.48774850e-01
1.35463226e+00 -9.20216814e-02 2.42786661e-01 7.79774785e-01
-5.74993864e-02 -8.15498292e-01 7.32660055e-01 7.27056146e-01
7.37698972e-01 -7.85952330e-01 7.63034880e-01 7.32767463e-01
-7.05552340e-01 6.25611842e-02 -6.86099768e-01 8.33095834e-02
2.76066631e-01 7.15934396e-01 -1.03150904e+00 8.86212587e-01
1.51689753e-01 8.73747051e-01 -1.07181191e+00 8.35885763e-01
-3.36701930e-01 9.45502996e-01 4.42537805e-03 -9.39749405e-02
1.41809853e-02 1.23896390e-01 7.73115396e-01 1.44570589e+00
4.82749641e-01 -3.68648581e-02 -9.50916484e-02 5.85786641e-01
-3.26532722e-01 1.51762828e-01 -2.22615466e-01 -3.82992953e-01
6.03780568e-01 6.56565249e-01 -2.43925810e-01 -6.72333300e-01
2.96557061e-02 8.31402123e-01 1.42053664e-01 4.79707807e-01
-7.30146706e-01 -1.65938169e-01 -3.19017544e-02 6.58861995e-02
6.23670444e-02 -1.34374455e-01 -6.75987840e-01 -1.40821826e+00
-4.60996255e-02 -8.94758523e-01 5.41664720e-01 -5.68688333e-01
-1.14735663e+00 2.98417151e-01 1.70511678e-01 -1.07382321e+00
-3.68793607e-01 1.85469076e-01 -9.40324306e-01 3.59697253e-01
-1.28933167e+00 -8.96669686e-01 -9.54756215e-02 7.59392381e-02
1.06760108e+00 -3.16577315e-01 5.86422980e-01 2.77384627e-03
-8.34965885e-01 9.27028060e-01 3.43740612e-01 -3.06202412e-01
1.01002979e+00 -1.09311163e+00 5.49462140e-01 1.06834626e+00
2.78371815e-02 6.85412169e-01 1.04532933e+00 -9.01034832e-01
-1.03877532e+00 -1.32261705e+00 9.38892543e-01 -5.61444700e-01
5.06197989e-01 2.66729861e-01 -1.24764752e+00 6.74292982e-01
4.96842414e-01 -8.26287627e-01 7.25311637e-01 2.72154421e-01
-6.92024752e-02 -4.02435772e-02 -6.90839589e-01 6.27545595e-01
8.30562890e-01 -2.63928890e-01 -1.17704487e+00 2.68001944e-01
9.23833787e-01 -3.05376768e-01 -6.95308566e-01 6.39642596e-01
3.59869033e-01 -7.98560143e-01 8.37795556e-01 -8.31946790e-01
8.76498699e-01 1.70549065e-01 3.28081138e-02 -1.55204332e+00
-2.69386560e-01 -4.80279237e-01 -7.06082463e-01 1.57215142e+00
4.66284096e-01 -2.43881211e-01 6.70941174e-01 3.24490368e-01
-5.49138665e-01 -5.12685180e-01 -6.76392794e-01 -1.01468301e+00
6.27687991e-01 -3.09277803e-01 4.85975146e-01 6.92074060e-01
3.03529918e-01 7.06367135e-01 -2.51834929e-01 -2.19493836e-01
3.84068131e-01 -3.52707282e-02 8.17768514e-01 -9.28453505e-01
-8.09782147e-02 -7.55924463e-01 1.83538571e-01 -1.41000378e+00
3.97369236e-01 -1.04992008e+00 -8.15091357e-02 -1.85237741e+00
4.83727962e-01 1.44987449e-01 -9.38363597e-02 1.79530248e-01
-6.80921853e-01 -3.54761183e-01 1.73943549e-01 3.58613938e-01
-1.00061047e+00 1.12261200e+00 1.38314271e+00 -4.25863892e-01
-5.96850812e-01 9.81037393e-02 -8.89846265e-01 5.86432457e-01
9.92458284e-01 -3.20217520e-01 -6.97631896e-01 -6.40165627e-01
2.76619285e-01 9.42597240e-02 -9.09645259e-02 -1.02379084e+00
4.47590917e-01 -1.74909696e-01 2.06377760e-01 -1.08469212e+00
-1.21926092e-01 1.39992774e-01 -2.70962626e-01 4.12775904e-01
-7.05722034e-01 4.36509997e-02 5.37330925e-01 6.09851360e-01
-2.60193855e-01 -5.16977131e-01 5.61549783e-01 6.40959740e-02
-4.63525087e-01 1.05828479e-01 -4.00857776e-01 5.55717468e-01
5.26229560e-01 -9.69749391e-02 -6.11973584e-01 -3.69504958e-01
-3.44447404e-01 4.71515357e-01 4.13973719e-01 5.24868727e-01
5.53468883e-01 -1.19105089e+00 -1.14101195e+00 -2.32395902e-01
4.22148071e-02 1.52384579e-01 7.29484022e-01 6.98000908e-01
-2.83731073e-01 6.30497813e-01 -8.84058475e-02 -3.81255895e-01
-1.07493472e+00 5.82396448e-01 -1.45547643e-01 -5.74108124e-01
-4.82071549e-01 6.84211969e-01 4.12045904e-02 1.85910519e-02
2.23830625e-01 -4.63935226e-01 -3.28236759e-01 4.36632782e-01
6.68314993e-01 9.85229433e-01 1.15005732e-01 -1.00178517e-01
-7.92831779e-02 8.83407220e-02 -4.76596981e-01 -1.17421344e-01
1.14618552e+00 -1.95473745e-01 -8.29044357e-02 6.43832266e-01
8.80425751e-01 2.02117994e-01 -1.10588646e+00 -2.67869264e-01
2.92746872e-01 1.45754874e-01 -1.61500379e-01 -7.18746245e-01
-6.34386063e-01 8.16836536e-01 -2.28632629e-01 1.94988057e-01
1.09019792e+00 -1.91764519e-01 1.04963028e+00 7.87723482e-01
-1.43219262e-01 -1.60580599e+00 5.96856892e-01 8.50759208e-01
1.32786834e+00 -9.98251796e-01 5.56192815e-01 1.99900363e-02
-9.06148255e-01 1.21029603e+00 5.86296201e-01 1.12591326e-01
7.17893615e-02 -1.78243294e-01 -2.98807710e-01 1.19951360e-01
-1.11026180e+00 4.41711694e-01 5.75090885e-01 3.54597360e-01
4.68438715e-01 3.10758129e-02 -6.53187096e-01 1.16364372e+00
-8.03263485e-01 -1.10739172e-01 8.59381020e-01 5.39082170e-01
-6.88265920e-01 -7.82565176e-01 -7.59105757e-02 6.25306129e-01
-3.17280143e-01 -2.78749853e-01 -4.83600467e-01 4.63473290e-01
-3.69794905e-01 1.21518660e+00 4.86207791e-02 -2.18266293e-01
4.47361887e-01 5.23626618e-02 5.10964632e-01 -9.05773640e-01
-4.84229386e-01 -2.38138914e-01 3.89521658e-01 -8.71041976e-03
-2.29635760e-01 -7.10946739e-01 -1.37342680e+00 -6.42693520e-01
-3.92093301e-01 4.22360420e-01 2.43008643e-01 8.38709652e-01
3.90782356e-01 8.52527916e-01 4.95082736e-01 -6.03532314e-01
-1.01050651e+00 -1.48326325e+00 -3.08280408e-01 2.58648872e-01
2.10678503e-01 -1.45204365e-01 -2.72376359e-01 2.24109903e-01] | [12.41802978515625, 9.408263206481934] |
1651ae81-80b4-414a-bfdf-c82008a0080a | near-lossless-deep-feature-compression-for | 1804.09963 | null | http://arxiv.org/abs/1804.09963v2 | http://arxiv.org/pdf/1804.09963v2.pdf | Near-Lossless Deep Feature Compression for Collaborative Intelligence | Collaborative intelligence is a new paradigm for efficient deployment of deep
neural networks across the mobile-cloud infrastructure. By dividing the network
between the mobile and the cloud, it is possible to distribute the
computational workload such that the overall energy and/or latency of the
system is minimized. However, this necessitates sending deep feature data from
the mobile to the cloud in order to perform inference. In this work, we examine
the differences between the deep feature data and natural image data, and
propose a simple and effective near-lossless deep feature compressor. The
proposed method achieves up to 5% bit rate reduction compared to HEVC-Intra and
even more against other popular image codecs. Finally, we suggest an approach
for reconstructing the input image from compressed deep features in the cloud,
that could serve to supplement the inference performed by the deep model. | ['Ivan V. Bajic', 'Hyomin Choi'] | 2018-04-26 | null | null | null | null | ['feature-compression'] | ['computer-vision'] | [-1.17072120e-01 -1.44608527e-01 1.19751222e-01 -2.42660165e-01
-1.38309732e-01 -2.82678932e-01 8.93051326e-02 -1.40465751e-01
-6.35995626e-01 3.81993771e-01 -8.76725018e-02 -3.22563738e-01
-1.33182615e-01 -1.04735184e+00 -6.47544503e-01 -8.76441121e-01
-4.60621677e-02 1.27308235e-01 1.58882678e-01 1.08962394e-01
3.02240044e-01 7.07902730e-01 -1.96055257e+00 2.59623677e-01
9.03308570e-01 1.50894845e+00 9.92156625e-01 8.39509130e-01
-1.52408078e-01 1.24859524e+00 -6.56599700e-01 -6.35879517e-01
2.46026292e-01 -1.25433147e-01 -7.45934248e-01 -2.04660937e-01
1.96763292e-01 -8.36551785e-01 -5.91526926e-01 1.23917615e+00
6.87070072e-01 1.66549429e-01 1.58443362e-01 -1.12551212e+00
-4.65692937e-01 2.69182324e-01 -2.27637544e-01 2.48993754e-01
-2.81768269e-03 -2.44175315e-01 4.31261033e-01 -8.64464641e-01
4.72212315e-01 6.83946848e-01 6.73030496e-01 4.34495777e-01
-3.71803492e-01 -6.50998533e-01 -3.56371999e-02 8.19255710e-01
-1.47076178e+00 -8.53236318e-01 7.79740036e-01 1.81047812e-01
1.12944376e+00 4.37275827e-01 1.05371654e+00 4.23382342e-01
3.42806667e-01 6.24548614e-01 5.02059817e-01 -3.82096142e-01
5.10430932e-01 5.23436703e-02 -3.41683418e-01 8.74265790e-01
3.04165423e-01 -2.92443931e-01 -5.78472853e-01 -8.35420862e-02
4.72211510e-01 3.92507404e-01 -2.42005199e-01 1.28143325e-01
-4.62693959e-01 5.26362598e-01 4.66955245e-01 2.00628161e-01
-8.42863202e-01 3.42426926e-01 4.99584138e-01 3.09774637e-01
7.03309655e-01 2.17360351e-02 -4.25908029e-01 -4.17796195e-01
-1.51420283e+00 -5.08655384e-02 1.10520124e+00 9.66458857e-01
7.58514285e-01 2.35020742e-01 3.46680164e-01 6.02772176e-01
2.74769694e-01 4.35914695e-01 7.30900824e-01 -1.46627092e+00
3.55202258e-01 3.00154209e-01 -3.47909153e-01 -1.25088441e+00
8.08282942e-02 -5.81987262e-01 -1.06240427e+00 6.86676949e-02
-2.86861509e-01 -2.75428802e-01 -4.79623318e-01 1.16005123e+00
2.75321245e-01 5.43088078e-01 1.54012203e-01 8.80842328e-01
6.80575371e-01 7.25629866e-01 -3.00755948e-01 -2.56587029e-01
9.90000010e-01 -1.29379880e+00 -8.01390886e-01 -3.98006886e-02
3.65377843e-01 -6.63389862e-01 5.71219146e-01 4.55382049e-01
-1.55248284e+00 -6.95263147e-01 -1.08364677e+00 -3.39767754e-01
-1.83306217e-01 1.42263755e-01 6.35402143e-01 5.89599729e-01
-1.61001122e+00 8.24689984e-01 -9.81595635e-01 -8.52602534e-03
4.98370081e-01 6.28714502e-01 -1.92762673e-01 -3.11002582e-01
-6.58216834e-01 7.09382117e-01 2.17480272e-01 2.49517053e-01
-8.08280110e-01 -4.67455953e-01 -5.63151300e-01 4.99851853e-01
-5.47147356e-02 -7.17979014e-01 1.11981106e+00 -1.21427178e+00
-1.66654539e+00 3.29609334e-01 -4.10073221e-01 -6.45246446e-01
3.46497834e-01 -2.18250006e-01 -1.02174453e-01 5.09844542e-01
-3.38219762e-01 6.19995296e-01 7.50679672e-01 -9.48487580e-01
-8.20164323e-01 -3.59341621e-01 9.81890261e-02 3.54663759e-01
-7.38246679e-01 -2.10937053e-01 -9.15339410e-01 -3.27370197e-01
9.64002237e-02 -8.66867781e-01 -2.27455005e-01 2.52302408e-01
-1.03851721e-01 2.72459704e-02 1.28563249e+00 -9.95972574e-01
9.39287603e-01 -2.27421737e+00 2.47816928e-02 1.14332043e-01
4.16078180e-01 4.81532812e-01 4.01477367e-02 6.61424175e-02
4.66193080e-01 -1.26971379e-01 1.06411554e-01 -7.70037770e-01
-4.64398742e-01 4.48951542e-01 -2.33240247e-01 4.95245576e-01
-1.91260040e-01 7.99344063e-01 -4.90905583e-01 -6.54610932e-01
-1.11946188e-01 6.72477603e-01 -1.04382455e+00 4.30560380e-01
1.99013025e-01 -5.80367371e-02 -2.49315336e-01 6.47132993e-01
9.76967573e-01 -1.35787219e-01 3.24852288e-01 -3.02850485e-01
-1.08929917e-01 2.36350521e-01 -7.35079527e-01 1.73579943e+00
-7.96366870e-01 9.58654404e-01 3.97012562e-01 -1.26450360e+00
8.18030596e-01 4.05509949e-01 6.52906537e-01 -7.43983388e-01
2.60804474e-01 2.13560447e-01 -2.02260450e-01 -7.05981791e-01
8.91527951e-01 1.93998680e-01 4.24590886e-01 3.40417802e-01
1.37555197e-01 1.87507905e-02 -9.14214402e-02 -6.16718791e-02
1.01329863e+00 -2.56018639e-02 -1.43952742e-01 -6.46831840e-02
3.75521928e-01 -2.58910149e-01 6.02519035e-01 4.45656955e-01
-2.41507411e-01 1.97049841e-01 -3.08741648e-02 -4.59314227e-01
-1.10651183e+00 -7.32397974e-01 2.19403207e-01 9.65355575e-01
2.13015974e-01 -3.58555496e-01 -1.02706075e+00 -2.43197158e-01
-2.66590923e-01 2.17430592e-01 -1.03255868e-01 -2.75730938e-01
-4.20004904e-01 -3.83328617e-01 5.64921260e-01 4.62766528e-01
1.20528007e+00 -7.71115363e-01 -1.05716336e+00 1.72580585e-01
-4.92487282e-01 -1.15273547e+00 -2.61681885e-01 3.02822351e-01
-1.20905793e+00 -7.43027091e-01 -5.45756280e-01 -8.72492373e-01
6.36658669e-01 5.31648815e-01 9.39331591e-01 4.30728585e-01
-5.10160439e-02 4.80204374e-02 -4.34804857e-01 -3.52675468e-01
-5.12356944e-02 -1.74039099e-02 -8.24334158e-04 -1.42037407e-01
1.75394803e-01 -7.46184647e-01 -7.98507512e-01 -3.05931807e-01
-7.06920385e-01 3.12004805e-01 4.04227853e-01 4.92194235e-01
8.23377550e-01 5.62102497e-01 2.27806941e-01 -5.62309325e-01
3.93815219e-01 -7.08758831e-01 -5.25642812e-01 1.06427811e-01
-5.53610861e-01 -1.18133098e-01 1.00035453e+00 -2.67375797e-01
-1.03669274e+00 1.18001960e-01 -4.66469616e-01 -7.14986861e-01
2.58456975e-01 3.96220624e-01 -2.40893215e-01 -3.62035275e-01
2.64980625e-02 6.07850611e-01 6.68745264e-02 -5.89789867e-01
1.15869679e-01 1.04296398e+00 6.55498564e-01 -5.40126525e-02
3.82334799e-01 5.60346305e-01 -1.00078605e-01 -8.49892855e-01
-3.60774010e-01 -1.09551542e-01 -2.89228499e-01 -2.57229924e-01
7.07165956e-01 -1.05106008e+00 -9.06881332e-01 5.47180533e-01
-1.27270687e+00 -2.58738250e-01 -3.36230338e-01 4.89421070e-01
-4.40717012e-01 2.97037363e-01 -7.92565048e-01 -7.84543395e-01
-7.93338358e-01 -1.26815701e+00 6.99687183e-01 2.44252771e-01
4.89725232e-01 -8.23180437e-01 -2.87004560e-01 3.67663562e-01
7.83237576e-01 -2.21801847e-01 6.23388171e-01 -2.93019682e-01
-7.35305071e-01 -1.84250310e-01 -3.37967515e-01 6.46948516e-01
-1.36144157e-03 1.19628096e-02 -1.11187994e+00 -3.79831165e-01
6.04892135e-01 -1.37469575e-01 7.10848749e-01 3.66042286e-01
1.87752295e+00 -6.92866385e-01 -7.40884021e-02 1.25272501e+00
1.66318035e+00 4.09450382e-01 7.45315611e-01 1.10746190e-01
6.39762163e-01 3.38228106e-01 2.91648269e-01 6.36718333e-01
6.19163930e-01 3.31520945e-01 6.65368557e-01 -1.05490990e-01
-2.99877435e-01 -1.86616257e-02 3.39270830e-01 1.59238625e+00
-2.55670428e-01 -2.70690292e-01 -4.82634902e-01 2.88015336e-01
-1.64955533e+00 -9.23192441e-01 2.58287549e-01 1.87711716e+00
6.27321601e-01 -2.04558209e-01 -3.56574148e-01 3.35200012e-01
4.92765099e-01 -2.73758434e-02 -6.63106680e-01 -6.50829077e-01
1.91572800e-01 2.43210241e-01 5.97475052e-01 2.26573497e-01
-7.08191395e-01 6.85077190e-01 6.54592133e+00 8.10000241e-01
-1.51250720e+00 5.49439490e-01 8.13377738e-01 -5.70410311e-01
-2.22743496e-01 -3.07371587e-01 -4.96070802e-01 7.09859610e-01
1.45819700e+00 -1.12861164e-01 8.36775482e-01 1.13961017e+00
2.00096264e-01 -1.93756893e-01 -7.37465978e-01 1.24330604e+00
1.26855031e-01 -1.65614450e+00 -6.39982373e-02 2.29841843e-01
5.87402940e-01 4.18998182e-01 -3.33920792e-02 7.52258748e-02
-1.08644091e-01 -5.27856231e-01 9.55360711e-01 5.99514127e-01
9.41499591e-01 -1.19621384e+00 9.52932715e-01 6.03670299e-01
-1.24245930e+00 -2.61832505e-01 -7.48656571e-01 -1.46203950e-01
-2.57215470e-01 7.54454672e-01 -4.33945239e-01 4.70159113e-01
1.12141097e+00 5.32849371e-01 -4.25769091e-01 9.74380314e-01
4.27684724e-01 3.44885647e-01 -1.79477960e-01 9.31320488e-02
-5.47239073e-02 -1.53029203e-01 -5.47279045e-03 9.97184873e-01
7.33667433e-01 2.94436533e-02 -1.10988198e-02 4.37360018e-01
-4.48133260e-01 -3.93444533e-03 -6.57705903e-01 1.86235726e-01
7.15366244e-01 1.13058174e+00 -6.43547654e-01 -4.90094751e-01
-3.84426564e-01 1.58476901e+00 4.62973475e-01 2.05563486e-01
-9.35350120e-01 -5.62774181e-01 7.57906854e-01 -2.94481575e-01
4.43309546e-01 -2.73363262e-01 -2.03287512e-01 -1.09011984e+00
1.91651285e-01 -6.49651825e-01 -2.48891711e-01 -7.74484456e-01
-5.41951478e-01 9.26088572e-01 -4.74286079e-01 -1.02013290e+00
-2.90660590e-01 -1.64037392e-01 -4.96912241e-01 7.97772586e-01
-1.75335586e+00 -7.46001899e-01 -6.18525803e-01 9.49132383e-01
5.63377678e-01 -2.99403518e-01 7.84524381e-01 7.81937182e-01
-5.08626223e-01 6.89926744e-01 3.25600475e-01 -1.62511662e-01
8.61406475e-02 -7.25501478e-01 3.66094202e-01 9.50870633e-01
9.79382545e-02 3.54175031e-01 5.17063022e-01 -4.54179108e-01
-1.90478659e+00 -1.18520284e+00 7.95833349e-01 4.46304053e-01
6.51035160e-02 -1.97462484e-01 -6.69510484e-01 1.19697638e-01
2.87493944e-01 3.49194884e-01 7.03629613e-01 -4.94795054e-01
3.19189504e-02 -5.24926960e-01 -1.56735265e+00 4.00161594e-01
9.60223734e-01 -7.51178861e-01 9.40344110e-02 2.55149961e-01
8.95819962e-01 -2.75023818e-01 -6.81831658e-01 3.31783257e-02
5.43803453e-01 -1.03852522e+00 8.04792047e-01 -8.68105292e-02
5.98889232e-01 -3.58215650e-03 -5.73155642e-01 -1.09804368e+00
-2.78198749e-01 -4.11232769e-01 -6.74118519e-01 9.70532954e-01
7.29665207e-03 -3.91681522e-01 1.16788447e+00 4.81038123e-01
-2.88183212e-01 -9.68603969e-01 -1.25632608e+00 -4.48864222e-01
-4.08393532e-01 -6.28389835e-01 6.98564053e-01 5.76936722e-01
-3.77360970e-01 -3.90287787e-01 -2.08375037e-01 1.80507660e-01
4.96726304e-01 -2.96983495e-02 4.24258769e-01 -9.07210290e-01
-4.45188552e-01 2.11283891e-03 -4.11852300e-01 -1.15108430e+00
2.38727078e-01 -9.22091484e-01 1.60734653e-02 -1.67147553e+00
2.18901783e-01 -6.73615575e-01 -2.66215026e-01 3.42465758e-01
2.52205312e-01 6.07854784e-01 4.52652425e-01 4.48249459e-01
-6.21743858e-01 5.69538951e-01 1.15254354e+00 4.64119017e-02
-6.57674484e-03 -2.44147405e-02 -6.73347950e-01 7.12175965e-01
8.05021644e-01 -4.45789665e-01 -6.60034180e-01 -9.58608985e-01
9.75003168e-02 1.56454310e-01 2.34077007e-01 -1.21984971e+00
8.46181333e-01 1.92054659e-01 4.41596091e-01 -5.66841245e-01
5.76284349e-01 -1.32672238e+00 1.62394240e-01 7.29799807e-01
6.47501275e-02 3.39170665e-01 -8.78917575e-02 4.04073477e-01
-3.82503748e-01 -3.85701805e-01 5.70202708e-01 -1.11120917e-01
-6.14872217e-01 2.72760659e-01 -4.87310976e-01 -4.51892167e-01
9.51997399e-01 -3.28075260e-01 -2.62450039e-01 -4.19611722e-01
-3.67737621e-01 -2.49080122e-01 3.27301204e-01 7.40036741e-02
1.07794082e+00 -1.14140773e+00 -2.81685174e-01 4.57327992e-01
-7.04900384e-01 2.33946107e-02 4.73533362e-01 7.23442376e-01
-1.10564113e+00 3.13016057e-01 -3.57330412e-01 -3.01136106e-01
-1.27296150e+00 4.43786770e-01 3.26575100e-01 -1.42015461e-02
-5.46968460e-01 8.95738482e-01 -3.68399441e-01 8.40328932e-02
4.07770753e-01 -1.92968503e-01 -4.38207351e-02 -1.11833818e-01
7.60345936e-01 6.94426000e-01 3.34774375e-01 -3.90204877e-01
-2.63904870e-01 2.46797353e-01 7.43713602e-02 1.32611290e-01
1.58655667e+00 -4.62886482e-01 -5.31238973e-01 -5.07730432e-02
1.59226573e+00 -8.86114612e-02 -1.32793117e+00 -6.36125281e-02
-4.11898077e-01 -3.66231650e-01 7.28920996e-01 -2.51645923e-01
-1.92754984e+00 6.93633616e-01 9.37495887e-01 1.55156463e-01
1.82220948e+00 -3.52155298e-01 1.24983275e+00 6.61277652e-01
5.23762941e-01 -1.21554220e+00 -8.15401971e-02 4.68613088e-01
4.28895026e-01 -1.00686896e+00 -4.57125250e-03 -1.24824390e-01
-3.82922441e-01 1.12699819e+00 4.01120991e-01 -2.90926814e-01
9.44012463e-01 8.32360685e-01 -3.36072415e-01 -3.24583463e-02
-9.33662415e-01 1.90696493e-01 -1.32647723e-01 4.54060316e-01
1.73814625e-01 -1.12478118e-02 -1.14925429e-02 4.96043831e-01
-4.08712327e-01 3.78317714e-01 4.40674990e-01 1.02932930e+00
-4.83320266e-01 -8.61726761e-01 -9.47586074e-02 5.79606712e-01
-7.61932373e-01 -2.62998730e-01 1.28714234e-01 6.33781999e-02
5.55852771e-01 1.12820876e+00 5.88347554e-01 -8.61415029e-01
-4.34817187e-02 -2.17882872e-01 3.46964598e-01 -2.53836602e-01
-7.11983860e-01 -4.13701832e-02 -8.05704147e-02 -8.12506020e-01
-2.97507703e-01 -1.54694974e-01 -1.16143739e+00 -8.76627684e-01
-3.23839575e-01 1.28522038e-01 1.33961463e+00 8.49895597e-01
7.04628468e-01 6.04077458e-01 8.81733954e-01 -1.15974247e+00
-2.01161444e-01 -5.62441409e-01 -5.37535906e-01 -2.16040716e-01
3.72474283e-01 -1.95922563e-03 -1.79543763e-01 2.12566599e-01] | [8.455527305603027, 2.8344573974609375] |
78d5486d-e233-483a-af81-535e63960f57 | ign-implicit-generative-networks | 2206.05860 | null | https://arxiv.org/abs/2206.05860v2 | https://arxiv.org/pdf/2206.05860v2.pdf | IGN : Implicit Generative Networks | In this work, we build recent advances in distributional reinforcement learning to give a state-of-art distributional variant of the model based on the IQN. We achieve this by using the GAN model's generator and discriminator function with the quantile regression to approximate the full quantile value for the state-action return distribution. We demonstrate improved performance on our baseline dataset - 57 Atari 2600 games in the ALE. Also, we use our algorithm to show the state-of-art training performance of risk-sensitive policies in Atari games with the policy optimization and evaluation. | ['Jianfen Zhang', 'Zhijun Yan', 'Feiyu Han', 'Tianyi Wu', 'Haozheng Luo'] | 2022-06-13 | null | null | null | null | ['distributional-reinforcement-learning'] | ['methodology'] | [-4.42857653e-01 2.36250401e-01 -3.82556587e-01 -3.23627144e-01
-1.41886401e+00 -6.87064469e-01 8.47129345e-01 -4.40640897e-01
-6.48089170e-01 1.33149731e+00 3.51460785e-01 -7.63559461e-01
-2.48019621e-01 -9.67793107e-01 -8.01799655e-01 -7.99313664e-01
-3.08605760e-01 1.07453489e+00 -1.31449327e-01 -5.23906350e-01
2.15323474e-02 -7.92944804e-03 -1.15839124e+00 1.08546056e-01
8.51101935e-01 1.03098035e+00 -3.79648328e-01 9.66647446e-01
2.20361993e-01 1.29757965e+00 -1.09565234e+00 -7.40733743e-01
4.64771807e-01 -5.06370664e-01 -4.45122212e-01 -7.02942431e-01
8.72675255e-02 -9.17480588e-01 -5.48001289e-01 1.15390873e+00
7.91657090e-01 4.60880309e-01 1.19889319e+00 -1.19784784e+00
-7.33766496e-01 9.51217055e-01 -5.97470403e-01 2.64011323e-01
1.26859779e-02 3.27068686e-01 1.12068450e+00 -1.74124271e-01
2.45990649e-01 1.41424930e+00 3.54174554e-01 8.54087591e-01
-1.21452618e+00 -6.97012901e-01 -4.51269820e-02 -1.19959772e-01
-1.00539005e+00 9.71197486e-02 2.58398145e-01 -2.66206801e-01
1.06979930e+00 -1.67154655e-01 5.66378117e-01 1.71691561e+00
3.82014126e-01 1.01569974e+00 1.44664264e+00 -3.15529436e-01
7.62834609e-01 -3.28770071e-01 -4.54908937e-01 3.23063254e-01
-1.61592841e-01 1.11896980e+00 2.32648849e-02 -5.16478777e-01
9.24565494e-01 -2.47377440e-01 4.09087598e-01 -2.29204401e-01
-3.57410848e-01 1.45331728e+00 8.07519183e-02 -3.49622875e-01
-4.69032377e-01 9.53020811e-01 5.60480177e-01 4.48316067e-01
6.99771583e-01 3.06108981e-01 -3.27953100e-01 -7.79607534e-01
-8.54479313e-01 8.24677169e-01 7.71700501e-01 9.58718002e-01
2.27209076e-01 7.16936290e-01 -9.45107758e-01 6.33496940e-01
2.44253576e-01 8.38259637e-01 4.11977112e-01 -1.33007944e+00
6.48816109e-01 -3.80407423e-01 5.06607413e-01 1.15406401e-01
2.57801022e-02 -3.78179252e-01 -3.88573229e-01 6.80408776e-01
5.34188867e-01 -7.92241693e-01 -1.04895794e+00 1.94846940e+00
-6.22188374e-02 4.43926990e-01 2.61173248e-01 5.31786740e-01
-1.23414956e-02 6.57379329e-01 6.14664078e-01 7.12081045e-02
7.29757428e-01 -6.20207429e-01 -4.79167789e-01 2.67453194e-01
4.36602861e-01 -3.26092780e-01 1.23596144e+00 4.54816252e-01
-1.17747903e+00 -2.07545027e-01 -7.26549208e-01 5.45612752e-01
-5.05982377e-02 -1.40615851e-01 4.47510749e-01 8.48396361e-01
-1.07887745e+00 7.70153224e-01 -7.78144002e-01 1.64913401e-01
5.43853700e-01 1.15720652e-01 3.92473936e-01 3.39852393e-01
-1.60629797e+00 9.84013736e-01 5.42341292e-01 -6.42010689e-01
-2.05550790e+00 -9.86216068e-01 -6.92965448e-01 1.42075848e-02
5.33974349e-01 -4.96903092e-01 2.08220816e+00 -6.32143199e-01
-2.22525883e+00 3.20702523e-01 6.72037363e-01 -1.08894765e+00
8.11769247e-01 -5.48374534e-01 -1.89250633e-01 -5.70061915e-02
-2.70346217e-02 3.80337447e-01 7.44349182e-01 -1.00093412e+00
-7.25185037e-01 -4.62967940e-02 1.46367431e-01 4.01497841e-01
1.89131752e-01 -1.71877351e-02 3.69082153e-01 -8.56889963e-01
-1.40030849e+00 -7.23549843e-01 -3.41411352e-01 -9.05471265e-01
-2.65678316e-01 -4.89142448e-01 2.63897210e-01 -6.42706454e-01
1.21570385e+00 -1.89023638e+00 -1.98861863e-02 3.13100308e-01
-3.05021167e-01 2.55626261e-01 -2.05238864e-01 6.03535831e-01
-1.19242303e-01 2.48553351e-01 1.53203094e-02 -4.44404125e-01
6.57997251e-01 4.78995591e-01 -1.04251480e+00 3.34268540e-01
-3.66679244e-02 1.08120918e+00 -9.62819040e-01 3.35748494e-02
2.26665616e-01 -5.96306995e-02 -7.04926610e-01 7.61908233e-01
-8.67973924e-01 2.69074738e-01 -4.06982988e-01 3.27154011e-01
2.67943144e-01 5.09532750e-01 -4.30097319e-02 6.63874567e-01
1.79964036e-01 2.69424051e-01 -5.80694318e-01 1.70932508e+00
-4.49884713e-01 -5.13183400e-02 -1.40102088e-01 -1.04755163e+00
7.40870297e-01 2.29325935e-01 4.08903927e-01 -8.46152544e-01
2.81838179e-01 1.01579256e-01 3.94769609e-02 -1.63751822e-02
4.89267886e-01 -4.97147620e-01 -5.36206126e-01 7.64034569e-01
2.06679612e-01 -3.16322178e-01 3.38122733e-02 1.09831706e-01
9.45992112e-01 7.94345438e-01 1.03616558e-01 -5.23692846e-01
-1.72410399e-01 -1.86008349e-01 1.35432348e-01 1.26306260e+00
-2.24067003e-01 2.98376620e-01 1.06752205e+00 -2.22844511e-01
-1.38329005e+00 -1.68541205e+00 6.40277788e-02 1.50295389e+00
-7.25242198e-01 -2.93362945e-01 -9.34762180e-01 -1.07503355e+00
2.21987262e-01 1.61864817e+00 -1.19186318e+00 -3.62291068e-01
-3.73065323e-01 -6.53327107e-01 1.02071893e+00 8.22778106e-01
3.76120865e-01 -1.21947432e+00 -5.96265793e-01 2.13004097e-01
3.59162360e-01 -3.70313346e-01 -4.16747987e-01 4.87081110e-01
-3.68341893e-01 -7.30326653e-01 -7.60661840e-01 -5.87677099e-02
-1.77188128e-01 -7.76694536e-01 1.58173335e+00 -7.06913531e-01
1.71251938e-01 5.83587170e-01 -1.39872938e-01 -8.72371554e-01
-7.70654261e-01 7.98914954e-02 -3.01446170e-02 -6.47802889e-01
9.62976962e-02 -5.76771677e-01 -5.86408615e-01 3.35269310e-02
-8.81629765e-01 -7.68818200e-01 2.49431148e-01 1.06226182e+00
6.15205050e-01 -1.01145737e-01 7.38662720e-01 -9.21673775e-01
1.10429704e+00 -7.15195119e-01 -1.04175818e+00 3.10734510e-01
-5.45850694e-01 2.79774159e-01 8.56393814e-01 -3.84429783e-01
-1.36181092e+00 -5.59888721e-01 -4.64233398e-01 -6.06431603e-01
-7.10835308e-02 2.55925924e-01 1.58712585e-02 7.14856923e-01
8.57948303e-01 4.12862934e-03 -6.59421310e-02 -1.00885034e-01
9.20494556e-01 5.47879815e-01 5.02224803e-01 -1.49012434e+00
5.79398513e-01 3.41271199e-02 -1.07606076e-01 1.50015242e-02
-1.17060006e+00 8.01608711e-02 1.81476519e-01 1.09857775e-03
9.34880733e-01 -1.05418134e+00 -9.10391629e-01 2.65725344e-01
-5.18347979e-01 -1.28298485e+00 -1.07158446e+00 3.15184563e-01
-1.56596053e+00 -2.39564881e-01 -6.87807977e-01 -1.19724762e+00
-4.38144326e-01 -1.00234854e+00 1.07171392e+00 2.05551594e-01
2.94466853e-01 -1.16835117e+00 8.74542654e-01 -3.76399636e-01
6.24227226e-01 1.62158504e-01 9.38651204e-01 -8.56261969e-01
8.42401385e-02 4.25712973e-01 2.48548225e-01 7.05549181e-01
-1.41237482e-01 -1.84592351e-01 -8.36598396e-01 -4.38788712e-01
-1.28718138e-01 -9.35986221e-01 8.90689611e-01 8.56758893e-01
1.41858482e+00 -3.01021487e-01 3.39407593e-01 7.30978549e-01
1.50840247e+00 4.70575631e-01 9.20503199e-01 4.48882937e-01
2.30941013e-01 2.35695578e-02 8.10131609e-01 7.51484692e-01
1.40572771e-01 4.69343692e-01 6.77689493e-01 2.33733550e-01
2.13371173e-01 -1.03985536e+00 8.42152655e-01 -1.25082701e-01
-1.45037040e-01 -2.77633071e-01 -6.91058278e-01 3.67908508e-01
-1.90385175e+00 -1.32830906e+00 7.23154485e-01 2.23564625e+00
1.01961911e+00 1.08407438e-01 8.88989687e-01 -7.42877662e-01
3.95971477e-01 8.87426734e-02 -7.77363837e-01 -8.72069359e-01
1.40331119e-01 9.87315774e-01 8.60021234e-01 6.45774662e-01
-1.00068212e+00 1.26067150e+00 8.10942078e+00 1.44598806e+00
-4.77086633e-01 1.52481407e-01 1.14049923e+00 -3.65319014e-01
-4.29255366e-01 -3.50147545e-01 -8.14329863e-01 5.02666175e-01
1.58212280e+00 -3.38156730e-01 9.26418960e-01 1.15202498e+00
8.88885930e-02 1.92293115e-02 -8.30136657e-01 6.09659076e-01
-3.49005938e-01 -1.01298344e+00 2.80439518e-02 3.62600118e-01
1.01689208e+00 2.05857083e-01 7.36143947e-01 1.17871046e+00
1.78395176e+00 -1.55767286e+00 7.92639852e-01 3.70287150e-01
1.14356613e+00 -1.63235295e+00 6.96104109e-01 2.56102145e-01
-4.92212772e-01 -3.05255979e-01 -8.31346154e-01 4.28287648e-02
-1.01409845e-01 1.01529017e-01 -7.91728079e-01 5.41748464e-01
5.64875424e-01 3.64017099e-01 1.66892428e-02 5.67171514e-01
-7.05584824e-01 1.24850380e+00 -1.57841057e-01 6.79400265e-02
6.99985743e-01 -5.57594895e-01 4.57868129e-01 8.62401783e-01
3.84279519e-01 -4.42204177e-02 3.04165572e-01 9.42705214e-01
-1.40178651e-01 -2.81343579e-01 -8.65309417e-01 3.81878093e-02
3.24802428e-01 9.40878272e-01 1.45310163e-01 -4.40373838e-01
-3.68980062e-03 4.51755673e-01 7.72944152e-01 5.24482608e-01
-1.30938911e+00 -8.01956728e-02 1.03028584e+00 -2.52726287e-01
5.73697627e-01 -5.95713481e-02 1.11210115e-01 -9.34497654e-01
-6.90567553e-01 -1.09169555e+00 8.44354331e-01 -5.54226100e-01
-1.70429933e+00 3.01538885e-01 5.02864778e-01 -7.75307178e-01
-1.12324011e+00 -7.77434647e-01 -8.33859324e-01 1.00291777e+00
-1.41479790e+00 -9.31832016e-01 8.28355074e-01 8.59220862e-01
2.26317167e-01 -6.76137686e-01 9.07347381e-01 -2.08572447e-01
-2.89178133e-01 8.45200062e-01 6.25188470e-01 2.54026473e-01
7.02232838e-01 -2.07088709e+00 5.47824204e-01 4.66868877e-01
7.16476962e-02 2.01596022e-01 6.93628013e-01 -6.99284077e-01
-9.20249760e-01 -1.15077126e+00 -4.38841611e-01 -6.77692592e-01
1.07226694e+00 -2.78172225e-01 -1.82290614e-01 1.06920660e+00
6.69320643e-01 -3.83325100e-01 6.37167990e-01 7.99894631e-02
-3.09500933e-01 5.84183484e-02 -1.24147022e+00 6.22680485e-01
7.63920486e-01 -3.77324492e-01 -7.02148199e-01 5.15075587e-02
7.26663768e-01 -5.63077509e-01 -7.84035563e-01 -2.63208486e-02
1.59997374e-01 -9.31765556e-01 8.64116073e-01 -1.42126596e+00
5.96449196e-01 1.50548279e-01 -4.17015731e-01 -2.15627027e+00
-1.07832365e-01 -1.17460823e+00 -2.73998022e-01 1.10973930e+00
2.02803820e-01 -6.77807987e-01 7.29919851e-01 3.47867191e-01
-2.23424986e-01 -6.92140937e-01 -1.32214570e+00 -1.04651833e+00
1.17845786e+00 -5.14985323e-01 8.68271530e-01 1.93163097e-01
-1.21399999e-01 -2.41900738e-02 -7.53706038e-01 -3.38826180e-01
8.37679505e-01 -1.08488262e-01 5.70297718e-01 -4.81415957e-01
-7.48854280e-01 -4.47338223e-01 8.08831900e-02 -8.36182296e-01
7.28128314e-01 -7.03378975e-01 1.35717407e-01 -1.05738676e+00
1.56206295e-01 -3.49050462e-01 -6.71670377e-01 4.14545834e-01
-1.47322923e-01 -1.03754178e-01 1.02678753e-01 -3.08972597e-01
-7.76916087e-01 1.08150434e+00 1.09765744e+00 9.54688434e-03
-4.02784441e-03 1.55580819e-01 -8.50693524e-01 3.93850565e-01
1.19528866e+00 -5.74950755e-01 -9.08825099e-01 -5.69939017e-02
1.93532720e-01 2.88133085e-01 2.02826306e-01 -6.41907334e-01
-5.14488339e-01 -5.56573629e-01 4.10856128e-01 -3.11573505e-01
1.25969499e-01 -3.51094365e-01 -1.74723431e-01 4.96955603e-01
-7.04145014e-01 4.66995120e-01 2.93681741e-01 7.03477025e-01
-1.06643364e-02 -6.08600080e-02 7.93172419e-01 -1.42015994e-01
-3.04708719e-01 5.40236294e-01 -4.48734701e-01 9.40683246e-01
1.12961602e+00 7.59743989e-01 -6.16162598e-01 -9.86161053e-01
-5.90205312e-01 2.92798012e-01 3.11313778e-01 -1.45086451e-02
3.22230875e-01 -1.45860553e+00 -1.02860963e+00 -1.51440054e-01
-2.85348177e-01 -2.60626286e-01 -4.53456631e-03 1.28413141e-01
-2.31883258e-01 1.28836408e-01 -4.21898991e-01 -1.47110090e-01
-2.32724383e-01 5.68673134e-01 7.53555179e-01 -9.23419058e-01
-2.84595013e-01 5.74106693e-01 1.68387711e-01 -4.40696180e-01
6.30259365e-02 8.60898271e-02 8.11249837e-02 -2.55271882e-01
6.84776545e-01 3.87716174e-01 -3.88893187e-01 5.75097315e-02
8.42808485e-02 1.30538289e-02 2.46218499e-02 -8.35928500e-01
1.20508194e+00 3.96829039e-01 4.08867180e-01 2.97549516e-01
6.74680293e-01 -8.57417583e-02 -1.91373050e+00 1.74557477e-01
-3.67581815e-01 -4.07633841e-01 -8.76301751e-02 -1.23474967e+00
-8.82456124e-01 8.04493725e-01 5.79774261e-01 1.24706529e-01
6.73829436e-01 -7.75476918e-02 4.05444741e-01 -8.92155617e-02
5.97868681e-01 -1.31860340e+00 1.73496410e-01 8.07174027e-01
6.71651363e-01 -8.35975051e-01 -1.87259704e-01 6.14099622e-01
-1.19366467e+00 7.63642490e-01 6.49798572e-01 -8.17321777e-01
7.37709463e-01 6.06862366e-01 6.31705299e-02 1.18300676e-01
-9.99291658e-01 -2.40171507e-01 2.49779578e-02 1.18850625e+00
4.80510108e-02 5.67012191e-01 -2.79655196e-02 6.43753707e-01
-5.33020198e-01 -2.67936081e-01 4.36819851e-01 6.27860546e-01
-9.28132832e-02 -1.47136223e+00 6.29721358e-02 4.41836506e-01
-9.29859698e-01 -2.21219346e-01 5.59206568e-02 9.68634844e-01
-6.02515876e-01 8.77163112e-01 9.91265848e-02 -2.44016200e-01
2.56995469e-01 1.87889218e-01 7.08885193e-01 -4.74189490e-01
-7.04673707e-01 1.45894811e-02 1.20137937e-01 -7.10638046e-01
5.47331385e-02 -5.41009188e-01 -1.03770006e+00 -5.10856748e-01
2.86971092e-01 3.45455855e-01 3.84539157e-01 6.97526693e-01
8.76743644e-02 4.44174796e-01 6.01376414e-01 -5.52234352e-01
-1.95883584e+00 -1.02527130e+00 -9.62192893e-01 4.56642628e-01
6.53100014e-03 -8.14915240e-01 -2.03727171e-01 -6.31937206e-01] | [4.071872234344482, 2.5417487621307373] |
9b6afa2c-6f14-433b-9196-fdc5cb12606e | imbalanced-classification-in-faulty-turbine | 2301.04049 | null | https://arxiv.org/abs/2301.04049v1 | https://arxiv.org/pdf/2301.04049v1.pdf | Imbalanced Classification In Faulty Turbine Data: New Proximal Policy Optimization | There is growing importance to detecting faults and implementing the best methods in industrial and real-world systems. We are searching for the most trustworthy and practical data-based fault detection methods proposed by artificial intelligence applications. In this paper, we propose a framework for fault detection based on reinforcement learning and a policy known as proximal policy optimization. As a result of the lack of fault data, one of the significant problems with the traditional policy is its weakness in detecting fault classes, which was addressed by changing the cost function. Using modified Proximal Policy Optimization, we can increase performance, overcome data imbalance, and better predict future faults. When our modified policy is implemented, all evaluation metrics will increase by $3\%$ to $4\%$ as compared to the traditional policy in the first benchmark, between $20\%$ and $55\%$ in the second benchmark, and between $6\%$ and $14\%$ in the third benchmark, as well as an improvement in performance and prediction speed compared to previous methods. | ['Arash Ghahremani', 'Mostafa Yari', 'Mahdi Aliyari Shoorehdeli', 'Mohammad Hossein Modirrousta'] | 2023-01-10 | null | null | null | null | ['imbalanced-classification', 'fault-detection'] | ['miscellaneous', 'miscellaneous'] | [-2.30521217e-01 -1.25125095e-01 -1.29372239e-01 -8.25325865e-03
-3.03345948e-01 7.42292181e-02 3.44006270e-02 5.04238248e-01
-3.15003455e-01 1.09733021e+00 -5.50610006e-01 -2.36487612e-01
-4.99778539e-01 -1.05391455e+00 -6.45240545e-01 -3.78139019e-01
-4.28886622e-01 4.55847889e-01 5.07224262e-01 -2.22875357e-01
6.18924677e-01 3.72061908e-01 -1.71903729e+00 6.16417751e-02
1.06807292e+00 1.38289583e+00 -9.87142045e-03 2.92172283e-01
-2.66001429e-02 6.16707385e-01 -9.29157376e-01 -6.38127849e-02
3.07900280e-01 -4.57449168e-01 -6.81780398e-01 -6.99679181e-02
-1.36633188e-01 -2.79177427e-01 -1.66153014e-02 1.12250495e+00
3.61179471e-01 4.64120470e-02 2.25233525e-01 -1.48259652e+00
-2.08108430e-03 5.40073693e-01 -6.04461074e-01 1.10545782e-02
2.80727983e-01 2.45692432e-01 5.70928097e-01 -4.56271648e-01
2.23376751e-01 9.33971286e-01 5.24138689e-01 3.86703312e-01
-9.16027963e-01 -3.65058750e-01 2.51679182e-01 5.83732426e-01
-1.20674717e+00 -1.14256412e-01 7.05035985e-01 -3.16182375e-01
1.18789148e+00 3.39000076e-01 5.31355262e-01 5.15856922e-01
3.73899907e-01 5.90560973e-01 1.09747696e+00 -6.17095828e-01
5.16572833e-01 -1.09087817e-01 3.16419937e-02 8.02420914e-01
5.13072133e-01 1.27666727e-01 -6.04377054e-02 -1.85966015e-01
6.11356854e-01 2.62138993e-01 -1.96704701e-01 -7.22705573e-02
-9.79531407e-01 6.64693832e-01 2.68975765e-01 3.57259363e-01
-4.53086913e-01 1.15596183e-01 6.16934478e-01 3.36652040e-01
4.12384242e-01 9.24824178e-01 -6.04228437e-01 -5.39887428e-01
-9.48856950e-01 2.80932844e-01 6.85518265e-01 4.10862148e-01
7.46589601e-01 4.20568109e-01 -1.65000390e-02 8.34527493e-01
1.15534082e-01 4.88995731e-01 6.82359457e-01 -1.13432252e+00
2.17602953e-01 9.99561071e-01 2.40370542e-01 -8.55527163e-01
-3.35042328e-01 -5.26379526e-01 -7.27123737e-01 7.09730148e-01
3.73882264e-01 -2.37243325e-01 -7.97472417e-01 1.42704511e+00
9.35033560e-02 8.88121724e-02 8.17823410e-03 6.37592852e-01
-9.25251916e-02 4.61690813e-01 -4.35457587e-01 -5.93080342e-01
1.05828452e+00 -9.41517591e-01 -5.56113899e-01 -2.54886657e-01
5.29927433e-01 -8.12291145e-01 9.66046274e-01 7.31328487e-01
-1.06091881e+00 -5.86577654e-01 -1.25413930e+00 8.55534494e-01
-1.71877384e-01 -1.54633701e-01 4.13226038e-01 6.32657647e-01
-7.95879424e-01 1.03529501e+00 -8.96827638e-01 -3.11978251e-01
1.60897493e-01 4.99960482e-01 9.79182571e-02 -3.60471606e-01
-1.11699069e+00 8.35283697e-01 3.70192796e-01 -1.43014386e-01
-8.78558040e-01 -5.18040776e-01 -5.64671218e-01 1.94767639e-01
7.08009481e-01 -3.25258493e-01 1.10860360e+00 -7.06488013e-01
-1.36380267e+00 -5.99596947e-02 1.08351953e-01 -6.93379283e-01
4.98997360e-01 -3.68745893e-01 -5.76724112e-01 -1.02072708e-01
5.43939508e-02 5.37774749e-02 6.16301775e-01 -1.04155278e+00
-9.27569568e-01 -4.10089046e-01 2.04316396e-02 -2.15874407e-02
-2.13964775e-01 -4.25761729e-01 -1.86322350e-02 -4.66672987e-01
5.22027053e-02 -7.35193133e-01 -3.23221266e-01 -2.68644035e-01
-1.79061338e-01 -2.97152847e-01 1.15791368e+00 -6.28176272e-01
1.31469965e+00 -2.18536401e+00 -2.76040763e-01 3.56381357e-01
-5.53410649e-02 4.99118596e-01 2.16042250e-01 4.02431309e-01
-5.64406626e-02 5.12666106e-02 -2.01963499e-01 8.83853287e-02
-6.60513490e-02 4.45927799e-01 -8.29260126e-02 2.69702852e-01
2.31047526e-01 2.86119610e-01 -8.30561817e-01 -1.63714252e-02
3.25799584e-01 -1.07325606e-01 -5.48556447e-01 3.02898616e-01
-3.07989359e-01 -7.60520548e-02 -3.56013417e-01 6.41689777e-01
5.39715827e-01 -1.74239606e-01 1.97863042e-01 1.65961161e-01
-2.67076679e-02 1.05524436e-01 -1.57434416e+00 1.17354810e+00
-3.36071700e-01 1.79949045e-01 2.92120725e-02 -1.58276606e+00
1.33369887e+00 4.58749324e-01 9.00434077e-01 -1.03060865e+00
-1.04806393e-01 3.62841874e-01 -4.57312316e-02 -4.63046193e-01
2.96294063e-01 3.92948985e-02 -3.35464254e-02 4.45722520e-01
-2.93464661e-01 -1.04214683e-01 5.23048460e-01 -2.85999149e-01
1.56813192e+00 -1.01268440e-01 9.42910090e-02 -4.57825869e-01
5.93786895e-01 9.21547040e-02 1.05206549e+00 5.49727201e-01
-4.11099166e-01 8.61145854e-02 7.84049392e-01 -7.13108003e-01
-6.70017183e-01 -7.26003289e-01 4.66939658e-02 6.48342669e-01
1.48373753e-01 -1.94676250e-01 -7.98444331e-01 -8.60609531e-01
3.35200638e-01 8.04793298e-01 -2.90195674e-01 -5.32261431e-01
-4.89240438e-01 -8.65013242e-01 2.93088198e-01 4.53609586e-01
8.25123787e-01 -1.09112763e+00 -8.33483279e-01 3.80091310e-01
7.22778663e-02 -6.64675891e-01 8.56706202e-02 4.06488538e-01
-9.80898738e-01 -1.18946481e+00 -4.60393190e-01 -3.00965518e-01
7.47682750e-01 -1.31986678e-01 1.11007118e+00 4.62925792e-01
-4.71883208e-01 -1.50143161e-01 -5.47915161e-01 -5.29727757e-01
-4.60634947e-01 -1.56117365e-01 1.99485719e-01 -2.35215992e-01
-4.91060540e-02 -2.47289494e-01 -7.42984295e-01 4.38734829e-01
-8.24945271e-01 -4.44363415e-01 6.83405399e-01 1.09250224e+00
3.00118834e-01 5.28632343e-01 9.56154823e-01 -1.03964651e+00
5.99077702e-01 -3.69443148e-01 -8.65499437e-01 1.42711475e-01
-1.55710781e+00 2.56852865e-01 8.42960298e-01 -8.85882080e-02
-7.52254784e-01 -2.61160582e-01 -1.12650447e-01 -4.69716102e-01
-6.38008043e-02 4.43594784e-01 -1.75928529e-02 1.46181583e-01
7.14676917e-01 7.10624903e-02 2.62977123e-01 -3.46199334e-01
-1.56519458e-01 4.42977250e-01 1.24233246e-01 -5.10902166e-01
4.08669174e-01 2.06044540e-01 1.39251668e-02 -3.49262357e-01
-2.92033106e-01 -6.80893287e-02 -2.62643043e-02 -1.79565996e-01
2.69090623e-01 -4.52731282e-01 -8.35296333e-01 4.37735170e-01
-6.04851484e-01 -2.82213807e-01 -5.26453137e-01 4.63899165e-01
-5.19387901e-01 4.35492754e-01 -5.60623229e-01 -7.37403214e-01
-4.67752814e-01 -1.47002995e+00 6.40319705e-01 1.20265976e-01
-2.27225170e-01 -6.32815182e-01 -1.63266867e-01 1.19769417e-01
7.13025510e-01 4.45088863e-01 1.16737771e+00 -3.01673383e-01
-3.03591400e-01 -3.06497365e-01 -9.69009399e-02 8.22462261e-01
3.85435700e-01 8.90426636e-02 -4.67166722e-01 -7.30790794e-01
2.34787226e-01 7.55339563e-02 5.28523445e-01 1.56178460e-01
1.00904071e+00 -3.38714093e-01 -3.65238070e-01 4.42269519e-02
1.60906184e+00 8.41207683e-01 6.71897173e-01 4.42696214e-01
1.82952687e-01 3.39543909e-01 9.19700861e-01 6.18225098e-01
-1.03573263e-01 5.43417811e-01 7.46962965e-01 -1.55918926e-01
6.82202280e-02 1.93345353e-01 3.73394698e-01 5.61585784e-01
2.00813189e-01 1.44817337e-01 -1.02021837e+00 6.07735872e-01
-2.00153708e+00 -6.32813811e-01 1.16719365e-01 2.44453478e+00
6.28350675e-01 7.20854700e-01 6.60897642e-02 7.87394881e-01
7.34620452e-01 -4.32539791e-01 -6.87607408e-01 -5.70770621e-01
4.72698390e-01 2.60172904e-01 3.58658314e-01 6.65941164e-02
-7.03759432e-01 5.04101336e-01 6.29230261e+00 6.64131105e-01
-1.58360767e+00 -1.91833764e-01 6.27098322e-01 -6.66249469e-02
2.32439533e-01 1.17640026e-01 -4.04864281e-01 9.14939642e-01
1.01876080e+00 -2.72921324e-01 4.12457049e-01 1.10234177e+00
1.46782726e-01 -5.30968428e-01 -1.06189835e+00 8.66017222e-01
-1.00448810e-01 -1.09595990e+00 -4.93861288e-01 -1.00556970e-01
7.91569710e-01 -1.78207755e-01 -2.44959950e-01 3.74059200e-01
2.58011401e-01 -7.27113008e-01 5.95810890e-01 3.94815177e-01
3.34031165e-01 -1.20451236e+00 1.21210563e+00 3.46504480e-01
-7.48752415e-01 -4.56001401e-01 -2.98321068e-01 -4.76760268e-01
-1.91113159e-01 1.24506569e+00 -9.62508142e-01 7.64681697e-01
9.52691376e-01 2.84643233e-01 -3.71881098e-01 1.10978973e+00
-1.71404138e-01 5.25350809e-01 -2.68551588e-01 -2.00662926e-01
-2.30014056e-01 3.00181329e-01 2.73184806e-01 6.11637175e-01
4.24973965e-01 -4.72317129e-01 5.82184374e-01 5.20836115e-01
-5.28843291e-02 -2.32078865e-01 -2.81898648e-01 1.52490303e-01
4.58703309e-01 9.01088536e-01 -8.72580051e-01 -5.10725200e-01
-3.48554045e-01 8.23258400e-01 7.74495974e-02 4.92298678e-02
-7.63217032e-01 -7.12937355e-01 6.27925038e-01 3.09078544e-01
1.60094306e-01 -7.34776780e-02 -3.22006941e-01 -6.92555308e-01
2.16437384e-01 -1.06566250e+00 4.02220309e-01 -3.08137983e-01
-1.15444505e+00 5.84024489e-01 -1.21478438e-01 -1.13268769e+00
-6.05735481e-01 -5.07884264e-01 -3.57946277e-01 7.26920485e-01
-1.26320934e+00 -2.79902190e-01 -2.37003192e-01 4.24392015e-01
3.98458868e-01 -8.71113241e-02 8.45307469e-01 5.35259962e-01
-7.39829957e-01 5.32683253e-01 2.79629469e-01 -1.84456095e-01
7.18210220e-01 -1.20308769e+00 1.26089349e-01 9.63934779e-01
-1.90692350e-01 1.10039048e-01 8.09781194e-01 -4.44503158e-01
-1.34257817e+00 -1.04467463e+00 4.18227196e-01 9.33869705e-02
3.45600009e-01 -1.13787707e-02 -1.05015731e+00 1.31679952e-01
4.94907573e-02 2.12293938e-01 1.68798968e-01 5.36640808e-02
2.55130716e-02 -7.44769871e-01 -1.61340117e+00 2.94687122e-01
5.09395003e-01 -1.20994493e-01 -3.29853892e-01 2.41278052e-01
4.46215689e-01 -3.29741538e-01 -1.25984001e+00 7.78945327e-01
1.83879778e-01 -1.01819181e+00 4.37714279e-01 -1.16545647e-01
1.99979365e-01 -3.23520839e-01 1.12652434e-02 -1.46552157e+00
-3.48341852e-01 -1.85711801e-01 -1.31921202e-01 1.01410294e+00
4.20367002e-01 -1.04480422e+00 8.82566631e-01 3.25834721e-01
-2.94768453e-01 -1.10091054e+00 -1.00127769e+00 -1.00631011e+00
-4.96775843e-02 -2.16671422e-01 6.40026510e-01 8.76679540e-01
1.70181170e-01 -2.40143370e-02 -9.57499370e-02 5.40249720e-02
4.45806593e-01 1.81966841e-01 4.15787399e-01 -1.00446105e+00
-5.99667668e-01 -4.65070307e-01 -4.33027983e-01 -3.05896878e-01
-3.27318311e-01 -3.17352176e-01 1.36537120e-01 -1.51520371e+00
-1.16896130e-01 -5.47216058e-01 -8.03356767e-01 7.47232497e-01
-2.56599963e-01 -3.54712340e-03 7.77344853e-02 1.36904925e-01
-3.95013511e-01 3.09734970e-01 9.44168031e-01 2.64436994e-02
8.71873796e-02 -9.12469849e-02 -5.07576644e-01 4.65746790e-01
9.07464027e-01 -5.04665673e-01 -3.15453708e-01 -3.86405557e-01
1.04436547e-01 3.10798019e-01 1.94626227e-02 -1.44285941e+00
1.14758439e-01 -1.85001403e-01 3.62943798e-01 -3.11180085e-01
-1.64843220e-02 -7.99003839e-01 6.68649226e-02 1.27373719e+00
2.18632236e-01 4.06666189e-01 2.59476393e-01 3.11236709e-01
-4.61220324e-01 -2.11561859e-01 8.53257954e-01 -1.49933606e-01
-8.75748873e-01 -1.47725567e-01 -1.81857482e-01 -2.52268136e-01
1.49048996e+00 -5.96658438e-02 -4.42288220e-01 -1.02956027e-01
-3.82465065e-01 3.60619783e-01 5.44186234e-01 5.21974325e-01
5.74041903e-01 -1.05540884e+00 -4.91363794e-01 4.04942244e-01
-8.43965337e-02 -2.50701774e-02 1.39557138e-01 7.21846998e-01
-5.35825908e-01 2.86700487e-01 -4.89310205e-01 -4.60951239e-01
-1.04179871e+00 6.73733234e-01 2.34734714e-01 -5.57711184e-01
-1.05537303e-01 6.28343105e-01 -5.82083285e-01 -3.86512242e-02
7.57517964e-02 -3.40223700e-01 9.59711820e-02 -3.30241770e-01
2.29408175e-01 9.08826292e-01 6.51251793e-01 1.62378818e-01
-5.83631337e-01 2.34995887e-01 -1.49843125e-02 1.38780996e-01
1.26216173e+00 2.65769958e-01 -3.13362628e-01 2.95218766e-01
7.74786532e-01 -3.23648065e-01 -1.19916546e+00 6.05620146e-02
3.32406610e-01 -4.76803005e-01 4.45025489e-02 -1.00657094e+00
-1.30219352e+00 6.30766273e-01 1.19710422e+00 6.78845525e-01
1.27187312e+00 -4.00889903e-01 6.33751631e-01 1.54814988e-01
8.78325760e-01 -1.22640872e+00 1.31173655e-01 4.16340262e-01
6.37802660e-01 -1.08774590e+00 8.06838274e-02 -2.35256612e-01
-2.42809594e-01 9.99357760e-01 8.20338607e-01 -3.82851750e-01
4.87453073e-01 4.12531972e-01 2.25332435e-02 -5.91634475e-02
-9.10764694e-01 9.75836590e-02 -1.28974736e-01 2.75386602e-01
2.71415353e-01 1.75253078e-01 -7.10367978e-01 3.27042282e-01
2.49034658e-01 1.36229992e-01 3.63840997e-01 1.41748798e+00
-5.54780960e-01 -1.64381814e+00 -5.16117215e-01 7.99007654e-01
-6.42707825e-01 3.84315193e-01 -1.95786878e-02 7.56434262e-01
3.40798378e-01 1.12008357e+00 9.37701091e-02 -5.55401087e-01
6.45167768e-01 1.40496656e-01 2.89758474e-01 -4.54855502e-01
-4.28746045e-01 -9.60269347e-02 -7.30216829e-03 -8.13739300e-01
-2.21389569e-02 -4.56579506e-01 -1.53467405e+00 -2.87539631e-01
-4.15023118e-01 5.57210207e-01 8.03082347e-01 9.91622329e-01
5.99064291e-01 1.04157901e+00 8.29124570e-01 -3.55222344e-01
-8.19144249e-01 -9.35872853e-01 -6.31756604e-01 4.34361935e-01
-8.40040203e-03 -9.08145607e-01 -2.84182280e-01 -3.76957625e-01] | [6.98496675491333, 2.444140672683716] |
c5483975-84ed-4f7f-970a-4098def0db82 | constructing-a-meta-learner-for-unsupervised | 2304.11438 | null | https://arxiv.org/abs/2304.11438v1 | https://arxiv.org/pdf/2304.11438v1.pdf | Constructing a meta-learner for unsupervised anomaly detection | Unsupervised anomaly detection (AD) is critical for a wide range of practical applications, from network security to health and medical tools. Due to the diversity of problems, no single algorithm has been found to be superior for all AD tasks. Choosing an algorithm, otherwise known as the Algorithm Selection Problem (ASP), has been extensively examined in supervised classification problems, through the use of meta-learning and AutoML, however, it has received little attention in unsupervised AD tasks. This research proposes a new meta-learning approach that identifies an appropriate unsupervised AD algorithm given a set of meta-features generated from the unlabelled input dataset. The performance of the proposed meta-learner is superior to the current state of the art solution. In addition, a mixed model statistical analysis has been conducted to examine the impact of the meta-learner components: the meta-model, meta-features, and the base set of AD algorithms, on the overall performance of the meta-learner. The analysis was conducted using more than 10,000 datasets, which is significantly larger than previous studies. Results indicate that a relatively small number of meta-features can be used to identify an appropriate AD algorithm, but the choice of a meta-model in the meta-learner has a considerable impact. | ['Andrew McCarren', 'Suzanne Little', 'Małgorzata Gutowska'] | 2023-04-22 | null | null | null | null | ['automl'] | ['methodology'] | [ 2.18831211e-01 -2.53188372e-01 -1.72079965e-01 -2.22347364e-01
-5.13170660e-01 3.18136154e-04 6.87949061e-01 6.26673341e-01
-5.73344707e-01 4.35146213e-01 -1.97470903e-01 -2.55056441e-01
-5.75047314e-01 -7.07113862e-01 -3.27514499e-01 -7.02765167e-01
-1.16455041e-01 3.20452482e-01 1.35213539e-01 1.96580708e-01
6.69743061e-01 4.64441240e-01 -2.26796484e+00 4.52930927e-01
1.07427192e+00 9.63005245e-01 -1.81368168e-03 5.62489033e-01
-4.61985648e-01 7.94893861e-01 -6.40959382e-01 -2.71131128e-01
2.59631902e-01 -7.10042357e-01 -7.42859244e-01 4.18324284e-02
3.73932272e-01 5.21750841e-03 1.83831915e-01 8.51107895e-01
5.45361698e-01 1.66051939e-01 9.26654935e-01 -1.44818711e+00
1.22796774e-01 4.18019950e-01 -2.74316370e-01 4.81849939e-01
1.14112310e-01 -7.45668709e-02 1.01550293e+00 -7.05591500e-01
3.51443142e-01 1.00181067e+00 6.28519058e-01 3.39573145e-01
-1.01110411e+00 -7.03496993e-01 -6.91179419e-03 3.79107028e-01
-1.30001819e+00 -1.68697104e-01 8.57548833e-01 -5.30928671e-01
8.61706913e-01 2.37472594e-01 6.14838004e-01 7.34446526e-01
5.33925593e-01 5.99409044e-01 1.17121804e+00 -9.42131281e-01
4.72397327e-01 4.11660969e-01 3.68464977e-01 5.55816591e-01
6.36305928e-01 9.41910967e-03 -3.83942962e-01 -5.94544411e-01
2.05397546e-01 6.63830265e-02 3.89424324e-01 -3.83525312e-01
-8.09773326e-01 9.64132667e-01 -6.13146611e-02 5.13560355e-01
-4.98816997e-01 -3.48522007e-01 6.14066243e-01 5.69854617e-01
5.16452134e-01 8.72547746e-01 -5.04605234e-01 -1.32949069e-01
-7.42625117e-01 1.99154884e-01 6.98806405e-01 3.59469354e-01
6.01818442e-01 1.30467657e-02 -3.82127054e-02 9.88789439e-01
3.85291666e-01 5.54847904e-02 7.97545910e-01 -5.20164847e-01
3.49158436e-01 1.16858709e+00 -2.65479475e-01 -9.89655375e-01
-3.77653331e-01 -4.34233367e-01 -5.27101636e-01 2.71987885e-01
3.02062958e-01 -2.96404392e-01 -8.88111889e-01 1.35298777e+00
3.45439047e-01 1.35343462e-01 -2.05959398e-02 4.39956665e-01
6.82659864e-01 6.00040972e-01 3.00315589e-01 -4.85398233e-01
9.00010645e-01 -7.67156780e-01 -5.31392336e-01 -3.52461368e-01
9.63011742e-01 -6.77407920e-01 6.41493082e-01 7.69070506e-01
-7.76130915e-01 -5.03412008e-01 -1.23891521e+00 6.95605099e-01
-5.65455973e-01 -4.67918031e-02 5.18662751e-01 6.34796977e-01
-5.78052223e-01 4.83661026e-01 -6.15391552e-01 -6.83814704e-01
3.50172371e-01 5.26326537e-01 -2.61573374e-01 -5.29007129e-02
-9.41518784e-01 1.09841168e+00 7.43480980e-01 -2.20561281e-01
-6.97592974e-01 -4.34391856e-01 -5.40417612e-01 -2.90335119e-01
3.43694657e-01 -4.83187854e-01 1.04926515e+00 -1.42873979e+00
-1.07616532e+00 5.67499459e-01 1.06408201e-01 -5.41914105e-01
4.71482575e-01 -1.93822965e-01 -8.07441831e-01 3.31233325e-03
-1.21571928e-01 1.49518266e-01 1.04775620e+00 -1.10210931e+00
-1.14056504e+00 -4.86708492e-01 -2.65315533e-01 3.53694081e-01
-4.83566284e-01 3.70008126e-02 1.88375711e-02 -6.69075251e-01
1.34769350e-01 -8.80546510e-01 -2.73096502e-01 -3.95184278e-01
1.14931324e-02 -2.42797747e-01 9.42857921e-01 -1.15821555e-01
1.75214171e+00 -2.27204895e+00 -2.56498344e-02 5.73920012e-01
4.10756841e-02 4.24677312e-01 8.80770907e-02 4.31749284e-01
-2.36746684e-01 1.34561017e-01 -4.42480832e-01 1.01639882e-01
-5.05089104e-01 1.36877447e-01 2.24862501e-01 4.67308909e-01
4.18323092e-02 9.28160474e-02 -7.11680174e-01 -5.31607926e-01
4.99085307e-01 6.99754804e-02 -4.32655185e-01 4.62640822e-01
1.89282489e-03 5.90979345e-02 -6.53621733e-01 5.80639839e-01
2.80939281e-01 -4.76044789e-02 1.94609482e-02 8.77534226e-02
-8.55526552e-02 -2.00691279e-02 -1.39962757e+00 1.23628461e+00
-3.16460162e-01 5.46707928e-01 -4.43759590e-01 -1.17729104e+00
1.07048333e+00 3.96894544e-01 9.09272671e-01 -5.65804601e-01
2.08665833e-01 5.29356062e-01 6.27235115e-01 -5.27693808e-01
2.70058304e-01 -4.46163304e-02 2.14869902e-01 6.52880132e-01
1.17885575e-01 2.87160248e-01 3.31749618e-01 -8.56790990e-02
1.21298575e+00 -2.47305527e-01 6.65625393e-01 -2.92087972e-01
7.66092479e-01 3.19358259e-01 4.61009562e-01 9.12852764e-01
-1.62418902e-01 4.45946336e-01 9.94011164e-02 -5.76490402e-01
-7.42163241e-01 -7.79575765e-01 -3.51213008e-01 1.09043837e+00
-1.14482082e-01 -3.12622428e-01 -7.55653083e-01 -8.57007086e-01
9.56779271e-02 8.83657694e-01 -5.66455245e-01 -5.35659909e-01
-3.74038637e-01 -1.23432314e+00 5.14269173e-01 1.90384418e-01
5.11442065e-01 -1.31174099e+00 -1.03666842e+00 3.29815805e-01
2.35195369e-01 -6.17855370e-01 7.73285851e-02 5.09752989e-01
-1.38586187e+00 -1.41366613e+00 -1.80529624e-01 -5.63649297e-01
8.05679381e-01 2.86530316e-01 1.04566073e+00 3.22031975e-01
-4.63464439e-01 4.89442915e-01 -6.96309209e-01 -7.71637022e-01
-6.29591286e-01 1.85945079e-01 1.62024468e-01 2.32072428e-01
7.06483841e-01 -1.78912222e-01 -1.07831798e-01 1.03755571e-01
-9.72927451e-01 -4.01231945e-01 9.77766693e-01 7.63653696e-01
3.82282376e-01 7.68979013e-01 6.89450264e-01 -1.35111213e+00
6.99154079e-01 -7.93219388e-01 -1.93872780e-01 3.11024666e-01
-1.19672287e+00 1.82309061e-01 4.68982100e-01 -4.66416240e-01
-9.31286812e-01 -1.32917553e-01 5.27425110e-02 -2.18116954e-01
-4.16091502e-01 6.84567928e-01 -1.88446835e-01 -1.00148350e-01
7.56412268e-01 -4.89903614e-02 2.59912103e-01 -4.45632815e-01
-1.84779212e-01 8.08021188e-01 2.51368005e-02 -4.86195207e-01
4.80952799e-01 9.65468003e-04 -2.44744606e-02 -1.11089730e+00
-7.58202493e-01 -6.50961101e-01 -7.46593595e-01 -2.39984050e-01
6.18402064e-01 -4.73099053e-01 1.03212893e-01 6.21810496e-01
-5.87633193e-01 -1.38283670e-02 -8.76031909e-03 7.49889791e-01
-2.57943630e-01 1.46281451e-01 4.02802117e-02 -8.71811390e-01
-3.87353778e-01 -1.23202455e+00 1.74707532e-01 1.10191405e-01
-5.81100821e-01 -1.02327681e+00 1.44001946e-01 2.94003218e-01
2.15511397e-01 3.47087950e-01 1.25163758e+00 -1.32334995e+00
-6.32302985e-02 -3.96409959e-01 2.33688548e-01 3.65467757e-01
4.93254274e-01 2.10125729e-01 -9.36528683e-01 -2.29107365e-01
7.88797364e-02 1.88755002e-02 6.81240976e-01 4.17178690e-01
1.23712707e+00 -2.62209386e-01 -2.81412601e-01 2.89585590e-01
1.37470543e+00 6.55805647e-01 4.69965219e-01 9.69834864e-01
5.47996163e-01 4.93130088e-01 1.05296683e+00 4.57570195e-01
-3.32583562e-02 2.69818991e-01 3.84794742e-01 1.50240690e-01
3.28128219e-01 1.64090231e-01 3.63834590e-01 5.59924304e-01
8.14037845e-02 -1.62027836e-01 -1.11416543e+00 3.29714626e-01
-1.83377862e+00 -9.98366714e-01 7.10098296e-02 2.42499328e+00
3.65378618e-01 5.37802100e-01 4.59868789e-01 8.57670844e-01
6.62359357e-01 -1.47033602e-01 -4.57606375e-01 -8.69111776e-01
2.47162864e-01 1.05396263e-01 3.28264564e-01 6.34831190e-02
-1.19971156e+00 3.00779730e-01 6.69153786e+00 6.77967429e-01
-1.08268774e+00 -2.16512963e-01 5.23395777e-01 2.43791610e-01
-3.59279253e-02 -1.09515101e-01 -6.29113317e-01 5.67590058e-01
1.20306313e+00 -9.07422379e-02 4.61630560e-02 8.58632624e-01
1.22009739e-01 -3.01597863e-01 -1.11947477e+00 8.30182195e-01
3.10027242e-01 -8.34687054e-01 3.81031871e-01 1.66709080e-01
7.28724003e-01 -2.88898885e-01 -8.84532705e-02 2.46293649e-01
-3.41744013e-02 -8.62726688e-01 4.57291603e-01 3.54220390e-01
1.86663061e-01 -1.15916455e+00 1.03428721e+00 4.55238193e-01
-7.71463990e-01 -6.14058018e-01 5.73109649e-03 -6.45186752e-02
-5.02014339e-01 5.20214260e-01 -8.56592000e-01 5.21012306e-01
6.14446163e-01 4.96127337e-01 -9.21678722e-01 1.30941319e+00
3.67318213e-01 8.30029666e-01 -3.76670435e-02 -4.37832028e-02
1.88270003e-01 -3.29603045e-03 5.44860721e-01 1.12960958e+00
2.71652699e-01 -4.56636921e-02 1.28036708e-01 2.21981779e-01
3.22261274e-01 5.60840845e-01 -7.50532031e-01 -2.50611395e-01
6.10775530e-01 9.72100437e-01 -8.11835587e-01 -8.40523392e-02
-6.44389391e-01 5.17968714e-01 1.26344621e-01 1.25225959e-02
-4.09955472e-01 -3.17881167e-01 4.17878181e-01 2.66019136e-01
-8.67434442e-02 1.88133955e-01 -5.40225685e-01 -6.13158226e-01
-1.45900592e-01 -1.34864604e+00 9.93185759e-01 -2.29312584e-01
-1.30057132e+00 5.20102143e-01 2.58662403e-01 -1.53918600e+00
-4.48439628e-01 -5.63353539e-01 -8.63759875e-01 5.15344739e-01
-9.14388776e-01 -7.92286873e-01 -2.88661838e-01 2.18299553e-01
6.53239906e-01 -8.54562402e-01 8.63257468e-01 2.76918441e-01
-7.34759927e-01 5.90451181e-01 1.89975306e-01 2.15988886e-02
7.30690479e-01 -1.19332290e+00 -2.54001856e-01 9.39429402e-01
1.53302982e-01 5.61091244e-01 6.26044929e-01 -7.10847855e-01
-1.27280486e+00 -1.02576983e+00 5.63652873e-01 -4.20547038e-01
5.82661927e-02 3.14520985e-01 -1.02614152e+00 3.34326893e-01
-1.40181765e-01 -9.51154977e-02 9.93210912e-01 2.67725468e-01
-6.40230253e-02 -2.99085587e-01 -1.45845306e+00 3.39388281e-01
4.16501045e-01 -1.45220220e-01 -7.02832639e-01 -3.55044007e-01
5.12718670e-02 -2.18229130e-01 -8.66741896e-01 7.16444612e-01
6.60677373e-01 -1.05798256e+00 8.57947588e-01 -5.48580647e-01
3.81350666e-01 -1.59638330e-01 -9.95988958e-03 -1.33046651e+00
-3.91086340e-01 9.36203152e-02 -4.11294699e-01 1.30237198e+00
5.35893142e-01 -6.26564443e-01 7.70735443e-01 7.05647409e-01
3.83481495e-02 -9.98448908e-01 -7.21310794e-01 -5.64563692e-01
-1.53014347e-01 -5.64031124e-01 6.21888518e-01 1.03959441e+00
-3.80853981e-01 1.93453982e-01 -1.10480390e-01 7.43095800e-02
6.39556944e-01 -1.92658991e-01 7.79786825e-01 -1.72396564e+00
1.61412768e-02 -6.37149990e-01 -6.70367599e-01 9.63565111e-02
-3.13651115e-02 -9.04311001e-01 -1.45026371e-01 -1.39199913e+00
1.37378633e-01 -6.99740827e-01 -7.50385523e-01 3.08839709e-01
-3.47716838e-01 -2.16204673e-01 -2.19198018e-02 3.23381454e-01
-3.19388777e-01 2.04845712e-01 6.30197406e-01 -7.51638785e-02
-6.81138992e-01 3.50934297e-01 -6.75795197e-01 9.44655716e-01
8.97683024e-01 -7.97148705e-01 -6.23602748e-01 -1.83191728e-02
-1.17676958e-01 -2.27217957e-01 -2.52082143e-02 -1.12876773e+00
2.15029657e-01 -2.58624971e-01 5.06506860e-01 -5.05527914e-01
-1.83718950e-01 -1.10149503e+00 1.00360841e-01 6.35194421e-01
-5.21631420e-01 5.04658818e-01 3.08500767e-01 5.77209115e-01
-1.87690496e-01 -8.31313252e-01 8.17449749e-01 -1.41852394e-01
-1.05515671e+00 1.51911020e-01 -5.20646989e-01 2.51614183e-01
1.24361157e+00 -7.16436803e-01 3.40105712e-01 1.16829753e-01
-3.83190602e-01 -2.39541996e-02 3.29522848e-01 6.22702658e-01
5.83586454e-01 -1.12988687e+00 -5.66745818e-01 3.73295605e-01
4.02990788e-01 -4.01123315e-02 -6.67595863e-02 5.19769728e-01
-5.39849341e-01 1.44596413e-01 -4.19398367e-01 -6.22319698e-01
-1.31459761e+00 2.70966768e-01 1.75724968e-01 -1.20783210e-01
-5.69533885e-01 3.62401783e-01 -5.36624014e-01 -2.90489703e-01
4.21877563e-01 9.49831754e-02 -4.60741848e-01 3.42250586e-01
4.81758207e-01 8.93292844e-01 3.51663381e-01 -5.17574251e-01
-3.13163131e-01 2.37784281e-01 -3.21233273e-01 8.59332364e-03
1.21711409e+00 8.75788257e-02 -7.88825601e-02 8.05430055e-01
8.60759258e-01 -3.60031486e-01 -5.42906523e-01 -7.47453272e-02
5.54935455e-01 -4.96800512e-01 2.35238954e-01 -6.88158095e-01
-8.05262744e-01 5.02203703e-01 1.00369525e+00 4.41722907e-02
1.22978461e+00 -5.55732667e-01 2.42456675e-01 4.47396249e-01
3.16331893e-01 -1.42178214e+00 2.34233975e-01 2.85665691e-01
2.72230685e-01 -1.55367243e+00 2.80065626e-01 -1.61018744e-02
-6.51641726e-01 1.16906583e+00 9.68153298e-01 -2.99627706e-02
8.76929581e-01 5.36978990e-02 2.41003428e-02 -1.55739427e-01
-7.92780876e-01 1.00434996e-01 5.21517634e-01 3.10304999e-01
6.31349087e-01 -1.27395093e-01 -4.92185980e-01 4.44403052e-01
1.80546120e-01 -1.31880855e-02 3.35219860e-01 1.25615954e+00
-5.78855038e-01 -1.26630414e+00 -4.07534570e-01 1.23409617e+00
-7.25839019e-01 2.58482128e-01 -4.78278548e-01 8.72988641e-01
3.80977958e-01 9.99843836e-01 1.25930458e-01 -5.15132308e-01
2.35428542e-01 3.49946827e-01 2.13168025e-01 -8.22904527e-01
-8.28613281e-01 -3.86030436e-01 -2.29933430e-02 -1.13264829e-01
-6.42400086e-01 -7.52213299e-01 -1.10024047e+00 -7.35945255e-02
-5.29242814e-01 4.67492759e-01 4.66037184e-01 1.32976317e+00
1.51081070e-01 3.78960192e-01 8.05151761e-01 -2.80667335e-01
-4.84637648e-01 -9.01965916e-01 -4.69736874e-01 3.03276747e-01
1.14797749e-01 -9.60910857e-01 -5.66199601e-01 -3.62984866e-01] | [8.305335998535156, 4.232251167297363] |
8ecf5a20-d03d-4d29-98d4-80ba27572985 | decoupling-visual-semantic-feature-learning | 2111.12351 | null | https://arxiv.org/abs/2111.12351v1 | https://arxiv.org/pdf/2111.12351v1.pdf | Decoupling Visual-Semantic Feature Learning for Robust Scene Text Recognition | Semantic information has been proved effective in scene text recognition. Most existing methods tend to couple both visual and semantic information in an attention-based decoder. As a result, the learning of semantic features is prone to have a bias on the limited vocabulary of the training set, which is called vocabulary reliance. In this paper, we propose a novel Visual-Semantic Decoupling Network (VSDN) to address the problem. Our VSDN contains a Visual Decoder (VD) and a Semantic Decoder (SD) to learn purer visual and semantic feature representation respectively. Besides, a Semantic Encoder (SE) is designed to match SD, which can be pre-trained together by additional inexpensive large vocabulary via a simple word correction task. Thus the semantic feature is more unbiased and precise to guide the visual feature alignment and enrich the final character representation. Experiments show that our method achieves state-of-the-art or competitive results on the standard benchmarks, and outperforms the popular baseline by a large margin under circumstances where the training set has a small size of vocabulary. | ['Wenyu Liu', 'Yongpan Wang', 'Qi Zheng', 'Bohan Li', 'Changxu Cheng'] | 2021-11-24 | null | null | null | null | ['scene-text-recognition'] | ['computer-vision'] | [ 2.58439898e-01 -3.10612798e-01 -2.19112337e-01 -5.47064424e-01
-5.32919765e-01 -2.05185592e-01 7.78025210e-01 -4.07134136e-03
-5.05659163e-01 3.31814110e-01 4.07070458e-01 4.43123542e-02
4.89854008e-01 -6.78978503e-01 -7.50228941e-01 -7.06309199e-01
9.41273391e-01 3.36662233e-01 4.22379225e-01 -2.07560688e-01
4.94298995e-01 -4.63071018e-02 -1.34075987e+00 3.20183963e-01
8.24577093e-01 1.05871749e+00 9.34941769e-01 3.00628155e-01
-5.55512547e-01 7.12284625e-01 -4.72175270e-01 -3.70267957e-01
7.91687295e-02 -5.29784143e-01 -6.45621657e-01 1.65623114e-01
2.92580098e-01 -2.38536671e-01 -5.06134391e-01 1.25612950e+00
6.67345941e-01 1.23417780e-01 7.68124104e-01 -1.00579083e+00
-9.63552654e-01 5.66937625e-01 -6.46855116e-01 4.89443466e-02
1.99541196e-01 3.92982252e-02 1.09383237e+00 -1.02868986e+00
5.40280819e-01 1.28846359e+00 3.77478451e-01 6.38873935e-01
-8.90118897e-01 -6.56993985e-01 4.05009508e-01 3.92199278e-01
-1.66963553e+00 -5.60947597e-01 8.88447583e-01 -3.66148651e-01
7.86455452e-01 5.63459620e-02 5.77430844e-01 1.25110316e+00
-1.57820344e-01 9.66275692e-01 6.04465365e-01 -4.65503722e-01
9.27589089e-02 3.67223352e-01 1.61510929e-01 4.86688375e-01
2.39910200e-01 -3.74590993e-01 -4.65738952e-01 4.01905924e-01
8.13619137e-01 9.31352228e-02 -3.61356318e-01 -5.34335852e-01
-1.04995549e+00 8.13753426e-01 5.30246437e-01 2.09181860e-01
-7.11020976e-02 1.51893705e-01 6.15325570e-01 1.68720812e-01
2.85681069e-01 1.95913851e-01 -2.00743407e-01 6.20180964e-02
-8.36396873e-01 -1.19711608e-01 4.61599320e-01 1.19127536e+00
6.76891565e-01 1.23225361e-01 -4.37638044e-01 1.00743127e+00
5.96119046e-01 6.86456144e-01 6.81725919e-01 -2.50889003e-01
7.86786020e-01 7.34701514e-01 -1.28302574e-01 -1.06013536e+00
-1.09913267e-01 -3.54335755e-01 -7.78531432e-01 -2.04743713e-01
2.24674046e-01 2.92982846e-01 -9.92340386e-01 1.72580528e+00
1.89648628e-01 2.05183581e-01 1.30357638e-01 1.17223382e+00
1.01082218e+00 8.03837538e-01 2.16582492e-01 2.72135325e-02
1.36499560e+00 -1.09661841e+00 -6.67345703e-01 -5.38097739e-01
5.58180809e-01 -7.66352236e-01 1.34609640e+00 6.90791234e-02
-7.05363929e-01 -5.84402621e-01 -1.27911878e+00 -4.29616451e-01
-3.03409845e-01 2.32129112e-01 1.30986750e-01 4.14201885e-01
-6.12319946e-01 1.98459685e-01 -5.28497875e-01 -6.01308942e-01
5.96327424e-01 1.19762294e-01 -3.54720056e-01 -2.53535479e-01
-1.12251914e+00 8.82329643e-01 5.59142411e-01 1.40317865e-02
-1.02311146e+00 -3.51559550e-01 -9.46274042e-01 6.51405156e-02
4.32397604e-01 -7.03649580e-01 9.49994981e-01 -1.52954447e+00
-1.38243639e+00 8.69621038e-01 -2.21802965e-01 -3.13504823e-02
4.14341748e-01 -1.36575893e-01 -3.35451335e-01 1.04301043e-01
2.40040869e-01 5.97431362e-01 1.05015099e+00 -1.29110074e+00
-5.39762735e-01 -3.61547470e-01 -7.76519924e-02 6.21063888e-01
-6.50189996e-01 9.49032679e-02 -9.13449943e-01 -8.81178796e-01
-9.62942280e-03 -4.35773671e-01 6.97895586e-02 1.53405592e-01
-4.27576303e-01 -2.38759816e-01 7.20439136e-01 -7.36319959e-01
1.29156733e+00 -2.34884739e+00 3.79953802e-01 8.61021951e-02
2.05706626e-01 4.66763794e-01 -2.20865354e-01 3.55238736e-01
8.17323700e-02 -1.21306561e-01 -1.27358556e-01 -4.07225221e-01
-9.52923447e-02 1.61148101e-01 -3.45740587e-01 4.91338253e-01
1.17373966e-01 9.80919778e-01 -8.88251424e-01 -7.46367872e-01
9.12099406e-02 3.20056677e-01 -5.38775444e-01 3.23560148e-01
-2.46874765e-01 2.09575832e-01 -7.89178669e-01 4.10922915e-01
5.98561764e-01 -5.93954265e-01 1.95761874e-01 -2.44722143e-01
4.77918684e-02 3.08643520e-01 -1.21021497e+00 1.99877930e+00
-3.62773627e-01 6.88343227e-01 -2.99644500e-01 -1.38975775e+00
1.09636855e+00 1.58496112e-01 -1.49764553e-01 -1.00812352e+00
4.29738909e-01 1.21107012e-01 -2.62474835e-01 -5.54204047e-01
3.27289015e-01 -1.24101758e-01 -1.39783006e-02 2.67437160e-01
1.31964669e-01 1.47348955e-01 -1.15942322e-01 2.32146487e-01
6.73223197e-01 1.19866967e-01 4.33784217e-01 -3.49065542e-01
8.53031635e-01 -8.20658132e-02 5.76416612e-01 5.33769727e-01
-7.36117736e-02 7.70226657e-01 4.46467280e-01 -2.20457628e-01
-1.39514863e+00 -8.37550402e-01 -5.35268560e-02 1.11466098e+00
7.49604881e-01 -4.85197216e-01 -7.58970737e-01 -7.62237549e-01
-1.17989555e-01 6.38781965e-01 -5.57708621e-01 -3.78056943e-01
-4.38095361e-01 -4.07240570e-01 4.59748000e-01 8.49329472e-01
7.45102465e-01 -9.36897635e-01 -1.31877795e-01 4.73017171e-02
-1.37549967e-01 -1.07426500e+00 -7.67686844e-01 7.80934468e-03
-5.61608374e-01 -9.71109807e-01 -8.61096442e-01 -1.12435007e+00
8.14514279e-01 5.77716827e-01 7.74875998e-01 1.49083778e-01
2.28925701e-02 -5.86284965e-04 -7.03604102e-01 -3.36383432e-01
-1.58986688e-01 9.82887298e-03 -1.99422881e-01 2.62277365e-01
6.96666718e-01 -1.70988277e-01 -5.67928314e-01 3.00133407e-01
-7.83043981e-01 4.55176115e-01 4.70140964e-01 1.09699667e+00
4.92169321e-01 -3.09540182e-01 4.43605363e-01 -7.02716708e-01
3.09206218e-01 -3.28028411e-01 -5.51438391e-01 4.47478235e-01
-5.29070795e-01 2.62591034e-01 8.86939228e-01 -3.45455945e-01
-1.03473306e+00 5.81557527e-02 -2.62530483e-02 -4.83424723e-01
-1.22813331e-02 3.73974442e-01 -6.65401340e-01 3.96918692e-02
3.32453787e-01 6.58798218e-01 -5.66372536e-02 -6.94903791e-01
3.64713758e-01 1.13593459e+00 2.72192806e-01 -3.15939575e-01
6.93094492e-01 2.91973740e-01 -3.39998960e-01 -7.91190505e-01
-8.72967422e-01 -4.90489542e-01 -6.31580055e-01 2.71533113e-02
8.91711891e-01 -1.24083555e+00 -2.14956641e-01 6.38007581e-01
-1.18079400e+00 -1.59545720e-01 -1.69605669e-02 4.29289252e-01
-3.17200214e-01 6.11816466e-01 -3.28730792e-01 -3.63140553e-01
-3.22499186e-01 -1.26858366e+00 1.22103775e+00 2.73959339e-01
1.11548662e-01 -8.90594482e-01 -9.05828327e-02 2.50765920e-01
1.77581161e-01 -2.99840480e-01 8.06847334e-01 -7.91957438e-01
-5.30376256e-01 -5.16190417e-02 -8.23442101e-01 5.05048156e-01
3.91783938e-02 -2.00175092e-01 -1.07673144e+00 -2.93687224e-01
-3.05635370e-02 -3.38635594e-01 1.05602431e+00 1.75023362e-01
1.17835772e+00 -5.10872900e-02 -5.09320736e-01 7.86620319e-01
1.65043092e+00 1.25922933e-01 7.68810511e-01 4.49348181e-01
1.21373093e+00 3.24752957e-01 6.20143592e-01 3.73266935e-01
5.23736477e-01 8.82886112e-01 2.29241461e-01 1.86163448e-02
-3.90681654e-01 -4.91363078e-01 3.23046863e-01 1.04340637e+00
2.28104070e-01 -5.16770780e-01 -9.27913547e-01 4.73019749e-01
-1.91390097e+00 -7.62854934e-01 1.75506424e-03 2.14121795e+00
9.29128468e-01 -1.04966030e-01 -1.08856410e-01 1.96264181e-02
1.02959776e+00 2.82931060e-01 -5.17951787e-01 -8.00202116e-02
-2.86791086e-01 -1.26458690e-01 5.72389245e-01 3.26671243e-01
-1.04757261e+00 1.20482838e+00 5.96516657e+00 1.28720725e+00
-1.19550645e+00 1.77107453e-01 3.32679451e-01 5.54047897e-02
-4.26937819e-01 7.81799294e-03 -9.87105787e-01 7.62624919e-01
5.38867295e-01 -1.32642567e-01 2.42615730e-01 8.79175782e-01
3.61102931e-02 1.87338218e-01 -9.46789920e-01 1.30050325e+00
4.85497475e-01 -1.31329095e+00 5.04109561e-01 -1.32902190e-01
4.91693735e-01 -2.17098799e-02 -2.04726219e-01 1.70289770e-01
-5.41862175e-02 -1.00594640e+00 9.02641475e-01 5.45112252e-01
1.27032018e+00 -6.64174497e-01 7.16877460e-01 3.48862618e-01
-1.36432707e+00 -2.69127935e-02 -7.57728159e-01 9.71432030e-02
-7.02568591e-02 2.28197977e-01 -3.57174635e-01 4.63963330e-01
4.31208551e-01 1.16683793e+00 -6.01788998e-01 8.97603750e-01
-3.46506745e-01 3.27974528e-01 6.52411999e-03 -3.37733716e-01
3.51568490e-01 -6.13150336e-02 3.51395965e-01 1.20216763e+00
8.82070288e-02 -1.15093529e-01 2.40956262e-01 6.94224656e-01
-1.98345438e-01 5.12088060e-01 -4.82880741e-01 -8.96395296e-02
4.93058264e-01 1.09933615e+00 -6.25629544e-01 -4.97081399e-01
-8.02913308e-01 1.25333595e+00 4.66002256e-01 1.80076629e-01
-8.84700954e-01 -6.60393357e-01 5.13336897e-01 -9.50443894e-02
6.31485343e-01 6.97299466e-02 -3.72373253e-01 -1.54680109e+00
5.96472397e-02 -8.03731620e-01 1.91590622e-01 -8.72455597e-01
-1.22839570e+00 4.59161103e-01 -3.50367635e-01 -1.39790332e+00
3.78035247e-01 -8.08515191e-01 -5.18265545e-01 9.30666566e-01
-1.66406596e+00 -1.19541430e+00 -5.74420094e-01 8.04841816e-01
8.21667254e-01 -4.33996260e-01 5.73068559e-01 5.12121379e-01
-7.39017606e-01 8.15301776e-01 3.67488921e-01 4.16434109e-01
8.89342844e-01 -9.98261571e-01 2.75683641e-01 9.37176108e-01
4.86514643e-02 3.24711233e-01 4.97282475e-01 -7.18693912e-01
-1.41250360e+00 -1.09690571e+00 9.19925034e-01 -4.45205361e-01
5.46250582e-01 -5.86433172e-01 -9.54935551e-01 6.39264703e-01
1.36297598e-01 -1.38735950e-01 5.58475375e-01 -2.52439290e-01
-5.98864973e-01 -1.26933053e-01 -6.59094512e-01 6.10172749e-01
1.07343161e+00 -7.38071859e-01 -7.88629651e-01 2.25431338e-01
9.22592223e-01 -2.72907645e-01 -4.44110513e-01 5.16834930e-02
4.85586226e-01 -5.28598845e-01 7.52948940e-01 -5.44085145e-01
5.61273158e-01 -4.58783597e-01 -3.93402904e-01 -1.03105235e+00
-3.65261704e-01 -4.75765169e-02 1.86541706e-01 1.39106536e+00
1.78112686e-01 -3.15950722e-01 5.20744443e-01 1.19992919e-01
-2.13194251e-01 -4.50142264e-01 -6.93286538e-01 -7.91492760e-01
-8.11640471e-02 -3.68179500e-01 6.01459146e-01 9.53240275e-01
1.96440262e-03 8.46288145e-01 -5.88694334e-01 -8.25019479e-02
4.01211023e-01 1.24349974e-01 6.72820449e-01 -1.11702538e+00
-5.14647961e-02 -6.08081162e-01 -5.58291316e-01 -1.59142828e+00
2.01095656e-01 -1.07480180e+00 1.36584938e-01 -1.49007058e+00
7.01673567e-01 -3.60140651e-01 -3.83769780e-01 4.09554660e-01
-4.61248785e-01 1.71828806e-01 1.51491001e-01 3.90715480e-01
-8.87908518e-01 9.83892977e-01 1.42096853e+00 -2.27964610e-01
1.27124384e-01 -4.60198462e-01 -9.65575874e-01 5.58666289e-01
6.30054533e-01 -4.08635199e-01 -4.28905874e-01 -6.63172960e-01
1.36413008e-01 -2.81652272e-01 2.65534729e-01 -6.38138354e-01
3.38979334e-01 -1.46872103e-01 5.00976145e-01 -4.00481582e-01
1.60749480e-01 -8.00618052e-01 -3.51325542e-01 1.97101414e-01
-4.34614778e-01 -2.25784600e-01 -8.23423043e-02 7.10346341e-01
-3.80336940e-01 -3.76185834e-01 7.73370922e-01 2.12178454e-02
-1.13713682e+00 3.03981751e-01 1.04030175e-02 2.37298429e-01
9.76153672e-01 -4.49309796e-01 -4.17980790e-01 -2.43257701e-01
-2.13222012e-01 3.71365309e-01 6.61254466e-01 6.70480907e-01
7.47860849e-01 -1.54709530e+00 -6.06950998e-01 3.98146629e-01
5.13064802e-01 -1.28233418e-01 1.92888409e-01 7.20046282e-01
-5.33871055e-01 4.23711509e-01 -2.08540112e-01 -5.65885067e-01
-1.17261481e+00 7.44535089e-01 2.03208089e-01 7.97929615e-02
-7.72129357e-01 9.90823865e-01 6.42274976e-01 -1.49882481e-01
3.55199993e-01 4.22040671e-02 -4.23053980e-01 1.69944093e-01
6.39709830e-01 3.37836966e-02 -2.98783481e-01 -9.55133915e-01
-5.18247962e-01 8.81628573e-01 -9.78473946e-02 1.78916752e-01
1.08903027e+00 -5.44924736e-01 4.93320338e-02 4.61654812e-01
1.40545249e+00 -2.73626000e-01 -1.25085318e+00 -6.21898890e-01
-1.91621915e-01 -5.63682318e-01 1.67613283e-01 -3.75998259e-01
-1.10295868e+00 1.14255333e+00 3.76483887e-01 -2.17669949e-01
1.11602497e+00 -1.24422824e-02 7.97893703e-01 3.37024450e-01
6.88892454e-02 -1.24251688e+00 1.74744487e-01 6.33537173e-01
7.50831664e-01 -1.39824665e+00 -1.12429112e-01 -3.73182446e-01
-1.13786232e+00 1.10706258e+00 7.33658612e-01 -2.68552482e-01
4.73829716e-01 -8.69673043e-02 4.89903763e-02 -8.32069516e-02
-6.12455070e-01 -3.71805429e-01 4.53024209e-01 4.77450639e-01
3.00205022e-01 -1.37512311e-01 -1.98507726e-01 8.57966423e-01
2.01481760e-01 -3.15792829e-01 1.57706663e-01 6.91224456e-01
-7.52519131e-01 -1.13744283e+00 -2.49021649e-02 2.54550815e-01
-2.26267502e-01 -4.39326048e-01 -1.96712762e-01 6.29864573e-01
1.75933167e-02 6.71129584e-01 2.53716052e-01 -4.16048855e-01
1.45138100e-01 5.19586131e-02 3.31816822e-01 -6.78456783e-01
-2.77217567e-01 2.29076236e-01 -6.25979081e-02 -6.12514675e-01
-2.89074153e-01 -4.97748226e-01 -1.17597651e+00 -1.18581705e-01
-5.17554700e-01 -8.48220289e-02 5.01497746e-01 1.20134711e+00
1.69590086e-01 5.52819073e-01 6.28878772e-01 -4.17970389e-01
-4.91388083e-01 -9.32774961e-01 -7.67007351e-01 6.04796171e-01
1.23309635e-01 -7.08596289e-01 -1.14247553e-01 1.23898655e-01] | [11.751325607299805, 2.113184928894043] |
cd60566d-a56d-4e3b-9d24-f202177d2d4c | kinnews-and-kirnews-benchmarking-cross | 2010.12174 | null | https://arxiv.org/abs/2010.12174v1 | https://arxiv.org/pdf/2010.12174v1.pdf | KINNEWS and KIRNEWS: Benchmarking Cross-Lingual Text Classification for Kinyarwanda and Kirundi | Recent progress in text classification has been focused on high-resource languages such as English and Chinese. For low-resource languages, amongst them most African languages, the lack of well-annotated data and effective preprocessing, is hindering the progress and the transfer of successful methods. In this paper, we introduce two news datasets (KINNEWS and KIRNEWS) for multi-class classification of news articles in Kinyarwanda and Kirundi, two low-resource African languages. The two languages are mutually intelligible, but while Kinyarwanda has been studied in Natural Language Processing (NLP) to some extent, this work constitutes the first study on Kirundi. Along with the datasets, we provide statistics, guidelines for preprocessing, and monolingual and cross-lingual baseline models. Our experiments show that training embeddings on the relatively higher-resourced Kinyarwanda yields successful cross-lingual transfer to Kirundi. In addition, the design of the created datasets allows for a wider use in NLP beyond text classification in future studies, such as representation learning, cross-lingual learning with more distant languages, or as base for new annotations for tasks such as parsing, POS tagging, and NER. The datasets, stopwords, and pre-trained embeddings are publicly available at https://github.com/Andrews2017/KINNEWS-and-KIRNEWS-Corpus . | ['Li Huang', 'Julia Kreutzer', 'Hong Qu', 'Rubungo Andre Niyongabo'] | 2020-10-23 | null | https://aclanthology.org/2020.coling-main.480 | https://aclanthology.org/2020.coling-main.480.pdf | coling-2020-8 | ['news-classification'] | ['natural-language-processing'] | [-2.72794992e-01 -1.25680894e-01 -5.34843922e-01 -4.34751779e-01
-1.08847499e+00 -8.94936144e-01 7.23765612e-01 4.53889370e-01
-9.87308085e-01 8.17184567e-01 8.30739200e-01 -7.35510111e-01
2.47145757e-01 -5.54324090e-01 -3.55581284e-01 -3.55954945e-01
1.27914861e-01 6.49255455e-01 -2.15456486e-01 -2.16541648e-01
9.15439576e-02 1.28935799e-01 -9.11717892e-01 3.62704664e-01
8.08441460e-01 3.13863337e-01 3.44115794e-01 5.45802832e-01
-3.36547673e-01 4.54442382e-01 -2.81461209e-01 -6.26723468e-01
5.32081127e-02 -1.71957687e-01 -1.10018337e+00 -5.10188520e-01
4.37005073e-01 -3.60396206e-02 -2.10895151e-01 6.16731703e-01
6.33188188e-01 3.99504714e-02 7.38431811e-01 -6.05060399e-01
-1.16430748e+00 1.20566809e+00 -4.40103024e-01 2.97179043e-01
1.98583156e-01 -2.19288856e-01 1.36949694e+00 -1.08692861e+00
9.89349365e-01 1.19505465e+00 7.85829961e-01 6.21000111e-01
-8.35564137e-01 -7.13011324e-01 1.40466645e-01 4.16308716e-02
-1.10165989e+00 -6.44294500e-01 4.79258716e-01 -3.23892564e-01
1.25976014e+00 -1.08097672e-01 2.44109839e-01 1.31619930e+00
1.01224799e-02 8.97345245e-01 1.22409463e+00 -9.16269183e-01
-2.40830004e-01 3.84088725e-01 3.44444394e-01 5.01770794e-01
4.01204258e-01 -9.74030793e-02 -5.15345633e-01 -4.04016785e-02
3.75341803e-01 -2.83058733e-01 -2.47763410e-01 2.52099216e-01
-1.39516473e+00 1.10991037e+00 8.05274248e-02 8.86179209e-01
-1.04413226e-01 -1.68500349e-01 7.57578373e-01 4.92035657e-01
1.00839889e+00 5.57818949e-01 -1.16120839e+00 -3.93728942e-01
-5.92217207e-01 -1.24454364e-01 9.84033346e-01 7.74844825e-01
4.47575063e-01 1.02927007e-01 3.92121434e-01 1.38209224e+00
3.18168700e-01 4.48410839e-01 8.74355853e-01 -2.37105846e-01
9.67413723e-01 3.43346447e-01 -1.94364369e-01 -5.98745704e-01
-3.78700376e-01 1.05711482e-01 -4.44013268e-01 -2.26102173e-01
6.78428233e-01 -5.95630646e-01 -9.15184796e-01 1.51395130e+00
7.68752843e-02 -3.78487796e-01 7.29006469e-01 5.48689961e-01
9.47960913e-01 1.07941437e+00 3.76405001e-01 -6.86444938e-02
1.76129103e+00 -1.01538169e+00 -5.46093941e-01 -4.63024169e-01
1.13062656e+00 -1.20185089e+00 1.20263565e+00 6.79018348e-02
-7.69400299e-01 -2.88881153e-01 -6.81650758e-01 -3.56209934e-01
-8.89987051e-01 3.55763376e-01 7.07422316e-01 6.17468417e-01
-8.21865559e-01 1.78360760e-01 -7.82889724e-01 -9.11633909e-01
3.21935713e-01 -1.03987932e-01 -8.87206674e-01 -3.85784179e-01
-1.45048320e+00 1.36718643e+00 6.95611238e-01 -2.44233564e-01
-4.13529187e-01 -6.97284043e-01 -1.15878320e+00 -3.06255221e-01
-1.13313317e-01 2.66061723e-01 1.18698204e+00 -8.79668534e-01
-1.27687943e+00 1.21025264e+00 -1.84745826e-02 -2.73212403e-01
3.07547182e-01 -5.50289869e-01 -6.57400608e-01 -1.96605623e-01
2.01628178e-01 5.91740012e-01 1.88100874e-01 -9.33875442e-01
-7.60956287e-01 -2.89685160e-01 -1.85703486e-01 3.86363387e-01
-5.45248151e-01 7.79557526e-01 -2.49484196e-01 -8.26600194e-01
-2.66024411e-01 -9.15728390e-01 -1.02283955e-01 -5.68527341e-01
-7.78525323e-02 -4.36517835e-01 6.05523169e-01 -1.23359728e+00
1.11102164e+00 -2.23655558e+00 -2.16367781e-01 -2.18722895e-01
-4.02678162e-01 3.83780569e-01 -3.44358504e-01 8.38777125e-01
-4.50758338e-02 5.49724877e-01 -5.12712188e-02 -1.70049280e-01
-2.19824567e-01 3.74117613e-01 -1.86691329e-01 4.94470567e-01
4.27576751e-01 8.44815135e-01 -9.04592872e-01 -3.10658693e-01
4.60506268e-02 5.41796029e-01 -2.22050622e-01 -7.95085914e-03
2.05286041e-01 2.80865282e-01 -2.97117323e-01 5.47766626e-01
3.19152445e-01 4.20773923e-01 4.36236203e-01 1.69194024e-02
-6.21024847e-01 9.58976924e-01 -7.41943598e-01 1.41387510e+00
-1.01483071e+00 8.27551723e-01 5.62757719e-03 -9.73589718e-01
7.98018992e-01 4.95283633e-01 1.41834974e-01 -4.97630149e-01
3.00997496e-01 5.99742353e-01 2.69417346e-01 -3.90660375e-01
7.79371560e-01 2.35087425e-03 -4.04132456e-01 6.41201258e-01
3.47128958e-01 -1.40430897e-01 4.62270558e-01 -7.81970844e-02
8.06367278e-01 2.27187812e-01 7.00090766e-01 -5.05865037e-01
1.89994752e-01 4.09652114e-01 5.09203374e-01 3.25138986e-01
-7.51794502e-02 3.99461478e-01 9.10522640e-02 -3.66691053e-01
-1.15386605e+00 -7.87763953e-01 -7.84571528e-01 1.54900682e+00
-4.74333644e-01 -3.96611959e-01 -3.27631265e-01 -9.36839700e-01
-5.68960831e-02 8.45973372e-01 -5.01893342e-01 4.52010572e-01
-8.80478144e-01 -9.65328395e-01 7.71108687e-01 5.33332467e-01
1.93705171e-01 -1.18300152e+00 -1.45247430e-02 3.90642822e-01
-2.49862731e-01 -1.11169219e+00 -4.53526437e-01 5.76343954e-01
-6.44164801e-01 -9.41713452e-01 -7.94828713e-01 -1.36707306e+00
5.20019412e-01 7.08552971e-02 1.02103388e+00 -3.10510814e-01
-3.24740522e-02 2.61434227e-01 -8.10853422e-01 -6.09893918e-01
-6.07508540e-01 4.31521714e-01 2.19588146e-01 -5.88317215e-01
6.81689203e-01 -7.60551170e-02 2.44179577e-01 5.91684505e-02
-6.93521440e-01 -1.53837383e-01 6.55223966e-01 9.07331765e-01
4.05080408e-01 -2.02013329e-01 7.89198458e-01 -1.32379818e+00
6.55271411e-01 -8.27605903e-01 -2.26923317e-01 3.18347305e-01
-2.02197105e-01 -1.11702301e-01 8.28134716e-01 -4.99267846e-01
-1.27039409e+00 -3.24036300e-01 -4.88578320e-01 5.83882093e-01
-3.97724032e-01 9.92033422e-01 2.68468615e-02 3.26228470e-01
7.42274940e-01 -2.12847903e-01 -5.25938034e-01 -7.02579260e-01
4.76902187e-01 1.23697317e+00 1.80548951e-01 -6.64847553e-01
5.77795148e-01 -1.62260786e-01 -8.84085178e-01 -1.35704863e+00
-9.75242972e-01 -6.10747874e-01 -9.12861228e-01 2.83096582e-01
9.58810210e-01 -1.14392924e+00 1.72078878e-01 4.26331013e-01
-1.17681086e+00 -5.69714010e-01 -1.40350401e-01 9.51631367e-01
2.19899360e-02 -5.54806180e-03 -8.41725767e-01 -5.94928563e-01
-3.63630772e-01 -7.84337163e-01 5.45057178e-01 9.97740775e-02
-5.50728858e-01 -1.56178427e+00 4.32843149e-01 3.98595721e-01
4.10761654e-01 -3.70582826e-02 1.14785826e+00 -1.38697970e+00
2.23535448e-01 -1.75454721e-01 -1.50327772e-01 5.67134023e-01
2.82324135e-01 1.78343400e-01 -8.44457567e-01 -2.53821999e-01
-4.41086441e-01 -8.25698614e-01 7.42603958e-01 9.99809057e-02
4.37385827e-01 -2.78868169e-01 -9.34604108e-02 2.82371938e-01
1.29471123e+00 2.60548770e-01 2.45449573e-01 6.30167961e-01
5.76447725e-01 7.17629731e-01 4.81781065e-01 1.62507612e-02
7.15518236e-01 3.32675755e-01 -4.20314699e-01 4.32992764e-02
-2.32827753e-01 -1.33808836e-01 8.32619190e-01 1.71846890e+00
-1.72924027e-02 -3.36183548e-01 -1.45028234e+00 9.53743339e-01
-1.57586133e+00 -7.09794164e-01 3.35603915e-02 1.96766293e+00
1.28414643e+00 -1.99805483e-01 -3.32039356e-01 -2.13633716e-01
5.43819070e-01 2.17403769e-01 -9.23773460e-03 -8.74860525e-01
-2.23961756e-01 4.61066604e-01 4.40256178e-01 6.29714727e-01
-1.38937593e+00 1.28838909e+00 5.40809631e+00 7.52773702e-01
-1.04783213e+00 4.13753211e-01 4.82981533e-01 3.05808812e-01
-2.49758795e-01 -7.90980533e-02 -1.26533329e+00 2.74488062e-01
1.27953231e+00 -1.32507920e-01 2.91156948e-01 8.20888221e-01
-9.23015177e-02 1.34912953e-01 -9.04638112e-01 6.11909032e-01
1.97434947e-01 -1.23364902e+00 -2.09281653e-01 -2.31842816e-01
7.48385668e-01 8.92402709e-01 -2.12637335e-01 6.66369557e-01
8.42540860e-01 -8.94434571e-01 5.49531102e-01 -2.42108911e-01
9.50710297e-01 -8.25382471e-01 1.10515678e+00 1.78765371e-01
-9.12522018e-01 2.32664734e-01 -5.67392945e-01 1.44021973e-01
1.32769227e-01 4.28236693e-01 -6.99055493e-01 5.89122355e-01
6.64826393e-01 7.74708688e-01 -5.01864851e-01 5.37993431e-01
-5.39505422e-01 1.15869045e+00 -3.74447376e-01 -3.11452389e-01
4.60456997e-01 -1.40924692e-01 2.14190975e-01 1.81543994e+00
3.74282628e-01 -6.75756037e-02 3.99582088e-01 -1.11825101e-01
-4.38541710e-01 9.06039178e-01 -6.80187643e-01 -5.69370329e-01
5.69351852e-01 1.38902402e+00 -7.06919730e-01 -1.00282416e-01
-8.36173236e-01 6.86087668e-01 8.30174804e-01 1.27380773e-01
-3.80529463e-01 -6.77482784e-01 6.65756524e-01 -5.05044833e-02
7.93092400e-02 -5.31932950e-01 -2.88605511e-01 -1.48189211e+00
-2.48267695e-01 -9.42457318e-01 5.14423549e-01 -1.95954293e-01
-1.77539754e+00 8.27885866e-01 3.78989801e-02 -7.01520383e-01
-2.92847812e-01 -1.08624458e+00 -6.25502527e-01 9.27695990e-01
-1.83175123e+00 -1.53685689e+00 2.72024065e-01 2.30023310e-01
8.11778486e-01 -5.58948219e-01 1.28781676e+00 3.70499879e-01
-5.83242536e-01 5.56787968e-01 6.11595929e-01 6.77358150e-01
1.07311606e+00 -1.14833355e+00 4.83755141e-01 7.45677471e-01
4.64092910e-01 5.38926125e-01 2.81014502e-01 -6.83139920e-01
-1.10639083e+00 -1.14652324e+00 1.52238512e+00 -4.88501906e-01
1.18620121e+00 -5.58700800e-01 -9.85622466e-01 9.84800220e-01
7.67306566e-01 -2.33313128e-01 1.25371933e+00 7.21297860e-01
-3.65264714e-01 2.96978563e-01 -9.36851859e-01 7.17464447e-01
6.51518583e-01 -5.45484900e-01 -9.60495293e-01 4.76816237e-01
4.71438795e-01 -2.32611760e-01 -9.24770355e-01 -1.44652709e-01
5.66351414e-01 -2.13236198e-01 5.47695518e-01 -8.66503835e-01
5.18241704e-01 1.51353270e-01 -2.38716066e-01 -1.77434027e+00
-2.41557837e-01 -3.66270810e-01 5.37943125e-01 1.81661367e+00
9.68910456e-01 -8.52662623e-01 2.47873396e-01 1.84908807e-01
-4.29635733e-01 -4.77382213e-01 -9.13630128e-01 -7.91085184e-01
8.27479601e-01 -7.47263253e-01 1.00848697e-01 1.58837211e+00
1.75305292e-01 5.31098425e-01 -1.60406962e-01 -1.51011432e-02
8.63129422e-02 -2.46307373e-01 5.29931247e-01 -1.15107691e+00
-4.47915606e-02 -2.52229422e-01 -1.32628351e-01 -6.08735800e-01
5.29869854e-01 -1.37281299e+00 1.76810265e-01 -1.67783666e+00
-5.59053756e-02 -7.98405766e-01 -1.16107270e-01 1.07141387e+00
-1.79594353e-01 2.14369878e-01 1.75884545e-01 9.07338411e-02
-1.54702198e-02 2.22958475e-01 7.41283059e-01 4.04161252e-02
-2.98176378e-01 -5.31388104e-01 -8.00025582e-01 8.83196533e-01
1.18055964e+00 -7.60657012e-01 5.29709943e-02 -8.36307943e-01
1.08038343e-01 -5.28478920e-01 -4.06553566e-01 -5.71512163e-01
-1.71977386e-01 -1.80177465e-01 4.13394362e-01 -2.79821485e-01
-8.12987909e-02 -6.25861764e-01 -3.62307459e-01 1.33358777e-01
-3.73132050e-01 3.83176684e-01 4.53454286e-01 -1.55752152e-01
-3.37425172e-01 -6.69685185e-01 8.35544705e-01 -1.56113073e-01
-7.88240969e-01 1.11220405e-01 -7.60817289e-01 6.97186291e-01
9.23373461e-01 7.76576400e-02 -5.13804734e-01 -4.63270433e-02
-4.52080250e-01 2.60393143e-01 2.87179798e-01 6.46085620e-01
2.31014192e-01 -1.22402394e+00 -1.19098008e+00 2.64458299e-01
1.65537804e-01 -2.50896275e-01 -9.70644951e-02 6.06663883e-01
-7.26392150e-01 4.14729625e-01 -2.55205154e-01 1.25065506e-01
-1.01240599e+00 5.41896373e-02 -2.64317244e-01 -4.84483838e-01
-4.10276026e-01 7.02545941e-01 -2.03590430e-02 -1.14275360e+00
-1.31000370e-01 -2.02839389e-01 -4.69532311e-01 5.41216791e-01
4.61202979e-01 2.18114913e-01 -1.86942238e-02 -9.41593170e-01
-4.12909508e-01 1.84093103e-01 -4.69926745e-01 -1.53917626e-01
1.71486378e+00 -9.23985690e-02 -1.37392089e-01 7.79090881e-01
1.22469938e+00 6.55533850e-01 -5.02159536e-01 -2.15327457e-01
3.75045657e-01 -2.49567762e-01 6.88657090e-02 -9.24527586e-01
-5.38115799e-01 8.86700094e-01 9.19606462e-02 1.08850449e-01
5.42462647e-01 2.17836648e-01 7.62680948e-01 5.43089986e-01
1.44200400e-01 -1.34910333e+00 -4.48175490e-01 1.33145452e+00
8.42454135e-01 -1.33654952e+00 -1.81802362e-01 -1.02894217e-01
-9.24972832e-01 1.27790248e+00 4.74553525e-01 9.00362432e-02
8.81014943e-01 4.29958254e-01 6.46536171e-01 1.38413042e-01
-6.00723684e-01 -2.36434773e-01 1.73688203e-01 7.33768582e-01
1.25665808e+00 1.18708909e-01 -5.08241355e-01 7.64548004e-01
-5.21149874e-01 -3.95206869e-01 6.38826787e-01 8.96564603e-01
-3.69412661e-01 -1.39185655e+00 -1.82899401e-01 5.90874612e-01
-1.06605577e+00 -6.16553664e-01 -2.47071758e-01 1.23804975e+00
7.84417540e-02 7.69446731e-01 2.45105907e-01 1.13079444e-01
2.49419421e-01 4.92358327e-01 1.28422484e-01 -1.07183695e+00
-6.54441655e-01 -4.99187931e-02 7.90504277e-01 2.32101068e-01
-4.02131081e-01 -9.34066355e-01 -1.14942718e+00 -2.77451634e-01
-3.69995177e-01 5.13297260e-01 8.64087164e-01 8.00382733e-01
6.70763031e-02 1.05359368e-01 3.73384833e-01 -8.14662278e-01
-3.18394572e-01 -1.33152223e+00 -4.92092013e-01 1.87895179e-01
-1.17614709e-01 -2.43961811e-01 -3.28654885e-01 7.31802359e-02] | [10.572056770324707, 9.901747703552246] |
76a25769-31ea-4f76-b37d-bafd3e99011c | underwater-ranker-learn-which-is-better-and | 2208.06857 | null | https://arxiv.org/abs/2208.06857v2 | https://arxiv.org/pdf/2208.06857v2.pdf | Underwater Ranker: Learn Which Is Better and How to Be Better | In this paper, we present a ranking-based underwater image quality assessment (UIQA) method, abbreviated as URanker. The URanker is built on the efficient conv-attentional image Transformer. In terms of underwater images, we specially devise (1) the histogram prior that embeds the color distribution of an underwater image as histogram token to attend global degradation and (2) the dynamic cross-scale correspondence to model local degradation. The final prediction depends on the class tokens from different scales, which comprehensively considers multi-scale dependencies. With the margin ranking loss, our URanker can accurately rank the order of underwater images of the same scene enhanced by different underwater image enhancement (UIE) algorithms according to their visual quality. To achieve that, we also contribute a dataset, URankerSet, containing sufficient results enhanced by different UIE algorithms and the corresponding perceptual rankings, to train our URanker. Apart from the good performance of URanker, we found that a simple U-shape UIE network can obtain promising performance when it is coupled with our pre-trained URanker as additional supervision. In addition, we also propose a normalization tail that can significantly improve the performance of UIE networks. Extensive experiments demonstrate the state-of-the-art performance of our method. The key designs of our method are discussed. We will release our dataset and code. | ['Chongyi Li', 'Weidong Zhang', 'Zhi Chai', 'Linghao Han', 'Xin Jin', 'Ruiqi Wu', 'Chunle Guo'] | 2022-08-14 | null | null | null | null | ['uie'] | ['computer-vision'] | [ 1.76556751e-01 -1.86646059e-01 6.53984308e-01 -4.73666579e-01
-9.05767202e-01 -2.85640061e-01 5.29347658e-02 -2.01761529e-01
-6.14251256e-01 3.35646957e-01 3.95596802e-01 6.18451647e-02
-2.83275694e-01 -8.58622849e-01 -8.47588956e-01 -9.62805212e-01
-3.90722543e-01 -3.24654877e-01 4.32605565e-01 -6.25168741e-01
3.58175188e-01 3.41008276e-01 -1.72156382e+00 2.84872562e-01
1.17684042e+00 1.27214730e+00 6.21937215e-01 8.38214755e-01
3.66804153e-01 8.87926161e-01 -5.64334452e-01 -4.39962447e-01
3.41273755e-01 -4.95227993e-01 -3.77282739e-01 -1.25844657e-01
5.84301174e-01 -7.55281031e-01 -6.67428911e-01 1.43434274e+00
9.72976506e-01 2.37102225e-01 4.84763861e-01 -1.02979720e+00
-7.69133985e-01 6.04349196e-01 -2.35656038e-01 3.41072619e-01
-1.28974780e-01 -3.36762108e-02 1.09296048e+00 -1.16412854e+00
2.51075059e-01 1.27399385e+00 7.02870905e-01 3.76970232e-01
-5.61084449e-01 -6.95459902e-01 1.38245627e-01 5.28935730e-01
-1.01799405e+00 -4.05297816e-01 7.90810347e-01 -7.14396089e-02
4.47531998e-01 1.19342417e-01 6.65576935e-01 6.04078948e-01
1.92110926e-01 8.73222232e-01 1.17094851e+00 -6.91118464e-02
2.73718804e-01 -3.89396131e-01 -3.10795638e-03 7.18262434e-01
-7.68650398e-02 1.24870099e-01 -5.19555271e-01 2.43655384e-01
7.48822510e-01 6.28460795e-02 -6.48755252e-01 2.50233366e-04
-7.78652668e-01 5.54352641e-01 8.06569993e-01 5.75615019e-02
-1.28765032e-01 1.78583056e-01 3.56338918e-01 4.38515961e-01
3.56737345e-01 2.83552200e-01 -5.24986863e-01 1.80739477e-01
-5.29140413e-01 -1.59126222e-01 4.09956038e-01 6.97474301e-01
1.00586426e+00 7.51191452e-02 -2.40141526e-01 9.93764818e-01
4.73442912e-01 7.11401284e-01 3.25760871e-01 -1.12607574e+00
3.65219623e-01 2.90736914e-01 1.33549944e-01 -9.38415110e-01
-3.79592001e-01 -1.88162953e-01 -8.46057594e-01 5.09547174e-01
4.84101521e-03 6.67155385e-02 -1.05467832e+00 1.50468647e+00
-1.44061267e-01 1.54480487e-01 5.15146136e-01 1.02649117e+00
1.06667399e+00 9.42314029e-01 -1.83393717e-01 -2.08858654e-01
1.14011991e+00 -9.94563758e-01 -8.21769178e-01 -1.35656223e-01
6.07250147e-02 -4.83323932e-01 1.18917918e+00 4.16455925e-01
-1.08955407e+00 -6.58095896e-01 -1.32363355e+00 -9.36451182e-02
-9.55317244e-02 1.87341154e-01 3.64379793e-01 3.72383147e-01
-1.40419650e+00 9.75271583e-01 -7.81655133e-01 1.13305878e-02
1.98740363e-01 1.20828569e-01 -2.53645629e-01 -3.56864721e-01
-1.44063067e+00 8.32981825e-01 2.60016710e-01 5.67248642e-01
-1.41948581e+00 -4.37770635e-01 -1.13518679e+00 1.83439434e-01
1.04467496e-01 -1.27676621e-01 1.05141938e+00 -7.35421658e-01
-1.58516729e+00 3.88718456e-01 7.78743774e-02 -7.98735991e-02
4.71881390e-01 -3.83276731e-01 -3.28060150e-01 6.19092882e-01
6.71304762e-02 4.17681217e-01 7.15080917e-01 -1.61417735e+00
-8.74281108e-01 -2.67488182e-01 1.98103800e-01 4.80129659e-01
-7.42544889e-01 1.24530457e-01 -9.54654753e-01 -6.43998086e-01
2.89001286e-01 -6.14790499e-01 -1.61436617e-01 3.01347613e-01
-2.22805925e-02 -1.62348226e-01 6.80557609e-01 -8.51795077e-01
1.09053063e+00 -2.42062116e+00 2.50090033e-01 3.26959277e-03
9.82561484e-02 1.70542777e-01 -5.71495056e-01 3.84276509e-01
1.76161200e-01 8.47107172e-02 -4.92515534e-01 -4.16519374e-01
9.07433331e-02 4.76978809e-01 -1.81784645e-01 6.32234693e-01
1.80444315e-01 5.94680607e-01 -1.21143913e+00 -5.45113146e-01
1.49678841e-01 4.41191018e-01 -4.47540581e-01 7.68172443e-01
3.39581072e-01 2.12222442e-01 -2.07677618e-01 8.35975349e-01
9.03152645e-01 5.73194809e-02 -1.38581157e-01 -8.34063292e-01
-2.67905623e-01 -1.91478938e-01 -9.95185792e-01 1.50789833e+00
-4.53112602e-01 7.15284169e-01 3.09770882e-01 -7.25243688e-01
9.87855494e-01 -2.37277504e-02 3.09085280e-01 -9.57136095e-01
-3.55788344e-03 4.05054539e-01 -2.39081770e-01 -7.16268063e-01
6.23114169e-01 -1.68522790e-01 1.61666632e-01 -1.31242648e-01
1.80540949e-01 -5.32578081e-02 3.47635359e-01 2.08531082e-01
8.63253057e-01 1.98572889e-01 -8.77324585e-03 -3.68558109e-01
3.28866601e-01 -6.10091209e-01 7.97671616e-01 7.64909267e-01
-3.78854007e-01 8.33383977e-01 1.85425505e-01 -3.59674066e-01
-9.56261218e-01 -1.04320085e+00 -4.07203317e-01 1.25049734e+00
1.00177324e+00 -2.80205347e-02 -4.68659312e-01 -4.44478124e-01
-3.55010509e-01 7.51022622e-02 -8.79373729e-01 -1.58341184e-01
-5.50169468e-01 -1.05124497e+00 4.42152709e-01 6.74201012e-01
8.52680624e-01 -1.13288939e+00 -3.86568546e-01 1.75683022e-01
-3.02168399e-01 -9.24221158e-01 -4.47727919e-01 3.75970960e-01
-7.26786137e-01 -9.60297823e-01 -9.85321999e-01 -9.01950598e-01
5.83006084e-01 4.70902234e-01 1.05157673e+00 3.93362582e-01
2.35705584e-01 1.59630105e-01 -9.24556434e-01 -1.52294442e-01
-2.66896456e-01 -6.65707231e-01 2.80346274e-02 -8.67847502e-02
-2.85443634e-01 -6.04392231e-01 -1.10994458e+00 5.48131764e-01
-1.37325978e+00 -1.33359298e-01 9.11243677e-01 9.92494404e-01
4.58056003e-01 2.22305715e-01 2.96052307e-01 -2.45450839e-01
1.98528349e-01 -2.84577549e-01 -3.94564569e-01 2.96117425e-01
-5.51813006e-01 2.65469924e-02 4.96211171e-01 -3.38579148e-01
-9.34809923e-01 -3.06090295e-01 -6.16915882e-01 -3.47365737e-01
2.88085043e-01 6.90083444e-01 -3.65414947e-01 -3.57497543e-01
3.41772377e-01 3.50293875e-01 -1.11251466e-01 -4.51819867e-01
1.27252012e-01 7.42257476e-01 8.63593638e-01 -4.34561193e-01
8.41807425e-01 5.07696927e-01 -3.53372514e-01 -7.12851346e-01
-1.04909945e+00 -4.02611017e-01 -2.45998263e-01 -5.64349115e-01
8.39630067e-01 -1.28850782e+00 -6.24011159e-01 8.45627666e-01
-9.76063967e-01 -6.97368443e-01 7.78695047e-02 3.55716556e-01
-2.12346479e-01 7.19459653e-01 -1.12971437e+00 -8.65900755e-01
-4.45940405e-01 -1.27612925e+00 1.16817117e+00 4.75888312e-01
9.96041358e-01 -6.77080333e-01 -5.35191894e-02 1.49938047e-01
5.55248618e-01 1.54078647e-01 5.11176884e-01 -1.21196993e-01
-4.49444264e-01 2.53196090e-01 -6.76369071e-01 7.94673324e-01
-1.17113972e-02 6.19341340e-03 -1.04408824e+00 -6.24479473e-01
-1.61404967e-01 -5.03470182e-01 1.25740433e+00 2.69976676e-01
1.13296020e+00 -1.81986079e-01 2.88779318e-01 9.17417943e-01
1.63168037e+00 1.77440926e-01 1.14737570e+00 7.15657115e-01
7.02197611e-01 6.67587996e-01 7.31034398e-01 4.92350876e-01
6.20468497e-01 4.62682873e-01 1.03355920e+00 -5.96560478e-01
-1.48939833e-01 -1.55147910e-01 8.18790376e-01 1.23074484e+00
-3.69898707e-01 -2.99184859e-01 -4.17844772e-01 6.16576672e-01
-1.63539231e+00 -8.22061300e-01 3.70848551e-02 1.83093655e+00
8.08184743e-01 -2.37687618e-01 -3.72346252e-01 6.56609982e-03
7.11905956e-01 2.05960304e-01 -6.25744402e-01 1.58941373e-01
-4.29055542e-01 -2.07652494e-01 6.63799763e-01 4.30452943e-01
-1.21055043e+00 6.66365206e-01 5.78476429e+00 8.47814143e-01
-8.53076041e-01 -1.09672271e-01 6.10729754e-01 6.05820775e-01
-3.40062350e-01 -2.88268656e-01 -6.00959897e-01 5.11726856e-01
4.98483926e-01 2.94206172e-01 3.54655296e-01 7.43223488e-01
2.60034293e-01 1.68582603e-01 -6.41400754e-01 8.98174405e-01
3.48774642e-01 -8.00172806e-01 5.95195219e-02 -1.52946919e-01
7.77353883e-01 2.03397945e-01 -9.91993323e-02 3.76613855e-01
3.91148955e-01 -7.84822047e-01 1.07605720e+00 5.32013237e-01
9.00500298e-01 -5.36887705e-01 1.29283929e+00 -1.76376253e-01
-1.40262234e+00 -3.60795736e-01 -7.88603544e-01 2.58024782e-02
9.70554873e-02 3.77845913e-01 1.23586975e-01 7.42090404e-01
1.50079703e+00 9.75719333e-01 -6.39089108e-01 1.24039841e+00
-4.60558683e-01 6.10078752e-01 -1.58600911e-01 1.88458696e-01
1.30312383e-01 -4.66157794e-01 2.60363907e-01 1.13773334e+00
7.28855848e-01 4.01604623e-01 3.04625351e-02 1.58321753e-01
-1.76344112e-01 1.57748964e-02 -1.00606613e-01 4.68466282e-01
2.96162695e-01 1.27805316e+00 -4.48262304e-01 -4.02138323e-01
-4.18716192e-01 1.05856478e+00 1.30022570e-01 3.98131967e-01
-7.04147398e-01 -6.10028386e-01 6.60928667e-01 -4.58286524e-01
4.42730129e-01 7.27924798e-03 1.57135293e-01 -1.27383780e+00
-5.07758558e-02 -7.62179732e-01 3.22063118e-01 -1.11955714e+00
-1.62624812e+00 1.05190647e+00 -2.14392051e-01 -1.71959770e+00
5.60885668e-01 -7.59314716e-01 -5.61321914e-01 4.80487823e-01
-2.41975284e+00 -1.02511311e+00 -9.14999604e-01 5.22416234e-01
5.62939644e-01 1.71512082e-01 4.22578722e-01 5.99214494e-01
-4.93745178e-01 5.53548217e-01 2.65207559e-01 4.25503850e-01
7.93734908e-01 -1.47626662e+00 -1.48674935e-01 1.18708098e+00
-2.85437107e-01 3.45549554e-01 8.57677341e-01 -4.08960938e-01
-1.24465263e+00 -1.22439027e+00 2.32537940e-01 7.89513066e-02
8.56862664e-01 1.09256543e-01 -1.15695786e+00 2.88118124e-01
2.26334393e-01 2.77971506e-01 5.32720029e-01 -4.81208354e-01
-1.56786487e-01 -4.91399229e-01 -7.31248081e-01 5.11278689e-01
1.03790545e+00 -3.61616343e-01 -5.07964253e-01 7.98780993e-02
8.97249877e-01 -5.08745074e-01 -1.13712502e+00 8.64077747e-01
5.54391086e-01 -1.24676430e+00 8.82456839e-01 1.08361878e-01
8.69044006e-01 -8.32457006e-01 -5.08408487e-01 -1.37924457e+00
-2.59443074e-01 -2.76150018e-01 1.98728204e-01 1.10587788e+00
2.15104029e-01 -3.32814813e-01 3.05428863e-01 3.13182712e-01
-8.27646852e-01 -6.04347825e-01 -7.56049335e-01 -5.23897111e-01
-1.22884855e-01 -2.65524864e-01 4.31217462e-01 6.32995665e-01
-4.00669962e-01 1.24491844e-02 -9.13772941e-01 8.83108079e-01
8.65065992e-01 1.50654793e-01 4.76651132e-01 -9.99265075e-01
-1.56811267e-01 -2.94725925e-01 -4.67464060e-01 -1.37204945e+00
-3.53121281e-01 -4.09190953e-01 7.72280931e-01 -1.85956550e+00
5.02739549e-01 -2.33568549e-01 -7.77493417e-01 6.22474134e-01
-7.06210077e-01 8.22887599e-01 2.20482200e-01 3.73783201e-01
-8.90889227e-01 1.17812753e+00 1.48367035e+00 -2.94395179e-01
-9.34202131e-03 -3.21278989e-01 -6.46091759e-01 5.82747936e-01
5.66041648e-01 -2.85679668e-01 -5.98811358e-02 -7.16296673e-01
3.20028365e-01 1.75089568e-01 1.97359279e-01 -1.02463090e+00
2.86415815e-01 -6.15878170e-03 3.52441907e-01 -3.74904186e-01
1.74237072e-01 -6.97973788e-01 -3.06692988e-01 3.79356891e-01
-2.76669174e-01 -8.45414994e-04 -8.38940069e-02 5.80203176e-01
-7.32536674e-01 -3.05883288e-01 1.00109398e+00 -7.17954012e-03
-1.30436552e+00 4.49391723e-01 -1.53909802e-01 -2.37143531e-01
4.52864796e-01 -1.10253461e-01 -5.32164693e-01 -4.71349806e-01
-4.14282769e-01 5.13353050e-01 5.71952879e-01 1.32320419e-01
1.10724545e+00 -1.09924173e+00 -9.97759283e-01 -1.23748727e-01
3.65484864e-01 5.03420234e-02 5.71765423e-01 5.82927763e-01
-7.67028093e-01 -6.54885292e-01 -3.49469930e-01 -3.72366697e-01
-1.09062874e+00 1.40314236e-01 4.99767512e-01 -1.41841799e-01
-7.03444660e-01 8.67636502e-01 4.80267107e-01 -3.93616021e-01
1.72998175e-01 -2.49534771e-01 -7.28627324e-01 3.24783400e-02
8.76148999e-01 2.84403116e-01 -2.47768480e-02 -5.65989196e-01
-1.86042622e-01 7.73141742e-01 3.35983068e-01 3.54541615e-02
1.68693149e+00 -4.77088749e-01 -3.59512478e-01 1.37216955e-01
1.23172402e+00 1.65472329e-02 -1.97392786e+00 -1.14334777e-01
-4.71826136e-01 -5.35538435e-01 2.97272772e-01 -6.51435971e-01
-1.43321717e+00 7.87887275e-01 1.03347659e+00 1.13516591e-01
1.67687154e+00 -5.47767393e-02 7.41388738e-01 2.98517942e-01
4.33975130e-01 -8.84741902e-01 5.44250607e-01 6.27493203e-01
1.07967901e+00 -1.56013882e+00 1.17590904e-01 -4.47011329e-02
-7.03197122e-01 1.18080747e+00 6.26657009e-01 -7.48893544e-02
3.85171473e-01 2.92812437e-01 6.46542072e-01 -1.49748951e-01
-2.67318785e-01 -4.39937025e-01 2.29817927e-01 5.21181345e-01
-4.27566022e-02 -3.03417504e-01 -1.53093159e-01 5.70702672e-01
8.71463716e-02 -3.18594456e-01 7.19456732e-01 6.83336675e-01
-7.58392572e-01 -6.71239316e-01 -2.74360627e-01 1.91794410e-01
-5.40097058e-01 -4.02066439e-01 3.91992092e-01 4.82179791e-01
1.45210341e-01 1.13589919e+00 -1.88360259e-01 -7.08469927e-01
5.20816088e-01 -8.07202697e-01 -1.67977903e-02 -1.21843815e-01
-2.01335564e-01 1.45511106e-01 -2.25311741e-01 -6.40531898e-01
-8.17351222e-01 -3.18215609e-01 -1.23611963e+00 2.18606770e-01
-3.42852205e-01 4.47091043e-01 5.95889509e-01 7.08009005e-01
-1.20561093e-01 6.11409783e-01 1.08425593e+00 -1.40731454e+00
-3.86545032e-01 -1.28737593e+00 -8.90273273e-01 5.18208802e-01
6.07665598e-01 -5.94181478e-01 -8.43857348e-01 1.87357336e-01] | [10.693957328796387, -3.527132272720337] |
25223227-ebf2-4f42-bd46-291df8e3a040 | convergence-theory-of-generalized-distributed | 2207.10969 | null | https://arxiv.org/abs/2207.10969v2 | https://arxiv.org/pdf/2207.10969v2.pdf | Convergence Theory of Generalized Distributed Subgradient Method with Random Quantization | The distributed subgradient method (DSG) is a widely discussed algorithm to cope with large-scale distributed optimization problems in the arising machine learning applications. Most exisiting works on DSG focus on ideal communication between the cooperative agents such that the shared information between agents is exact and perfect. This assumption, however, could lead to potential privacy concerns and is not feasible when the wireless transmission links are not of good quality. To overcome the challenge, a common approach is to quantize the data locally before transmission, which avoids exposure of raw data and significantly reduces the size of data. Compared with perfect data, quantization poses fundamental challenges on loss of data accuracy, which further impacts the convergence of the algorithms. To settle the problem, we propose a generalized distributed subgradient method with random quantization, which can be intepreted as a two time-scale stochastic approximation method. We provide comprehensive results on the convergence of the algorithm and derive upper bounds on the convergence rates in terms of the quantization bit, stepsizes and the number of network agents. Our results extend the existing results, where only special cases are considered and general conclusions for the convergence rates are missing. Finally, numerical simulations are conducted on linear regression problems to support our theoretical results. | ['Yong Ren', 'Jun Du', 'Zhaoyue Xia'] | 2022-07-22 | null | null | null | null | ['distributed-optimization'] | ['methodology'] | [-6.19594716e-02 2.94602126e-01 -2.55325854e-01 -2.91628927e-01
-8.27948928e-01 -3.98696333e-01 3.89463641e-02 3.78560632e-01
-6.82644606e-01 1.27975738e+00 -3.02311927e-01 -1.53065726e-01
-3.92340273e-01 -8.90031815e-01 -5.77351332e-01 -1.32094967e+00
-5.30386746e-01 1.02184638e-01 -2.82108009e-01 -1.02058180e-01
6.98877126e-02 1.10457577e-01 -8.04216921e-01 -5.17343342e-01
1.05399275e+00 1.33925509e+00 1.15133584e-01 3.76232564e-01
9.37783420e-02 5.08141577e-01 -1.04528391e+00 -5.04550993e-01
7.40632594e-01 -3.66562963e-01 -4.22022194e-01 2.02560365e-01
-1.15221463e-01 -6.18171573e-01 -2.15001926e-01 1.64831638e+00
5.35823226e-01 1.80133227e-02 3.12311679e-01 -1.58336389e+00
-3.69690925e-01 9.33504581e-01 -8.55224192e-01 5.97849078e-02
1.63940638e-02 -3.93497139e-01 8.98971498e-01 -3.82486910e-01
3.52467954e-01 1.03706336e+00 6.32844210e-01 4.10908312e-01
-7.57459223e-01 -8.01486611e-01 2.13628873e-01 2.03113928e-01
-1.61890244e+00 -2.88936734e-01 6.89426899e-01 3.13925259e-02
1.56416535e-01 4.44673598e-01 6.17620289e-01 3.78134131e-01
2.24776581e-01 6.82935894e-01 1.01213384e+00 -3.78523439e-01
7.31512964e-01 3.57155859e-01 -1.95631340e-01 5.21123946e-01
6.24848545e-01 -6.69466853e-02 -4.66495693e-01 -5.22586942e-01
5.99525213e-01 1.43864676e-01 -6.67799652e-01 -4.39786792e-01
-7.47620046e-01 1.10862637e+00 5.54565072e-01 1.95116729e-01
-6.53379798e-01 7.73305520e-02 2.55583942e-01 6.66433275e-01
7.84198761e-01 -4.51944508e-02 -2.40981311e-01 -7.29720369e-02
-7.61662543e-01 2.96019971e-01 1.00809705e+00 1.04571462e+00
7.43916392e-01 -3.73569094e-02 1.10935317e-02 4.58149225e-01
3.20705414e-01 5.14268994e-01 1.33242130e-01 -1.13116717e+00
9.08299565e-01 1.94823235e-01 4.19985801e-01 -1.35673380e+00
-1.92945302e-01 -6.13217294e-01 -1.35008144e+00 1.74367353e-01
4.70616609e-01 -9.30466294e-01 -3.60583030e-02 1.73962748e+00
5.66918969e-01 -3.76620367e-02 2.21112683e-01 1.23881626e+00
1.37353718e-01 7.77755857e-01 -3.21218878e-01 -8.78018916e-01
7.70438910e-01 -6.14156067e-01 -1.08725178e+00 1.42743558e-01
6.76187217e-01 -2.95771211e-01 4.05296981e-01 4.51904237e-01
-1.21051049e+00 6.70588240e-02 -1.12403011e+00 4.14605856e-01
-1.83683798e-01 -1.78253025e-01 7.60533392e-01 1.02666879e+00
-1.15558755e+00 4.53299642e-01 -7.27833211e-01 -2.09673434e-01
5.21104753e-01 5.47940016e-01 2.73501198e-03 -3.06848232e-02
-1.20598900e+00 4.39355165e-01 2.92967886e-01 4.98889297e-01
-5.73903501e-01 -6.46464586e-01 -5.14892459e-01 1.09480038e-01
5.57952046e-01 -4.52980876e-01 1.01889336e+00 -9.85419393e-01
-1.65194631e+00 2.21196324e-01 5.05849347e-03 -7.90402949e-01
9.87056971e-01 1.44373462e-01 1.50731117e-01 7.96278566e-02
-6.05998747e-02 6.07492551e-02 5.81459522e-01 -9.69466984e-01
-8.86956513e-01 -4.86962914e-01 2.64240772e-01 4.84625190e-01
-8.30098212e-01 -2.61175454e-01 -2.51095176e-01 -5.82952738e-01
-2.08841875e-01 -7.08351374e-01 -7.79848337e-01 5.08185863e-01
-3.04026932e-01 -8.46538022e-02 6.70275033e-01 -4.26215619e-01
1.12675607e+00 -2.13362074e+00 2.23544352e-02 4.04182196e-01
2.97254801e-01 3.27403769e-02 1.52384087e-01 5.52344382e-01
3.72720420e-01 3.30634825e-02 -2.04798639e-01 -6.37067735e-01
8.07827339e-02 2.39367947e-01 -1.03308477e-01 9.63088334e-01
-6.61006570e-01 3.67270976e-01 -8.52376759e-01 -2.65711993e-01
-1.61428377e-02 2.23037139e-01 -4.48713571e-01 2.24932611e-01
1.20828666e-01 2.16231972e-01 -8.71482372e-01 2.10382298e-01
1.09886968e+00 -2.31116667e-01 3.08328807e-01 -1.92374028e-02
-5.50043471e-02 -2.30679020e-01 -1.45938039e+00 1.48078966e+00
-4.06059533e-01 2.97983885e-01 1.03447104e+00 -1.46667528e+00
6.88171625e-01 4.76429075e-01 7.21585929e-01 -3.40696841e-01
2.13925138e-01 2.18054026e-01 -4.58252579e-01 -2.78458655e-01
2.64803857e-01 -8.67953598e-02 -4.77412827e-02 4.60982233e-01
-4.91628319e-01 1.41164184e-01 9.78267714e-02 2.04191551e-01
8.18275094e-01 -8.10833216e-01 6.29104793e-01 -3.53730708e-01
3.74405473e-01 -2.51457244e-01 7.58092105e-01 8.60252380e-01
-3.51522744e-01 -1.33019155e-02 4.26693916e-01 -1.94505781e-01
-7.60546029e-01 -6.17864788e-01 1.26061261e-01 7.43557692e-01
5.15745580e-01 -3.63842070e-01 -9.26504433e-01 -5.44176519e-01
1.03553362e-01 2.87817776e-01 -5.72331309e-01 1.25849277e-01
1.00849465e-01 -1.20648551e+00 3.28089654e-01 -3.51567641e-02
8.86055827e-01 -2.34311938e-01 -5.86643398e-01 2.17263997e-01
-1.86385676e-01 -1.07961726e+00 -4.76948023e-01 4.15207148e-02
-7.15258777e-01 -9.48343217e-01 -8.81884158e-01 -4.22645748e-01
7.76576757e-01 5.56595445e-01 4.95517850e-01 1.87798560e-01
5.72550632e-02 3.78536582e-01 -3.28598678e-01 -5.88201582e-01
-7.65346438e-02 -4.27008122e-02 1.34523749e-01 2.66167223e-01
1.94080221e-03 -5.84237576e-01 -6.63268983e-01 1.82586342e-01
-8.28735113e-01 -2.18327895e-01 5.42326152e-01 7.06538558e-01
4.10310447e-01 5.13779163e-01 7.75884449e-01 -7.43150115e-01
8.80184352e-01 -8.14764321e-01 -1.11526585e+00 1.82088882e-01
-6.36914432e-01 -1.49777755e-01 9.97432351e-01 -1.07053272e-01
-8.32959235e-01 -1.74574733e-01 2.03987584e-01 -2.66488224e-01
3.86916906e-01 5.24442971e-01 -2.30983555e-01 -5.82751155e-01
1.82339922e-01 2.30464771e-01 2.41398335e-01 -2.08111987e-01
3.87087584e-01 9.20229375e-01 -3.16057876e-02 -3.51445854e-01
8.02685499e-01 7.13945150e-01 2.30808243e-01 -1.04786170e+00
-5.44057906e-01 -2.00858265e-01 -5.38743623e-02 -2.49069668e-02
4.09330547e-01 -9.16139305e-01 -1.16339660e+00 4.21209633e-01
-1.05391729e+00 -1.20218769e-01 -1.94961429e-01 5.73170841e-01
-5.32998562e-01 5.70903420e-01 -5.11121690e-01 -1.17399096e+00
-4.18521523e-01 -8.59514177e-01 4.25581783e-01 2.59305000e-01
4.28779185e-01 -1.31157351e+00 -2.47092381e-01 1.26361907e-01
6.18901968e-01 3.92112762e-01 2.88541883e-01 -4.10353750e-01
-7.18906403e-01 -2.12527156e-01 -5.64794689e-02 2.22366273e-01
2.79832900e-01 -4.92734790e-01 -3.72766465e-01 -7.82764196e-01
4.68616575e-01 -2.80289441e-01 2.36941949e-01 4.73679394e-01
1.20422113e+00 -9.63231266e-01 -3.46012563e-01 6.88661516e-01
1.65472937e+00 7.04041496e-02 2.26297341e-02 1.33882642e-01
2.05964655e-01 3.49313915e-01 7.28995204e-01 1.29025710e+00
5.19800961e-01 5.82281172e-01 6.71653867e-01 -1.29200488e-01
5.54704130e-01 5.01867048e-02 2.36933753e-01 6.50883734e-01
2.64687836e-03 -6.04717016e-01 -2.33962759e-01 4.19928282e-01
-1.92580640e+00 -7.69308031e-01 1.64830059e-01 2.36603975e+00
7.89884627e-01 -3.73340607e-01 1.44068688e-01 2.68185228e-01
9.22469735e-01 1.08865492e-01 -5.79987764e-01 -2.38256305e-01
-9.22996625e-02 -3.84365380e-01 1.13507116e+00 6.01888299e-01
-9.51143861e-01 3.63238186e-01 6.50426054e+00 9.97174978e-01
-1.10061026e+00 2.24040836e-01 8.28723192e-01 -2.34097019e-01
-9.37031060e-02 -3.95202249e-01 -6.09979689e-01 6.42042875e-01
6.82485342e-01 -6.48807466e-01 8.10778737e-01 8.08691144e-01
6.44976079e-01 -1.47999734e-01 -7.84691989e-01 1.18902516e+00
-1.25100926e-01 -1.16187835e+00 -2.81298429e-01 3.97287607e-01
8.12075317e-01 7.25644268e-03 1.23022772e-01 -2.71825373e-01
5.20565629e-01 -6.80455089e-01 4.42760468e-01 1.53906792e-01
4.07105714e-01 -9.98119235e-01 9.64163125e-01 4.91896659e-01
-1.01728451e+00 -3.73452157e-01 -8.18607211e-01 -3.26999158e-01
1.73172638e-01 9.31368172e-01 -5.40945053e-01 1.04847372e+00
7.24378884e-01 6.34836018e-01 -5.47892973e-02 1.23663723e+00
7.99991414e-02 4.47211057e-01 -8.27466428e-01 -4.05052185e-01
2.98295587e-01 -5.01210153e-01 6.13245487e-01 6.63556695e-01
4.71709698e-01 3.37773353e-01 3.67382348e-01 5.06631672e-01
-1.26201779e-01 2.86640286e-01 -4.69713300e-01 3.01683784e-01
9.19510603e-01 1.09199941e+00 -3.70097220e-01 -1.36712775e-01
-5.86185038e-01 9.31656599e-01 3.72487158e-01 6.46887600e-01
-5.84453225e-01 -5.65671861e-01 6.93927407e-01 -3.72040719e-01
3.22744876e-01 -3.05386901e-01 -1.86546519e-01 -9.81980264e-01
4.87354845e-01 -6.33167386e-01 4.78623182e-01 6.53948635e-02
-1.25964522e+00 1.96206272e-01 -2.14624614e-01 -1.28458357e+00
-2.29766160e-01 -1.39268292e-02 -4.33005422e-01 5.37631750e-01
-1.64865530e+00 -6.16015017e-01 -1.75766483e-01 8.16443622e-01
1.74537823e-01 -1.89490207e-02 6.08377039e-01 4.45217639e-01
-6.37746036e-01 9.98588681e-01 8.46793294e-01 4.70871180e-02
2.64652312e-01 -1.00358379e+00 -4.55963165e-01 7.85125852e-01
-2.35973328e-01 3.43976676e-01 8.72864008e-01 -1.78174481e-01
-1.71740425e+00 -1.13971758e+00 4.35135126e-01 4.55347240e-01
6.25708938e-01 -3.20047945e-01 -4.63129044e-01 4.42621350e-01
2.63960183e-01 3.04087549e-01 7.50190139e-01 -2.42714420e-01
3.95741731e-01 -5.50574899e-01 -1.56334054e+00 3.60056758e-01
6.77693129e-01 -1.98864541e-03 9.00814831e-02 5.77945352e-01
3.11473787e-01 -2.37103179e-01 -9.76285875e-01 -2.33345971e-01
1.83518872e-01 -9.20690417e-01 5.85623980e-01 -1.38535380e-01
-2.89495081e-01 -1.62046477e-01 -2.59945840e-01 -1.55554199e+00
1.59424037e-01 -1.27353776e+00 3.26137878e-02 1.17019582e+00
1.75878868e-01 -1.01818252e+00 1.07946599e+00 7.59311438e-01
4.62227374e-01 -7.45858610e-01 -1.41691077e+00 -9.38848495e-01
2.22871080e-02 -6.24182224e-02 5.23905098e-01 9.06549275e-01
5.14554799e-01 -1.52170062e-01 -6.01001322e-01 4.35045838e-01
1.09710550e+00 3.14227641e-02 6.69462621e-01 -8.12120140e-01
-8.23934451e-02 -1.09541066e-01 -4.29524779e-01 -1.42288995e+00
5.96350208e-02 -5.64515412e-01 1.02662668e-02 -1.17251456e+00
-5.57220392e-02 -8.16683650e-01 -7.51788542e-02 1.55072883e-01
-3.54970656e-02 -1.00239120e-01 3.68330061e-01 1.94058880e-01
-8.88812304e-01 9.46001351e-01 1.08938181e+00 -8.22050199e-02
-8.96451995e-02 4.81028467e-01 -6.94261432e-01 4.92643952e-01
7.77949691e-01 -4.84303385e-01 -7.66322076e-01 -5.38205326e-01
1.38865352e-01 4.17274117e-01 1.53481187e-02 -7.89252520e-01
6.38716638e-01 -2.92955816e-01 -1.19505621e-01 -2.60300875e-01
3.42447847e-01 -1.33036911e+00 6.18400723e-02 6.27977371e-01
-3.87824744e-01 -2.26285294e-01 -4.36352760e-01 9.52189267e-01
-2.54008651e-01 -2.68389881e-01 7.24070787e-01 -3.51059213e-02
-3.78294364e-02 6.53035343e-01 -4.18466806e-01 -7.46172592e-02
1.40354645e+00 -7.12056505e-03 8.77090022e-02 -9.58011746e-01
-5.01030743e-01 7.41739690e-01 2.68580884e-01 -1.62237138e-01
4.62451369e-01 -1.32168925e+00 -7.79455721e-01 -8.31972659e-02
-5.19150913e-01 6.01001531e-02 6.42127469e-02 8.54446948e-01
-2.71841109e-01 5.27074337e-01 1.60266355e-01 -2.77690947e-01
-1.02845418e+00 7.01887846e-01 4.40685093e-01 -1.03340149e-01
-3.71813387e-01 6.82499230e-01 -4.40926179e-02 4.29071533e-03
6.70350850e-01 -3.24627042e-01 1.48570668e-02 -6.92527220e-02
7.91772723e-01 5.74834168e-01 -2.74401661e-02 -1.54428512e-01
-1.85174018e-01 4.05772507e-01 -1.60503611e-02 -8.87032002e-02
1.41240132e+00 -8.64854813e-01 -2.23615721e-01 8.10938403e-02
1.40844429e+00 -9.48571321e-03 -1.38805735e+00 -4.51124489e-01
-4.10070270e-01 -6.48643196e-01 1.89578235e-01 -2.53690362e-01
-1.69882202e+00 5.83732903e-01 5.37896872e-01 6.33186758e-01
1.29321146e+00 -3.82309884e-01 6.04315639e-01 5.57346404e-01
1.01085126e+00 -1.09845114e+00 -4.39483970e-01 3.30760628e-02
6.13422394e-01 -1.11777198e+00 2.03721598e-01 -6.76941931e-01
-3.02734315e-01 1.09670496e+00 1.95615202e-01 -1.26937211e-01
7.75286615e-01 2.84119487e-01 -8.32558796e-02 1.86005071e-01
-6.13680780e-01 2.85848439e-01 -4.37931895e-01 6.52618527e-01
-5.60650835e-03 5.53993843e-02 -8.50045562e-01 6.93331897e-01
-1.03410996e-01 8.33024911e-04 8.12275827e-01 9.23493028e-01
-3.21678579e-01 -9.27979589e-01 -4.89312023e-01 2.02764019e-01
-8.42576802e-01 1.26136094e-01 2.27333289e-02 4.77092564e-01
-1.31592005e-01 1.25823534e+00 -1.35046214e-01 1.67753145e-01
-6.43101856e-02 -7.53163874e-01 1.87963337e-01 7.17343623e-03
-2.10574463e-01 -1.20860390e-01 -9.26447734e-02 -4.86849725e-01
-6.09173954e-01 -4.03716743e-01 -1.05846357e+00 -6.73271656e-01
-4.96639490e-01 8.64696383e-01 6.53137326e-01 8.65229428e-01
3.68658006e-01 5.97264059e-02 1.21491241e+00 -6.35144770e-01
-1.09544909e+00 -6.63569808e-01 -9.18244243e-01 -5.24306558e-02
6.18591189e-01 -3.69259566e-01 -6.20759666e-01 -2.90857494e-01] | [6.166580677032471, 4.959129810333252] |
15ff5e53-835d-45b0-a6fb-a6183df8cbb2 | hierarchical-modeling-of-molecular-energies | 1710.00017 | null | http://arxiv.org/abs/1710.00017v1 | http://arxiv.org/pdf/1710.00017v1.pdf | Hierarchical modeling of molecular energies using a deep neural network | We introduce the Hierarchically Interacting Particle Neural Network (HIP-NN)
to model molecular properties from datasets of quantum calculations. Inspired
by a many-body expansion, HIP-NN decomposes properties, such as energy, as a
sum over hierarchical terms. These terms are generated from a neural network--a
composition of many nonlinear transformations--acting on a representation of
the molecule. HIP-NN achieves state-of-the-art performance on a dataset of 131k
ground state organic molecules, and predicts energies with 0.26 kcal/mol mean
absolute error. With minimal tuning, our model is also competitive on a dataset
of molecular dynamics trajectories. In addition to enabling accurate energy
predictions, the hierarchical structure of HIP-NN helps to identify regions of
model uncertainty. | ['Nicholas Lubbers', 'Kipton Barros', 'Justin S. Smith'] | 2017-09-29 | null | null | null | null | ['formation-energy'] | ['miscellaneous'] | [ 2.38074884e-01 1.76793322e-01 -2.52584428e-01 -3.62593353e-01
-7.82219648e-01 -3.86393577e-01 6.02235138e-01 6.60385013e-01
-3.35756391e-01 1.13563776e+00 3.35796475e-01 -6.08392358e-01
-7.07184598e-02 -1.25559855e+00 -1.24124062e+00 -1.08552289e+00
-4.14843857e-01 6.30694568e-01 -2.09279303e-02 -2.34259695e-01
3.28671843e-01 7.08646119e-01 -1.22924149e+00 4.26386863e-01
9.95838404e-01 8.36926162e-01 -4.47534293e-01 4.66862500e-01
4.16822284e-01 7.71703959e-01 1.27219096e-01 -1.20842300e-01
1.13835856e-01 -3.17297369e-01 -9.79263008e-01 -1.07796443e+00
4.00065809e-01 -2.49742847e-02 -7.31144845e-01 1.08510053e+00
4.28504109e-01 6.72392249e-01 1.18715107e+00 -7.29723930e-01
-9.73947227e-01 6.66007280e-01 1.11101024e-01 -1.18350983e-01
2.66713649e-01 3.69912684e-01 1.26580489e+00 -9.50393081e-01
6.54264033e-01 1.27786076e+00 9.81289566e-01 5.18004119e-01
-1.98089695e+00 -6.60230577e-01 -4.83839273e-01 1.53531522e-01
-1.57704353e+00 -3.45175534e-01 3.97969067e-01 -5.23800671e-01
2.01213336e+00 8.41016024e-02 5.86148322e-01 8.82070303e-01
8.62560272e-01 4.59187105e-03 6.27556086e-01 -3.93619418e-01
5.52643299e-01 -3.57271940e-01 4.72393364e-01 6.25444114e-01
3.29034716e-01 5.97895145e-01 -5.24882913e-01 -7.94627666e-01
4.12019163e-01 -1.84307843e-01 -3.05131227e-02 -4.80191976e-01
-1.09933615e+00 9.81955349e-01 7.96359181e-01 -2.69486159e-01
-5.65433800e-01 5.32798529e-01 4.06109065e-01 -3.96670938e-01
4.79267091e-01 8.82713556e-01 -6.98598564e-01 -1.45029992e-01
-6.70187950e-01 6.51719153e-01 1.01480258e+00 3.84003371e-01
7.82258332e-01 -3.77908945e-02 -5.07478081e-02 2.57965684e-01
5.31673372e-01 3.74450177e-01 1.16061084e-02 -8.26303899e-01
1.78308234e-01 4.29749429e-01 1.55657694e-01 -6.47654593e-01
-6.29744709e-01 1.31253168e-01 -1.07058775e+00 2.09984496e-01
1.37586415e-01 -6.36109933e-02 -9.75511134e-01 1.73118341e+00
2.73372799e-01 -2.68267870e-01 1.20046556e-01 3.19552332e-01
9.52107310e-01 1.20417976e+00 4.96502191e-01 -2.74662584e-01
8.43322933e-01 -8.68554294e-01 -3.30543578e-01 4.32238877e-01
6.81677639e-01 -9.51857865e-02 3.67955983e-01 3.37326944e-01
-1.47348046e+00 -5.06834686e-01 -9.75565970e-01 -3.84621918e-01
-7.14974403e-01 -5.11776268e-01 9.56577837e-01 3.40003997e-01
-9.55578685e-01 1.73700941e+00 -1.07750404e+00 -1.16256177e-02
4.86166924e-02 1.05982447e+00 -5.31990290e-01 4.44258451e-01
-1.45744812e+00 1.08263767e+00 9.50980127e-01 -1.51755109e-01
-8.66212606e-01 -1.31590652e+00 -8.59144568e-01 2.75894105e-01
-1.65402189e-01 -9.16319311e-01 1.23509979e+00 -1.75504506e-01
-1.67147136e+00 1.44243881e-01 -4.37337667e-01 -7.66331017e-01
-3.39187905e-02 1.77056819e-01 -3.21219057e-01 1.18363658e-02
-2.24804878e-01 8.97823989e-01 3.62815261e-01 -9.64200497e-01
9.42536965e-02 -9.14701745e-02 -7.31519833e-02 -2.00218358e-03
2.82886326e-02 -1.40003890e-01 1.28662169e-01 -1.53895244e-02
1.00136146e-01 -1.15942287e+00 -7.24970162e-01 -5.71696699e-01
-6.04576647e-01 -3.91445398e-01 1.27234817e-01 -4.49261725e-01
1.13183415e+00 -1.45131123e+00 5.54634750e-01 4.39936399e-01
4.89958793e-01 -3.26418467e-02 8.04678798e-02 8.57965291e-01
-7.75751233e-01 3.69206607e-01 -2.94585288e-01 -6.89730123e-02
2.67469138e-01 -1.01512536e-01 -1.81961730e-01 3.44861716e-01
8.61112177e-02 1.28107166e+00 -7.65171170e-01 -8.00518170e-02
3.95893455e-01 8.07129860e-01 -1.00177884e+00 -1.81705020e-02
-4.98879522e-01 3.58733207e-01 -1.19755208e-01 4.73940641e-01
5.91711640e-01 -5.01157284e-01 -7.10981665e-03 -4.63182718e-01
-2.22683027e-01 6.88656151e-01 -6.09508574e-01 1.47701371e+00
3.47225135e-03 9.50386673e-02 -4.97775853e-01 -6.29079580e-01
4.98718023e-01 1.91828936e-01 7.18559325e-01 -6.01547360e-01
2.88068838e-02 7.51336440e-02 4.90701944e-01 -1.07814729e-01
7.11038828e-01 -4.66075867e-01 -9.06916410e-02 -1.01669384e-02
2.81339675e-01 -3.28704745e-01 3.97159636e-01 1.62567854e-01
1.06144178e+00 1.44120738e-01 5.59748948e-01 -5.50109386e-01
4.70579684e-01 4.52499315e-02 3.53507787e-01 8.80957127e-01
1.84146520e-02 3.73032749e-01 2.94891864e-01 -9.61978614e-01
-1.49043822e+00 -1.01857483e+00 -6.61890507e-01 9.76916134e-01
-2.04978347e-01 -9.44626570e-01 -8.13371181e-01 -1.65148914e-01
3.75718921e-01 8.02338660e-01 -8.13090324e-01 -5.71746171e-01
-3.08693171e-01 -1.31241536e+00 3.57334971e-01 2.94494122e-01
-2.65823081e-02 -1.42536592e+00 1.11832775e-01 5.91401160e-01
3.21790457e-01 -6.93576396e-01 -1.71156824e-01 7.53505886e-01
-6.22255564e-01 -8.92599285e-01 -3.16197313e-02 -2.29919136e-01
4.20113534e-01 -3.18570524e-01 1.39130306e+00 -9.90868807e-02
-3.07800800e-01 -1.73242748e-01 7.21153989e-02 -2.79839486e-01
-9.52483952e-01 1.52884901e-01 6.89507782e-01 -5.68573475e-01
4.74333793e-01 -8.01274955e-01 -6.56935036e-01 -2.40987867e-01
-5.58484614e-01 7.19777122e-02 1.98839664e-01 8.06598365e-01
6.63961709e-01 2.21276313e-01 3.32761966e-02 -5.13406396e-01
6.13660634e-01 -3.56848180e-01 -6.81209564e-01 2.56683435e-02
-7.13834822e-01 4.92613226e-01 9.43166018e-01 -1.47739291e-01
-7.39053428e-01 1.61750630e-01 -4.16428536e-01 5.75313456e-02
-2.95218289e-01 4.37771767e-01 -2.40729805e-02 -5.29331863e-01
8.34054291e-01 1.78895950e-01 -5.05858183e-01 -2.14268744e-01
4.51607376e-01 2.55177647e-01 4.18152243e-01 -1.03431773e+00
6.05366945e-01 1.13999797e-02 7.81707764e-01 -8.06832492e-01
-5.82649112e-01 -8.89286995e-02 -9.37739015e-01 3.20052892e-01
1.12167239e+00 -7.39959180e-01 -1.68732393e+00 2.52624154e-01
-1.25141227e+00 -4.24391419e-01 -2.06932858e-01 5.03447473e-01
-4.79136765e-01 4.06511188e-01 -1.04814398e+00 -7.06664145e-01
-7.94886708e-01 -1.35985363e+00 9.97553766e-01 2.70801604e-01
-3.71117383e-01 -9.89924073e-01 4.49433863e-01 2.24044055e-01
2.78863668e-01 2.48154476e-01 1.38494408e+00 -5.62216580e-01
-6.09902978e-01 -1.89816058e-01 -1.52723923e-01 8.50122645e-02
-1.33153066e-01 4.30229485e-01 -8.74092937e-01 -3.30619186e-01
-5.99396169e-01 -5.27844906e-01 1.06642377e+00 7.59968936e-01
1.42129111e+00 -3.47311467e-01 -4.93429035e-01 7.94641256e-01
1.26563191e+00 5.02577782e-01 5.42459011e-01 1.40538290e-01
1.01334894e+00 1.16263755e-01 -1.41146421e-01 2.76180536e-01
1.98097438e-01 5.38567901e-01 5.17669559e-01 8.72020349e-02
5.01829982e-01 -3.12429011e-01 3.80227208e-01 8.18410814e-01
-5.07488430e-01 -1.34875298e-01 -1.21782053e+00 -2.19896406e-01
-1.75314701e+00 -1.18202138e+00 -3.54497552e-01 2.09340453e+00
9.97391522e-01 2.50379741e-01 2.47186478e-02 -3.30049306e-01
2.20658675e-01 9.96569470e-02 -9.86910403e-01 -7.35052168e-01
2.05270663e-01 5.49576819e-01 8.06452692e-01 8.69897783e-01
-1.21233976e+00 9.82325494e-01 7.85874987e+00 8.19952965e-01
-8.71663153e-01 -2.56497055e-01 6.71726763e-01 -4.96918745e-02
-1.78030804e-01 7.56605417e-02 -9.83807147e-01 5.76002479e-01
1.60928869e+00 -3.22357446e-01 7.25913346e-01 6.98795080e-01
2.58307129e-01 1.62760735e-01 -1.33777905e+00 6.24657929e-01
-4.74796265e-01 -1.90796375e+00 4.10661817e-01 3.84359777e-01
9.41960096e-01 4.53271836e-01 -4.71043363e-02 4.56038535e-01
5.83505094e-01 -1.58329630e+00 4.42254782e-01 6.70046091e-01
6.31740749e-01 -1.03426909e+00 4.48426932e-01 2.25375518e-01
-9.84016061e-01 2.33542562e-01 -5.70570946e-01 -3.28788966e-01
3.48086134e-02 4.38540936e-01 -5.48111975e-01 5.36585391e-01
5.04810393e-01 6.04255140e-01 -2.10727870e-01 6.99432671e-01
8.03301707e-02 5.21984100e-01 -4.00097072e-01 -1.99837387e-01
3.13398421e-01 -7.21634686e-01 6.04860745e-02 1.17059469e+00
4.68717255e-02 5.56529582e-01 1.03436649e-01 1.19131112e+00
-3.84355783e-01 -4.33120094e-02 -5.18359959e-01 -2.74279177e-01
5.05624533e-01 1.19277894e+00 -4.53945428e-01 -4.02519733e-01
-2.98984032e-02 6.02962196e-01 4.48391616e-01 2.42029354e-01
-9.82946754e-01 -3.39290828e-01 8.78024220e-01 -1.15833089e-01
6.22670949e-02 -3.98880452e-01 1.46428898e-01 -9.24701929e-01
-5.76950252e-01 -7.94327557e-01 -1.98041260e-01 -7.66509175e-01
-1.22248137e+00 4.22338635e-01 7.09731504e-02 -4.92891729e-01
-1.11085556e-01 -1.08358097e+00 -6.66680396e-01 1.02354550e+00
-9.95142460e-01 -8.40511620e-01 2.12434828e-01 1.58082843e-02
7.00632203e-03 8.61591995e-02 1.36393571e+00 6.42744750e-02
-8.12533379e-01 1.83507055e-01 6.09272540e-01 -2.83224255e-01
4.73178893e-01 -1.62506139e+00 1.05842412e+00 2.19427139e-01
-1.21153593e-01 1.19438338e+00 9.16353106e-01 -7.99277842e-01
-1.53635859e+00 -1.05791283e+00 5.92666030e-01 -7.16235816e-01
8.64122033e-01 -3.99964571e-01 -1.03091180e+00 4.22583789e-01
1.12347879e-01 2.87809640e-01 9.52689707e-01 4.05596972e-01
-3.07337016e-01 4.43857849e-01 -1.10224497e+00 4.45593357e-01
1.12921870e+00 -8.64736438e-01 -5.20522177e-01 7.86360681e-01
1.07595408e+00 -5.92916548e-01 -1.41636741e+00 6.62707269e-01
6.67604923e-01 -9.08600450e-01 1.56766403e+00 -1.21485305e+00
6.19970977e-01 4.59048292e-03 -1.68456271e-01 -1.22264433e+00
-9.97288942e-01 -6.71749711e-01 -4.86923009e-01 3.61413419e-01
6.70616031e-01 -4.57907706e-01 9.42905545e-01 1.30083358e+00
-1.65157169e-01 -9.74853694e-01 -9.20071781e-01 -6.85398817e-01
7.36540198e-01 -2.82172561e-01 7.06097066e-01 6.71947777e-01
3.12039524e-01 5.63699841e-01 -2.77992308e-01 2.35383064e-01
7.24611342e-01 -1.66623190e-01 3.36171180e-01 -1.33482063e+00
-2.93425769e-01 -3.67623717e-01 -2.50216246e-01 -7.23596513e-01
5.88257670e-01 -1.12254024e+00 -2.90037338e-02 -1.36838651e+00
6.27458513e-01 -3.83793823e-02 -4.97840196e-01 6.59064531e-01
3.38769667e-02 1.80382192e-01 5.72251156e-02 6.31040037e-02
-7.14806318e-01 9.48700845e-01 8.91085744e-01 -2.17132002e-01
-3.79709542e-01 -4.50701177e-01 -3.85830700e-01 9.02988672e-01
9.01792645e-01 -6.36306703e-01 8.59470889e-02 2.66748846e-01
4.73560989e-01 1.22647658e-01 3.02878559e-01 -1.27149141e+00
2.06938714e-01 -2.17316270e-01 5.82053006e-01 -9.02876556e-01
6.29628420e-01 -3.83493006e-01 3.26145142e-01 5.82589030e-01
-5.05232930e-01 -5.11141084e-02 3.72249782e-01 4.01382029e-01
9.01188478e-02 -1.12171926e-01 7.53999889e-01 -2.91151613e-01
-1.10262200e-01 5.80183744e-01 -3.78849804e-01 -3.50197732e-01
6.31562054e-01 2.69835386e-02 -3.22419375e-01 4.59686443e-02
-7.96432376e-01 -2.15855595e-02 5.33677995e-01 3.37517224e-02
1.27756447e-01 -1.33542740e+00 -6.29321858e-02 4.99140583e-02
-7.97432587e-02 1.96879301e-02 2.78945267e-01 4.64940131e-01
-6.87371254e-01 1.10679710e+00 1.49035146e-02 -2.77383775e-01
-1.08858156e+00 6.73135936e-01 6.26758814e-01 -4.05586302e-01
-3.33943754e-01 7.63114035e-01 2.06669673e-01 -8.43964040e-01
-1.63935766e-01 -6.35336280e-01 2.96885997e-01 -1.40985057e-01
3.10119241e-01 3.11966300e-01 1.41698301e-01 -8.10352504e-01
-4.52426106e-01 4.60414410e-01 -1.95437700e-01 2.35224232e-01
1.41159940e+00 6.52426541e-01 -4.88948196e-01 2.08524600e-01
1.43630958e+00 -2.56117046e-01 -1.18283963e+00 1.39956474e-01
-1.85423717e-01 4.28690225e-01 2.15879679e-01 -7.61697412e-01
-1.42590359e-01 8.12429726e-01 3.53229940e-01 2.04039022e-01
1.98004410e-01 -3.17155004e-01 7.50527680e-01 1.24889970e+00
4.12603945e-01 -8.06359172e-01 -2.36391872e-01 1.02140367e+00
5.50438046e-01 -1.40659726e+00 2.80201524e-01 -7.32716098e-02
-7.32103884e-02 1.30586612e+00 4.18358058e-01 -1.04758414e-02
6.09823465e-01 -2.43619997e-02 -7.86149263e-01 -3.55703354e-01
-7.50006616e-01 2.44268253e-01 5.89900851e-01 1.68375447e-01
6.95970774e-01 1.83027461e-01 2.07904503e-01 4.33048427e-01
-3.37198853e-01 -1.43361734e-02 3.02638978e-01 6.22133851e-01
-5.00276268e-01 -1.27582240e+00 -2.42746562e-01 5.24649501e-01
-2.61111051e-01 -6.11341774e-01 -3.86902690e-01 6.92387462e-01
2.60764092e-01 6.64401174e-01 -1.43508881e-01 -3.56263429e-01
1.40719011e-01 4.77146029e-01 5.75265527e-01 -6.43644214e-01
-6.69428468e-01 -5.25078058e-01 8.19503143e-02 -7.48434544e-01
-2.74566680e-01 -3.76771778e-01 -1.61296582e+00 -8.27845216e-01
-2.70136625e-01 6.77475393e-01 4.71055359e-01 7.50689030e-01
4.90273952e-01 5.49454391e-01 2.62396127e-01 -1.58663106e+00
-6.89901114e-01 -9.08988535e-01 -4.73773003e-01 2.75742441e-01
4.16568577e-01 -4.61159378e-01 -3.49132657e-01 -3.95741522e-01] | [5.31348180770874, 5.450888156890869] |
3d62d8b9-1fe1-4ed6-89b7-00b0472249a7 | discriminative-appearance-modeling-with-multi | 2101.12159 | null | https://arxiv.org/abs/2101.12159v1 | https://arxiv.org/pdf/2101.12159v1.pdf | Discriminative Appearance Modeling with Multi-track Pooling for Real-time Multi-object Tracking | In multi-object tracking, the tracker maintains in its memory the appearance and motion information for each object in the scene. This memory is utilized for finding matches between tracks and detections and is updated based on the matching result. Many approaches model each target in isolation and lack the ability to use all the targets in the scene to jointly update the memory. This can be problematic when there are similar looking objects in the scene. In this paper, we solve the problem of simultaneously considering all tracks during memory updating, with only a small spatial overhead, via a novel multi-track pooling module. We additionally propose a training strategy adapted to multi-track pooling which generates hard tracking episodes online. We show that the combination of these innovations results in a strong discriminative appearance model, enabling the use of greedy data association to achieve online tracking performance. Our experiments demonstrate real-time, state-of-the-art performance on public multi-object tracking (MOT) datasets. | ['James M. Rehg', 'Mazen Alotaibi', 'Li Fuxin', 'Chanho Kim'] | 2021-01-28 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Kim_Discriminative_Appearance_Modeling_With_Multi-Track_Pooling_for_Real-Time_Multi-Object_Tracking_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Kim_Discriminative_Appearance_Modeling_With_Multi-Track_Pooling_for_Real-Time_Multi-Object_Tracking_CVPR_2021_paper.pdf | cvpr-2021-1 | ['real-time-multi-object-tracking'] | ['computer-vision'] | [-1.33815885e-01 -4.75600719e-01 -2.45792344e-01 5.47877848e-02
-7.65879929e-01 -8.24787617e-01 4.66375262e-01 2.78646559e-01
-6.10831082e-01 5.23619950e-01 -1.55586556e-01 2.06643060e-01
1.38052061e-01 -4.71648604e-01 -1.01577926e+00 -5.11538744e-01
-1.01423703e-01 5.26023924e-01 1.05933511e+00 4.19211179e-01
-3.35668065e-02 7.43472755e-01 -1.60033834e+00 2.40531802e-01
3.22544187e-01 8.96750450e-01 5.50982475e-01 8.87167394e-01
-2.29717791e-02 7.23803461e-01 -5.95572710e-01 -3.22267681e-01
3.85937303e-01 -1.64984204e-02 -2.56692231e-01 1.67981282e-01
1.11616480e+00 -3.42132181e-01 -4.95034695e-01 8.72178257e-01
4.09965426e-01 1.66907474e-01 9.31693316e-02 -1.23586476e+00
-1.77125260e-01 1.99277237e-01 -7.61273086e-01 6.05986774e-01
-1.11531345e-02 7.25815222e-02 7.10885525e-01 -7.77053654e-01
7.77839720e-01 1.08767319e+00 9.73746300e-01 4.68981355e-01
-1.16830456e+00 -7.87855864e-01 5.20827472e-01 1.94095626e-01
-1.20935714e+00 -5.18715918e-01 1.94648847e-01 -3.86533052e-01
9.01891172e-01 2.15408042e-01 7.76841581e-01 4.99119490e-01
3.84330511e-01 7.51088798e-01 6.59811497e-01 -2.46424079e-01
-1.53834045e-01 2.01041058e-01 2.86224365e-01 7.65450478e-01
6.52178109e-01 1.46593511e-01 -7.88528979e-01 -3.08487594e-01
6.13962412e-01 1.97149083e-01 5.27018905e-02 -6.05416238e-01
-1.24690413e+00 4.99303937e-01 4.62490171e-01 3.77294570e-01
-4.51357931e-01 5.20803452e-01 1.33395776e-01 2.35554338e-01
2.17149332e-01 -1.48647965e-03 -2.52012402e-01 4.35585529e-02
-1.13280189e+00 2.85386801e-01 4.74219471e-01 1.03137553e+00
8.24978352e-01 -1.57622352e-01 -4.76983160e-01 3.20564359e-01
4.67090368e-01 7.91150093e-01 -3.71491984e-02 -6.99717820e-01
1.71752796e-01 4.68489856e-01 3.69445831e-01 -7.71955013e-01
-4.57408816e-01 -6.35518551e-01 -1.25881985e-01 3.38201374e-01
4.86063868e-01 2.74005830e-02 -9.10278320e-01 1.87465572e+00
8.47540855e-01 7.93916643e-01 -3.57127756e-01 6.27634943e-01
5.68710208e-01 3.65141273e-01 2.20913887e-01 -1.75592259e-01
1.46341741e+00 -1.13338339e+00 -5.57639837e-01 -4.28865910e-01
3.14823747e-01 -9.53693628e-01 1.17000556e-02 1.86817124e-02
-1.16426933e+00 -8.45309079e-01 -9.45414484e-01 2.37853885e-01
-2.60736912e-01 3.67149085e-01 5.11205494e-01 6.43361390e-01
-1.09756744e+00 5.50807178e-01 -1.08744824e+00 -5.21032572e-01
3.35728526e-01 6.78248107e-01 -2.76116371e-01 6.05543107e-02
-4.64590222e-01 1.16595805e+00 4.13696736e-01 8.91577378e-02
-8.72713566e-01 -6.81606293e-01 -5.18724978e-01 -1.33953452e-01
4.23507810e-01 -7.58294642e-01 1.26756144e+00 -7.36096203e-01
-1.01025259e+00 7.80653000e-01 -5.60881138e-01 -6.03286147e-01
5.10610044e-01 -4.12469625e-01 -4.39071655e-01 -5.96264489e-02
1.00288250e-01 7.49444962e-01 9.47008073e-01 -1.10370874e+00
-1.23799682e+00 -3.34476948e-01 -7.64126256e-02 1.86629146e-02
-2.25688219e-01 2.74024963e-01 -1.17788303e+00 -5.59635043e-01
6.54806793e-02 -1.16521776e+00 -2.58875519e-01 3.61752898e-01
7.21216649e-02 2.25339737e-02 1.14831483e+00 -5.79245508e-01
1.07780659e+00 -2.15406322e+00 7.78850392e-02 1.04880251e-01
2.90588051e-01 3.28396112e-01 -5.30872047e-02 5.64127229e-02
4.35376853e-01 -5.76409340e-01 5.21143854e-01 -8.89056027e-01
-1.96411923e-01 1.62383929e-01 -2.11994782e-01 7.76437461e-01
1.20290458e-01 9.57634509e-01 -8.02755654e-01 -7.30436623e-01
3.27537745e-01 4.04029042e-01 -5.85984588e-01 4.62619178e-02
-2.66142488e-01 5.97224355e-01 -4.69477862e-01 5.45083344e-01
8.43568742e-01 -4.77558225e-01 1.84217453e-01 -1.89446613e-01
-5.23616493e-01 -4.89197485e-02 -1.54693949e+00 1.72574568e+00
-1.28224969e-01 6.80424869e-01 1.08371675e-01 -1.82400927e-01
4.82668400e-01 2.39707887e-01 6.84742033e-01 -4.63885218e-01
-2.66126040e-02 1.22354634e-01 -6.58538938e-02 2.05973521e-01
8.85679066e-01 1.34783581e-01 1.23651505e-01 4.55283552e-01
1.71453252e-01 7.01859653e-01 3.42111021e-01 1.84640810e-01
1.30151880e+00 1.69533610e-01 -2.71687079e-02 9.34247300e-02
2.44785428e-01 2.25266874e-01 7.41039932e-01 1.20581234e+00
-2.80256838e-01 1.87756792e-01 -4.17091519e-01 -3.52906823e-01
-8.99154842e-01 -1.22812355e+00 1.23961018e-02 1.22460115e+00
4.31113392e-01 -4.81177092e-01 -2.78441668e-01 -4.12917107e-01
2.06662014e-01 1.24695621e-01 -6.46811187e-01 2.02470329e-02
-9.11080062e-01 -5.38342297e-01 3.48696947e-01 6.27420247e-01
2.01198652e-01 -8.49031866e-01 -1.17340505e+00 5.50256312e-01
1.31545261e-01 -1.16541040e+00 -6.55851781e-01 1.35617390e-01
-8.81207168e-01 -9.64658976e-01 -4.67754960e-01 -5.27501523e-01
5.14399111e-01 6.90076828e-01 9.35422480e-01 2.11233616e-01
-4.43896055e-01 7.00833321e-01 -8.20411816e-02 -3.47863883e-01
-2.67516106e-01 -2.01118231e-01 6.74777105e-02 1.66306272e-02
2.00849727e-01 -1.18439414e-01 -3.56775433e-01 3.62332493e-01
-3.98386687e-01 -8.66417773e-03 6.28547430e-01 4.41301018e-01
7.33224213e-01 -1.65734634e-01 1.48587927e-01 -3.32210749e-01
-2.57069141e-01 -2.42258742e-01 -1.15778005e+00 4.83275086e-01
-6.10306626e-03 1.58067851e-03 -2.66676918e-02 -8.68903160e-01
-8.68984878e-01 5.85264683e-01 4.01975393e-01 -8.26898992e-01
2.09933028e-01 1.98551379e-02 1.20440565e-01 -6.04363918e-01
1.77967712e-01 1.95351914e-01 -1.54616639e-01 -6.04153812e-01
3.82499307e-01 1.14656128e-01 6.56908691e-01 -2.83720762e-01
9.59949315e-01 8.68483961e-01 6.61007389e-02 -5.73601246e-01
-7.71457851e-01 -9.50912297e-01 -8.38902712e-01 -6.96486712e-01
8.73467624e-01 -1.21256745e+00 -7.94425189e-01 3.19658041e-01
-1.28008378e+00 -2.48520579e-02 -2.27862045e-01 6.89340830e-01
-2.83154756e-01 2.77247280e-01 -5.25494158e-01 -9.34091985e-01
-2.11371303e-01 -9.68435466e-01 1.15745890e+00 4.92193609e-01
8.37885812e-02 -7.29788303e-01 3.68745327e-01 -1.14036955e-01
4.98629361e-01 -5.04211104e-03 -2.30238810e-02 -4.27053213e-01
-1.37550390e+00 -3.16602379e-01 -1.95077017e-01 -3.68483156e-01
-3.05480491e-02 -1.24890812e-01 -1.05586827e+00 -7.49767303e-01
-3.62059385e-01 9.62786302e-02 1.15379548e+00 5.69725215e-01
5.30587971e-01 2.56636888e-01 -1.01586664e+00 3.24510008e-01
1.54424095e+00 5.41233011e-02 2.09374249e-01 3.93400759e-01
7.37799585e-01 -8.39320850e-03 8.43761444e-01 5.41681945e-01
2.78754354e-01 1.21171796e+00 2.84638405e-01 6.10092469e-03
-4.94943529e-01 6.97878227e-02 5.14109075e-01 3.92851591e-01
1.34304672e-01 -1.13371648e-01 -5.88857710e-01 6.25406682e-01
-2.18423748e+00 -1.34020245e+00 -2.91573107e-01 2.54821968e+00
5.04126847e-01 2.58288622e-01 3.61656666e-01 -6.31655633e-01
8.84277880e-01 -2.30118912e-02 -4.86963511e-01 3.38522404e-01
-6.71417043e-02 1.11408293e-01 8.90754223e-01 3.72820228e-01
-1.45624220e+00 8.82624924e-01 6.53513956e+00 5.10834157e-01
-1.02808201e+00 5.06671011e-01 -3.01532149e-01 -6.00794733e-01
3.14305186e-01 7.02522993e-02 -1.60476017e+00 4.00161654e-01
9.98480737e-01 2.37813070e-02 6.19608574e-02 6.80639923e-01
-1.00573361e-01 -3.03905278e-01 -1.12279856e+00 9.19000804e-01
6.78227767e-02 -1.48285162e+00 -3.12087566e-01 1.40948251e-01
5.59842825e-01 4.89672333e-01 -4.84509431e-02 1.76663592e-01
3.61509085e-01 -3.06323767e-01 1.13579381e+00 6.20545685e-01
2.50444025e-01 -3.93646866e-01 2.90873617e-01 1.58608302e-01
-1.94907844e+00 -1.78992316e-01 -3.65323395e-01 2.18914464e-01
3.91994387e-01 2.44935110e-01 -6.00942910e-01 7.56550610e-01
8.56433094e-01 6.57053888e-01 -8.63409817e-01 1.90309131e+00
4.63171810e-01 1.20667003e-01 -6.37525141e-01 2.53254056e-01
1.36241868e-01 3.46019417e-01 6.98319256e-01 1.23839653e+00
3.34974557e-01 -4.15078312e-01 6.88684464e-01 5.54769039e-01
1.95124432e-01 -2.90215731e-01 -3.60666603e-01 1.23983234e-01
6.13388240e-01 1.30139387e+00 -1.02681792e+00 -5.31458437e-01
-8.41832340e-01 8.95979106e-01 4.71455455e-01 -4.12070705e-03
-1.26425076e+00 2.48397827e-01 8.23933303e-01 -7.74914548e-02
1.05723548e+00 -2.83897400e-01 2.98657149e-01 -9.59158897e-01
1.60693228e-01 -4.74195004e-01 6.60018146e-01 -3.77244920e-01
-1.05422044e+00 3.69487286e-01 -1.48445442e-01 -1.46922684e+00
4.54889089e-02 -4.94937271e-01 -3.80917698e-01 7.91108608e-01
-1.37480450e+00 -1.54710519e+00 -3.05952162e-01 5.83172143e-01
2.32831046e-01 4.46518920e-02 5.06898344e-01 8.98196995e-01
-4.96065021e-01 4.85806942e-01 9.15999413e-02 -6.77365959e-02
8.70057285e-01 -9.59795654e-01 4.69635814e-01 1.08003104e+00
3.81811053e-01 5.73197186e-01 8.01116765e-01 -9.92818356e-01
-1.62816668e+00 -1.32868338e+00 8.28972399e-01 -8.05195332e-01
6.80757761e-01 -3.51943284e-01 -8.86650741e-01 9.89067256e-01
8.12394097e-02 2.24377498e-01 6.08225763e-01 1.64158791e-01
-2.46560335e-01 4.86789048e-02 -8.27285588e-01 3.36686969e-01
9.91966009e-01 -2.87840128e-01 -4.93550628e-01 2.18488336e-01
4.66453880e-01 -6.63752139e-01 -6.35905921e-01 1.79563507e-01
7.75289237e-01 -5.84668458e-01 1.09901381e+00 -4.93832827e-01
-6.99151218e-01 -9.50857222e-01 -4.99234982e-02 -7.40484416e-01
-6.78452730e-01 -3.90212446e-01 -5.50030768e-01 1.08511364e+00
1.69034109e-01 -3.84567022e-01 8.62402439e-01 3.29128981e-01
-1.83691218e-01 -1.12669230e-01 -1.16215587e+00 -9.68238175e-01
-6.04016542e-01 -2.35256240e-01 2.51229525e-01 4.87357944e-01
-4.70133454e-01 2.39001010e-02 -6.35863602e-01 7.78692424e-01
9.59665358e-01 6.41960800e-01 1.08260202e+00 -1.40999889e+00
-7.34669209e-01 -2.83651263e-01 -5.64349174e-01 -1.20064461e+00
-5.69973178e-02 -6.92822635e-01 2.99863279e-01 -1.34756124e+00
7.00203896e-01 -5.80887020e-01 -4.95595694e-01 4.91971880e-01
-2.82693118e-01 4.23357576e-01 6.49462521e-01 4.18273300e-01
-1.34940565e+00 8.30252245e-02 7.97077179e-01 -1.19897209e-01
-1.25062242e-01 1.16098203e-01 -1.54540271e-01 2.87814677e-01
3.15908760e-01 -8.80515039e-01 1.50853485e-01 -4.68847603e-01
-2.56118923e-01 -1.82185397e-01 7.15810478e-01 -1.56749690e+00
7.51974761e-01 2.29957536e-01 7.29799092e-01 -1.22572076e+00
8.12713921e-01 -1.03605890e+00 7.82901466e-01 7.86275923e-01
1.69542477e-01 2.92443186e-01 8.73395681e-01 8.80184472e-01
7.97795057e-02 -1.00554526e-01 7.90924489e-01 3.75746936e-02
-1.03815353e+00 4.50898618e-01 -9.98542309e-02 -4.84808832e-01
1.21235406e+00 -3.44611883e-01 -2.85677463e-01 1.49782166e-01
-8.11591804e-01 5.15348077e-01 6.46943510e-01 5.77316046e-01
1.73510358e-01 -1.40998232e+00 -4.91331488e-01 4.27170396e-02
6.59137592e-02 -5.88624656e-01 4.03684169e-01 1.07248688e+00
2.07958296e-02 5.12810349e-01 -3.15364808e-01 -8.90999615e-01
-1.82744753e+00 7.19539583e-01 3.73808116e-01 -3.98506969e-01
-8.02639246e-01 7.61646926e-01 1.56589344e-01 1.06666863e-01
4.04820561e-01 -1.94036543e-01 1.72714308e-01 5.71106337e-02
8.69450092e-01 1.63560435e-01 5.08173555e-02 -8.16991746e-01
-6.12053156e-01 7.44528592e-01 -2.63745576e-01 -1.42281070e-01
1.05248976e+00 -1.29957065e-01 1.09074950e-01 4.44728851e-01
5.70984542e-01 2.84671456e-01 -1.51469719e+00 -4.59912896e-01
1.43948659e-01 -7.81096399e-01 8.24436173e-02 -4.89931315e-01
-1.00251722e+00 2.90943831e-01 1.03832924e+00 -1.01465143e-01
7.69357681e-01 2.65510648e-01 6.42268419e-01 3.12057376e-01
7.60239840e-01 -8.85783374e-01 -2.34807748e-02 3.94559890e-01
2.67968684e-01 -1.11153066e+00 1.21105410e-01 -2.36919999e-01
-1.96447372e-01 8.19042027e-01 6.22309864e-01 3.65714878e-02
5.99760771e-01 5.06516576e-01 -6.68310225e-02 -2.30819941e-01
-8.92817974e-01 -6.09855711e-01 3.21662545e-01 4.26170319e-01
1.28068849e-01 -2.31721863e-01 2.30244517e-01 -7.48139918e-02
5.23719311e-01 -1.28323538e-02 -4.88844104e-02 1.41498470e+00
-8.16166878e-01 -1.25049436e+00 -7.03625500e-01 3.31784844e-01
-3.91466856e-01 2.27548763e-01 -1.43243298e-01 7.52414227e-01
2.89391905e-01 7.47470915e-01 4.15434450e-01 -1.70021221e-01
3.65699738e-01 -7.66417608e-02 1.03974974e+00 -5.97225606e-01
-9.41982746e-01 3.20219994e-01 -3.44597735e-02 -7.66961038e-01
-5.56693673e-01 -1.02048647e+00 -1.17174256e+00 -3.62524271e-01
-6.01518273e-01 -1.20964907e-01 5.28601229e-01 7.47383416e-01
5.54401219e-01 6.78104877e-01 2.77477503e-02 -1.04577529e+00
-2.25289270e-01 -6.75641716e-01 -4.11794662e-01 2.29041293e-01
5.26223361e-01 -1.15422010e+00 2.65863806e-01 -1.13118009e-03] | [6.378493309020996, -2.0497119426727295] |
8a2a1c76-0e6e-44b6-9691-1ca827958ef7 | no-spare-parts-sharing-part-detectors-for | 1510.04908 | null | http://arxiv.org/abs/1510.04908v2 | http://arxiv.org/pdf/1510.04908v2.pdf | No Spare Parts: Sharing Part Detectors for Image Categorization | This work aims for image categorization using a representation of distinctive
parts. Different from existing part-based work, we argue that parts are
naturally shared between image categories and should be modeled as such. We
motivate our approach with a quantitative and qualitative analysis by
backtracking where selected parts come from. Our analysis shows that in
addition to the category parts defining the class, the parts coming from the
background context and parts from other image categories improve categorization
performance. Part selection should not be done separately for each category,
but instead be shared and optimized over all categories. To incorporate part
sharing between categories, we present an algorithm based on AdaBoost to
jointly optimize part sharing and selection, as well as fusion with the global
image representation. We achieve results competitive to the state-of-the-art on
object, scene, and action categories, further improving over deep convolutional
neural networks. | ['Cees G. M. Snoek', 'Jan C. van Gemert', 'Pascal Mettes'] | 2015-10-16 | null | null | null | null | ['image-categorization'] | ['computer-vision'] | [ 3.50774050e-01 2.06112385e-01 -2.81167418e-01 -6.58363283e-01
-4.26618785e-01 -7.19747484e-01 8.80703151e-01 4.58139509e-01
-2.32448488e-01 4.36068289e-02 5.00218689e-01 2.78787047e-01
-2.06192359e-01 -6.21048868e-01 -5.50388277e-01 -5.88582933e-01
1.31632537e-01 4.76642966e-01 5.17896116e-01 -2.52730101e-01
2.88933665e-01 6.66482091e-01 -2.08015513e+00 9.25686538e-01
3.03572327e-01 1.27607930e+00 -1.08793914e-01 3.25363606e-01
-5.44572413e-01 1.36520624e+00 -6.01577997e-01 -2.45520890e-01
4.17882085e-01 -5.56710005e-01 -1.27062762e+00 8.69102240e-01
8.78574073e-01 -2.19503045e-01 -4.07977635e-03 1.10257220e+00
1.11268297e-01 2.51846135e-01 8.83179665e-01 -1.55315924e+00
-4.12117302e-01 8.38236272e-01 -4.24742222e-01 -9.53162611e-02
1.82436153e-01 5.78548722e-02 1.01268291e+00 -6.43611968e-01
7.28385329e-01 1.60199678e+00 7.01985717e-01 5.91204464e-01
-1.31861365e+00 -2.14499727e-01 8.24981511e-01 2.14715749e-01
-1.24003065e+00 -4.88041610e-01 7.87493110e-01 -7.78029621e-01
9.57897425e-01 3.83710295e-01 7.03788877e-01 5.20996153e-01
-9.32376310e-02 1.23458517e+00 1.16975677e+00 -5.10639131e-01
3.28414768e-01 3.11402142e-01 7.08500206e-01 5.19708276e-01
1.49080679e-01 -3.12982678e-01 -2.39793703e-01 -4.51949686e-02
3.70096654e-01 2.95080930e-01 7.66086280e-02 -1.14122605e+00
-1.13805807e+00 7.32291937e-01 8.15073967e-01 4.42847729e-01
-5.06283581e-01 5.81978202e-01 5.27760267e-01 6.94729090e-02
3.91489595e-01 1.70119107e-01 -6.59806669e-01 2.93809026e-01
-8.22857857e-01 1.98910981e-01 5.90199411e-01 8.96515965e-01
1.17043495e+00 -3.58058423e-01 -3.81664515e-01 1.08521366e+00
2.99046189e-01 1.30063340e-01 5.02098501e-01 -1.40807438e+00
-1.53316841e-01 1.20707631e+00 -2.34122008e-01 -6.50500298e-01
-4.35187250e-01 -2.03547478e-01 -4.82567668e-01 5.14920115e-01
4.31542277e-01 3.43934417e-01 -1.28432989e+00 1.64862728e+00
3.84665757e-01 -5.43491483e-01 -3.54910076e-01 5.80434203e-01
9.96614099e-01 -1.54421598e-01 5.04313469e-01 3.05592656e-01
1.64894879e+00 -1.12875056e+00 -3.85101229e-01 -4.50766295e-01
7.17542350e-01 -5.85474193e-01 4.71993744e-01 4.15415376e-01
-8.60988796e-01 -9.62209105e-01 -7.24458516e-01 -6.11969754e-02
-8.27737629e-01 8.22990090e-02 1.03242755e+00 7.12157309e-01
-1.21707749e+00 6.01790249e-01 -5.21986187e-01 -7.93124139e-01
9.94547844e-01 2.80979753e-01 -5.80328286e-01 -9.28030610e-02
-5.19417882e-01 9.04163301e-01 5.40123999e-01 -3.56424451e-01
-9.16947305e-01 -5.26969194e-01 -8.55242729e-01 -1.04449026e-01
2.67699420e-01 -7.87647069e-01 1.29465246e+00 -1.73929214e+00
-9.87679780e-01 1.30816925e+00 -7.67053664e-02 -5.38246095e-01
1.71831578e-01 7.71065010e-03 1.46395797e-02 3.57711256e-01
2.01690719e-01 1.38301742e+00 7.27310419e-01 -1.68720043e+00
-1.02247596e+00 -3.66565287e-01 4.84073609e-01 3.18881363e-01
-8.69258940e-02 2.44893715e-01 -3.84309441e-01 -4.44939107e-01
2.95314908e-01 -8.39740753e-01 -3.09044749e-01 5.06520748e-01
-1.95902273e-01 -3.68425906e-01 6.58464372e-01 -3.36839020e-01
7.95093119e-01 -2.12563682e+00 7.39506707e-02 -1.27459124e-01
2.63658643e-01 -1.55404136e-01 -1.77852124e-01 2.03649789e-01
-2.98599720e-01 1.40902430e-01 -1.86978191e-01 -2.17982784e-01
9.81076583e-02 3.00976127e-01 7.57366940e-02 6.01856589e-01
3.26056868e-01 8.34118903e-01 -6.83943331e-01 -6.85445130e-01
5.83076239e-01 2.41723135e-01 -4.69719946e-01 -2.18798310e-01
-1.24783643e-01 -2.45081961e-01 -2.59195238e-01 9.60915983e-01
8.19966316e-01 -1.27850562e-01 -1.33527610e-02 -3.46048206e-01
5.72272204e-02 1.29643112e-01 -1.33449674e+00 1.55684078e+00
-1.67935535e-01 4.42259043e-01 2.92477220e-01 -1.31598413e+00
6.02394402e-01 -5.96044585e-02 7.90255845e-01 -5.42837441e-01
3.27432603e-01 -5.35900928e-02 1.77013338e-01 -1.11551866e-01
3.26529890e-01 -1.25846595e-01 -1.14621498e-01 5.05356967e-01
5.22134602e-01 -4.15639043e-01 3.62506241e-01 3.37090492e-01
1.06977832e+00 4.49353367e-01 6.11206949e-01 -5.18487751e-01
5.87245643e-01 5.25693119e-01 2.15592459e-01 8.95990014e-01
-6.93471789e-01 7.54398882e-01 3.12574089e-01 -3.83485287e-01
-7.47595310e-01 -7.53235400e-01 -3.09681863e-01 1.58816683e+00
5.13593614e-01 -2.45495424e-01 -9.15273845e-01 -1.12198412e+00
2.54568607e-01 4.85123545e-01 -1.08299983e+00 -1.98391750e-01
-3.80135179e-01 -5.20740986e-01 1.17865570e-01 8.36138070e-01
5.06391943e-01 -1.19785297e+00 -7.40506053e-01 2.15870619e-01
-8.60338360e-02 -8.15375984e-01 -1.49574712e-01 7.31508791e-01
-8.19564223e-01 -1.35732019e+00 -4.50944036e-01 -4.79306906e-01
6.17630839e-01 7.58922458e-01 1.40939116e+00 1.91061705e-01
-4.60085481e-01 9.78673637e-01 -6.25413060e-01 -5.99037826e-01
-4.38036889e-01 -4.21155035e-01 -3.95363510e-01 1.42742202e-01
5.98568797e-01 -6.81604370e-02 -5.76226234e-01 5.09521663e-01
-7.80923486e-01 1.13932483e-01 7.41083801e-01 6.37194753e-01
6.56184673e-01 2.68458486e-01 -1.05994523e-01 -7.92569757e-01
-1.87489107e-01 -3.61456305e-01 -1.06739447e-01 3.44651014e-01
-1.32324964e-01 -1.30098686e-01 1.51091620e-01 -5.27968645e-01
-1.09518063e+00 6.30017161e-01 2.25588024e-01 -3.21778953e-01
-8.42104673e-01 -1.71228915e-01 -3.62260222e-01 -3.10887665e-01
8.25984955e-01 2.30951142e-02 7.79691562e-02 -5.98043144e-01
7.34587431e-01 7.23237455e-01 1.45053446e-01 -4.32041287e-01
2.90325940e-01 1.01571429e+00 -1.21087983e-01 -6.76084220e-01
-8.61658573e-01 -1.03807080e+00 -1.20740783e+00 -4.77392793e-01
8.99367750e-01 -1.14227974e+00 -3.02547038e-01 4.69967663e-01
-9.48677361e-01 -3.48234951e-01 -8.63802314e-01 8.43447372e-02
-6.77443326e-01 3.76892060e-01 -3.97311956e-01 -7.64421642e-01
7.48685151e-02 -1.01919711e+00 1.42739272e+00 8.04922357e-02
-2.88934499e-01 -8.02648604e-01 -3.56156975e-01 4.35989380e-01
3.09130996e-01 1.79139465e-01 5.69417655e-01 -9.68449354e-01
-4.94537145e-01 -3.43065619e-01 -4.66923773e-01 4.85215366e-01
2.41706371e-01 -5.46942949e-02 -1.23904610e+00 9.01252404e-02
-5.75089492e-02 -3.65147442e-01 1.37605059e+00 5.13000429e-01
1.13750005e+00 -7.08768591e-02 -5.85384965e-01 1.74332365e-01
1.48247206e+00 3.72841060e-02 5.42073429e-01 5.04795372e-01
7.66214073e-01 1.12948442e+00 6.21040642e-01 2.65126318e-01
1.95334718e-01 5.58367312e-01 6.24785900e-01 -1.30073652e-01
-9.39467430e-01 1.86646909e-01 1.44444123e-01 5.40302545e-02
1.94576323e-01 -5.44757992e-02 -7.39853263e-01 8.26427162e-01
-1.90175009e+00 -1.18597496e+00 -3.40883285e-01 1.83172238e+00
5.66389501e-01 1.04194589e-01 5.11615276e-01 2.38072231e-01
7.26260841e-01 -1.56925753e-01 -3.76713008e-01 -2.96868116e-01
-2.52429873e-01 1.55178055e-01 5.63279212e-01 2.54923940e-01
-1.62904108e+00 7.97572434e-01 7.42814589e+00 1.05019724e+00
-7.41007686e-01 1.33602604e-01 7.25580454e-01 2.83164054e-01
-5.59451059e-02 3.30973029e-01 -6.70668423e-01 1.05346985e-01
3.96826863e-01 3.23253930e-01 1.00314766e-01 1.27463222e+00
-3.61588538e-01 -3.49852115e-01 -1.22164130e+00 6.85384452e-01
3.45049888e-01 -9.46166575e-01 1.09005265e-01 1.24357276e-01
7.01414287e-01 3.98084037e-02 -4.74787414e-01 3.20926607e-01
6.40120864e-01 -7.32513964e-01 1.37739527e+00 5.11541009e-01
1.13970183e-01 -3.14281195e-01 7.09574938e-01 4.89664041e-02
-1.36454284e+00 -4.65184510e-01 -4.34715062e-01 -1.86037645e-01
-3.51731062e-01 4.23584580e-01 -3.37575108e-01 4.62156385e-01
8.45470905e-01 8.33721042e-01 -1.23664272e+00 1.11035597e+00
8.30480978e-02 3.92680049e-01 -2.26559594e-01 1.43621176e-01
1.25377953e-01 2.42575660e-01 1.08269922e-01 1.40616763e+00
-2.61688530e-01 -6.44348422e-03 6.31765127e-01 7.49131560e-01
2.51877427e-01 -1.98650938e-02 -3.87762547e-01 4.13491018e-02
6.35685697e-02 1.49358559e+00 -1.33578289e+00 -8.40506613e-01
-5.81954539e-01 9.41030979e-01 1.15402982e-01 6.87773377e-02
-5.13997674e-01 -3.65790218e-01 8.60661447e-01 1.45742342e-01
5.14132261e-01 2.33925432e-01 -5.33967614e-01 -1.02251017e+00
-3.65059674e-01 -7.67870724e-01 6.42599344e-01 -7.93291807e-01
-1.45997071e+00 4.16827232e-01 5.09430766e-01 -1.46816373e+00
-6.12916949e-04 -9.73165214e-01 -2.64943838e-01 4.60364997e-01
-1.13817835e+00 -1.65496492e+00 -4.83401150e-01 5.24457753e-01
6.60582125e-01 1.25985378e-02 7.24773169e-01 -3.15526798e-02
-2.46385604e-01 1.48894250e-01 -2.37955883e-01 2.49705806e-01
4.67220724e-01 -1.37082314e+00 4.95212711e-02 7.68900692e-01
3.27652663e-01 5.87120414e-01 6.43914580e-01 -3.29418898e-01
-8.22291434e-01 -9.68516469e-01 5.52184880e-01 -7.25418508e-01
5.49657583e-01 -3.63846302e-01 -6.58152163e-01 4.34730142e-01
3.37071568e-01 2.91087121e-01 5.26500344e-01 2.42204487e-01
-7.92262852e-01 -1.15445293e-01 -1.35648894e+00 3.36684585e-01
1.33498299e+00 -4.38948750e-01 -8.11319232e-01 3.53392124e-01
3.73710632e-01 2.59817131e-02 -5.80068111e-01 4.28412706e-01
5.43719828e-01 -1.28617811e+00 1.25913751e+00 -5.73079348e-01
2.94648260e-01 -5.11157334e-01 -5.15637279e-01 -9.52187002e-01
-7.03823566e-01 1.58110663e-01 4.21619695e-03 1.39507806e+00
-1.26867265e-01 -2.25554258e-01 7.50246227e-01 6.20988309e-01
-2.82363772e-01 -3.64374220e-01 -7.08392084e-01 -8.41060340e-01
2.07074523e-01 -4.80895042e-01 5.55916846e-01 5.86962640e-01
-1.91247717e-01 2.48051882e-01 1.65922627e-01 -3.51331234e-01
6.72701836e-01 5.46741784e-01 7.61045456e-01 -1.36942172e+00
-2.72658259e-01 -1.05238366e+00 -9.55565333e-01 -6.23379946e-01
9.38012898e-02 -8.44881475e-01 4.49084222e-01 -2.01110101e+00
6.88882887e-01 -3.50842893e-01 -4.96764779e-01 1.13962245e+00
-1.19939469e-01 7.18123257e-01 3.95073771e-01 2.85058081e-01
-9.46930826e-01 2.11349443e-01 9.33486521e-01 -6.36873245e-01
1.74198136e-01 -3.00481051e-01 -1.22781587e+00 9.02163625e-01
5.33523142e-01 -4.06127065e-01 -1.94170833e-01 -8.49904940e-02
-3.66766185e-01 -5.90717018e-01 5.42510867e-01 -1.33257496e+00
-6.02377802e-02 -2.97155172e-01 7.40712345e-01 -5.19321382e-01
4.36266512e-01 -1.02047253e+00 9.96184349e-02 5.46090901e-01
-4.53030884e-01 -7.28877187e-01 1.23480238e-01 4.23043996e-01
-2.91251093e-01 -3.06527376e-01 1.12406421e+00 -5.57324767e-01
-1.27090049e+00 -5.82084097e-02 -5.18162668e-01 -2.22659558e-01
1.06407416e+00 -6.25657678e-01 -3.46629709e-01 -1.51548654e-01
-8.83028209e-01 2.08063629e-02 7.32002378e-01 4.52577114e-01
1.38224229e-01 -1.30060756e+00 -4.18048620e-01 -1.38704643e-01
6.29191875e-01 -4.47688639e-01 3.92728478e-01 6.44649267e-01
-1.71889633e-01 3.46662849e-01 -4.17604893e-01 -7.63689339e-01
-1.44083345e+00 9.92406011e-01 5.32857776e-01 1.48671359e-01
-3.65207732e-01 1.05119085e+00 6.38836443e-01 -5.54113686e-01
2.86141306e-01 -4.00145322e-01 -4.95392025e-01 4.88557965e-01
4.30603385e-01 1.29027486e-01 1.11911014e-01 -1.14413047e+00
-8.15599620e-01 7.90212035e-01 1.43548608e-01 1.14688396e-01
1.03235281e+00 -3.13227981e-01 -4.48587120e-01 4.03845787e-01
1.21623015e+00 -1.27023458e-01 -1.10469687e+00 -2.26537079e-01
-2.25329734e-02 -4.67619926e-01 2.51643639e-02 -1.09975111e+00
-9.67226326e-01 7.45103598e-01 8.44863117e-01 3.16864133e-01
1.21165252e+00 5.03279746e-01 1.91143733e-02 2.50767082e-01
5.36322474e-01 -1.17103434e+00 1.82960421e-01 3.49415600e-01
8.53080451e-01 -1.34677017e+00 3.96484375e-01 -5.35994530e-01
-8.13750505e-01 1.05164850e+00 6.14568830e-01 -3.98601830e-01
8.96001160e-01 -1.71658653e-03 1.28177509e-01 -2.89183050e-01
-6.04544759e-01 -9.84582841e-01 7.25380063e-01 8.38827252e-01
5.08002460e-01 9.77173820e-02 -1.45487234e-01 4.89671707e-01
3.30111921e-01 -2.77912259e-01 1.85711846e-01 1.24316609e+00
-8.09097350e-01 -1.24728692e+00 -4.37283367e-01 6.60565674e-01
-3.14207256e-01 2.70732194e-01 -1.01984859e+00 8.75127673e-01
9.12489951e-01 1.11730981e+00 2.67092377e-01 -3.02015126e-01
2.96842486e-01 1.58422202e-01 7.35020220e-01 -8.41917276e-01
-1.01963449e+00 1.09009773e-01 1.30023330e-01 -8.26720834e-01
-9.89821255e-01 -8.83152187e-01 -1.15469539e+00 1.10050030e-01
-4.90716606e-01 -1.92537650e-01 6.75151825e-01 8.66738260e-01
1.30958885e-01 6.56205297e-01 4.22391325e-01 -1.15484166e+00
-1.36878669e-01 -1.02406049e+00 -6.32616043e-01 8.35785449e-01
1.97656244e-01 -9.74294424e-01 -5.13451636e-01 2.16381654e-01] | [9.500438690185547, 1.13686203956604] |
84a6648c-f71d-43be-aa73-0d0b9bf7e4ac | bandit-social-learning-exploration-under | 2302.07425 | null | https://arxiv.org/abs/2302.07425v3 | https://arxiv.org/pdf/2302.07425v3.pdf | Bandit Social Learning: Exploration under Myopic Behavior | We study social learning dynamics where the agents collectively follow a simple multi-armed bandit protocol. Agents arrive sequentially, choose arms and receive associated rewards. Each agent observes the full history (arms and rewards) of the previous agents, and there are no private signals. While collectively the agents face exploration-exploitation tradeoff, each agent acts myopically, without regards to exploration. Motivating scenarios concern reviews and ratings on online platforms. We allow a wide range of myopic behaviors that are consistent with (parameterized) confidence intervals, including the "unbiased" behavior as well as various behaviorial biases. While extreme versions of these behaviors correspond to well-known bandit algorithms, we prove that more moderate versions lead to stark exploration failures, and consequently to regret rates that are linear in the number of agents. We provide matching upper bounds on regret by analyzing "moderately optimistic" agents. As a special case of independent interest, we obtain a general result on failure of the greedy algorithm in multi-armed bandits. This is the first such result in the literature, to the best of our knowledge. | ['Aleksandrs Slivkins', 'Suho Shin', 'Mohammadtaghi Hajiaghayi', 'Kiarash Banihashem'] | 2023-02-15 | null | null | null | null | ['multi-armed-bandits'] | ['miscellaneous'] | [-2.33320206e-01 5.74667156e-01 -8.27616155e-01 -1.67425215e-01
-7.55426526e-01 -1.01876867e+00 3.01515162e-01 6.08201362e-02
-6.08589470e-01 1.23974180e+00 4.00831997e-02 -4.16360736e-01
-6.07461989e-01 -6.28462553e-01 -8.82611394e-01 -9.24337745e-01
-2.98486084e-01 8.98989201e-01 -3.34428310e-01 -1.41452253e-01
2.59153843e-01 2.75793314e-01 -1.15661275e+00 -1.60427913e-01
6.70533776e-01 1.18510163e+00 -3.47440183e-01 8.21966887e-01
5.30462414e-02 8.90576899e-01 -4.68023807e-01 -6.85699046e-01
7.39453733e-01 -2.20548511e-01 -6.40304983e-01 3.94049138e-01
-2.49199718e-01 -6.57879770e-01 -6.15259819e-02 1.39863348e+00
1.65231988e-01 8.74751434e-02 4.04632837e-01 -1.31004298e+00
-9.10499096e-01 1.43221867e+00 -1.14658904e+00 2.20576286e-01
1.46180347e-01 -6.45531192e-02 1.27770972e+00 1.47786215e-01
3.98417413e-01 1.19515193e+00 2.12682351e-01 5.03959477e-01
-1.20843935e+00 -3.47867996e-01 6.70936227e-01 -2.57292807e-01
-4.02025402e-01 -3.52380306e-01 5.90415061e-01 -1.69328213e-01
2.23804981e-01 4.86612260e-01 8.16505075e-01 8.73745799e-01
-2.61353165e-01 1.13029218e+00 1.39074659e+00 -5.10946989e-01
7.48012960e-01 4.15931880e-01 5.19005358e-01 3.18764269e-01
9.19508398e-01 2.07439154e-01 -4.68844563e-01 -5.57026982e-01
5.22529662e-01 2.47271761e-01 -1.35721743e-01 -5.10933936e-01
-9.37248766e-01 9.07927275e-01 1.43543348e-01 -1.58306941e-01
-8.94286275e-01 4.54927474e-01 -2.79870294e-02 9.06388164e-01
5.85921466e-01 2.66525894e-01 -3.44867796e-01 -8.99396390e-02
-4.31591213e-01 4.15775865e-01 1.27063072e+00 9.56823051e-01
7.51697302e-01 -1.56665325e-01 -2.87291288e-01 5.25439024e-01
9.20341760e-02 7.34443307e-01 8.68486613e-02 -1.43824756e+00
7.33030975e-01 2.01888934e-01 1.27377665e+00 -4.40428048e-01
-4.27038401e-01 -8.03357542e-01 -4.08430696e-01 4.50438768e-01
7.81238317e-01 -7.77412295e-01 -3.88722181e-01 1.96621442e+00
3.44924271e-01 -4.48496640e-01 1.34753868e-01 9.79072392e-01
-2.14654177e-01 3.05838585e-01 -3.70724887e-01 -9.55159426e-01
1.03548098e+00 -9.50877845e-01 -6.67390585e-01 -3.24501693e-01
4.09142554e-01 -4.43264693e-01 4.94761527e-01 7.47982085e-01
-1.60647166e+00 5.15845478e-01 -7.44876504e-01 5.68796039e-01
2.13648211e-02 -1.83293596e-01 9.35118318e-01 1.03298461e+00
-9.63436902e-01 5.33875287e-01 -6.54925048e-01 -4.16348837e-02
3.77650440e-01 1.97546303e-01 4.32397664e-01 2.30021954e-01
-6.59242928e-01 5.86300075e-01 -9.12644193e-02 1.53820649e-01
-9.55631733e-01 -4.59087849e-01 1.70481447e-02 -1.68335601e-03
1.02176607e+00 -7.23569453e-01 1.97953761e+00 -1.42150879e+00
-1.68835723e+00 5.91588438e-01 1.22761540e-02 -6.61404490e-01
1.16239309e+00 -2.67267942e-01 1.94811463e-01 3.36871222e-02
1.80105135e-01 -2.34685969e-02 5.32496512e-01 -1.24084997e+00
-7.82398224e-01 -6.54228330e-01 6.41794503e-01 4.71861482e-01
-3.23355585e-01 -5.93559593e-02 2.17191890e-01 -2.46699333e-01
-1.09236091e-01 -1.26269615e+00 -5.09313047e-01 -1.98182970e-01
-6.22556090e-01 5.89684863e-03 -2.44657263e-01 4.76254858e-02
9.30279016e-01 -1.70488107e+00 -6.03874102e-02 4.37849313e-01
7.70497099e-02 -4.62274343e-01 3.23100910e-02 4.88894045e-01
2.99193859e-01 3.58876258e-01 2.60079592e-01 -5.39461792e-01
5.10139108e-01 4.60242555e-02 -6.18662357e-01 8.33491802e-01
-7.30975449e-01 7.65733004e-01 -7.25620747e-01 2.85143178e-04
-4.18883622e-01 -6.26207948e-01 -5.23348510e-01 1.80609506e-02
-4.49614316e-01 6.16592541e-02 -1.09427667e+00 4.74029064e-01
5.99827409e-01 -6.39615715e-01 2.59636819e-01 6.28889084e-01
-2.86199391e-01 9.57732871e-02 -1.21655679e+00 9.44378316e-01
-1.47320509e-01 8.34911093e-02 6.10678494e-01 -9.61502850e-01
1.90758198e-01 2.67955273e-01 2.47000203e-01 -1.13994807e-01
4.03255790e-01 4.25073653e-01 -1.03151888e-01 -3.28429699e-01
5.34896970e-01 -1.96919277e-01 -8.61248560e-03 1.34578168e+00
-6.05054200e-01 2.15807393e-01 2.25158617e-01 3.49489152e-01
8.05907369e-01 -4.01254535e-01 3.25084031e-01 -3.86575431e-01
-1.64216802e-01 1.84462041e-01 3.87806714e-01 1.61215627e+00
-1.74302056e-01 -1.48138367e-02 9.63922381e-01 -3.21019769e-01
-9.88009036e-01 -7.46647954e-01 3.21007460e-01 1.40442276e+00
3.99821341e-01 3.92710030e-01 -6.68946862e-01 -2.78746784e-01
5.02216160e-01 8.35246325e-01 -1.14651000e+00 3.82555038e-01
2.79313605e-02 -1.00285208e+00 -1.08275346e-01 1.61874190e-01
3.62972856e-01 -8.03826928e-01 -7.08841324e-01 1.87561691e-01
1.73641711e-01 -7.49801397e-01 -5.47691941e-01 3.55570838e-02
-8.83227527e-01 -1.12008929e+00 -8.16109180e-01 7.71267805e-05
6.07282102e-01 5.45291960e-01 8.22356641e-01 -3.16552445e-02
2.59151429e-01 6.24377549e-01 -2.36221582e-01 -7.31633425e-01
-1.92340761e-01 1.95142720e-02 1.55302703e-01 1.45569965e-01
1.98373318e-01 -3.27213585e-01 -1.00803125e+00 2.92671084e-01
-6.54525280e-01 -1.65421262e-01 5.00486255e-01 4.94679987e-01
3.30544591e-01 -3.16030562e-01 8.01183999e-01 -1.01553118e+00
8.64342451e-01 -7.44819522e-01 -1.19553041e+00 4.54265207e-01
-6.13889694e-01 -3.50539424e-02 5.59494913e-01 -6.14970028e-01
-1.20750964e+00 -3.07356626e-01 8.47581208e-01 -3.59830237e-03
1.99718177e-01 4.51275885e-01 4.24831450e-01 -4.96587669e-03
6.64030194e-01 -2.80485414e-02 2.35148564e-01 -5.51801383e-01
3.51483881e-01 6.73878372e-01 7.29988664e-02 -8.92034531e-01
4.59770441e-01 8.55012178e-01 8.53889659e-02 -4.38675582e-01
-1.21259320e+00 4.22874428e-02 4.66283888e-01 -1.49704993e-01
2.21267670e-01 -8.07802081e-01 -1.42488909e+00 3.95159006e-01
-9.17688608e-01 -3.89442921e-01 -4.43586588e-01 7.34786093e-01
-9.86245334e-01 2.10537881e-01 -6.46873713e-01 -1.81283236e+00
-8.64937827e-02 -8.43727350e-01 3.19257975e-01 5.27310073e-01
2.56895781e-01 -7.12210596e-01 1.10074639e-01 3.85396063e-01
4.64119315e-01 -1.70770556e-01 2.64325976e-01 -8.12604666e-01
-9.71773207e-01 -1.25955611e-01 5.77061363e-02 -1.28370911e-01
-1.72766164e-01 -1.27537683e-01 -6.29891098e-01 -4.40201581e-01
1.40786216e-01 -5.48493385e-01 7.11882651e-01 9.83200729e-01
8.11539471e-01 -9.28268075e-01 -4.19655055e-01 1.06401749e-01
1.11067510e+00 2.88619608e-01 -3.70149016e-02 6.12923145e-01
-1.25313148e-01 6.12150967e-01 5.39755523e-01 1.15806103e+00
2.75666982e-01 3.22416812e-01 7.78653204e-01 3.62064809e-01
8.36598694e-01 -8.74553844e-02 3.39641690e-01 -9.60628912e-02
-1.74143985e-01 -3.68142486e-01 -3.18031520e-01 7.56507814e-01
-2.18750477e+00 -1.04134965e+00 2.76823342e-01 2.49394584e+00
8.95613909e-01 1.72868431e-01 7.74564683e-01 -2.91825235e-01
9.76177812e-01 2.49152780e-02 -1.22163844e+00 -4.07098919e-01
-2.90198892e-01 -5.12318134e-01 1.18750811e+00 7.22337306e-01
-6.77798986e-01 4.22509104e-01 6.91225100e+00 5.58726907e-01
-5.20985365e-01 2.77312309e-01 1.13160002e+00 -1.09865046e+00
-5.93665302e-01 1.58017352e-01 -8.18801880e-01 3.69532526e-01
7.02899039e-01 -5.27528942e-01 1.03886092e+00 1.11471701e+00
2.17053354e-01 -3.89951885e-01 -1.08492947e+00 7.25121319e-01
-4.17570919e-01 -1.39291275e+00 -5.99055588e-01 4.32718039e-01
1.32809150e+00 4.56449762e-02 3.35017502e-01 -1.47418395e-01
1.36357999e+00 -5.57508349e-01 1.08896685e+00 4.20631140e-01
1.54273644e-01 -9.43222821e-01 3.14804971e-01 6.13346577e-01
-2.96708047e-01 -8.01122367e-01 -3.58649403e-01 -5.44265449e-01
3.38930845e-01 6.86408222e-01 -1.40793428e-01 2.54856080e-01
5.12084723e-01 2.43220583e-01 2.56450772e-01 1.22760916e+00
2.22265292e-02 4.59511012e-01 -6.92822039e-01 -6.17388725e-01
5.81872821e-01 -5.66220164e-01 6.43264234e-01 3.94619167e-01
4.05257195e-01 3.74371529e-01 1.39452010e-01 7.00975418e-01
-4.01542276e-01 1.43775210e-01 -2.12976053e-01 -1.78481609e-01
6.55401945e-01 9.64297354e-01 -6.91766739e-01 -4.53438371e-01
-3.30702543e-01 4.60574508e-01 4.39505994e-01 6.10673070e-01
-4.69495595e-01 1.42407000e-01 6.97887838e-01 -1.60241142e-01
3.60793859e-01 3.18510354e-01 -5.35004616e-01 -1.10975122e+00
2.56624103e-01 -5.47054768e-01 6.91477299e-01 -4.13021535e-01
-1.58841097e+00 3.00970431e-02 2.48457305e-02 -8.76406670e-01
-3.47314030e-01 -2.51470476e-01 -4.95930284e-01 4.27589983e-01
-1.64982140e+00 -5.35948217e-01 3.86809528e-01 2.04063967e-01
1.78181022e-01 -1.30400762e-01 2.39420235e-01 -4.53539401e-01
-6.30911708e-01 3.50985587e-01 8.55859458e-01 -4.39854383e-01
1.70698836e-01 -1.17980981e+00 -1.75272465e-01 4.12995815e-01
-1.91484183e-01 5.73397994e-01 1.11915731e+00 -4.79602069e-01
-1.65164757e+00 -6.14581227e-01 1.25945523e-01 1.13503328e-02
1.09757340e+00 1.42973542e-01 -2.68604279e-01 9.84023154e-01
2.79244274e-01 -2.19780892e-01 5.64109087e-01 4.24004555e-01
-2.15430319e-01 -2.19326958e-01 -1.22593629e+00 7.89104581e-01
9.95194793e-01 -1.58376414e-02 -2.98089862e-01 6.37296081e-01
4.73153651e-01 -3.48815084e-01 -4.11530018e-01 -1.76609874e-01
7.15684533e-01 -1.23886263e+00 4.81771350e-01 -9.27906454e-01
7.73860961e-02 6.31407052e-02 -1.14607170e-01 -1.44505930e+00
-1.22756548e-01 -1.37841749e+00 -1.08378425e-01 7.67621696e-01
7.40744531e-01 -1.01432896e+00 1.15672958e+00 1.01629651e+00
3.38325232e-01 -5.13722479e-01 -1.14845645e+00 -9.23358083e-01
3.99321288e-01 -5.82943037e-02 7.40137219e-01 5.79284608e-01
5.24163365e-01 -2.51345605e-01 -6.94101572e-01 7.84184635e-02
1.05086708e+00 8.12732577e-01 6.32088721e-01 -1.06332994e+00
-7.78100550e-01 -7.61367440e-01 6.14689529e-01 -1.40076530e+00
-4.51481156e-03 -2.88150012e-01 -2.27745771e-01 -1.10885847e+00
6.92086816e-01 -5.99416256e-01 -3.28453660e-01 1.96687460e-01
1.20943688e-01 -6.24919906e-02 3.10723335e-01 2.89134532e-01
-1.08478928e+00 2.49834895e-01 1.06506240e+00 -2.33736560e-02
-3.44421804e-01 5.97582996e-01 -1.39577401e+00 5.71826935e-01
8.88913810e-01 -4.19826448e-01 -3.65513563e-01 -4.26606208e-01
8.90564919e-01 8.20785463e-01 1.73635960e-01 -5.88641092e-02
3.81610602e-01 -7.37222552e-01 -7.86970332e-02 -5.70305705e-01
2.04650745e-01 -5.52648723e-01 3.88900340e-01 5.29075921e-01
-9.54214394e-01 -1.21790916e-01 -4.35307741e-01 1.01787496e+00
4.36309874e-01 -4.84546810e-01 6.80892408e-01 -4.24189538e-01
3.67198378e-01 2.59460241e-01 -4.49465126e-01 -9.85311065e-03
1.19498670e+00 -1.10483244e-02 -7.78367162e-01 -1.21553493e+00
-8.83962572e-01 6.28321767e-01 5.59639633e-01 -1.76620901e-01
-1.59714948e-02 -9.93705690e-01 -7.41396546e-01 -3.87283355e-01
-1.41439900e-01 -3.25886637e-01 4.84995931e-01 9.31634605e-01
1.07311904e-01 2.92186230e-01 9.45800170e-02 -3.92498821e-02
-7.89519548e-01 9.02347028e-01 4.09971952e-01 -2.58060187e-01
-4.82854694e-02 6.56917393e-01 1.65075123e-01 1.85336769e-01
4.27696615e-01 -9.44126695e-02 1.48326293e-01 2.78020233e-01
7.40058124e-01 7.72193432e-01 -3.71637672e-01 1.33705316e-02
-2.65900325e-02 2.42624193e-01 -4.60484743e-01 -5.99365532e-01
1.42685628e+00 -6.88398540e-01 -7.90290833e-02 4.45865095e-01
5.28318584e-01 2.18300313e-01 -1.40042388e+00 -5.13347149e-01
-1.07885525e-01 -5.63340127e-01 -1.01345919e-01 -8.69744301e-01
-1.17752576e+00 2.31295064e-01 3.61301936e-02 1.11391461e+00
7.31743217e-01 1.07861280e-01 1.05183028e-01 5.72174430e-01
7.40232885e-01 -1.34280694e+00 -8.83365795e-02 5.07708527e-02
7.11693048e-01 -1.18176126e+00 1.22439690e-01 7.97706172e-02
-7.69784987e-01 6.80697083e-01 9.65973437e-02 -3.43397230e-01
5.24068952e-01 3.44068334e-02 -2.23070502e-01 5.15408032e-02
-1.11756265e+00 -2.22055480e-01 -4.41621572e-01 1.34260431e-01
-1.53437644e-01 3.59460145e-01 -5.83955824e-01 8.40753615e-01
-1.93909481e-01 1.38644218e-01 1.05393434e+00 7.48889863e-01
-7.89170921e-01 -8.51548731e-01 -6.38347387e-01 7.06519604e-01
-1.02000749e+00 2.13823676e-01 -3.48462462e-01 4.48611826e-01
-9.41152155e-01 1.23544776e+00 -2.33508032e-02 5.11280000e-01
2.45991144e-02 -4.02303964e-01 4.15870458e-01 -1.76073715e-01
-4.48310167e-01 2.80903667e-01 2.61565655e-01 -3.53527039e-01
-4.88936931e-01 -8.53565156e-01 -6.40436590e-01 -7.22679734e-01
-4.38201904e-01 5.19769073e-01 5.19727886e-01 7.67325282e-01
2.45772406e-01 -1.97164774e-01 1.05885351e+00 -6.39780164e-01
-1.62494731e+00 -7.95597315e-01 -1.08979654e+00 2.36927852e-01
8.80488098e-01 -5.18598735e-01 -8.44032049e-01 -6.80692196e-01] | [4.438459396362305, 3.2105484008789062] |
ba54cba2-aab5-47ed-b131-4e3970607dc1 | sofamyroom-a-fast-and-multiplatform-shoebox | 2106.12992 | null | https://arxiv.org/abs/2106.12992v1 | https://arxiv.org/pdf/2106.12992v1.pdf | SofaMyRoom: a fast and multiplatform "shoebox" room simulator for binaural room impulse response dataset generation | This paper introduces a shoebox room simulator able to systematically generate synthetic datasets of binaural room impulse responses (BRIRs) given an arbitrary set of head-related transfer functions (HRTFs). The evaluation of machine hearing algorithms frequently requires BRIR datasets in order to simulate the acoustics of any environment. However, currently available solutions typically consider only HRTFs measured on dummy heads, which poorly characterize the high variability in spatial sound perception. Our solution allows to integrate a room impulse response (RIR) simulator with different HRTF sets represented in Spatially Oriented Format for Acoustics (SOFA). The source code and the compiled binaries for different operating systems allow to both advanced and non-expert users to benefit from our toolbox, see https://github.com/spatialaudiotools/sofamyroom/ . | ['Federico Avanzini', 'Michele Geronazzo', 'Daniele Bianchi', 'Roberto Barumerli'] | 2021-06-24 | null | null | null | null | ['room-impulse-response'] | ['audio'] | [-3.40277642e-01 -3.46140057e-01 1.30064273e+00 -4.44538504e-01
-1.30354178e+00 -6.61205769e-01 1.89113945e-01 3.23358849e-02
-3.35043192e-01 6.80422068e-01 1.85402140e-01 -4.30175215e-01
-2.05049053e-01 -5.77018738e-01 -5.69104612e-01 -7.72809803e-01
-2.20826715e-01 3.24336082e-01 2.55318224e-01 -3.86085153e-01
-1.97169378e-01 4.45705712e-01 -2.26825309e+00 2.07548648e-01
7.06441760e-01 8.53246748e-01 7.24588752e-01 1.21192789e+00
3.10621858e-01 3.52801353e-01 -1.00087428e+00 2.63036281e-01
1.66709661e-01 -2.10799143e-01 -2.69800156e-01 -7.51071513e-01
3.74029458e-01 -4.75587137e-02 -1.02906741e-01 7.82069147e-01
1.34818900e+00 6.25446320e-01 5.16197145e-01 -9.96351123e-01
-8.82463828e-02 4.36637670e-01 5.00702500e-01 8.17562193e-02
8.01114857e-01 4.12225306e-01 5.79692245e-01 -7.43465364e-01
1.32434079e-02 9.58674908e-01 5.97032845e-01 3.70306879e-01
-1.19193041e+00 -8.57750177e-01 -6.11428201e-01 2.01731488e-01
-1.69123495e+00 -7.54537106e-01 5.85815430e-01 -3.26319605e-01
6.09427631e-01 7.79043615e-01 5.43187201e-01 1.34752870e+00
-3.50444645e-01 -2.14467328e-02 1.52134097e+00 -4.93054807e-01
6.22256696e-01 4.37748224e-01 1.81282535e-01 1.78219825e-01
-1.13884211e-01 3.56563866e-01 -5.11139870e-01 -2.24702775e-01
4.95724350e-01 -7.13979125e-01 -7.73059011e-01 -8.88730492e-03
-1.09593797e+00 3.87744337e-01 3.05569142e-01 5.01168549e-01
-3.75550032e-01 1.19315349e-02 -7.53904227e-04 3.55523109e-01
5.61105348e-02 3.69271338e-01 -3.29638660e-01 -2.11266503e-01
-7.02611268e-01 5.73287904e-01 8.25247228e-01 8.41589928e-01
6.78767860e-01 2.70544648e-01 -2.22024694e-01 1.34998989e+00
3.56660336e-01 9.88845408e-01 3.30740511e-01 -9.00601625e-01
2.25978601e-03 -5.06701350e-01 3.02735686e-01 -4.81026053e-01
-5.55054784e-01 -4.68881398e-01 -3.79219919e-01 3.77405763e-01
7.08793104e-01 -1.47356331e-01 -8.26119959e-01 1.56024909e+00
4.06531334e-01 6.36483207e-02 -4.06958498e-02 9.80734348e-01
1.13026011e+00 6.17541790e-01 -2.94341266e-01 7.90202022e-02
1.45038891e+00 -3.53650749e-01 -5.08170664e-01 -1.14366926e-01
2.81308740e-01 -1.03855157e+00 1.61938787e+00 6.38573229e-01
-1.16605294e+00 -7.00632572e-01 -6.86508775e-01 2.49458328e-01
-5.38881242e-01 -2.03651950e-01 8.13188702e-02 1.03179967e+00
-1.36658835e+00 1.91832349e-01 -5.81468165e-01 -1.71941772e-01
-3.52945507e-01 1.13082707e-01 -2.43897557e-01 6.62204400e-02
-1.22881866e+00 7.13326693e-01 -1.35791406e-01 2.79322177e-01
-1.25133049e+00 -8.96858275e-01 -8.95102501e-01 8.92636850e-02
-7.97464140e-03 -5.12032509e-01 1.79323757e+00 -1.84369311e-01
-1.70586705e+00 5.59613466e-01 -1.28578961e-01 -1.28491566e-01
5.38797915e-01 -1.21459082e-01 -6.23217285e-01 -1.65890753e-01
-3.32213402e-01 2.07490817e-01 6.92798734e-01 -1.54691362e+00
1.31148398e-02 -2.53482796e-02 -2.83845484e-01 7.78115634e-03
2.56083041e-01 3.58062834e-01 -3.56055759e-02 -3.79947543e-01
-1.85658470e-01 -5.45993924e-01 -1.11876108e-01 -5.74203551e-01
-5.92730224e-01 2.64750868e-01 4.98992810e-03 -8.32873285e-01
1.01543224e+00 -2.14532495e+00 -2.75559098e-01 3.89995307e-01
-2.80885071e-01 -1.19828626e-04 -1.70943841e-01 2.84777015e-01
-3.91307950e-01 -3.94572467e-01 -1.90674350e-01 -4.52943921e-01
3.84494692e-01 -2.96719313e-01 -3.86555463e-01 4.01726633e-01
-7.34331787e-01 2.10384682e-01 -6.50673032e-01 -2.58224130e-01
5.37831366e-01 9.51606333e-01 -6.03191316e-01 7.08731830e-01
6.94098696e-02 9.20689285e-01 -6.31586090e-02 4.18256044e-01
8.92437100e-01 6.36849344e-01 -4.05945182e-01 -1.76607706e-02
-4.92716432e-01 6.33602381e-01 -1.32466412e+00 1.42285538e+00
-1.18799675e+00 4.67340678e-01 6.11458302e-01 -2.57984161e-01
1.11838031e+00 4.89029884e-01 -1.06189981e-01 -7.05062389e-01
1.46561071e-01 5.54641187e-01 2.08732128e-01 -3.16789806e-01
2.95461535e-01 -1.74838260e-01 -1.39041962e-02 4.17166173e-01
2.82430723e-02 -7.60791659e-01 -3.25489491e-01 -3.17759275e-01
1.10067832e+00 -1.25025779e-01 1.05274795e-02 -3.60365689e-01
5.16637862e-01 -5.41390717e-01 -7.99827278e-02 8.36930752e-01
-1.81470402e-02 9.55730140e-01 -1.53595299e-01 1.15120351e-01
-8.68817568e-01 -1.59858823e+00 -5.11299312e-01 1.29870415e+00
-5.13128102e-01 -2.21049972e-02 -1.12678957e+00 4.06903416e-01
-3.69174302e-01 1.21750951e+00 -3.44033986e-01 1.72397926e-01
-3.77558559e-01 -2.84185886e-01 8.03457499e-01 2.87570655e-02
4.38335724e-02 -1.63563764e+00 -8.06699932e-01 4.08650115e-02
-2.38594741e-01 -8.93162191e-01 -1.47036493e-01 3.99465352e-01
-1.53650224e-01 -3.31962287e-01 -8.58708620e-01 -3.47042292e-01
2.03354686e-01 7.98314065e-02 1.12643540e+00 -2.61769295e-01
-7.32735753e-01 9.06349123e-01 -3.55654448e-01 -7.79397786e-01
-5.68052471e-01 -3.49522591e-01 3.13934505e-01 -2.21650183e-01
-2.93915302e-01 -1.06019652e+00 -6.09963536e-01 5.17634988e-01
-7.45442450e-01 -3.87835234e-01 -1.45357311e-01 1.65734157e-01
3.58380407e-01 -9.91367325e-02 5.51898181e-01 -2.44812712e-01
9.24512506e-01 -2.88041860e-01 -8.97505999e-01 7.19316080e-02
2.55308636e-02 -2.23839641e-01 7.56498218e-01 -2.35750884e-01
-1.17506230e+00 -3.21343333e-01 -6.86842382e-01 -2.08082944e-01
-1.06694639e+00 -9.28577781e-02 -4.79919463e-01 1.41904086e-01
9.93495464e-01 4.41329688e-01 -3.34859401e-01 -7.03375518e-01
1.48324400e-01 9.60360646e-01 7.59384692e-01 -8.17510068e-01
8.65352571e-01 1.10943139e-01 -3.55619758e-01 -1.17540324e+00
-3.07780266e-01 -4.86648440e-01 -1.50126681e-01 -4.94027972e-01
7.42734611e-01 -7.93950915e-01 -9.53879058e-01 5.26790738e-01
-1.09147823e+00 -9.18419242e-01 -1.42597839e-01 8.60287309e-01
-8.00753653e-01 -6.12182841e-02 -2.06980303e-01 -1.41683006e+00
-2.87575901e-01 -1.04595470e+00 9.31117773e-01 1.98990300e-01
-3.62707496e-01 -7.89242089e-01 4.08802748e-01 3.00832063e-01
8.50376546e-01 -8.38759765e-02 5.62265635e-01 -3.36788148e-01
-3.57016981e-01 -1.84946045e-01 4.59494770e-01 3.43766659e-01
-5.21161295e-02 -2.22689286e-02 -1.99010479e+00 -1.72791913e-01
6.66348636e-02 -3.18996012e-01 2.92804062e-01 8.03399444e-01
1.31931555e+00 -2.91398950e-02 3.02965879e-01 6.60702407e-01
1.04309905e+00 1.20864838e-01 9.31499422e-01 2.10250497e-01
1.35165557e-01 9.26404953e-01 5.57296038e-01 7.01018035e-01
1.00452289e-01 8.21739376e-01 5.11941791e-01 -2.26623446e-01
-6.37944788e-02 1.69577952e-02 3.24425787e-01 5.39822042e-01
-3.34486395e-01 -3.09251547e-01 -9.74938393e-01 4.64966863e-01
-9.41075444e-01 -9.02680874e-01 -2.39377141e-01 2.81101632e+00
6.49656177e-01 -1.71235263e-01 1.16782822e-01 3.97799969e-01
6.42327070e-01 2.63935123e-02 6.51290864e-02 -6.83468461e-01
9.07904059e-02 8.56115162e-01 2.46433765e-01 1.00773013e+00
-5.31073749e-01 4.09784824e-01 5.94134045e+00 5.68993330e-01
-1.04786623e+00 1.59932718e-01 4.59023342e-02 -1.11661308e-01
-4.46798086e-01 -3.96733463e-01 -5.95837235e-01 2.49473706e-01
1.50995219e+00 -1.57332435e-01 1.02043927e+00 7.54878223e-01
6.87834918e-01 -2.49216601e-01 -9.15793777e-01 1.03665113e+00
-1.76838055e-01 -6.39645517e-01 -6.83765829e-01 -2.07855888e-02
-1.78178132e-01 2.21942201e-01 4.38012034e-01 3.36257815e-01
3.46075892e-01 -1.32026553e+00 9.36377704e-01 6.16518021e-01
8.90674651e-01 -5.64404428e-01 3.90586138e-01 2.90595055e-01
-1.11862302e+00 6.95785880e-02 -4.17866975e-01 3.24229240e-01
2.47372657e-01 4.76099819e-01 -1.22349501e+00 4.57435340e-01
1.19279194e+00 -4.54828203e-01 -5.67430019e-01 1.58541524e+00
-4.04330380e-02 8.88882518e-01 -6.86544478e-01 1.29831076e-01
-2.63263166e-01 7.41946176e-02 7.09754467e-01 1.39286482e+00
8.38035345e-01 -1.39533564e-01 -5.35756767e-01 9.36620057e-01
3.96252722e-01 3.21332902e-01 -6.35435045e-01 5.80844283e-01
7.16771901e-01 1.44882262e+00 -2.46006846e-01 2.96200424e-01
2.11785793e-01 5.85334361e-01 -3.60160351e-01 6.76458836e-01
-1.11222780e+00 -6.92344606e-01 8.80094528e-01 3.60449046e-01
1.64063513e-01 -2.68192917e-01 4.71795723e-02 -4.43127900e-01
-8.55473578e-02 -9.75757301e-01 6.82113245e-02 -1.46608126e+00
-8.27026010e-01 1.03116822e+00 2.97426939e-01 -1.15165782e+00
-2.40876347e-01 -5.80691993e-01 -8.64054859e-01 1.33201492e+00
-1.23372030e+00 -7.50424981e-01 -6.86824501e-01 8.54281306e-01
1.97796881e-01 1.83623552e-01 1.22633636e+00 6.08674169e-01
-1.85235783e-01 5.97970545e-01 -3.93197909e-02 -2.06176609e-01
9.26820874e-01 -1.33391738e+00 1.79794669e-01 3.41804683e-01
-1.64842695e-01 6.97451711e-01 1.41986072e+00 1.29361466e-01
-7.78945506e-01 -8.11181426e-01 3.63592893e-01 -4.84874517e-01
4.82682675e-01 -7.91461825e-01 -1.15889561e+00 3.93140763e-01
1.92989439e-01 1.77552491e-01 8.68185282e-01 1.35071903e-01
-4.57615107e-01 -3.68328094e-01 -1.06880665e+00 4.84238446e-01
6.46434367e-01 -5.58741629e-01 -3.30085188e-01 2.42310673e-01
5.34960628e-01 -3.56484771e-01 -6.90936446e-01 1.49824232e-01
4.94531482e-01 -1.45932531e+00 1.20699954e+00 1.13770761e-01
-2.90384620e-01 -5.54927051e-01 -5.02286375e-01 -1.54898250e+00
1.16632923e-01 -7.79993892e-01 4.29906607e-01 1.21234202e+00
3.46746057e-01 -9.37770069e-01 -1.21359311e-01 3.06781471e-01
-4.37437564e-01 -5.74118569e-02 -1.11948287e+00 -8.87492120e-01
-2.49816906e-02 -1.16325772e+00 7.68182158e-01 3.70110393e-01
-2.76148766e-01 5.15591912e-02 -1.63673013e-01 6.88035488e-01
6.32970810e-01 -1.71474010e-01 9.93964016e-01 -1.17324018e+00
-6.46914601e-01 -2.64287800e-01 -1.25140071e-01 -5.09450316e-01
-2.08108891e-02 -4.84374374e-01 6.94506824e-01 -1.43587661e+00
-5.30182660e-01 -8.51840675e-01 -2.74011582e-01 2.20609441e-01
2.82041013e-01 1.76524729e-01 -1.59410145e-02 -4.30815786e-01
-1.46083593e-01 4.50174540e-01 9.54620898e-01 3.60617727e-01
-3.11603755e-01 5.27453125e-01 -1.68146566e-01 7.12140560e-01
8.36390018e-01 -4.79191124e-01 -3.46382141e-01 -1.19828708e-01
-1.39443604e-02 2.36653820e-01 8.29193830e-01 -1.67851257e+00
1.16586037e-01 8.65217745e-02 6.02305904e-02 -4.45232362e-01
7.88770676e-01 -7.82724679e-01 5.33510745e-01 8.45323652e-02
-2.17914894e-01 -1.25952557e-01 4.96470690e-01 4.09228131e-02
-1.09286860e-01 -1.12479679e-01 9.81907487e-01 -1.16396815e-01
-3.25746313e-02 -1.48533523e-01 -6.61545038e-01 -8.71816054e-02
4.97997254e-01 1.82694884e-03 -2.70685971e-01 -9.73031759e-01
-9.03405547e-01 -4.56162274e-01 4.14984703e-01 -4.46854252e-03
5.72632253e-01 -7.60638893e-01 -6.65517807e-01 3.41861248e-01
1.60130441e-01 2.86803115e-02 8.06472421e-01 6.44875586e-01
-6.18916810e-01 5.07345200e-01 -1.60366103e-01 -4.64786142e-01
-1.44574893e+00 2.06258625e-01 9.83512282e-01 2.50165731e-01
-4.17103112e-01 1.01553988e+00 5.14855206e-01 -7.93373883e-01
4.72094297e-01 -4.79433000e-01 -1.18556924e-01 -2.54053056e-01
6.58611476e-01 3.43570411e-01 4.48857933e-01 -5.62521160e-01
-4.35628086e-01 4.05203760e-01 8.19051921e-01 -6.18758261e-01
1.12634170e+00 -3.90282273e-02 1.35696530e-01 6.69331372e-01
9.35190916e-01 7.20768631e-01 -8.50474238e-01 1.30214810e-01
-3.40846360e-01 -4.20116961e-01 3.05926125e-03 -9.21298742e-01
-4.20429200e-01 1.15702939e+00 1.02987576e+00 1.76114380e-01
1.33807087e+00 2.13301867e-01 3.47675532e-01 4.04444724e-01
8.39277327e-01 -6.82499230e-01 -1.78376839e-01 3.92893821e-01
1.14437306e+00 -5.72207153e-01 -5.75420618e-01 -2.03386039e-01
-5.71650445e-01 8.64781618e-01 2.20713690e-01 3.47717762e-01
8.52874935e-01 7.46244073e-01 6.55512750e-01 1.72729686e-01
-5.69864154e-01 -4.77037579e-01 1.47708863e-01 1.08952022e+00
6.32380009e-01 2.09186167e-01 2.79764801e-01 6.04473352e-01
-1.05243087e+00 -4.61556792e-01 5.03277898e-01 3.61803651e-01
-5.72540820e-01 -9.11466956e-01 -1.29562163e+00 -1.37021050e-01
-4.68492121e-01 -3.45560223e-01 1.48409814e-01 3.60312462e-01
2.36043930e-02 1.13804018e+00 3.99089791e-02 -2.98131704e-01
7.20037043e-01 2.95892596e-01 5.20427346e-01 -5.73495388e-01
-7.24884450e-01 2.84449697e-01 9.93318558e-02 -3.67362440e-01
9.26168784e-02 -6.09822750e-01 -1.20041406e+00 -1.45920232e-01
1.16116755e-01 3.53641808e-01 1.08188009e+00 1.50969774e-01
-2.72809085e-03 9.18379843e-01 5.50511062e-01 -1.06137002e+00
-2.81615824e-01 -1.33883226e+00 -1.06584430e+00 1.16226010e-01
5.51413894e-01 -5.74704766e-01 -8.42970908e-01 -3.00109908e-02] | [15.14090633392334, 5.847815990447998] |
af46d40b-413c-49ab-bc2a-ddc6913183a2 | augmenting-task-oriented-dialogue-systems | 2210.13344 | null | https://arxiv.org/abs/2210.13344v1 | https://arxiv.org/pdf/2210.13344v1.pdf | Augmenting Task-Oriented Dialogue Systems with Relation Extraction | The standard task-oriented dialogue pipeline uses intent classification and slot-filling to interpret user utterances. While this approach can handle a wide range of queries, it does not extract the information needed to handle more complex queries that contain relationships between slots. We propose integration of relation extraction into this pipeline as an effective way to expand the capabilities of dialogue systems. We evaluate our approach by using an internal dataset with slot and relation annotations spanning three domains. Finally, we show how slot-filling annotation schemes can be simplified once the expressive power of relation annotations is available, reducing the number of slots while still capturing the user's intended meaning. | ['Jonathan K. Kummerfeld', 'Kevin Leach', 'Zhenguo Chen', 'Andrew Lee'] | 2022-10-24 | null | null | null | null | ['intent-classification', 'slot-filling', 'task-oriented-dialogue-systems'] | ['natural-language-processing', 'natural-language-processing', 'natural-language-processing'] | [ 2.05823109e-01 1.02437067e+00 -1.45980209e-01 -6.61767900e-01
-5.62259138e-01 -7.49814391e-01 7.52714515e-01 6.15732133e-01
-4.18720514e-01 9.45144773e-01 7.08773851e-01 -6.46919668e-01
2.88167112e-02 -8.31355393e-01 2.17957854e-01 3.65611881e-01
2.99131814e-02 9.31865990e-01 8.16137493e-01 -9.65308964e-01
5.88770164e-03 8.82219300e-02 -1.48614097e+00 7.42553294e-01
7.16719806e-01 8.41371953e-01 -3.22522372e-02 7.24659681e-01
-8.79996240e-01 1.00777388e+00 -8.11423004e-01 -2.24307626e-01
-2.11559206e-01 -3.81562114e-01 -1.62579489e+00 8.57186913e-02
-2.08663106e-01 -3.73833239e-01 2.82445103e-02 3.94407898e-01
2.75749683e-01 2.95065582e-01 3.55201244e-01 -1.12435889e+00
2.15317473e-01 7.46084988e-01 4.54048634e-01 7.62286931e-02
1.20932710e+00 -2.75390208e-01 1.37141144e+00 -3.35980415e-01
9.11522210e-01 1.37496400e+00 5.78835070e-01 6.07070923e-01
-1.14955819e+00 -1.00531556e-01 1.64802186e-02 -1.72004819e-01
-7.46090114e-01 -7.55271435e-01 5.91553509e-01 -3.68150800e-01
1.72942615e+00 7.85046816e-01 5.17039239e-01 6.94734752e-01
-5.39135575e-01 8.07116628e-01 8.62873495e-01 -8.74559522e-01
1.37112841e-01 5.30258536e-01 7.36250699e-01 5.95735073e-01
-2.72089481e-01 -5.15418470e-01 -5.34366727e-01 -6.45698249e-01
4.34473783e-01 -4.51920092e-01 -9.60672945e-02 -1.96232319e-01
-6.91228449e-01 1.00921309e+00 -3.08742613e-01 5.32593071e-01
-5.07094748e-02 -5.46889663e-01 8.78252745e-01 5.15761197e-01
6.85270011e-01 9.05778050e-01 -8.66115272e-01 -5.48906088e-01
-3.49058002e-01 5.35024226e-01 1.77634299e+00 1.08590865e+00
8.05256188e-01 -5.42704403e-01 -3.37621272e-01 1.24900210e+00
2.79068917e-01 -1.38165504e-01 3.75747323e-01 -1.15377593e+00
4.39969271e-01 1.23665392e+00 5.35913408e-01 -6.66410267e-01
-9.09167528e-01 1.36343434e-01 -4.45643552e-02 -3.95250678e-01
5.30147910e-01 -3.84495825e-01 -4.55531061e-01 1.38501430e+00
4.29957658e-01 -7.64043093e-01 3.92437249e-01 5.15614271e-01
1.09145319e+00 3.16522509e-01 4.62153524e-01 -4.38394576e-01
2.01596975e+00 -6.16510451e-01 -1.10096395e+00 -4.96545762e-01
1.32656574e+00 -6.97179496e-01 1.02583683e+00 -1.50183558e-01
-9.43510532e-01 -1.45304054e-01 -7.50221908e-01 -4.94022310e-01
-4.67473835e-01 -1.53858811e-01 1.22333670e+00 7.67130613e-01
-8.11441362e-01 1.64228141e-01 -5.08072138e-01 -6.26149237e-01
-1.27797320e-01 4.48173642e-01 -3.97720098e-01 3.71317416e-01
-1.73338127e+00 1.00799608e+00 6.62816107e-01 -6.12390339e-01
1.69332817e-01 -3.89505774e-01 -1.24001384e+00 1.34943530e-01
6.71207845e-01 -5.47545373e-01 1.97182083e+00 -3.53900164e-01
-1.50259829e+00 8.28080654e-01 -4.41302270e-01 -4.40694362e-01
2.47549444e-01 -2.26439223e-01 -2.27304950e-01 -1.30959405e-02
2.81787992e-01 6.27874672e-01 1.46588996e-01 -7.92066336e-01
-8.48188162e-01 -3.90201837e-01 6.73146904e-01 6.63828313e-01
-2.88624883e-01 3.75421733e-01 -4.32527930e-01 -3.34649682e-02
1.44570023e-01 -7.55662918e-01 -1.33406788e-01 -4.44232017e-01
-4.11304832e-01 -7.62603104e-01 9.40348685e-01 -6.47993982e-01
1.60356104e+00 -1.98420119e+00 -1.95320159e-01 -7.34891221e-02
2.07203239e-01 2.21878886e-01 2.93082356e-01 1.01632822e+00
2.13907376e-01 1.27211988e-01 -6.81625456e-02 -4.60842729e-01
2.93132514e-01 8.59919608e-01 -4.08007950e-01 -3.55782688e-01
2.19192401e-01 8.82577538e-01 -6.60195649e-01 -6.25609279e-01
2.63477057e-01 4.75720316e-02 -6.85699642e-01 5.05173266e-01
-8.90918374e-01 2.41319939e-01 -7.13440299e-01 3.38213056e-01
2.51915623e-02 -1.65481448e-01 6.57261670e-01 1.13710418e-01
-3.20516638e-02 1.27643180e+00 -1.05107641e+00 1.63556433e+00
-6.13981009e-01 5.60370386e-01 1.78270698e-01 -7.73444116e-01
8.30132663e-01 6.70044541e-01 2.93094605e-01 -5.30800581e-01
-5.11913858e-02 1.34266287e-01 -1.34598672e-01 -7.38146782e-01
9.17347252e-01 -2.59016186e-01 -6.18318796e-01 7.15304077e-01
2.17020363e-01 -1.82931021e-01 3.90804976e-01 2.09757924e-01
1.11845052e+00 -7.85809569e-03 8.07541132e-01 -1.06297649e-01
6.05986297e-01 4.86360162e-01 2.94525921e-01 5.95947266e-01
8.14217553e-02 -1.28618583e-01 9.19570208e-01 -4.87486333e-01
-8.44892263e-01 -3.26295137e-01 -2.22725749e-01 1.80583107e+00
-1.79924622e-01 -9.44297671e-01 -6.46831334e-01 -8.73805225e-01
-2.26252392e-01 9.41284418e-01 -2.56908208e-01 2.50337780e-01
-5.30004203e-01 -3.78517419e-01 8.51516485e-01 1.20554239e-01
6.09373271e-01 -9.79847491e-01 -9.81159687e-01 4.43285108e-01
-6.80462956e-01 -1.31836474e+00 2.48331890e-01 4.60380793e-01
-6.05050504e-01 -1.06884599e+00 1.54792266e-02 -5.34899354e-01
1.56132936e-01 -2.10665628e-01 1.38389158e+00 2.49565214e-01
3.69593389e-02 5.09342134e-01 -5.34500837e-01 -4.38211292e-01
-6.79020882e-01 5.23080647e-01 -4.47926909e-01 -6.76069975e-01
7.81547844e-01 -2.34448329e-01 -4.63832505e-02 3.60459328e-01
-7.98939168e-01 1.48426443e-01 -1.60540506e-01 8.44499111e-01
-2.12523445e-01 -2.07330082e-02 3.57548803e-01 -1.50749838e+00
1.24072206e+00 -3.05819839e-01 -8.16375762e-02 2.43530095e-01
-4.65209186e-01 4.06555057e-01 3.35078210e-01 -8.71116742e-02
-1.36945415e+00 7.98089802e-02 -6.30086124e-01 7.50564158e-01
-5.12011766e-01 4.78793591e-01 -2.76022196e-01 2.98888624e-01
7.84912467e-01 -1.43918499e-01 8.79746601e-02 -6.60585165e-01
5.33678651e-01 8.97927523e-01 2.09245786e-01 -7.20440984e-01
1.52741328e-01 6.72574714e-02 -3.04974943e-01 -9.74610209e-01
-9.57527816e-01 -8.98429930e-01 -6.44914806e-01 1.55860394e-01
7.74856329e-01 -6.30288959e-01 -7.78375626e-01 -2.52953798e-01
-1.32321799e+00 -3.41865093e-01 -5.33258021e-01 -7.74595886e-03
-5.74763834e-01 6.06287777e-01 -7.40813613e-01 -1.07166195e+00
-1.76642403e-01 -9.23818767e-01 1.10622013e+00 2.32297331e-02
-1.12109566e+00 -1.29476976e+00 8.20032582e-02 5.38803577e-01
3.53682637e-01 -2.38572359e-01 1.12247014e+00 -1.60624278e+00
-3.20571102e-02 -1.77176461e-01 -1.17535084e-01 -3.66601646e-01
2.14312792e-01 -5.91549873e-01 -9.57506299e-01 3.72094691e-01
-1.08600639e-01 -5.27469814e-01 3.25848430e-01 -2.98656166e-01
3.16143364e-01 -5.19958317e-01 -3.63943338e-01 -2.01694041e-01
7.32084394e-01 4.46039498e-01 5.38123429e-01 4.88081306e-01
6.89631626e-02 1.10314286e+00 8.39550853e-01 5.68277717e-01
7.18289733e-01 1.02701676e+00 -7.20451325e-02 -7.01073930e-02
3.22009116e-01 -3.46649319e-01 -2.84082234e-01 1.23485938e-01
2.64796108e-01 -7.69485831e-02 -1.06896484e+00 1.96145803e-01
-2.00968313e+00 -6.47961617e-01 -4.77534384e-02 1.57330680e+00
1.29572010e+00 2.84412622e-01 4.24562097e-01 3.12669337e-01
1.02556556e-01 2.37713709e-01 -7.85514414e-02 -7.17979014e-01
2.98242271e-01 1.53553009e-01 5.70700243e-02 1.00503802e+00
-1.05900919e+00 1.30971396e+00 7.44185877e+00 2.97123820e-01
-5.86726248e-01 7.17844367e-02 2.69524634e-01 6.99210107e-01
-4.28815395e-01 3.03134263e-01 -9.08917785e-01 -5.26267365e-02
9.69560385e-01 1.12825194e-02 5.11840396e-02 9.52510417e-01
-8.11468437e-02 -4.66440946e-01 -1.15281296e+00 4.83633697e-01
-3.00528258e-01 -1.21288037e+00 -2.84321398e-01 -1.35676697e-01
-3.36329222e-01 -5.65482676e-01 -5.81705272e-01 7.67634451e-01
4.26479697e-01 -7.96038628e-01 3.25039029e-02 1.77270234e-01
5.98523736e-01 -2.62349218e-01 7.49670327e-01 7.29807556e-01
-7.99183071e-01 -7.04048807e-03 -1.49239153e-01 -8.30780268e-01
1.95163280e-01 2.39728883e-01 -1.59193647e+00 3.07367146e-01
2.46390358e-01 1.05327316e-01 -5.89126348e-01 4.24913168e-01
-1.05611883e-01 3.68692964e-01 -6.35696769e-01 -2.57322550e-01
2.69419122e-02 -2.89918110e-03 6.18080676e-01 1.57123518e+00
-3.37556571e-01 5.18812776e-01 4.80744988e-01 3.57535601e-01
3.29813749e-01 2.44945586e-01 -7.85927474e-01 -1.72298960e-02
6.08062565e-01 1.03938663e+00 -6.73018217e-01 -5.64926863e-01
-4.01320934e-01 8.84849131e-01 2.63339877e-01 2.81362291e-02
-3.77716799e-03 -4.46166575e-01 6.02499783e-01 1.10460989e-01
-1.62019268e-01 -3.17584306e-01 -2.56204188e-01 -1.02712715e+00
-6.74477518e-02 -8.93817365e-01 7.26951003e-01 -6.33012891e-01
-6.21387661e-01 5.93425155e-01 4.79834735e-01 -5.84895849e-01
-1.13275647e+00 -5.12610376e-01 -2.71936446e-01 7.89350152e-01
-1.00162423e+00 -1.07513118e+00 -7.90097378e-03 3.16503495e-01
7.78266132e-01 3.45489010e-02 1.61683333e+00 5.84462695e-02
-1.21027067e-01 2.20313370e-01 -8.25703204e-01 3.06931078e-01
5.36464036e-01 -1.49795294e+00 3.47225606e-01 2.37321109e-01
1.58952121e-02 8.53984177e-01 1.09615195e+00 -5.18463552e-01
-9.84348178e-01 -4.13258761e-01 1.49162638e+00 -5.79110086e-01
6.09177232e-01 -5.24583220e-01 -1.15337288e+00 8.04442525e-01
4.03997928e-01 -4.31575924e-01 1.12215936e+00 7.86022305e-01
-4.38108236e-01 5.19711971e-01 -1.04678750e+00 4.18250322e-01
9.46008146e-01 -7.50226021e-01 -1.26791191e+00 3.74507815e-01
1.09483075e+00 -7.56345451e-01 -1.07966280e+00 3.19330037e-01
3.15508097e-01 -9.78202820e-01 7.54530966e-01 -8.71338248e-01
5.28609902e-02 -5.25108688e-02 -1.74114943e-01 -9.79383171e-01
2.16161415e-01 -7.44500458e-01 -1.03303291e-01 1.36057508e+00
8.03651571e-01 -5.93914390e-01 6.48764729e-01 1.50222659e+00
-1.03237301e-01 -4.44265664e-01 -9.45249021e-01 -1.66174948e-01
-2.66266942e-01 -7.71066487e-01 6.51388347e-01 8.82717967e-01
1.19293022e+00 1.16022027e+00 -9.14334580e-02 -3.41411144e-01
-8.70148391e-02 -4.50300165e-02 7.38245547e-01 -1.50090396e+00
-4.06009346e-01 2.31432598e-02 -2.18227819e-01 -1.31916428e+00
4.01174366e-01 -5.12475491e-01 -1.10497586e-02 -1.74961126e+00
-3.80752206e-01 -8.03800225e-01 5.45805037e-01 1.01804185e+00
5.61119616e-02 -1.69029623e-01 -1.46741066e-02 2.47634336e-01
-9.98717189e-01 1.45676583e-01 7.58806944e-01 6.24832138e-02
-6.01942897e-01 2.39856184e-01 -8.01414013e-01 7.52650380e-01
7.03608930e-01 1.38146067e-02 -6.80476487e-01 8.00041631e-02
3.09859216e-01 4.48901534e-01 -1.51382759e-01 -6.02982044e-01
2.18533099e-01 -1.38568431e-01 -6.73859045e-02 -2.80152231e-01
4.68378097e-01 -8.53151143e-01 -8.73738378e-02 1.32923871e-01
-8.57459128e-01 -3.28150868e-01 3.39584559e-01 7.67346472e-02
-3.40760201e-01 -4.30893123e-01 2.43780985e-01 -3.41801137e-01
-8.04972351e-01 -4.30602014e-01 -8.86657000e-01 2.90228367e-01
6.93540692e-01 -7.46221617e-02 -2.94898957e-01 -5.96638501e-01
-1.04912460e+00 3.10335606e-01 3.25444072e-01 4.61406022e-01
2.90746629e-01 -7.99363971e-01 -7.04193860e-02 2.90925920e-01
4.68131065e-01 3.60904820e-02 -2.26237491e-01 2.87058234e-01
-3.47922623e-01 8.66327941e-01 -6.81830272e-02 -4.77211773e-01
-1.57549906e+00 1.90687284e-01 2.93310285e-01 -7.26346850e-01
-7.08881736e-01 4.55483973e-01 -2.17772320e-01 -8.60152960e-01
2.70735353e-01 -2.67577291e-01 -8.37904215e-01 3.36273015e-01
6.25329673e-01 -7.50243068e-02 1.83917731e-01 -5.24163663e-01
-3.31844062e-01 -2.10133195e-01 -1.15731545e-01 -4.96439546e-01
9.60966587e-01 -4.55184042e-01 -2.50139296e-01 6.55776501e-01
8.65216672e-01 1.17697176e-02 -6.46505535e-01 -3.66987884e-01
7.80942380e-01 -2.09064513e-01 -5.10811985e-01 -7.46827602e-01
1.68252900e-01 5.57333410e-01 1.03666335e-01 1.02947271e+00
9.15110946e-01 3.11639488e-01 7.39617407e-01 7.56703794e-01
5.25985479e-01 -1.15391433e+00 -3.03596675e-01 9.92523015e-01
6.37380540e-01 -1.16083515e+00 -5.11144064e-02 -9.19181168e-01
-8.66424322e-01 1.08032155e+00 5.38255692e-01 6.82028413e-01
4.22237247e-01 5.08078456e-01 2.08649695e-01 -3.76228392e-01
-1.06973529e+00 -5.73104858e-01 9.44970399e-02 7.53563225e-01
9.47050095e-01 -1.32709844e-02 -6.59339309e-01 5.90434253e-01
-4.80388284e-01 -2.10834652e-01 4.05984402e-01 1.23790634e+00
-7.66804516e-01 -1.60653973e+00 -9.81422588e-02 4.71765608e-01
-5.75549424e-01 -7.15320706e-02 -8.19401264e-01 7.98077822e-01
-2.51305163e-01 1.30414224e+00 9.06466991e-02 -2.85199851e-01
5.72562516e-01 7.04868555e-01 7.92698711e-02 -1.28274632e+00
-7.33356714e-01 -5.72262667e-02 1.34398472e+00 -5.23941517e-01
-5.12690425e-01 -3.79428059e-01 -1.44369590e+00 1.54024169e-01
-3.11104923e-01 7.37050831e-01 2.51816392e-01 1.30140936e+00
2.29090407e-01 1.88909337e-01 1.58729583e-01 -2.68529087e-01
-3.59811485e-01 -1.07967079e+00 -2.78156579e-01 5.20955145e-01
1.20642550e-01 -6.53954566e-01 4.50059101e-02 -6.01813421e-02] | [12.641746520996094, 7.762259483337402] |
3f023d06-6ec9-46ed-8d53-1c1a8a1311cb | metric-learning-for-image-registration | 1904.09524 | null | http://arxiv.org/abs/1904.09524v1 | http://arxiv.org/pdf/1904.09524v1.pdf | Metric Learning for Image Registration | Image registration is a key technique in medical image analysis to estimate
deformations between image pairs. A good deformation model is important for
high-quality estimates. However, most existing approaches use ad-hoc
deformation models chosen for mathematical convenience rather than to capture
observed data variation. Recent deep learning approaches learn deformation
models directly from data. However, they provide limited control over the
spatial regularity of transformations. Instead of learning the entire
registration approach, we learn a spatially-adaptive regularizer within a
registration model. This allows controlling the desired level of regularity and
preserving structural properties of a registration model. For example,
diffeomorphic transformations can be attained. Our approach is a radical
departure from existing deep learning approaches to image registration by
embedding a deep learning model in an optimization-based registration algorithm
to parameterize and data-adapt the registration model itself. | ['Francois-Xavier Vialard', 'Marc Niethammer', 'Roland Kwitt'] | 2019-04-21 | metric-learning-for-image-registration-1 | http://openaccess.thecvf.com/content_CVPR_2019/html/Niethammer_Metric_Learning_for_Image_Registration_CVPR_2019_paper.html | http://openaccess.thecvf.com/content_CVPR_2019/papers/Niethammer_Metric_Learning_for_Image_Registration_CVPR_2019_paper.pdf | cvpr-2019-6 | ['deformable-medical-image-registration', 'diffeomorphic-medical-image-registration'] | ['medical', 'medical'] | [ 7.71073028e-02 2.21000209e-01 -3.61064345e-01 -7.86453068e-01
-1.00317192e+00 -5.79122186e-01 5.37649035e-01 1.37156278e-01
-5.17504156e-01 1.93036050e-01 4.64737952e-01 1.31607890e-01
-1.74143270e-01 -7.75363088e-01 -7.27703691e-01 -6.86142266e-01
-2.33892262e-01 6.09613359e-01 -2.80725881e-02 -2.16069728e-01
2.28284389e-01 7.61089623e-01 -6.11313343e-01 -7.11007342e-02
7.65081227e-01 4.46199834e-01 3.72023508e-02 5.30155003e-01
2.17505485e-01 1.49727628e-01 -5.06734923e-02 8.42534564e-03
5.96349657e-01 -5.81991553e-01 -1.07648897e+00 1.24735028e-01
6.42731667e-01 -2.61818498e-01 -5.21449506e-01 1.02114570e+00
5.50948858e-01 1.04180172e-01 6.17331326e-01 -7.64359832e-01
-7.61685610e-01 3.27740669e-01 -4.80086803e-01 3.02415401e-01
1.46685258e-01 1.42927930e-01 8.57297421e-01 -5.83544374e-01
7.52937257e-01 9.14833486e-01 9.85951304e-01 5.28021455e-01
-1.62662625e+00 -3.19793850e-01 -3.30718756e-01 -2.81375796e-01
-1.41077816e+00 -5.52286208e-01 9.46444452e-01 -6.29328966e-01
5.08421004e-01 2.91981399e-01 8.55898678e-01 4.79707927e-01
8.41285288e-01 2.33429134e-01 1.02866030e+00 -1.88067004e-01
-4.18611951e-02 -5.68238020e-01 -1.14862636e-01 8.11933279e-01
9.37665580e-04 2.69930184e-01 -1.95601702e-01 -1.99595228e-01
1.36573875e+00 1.71138182e-01 -2.96844155e-01 -6.22781813e-01
-1.49323571e+00 7.94927597e-01 8.98111045e-01 5.02257586e-01
-4.62932110e-01 3.56248856e-01 1.21116757e-01 3.35848391e-01
5.24413049e-01 7.65043736e-01 -4.08622146e-01 2.12363571e-01
-1.06464267e+00 3.63020837e-01 3.30458075e-01 4.84077871e-01
9.42359388e-01 -2.37893574e-02 -2.65663415e-01 5.13626575e-01
4.25371200e-01 1.45287901e-01 7.35139012e-01 -1.16198623e+00
8.50887075e-02 4.44798887e-01 -1.32202312e-01 -1.03332746e+00
-6.31865621e-01 -2.14991063e-01 -8.57503057e-01 3.94253761e-01
5.71250200e-01 -1.89277763e-03 -1.05748296e+00 1.86177683e+00
3.33708405e-01 3.78477097e-01 -2.98364937e-01 8.75219643e-01
5.70202351e-01 1.34149283e-01 -5.62447356e-03 5.14573082e-02
1.17085040e+00 -6.18772686e-01 -6.49860561e-01 -2.15891689e-01
7.54513800e-01 -9.55245495e-01 1.03728235e+00 -2.50952005e-01
-1.37128150e+00 -3.07886183e-01 -9.35722113e-01 -3.64203364e-01
-5.09157777e-02 -4.22187477e-01 4.52643424e-01 2.73297638e-01
-1.32565200e+00 7.66179442e-01 -1.35210729e+00 -1.98907614e-01
6.69589281e-01 8.36505234e-01 -6.62827551e-01 1.09885544e-01
-8.99493337e-01 1.12044668e+00 -1.15476665e-03 2.33286351e-01
-6.86728358e-01 -1.07493830e+00 -9.90794420e-01 -2.87774324e-01
-2.75027663e-01 -8.45854998e-01 9.92855906e-01 -1.00839186e+00
-1.44970167e+00 1.46376586e+00 -4.60400395e-02 -1.98662177e-01
5.26617885e-01 7.01634362e-02 3.43234204e-02 -1.22061253e-01
-2.44693980e-02 6.95793033e-01 7.26982892e-01 -1.11442542e+00
1.86800823e-01 -3.47000271e-01 -9.87914354e-02 2.55727798e-01
6.19041920e-02 1.47470072e-01 -2.59618551e-01 -8.75189841e-01
4.53319490e-01 -1.08095860e+00 -6.62908018e-01 4.56183881e-01
-8.55446234e-02 2.50933319e-01 6.10046029e-01 -8.11186552e-01
7.14035869e-01 -1.97822261e+00 2.75919259e-01 3.60219002e-01
4.62204963e-01 -2.93304790e-02 -3.88226688e-01 1.27388642e-03
-1.51126549e-01 6.01582378e-02 -4.94847476e-01 -2.85731405e-01
-2.92083651e-01 2.76080161e-01 2.32385714e-02 1.04595017e+00
1.83424011e-01 1.34752321e+00 -8.56177568e-01 -4.16998327e-01
2.02804402e-01 8.17622542e-01 -8.92987430e-01 3.99546027e-01
3.26691382e-02 1.13991845e+00 -5.50803900e-01 2.24437743e-01
6.08833611e-01 -1.25305042e-01 1.76050723e-01 -5.20236671e-01
8.12692121e-02 3.02380383e-01 -8.85715365e-01 2.13220811e+00
-4.59910959e-01 5.05819917e-01 1.33943349e-01 -1.26767993e+00
7.85718501e-01 3.67523760e-01 1.17098427e+00 -3.69030386e-01
3.69677693e-01 2.01407179e-01 1.93924382e-01 -3.75668943e-01
1.42221585e-01 -3.47345918e-01 6.04954213e-02 5.65340638e-01
-9.99315232e-02 -4.06714797e-01 -4.43184406e-01 -3.07374656e-01
9.33077633e-01 1.47706926e-01 3.14945936e-01 -7.14172184e-01
2.66129196e-01 -7.02649429e-02 6.76260769e-01 3.35968584e-01
-2.00869009e-01 1.15624630e+00 2.37700101e-02 -9.40600574e-01
-1.17098486e+00 -1.21489060e+00 -4.41288859e-01 5.86962223e-01
9.55733880e-02 -3.81375663e-03 -8.97394001e-01 -5.30481815e-01
-9.46478099e-02 -1.34563625e-01 -8.01891744e-01 -4.12004381e-01
-1.19083321e+00 -9.06608462e-01 5.04536688e-01 4.75960016e-01
3.81401509e-01 -1.09294736e+00 -2.68584818e-01 2.52874792e-01
-1.18934967e-01 -9.30130541e-01 -1.28649962e+00 -8.08468908e-02
-1.25272524e+00 -1.17396855e+00 -5.64532280e-01 -1.20550776e+00
1.26069129e+00 2.62770080e-03 1.15489423e+00 4.71456081e-01
-4.40830678e-01 3.61612648e-01 2.10380092e-01 4.88454178e-02
-6.78392649e-01 1.56637087e-01 2.64624655e-02 -3.28880847e-02
1.11024946e-01 -8.50363433e-01 -7.68842995e-01 2.50112385e-01
-9.43674266e-01 -3.04047287e-01 4.17766958e-01 8.47098529e-01
9.15642560e-01 -3.47808689e-01 1.79359868e-01 -8.13258886e-01
6.66846693e-01 -1.80563271e-01 -5.08187592e-01 2.33400777e-01
-5.77403843e-01 3.15567791e-01 2.36219004e-01 -5.71317375e-01
-5.90887010e-01 3.38274837e-01 -2.66474754e-01 -3.87614906e-01
1.34278387e-02 5.16322792e-01 -1.82033405e-02 -5.98947227e-01
7.15717375e-01 3.50013599e-02 4.90618795e-01 -2.38262728e-01
3.57532978e-01 -1.16361082e-02 7.09532619e-01 -6.22041285e-01
9.95953441e-01 6.26097977e-01 1.59640744e-01 -4.61368829e-01
-5.85342228e-01 -1.18846223e-01 -1.38328743e+00 7.47534111e-02
1.11031759e+00 -6.89081669e-01 -4.73283887e-01 4.96153772e-01
-1.09007657e+00 -8.03421140e-01 -5.38797677e-01 6.16591036e-01
-8.20414245e-01 2.30480582e-01 -6.01484537e-01 2.72110790e-01
-3.68116885e-01 -1.42992818e+00 1.18043303e+00 -7.46513680e-02
-4.34045136e-01 -1.64502060e+00 4.95523155e-01 5.45311943e-02
6.11673236e-01 5.67082167e-01 7.09272683e-01 -3.60688835e-01
-5.82782209e-01 -3.69068921e-01 9.98644754e-02 3.07721924e-02
7.47819781e-01 -3.06116074e-01 -6.15802824e-01 -4.29947913e-01
1.87687218e-01 -7.40499049e-02 4.83335197e-01 7.83164918e-01
1.28377366e+00 -4.22424465e-01 -3.18157136e-01 1.21941483e+00
1.21370947e+00 -1.72889471e-01 7.94639647e-01 1.95952773e-01
9.53086913e-01 3.85654092e-01 -1.90459508e-02 -1.39688641e-01
5.27089298e-01 9.95302498e-01 5.00531346e-02 -5.71532667e-01
-4.56119090e-01 -5.70417531e-02 8.80599916e-02 8.61939013e-01
-2.29757637e-01 4.91220146e-01 -1.07267582e+00 3.40071827e-01
-1.54761875e+00 -7.34097779e-01 8.11380818e-02 2.35088325e+00
1.31190920e+00 -2.45715916e-01 -1.32659059e-02 -4.13768709e-01
5.78482628e-01 2.38632202e-01 -3.92576635e-01 -7.26673901e-02
1.33354038e-01 2.28747532e-01 4.45688874e-01 9.68894720e-01
-1.10433698e+00 9.26505983e-01 7.72581816e+00 1.91257089e-01
-1.41447508e+00 1.90999687e-01 5.05463362e-01 1.37377024e-01
-5.25664926e-01 -1.40496308e-03 -3.65976214e-01 2.17800647e-01
5.41066229e-01 -1.29506156e-01 6.01303637e-01 2.91778088e-01
4.28023547e-01 4.23512399e-01 -1.36602664e+00 7.97554791e-01
1.07355770e-02 -1.59625804e+00 -6.06687702e-02 3.95597100e-01
6.71988010e-01 2.23614991e-01 6.54756278e-02 -4.00153607e-01
4.86752510e-01 -1.26941109e+00 3.13576311e-01 7.68331647e-01
7.42313564e-01 -3.55511576e-01 4.16047156e-01 -2.00264417e-02
-1.03710127e+00 5.38852274e-01 -3.45379561e-01 2.10759372e-01
2.45444238e-01 2.78212249e-01 -5.14315724e-01 1.44429773e-01
4.85651731e-01 7.92171121e-01 -4.13500428e-01 9.29538250e-01
-4.44412567e-02 3.12312782e-01 -1.72442853e-01 8.07395220e-01
-4.60822619e-02 -4.97679979e-01 5.28487980e-01 9.91555512e-01
1.61668748e-01 8.40601772e-02 4.96925175e-01 1.01301098e+00
-3.50204051e-01 4.76592313e-03 -6.69891596e-01 3.50079238e-01
1.95966274e-01 1.26025355e+00 -7.26755381e-01 1.36025235e-01
-3.16216797e-01 8.88391495e-01 4.50011641e-01 1.76460236e-01
-5.59441566e-01 8.99979752e-03 1.01089036e+00 4.88360137e-01
-1.35351688e-01 -6.69995844e-01 -5.18900931e-01 -1.28923523e+00
1.92969292e-02 -8.34206402e-01 1.74642414e-01 -3.46894324e-01
-1.28332865e+00 5.26752174e-01 -4.67425259e-03 -1.19930327e+00
-2.22148255e-01 -2.90194899e-01 -9.06675577e-01 1.01539886e+00
-1.33411813e+00 -1.42394423e+00 -2.34508440e-01 7.51211703e-01
1.52190357e-01 2.09442191e-02 9.54678714e-01 2.43999094e-01
-1.06224671e-01 6.52751744e-01 -5.54076172e-02 5.19103646e-01
9.15920734e-01 -1.20138180e+00 5.10033250e-01 7.97489405e-01
2.28823513e-01 9.45606828e-01 5.54108858e-01 -5.85289598e-01
-1.36656833e+00 -1.01517868e+00 8.51731002e-01 -7.85320282e-01
6.20424926e-01 -2.66637743e-01 -1.20518720e+00 1.08722055e+00
1.52379483e-01 8.05812418e-01 6.03773057e-01 -8.89278203e-02
-2.57106453e-01 -1.54164553e-01 -1.31358194e+00 5.54849923e-01
9.57413554e-01 -6.62503123e-01 -5.39231539e-01 4.72359389e-01
4.29425240e-01 -9.32741284e-01 -1.50937057e+00 2.80093044e-01
6.88617706e-01 -3.36415201e-01 1.32190752e+00 -7.90854335e-01
1.64344370e-01 -3.00017446e-01 2.04938710e-01 -1.43192947e+00
-4.84801322e-01 -7.96989679e-01 2.28314534e-01 8.64422321e-01
1.39189541e-01 -5.70030153e-01 8.39321434e-01 1.00436854e+00
-1.98818326e-01 -5.35178423e-01 -9.17024553e-01 -6.10448778e-01
6.97328389e-01 -2.63282191e-02 6.63465679e-01 1.40099871e+00
-2.02867404e-01 -2.25961670e-01 -1.19211346e-01 2.55805135e-01
6.96824074e-01 -5.76764755e-02 6.34191275e-01 -1.06700742e+00
-9.88686979e-02 -6.05615616e-01 -5.84040463e-01 -7.79351413e-01
4.61000621e-01 -1.36773300e+00 2.62557656e-01 -1.33761489e+00
8.04434940e-02 -7.82001317e-01 -2.46871576e-01 5.84200799e-01
-2.42014930e-01 6.10941350e-01 -1.92520246e-01 6.40045881e-01
2.81839482e-02 3.64217222e-01 1.63419759e+00 -1.14087842e-01
-4.21998173e-01 -2.75965035e-02 -7.26335347e-01 7.50486910e-01
9.78625655e-01 -6.44583702e-01 -3.46077740e-01 -8.37814450e-01
1.54870553e-02 -2.29685187e-01 5.00134766e-01 -5.82553208e-01
2.57579505e-01 -3.29109520e-01 4.24483329e-01 2.01090500e-01
-1.06858999e-01 -6.94489598e-01 2.37360984e-01 5.35433948e-01
-5.45563221e-01 2.66861618e-01 -6.15016259e-02 1.63657784e-01
-2.41304085e-01 3.16918790e-02 1.14931989e+00 -8.29637572e-02
-2.39095122e-01 8.91573846e-01 -1.27213523e-01 2.01430857e-01
6.80441678e-01 -1.52343720e-01 1.03077859e-01 -2.33370766e-01
-8.80860925e-01 -9.50113237e-02 7.40872979e-01 3.94518197e-01
5.64076245e-01 -1.64144552e+00 -8.31858695e-01 3.80533665e-01
-5.01707196e-02 3.44634622e-01 -1.39429331e-01 1.01078570e+00
-7.07222462e-01 -1.38473019e-01 -3.93271327e-01 -6.24676526e-01
-9.90716159e-01 3.12097907e-01 9.13412035e-01 -3.01240414e-01
-9.19660509e-01 5.68132102e-01 2.21380711e-01 -9.20372248e-01
-2.19059065e-01 -4.20481175e-01 -8.76291022e-02 -4.86473769e-01
2.36335725e-01 -1.20906949e-01 1.67989433e-01 -9.22027946e-01
-3.69292110e-01 1.13622105e+00 9.34462324e-02 -3.16337124e-02
1.46150744e+00 -5.95989563e-02 -4.33599025e-01 7.90397897e-02
1.47763777e+00 7.85598438e-03 -1.47110963e+00 -4.70303357e-01
5.05190417e-02 -4.42974746e-01 3.20784956e-01 -3.21707398e-01
-1.44318736e+00 5.89872301e-01 6.86905086e-01 -2.14514643e-01
9.82169569e-01 8.00647810e-02 8.41371953e-01 1.94219604e-01
2.26732522e-01 -6.93703115e-01 7.79492930e-02 2.92827308e-01
1.17186224e+00 -1.40751195e+00 2.65511036e-01 -1.94628254e-01
-2.47551605e-01 1.12314212e+00 3.07782382e-01 -6.92337334e-01
1.13485086e+00 5.73541701e-01 4.15963888e-01 -4.06871647e-01
1.98765527e-02 2.39938095e-01 9.00409043e-01 7.09682643e-01
9.20414031e-01 -1.51337162e-01 -2.83207387e-01 1.20632418e-01
-3.37150037e-01 -1.79783422e-02 2.47252584e-01 7.71202683e-01
-1.41739696e-01 -1.59484804e+00 -2.32655063e-01 3.35162252e-01
-4.80817944e-01 -4.56745131e-03 -1.34178251e-02 7.48204231e-01
-1.21866100e-01 2.46713310e-01 3.63962710e-01 -1.37590393e-01
2.13611349e-01 -2.16166541e-01 8.54294479e-01 -7.03046501e-01
-6.91050351e-01 1.81689888e-01 -5.05412936e-01 -8.10269654e-01
-7.67399788e-01 -9.66529429e-01 -1.38354921e+00 -2.73438245e-01
-5.07090352e-02 -1.53998032e-01 4.16625619e-01 1.06222701e+00
3.11615705e-01 3.44952613e-01 7.26820767e-01 -1.01216519e+00
-3.23947757e-01 -5.85723817e-01 -3.60477865e-01 7.08356500e-01
4.96200383e-01 -4.16130662e-01 2.49772109e-02 5.24765730e-01] | [13.96570873260498, -2.5605995655059814] |
7da50679-c919-478a-85e7-66afee9e35aa | bi-link-bridging-inductive-link-predictions | 2210.14463 | null | https://arxiv.org/abs/2210.14463v1 | https://arxiv.org/pdf/2210.14463v1.pdf | Bi-Link: Bridging Inductive Link Predictions from Text via Contrastive Learning of Transformers and Prompts | Inductive knowledge graph completion requires models to comprehend the underlying semantics and logic patterns of relations. With the advance of pretrained language models, recent research have designed transformers for link prediction tasks. However, empirical studies show that linearizing triples affects the learning of relational patterns, such as inversion and symmetry. In this paper, we propose Bi-Link, a contrastive learning framework with probabilistic syntax prompts for link predictions. Using grammatical knowledge of BERT, we efficiently search for relational prompts according to learnt syntactical patterns that generalize to large knowledge graphs. To better express symmetric relations, we design a symmetric link prediction model, establishing bidirectional linking between forward prediction and backward prediction. This bidirectional linking accommodates flexible self-ensemble strategies at test time. In our experiments, Bi-Link outperforms recent baselines on link prediction datasets (WN18RR, FB15K-237, and Wikidata5M). Furthermore, we construct Zeshel-Ind as an in-domain inductive entity linking the environment to evaluate Bi-Link. The experimental results demonstrate that our method yields robust representations which can generalize under domain shift. | ['Mobarakol Islam', 'Shihao Liang', 'Bohua Peng'] | 2022-10-26 | null | null | null | null | ['inductive-knowledge-graph-completion'] | ['knowledge-base'] | [-1.40414327e-01 7.84201622e-01 -8.25563967e-01 -4.84654933e-01
-1.46113455e-01 -5.50064445e-01 6.10652804e-01 1.97950438e-01
6.42044395e-02 8.49867344e-01 4.66337830e-01 -6.69923365e-01
-5.23624361e-01 -1.38318920e+00 -1.22505379e+00 2.28783265e-01
-3.33965749e-01 9.25070524e-01 4.87424940e-01 -5.70904315e-01
-2.19108284e-01 -2.92415526e-02 -1.11141384e+00 6.47460938e-01
1.11796916e+00 7.09262908e-01 -5.08693606e-02 2.41391659e-01
-4.40724552e-01 1.47532010e+00 5.88543192e-02 -1.13738930e+00
-1.19776942e-01 -1.26789674e-01 -1.31917620e+00 -9.08616424e-01
3.26047897e-01 -1.06079362e-01 -6.55152857e-01 6.80001378e-01
1.58677623e-01 -9.57875978e-03 7.24632859e-01 -1.46471393e+00
-1.19738305e+00 1.73719323e+00 -1.75686255e-01 6.29502982e-02
6.77376091e-01 -2.63874352e-01 1.81234920e+00 -8.18421543e-01
1.11343467e+00 1.49290538e+00 1.06162262e+00 3.40380311e-01
-1.64389408e+00 -8.74769568e-01 3.84812832e-01 7.88601875e-01
-1.31438291e+00 -2.19332397e-01 8.36644769e-01 -3.91194105e-01
1.24141788e+00 -1.17237993e-01 4.89730507e-01 1.33971107e+00
5.52277192e-02 6.99304223e-01 7.51763225e-01 -5.03494382e-01
-3.11776012e-01 -1.50737375e-01 5.75682938e-01 8.92501712e-01
5.24008751e-01 1.97033584e-01 -8.94797325e-01 -6.99419305e-02
4.14726764e-01 -5.00302076e-01 -2.62444854e-01 -5.72352707e-01
-1.13187921e+00 5.42797387e-01 8.06855679e-01 -1.08424202e-01
2.13223156e-02 2.15654433e-01 3.92577708e-01 5.34889042e-01
2.41024300e-01 5.85942566e-01 -8.42241049e-01 2.57005036e-01
-1.35692805e-01 3.22066367e-01 1.13756013e+00 1.49377549e+00
9.67365205e-01 -4.56122845e-01 -2.27853119e-01 7.60081947e-01
6.02955043e-01 3.86291504e-01 1.13149278e-01 -8.41946363e-01
9.27371025e-01 8.22183073e-01 -3.56805086e-01 -1.13429320e+00
-6.31328404e-01 -4.41087514e-01 -4.85032588e-01 -6.43998623e-01
2.63561279e-01 7.84999784e-03 -4.84162062e-01 1.97666717e+00
9.30229127e-02 3.72610807e-01 4.19783026e-01 4.12574470e-01
1.32760608e+00 4.45214510e-01 4.76280481e-01 1.39966637e-01
1.00141573e+00 -7.42603600e-01 -5.64505637e-01 -2.73600638e-01
1.24786413e+00 -2.29425907e-01 9.78326976e-01 -1.36876274e-02
-8.28987300e-01 -4.02398109e-01 -8.92161071e-01 -3.46630722e-01
-5.01087070e-01 -2.79735476e-01 1.17964983e+00 1.41855150e-01
-8.47976208e-01 6.54181182e-01 -4.19089556e-01 -4.77345973e-01
3.48294258e-01 3.22816223e-01 -4.49632287e-01 -2.03426704e-01
-1.99312556e+00 1.12374473e+00 1.12302315e+00 -1.55526325e-01
-3.78857970e-01 -1.10885835e+00 -1.11684680e+00 1.59235969e-01
4.76248860e-01 -9.98751521e-01 9.25740540e-01 -3.96701485e-01
-1.14912426e+00 8.70539665e-01 -1.14352569e-01 -5.95399022e-01
-9.56828520e-03 -2.55017966e-01 -6.67215049e-01 -2.76119798e-01
1.22074768e-01 7.57742584e-01 1.44571930e-01 -1.26902783e+00
-3.68888885e-01 -1.13479555e-01 4.54182595e-01 1.86628982e-01
-3.31039846e-01 -2.98685968e-01 -3.38497519e-01 -5.19840539e-01
2.57305741e-01 -7.89892197e-01 2.95752168e-01 -5.81280112e-01
-9.45433497e-01 -7.83823669e-01 3.53704691e-01 -4.57717687e-01
1.41961956e+00 -1.78374946e+00 2.30427444e-01 5.36114812e-01
1.97427168e-01 -8.45478624e-02 -2.72701979e-01 4.67230707e-01
-4.42745686e-01 3.20834011e-01 3.57228667e-01 3.82363088e-02
1.81790575e-01 6.05564654e-01 -7.48210967e-01 -2.48157382e-01
2.46563688e-01 1.50091934e+00 -9.62333739e-01 -7.87829518e-01
-3.65532488e-01 -5.82059734e-02 -1.02321696e+00 1.58333644e-01
-7.09728062e-01 2.62101702e-02 -3.54891628e-01 5.54068148e-01
3.16653252e-01 -5.53601444e-01 9.25368667e-01 -6.11315727e-01
6.05213642e-01 1.11171794e+00 -8.96537542e-01 1.78027856e+00
-3.01081330e-01 2.77274609e-01 -6.63918972e-01 -9.87072289e-01
1.12275517e+00 6.83906451e-02 1.31476209e-01 -8.75264823e-01
-3.97165209e-01 8.84474367e-02 1.76576898e-01 -4.56501216e-01
4.46783304e-01 -2.95274928e-02 -7.40518142e-03 3.68312985e-01
3.57767999e-01 1.36996627e-01 3.40159655e-01 8.25627387e-01
1.19120955e+00 8.01333427e-01 1.20478161e-01 -1.14612676e-01
2.76809901e-01 1.01664819e-01 7.79630244e-01 5.99014819e-01
2.99152166e-01 -2.31280044e-01 8.14991295e-01 -4.70772356e-01
-5.69768727e-01 -1.36652291e+00 -5.67539223e-02 1.39562225e+00
2.00752988e-01 -9.90382373e-01 -6.50242940e-02 -1.00187051e+00
3.99538934e-01 9.97897387e-01 -4.72120553e-01 -4.64012653e-01
-8.52379799e-01 -4.86769468e-01 8.35571766e-01 8.29372168e-01
2.83934593e-01 -1.05978203e+00 5.60837269e-01 2.25991011e-01
-3.68928432e-01 -1.23117054e+00 1.02823183e-01 4.15571153e-01
-7.93376744e-01 -1.32891572e+00 4.22600091e-01 -1.02856815e+00
5.29523671e-01 -2.65735954e-01 1.78761458e+00 1.30370349e-01
2.86709279e-01 3.28115582e-01 -3.85157257e-01 -8.47399235e-02
-3.31787914e-01 4.80229706e-01 1.16626605e-01 -6.43096983e-01
6.30217969e-01 -9.16761518e-01 -4.13210802e-02 3.64549398e-01
-3.16626787e-01 2.86534071e-01 4.23831522e-01 7.87994683e-01
4.69317377e-01 2.78189336e-03 7.55140781e-01 -1.38991904e+00
7.20891118e-01 -7.44020164e-01 -3.11113507e-01 8.18073988e-01
-9.50903893e-01 5.48927188e-01 5.22226810e-01 -1.53695270e-01
-1.27170706e+00 -3.11237782e-01 2.88805336e-01 -4.83964533e-02
2.36826748e-01 1.02637243e+00 -4.19719100e-01 2.08232865e-01
9.31950033e-01 -1.13962710e-01 -2.94104755e-01 -2.60977060e-01
9.22274530e-01 2.29778588e-02 5.15607655e-01 -1.50676811e+00
1.07595360e+00 -1.63177967e-01 1.45862922e-01 -4.33600731e-02
-1.23000491e+00 -1.16879800e-02 -7.40555823e-01 2.06632107e-01
5.10066450e-01 -1.08049667e+00 -8.89100909e-01 -2.32410312e-01
-1.29224253e+00 -6.24139905e-01 -1.84192918e-02 2.93509245e-01
-3.85758638e-01 1.71907082e-01 -7.12968767e-01 -1.01527408e-01
-7.58040547e-02 -5.85301399e-01 4.44496542e-01 -1.40846834e-01
-6.80224478e-01 -1.42298663e+00 1.55876502e-01 3.34450752e-01
5.11852913e-02 -1.52348682e-01 1.78833365e+00 -9.26339090e-01
-8.27010036e-01 4.27987814e-01 -4.06576872e-01 -9.78838801e-02
-4.41822000e-02 -2.30855033e-01 -5.72294056e-01 2.85207719e-01
-1.18399036e+00 -7.21712172e-01 9.97095108e-01 -1.71519086e-01
1.18762338e+00 -4.69418943e-01 -8.69899333e-01 6.81114495e-01
9.93429720e-01 -1.05850421e-01 6.27963483e-01 4.90661621e-01
1.02316904e+00 7.08147466e-01 4.38602477e-01 -5.84616922e-02
1.30658579e+00 5.32105565e-01 2.06030369e-01 4.27626580e-01
-3.97693068e-01 -1.14233279e+00 3.12368721e-01 7.38271594e-01
-2.08080202e-01 -8.09664726e-02 -1.09869289e+00 4.65246320e-01
-1.96676624e+00 -1.10333586e+00 -2.46961638e-01 1.66294229e+00
1.56774843e+00 2.98716694e-01 -1.68357342e-01 -1.59400702e-01
5.19310117e-01 -1.14473522e-01 -4.10207182e-01 -8.52761269e-02
-2.71936595e-01 2.73924381e-01 3.64696652e-01 6.72425807e-01
-9.00472164e-01 1.38585937e+00 5.92436123e+00 7.35888064e-01
-5.02886176e-01 -2.00887218e-01 2.18576595e-01 3.23903829e-01
-1.01939559e+00 4.26606596e-01 -1.20305395e+00 1.36477947e-01
7.20801592e-01 -2.18163386e-01 5.57731986e-01 6.55511200e-01
-4.80391294e-01 4.08941865e-01 -1.72516930e+00 5.56054413e-01
-2.30706409e-01 -1.61313856e+00 4.00111467e-01 -1.98691949e-01
6.31613255e-01 -5.01304716e-02 -1.37595296e-01 9.99477684e-01
1.06928265e+00 -1.12771964e+00 4.54648584e-01 7.42787182e-01
5.41923881e-01 -5.78637719e-01 4.44080472e-01 6.93856701e-02
-1.23088646e+00 -4.50829938e-02 -2.78161556e-01 -1.29196987e-01
-3.16939736e-03 5.47348499e-01 -1.14581001e+00 9.67229009e-01
5.10276973e-01 1.15562630e+00 -7.47452796e-01 3.93357873e-01
-1.07675755e+00 7.18003213e-01 -1.63710833e-01 3.79475541e-02
-3.40992033e-01 7.01730996e-02 2.17821479e-01 1.11425519e+00
6.79034814e-02 1.94607571e-01 2.50243485e-01 1.08669031e+00
-5.52576780e-01 -7.15274140e-02 -8.58433425e-01 -7.82684460e-02
1.00031221e+00 8.41850877e-01 -5.13749607e-02 -1.25864923e-01
-5.15587270e-01 3.24798197e-01 1.05060101e+00 6.51115060e-01
-7.22695708e-01 -1.48466870e-01 4.81592447e-01 2.78711349e-01
1.97340697e-01 -1.06048785e-01 -4.13619429e-01 -1.19187546e+00
-3.26888859e-02 -4.70103681e-01 8.32988977e-01 -8.48526657e-01
-1.60253203e+00 1.45642027e-01 3.81279647e-01 -6.25065088e-01
-3.24254215e-01 -8.04964662e-01 -5.61698735e-01 7.28232801e-01
-1.65862715e+00 -1.66095924e+00 5.74011281e-02 6.05466962e-01
-2.15085283e-01 -2.96293527e-01 9.48509037e-01 1.45970866e-01
-4.74884331e-01 8.69337857e-01 -5.22362173e-01 5.79020441e-01
8.87742043e-01 -1.27337706e+00 4.04491633e-01 4.51429248e-01
3.06384593e-01 1.17043436e+00 4.76446033e-01 -1.04720545e+00
-1.39751029e+00 -1.31153190e+00 1.37083888e+00 -6.95854545e-01
1.20143688e+00 -1.35783002e-01 -1.05259120e+00 1.52476156e+00
1.06865764e-01 2.08743364e-01 9.42338467e-01 1.32198489e+00
-1.21224797e+00 -3.00351322e-01 -6.12272322e-01 5.94191730e-01
1.88074148e+00 -7.15224624e-01 -1.05164742e+00 2.83969462e-01
1.16333067e+00 -3.99943650e-01 -1.30449128e+00 7.22851813e-01
3.98110896e-01 -6.00655317e-01 1.24644744e+00 -1.20821941e+00
8.65677476e-01 -2.32072383e-01 -1.83470041e-01 -1.54656708e+00
-6.60468102e-01 -3.38364184e-01 -7.37420261e-01 1.51243222e+00
1.10378075e+00 -6.37003422e-01 7.56947815e-01 5.27549982e-01
-1.49158701e-01 -8.06188285e-01 -4.68859971e-01 -9.48737264e-01
2.33040959e-01 -5.61666012e-01 6.78400636e-01 1.34247243e+00
7.41209984e-01 7.94984221e-01 4.20665517e-02 2.91180938e-01
6.58330262e-01 4.05281246e-01 7.42364049e-01 -1.63236821e+00
-4.83262062e-01 -1.78705081e-01 -2.37954795e-01 -1.07443440e+00
1.07685733e+00 -1.76470530e+00 -5.15899539e-01 -1.81289959e+00
2.63353258e-01 -9.27748501e-01 -3.48641962e-01 1.04153049e+00
-2.26385400e-01 -2.80654788e-01 -1.27202258e-01 5.96914776e-02
-8.02267075e-01 5.65560937e-01 1.17836773e+00 -3.25865149e-01
7.74768218e-02 -3.81826788e-01 -9.73468542e-01 6.59245610e-01
6.30359828e-01 -3.14161688e-01 -8.63783479e-01 -6.03264749e-01
1.13042235e+00 -1.79763272e-01 4.75513577e-01 -4.20051098e-01
4.99351531e-01 -1.73343569e-01 3.27596575e-01 -5.17148376e-01
9.44004059e-02 -6.86150312e-01 8.69877562e-02 1.04488060e-01
-7.49604642e-01 -1.98449939e-01 1.65368438e-01 6.87061787e-01
-2.94495057e-02 -6.17846288e-02 3.31308581e-02 7.11399466e-02
-1.10125411e+00 1.74781203e-01 4.55770195e-01 5.01190305e-01
7.61114120e-01 1.61737010e-01 -9.64335084e-01 -1.58710465e-01
-1.00196755e+00 7.31968462e-01 3.91751304e-02 5.73792636e-01
5.32988548e-01 -1.76253486e+00 -5.75746417e-01 -1.20174900e-01
5.62110364e-01 2.10652977e-01 -1.66420624e-01 7.05536723e-01
-9.76494625e-02 4.14963365e-01 3.58294174e-02 -2.44887859e-01
-8.56483221e-01 5.92852652e-01 1.78953931e-01 -5.49984396e-01
-4.96419936e-01 9.80865479e-01 -1.79709326e-02 -9.89997804e-01
1.37751505e-01 -4.50045735e-01 -3.10313582e-01 2.06320528e-02
2.52596661e-02 1.76661223e-01 -3.15604731e-02 -1.33583859e-01
-3.99878144e-01 2.68943191e-01 -2.84831166e-01 3.46275032e-01
1.36758840e+00 5.44606000e-02 -5.24579048e-01 1.99982405e-01
8.00791025e-01 -3.57480720e-02 -7.57505774e-01 -7.11950779e-01
6.42309010e-01 -1.31724775e-02 -3.60019475e-01 -1.11553645e+00
-6.43535495e-01 4.57612157e-01 -3.45237046e-01 -8.23206007e-02
5.79084635e-01 4.71429974e-01 5.01908958e-01 9.79085267e-01
6.27529442e-01 -8.40794086e-01 3.59229557e-02 1.06936073e+00
8.40236604e-01 -1.13988304e+00 3.10201943e-02 -9.19496179e-01
-4.72663641e-01 9.65481699e-01 1.18019843e+00 2.32955158e-01
6.89570069e-01 1.89688116e-01 -5.30678332e-01 -3.30480486e-01
-1.13000023e+00 -2.57223904e-01 4.96413767e-01 9.71370101e-01
7.83352435e-01 2.35474572e-01 -4.86264415e-02 9.54005659e-01
-4.51911926e-01 -7.72824064e-02 9.11215544e-02 4.27266359e-01
-2.32729226e-01 -1.45706189e+00 -2.66360957e-02 4.62174237e-01
1.04256392e-01 -5.60346186e-01 -5.47411203e-01 8.88150573e-01
1.64225399e-01 6.44690752e-01 -1.03926972e-01 -6.40209556e-01
4.68404144e-01 4.41494137e-01 7.19205260e-01 -7.68763185e-01
-1.60452276e-01 -8.39758277e-01 8.61478090e-01 -5.68938255e-01
-3.20355326e-01 -3.96599263e-01 -1.73301435e+00 -4.65594590e-01
-1.14478670e-01 1.94185287e-01 -1.16708241e-01 1.01383007e+00
5.21014392e-01 7.10044622e-01 -2.23932602e-02 2.97018606e-02
-3.89627278e-01 -9.06863093e-01 -4.07984674e-01 5.10685742e-01
-3.81911308e-01 -9.55348015e-01 1.24214076e-01 1.32501259e-01] | [8.906241416931152, 7.869927406311035] |
16718075-4f02-4020-be30-2fc5a3811cfb | how-well-do-multi-hop-reading-comprehension-1 | 2210.05208 | null | https://arxiv.org/abs/2210.05208v1 | https://arxiv.org/pdf/2210.05208v1.pdf | How Well Do Multi-hop Reading Comprehension Models Understand Date Information? | Several multi-hop reading comprehension datasets have been proposed to resolve the issue of reasoning shortcuts by which questions can be answered without performing multi-hop reasoning. However, the ability of multi-hop models to perform step-by-step reasoning when finding an answer to a comparison question remains unclear. It is also unclear how questions about the internal reasoning process are useful for training and evaluating question-answering (QA) systems. To evaluate the model precisely in a hierarchical manner, we first propose a dataset, \textit{HieraDate}, with three probing tasks in addition to the main question: extraction, reasoning, and robustness. Our dataset is created by enhancing two previous multi-hop datasets, HotpotQA and 2WikiMultiHopQA, focusing on multi-hop questions on date information that involve both comparison and numerical reasoning. We then evaluate the ability of existing models to understand date information. Our experimental results reveal that the multi-hop models do not have the ability to subtract two dates even when they perform well in date comparison and number subtraction tasks. Other results reveal that our probing questions can help to improve the performance of the models (e.g., by +10.3 F1) on the main QA task and our dataset can be used for data augmentation to improve the robustness of the models. | ['Akiko Aizawa', 'Saku Sugawara', 'Xanh Ho'] | 2022-10-11 | null | null | null | null | ['multi-hop-reading-comprehension'] | ['natural-language-processing'] | [ 1.90544754e-01 4.56478745e-01 2.25686207e-01 -6.89113379e-01
-1.22785246e+00 -8.89889956e-01 4.45468545e-01 5.52387238e-01
-4.43426937e-01 5.04629135e-01 3.07563297e-03 -8.32221508e-01
-4.97970402e-01 -1.12049580e+00 -9.69630718e-01 1.03986859e-01
3.30911368e-01 7.69211769e-01 4.76067811e-01 -8.26376677e-01
5.05550086e-01 -8.66767988e-02 -1.67721546e+00 8.35337102e-01
1.48057246e+00 9.59083200e-01 7.93455541e-02 1.09261525e+00
-5.46179771e-01 1.26242709e+00 -1.09802568e+00 -6.66655362e-01
1.07160896e-01 -4.69531775e-01 -1.40599537e+00 -8.00182819e-01
9.29178238e-01 -8.71216834e-01 2.86125112e-02 7.22756505e-01
2.58500218e-01 7.51879513e-02 4.26916242e-01 -1.29169023e+00
-1.03716326e+00 9.71973300e-01 -2.77321935e-01 4.85796630e-01
1.02442169e+00 7.60641247e-02 1.26499450e+00 -4.86156940e-01
3.40998381e-01 1.35550869e+00 7.31370330e-01 4.53511357e-01
-6.59057736e-01 -3.85038644e-01 2.61889189e-01 8.67666006e-01
-8.37894142e-01 -1.92366481e-01 6.05554879e-01 -1.20714292e-01
1.12910950e+00 7.24591434e-01 1.22760970e-03 8.96706760e-01
-2.04818472e-01 8.31977248e-01 1.17832482e+00 -5.20240963e-01
7.15236738e-02 -2.72141874e-01 7.29546785e-01 7.72856295e-01
3.64536699e-03 -3.90929192e-01 -8.02204549e-01 1.04403175e-01
4.38906074e-01 -5.53584099e-01 -4.14478153e-01 4.01340038e-01
-1.01201534e+00 7.91981637e-01 4.14501607e-01 1.38286892e-02
-1.36913449e-01 -1.86369300e-01 1.19089060e-01 7.69170165e-01
-3.32712010e-02 1.00171328e+00 -1.12872744e+00 -3.55513304e-01
-7.33328640e-01 6.81324005e-01 1.13880563e+00 8.95197988e-01
5.05783677e-01 -5.42054355e-01 -5.14318049e-01 7.18239963e-01
-1.37525141e-01 4.62324977e-01 2.70465225e-01 -1.22015297e+00
1.01059389e+00 9.29452896e-01 9.87481028e-02 -8.33014131e-01
-6.85355484e-01 -8.97416621e-02 -2.80681670e-01 -2.14268774e-01
1.13476920e+00 -3.74042094e-02 -8.40095699e-01 1.90196645e+00
1.68113679e-01 -1.23734869e-01 5.01285382e-02 7.18181670e-01
1.38962853e+00 6.07927859e-01 -4.29224856e-02 1.35179907e-01
1.85499442e+00 -1.13133311e+00 -8.07243407e-01 -4.45552319e-01
8.43382418e-01 -7.13914990e-01 1.73205221e+00 4.34942514e-01
-1.47635365e+00 -6.53597057e-01 -9.31087792e-01 -7.35559225e-01
-6.27335191e-01 -1.28107190e-01 7.02537537e-01 6.16956592e-01
-8.48095655e-01 4.02016282e-01 -4.13765877e-01 -1.93087995e-01
7.32485130e-02 4.89287414e-02 -1.79582543e-03 -4.71365392e-01
-1.72207034e+00 1.12233102e+00 2.84107924e-01 -2.54214823e-01
-1.74081177e-01 -1.19150507e+00 -8.86907458e-01 3.56225848e-01
6.46810114e-01 -9.18908596e-01 1.78610349e+00 -2.09485039e-01
-1.17919028e+00 6.62044227e-01 -2.75601685e-01 -4.51548725e-01
5.73515713e-01 -4.39990461e-01 -2.71035850e-01 5.55429459e-01
4.02540505e-01 7.76755631e-01 3.85935068e-01 -8.42948616e-01
-7.16495097e-01 -5.58393955e-01 8.52945983e-01 1.46900028e-01
-4.34848405e-02 -8.20038542e-02 -4.24982220e-01 -5.20799160e-01
3.93048257e-01 -5.12225628e-01 4.78481829e-01 -7.94139728e-02
-2.99686581e-01 -4.97509539e-01 5.78162909e-01 -1.30927885e+00
1.49901795e+00 -1.56372833e+00 -2.18195751e-01 -2.11820245e-01
1.52404055e-01 -1.17431387e-01 -4.37713683e-01 3.01207721e-01
6.06576838e-02 3.03378016e-01 -3.09802890e-01 1.14936516e-01
1.51614591e-01 2.15803638e-01 -3.81501794e-01 -3.20691764e-01
4.70532000e-01 1.07310760e+00 -7.77425945e-01 -3.89823496e-01
-2.44735941e-01 -2.95087665e-01 -7.43802369e-01 2.86503971e-01
-8.30014169e-01 5.52355647e-02 -2.26152778e-01 8.45243454e-01
8.02435458e-01 -3.17577899e-01 -2.14956775e-01 4.35575172e-02
2.92737782e-01 9.10722971e-01 -1.12858975e+00 1.70801795e+00
-4.30469662e-01 4.71775651e-01 -1.59905240e-01 -6.50844276e-01
6.14576101e-01 -1.49917081e-01 -6.70995340e-02 -1.33260489e+00
-6.88352361e-02 9.09646377e-02 3.50489110e-01 -7.60160446e-01
8.13317955e-01 3.90963137e-01 -2.40669906e-01 4.65909451e-01
-2.04898104e-01 -2.79995054e-01 7.86561787e-01 4.85474825e-01
1.39082563e+00 5.81969060e-02 -1.29999578e-01 1.31432027e-01
5.92036188e-01 3.43668640e-01 3.08370709e-01 1.15947211e+00
-6.13777898e-02 6.11700118e-01 7.23888516e-01 -6.28779968e-03
-7.02483475e-01 -1.14882934e+00 -2.57653482e-02 1.50513959e+00
3.62693258e-02 -5.44573724e-01 -7.94869483e-01 -8.20027053e-01
-2.09741239e-02 1.22815025e+00 -5.06394506e-01 -1.75694883e-01
-6.85399771e-01 -5.37324131e-01 9.25907195e-01 7.95210898e-01
9.49819744e-01 -7.32785583e-01 -8.28897595e-01 1.12328947e-01
-9.55668867e-01 -1.25783861e+00 -1.02492049e-01 3.94830629e-02
-7.84258485e-01 -1.47783267e+00 -2.73946434e-01 -6.92425549e-01
4.62653637e-01 1.96551114e-01 1.59475124e+00 5.31745553e-01
1.05936036e-01 5.91312289e-01 -6.00265741e-01 -7.87714064e-01
-3.72264832e-01 3.51822972e-01 -5.35772979e-01 -9.07919824e-01
6.97900236e-01 -1.83771610e-01 -4.48124498e-01 4.17022586e-01
-8.94918799e-01 6.94118738e-02 4.46453005e-01 6.48441613e-01
3.15688550e-02 -7.64064193e-02 7.58325994e-01 -1.02844644e+00
9.49020565e-01 -3.98395509e-01 -4.70208734e-01 6.95628285e-01
-5.22652149e-01 3.84011805e-01 6.77470565e-01 -1.89033538e-01
-1.24807703e+00 -6.36581838e-01 -4.15935308e-01 3.50921303e-01
-7.43528157e-02 6.55482769e-01 -1.21307611e-01 2.69899547e-01
8.31385314e-01 -1.04386378e-02 -2.61109412e-01 -3.84197384e-01
5.57183325e-01 3.84250045e-01 9.07951355e-01 -9.92627680e-01
8.39984655e-01 6.38755411e-02 -3.17205638e-01 -4.19061780e-01
-1.23887181e+00 -2.87904203e-01 -3.52458477e-01 -2.17976701e-02
8.61592472e-01 -7.52394795e-01 -1.07868433e+00 4.86497492e-01
-1.28441477e+00 -3.13681334e-01 -7.73652866e-02 9.09110457e-02
-5.05335808e-01 3.56107026e-01 -9.14231420e-01 -6.19420230e-01
-2.12440550e-01 -7.02778399e-01 8.82195234e-01 3.92191678e-01
-6.16307616e-01 -8.21194053e-01 -2.91208595e-01 1.12295091e+00
4.08113331e-01 -8.41687024e-02 1.87168622e+00 -9.25202787e-01
-8.42610657e-01 8.85067508e-02 -1.39006585e-01 -8.25540442e-03
-2.37395823e-01 -7.43274912e-02 -9.98929024e-01 3.93091887e-01
2.59612083e-01 -6.85135961e-01 8.32786798e-01 1.31696805e-01
1.25326467e+00 -2.65912622e-01 2.78033167e-01 2.39292637e-01
9.55129206e-01 1.50814414e-01 9.98482049e-01 7.06410229e-01
2.71383017e-01 8.17304730e-01 7.59245872e-01 1.62128553e-01
1.23935103e+00 1.56184822e-01 3.78714293e-01 2.34990850e-01
-9.93900746e-03 -3.25343013e-01 3.12502608e-02 7.72082806e-01
3.38088155e-01 -1.99554294e-01 -1.14253628e+00 6.22082591e-01
-1.59005117e+00 -8.53200138e-01 -6.37158871e-01 1.89944208e+00
1.16067171e+00 3.99037808e-01 -2.07851797e-01 5.24239361e-01
2.25984111e-01 3.82793397e-02 -6.02359235e-01 -7.27870226e-01
-2.10166663e-01 4.74222481e-01 -3.06205847e-03 4.72987086e-01
-7.05072761e-01 7.99321890e-01 6.20417023e+00 3.65239769e-01
-5.60947597e-01 -2.93641776e-01 6.04503810e-01 2.82646894e-01
-6.41428888e-01 9.47399884e-02 -6.79398179e-01 7.93397874e-02
9.58553970e-01 1.33253813e-01 4.56040025e-01 4.33909893e-01
-1.87236756e-01 -6.43640995e-01 -1.60139966e+00 6.72737002e-01
1.59467369e-01 -1.29610085e+00 2.12568581e-01 -7.83279002e-01
5.01163781e-01 -7.03429699e-01 -2.66183261e-03 1.03604341e+00
2.62863249e-01 -1.09093177e+00 6.01141691e-01 6.47257805e-01
3.64472121e-01 -4.65837061e-01 7.25269794e-01 7.08342731e-01
-8.69781137e-01 -3.10552448e-01 -1.22140720e-01 -5.59174359e-01
-5.57376351e-03 1.82323262e-01 -7.96881318e-01 7.40848482e-01
9.18029547e-01 -4.45052497e-02 -1.31956708e+00 8.20752561e-01
-6.71569347e-01 4.59236056e-01 -3.76711160e-01 -1.56731308e-01
3.65979178e-03 6.41304478e-02 7.95621574e-02 5.06704926e-01
2.12948218e-01 4.76880044e-01 -2.40752727e-01 9.55360413e-01
-3.25011075e-01 -1.06643513e-01 8.24578777e-02 1.79475602e-02
7.56450295e-01 7.35195339e-01 -4.01626855e-01 -4.21256393e-01
-5.94370365e-01 7.13548124e-01 5.55476844e-01 2.42515475e-01
-1.02569652e+00 -5.97848117e-01 3.86842906e-01 1.01369947e-01
1.42481148e-01 -1.26457855e-01 -8.65824163e-01 -1.13567972e+00
6.05655253e-01 -1.29454279e+00 9.42801118e-01 -1.42679048e+00
-1.32193005e+00 1.36612132e-01 1.59924865e-01 -5.47583461e-01
-3.82535696e-01 -7.03123987e-01 -6.94954455e-01 9.77380693e-01
-1.83125532e+00 -8.56806874e-01 -6.99508369e-01 5.80491304e-01
6.79277301e-01 3.28315437e-01 8.36487412e-01 2.19311774e-01
-3.14003140e-01 8.66753936e-01 -4.48712051e-01 2.50578314e-01
9.94372427e-01 -1.69683969e+00 5.26038766e-01 9.09949601e-01
-1.13659203e-01 8.75088692e-01 6.23452246e-01 -5.61418712e-01
-1.47146285e+00 -6.20422125e-01 1.12568855e+00 -9.90611732e-01
6.59504712e-01 -1.20153308e-01 -1.41983330e+00 5.86916685e-01
2.43469387e-01 -7.95652628e-01 5.15953779e-01 3.98523092e-01
-7.51727164e-01 -1.59060787e-02 -1.22652578e+00 7.64154613e-01
9.71501231e-01 -5.26591480e-01 -1.58482409e+00 4.07145053e-01
1.11311853e+00 -7.78274894e-01 -7.59490609e-01 4.79528725e-01
1.84392959e-01 -9.58214998e-01 9.45919931e-01 -7.94403195e-01
1.01169193e+00 -1.49060056e-01 -6.03486300e-02 -1.06735504e+00
5.79936318e-02 -2.09456012e-01 -3.20287704e-01 1.29615605e+00
7.55701959e-01 -4.86377805e-01 6.94605708e-01 9.89843071e-01
-4.18988727e-02 -6.20272875e-01 -6.50320053e-01 -5.72153509e-01
4.61062223e-01 -6.38436615e-01 8.98219407e-01 9.12478626e-01
-1.38290748e-01 5.58850586e-01 3.70986879e-01 4.80583847e-01
1.22051917e-01 1.48468807e-01 7.89515138e-01 -1.33911836e+00
-3.36021274e-01 -4.53871101e-01 2.46490315e-01 -1.33871770e+00
-3.93398479e-02 -6.56212389e-01 -1.36372909e-01 -2.10354614e+00
-1.92416191e-01 -1.95308790e-01 1.93363130e-01 5.21000624e-01
-6.71270311e-01 -4.81891632e-01 3.00173253e-01 -9.30926651e-02
-4.46040601e-01 1.82944208e-01 1.28313255e+00 -9.67157260e-02
-9.47464406e-02 -3.38953659e-02 -1.10347223e+00 7.30952144e-01
6.92414224e-01 -1.99443758e-01 -6.13208294e-01 -9.22561824e-01
6.90361917e-01 4.06390339e-01 3.62496704e-01 -9.30984199e-01
6.65446341e-01 -1.23684727e-01 4.27374691e-01 -8.67276192e-01
2.09845468e-01 -5.67851067e-01 -7.92569458e-01 1.82836041e-01
-5.64525306e-01 7.39487886e-01 4.30857509e-01 1.82354346e-01
-2.60883689e-01 -6.10010445e-01 2.14999899e-01 -3.84378165e-01
-7.80133009e-01 -3.84035230e-01 -8.53906199e-02 7.60915637e-01
6.57878757e-01 -3.32155190e-02 -1.17012262e+00 -5.78845859e-01
-4.47897911e-01 7.91412413e-01 8.66306946e-03 5.30451298e-01
5.01598239e-01 -8.70154679e-01 -7.91062355e-01 -4.86572012e-02
3.68792444e-01 2.22348541e-01 4.03704405e-01 5.80200672e-01
-7.73326457e-01 5.09701669e-01 -2.47533858e-01 -4.72364902e-01
-1.12881911e+00 4.81838405e-01 2.16356501e-01 -4.93760735e-01
-1.30493402e-01 1.02432323e+00 -3.73478830e-01 -8.95475924e-01
3.91519755e-01 -9.12303329e-01 -3.20840806e-01 2.84677565e-01
6.65615380e-01 5.23335576e-01 5.00167906e-01 3.20738882e-01
-2.06201658e-01 4.74016756e-01 -3.56468886e-01 -1.42363891e-01
1.09301162e+00 -3.04444637e-02 -2.41929293e-01 3.55566949e-01
7.69636095e-01 8.41468871e-02 -7.36269236e-01 -9.30379704e-02
3.19893360e-01 -3.17766100e-01 -5.25888562e-01 -1.57613420e+00
-4.44506764e-01 8.36943865e-01 3.92411277e-02 4.24897909e-01
1.34674537e+00 -1.01957582e-01 1.04044366e+00 9.61387455e-01
6.87849596e-02 -1.27628982e+00 2.88227171e-01 1.03806090e+00
1.00700676e+00 -1.47049117e+00 -2.07780123e-01 -6.20511055e-01
-4.29325312e-01 1.21797264e+00 1.33567619e+00 4.17979598e-01
4.59710807e-02 1.19218439e-01 2.55641997e-01 -2.07801908e-01
-1.00082982e+00 -2.63551176e-01 4.17234391e-01 4.32385981e-01
3.92641813e-01 -3.78984511e-01 -3.41404289e-01 1.02408528e+00
-9.04325902e-01 -3.28683704e-01 7.67146707e-01 1.22268307e+00
-4.63735759e-01 -8.21346760e-01 -6.92983568e-01 6.12213492e-01
-2.53363103e-01 -4.16192174e-01 -7.11248696e-01 1.04071069e+00
-2.11133078e-01 1.39556527e+00 1.32937322e-03 -9.35818851e-02
7.59112179e-01 6.06567860e-01 5.53444803e-01 -4.85938758e-01
-8.76703322e-01 -8.42774212e-01 3.64774883e-01 -3.25182766e-01
1.53215062e-02 -3.29236209e-01 -1.55036521e+00 -5.55750072e-01
-9.39445049e-02 -4.91247103e-02 4.51298356e-01 1.16233802e+00
1.59196019e-01 9.61402893e-01 2.48477869e-02 4.16279465e-01
-9.04098690e-01 -9.45142388e-01 9.46940631e-02 5.01780689e-01
1.91456273e-01 -1.68837875e-01 -3.36518079e-01 -1.27772555e-01] | [10.99449634552002, 7.931440353393555] |
1465c786-81ef-4c3a-b95a-7058c2d402c0 | a-review-of-testing-object-based-environment | 2102.08460 | null | https://arxiv.org/abs/2102.08460v1 | https://arxiv.org/pdf/2102.08460v1.pdf | A Review of Testing Object-Based Environment Perception for Safe Automated Driving | Safety assurance of automated driving systems must consider uncertain environment perception. This paper reviews literature addressing how perception testing is realized as part of safety assurance. We focus on testing for verification and validation purposes at the interface between perception and planning, and structure our analysis along the three axes 1) test criteria and metrics, 2) test scenarios, and 3) reference data. Furthermore, the analyzed literature includes related safety standards, safety-independent perception algorithm benchmarking, and sensor modeling. We find that the realization of safety-aware perception testing remains an open issue since challenges concerning the three testing axes and their interdependencies currently do not appear to be sufficiently solved. | ['Lutz Eckstein', 'Maike Scholtes', 'Michael Hoss'] | 2021-02-16 | null | null | null | null | ['sensor-modeling'] | ['computer-vision'] | [ 3.53169352e-01 3.49330723e-01 -5.82356080e-02 -6.09286845e-01
-4.87181365e-01 -8.91151190e-01 5.52164495e-01 5.30555487e-01
-1.24840371e-01 4.52598542e-01 -1.76743716e-01 -8.82172585e-01
-6.16574883e-01 -7.62174189e-01 -5.44497073e-01 -2.09641859e-01
1.27267361e-01 1.20808408e-01 5.99620283e-01 -4.26923960e-01
3.57641548e-01 6.03148878e-01 -2.36991763e+00 -4.88518290e-02
8.56414676e-01 1.23273718e+00 -2.97484934e-01 6.92163646e-01
4.82860982e-01 4.76604551e-01 -6.28050447e-01 -2.50182629e-01
2.57751703e-01 -1.22130334e-01 -5.50101519e-01 1.08821899e-01
1.71405897e-01 -1.83258325e-01 1.66813806e-01 1.08669233e+00
1.43938735e-01 -1.10302426e-01 2.35953733e-01 -2.23226905e+00
-1.97674930e-02 3.07245068e-02 2.86302954e-01 -1.37502581e-01
3.69769394e-01 7.42061913e-01 3.14603657e-01 -3.63220692e-01
1.84983522e-01 9.71653998e-01 4.01233494e-01 3.86704803e-01
-1.02822042e+00 -4.71433699e-01 1.56325787e-01 3.30822498e-01
-1.27533758e+00 -4.64999497e-01 7.37186968e-01 -7.86351204e-01
1.24419808e+00 5.25911927e-01 7.65589893e-01 1.15613568e+00
7.80229568e-01 6.66063577e-02 1.17454338e+00 -2.55799502e-01
7.45412827e-01 7.21118748e-01 6.20031118e-01 1.29873827e-01
9.29799378e-01 1.02465439e+00 -3.12270433e-01 1.32296942e-02
6.64480552e-02 -8.09326053e-01 7.55404383e-02 -7.16009140e-01
-6.07325912e-01 4.45215017e-01 -2.58637667e-01 -1.04961418e-01
-1.40775472e-01 1.62325934e-01 6.53248966e-01 2.95649439e-01
-2.85106421e-01 2.84246951e-01 -5.18632293e-01 -4.41245705e-01
-1.47846520e-01 4.34079111e-01 6.91706717e-01 1.21917582e+00
6.14924848e-01 3.15337330e-01 -1.63383588e-01 2.73793995e-01
7.02613294e-01 5.70420325e-01 -3.25865448e-01 -1.32543826e+00
1.61827892e-01 4.87067550e-01 5.19555748e-01 -7.74082601e-01
-4.36935723e-01 -2.06549332e-01 -1.09912716e-01 1.02993429e+00
-1.15348130e-01 -2.60864377e-01 -8.13590944e-01 1.56101418e+00
7.86551684e-02 -1.78374097e-01 4.91506219e-01 5.53673863e-01
6.51762068e-01 2.59604067e-01 4.20603693e-01 -3.89024556e-01
1.05856562e+00 -4.73934472e-01 -9.09925163e-01 -6.02150679e-01
3.86291265e-01 -5.05980074e-01 9.82293963e-01 7.66240895e-01
-1.20046151e+00 -9.59463358e-01 -1.71038961e+00 4.84593093e-01
-5.32400966e-01 -7.64063537e-01 4.28133190e-01 1.34757173e+00
-1.17277193e+00 -4.87342989e-03 -7.34534919e-01 -3.22533488e-01
-2.76219368e-01 1.61567524e-01 -1.56444520e-01 1.17310926e-01
-1.30799413e+00 1.39225817e+00 3.03987980e-01 1.66823313e-01
-1.24342096e+00 -5.60896754e-01 -1.25444567e+00 -3.55064332e-01
1.49562210e-01 -4.32322115e-01 1.41199386e+00 -2.88989902e-01
-1.37359810e+00 3.01914752e-01 3.76031213e-02 -3.87903690e-01
5.02371848e-01 -7.76412934e-02 -1.19175184e+00 -3.04625213e-01
3.13955471e-02 3.42999786e-01 3.27433586e-01 -1.87977350e+00
-9.22959924e-01 -9.14174989e-02 1.19951777e-01 -1.86712027e-01
5.00115156e-01 -6.30877167e-02 4.93485965e-02 3.29294592e-01
-3.02448273e-02 -8.48887980e-01 -3.39089096e-01 -4.94163632e-01
-4.50289659e-02 8.50537643e-02 7.50580728e-01 -1.82376727e-01
1.22026956e+00 -2.34730339e+00 -6.65432215e-01 6.05289400e-01
-2.84445137e-01 5.17796129e-02 1.62912831e-01 7.21670806e-01
-1.43591166e-01 -7.25309504e-03 -7.12373853e-02 1.40810505e-01
5.41904807e-01 2.23210841e-01 -3.85003895e-01 3.48202646e-01
2.11294934e-01 5.85103452e-01 -7.28750467e-01 -2.27437973e-01
1.02477968e+00 1.67181298e-01 -2.49210939e-01 1.65585324e-01
-6.07720837e-02 2.28920266e-01 -3.16109926e-01 8.20301890e-01
8.44052732e-01 7.22259641e-01 -4.22249660e-02 -6.84065595e-02
-6.80948853e-01 3.44213158e-01 -1.19635546e+00 1.18377197e+00
-3.09224635e-01 3.38864595e-01 2.74464358e-02 -4.69510227e-01
9.60039735e-01 4.83449191e-01 2.73490846e-01 -1.16643798e+00
3.06939662e-01 2.49780625e-01 3.65422875e-01 -8.85111332e-01
5.07081509e-01 -1.73398584e-01 -5.60009778e-01 -1.48138747e-01
-3.09889346e-01 -6.53716564e-01 2.12176532e-01 -3.71843487e-01
1.20542371e+00 1.75077230e-01 1.26411885e-01 -3.32684040e-01
7.37912536e-01 3.64216805e-01 7.00430870e-01 5.73924422e-01
-9.14844334e-01 7.53464773e-02 5.98753452e-01 -1.78846195e-01
-7.69902647e-01 -1.28979528e+00 -4.33564693e-01 4.61468279e-01
6.17577434e-01 -3.74878347e-01 -7.75989115e-01 -4.71454203e-01
1.78712606e-01 1.72945929e+00 -5.70143461e-01 -6.64693415e-01
-3.72447297e-02 -1.58891365e-01 2.97549427e-01 6.04612708e-01
1.72892004e-01 -9.56372857e-01 -1.53957140e+00 1.19364470e-01
2.05215305e-01 -1.20998323e+00 4.82780069e-01 3.59839499e-01
-4.17293668e-01 -1.19138598e+00 7.39935160e-01 -2.38260031e-01
3.45089942e-01 2.88881898e-01 1.00690877e+00 -1.06910512e-01
1.55847529e-02 9.74103451e-01 -2.76333183e-01 -1.08583379e+00
-9.04099464e-01 -9.22842860e-01 1.70336917e-01 -4.85845149e-01
5.57802141e-01 -4.62533563e-01 -4.14699167e-01 8.46683741e-01
-7.68790305e-01 -4.86058682e-01 5.12199104e-01 1.15912110e-01
5.92547238e-01 4.22765106e-01 6.24249160e-01 -1.87822402e-01
9.32696760e-01 -1.45862728e-01 -1.06488419e+00 3.12547743e-01
-1.08972728e+00 -5.23815155e-01 -3.66910957e-02 1.21278606e-01
-7.64223099e-01 -3.14445719e-02 -2.98298866e-01 -9.78465974e-02
-9.74224985e-01 2.66081929e-01 -6.36090815e-01 -1.86430067e-01
7.18117714e-01 -1.57240495e-01 2.33610868e-01 2.76456505e-01
5.31283161e-03 4.02714014e-01 4.67866570e-01 -5.04618049e-01
7.79649377e-01 1.12614542e-01 1.68133140e-01 -4.57826942e-01
-1.80529431e-01 -1.88651815e-01 -3.11623156e-01 -8.00616562e-01
1.00334251e+00 -4.74974155e-01 -1.20903230e+00 4.30237427e-02
-1.11542082e+00 -3.90527934e-01 -6.68481886e-01 5.76023281e-01
-9.54647303e-01 2.22645298e-01 -2.74204947e-02 -1.40853322e+00
2.41736233e-01 -1.67121422e+00 8.18116665e-01 -2.40343302e-01
-5.72805524e-01 -4.87523109e-01 1.26518369e-01 4.83212858e-01
4.03675854e-01 5.81637204e-01 8.01699460e-01 -3.65908086e-01
-5.45686007e-01 -5.58066368e-01 1.70564428e-01 6.97195768e-01
-9.38313678e-02 2.41996273e-01 -1.31867182e+00 -1.83643028e-01
3.46132547e-01 -1.84803307e-01 1.02552585e-01 2.68412888e-01
8.69112194e-01 1.70492813e-01 -9.00137499e-02 -9.81542468e-02
1.35449731e+00 8.64708424e-01 8.77430022e-01 4.99250680e-01
-2.72179432e-02 1.30136812e+00 1.52328265e+00 1.89891905e-01
2.67499059e-01 4.51003313e-01 1.12644768e+00 1.79566175e-01
1.09979667e-01 -1.25710294e-01 5.05956352e-01 2.64869511e-01
2.07116500e-01 -2.71601677e-01 -1.11571538e+00 4.20293927e-01
-1.74993467e+00 -7.77963459e-01 -3.85696948e-01 2.36134481e+00
-5.11571299e-03 9.69545543e-01 2.44472250e-02 8.08130801e-01
4.38150734e-01 -4.73166943e-01 -5.55708289e-01 -1.08705938e+00
3.65665853e-01 -8.52232203e-02 4.77409571e-01 8.36726904e-01
-9.21706915e-01 4.42492217e-01 7.75756073e+00 -3.56370322e-02
-6.41921878e-01 1.14657342e-01 2.57346153e-01 2.79109865e-01
-5.08770406e-01 2.61715591e-01 -5.38196564e-01 1.73695490e-01
1.37664092e+00 -1.83020055e-01 -1.29718393e-01 9.45146859e-01
5.36209881e-01 -5.16412914e-01 -1.29985237e+00 5.58949411e-01
-1.38356179e-01 -7.26749957e-01 -2.89720327e-01 1.39738604e-01
2.33494982e-01 -2.52555579e-01 1.56682625e-01 4.21930313e-01
3.41809988e-02 -1.03617406e+00 1.40852296e+00 4.35211241e-01
3.94094169e-01 -1.02136970e+00 1.07287240e+00 6.83197901e-02
-8.58894348e-01 -3.57151955e-01 1.38928108e-02 -3.96819115e-01
5.50397098e-01 4.03957963e-01 -7.00695872e-01 6.85496867e-01
8.65686774e-01 -1.66573580e-02 -5.63345432e-01 1.10826695e+00
-2.81827360e-01 4.46158201e-01 1.52193293e-01 -2.73266118e-02
1.43561922e-02 -1.66578650e-01 4.53943253e-01 1.02374864e+00
2.58545846e-01 -1.57034948e-01 2.52895653e-02 8.53018403e-01
1.24350166e+00 -3.29984635e-01 -1.00146651e+00 2.21374631e-01
4.42137241e-01 8.84813666e-01 -5.68872154e-01 6.03556633e-02
-3.69219035e-01 4.98067066e-02 -5.95279217e-01 2.60364115e-01
-1.17052209e+00 -2.42640257e-01 1.22652400e+00 2.85970658e-01
-1.92819178e-01 -4.58742291e-01 -7.68937945e-01 2.50061095e-01
1.82358041e-01 -8.34531784e-01 1.56866550e-01 -8.70647490e-01
-7.76224673e-01 7.30807126e-01 5.91271877e-01 -1.64213920e+00
-3.27033848e-01 -7.58653462e-01 -2.98923612e-01 6.86542332e-01
-1.23429513e+00 -8.55060399e-01 -5.93365312e-01 1.52226448e-01
1.98555663e-01 1.03785940e-01 8.35284531e-01 2.30331659e-01
-3.87883395e-01 4.20075715e-01 -9.57728565e-01 -8.29604506e-01
4.57546651e-01 -8.04207325e-01 2.45267704e-01 1.13833225e+00
-8.18179071e-01 5.23264468e-01 1.42145395e+00 -8.08673322e-01
-1.50891984e+00 -9.80134010e-01 4.73848671e-01 -9.37657177e-01
6.70654178e-01 -4.60789472e-01 -6.15595162e-01 5.06798029e-01
4.37873125e-01 -4.78530407e-01 7.77214885e-01 -1.05333235e-02
-1.60464957e-01 -4.49275970e-01 -1.44446695e+00 5.87174654e-01
1.00047970e+00 -4.47969466e-01 -5.25563896e-01 1.15249353e-02
9.40829694e-01 -2.48904884e-01 -8.85988414e-01 1.10801959e+00
4.28181797e-01 -1.38226461e+00 7.39955604e-01 -2.51787812e-01
-1.55121669e-01 -7.56458700e-01 -5.43035686e-01 -8.60547364e-01
-1.75376803e-01 -4.46518064e-01 3.56833428e-01 1.07417929e+00
7.77330935e-01 -1.04084027e+00 3.66094112e-01 1.13046217e+00
-9.62695420e-01 -3.22481900e-01 -9.70710874e-01 -1.04597330e+00
-2.23393127e-01 -1.49714243e+00 8.01397324e-01 4.51440632e-01
3.91666532e-01 1.22090884e-01 2.54614800e-01 8.27122629e-01
5.80991089e-01 -3.50970089e-01 8.06442797e-01 -1.08728909e+00
8.87991488e-02 -4.54225034e-01 -9.64271009e-01 -7.37060457e-02
-1.59756884e-01 -1.59495130e-01 6.93936229e-01 -1.80455744e+00
-3.52619261e-01 -2.85831481e-01 -3.75981361e-01 3.02430302e-01
4.28224266e-01 -3.50088030e-01 -2.78168440e-01 -6.39341652e-01
-3.01554859e-01 4.02953833e-01 8.47214162e-01 -1.40162691e-01
9.47728287e-03 8.39516297e-02 -8.12641621e-01 6.03073537e-01
1.03477335e+00 -3.50167789e-02 -1.25934064e+00 -7.92053640e-02
2.41166964e-01 1.85083792e-01 7.22924292e-01 -1.47945464e+00
3.02484304e-01 -6.36469901e-01 -2.18111396e-01 -9.01714265e-01
4.97017622e-01 -1.44708717e+00 5.90242445e-01 6.43937230e-01
-1.50786027e-01 5.65268993e-01 8.26766551e-01 4.63825405e-01
-2.89247245e-01 -1.68802187e-01 5.84989607e-01 4.86457109e-01
-1.07105196e+00 -2.21592084e-01 -8.56902778e-01 -5.25167167e-01
1.67346036e+00 -7.74360120e-01 -5.25685847e-01 -4.58253212e-02
-6.69417620e-01 2.99449027e-01 6.05462134e-01 7.41476595e-01
7.96472728e-01 -1.05066383e+00 -1.99115396e-01 4.52743858e-01
6.72061682e-01 -2.08289459e-01 3.36577475e-01 7.37548947e-01
-2.54865497e-01 6.33115113e-01 -5.05235434e-01 -8.00018787e-01
-1.02380323e+00 9.31299150e-01 4.51667994e-01 5.23028433e-01
-1.32551745e-01 4.63140875e-01 1.76545009e-01 -3.40877265e-01
4.79838997e-01 -4.59286451e-01 -1.19368710e-01 -4.73642617e-01
3.22430551e-01 3.81958663e-01 5.90026498e-01 -5.28547049e-01
-7.20728338e-01 3.72426778e-01 6.18094325e-01 -5.66204846e-01
5.74585795e-01 -1.56430304e-01 -3.33396792e-02 7.39441991e-01
4.56993937e-01 -1.39452517e-01 -1.01646709e+00 9.21054065e-01
2.54645348e-01 -4.20963407e-01 -1.19411834e-01 -1.07783175e+00
-4.35391754e-01 6.25433683e-01 1.04282558e+00 6.71432614e-01
1.11242163e+00 -1.66529700e-01 2.23609246e-02 9.21356380e-02
8.03174913e-01 -1.51271737e+00 -1.81819737e-01 2.45680079e-01
1.26539147e+00 -1.03166366e+00 -2.86956996e-01 -9.43732142e-01
-7.65585661e-01 5.31675339e-01 1.13980639e+00 3.81362230e-01
1.08501399e+00 6.76677644e-01 3.75998557e-01 -1.98116571e-01
-9.25527215e-01 -1.90863132e-01 1.06486157e-01 1.11659706e+00
1.00385942e-01 1.66608602e-01 -4.11696970e-01 4.89075363e-01
-3.11918348e-01 -1.17344283e-01 5.22892833e-01 1.26466930e+00
-6.75597250e-01 -1.15072858e+00 -5.58943033e-01 1.35850459e-01
2.60835499e-01 5.02907753e-01 -3.73643994e-01 1.04067338e+00
5.41220129e-01 1.80376804e+00 -2.08139375e-01 -1.14826393e+00
1.25006008e+00 -9.66877192e-02 2.59180337e-01 -4.60969895e-01
-3.90482634e-01 -2.64444113e-01 8.78718436e-01 -9.15637910e-01
-1.37227371e-01 -9.27163959e-01 -1.15006721e+00 -6.24891557e-02
1.30383119e-01 3.50154459e-01 1.13917780e+00 7.00344384e-01
3.16876113e-01 1.08274889e+00 5.95978379e-01 -3.29209954e-01
-5.22341669e-01 -4.40629244e-01 -3.73516351e-01 -1.07693091e-01
2.91373104e-01 -9.91875768e-01 -4.84254032e-01 -2.75730520e-01] | [5.533448696136475, 1.3370546102523804] |
ef3dc062-12f8-45c3-9b2b-9d547154943a | hierarchical-representations-and-explicit | 2108.01176 | null | https://arxiv.org/abs/2108.01176v2 | https://arxiv.org/pdf/2108.01176v2.pdf | Hierarchical Representations and Explicit Memory: Learning Effective Navigation Policies on 3D Scene Graphs using Graph Neural Networks | Representations are crucial for a robot to learn effective navigation policies. Recent work has shown that mid-level perceptual abstractions, such as depth estimates or 2D semantic segmentation, lead to more effective policies when provided as observations in place of raw sensor data (e.g., RGB images). However, such policies must still learn latent three-dimensional scene properties from mid-level abstractions. In contrast, high-level, hierarchical representations such as 3D scene graphs explicitly provide a scene's geometry, topology, and semantics, making them compelling representations for navigation. In this work, we present a reinforcement learning framework that leverages high-level hierarchical representations to learn navigation policies. Towards this goal, we propose a graph neural network architecture and show how to embed a 3D scene graph into an agent-centric feature space, which enables the robot to learn policies for low-level action in an end-to-end manner. For each node in the scene graph, our method uses features that capture occupancy and semantic content, while explicitly retaining memory of the robot trajectory. We demonstrate the effectiveness of our method against commonly used visuomotor policies in a challenging object search task. These experiments and supporting ablation studies show that our method leads to more effective object search behaviors, exhibits improved long-term memory, and successfully leverages hierarchical information to guide its navigation objectives. | ['Luca Carlone', 'J. Daniel Griffith', 'Nathan Hughes', 'Lisa Peng', 'Zachary Ravichandran'] | 2021-08-02 | null | null | null | null | ['2d-semantic-segmentation'] | ['computer-vision'] | [ 7.87840337e-02 4.16091025e-01 -3.61618429e-01 -3.30318362e-01
-4.71412838e-01 -4.65160847e-01 6.74071372e-01 3.86585295e-01
-5.59689701e-01 3.36573869e-01 4.26577538e-01 -2.52290249e-01
-9.31769833e-02 -8.74896169e-01 -9.74207282e-01 -4.23023194e-01
-2.24489316e-01 4.38787878e-01 1.89641997e-01 -1.78012341e-01
4.80309695e-01 6.43835604e-01 -1.64783013e+00 -1.49198472e-01
7.48150766e-01 9.19552803e-01 6.58674896e-01 6.71989858e-01
1.88679155e-02 9.57963467e-01 -1.14921816e-01 5.53191125e-01
3.16446245e-01 -7.06915334e-02 -8.42938721e-01 2.30128169e-01
5.97994030e-01 -6.43689990e-01 -6.59255922e-01 9.70517337e-01
9.65479687e-02 5.66840947e-01 6.38913453e-01 -1.11732781e+00
-6.03463292e-01 1.65646657e-01 -3.02083254e-01 -5.29722609e-02
3.60026568e-01 5.90820074e-01 1.09717333e+00 -4.28608745e-01
5.84461749e-01 1.69575715e+00 3.70740741e-01 4.84865457e-01
-1.37976742e+00 -3.08752418e-01 6.70594633e-01 6.46100044e-02
-8.89552832e-01 -2.84130722e-01 7.46906102e-01 -4.37321603e-01
1.39269948e+00 -3.51431251e-01 9.95787561e-01 1.11102605e+00
4.66156989e-01 9.73132730e-01 9.90639567e-01 -9.18844193e-02
7.16034830e-01 -4.37474787e-01 1.82434320e-01 1.11428320e+00
2.18854755e-01 3.82086426e-01 -6.01889312e-01 -1.67599265e-02
1.20705283e+00 2.61559695e-01 -6.10840507e-02 -1.26120150e+00
-1.05191660e+00 9.41019535e-01 1.18928313e+00 -2.16534287e-02
-5.29774427e-01 8.25862586e-01 -2.52998085e-03 -1.06528394e-01
-1.09287970e-01 7.84983635e-01 -3.05130571e-01 -2.25577295e-01
-2.09771097e-01 3.11503589e-01 5.75113416e-01 1.06465697e+00
1.12650931e+00 2.64600068e-01 -2.83833724e-02 5.91006577e-01
6.95132375e-01 5.84895074e-01 4.16354686e-01 -1.65069139e+00
1.80945814e-01 6.32654905e-01 1.04872085e-01 -1.08786356e+00
-7.72340655e-01 -2.05262363e-01 -2.44275779e-01 5.57815313e-01
9.77624394e-03 3.65467697e-01 -1.60026944e+00 2.05330014e+00
1.70733064e-01 1.00556950e-04 1.35780841e-01 1.12231946e+00
4.16324854e-01 4.71462607e-01 2.76182294e-01 4.24191058e-01
1.03685904e+00 -1.15216994e+00 -2.94736862e-01 -7.23936558e-01
7.73369670e-01 2.17573822e-01 1.26365113e+00 2.12891936e-01
-1.07618344e+00 -4.78154719e-01 -1.06919825e+00 -3.47643554e-01
-4.92856950e-01 -3.36600989e-01 8.77363443e-01 1.50032520e-01
-1.43690181e+00 3.79339844e-01 -1.33658898e+00 -8.03761661e-01
6.65836394e-01 4.55421716e-01 -5.01070380e-01 -1.80096969e-01
-7.09853709e-01 1.07186139e+00 5.39642751e-01 -6.86740130e-02
-1.54029083e+00 -2.89753944e-01 -1.44761646e+00 4.87979464e-02
3.85417372e-01 -1.00260341e+00 1.57952750e+00 -2.68595219e-01
-1.59207654e+00 5.64484715e-01 -1.52227104e-01 -5.13706684e-01
-1.12555541e-01 -3.90122354e-01 5.60714960e-01 3.73917907e-01
2.78641701e-01 1.33931005e+00 7.95960009e-01 -1.42347443e+00
-7.02310741e-01 -6.48876667e-01 6.06566966e-01 6.25690460e-01
-1.14672095e-01 -1.03259540e+00 -6.84735060e-01 -1.24103174e-01
5.04680216e-01 -1.12328339e+00 -7.85416007e-01 1.54584378e-01
-2.90752560e-01 -1.27837166e-01 6.63868308e-01 -3.75871539e-01
3.54241997e-01 -2.04442334e+00 5.12104869e-01 1.66239336e-01
3.68010193e-01 -4.10310686e-01 -2.25366250e-01 1.33208945e-01
5.22468984e-01 7.68571720e-02 -2.27464393e-01 -3.51287276e-01
2.40135640e-01 6.05683267e-01 -2.71396488e-01 4.95784670e-01
7.79062808e-02 1.19159138e+00 -1.20573509e+00 -2.48945542e-02
6.14689410e-01 3.19156438e-01 -9.78392005e-01 1.98549286e-01
-6.63956225e-01 5.87787032e-01 -7.94458330e-01 4.93591189e-01
4.04913686e-02 -2.49430075e-01 4.54866737e-02 -2.65458524e-02
4.71891575e-02 4.80289757e-01 -6.31602585e-01 2.44242620e+00
-7.20945716e-01 5.20561934e-01 1.37217939e-01 -8.44881237e-01
7.44913816e-01 -3.18011642e-01 2.70283014e-01 -1.23752117e+00
8.94262865e-02 -2.87483260e-02 -2.89565206e-01 -2.48698220e-01
6.29850566e-01 2.99313009e-01 -4.39149201e-01 1.53204501e-01
1.41355172e-01 -7.75726318e-01 -1.40217856e-01 3.77793252e-01
1.33297479e+00 5.60209215e-01 2.39246175e-01 -8.90434608e-02
-8.10389668e-02 3.16119611e-01 2.70309508e-01 1.18311167e+00
-2.83243835e-01 1.98834360e-01 3.18939984e-01 -3.01791906e-01
-9.75143433e-01 -1.27386951e+00 3.74281198e-01 1.06049275e+00
6.11908674e-01 -3.32412779e-01 -4.58650738e-01 -4.40529257e-01
3.00056189e-01 8.57936263e-01 -7.90742218e-01 -6.10039592e-01
-5.52687943e-01 -3.96636873e-02 1.52295053e-01 7.28427052e-01
5.52642345e-01 -1.04561281e+00 -1.35789180e+00 2.79573351e-01
2.17717320e-01 -1.27128947e+00 -1.62901357e-01 6.69948041e-01
-1.17946386e+00 -1.04378867e+00 -1.46621808e-01 -7.84627438e-01
8.31962049e-01 5.59645593e-01 9.83781338e-01 -3.53075936e-02
-4.11608964e-01 9.96401250e-01 -3.31051409e-01 7.44691715e-02
-6.37947917e-02 2.15739230e-04 1.25323012e-01 -7.05244839e-01
5.51790968e-02 -7.18361378e-01 -6.95566177e-01 -4.00165431e-02
-7.60294199e-01 1.52049869e-01 8.23908806e-01 7.32129037e-01
6.92193389e-01 -2.16730937e-01 1.68812722e-01 -5.79871595e-01
5.48550665e-01 -3.39362025e-01 -7.23981559e-01 -3.10710669e-01
-5.48455119e-01 6.01925790e-01 4.11591470e-01 -4.44152728e-02
-7.81943858e-01 2.86273390e-01 1.09171346e-01 -6.09536707e-01
-4.55716759e-01 5.41138709e-01 -4.37353440e-02 -5.05948626e-02
6.69373274e-01 2.26290196e-01 1.35594411e-02 -2.65572429e-01
8.68877530e-01 1.69219881e-01 6.23930991e-01 -7.74434328e-01
5.37850380e-01 5.31310916e-01 8.61219019e-02 -7.88827181e-01
-6.60091519e-01 -4.59798485e-01 -6.69835687e-01 6.27621934e-02
9.64272141e-01 -1.05651832e+00 -9.81300771e-01 2.28492424e-01
-9.30870652e-01 -1.04273868e+00 -2.20347807e-01 4.99493569e-01
-1.26804113e+00 -1.01718850e-01 -6.13828540e-01 -8.06715250e-01
3.14188927e-01 -1.29985905e+00 1.47490382e+00 2.51163006e-01
-1.93057969e-01 -8.64554524e-01 -3.11194342e-02 7.47847855e-02
1.55563444e-01 2.65381157e-01 1.15805197e+00 -1.45731285e-01
-1.02432621e+00 1.63836032e-01 -2.15658784e-01 -2.88389206e-01
1.49248019e-01 -6.02166057e-01 -9.24135566e-01 -3.89431179e-01
-2.32103527e-01 -6.86886311e-01 1.12862933e+00 4.73744750e-01
1.18985248e+00 -1.98907375e-01 -5.43300927e-01 7.95719147e-01
1.29694879e+00 2.47209534e-01 2.59690225e-01 5.57079375e-01
8.62935066e-01 6.16390467e-01 5.39726317e-01 2.66022205e-01
7.72391081e-01 5.37872970e-01 1.08699286e+00 8.74959230e-02
-6.21712208e-02 -6.96611047e-01 3.30085695e-01 2.38332868e-01
4.48069602e-01 -2.73294710e-02 -9.80988681e-01 5.18503904e-01
-2.14518380e+00 -5.64547479e-01 5.22188067e-01 1.84483254e+00
5.12498736e-01 4.00395274e-01 -1.07023366e-01 -4.43875134e-01
5.28445132e-02 4.34754282e-01 -1.23911774e+00 -4.06294465e-01
1.86974004e-01 -2.30753258e-01 7.51378417e-01 7.74248958e-01
-1.02545202e+00 1.53339672e+00 6.16783476e+00 1.20848119e-01
-9.27623749e-01 -1.75084502e-01 1.76133230e-01 -1.15153424e-01
-5.38812459e-01 -7.36381188e-02 -6.21578217e-01 -1.87792584e-01
7.87265182e-01 2.07339376e-01 7.43970335e-01 1.09082139e+00
2.79241294e-01 -3.71876866e-01 -1.46815574e+00 1.06646097e+00
-5.42919077e-02 -1.28481352e+00 1.89105660e-01 3.62432629e-01
4.48243141e-01 3.11867267e-01 3.32139164e-01 6.92043364e-01
1.15503883e+00 -1.26938438e+00 8.39944661e-01 2.56163150e-01
3.05098593e-01 -5.43534040e-01 6.69038622e-03 5.33569217e-01
-1.13779056e+00 -4.04960573e-01 -3.60189736e-01 -2.57475734e-01
4.16512676e-02 -2.40742713e-02 -1.03130007e+00 1.43104166e-01
7.13871300e-01 1.02280974e+00 -4.96804059e-01 8.64520073e-01
-3.85617822e-01 1.10195719e-01 -4.28489536e-01 -1.57105967e-01
8.41450691e-01 7.46053010e-02 4.09205288e-01 7.44224072e-01
3.90220471e-02 2.34457701e-01 6.56070113e-01 1.01978970e+00
3.02717108e-02 -4.43445534e-01 -1.10262477e+00 -1.42027006e-01
4.82383400e-01 8.14838588e-01 -7.54827559e-01 -1.07873887e-01
-1.78683758e-01 1.07048869e+00 8.43331635e-01 8.02724838e-01
-4.44575965e-01 7.33899102e-02 1.01646876e+00 -3.18757653e-01
3.77955407e-01 -9.56370771e-01 -4.12981600e-01 -8.69873822e-01
-2.81427115e-01 -6.46006048e-01 1.20077759e-01 -1.07943058e+00
-7.39860952e-01 3.34722430e-01 -1.20835021e-01 -7.23577261e-01
-5.80617666e-01 -9.28263247e-01 -1.46140575e-01 7.00754642e-01
-1.58835661e+00 -1.02894139e+00 -4.88855749e-01 6.21406138e-01
7.12451339e-01 5.59524633e-02 9.40632641e-01 -4.84148562e-01
-9.08771381e-02 4.00050208e-02 -1.83341771e-01 -2.01327959e-04
1.23705596e-01 -1.34341371e+00 5.09259999e-01 5.18754542e-01
3.40090334e-01 7.92531729e-01 6.75010085e-01 -6.88031614e-01
-1.97838950e+00 -8.52339685e-01 -1.86200261e-01 -6.36908829e-01
4.52526242e-01 -4.91647094e-01 -6.05860412e-01 9.24654424e-01
-1.46344036e-01 -7.26712355e-03 2.75154561e-01 3.35927159e-01
-5.58924913e-01 1.99227303e-01 -9.18850720e-01 1.00715411e+00
1.40363598e+00 -8.25664878e-01 -7.30029047e-01 1.11495860e-01
1.18022704e+00 -5.92325985e-01 -4.93252397e-01 3.57472539e-01
5.90999305e-01 -7.53573239e-01 1.08623934e+00 -7.59982824e-01
2.33261883e-01 -2.87962645e-01 -5.61075032e-01 -1.54224563e+00
-5.58521569e-01 -2.19509363e-01 -3.80670756e-01 2.12088749e-01
2.64370143e-01 -3.96492928e-01 1.02231288e+00 7.30125368e-01
-3.84027183e-01 -6.82644129e-01 -7.74463475e-01 -6.05572104e-01
-4.41284589e-02 -5.92244744e-01 2.01931834e-01 4.86581385e-01
-8.56057331e-02 4.14174557e-01 4.04749624e-02 5.27843714e-01
7.28747487e-01 3.22084635e-01 8.73872995e-01 -1.16927660e+00
-2.34310851e-02 -6.00145996e-01 -6.62210941e-01 -1.74959815e+00
5.94475389e-01 -8.36575925e-01 6.85239077e-01 -2.09509110e+00
-7.64352456e-02 -5.75455725e-01 -3.87512147e-01 7.05985963e-01
1.20225906e-01 -1.53807044e-01 3.05567473e-01 1.18746288e-01
-8.04190576e-01 9.63542640e-01 1.26226854e+00 -3.91378701e-01
-4.42526460e-01 -5.40525675e-01 -7.42961049e-01 7.82987118e-01
8.10016155e-01 -9.83946249e-02 -7.25267470e-01 -8.93428564e-01
-5.56310527e-02 2.61154654e-03 6.07488394e-01 -1.08787882e+00
4.44791466e-01 -3.21386844e-01 5.16470015e-01 -5.04778683e-01
8.23978841e-01 -8.85404348e-01 -6.16845310e-01 6.11845016e-01
-5.28844357e-01 -7.42770284e-02 3.88729244e-01 1.06906259e+00
6.87908977e-02 2.29314759e-01 4.56058979e-01 -3.70224297e-01
-1.44029033e+00 5.10785401e-01 -4.55084801e-01 -7.50608221e-02
8.33699644e-01 -4.74157751e-01 -2.66367137e-01 -4.59290177e-01
-7.52158463e-01 6.27858818e-01 8.60731959e-01 7.33317196e-01
8.70588183e-01 -1.15289259e+00 -7.80905113e-02 2.55473346e-01
2.65646249e-01 4.29329216e-01 -4.94354777e-02 4.48548496e-01
-4.30341780e-01 6.75229728e-01 -5.91456950e-01 -1.01242805e+00
-5.58969498e-01 6.38559282e-01 3.24353784e-01 1.89717337e-01
-8.66721272e-01 8.86078119e-01 7.76550412e-01 -7.58407950e-01
4.15264487e-01 -7.06450045e-01 -7.09320158e-02 -3.71671408e-01
8.65931809e-02 -1.68370888e-01 -3.71328533e-01 -5.35393536e-01
-3.22065413e-01 5.47573984e-01 3.21385600e-02 -4.44397539e-01
1.25101399e+00 -4.16952044e-01 2.64517039e-01 6.49144888e-01
1.14197624e+00 -3.51881534e-01 -2.04138947e+00 -2.10611552e-01
1.22250915e-01 -4.71066594e-01 3.03218395e-01 -6.00193322e-01
-7.00340807e-01 9.27272320e-01 4.28926885e-01 -1.00952201e-01
6.56881869e-01 2.14608014e-01 6.60069346e-01 9.70327556e-01
9.36500251e-01 -9.85050380e-01 5.98101079e-01 8.72770011e-01
7.27657318e-01 -1.30968904e+00 -2.38523588e-01 -1.11828476e-01
-4.99179155e-01 9.28730011e-01 9.86209035e-01 -3.10367435e-01
3.25814962e-01 -5.86979166e-02 -1.31391481e-01 -3.38729352e-01
-8.42509449e-01 -5.52035928e-01 2.46575195e-02 8.71405721e-01
-1.89162821e-01 -4.59769927e-02 6.76431894e-01 1.12370290e-01
-2.67070562e-01 -4.23816890e-01 3.20266336e-01 1.28133905e+00
-1.04689670e+00 -6.04507685e-01 -2.89118178e-02 3.76442164e-01
3.51400614e-01 7.31135756e-02 -3.00896525e-01 7.20902443e-01
-3.69065225e-01 8.64930093e-01 1.62295818e-01 -3.95535409e-01
3.91537666e-01 -7.39302561e-02 7.35854805e-01 -9.86318231e-01
4.43948470e-02 -2.30031922e-01 -8.34141448e-02 -1.31080616e+00
-1.94736719e-01 -4.29383814e-01 -1.90703142e+00 1.06563844e-01
2.75691748e-01 -4.14033607e-02 6.97958291e-01 8.84767830e-01
4.80612874e-01 7.04186738e-01 2.10249424e-01 -1.24965084e+00
-3.34173828e-01 -5.28842807e-01 -3.67569685e-01 3.78351092e-01
8.21446359e-01 -1.12450778e+00 -1.10357516e-01 -1.04674347e-01] | [4.57877254486084, 0.6187423467636108] |
d9e700fe-9775-42e9-98f7-3c945752911f | l-verse-bidirectional-generation-between | 2111.11133 | null | https://arxiv.org/abs/2111.11133v10 | https://arxiv.org/pdf/2111.11133v10.pdf | L-Verse: Bidirectional Generation Between Image and Text | Far beyond learning long-range interactions of natural language, transformers are becoming the de-facto standard for many vision tasks with their power and scalability. Especially with cross-modal tasks between image and text, vector quantized variational autoencoders (VQ-VAEs) are widely used to make a raw RGB image into a sequence of feature vectors. To better leverage the correlation between image and text, we propose L-Verse, a novel architecture consisting of feature-augmented variational autoencoder (AugVAE) and bidirectional auto-regressive transformer (BiART) for image-to-text and text-to-image generation. Our AugVAE shows the state-of-the-art reconstruction performance on ImageNet1K validation set, along with the robustness to unseen images in the wild. Unlike other models, BiART can distinguish between image (or text) as a conditional reference and a generation target. L-Verse can be directly used for image-to-text or text-to-image generation without any finetuning or extra object detection framework. In quantitative and qualitative experiments, L-Verse shows impressive results against previous methods in both image-to-text and text-to-image generation on MS-COCO Captions. We furthermore assess the scalability of L-Verse architecture on Conceptual Captions and present the initial result of bidirectional vision-language representation learning on general domain. | ['Kyunghoon Bae', 'Honglak Lee', 'Seung Hwan Kim', 'Soonyoung Lee', 'Yewon Seo', 'Sangyun Kim', 'Sihaeng Lee', 'Gwangmo Song', 'TaeHoon Kim'] | 2021-11-22 | null | http://openaccess.thecvf.com//content/CVPR2022/html/Kim_L-Verse_Bidirectional_Generation_Between_Image_and_Text_CVPR_2022_paper.html | http://openaccess.thecvf.com//content/CVPR2022/papers/Kim_L-Verse_Bidirectional_Generation_Between_Image_and_Text_CVPR_2022_paper.pdf | cvpr-2022-1 | ['zero-shot-text-to-image-generation'] | ['natural-language-processing'] | [ 2.39320889e-01 2.22411454e-01 2.49002159e-01 -3.24157923e-01
-1.01920807e+00 -6.11304700e-01 1.33708251e+00 -7.64754951e-01
-2.80874997e-01 5.78718662e-01 2.75175124e-01 -2.09269613e-01
3.78814787e-01 -7.65459895e-01 -1.34004545e+00 -8.43550444e-01
6.35317922e-01 6.81822956e-01 -1.56822009e-03 -4.84408587e-01
-3.16464871e-01 1.73917234e-01 -1.46395445e+00 6.37073278e-01
3.57458591e-01 1.20136213e+00 5.23059130e-01 8.94141734e-01
-7.43219852e-02 1.03525198e+00 -4.66394424e-01 -6.88142061e-01
3.72381032e-01 -6.28904939e-01 -6.68006599e-01 2.82219827e-01
8.66466403e-01 -5.52129805e-01 -7.35156298e-01 7.76730180e-01
5.49902201e-01 -5.00773676e-02 9.94813204e-01 -1.59939253e+00
-1.26346624e+00 5.40166974e-01 -4.95395780e-01 -1.67255178e-01
2.87601084e-01 7.41146266e-01 1.04372621e+00 -1.29674029e+00
9.36207235e-01 1.57096815e+00 4.07214910e-01 9.48352158e-01
-1.54883003e+00 -6.82730913e-01 -3.31148922e-01 1.96375117e-01
-1.05457354e+00 -5.96461296e-01 4.55979019e-01 -5.51799357e-01
1.15378547e+00 9.76781324e-02 7.63419867e-01 2.14467335e+00
2.75528312e-01 1.10052025e+00 1.18880546e+00 -2.29334861e-01
-1.42641753e-01 2.71826148e-01 -6.03348613e-01 6.06862843e-01
-3.46343577e-01 6.21126771e-01 -7.95302391e-01 2.78535485e-01
1.00088227e+00 -3.88893485e-01 -3.41066480e-01 -2.99794406e-01
-1.44723177e+00 1.07085025e+00 7.10683584e-01 -7.91345537e-02
-5.01449108e-01 6.99006617e-01 1.95782900e-01 8.56612995e-02
7.60763511e-02 4.19840664e-01 -9.26555842e-02 1.66755752e-03
-9.84778106e-01 3.69986057e-01 4.18587416e-01 9.53979790e-01
4.66804177e-01 7.25416183e-01 -6.87916815e-01 6.91103101e-01
3.15696090e-01 1.05541158e+00 5.28993845e-01 -9.90588307e-01
4.59163308e-01 2.80348659e-01 -5.19558005e-02 -3.45311314e-01
1.58391193e-01 -2.23675698e-01 -1.04529762e+00 4.22785670e-01
2.53776222e-01 -1.07497945e-01 -1.42691445e+00 1.69458508e+00
6.57788515e-02 -1.02970256e-02 4.45197850e-01 1.02094591e+00
1.12418568e+00 1.16686130e+00 7.59306923e-02 3.33647765e-02
1.26097536e+00 -9.66626167e-01 -5.53706765e-01 -2.87752211e-01
-1.13510462e-02 -7.53281713e-01 1.15533388e+00 2.22101346e-01
-1.23431015e+00 -7.62116134e-01 -6.81922436e-01 -5.39379537e-01
-3.17490518e-01 1.47219107e-01 4.06899124e-01 2.64826417e-01
-1.38878930e+00 1.45889014e-01 -6.17230952e-01 -3.20234895e-01
4.67637658e-01 3.05557121e-02 -5.95468044e-01 1.43174529e-02
-1.08088052e+00 8.43841136e-01 3.65161479e-01 -3.94343287e-02
-1.53448868e+00 -8.77166748e-01 -1.08147800e+00 -1.69223830e-01
-1.27329350e-01 -1.23022664e+00 1.14044893e+00 -1.14019716e+00
-1.61279535e+00 9.20333862e-01 -6.87461570e-02 -7.88864195e-01
8.44249606e-01 5.04056998e-02 -1.28083527e-01 2.26902217e-01
3.11829209e-01 1.67618155e+00 1.42044711e+00 -1.47670662e+00
-4.83810782e-01 -2.62821559e-02 -1.71476733e-02 2.35747069e-01
-3.79686314e-03 -3.94428939e-01 -4.65079367e-01 -7.07771540e-01
-2.54566401e-01 -8.90726745e-01 1.88427687e-01 2.21706733e-01
-6.00413680e-01 -4.31593567e-01 1.07222939e+00 -5.98446190e-01
2.02075168e-01 -2.00376606e+00 4.05021727e-01 -4.32507575e-01
1.88034609e-01 1.66612387e-01 -5.87829292e-01 3.13156217e-01
-2.71735042e-01 -4.89236191e-02 -1.75884008e-01 -5.64464688e-01
1.21312380e-01 4.03078467e-01 -8.65073323e-01 2.08034351e-01
6.38995290e-01 1.63596940e+00 -8.21177840e-01 -5.60567975e-01
6.30095184e-01 8.64710152e-01 -3.80160898e-01 4.67593193e-01
-5.14938116e-01 4.87332970e-01 -8.78440663e-02 4.00245816e-01
3.13504577e-01 -4.27802205e-01 -4.03501630e-01 -5.66953897e-01
2.41592944e-01 -2.00994536e-01 -4.22961354e-01 1.59035051e+00
-6.62683666e-01 1.17286706e+00 -1.47951230e-01 -7.02182472e-01
5.14347672e-01 3.56927335e-01 2.46103182e-01 -1.04674995e+00
1.01000614e-01 -5.26812226e-02 -2.14258686e-01 -4.10380542e-01
7.13542938e-01 -2.56840259e-01 7.38834292e-02 1.75366119e-01
5.47683418e-01 -7.08355665e-01 1.74931705e-01 5.41668355e-01
6.70126677e-01 4.72096711e-01 -1.46897674e-01 1.43725291e-01
1.75707370e-01 8.81351456e-02 -1.51311427e-01 7.82446146e-01
4.80363518e-02 1.08938706e+00 2.20862657e-01 -1.41834557e-01
-1.56157434e+00 -1.49372137e+00 -7.37952366e-02 9.60820198e-01
-1.17637023e-01 -1.35971189e-01 -6.63317502e-01 -5.10122001e-01
-1.26442373e-01 1.23603570e+00 -8.69814873e-01 -7.25263655e-02
-1.32734001e-01 -4.02799875e-01 9.12430048e-01 4.41622764e-01
6.06457591e-01 -1.36954451e+00 -4.40947264e-01 -2.93787313e-03
-5.49758255e-01 -1.58338690e+00 -3.84978414e-01 4.40315008e-02
-3.87068868e-01 -7.48457372e-01 -1.04864764e+00 -6.78731859e-01
4.05935764e-01 5.51784225e-02 1.30295050e+00 -4.79859680e-01
-4.33903098e-01 9.00046229e-01 -2.55595744e-01 -3.01808000e-01
-8.61346662e-01 -4.72483516e-01 -3.57378386e-02 1.94009438e-01
-2.38718301e-01 -3.73140246e-01 -6.32598341e-01 1.56025887e-01
-1.13272238e+00 4.14713144e-01 7.13671327e-01 1.28754127e+00
6.58212960e-01 -4.71233368e-01 4.97591570e-02 -2.97184020e-01
4.76184487e-01 -3.15865993e-01 -6.18791759e-01 2.75957465e-01
-2.89227098e-01 2.70160198e-01 6.05671644e-01 -5.62813044e-01
-1.17829919e+00 1.98288590e-01 -1.30891189e-01 -1.08282804e+00
4.86679077e-02 3.53279598e-02 9.46923345e-02 1.29134357e-01
9.51789320e-01 9.46862340e-01 -6.58648536e-02 2.65909284e-01
1.01770151e+00 5.43471754e-01 1.02312541e+00 -4.80463356e-01
9.79799271e-01 6.25230014e-01 -7.69220386e-03 -1.03816175e+00
-6.53250098e-01 9.41500627e-03 -2.42177933e-01 -2.07763404e-01
1.51405561e+00 -1.34206915e+00 -7.25971222e-01 5.70228577e-01
-1.46983016e+00 -8.02184343e-01 -4.69719082e-01 8.31980705e-02
-8.21392894e-01 3.19680274e-02 -4.75337356e-01 -6.48218870e-01
-4.44539398e-01 -1.41715813e+00 1.61799419e+00 1.29761696e-01
1.32046252e-01 -9.56618965e-01 -2.52452970e-01 7.46662855e-01
3.34689528e-01 2.99555957e-01 5.65235794e-01 -6.32445589e-02
-7.70732880e-01 2.47172147e-01 -4.67328668e-01 6.07091963e-01
-3.67157936e-01 -3.33160013e-02 -1.12981820e+00 -3.61195207e-01
-3.59554082e-01 -1.06729758e+00 1.09088123e+00 4.45226878e-01
8.76816630e-01 -2.94602603e-01 -8.09649453e-02 8.53558004e-01
1.27582753e+00 -2.53217429e-01 9.58111048e-01 1.20076686e-01
1.05086279e+00 3.41115564e-01 2.23393917e-01 1.56921491e-01
5.57761192e-01 7.26236880e-01 6.85297668e-01 -2.95522839e-01
-8.11058998e-01 -6.40327156e-01 8.02448988e-01 2.94968128e-01
1.55767441e-01 -7.59993076e-01 -7.17480123e-01 6.65292084e-01
-1.60272610e+00 -1.05323064e+00 -2.02971712e-01 1.64761627e+00
7.32427001e-01 -1.76112384e-01 -1.98641613e-01 -3.69743913e-01
4.35313046e-01 2.50806183e-01 -7.03785956e-01 -1.59268484e-01
-5.26216507e-01 1.53385669e-01 4.20226455e-01 3.53240460e-01
-8.58459651e-01 1.43135881e+00 6.02971220e+00 1.07338667e+00
-1.43944454e+00 1.92714185e-01 7.17480242e-01 -1.76595584e-01
-3.77491355e-01 -2.75049746e-01 -6.65334761e-01 1.62817806e-01
8.97130549e-01 1.56748652e-01 6.11464381e-01 7.60658145e-01
4.11572270e-02 1.43571854e-01 -1.24693787e+00 1.52584076e+00
1.54175997e-01 -1.66879439e+00 8.24504673e-01 2.05233544e-02
8.92391860e-01 4.25370425e-01 4.98631835e-01 5.22374213e-01
7.25865185e-01 -1.33496714e+00 1.23989785e+00 3.96251470e-01
1.37410343e+00 -1.88084289e-01 3.30909133e-01 5.05776554e-02
-9.89855230e-01 7.93169588e-02 -3.68445337e-01 4.35719818e-01
3.34090441e-01 1.77303836e-01 -1.03133500e+00 4.46953446e-01
8.92224848e-01 7.74641275e-01 -4.87559378e-01 1.88056782e-01
-4.33422625e-01 5.16109228e-01 -2.35874236e-01 1.65915787e-01
5.62551498e-01 -3.65151837e-02 7.58154631e-01 9.73189235e-01
2.93676555e-01 -4.37687375e-02 -1.02296300e-01 1.49355245e+00
-2.79904425e-01 -3.21003318e-01 -7.49461293e-01 -3.96035343e-01
1.86695866e-02 1.21461558e+00 -3.07899177e-01 -4.22162473e-01
-1.08251289e-01 1.56712937e+00 1.58880591e-01 6.75801158e-01
-9.28213537e-01 2.66561527e-02 5.41764259e-01 8.26009065e-02
5.05153537e-01 -9.06093717e-02 2.10171491e-01 -1.18634093e+00
-1.16622329e-01 -9.15819466e-01 -2.75706928e-02 -1.67239213e+00
-1.48334527e+00 9.07743454e-01 -2.24955734e-02 -1.01115060e+00
-6.60711050e-01 -8.60939264e-01 -3.73121381e-01 7.87083626e-01
-1.30755746e+00 -1.92822981e+00 -4.10545141e-01 1.10913134e+00
8.88608396e-01 -3.68119866e-01 6.00780189e-01 -1.86135158e-01
-7.54681528e-02 6.00776255e-01 -5.59536070e-02 4.03364331e-01
5.85780263e-01 -1.30956709e+00 5.77177823e-01 9.39289510e-01
6.60506368e-01 1.98927715e-01 7.66073883e-01 -4.20373112e-01
-1.59951758e+00 -1.33061981e+00 2.77579784e-01 -8.44427049e-01
6.62873507e-01 -6.97941780e-01 -3.65445018e-01 8.58365297e-01
7.24517107e-01 2.44053677e-01 1.32939769e-02 -6.02810740e-01
-5.97807169e-01 3.59451510e-02 -8.11028719e-01 7.75455356e-01
8.55841339e-01 -8.33702803e-01 -4.71315503e-01 5.38738132e-01
9.73373115e-01 -7.13065803e-01 -6.70747340e-01 1.30130380e-01
6.26422763e-01 -1.11334562e+00 1.19544458e+00 -3.75273377e-01
1.16317475e+00 -1.57017559e-01 -4.58948016e-01 -1.35124350e+00
-1.28166616e-01 -7.29682148e-01 1.38427928e-01 1.14248145e+00
3.95077229e-01 -3.91678691e-01 4.95230556e-01 2.44626150e-01
-3.42773795e-02 -4.59856719e-01 -1.01214981e+00 -6.36119723e-01
1.33656561e-01 -6.66380525e-01 2.20192835e-01 6.75106943e-01
-7.64144242e-01 8.47876132e-01 -6.65303528e-01 -6.18190095e-02
7.96306193e-01 -1.79802120e-01 1.06916130e+00 -6.30742788e-01
-5.60282409e-01 -3.91650885e-01 -3.42490882e-01 -1.08803964e+00
2.68816054e-01 -1.10291052e+00 2.52682805e-01 -1.73130822e+00
9.92459580e-02 3.21927145e-02 1.05572835e-01 5.46853781e-01
4.46529463e-02 7.30455577e-01 5.95600247e-01 1.49870455e-01
-4.12414402e-01 1.04567039e+00 1.61522925e+00 -5.70330143e-01
1.07656449e-01 -5.13116837e-01 -3.87130678e-01 2.36161441e-01
1.66162953e-01 -3.31640571e-01 -5.14471292e-01 -5.98334312e-01
2.55901068e-01 3.48663807e-01 8.72370541e-01 -7.49624550e-01
-5.46425954e-02 1.14168255e-02 7.63251126e-01 -5.08260310e-01
7.94096529e-01 -5.84429443e-01 2.46948406e-01 2.63316661e-01
-2.87449747e-01 -7.82430321e-02 1.49292722e-01 6.86797857e-01
-1.56506926e-01 2.91770548e-01 7.69844294e-01 -2.17806399e-01
-7.94016123e-01 4.13751125e-01 -2.13073090e-01 7.36331120e-02
9.07745481e-01 -1.99320570e-01 -4.10751462e-01 -8.56694400e-01
-5.92157304e-01 1.72946423e-01 4.17215824e-01 8.52383614e-01
1.02715445e+00 -1.29284132e+00 -1.07268631e+00 1.84707046e-01
3.58736336e-01 -7.68532790e-03 2.45414793e-01 5.47807693e-01
-3.57495189e-01 4.65059817e-01 -2.43494570e-01 -1.22415757e+00
-1.05462289e+00 4.62326854e-01 5.08506954e-01 -1.15115985e-01
-5.69702446e-01 1.12562788e+00 7.37016678e-01 -3.77689004e-01
-1.09778838e-02 -8.27530324e-02 1.25214890e-01 -7.17461631e-02
3.27247232e-01 -1.60311535e-01 -3.66097569e-01 -1.04144645e+00
2.59173978e-02 3.88916880e-01 1.01484455e-01 -6.76957369e-01
1.21498430e+00 -6.82902038e-02 -2.20923945e-02 3.26742321e-01
1.16345024e+00 -4.94200557e-01 -1.60796380e+00 -2.35181558e-03
-8.27205420e-01 -9.65422913e-02 3.00632477e-01 -9.33618963e-01
-1.29380202e+00 1.15679646e+00 8.05528939e-01 -1.86501205e-01
7.41709590e-01 2.04884112e-01 8.85385275e-01 3.08581471e-01
1.96132466e-01 -6.76573634e-01 6.54758871e-01 3.79177421e-01
1.49292731e+00 -1.49171305e+00 -3.35295677e-01 1.65293470e-01
-1.13445282e+00 1.02925825e+00 6.62500679e-01 -3.56530212e-02
2.24653721e-01 -3.63027081e-02 2.58181304e-01 -1.68303594e-01
-9.09487426e-01 -2.41667569e-01 4.14502174e-01 9.64431763e-01
1.32425681e-01 3.26520391e-02 5.83851576e-01 1.00052007e-01
-6.06791019e-01 -1.79419562e-01 5.92642725e-01 2.17585385e-01
2.85769671e-01 -8.72995496e-01 -4.40042019e-01 3.26033473e-01
-1.47045210e-01 -4.18332696e-01 -3.33629012e-01 8.43385816e-01
2.63427719e-02 1.00198555e+00 2.35520259e-01 -3.55366975e-01
3.95868830e-02 -3.96444052e-02 8.14686894e-01 -2.97735751e-01
-5.49375296e-01 8.80179331e-02 -2.20960662e-01 -7.54963160e-01
-3.12928259e-01 -3.73693585e-01 -1.05524170e+00 -1.24993049e-01
-1.74967095e-01 -3.23681712e-01 8.58803630e-01 8.29133689e-01
2.87908286e-01 7.51265466e-01 2.51600534e-01 -1.22090781e+00
-4.47835594e-01 -9.78241384e-01 -3.07537377e-01 7.22760618e-01
6.06941104e-01 -4.97978270e-01 -2.60708958e-01 4.62464482e-01] | [11.03407096862793, 1.2136867046356201] |
f95bb49f-7c76-46e5-b5e9-5ac42319324b | make-an-audio-text-to-audio-generation-with | 2301.12661 | null | https://arxiv.org/abs/2301.12661v1 | https://arxiv.org/pdf/2301.12661v1.pdf | Make-An-Audio: Text-To-Audio Generation with Prompt-Enhanced Diffusion Models | Large-scale multimodal generative modeling has created milestones in text-to-image and text-to-video generation. Its application to audio still lags behind for two main reasons: the lack of large-scale datasets with high-quality text-audio pairs, and the complexity of modeling long continuous audio data. In this work, we propose Make-An-Audio with a prompt-enhanced diffusion model that addresses these gaps by 1) introducing pseudo prompt enhancement with a distill-then-reprogram approach, it alleviates data scarcity with orders of magnitude concept compositions by using language-free audios; 2) leveraging spectrogram autoencoder to predict the self-supervised audio representation instead of waveforms. Together with robust contrastive language-audio pretraining (CLAP) representations, Make-An-Audio achieves state-of-the-art results in both objective and subjective benchmark evaluation. Moreover, we present its controllability and generalization for X-to-Audio with "No Modality Left Behind", for the first time unlocking the ability to generate high-definition, high-fidelity audios given a user-defined modality input. Audio samples are available at https://Text-to-Audio.github.io | ['Zhou Zhao', 'Xiang Yin', 'Jinglin Liu', 'Zhenhui Ye', 'Mingze Li', 'Luping Liu', 'Yi Ren', 'Dongchao Yang', 'Jiawei Huang', 'Rongjie Huang'] | 2023-01-30 | null | null | null | null | ['audio-generation', 'video-generation', 'text-to-video-generation'] | ['audio', 'computer-vision', 'natural-language-processing'] | [ 3.15253764e-01 -1.87035743e-02 9.91943628e-02 -1.88315123e-01
-1.56357002e+00 -4.48249847e-01 6.02378666e-01 -1.07879117e-01
-4.45061326e-02 5.93120039e-01 4.80638146e-01 -1.10441722e-01
-1.51020795e-01 -6.42485499e-01 -7.15710580e-01 -6.81119680e-01
-2.98015419e-02 3.31728905e-01 -1.57328054e-01 -3.57124716e-01
-2.88286895e-01 -9.84653234e-02 -1.87917781e+00 6.29482865e-01
7.37531900e-01 1.09807825e+00 3.26194942e-01 1.02028835e+00
2.34265756e-02 7.90740967e-01 -4.90338326e-01 -3.22380394e-01
-2.90174708e-02 -6.91317976e-01 -2.58072913e-01 4.15038504e-02
3.77433062e-01 -4.10664946e-01 -4.44158733e-01 5.32025278e-01
8.55829298e-01 6.63020089e-02 5.58680952e-01 -1.37216520e+00
-7.72352338e-01 8.00956309e-01 -3.79644930e-01 -2.03473717e-01
6.18433595e-01 2.91530997e-01 1.26525009e+00 -1.01161945e+00
5.54102838e-01 1.12458932e+00 5.62262356e-01 7.30927348e-01
-1.40250421e+00 -8.34716916e-01 -2.12621018e-01 1.14265509e-01
-1.54210591e+00 -6.89211249e-01 9.64703500e-01 -5.23684740e-01
7.94629931e-01 3.75442386e-01 8.31782639e-01 1.64565217e+00
2.32315883e-02 6.71085298e-01 9.47386086e-01 -3.91950846e-01
1.31459430e-01 3.59331258e-02 -5.60863078e-01 3.73941064e-01
-6.30111992e-01 3.37679774e-01 -1.12285805e+00 -5.69019243e-02
5.64223647e-01 -1.55212745e-01 -3.32105041e-01 1.22066766e-01
-1.28131890e+00 8.44676197e-01 3.15504074e-02 3.94727528e-01
-3.61566126e-01 3.37426722e-01 4.04226393e-01 4.46975082e-01
3.73987645e-01 2.08384618e-01 -5.73895387e-02 -5.59721768e-01
-1.37377262e+00 2.60264188e-01 4.97525662e-01 8.03196311e-01
4.44179267e-01 7.86065876e-01 -5.16638644e-02 8.71023238e-01
1.83938220e-01 8.40644181e-01 6.60862863e-01 -8.78599405e-01
5.28084278e-01 -1.45300161e-04 -1.69719756e-01 -6.72368705e-01
-3.13235551e-01 -5.00891805e-01 -1.00411570e+00 9.42616761e-02
1.29552796e-01 -2.59198219e-01 -7.53957331e-01 1.75867891e+00
1.18992567e-01 1.84023753e-01 2.76927762e-02 8.54954660e-01
9.11151767e-01 1.11296856e+00 -1.37803480e-01 -2.73540586e-01
1.31612492e+00 -7.11861074e-01 -8.62753689e-01 2.02665523e-01
8.30084532e-02 -8.96049321e-01 1.34369731e+00 7.28442669e-01
-1.22770798e+00 -9.04904842e-01 -9.41319227e-01 -1.00083284e-01
-1.47106051e-01 2.51472145e-01 3.66904497e-01 6.17279530e-01
-1.13084137e+00 4.17225152e-01 -6.46277070e-01 -1.12918973e-01
1.28608257e-01 1.28070086e-01 -3.92857462e-01 1.46113917e-01
-1.26208889e+00 1.35260537e-01 3.27543736e-01 -2.04455644e-01
-1.34162891e+00 -1.08650839e+00 -8.35548401e-01 7.15257600e-04
3.55710745e-01 -8.09680223e-01 1.15421784e+00 -9.79709744e-01
-1.94259381e+00 3.93285543e-01 -7.55065307e-02 -5.32103181e-01
5.43273926e-01 -4.31845427e-01 -8.43572140e-01 4.93742704e-01
-2.60236654e-02 1.18299973e+00 1.35844934e+00 -1.22458410e+00
-4.72513765e-01 1.29399523e-01 -3.25182289e-01 1.16335124e-01
-7.50625372e-01 -1.02703601e-01 -4.05624300e-01 -9.96519029e-01
-1.72985733e-01 -8.99470210e-01 1.78218871e-01 -1.56322002e-01
-3.95583510e-01 1.51057810e-01 7.88398862e-01 -7.70484388e-01
1.32874012e+00 -2.40379786e+00 1.65629968e-01 -5.55428416e-02
3.05294037e-01 -1.11715131e-01 -4.52803105e-01 8.57491672e-01
-2.10386321e-01 -6.04350530e-02 -9.80109163e-03 -5.60062468e-01
1.80332124e-01 -5.15399911e-02 -7.83675969e-01 5.54845631e-02
3.63309026e-01 8.09257627e-01 -8.00433517e-01 -4.96632218e-01
2.28950053e-01 9.64995146e-01 -7.32339799e-01 3.05670142e-01
-3.05698127e-01 6.43492699e-01 -7.26110712e-02 6.61200583e-01
2.96521991e-01 -1.45928457e-01 -6.25688285e-02 -3.06192040e-01
-1.03569478e-01 2.50388831e-01 -1.17351818e+00 2.15318918e+00
-6.01134360e-01 7.36400306e-01 1.52249217e-01 -5.02140939e-01
9.82167244e-01 7.18820393e-01 7.18318760e-01 -6.38875663e-01
4.91765477e-02 3.20489973e-01 -3.67785156e-01 -3.98944587e-01
7.57445931e-01 -4.25043494e-01 -4.26568724e-02 3.23964804e-01
5.31211793e-01 -4.93235171e-01 1.97015122e-01 3.35384130e-01
8.15679252e-01 1.15172997e-01 -2.56799132e-01 1.19197614e-01
2.17715919e-01 -5.28429091e-01 1.06402300e-01 4.80166972e-01
3.15616637e-01 1.03913701e+00 1.88380077e-01 2.23037884e-01
-1.01864338e+00 -1.38361418e+00 5.98277077e-02 1.28287899e+00
-4.28542405e-01 -8.51487756e-01 -6.47238433e-01 -7.51685798e-02
-1.50487661e-01 7.46205509e-01 -5.29508591e-01 -7.94442818e-02
-1.60958424e-01 -3.84703100e-01 8.87994409e-01 3.76851112e-01
5.42070307e-02 -7.84889460e-01 -2.22233325e-01 4.47668821e-01
-4.42624241e-01 -1.06804335e+00 -5.35750508e-01 8.25890377e-02
-6.82593524e-01 -5.56705654e-01 -1.10311866e+00 -4.77532566e-01
1.75578639e-01 -3.96408252e-02 9.21034515e-01 -5.94921231e-01
-1.87709212e-01 6.66463673e-01 -3.08523685e-01 -2.99642622e-01
-6.34258866e-01 1.02987252e-02 1.78028643e-01 3.54661047e-01
-2.26349756e-01 -1.02124465e+00 -4.84387010e-01 1.13105267e-01
-1.10425019e+00 3.70243639e-01 5.47139227e-01 8.92975450e-01
6.35176897e-01 3.23456600e-02 9.54326689e-01 -2.07592130e-01
8.40498924e-01 -4.97414678e-01 -2.18284324e-01 -1.07685752e-01
-5.26022017e-01 -2.21089765e-01 6.53802514e-01 -7.99861670e-01
-9.89554107e-01 -1.01439431e-01 -4.40545976e-01 -7.80898809e-01
-2.73920111e-02 4.10186440e-01 -6.21015839e-02 4.51990753e-01
6.89915419e-01 4.72286135e-01 -6.58319145e-03 -4.41260159e-01
7.29146779e-01 7.35069454e-01 8.41932595e-01 -6.75373316e-01
8.12780499e-01 4.78934139e-01 -1.93000495e-01 -9.35482502e-01
-3.66840094e-01 -7.02918395e-02 -9.86622348e-02 -4.89696920e-01
8.62283587e-01 -1.37586820e+00 -6.68251336e-01 4.66572493e-01
-8.55383456e-01 -3.89561743e-01 -6.39398992e-01 6.19806588e-01
-6.83879018e-01 1.66654304e-01 -8.09573770e-01 -9.62742567e-01
-5.21433890e-01 -9.22792315e-01 1.32214141e+00 5.26658967e-02
-2.74369806e-01 -6.85359240e-01 1.09023243e-01 4.64879513e-01
5.41015208e-01 1.85495630e-01 4.59540993e-01 -2.77338415e-01
-5.53340495e-01 -8.71101320e-02 2.02635154e-01 5.68358839e-01
-4.50631343e-02 7.44019300e-02 -1.28942299e+00 -3.03766310e-01
-1.47120133e-01 -6.93986356e-01 7.60635138e-01 3.29447508e-01
1.06148303e+00 -2.63736665e-01 1.41952693e-01 5.81024170e-01
1.02726996e+00 3.18704061e-02 5.56192517e-01 -3.16561937e-01
6.07310414e-01 3.68379474e-01 2.80606061e-01 8.07746708e-01
3.71425718e-01 7.20298946e-01 1.56995252e-01 -1.05831899e-01
-5.63575089e-01 -7.85459101e-01 6.56626999e-01 1.45437372e+00
6.69092685e-02 -4.26397443e-01 -6.26601934e-01 6.43656909e-01
-1.43828607e+00 -9.71938133e-01 -6.54137228e-03 2.09221864e+00
1.15149045e+00 4.27800752e-02 2.52773017e-01 3.60661000e-01
5.35462022e-01 1.56114236e-01 -2.93691605e-01 7.91138113e-02
-3.70673597e-01 2.80506015e-01 -1.95482567e-01 4.85722035e-01
-8.58139098e-01 6.11321092e-01 5.59267521e+00 1.46187222e+00
-1.44869542e+00 4.53125536e-01 4.60023701e-01 -5.98787427e-01
-6.41037941e-01 -1.75642446e-01 -6.37993038e-01 4.67742741e-01
1.26306129e+00 -2.89436933e-02 6.22648954e-01 4.44812387e-01
3.55604678e-01 1.93230882e-01 -1.13682473e+00 1.21627748e+00
8.42471570e-02 -1.31446099e+00 2.08727449e-01 -5.08528687e-02
7.01460123e-01 1.98783656e-03 4.22351688e-01 2.88906962e-01
-3.12144104e-02 -1.00738227e+00 1.20005846e+00 5.22698104e-01
1.36110854e+00 -6.04710460e-01 9.61653516e-02 1.30081341e-01
-1.33500898e+00 -1.11028249e-03 1.57403901e-01 2.04196364e-01
4.77470368e-01 8.14575791e-01 -7.46115029e-01 7.42502272e-01
4.95842725e-01 5.36650896e-01 -2.11482644e-01 5.13523757e-01
-1.02618851e-01 9.98146415e-01 -4.21488374e-01 1.11722305e-01
1.18916668e-01 4.32253256e-02 7.47553051e-01 1.39959681e+00
7.90464878e-01 -1.57925248e-01 7.60169476e-02 9.62806404e-01
-1.14292935e-01 1.99547946e-01 -5.26767612e-01 -4.51521575e-01
4.06948984e-01 1.12361872e+00 -3.34113777e-01 -1.08053707e-01
-2.54194178e-02 9.18186843e-01 -2.84695119e-01 4.81714994e-01
-9.92233396e-01 -3.79919469e-01 2.50136614e-01 4.14859444e-01
2.31765538e-01 -3.29780549e-01 7.71394074e-02 -9.55687821e-01
9.33273509e-02 -1.05199158e+00 2.48046875e-01 -1.23367560e+00
-1.26576078e+00 8.51346433e-01 -4.58532386e-02 -1.53683066e+00
-6.93282127e-01 -1.75412953e-01 -2.87357002e-01 6.78761959e-01
-1.31215870e+00 -1.48644090e+00 -2.30197638e-01 8.75868797e-01
6.90391600e-01 -3.97821188e-01 9.71372724e-01 7.42840886e-01
1.35851661e-02 6.61949098e-01 1.18101493e-01 -1.56293780e-01
9.00174499e-01 -9.03633535e-01 1.23083144e-01 4.99311209e-01
4.15906280e-01 3.47155660e-01 7.51842082e-01 -3.30461353e-01
-1.59872603e+00 -8.67643774e-01 6.00458443e-01 -2.46886849e-01
8.94700468e-01 -7.56988883e-01 -6.94949806e-01 3.01656932e-01
4.90057290e-01 -2.68667877e-01 8.59278262e-01 1.73381492e-02
-4.81516778e-01 -4.61227983e-01 -6.77461147e-01 4.93904501e-01
7.54554391e-01 -9.79710758e-01 -1.80992395e-01 1.49712235e-01
9.75990772e-01 -4.62512165e-01 -9.77524817e-01 1.65556133e-01
5.56309402e-01 -9.62832272e-01 9.04607296e-01 -1.16844483e-01
8.15584660e-01 -2.96073616e-01 -4.20444876e-01 -1.23775244e+00
8.40212405e-02 -1.33222544e+00 -1.87887564e-01 1.80653965e+00
3.75519037e-01 -4.31027830e-01 4.20237988e-01 -1.03764288e-01
-2.01602310e-01 -6.07903540e-01 -1.16612411e+00 -7.72934675e-01
-9.22313631e-02 -9.40916359e-01 4.81631905e-01 7.63971984e-01
7.39680603e-02 5.10213196e-01 -9.27596867e-01 -5.83502948e-02
4.13039953e-01 8.61862116e-03 8.11470211e-01 -8.57829750e-01
-8.06700945e-01 -3.47424597e-01 -1.47129267e-01 -1.20176888e+00
-6.92757517e-02 -8.41428399e-01 -3.77323553e-02 -1.08727384e+00
-8.82377103e-02 -2.70930201e-01 -8.00843760e-02 4.61521924e-01
2.01180920e-01 3.64926994e-01 5.07759631e-01 -3.77480425e-02
-4.80960399e-01 1.06169116e+00 1.02486634e+00 -2.51311958e-01
-2.24674970e-01 -1.73151210e-01 -5.62167645e-01 2.76556224e-01
6.43529296e-01 -4.82045382e-01 -7.11064637e-01 -2.12523758e-01
3.18410039e-01 4.86744523e-01 4.97980356e-01 -1.39272809e+00
2.43030861e-02 1.40520826e-01 1.19040452e-01 -4.79001313e-01
9.53173518e-01 -5.75540364e-01 5.89628935e-01 7.81524554e-02
-4.80063319e-01 -9.08123031e-02 3.99336994e-01 6.32898271e-01
-5.39938509e-01 2.54782606e-02 5.01670480e-01 3.15024793e-01
-2.71677315e-01 5.89728802e-02 -4.29762304e-01 1.25643373e-01
5.15739322e-01 -1.81871206e-02 -2.49308720e-01 -1.09806180e+00
-6.69978082e-01 -1.88710526e-01 6.56285658e-02 6.36171341e-01
7.33893931e-01 -1.50089920e+00 -1.05809081e+00 3.41450691e-01
1.12434447e-01 -3.44476312e-01 8.12203825e-01 8.22243989e-01
-4.42623980e-02 3.43786150e-01 -1.33999866e-02 -7.68063307e-01
-1.18974602e+00 3.17894459e-01 -5.43964319e-02 -1.63501725e-01
-5.37749887e-01 7.05509365e-01 3.06626439e-01 -2.19081402e-01
2.63948113e-01 -1.51772678e-01 2.42673576e-01 1.76456571e-01
4.66778129e-01 1.85963303e-01 1.67579949e-02 -4.94445264e-01
1.84577657e-03 3.29658747e-01 3.01184773e-01 -8.10921967e-01
1.35016274e+00 -5.85589595e-02 3.49519253e-01 7.95528173e-01
1.17843902e+00 4.52916712e-01 -1.36661911e+00 -1.11726243e-02
-7.20611274e-01 -2.52226770e-01 1.77196294e-01 -9.06429946e-01
-1.03654110e+00 1.23464036e+00 7.47262776e-01 1.82177737e-01
1.40044677e+00 -3.29449512e-02 1.06376457e+00 1.40243962e-01
1.91360369e-01 -1.08276856e+00 5.74208379e-01 2.72935957e-01
1.20298755e+00 -7.62275040e-01 -2.26916447e-01 -6.91724643e-02
-9.11551297e-01 1.01068747e+00 1.30093783e-01 4.38117057e-01
4.93780643e-01 5.82300782e-01 1.89202219e-01 -1.07713102e-04
-1.05963349e+00 -1.13769621e-01 4.23492312e-01 7.19013631e-01
4.41230774e-01 7.37840086e-02 1.55906424e-01 8.43823731e-01
-5.18795848e-01 1.57212779e-01 3.05650175e-01 5.86166263e-01
-1.60890758e-01 -1.09390521e+00 -5.76288164e-01 1.66778922e-01
-3.52536619e-01 -3.66227627e-01 -8.82567018e-02 5.28869867e-01
-2.75190864e-02 1.11161256e+00 -4.40024473e-02 -6.54887080e-01
9.47138444e-02 2.10885510e-01 3.82839561e-01 -1.14305995e-01
-5.74389815e-01 7.88586140e-01 2.55256481e-02 -3.09734643e-01
-1.46967456e-01 -5.69633901e-01 -1.20989788e+00 -2.82291085e-01
-2.47575611e-01 1.57269061e-01 8.30000460e-01 4.17591155e-01
7.75643468e-01 7.93808818e-01 6.87068343e-01 -1.06744766e+00
-5.18157005e-01 -1.04606044e+00 -7.27503955e-01 3.60894769e-01
3.80031735e-01 -2.78155267e-01 -2.32152849e-01 4.34125930e-01] | [15.478331565856934, 5.532462120056152] |
878c9b8e-04db-4c1a-a847-821b515681e1 | medfmc-a-real-world-dataset-and-benchmark-for | 2306.09579 | null | https://arxiv.org/abs/2306.09579v1 | https://arxiv.org/pdf/2306.09579v1.pdf | MedFMC: A Real-world Dataset and Benchmark For Foundation Model Adaptation in Medical Image Classification | Foundation models, often pre-trained with large-scale data, have achieved paramount success in jump-starting various vision and language applications. Recent advances further enable adapting foundation models in downstream tasks efficiently using only a few training samples, e.g., in-context learning. Yet, the application of such learning paradigms in medical image analysis remains scarce due to the shortage of publicly accessible data and benchmarks. In this paper, we aim at approaches adapting the foundation models for medical image classification and present a novel dataset and benchmark for the evaluation, i.e., examining the overall performance of accommodating the large-scale foundation models downstream on a set of diverse real-world clinical tasks. We collect five sets of medical imaging data from multiple institutes targeting a variety of real-world clinical tasks (22,349 images in total), i.e., thoracic diseases screening in X-rays, pathological lesion tissue screening, lesion detection in endoscopy images, neonatal jaundice evaluation, and diabetic retinopathy grading. Results of multiple baseline methods are demonstrated using the proposed dataset from both accuracy and cost-effective perspectives. | ['Shaoting Zhang', 'Yu Qiao', 'Kang Li', 'Jie Zhao', 'Qi Duan', 'Tian Shen', 'Junjun He', 'Jun Shen', 'Xiangyu Gao', 'Xiaoqiang Liu', 'Qian Da', 'Mengzhang Li', 'Lilong Wang', 'Xiaosong Wang', 'Dequan Wang'] | 2023-06-16 | null | null | null | null | ['medical-image-classification', 'diabetic-retinopathy-grading'] | ['medical', 'medical'] | [ 5.29801965e-01 -5.23650311e-02 -3.47961396e-01 -5.86256981e-01
-1.15718269e+00 -3.76759559e-01 2.75178432e-01 4.49190617e-01
-5.75881541e-01 4.42617387e-01 1.51693150e-01 -5.69172323e-01
-2.11157009e-01 -3.77867788e-01 -7.60704398e-01 -6.56400144e-01
-1.76711962e-01 3.01615983e-01 1.76512256e-01 2.12122172e-01
3.08817532e-02 3.05019319e-01 -1.31974483e+00 1.08874333e+00
9.44963455e-01 9.58347023e-01 4.24727708e-01 8.99660945e-01
3.25354666e-01 8.79841983e-01 -3.63417953e-01 -4.63077247e-01
6.75173942e-03 -4.83302981e-01 -7.89651096e-01 2.16698982e-02
6.59911454e-01 -5.04258394e-01 -8.47157761e-02 9.08378124e-01
9.85330999e-01 -2.15312824e-01 4.64993626e-01 -6.39405429e-01
-7.84092486e-01 5.94481111e-01 -3.62805843e-01 5.76268613e-01
1.42705724e-01 2.29561225e-01 7.82683730e-01 -7.47457147e-01
6.80222213e-01 9.02453482e-01 8.96254539e-01 8.63932312e-01
-1.04306293e+00 -3.53932440e-01 -1.88328162e-01 2.90190369e-01
-8.82163644e-01 -2.69939512e-01 2.29302511e-01 -6.29559934e-01
8.50668073e-01 5.54196000e-01 4.42764223e-01 1.26374912e+00
2.18795940e-01 8.13115537e-01 1.26303375e+00 -4.80969101e-01
9.95584484e-03 2.15321183e-01 8.85213986e-02 9.20484841e-01
1.92920640e-01 1.11164257e-01 -4.31432247e-01 -2.59729534e-01
5.01154900e-01 1.71543196e-01 -4.24322128e-01 -8.70793909e-02
-1.43619168e+00 6.85860336e-01 3.36295426e-01 6.90818578e-02
-2.83350170e-01 -2.61869371e-01 5.42380810e-01 3.61209720e-01
3.64681154e-01 2.66453713e-01 -8.07129860e-01 2.26585418e-01
-7.25427508e-01 -1.08426876e-01 7.28740573e-01 9.95176733e-01
2.80338824e-01 -4.91108984e-01 -5.46880841e-01 8.87192428e-01
2.12745175e-01 1.95446089e-01 9.12640035e-01 -7.14016914e-01
4.09346074e-01 5.84298134e-01 -2.01359004e-01 -3.51380765e-01
-6.07630670e-01 -5.39468169e-01 -9.71479297e-01 3.05872201e-03
4.37344491e-01 -2.78903008e-01 -1.31783104e+00 1.55978525e+00
5.12759149e-01 3.82382333e-01 7.27949571e-03 7.11845279e-01
1.54888511e+00 2.37721652e-01 3.40970188e-01 -1.93519592e-01
1.52779830e+00 -1.33112812e+00 -3.19646120e-01 2.00871695e-02
1.00983548e+00 -8.38832438e-01 1.21233201e+00 5.41197360e-01
-1.07188141e+00 -4.83880520e-01 -6.24120176e-01 -6.98885843e-02
-2.76105672e-01 6.39928341e-01 6.47319913e-01 4.76738513e-01
-1.03624022e+00 4.25201148e-01 -9.13705289e-01 -4.60897148e-01
5.70046484e-01 5.02896607e-01 -3.88968527e-01 -2.71605641e-01
-7.52911985e-01 8.80993307e-01 7.91345686e-02 8.35739896e-02
-1.15157354e+00 -1.10760105e+00 -5.68599284e-01 -1.97074279e-01
2.51890033e-01 -1.00213969e+00 1.28633618e+00 -8.05812776e-01
-1.20064330e+00 1.48718667e+00 2.50101089e-01 -3.22468102e-01
6.86874688e-01 -2.96911001e-01 -3.42234552e-01 2.55855411e-01
6.62590042e-02 6.83055878e-01 4.39623624e-01 -7.96591401e-01
-9.18676436e-01 -1.97426155e-01 1.56985670e-01 1.75726309e-03
-4.64412242e-01 2.73097843e-01 -4.91717368e-01 -5.70222020e-01
-2.30520368e-01 -1.02465570e+00 -6.33395672e-01 1.83746129e-01
-3.94477099e-01 -5.20103946e-02 4.29419935e-01 -5.78934550e-01
1.14040411e+00 -1.97912049e+00 -1.37905432e-02 -2.03704149e-01
3.14486235e-01 5.11062443e-01 -2.58646041e-01 7.94245005e-02
-2.55966693e-01 2.17588663e-01 1.08808337e-03 -2.08674058e-01
-5.76437294e-01 3.74589488e-02 1.03276730e-01 3.88372719e-01
1.66902289e-01 9.42534089e-01 -9.91702735e-01 -7.37533391e-01
3.17552388e-01 3.06700468e-01 -8.34026873e-01 3.50272864e-01
-2.96135843e-02 7.99803078e-01 -7.03729749e-01 1.01256967e+00
1.66262731e-01 -7.92999566e-01 3.86117510e-02 -4.82930750e-01
-9.91009474e-02 6.91679344e-02 -4.65490609e-01 1.94369912e+00
-4.74924356e-01 2.85638928e-01 4.79238965e-02 -1.21701717e+00
2.08498210e-01 5.35451293e-01 7.19486833e-01 -4.47118133e-01
7.12327957e-02 4.29327071e-01 3.60996217e-01 -1.09893346e+00
-2.02839032e-01 1.19643204e-01 3.89457315e-01 3.51425484e-02
1.55495778e-01 2.23001868e-01 4.23773497e-01 -2.46394292e-01
1.27301383e+00 -1.95826292e-02 3.76414627e-01 -2.25788727e-01
6.47599876e-01 3.49525362e-01 3.46832216e-01 9.07081664e-01
-4.64956492e-01 7.16145158e-01 2.41612121e-01 -7.58532524e-01
-7.03835368e-01 -8.85405123e-01 -5.13171136e-01 1.27406883e+00
-2.26129696e-01 -2.16100171e-01 -5.97965360e-01 -7.41749644e-01
1.82178263e-02 4.84258458e-02 -9.29724693e-01 -1.67916372e-01
-6.76983893e-01 -1.42557740e+00 4.14848506e-01 5.36365092e-01
1.00127049e-01 -1.08526826e+00 -8.27431679e-01 1.82951167e-01
1.73201505e-02 -1.12060034e+00 -2.98608810e-01 1.51464671e-01
-1.29791713e+00 -1.33488452e+00 -9.93919253e-01 -1.09482133e+00
8.39780927e-01 1.04326025e-01 1.26212251e+00 2.79015929e-01
-1.14809990e+00 3.89486700e-01 -3.99741203e-01 -5.76548040e-01
-6.13461494e-01 4.39240560e-02 -3.50069016e-01 -2.02748433e-01
2.78166592e-01 -2.06390828e-01 -1.04875898e+00 2.32704520e-01
-7.42028534e-01 2.31977358e-01 1.00489819e+00 1.18893683e+00
9.21340525e-01 -7.47232437e-01 6.76167190e-01 -1.32100785e+00
4.14020002e-01 -5.84072411e-01 -4.28622544e-01 6.32244110e-01
-6.92696452e-01 -5.40222824e-01 4.00817245e-01 -5.24052858e-01
-8.81291211e-01 -1.83952436e-01 -2.05384836e-01 -3.12376469e-01
-2.27329522e-01 8.81633759e-01 5.82250953e-01 -9.88539457e-02
9.25382197e-01 -3.48158479e-02 -1.42999962e-01 -6.18658543e-01
2.18321353e-01 6.21041954e-01 5.79724848e-01 -5.17260373e-01
1.94450125e-01 4.53334957e-01 1.09435730e-01 -5.24362206e-01
-1.31269395e+00 -6.37938499e-01 -4.89652574e-01 9.06206518e-02
1.00019574e+00 -1.05253828e+00 -2.93880165e-01 2.87629217e-01
-9.03792679e-01 -4.09294695e-01 -3.05890143e-01 8.73457551e-01
-4.50868875e-01 1.63827524e-01 -9.12674785e-01 -1.75276250e-01
-6.85510695e-01 -1.45178890e+00 9.80804324e-01 1.96472570e-01
1.38922602e-01 -9.59892929e-01 4.89877984e-02 4.42358315e-01
3.28287065e-01 5.15304744e-01 1.32693946e+00 -5.68468034e-01
-4.67916012e-01 -2.21764639e-01 -2.69215614e-01 5.11842370e-01
4.41254616e-01 2.85060611e-02 -1.01310503e+00 -3.84505481e-01
-7.14942068e-02 -5.27186215e-01 9.47869837e-01 6.95447505e-01
1.70484924e+00 1.86491713e-01 -3.86039168e-01 1.00441873e+00
1.24966085e+00 2.57922471e-01 2.85933495e-01 2.97188610e-01
5.54163575e-01 5.74713886e-01 5.88993907e-01 3.46844941e-01
1.85882732e-01 1.95443228e-01 3.98753881e-01 -5.59355438e-01
-5.77161789e-01 1.53627291e-01 -1.12370394e-01 7.04263270e-01
-2.97808796e-01 -9.71240252e-02 -1.25109184e+00 7.18750000e-01
-1.68376732e+00 -4.98462409e-01 -8.31718817e-02 1.95627022e+00
1.13941526e+00 -5.13894185e-02 -1.37678787e-01 -4.70279545e-01
7.16664016e-01 -4.43189234e-01 -6.76598370e-01 -3.71689573e-02
1.19214974e-01 9.86171290e-02 2.10000485e-01 -1.20677404e-01
-1.29006302e+00 4.66170013e-01 6.68014193e+00 5.27105510e-01
-1.36610222e+00 3.22427899e-01 1.15846992e+00 -8.67623016e-02
8.49538669e-02 -4.83562946e-01 -6.13401592e-01 3.31190109e-01
9.39363956e-01 9.41885561e-02 1.01758866e-02 9.72781479e-01
1.87676445e-01 1.01236224e-01 -1.65585959e+00 8.39096606e-01
-1.05938114e-01 -1.66061974e+00 2.25953177e-01 -3.16334188e-01
1.02470803e+00 6.85152113e-01 4.00065839e-01 2.08751589e-01
-2.06406247e-02 -1.04925156e+00 2.52227366e-01 3.24509680e-01
1.36398125e+00 6.82948902e-02 8.17471921e-01 3.07386905e-01
-5.97763360e-01 -1.15650482e-01 -1.87665790e-01 5.27367711e-01
-1.83817521e-01 4.13982868e-01 -8.41832638e-01 4.23439354e-01
1.03080928e+00 5.84623039e-01 -5.63227654e-01 1.35294259e+00
1.29676059e-01 8.20380211e-01 -1.25661567e-01 2.52500534e-01
2.84982651e-01 -4.16191295e-02 1.94604620e-01 1.61018813e+00
3.23945105e-01 1.19304143e-01 1.94425449e-01 4.90651220e-01
-1.45657882e-01 3.24394166e-01 -3.67826521e-01 5.48017696e-02
1.90239437e-02 1.59796309e+00 -5.68114698e-01 -4.83132839e-01
-9.67583239e-01 3.24105054e-01 9.77168903e-02 3.24347764e-01
-8.57178867e-01 5.77989295e-02 2.15143576e-01 3.77747156e-02
1.73104312e-02 4.08555210e-01 -9.37871113e-02 -1.18862414e+00
-2.21072599e-01 -1.10117316e+00 8.19947243e-01 -5.24856925e-01
-1.41469073e+00 5.68795204e-01 6.16217218e-02 -1.27660275e+00
-1.14720024e-01 -9.69815314e-01 -6.45645022e-01 7.53434002e-01
-2.09370637e+00 -1.27197862e+00 -5.97635210e-01 7.41530001e-01
6.31973922e-01 -3.18201870e-01 1.01392770e+00 6.94764793e-01
-6.48778737e-01 7.03680873e-01 2.19023749e-01 1.09968849e-01
1.07248771e+00 -1.17768645e+00 -1.07047223e-01 4.16360229e-01
-1.06173195e-01 7.49956965e-01 2.93812931e-01 -3.53324980e-01
-1.40771842e+00 -1.32118690e+00 4.68069345e-01 -5.48319221e-01
5.71307182e-01 1.02624103e-01 -6.47017956e-01 7.95192182e-01
-5.01706935e-02 7.34344959e-01 1.15620172e+00 -6.22257963e-02
-6.38365224e-02 -1.71371534e-01 -1.24387765e+00 5.52809000e-01
1.02793264e+00 -2.59881884e-01 -3.99886608e-01 7.86485672e-01
5.05271375e-01 -8.42661440e-01 -1.09233010e+00 9.25566494e-01
5.85390210e-01 -8.13377261e-01 1.16008055e+00 -1.18847299e+00
8.60646248e-01 1.90505728e-01 -9.96816680e-02 -8.59437764e-01
-1.56670302e-01 -6.65265262e-01 -1.41093686e-01 7.92634547e-01
5.80889046e-01 -5.70029855e-01 8.88221323e-01 3.09159786e-01
-5.44009864e-01 -1.26496291e+00 -7.48762488e-01 -2.71566331e-01
6.15199804e-02 -1.06747113e-01 1.67513981e-01 8.80161464e-01
-3.67356151e-01 1.01162933e-01 3.67264939e-03 1.29268512e-01
5.44647872e-01 3.54260504e-01 4.70354885e-01 -9.16350186e-01
-5.33486664e-01 -2.83683896e-01 -1.21067636e-01 -7.59040534e-01
-3.96971643e-01 -1.07336771e+00 -1.27821416e-01 -1.54623008e+00
5.49897194e-01 -5.77563226e-01 -8.55107129e-01 4.86196488e-01
-4.71266687e-01 2.58883893e-01 -2.25567922e-01 3.51936013e-01
-6.76278889e-01 -1.31145865e-01 1.51583672e+00 -1.76443368e-01
5.66148162e-02 2.85132140e-01 -7.55899072e-01 1.06379545e+00
5.32127261e-01 -4.25639898e-01 -4.88633633e-01 -8.16441238e-01
-6.73955455e-02 1.05847612e-01 3.42770666e-01 -8.28479886e-01
1.28911451e-01 -1.45839289e-01 3.97033155e-01 -4.05693740e-01
-1.51286945e-01 -5.59184492e-01 -2.57121146e-01 8.42938304e-01
-5.06821275e-01 1.11154784e-02 2.19431847e-01 5.48691571e-01
-2.89362937e-01 -2.91066438e-01 8.88250768e-01 -5.02517402e-01
-7.99429953e-01 5.57929933e-01 1.61905915e-01 2.61468977e-01
1.04280412e+00 -1.54975289e-02 -5.12222290e-01 2.21852168e-01
-1.03718412e+00 1.88103542e-01 -5.34882359e-02 2.64246374e-01
5.84045708e-01 -8.62868011e-01 -1.08884180e+00 1.36959761e-01
3.20014954e-01 1.85847372e-01 3.60743135e-01 1.44618142e+00
-7.03967094e-01 4.58119601e-01 -1.15505815e-01 -1.03706992e+00
-1.50241792e+00 4.07827675e-01 2.65673071e-01 -6.11411631e-01
-7.57697940e-01 1.17149162e+00 5.83046794e-01 -4.25674051e-01
2.50703037e-01 -8.39211583e-01 -2.38832504e-01 -9.19477269e-02
5.42601049e-01 1.34113818e-01 3.76881450e-01 -8.01570043e-02
-3.72272283e-01 6.32417977e-01 -3.89190644e-01 6.14887774e-01
1.52682233e+00 6.49711341e-02 -7.08836168e-02 1.95568651e-01
1.11679459e+00 -3.47866684e-01 -8.91514599e-01 -3.00684780e-01
1.17315851e-01 -1.81728855e-01 8.25241581e-02 -1.18734813e+00
-1.02093589e+00 9.94497180e-01 9.93175805e-01 -6.43738061e-02
1.36947906e+00 4.54570465e-02 5.34402549e-01 5.95654905e-01
2.68814713e-01 -8.42060864e-01 9.57707688e-02 8.52984861e-02
7.66834915e-01 -1.75240290e+00 1.78701743e-01 -4.06847060e-01
-4.92631257e-01 1.13065898e+00 5.56860507e-01 1.65625170e-01
6.68360054e-01 2.11265802e-01 3.69859278e-01 -2.28358507e-01
-1.03453434e+00 6.76910281e-02 4.82795805e-01 5.06254494e-01
9.00255203e-01 -1.27323428e-02 -3.25107843e-01 6.68724418e-01
3.38042229e-01 3.67790312e-01 3.82366210e-01 9.97372150e-01
-1.96844906e-01 -1.08700478e+00 1.50077548e-02 9.03100014e-01
-1.19121718e+00 -3.60329419e-01 9.42890197e-02 8.10397267e-01
2.74752766e-01 5.92540860e-01 -4.83620226e-01 2.35992372e-02
4.13883954e-01 -2.55495012e-01 5.33860862e-01 -1.06955040e+00
-7.68994033e-01 1.06978469e-01 1.03336282e-01 -5.82964063e-01
-6.90538824e-01 -6.18501782e-01 -1.02138472e+00 3.94703180e-01
-2.43498042e-01 -1.97669655e-01 6.62800789e-01 6.85316622e-01
2.74242640e-01 7.39240468e-01 3.21383089e-01 -3.87983024e-01
-1.01931274e+00 -8.84577155e-01 -1.27194509e-01 5.22516310e-01
4.82345641e-01 -2.56594688e-01 -1.86910942e-01 4.68402535e-01] | [14.851948738098145, -2.4271295070648193] |
4e808bd1-20e0-4225-8673-bb8f07a6d669 | quadtree-driven-lossy-event-compression | 2005.00974 | null | https://arxiv.org/abs/2005.00974v2 | https://arxiv.org/pdf/2005.00974v2.pdf | Lossy Event Compression based on Image-derived Quad Trees and Poisson Disk Sampling | With several advantages over conventional RGB cameras, event cameras have provided new opportunities for tackling visual tasks under challenging scenarios with fast motion, high dynamic range, and/or power constraint. Yet unlike image/video compression, the performance of event compression algorithm is far from satisfying and practical. The main challenge for compressing events is the unique event data form, i.e., a stream of asynchronously fired event tuples each encoding the 2D spatial location, timestamp, and polarity (denoting an increase or decrease in brightness). Since events only encode temporal variations, they lack spatial structure which is crucial for compression. To address this problem, we propose a novel event compression algorithm based on a quad tree (QT) segmentation map derived from the adjacent intensity images. The QT informs 2D spatial priority within the 3D space-time volume. In the event encoding step, events are first aggregated over time to form polarity-based event histograms. The histograms are then variably sampled via Poisson Disk Sampling prioritized by the QT based segmentation map. Next, differential encoding and run length encoding are employed for encoding the spatial and polarity information of the sampled events, respectively, followed by Huffman encoding to produce the final encoded events. Our Poisson Disk Sampling based Lossy Event Compression (PDS-LEC) algorithm performs rate-distortion based optimal allocation. On average, our algorithm achieves greater than 6x compression compared to the state of the art. | ['Zihao W. Wang', 'Srutarshi Banerjee', 'Aggelos Katsaggelos', 'Henry H. Chopp', 'Oliver Cossairt'] | 2020-05-03 | null | null | null | null | ['event-based-vision'] | ['computer-vision'] | [ 8.12292457e-01 -5.64065516e-01 -1.65130869e-01 -2.80241072e-01
-6.88847959e-01 -4.08740669e-01 3.70171994e-01 5.16815543e-01
-6.39371574e-01 7.61584342e-01 1.48082033e-01 1.46325082e-01
-2.25836873e-01 -1.03671575e+00 -6.48345709e-01 -1.00534582e+00
-3.67289513e-01 1.64584145e-01 6.69510543e-01 5.53514123e-01
3.09604466e-01 7.36885428e-01 -1.93082952e+00 3.40984941e-01
5.03085852e-01 1.61859465e+00 3.51478189e-01 8.35613728e-01
-3.47023278e-01 9.85795319e-01 -7.80243397e-01 3.45587060e-02
1.10211030e-01 -5.53776264e-01 -4.25751090e-01 1.38097942e-01
-5.40886782e-02 -7.11678147e-01 -7.19759524e-01 6.88193917e-01
5.83471239e-01 3.44425231e-01 4.34490770e-01 -1.33753836e+00
1.62184592e-02 2.36822292e-01 -6.87835038e-01 6.09652162e-01
6.28109813e-01 6.71020672e-02 6.04681790e-01 -6.42634451e-01
8.55202615e-01 8.45123827e-01 1.56022012e-01 8.87676403e-02
-8.56196642e-01 -4.26692307e-01 -2.81407088e-01 5.73651254e-01
-1.60058701e+00 -2.40009040e-01 7.17651129e-01 1.24410436e-01
1.20582056e+00 5.21241069e-01 1.08250415e+00 5.16262054e-01
4.42476273e-01 7.13065207e-01 7.02552199e-01 -1.08481944e-01
7.02728510e-01 -5.57652891e-01 -4.85792220e-01 1.23909377e-01
-4.22561206e-02 -1.20206080e-01 -1.06834030e+00 -1.31034896e-01
7.99385369e-01 3.17928016e-01 -4.93589520e-01 -6.58193678e-02
-1.29152048e+00 4.13732380e-01 2.05090985e-01 -1.87628195e-01
-8.32305908e-01 4.14033890e-01 5.44667065e-01 7.26102367e-02
2.15953052e-01 -4.10863131e-01 -8.85589570e-02 -6.49706006e-01
-1.32125092e+00 4.33943897e-01 5.45218766e-01 9.61740434e-01
5.88620126e-01 -3.05713952e-01 -4.69625711e-01 4.94959563e-01
7.72277787e-02 5.99113345e-01 3.10869724e-01 -1.13012910e+00
4.60008174e-01 4.26328629e-01 -3.63065973e-02 -1.04702699e+00
-6.29310608e-02 2.87517577e-01 -9.93178487e-01 1.21627875e-01
1.33985311e-01 3.06000113e-01 -7.27605879e-01 1.29304874e+00
4.37375933e-01 4.01057214e-01 -1.58661511e-02 8.29101920e-01
5.61441779e-01 1.31437147e+00 -4.54484206e-03 -9.38228846e-01
1.45489621e+00 -1.61007479e-01 -9.90159869e-01 2.87769288e-01
-7.98570439e-02 -7.06912994e-01 7.37579465e-01 4.83813465e-01
-1.57716000e+00 -2.59441972e-01 -1.02727640e+00 -2.25319013e-01
-1.15048982e-01 -5.19661665e-01 1.89902291e-01 2.25186944e-01
-7.63395548e-01 3.26616585e-01 -8.90509367e-01 -1.14723362e-01
5.57892382e-01 2.84532100e-01 -4.58286218e-02 -2.99896508e-01
-9.46795642e-01 1.51179522e-01 4.75128502e-01 -4.24159497e-01
-7.71459401e-01 -6.20417535e-01 -7.21064031e-01 1.11626081e-01
1.37556359e-01 -2.72939116e-01 1.06485724e+00 -3.39304864e-01
-1.08768106e+00 6.85416579e-01 -5.72192907e-01 -7.40429461e-01
2.24361002e-01 2.93360651e-01 -1.91089973e-01 8.32595408e-01
-1.32683262e-01 7.46082127e-01 7.85995901e-01 -8.03431988e-01
-9.62721229e-01 -2.73323327e-01 -1.78452358e-01 4.98245448e-01
-2.80427575e-01 3.27634439e-02 -7.92512238e-01 -6.34167790e-01
2.11566180e-01 -4.66457546e-01 5.31999692e-02 5.13586223e-01
-7.79663846e-02 -1.27865866e-01 1.20652699e+00 -3.25310200e-01
1.68625152e+00 -2.69536257e+00 3.41778025e-02 1.16712831e-01
1.50213823e-01 -9.78683084e-02 3.54152173e-01 3.09789509e-01
2.42147639e-01 -2.37494260e-01 -4.87608820e-01 -2.70324647e-01
-2.38244891e-01 4.84956831e-01 -6.10030711e-01 4.66277540e-01
4.69402131e-03 5.43959200e-01 -1.04039013e+00 -9.87637103e-01
6.54461384e-01 7.19853580e-01 -2.67922729e-01 4.30614710e-01
-2.36963555e-01 1.39014363e-01 -2.44689837e-01 7.31185675e-01
9.08696055e-01 -1.95131943e-01 -1.80475816e-01 -4.53842163e-01
-3.05623353e-01 3.06026220e-01 -1.35513580e+00 1.63135254e+00
-5.15934117e-02 7.41876900e-01 -2.28912070e-01 -5.85752547e-01
8.35604250e-01 4.58011329e-01 1.03732359e+00 -1.04803753e+00
1.74543727e-02 -3.77679057e-02 -7.72616446e-01 -4.09511596e-01
9.45498526e-01 1.02550142e-01 -1.40487909e-01 4.96749759e-01
-2.95283705e-01 -3.46261322e-01 3.48517567e-01 2.74352461e-01
1.30247223e+00 -1.13837488e-01 3.78996551e-01 4.45999980e-01
1.67617843e-01 -7.05921266e-04 5.53910851e-01 5.58774352e-01
-3.65925759e-01 9.42746937e-01 4.19557840e-01 -4.25189078e-01
-1.19139254e+00 -1.40033424e+00 -4.24589813e-01 5.15863061e-01
6.06471002e-01 -5.78701496e-01 -5.35212219e-01 -9.14233178e-02
-2.10383505e-01 6.25988901e-01 -1.84241623e-01 -2.44408380e-02
-5.96109629e-01 -7.55855083e-01 4.17862087e-01 3.47895682e-01
7.56184518e-01 -1.12713993e+00 -1.62042499e+00 5.63648462e-01
-3.28216970e-01 -1.21439230e+00 -4.96770710e-01 3.39274496e-01
-8.61669302e-01 -7.49288082e-01 -4.94809389e-01 -3.62327397e-01
3.07958931e-01 4.02748048e-01 1.00177133e+00 -3.56435776e-01
-6.37812495e-01 3.59522164e-01 -5.50741673e-01 -3.44287992e-01
1.17236279e-01 -5.65144181e-01 -3.56583744e-01 6.34092242e-02
3.29076707e-01 -7.53855109e-01 -1.15041447e+00 7.44145364e-02
-1.62100279e+00 2.23756544e-02 1.76612467e-01 3.81134540e-01
1.38186097e+00 6.95077837e-01 7.53436387e-02 -1.82990015e-01
2.78819531e-01 -4.95443344e-01 -5.03673375e-01 -3.02872490e-02
-1.13322631e-01 -2.82168269e-01 5.41097283e-01 -3.39692891e-01
-8.99807632e-01 1.56585425e-01 1.33819550e-01 -4.50269401e-01
-7.78807998e-02 1.48206934e-01 2.49797069e-02 2.05755353e-01
2.95374125e-01 6.19893134e-01 -1.55109182e-01 1.66574687e-01
1.76567599e-01 7.16142952e-01 8.89658511e-01 -1.96068078e-01
1.61772341e-01 1.01787019e+00 1.33256763e-01 -8.41565132e-01
-1.06713012e-01 -4.62995917e-01 -1.95674717e-01 -4.99884993e-01
1.03948879e+00 -7.75912285e-01 -5.98141789e-01 5.99507213e-01
-1.22862840e+00 -1.63506567e-01 -8.12031269e-01 4.12537575e-01
-7.48876035e-01 3.28989416e-01 -7.58760989e-01 -1.01957726e+00
-2.46531337e-01 -1.05902064e+00 1.28901386e+00 4.17125821e-01
-5.42152971e-02 -4.59220886e-01 -1.24554209e-01 -2.25984961e-01
3.55906576e-01 5.56891918e-01 6.68001175e-01 1.21794209e-01
-1.10285568e+00 -2.17903897e-01 -3.62189770e-01 -1.44300148e-01
5.64589389e-02 4.45892513e-02 -6.81560040e-01 -3.65870111e-02
1.11241728e-01 -1.71223730e-01 5.99314272e-01 6.93306983e-01
1.74681127e+00 -1.47625938e-01 -3.33901554e-01 8.40837777e-01
1.59245265e+00 6.53728485e-01 1.13288486e+00 1.31117776e-01
3.31095248e-01 1.08927302e-01 6.95997596e-01 1.34081018e+00
4.49624330e-01 6.29902959e-01 6.43225491e-01 1.66100889e-01
-2.18410000e-01 -9.37382430e-02 2.75704622e-01 4.85183179e-01
1.31048247e-01 -8.64076495e-01 -4.19790715e-01 5.51304162e-01
-1.54905570e+00 -1.24715221e+00 -1.07297212e-01 2.65496540e+00
9.88753140e-01 1.29247144e-01 -4.59160954e-02 6.04205012e-01
7.16461360e-01 5.46200812e-01 -8.16257477e-01 -4.66796130e-01
-4.54385370e-01 2.40239114e-01 7.67184854e-01 6.47188947e-02
-7.99707234e-01 2.85185575e-01 5.91085434e+00 1.14422882e+00
-1.07229400e+00 -1.85145378e-01 6.23112619e-01 -6.67996049e-01
-4.00677234e-01 -1.27307788e-01 -6.80445910e-01 9.14475143e-01
1.02641523e+00 -3.01061988e-01 5.20434320e-01 1.97905287e-01
4.84332800e-01 -8.48324776e-01 -7.10591197e-01 1.47799110e+00
-1.52430532e-03 -1.31362450e+00 1.71306610e-01 1.10417441e-01
5.03058612e-01 -2.17290044e-01 -1.02065951e-01 -4.29841995e-01
-2.30954036e-01 -7.89939225e-01 8.11752915e-01 5.26633799e-01
1.20056713e+00 -8.82849038e-01 4.06868070e-01 1.11999154e-01
-1.60585308e+00 1.49576087e-03 -3.53600651e-01 1.03135921e-01
8.81007791e-01 1.05983472e+00 -4.40537959e-01 4.30614710e-01
1.10121858e+00 7.13466048e-01 -3.05453036e-02 1.10166228e+00
2.74778903e-01 5.53930044e-01 -8.76026988e-01 8.65629017e-02
-4.42517884e-02 -3.61207314e-02 6.35856390e-01 1.13812304e+00
6.92626357e-01 5.65658212e-01 -1.74696714e-01 4.77125019e-01
-1.22629493e-01 -2.76515841e-01 -5.93096495e-01 1.75451413e-01
1.04982591e+00 8.00832808e-01 -9.76713419e-01 -5.01195848e-01
-4.83970493e-01 1.34147334e+00 -1.53191537e-01 2.83169091e-01
-9.43241954e-01 -5.99121928e-01 3.64495188e-01 1.19925134e-01
5.66719115e-01 -3.46365243e-01 -3.54356080e-01 -7.16894329e-01
3.04100454e-01 -4.47753429e-01 6.74990654e-01 -9.29639101e-01
-7.62773633e-01 4.08459157e-01 2.43545219e-01 -1.40298331e+00
-2.61914790e-01 4.32512164e-02 -3.82769972e-01 4.81444210e-01
-1.47944391e+00 -4.38273668e-01 -6.87103689e-01 9.56812382e-01
5.92132866e-01 2.90366113e-01 5.71220160e-01 5.04710019e-01
-2.67703980e-01 1.94540307e-01 6.26155436e-02 -4.02608573e-01
5.67184448e-01 -9.46047425e-01 6.14641123e-02 7.88699806e-01
-1.01143532e-01 -2.74768054e-01 4.90763068e-01 -6.60602033e-01
-1.74413812e+00 -1.13068378e+00 9.04540479e-01 1.97249249e-01
1.55776694e-01 -1.40054762e-01 -9.28235292e-01 1.55075967e-01
1.31567419e-01 3.45729589e-01 6.05073571e-01 -1.10315919e+00
4.81490381e-02 -4.74082321e-01 -1.47625422e+00 3.33787918e-01
9.22168911e-01 -5.50254524e-01 -2.32510000e-01 7.27212429e-02
7.74013221e-01 -6.64907694e-01 -8.71304154e-01 3.65646750e-01
3.69687945e-01 -1.15278471e+00 1.07360780e+00 3.66130531e-01
5.40419936e-01 -6.94836736e-01 -5.38557708e-01 -4.39245164e-01
7.72198439e-02 -6.09256506e-01 -4.68825251e-01 1.10302103e+00
-2.33295366e-01 -2.82966327e-02 6.42675579e-01 5.33590138e-01
4.78061661e-02 -8.45231652e-01 -1.46574855e+00 -4.23320115e-01
-8.95633876e-01 -8.25537920e-01 6.48594737e-01 3.62458944e-01
-1.89535767e-01 -3.12930167e-01 -2.18311369e-01 1.00691199e-01
9.13551092e-01 2.01303944e-01 3.02662134e-01 -6.50273502e-01
3.99456434e-02 -2.41910025e-01 -6.45017862e-01 -1.24595428e+00
-5.22351205e-01 -5.42336226e-01 1.76923692e-01 -1.50176370e+00
2.31512010e-01 -4.12981868e-01 -1.62840173e-01 6.46762699e-02
3.66303734e-02 5.11108100e-01 6.04708306e-02 3.24813843e-01
-8.85594010e-01 6.04206979e-01 8.92439544e-01 9.14307311e-02
-2.99480647e-01 -4.67213005e-01 2.22764909e-01 2.54906803e-01
5.38849652e-01 -4.39401388e-01 -5.45484602e-01 -4.38149601e-01
3.70918602e-01 6.25620902e-01 6.43105626e-01 -1.22446120e+00
6.63727283e-01 -2.10514128e-01 4.36275363e-01 -1.24474132e+00
5.03833413e-01 -9.98849154e-01 4.84296799e-01 2.03161940e-01
-1.52785599e-01 3.01104099e-01 1.20918632e-01 7.53152847e-01
-4.15322125e-01 -5.25410008e-03 6.19099915e-01 4.65929396e-02
-7.89244235e-01 8.04774582e-01 -6.17539525e-01 2.71444097e-02
1.35918331e+00 -6.96346641e-01 3.53777246e-03 -3.77599478e-01
-3.06602359e-01 9.22112018e-02 6.57187402e-01 -9.35973451e-02
1.13866150e+00 -1.48275208e+00 -5.69167018e-01 3.45390230e-01
9.63058136e-03 5.37758470e-01 5.77753663e-01 5.65744221e-01
-6.60904646e-01 6.28620163e-02 -1.55228928e-01 -9.59261119e-01
-1.16344726e+00 4.40666735e-01 -1.58037826e-01 -5.42922318e-02
-8.86898518e-01 7.41437078e-01 -1.45513877e-01 7.17784286e-01
4.41254228e-01 -3.25101078e-01 9.91734862e-02 1.26094460e-01
9.46831346e-01 5.90750694e-01 2.96505112e-02 -4.16773260e-01
-2.17787355e-01 5.00587702e-01 2.61508077e-01 -5.16434550e-01
1.19017887e+00 -5.11907637e-01 -1.69660747e-01 4.73456562e-01
1.24541569e+00 -2.38226846e-01 -1.57005382e+00 -3.28601897e-02
-4.63918447e-01 -7.98345625e-01 1.36389792e-01 -2.96561718e-01
-1.06308389e+00 7.05140531e-01 6.84039712e-01 4.31234896e-01
1.97454119e+00 3.41566913e-02 1.43474233e+00 -2.83420652e-01
4.53295857e-01 -1.07507622e+00 -3.11035775e-02 2.13322788e-01
5.73979795e-01 -7.18623817e-01 4.47705351e-02 -3.09091568e-01
-4.09856588e-01 1.13875079e+00 9.20919627e-02 1.01278216e-01
5.15893638e-01 7.43943751e-01 -5.57288408e-01 -1.60434440e-01
-8.22494090e-01 6.80554211e-02 -1.96765170e-01 6.46902978e-01
1.72804102e-01 -3.68695557e-02 -3.40277046e-01 7.95133319e-03
-5.72225191e-02 3.08626920e-01 3.45843583e-01 1.20946908e+00
-4.76092905e-01 -8.38216305e-01 -6.24674857e-01 6.72544837e-01
-4.03691918e-01 -4.04329039e-02 3.57603520e-01 2.33839810e-01
2.22291932e-01 8.16401422e-01 8.48291039e-01 -2.11055070e-01
2.98182160e-01 -2.95338452e-01 4.86386210e-01 -7.85194570e-04
-1.72247678e-01 3.96479294e-02 -3.46356004e-01 -9.38873529e-01
-5.20965338e-01 -8.94052029e-01 -1.76605690e+00 -4.31147397e-01
2.42945656e-01 -1.25561818e-01 5.81463635e-01 5.95446944e-01
4.47109342e-01 4.87722725e-01 8.48750055e-01 -7.80008078e-01
1.53409481e-01 -2.60236055e-01 -5.65600455e-01 5.88190019e-01
4.56651896e-01 -3.44898999e-01 -4.49545503e-01 4.56271201e-01] | [8.983074188232422, -1.5410114526748657] |
ea5f4d6c-5003-492b-a4a1-2d4c2c8a188b | temporal-information-extraction-from-clinical | null | null | https://aclanthology.org/E17-2117 | https://aclanthology.org/E17-2117.pdf | Temporal information extraction from clinical text | In this paper, we present a method for temporal relation extraction from clinical narratives in French and in English. We experiment on two comparable corpora, the MERLOT corpus and the THYME corpus, and show that a common approach can be used for both languages. | ["Aur{\\'e}lie N{\\'e}v{\\'e}ol", 'Olivier Ferret', 'Xavier Tannier', 'Julien Tourille'] | 2017-04-01 | null | null | null | eacl-2017-4 | ['temporal-relation-extraction', 'temporal-information-extraction'] | ['natural-language-processing', 'natural-language-processing'] | [ 3.19504291e-02 3.99677306e-01 -6.18927598e-01 -2.47309312e-01
-7.81024575e-01 -4.47129667e-01 7.99253225e-01 2.81253010e-01
-7.26351082e-01 1.32618570e+00 5.46673417e-01 -4.16369379e-01
-2.68750936e-01 -3.52935106e-01 2.51772642e-01 -1.87788367e-01
-1.16833888e-01 5.70204377e-01 5.42280316e-01 -3.97706240e-01
-2.18427535e-02 4.55222577e-01 -6.45029485e-01 9.61156547e-01
1.89969093e-01 3.37505668e-01 -1.17003351e-01 6.01561546e-01
3.21714096e-02 1.66570139e+00 -4.36656743e-01 -7.53512263e-01
-4.38129276e-01 -7.17549741e-01 -1.65848470e+00 -8.11960474e-02
-5.88053226e-01 1.26317397e-01 -7.15911090e-01 5.49167693e-01
2.30482802e-01 -2.60648340e-01 8.03371012e-01 -6.84185326e-01
1.56588048e-01 9.76557851e-01 -2.11163521e-01 7.81639218e-01
9.73862410e-01 -4.67330128e-01 5.92381835e-01 -4.52983499e-01
1.85874796e+00 1.00051844e+00 7.36783385e-01 8.44568551e-01
-8.39417338e-01 -3.51293564e-01 -6.15066662e-02 3.86679560e-01
-1.26618981e+00 -4.76689786e-01 4.50567067e-01 -6.08041227e-01
1.39218748e+00 3.47149968e-01 8.93596590e-01 1.51892769e+00
9.83091831e-01 7.25374043e-01 1.47665524e+00 -7.51640916e-01
-2.56263316e-01 1.22549020e-01 3.31716716e-01 5.57235956e-01
-2.90856898e-01 2.29452595e-01 -7.57170618e-01 -3.89032871e-01
6.43606067e-01 -5.70522726e-01 -2.68391132e-01 3.91269624e-01
-1.49646401e+00 7.47530222e-01 -3.28257412e-01 1.04458356e+00
-3.19994599e-01 5.47411591e-02 1.11345434e+00 4.12484646e-01
5.53046644e-01 1.25274733e-01 -2.48544186e-01 -4.30825353e-01
-8.37183297e-01 4.21943992e-01 1.11203289e+00 8.43225002e-01
-5.13632417e-01 -7.33641565e-01 1.56779647e-01 4.67619061e-01
1.29698485e-01 -1.57419994e-01 6.88271582e-01 -7.59052515e-01
3.90060037e-01 1.24076158e-01 -5.74587695e-02 -6.93809927e-01
-8.30595791e-01 3.12636822e-01 -4.09532696e-01 -5.07617056e-01
3.66400748e-01 -1.57433793e-01 -5.02721727e-01 1.55223632e+00
4.07112949e-02 -2.66144335e-01 6.84744656e-01 3.34210306e-01
1.27137411e+00 2.65046090e-01 2.08269209e-01 -1.16718566e+00
1.63926077e+00 -7.38358140e-01 -1.71418810e+00 -4.88874950e-02
1.04563141e+00 -1.10686791e+00 2.22241152e-02 6.07837737e-01
-1.33726299e+00 2.55127430e-01 -6.65339649e-01 -9.25023779e-02
-3.10000479e-01 2.07391977e-01 7.12010086e-01 6.92536160e-02
-8.78530681e-01 6.36749864e-01 -1.24689114e+00 -9.43710983e-01
6.94571882e-02 2.11665064e-01 -6.85496569e-01 2.27439359e-01
-1.30249953e+00 1.43869877e+00 8.48532915e-01 4.79993820e-02
-4.33282465e-01 5.27371652e-02 -6.39448225e-01 -7.83222198e-01
2.48110503e-01 -3.27670902e-01 1.78486216e+00 -2.98283905e-01
-1.16294003e+00 1.56173766e+00 -1.68223068e-01 -5.59469521e-01
5.43342173e-01 1.19379461e-01 -1.14873374e+00 4.06615555e-01
8.16266164e-02 6.67987466e-02 1.36349574e-01 -6.69310391e-01
-5.34865499e-01 -1.67935058e-01 -7.30354190e-02 -7.47458637e-02
1.04567796e-01 1.02543640e+00 -3.42044204e-01 -1.05681467e+00
1.81732148e-01 -8.55186343e-01 -3.99345607e-01 -3.96089792e-01
-3.21504831e-01 -4.75851238e-01 2.42374882e-01 -1.05478990e+00
1.91660094e+00 -1.89404464e+00 3.57822388e-01 1.16649959e-02
2.95037001e-01 -4.06500518e-01 4.49418962e-01 1.02885687e+00
-5.27275145e-01 1.58048809e-01 -3.76812339e-01 7.82235339e-02
-4.89694297e-01 8.59595120e-01 -1.63401544e-01 6.08871818e-01
1.19125508e-01 8.76981497e-01 -1.15272713e+00 -1.33585131e+00
-3.65532070e-01 4.08905819e-02 -3.62287849e-01 -1.55278966e-01
1.11876383e-01 4.09935057e-01 -6.42932832e-01 4.32917893e-01
-1.82189360e-01 -6.88449591e-02 9.11863208e-01 -1.44345323e-02
-3.66846085e-01 9.93531287e-01 -7.04919338e-01 1.72634542e+00
-1.50552601e-01 7.69052625e-01 -1.31779686e-01 -5.97478569e-01
6.23470247e-01 1.45936120e+00 5.85485160e-01 -3.44765782e-01
4.05836076e-01 4.51415658e-01 1.24792285e-01 -7.86602378e-01
4.79558259e-01 -8.30486357e-01 -3.45916331e-01 6.18990242e-01
9.13734436e-02 -2.02938393e-01 9.38427329e-01 2.82728493e-01
1.13649821e+00 6.90090880e-02 1.24166644e+00 -3.41188937e-01
8.07684839e-01 4.54949945e-01 6.29960716e-01 -1.15536161e-01
-2.78590441e-01 9.22508761e-02 9.71049249e-01 -6.64916515e-01
-6.48561656e-01 -6.47764802e-01 -6.33284926e-01 8.93690363e-02
-4.09844369e-01 -1.27067590e+00 -5.88431478e-01 -1.02819610e+00
-7.84248590e-01 5.35320759e-01 -9.23604846e-01 2.82223374e-01
-1.15623462e+00 -9.22141194e-01 8.25763762e-01 5.83668172e-01
-2.80158311e-01 -1.38819194e+00 -6.85592353e-01 5.40633559e-01
-6.22032106e-01 -1.50208068e+00 -2.98680425e-01 2.81767905e-01
-9.62791324e-01 -1.19592166e+00 -2.73095906e-01 -7.89629281e-01
1.95591554e-01 -7.05314636e-01 1.16900909e+00 -1.05909988e-01
-1.91438213e-01 2.17042223e-01 -5.26939511e-01 -3.36534351e-01
-1.14882660e+00 -2.85494141e-02 -2.61519998e-02 -7.84157932e-01
3.93466294e-01 -4.99639302e-01 2.78590083e-01 2.09701478e-01
-8.59455705e-01 4.08334918e-02 7.19574839e-02 8.94363403e-01
2.50366569e-01 -1.50203362e-01 3.85399878e-01 -1.43529093e+00
9.08286095e-01 -4.47365671e-01 2.65308376e-03 4.20094980e-03
-4.97120827e-01 -9.80709940e-02 1.05039947e-01 -3.66115570e-01
-1.01511216e+00 -1.29370913e-01 -6.34473145e-01 2.24880114e-01
9.95462313e-02 9.49837744e-01 5.02627850e-01 3.60074133e-01
9.09765542e-01 -1.76203668e-01 -1.55634284e-01 -4.04624522e-01
2.54683107e-01 4.67094988e-01 6.34645879e-01 -1.97605044e-01
1.43835753e-01 5.62817633e-01 -3.78218815e-02 -6.10338688e-01
-3.73824090e-01 -3.99680197e-01 -8.41076851e-01 -3.18696737e-01
1.03879488e+00 -8.24145794e-01 -2.60052830e-01 1.83543805e-02
-1.67870319e+00 -5.93837239e-02 -5.66579163e-01 9.45851803e-01
-8.29196155e-01 2.12481186e-01 -1.20844769e+00 -5.74065268e-01
-7.09500685e-02 -7.96399534e-01 5.92150390e-01 -3.28058362e-01
-9.92544591e-01 -1.51184154e+00 7.18285441e-01 -2.11042613e-01
-5.27359724e-01 4.96009529e-01 8.51794720e-01 -6.77804351e-01
2.47626394e-01 -1.03541680e-01 1.79713368e-01 -3.80747169e-01
2.81550378e-01 4.96973768e-02 -5.25518715e-01 2.74888258e-02
4.65976566e-01 -4.62772727e-01 5.33625603e-01 9.21595842e-02
3.37462574e-01 -2.52202183e-01 -9.34380412e-01 1.13998756e-01
1.30276167e+00 8.66240621e-01 7.63053596e-01 3.08354527e-01
-6.41858345e-03 8.09628010e-01 8.30249012e-01 6.62960410e-02
2.77273238e-01 6.14652753e-01 -4.69505459e-01 1.61604285e-01
-1.67931840e-02 6.48849607e-02 2.29632437e-01 1.65448081e+00
-5.14635861e-01 -3.39420676e-01 -1.33756244e+00 9.28764582e-01
-1.84486735e+00 -9.07175958e-01 -3.45511973e-01 1.10940242e+00
1.40733314e+00 2.12266386e-01 2.18502074e-01 3.20991874e-01
3.57621282e-01 1.35277644e-01 4.96549785e-01 -7.72341311e-01
-2.32431307e-01 3.67399842e-01 2.07242638e-01 5.31298459e-01
-9.59588408e-01 9.45877016e-01 8.48830223e+00 8.29442024e-01
-6.71928823e-01 5.66599488e-01 2.66683966e-01 8.44636559e-02
-1.13240592e-01 -5.95269911e-02 -5.65643549e-01 -1.25264868e-01
1.58886576e+00 -4.89375651e-01 -2.29313239e-01 2.23773494e-01
2.87373036e-01 -1.59059882e-01 -1.14596915e+00 9.38508391e-01
3.54499333e-02 -1.60160482e+00 -6.36835635e-01 5.35269715e-02
4.60153162e-01 -2.63806581e-01 -5.42516291e-01 -2.40884781e-01
1.00338168e-01 -9.67968762e-01 7.63634562e-01 5.89673519e-01
1.00108886e+00 -4.49737579e-01 9.56787705e-01 7.24544078e-02
-1.25454140e+00 3.02004755e-01 1.63938239e-01 6.68710694e-02
7.02441573e-01 8.39157626e-02 -9.77477789e-01 1.15958202e+00
4.18991446e-01 1.09068906e+00 -1.43212095e-01 4.95918155e-01
-5.68319619e-01 5.16903520e-01 -9.31726247e-02 -8.10264945e-02
3.24874938e-01 -5.78919686e-02 6.76932871e-01 1.47663593e+00
1.08369015e-01 7.01148033e-01 -2.59269774e-02 2.84279436e-01
3.59623462e-01 5.62873363e-01 -8.08864534e-01 -6.41900122e-01
-1.96302772e-01 1.02872837e+00 -1.15642273e+00 -3.44473690e-01
-5.44722080e-01 8.93673003e-01 -8.54985639e-02 -1.25071794e-01
-7.05115020e-01 -3.80916633e-02 -2.88699530e-02 2.16435775e-01
-1.13865189e-01 -7.99276233e-02 9.32038650e-02 -1.12629032e+00
-2.31495589e-01 -1.13049483e+00 8.96161973e-01 -6.46699846e-01
-1.10285544e+00 1.43902993e+00 3.82397413e-01 -1.25062430e+00
-9.13263559e-01 -6.42612219e-01 -2.20120743e-01 7.31106758e-01
-8.55695248e-01 -1.08894098e+00 6.20556951e-01 7.08696187e-01
4.56435740e-01 -8.23826641e-02 1.08540976e+00 5.81109405e-01
-5.07434964e-01 8.37660208e-02 -3.37413400e-01 1.24274328e-01
9.88840938e-01 -9.75381792e-01 3.03457081e-01 5.29709637e-01
7.72857456e-04 6.86163545e-01 7.15462565e-01 -9.14812744e-01
-7.08268285e-01 -5.24423778e-01 1.98759615e+00 -4.91867810e-01
1.15227246e+00 1.66960154e-02 -6.29203498e-01 1.00939894e+00
6.65966690e-01 -4.63253081e-01 6.91747606e-01 -6.85923696e-02
7.18505755e-02 6.26232207e-01 -1.01065636e+00 6.37315810e-01
1.11715102e+00 -7.07451284e-01 -1.24701786e+00 8.19784820e-01
5.65177977e-01 -6.85657501e-01 -1.45915139e+00 4.18171674e-01
5.04493713e-01 -5.18835545e-01 4.77529049e-01 -9.28633630e-01
5.98706543e-01 -8.80968496e-02 4.16846350e-02 -8.48880529e-01
8.27706158e-02 -1.09085631e+00 4.32880782e-02 8.84799480e-01
7.43464589e-01 -5.66772401e-01 2.90563911e-01 4.46423329e-02
-1.51183382e-02 -8.82587373e-01 -1.41972780e+00 -5.19133747e-01
2.47434422e-01 -6.23328328e-01 -4.61320616e-02 1.20660651e+00
9.13767457e-01 6.79155231e-01 -8.02238584e-02 -4.66692209e-01
-1.65034533e-01 -1.31127238e-01 -1.17497696e-02 -9.76228476e-01
-5.19813709e-02 -1.68900937e-01 -3.13050985e-01 -2.76299804e-01
3.32056284e-01 -1.03216803e+00 -2.20844954e-01 -1.51528072e+00
3.29499006e-01 -3.47574651e-02 2.28427172e-01 5.81063628e-01
3.18617880e-01 1.55778453e-01 -1.68428272e-01 1.87191412e-01
-1.96920678e-01 6.26122430e-02 1.13412893e+00 1.14620082e-01
-1.04040489e-01 -4.19428080e-01 -2.52045870e-01 1.06515443e+00
4.60945129e-01 -1.07224786e+00 -2.43578315e-01 2.83281714e-01
6.01702273e-01 7.48826444e-01 -8.47023949e-02 -4.05815482e-01
3.07926983e-01 -2.27072805e-01 -2.43586481e-01 -9.37827349e-01
6.87547326e-02 -5.13765633e-01 6.70590520e-01 8.58131349e-01
-1.92519382e-01 7.34104931e-01 5.99608302e-01 2.52814054e-01
-6.24563098e-01 -6.41740024e-01 7.95630336e-01 -3.52033496e-01
-3.34384352e-01 -1.80833921e-01 -9.06283617e-01 2.93092519e-01
1.12937117e+00 1.19714141e-01 -2.60667980e-01 -9.58560333e-02
-1.23737609e+00 4.96449918e-02 3.47698450e-01 1.36982322e-01
7.07479000e-01 -1.40201616e+00 -9.46437836e-01 -3.09702963e-01
1.84621178e-02 -8.54363203e-01 4.49458770e-02 1.51100874e+00
-8.78784180e-01 8.06090713e-01 -3.61535847e-02 -1.72401547e-01
-1.61328864e+00 8.88719797e-01 3.39470536e-01 -1.04611492e+00
-8.32948089e-01 4.31669056e-01 -8.33130181e-02 -3.40183564e-02
-3.68576705e-01 -4.86397207e-01 -7.21984863e-01 4.69648689e-01
4.89184618e-01 -4.60079592e-03 -7.28452504e-02 -8.70110512e-01
-8.32928777e-01 2.70147473e-01 -1.65400296e-01 -9.06125546e-01
1.44495058e+00 -2.24093790e-03 -7.37464905e-01 9.64565039e-01
9.10534561e-01 4.17205930e-01 1.23723239e-01 7.83120934e-03
5.67914307e-01 2.74457335e-01 -1.73974097e-01 -6.15887642e-01
-6.48271680e-01 1.41279846e-01 2.53852010e-01 3.72695535e-01
1.27270353e+00 4.05131072e-01 6.03152394e-01 9.95452031e-02
2.46163100e-01 -1.04553354e+00 -3.85331124e-01 5.38142383e-01
1.03657556e+00 -3.95436198e-01 3.18881094e-01 -9.51255679e-01
-7.20587075e-01 1.34786940e+00 -2.20646575e-01 9.11789834e-02
7.93908536e-01 7.42412627e-01 1.81333005e-01 -6.98451638e-01
-1.40588677e+00 -5.05685359e-02 2.88929343e-01 1.60929158e-01
1.18004000e+00 -4.28421050e-02 -1.70909441e+00 7.68557429e-01
-2.52486318e-01 4.08784240e-01 9.83118474e-01 1.44962215e+00
4.39210147e-01 -1.73419881e+00 -4.08538759e-01 1.33386016e-01
-1.19735527e+00 -2.54027933e-01 -7.48696804e-01 1.47380435e+00
2.91326106e-01 8.01641405e-01 -1.26005664e-01 -2.02508494e-01
4.07003939e-01 3.23601246e-01 1.04214191e+00 -6.38782740e-01
-9.17156458e-01 7.20560014e-01 1.26974821e+00 -5.45420349e-01
-1.32122719e+00 -1.16855812e+00 -1.35929966e+00 4.43283468e-02
-2.47844934e-01 4.55374092e-01 7.20711499e-02 9.96128142e-01
-6.20480716e-01 9.19102192e-01 3.27261202e-02 1.70562059e-01
2.28427783e-01 -1.01427984e+00 -7.06498802e-01 -5.22968173e-02
-3.63622978e-02 -6.84922040e-01 -5.91237694e-02 4.70427811e-01] | [8.571464538574219, 8.991735458374023] |
cedc9ade-3071-4f19-9458-6e0441b5a1f1 | mist-multiple-instance-self-training | 2104.01633 | null | https://arxiv.org/abs/2104.01633v1 | https://arxiv.org/pdf/2104.01633v1.pdf | MIST: Multiple Instance Self-Training Framework for Video Anomaly Detection | Weakly supervised video anomaly detection (WS-VAD) is to distinguish anomalies from normal events based on discriminative representations. Most existing works are limited in insufficient video representations. In this work, we develop a multiple instance self-training framework (MIST)to efficiently refine task-specific discriminative representations with only video-level annotations. In particular, MIST is composed of 1) a multiple instance pseudo label generator, which adapts a sparse continuous sampling strategy to produce more reliable clip-level pseudo labels, and 2) a self-guided attention boosted feature encoder that aims to automatically focus on anomalous regions in frames while extracting task-specific representations. Moreover, we adopt a self-training scheme to optimize both components and finally obtain a task-specific feature encoder. Extensive experiments on two public datasets demonstrate the efficacy of our method, and our method performs comparably to or even better than existing supervised and weakly supervised methods, specifically obtaining a frame-level AUC 94.83% on ShanghaiTech. | ['Wei-Shi Zheng', 'Fa-Ting Hong', 'Jia-Chang Feng'] | 2021-04-04 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Feng_MIST_Multiple_Instance_Self-Training_Framework_for_Video_Anomaly_Detection_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Feng_MIST_Multiple_Instance_Self-Training_Framework_for_Video_Anomaly_Detection_CVPR_2021_paper.pdf | cvpr-2021-1 | ['anomaly-detection-in-surveillance-videos', 'anomaly-detection-in-surveillance-videos'] | ['computer-vision', 'methodology'] | [ 4.11682487e-01 -1.46397933e-01 -1.46218613e-01 -6.14837289e-01
-1.05624354e+00 -2.24294379e-01 4.91744101e-01 7.39235803e-02
-2.28395090e-01 4.57877606e-01 2.89197594e-01 7.29042962e-02
4.03858006e-01 -3.43791932e-01 -9.40694273e-01 -6.09881997e-01
-1.96269572e-01 1.03099950e-01 3.01275730e-01 1.06059954e-01
7.04870000e-02 6.10672534e-02 -1.55096745e+00 4.56958383e-01
1.04686689e+00 1.28706193e+00 -1.59117445e-01 5.76040983e-01
1.80461600e-01 1.35117328e+00 -3.68732065e-01 -2.03632042e-01
1.59998149e-01 -6.71870828e-01 -7.48202264e-01 6.27685368e-01
4.98999864e-01 -6.95346117e-01 -4.71262395e-01 9.45094168e-01
2.22156331e-01 1.79169506e-01 6.81410372e-01 -1.18140793e+00
-5.36498427e-01 1.06588729e-01 -7.37666607e-01 7.56017208e-01
3.61626565e-01 4.20533985e-01 1.12891006e+00 -8.76093566e-01
4.24257100e-01 8.94031107e-01 4.29935336e-01 6.18364871e-01
-1.08694911e+00 -5.29944599e-01 5.38957238e-01 5.23146272e-01
-1.34792185e+00 -5.09861052e-01 9.05040026e-01 -5.48370183e-01
8.28886569e-01 6.23313673e-02 4.35342044e-01 1.57180357e+00
-2.55756322e-02 1.10893953e+00 5.50674379e-01 -1.47051141e-01
2.47254953e-01 -1.46480292e-01 2.64263507e-02 9.04985845e-01
1.66499585e-01 -2.29168594e-01 -5.32532454e-01 -3.21769774e-01
6.71042919e-01 1.69654965e-01 -2.98803568e-01 -4.72113550e-01
-1.18279469e+00 6.82264388e-01 7.14196265e-02 3.14811200e-01
-6.66464210e-01 -1.28498941e-03 7.57811248e-01 2.88220376e-01
9.39194202e-01 2.33270317e-01 -3.50591332e-01 -2.84128517e-01
-9.09188688e-01 1.43184423e-01 2.79250294e-01 8.38396788e-01
6.33923411e-01 4.56604838e-01 -7.01739430e-01 9.65561092e-01
2.00267524e-01 1.05566874e-01 7.05222130e-01 -8.77194345e-01
3.72472405e-01 5.82358301e-01 -1.12924680e-01 -7.64731765e-01
-1.83717981e-01 -6.26173496e-01 -7.10623682e-01 -8.65659118e-02
1.65783241e-01 -1.23871759e-01 -9.57102239e-01 1.92201638e+00
1.89215571e-01 9.80910301e-01 -5.82449175e-02 9.12029684e-01
5.99814534e-01 5.99644363e-01 3.17004025e-01 -2.97002107e-01
1.14065754e+00 -1.18587995e+00 -6.63793266e-01 -2.00590342e-01
8.89677644e-01 -3.22488219e-01 8.57630014e-01 2.49650016e-01
-1.09880483e+00 -6.52466595e-01 -9.83890712e-01 1.54420123e-01
1.80116206e-01 2.19438151e-01 2.03898266e-01 2.53406584e-01
-8.40802491e-01 3.87901932e-01 -9.15298641e-01 -1.46482736e-01
9.91273224e-01 4.28948440e-02 -3.99035364e-01 -1.50543720e-01
-9.12079632e-01 4.16039288e-01 3.45224947e-01 -1.27290254e-02
-1.44731975e+00 -6.83821559e-01 -1.29091048e+00 3.48045193e-02
5.66812694e-01 -6.23290002e-01 1.17264998e+00 -1.40461779e+00
-1.32688224e+00 8.85923266e-01 -4.61773455e-01 -7.22747028e-01
4.84386414e-01 -4.48314488e-01 -4.33562756e-01 5.27352512e-01
2.97966003e-01 5.33484399e-01 1.23466027e+00 -9.65717196e-01
-8.79270315e-01 -1.57760054e-01 -1.29039705e-01 2.52990067e-01
-5.78251243e-01 5.45294099e-02 -7.52526820e-01 -1.01077235e+00
-9.57296938e-02 -5.76807261e-01 -2.70493299e-01 -2.02451766e-01
-3.49712789e-01 -4.33460116e-01 8.53335738e-01 -8.33584487e-01
1.27676988e+00 -2.33775735e+00 1.98141158e-01 1.44819707e-01
4.61219877e-01 4.04016584e-01 -2.45565325e-01 -1.66374862e-01
-3.07961226e-01 -2.43798599e-01 -3.30778331e-01 -5.26103795e-01
-3.22557986e-01 1.33437052e-01 -1.37266234e-01 6.44909263e-01
8.67107213e-01 8.07557285e-01 -1.15454662e+00 -5.46051085e-01
1.85901001e-01 2.66198158e-01 -7.59006321e-01 6.65684938e-01
-2.02177420e-01 6.31891906e-01 -7.00809121e-01 1.04853463e+00
3.98784935e-01 -3.19551319e-01 -2.09745929e-01 -3.19002271e-02
1.58662975e-01 4.40411121e-02 -7.66575694e-01 1.79234910e+00
-9.08632204e-02 5.53284824e-01 -2.90285498e-01 -1.49496412e+00
8.18927169e-01 4.09542233e-01 7.64347374e-01 -6.76380873e-01
8.66036788e-02 1.17913581e-01 -3.35727394e-01 -8.63456666e-01
1.62376806e-01 3.74117106e-01 2.01793730e-01 2.53875881e-01
4.91373628e-01 6.06063902e-01 2.66161352e-01 3.86924714e-01
1.52548313e+00 4.49801296e-01 1.92212939e-01 -1.56145832e-02
9.60818946e-01 -3.11483473e-01 9.62411344e-01 6.39353693e-01
-5.06279349e-01 9.85965550e-01 7.62390912e-01 -6.41730249e-01
-9.58537102e-01 -8.84605169e-01 2.22370643e-02 1.04876590e+00
4.09100763e-02 -5.60596585e-01 -7.49040604e-01 -1.33108592e+00
-1.75995022e-01 6.75635338e-01 -8.47148716e-01 -5.63954711e-01
-5.96663117e-01 -5.98279774e-01 3.67460072e-01 6.48370683e-01
3.34481597e-01 -1.08413339e+00 -4.17111874e-01 2.04993695e-01
-4.20727670e-01 -1.33296788e+00 -6.54495537e-01 4.63681668e-02
-7.78274059e-01 -1.17299449e+00 -7.63705611e-01 -6.99501395e-01
8.62032712e-01 2.72844672e-01 1.13292575e+00 1.42813772e-01
-2.15610966e-01 4.92567360e-01 -7.61978924e-01 -1.28664821e-01
-2.10786402e-01 -2.42847651e-02 3.93467583e-02 6.52657866e-01
5.17073989e-01 -4.34222072e-01 -6.16866887e-01 2.29290783e-01
-9.43511903e-01 -1.84450865e-01 5.59297025e-01 8.93816710e-01
8.03649902e-01 -2.86115825e-01 7.13332653e-01 -9.33015168e-01
1.83062136e-01 -9.52904105e-01 -2.76692450e-01 -2.78054532e-02
-4.00328904e-01 -3.84472869e-02 6.72883630e-01 -3.94104838e-01
-1.04163420e+00 3.35038006e-01 -1.50399327e-01 -9.98004973e-01
-4.94742274e-01 1.14036329e-01 -1.03064939e-01 2.40991414e-01
6.07398927e-01 5.86388707e-01 -3.93349007e-02 -2.87447512e-01
-1.12575367e-01 4.49923426e-01 6.93233192e-01 -6.48835063e-01
8.76132846e-01 4.38988030e-01 -4.14348781e-01 -6.73574865e-01
-1.17236507e+00 -5.58461189e-01 -5.67796111e-01 -2.34613404e-01
9.61862743e-01 -1.30044436e+00 -5.05060963e-02 4.66185898e-01
-6.51776016e-01 -2.70692974e-01 -5.33027470e-01 4.77528632e-01
-6.81709468e-01 4.65419948e-01 -5.45990825e-01 -7.30587959e-01
-1.99921861e-01 -1.11146665e+00 1.34407842e+00 1.87255159e-01
-1.43002868e-01 -7.31906772e-01 1.59458637e-01 3.27871472e-01
2.51040727e-01 5.01132905e-01 4.18863535e-01 -9.52900589e-01
-3.89805883e-01 -2.64119297e-01 -3.03427219e-01 7.43794143e-01
1.59101993e-01 -1.13955721e-01 -1.01610756e+00 -3.27457249e-01
-2.31433272e-01 -5.14857888e-01 1.16534483e+00 3.96430612e-01
1.67954397e+00 -1.63557649e-01 -1.59426391e-01 8.05870831e-01
9.77730155e-01 3.54724713e-02 6.60378575e-01 4.52935338e-01
8.66174102e-01 2.84848541e-01 8.00390303e-01 6.37984931e-01
3.60012323e-01 5.93029082e-01 5.52587926e-01 -1.07293718e-01
-4.29145247e-03 -1.16582498e-01 6.65441930e-01 4.05983478e-01
-2.63510317e-01 -7.27112293e-02 -5.21357834e-01 7.25112677e-01
-1.96985245e+00 -1.03809023e+00 4.30093333e-02 2.03348446e+00
6.41448379e-01 1.05219960e-01 4.61728483e-01 1.79390553e-02
6.52454436e-01 3.66197079e-01 -7.26528347e-01 -4.89193499e-02
-4.06305045e-02 -1.57237053e-02 8.29595402e-02 -7.22955912e-02
-1.62998497e+00 8.71127963e-01 5.35816431e+00 7.74652064e-01
-8.10801566e-01 1.53465286e-01 1.01934278e+00 -3.07548821e-01
-4.83583920e-02 -2.25351200e-01 -4.42001313e-01 7.96509147e-01
9.79744017e-01 1.35660367e-02 -4.50549833e-02 1.01625717e+00
2.16635868e-01 2.96276242e-01 -1.06783080e+00 9.85902488e-01
4.32142079e-01 -1.08538079e+00 3.12525421e-01 5.38637163e-03
7.31247783e-01 -2.81651914e-02 -7.97348246e-02 5.01336753e-01
-3.70423757e-02 -7.36462474e-01 7.45509446e-01 6.31820321e-01
7.77476072e-01 -8.23282540e-01 8.49462330e-01 2.87327613e-03
-1.22074091e+00 -2.16367275e-01 -3.03157955e-01 2.11935282e-01
1.53817162e-01 5.54377496e-01 -6.03205562e-01 6.11826718e-01
9.27370012e-01 1.31850612e+00 -6.85584009e-01 1.07640874e+00
-9.79751348e-02 1.13018000e+00 6.57538101e-02 5.03417790e-01
3.68588686e-01 -7.57998973e-02 7.07727134e-01 1.38128698e+00
1.90284967e-01 -3.13285254e-02 3.28571439e-01 6.38316512e-01
-1.34319440e-01 1.84245989e-01 -4.88561660e-01 8.25185552e-02
1.06459066e-01 1.30796075e+00 -3.89338136e-01 -4.51443672e-01
-8.14257860e-01 1.37797761e+00 4.18525219e-01 4.86787826e-01
-1.14047861e+00 -1.52897552e-01 8.73156488e-01 -6.46683052e-02
4.54250872e-01 1.02328040e-01 1.53009340e-01 -1.50567901e+00
1.72845423e-01 -9.32209671e-01 7.91499674e-01 -4.73709404e-01
-1.27622795e+00 7.51023829e-01 -2.78919369e-01 -1.76179922e+00
-4.41599071e-01 -2.03652099e-01 -7.52007544e-01 3.66558343e-01
-1.63320231e+00 -1.10814798e+00 -5.17882824e-01 8.88433158e-01
1.02924395e+00 -4.43281531e-01 6.33705318e-01 3.48304629e-01
-1.21858907e+00 8.40551257e-01 -2.32809901e-01 3.32255095e-01
7.66437471e-01 -1.25846076e+00 1.45521745e-01 1.25868499e+00
1.64211914e-01 -3.02137192e-02 4.60580200e-01 -5.62005162e-01
-9.63490784e-01 -1.56902504e+00 4.06351000e-01 -4.31683004e-01
4.83140141e-01 -2.34072477e-01 -1.27473295e+00 9.14858639e-01
-1.20868973e-01 5.78667819e-01 6.93217099e-01 -5.71854711e-02
-5.30545413e-01 -1.04137287e-01 -1.07849514e+00 2.69580543e-01
1.33786106e+00 -2.95798153e-01 -5.10543466e-01 4.40132618e-01
6.30718410e-01 -5.35605669e-01 -5.27120829e-01 6.22198343e-01
2.65263915e-01 -9.56627429e-01 8.66307557e-01 -8.36737573e-01
5.93304873e-01 -3.50376934e-01 -1.45649880e-01 -1.11893809e+00
-5.43349266e-01 -4.61939573e-01 -7.03678489e-01 1.42539024e+00
1.60451382e-01 -4.77775276e-01 6.86920464e-01 2.21742377e-01
-5.51169634e-01 -8.57091725e-01 -7.62549400e-01 -7.16807067e-01
-5.72373271e-01 -5.32684267e-01 3.87706220e-01 1.04751050e+00
-2.43098900e-01 -5.97058004e-03 -6.36487007e-01 4.47729826e-01
5.86214602e-01 -2.25555658e-01 5.95795989e-01 -1.02022457e+00
-2.13535443e-01 -2.63432026e-01 -7.25273371e-01 -9.46742356e-01
4.90004808e-01 -6.92620933e-01 8.27323645e-02 -9.10132766e-01
3.54166061e-01 -1.16638951e-01 -8.63728821e-01 6.37874663e-01
-5.94994664e-01 4.69870687e-01 -4.47726101e-01 2.13539466e-01
-1.26651311e+00 7.98800945e-01 8.52677941e-01 -3.01644728e-02
-8.90298709e-02 6.03669249e-02 -8.34650517e-01 8.03136289e-01
6.43954515e-01 -3.12543154e-01 -3.86320561e-01 -3.34573597e-01
-2.91732043e-01 -2.37086803e-01 3.37838531e-01 -1.16885650e+00
-2.92677999e-01 3.87176895e-03 6.50718510e-01 -3.61334026e-01
4.91650701e-02 -6.35291815e-01 -5.15457690e-01 2.82363474e-01
-4.77452129e-01 -1.46120831e-01 2.16623731e-02 9.34797049e-01
-4.55752373e-01 -1.11125022e-01 7.56413519e-01 8.65021497e-02
-9.89934444e-01 7.30912745e-01 -4.07519639e-01 1.80173546e-01
1.35408080e+00 -6.81699365e-02 -8.86865407e-02 -5.19564152e-01
-5.75095952e-01 4.48482633e-01 2.95783192e-01 6.67458534e-01
6.73201501e-01 -1.46348429e+00 -1.10190463e+00 4.60013390e-01
5.41586161e-01 4.54906374e-02 4.89036471e-01 8.72653067e-01
-3.13592941e-01 1.03644796e-01 -1.25511810e-01 -8.39760423e-01
-1.12074494e+00 4.29299951e-01 2.31137827e-01 -1.63932875e-01
-9.23515320e-01 9.59521949e-01 3.13176602e-01 2.23729953e-01
3.30708057e-01 -1.02944873e-01 -4.69470829e-01 -8.67153257e-02
9.99650180e-01 2.63464808e-01 -3.40768732e-02 -8.47771168e-01
-4.23333287e-01 2.27598056e-01 -3.59716117e-01 3.78502578e-01
1.42945504e+00 -5.55067249e-02 2.48254910e-01 2.76054621e-01
1.23721933e+00 -1.41363010e-01 -1.77332735e+00 -4.56591994e-01
-3.06680799e-02 -6.45310402e-01 2.00055689e-02 -3.20638299e-01
-1.34518588e+00 4.50020939e-01 6.99287951e-01 1.30908126e-02
1.45958853e+00 1.99179411e-01 9.87166405e-01 1.11012004e-01
-7.25704357e-02 -9.51086938e-01 5.02819180e-01 3.07844639e-01
6.65802002e-01 -1.37738502e+00 -3.23653221e-01 -2.42954776e-01
-9.15913999e-01 9.00824368e-01 1.02791667e+00 -4.39988613e-01
2.79304236e-01 -1.60529479e-01 -1.10637955e-01 -1.50630236e-01
-8.76269281e-01 -3.68449807e-01 6.07588708e-01 6.33802414e-01
3.67910028e-01 -3.38222235e-01 -5.78753352e-02 6.78638279e-01
3.92187476e-01 4.34321575e-02 3.26445460e-01 8.22550893e-01
-4.16476727e-01 -8.45907450e-01 -6.95407242e-02 8.83747399e-01
-6.65100694e-01 1.71390340e-01 2.61584464e-02 2.63920128e-01
1.46592289e-01 8.84224236e-01 3.90264839e-01 -5.64171135e-01
3.50746036e-01 2.71980107e-01 2.02091057e-02 -5.92689693e-01
-1.52167529e-01 4.76392694e-02 9.70319137e-02 -1.11597347e+00
-2.70505846e-01 -9.49989796e-01 -1.00873208e+00 2.41333857e-01
-7.57934079e-02 7.71199837e-02 7.40548596e-02 1.14280331e+00
5.70193768e-01 8.23498905e-01 1.02655280e+00 -8.04510355e-01
-2.11194530e-01 -9.34510291e-01 -4.56795365e-01 9.21807945e-01
6.21328294e-01 -8.44377160e-01 -4.17433947e-01 2.23866433e-01] | [7.851104259490967, 1.5926729440689087] |
4aaa129e-6b6d-4b9e-b47d-12a27fa3b330 | transtrack-multiple-object-tracking-with | 2012.15460 | null | https://arxiv.org/abs/2012.15460v2 | https://arxiv.org/pdf/2012.15460v2.pdf | TransTrack: Multiple Object Tracking with Transformer | In this work, we propose TransTrack, a simple but efficient scheme to solve the multiple object tracking problems. TransTrack leverages the transformer architecture, which is an attention-based query-key mechanism. It applies object features from the previous frame as a query of the current frame and introduces a set of learned object queries to enable detecting new-coming objects. It builds up a novel joint-detection-and-tracking paradigm by accomplishing object detection and object association in a single shot, simplifying complicated multi-step settings in tracking-by-detection methods. On MOT17 and MOT20 benchmark, TransTrack achieves 74.5\% and 64.5\% MOTA, respectively, competitive to the state-of-the-art methods. We expect TransTrack to provide a novel perspective for multiple object tracking. The code is available at: \url{https://github.com/PeizeSun/TransTrack}. | ['Ping Luo', 'Changhu Wang', 'Zehuan Yuan', 'Enze Xie', 'Jinkun Cao', 'Rufeng Zhang', 'Yi Jiang', 'Peize Sun'] | 2020-12-31 | null | null | null | null | ['multiple-object-tracking-with-transformer'] | ['computer-vision'] | [-2.52421945e-01 -5.40594220e-01 -4.06248242e-01 -5.47786355e-02
-9.69415724e-01 -5.02909184e-01 5.00867069e-01 -1.67087734e-01
-5.03870010e-01 4.92311627e-01 -1.26491398e-01 3.81893814e-02
1.06485486e-01 -6.02962196e-01 -8.63156021e-01 -5.58100879e-01
1.43096549e-02 5.77854037e-01 1.08003342e+00 6.32546097e-02
4.53377962e-02 3.58600706e-01 -1.59362948e+00 1.48310259e-01
2.57282734e-01 1.05559790e+00 5.28126121e-01 9.56100464e-01
8.90902504e-02 8.34304512e-01 -4.53265667e-01 -4.94601130e-01
2.70681739e-01 -5.42298630e-02 -7.07249522e-01 -2.56124228e-01
9.36190426e-01 -5.90982735e-01 -7.51985133e-01 9.67195630e-01
5.21252573e-01 1.72140554e-01 2.03680605e-01 -1.49703288e+00
-8.60496521e-01 4.58823681e-01 -7.93794394e-01 9.61279869e-01
2.33667076e-01 4.97884959e-01 1.12809741e+00 -1.23561442e+00
4.57612425e-01 1.45157242e+00 6.68718815e-01 5.99922240e-01
-1.11009157e+00 -1.04720938e+00 3.41156751e-01 3.82451981e-01
-1.56102741e+00 -5.98685980e-01 1.37549475e-01 -4.12342966e-01
6.91662431e-01 3.14848542e-01 6.84081018e-01 7.85858631e-01
3.03131039e-03 1.18122518e+00 4.98725682e-01 -1.15962941e-02
-3.48316073e-01 -2.51654476e-01 3.88974756e-01 8.57363999e-01
4.04310316e-01 2.35577345e-01 -6.07556462e-01 -4.93468791e-02
5.61784148e-01 4.67439324e-01 -1.79784283e-01 -2.31251553e-01
-1.50387704e+00 6.24616146e-01 7.60878444e-01 1.54552281e-01
-3.75554383e-01 7.20424235e-01 2.37059683e-01 -1.60218757e-02
2.01887622e-01 4.75080349e-02 -2.71754295e-01 -1.63441360e-01
-7.55982637e-01 3.86036783e-01 4.44830894e-01 1.30828261e+00
6.89602256e-01 -8.81872252e-02 -7.65451312e-01 2.60411650e-01
4.82947648e-01 9.54068601e-01 3.41477729e-02 -1.01041996e+00
2.99151629e-01 5.23527980e-01 2.97127664e-01 -7.61977971e-01
-9.54122916e-02 -4.45729047e-01 -2.65172809e-01 5.69572859e-02
3.27734411e-01 6.52270541e-02 -8.51417959e-01 1.65077543e+00
7.23079383e-01 8.16611230e-01 -2.65391350e-01 1.02718067e+00
1.01035333e+00 4.61935937e-01 2.25075811e-01 -1.09169632e-02
1.68415499e+00 -1.40595257e+00 -7.32144713e-01 -5.08514873e-04
3.18976343e-01 -8.18881214e-01 8.45099449e-01 5.32057881e-02
-1.17402112e+00 -8.51544738e-01 -6.67491972e-01 -1.11002102e-01
-2.66869694e-01 2.33731419e-01 4.82640266e-01 3.22453022e-01
-8.88973415e-01 2.96637625e-01 -1.02901411e+00 -4.45752561e-01
7.66580462e-01 3.26639861e-01 2.56890170e-02 -1.75768390e-01
-7.73239434e-01 6.67693555e-01 2.69139111e-01 3.02308686e-02
-1.40929425e+00 -8.62852693e-01 -5.33525825e-01 4.77145575e-02
9.17667329e-01 -9.19054627e-01 1.70771754e+00 -2.27331460e-01
-1.07184243e+00 8.05702865e-01 -4.41426575e-01 -4.82231617e-01
4.85219419e-01 -8.42086017e-01 -3.86704922e-01 6.10954948e-02
4.87511307e-01 7.75741279e-01 9.03945148e-01 -9.01841402e-01
-1.11630487e+00 -1.43106580e-01 9.28538442e-02 -5.29911034e-02
-7.94733092e-02 3.56892467e-01 -1.31753945e+00 -5.22756100e-01
-2.85442948e-01 -1.06469715e+00 -1.02058630e-02 5.42733014e-01
-3.61043394e-01 -5.16674876e-01 1.29756546e+00 -2.31318489e-01
1.30791664e+00 -2.15817118e+00 -5.54261468e-02 -4.79509264e-01
6.63057327e-01 5.74265122e-01 -1.40699863e-01 2.28433192e-01
4.09330547e-01 -1.85927242e-01 3.48575056e-01 -7.22635031e-01
9.76946428e-02 -3.31454165e-02 -2.96126544e-01 4.89580721e-01
1.99717432e-01 1.27455890e+00 -1.07010889e+00 -7.18503654e-01
3.20908964e-01 4.17065650e-01 -4.08941746e-01 2.18445629e-01
-4.28075224e-01 2.77526766e-01 -6.17771626e-01 8.96918058e-01
7.04447508e-01 -7.15990841e-01 -1.63058609e-01 -2.51161635e-01
-2.81553566e-01 4.33562789e-03 -1.15318263e+00 1.59593189e+00
6.81192130e-02 7.65941739e-01 3.86283286e-02 -3.07810128e-01
5.50802588e-01 1.01254761e-01 5.91495335e-01 -7.37993300e-01
2.17015415e-01 -1.47207916e-01 -1.54951781e-01 -3.70501876e-01
7.24855542e-01 3.75538945e-01 1.01129889e-01 2.56509930e-01
1.11351319e-01 4.66184229e-01 3.76493722e-01 4.66620862e-01
1.30657828e+00 1.78474128e-01 1.74582005e-01 -1.21379524e-01
3.21658373e-01 9.46760252e-02 8.24010313e-01 1.15635014e+00
-6.31076396e-01 3.47929955e-01 -2.09111840e-01 -5.26446939e-01
-6.38805568e-01 -1.33414769e+00 1.01642706e-01 1.50771880e+00
6.09589040e-01 -6.79628670e-01 -2.41845429e-01 -6.75250828e-01
3.76801252e-01 2.44952932e-01 -6.23897314e-01 2.85925157e-02
-7.36340702e-01 -3.83368790e-01 4.35503334e-01 7.08112240e-01
3.46320868e-01 -1.03210342e+00 -1.06657755e+00 2.73762107e-01
-2.38830924e-01 -1.26859128e+00 -9.30717170e-01 -2.18508855e-01
-6.08746409e-01 -1.35137558e+00 -8.09479177e-01 -4.97903407e-01
2.37056375e-01 9.42386448e-01 1.30418873e+00 5.62979579e-01
-5.26459992e-01 4.87434775e-01 -3.33299965e-01 -6.05915487e-01
1.15728170e-01 -2.45182719e-02 1.33502513e-01 -2.20366288e-03
4.97115731e-01 -1.38475984e-01 -8.74795973e-01 6.04334354e-01
-4.66920555e-01 1.78902633e-02 5.67757308e-01 5.45813441e-01
7.25740492e-01 -4.35791641e-01 1.93450913e-01 -3.61087680e-01
-4.17398326e-02 -5.68289280e-01 -9.09029603e-01 3.13750118e-01
-2.41325200e-01 -3.24312747e-01 5.14356866e-02 -6.05887771e-01
-6.37514889e-01 6.05848953e-02 1.21348746e-01 -9.67556179e-01
-3.85251530e-02 -1.38079569e-01 1.69236958e-01 -2.37042964e-01
4.07362789e-01 3.95822376e-01 -2.65823841e-01 -7.34917581e-01
3.47149849e-01 2.16114908e-01 7.08745897e-01 -4.50331867e-01
9.51316357e-01 7.34600961e-01 -1.99278921e-01 -4.20868248e-01
-1.15241182e+00 -9.71057534e-01 -4.61913466e-01 -5.22029400e-01
9.48044538e-01 -1.16971397e+00 -1.31616092e+00 4.27325994e-01
-1.20496309e+00 -3.91921133e-01 -2.06085712e-01 4.69988644e-01
-2.57112145e-01 1.65252298e-01 -4.91635591e-01 -7.54533648e-01
-5.35205364e-01 -1.10595572e+00 1.40041864e+00 5.81243992e-01
2.25634113e-01 -4.98494089e-01 9.54940021e-02 2.65043646e-01
5.64825654e-01 -1.04147568e-01 -8.36302191e-02 -5.46288550e-01
-1.40717280e+00 6.50676247e-03 -4.95293796e-01 -3.27595651e-01
-1.25596882e-03 2.73898207e-02 -8.25222671e-01 -7.33377814e-01
-5.19037068e-01 -2.63273388e-01 1.14129674e+00 3.49877536e-01
9.48406398e-01 -1.01977870e-01 -9.46792781e-01 6.14087462e-01
1.42350554e+00 2.32540265e-01 3.47745687e-01 2.89790004e-01
6.82777345e-01 -2.37029657e-01 9.31501746e-01 5.67684829e-01
7.52730906e-01 1.19111574e+00 6.65479541e-01 -5.10068573e-02
-4.84971851e-01 -1.41471297e-01 3.33867252e-01 4.79949743e-01
5.82597964e-02 -2.36548394e-01 -9.79042530e-01 7.80771494e-01
-2.14816380e+00 -1.29877388e+00 -3.57296646e-01 1.94121659e+00
4.78037924e-01 2.37196699e-01 6.23612106e-01 -4.98949021e-01
9.25485969e-01 1.41894042e-01 -7.09048271e-01 6.56245530e-01
2.32440829e-01 -5.14400974e-02 4.96813953e-01 3.60358179e-01
-1.31980550e+00 1.11481094e+00 6.02602005e+00 7.50745893e-01
-8.92521322e-01 6.01381183e-01 -1.67294711e-01 -3.64933223e-01
3.28803241e-01 -2.67961109e-03 -1.55135357e+00 6.23718619e-01
7.00901210e-01 -2.67535806e-01 1.43338665e-01 8.57302248e-01
-5.91365807e-02 -1.36206925e-01 -8.32252026e-01 9.76329982e-01
-4.69428487e-02 -1.57655048e+00 -6.64849207e-02 -2.03698695e-01
3.40072960e-01 5.78591168e-01 1.21203775e-03 5.92764318e-01
4.76214767e-01 -4.69423741e-01 1.00460064e+00 6.84949577e-01
4.12630171e-01 -2.08571300e-01 4.71185535e-01 2.22486198e-01
-2.14893103e+00 -1.68911919e-01 -1.99139699e-01 1.02555163e-01
3.92643452e-01 1.10180877e-01 -6.63111806e-01 5.92830241e-01
1.29824436e+00 9.18770134e-01 -8.33226323e-01 1.71442783e+00
1.26587465e-01 4.90197569e-01 -4.55706179e-01 1.71819907e-02
1.23538733e-01 3.10362548e-01 8.97643268e-01 1.30390227e+00
2.07269102e-01 1.40340507e-01 7.53942251e-01 8.53110015e-01
-1.21463314e-01 -3.16204101e-01 -3.63195688e-01 2.07814053e-01
8.14975917e-01 1.33611012e+00 -7.85006642e-01 -6.37683868e-01
-6.46064579e-01 7.91316092e-01 3.88665348e-01 1.59836367e-01
-1.35424232e+00 -3.02459419e-01 8.66278589e-01 2.05831937e-02
1.08342564e+00 -1.86062098e-01 4.01538193e-01 -1.05932772e+00
5.09825572e-02 -5.15015900e-01 7.64188290e-01 -8.38734269e-01
-1.02139211e+00 5.19278765e-01 -3.41991410e-02 -1.37324905e+00
4.19033676e-01 -4.73958880e-01 -6.88946903e-01 3.98630857e-01
-1.59172487e+00 -1.45425284e+00 -6.92493737e-01 7.42392957e-01
6.74544215e-01 3.97326276e-02 4.23534125e-01 6.99276149e-01
-7.25109637e-01 6.48948610e-01 -5.32467254e-02 2.88814545e-01
6.69058859e-01 -1.10451329e+00 6.57003403e-01 9.82732952e-01
2.96281010e-01 5.26124954e-01 4.92921740e-01 -6.41243756e-01
-1.69964862e+00 -1.33070123e+00 5.07264137e-01 -8.80051315e-01
7.55457938e-01 -2.72014827e-01 -8.08468223e-01 1.03518879e+00
2.47842923e-01 5.09711146e-01 3.77279669e-01 -9.37461257e-02
-2.50509053e-01 -1.89594746e-01 -6.73020840e-01 4.27838743e-01
1.31894839e+00 -3.00811559e-01 -4.89267558e-01 2.71101892e-01
1.08682191e+00 -6.67316794e-01 -8.78114223e-01 3.07355911e-01
6.04970932e-01 -8.08305800e-01 1.29705417e+00 -4.76171434e-01
-4.15162832e-01 -8.15321565e-01 -1.91047594e-01 -5.83613813e-01
-6.68556154e-01 -8.25527549e-01 -8.39072526e-01 1.15846789e+00
1.88387021e-01 -4.33263958e-01 7.12767243e-01 7.02540725e-02
-2.47344345e-01 -7.35030055e-01 -9.68639553e-01 -1.00996602e+00
-4.44961548e-01 -3.57468754e-01 6.57845140e-01 5.71545601e-01
-6.90552533e-01 2.10311174e-01 -5.53649366e-01 4.37036812e-01
1.04408216e+00 4.14362103e-01 1.12660396e+00 -1.24894547e+00
-3.43586743e-01 -3.94063026e-01 -3.78597766e-01 -1.50180459e+00
-2.48814180e-01 -7.37981200e-01 7.18208998e-02 -1.43749988e+00
5.66448867e-01 -4.81606781e-01 -6.60707116e-01 6.35693908e-01
-3.29475641e-01 5.49903095e-01 8.63993227e-01 5.05376995e-01
-1.66023827e+00 6.07862413e-01 1.27633536e+00 -1.56469390e-01
1.06566526e-01 2.35452160e-01 -5.23052275e-01 4.55717564e-01
6.93395019e-01 -9.40653086e-01 1.02824681e-01 -6.64665699e-01
-3.19630623e-01 -1.45040760e-02 8.73731256e-01 -1.33648717e+00
7.49182224e-01 -1.17529877e-01 2.08110124e-01 -1.10179043e+00
6.94521248e-01 -7.85539150e-01 2.34807909e-01 7.47038245e-01
-2.85321544e-03 5.13643503e-01 4.00901318e-01 9.06612813e-01
7.78100938e-02 2.13004708e-01 7.65702724e-01 -3.31704803e-02
-1.27345514e+00 8.20337176e-01 2.11310983e-02 2.51279414e-01
1.32618582e+00 -1.07406108e-02 -6.27682626e-01 1.07187708e-03
-6.37996852e-01 7.40458250e-01 2.36609161e-01 7.80983210e-01
5.86525917e-01 -1.65909564e+00 -7.28930771e-01 -1.29515618e-01
3.38803858e-01 -1.25550434e-01 1.80064529e-01 1.30858278e+00
-7.24757090e-02 5.25518119e-01 -1.25792965e-01 -1.00898612e+00
-1.70431387e+00 7.83008695e-01 2.38390505e-01 -1.04783110e-01
-9.31947768e-01 1.02305257e+00 2.92512357e-01 1.03123397e-01
5.00054479e-01 -1.74999848e-01 1.15567371e-01 -2.64167279e-01
9.27109718e-01 4.40650463e-01 -4.62997049e-01 -6.71569109e-01
-6.93710506e-01 6.67473614e-01 -3.91304404e-01 1.71208784e-01
1.01570630e+00 -1.73363283e-01 2.25233704e-01 3.05033326e-01
7.64338672e-01 -1.91876188e-01 -1.40075767e+00 -7.27903247e-01
9.39236488e-03 -8.98121178e-01 -7.28930831e-02 -6.92696512e-01
-1.16935623e+00 5.76751351e-01 8.47956538e-01 8.71868506e-02
8.59849095e-01 4.92402703e-01 8.76507640e-01 4.96871233e-01
4.96515810e-01 -5.69893956e-01 4.42827135e-01 5.20380020e-01
7.23430037e-01 -1.35350347e+00 -8.25227201e-02 -1.85821310e-01
-3.32768112e-01 7.11550236e-01 1.11414826e+00 7.65463337e-02
6.67825103e-01 3.81641656e-01 -2.29742303e-02 -5.16332626e-01
-9.48769867e-01 -7.56798506e-01 3.24338138e-01 4.01250064e-01
6.74837381e-02 -1.08688973e-01 2.64056832e-01 2.28574753e-01
4.71315593e-01 2.46500626e-01 -1.38513725e-02 1.10915744e+00
-5.84912598e-01 -8.18492711e-01 -4.72359955e-01 3.19631368e-01
-3.94994676e-01 -1.07203098e-02 -7.24820569e-02 9.30898070e-01
2.33233228e-01 7.84914613e-01 1.01429969e-01 -4.04069096e-01
4.51920211e-01 -3.40435296e-01 3.48009497e-01 -5.42118490e-01
-5.38727641e-01 6.56436896e-03 -1.31452546e-01 -9.02187765e-01
-4.97041047e-01 -7.15901077e-01 -1.25107443e+00 -4.40149248e-01
-6.58732474e-01 1.12415934e-02 1.23936646e-01 6.52165830e-01
7.82477379e-01 6.93379521e-01 1.22373313e-01 -1.15153587e+00
-3.48666430e-01 -8.10446382e-01 -8.52004886e-02 3.81584406e-01
7.24732101e-01 -1.18363762e+00 1.13533691e-01 -1.29884362e-01] | [6.3188157081604, -2.0747413635253906] |
54e11af7-0d07-4643-bc72-a133182a9a59 | poster-v2-a-simpler-and-stronger-facial | 2301.12149 | null | https://arxiv.org/abs/2301.12149v2 | https://arxiv.org/pdf/2301.12149v2.pdf | POSTER++: A simpler and stronger facial expression recognition network | Facial expression recognition (FER) plays an important role in a variety of real-world applications such as human-computer interaction. POSTER achieves the state-of-the-art (SOTA) performance in FER by effectively combining facial landmark and image features through two-stream pyramid cross-fusion design. However, the architecture of POSTER is undoubtedly complex. It causes expensive computational costs. In order to relieve the computational pressure of POSTER, in this paper, we propose POSTER++. It improves POSTER in three directions: cross-fusion, two-stream, and multi-scale feature extraction. In cross-fusion, we use window-based cross-attention mechanism replacing vanilla cross-attention mechanism. We remove the image-to-landmark branch in the two-stream design. For multi-scale feature extraction, POSTER++ combines images with landmark's multi-scale features to replace POSTER's pyramid design. Extensive experiments on several standard datasets show that our POSTER++ achieves the SOTA FER performance with the minimum computational cost. For example, POSTER++ reached 92.21% on RAF-DB, 67.49% on AffectNet (7 cls) and 63.77% on AffectNet (8 cls), respectively, using only 8.4G floating point operations (FLOPs) and 43.7M parameters (Param). This demonstrates the effectiveness of our improvements. | ['Aibin Huang', 'Binling Nie', 'Yuanqi Chang', 'Xuesong Yin', 'Rui Xu', 'Jiawei Mao'] | 2023-01-28 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [-4.53410707e-02 -3.48692298e-01 3.82322110e-02 -5.77516258e-01
-7.76329815e-01 1.78536654e-01 1.69382274e-01 -3.92960757e-02
-7.12654471e-01 3.16095203e-01 1.56494360e-02 4.76695597e-03
2.95964807e-01 -6.76511109e-01 -5.32869220e-01 -5.52460015e-01
-1.53218225e-01 -2.99270272e-01 2.43121460e-01 -4.57227260e-01
8.46612155e-02 5.79952598e-01 -1.66536713e+00 6.77653372e-01
5.39017975e-01 1.59819746e+00 -1.61941379e-01 4.18094546e-01
-2.38695055e-01 5.33169210e-01 -6.71645761e-01 -8.15565646e-01
9.56134424e-02 -4.82594455e-03 -3.21484178e-01 -8.72921422e-02
6.18509829e-01 -4.41304207e-01 -2.97496945e-01 8.87757242e-01
9.56898510e-01 -2.97546059e-01 1.87196165e-01 -1.54606986e+00
-2.55890042e-01 1.15990341e-01 -1.45973861e+00 1.95353344e-01
1.64757416e-01 9.91640016e-02 7.18166113e-01 -1.24866354e+00
9.67764929e-02 1.82591200e+00 6.79702640e-01 3.89647067e-01
-9.49242890e-01 -1.25101161e+00 1.74446270e-01 2.82468200e-01
-1.82092583e+00 -7.06590891e-01 4.93105263e-01 -2.05223877e-02
8.17139447e-01 2.91810989e-01 5.63563943e-01 5.78575253e-01
1.49392799e-01 1.13986290e+00 1.00018466e+00 -1.18943557e-01
-9.32365209e-02 -2.36054003e-01 3.32252346e-02 1.02491260e+00
-3.49073000e-02 -3.14033031e-01 -8.98136079e-01 -3.18077981e-01
8.04050982e-01 4.63266708e-02 -3.01254332e-01 4.24709201e-01
-6.22358799e-01 7.89510667e-01 5.13343275e-01 1.15256973e-01
-4.75566417e-01 4.45309907e-01 5.88073194e-01 1.06845014e-01
3.80883157e-01 -2.71833032e-01 -5.12010932e-01 -1.15235016e-01
-8.69420946e-01 3.71132568e-02 2.00020134e-01 8.64522874e-01
7.66249418e-01 2.72903353e-01 -2.20504373e-01 1.05633259e+00
4.64592367e-01 8.09263587e-01 4.39355522e-01 -7.81134129e-01
3.73874277e-01 5.73880255e-01 -3.08310568e-01 -1.24152362e+00
-4.13042754e-01 -2.62527704e-01 -1.05563617e+00 1.01054981e-01
7.33945593e-02 -2.66005397e-01 -8.08247626e-01 1.74939442e+00
3.44477147e-01 3.37238610e-01 -2.86168158e-01 8.12664270e-01
1.03504491e+00 9.04591262e-01 3.32559705e-01 -2.63176084e-01
1.82628369e+00 -9.32316840e-01 -6.26540601e-01 -6.76611159e-03
4.34185266e-01 -9.48594213e-01 9.44915533e-01 5.00971079e-01
-1.18255556e+00 -7.59363890e-01 -9.34756696e-01 -9.22770575e-02
-1.94732129e-04 3.56573880e-01 8.42812002e-01 7.38978565e-01
-1.01689267e+00 3.29875559e-01 -6.03181541e-01 -4.69704531e-02
7.76256979e-01 7.75227129e-01 -4.57261890e-01 8.47703516e-02
-1.11047828e+00 2.21119463e-01 -5.77652939e-02 2.10099608e-01
-2.16531724e-01 -8.22010934e-01 -8.81776571e-01 2.73733705e-01
2.68107384e-01 -3.50537896e-01 1.15399933e+00 -1.18180716e+00
-1.48718214e+00 5.86998224e-01 -5.90832472e-01 -2.48210981e-01
8.77110064e-02 -3.73365194e-01 -6.75584912e-01 3.33158433e-01
-1.99021563e-01 1.16196537e+00 8.97889256e-01 -8.64012599e-01
-7.22851038e-01 -6.34498417e-01 -1.48672357e-01 8.41188878e-02
-4.46118176e-01 3.89989674e-01 -1.11416245e+00 -5.98293841e-01
-7.31308088e-02 -8.55340600e-01 -4.98752519e-02 4.09435511e-01
1.52909793e-02 -3.28888327e-01 1.08419991e+00 -5.19544065e-01
1.53793716e+00 -2.60934806e+00 -6.42565310e-01 8.71023536e-02
2.39834771e-01 6.91242993e-01 -2.81148732e-01 -1.00437485e-01
-6.68708012e-02 1.26377419e-01 7.87708014e-02 -5.92869341e-01
-1.87509075e-01 -8.15853998e-02 8.21866393e-02 4.76470530e-01
5.22445560e-01 8.93345952e-01 -4.06939328e-01 -7.61325955e-01
4.09642123e-02 9.02355373e-01 -9.07049417e-01 -1.36041179e-01
2.99489796e-01 -4.92264219e-02 -4.24640417e-01 9.56561744e-01
1.16864002e+00 -9.52368081e-02 -8.93166959e-02 -6.49008512e-01
-4.90559600e-02 -2.66895175e-01 -1.19709909e+00 1.62294304e+00
-6.33660436e-01 5.01354218e-01 4.27769333e-01 -4.68354017e-01
9.65600193e-01 1.93707392e-01 5.50675809e-01 -9.55751002e-01
5.40793240e-01 1.70383736e-01 -1.15788437e-01 -3.49367470e-01
4.14478093e-01 -1.13493808e-01 -1.58691481e-01 -5.29772006e-02
7.48019069e-02 4.52880055e-01 1.50019079e-01 1.01047993e-01
8.80501390e-01 -1.59509629e-01 2.24966273e-01 -1.87421784e-01
8.20079088e-01 -6.59620047e-01 8.24023485e-01 1.40636444e-01
-3.99237454e-01 4.96037930e-01 6.00850463e-01 -2.91091412e-01
-2.97121823e-01 -7.41322219e-01 -1.51303887e-01 1.20639718e+00
-6.46222290e-03 -7.34618485e-01 -7.59594977e-01 -4.87385035e-01
3.04863136e-02 1.09992005e-01 -4.70687002e-01 -7.61487857e-02
-5.16422153e-01 -8.46855879e-01 8.02291572e-01 5.78777850e-01
1.00342309e+00 -9.18790162e-01 -7.87041724e-01 9.76094976e-02
-1.25207111e-01 -1.06346393e+00 -7.39888906e-01 -3.74004662e-01
-6.14403367e-01 -8.15705299e-01 -7.78491497e-01 -5.82754612e-01
5.92802584e-01 4.33188438e-01 8.54696393e-01 2.04092801e-01
-3.94151568e-01 7.98545498e-03 -1.49755076e-01 -4.01458561e-01
5.14548481e-01 -2.64124900e-01 -1.06598824e-01 5.24966598e-01
5.56265295e-01 -5.37172318e-01 -7.65236914e-01 4.23429519e-01
-8.09823871e-01 -5.27635179e-02 7.76339471e-01 9.44950342e-01
5.69236994e-01 -1.79178938e-01 5.40611207e-01 -4.68464613e-01
4.51376110e-01 -3.83040994e-01 -4.90669399e-01 -4.15881090e-02
-1.39965966e-01 -2.69040138e-01 4.00846928e-01 -3.84757876e-01
-1.02877808e+00 1.49607271e-01 -4.47867215e-01 -6.00465477e-01
6.16338253e-02 2.62257963e-01 -4.60896999e-01 -1.54559270e-01
2.19485432e-01 4.77313325e-02 1.72027394e-01 -3.49839419e-01
1.95376836e-02 1.01306319e+00 4.43059415e-01 -5.93495071e-01
1.30514637e-01 4.31121141e-01 4.91454341e-02 -9.46223319e-01
-5.36962628e-01 -3.16539586e-01 -6.45126626e-02 -1.67521149e-01
6.52420163e-01 -1.37836957e+00 -1.44467425e+00 7.13746011e-01
-1.23714650e+00 1.32878155e-01 2.25920752e-01 3.39228123e-01
-6.50304481e-02 1.34820282e-01 -6.42221332e-01 -9.01848316e-01
-8.47726345e-01 -1.32286024e+00 1.46803784e+00 6.07422173e-01
-1.42040461e-01 -2.23434612e-01 -6.62990034e-01 2.88588703e-01
4.55391526e-01 9.55745876e-02 3.03799897e-01 -5.86498640e-02
-2.66665339e-01 -9.62717924e-03 -8.71249855e-01 3.78663719e-01
-8.60370249e-02 2.22915201e-03 -1.25631106e+00 -1.73725158e-01
-2.33245209e-01 -2.60890454e-01 7.81408846e-01 1.98761448e-01
1.58499908e+00 1.47507992e-02 -1.78349748e-01 7.58756638e-01
1.46592569e+00 4.81142581e-01 8.26430023e-01 -1.28903747e-01
4.27475035e-01 2.99508959e-01 6.37811363e-01 8.38684320e-01
3.57692659e-01 8.09793413e-01 3.29731494e-01 -3.61802846e-01
-2.12007940e-01 -1.51212364e-02 5.66715479e-01 5.33941567e-01
1.43195968e-02 -6.76167607e-02 -6.90446198e-01 3.31164002e-01
-1.67969918e+00 -8.20394635e-01 -2.07408458e-01 1.91321421e+00
7.86672413e-01 2.54645310e-02 1.10693790e-01 2.04817101e-01
6.28583372e-01 1.63214073e-01 -3.38856041e-01 -5.15788078e-01
-1.87944219e-01 6.33458257e-01 3.96912634e-01 3.62824440e-01
-1.00921679e+00 8.94003391e-01 5.23954964e+00 1.42584634e+00
-1.58964825e+00 1.59988910e-01 1.01829684e+00 -3.27590704e-01
3.00973743e-01 -5.29709160e-01 -1.22121823e+00 6.84789717e-01
8.85924459e-01 8.57663825e-02 -4.79401415e-03 7.96265364e-01
1.91989675e-01 -4.29653138e-01 -5.41565418e-01 1.61900425e+00
2.24531725e-01 -1.17248321e+00 8.57984424e-02 1.12532359e-02
1.15247272e-01 -2.19391063e-01 6.78928755e-03 3.67303848e-01
-2.56519824e-01 -8.88257861e-01 5.26985943e-01 2.27910310e-01
1.11319697e+00 -1.17143619e+00 9.15820122e-01 -2.04537168e-01
-1.47878551e+00 6.02610111e-02 -1.75788581e-01 9.18791816e-02
1.92449972e-01 6.73277736e-01 -3.01560551e-01 3.27493638e-01
9.90881741e-01 4.95851964e-01 -3.58670801e-01 7.53241718e-01
5.67487851e-02 4.16301847e-01 -7.73711801e-01 -6.05812073e-02
1.94778159e-01 1.42273188e-01 1.09769702e-01 1.38799763e+00
3.35357696e-01 4.06692773e-01 5.72881438e-02 1.23230845e-01
-2.22181693e-01 2.57378578e-01 -3.98284383e-02 3.07955652e-01
3.16627890e-01 1.38159359e+00 -4.12784666e-01 -3.05755287e-01
-4.16676939e-01 1.04684603e+00 2.07387898e-02 -1.87205262e-02
-1.31105053e+00 -7.08089650e-01 1.00688481e+00 1.63672045e-01
4.51655835e-01 -1.30365148e-01 -1.47190213e-01 -8.20018768e-01
8.29230770e-02 -8.19580853e-01 4.18331712e-01 -8.27956200e-01
-8.23636651e-01 8.21268201e-01 -1.97494715e-01 -8.28929007e-01
4.73149538e-01 -5.41204214e-01 -5.39802670e-01 6.63279414e-01
-1.75365996e+00 -1.15745676e+00 -5.08811176e-01 1.02830219e+00
4.41176623e-01 2.56749652e-02 7.21692324e-01 9.30793941e-01
-8.96507621e-01 1.25267136e+00 -4.14825976e-01 2.50619620e-01
8.51136923e-01 -6.35586202e-01 2.24717140e-01 4.60190505e-01
-1.93343535e-01 6.82065070e-01 1.82107478e-01 -2.96343654e-01
-1.47658336e+00 -9.80620265e-01 8.40345860e-01 4.59285557e-01
4.78033215e-01 -2.47546569e-01 -6.76128626e-01 1.83161110e-01
1.69093251e-01 4.58477050e-01 8.46361697e-01 -1.13782333e-02
-5.07790864e-01 -6.93489850e-01 -1.30016637e+00 7.75163233e-01
1.02826548e+00 -1.16223127e-01 9.39956680e-02 -1.05540499e-01
3.89364690e-01 -2.87987292e-01 -9.10232663e-01 5.30993044e-01
6.93187177e-01 -1.12788701e+00 8.25374305e-01 -1.68961823e-01
3.34517986e-01 -3.95432800e-01 -3.24745804e-01 -8.63327742e-01
-2.38848329e-01 -9.52385545e-01 4.18719724e-02 1.28366983e+00
1.12926513e-01 -6.81589484e-01 7.21555173e-01 4.42044020e-01
-1.21100936e-02 -1.32028806e+00 -1.01362252e+00 -4.69232827e-01
-4.80455130e-01 -5.40621579e-01 8.37632418e-01 5.30538559e-01
-1.47458255e-01 5.35997689e-01 -3.57895553e-01 -1.32358242e-02
2.90477633e-01 -1.05461657e-01 8.78536284e-01 -8.80618751e-01
-8.73772874e-02 -5.08421481e-01 -6.14944696e-01 -1.29347432e+00
-5.37345186e-02 -3.39910507e-01 -3.01957667e-01 -8.94732237e-01
8.45237821e-02 -4.65821594e-01 -4.06747460e-01 9.44912791e-01
-1.35861695e-01 7.01423705e-01 5.67125440e-01 -2.61037558e-01
-6.67530000e-01 8.82852018e-01 1.15917003e+00 6.20356463e-02
-4.37914059e-02 -4.47144836e-01 -8.74781489e-01 9.74704206e-01
8.80851984e-01 -3.18329722e-01 -2.56658614e-01 -5.28489113e-01
-1.91518158e-01 7.79557228e-02 1.51543692e-01 -1.05703759e+00
1.57574072e-01 2.39680707e-03 5.40174425e-01 -5.83795726e-01
9.22086298e-01 -6.74195826e-01 -2.30906188e-01 4.51263428e-01
3.48543674e-01 2.43642613e-01 7.31551170e-01 2.38989159e-01
-5.37379861e-01 4.12775904e-01 1.09996533e+00 2.13387966e-01
-8.03874612e-01 5.59856057e-01 -1.89565733e-01 -2.04468086e-01
1.03358650e+00 -1.50495321e-01 -2.82943308e-01 -3.43887806e-01
-5.42903125e-01 3.13324779e-01 -7.24900886e-02 2.77171701e-01
7.44139552e-01 -1.41272736e+00 -6.98334575e-01 4.21510816e-01
-6.35261163e-02 -1.21470526e-01 6.54794574e-01 1.02835000e+00
-4.36929882e-01 2.55817980e-01 -3.39508116e-01 -5.71144104e-01
-1.82228398e+00 -1.80542078e-02 1.33143574e-01 -1.21002145e-01
-2.57269412e-01 1.20584679e+00 3.67535025e-01 2.58237392e-01
1.54829726e-01 9.02820658e-03 9.80446208e-03 2.24514708e-01
1.03804052e+00 2.97207803e-01 4.65221331e-02 -7.76213884e-01
-5.64009607e-01 7.43988693e-01 -2.40407765e-01 -3.78297977e-02
1.16507471e+00 7.40822852e-02 -2.03732386e-01 -1.49051696e-01
1.55337656e+00 1.53224260e-01 -1.14744329e+00 -1.29216745e-01
-4.88646507e-01 -6.66738987e-01 3.84555429e-01 -6.49949849e-01
-1.66383529e+00 1.02002108e+00 7.85193503e-01 -5.60396791e-01
1.75324130e+00 -4.30962622e-01 1.18969822e+00 4.35432009e-02
4.89962488e-01 -8.24066579e-01 6.96760863e-02 3.23955417e-01
9.79553640e-01 -1.06783831e+00 1.68099001e-01 -7.46782005e-01
-5.99084139e-01 9.21098411e-01 8.67309868e-01 -9.13584158e-02
1.03651273e+00 4.52569753e-01 2.11902503e-02 -8.83199126e-02
-7.78277695e-01 -1.84708044e-01 1.16708502e-01 2.30262816e-01
4.75582689e-01 -1.15583830e-01 -4.33634579e-01 8.71221244e-01
-8.88856128e-03 2.84106553e-01 1.38905481e-01 1.00654101e+00
-3.52381259e-01 -9.46196854e-01 -4.13915902e-01 2.33526051e-01
-7.46966302e-01 -3.24472368e-01 1.76847130e-02 7.19927073e-01
4.75839138e-01 1.00482690e+00 2.22735748e-01 -5.43168068e-01
3.78301799e-01 -1.63240656e-01 4.15057987e-01 -1.52707100e-01
-7.30681956e-01 4.81587857e-01 1.44672364e-01 -8.83132517e-01
-3.74244958e-01 -4.91348803e-01 -1.39213920e+00 -6.63170099e-01
-2.71875352e-01 -1.56922378e-02 7.65704632e-01 5.25142312e-01
1.01867449e+00 4.21982378e-01 5.38767397e-01 -6.55652344e-01
-9.45386291e-02 -9.41763401e-01 -4.33226913e-01 1.46705315e-01
4.03433703e-02 -7.34398484e-01 4.83665392e-02 -2.02729851e-02] | [13.538862228393555, 1.4715653657913208] |
d30b3a8c-0507-4510-b79b-d99220c9f42a | multimodal-interactive-lung-lesion | 2301.09914 | null | https://arxiv.org/abs/2301.09914v1 | https://arxiv.org/pdf/2301.09914v1.pdf | Multimodal Interactive Lung Lesion Segmentation: A Framework for Annotating PET/CT Images based on Physiological and Anatomical Cues | Recently, deep learning enabled the accurate segmentation of various diseases in medical imaging. These performances, however, typically demand large amounts of manual voxel annotations. This tedious process for volumetric data becomes more complex when not all required information is available in a single imaging domain as is the case for PET/CT data. We propose a multimodal interactive segmentation framework that mitigates these issues by combining anatomical and physiological cues from PET/CT data. Our framework utilizes the geodesic distance transform to represent the user annotations and we implement a novel ellipsoid-based user simulation scheme during training. We further propose two annotation interfaces and conduct a user study to estimate their usability. We evaluated our model on the in-domain validation dataset and an unseen PET/CT dataset. We make our code publicly available: https://github.com/verena-hallitschke/pet-ct-annotate. | ['Rainer Stiefelhagen', 'Jens Kleesiek', 'Constantin Seibold', 'Lars Heiliger', 'Moon Kim', 'Zdravko Marinov', 'Philipp Kataliakos', 'Tobias Schlumberger', 'Verena Jasmin Hallitschke'] | 2023-01-24 | null | null | null | null | ['interactive-segmentation', 'user-simulation'] | ['computer-vision', 'natural-language-processing'] | [-5.89661747e-02 2.10968122e-01 1.01507433e-01 -7.63420224e-01
-9.53337967e-01 -7.84649074e-01 -9.27903503e-03 3.55449617e-01
-8.08236241e-01 6.66135430e-01 -1.23749413e-01 -5.88804305e-01
1.98701054e-01 -5.27027428e-01 -4.45011765e-01 -5.10006011e-01
1.86279286e-02 6.93748295e-01 4.39722866e-01 2.21334085e-01
-4.75996174e-02 4.45677370e-01 -7.80127764e-01 4.52043474e-01
9.70466077e-01 9.03199613e-01 6.55455768e-01 6.48367167e-01
1.12336859e-01 2.42066339e-01 -3.74720156e-01 -2.42789507e-01
2.24903718e-01 -2.23854616e-01 -8.99911642e-01 4.38546576e-02
-6.89218566e-02 -5.92951238e-01 -6.30498230e-02 9.28473175e-01
8.95540535e-01 -6.91651106e-02 5.22534013e-01 -1.13189936e+00
-3.13059509e-01 3.56890589e-01 -5.59676945e-01 3.22758198e-01
2.73713052e-01 1.85354516e-01 4.10399884e-01 -6.97821915e-01
6.59258544e-01 5.78624010e-01 6.86804235e-01 5.70561707e-01
-1.12235880e+00 -5.17881870e-01 -2.00949550e-01 -1.75389051e-02
-1.37202597e+00 -1.24995418e-01 3.92960906e-01 -6.84487104e-01
6.27567530e-01 4.43190277e-01 8.07608485e-01 1.10023379e+00
1.78481176e-01 7.83052564e-01 1.09801316e+00 -8.17636028e-02
2.21053109e-01 4.61261183e-01 3.44097540e-02 7.13488519e-01
1.24518529e-01 -2.73621023e-01 2.03639030e-01 -2.08608553e-01
1.18827569e+00 -1.24577843e-01 -5.85951805e-01 -5.02802670e-01
-1.21114337e+00 7.89387882e-01 4.57364529e-01 4.29480612e-01
-3.47045541e-01 7.73934722e-02 5.23647368e-01 -7.66168162e-02
4.03088480e-01 2.23842174e-01 -4.24877048e-01 -1.53271079e-01
-1.03961062e+00 -1.12681828e-01 7.45142221e-01 7.75984645e-01
1.43408567e-01 -3.78513902e-01 -2.38548234e-01 7.41675198e-01
3.72419655e-01 1.25137672e-01 5.52490473e-01 -7.61981189e-01
1.53502792e-01 2.41794854e-01 2.97476232e-01 -5.98975241e-01
-8.82683039e-01 -3.91180038e-01 -6.14622355e-01 -1.54062677e-02
6.03329659e-01 -3.27089816e-01 -9.70372081e-01 1.26622057e+00
5.50207376e-01 -7.41414279e-02 -4.99360949e-01 1.35846937e+00
1.10733485e+00 1.16766259e-01 5.51310420e-01 -6.18954301e-02
1.44625974e+00 -6.79240882e-01 -5.42099893e-01 1.28609538e-01
9.58840370e-01 -6.03993297e-01 1.46748102e+00 3.64805192e-01
-1.08143139e+00 -3.22703481e-01 -8.85846972e-01 -1.31832868e-01
-1.41421571e-01 4.62207139e-01 7.31001198e-01 1.00018525e+00
-1.08839750e+00 5.58977365e-01 -1.33249140e+00 -3.48176539e-01
6.84412181e-01 5.93243420e-01 -2.85978317e-01 1.90324694e-01
-9.16238785e-01 7.75679946e-01 3.92484814e-01 -5.17469682e-02
-8.23788643e-01 -7.79237509e-01 -6.66234314e-01 -2.17354789e-01
3.61918509e-01 -7.81605244e-01 1.58351922e+00 -7.36102879e-01
-1.31192720e+00 1.06033659e+00 9.18486267e-02 -1.32370561e-01
1.05692518e+00 -1.01829404e-02 -1.24466002e-01 3.23691994e-01
-5.53417057e-02 8.22566092e-01 3.35524410e-01 -1.22871017e+00
-6.56652451e-02 -4.16548163e-01 -9.02465917e-03 2.50184268e-01
3.51148881e-02 -1.44803850e-02 -5.57406068e-01 -4.37131852e-01
1.38267562e-01 -9.23027098e-01 -5.48698485e-01 2.70137250e-01
-4.79058295e-01 1.65282190e-01 6.06253922e-01 -7.97053039e-01
1.02264166e+00 -1.85422146e+00 -2.61577100e-01 3.04041713e-01
5.16234517e-01 4.96566929e-02 2.76804537e-01 -1.01037733e-01
-1.72522515e-01 2.43570134e-01 -4.98378694e-01 -2.59385377e-01
-1.00007474e-01 5.79975024e-02 2.96899021e-01 5.18129826e-01
-3.01611811e-01 1.02530873e+00 -7.98348427e-01 -8.25621486e-01
3.21693093e-01 5.48599124e-01 -4.70604867e-01 1.53806224e-01
-2.93896440e-02 1.21979308e+00 -6.52125418e-01 7.23205149e-01
6.75721586e-01 -5.33865929e-01 3.13219607e-01 -4.26140249e-01
2.07121093e-02 -3.23334672e-02 -8.50024164e-01 2.12222314e+00
-3.96753818e-01 2.27910936e-01 1.42006293e-01 -7.97975421e-01
3.33908707e-01 5.92788637e-01 8.58026922e-01 -6.97759867e-01
6.25007391e-01 1.85905293e-01 1.06206052e-01 -8.56309652e-01
3.63954991e-01 -1.39412373e-01 5.98889329e-02 5.37441969e-01
-1.31528765e-01 -4.09409314e-01 -2.80834176e-02 1.52344272e-01
9.76563692e-01 3.12677175e-01 2.62014449e-01 -2.62448817e-01
1.64135888e-01 1.13936178e-01 2.62816221e-01 6.61502242e-01
-4.94868755e-01 9.14066970e-01 3.99157435e-01 -3.51429760e-01
-1.01249051e+00 -9.71829772e-01 -4.46353942e-01 8.23590159e-01
1.15576901e-01 -3.73234421e-01 -1.10861683e+00 -9.01839554e-01
-3.16332042e-01 8.09590876e-01 -5.51248491e-01 3.46048504e-01
-9.67549980e-02 -8.83721113e-01 5.38394749e-01 6.81629658e-01
2.16731116e-01 -8.85338485e-01 -1.03781760e+00 1.78431869e-01
-2.51577526e-01 -1.26732218e+00 -3.45085263e-01 9.16650370e-02
-9.77767527e-01 -1.05022144e+00 -9.17677045e-01 -3.88535738e-01
7.86123455e-01 -3.11937630e-01 9.97179925e-01 6.47597685e-02
-6.92910314e-01 5.21761119e-01 -1.90350071e-01 -2.75820255e-01
-2.52269089e-01 4.75509241e-02 -2.55594671e-01 -4.51373518e-01
1.84299480e-02 -5.91901898e-01 -1.12096107e+00 6.18423879e-01
-8.87304008e-01 4.25733030e-01 3.24595034e-01 5.15262842e-01
9.10231292e-01 -4.90010381e-01 3.13511074e-01 -1.11038005e+00
7.55976915e-01 -5.74951172e-01 -6.35242522e-01 1.62587270e-01
-9.03465003e-02 -3.35471421e-01 1.69106200e-01 -4.72272336e-01
-9.25921977e-01 4.19417322e-01 -5.36749601e-01 -2.64562100e-01
-4.51386988e-01 5.86089134e-01 1.51604563e-01 -2.10861579e-01
7.78464139e-01 -1.43087626e-01 -3.09207916e-01 -3.99220586e-01
1.40414879e-01 6.58674538e-01 3.47178787e-01 -5.47936380e-01
2.45263726e-01 5.21831989e-01 -2.62360573e-01 -5.36634922e-01
-3.46239984e-01 -3.05996656e-01 -9.64836597e-01 -3.74953896e-01
1.12527657e+00 -5.46960175e-01 -8.20236564e-01 -1.65468715e-02
-1.15204585e+00 -7.09115744e-01 -2.47418866e-01 6.66952133e-01
-4.93183285e-01 4.91935432e-01 -6.63979650e-01 -6.23057723e-01
-3.26959759e-01 -1.69362617e+00 1.02263319e+00 1.12737052e-01
-4.75282013e-01 -1.05736411e+00 -4.86596897e-02 4.22767431e-01
4.50290322e-01 5.16387880e-01 7.21225441e-01 -7.75080562e-01
-3.68935376e-01 -2.57202506e-01 -4.22724813e-01 -5.12800179e-02
-3.10024284e-02 -2.80865818e-01 -9.21548486e-01 -9.45622772e-02
1.00196309e-01 -4.10786659e-01 3.10335875e-01 6.26560032e-01
1.82957995e+00 2.50848085e-01 -4.01669621e-01 5.67451954e-01
1.24774373e+00 2.34777823e-01 5.43646812e-01 -3.74657800e-03
8.84072006e-01 3.82136822e-01 5.79788387e-01 6.09506369e-01
5.30185580e-01 6.10922694e-01 3.13871533e-01 -4.79861587e-01
2.83912987e-01 2.04552755e-01 -2.64053226e-01 6.61496043e-01
-3.96708697e-01 -3.68465036e-01 -1.36998403e+00 4.13272619e-01
-1.60861492e+00 -3.36191148e-01 -3.92641425e-01 2.11098170e+00
8.91153395e-01 -1.19833864e-01 1.60262212e-01 -2.22421974e-01
5.39634049e-01 -4.23299968e-01 -5.09490192e-01 -1.93731248e-01
5.38408339e-01 3.55550945e-01 5.97920895e-01 4.25527662e-01
-1.16535246e+00 6.94880784e-01 5.96015310e+00 7.50337064e-01
-1.34953129e+00 7.25233614e-01 9.54117298e-01 -1.00040480e-01
-3.00802916e-01 -3.53517801e-01 -1.00965656e-01 4.27486897e-01
8.89042497e-01 2.02403426e-01 -4.31096256e-02 7.67326891e-01
5.65908968e-01 -6.20743811e-01 -1.28408289e+00 1.09380412e+00
-3.75767827e-01 -9.63147402e-01 -4.18804526e-01 7.75556490e-02
2.89129913e-01 2.72285342e-01 4.90933768e-02 -1.98579524e-02
6.29504174e-02 -1.19592845e+00 4.46795464e-01 4.67588097e-01
1.04673064e+00 -3.23524952e-01 5.80879748e-01 3.81542742e-01
-8.49219382e-01 5.00140429e-01 -9.80124697e-02 5.25537074e-01
3.04189771e-01 4.46945280e-01 -1.20687735e+00 5.87572336e-01
6.26544237e-01 1.63123891e-01 -5.14215052e-01 1.36715817e+00
-2.11242121e-02 4.95050937e-01 -4.54634219e-01 3.02267730e-01
1.65472612e-01 -1.87706664e-01 3.35870653e-01 1.26809025e+00
3.48286361e-01 3.72362673e-01 3.31981122e-01 1.03626072e+00
5.00886068e-02 4.33810562e-01 -2.11982876e-01 1.24593303e-01
5.38920611e-02 1.64499259e+00 -1.34216547e+00 -2.77890861e-01
-3.64507943e-01 1.26468742e+00 4.50015776e-02 2.74436772e-01
-1.34915602e+00 1.04022369e-01 -6.75533265e-02 4.21714246e-01
-1.48269698e-01 -2.81304210e-01 -4.90268081e-01 -1.06151855e+00
-1.55324265e-01 -4.07099932e-01 3.97224963e-01 -1.00420725e+00
-1.00857043e+00 8.28228831e-01 2.81955034e-01 -1.03469777e+00
-4.66237999e-02 -5.79934120e-01 -4.31109846e-01 7.87840843e-01
-1.05621004e+00 -9.76978719e-01 -7.30885565e-01 6.57944977e-01
2.21497506e-01 4.33206797e-01 9.05220985e-01 7.43663788e-01
-4.26795572e-01 5.07230043e-01 -2.45754302e-01 7.09697902e-02
7.08657384e-01 -1.36907983e+00 1.78653955e-01 1.79745033e-01
-2.03024179e-01 3.75824481e-01 5.41971207e-01 -6.90775096e-01
-8.81115735e-01 -7.67091870e-01 1.75370425e-01 -4.63531315e-01
5.30689597e-01 -3.37212443e-01 -9.08128083e-01 7.99436510e-01
1.86248407e-01 3.72935683e-01 8.58005345e-01 -3.14502031e-01
2.92266101e-01 4.35131550e-01 -1.59887934e+00 4.07935619e-01
9.16232467e-01 -2.97267348e-01 -1.79807678e-01 6.19508803e-01
4.17338848e-01 -1.02922201e+00 -1.12827194e+00 3.83986890e-01
5.49872994e-01 -9.03707922e-01 6.99167013e-01 -3.06578040e-01
1.58777013e-01 -2.33996645e-01 9.00948644e-02 -1.23706555e+00
1.88628778e-01 -2.09851161e-01 3.18750799e-01 5.74234366e-01
5.99737167e-01 -6.24212503e-01 8.11244369e-01 1.12393975e+00
-2.61619538e-01 -9.82091725e-01 -8.59058857e-01 -3.28245670e-01
1.15171885e-02 -8.44468117e-01 2.51029134e-01 1.08220017e+00
1.21074654e-01 -1.59705907e-01 6.29233345e-02 1.23948470e-01
3.91611725e-01 -3.21194321e-01 3.77976775e-01 -1.03713262e+00
-4.59951490e-01 -1.75484717e-01 -2.63503522e-01 -6.94226623e-01
-3.17684352e-01 -1.01434457e+00 -1.25181943e-01 -1.67934453e+00
3.50594580e-01 -6.64063215e-01 -1.64181098e-01 5.82192779e-01
1.83064163e-01 5.90210736e-01 1.00225536e-02 1.45044431e-01
-6.05983257e-01 2.97662318e-01 1.63363862e+00 2.20898852e-01
-2.09496826e-01 -2.61313766e-02 -3.37488204e-01 8.79309714e-01
1.02991140e+00 -5.07932305e-01 -4.78798896e-01 -4.33099806e-01
-4.22733091e-02 4.29808736e-01 6.33144796e-01 -9.07497764e-01
1.70098230e-01 1.16862819e-01 6.93221688e-01 -5.81568837e-01
3.84887278e-01 -1.07691669e+00 3.18538994e-01 3.50581378e-01
-1.70535535e-01 -7.21013397e-02 5.21022677e-01 1.78823486e-01
1.38376102e-01 -2.26888716e-01 8.08552921e-01 -3.84677768e-01
-3.41636300e-01 4.89150971e-01 -4.02439684e-01 -1.31530371e-02
1.25565112e+00 -2.11651221e-01 8.93047005e-02 -4.02377248e-01
-1.53987789e+00 2.28172004e-01 3.98260206e-01 -1.55952945e-01
6.00237668e-01 -1.03910482e+00 -4.38120246e-01 -2.73998939e-02
-6.82755858e-02 4.20483537e-02 5.62551796e-01 1.55925465e+00
-8.73274744e-01 1.95575148e-01 -3.02773178e-01 -8.50060761e-01
-1.21301031e+00 1.89974919e-01 7.52100825e-01 -3.03083807e-01
-6.08633041e-01 6.70558929e-01 4.02978480e-01 -4.57521439e-01
2.05008890e-02 -6.25808954e-01 2.87805032e-02 -9.15159658e-02
1.71771958e-01 1.33620098e-01 2.55776227e-01 -4.38457757e-01
-4.07988250e-01 1.63278997e-01 -1.71280280e-01 -4.32123750e-01
1.16830850e+00 -1.71025395e-01 1.33691043e-01 2.37219036e-01
1.06608796e+00 -2.64500767e-01 -1.16865921e+00 7.11576268e-02
2.04943959e-02 -3.54703873e-01 2.86596596e-01 -1.17665231e+00
-1.15272367e+00 8.59151006e-01 9.99078751e-01 1.40753329e-01
1.00240839e+00 1.78098008e-01 7.59201646e-01 -2.11629570e-02
3.60114634e-01 -1.05423617e+00 -2.00491756e-01 1.22639619e-01
8.80001485e-01 -1.45293176e+00 -7.49438256e-03 -6.25422418e-01
-1.07553220e+00 9.76494074e-01 6.84760392e-01 2.47388765e-01
7.20382273e-01 5.57287157e-01 1.45750672e-01 -4.19239521e-01
-1.50380850e-01 -6.86842799e-02 4.28666204e-01 4.74877030e-01
8.64915550e-01 2.01556250e-01 -4.53625917e-01 9.05616701e-01
-9.41896588e-02 3.53185415e-01 3.87128800e-01 1.08393359e+00
-3.35343555e-02 -1.08932650e+00 -3.30384403e-01 4.52304661e-01
-6.67532504e-01 -1.14116728e-01 -1.51479736e-01 9.56662834e-01
1.53616490e-02 4.83079433e-01 -7.57603496e-02 -6.47235811e-02
2.23032176e-01 8.64438564e-02 7.41872787e-01 -8.01489234e-01
-6.56826496e-01 3.66718471e-01 -2.19055694e-02 -7.90432394e-01
-2.89538503e-01 -5.21455348e-01 -1.60940564e+00 -8.27195942e-02
-4.22205962e-02 -4.56397533e-02 9.48252022e-01 6.90631270e-01
2.09083796e-01 6.86893344e-01 1.14332192e-01 -1.00176036e+00
-3.30149718e-02 -9.29288089e-01 -4.93141800e-01 2.18499392e-01
-7.54724517e-02 -6.31435037e-01 1.50795698e-01 3.61053087e-02] | [14.672104835510254, -2.2943010330200195] |
a3dcea54-c619-4300-81ca-fb580e853a36 | motion-guided-3d-pose-estimation-from-videos | 2004.13985 | null | https://arxiv.org/abs/2004.13985v1 | https://arxiv.org/pdf/2004.13985v1.pdf | Motion Guided 3D Pose Estimation from Videos | We propose a new loss function, called motion loss, for the problem of monocular 3D Human pose estimation from 2D pose. In computing motion loss, a simple yet effective representation for keypoint motion, called pairwise motion encoding, is introduced. We design a new graph convolutional network architecture, U-shaped GCN (UGCN). It captures both short-term and long-term motion information to fully leverage the additional supervision from the motion loss. We experiment training UGCN with the motion loss on two large scale benchmarks: Human3.6M and MPI-INF-3DHP. Our model surpasses other state-of-the-art models by a large margin. It also demonstrates strong capacity in producing smooth 3D sequences and recovering keypoint motion. | ['Yuanjun Xiong', 'Jingbo Wang', 'Sijie Yan', 'Dahua Lin'] | 2020-04-29 | null | https://www.ecva.net/papers/eccv_2020/papers_ECCV/html/2054_ECCV_2020_paper.php | https://www.ecva.net/papers/eccv_2020/papers_ECCV/papers/123580749.pdf | eccv-2020-8 | ['monocular-3d-human-pose-estimation'] | ['computer-vision'] | [-3.41620862e-01 -2.88054854e-01 -4.97160465e-01 -1.18576132e-01
-6.23302937e-01 -4.21703279e-01 4.60448921e-01 -5.25174201e-01
-5.28625846e-01 5.50415397e-01 4.96118426e-01 -2.76592728e-02
2.86030889e-01 -3.57862234e-01 -9.38328028e-01 -3.25320214e-01
-4.27160293e-01 2.52389073e-01 2.45698750e-01 -1.94646046e-01
-5.97307049e-02 6.76105738e-01 -1.09049881e+00 -4.07780856e-02
2.20829174e-01 8.22694957e-01 1.09720483e-01 1.09506989e+00
3.82429123e-01 1.00097096e+00 -4.60341930e-01 -1.88675389e-01
4.86663043e-01 -3.00807387e-01 -8.63622487e-01 -9.47821364e-02
8.69188011e-01 -8.32432985e-01 -1.14926374e+00 6.02576494e-01
8.84178400e-01 2.47832820e-01 4.98258382e-01 -1.33150268e+00
-5.28502345e-01 2.96882782e-02 -5.36262393e-01 1.07708745e-01
8.04020643e-01 5.97164154e-01 8.83702099e-01 -9.12785172e-01
1.25146008e+00 1.61469102e+00 9.51243937e-01 7.63552964e-01
-8.70435774e-01 -3.53848577e-01 1.90644711e-01 1.67660475e-01
-1.33437645e+00 -1.18138514e-01 7.68041670e-01 -4.52057034e-01
1.15177619e+00 1.22703925e-01 7.97993004e-01 1.42486310e+00
5.12629807e-01 1.26260912e+00 3.46353471e-01 1.03454448e-01
-2.94471115e-01 -8.59614313e-01 -2.73495615e-01 1.06448901e+00
1.03354953e-01 3.51779193e-01 -6.98609352e-01 -7.73823932e-02
1.18120480e+00 -5.51028959e-02 -5.47338784e-01 -7.57863402e-01
-1.50295806e+00 5.78371644e-01 7.35489011e-01 -1.83589593e-01
-9.79691371e-02 9.71227288e-01 3.06718260e-01 1.76263168e-01
4.86557961e-01 9.17736962e-02 -5.28395772e-01 -3.73187125e-01
-5.96552849e-01 8.61090779e-01 6.45893514e-01 1.19373167e+00
3.79311413e-01 -3.98069248e-02 -3.42614889e-01 3.55155349e-01
4.02093410e-01 7.54089057e-01 1.53066918e-01 -1.27967620e+00
5.72951794e-01 3.56085569e-01 2.24571079e-01 -1.24527538e+00
-6.73715174e-01 -3.54329675e-01 -8.48217249e-01 2.62876123e-01
3.66822571e-01 -1.00771166e-01 -1.19304156e+00 1.96249902e+00
4.47747141e-01 3.17786872e-01 -2.59206802e-01 1.29144323e+00
1.14146197e+00 5.06415963e-01 -1.49522170e-01 5.07103682e-01
8.09295475e-01 -1.30325162e+00 -5.18974245e-01 -2.93244213e-01
7.45732367e-01 -4.63644922e-01 9.30998623e-01 2.12257709e-02
-1.26244855e+00 -7.10560620e-01 -1.06145084e+00 -6.80412650e-01
1.41662285e-01 -4.82548587e-02 7.03339994e-01 1.20270280e-02
-1.10016108e+00 8.67105544e-01 -9.91795599e-01 -2.38431871e-01
1.04077369e-01 2.83267707e-01 -6.96964741e-01 -3.76936555e-01
-1.19430649e+00 6.34508133e-01 -1.03925474e-01 2.06173450e-01
-1.11850822e+00 -7.42048085e-01 -1.27041495e+00 -2.82484680e-01
2.99024940e-01 -1.48403072e+00 1.23068511e+00 -3.86222936e-02
-1.56848121e+00 9.75615680e-01 -9.77762938e-02 -4.81251985e-01
1.09965146e+00 -6.89212084e-01 -4.51284647e-02 3.38488221e-01
1.32183298e-01 1.21093619e+00 7.85765469e-01 -1.01840889e+00
-3.01851571e-01 -3.27701211e-01 1.09399810e-01 4.49362427e-01
4.19064134e-01 -3.37352455e-01 -1.06390893e+00 -9.93571937e-01
6.13605455e-02 -1.24858022e+00 -3.02777141e-01 4.43911374e-01
-5.69507003e-01 -1.10992178e-01 6.91688418e-01 -7.93199241e-01
1.06276679e+00 -1.77714896e+00 6.92144573e-01 -1.02895506e-01
4.24419582e-01 3.52952331e-02 -2.52025187e-01 3.41680586e-01
2.98022807e-01 -1.70666993e-01 -2.21022978e-01 -6.81834280e-01
1.56555042e-01 2.02664316e-01 5.13531119e-02 5.91364324e-01
1.53523073e-01 1.54605937e+00 -9.96139407e-01 -3.42531800e-01
1.71006739e-01 6.57997608e-01 -8.64821136e-01 2.63211280e-01
-1.89448684e-01 6.57686949e-01 -1.96226150e-01 7.30612516e-01
5.54366589e-01 -5.15331805e-01 -1.11095034e-01 -9.61716920e-02
3.33368212e-01 5.81654832e-02 -9.25574243e-01 2.62524867e+00
-3.92958336e-02 6.33382618e-01 -1.24529928e-01 -3.13059747e-01
5.93383908e-01 1.10561013e-01 6.88844204e-01 -4.05673534e-01
5.20917140e-02 -6.49221567e-03 -4.30537283e-01 -3.64502192e-01
7.16186881e-01 3.91008198e-01 -2.45385841e-01 7.75507092e-02
2.15586483e-01 -9.24327075e-02 -2.46185824e-01 2.64437467e-01
1.30262041e+00 7.48781323e-01 1.04753293e-01 -3.82398511e-03
2.78038651e-01 -6.75971806e-02 6.74471319e-01 5.07001936e-01
-4.71329689e-01 1.10296953e+00 4.14679945e-01 -6.85934424e-01
-1.05399382e+00 -1.28200150e+00 5.71366429e-01 7.64281094e-01
4.75413471e-01 -5.53164721e-01 -3.78689498e-01 -8.41640949e-01
5.12932658e-01 -1.33936048e-01 -5.47698438e-01 -2.78456122e-01
-1.11898518e+00 -1.36829898e-01 6.63530529e-01 9.00836706e-01
5.84451079e-01 -7.69538999e-01 -6.19131386e-01 3.30744535e-02
-4.60740179e-01 -1.37517369e+00 -1.22383928e+00 -3.58342677e-01
-8.08218241e-01 -1.08427703e+00 -1.23372197e+00 -6.76313102e-01
2.49272063e-01 4.33897346e-01 1.45559156e+00 -1.01816999e-02
-3.18165421e-01 5.30045927e-01 -2.20896140e-01 -8.90878513e-02
6.14598580e-02 1.84316084e-01 7.99154192e-02 -4.73401159e-01
1.51698261e-01 -4.78687733e-01 -1.08887434e+00 3.86158794e-01
-5.47966361e-01 8.67682248e-02 4.72670943e-01 7.91881800e-01
6.58544660e-01 -6.08426630e-01 5.66712022e-02 -4.21338052e-01
2.26401612e-01 -2.69240677e-01 -2.94822603e-01 -1.23360977e-01
-2.51559336e-02 2.55737782e-01 1.39662921e-01 -3.24137717e-01
-5.82155168e-01 1.80208609e-01 -2.98111260e-01 -9.48719501e-01
4.26573418e-02 2.67673492e-01 -1.34870246e-01 -3.78557831e-01
5.04525423e-01 -1.08298190e-01 1.90502219e-02 -5.45289874e-01
6.27647519e-01 2.23371796e-02 1.11130428e+00 -3.91449571e-01
9.71206069e-01 7.40453064e-01 3.96897912e-01 -6.49912536e-01
-5.20356596e-01 -6.31201029e-01 -7.75747180e-01 -1.38384208e-01
1.24217963e+00 -1.29009831e+00 -8.94030273e-01 8.28353703e-01
-1.48661387e+00 -6.37460232e-01 -1.15856804e-01 6.13713443e-01
-8.62913668e-01 5.35905838e-01 -9.60711002e-01 -4.16460276e-01
-4.66957092e-01 -1.03535962e+00 1.66671693e+00 -2.43647352e-01
-2.43918195e-01 -9.63748395e-01 1.51891366e-01 8.39796364e-02
2.83327214e-02 8.53035331e-01 5.38735449e-01 5.88425808e-02
-9.47146654e-01 -2.68185019e-01 -9.06013772e-02 2.22964168e-01
-1.58006698e-01 -3.56655121e-01 -5.65778792e-01 -6.42072558e-01
-3.05718958e-01 -4.66150194e-01 1.10189331e+00 5.40573120e-01
8.84769261e-01 1.20883435e-02 -3.68195504e-01 1.29060531e+00
1.06836760e+00 -2.49031693e-01 6.52770638e-01 2.32485741e-01
1.51177979e+00 3.01965237e-01 5.02015889e-01 3.17315429e-01
8.16794515e-01 9.07995999e-01 4.16858971e-01 -5.06003648e-02
-4.66921538e-01 -6.71339810e-01 3.90358299e-01 8.88725877e-01
-2.27561802e-01 -1.95371762e-01 -9.57167208e-01 3.72550160e-01
-2.09700799e+00 -7.35701025e-01 6.67837188e-02 1.90888143e+00
4.70954567e-01 2.09339842e-01 3.54648620e-01 -1.29656360e-01
4.03866500e-01 5.59536636e-01 -7.08490729e-01 1.09805435e-01
-1.81313396e-01 -1.58032104e-01 6.84216440e-01 7.31112123e-01
-1.27276683e+00 1.01540673e+00 6.65342999e+00 4.65985000e-01
-9.47185099e-01 -2.40979746e-01 3.13858718e-01 -2.97866672e-01
-2.52393067e-01 -2.91614652e-01 -8.28037739e-01 2.37803206e-01
5.17319798e-01 8.88723060e-02 1.57142729e-01 8.08702409e-01
-4.38638544e-03 1.74517468e-01 -1.27667904e+00 1.36313117e+00
2.04570696e-01 -1.26651812e+00 1.49119258e-01 8.67455080e-02
7.58312225e-01 2.28246540e-01 -1.55102145e-02 8.14115480e-02
3.15249681e-01 -1.23571110e+00 9.08257425e-01 5.42969465e-01
9.47802186e-01 -8.09956968e-01 5.51511109e-01 2.63780892e-01
-1.69362152e+00 2.35702991e-01 -2.01461270e-01 -4.88416478e-02
4.68800008e-01 1.65721521e-01 -3.85461628e-01 8.18054974e-01
5.93889654e-01 1.15611315e+00 -4.07075405e-01 1.02122474e+00
-3.86680067e-01 -4.22865935e-02 -1.76082522e-01 2.54297644e-01
4.54396844e-01 1.68292582e-01 8.53188932e-01 1.04881358e+00
2.22290158e-01 -9.21073928e-02 4.39676732e-01 6.05893791e-01
-2.29849234e-01 -4.71100152e-01 -8.16665947e-01 1.77708700e-01
2.88689464e-01 8.03291917e-01 -2.69144505e-01 -1.03601746e-01
-3.63288790e-01 1.48715580e+00 3.35043848e-01 4.01024789e-01
-8.62275422e-01 -3.03726733e-01 1.11238647e+00 4.51886021e-02
3.05589646e-01 -8.72201324e-01 7.33074844e-02 -1.26482022e+00
2.44359076e-01 -7.01638460e-01 2.99735904e-01 -7.78341174e-01
-1.21463072e+00 4.28372473e-01 1.61857009e-01 -1.30726457e+00
-5.60967624e-01 -8.03204179e-01 -3.10331613e-01 7.37147570e-01
-1.37256861e+00 -1.44212341e+00 -6.46173239e-01 7.60258019e-01
3.35793793e-01 7.50033930e-02 3.99512321e-01 2.77443469e-01
-1.46494374e-01 7.21166193e-01 -4.57841396e-01 3.67742896e-01
8.93995941e-01 -1.41573203e+00 1.40546632e+00 6.61934316e-01
-2.69866474e-02 3.79519045e-01 4.74545985e-01 -7.64369249e-01
-1.75101459e+00 -1.18645084e+00 8.07888389e-01 -9.76852536e-01
3.36387575e-01 -4.98801887e-01 -5.48484087e-01 8.39768112e-01
-1.67740270e-01 3.56445730e-01 1.45181954e-01 -3.97116035e-01
-3.99262667e-01 3.78805101e-01 -9.49613273e-01 7.98875034e-01
2.00690627e+00 -6.84625030e-01 -4.11905646e-01 1.86498731e-01
1.41378665e+00 -9.85898793e-01 -9.06326771e-01 6.01495206e-01
7.13940322e-01 -6.76102936e-01 1.52041185e+00 -7.81584620e-01
5.34223080e-01 -7.98848942e-02 -2.61834979e-01 -1.25489950e+00
-4.97311085e-01 -1.07060635e+00 -7.69689083e-01 3.11819643e-01
-5.03262579e-02 -5.96849024e-02 1.32357728e+00 3.24153930e-01
-1.82729125e-01 -9.54858184e-01 -8.16083133e-01 -1.03278625e+00
1.93334177e-01 -3.41506541e-01 5.70423365e-01 6.86214149e-01
-4.15218711e-01 2.60387540e-01 -9.29581344e-01 -3.85097340e-02
8.33041131e-01 -1.77816316e-01 1.35754490e+00 -9.46681976e-01
-4.65993106e-01 -2.52664953e-01 -8.76379728e-01 -1.89274836e+00
1.99447170e-01 -7.78928041e-01 2.53480040e-02 -1.50040007e+00
-3.15279216e-02 1.96340770e-01 -7.07814172e-02 2.15111747e-02
-4.39151704e-01 2.74044901e-01 4.39121127e-01 1.72922567e-01
-6.43291891e-01 6.90462887e-01 1.83273757e+00 -8.57195482e-02
-1.97249264e-01 -2.13339314e-01 -5.40482737e-02 8.33993018e-01
4.64950264e-01 -1.49469838e-01 -4.39630002e-01 -8.32134604e-01
4.60997149e-02 3.04878145e-01 7.78292298e-01 -1.19631147e+00
3.26258689e-01 -1.01599373e-01 5.46732783e-01 -1.03441703e+00
5.34176826e-01 -4.22620088e-01 1.68453619e-01 5.42426348e-01
-2.36980468e-01 6.03497267e-01 -1.56739652e-02 8.22281361e-01
-1.24302521e-01 6.41378582e-01 4.60493445e-01 -2.87445843e-01
-1.19888496e+00 1.02133250e+00 2.75221229e-01 5.30078471e-01
6.64296389e-01 -8.92785266e-02 -2.68795609e-01 -6.65338755e-01
-6.47523820e-01 4.21792239e-01 6.54202163e-01 7.42116511e-01
9.93840456e-01 -1.96829545e+00 -7.17604041e-01 -3.81950811e-02
2.51146406e-01 2.77191222e-01 2.60388702e-01 7.52955079e-01
-1.11305928e+00 4.84254569e-01 -1.74553901e-01 -8.18106592e-01
-1.11198938e+00 4.30675000e-01 4.05054152e-01 -3.88409823e-01
-1.03866553e+00 1.12492502e+00 3.07598293e-01 -5.83780527e-01
5.22794127e-01 -5.46695709e-01 3.79716158e-01 -6.12892509e-01
4.09181029e-01 5.91531932e-01 -2.20442951e-01 -8.36784303e-01
-5.60339808e-01 8.65259409e-01 3.51136088e-01 -1.83799550e-01
1.03975141e+00 -1.70415163e-01 2.67986357e-01 2.70838380e-01
1.74604750e+00 -2.06823558e-01 -1.78744531e+00 -1.96928456e-01
-1.52755991e-01 -6.87182665e-01 -3.31419080e-01 -4.48344767e-01
-1.01705742e+00 8.10086429e-01 3.89017910e-01 -6.74540341e-01
7.52178729e-01 -1.06723331e-01 1.36067653e+00 5.37449837e-01
6.93998635e-01 -9.00599658e-01 5.58095574e-01 7.87403882e-01
1.04457045e+00 -1.28400719e+00 4.83621210e-02 -4.20582563e-01
-5.99403381e-01 7.78893888e-01 7.32744455e-01 -4.11486179e-01
5.22462547e-01 9.82419550e-02 7.41890892e-02 -2.37210020e-01
-6.22151673e-01 -2.06752568e-01 9.23843503e-01 7.07083344e-01
3.63194674e-01 4.34466563e-02 -1.07321322e-01 2.92722732e-01
-2.39041016e-01 1.25511527e-01 1.88939199e-02 1.19858575e+00
-1.53521314e-01 -8.62120450e-01 -8.06786194e-02 -1.46156773e-01
-3.98172379e-01 9.33181867e-02 -6.30132496e-01 1.09820032e+00
-1.23432532e-01 6.12036407e-01 -1.75214913e-02 -9.10776019e-01
5.67850471e-01 -3.17825556e-01 7.59333491e-01 -3.04822475e-01
-2.70112365e-01 -4.70377840e-02 2.21778035e-01 -1.32913959e+00
-3.47344041e-01 -2.33097911e-01 -1.16825736e+00 -8.20132256e-01
3.30124050e-01 -3.90522659e-01 2.78378755e-01 6.40690088e-01
5.51170468e-01 5.15361786e-01 3.07662278e-01 -1.44880188e+00
-4.77322459e-01 -6.42068267e-01 -4.44325179e-01 8.81816149e-01
6.68836713e-01 -7.14565516e-01 -1.37189284e-01 -2.18176186e-01] | [7.237828731536865, -0.48091065883636475] |
1cb186ed-0081-4c0a-a8cd-8e6731c349c0 | deep-long-short-term-memory-adaptive | 1711.08016 | null | http://arxiv.org/abs/1711.08016v1 | http://arxiv.org/pdf/1711.08016v1.pdf | Deep Long Short-Term Memory Adaptive Beamforming Networks For Multichannel Robust Speech Recognition | Far-field speech recognition in noisy and reverberant conditions remains a
challenging problem despite recent deep learning breakthroughs. This problem is
commonly addressed by acquiring a speech signal from multiple microphones and
performing beamforming over them. In this paper, we propose to use a recurrent
neural network with long short-term memory (LSTM) architecture to adaptively
estimate real-time beamforming filter coefficients to cope with non-stationary
environmental noise and dynamic nature of source and microphones positions
which results in a set of timevarying room impulse responses. The LSTM adaptive
beamformer is jointly trained with a deep LSTM acoustic model to predict senone
labels. Further, we use hidden units in the deep LSTM acoustic model to assist
in predicting the beamforming filter coefficients. The proposed system achieves
7.97% absolute gain over baseline systems with no beamforming on CHiME-3 real
evaluation set. | ['Hakan Erdogan', 'John R. Hershey', 'Zhong Meng', 'Shinji Watanabe'] | 2017-11-21 | null | null | null | null | ['robust-speech-recognition'] | ['speech'] | [ 2.13110119e-01 -5.40562749e-01 8.53974164e-01 -5.93477190e-01
-1.27036572e+00 -4.96645629e-01 1.67359874e-01 -6.46499634e-01
-4.18668866e-01 4.51784223e-01 7.29871154e-01 -4.86622870e-01
1.60435531e-02 -2.66113549e-01 -8.11375916e-01 -9.45260882e-01
-2.85324693e-01 -7.73472413e-02 -2.12300941e-01 -7.27867708e-02
-1.89271986e-01 5.21919727e-01 -1.13300312e+00 5.63206494e-01
1.91310689e-01 1.17344391e+00 7.77800798e-01 1.39159465e+00
2.41393164e-01 8.48844528e-01 -1.00132310e+00 3.31271589e-01
1.18213646e-01 -1.57974482e-01 -3.23922873e-01 -6.14226460e-02
5.20600438e-01 -4.74393576e-01 -5.92059791e-01 8.62508714e-01
1.16211581e+00 7.11453080e-01 2.22281188e-01 -3.25965434e-01
-2.94706017e-01 8.56312156e-01 2.28971262e-02 6.01455748e-01
2.16821074e-01 4.30883467e-02 7.95635819e-01 -1.34323454e+00
-3.30757111e-01 1.19654310e+00 7.12918639e-01 3.94319713e-01
-7.31263697e-01 -7.42200732e-01 2.70563245e-01 4.34118271e-01
-1.20946252e+00 -1.10129905e+00 6.48625195e-01 -1.76362306e-01
1.51765835e+00 3.15798938e-01 1.24372758e-01 1.50881231e+00
3.93228590e-01 4.09184486e-01 7.48333752e-01 -4.74550158e-01
2.05926403e-01 -3.74608934e-01 3.46262529e-02 3.83466095e-01
-7.36993074e-01 4.85327654e-02 -7.15589762e-01 -1.42260985e-02
6.26018405e-01 7.18205236e-03 -5.53481221e-01 6.26237452e-01
-1.31585836e+00 3.57233167e-01 6.98899031e-01 6.35905087e-01
-6.78626895e-01 3.84550095e-01 9.82135013e-02 4.21504289e-01
5.01066267e-01 4.12840277e-01 -7.08970428e-01 2.51984522e-02
-9.55198705e-01 -1.76249743e-01 6.20101035e-01 5.32674193e-01
5.87954074e-02 8.88296008e-01 -1.54479235e-01 1.37760556e+00
7.83262312e-01 8.26893032e-01 8.59085143e-01 -8.09971392e-01
6.30125105e-01 -6.38958275e-01 1.77896529e-01 -8.24386537e-01
-4.38937724e-01 -1.30302548e+00 -9.51266468e-01 5.43024018e-02
4.47173417e-02 -6.23083055e-01 -1.22708464e+00 1.65784287e+00
-4.88044471e-02 6.06861949e-01 1.62181348e-01 9.56176937e-01
6.43302977e-01 1.10843420e+00 -3.28175008e-01 -1.87993422e-01
7.49979317e-01 -1.24535632e+00 -1.04869270e+00 -6.89237475e-01
1.08202927e-01 -1.05990601e+00 7.67163277e-01 6.93036079e-01
-1.12746644e+00 -8.64676833e-01 -1.12801886e+00 1.61883757e-01
-1.66094303e-01 1.16722628e-01 -8.93283784e-02 6.78873837e-01
-1.32578576e+00 1.99461937e-01 -1.00171661e+00 1.53556392e-01
-1.64587036e-01 6.04103148e-01 -4.11140136e-02 -2.65953802e-02
-1.28323650e+00 5.57917595e-01 -6.38167635e-02 1.03392041e+00
-1.28557193e+00 -4.37314332e-01 -7.21881866e-01 3.34498525e-01
2.22206973e-02 -5.29830217e-01 1.82073331e+00 -8.97528052e-01
-2.02799988e+00 -1.52704790e-01 -4.94241774e-01 -5.83736479e-01
-1.32756783e-02 -5.84852755e-01 -9.34885502e-01 -4.28445697e-01
-4.60579753e-01 3.26609351e-02 1.01254809e+00 -1.11373568e+00
-5.19953907e-01 -1.31520391e-01 -4.82206732e-01 3.23663950e-01
-5.38954616e-01 3.03479228e-02 6.58197924e-02 -8.62149239e-01
5.52094758e-01 -6.36455536e-01 -4.31924701e-01 -7.22522557e-01
-4.80873853e-01 1.27294183e-01 5.52619100e-01 -1.12376392e+00
1.25505209e+00 -2.13290501e+00 -6.41147643e-02 1.62121102e-01
-1.94137022e-01 4.35656905e-01 -2.69192696e-01 2.30981529e-01
-1.46851495e-01 -3.67299050e-01 1.93712384e-01 -7.51869619e-01
-4.71093953e-02 6.57832697e-02 -7.67003417e-01 5.07715642e-01
-3.34118336e-01 4.04041827e-01 -6.90239549e-01 2.45756641e-01
4.34442312e-01 9.23722625e-01 -3.07796270e-01 6.11283898e-01
7.20540360e-02 6.72104657e-01 -4.30097729e-02 3.73270363e-01
5.67894995e-01 2.20797688e-01 1.08898068e-02 -1.47221044e-01
-1.13873281e-01 6.38033509e-01 -1.24656081e+00 1.91740215e+00
-9.61131096e-01 8.94058466e-01 6.80326641e-01 -7.57521451e-01
9.27943528e-01 9.23869550e-01 -4.01048660e-02 -7.25227475e-01
2.30828166e-01 4.07656342e-01 3.13444376e-01 -5.60662389e-01
2.04684660e-01 -2.19094262e-01 3.55935365e-01 1.39405996e-01
1.91228405e-01 3.50192994e-01 -6.92535818e-01 -5.84112644e-01
1.26905608e+00 -3.04292530e-01 -2.52168745e-01 1.49203986e-01
6.27621889e-01 -9.78858888e-01 4.35743988e-01 8.41197014e-01
8.52859244e-02 8.84810209e-01 -6.85191095e-01 -2.65326142e-01
-5.13473451e-01 -1.12268066e+00 2.47970536e-01 1.76509631e+00
-7.12359250e-01 4.96254712e-02 -6.34741187e-01 4.99677286e-02
-5.53020298e-01 9.80334997e-01 -3.51059526e-01 7.31509477e-02
-1.10483611e+00 -4.91113335e-01 4.25855637e-01 6.75168335e-01
3.45090181e-01 -9.18076336e-01 -3.42038810e-01 6.93591952e-01
-2.81938642e-01 -1.01390994e+00 -6.44982100e-01 6.12846851e-01
-5.08056641e-01 -1.80911392e-01 -8.95344734e-01 -9.95354533e-01
2.55320162e-01 4.41476047e-01 8.82161021e-01 -4.07134742e-01
1.83920175e-01 2.20529452e-01 -1.78640455e-01 -4.78601068e-01
-2.15087146e-01 4.45510335e-02 3.09600234e-01 3.83230329e-01
-2.91572005e-01 -8.48112464e-01 -7.43221700e-01 4.31213558e-01
-5.10554314e-01 -3.09191167e-01 4.43210095e-01 9.41898346e-01
3.75144035e-01 1.60593063e-01 5.78224957e-01 -2.32354105e-01
6.38922334e-01 -4.63473499e-01 -3.97732288e-01 1.84745029e-01
1.01572916e-01 -1.65720537e-01 8.63216579e-01 -4.74321365e-01
-1.62802792e+00 -4.83370312e-02 -7.78062224e-01 -5.14732361e-01
-2.31816322e-01 6.35220468e-01 -4.39312845e-01 1.78339601e-01
5.44123292e-01 1.34236708e-01 -6.89617455e-01 -8.75458837e-01
1.56556368e-01 1.12756717e+00 6.99274600e-01 -1.14055343e-01
4.37506586e-01 2.83153746e-02 -3.57555062e-01 -9.34958339e-01
-9.01626825e-01 -4.70780581e-01 -4.54016626e-01 -2.10258305e-01
6.73271835e-01 -1.07632065e+00 -5.07838726e-01 7.76565611e-01
-1.26451325e+00 -6.91460848e-01 2.91747034e-01 9.90159214e-01
-1.27494425e-01 -2.95350939e-01 -7.84206212e-01 -1.22672999e+00
-5.48346579e-01 -1.25777757e+00 7.89513648e-01 -8.57932195e-02
6.04697317e-02 -1.21658182e+00 1.12403758e-01 4.02849913e-01
9.66901839e-01 -1.65256053e-01 2.82533824e-01 -5.08146703e-01
-4.11316693e-01 -2.99799085e-01 5.16043842e-01 7.93946683e-01
2.59341091e-01 -4.49736238e-01 -1.62633300e+00 -5.01252294e-01
6.38835073e-01 3.33293714e-02 8.49681199e-01 1.04740274e+00
1.17803836e+00 -4.62681234e-01 2.89554019e-02 8.58108521e-01
9.52798665e-01 6.00400746e-01 3.05145711e-01 1.71005595e-02
8.29444587e-01 3.59595776e-01 9.21277180e-02 2.82500237e-01
7.11834878e-02 4.22831237e-01 3.21323603e-01 -1.80908337e-01
-3.06001008e-01 -3.25946882e-02 6.98180735e-01 1.57685030e+00
3.74051422e-01 -8.80700409e-01 -9.98434722e-01 4.82641131e-01
-1.48375189e+00 -9.73632634e-01 8.74134451e-02 2.09877920e+00
4.77623194e-01 9.81725529e-02 -4.66609091e-01 2.82083571e-01
5.22463202e-01 3.58039320e-01 -5.00882506e-01 -4.28523391e-01
-1.10421717e-01 4.47869748e-01 4.60393667e-01 1.08081782e+00
-9.34316814e-01 4.88637000e-01 6.37877560e+00 6.07219398e-01
-1.68135250e+00 4.50379103e-01 7.47157872e-01 -4.57271367e-01
-1.15508661e-01 -6.81032002e-01 -4.57242608e-01 1.11027092e-01
1.44785500e+00 5.19133508e-01 6.92667425e-01 4.34658557e-01
6.21961772e-01 3.96805942e-01 -1.05789232e+00 9.66920018e-01
1.52476206e-01 -8.05446446e-01 -5.41812181e-01 -1.58862904e-01
6.21158957e-01 6.83956444e-01 6.51822388e-01 2.67733991e-01
3.30594212e-01 -1.49448836e+00 8.15450728e-01 6.28151178e-01
5.61958015e-01 -4.63116020e-01 5.92107177e-01 4.53141183e-01
-1.17476368e+00 -5.31563938e-01 -2.37564802e-01 -3.39480639e-01
3.78881454e-01 5.50337374e-01 -1.36948133e+00 1.16738848e-01
8.10106277e-01 3.15822244e-01 2.51015164e-02 1.18372655e+00
-1.08684041e-01 1.36472094e+00 -3.38032275e-01 2.20908239e-01
2.80939013e-01 2.38915667e-01 6.34187877e-01 1.34861171e+00
7.47243464e-01 1.11597113e-01 2.16349158e-02 3.09312224e-01
-2.50105619e-01 -3.79793435e-01 -6.11057281e-02 2.42303506e-01
4.84706730e-01 1.01740956e+00 -3.37959796e-01 -3.65758650e-02
-2.21138090e-01 8.76335979e-01 5.79275303e-02 1.11685419e+00
-6.88634396e-01 -1.97773650e-01 6.20951116e-01 -1.84412837e-01
4.26705152e-01 -6.81846499e-01 -1.37927875e-01 -8.58097970e-01
-9.21663363e-03 -7.76307821e-01 4.39146124e-02 -1.09582937e+00
-1.15700662e+00 1.18340921e+00 -4.45628822e-01 -8.87485266e-01
-5.10051966e-01 -6.57331645e-01 -7.93990552e-01 1.37635100e+00
-1.41033757e+00 -1.07751477e+00 -6.32043276e-03 6.60326362e-01
1.20768178e+00 -2.16755867e-01 9.85148549e-01 6.13149822e-01
-5.83661079e-01 5.15795887e-01 5.33199370e-01 1.53483957e-01
8.04241776e-01 -1.28630459e+00 7.14861333e-01 1.13012552e+00
3.37435514e-01 8.24100018e-01 9.95952308e-01 -1.50222301e-01
-1.31884265e+00 -1.21336257e+00 7.86199450e-01 -1.72873050e-01
5.52915454e-01 -7.25764394e-01 -9.40713942e-01 8.21340919e-01
5.65415800e-01 2.22136974e-01 7.85766542e-01 3.84007357e-02
-1.38281539e-01 -4.48247194e-01 -7.10848212e-01 2.33612061e-01
7.77621090e-01 -6.18438900e-01 -4.71407741e-01 4.65523511e-01
7.60971010e-01 -7.74905860e-01 -4.24275935e-01 1.38099045e-01
4.19811457e-01 -7.42712080e-01 1.15880966e+00 -1.49895355e-01
-3.95255208e-01 -2.11215734e-01 -7.46766329e-01 -1.65355921e+00
-4.76010054e-01 -9.47239220e-01 -1.35670409e-01 1.06108499e+00
6.17514014e-01 -7.35426724e-01 4.20041591e-01 2.99114019e-01
-7.53076851e-01 -5.15679002e-01 -1.24244642e+00 -3.80335867e-01
-5.02575673e-02 -7.47798920e-01 5.21715462e-01 4.06651169e-01
-6.67339921e-01 4.48522598e-01 -6.99253559e-01 7.27637649e-01
3.47191721e-01 -3.81479263e-01 2.76830167e-01 -6.30134881e-01
-6.48990750e-01 -1.55827567e-01 2.99035370e-01 -1.57080877e+00
2.42799640e-01 -3.96318972e-01 7.99956501e-01 -1.70110941e+00
-7.36233950e-01 -1.33163631e-01 -9.04743195e-01 2.42113993e-01
-7.05820769e-02 -5.64517230e-02 -7.68107176e-02 -2.65621871e-01
-4.48713958e-01 6.44905150e-01 7.22409308e-01 -1.32125050e-01
-2.82292694e-01 5.99337161e-01 -1.88922271e-01 7.22998500e-01
6.59279168e-01 -3.19457442e-01 -4.27692443e-01 -1.28372836e+00
-2.78079472e-02 4.23314720e-01 2.45998189e-01 -1.36118770e+00
6.52153075e-01 2.52969712e-01 6.71064794e-01 -5.13918221e-01
8.11010718e-01 -8.54500830e-01 1.25304461e-01 2.87108004e-01
-6.68424487e-01 1.52855635e-01 2.24821433e-01 4.95660573e-01
-3.11189115e-01 1.07745133e-01 4.51493740e-01 -1.02452308e-01
-3.67862552e-01 8.23099315e-02 -8.46868396e-01 -5.60809135e-01
3.42913717e-02 1.74919456e-01 9.56567079e-02 -7.11820543e-01
-9.83403742e-01 -3.57403941e-02 -5.86998284e-01 3.96020979e-01
7.28894353e-01 -1.21003020e+00 -8.31267774e-01 4.54661965e-01
-6.76493287e-01 -3.94249558e-02 5.54499328e-01 4.43211049e-01
-1.45260856e-01 5.53016961e-01 3.49154443e-01 -4.94634718e-01
-1.23897707e+00 1.18901849e-01 9.64230895e-01 6.27306998e-02
-1.74336642e-01 1.80539012e+00 1.02366105e-01 -4.19422865e-01
8.04973423e-01 -7.12574959e-01 -1.65885359e-01 -3.10965657e-01
7.20135689e-01 4.25036401e-01 6.44776523e-01 -7.11852372e-01
-3.55399281e-01 2.40288213e-01 2.20552370e-01 -6.73193693e-01
1.43749332e+00 -4.96539950e-01 1.21641234e-01 7.71772385e-01
1.38079524e+00 3.42550129e-01 -1.22355890e+00 -5.76173961e-01
-2.05781594e-01 -2.27591291e-01 7.85360992e-01 -1.24725366e+00
-1.02744067e+00 1.26952267e+00 1.02930307e+00 -2.32069939e-02
1.38766789e+00 -5.44575393e-01 1.06004739e+00 6.97012484e-01
6.26151487e-02 -8.00411224e-01 1.53909414e-03 8.86402965e-01
1.12939692e+00 -9.81182337e-01 -6.90624714e-01 3.86207938e-01
-7.60768652e-02 1.30489957e+00 3.62893403e-01 1.06339544e-01
8.53422701e-01 5.47059834e-01 6.60398722e-01 2.20425978e-01
-8.29946697e-01 2.39267841e-01 5.69877505e-01 4.57280636e-01
6.71418130e-01 5.18362112e-02 8.11960280e-01 7.18837440e-01
-5.43695927e-01 -4.53137636e-01 1.08435079e-01 6.56579792e-01
-6.80101752e-01 -8.79480660e-01 -9.22735929e-01 1.40124917e-01
-6.04573429e-01 -4.13182169e-01 6.26541749e-02 -3.78842860e-01
4.10462096e-02 1.62520957e+00 -4.75880364e-03 -6.17795646e-01
4.72210944e-01 1.47077993e-01 1.86922863e-01 -5.70178568e-01
-9.59928870e-01 8.84086907e-01 -5.97628132e-02 -2.77227491e-01
-1.22140855e-01 -4.78393167e-01 -1.07573712e+00 1.68557614e-01
-3.71820331e-01 1.12706646e-01 9.27272618e-01 1.05849051e+00
2.60393411e-01 1.27647281e+00 8.26746404e-01 -1.22108507e+00
-7.97196150e-01 -1.31341326e+00 -3.78926218e-01 -2.94536591e-01
1.16635370e+00 -1.16948001e-01 -4.11936998e-01 9.25864875e-02] | [14.944487571716309, 5.979240417480469] |
eac10370-3985-4215-a6ae-6174e8c8f684 | joint-vertebrae-identification-and | 1812.03500 | null | http://arxiv.org/abs/1812.03500v1 | http://arxiv.org/pdf/1812.03500v1.pdf | Joint Vertebrae Identification and Localization in Spinal CT Images by Combining Short- and Long-Range Contextual Information | Automatic vertebrae identification and localization from arbitrary CT images
is challenging. Vertebrae usually share similar morphological appearance.
Because of pathology and the arbitrary field-of-view of CT scans, one can
hardly rely on the existence of some anchor vertebrae or parametric methods to
model the appearance and shape. To solve the problem, we argue that one should
make use of the short-range contextual information, such as the presence of
some nearby organs (if any), to roughly estimate the target vertebrae; due to
the unique anatomic structure of the spine column, vertebrae have fixed
sequential order which provides the important long-range contextual information
to further calibrate the results.
We propose a robust and efficient vertebrae identification and localization
system that can inherently learn to incorporate both the short-range and
long-range contextual information in a supervised manner. To this end, we
develop a multi-task 3D fully convolutional neural network (3D FCN) to
effectively extract the short-range contextual information around the target
vertebrae. For the long-range contextual information, we propose a multi-task
bidirectional recurrent neural network (Bi-RNN) to encode the spatial and
contextual information among the vertebrae of the visible spine column. We
demonstrate the effectiveness of the proposed approach on a challenging dataset
and the experimental results show that our approach outperforms the
state-of-the-art methods by a significant margin. | ['Jiebo Luo', 'Haofu Liao', 'Addisu Mesfin'] | 2018-12-09 | null | null | null | null | ['joint-vertebrae-identification-and'] | ['medical'] | [-2.28993759e-01 -2.08603665e-01 -2.76193678e-01 -3.10603887e-01
-1.03698564e+00 -2.92038053e-01 1.59229547e-01 1.01417992e-02
-4.27556425e-01 2.77841926e-01 2.31271490e-01 -1.12763099e-01
-3.52537990e-01 -5.74346483e-01 -7.90273666e-01 -7.07225382e-01
9.56947580e-02 6.48723543e-01 5.40171504e-01 -9.25556049e-02
-4.79112659e-03 7.84266055e-01 -1.40189946e+00 7.15462118e-02
5.49530983e-01 9.39771891e-01 4.59079683e-01 3.70091498e-01
-1.55138880e-01 8.53124321e-01 -4.39002752e-01 1.08097091e-01
9.20043066e-02 -3.32175285e-01 -8.35784912e-01 7.45205060e-02
3.26013029e-01 -4.09683853e-01 -3.90705854e-01 8.44756007e-01
6.77612424e-01 5.84352687e-02 7.94415653e-01 -6.31316841e-01
-5.21242201e-01 5.45772552e-01 -1.00548875e+00 2.96116561e-01
-1.88564524e-01 -1.08106568e-01 7.40086257e-01 -1.01368773e+00
4.52193886e-01 9.78227854e-01 7.09548235e-01 4.79287893e-01
-6.47040308e-01 -7.54722059e-01 8.67937133e-02 3.73353548e-02
-1.55176580e+00 7.91422650e-02 9.41722512e-01 -3.03329796e-01
-8.58229920e-02 -5.87056726e-02 7.30365276e-01 9.86761034e-01
3.48356843e-01 8.12447548e-01 1.22221029e+00 -2.40841717e-01
-7.71403313e-02 -4.64368492e-01 1.05651906e-02 1.24093509e+00
-3.54041718e-02 -1.20879315e-01 -1.36904791e-01 2.36120671e-01
1.34994125e+00 4.21216160e-01 -6.41705915e-02 -4.90093738e-01
-1.09555805e+00 6.00801647e-01 9.41713512e-01 4.74621147e-01
-3.10945392e-01 3.49580020e-01 3.54557186e-01 -2.77633399e-01
3.32935974e-02 -6.71782941e-02 -2.98867345e-01 2.80258328e-01
-8.56311977e-01 -9.11720693e-02 2.09713057e-01 5.47091424e-01
5.26573300e-01 1.40245417e-02 -1.13475114e-01 1.13942993e+00
3.97316128e-01 5.73600471e-01 9.19233084e-01 -8.28990459e-01
2.79052615e-01 5.44258952e-01 -3.18187863e-01 -7.31166422e-01
-7.31285274e-01 -6.19393885e-01 -8.25661957e-01 1.61969662e-01
5.72374821e-01 1.53658211e-01 -1.28593171e+00 1.67597497e+00
7.44725168e-01 2.84422606e-01 -2.95454800e-01 1.12755072e+00
9.94912565e-01 3.74161303e-01 -2.08363086e-01 4.16660346e-02
1.44372618e+00 -1.07176757e+00 -4.22445744e-01 -3.40336859e-02
3.89704466e-01 -6.12976134e-01 1.22390640e+00 -6.06824271e-02
-9.01312947e-01 -6.11367762e-01 -9.39512849e-01 -1.76327467e-01
-1.59073323e-01 4.64247406e-01 3.66427153e-01 3.07668969e-02
-5.98885715e-01 2.93831885e-01 -1.34307194e+00 1.21712856e-01
2.58604914e-01 3.80013347e-01 -4.13928926e-01 9.80314314e-02
-1.08413863e+00 7.59463727e-01 3.24831903e-02 4.87529993e-01
-9.31037962e-01 -3.87288392e-01 -8.41673076e-01 -1.61370769e-01
4.90407586e-01 -6.84216499e-01 1.12417364e+00 -5.95281661e-01
-1.56432867e+00 8.85337234e-01 1.14190998e-02 -1.66582018e-01
4.70695257e-01 -2.56962240e-01 -4.24029119e-02 5.29701233e-01
2.94522673e-01 4.50416714e-01 6.77147269e-01 -1.34607768e+00
-5.46796322e-01 -7.15808332e-01 -8.26633945e-02 3.33422780e-01
-1.07467040e-01 -9.93908122e-02 -9.18045938e-01 -8.85425985e-01
5.37947774e-01 -1.17764831e+00 -4.33609456e-01 3.57526630e-01
-4.62695479e-01 -1.86032936e-01 7.81629860e-01 -6.54151797e-01
1.02388012e+00 -1.93613684e+00 1.78520635e-01 2.68647313e-01
2.69873798e-01 3.22904885e-02 2.04912484e-01 -1.13989972e-01
-3.61003466e-02 -2.00556666e-01 -3.22053760e-01 -2.34939843e-01
-4.61951137e-01 5.69608390e-01 -1.71546489e-02 7.13929117e-01
-3.15395236e-01 9.25515056e-01 -8.22643578e-01 -1.07784438e+00
3.16913545e-01 6.84150398e-01 -1.39731586e-01 1.84644759e-01
1.95984155e-01 6.89752638e-01 -9.29598987e-01 8.62951756e-01
2.69552469e-01 -5.33366323e-01 -1.97552532e-01 -3.56196761e-01
1.31314039e-01 -8.74524117e-02 -1.19525564e+00 2.12395358e+00
-7.26237178e-01 3.59646231e-02 1.94699243e-01 -9.48890746e-01
6.58354938e-01 3.20206106e-01 1.00160253e+00 -5.27390361e-01
2.83846200e-01 5.18235147e-01 1.05549343e-01 -4.32045758e-01
-1.64236724e-01 -1.31804198e-01 -1.73624009e-01 5.84475100e-01
2.43243258e-02 -1.82959959e-01 -2.76154339e-01 6.54824674e-02
7.53968418e-01 8.90300646e-02 2.06511796e-01 -2.05448687e-01
6.68340027e-01 -5.28222203e-01 6.09377623e-01 4.73102182e-01
-2.07871154e-01 1.03829348e+00 -8.60446095e-02 -5.21371961e-01
-8.00272763e-01 -1.32423806e+00 -3.17288011e-01 7.35601187e-01
4.00191516e-01 -9.25187767e-02 -7.60932982e-01 -8.83799791e-01
-1.04433775e-01 -1.01124331e-01 -9.55363095e-01 -1.85074642e-01
-1.05755508e+00 -4.54386622e-01 4.92445856e-01 8.50089729e-01
3.82740766e-01 -9.10602927e-01 -7.96645820e-01 1.33027017e-01
-1.71590313e-01 -1.04074383e+00 -6.67087615e-01 4.13737893e-01
-9.67448413e-01 -9.69819963e-01 -1.00745499e+00 -8.40306163e-01
8.32602561e-01 1.02470696e-01 6.88023746e-01 3.11241001e-01
-3.12666863e-01 1.69856563e-01 -1.68255702e-01 -1.66208073e-01
-4.86590378e-02 1.05337642e-01 -9.14138034e-02 -9.60329548e-02
-3.15717161e-01 -5.99122167e-01 -8.48890841e-01 5.23845017e-01
-8.94173443e-01 1.03274010e-01 8.70193601e-01 1.05174565e+00
1.27707136e+00 1.12561751e-02 4.03826147e-01 -9.29881871e-01
2.63561308e-01 -4.60059553e-01 -3.75498861e-01 3.92625004e-01
-3.49433631e-01 2.03590155e-01 6.17499113e-01 -5.90887129e-01
-7.08674610e-01 3.17811519e-01 -4.81685638e-01 -7.67350495e-01
-1.25952866e-02 5.30601859e-01 -5.66703230e-02 -1.65314466e-01
2.83678144e-01 2.58401304e-01 -3.37237231e-02 -5.84361255e-01
4.14916277e-01 5.23754716e-01 1.09729671e+00 -9.03191507e-01
6.51212394e-01 7.55058944e-01 3.29433411e-01 -3.53543371e-01
-1.27956069e+00 -5.33307612e-01 -1.00453293e+00 -4.48510110e-01
9.23050821e-01 -7.62190878e-01 -4.08564806e-01 3.68925154e-01
-7.68209994e-01 -2.32952297e-01 -2.90090650e-01 6.27473116e-01
-5.21189272e-01 4.85719830e-01 -9.92219746e-01 -3.84112835e-01
-5.54557800e-01 -1.65740705e+00 1.43834233e+00 2.66170472e-01
1.84779078e-01 -8.05698276e-01 -2.86031030e-02 4.67281252e-01
2.44770795e-02 4.53589559e-01 8.63722920e-01 -5.12229025e-01
-3.92367542e-01 -9.45545807e-02 -7.43513107e-02 8.68003666e-02
3.83115143e-01 6.35822192e-02 -5.05638301e-01 -1.16580382e-01
1.15321122e-01 -4.43400949e-01 6.93068087e-01 6.74300611e-01
1.73268211e+00 1.00577332e-01 -3.44574630e-01 8.74611199e-01
1.16775703e+00 -1.72880784e-01 9.74977836e-02 2.15425938e-01
1.17010224e+00 5.44179797e-01 5.67219794e-01 2.53420681e-01
6.42745912e-01 7.36415982e-01 5.72677791e-01 -3.04546058e-01
-4.93585378e-01 -4.27529544e-01 -9.04010385e-02 1.30634201e+00
-9.45570469e-02 1.75009847e-01 -1.04172504e+00 5.02657712e-01
-1.61412907e+00 -4.74574298e-01 -1.24349408e-01 2.06509423e+00
9.54115987e-01 1.00220159e-01 -2.46164408e-02 -2.96745263e-03
7.21274257e-01 5.14209569e-02 -6.96286321e-01 -5.43219075e-02
1.50141656e-01 2.20304146e-01 5.95602632e-01 2.71025985e-01
-1.03541768e+00 7.29669750e-01 6.12724161e+00 1.14034784e+00
-1.56236088e+00 1.35193557e-01 5.08721113e-01 -5.21695949e-02
-7.30714500e-02 -4.04828668e-01 -6.65546298e-01 7.27837086e-01
6.15525961e-01 3.65280956e-01 -3.73077020e-03 8.43364477e-01
2.67041400e-02 1.90898925e-01 -8.24389696e-01 7.53835082e-01
3.98851521e-02 -1.04901230e+00 3.71700409e-03 4.71193232e-02
6.63395941e-01 2.23536447e-01 2.04379350e-01 7.07606003e-02
2.77114540e-01 -1.00387979e+00 8.33627284e-01 6.15439534e-01
8.91784787e-01 -8.02296817e-01 6.00973964e-01 4.92209941e-01
-1.49973583e+00 5.38553635e-04 -3.49782586e-01 4.91380692e-01
1.63221896e-01 4.39214349e-01 -5.71477950e-01 6.65321827e-01
8.68740737e-01 7.00160086e-01 -6.21131241e-01 1.02457893e+00
-5.54184258e-01 5.15587389e-01 -6.19504571e-01 3.64268035e-01
2.83036679e-01 -1.89078420e-01 2.03071654e-01 7.59839237e-01
3.83096427e-01 3.15846443e-01 2.23806441e-01 5.71097314e-01
-1.98635101e-01 2.20979631e-01 -2.70597607e-01 5.85979283e-01
4.63158339e-01 1.28985357e+00 -1.16778517e+00 -1.70661688e-01
-4.40161437e-01 7.15527952e-01 5.08990347e-01 1.26840040e-01
-8.65786493e-01 1.06649272e-01 1.24861017e-01 7.96580911e-02
2.87486941e-01 -1.17349647e-01 -2.03235060e-01 -1.03551352e+00
1.04535006e-01 -6.98348165e-01 7.27856040e-01 -8.14227223e-01
-1.17027676e+00 6.39570594e-01 -1.68051347e-01 -1.23217165e+00
-3.19910377e-01 -3.67821276e-01 -5.36279082e-01 4.69372511e-01
-1.21503353e+00 -1.67171168e+00 -2.43125126e-01 7.94768214e-01
5.82501590e-01 2.71957338e-01 5.88311791e-01 2.51403004e-01
-5.95695496e-01 5.56726396e-01 1.70614704e-01 5.52791595e-01
8.14754844e-01 -1.33834887e+00 -2.33208850e-01 6.94275022e-01
3.47312897e-01 7.04785287e-01 3.17207813e-01 -5.62813342e-01
-1.17363608e+00 -1.01206934e+00 1.41762033e-01 -5.29270649e-01
5.83046317e-01 -2.20127612e-01 -1.03278744e+00 6.99877501e-01
-4.21235174e-01 7.52669930e-01 4.44932818e-01 -1.41663745e-01
-2.38374487e-01 -3.63277882e-01 -7.32406318e-01 4.35615361e-01
8.37574363e-01 -6.59713626e-01 -8.13196778e-01 1.92104116e-01
5.87159634e-01 -9.15594459e-01 -1.07551682e+00 1.01935399e+00
8.39336097e-01 -6.69219136e-01 1.32929254e+00 -1.03543393e-01
6.78563654e-01 -6.15339637e-01 2.23676879e-02 -9.50578809e-01
4.62854095e-02 1.75927892e-01 -1.65457696e-01 9.71234202e-01
9.66678560e-02 -1.96303695e-01 8.75629604e-01 3.12331080e-01
-4.40952152e-01 -1.52449727e+00 -1.10776734e+00 -3.60158175e-01
4.34533119e-01 -2.61265069e-01 5.23695350e-01 7.84962416e-01
-5.07776797e-01 2.72672325e-01 -4.08338100e-01 3.19651961e-01
3.79158467e-01 7.92548239e-01 5.81031680e-01 -1.07129145e+00
-3.31766099e-01 -3.84625852e-01 -3.83977711e-01 -1.31741643e+00
2.09750190e-01 -8.29535007e-01 1.65037643e-02 -1.59637880e+00
3.36154759e-01 -6.25984669e-01 -7.65507579e-01 4.91270989e-01
-2.86812276e-01 4.02698129e-01 -1.77431047e-01 5.52232862e-01
-5.45633972e-01 5.72239816e-01 1.93434787e+00 -6.36533201e-02
1.74716100e-01 1.49722680e-01 -3.95012856e-01 1.06599784e+00
2.96536088e-01 -5.64514220e-01 -4.95349646e-01 -6.03726387e-01
-1.73929751e-01 4.34303790e-01 2.44562224e-01 -8.95242989e-01
5.19273221e-01 7.87123572e-03 6.41501188e-01 -9.54054415e-01
4.73767102e-01 -1.07010221e+00 -8.63735154e-02 5.31891525e-01
-3.49197537e-01 1.90682456e-01 -2.11828455e-01 5.81441045e-01
-3.91393781e-01 -1.85108468e-01 8.16886902e-01 -3.70604336e-01
-4.89270300e-01 7.36917555e-01 -2.35838622e-01 3.82584147e-02
1.04106355e+00 -1.47329912e-01 1.44349322e-01 -2.80645907e-01
-7.42762446e-01 3.66496921e-01 5.17376125e-01 3.06435585e-01
7.07982600e-01 -1.30680835e+00 -4.65033323e-01 8.93864855e-02
9.81681719e-02 5.50597012e-01 3.87160838e-01 9.69672382e-01
-6.02832079e-01 9.22891647e-02 -3.16204667e-01 -9.85263824e-01
-1.09456348e+00 5.91281593e-01 7.51546264e-01 -3.86470914e-01
-9.62003112e-01 8.49389493e-01 5.41564524e-01 -6.04271472e-01
1.53422102e-01 -4.82114464e-01 -4.36297923e-01 -3.30101848e-01
1.59741417e-01 1.58017334e-02 -7.31969178e-02 -1.07006729e+00
-3.13991755e-01 1.26090133e+00 -1.17610125e-02 1.12597682e-01
1.26858342e+00 -2.69897342e-01 -6.85814917e-02 6.31365299e-01
1.28403699e+00 2.97533840e-01 -1.25270271e+00 -5.07557452e-01
-3.07598621e-01 -2.71771640e-01 3.71994749e-02 -6.98528469e-01
-1.66467345e+00 1.13222587e+00 6.07875288e-01 -4.97898847e-01
1.08837306e+00 3.13651443e-01 9.97575104e-01 3.31997126e-02
5.39768875e-01 -9.01758790e-01 1.78596199e-01 2.79495835e-01
5.87557912e-01 -1.17361915e+00 9.75423083e-02 -3.59988451e-01
-4.61287409e-01 1.24309826e+00 8.49301577e-01 -1.75651327e-01
8.37318897e-01 3.43133062e-01 4.24916744e-01 -5.63456237e-01
-2.75050640e-01 -1.33921608e-01 5.98436475e-01 2.00541154e-01
3.02605093e-01 -8.00297260e-02 -2.19275594e-01 7.24257827e-01
-2.72753417e-01 -3.41465920e-01 1.87408656e-01 7.61236012e-01
-2.83312112e-01 -1.09544873e+00 -4.23227310e-01 4.59631532e-01
-8.87053668e-01 3.04500580e-01 -5.08800261e-02 9.92654383e-01
2.71598488e-01 3.96387339e-01 -1.15071304e-01 -1.06436037e-01
1.58388525e-01 -3.89722615e-01 4.38165635e-01 -5.71526289e-01
-4.47915405e-01 4.16539133e-01 -5.15535831e-01 -5.91287017e-01
-3.71973753e-01 -5.11600912e-01 -2.02049088e+00 2.38893941e-01
-4.10749376e-01 1.04336850e-01 6.04964733e-01 1.23249662e+00
-1.15377650e-01 7.53481209e-01 6.33605003e-01 -6.52049541e-01
-4.94554400e-01 -7.22198248e-01 -6.36600137e-01 5.48320293e-01
3.99537265e-01 -1.01346028e+00 -2.39344109e-02 -9.71036851e-02] | [14.684100151062012, -2.2649431228637695] |
54343750-771f-4180-8b80-f985ce6acf3e | mlrg-deep-curvature | 1912.09656 | null | https://arxiv.org/abs/1912.09656v2 | https://arxiv.org/pdf/1912.09656v2.pdf | Deep Curvature Suite | We present MLRG Deep Curvature suite, a PyTorch-based, open-source package for analysis and visualisation of neural network curvature and loss landscape. Despite of providing rich information into properties of neural network and useful for a various designed tasks, curvature information is still not made sufficient use for various reasons, and our method aims to bridge this gap. We present a primer, including its main practical desiderata and common misconceptions, of \textit{Lanczos algorithm}, the theoretical backbone of our package, and present a series of examples based on synthetic toy examples and realistic modern neural networks tested on CIFAR datasets, and show the superiority of our package against existing competing approaches for the similar purposes. | ['Xingchen Wan', 'Timur Garipov', 'Diego Granziol'] | 2019-12-20 | null | https://openreview.net/forum?id=86t2GlfzFo | https://openreview.net/pdf?id=86t2GlfzFo | null | ['misconceptions'] | ['miscellaneous'] | [-1.62811249e-01 1.61850184e-01 1.83557943e-01 -5.61580837e-01
-2.10600078e-01 -4.24100280e-01 4.64901567e-01 -1.26669765e-01
-6.07390702e-01 8.19816530e-01 -6.20493293e-03 -8.04505408e-01
-4.87715214e-01 -2.34767467e-01 -5.98228514e-01 -1.14742482e+00
-5.94752550e-01 -7.19492789e-03 3.59140813e-01 -2.87518114e-01
4.25017983e-01 5.61987102e-01 -1.18868053e+00 1.04224265e-01
6.26420796e-01 9.98702884e-01 -5.29207215e-02 6.19411290e-01
4.71520036e-01 6.24211550e-01 -1.21073544e-01 -8.36648285e-01
3.63685966e-01 -3.52778472e-02 -8.44994247e-01 -7.24575818e-01
4.74099874e-01 -1.94981545e-01 -3.29909056e-01 1.11096454e+00
5.91986477e-01 1.71977341e-01 1.03236628e+00 -1.07304811e+00
-5.47630906e-01 5.68891585e-01 -5.66710413e-01 4.24427927e-01
-4.70710605e-01 1.31450802e-01 1.25409698e+00 -7.12401271e-01
3.69764179e-01 1.12706029e+00 1.12265217e+00 7.26716280e-01
-1.18463016e+00 -5.37118196e-01 1.80815250e-01 6.23726621e-02
-9.29354072e-01 -4.38647836e-01 6.92408979e-01 -4.39993501e-01
8.40587616e-01 4.86838251e-01 5.21459222e-01 1.18954110e+00
2.28920966e-01 1.01225531e+00 8.68105769e-01 1.21683784e-01
3.23921084e-01 2.41780788e-01 6.34440184e-01 6.36510968e-01
-1.96604393e-02 3.09302151e-01 -1.40994206e-01 -4.11023498e-01
1.07633817e+00 -3.31869215e-01 -4.90170956e-01 -6.12441480e-01
-8.98685813e-01 8.93624425e-01 8.69487107e-01 -4.98588458e-02
-1.20114140e-01 3.39402944e-01 5.41153491e-01 3.81973326e-01
4.20010507e-01 2.25897446e-01 -6.82110727e-01 -2.80601710e-01
-6.60658240e-01 7.25359857e-01 6.13381147e-01 6.87589884e-01
2.87214667e-01 2.10937355e-02 1.61361486e-01 9.71971571e-01
5.64721763e-01 -1.34021163e-01 5.22510171e-01 -8.57483149e-01
3.60195339e-01 3.66877049e-01 -2.91402191e-01 -1.05168509e+00
-7.11290538e-01 -6.24216497e-01 -1.24595797e+00 4.95869637e-01
8.54799628e-01 -4.46246207e-01 -4.82648879e-01 1.66112208e+00
7.69284442e-02 -1.79737717e-01 2.31731869e-02 8.85415435e-01
8.67647290e-01 4.26820248e-01 -1.27724499e-01 1.13913111e-01
9.33428109e-01 -9.59108829e-01 -1.98162839e-01 -2.04386562e-02
8.86813641e-01 -3.72740388e-01 9.77296770e-01 5.22382140e-01
-1.22410202e+00 -2.32757218e-02 -9.85772252e-01 2.58415919e-02
-5.93610704e-01 1.05347343e-01 1.08623958e+00 8.15890312e-01
-1.38036120e+00 1.07446563e+00 -7.92158902e-01 6.09276183e-02
9.01538968e-01 3.36115688e-01 -2.87440956e-01 5.14995396e-01
-8.88448060e-01 5.89694679e-01 3.31469953e-01 2.98332721e-01
-4.00477797e-01 -7.80743122e-01 -5.34986615e-01 -1.29364461e-01
6.18302077e-03 -4.21666682e-01 1.44704282e+00 -7.99137533e-01
-1.41613221e+00 6.26778245e-01 1.65845647e-01 -8.22553277e-01
8.54235232e-01 -1.92611411e-01 5.57413474e-02 -1.76606059e-01
-8.47530067e-01 7.07579970e-01 3.85014117e-01 -9.22406018e-01
-2.68072903e-01 -4.55038518e-01 -1.80654928e-01 2.38320753e-01
-2.27678075e-01 3.68823633e-02 -4.05111998e-01 -7.31311142e-01
-5.64595908e-02 -7.80299067e-01 -5.14788210e-01 1.51229307e-01
-6.07214212e-01 -2.38189280e-01 4.99905229e-01 -4.80620205e-01
1.15516698e+00 -1.81341720e+00 -1.28141686e-01 4.75380778e-01
5.22807777e-01 3.46817970e-01 1.58302635e-01 3.10090214e-01
-3.01685214e-01 2.36681879e-01 -6.88418508e-01 -1.52810618e-01
-1.21505717e-02 -1.71386436e-01 -3.17283392e-01 6.79582655e-01
-1.34206057e-01 8.60442996e-01 -5.87471485e-01 -5.09135649e-02
-1.37302857e-02 5.74196994e-01 -6.68810248e-01 -3.66993457e-01
5.75897917e-02 2.00215369e-01 -3.57904822e-01 3.96706522e-01
7.88672209e-01 -1.31579667e-01 -2.21639305e-01 -3.98390964e-02
-9.52871814e-02 2.50427961e-01 -9.21677589e-01 1.33751392e+00
3.95539142e-02 1.07700264e+00 1.72244608e-01 -1.06597984e+00
7.07925081e-01 1.07341453e-01 2.27105454e-01 -2.24283338e-01
1.62753806e-01 5.34482040e-02 2.80018449e-01 -2.73294985e-01
7.67432153e-02 3.56244236e-01 3.95079434e-01 2.82927036e-01
-6.72847629e-02 2.84048736e-01 -1.98751271e-01 1.26323715e-01
1.13826597e+00 1.59539387e-01 4.38241996e-02 -7.67777264e-01
3.61521542e-01 -3.10807735e-01 2.01406658e-01 5.06546915e-01
-5.86165428e-01 6.76039815e-01 1.02373135e+00 -7.23977149e-01
-1.24578333e+00 -1.00241089e+00 -6.60388470e-01 1.09729803e+00
-3.03333819e-01 -2.81393975e-01 -1.11052370e+00 -6.22613728e-01
-9.60055459e-03 5.71559250e-01 -9.44930315e-01 3.52077410e-02
-3.88599515e-01 -1.26943851e+00 8.88266623e-01 4.08790231e-01
5.51433742e-01 -1.28827655e+00 -5.82966089e-01 -2.78014421e-01
5.35348833e-01 -6.81875110e-01 -1.52660832e-01 1.88977420e-01
-1.28176856e+00 -1.10751879e+00 -8.60729575e-01 -7.92367458e-01
6.23177886e-01 4.89444360e-02 9.81773734e-01 -4.05193307e-02
-2.98698038e-01 -1.65866435e-01 3.00742626e-01 -2.75503397e-01
-6.79944083e-02 1.14051573e-01 8.20568204e-02 -3.56601894e-01
1.54746354e-01 -8.59161735e-01 -9.34473157e-01 3.95609945e-01
-7.81514943e-01 2.71611363e-01 4.89232272e-01 5.73916674e-01
2.98891589e-03 -2.02226430e-01 8.27097237e-01 -9.88580227e-01
1.23353744e+00 -5.20554185e-01 -7.92211294e-01 2.94131227e-02
-7.17509985e-01 7.67470151e-02 6.39553010e-01 -1.02198184e-01
-8.81266057e-01 -1.36162087e-01 -7.86901951e-01 -1.24886349e-01
-2.32807159e-01 3.12692434e-01 8.96179974e-02 -3.25803161e-01
8.65342915e-01 -1.38529763e-02 8.56174976e-02 -7.22615480e-01
3.13618422e-01 4.63079542e-01 5.53623617e-01 -4.40793782e-01
5.02392709e-01 3.38955641e-01 2.03833394e-02 -1.00214100e+00
-3.98421466e-01 -2.91602910e-01 -7.39721656e-01 -1.28175378e-01
4.95680630e-01 -4.37513709e-01 -1.21776581e+00 6.48081839e-01
-1.02693331e+00 -4.96849835e-01 2.14511022e-01 1.44135356e-01
-5.72404683e-01 3.19539160e-01 -6.93138599e-01 -9.34331059e-01
-6.08105779e-01 -1.31650364e+00 3.68471324e-01 4.30546373e-01
1.99474186e-01 -1.64958823e+00 3.00524861e-01 2.73722783e-02
5.68447292e-01 4.75044250e-01 8.29375207e-01 -7.77389705e-01
-1.98303908e-01 -5.46787530e-02 -4.77598846e-01 5.31777203e-01
-3.92371446e-01 3.28099608e-01 -1.46311593e+00 -1.52950361e-01
-1.90396339e-01 -5.19500911e-01 1.22723842e+00 5.96215725e-01
1.68057024e+00 -5.70930839e-01 -2.96885073e-01 8.47198784e-01
1.37641466e+00 -2.42313780e-02 6.43015325e-01 3.10013831e-01
5.01369894e-01 7.28041887e-01 -2.22259033e-02 3.23831618e-01
2.29815185e-01 3.89096409e-01 6.88468874e-01 -1.40453696e-01
5.23178756e-01 3.00182980e-02 2.71279424e-01 6.99620962e-01
-3.83985758e-01 1.71816126e-01 -1.13270712e+00 2.50125647e-01
-2.03890824e+00 -9.71515477e-01 -3.94630224e-01 2.15267134e+00
1.06163955e+00 4.38985497e-01 4.85793382e-01 7.49388784e-02
3.70448053e-01 7.45489495e-03 -7.67583609e-01 -6.12397075e-01
-1.50667861e-01 -2.70234853e-01 6.53009295e-01 4.51064497e-01
-1.09282172e+00 8.37987900e-01 7.39552450e+00 1.02183247e+00
-8.54131877e-01 -4.14968222e-01 1.16379344e+00 8.78536701e-02
-1.00086264e-01 -1.65224329e-01 -7.33208060e-01 1.71361133e-01
9.82100725e-01 -1.31157592e-01 4.29929048e-01 9.19538617e-01
1.13399714e-01 -8.08139890e-02 -1.11844838e+00 8.78026724e-01
-4.13928390e-01 -1.52997077e+00 -4.06702220e-01 7.65360743e-02
4.03470278e-01 7.35471904e-01 5.49317956e-01 1.55684620e-01
4.98051018e-01 -1.17135143e+00 3.37494195e-01 5.48336208e-01
3.75827700e-01 -9.97625232e-01 5.85940003e-01 5.37591517e-01
-7.65618205e-01 6.74435645e-02 -7.64165163e-01 -1.34737313e-01
-3.95202875e-01 6.49254441e-01 -6.98849857e-01 1.73594385e-01
8.77488971e-01 8.00710320e-01 -9.07412589e-01 1.51808572e+00
1.78245932e-01 7.38822043e-01 -3.01548094e-01 -3.74382645e-01
5.33459842e-01 -4.92217302e-01 5.66715598e-01 1.49491310e+00
-3.35662693e-01 -2.78889984e-01 -7.09537625e-01 1.17267120e+00
1.93834137e-02 2.01588452e-01 -8.60636055e-01 2.95228720e-01
9.69821811e-02 1.64926171e+00 -5.91839850e-01 5.59869818e-02
-1.79643825e-01 3.95397007e-01 7.43304133e-01 5.85140228e-01
-9.66541827e-01 -7.03477859e-01 8.83316338e-01 -2.09097922e-01
-8.39521587e-02 -3.24105829e-01 -5.31377137e-01 -9.62424219e-01
-1.85861569e-02 -7.01750040e-01 1.78258449e-01 -5.35378814e-01
-1.17514718e+00 5.81024349e-01 1.41389370e-01 -7.42882967e-01
-3.34105976e-02 -1.08213544e+00 -1.13241255e+00 7.51298368e-01
-1.38544643e+00 -4.71847624e-01 1.36991203e-01 4.50942218e-01
3.89915407e-01 -1.96633115e-01 5.50059497e-01 1.12944998e-01
-1.00788307e+00 8.59436989e-01 8.00768375e-01 2.64511764e-01
4.06198561e-01 -1.44866502e+00 1.03849638e+00 2.18072414e-01
-2.37511378e-02 9.24950242e-01 8.33968461e-01 -1.01916373e-01
-1.08168888e+00 -9.44939673e-01 2.75937527e-01 -4.41097379e-01
8.65316153e-01 -5.48340857e-01 -1.03757918e+00 6.57905757e-01
2.18043864e-01 -1.76397234e-01 5.71477711e-01 3.36899072e-01
-1.94441929e-01 6.48374632e-02 -9.99773920e-01 1.09075427e+00
1.19425929e+00 -3.22222076e-02 -4.95093942e-01 3.32912773e-01
4.27294254e-01 -2.75240839e-01 -7.89680958e-01 5.39896488e-01
8.06676447e-01 -1.72447443e+00 1.03414106e+00 -7.11893082e-01
3.88450056e-01 3.50658536e-01 2.02028647e-01 -1.60864151e+00
-1.37312099e-01 -8.92457306e-01 -7.30505586e-02 9.98563409e-01
8.25223207e-01 -9.50891614e-01 1.05654109e+00 8.11949432e-01
-1.09116428e-01 -1.43376791e+00 -1.03904319e+00 -5.15021980e-01
7.98416734e-01 -7.21681356e-01 9.47173759e-02 7.28257179e-01
3.47637653e-01 2.99341321e-01 3.96348499e-02 -2.81422406e-01
7.63281047e-01 -4.32971448e-01 6.69026434e-01 -1.44846702e+00
-1.09462105e-01 -1.12983847e+00 -2.67575920e-01 -1.17000496e+00
-7.89392740e-02 -8.71256828e-01 -3.34473342e-01 -1.12511480e+00
7.61995651e-03 -5.77299237e-01 -1.92472696e-01 4.27970767e-01
1.83622733e-01 6.65310025e-02 -1.38557583e-01 2.97561646e-01
-4.19481009e-01 5.63741148e-01 1.27836657e+00 2.02802673e-01
-1.88673437e-01 5.56233346e-01 -7.73729205e-01 1.22224748e+00
1.20232189e+00 -7.60970488e-02 -5.33579528e-01 -2.77331114e-01
5.75281441e-01 -3.64599317e-01 6.18572056e-01 -1.09396553e+00
4.97011729e-02 -4.33610938e-02 5.05098224e-01 -5.41167259e-01
3.46072763e-01 -5.65160096e-01 -4.92152929e-01 4.35980439e-01
-7.36175656e-01 2.94720352e-01 5.82137048e-01 5.00082910e-01
3.43634933e-01 -2.26654723e-01 1.08218944e+00 -1.19428309e-02
-1.02615424e-01 4.98637348e-01 -1.21818177e-01 1.25297129e-01
7.53369391e-01 -1.13125376e-01 -5.30668795e-01 -3.24222177e-01
-6.39388204e-01 2.11287811e-01 3.03791374e-01 2.09141701e-01
6.28160655e-01 -1.21876121e+00 -3.97608012e-01 3.86443317e-01
-1.51163653e-01 -2.54965741e-02 1.94616076e-02 1.03663754e+00
-6.55908644e-01 6.39907002e-01 -3.56777608e-01 -4.83576089e-01
-1.14756620e+00 1.87831208e-01 6.93472326e-01 -1.08051307e-01
-8.97581041e-01 1.08681452e+00 5.78246474e-01 -5.79772830e-01
7.55001783e-01 -6.30423903e-01 -6.22051537e-01 -4.09908265e-01
5.36196470e-01 4.09828216e-01 1.50346562e-01 -3.16458970e-01
-1.59431189e-01 4.04370964e-01 -2.82752901e-01 -1.44991115e-01
1.52263558e+00 3.21924724e-02 6.41831085e-02 6.05952084e-01
1.34803879e+00 -4.86304045e-01 -1.59530401e+00 -1.84319600e-01
6.71522245e-02 1.51600561e-03 1.81169078e-01 -6.72429204e-01
-1.18438756e+00 1.17205453e+00 6.11529887e-01 3.66984606e-01
9.39646721e-01 -3.39889526e-01 3.57395619e-01 9.47300196e-01
-8.64853635e-02 -1.07874095e+00 -3.40910554e-01 7.86822736e-01
1.00764108e+00 -1.03549957e+00 -1.22091837e-01 -1.82957090e-02
-8.94619167e-01 1.27901065e+00 6.34291172e-01 -3.63247037e-01
9.97773647e-01 1.16993897e-01 5.11334538e-02 -2.46369079e-01
-1.21577358e+00 3.18138570e-01 3.78688782e-01 6.63471878e-01
6.89956248e-01 -4.42423910e-01 -3.51106841e-03 5.29575169e-01
-5.47074795e-01 -3.39985013e-01 3.21773410e-01 5.33326626e-01
-6.27211571e-01 -5.40847957e-01 -1.23593286e-01 5.37952542e-01
-8.01045358e-01 -2.72595167e-01 -4.60130960e-01 8.81133854e-01
-4.26595688e-01 2.25225851e-01 -1.45507455e-01 -3.74834865e-01
4.58714116e-04 -2.79586464e-02 3.79791617e-01 9.35778692e-02
-7.44343758e-01 -1.59339905e-01 -4.20253165e-02 -4.48390841e-01
1.58251867e-01 -6.77627325e-01 -1.14738345e+00 -4.80080158e-01
-2.56358206e-01 9.75916386e-02 9.64778364e-01 8.08877647e-01
1.32339135e-01 2.82290757e-01 5.08936644e-01 -9.93637621e-01
-7.77821243e-01 -9.27546382e-01 -6.15701795e-01 1.40050158e-01
5.87850451e-01 -5.98588288e-01 -6.22042060e-01 -1.70090020e-01] | [7.945828437805176, 3.581557512283325] |
d09c454f-a061-4d54-a20f-264714b7db26 | a-causal-based-framework-for-multimodal | 2008.02171 | null | https://arxiv.org/abs/2008.02171v1 | https://arxiv.org/pdf/2008.02171v1.pdf | A Causal-based Framework for Multimodal Multivariate Time Series Validation Enhanced by Unsupervised Deep Learning as an Enabler for Industry 4.0 | An advanced conceptual validation framework for multimodal multivariate time series defines a multi-level contextual anomaly detection ranging from an univariate context definition, to a multimodal abstract context representation learnt by an Autoencoder from heterogeneous data (images, time series, sounds, etc.) associated to an industrial process. Each level of the framework is either applicable to historical data and/or live data. The ultimate level is based on causal discovery to identify causal relations in observational data in order to exclude biased data to train machine learning models and provide means to the domain expert to discover unknown causal relations in the underlying process represented by the data sample. A Long Short-Term Memory Autoencoder is successfully evaluated on multivariate time series to validate the learnt representation of abstract contexts associated to multiple assets of a blast furnace. A research roadmap is identified to combine causal discovery and representation learning as an enabler for unsupervised Root Cause Analysis applied to the process industry. | ['Cedric Schockaert'] | 2020-08-05 | null | null | null | null | ['contextual-anomaly-detection'] | ['miscellaneous'] | [ 6.51001573e-01 1.91352606e-01 2.34307438e-01 -7.04768822e-02
-3.59238744e-01 -3.67385775e-01 9.11894441e-01 9.06283677e-01
2.83442497e-01 3.58325481e-01 3.94505620e-01 -5.66375017e-01
-9.14867401e-01 -1.03898859e+00 -6.62684619e-01 -9.20597494e-01
-5.93906105e-01 3.63976091e-01 -3.58270049e-01 -1.98178947e-01
2.89430797e-01 7.25013435e-01 -1.69774115e+00 7.01853395e-01
1.42263874e-01 1.00719631e+00 9.18289423e-02 7.56188214e-01
-2.72659302e-01 1.16365480e+00 -7.27964342e-01 7.84280524e-02
-2.26735413e-01 -4.51623678e-01 -5.54924071e-01 -2.78778560e-02
1.51457489e-01 1.64688304e-01 -1.97889358e-01 9.57863033e-01
2.01054126e-01 9.64840055e-02 8.13246965e-01 -1.32775569e+00
-6.24481380e-01 6.93189263e-01 -1.26202524e-01 3.87643754e-01
4.43873852e-01 4.87449877e-02 9.34893608e-01 -7.25357652e-01
4.83931154e-01 1.41827250e+00 4.61850703e-01 -2.56496836e-02
-1.34867120e+00 -2.59006470e-01 2.13256463e-01 3.04744154e-01
-8.26534331e-01 1.29392803e-01 8.92090738e-01 -8.00511301e-01
8.10185373e-01 3.00945878e-01 6.57612860e-01 1.59742713e+00
6.93172574e-01 3.10682565e-01 8.91618550e-01 -4.48583007e-01
6.71686769e-01 -4.84340370e-01 6.33676127e-02 4.22764838e-01
5.59874997e-02 5.32057405e-01 -7.32214630e-01 -4.63814706e-01
6.83228016e-01 3.03477764e-01 9.52566192e-02 -4.17782515e-02
-1.24109006e+00 1.04200339e+00 2.82008827e-01 6.14589512e-01
-9.62736785e-01 1.25393137e-01 8.14502299e-01 7.39377677e-01
2.42420971e-01 6.96417928e-01 -7.85384059e-01 1.39861524e-01
-4.68316674e-01 1.67601451e-01 8.45137000e-01 2.16600016e-01
3.73743445e-01 5.40736794e-01 -2.20220424e-02 3.84169221e-01
4.07945931e-01 5.37716210e-01 5.67414165e-01 -8.57419729e-01
-3.79601605e-02 8.81481946e-01 -2.92823166e-01 -9.71327364e-01
-3.11956137e-01 -2.69064337e-01 -7.34068990e-01 5.83548248e-01
2.66505718e-01 -2.75152028e-01 -7.20116615e-01 1.36186111e+00
2.98175395e-01 5.50347626e-01 2.88436025e-01 6.52634382e-01
7.42729783e-01 7.99357057e-01 2.52426386e-01 -2.75875807e-01
1.29299247e+00 1.25964120e-01 -8.24542522e-01 1.42969549e-01
3.02290678e-01 -3.91764283e-01 7.05877960e-01 7.48199403e-01
-4.56566960e-01 -4.45910007e-01 -1.22476232e+00 5.55122137e-01
-8.63989830e-01 -3.04403067e-01 4.26217973e-01 1.43654019e-01
-2.70339310e-01 8.26927662e-01 -5.45341194e-01 -4.74891782e-01
-3.70968617e-02 1.59536570e-01 -4.47600186e-01 2.39899009e-02
-1.55595899e+00 6.63940132e-01 7.11454511e-01 2.24099383e-01
-1.39138317e+00 -8.32129836e-01 -9.30123210e-01 3.39506567e-02
3.39221865e-01 -5.04636288e-01 8.66099954e-01 -8.98160398e-01
-1.08871663e+00 2.75142431e-01 2.64115870e-01 -6.57599032e-01
2.15583637e-01 -3.50901544e-01 -1.05871832e+00 3.42758708e-02
3.91056528e-03 -3.63912404e-01 1.26407290e+00 -1.43306112e+00
-6.56548381e-01 -4.11702126e-01 -3.75220239e-01 -4.92059231e-01
-1.93766922e-01 -5.72802238e-02 5.19216955e-01 -7.54312456e-01
1.55200914e-01 -7.05176294e-01 -1.77894145e-01 -7.92164385e-01
-4.14027333e-01 -2.31986821e-01 1.46110165e+00 -9.17443514e-01
1.02996528e+00 -2.07349539e+00 3.67195904e-01 5.08243322e-01
-1.43802390e-01 -4.49707448e-01 -8.93945545e-02 7.70408988e-01
-9.70827818e-01 -2.07290929e-02 -4.30839747e-01 3.47395271e-01
-1.87396184e-01 3.93549055e-01 -8.20887864e-01 4.75404859e-01
5.88002563e-01 3.88025016e-01 -1.03877640e+00 1.94839519e-02
6.14007294e-01 3.60949099e-01 1.12538472e-01 3.43255222e-01
-4.63306010e-01 4.92081016e-01 -4.70953792e-01 7.74662256e-01
-6.01170771e-03 -7.23044993e-03 2.53992021e-01 -1.68588638e-01
-2.54957646e-01 -2.16981620e-01 -1.16278267e+00 1.50650334e+00
-4.78267819e-01 5.35750031e-01 -2.40590841e-01 -1.16539478e+00
1.04641807e+00 8.86864007e-01 7.72645175e-01 -5.56483984e-01
1.41068280e-01 2.47363355e-02 -1.35517612e-01 -9.50581431e-01
2.27896422e-01 -3.64766568e-01 -1.36372328e-01 2.30404451e-01
2.98495859e-01 1.09171554e-01 -2.10981630e-02 -1.36086017e-01
1.33003354e+00 7.34739900e-02 6.27747998e-02 -2.98433691e-01
6.67919278e-01 6.35497943e-02 4.47498798e-01 5.70480287e-01
1.84206709e-01 3.59698325e-01 7.60719657e-01 -9.10174668e-01
-9.71171498e-01 -1.23652065e+00 -7.06846863e-02 7.65985668e-01
-3.11118543e-01 -3.05273503e-01 -8.08323622e-02 -6.30906820e-01
1.87800318e-01 1.11899841e+00 -1.00371027e+00 -3.71326745e-01
-3.19908828e-01 -6.72006190e-01 3.12351674e-01 4.33781177e-01
-2.63231039e-01 -1.20550656e+00 -7.91588306e-01 5.45530558e-01
2.14618996e-01 -6.98280871e-01 7.80287862e-01 6.88210905e-01
-1.06855989e+00 -1.32342887e+00 6.34365156e-02 -2.00550184e-01
5.43947667e-02 -7.31375396e-01 1.08261669e+00 -3.40586543e-01
-4.84136909e-01 7.13194847e-01 -2.01146781e-01 -9.41996336e-01
-9.11372602e-01 -4.32898670e-01 3.06155443e-01 2.04579398e-01
1.54203117e-01 -7.33627021e-01 -2.57059991e-01 -7.97830056e-03
-1.14398873e+00 -8.63649070e-01 5.17628372e-01 9.39876139e-01
6.02799296e-01 6.44909680e-01 8.48738968e-01 -3.00486594e-01
6.44371390e-01 -1.01075470e+00 -4.82068479e-01 2.48735502e-01
-6.11175060e-01 1.29629567e-01 3.78020495e-01 -5.19571483e-01
-1.19779885e+00 -6.21169917e-02 2.46364906e-01 -7.96761334e-01
-8.56417418e-01 1.14014387e+00 -1.46406785e-01 8.13610196e-01
7.59744108e-01 2.04897057e-02 9.18704122e-02 -4.54465926e-01
3.55035871e-01 6.04697950e-02 8.70264411e-01 -8.81968796e-01
6.45944536e-01 5.18281877e-01 4.71251965e-01 -1.07024276e+00
-2.99603701e-01 -1.65104404e-01 -7.21861303e-01 -6.40163183e-01
1.17239738e+00 -8.04703712e-01 -6.22595489e-01 -1.14944965e-01
-1.09588385e+00 -9.94754136e-02 -6.85029209e-01 7.63750136e-01
-4.69227761e-01 -1.07164986e-01 -5.81022561e-01 -9.30076778e-01
-2.22260524e-02 -5.28502703e-01 1.04525518e+00 -1.65872663e-01
-6.38995230e-01 -1.14488256e+00 3.93202990e-01 2.88359728e-02
-9.75946430e-03 9.52657759e-01 1.42094612e+00 -1.09284854e+00
-1.57020807e-01 -6.96522832e-01 2.93327034e-01 2.26897538e-01
3.96467537e-01 4.30928528e-01 -1.10928643e+00 -6.60065040e-02
3.30117524e-01 -2.43029259e-02 6.20398939e-01 3.63780022e-01
1.04825640e+00 -2.03629583e-01 -1.35150939e-01 -1.00882955e-01
1.28870821e+00 5.43821096e-01 4.64556694e-01 3.94574165e-01
6.56708360e-01 1.11292422e+00 5.55060506e-01 5.41199028e-01
-4.65568334e-01 -1.25500053e-01 1.08256125e+00 6.39722049e-02
4.62398678e-01 -9.36484244e-03 5.49144745e-01 5.43372273e-01
-3.69722992e-01 -3.04080024e-02 -1.10474265e+00 5.81729889e-01
-1.93692374e+00 -1.45804775e+00 -6.60493493e-01 2.16360664e+00
4.51444089e-02 7.54338875e-02 -1.46536455e-01 4.81937140e-01
8.30420852e-01 1.36922086e-02 -3.51432115e-01 -4.83714014e-01
-2.52757490e-01 8.97400603e-02 2.23255288e-02 2.02207521e-01
-1.29468429e+00 1.53751418e-01 6.09350824e+00 3.48882169e-01
-1.08879495e+00 -2.86320169e-02 3.04916084e-01 6.56617153e-03
-2.68946111e-01 1.57647170e-02 1.72519132e-01 2.60164496e-02
1.50451338e+00 -2.80629005e-03 4.69169728e-02 7.88086593e-01
2.58141726e-01 2.74479091e-01 -1.37329531e+00 5.88391542e-01
-3.09189886e-01 -1.34803581e+00 2.61943012e-01 3.11609566e-01
5.85495055e-01 7.29570612e-02 -4.46212292e-02 2.32341543e-01
4.96649176e-01 -1.09915078e+00 7.60979950e-01 1.03698289e+00
3.96342874e-02 -7.61542499e-01 8.93152356e-01 -8.79264846e-02
-1.00260365e+00 -6.78072870e-01 3.74085903e-02 -1.47211835e-01
1.16268165e-01 8.38446140e-01 -8.05946469e-01 1.10141289e+00
9.30145264e-01 7.59670734e-01 -4.05670583e-01 5.48895001e-01
-1.35714889e-01 8.12448680e-01 -6.06532991e-02 2.85364658e-01
7.09699690e-02 -1.43772110e-01 1.01878858e+00 1.03997004e+00
5.60914040e-01 -2.31693879e-01 -1.08488845e-02 9.38008666e-01
4.77778524e-01 -1.48026288e-01 -1.17316377e+00 -3.45957309e-01
1.06334805e-01 1.01570451e+00 -5.37029743e-01 -2.26025388e-01
-3.19369584e-01 7.21685708e-01 -4.78884369e-01 6.48312628e-01
-6.08196855e-01 -2.84964219e-02 6.60266221e-01 -1.21144637e-01
1.60770938e-01 -1.30753726e-01 -1.49621457e-01 -7.99385846e-01
-3.62404644e-01 -8.36021245e-01 1.05816770e+00 -9.23150063e-01
-1.76941597e+00 4.16502327e-01 1.52535304e-01 -1.22561193e+00
-5.86921334e-01 -8.38084936e-01 -9.39417720e-01 8.78357112e-01
-8.30950379e-01 -1.16790557e+00 -1.06589884e-01 7.31738567e-01
4.41330642e-01 -6.73911154e-01 1.20781040e+00 -1.44457847e-01
-6.09695375e-01 -6.90886855e-01 4.42790389e-02 5.57115227e-02
4.75209594e-01 -1.55218410e+00 -2.71071851e-01 9.48038995e-01
8.34175050e-02 6.62018120e-01 1.04562056e+00 -9.99832749e-01
-1.57399750e+00 -1.14672995e+00 4.23925906e-01 -5.51403582e-01
1.24851990e+00 1.38505563e-01 -1.16028929e+00 5.68080723e-01
4.49719638e-01 -6.66767135e-02 8.67214501e-01 3.56201172e-01
-2.13122725e-01 -2.02940390e-01 -9.11409974e-01 4.36152577e-01
3.17088246e-01 -9.01133239e-01 -1.08875763e+00 6.74910843e-02
7.79890597e-01 2.23465294e-01 -1.56698513e+00 4.12259787e-01
4.38211054e-01 -6.63018048e-01 9.72244203e-01 -1.04872620e+00
9.27272618e-01 -6.26242220e-01 -5.61323404e-01 -1.32103503e+00
-2.55541444e-01 -2.97553778e-01 -5.36027730e-01 1.23207676e+00
3.42173308e-01 -3.93644392e-01 2.47168481e-01 3.60487819e-01
-2.24621579e-01 -2.43109226e-01 -1.12938166e+00 -6.72419250e-01
-4.93085459e-02 -1.09232807e+00 8.09941888e-01 1.38671708e+00
-1.71738937e-01 3.33957374e-01 -2.67575635e-03 7.81305313e-01
7.04699695e-01 2.08793849e-01 3.20351034e-01 -1.69558859e+00
-6.41428540e-03 -4.29122835e-01 -5.25580585e-01 5.04518688e-01
1.73748970e-01 -6.60071492e-01 -2.93531507e-01 -1.19357574e+00
-5.43575525e-01 2.14855745e-01 -8.89261663e-01 2.44182587e-01
2.63025612e-01 -3.83099288e-01 -8.99428725e-02 -5.90197705e-02
2.62655079e-01 6.58822536e-01 6.01034999e-01 -6.08472347e-01
-1.65903360e-01 -2.88331509e-01 -2.05652669e-01 8.27791929e-01
6.94881558e-01 -4.91683275e-01 -4.58081424e-01 1.88484475e-01
5.61786234e-01 4.15776670e-01 1.07699394e+00 -9.01416957e-01
3.64683568e-02 -3.25136751e-01 6.08069181e-01 -6.62371814e-01
-2.14747980e-01 -1.37367404e+00 4.76373404e-01 5.59691250e-01
-3.52380276e-01 3.12132090e-01 2.69591659e-01 9.73327219e-01
-6.26510024e-01 -5.72082587e-02 6.07219245e-03 -1.07907139e-01
-9.78022039e-01 -5.52746467e-02 -5.88399649e-01 -7.20252812e-01
8.51456583e-01 2.42208187e-02 -1.67713314e-01 -2.02736467e-01
-1.19915056e+00 -7.16614202e-02 -1.76699117e-01 7.32132554e-01
6.95097566e-01 -1.45502102e+00 -8.68990004e-01 1.66223094e-01
4.42731768e-01 -2.30247617e-01 4.52477843e-01 5.77214897e-01
-1.44865438e-01 1.10993840e-01 -1.00520626e-01 -8.44405830e-01
-9.22218263e-01 8.92185450e-01 3.45420718e-01 4.93996893e-04
-7.46756315e-01 3.45378190e-01 -1.57027654e-02 -1.78241223e-01
-6.15812242e-02 -5.50571144e-01 -3.45160991e-01 5.17213643e-01
5.40601730e-01 4.07553285e-01 3.22947562e-01 -4.02123749e-01
-1.88325077e-01 2.80440673e-02 6.28512681e-01 -2.09767833e-01
1.69439662e+00 1.30741194e-01 -3.67080688e-01 1.27645445e+00
8.79372478e-01 -2.85551846e-01 -1.11357665e+00 1.71481654e-01
5.30128181e-01 1.37994185e-01 2.95932591e-01 -9.92536187e-01
-8.83324623e-01 6.52828395e-01 1.11021483e+00 8.25069308e-01
1.00864708e+00 6.14537932e-02 -2.11741813e-02 6.03852332e-01
-2.36034453e-01 -1.07925260e+00 1.78151712e-01 5.54786026e-01
1.60932422e+00 -1.15336168e+00 -1.74448386e-01 1.75987393e-01
-4.85872775e-01 1.75880587e+00 5.95025048e-02 -4.14722003e-02
9.65268552e-01 3.54921728e-01 7.29633048e-02 -9.72581804e-01
-8.14134240e-01 -4.69851727e-03 3.62989306e-01 5.25673389e-01
4.65782791e-01 3.99116687e-02 3.31522107e-01 5.45270324e-01
8.37837234e-02 -1.25178322e-01 3.85171890e-01 8.79711151e-01
6.23307675e-02 -8.01805079e-01 -9.46326613e-01 3.41782480e-01
-4.81866598e-01 1.23763770e-01 -4.10379738e-01 9.47142601e-01
3.25147033e-01 1.11481035e+00 4.21956688e-01 -4.63677883e-01
5.69155455e-01 7.20288455e-01 5.33080213e-02 -2.66136855e-01
-6.19225502e-01 4.01096307e-02 2.43409082e-01 -5.26487112e-01
-5.98765671e-01 -1.07933021e+00 -1.25047576e+00 9.26994979e-02
-1.09346127e-02 -1.03712507e-01 8.41423631e-01 1.10639036e+00
2.90459078e-02 1.36045933e+00 5.75022995e-01 -6.84352934e-01
1.77060235e-02 -1.08381677e+00 -7.69390583e-01 5.83683908e-01
6.88005149e-01 -5.91224194e-01 -4.40589815e-01 4.88407820e-01] | [7.086830139160156, 2.66949462890625] |
c8eb23c5-ecd0-4ba2-8ca6-a8366d54b42b | direct-simultaneous-speech-to-speech | 2110.08250 | null | https://arxiv.org/abs/2110.08250v2 | https://arxiv.org/pdf/2110.08250v2.pdf | Direct Simultaneous Speech-to-Speech Translation with Variational Monotonic Multihead Attention | We present a direct simultaneous speech-to-speech translation (Simul-S2ST) model, Furthermore, the generation of translation is independent from intermediate text representations. Our approach leverages recent progress on direct speech-to-speech translation with discrete units, in which a sequence of discrete representations, instead of continuous spectrogram features, learned in an unsupervised manner, are predicted from the model and passed directly to a vocoder for speech synthesis on-the-fly. We also introduce the variational monotonic multihead attention (V-MMA), to handle the challenge of inefficient policy learning in speech simultaneous translation. The simultaneous policy then operates on source speech features and target discrete units. We carry out empirical studies to compare cascaded and direct approach on the Fisher Spanish-English and MuST-C English-Spanish datasets. Direct simultaneous model is shown to outperform the cascaded model by achieving a better tradeoff between translation quality and latency. | ['Phillip Koehn', 'Juan Pino', 'Wei-Ning Hsu', 'Peng-Jen Chen', 'Yun Tang', 'Ann Lee', 'Danni Liu', 'Hongyu Gong', 'Xutai Ma'] | 2021-10-15 | null | null | null | null | ['speech-to-speech-translation'] | ['speech'] | [ 4.01315480e-01 3.27391237e-01 -2.45236441e-01 -2.90890515e-01
-1.68239236e+00 -7.43415236e-01 1.10448015e+00 -3.68561417e-01
-9.23660398e-02 9.36383247e-01 6.63081467e-01 -7.76827991e-01
6.60257339e-01 -3.80041122e-01 -1.09269702e+00 -5.83838105e-01
4.44412082e-01 6.95215642e-01 -1.97515458e-01 -2.80027300e-01
-2.62877405e-01 1.42074451e-01 -1.10815752e+00 6.87620223e-01
6.50971055e-01 6.73604369e-01 5.39325058e-01 1.12032700e+00
-1.20234385e-01 7.03177571e-01 -5.94697833e-01 -3.71213704e-01
2.78625190e-01 -9.83895719e-01 -7.46522129e-01 6.09684587e-02
1.58678696e-01 -3.68508637e-01 -2.84650505e-01 6.90948963e-01
7.52888381e-01 3.33547741e-01 8.22624266e-01 -9.22058582e-01
-1.10120392e+00 8.23581278e-01 -9.31819454e-02 2.44764149e-01
3.71772975e-01 3.33623439e-01 1.23213434e+00 -1.36315131e+00
6.29846156e-01 1.76969266e+00 9.78267044e-02 6.17091119e-01
-1.38002801e+00 -4.05324847e-01 1.57883912e-01 -2.19805300e-01
-9.94020939e-01 -1.21099293e+00 3.22851777e-01 -1.24384530e-01
1.68296838e+00 2.12108433e-01 3.25432271e-01 1.79069376e+00
2.35552460e-01 9.44025695e-01 1.06348395e+00 -6.52947068e-01
1.45400167e-01 2.50618964e-01 -4.71136808e-01 6.25718534e-01
-5.72318494e-01 6.72596335e-01 -9.36484277e-01 -8.40615761e-03
7.39405811e-01 -4.01931524e-01 -1.07319757e-01 2.70597845e-01
-1.56045139e+00 8.36505890e-01 3.41523886e-02 -9.49567929e-03
-5.62059879e-01 4.18058276e-01 3.84901643e-01 9.27786112e-01
7.40948856e-01 1.29374564e-01 -7.06667304e-01 -3.64259362e-01
-1.15020096e+00 1.14590131e-01 7.29416966e-01 1.32701874e+00
4.82679158e-01 7.01636910e-01 -7.98079193e-01 6.64547563e-01
5.55278003e-01 9.65942681e-01 8.60313475e-01 -7.17346907e-01
9.18967187e-01 -2.61128157e-01 7.48901963e-02 -7.02060461e-02
3.15883547e-01 -3.51490527e-01 -5.70465207e-01 -2.94440724e-02
-9.58329290e-02 -4.83978182e-01 -1.12691188e+00 1.68775392e+00
2.13142395e-01 2.71782011e-01 6.12009823e-01 8.24714124e-01
6.28549337e-01 1.41406190e+00 -1.49367109e-01 -5.09989798e-01
9.86522317e-01 -1.50019383e+00 -9.37924862e-01 -2.13062957e-01
3.69532138e-01 -1.26868892e+00 1.18526554e+00 -9.26098973e-02
-1.57217228e+00 -7.39585698e-01 -7.59822845e-01 -3.74766707e-01
-1.92604229e-01 3.54620546e-01 7.25992173e-02 3.35938513e-01
-1.48876166e+00 5.31341732e-01 -1.08872688e+00 -2.96760231e-01
-7.93972835e-02 4.80752945e-01 -3.67712900e-02 5.33571482e-01
-1.23590243e+00 9.71087694e-01 3.67394499e-02 -1.59386009e-01
-1.51485574e+00 -4.30135220e-01 -9.36989129e-01 1.21654630e-01
1.16016574e-01 -9.15962934e-01 1.90111804e+00 -1.28778839e+00
-2.59560609e+00 4.73045230e-01 -7.00233340e-01 -6.84318483e-01
5.06922126e-01 -2.67112523e-01 -1.99991256e-01 -1.26057509e-02
2.23826095e-02 7.19711006e-01 1.32554018e+00 -1.03572285e+00
-5.13790011e-01 1.47613317e-01 -2.53164232e-01 5.92924356e-01
1.15773328e-01 2.11105749e-01 -1.92867503e-01 -8.23950887e-01
-2.57360458e-01 -1.01660025e+00 3.56678269e-03 -3.41633081e-01
-4.90785003e-01 -2.19868094e-01 7.27978766e-01 -9.71947670e-01
9.59061921e-01 -1.89185929e+00 7.85177708e-01 -2.81471223e-01
-4.50441152e-01 1.09509595e-01 -5.19233227e-01 7.51276374e-01
9.23814625e-02 6.88400492e-02 -1.23944372e-01 -1.07153642e+00
8.69091600e-02 3.16513002e-01 -7.95457542e-01 2.37067565e-01
5.37866771e-01 1.38115048e+00 -7.07565486e-01 -3.86112034e-01
8.91119540e-02 5.32804906e-01 -5.02023041e-01 6.77072883e-01
-5.69264412e-01 5.33703864e-01 -2.27228701e-01 6.35667920e-01
7.05884211e-03 1.48844847e-03 1.20166190e-01 3.82474452e-01
-1.67051435e-01 9.57733333e-01 -5.12040496e-01 1.89559567e+00
-9.12045836e-01 6.17084920e-01 1.03405938e-01 -8.85000944e-01
7.16385186e-01 9.85449255e-01 -1.97730567e-02 -7.32017636e-01
5.46873584e-02 3.58573288e-01 -1.55187815e-01 -2.54857957e-01
4.34728444e-01 -1.65122956e-01 -1.24541065e-02 5.13173580e-01
4.07819957e-01 -3.20160985e-01 -3.02470833e-01 1.56478181e-01
6.76177442e-01 6.97175682e-01 1.86067164e-01 -1.56872556e-01
2.69941241e-01 3.77591141e-02 3.03246707e-01 5.99476457e-01
-2.33696215e-02 6.35778010e-01 2.47170553e-01 1.36417314e-01
-1.34640312e+00 -1.28164244e+00 5.23672044e-01 1.43372035e+00
-3.96463722e-01 -2.99714804e-01 -9.26039517e-01 -5.39514065e-01
-2.09047824e-01 9.48872447e-01 -3.94008577e-01 -2.67020110e-02
-7.29596376e-01 -2.43647680e-01 5.61421752e-01 4.33393538e-01
-9.64401811e-02 -1.25280404e+00 -1.88031241e-01 4.90581274e-01
-2.99326777e-01 -1.25049806e+00 -1.00420642e+00 2.63088644e-01
-9.49669480e-01 -5.41670993e-02 -9.05546486e-01 -8.40329111e-01
2.73706228e-01 1.65808737e-01 1.11103296e+00 -4.42344218e-01
3.44018877e-01 1.08679004e-01 -3.74943405e-01 -2.24022031e-01
-1.16793728e+00 3.70694578e-01 3.08363974e-01 1.46687016e-01
-8.71842727e-02 -5.22373736e-01 -1.89854816e-01 -7.74614587e-02
-5.19370973e-01 3.47662449e-01 8.82939160e-01 1.10618341e+00
7.50370443e-01 -9.34138238e-01 6.23069048e-01 -4.47500795e-01
8.88127923e-01 -6.49562716e-01 -5.72400570e-01 2.32996449e-01
-5.28807998e-01 4.07644272e-01 9.10895228e-01 -5.41401982e-01
-1.04293060e+00 1.21681564e-01 -1.12708114e-01 -7.79480696e-01
9.56325904e-02 2.76295394e-01 -3.03631332e-02 5.59037149e-01
3.58675539e-01 7.50067592e-01 9.38629508e-02 -4.34161514e-01
7.49236584e-01 1.01326442e+00 4.66479748e-01 -5.59390545e-01
8.31855834e-01 -1.18731044e-01 -4.29016531e-01 -7.57355034e-01
-2.56474048e-01 -2.98997983e-02 -5.78352511e-01 3.05712700e-01
1.02347708e+00 -1.21531868e+00 -2.57621706e-01 1.42792434e-01
-1.62732255e+00 -9.12980616e-01 -1.76624760e-01 5.30761063e-01
-9.61746037e-01 -1.91632472e-02 -9.81236577e-01 -8.73665690e-01
-6.48217618e-01 -1.53448439e+00 1.61895072e+00 -2.57552087e-01
-1.69142947e-01 -9.04998958e-01 4.15808614e-03 1.73382580e-01
5.33490896e-01 -2.44636387e-01 9.42680895e-01 -7.52016425e-01
-7.13964462e-01 3.86934072e-01 1.22948818e-01 4.18624192e-01
2.12262020e-01 1.62916817e-02 -9.71527100e-01 -5.57452321e-01
-3.17260265e-01 -3.90286446e-01 6.65965199e-01 4.18037832e-01
3.61674577e-01 -6.43702269e-01 -1.95460506e-02 6.45972490e-01
1.08499408e+00 1.37194723e-01 1.84491858e-01 -3.18504572e-01
5.62878132e-01 2.83861220e-01 3.49388570e-01 2.99068205e-02
2.95361966e-01 8.61840248e-01 3.08152344e-02 -1.74849242e-01
-6.45311475e-01 -5.96242666e-01 1.37529087e+00 1.67796957e+00
9.80634838e-02 -5.72287917e-01 -5.38100064e-01 5.21593630e-01
-1.75330150e+00 -8.17206383e-01 3.12484801e-01 2.02685952e+00
9.46990907e-01 -3.70801799e-02 4.86115254e-02 -3.58674645e-01
5.59424877e-01 1.06972866e-01 -4.10423398e-01 -9.80920494e-01
-2.33813096e-02 5.25407672e-01 3.55254441e-01 1.21724236e+00
-6.40709639e-01 1.67756331e+00 6.85054541e+00 8.10233891e-01
-1.30240273e+00 6.77860498e-01 6.09705329e-01 -3.16611826e-01
-4.90605563e-01 2.05371846e-02 -8.09763849e-01 3.78725022e-01
1.75496078e+00 -1.96700066e-01 9.58273411e-01 5.14818966e-01
7.45956004e-01 6.34505332e-01 -1.35382330e+00 7.19988585e-01
-9.22927633e-02 -1.47665775e+00 3.37141186e-01 1.25366673e-01
9.34614837e-01 3.67131442e-01 2.24519864e-01 5.64115882e-01
7.30169177e-01 -1.02154195e+00 1.17587483e+00 1.94975272e-01
1.24675286e+00 -7.11550117e-01 1.73897445e-01 3.98707509e-01
-1.16143465e+00 9.80975106e-02 -2.14613706e-01 2.59483159e-02
4.04106677e-01 -8.22292194e-02 -1.15562844e+00 5.54665327e-01
1.11983143e-01 5.43587327e-01 1.30859539e-01 -2.93205008e-02
-2.83371061e-01 9.92714942e-01 -7.01086968e-02 -1.87863603e-01
7.23207295e-01 -1.69818282e-01 6.41055882e-01 1.38366640e+00
5.81331074e-01 -1.06411166e-01 3.88748080e-01 9.36731756e-01
-2.70894498e-01 2.92920649e-01 -6.76046789e-01 -3.88883412e-01
6.50757790e-01 7.85515130e-01 -2.44615793e-01 -7.13466704e-01
-5.76273978e-01 1.64897537e+00 3.34321499e-01 7.55546272e-01
-8.32833707e-01 6.92563727e-02 8.01037729e-01 -1.63559228e-01
5.19353688e-01 -4.81794953e-01 2.66956855e-02 -1.25412250e+00
-1.62818283e-01 -1.26526296e+00 -2.34341383e-01 -7.54982114e-01
-8.95220399e-01 9.79606926e-01 -2.86033839e-01 -1.03285718e+00
-8.98246646e-01 -2.98095524e-01 -5.03107786e-01 1.38999903e+00
-1.53373742e+00 -1.45020092e+00 5.21945536e-01 6.52540147e-01
1.49710357e+00 -5.70369720e-01 1.09717059e+00 -1.54708028e-02
-4.74606603e-01 7.02331364e-01 3.60442787e-01 -4.53653708e-02
6.70157909e-01 -1.11726582e+00 1.26001680e+00 9.39601243e-01
6.47283852e-01 3.86375546e-01 6.08161628e-01 -6.15418017e-01
-1.65810943e+00 -1.21524155e+00 1.15058708e+00 -5.56928933e-01
5.43964028e-01 -7.55995214e-01 -4.00879353e-01 1.02121913e+00
8.78013909e-01 -1.72041178e-01 3.19514871e-01 -4.59835470e-01
-7.35597759e-02 2.06420824e-01 -7.17708707e-01 6.21408999e-01
1.01437652e+00 -9.94567871e-01 -5.70648670e-01 4.79065746e-01
1.48867702e+00 -5.09099901e-01 -6.15060210e-01 -1.79358780e-01
2.82474726e-01 -3.98807019e-01 7.42320001e-01 -1.02107751e+00
6.07336879e-01 4.76712734e-02 -6.58135831e-01 -1.76595795e+00
-1.09721549e-01 -1.44748938e+00 -3.81475270e-01 1.04124975e+00
1.03701532e+00 -5.29935300e-01 2.49859363e-01 -2.35853150e-01
-6.40314043e-01 -6.84388518e-01 -1.11562037e+00 -7.50537395e-01
4.25556213e-01 -1.41570866e-01 6.72383070e-01 5.54903328e-01
-2.86334127e-01 8.24871182e-01 -7.62589574e-01 3.49352598e-01
2.29580432e-01 1.47275059e-02 6.24640048e-01 -4.34948713e-01
-8.86650443e-01 -2.73663044e-01 2.43650898e-01 -1.51981640e+00
3.98486286e-01 -1.23129153e+00 3.16753536e-01 -1.31317282e+00
-1.85653090e-01 1.40787080e-01 -1.80665985e-01 3.04690719e-01
-1.32830560e-01 -8.43956992e-02 2.40561768e-01 2.83081174e-01
-1.95375327e-02 9.65194106e-01 1.37019622e+00 -1.34883553e-01
-3.39532703e-01 5.90883456e-02 -3.48711133e-01 -1.81370880e-03
5.78746140e-01 -5.63319504e-01 -6.93156540e-01 -8.77383411e-01
-3.26825500e-01 7.95971870e-01 -9.73857120e-02 -5.23545802e-01
1.05172284e-01 -2.72382647e-01 1.04657253e-02 -2.37432301e-01
6.44692063e-01 -3.29862803e-01 -1.54840082e-01 3.77696067e-01
-7.63704836e-01 5.47462940e-01 -7.55530745e-02 4.94842529e-01
-2.42770687e-01 3.64489883e-01 6.40484333e-01 -1.48573160e-01
-1.34371489e-01 3.74926567e-01 -6.30113959e-01 -1.51413828e-01
6.83248281e-01 2.32925966e-01 7.96233192e-02 -7.29444563e-01
-9.85865951e-01 -1.36799037e-01 5.84209561e-02 7.92187393e-01
6.32192135e-01 -1.52813947e+00 -1.22431004e+00 4.36082006e-01
-2.76382923e-01 -2.99975157e-01 -4.56334680e-01 8.37510526e-01
-2.03146785e-01 6.79301023e-01 2.30860576e-01 -5.85125625e-01
-1.12443733e+00 5.21209061e-01 3.19437116e-01 -1.10878818e-01
-5.57347000e-01 9.93079960e-01 -1.07592260e-02 -5.57636023e-01
1.56995863e-01 -6.20105863e-01 4.94986475e-01 -1.54722124e-01
2.64180034e-01 1.52639970e-01 1.36224985e-01 -9.11219001e-01
-1.36224329e-01 1.15630239e-01 -9.76325362e-04 -1.07554722e+00
1.09783006e+00 -3.85913789e-01 2.35853150e-01 6.38744950e-01
1.22751594e+00 5.31884357e-02 -1.33157301e+00 -3.91282439e-01
-2.03917205e-01 -8.46796036e-02 1.01178214e-01 -8.54247987e-01
-8.22135568e-01 1.15695655e+00 3.92064214e-01 -1.22564740e-01
6.39679909e-01 -1.45823643e-01 1.09584212e+00 3.84582192e-01
9.55251530e-02 -1.05217326e+00 2.23247305e-01 6.47719502e-01
1.03017318e+00 -1.18208838e+00 -6.50345087e-01 -3.45121510e-02
-9.63340402e-01 1.03637421e+00 1.30122855e-01 -7.23913983e-02
2.90402859e-01 5.23673475e-01 4.34970140e-01 4.54133540e-01
-1.45142806e+00 -1.96119875e-01 2.64361084e-01 4.68304276e-01
7.28335619e-01 3.45856190e-01 6.68487772e-02 3.27825129e-01
-4.12896216e-01 -2.39196792e-01 2.31981531e-01 6.26581788e-01
-3.34489703e-01 -1.31365299e+00 -3.33333880e-01 -2.67023351e-02
-3.95014793e-01 -6.73024416e-01 -5.48631489e-01 3.41220856e-01
-4.03491586e-01 1.20262241e+00 5.23288958e-02 -2.81519592e-01
1.60453096e-01 4.84263957e-01 3.61375928e-01 -8.38272154e-01
-7.69355237e-01 5.78314722e-01 2.53717273e-01 -6.25742376e-01
8.65084305e-02 -6.98644817e-01 -1.21881759e+00 -2.42375076e-01
-2.52314478e-01 1.79731235e-01 8.22246194e-01 1.06666434e+00
6.36356592e-01 6.89050198e-01 9.59294677e-01 -1.07823336e+00
-1.19297957e+00 -1.23135734e+00 7.83349052e-02 1.61668822e-01
9.34129655e-01 -1.39581278e-01 -1.36925235e-01 2.27910981e-01] | [14.550394058227539, 7.110716342926025] |
e9cd64a6-0845-4244-8288-12a816db636b | flat-latent-manifolds-for-music-improvisation | 2202.12243 | null | https://arxiv.org/abs/2202.12243v3 | https://arxiv.org/pdf/2202.12243v3.pdf | Flat Latent Manifolds for Human-machine Co-creation of Music | The use of machine learning in artistic music generation leads to controversial discussions of the quality of art, for which objective quantification is nonsensical. We therefore consider a music-generating algorithm as a counterpart to a human musician, in a setting where reciprocal interplay is to lead to new experiences, both for the musician and the audience. To obtain this behaviour, we resort to the framework of recurrent Variational Auto-Encoders (VAE) and learn to generate music, seeded by a human musician. In the learned model, we generate novel musical sequences by interpolation in latent space. Standard VAEs however do not guarantee any form of smoothness in their latent representation. This translates into abrupt changes in the generated music sequences. To overcome these limitations, we regularise the decoder and endow the latent space with a flat Riemannian manifold, i.e., a manifold that is isometric to the Euclidean space. As a result, linearly interpolating in the latent space yields realistic and smooth musical changes that fit the type of machine--musician interactions we aim for. We provide empirical evidence for our method via a set of experiments on music datasets and we deploy our model for an interactive jam session with a professional drummer. The live performance provides qualitative evidence that the latent representation can be intuitively interpreted and exploited by the drummer to drive the interplay. Beyond the musical application, our approach showcases an instance of human-centred design of machine-learning models, driven by interpretability and the interaction with the end user. | ['Patrick van der Smagt', 'Luciano Pinna', 'Mathis Nitschke', 'Francesco Ferroni', 'Djalel Benbouzid', 'Nutan Chen'] | 2022-02-23 | null | null | null | null | ['music-generation', 'music-generation'] | ['audio', 'music'] | [ 2.89178878e-01 6.47243679e-01 2.99317569e-01 1.02051713e-01
-4.18501616e-01 -8.12692761e-01 1.01716959e+00 -4.90682214e-01
-2.37650815e-02 4.75580752e-01 4.92921323e-01 1.51418373e-01
-2.96313584e-01 -7.68224061e-01 -8.71348739e-01 -8.44821215e-01
9.02065448e-03 3.87108296e-01 -4.38130438e-01 -4.60999042e-01
1.50497109e-01 2.15468317e-01 -1.60420632e+00 4.59371835e-01
5.42961895e-01 3.48969549e-01 1.92462876e-01 9.28651989e-01
5.08510433e-02 5.16138017e-01 -4.79341149e-01 -6.24431193e-01
3.47685188e-01 -8.68442893e-01 -7.68228292e-01 3.09606701e-01
-1.22534432e-01 2.53500938e-01 2.47086547e-02 8.36365819e-01
3.36065114e-01 4.02356423e-02 9.01698112e-01 -9.39591527e-01
-6.32417142e-01 1.14810824e+00 -1.48434103e-01 -5.19967198e-01
3.95440459e-01 2.62167007e-01 1.31524754e+00 -7.11248338e-01
8.61499488e-01 1.06076431e+00 5.98269641e-01 7.52524853e-01
-1.76816273e+00 -1.48198068e-01 -5.60421765e-01 -2.74832040e-01
-1.11540830e+00 -5.46860337e-01 1.02392280e+00 -6.70163095e-01
8.00240785e-02 6.71027660e-01 1.00312376e+00 1.46764719e+00
-1.19849883e-01 7.90332556e-01 9.04631972e-01 -5.94046712e-01
3.17406178e-01 3.62218261e-01 -5.44299543e-01 1.00108169e-01
-2.20647156e-01 4.15888846e-01 -5.88142335e-01 2.62323581e-03
1.05999637e+00 -3.64008278e-01 -2.66786754e-01 -6.04176044e-01
-1.60068607e+00 7.89332151e-01 2.31332615e-01 8.60826194e-01
-6.01243138e-01 3.63127828e-01 2.84088880e-01 5.15185714e-01
2.01336205e-01 8.88977468e-01 -1.04692176e-01 -4.49476361e-01
-1.12589836e+00 4.81950313e-01 8.89001667e-01 5.23591995e-01
3.25948626e-01 1.96252555e-01 -1.48976490e-01 5.41490376e-01
3.60233188e-01 1.04300261e-01 5.06942689e-01 -1.30137479e+00
-6.81031123e-02 4.24812019e-01 1.90564871e-01 -9.18674529e-01
2.19037803e-03 -7.74699390e-01 -7.72572637e-01 5.10739982e-01
5.67171752e-01 -1.13346223e-02 -1.17180333e-01 1.93699336e+00
5.37855737e-02 3.28802109e-01 1.42136157e-01 1.09807193e+00
4.16403450e-02 4.96320128e-01 -2.26528898e-01 -2.13063464e-01
1.06814671e+00 -5.02351642e-01 -5.91809809e-01 3.47429901e-01
6.51285052e-01 -8.49639297e-01 1.59205174e+00 7.20127106e-01
-1.55986130e+00 -7.13262916e-01 -9.86665905e-01 2.07508370e-01
1.30974561e-01 -4.16575707e-02 4.65084046e-01 5.96465766e-01
-9.45839703e-01 1.14324272e+00 -4.98842537e-01 -1.31553501e-01
3.93033214e-02 8.59057307e-02 -1.03007644e-01 8.00127387e-01
-1.08862960e+00 5.72347283e-01 2.72961229e-01 2.04110309e-01
-7.47589886e-01 -6.59792125e-01 -3.62581342e-01 -1.08394012e-01
-1.27304435e-01 -1.13535440e+00 1.27536511e+00 -1.68525290e+00
-1.97685695e+00 1.12586224e+00 3.28708023e-01 -4.25661385e-01
1.05073452e+00 -7.06051514e-02 -1.34086832e-01 -2.50714332e-01
-1.06505975e-01 5.01065433e-01 1.12625396e+00 -1.56638944e+00
-1.50618538e-01 -1.10644766e-03 1.22621834e-01 7.21190944e-02
-1.42888129e-01 -3.03632915e-01 -2.01613400e-02 -9.63186741e-01
-1.22541316e-01 -1.03503835e+00 -1.79887027e-01 -3.25889647e-01
-6.02847517e-01 5.46448268e-02 3.33761811e-01 -3.01289439e-01
1.30597806e+00 -2.40824151e+00 9.55327332e-01 3.27005535e-01
1.49344340e-01 -1.47037134e-01 -2.70608105e-02 6.25456572e-01
-2.93233335e-01 8.79666135e-02 -3.06825578e-01 -6.47680461e-01
5.00398099e-01 1.72715232e-01 -6.33134425e-01 4.29743946e-01
-1.62670404e-01 1.04931557e+00 -8.25483441e-01 -8.34430158e-02
5.61088100e-02 6.30398870e-01 -9.05957460e-01 3.11556548e-01
-5.11981428e-01 1.11849153e+00 -4.13761765e-01 -6.38607666e-02
3.47687155e-02 -1.04865439e-01 9.99238864e-02 -6.65156320e-02
-2.99329758e-01 2.24133790e-01 -1.30844939e+00 2.25057769e+00
-6.53489947e-01 4.73636448e-01 1.97717354e-01 -8.65265012e-01
8.88513386e-01 5.62204659e-01 3.83343905e-01 -2.63734937e-01
2.63001323e-01 4.17738348e-01 1.81357637e-01 -5.12242973e-01
3.80283028e-01 -6.25450909e-01 -7.60901570e-02 8.02461088e-01
-1.10951379e-01 -2.94252932e-01 -1.11914240e-01 -1.98944332e-03
8.32251072e-01 6.61250174e-01 -2.84502488e-02 -3.71459872e-01
3.66339743e-01 -3.19579333e-01 -1.85145997e-02 5.41370809e-01
4.18911606e-01 6.05577111e-01 5.50982416e-01 -2.76006252e-01
-1.44508517e+00 -1.20772338e+00 -4.28957082e-02 9.64120030e-01
-3.52252990e-01 -4.31648403e-01 -1.08206844e+00 -5.43726832e-02
-3.39971721e-01 1.01245916e+00 -6.99846864e-01 -3.08346510e-01
-5.35167873e-01 -2.17761427e-01 5.68079650e-01 -2.64058590e-01
-7.48909488e-02 -1.42592847e+00 -8.78982663e-01 4.26387548e-01
-2.34720841e-01 -5.39939702e-01 -5.11579037e-01 -1.94993779e-01
-9.38985586e-01 -6.69003427e-01 -8.17709506e-01 -4.72311318e-01
2.14269325e-01 -4.50301141e-01 1.28361058e+00 -1.97398394e-01
-2.35295415e-01 6.11874282e-01 -2.79952914e-01 -1.82976618e-01
-1.10940158e+00 -1.34670530e-02 1.87648997e-01 5.49660206e-01
-2.06804991e-01 -1.32438767e+00 -6.94266915e-01 1.61522254e-01
-1.19756615e+00 5.87710500e-01 4.95112151e-01 6.90609634e-01
2.15604052e-01 -2.14181036e-01 4.23450053e-01 -6.27759874e-01
7.44452655e-01 -4.02425587e-01 -3.42965633e-01 -1.71405569e-01
-2.62977242e-01 4.86247480e-01 4.26626801e-01 -6.76252246e-01
-7.55155146e-01 1.42300474e-02 -1.65610528e-03 -3.48950446e-01
-1.16732903e-01 4.50421393e-01 -9.99215692e-02 3.27471465e-01
1.04572964e+00 2.05442369e-01 4.04119194e-02 -6.10554278e-01
1.00834370e+00 5.25095761e-01 6.97255135e-01 -7.49412775e-01
9.66525018e-01 4.45672035e-01 2.11433228e-03 -6.92385018e-01
-4.54831332e-01 1.33811682e-01 -7.87292242e-01 -4.55697417e-01
8.38816702e-01 -5.14346659e-01 -1.07494581e+00 9.86906439e-02
-1.26756787e+00 -6.19330525e-01 -1.12697804e+00 4.36396182e-01
-1.32081079e+00 9.60796475e-02 -4.48406994e-01 -1.07232130e+00
-1.70262769e-01 -9.26546276e-01 1.10841048e+00 -2.28066236e-01
-7.79110551e-01 -1.12498546e+00 3.77704948e-01 1.29908085e-01
3.40535015e-01 6.20426834e-01 8.83296311e-01 -1.58272281e-01
-5.75824380e-01 9.09773912e-03 5.60666084e-01 2.15404063e-01
1.18471943e-01 5.10948990e-03 -1.10340798e+00 1.26201853e-01
2.43746713e-01 -7.80109465e-02 4.76032645e-01 2.15165660e-01
7.37393320e-01 -5.67521334e-01 1.82739183e-01 4.51422244e-01
1.08406055e+00 -2.06099331e-01 7.17154741e-01 2.51972437e-01
6.33683324e-01 1.03907287e+00 3.97928096e-02 6.09088063e-01
-3.04447226e-02 1.05100882e+00 3.28976244e-01 1.17322616e-02
-2.42350660e-02 -5.29220402e-01 7.05352902e-01 9.81879532e-01
-5.00437379e-01 2.44371906e-01 -6.14961982e-01 4.45952266e-01
-1.89317536e+00 -1.21579516e+00 -2.68478096e-01 2.32756972e+00
1.15750873e+00 1.24080829e-01 4.11481559e-01 4.36922014e-01
5.05540252e-01 1.40133463e-02 -7.96818733e-02 -5.18004179e-01
-4.92327511e-02 2.28877753e-01 -5.01011163e-02 5.67864776e-01
-5.98022759e-01 7.57750392e-01 5.91376972e+00 7.56886899e-01
-1.09707737e+00 1.45286605e-01 2.69862115e-01 -2.67097384e-01
-7.52733707e-01 1.22330531e-01 -3.17138769e-02 3.82524699e-01
1.05530465e+00 -1.57884717e-01 9.77439940e-01 4.87936229e-01
7.04886317e-01 4.84336317e-01 -1.47253740e+00 9.77151513e-01
-2.65806347e-01 -1.51549530e+00 8.80381167e-02 3.28638107e-01
5.96901715e-01 -5.14486492e-01 4.15552288e-01 1.57034382e-01
1.55462191e-01 -1.28839803e+00 1.25731599e+00 1.04587889e+00
7.56378651e-01 -5.55802584e-01 3.92208956e-02 5.41890144e-01
-8.02447081e-01 8.60735178e-02 1.75602838e-01 -2.60579824e-01
3.73606741e-01 4.44546282e-01 -5.38722515e-01 3.13742101e-01
9.84104052e-02 6.02674901e-01 -1.40262261e-01 5.91822565e-01
-2.63844635e-02 7.71980643e-01 -8.81797969e-02 2.29840294e-01
2.02382460e-01 -7.26374388e-01 1.20563447e+00 1.02216995e+00
5.27584016e-01 -2.03929529e-01 -3.60181272e-01 1.60159683e+00
1.70578450e-01 2.89909273e-01 -7.89935827e-01 -3.22525263e-01
-2.21389204e-01 1.18445730e+00 -4.16509479e-01 -2.76117325e-02
2.00526237e-01 1.20769942e+00 -5.00008576e-02 3.48101795e-01
-7.41674781e-01 -2.37467177e-02 5.67746341e-01 4.15421724e-01
-6.23841882e-02 -2.64332205e-01 -5.13263345e-01 -1.20794392e+00
3.70859704e-03 -1.06182075e+00 -2.03670070e-01 -9.19837475e-01
-1.16313601e+00 4.98715580e-01 -2.28642583e-01 -1.27369940e+00
-6.69864595e-01 -1.97592348e-01 -7.44374216e-01 8.29261661e-01
-7.05722809e-01 -1.22489047e+00 1.42558441e-01 4.41432327e-01
3.52103412e-01 -9.61717740e-02 9.84368801e-01 2.26258695e-01
-3.75703350e-02 3.31219107e-01 -5.13321348e-02 -1.49796262e-01
3.71017814e-01 -1.49731445e+00 4.00603861e-01 3.87065768e-01
8.76049280e-01 7.03400850e-01 1.32553256e+00 -2.51070499e-01
-1.46063173e+00 -8.09717357e-01 8.44099641e-01 -7.57621408e-01
8.85498524e-01 -5.24752855e-01 -7.67353773e-01 4.66513336e-01
2.21415594e-01 -5.67562521e-01 7.27006018e-01 9.81511027e-02
-1.28584325e-01 2.62071550e-01 -6.22394979e-01 1.03246677e+00
1.20020068e+00 -6.69914186e-01 -6.73319936e-01 1.79101214e-01
6.08500481e-01 -7.93449432e-02 -8.23880613e-01 9.52526107e-02
7.01471686e-01 -1.14698267e+00 8.92424643e-01 -6.18614554e-01
5.84435761e-01 -3.14429849e-01 -7.90482387e-02 -1.33645844e+00
-2.01739118e-01 -1.27554238e+00 -5.85751757e-02 1.23705375e+00
3.33647370e-01 -1.42691374e-01 6.47555292e-01 3.17489266e-01
-1.36330768e-01 -4.75255191e-01 -8.18864822e-01 -5.18594861e-01
1.66121811e-01 -8.11849475e-01 5.47912955e-01 9.46269989e-01
1.94692001e-01 3.56766373e-01 -5.92124581e-01 -3.14861566e-01
6.26195908e-01 1.69174895e-01 1.01094079e+00 -1.30763721e+00
-1.08160317e+00 -8.97247791e-01 -2.78113037e-01 -9.12390411e-01
8.63620266e-02 -1.19573927e+00 -1.54300764e-01 -9.32393372e-01
-1.36402622e-01 -3.30252677e-01 -8.04353580e-02 -1.52463853e-01
3.78050417e-01 3.25804681e-01 3.64420861e-01 5.05670786e-01
-1.45100862e-01 6.24231398e-01 1.39772820e+00 9.59875211e-02
-5.75858295e-01 2.14642793e-01 -6.49674058e-01 7.58152544e-01
5.80866873e-01 -3.92444015e-01 -3.07833999e-01 -4.22271267e-02
9.43515778e-01 2.51701087e-01 7.26635396e-01 -7.40184605e-01
-1.43725559e-01 -7.54939914e-02 -4.72926162e-02 1.69909239e-01
3.61881793e-01 -7.73542523e-01 9.39820647e-01 4.95153040e-01
-7.42560565e-01 -4.56529409e-01 -2.16218516e-01 3.67509335e-01
-9.73435789e-02 -2.67852843e-01 6.21635675e-01 -6.69916719e-02
-1.63326424e-03 -3.03192083e-02 -2.87945569e-01 -8.98756236e-02
6.89400375e-01 -3.15957665e-01 4.88999814e-01 -5.68369627e-01
-1.29234719e+00 -4.65439349e-01 4.92219627e-01 4.19591755e-01
1.72392577e-01 -1.56515360e+00 -9.59102273e-01 2.01250643e-01
-5.51695675e-02 -3.70459318e-01 1.52167872e-01 8.58762264e-01
-2.53932029e-01 2.74447799e-02 -8.50840360e-02 -6.48575723e-01
-8.73989701e-01 4.48637277e-01 4.93182123e-01 -1.56055808e-01
-8.00162911e-01 4.60024655e-01 3.12904716e-01 -2.56952882e-01
9.51077044e-02 -2.34796464e-01 -1.79253861e-01 2.14385137e-01
1.35888085e-01 1.45698890e-01 -4.80166674e-01 -7.11154699e-01
3.34036440e-01 5.56214154e-01 7.40855813e-01 -7.55524874e-01
1.27202177e+00 -5.18138967e-02 -1.64314002e-01 1.18022263e+00
9.47456419e-01 4.31456566e-01 -1.22122157e+00 9.67633352e-02
1.85227484e-01 -2.67560810e-01 -3.07187140e-01 -4.73094225e-01
-6.60650134e-01 8.76360953e-01 4.60378617e-01 5.73082745e-01
7.30456769e-01 4.04254682e-02 4.98633295e-01 6.24227636e-02
3.84193629e-01 -9.12461638e-01 3.05227548e-01 1.11886814e-01
1.41391480e+00 -6.26453161e-01 -6.23130322e-01 2.19623938e-01
-6.61631167e-01 1.13180947e+00 -2.73336560e-01 -3.73146802e-01
5.29033065e-01 -3.29743809e-04 6.89816922e-02 -1.49690226e-01
-6.45780504e-01 2.77955700e-02 4.85597908e-01 3.49613786e-01
6.16072536e-01 3.81623715e-01 -3.51557702e-01 8.48393261e-01
-9.81847405e-01 1.14172280e-01 5.75767219e-01 2.63261646e-01
-2.01738819e-01 -1.32875586e+00 -4.49621767e-01 -2.08805904e-01
-4.23352748e-01 3.53961587e-02 -3.67293656e-01 4.66149330e-01
3.38902950e-01 7.92810559e-01 -5.59172332e-02 -3.32570732e-01
3.11507374e-01 3.17190170e-01 5.96166492e-01 -5.26237428e-01
-7.49153912e-01 4.05757636e-01 -1.19336568e-01 -4.91805226e-01
-5.24652243e-01 -6.99430287e-01 -1.08548570e+00 -3.05286974e-01
-2.31161178e-03 4.52146649e-01 7.96933591e-01 7.53083348e-01
1.66519612e-01 7.16199338e-01 6.89254880e-01 -1.17662621e+00
-5.92352808e-01 -8.83319855e-01 -8.93482804e-01 7.77809083e-01
2.51228929e-01 -3.16541374e-01 -5.11557877e-01 5.03083706e-01] | [15.854843139648438, 5.608773708343506] |
0fac4010-87cc-4353-99c3-d61f41dda9dd | graphsaint-graph-sampling-based-inductive | 1907.04931 | null | https://arxiv.org/abs/1907.04931v4 | https://arxiv.org/pdf/1907.04931v4.pdf | GraphSAINT: Graph Sampling Based Inductive Learning Method | Graph Convolutional Networks (GCNs) are powerful models for learning representations of attributed graphs. To scale GCNs to large graphs, state-of-the-art methods use various layer sampling techniques to alleviate the "neighbor explosion" problem during minibatch training. We propose GraphSAINT, a graph sampling based inductive learning method that improves training efficiency and accuracy in a fundamentally different way. By changing perspective, GraphSAINT constructs minibatches by sampling the training graph, rather than the nodes or edges across GCN layers. Each iteration, a complete GCN is built from the properly sampled subgraph. Thus, we ensure fixed number of well-connected nodes in all layers. We further propose normalization technique to eliminate bias, and sampling algorithms for variance reduction. Importantly, we can decouple the sampling from the forward and backward propagation, and extend GraphSAINT with many architecture variants (e.g., graph attention, jumping connection). GraphSAINT demonstrates superior performance in both accuracy and training time on five large graphs, and achieves new state-of-the-art F1 scores for PPI (0.995) and Reddit (0.970). | ['Ajitesh Srivastava', 'Viktor Prasanna', 'Rajgopal Kannan', 'Hongkuan Zhou', 'Hanqing Zeng'] | 2019-07-10 | null | https://openreview.net/forum?id=BJe8pkHFwS | https://openreview.net/pdf?id=BJe8pkHFwS | iclr-2020-1 | ['graph-sampling'] | ['graphs'] | [-9.36560705e-02 5.95906854e-01 -4.09577191e-01 -4.90505248e-01
-5.91731310e-01 -5.61367452e-01 3.62594336e-01 2.17432261e-01
2.68738549e-02 9.25737679e-01 -6.02085143e-03 -2.25435749e-01
-2.24328369e-01 -1.15300786e+00 -1.14254248e+00 -7.03218758e-01
-4.54907984e-01 5.02344429e-01 3.30421060e-01 5.28494157e-02
-2.32914705e-02 5.56997836e-01 -8.58602464e-01 4.99606729e-02
9.50700641e-01 6.48207784e-01 6.82414621e-02 8.08414698e-01
-2.61980563e-01 9.29028034e-01 -6.56311572e-01 -6.26833916e-01
-4.69586253e-02 -2.23401889e-01 -8.33566606e-01 -2.32714504e-01
7.06952751e-01 1.30443899e-02 -7.82308161e-01 1.06808734e+00
6.07461214e-01 8.19189474e-02 6.33850515e-01 -1.46979094e+00
-1.13754559e+00 1.34532726e+00 -8.19849908e-01 1.90660447e-01
-1.47798851e-01 -1.43128961e-01 1.24972534e+00 -5.34186840e-01
5.15594542e-01 1.29884005e+00 1.00639272e+00 3.93649459e-01
-1.29073393e+00 -7.74849474e-01 7.41225421e-01 -8.21783766e-02
-1.33961523e+00 4.72626388e-02 7.25660026e-01 -1.60149142e-01
1.08927822e+00 2.28661388e-01 6.04972124e-01 1.09928977e+00
4.40443568e-02 8.18460524e-01 5.75179636e-01 -1.15403950e-01
-2.34077871e-02 -2.00705156e-01 5.34223616e-01 9.95143414e-01
6.65007651e-01 -2.36667380e-01 -2.52830446e-01 -5.25309481e-02
9.55625594e-01 3.09844278e-02 -2.97862798e-01 -2.40380332e-01
-7.95219541e-01 7.66217470e-01 1.18294048e+00 -4.77238335e-02
-2.63485759e-01 6.24405324e-01 5.48794270e-01 2.55209178e-01
4.21118259e-01 2.11252674e-01 -7.71859884e-01 3.68495345e-01
-6.35124624e-01 2.21093111e-02 8.90156090e-01 1.45367515e+00
9.17231023e-01 2.55465060e-01 -4.77844208e-01 8.80213857e-01
3.34469140e-01 1.84950605e-01 2.77665019e-01 -3.67514968e-01
4.79305863e-01 9.61978853e-01 -5.27815819e-01 -9.87965882e-01
-3.83589655e-01 -1.01715744e+00 -1.18723607e+00 -3.90669227e-01
9.88873467e-02 -2.64697194e-01 -1.38859248e+00 1.85254014e+00
2.81797677e-01 3.93554628e-01 -2.14244545e-01 5.92924595e-01
1.35377502e+00 5.52221596e-01 1.82961896e-01 7.06446543e-02
1.09188759e+00 -1.19785774e+00 -4.20106739e-01 -3.37034613e-01
7.06950545e-01 -1.02518439e-01 1.09952331e+00 3.34251612e-01
-7.77247548e-01 -4.20917571e-01 -1.05441380e+00 -2.94527143e-01
-5.42887151e-01 2.40118578e-01 1.02334070e+00 4.30492699e-01
-1.15699947e+00 1.08000410e+00 -7.51316369e-01 -3.36219996e-01
7.52812743e-01 5.75078607e-01 -4.56840217e-01 -1.12594195e-01
-1.08149087e+00 3.03630799e-01 4.86726671e-01 7.34686777e-02
-1.03506815e+00 -7.31373370e-01 -9.05312121e-01 4.19549823e-01
3.64618659e-01 -5.80651104e-01 7.95599878e-01 -8.91034961e-01
-1.13027835e+00 8.27720821e-01 5.56663349e-02 -7.04603136e-01
7.04045147e-02 -1.81910083e-01 -2.23551616e-01 -2.20356092e-01
-3.80079225e-02 6.42952442e-01 5.62888563e-01 -8.98296893e-01
-2.21879646e-01 -4.27556634e-01 4.31711748e-02 2.99729090e-02
-3.02180588e-01 -3.42103928e-01 -8.19362402e-01 -5.66527665e-01
2.94285379e-02 -8.64187658e-01 -1.89712867e-01 -3.49115312e-01
-8.81223798e-01 -5.05666673e-01 6.24250531e-01 -6.47803247e-01
1.40852273e+00 -1.85304022e+00 3.87149937e-02 2.73578703e-01
8.32141936e-01 3.25915992e-01 -1.85369864e-01 3.97486418e-01
-2.98399806e-01 3.83079261e-01 -4.40871119e-02 -1.52471274e-01
6.31348579e-04 3.29379529e-01 1.39560476e-01 5.11105895e-01
3.37142408e-01 1.21615624e+00 -9.63469625e-01 -3.06529194e-01
-1.46043003e-01 4.58445460e-01 -5.93871176e-01 1.88809797e-01
-3.47084373e-01 6.35049567e-02 -4.22945380e-01 5.15397787e-01
9.21513975e-01 -8.08233738e-01 4.32385981e-01 -1.66006878e-01
4.49951649e-01 4.98937368e-01 -1.05332363e+00 1.61527193e+00
-1.55145526e-01 5.28379560e-01 -1.39092892e-01 -1.27061164e+00
1.11136043e+00 -1.50933368e-02 4.45580333e-02 -2.58435726e-01
9.04956982e-02 -6.97935075e-02 1.27598330e-01 -1.10214978e-01
2.03008518e-01 6.60755754e-01 1.02474801e-01 1.29860520e-01
5.78554332e-01 4.51807767e-01 3.58903736e-01 5.19239545e-01
1.35003328e+00 2.90399432e-01 2.79228985e-01 -5.36435246e-01
1.93849131e-01 -2.75436997e-01 5.97237825e-01 7.57078528e-01
1.30429110e-02 5.19711494e-01 1.01612031e+00 -5.63891470e-01
-8.69367301e-01 -9.96755481e-01 3.35279703e-01 1.30421770e+00
-2.03187197e-01 -6.04620576e-01 -8.17264974e-01 -1.03249145e+00
2.70104080e-01 4.52752739e-01 -9.23859417e-01 -3.88972074e-01
-6.18402064e-01 -9.99105036e-01 6.38238549e-01 6.26729786e-01
4.94826883e-01 -9.79263842e-01 3.17042261e-01 3.07966858e-01
2.98385382e-01 -8.85822177e-01 -7.50894725e-01 4.18508559e-01
-1.04564178e+00 -1.33008754e+00 -4.93276864e-01 -1.03287387e+00
9.33341444e-01 1.59128726e-01 1.53418016e+00 1.83611497e-01
-1.94415987e-01 -2.69204706e-01 -3.04865152e-01 -2.57818371e-01
-8.86006206e-02 6.59897387e-01 -3.88618857e-01 -2.04044506e-01
2.36734062e-01 -9.67698216e-01 -5.35444140e-01 -1.11783095e-01
-4.37881023e-01 4.99036014e-02 7.10628688e-01 8.00078511e-01
6.91376209e-01 -1.56290129e-01 7.79859364e-01 -1.55414057e+00
7.22217381e-01 -5.89531958e-01 -7.28225946e-01 3.05797547e-01
-8.00852060e-01 4.00088221e-01 8.61478865e-01 -5.00690877e-01
-4.61391002e-01 -2.47230664e-01 -6.86172470e-02 -5.45512736e-01
2.01381072e-01 6.02679253e-01 -1.67938411e-01 -1.11256070e-01
7.75502801e-01 8.95572305e-02 -1.70235142e-01 -5.24186134e-01
6.71330810e-01 3.18480343e-01 3.95827472e-01 -4.05544609e-01
5.58811009e-01 -6.24643732e-03 2.77974214e-02 -5.07390082e-01
-7.67045259e-01 -3.11084479e-01 -3.62270653e-01 5.00804372e-02
4.86510038e-01 -1.02723801e+00 -8.81268501e-01 5.15359402e-01
-9.29545403e-01 -5.14336646e-01 -6.71135858e-02 2.27583095e-01
1.23519689e-01 8.60796496e-02 -8.61779332e-01 -5.45624256e-01
-8.73264849e-01 -7.52397835e-01 1.03318548e+00 4.61556852e-01
1.13841243e-01 -1.06413579e+00 -8.31633434e-02 -3.23820263e-01
3.43224019e-01 5.53564012e-01 9.51677620e-01 -1.00981712e+00
-6.45249367e-01 -2.64483660e-01 -6.22386515e-01 3.54021698e-01
2.08296567e-01 3.28279026e-02 -1.07481170e+00 -5.54528892e-01
-7.79470801e-01 -2.49811158e-01 1.07094634e+00 6.20768905e-01
1.23944533e+00 -5.34059167e-01 -7.83560991e-01 8.23631406e-01
1.53828406e+00 -1.72579214e-01 5.85995317e-01 -2.12411225e-01
1.36771250e+00 1.40353948e-01 -1.23293675e-01 -8.38646758e-03
4.86834019e-01 3.34701128e-02 4.58089799e-01 -1.73345953e-01
-4.97455269e-01 -6.64485097e-01 -3.30037214e-02 1.00056076e+00
1.66550323e-01 -5.71183622e-01 -6.39014244e-01 6.15596116e-01
-1.84634221e+00 -5.05639136e-01 -3.45746845e-01 2.23051381e+00
7.97763109e-01 2.71088541e-01 1.98605545e-02 -1.91530362e-01
9.82693553e-01 1.73789367e-01 -8.47844720e-01 -1.88300401e-01
-1.08098790e-01 3.29867840e-01 8.61690104e-01 4.27769244e-01
-1.07790613e+00 1.15069735e+00 6.04383802e+00 7.71516085e-01
-9.29061174e-01 -1.53260976e-01 1.01213813e+00 1.53446749e-01
-3.76490116e-01 2.12206680e-04 -8.83385241e-01 3.31474632e-01
1.07425404e+00 -7.47386813e-02 6.29228652e-01 1.05925930e+00
-1.05943084e-01 4.84211355e-01 -1.06408262e+00 7.67280042e-01
-1.10174179e-01 -1.45370293e+00 1.73640952e-01 7.83904791e-02
7.72000611e-01 5.94272554e-01 -4.33781117e-01 6.94469392e-01
9.69282866e-01 -1.26280236e+00 1.90050393e-01 3.64951789e-01
7.88338721e-01 -7.89715290e-01 6.10123575e-01 7.08338469e-02
-1.44405532e+00 1.52889773e-01 -6.55242920e-01 4.43181507e-02
-3.18841398e-01 9.68386292e-01 -9.80417490e-01 7.75668323e-01
7.31046259e-01 9.07866478e-01 -6.68965042e-01 9.87319946e-01
-4.29469019e-01 8.72132003e-01 -4.13206130e-01 -3.60451519e-01
2.27232978e-01 -1.55713394e-01 1.82889611e-01 1.18860924e+00
2.84189899e-02 -2.51229018e-01 2.33737260e-01 1.00717604e+00
-7.90802062e-01 4.99898531e-02 -6.54807985e-01 -2.83945620e-01
8.63567412e-01 1.42151165e+00 -6.21786833e-01 -4.35791373e-01
-2.66979814e-01 8.59274805e-01 1.13706410e+00 5.20305336e-01
-9.52290952e-01 -7.78680086e-01 5.45564175e-01 -4.85086516e-02
4.83345032e-01 -3.97666954e-02 -5.36346100e-02 -8.79929662e-01
-1.03612773e-01 -7.32464433e-01 6.72190368e-01 -7.23284841e-01
-1.43992984e+00 6.69387221e-01 -1.94162965e-01 -5.63017547e-01
1.93067223e-01 -6.08150601e-01 -8.10365736e-01 9.90460694e-01
-1.41395855e+00 -1.22396886e+00 -5.01877308e-01 3.49499553e-01
1.01566941e-01 4.38269451e-02 7.27738440e-01 3.40729773e-01
-8.11301172e-01 8.85293067e-01 1.57583971e-02 5.06348193e-01
4.72802758e-01 -1.60455215e+00 1.17741942e+00 8.18560064e-01
2.52301574e-01 8.44060838e-01 2.49493986e-01 -7.68563628e-01
-1.50179279e+00 -1.59546185e+00 6.02403164e-01 -3.06237042e-02
4.58519369e-01 -6.34443641e-01 -9.92907226e-01 1.17389774e+00
1.01150386e-01 4.54298824e-01 2.46959478e-01 5.11757195e-01
-5.14350832e-01 -3.47654819e-01 -1.15653193e+00 5.81332088e-01
1.39831686e+00 -3.16282123e-01 4.50673960e-02 4.83982801e-01
1.28485096e+00 -6.54035032e-01 -1.02423191e+00 2.69235641e-01
2.85718560e-01 -5.95859826e-01 8.21702242e-01 -9.30357277e-01
-7.70344958e-02 -1.76344916e-01 1.21277943e-01 -1.44380760e+00
-7.52776563e-01 -9.80907500e-01 -4.19445366e-01 1.32331204e+00
7.61295378e-01 -9.35750604e-01 1.18161464e+00 1.36174470e-01
-4.18693870e-01 -9.32200909e-01 -4.46240574e-01 -5.68733335e-01
3.88759002e-02 -5.90588665e-03 9.27976966e-01 1.02522469e+00
-1.03275463e-01 8.56561303e-01 -2.55492687e-01 4.25414324e-01
6.94677055e-01 3.16259228e-02 8.88803840e-01 -1.47839117e+00
-3.99810493e-01 -3.48988414e-01 -3.61606598e-01 -1.08394587e+00
1.27986574e-03 -1.29086590e+00 -1.84489295e-01 -1.88389349e+00
3.11579108e-01 -4.95234400e-01 -4.14073020e-01 7.32895911e-01
-5.82846940e-01 -4.93104570e-02 -2.72003800e-01 -7.48482570e-02
-5.37255585e-01 3.39220285e-01 1.28698492e+00 -1.60908625e-01
-1.23351738e-01 -1.50124431e-01 -1.06309271e+00 3.96843821e-01
1.00000131e+00 -4.79723006e-01 -8.11707318e-01 -5.17980814e-01
2.32015654e-01 -3.81616950e-01 7.88788870e-02 -7.17587769e-01
8.88787284e-02 1.60019502e-01 4.83844310e-01 -3.99547487e-01
-1.72485143e-01 -2.38391131e-01 2.17135280e-01 3.33736151e-01
-3.51793826e-01 1.50255129e-01 2.82905877e-01 7.66858876e-01
1.96471483e-01 1.07963709e-02 6.82514668e-01 -3.38053823e-01
-3.77851337e-01 9.26245630e-01 2.91587353e-01 4.26379830e-01
6.37973487e-01 1.71509776e-02 -6.55960739e-01 -1.54399931e-01
-8.10657084e-01 4.21684653e-01 5.14849350e-02 2.16652527e-01
3.36009860e-01 -1.36181879e+00 -7.73568630e-01 2.88095176e-01
1.77672878e-02 5.69693863e-01 1.29380515e-02 6.70311987e-01
-7.03850865e-01 5.76076582e-02 2.82115489e-01 -4.37709004e-01
-9.41738188e-01 4.81789917e-01 5.06562710e-01 -5.78912079e-01
-8.08798552e-01 1.41560125e+00 1.60233453e-01 -8.33084106e-01
5.06223917e-01 -6.91268504e-01 -8.86607766e-02 -2.72527069e-01
2.58317262e-01 3.54773790e-01 1.86978847e-01 -5.50885685e-02
-3.54724109e-01 1.01767384e-01 -4.68603104e-01 7.40885735e-01
1.36128974e+00 2.05020487e-01 -1.79587126e-01 1.76971301e-01
1.38925457e+00 -1.92282900e-01 -1.25303245e+00 -2.27330908e-01
-8.38755667e-02 8.59736465e-03 2.89245863e-02 -7.73056030e-01
-1.49976742e+00 6.64154649e-01 2.47058794e-01 4.85744923e-01
8.30492675e-01 2.52017024e-04 6.45655632e-01 2.00943187e-01
2.30281338e-01 -6.38636529e-01 -1.84627205e-01 4.04125184e-01
7.67525613e-01 -1.10621119e+00 1.18630081e-01 -6.24981403e-01
-2.79142618e-01 9.28418756e-01 8.00722063e-01 -6.89902365e-01
6.82161391e-01 2.84167141e-01 -4.53425199e-01 -5.04071951e-01
-9.23406720e-01 7.13444054e-02 2.24035114e-01 8.74781072e-01
7.87636995e-01 4.15578574e-01 5.44795953e-02 6.41065717e-01
-1.27500474e-01 -1.62779108e-01 3.40570539e-01 3.98035020e-01
-3.51728916e-01 -8.21794987e-01 2.51116365e-01 9.35577810e-01
-3.49063367e-01 -5.63155830e-01 -6.02522612e-01 9.62877333e-01
-1.84394643e-01 5.40152013e-01 -3.12203169e-02 -4.39428061e-01
3.43602061e-01 -1.35749936e-01 4.96158659e-01 -6.85595512e-01
-5.50231695e-01 -1.38123482e-01 4.99047220e-01 -5.50318182e-01
3.45794559e-02 -2.19462007e-01 -1.37978077e+00 -5.40218353e-01
-7.37235010e-01 2.58751392e-01 2.24053919e-01 5.44139266e-01
7.23734498e-01 1.13893008e+00 3.66348565e-01 -5.84477067e-01
-4.21892464e-01 -1.43861675e+00 -8.45776856e-01 2.21723512e-01
2.08868533e-01 -3.39604020e-01 -2.69196838e-01 -3.34148735e-01] | [6.976623058319092, 6.296162128448486] |
dd370890-665a-40c1-a1f8-5b000ed1e77b | optimal-graph-filters-for-clustering | 2211.04634 | null | https://arxiv.org/abs/2211.04634v1 | https://arxiv.org/pdf/2211.04634v1.pdf | Optimal Graph Filters for Clustering Attributed Graphs | Many real-world systems can be represented as graphs where the different entities are presented by nodes and their interactions by edges. An important task in studying large datasets is graph clustering. While there has been a lot of work on graph clustering using the connectivity between the nodes, many real-world networks also have node attributes. Clustering attributed graphs requires joint modeling of graph structure and node attributes. Recent work has focused on graph convolutional networks and graph convolutional filters to combine structural and content information. However, these methods are mostly limited to lowpass filtering and do not explicitly optimize the filters for the clustering task. In this paper, we introduce a graph signal processing based approach, where we design polynomial graph filters optimized for clustering. The proposed approach is formulated as a two-step iterative optimization problem where graph filters that are interpretable and optimal for the given data are learned while maximizing the separation between different clusters. The proposed approach is evaluated on attributed networks and compared to the state-of-the-art graph convolutional network approaches. | ['Selin Aviyente', 'Meiby Ortiz-Bouza'] | 2022-11-09 | null | null | null | null | ['graph-clustering'] | ['graphs'] | [-1.15553454e-01 4.16539073e-01 1.89759210e-01 -5.19643843e-01
1.53648823e-01 -3.91244709e-01 5.03498971e-01 7.46890962e-01
-2.92414814e-01 5.21899052e-02 -2.30926252e-03 -4.75526899e-02
-5.86702526e-01 -9.93305624e-01 -5.71233690e-01 -4.45747644e-01
-5.37560821e-01 5.80213487e-01 1.96155116e-01 1.28224507e-01
-6.69998005e-02 6.46934927e-01 -1.14503884e+00 1.93883732e-01
8.59531045e-01 7.11587608e-01 2.35233419e-02 9.26763535e-01
-1.98503688e-01 9.54190016e-01 -5.12836754e-01 -2.55169690e-01
1.73675969e-01 -3.90811443e-01 -6.44438326e-01 4.39780086e-01
5.20153999e-01 3.99817407e-01 -5.63240230e-01 1.33158576e+00
4.44244057e-01 3.21732998e-01 6.18477404e-01 -1.47558928e+00
-8.35686266e-01 9.02864754e-01 -3.61536413e-01 5.89067601e-02
-1.84512399e-02 -1.75811484e-01 1.32858145e+00 -4.97569054e-01
4.97160524e-01 1.42940760e+00 5.85279584e-01 1.00723617e-01
-1.50747025e+00 -4.70720649e-01 4.84543800e-01 3.50905031e-01
-1.46933675e+00 9.13231745e-02 9.67987359e-01 -6.35483563e-01
8.53023946e-01 9.86942053e-02 8.49871635e-01 3.76787901e-01
-1.24345319e-02 5.54927111e-01 5.52032232e-01 -2.65313745e-01
1.86801851e-01 -6.41236454e-02 5.28041124e-01 9.89223301e-01
3.42055738e-01 -3.59592885e-01 -2.73389965e-01 8.06522593e-02
6.02422178e-01 1.76766902e-01 -3.32036972e-01 -5.62309742e-01
-1.01447535e+00 9.19563115e-01 9.30972457e-01 3.49880993e-01
-6.20719254e-01 5.85657418e-01 1.34728640e-01 2.38193601e-01
5.14644682e-01 3.26624483e-01 -2.90513188e-02 7.22510397e-01
-9.63550568e-01 2.47280020e-02 1.07479417e+00 9.06346798e-01
1.07277691e+00 9.40888375e-02 -6.12737797e-02 5.86332262e-01
7.21172154e-01 8.02227482e-02 -9.40627083e-02 -6.55475795e-01
2.58056015e-01 1.04083848e+00 -4.60697800e-01 -1.66690135e+00
-8.13696206e-01 -6.62944078e-01 -1.06631303e+00 5.16191572e-02
3.44944447e-01 -2.29085371e-01 -1.04001737e+00 1.51012456e+00
1.38798922e-01 7.28897154e-01 -1.79499164e-01 8.39319229e-01
1.34668994e+00 5.98783851e-01 3.16248126e-02 5.74962646e-02
1.19836295e+00 -7.98225582e-01 -9.06511426e-01 -1.83002815e-01
3.37257504e-01 -5.08578598e-01 6.44956827e-01 2.52367079e-01
-8.79709482e-01 -4.98961449e-01 -9.95591283e-01 3.00248206e-01
-4.73794788e-01 2.29849532e-01 6.87828302e-01 6.22158587e-01
-1.25301492e+00 7.01507926e-01 -9.54971015e-01 -6.82907760e-01
3.81383359e-01 8.06504786e-01 -3.41875643e-01 1.78492710e-01
-9.81598675e-01 4.84086394e-01 6.08460128e-01 4.54218358e-01
-6.61511779e-01 -2.98925281e-01 -8.91700983e-01 5.07465541e-01
4.88917470e-01 -6.69626772e-01 6.31398141e-01 -1.25516737e+00
-1.10637176e+00 6.29411697e-01 2.43389040e-01 -6.12770557e-01
3.05547807e-02 5.22904247e-02 -4.61225450e-01 1.97963625e-01
-1.93726256e-01 4.63531822e-01 8.97580147e-01 -1.14683867e+00
-4.40038741e-01 -3.19163442e-01 1.52439907e-01 2.63700843e-01
-6.32152855e-01 -2.25290418e-01 -8.24953198e-01 -5.93180299e-01
3.11244994e-01 -9.25860643e-01 -5.06169975e-01 -2.28366584e-01
-7.06841826e-01 -2.24856079e-01 9.36719418e-01 -4.69600201e-01
1.41104412e+00 -2.11293530e+00 2.62285084e-01 8.43244433e-01
9.44850385e-01 7.98712075e-02 -1.52150691e-01 6.40775800e-01
-9.85979512e-02 2.89012700e-01 -1.34067059e-01 -2.54831254e-01
7.56013812e-03 1.18506297e-01 2.09686875e-01 7.62064457e-01
2.72875905e-01 8.72299135e-01 -8.10474277e-01 -4.12306547e-01
4.28933203e-01 7.14747667e-01 -5.70610046e-01 2.64471471e-01
-1.43017262e-01 3.55966359e-01 -4.31758851e-01 1.95655599e-01
6.51624680e-01 -6.08724236e-01 5.20105243e-01 -5.39511859e-01
2.37760454e-01 -2.21141115e-01 -1.54092276e+00 1.32083058e+00
4.57592793e-02 8.56134593e-01 4.65466678e-01 -1.50474596e+00
9.70460236e-01 1.89807370e-01 9.11160469e-01 4.73004691e-02
4.00900334e-01 -1.18603110e-01 4.69274879e-01 -3.40150923e-01
2.27544442e-01 2.96320796e-01 1.88021585e-01 1.85894936e-01
4.02212232e-01 -5.44133503e-03 3.85693997e-01 5.37397385e-01
1.37785876e+00 -3.33861440e-01 7.10174590e-02 -3.73604834e-01
6.25397921e-01 -1.82118014e-01 4.31559294e-01 5.79811156e-01
7.42652342e-02 4.00741607e-01 7.34117270e-01 -4.38192874e-01
-5.51947832e-01 -9.81459022e-01 3.96245092e-01 5.99285483e-01
1.33747384e-01 -7.53159821e-01 -8.70446563e-01 -5.43877959e-01
3.55153643e-02 2.31634319e-01 -6.23811960e-01 -7.48698190e-02
-4.73646224e-01 -7.65959561e-01 3.21968198e-01 3.80023420e-01
4.06872839e-01 -1.05386949e+00 -1.72367185e-01 3.55935603e-01
9.81915891e-02 -1.32597828e+00 -4.50661987e-01 2.44483158e-01
-7.01055825e-01 -1.38465774e+00 -1.76180616e-01 -1.00776327e+00
9.88727689e-01 8.74459594e-02 1.20429385e+00 4.49309826e-01
-3.31014216e-01 6.50335014e-01 -3.31692934e-01 -3.59808624e-01
-1.27778709e-01 1.77053750e-01 -2.88110197e-01 6.74567461e-01
3.06263596e-01 -6.46226108e-01 -4.35463369e-01 -1.10519053e-02
-8.12910557e-01 -3.93145420e-02 6.93787098e-01 7.28117228e-01
6.35213077e-01 5.19939959e-01 2.76962429e-01 -1.33577669e+00
9.94407177e-01 -5.00309110e-01 -6.04871869e-01 2.56536096e-01
-4.82053041e-01 2.37797573e-01 8.70413780e-01 -4.00500447e-01
-4.38116521e-01 4.42532986e-01 1.98375106e-01 -7.36226499e-01
-1.21495835e-01 8.54889512e-01 -3.18607867e-01 -2.76671767e-01
4.54844296e-01 -2.75856405e-01 -1.13216126e-02 -9.12482888e-02
6.04547620e-01 2.49994740e-01 2.52740115e-01 -2.99940079e-01
7.22512305e-01 4.62673068e-01 3.18784118e-01 -1.21379554e+00
-4.16188419e-01 -7.13284552e-01 -7.56113589e-01 -5.61448932e-01
1.19373119e+00 -5.66124737e-01 -1.04124928e+00 3.09019506e-01
-9.19050694e-01 -2.27432847e-01 -6.09091781e-02 5.84941626e-01
-2.96977550e-01 5.21088183e-01 -4.62428004e-01 -5.85783899e-01
-2.33422711e-01 -1.12235153e+00 7.51331329e-01 1.94281831e-01
-2.90012918e-02 -1.41558623e+00 -1.68158576e-01 -1.98469892e-01
2.39563614e-01 5.62258065e-01 8.59163642e-01 -8.62510800e-01
-7.98470557e-01 -2.74065107e-01 -3.33546311e-01 1.66521892e-01
1.36233896e-01 1.93396360e-01 -6.42468452e-01 -3.52950364e-01
-4.71300542e-01 1.76211014e-01 8.95291865e-01 8.32596481e-01
1.09056330e+00 -1.63092360e-01 -4.47464108e-01 7.94508100e-01
1.43146324e+00 -1.80202220e-02 3.39474529e-01 -1.60605505e-01
1.23914731e+00 6.62836134e-01 9.34720621e-04 2.05797106e-01
3.67637396e-01 4.37533587e-01 8.18391263e-01 -4.56469148e-01
-4.50243354e-02 6.97816908e-02 -2.04966497e-02 8.09002042e-01
-7.07788318e-02 -6.89812422e-01 -1.03003085e+00 4.95357275e-01
-2.24470019e+00 -9.52778935e-01 -7.63813198e-01 1.91532767e+00
-5.92621900e-02 1.88615173e-01 -2.71065254e-03 1.43235564e-01
1.12467575e+00 1.45130977e-01 -3.62190574e-01 -2.15510607e-01
-1.12430686e-02 2.83151478e-01 5.62200189e-01 5.63051641e-01
-1.34914970e+00 8.41433525e-01 5.73013687e+00 3.56091678e-01
-1.00345504e+00 -2.20052987e-01 4.16296750e-01 3.51834714e-01
-1.23895392e-01 1.76394984e-01 -2.91478813e-01 1.25738919e-01
8.02409947e-01 -1.62229598e-01 5.60236514e-01 5.51167607e-01
2.38163084e-01 3.01247776e-01 -1.16903126e+00 9.13781285e-01
-5.37254959e-02 -1.22987461e+00 -5.17914183e-02 9.39390063e-02
5.36915123e-01 -1.13718309e-01 -2.75564045e-02 7.10271299e-02
5.60591459e-01 -1.29538906e+00 4.55513179e-01 6.33944333e-01
9.81773436e-02 -6.95230722e-01 5.63683927e-01 1.94404826e-01
-1.82035673e+00 -6.76160380e-02 -2.76853323e-01 6.23119250e-02
4.43335958e-02 6.90744698e-01 -9.60772812e-01 6.40761197e-01
6.77627325e-01 1.22296894e+00 -6.99756801e-01 1.15578866e+00
-3.11637431e-01 8.18174303e-01 -3.32825989e-01 -2.03747347e-01
3.70104253e-01 -4.52865899e-01 5.49807250e-01 1.25159514e+00
1.10032566e-01 -4.82481271e-02 6.26817346e-01 1.01451802e+00
-2.59747684e-01 3.06779295e-01 -5.83205938e-01 -4.72815156e-01
2.71510750e-01 1.61629450e+00 -1.38983607e+00 -2.26156324e-01
-6.27349555e-01 7.86403477e-01 5.70666254e-01 5.01206875e-01
-6.54912949e-01 -6.13396049e-01 4.21232998e-01 1.96881115e-01
2.83793658e-01 -4.20170814e-01 -5.55119738e-02 -8.94249618e-01
-2.48989880e-01 -4.98113543e-01 5.97127616e-01 -5.68827868e-01
-1.40921581e+00 6.33337140e-01 -1.22029103e-01 -9.84873056e-01
1.57977849e-01 -6.55069530e-01 -7.98696458e-01 8.19362640e-01
-1.09998143e+00 -1.30361843e+00 -6.92575693e-01 8.94244194e-01
1.23905547e-01 -1.09626375e-01 5.72539032e-01 3.81549090e-01
-4.09496248e-01 2.34869495e-01 -7.62881339e-02 4.96558964e-01
3.74631315e-01 -1.53917074e+00 4.65771914e-01 1.00065613e+00
5.38568795e-01 6.54027224e-01 5.76693177e-01 -7.88835049e-01
-1.40100002e+00 -1.40164983e+00 6.11802220e-01 3.12763229e-02
7.28354335e-01 -5.69888353e-01 -9.14902270e-01 7.74510264e-01
4.05466586e-01 4.24988776e-01 4.41527218e-01 2.16023430e-01
-8.05266649e-02 -1.46111175e-01 -8.85481417e-01 3.51602405e-01
8.48586261e-01 -2.62669742e-01 2.25737561e-02 3.26540828e-01
4.66018289e-01 -1.46686926e-01 -9.76300120e-01 1.99160188e-01
1.99558198e-01 -7.34820426e-01 8.54152560e-01 -6.38732433e-01
-7.47854337e-02 -6.79082334e-01 1.54499009e-01 -1.65203571e+00
-6.02568507e-01 -4.95374948e-01 -2.53314134e-02 1.03861094e+00
3.52341712e-01 -4.46109116e-01 9.74274278e-01 3.25982004e-01
-1.19867973e-01 -4.60081667e-01 -3.26485395e-01 -4.11501646e-01
-5.42791307e-01 -3.37088913e-01 4.28464323e-01 1.01202810e+00
-1.68020621e-01 7.05630064e-01 -1.41501591e-01 6.46556258e-01
8.58282626e-01 1.81137696e-02 6.79075181e-01 -1.72385561e+00
-2.95388967e-01 -6.16379917e-01 -1.07586432e+00 -5.98393202e-01
2.97999233e-01 -1.23821390e+00 -1.38658866e-01 -2.07939982e+00
-1.24751255e-01 -3.28370005e-01 -1.86338410e-01 3.36120516e-01
-2.45111510e-01 -1.09381393e-01 2.65519768e-01 -1.22943670e-01
-6.36540234e-01 2.54444927e-01 9.52892959e-01 -4.65877801e-01
-3.43693584e-01 -3.25794220e-02 -4.32101816e-01 6.15902483e-01
7.81475484e-01 -3.58934551e-01 -8.04300308e-01 -4.65822428e-01
3.59304786e-01 2.45934166e-02 4.48428601e-01 -1.24419165e+00
6.17765427e-01 1.64582208e-01 1.35645077e-01 -5.81978738e-01
2.18128860e-01 -1.19661474e+00 5.75762868e-01 3.35386455e-01
-3.13079327e-01 1.54337123e-01 -1.63774177e-01 8.89694154e-01
-3.94342214e-01 1.42605737e-01 7.97245622e-01 -2.54546881e-01
-7.22339034e-01 6.45788491e-01 -3.92121702e-01 -2.15836652e-02
1.13671815e+00 -1.16572097e-01 -2.07877327e-02 -6.91041172e-01
-1.36993301e+00 4.21668053e-01 1.39334202e-01 2.41543815e-01
7.56013930e-01 -1.17297471e+00 -7.42161036e-01 7.00517446e-02
-1.91131860e-01 4.88966964e-02 9.35494155e-02 7.46094525e-01
-6.53818488e-01 4.44336869e-02 -8.53272453e-02 -7.52534389e-01
-1.47095287e+00 5.79292774e-01 5.18750787e-01 -3.75363439e-01
-6.26857519e-01 8.89743924e-01 1.68677494e-01 -4.74431992e-01
3.32442045e-01 -3.74433130e-01 -6.73748434e-01 1.15618147e-01
-1.35552824e-01 2.50368893e-01 7.44473711e-02 -8.11116457e-01
-3.12605828e-01 5.23985982e-01 1.82422131e-01 2.01454803e-01
1.40656662e+00 5.38362339e-02 -3.94365758e-01 2.48060033e-01
1.04160547e+00 -2.05386564e-01 -7.49736488e-01 -1.77012235e-01
2.35867903e-01 -1.76065177e-01 1.71105310e-01 -1.95367917e-01
-1.64282131e+00 1.04084599e+00 5.53132296e-01 7.71555781e-01
1.13762701e+00 -2.11617704e-02 3.33775312e-01 4.67632622e-01
-2.81301830e-02 -9.79795337e-01 1.53252825e-01 4.66611803e-01
5.44890106e-01 -1.03913438e+00 1.34346858e-01 -8.54442596e-01
-3.46628547e-01 1.21038580e+00 6.34209692e-01 -7.07406640e-01
1.20044923e+00 1.59013867e-01 -1.52266473e-01 -8.00741732e-01
-5.02059042e-01 -5.04206300e-01 5.80264747e-01 6.46982193e-01
6.47121787e-01 1.84906498e-01 -2.29937360e-01 5.07190406e-01
-3.69772799e-02 -4.00745928e-01 4.02290761e-01 4.99015629e-01
-3.76823902e-01 -1.01881802e+00 -2.06854492e-01 8.67200434e-01
-5.72902322e-01 -8.06072354e-03 -6.43598258e-01 7.79100418e-01
1.06083537e-02 1.19250286e+00 1.45542338e-01 -5.34847319e-01
3.81611943e-01 -4.24679995e-01 2.22850174e-01 -8.34152877e-01
-8.59147549e-01 1.48579925e-01 1.09040774e-02 -4.57215935e-01
-7.51187623e-01 -3.99361283e-01 -1.50046003e+00 -1.11390248e-01
-5.18387020e-01 2.87015766e-01 5.07975101e-01 5.74064136e-01
2.29750678e-01 1.02026665e+00 3.47398251e-01 -7.53630638e-01
1.82624146e-01 -7.63795912e-01 -8.81025791e-01 5.52316248e-01
1.78143680e-01 -5.50704956e-01 -2.13270679e-01 1.59044608e-01] | [7.171375274658203, 6.0998101234436035] |
05ada3e2-b9e4-4500-8caf-652fe9d4a84b | hrel-filter-pruning-based-on-high-relevance | 2202.10716 | null | https://arxiv.org/abs/2202.10716v1 | https://arxiv.org/pdf/2202.10716v1.pdf | HRel: Filter Pruning based on High Relevance between Activation Maps and Class Labels | This paper proposes an Information Bottleneck theory based filter pruning method that uses a statistical measure called Mutual Information (MI). The MI between filters and class labels, also called \textit{Relevance}, is computed using the filter's activation maps and the annotations. The filters having High Relevance (HRel) are considered to be more important. Consequently, the least important filters, which have lower Mutual Information with the class labels, are pruned. Unlike the existing MI based pruning methods, the proposed method determines the significance of the filters purely based on their corresponding activation map's relationship with the class labels. Architectures such as LeNet-5, VGG-16, ResNet-56\textcolor{myblue}{, ResNet-110 and ResNet-50 are utilized to demonstrate the efficacy of the proposed pruning method over MNIST, CIFAR-10 and ImageNet datasets. The proposed method shows the state-of-the-art pruning results for LeNet-5, VGG-16, ResNet-56, ResNet-110 and ResNet-50 architectures. In the experiments, we prune 97.98 \%, 84.85 \%, 76.89\%, 76.95\%, and 63.99\% of Floating Point Operation (FLOP)s from LeNet-5, VGG-16, ResNet-56, ResNet-110, and ResNet-50 respectively.} The proposed HRel pruning method outperforms recent state-of-the-art filter pruning methods. Even after pruning the filters from convolutional layers of LeNet-5 drastically (i.e. from 20, 50 to 2, 3, respectively), only a small accuracy drop of 0.52\% is observed. Notably, for VGG-16, 94.98\% parameters are reduced, only with a drop of 0.36\% in top-1 accuracy. \textcolor{myblue}{ResNet-50 has shown a 1.17\% drop in the top-5 accuracy after pruning 66.42\% of the FLOPs.} In addition to pruning, the Information Plane dynamics of Information Bottleneck theory is analyzed for various Convolutional Neural Network architectures with the effect of pruning. | ['SH Shabbeer Basha', 'Shiv Ram Dubey', 'Mrinmoy Ghorai', 'CH Sarvani'] | 2022-02-22 | null | null | null | null | ['information-plane'] | ['methodology'] | [-1.05215590e-02 1.53403208e-01 4.01228994e-01 -1.99378714e-01
2.20274609e-02 -2.62935191e-01 -3.90012786e-02 1.75281599e-01
-9.26998496e-01 9.13368702e-01 -2.54104018e-01 -4.37370718e-01
-5.08421123e-01 -1.10885823e+00 -5.54334462e-01 -4.41929072e-01
-1.28527343e-01 -7.32936338e-02 6.43099785e-01 -1.93953753e-01
2.55649209e-01 3.65792334e-01 -1.88531566e+00 6.02502525e-01
7.40408182e-01 1.41849267e+00 2.73220032e-01 6.48215592e-01
7.90506154e-02 5.90618551e-01 -8.56339037e-01 -5.72101474e-01
6.11423314e-01 -2.12170199e-01 -5.72732329e-01 -5.27672112e-01
6.93490207e-01 -6.95583075e-02 -6.04449689e-01 1.44496083e+00
4.41578716e-01 1.86422274e-01 4.88705963e-01 -7.93505669e-01
-1.38229311e-01 1.00775802e+00 -6.67234600e-01 8.01344812e-01
-4.28745240e-01 -7.07192943e-02 9.03638124e-01 -7.46680379e-01
4.05845225e-01 1.09940481e+00 6.52470291e-01 4.28858250e-01
-1.07329893e+00 -1.33104002e+00 3.90078723e-01 -1.68804713e-02
-1.61238456e+00 1.85343549e-02 1.91183642e-01 -5.56586385e-01
1.08290768e+00 3.04881096e-01 7.59558380e-01 2.57429391e-01
4.33432639e-01 3.02353770e-01 1.18243980e+00 -3.76067251e-01
1.12097025e-01 2.04843014e-01 7.85715163e-01 1.02521610e+00
7.32010305e-01 1.78388923e-01 -6.73500597e-01 1.95964739e-01
8.19711745e-01 -9.03107505e-03 -2.79459119e-01 6.79156482e-01
-5.57158709e-01 5.28624296e-01 6.57531559e-01 1.66134775e-01
-1.65266916e-01 1.41946092e-01 3.13275933e-01 4.84611154e-01
4.79061961e-01 2.79443622e-01 -6.15552187e-01 2.59599388e-01
-1.11491072e+00 1.64020836e-01 8.01469922e-01 1.25704312e+00
8.58794987e-01 3.35604161e-01 -3.67766887e-01 7.62490392e-01
4.02723134e-01 3.60969961e-01 2.64910340e-01 -5.17640769e-01
5.88245749e-01 9.64558482e-01 -2.80038059e-01 -9.22254205e-01
-5.24312377e-01 -1.02974248e+00 -1.06243849e+00 5.59510052e-01
3.10176730e-01 -4.12777871e-01 -1.26153684e+00 1.64905107e+00
-2.84842044e-01 5.64913824e-02 1.14855841e-01 6.15929365e-01
1.28894258e+00 4.23020452e-01 1.67771429e-01 -1.59142911e-01
1.60033035e+00 -5.31504273e-01 -2.89973766e-01 -1.27774805e-01
2.27046728e-01 -1.09395492e+00 6.04578078e-01 5.32671809e-01
-1.18805575e+00 -8.94027948e-01 -1.31502092e+00 4.39240009e-01
-5.43709755e-01 5.18427134e-01 4.20408607e-01 7.44473279e-01
-1.04952610e+00 9.73575950e-01 -5.84664464e-01 -1.43404454e-01
7.28743970e-01 8.09870899e-01 1.29209101e-01 5.15988655e-02
-1.00187814e+00 6.97944343e-01 7.38402307e-01 -7.70571157e-02
-6.56286001e-01 -9.33384061e-01 -2.53239095e-01 4.35522825e-01
1.99050549e-02 -4.54233468e-01 7.55216479e-01 -6.85593545e-01
-1.00296366e+00 5.61497092e-01 2.15824042e-02 -7.00681984e-01
2.76905656e-01 -3.31047177e-01 -5.75832665e-01 1.32395655e-01
-2.16424271e-01 9.55033064e-01 3.81066412e-01 -9.16940928e-01
-1.29352939e+00 -3.09597105e-01 2.66879201e-01 1.43284658e-02
-4.05133039e-01 4.00931686e-02 -3.11136186e-01 -7.52977192e-01
5.44546545e-01 -6.56718910e-01 -2.07140043e-01 -4.81359661e-01
-6.15085423e-01 -1.20070996e-02 6.32864833e-01 -3.11670929e-01
1.58999324e+00 -2.14197302e+00 -5.81759453e-01 4.35394853e-01
5.10874331e-01 7.45581329e-01 1.30840093e-01 -1.39628723e-01
4.21411023e-02 3.70787263e-01 3.53111215e-02 1.73368752e-01
-4.29506630e-01 -7.74481967e-02 -5.14716394e-02 9.76072624e-02
1.37249529e-01 2.28101701e-01 -5.26051342e-01 -1.77415073e-01
2.59843141e-01 4.08967912e-01 -6.75264239e-01 -2.77332276e-01
1.25615597e-01 1.40050411e-01 -3.48632425e-01 3.20509791e-01
9.61519778e-01 5.81897888e-03 -1.03944339e-01 -4.95102555e-01
-5.36501646e-01 2.74288803e-02 -1.32286119e+00 1.04079866e+00
-1.07397974e-01 8.25803578e-01 -2.27919251e-01 -6.70896351e-01
1.32571495e+00 2.13588938e-01 3.15454841e-01 -7.29459703e-01
4.52588201e-01 2.26418182e-01 7.43697166e-01 -4.38953675e-02
5.12058616e-01 2.31432766e-01 3.13612819e-01 5.61166406e-02
3.22537571e-01 3.84169519e-01 5.61263263e-01 2.65832752e-01
1.32035005e+00 -4.75405812e-01 -3.33134197e-02 -9.37029898e-01
6.45070672e-01 -2.19658524e-01 8.21649849e-01 1.04793799e+00
-1.27459280e-02 5.12907863e-01 4.18902338e-01 -4.22394663e-01
-8.14689696e-01 -9.06726837e-01 -4.44718599e-01 8.42134416e-01
1.91678047e-01 -8.01438570e-01 -9.44861114e-01 -2.48604745e-01
-8.69874060e-02 5.11065364e-01 -4.41884041e-01 -2.56049335e-01
-4.81273085e-01 -8.98330033e-01 9.65229273e-01 4.43268031e-01
1.44045055e+00 -1.10072327e+00 -9.13371563e-01 3.27116400e-01
3.77353638e-01 -9.22925413e-01 2.09362194e-01 3.08238447e-01
-1.24220991e+00 -1.27823591e+00 -2.70700961e-01 -7.26203382e-01
7.96720743e-01 -2.44051684e-02 1.14240539e+00 2.43511647e-01
-5.45027792e-01 -4.22080979e-02 -2.81303376e-01 -6.23253763e-01
7.39943385e-02 4.37685102e-02 -9.05998051e-02 -3.02553684e-01
4.90359783e-01 -6.27606511e-01 -9.58614588e-01 6.39636040e-01
-5.71590960e-01 6.36137798e-02 7.27922678e-01 7.10089147e-01
8.55131328e-01 5.54075181e-01 4.46790457e-01 -1.27864861e+00
4.45644468e-01 -5.93755618e-02 -7.58837700e-01 1.12812787e-01
-9.02914822e-01 -8.11595842e-02 8.13428462e-01 -3.81764680e-01
-1.08801258e+00 -1.04822226e-01 7.02845678e-02 -2.60010570e-01
-1.16343521e-01 2.61810392e-01 1.92995425e-02 -1.34002492e-01
8.50731254e-01 -1.28577888e-01 -6.33408010e-01 -6.31615520e-01
-6.16274849e-02 2.96856999e-01 4.49927777e-01 -2.33741716e-01
4.78573620e-01 3.44035953e-01 -1.68695878e-02 -9.23836172e-01
-6.69922292e-01 -3.03180933e-01 -3.30186367e-01 -1.33283123e-01
7.70637512e-01 -8.48551035e-01 -8.50724936e-01 4.37030762e-01
-9.64982033e-01 -1.97697897e-02 -2.13375792e-01 8.47092330e-01
1.99064314e-02 -3.47365886e-01 -6.48121238e-01 -8.31160307e-01
-7.24294782e-01 -1.32149851e+00 3.34257066e-01 7.97597468e-01
-5.10755442e-02 -4.75827932e-01 -7.55611241e-01 -1.91234246e-01
4.50907379e-01 6.27316758e-02 1.08223498e+00 -6.52418375e-01
-7.06578434e-01 -4.10564318e-02 -7.84804821e-01 6.56323969e-01
-2.30110034e-01 1.61147155e-02 -9.65642333e-01 -3.24618876e-01
-2.51807332e-01 2.46850654e-01 1.32904160e+00 7.10591435e-01
1.17959607e+00 -1.38730928e-01 -4.12102968e-01 8.62694919e-01
1.81164181e+00 6.58196688e-01 6.60435677e-01 4.50319834e-02
2.95619100e-01 1.02195278e-01 4.68715012e-01 5.72427571e-01
-2.58869320e-01 8.55760425e-02 4.47084069e-01 -7.53960162e-02
-5.43485105e-01 1.18365280e-01 -2.13876702e-02 7.58500278e-01
-4.08026665e-01 -3.51927996e-01 -5.97216487e-01 5.24977781e-02
-1.49611354e+00 -7.86769807e-01 -4.96793211e-01 2.23439717e+00
5.61885417e-01 7.60703683e-01 -4.75727439e-01 4.29436237e-01
7.91211605e-01 -1.81250066e-01 -4.75132316e-01 -4.42704976e-01
-1.18288226e-01 8.12403500e-01 9.47598338e-01 2.97570914e-01
-1.11148155e+00 1.07245350e+00 4.80444908e+00 7.45027602e-01
-9.35739160e-01 -2.79332325e-02 8.17118287e-01 -1.19771212e-01
3.12752575e-01 -3.45005235e-03 -1.50805342e+00 4.51055944e-01
7.18097925e-01 7.20767528e-02 3.33440989e-01 6.51089489e-01
1.25140712e-01 -5.72806418e-01 -9.37782526e-01 9.75038052e-01
-2.97782093e-01 -1.34767115e+00 7.13571310e-02 -5.22572398e-02
7.12134182e-01 1.76972687e-01 4.77452241e-02 3.93434554e-01
4.47178870e-01 -9.54377711e-01 7.95129955e-01 6.40719891e-01
8.16988051e-01 -1.10774648e+00 9.40733016e-01 9.80662629e-02
-1.40738177e+00 -2.12494463e-01 -1.00395012e+00 -2.26995841e-01
-2.32297912e-01 9.45241034e-01 -4.77472544e-01 3.03820223e-01
1.32979310e+00 3.97643834e-01 -5.73011160e-01 1.32021046e+00
-2.61055499e-01 9.33501542e-01 -4.18919027e-01 -1.35345623e-01
1.40606537e-01 -1.82317104e-02 5.20112336e-01 1.41949105e+00
2.35125214e-01 5.50931275e-01 -7.68185686e-03 8.79777730e-01
-2.84265280e-01 -6.80178627e-02 -1.75522313e-01 4.20463890e-01
6.73508108e-01 1.11994982e+00 -1.13847911e+00 -5.29172480e-01
-2.47611016e-01 2.07523450e-01 6.96331486e-02 4.16220486e-01
-7.86603689e-01 -8.75546396e-01 7.72349298e-01 7.61913657e-02
2.96586335e-01 1.44791737e-01 -5.27072549e-01 -6.31418705e-01
-7.39997849e-02 -6.19548023e-01 4.66719329e-01 -2.75930792e-01
-9.04017389e-01 1.12925446e+00 1.33267716e-01 -9.91205394e-01
5.42482078e-01 -7.81725049e-01 -5.58456540e-01 1.09701371e+00
-1.17606151e+00 -3.53402913e-01 -5.56426644e-01 5.22501767e-01
4.81972694e-01 -6.33966923e-01 5.20786643e-01 5.57072461e-01
-5.93301654e-01 9.13832307e-01 -9.63778421e-03 3.49994212e-01
8.54729488e-02 -9.59409893e-01 2.84347713e-01 8.87361944e-01
-5.37274666e-02 8.27214599e-01 3.94093245e-01 -5.87360740e-01
-6.77951515e-01 -1.19066012e+00 5.55431902e-01 3.72281551e-01
1.93175122e-01 -2.69737661e-01 -6.96019173e-01 4.08397794e-01
8.57015923e-02 1.41273782e-01 3.91957760e-01 -4.17433828e-02
-1.16380699e-01 -5.54071188e-01 -1.29991674e+00 5.65864027e-01
1.28419447e+00 6.38350099e-02 -2.61649638e-01 5.68139441e-02
5.57934582e-01 -3.17641199e-01 -9.68939960e-01 5.48589826e-01
5.94954073e-01 -1.34556305e+00 8.48927259e-01 -2.75465488e-01
1.07165210e-01 -2.23105878e-01 -3.24223369e-01 -8.58741820e-01
-6.00882947e-01 -4.01809901e-01 2.64997870e-01 1.23249185e+00
7.65840054e-01 -8.68353426e-01 8.98843527e-01 2.65026838e-01
-4.17078674e-01 -7.64205873e-01 -8.21751773e-01 -6.02158546e-01
2.41655111e-02 -3.73732716e-01 4.48189825e-01 3.02321136e-01
-5.07257581e-01 1.78022221e-01 1.29959077e-01 3.10483873e-01
5.12675881e-01 -4.10822570e-01 3.41170639e-01 -1.56501782e+00
-1.91155016e-01 -6.81637108e-01 -6.10656381e-01 -9.27683353e-01
-4.46224660e-01 -7.61552095e-01 -2.95575857e-01 -1.31338561e+00
-1.92520476e-03 -7.78870225e-01 -8.45460296e-01 4.68573362e-01
1.36917770e-01 3.66443664e-01 3.07498515e-01 6.03662468e-02
-1.51604101e-01 -1.30121768e-01 1.23575068e+00 -1.36026636e-01
-3.34313720e-01 1.91738799e-01 -5.32833040e-01 1.07039464e+00
9.95680094e-01 -6.33021057e-01 -6.29736841e-01 -5.71582913e-01
7.75007084e-02 -3.61053675e-01 2.83264607e-01 -1.67736948e+00
3.97863269e-01 3.96354496e-01 8.24638903e-01 -8.58872592e-01
2.72275031e-01 -6.86449289e-01 1.05197486e-02 7.38411963e-01
-2.90681779e-01 2.40382940e-01 5.32413900e-01 5.06904900e-01
-1.93202406e-01 -4.07093287e-01 1.05792415e+00 -3.69385898e-01
-7.40044475e-01 2.15338692e-01 -3.84337008e-01 1.00157507e-01
8.09103549e-01 -4.90386516e-01 -4.85567451e-01 2.84642190e-01
-9.85342085e-01 -2.77527012e-02 -3.82781535e-01 -1.03580148e-03
8.95412385e-01 -7.96908557e-01 -7.08302021e-01 3.91346186e-01
-2.50912070e-01 9.97435227e-02 1.80672139e-01 6.24739110e-01
-6.69371784e-01 3.90264302e-01 -3.91271442e-01 -4.81762052e-01
-1.42578018e+00 -9.01751891e-02 3.33809495e-01 -3.45635504e-01
-6.99864507e-01 1.33109426e+00 3.68600160e-01 2.39188567e-01
4.01495665e-01 -7.38381684e-01 -4.88480866e-01 -6.27935976e-02
4.24137443e-01 7.57132649e-01 4.29235220e-01 -2.38368392e-01
-3.38793725e-01 4.96750623e-01 -2.77400374e-01 -1.08510796e-02
1.14235508e+00 3.10062081e-01 -6.62549511e-02 -1.48911729e-01
7.78818071e-01 -2.25723922e-01 -1.02380633e+00 -7.21364841e-02
-1.64025426e-02 -2.85968781e-01 1.94965407e-01 -1.07794714e+00
-1.61752105e+00 8.64854574e-01 1.07837200e+00 -1.67463034e-01
1.32758296e+00 -4.25732940e-01 4.73002821e-01 3.01183641e-01
4.77408022e-01 -1.16061175e+00 -2.14671254e-01 8.95543814e-01
5.11223376e-01 -6.67469800e-01 2.66219497e-01 -5.53629518e-01
-1.14294082e-01 1.19708228e+00 1.07334769e+00 -4.53708261e-01
1.22887421e+00 4.19222206e-01 -3.01807940e-01 -3.99126738e-01
-7.11697221e-01 -2.48884618e-01 2.85819948e-01 1.84393436e-01
6.12258792e-01 1.38914511e-01 -6.42800391e-01 7.86762476e-01
-6.68331265e-01 -2.45087922e-01 2.85498500e-01 8.84567499e-01
-8.03264797e-01 -6.72646582e-01 -7.32682571e-02 1.00607324e+00
-7.56854057e-01 -4.92759377e-01 -2.19996050e-02 1.02641904e+00
7.66925454e-01 1.16943455e+00 2.61732578e-01 -8.03930759e-01
4.67058092e-01 -1.52749345e-01 3.82452458e-01 -7.02648342e-01
-1.15452814e+00 1.30739301e-01 -1.33626899e-02 -2.93861926e-01
-8.30140561e-02 -1.48581028e-01 -1.48037779e+00 -4.58691061e-01
-5.81585407e-01 2.05774859e-01 5.10408461e-01 6.41226470e-01
2.69972235e-01 1.11315393e+00 -3.02609559e-02 -1.55371845e-01
-1.63084269e-01 -1.01036727e+00 -7.63527513e-01 5.33366427e-02
-3.87349725e-01 -7.93938994e-01 -2.05505013e-01 -1.09661303e-01] | [8.559946060180664, 3.0246477127075195] |
99a26a20-840d-4931-a740-3efe387edf06 | take-the-hint-improving-arabic-diacritization | 2306.03557 | null | https://arxiv.org/abs/2306.03557v1 | https://arxiv.org/pdf/2306.03557v1.pdf | Take the Hint: Improving Arabic Diacritization with Partially-Diacritized Text | Automatic Arabic diacritization is useful in many applications, ranging from reading support for language learners to accurate pronunciation predictor for downstream tasks like speech synthesis. While most of the previous works focused on models that operate on raw non-diacritized text, production systems can gain accuracy by first letting humans partly annotate ambiguous words. In this paper, we propose 2SDiac, a multi-source model that can effectively support optional diacritics in input to inform all predictions. We also introduce Guided Learning, a training scheme to leverage given diacritics in input with different levels of random masking. We show that the provided hints during test affect more output positions than those annotated. Moreover, experiments on two common benchmarks show that our approach i) greatly outperforms the baseline also when evaluated on non-diacritized text; and ii) achieves state-of-the-art results while reducing the parameter count by over 60%. | ['Mohammad Zeineldeen', 'Nick Rossenbach', 'Mattia Di Gangi', 'Parnia Bahar'] | 2023-06-06 | null | null | null | null | ['speech-synthesis'] | ['speech'] | [ 3.52242976e-01 3.43981206e-01 -2.13291466e-01 -3.97531211e-01
-1.06020224e+00 -7.96360970e-01 6.71522319e-01 3.39025617e-01
-5.46944678e-01 8.24068189e-01 3.76786828e-01 -9.96628702e-01
3.34291935e-01 -6.71490610e-01 -7.80276537e-01 -3.72509271e-01
2.31903493e-01 6.33440673e-01 5.45690298e-01 -7.74830639e-01
2.24366218e-01 1.22104488e-01 -1.28774238e+00 8.29848588e-01
1.23005450e+00 5.53465188e-01 2.51756459e-01 7.31335759e-01
-1.15802132e-01 7.96800554e-01 -1.03030205e+00 -6.98658705e-01
6.23030290e-02 -5.52496016e-01 -9.15294111e-01 -2.12067083e-01
2.43117973e-01 -3.59793067e-01 3.15830737e-01 7.52058744e-01
3.87375891e-01 1.01151034e-01 8.22424769e-01 -5.71460724e-01
-7.14222968e-01 1.51070595e+00 -2.36698493e-01 1.61453754e-01
4.74019498e-01 -3.57101187e-02 1.34916329e+00 -1.28520393e+00
2.27655843e-01 1.29147959e+00 3.28870147e-01 7.55027652e-01
-1.11338377e+00 -3.37387651e-01 4.41593468e-01 1.78431779e-01
-9.87477183e-01 -4.32134479e-01 4.39967722e-01 -1.02475673e-01
1.16617346e+00 5.24817109e-01 2.23205268e-01 9.67400134e-01
-2.26237655e-01 1.27025247e+00 1.26592970e+00 -1.12191904e+00
-5.75058497e-02 9.46694016e-02 4.53734130e-01 6.25873268e-01
-1.31139174e-01 -2.21485004e-01 -7.73252487e-01 3.67450684e-01
2.30673835e-01 -5.23809850e-01 -3.10231864e-01 5.83254457e-01
-1.25570524e+00 6.85376585e-01 -6.13392740e-02 3.13748479e-01
-6.09108321e-02 -3.03758740e-01 3.31772119e-01 6.50586724e-01
3.36453646e-01 5.75373113e-01 -8.32252145e-01 -4.14947271e-01
-8.91077459e-01 -3.16862650e-02 8.96187246e-01 9.74981189e-01
4.42604423e-01 2.77329415e-01 -5.51687837e-01 9.47096586e-01
3.44339252e-01 6.10962093e-01 7.90341496e-01 -6.27488911e-01
8.76148462e-01 5.22038937e-01 4.96651046e-02 -1.76355988e-01
-5.47053926e-02 -1.94900081e-01 -5.34653962e-01 1.30025283e-01
1.04201019e+00 -3.46394718e-01 -1.09252930e+00 1.58784103e+00
-1.59939863e-02 -1.43031567e-01 2.97811687e-01 6.95910037e-01
7.07365811e-01 1.09272218e+00 -8.52017477e-02 -2.26095453e-01
1.30282366e+00 -1.41955805e+00 -8.81960928e-01 -5.62131941e-01
8.57678771e-01 -1.01792300e+00 1.77776384e+00 9.17075515e-01
-1.50050628e+00 -4.77614403e-01 -1.23530400e+00 -1.63883612e-01
-4.00543451e-01 3.97186875e-01 2.44939700e-01 9.93982434e-01
-1.01144397e+00 3.95616561e-01 -8.54698241e-01 2.31669158e-01
-1.74116179e-01 4.09967214e-01 -1.00556873e-01 4.30259891e-02
-1.27591980e+00 1.10236692e+00 4.20294285e-01 3.37731801e-02
-6.62691653e-01 -4.36831534e-01 -7.56415009e-01 2.86367405e-02
3.02847624e-01 4.39753942e-02 1.71938503e+00 -1.11840451e+00
-2.09077406e+00 8.18101704e-01 -2.22852781e-01 -5.99076092e-01
4.19312716e-01 -7.09748089e-01 -1.96534052e-01 -1.41680717e-01
-3.73752534e-01 6.26875222e-01 7.30019867e-01 -1.01247311e+00
-7.65828192e-01 -1.65218160e-01 -4.56828885e-02 3.96972746e-01
-4.67223406e-01 3.76325130e-01 -3.30530345e-01 -9.51480985e-01
1.14556640e-01 -9.24306273e-01 -3.88664640e-02 -6.97737396e-01
-6.67462587e-01 -5.00333428e-01 5.28758049e-01 -9.99311626e-01
1.61437345e+00 -1.67586124e+00 2.56688476e-01 1.94162652e-01
-4.38712209e-01 7.09806919e-01 -1.21901810e-01 3.62320662e-01
2.15296894e-01 6.99429661e-02 -2.73085207e-01 -5.43221176e-01
1.30682617e-01 2.26604834e-01 -5.85769653e-01 -1.04768477e-01
5.38704693e-01 8.87939632e-01 -8.72928917e-01 -1.29054680e-01
7.36993775e-02 1.51689932e-01 -6.05516553e-01 3.12299967e-01
-6.80362344e-01 3.59812289e-01 1.75154470e-02 5.49153686e-01
2.61932760e-01 1.60828605e-01 2.52087146e-01 5.13732851e-01
-8.22891891e-02 1.23123527e+00 -1.04348862e+00 1.43405521e+00
-6.65318549e-01 4.15260285e-01 -7.77358189e-02 -7.39166081e-01
7.93433845e-01 4.32068944e-01 -4.48344648e-01 -4.67027932e-01
2.05329731e-01 5.89347243e-01 5.28113544e-01 5.71218021e-02
6.91161335e-01 2.49807835e-01 -8.73124152e-02 6.56767249e-01
4.46887240e-02 -2.65111208e-01 3.37897658e-01 1.30840465e-01
8.75812888e-01 2.77922988e-01 3.14693660e-01 -2.15888977e-01
8.75967860e-01 -1.41071633e-01 1.93787236e-02 7.52649963e-01
9.95455310e-02 6.94076836e-01 4.31632668e-01 -6.52555972e-02
-5.70560336e-01 -1.08300650e+00 1.33739084e-01 1.91716838e+00
-2.39172384e-01 -6.02709055e-01 -1.09581900e+00 -9.33268011e-01
-2.86140412e-01 1.36768591e+00 -3.45962137e-01 4.58787903e-02
-1.15345728e+00 -6.88849568e-01 7.25025952e-01 6.33455753e-01
1.45738631e-01 -1.33798349e+00 -1.58670500e-01 3.16600293e-01
-2.47661695e-01 -9.16913509e-01 -4.06440854e-01 5.71220100e-01
-6.57136619e-01 -6.04334652e-01 -6.82624459e-01 -1.15279603e+00
5.18000126e-01 -1.98300891e-02 1.26312208e+00 3.47023994e-01
4.89395887e-01 -1.29215434e-01 -6.88533306e-01 -7.47115433e-01
-9.48354244e-01 4.42497283e-01 -1.46820933e-01 -1.38708591e-01
5.09077251e-01 -1.07019514e-01 -3.54509689e-02 3.12981129e-01
-6.69117033e-01 2.25085020e-01 5.55773079e-01 8.94869208e-01
4.31650430e-01 -6.12038791e-01 8.15364480e-01 -1.37863588e+00
6.29805982e-01 -2.46657699e-01 -4.65474248e-01 3.32045645e-01
-6.48763120e-01 4.13422644e-01 9.85272646e-01 -4.56982434e-01
-1.16265309e+00 1.15731154e-02 -6.99128330e-01 4.01822507e-01
-2.03131035e-01 5.59751451e-01 -3.79240960e-01 5.88909805e-01
8.22909951e-01 2.80499876e-01 -1.40208770e-02 -5.53940952e-01
5.80980062e-01 9.21620131e-01 4.92942601e-01 -6.83604598e-01
6.69713974e-01 -2.20572561e-01 -4.24210042e-01 -6.76152766e-01
-1.14206028e+00 -2.71934360e-01 -7.17895210e-01 1.42060980e-01
4.34205681e-01 -8.24581981e-01 -3.90690506e-01 6.35847509e-01
-1.24069250e+00 -8.58317554e-01 -1.08263895e-01 1.98489919e-01
-3.11083019e-01 2.74873167e-01 -1.02345347e+00 -7.50062943e-01
-3.63485634e-01 -1.28513014e+00 8.90558124e-01 5.95614128e-02
-3.22432518e-01 -1.02926719e+00 -2.05500588e-01 3.19465935e-01
4.09817994e-01 -6.00712776e-01 1.07179058e+00 -1.20633912e+00
-4.81145829e-01 9.19196755e-02 1.52014285e-01 7.36585855e-01
1.75756499e-01 -6.05512597e-02 -1.15867233e+00 -2.10723821e-02
-5.21875858e-01 -5.51041782e-01 7.93034673e-01 5.86658046e-02
8.92183423e-01 -5.93187809e-01 2.22771287e-01 1.72883391e-01
5.94530404e-01 6.60377443e-02 6.83723748e-01 3.86795312e-01
5.85655332e-01 6.31446719e-01 8.97456825e-01 1.73725262e-01
4.58380699e-01 6.92858934e-01 4.45479035e-01 1.45834178e-01
-3.56368661e-01 -3.68193209e-01 9.80798662e-01 1.22023761e+00
-1.21653294e-02 -6.26484036e-01 -8.41801941e-01 7.10679114e-01
-1.62121344e+00 -5.05398750e-01 -4.34690803e-01 2.24640727e+00
1.67880297e+00 3.45318913e-01 1.51485041e-01 6.08899593e-01
5.71840107e-01 -1.20444059e-01 1.14439406e-01 -9.31918919e-01
-2.31586650e-01 8.37002695e-01 1.58440143e-01 1.02015793e+00
-8.83653581e-01 1.66596079e+00 5.93448305e+00 9.26562548e-01
-1.04047632e+00 1.99903294e-01 6.78350806e-01 3.19801897e-01
-4.05959219e-01 -2.89253265e-01 -1.36964631e+00 3.14300805e-01
1.26934755e+00 1.89848721e-01 4.18456465e-01 6.51768208e-01
8.72251466e-02 2.72991788e-02 -1.13012075e+00 3.55509907e-01
9.61427614e-02 -9.67269599e-01 1.98021680e-01 -4.59740043e-01
8.75573218e-01 -2.87274957e-01 2.61744499e-01 5.13808250e-01
5.32761753e-01 -1.10791218e+00 9.41306412e-01 -4.44672927e-02
8.73829484e-01 -8.49985063e-01 6.64955556e-01 5.59257448e-01
-5.68565011e-01 9.53821540e-02 -1.06425747e-01 -4.51478064e-01
-7.31314020e-03 2.27401063e-01 -1.48101425e+00 1.26476139e-01
5.69308624e-02 1.68385446e-01 -8.31441283e-01 7.55777240e-01
-1.28721893e+00 1.63596451e+00 -3.28725845e-01 -5.47366440e-01
4.31125432e-01 -1.20635688e-01 3.69283676e-01 1.62431252e+00
2.77274758e-01 -1.01320058e-01 -2.87079532e-02 1.64419860e-01
2.17916369e-02 6.28297627e-01 -2.34678477e-01 1.26699701e-01
6.03199124e-01 9.43655670e-01 -5.56450903e-01 -3.44745725e-01
-4.60273415e-01 1.25412333e+00 4.97016996e-01 1.21229559e-01
-6.84019744e-01 -4.31627750e-01 3.42334062e-01 1.22101165e-01
4.09751266e-01 -3.98240089e-01 -4.27590102e-01 -8.28256428e-01
-6.60293624e-02 -1.28544414e+00 3.72267008e-01 -4.68181401e-01
-9.32822764e-01 8.40236664e-01 -2.40550652e-01 -9.57412183e-01
-9.12387490e-01 -1.18052602e+00 -4.10357922e-01 1.08510828e+00
-1.64164460e+00 -1.18465960e+00 1.70804605e-01 3.35978925e-01
9.76934314e-01 -3.63293499e-01 1.14477015e+00 5.88574931e-02
-4.04298365e-01 1.13905489e+00 4.84095477e-02 3.52651238e-01
9.86803830e-01 -1.77361965e+00 5.98285079e-01 1.31051910e+00
5.90254366e-01 6.51895523e-01 6.27152860e-01 -4.47166324e-01
-1.10026371e+00 -1.02402306e+00 1.47716463e+00 -6.69558167e-01
5.99094510e-01 -3.74905288e-01 -9.56900001e-01 9.07980144e-01
5.37415445e-01 -3.01565915e-01 6.47497654e-01 2.15304866e-01
-1.50475264e-01 7.37561211e-02 -6.30888045e-01 9.29829121e-01
7.35669136e-01 -1.94344074e-01 -7.91260183e-01 4.06092435e-01
8.50520015e-01 -8.04302871e-01 -2.64732808e-01 7.18887523e-02
3.06773186e-01 -8.80389452e-01 6.75357699e-01 -8.49656045e-01
5.79084814e-01 -2.77160376e-01 -2.00259805e-01 -1.80058908e+00
6.17224500e-02 -9.91949201e-01 -3.39666665e-01 1.16164279e+00
1.08101773e+00 -3.88879806e-01 5.68505049e-01 1.61858931e-01
-7.13554561e-01 -6.64900362e-01 -6.77707791e-01 -6.36297822e-01
3.64860535e-01 -6.40408337e-01 4.20610905e-01 5.86507797e-01
3.05130810e-01 7.28838742e-01 -1.51225790e-01 9.79593769e-02
-1.84150413e-01 -2.49139383e-01 5.70428729e-01 -8.75376642e-01
-4.81732070e-01 -4.45432305e-01 7.48302564e-02 -1.60009444e+00
3.41636688e-01 -9.92981911e-01 4.48404729e-01 -1.13480854e+00
-5.85600078e-01 -6.88491762e-01 -1.18212193e-01 8.94055367e-01
-6.84404314e-01 4.43068951e-01 1.69214353e-01 -3.46183449e-01
-4.14530843e-01 2.16695875e-01 1.04920304e+00 9.17126089e-02
-4.44792569e-01 5.78191102e-01 -6.72265887e-01 8.06710541e-01
9.04859185e-01 -2.53745586e-01 -3.43238860e-01 -7.63390660e-01
1.85486197e-01 -1.03597097e-01 -2.38858670e-01 -8.55747700e-01
6.04968937e-03 -9.95108560e-02 -1.83127709e-02 -5.48918486e-01
2.53354758e-01 -3.73566031e-01 -8.40850294e-01 3.20121199e-01
-5.40347695e-01 3.60541433e-01 3.05619031e-01 -1.72939539e-01
-1.75443575e-01 -8.37818205e-01 6.81867599e-01 2.99088676e-02
-4.42696542e-01 -2.24344358e-01 -7.81876147e-01 3.71062160e-01
5.71025252e-01 6.72094151e-02 -3.52091104e-01 -4.61910337e-01
-6.04931593e-01 -1.40962631e-01 1.02259420e-01 3.56715083e-01
4.57340211e-01 -8.89085352e-01 -9.65480328e-01 4.28993106e-01
-1.48203582e-01 5.74533641e-02 -5.67678511e-01 5.94200194e-01
-6.92861736e-01 6.06507301e-01 7.58255348e-02 -3.55975330e-01
-1.38542008e+00 7.23433867e-02 4.87016030e-02 -4.24547315e-01
-1.60554260e-01 1.36698508e+00 -1.87718868e-01 -7.10131407e-01
6.29029572e-01 -6.42536998e-01 -3.07278007e-01 1.72145441e-01
7.55264223e-01 2.94030786e-01 7.04468071e-01 -5.52402794e-01
-1.14242829e-01 -1.53714508e-01 -3.88936579e-01 -6.54418290e-01
9.64323580e-01 -2.34661605e-02 3.74425501e-02 6.33817554e-01
4.85709667e-01 7.75854409e-01 -1.04139197e+00 -3.78185093e-01
3.76140177e-01 -2.00268943e-02 -1.95146278e-01 -1.40168405e+00
-3.96352828e-01 1.23494458e+00 -7.33685866e-02 2.41911218e-01
1.06635070e+00 -3.52792859e-01 7.75147259e-01 7.56453216e-01
1.26426592e-01 -1.10778916e+00 2.74519265e-01 1.31920254e+00
8.89967203e-01 -1.18105698e+00 -4.64187294e-01 -5.84034443e-01
-1.07212591e+00 1.12235045e+00 8.52399588e-01 1.14788927e-01
3.83976758e-01 4.06719953e-01 4.81275737e-01 5.52580118e-01
-8.54267240e-01 -2.67059654e-01 4.54539627e-01 4.99841690e-01
1.13216925e+00 9.98029932e-02 -4.98255789e-01 1.05348909e+00
-8.04499030e-01 -7.05080986e-01 7.44441152e-01 8.52533281e-01
-6.25606358e-01 -1.67007208e+00 -4.25016046e-01 3.30186605e-01
-7.17835665e-01 -7.52817750e-01 -5.90753376e-01 6.47553384e-01
-7.30488077e-02 1.09142613e+00 -1.94503758e-02 -1.68680057e-01
1.77858278e-01 4.89555240e-01 6.00225091e-01 -1.18573952e+00
-9.39782798e-01 4.91358303e-02 3.44382614e-01 5.41029789e-04
2.10772038e-01 -7.05736876e-01 -1.60672331e+00 1.15814358e-01
-5.37970126e-01 3.13825905e-01 5.19024789e-01 1.19760418e+00
-1.43962726e-01 3.67768139e-01 5.28094530e-01 -6.76302910e-01
-9.01798069e-01 -1.49061942e+00 -2.71015555e-01 -1.94954011e-03
3.22122991e-01 -1.67089060e-01 -2.88860172e-01 1.70775965e-01] | [10.860761642456055, 10.28982162475586] |
f7b72f2d-354f-452b-8ba5-ee0b4c8b63dd | automated-topical-component-extraction-using-1 | 2008.01809 | null | https://arxiv.org/abs/2008.01809v1 | https://arxiv.org/pdf/2008.01809v1.pdf | Automated Topical Component Extraction Using Neural Network Attention Scores from Source-based Essay Scoring | While automated essay scoring (AES) can reliably grade essays at scale, automated writing evaluation (AWE) additionally provides formative feedback to guide essay revision. However, a neural AES typically does not provide useful feature representations for supporting AWE. This paper presents a method for linking AWE and neural AES, by extracting Topical Components (TCs) representing evidence from a source text using the intermediate output of attention layers. We evaluate performance using a feature-based AES requiring TCs. Results show that performance is comparable whether using automatically or manually constructed TCs for 1) representing essays as rubric-based features, 2) grading essays. | ['Haoran Zhang', 'Diane Litman'] | 2020-08-04 | automated-topical-component-extraction-using | https://aclanthology.org/2020.acl-main.759 | https://aclanthology.org/2020.acl-main.759.pdf | acl-2020-6 | ['automated-essay-scoring', 'automated-writing-evaluation'] | ['natural-language-processing', 'natural-language-processing'] | [-5.39897159e-02 1.68714561e-02 -3.91778089e-02 -3.84067923e-01
-1.04736710e+00 -7.78634250e-01 6.38061643e-01 6.99559212e-01
-5.09780049e-01 8.38286817e-01 6.06326342e-01 -3.78027201e-01
-3.40685278e-01 -7.54701912e-01 -2.52837509e-01 -4.57565933e-02
6.42932534e-01 1.56853959e-01 -1.83410242e-01 -3.08248132e-01
1.21946156e+00 9.19187516e-02 -1.14048362e+00 5.39305568e-01
1.15555787e+00 9.74026442e-01 -2.37164367e-02 1.16100383e+00
-6.76145673e-01 1.18275154e+00 -1.40546393e+00 -8.25740814e-01
-1.30528480e-01 -6.70194387e-01 -8.38956177e-01 -3.95831198e-01
1.11124098e+00 -2.77721494e-01 -1.79366678e-01 9.71560955e-01
2.44461462e-01 1.24369182e-01 1.10114801e+00 -9.07721221e-01
-1.40983891e+00 7.14365304e-01 -1.67917401e-01 4.16372418e-01
7.26368725e-01 1.20787583e-01 1.56457567e+00 -1.13755727e+00
4.20171380e-01 5.12132764e-01 9.12089884e-01 5.46030462e-01
-8.68767440e-01 -5.70195615e-01 -1.99919537e-01 3.95322174e-01
-6.40729129e-01 -2.04759210e-01 9.24597919e-01 -4.49335098e-01
1.12567079e+00 3.21460634e-01 7.61757791e-01 1.02046990e+00
5.90064526e-01 1.08054352e+00 1.46791959e+00 -6.18346334e-01
1.61075652e-01 1.80642560e-01 1.19769871e+00 7.55483329e-01
2.67236710e-01 -2.71392673e-01 -1.12601697e+00 -1.95878312e-01
1.84032917e-01 -5.16546845e-01 -2.13960916e-01 7.29109287e-01
-8.88242722e-01 7.90393054e-01 2.41080284e-01 2.56670825e-02
-5.64848185e-01 1.53324038e-01 5.10804832e-01 9.13976610e-01
3.50132614e-01 1.31720328e+00 -3.94172072e-01 -4.91326541e-01
-1.51497054e+00 4.29898292e-01 9.43528891e-01 3.07324797e-01
4.58906621e-01 2.59770721e-01 -9.62264299e-01 9.44578290e-01
9.53044295e-02 1.89523682e-01 1.04349136e+00 -8.08341086e-01
6.01523697e-01 9.12981272e-01 1.52807727e-01 -1.07733488e+00
-3.79585326e-01 -4.21679795e-01 -3.29479426e-01 1.24211006e-01
2.80877531e-01 -1.66558981e-01 -5.64749658e-01 1.45275235e+00
-5.16070127e-01 -4.95344907e-01 -7.31316358e-02 7.57001877e-01
1.37965822e+00 6.53804123e-01 1.00273825e-02 8.05835724e-02
1.23920345e+00 -8.60715628e-01 -9.43355680e-01 -8.64866525e-02
5.79235017e-01 -7.86335349e-01 1.52457070e+00 8.17603052e-01
-1.61623347e+00 -6.31228685e-01 -1.44312167e+00 -3.30024123e-01
-5.05053163e-01 7.49000669e-01 2.42575765e-01 7.81135023e-01
-1.25615561e+00 5.04307747e-01 -2.34289363e-01 2.59325147e-01
5.18600702e-01 2.47334540e-01 -1.86237454e-01 4.91884053e-01
-1.45138907e+00 1.32715285e+00 -1.09315723e-01 -2.16091871e-01
-5.36430657e-01 -9.43880916e-01 -8.52936506e-01 5.01223326e-01
-3.68208587e-01 -4.55435723e-01 1.44618714e+00 -8.26488018e-01
-1.85231364e+00 6.26406789e-01 -5.80718592e-02 -4.22741383e-01
9.46930572e-02 -3.63197744e-01 -1.40999839e-01 2.70064678e-02
-1.19353913e-01 3.51658940e-01 5.31900644e-01 -4.86549735e-01
-2.99860746e-01 -1.87455401e-01 1.90690503e-01 3.89310092e-01
-1.20455599e+00 2.12821394e-01 2.72102416e-01 -8.49908650e-01
-2.49264732e-01 -3.95407379e-01 2.73851961e-01 -2.86607087e-01
-3.71720314e-01 -1.01576138e+00 4.66351777e-01 -1.16100311e+00
1.87227190e+00 -1.44300354e+00 -1.02554180e-03 3.12478781e-01
3.37230593e-01 1.41031504e-01 -3.24877828e-01 2.45317087e-01
2.80211657e-01 1.58967957e-01 -1.08394250e-01 -1.47881031e-01
3.87725919e-01 -5.25786102e-01 -4.02500957e-01 -7.43669420e-02
3.70513976e-01 1.17870665e+00 -8.09273243e-01 -2.14641631e-01
-2.55474836e-01 -7.98441097e-02 -3.34049731e-01 3.28735709e-01
-1.91141367e-01 -5.59366643e-01 -1.50539219e-01 5.13484478e-01
1.44066112e-02 -3.29364419e-01 -2.33571246e-01 2.73712784e-01
-4.77589741e-02 1.14653897e+00 -7.09356368e-01 1.44125700e+00
-6.99463069e-01 1.50616217e+00 -3.97893697e-01 -2.85338223e-01
1.53149676e+00 3.01425487e-01 -2.34552443e-01 -6.86199367e-01
-5.56242727e-02 2.12581038e-01 5.46986386e-02 -2.00680226e-01
1.13499260e+00 3.13805342e-01 -2.44042143e-01 1.51192391e+00
4.12988454e-01 -4.53528315e-01 5.44552267e-01 6.27983153e-01
1.45859385e+00 1.79042369e-02 4.46932554e-01 -3.86188716e-01
6.32213295e-01 1.40972603e-02 -7.98054412e-02 8.14382255e-01
5.13453793e-04 6.21011138e-01 6.56232238e-01 -4.29021269e-01
-1.03185713e+00 -8.22253227e-01 -9.15293768e-03 1.12471330e+00
-7.13076711e-01 -9.89127576e-01 -7.49419689e-01 -8.79773140e-01
-1.55073414e-02 1.03824794e+00 -8.42988312e-01 -2.50508875e-01
-2.86179155e-01 -6.90028369e-01 8.33000839e-01 8.07098448e-01
3.17849517e-01 -1.31753588e+00 -7.68465519e-01 3.94950032e-01
-1.37662649e-01 -2.23958150e-01 -5.90252280e-01 4.55095381e-01
-6.57411993e-01 -9.77857471e-01 -7.99453437e-01 -4.21141207e-01
5.05259395e-01 2.42117252e-02 1.43185866e+00 3.77340108e-01
-1.66660368e-01 6.05550587e-01 -7.16056377e-02 -7.99453139e-01
-4.06416595e-01 2.96843946e-01 -5.74829467e-02 -5.29810429e-01
1.06637692e+00 -1.49068370e-01 -3.35040838e-01 -2.71512657e-01
-5.67129910e-01 -2.50490874e-01 4.78660703e-01 1.02617908e+00
-9.86118093e-02 -7.00514257e-01 9.44043398e-01 -8.44874322e-01
2.01782322e+00 -3.90374243e-01 -2.65896559e-01 4.22334015e-01
-9.88751769e-01 3.68776619e-02 7.62098193e-01 -2.26717681e-01
-1.05215454e+00 -6.26652479e-01 4.45588194e-02 8.17359686e-02
7.86180571e-02 9.86108601e-01 6.85527742e-01 -6.53998852e-02
1.22472763e+00 3.25465322e-01 -1.74055800e-01 -2.19039589e-01
-1.11550413e-01 9.25999701e-01 5.66442788e-01 -8.43251646e-01
3.33774149e-01 -4.83248860e-01 -2.99555391e-01 -2.94903964e-01
-1.31325245e+00 -2.60339081e-01 -5.45243859e-01 -4.97226149e-01
2.69308567e-01 -5.77405334e-01 -5.81018806e-01 1.78332418e-01
-1.33431840e+00 -1.82033867e-01 -2.31580049e-01 3.93246204e-01
-2.65172333e-01 3.36927354e-01 -1.09490800e+00 -5.58558762e-01
-8.82945180e-01 -7.15795636e-01 5.61323166e-01 6.18374944e-01
-1.11228585e+00 -1.03589809e+00 7.20387638e-01 3.92492831e-01
5.30127347e-01 -2.03694865e-01 1.00673485e+00 -8.99410605e-01
1.17515177e-01 -4.24983680e-01 -2.20045149e-01 4.77929652e-01
-2.11194247e-01 3.08349252e-01 -1.20255697e+00 2.04537645e-01
-8.74868855e-02 -9.24037039e-01 1.02409220e+00 4.03612167e-01
1.59035778e+00 -5.61923027e-01 6.11884058e-01 2.02414289e-01
1.10748291e+00 -1.76120564e-01 5.70350587e-01 5.05830646e-01
3.35098982e-01 5.56062877e-01 2.56539404e-01 5.21966040e-01
4.02514607e-01 1.99186757e-01 -2.41103709e-01 4.08721179e-01
-3.25498223e-01 -2.94486340e-02 9.82711196e-01 1.30518174e+00
-6.74811676e-02 -1.08507141e-01 -1.18605769e+00 5.84767997e-01
-1.48749542e+00 -9.28738713e-01 -4.92176443e-01 1.75301766e+00
1.41477585e+00 9.29407850e-02 -1.19926378e-01 4.09847051e-01
3.15342933e-01 1.25583023e-01 -3.59173030e-01 -1.26798224e+00
-2.16536894e-01 8.14360023e-01 -6.00258596e-02 5.09252906e-01
-6.99765265e-01 4.29366797e-01 7.02178097e+00 5.36404014e-01
-7.11340129e-01 -5.25642596e-02 4.90833253e-01 -2.49049187e-01
-6.53379798e-01 -4.69056636e-01 -7.28702545e-01 4.85605508e-01
1.34015930e+00 -5.55440009e-01 1.06316525e-02 6.44519627e-01
-1.61748186e-01 -7.87289813e-02 -1.10809922e+00 5.51725924e-01
5.00419319e-01 -1.54318178e+00 2.79485703e-01 -3.50112468e-01
1.04961920e+00 -3.82008582e-01 4.23802525e-01 5.42552888e-01
6.61790133e-01 -1.19604349e+00 8.78661513e-01 8.11515450e-01
7.99763739e-01 -6.76482558e-01 9.08391178e-01 6.54604286e-02
-5.47287226e-01 -8.15329999e-02 -5.65722942e-01 -5.41450799e-01
-5.85765064e-01 3.40580791e-01 -1.07481146e+00 7.87486061e-02
3.29471439e-01 6.94815159e-01 -1.26602089e+00 1.04465389e+00
-7.84600079e-01 1.06236613e+00 1.22181073e-01 -8.06652248e-01
6.17639840e-01 5.38837798e-02 3.51661325e-01 1.69844329e+00
4.59355325e-01 9.51116066e-03 -3.98120850e-01 1.12729096e+00
-4.41253155e-01 2.66239494e-01 -2.22999543e-01 -2.65282333e-01
4.49619025e-01 1.44482803e+00 -3.48822057e-01 -3.49113673e-01
-2.70680934e-01 7.20967829e-01 5.58028698e-01 1.53235465e-01
-3.19403529e-01 -9.32984829e-01 4.81212661e-02 -3.10393572e-01
-3.55651826e-01 1.24169886e-02 -1.07881010e+00 -1.03591681e+00
6.88393414e-02 -7.89674282e-01 4.91425633e-01 -1.16814184e+00
-1.52851880e+00 3.98662835e-01 -7.07016230e-01 -7.22018301e-01
-2.90798068e-01 -1.03574908e+00 -1.31859529e+00 1.23240244e+00
-1.35411525e+00 -5.73034346e-01 -5.25605023e-01 2.00328559e-01
9.00307596e-01 -6.10412776e-01 1.10217512e+00 -2.04226002e-01
-3.89405638e-01 8.95298004e-01 2.02938379e-03 3.25672954e-01
1.08052433e+00 -2.07411003e+00 1.14934064e-01 6.61518097e-01
4.68744040e-01 8.50232899e-01 4.11439687e-01 -5.42607963e-01
-8.84697139e-01 -5.84010422e-01 1.23198450e+00 -9.24724936e-01
7.79192865e-01 2.82494146e-02 -1.14854920e+00 3.77344877e-01
6.74657643e-01 -7.85429180e-01 1.22225595e+00 5.64342141e-01
-6.15758002e-01 2.02922404e-01 -7.67425537e-01 5.28402150e-01
1.86270416e-01 -9.36490297e-01 -1.28279626e+00 1.49975136e-01
4.00801450e-02 -2.91386127e-01 -8.83321345e-01 -1.03219748e-01
7.46885538e-01 -6.01761222e-01 4.38757509e-01 -9.49881256e-01
1.40271986e+00 1.86536256e-02 3.23805064e-01 -1.88927722e+00
-6.43199146e-01 -4.81817186e-01 -6.52091682e-01 8.58235836e-01
6.28194392e-01 -3.18406895e-03 8.78445387e-01 7.14556217e-01
-4.10783708e-01 -9.42107975e-01 -5.12441456e-01 -3.09304237e-01
4.75334167e-01 -2.74810106e-01 6.44476593e-01 1.13000321e+00
7.27310777e-01 4.61632729e-01 1.86287239e-01 -5.82596064e-01
2.83583462e-01 -1.21158555e-01 3.80299181e-01 -1.53677034e+00
1.77993983e-01 -1.32941973e+00 -1.76593274e-01 -4.24159437e-01
3.25827241e-01 -1.12361884e+00 -9.92849562e-03 -1.78083014e+00
5.39599419e-01 7.13152392e-03 -6.94265068e-01 5.59821844e-01
-5.45797288e-01 4.84271437e-01 3.29241790e-02 3.93765833e-04
-6.17188871e-01 3.61587137e-01 9.53372717e-01 -3.04930151e-01
1.69217497e-01 -2.00406089e-01 -8.88740778e-01 5.95530152e-01
9.89511192e-01 -2.68855870e-01 -1.65604264e-01 -4.06110197e-01
9.97331262e-01 -1.29696622e-01 1.43058836e-01 -8.27491641e-01
7.64904261e-01 -1.82987880e-02 9.09363985e-01 -5.35653710e-01
-1.29261866e-01 9.91020650e-02 -6.91199720e-01 2.13967323e-01
-1.24775398e+00 4.68792260e-01 1.90207958e-01 4.01406139e-02
-3.93201828e-01 -1.04956603e+00 4.22625333e-01 -1.94660991e-01
-3.19162935e-01 -1.39297768e-01 -6.71478868e-01 -4.05875519e-02
2.81824976e-01 -2.62647063e-01 -7.86525071e-01 -4.99869645e-01
-3.57334703e-01 1.16923556e-01 6.03445172e-02 3.53718966e-01
1.00557435e+00 -1.36076033e+00 -1.24320495e+00 -1.69640556e-02
7.92733729e-02 -6.55144691e-01 -8.02475810e-02 6.57104254e-01
-5.06871521e-01 8.13881695e-01 -5.80729544e-01 2.93163839e-03
-1.33013248e+00 -4.19504166e-01 8.88132975e-02 -5.76850891e-01
-1.57761589e-01 1.10151613e+00 -3.38734657e-01 -4.00903314e-01
2.28168994e-01 -3.64110321e-02 -8.11820865e-01 4.59025532e-01
1.00530469e+00 6.23369634e-01 5.25762856e-01 7.67987743e-02
3.27429712e-01 1.21014461e-01 -4.18077201e-01 -4.08899993e-01
1.30926442e+00 3.57435018e-01 -2.15707526e-01 6.80994391e-01
5.22612631e-01 2.10051388e-01 -8.63992631e-01 -4.85455096e-02
3.51219118e-01 -3.13611180e-01 3.87684613e-01 -1.54157555e+00
-5.38134456e-01 1.20166636e+00 -2.21259017e-02 2.44772136e-01
6.71110988e-01 -7.35999584e-01 6.92367792e-01 9.78697240e-01
-8.69848207e-02 -1.62557566e+00 4.60302025e-01 8.75763297e-01
1.18744671e+00 -9.39193785e-01 2.03817829e-01 5.43606818e-01
-7.47476995e-01 1.90737319e+00 1.01864409e+00 -4.26874101e-01
2.07840115e-01 2.45763838e-01 -4.07758020e-02 -3.99299026e-01
-1.22964704e+00 5.94204545e-01 1.12761438e+00 6.98521957e-02
1.03198147e+00 -6.01095036e-02 -6.84677899e-01 1.23447788e+00
-6.46711349e-01 -1.71459973e-01 1.22029293e+00 5.76348662e-01
-6.58447683e-01 -7.56935894e-01 -3.97071004e-01 1.37462986e+00
-6.77510083e-01 -4.50220466e-01 -1.13527942e+00 3.00868362e-01
-5.37693739e-01 6.19943321e-01 2.01363936e-02 -3.17277372e-01
-1.28699690e-01 7.50864685e-01 6.46347404e-01 -8.59632254e-01
-1.59885049e+00 -8.53102684e-01 1.77390844e-01 -1.31962433e-01
-2.58032810e-02 -6.33135676e-01 -7.74897695e-01 -2.88208336e-01
-5.26509047e-01 3.75124633e-01 8.03474188e-01 7.49325514e-01
8.40142593e-02 7.67088175e-01 4.24875826e-01 -2.26322800e-01
-9.69248235e-01 -1.47571802e+00 -3.28981608e-01 1.50704086e-01
1.75811350e-01 -4.54194099e-01 -4.18759555e-01 -1.05368294e-01] | [11.300117492675781, 9.338724136352539] |
8257483b-35e1-4996-8477-923ed9404a45 | auto-weighted-multi-view-feature-selection | 2104.04906 | null | https://arxiv.org/abs/2104.04906v1 | https://arxiv.org/pdf/2104.04906v1.pdf | Auto-weighted Multi-view Feature Selection with Graph Optimization | In this paper, we focus on the unsupervised multi-view feature selection which tries to handle high dimensional data in the field of multi-view learning. Although some graph-based methods have achieved satisfactory performance, they ignore the underlying data structure across different views. Besides, their pre-defined laplacian graphs are sensitive to the noises in the original data space, and fail to get the optimal neighbor assignment. To address the above problems, we propose a novel unsupervised multi-view feature selection model based on graph learning, and the contributions are threefold: (1) during the feature selection procedure, the consensus similarity graph shared by different views is learned. Therefore, the proposed model can reveal the data relationship from the feature subset. (2) a reasonable rank constraint is added to optimize the similarity matrix to obtain more accurate information; (3) an auto-weighted framework is presented to assign view weights adaptively, and an effective alternative iterative algorithm is proposed to optimize the problem. Experiments on various datasets demonstrate the superiority of the proposed method compared with the state-of-the-art methods. | ['Xuelong Li', 'Mulin Chen', 'Xu Jiang', 'Qi Wang'] | 2021-04-11 | null | null | null | null | ['multi-view-learning'] | ['computer-vision'] | [-1.00090824e-01 -4.50102806e-01 -2.16713682e-01 -3.33473146e-01
-4.62746263e-01 -2.19648153e-01 2.07474247e-01 -1.38993442e-01
4.13546860e-02 2.42911443e-01 4.38509375e-01 3.23950648e-01
-7.21850514e-01 -7.75431752e-01 -1.45812500e-02 -9.19489145e-01
2.61787057e-01 1.05853364e-01 2.78669029e-01 -2.35808060e-01
4.21067357e-01 1.56971008e-01 -1.34518206e+00 2.10702904e-02
1.09055579e+00 8.50508928e-01 3.19473475e-01 -1.01068765e-01
-4.40876037e-02 4.26308453e-01 -9.34584215e-02 -1.34124309e-01
2.10316837e-01 -6.64951444e-01 -3.76460820e-01 6.81876600e-01
2.28180319e-01 1.53884845e-04 -3.34324241e-01 1.49214625e+00
6.54141128e-01 7.67645687e-02 6.06023669e-01 -1.23824298e+00
-6.72090471e-01 4.03095514e-01 -1.03536391e+00 5.98593205e-02
2.24286854e-01 -1.97252646e-01 1.26713765e+00 -1.30206859e+00
5.65168738e-01 1.28423738e+00 2.41198242e-01 1.05814494e-01
-1.18899536e+00 -6.32849514e-01 5.17063737e-01 5.82138896e-01
-1.61414993e+00 -2.53167182e-01 1.30140030e+00 -3.73542249e-01
1.35602206e-01 1.94995314e-01 8.61614585e-01 7.76802301e-01
1.39432892e-01 6.04875267e-01 1.15476120e+00 -8.27724859e-02
7.72480713e-03 6.88543171e-02 1.82137087e-01 9.55773115e-01
3.30957264e-01 -2.66894192e-01 -4.39014941e-01 -4.47612256e-01
4.66251642e-01 4.30718392e-01 -6.19089365e-01 -1.06063843e+00
-1.36561453e+00 9.59655285e-01 3.88479829e-01 3.88808608e-01
-3.45368892e-01 -4.62822676e-01 5.61926365e-01 2.58631706e-01
4.17758554e-01 8.23131576e-03 -2.29512036e-01 3.49947751e-01
-5.56261420e-01 -2.37090558e-01 4.52964038e-01 9.25647199e-01
1.00655508e+00 3.79838608e-02 1.48091406e-01 9.91868079e-01
5.48372209e-01 3.83680910e-01 3.95868361e-01 -5.06673098e-01
8.19553375e-01 1.13824213e+00 -1.83791056e-01 -1.65841234e+00
-5.03369570e-01 -4.77396846e-01 -1.22383058e+00 -1.86099745e-02
1.34808267e-03 -1.55242324e-01 -4.96422231e-01 1.40940833e+00
4.55624521e-01 8.60376731e-02 -1.53853493e-02 1.00269222e+00
9.09600556e-01 4.57906544e-01 -4.13764983e-01 -6.39051676e-01
1.12245440e+00 -7.80226111e-01 -7.56701708e-01 9.03171450e-02
2.14752585e-01 -7.03138351e-01 8.90861273e-01 4.01035577e-01
-7.70506561e-01 -5.64570665e-01 -1.25211740e+00 4.17681634e-01
-2.90531479e-03 1.94380686e-01 5.26324570e-01 3.58789235e-01
-6.27396822e-01 2.56503761e-01 -5.69532335e-01 -3.65797967e-01
1.81208044e-01 3.77867550e-01 -5.58787644e-01 -3.65819663e-01
-1.06691444e+00 4.64807928e-01 5.23655832e-01 3.10214430e-01
-4.19594437e-01 -3.14392120e-01 -7.55332410e-01 6.68976232e-02
7.86845386e-01 -7.92040765e-01 3.58220905e-01 -9.17583525e-01
-1.30170166e+00 3.79639506e-01 -2.28013650e-01 4.07835364e-01
3.97570431e-01 8.27448815e-02 -7.38098621e-01 2.33600870e-01
1.86218083e-01 -1.60583362e-01 9.60806966e-01 -1.46659851e+00
-6.70471072e-01 -7.23045230e-01 -1.42714575e-01 5.23744822e-01
-6.87816918e-01 -3.86740267e-01 -8.12116981e-01 -5.96677542e-01
8.37914407e-01 -8.71139824e-01 -3.24266046e-01 -3.02427977e-01
-4.44818705e-01 -8.87830406e-02 1.05860496e+00 -2.71496236e-01
1.48992968e+00 -2.27952385e+00 7.91294515e-01 6.09930098e-01
5.61318457e-01 -3.16739045e-02 -5.35135530e-02 7.94257700e-01
-1.13833234e-01 -2.41487511e-02 -1.32107168e-01 1.53839797e-01
-3.95640373e-01 -2.27530673e-02 9.78991538e-02 7.38919795e-01
-2.26916373e-01 2.68831789e-01 -8.70963514e-01 -6.94937408e-01
1.95860267e-01 3.16933841e-01 -3.41731071e-01 1.17565535e-01
4.39012557e-01 5.42302251e-01 -1.07908428e+00 5.25630295e-01
8.15457463e-01 -4.59168822e-01 3.29284877e-01 -7.72248030e-01
-5.40930480e-02 -3.94853026e-01 -1.78257966e+00 1.69391155e+00
-1.51610509e-01 -2.58811802e-01 2.66634703e-01 -1.02657020e+00
9.28577900e-01 2.49182224e-01 8.47087562e-01 -2.79983163e-01
9.52822492e-02 1.90427184e-01 1.24491163e-01 -5.90805054e-01
1.56863227e-01 1.04840882e-02 9.70039889e-02 4.72499222e-01
2.38929596e-02 1.32911101e-01 1.38965532e-01 2.65966088e-01
6.51038587e-01 -1.72886059e-01 5.93553185e-01 -4.01926398e-01
1.23275208e+00 -2.87198454e-01 1.03427076e+00 1.75600022e-01
2.58042421e-02 5.83537281e-01 3.93827111e-01 -4.95199323e-01
-5.99205852e-01 -7.60239482e-01 9.38654470e-04 7.11306274e-01
6.88880801e-01 -5.84256828e-01 -4.17302459e-01 -1.01644218e+00
-1.04232587e-01 2.53358215e-01 -4.35042411e-01 -3.11750412e-01
-3.83229822e-01 -8.92392099e-01 -2.52957731e-01 2.40595043e-01
6.75772190e-01 -4.27487761e-01 -5.73130995e-02 1.97779939e-01
-3.01607758e-01 -7.20887721e-01 -7.91821778e-01 -2.99399793e-01
-1.00703537e+00 -1.39974058e+00 -5.82262337e-01 -7.69780099e-01
1.02913535e+00 9.61558998e-01 6.39197886e-01 2.92983115e-01
5.07810563e-02 5.35541773e-01 -5.37177801e-01 1.38319954e-02
4.31263410e-02 8.62523168e-02 2.44199157e-01 7.21440911e-01
3.24229211e-01 -7.87593544e-01 -6.59707546e-01 6.34785831e-01
-8.25373113e-01 5.44522814e-02 6.80169284e-01 1.21595144e+00
9.08209085e-01 3.04210126e-01 6.82359934e-01 -1.02653873e+00
6.10248983e-01 -3.85321587e-01 -5.57340860e-01 5.12329400e-01
-1.03707182e+00 4.78086919e-02 7.96701789e-01 -1.53522894e-01
-9.59652424e-01 2.08971664e-01 2.61922687e-01 -5.45221269e-01
1.84747353e-01 7.08680034e-01 -6.85094416e-01 -3.63102078e-01
7.82978982e-02 4.69572634e-01 1.15182556e-01 -4.20093149e-01
3.77260894e-01 4.30263877e-01 -1.01297647e-01 -2.21169338e-01
9.33471978e-01 6.28494799e-01 2.69313276e-01 -6.09933376e-01
-8.01493168e-01 -5.74176252e-01 -9.95071948e-01 -2.36902490e-01
7.20331550e-01 -8.71904433e-01 -6.37162685e-01 3.43648702e-01
-6.70817435e-01 7.00138450e-01 2.30649740e-01 7.80255735e-01
-3.91652703e-01 8.88616145e-01 -2.70798862e-01 -5.50427556e-01
-2.77706593e-01 -1.16230476e+00 6.62598312e-01 2.89548367e-01
3.21773380e-01 -9.78290856e-01 -1.09550580e-01 3.09779495e-01
8.32481757e-02 1.66426182e-01 1.05096805e+00 -6.79907739e-01
-4.17412728e-01 -6.40552640e-02 -2.07470357e-01 7.32010901e-02
6.94328666e-01 6.49444535e-02 -4.88717109e-01 -5.82470000e-01
2.72393972e-01 -2.89404429e-02 7.61737525e-01 1.49496749e-01
1.09134459e+00 -7.70321712e-02 -4.82589930e-01 6.00406110e-01
1.47643566e+00 1.36146814e-01 1.60272136e-01 1.97759330e-01
1.10442483e+00 5.95735967e-01 8.07073355e-01 7.08607972e-01
4.56300616e-01 5.11947155e-01 5.30965507e-01 -9.47492197e-02
3.45445067e-01 -2.66060919e-01 4.46482413e-02 1.57164252e+00
-1.25622898e-01 -4.07410450e-02 -4.94502842e-01 3.77280772e-01
-2.16379929e+00 -9.68648791e-01 -3.50963026e-01 2.31284690e+00
1.42743200e-01 9.01504606e-02 1.47453249e-01 6.79927692e-02
9.77277696e-01 6.06783926e-01 -6.67487085e-01 2.93696463e-01
-1.70378298e-01 -6.17463291e-01 2.00581521e-01 2.41895780e-01
-1.21479511e+00 5.21416783e-01 5.20371723e+00 7.72340298e-01
-9.52614844e-01 -4.04362343e-02 2.10989028e-01 1.40973672e-01
-5.35239637e-01 1.34933844e-01 -5.34471393e-01 3.02267224e-01
-5.79891540e-02 -4.65973347e-01 4.36512619e-01 7.15245903e-01
2.11191922e-01 1.91129938e-01 -7.20258951e-01 1.24429715e+00
5.20827532e-01 -7.91257024e-01 2.84975171e-01 1.43854409e-01
7.04475403e-01 -2.60399133e-01 -9.04551055e-03 -3.60110961e-02
-8.37718621e-02 -4.40434128e-01 2.65235156e-01 7.17592776e-01
5.06898463e-01 -1.16545796e+00 6.84468210e-01 4.11148965e-01
-1.68247700e+00 -1.56292379e-01 -6.69486582e-01 2.56030917e-01
1.66622177e-01 7.70323575e-01 -2.19010815e-01 1.36881053e+00
4.30825919e-01 1.26629972e+00 -8.68524492e-01 9.44447100e-01
-8.11472833e-02 1.84492990e-01 3.51228230e-02 8.26874748e-02
1.80482313e-01 -8.12570035e-01 7.30079293e-01 5.25942564e-01
3.75252128e-01 2.38686502e-01 7.25827277e-01 4.48798001e-01
2.42242321e-01 6.23335183e-01 -6.12200260e-01 1.79825053e-01
3.11739951e-01 1.58299780e+00 -7.14819670e-01 -1.09513819e-01
-9.72620368e-01 1.05235434e+00 3.79113793e-01 3.30042809e-01
-4.41023380e-01 -4.83515382e-01 3.20678204e-01 -1.20213859e-01
4.16645020e-01 -1.05298460e-01 -6.90095942e-04 -1.60507858e+00
5.26096076e-02 -9.53444719e-01 7.65505731e-01 -3.74970943e-01
-1.76744843e+00 4.91274565e-01 -1.53634548e-01 -2.00110197e+00
4.94065695e-02 -2.00820848e-01 -5.76550722e-01 7.60394931e-01
-1.33320534e+00 -1.10228586e+00 -3.33147466e-01 8.77160668e-01
4.11922455e-01 -4.00148451e-01 6.84145868e-01 3.56120497e-01
-5.80962658e-01 3.47928286e-01 3.05839777e-01 -4.02719937e-02
8.25488925e-01 -1.03674650e+00 -1.93766311e-01 7.73287296e-01
1.94579065e-01 6.40452325e-01 3.11643600e-01 -6.95901513e-01
-1.96130538e+00 -9.38116312e-01 3.65777820e-01 4.32345010e-02
6.83369219e-01 -4.52333651e-02 -9.76894379e-01 4.33658361e-01
1.69796407e-01 2.55979121e-01 9.80472028e-01 2.11225793e-01
-2.34659269e-01 -3.45470309e-01 -8.56770277e-01 5.41948736e-01
1.04428363e+00 -3.12235206e-01 -4.53382492e-01 2.61372745e-01
3.58457983e-01 -2.15810463e-01 -1.10125685e+00 5.05207539e-01
5.23585677e-01 -1.18473804e+00 7.87725329e-01 -4.92327720e-01
7.95133039e-02 -7.00729251e-01 -1.28722444e-01 -1.66727757e+00
-8.70902181e-01 -3.33681613e-01 6.11838438e-02 1.48127604e+00
1.99559301e-01 -8.79170299e-01 5.27601480e-01 1.00996792e-01
1.10629983e-01 -8.58557940e-01 -8.68128121e-01 -5.52387297e-01
-5.84772229e-01 1.74746722e-01 5.22416174e-01 1.16605794e+00
-5.04832007e-02 7.40694940e-01 -7.08395422e-01 4.23278660e-01
8.46067786e-01 7.87173510e-01 7.02730179e-01 -1.46915483e+00
-2.28571579e-01 -3.17467839e-01 -5.86010098e-01 -7.15268850e-01
1.98921282e-02 -7.98050165e-01 -3.37592870e-01 -1.63612592e+00
5.13659894e-01 -2.12280527e-01 -6.50887251e-01 2.22834386e-03
-6.00651264e-01 -2.77445257e-01 2.07556024e-01 5.15447259e-01
-7.73575842e-01 8.39733124e-01 1.47959185e+00 -6.78779185e-02
-1.31326497e-01 2.43668810e-01 -9.03847575e-01 8.40537190e-01
4.43180323e-01 -3.55642676e-01 -6.48645163e-01 -2.70474046e-01
2.38182116e-02 3.68836373e-01 -5.88690713e-02 -8.03372979e-01
3.22216809e-01 -3.41952503e-01 3.78963262e-01 -6.96787775e-01
1.99110061e-01 -1.28847671e+00 9.91062820e-02 3.71072412e-01
5.63823991e-02 2.68879384e-01 -5.40380418e-01 1.06779277e+00
-4.42902386e-01 -8.56809691e-02 6.85535312e-01 -2.57622302e-01
-7.70707250e-01 6.72404230e-01 4.25145440e-02 1.66886616e-02
9.76025999e-01 -1.07309252e-01 -9.00995210e-02 -4.58075076e-01
-6.57932281e-01 6.01674199e-01 6.03497565e-01 5.03790617e-01
8.07558119e-01 -1.78014529e+00 -7.06285000e-01 2.72537768e-01
5.46245873e-01 -3.03506881e-01 4.78992194e-01 1.03612006e+00
2.93344036e-02 6.55752746e-03 -1.21456131e-01 -5.78756094e-01
-1.36921322e+00 8.69989634e-01 4.45942730e-02 -1.83430538e-01
-6.57673180e-01 2.78199404e-01 3.51511151e-01 -3.59500349e-01
-1.10883825e-01 4.15741384e-01 -7.22607195e-01 4.27827209e-01
4.63004351e-01 4.88067210e-01 -1.64566323e-01 -9.29370105e-01
-4.20436054e-01 1.15870106e+00 -1.62420347e-01 1.30012542e-01
1.48961484e+00 -5.42131007e-01 -3.74879479e-01 6.59921944e-01
1.19240737e+00 2.95663983e-01 -8.80371392e-01 -4.39545482e-01
-1.97283357e-01 -7.68392980e-01 -9.19042388e-04 -1.44049376e-01
-1.38306427e+00 7.91661024e-01 5.85378826e-01 3.03133070e-01
1.25988400e+00 -3.04340929e-01 5.83238363e-01 3.61407727e-01
5.54783285e-01 -1.01396668e+00 5.43958955e-02 3.35850924e-01
7.54874170e-01 -1.23959291e+00 5.08698225e-01 -8.23795795e-01
-9.17348564e-01 1.24463785e+00 8.21974874e-01 -1.67531580e-01
7.14476824e-01 -4.00743812e-01 -7.42475986e-02 -5.34781575e-01
-4.75449800e-01 -2.23788261e-01 4.41333026e-01 3.05426568e-01
2.18842641e-01 -2.21591651e-01 -6.57179773e-01 5.82425475e-01
3.64930898e-01 -3.76469940e-01 3.73073310e-01 7.65466988e-01
-3.59221131e-01 -1.12059450e+00 -3.18951994e-01 5.58512449e-01
-3.32561731e-01 1.98132262e-01 -4.04594213e-01 6.16466761e-01
-1.11471683e-01 1.05986333e+00 -5.74941337e-01 -6.69361413e-01
6.23408258e-01 -1.30539253e-01 1.32435933e-01 -6.48581982e-01
-3.17047894e-01 5.26741147e-01 -2.66867340e-01 -5.05394638e-01
-7.62199700e-01 -6.86943591e-01 -9.79082346e-01 1.39599815e-01
-6.72649920e-01 3.88595581e-01 9.61212739e-02 8.00208032e-01
5.26650727e-01 3.77730489e-01 1.30195403e+00 -6.71788931e-01
-5.97786307e-01 -6.01541579e-01 -1.03452575e+00 5.64724088e-01
5.13846688e-02 -9.13994968e-01 -5.08670926e-01 -2.56893724e-01] | [8.206299781799316, 4.628712177276611] |
72794340-b69b-49f2-9d8b-dc0c660b7df3 | generalizing-hand-segmentation-in-egocentric | null | null | http://openaccess.thecvf.com/content_CVPR_2020/html/Cai_Generalizing_Hand_Segmentation_in_Egocentric_Videos_With_Uncertainty-Guided_Model_Adaptation_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Cai_Generalizing_Hand_Segmentation_in_Egocentric_Videos_With_Uncertainty-Guided_Model_Adaptation_CVPR_2020_paper.pdf | Generalizing Hand Segmentation in Egocentric Videos With Uncertainty-Guided Model Adaptation | Although the performance of hand segmentation in egocentric videos has been significantly improved by using CNNs, it still remains a challenging issue to generalize the trained models to new domains, e.g., unseen environments. In this work, we solve the hand segmentation generalization problem without requiring segmentation labels in the target domain. To this end, we propose a Bayesian CNN-based model adaptation framework for hand segmentation, which introduces and considers two key factors: 1) prediction uncertainty when the model is applied in a new domain and 2) common information about hand shapes shared across domains. Consequently, we propose an iterative self-training method for hand segmentation in the new domain, which is guided by the model uncertainty estimated by a Bayesian CNN. We further use an adversarial component in our framework to utilize shared information about hand shapes to constrain the model adaptation process. Experiments on multiple egocentric datasets show that the proposed method significantly improves the generalization performance of hand segmentation.
| [' Yoichi Sato', ' Feng Lu', 'Minjie Cai'] | 2020-06-01 | null | null | null | cvpr-2020-6 | ['hand-segmentation'] | ['computer-vision'] | [ 1.34415075e-01 -5.28959557e-03 -1.66973129e-01 -3.85632128e-01
-3.20170671e-01 -5.95590234e-01 1.14698648e-01 -6.94355011e-01
-2.32950285e-01 7.53788352e-01 1.31088406e-01 2.79550433e-01
-1.44968191e-02 -7.77430356e-01 -7.87686169e-01 -7.61524439e-01
5.79540491e-01 5.98705053e-01 4.09758359e-01 1.93825662e-01
-6.59235613e-03 4.14675981e-01 -1.10135138e+00 -1.74819663e-01
1.04966211e+00 8.10316503e-01 5.64972341e-01 3.90401065e-01
1.18930057e-01 3.03234428e-01 -3.49964947e-01 -2.07443595e-01
3.24059695e-01 -3.11277866e-01 -1.09070396e+00 5.06427288e-01
2.58960038e-01 -7.32638240e-01 -5.66276908e-01 1.22059202e+00
4.81298476e-01 5.20128012e-01 8.23248446e-01 -1.00332987e+00
-6.67480052e-01 5.73662579e-01 -5.79552352e-01 -8.43131635e-03
8.12876374e-02 2.58551449e-01 5.98929703e-01 -6.21309996e-01
7.94648349e-01 1.25022125e+00 4.37660664e-01 8.44142318e-01
-1.13258028e+00 -7.49229491e-01 7.08618402e-01 1.88129589e-01
-1.50291276e+00 -6.48878422e-03 1.12604952e+00 -6.32036805e-01
3.30772996e-01 -3.08583319e-01 6.93095446e-01 1.43473423e+00
-2.47532636e-01 1.28419530e+00 9.41448033e-01 -1.26713589e-01
4.51831520e-01 7.78335631e-02 -1.63111258e-02 4.06429440e-01
1.59692056e-02 -1.01582564e-01 -7.46611878e-02 1.47437155e-01
1.14123249e+00 -4.71262299e-02 -2.38598272e-01 -5.38693249e-01
-9.61718023e-01 4.40330386e-01 6.26411855e-01 2.44879559e-01
-5.02577603e-01 2.03656062e-01 3.33337933e-01 -3.97281915e-01
4.01462585e-01 5.74429892e-02 -6.29144371e-01 -1.77113917e-02
-9.36270654e-01 4.43951666e-01 5.15772104e-01 1.53149474e+00
6.81456566e-01 3.22447531e-03 -2.36703128e-01 8.62520576e-01
3.13155293e-01 5.80489814e-01 2.83424467e-01 -8.11825097e-01
4.91174906e-01 3.92871916e-01 9.18334052e-02 -9.11665678e-01
-3.92157257e-01 -6.48175716e-01 -6.58535480e-01 -1.68908402e-01
6.18358135e-01 -3.42271417e-01 -1.19910526e+00 2.11524129e+00
4.99169141e-01 4.95655835e-01 -1.30112857e-01 1.12014222e+00
5.49756706e-01 3.11166853e-01 3.16584945e-01 8.51814821e-02
9.66207206e-01 -1.00039065e+00 -7.27832913e-01 -2.83000678e-01
2.69349009e-01 -5.43087959e-01 9.87467170e-01 2.96354622e-01
-8.49300861e-01 -7.46335387e-01 -7.41530240e-01 1.08795069e-01
-8.71576443e-02 1.23702280e-01 2.99258739e-01 6.27585769e-01
-4.80878651e-01 5.69388688e-01 -9.24063504e-01 -4.97664392e-01
7.56299853e-01 3.60375404e-01 -7.68259764e-02 -1.78234518e-01
-9.89223361e-01 3.76242578e-01 8.31616640e-01 2.18895391e-01
-1.02971363e+00 -3.90395850e-01 -7.80199766e-01 2.73100892e-03
7.18960762e-01 -8.35398793e-01 1.24920952e+00 -9.72762525e-01
-1.62157309e+00 4.05294061e-01 -2.40829587e-01 -3.81436124e-02
7.45396614e-01 -4.44444507e-01 5.60891032e-02 3.25283371e-02
1.30163446e-01 6.97516382e-01 1.12699437e+00 -1.57259667e+00
-5.71263313e-01 -6.56584740e-01 2.40620404e-01 2.44187325e-01
-2.71201223e-01 -4.54887122e-01 -8.27617884e-01 -9.49804366e-01
2.90823817e-01 -1.22729468e+00 -1.17096908e-01 -3.17186594e-01
-6.44256949e-01 -3.80858094e-01 9.25599813e-01 -9.15982366e-01
9.67382133e-01 -1.83840990e+00 6.52968347e-01 3.79687518e-01
-1.52216768e-02 1.96268409e-01 -6.52135760e-02 -8.51057768e-02
2.42511034e-01 1.00785717e-01 -3.31371754e-01 -4.51439172e-01
-1.70063041e-02 4.13580209e-01 -3.62940609e-01 6.67653456e-02
1.83260933e-01 8.45028877e-01 -1.00107265e+00 -6.42931700e-01
1.91863015e-01 4.69773740e-01 -6.87412024e-01 3.94080848e-01
-5.61018050e-01 1.08607554e+00 -8.66053224e-01 5.97808897e-01
9.99563515e-01 -1.72859594e-01 9.68490317e-02 -3.76146078e-01
3.61091018e-01 -2.84234226e-01 -1.20780396e+00 2.19177771e+00
-4.60572779e-01 -4.06969413e-02 -2.69641131e-01 -7.96025872e-01
7.47179449e-01 2.85645813e-01 2.26024330e-01 -8.41994360e-02
4.41029340e-01 -5.07126600e-02 -1.38395205e-01 -6.54952407e-01
2.81598389e-01 -1.05914697e-01 2.16535717e-01 5.13443470e-01
2.49482885e-01 -2.89053589e-01 -1.41914666e-01 -1.10722370e-01
4.94412690e-01 7.22352803e-01 -1.02847330e-01 -4.43363003e-02
6.22294545e-01 -1.08089924e-01 9.72332954e-01 6.66482627e-01
-4.87132281e-01 8.29223454e-01 2.29058430e-01 -2.77328342e-01
-9.89924550e-01 -1.08960092e+00 -1.59055755e-01 8.87984335e-01
4.04775620e-01 7.83194080e-02 -1.13369000e+00 -1.09476733e+00
-8.15708488e-02 6.25405967e-01 -6.86186969e-01 -1.79510430e-01
-7.18447864e-01 -4.27457452e-01 2.41778821e-01 1.23245716e+00
1.01223838e+00 -1.05425870e+00 -1.95578650e-01 2.46562257e-01
-4.44899648e-01 -1.42093468e+00 -6.69992566e-01 -3.45709562e-01
-9.86199677e-01 -1.01828671e+00 -1.32611585e+00 -8.77758265e-01
7.39880443e-01 4.14484739e-02 5.17435431e-01 -3.59957397e-01
-6.72181621e-02 7.46490300e-01 -3.26610595e-01 -1.89189449e-01
3.68900001e-02 4.65332001e-01 1.21224605e-01 1.96075186e-01
4.74567205e-01 -5.74100018e-01 -7.21924782e-01 5.22143900e-01
-6.82154179e-01 -1.30176023e-02 4.93727505e-01 9.55932438e-01
7.22128451e-01 7.96549860e-03 7.50830829e-01 -7.53824115e-01
4.24602240e-01 -3.41362357e-01 -3.99250537e-01 3.36560905e-01
-2.98380464e-01 -1.33415442e-02 5.45544982e-01 -7.24037051e-01
-1.60877740e+00 4.66358334e-01 -3.50273736e-02 -7.07237184e-01
-3.85832131e-01 3.99560422e-01 -6.05045021e-01 7.44327158e-02
4.08544004e-01 3.40321928e-01 -1.84055641e-01 -7.00607240e-01
4.41694021e-01 7.09679425e-01 6.08715415e-01 -1.01508355e+00
8.29773724e-01 6.23360038e-01 -1.61848694e-01 -6.14360929e-01
-8.18920374e-01 -2.82060742e-01 -1.11206186e+00 -2.09822312e-01
1.24115336e+00 -8.88503075e-01 -5.33408523e-01 8.77406478e-01
-1.32063639e+00 -5.17710507e-01 -9.45070088e-02 5.43971062e-01
-8.20115864e-01 6.70796514e-01 -2.08081454e-01 -7.49864519e-01
-5.26203029e-02 -1.39196932e+00 9.54772055e-01 5.10833025e-01
-1.09422974e-01 -1.01450980e+00 -3.43950205e-02 5.36488771e-01
6.34869263e-02 1.46094203e-01 7.53664970e-01 -6.34300292e-01
-8.35595191e-01 -1.52381733e-01 -3.48452747e-01 6.83563232e-01
2.72958636e-01 -3.78795207e-01 -1.09854198e+00 -2.79142916e-01
-3.56561653e-02 -2.86040694e-01 6.95953429e-01 4.65990722e-01
1.60501206e+00 -5.24841882e-02 -5.13774812e-01 6.43810570e-01
8.44403863e-01 2.06952065e-01 5.60902297e-01 1.24506809e-01
9.90536034e-01 6.24445319e-01 7.27384627e-01 4.83659506e-01
4.35681820e-01 5.67818046e-01 3.00865889e-01 3.29360723e-01
-7.83397555e-02 -4.61494058e-01 -2.29111224e-01 5.08482277e-01
-4.09887195e-01 -1.63042396e-01 -8.77659976e-01 7.39647806e-01
-1.97702396e+00 -5.92445731e-01 4.97834921e-01 1.89544904e+00
6.25768840e-01 7.13846553e-03 8.71946290e-02 -2.13522851e-01
1.01392388e+00 2.10116729e-02 -1.11744523e+00 2.00779796e-01
2.17893347e-01 1.48602709e-01 1.84470132e-01 2.50821710e-01
-1.33968413e+00 1.34232032e+00 5.61848736e+00 7.54391253e-01
-9.15046155e-01 9.55991633e-03 3.16921622e-01 2.53968596e-01
7.68974200e-02 7.53015000e-03 -7.10618913e-01 4.43486333e-01
-1.65577620e-01 -1.14535159e-02 5.50157130e-01 1.24162352e+00
1.24771066e-01 1.60014823e-01 -1.06163275e+00 1.00049055e+00
2.06130624e-01 -6.72361672e-01 1.89490691e-01 -1.02334358e-01
9.29427564e-01 -3.91807914e-01 2.08765522e-01 2.46678814e-01
7.01196790e-02 -7.96231091e-01 5.95864654e-01 6.16039455e-01
5.50459146e-01 -7.72525668e-01 6.13667488e-01 4.23469514e-01
-1.13526106e+00 -7.91278109e-02 -3.21267903e-01 2.43458897e-01
1.67074293e-01 1.10311151e-01 -8.54653895e-01 4.59444284e-01
7.80441940e-01 8.48506033e-01 -3.71355385e-01 9.50717628e-01
-5.54503977e-01 4.76054847e-01 -2.13163823e-01 2.70137101e-01
6.76202029e-02 -1.74015969e-01 7.01082051e-01 7.66884208e-01
1.31216377e-01 1.72323063e-01 2.35910952e-01 1.30909932e+00
3.59746218e-02 -6.46753833e-02 -3.91051173e-01 1.22254156e-01
4.37699676e-01 9.64856684e-01 -6.32205367e-01 -3.40964675e-01
-2.26579592e-01 1.24852467e+00 2.18123585e-01 7.76429951e-01
-7.82467365e-01 -3.55951875e-01 4.56643522e-01 -2.08870307e-01
5.45327067e-01 -2.55458564e-01 -3.20126384e-01 -1.39068604e+00
1.68471605e-01 -5.66716850e-01 2.15312436e-01 -8.20976317e-01
-1.44388855e+00 5.10265589e-01 1.28643572e-01 -1.06346059e+00
-2.42560998e-01 -5.41630208e-01 -4.67293054e-01 1.01963639e+00
-1.27631652e+00 -1.49298716e+00 -6.50924742e-01 7.44023263e-01
6.25770509e-01 -1.63196385e-01 4.88351345e-01 6.67384118e-02
-6.72188640e-01 6.46376967e-01 1.98266003e-02 4.56811845e-01
7.53571570e-01 -1.15228283e+00 3.50462437e-01 7.13890612e-01
-2.60029763e-01 7.14529932e-01 5.37993789e-01 -1.02314115e+00
-8.79778028e-01 -1.28278267e+00 1.74146041e-01 -3.41869026e-01
3.14947724e-01 -2.69143879e-01 -1.00051510e+00 1.07769954e+00
-1.57989323e-01 -1.08036660e-01 5.34028292e-01 1.68175548e-01
-3.94352734e-01 8.94838721e-02 -1.31069994e+00 6.97287619e-01
1.41865325e+00 -5.56717157e-01 -7.52803683e-01 3.15221697e-01
5.73017538e-01 -5.81638694e-01 -8.13118339e-01 6.05082452e-01
8.01407754e-01 -5.35598457e-01 8.72622669e-01 -8.84464264e-01
3.21363628e-01 -2.28955328e-01 -2.12076038e-01 -1.38464034e+00
-2.74014831e-01 -3.07307929e-01 -9.24971029e-02 1.27619505e+00
-8.81316513e-02 -5.55610716e-01 1.15936530e+00 8.90605509e-01
-8.15315396e-02 -5.30734301e-01 -9.04429436e-01 -1.07417810e+00
4.27016497e-01 -4.31175023e-01 8.09179485e-01 7.38000035e-01
-1.82026446e-01 4.62662019e-02 -4.14479733e-01 3.18808943e-01
8.92762899e-01 1.36423796e-01 1.02382100e+00 -1.14895761e+00
-2.43766978e-01 -9.39931720e-02 -4.43523616e-01 -1.66637933e+00
5.85934043e-01 -7.08511114e-01 2.48154968e-01 -1.42362571e+00
4.39256132e-01 -4.59314913e-01 -2.12033570e-01 3.09287757e-01
-2.70282894e-01 -8.35514888e-02 2.24790558e-01 3.09513479e-01
-3.90890062e-01 7.99917877e-01 1.72735643e+00 -2.13707730e-01
-4.30457264e-01 1.92540959e-01 -3.77940446e-01 9.81493413e-01
7.79237092e-01 -3.90066504e-01 -6.10526025e-01 -5.71545839e-01
-2.43821993e-01 -1.27000496e-01 5.27542830e-01 -1.14145553e+00
2.39091620e-01 -2.05872327e-01 4.59283769e-01 -8.19756925e-01
3.68700475e-01 -9.57752943e-01 -2.43620694e-01 1.56249240e-01
-1.30124971e-01 -6.12026215e-01 2.61113137e-01 9.54202890e-01
5.12578860e-02 -3.11171114e-01 7.01484621e-01 -7.86873177e-02
-8.25713634e-01 6.15843773e-01 -1.13308690e-01 1.67724594e-01
1.04875505e+00 -2.93298781e-01 6.07906394e-02 -2.85316765e-01
-1.27299058e+00 3.85867387e-01 4.32705790e-01 5.81880271e-01
4.82346952e-01 -1.28066027e+00 -3.51725072e-01 1.45111859e-01
1.83834448e-01 5.22330165e-01 6.89342737e-01 6.12685204e-01
-2.21093357e-01 2.77691454e-01 -2.95097560e-01 -9.51751232e-01
-9.91207719e-01 7.16024041e-01 2.92849004e-01 1.10255018e-01
-5.08126557e-01 8.76042724e-01 6.16548896e-01 -5.60231626e-01
3.76305372e-01 -3.49937648e-01 -1.87729195e-01 -1.91733032e-01
2.01369345e-01 3.97171050e-01 -4.34990853e-01 -7.13158786e-01
-2.55206555e-01 1.00624812e+00 -3.87923390e-01 -2.70097312e-02
1.03209257e+00 -2.78658390e-01 1.79868534e-01 3.79888207e-01
9.81143177e-01 -4.29159224e-01 -1.76604998e+00 -4.74851608e-01
-2.78190792e-01 -7.17060447e-01 -9.22831073e-02 -8.64320576e-01
-1.18428659e+00 1.02101409e+00 7.29484439e-01 -4.81742531e-01
9.93991613e-01 1.32068276e-01 1.11273706e+00 4.74644810e-01
4.67081308e-01 -1.35374248e+00 2.76654065e-01 5.39819419e-01
9.13858056e-01 -1.19719064e+00 -1.83237404e-01 -6.29967630e-01
-7.12489367e-01 1.04186964e+00 9.07144904e-01 -1.74172446e-01
8.14690173e-01 -4.05178845e-01 -5.12295626e-02 1.27422631e-01
1.77559093e-01 -2.37185195e-01 3.24507475e-01 8.26548934e-01
-4.82663997e-02 5.81259504e-02 -9.26496014e-02 1.06387269e+00
6.56784400e-02 4.97420400e-01 2.02547759e-01 8.46760035e-01
-5.19053116e-02 -1.05668771e+00 -3.62988830e-01 1.93407789e-01
-1.69336185e-01 2.68988699e-01 -1.99353725e-01 8.18860888e-01
3.28827918e-01 6.21051192e-01 -1.29345134e-01 -6.01696074e-01
3.29511017e-01 2.02511355e-01 7.17857063e-01 -6.29506886e-01
-7.17420354e-02 1.08603314e-01 -4.31061655e-01 -3.09133589e-01
-4.86367434e-01 -7.27193296e-01 -1.21196580e+00 8.16468745e-02
-5.45355439e-01 -2.15441659e-01 4.70865339e-01 1.23327088e+00
2.62931436e-01 5.84926963e-01 3.38020593e-01 -1.00645304e+00
-7.53317297e-01 -1.04917371e+00 -8.24963808e-01 3.97903055e-01
2.69703642e-02 -1.16323936e+00 -2.59871274e-01 2.09103331e-01] | [9.538701057434082, 1.2870944738388062] |
efdeb7dd-681c-4b04-b9ed-1a5ccf9e961c | opensr-open-modality-speech-recognition-via | 2306.06410 | null | https://arxiv.org/abs/2306.06410v1 | https://arxiv.org/pdf/2306.06410v1.pdf | OpenSR: Open-Modality Speech Recognition via Maintaining Multi-Modality Alignment | Speech Recognition builds a bridge between the multimedia streaming (audio-only, visual-only or audio-visual) and the corresponding text transcription. However, when training the specific model of new domain, it often gets stuck in the lack of new-domain utterances, especially the labeled visual utterances. To break through this restriction, we attempt to achieve zero-shot modality transfer by maintaining the multi-modality alignment in phoneme space learned with unlabeled multimedia utterances in the high resource domain during the pre-training \cite{shi2022learning}, and propose a training system Open-modality Speech Recognition (\textbf{OpenSR}) that enables the models trained on a single modality (e.g., audio-only) applicable to more modalities (e.g., visual-only and audio-visual). Furthermore, we employ a cluster-based prompt tuning strategy to handle the domain shift for the scenarios with only common words in the new domain utterances. We demonstrate that OpenSR enables modality transfer from one to any in three different settings (zero-, few- and full-shot), and achieves highly competitive zero-shot performance compared to the existing few-shot and full-shot lip-reading methods. To the best of our knowledge, OpenSR achieves the state-of-the-art performance of word error rate in LRS2 on audio-visual speech recognition and lip-reading with 2.7\% and 25.0\%, respectively. The code and demo are available at https://github.com/Exgc/OpenSR. | ['Zhou Zhao', 'Xinyu Duan', 'Wang Lin', 'Linjun Li', 'Tao Jin', 'Xize Cheng'] | 2023-06-10 | null | null | null | null | ['visual-speech-recognition', 'audio-visual-speech-recognition'] | ['speech', 'speech'] | [ 3.08231205e-01 -6.96701035e-02 -3.49190801e-01 -2.38314003e-01
-1.64587200e+00 -4.86418158e-01 3.03789109e-01 -3.31901729e-01
-3.04453999e-01 5.34875631e-01 2.51970828e-01 -3.21485341e-01
3.27527732e-01 -3.33822191e-01 -8.56263459e-01 -7.31705725e-01
4.99208421e-01 2.62865782e-01 3.34258348e-01 -1.58548951e-01
-1.01779252e-01 -6.28209189e-02 -1.97141182e+00 4.51455474e-01
5.81408739e-01 9.78544474e-01 5.32868028e-01 8.19405615e-01
-3.85104626e-01 2.32354745e-01 -2.88666993e-01 -3.02118003e-01
-3.88176776e-02 -2.94590861e-01 -6.74591720e-01 6.18992448e-02
2.48370200e-01 -4.48206574e-01 -6.31750405e-01 8.78218591e-01
1.08749747e+00 3.57171059e-01 5.89172184e-01 -1.51006114e+00
-8.41796577e-01 3.96604657e-01 -6.60263777e-01 -1.43971266e-02
6.29492283e-01 3.43405545e-01 5.75056851e-01 -1.29061019e+00
5.44889867e-01 1.18946052e+00 3.31250072e-01 1.11833799e+00
-8.12211037e-01 -9.21918809e-01 3.79151739e-02 3.67563099e-01
-1.59131467e+00 -1.33583236e+00 6.19289935e-01 -2.65074849e-01
8.37768912e-01 2.54406214e-01 1.26425833e-01 1.61859882e+00
-5.88582814e-01 8.13812256e-01 9.03345764e-01 -4.68102217e-01
1.54032484e-01 3.40260744e-01 -1.08137801e-02 1.38888255e-01
-4.92237896e-01 3.48133966e-02 -1.03465796e+00 9.55124870e-02
2.77064919e-01 -6.91926405e-02 -4.66243804e-01 3.57545353e-02
-1.14991355e+00 4.71034020e-01 -3.56486857e-01 2.24881157e-01
-5.44577651e-02 -8.88510793e-02 6.92661762e-01 4.59267944e-01
4.18704808e-01 -4.07043695e-01 -6.35780454e-01 -6.19237006e-01
-1.02511239e+00 -4.26632404e-01 4.54866141e-01 1.32414806e+00
7.71863699e-01 1.96005836e-01 -2.94124991e-01 1.40249658e+00
4.07958567e-01 9.29333746e-01 7.33372629e-01 -7.55895197e-01
7.40389645e-01 -3.38188000e-02 -1.57604858e-01 -1.50796115e-01
2.39032842e-02 2.86439080e-02 -7.58768260e-01 -2.99606323e-01
1.44591972e-01 -2.41186738e-01 -1.13959086e+00 1.93801546e+00
5.60405314e-01 5.33575058e-01 6.11392200e-01 7.00978756e-01
1.55521667e+00 1.06349730e+00 -8.62595066e-03 -5.91900766e-01
1.38728893e+00 -9.89405334e-01 -8.48513007e-01 2.14262120e-02
4.69417810e-01 -1.00021791e+00 1.51716590e+00 1.39274433e-01
-1.06889904e+00 -7.45750785e-01 -7.01894820e-01 -1.06442750e-01
-3.67465973e-01 3.38247158e-02 -2.57165641e-01 7.44831920e-01
-1.14037216e+00 -1.12755552e-01 -3.84723067e-01 -6.01232469e-01
1.77384362e-01 3.25393528e-02 -6.30436242e-01 -3.78176361e-01
-1.49144864e+00 3.82280618e-01 1.90747038e-01 -3.45098674e-01
-1.15233397e+00 -7.16077983e-01 -8.87476802e-01 5.28984554e-02
7.00961709e-01 -3.72394383e-01 1.45429158e+00 -1.04243171e+00
-1.87735605e+00 9.25746381e-01 -6.06801450e-01 6.60301745e-02
2.65242368e-01 1.51670873e-01 -7.51424134e-01 3.32015723e-01
-1.10074773e-01 8.66099775e-01 1.18332267e+00 -1.24885535e+00
-6.20859802e-01 -2.00882897e-01 -1.16685115e-01 3.99575233e-01
-6.40537322e-01 3.13605309e-01 -7.49810517e-01 -5.75966239e-01
-1.78242356e-01 -7.08072543e-01 6.61628187e-01 -5.92862293e-02
-2.84409195e-01 -3.00272048e-01 9.03005719e-01 -8.15378428e-01
1.16184211e+00 -2.69418001e+00 1.69447321e-03 -2.49852076e-01
-2.14029759e-01 4.66359586e-01 -4.58856136e-01 5.77543199e-01
-1.19087100e-01 1.56546399e-01 -1.60342187e-01 -6.50262237e-01
6.86867535e-02 1.79987445e-01 -2.69896060e-01 2.03067079e-01
-1.08142503e-01 6.69786632e-01 -7.02017963e-01 -7.08985925e-01
2.47223958e-01 6.28640473e-01 -3.32490325e-01 3.87352973e-01
3.79213393e-02 6.08645082e-01 -1.36265196e-02 8.83716166e-01
8.58837187e-01 -6.44062683e-02 -9.59468260e-02 -1.34982035e-01
-9.21822060e-03 9.85698551e-02 -1.22216380e+00 2.08065128e+00
-7.16942430e-01 4.68799084e-01 2.44422063e-01 -7.82716334e-01
7.76904404e-01 9.65347409e-01 4.18646067e-01 -7.37953901e-01
-6.56240135e-02 2.98407227e-01 -3.75544280e-01 -8.42666149e-01
4.14657027e-01 -2.81846464e-01 -6.31803796e-02 2.59238422e-01
6.03246510e-01 1.22378223e-01 -1.81586608e-01 1.48078308e-01
7.22316027e-01 -1.11093700e-01 4.98069711e-02 2.90350080e-01
6.04336202e-01 -5.44953883e-01 4.49644685e-01 4.87364978e-01
-4.16882455e-01 8.41822684e-01 9.84478444e-02 6.00611806e-01
-9.24563944e-01 -1.26085901e+00 -1.27688095e-01 1.64978409e+00
1.33483127e-01 -1.98323235e-01 -7.44425833e-01 -4.51271057e-01
-1.88424677e-01 7.23485351e-01 -2.71429777e-01 -3.40518773e-01
-1.14951007e-01 -2.05534287e-02 9.15768862e-01 3.57189268e-01
4.64505970e-01 -9.95246589e-01 -7.74713308e-02 -6.30689934e-02
-6.13673985e-01 -1.18237495e+00 -8.47808063e-01 -3.72890458e-02
-3.03292811e-01 -6.70813739e-01 -1.31699526e+00 -1.01433682e+00
2.18644857e-01 5.15792489e-01 5.45207798e-01 -3.22276324e-01
7.14355782e-02 8.51809025e-01 -7.83982933e-01 -7.15532899e-02
-5.86539030e-01 7.78348595e-02 1.66301087e-01 3.92061353e-01
3.63557547e-01 -4.26550180e-01 -4.17960465e-01 5.26058793e-01
-8.60112667e-01 -6.61548693e-03 3.11099827e-01 9.16269660e-01
6.00003541e-01 -3.15192103e-01 1.08787370e+00 -2.02549458e-01
3.88097614e-01 -7.36489773e-01 -1.40875489e-01 5.56552231e-01
-2.96646059e-01 -4.99769598e-01 4.18833405e-01 -8.37294757e-01
-1.20109487e+00 -1.12826101e-01 -5.14628828e-01 -9.89547908e-01
-6.10542834e-01 3.27853113e-01 -5.13940692e-01 1.82746187e-01
3.15834105e-01 6.07429683e-01 6.64239451e-02 -6.70698464e-01
5.70123196e-01 1.50522768e+00 5.93592584e-01 -3.82974267e-01
5.62283397e-01 2.05028817e-01 -6.58328831e-01 -1.19276834e+00
-2.71200120e-01 -8.93304169e-01 -2.36767054e-01 -4.63908076e-01
8.46620739e-01 -1.17455006e+00 -5.68575621e-01 8.79344046e-01
-1.04655349e+00 -4.52602714e-01 -3.82843435e-01 4.44951385e-01
-6.17598176e-01 4.86103207e-01 -3.56031865e-01 -9.73649740e-01
-2.98628151e-01 -1.28085065e+00 1.37040949e+00 2.12615192e-01
1.86951652e-01 -6.35269582e-01 1.26342969e-02 5.67118824e-01
2.47972742e-01 -5.31220436e-01 4.47207510e-01 -8.85124862e-01
-9.86279845e-02 1.51710182e-01 -2.27013245e-01 6.03339136e-01
1.19523399e-01 -2.42494773e-02 -1.47846198e+00 -4.80425596e-01
-2.02856883e-01 -7.22441792e-01 7.46015728e-01 2.54764974e-01
1.06049383e+00 -1.50496334e-01 -1.46191135e-01 4.75676358e-01
1.11676812e+00 4.16253030e-01 7.48624265e-01 -2.75400698e-01
5.15654743e-01 5.82707822e-01 7.97326386e-01 5.20260632e-01
5.38852453e-01 8.84411752e-01 3.43093723e-02 -7.03925192e-02
-4.84820813e-01 -5.57042599e-01 7.37411797e-01 1.17927301e+00
3.34337860e-01 -5.56694508e-01 -8.03998053e-01 7.09243298e-01
-1.66317487e+00 -9.14070129e-01 2.96995789e-01 2.40412569e+00
9.92421389e-01 -2.80676603e-01 2.18818471e-01 2.03628540e-01
1.12116563e+00 1.16931394e-01 -6.52680278e-01 -3.26937556e-01
-1.67136341e-01 1.05378352e-01 2.00109273e-01 5.48366010e-01
-8.48497152e-01 1.04606009e+00 5.30101871e+00 1.41403627e+00
-1.26180613e+00 7.97588408e-01 3.60166460e-01 -3.87156725e-01
-3.49859595e-01 -2.97973216e-01 -9.30151880e-01 7.61485040e-01
1.24244177e+00 -6.50940835e-02 6.08259439e-01 5.61490774e-01
2.86588848e-01 5.22761010e-02 -1.08234334e+00 1.44504213e+00
3.14323604e-01 -8.91526401e-01 -1.79401468e-02 -1.40711114e-01
4.20251399e-01 1.47324905e-01 2.99878061e-01 6.39700770e-01
-2.81936347e-01 -7.81841338e-01 7.81468034e-01 3.74063760e-01
1.86396170e+00 -4.78067845e-01 3.86034697e-01 4.34078068e-01
-1.32666898e+00 9.59504955e-03 -1.08283788e-01 5.40338814e-01
3.75774294e-01 1.36285484e-01 -6.28711879e-01 5.00592053e-01
8.00935268e-01 4.60511476e-01 -6.43979758e-02 6.60577476e-01
5.93940355e-02 7.95061469e-01 -2.57489473e-01 2.45453000e-01
-9.75157842e-02 3.67460936e-01 7.02190101e-01 1.23803461e+00
6.44392908e-01 7.81730860e-02 2.82736346e-02 3.05750400e-01
-3.15288305e-01 2.89068341e-01 -6.80468082e-01 1.58279967e-02
8.79203260e-01 8.16543102e-01 -1.54789448e-01 -5.26441216e-01
-6.96278811e-01 9.66467977e-01 -1.48576438e-01 6.59035504e-01
-9.75959659e-01 -4.05436099e-01 6.22232378e-01 9.30350507e-04
4.01150554e-01 1.87979788e-01 2.04370826e-01 -1.25462246e+00
1.03162214e-01 -9.90067065e-01 3.60568047e-01 -1.05255318e+00
-1.23335361e+00 3.43974084e-01 1.44642636e-01 -1.31738269e+00
-2.79010206e-01 -2.61008412e-01 -4.20993745e-01 8.00588727e-01
-1.68660009e+00 -1.15194297e+00 -1.49758801e-01 1.19253051e+00
1.02028537e+00 -4.09881741e-01 7.98750877e-01 8.21894884e-01
-4.42946821e-01 1.29121125e+00 3.74715537e-01 -6.96961284e-02
1.28793943e+00 -5.41428387e-01 1.04539871e-01 5.01992226e-01
-7.29030296e-02 1.56614676e-01 4.30842727e-01 -4.22878772e-01
-1.38609016e+00 -1.00111926e+00 8.02839518e-01 7.28507489e-02
4.73288447e-01 -4.72117573e-01 -1.07989681e+00 4.24633294e-01
3.10474783e-01 -3.54781933e-02 1.00807714e+00 -1.09604180e-01
-5.91871142e-01 -2.94489741e-01 -1.31535733e+00 4.23492610e-01
1.08960938e+00 -1.02256167e+00 -5.05587220e-01 1.32493883e-01
1.32201338e+00 -2.83684105e-01 -8.38182032e-01 4.62183177e-01
5.43389797e-01 -5.80607176e-01 9.36279833e-01 -4.53044742e-01
1.49116710e-01 -8.41087252e-02 -7.21825004e-01 -1.12167037e+00
2.70304412e-01 -7.35913455e-01 -1.25190318e-01 1.79607904e+00
3.75577152e-01 -7.87390590e-01 2.82426953e-01 6.22107349e-02
-4.41575557e-01 -4.02272969e-01 -1.56132090e+00 -8.58004093e-01
1.36508077e-01 -6.53928995e-01 6.10782027e-01 9.74589944e-01
1.89567685e-01 3.30367267e-01 -6.15904868e-01 1.40146852e-01
3.21295410e-01 -2.45131060e-01 7.09983289e-01 -7.49153674e-01
-4.57333684e-01 -5.34976088e-02 -7.53739402e-02 -1.26148272e+00
3.55873495e-01 -8.79657328e-01 2.06324890e-01 -1.28282773e+00
3.30378622e-01 -2.66251236e-01 -4.73137081e-01 6.81453228e-01
8.18111300e-02 2.77575731e-01 4.15140361e-01 1.76264077e-01
-7.08757877e-01 6.91151202e-01 1.10822725e+00 -2.39197627e-01
-2.06718475e-01 -1.69847067e-02 -4.81713653e-01 4.30550367e-01
8.68665040e-01 -3.33829254e-01 -5.31849027e-01 -3.69029254e-01
-2.31591612e-01 4.83879358e-01 2.56439626e-01 -7.41868496e-01
2.60360211e-01 -1.21250130e-01 -3.01147133e-01 -6.61283493e-01
7.56968319e-01 -7.00803518e-01 3.20705660e-02 -1.13709860e-01
-4.03239310e-01 -4.47126269e-01 3.97584647e-01 5.47537148e-01
-3.11484456e-01 -2.69205928e-01 9.32234049e-01 1.47579104e-01
-8.69092107e-01 3.86271954e-01 -3.56341839e-01 3.13693911e-01
9.58516061e-01 -3.09707522e-01 -5.04698336e-01 -6.24896407e-01
-1.00001597e+00 2.90115446e-01 3.02482277e-01 6.12960994e-01
9.37107623e-01 -1.45189333e+00 -7.67870486e-01 2.06906945e-01
4.86225367e-01 -2.45757356e-01 9.40034270e-01 8.87395620e-01
2.21214160e-01 1.65557757e-01 -3.82429287e-02 -8.12523007e-01
-1.63613999e+00 6.80528283e-01 1.50599614e-01 2.08429247e-01
-3.20789993e-01 8.88800919e-01 3.47327292e-01 -5.64440131e-01
6.61351144e-01 1.25633612e-01 -5.00562452e-02 2.10544989e-01
5.48420966e-01 5.26543856e-01 3.68390838e-03 -1.02146697e+00
-4.05258656e-01 6.97268426e-01 2.08928704e-01 -5.11594892e-01
8.69468331e-01 -6.85991764e-01 3.28222156e-01 9.62648749e-01
1.51023257e+00 -7.38411769e-02 -1.11660337e+00 -4.49266821e-01
-7.59974897e-01 -3.66266787e-01 -1.06520385e-01 -8.47118497e-01
-8.95332634e-01 1.35875809e+00 1.01483452e+00 -8.50239694e-02
1.29315889e+00 3.08268189e-01 1.07342434e+00 9.39114615e-02
3.05277258e-01 -1.30963647e+00 1.79423735e-01 4.57712352e-01
8.47974598e-01 -1.36016393e+00 -7.07794487e-01 -1.91981003e-01
-9.13788021e-01 7.15113878e-01 4.81079519e-01 7.26197660e-01
7.74791777e-01 1.27397135e-01 1.82100952e-01 3.90689969e-01
-8.73323441e-01 -4.52281326e-01 1.57111526e-01 9.05110955e-01
1.52101725e-01 7.65894800e-02 2.21614808e-01 5.49569845e-01
1.39445946e-01 1.45747438e-01 2.35590011e-01 6.84564829e-01
-5.06465495e-01 -9.10419226e-01 -5.52129567e-01 1.65551022e-01
-3.50281805e-01 -3.60492736e-01 -7.31591731e-02 3.88335198e-01
1.68864980e-01 1.40340304e+00 -1.73172001e-02 -5.06989181e-01
3.10814083e-01 6.54881954e-01 2.26575255e-01 -6.32354379e-01
-2.09233865e-01 5.28057992e-01 6.38831779e-02 -3.25047404e-01
-3.43423188e-01 -7.05648541e-01 -1.10822320e+00 -3.15901726e-01
-4.32226896e-01 -5.11083826e-02 7.43328750e-01 6.37901127e-01
4.79429066e-01 4.69786137e-01 8.35338056e-01 -6.78366601e-01
-5.32218397e-01 -1.15089643e+00 -6.16869450e-01 3.00765842e-01
5.15428841e-01 -6.59129202e-01 -5.43002963e-01 2.19853535e-01] | [14.297340393066406, 5.130397796630859] |
cd61dc2e-a5df-40f1-93b2-bd68b450e8da | skillgpt-a-restful-api-service-for-skill | 2304.11060 | null | https://arxiv.org/abs/2304.11060v1 | https://arxiv.org/pdf/2304.11060v1.pdf | SkillGPT: a RESTful API service for skill extraction and standardization using a Large Language Model | We present SkillGPT, a tool for skill extraction and standardization (SES) from free-style job descriptions and user profiles with an open-source Large Language Model (LLM) as backbone. Most previous methods for similar tasks either need supervision or rely on heavy data-preprocessing and feature engineering. Directly prompting the latest conversational LLM for standard skills, however, is slow, costly and inaccurate. In contrast, SkillGPT utilizes a LLM to perform its tasks in steps via summarization and vector similarity search, to balance speed with precision. The backbone LLM of SkillGPT is based on Llama, free for academic use and thus useful for exploratory research and prototype development. Hence, our cost-free SkillGPT gives users the convenience of conversational SES, efficiently and reliably. | ['Tijl De Bie', 'Bo Kang', 'Nan Li'] | 2023-04-17 | null | null | null | null | ['feature-engineering'] | ['methodology'] | [ 1.73952147e-01 1.36954561e-01 -2.93540478e-01 -4.12762523e-01
-8.65507901e-01 -5.51576912e-01 6.68396175e-01 4.05724972e-01
-4.97741878e-01 8.04291308e-01 3.20844382e-01 -5.22332907e-01
-3.96011710e-01 -3.54010224e-01 -5.81348166e-02 -1.29293799e-01
1.91305667e-01 8.36476505e-01 2.58718636e-02 -4.62806225e-01
4.41731721e-01 8.13155323e-02 -1.53677273e+00 2.06124559e-01
1.02954030e+00 5.16115248e-01 5.47159016e-01 9.52792168e-01
-6.62551820e-01 8.76937032e-01 -6.46791518e-01 -7.84748912e-01
-3.01670860e-02 -3.94433379e-01 -1.28397584e+00 -1.68048248e-01
3.92928183e-01 -1.64626725e-03 -1.38023943e-02 6.08930171e-01
6.54741704e-01 4.16182041e-01 5.56102872e-01 -1.29536438e+00
-4.05632287e-01 5.22907197e-01 -4.14378010e-03 1.24297198e-03
9.16557491e-01 4.91317473e-02 8.68338048e-01 -7.09946156e-01
7.19593704e-01 1.04141355e+00 9.90761161e-01 6.44758701e-01
-1.25727177e+00 -6.15488112e-01 -4.15619351e-02 -1.26834601e-01
-1.16526794e+00 -7.05847144e-01 4.75742519e-01 -5.76007605e-01
1.15384650e+00 5.88305712e-01 7.43916392e-01 1.38716316e+00
1.29417539e-01 5.50791144e-01 1.28318918e+00 -3.38892788e-01
2.47549042e-02 5.24998248e-01 7.31902540e-01 9.27827954e-01
2.14112401e-01 -4.75643158e-01 -8.94780993e-01 -4.65558499e-01
8.00493300e-01 -8.90601948e-02 -2.48296373e-02 -1.22583531e-01
-1.24979699e+00 4.52289075e-01 -3.90223473e-01 4.33316708e-01
-2.91878164e-01 -3.92128050e-01 4.90392268e-01 9.44937646e-01
4.75029796e-01 1.00811911e+00 -6.97998762e-01 -1.10087740e+00
-1.12066650e+00 5.04028141e-01 1.33474672e+00 1.11985803e+00
5.89684010e-01 -4.87123162e-01 -3.98277640e-01 1.18695629e+00
3.47158052e-02 3.91286254e-01 8.26227486e-01 -1.14536798e+00
3.43882918e-01 7.37273455e-01 1.70577303e-01 -1.02258027e+00
-6.31011009e-01 -3.42301756e-01 -5.87766588e-01 -3.60503465e-01
5.88837445e-01 -1.70452610e-01 -3.60133469e-01 1.41578531e+00
-8.30325261e-02 8.11081380e-02 6.85727075e-02 3.44941735e-01
1.04649425e+00 2.00286776e-01 1.98253348e-01 -3.67834151e-01
1.34033215e+00 -9.76667106e-01 -8.09070885e-01 -4.85504568e-01
9.90302026e-01 -8.38980436e-01 1.45247924e+00 3.72778594e-01
-1.29617834e+00 -6.86900854e-01 -6.49436235e-01 -2.75466144e-01
-3.79524440e-01 -1.05650648e-01 9.41763997e-01 7.99607515e-01
-1.13791323e+00 1.02826059e+00 -5.57890832e-01 -6.78231478e-01
1.39771432e-01 2.88187027e-01 -5.95544040e-01 7.50666931e-02
-1.20612299e+00 1.03836918e+00 -1.73328947e-02 -4.43946928e-01
-5.84962894e-04 -6.97639704e-01 -1.04734862e+00 -5.04262745e-02
3.60437125e-01 -8.45776558e-01 1.67358661e+00 -5.29039919e-01
-1.84395278e+00 1.14562929e+00 -2.38801345e-01 -4.91943955e-02
3.59278560e-01 -2.35602543e-01 -2.82831788e-01 -3.12334776e-01
1.61711872e-01 5.43593429e-02 6.60176039e-01 -5.80588222e-01
-4.15869087e-01 1.23499951e-03 9.66813490e-02 3.52047712e-01
-5.56955934e-01 1.85883924e-01 -6.23329461e-01 -2.25599796e-01
-8.13048407e-02 -7.74969161e-01 -3.16722274e-01 -7.47934282e-01
-1.59808561e-01 -5.59559524e-01 2.33307388e-02 -9.22538042e-01
1.66557515e+00 -1.86261678e+00 2.16446832e-01 3.63568574e-01
4.29063529e-01 3.04644316e-01 -8.50053281e-02 6.18884921e-01
1.57288268e-01 8.00777972e-02 2.31019646e-01 -7.65112400e-01
2.63072908e-01 1.99209541e-01 1.59934536e-01 1.46038145e-01
-2.58986473e-01 9.93883431e-01 -1.27081776e+00 -6.84217155e-01
8.19606408e-02 -1.91182852e-01 -6.36504412e-01 4.34712023e-01
1.71606109e-01 4.17771697e-01 -2.14603096e-01 3.81168008e-01
2.39109993e-01 -3.90460223e-01 3.73825818e-01 2.82031447e-01
-1.96502686e-01 6.47469878e-01 -1.05476236e+00 2.18375015e+00
-1.05285788e+00 4.80248392e-01 2.26725608e-01 -6.03707552e-01
9.56077278e-01 2.11652339e-01 3.39681804e-01 -2.92031109e-01
1.03702419e-01 2.07093686e-01 -2.95795172e-01 -4.39826399e-01
8.31893563e-01 2.12314144e-01 -5.09086907e-01 7.54804552e-01
3.72088462e-01 -2.87815452e-01 5.62837899e-01 5.37174881e-01
1.23324203e+00 2.63609082e-01 5.44000745e-01 -3.22964817e-01
5.46774030e-01 -6.29911572e-02 3.99154842e-01 1.01248169e+00
-3.02370414e-02 1.51775494e-01 4.49430138e-01 5.58886342e-02
-7.42715538e-01 -8.73664856e-01 1.55786248e-02 1.57172263e+00
-4.57049012e-01 -1.21742392e+00 -8.54274929e-01 -6.48846686e-01
5.99422753e-02 6.22643530e-01 2.56962553e-02 -4.71600741e-02
-2.30076447e-01 5.99117205e-02 5.49301088e-01 1.57922924e-01
3.74872923e-01 -1.07712054e+00 -1.89106107e-01 2.65731573e-01
-3.96478236e-01 -1.01729393e+00 -6.62494838e-01 -4.68890220e-02
-5.70933342e-01 -8.46916378e-01 -6.99740171e-01 -7.98102319e-01
4.98925358e-01 2.16826275e-01 1.42727721e+00 1.70771778e-01
-1.36481956e-01 5.32949150e-01 -1.91820458e-01 -4.05721605e-01
-5.85452020e-01 4.11280751e-01 4.14301753e-01 -5.27206302e-01
5.41493893e-01 -8.88135374e-01 -1.29480615e-01 2.10937679e-01
-1.19369090e-01 4.08017218e-01 6.52570784e-01 6.39738560e-01
-5.60093969e-02 -2.68051296e-01 6.27652168e-01 -1.39202118e+00
1.25079656e+00 -4.86015886e-01 -9.34731960e-02 1.14661008e-01
-9.46252584e-01 9.63533204e-03 3.31370324e-01 -3.67794305e-01
-9.89987791e-01 -2.89063990e-01 -5.28421998e-01 -5.20303696e-02
-3.32091480e-01 5.73727906e-01 1.66355804e-01 6.61763176e-03
7.70357966e-01 3.23944747e-01 1.99164271e-01 -7.62080371e-01
2.83242285e-01 1.32304573e+00 6.99914157e-01 -6.22063696e-01
5.50168931e-01 -3.49995732e-01 -4.11003858e-01 -1.05793822e+00
-1.01232445e+00 -9.26495135e-01 -6.65157080e-01 -1.91103145e-01
3.86018306e-01 -7.71616936e-01 -1.00441468e+00 4.22316402e-01
-9.33805525e-01 -4.26424444e-01 -1.93085298e-01 2.48794228e-01
-7.47021914e-01 5.36137760e-01 -6.60583198e-01 -8.43228757e-01
-5.02079129e-01 -6.96329594e-01 9.80816066e-01 1.34549588e-01
-1.30263293e+00 -1.11664307e+00 -7.97719806e-02 8.20156217e-01
6.36892080e-01 -2.89914399e-01 6.86078131e-01 -8.84750545e-01
1.61139816e-01 -4.94473189e-01 1.13207586e-01 2.17710733e-01
2.31666058e-01 -5.53235054e-01 -8.08387101e-01 -2.27269590e-01
-2.94217765e-01 -5.13423741e-01 2.30449632e-01 -1.19673081e-01
1.20430386e+00 -2.88898289e-01 -1.98807538e-01 5.31575859e-01
5.80457509e-01 -2.94516623e-01 2.87548959e-01 2.41992816e-01
4.95599508e-01 7.17743397e-01 7.44761050e-01 5.19777000e-01
4.50892657e-01 4.98044550e-01 -7.74282694e-01 1.26593471e-01
9.87957269e-02 -4.44851726e-01 6.21709049e-01 1.31117809e+00
-3.04737955e-01 1.95527256e-01 -8.85590315e-01 3.55841279e-01
-1.96295726e+00 -9.31947768e-01 -1.03161044e-01 2.15846539e+00
1.29935825e+00 4.92686391e-01 2.80196398e-01 4.95706983e-02
2.11040340e-02 8.76038801e-03 -1.55197084e-01 -8.25720787e-01
3.58836114e-01 3.14461052e-01 3.71925592e-01 5.25510907e-01
-6.26093209e-01 9.18706596e-01 6.83376789e+00 9.50559497e-01
-4.88086045e-01 2.71755308e-01 9.56910774e-02 -1.74759045e-01
-2.04072446e-01 -9.55609530e-02 -1.03574109e+00 4.25884515e-01
1.41000009e+00 -6.86956108e-01 5.85682988e-01 9.42014396e-01
1.96970910e-01 -1.80503413e-01 -1.23700345e+00 1.40853715e+00
-6.41769245e-02 -1.30621827e+00 -2.79049277e-01 -1.21242136e-01
4.39854354e-01 -2.41109610e-01 -2.88950115e-01 9.11971331e-01
5.34341156e-01 -7.79416084e-01 2.60548323e-01 7.65979052e-01
9.36810017e-01 -7.75645196e-01 7.50976682e-01 8.07896674e-01
-1.02209663e+00 -7.87064782e-04 -1.21816613e-01 -5.57953298e-01
3.00861225e-02 4.06949103e-01 -7.63047934e-01 5.83622277e-01
2.75460094e-01 6.42146766e-01 -5.90456963e-01 3.39623064e-01
-1.20604308e-02 5.34702837e-01 1.10675484e-01 -1.60404831e-01
-4.33274470e-02 -2.66609251e-01 5.17055154e-01 1.58560050e+00
1.09516419e-01 -1.69562966e-01 6.97647214e-01 3.92334402e-01
1.71035230e-01 4.06991303e-01 -7.30472505e-01 -5.59665382e-01
7.24473417e-01 1.56859088e+00 -3.78540516e-01 -5.22903919e-01
-4.41728920e-01 1.27403295e+00 4.46835965e-01 1.11567145e-02
-3.04440886e-01 -7.71679044e-01 9.89223599e-01 3.84614617e-01
-4.63287771e-01 -5.24971128e-01 -2.75183201e-01 -1.03501785e+00
-2.42176101e-01 -1.13116157e+00 3.08500469e-01 -5.37204027e-01
-1.37150109e+00 4.66625959e-01 1.34584792e-02 -8.28547120e-01
-7.86387682e-01 -4.97674108e-01 -4.05930221e-01 1.17777836e+00
-8.90427947e-01 -9.73779738e-01 -2.94252247e-01 5.91871798e-01
7.50360072e-01 -2.73468256e-01 1.17913508e+00 3.02167892e-01
-5.83929360e-01 7.51300573e-01 -9.75314900e-02 -1.64854497e-01
1.08099616e+00 -1.53639412e+00 7.13444233e-01 2.87510723e-01
-1.65226698e-01 1.37491882e+00 8.48102868e-01 -8.82491112e-01
-1.32378280e+00 -8.58195603e-01 1.71964192e+00 -9.19460952e-01
8.64493966e-01 -7.05835819e-01 -6.44363463e-01 5.32536745e-01
9.86964032e-02 -7.56446421e-01 9.41548765e-01 9.66659725e-01
1.24134563e-01 1.56431496e-01 -8.12271833e-01 7.52944410e-01
1.57312155e+00 -8.25424910e-01 -7.92209923e-01 7.29844153e-01
6.04507089e-01 -5.41586757e-01 -1.21462154e+00 -3.44258487e-01
4.97379780e-01 -6.30931735e-01 8.90191078e-01 -6.83727801e-01
2.00145811e-01 3.32149178e-01 4.26398307e-01 -1.55885422e+00
-6.39582753e-01 -1.23605156e+00 -9.41670388e-02 1.31115198e+00
3.84904623e-01 -4.69125509e-01 5.90036094e-01 9.34336305e-01
-2.98868090e-01 -4.99803543e-01 -4.13452148e-01 -8.65069866e-01
-3.31968158e-01 -7.93855369e-01 5.09614289e-01 1.08864033e+00
7.06466138e-01 7.43732750e-01 -5.43294251e-01 -4.27809834e-01
3.52054536e-01 1.66092530e-01 1.15565550e+00 -1.84943330e+00
-4.32655603e-01 -3.53930920e-01 -6.31701276e-02 -1.09446478e+00
3.30386907e-01 -1.04415834e+00 -3.71464305e-02 -1.60903072e+00
2.15789050e-01 -3.34914267e-01 -3.36182266e-02 5.53028464e-01
-1.84800133e-01 1.84786208e-02 -1.79661196e-02 4.64660674e-02
-8.97754014e-01 -1.11925252e-01 1.11562312e+00 2.57823765e-01
-5.97963631e-01 3.08998883e-01 -1.07976437e+00 9.66022968e-01
7.75801659e-01 -2.99520016e-01 -5.00083804e-01 2.97049463e-01
3.68678123e-01 2.71626972e-02 -1.65241007e-02 -9.90191698e-01
3.57705563e-01 -3.54899347e-01 2.30788961e-02 -1.40177444e-01
3.58261019e-01 -3.00932318e-01 -7.72402808e-02 1.00982279e-01
-4.28986996e-01 1.36091441e-01 8.36837441e-02 2.02512026e-01
1.22907683e-01 -2.96367377e-01 2.90026307e-01 -4.31551665e-01
-7.66791642e-01 1.58782497e-01 -7.72228122e-01 -8.59884024e-02
5.58654130e-01 -2.48176128e-01 -1.98292509e-02 -6.27314866e-01
-9.13698971e-01 3.86934757e-01 3.71974677e-01 6.53552651e-01
3.34730119e-01 -1.09846628e+00 -4.25304890e-01 5.09454370e-01
2.08702177e-01 -3.89276177e-01 1.90929160e-01 1.03004885e+00
-5.42307496e-02 6.26659870e-01 -7.62500763e-02 -1.29461423e-01
-1.71706450e+00 3.26454520e-01 -1.04575679e-01 -5.21433294e-01
-5.32838702e-01 8.60090673e-01 -5.12356341e-01 -7.61934578e-01
2.65727341e-01 -1.69088706e-01 -5.47048807e-01 2.08143935e-01
7.64947593e-01 5.28732896e-01 2.99899876e-01 -3.18841189e-01
-1.71757996e-01 2.95678433e-02 -8.79550353e-02 -2.59474039e-01
1.12126219e+00 -3.81911755e-01 -1.92505941e-01 8.75057101e-01
8.79547656e-01 2.85003394e-01 -5.36139786e-01 -4.98283565e-01
3.74627054e-01 -5.17587543e-01 -1.34115396e-02 -6.28430367e-01
-6.70074597e-02 3.02096248e-01 7.76057616e-02 2.33553067e-01
7.04939663e-01 -7.00103715e-02 8.95146430e-01 8.65755022e-01
5.56408048e-01 -1.38064802e+00 -5.38558252e-02 1.01313519e+00
7.62808502e-01 -1.08081615e+00 -1.30641431e-01 -3.45104218e-01
-7.24611878e-01 8.49660754e-01 5.97590029e-01 4.54003692e-01
5.28930843e-01 2.94602334e-01 1.86616048e-01 -2.06749246e-01
-9.67417061e-01 -2.76898652e-01 3.56192023e-01 5.76386333e-01
9.09204185e-01 -1.90346446e-02 -6.68458879e-01 1.16822302e+00
-7.09421456e-01 2.08704576e-01 2.97954887e-01 1.18196976e+00
-4.69346017e-01 -1.52675200e+00 -1.43353447e-01 9.52635527e-01
-2.95770466e-01 -2.36643657e-01 -5.38688779e-01 4.30909216e-01
-1.21011779e-01 1.10946894e+00 -2.42331764e-03 -6.21349156e-01
6.26214087e-01 6.13584816e-01 3.62296075e-01 -1.06316972e+00
-8.55501592e-01 -5.33202589e-01 8.25108945e-01 -8.50610495e-01
-1.08933613e-01 -7.89079070e-01 -9.58192885e-01 -7.58956373e-01
-4.79073822e-02 5.65169096e-01 4.75102693e-01 1.01994586e+00
5.42368889e-01 3.53599846e-01 4.33228642e-01 -7.35533953e-01
-7.80887902e-01 -1.40379262e+00 -7.41232336e-01 5.78979433e-01
-4.81225289e-02 -5.86044729e-01 1.41177615e-02 -1.10900320e-01] | [12.726411819458008, 7.994873523712158] |
3fe7e3f5-5e0a-4058-8d04-93e7518fbba0 | attention-based-multi-modal-new-product-sales | null | null | https://dl.acm.org/doi/10.1145/3394486.3403362 | https://dl.acm.org/doi/pdf/10.1145/3394486.3403362 | Attention based Multi-Modal New Product Sales Time-series Forecasting | Trend driven retail industries such as fashion, launch substantial new products every season. In such a scenario, an accurate demand forecast for these newly launched products is vital for efficient downstream supply chain planning like assortment planning and stock allocation. While classical time-series forecasting algorithms can be used for existing products to forecast the sales, new products do not have any historical time-series data to base the forecast on. In this paper, we propose and empirically evaluate several novel attention-based multi-modal encoder-decoder models to forecast the sales for a new product purely based on product images, any available product attributes and also external factors like holidays, events, weather, and discount. We experimentally validate our approaches on a large fashion dataset and report the improvements in achieved accuracy and enhanced model interpretability as compared to existing k-nearest neighbor based baseline approaches. | ['Vikas Raykar', 'Satyam Dwivedi', 'SURYA SHRAVAN KUMAR SAJJA', 'Sumanta Mukherjee', 'Kushagra Manglik', 'Vijay Ekambaram'] | 2020-08-23 | null | null | null | acm-sigkdd-international-conference-on-3 | ['new-product-sales-forecasting', 'short-observation-new-product-sales'] | ['time-series', 'time-series'] | [-1.58221200e-01 -3.00299168e-01 -9.19664085e-01 -9.94096458e-01
-6.28576040e-01 -7.90392518e-01 6.82423413e-01 5.02260149e-01
-1.58161119e-01 2.84200877e-01 6.54121339e-01 -3.42332363e-01
-1.16039298e-01 -7.10941195e-01 -6.86164439e-01 -3.67501825e-01
-5.41993454e-02 8.48332942e-01 -4.06402111e-01 -4.07129496e-01
2.11132959e-01 3.41011792e-01 -1.28874576e+00 3.47997665e-01
3.75498295e-01 1.41680896e+00 2.43366241e-01 3.97833973e-01
-7.83096775e-02 6.77891612e-01 -8.51763561e-02 -9.22914565e-01
6.08243048e-01 4.71099466e-02 -2.10177705e-01 4.17350590e-01
-7.22297430e-02 -5.70984304e-01 -6.16738319e-01 6.89011514e-01
1.18987329e-01 1.88718840e-01 7.01714337e-01 -1.00911379e+00
-1.12483907e+00 1.18197858e+00 -5.79514265e-01 3.12053442e-01
3.91622521e-02 1.35188207e-01 1.21126127e+00 -7.53694594e-01
6.03993416e-01 7.65837908e-01 5.78281760e-01 -1.47079781e-01
-1.24649966e+00 -4.56185818e-01 6.79495156e-01 3.14342886e-01
-1.21425974e+00 -2.29407862e-01 9.70791817e-01 -2.37660557e-01
9.11359131e-01 1.95537478e-01 6.81278467e-01 1.19597125e+00
3.62942785e-01 8.70613098e-01 7.35075593e-01 -7.53493980e-02
1.53261736e-01 3.70963812e-01 -5.59248589e-02 -6.66699111e-02
-2.50786990e-01 2.34474853e-01 -4.28585947e-01 4.45034564e-01
5.12063563e-01 4.03123379e-01 3.67896646e-01 1.10445403e-01
-1.31276226e+00 9.81199563e-01 5.76272070e-01 3.89082544e-02
-9.88542080e-01 8.15219358e-02 5.06535709e-01 4.19137985e-01
9.00223136e-01 3.96139622e-01 -8.73306811e-01 -2.28035510e-01
-1.23059964e+00 2.92732358e-01 7.56747246e-01 1.19101512e+00
2.72577673e-01 1.86032310e-01 1.00551220e-02 5.00674844e-01
2.02303305e-01 5.24055839e-01 4.14222270e-01 -5.07562697e-01
4.31036472e-01 1.71057194e-01 2.51650870e-01 -1.04683530e+00
-4.52412367e-01 -7.61543453e-01 -6.61195815e-01 -5.95534682e-01
7.89431632e-02 2.55830004e-03 -1.12900794e+00 1.12297451e+00
3.64785106e-03 2.69370675e-01 -2.47767083e-02 8.58446717e-01
4.84099537e-01 9.92322683e-01 1.11674309e-01 -3.02886933e-01
1.12376440e+00 -8.83658826e-01 -8.04553390e-01 -1.18663058e-01
4.57943887e-01 -9.94041264e-01 5.74459851e-01 4.55495864e-01
-8.23222041e-01 -6.20497227e-01 -7.58543849e-01 9.66839790e-02
-6.20536327e-01 1.44118160e-01 1.12579739e+00 2.48588100e-01
-3.63980681e-01 5.14032781e-01 -6.03047490e-01 -2.25220963e-01
2.44010434e-01 2.85697967e-01 -3.77837033e-03 -2.50544518e-01
-1.11938643e+00 8.93709242e-01 3.48450065e-01 4.44200605e-01
-1.01751304e+00 -8.54088724e-01 -8.39539647e-01 3.01828515e-02
8.48152861e-02 -2.46408015e-01 1.44204640e+00 -7.60993600e-01
-1.21571231e+00 3.52584153e-01 -3.35337408e-02 -7.31751978e-01
2.13633493e-01 9.40594375e-02 -1.05193973e+00 -6.24527752e-01
-1.47385225e-01 6.70288801e-01 7.08704650e-01 -1.05505478e+00
-9.65774775e-01 -4.24377441e-01 -9.40959752e-02 2.35736519e-01
-1.89824164e-01 -8.93372018e-03 -4.38323081e-01 -8.49809170e-01
-7.48330057e-02 -1.11182475e+00 -5.29918134e-01 -5.61346889e-01
-5.22502005e-01 1.88348405e-02 4.29571122e-01 -7.97703207e-01
1.03939342e+00 -2.25038314e+00 -1.60366818e-01 2.61151969e-01
-2.06364617e-01 -2.66771853e-01 -1.87443748e-01 6.00380540e-01
-1.31734945e-02 -3.55234027e-01 3.64341021e-01 -2.80299306e-01
3.71267140e-01 3.34807694e-01 -7.54770577e-01 2.69815326e-01
-7.03001544e-02 1.14038765e+00 -7.25051284e-01 1.37652397e-01
3.37787569e-01 5.33265471e-01 -2.44689882e-01 -1.15828581e-01
-5.79885244e-01 5.04690230e-01 -2.90276736e-01 9.08441782e-01
5.71517885e-01 -2.61216581e-01 1.10695519e-01 -4.08530176e-01
-1.62535802e-01 3.07450086e-01 -8.92864048e-01 1.76975834e+00
-8.52607667e-01 5.06833375e-01 -4.36156929e-01 -8.37539673e-01
9.80012059e-01 7.40178823e-02 5.33087671e-01 -9.30273831e-01
3.10862124e-01 1.20020688e-01 -2.57411897e-02 -2.05163866e-01
9.55853999e-01 -2.31924549e-01 -3.81031841e-01 9.45288092e-02
-6.46539479e-02 2.08994582e-01 1.58909082e-01 -3.16332169e-02
5.82601488e-01 -3.23073268e-02 -1.56113908e-01 2.72944868e-02
-1.29512891e-01 3.77991199e-01 5.08606672e-01 2.21787900e-01
2.06584126e-01 4.60557550e-01 1.31662264e-01 -9.46892321e-01
-1.45804369e+00 -1.05242300e+00 -2.87999123e-01 1.23334968e+00
9.41665024e-02 -2.69059479e-01 -1.24371581e-01 -5.52488863e-01
3.65319401e-01 1.19738889e+00 -3.95575494e-01 -2.62126867e-02
-1.51757240e-01 -7.52427042e-01 -1.53211012e-01 9.28664744e-01
8.32507983e-02 -6.67646646e-01 1.29108906e-01 5.50256670e-01
1.79491844e-02 -1.21506619e+00 -7.44845569e-01 4.00893003e-01
-6.95067108e-01 -2.97231764e-01 -7.90988863e-01 -7.46445239e-01
3.70493323e-01 -2.19783522e-02 1.10452914e+00 -7.82143533e-01
-9.62196290e-02 3.99154909e-02 -2.38681614e-01 -6.15552664e-01
-1.46773830e-01 3.75213802e-01 2.04487532e-01 2.42985487e-01
9.01907265e-01 -4.20509160e-01 -7.84095228e-01 3.13732445e-01
-6.79596901e-01 -7.07567483e-02 8.19002569e-01 5.33561945e-01
8.26862931e-01 5.44232786e-01 8.40254605e-01 -8.71525586e-01
5.35234392e-01 -8.50800872e-01 -8.04132879e-01 4.88084294e-02
-1.08215952e+00 -1.28678963e-01 5.67286909e-01 -4.52209741e-01
-1.15917325e+00 3.22621226e-01 -2.61326879e-01 -4.27067280e-01
-1.34472594e-01 9.87281621e-01 2.53863484e-01 6.54379010e-01
9.53440145e-02 3.98877203e-01 -4.52898294e-01 -7.81770051e-01
7.07946658e-01 9.13666725e-01 5.99839866e-01 7.32912719e-02
6.15512788e-01 3.06681097e-01 -2.73359209e-01 -3.29216838e-01
-8.86017203e-01 -6.14677489e-01 -7.25338519e-01 -2.34316885e-02
6.20101213e-01 -1.35672843e+00 -6.74149513e-01 1.84561715e-01
-9.33353245e-01 6.35745823e-02 -4.98302639e-01 9.26211536e-01
-4.34273154e-01 -3.87393415e-01 -7.14171052e-01 -8.14217150e-01
-1.62811384e-01 -1.15896416e+00 1.17491865e+00 -1.69530913e-01
-2.00771138e-01 -1.16892958e+00 -1.30118132e-01 4.68250632e-01
5.65538287e-01 7.80473053e-02 9.66514766e-01 -8.17452371e-01
-5.76454341e-01 -6.81890368e-01 -1.02501415e-01 2.18331516e-01
1.42471462e-01 -2.72339433e-01 -7.10834265e-01 -1.45727426e-01
-3.71943027e-01 1.00932188e-01 8.81975591e-01 7.26063907e-01
9.80708301e-01 -4.13780749e-01 -4.23352480e-01 4.70529377e-01
1.39644623e+00 4.53307360e-01 4.47491318e-01 9.42646265e-02
6.06041849e-01 6.47182047e-01 8.86189997e-01 7.69908130e-01
8.63395274e-01 7.44753718e-01 4.29819942e-01 5.50707094e-02
2.59277642e-01 -6.37806594e-01 3.28869224e-01 1.19363451e+00
5.33801690e-02 -5.11811078e-01 -4.99624640e-01 9.22930539e-01
-1.78477764e+00 -7.94584215e-01 -8.31123367e-02 2.09170985e+00
5.44175565e-01 1.67628512e-01 2.42818475e-01 -3.10473859e-01
5.02622426e-01 8.97846520e-02 -9.85716462e-01 -6.29495203e-01
-2.01065823e-01 -2.16340348e-01 1.20378232e+00 6.54235855e-02
-1.13329637e+00 8.93729270e-01 7.11611366e+00 7.83232927e-01
-1.03181851e+00 6.48589656e-02 1.08947146e+00 -5.00002325e-01
-7.21381903e-01 -2.15177730e-01 -1.09316480e+00 6.95262492e-01
1.38189745e+00 -1.58546731e-01 6.54703081e-01 9.51881647e-01
3.17302257e-01 3.09066683e-01 -1.24157274e+00 9.99129117e-01
-2.97576021e-02 -1.55777967e+00 -1.07274897e-01 2.71945745e-01
9.23270285e-01 2.66263813e-01 7.43320107e-01 3.71573150e-01
5.34075022e-01 -9.25035357e-01 7.60805190e-01 5.86624086e-01
5.53185225e-01 -1.00558400e+00 8.62242579e-01 1.70627654e-01
-1.09711623e+00 -3.09253216e-01 -1.85951650e-01 -7.28485510e-02
7.37910092e-01 7.97433376e-01 -7.41465867e-01 5.43345451e-01
4.47628498e-01 1.02506149e+00 -1.03143267e-01 7.01455295e-01
2.10360065e-01 5.31310439e-01 -3.79654378e-01 1.53918818e-01
5.54661393e-01 -2.53168195e-01 1.10736638e-01 8.28294337e-01
5.72888196e-01 -9.54756811e-02 2.55621165e-01 5.81032753e-01
-2.50707269e-01 9.56278071e-02 -5.18315136e-01 -4.02768612e-01
1.95574626e-01 1.16635764e+00 -6.45333290e-01 -6.17170334e-02
-8.65798771e-01 1.12589550e+00 -3.62987965e-01 4.55724269e-01
-1.00330102e+00 -8.83877054e-02 9.76096034e-01 3.81445765e-01
8.08007419e-01 -3.60654294e-01 -1.08986497e-01 -1.11750209e+00
1.09266248e-02 -6.94281220e-01 3.20236117e-01 -6.45689964e-01
-1.74917996e+00 6.37834966e-01 -1.02326818e-01 -1.27769756e+00
-5.05914629e-01 -5.41319549e-01 -8.34863484e-02 6.43239439e-01
-1.76432753e+00 -1.64089358e+00 3.49275380e-01 3.27213645e-01
9.33298051e-01 -3.08613420e-01 5.52431047e-01 5.10277152e-01
-4.36221182e-01 4.29264873e-01 6.82785273e-01 -2.49620974e-02
5.50544977e-01 -1.08491647e+00 6.35719955e-01 4.60679799e-01
5.01900494e-01 3.38065177e-01 7.86556005e-01 -7.64256477e-01
-1.67621183e+00 -1.29026639e+00 1.11448753e+00 -3.53550136e-01
9.19988573e-01 -4.29568887e-01 -3.26022267e-01 1.23388672e+00
2.73752749e-01 -1.86306894e-01 9.53988016e-01 5.24842858e-01
-4.40379024e-01 -5.19416392e-01 -9.53339577e-01 2.68978596e-01
6.49777591e-01 -2.95349002e-01 -4.68453467e-02 6.71455026e-01
9.54922140e-01 -3.61795634e-01 -1.44403350e+00 1.21932298e-01
7.69930124e-01 -3.62020016e-01 8.62441242e-01 -7.75065482e-01
5.55226266e-01 -2.96636927e-03 -4.76322472e-01 -1.61051726e+00
-6.08676314e-01 -4.31926608e-01 -8.88208821e-02 1.34806824e+00
8.75782549e-01 -3.29147041e-01 8.12025726e-01 1.00075233e+00
-9.89978909e-02 -6.90150976e-01 -6.09588027e-01 -7.92545378e-01
-1.86506405e-01 -9.02852535e-01 1.27949572e+00 9.16869164e-01
1.63580861e-03 3.31406742e-01 -8.54771554e-01 2.87367254e-01
3.78323048e-01 7.32455969e-01 4.05141264e-01 -1.02463067e+00
-2.37933397e-01 -1.95206702e-01 -6.24256790e-01 -1.34318161e+00
3.37134376e-02 -8.86958122e-01 -2.28652626e-01 -1.34032404e+00
3.85515839e-02 -5.43250620e-01 -6.74219191e-01 1.86684147e-01
4.76427048e-01 3.13176394e-01 1.18863992e-01 9.76825431e-02
-5.12158513e-01 4.94277954e-01 1.02819431e+00 -4.77462232e-01
-2.03981623e-01 3.44626695e-01 -7.34791875e-01 4.60838288e-01
6.08068049e-01 -2.50426531e-01 -4.34001476e-01 -5.44583738e-01
5.65009654e-01 2.39325427e-02 -1.01984426e-01 -3.90771300e-01
2.52049148e-01 -2.56112039e-01 6.01519048e-01 -9.93669450e-01
5.09975612e-01 -1.20852470e+00 5.23911119e-01 4.79459986e-02
-6.79785907e-01 3.45657408e-01 3.68893519e-02 9.50850189e-01
-3.44966173e-01 5.69518507e-02 -6.95403367e-02 6.67629531e-03
-1.16747236e+00 5.95391810e-01 -4.58306968e-01 -6.94864333e-01
1.15213132e+00 -6.44654930e-02 2.21079905e-02 -7.50943184e-01
-9.94415998e-01 3.80950212e-01 1.88517675e-01 9.97704983e-01
4.59239393e-01 -1.56657195e+00 -8.73578906e-01 1.29074246e-01
4.36740428e-01 -4.67905015e-01 2.71900892e-01 7.20205903e-01
-2.29468480e-01 8.58784139e-01 4.32302281e-02 -2.14668006e-01
-6.02348983e-01 1.13811445e+00 -2.23377094e-01 -2.73340195e-01
-3.18720222e-01 6.54196918e-01 -3.18932757e-02 -3.27320874e-01
1.41726643e-01 -7.47258127e-01 -1.67111948e-01 3.33020627e-01
4.33716625e-01 1.50654182e-01 2.65185148e-01 -9.50091898e-01
-6.90853298e-02 1.72112793e-01 -4.30480152e-01 1.43821135e-01
1.77989697e+00 -5.32448232e-01 4.50078845e-01 7.25197434e-01
1.23535514e+00 -4.69650567e-01 -1.34541833e+00 -4.46364284e-01
4.84231859e-02 -7.48448551e-01 6.43503368e-01 -1.03674924e+00
-1.50533593e+00 4.40797687e-01 5.67824543e-01 3.72049570e-01
1.25503528e+00 1.67029232e-01 1.44127810e+00 -2.09353436e-02
4.82396156e-01 -1.08602035e+00 -6.02500916e-01 1.27033487e-01
5.51793277e-01 -1.56199646e+00 -1.46597192e-01 -2.63201207e-01
-1.14056170e+00 7.72892416e-01 -1.20488545e-02 1.45974532e-01
9.78396595e-01 2.79629409e-01 7.92781711e-02 -1.37546897e-01
-9.87542689e-01 -3.68908674e-01 3.60440701e-01 5.72233856e-01
3.79084021e-01 3.71591657e-01 1.90993324e-01 8.22611630e-01
-1.70535237e-01 4.93344441e-02 1.48811981e-01 3.91651154e-01
8.22962746e-02 -1.06790352e+00 2.29642317e-01 8.80667746e-01
-4.82843667e-01 -2.79979587e-01 1.79793090e-01 4.48553056e-01
3.84051539e-02 7.72880912e-01 6.60930336e-01 -5.50964892e-01
3.60171616e-01 -1.86704323e-01 3.12360466e-01 -4.09055501e-01
-6.07582688e-01 3.04361314e-01 2.44758070e-01 -2.67166972e-01
-1.56861007e-01 -9.47612166e-01 -8.41210485e-01 -6.75017715e-01
-4.48155999e-01 -1.37955338e-01 1.16842949e+00 8.12132001e-01
6.57020986e-01 4.52440649e-01 1.00614381e+00 -7.67141104e-01
-5.76868296e-01 -8.03210437e-01 -9.98784363e-01 3.73741150e-01
1.71565816e-01 -3.59559894e-01 1.12649798e-01 3.45086753e-01] | [7.1470794677734375, 2.902189016342163] |
6d22ec18-a90a-4ae4-9c21-55d4a08a709b | adversarial-attacks-on-machine-learning | 2002.09565 | null | https://arxiv.org/abs/2002.09565v4 | https://arxiv.org/pdf/2002.09565v4.pdf | Adversarial Attacks on Machine Learning Systems for High-Frequency Trading | Algorithmic trading systems are often completely automated, and deep learning is increasingly receiving attention in this domain. Nonetheless, little is known about the robustness properties of these models. We study valuation models for algorithmic trading from the perspective of adversarial machine learning. We introduce new attacks specific to this domain with size constraints that minimize attack costs. We further discuss how these attacks can be used as an analysis tool to study and evaluate the robustness properties of financial models. Finally, we investigate the feasibility of realistic adversarial attacks in which an adversarial trader fools automated trading systems into making inaccurate predictions. | ['Avi Schwarzschild', 'Tom Goldstein', 'Micah Goldblum', 'Ankit B. Patel'] | 2020-02-21 | null | null | null | null | ['algorithmic-trading'] | ['time-series'] | [-2.00465783e-01 1.85428560e-01 -1.49131306e-02 -1.25960290e-01
-4.27898139e-01 -1.39010203e+00 6.93715036e-01 8.63728598e-02
-2.87786573e-01 6.04976356e-01 -3.74264985e-01 -7.96568513e-01
1.07853167e-01 -9.67825413e-01 -8.50134313e-01 -3.71085882e-01
-5.30877829e-01 5.81791699e-01 6.21392615e-02 -1.80891335e-01
2.14613199e-01 7.99830317e-01 -4.37097609e-01 1.29915297e-01
4.54655141e-01 1.28816843e+00 -1.10270560e+00 7.13871479e-01
1.48272067e-01 1.17240083e+00 -9.97389078e-01 -1.10971177e+00
1.20276511e+00 -2.87842780e-01 -5.46679378e-01 -2.65388042e-01
2.53060579e-01 -8.24143112e-01 -3.58607918e-01 1.36053300e+00
1.51464239e-01 -2.04932272e-01 4.71536487e-01 -1.65429366e+00
-5.20561397e-01 1.02212405e+00 -3.50850821e-01 2.28209034e-01
-2.94791728e-01 5.97987831e-01 1.16319036e+00 -1.48357913e-01
3.20511132e-01 1.12898171e+00 6.19970202e-01 2.66158700e-01
-1.21200645e+00 -1.06036198e+00 -4.22770940e-02 -3.73443127e-01
-5.59478879e-01 -2.04363823e-01 8.57295692e-01 -4.32744563e-01
6.66789174e-01 1.99984625e-01 6.00582182e-01 1.09491932e+00
6.85340762e-01 6.67399108e-01 1.28496957e+00 -4.98063378e-02
3.78878027e-01 3.17501515e-01 4.81466129e-02 4.02528882e-01
5.19291937e-01 1.10096943e+00 -1.50843829e-01 -7.56841958e-01
7.21611261e-01 -1.44727483e-01 2.42355511e-01 -6.46924913e-01
-7.39221275e-01 1.50028229e+00 4.63465631e-01 4.27956581e-02
-2.43555620e-01 6.00630760e-01 6.64602399e-01 1.08025241e+00
4.65511769e-01 1.26075017e+00 -6.94288969e-01 1.41534314e-01
-8.01835418e-01 4.92406428e-01 1.13374186e+00 6.16411507e-01
6.66577518e-02 7.26859391e-01 3.66436213e-01 -2.55177189e-02
1.10328026e-01 4.78877246e-01 2.02441901e-01 -1.30516303e+00
3.35264951e-01 5.70682175e-02 3.65721196e-01 -8.49236548e-01
-1.54189810e-01 -5.63783586e-01 -3.94899130e-01 6.98912501e-01
6.30502224e-01 -5.86865008e-01 -4.38481569e-01 1.45072603e+00
-1.35122523e-01 1.84285015e-01 2.12794274e-01 5.90570629e-01
-3.85822892e-01 4.00952071e-01 -2.36459911e-01 -2.23132566e-01
6.59930885e-01 -8.08252215e-01 -2.53223151e-01 -3.33426371e-02
3.93614113e-01 -6.00022197e-01 5.64244926e-01 5.36448777e-01
-1.11478364e+00 5.54943178e-03 -1.27380776e+00 5.01893580e-01
-4.85313535e-01 -8.17885816e-01 8.15204024e-01 1.17647421e+00
-6.55858457e-01 9.18425620e-01 -1.01577890e+00 5.53334057e-01
5.58983445e-01 4.97210950e-01 7.16546774e-02 6.86436415e-01
-1.54233456e+00 1.07491291e+00 1.74926773e-01 -3.37387212e-02
-1.20629418e+00 -8.20848346e-01 -5.46801805e-01 2.05065720e-02
4.23368156e-01 -3.57466191e-01 1.71665359e+00 -1.44967437e+00
-1.52248752e+00 4.92907852e-01 7.58652508e-01 -1.49760330e+00
1.30550563e+00 -2.57236421e-01 -5.19847453e-01 6.92588016e-02
-3.05815220e-01 -2.95663811e-02 1.18217933e+00 -9.31772411e-01
-2.57526428e-01 -1.57083362e-01 4.24284458e-01 -3.38343829e-01
-1.49453789e-01 2.28488684e-01 9.02168870e-01 -1.07391095e+00
-5.93525946e-01 -1.06744587e+00 -3.20410192e-01 8.09845179e-02
-3.80874127e-01 4.21171218e-01 6.95497036e-01 -3.23090583e-01
8.21007311e-01 -2.15406919e+00 -4.14005339e-01 6.52539849e-01
1.80128455e-01 3.55598390e-01 1.09453268e-01 3.06007981e-01
-9.38011557e-02 5.91929436e-01 -3.26461434e-01 1.33021027e-01
4.62171018e-01 -9.62329581e-02 -1.19413877e+00 5.76895893e-01
2.32148930e-01 1.21687210e+00 -6.68953836e-01 1.22492775e-01
1.10818259e-01 -2.12505892e-01 -5.68053305e-01 3.71883839e-01
-5.11325121e-01 1.43649995e-01 -6.79277062e-01 5.66189706e-01
4.76777226e-01 8.95005763e-02 9.57185626e-02 3.72181535e-01
2.06396237e-01 1.11965381e-01 -7.21451342e-01 5.19161999e-01
-4.17678773e-01 5.10853291e-01 2.25861982e-01 -8.42255414e-01
6.26548171e-01 -1.38695519e-02 2.09432065e-01 -4.17934090e-01
5.40425539e-01 3.24808091e-01 4.55645472e-01 3.90138268e-01
1.24454513e-01 -6.88720703e-01 -4.12288547e-01 1.12802577e+00
-3.78970712e-01 -3.88836622e-01 -2.30249718e-01 6.10439330e-02
1.06191099e+00 -3.64209056e-01 2.58420050e-01 -1.15478814e-01
4.64154258e-02 9.83306170e-02 2.89585263e-01 1.04125690e+00
-5.06513834e-01 -1.46134011e-02 9.03339863e-01 -7.79007733e-01
-1.18706167e+00 -1.23404980e+00 3.71550396e-02 8.33878756e-01
-1.40474245e-01 2.04391375e-01 -6.52792871e-01 -1.16877413e+00
9.04856205e-01 8.38292539e-01 -8.49747598e-01 -5.52864134e-01
-5.28288782e-01 -6.80681884e-01 1.04725087e+00 6.59054399e-01
5.23214459e-01 -1.04602301e+00 -7.43421495e-01 2.25341916e-01
7.12860048e-01 -8.16071868e-01 -5.49852669e-01 3.24322462e-01
-7.20647454e-01 -1.34066248e+00 -3.23378235e-01 -1.69617236e-01
3.01925808e-01 -2.97277302e-01 1.15199888e+00 -7.20486417e-02
-8.89307633e-02 3.12318116e-01 1.55727556e-02 -7.31236160e-01
-1.14605224e+00 -2.96990927e-02 2.92625725e-01 -6.83639105e-03
1.93836182e-01 -3.52051973e-01 -4.30893719e-01 2.55386651e-01
-9.30463910e-01 -6.63963854e-01 3.33477736e-01 6.85542226e-01
9.76616517e-02 2.84555793e-01 7.08485961e-01 -1.36506522e+00
8.92290533e-01 -5.41163683e-01 -1.38072443e+00 2.94999152e-01
-8.38055491e-01 2.04016060e-01 1.10307968e+00 -5.91498911e-01
-6.75526381e-01 -2.41812050e-01 2.30816305e-01 -7.40553379e-01
3.03947777e-01 2.12288976e-01 -6.04992881e-02 -6.43394232e-01
5.52746594e-01 -9.54092145e-02 2.76315272e-01 -2.40865663e-01
4.91051197e-01 1.97006002e-01 1.85692713e-01 -5.24817109e-01
1.35968959e+00 3.14495534e-01 1.13808371e-01 -2.12548628e-01
-6.11505151e-01 5.87983072e-01 -2.95280010e-01 6.60225078e-02
2.68310964e-01 -4.63773966e-01 -8.54737639e-01 7.34963119e-01
-7.42784142e-01 -6.00078344e-01 -4.50268060e-01 9.57968310e-02
-7.86758840e-01 1.60867527e-01 -9.65853751e-01 -6.57460749e-01
-2.12624043e-01 -1.06936729e+00 1.44690275e-01 -2.69215822e-01
-3.82980347e-01 -1.20016265e+00 1.58911943e-01 1.80660278e-01
6.35600090e-01 5.15371978e-01 1.07978129e+00 -1.62861145e+00
-6.88957155e-01 -5.49355328e-01 1.45741284e-01 7.27466166e-01
-1.54119238e-01 7.33079016e-02 -8.09443176e-01 -3.58454883e-01
5.49138367e-01 -4.89558011e-01 5.93419731e-01 7.31903017e-02
9.14622605e-01 -9.44424391e-01 1.72025830e-01 6.39600694e-01
1.12118614e+00 4.43897843e-01 1.45697445e-01 6.32411182e-01
2.60532886e-01 5.16982317e-01 5.61754465e-01 3.59079838e-01
-5.05711079e-01 2.03190669e-01 5.25405705e-01 2.64717668e-01
7.11871922e-01 -3.40380788e-01 5.00677705e-01 6.35026172e-02
3.49142045e-01 -1.39532804e-01 -8.56688261e-01 5.22363707e-02
-1.45263779e+00 -9.75236058e-01 4.80682820e-01 1.99107230e+00
7.94248044e-01 8.13958406e-01 4.19674695e-01 -6.07856642e-03
6.50730908e-01 3.31847876e-01 -1.00895965e+00 -9.22609925e-01
1.32776760e-02 4.86893862e-01 1.07524145e+00 5.61723828e-01
-1.34140170e+00 9.71893847e-01 7.66428185e+00 4.96602207e-01
-9.49964225e-01 -8.12837780e-02 9.55200851e-01 -4.33406889e-01
-2.21365228e-01 -6.88160071e-03 -2.41392270e-01 6.73792839e-01
1.13798881e+00 -6.46815240e-01 6.15396976e-01 1.18914664e+00
-1.49511725e-01 5.89045882e-01 -1.30251575e+00 1.95228577e-01
-4.43191528e-01 -1.31186533e+00 2.45142117e-01 3.27654570e-01
6.53394222e-01 -1.88815713e-01 7.56921113e-01 4.34941232e-01
1.19275224e+00 -1.13878989e+00 7.81626284e-01 1.81439340e-01
2.89966792e-01 -1.36852920e+00 9.00036156e-01 1.93664432e-01
-5.88515341e-01 -2.56745428e-01 -1.28070295e-01 -1.76478952e-01
-5.61605059e-02 2.44900763e-01 -5.63222528e-01 6.04557432e-02
2.02761665e-01 3.31180662e-01 -4.38832313e-01 5.16479790e-01
-2.59334087e-01 7.29823470e-01 -3.33703279e-01 8.98753703e-02
6.53455555e-01 -1.95608690e-01 4.29904252e-01 6.68512404e-01
-1.10499719e-02 -9.60041583e-02 1.40665188e-01 1.10295630e+00
-3.55916977e-01 -4.23633426e-01 -9.28701520e-01 -5.87204099e-01
3.69242758e-01 7.63813972e-01 -6.52841508e-01 -2.79368870e-02
-2.76061416e-01 6.42310202e-01 9.35542062e-02 1.40897185e-01
-1.02342737e+00 -2.79564530e-01 8.59254360e-01 -1.58945229e-02
2.20282570e-01 -1.08859383e-01 -4.34516788e-01 -1.05222940e+00
-1.54086292e-01 -1.22398663e+00 4.08082575e-01 -2.35901892e-01
-1.71701157e+00 3.19258869e-01 -3.71350080e-01 -1.04247069e+00
-6.02741420e-01 -6.42723680e-01 -7.49674559e-01 4.68999445e-01
-1.10301661e+00 -7.81260073e-01 8.00917685e-01 5.56317985e-01
1.07010059e-01 -7.08873689e-01 4.71800566e-01 -4.48389590e-01
-3.69365305e-01 8.83129179e-01 5.05569041e-01 4.90503550e-01
4.07156914e-01 -1.30106413e+00 9.78132069e-01 9.06019330e-01
3.08197469e-01 5.59460402e-01 7.25150764e-01 -6.96457028e-01
-1.29209995e+00 -1.24078941e+00 6.74764439e-02 -8.04655015e-01
1.58941960e+00 -3.45746875e-01 -8.10558081e-01 1.10548401e+00
1.72847793e-01 4.73235585e-02 5.60068727e-01 -2.51883149e-01
-7.65324831e-01 -5.99966533e-02 -1.51722407e+00 5.06905138e-01
4.34416890e-01 -7.97643483e-01 -5.97104967e-01 1.27216831e-01
9.56362009e-01 -1.30219877e-01 -8.62149358e-01 1.74736008e-01
5.60133398e-01 -1.00463080e+00 8.47080529e-01 -1.02037966e+00
2.69766271e-01 2.23991975e-01 1.41206905e-01 -1.32984257e+00
1.15331650e-01 -1.15798843e+00 -4.90977108e-01 8.41904163e-01
5.66175044e-01 -1.25545633e+00 7.93914795e-01 8.70540380e-01
4.90889966e-01 -4.50923383e-01 -9.39333797e-01 -1.25774145e+00
9.60467219e-01 -1.96399167e-01 7.83709228e-01 1.16871560e+00
-1.32537251e-02 -2.86655784e-01 -3.93340409e-01 5.79620637e-02
7.58894861e-01 4.19015586e-01 5.25288880e-01 -8.95486116e-01
-6.26393199e-01 -8.50285530e-01 -3.02336395e-01 -2.10535377e-01
5.85818410e-01 -6.19126678e-01 -3.72607023e-01 -2.43444264e-01
-2.25862741e-01 -2.24899456e-01 -6.13197565e-01 2.51311004e-01
2.49364391e-01 2.03378424e-01 6.26841664e-01 2.51486450e-01
-1.91705912e-01 2.54475355e-01 7.35780001e-01 -4.27110761e-01
9.59271863e-02 5.61932027e-01 -7.53937304e-01 8.56782019e-01
1.18598521e+00 -5.33816755e-01 -3.13001305e-01 3.00280862e-02
3.01551521e-01 -1.64369456e-02 4.17671859e-01 -5.83803713e-01
-4.61328030e-02 -4.05065328e-01 2.56399125e-01 -2.18312405e-02
-1.31906688e-01 -1.14824712e+00 1.07590646e-01 1.01429605e+00
-7.33715892e-01 3.20023030e-01 1.29538074e-01 5.51191509e-01
-1.92835763e-01 -9.05097872e-02 1.09160244e+00 -2.68694729e-01
-5.40533103e-02 4.73997533e-01 -4.63013798e-01 3.79818022e-01
1.38508904e+00 2.63763279e-01 -3.34800065e-01 -6.56700134e-01
-8.39245141e-01 3.79884303e-01 6.94783747e-01 2.53607035e-01
3.05732608e-01 -1.00057125e+00 -5.98078430e-01 2.63769656e-01
-3.39203566e-01 -6.29540086e-01 -3.06543499e-01 1.90727457e-01
-8.88288975e-01 2.91699886e-01 -2.63183415e-01 3.18241954e-01
-7.51093090e-01 1.18041801e+00 7.73318172e-01 -4.54959989e-01
-2.64953136e-01 5.30435741e-01 2.24209785e-01 -2.25502193e-01
1.04760237e-01 -2.10764155e-01 5.62589347e-01 -7.89854750e-02
3.11518967e-01 5.06980777e-01 -2.17331260e-01 -9.69715789e-02
-2.48002455e-01 -4.68821079e-02 -4.04485226e-01 -3.60984296e-01
1.12791085e+00 4.31911170e-01 -2.78648324e-02 3.19451183e-01
1.01796186e+00 2.82761976e-02 -1.31146038e+00 -5.84742092e-02
2.94850498e-01 -3.28477502e-01 -3.14368308e-01 -1.04285836e+00
-1.32642114e+00 9.69151318e-01 2.04667702e-01 7.64539838e-01
7.58172035e-01 -4.62922245e-01 8.55134547e-01 5.70203543e-01
5.38383186e-01 -9.79117572e-01 -9.46478173e-02 5.38216114e-01
7.95336366e-01 -8.64870191e-01 -6.79567754e-02 1.79504216e-01
-7.71049500e-01 9.85790312e-01 3.11124504e-01 -9.18759286e-01
9.51389253e-01 6.92488313e-01 3.00246090e-01 1.42340899e-01
-7.18605578e-01 5.37702560e-01 -1.86854213e-01 3.54914397e-01
-2.83996463e-01 3.33101630e-01 1.90231934e-01 8.42484057e-01
-6.71772480e-01 -2.96311140e-01 8.06003392e-01 9.36300337e-01
-2.15548441e-01 -1.02806854e+00 -2.63309121e-01 4.82442230e-01
-9.30849493e-01 -1.82863533e-01 -1.06220496e+00 1.10120308e+00
-4.30714488e-01 7.53643870e-01 -1.72903351e-02 -3.88336122e-01
1.39068305e-01 2.32323796e-01 1.78349808e-01 -3.12983185e-01
-1.20362747e+00 -2.41967216e-01 5.80935366e-02 -6.32126272e-01
1.19957559e-01 -5.92127800e-01 -7.70949125e-01 -8.87259126e-01
-2.55002528e-01 2.87131876e-01 1.32349387e-01 7.04373717e-01
1.40107661e-01 1.54942632e-01 1.27234197e+00 -2.49860898e-01
-1.54393160e+00 -4.90047485e-01 -9.53055024e-01 1.28724501e-01
6.51777327e-01 -4.29321617e-01 -1.00871515e+00 -3.75990003e-01] | [5.69517707824707, 7.610881328582764] |
431eb38f-ac34-4bf2-91e2-4cb5a4c02b97 | privacy-preserving-representations-are-not | 2305.04603 | null | https://arxiv.org/abs/2305.04603v1 | https://arxiv.org/pdf/2305.04603v1.pdf | Privacy-Preserving Representations are not Enough -- Recovering Scene Content from Camera Poses | Visual localization is the task of estimating the camera pose from which a given image was taken and is central to several 3D computer vision applications. With the rapid growth in the popularity of AR/VR/MR devices and cloud-based applications, privacy issues are becoming a very important aspect of the localization process. Existing work on privacy-preserving localization aims to defend against an attacker who has access to a cloud-based service. In this paper, we show that an attacker can learn about details of a scene without any access by simply querying a localization service. The attack is based on the observation that modern visual localization algorithms are robust to variations in appearance and geometry. While this is in general a desired property, it also leads to algorithms localizing objects that are similar enough to those present in a scene. An attacker can thus query a server with a large enough set of images of objects, \eg, obtained from the Internet, and some of them will be localized. The attacker can thus learn about object placements from the camera poses returned by the service (which is the minimal information returned by such a service). In this paper, we develop a proof-of-concept version of this attack and demonstrate its practical feasibility. The attack does not place any requirements on the localization algorithm used, and thus also applies to privacy-preserving representations. Current work on privacy-preserving representations alone is thus insufficient. | ['Zuzana Kukelova', 'Fredrik Kahl', 'Torsten Sattler', 'Kunal Chelani'] | 2023-05-08 | null | null | null | null | ['visual-localization'] | ['computer-vision'] | [-3.38222831e-02 -3.90277356e-01 -6.84285397e-03 -2.43045717e-01
-8.25024128e-01 -1.32227135e+00 3.14960837e-01 3.68983924e-01
-5.39957583e-01 2.96950042e-01 -4.78492707e-01 -4.78272259e-01
1.93039551e-01 -6.66075587e-01 -8.68086338e-01 -8.31366718e-01
-1.96011424e-01 4.00756985e-01 3.29194307e-01 1.46651119e-01
4.48125452e-01 1.40973735e+00 -1.05183554e+00 -2.12272272e-01
5.04953824e-02 1.22290146e+00 8.91603231e-02 9.98145700e-01
1.89464986e-01 4.63746756e-01 -7.82096922e-01 -3.34933609e-01
6.49039268e-01 6.07943162e-02 -5.89588821e-01 1.38785496e-01
4.46245283e-01 -5.86433411e-01 -3.98654610e-01 1.39147592e+00
3.86294276e-01 3.95019986e-02 1.39423534e-01 -1.73604453e+00
-3.99969548e-01 -1.78405613e-01 -6.48481131e-01 2.80223906e-01
6.76555812e-01 -2.37740159e-01 6.11075163e-01 -3.76594245e-01
7.40497828e-01 7.75043964e-01 4.33451027e-01 4.36576575e-01
-1.16025817e+00 -6.84216142e-01 6.22765496e-02 1.39101580e-01
-1.99701452e+00 -4.28794444e-01 6.22372925e-01 -6.17965609e-02
4.98353899e-01 6.69615507e-01 1.81112841e-01 7.30581939e-01
2.01996565e-01 3.31770152e-01 1.06935310e+00 -3.15081120e-01
4.73174661e-01 6.06701434e-01 -1.36859357e-01 4.62533206e-01
6.13248050e-01 -6.21302202e-02 -3.66881400e-01 -9.48715806e-01
8.28564286e-01 2.32379436e-01 -5.93372881e-01 -1.27548075e+00
-8.46031189e-01 8.29921424e-01 4.54851359e-01 -1.14703430e-02
-2.13646591e-01 3.27448338e-01 3.21216732e-01 3.83287579e-01
2.32612967e-01 2.37111911e-01 -4.04929638e-01 5.06615527e-02
-6.48593903e-01 2.40430817e-01 1.01854515e+00 1.29103887e+00
7.29216695e-01 -3.79249215e-01 6.81415379e-01 -1.56499222e-01
2.11777344e-01 6.72289312e-01 -1.17630035e-01 -9.68062341e-01
1.75191417e-01 1.33486375e-01 4.70274150e-01 -1.51283991e+00
1.50261819e-01 1.09559938e-01 -4.24025208e-01 4.08599734e-01
3.52339357e-01 -8.89284089e-02 -3.24249655e-01 1.79580462e+00
5.74622750e-01 3.57293695e-01 7.26043284e-02 9.41562831e-01
1.09566748e-01 5.41589320e-01 -2.68072754e-01 -2.00675860e-01
1.44432056e+00 -2.50343293e-01 -3.34879726e-01 8.09753779e-03
3.41338247e-01 -6.77584291e-01 2.70502120e-01 2.85989374e-01
-6.04137897e-01 -5.59946820e-02 -1.11516714e+00 1.10317081e-01
-6.04485452e-01 -2.37852216e-01 4.61265147e-01 1.05409563e+00
-1.23712432e+00 1.09665744e-01 -8.55229259e-01 -7.87982225e-01
2.67836452e-01 7.80408740e-01 -8.02411675e-01 -2.30210707e-01
-6.92831099e-01 8.12085569e-01 8.29912648e-02 -2.36675873e-01
-8.45964551e-01 -2.51091987e-01 -7.70553052e-01 -3.74792255e-02
4.73947704e-01 -4.03931916e-01 1.08204651e+00 -7.50925004e-01
-8.23347449e-01 9.13917840e-01 -2.51398355e-01 -6.85719669e-01
6.81812227e-01 1.28269186e-02 -2.94732213e-01 5.96610606e-01
2.86174297e-01 2.51460433e-01 1.03841233e+00 -1.42998660e+00
-7.64030457e-01 -8.54648352e-01 3.58098954e-01 9.12512690e-02
-1.41344145e-01 2.67138124e-01 -5.23710489e-01 -1.06268376e-01
2.89451510e-01 -1.13351691e+00 -4.19231772e-01 5.18137157e-01
-2.65292138e-01 2.90828437e-01 1.42318022e+00 -3.83778811e-01
4.19446290e-01 -2.43543077e+00 -5.73061585e-01 7.03134120e-01
2.43706435e-01 5.33575155e-02 2.27065831e-01 4.83906388e-01
3.45812477e-02 3.22816938e-01 2.95018643e-01 -4.58330363e-01
-6.59305975e-02 1.99881569e-01 -7.59343684e-01 1.46792006e+00
-4.14247364e-01 3.22309583e-01 -7.84686983e-01 -2.67390013e-01
4.03026551e-01 6.20818675e-01 -4.79720533e-01 1.43520638e-01
1.14544593e-01 5.20386338e-01 -6.53736472e-01 6.04252100e-01
1.14541411e+00 -1.10132605e-01 1.66658640e-01 1.19198129e-01
-7.53438519e-03 1.99353751e-02 -1.24731112e+00 1.59025311e+00
-3.26468349e-01 5.33127844e-01 5.51109791e-01 -7.45968342e-01
8.19514036e-01 5.46941817e-01 6.01213813e-01 -2.73699965e-02
2.43188828e-01 1.92210019e-01 -5.66794395e-01 1.17268376e-02
5.10289371e-01 4.09070998e-01 -3.01613212e-01 5.80783486e-01
-4.33917344e-01 9.49570313e-02 -5.78060329e-01 2.46926412e-01
1.29530990e+00 -2.55641103e-01 3.25879395e-01 1.81035139e-02
5.17286897e-01 -1.09792901e-02 3.08533490e-01 1.12973332e+00
-3.80311161e-01 4.53182667e-01 2.11627349e-01 -3.66472006e-01
-8.80468607e-01 -1.12329626e+00 -9.12038144e-03 5.71447611e-01
7.28322506e-01 -5.18263042e-01 -5.62236726e-01 -8.76260579e-01
1.45453498e-01 4.09475148e-01 -3.64217192e-01 -7.74837211e-02
-3.16362679e-01 4.78930213e-02 5.32456934e-01 1.92567080e-01
5.18914044e-01 -4.34959173e-01 -1.02998435e+00 -1.66732162e-01
-2.01906413e-02 -1.26435411e+00 -6.10596299e-01 4.63142283e-02
-5.96543968e-01 -1.28125191e+00 -3.14909071e-01 -5.03780365e-01
1.12546980e+00 9.66110706e-01 6.14112973e-01 6.38927892e-02
-1.97666526e-01 1.01584291e+00 -2.32944965e-01 -3.67289901e-01
-2.15906516e-01 -1.66410431e-01 3.18655491e-01 2.25622430e-01
6.29634798e-01 -5.62676370e-01 -4.85959619e-01 3.05123448e-01
-7.45942235e-01 -8.46151590e-01 1.37915447e-01 1.79192349e-01
7.14616776e-01 4.18384761e-01 -2.38559678e-01 -7.42768645e-01
2.61149496e-01 -4.15261865e-01 -1.24024427e+00 1.14917807e-01
-3.22592348e-01 -3.46050262e-01 4.65206742e-01 -4.84576613e-01
-2.83423483e-01 6.38558626e-01 2.27015600e-01 -9.15763021e-01
-4.90603030e-01 -1.02098808e-01 -5.42508841e-01 -8.67838562e-01
2.15388849e-01 4.04969931e-01 5.85818514e-02 -3.88630837e-01
4.13430452e-01 6.94138408e-01 5.95030069e-01 -3.64384860e-01
1.14210975e+00 9.52900946e-01 2.93372542e-01 -9.54610169e-01
-3.24671455e-02 -8.14231575e-01 -3.61383289e-01 8.22891742e-02
6.37679100e-01 -8.58761489e-01 -1.37821114e+00 3.73693258e-02
-1.37796688e+00 5.69952130e-01 -2.53701359e-01 2.32456580e-01
-6.96388841e-01 6.42163575e-01 -6.08679354e-02 -1.01498282e+00
8.65779817e-02 -1.29655457e+00 8.28682721e-01 1.18101545e-01
-1.32589857e-03 -7.16317236e-01 -2.51912743e-01 1.19017035e-01
3.37960660e-01 4.28141803e-01 4.20992166e-01 -1.03386819e+00
-1.09951401e+00 -1.04503703e+00 -1.52894482e-01 6.28663227e-02
1.26009837e-01 -3.93030465e-01 -9.46045995e-01 -7.43062377e-01
5.14123380e-01 1.58317015e-01 -1.49264233e-02 2.78342366e-01
1.09357357e+00 -6.33415639e-01 -4.82873887e-01 7.65100658e-01
1.69067442e+00 2.57118583e-01 4.27952349e-01 4.46685880e-01
5.56996286e-01 3.39303076e-01 6.13171220e-01 4.31491196e-01
2.51717418e-01 8.50651979e-01 8.21682453e-01 6.44563437e-02
7.53645718e-01 -3.36385548e-01 2.37960279e-01 -1.48903117e-01
3.49725634e-01 -2.30729267e-01 -4.99214143e-01 3.26382995e-01
-1.78289950e+00 -7.69398868e-01 1.71011358e-01 2.60308862e+00
1.07128412e-01 -2.65302002e-01 3.88967693e-02 -2.93263346e-01
8.86995733e-01 2.00231746e-01 -6.98198318e-01 -5.39712250e-01
1.56681135e-01 -8.23814496e-02 1.43164241e+00 4.07924354e-01
-1.29808748e+00 7.43405163e-01 5.75795746e+00 4.54523981e-01
-1.17414534e+00 2.32127860e-01 4.23167974e-01 4.97827828e-02
5.81497252e-02 3.78527015e-01 -7.63167083e-01 2.57604897e-01
8.56195331e-01 -4.45687056e-01 4.17112589e-01 1.49082232e+00
-3.62528348e-03 -2.32704163e-01 -1.14037669e+00 1.17613900e+00
3.59947234e-01 -1.28135192e+00 -2.60670453e-01 7.18707621e-01
1.28453240e-01 -2.53689289e-02 2.37315923e-01 -4.64521319e-01
3.51509064e-01 -7.38019049e-01 7.12350309e-01 -1.36831462e-01
8.12670529e-01 -1.15309525e+00 7.14362860e-01 6.16040230e-01
-1.27641642e+00 9.82514247e-02 -6.66418672e-01 3.73718113e-01
5.56415804e-02 -1.08269956e-02 -9.43741441e-01 4.91851270e-01
8.73482585e-01 1.73770130e-01 -5.33377886e-01 1.11390996e+00
-1.84022382e-01 1.93955719e-01 -7.16881752e-01 5.65841682e-02
1.92838982e-02 -1.73799813e-01 8.75136316e-01 8.07328165e-01
3.72287661e-01 1.24618098e-01 3.19162458e-01 6.12878442e-01
-1.12508737e-01 -1.69652831e-02 -1.30024564e+00 2.18873903e-01
7.98678637e-01 1.00539386e+00 -8.04267526e-01 1.75478920e-01
-4.04848516e-01 1.41070187e+00 -1.02999322e-01 1.81495994e-01
-7.18668640e-01 -3.77070785e-01 1.11635554e+00 2.92857528e-01
5.37199974e-01 -6.33929431e-01 8.50369781e-02 -1.11227047e+00
1.51567506e-02 -8.55483592e-01 4.51015919e-01 -5.62743723e-01
-9.77150142e-01 4.48221713e-01 -8.08334351e-02 -1.32950890e+00
-1.73060194e-01 -3.52331012e-01 -2.30986953e-01 8.46595407e-01
-1.30996406e+00 -1.04538774e+00 3.88205908e-02 1.09224081e+00
-7.93776438e-02 -9.46522632e-04 1.15337205e+00 9.80929211e-02
2.84776725e-02 6.00376129e-01 2.73691565e-02 3.47127497e-01
5.57872772e-01 -9.35413599e-01 3.28711510e-01 1.33825362e+00
6.85742557e-01 9.93247449e-01 7.78618991e-01 -5.44697285e-01
-1.88553154e+00 -9.12688255e-01 8.80612612e-01 -6.81595325e-01
4.91950035e-01 -6.91695333e-01 -5.29332697e-01 9.15166020e-01
-1.53110147e-01 4.20810372e-01 5.77992499e-01 -4.29059148e-01
-5.74199378e-01 -2.04701796e-01 -1.68623316e+00 4.70802665e-01
4.86823678e-01 -9.62733924e-01 -8.74265656e-02 4.74926591e-01
5.86840868e-01 -4.21514124e-01 -4.39344764e-01 -7.18650147e-02
2.31018588e-01 -9.23068523e-01 1.15272665e+00 -3.74185801e-01
-7.48910666e-01 -6.84953809e-01 -7.08054841e-01 -6.49336457e-01
1.31775692e-01 -8.84269416e-01 -6.61409423e-02 1.17831123e+00
-1.75074875e-01 -9.43458855e-01 1.20719528e+00 9.79232192e-01
7.29454756e-01 3.17451693e-02 -1.46616852e+00 -8.37150037e-01
-3.91743749e-01 -4.42785054e-01 6.83350801e-01 8.07424188e-01
-3.82303804e-01 -3.24407458e-01 -3.73858243e-01 1.36738348e+00
9.00661707e-01 4.27790061e-02 1.12677932e+00 -8.77175093e-01
-1.80868711e-02 2.83167273e-01 -1.09760416e+00 -1.08922124e+00
1.42516851e-01 -5.32565236e-01 -5.55800907e-02 -8.88956010e-01
-1.49083793e-01 -4.80621457e-01 -2.15380222e-01 2.99977690e-01
5.53979456e-01 3.27992916e-01 3.35586876e-01 4.25705731e-01
-5.36530018e-01 -9.38587785e-02 3.21640074e-01 6.24953844e-02
5.05073033e-02 6.03555262e-01 -6.72722936e-01 7.15780079e-01
7.85123825e-01 -7.66124606e-01 -5.06977618e-01 -1.60666883e-01
-1.39430523e-01 1.34488180e-01 7.49479830e-01 -8.26266885e-01
6.56949937e-01 -2.00284533e-02 3.16706866e-01 -5.55043042e-01
6.05049312e-01 -1.84902894e+00 3.80476624e-01 5.24954975e-01
-4.91988882e-02 4.17228401e-01 -4.57518660e-02 9.23265100e-01
7.88826197e-02 -1.74585685e-01 7.36670017e-01 -1.63023740e-01
-7.74495602e-01 5.59050858e-01 -3.02523851e-01 -5.30162990e-01
1.43880498e+00 -4.60911095e-01 -1.81220487e-01 -8.70851159e-01
-5.21271288e-01 -7.01349527e-02 1.24045420e+00 2.00180158e-01
7.65373349e-01 -9.04943526e-01 -1.12390034e-01 3.29599589e-01
2.38085657e-01 -2.06337109e-01 -1.55854404e-01 5.56394637e-01
-8.12640011e-01 2.58389890e-01 4.30608317e-02 -5.81104636e-01
-1.70302796e+00 1.07545114e+00 1.55318469e-01 2.71585494e-01
-6.94616079e-01 7.56741166e-01 3.51769626e-02 -4.23447415e-02
4.56633061e-01 2.03900263e-01 3.74456972e-01 -3.80955398e-01
7.77402103e-01 9.91460681e-02 -6.75262809e-02 -1.03493249e+00
-8.51662338e-01 4.37305003e-01 -2.29361698e-01 -2.72871733e-01
1.02585542e+00 -5.39808452e-01 -2.00508058e-01 -9.13951173e-02
1.41674948e+00 6.01206720e-01 -1.11669815e+00 -9.86402407e-02
2.33040527e-02 -1.17543089e+00 -2.33577378e-02 -4.24754918e-02
-1.23306632e+00 5.78277349e-01 8.42257917e-01 3.52542877e-01
9.99124885e-01 1.11153856e-01 5.98360360e-01 4.22406286e-01
1.26392937e+00 -5.98224044e-01 -4.10776436e-01 -8.28221217e-02
3.32776040e-01 -9.56051409e-01 1.76394925e-01 -5.42637527e-01
-4.52266097e-01 8.66596043e-01 1.29374474e-01 -4.00063723e-01
7.29665518e-01 4.18204248e-01 2.16824174e-01 -1.56994723e-02
-2.19955727e-01 1.62876621e-01 -2.87796319e-01 1.10197711e+00
-5.53209722e-01 -9.07439217e-02 2.86152571e-01 1.37237355e-01
-1.93576645e-02 -3.21377993e-01 6.01584554e-01 1.32938910e+00
-3.09576750e-01 -1.18584573e+00 -7.66216755e-01 -1.65287942e-01
-6.20701194e-01 2.87151337e-01 -5.03727853e-01 7.15072215e-01
7.40017323e-03 1.09940386e+00 -8.63953754e-02 -2.54633754e-01
1.50728837e-01 -3.43616337e-01 1.48879021e-01 -5.53824782e-01
-3.26337099e-01 -3.69998105e-02 -3.23980242e-01 -9.94925916e-01
-1.83105245e-01 -8.73873353e-01 -8.89036477e-01 -7.02410340e-01
-4.75548625e-01 4.16673154e-01 1.09466231e+00 2.78187692e-01
4.08144474e-01 -3.18408161e-01 1.14320958e+00 -6.81406379e-01
-6.12743676e-01 1.48985729e-01 -9.51742470e-01 3.32220495e-02
5.84457040e-01 -3.28429759e-01 -6.19765580e-01 -1.21352121e-01] | [7.55098819732666, -2.1384921073913574] |
0ea99eaa-be6c-4cb7-9f85-8022148fe19e | gleam-greedy-learning-for-large-scale | 2207.08393 | null | https://arxiv.org/abs/2207.08393v1 | https://arxiv.org/pdf/2207.08393v1.pdf | GLEAM: Greedy Learning for Large-Scale Accelerated MRI Reconstruction | Unrolled neural networks have recently achieved state-of-the-art accelerated MRI reconstruction. These networks unroll iterative optimization algorithms by alternating between physics-based consistency and neural-network based regularization. However, they require several iterations of a large neural network to handle high-dimensional imaging tasks such as 3D MRI. This limits traditional training algorithms based on backpropagation due to prohibitively large memory and compute requirements for calculating gradients and storing intermediate activations. To address this challenge, we propose Greedy LEarning for Accelerated MRI (GLEAM) reconstruction, an efficient training strategy for high-dimensional imaging settings. GLEAM splits the end-to-end network into decoupled network modules. Each module is optimized in a greedy manner with decoupled gradient updates, reducing the memory footprint during training. We show that the decoupled gradient updates can be performed in parallel on multiple graphical processing units (GPUs) to further reduce training time. We present experiments with 2D and 3D datasets including multi-coil knee, brain, and dynamic cardiac cine MRI. We observe that: i) GLEAM generalizes as well as state-of-the-art memory-efficient baselines such as gradient checkpointing and invertible networks with the same memory footprint, but with 1.3x faster training; ii) for the same memory footprint, GLEAM yields 1.1dB PSNR gain in 2D and 1.8 dB in 3D over end-to-end baselines. | ['Mert Pilanci', 'Morteza Mardani', 'John M Pauly', 'Shreyas Vasanawala', 'Christopher M Sandino', 'Arjun D Desai', 'Tolga Ergen', 'Arda Sahiner', 'Batu Ozturkler'] | 2022-07-18 | null | null | null | null | ['mri-reconstruction'] | ['computer-vision'] | [ 1.18744366e-01 -3.08264215e-02 -1.74608409e-01 -5.52563965e-01
-9.16328728e-01 -2.97010923e-03 -1.24321310e-02 5.94008863e-02
-9.82110977e-01 4.34744239e-01 3.10163736e-01 -6.04992390e-01
3.19449529e-02 -3.34232211e-01 -8.60585272e-01 -6.95209801e-01
-8.10695410e-01 4.57878739e-01 2.33799919e-01 2.40791187e-01
-6.08489662e-02 6.12144351e-01 -7.36246645e-01 2.86640376e-01
4.57977176e-01 1.00689292e+00 1.88386634e-01 8.74861419e-01
1.72778606e-01 1.02415788e+00 -3.67553681e-02 2.30735853e-01
2.59293586e-01 -1.60978183e-01 -7.67765939e-01 -2.42049649e-01
6.89501524e-01 -8.14077377e-01 -6.77680135e-01 7.40431070e-01
9.82462406e-01 2.14103073e-01 1.14901192e-01 -6.69184983e-01
-1.59034312e-01 6.09261692e-01 -7.97196269e-01 5.63385606e-01
-3.80573332e-01 2.29378849e-01 3.40504467e-01 -9.55515087e-01
5.42076766e-01 7.73401022e-01 1.12802935e+00 5.93956411e-01
-1.44321585e+00 -5.03663063e-01 3.60183567e-02 -1.73746068e-02
-9.75350559e-01 -3.55449557e-01 6.51852250e-01 -1.64600775e-01
1.23033690e+00 1.78000391e-01 7.02816069e-01 6.85204148e-01
5.74385345e-01 6.79525197e-01 1.20010436e+00 -9.93464440e-02
2.02886075e-01 -6.10277355e-01 4.60932434e-01 9.94147956e-01
1.18125550e-01 2.04933718e-01 -5.02623439e-01 -3.24992269e-01
1.10810721e+00 -7.85347819e-02 -3.53669733e-01 -2.85629809e-01
-1.30972242e+00 8.60699654e-01 7.93444514e-01 -4.40052561e-02
-6.09384954e-01 5.41019201e-01 9.13824618e-01 1.46753967e-01
5.60206711e-01 1.99323684e-01 -3.61088067e-01 -1.65394694e-01
-1.10105312e+00 1.00673333e-01 7.98070490e-01 3.94619942e-01
4.13617969e-01 3.76113147e-01 -1.11846089e-01 9.49988186e-01
1.06900372e-01 4.44428951e-01 5.96535981e-01 -1.07015526e+00
3.56629670e-01 -7.13627860e-02 -4.20383871e-01 -9.06288564e-01
-1.05844235e+00 -8.21516991e-01 -1.49307668e+00 3.84917378e-01
2.21013680e-01 -4.05770183e-01 -1.25438118e+00 1.76062059e+00
5.67993581e-01 3.06701124e-01 -4.57910568e-01 1.21990526e+00
7.32484996e-01 5.90965033e-01 5.60934506e-02 -1.80274442e-01
1.41085279e+00 -1.38121569e+00 -2.93975294e-01 -5.59034765e-01
8.75436127e-01 -4.71684694e-01 1.20554996e+00 3.02776545e-01
-1.38792896e+00 -1.88590318e-01 -1.12115455e+00 -1.30038425e-01
4.34419721e-01 3.99731137e-02 9.50432062e-01 4.31698412e-01
-1.32197011e+00 9.15949345e-01 -1.80440843e+00 3.35412800e-01
4.80923444e-01 6.94911718e-01 -9.66788828e-02 1.54415546e-02
-6.94346964e-01 6.23147011e-01 2.11847484e-01 2.81127453e-01
-1.03359175e+00 -1.29085410e+00 -6.76014602e-01 -6.71992078e-02
2.38069043e-01 -8.38190317e-01 1.34499013e+00 -5.76997340e-01
-1.53138530e+00 8.75133336e-01 2.00445801e-02 -7.72871614e-01
5.14013052e-01 -3.39737892e-01 -1.41812161e-01 3.63172054e-01
-1.44654168e-02 6.23045325e-01 7.38578796e-01 -5.72429001e-01
5.10333218e-02 -4.79119182e-01 -4.59413499e-01 1.59351528e-01
-2.39105806e-01 4.49140668e-02 -6.54625893e-01 -7.56604791e-01
5.50679982e-01 -1.30407190e+00 -7.44671524e-01 3.91961724e-01
-2.13759482e-01 4.32578832e-01 4.97316241e-01 -9.55333769e-01
8.38594675e-01 -1.95700765e+00 -1.33370489e-01 2.19517022e-01
6.99441671e-01 -1.03909837e-03 1.10543743e-02 -4.64542866e-01
-3.25783759e-01 -4.25564349e-01 -4.03664380e-01 -4.33643013e-01
-3.83232415e-01 2.63154060e-01 -8.38566851e-03 6.67774260e-01
-2.39772528e-01 9.41606998e-01 -7.57333934e-01 -4.26124334e-01
-9.98319760e-02 7.14693248e-01 -9.44775105e-01 1.08819775e-01
2.39664555e-01 7.40306616e-01 -2.64315575e-01 3.80276769e-01
6.86621368e-01 -7.41483271e-01 4.22173381e-01 -5.77060759e-01
8.37778449e-02 3.78695697e-01 -1.01668727e+00 2.14969277e+00
-7.24380672e-01 4.29660082e-01 4.59231377e-01 -1.15924060e+00
2.78964877e-01 1.90563515e-01 7.23663330e-01 -9.21602011e-01
3.89143936e-02 3.80937725e-01 -5.24883531e-02 -2.13572189e-01
2.49004573e-01 -1.80337548e-01 1.13377072e-01 7.93903589e-01
-3.65663953e-02 5.62129617e-02 9.80742835e-03 2.58940578e-01
1.26334381e+00 -1.07363485e-01 -1.44178018e-01 -4.76392508e-01
6.23349473e-02 -2.83475649e-02 5.73420942e-01 1.05184710e+00
-5.36352098e-02 4.77178186e-01 3.76570255e-01 -1.01883399e+00
-1.07698739e+00 -1.05926514e+00 -1.04770288e-01 1.23211443e+00
-2.98469394e-01 -7.72070810e-02 -6.73900068e-01 -3.15620095e-01
-2.10514098e-01 2.68502474e-01 -4.31649894e-01 9.49398354e-02
-1.29007900e+00 -1.37611663e+00 5.57566226e-01 6.87600911e-01
6.02819085e-01 -7.41281331e-01 -9.84384000e-01 4.60118651e-01
-8.97345226e-03 -1.14450121e+00 -7.10662603e-01 4.68258470e-01
-1.66829002e+00 -6.68772280e-01 -8.10105979e-01 -6.74330950e-01
9.03898656e-01 7.27407411e-02 1.43413663e+00 1.15223452e-01
-5.03031731e-01 7.36249015e-02 3.69102955e-01 2.14185342e-01
-3.13504726e-01 1.73673794e-01 6.75444826e-02 -5.80745101e-01
-3.91031355e-01 -1.03325951e+00 -1.06407797e+00 4.97147627e-02
-7.61436760e-01 4.81744230e-01 8.25605094e-01 1.14549196e+00
8.99640322e-01 -5.75929224e-01 9.91192088e-02 -9.91958737e-01
5.34569025e-01 -1.82245955e-01 -6.42367363e-01 -5.78379892e-02
-7.91151047e-01 4.05644149e-01 5.39347291e-01 -6.00003362e-01
-7.43372798e-01 2.09908426e-01 -4.17041630e-01 -3.73489410e-01
2.06116110e-01 6.66220129e-01 7.34115303e-01 -6.13054931e-01
8.13145638e-01 2.67190129e-01 2.62068629e-01 -4.73415494e-01
2.48487607e-01 8.53128433e-02 7.77551293e-01 -8.23140502e-01
1.94507882e-01 5.88853419e-01 2.44912684e-01 -4.92693275e-01
-7.47095346e-01 -9.88131538e-02 -1.77344412e-01 -1.55564144e-01
8.28018725e-01 -1.09621215e+00 -8.13518286e-01 6.82479441e-01
-9.89683390e-01 -7.95148551e-01 -1.34699732e-01 8.69567811e-01
-4.90072906e-01 4.69683647e-01 -1.49373281e+00 -4.33842950e-02
-1.18234313e+00 -1.39827120e+00 7.39147067e-01 -9.76182222e-02
-3.50921601e-02 -8.53328526e-01 1.00027613e-01 1.09137639e-01
9.18154478e-01 2.80804843e-01 1.04562962e+00 -1.39764965e-01
-5.05894899e-01 -6.85956851e-02 -4.39658791e-01 3.07718813e-01
-3.63881379e-01 -9.47844207e-01 -6.32200360e-01 -6.68200374e-01
4.78288114e-01 -5.79471827e-01 8.71941805e-01 7.81679988e-01
1.47723246e+00 -3.39798868e-01 -2.79284511e-02 1.22481894e+00
1.39316177e+00 -1.63701966e-01 3.42926383e-01 2.03042313e-01
8.66638839e-01 -1.68267623e-01 -1.66286275e-01 2.60926247e-01
2.26910740e-01 5.34826815e-01 1.33491322e-01 -4.14784104e-01
-4.10699546e-01 1.71055242e-01 1.10271506e-01 1.43984842e+00
-2.29910016e-02 6.11477435e-01 -1.22250831e+00 3.49932134e-01
-1.68472660e+00 -3.15734595e-01 -8.94987583e-02 2.15484905e+00
1.19027853e+00 2.93973625e-01 8.52426142e-02 -3.41770709e-01
3.42509657e-01 2.81946570e-01 -8.45481575e-01 -2.03323409e-01
2.35789567e-01 4.85043049e-01 8.45008910e-01 6.28832340e-01
-1.02507973e+00 4.72372919e-01 6.25619268e+00 7.29877710e-01
-1.67898202e+00 7.75758088e-01 1.15463805e+00 -6.34177566e-01
2.79742450e-01 -2.14439765e-01 -4.60967273e-01 1.99387431e-01
9.65429425e-01 3.03676188e-01 5.61797261e-01 1.03800333e+00
2.11713284e-01 -1.01178728e-01 -9.56155419e-01 1.22186577e+00
-3.85891572e-02 -1.66637766e+00 -2.52075136e-01 -1.33319885e-01
7.18381047e-01 9.03504550e-01 -8.52350369e-02 1.69842109e-01
1.18678227e-01 -8.82400513e-01 3.10009122e-01 1.43579662e-01
8.90086353e-01 -5.97302675e-01 4.41856384e-01 1.98795125e-01
-8.29865098e-01 2.87570208e-01 -2.32575148e-01 -3.85547318e-02
4.24256355e-01 1.12780201e+00 -5.57567477e-01 9.41589698e-02
8.47049117e-01 2.07540616e-01 -3.31555516e-01 9.36303437e-01
-5.23738824e-02 8.68258238e-01 -6.31224215e-01 1.35034099e-01
5.24083257e-01 -6.96583241e-02 4.61467892e-01 1.33808649e+00
2.42386255e-02 1.23256549e-01 3.63224357e-01 6.57534540e-01
-2.28566319e-01 -1.43936411e-01 1.20361056e-02 3.27266932e-01
-8.04648455e-03 1.32652473e+00 -9.77226377e-01 -4.97860312e-01
-3.06170940e-01 1.09289229e+00 5.91011107e-01 2.79192567e-01
-9.77237642e-01 -2.48720825e-01 2.86532193e-01 2.01797873e-01
1.02197252e-01 -6.89676225e-01 -6.09796286e-01 -1.12601328e+00
2.49534294e-01 -8.17769349e-01 3.01256925e-01 -4.39863235e-01
-1.15352261e+00 7.97148466e-01 -2.42154166e-01 -7.39775538e-01
-3.01962376e-01 -5.68084896e-01 -3.44452351e-01 6.98215067e-01
-1.41054583e+00 -7.55773187e-01 -3.54167372e-01 5.15028775e-01
1.83365867e-01 2.68800974e-01 8.19505751e-01 7.11792231e-01
-3.65945756e-01 5.42227149e-01 8.43878910e-02 3.82665634e-01
4.64461803e-01 -1.02463746e+00 5.78367174e-01 7.68819392e-01
-2.50087023e-01 5.69928288e-01 3.32971722e-01 -6.77162707e-01
-1.77415931e+00 -1.01127350e+00 5.48976123e-01 2.97364891e-01
7.54623771e-01 -2.13938594e-01 -9.56960797e-01 6.23983383e-01
-6.78527281e-02 7.35109746e-01 4.26287502e-01 2.99821198e-01
-2.86174655e-01 -1.81269065e-01 -9.55632567e-01 6.07878327e-01
1.12652779e+00 -5.52778900e-01 -7.37277493e-02 8.79123211e-01
6.41452134e-01 -1.29578745e+00 -1.03201973e+00 3.94842654e-01
4.94001687e-01 -7.65543699e-01 1.20979226e+00 -4.99848098e-01
5.24177194e-01 4.41432334e-02 2.49687269e-01 -9.87930179e-01
-4.34885234e-01 -6.16127968e-01 -4.82453555e-01 -4.30253446e-02
2.82324344e-01 -5.81205368e-01 1.09492362e+00 8.09488475e-01
-4.55830872e-01 -1.33214998e+00 -1.22875226e+00 -7.44704545e-01
9.19580311e-02 -7.33625233e-01 3.74933258e-02 8.16715598e-01
-3.28893989e-01 3.61869931e-01 -5.21995068e-01 1.27797276e-01
1.03511572e+00 8.67658257e-02 3.37519675e-01 -4.86611724e-01
-8.53954494e-01 -4.80859101e-01 -2.31446460e-01 -1.33270085e+00
-1.65380284e-01 -1.10933602e+00 3.01901381e-02 -1.12149596e+00
3.89221758e-01 -6.49657369e-01 -3.47304374e-01 6.60147071e-01
-1.52022198e-01 5.18542111e-01 1.46159977e-01 2.19360188e-01
-5.32262206e-01 2.81554431e-01 1.15923846e+00 -3.89084220e-02
-2.53685057e-01 -3.53563488e-01 -3.40473652e-01 7.72084892e-01
6.51511729e-01 -7.40421176e-01 -3.20553631e-01 -1.17850280e+00
1.23738773e-01 5.08501887e-01 6.00549698e-01 -1.27126181e+00
5.77264786e-01 3.62794816e-01 5.84504724e-01 -3.87586951e-01
2.61707306e-01 -3.36475432e-01 -3.79755944e-02 9.38608050e-01
-3.40984583e-01 4.02056873e-01 4.80177015e-01 2.11580247e-01
5.96534461e-02 6.31317124e-02 1.04935670e+00 -2.61263609e-01
-3.48371416e-01 6.39843404e-01 -2.68456489e-01 3.45017314e-01
3.02076906e-01 1.51287645e-01 1.57133579e-01 -2.05213517e-01
-1.10876155e+00 7.32835755e-02 -4.03409526e-02 -1.02214724e-01
7.06111670e-01 -1.17160606e+00 -8.30860972e-01 1.76446289e-01
-6.02893651e-01 6.37074560e-02 7.22823799e-01 1.36702013e+00
-9.71654177e-01 4.92332652e-02 -1.13103159e-01 -1.06269944e+00
-1.06156540e+00 1.81705862e-01 6.77913308e-01 -9.44104135e-01
-1.17942178e+00 9.52514410e-01 1.23005934e-01 -6.55375838e-01
3.02149624e-01 -4.41320866e-01 6.93720341e-01 -6.58180237e-01
6.84947133e-01 3.00791711e-01 4.54498500e-01 2.96797957e-02
-4.73111510e-01 3.70214134e-01 -4.40962374e-01 -9.68387648e-02
1.72140181e+00 3.60096425e-01 -3.24597359e-01 9.04040039e-02
1.45221055e+00 -3.67948651e-01 -1.33094227e+00 -4.36923951e-01
-2.47903958e-01 -1.45878941e-01 7.67387569e-01 -8.68197680e-01
-1.57432246e+00 7.77901769e-01 1.13179684e+00 -5.51491618e-01
1.21743000e+00 -3.17269176e-01 1.33086073e+00 6.80488110e-01
2.47014135e-01 -1.03181159e+00 1.19430218e-02 5.46040058e-01
8.09768915e-01 -1.10265195e+00 4.47074145e-01 -2.06246391e-01
-3.45989347e-01 1.07058072e+00 3.75807822e-01 -3.34510386e-01
8.40011239e-01 7.89293587e-01 -3.81379277e-02 -4.03071821e-01
-5.39646387e-01 5.98345757e-01 1.94015250e-01 6.58316687e-02
5.08936405e-01 1.22491866e-01 -3.50968182e-01 2.85990328e-01
2.98131648e-02 3.44395220e-01 9.77114215e-02 1.09328926e+00
-1.25758186e-01 -6.59752965e-01 -9.15549025e-02 7.15215206e-01
-5.94302773e-01 -4.06942606e-01 5.76615930e-01 5.83540320e-01
-4.71868217e-01 1.46370262e-01 1.01389490e-01 -8.12260062e-03
1.32121637e-01 -1.58785447e-01 7.04774320e-01 -2.48419985e-01
-8.41390848e-01 1.18476525e-01 4.52588201e-02 -1.04578578e+00
-1.99115761e-02 -2.98878163e-01 -1.57112777e+00 -4.19863969e-01
6.47427514e-02 -8.33092034e-02 9.32688236e-01 7.38229573e-01
7.63197124e-01 6.90408468e-01 3.10434014e-01 -1.15051413e+00
-8.72516990e-01 -6.74174190e-01 -3.23666453e-01 2.99431294e-01
2.41789654e-01 -1.98501363e-01 -7.46024773e-02 -2.76063532e-01] | [13.500245094299316, -2.381458044052124] |
73e117ff-26b5-4332-ad2e-0ef565b8fde3 | real-time-object-detection-yolov1-re | 2305.17786 | null | https://arxiv.org/abs/2305.17786v1 | https://arxiv.org/pdf/2305.17786v1.pdf | Real-time Object Detection: YOLOv1 Re-Implementation in PyTorch | Real-time object detection is a crucial problem to solve when in comes to computer vision systems that needs to make appropriate decision based on detection in a timely manner. I have chosen the YOLO v1 architecture to implement it using PyTorch framework, with goal to familiarize with entire object detection pipeline I attempted different techniques to modify the original architecture to improve the results. Finally, I compare the metrics of my implementation to the original. | ['Michael Shenoda'] | 2023-05-28 | null | null | null | null | ['real-time-object-detection'] | ['computer-vision'] | [-2.83893608e-02 -4.05656546e-01 3.21754754e-01 -3.44733387e-01
-1.41996428e-01 -4.59724844e-01 4.10855114e-01 -1.09223559e-01
-6.63787186e-01 5.24875782e-02 -4.18579012e-01 -5.12781143e-01
1.11688957e-01 -6.16514921e-01 -3.32509518e-01 -3.79708976e-01
5.21003678e-02 1.20447673e-01 1.14599144e+00 -2.19427198e-01
6.13572061e-01 7.47107506e-01 -1.81901598e+00 3.95312965e-01
-4.91575226e-02 7.90886164e-01 4.14865315e-01 1.55228555e+00
-3.50548699e-02 8.91341448e-01 -6.52712166e-01 1.00361757e-01
6.14092469e-01 -1.33273348e-01 -7.36428738e-01 1.81115270e-01
6.28230691e-01 -3.04070920e-01 -2.04938307e-01 1.15071249e+00
5.32703042e-01 -6.96613267e-03 5.75168490e-01 -1.30988646e+00
-5.03387004e-02 1.71159416e-01 -6.49123311e-01 1.18757951e+00
2.14579895e-01 5.61861217e-01 3.75949204e-01 -8.91432762e-01
5.87918818e-01 1.29293478e+00 6.49332166e-01 2.92008281e-01
-1.00924599e+00 -5.38748920e-01 -3.49404179e-02 6.60177827e-01
-1.31878626e+00 -5.22178888e-01 2.63415724e-01 -5.04346907e-01
1.17274916e+00 4.24019426e-01 3.75894189e-01 6.48138642e-01
2.77391851e-01 8.09178054e-01 1.00694692e+00 -7.33083308e-01
-5.07042594e-02 3.24523628e-01 5.45185983e-01 6.29629493e-01
2.73302794e-01 1.80838183e-01 -1.21040590e-01 2.44449824e-01
7.41757095e-01 -6.66804984e-02 1.12702347e-01 -2.27851480e-01
-1.16673100e+00 9.05017793e-01 4.79972959e-01 2.78091580e-01
-1.42647028e-01 2.25325942e-01 8.25548828e-01 2.41676539e-01
-2.66404480e-01 2.81766623e-01 -4.09027457e-01 8.06345120e-02
-8.28501999e-01 2.43851870e-01 5.39555728e-01 8.37191403e-01
3.34358722e-01 -6.58443421e-02 -1.20570496e-01 1.58295467e-01
7.60241389e-01 3.21155936e-01 4.67259496e-01 -1.08863628e+00
5.49857579e-02 6.36233926e-01 3.65167111e-02 -8.09716582e-01
-5.70709467e-01 -2.10812137e-01 -2.43062526e-01 1.10231841e+00
7.77067900e-01 7.18837380e-02 -1.14831281e+00 6.15997434e-01
5.87398171e-01 1.14272147e-01 -1.88262507e-01 9.42507267e-01
1.07037306e+00 5.81600666e-01 2.93076336e-01 7.44639486e-02
1.52700019e+00 -1.01989031e+00 -5.11536360e-01 -2.99335599e-01
6.24509573e-01 -1.25232077e+00 6.36319816e-01 6.18539512e-01
-6.84553862e-01 -9.30920362e-01 -1.48653972e+00 -1.69692203e-01
-4.03090656e-01 3.44978839e-01 5.36223173e-01 9.24744487e-01
-9.24963713e-01 5.89886248e-01 -8.28433275e-01 -6.65277541e-01
7.48746854e-04 4.00053442e-01 -3.45799625e-01 2.07521334e-01
-5.79848886e-01 1.04576802e+00 9.86187041e-01 9.21816006e-02
-6.24359369e-01 -2.05044493e-01 -5.06824195e-01 -2.72450358e-01
4.59336579e-01 -6.56008184e-01 1.60028827e+00 -7.46585488e-01
-1.31425774e+00 1.10971582e+00 1.03550903e-01 -6.93258107e-01
6.24248385e-01 -1.28893524e-01 -1.79537743e-01 3.34293157e-01
5.99441566e-02 7.04489410e-01 1.14424670e+00 -6.23666465e-01
-1.47516692e+00 -3.21375728e-01 -3.66869941e-03 7.98344612e-02
3.70906919e-01 5.27512372e-01 -4.25767779e-01 -3.46618176e-01
1.43111944e-01 -7.19149709e-01 -2.56584585e-01 2.84161270e-01
-1.24686189e-01 -4.68320131e-01 1.21275806e+00 -3.53907526e-01
8.09243619e-01 -2.27040005e+00 -3.96868765e-01 -1.27033740e-01
2.71374047e-01 7.05146849e-01 1.58862740e-01 -2.04277765e-02
-1.66638017e-01 -2.97923625e-01 2.85729110e-01 -1.79228395e-01
-3.01182628e-01 6.71500713e-02 -2.17279449e-01 6.47184908e-01
1.29822344e-01 4.58355069e-01 -6.36576951e-01 -8.20205927e-01
8.60479295e-01 2.78335869e-01 -9.61361006e-02 2.60686755e-01
-2.32140422e-02 -2.72086531e-01 -4.22729760e-01 7.79542565e-01
6.44860148e-01 2.14848608e-01 -4.13453519e-01 -1.85235277e-01
-5.49984932e-01 -1.16088770e-01 -1.65820622e+00 1.07123005e+00
2.10336328e-01 1.14176464e+00 1.82282910e-01 -1.01078176e+00
6.17394805e-01 2.34560952e-01 1.68030426e-01 -5.81468582e-01
4.73656774e-01 -1.03334948e-01 2.06127465e-01 -5.84504187e-01
4.26473439e-01 2.30147690e-01 4.21770751e-01 1.88758075e-01
-8.70916918e-02 1.23163491e-01 5.27769983e-01 3.48797351e-01
1.13242173e+00 3.95095795e-01 6.62351906e-01 -1.76333100e-01
6.35909200e-01 7.49563396e-01 2.16114044e-01 1.17341232e+00
-7.36301899e-01 7.70122826e-01 1.86070129e-01 -6.75901651e-01
-1.00334716e+00 -9.03729737e-01 -2.85855293e-01 1.19776285e+00
-3.19303349e-02 -2.78359860e-01 -6.44909978e-01 -6.57206416e-01
-3.16910446e-01 4.67315197e-01 -6.18960977e-01 1.56540141e-01
-6.68926716e-01 -7.22487390e-01 5.60256243e-01 4.81251180e-01
6.22374833e-01 -1.31402647e+00 -1.57367766e+00 3.27535599e-01
5.36037803e-01 -1.07144177e+00 -8.71894043e-03 3.58263642e-01
-9.38830972e-01 -1.33827257e+00 -4.74948198e-01 -6.47024155e-01
2.32028186e-01 7.85469770e-01 1.14799726e+00 2.91997254e-01
-1.17914867e+00 3.50893378e-01 -5.31353652e-01 -8.60969603e-01
-3.54871064e-01 -1.72077879e-01 -2.01372817e-01 -6.10921860e-01
5.81277788e-01 2.80244797e-01 -6.66168869e-01 2.20757410e-01
-9.49596643e-01 -2.40592510e-01 5.03642321e-01 2.18895882e-01
3.15300077e-01 1.25065520e-01 -3.07027340e-01 -6.65095389e-01
2.34018818e-01 -6.35512099e-02 -9.19063091e-01 1.13799170e-01
-5.36886692e-01 -9.05688405e-02 1.37855448e-02 -2.02718154e-01
-7.38042831e-01 6.37787819e-01 -2.68494666e-01 -2.10000366e-01
-3.69092375e-01 -3.08674127e-01 2.12326184e-01 -3.28301132e-01
7.85575628e-01 -8.77053514e-02 -6.12831302e-03 -5.59871376e-01
3.90169650e-01 6.75623834e-01 7.62662590e-01 4.44726460e-02
4.96378690e-01 5.59063792e-01 -7.51196221e-02 -1.00984693e+00
-5.81541598e-01 -1.06629860e+00 -9.04545486e-01 -2.65907019e-01
8.75823736e-01 -7.72394061e-01 -8.29724610e-01 4.85132545e-01
-1.22554898e+00 -2.11429894e-01 -1.60325933e-02 3.69063944e-01
-2.18414262e-01 3.49852562e-01 -4.60770428e-01 -7.01193750e-01
-3.99795711e-01 -1.04701507e+00 9.00171340e-01 5.30035794e-01
8.47045332e-02 -6.08707666e-01 2.12647766e-01 7.12828413e-02
4.27036285e-01 1.29914701e-01 1.36419505e-01 -2.65990168e-01
-5.96643150e-01 -3.81538987e-01 -6.24767900e-01 2.47715428e-01
-1.76125616e-01 6.20815814e-01 -1.22619545e+00 -8.80841687e-02
1.58839792e-01 2.64843404e-01 9.98567224e-01 5.21034002e-01
8.57702851e-01 2.25038335e-01 -6.07227683e-01 4.17214483e-01
1.53439438e+00 1.77238107e-01 6.48083687e-01 1.15469408e+00
4.14597660e-01 3.36010635e-01 8.69417012e-01 2.17615530e-01
1.52682975e-01 7.44585335e-01 4.33979154e-01 1.31198406e-01
-4.36982751e-01 4.14996892e-01 4.37915921e-01 -8.19612592e-02
1.17369920e-01 -1.80708617e-02 -6.77850544e-01 3.60770762e-01
-1.85016811e+00 -1.13852084e+00 -7.78659582e-01 1.76630998e+00
4.14665133e-01 5.96434236e-01 4.51444089e-01 4.64140475e-01
5.01112700e-01 -3.15170795e-01 -9.11400616e-02 -4.63991910e-01
2.19049692e-01 5.13166524e-02 7.10083663e-01 3.43390375e-01
-1.44794166e+00 1.00997341e+00 7.86948538e+00 5.67169130e-01
-1.46313679e+00 1.55236617e-01 5.22059388e-02 -7.16273412e-02
9.27842379e-01 1.00439288e-01 -1.34677470e+00 3.17494512e-01
9.56378877e-01 -1.52122006e-02 -1.47377372e-01 1.23044538e+00
1.95662871e-01 -6.67494595e-01 -9.76017833e-01 9.11236286e-01
-9.89778489e-02 -1.20097697e+00 -4.32907134e-01 -4.13169086e-01
-5.25585674e-02 2.61441112e-01 -3.62688988e-01 5.50834179e-01
2.19723955e-01 -9.04762208e-01 8.76838744e-01 3.42887789e-01
-1.01059070e-03 -4.16192979e-01 8.27246308e-01 4.17584419e-01
-1.30087006e+00 7.48099089e-02 -6.63867414e-01 -5.17174304e-01
-8.44593942e-02 2.35790357e-01 -1.59931672e+00 -1.85720157e-02
1.15027916e+00 2.40096867e-01 -1.29559493e+00 1.78128433e+00
-1.65068179e-01 4.15107131e-01 -6.46043181e-01 2.01600175e-02
1.68945894e-01 3.10194582e-01 6.95228517e-01 1.65255940e+00
4.29952424e-03 7.01697692e-02 3.06685120e-01 3.35470557e-01
7.03847826e-01 -2.25931313e-02 -5.67739189e-01 5.30418634e-01
1.09245479e-01 1.50731945e+00 -1.25558865e+00 -5.54978311e-01
-3.28811824e-01 9.05476451e-01 -4.22358997e-02 -1.62887976e-01
-5.40218115e-01 -5.85667193e-01 3.89024287e-01 1.92669064e-01
4.65505362e-01 -4.63782609e-01 -3.86709541e-01 -5.74770033e-01
-2.66216964e-01 -6.61730886e-01 7.78440177e-01 -1.00151813e+00
-8.21767569e-01 6.33366525e-01 2.52558261e-01 -1.04866862e+00
-2.03640476e-01 -1.14019883e+00 -8.86718214e-01 6.19885325e-01
-1.30547035e+00 -9.73834336e-01 -3.45580280e-01 3.81399810e-01
8.44645977e-01 -3.67118604e-02 4.54357833e-01 3.60592246e-01
-7.89489567e-01 2.57398039e-01 -2.37018287e-01 3.58925879e-01
8.12954307e-01 -1.38732684e+00 5.15894651e-01 1.33947492e+00
3.38046879e-01 4.01361614e-01 1.20869398e+00 -3.26519996e-01
-1.28591502e+00 -7.76026964e-01 4.16172832e-01 -7.32255340e-01
5.16373217e-01 1.50673568e-01 -8.12206805e-01 7.18792200e-01
4.16757703e-01 2.45509669e-01 1.62245616e-01 -4.33191717e-01
-5.78001328e-02 1.09707974e-01 -1.14073288e+00 2.48774245e-01
4.34147924e-01 3.54016535e-02 -1.00858164e+00 2.85788983e-01
5.57474732e-01 -5.71523070e-01 -2.80280888e-01 1.27811164e-01
4.01687264e-01 -1.24078560e+00 1.18507326e+00 -3.70908320e-01
-3.12171280e-01 -9.76732492e-01 2.18498893e-02 -6.33977115e-01
-3.97856295e-01 -2.64100462e-01 8.11503530e-02 9.32898045e-01
3.28907520e-02 -3.50046128e-01 8.22464406e-01 4.03006643e-01
1.11891285e-01 3.00232023e-02 -9.45416570e-01 -6.27970755e-01
-4.14074987e-01 -5.86502850e-01 -9.76748094e-02 2.61188418e-01
-4.95927393e-01 4.13243681e-01 3.27486433e-02 3.72776628e-01
8.31212223e-01 1.02683529e-01 1.03252804e+00 -1.13742995e+00
-2.64802307e-01 -6.58015966e-01 -8.57627690e-01 -5.23324490e-01
-6.97917223e-01 -4.47802246e-01 1.24608502e-01 -1.40468764e+00
1.15320787e-01 -2.37468377e-01 -1.23693198e-01 4.01046574e-01
-1.99560642e-01 9.76114333e-01 5.09233236e-01 1.96002364e-01
-7.21091270e-01 -2.03391448e-01 6.63869441e-01 1.00906733e-02
-1.73053280e-01 2.13435262e-01 -3.78715307e-01 9.19606388e-01
5.97494841e-01 -8.91099334e-01 2.02732518e-01 -2.13652283e-01
1.19281746e-01 -3.62512320e-01 5.66802621e-01 -1.54541850e+00
3.32076192e-01 -6.40523853e-03 6.48511469e-01 -1.05452514e+00
4.93408814e-02 -7.42477953e-01 -1.70774803e-01 8.87106895e-01
1.83638632e-01 2.73765534e-01 4.06839520e-01 3.85686636e-01
-1.47645315e-02 -7.35331953e-01 1.09366655e+00 -3.52806747e-01
-1.81438112e+00 -6.03357852e-02 -7.27444708e-01 -4.52346683e-01
1.50759923e+00 -5.60040355e-01 -1.84971243e-01 3.12737048e-01
-9.59466994e-01 2.68283427e-01 3.29364002e-01 5.92366695e-01
4.61580932e-01 -8.81731689e-01 -6.77934289e-01 1.29007041e-01
2.57461429e-01 -1.96021348e-01 -1.11496322e-01 7.64984488e-01
-1.37693906e+00 4.90351647e-01 -5.77480853e-01 -9.46471632e-01
-2.06504345e+00 6.60990179e-01 3.30508679e-01 3.81667651e-02
-1.12878656e+00 8.12751710e-01 -2.31878430e-01 9.17181820e-02
4.41423535e-01 -3.72847915e-01 -7.53176749e-01 5.96516654e-02
1.22022653e+00 5.99705994e-01 3.04846138e-01 -2.43936047e-01
-6.04161024e-01 5.25095582e-01 -2.20816538e-01 -8.44727457e-02
9.35727596e-01 -3.31553131e-01 7.23338500e-02 2.62157798e-01
6.97299302e-01 -3.34381700e-01 -1.15753531e+00 -1.38881262e-02
4.43236142e-01 -6.89546227e-01 4.09300357e-01 -4.46018219e-01
-6.69005513e-01 8.85725021e-01 1.39195216e+00 3.56827170e-01
7.83073366e-01 -1.81438789e-01 1.20319970e-01 6.59624100e-01
1.66045889e-01 -9.98933613e-01 -3.89271113e-03 5.96503019e-01
5.29505134e-01 -1.55275989e+00 6.67719364e-01 -3.53098065e-01
-4.60880578e-01 1.71189439e+00 7.11196482e-01 -5.23797035e-01
6.15575612e-01 3.10119510e-01 4.80966151e-01 -2.93659538e-01
-5.42322755e-01 -6.11598134e-01 2.36501873e-01 6.98097765e-01
3.93153071e-01 -6.11189343e-02 -2.45500013e-01 -1.38461277e-01
1.43000588e-01 5.79266436e-02 7.01088488e-01 1.13266528e+00
-1.07873178e+00 -1.07064545e+00 -1.10395229e+00 2.46295005e-01
-5.70158958e-01 2.47044086e-01 -2.64066070e-01 1.05133808e+00
2.58497119e-01 9.80147123e-01 6.90129027e-02 -8.92789289e-02
3.21364611e-01 -1.54425278e-01 4.34832543e-01 -8.60211313e-01
-6.61847770e-01 1.19284511e-01 -2.39944205e-01 -6.88706458e-01
-2.11600229e-01 -6.60377085e-01 -1.27875364e+00 -7.83959180e-02
-3.28640342e-01 -3.29920471e-01 9.01873648e-01 1.07393634e+00
1.11783765e-01 6.09511971e-01 5.75476997e-02 -1.22616661e+00
-3.80061895e-01 -1.13258135e+00 -2.94505656e-01 3.20653655e-02
3.68840933e-01 -4.54192936e-01 -1.07640259e-01 2.27044091e-01] | [8.541923522949219, -0.501251757144928] |
ab191fde-1f2f-42bd-b37a-1cf3732e52f4 | a-model-based-solution-to-the-offline-multi | 2305.17198 | null | https://arxiv.org/abs/2305.17198v1 | https://arxiv.org/pdf/2305.17198v1.pdf | A Model-Based Solution to the Offline Multi-Agent Reinforcement Learning Coordination Problem | Training multiple agents to coordinate is an important problem with applications in robotics, game theory, economics, and social sciences. However, most existing Multi-Agent Reinforcement Learning (MARL) methods are online and thus impractical for real-world applications in which collecting new interactions is costly or dangerous. While these algorithms should leverage offline data when available, doing so gives rise to the offline coordination problem. Specifically, we identify and formalize the strategy agreement (SA) and the strategy fine-tuning (SFT) challenges, two coordination issues at which current offline MARL algorithms fail. To address this setback, we propose a simple model-based approach that generates synthetic interaction data and enables agents to converge on a strategy while fine-tuning their policies accordingly. Our resulting method, Model-based Offline Multi-Agent Proximal Policy Optimization (MOMA-PPO), outperforms the prevalent learning methods in challenging offline multi-agent MuJoCo tasks even under severe partial observability and with learned world models. | ['Amy Zhang', 'Derek Nowrouzezahrai', 'Jakob Foerster', 'Paul Barde'] | 2023-05-26 | null | null | null | null | ['multi-agent-reinforcement-learning'] | ['methodology'] | [-2.98015475e-01 1.87171504e-01 -2.55277067e-01 3.37939650e-01
-6.62627697e-01 -6.59018934e-01 8.07102382e-01 4.42659497e-01
-6.68794096e-01 1.39611888e+00 -1.50485784e-01 -2.94347674e-01
-5.18211484e-01 -5.58096647e-01 -7.65768886e-01 -7.51237869e-01
-4.06648219e-01 9.31627512e-01 -2.24532932e-02 -5.67107022e-01
2.21003145e-01 1.13421038e-01 -1.22458351e+00 -5.04782796e-01
1.14316380e+00 5.49796045e-01 2.55475909e-01 7.79935896e-01
3.41837198e-01 1.16788125e+00 -6.31828666e-01 -7.53135905e-02
6.54802382e-01 -3.73953104e-01 -7.20352232e-01 1.46176994e-01
-3.75001252e-01 -5.48549116e-01 -7.62603879e-02 1.00450015e+00
5.80577135e-01 4.47701722e-01 2.34702319e-01 -1.68767822e+00
-2.63196260e-01 8.57622504e-01 -6.55543685e-01 7.34708644e-03
3.05147350e-01 4.33568567e-01 8.50939512e-01 -3.21996421e-01
6.78070664e-01 1.26796281e+00 4.78596717e-01 6.49138808e-01
-1.10263026e+00 -2.59265035e-01 3.51662070e-01 1.60719946e-01
-8.48968804e-01 -2.04692498e-01 5.98287404e-01 -3.41963351e-01
6.35892510e-01 -1.32997990e-01 8.40980351e-01 1.08583832e+00
1.08565539e-01 8.23172510e-01 1.43374264e+00 -3.52963716e-01
7.21833050e-01 -2.16882691e-01 -6.35607779e-01 6.01356626e-01
4.48909760e-01 4.42371756e-01 -4.14786220e-01 -3.34935695e-01
8.57294917e-01 -1.69304162e-01 1.19463727e-01 -6.40589893e-01
-1.39552701e+00 8.34904313e-01 1.41062826e-01 -8.70281756e-02
-8.83291245e-01 3.58023286e-01 4.02662247e-01 9.67593968e-01
2.40688071e-01 8.84811401e-01 -6.34480059e-01 -4.46089655e-01
-2.98968554e-01 8.07794750e-01 1.21466053e+00 8.43430996e-01
8.52675378e-01 2.02650547e-01 2.43327439e-01 6.10110223e-01
2.29169965e-01 2.97552019e-01 2.29288727e-01 -1.50934041e+00
4.32124317e-01 4.35173035e-01 8.70803058e-01 -7.22155690e-01
-5.86829484e-01 -2.84106553e-01 -4.49854285e-01 5.92185736e-01
6.93411410e-01 -8.98409367e-01 -2.90950507e-01 1.84855545e+00
8.17996204e-01 2.06379592e-01 3.51867825e-01 9.63667750e-01
5.23850247e-02 3.65065306e-01 -1.06191844e-01 -6.30275846e-01
8.35693896e-01 -1.23664629e+00 -6.09378636e-01 -3.70495975e-01
5.42081892e-01 -4.57098991e-01 7.97486424e-01 3.62536997e-01
-1.19398928e+00 2.93374211e-02 -8.95156682e-01 6.65563345e-01
-1.61690667e-01 -4.18896645e-01 8.73212457e-01 3.21978748e-01
-1.10279703e+00 6.81726515e-01 -9.59417880e-01 -4.42727476e-01
1.38180070e-02 6.41663969e-01 -1.40616953e-01 2.25255430e-01
-9.93411779e-01 1.13272202e+00 3.81377310e-01 -4.90282625e-02
-1.42071104e+00 -5.02856910e-01 -6.05469108e-01 -1.39475018e-01
1.29124939e+00 -6.76168263e-01 1.79651403e+00 -1.07225084e+00
-2.30031013e+00 1.96566328e-01 4.80792344e-01 -6.17494285e-01
8.26094568e-01 2.55723335e-02 1.63093209e-01 -7.81889260e-02
2.77010232e-01 4.51779634e-01 8.36429536e-01 -1.48443067e+00
-8.31928551e-01 -6.19918108e-02 7.76012182e-01 7.45005846e-01
9.71132983e-03 -2.82685161e-01 2.16229573e-01 -2.52952009e-01
-4.88241076e-01 -1.13137484e+00 -8.15679908e-01 -5.23907304e-01
6.03778346e-04 -4.19808745e-01 5.77838600e-01 -1.61718935e-01
6.52521074e-01 -1.56122494e+00 5.53206503e-01 2.15699852e-01
3.59537542e-01 1.88697800e-01 -3.79636258e-01 1.01967227e+00
6.07583642e-01 -1.45862222e-01 2.70534635e-01 -3.64813149e-01
4.01653141e-01 4.65057462e-01 5.82262035e-03 5.16339362e-01
-2.74571359e-01 8.57872188e-01 -1.44750607e+00 -3.22382271e-01
2.86419958e-01 -2.84485728e-01 -8.22993398e-01 4.00519252e-01
-6.52452111e-01 9.07347918e-01 -7.39150703e-01 5.60063958e-01
1.87649578e-01 -1.72800899e-01 6.16533995e-01 5.49736798e-01
-4.20502841e-01 -6.48922622e-02 -1.27872014e+00 1.49457026e+00
-6.17791235e-01 2.26010308e-01 6.32342100e-01 -1.10640681e+00
5.46375513e-01 3.98898542e-01 1.04869032e+00 -5.79889894e-01
4.01927799e-01 3.37071687e-01 3.56229335e-01 -4.64658648e-01
4.97606248e-01 8.48682150e-02 -1.63851380e-01 7.69376695e-01
4.77247909e-02 -2.07819417e-01 4.92331326e-01 1.88986242e-01
1.21826482e+00 3.36806029e-01 5.62988877e-01 -1.71213537e-01
2.63315976e-01 3.55913699e-01 7.79469907e-01 1.29652894e+00
-6.73087776e-01 -2.58737445e-01 7.14766979e-01 -4.68199968e-01
-1.08454859e+00 -5.81804931e-01 8.07528257e-01 1.32226193e+00
4.86656278e-01 -2.63042182e-01 -6.88603342e-01 -4.94284987e-01
1.44379690e-01 3.29226255e-01 -5.46541572e-01 4.16335240e-02
-6.75592422e-01 -5.41821659e-01 5.80116846e-02 -3.51013839e-02
4.46674436e-01 -1.17124414e+00 -8.96608412e-01 7.64451146e-01
-1.05247602e-01 -1.02319980e+00 -5.06120622e-01 -7.49370381e-02
-3.32546353e-01 -1.37566614e+00 -5.58394551e-01 -4.87389833e-01
5.18477321e-01 1.70450553e-01 9.45147634e-01 -2.50032488e-02
2.85367556e-02 8.67431343e-01 -4.55540389e-01 -4.98831242e-01
-6.12923384e-01 1.46451905e-01 5.39834440e-01 1.66579336e-02
-7.40568861e-02 -6.51546299e-01 -5.29396415e-01 2.49232292e-01
-5.83479464e-01 9.55256075e-02 5.38864434e-01 1.10977519e+00
2.90107399e-01 -5.96075729e-02 1.04066896e+00 -6.88774168e-01
1.26944172e+00 -5.88535905e-01 -1.16842365e+00 3.02065402e-01
-6.08356714e-01 -1.77250262e-02 9.87688124e-01 -7.62320876e-01
-7.84574270e-01 -5.92847243e-02 4.28067923e-01 -3.96648526e-01
1.03225328e-01 6.75916910e-01 3.28718573e-01 -3.65543604e-01
6.24841750e-01 7.41388649e-02 4.54496652e-01 3.42066884e-02
2.74624377e-01 3.83640885e-01 1.77158695e-02 -1.01496196e+00
7.37676084e-01 2.62024432e-01 6.08735383e-02 -5.94343662e-01
-6.58777952e-01 -2.58421391e-01 -2.25456938e-01 -4.78440195e-01
4.31261212e-01 -8.93955767e-01 -1.49594641e+00 4.66865391e-01
-9.35717821e-01 -9.80519235e-01 -3.34324121e-01 4.80016857e-01
-1.15618086e+00 1.60774156e-01 -5.26000381e-01 -1.03348136e+00
-1.46408137e-02 -1.30481875e+00 5.08612812e-01 4.70903397e-01
2.20137805e-01 -1.23131025e+00 5.23735225e-01 1.13423347e-01
6.49087548e-01 3.34034890e-01 4.11858857e-01 -5.41009724e-01
-6.28302872e-01 1.62523553e-01 3.66509169e-01 -1.90917283e-01
5.75664714e-02 -3.34913850e-01 -2.50798255e-01 -7.67172396e-01
-1.90890998e-01 -8.97677779e-01 -8.49140659e-02 4.41843867e-01
5.23267269e-01 -6.65320337e-01 -8.46267715e-02 1.01228140e-01
1.36222041e+00 2.44181290e-01 -1.23956636e-01 6.99077845e-01
2.82307386e-01 4.71868694e-01 8.29904914e-01 1.05801117e+00
9.45617676e-01 6.74083531e-01 7.66530216e-01 2.90908039e-01
4.15239573e-01 -3.41122776e-01 6.06874943e-01 7.06902146e-01
-2.94963449e-01 -1.06439933e-01 -8.43748450e-01 5.55463552e-01
-2.63553762e+00 -1.00174153e+00 4.68784571e-01 2.12202573e+00
8.07812810e-01 -1.70408174e-01 5.92490554e-01 -4.32399869e-01
5.96330345e-01 -6.01399317e-02 -9.72034752e-01 -3.13194394e-01
1.19825833e-01 -2.00166613e-01 6.53289795e-01 6.86546803e-01
-8.67620766e-01 1.11955869e+00 5.93572617e+00 5.15007913e-01
-8.21873248e-01 3.51555258e-01 2.91909903e-01 -4.29298766e-02
5.13973385e-02 2.55219728e-01 -3.36703449e-01 2.87372708e-01
7.57297814e-01 -4.44483876e-01 1.47874570e+00 8.87044191e-01
8.25999677e-01 -3.40249181e-01 -1.06512892e+00 9.45197046e-01
-3.81504983e-01 -1.36499751e+00 -4.25826639e-01 1.73935086e-01
1.12924135e+00 2.91910261e-01 -1.58227175e-01 6.69276834e-01
1.28031588e+00 -6.96806014e-01 7.46320367e-01 1.32859290e-01
1.88971221e-01 -7.77173758e-01 4.77252245e-01 8.47526193e-01
-9.71762836e-01 -5.58499157e-01 -2.74973780e-01 -5.57533324e-01
1.60490498e-01 -3.68699692e-02 -9.30948019e-01 4.86764312e-01
2.74662733e-01 4.57854867e-01 -3.07864770e-02 9.87572610e-01
-2.33236507e-01 2.76222527e-01 -3.90994191e-01 -4.54910278e-01
5.99275231e-01 -6.19085371e-01 7.92782903e-01 2.38505229e-01
6.42098561e-02 -4.19808254e-02 8.94712985e-01 7.08929658e-01
-5.73907755e-02 -1.13937542e-01 -6.60885632e-01 -2.50825107e-01
7.93166518e-01 1.31575108e+00 -5.94346166e-01 -9.71352309e-02
-3.06678802e-01 6.17960036e-01 7.20608950e-01 5.03485382e-01
-7.04420149e-01 -5.72448932e-02 8.55811477e-01 -4.11101639e-01
-8.99766386e-02 -6.04860246e-01 1.13013133e-01 -1.23166502e+00
-7.69751966e-02 -1.23311794e+00 1.50303245e-01 -1.73138440e-01
-1.23206890e+00 7.79899284e-02 -1.89004228e-01 -1.26413071e+00
-7.96852112e-01 -2.52625048e-01 -5.51080108e-01 1.32272020e-01
-1.58852720e+00 -1.19448566e+00 7.98121095e-02 5.77255428e-01
6.29323483e-01 -4.47481275e-01 6.54256046e-01 7.66549781e-02
-6.17127597e-01 2.17959017e-01 4.60359275e-01 -2.26236567e-01
4.50633079e-01 -1.47597408e+00 -1.27377167e-01 6.80289447e-01
-2.46907622e-01 2.60628402e-01 8.63133252e-01 -4.74953800e-01
-2.03240633e+00 -8.91088724e-01 1.30756214e-01 -1.45314142e-01
1.10904348e+00 -3.06665897e-01 -1.25473350e-01 6.86154962e-01
3.51321936e-01 -3.43581885e-02 1.73950285e-01 4.05679122e-02
2.37835482e-01 -6.47981046e-03 -9.32929158e-01 9.52971339e-01
8.19825590e-01 -1.18304193e-01 -1.25171944e-01 6.33439004e-01
6.46120906e-01 -5.66130877e-01 -8.72247159e-01 -1.41260609e-01
3.08839530e-01 -7.14217365e-01 6.08337343e-01 -7.91866302e-01
8.81915838e-02 -3.14676374e-01 1.32832006e-01 -2.06398559e+00
-7.86080584e-03 -1.62553954e+00 -2.31752723e-01 6.70119107e-01
2.27904096e-01 -9.29907620e-01 6.68896496e-01 4.04287577e-01
-3.88597436e-02 -5.44957519e-01 -1.00800145e+00 -9.47011411e-01
1.65140852e-01 3.94194461e-02 6.17660165e-01 9.84773397e-01
4.16985005e-01 1.91276059e-01 -7.30931222e-01 1.80526957e-01
9.26047981e-01 7.94618875e-02 1.29255772e+00 -8.86307001e-01
-6.74098730e-01 -5.16597152e-01 1.92378104e-01 -1.02242827e+00
3.86430353e-01 -3.46452713e-01 1.91491395e-01 -1.50875306e+00
-1.61489889e-01 -8.81896496e-01 -1.29891923e-02 4.16129529e-01
-4.15985845e-03 -4.54974145e-01 4.79427159e-01 2.98263699e-01
-1.24319029e+00 7.49194622e-01 1.47390974e+00 -3.55766118e-02
-5.17809927e-01 1.17022991e-01 -4.98179823e-01 5.48526525e-01
8.84014606e-01 -2.60282516e-01 -4.94095534e-01 -5.38871467e-01
4.81400102e-01 6.10298872e-01 2.07611740e-01 -7.94112265e-01
6.06619596e-01 -9.61561739e-01 -5.55267274e-01 1.41193777e-01
2.75999904e-01 -7.17409968e-01 2.85945870e-02 8.24726462e-01
-3.14211726e-01 3.06047022e-01 -3.42095405e-01 8.31680119e-01
-7.83276930e-03 -1.28602728e-01 5.60093164e-01 -4.63979274e-01
-6.76305294e-01 4.53667253e-01 -6.91470921e-01 3.06790859e-01
1.45794737e+00 2.33613059e-01 -4.31368321e-01 -8.36658418e-01
-6.41486049e-01 9.75254774e-01 3.71728987e-01 2.50020236e-01
3.12445939e-01 -1.00226355e+00 -7.89723516e-01 -1.80482239e-01
-1.80539206e-01 4.67595942e-02 3.26427929e-02 1.06815541e+00
-3.86267841e-01 8.76415521e-02 -2.71569759e-01 -1.79919168e-01
-5.63364267e-01 3.83311778e-01 6.66057348e-01 -6.13734961e-01
-3.40454936e-01 3.67156386e-01 -3.42376471e-01 -7.86936581e-01
2.10640341e-01 8.69880710e-03 -8.96841437e-02 -1.43253177e-01
2.53402919e-01 4.97745425e-01 -2.51918674e-01 -1.60805658e-01
7.64367133e-02 -9.08286124e-02 -6.36519678e-03 -3.68520290e-01
1.46813321e+00 -3.77255410e-01 -1.02888085e-01 2.60027736e-01
2.65570372e-01 -2.48511091e-01 -1.72556567e+00 -2.89232016e-01
2.46953562e-01 -3.07431370e-01 -3.11529666e-01 -6.52457714e-01
-6.58895433e-01 2.07657769e-01 1.49332747e-01 6.16256356e-01
6.03993833e-01 -2.90320903e-01 5.56644440e-01 8.84267271e-01
1.02070761e+00 -1.65145314e+00 3.26349258e-01 8.77768874e-01
5.50745368e-01 -1.49374163e+00 -2.05843166e-01 3.38106751e-02
-7.93500602e-01 8.83363664e-01 8.77226412e-01 -4.56934094e-01
4.59742516e-01 2.32447058e-01 1.06493667e-01 -2.01538987e-02
-1.15890718e+00 -4.68781590e-01 -7.06575096e-01 7.61841834e-01
-4.22014624e-01 2.95168310e-01 -3.97878110e-01 2.76470780e-01
-7.75562823e-02 -5.67288399e-02 9.39894021e-01 1.27525890e+00
-3.47114563e-01 -1.30894232e+00 -2.91289032e-01 3.94750446e-01
-1.31053239e-01 5.07032037e-01 -2.61472911e-01 8.05677235e-01
-2.28827536e-01 1.16252017e+00 -2.40869403e-01 -1.25327781e-01
2.87774783e-02 -4.92507637e-01 5.48814952e-01 -4.68560040e-01
-6.76243544e-01 2.18331963e-02 1.47708416e-01 -6.18820906e-01
-7.08658755e-01 -5.68653464e-01 -1.12035203e+00 -4.46972281e-01
-2.51488239e-01 3.61119539e-01 5.23268223e-01 1.30178952e+00
5.01582503e-01 4.01160538e-01 9.95678186e-01 -1.20934463e+00
-1.28008986e+00 -8.81858051e-01 -4.23210979e-01 2.42819056e-01
5.12005985e-01 -1.11422122e+00 -1.33911863e-01 -3.53961110e-01] | [3.7752108573913574, 2.0318379402160645] |
3c6d963f-3104-4ca8-89f5-1788fbab2684 | weakly-supervised-regional-and-temporal | 2204.00379 | null | https://arxiv.org/abs/2204.00379v1 | https://arxiv.org/pdf/2204.00379v1.pdf | Weakly Supervised Regional and Temporal Learning for Facial Action Unit Recognition | Automatic facial action unit (AU) recognition is a challenging task due to the scarcity of manual annotations. To alleviate this problem, a large amount of efforts has been dedicated to exploiting various weakly supervised methods which leverage numerous unlabeled data. However, many aspects with regard to some unique properties of AUs, such as the regional and relational characteristics, are not sufficiently explored in previous works. Motivated by this, we take the AU properties into consideration and propose two auxiliary AU related tasks to bridge the gap between limited annotations and the model performance in a self-supervised manner via the unlabeled data. Specifically, to enhance the discrimination of regional features with AU relation embedding, we design a task of RoI inpainting to recover the randomly cropped AU patches. Meanwhile, a single image based optical flow estimation task is proposed to leverage the dynamic change of facial muscles and encode the motion information into the global feature representation. Based on these two self-supervised auxiliary tasks, local features, mutual relation and motion cues of AUs are better captured in the backbone network. Furthermore, by incorporating semi-supervised learning, we propose an end-to-end trainable framework named weakly supervised regional and temporal learning (WSRTL) for AU recognition. Extensive experiments on BP4D and DISFA demonstrate the superiority of our method and new state-of-the-art performances are achieved. | ['ShiLiang Pu', 'Chunmao Wang', 'Qiang Li', 'Jingjing Wang', 'Jingwei Yan'] | 2022-04-01 | null | null | null | null | ['facial-action-unit-detection'] | ['computer-vision'] | [ 1.95209637e-01 1.27223834e-01 -6.42381489e-01 -3.04217845e-01
-5.71274936e-01 -1.45285517e-01 3.63714933e-01 -7.12833226e-01
-1.23371691e-01 6.71383381e-01 5.34295440e-01 4.19604957e-01
7.43438303e-02 -4.29789126e-01 -5.10344326e-01 -1.07347751e+00
1.68168500e-01 -1.92803532e-01 -9.80989542e-03 -1.71753302e-01
-1.27011389e-01 3.59302074e-01 -1.37965333e+00 2.94001959e-02
7.76488364e-01 1.38952100e+00 -4.52114493e-02 3.67766283e-02
1.49384975e-01 9.80476499e-01 5.31422757e-02 -5.66956699e-02
4.34856832e-01 -7.82174647e-01 -6.75665200e-01 5.32587051e-01
3.45076382e-01 -6.68942511e-01 -7.68826187e-01 8.42973650e-01
5.59624255e-01 1.26681983e-01 5.52216470e-01 -1.22751093e+00
-5.13052464e-01 1.05058521e-01 -8.23465228e-01 1.53696522e-01
2.22609773e-01 4.11791861e-01 1.05734134e+00 -1.06168294e+00
6.50618196e-01 1.04609656e+00 3.46076906e-01 7.52189636e-01
-1.09346902e+00 -7.72695601e-01 2.64115781e-01 2.06255093e-01
-1.33067584e+00 -7.63472736e-01 1.25214159e+00 -3.37427586e-01
4.24840003e-01 -1.66534670e-02 5.86708307e-01 1.25924993e+00
-3.95749658e-01 1.10658872e+00 1.09546602e+00 -2.00012490e-01
-1.14723027e-01 -1.00305006e-01 -2.29828954e-01 1.27771556e+00
-3.22708458e-01 2.38651037e-01 -7.08280146e-01 1.13223150e-01
1.01813650e+00 1.45870849e-01 -4.45609689e-01 -4.10031617e-01
-1.07031047e+00 6.29233599e-01 3.83215308e-01 2.68650562e-01
-3.04743916e-01 4.23939340e-02 5.04915833e-01 5.42739145e-02
8.22834551e-01 -1.47743508e-01 -3.44770759e-01 -2.38305792e-01
-7.66793132e-01 -2.22034842e-01 2.73757249e-01 7.28410840e-01
9.78732288e-01 2.65443921e-01 -2.64835596e-01 8.95764649e-01
3.89956206e-01 3.22115183e-01 5.06960034e-01 -9.76804137e-01
4.38029379e-01 8.09620559e-01 -1.93908781e-01 -1.16413927e+00
-2.80878991e-01 -3.94007653e-01 -9.52179193e-01 9.35218483e-03
3.43716621e-01 -1.60233602e-01 -7.31534481e-01 1.93597615e+00
4.79027569e-01 6.47495866e-01 -4.33430187e-02 1.15712571e+00
8.45902205e-01 4.43590552e-01 -2.24356465e-02 -5.85087180e-01
1.04455113e+00 -1.35118306e+00 -7.19870925e-01 -6.57402351e-02
7.80452013e-01 -5.59293389e-01 8.09073865e-01 7.46424645e-02
-9.53862250e-01 -7.04569459e-01 -7.56267965e-01 -6.25296906e-02
1.33006260e-01 4.92837369e-01 6.98274195e-01 2.52509892e-01
-6.69926584e-01 4.11032081e-01 -1.04272723e+00 -1.95316285e-01
8.42930496e-01 3.61969322e-01 -6.10876322e-01 -1.99059218e-01
-1.09169054e+00 3.85785550e-01 9.39094126e-02 6.62116349e-01
-1.01506233e+00 -3.93861651e-01 -1.04745615e+00 -2.44995430e-01
5.43825984e-01 -4.98902261e-01 8.00294995e-01 -1.20872653e+00
-1.70464671e+00 8.54909956e-01 -2.92893767e-01 -6.40825136e-03
5.19533515e-01 -2.34137811e-02 -2.14812845e-01 5.84796727e-01
1.06819682e-01 6.20135486e-01 9.95820463e-01 -1.02098286e+00
-5.54668307e-01 -5.86091876e-01 -5.84407058e-03 2.06080422e-01
-7.64003992e-01 -4.92763519e-02 -5.75470805e-01 -8.95162404e-01
2.81792842e-02 -9.74688530e-01 -4.68163155e-02 4.89891708e-01
-2.13790998e-01 -3.45858991e-01 9.90276158e-01 -5.95559180e-01
1.22856700e+00 -2.30525851e+00 4.50690210e-01 -4.45301011e-02
3.37487042e-01 3.89033139e-01 -3.65058511e-01 4.22489755e-02
-5.62815927e-02 -2.76685029e-01 -2.87923157e-01 -4.55623984e-01
-1.75583646e-01 3.17509055e-01 -1.55667722e-01 7.27824509e-01
5.89737117e-01 1.04250181e+00 -1.06336105e+00 -7.50777781e-01
2.25919470e-01 3.61458123e-01 -5.28879404e-01 4.97687638e-01
-4.99944016e-02 9.92805421e-01 -8.69810998e-01 9.51748550e-01
4.99112844e-01 -1.16337121e-01 -2.27132142e-01 -4.57591087e-01
7.25675598e-02 -1.33771211e-01 -8.78400981e-01 2.11218524e+00
-3.36853951e-01 2.98311055e-01 2.51254618e-01 -1.40196133e+00
8.71527612e-01 4.54780638e-01 1.04910004e+00 -5.66398621e-01
3.31389785e-01 3.09414774e-01 -1.58012152e-01 -8.17703068e-01
-1.80478513e-01 -1.53002813e-01 4.07655090e-01 5.77492237e-01
2.77908921e-01 3.85808945e-01 -5.60740428e-03 -1.07977748e-01
9.61909771e-01 8.01202655e-01 1.24547414e-01 8.36549848e-02
7.63547719e-01 -3.61692429e-01 9.89123344e-01 -6.91648200e-02
-7.99208462e-01 7.39229023e-01 4.54216361e-01 -3.84617120e-01
-5.89474082e-01 -6.79452419e-01 -1.67994365e-01 1.01544929e+00
2.30056867e-01 -3.05244029e-01 -6.41032875e-01 -1.13252378e+00
-1.68554172e-01 -2.43569180e-01 -1.01287568e+00 -4.38778013e-01
-6.92549109e-01 -6.98801577e-01 4.48705077e-01 7.12880671e-01
7.77640581e-01 -1.05662441e+00 -2.49228790e-01 1.48375958e-01
-3.37601960e-01 -1.43362081e+00 -7.71309614e-01 -2.12315992e-01
-8.64652634e-01 -1.04712594e+00 -9.67734277e-01 -9.14654136e-01
8.19807410e-01 5.05333006e-01 5.30948043e-01 8.89883265e-02
-3.64951372e-01 3.16597939e-01 -4.38977331e-01 2.11178809e-01
7.95203969e-02 -2.05390483e-01 8.57521296e-02 7.49822855e-01
3.00853491e-01 -8.14958632e-01 -1.04610765e+00 6.45844996e-01
-8.98790002e-01 7.88784325e-02 8.25610816e-01 1.06428611e+00
6.19413614e-01 -3.76533926e-01 6.03333175e-01 -5.68112433e-01
5.74672893e-02 -4.76843745e-01 7.50100939e-03 7.55564570e-02
-2.15203911e-01 3.82793732e-02 5.64977109e-01 -5.56973934e-01
-1.21595109e+00 3.63763183e-01 -5.26666678e-02 -8.38909686e-01
-1.03756264e-01 3.34444940e-01 -3.36276293e-01 -3.56657296e-01
3.07313025e-01 4.25078362e-01 5.14081299e-01 -2.45902345e-01
3.79540473e-01 6.98905110e-01 3.66147637e-01 -6.07149363e-01
8.64566207e-01 8.67247581e-01 -1.97169464e-02 -7.92957067e-01
-1.33891928e+00 -5.16529024e-01 -7.92435765e-01 -3.76258314e-01
9.15611923e-01 -1.12476337e+00 -4.90474194e-01 5.09764373e-01
-8.22267532e-01 -3.23169172e-01 -2.23365828e-01 5.92061162e-01
-7.24510849e-01 7.52051711e-01 -6.16931438e-01 -6.47569299e-01
-3.32894415e-01 -1.22160292e+00 1.23858345e+00 1.64683402e-01
-1.05310520e-02 -8.13274384e-01 5.08857556e-02 7.03078270e-01
1.02580436e-01 4.72129583e-01 5.06890774e-01 -2.41686583e-01
-5.74689031e-01 -7.88899809e-02 -5.18458128e-01 7.85478592e-01
4.94422138e-01 -1.93453327e-01 -9.99923766e-01 -2.28463769e-01
-1.36145316e-02 -8.62409055e-01 9.39547360e-01 1.12771288e-01
1.13285124e+00 -2.22163022e-01 -2.30837911e-01 7.02362061e-01
9.21083093e-01 -1.75497785e-01 3.93146247e-01 -6.43114075e-02
1.02502418e+00 8.46246898e-01 8.80644619e-01 5.99641681e-01
1.85789809e-01 7.00131536e-01 5.37693799e-01 -2.39705503e-01
-2.81952590e-01 -2.90011913e-01 6.00772977e-01 8.01469624e-01
-5.96874297e-01 3.63871723e-01 -3.08447063e-01 4.54302847e-01
-2.04227996e+00 -8.54708850e-01 3.25413704e-01 1.86638987e+00
9.22115982e-01 -9.45471078e-02 2.77046651e-01 -8.04032311e-02
5.17870247e-01 5.57836771e-01 -6.43407524e-01 3.93038422e-01
-1.72688708e-01 -2.34491415e-02 2.50511244e-03 7.07978681e-02
-1.05885112e+00 1.16598988e+00 4.85176659e+00 9.43060696e-01
-1.25444722e+00 1.83788687e-01 7.96600997e-01 -4.87908944e-02
-4.29376215e-02 -9.44193378e-02 -6.47418261e-01 5.08493006e-01
4.88456309e-01 3.46951455e-01 2.94430673e-01 6.46049738e-01
5.49756289e-01 2.58361936e-01 -9.99848187e-01 1.09579206e+00
3.82638812e-01 -9.99790907e-01 -8.34354758e-02 2.03249067e-01
8.46402884e-01 -2.60199249e-01 -2.68156789e-02 2.88958669e-01
-1.15025081e-01 -8.76996756e-01 3.68935049e-01 5.24709463e-01
9.02704716e-01 -6.10248268e-01 6.00936294e-01 1.34705603e-01
-1.35638678e+00 -7.17576221e-02 -4.05936956e-01 -1.34008855e-01
4.17386591e-02 1.61780357e-01 -3.32639724e-01 5.73436201e-01
5.72346449e-01 1.43166828e+00 -3.79440457e-01 5.89112759e-01
-3.59906286e-01 7.25584507e-01 -1.75870776e-01 3.17271471e-01
4.50582176e-01 -4.89305347e-01 4.11206901e-01 7.58833647e-01
-2.18927264e-02 2.39821374e-01 1.93266690e-01 8.83767188e-01
-1.12771809e-01 2.26106733e-01 -6.01923108e-01 -1.24113135e-01
-9.61556360e-02 1.44394433e+00 -3.62804681e-01 7.21446201e-02
-7.60437608e-01 1.16475117e+00 4.96500403e-01 4.50986743e-01
-7.65104115e-01 2.44258083e-02 6.77441716e-01 -9.74836051e-02
2.00576633e-01 -1.85728714e-01 2.29086339e-01 -1.49513412e+00
2.51522273e-01 -7.59562373e-01 3.49252015e-01 -4.93820131e-01
-1.24825132e+00 4.54598516e-01 -3.21194977e-01 -1.64007115e+00
-1.09919451e-01 -4.78947610e-01 -5.98822236e-01 4.42860156e-01
-1.79015064e+00 -1.57274687e+00 -6.24371111e-01 9.07892764e-01
6.70920372e-01 -1.95809573e-01 6.43183947e-01 4.03749317e-01
-1.06354952e+00 7.04434276e-01 -1.97380006e-01 5.25082290e-01
9.00350451e-01 -7.83351123e-01 -3.96568745e-01 8.45826685e-01
1.27323613e-01 4.04926866e-01 1.54319510e-01 -3.15027148e-01
-1.43303430e+00 -1.29525340e+00 3.40847552e-01 -1.52865916e-01
8.64713669e-01 -2.57972628e-01 -8.15534174e-01 5.94785035e-01
-2.69760918e-02 8.92352223e-01 7.77077079e-01 -3.35009933e-01
-3.37543994e-01 -4.95775968e-01 -7.21416295e-01 5.49707115e-01
1.50276232e+00 -5.42429209e-01 -3.00661594e-01 3.77373993e-01
4.40734565e-01 -1.51693299e-01 -9.24144089e-01 5.59310734e-01
5.45602500e-01 -8.71194005e-01 7.31892824e-01 -7.55562663e-01
7.02069521e-01 -1.95457369e-01 2.95990426e-03 -1.02557850e+00
2.39587501e-02 -9.00875330e-01 -3.76745403e-01 1.46211624e+00
-5.99075528e-03 -4.98162419e-01 1.12123382e+00 4.00435776e-01
-1.28370211e-01 -1.12314868e+00 -9.50039327e-01 -5.44995487e-01
-2.49996960e-01 -2.25398451e-01 3.92269045e-02 1.02681839e+00
-3.61120850e-02 3.59314799e-01 -7.50938833e-01 -5.76110855e-02
5.80661952e-01 3.81433368e-01 7.85086989e-01 -9.41078126e-01
-1.85528025e-01 -2.74792224e-01 -6.13223732e-01 -1.38752770e+00
5.80610633e-01 -8.54633987e-01 7.09639415e-02 -1.07337844e+00
2.66611725e-01 -3.93819392e-01 -4.24334943e-01 7.29576707e-01
-1.87070310e-01 6.48904622e-01 -5.97493500e-02 5.65221131e-01
-7.35733986e-01 1.32025111e+00 1.71706843e+00 -5.80260344e-02
-1.01699591e-01 -9.07491297e-02 -5.71947992e-01 8.03268015e-01
4.11547780e-01 -1.31084189e-01 -3.56208652e-01 -3.05179298e-01
-3.32131982e-01 2.29323998e-01 3.68099242e-01 -7.25174665e-01
8.51728022e-03 -2.17646763e-01 1.59781158e-01 -2.17455894e-01
4.26325113e-01 -8.51008952e-01 -6.37390256e-01 2.32910335e-01
-1.68791965e-01 -4.69413459e-01 -1.75376981e-01 9.49580431e-01
-6.35169923e-01 2.21423835e-01 8.48619998e-01 -1.38377873e-02
-7.96309114e-01 1.01841617e+00 9.31534395e-02 5.39873205e-02
1.22039413e+00 -3.51119220e-01 7.25356955e-03 -2.47376889e-01
-7.16219902e-01 3.43917698e-01 1.65527046e-01 4.74500090e-01
6.87637269e-01 -1.43596363e+00 -7.39235759e-01 3.07359040e-01
3.64999920e-01 1.80694029e-01 4.36766207e-01 1.41129041e+00
-2.97832459e-01 3.19406152e-01 -4.03090477e-01 -5.93998730e-01
-1.06046784e+00 3.87139291e-01 2.93868840e-01 -1.22969523e-01
-5.86057842e-01 7.55285203e-01 4.85444129e-01 -2.22458050e-01
2.50562370e-01 1.61465034e-01 -3.44849467e-01 1.27962291e-01
4.30459112e-01 2.00538784e-01 -2.28317425e-01 -1.02835703e+00
-1.68321759e-01 8.53707492e-01 -3.44276838e-02 3.13135594e-01
1.43601811e+00 -3.48895550e-01 -1.26857206e-01 9.28644463e-02
1.56657207e+00 -6.45547956e-02 -1.83672762e+00 -6.04911685e-01
-3.60674232e-01 -5.40998280e-01 1.43531203e-01 -1.61421627e-01
-1.63717842e+00 1.06461012e+00 5.22209227e-01 -3.86884123e-01
1.32024252e+00 -3.02786529e-02 1.01494038e+00 2.45557260e-02
2.13975549e-01 -8.79480660e-01 5.64848006e-01 -2.31982907e-03
7.51377702e-01 -1.58621955e+00 -2.44527504e-01 -6.87675476e-01
-6.69857442e-01 1.23550737e+00 9.40158367e-01 -1.77632123e-01
6.22483134e-01 -1.72336996e-01 1.42459303e-01 -7.04370514e-02
-4.30880964e-01 -4.70406353e-01 3.81147772e-01 4.31730688e-01
3.25875103e-01 -4.49679166e-01 -4.12134498e-01 5.68981290e-01
3.80662233e-01 2.24635154e-01 4.17260714e-02 9.42496896e-01
-1.37142569e-01 -9.14404988e-01 1.84727371e-01 1.83346942e-01
-4.58529741e-01 1.65649608e-01 -3.26899648e-01 7.70933747e-01
9.80181769e-02 7.69869447e-01 -1.98314652e-01 -4.49909776e-01
3.83581035e-02 -1.79596260e-01 4.69659925e-01 -5.72060108e-01
-7.00098053e-02 4.05219108e-01 -2.98704766e-02 -8.74904394e-01
-7.98414946e-01 -5.08117139e-01 -1.14662492e+00 1.59467638e-01
-2.18738109e-01 -6.63629770e-02 4.25027870e-02 1.16745520e+00
4.32857633e-01 1.88594073e-01 1.04557014e+00 -9.18658555e-01
-4.87293094e-01 -1.00327229e+00 -6.77102447e-01 6.35600984e-01
4.11322385e-01 -9.55114067e-01 -5.01363993e-01 8.45816582e-02] | [13.623042106628418, 1.5677108764648438] |
a9de318d-8be3-4621-95d5-65362f3955cf | sequencer-sequence-to-sequence-learning-for | 1901.01808 | null | https://arxiv.org/abs/1901.01808v3 | https://arxiv.org/pdf/1901.01808v3.pdf | SequenceR: Sequence-to-Sequence Learning for End-to-End Program Repair | This paper presents a novel end-to-end approach to program repair based on sequence-to-sequence learning. We devise, implement, and evaluate a system, called SequenceR, for fixing bugs based on sequence-to-sequence learning on source code. This approach uses the copy mechanism to overcome the unlimited vocabulary problem that occurs with big code. Our system is data-driven; we train it on 35,578 samples, carefully curated from commits to open-source repositories. We evaluate it on 4,711 independent real bug fixes, as well on the Defects4J benchmark used in program repair research. SequenceR is able to perfectly predict the fixed line for 950/4711 testing samples, and find correct patches for 14 bugs in Defects4J. It captures a wide range of repair operators without any domain-specific top-down design. | ['Martin Monperrus', 'Louis-Noël Pouchet', 'Zimin Chen', 'Steve Kommrusch', 'Denys Poshyvanyk', 'Michele Tufano'] | 2018-12-24 | null | null | null | null | ['program-repair', 'program-repair'] | ['computer-code', 'reasoning'] | [-1.78497970e-01 -1.53619170e-01 -2.64932126e-01 -2.81182110e-01
-1.38157988e+00 -8.10349762e-01 -4.08958405e-01 3.06176633e-01
2.82133579e-01 5.25451422e-01 2.02079322e-02 -7.34039903e-01
3.56218852e-02 -5.60711741e-01 -1.23259580e+00 8.56051818e-02
-3.49537522e-01 -9.86859873e-02 4.51101929e-01 -3.98100197e-01
9.13618743e-01 -4.55244869e-01 -1.56770432e+00 7.09514022e-01
9.92766500e-01 9.09774005e-02 3.64154667e-01 1.21524608e+00
3.48326899e-02 9.96421456e-01 -1.11222172e+00 -5.48341155e-01
-1.72140390e-01 -1.19662598e-01 -8.30472052e-01 -4.67410922e-01
6.52669013e-01 -8.77216458e-03 2.13128328e-01 1.18206716e+00
3.54022354e-01 -5.96620798e-01 2.34201446e-01 -8.49156559e-01
-9.05690551e-01 8.61441493e-01 -6.40957057e-01 3.19568008e-01
1.00560701e+00 3.42151463e-01 1.35576451e+00 -8.42040360e-01
8.67390573e-01 1.05761433e+00 1.27129221e+00 6.25435710e-01
-1.28344297e+00 -2.08707199e-01 -4.69369739e-01 -1.06700025e-01
-1.22761798e+00 -3.81487720e-02 1.12715781e-01 -1.11498606e+00
1.85988438e+00 2.26785034e-01 2.14879364e-01 1.21172440e+00
8.89811158e-01 3.35648268e-01 5.00397086e-01 -7.75856078e-01
2.90958166e-01 -3.63926381e-01 7.85718501e-01 1.30037451e+00
3.22824776e-01 1.47773132e-01 -1.80458948e-01 -1.02239883e+00
1.38520986e-01 -5.37266061e-02 -3.84382099e-01 3.68333869e-02
-7.51536131e-01 7.57712185e-01 1.16606876e-02 2.59024650e-01
1.76080033e-01 5.01623809e-01 7.65795469e-01 8.19871843e-01
2.58140743e-01 8.24286461e-01 -1.02558231e+00 -6.58314407e-01
-6.73827291e-01 4.80837107e-01 9.79122758e-01 1.25244689e+00
1.00059128e+00 3.45160216e-02 -2.46910751e-01 7.91501284e-01
4.25550163e-01 3.91816378e-01 6.36054933e-01 -6.78974867e-01
8.53192568e-01 9.19017851e-01 4.08347733e-02 -8.98153782e-01
-2.98468679e-01 -1.26273766e-01 5.36221340e-02 1.18504763e-01
6.72331154e-02 4.77903802e-03 -7.12607861e-01 1.30145419e+00
-6.58440366e-02 -6.61519647e-04 -3.49891692e-01 3.88926417e-01
7.24040091e-01 3.15546870e-01 -3.92194957e-01 2.18119547e-01
1.00933087e+00 -1.01754439e+00 -2.56524414e-01 -4.46513772e-01
1.50415504e+00 -8.19567025e-01 1.52585113e+00 6.41515315e-01
-7.93369830e-01 -4.86887842e-01 -1.12774193e+00 4.92581949e-02
-6.87623695e-02 3.27134997e-01 5.16474128e-01 8.67507577e-01
-1.31949914e+00 7.55712867e-01 -6.81982696e-01 -1.84003904e-01
1.69366226e-01 8.62234831e-02 -4.99935895e-01 -4.12321121e-01
-4.72713619e-01 5.77260733e-01 2.19553918e-01 -4.12316442e-01
-1.30720031e+00 -1.17169559e+00 -9.41655695e-01 9.30107161e-02
4.13152844e-01 -4.41359997e-01 1.60102069e+00 -5.49899876e-01
-7.55991161e-01 7.21337080e-01 -9.21818838e-02 -2.02261046e-01
-2.40406573e-01 -4.45591092e-01 -3.25534046e-01 -5.05613625e-01
3.39086860e-01 -1.74883693e-01 8.68995726e-01 -9.70578432e-01
-5.22377312e-01 5.10546472e-03 2.05741391e-01 -1.14188421e+00
-1.06771991e-01 4.56227124e-01 -7.03535378e-02 -6.38439775e-01
-5.26896954e-01 -8.71518850e-01 -6.50689900e-02 -6.73479736e-01
-5.84975779e-01 -1.65618539e-01 1.77623555e-01 -1.21804380e+00
1.80222940e+00 -2.20381999e+00 3.75175774e-01 6.38006702e-02
3.75685185e-01 2.61522949e-01 -7.39266396e-01 8.38859320e-01
-6.06517494e-01 4.10215825e-01 -4.65310961e-01 -1.08748026e-01
5.62354214e-02 -1.35328680e-01 -4.14392382e-01 3.70724678e-01
3.26483995e-01 8.81450891e-01 -1.00302589e+00 -3.52965482e-02
-4.85474050e-01 -2.06978008e-01 -1.35237980e+00 3.23668718e-01
-8.28422606e-01 -3.53999615e-01 -1.91835612e-01 1.13903427e+00
4.21309322e-01 -3.39151137e-02 -5.82689457e-02 7.02504456e-01
-6.33860826e-02 4.45194900e-01 -6.01754785e-01 2.07488561e+00
-5.99088132e-01 5.50624549e-01 -1.95973977e-01 -6.15546107e-01
1.16465306e+00 2.16721654e-01 -2.26885557e-01 -4.30967420e-01
-3.92024964e-01 5.79267323e-01 1.00689404e-01 -1.24234354e+00
3.37322086e-01 4.86509442e-01 -7.40026951e-01 4.25627440e-01
2.93862402e-01 4.66646962e-02 4.42999393e-01 1.60605595e-01
2.45551348e+00 1.36746481e-01 2.75555193e-01 -2.73407996e-01
2.75118530e-01 2.13650629e-01 7.01077938e-01 8.16558838e-01
3.09484839e-01 6.66226864e-01 1.04429853e+00 -5.78524530e-01
-1.09040225e+00 -7.59097815e-01 2.00008154e-01 1.13571763e+00
-7.44661152e-01 -1.24380624e+00 -1.02965403e+00 -1.02208745e+00
1.46390557e-01 7.30714023e-01 -5.75907469e-01 -6.05756044e-01
-5.78585505e-01 -3.66631776e-01 8.75792623e-01 3.52082074e-01
-4.63593215e-01 -1.11809993e+00 -3.63073766e-01 3.61213982e-01
-7.27853850e-02 -1.55550733e-01 -6.91743374e-01 1.93185642e-01
-4.56530601e-01 -1.57194376e+00 -5.38965911e-02 -8.97250235e-01
4.98119295e-01 1.65305100e-02 1.80184031e+00 9.69093680e-01
-9.38094676e-01 3.49135756e-01 -7.10710049e-01 -3.34789120e-02
-1.02655697e+00 2.57644057e-01 -3.81290376e-01 -8.69454503e-01
3.57266486e-01 -3.05183887e-01 3.68974298e-01 1.66057199e-01
-6.63261175e-01 -8.63586068e-01 4.12189007e-01 1.22274661e+00
1.78966671e-01 -8.95328075e-03 5.42458653e-01 -1.01540947e+00
7.93075621e-01 -8.16900253e-01 -1.02077699e+00 5.63977361e-01
-3.26259315e-01 -4.21221042e-03 6.54571891e-01 1.37266499e-04
-6.59098625e-01 -1.14627115e-01 -5.76279640e-01 -2.43106037e-01
1.55223925e-02 9.95767236e-01 6.17982931e-02 -2.60082662e-01
1.28936887e+00 -4.92639318e-02 -2.84829885e-01 -7.63503969e-01
2.54728645e-02 6.85613692e-01 5.14415562e-01 -8.61366212e-01
7.76904762e-01 -6.70221508e-01 -6.11995161e-01 -6.08871937e-01
-6.35993123e-01 -5.16121686e-01 -3.37749362e-01 2.07079366e-01
3.93485844e-01 -8.24638247e-01 -5.78282952e-01 3.79705518e-01
-1.47935617e+00 -5.94556868e-01 7.60327792e-03 -1.65865183e-01
-4.78330344e-01 3.80804449e-01 -8.73601735e-01 -4.19183582e-01
-3.51600438e-01 -1.37865651e+00 1.43461394e+00 -2.13244498e-01
-5.14213681e-01 -8.39886367e-01 6.22811913e-01 1.02449708e-01
3.30000609e-01 1.91375181e-01 1.67398906e+00 -3.11637074e-01
-7.70631075e-01 -4.51511502e-01 2.38114387e-01 4.23515201e-01
-5.12609631e-02 6.12355232e-01 -5.77187479e-01 -5.98354995e-01
-6.10670894e-02 -4.40212816e-01 7.96288848e-01 1.21835489e-02
1.13860965e+00 -2.35346168e-01 -3.41606170e-01 3.48348826e-01
1.79916811e+00 -1.85474232e-01 9.64158475e-01 3.56971741e-01
7.12928891e-01 3.22186381e-01 8.86765480e-01 5.74694097e-01
2.68401921e-01 4.30679440e-01 8.17748845e-01 6.00385308e-01
4.39174213e-02 -2.96995670e-01 8.74575317e-01 8.61328781e-01
3.66517007e-01 -1.96432337e-01 -1.50168753e+00 8.85618269e-01
-1.60655344e+00 -7.48115420e-01 -6.87317908e-01 2.13329697e+00
1.06763637e+00 1.73077002e-01 -2.72877477e-02 1.01480119e-01
4.72642362e-01 -3.65683645e-01 -5.61402887e-02 -5.91234922e-01
4.13791329e-01 4.96854782e-01 1.60998210e-01 4.03819263e-01
-7.37188160e-01 7.80931532e-01 6.89367342e+00 8.61262679e-01
-7.14916527e-01 2.85915762e-01 4.38250788e-03 1.62240565e-01
-5.27634859e-01 3.85542750e-01 -9.63364065e-01 5.19311845e-01
1.35766137e+00 -2.68069170e-02 5.54134667e-01 1.36399031e+00
-3.11629802e-01 -1.95306793e-01 -1.35855412e+00 4.78661001e-01
2.01450959e-02 -1.63980222e+00 -7.00453877e-01 -1.78728029e-01
9.58993673e-01 1.74724787e-01 -1.88863844e-01 7.39601254e-01
3.72528881e-01 -1.11284828e+00 7.98278451e-01 7.19691753e-01
7.30628014e-01 -7.34615624e-01 9.61050212e-01 6.34929121e-01
-7.20840096e-01 -5.59942245e-01 -7.53598869e-01 -1.42824560e-01
-4.07589614e-01 9.25785184e-01 -1.06228387e+00 3.81986439e-01
9.70012188e-01 9.88257527e-01 -1.29868460e+00 1.33184648e+00
-3.71950299e-01 9.65387106e-01 2.95996845e-01 -4.42142747e-02
-3.29271317e-01 4.15065438e-01 3.39887172e-01 1.39948928e+00
4.84591782e-01 -6.59610510e-01 1.15139388e-01 1.24252629e+00
-4.94074002e-02 -2.35011920e-01 -6.85406446e-01 -2.08106786e-01
5.82248628e-01 1.04933357e+00 -2.21947119e-01 -4.42217514e-02
-4.58221018e-01 4.89038646e-01 4.88432735e-01 5.93376579e-03
-1.01110804e+00 -7.33618021e-01 6.80722296e-01 8.26267898e-02
5.49184561e-01 -4.48282138e-02 -1.81763202e-01 -1.08667982e+00
4.95885849e-01 -1.71866989e+00 1.13554031e-01 -6.32301033e-01
-1.38240588e+00 3.74029547e-01 -2.08756343e-01 -9.74165022e-01
-2.53138036e-01 -5.99871576e-01 -1.02575433e+00 8.85078669e-01
-1.08753264e+00 -6.50849521e-01 -1.06935851e-01 3.93808931e-02
6.51656091e-01 -4.28878963e-01 8.51773918e-01 5.43861926e-01
-6.25634134e-01 8.91982436e-01 3.49072069e-02 -8.41590613e-02
7.95308888e-01 -1.62780297e+00 1.24166155e+00 8.84456515e-01
-9.62606817e-02 1.30877340e+00 5.97965419e-01 -1.03715920e+00
-1.85265040e+00 -1.36809719e+00 1.01709890e+00 -1.05021918e+00
9.20995951e-01 -6.81057632e-01 -1.25674868e+00 8.86196375e-01
-8.17349181e-03 -7.97075499e-03 5.98070264e-01 4.35570747e-01
-8.35364342e-01 2.88779467e-01 -1.06702399e+00 9.09032151e-02
1.02336025e+00 -8.08280051e-01 -6.71354949e-01 5.94331861e-01
1.15894616e+00 -7.10924685e-01 -1.10232091e+00 5.52178500e-03
-6.80696890e-02 -1.07598770e+00 7.72822022e-01 -6.50138617e-01
1.04057980e+00 -3.51691097e-01 -2.10731640e-01 -1.46945989e+00
-5.32739878e-01 -7.82701731e-01 -9.28378627e-02 1.43419802e+00
7.22037196e-01 -5.70539057e-01 3.85130316e-01 6.25893995e-02
-7.45874166e-01 -5.80859125e-01 -6.49542809e-01 -8.50768685e-01
2.52394322e-02 -3.72030258e-01 5.85517406e-01 8.05452645e-01
2.42181256e-01 6.36753440e-02 -2.96423465e-01 3.58986408e-01
2.04503298e-01 7.19046453e-04 9.05147254e-01 -9.62699771e-01
-1.01986229e+00 -6.13002665e-03 -5.16303658e-01 -4.95327055e-01
4.60812539e-01 -9.74880636e-01 4.60881740e-01 -9.35371697e-01
2.76588976e-01 -3.57645571e-01 2.75223434e-01 8.56694043e-01
-3.82776260e-01 -1.46427557e-01 -4.38838512e-01 -8.93333331e-02
-5.23432314e-01 5.38963601e-02 4.93424684e-01 -3.73116732e-01
2.56763279e-01 -8.92318636e-02 -5.67141414e-01 6.37107193e-01
5.43957114e-01 -1.00229132e+00 -1.08943857e-01 -7.22198308e-01
8.74046564e-01 3.93272370e-01 3.66179556e-01 -1.01950443e+00
-1.04378656e-01 5.20712249e-02 -3.66553545e-01 -3.49997789e-01
-5.23115218e-01 -2.74012744e-01 1.53803751e-01 7.32308865e-01
-3.23787034e-01 4.96886611e-01 3.78311425e-01 6.38873994e-01
-1.55311868e-01 -1.23368096e+00 3.88139963e-01 -2.64802396e-01
-8.85254085e-01 -2.04651862e-01 -3.43125761e-01 3.08348507e-01
9.37560856e-01 2.51069874e-01 -9.52200949e-01 6.03231013e-01
-1.60187840e-01 2.77091321e-02 9.82760549e-01 9.28608060e-01
5.14376163e-01 -9.53016222e-01 -6.32463276e-01 4.02236164e-01
6.77206635e-01 -4.14427400e-01 1.78340256e-01 4.26698238e-01
-8.76679003e-01 2.67709494e-01 1.24322966e-01 -6.94901884e-01
-1.35060489e+00 8.51532638e-01 1.82384923e-01 -1.97559193e-01
-5.02256155e-01 1.21897984e+00 -4.61480856e-01 -9.31969345e-01
-1.78197008e-02 -7.24947274e-01 3.94849539e-01 -5.57961404e-01
7.54614294e-01 2.58175045e-01 6.49367273e-01 1.70059144e-01
-4.55256492e-01 4.00087833e-01 -1.99448079e-01 6.46427512e-01
1.55013919e+00 6.10505879e-01 -8.04651499e-01 5.55741310e-01
1.27637839e+00 2.00687096e-01 -6.76771939e-01 2.36831561e-01
8.45689654e-01 -8.80146027e-01 -3.82747829e-01 -9.12845969e-01
-7.53475964e-01 7.77794182e-01 3.47416401e-01 2.29113594e-01
6.12252116e-01 1.04974946e-02 6.35233700e-01 6.83860958e-01
9.53095019e-01 -6.49809837e-01 4.02188599e-01 8.81414771e-01
8.96030545e-01 -1.05162537e+00 -3.98717791e-01 -2.43067503e-01
1.44774526e-01 1.16527629e+00 9.23725426e-01 -4.23346072e-01
1.51143864e-01 7.40128517e-01 -5.00703037e-01 -3.41138572e-01
-1.34328902e+00 5.84265411e-01 -1.13530517e-01 8.34097385e-01
7.24861860e-01 -1.40192866e-01 -1.54215619e-01 6.51460528e-01
-1.48215190e-01 4.25120853e-02 1.18638098e+00 1.42291248e+00
-6.34670079e-01 -1.40367019e+00 -6.36853039e-01 5.34779668e-01
-4.50770199e-01 -4.32597488e-01 -2.56887972e-01 3.91755074e-01
1.76315546e-01 8.65231752e-01 -4.42491382e-01 -6.12531245e-01
4.74405557e-01 8.16095546e-02 5.40106893e-01 -1.46154070e+00
-1.05367637e+00 -3.44953835e-01 3.25989038e-01 -8.24940503e-01
5.37371933e-01 -6.24301493e-01 -9.94029641e-01 -2.41817728e-01
-5.51873386e-01 1.94922127e-02 4.77064341e-01 6.34520352e-01
7.73706019e-01 8.97605240e-01 4.58701789e-01 -3.48118752e-01
-9.94263530e-01 -9.06400144e-01 -3.78669024e-01 7.79681429e-02
6.74967706e-01 -4.40058112e-01 -3.49036247e-01 2.47905582e-01] | [7.628413677215576, 7.717782497406006] |
72ee643d-9a09-47b6-9267-cef90e517f47 | continual-learning-in-open-vocabulary | 2307.01430 | null | https://arxiv.org/abs/2307.01430v1 | https://arxiv.org/pdf/2307.01430v1.pdf | Continual Learning in Open-vocabulary Classification with Complementary Memory Systems | We introduce a method for flexible continual learning in open-vocabulary image classification, drawing inspiration from the complementary learning systems observed in human cognition. We propose a "tree probe" method, an adaption of lazy learning principles, which enables fast learning from new examples with competitive accuracy to batch-trained linear models. Further, we propose a method to combine predictions from a CLIP zero-shot model and the exemplar-based model, using the zero-shot estimated probability that a sample's class is within any of the exemplar classes. We test in data incremental, class incremental, and task incremental settings, as well as ability to perform flexible inference on varying subsets of zero-shot and learned categories. Our proposed method achieves a good balance of learning speed, target task effectiveness, and zero-shot effectiveness. | ['Derek Hoiem', 'Yao Xiao', 'Weijie Lyu', 'Zhen Zhu'] | 2023-07-04 | null | null | null | null | ['continual-learning'] | ['methodology'] | [ 3.75582665e-01 1.05483569e-01 -4.21887815e-01 -6.45841360e-01
-8.78283858e-01 -3.39034528e-01 6.40911937e-01 3.06332409e-01
-5.40404499e-01 8.00189972e-01 -4.50653881e-02 -1.75755844e-02
-3.87575626e-01 -6.17032945e-01 -7.87682176e-01 -5.10090470e-01
-2.19955817e-01 6.40040100e-01 6.39084399e-01 8.56868178e-02
2.42639318e-01 1.92601576e-01 -1.99962139e+00 5.00800252e-01
6.69754028e-01 9.72006023e-01 2.86784321e-01 7.90451407e-01
-1.21546194e-01 9.14108992e-01 -3.42124581e-01 -4.24781561e-01
2.36490190e-01 -1.24479413e-01 -8.49758089e-01 1.12240613e-01
7.05598950e-01 -3.82629752e-01 -1.68729469e-01 7.84196436e-01
2.87084103e-01 4.85252321e-01 8.34668696e-01 -1.14254117e+00
-8.87474656e-01 5.43349743e-01 -1.94398224e-01 4.48097289e-01
1.52241692e-01 3.09654623e-01 9.82493103e-01 -1.08307827e+00
8.03291261e-01 1.17064476e+00 6.17217660e-01 7.11598337e-01
-1.30091596e+00 -5.68644702e-01 4.49515432e-01 4.04216498e-01
-1.31581974e+00 -6.07011318e-01 4.33777332e-01 -6.58302605e-01
9.40748394e-01 9.79841053e-02 6.90356612e-01 1.04028857e+00
1.63068444e-01 9.25712287e-01 9.33855474e-01 -7.12153137e-01
4.91256058e-01 4.42699462e-01 6.19002402e-01 7.36889720e-01
-5.58823794e-02 -2.14304030e-02 -6.45662129e-01 -9.23224464e-02
3.76172990e-01 4.47824985e-01 -9.04052407e-02 -7.12176025e-01
-9.17649329e-01 8.38204503e-01 4.53165323e-01 1.79476775e-02
-1.70560345e-01 1.42652869e-01 3.91293138e-01 3.85638416e-01
7.23423064e-01 2.89911896e-01 -6.17985785e-01 8.16875622e-02
-9.15253043e-01 1.23518765e-01 6.73532426e-01 1.05543375e+00
8.16665351e-01 -6.19409606e-02 -5.39400578e-01 9.56370413e-01
-3.19992974e-02 3.40630531e-01 9.07366216e-01 -8.61547649e-01
-2.59219468e-01 3.57944608e-01 -4.75023910e-02 -1.40475065e-01
-2.04228118e-01 -5.40382922e-01 -2.98016787e-01 2.54433692e-01
1.53835967e-01 1.86168730e-01 -1.12172520e+00 1.73948586e+00
3.89004201e-01 3.25319946e-01 -8.92873295e-03 2.69477904e-01
4.49547797e-01 3.87092799e-01 3.77346724e-01 -6.80639446e-01
1.03982103e+00 -1.10045946e+00 -3.55248988e-01 -9.19094309e-02
6.64245903e-01 -2.26544008e-01 1.48891187e+00 4.20788676e-01
-8.05572510e-01 -7.97573268e-01 -9.11223054e-01 -5.20127602e-02
-4.75021094e-01 -3.97680759e-01 5.96885502e-01 4.23015058e-01
-8.37529957e-01 6.90383375e-01 -5.58839142e-01 -2.48112187e-01
7.80990124e-01 1.16388619e-01 -2.28798147e-02 -3.01284969e-01
-7.06717312e-01 8.12200487e-01 4.83736902e-01 -7.97396004e-01
-1.30948412e+00 -8.48603308e-01 -5.58557630e-01 2.23536536e-01
6.17356479e-01 -6.53327703e-01 1.59606636e+00 -1.03247547e+00
-1.33560514e+00 9.23131049e-01 -2.02575594e-01 -8.15736711e-01
3.60911459e-01 -3.48754317e-01 -9.50302109e-02 -4.90223505e-02
-6.50497004e-02 9.39326704e-01 1.16939294e+00 -9.16829050e-01
-7.75835812e-01 -4.14260954e-01 1.58610031e-01 2.57858932e-01
-5.45317292e-01 -4.01178271e-01 2.12783888e-02 -4.57429469e-01
-2.56779283e-01 -8.22054386e-01 -1.01027705e-01 1.98423877e-01
1.25187010e-01 -7.16904223e-01 7.56808579e-01 -1.99123308e-01
1.15318906e+00 -2.10997939e+00 -5.89420721e-02 -2.21624523e-01
2.74008185e-01 3.22956622e-01 -1.25432149e-01 -2.71223206e-02
1.00453645e-01 -1.44243374e-01 7.02805966e-02 -2.63632804e-01
-9.84017551e-02 2.36069188e-01 -6.17526412e-01 -1.04970194e-01
-1.00300625e-01 9.89481628e-01 -1.17758262e+00 -4.42102641e-01
1.46079019e-01 1.17001794e-01 -6.19435072e-01 1.90896332e-01
-4.97324198e-01 -8.74675438e-02 9.79922712e-03 4.44136918e-01
1.85240552e-01 -5.10181487e-01 7.91864172e-02 5.09002320e-02
2.51625538e-01 2.02557873e-02 -8.15541208e-01 1.50744319e+00
-5.05318463e-01 4.48089689e-01 -7.50752330e-01 -9.26542878e-01
7.87198365e-01 2.00143889e-01 -6.56509399e-02 -5.06857693e-01
-1.84535384e-01 -2.53425781e-02 8.23025778e-02 -4.59278077e-01
3.18669677e-01 -3.70775491e-01 4.06920575e-02 4.25685912e-01
8.05948317e-01 -2.03477055e-01 1.77694380e-01 2.57494092e-01
9.38030899e-01 2.59664416e-01 7.98280656e-01 -2.24086329e-01
7.13069588e-02 -1.54023886e-01 2.09791586e-01 1.16671324e+00
-3.07864070e-01 2.72996902e-01 1.75197437e-01 -8.09037566e-01
-9.49758291e-01 -1.30369782e+00 -3.34731579e-01 1.97416532e+00
-9.77324545e-02 -3.77620429e-01 -6.31490469e-01 -8.98650408e-01
-4.75989655e-02 1.04883492e+00 -8.97476494e-01 -4.48729962e-01
-9.52728167e-02 -3.04280043e-01 -3.82472165e-02 4.56554264e-01
1.12792656e-01 -1.07814837e+00 -8.24917674e-01 1.07705556e-02
1.66360050e-01 -7.25946665e-01 -4.21378672e-01 4.73906070e-01
-9.11210656e-01 -9.12267447e-01 -4.76119995e-01 -7.30279267e-01
5.02908766e-01 5.32673717e-01 1.10268831e+00 -1.84264325e-04
-5.60104549e-01 6.92857862e-01 -3.90009284e-01 -6.20116651e-01
-2.25743845e-01 1.72092989e-02 3.92447561e-01 -2.88185850e-02
6.17008448e-01 -3.47582906e-01 -4.46471244e-01 1.26830563e-01
-6.85669661e-01 1.85664594e-01 3.89563411e-01 9.48056579e-01
7.73953140e-01 -2.24460110e-01 8.64737928e-01 -1.10876465e+00
5.29761314e-01 -6.53068185e-01 -3.40111524e-01 7.21402705e-01
-9.28411305e-01 1.87528282e-01 6.27460659e-01 -1.01315570e+00
-1.00468552e+00 6.70438707e-02 2.62476802e-01 -7.48117089e-01
-9.25780535e-02 1.88545972e-01 2.53785700e-01 -3.05284411e-02
1.13228440e+00 4.79195386e-01 -1.40231535e-01 -2.48425230e-01
8.32081020e-01 7.42778838e-01 4.24153686e-01 -5.60646296e-01
3.82888466e-01 3.20154697e-01 -4.72451687e-01 -8.63167107e-01
-1.36657429e+00 -4.84578639e-01 -9.42401767e-01 -3.14257532e-01
5.60091615e-01 -9.44022417e-01 -4.50080812e-01 3.36231917e-01
-8.03524137e-01 -7.71584392e-01 -9.29595590e-01 2.79346555e-01
-7.70056784e-01 5.74704707e-02 -3.00776660e-01 -8.76897812e-01
-3.64567280e-01 -8.94634247e-01 9.77721453e-01 3.45154613e-01
-3.46708924e-01 -9.64740157e-01 4.32799663e-03 1.26875013e-01
3.91708106e-01 -4.21691418e-01 9.10652280e-01 -1.20427513e+00
-6.15367949e-01 -3.48437697e-01 4.31400239e-02 3.80080134e-01
-4.04587090e-02 -3.38016033e-01 -1.19817042e+00 -4.35829163e-01
-4.87437062e-02 -1.00453079e+00 9.86308992e-01 3.68601114e-01
1.40254879e+00 -4.09504563e-01 -4.58463550e-01 3.45101058e-01
1.35001338e+00 1.01643071e-01 3.10357630e-01 -4.97495010e-02
2.99971849e-01 2.67783642e-01 6.44988954e-01 6.17244363e-01
2.44029760e-01 6.64836109e-01 -3.43664624e-02 6.92129433e-01
-3.45534861e-01 -4.14132178e-01 1.47436112e-01 4.54400271e-01
1.28013849e-01 -1.31277099e-01 -7.93887615e-01 6.16029680e-01
-1.75372076e+00 -1.17698014e+00 5.29723883e-01 2.50739646e+00
9.49776113e-01 5.54014564e-01 1.24791242e-01 -1.65095657e-01
4.38990772e-01 -9.57910493e-02 -9.47295606e-01 -2.34723508e-01
3.09826702e-01 3.80450070e-01 6.81432709e-02 6.22019768e-01
-1.04185176e+00 1.13672471e+00 7.65735769e+00 1.18196344e+00
-8.31049204e-01 4.49618220e-01 6.93084896e-01 -4.89452392e-01
-1.78529263e-01 -8.67994502e-02 -1.13935089e+00 3.03830117e-01
1.15894008e+00 -4.94548500e-01 4.68888015e-01 1.29267526e+00
-4.17256117e-01 -6.74242387e-03 -1.43085349e+00 7.55277574e-01
4.51065809e-01 -1.42451191e+00 3.68448466e-01 -1.88229606e-01
8.12281549e-01 1.27167419e-01 1.41705662e-01 9.76916134e-01
5.60810924e-01 -6.87846720e-01 6.79863691e-01 6.94311142e-01
9.75816190e-01 -4.21013385e-01 1.07509024e-01 7.94532359e-01
-8.28567982e-01 -6.38913453e-01 -4.90113646e-01 -2.56327540e-01
-2.02518851e-01 2.24725619e-01 -1.04432368e+00 -1.62738815e-01
6.31483555e-01 5.79610169e-01 -6.66382492e-01 1.05568111e+00
-1.57833606e-01 6.14683747e-01 -8.34968463e-02 -2.02660307e-01
-2.99232081e-02 3.99262875e-01 3.08454037e-01 1.13855958e+00
6.67859465e-02 2.30475828e-01 4.22202885e-01 3.97917032e-01
-1.99151471e-01 1.28236994e-01 -7.22939610e-01 3.72833908e-01
8.30997646e-01 1.12542379e+00 -6.70696378e-01 -9.92164314e-01
-3.11204582e-01 9.63877738e-01 9.16235566e-01 1.86161458e-01
-5.70809484e-01 -1.05392054e-01 3.13653439e-01 1.87361717e-01
4.85271305e-01 9.26927999e-02 -7.07717091e-02 -1.19551349e+00
-2.31826276e-01 -5.50569296e-01 5.80225706e-01 -9.79177594e-01
-1.39146209e+00 4.61314946e-01 2.78933853e-01 -1.15572560e+00
-3.79530370e-01 -5.52853346e-01 -5.26755035e-01 2.56728053e-01
-1.23671079e+00 -1.01190829e+00 -2.41441324e-01 5.47234118e-01
1.20760179e+00 -4.26343620e-01 1.14549744e+00 -1.11047991e-01
-1.83268458e-01 8.47044528e-01 2.66786426e-01 -3.22842032e-01
6.55870914e-01 -1.11011815e+00 2.28079900e-01 3.44651103e-01
4.61377680e-01 5.89855790e-01 5.74515939e-01 -5.37481964e-01
-7.47806072e-01 -1.15151095e+00 5.39620399e-01 -7.55090773e-01
5.97146034e-01 -3.74437451e-01 -1.13077462e+00 1.03939784e+00
-1.23317696e-01 3.97962302e-01 8.77057433e-01 4.33509320e-01
-7.99361169e-01 -2.05572873e-01 -1.18013847e+00 5.64189970e-01
1.30185604e+00 -5.59601665e-01 -9.81061995e-01 6.45509005e-01
1.14690316e+00 -1.33888125e-01 -6.62905812e-01 2.76755095e-01
8.96003306e-01 -8.04599583e-01 9.84480083e-01 -9.81018603e-01
2.46112317e-01 2.33839720e-01 -2.57811785e-01 -1.24808848e+00
-6.71195328e-01 -2.12428972e-01 -6.97227061e-01 7.93536365e-01
2.48078227e-01 -3.02221060e-01 5.62801003e-01 5.38202524e-01
-1.28831223e-01 -9.41321731e-01 -9.44416940e-01 -9.75376725e-01
1.05487509e-02 -4.02176678e-01 2.17677489e-01 7.74861217e-01
1.92663610e-01 7.46078372e-01 -4.96453494e-01 -3.34878445e-01
7.04088867e-01 1.09791152e-01 7.19578564e-01 -1.50667167e+00
-5.42212188e-01 -2.16047898e-01 -4.43992674e-01 -1.06057155e+00
7.28202462e-02 -1.03217268e+00 7.48124868e-02 -1.06395590e+00
6.85003221e-01 -3.53667200e-01 -7.36145139e-01 5.70207715e-01
-3.70653689e-01 1.93395898e-01 2.43199900e-01 4.92380232e-01
-1.35818923e+00 3.78524691e-01 9.28895414e-01 -8.41701552e-02
-3.11844498e-01 5.96104935e-02 -5.62871099e-01 8.02374005e-01
4.39675450e-01 -4.92709637e-01 -7.20759928e-01 -1.55688599e-01
-2.87375487e-02 -1.64874271e-01 2.92346656e-01 -1.21338058e+00
3.23674232e-01 -1.82265028e-01 3.95739704e-01 -2.44054660e-01
3.47127974e-01 -5.07424295e-01 -2.91632682e-01 5.33843696e-01
-9.29598331e-01 -6.46032155e-01 -7.19470670e-04 1.10130966e+00
3.48521739e-01 -3.82402599e-01 1.04684472e+00 -3.98075461e-01
-1.15261531e+00 4.34088260e-01 -1.49042502e-01 1.67596146e-01
1.29225290e+00 -2.76330829e-01 -3.85801464e-01 -6.98090494e-02
-1.15959918e+00 2.08270952e-01 2.87296236e-01 5.83216488e-01
7.22750187e-01 -1.22830033e+00 -5.20451188e-01 2.09369361e-01
6.80722535e-01 -2.51998901e-01 4.22057271e-01 5.02593338e-01
8.94116238e-02 2.63108492e-01 -1.38394907e-01 -7.15972066e-01
-1.13435161e+00 1.09710097e+00 2.11843669e-01 -2.00027436e-01
-5.87095201e-01 1.09487987e+00 3.52648854e-01 -4.67613161e-01
4.79438782e-01 6.00841045e-02 -4.30896282e-02 5.44015244e-02
9.86502707e-01 2.46133611e-01 -1.92521676e-01 -2.26395894e-02
1.04482882e-02 2.91188896e-01 -4.49076056e-01 2.14821100e-02
1.17568684e+00 -2.28754580e-01 3.94798756e-01 1.33693051e+00
9.60690498e-01 -5.48517108e-01 -1.47566378e+00 -8.66418540e-01
-1.50678465e-02 -4.64746565e-01 -1.89854115e-01 -9.77306545e-01
-1.51017621e-01 8.67017508e-01 9.88762140e-01 2.81367730e-02
6.83716178e-01 1.92901403e-01 5.39116621e-01 8.29530835e-01
6.43087745e-01 -1.15372443e+00 5.98767042e-01 4.93474007e-01
5.81424952e-01 -1.40098083e+00 4.36427891e-02 8.03923309e-02
-5.93598127e-01 9.24313009e-01 7.64473379e-01 -1.60618380e-01
9.01426375e-01 -8.29580277e-02 -3.07410657e-01 -3.32206264e-02
-1.42465127e+00 -2.37461388e-01 3.51657391e-01 5.87485492e-01
2.36193731e-01 1.84558287e-01 -4.08784077e-02 5.08696973e-01
5.23689687e-02 5.17432988e-01 2.30574459e-01 7.86164701e-01
-1.22770429e+00 -5.10206759e-01 8.53611231e-02 8.85936141e-01
-5.53465225e-02 -2.93364704e-01 1.01096161e-01 5.19417822e-01
3.38820368e-01 4.30034429e-01 2.76733220e-01 -3.61965448e-01
-4.13927399e-02 8.10357749e-01 7.78364122e-01 -1.12646747e+00
6.30022660e-02 -3.19160372e-01 -2.91038692e-01 -2.84686506e-01
-1.40286520e-01 -5.19296050e-01 -9.18779135e-01 2.08374053e-01
-5.42552292e-01 -1.26166707e-02 3.25966358e-01 1.20896757e+00
3.20462227e-01 3.65506351e-01 5.71949661e-01 -7.94915378e-01
-9.04317200e-01 -1.17342365e+00 -4.69947755e-01 5.34788251e-01
2.75061965e-01 -7.92623043e-01 -3.33628654e-01 3.40989918e-01] | [9.911911010742188, 3.145718812942505] |
705e5b98-04ce-445a-929d-f645e533ed13 | a-new-generation-of-perspective-api-efficient | 2202.11176 | null | https://arxiv.org/abs/2202.11176v1 | https://arxiv.org/pdf/2202.11176v1.pdf | A New Generation of Perspective API: Efficient Multilingual Character-level Transformers | On the world wide web, toxic content detectors are a crucial line of defense against potentially hateful and offensive messages. As such, building highly effective classifiers that enable a safer internet is an important research area. Moreover, the web is a highly multilingual, cross-cultural community that develops its own lingo over time. As such, it is crucial to develop models that are effective across a diverse range of languages, usages, and styles. In this paper, we present the fundamentals behind the next version of the Perspective API from Google Jigsaw. At the heart of the approach is a single multilingual token-free Charformer model that is applicable across a range of languages, domains, and tasks. We demonstrate that by forgoing static vocabularies, we gain flexibility across a variety of settings. We additionally outline the techniques employed to make such a byte-level model efficient and feasible for productionization. Through extensive experiments on multilingual toxic comment classification benchmarks derived from real API traffic and evaluation on an array of code-switching, covert toxicity, emoji-based hate, human-readable obfuscation, distribution shift, and bias evaluation settings, we show that our proposed approach outperforms strong baselines. Finally, we present our findings from deploying this system in production. | ['Lucy Vasserman', 'Donald Metzler', 'Jai Gupta', 'Jeffrey Sorensen', 'Yi Tay', 'Vinh Q. Tran', 'Alyssa Lees'] | 2022-02-22 | null | null | null | null | ['toxic-comment-classification'] | ['natural-language-processing'] | [-2.55220354e-01 -6.49550915e-01 -4.07258749e-01 4.42237109e-02
-9.24469233e-01 -1.11751747e+00 7.66970515e-01 2.58715928e-01
-3.28628808e-01 6.07028008e-01 3.24823022e-01 -5.88947594e-01
2.45218843e-01 -4.85364646e-01 -4.96480316e-01 -2.82102883e-01
-1.31798415e-02 -8.82017910e-02 3.10911179e-01 -5.73921740e-01
3.60452056e-01 2.78370440e-01 -1.09671462e+00 5.29875994e-01
1.07563162e+00 7.43351042e-01 -3.77659738e-01 6.10053718e-01
-3.67417410e-02 1.09446561e+00 -8.68932962e-01 -1.36247182e+00
2.00110957e-01 -7.58170113e-02 -6.10919476e-01 -3.53755623e-01
7.53267467e-01 -2.69789428e-01 -2.70914853e-01 1.31827450e+00
5.27589500e-01 -2.69724578e-01 6.48114502e-01 -1.25206447e+00
-1.09271324e+00 8.06976438e-01 -5.40654898e-01 2.61097103e-01
3.26015055e-01 2.56989270e-01 1.07652664e+00 -6.60006166e-01
7.21485496e-01 1.26701677e+00 7.31351912e-01 5.88423967e-01
-1.12402189e+00 -1.02349150e+00 9.98985469e-02 8.41089338e-02
-1.09408033e+00 -3.45961899e-01 6.59236670e-01 -8.13478827e-01
7.49582708e-01 4.38647717e-01 2.13920280e-01 2.10029793e+00
3.39358747e-01 9.24270272e-01 1.23470199e+00 -2.98203468e-01
-8.67442116e-02 6.66160464e-01 1.49916023e-01 8.30795050e-01
3.97833288e-01 -3.33442748e-01 -5.75365424e-01 -5.16394258e-01
-2.56608129e-01 -5.36518581e-02 -2.92466640e-01 -2.15856597e-01
-7.91643262e-01 1.27787864e+00 1.07791908e-01 2.74585515e-01
2.71674663e-01 4.74770181e-02 8.55542481e-01 2.84910798e-01
7.39519060e-01 6.50196016e-01 -2.30139166e-01 -4.21402872e-01
-7.06697047e-01 2.57120490e-01 1.24278283e+00 9.03473914e-01
2.39813179e-01 -7.79690742e-02 -9.45047438e-02 1.07024503e+00
1.30412996e-01 8.96249294e-01 5.59601545e-01 -6.71996474e-01
7.66838253e-01 3.95644188e-01 -4.84541990e-02 -1.21135902e+00
-1.12352915e-01 -4.34594929e-01 -3.44023287e-01 -1.31562114e-01
4.23120826e-01 -4.13537502e-01 -3.28460604e-01 1.54791212e+00
1.32265076e-01 -3.73362720e-01 -3.92004013e-01 4.71985847e-01
3.08296025e-01 4.56545085e-01 3.01812381e-01 1.66324675e-01
1.47756016e+00 -9.29714978e-01 -6.25640094e-01 -8.74256119e-02
1.08672547e+00 -9.18076813e-01 1.30815864e+00 5.11672199e-01
-5.59042335e-01 8.58954936e-02 -1.04290545e+00 -3.14975858e-01
-1.01600492e+00 -4.29767579e-01 5.78613520e-01 1.29056847e+00
-6.90918684e-01 2.31302306e-01 -3.78191710e-01 -5.10722101e-01
3.00800890e-01 -2.34951138e-01 -2.61372507e-01 -4.45936173e-02
-1.43883133e+00 1.21128380e+00 -5.04556708e-02 -5.30108511e-01
-6.70777798e-01 -6.43653691e-01 -5.72489262e-01 -2.00899020e-01
2.86282063e-01 -3.81690217e-04 1.30714560e+00 -1.00766134e+00
-1.10957670e+00 7.04752207e-01 2.48860672e-01 -3.76989782e-01
4.69082683e-01 -4.14656639e-01 -7.59776413e-01 -1.81600824e-01
4.57437858e-02 5.98510206e-02 9.45179880e-01 -9.99347329e-01
-4.86572385e-01 -1.52003959e-01 2.65933096e-01 -1.68910414e-01
-1.08175647e+00 6.66785717e-01 -1.47754446e-01 -7.94291079e-01
-1.18023908e+00 -1.03026652e+00 4.03170466e-01 -2.54335225e-01
-5.50682843e-01 -5.95344172e-04 1.03457189e+00 -9.83672380e-01
1.83113265e+00 -2.29617286e+00 -2.20381953e-02 1.62856624e-01
3.65298897e-01 4.06319320e-01 -6.07247464e-02 7.72400796e-01
2.26938322e-01 8.02691817e-01 -1.73892885e-01 -4.04942900e-01
2.72741675e-01 -2.65084654e-01 -4.52469587e-01 6.08800173e-01
-8.81480873e-02 7.61028409e-01 -9.66961861e-01 -4.62553203e-01
-2.16783732e-01 4.29466486e-01 -8.65281343e-01 -4.78496887e-02
-2.22216234e-01 8.25046096e-03 -4.60875034e-01 8.76680195e-01
3.29802483e-01 -3.66794765e-02 8.42203721e-02 2.47468591e-01
-1.92216322e-01 3.54630828e-01 -4.93234515e-01 1.42906594e+00
-7.87864745e-01 9.72418964e-01 1.53875113e-01 -1.77488953e-01
5.71173072e-01 5.16444296e-02 1.34303004e-01 -7.42025614e-01
3.94244671e-01 3.39307725e-01 2.66150419e-05 -6.44850194e-01
5.52513719e-01 2.47845352e-01 -4.33438659e-01 6.51571393e-01
-6.87431395e-02 1.75583720e-01 3.97973657e-01 5.25418162e-01
1.34753263e+00 -2.76296705e-01 2.69550711e-01 -2.16450244e-01
5.51791310e-01 -1.98764279e-01 1.73832342e-01 6.16827428e-01
-5.53523004e-01 2.02161714e-01 7.80832171e-01 -4.17775482e-01
-1.00438023e+00 -8.44428539e-01 -3.14943314e-01 1.60650218e+00
-2.45185629e-01 -5.99011898e-01 -8.71496260e-01 -1.15385497e+00
2.14020431e-01 8.23003471e-01 -5.94260216e-01 -7.04765767e-02
-4.37700480e-01 -8.49974573e-01 1.05960202e+00 -1.50549978e-01
4.42143142e-01 -6.92931175e-01 -1.87717527e-01 5.02578244e-02
-3.01519573e-01 -1.20215905e+00 -8.04750502e-01 -6.86770007e-02
-3.01440321e-02 -1.21153224e+00 -3.47600937e-01 -4.82531160e-01
6.41728044e-02 3.08483094e-01 1.08356154e+00 2.32537031e-01
-1.57049075e-01 2.51799762e-01 -7.06865489e-01 -5.24062335e-01
-9.09762025e-01 5.36863387e-01 1.08715352e-02 1.40017971e-01
6.42607093e-01 -2.93950379e-01 -2.13357463e-01 9.30715948e-02
-9.65030193e-01 -2.23391652e-01 2.19069228e-01 5.53874671e-01
-3.99442106e-01 -2.34196201e-01 4.80042756e-01 -1.35170007e+00
9.85246718e-01 -1.09905779e+00 -4.17001963e-01 3.21166247e-01
-3.76267344e-01 -1.49836749e-01 1.04620445e+00 -5.40929794e-01
-1.00026619e+00 -5.11328876e-01 -1.67922348e-01 -4.29092385e-02
8.61003026e-02 1.91062704e-01 -7.77079388e-02 -2.19641015e-01
1.10210145e+00 3.94690409e-02 -1.20130427e-01 -5.72527468e-01
7.43238747e-01 1.31532753e+00 5.48539571e-02 -7.10669160e-01
9.94030714e-01 3.76180887e-01 -6.46006048e-01 -9.72176433e-01
-7.84137666e-01 -5.11974394e-01 -3.50836664e-01 -1.50596708e-01
8.52901816e-01 -8.38344514e-01 -6.38010085e-01 6.40788853e-01
-1.25707936e+00 -2.75265127e-01 4.78051960e-01 1.05695955e-01
4.80906926e-02 6.10193610e-01 -1.04047084e+00 -5.97754180e-01
-4.86113995e-01 -1.26947474e+00 7.32504725e-01 -2.72761643e-01
-4.70776796e-01 -1.31221068e+00 3.80947053e-01 6.32436693e-01
6.31702363e-01 3.19333524e-01 1.08541596e+00 -1.11845207e+00
-2.51215786e-01 -2.73436397e-01 -3.06190610e-01 5.45127332e-01
4.19288389e-02 3.60443920e-01 -9.19159174e-01 -3.89263302e-01
-7.62007162e-02 -6.56338334e-01 5.45698702e-01 -5.24447620e-01
1.10855889e+00 -7.76125133e-01 -1.16451740e-01 6.44681036e-01
1.30867231e+00 -1.38706639e-01 2.67050028e-01 5.36334753e-01
8.64359796e-01 5.85700095e-01 7.35132918e-02 4.19193000e-01
4.71502542e-01 7.15884387e-01 4.07213986e-01 2.81429768e-01
-1.22168317e-01 -4.20091331e-01 8.69552672e-01 1.14311004e+00
3.48363757e-01 -5.54105937e-01 -9.93366420e-01 3.98344696e-01
-1.37954855e+00 -1.27816188e+00 -1.62533775e-01 1.98750067e+00
1.09288514e+00 8.00598711e-02 4.07356113e-01 -2.47182578e-01
6.10920966e-01 3.17827255e-01 -3.26574981e-01 -1.05614483e+00
-5.35547882e-02 -5.38773350e-02 8.01682293e-01 4.71492440e-01
-1.18448317e+00 1.01105523e+00 5.98624563e+00 1.08009553e+00
-1.15055835e+00 5.71732879e-01 4.50323045e-01 -2.75399446e-01
-3.86073709e-01 -3.90149236e-01 -8.89541805e-01 9.91148770e-01
1.28419185e+00 -1.63195267e-01 8.51476192e-01 1.01520097e+00
-5.31921685e-02 2.65684575e-01 -9.03822482e-01 7.32401252e-01
6.84877992e-01 -1.05878866e+00 -3.70383933e-02 1.04342744e-01
7.88156807e-01 4.78848606e-01 4.03265804e-01 5.46845257e-01
6.93202138e-01 -8.08631957e-01 9.76614594e-01 -2.54772305e-01
7.82286227e-01 -6.05327189e-01 4.36171532e-01 3.51000100e-01
-6.51701629e-01 -4.06983644e-01 2.10256074e-02 1.26371950e-01
-1.52269289e-01 4.32115406e-01 -5.74893951e-01 8.05673599e-02
6.33698642e-01 7.65162945e-01 -1.02562559e+00 8.23422194e-01
-2.50653923e-01 7.35535324e-01 9.17901322e-02 -3.85977954e-01
1.94067881e-01 4.84635495e-02 4.65852916e-01 1.96041667e+00
1.23137139e-01 -4.69258368e-01 -3.53181139e-02 5.63012958e-01
-6.01355493e-01 3.44212055e-01 -1.02934754e+00 -2.64057130e-01
6.66389525e-01 1.34184074e+00 -2.73890793e-01 -1.42611181e-02
-9.22259569e-01 8.69742036e-01 7.15498149e-01 2.85236627e-01
-1.26615000e+00 -5.25904834e-01 9.53021407e-01 9.16231349e-02
-9.33465436e-02 -2.84459293e-01 -6.30518422e-02 -1.53634202e+00
-1.73888788e-01 -1.69314253e+00 3.22210550e-01 -3.07910591e-01
-1.62609112e+00 5.17432988e-01 -9.06212181e-02 -9.44772124e-01
-8.28915238e-02 -7.39493132e-01 -2.79455185e-01 5.98535180e-01
-1.50781584e+00 -1.16317034e+00 1.26035586e-01 5.50924301e-01
6.17329180e-01 -1.90085024e-01 5.97310901e-01 7.25787997e-01
-8.24688137e-01 8.17896783e-01 3.72764587e-01 6.39956236e-01
1.08625734e+00 -1.20642471e+00 5.27575970e-01 9.33115423e-01
1.71852350e-01 9.25761044e-01 5.71018398e-01 -6.80326104e-01
-1.29710233e+00 -1.16098857e+00 8.96984577e-01 -1.09347606e+00
1.53791142e+00 -1.00976288e+00 -7.93472648e-01 7.25625396e-01
5.11653423e-01 -4.99398530e-01 9.84303176e-01 3.05398643e-01
-1.21006322e+00 6.28971905e-02 -9.64002430e-01 8.85983109e-01
9.70668495e-01 -9.64402497e-01 -2.80974656e-01 7.07533956e-01
7.52987266e-01 -2.42638901e-01 -7.98928082e-01 -3.74358267e-01
7.34065890e-01 -1.06876552e+00 5.39342463e-01 -8.56256485e-01
5.27573526e-01 1.63796484e-01 -2.54474789e-01 -1.39703763e+00
-2.66964555e-01 -9.13001597e-01 -1.76268220e-01 1.33638501e+00
6.50818646e-01 -6.79503381e-01 2.25276530e-01 6.70327783e-01
1.43566225e-02 -4.50114250e-01 -7.32283950e-01 -8.38893712e-01
5.70537269e-01 -5.60184419e-01 3.72773081e-01 1.29176319e+00
3.63533825e-01 3.10347199e-01 -5.83381355e-01 -1.11168943e-01
6.62720621e-01 -3.94308507e-01 8.98344219e-01 -9.76784289e-01
-1.21483520e-01 -5.52834988e-01 -7.11481869e-02 -6.76713884e-01
4.27984536e-01 -1.16489148e+00 -4.40690190e-01 -7.54764378e-01
5.37200093e-01 -2.82698482e-01 -8.84641930e-02 4.29677516e-01
3.09505817e-02 3.29626620e-01 5.57404935e-01 2.01526895e-01
-8.01473618e-01 2.74609923e-02 8.54070246e-01 -2.60927200e-01
2.58617818e-01 -3.41407835e-01 -9.28069711e-01 7.81916440e-01
8.14752758e-01 -7.11829782e-01 -2.58262753e-01 -5.38223743e-01
6.73084080e-01 -5.92608690e-01 4.12599891e-02 -6.96922362e-01
-2.90081389e-02 -1.45938858e-01 -2.13315636e-01 1.21930949e-01
1.34097368e-01 -6.91470921e-01 -4.11326528e-01 3.91086370e-01
-5.33373833e-01 3.51583987e-01 -3.16113909e-03 5.33904254e-01
1.87343687e-01 -1.41833112e-01 7.81312883e-01 -5.60019985e-02
-4.06700879e-01 3.31395954e-01 -4.28947598e-01 7.83841848e-01
1.00872087e+00 2.17797101e-01 -1.08465660e+00 -3.63520980e-01
1.02636479e-01 3.27241956e-03 8.18115652e-01 9.66222107e-01
-2.25365404e-02 -1.06826806e+00 -7.73913145e-01 -1.87534746e-02
4.52903420e-01 -1.17569792e+00 -1.67850837e-01 6.13798738e-01
-5.84687173e-01 4.97262031e-01 -1.05085373e-01 -2.42749695e-02
-1.23273981e+00 7.22074628e-01 2.42710546e-01 -2.63371587e-01
-1.40173167e-01 5.27045310e-01 -1.07209377e-01 -4.83842075e-01
1.89191252e-01 -5.35677746e-02 -1.64550412e-02 1.39616713e-01
7.70224512e-01 5.62518954e-01 2.27067396e-01 -9.24130917e-01
-2.56381541e-01 1.78196970e-02 -2.40845695e-01 1.02599323e-01
8.65303993e-01 -1.23717345e-01 -2.25566894e-01 3.87732834e-01
1.62459350e+00 8.86774838e-01 -7.94476867e-01 -2.04701889e-02
1.45714000e-01 -5.90285301e-01 -2.48600200e-01 -1.01666462e+00
-7.86328316e-01 8.38479400e-01 1.12565123e-01 6.02747202e-01
5.15053451e-01 -2.65183687e-01 1.11709154e+00 2.65671134e-01
4.91998821e-01 -1.12114906e+00 1.02204509e-01 6.09094262e-01
8.11500728e-01 -1.22901976e+00 -1.28584821e-02 -2.99687803e-01
-6.91612899e-01 8.22180986e-01 5.15728056e-01 3.28924447e-01
6.21780396e-01 2.28264138e-01 2.17060715e-01 -5.72727732e-02
-8.64673793e-01 7.77219087e-02 1.79341003e-01 6.28122687e-01
7.28967905e-01 1.85312286e-01 -4.46775109e-01 3.95331025e-01
-3.06927711e-01 -5.55061877e-01 8.17306042e-01 6.80664837e-01
-3.78106773e-01 -1.23159468e+00 -3.40359122e-01 3.83973688e-01
-1.11361456e+00 -4.29148465e-01 -6.59280002e-01 1.04878891e+00
1.23071969e-01 1.17738187e+00 -4.43776757e-01 -5.96516311e-01
2.38620304e-02 1.85159594e-01 7.98821300e-02 -6.83483839e-01
-1.01301539e+00 -5.63727677e-01 3.98743421e-01 -5.32801211e-01
1.12794779e-01 -5.45888126e-01 -5.04459441e-01 -9.79557812e-01
-2.31407970e-01 2.88377050e-02 8.21714938e-01 5.95800519e-01
4.40036625e-01 -1.21503986e-01 6.99171722e-01 -1.43898889e-01
-7.51600444e-01 -8.87693942e-01 -4.10817534e-01 8.52172494e-01
2.23918051e-01 -5.18646121e-01 -7.00212002e-01 -1.69172525e-01] | [8.801642417907715, 10.562004089355469] |
527ba3e8-9989-4afa-9e75-a278185b1107 | tile2vec-unsupervised-representation-learning | 1805.02855 | null | http://arxiv.org/abs/1805.02855v2 | http://arxiv.org/pdf/1805.02855v2.pdf | Tile2Vec: Unsupervised representation learning for spatially distributed data | Geospatial analysis lacks methods like the word vector representations and
pre-trained networks that significantly boost performance across a wide range
of natural language and computer vision tasks. To fill this gap, we introduce
Tile2Vec, an unsupervised representation learning algorithm that extends the
distributional hypothesis from natural language -- words appearing in similar
contexts tend to have similar meanings -- to spatially distributed data. We
demonstrate empirically that Tile2Vec learns semantically meaningful
representations on three datasets. Our learned representations significantly
improve performance in downstream classification tasks and, similar to word
vectors, visual analogies can be obtained via simple arithmetic in the latent
space. | ['Sherrie Wang', 'George Azzari', 'Stefano Ermon', 'Neal Jean', 'David Lobell', 'Anshul Samar'] | 2018-05-08 | null | null | null | null | ['visual-analogies'] | ['computer-vision'] | [ 6.29293360e-03 -1.81624100e-01 -1.91203147e-01 -4.76120234e-01
-4.83684123e-01 -6.27469540e-01 1.18240547e+00 5.57969868e-01
-3.68209273e-01 4.99434859e-01 1.07963729e+00 -7.79974818e-01
-4.82246839e-02 -1.11822736e+00 -4.33182329e-01 -4.52074260e-01
-2.84026980e-01 -1.86874419e-02 3.06772604e-03 -2.24527687e-01
2.99708247e-01 4.64243054e-01 -1.35368836e+00 3.09313953e-01
3.39340717e-01 5.95153809e-01 9.71342325e-02 3.99847418e-01
-6.33609712e-01 8.13111424e-01 -2.64505416e-01 -1.28670529e-01
1.93120018e-01 -1.55297577e-01 -5.86302817e-01 -4.99391645e-01
4.53162581e-01 -1.16154425e-01 -5.60128212e-01 8.52260530e-01
2.05019802e-01 5.85132122e-01 1.36832011e+00 -1.01085877e+00
-1.63970816e+00 4.45016772e-01 -8.36105824e-01 5.71614802e-01
1.38054892e-01 1.43328100e-01 1.44821191e+00 -1.18966234e+00
6.97666407e-01 1.57073617e+00 1.02880609e+00 5.01630679e-02
-1.43373477e+00 -3.46841902e-01 -2.03120168e-02 2.60059554e-02
-1.55615497e+00 4.12445143e-02 3.11339021e-01 -7.57119179e-01
1.63947475e+00 1.60843670e-01 4.81868595e-01 1.10743916e+00
4.17074829e-01 4.53161091e-01 7.65959263e-01 -4.49830711e-01
1.77844137e-01 -2.62648433e-01 2.82656342e-01 2.01216087e-01
3.82854789e-01 2.11883068e-01 -3.84406447e-01 -2.18229815e-01
9.72387791e-01 4.42958564e-01 6.55235946e-02 -3.56207758e-01
-1.28099251e+00 1.22401655e+00 9.71202970e-01 6.25816882e-01
-4.77175087e-01 8.55920196e-01 4.37090129e-01 3.67109999e-02
8.13324332e-01 7.01528549e-01 -1.43401057e-01 1.82053939e-01
-7.15148926e-01 1.08404957e-01 1.24457479e-01 6.86584175e-01
9.21399176e-01 4.51538622e-01 -1.98252261e-01 8.43756795e-01
3.65705669e-01 7.19401538e-01 7.68494725e-01 -4.81827646e-01
1.51749983e-01 4.17591661e-01 -2.10055530e-01 -1.43790793e+00
-1.55864894e-01 -1.15357652e-01 -7.07908571e-01 2.24423379e-01
1.39044136e-01 1.53802574e-01 -1.03638113e+00 1.61599147e+00
-1.13011137e-01 3.40739042e-01 1.63435459e-01 7.26573765e-01
7.17844069e-01 9.59781408e-01 5.19838095e-01 3.42425555e-01
1.38096178e+00 -5.39087892e-01 -5.71720898e-01 -4.78806406e-01
5.85575521e-01 -5.71060300e-01 1.25009334e+00 -1.46862552e-01
-4.91695434e-01 -5.18691123e-01 -1.03739762e+00 -4.42018777e-01
-1.14666724e+00 -5.28763235e-01 1.06275153e+00 3.75461429e-01
-1.39610434e+00 4.47694600e-01 -6.25430346e-01 -8.40954542e-01
5.81756175e-01 -2.45895714e-01 -5.30149877e-01 -1.84586574e-03
-1.00039053e+00 1.16336644e+00 3.14226061e-01 -5.14744401e-01
-7.65467584e-01 -8.48463535e-01 -1.31763673e+00 2.39072770e-01
-4.75362688e-01 -6.78973854e-01 9.41084743e-01 -7.05606461e-01
-4.57350820e-01 9.32368696e-01 -3.46186876e-01 -5.23748636e-01
-2.15830520e-01 -3.59568536e-01 -3.54378164e-01 -2.22334862e-02
6.32171869e-01 8.27615380e-01 6.66581929e-01 -1.10061705e+00
-2.19332144e-01 -2.86533713e-01 -3.74782771e-01 -5.07229008e-02
-5.99940717e-01 -1.78504568e-02 -9.08397287e-02 -1.31161416e+00
-3.24601978e-02 -4.47030306e-01 -6.78702116e-01 4.14568543e-01
-1.04797222e-01 -3.93707901e-01 6.38026178e-01 -4.47657198e-01
9.43809092e-01 -2.34330511e+00 -6.03047721e-02 3.57377529e-01
5.27132273e-01 -1.04537830e-01 -4.21063304e-01 5.89011669e-01
-5.20853937e-01 4.56164241e-01 -2.99781293e-01 2.26716191e-01
-1.18754590e-02 4.72279638e-01 -8.52006376e-01 6.79170191e-01
3.76935929e-01 1.33105576e+00 -1.36908233e+00 -1.59108400e-01
5.84017038e-01 5.34482181e-01 -2.99112827e-01 -1.17151126e-01
-5.00776917e-02 -8.77218395e-02 -2.61613101e-01 2.60049820e-01
4.49950069e-01 -3.68386000e-01 1.50143057e-01 1.55646041e-01
-9.54626799e-02 2.87608325e-01 -4.38741773e-01 1.80530035e+00
-5.04756451e-01 1.45757818e+00 -6.96303189e-01 -1.16708577e+00
8.71059537e-01 3.44855599e-02 1.54594570e-01 -8.64843726e-01
7.11236224e-02 -2.63699442e-01 -4.01851356e-01 -4.90456134e-01
7.14226723e-01 -5.31886399e-01 -3.13984066e-01 6.88947022e-01
6.97128382e-03 -1.70245290e-01 -4.13715154e-01 3.66804421e-01
1.01799691e+00 -1.21631809e-01 7.94052601e-01 -6.39010489e-01
-2.63934463e-01 3.19774896e-01 1.45562306e-01 6.44368708e-01
4.27476205e-02 4.52830493e-01 1.81135014e-01 -7.95826137e-01
-1.13937104e+00 -1.75354052e+00 -2.56821901e-01 1.50913930e+00
5.14465198e-02 -5.39556980e-01 -4.95529734e-03 -3.44234616e-01
5.07623732e-01 1.16369402e+00 -1.07159102e+00 -6.35136245e-03
-1.36581808e-01 -4.35500175e-01 7.68838167e-01 1.26089251e+00
-3.67889315e-01 -9.07695293e-01 -6.95316613e-01 -1.43564232e-02
3.71164352e-01 -5.51208556e-01 1.67519096e-02 3.41673136e-01
-7.87303269e-01 -7.51273274e-01 -8.72366846e-01 -7.88590908e-01
4.81792301e-01 7.60616124e-01 1.39728284e+00 -1.32576048e-01
-7.75636852e-01 4.11320508e-01 -5.00021696e-01 -4.57043499e-01
8.48780498e-02 -4.15479898e-01 1.76823586e-01 -4.03197467e-01
9.80967641e-01 -7.10666716e-01 -4.43388969e-01 -8.86195302e-02
-1.01409197e+00 -2.41027549e-01 2.88469940e-01 7.17056334e-01
5.12590826e-01 -4.54606354e-01 3.32748860e-01 -4.43460286e-01
8.58218670e-01 -9.88263786e-01 -1.08844258e-01 2.99494237e-01
-3.17058712e-01 3.52871865e-01 2.93806434e-01 -3.82862151e-01
-6.25482321e-01 -2.06846461e-01 2.14122478e-02 -4.81617182e-01
-2.39491254e-01 7.87708819e-01 4.44822192e-01 3.13240528e-01
1.23312747e+00 2.15838879e-01 -2.29297549e-01 -3.03504437e-01
1.30687559e+00 3.86836231e-01 6.24182820e-01 -5.27196646e-01
8.55830431e-01 6.68138564e-01 -7.60730058e-02 -1.28063023e+00
-3.97036791e-01 -7.15098917e-01 -5.91744781e-01 1.88220084e-01
1.32942629e+00 -1.03246164e+00 -2.23462909e-01 -2.96399742e-01
-1.28580534e+00 -1.89485446e-01 -5.25850356e-01 5.86986780e-01
-2.12620184e-01 2.92541832e-01 -3.37558776e-01 -6.00845337e-01
-2.74042487e-02 -5.58999717e-01 1.04014361e+00 -6.53211251e-02
-7.57636368e-01 -1.44622922e+00 4.35289949e-01 -6.81510687e-01
6.39883518e-01 3.78216118e-01 1.15878499e+00 -6.37214303e-01
-9.04352814e-02 1.40804380e-01 -6.64242983e-01 -7.60099292e-02
4.47824627e-01 -1.27705485e-01 -1.05780303e+00 -1.27114840e-02
-4.98824239e-01 -2.16746822e-01 1.45660353e+00 6.09751761e-01
1.12744617e+00 3.21271494e-02 -5.87588191e-01 6.66247845e-01
1.57182086e+00 -1.76936984e-01 6.51828110e-01 3.63012850e-01
7.13772178e-01 6.73245728e-01 -1.47563266e-02 3.62578541e-01
1.03493325e-01 2.37055853e-01 3.09248269e-01 -1.76788449e-01
-1.33107722e-01 -5.78121901e-01 5.49047291e-02 4.11113143e-01
-2.71776337e-02 -1.83771208e-01 -1.31115949e+00 1.12837517e+00
-1.69019103e+00 -1.16655958e+00 -3.20892602e-01 1.70351350e+00
3.67087454e-01 -3.56098056e-01 -3.19674104e-01 -2.85012335e-01
6.73583984e-01 6.35454297e-01 -3.99963677e-01 -4.72857922e-01
-3.49258751e-01 5.89877129e-01 6.95718348e-01 4.17914361e-01
-1.01521778e+00 1.35119927e+00 7.93591404e+00 8.11366141e-01
-1.00277758e+00 2.81748027e-01 2.96897650e-01 -5.60678169e-02
-1.04414976e+00 -1.92140818e-01 1.67359412e-02 -9.10149068e-02
6.83013737e-01 -4.35801059e-01 -7.70460209e-03 8.67996395e-01
9.30671394e-02 2.68715262e-01 -1.13309956e+00 1.11220491e+00
7.88387954e-02 -1.83834696e+00 5.83218932e-01 1.17133588e-01
7.01856315e-01 3.75893146e-01 5.05594373e-01 1.78422779e-01
9.67727602e-01 -1.61478603e+00 5.20680070e-01 3.18915159e-01
1.04494643e+00 -5.28022110e-01 3.86702865e-01 -4.14966911e-01
-1.52665007e+00 1.18477263e-01 -9.34655309e-01 -5.34817994e-01
1.93114415e-01 4.02364284e-01 -7.42519140e-01 2.16500774e-01
7.79726088e-01 1.10318220e+00 -6.83512449e-01 7.70087779e-01
-5.85951030e-01 6.20433927e-01 -3.43278348e-02 -1.69121131e-01
6.93564415e-01 4.75361980e-02 1.89161062e-01 1.62921095e+00
4.39087212e-01 1.29335836e-01 -1.82580620e-01 1.03539014e+00
1.14784904e-01 2.10266769e-01 -1.63866472e+00 -3.04008007e-01
4.82785195e-01 7.64990747e-01 -7.77250707e-01 -4.02024299e-01
-6.00311458e-01 9.01799917e-01 4.82198507e-01 6.64816260e-01
-6.85965180e-01 -4.54004109e-01 1.19884777e+00 -3.45659554e-02
2.99460560e-01 -5.80814660e-01 -6.74650013e-01 -1.05722165e+00
-5.88584602e-01 -8.53008479e-02 3.41913730e-01 -1.22741771e+00
-1.95780957e+00 4.67537284e-01 -1.31353196e-02 -1.18668401e+00
-3.64936948e-01 -9.30457056e-01 -9.56430495e-01 9.64804709e-01
-1.28943050e+00 -1.12753904e+00 -7.76182711e-02 6.83840990e-01
5.50967813e-01 -3.24805856e-01 1.05406249e+00 -2.05188856e-01
2.03113984e-02 5.37043989e-01 3.92615527e-01 3.99529099e-01
6.03432119e-01 -1.30488646e+00 1.05588686e+00 7.34845817e-01
6.96423054e-01 1.12796688e+00 4.50374812e-01 -6.91036046e-01
-1.09293294e+00 -1.24308193e+00 8.97054970e-01 -4.78648961e-01
1.17709541e+00 -2.23963022e-01 -9.20166254e-01 1.01566780e+00
5.04515529e-01 1.80058002e-01 1.34953845e+00 5.00331998e-01
-1.12802064e+00 5.23401499e-01 -8.19408476e-01 8.10920358e-01
1.03767264e+00 -1.14081752e+00 -1.31639850e+00 2.77114689e-01
8.76425505e-01 4.43588376e-01 -5.62349498e-01 -1.30455360e-01
6.62733495e-01 -5.39106607e-01 1.41710293e+00 -1.28804374e+00
7.27150619e-01 -1.90250471e-01 -8.93078089e-01 -1.56076825e+00
-7.79077172e-01 4.49049734e-02 4.62926656e-01 8.01214933e-01
4.53684181e-01 -6.68468654e-01 4.10117179e-01 2.47682899e-01
2.47327369e-02 -2.17898101e-01 -8.09876621e-01 -9.26635027e-01
6.60516143e-01 -1.01677716e+00 7.07640886e-01 1.57870424e+00
2.81025738e-01 4.01872218e-01 -5.99824116e-02 1.41495988e-01
5.51527202e-01 8.63306504e-03 5.00882745e-01 -1.26604533e+00
6.37217090e-02 -7.43543327e-01 -9.82790232e-01 -1.06248736e+00
3.93195927e-01 -1.37041318e+00 -4.07245420e-02 -1.85346985e+00
1.32248044e-01 -1.89044729e-01 -4.30424958e-01 6.90586865e-01
-1.83195844e-01 4.68326300e-01 -4.25223596e-02 1.87356725e-01
-2.86205262e-01 5.85835218e-01 5.94569266e-01 -4.92166847e-01
2.80904174e-02 -9.54024374e-01 -8.30632508e-01 7.20174074e-01
5.93296468e-01 -3.85188311e-01 -5.57097077e-01 -1.02033913e+00
4.22125041e-01 -3.77415597e-01 5.81641078e-01 -5.82858205e-01
-1.04858531e-02 -4.52652931e-01 8.14478159e-01 -4.47475851e-01
6.19091168e-02 -6.88886940e-01 -1.81200430e-01 3.26508850e-01
-5.47120035e-01 4.34202045e-01 6.58025026e-01 8.63698125e-01
-4.76538241e-02 1.55231833e-01 4.07645494e-01 -1.81635797e-01
-1.36357164e+00 4.25346643e-02 -9.20214295e-01 -2.48565637e-02
1.09634030e+00 -3.96108299e-01 -6.15887761e-01 -5.29105961e-01
-5.66729307e-01 3.48637253e-03 5.46791077e-01 7.39340901e-01
9.92488205e-01 -1.72281122e+00 -8.18240523e-01 1.94224253e-01
5.64423203e-01 -3.97408515e-01 -1.25554994e-01 1.93343595e-01
-7.66979277e-01 3.94924074e-01 -5.14322698e-01 -5.97100973e-01
-5.52783012e-01 7.57021904e-01 -2.18109302e-02 2.67782569e-01
-8.08459878e-01 1.02454162e+00 8.09061587e-01 -3.40398848e-01
-2.80466914e-01 -5.14739096e-01 -2.34496340e-01 3.31119329e-01
8.86835515e-01 1.63850799e-01 -4.35380399e-01 -8.11767638e-01
-5.24086177e-01 7.18933702e-01 1.92030162e-01 -3.33416998e-01
1.61746252e+00 1.59969389e-01 -2.46249691e-01 6.55036628e-01
1.53036594e+00 -1.61898509e-01 -6.81267977e-01 -2.87297457e-01
2.07624927e-01 -6.05415404e-01 1.58563107e-01 -2.53134638e-01
-7.08692729e-01 1.21406794e+00 6.83790565e-01 1.38724566e-01
5.33838749e-01 4.66939300e-01 5.20729601e-01 5.11948228e-01
1.17349982e-01 -5.74183822e-01 1.23354875e-01 6.93373442e-01
8.77982140e-01 -1.07277238e+00 1.68990165e-01 -2.84473728e-02
-6.49702132e-01 1.05734980e+00 2.50834107e-01 -3.90389115e-01
9.20069277e-01 1.95834860e-01 2.43176550e-01 -6.11289501e-01
-5.27206004e-01 -5.05255997e-01 4.41745549e-01 1.14225328e+00
6.51121080e-01 2.50654787e-01 -5.16240075e-02 3.00093800e-01
-2.50634521e-01 -5.34597337e-01 1.60144746e-01 8.04921091e-01
-5.75243890e-01 -5.78729033e-01 -4.03353751e-01 4.96965230e-01
-1.82319894e-01 -6.11475229e-01 -5.19984543e-01 6.32809281e-01
-1.48882031e-01 8.81210029e-01 8.57699633e-01 -4.95163202e-01
-4.06840295e-02 9.12877545e-02 5.16101457e-02 -7.94420481e-01
-3.71214807e-01 -1.19967237e-01 -2.38376975e-01 -6.14788353e-01
-3.47241104e-01 -4.13329154e-01 -1.30496573e+00 -3.82203132e-01
7.37464204e-02 -2.24673972e-02 5.53036332e-01 8.84733498e-01
2.46178716e-01 4.91572738e-01 2.07202539e-01 -9.85704422e-01
-1.29174963e-01 -9.82425809e-01 -7.32781768e-01 7.07496524e-01
1.40573859e-01 -6.58488274e-01 -2.55620033e-01 1.03493355e-01] | [10.523207664489746, 2.1367335319519043] |
d4cd5b5c-acbc-4e2b-aabd-b62e87601138 | learning-predictive-representations-for | 2003.05436 | null | https://arxiv.org/abs/2003.05436v1 | https://arxiv.org/pdf/2003.05436v1.pdf | Learning Predictive Representations for Deformable Objects Using Contrastive Estimation | Using visual model-based learning for deformable object manipulation is challenging due to difficulties in learning plannable visual representations along with complex dynamic models. In this work, we propose a new learning framework that jointly optimizes both the visual representation model and the dynamics model using contrastive estimation. Using simulation data collected by randomly perturbing deformable objects on a table, we learn latent dynamics models for these objects in an offline fashion. Then, using the learned models, we use simple model-based planning to solve challenging deformable object manipulation tasks such as spreading ropes and cloths. Experimentally, we show substantial improvements in performance over standard model-based learning techniques across our rope and cloth manipulation suite. Finally, we transfer our visual manipulation policies trained on data purely collected in simulation to a real PR2 robot through domain randomization. | ['Pieter Abbeel', 'Wilson Yan', 'Lerrel Pinto', 'Ashwin Vangipuram'] | 2020-03-11 | null | null | null | null | ['deformable-object-manipulation'] | ['robots'] | [ 1.67885140e-01 1.78246126e-01 -4.18179691e-01 9.61140394e-02
-5.52981138e-01 -9.39257801e-01 5.47592819e-01 -1.73309535e-01
-1.47993237e-01 6.16321385e-01 8.65828916e-02 6.91747591e-02
-1.32960364e-01 -4.20451283e-01 -1.29573143e+00 -4.90674078e-01
-3.81399214e-01 1.04583168e+00 2.09711298e-01 -1.50966555e-01
1.33758232e-01 8.61387134e-01 -1.27158272e+00 2.20124975e-01
4.52264190e-01 4.65247244e-01 5.61533451e-01 1.13052857e+00
3.36143553e-01 9.93136466e-01 -1.91280872e-01 3.46047431e-01
6.03557646e-01 8.57550949e-02 -7.16915667e-01 5.51243603e-01
6.07557118e-01 -7.33092427e-01 -5.75325906e-01 5.97792923e-01
5.44912145e-02 4.84915167e-01 9.92076039e-01 -1.29201531e+00
-7.03661799e-01 4.24495906e-01 -5.73554218e-01 -3.19729447e-01
1.85342118e-01 8.45918953e-01 6.81921184e-01 -5.07476330e-01
1.13519430e+00 1.85489571e+00 2.66357243e-01 9.83281314e-01
-1.53684533e+00 -3.26306731e-01 5.73728025e-01 -1.69352338e-01
-7.90470898e-01 -2.19504446e-01 5.48801064e-01 -8.39865327e-01
1.03162611e+00 -9.15202647e-02 8.37407708e-01 1.44777679e+00
3.10913175e-01 9.05603647e-01 9.52519953e-01 -5.42726442e-02
2.05829963e-01 -3.74340415e-01 -1.73435241e-01 1.23519957e+00
1.63555026e-01 2.74811000e-01 -2.37322807e-01 -4.13969815e-01
1.47357178e+00 4.73788418e-02 -1.97666474e-02 -1.01745081e+00
-1.31625032e+00 5.07386327e-01 5.95068216e-01 -4.94078398e-01
-3.24892014e-01 1.04466462e+00 3.17604020e-02 2.54303962e-02
1.43114120e-01 5.27833045e-01 -6.11751974e-01 -1.64246820e-02
-2.84647971e-01 8.72619271e-01 1.03139889e+00 1.19960785e+00
5.37353635e-01 2.48789981e-01 -3.35369855e-01 5.04983962e-01
6.00475132e-01 6.94855928e-01 -2.27402002e-01 -1.46723866e+00
4.13619399e-01 1.78050831e-01 6.60434484e-01 -8.94658267e-01
-3.33943427e-01 4.99636590e-01 -4.16313082e-01 9.03680325e-01
5.04814923e-01 -2.97610402e-01 -1.73951590e+00 1.62273920e+00
3.95847261e-01 1.27128288e-01 -8.62395838e-02 9.58556235e-01
2.17366844e-01 8.32792163e-01 2.71913409e-01 1.22509941e-01
8.44202280e-01 -1.31120205e+00 -4.69591111e-01 -2.67786175e-01
7.94617906e-02 -4.16922152e-01 1.11850727e+00 3.16151768e-01
-1.32763195e+00 -3.28207910e-01 -8.22879970e-01 -2.07119152e-01
2.01860722e-02 1.41282454e-01 6.68156862e-01 -2.07962424e-01
-9.63006377e-01 8.86342585e-01 -1.69703007e+00 -4.42925364e-01
6.44386292e-01 4.79902178e-01 -1.72338501e-01 5.99867553e-02
-2.68319130e-01 9.60430741e-01 2.36019209e-01 -1.26607239e-01
-2.12027454e+00 -5.46121716e-01 -9.27245378e-01 -7.70531297e-02
6.22833192e-01 -7.90844381e-01 1.42687464e+00 -5.04351795e-01
-1.91792309e+00 7.33213663e-01 -5.44944108e-02 -2.87774384e-01
8.45376730e-01 -3.12462777e-01 5.92895448e-01 5.64309023e-02
-2.60259241e-01 9.11018789e-01 1.28834903e+00 -2.12404275e+00
-1.75444111e-01 5.71337529e-02 3.79709631e-01 3.48523706e-01
2.21664011e-01 -3.38143766e-01 -4.70568806e-01 -6.83978856e-01
-1.26914173e-01 -1.48841012e+00 -6.11595452e-01 9.40331638e-01
-4.32377249e-01 -7.60717019e-02 1.12078822e+00 -7.26631820e-01
3.56143832e-01 -1.72548091e+00 1.02090216e+00 1.35569543e-01
2.85786122e-01 7.53476024e-02 -4.24441844e-01 5.20130157e-01
4.39566463e-01 6.82748035e-02 -3.02754879e-01 -3.47287029e-01
2.13867500e-01 5.30824006e-01 -4.80323374e-01 3.93230587e-01
3.43291491e-01 1.13692784e+00 -9.95870829e-01 -2.45969713e-01
1.93795845e-01 1.83283597e-01 -7.29086399e-01 4.87012565e-01
-1.19543338e+00 7.65637517e-01 -5.77794135e-01 9.05736327e-01
4.43776190e-01 -3.60609084e-01 5.05543470e-01 7.92125240e-03
1.27751138e-02 -2.20335230e-01 -8.40255260e-01 1.79942858e+00
-4.14437711e-01 3.97489905e-01 4.98237073e-01 -6.39818311e-01
6.06936991e-01 -4.20310460e-02 3.94719571e-01 1.21289179e-01
-1.45425871e-01 -2.22593427e-01 -1.90465510e-01 -8.83457363e-01
3.68145198e-01 2.36344449e-02 -1.87963501e-01 4.67716604e-01
7.92860314e-02 -8.24196637e-01 -1.36500567e-01 1.04534321e-01
1.19768775e+00 8.77233624e-01 -7.40953907e-02 -3.02598756e-02
-3.68800730e-01 3.72755438e-01 2.49686807e-01 8.27760637e-01
4.63822708e-02 4.69404936e-01 4.67502654e-01 -4.56822067e-01
-1.43675005e+00 -1.60285008e+00 4.12643075e-01 9.91817296e-01
3.39487374e-01 -1.10579602e-01 -4.32057172e-01 -7.59780407e-01
5.51405847e-01 4.68091547e-01 -6.74230039e-01 -1.23548612e-01
-1.02682519e+00 -1.70621067e-01 2.30919659e-01 7.09324062e-01
-1.97868012e-02 -1.26814795e+00 -7.51564801e-01 1.08770289e-01
2.33920977e-01 -9.54462290e-01 -3.62589866e-01 5.71784414e-02
-7.55449414e-01 -9.81673717e-01 -7.62478173e-01 -8.98663104e-01
8.71946394e-01 2.26963252e-01 9.00762796e-01 7.82725140e-02
-6.72553122e-01 9.57033515e-01 -6.80989726e-03 -3.67239296e-01
-7.25268424e-01 -8.32948610e-02 1.29288524e-01 -5.33910811e-01
-6.90098763e-01 -3.87294531e-01 -2.07052767e-01 2.30254345e-02
-6.58914328e-01 1.83291078e-01 3.31247360e-01 6.70917690e-01
6.44238830e-01 -4.20685560e-01 -5.96737266e-02 -3.71656328e-01
4.69271183e-01 -4.23995972e-01 -9.83947635e-01 4.30916339e-01
1.33857936e-01 2.79669106e-01 2.88451701e-01 -1.09869254e+00
-9.77204621e-01 3.99049461e-01 5.56011558e-01 -1.05037320e+00
-5.68780825e-02 9.47823748e-02 2.18554214e-01 -3.20043474e-01
5.16254008e-01 -2.15405315e-01 1.31540120e-01 -4.70519274e-01
5.67465186e-01 5.79819120e-02 2.85154611e-01 -1.08251047e+00
1.27243555e+00 6.78452730e-01 3.62195745e-02 -6.28333628e-01
-5.45512855e-01 1.45787403e-01 -5.30547023e-01 -2.04284623e-01
1.04100001e+00 -7.87896514e-01 -1.14454556e+00 5.94067693e-01
-1.21471417e+00 -1.31909120e+00 -3.96224648e-01 3.54189813e-01
-1.18420565e+00 1.45743400e-01 -7.58652568e-01 -7.31614053e-01
4.05005217e-02 -1.09005642e+00 1.39759600e+00 1.15003698e-01
-1.91874683e-01 -1.01049185e+00 1.36729404e-01 7.91054219e-02
3.03622812e-01 7.60579824e-01 1.14004683e+00 2.54482366e-02
-1.29763353e+00 1.78073972e-01 4.92274668e-03 2.36093104e-02
3.69401693e-01 2.54824042e-01 -4.47097689e-01 -6.43595636e-01
-5.25442183e-01 -9.66956973e-01 8.44510734e-01 4.60903317e-01
1.38063943e+00 -6.84316814e-01 -7.76304066e-01 6.53197944e-01
1.28983879e+00 -9.69790574e-03 8.86050016e-02 6.90154673e-04
1.13079357e+00 4.54729557e-01 7.44884074e-01 6.78010285e-01
4.38321978e-01 5.86524487e-01 6.88228548e-01 1.05425760e-01
-1.52031213e-01 -4.85820651e-01 4.25840646e-01 1.40479788e-01
-3.13343614e-01 -4.72414583e-01 -1.10334623e+00 6.74378932e-01
-2.06029272e+00 -7.81802058e-01 4.13102925e-01 1.79916036e+00
8.62564564e-01 -5.15363812e-02 1.74906850e-01 -8.65271091e-01
4.41369981e-01 2.11793050e-01 -1.01222765e+00 -2.31411546e-01
4.58113700e-01 2.39340309e-02 6.87304556e-01 7.85716236e-01
-1.15146255e+00 1.39348543e+00 6.86994028e+00 4.15897846e-01
-1.11040378e+00 -2.22208187e-01 7.52777904e-02 -2.94984221e-01
-3.04458678e-01 4.56683040e-02 -7.22169816e-01 7.93304890e-02
2.91359037e-01 -1.46171376e-01 1.12037480e+00 8.44218016e-01
3.23047191e-01 -1.80779826e-02 -1.28779852e+00 6.18362069e-01
-1.21694736e-01 -1.38964713e+00 2.31992692e-01 1.59188598e-01
9.44364607e-01 1.23693205e-01 2.04248890e-01 3.15059453e-01
1.27151072e+00 -1.22336352e+00 9.13887620e-01 7.96213269e-01
5.22628665e-01 -2.87070721e-01 -3.59436840e-01 5.40233195e-01
-9.49407876e-01 -1.83743820e-01 -3.25866878e-01 -6.77090809e-02
3.34265500e-01 -3.36485922e-01 -1.02239311e+00 -2.29742136e-02
5.97112238e-01 7.71109223e-01 1.17486328e-01 7.79270768e-01
-2.92553872e-01 3.77494961e-01 -4.37726587e-01 9.14878845e-02
2.55778730e-01 -2.92371195e-02 9.32396829e-01 9.27740514e-01
-7.48745650e-02 -7.19689131e-02 8.42401922e-01 1.14450860e+00
-2.00637445e-01 -7.16889262e-01 -9.47709501e-01 -2.95884788e-01
2.69956887e-01 1.13756216e+00 -6.57148063e-01 -3.70559931e-01
1.23417936e-01 9.15494084e-01 7.83278048e-01 7.25908756e-01
-1.09075332e+00 2.84689814e-01 9.38963771e-01 7.88718387e-02
5.47416568e-01 -1.11204350e+00 2.12996826e-01 -1.30558443e+00
-1.23436525e-01 -8.87475491e-01 -3.29749554e-01 -7.41521955e-01
-1.34242070e+00 -8.64404887e-02 5.83002746e-01 -1.02597570e+00
-3.04895669e-01 -7.62950599e-01 -2.70318985e-01 5.94557822e-01
-1.19739068e+00 -1.54267979e+00 -3.80420059e-01 4.20643181e-01
8.91698897e-01 1.50983155e-01 6.56493485e-01 -5.47197223e-01
-1.63539678e-01 3.70639819e-03 -9.33589879e-03 1.68109145e-02
5.21533966e-01 -1.31385505e+00 7.66307473e-01 2.61876047e-01
-2.34462142e-01 5.34345448e-01 6.02104843e-01 -1.05799186e+00
-1.82883716e+00 -1.19406950e+00 -3.25671494e-01 -7.32765853e-01
7.04122484e-01 -6.60796285e-01 -9.13347602e-01 1.32471156e+00
1.24156885e-01 3.30232471e-01 -9.74316597e-02 -4.32333946e-01
-1.84595466e-01 4.68658179e-01 -1.25574446e+00 8.06558192e-01
1.33396244e+00 -3.50813657e-01 -4.82942313e-01 6.46380782e-01
9.09127057e-01 -1.12078094e+00 -8.33531737e-01 5.12312531e-01
8.23835611e-01 1.47879958e-01 1.18698835e+00 -1.26792479e+00
4.29878324e-01 -2.88648874e-01 -1.74459040e-01 -1.34937370e+00
-3.58044237e-01 -8.39513361e-01 -6.88250721e-01 7.66926765e-01
7.38357455e-02 -1.34529769e-01 8.00296962e-01 6.51577353e-01
6.18521199e-02 -6.20258629e-01 -4.48484868e-01 -9.93334353e-01
1.85391456e-01 1.24500982e-01 -1.70397328e-03 8.87388468e-01
-4.83919889e-01 -1.16198942e-01 -6.97041214e-01 3.95072252e-01
8.16459060e-01 1.30271494e-01 1.04750359e+00 -7.74598062e-01
-7.07408130e-01 -2.52978094e-02 -8.06619450e-02 -1.28166330e+00
5.59305251e-01 -7.55081475e-01 5.52212715e-01 -1.81596470e+00
3.40608031e-01 -5.84360003e-01 3.49946111e-01 9.41492677e-01
-3.40898223e-02 -3.64602804e-01 6.93177342e-01 2.22936526e-01
-4.86149669e-01 6.81029022e-01 1.78240120e+00 -4.50783014e-01
-3.60166490e-01 -2.05863640e-01 7.74935856e-02 1.01449072e+00
6.50200605e-01 -4.81217384e-01 -4.65915024e-01 -9.72845137e-01
-1.96181938e-01 3.24695289e-01 8.39026809e-01 -8.03612232e-01
-8.39222744e-02 -9.23311412e-01 3.55628252e-01 -2.80118793e-01
6.32635415e-01 -8.73615861e-01 -1.68557540e-01 7.43378520e-01
-6.09230101e-01 -4.46755774e-02 5.60600579e-01 1.10826397e+00
5.82765698e-01 6.50015697e-02 8.16871166e-01 -4.85798717e-01
-7.01658070e-01 5.64373910e-01 -6.41491711e-01 -1.13681657e-02
1.30051136e+00 3.18835527e-01 -3.96165311e-01 -2.71829069e-01
-1.25315678e+00 5.31803310e-01 1.07077301e+00 7.45979607e-01
6.29283309e-01 -1.08021724e+00 -5.68525016e-01 -1.01523072e-01
-1.70533329e-01 2.38757446e-01 1.91946477e-02 1.58851460e-01
-8.06482434e-01 -2.13592798e-01 -4.89020973e-01 -7.62627304e-01
-1.16259289e+00 5.83335757e-01 3.81996781e-01 -3.63286287e-01
-5.73886573e-01 8.75230074e-01 3.19156319e-01 -8.15933883e-01
3.10884953e-01 -7.47591138e-01 2.52033263e-01 -5.06470799e-01
-8.07496682e-02 4.27827537e-01 -7.46821702e-01 -1.23565711e-01
-2.17313796e-01 7.73942530e-01 -7.50226602e-02 -3.21404010e-01
1.35254765e+00 1.86623663e-01 8.47460143e-03 4.65216845e-01
8.75627697e-01 -2.28152588e-01 -2.33623791e+00 1.55816004e-01
-1.18481278e-01 -3.46747220e-01 -4.62947726e-01 -7.30632901e-01
-5.87616563e-01 7.34548807e-01 3.87714326e-01 -2.10263938e-01
3.49033803e-01 2.05724597e-01 5.47542751e-01 1.01392329e+00
6.42783880e-01 -8.22439432e-01 7.80019343e-01 6.97708726e-01
1.33333504e+00 -1.21829903e+00 1.15429588e-01 -3.40339929e-01
-6.71778142e-01 1.08502650e+00 9.21603441e-01 -7.71550417e-01
5.06722391e-01 7.78254330e-01 -1.43074334e-01 -2.41215140e-01
-1.07263350e+00 -1.96042787e-02 1.27276912e-01 7.95474648e-01
-2.87418813e-01 -4.09421232e-03 4.61907029e-01 -1.88125402e-01
3.13342243e-01 1.64318725e-01 4.50900227e-01 1.56092656e+00
-5.33978283e-01 -1.00050735e+00 -2.20000073e-01 1.24163322e-01
4.51662280e-02 3.63258511e-01 -4.26996469e-01 9.75774169e-01
-2.73209780e-01 3.67371053e-01 2.54637897e-02 -5.75097948e-02
2.54290879e-01 -3.66821975e-01 1.15489912e+00 -9.36148345e-01
-3.35837662e-01 -1.06288381e-01 -3.20246480e-02 -7.21520007e-01
-2.97204882e-01 -6.42497897e-01 -1.47745514e+00 3.80198681e-03
1.75768599e-01 -5.97342551e-01 6.41370714e-01 8.86082828e-01
3.49808484e-01 5.55873394e-01 3.91701519e-01 -1.83408487e+00
-9.83981371e-01 -6.42305970e-01 -2.23389819e-01 3.07724237e-01
7.38234103e-01 -8.27608407e-01 -6.19048141e-02 4.56221402e-01] | [4.835115432739258, 0.4424383342266083] |
c4d89199-0029-478b-9c58-bcfd05df75b4 | occlusion-handling-using-semantic | 1707.09603 | null | http://arxiv.org/abs/1707.09603v1 | http://arxiv.org/pdf/1707.09603v1.pdf | Occlusion Handling using Semantic Segmentation and Visibility-Based Rendering for Mixed Reality | Real-time occlusion handling is a major problem in outdoor mixed reality
system because it requires great computational cost mainly due to the
complexity of the scene. Using only segmentation, it is difficult to accurately
render a virtual object occluded by complex objects such as trees, bushes etc.
In this paper, we propose a novel occlusion handling method for real-time,
outdoor, and omni-directional mixed reality system using only the information
from a monocular image sequence. We first present a semantic segmentation
scheme for predicting the amount of visibility for different type of objects in
the scene. We also simultaneously calculate a foreground probability map using
depth estimation derived from optical flow. Finally, we combine the
segmentation result and the probability map to render the computer generated
object and the real scene using a visibility-based rendering method. Our
results show great improvement in handling occlusions compared to existing
blending based methods. | ['Yasuhide Okamoto', 'Taiki Fukiage', 'Takeshi Oishi', 'Menandro Roxas', 'Tomoki Hori'] | 2017-07-30 | null | null | null | null | ['occlusion-handling'] | ['computer-vision'] | [ 3.46474797e-01 -3.39312792e-01 4.94364500e-01 -2.72244334e-01
-2.35461622e-01 -5.54023623e-01 3.05158377e-01 -1.21670976e-01
-2.49526411e-01 9.89258826e-01 -1.94097057e-01 -3.32626581e-01
3.33719701e-02 -1.26297414e+00 -3.70781839e-01 -6.08140826e-01
2.85193890e-01 6.30785406e-01 8.96663368e-01 2.08780840e-02
1.47992983e-01 6.49552226e-01 -1.80369198e+00 -2.26917136e-02
1.16735482e+00 7.52409875e-01 8.51068020e-01 1.06142700e+00
-4.28234816e-01 7.67334759e-01 -5.62156379e-01 6.47238716e-02
4.12922174e-01 -3.48508835e-01 -4.02955949e-01 4.64962363e-01
5.80815613e-01 -5.84499061e-01 -2.69425958e-01 9.01081681e-01
4.07182813e-01 3.57199162e-01 4.46375996e-01 -1.14062095e+00
8.10481384e-02 -1.56180903e-01 -7.06459343e-01 3.74102265e-01
6.80067599e-01 -4.41628732e-02 2.20867738e-01 -4.89484638e-01
8.00074697e-01 1.10027111e+00 1.33966520e-01 3.45260799e-02
-8.89593959e-01 -4.26654547e-01 4.08339381e-01 6.52584955e-02
-1.18256545e+00 -1.33758858e-01 8.17439020e-01 -4.30729032e-01
5.83465397e-01 6.96772516e-01 1.02159369e+00 6.12332821e-02
1.88688979e-01 5.41823149e-01 1.38267899e+00 -2.26951167e-01
2.14666203e-01 2.10569590e-01 3.08370829e-01 7.66680360e-01
2.08933592e-01 1.86471492e-01 2.57906667e-03 -1.15025468e-01
8.93355012e-01 -5.39617501e-02 -4.48520541e-01 -4.72705305e-01
-1.06135798e+00 2.52017707e-01 2.81376123e-01 8.81423727e-02
-3.81423533e-01 5.33440001e-02 -1.16577193e-01 -1.21009268e-01
4.88996416e-01 -1.18114538e-01 -3.92517388e-01 -4.11959141e-02
-1.04209507e+00 3.24535012e-01 7.34441280e-01 7.63771951e-01
8.22537065e-01 1.87237114e-01 2.06281841e-01 5.63930094e-01
6.19623065e-01 6.03124738e-01 -1.84238970e-01 -1.18678510e+00
3.43866467e-01 4.24805820e-01 3.55711311e-01 -1.12283802e+00
-2.02782348e-01 -4.71058130e-01 -4.44210410e-01 8.42969120e-01
4.84128863e-01 1.50227612e-02 -1.14250743e+00 1.11623311e+00
9.91830111e-01 4.99435335e-01 1.56996343e-02 9.53639925e-01
1.05705595e+00 9.26675737e-01 -1.52828887e-01 -4.68349665e-01
1.15669978e+00 -1.08585441e+00 -9.12413597e-01 -2.64484376e-01
2.41418004e-01 -1.15078735e+00 5.59032857e-01 6.97858930e-01
-1.02840137e+00 -5.51511347e-01 -1.08602309e+00 -1.56009421e-01
-2.34822035e-01 7.84991682e-03 8.58078241e-01 9.05886710e-01
-7.71551847e-01 2.37020478e-01 -7.95065582e-01 -1.19954482e-01
9.17507112e-02 3.12073559e-01 -9.10038725e-02 -3.72460663e-01
-5.31106830e-01 7.72019446e-01 2.59229153e-01 2.84220368e-01
-7.13943779e-01 -4.89010274e-01 -7.49562919e-01 -2.22857118e-01
2.91155636e-01 -7.32706726e-01 5.98562062e-01 -8.12854648e-01
-1.47768342e+00 8.04675579e-01 -5.20238996e-01 1.11802831e-01
7.49399900e-01 -1.74633324e-01 -2.10765347e-01 9.35181007e-02
1.16830654e-01 3.06912750e-01 3.88975859e-01 -1.71236324e+00
-9.11489010e-01 -3.98942173e-01 3.81194025e-01 6.00513101e-01
7.46372521e-01 -1.47170909e-02 -4.12039906e-01 -1.34364933e-01
7.42228866e-01 -6.19866133e-01 -5.24667799e-01 3.41286361e-01
-3.64028662e-01 2.48661146e-01 1.20303273e+00 -9.80628490e-01
8.07407379e-01 -2.20984674e+00 -3.55942585e-02 5.91787770e-02
2.56705314e-01 1.90329313e-01 2.89914995e-01 -9.85328779e-02
2.90629834e-01 -3.01192135e-01 -3.38526696e-01 -4.00233299e-01
-4.79305476e-01 4.09549654e-01 -2.25117832e-01 4.88660038e-01
-5.18137455e-01 3.08306128e-01 -1.08913791e+00 -9.26113665e-01
9.53275859e-01 6.40070915e-01 -4.50226873e-01 3.78346503e-01
-2.31231362e-01 8.17590892e-01 -2.66662985e-01 6.91348732e-01
1.36163259e+00 3.64599615e-01 -9.78862494e-02 -1.10421054e-01
-5.20318329e-01 1.41200826e-01 -1.78833008e+00 1.35351133e+00
-6.59316540e-01 6.75056577e-01 3.80076706e-01 -4.47650582e-01
8.41807663e-01 2.00136364e-01 4.15509939e-01 -4.86134678e-01
2.00158671e-01 2.22968906e-01 -3.17857951e-01 -4.29755867e-01
9.03560460e-01 -3.34846191e-02 5.29405951e-01 1.65703952e-01
-4.45473671e-01 -6.38066232e-01 2.04075918e-01 1.53716266e-01
7.02139199e-01 6.24490380e-01 1.04781695e-01 -5.87515794e-02
3.26535285e-01 2.14227706e-01 8.26715648e-01 5.83061278e-01
-2.35761389e-01 8.46690655e-01 2.31050346e-02 -3.83114457e-01
-7.35967278e-01 -1.24522769e+00 -2.56707817e-01 4.36073184e-01
8.76890540e-01 1.42236752e-02 -3.95280391e-01 -3.90652306e-02
-2.91834980e-01 6.48245513e-01 -1.28989071e-01 5.61994076e-01
-7.20461309e-01 -1.00203264e+00 -3.81411374e-01 6.78511336e-02
9.02036428e-01 -6.66360557e-01 -7.67950535e-01 3.78073990e-01
-4.48976636e-01 -1.34989607e+00 -5.05537959e-03 -1.17434107e-01
-1.20778859e+00 -9.61759925e-01 -4.92095947e-01 -6.87940776e-01
5.54882765e-01 8.80347669e-01 1.02824283e+00 1.92853704e-01
-5.87317109e-01 1.25427380e-01 3.69493589e-02 -1.23199217e-01
-1.05706148e-01 -6.33498728e-01 -5.23376584e-01 -2.99412869e-02
-3.03622842e-01 -6.55935943e-01 -6.82876050e-01 5.06395578e-01
-6.81948125e-01 6.04725301e-01 -1.35505334e-01 1.74313083e-01
6.76025152e-01 5.29425383e-01 -3.08890522e-01 -7.20098734e-01
-6.84381723e-02 -3.21373075e-01 -1.05648220e+00 4.34005111e-02
9.64516997e-02 -6.14753842e-01 1.05927221e-01 -3.49463433e-01
-1.55959857e+00 2.74745852e-01 3.21869589e-02 -8.78332630e-02
-3.22249860e-01 7.45336860e-02 -4.96972263e-01 -2.43323714e-01
1.13346249e-01 9.30305198e-02 -5.16517043e-01 -4.33280528e-01
2.56893992e-01 5.96681535e-01 5.50452650e-01 -3.65022749e-01
6.93992436e-01 1.08607054e+00 1.44501179e-01 -1.15925777e+00
-5.54758608e-01 -6.14886701e-01 -7.37043977e-01 -5.08262753e-01
1.22717905e+00 -8.61989677e-01 -5.35565853e-01 5.25145710e-01
-1.41701925e+00 -3.78387958e-01 -2.91074701e-02 7.66942263e-01
-3.46761703e-01 6.75624728e-01 -4.42336440e-01 -1.28346729e+00
1.90394238e-01 -1.29798388e+00 1.08187938e+00 4.13447142e-01
3.84502351e-01 -9.14941669e-01 3.20359506e-02 7.26826012e-01
1.34976342e-01 5.90652704e-01 4.77725863e-01 4.81660217e-01
-1.58468473e+00 2.42800843e-02 -6.11439168e-01 4.73823547e-02
1.45306662e-01 4.60702121e-01 -1.02429545e+00 1.92942575e-01
2.44184896e-01 3.69961888e-01 5.88121295e-01 7.37880707e-01
8.30573499e-01 2.26973072e-01 -4.20824677e-01 8.33210945e-01
1.84384322e+00 5.90401471e-01 9.70360219e-01 1.63309172e-01
1.05753779e+00 5.77036858e-01 1.04869902e+00 2.77183264e-01
4.95560259e-01 8.56301010e-01 4.95375544e-01 -3.13859522e-01
-4.54595625e-01 3.48474868e-02 -1.70017332e-01 8.18998873e-01
-2.48484895e-01 -5.81554711e-01 -6.15333974e-01 4.75118726e-01
-1.69681478e+00 -8.44642580e-01 -1.07323182e+00 2.41886210e+00
3.03201586e-01 1.70501489e-02 -3.15645814e-01 3.55283022e-01
7.56188929e-01 8.51780772e-02 -2.35248692e-02 -3.32186311e-01
-3.19397658e-01 -1.85134995e-03 5.56406915e-01 1.29486394e+00
-9.34799969e-01 9.08673584e-01 6.44376230e+00 6.59629583e-01
-9.63925004e-01 2.98846215e-01 3.69510442e-01 1.41558424e-01
-4.32169676e-01 4.76683617e-01 -6.88347697e-01 4.76534396e-01
3.04691136e-01 1.67679787e-01 2.85260201e-01 5.19340575e-01
3.09932351e-01 -1.21005869e+00 -4.37320203e-01 9.72964168e-01
1.50570329e-02 -8.64171803e-01 -2.89815217e-01 3.11531484e-01
8.47049236e-01 -1.42896712e-01 -3.98040771e-01 -2.59315372e-01
3.84881496e-01 -6.10659838e-01 6.67007387e-01 4.47979331e-01
1.20642409e-01 -2.18248978e-01 4.82585967e-01 4.46182072e-01
-1.39431953e+00 3.80482227e-01 -4.07546401e-01 -4.54663366e-01
8.34014058e-01 9.92573380e-01 -4.36222136e-01 6.02216721e-01
3.00972879e-01 2.65493393e-01 -2.56889373e-01 1.60597742e+00
-3.91787477e-02 1.38845772e-01 -6.83675110e-01 2.95059323e-01
-1.00792293e-02 -7.46675134e-01 7.50706315e-01 7.65175283e-01
1.39318064e-01 3.92345697e-01 4.61746037e-01 7.27763951e-01
6.75140977e-01 1.15584701e-01 -5.79641283e-01 5.35177648e-01
3.22417496e-03 1.12859523e+00 -1.23493516e+00 -6.18002355e-01
-2.08342105e-01 1.09475136e+00 -2.02130035e-01 5.30227423e-01
-1.01030409e+00 -1.30018637e-01 6.32826388e-02 2.01020539e-01
-8.87862667e-02 -7.13921905e-01 -4.10183161e-01 -1.30378616e+00
4.74099182e-02 -2.01864138e-01 -8.76186416e-02 -1.37106991e+00
-3.48995984e-01 6.83664501e-01 4.32398766e-02 -1.13451946e+00
3.34555775e-01 -3.07049304e-01 -4.25275564e-01 1.00796533e+00
-1.56894314e+00 -9.28402722e-01 -8.81900966e-01 2.87889540e-01
7.38900900e-01 4.21959698e-01 4.52108443e-01 7.26492763e-01
-3.54038417e-01 -4.23503339e-01 -4.29523438e-02 -3.62690717e-01
1.03383832e-01 -1.07657635e+00 1.91432182e-02 1.14093590e+00
-4.18997072e-02 8.10229704e-02 9.80371058e-01 -9.25861359e-01
-9.92302835e-01 -6.68124616e-01 7.63624609e-01 -2.98528284e-01
9.66240615e-02 -3.01008314e-01 -7.12511718e-01 4.28277224e-01
-2.83297244e-02 1.65522814e-01 2.15183318e-01 -3.73634756e-01
3.20477813e-01 6.98523596e-02 -1.47526526e+00 5.19701302e-01
1.10150969e+00 -1.54229507e-01 -2.58320779e-01 5.06903529e-01
7.35021532e-01 -8.93132329e-01 -3.63461584e-01 4.97797102e-01
6.29215240e-01 -1.26201999e+00 1.19652307e+00 1.56437874e-01
1.47260293e-01 -8.90085757e-01 -4.67534006e-01 -7.75791466e-01
-1.59400608e-02 -2.95511812e-01 2.45061904e-01 9.35093403e-01
-8.31610039e-02 -7.99318790e-01 8.27276886e-01 7.22740233e-01
-1.10830821e-01 -2.68210679e-01 -9.13318336e-01 -5.45469582e-01
-7.87134528e-01 -3.52414787e-01 3.83167922e-01 8.73687208e-01
-8.11293840e-01 -9.82801989e-02 -2.75519550e-01 7.06606746e-01
9.73396599e-01 4.24925566e-01 1.02914119e+00 -1.33488441e+00
-3.71431500e-01 9.40491632e-02 -6.29558146e-01 -1.24529016e+00
-2.88824737e-01 -3.13924193e-01 6.23265654e-02 -2.07942319e+00
1.36330023e-01 -1.08780932e+00 3.17397058e-01 -3.58260989e-01
-2.23263234e-01 5.55483222e-01 1.98393956e-01 9.40477848e-02
-3.47445130e-01 3.79798323e-01 1.62106001e+00 1.60008550e-01
-4.74934429e-01 1.36950538e-01 1.17277563e-01 8.85101080e-01
6.40318632e-01 -5.26978254e-01 -5.02222538e-01 -6.82353079e-01
7.61111313e-03 5.48520803e-01 5.71028292e-01 -1.12154746e+00
-1.21701993e-01 -4.89641190e-01 2.58775622e-01 -1.14956999e+00
9.02825832e-01 -1.00304532e+00 6.95496619e-01 4.29120421e-01
4.61774737e-01 -2.71309704e-01 8.39761421e-02 5.32476544e-01
3.78463008e-02 -3.89861882e-01 7.27438629e-01 -4.06282574e-01
-6.19404972e-01 1.62052065e-01 -3.11377019e-01 -2.69859731e-01
1.13399267e+00 -7.18310058e-01 -2.94294149e-01 -2.88607836e-01
-9.11177039e-01 7.05456315e-03 6.10390365e-01 6.63789734e-02
6.52539253e-01 -9.23619509e-01 -3.83303016e-01 9.02006254e-02
-4.00680363e-01 3.06356549e-01 5.23362696e-01 5.73804796e-01
-1.42109203e+00 1.25373855e-01 -1.83578670e-01 -7.94539571e-01
-1.56343651e+00 4.24227268e-01 4.10234094e-01 2.13800110e-02
-6.97978139e-01 6.96980953e-01 6.18542254e-01 -4.43154156e-01
-1.16832890e-02 -5.06576478e-01 -1.13619857e-01 -3.25496763e-01
3.64473760e-01 7.66555846e-01 -1.68992847e-01 -7.90932357e-01
-2.14389473e-01 9.77863550e-01 5.50226986e-01 -5.00648022e-01
9.11259294e-01 -5.99363506e-01 -2.53955424e-01 5.64627886e-01
8.68372142e-01 3.63797724e-01 -1.17421675e+00 9.83731356e-03
-6.29918098e-01 -1.19833386e+00 5.27574420e-01 -6.02172613e-01
-1.11206663e+00 8.72232318e-01 8.97224188e-01 -1.56411286e-02
1.01329255e+00 -3.55148584e-01 6.94513500e-01 -3.11028957e-01
8.32524717e-01 -8.10078740e-01 -4.48743790e-01 8.04136172e-02
4.45737451e-01 -1.09639001e+00 4.59121287e-01 -1.40065300e+00
-2.50175565e-01 7.91633844e-01 6.31130815e-01 -2.14496702e-01
8.45883667e-01 3.07055861e-01 2.12469473e-01 -9.74680707e-02
-2.35202461e-01 -4.15301532e-01 -8.24887976e-02 7.62272894e-01
1.62139207e-01 2.25124300e-01 -3.47910047e-01 -3.44458222e-01
-5.58362752e-02 6.51964694e-02 9.33310986e-01 1.10626113e+00
-6.96155965e-01 -9.38561499e-01 -8.14490736e-01 2.26262718e-01
-2.06252277e-01 7.57288262e-02 2.63745755e-01 4.89592075e-01
4.76555258e-01 1.12826276e+00 1.70894489e-01 1.13475844e-01
3.19375768e-02 -4.99779999e-01 7.75473356e-01 -6.23697579e-01
-1.67223066e-01 2.84272760e-01 1.40269190e-01 -5.48226774e-01
-6.75193250e-01 -5.69918096e-01 -1.11460888e+00 -1.42931670e-01
-6.06727302e-01 -7.09333792e-02 1.15524244e+00 7.17659771e-01
-7.63605163e-02 7.08159685e-01 3.51698071e-01 -1.00236678e+00
7.01306641e-01 -4.97114629e-01 -7.45719790e-01 -7.80289769e-02
3.74594569e-01 -8.52553129e-01 -5.28517544e-01 3.05496212e-02] | [9.16026782989502, -2.4411818981170654] |
5f19aa87-6012-4858-82fd-fbcdd595749c | fish-sounds-towards-the-evaluation-of-marine | 2201.05013 | null | https://arxiv.org/abs/2201.05013v2 | https://arxiv.org/pdf/2201.05013v2.pdf | Fish sounds: towards the evaluation of marine acoustic biodiversity through data-driven audio source separation | The marine ecosystem is changing at an alarming rate, exhibiting biodiversity loss and the migration of tropical species to temperate basins. Monitoring the underwater environments and their inhabitants is of fundamental importance to understand the evolution of these systems and implement safeguard policies. However, assessing and tracking biodiversity is often a complex task, especially in large and uncontrolled environments, such as the oceans. One of the most popular and effective methods for monitoring marine biodiversity is passive acoustics monitoring (PAM), which employs hydrophones to capture underwater sound. Many aquatic animals produce sounds characteristic of their own species; these signals travel efficiently underwater and can be detected even at great distances. Furthermore, modern technologies are becoming more and more convenient and precise, allowing for very accurate and careful data acquisition. To date, audio captured with PAM devices is frequently manually processed by marine biologists and interpreted with traditional signal processing techniques for the detection of animal vocalizations. This is a challenging task, as PAM recordings are often over long periods of time. Moreover, one of the causes of biodiversity loss is sound pollution; in data obtained from regions with loud anthropic noise, it is hard to separate the artificial from the fish sound manually. Nowadays, machine learning and, in particular, deep learning represents the state of the art for processing audio signals. Specifically, sound separation networks are able to identify and separate human voices and musical instruments. In this work, we show that the same techniques can be successfully used to automatically extract fish vocalizations in PAM recordings, opening up the possibility for biodiversity monitoring at a large scale. | ['Silvia Zuffi', 'Emanuele Rodolà', 'Nicola Zonca', 'Michele Mancusi'] | 2022-01-13 | null | null | null | null | ['audio-source-separation'] | ['audio'] | [ 1.53200299e-01 -5.11937559e-01 6.15396261e-01 1.86290637e-01
-8.87751207e-02 -7.26221919e-01 2.18604207e-01 6.13542974e-01
-9.44261968e-01 5.81347644e-01 -6.21637590e-02 3.17592174e-02
-7.11687356e-02 -9.37050104e-01 -4.55507487e-01 -1.00008047e+00
-5.92906892e-01 1.82847187e-01 3.69191885e-01 -3.72602642e-01
5.54789342e-02 8.27528358e-01 -2.12523150e+00 -2.69372076e-01
6.82406604e-01 8.03391516e-01 5.97537041e-01 8.23385239e-01
-3.67939830e-01 -5.82070611e-02 -7.51641572e-01 -3.30021679e-01
8.60535353e-02 -4.37599003e-01 -2.60458946e-01 -3.81621242e-01
1.05535060e-01 -2.64413238e-01 3.45229477e-01 1.26951313e+00
5.59169769e-01 3.07387710e-01 4.37788993e-01 -3.98813367e-01
1.07309923e-01 7.85258234e-01 -2.52829045e-01 9.28185210e-02
-5.56785688e-02 1.82412222e-01 8.97519231e-01 -5.16532004e-01
-4.16651145e-02 1.07325971e+00 6.92277193e-01 4.73791093e-01
-1.21020985e+00 -7.56455183e-01 -1.57416731e-01 2.84326524e-01
-1.17192614e+00 -5.53172827e-01 6.32645845e-01 -7.39192665e-01
4.31676626e-01 3.98940921e-01 1.23554528e+00 7.34497547e-01
7.72892013e-02 2.98048615e-01 8.38986099e-01 -9.03068930e-02
3.48165631e-01 -7.79361427e-02 -3.66213977e-01 1.19317792e-01
4.16175336e-01 -4.39979583e-02 -4.58923727e-01 -2.35806480e-01
4.20463532e-01 2.61059523e-01 -7.86228001e-01 3.46173465e-01
-9.04246271e-01 5.29970884e-01 3.34244996e-01 4.80155587e-01
-3.10592502e-01 3.78867164e-02 3.31331015e-01 1.15931764e-01
3.22160184e-01 6.83695734e-01 -4.48698223e-01 -5.41770399e-01
-9.55875337e-01 -8.70857090e-02 9.39953268e-01 6.69985041e-02
5.91768682e-01 2.58839011e-01 9.28374946e-01 1.15641224e+00
2.99915850e-01 8.98740053e-01 6.40756488e-01 -6.48571372e-01
-1.59191862e-01 3.49201143e-01 -5.01499660e-02 -1.14783633e+00
-6.27072632e-01 -5.41113436e-01 -1.03098702e+00 3.23346883e-01
4.97982025e-01 -1.27908841e-01 -2.25802362e-01 1.46918643e+00
2.63137221e-01 -8.44575316e-02 1.54985756e-01 9.02100205e-01
6.51614547e-01 7.65195549e-01 4.02251445e-02 -3.35687280e-01
1.47978306e+00 -1.71612740e-01 -4.13181901e-01 -1.96820110e-01
9.33914781e-02 -4.87334430e-01 9.46209550e-01 6.57487690e-01
-5.10835588e-01 -3.78063828e-01 -8.17267537e-01 5.43737948e-01
-4.29910988e-01 -2.08228111e-01 3.73494536e-01 5.61220646e-01
-5.28023005e-01 7.92241395e-01 -1.11798060e+00 -2.68691808e-01
2.01944709e-01 1.21275358e-01 -5.87224066e-01 3.87127787e-01
-9.97720003e-01 4.83282387e-01 -5.54372706e-02 5.96465647e-01
-9.33131993e-01 -5.43835402e-01 -7.46077299e-01 3.04336607e-01
-3.44526954e-02 9.82399136e-02 1.07924283e+00 -9.49899554e-01
-1.57839465e+00 6.71519279e-01 2.95648485e-01 -4.62778330e-01
3.74719650e-01 -4.99292612e-01 -2.79980272e-01 1.95401818e-01
-1.34515941e-01 1.58277825e-01 7.70976782e-01 -9.70014691e-01
-6.88365698e-01 -5.28210402e-01 -4.25588429e-01 -1.32716268e-01
-7.28668630e-01 1.75272953e-02 3.55221301e-01 -4.55029398e-01
3.91631797e-02 -6.65717244e-01 -2.75374085e-01 2.59312898e-01
3.38210724e-02 8.71096179e-02 4.80435193e-01 -6.39380395e-01
8.16208422e-01 -2.28281426e+00 2.71249473e-01 -5.11437580e-02
-1.19925123e-02 4.16892201e-01 6.83396459e-02 5.84140837e-01
4.98501301e-01 1.70128584e-01 -6.27861321e-01 -2.40123302e-01
-3.44927520e-01 2.79426754e-01 -1.24271013e-01 6.69184327e-01
-1.30145445e-01 -9.34214815e-02 -1.06032085e+00 -2.21741021e-01
4.77618575e-02 6.26659453e-01 -4.60719705e-01 3.09296966e-01
1.42513318e-02 7.98379362e-01 -2.69455127e-02 7.08010554e-01
4.86125290e-01 5.54801643e-01 6.43318295e-02 1.93800732e-01
-9.05621409e-01 1.14965066e-02 -1.03492010e+00 1.17505002e+00
-6.70994222e-01 8.42640877e-01 7.98901498e-01 -9.68499064e-01
1.17225492e+00 1.86238408e-01 2.97500998e-01 -3.18308234e-01
2.68128872e-01 5.50253332e-01 1.79779768e-01 -8.69584382e-01
4.88538384e-01 -7.24879742e-01 9.36038867e-02 -4.68586385e-02
-1.94622234e-01 -1.52917579e-01 -6.71915039e-02 -6.08594179e-01
6.79463387e-01 -4.02583808e-01 3.15476179e-01 -2.83669561e-01
5.44505417e-01 -2.57073551e-01 7.18150854e-01 3.62773478e-01
-1.75967485e-01 2.90106565e-01 1.17916510e-01 -4.53230292e-01
-5.01194000e-01 -7.37713099e-01 -5.31571686e-01 1.14755940e+00
5.81910498e-02 1.74501956e-01 -4.54363376e-01 1.39947310e-01
9.16809961e-02 1.28230706e-01 -3.81943107e-01 -1.04068793e-01
-4.73124117e-01 -7.11842537e-01 8.38627756e-01 2.06572458e-01
4.14005697e-01 -1.37418950e+00 -1.12545323e+00 5.80364347e-01
-1.25314370e-01 -9.41037476e-01 2.67124861e-01 3.82185847e-01
-8.26664090e-01 -8.09065342e-01 -8.25795770e-01 -6.30195141e-01
2.26276115e-01 1.78900674e-01 7.48063743e-01 2.41230875e-01
-4.49016988e-01 -8.70331302e-02 -5.33478498e-01 -6.39550030e-01
-5.70578456e-01 -7.74960741e-02 5.03850996e-01 2.38949299e-01
-1.02629494e-02 -1.00632966e+00 -5.53521276e-01 3.50681752e-01
-9.01162386e-01 -4.93299097e-01 6.26467586e-01 6.54245496e-01
3.61700326e-01 1.53466627e-01 6.57238841e-01 -2.82545656e-01
8.60818774e-02 -4.86115545e-01 -8.00409913e-01 -2.11631358e-01
3.56119782e-01 -3.28715742e-01 1.05202866e+00 -4.23053324e-01
-4.89984542e-01 7.57103562e-02 -7.19922900e-01 1.90873757e-01
-6.27783835e-01 6.62895679e-01 -2.08910197e-01 -1.62471458e-01
6.47620082e-01 4.40805912e-01 9.84259024e-02 -9.36040342e-01
-3.22569907e-01 1.05224490e+00 5.77063859e-01 -2.17636943e-01
6.36755764e-01 6.57780170e-01 4.57110740e-02 -1.97388363e+00
-4.21409667e-01 -5.56187332e-01 -4.26633388e-01 -5.29519796e-01
8.26577544e-01 -6.00990891e-01 -9.74053383e-01 8.22440207e-01
-8.09843481e-01 -2.74560183e-01 -1.41820282e-01 8.61694813e-01
1.17327482e-01 4.71996874e-01 -4.05301452e-01 -1.26430821e+00
-4.63875234e-01 -1.14909303e+00 8.29084694e-01 5.16716003e-01
5.07679172e-02 -6.93767309e-01 4.73643512e-01 -2.71348003e-02
6.07983947e-01 2.65555978e-01 4.94189590e-01 -4.67851847e-01
1.71335414e-02 -2.55166441e-01 3.09893042e-01 4.47733402e-01
4.96216297e-01 4.49151337e-01 -1.01587903e+00 -1.28387213e-01
1.70319602e-01 -6.90861568e-02 8.88977230e-01 1.89650208e-01
6.83752596e-01 -1.36991709e-01 1.36822075e-01 6.06197953e-01
1.16413081e+00 2.47857735e-01 3.02608192e-01 3.81101161e-01
2.96850920e-01 1.14882302e+00 2.58753330e-01 4.77758527e-01
-2.46241246e-03 5.35809457e-01 9.59621489e-01 2.14729950e-01
3.44721049e-01 1.56497002e-01 5.27890027e-01 9.10605550e-01
-3.40007514e-01 4.38673981e-02 -9.66689825e-01 8.41904223e-01
-1.18706656e+00 -1.05792224e+00 -4.04212654e-01 2.56563544e+00
7.73168862e-01 -1.94052249e-01 2.09332377e-01 4.59237039e-01
5.35982490e-01 -2.23896056e-01 -1.90964356e-01 -2.16286018e-01
-1.67768553e-01 1.77788541e-01 4.24138367e-01 3.95771384e-01
-9.67031598e-01 5.75298488e-01 5.08185196e+00 9.64737833e-02
-1.54503655e+00 -2.40366265e-01 -3.40976059e-01 3.48047502e-02
-4.84709218e-02 -2.49672011e-01 -6.57660782e-01 5.32521248e-01
9.30762291e-01 3.51715297e-01 3.40765029e-01 7.84764528e-01
5.02699435e-01 -1.93935961e-01 -5.52784681e-01 7.89421141e-01
-1.66488647e-01 -7.68287361e-01 -9.13869366e-02 2.32263327e-01
9.35528129e-02 1.15903236e-01 -3.20263833e-01 -8.37678984e-02
-1.97640955e-01 -7.55614519e-01 7.14856863e-01 5.45480788e-01
4.54036206e-01 -8.28681946e-01 9.28263068e-01 6.08460426e-01
-1.33758688e+00 -2.41680548e-01 -6.92550778e-01 -4.91120666e-01
3.45751375e-01 8.52705002e-01 -4.07530099e-01 1.68020442e-01
1.22444510e+00 6.57116354e-01 -3.98267239e-01 1.69216835e+00
-3.10604721e-01 7.77279973e-01 -6.32149756e-01 -4.13296729e-01
-4.17176224e-02 -4.86701250e-01 9.71101999e-01 1.30240953e+00
5.79214633e-01 -2.06878468e-01 -1.69948906e-01 5.00879765e-01
-7.00934231e-03 2.63594568e-01 -4.49959517e-01 -3.23637486e-01
2.93658346e-01 1.28101778e+00 -8.25322270e-01 -4.39589955e-02
-4.80809584e-02 4.10135448e-01 -2.14736342e-01 -1.68169186e-01
-3.65925878e-01 -5.57953298e-01 1.21054852e+00 -1.17950216e-01
6.15839586e-02 -6.16983771e-01 1.28548279e-01 -9.04534042e-01
-3.03401381e-01 -5.99340856e-01 8.20546374e-02 -4.10237089e-02
-1.22757947e+00 6.17383540e-01 -3.98642927e-01 -1.56398916e+00
1.92445636e-01 -4.91360962e-01 -6.10932231e-01 5.89130163e-01
-1.70085788e+00 -3.89046520e-01 -6.58616841e-01 1.74549282e-01
4.30314392e-01 2.19038621e-01 9.83799696e-01 4.96794671e-01
-4.19407725e-01 1.21442840e-01 5.68646014e-01 2.15360567e-01
4.77416396e-01 -1.10203648e+00 -5.09305932e-02 6.68505192e-01
2.57526755e-01 3.20143998e-01 1.02167523e+00 -4.11980838e-01
-1.37044680e+00 -8.28710914e-01 4.18967038e-01 3.43144298e-01
1.02485120e+00 -4.38030571e-01 -1.28019333e+00 -5.81885464e-02
-2.77201205e-01 -2.70343184e-01 1.23225999e+00 -2.63491362e-01
5.09972349e-02 -5.36456466e-01 -8.49617898e-01 3.51072401e-01
5.35926163e-01 -2.44435444e-01 -5.55789888e-01 1.71214968e-01
2.04916015e-01 1.09816305e-01 -7.89583683e-01 3.34386021e-01
1.08845699e+00 -9.17551935e-01 6.29363596e-01 -2.40397558e-01
2.01553121e-01 -5.72537065e-01 -5.50023355e-02 -1.40570474e+00
1.15292445e-01 -2.99695969e-01 6.55621111e-01 1.41544437e+00
1.69381782e-01 -7.94043720e-01 3.93930405e-01 -2.69339204e-01
-2.47561812e-01 1.39170170e-01 -1.13309801e+00 -9.45079327e-01
7.63162300e-02 -3.20144594e-01 2.99349815e-01 6.18399739e-01
-1.10099323e-01 1.65723357e-02 -4.48849797e-01 5.86496115e-01
8.10725272e-01 1.54160842e-01 8.45924914e-01 -1.95357680e+00
-4.75991249e-01 -7.94594347e-01 -8.25483203e-01 -5.29095590e-01
-6.23248927e-02 -2.51266778e-01 6.32610977e-01 -1.30919540e+00
-5.52627683e-01 -5.38398214e-02 1.15662992e-01 2.85147756e-01
2.75453590e-02 4.84901637e-01 -2.00438555e-02 3.18203807e-01
2.79408246e-01 8.29323232e-01 8.68233919e-01 -6.97767884e-02
-4.07389551e-01 3.41796130e-01 -2.33229905e-01 9.97205019e-01
8.16320240e-01 -6.82295144e-01 3.34782839e-01 -3.94175619e-01
4.48421061e-01 -1.55679509e-01 2.11567238e-01 -1.36688602e+00
3.14155221e-01 -4.27345373e-02 -1.95787951e-01 -1.24993376e-01
3.60375762e-01 -1.16219437e+00 3.15566540e-01 9.31728959e-01
1.12298802e-01 -3.87999773e-01 2.80040860e-01 6.92675710e-01
-5.42239249e-01 -4.55644816e-01 1.13508964e+00 -1.59301862e-01
-4.87126738e-01 -1.82974502e-01 -1.07992780e+00 -1.79667562e-01
5.32153308e-01 -3.10858432e-03 -3.52959484e-01 -2.32401460e-01
-5.20358860e-01 1.24713913e-01 4.32227761e-01 1.71930626e-01
6.02998376e-01 -4.29998696e-01 -8.79682660e-01 1.59339949e-01
2.87368912e-02 2.23747224e-01 5.32311916e-01 9.40504849e-01
-1.12631917e+00 -4.31217790e-01 -3.77290815e-01 -7.13325918e-01
-1.37304497e+00 9.86451283e-02 5.15989840e-01 4.43372905e-01
-7.62001157e-01 8.94189596e-01 1.03820018e-01 -8.85096565e-02
2.64550209e-01 -3.59670818e-01 -7.92662978e-01 6.28619373e-01
9.81282234e-01 4.40000564e-01 -6.88005090e-02 -7.11645484e-01
-3.80317628e-01 9.18294311e-01 5.32953858e-01 1.25200063e-01
1.74671113e+00 -9.28760506e-03 -4.92029428e-01 8.07044268e-01
9.62050676e-01 4.65277731e-01 -1.14423275e+00 2.22407550e-01
-1.70053929e-01 -3.78857821e-01 -1.01994999e-01 -2.46427625e-01
-1.06584024e+00 1.35180938e+00 4.48446572e-01 8.97610068e-01
1.15983009e+00 -4.85249348e-02 6.43297970e-01 5.51637411e-01
5.22417545e-01 -6.85507476e-01 -3.50404322e-01 3.96532714e-01
8.85110319e-01 -1.01557982e+00 -2.89115846e-01 9.66535732e-02
-1.31815627e-01 1.42922628e+00 4.75845784e-02 4.28075567e-02
7.01109052e-01 4.59531277e-01 2.97549903e-01 1.21261880e-01
-1.68779522e-01 -4.16334778e-01 -2.50361204e-01 5.82776546e-01
1.03232272e-01 1.83171287e-01 -2.58623421e-01 6.07611954e-01
-6.37287915e-01 -6.85526967e-01 6.82244778e-01 7.24776387e-01
-9.61087227e-01 -7.79023886e-01 -6.67715192e-01 1.71301484e-01
-7.24636376e-01 -3.86886261e-02 -5.03119588e-01 3.34067285e-01
-6.65890276e-02 9.87181067e-01 -1.02565579e-01 -4.25323397e-01
5.54094136e-01 -6.32695407e-02 -1.46492109e-01 -4.77026016e-01
-7.57087648e-01 1.38555914e-01 -2.41654173e-01 8.20854753e-02
-6.93107307e-01 -8.05588186e-01 -1.26916993e+00 -1.13173407e-02
-5.14486611e-01 8.14509511e-01 1.17066181e+00 7.36472487e-01
-1.59173951e-01 4.55223143e-01 8.86336148e-01 -1.39446151e+00
-2.59727895e-01 -1.09330976e+00 -1.00139213e+00 9.30847484e-04
5.03731549e-01 -6.93776667e-01 -1.10624933e+00 -4.95034866e-02] | [8.567749977111816, -1.18271803855896] |
b69ed3ea-073f-4d0f-a159-c76bac17b514 | the-drunkard-s-odometry-estimating-camera | 2306.16917 | null | https://arxiv.org/abs/2306.16917v1 | https://arxiv.org/pdf/2306.16917v1.pdf | The Drunkard's Odometry: Estimating Camera Motion in Deforming Scenes | Estimating camera motion in deformable scenes poses a complex and open research challenge. Most existing non-rigid structure from motion techniques assume to observe also static scene parts besides deforming scene parts in order to establish an anchoring reference. However, this assumption does not hold true in certain relevant application cases such as endoscopies. Deformable odometry and SLAM pipelines, which tackle the most challenging scenario of exploratory trajectories, suffer from a lack of robustness and proper quantitative evaluation methodologies. To tackle this issue with a common benchmark, we introduce the Drunkard's Dataset, a challenging collection of synthetic data targeting visual navigation and reconstruction in deformable environments. This dataset is the first large set of exploratory camera trajectories with ground truth inside 3D scenes where every surface exhibits non-rigid deformations over time. Simulations in realistic 3D buildings lets us obtain a vast amount of data and ground truth labels, including camera poses, RGB images and depth, optical flow and normal maps at high resolution and quality. We further present a novel deformable odometry method, dubbed the Drunkard's Odometry, which decomposes optical flow estimates into rigid-body camera motion and non-rigid scene deformations. In order to validate our data, our work contains an evaluation of several baselines as well as a novel tracking error metric which does not require ground truth data. Dataset and code: https://davidrecasens.github.io/TheDrunkard'sOdometry/ | ['Javier Civera', 'Marc Pollefeys', 'Martin R. Oswald', 'David Recasens'] | 2023-06-29 | null | null | null | null | ['optical-flow-estimation', '6d-pose-estimation-using-rgbd', 'visual-navigation'] | ['computer-vision', 'computer-vision', 'robots'] | [-4.20009643e-02 -1.23269215e-01 6.53473511e-02 -2.06835270e-01
-5.66731691e-01 -1.03799546e+00 4.45439219e-01 -2.97978818e-01
-3.65680248e-01 6.18633091e-01 1.11012608e-01 7.83309042e-02
-5.98522089e-02 -5.94378710e-01 -8.84165525e-01 -7.85918534e-01
8.43758956e-02 6.22761071e-01 4.42310005e-01 -2.62785137e-01
4.36961204e-02 4.72412765e-01 -1.19712162e+00 -2.48592824e-01
7.74797022e-01 4.28958416e-01 2.77152479e-01 8.20576787e-01
4.66883153e-01 4.87682462e-01 7.70029612e-03 -3.53462875e-01
6.28894508e-01 -1.30270898e-01 -8.46782804e-01 1.32389158e-01
7.92602599e-01 -4.99545485e-01 -4.99564201e-01 1.20000386e+00
5.36245584e-01 4.36927527e-01 3.25223833e-01 -1.05813265e+00
-2.67288506e-01 -1.21247970e-01 -2.10881203e-01 8.42388570e-02
8.44893217e-01 3.66182774e-01 4.84785527e-01 -7.16942608e-01
1.25172091e+00 1.12823308e+00 8.23492408e-01 7.29094923e-01
-1.06647909e+00 -2.17856303e-01 1.01899780e-01 -2.46531546e-01
-1.22380960e+00 -2.42621720e-01 6.60971701e-01 -8.60785186e-01
5.00650585e-01 3.05251658e-01 7.43045509e-01 1.38977802e+00
2.55350202e-01 4.60823536e-01 8.64372373e-01 -4.49133962e-02
1.19631819e-01 -1.58815533e-02 -4.31350768e-02 1.00464857e+00
5.40795743e-01 1.18971691e-01 -2.82724410e-01 8.25515240e-02
1.12864482e+00 2.47895300e-01 -7.59173751e-01 -1.04217529e+00
-1.74729049e+00 6.01590395e-01 5.20858228e-01 8.61025006e-02
-1.07421733e-01 2.17934340e-01 6.90223947e-02 1.68922901e-01
1.70876697e-01 1.98107615e-01 -3.09408695e-01 -1.97774619e-01
-6.13908529e-01 3.87778342e-01 9.01539147e-01 1.15050900e+00
9.02387142e-01 -9.63330194e-02 2.33630419e-01 6.57990947e-02
4.96641189e-01 6.70160353e-01 5.24751663e-01 -1.16198695e+00
4.76158679e-01 5.23682237e-01 4.93012071e-01 -1.26528287e+00
-4.60430652e-01 -1.44849434e-01 -7.18140483e-01 1.71496779e-01
6.23757720e-01 4.56969924e-02 -8.96524131e-01 1.57827377e+00
7.87428021e-01 4.85210329e-01 9.03210193e-02 1.30928481e+00
8.83608818e-01 2.50210524e-01 -5.03837705e-01 5.60942627e-02
1.06039846e+00 -1.06091046e+00 -5.35663188e-01 -3.64327654e-02
6.39783382e-01 -7.83799291e-01 1.04171276e+00 3.34453642e-01
-9.74581897e-01 -4.62010801e-01 -8.49080920e-01 -3.04963917e-01
-5.33672981e-02 -1.03720538e-01 4.30979282e-01 5.11139095e-01
-9.68703389e-01 5.64413667e-01 -1.30429614e+00 -7.04284847e-01
-8.76599029e-02 3.63333046e-01 -7.41638899e-01 -2.88338333e-01
-8.29299271e-01 8.86854231e-01 1.21428698e-01 1.75905138e-01
-1.19351208e+00 -6.02440894e-01 -1.17235911e+00 -7.59967566e-01
2.40503967e-01 -1.19076490e+00 1.04751384e+00 -5.07894754e-01
-1.60092926e+00 1.25536156e+00 -6.16840273e-02 -2.36046195e-01
1.05254281e+00 -4.12866592e-01 1.39558390e-02 2.17765886e-02
-7.03486893e-03 3.35301250e-01 5.84669828e-01 -1.28875017e+00
-2.51603454e-01 -4.91305172e-01 3.29494655e-01 4.45065111e-01
2.36125097e-01 -5.25762796e-01 -7.33914793e-01 -5.51494598e-01
4.48445976e-01 -1.59984231e+00 -4.64890897e-01 9.07334834e-02
-5.10496974e-01 5.31045735e-01 6.27495944e-01 -4.83024001e-01
8.16722214e-01 -1.87177098e+00 6.03586376e-01 -1.13270715e-01
2.08088383e-01 3.74211520e-02 1.11495480e-01 1.78411648e-01
1.92937687e-01 -2.44719297e-01 -3.64567637e-01 -5.01661003e-01
-3.49810906e-02 4.99566555e-01 -8.58876556e-02 1.04413700e+00
-2.74034709e-01 8.19580257e-01 -1.23753107e+00 -2.74175256e-01
6.08238816e-01 6.13528669e-01 -7.47179270e-01 1.75617620e-01
1.63725682e-03 1.33258820e+00 -4.87746388e-01 7.22034216e-01
7.41080463e-01 -8.31316039e-02 -1.53523460e-01 -3.69998097e-01
-2.50140131e-01 8.83051567e-03 -1.67910516e+00 2.54857421e+00
-2.73563564e-01 3.85300994e-01 1.77864462e-01 -6.31751478e-01
6.53089821e-01 2.28067115e-02 8.14283192e-01 -3.26884896e-01
8.52314681e-02 3.18772376e-01 -2.89777577e-01 -7.95375705e-01
6.60103798e-01 -8.46677721e-02 -1.32747784e-01 -3.70793976e-02
-9.44369137e-02 -5.04866302e-01 -7.72482157e-02 -6.25335053e-02
1.29291725e+00 7.04366088e-01 1.60526901e-01 -3.93750966e-01
7.36037016e-01 3.20709765e-01 6.18886530e-01 3.09066027e-01
-3.12186807e-01 9.86396432e-01 -1.11204624e-01 -7.08527446e-01
-1.04042006e+00 -1.36384141e+00 -2.19739497e-01 2.21920639e-01
8.02046716e-01 -8.28047097e-02 -8.34708273e-01 -4.45970595e-01
9.82272774e-02 1.63992509e-01 -6.62744284e-01 5.42542972e-02
-7.29527116e-01 -7.40506411e-01 3.36157799e-01 -2.05965992e-02
4.81307536e-01 -5.57706773e-01 -7.59497583e-01 2.89445575e-02
-3.63264084e-01 -1.49597275e+00 -6.09318614e-01 -2.84770131e-01
-8.24308157e-01 -1.44976294e+00 -7.74620116e-01 -6.24523282e-01
8.73129904e-01 3.79190832e-01 9.96977329e-01 -3.73637348e-01
-4.20929104e-01 8.36222172e-01 -2.40528241e-01 -2.76894821e-03
-3.41403186e-01 -4.16264199e-02 2.86691517e-01 5.30777462e-02
-1.95328772e-01 -4.66380328e-01 -1.07107091e+00 7.51314461e-01
-8.21156144e-01 -5.44684902e-02 1.10448413e-01 4.14351493e-01
8.87372911e-01 -4.34875131e-01 -5.13008952e-01 -7.01027572e-01
3.29574831e-02 -4.49288398e-01 -8.31343293e-01 -8.96975473e-02
-1.16457306e-01 -2.74711754e-02 2.78369069e-01 -4.19186354e-01
-7.52474070e-01 4.82220203e-01 -1.99286938e-02 -7.13255227e-01
-8.34820271e-02 5.74983545e-02 -1.06018037e-01 -3.67692769e-01
7.78366685e-01 1.11382455e-01 6.44972771e-02 -3.75592530e-01
3.57697397e-01 7.29242936e-02 7.93782592e-01 -5.54821730e-01
1.18216419e+00 1.13492548e+00 1.97908416e-01 -6.95869088e-01
-6.44160628e-01 -7.72055924e-01 -1.03051436e+00 -2.58154184e-01
1.05319488e+00 -1.12335002e+00 -8.78540456e-01 4.60488856e-01
-1.02230895e+00 -4.92805094e-01 -3.99239510e-01 9.68765020e-01
-8.68458271e-01 5.65932393e-01 -4.82969105e-01 -4.57660317e-01
-1.28130943e-01 -1.51461124e+00 1.22330165e+00 -2.75594890e-02
-1.74940348e-01 -1.37535810e+00 6.10290587e-01 3.77909124e-01
7.96065405e-02 1.13560402e+00 3.73629644e-03 -9.54085216e-03
-1.13297141e+00 -1.96718052e-01 3.86173695e-01 1.63599208e-01
1.58818007e-01 1.12296967e-02 -7.73559451e-01 -6.28986835e-01
2.00274363e-01 6.94104061e-02 5.97046435e-01 4.12565768e-01
4.51436281e-01 -1.46899298e-01 -2.79052496e-01 1.26161802e+00
1.64315474e+00 -1.12833269e-01 6.23425901e-01 4.80358541e-01
1.17049980e+00 5.27596533e-01 7.45281458e-01 3.39933127e-01
7.18181849e-01 9.81435359e-01 9.40367520e-01 1.97434388e-02
-1.52962562e-02 -1.64730296e-01 6.29525363e-01 1.03875017e+00
-5.26301861e-01 -1.53071210e-01 -9.96564388e-01 4.98973668e-01
-1.77811086e+00 -7.10363805e-01 -5.88014305e-01 2.67156911e+00
4.75514203e-01 -2.59127289e-01 -2.11190581e-02 -1.31634995e-01
5.12872934e-01 3.44720222e-02 -5.47517598e-01 4.37383130e-02
-4.13122438e-02 -2.01687023e-01 7.23120928e-01 9.08010781e-01
-1.10558248e+00 7.71957457e-01 4.67736197e+00 1.32801831e-01
-1.17946172e+00 2.79734612e-01 -2.15181828e-01 -5.65588549e-02
-4.16956246e-01 8.82457122e-02 -9.21275020e-01 4.54054832e-01
6.83123827e-01 7.50855682e-03 4.35210973e-01 5.84033251e-01
3.60844254e-01 -1.51646048e-01 -1.07435262e+00 1.14204526e+00
1.62133738e-01 -1.07919586e+00 -2.04030484e-01 1.90729052e-01
9.20128882e-01 4.54233468e-01 2.59600580e-02 -1.34069860e-01
5.76121986e-01 -7.02265084e-01 7.43944466e-01 7.77528226e-01
6.45608783e-01 -1.46091416e-01 4.63721693e-01 4.15587395e-01
-1.14696860e+00 3.15869361e-01 -3.28781098e-01 2.29337722e-01
4.33717757e-01 4.86349821e-01 -5.88066459e-01 8.80343139e-01
8.03966880e-01 1.11029577e+00 -3.15513611e-01 1.11115086e+00
-1.08113714e-01 -3.04972660e-02 -3.91093552e-01 4.43103939e-01
3.52494568e-01 -4.70886379e-01 9.73549187e-01 1.10074794e+00
6.47761524e-01 -7.56214783e-02 1.62979543e-01 6.89067423e-01
2.06165075e-01 -8.91227946e-02 -9.80811357e-01 6.34141564e-01
-7.82191195e-03 1.30933595e+00 -6.40225112e-01 4.66041174e-03
-3.34953845e-01 1.30543387e+00 -4.98798490e-02 3.00839245e-01
-1.00940585e+00 1.72758266e-01 1.00534761e+00 4.56746548e-01
-1.17719010e-01 -7.98680604e-01 3.47146720e-01 -1.76703668e+00
2.16391876e-01 -4.63331044e-01 2.51356214e-01 -7.67388642e-01
-7.46680558e-01 4.88726497e-01 -2.19975542e-02 -1.82656527e+00
-2.18905553e-01 -6.64799988e-01 -1.74234465e-01 5.91113508e-01
-1.61988652e+00 -1.20191264e+00 -1.18537092e+00 1.03034914e+00
4.82448280e-01 4.25249785e-01 6.46967888e-01 4.68778402e-01
-4.67330515e-01 9.79690701e-02 3.07395875e-01 7.02878162e-02
8.83107543e-01 -1.37612331e+00 4.09368813e-01 9.09245074e-01
7.54296407e-02 7.49178171e-01 8.40841532e-01 -4.92633253e-01
-1.88938236e+00 -1.27961135e+00 1.95185974e-01 -1.27935219e+00
5.50054729e-01 -3.22989821e-01 -7.13748991e-01 1.05477977e+00
-2.74605721e-01 5.48461974e-01 1.03833079e-01 -6.57044172e-01
1.45094320e-01 6.82648942e-02 -1.23897791e+00 4.29219812e-01
1.43691480e+00 -3.19317490e-01 -3.82276714e-01 5.15330434e-01
7.24363685e-01 -1.31221366e+00 -1.15072536e+00 5.14201641e-01
5.96012950e-01 -1.08915520e+00 1.10166121e+00 -1.83668733e-01
1.10384271e-01 -6.67201936e-01 -3.56332123e-01 -1.04809356e+00
1.65670022e-01 -1.02364194e+00 -2.27790233e-02 8.17595899e-01
-9.66498554e-02 -6.65659964e-01 9.10645545e-01 6.49168253e-01
-4.18000996e-01 -3.23436618e-01 -1.07814860e+00 -9.67427731e-01
-1.35850698e-01 -4.93210495e-01 2.21930727e-01 1.27494287e+00
-5.14316201e-01 -2.02318981e-01 -3.29812706e-01 5.33008575e-01
8.11977327e-01 -1.34777389e-02 1.28733814e+00 -1.02674186e+00
-1.70575231e-02 1.49478242e-01 -8.57669234e-01 -1.22298694e+00
1.30830808e-02 -9.72195148e-01 -5.54856332e-03 -1.58414495e+00
-2.16223255e-01 -3.14868689e-01 2.58884937e-01 -5.52136302e-02
2.32422680e-01 3.23552787e-01 -7.35341311e-02 5.59780002e-01
-6.31728530e-01 3.70434642e-01 1.68116891e+00 1.21804982e-01
-3.00156206e-01 1.18283018e-01 1.57817110e-01 1.03230131e+00
3.08342248e-01 -3.30977380e-01 -5.69332123e-01 -6.24664962e-01
2.72018313e-01 2.07859427e-01 8.57704520e-01 -1.12987971e+00
3.34393173e-01 -2.89178729e-01 -7.98487738e-02 -3.09248537e-01
3.55244279e-01 -9.56822395e-01 8.31306517e-01 7.22957313e-01
1.93294689e-01 3.35151970e-01 1.29710793e-01 7.29841590e-01
-1.62712395e-01 -4.14324589e-02 9.21508968e-01 -4.60127831e-01
-8.97974074e-01 6.76723063e-01 6.94396645e-02 2.79721826e-01
1.22200465e+00 -5.41934788e-01 -2.38518149e-01 -9.46375281e-02
-1.03831184e+00 7.09861517e-02 1.26619411e+00 5.09865463e-01
5.98178625e-01 -1.27940249e+00 -5.16011834e-01 2.88434356e-01
1.66274980e-01 6.68671370e-01 3.96819770e-01 1.22763908e+00
-1.28939366e+00 3.30720574e-01 -1.73836544e-01 -1.01420176e+00
-9.86983776e-01 5.57469904e-01 7.66675115e-01 -1.73325054e-02
-8.71964276e-01 6.18946373e-01 3.18284243e-01 -7.44176745e-01
-8.85583237e-02 -8.45379353e-01 8.36167336e-02 -2.01276273e-01
1.30973496e-02 4.23155278e-01 1.03524439e-01 -1.03492439e+00
-5.16782105e-01 1.32532156e+00 4.86994147e-01 -2.29234770e-02
1.06923378e+00 -5.71297884e-01 6.77667260e-02 3.92103314e-01
1.23866189e+00 1.48766026e-01 -1.49508035e+00 -4.29325737e-02
-2.62954444e-01 -6.87720239e-01 -2.90062279e-01 -1.82216898e-01
-9.78100657e-01 7.60145128e-01 6.83065951e-01 -8.18574578e-02
7.69202888e-01 -1.30667418e-01 7.17545092e-01 3.35057139e-01
8.16697538e-01 -5.88854492e-01 4.00342159e-02 4.50503618e-01
9.24838364e-01 -1.40441287e+00 4.28024307e-02 -5.95132291e-01
-5.03170669e-01 9.62966681e-01 5.48946321e-01 -3.14561129e-01
5.60712218e-01 1.14590123e-01 1.14661865e-01 -1.51286304e-01
-2.18560860e-01 -1.81731403e-01 2.57632703e-01 5.93514562e-01
8.94257426e-02 -2.10574716e-01 1.45175800e-01 -1.88590516e-03
-3.06669414e-01 -1.45256547e-02 6.50733411e-01 7.63406992e-01
-3.28408591e-02 -9.45493996e-01 -4.44186300e-01 -3.30114931e-01
-3.16895396e-01 2.96041995e-01 3.36549841e-02 1.10394776e+00
1.39429703e-01 4.21816617e-01 -1.44362390e-01 -2.66998231e-01
7.21129894e-01 -5.16922832e-01 7.57485092e-01 -5.46475291e-01
-4.06128675e-01 -4.94961105e-02 -1.97386444e-01 -1.03760755e+00
-8.78055394e-01 -7.96970606e-01 -1.41779435e+00 -3.77884835e-01
3.41940857e-02 -1.49642140e-01 8.15589786e-01 4.66410935e-01
1.31877020e-01 4.03272510e-01 4.10232186e-01 -1.28661644e+00
-3.36811483e-01 -5.03110945e-01 -4.60446030e-01 8.95184994e-01
8.23774636e-01 -8.32150221e-01 -5.76631844e-01 4.25168246e-01] | [8.416228294372559, -2.0495898723602295] |
9ebaff42-79a7-4ce4-a0ec-4d1cd8767df9 | plujagh-at-semeval-2016-task-11-simple-system | null | null | https://aclanthology.org/S16-1146 | https://aclanthology.org/S16-1146.pdf | PLUJAGH at SemEval-2016 Task 11: Simple System for Complex Word Identification | null | ["Krzysztof Wr{\\'o}bel"] | 2016-06-01 | null | null | null | semeval-2016-6 | ['complex-word-identification'] | ['natural-language-processing'] | [-8.63703638e-02 1.71006292e-01 -6.22772932e-01 -4.08054382e-01
-8.41685571e-03 -9.08429027e-01 6.55310392e-01 -6.53472245e-01
-2.85945535e-01 1.06888819e+00 -4.63127941e-02 -1.01159286e+00
-3.91567826e-01 -9.63214397e-01 -4.95059669e-01 -6.31337762e-01
-9.79754329e-01 7.25764990e-01 3.30370307e-01 -6.93831444e-01
7.03166842e-01 7.88774848e-01 -1.68942046e+00 7.18545914e-01
7.04417467e-01 8.52217197e-01 2.49141872e-01 1.14950800e+00
-1.95044339e-01 1.55633950e+00 -7.48382092e-01 -5.46825826e-01
3.13719302e-01 -1.23176083e-01 -7.22945035e-01 -1.01074085e-01
9.28529128e-02 -8.59008506e-02 -2.09758401e-01 9.22211111e-01
5.37373662e-01 4.49454933e-02 1.08379531e+00 -1.42548037e+00
-5.91619551e-01 6.10313773e-01 -4.01565880e-02 1.21627934e-01
1.03678203e+00 -5.39447069e-01 1.19919395e+00 -1.13026452e+00
7.20913768e-01 1.26888943e+00 8.66221786e-01 5.44149756e-01
-1.22286928e+00 -1.94712028e-01 -3.26822817e-01 -9.51717794e-02
-1.46558487e+00 -3.25250506e-01 4.25783843e-02 -2.08119690e-01
1.66093647e+00 1.26596653e+00 1.20609856e+00 1.01401424e+00
1.26658809e+00 8.34431887e-01 1.04267764e+00 -5.13792276e-01
3.35295945e-01 3.66983831e-01 1.54683650e-01 6.33519173e-01
8.40953708e-01 5.26628852e-01 -7.06372619e-01 -9.13127720e-01
9.33553874e-01 -2.94925272e-01 1.71355158e-01 -5.05680561e-01
-9.05919552e-01 6.91228509e-01 1.78732842e-01 3.83959889e-01
-1.39880210e-01 9.89067405e-02 1.26390755e-01 5.30987144e-01
-2.58292928e-02 6.47037446e-01 -9.11868811e-01 -1.33165747e-01
-8.71728659e-01 5.10332465e-01 1.25398111e+00 1.52653182e+00
1.24482810e-01 2.94908643e-01 -9.34252143e-02 3.17179203e-01
8.92314315e-01 1.01808000e+00 4.28362608e-01 -1.36146402e+00
-6.87414408e-02 1.72361732e-01 5.01781464e-01 -8.52631688e-01
-6.33224547e-01 -9.64177120e-03 -8.93263519e-01 4.49267089e-01
3.49161088e-01 4.57367361e-01 -8.02827001e-01 5.07305264e-01
4.33481112e-02 -2.34125629e-01 4.53833073e-01 5.55570945e-02
4.99930978e-01 3.76208365e-01 -1.34477139e-01 -5.73289394e-01
1.06082785e+00 -1.36716676e+00 -1.35299087e+00 2.33215362e-01
9.05734658e-01 -1.07320261e+00 4.35900748e-01 5.33875942e-01
-1.55548143e+00 -1.37560293e-01 -1.08699942e+00 1.78573877e-01
-7.27255583e-01 -3.14239264e-01 8.57801437e-01 1.43120694e+00
-1.60129595e+00 9.73287821e-01 -4.91727620e-01 6.59165755e-02
1.20568443e-02 8.24621081e-01 -2.64718989e-03 4.62812334e-01
-1.33193445e+00 1.08501506e+00 2.22979754e-01 -1.21242590e-01
-1.65216476e-01 -2.13068098e-01 -8.23704481e-01 -5.63443303e-01
-4.78693932e-01 -5.29636025e-01 1.44139910e+00 -2.59346128e-01
-1.65295815e+00 9.71794367e-01 -1.42069459e-01 -1.97814897e-01
6.14786744e-01 -1.28011424e-02 -8.31891418e-01 2.42498964e-01
-1.89849049e-01 5.76383233e-01 9.28263724e-01 -1.35132408e+00
-7.59897232e-01 -1.67359829e-01 -1.23336017e-01 2.66287565e-01
-1.25510961e-01 1.89734384e-01 2.11616129e-01 -1.12999000e-01
3.27147305e-01 -7.20919967e-01 -2.53068686e-01 -5.32041907e-01
-1.46512717e-01 -7.10518599e-01 7.70373225e-01 -4.51523662e-01
1.83705616e+00 -1.67618537e+00 -2.06720144e-01 3.98590982e-01
3.57815564e-01 -1.24705513e-03 2.40583986e-01 1.08380008e+00
-2.76906848e-01 7.32199550e-01 4.11965609e-01 -1.10722095e-01
2.24991128e-01 4.85861301e-01 -4.16602850e-01 3.05609167e-01
-1.29282743e-01 1.15307164e+00 -1.16605783e+00 -5.23096442e-01
4.11106765e-01 9.42391157e-02 -4.58732933e-01 4.79237735e-01
2.99364805e-01 1.47170946e-01 -3.56553018e-01 1.39399457e+00
1.15709066e+00 -1.31984919e-01 1.45911396e-01 5.30878425e-01
-3.79135728e-01 3.55090618e-01 -6.96863770e-01 1.05554795e+00
7.08333924e-02 5.00173986e-01 1.02364108e-01 -7.94621468e-01
3.33247900e-01 8.61540735e-01 4.37155962e-01 -9.67555881e-01
-2.26398129e-02 6.92409754e-01 1.12803578e-01 -6.07703328e-01
7.58228302e-01 3.94563079e-02 -3.64872098e-01 6.40070081e-01
-2.37588286e-01 -6.59476995e-01 9.64643434e-02 3.08100313e-01
6.22585893e-01 -5.18246442e-02 5.62923312e-01 -1.05613089e+00
7.86340594e-01 -1.82965681e-01 -1.81299388e-01 1.03415680e+00
-3.09923887e-01 3.19085121e-01 2.99841821e-01 -6.65102363e-01
-6.45341039e-01 -1.12307119e+00 -4.89381433e-01 1.30636716e+00
3.24267983e-01 -4.39044595e-01 -9.54439282e-01 -2.49762803e-01
1.77620783e-01 6.89606130e-01 -5.90509653e-01 3.84124845e-01
-5.03739953e-01 -8.32535863e-01 7.39044368e-01 3.45434904e-01
-5.07752821e-02 -1.33414865e+00 -6.58416986e-01 1.25490099e-01
-2.20292807e-01 -6.63697243e-01 -6.23428151e-02 4.48765576e-01
-1.35989368e+00 -5.18594682e-01 -6.66252747e-02 -8.20914626e-01
5.87345481e-01 2.46782884e-01 1.27047324e+00 5.39230824e-01
-2.31483161e-01 4.26904231e-01 -1.21292919e-01 -4.95818377e-01
-4.59671497e-01 -8.00336525e-02 5.28869390e-01 -5.87835789e-01
5.19427478e-01 -2.50617653e-01 -7.29350567e-01 5.37953973e-01
-6.88540697e-01 1.62748516e-01 1.79803044e-01 1.04410267e+00
1.35816500e-01 -9.34035778e-02 1.22507080e-01 -6.38007045e-01
8.72274399e-01 -1.69219792e-01 -3.78732830e-01 5.77745810e-02
-6.77108407e-01 -3.74140263e-01 3.21430594e-01 -3.25342178e-01
-1.01981449e+00 -4.87835288e-01 -9.82677937e-02 2.45538145e-01
1.11353043e-02 -1.46784872e-01 6.47139177e-02 -5.24923325e-01
8.02199244e-01 9.25758183e-02 1.99174434e-02 -6.80815242e-03
3.01039815e-01 7.09525108e-01 -6.82967342e-03 -6.68678164e-01
8.44880998e-01 4.91470337e-01 7.98524171e-02 -9.57177758e-01
-1.52186140e-01 -2.60129690e-01 -9.51962709e-01 -6.54426932e-01
6.56643391e-01 -6.78531289e-01 -9.10833478e-01 3.91110867e-01
-9.38691139e-01 -3.38627815e-01 -3.91645581e-01 4.25431967e-01
-1.01278400e+00 2.75717527e-02 -3.90154392e-01 -1.27895141e+00
-5.10977268e-01 -1.02017939e+00 9.43384409e-01 5.30070923e-02
-5.10597289e-01 -1.26927447e+00 5.87685481e-02 2.71537274e-01
1.81734428e-01 -1.73075795e-01 6.90226793e-01 -2.38256708e-01
-4.24233019e-01 -1.53791070e-01 2.34436691e-02 -1.39755070e-01
1.70832314e-02 4.95917559e-01 -9.81751978e-01 -5.31145096e-01
6.65065646e-02 -1.92070693e-01 -1.08835101e-01 6.52520418e-01
5.91872573e-01 -2.29931593e-01 -8.56000841e-01 5.40386558e-01
1.38545322e+00 3.85070026e-01 5.32770038e-01 7.28214979e-01
1.41836226e-01 5.53460240e-01 9.17806149e-01 4.63203549e-01
1.30579369e-02 3.28798652e-01 2.40537539e-01 1.49327129e-01
1.11720070e-01 -1.54819340e-01 3.77893507e-01 1.16112018e+00
-8.18235934e-01 -2.69281328e-01 -5.07867396e-01 4.42987174e-01
-1.72482407e+00 -1.40330648e+00 -4.32368398e-01 6.90478683e-01
6.25676990e-01 1.56016424e-01 -1.48347050e-01 3.35214496e-01
4.99015123e-01 -2.03574806e-01 -1.19133167e-01 -1.06291151e+00
-1.43546045e-01 3.15233678e-01 7.37729073e-01 1.00061214e+00
-7.20721722e-01 1.03317809e+00 1.29781246e+01 1.02230716e+00
2.21112028e-01 1.03134915e-01 5.16071796e-01 3.48020852e-01
-4.36954498e-01 -4.56139445e-02 -1.04416132e+00 2.72933897e-02
1.38140702e+00 -4.30666685e-01 6.85999811e-01 5.44219851e-01
3.44648361e-01 -4.23268199e-01 -1.26188684e+00 5.26221812e-01
9.73738134e-02 -1.40886843e+00 -2.83300440e-04 6.85225725e-01
7.73699820e-01 -5.08050561e-01 6.22419357e-01 3.24184299e-01
6.09259963e-01 -1.14389277e+00 8.60300779e-01 2.53660440e-01
1.03040910e+00 -6.05088234e-01 5.67372203e-01 1.68872893e-01
-1.14389896e+00 -2.20873043e-01 -8.77727985e-01 -1.00755692e+00
3.93533185e-02 -1.81779593e-01 -4.29956943e-01 3.48861217e-01
9.58353162e-01 2.99398601e-01 -3.93658698e-01 9.95779395e-01
-4.78694476e-02 1.04875881e-02 -2.71853864e-01 -4.48467314e-01
4.83122796e-01 -3.54241252e-01 4.66730654e-01 1.00164843e+00
2.48499006e-01 3.51035744e-01 -9.84472036e-02 4.01770771e-01
5.45058846e-01 3.29446048e-02 -1.19659424e+00 -1.78908288e-01
2.83276141e-01 9.16795909e-01 -4.83487815e-01 -4.22520459e-01
-2.00212970e-01 8.62069130e-01 -3.55488248e-02 5.01107454e-01
-6.11489356e-01 -4.35615242e-01 9.72222984e-01 -1.27327025e-01
-1.14700586e-01 -3.48497719e-01 -6.23769283e-01 -7.30352640e-01
-5.89872956e-01 -4.54965204e-01 5.93606755e-02 -5.53365827e-01
-1.39813089e+00 5.79277515e-01 -2.27688253e-02 -1.40553558e+00
-6.99901402e-01 -1.27676582e+00 -4.76714373e-01 4.92853165e-01
-1.11898029e+00 -1.10984349e+00 2.50124663e-01 4.52870727e-01
1.64141744e-01 -5.34416080e-01 1.39563632e+00 3.57715860e-02
1.00637585e-01 9.24474537e-01 6.69434488e-01 -7.37814724e-01
5.56605101e-01 -1.27867436e+00 5.68737745e-01 -1.37897313e-01
-4.31265175e-01 9.05828118e-01 6.28349900e-01 -5.39804697e-01
-1.41196322e+00 -3.66917729e-01 1.08350635e+00 -9.83769417e-01
6.55218959e-01 -3.86345625e-01 4.23767231e-02 7.88592756e-01
7.15902448e-01 -6.18741751e-01 8.21781039e-01 -1.83753878e-01
1.80774391e-01 5.75296998e-01 -1.39248300e+00 6.12354755e-01
1.66275799e+00 -4.63594139e-01 -6.25784039e-01 7.60327101e-01
8.13696027e-01 -6.94087505e-01 -1.30082703e+00 3.34633321e-01
8.65424156e-01 -8.75409484e-01 1.61978090e+00 -1.32660246e+00
-4.43697497e-02 2.81152606e-01 -2.61993498e-01 -9.32519078e-01
-5.96193194e-01 -1.23518765e+00 -5.33532679e-01 -5.83747849e-02
5.96577883e-01 -1.13057327e+00 3.42365682e-01 8.74560475e-01
-2.82833427e-01 -6.42737269e-01 -1.06996536e+00 -1.32016802e+00
-3.35779637e-02 -1.45572275e-01 4.90409225e-01 7.63798356e-01
6.83744550e-01 1.09839931e-01 -6.36873543e-02 -1.00294888e-01
5.33176839e-01 9.55312885e-03 4.41501856e-01 -1.34294486e+00
3.86843324e-01 -5.75816095e-01 -3.07655483e-01 -9.45992947e-01
-8.85957032e-02 -8.12076271e-01 -6.53862000e-01 -1.28511906e+00
-8.32044985e-03 -1.91056758e-01 -1.11109078e-01 -1.65725678e-01
3.67937148e-01 2.13746816e-01 1.20859891e-02 1.03788137e-01
-3.71160030e-01 6.18435517e-02 1.29639816e+00 6.91750320e-05
-1.62315920e-01 4.85058486e-01 -4.81304944e-01 7.84440815e-01
8.58408585e-02 -2.99253196e-01 -6.78878546e-01 6.11881316e-02
6.69384480e-01 4.61409837e-02 3.24159935e-02 -7.42885649e-01
5.37211418e-01 -3.75702560e-01 4.78586555e-01 -1.32223868e+00
1.30741090e-01 -9.61415648e-01 6.95283338e-02 9.40189242e-01
2.74610907e-01 1.20424610e-02 8.66204947e-02 5.39500564e-02
-1.44871444e-01 -5.70943117e-01 9.21121240e-01 -4.07591403e-01
-4.92852688e-01 -5.20386267e-03 -1.03226590e+00 8.97834301e-02
9.92593169e-01 -7.84614205e-01 -3.59281451e-01 -4.20183957e-01
-8.29068601e-01 -1.95836127e-02 6.50830388e-01 3.03609259e-02
7.30431557e-01 -1.51530886e+00 -2.30721906e-01 7.18729138e-01
-3.22939813e-01 -3.74400020e-01 -1.70157343e-01 6.58265352e-01
-1.32361674e+00 1.02442718e+00 -5.41665435e-01 -4.55340147e-01
-1.14228773e+00 4.78126436e-01 4.28307921e-01 -2.41845414e-01
-2.15481281e-01 1.11954463e+00 2.71224789e-02 -8.23025763e-01
1.85185194e-01 -9.21545625e-02 -7.54407048e-01 3.55081353e-03
6.88606799e-01 1.05194807e+00 -2.91290224e-01 -6.04341030e-01
-4.56784427e-01 6.65885091e-01 2.32151806e-01 -2.87484169e-01
9.17833567e-01 -2.19243199e-01 -9.89108324e-01 4.28274393e-01
8.38715494e-01 -1.36269778e-01 -3.69319022e-02 4.16855574e-01
1.25943512e-01 -8.02164078e-01 -4.32406247e-01 -3.55811834e-01
-1.85641110e-01 5.54822803e-01 5.34874737e-01 8.99602413e-01
8.57008278e-01 -3.02566767e-01 8.18335712e-01 9.66778398e-01
5.72402716e-01 -1.68019545e+00 -2.13140488e-01 6.89524531e-01
9.12339568e-01 -9.28350806e-01 5.44190466e-01 -7.27165341e-01
-4.14997995e-01 1.32979155e+00 4.68304873e-01 -1.55325383e-01
1.27306652e+00 5.74917436e-01 1.14069022e-02 -3.70670199e-01
-9.44949508e-01 1.12705544e-01 3.75366658e-01 1.11147714e+00
5.06513238e-01 5.07374525e-01 -9.66500878e-01 3.21953118e-01
-7.28706717e-01 -2.34555230e-01 4.90474731e-01 1.41972518e+00
-6.43810987e-01 -1.20391917e+00 -7.23931909e-01 4.61561680e-01
-5.49773455e-01 -1.16372630e-01 -5.28106689e-01 8.46754074e-01
-5.76629937e-02 1.49448860e+00 -1.97535474e-03 -5.03491640e-01
4.11356747e-01 1.41089618e-01 7.21762300e-01 -1.23501487e-01
-9.04846430e-01 4.15413082e-01 3.84890139e-01 -1.23056793e+00
-8.58632207e-01 -1.05834293e+00 -1.40667629e+00 -1.19437599e+00
-5.12782812e-01 1.89310342e-01 3.83317530e-01 3.90289724e-01
-2.06836104e-01 2.85260603e-02 9.80917513e-01 -1.07949340e+00
-3.90341938e-01 -9.39418614e-01 -1.00026262e+00 -7.84516707e-02
2.89751232e-01 -8.00943017e-01 -7.83523321e-01 2.83909619e-01] | [-7.2968339920043945, 3.6863887310028076] |
777187ae-3482-495d-a55b-82037e705873 | deformable-gans-for-pose-based-human-image | 1801.00055 | null | http://arxiv.org/abs/1801.00055v2 | http://arxiv.org/pdf/1801.00055v2.pdf | Deformable GANs for Pose-based Human Image Generation | In this paper we address the problem of generating person images conditioned
on a given pose. Specifically, given an image of a person and a target pose, we
synthesize a new image of that person in the novel pose. In order to deal with
pixel-to-pixel misalignments caused by the pose differences, we introduce
deformable skip connections in the generator of our Generative Adversarial
Network. Moreover, a nearest-neighbour loss is proposed instead of the common
L1 and L2 losses in order to match the details of the generated image with the
target image. We test our approach using photos of persons in different poses
and we compare our method with previous work in this area showing
state-of-the-art results in two benchmarks. Our method can be applied to the
wider field of deformable object generation, provided that the pose of the
articulated object can be extracted using a keypoint detector. | ['Nicu Sebe', 'Stephane Lathuiliere', 'Enver Sangineto', 'Aliaksandr Siarohin'] | 2017-12-29 | deformable-gans-for-pose-based-human-image-1 | http://openaccess.thecvf.com/content_cvpr_2018/html/Siarohin_Deformable_GANs_for_CVPR_2018_paper.html | http://openaccess.thecvf.com/content_cvpr_2018/papers/Siarohin_Deformable_GANs_for_CVPR_2018_paper.pdf | cvpr-2018-6 | ['gesture-to-gesture-translation', 'pose-transfer'] | ['computer-vision', 'computer-vision'] | [ 6.11132920e-01 3.64956111e-01 4.21880186e-01 -1.86073244e-01
-7.40959525e-01 -7.57527232e-01 7.36279309e-01 -2.10325584e-01
-5.71101964e-01 8.38477314e-01 4.64624204e-02 4.41501200e-01
2.01655179e-01 -7.75836706e-01 -1.05885470e+00 -6.98540509e-01
2.67390311e-01 7.13638961e-01 2.44405210e-01 -1.26196325e-01
-1.17716759e-01 7.93854773e-01 -1.30489016e+00 1.69909373e-01
4.98570502e-01 6.72868609e-01 7.01937377e-02 7.64692962e-01
5.63005507e-01 1.33270562e-01 -9.08089280e-01 -7.12215781e-01
7.14179158e-01 -5.56018591e-01 -8.18989396e-01 4.78465259e-01
1.00105155e+00 -2.11935356e-01 -4.55094904e-01 9.75184739e-01
7.21644104e-01 3.12984854e-01 7.69888401e-01 -1.12669921e+00
-1.48710147e-01 2.65941650e-01 -5.17441034e-01 -1.28438592e-01
6.56015694e-01 3.73771578e-01 5.37406385e-01 -6.56472802e-01
1.04678011e+00 1.36277437e+00 5.70642948e-01 8.76815796e-01
-1.47078991e+00 -4.41478401e-01 1.16106644e-01 -6.11939467e-02
-1.25788963e+00 -4.04421479e-01 7.72692025e-01 -4.37632889e-01
3.58491778e-01 8.66718516e-02 7.15317190e-01 1.36266398e+00
1.40731692e-01 4.69836354e-01 9.97063220e-01 -4.69172418e-01
1.39265023e-02 -7.36649707e-02 -6.78889692e-01 6.97964907e-01
1.17083684e-01 2.64916658e-01 -1.44570619e-01 4.12915237e-02
9.52240348e-01 -1.03844739e-01 -2.78985381e-01 -5.92173994e-01
-1.25814176e+00 6.87133849e-01 6.86564744e-01 1.74554870e-01
-4.90421623e-01 4.27363873e-01 -6.21308163e-02 -5.43321893e-02
3.16643536e-01 3.99572492e-01 -1.30276717e-02 2.94082969e-01
-9.84818101e-01 7.74579525e-01 7.66590476e-01 7.29698777e-01
5.46878755e-01 -4.81170453e-02 -6.46589220e-01 3.36036593e-01
1.73834339e-01 5.19748449e-01 5.87058067e-02 -9.71582353e-01
5.54411769e-01 2.94411808e-01 3.13657492e-01 -8.65656376e-01
-6.56766370e-02 -4.10389125e-01 -6.79408789e-01 5.02075136e-01
8.07879984e-01 -3.38621318e-01 -1.30764890e+00 1.78207684e+00
5.74217081e-01 3.31330806e-01 9.27445218e-02 1.01914728e+00
5.61229587e-01 5.08412063e-01 -5.70833348e-02 2.65486956e-01
1.16862273e+00 -8.57062638e-01 -3.62028629e-01 -6.03945732e-01
-2.71417618e-01 -7.53811538e-01 5.74707270e-01 1.18668385e-01
-1.42906547e+00 -7.33759403e-01 -9.13421452e-01 -5.60758170e-03
-1.17133118e-01 1.95877582e-01 8.78978055e-03 5.47132254e-01
-9.00458753e-01 7.34581530e-01 -7.41095901e-01 -4.28299546e-01
3.84096116e-01 3.39678407e-01 -4.77302849e-01 -2.50646546e-02
-1.11142540e+00 9.76046085e-01 3.85160059e-01 1.35063261e-01
-1.14340365e+00 -5.53286552e-01 -9.28600848e-01 -1.71952605e-01
4.01261479e-01 -1.02180028e+00 9.16743636e-01 -1.06921661e+00
-1.43715823e+00 1.00840628e+00 1.76987663e-01 -5.55434525e-01
1.23031807e+00 -1.75785571e-01 -3.82577479e-02 2.82545984e-01
4.25043516e-02 1.04996729e+00 1.23273981e+00 -1.36775780e+00
-3.35501939e-01 -3.55000019e-01 1.33325636e-01 3.35191548e-01
1.70674592e-01 -3.00640851e-01 -6.95834458e-01 -9.59829569e-01
-2.38510773e-01 -1.35642374e+00 -3.55965853e-01 2.18085021e-01
-7.17669249e-01 2.11783394e-01 7.29632795e-01 -8.50818217e-01
4.67651457e-01 -2.07751369e+00 7.06534326e-01 2.91870058e-01
-1.40694350e-01 2.34922394e-01 -2.37246200e-01 3.30416828e-01
-2.93843634e-02 -1.84744030e-01 -4.19138581e-01 -6.83841527e-01
-7.60002062e-02 1.57496512e-01 -1.85609132e-01 6.18149698e-01
3.47804636e-01 9.35148597e-01 -7.28849947e-01 -3.83896798e-01
3.40193510e-01 7.75875926e-01 -4.96513695e-01 3.23557675e-01
-2.59703398e-01 7.78054953e-01 -2.54215658e-01 1.12350486e-01
5.78004718e-01 1.18727610e-01 -8.17565024e-02 -3.14512759e-01
2.87808985e-01 -4.55201790e-02 -1.43250084e+00 1.83911610e+00
-2.75652349e-01 3.31863582e-01 -7.04715028e-02 -6.69845700e-01
7.53972888e-01 3.78196031e-01 2.93525994e-01 -1.48992330e-01
8.55281726e-02 -7.21579641e-02 -3.49379703e-02 -1.36212558e-01
2.57831752e-01 -2.41743401e-01 -4.90421392e-02 1.42181233e-01
1.07583165e-01 -3.62480462e-01 3.40529412e-01 -3.94741707e-02
7.56285250e-01 3.13899010e-01 -4.35411260e-02 2.28832290e-03
5.41862726e-01 -3.65669161e-01 1.94210947e-01 5.85512102e-01
2.32852042e-01 1.19269872e+00 2.58179158e-01 -3.82065445e-01
-1.35033298e+00 -1.27174985e+00 3.10718156e-02 4.09792751e-01
5.74918427e-02 3.37610133e-02 -9.96631384e-01 -8.51728141e-01
-3.75095308e-02 3.92281950e-01 -7.71360755e-01 -9.99941081e-02
-8.12310278e-01 -3.41597229e-01 6.06248379e-01 5.64585686e-01
6.58730507e-01 -1.27758276e+00 -7.34385073e-01 3.02797586e-01
-1.79309145e-01 -1.32879269e+00 -7.57908881e-01 -2.45570540e-01
-5.85937440e-01 -1.13450372e+00 -1.19099712e+00 -7.52698064e-01
1.02787030e+00 -3.36660892e-01 1.03501725e+00 -1.64456934e-01
-4.65993464e-01 3.54867160e-01 -8.61620456e-02 -1.77270114e-01
-6.44781113e-01 1.68808401e-02 -5.76931015e-02 2.05305010e-01
-3.78811508e-01 -2.36285999e-01 -7.44677901e-01 2.28109807e-01
-1.08376920e+00 -3.60614024e-02 4.74439293e-01 7.27848470e-01
4.75917220e-01 1.86357111e-01 1.59459710e-01 -7.09598601e-01
3.28137875e-01 -3.04505024e-02 -5.98108768e-01 1.59655407e-01
-1.00403339e-01 2.18053713e-01 3.64497066e-01 -5.70401013e-01
-8.92705977e-01 6.13081336e-01 -1.42949343e-01 -4.03926522e-01
-2.83732116e-01 -1.17151394e-01 -4.00496036e-01 -1.50491282e-01
6.05840743e-01 4.41310517e-02 5.78059080e-05 -3.63343656e-01
4.84415859e-01 5.81047870e-02 9.05853271e-01 -6.17097318e-01
1.31539202e+00 6.50122285e-01 2.42047250e-01 -3.88215303e-01
-5.95433950e-01 2.83404067e-03 -7.62082398e-01 -2.69441575e-01
1.15225589e+00 -8.04479897e-01 -5.06262779e-01 6.76646590e-01
-1.35103476e+00 -1.83624282e-01 -6.13616705e-01 3.37742925e-01
-7.30393827e-01 2.64063686e-01 -3.29892933e-01 -5.14743805e-01
-3.26276094e-01 -1.36674702e+00 1.34610367e+00 3.60908300e-01
-1.98483571e-01 -9.44633484e-01 8.71594846e-02 3.75297129e-01
1.53464153e-01 7.65491426e-01 4.25256521e-01 -4.98366505e-01
-6.60055578e-01 -3.74119073e-01 2.10947454e-01 6.18114293e-01
1.37058511e-01 -2.34163597e-01 -8.71582031e-01 -6.82239592e-01
-3.24661523e-01 -2.34538540e-01 7.36849785e-01 2.48092696e-01
7.02757537e-01 -4.83900845e-01 -3.84295225e-01 5.07146537e-01
1.27427006e+00 -4.55519743e-03 8.98469687e-01 2.10285336e-01
8.98416460e-01 6.65588677e-01 3.00142884e-01 2.00974658e-01
2.89863925e-02 1.01424575e+00 4.18663293e-01 -3.07513297e-01
-4.67334747e-01 -4.24356490e-01 2.89676279e-01 -4.19469357e-01
-3.07151079e-01 -5.31870604e-01 -6.57449365e-01 5.84868491e-01
-1.55786252e+00 -1.04939353e+00 2.70416737e-01 2.46064830e+00
7.33871639e-01 1.31372660e-01 3.82403642e-01 9.02176648e-02
9.21490788e-01 1.73094794e-01 -4.08613980e-01 6.78882897e-02
5.98150790e-02 4.77927238e-01 4.58413869e-01 5.76213479e-01
-1.12596834e+00 8.11046839e-01 6.02434540e+00 4.53610748e-01
-1.11298966e+00 -1.48740545e-01 5.31084001e-01 -1.42547369e-01
1.12164272e-02 -1.33137748e-01 -8.68742704e-01 4.72024888e-01
4.91564333e-01 1.35113776e-01 2.78801888e-01 4.62572217e-01
-9.58503336e-02 1.00961983e-01 -1.24873066e+00 7.30844140e-01
2.32181892e-01 -1.09844601e+00 1.42324358e-01 1.54373810e-01
9.15495455e-01 -4.40187484e-01 3.12139690e-01 -1.79925412e-01
1.36593834e-01 -1.13944864e+00 9.00456905e-01 7.54361808e-01
7.06023753e-01 -9.08539653e-01 6.78196371e-01 1.41470224e-01
-9.58216608e-01 3.07427287e-01 -1.14071937e-02 1.49111792e-01
3.82841438e-01 2.94770688e-01 -1.00705349e+00 5.38484931e-01
4.90307629e-01 3.68434936e-01 -5.69052458e-01 1.07583117e+00
-4.49549019e-01 9.56058726e-02 -3.79366606e-01 5.09633422e-01
1.18866853e-01 -5.29049784e-02 8.71308208e-01 1.02508938e+00
4.28116649e-01 -3.21482152e-01 3.03287923e-01 1.03525507e+00
-2.22139746e-01 -1.96452603e-01 -4.93629545e-01 2.14737892e-01
1.79383829e-01 1.17824793e+00 -7.13218033e-01 -3.27192396e-01
9.92844254e-02 1.58375311e+00 3.27169523e-02 3.39421242e-01
-8.18976343e-01 -1.75926059e-01 4.40980434e-01 5.14093935e-01
6.08724535e-01 -8.11634883e-02 2.85941362e-01 -8.71931553e-01
2.49830082e-01 -9.18417931e-01 1.75993681e-01 -7.11403549e-01
-1.12546349e+00 7.22355306e-01 3.35668385e-01 -1.12561977e+00
-6.50549591e-01 -3.49661887e-01 -5.90339422e-01 1.14319229e+00
-1.17033315e+00 -1.36489213e+00 -4.26499933e-01 5.32578468e-01
4.26053464e-01 4.53052483e-02 6.48687422e-01 2.04596013e-01
-1.23610817e-01 4.66435462e-01 -3.25313300e-01 3.75252396e-01
8.65611613e-01 -1.29921532e+00 8.10837626e-01 9.54016984e-01
2.21500039e-01 3.18329334e-01 8.90794396e-01 -7.34485567e-01
-8.63359571e-01 -1.30126333e+00 6.93100989e-01 -5.93231618e-01
1.37769341e-01 -5.00088692e-01 -6.69407070e-01 7.72972465e-01
2.06634566e-01 2.66844094e-01 -2.67000031e-02 -5.79502046e-01
-7.27618784e-02 2.99789831e-02 -1.43851161e+00 5.14033437e-01
8.95212293e-01 -2.77306587e-01 -4.52346414e-01 3.74051839e-01
3.99961382e-01 -9.58777785e-01 -7.74813831e-01 5.07772982e-01
5.95322847e-01 -6.66014314e-01 1.28114152e+00 -5.18012047e-01
3.60053688e-01 -4.83550489e-01 1.85370594e-01 -1.59929335e+00
-1.62251368e-01 -6.62109733e-01 2.35539079e-01 1.22516203e+00
1.67695373e-01 -3.36274415e-01 8.72262299e-01 4.45533842e-01
2.42686212e-01 -5.07573545e-01 -9.98553038e-01 -7.73294806e-01
1.65413216e-01 1.34602487e-01 4.69684869e-01 2.99133033e-01
-6.79161131e-01 2.67908990e-01 -6.57155931e-01 3.22171658e-01
7.51055300e-01 -3.08994222e-02 9.59840655e-01 -1.02302337e+00
-5.97657382e-01 -1.16548501e-02 -7.67736197e-01 -8.69281352e-01
1.49264097e-01 -6.08406126e-01 1.57346249e-01 -1.40874147e+00
1.81076318e-01 -1.72714487e-01 2.33679220e-01 3.16281557e-01
-3.85662407e-01 7.57955492e-01 6.31519139e-01 -9.19834897e-02
-5.74858338e-02 3.27879488e-01 1.53847659e+00 -3.64924014e-01
-9.39634219e-02 2.28001848e-01 -2.38091707e-01 7.03477204e-01
5.54028571e-01 -4.06454891e-01 -2.63174385e-01 -2.89638937e-01
-7.63204843e-02 1.49983019e-01 1.08514166e+00 -1.24330175e+00
-1.64420810e-02 1.82409838e-01 9.02465641e-01 -3.09501588e-01
7.13848114e-01 -9.26035583e-01 5.42861581e-01 6.51034474e-01
-3.42618018e-01 3.02396566e-02 1.69831038e-01 5.26068866e-01
2.69354321e-02 -2.53773808e-01 1.06463575e+00 -1.52055472e-01
-3.62207204e-01 3.95984262e-01 1.03453442e-01 1.71064615e-01
1.21192575e+00 -1.74103752e-01 -1.15716554e-01 -4.47435915e-01
-9.94807601e-01 -1.48878485e-01 7.63230562e-01 5.49580991e-01
6.71003699e-01 -1.34828496e+00 -9.37791169e-01 3.15094441e-01
6.34507323e-03 1.55940741e-01 1.47077352e-01 4.76679742e-01
-5.97009957e-01 2.63395738e-02 -3.48821789e-01 -4.62656409e-01
-1.40325570e+00 5.22282779e-01 6.40950263e-01 -2.86733806e-01
-6.80067897e-01 6.23181045e-01 3.33418429e-01 -2.63554990e-01
2.73222089e-01 -1.67937145e-01 7.44528770e-02 -1.86116278e-01
3.10000688e-01 2.60916024e-01 -5.38448878e-02 -9.85513985e-01
-3.19995105e-01 7.41809309e-01 4.53027412e-02 -4.91485327e-01
1.18437886e+00 2.83079237e-01 3.62436414e-01 1.42382845e-01
1.08184028e+00 2.57760525e-01 -1.75602043e+00 -1.01302452e-01
-6.00808263e-01 -5.62963247e-01 -4.68574405e-01 -8.56706321e-01
-1.25078392e+00 5.93518436e-01 8.13969076e-01 -2.86913604e-01
8.89901578e-01 9.15377066e-02 6.87689662e-01 -6.15115948e-02
2.43064418e-01 -7.57926583e-01 2.81351388e-01 1.07483692e-01
1.09363258e+00 -1.00350070e+00 8.66306759e-03 -3.54566664e-01
-5.69745481e-01 8.59182239e-01 4.52527463e-01 -5.50266802e-01
1.60567552e-01 1.60258025e-01 6.51161224e-02 2.81686671e-02
-1.21157363e-01 -1.48735657e-01 6.81374133e-01 7.62363434e-01
2.22932100e-01 -1.62275180e-01 -1.21393546e-01 -1.80651709e-01
-3.50047469e-01 -5.86974993e-02 3.88269722e-01 7.64450312e-01
-3.38178687e-02 -1.38954365e+00 -5.93808591e-01 -8.29401016e-02
-5.00276268e-01 9.38173905e-02 -5.39286315e-01 8.70732307e-01
3.82439584e-01 6.31997347e-01 1.21503830e-01 -5.13306074e-02
6.02380812e-01 -2.34624953e-03 9.25566614e-01 -6.40556157e-01
-6.26567602e-01 1.37341917e-01 -1.21344917e-01 -4.86691713e-01
-5.19598901e-01 -8.45272303e-01 -1.15824997e+00 -3.46819982e-02
5.30596115e-02 -1.05153993e-01 5.94160974e-01 7.89737463e-01
9.20072943e-02 5.89690089e-01 3.18724364e-01 -1.22044587e+00
-4.68657345e-01 -9.56317961e-01 -3.60656142e-01 9.99149323e-01
3.23315501e-01 -4.99808580e-01 -1.79962024e-01 4.39784497e-01] | [11.70260238647461, -0.7311621904373169] |
69234a9f-e266-4a53-aee2-0c835b52bcc8 | learning-to-do-or-learning-while-doing | 2306.03739 | null | https://arxiv.org/abs/2306.03739v1 | https://arxiv.org/pdf/2306.03739v1.pdf | Learning to Do or Learning While Doing: Reinforcement Learning and Bayesian Optimisation for Online Continuous Tuning | Online tuning of real-world plants is a complex optimisation problem that continues to require manual intervention by experienced human operators. Autonomous tuning is a rapidly expanding field of research, where learning-based methods, such as Reinforcement Learning-trained Optimisation (RLO) and Bayesian optimisation (BO), hold great promise for achieving outstanding plant performance and reducing tuning times. Which algorithm to choose in different scenarios, however, remains an open question. Here we present a comparative study using a routine task in a real particle accelerator as an example, showing that RLO generally outperforms BO, but is not always the best choice. Based on the study's results, we provide a clear set of criteria to guide the choice of algorithm for a given tuning task. These can ease the adoption of learning-based autonomous tuning solutions to the operation of complex real-world plants, ultimately improving the availability and pushing the limits of operability of these facilities, thereby enabling scientific and engineering advancements. | ['Holger Schlarb', 'Florian Burkart', 'Thomas Vinatier', 'Frank Mayet', 'Hannes Dinter', 'Willi Kuropka', 'Erik Bründermann', 'Oliver Stein', 'Andrea Santamaria Garcia', 'Annika Eichler', 'Chenran Xu', 'Jan Kaiser'] | 2023-06-06 | null | null | null | null | ['bayesian-optimisation', 'open-question'] | ['methodology', 'natural-language-processing'] | [ 3.20156723e-01 -1.40398711e-01 -2.76206851e-01 -1.29791424e-01
-2.48063996e-01 -7.76342630e-01 4.10100788e-01 4.50731486e-01
-2.35734761e-01 7.24150538e-01 -3.18023950e-01 -8.16589653e-01
-8.10851395e-01 -7.64176846e-01 -2.67046511e-01 -1.05467999e+00
6.64240820e-03 7.81695187e-01 3.24797392e-01 -4.55211610e-01
5.54343224e-01 7.66615152e-01 -1.53259480e+00 -3.07046503e-01
6.97755337e-01 8.64045441e-01 6.29230618e-01 7.88379312e-01
1.56776801e-01 3.93188000e-01 -5.64274013e-01 1.48539901e-01
2.61970848e-01 -3.53112131e-01 -7.09248483e-01 9.09589753e-02
-5.60203016e-01 1.16630502e-01 2.63266385e-01 6.46281779e-01
8.19611549e-01 3.44527185e-01 5.17582715e-01 -6.93389416e-01
-2.06228003e-01 3.50381136e-01 -1.75630450e-01 2.34481871e-01
9.50277299e-02 5.96883416e-01 9.51511502e-01 -8.65322575e-02
6.18169233e-02 9.46821868e-01 4.52492535e-01 8.01503509e-02
-1.59912002e+00 -2.58326024e-01 -1.08063690e-01 5.92181645e-02
-1.25336385e+00 -5.30061305e-01 4.51934725e-01 -5.54445267e-01
1.03527260e+00 2.63005674e-01 6.18996561e-01 3.94976228e-01
4.83900964e-01 2.78225154e-01 9.73111808e-01 -5.81807494e-01
7.67135322e-01 8.29312205e-03 -7.89485037e-01 3.98309052e-01
2.46969104e-01 6.69617176e-01 -4.18201208e-01 -1.66739479e-01
8.17601621e-01 -5.55626810e-01 -2.38393903e-01 -9.32350874e-01
-1.14953566e+00 1.11611378e+00 4.05618727e-01 3.55477035e-01
-5.62811732e-01 1.20045081e-01 5.39665997e-01 2.99309939e-01
1.04504086e-01 1.27491844e+00 -8.43102813e-01 -3.11754942e-01
-7.50065029e-01 4.07042980e-01 9.82848346e-01 4.89234447e-01
4.20363516e-01 2.95815885e-01 -1.20763816e-01 8.15357149e-01
3.99284691e-01 2.96121895e-01 3.63754988e-01 -1.05683148e+00
-3.11178919e-02 4.90528613e-01 5.36406279e-01 -6.11065805e-01
-5.34433842e-01 -6.13166809e-01 -3.27804327e-01 5.46186388e-01
4.79569077e-01 -1.87875807e-01 -4.98632401e-01 1.08913171e+00
3.77092630e-01 -5.40833175e-01 -3.41333449e-02 7.38806367e-01
-6.80638701e-02 5.31783760e-01 2.12808907e-01 -4.89676982e-01
1.04757130e+00 -6.64837539e-01 -5.68970919e-01 -4.45866466e-01
5.72042346e-01 -1.04846406e+00 8.28378439e-01 5.69805086e-01
-7.26356089e-01 -5.24226964e-01 -1.15301073e+00 8.22256505e-01
-4.36971635e-01 3.54853347e-02 9.96345520e-01 9.65797961e-01
-6.52351081e-01 1.12673271e+00 -9.72747147e-01 -5.22784889e-01
8.14855248e-02 7.79943705e-01 1.76595941e-01 1.20574616e-01
-9.22341645e-01 1.33017373e+00 9.45686877e-01 3.68839055e-01
-8.93792093e-01 -4.72493678e-01 -5.41419685e-01 1.90167218e-01
8.99127424e-01 -5.47732532e-01 1.65187597e+00 -2.98222691e-01
-2.08099985e+00 2.77792841e-01 3.01986933e-01 -4.02055442e-01
2.27449775e-01 5.58369653e-03 -3.50675374e-01 -3.45381469e-01
-1.26190379e-01 4.32326674e-01 9.95012105e-01 -1.04221916e+00
-8.10135365e-01 -2.44975775e-01 9.43832621e-02 3.89431715e-01
-8.80059898e-02 -8.18851590e-02 3.71221453e-01 -5.79985790e-02
1.81262225e-01 -1.12107265e+00 -5.60798943e-01 -2.40612626e-01
1.18624374e-01 -3.23961854e-01 7.28877664e-01 -3.11755925e-01
1.16153908e+00 -1.89161325e+00 2.29558721e-01 2.43083477e-01
-4.68486905e-01 3.12586397e-01 2.85966069e-01 7.09301889e-01
1.18032612e-01 -1.18836947e-01 -2.31482953e-01 4.48293865e-01
1.16286702e-01 4.46347356e-01 5.98783009e-02 2.86632925e-01
2.74414212e-01 4.59672987e-01 -1.18644631e+00 4.93132882e-02
6.50267541e-01 -8.10429454e-02 -2.71610916e-01 2.88053066e-01
-4.34991539e-01 8.70596886e-01 -6.59429669e-01 7.05565333e-01
1.14533026e-02 -2.83615962e-02 2.86089152e-01 7.06969798e-02
-4.50197071e-01 -2.17625522e-03 -1.35294271e+00 1.35786617e+00
-6.34183049e-01 2.85735160e-01 2.18718767e-01 -1.05502260e+00
1.26739717e+00 3.90566021e-01 5.67701817e-01 -4.38096732e-01
2.05839038e-01 4.69820410e-01 5.33768237e-01 -5.26370108e-01
3.32788676e-01 -2.09957436e-01 1.26243532e-01 1.53760180e-01
7.28433952e-02 -1.04507577e+00 4.77271557e-01 -5.05619109e-01
1.14243913e+00 4.09118563e-01 7.32414186e-01 -7.40875661e-01
6.17363989e-01 3.28526199e-01 4.76992577e-01 7.75119424e-01
-4.56649929e-01 -2.02896774e-01 2.70830482e-01 -4.81181741e-01
-1.02434385e+00 -5.80970347e-01 -3.38142782e-01 1.37235069e+00
-6.78897426e-02 -7.98082426e-02 -2.85018265e-01 -1.48282766e-01
2.60710746e-01 9.79305446e-01 -9.60219279e-02 -2.02677652e-01
-2.67864078e-01 -9.40994978e-01 -1.06806785e-01 2.47117177e-01
1.52619511e-01 -1.26348722e+00 -1.06007934e+00 7.20952272e-01
2.57961899e-01 -6.98741794e-01 4.75728028e-02 1.08579862e+00
-1.06854975e+00 -7.19611704e-01 -4.72041100e-01 -3.24045002e-01
3.08066159e-01 7.70735964e-02 1.18303549e+00 -8.73126760e-02
-4.16423440e-01 2.06068024e-01 -3.62265468e-01 -8.22914541e-01
-8.09065998e-01 4.35430855e-01 7.81765655e-02 -5.78113496e-01
4.98614274e-02 -4.07506526e-01 -5.99206686e-01 7.61530757e-01
-6.09065473e-01 -4.53919351e-01 7.33520269e-01 1.09521985e+00
4.69374508e-01 7.57782459e-01 7.09358037e-01 -7.63368666e-01
8.00761998e-01 -2.63593376e-01 -1.17413545e+00 3.81809711e-01
-9.19795156e-01 2.47944266e-01 4.93361562e-01 -3.17836612e-01
-8.81420791e-01 4.38332468e-01 5.05689457e-02 1.94203183e-01
-3.18320394e-01 5.62037945e-01 -7.07084239e-02 -4.99573141e-01
9.43780065e-01 -1.22602701e-01 5.98457195e-02 -1.13553725e-01
1.26673236e-01 5.31705260e-01 2.65023291e-01 -5.46412230e-01
7.29612052e-01 -6.69652224e-02 1.65758029e-01 -8.35334539e-01
-7.61472583e-01 -6.29432440e-01 -6.91111624e-01 -4.18614298e-01
6.19710386e-01 -5.15625775e-01 -9.58764315e-01 -2.53977608e-02
-4.08490539e-01 -6.82973027e-01 -2.71329790e-01 4.17619854e-01
-9.88707304e-01 -8.43223780e-02 9.64081362e-02 -9.07396793e-01
-1.58349231e-01 -1.53071046e+00 1.05028379e+00 5.47944188e-01
-5.36129415e-01 -9.62963462e-01 6.78837672e-02 3.65315467e-01
8.08903039e-01 2.31710449e-01 9.80786026e-01 -3.27282548e-01
-4.63052481e-01 -2.51566470e-01 3.22261095e-01 5.54964021e-02
3.76532644e-01 2.86916047e-01 -8.18611860e-01 -5.82208216e-01
1.45136878e-01 -4.07246470e-01 2.39021227e-01 5.05350411e-01
1.07276642e+00 3.21097404e-01 -3.41973335e-01 1.81516483e-01
1.38257873e+00 6.36399388e-01 2.46847808e-01 6.36797667e-01
6.21199347e-02 4.86858815e-01 1.16589618e+00 6.40970349e-01
-3.65961015e-01 7.40608454e-01 7.64655828e-01 8.15779567e-02
5.17874479e-01 3.39780264e-02 1.14180170e-01 4.61477846e-01
-1.45978451e-01 -1.42157555e-01 -9.46132362e-01 3.65363538e-01
-1.91162765e+00 -8.17839921e-01 2.60872960e-01 2.56085277e+00
5.26128411e-01 3.81285667e-01 5.89001998e-02 4.10467952e-01
5.99367261e-01 -2.08903644e-02 -8.64432216e-01 -6.24597311e-01
3.69488955e-01 1.68448254e-01 9.18633401e-01 1.50247127e-01
-1.02950644e+00 6.50106430e-01 6.73334503e+00 8.02184522e-01
-1.02381229e+00 -3.00067633e-01 3.24876815e-01 2.95197457e-01
2.41211995e-01 2.70772755e-01 -7.07704544e-01 2.29677424e-01
1.16351163e+00 -2.25773081e-01 1.05513310e+00 8.92447233e-01
5.79883873e-01 -6.40094221e-01 -9.57432449e-01 4.31387216e-01
-5.77633858e-01 -1.07782459e+00 -6.57629788e-01 5.27702905e-02
6.01995826e-01 -2.87275285e-01 -1.27827555e-01 4.80431944e-01
6.25304639e-01 -8.96016538e-01 3.89935434e-01 1.22222967e-01
1.67665809e-01 -9.61525679e-01 8.28941047e-01 4.49841052e-01
-1.03505719e+00 -6.88512504e-01 -4.60544378e-01 4.57939841e-02
3.10827583e-01 6.49939954e-01 -1.13985848e+00 7.99422562e-01
7.30369985e-01 1.18861206e-01 -5.11179745e-01 1.33485806e+00
-2.32743472e-01 6.50385201e-01 -3.60783279e-01 -3.10473979e-01
2.67668426e-01 -1.18029222e-01 4.38102186e-01 6.36765063e-01
2.38567367e-01 -1.38489947e-01 2.81587780e-01 7.02669203e-01
6.33778930e-01 6.31649047e-02 -5.44414699e-01 -3.48851949e-01
5.82967937e-01 1.32202494e+00 -1.18222868e+00 1.43208057e-01
8.00468549e-02 4.64083761e-01 1.67576298e-02 -1.96308941e-01
-6.57226503e-01 -3.13611448e-01 3.12614173e-01 -1.50882840e-01
4.85065967e-01 -5.19083261e-01 -3.51164907e-01 -1.56054020e-01
-5.27811170e-01 -1.02634692e+00 3.85539681e-01 -6.42048180e-01
-8.39254856e-01 3.47220339e-02 8.78673643e-02 -9.49315965e-01
-3.97393912e-01 -8.59867990e-01 -4.93136615e-01 7.11362362e-01
-1.17141902e+00 -4.63257164e-01 -2.82474846e-01 -1.74530625e-01
7.14500606e-01 -3.67405027e-01 1.07736015e+00 -1.30057231e-01
-4.57556665e-01 -1.50126815e-01 6.16362989e-01 -7.86111593e-01
6.77201867e-01 -1.36054635e+00 2.97854021e-02 2.63538808e-01
5.02622314e-03 2.14612573e-01 1.13437021e+00 -5.02288759e-01
-1.51794052e+00 -6.89386249e-01 3.09433073e-01 -2.57235020e-01
9.05372560e-01 7.00516030e-02 -7.08014905e-01 1.01440348e-01
1.03367209e-01 -5.47879040e-01 5.77319205e-01 3.27206343e-01
6.28432095e-01 -1.88950732e-01 -1.09160686e+00 5.56130171e-01
5.53124905e-01 -2.22090766e-01 -1.75180852e-01 7.03981757e-01
5.06115317e-01 -5.80929160e-01 -1.40294576e+00 7.29448318e-01
3.33203822e-01 -7.06426561e-01 8.50999832e-01 -2.53032148e-01
-1.06090702e-01 -5.73476911e-01 1.73031822e-01 -1.72787964e+00
-6.27043486e-01 -9.58560109e-01 9.78738219e-02 1.04465783e+00
4.40116405e-01 -5.91362059e-01 6.46913648e-01 6.22231603e-01
-7.96570722e-03 -8.26181948e-01 -6.27715170e-01 -8.84887755e-01
-1.88797668e-01 2.66763400e-02 6.55485153e-01 5.22938609e-01
-1.07294671e-01 2.61618853e-01 -3.20321769e-02 2.69679695e-01
3.99866790e-01 -4.04216982e-02 6.58616662e-01 -1.22613871e+00
-5.15281677e-01 -4.61925060e-01 -3.23981375e-01 -3.56599510e-01
-1.83569059e-01 -3.36995244e-01 5.22219658e-01 -1.19474721e+00
-5.51025629e-01 -6.11183524e-01 -1.67274371e-01 2.77501345e-01
-1.11500807e-01 -3.19293946e-01 -8.59320164e-02 4.36437875e-03
-7.52590299e-02 5.21687806e-01 1.36220837e+00 -9.17809457e-02
-5.67226529e-01 5.60640931e-01 -5.32435715e-01 4.96681601e-01
1.08033037e+00 -5.41321278e-01 -5.78345835e-01 -6.34755008e-03
1.68724820e-01 2.12146252e-01 -1.49704263e-01 -9.97449398e-01
2.30058685e-01 -5.89636981e-01 3.56921941e-01 -1.26418650e-01
1.37023121e-01 -1.01270771e+00 3.61186266e-01 7.79828489e-01
-2.46767908e-01 2.56641895e-01 4.43753242e-01 6.32223964e-01
8.79828259e-02 -8.69033754e-01 9.30124760e-01 -1.73918888e-01
-8.24747920e-01 -1.30974129e-01 -4.81445402e-01 -3.44170243e-01
1.24554265e+00 -7.69962072e-02 8.29942152e-02 -1.02185629e-01
-9.32846546e-01 4.44491804e-01 1.16610579e-01 2.20271632e-01
-1.53672516e-01 -8.16851020e-01 -5.42404354e-01 7.01002851e-02
1.91087827e-01 4.71918993e-02 -1.94837227e-02 6.44178569e-01
-6.97415829e-01 5.60477793e-01 -2.85745680e-01 -8.33464026e-01
-1.01658678e+00 5.53682685e-01 3.70270640e-01 -4.78949308e-01
-1.48367643e-01 6.44091249e-01 -4.48189974e-01 -5.40101647e-01
6.17904663e-02 -2.59143859e-02 -1.40248403e-01 -1.04185477e-01
2.74525061e-02 4.21817511e-01 5.50591409e-01 -1.09710418e-01
-4.87334244e-02 8.78941417e-02 6.43627793e-02 -8.60302988e-03
1.14492762e+00 1.01405174e-01 1.39692679e-01 4.16812927e-01
1.41821399e-01 -6.21104836e-01 -1.30514681e+00 1.13212995e-01
5.24957538e-01 -7.00654089e-01 4.49916571e-01 -8.66550386e-01
-7.30040073e-01 3.99951935e-01 7.53176987e-01 5.47598064e-01
1.05285954e+00 -3.55078816e-01 -2.49560550e-02 7.50868499e-01
5.71434557e-01 -1.42868125e+00 -1.37518495e-01 3.08673888e-01
8.78683269e-01 -1.20942557e+00 3.68558288e-01 -1.53310731e-01
-5.42800069e-01 1.23202229e+00 4.69444305e-01 1.13658845e-01
5.25749981e-01 1.65879592e-01 -7.03549534e-02 -1.08140074e-01
-8.55505407e-01 -3.72448802e-01 -1.56171158e-01 6.37446702e-01
4.25423533e-01 2.22102791e-01 -2.57909358e-01 -3.31521600e-01
-1.89421445e-01 6.42662635e-03 1.88560322e-01 1.27014387e+00
-8.05727720e-01 -1.36174881e+00 -6.54751122e-01 6.61651611e-01
-7.99860209e-02 3.58432114e-01 -7.53171742e-02 6.84983253e-01
-9.42297932e-03 1.10118973e+00 -3.98990035e-01 8.74219909e-02
5.73345602e-01 1.18254595e-01 5.04005134e-01 -6.94667518e-01
-7.19137549e-01 5.50159775e-02 1.60704777e-01 -2.05347195e-01
-2.47714132e-01 -7.18343854e-01 -6.86298013e-01 -2.76225835e-01
-9.42885578e-01 2.00042278e-01 1.12921488e+00 9.42527056e-01
1.17413551e-01 9.93794560e-01 7.34843612e-01 -1.21501267e+00
-1.32223475e+00 -7.36180723e-01 -6.86541557e-01 -2.91416556e-01
-1.77358434e-01 -1.08817506e+00 -2.49316737e-01 -1.66729167e-01] | [4.996670722961426, 2.378217935562134] |
827a6e52-e6b7-40a0-a917-c426ed3dcd75 | curating-a-covid-19-data-repository-and | 2005.07882 | null | https://arxiv.org/abs/2005.07882v2 | https://arxiv.org/pdf/2005.07882v2.pdf | Curating a COVID-19 data repository and forecasting county-level death counts in the United States | As the COVID-19 outbreak evolves, accurate forecasting continues to play an extremely important role in informing policy decisions. In this paper, we present our continuous curation of a large data repository containing COVID-19 information from a range of sources. We use this data to develop predictions and corresponding prediction intervals for the short-term trajectory of COVID-19 cumulative death counts at the county-level in the United States up to two weeks ahead. Using data from January 22 to June 20, 2020, we develop and combine multiple forecasts using ensembling techniques, resulting in an ensemble we refer to as Combined Linear and Exponential Predictors (CLEP). Our individual predictors include county-specific exponential and linear predictors, a shared exponential predictor that pools data together across counties, an expanded shared exponential predictor that uses data from neighboring counties, and a demographics-based shared exponential predictor. We use prediction errors from the past five days to assess the uncertainty of our death predictions, resulting in generally-applicable prediction intervals, Maximum (absolute) Error Prediction Intervals (MEPI). MEPI achieves a coverage rate of more than 94% when averaged across counties for predicting cumulative recorded death counts two weeks in the future. Our forecasts are currently being used by the non-profit organization, Response4Life, to determine the medical supply need for individual hospitals and have directly contributed to the distribution of medical supplies across the country. We hope that our forecasts and data repository at https://covidseverity.com can help guide necessary county-specific decision-making and help counties prepare for their continued fight against COVID-19. | ['Tiffany Tang', 'Yan Shuo Tan', 'Raaz Dwivedi', 'Briton Park', 'Yu Wang', 'Xiao Li', 'Robert Netzorg', 'Bin Yu', 'Rebecca L. Barter', 'James Duncan', 'Nick Altieri', 'Karl Kumbier', 'Chao Zhang', 'Chandan Singh'] | 2020-05-16 | null | null | null | null | ['covid-19-tracking'] | ['time-series'] | [-3.27382088e-01 -1.10875838e-01 -3.47032756e-01 -4.04358625e-01
-9.58318591e-01 -4.66930270e-01 5.36479712e-01 9.17345643e-01
-3.56089294e-01 1.00392139e+00 8.75397563e-01 -9.40620184e-01
-1.75211698e-01 -8.35713983e-01 -3.17107230e-01 -3.67927074e-01
-3.91743064e-01 3.94904256e-01 -2.62424916e-01 -7.72921070e-02
-2.80445572e-02 4.86167014e-01 -7.12723494e-01 1.32116050e-01
1.12924910e+00 5.66604316e-01 2.37926636e-02 6.36208534e-01
3.84894073e-01 7.15667188e-01 -2.82115221e-01 3.13653611e-03
1.57260939e-01 -1.57562017e-01 -2.55667090e-01 -8.79160702e-01
-4.69187558e-01 -6.73868656e-01 -6.69286028e-02 2.90248781e-01
3.17165941e-01 -2.23707497e-01 9.59239364e-01 -1.06172609e+00
-2.70234644e-01 5.25847912e-01 -4.12416250e-01 4.14699018e-01
2.07684144e-01 3.12141359e-01 2.59167939e-01 -6.42708540e-01
5.55280268e-01 8.72298896e-01 1.41171980e+00 1.26714602e-01
-1.35753369e+00 -8.75705361e-01 3.05545032e-02 -2.41053388e-01
-1.47128439e+00 -4.01615292e-01 1.39726158e-02 -1.01367402e+00
1.24182093e+00 2.45071694e-01 7.18599200e-01 9.02467668e-01
8.47141564e-01 -3.23529057e-02 9.63906586e-01 -6.89267321e-03
1.24101043e-01 9.21821520e-02 1.55139878e-01 2.56288238e-02
5.27526081e-01 4.72952873e-01 9.95532796e-02 -8.91138375e-01
5.51503658e-01 5.62329173e-01 -1.26704365e-01 4.20448899e-01
-1.27386212e+00 9.12656903e-01 1.80595145e-01 -1.60501823e-01
-8.97998154e-01 9.84029248e-02 2.15048432e-01 -6.10548146e-02
9.20506179e-01 2.54469246e-01 -9.51455116e-01 -1.45142913e-01
-9.90314126e-01 4.17463064e-01 7.39979863e-01 3.64684820e-01
3.13258976e-01 -5.47937341e-02 -8.55101943e-02 6.46251023e-01
1.71886727e-01 1.11188686e+00 -1.57119423e-01 -8.87216210e-01
3.19579899e-01 2.15499505e-01 6.35046005e-01 -1.11049008e+00
-9.93168533e-01 -2.79021323e-01 -9.00387824e-01 -2.54451126e-01
3.57337475e-01 -9.85232294e-01 -6.00159466e-01 1.80915809e+00
1.66369990e-01 2.52191812e-01 1.10329449e-01 4.02346104e-01
3.20150793e-01 9.60493207e-01 5.57832837e-01 -4.72042233e-01
1.12842357e+00 -1.14531688e-01 -4.57854658e-01 1.11063063e-01
8.02399695e-01 -5.42174995e-01 -4.91222069e-02 -2.68553458e-02
-5.78366637e-01 -4.38361466e-02 -3.95525277e-01 7.89173543e-01
-3.73636872e-01 -1.80578921e-02 3.43767703e-01 5.70467860e-02
-9.04668868e-01 3.22811991e-01 -1.18432510e+00 -6.98164582e-01
2.24903136e-01 -2.28849240e-02 -1.69858456e-01 1.84573591e-01
-1.31438553e+00 1.12622130e+00 7.75350034e-02 1.71519183e-02
-6.72799110e-01 -1.26410937e+00 -8.22514057e-01 -1.31224953e-02
-3.17986101e-01 -7.38892794e-01 8.37987244e-01 -1.57511175e-01
-3.50602448e-01 3.08174074e-01 -3.61033589e-01 -7.76653886e-01
4.11406070e-01 7.66696483e-02 -7.15324581e-01 -2.19588250e-01
5.07265568e-01 3.06029230e-01 1.33761585e-01 -8.65765929e-01
-7.99506605e-01 -4.15427089e-01 -7.47897446e-01 -4.21131141e-02
-5.77118136e-02 3.98582131e-01 3.84157866e-01 -8.57243359e-01
-3.36992592e-01 -8.95699322e-01 -7.56576061e-01 -6.79639637e-01
-1.18578836e-01 -5.40356599e-02 3.46548498e-01 -1.38950288e+00
1.57490659e+00 -1.90239358e+00 -4.88766491e-01 1.30403325e-01
9.44551080e-02 -1.49667710e-01 8.13930258e-02 8.36433649e-01
-1.06408894e-01 1.39970258e-01 -2.80074269e-01 -7.29767159e-02
-5.61151505e-01 -1.19056560e-01 -5.20115137e-01 5.48849761e-01
3.81502450e-01 5.99963963e-01 -9.84777331e-01 -1.38412356e-01
2.99537152e-01 4.10854489e-01 -4.61465627e-01 8.32257196e-02
2.03526035e-01 7.07157969e-01 -3.08946520e-01 5.32932699e-01
7.46782899e-01 -2.65847653e-01 1.27776310e-01 3.20362240e-01
-5.37917972e-01 8.49821195e-02 -4.51401025e-01 7.24664986e-01
-4.71814007e-01 3.49509716e-01 2.63104471e-03 -6.54670119e-01
1.08480942e+00 3.69670004e-01 9.54995453e-01 -2.16875926e-01
-2.40893617e-01 2.40853339e-01 -6.25470728e-02 -2.21370533e-01
3.67639542e-01 -1.87274128e-01 -3.54277670e-01 6.34754181e-01
-5.40762722e-01 1.17464006e-01 -9.78532657e-02 1.70879856e-01
1.19958901e+00 -1.64075181e-01 5.68857014e-01 -6.52936101e-01
-1.17101278e-02 6.72547162e-01 7.94003367e-01 4.11459565e-01
-2.64467925e-01 2.44481564e-01 5.66556096e-01 -7.90526807e-01
-1.13503850e+00 -1.13113773e+00 -7.76930809e-01 4.41796929e-01
-4.26461667e-01 -1.63908005e-01 -1.09711416e-01 -2.02588186e-01
4.33673710e-01 1.02146924e+00 -7.85123885e-01 9.73654818e-03
-3.72149557e-01 -1.14377129e+00 5.20061672e-01 4.97790933e-01
9.99580100e-02 -5.28958380e-01 -8.83011699e-01 6.53290451e-01
-2.92288750e-01 -5.79415500e-01 -2.80159146e-01 2.64712662e-01
-6.51669562e-01 -1.11353254e+00 -9.80891883e-01 -2.42509574e-01
4.89950061e-01 -1.29985943e-01 9.47842836e-01 -3.09481740e-01
-1.40898794e-01 3.49972904e-01 -2.48817429e-02 -8.21803570e-01
-4.25924659e-01 -1.58658266e-01 3.41474235e-01 -6.78655982e-01
5.69206774e-01 -1.99764416e-01 -7.97706306e-01 5.65731414e-02
-2.34086603e-01 2.19842214e-02 1.77171037e-01 7.15693533e-01
5.00903070e-01 -3.92691910e-01 1.36144388e+00 -6.84292495e-01
6.85544133e-01 -1.38845670e+00 -6.42965853e-01 1.49413392e-01
-7.34623551e-01 -2.78051376e-01 6.07352376e-01 -6.62985146e-02
-1.01591825e+00 -7.61808008e-02 1.27649486e-01 -7.62956217e-02
-1.61808744e-01 1.04284334e+00 8.64569008e-01 7.99686015e-01
5.05866706e-01 7.12093636e-02 8.89319852e-02 -3.29807132e-01
-2.00948328e-01 8.56473207e-01 4.98313725e-01 -2.74158686e-01
3.67460579e-01 3.49015504e-01 -1.81728661e-01 -4.67538297e-01
-2.70465374e-01 -5.26114225e-01 -2.07731009e-01 -1.58299878e-01
9.40351963e-01 -1.51962352e+00 -6.08989716e-01 5.41204751e-01
-1.05389297e+00 -6.01825297e-01 1.93513781e-01 8.39527547e-01
-4.15526122e-01 -2.49305815e-01 -6.14507437e-01 -9.81313884e-01
-4.72653687e-01 -8.58055830e-01 6.15029931e-01 -1.72615163e-02
-7.79928565e-01 -1.38985789e+00 7.97443986e-01 -1.40875056e-01
7.91254878e-01 8.97094727e-01 9.67256069e-01 -7.24817216e-01
7.26870671e-02 -1.39044642e-01 -2.26027325e-01 -5.74470125e-02
3.63695592e-01 2.12267712e-01 -3.50042313e-01 -2.63270408e-01
-4.96111870e-01 1.70162559e-01 8.28713715e-01 1.01686811e+00
8.51718664e-01 -5.20936310e-01 -8.85043681e-01 4.78086233e-01
1.43007863e+00 5.32502532e-01 1.88901141e-01 1.49653852e-01
6.17517456e-02 6.84628427e-01 6.29698873e-01 9.99757171e-01
8.12536895e-01 1.98895514e-01 1.15368880e-01 -1.55957997e-01
4.16995615e-01 -3.33599359e-01 1.84189156e-02 4.50461864e-01
-2.63829559e-01 1.78302139e-01 -1.76992214e+00 1.05195177e+00
-1.64510918e+00 -1.18280661e+00 -3.67185771e-01 2.49844074e+00
7.45993137e-01 -2.70067215e-01 7.94944093e-02 -7.24881172e-01
5.10716319e-01 -1.44064263e-01 -6.30715132e-01 -6.55543983e-01
1.95141539e-01 -1.68025777e-01 1.01774180e+00 4.30591315e-01
-1.10253692e+00 2.93126315e-01 7.42741966e+00 2.86871284e-01
-1.10543537e+00 -1.09328486e-01 1.27872324e+00 2.38037389e-02
-4.38521683e-01 2.70523783e-02 -7.35549748e-01 6.17361784e-01
1.57843089e+00 -6.79570556e-01 2.52724648e-01 7.08602190e-01
8.87560725e-01 -2.53631204e-01 -5.74778199e-01 3.86838257e-01
-3.72982502e-01 -1.63359678e+00 -6.80506527e-01 2.49556884e-01
1.01745689e+00 5.55864215e-01 2.25275140e-02 2.88979650e-01
9.53915715e-01 -1.12331736e+00 3.34968179e-01 1.03470147e+00
1.23699975e+00 -1.05861986e+00 1.03557909e+00 4.62571740e-01
-1.16125298e+00 -1.27881989e-01 -4.48231399e-02 -1.78307980e-01
4.75639015e-01 8.35573435e-01 -1.12480187e+00 3.00075144e-01
8.45651150e-01 8.91745746e-01 -8.56412053e-02 7.23922193e-01
4.35181081e-01 9.08557236e-01 -3.69708538e-01 2.90184289e-01
1.22055173e-01 -1.02295227e-01 3.26726258e-01 1.08721173e+00
7.41428792e-01 5.67210734e-01 -6.34005293e-02 6.55313849e-01
3.18624973e-01 -1.15055181e-01 -7.50467122e-01 1.14412963e-01
8.46676886e-01 9.29669619e-01 -1.98535711e-01 -3.22875947e-01
-3.95915031e-01 2.61415690e-01 2.62007676e-03 5.43870866e-01
-7.97995329e-01 -3.12881619e-01 1.03398705e+00 3.59606594e-01
-1.04760900e-01 -1.91663116e-01 -5.09463131e-01 -8.58374476e-01
-6.75526559e-01 -5.09772480e-01 3.89425844e-01 -5.13355613e-01
-1.37924564e+00 5.21631479e-01 4.19714659e-01 -1.08590603e+00
-7.64545918e-01 -1.28659144e-01 -8.22215259e-01 1.45323193e+00
-1.21590376e+00 -8.31800878e-01 -7.70537322e-03 2.15429574e-01
1.77115589e-01 1.14056475e-01 1.04133582e+00 -1.43017933e-01
-6.21948779e-01 2.11715162e-01 7.48714030e-01 6.13670237e-02
8.90863597e-01 -7.01924980e-01 2.86247492e-01 4.42582011e-01
-9.16028798e-01 7.95985878e-01 6.58935189e-01 -1.17005551e+00
-6.84643984e-01 -1.58790600e+00 1.17159700e+00 -6.07089460e-01
8.22508574e-01 1.54120058e-01 -6.45417869e-01 8.86777163e-01
-4.25170362e-02 -3.42877477e-01 9.71073627e-01 1.22733861e-01
1.95927620e-01 -1.54323682e-01 -1.34869909e+00 1.80080995e-01
3.13152283e-01 -1.00086480e-01 -3.53415757e-01 2.46873260e-01
6.45427346e-01 1.87345631e-02 -1.42004561e+00 5.31733155e-01
6.58217430e-01 -6.53517246e-01 6.83572054e-01 -6.79342926e-01
5.27332604e-01 1.39696538e-01 -2.42277071e-01 -1.77181196e+00
-5.81672609e-01 -2.58501619e-01 3.53631198e-01 8.37904871e-01
7.34429121e-01 -1.25737321e+00 2.06176832e-01 8.96866679e-01
-6.84225634e-02 -9.97183800e-01 -9.57728207e-01 -5.56769788e-01
6.11794949e-01 -3.01114082e-01 1.03318584e+00 1.01967764e+00
2.47333735e-01 -4.00174886e-01 -4.20453250e-01 4.31936055e-01
5.97531676e-01 -9.93702747e-03 5.38611412e-01 -1.12639391e+00
4.06039245e-02 -4.95750792e-02 8.49036779e-03 -3.25818479e-01
-1.70925766e-01 -4.02800292e-01 -2.93304265e-01 -1.74715161e+00
5.84722042e-01 -1.00112545e+00 -2.90372133e-01 7.81595170e-01
-2.83291310e-01 -1.66364446e-01 -1.28862914e-02 4.28167373e-01
1.69074520e-01 2.53940970e-01 5.20036399e-01 7.96723738e-02
-3.46511841e-01 4.84384634e-02 -6.44024372e-01 5.34788430e-01
1.05359745e+00 -5.42134762e-01 -4.47204560e-02 -2.27053612e-01
6.27031401e-02 6.92127228e-01 3.74853522e-01 -9.29927349e-01
3.98393646e-02 -8.36988568e-01 6.80912137e-01 -1.10125673e+00
-7.85578936e-02 -7.74962842e-01 9.35993791e-01 1.13053250e+00
-1.57855496e-01 4.10242975e-01 5.61288536e-01 5.15540302e-01
7.00006783e-02 5.28606296e-01 3.91617298e-01 1.00189842e-01
-6.23132549e-02 3.69168460e-01 -8.71331215e-01 -7.55588114e-02
1.39129972e+00 1.67660370e-01 -6.37074113e-01 -5.91845393e-01
-7.07915723e-01 7.31841266e-01 5.30250967e-01 1.76185206e-01
3.25135529e-01 -1.13410640e+00 -1.39033830e+00 8.96273479e-02
5.91627061e-02 -4.27065343e-01 4.08768535e-01 1.23046994e+00
-5.87215483e-01 7.95162618e-01 -8.48614275e-02 -3.23888630e-01
-7.52318501e-01 5.12004435e-01 3.39512229e-01 -3.98402780e-01
-3.37047964e-01 3.49960804e-01 7.47623965e-02 -3.94576550e-01
-4.21339363e-01 -3.36572707e-01 -3.37456428e-02 3.26166414e-02
7.30783105e-01 5.86432457e-01 -4.62315142e-01 -6.81006193e-01
-7.16421187e-01 1.69844955e-01 2.71863937e-01 3.38703506e-02
1.66967797e+00 -1.84225798e-01 -1.67136490e-02 6.35760725e-01
1.01375723e+00 -1.75477192e-02 -1.29196846e+00 3.35183710e-01
-2.39208147e-01 -8.96847174e-02 -1.81409016e-01 -1.11637831e+00
-5.43800652e-01 4.26284790e-01 5.31947076e-01 -2.71244789e-03
1.06191909e+00 -2.67182868e-02 7.96648860e-01 -2.88815498e-01
3.95914197e-01 -6.50374711e-01 -8.88664663e-01 3.15828264e-01
9.34235275e-01 -1.25347674e+00 1.85979575e-01 1.37426347e-01
-7.07973063e-01 7.91407108e-01 7.00636208e-02 5.79641834e-02
1.04815280e+00 4.26401436e-01 7.24390298e-02 1.49787605e-01
-1.28322947e+00 3.39383394e-01 -1.44631982e-01 7.88606048e-01
4.04994607e-01 8.41317415e-01 -5.98559499e-01 9.10461783e-01
8.09165165e-02 1.69852316e-01 3.82775426e-01 3.58163297e-01
-2.95403302e-01 -7.42822587e-01 -7.12659955e-01 1.25793135e+00
-4.12926525e-01 -4.24874991e-01 1.30440921e-01 5.87666273e-01
1.27783522e-01 1.11594212e+00 4.03559864e-01 -2.24828243e-01
5.36002265e-03 -5.59617206e-02 -4.19865698e-01 -1.88735604e-01
-4.35360998e-01 -1.76359162e-01 3.56469601e-01 -3.75684708e-01
-5.05761169e-02 -1.08091915e+00 -1.22814441e+00 -8.11771691e-01
2.58536898e-02 2.47747123e-01 6.36868298e-01 5.64259708e-01
7.78217912e-01 2.22513273e-01 8.31042886e-01 -5.84049344e-01
-7.64561176e-01 -1.09635389e+00 -3.01043779e-01 -1.88734695e-01
4.50730205e-01 -3.83170694e-01 -4.96442556e-01 -1.55277938e-01] | [6.051515579223633, 4.4234700202941895] |
73815919-85a8-4e04-b6d3-22849c0f0bd6 | iteratively-improving-speech-recognition-and | 2305.15055 | null | https://arxiv.org/abs/2305.15055v1 | https://arxiv.org/pdf/2305.15055v1.pdf | Iteratively Improving Speech Recognition and Voice Conversion | Many existing works on voice conversion (VC) tasks use automatic speech recognition (ASR) models for ensuring linguistic consistency between source and converted samples. However, for the low-data resource domains, training a high-quality ASR remains to be a challenging task. In this work, we propose a novel iterative way of improving both the ASR and VC models. We first train an ASR model which is used to ensure content preservation while training a VC model. In the next iteration, the VC model is used as a data augmentation method to further fine-tune the ASR model and generalize it to diverse speakers. By iteratively leveraging the improved ASR model to train VC model and vice-versa, we experimentally show improvement in both the models. Our proposed framework outperforms the ASR and one-shot VC baseline models on English singing and Hindi speech domains in subjective and objective evaluations in low-data resource settings. | ['Onoe Naoyuki', 'Naoya Takahashi', 'Mayank Kumar Singh'] | 2023-05-24 | null | null | null | null | ['voice-conversion', 'voice-conversion', 'automatic-speech-recognition'] | ['audio', 'speech', 'speech'] | [ 3.93814653e-01 -2.53808171e-01 -1.40180081e-01 -3.71475130e-01
-1.11026406e+00 -6.69119656e-01 4.66999650e-01 -2.80544192e-01
-3.08795005e-01 6.53036535e-01 5.04928291e-01 -4.57959592e-01
3.90764683e-01 -3.37129295e-01 -5.46166301e-01 -3.94866973e-01
4.03030038e-01 3.84531230e-01 2.06313923e-01 -3.52210522e-01
-5.92764169e-02 2.98074186e-01 -1.48268545e+00 2.13465303e-01
1.07414019e+00 6.52083993e-01 5.27974784e-01 1.00784004e+00
-2.31433496e-01 7.06230342e-01 -7.90154219e-01 -1.50330752e-01
3.20092559e-01 -8.64348650e-01 -8.11696649e-01 1.13169022e-01
3.07787865e-01 -7.93372691e-02 -3.61364603e-01 8.60883176e-01
8.63191903e-01 7.16384888e-01 2.78529793e-01 -7.95561850e-01
-8.96577954e-01 8.98440421e-01 -2.18032598e-01 2.72278398e-01
2.77243346e-01 -8.96716863e-03 1.10786927e+00 -9.58970070e-01
4.59202558e-01 1.21752405e+00 3.12506378e-01 1.08650744e+00
-1.16071415e+00 -7.43556142e-01 2.44454235e-01 9.82736796e-02
-1.28756189e+00 -1.09068084e+00 8.70649934e-01 9.77021549e-03
8.45442414e-01 5.04664302e-01 3.95814419e-01 8.60972047e-01
-7.57207334e-01 8.31999183e-01 1.01613963e+00 -6.61808729e-01
1.76089019e-01 3.12042147e-01 -1.00051150e-01 1.67027518e-01
-4.37960654e-01 2.94845458e-02 -9.07191157e-01 1.29091159e-01
6.25981271e-01 -4.77778584e-01 -4.78318721e-01 2.33846277e-01
-9.70841110e-01 6.89587057e-01 2.18047202e-01 4.83586818e-01
-2.85263360e-01 -1.58429787e-01 3.35766405e-01 5.78074515e-01
5.94537914e-01 6.79311454e-01 -4.50751930e-01 -4.74306464e-01
-1.22458100e+00 -6.63199723e-02 7.12636232e-01 1.10552263e+00
2.47560129e-01 4.51018721e-01 -4.34563160e-01 1.76120484e+00
1.81135133e-01 3.82031381e-01 1.01051688e+00 -7.71433055e-01
6.55987740e-01 2.55475789e-01 -1.21884398e-01 -2.29196385e-01
1.86553955e-01 -6.98517501e-01 -6.33087695e-01 -1.21600479e-01
1.31932469e-02 4.82515581e-02 -1.07582140e+00 1.78245723e+00
2.90517658e-01 3.52301985e-01 3.10491413e-01 7.99363136e-01
8.78506780e-01 9.88556802e-01 -1.49772331e-01 -5.48807442e-01
9.24381316e-01 -1.37838709e+00 -1.09206128e+00 -2.67653555e-01
3.42446059e-01 -1.07217884e+00 1.71153927e+00 3.10299486e-01
-1.19908178e+00 -8.65104437e-01 -1.16469967e+00 -1.61941677e-01
-3.25061269e-02 4.55447733e-01 9.84541327e-02 8.18448007e-01
-1.02997863e+00 4.50898588e-01 -6.80681825e-01 -2.24996626e-01
2.67263830e-01 2.30292708e-01 -2.11854443e-01 -6.76756203e-02
-1.23584473e+00 5.01203954e-01 1.01660289e-01 -6.28557336e-03
-1.11911213e+00 -7.37905622e-01 -7.29090929e-01 -4.55479249e-02
3.85996968e-01 -1.32391721e-01 1.56992745e+00 -8.39218616e-01
-2.20288420e+00 6.73870981e-01 -3.75539660e-01 -4.06694382e-01
4.90742296e-01 -3.14729005e-01 -6.34494364e-01 -1.45626470e-01
-2.44618908e-01 3.65428269e-01 9.62929428e-01 -1.21235645e+00
-6.58372939e-01 -1.13628000e-01 -1.85972169e-01 4.87206727e-01
-5.64366639e-01 1.91304445e-01 -6.10909820e-01 -9.17645633e-01
6.21500127e-02 -8.66050124e-01 5.76123036e-02 -4.51577783e-01
-4.87579048e-01 -1.29429400e-01 6.90902412e-01 -1.13611770e+00
1.69988108e+00 -2.31208205e+00 2.41480276e-01 -4.43084240e-02
-2.93002844e-01 8.67451310e-01 -3.77234191e-01 1.15554638e-01
1.85265034e-01 1.35833040e-01 -4.02537882e-01 -7.47711182e-01
-1.52239859e-01 1.85063913e-01 -4.35272008e-01 8.81693140e-02
2.99408555e-01 4.55793709e-01 -7.31006384e-01 -3.53910118e-01
1.58869907e-01 5.11981785e-01 -6.50434613e-01 7.82417119e-01
-9.68179703e-02 5.82370341e-01 1.70810357e-01 3.97178560e-01
5.44711351e-01 2.10936219e-01 2.16489583e-01 2.65403204e-02
-1.21748582e-01 8.56952608e-01 -1.19742310e+00 1.64588225e+00
-1.13251650e+00 5.87250769e-01 2.86241829e-01 -8.31088543e-01
1.22853100e+00 6.25536799e-01 3.99366505e-02 -6.73650324e-01
-9.70370993e-02 4.84827489e-01 1.77386016e-01 -2.16594249e-01
6.72866762e-01 -3.89853120e-01 1.17674768e-01 3.88537943e-01
2.62473017e-01 -3.55336219e-01 6.03778102e-02 -2.36029010e-02
6.95571423e-01 1.49288177e-01 1.95614234e-01 -2.99794339e-02
9.67222512e-01 -2.07321376e-01 6.16627395e-01 5.93172014e-01
-2.23256543e-01 9.48273838e-01 -2.00508848e-01 3.29615444e-01
-9.66220796e-01 -1.05748212e+00 -4.13833596e-02 1.36976838e+00
-3.01637858e-01 -4.06399429e-01 -8.29879820e-01 -4.85843956e-01
-2.92969227e-01 1.00995803e+00 -2.40211755e-01 -2.67860919e-01
-8.33046436e-01 -3.99489105e-01 7.26265430e-01 5.43526411e-01
3.63533020e-01 -1.07188225e+00 2.52501577e-01 5.19857943e-01
-3.91366363e-01 -1.14246356e+00 -1.00994551e+00 1.40083611e-01
-6.15564466e-01 -4.48500842e-01 -7.85236299e-01 -9.47931170e-01
2.02010334e-01 3.34851772e-01 8.44577849e-01 1.40411571e-01
-2.87948758e-04 6.57402426e-02 -6.47649646e-01 -2.37193331e-01
-1.11414421e+00 2.85273969e-01 3.47328097e-01 1.27944499e-01
5.86837418e-02 -5.09017348e-01 -2.19011515e-01 2.16728866e-01
-7.41988242e-01 -1.73329502e-01 2.58669704e-01 9.46389973e-01
8.80927742e-01 -1.09404340e-01 1.07781613e+00 -8.51446748e-01
8.97285044e-01 -2.00565502e-01 -4.39320356e-01 3.64214242e-01
-5.66749394e-01 -1.73404552e-02 1.06854534e+00 -7.06764519e-01
-1.20599496e+00 5.73808774e-02 -5.64426124e-01 -5.08855045e-01
3.80478799e-02 3.36146265e-01 -4.96879578e-01 2.72927761e-01
5.36047816e-01 4.57796305e-01 -9.72836465e-02 -8.97136927e-01
5.60196102e-01 1.51277125e+00 6.94217086e-01 -5.29522359e-01
8.79861772e-01 -1.26786381e-01 -6.16467357e-01 -1.12685966e+00
-7.23101676e-01 -5.83647370e-01 -6.74697042e-01 7.45997904e-03
5.79708636e-01 -9.34263051e-01 -1.58394247e-01 3.11861575e-01
-1.00590372e+00 -4.42326277e-01 -4.70755070e-01 5.95097542e-01
-5.62403977e-01 1.88650504e-01 -4.73014921e-01 -1.08774102e+00
-4.98674542e-01 -1.10866416e+00 8.23060930e-01 2.01866075e-01
-5.11864685e-02 -8.21046114e-01 3.18368435e-01 5.48217118e-01
5.70491195e-01 -5.54410875e-01 6.46219850e-01 -9.56566334e-01
-1.76520973e-01 -7.44596422e-02 3.03645968e-01 8.84071350e-01
6.18680656e-01 -1.67361759e-02 -1.28623343e+00 -2.78262079e-01
-4.32402119e-02 -3.32755744e-01 6.48553550e-01 1.29068390e-01
1.10199833e+00 -4.03992265e-01 1.71959862e-01 6.01741612e-01
8.87887359e-01 3.60782862e-01 5.21476626e-01 4.14846204e-02
6.24851227e-01 4.71876830e-01 7.62789130e-01 9.91568938e-02
1.00428134e-01 8.83260548e-01 -2.47268245e-01 9.56040248e-03
-7.67351985e-01 -5.48784673e-01 6.26291037e-01 1.65063608e+00
-2.62333695e-02 -2.56106913e-01 -5.63153803e-01 7.48533189e-01
-1.45651507e+00 -8.54340911e-01 2.13978082e-01 2.58975673e+00
1.42913318e+00 -1.30136147e-01 3.79424840e-01 2.41092443e-01
9.66303885e-01 1.60510719e-01 -3.93710941e-01 -4.47577268e-01
-1.22427009e-01 4.77059156e-01 -1.32342368e-01 8.35961878e-01
-7.64749944e-01 1.38879585e+00 5.98588419e+00 9.84826982e-01
-1.28816116e+00 4.05343026e-01 3.50764006e-01 -2.86427438e-01
-3.45700473e-01 -1.22479223e-01 -8.24723125e-01 2.61837244e-01
1.03970933e+00 -1.86068743e-01 1.14592195e+00 7.55448878e-01
3.61774594e-01 4.98653740e-01 -9.48970079e-01 9.08242881e-01
1.30932704e-01 -1.10369098e+00 1.72686875e-01 -2.39637956e-01
6.27874255e-01 -5.26175611e-02 1.84106827e-03 5.80686748e-01
2.68537551e-01 -9.66227174e-01 8.02233338e-01 1.51959690e-03
1.13232780e+00 -7.24297404e-01 5.02764583e-01 2.76576132e-01
-1.33066642e+00 1.43197462e-01 -2.97552019e-01 3.42079401e-01
1.92615286e-01 -3.10096256e-02 -1.11445713e+00 4.38516021e-01
3.49733323e-01 4.17683065e-01 -4.68584031e-01 8.19270492e-01
-2.99910843e-01 1.23707175e+00 -1.16386779e-01 5.83442859e-03
-9.11118314e-02 -6.81982860e-02 7.38001406e-01 1.36838746e+00
2.55046815e-01 3.38072097e-03 6.98580742e-02 7.28394270e-01
-4.86429185e-01 5.16862452e-01 -1.28171042e-01 -4.10898238e-01
1.00947249e+00 1.08290195e+00 -1.91074878e-01 -2.47391924e-01
-2.50943810e-01 1.06666696e+00 4.57319379e-01 3.00893694e-01
-5.41862965e-01 -4.64635283e-01 7.78348565e-01 4.40348834e-02
2.64635682e-01 -2.31976569e-01 -1.34666041e-01 -1.17395699e+00
-5.10227168e-03 -1.11535597e+00 2.19780758e-01 -5.07935107e-01
-1.17926717e+00 1.07412124e+00 -4.77345079e-01 -1.37471414e+00
-4.21898901e-01 -2.00588599e-01 -6.62737250e-01 1.07347882e+00
-1.79590142e+00 -1.13023293e+00 -1.43841296e-01 6.88236475e-01
1.05910361e+00 -6.29764259e-01 8.19349408e-01 3.88558984e-01
-6.36395097e-01 9.72083449e-01 1.22851744e-01 1.12298345e-02
8.49335194e-01 -1.27842438e+00 5.12489319e-01 1.11361194e+00
4.69405890e-01 7.17660964e-01 6.47130132e-01 -6.08819366e-01
-1.15905094e+00 -1.29331338e+00 7.95646906e-01 -4.92534824e-02
4.59193766e-01 -5.81929743e-01 -1.19748604e+00 4.86625999e-01
1.32943407e-01 4.15403163e-03 8.81577015e-01 1.71961606e-01
-2.93945879e-01 -3.27328652e-01 -1.00951982e+00 4.73198354e-01
1.08764195e+00 -8.37883413e-01 -6.66720927e-01 8.50776583e-02
1.14911437e+00 -3.77277792e-01 -8.80649626e-01 8.15976411e-02
2.85096914e-01 -3.88695925e-01 7.39404559e-01 -7.36736238e-01
8.41109008e-02 -4.30239826e-01 -5.54333806e-01 -1.66314530e+00
-1.17205165e-01 -1.03316963e+00 -1.44346699e-01 1.82410836e+00
6.53562427e-01 -1.89143687e-01 3.30307394e-01 1.82952166e-01
-5.04214585e-01 -2.15311423e-01 -1.00312459e+00 -1.00786030e+00
-6.62368387e-02 -4.88284856e-01 7.12763131e-01 9.54822063e-01
-1.38504103e-01 5.82267165e-01 -5.19514501e-01 6.12972230e-02
2.66746730e-01 -7.11882636e-02 8.13320994e-01 -8.99803162e-01
-5.25173604e-01 -1.47820547e-01 1.23251431e-01 -1.16602206e+00
1.78889006e-01 -1.13125527e+00 4.22069103e-01 -1.22377133e+00
-1.55229762e-01 -5.86657703e-01 -3.46138597e-01 2.69979000e-01
-5.09626448e-01 1.92573398e-01 4.68633235e-01 1.96684688e-01
-2.52388388e-01 7.86291718e-01 1.21245646e+00 6.82597533e-02
-8.32944334e-01 3.86795700e-01 -6.66899085e-01 2.06054896e-01
9.13458884e-01 -4.55102891e-01 -5.43497324e-01 -3.69254082e-01
-4.82366771e-01 1.01971170e-02 -1.33665189e-01 -9.30072129e-01
1.40281022e-01 -1.68164968e-01 -2.40189340e-02 -4.52039421e-01
5.04269958e-01 -5.47656596e-01 -1.84860170e-01 1.40515596e-01
-6.23688459e-01 -3.56068641e-01 2.15254456e-01 3.96468133e-01
-4.25191194e-01 -3.53836596e-01 1.10370934e+00 2.23594084e-02
-3.90166789e-01 2.21050486e-01 -1.85349569e-01 3.19384366e-01
5.19960105e-01 6.21446036e-02 -8.96711126e-02 -5.72831333e-01
-7.63266802e-01 1.18523249e-02 1.88555047e-01 7.90660560e-01
6.74678445e-01 -1.44747508e+00 -9.47974741e-01 4.43761408e-01
8.54691043e-02 -1.87795848e-01 1.78036466e-01 4.56572443e-01
-1.16042174e-01 2.68666863e-01 4.21795473e-02 -3.04341972e-01
-1.53843141e+00 4.19745147e-01 3.85188937e-01 -6.94292113e-02
-2.95142472e-01 9.12963450e-01 -1.53261110e-01 -8.80412221e-01
5.10538280e-01 -1.47732332e-01 -2.91125387e-01 -7.45114163e-02
7.14114547e-01 4.17773724e-01 2.72905767e-01 -7.90041983e-01
-2.71949768e-01 2.41242558e-01 -1.53996438e-01 -6.20299399e-01
1.43627739e+00 -4.53063399e-01 1.94045171e-01 6.98931694e-01
1.08456552e+00 4.34939444e-01 -1.06147385e+00 -4.90046710e-01
-1.26116455e-01 -4.90020275e-01 1.72334164e-01 -8.46096456e-01
-8.79990458e-01 1.07417965e+00 4.50284034e-01 6.88916072e-02
1.24865258e+00 -4.51335721e-02 8.50260198e-01 2.60217786e-01
-1.62686631e-02 -1.42195547e+00 5.33727929e-02 5.29817522e-01
1.22805107e+00 -1.07470965e+00 -4.46108967e-01 -4.60701883e-01
-1.03989375e+00 1.04466248e+00 6.67734742e-01 9.97679383e-02
4.33714688e-01 1.35501295e-01 2.50892907e-01 5.34822047e-01
-6.61497712e-01 -3.96014184e-01 5.17484367e-01 7.18586266e-01
7.36232340e-01 1.67824045e-01 -2.22066224e-01 8.50589693e-01
-6.27672374e-01 -5.17559238e-02 5.26809871e-01 5.76537371e-01
-4.87049580e-01 -1.35870063e+00 -2.48670757e-01 2.05474079e-01
-4.97104734e-01 -2.96577513e-01 -5.31886399e-01 3.92437726e-01
-2.87331849e-01 1.37350202e+00 -3.73750925e-02 -6.77429497e-01
6.03688657e-01 2.02647030e-01 2.84487963e-01 -9.77926672e-01
-7.71260440e-01 3.64485443e-01 3.81202370e-01 -2.02485740e-01
-2.31490791e-01 -6.50692463e-01 -1.22430301e+00 1.04806595e-01
-5.85784972e-01 2.59131819e-01 8.00666988e-01 8.20306659e-01
2.03928515e-01 7.61080146e-01 1.09290636e+00 -3.40179443e-01
-9.35429692e-01 -1.29463243e+00 -4.95466024e-01 4.37826931e-01
5.98254442e-01 -3.15456510e-01 -3.65447670e-01 1.61838576e-01] | [14.578495025634766, 6.606527805328369] |
81b8ee9d-1b89-4b40-b09b-1876ce4451bb | ego-exo-transferring-visual-representations | 2104.07905 | null | https://arxiv.org/abs/2104.07905v1 | https://arxiv.org/pdf/2104.07905v1.pdf | Ego-Exo: Transferring Visual Representations from Third-person to First-person Videos | We introduce an approach for pre-training egocentric video models using large-scale third-person video datasets. Learning from purely egocentric data is limited by low dataset scale and diversity, while using purely exocentric (third-person) data introduces a large domain mismatch. Our idea is to discover latent signals in third-person video that are predictive of key egocentric-specific properties. Incorporating these signals as knowledge distillation losses during pre-training results in models that benefit from both the scale and diversity of third-person video data, as well as representations that capture salient egocentric properties. Our experiments show that our Ego-Exo framework can be seamlessly integrated into standard video models; it outperforms all baselines when fine-tuned for egocentric activity recognition, achieving state-of-the-art results on Charades-Ego and EPIC-Kitchens-100. | ['Kristen Grauman', 'Bo Xiong', 'Tushar Nagarajan', 'Yanghao Li'] | 2021-04-16 | null | http://openaccess.thecvf.com//content/CVPR2021/html/Li_Ego-Exo_Transferring_Visual_Representations_From_Third-Person_to_First-Person_Videos_CVPR_2021_paper.html | http://openaccess.thecvf.com//content/CVPR2021/papers/Li_Ego-Exo_Transferring_Visual_Representations_From_Third-Person_to_First-Person_Videos_CVPR_2021_paper.pdf | cvpr-2021-1 | ['egocentric-activity-recognition'] | ['computer-vision'] | [-2.39057407e-01 -1.64005496e-02 -5.91186285e-01 -3.47529024e-01
-4.12158281e-01 -6.61606491e-01 1.09161162e+00 -5.28798521e-01
-3.14022034e-01 4.87421930e-01 1.24617290e+00 3.27207536e-01
-1.11146495e-01 -5.21654189e-01 -1.00024962e+00 -4.11879569e-01
-3.56809288e-01 3.45036685e-01 -3.38663347e-02 -7.66521646e-03
-1.36630088e-02 2.91741908e-01 -1.53632915e+00 5.64504087e-01
2.33217537e-01 6.76887095e-01 -5.47655486e-02 8.21924686e-01
6.01823747e-01 1.35375142e+00 2.29328796e-02 -1.42840192e-01
4.63095993e-01 -2.54074037e-01 -7.79837966e-01 1.45734325e-01
8.85029435e-01 -7.59901345e-01 -1.18807709e+00 6.13121092e-01
4.68063861e-01 3.60001028e-01 8.76846731e-01 -1.10881376e+00
-5.86037040e-01 5.40443063e-01 -5.96871495e-01 5.47995567e-01
7.06468821e-01 3.61807287e-01 1.15550661e+00 -7.43153870e-01
1.22151303e+00 1.38006890e+00 5.83481789e-01 3.82361352e-01
-1.27498543e+00 -6.19899631e-01 4.05920237e-01 6.30944669e-01
-1.15619612e+00 -9.41505432e-01 8.72436345e-01 -7.34473169e-01
1.31559420e+00 -2.08835602e-01 8.30685318e-01 1.98457325e+00
2.14571003e-02 1.19014287e+00 5.32427728e-01 9.06997249e-02
6.58821240e-02 -1.47385597e-01 -2.26422772e-01 6.48393154e-01
1.80155054e-01 2.13124171e-01 -9.25640047e-01 1.12689041e-01
9.81197298e-01 -3.62549983e-02 -2.45955154e-01 -1.24434900e+00
-1.64013469e+00 7.18524098e-01 3.67483050e-01 -2.89274193e-02
-7.07513094e-01 5.31592250e-01 7.77167618e-01 3.48655790e-01
4.59564984e-01 6.16278172e-01 -3.93676460e-01 -6.99819148e-01
-9.26251829e-01 6.07478857e-01 5.91603160e-01 1.29132879e+00
7.54350960e-01 3.28887224e-01 -2.18781739e-01 4.82767224e-01
-1.75358489e-01 3.50491166e-01 6.34475768e-01 -1.37655389e+00
4.37609792e-01 2.57123351e-01 -3.97885852e-02 -8.48854601e-01
-3.47851783e-01 -4.59238321e-01 -1.98931426e-01 -8.26142132e-02
2.81039506e-01 -2.02007547e-01 -7.39131689e-01 2.04195738e+00
1.11182243e-01 5.65981567e-01 6.74447566e-02 8.26596737e-01
8.62240672e-01 2.32305363e-01 1.87101260e-01 2.16556892e-01
1.29515600e+00 -1.03575325e+00 -3.39312881e-01 -4.95854825e-01
6.66056335e-01 -1.57898515e-01 8.76405656e-01 7.80655295e-02
-1.16942191e+00 -5.99263608e-01 -8.05968225e-01 -1.60306081e-01
-1.52975231e-01 5.22916988e-02 9.17101204e-01 3.07018429e-01
-8.76227796e-01 5.39664865e-01 -8.15248251e-01 -7.56923079e-01
7.70513296e-01 2.68355589e-02 -9.26734149e-01 -5.35395630e-02
-8.82294238e-01 7.58900285e-01 4.81571108e-01 -6.74624979e-01
-1.51896560e+00 -1.07046235e+00 -1.16748941e+00 1.74924582e-01
5.20976126e-01 -1.18801546e+00 1.07543278e+00 -1.16192031e+00
-1.15924442e+00 1.06866729e+00 -1.88375428e-01 -6.17326856e-01
7.52259552e-01 -6.34329975e-01 -4.11773086e-01 5.76848507e-01
3.39032441e-01 9.35143352e-01 9.98226583e-01 -9.11338449e-01
-4.50794727e-01 -2.61972070e-01 5.09331703e-01 5.53605914e-01
-3.47258329e-01 -1.64050549e-01 -4.61339742e-01 -7.41610229e-01
-1.59934744e-01 -8.20492029e-01 2.07864866e-01 -2.71526068e-01
-1.25922441e-01 3.74170505e-02 7.70445585e-01 -5.75446427e-01
8.24314475e-01 -2.11420488e+00 4.53215271e-01 -1.82218894e-01
3.22483212e-01 -9.22427922e-02 -4.72339541e-01 3.91553551e-01
-4.50961679e-01 -3.13882321e-01 6.53215349e-01 -2.61268139e-01
-1.84989423e-02 -8.20307992e-03 -3.09601635e-01 6.43093884e-01
1.86455160e-01 1.21841037e+00 -1.27454090e+00 -3.22521597e-01
3.69098574e-01 3.85646701e-01 -1.19673347e+00 1.91112384e-01
-1.48544684e-01 4.95970488e-01 -3.85605961e-01 7.49849617e-01
2.17027068e-01 -2.28330106e-01 1.55089200e-01 -3.83238763e-01
2.83084333e-01 4.04679030e-01 -7.48446465e-01 2.14924717e+00
-2.44591519e-01 9.34166133e-01 -2.78832406e-01 -9.02817965e-01
3.80991369e-01 4.25951481e-01 8.30252528e-01 -6.22917831e-01
8.58266279e-02 -2.28453532e-01 -2.26826653e-01 -7.06409812e-01
6.12782121e-01 4.27701250e-02 -2.47144373e-03 4.06626463e-01
7.99974561e-01 2.21638992e-01 2.79120326e-01 4.96503532e-01
1.28751421e+00 9.41013575e-01 4.73738492e-01 -2.62181610e-01
2.64173031e-01 -4.27015163e-02 5.63883066e-01 7.83028841e-01
-3.72026831e-01 6.09027505e-01 7.99547732e-01 -4.89447862e-01
-1.22255421e+00 -1.33932769e+00 2.87363529e-01 1.26467371e+00
-1.46426037e-01 -7.87261963e-01 -5.72006285e-01 -8.28286052e-01
6.59600645e-02 7.50699103e-01 -8.58771920e-01 -4.59314346e-01
-6.32505178e-01 -1.04177348e-01 5.19189894e-01 8.21033478e-01
4.09733087e-01 -8.52493465e-01 -4.43889439e-01 1.55520970e-02
-3.92654151e-01 -1.56011748e+00 -3.46749187e-01 -7.39708543e-02
-8.56475651e-01 -1.23657024e+00 -6.61804676e-01 -3.15046698e-01
3.64243627e-01 7.79132485e-01 1.51644421e+00 -6.95637405e-01
-1.21993661e-01 9.12205338e-01 -3.46836925e-01 -1.59800336e-01
2.26715486e-02 5.80393001e-02 5.37500739e-01 -8.10880437e-02
9.23320234e-01 -1.01495183e+00 -7.21624017e-01 1.87714815e-01
-2.77330160e-01 1.77172437e-01 3.67997497e-01 7.81022906e-01
1.21097215e-01 -5.26305377e-01 5.40222466e-01 -8.33887815e-01
7.02818558e-02 -8.90654981e-01 -1.03056908e-01 -2.59403646e-01
-8.39980096e-02 -1.58832163e-01 6.26697302e-01 -4.55010086e-01
-1.12422740e+00 8.81411955e-02 3.77252311e-01 -1.12716067e+00
-3.18092316e-01 2.10281163e-01 -1.18556291e-01 1.74830049e-01
9.70024765e-01 2.42839307e-01 -1.44470051e-01 -3.92492712e-01
8.14332247e-01 9.91215706e-02 9.04392242e-01 -7.59474039e-01
7.61124074e-01 9.21887398e-01 -9.91224051e-02 -9.83056307e-01
-8.44364107e-01 -7.56137073e-01 -9.26874459e-01 -1.57282993e-01
8.81410241e-01 -1.87931836e+00 -6.13635004e-01 1.74984708e-01
-7.44020700e-01 -2.34674037e-01 -6.61953986e-01 7.70457983e-01
-1.25881469e+00 4.96479720e-01 -4.14438635e-01 -2.83962816e-01
1.35902956e-01 -7.50663698e-01 1.02265739e+00 -1.23324826e-01
-6.42346442e-01 -1.02761805e+00 3.04101408e-01 5.26146472e-01
-6.89175399e-03 2.47618169e-01 5.20519853e-01 -5.33056319e-01
-8.17409754e-01 -2.36560956e-01 -2.48456359e-01 1.46229059e-01
4.34767790e-02 -4.04057235e-01 -1.26408553e+00 -4.50314879e-01
-2.29871064e-01 -7.32062042e-01 1.20408380e+00 5.64743280e-01
1.04742670e+00 -3.09643328e-01 -2.93520510e-01 1.22524655e+00
9.99742031e-01 -4.30922538e-01 7.41111338e-01 4.90603179e-01
1.00427508e+00 5.69267511e-01 5.41861713e-01 5.84257483e-01
5.95897377e-01 6.06994331e-01 3.76191705e-01 2.33842194e-01
-1.72121987e-01 -6.55055225e-01 6.31503046e-01 2.68580705e-01
-5.84534943e-01 -1.13091283e-02 -4.55103844e-01 9.21385884e-01
-2.03228259e+00 -1.86391079e+00 4.18916881e-01 1.84333956e+00
4.88436341e-01 -2.56153420e-02 5.36440849e-01 -4.88886416e-01
1.56370178e-01 7.82364666e-01 -6.05821848e-01 -1.02238776e-03
-2.74874866e-01 -2.71466523e-01 5.46489716e-01 6.59327507e-02
-1.49585438e+00 1.23313618e+00 7.12697744e+00 3.90335768e-01
-8.31350982e-01 6.97882846e-02 1.27383694e-01 -7.87819862e-01
-1.56841189e-01 2.22099386e-02 -5.66279531e-01 2.64518052e-01
8.01731288e-01 -2.74629623e-01 5.49629569e-01 1.24016619e+00
3.92548501e-01 1.45231441e-01 -1.72081709e+00 1.24250388e+00
2.98459947e-01 -1.54801571e+00 1.75241441e-01 2.40276292e-01
1.02461445e+00 4.38126296e-01 -8.33914280e-02 5.49919963e-01
4.76615191e-01 -8.74159157e-01 8.16908598e-01 5.52152634e-01
7.13625073e-01 -4.93182212e-01 3.16181958e-01 1.94649681e-01
-1.11008668e+00 -2.42818013e-01 -5.01789749e-01 -2.80443728e-01
9.73547325e-02 3.71496044e-02 -7.07735777e-01 4.78189349e-01
8.29558313e-01 1.58028972e+00 -5.83719015e-01 6.37570143e-01
-2.03743786e-01 5.13411283e-01 -1.38289690e-01 4.90694851e-01
3.25546294e-01 -1.20351210e-01 9.16663468e-01 1.44124246e+00
1.10397346e-01 -6.81588203e-02 -1.30379811e-01 9.65299666e-01
-3.40260834e-01 -2.62211472e-01 -1.20876813e+00 -2.91304171e-01
2.03124791e-01 1.02868974e+00 -7.55428597e-02 -6.06595457e-01
-7.16972351e-01 1.05529606e+00 5.49220383e-01 7.25475073e-01
-8.24030817e-01 9.82614309e-02 1.41596663e+00 2.18080059e-01
5.69501400e-01 -2.38313466e-01 1.69693366e-01 -1.91226804e+00
-1.70081154e-01 -9.70774233e-01 4.87002134e-01 -9.52100396e-01
-1.00948584e+00 8.22263956e-02 2.80578822e-01 -1.41968501e+00
-9.57328260e-01 -6.74885869e-01 -6.04963243e-01 4.71739501e-01
-1.32356679e+00 -1.63479424e+00 -5.68502784e-01 8.84829700e-01
7.31996000e-01 -6.89958036e-01 6.04954004e-01 1.86048120e-01
-2.08875194e-01 5.08630216e-01 1.35937482e-01 3.18581045e-01
9.03954625e-01 -1.16677988e+00 6.18331075e-01 8.27415705e-01
4.95010078e-01 8.37547302e-01 6.61557794e-01 -5.49604714e-01
-1.55216873e+00 -1.03580046e+00 6.07179940e-01 -1.04424179e+00
7.38832712e-01 -5.66530883e-01 -4.24753696e-01 1.50269675e+00
3.99116985e-02 -6.02118485e-02 6.20841563e-01 5.90619504e-01
-9.38958764e-01 1.08945616e-01 -7.10928679e-01 8.72807801e-01
1.65267622e+00 -1.04553473e+00 -7.94716239e-01 3.44170421e-01
4.49406177e-01 -1.90672129e-01 -7.86901176e-01 2.36339435e-01
8.69338751e-01 -1.11452031e+00 1.52177835e+00 -1.19709170e+00
7.95639396e-01 -3.34271006e-02 -2.22992659e-01 -1.39829123e+00
-8.52481008e-01 -7.04429626e-01 -6.09341085e-01 9.31908250e-01
-2.67993391e-01 -1.23804487e-01 1.00289214e+00 1.40820235e-01
-3.64164710e-02 -1.24270372e-01 -6.05433404e-01 -8.01700830e-01
-1.16678037e-01 -3.95099014e-01 2.88933992e-01 1.07570696e+00
2.69261926e-01 4.71552074e-01 -8.41366589e-01 -2.91150361e-01
7.21176803e-01 -1.87800542e-01 1.40682542e+00 -1.00035429e+00
-4.49693978e-01 -2.67969549e-01 -9.34124649e-01 -1.38048875e+00
4.30317909e-01 -7.72185028e-01 -6.29723608e-01 -1.22123420e+00
5.85826933e-01 3.86934608e-01 -3.38526398e-01 1.83943942e-01
7.28413239e-02 2.58037657e-01 1.30120382e-01 1.94995120e-01
-8.19319725e-01 7.26419389e-01 1.08418465e+00 9.83030256e-03
-3.46181244e-02 -4.03145164e-01 -7.95299292e-01 1.00020754e+00
2.96812713e-01 -1.42061457e-01 -7.57494926e-01 -5.13616800e-01
1.35752812e-01 -1.01426773e-01 6.30820572e-01 -1.18314350e+00
9.74357575e-02 -1.56799629e-01 7.81205416e-01 -5.19147515e-01
7.96173275e-01 -5.90600371e-01 -7.71669950e-03 9.39577445e-03
-2.36512721e-01 -1.97622165e-01 -4.39437963e-02 1.00894380e+00
-1.37233511e-01 3.22188497e-01 6.08553767e-01 -4.63590831e-01
-1.27107525e+00 4.01898026e-01 -2.74415702e-01 4.03859705e-01
9.89125669e-01 -5.77407658e-01 -4.98458087e-01 -9.02569771e-01
-6.05127752e-01 2.23002642e-01 8.23125780e-01 8.58812988e-01
3.03624004e-01 -1.38093460e+00 -7.50687718e-01 2.88955390e-01
4.56605017e-01 -4.86500800e-01 5.73314250e-01 8.21978569e-01
-4.49574828e-01 6.84196055e-01 -7.62108684e-01 -7.85126269e-01
-1.13678181e+00 7.73023248e-01 2.00523317e-01 7.13590533e-03
-9.58761930e-01 7.19111562e-01 9.19895411e-01 -3.23090583e-01
1.08768985e-01 1.14740562e-02 -2.31440008e-01 1.46506071e-01
4.60476041e-01 3.89682770e-01 -4.60069835e-01 -9.13011909e-01
-3.52654785e-01 4.62272733e-01 -7.97550604e-02 -1.11254819e-01
1.46155739e+00 -1.82186231e-01 4.20795768e-01 2.27571085e-01
1.31238508e+00 -4.18883115e-02 -1.76691997e+00 -4.57019210e-01
-2.73299068e-01 -8.38894248e-01 1.47200571e-02 -4.17865962e-01
-7.84813881e-01 8.44432771e-01 1.46813989e-01 -7.10365236e-01
7.49457955e-01 1.37013689e-01 4.05711859e-01 7.57079244e-01
5.25780499e-01 -1.16324961e+00 5.40661275e-01 7.15385079e-01
7.67162383e-01 -1.34353268e+00 4.27378833e-01 -8.71887729e-02
-1.02487385e+00 9.10441279e-01 6.41434669e-01 -5.78722656e-01
5.39819479e-01 -4.06976521e-01 -2.80389965e-01 -3.70702714e-01
-1.02608955e+00 -2.47745514e-01 4.07447845e-01 9.21189189e-01
2.91928113e-01 -2.67792065e-02 2.57814139e-01 5.47136188e-01
-2.82739550e-01 1.30320475e-01 5.49472749e-01 7.14363635e-01
-1.74552709e-01 -4.27704245e-01 1.31676838e-01 3.00455600e-01
-3.80588353e-01 9.24869925e-02 -3.21603984e-01 9.78570700e-01
-1.74448222e-01 5.44132411e-01 2.81526417e-01 -4.55605686e-01
2.02763468e-01 1.09625436e-01 7.12769985e-01 -5.02755284e-01
-1.81743518e-01 1.93407774e-01 5.14532447e-01 -1.13061368e+00
-4.60325360e-01 -8.91221762e-01 -6.49951279e-01 -6.47299349e-01
2.57915676e-01 -2.37401143e-01 1.73872247e-01 7.24432468e-01
5.69020689e-01 5.24979711e-01 4.24455345e-01 -1.31197786e+00
-4.66732621e-01 -1.00705612e+00 -5.94131231e-01 8.38376701e-01
2.95917630e-01 -9.64113712e-01 -3.50351602e-01 3.36310029e-01] | [8.435591697692871, 0.5856999158859253] |
05e3e6a2-dfe5-4a3d-97a2-781dc9116361 | using-the-web-as-an-implicit-training-set | 1912.01113 | null | https://arxiv.org/abs/1912.01113v1 | https://arxiv.org/pdf/1912.01113v1.pdf | Using the Web as an Implicit Training Set: Application to Noun Compound Syntax and Semantics | An important characteristic of English written text is the abundance of noun compounds - sequences of nouns acting as a single noun, e.g., colon cancer tumor suppressor protein. While eventually mastered by domain experts, their interpretation poses a major challenge for automated analysis. Understanding noun compounds' syntax and semantics is important for many natural language applications, including question answering, machine translation, information retrieval, and information extraction. I address the problem of noun compounds syntax by means of novel, highly accurate unsupervised and lightly supervised algorithms using the Web as a corpus and search engines as interfaces to that corpus. Traditionally the Web has been viewed as a source of page hit counts, used as an estimate for n-gram word frequencies. I extend this approach by introducing novel surface features and paraphrases, which yield state-of-the-art results for the task of noun compound bracketing. I also show how these kinds of features can be applied to other structural ambiguity problems, like prepositional phrase attachment and noun phrase coordination. I address noun compound semantics by automatically generating paraphrasing verbs and prepositions that make explicit the hidden semantic relations between the nouns in a noun compound. I also demonstrate how these paraphrasing verbs can be used to solve various relational similarity problems, and how paraphrasing noun compounds can improve machine translation. | ['Preslav Nakov'] | 2019-11-23 | null | null | null | null | ['prepositional-phrase-attachment'] | ['natural-language-processing'] | [ 3.65775913e-01 6.88637272e-02 -6.06345892e-01 -2.92647183e-01
-6.77500308e-01 -1.00060189e+00 6.24499977e-01 8.70983422e-01
-7.00333834e-01 7.53052711e-01 4.72572356e-01 -5.74772179e-01
-1.46671131e-01 -8.81089807e-01 -6.63588762e-01 -3.81972402e-01
4.88952659e-02 7.76574433e-01 4.06886697e-01 -7.58444607e-01
3.25999677e-01 3.31885844e-01 -1.41795087e+00 6.10032916e-01
7.80301809e-01 4.90122706e-01 3.59339416e-01 1.04968295e-01
-7.86120772e-01 1.16914108e-01 -5.86396754e-01 -7.87572980e-01
-2.42464896e-02 -1.45530820e-01 -1.02246261e+00 -6.48511574e-02
-1.11989185e-01 4.67702001e-01 3.25184464e-01 1.22291362e+00
1.62393153e-02 -3.58410403e-02 6.73598528e-01 -1.04227269e+00
-3.02661777e-01 7.57689059e-01 -1.97766468e-01 3.34113061e-01
8.30143332e-01 -4.41338062e-01 1.91699278e+00 -1.25384343e+00
8.44765902e-01 1.30942810e+00 3.47953707e-01 4.34148580e-01
-1.26831913e+00 -3.05562407e-01 1.60390604e-03 1.46685228e-01
-1.33189845e+00 -8.02167803e-02 4.19296384e-01 -3.44890863e-01
1.41424668e+00 3.65267366e-01 4.57388103e-01 9.85389531e-01
2.29270816e-01 7.61715829e-01 6.95121765e-01 -8.56513977e-01
6.77073300e-02 1.12594180e-01 4.89060760e-01 5.36035001e-01
4.34823543e-01 -1.84092984e-01 -5.44840693e-01 -3.89800072e-01
3.70017678e-01 -1.44669861e-01 -1.15065686e-01 -1.43315434e-01
-1.45036435e+00 1.15517080e+00 1.70499042e-01 6.53334618e-01
-1.84676632e-01 -1.38399154e-01 4.88188446e-01 2.89785773e-01
2.60182768e-01 7.06333458e-01 -8.48528206e-01 -2.17581168e-02
-2.69292414e-01 6.70848489e-01 1.20660555e+00 1.11481321e+00
7.85478890e-01 -7.06123471e-01 3.65563482e-01 9.28984463e-01
3.09734792e-01 3.73957396e-01 6.38819575e-01 -6.10397279e-01
6.19657457e-01 8.04252625e-01 4.54324223e-02 -7.72767007e-01
-4.01381761e-01 -2.68006530e-02 -8.71369615e-02 -3.97906810e-01
7.42362976e-01 3.64827961e-01 -4.31464195e-01 1.86205232e+00
3.41836452e-01 -6.77738965e-01 2.99510479e-01 5.98226011e-01
5.42614877e-01 7.76644111e-01 3.89286548e-01 -3.92226160e-01
1.97796011e+00 -5.30603707e-01 -6.46713436e-01 -4.79939103e-01
9.60875750e-01 -1.03804076e+00 1.26672506e+00 6.20175712e-02
-1.04135191e+00 5.02745174e-02 -6.19803190e-01 -2.71171302e-01
-7.54920185e-01 -5.71033955e-01 7.67336071e-01 4.15312231e-01
-4.59809393e-01 5.42010963e-01 -5.12291551e-01 -5.28404653e-01
-4.83454904e-03 5.40589750e-01 -4.55760926e-01 3.15576084e-02
-1.34021676e+00 9.51176763e-01 5.53517997e-01 -4.81158763e-01
1.56220943e-01 -6.49985552e-01 -1.05618596e+00 3.97504829e-02
5.55953324e-01 -5.88451922e-01 1.51343775e+00 -9.54566956e-01
-8.30551863e-01 1.21427286e+00 -4.88389611e-01 -2.71142215e-01
-3.14304948e-01 -1.32496014e-01 -3.34175080e-01 -1.02495298e-01
6.63313985e-01 2.91573316e-01 4.69714910e-01 -8.03672671e-01
-7.37221837e-01 -6.26690269e-01 8.42939876e-03 4.65692878e-01
-2.22571060e-01 7.87652969e-01 -8.85957479e-02 -7.90654540e-01
4.63274688e-01 -7.65472114e-01 -1.06899694e-01 -2.69122511e-01
-4.17206436e-01 -8.26157570e-01 3.12555999e-01 -3.02267700e-01
1.20254779e+00 -2.00436902e+00 4.81851757e-01 5.68529546e-01
1.70074984e-01 -7.80660808e-02 -1.42164588e-01 8.04934144e-01
-4.32358295e-01 3.87779653e-01 -3.51544321e-01 2.68993258e-01
1.98826268e-01 6.46557629e-01 -3.66154253e-01 4.18088026e-02
3.42804193e-01 1.18018830e+00 -9.04949844e-01 -3.05780172e-01
-2.72002488e-01 -7.41583183e-02 -3.39344561e-01 -2.06966460e-01
-6.69453204e-01 -2.44324192e-01 -5.01459837e-01 7.52128661e-01
-8.09155777e-03 -3.00828964e-01 5.25962889e-01 -6.80485368e-02
-3.34110260e-02 8.74125004e-01 -1.01766288e+00 1.44033670e+00
-2.55201697e-01 1.03692010e-01 -1.42110869e-01 -9.98740911e-01
5.82018316e-01 2.42537916e-01 3.93356144e-01 -6.16850555e-01
3.38925757e-02 7.59846568e-01 2.44638085e-01 -4.70042855e-01
3.98715913e-01 -3.75464052e-01 -5.50892293e-01 3.46346498e-01
-2.74892986e-01 -3.05126816e-01 7.22733736e-01 2.27427617e-01
1.12575865e+00 -7.33596608e-02 8.41752291e-01 -4.24861133e-01
5.20353138e-01 6.09060407e-01 3.53966624e-01 4.37276959e-01
5.10574400e-01 2.53625721e-01 6.36211097e-01 -4.19358581e-01
-1.27886260e+00 -1.22092366e+00 -4.15226936e-01 1.37532163e+00
-3.07297185e-02 -7.22514391e-01 -6.04831398e-01 -5.28897285e-01
1.19504429e-01 5.66575170e-01 -1.72796905e-01 6.71752840e-02
-9.04934227e-01 -9.73854303e-01 3.66062224e-01 6.05890512e-01
-2.32281476e-01 -9.90061879e-01 -2.70222157e-01 4.80581611e-01
-3.21266651e-01 -1.18780315e+00 -4.23320770e-01 5.95730186e-01
-8.37321222e-01 -1.07080650e+00 -3.97867709e-01 -1.10463595e+00
4.13513005e-01 3.69031727e-02 1.48076844e+00 3.41669261e-01
-2.15352654e-01 1.15424156e-01 -4.87015754e-01 -6.32610857e-01
-7.55778849e-01 2.15620950e-01 2.40147367e-01 -5.07700145e-01
1.06959033e+00 -4.97045159e-01 6.91508353e-02 2.63846397e-01
-1.23400235e+00 -2.86235005e-01 4.26671237e-01 1.02629423e+00
6.19290948e-01 -3.63230556e-01 2.89408237e-01 -1.29968309e+00
8.38627636e-01 -6.12210751e-01 -6.35728121e-01 1.64097846e-01
-3.96819621e-01 4.48089212e-01 3.91528457e-01 -3.95442098e-01
-5.45210898e-01 1.41562894e-01 -3.30131948e-01 4.54172373e-01
-8.33495930e-02 9.50296760e-01 -4.69985038e-01 3.25133920e-01
8.61470044e-01 1.72540665e-01 8.24751258e-02 -4.45798784e-01
3.00570488e-01 6.01707876e-01 4.20739859e-01 -8.13608468e-01
9.19681847e-01 6.34294003e-02 1.87795773e-01 -1.13043988e+00
-1.02744520e+00 -9.39394176e-01 -4.89032865e-01 5.90007007e-01
8.99961054e-01 -7.82805681e-01 -5.91814697e-01 -5.21122776e-02
-1.53504670e+00 1.41193390e-01 -3.77211034e-01 1.81288958e-01
-4.79038835e-01 6.25612855e-01 -5.96431255e-01 -1.86422944e-01
-3.09581608e-02 -8.97934794e-01 1.24834025e+00 -1.99236482e-01
-8.62928748e-01 -1.09503090e+00 2.64056064e-02 3.23228866e-01
-8.64023063e-03 -2.35769257e-01 1.92475367e+00 -1.35626316e+00
-5.31338096e-01 -6.94166049e-02 -1.20983668e-01 -5.49768396e-02
1.61514014e-01 -4.08044487e-01 -3.64397347e-01 2.96326548e-01
-1.33072242e-01 -6.37188777e-02 4.76526618e-01 -9.69387591e-02
5.81782281e-01 -5.72851181e-01 -4.50248957e-01 1.61225796e-01
1.30995178e+00 6.60323650e-02 5.23573160e-01 5.25712132e-01
4.07889605e-01 1.04671514e+00 4.96277720e-01 7.02175032e-03
2.68586248e-01 7.97747552e-01 2.78610349e-01 3.16017002e-01
2.62431949e-01 -1.56209350e-01 1.92894205e-01 9.10009325e-01
1.38589501e-01 -2.61268586e-01 -1.19212377e+00 5.77434182e-01
-1.79146171e+00 -8.47375989e-01 -3.82599205e-01 2.00618672e+00
1.34682548e+00 1.75678149e-01 1.34062707e-01 4.05407488e-01
5.60856938e-01 -2.90936679e-01 -3.23368912e-03 -3.58896196e-01
-3.08944076e-01 7.11164713e-01 3.93503755e-01 7.40809083e-01
-9.04155552e-01 1.31204784e+00 5.90077686e+00 8.52534831e-01
-5.27569115e-01 -1.70901697e-02 1.83606707e-02 5.82099080e-01
-6.54527545e-01 1.10457532e-01 -1.06982219e+00 2.75002182e-01
8.18059921e-01 -2.93219179e-01 4.70928878e-01 7.62889504e-01
-3.36689949e-02 -2.30308220e-01 -1.57253969e+00 8.84677708e-01
-7.43375868e-02 -1.46655905e+00 3.25982541e-01 3.65059301e-02
3.06081861e-01 -1.41298845e-01 -2.12401837e-01 -4.57399264e-02
7.48171955e-02 -1.04513454e+00 5.82192779e-01 -4.25330102e-02
5.50428867e-01 -5.55706024e-01 6.59066856e-01 3.11069101e-01
-1.09745812e+00 -6.53303564e-02 -3.31358314e-01 -3.09244096e-01
7.86760449e-02 3.17064553e-01 -8.29321325e-01 2.71706253e-01
3.32391143e-01 4.36363131e-01 -2.65052468e-01 7.24984825e-01
-4.28527802e-01 2.59204865e-01 -6.47359312e-01 -7.06270397e-01
2.21482128e-01 -2.52160251e-01 8.87825072e-01 1.12451077e+00
-6.41290322e-02 2.75583446e-01 2.48184294e-01 7.24358916e-01
-7.91466832e-02 6.81115866e-01 -8.55032802e-01 -3.03175807e-01
4.58527535e-01 8.40096712e-01 -8.57852042e-01 -2.31448174e-01
-8.11121166e-01 8.47332895e-01 3.32197905e-01 2.72844821e-01
-3.49978477e-01 -5.80634058e-01 8.90314698e-01 4.28562313e-01
-1.89826060e-02 -3.75976503e-01 -1.45950049e-01 -1.08157313e+00
3.79614621e-01 -1.07780588e+00 3.74358505e-01 -5.93205333e-01
-1.66244268e+00 4.10704583e-01 3.57579321e-01 -8.81550491e-01
-6.71511531e-01 -1.15529954e+00 -2.59967715e-01 8.23766887e-01
-1.31598425e+00 -7.49255478e-01 3.12483609e-01 4.99366909e-01
7.75180876e-01 -1.41224459e-01 1.17263758e+00 1.38749972e-01
-9.43948850e-02 2.17122480e-01 -6.34168508e-03 2.25272179e-01
4.76742983e-01 -1.36449742e+00 5.14773607e-01 5.21586120e-01
6.49395585e-01 9.68609869e-01 8.53162646e-01 -7.00238049e-01
-1.47867727e+00 -7.48840928e-01 1.81672323e+00 -7.09705293e-01
1.12893534e+00 -5.94913363e-01 -9.58655775e-01 6.08304858e-01
-1.53882459e-01 -2.14053228e-01 9.24100518e-01 2.89015979e-01
-5.60778379e-01 2.48655826e-01 -7.24247813e-01 8.84908319e-01
1.06075525e+00 -4.43664908e-01 -1.18147278e+00 1.04597127e+00
1.08701742e+00 -3.33219051e-01 -5.80769539e-01 5.62070087e-02
3.42017651e-01 -3.81367266e-01 1.00862241e+00 -1.07338500e+00
5.38080454e-01 -1.08447842e-01 -5.12427390e-01 -1.00820208e+00
-1.69460215e-02 -6.33096993e-01 2.38545761e-01 6.31722987e-01
1.08990180e+00 -4.45163041e-01 7.73061395e-01 7.27631330e-01
-1.53452337e-01 -6.75946414e-01 -9.75176275e-01 -9.84868288e-01
2.69372761e-01 -4.33967948e-01 3.91870618e-01 7.26998091e-01
6.08203530e-01 1.00292099e+00 4.59095240e-01 -1.84030652e-01
3.30633730e-01 9.64412317e-02 4.26034749e-01 -1.57774377e+00
-3.63491029e-01 -4.00213212e-01 -4.40606624e-01 -1.06816328e+00
4.35530007e-01 -1.43059361e+00 -1.65388882e-01 -1.19466007e+00
8.22661594e-02 -4.62917894e-01 4.41970199e-01 5.33214331e-01
1.87071245e-02 1.02270037e-01 -6.24099113e-02 2.65764087e-01
-1.50982007e-01 7.02774301e-02 7.94052005e-01 -2.44341176e-02
-2.20554322e-01 1.92111544e-02 -6.60480499e-01 9.69622791e-01
6.99077010e-01 -9.23232079e-01 -6.14957474e-02 -4.28137839e-01
9.71242130e-01 -1.97307855e-01 2.05897376e-01 -2.69741267e-01
1.50346771e-01 -4.82951015e-01 -6.41879961e-02 -1.37451172e-01
3.37634027e-01 -9.82801914e-01 -1.56888694e-01 4.54962581e-01
-3.12678069e-01 7.14071453e-01 1.09763756e-01 6.26185179e-01
-3.08603853e-01 -8.36844683e-01 3.60381722e-01 -6.30238831e-01
-4.94561583e-01 -1.32010534e-01 -6.74178541e-01 4.91508573e-01
8.11178505e-01 -1.55524835e-01 -1.23222493e-01 -1.78446565e-02
-5.72933137e-01 -8.09895620e-02 3.87873143e-01 4.01728034e-01
6.10696971e-01 -1.06371367e+00 -4.97002363e-01 8.51393119e-02
3.77839446e-01 -9.98086110e-02 -7.62864232e-01 4.96418983e-01
-5.26602089e-01 3.01361740e-01 1.61285922e-01 -5.51932454e-01
-1.33051670e+00 6.02730155e-01 -4.33690585e-02 -3.59127343e-01
-4.45899993e-01 7.02369869e-01 6.57811686e-02 -4.37047660e-01
1.77012205e-01 -5.84048867e-01 5.22031076e-02 1.06104501e-01
4.18855071e-01 -2.66616344e-01 3.25099796e-01 -6.51184738e-01
-5.11700571e-01 6.00605726e-01 -2.78890967e-01 -2.48007447e-01
1.28711092e+00 -7.20317736e-02 -5.95397830e-01 4.22950715e-01
1.05522323e+00 2.51756251e-01 8.71774852e-02 -4.78215069e-01
9.75793302e-01 -1.09656200e-01 -7.49255598e-01 -5.20611048e-01
-1.73517630e-01 5.17530978e-01 -1.86035961e-01 3.18641067e-01
6.33003712e-01 6.19666636e-01 9.57399964e-01 1.07673168e+00
3.44039530e-01 -9.22239900e-01 1.08826011e-01 9.35297072e-01
5.93328595e-01 -1.10118413e+00 -1.66077122e-01 -1.12555444e+00
-3.08951557e-01 1.21693361e+00 9.67777446e-02 1.29910126e-01
5.20907104e-01 3.83661538e-01 -9.42813680e-02 -3.77750099e-01
-5.63262582e-01 -3.21532309e-01 2.65156925e-01 3.69394958e-01
6.51318371e-01 -8.82604495e-02 -1.02823710e+00 7.21949220e-01
-4.71596032e-01 -6.01076245e-01 5.20453393e-01 1.14268005e+00
-5.75607121e-01 -1.87097120e+00 -3.21586519e-01 5.26573181e-01
-8.77532125e-01 -7.66001523e-01 -5.35905242e-01 8.69005382e-01
-1.55930251e-01 6.20997727e-01 -3.09029520e-02 5.85972071e-02
3.45213592e-01 4.88440245e-01 4.87373203e-01 -1.32633793e+00
-7.19625950e-01 -3.72263305e-02 3.76328290e-01 -3.57701749e-01
-4.42464799e-01 -7.27197528e-01 -1.61066985e+00 6.58294186e-02
-3.95778656e-01 5.53859055e-01 8.40324044e-01 1.20298636e+00
-4.48875129e-02 -1.22413091e-01 1.44298553e-01 -2.67178208e-01
-9.45951521e-01 -6.69419050e-01 -3.48839134e-01 7.14943647e-01
-8.10267180e-02 -5.08418500e-01 -2.86416590e-01 1.07394993e-01] | [10.345478057861328, 9.201251029968262] |
6a4eb7ff-1a7e-47d2-9881-2209ca6469ee | dehin-a-decentralized-framework-for-embedding | 2201.02757 | null | https://arxiv.org/abs/2201.02757v1 | https://arxiv.org/pdf/2201.02757v1.pdf | DeHIN: A Decentralized Framework for Embedding Large-scale Heterogeneous Information Networks | Modeling heterogeneity by extraction and exploitation of high-order information from heterogeneous information networks (HINs) has been attracting immense research attention in recent times. Such heterogeneous network embedding (HNE) methods effectively harness the heterogeneity of small-scale HINs. However, in the real world, the size of HINs grow exponentially with the continuous introduction of new nodes and different types of links, making it a billion-scale network. Learning node embeddings on such HINs creates a performance bottleneck for existing HNE methods that are commonly centralized, i.e., complete data and the model are both on a single machine. To address large-scale HNE tasks with strong efficiency and effectiveness guarantee, we present \textit{Decentralized Embedding Framework for Heterogeneous Information Network} (DeHIN) in this paper. In DeHIN, we generate a distributed parallel pipeline that utilizes hypergraphs in order to infuse parallelization into the HNE task. DeHIN presents a context preserving partition mechanism that innovatively formulates a large HIN as a hypergraph, whose hyperedges connect semantically similar nodes. Our framework then adopts a decentralized strategy to efficiently partition HINs by adopting a tree-like pipeline. Then, each resulting subnetwork is assigned to a distributed worker, which employs the deep information maximization theorem to locally learn node embeddings from the partition it receives. We further devise a novel embedding alignment scheme to precisely project independently learned node embeddings from all subnetworks onto a common vector space, thus allowing for downstream tasks like link prediction and node classification. | ['Kai Zheng', 'Zi Huang', 'Tong Chen', 'Hongzhi Yin', 'Mubashir Imran'] | 2022-01-08 | null | null | null | null | ['network-embedding'] | ['methodology'] | [-2.90216476e-01 5.51526666e-01 -4.72822756e-01 1.33363418e-02
-4.00657982e-01 -6.94801629e-01 3.38399738e-01 1.97312623e-01
-1.82568625e-01 4.00956929e-01 2.20920220e-01 -3.24683011e-01
-1.88998580e-01 -1.30698419e+00 -5.01141489e-01 -5.40420294e-01
-1.90180451e-01 8.65897417e-01 4.38491285e-01 -1.03061236e-02
-3.82802248e-01 4.49403793e-01 -1.01783311e+00 -3.03729111e-03
2.73121029e-01 6.89378798e-01 1.97823584e-01 9.09141302e-01
-2.98734933e-01 6.10813498e-01 -2.39820644e-01 -7.89375424e-01
4.88555074e-01 1.39313132e-01 -1.03248167e+00 -4.30900484e-01
1.45801812e-01 -2.75019139e-01 -1.06014228e+00 7.62345731e-01
6.00718439e-01 -2.75444895e-01 5.28051794e-01 -1.79509759e+00
-4.28719103e-01 1.09985745e+00 -5.43052673e-01 -4.10147384e-02
-1.01160303e-01 1.75289661e-01 1.47477555e+00 -5.51805317e-01
1.01715422e+00 1.08265424e+00 9.14597332e-01 3.86251509e-01
-1.58506370e+00 -6.63069427e-01 1.66748375e-01 -9.50393155e-02
-1.48143077e+00 -1.38418987e-01 8.73209774e-01 -4.43644881e-01
1.02469862e+00 4.17230666e-01 8.35225582e-01 1.06005073e+00
-7.03772157e-02 8.47800195e-01 2.91331202e-01 1.39792308e-01
6.13482222e-02 2.25751147e-01 1.69071242e-01 7.48056114e-01
3.76567990e-01 -3.19709092e-01 -1.98722586e-01 -3.05196792e-01
5.29232442e-01 4.24108148e-01 6.87588379e-03 -6.15833521e-01
-1.26972866e+00 7.69122422e-01 9.90208745e-01 3.09334636e-01
-3.56212348e-01 3.32543194e-01 8.43845427e-01 3.95569056e-01
4.44218785e-01 1.67781726e-01 -7.15226591e-01 2.13112816e-01
-7.52642989e-01 -5.58162034e-02 1.24800038e+00 1.08808565e+00
1.09977424e+00 -4.99860048e-01 -2.46730521e-02 5.78071415e-01
4.39541191e-01 2.42045641e-01 4.52173352e-02 -7.71259069e-01
7.35464454e-01 1.03219330e+00 -5.00249326e-01 -1.48403060e+00
-6.02226734e-01 -4.74424422e-01 -1.36509943e+00 -2.47724980e-01
-1.39402986e-01 -1.39267713e-01 -2.63096243e-01 1.91107917e+00
7.75051653e-01 1.27078846e-01 3.20506841e-02 6.38360202e-01
7.83688843e-01 7.30713904e-01 -1.02729723e-01 3.70628476e-01
1.22512293e+00 -1.49328053e+00 -2.20517159e-01 1.30693287e-01
9.94759321e-01 -1.90767258e-01 6.08947933e-01 -2.64825881e-01
-8.89040053e-01 -2.62838393e-01 -1.01380134e+00 -4.18588638e-01
-6.05366349e-01 -4.37099010e-01 8.11025083e-01 2.38757059e-01
-1.37905240e+00 5.38370371e-01 -8.71458352e-01 -3.75043631e-01
5.11556447e-01 6.89787030e-01 -8.20092201e-01 -3.44024859e-02
-1.10235262e+00 3.87961566e-01 6.44389629e-01 6.81184754e-02
-7.80699611e-01 -1.12751496e+00 -1.10545158e+00 6.14559412e-01
4.68812883e-01 -1.22682118e+00 6.94087505e-01 -4.74286020e-01
-1.19590998e+00 7.19420075e-01 -5.22811823e-02 -1.51245683e-01
2.99524695e-01 4.34135705e-01 -2.37363636e-01 1.59100220e-01
8.35564807e-02 7.54396915e-01 4.25982058e-01 -1.20344412e+00
-4.53238964e-01 -3.28942716e-01 1.62151605e-01 -1.48597434e-01
-9.98708129e-01 -1.69059873e-01 -8.96896720e-01 -3.29497814e-01
4.77993265e-02 -9.86130714e-01 -3.55885148e-01 1.02111287e-01
-9.19356167e-01 -3.82338524e-01 8.58319521e-01 -4.67455238e-01
1.59665966e+00 -2.06374669e+00 5.52065253e-01 6.50299668e-01
1.37448180e+00 8.65526572e-02 -4.60811853e-01 8.24254334e-01
2.03263480e-02 5.05975783e-01 -1.05023980e-01 -6.36097491e-01
5.10211885e-01 2.10028142e-01 1.36501119e-01 2.82405347e-01
5.97861893e-02 1.29876673e+00 -9.65495467e-01 -7.28525937e-01
-4.34621684e-02 3.62967521e-01 -8.96869481e-01 3.86391878e-01
-5.70239089e-02 -4.07218486e-02 -4.51778948e-01 4.08515811e-01
7.77434111e-01 -7.26037025e-01 5.78835905e-01 -4.47909683e-01
1.13106668e-01 6.00862391e-02 -1.07372570e+00 1.72268713e+00
-7.58997679e-01 5.72341681e-01 4.14372832e-01 -9.68253970e-01
6.27259493e-01 3.00652862e-01 8.44947040e-01 -7.05599114e-02
-8.59204009e-02 8.75782520e-02 -1.30435880e-02 -3.53013486e-01
2.13737965e-01 2.16076136e-01 -2.96407282e-01 5.91455281e-01
2.35556409e-01 3.20071578e-01 1.95306003e-01 8.23183656e-01
1.94379711e+00 -5.18441737e-01 8.59585181e-02 -1.48873001e-01
2.70372272e-01 -1.43465117e-01 5.69662511e-01 5.34530342e-01
-9.26829875e-02 2.36866757e-01 9.70611870e-01 -5.23195863e-01
-1.36215365e+00 -1.21382999e+00 1.63659632e-01 1.06170440e+00
1.84301689e-01 -9.82360840e-01 -5.49382091e-01 -1.08734167e+00
3.47496152e-01 -3.27957690e-01 -5.92114031e-01 -1.89435527e-01
-4.11215335e-01 -5.43034434e-01 5.08149266e-01 4.43398058e-01
2.58738220e-01 -8.00982773e-01 1.39153495e-01 2.83965498e-01
1.47410259e-01 -9.49225605e-01 -5.28158307e-01 1.90482616e-01
-4.99380708e-01 -1.15244079e+00 -5.02032638e-01 -9.53526497e-01
7.07488358e-01 3.27444464e-01 1.29988098e+00 2.77680278e-01
-4.12680179e-01 2.25488171e-01 -6.55146277e-06 4.12766308e-01
-2.22348645e-01 1.07124925e+00 -1.23874158e-01 -1.84529737e-01
3.75214815e-01 -9.59625244e-01 -8.13952982e-01 1.80989847e-01
-1.15663731e+00 3.66754234e-01 8.02087247e-01 8.55729580e-01
3.47389489e-01 2.17480659e-01 3.54741007e-01 -1.27185333e+00
3.89247864e-01 -1.25495970e+00 -5.64058483e-01 2.76715070e-01
-4.40010309e-01 1.39510259e-01 9.96581614e-01 -3.05516094e-01
-4.75764424e-01 -2.00521886e-01 -7.21279979e-02 -4.82556969e-01
1.61484644e-01 6.58695579e-01 -5.04253149e-01 7.12161511e-02
2.02094987e-01 -2.53764153e-01 -7.81439319e-02 -5.02089083e-01
5.08718967e-01 8.49350929e-01 2.09465504e-01 -4.96171534e-01
1.10473967e+00 3.73186499e-01 2.47598976e-01 -5.57831705e-01
-4.63995606e-01 -6.56438053e-01 -6.20935798e-01 2.13269517e-01
5.95411956e-01 -8.65826011e-01 -1.06739235e+00 1.12284958e-01
-1.29387748e+00 -4.13890749e-01 -8.76557827e-02 1.18446589e-01
-1.00869253e-01 1.46060660e-01 -9.86522377e-01 -1.71631560e-01
-4.33587581e-01 -1.11930907e+00 1.07885563e+00 -3.40888426e-02
-2.01696396e-01 -1.32629669e+00 5.11286259e-01 2.45167792e-01
5.99035084e-01 2.98556294e-02 1.05795252e+00 -8.08588207e-01
-9.19455886e-01 -3.08792055e-01 -6.80834532e-01 -3.34651135e-02
-2.16777146e-01 5.60087599e-02 -7.76156604e-01 -4.34455782e-01
-8.03401887e-01 -1.40246332e-01 8.30622017e-01 -1.98643863e-01
1.32864165e+00 -5.01830876e-01 -8.27481031e-01 1.10478187e+00
1.70210445e+00 -5.84113121e-01 2.83937752e-01 2.97891140e-01
1.23045647e+00 7.37382650e-01 -3.58516037e-01 2.75869370e-01
8.82948160e-01 6.60798848e-01 4.95520920e-01 -4.99604762e-01
-1.38671353e-01 -5.96689999e-01 1.81213140e-01 1.32587624e+00
5.02970159e-01 -4.89655972e-01 -1.02343512e+00 6.39287293e-01
-1.97226059e+00 -6.81934059e-01 1.89633481e-02 1.65433133e+00
7.29298651e-01 -1.34709254e-01 1.55761227e-01 -2.03545585e-01
9.26882386e-01 3.40534508e-01 -7.28494465e-01 -5.65211512e-02
1.09620579e-01 -1.40658915e-01 6.44868433e-01 4.18660522e-01
-1.06623363e+00 9.23621833e-01 5.24965143e+00 7.84995615e-01
-7.92134106e-01 2.69556314e-01 5.21553934e-01 -9.34130996e-02
-8.99600744e-01 1.72020823e-01 -8.05350840e-01 5.92557907e-01
1.08995020e+00 -2.20922515e-01 5.13664842e-01 8.99569511e-01
-2.62450606e-01 6.56046093e-01 -1.17758226e+00 9.07692850e-01
-3.47655207e-01 -1.60885167e+00 -8.28535482e-02 3.43036026e-01
8.67687285e-01 6.29530132e-01 -1.05131544e-01 5.08268476e-01
7.49701083e-01 -8.69853735e-01 1.65612727e-01 2.49076813e-01
7.42001116e-01 -6.27810419e-01 7.71190703e-01 2.82145083e-01
-1.72192895e+00 -1.15695491e-01 -5.80051839e-01 3.80131036e-01
1.99337170e-01 8.44704688e-01 -6.60196126e-01 6.50192022e-01
4.59096909e-01 7.57939637e-01 -4.81524646e-01 8.00963283e-01
7.26939365e-02 3.59045655e-01 -6.58725023e-01 1.75068632e-01
2.01083377e-01 -2.08856195e-01 4.46529925e-01 1.16830373e+00
7.67238736e-02 -3.79367739e-01 4.17269468e-01 8.66632223e-01
-7.54774928e-01 5.48487119e-02 -9.46282387e-01 -1.90095678e-01
8.70458543e-01 1.87100804e+00 -6.38309598e-01 -2.97489911e-01
-7.52830982e-01 1.07774055e+00 9.59850788e-01 3.82069737e-01
-7.74607062e-01 -6.78421080e-01 8.91314209e-01 1.14110939e-01
2.69062966e-01 -3.07743996e-02 4.20075022e-02 -1.33447659e+00
1.17903732e-01 -4.27733719e-01 4.51815039e-01 -1.79690719e-01
-1.56653762e+00 7.64637232e-01 -2.61573076e-01 -7.84748256e-01
1.33426309e-01 -5.00079513e-01 -6.91225946e-01 7.35473096e-01
-1.52838826e+00 -1.54928744e+00 -2.94750959e-01 3.82363528e-01
-7.95598328e-02 -1.71986833e-01 7.68636823e-01 6.53764486e-01
-1.15201688e+00 8.79430115e-01 1.78891405e-01 4.19008732e-01
3.95615458e-01 -1.33197045e+00 5.90021670e-01 4.66664553e-01
5.19049652e-02 8.32666516e-01 9.55728740e-02 -6.19496882e-01
-1.66586697e+00 -1.38999975e+00 1.01105785e+00 -2.32375503e-01
1.17545485e+00 -8.66106570e-01 -6.66938782e-01 9.19651389e-01
-1.58180930e-02 6.26428068e-01 9.08918917e-01 4.13247645e-01
-6.53498948e-01 -3.74283373e-01 -1.03515017e+00 7.27273226e-01
1.34525061e+00 -8.10398638e-01 7.59719461e-02 3.26642066e-01
1.28132594e+00 1.35618672e-01 -1.53829086e+00 2.41397426e-01
5.47469020e-01 -6.79338992e-01 9.42692161e-01 -8.85460377e-01
4.18940037e-01 -3.10703546e-01 -4.60957410e-03 -1.25593936e+00
-6.64284825e-01 -8.40555012e-01 -4.55471069e-01 1.44976056e+00
6.16877377e-01 -7.88528979e-01 1.12435508e+00 5.69299936e-01
-4.23644856e-02 -9.04996097e-01 -7.07973599e-01 -5.03857613e-01
-1.42607227e-01 -1.38242111e-01 1.04288232e+00 9.00558412e-01
1.54886380e-01 5.89293659e-01 -3.02320197e-02 3.17886204e-01
5.67570388e-01 5.70693500e-02 1.10382950e+00 -1.38867486e+00
-5.39670348e-01 -5.31580627e-01 -8.55915070e-01 -1.11987281e+00
6.18577600e-01 -1.48205256e+00 -3.63914698e-01 -1.55562007e+00
7.37450838e-01 -8.20092201e-01 -3.75842959e-01 6.38806343e-01
-7.58573338e-02 2.01140344e-01 2.88317859e-01 3.27915907e-01
-9.77183998e-01 5.66649079e-01 9.03776884e-01 -4.03868496e-01
4.45048064e-02 -2.28498176e-01 -6.57171428e-01 4.14270371e-01
6.94247842e-01 -5.80948114e-01 -4.41115499e-01 -6.57186091e-01
6.88168526e-01 -2.42509805e-02 2.29868978e-01 -7.17395484e-01
5.57770967e-01 3.40811670e-01 8.84098858e-02 -3.46178383e-01
7.27861226e-02 -1.35130811e+00 4.45426822e-01 2.30548874e-01
-3.24648082e-01 2.51707405e-01 -1.55621856e-01 8.20159495e-01
-1.64165109e-01 1.11964673e-01 4.57403809e-01 1.25441045e-01
-3.67807686e-01 9.77076650e-01 7.42705464e-02 1.08732328e-01
1.37263548e+00 2.38019064e-01 -5.10443747e-01 3.55740339e-02
-6.42800570e-01 6.30880952e-01 7.53824055e-01 2.23078698e-01
3.08421582e-01 -1.40066302e+00 -5.68156481e-01 3.32716495e-01
2.11135685e-01 4.78260756e-01 3.98673147e-01 9.10023570e-01
-5.69388747e-01 2.45994598e-01 2.45700195e-01 -3.39837611e-01
-1.00937450e+00 7.68195868e-01 7.66039938e-02 -7.95889080e-01
-7.63930678e-01 7.10603774e-01 3.61815125e-01 -9.11084592e-01
2.05748767e-01 5.40464707e-02 1.71948984e-01 1.08214887e-02
2.95318216e-01 4.69313622e-01 -1.33144170e-01 -3.65161091e-01
-1.66464448e-01 1.10463023e-01 -1.52698770e-01 1.70604303e-01
1.51920772e+00 -1.95726544e-01 -7.55332649e-01 2.20346808e-01
2.04852700e+00 -3.68451215e-02 -1.03447235e+00 -3.25325519e-01
1.05806254e-01 -1.87218234e-01 9.53466445e-02 -2.16634348e-01
-1.46830428e+00 6.73704267e-01 -4.24883589e-02 4.39350545e-01
7.67734110e-01 4.33954388e-01 1.22693110e+00 4.43800628e-01
1.62234277e-01 -7.23344088e-01 -2.29758352e-01 3.87524813e-01
2.56636053e-01 -1.06009650e+00 -3.87395203e-01 -4.40830201e-01
-1.63823813e-01 1.09305453e+00 7.04550683e-01 2.30292585e-02
9.91212904e-01 5.68562388e-01 -4.04768616e-01 -4.19461131e-01
-1.09842968e+00 -2.52072904e-02 -8.73075947e-02 5.53184450e-01
2.81490143e-02 1.78982243e-01 1.63302407e-01 7.99504757e-01
-4.11742367e-02 -3.60405713e-01 3.80872041e-02 4.76925135e-01
-1.03613466e-01 -1.40827048e+00 1.89260453e-01 6.03627980e-01
2.56987358e-03 -2.22206861e-01 -2.74846017e-01 6.90657020e-01
1.00226313e-01 3.98291230e-01 2.20086858e-01 -7.43992448e-01
3.95305008e-02 -2.86740690e-01 -1.00255810e-01 -8.14060450e-01
-6.92496717e-01 -3.44559461e-01 -1.14606030e-01 -5.72772682e-01
2.15283304e-01 -1.39554963e-01 -1.10797572e+00 -8.07991743e-01
-1.27696112e-01 1.37327537e-01 5.59932470e-01 5.02874911e-01
8.28866661e-01 5.85526049e-01 1.01423967e+00 -8.40354025e-01
-3.70234698e-01 -7.72145748e-01 -6.53593600e-01 3.80299628e-01
8.02352726e-02 -3.25592726e-01 -5.59111476e-01 -5.20276010e-01] | [7.213281631469727, 6.221591949462891] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.