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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
57dc3ecd-3dbb-49db-b84d-c4da58f208c2 | hopretriever-retrieve-hops-over-wikipedia-to | 2012.15534 | null | https://arxiv.org/abs/2012.15534v1 | https://arxiv.org/pdf/2012.15534v1.pdf | HopRetriever: Retrieve Hops over Wikipedia to Answer Complex Questions | Collecting supporting evidence from large corpora of text (e.g., Wikipedia) is of great challenge for open-domain Question Answering (QA). Especially, for multi-hop open-domain QA, scattered evidence pieces are required to be gathered together to support the answer extraction. In this paper, we propose a new retrieval target, hop, to collect the hidden reasoning evidence from Wikipedia for complex question answering. Specifically, the hop in this paper is defined as the combination of a hyperlink and the corresponding outbound link document. The hyperlink is encoded as the mention embedding which models the structured knowledge of how the outbound link entity is mentioned in the textual context, and the corresponding outbound link document is encoded as the document embedding representing the unstructured knowledge within it. Accordingly, we build HopRetriever which retrieves hops over Wikipedia to answer complex questions. Experiments on the HotpotQA dataset demonstrate that HopRetriever outperforms previously published evidence retrieval methods by large margins. Moreover, our approach also yields quantifiable interpretations of the evidence collection process. | ['Bingquan Liu', 'Zhenzhou Ji', 'Chengjie Sun', 'Qun Liu', 'Xin Jiang', 'Lifeng Shang', 'Xiaoguang Li', 'Shaobo Li'] | 2020-12-31 | null | null | null | null | ['document-embedding'] | ['methodology'] | [-3.83961231e-01 8.04023445e-01 -4.44917083e-01 -4.32625934e-02
-1.34139633e+00 -7.85907030e-01 6.13213897e-01 9.25583959e-01
-3.63611609e-01 1.08059180e+00 6.56513691e-01 -4.15813178e-01
-6.26661122e-01 -8.48044097e-01 -1.16424549e+00 -2.88009405e-01
1.55456081e-01 7.16398716e-01 7.64977038e-01 -4.50744241e-01
2.56117702e-01 -2.61016101e-01 -1.04705513e+00 3.91460717e-01
1.31975210e+00 1.16665971e+00 1.52783826e-01 2.41476581e-01
-6.26569271e-01 1.05365205e+00 -6.60183191e-01 -8.67711186e-01
-3.25191438e-01 -2.26327460e-02 -1.28676772e+00 -3.49844992e-01
7.77165651e-01 -2.04653487e-01 -3.51475418e-01 1.09337330e+00
2.43313998e-01 -1.89887017e-01 7.96178162e-01 -1.13919032e+00
-7.26131976e-01 9.73384202e-01 -2.09947929e-01 4.60727036e-01
5.57189167e-01 -1.37225956e-01 1.64831316e+00 -9.78076935e-01
1.06652927e+00 1.12104809e+00 2.08050832e-01 1.14467807e-01
-6.12839937e-01 -9.11886916e-02 -5.38893230e-02 7.34172404e-01
-1.03441167e+00 -1.78712327e-02 7.28761494e-01 -2.20345944e-01
7.22635984e-01 3.22282553e-01 4.48085517e-01 8.70239973e-01
-1.74513996e-01 1.07801688e+00 7.96120346e-01 -5.87401509e-01
2.00117126e-01 2.74614453e-01 8.47746789e-01 1.01901984e+00
6.95805490e-01 -6.35056496e-01 -4.11631674e-01 -4.33721185e-01
8.79777521e-02 -4.71794844e-01 -5.45362830e-01 -1.74186528e-01
-1.08431602e+00 8.45116675e-01 7.13967502e-01 1.31332606e-01
-5.26528597e-01 3.33288759e-02 4.08249617e-01 1.39372453e-01
1.64753601e-01 4.72165316e-01 -5.42295873e-01 1.41493663e-01
-4.20831829e-01 4.36934918e-01 1.18140733e+00 9.33702290e-01
8.65906894e-01 -8.74517798e-01 -4.68113422e-01 9.19876337e-01
6.28243446e-01 5.77227652e-01 1.08489729e-01 -1.27373934e+00
1.33807778e+00 1.11051202e+00 3.00520301e-01 -1.16796303e+00
-3.80880162e-02 -3.33011508e-01 -1.89488843e-01 -7.20549524e-01
4.75110441e-01 -1.97279289e-01 -3.34607363e-01 1.45468020e+00
8.03080797e-01 -9.82136372e-03 4.43296373e-01 9.87805009e-01
1.36770296e+00 8.93827200e-01 -8.33021626e-02 1.15266345e-01
1.91703880e+00 -1.32022285e+00 -9.77144480e-01 -4.03854668e-01
8.07090700e-01 -5.08356273e-01 9.92044985e-01 4.98559624e-02
-9.36228454e-01 1.24654099e-01 -1.26686800e+00 -5.86027622e-01
-3.96363229e-01 1.64107278e-01 1.40875876e-01 -1.83756009e-01
-4.07818496e-01 2.97062565e-02 -6.14302196e-02 -4.30979319e-02
1.77111402e-01 -3.50517482e-01 -3.99606049e-01 -6.71985447e-01
-1.71692240e+00 1.01463687e+00 7.00302362e-01 1.48657262e-01
-6.01217330e-01 -5.32870412e-01 -8.92315567e-01 3.82856190e-01
1.13773131e+00 -9.20468032e-01 1.15714478e+00 7.75083452e-02
-7.22614586e-01 6.45802081e-01 -1.79240018e-01 -4.14064288e-01
1.46338373e-01 -3.58034998e-01 -5.48735738e-01 7.23959744e-01
4.90597010e-01 2.78021544e-01 5.97824872e-01 -1.31735003e+00
-5.19677520e-01 -4.36724961e-01 6.43083453e-01 1.48669913e-01
-5.32717824e-01 -1.84929594e-01 -8.44633579e-01 -2.76884317e-01
4.56068739e-02 -7.16489971e-01 2.28074774e-01 -1.37527898e-01
-6.65124893e-01 -9.59057033e-01 6.82787657e-01 -9.17589486e-01
1.63255715e+00 -1.99298728e+00 3.64951879e-01 2.16889873e-01
6.34351373e-01 1.41507745e-01 -3.65627259e-02 7.13930130e-01
5.96384406e-01 9.70311761e-02 -2.68976450e-01 2.91747183e-01
4.78553355e-01 5.10105491e-01 -7.21328974e-01 -1.14845239e-01
2.61423647e-01 1.11072445e+00 -1.26021540e+00 -1.13117921e+00
-5.47471046e-01 -4.62683663e-02 -3.64476979e-01 5.40765040e-02
-8.68445575e-01 -2.85708338e-01 -9.97442663e-01 7.82838464e-01
2.96882302e-01 -7.13536143e-01 9.31339189e-02 -5.27207911e-01
2.72918463e-01 7.18619645e-01 -8.80938530e-01 1.51617873e+00
-2.39501432e-01 5.23586571e-01 4.00834344e-02 -4.46793735e-01
7.53148615e-01 2.95409322e-01 -1.79190427e-01 -6.33228719e-01
-2.06302464e-01 5.98201692e-01 -1.69989288e-01 -9.16957200e-01
7.40604460e-01 3.06633592e-01 -2.28122160e-01 4.37965006e-01
1.44668072e-01 -4.71081994e-02 7.73001790e-01 1.01153326e+00
1.42038608e+00 -3.77914488e-01 1.49021670e-01 -2.80974098e-02
7.77921081e-01 3.84928674e-01 2.98520803e-01 6.19633794e-01
7.57156312e-02 -8.48549828e-02 8.20535004e-01 1.36619598e-01
-8.62768352e-01 -1.11365342e+00 -5.19830734e-02 6.24068439e-01
3.42901736e-01 -8.61266434e-01 -5.68982661e-01 -1.18325996e+00
2.97461271e-01 7.14758575e-01 -5.32179415e-01 -1.21447839e-01
-7.10911274e-01 1.85151324e-02 6.11107171e-01 3.79511654e-01
5.06754816e-01 -9.88894045e-01 -1.50320813e-01 2.27145121e-01
-9.70344365e-01 -1.26374781e+00 -3.53504241e-01 -1.02515280e-01
-6.81159496e-01 -1.48230469e+00 -6.48458004e-01 -7.13701189e-01
5.54640234e-01 1.88679427e-01 1.36091089e+00 4.75921512e-01
6.12299517e-02 4.98102516e-01 -5.92086017e-01 -3.01885694e-01
-4.61605601e-02 -5.93869649e-02 -3.80988389e-01 -2.33242407e-01
4.74263370e-01 -3.48137617e-02 -6.75449848e-01 2.01159403e-01
-1.08231235e+00 -5.12677073e-01 5.36893249e-01 8.86157990e-01
5.99355042e-01 -2.22182069e-02 7.12630808e-01 -7.81123698e-01
1.18615639e+00 -1.11617959e+00 -3.74743491e-01 9.52067673e-01
-5.89378297e-01 4.85115170e-01 3.34369212e-01 -1.41058072e-01
-1.02322149e+00 -8.83520067e-01 3.72381806e-02 -1.48348287e-01
3.94722342e-01 1.28186488e+00 -2.59673167e-02 4.01212960e-01
6.82623804e-01 -7.93117136e-02 -4.03077900e-01 -4.76375312e-01
7.38897979e-01 7.03799367e-01 3.71707886e-01 -1.09466779e+00
7.81274378e-01 3.93454254e-01 -2.10454717e-01 -6.13597095e-01
-1.50737047e+00 -6.58398390e-01 -1.38244703e-01 -2.73730487e-01
7.35514224e-01 -6.52097106e-01 -3.74538749e-01 -3.85793924e-01
-1.48144686e+00 1.29679725e-01 -3.47514898e-01 3.76939416e-01
-2.03192696e-01 5.63502014e-01 -6.76925421e-01 -5.37368476e-01
-3.84128332e-01 -7.69688129e-01 1.19760621e+00 1.56537145e-01
-1.48869827e-01 -1.01211536e+00 3.90982985e-01 1.09625053e+00
-1.06050886e-01 -1.04558885e-01 1.50447845e+00 -7.87767231e-01
-9.76801753e-01 -3.42743337e-01 -4.21844989e-01 8.00848529e-02
-3.36688668e-01 -2.13979036e-01 -5.07725596e-01 2.24056229e-01
-2.32188985e-01 -9.19765294e-01 8.65863919e-01 -2.35698044e-01
8.01035285e-01 -8.19186807e-01 -1.99098289e-01 -3.87561142e-01
1.17356241e+00 -4.99731511e-01 4.64001089e-01 6.37172222e-01
3.74843031e-01 8.52451682e-01 9.34453130e-01 1.06117621e-01
8.95347059e-01 4.99616683e-01 3.62617850e-01 5.23238122e-01
-1.89122736e-01 -6.37771130e-01 2.16001049e-01 1.31077874e+00
1.62934244e-01 -3.88854802e-01 -9.47799742e-01 6.92196369e-01
-1.86596656e+00 -1.01820958e+00 -5.23952127e-01 1.62452435e+00
1.11892462e+00 6.38486668e-02 -3.85003686e-01 4.30635102e-02
3.62164646e-01 1.50705457e-01 -4.15304869e-01 2.06679761e-01
-3.10955137e-01 -1.42053619e-01 -8.86087716e-02 6.55302584e-01
-5.18593431e-01 6.61200166e-01 4.87171602e+00 1.09749317e+00
-1.38815448e-01 2.61578947e-01 -1.97641224e-01 3.02773595e-01
-8.28252435e-01 3.34679216e-01 -1.05206501e+00 3.99071753e-01
5.15120625e-01 -3.75490934e-01 -6.45400435e-02 6.28398180e-01
-4.39596355e-01 -3.12192202e-01 -9.05513227e-01 4.42847729e-01
2.96746790e-01 -1.44658196e+00 3.66734475e-01 2.84466781e-02
6.11791492e-01 -1.18714795e-01 -1.51272014e-01 5.06644011e-01
3.46277386e-01 -3.68940651e-01 6.40925884e-01 5.66041470e-01
3.00170124e-01 -2.77912498e-01 9.48075294e-01 6.99957192e-01
-1.02757204e+00 -2.31498256e-01 -3.45446527e-01 5.50607443e-01
4.57462996e-01 8.79417002e-01 -9.17789340e-01 8.21740568e-01
5.27183652e-01 3.28510046e-01 -5.31867564e-01 8.89759243e-01
-9.56974685e-01 7.21300960e-01 -2.16377914e-01 -5.32583416e-01
6.22485280e-01 -1.92680612e-01 7.22542822e-01 1.00422370e+00
3.04910690e-01 4.54849452e-01 -6.63494170e-02 8.79309297e-01
-7.03465402e-01 1.74543172e-01 -4.87386286e-01 -4.27503526e-01
7.13923156e-01 1.22726381e+00 -8.52676034e-02 -6.90563917e-01
-3.26402754e-01 5.67639828e-01 9.41286206e-01 3.44761103e-01
-7.63244331e-01 -5.69518447e-01 1.47398869e-02 -2.57453829e-01
3.05516690e-01 -4.18138457e-03 3.37749571e-01 -1.20573676e+00
8.01801383e-01 -6.93612635e-01 9.84450281e-01 -1.19561148e+00
-1.58748376e+00 4.55425709e-01 2.82051247e-02 -9.71175373e-01
-2.03381032e-01 -4.30985719e-01 -2.22384483e-01 6.24509692e-01
-1.99884307e+00 -7.67383695e-01 -2.48827443e-01 5.28327882e-01
3.57026160e-01 5.52150384e-02 5.93178988e-01 3.99568707e-01
-4.72401798e-01 2.17957690e-01 2.49756547e-03 2.51155734e-01
5.99147797e-01 -1.24423528e+00 -4.26292568e-01 5.58785677e-01
4.15272236e-01 9.62918639e-01 5.25194764e-01 -8.66452336e-01
-1.58838511e+00 -8.03174317e-01 1.20005405e+00 -7.37372041e-01
1.32213235e+00 1.94230840e-01 -1.19559634e+00 5.71739733e-01
4.27886486e-01 -8.49550217e-02 6.52605712e-01 3.08253616e-01
-9.18217182e-01 -3.73163149e-02 -8.79356444e-01 5.65194309e-01
6.06507003e-01 -8.25642347e-01 -1.55025542e+00 6.15040600e-01
1.15433621e+00 -2.63250113e-01 -9.35966074e-01 3.45236838e-01
5.47438040e-02 -2.69878328e-01 9.26647484e-01 -8.19065571e-01
9.55155015e-01 -4.14178342e-01 -3.81350160e-01 -1.13432002e+00
1.59957826e-01 -1.76511705e-02 -1.08141899e+00 1.16512573e+00
1.01527834e+00 -3.69428754e-01 3.75247151e-01 4.68397558e-01
-3.06612164e-01 -9.02114213e-01 -9.97291625e-01 -7.36296475e-01
-1.73494235e-01 -5.56312352e-02 3.32404584e-01 9.21733201e-01
5.43131649e-01 6.95193291e-01 1.56274244e-01 3.99117023e-01
7.70956039e-01 3.56796414e-01 5.37308931e-01 -1.13912904e+00
-1.93391323e-01 -5.40181622e-02 1.10418923e-01 -1.41850603e+00
2.38347366e-01 -1.13134909e+00 6.17058091e-02 -2.14824486e+00
3.78233433e-01 -6.00139499e-01 -1.28941089e-01 1.93158329e-01
-4.05337989e-01 -2.05338821e-01 -5.50535321e-02 2.94794858e-01
-1.18925786e+00 7.12897360e-01 1.47029507e+00 -6.01309180e-01
4.54880953e-01 -6.82008982e-01 -7.29667008e-01 6.24747813e-01
3.81482631e-01 -5.80898464e-01 -4.17562336e-01 -7.32490957e-01
1.00090408e+00 3.13834548e-01 5.44955492e-01 -4.18395698e-01
7.06199825e-01 8.33802372e-02 -3.64333361e-01 -5.97585142e-01
5.67920387e-01 -8.89392972e-01 -5.41078985e-01 1.92698702e-01
-5.17440856e-01 -6.27121702e-02 -2.05746353e-01 9.38462257e-01
-6.40109658e-01 -7.81579912e-01 -2.12486595e-01 -3.65434401e-02
-5.69071949e-01 9.52028632e-02 1.60618961e-01 9.04403925e-01
5.48127711e-01 4.48795438e-01 -1.27920961e+00 -2.10423604e-01
-4.81876343e-01 7.72340775e-01 -1.35287076e-01 2.87459701e-01
9.09473479e-01 -1.43083167e+00 -7.72517622e-01 -6.19203687e-01
6.52116060e-01 1.15635000e-01 2.18657151e-01 9.44664299e-01
-2.01833859e-01 7.83376515e-01 4.17174578e-01 -2.34825671e-01
-1.07493389e+00 4.72341031e-01 -2.57922620e-01 -6.15479112e-01
-6.65143788e-01 6.37233615e-01 -3.50666970e-01 -4.31691378e-01
1.87547207e-01 -2.51796722e-01 -5.31192899e-01 2.88035840e-01
6.05262697e-01 4.24589008e-01 1.17413767e-01 -1.60500556e-01
-3.84138301e-02 3.02308708e-01 -1.29067823e-01 -2.71998227e-01
1.14833152e+00 -1.22273155e-01 -6.96153224e-01 4.77903187e-01
1.04080367e+00 4.30312604e-01 -6.42372787e-01 -6.39246702e-01
4.64920908e-01 -2.22249642e-01 -9.03524831e-02 -1.03801417e+00
-4.90244418e-01 6.70722663e-01 -2.23621845e-01 3.50294977e-01
4.17685539e-01 7.96149731e-01 1.11948395e+00 1.14589667e+00
3.79921794e-01 -1.07955360e+00 3.80685449e-01 7.09363461e-01
1.15339851e+00 -1.23035157e+00 -5.94177060e-02 -5.68480611e-01
-5.53148150e-01 8.45336437e-01 5.77265024e-01 2.98029035e-01
3.08900565e-01 -2.70867646e-01 -1.79657772e-01 -1.10360873e+00
-8.56040418e-01 -2.28814393e-01 6.45426989e-01 -5.68791702e-02
-7.21073300e-02 -1.21797979e-01 -5.72526813e-01 6.06094837e-01
-1.25874039e-02 -2.26222262e-01 2.68005133e-01 8.53923142e-01
-8.17167878e-01 -6.75356030e-01 -2.97760725e-01 4.97862130e-01
-2.10051969e-01 -2.78568894e-01 -7.19195127e-01 6.70086265e-01
-7.60768130e-02 1.08345687e+00 -4.34792429e-01 7.43465945e-02
3.50648075e-01 2.91122615e-01 4.16839093e-01 -4.62462068e-01
-2.39291430e-01 -5.54466605e-01 6.96174145e-01 -1.99078381e-01
-3.35517287e-01 -2.02048004e-01 -1.50477219e+00 8.51972848e-02
-5.87716281e-01 9.45433199e-01 6.45986021e-01 1.16746366e+00
3.52062106e-01 3.03589970e-01 2.13380024e-01 3.46030235e-01
-7.66107202e-01 -9.68438387e-01 -3.81764412e-01 4.38550711e-01
3.41010451e-01 -7.96860695e-01 -6.10460103e-01 -3.19931537e-01] | [10.810441017150879, 7.902409076690674] |
1d9c0360-49b3-41ec-8f3a-d8da687fd0fc | ask-adaptively-selecting-key-local-features | 2110.07703 | null | https://arxiv.org/abs/2110.07703v1 | https://arxiv.org/pdf/2110.07703v1.pdf | ASK: Adaptively Selecting Key Local Features for RGB-D Scene Recognition | Indoor scene images usually contain scattered objects and various scene layouts, which make RGB-D scene classification a challenging task. Existing methods still have limitations for classifying scene images with great spatial variability. Thus, how to extract local patch-level features effectively using only image labels is still an open problem for RGB-D scene recognition. In this paper, we propose an efficient framework for RGB-D scene recognition, which adaptively selects important local features to capture the great spatial variability of scene images. Specifically, we design a differentiable local feature selection (DLFS) module, which can extract the appropriate number of key local scenerelated features. Discriminative local theme-level and object-level representations can be selected with the DLFS module from the spatially-correlated multi-modal RGB-D features. We take advantage of the correlation between RGB and depth modalities to provide more cues for selecting local features. To ensure that discriminative local features are selected, the variational mutual information maximization loss is proposed. Additionally, the DLFS module can be easily extended to select local features of different scales. By concatenating the local-orderless and global structured multi-modal features, the proposed framework can achieve state-of-the-art performance on public RGB-D scene recognition datasets. | ['Qi Wang', 'Yuan Yuan', 'Zhitong Xiong'] | 2021-10-14 | null | null | null | null | ['scene-recognition'] | ['computer-vision'] | [ 2.70138204e-01 -6.89816535e-01 -2.58030504e-01 -6.90648019e-01
-1.01775289e+00 -4.38390225e-01 4.65509444e-01 1.08896419e-01
-2.46195018e-01 2.78086156e-01 7.44064059e-03 1.50061667e-01
-5.67297518e-01 -8.41627955e-01 -2.90444672e-01 -1.16073608e+00
1.75718501e-01 -1.98637858e-01 2.99657285e-01 9.65368822e-02
4.14425790e-01 9.37416792e-01 -1.78533602e+00 3.44319224e-01
6.39443457e-01 1.58161271e+00 7.15902150e-01 4.43758816e-01
-4.52271342e-01 7.75354266e-01 -3.05387199e-01 3.90687972e-01
3.19099605e-01 -4.10590887e-01 -5.65590262e-01 6.55733526e-01
5.03639102e-01 -1.77440837e-01 -4.45218772e-01 9.10705209e-01
6.43348217e-01 3.16572100e-01 6.15614057e-01 -1.03399324e+00
-4.04788077e-01 -3.26877177e-01 -4.84188437e-01 4.02707845e-01
5.28160572e-01 1.34817705e-01 9.64720964e-01 -1.04352701e+00
3.62658143e-01 1.14754808e+00 3.09824556e-01 1.22861974e-01
-1.02463412e+00 -3.65258932e-01 3.71039599e-01 3.48791420e-01
-1.61804521e+00 -3.95335883e-01 1.40294111e+00 -3.32263231e-01
7.45480597e-01 5.25326788e-01 5.00341058e-01 5.54728448e-01
-3.23846787e-02 9.74997759e-01 1.38487887e+00 -3.46675366e-01
1.56058475e-01 4.12370674e-02 -3.19295079e-02 6.44888818e-01
-1.36544108e-01 -1.91365197e-01 -8.22753668e-01 5.71385026e-02
8.93081009e-01 5.71411610e-01 -3.49749684e-01 -6.18383527e-01
-1.21210361e+00 6.19501948e-01 7.97661185e-01 5.21210790e-01
-4.08309758e-01 -1.10792466e-01 2.02188082e-02 3.85121554e-02
3.48136306e-01 9.14596692e-02 -4.94801700e-01 -2.92332135e-02
-6.45628512e-01 -1.24665082e-01 3.02670319e-02 8.03244472e-01
1.32759404e+00 -4.68513131e-01 -2.31188610e-01 1.07206070e+00
5.42321026e-01 7.13913679e-01 2.63074458e-01 -8.22344661e-01
5.50636888e-01 9.85724747e-01 -1.45951137e-01 -1.34777939e+00
-5.47218442e-01 -1.89324766e-01 -9.50424671e-01 -4.37076055e-02
5.76448068e-02 5.17400205e-01 -8.55712175e-01 1.15244710e+00
6.70064509e-01 3.85635942e-02 -8.99056792e-02 1.14212322e+00
1.08282053e+00 6.67088687e-01 -1.58115804e-01 -1.55911729e-01
1.03192866e+00 -4.71776515e-01 -3.67913902e-01 -2.17287093e-01
2.94112355e-01 -7.48609900e-01 1.21520436e+00 -8.20066221e-03
-5.53494394e-01 -6.58893049e-01 -9.05641615e-01 -2.72786051e-01
-3.73560518e-01 2.60781109e-01 7.97903776e-01 4.62404609e-01
-5.86794376e-01 1.40027791e-01 -7.71986604e-01 -2.88297892e-01
5.47933400e-01 4.13816243e-01 -5.30831218e-01 -5.67967594e-01
-7.62072682e-01 4.40052748e-01 2.61728555e-01 3.16086739e-01
-6.31083548e-01 -3.14638764e-01 -9.81668890e-01 -5.51688671e-01
2.77894229e-01 -3.17029417e-01 5.99106550e-01 -5.79196453e-01
-1.52152896e+00 9.86355901e-01 -3.86738330e-01 6.42171443e-01
2.54314020e-02 2.68968016e-01 -3.62205207e-01 4.05678272e-01
1.57484204e-01 4.66252752e-02 7.69178212e-01 -1.36175776e+00
-7.48128712e-01 -7.15762198e-01 2.06170212e-02 5.35494149e-01
-3.59647155e-01 -2.34522521e-01 -6.74052477e-01 -4.56019670e-01
8.73340786e-01 -5.77076852e-01 -2.69632876e-01 2.15289190e-01
-4.64212149e-01 -2.39332587e-01 7.60215700e-01 -2.67213017e-01
9.70901728e-01 -2.35312366e+00 3.11128587e-01 4.65804636e-01
7.72039741e-02 -2.24196211e-01 -7.52286911e-02 8.14208612e-02
3.90896022e-01 -2.64065832e-01 -2.92812526e-01 -3.00869465e-01
-7.55622312e-02 2.51348019e-01 6.62585795e-02 7.66249418e-01
1.68391347e-01 7.49097943e-01 -8.13117623e-01 -6.65376723e-01
8.16697478e-01 3.96425962e-01 -4.14677024e-01 2.98106164e-01
6.73475415e-02 7.21402287e-01 -1.26359522e+00 1.02044642e+00
9.33064163e-01 -2.60904580e-01 -2.83729166e-01 -3.73129100e-01
-2.61656493e-01 1.40740335e-01 -1.40052497e+00 2.08238912e+00
-6.59864962e-01 3.41632843e-01 -1.66137829e-01 -1.04353631e+00
1.17382038e+00 -2.90797591e-01 7.07649112e-01 -1.02364480e+00
9.02517289e-02 3.02952796e-01 -7.34261155e-01 -7.33820140e-01
1.14899471e-01 -9.60768480e-03 -2.79078901e-01 1.46157369e-01
1.77647767e-03 -4.29886639e-01 -4.55468208e-01 -2.62861431e-01
9.62870419e-01 -3.72885563e-03 2.69779474e-01 -6.73091486e-02
7.48981416e-01 -8.04489329e-02 4.79712784e-01 6.18911862e-01
-2.90431440e-01 6.85387671e-01 -1.31398395e-01 -3.74177307e-01
-5.30210614e-01 -1.05506277e+00 -4.41298217e-01 7.76506960e-01
7.56400347e-01 -2.36151978e-01 -9.04472321e-02 -7.30141759e-01
1.30831376e-01 1.81460172e-01 -4.51840043e-01 -2.23389938e-01
-4.51650113e-01 -5.98774970e-01 6.92527890e-02 2.36769632e-01
8.15718234e-01 -7.76760042e-01 -6.93351388e-01 9.67269880e-05
-7.65685365e-02 -1.00633872e+00 -2.86576658e-01 4.23058897e-01
-8.92672837e-01 -9.75497186e-01 -5.36567330e-01 -8.31125200e-01
7.83107579e-01 9.12175238e-01 7.03901768e-01 -1.00209586e-01
-4.25390512e-01 5.53025126e-01 -5.23123145e-01 2.72356689e-01
3.13107133e-01 3.97143327e-02 -1.43497959e-01 2.62295097e-01
4.81038958e-01 -5.20490348e-01 -8.04788888e-01 4.37885940e-01
-9.57678199e-01 8.60400498e-02 6.49847269e-01 8.17066014e-01
1.16288102e+00 2.41380602e-01 8.67538899e-02 -2.33573645e-01
2.91992147e-02 -3.19802612e-01 -3.63870651e-01 3.47199231e-01
-5.29054478e-02 -2.29245368e-02 5.34011483e-01 -1.85000911e-01
-9.51380432e-01 2.91703463e-01 -1.27251551e-01 -4.39346999e-01
-5.04738390e-01 4.59988981e-01 -6.78989589e-01 -4.40518856e-01
3.01162779e-01 6.94393814e-01 -4.42717075e-01 -6.32947028e-01
3.19527030e-01 8.99944246e-01 6.01674914e-02 -6.10194266e-01
7.60122955e-01 4.86531824e-01 1.57751784e-01 -1.16344333e+00
-9.87709463e-01 -8.28821063e-01 -8.84594560e-01 -2.15039775e-01
7.88948596e-01 -9.66067314e-01 -6.64928734e-01 6.65912271e-01
-7.57096887e-01 -2.68429399e-01 -2.34122679e-01 4.05626446e-01
-3.88651401e-01 2.03937978e-01 -3.06913644e-01 -7.97256768e-01
3.35154295e-01 -1.25155139e+00 1.59697413e+00 3.81638736e-01
4.30227667e-01 -7.52046883e-01 1.03479018e-02 2.72636563e-01
3.34712148e-01 3.36360455e-01 8.06529343e-01 2.49823984e-02
-1.00950444e+00 -1.47426113e-01 -4.42483068e-01 3.44539791e-01
7.48594999e-01 -2.44198605e-01 -1.00960672e+00 -3.32655683e-02
1.64548140e-02 -4.34191972e-01 9.25728440e-01 3.77368718e-01
1.58181155e+00 -1.27915274e-02 -2.58077890e-01 9.51481104e-01
1.63977897e+00 4.67133671e-02 3.56066376e-01 4.65501428e-01
1.06284058e+00 4.07022595e-01 8.36302161e-01 7.46967614e-01
5.28806388e-01 8.09051096e-01 3.32776546e-01 -7.39168748e-02
3.27444561e-02 -3.01650941e-01 1.80139720e-01 9.43765223e-01
3.16874012e-02 1.12029478e-01 -8.17784250e-01 2.24051088e-01
-1.65239751e+00 -7.26364374e-01 1.20742753e-01 2.11771512e+00
7.33498693e-01 -1.05231211e-01 -1.28116995e-01 1.26963705e-01
5.44473171e-01 4.95615691e-01 -7.09574282e-01 2.80212820e-01
-5.19793808e-01 1.22890860e-01 4.48155969e-01 3.25146914e-01
-1.35512388e+00 6.54851437e-01 5.22860575e+00 1.31042111e+00
-1.19368863e+00 -1.17550261e-01 6.02945089e-01 -3.54457311e-02
-4.86382097e-01 -1.14432842e-01 -7.64036298e-01 3.97686511e-01
1.39909118e-01 2.64560670e-01 2.95380324e-01 8.84388864e-01
2.04695076e-01 -2.01797187e-01 -7.63556421e-01 1.56327426e+00
1.25059277e-01 -1.11555266e+00 1.11728208e-02 1.08186930e-01
8.15970540e-01 -3.15038003e-02 6.69130459e-02 -1.13248900e-01
-1.88116193e-01 -7.69051433e-01 6.51543915e-01 8.40068340e-01
7.09821343e-01 -7.91364908e-01 3.65015507e-01 2.75743127e-01
-1.61525965e+00 -2.10298926e-01 -6.45358682e-01 1.56957000e-01
-1.25205994e-01 9.54166234e-01 -9.82128903e-02 6.89709127e-01
8.82804513e-01 1.22924268e+00 -7.60875762e-01 1.13900542e+00
-7.80223459e-02 6.52109236e-02 -5.58388591e-01 -1.79647014e-01
1.92798346e-01 -3.08095902e-01 2.53613412e-01 1.01345408e+00
4.07216161e-01 2.47923046e-01 5.08777797e-01 6.43949270e-01
1.24469213e-01 2.84160674e-01 -5.91268659e-01 1.96879461e-01
4.63881761e-01 1.23391604e+00 -7.72706926e-01 1.05070449e-01
-3.90757352e-01 1.31608057e+00 2.16569901e-01 3.85998756e-01
-4.23799008e-01 -3.85246754e-01 7.18710840e-01 -2.21323162e-01
3.37542027e-01 -7.09311962e-01 -1.44958183e-01 -1.34054840e+00
1.74565181e-01 -5.56612849e-01 3.40929627e-01 -6.24412656e-01
-1.48864555e+00 4.31873620e-01 -1.75920978e-01 -1.59799111e+00
2.91589797e-01 -7.31773257e-01 -2.66792178e-01 7.22431660e-01
-1.82845676e+00 -1.33764458e+00 -8.21656346e-01 1.18079340e+00
3.84673297e-01 2.18677759e-01 6.00948155e-01 3.38196635e-01
-5.78123927e-01 3.27596873e-01 2.17593238e-01 -4.41367365e-02
4.31387395e-01 -1.08773708e+00 -4.66552436e-01 6.69083118e-01
7.44017363e-02 4.78849590e-01 1.22750573e-01 -2.33597308e-01
-1.84986234e+00 -1.25127602e+00 4.84774888e-01 -2.51761675e-01
2.45439425e-01 -4.09536570e-01 -4.66941863e-01 1.11593232e-01
-5.51378012e-01 5.20521224e-01 8.00687373e-01 -8.16262811e-02
-2.17210442e-01 -5.87603688e-01 -1.10848558e+00 2.54814118e-01
1.23884368e+00 -1.05502295e+00 -2.79660523e-01 4.22592819e-01
5.35044611e-01 -3.70249420e-01 -1.04506052e+00 5.57907224e-01
3.37953895e-01 -1.00135374e+00 1.25865173e+00 2.75877845e-02
3.26521844e-01 -7.12313950e-01 -9.22738194e-01 -8.95949602e-01
-2.41698310e-01 -2.10277233e-02 7.07545206e-02 1.38626754e+00
-1.38083417e-02 -4.88486379e-01 5.47831178e-01 4.42370713e-01
-1.56208292e-01 -6.97465420e-01 -1.20306420e+00 -5.32476664e-01
-4.21084195e-01 -6.46984994e-01 5.92165232e-01 9.16309237e-01
-3.88434976e-01 1.09712947e-02 -1.51160941e-01 2.21873030e-01
7.68661976e-01 8.53915095e-01 8.05737495e-01 -9.92020369e-01
-9.93454307e-02 -5.28936505e-01 -8.07334721e-01 -1.60268354e+00
9.87656862e-02 -1.01029444e+00 2.56995678e-01 -1.55405760e+00
3.48205924e-01 -9.98413086e-01 -7.35567391e-01 4.42914277e-01
-1.84104785e-01 3.52969110e-01 2.09449232e-02 4.74206388e-01
-8.22473645e-01 1.08281052e+00 1.60110569e+00 -2.83737212e-01
-3.40477765e-01 1.24648465e-02 -3.74815404e-01 3.54289144e-01
6.00503743e-01 -2.58296609e-01 -2.31279716e-01 -2.80949265e-01
1.44004291e-02 -1.69163480e-01 4.84297812e-01 -9.37131643e-01
2.79596984e-01 -7.23382294e-01 8.07567537e-01 -7.63590991e-01
5.50679922e-01 -8.85574758e-01 -1.37173072e-01 3.25774476e-02
-1.32010534e-01 -4.45452124e-01 6.99012205e-02 5.22096038e-01
-4.95133162e-01 2.26234287e-01 7.14346170e-01 -1.72482654e-01
-1.17218769e+00 7.66600847e-01 -4.64210808e-02 -3.36357832e-01
1.01009440e+00 -4.61670667e-01 -7.42454976e-02 -7.78609235e-03
-4.17703897e-01 3.45093198e-02 6.42002106e-01 3.93604428e-01
1.02474439e+00 -1.69136214e+00 -1.94055349e-01 4.87121254e-01
6.43013775e-01 3.20921987e-01 6.57927930e-01 8.03009152e-01
-3.96006614e-01 2.93732107e-01 -1.85917854e-01 -1.07351351e+00
-9.80428815e-01 3.11278492e-01 4.29365188e-01 2.39179641e-01
-5.07931232e-01 1.06527424e+00 1.34494931e-01 -6.78821623e-01
-8.63006040e-02 -4.62341487e-01 -8.04820359e-02 4.64777574e-02
3.30922008e-01 1.45816773e-01 6.50425255e-02 -1.04298115e+00
-8.53150308e-01 1.27384150e+00 4.96502340e-01 2.96332449e-01
1.44745076e+00 -6.56108141e-01 -3.04436743e-01 7.86197364e-01
1.68340266e+00 8.07294808e-03 -1.29850054e+00 -5.73245943e-01
-1.56653851e-01 -1.04291177e+00 4.95808542e-01 -3.59174520e-01
-1.16798902e+00 9.46511984e-01 8.43270659e-01 2.82464288e-02
1.65310967e+00 3.69110048e-01 5.11514068e-01 2.63082296e-01
7.55008042e-01 -8.90229940e-01 3.17161024e-01 2.81544238e-01
6.85883939e-01 -1.45554554e+00 2.38045692e-01 -3.30543786e-01
-3.91601264e-01 1.18159807e+00 4.35284942e-01 -5.27893342e-02
9.20580804e-01 -2.73354471e-01 -1.54959988e-02 -4.16061342e-01
-2.85097510e-02 -5.36996901e-01 6.81518972e-01 3.36376667e-01
5.10311089e-02 1.47877827e-01 2.96632737e-01 4.66907203e-01
1.79474860e-01 -4.02302265e-01 -2.29099482e-01 1.01148665e+00
-6.36967242e-01 -1.06030667e+00 -3.39878857e-01 4.95026916e-01
-1.02303959e-02 1.90828562e-01 -1.40455902e-01 5.11320472e-01
4.63909924e-01 1.01454425e+00 -1.76511094e-01 -6.35436535e-01
3.48637462e-01 -3.31858337e-01 7.00461864e-01 -4.88975585e-01
6.85350001e-02 1.67622671e-01 -3.92828196e-01 -9.85232592e-01
-9.57858264e-01 -8.30273092e-01 -1.07418799e+00 -5.03702238e-02
-5.03376544e-01 -2.83518046e-01 7.15334237e-01 1.03285289e+00
2.79903352e-01 3.24173450e-01 1.24938476e+00 -1.20745718e+00
3.50716896e-02 -5.19352496e-01 -9.42055821e-01 3.57489198e-01
6.32460833e-01 -8.89855266e-01 -4.33449060e-01 -2.19161019e-01] | [9.574338912963867, -0.9518435001373291] |
5052bf52-4796-499d-9d38-39fa4c3d7ccb | learning-based-quality-control-for-cardiac-mr | 1803.09354 | null | http://arxiv.org/abs/1803.09354v2 | http://arxiv.org/pdf/1803.09354v2.pdf | Learning-Based Quality Control for Cardiac MR Images | The effectiveness of a cardiovascular magnetic resonance (CMR) scan depends
on the ability of the operator to correctly tune the acquisition parameters to
the subject being scanned and on the potential occurrence of imaging artefacts
such as cardiac and respiratory motion. In the clinical practice, a quality
control step is performed by visual assessment of the acquired images: however,
this procedure is strongly operator-dependent, cumbersome and sometimes
incompatible with the time constraints in clinical settings and large-scale
studies. We propose a fast, fully-automated, learning-based quality control
pipeline for CMR images, specifically for short-axis image stacks. Our pipeline
performs three important quality checks: 1) heart coverage estimation, 2)
inter-slice motion detection, 3) image contrast estimation in the cardiac
region. The pipeline uses a hybrid decision forest method - integrating both
regression and structured classification models - to extract landmarks as well
as probabilistic segmentation maps from both long- and short-axis images as a
basis to perform the quality checks. The technique was tested on up to 3000
cases from the UK Biobank as well as on 100 cases from the UK Digital Heart
Project, and validated against manual annotations and visual inspections
performed by expert interpreters. The results show the capability of the
proposed pipeline to correctly detect incomplete or corrupted scans (e.g. on UK
Biobank, sensitivity and specificity respectively 88% and 99% for heart
coverage estimation, 85% and 95% for motion detection), allowing their
exclusion from the analysed dataset or the triggering of a new acquisition. | ["Declan P. O'Regan", 'Jonathan Passerat-Palmbach', 'Wenjia Bai', 'Stuart Cook', 'Hideaki Suzuki', 'Giacomo Tarroni', 'Daniel Rueckert', 'Antonio de Marvao', 'Paul M. Matthews', 'Ozan Oktay', 'Andreas Schuh', 'Ben Glocker'] | 2018-03-25 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 5.50045609e-01 1.08080372e-01 2.93662399e-01 -3.67123395e-01
-9.61743414e-01 -8.95434916e-01 3.09004128e-01 7.95193493e-01
-6.69213057e-01 7.01685965e-01 -6.97373599e-02 -4.82003719e-01
-4.56535637e-01 -4.55944955e-01 -2.82472402e-01 -6.77552879e-01
-3.07659090e-01 9.86369491e-01 7.52678752e-01 6.45644665e-01
2.77552217e-01 7.70529807e-01 -8.50088596e-01 1.66105226e-01
6.52314365e-01 7.56110370e-01 4.93418247e-01 1.08563960e+00
2.64366031e-01 7.92494535e-01 -3.40566874e-01 -6.99771866e-02
1.32057339e-01 -5.45845211e-01 -7.11046517e-01 2.49858677e-01
-2.21791640e-02 -2.34237120e-01 4.39189017e-01 6.13009453e-01
8.64166677e-01 -2.03318447e-01 6.09307587e-01 -5.86410999e-01
3.45824808e-01 4.83914524e-01 -3.41063648e-01 6.10579848e-01
-4.19092970e-03 4.70870376e-01 3.54411691e-01 -5.65646410e-01
7.28279591e-01 3.36471915e-01 7.94593930e-01 -1.15728252e-01
-1.25838470e+00 -3.45037013e-01 -7.14924455e-01 3.01514864e-02
-1.10029638e+00 -4.27709281e-01 3.45841557e-01 -9.74823773e-01
6.44927680e-01 2.43200272e-01 7.04631984e-01 3.08239192e-01
3.50264817e-01 -6.45493269e-02 1.43956602e+00 -5.74727535e-01
4.35064316e-01 9.12113711e-02 -1.24553502e-01 4.61623251e-01
2.50511467e-01 1.43716350e-01 -3.83113921e-02 -2.85195202e-01
9.72390234e-01 -4.98096377e-01 -3.62437338e-01 -7.03556061e-01
-1.50783896e+00 6.13434076e-01 6.63048625e-02 5.68675697e-01
-7.05933928e-01 -2.22637072e-01 7.87550449e-01 -7.17513263e-02
-5.51265404e-02 2.05806881e-01 -2.81128347e-01 -1.61174208e-01
-1.32910013e+00 -2.36119822e-01 3.87977809e-01 3.11857730e-01
1.77936807e-01 -1.96530208e-01 -3.75369370e-01 7.02210307e-01
4.70642000e-01 3.72613847e-01 6.60860538e-01 -7.71577656e-01
9.82100740e-02 4.37810093e-01 7.62643013e-03 -8.17297220e-01
-7.29577899e-01 -5.50922275e-01 -8.96581173e-01 3.16839099e-01
5.99378586e-01 3.37459525e-04 -9.65545237e-01 1.22232938e+00
6.34415030e-01 2.89128963e-02 -2.33839035e-01 1.16311800e+00
4.34034944e-01 9.66840014e-02 3.59597772e-01 -5.78816056e-01
1.48914409e+00 -4.00539964e-01 -5.29110849e-01 8.52096751e-02
7.73093581e-01 -9.05537248e-01 6.23405993e-01 5.19838393e-01
-1.14565539e+00 -5.19881606e-01 -1.10551226e+00 7.20043957e-01
3.76860112e-01 4.29833412e-01 1.47445560e-01 9.56255555e-01
-8.15562308e-01 6.75815105e-01 -1.16760409e+00 -9.47151780e-02
4.47820157e-01 4.18394774e-01 -5.49108803e-01 -1.87334836e-01
-8.77472103e-01 1.10113502e+00 5.47457933e-01 3.84331316e-01
-6.63031757e-01 -7.99380839e-01 -7.50405848e-01 -3.04221243e-01
3.15646559e-01 -6.20114684e-01 8.07358921e-01 -8.68986309e-01
-1.34238565e+00 1.23817158e+00 2.61227131e-01 -5.80098093e-01
1.20995772e+00 1.15007952e-01 -2.28238001e-01 5.98533750e-01
2.23247170e-01 1.73635677e-01 7.39232004e-01 -8.15360725e-01
-2.76878655e-01 -5.82656562e-01 -5.66970706e-01 -4.05393429e-02
6.37794077e-01 1.95539534e-01 -2.39110500e-01 -6.07386291e-01
3.67785364e-01 -1.00543118e+00 -2.55813360e-01 -1.11643441e-01
-2.34501064e-01 5.64821780e-01 4.57519323e-01 -1.33268917e+00
1.11070764e+00 -1.96079278e+00 -2.32118323e-01 4.70750511e-01
2.82355279e-01 4.18909937e-01 5.53349316e-01 -2.48189047e-01
-3.82305026e-01 -9.55378264e-02 -6.73890054e-01 2.81951755e-01
-3.76406699e-01 -1.39301971e-01 3.77688974e-01 8.82877350e-01
3.71266492e-02 5.96462786e-01 -8.58340621e-01 -7.79704332e-01
7.84401834e-01 5.48563182e-01 -1.46495089e-01 3.85258794e-01
4.64139342e-01 1.11148262e+00 -1.36995062e-01 2.41430640e-01
4.93642956e-01 -5.20457290e-02 6.96761787e-01 -2.57869333e-01
-1.55860454e-01 1.31889492e-01 -1.47423935e+00 1.58550620e+00
-4.59720612e-01 2.58971214e-01 1.89165741e-01 -9.40185130e-01
8.47594440e-01 7.46811032e-01 5.30318916e-01 -5.68322241e-01
1.29423171e-01 5.30673802e-01 4.36323375e-01 -6.34783685e-01
-2.18168259e-01 -5.45927644e-01 3.34048957e-01 4.90169019e-01
-8.24277028e-02 -2.00426385e-01 6.19590655e-02 -2.75912415e-02
1.03918254e+00 2.76084691e-01 5.28353453e-01 -4.19898599e-01
8.30416203e-01 -3.19253393e-02 5.72833419e-01 6.28236890e-01
-5.46831310e-01 9.97378290e-01 5.21554947e-01 -4.83676821e-01
-1.14880168e+00 -8.30330610e-01 -5.29199898e-01 2.14991748e-01
-4.42552358e-01 9.60864127e-02 -6.88236296e-01 -7.04962194e-01
-4.33577687e-01 4.44152445e-01 -4.04552907e-01 1.08956888e-01
-8.09120178e-01 -9.38666105e-01 2.34444320e-01 2.94275880e-01
6.55809939e-02 -1.15245700e+00 -1.66042984e+00 6.24869704e-01
-1.87927797e-01 -1.08402777e+00 -6.21442031e-03 2.34757170e-01
-1.20774519e+00 -1.32222795e+00 -8.85603249e-01 -1.79951519e-01
3.81681055e-01 -5.07479608e-01 1.26702595e+00 1.04264155e-01
-8.59130442e-01 1.36059180e-01 -2.22225383e-01 -8.55201930e-02
-8.25648785e-01 -2.30260491e-01 -4.45723534e-01 -1.57207817e-01
-3.89362246e-01 -4.75222677e-01 -7.96188474e-01 4.22709316e-01
-8.46164286e-01 -7.39079043e-02 8.07331264e-01 6.95876360e-01
7.55671978e-01 -1.53442606e-01 4.62734312e-01 -1.03992260e+00
1.33806929e-01 -1.43389702e-01 -8.95599842e-01 3.38261425e-01
-5.03497183e-01 -1.91172004e-01 1.45527214e-01 -4.02374705e-03
-5.82314670e-01 3.76430601e-01 -1.80734053e-01 -1.71049044e-01
-5.56816995e-01 4.04928833e-01 9.26102847e-02 -7.42499679e-02
6.45484507e-01 6.54854625e-02 6.19889200e-02 -1.45908281e-01
-2.40441244e-02 3.61156285e-01 6.18255019e-01 -2.45820165e-01
4.29044783e-01 3.10448259e-01 3.99195462e-01 -7.49187589e-01
-7.25849345e-02 -5.31252086e-01 -1.25611174e+00 -4.69688773e-01
1.07022011e+00 -3.54509264e-01 -3.45259994e-01 2.09751502e-01
-9.24018383e-01 -1.83326438e-01 -2.62258470e-01 7.52749860e-01
-4.26291913e-01 7.07082927e-01 -3.49941164e-01 -8.69889915e-01
-7.20758617e-01 -1.55776203e+00 7.57691026e-01 -1.16824023e-01
-3.64335150e-01 -9.89479244e-01 1.37192890e-01 5.00894427e-01
5.47028720e-01 7.91365802e-01 1.03745425e+00 -6.69029891e-01
-4.46165293e-01 -3.29078525e-01 -8.04182142e-02 3.09426874e-01
-1.51615711e-02 4.39600535e-02 -8.14373612e-01 1.52170332e-02
1.40669078e-01 6.18701288e-03 4.27042574e-01 9.54210401e-01
6.01305306e-01 3.52209151e-01 6.97398931e-02 2.76029021e-01
1.45336795e+00 3.06093216e-01 6.75010204e-01 2.86199450e-01
3.27247173e-01 6.11792803e-01 5.07294059e-01 3.53008777e-01
-3.31474021e-02 6.96149707e-01 3.48967463e-01 -9.12598372e-02
-1.49505243e-01 3.97406161e-01 -1.80568174e-01 4.11855131e-01
-2.68943697e-01 2.70632982e-01 -1.38947415e+00 6.53521240e-01
-1.41758502e+00 -5.02494991e-01 -7.83352435e-01 2.79252625e+00
5.57619810e-01 4.91852999e-01 3.61254156e-01 5.24552464e-01
9.99234140e-01 -2.37826645e-01 -1.17101446e-01 -1.74354672e-01
1.61726847e-01 4.17282760e-01 5.87896824e-01 4.72635567e-01
-1.11204052e+00 1.36243939e-01 5.83183908e+00 1.73412859e-01
-1.15363872e+00 4.55873549e-01 9.30980861e-01 1.23042256e-01
3.06295544e-01 3.93658131e-02 -1.76824763e-01 4.03876394e-01
1.09487474e+00 4.96271044e-01 -2.75768787e-02 4.74510580e-01
4.96721715e-01 -6.55070424e-01 -6.87205195e-01 7.77269185e-01
-1.54944122e-01 -1.28552687e+00 -6.02287710e-01 -1.10498942e-01
1.85088128e-01 4.35739197e-02 -5.68385899e-01 -2.84779400e-01
-3.66821647e-01 -9.39038515e-01 7.34858930e-01 5.26609123e-01
1.08142030e+00 -5.63970923e-01 1.01049936e+00 3.23774248e-01
-8.29764724e-01 2.13701606e-01 1.41390771e-01 4.39889103e-01
5.04135728e-01 1.02158594e+00 -1.11947775e+00 6.88597202e-01
4.49941576e-01 1.31790116e-01 -5.41268528e-01 1.29611683e+00
-2.46712938e-01 6.53358638e-01 -3.34448725e-01 6.56895638e-01
4.20712382e-02 -4.59381975e-02 5.84937871e-01 1.28571582e+00
1.40597016e-01 1.45709291e-01 -8.93130749e-02 7.49898434e-01
5.66633642e-01 2.86771119e-01 -2.00941716e-03 2.50714809e-01
1.24329627e-01 1.60427821e+00 -1.43976390e+00 -3.80509883e-01
-6.50142580e-02 5.44757783e-01 -8.47320110e-02 -1.54553816e-01
-6.91367090e-01 -1.40305057e-01 -5.38303852e-01 7.40630686e-01
2.69219965e-01 5.17755859e-02 -3.81323159e-01 -6.70084715e-01
1.60255805e-01 -7.66128659e-01 6.26684487e-01 -6.43601596e-01
-6.75035357e-01 7.53911316e-01 6.11964911e-02 -8.48457754e-01
-5.11553943e-01 -4.02526945e-01 -5.58765948e-01 1.27107680e+00
-1.31006980e+00 -7.08776951e-01 -2.06569090e-01 4.94393036e-02
6.60805106e-02 1.30305171e-01 7.27744460e-01 4.61988240e-01
-2.10858852e-01 5.07947765e-02 -3.85440022e-01 1.14725277e-01
5.60214996e-01 -1.38197482e+00 -2.31758669e-01 8.04744244e-01
-2.43888348e-01 2.42887065e-01 5.87681949e-01 -6.70395434e-01
-7.20906436e-01 -6.91078663e-01 8.42126071e-01 -2.38418505e-01
3.38831961e-01 1.62431359e-01 -9.60237026e-01 2.50688761e-01
-8.76893923e-02 5.08615911e-01 7.85254359e-01 -5.04454970e-01
2.69475818e-01 1.44155681e-01 -1.41151297e+00 -3.09800953e-01
2.41600990e-01 -6.80530295e-02 -6.01531982e-01 1.04345225e-01
-2.28315800e-01 -6.46191120e-01 -1.30697775e+00 5.35177529e-01
7.52553165e-01 -1.34725630e+00 7.99744368e-01 -1.39794394e-01
1.65182441e-01 -4.33032602e-01 3.44168872e-01 -6.54085517e-01
-1.83816895e-01 -4.03712034e-01 4.11227167e-01 9.87450063e-01
2.52100021e-01 -4.38949734e-01 5.62679708e-01 6.55414522e-01
3.60300429e-02 -4.69859779e-01 -1.25954330e+00 -1.03010632e-01
-2.11887360e-01 -5.32430530e-01 -8.84995088e-02 9.40688610e-01
-3.83382857e-01 9.69395861e-02 7.70537704e-02 2.97393233e-01
7.28760064e-01 -1.34001836e-01 3.37571800e-01 -1.28976476e+00
-4.95537192e-01 -4.02072787e-01 -6.49772525e-01 9.07479227e-02
-3.86989295e-01 -7.03497410e-01 -7.83953071e-02 -1.48243296e+00
2.64330536e-01 -7.14665055e-01 -3.79428446e-01 5.97661873e-03
-2.18966380e-01 3.95556152e-01 2.92192828e-02 3.22629780e-01
-1.34578183e-01 -2.84614623e-01 1.02019918e+00 3.87121052e-01
-1.39785901e-01 2.63457835e-01 6.18596468e-03 6.51450098e-01
4.22082543e-01 -7.77091265e-01 -2.78924227e-01 6.69537187e-02
3.38596404e-01 6.93462551e-01 7.62578607e-01 -1.19188726e+00
-1.51555568e-01 3.04914385e-01 6.18351877e-01 -6.76703811e-01
-3.24170351e-01 -9.05435741e-01 6.24491930e-01 9.63569164e-01
-2.06985906e-01 1.39763966e-01 1.35520086e-01 2.09440008e-01
-7.03744143e-02 -6.34359241e-01 1.39310539e+00 -5.12078762e-01
-1.21721193e-01 -5.23852929e-02 -3.74151796e-01 1.18388504e-01
1.02178478e+00 -3.17411989e-01 3.05242747e-01 -1.07408471e-01
-1.42216158e+00 -1.16033375e-01 2.16365740e-01 -2.00762302e-01
2.94770151e-01 -8.19191277e-01 -9.57396150e-01 -1.88729316e-02
-1.17688932e-01 8.80123600e-02 7.56563783e-01 1.69809163e+00
-1.11342847e+00 3.28260809e-01 -2.00977832e-01 -1.05509424e+00
-1.22179604e+00 4.27916080e-01 7.88486779e-01 -7.98808873e-01
-8.49235713e-01 3.65793914e-01 -8.02586302e-02 -1.75055400e-01
-1.60492375e-01 -3.99336755e-01 -2.76977599e-01 5.11147901e-02
4.35850203e-01 2.71636873e-01 5.70394635e-01 -9.30437386e-01
-6.32219255e-01 6.01237178e-01 2.82619864e-01 -2.22149178e-01
1.17708385e+00 -2.17221111e-01 6.40568510e-02 4.60368395e-01
6.90636575e-01 -1.88556239e-01 -1.03559363e+00 3.00714523e-02
1.46548197e-01 -3.18419814e-01 4.24025685e-01 -1.12589502e+00
-9.41367507e-01 1.10981512e+00 1.09011400e+00 7.76236281e-02
1.07479441e+00 -4.26414579e-01 2.21312895e-01 -6.64757609e-01
9.01030973e-02 -8.11146915e-01 -4.70019996e-01 -1.46475211e-01
6.45644963e-01 -1.14243662e+00 3.65373552e-01 -3.68732065e-01
-8.18411171e-01 1.30058753e+00 -1.12812065e-01 1.48646832e-01
4.75365698e-01 2.03769088e-01 2.01300576e-01 -2.08892211e-01
-3.39240700e-01 1.83802247e-01 4.06613588e-01 6.41021311e-01
7.00661719e-01 7.50648156e-02 -4.85008001e-01 2.20563814e-01
3.54085386e-01 2.30805442e-01 3.97489011e-01 1.04272401e+00
-2.73998022e-01 -7.72283733e-01 -6.21741354e-01 3.63648117e-01
-8.46118629e-01 3.49413641e-02 1.13401048e-01 7.09045768e-01
1.88084528e-01 5.22122145e-01 -2.75240153e-01 3.68763864e-01
2.91449964e-01 3.29485446e-01 5.21644711e-01 -5.44828713e-01
-1.05987549e+00 4.98930216e-01 7.48782307e-02 -2.91284800e-01
-6.36752486e-01 -9.74841893e-01 -1.27683151e+00 3.89408976e-01
-3.50917250e-01 2.74924226e-02 9.61252630e-01 1.10750926e+00
1.00291476e-01 7.47531950e-01 3.59746605e-01 -6.29224539e-01
-2.69804567e-01 -1.07668304e+00 -5.02064168e-01 3.18838626e-01
1.34710342e-01 -5.53778470e-01 -2.57833034e-01 3.43534589e-01] | [14.076196670532227, -2.521127700805664] |
8375fc97-294c-4b4b-9bee-aa4f9fc35370 | performance-optimization-using-multimodal | 2304.12568 | null | https://arxiv.org/abs/2304.12568v2 | https://arxiv.org/pdf/2304.12568v2.pdf | Performance Optimization using Multimodal Modeling and Heterogeneous GNN | Growing heterogeneity and configurability in HPC architectures has made auto-tuning applications and runtime parameters on these systems very complex. Users are presented with a multitude of options to configure parameters. In addition to application specific solutions, a common approach is to use general purpose search strategies, which often might not identify the best configurations or their time to convergence is a significant barrier. There is, thus, a need for a general purpose and efficient tuning approach that can be easily scaled and adapted to various tuning tasks. We propose a technique for tuning parallel code regions that is general enough to be adapted to multiple tasks. In this paper, we analyze IR-based programming models to make task-specific performance optimizations. To this end, we propose the Multimodal Graph Neural Network and Autoencoder (MGA) tuner, a multimodal deep learning based approach that adapts Heterogeneous Graph Neural Networks and Denoizing Autoencoders for modeling IR-based code representations that serve as separate modalities. This approach is used as part of our pipeline to model a syntax, semantics, and structure-aware IR-based code representation for tuning parallel code regions/kernels. We extensively experiment on OpenMP and OpenCL code regions/kernels obtained from PolyBench, Rodinia, STREAM, DataRaceBench, AMD SDK, NPB, NVIDIA SDK, Parboil, SHOC, and LULESH benchmarks. We apply our multimodal learning techniques to the tasks of i) optimizing the number of threads, scheduling policy and chunk size in OpenMP loops and, ii) identifying the best device for heterogeneous device mapping of OpenCL kernels. Our experiments show that this multimodal learning based approach outperforms the state-of-the-art in all experiments. | ['Ali Jannesari', 'Eduardo Cesar', 'Anna Sikora', 'Ali TehraniJamsaz', 'Jordi Alcaraz', 'Akash Dutta'] | 2023-04-25 | null | null | null | null | ['multimodal-deep-learning'] | ['natural-language-processing'] | [-7.02497423e-01 -8.18034470e-01 -2.22778141e-01 -3.54251444e-01
-5.16222119e-01 -6.24925494e-01 2.25785226e-01 3.61168653e-01
-1.77196845e-01 3.26022714e-01 7.84620643e-02 -5.32615781e-01
-2.32579887e-01 -8.58945608e-01 -6.79940999e-01 -7.51016855e-01
-1.04775667e-01 7.50310481e-01 1.08558536e-01 -4.51959461e-01
3.26334953e-01 6.46301210e-01 -1.83534408e+00 6.73408449e-01
5.34553587e-01 7.81729460e-01 3.78359407e-01 9.60833609e-01
-4.35523897e-01 1.02050543e+00 -4.08182353e-01 2.22069606e-01
3.55996229e-02 -1.75913438e-01 -8.91925633e-01 -5.50275147e-01
9.09426808e-02 1.35565251e-01 9.36328694e-02 8.42527986e-01
7.55924761e-01 2.34439254e-01 6.31327331e-01 -9.77114797e-01
-2.24543363e-01 1.06993318e+00 -6.95216477e-01 2.79242396e-01
-2.03554899e-01 2.14990169e-01 6.99796438e-01 -5.46165407e-01
3.77320170e-01 1.12946987e+00 6.15483463e-01 1.38653502e-01
-1.27668321e+00 -3.68065000e-01 -2.27203459e-01 6.70802668e-02
-1.60509264e+00 -2.35099658e-01 5.69711924e-01 -8.73409271e-01
1.60080111e+00 2.58529276e-01 4.63610023e-01 8.90507400e-01
6.55364931e-01 1.21290617e-01 7.83170760e-01 -6.12032235e-01
4.96006697e-01 3.48731160e-01 3.92283261e-01 5.87321401e-01
-7.69950151e-02 2.39089787e-01 -4.16853726e-01 -4.57400203e-01
6.70984447e-01 -1.35356158e-01 -4.93132956e-02 -5.58822215e-01
-1.30354273e+00 1.00891221e+00 5.75388253e-01 3.79269272e-01
-2.71368653e-01 3.43564481e-01 1.07441175e+00 4.08679426e-01
2.97020495e-01 7.74598956e-01 -8.13971519e-01 -3.32548022e-01
-8.79787564e-01 3.43453854e-01 1.11015511e+00 7.34382868e-01
1.00514209e+00 2.89211720e-01 -4.82783735e-01 1.03610051e+00
1.61857620e-01 3.76519054e-01 7.61669099e-01 -6.86126113e-01
3.10233403e-02 6.38478696e-01 -4.92096037e-01 -8.09006870e-01
-7.55002260e-01 -4.29477215e-01 -8.33820939e-01 5.24465203e-01
-1.46768555e-01 -1.80905297e-01 -5.62886596e-01 1.31380093e+00
3.70194554e-01 -8.86853188e-02 2.91980933e-02 9.10332143e-01
8.85375559e-01 8.75144660e-01 2.44433314e-01 3.22831869e-01
1.30676007e+00 -1.22907329e+00 -1.11140609e-01 2.03738615e-01
1.12548399e+00 -1.06057000e+00 1.19816291e+00 2.62521118e-01
-6.53917372e-01 -9.75744009e-01 -1.03991795e+00 1.00799929e-03
-5.99933028e-01 3.31899226e-01 7.70380378e-01 5.14367163e-01
-1.15033841e+00 6.09641075e-01 -1.01450419e+00 -3.71938527e-01
-2.83890396e-01 6.88518524e-01 1.29636571e-01 3.79477799e-01
-6.94338858e-01 6.00955606e-01 9.71016228e-01 -4.73158300e-01
-7.39114583e-01 -9.59589958e-01 -6.36766553e-01 3.95035148e-01
8.11588615e-02 -9.73964274e-01 9.57349658e-01 -1.09628880e+00
-1.73382390e+00 7.48359203e-01 3.45419317e-01 -3.52929652e-01
-9.27020423e-03 1.27990261e-01 -4.56660986e-01 -5.20800710e-01
-2.66036510e-01 5.99953771e-01 8.98851514e-01 -1.16440785e+00
-2.24090353e-01 -1.80757210e-01 -1.92826569e-01 1.35430723e-01
-2.99870133e-01 1.94075793e-01 -3.86811107e-01 -3.87341887e-01
-6.48480892e-01 -8.41058552e-01 -9.29589272e-02 -7.86017776e-01
-3.61947715e-01 4.92562987e-02 8.00841212e-01 -2.50701874e-01
1.31232202e+00 -2.20596075e+00 5.70383430e-01 4.25892472e-01
2.08014250e-01 5.84928989e-02 -1.45285055e-01 3.98362100e-01
-2.53569275e-01 -6.05284311e-02 1.53177276e-01 5.58389388e-02
8.87606069e-02 1.54974908e-01 -3.16552758e-01 3.93618941e-02
-1.66604847e-01 5.90304136e-01 -4.26775932e-01 -4.65816915e-01
4.70568448e-01 6.08604908e-01 -7.96018839e-01 5.05465925e-01
-4.82325912e-01 4.84406620e-01 -4.10040796e-01 4.96623129e-01
6.43653512e-01 -5.58855712e-01 1.25614703e-01 -2.41602376e-01
-4.81385499e-01 -1.09296024e-01 -1.12342584e+00 1.69789553e+00
-9.86919343e-01 4.13085103e-01 -3.36643718e-02 -9.81911778e-01
1.09866047e+00 -1.11370958e-01 3.67170513e-01 -5.57455599e-01
3.25759828e-01 2.10220382e-01 -2.83759292e-02 -6.12740934e-01
7.02821314e-01 2.54273385e-01 -2.19884440e-01 5.69744408e-01
6.05109483e-02 -6.48896396e-02 5.84583282e-02 -1.15444966e-01
1.16517615e+00 2.38320917e-01 2.84777611e-01 -7.97810256e-01
6.86410904e-01 3.63920897e-01 3.74999009e-02 7.30646610e-01
3.80699456e-01 2.72942871e-01 6.09225452e-01 -9.12072480e-01
-1.23915601e+00 -7.55713642e-01 -1.93751231e-01 1.96962404e+00
-3.01057845e-01 -6.02872133e-01 -8.96739781e-01 -1.94560900e-01
-5.37467971e-02 6.12912118e-01 -5.86847305e-01 -3.13754031e-03
-6.62266433e-01 -9.14161086e-01 4.12434489e-01 3.97184789e-01
3.09173197e-01 -1.18027699e+00 -8.60037744e-01 1.96803510e-01
4.58963811e-01 -7.23869920e-01 -2.73803264e-01 6.36715114e-01
-7.56119370e-01 -7.17485905e-01 -2.97814578e-01 -6.44614875e-01
1.17359206e-01 -6.71934038e-02 1.84064674e+00 2.68523455e-01
-3.70448470e-01 1.12768009e-01 -2.81336516e-01 -1.26554310e-01
-7.72347271e-01 8.05554807e-01 -1.33866727e-01 -2.82646149e-01
1.88052714e-01 -5.55571020e-01 -3.31144482e-01 2.16782406e-01
-8.75781119e-01 2.35726520e-01 4.23916340e-01 9.80526090e-01
8.29764366e-01 -9.05383304e-02 5.25091924e-02 -1.01073468e+00
7.56008983e-01 -8.47673655e-01 -1.04374015e+00 3.50756526e-01
-6.64917946e-01 5.80024123e-01 1.07191622e+00 -5.26509106e-01
-8.65565717e-01 1.03073142e-01 -3.30533773e-01 -7.70389438e-01
-3.08301151e-01 7.22112060e-01 3.98061350e-02 -4.70387220e-01
1.23290145e+00 -1.19140998e-01 -5.31644560e-02 -3.86280119e-01
3.53460908e-01 7.51260757e-01 3.12280655e-01 -1.26923525e+00
2.59379834e-01 2.36760788e-02 1.59393605e-02 -8.83278012e-01
-3.39233041e-01 -4.43967998e-01 -5.11053085e-01 -5.43975942e-02
9.45579231e-01 -9.72879410e-01 -8.28006089e-01 3.15036833e-01
-1.03075683e+00 -7.81555712e-01 -6.97245896e-02 3.01076770e-01
-6.62631273e-01 -5.29566742e-02 -6.16776228e-01 -1.42048880e-01
-6.56861961e-01 -1.74670660e+00 1.18201077e+00 3.08749616e-01
-2.38496944e-01 -1.21739614e+00 5.38185179e-01 9.05709900e-03
1.11681807e+00 1.89990208e-01 1.35583198e+00 -6.87222183e-01
-5.70339143e-01 2.37659171e-01 -3.70315820e-01 -2.62310845e-03
-3.33574563e-01 3.22527200e-01 -8.64941120e-01 -4.01081055e-01
-2.41599977e-01 -3.81325901e-01 6.03117168e-01 4.76466268e-01
1.66319549e+00 -4.60600592e-02 -2.69968122e-01 1.39515030e+00
1.82281744e+00 -4.29075509e-02 3.65461320e-01 4.25734073e-01
1.15768993e+00 2.04502240e-01 1.10682055e-01 6.80669427e-01
1.83408111e-01 1.09111488e+00 3.86842221e-01 -1.45830691e-01
1.56903267e-01 3.50978404e-01 3.88824940e-01 1.09640765e+00
-4.12681550e-02 -1.66194364e-01 -1.33215415e+00 3.86702865e-02
-1.93455338e+00 -4.59397763e-01 -2.98104882e-01 2.26737881e+00
5.96214652e-01 -2.15332866e-01 1.35322079e-01 -4.24398780e-01
4.10319149e-01 -5.02243601e-02 -4.97451872e-01 -1.15406585e+00
1.07266806e-01 4.88962591e-01 5.69431365e-01 2.71494746e-01
-1.02202058e+00 1.03951001e+00 5.51512384e+00 1.11935508e+00
-1.68924570e+00 3.27449918e-01 6.43893361e-01 2.95469798e-02
-7.47536272e-02 -1.19817987e-01 -8.82014096e-01 4.00495946e-01
1.60097897e+00 4.94684651e-02 1.03845096e+00 1.22230625e+00
-5.27954921e-02 -2.28682030e-02 -1.26208723e+00 1.14205801e+00
-5.83115332e-02 -1.71673620e+00 -1.87112927e-01 -1.38541147e-01
7.20871210e-01 6.14326060e-01 -7.75350630e-02 6.58818364e-01
4.62793440e-01 -1.21982110e+00 5.46279848e-01 3.32570255e-01
7.10818231e-01 -9.54510033e-01 8.98608804e-01 7.40945041e-02
-1.13419247e+00 4.17000391e-02 -5.03120124e-01 8.70379712e-03
-4.82410043e-01 4.33749646e-01 -5.24582326e-01 5.13971210e-01
1.06468177e+00 4.46528494e-01 -6.58832252e-01 7.81555116e-01
4.18271184e-01 5.24484813e-01 -1.32403016e-01 -1.99591041e-01
-2.66021229e-02 4.69138473e-02 1.31288216e-01 1.52945876e+00
4.62817430e-01 -3.25476676e-01 3.46660763e-01 1.23316634e+00
1.33428931e-01 5.93431592e-01 -4.83329982e-01 3.58943567e-02
1.74432889e-01 1.51157069e+00 -7.56839991e-01 -1.68905810e-01
-3.66139174e-01 7.10483611e-01 6.39503062e-01 3.75606865e-01
-9.42279637e-01 -2.87667155e-01 8.36002707e-01 -8.28307569e-02
1.96416005e-01 -2.02438623e-01 -4.31420118e-01 -1.08430088e+00
-5.17483354e-01 -1.09302115e+00 5.14077008e-01 -5.48159301e-01
-1.22289562e+00 8.04148078e-01 5.77542037e-02 -9.41807926e-01
-4.78132516e-01 -7.21355617e-01 -7.30410397e-01 1.05719292e+00
-1.03195429e+00 -1.01291776e+00 -5.69335341e-01 4.56028610e-01
3.38327318e-01 -6.92564487e-01 9.22457099e-01 2.52917528e-01
-7.37017632e-01 5.21288574e-01 5.43054044e-01 -1.69486985e-01
6.65886104e-01 -1.11666763e+00 9.41339582e-02 2.59878844e-01
-8.74116346e-02 4.56610620e-01 7.46853590e-01 -3.17496866e-01
-1.93540132e+00 -1.23251784e+00 1.16069190e-01 -1.94911003e-01
8.40558887e-01 -3.32967401e-01 -1.29895127e+00 3.88218969e-01
4.83351529e-01 7.65487105e-02 7.25241840e-01 4.32810158e-01
-3.14566076e-01 -1.33549601e-01 -5.91110945e-01 1.86840966e-01
3.66936088e-01 -3.65632325e-01 3.35651748e-02 5.07336080e-01
5.97037494e-01 -8.66855800e-01 -1.31962669e+00 2.36600578e-01
1.99396953e-01 -1.05826378e+00 1.10462356e+00 -4.28357512e-01
5.61249018e-01 -3.26716244e-01 -3.07232291e-01 -1.41614056e+00
-4.52269197e-01 -2.10777938e-01 6.44748211e-02 1.13458955e+00
2.54331857e-01 -3.66068393e-01 3.95650685e-01 2.80084431e-01
-2.52512068e-01 -6.53718710e-01 -6.26887739e-01 -5.68402171e-01
3.88029993e-01 -3.40787411e-01 9.83628750e-01 1.04739404e+00
-2.11972058e-01 5.28198004e-01 -1.84230417e-01 9.22390521e-02
2.43421569e-01 4.29621071e-01 1.00479448e+00 -9.78090346e-01
-9.86018002e-01 -9.39787686e-01 -2.52287328e-01 -3.10929984e-01
4.76944417e-01 -1.17634666e+00 -2.53368884e-01 -7.64495313e-01
4.20209587e-01 -7.60732114e-01 -3.03612560e-01 4.03088599e-01
1.91194296e-01 -2.17017218e-01 -2.70484667e-02 2.50052094e-01
-5.71214199e-01 3.14485371e-01 6.30513966e-01 -1.89967617e-01
-6.50441408e-01 -3.59149933e-01 -2.45963514e-01 3.02708983e-01
8.97987664e-01 -3.45179290e-01 -2.09255949e-01 -6.55319333e-01
6.97091401e-01 1.30696520e-01 2.81886637e-01 -1.33972800e+00
1.51318878e-01 -1.27708629e-01 2.86415428e-01 -1.91183388e-01
-2.43962988e-01 -5.93654096e-01 5.99448025e-01 2.09688291e-01
-2.59453803e-01 4.91838634e-01 6.73460782e-01 5.90658672e-02
-1.70478985e-01 -2.40248844e-01 8.72931957e-01 -7.57790506e-02
-9.04745221e-01 1.15996212e-01 -3.52549762e-01 -4.97877859e-02
8.88303399e-01 3.55759144e-01 -4.85859603e-01 1.52889997e-01
-4.48719978e-01 1.54439732e-01 7.57035434e-01 4.88422632e-01
2.11823046e-01 -1.11181235e+00 -6.85095906e-01 4.54475224e-01
3.75007331e-01 -2.36351758e-01 5.59039235e-01 6.95891380e-01
-1.12850225e+00 2.41508037e-01 -5.69119334e-01 -7.83762395e-01
-1.14464962e+00 9.68828678e-01 6.11718059e-01 -5.43081760e-01
-3.90222371e-01 4.98612642e-01 1.34511158e-01 -7.69638300e-01
-6.84458688e-02 -2.62996048e-01 -2.32163668e-01 -3.34289253e-01
2.76201904e-01 2.86369145e-01 5.27062058e-01 -5.82327843e-01
-1.80805519e-01 7.41618276e-01 1.96108773e-01 3.31241548e-01
1.20769072e+00 3.03476095e-01 -6.77239835e-01 4.89815891e-01
1.34238398e+00 -2.07513839e-01 -5.52164793e-01 9.79509279e-02
-1.44658908e-01 -1.06824227e-02 4.15666908e-01 -6.13634586e-01
-1.31968093e+00 9.80355144e-01 1.01835477e+00 1.24878407e-01
1.13845193e+00 8.48075524e-02 3.77614886e-01 3.41874838e-01
3.94598007e-01 -1.06159008e+00 -2.43013576e-01 8.14452171e-01
7.42289960e-01 -1.25481033e+00 -1.28546104e-01 -4.92309825e-03
-4.01995540e-01 1.64696133e+00 7.63590276e-01 1.22534424e-01
8.09717357e-01 7.94354141e-01 3.56642227e-03 -3.57205987e-01
-9.10263658e-01 9.41267312e-02 3.77923250e-01 2.14144409e-01
8.28372657e-01 5.22821426e-01 2.67562922e-02 2.86936551e-01
-2.29502022e-01 -9.51875821e-02 2.20301211e-01 7.72099733e-01
-3.07846069e-01 -1.32393491e+00 -6.26945734e-01 5.75686216e-01
-2.05159336e-01 -1.65250659e-01 2.55685627e-01 6.18368864e-01
2.34706905e-02 3.90076786e-01 1.96689978e-01 -7.17215657e-01
2.23621484e-02 6.04685508e-02 3.14808309e-01 -5.17885566e-01
-1.30495763e+00 -1.43274233e-01 -2.54137516e-01 -6.39789879e-01
1.65324524e-01 -3.31971586e-01 -1.21162331e+00 -6.16192579e-01
1.21044435e-01 4.97252606e-02 9.45884287e-01 5.69343686e-01
8.07642698e-01 9.58579242e-01 3.01515490e-01 -1.18516564e+00
-3.93612981e-01 -6.10081494e-01 -3.62053633e-01 2.42515475e-01
-1.12695858e-01 -5.83416343e-01 -1.70185361e-02 -1.15853719e-01] | [6.266903877258301, 3.738947868347168] |
1a3c9013-f467-4d52-b848-acf3304feb01 | monocular-visual-inertial-depth-estimation | 2303.12134 | null | https://arxiv.org/abs/2303.12134v1 | https://arxiv.org/pdf/2303.12134v1.pdf | Monocular Visual-Inertial Depth Estimation | We present a visual-inertial depth estimation pipeline that integrates monocular depth estimation and visual-inertial odometry to produce dense depth estimates with metric scale. Our approach performs global scale and shift alignment against sparse metric depth, followed by learning-based dense alignment. We evaluate on the TartanAir and VOID datasets, observing up to 30% reduction in inverse RMSE with dense scale alignment relative to performing just global alignment alone. Our approach is especially competitive at low density; with just 150 sparse metric depth points, our dense-to-dense depth alignment method achieves over 50% lower iRMSE over sparse-to-dense depth completion by KBNet, currently the state of the art on VOID. We demonstrate successful zero-shot transfer from synthetic TartanAir to real-world VOID data and perform generalization tests on NYUv2 and VCU-RVI. Our approach is modular and is compatible with a variety of monocular depth estimation models. Video: https://youtu.be/IMwiKwSpshQ Code: https://github.com/isl-org/VI-Depth | ['Vladlen Koltun', 'Matthias Müller', 'René Ranftl', 'Diana Wofk'] | 2023-03-21 | null | null | null | null | ['depth-completion'] | ['computer-vision'] | [-4.72489744e-02 1.90736979e-01 -1.60497382e-01 -4.80311841e-01
-1.09466422e+00 -4.73504871e-01 5.54934561e-01 -1.21212058e-01
-4.88300890e-01 8.52860808e-01 5.57451487e-01 2.47524399e-03
3.01598072e-01 -7.61278152e-01 -9.41368043e-01 -2.12298617e-01
-9.34819803e-02 8.55553150e-01 2.44942009e-01 -7.17649534e-02
3.70554000e-01 3.73002887e-01 -1.56656528e+00 -9.19569433e-02
6.96860790e-01 1.09987187e+00 1.79042771e-01 1.26335144e+00
2.24122450e-01 8.48347366e-01 -1.91126555e-01 -1.51691223e-02
5.93417168e-01 -8.44343230e-02 -7.68455029e-01 -2.19819218e-01
1.33483481e+00 -8.93025517e-01 -7.63404131e-01 6.84872270e-01
8.53272676e-01 9.45989601e-03 3.54470402e-01 -9.66592312e-01
-1.75887331e-01 -9.94765684e-02 -6.25956893e-01 2.94396967e-01
1.01099980e+00 2.97603369e-01 8.49493802e-01 -1.07935917e+00
1.17330122e+00 1.21528184e+00 9.07894015e-01 3.51703137e-01
-1.18991351e+00 -4.79295373e-01 -7.12207258e-02 9.10917073e-02
-1.37020707e+00 -6.20091677e-01 2.57253259e-01 -4.61992979e-01
1.54845107e+00 -9.46498737e-02 9.98785079e-01 1.17748833e+00
2.93337852e-01 5.91153383e-01 7.59849489e-01 1.62080745e-03
2.58133531e-01 -5.84137797e-01 -3.64118338e-01 7.66299069e-01
3.22025061e-01 3.13805580e-01 -9.73269820e-01 2.02610586e-02
1.08244383e+00 -1.13116607e-01 -3.44756931e-01 -7.12997437e-01
-1.41699541e+00 7.91988194e-01 7.26076305e-01 -4.54425037e-01
-1.31699935e-01 7.67739892e-01 1.67488873e-01 2.12908015e-01
6.78878486e-01 4.20420289e-01 -5.16476572e-01 -6.46125972e-01
-8.38378251e-01 5.35941243e-01 7.04174340e-01 1.29339182e+00
1.36824036e+00 2.29573220e-01 5.12571216e-01 3.95635545e-01
4.31464642e-01 1.09098148e+00 3.67431164e-01 -1.73521507e+00
7.06056178e-01 2.29656488e-01 9.12710279e-02 -6.76918626e-01
-5.56485057e-01 -1.42552093e-01 -4.92561191e-01 5.62643826e-01
3.79894257e-01 -2.01304629e-01 -1.21698773e+00 1.34407735e+00
5.06993890e-01 4.39528197e-01 1.26456127e-01 1.01217222e+00
9.54311669e-01 4.59162295e-01 -4.62841362e-01 3.64758819e-01
6.67432368e-01 -9.34600115e-01 -2.45117471e-01 -9.90558803e-01
7.37990260e-01 -6.88754320e-01 9.24398899e-01 2.91845798e-01
-1.00106204e+00 -2.43599519e-01 -1.28433967e+00 -8.60315084e-01
7.60058463e-02 -2.76643902e-01 7.93321729e-01 1.89302966e-01
-1.47442973e+00 6.45555913e-01 -1.08108437e+00 -5.81965566e-01
1.65483549e-01 3.39804441e-01 -6.16818309e-01 -5.33349395e-01
-8.62099409e-01 5.76091647e-01 1.35727718e-01 -6.81999996e-02
-1.20328975e+00 -8.89832079e-01 -1.54459512e+00 -5.66523433e-01
-6.86551630e-02 -1.16889966e+00 1.35142744e+00 -5.30231833e-01
-1.28292549e+00 9.02461112e-01 -3.89053315e-01 -5.20799398e-01
7.78362155e-01 -5.98843336e-01 4.39347625e-01 3.16585481e-01
4.56945568e-01 1.34034145e+00 3.33559096e-01 -1.22640693e+00
-6.95399284e-01 -5.03619313e-01 1.52135730e-01 7.47621655e-01
3.82217616e-01 -7.04495251e-01 -6.28014326e-01 -3.96017656e-02
7.66691625e-01 -8.57455254e-01 -2.22074986e-01 3.69023323e-01
-2.08200693e-01 5.61310768e-01 5.52540541e-01 -6.62521839e-01
5.37921429e-01 -1.74181104e+00 3.47999334e-01 1.70278084e-02
5.03842294e-01 -5.70121467e-01 -7.46030360e-02 3.95973414e-01
4.21222776e-01 -3.74938101e-01 -1.02031521e-01 -1.12895298e+00
-1.35407239e-01 3.45315039e-01 -2.43452378e-02 8.18652689e-01
-3.20299119e-01 9.34319913e-01 -1.03356636e+00 -2.49632850e-01
7.02742040e-01 6.41517460e-01 -1.00408602e+00 2.66896516e-01
-1.30486339e-01 6.36094749e-01 -1.00131938e-02 1.14677608e+00
7.47734249e-01 -1.90012068e-01 -3.34836356e-02 -1.49593696e-01
-1.38470754e-01 3.87061477e-01 -1.11173868e+00 2.57630134e+00
-4.22229409e-01 1.02281749e+00 1.97838619e-01 -1.39496773e-01
7.50672817e-01 -2.34727055e-01 4.53525245e-01 -8.51694584e-01
-4.71674167e-02 4.07276422e-01 -4.06915218e-01 -8.70387852e-02
9.31905150e-01 7.04716891e-02 -2.38312576e-02 -1.31682739e-01
1.55263752e-01 -9.24777031e-01 -5.88984787e-02 6.01195097e-01
1.35135126e+00 7.37180710e-01 3.00605625e-01 -1.03187032e-01
-8.63582361e-03 2.20534820e-02 5.71481526e-01 5.26162624e-01
-6.38424158e-02 1.21862555e+00 2.30030954e-01 -4.73313361e-01
-1.45930707e+00 -1.41678774e+00 -1.38013810e-01 4.62051570e-01
5.62971354e-01 -5.45619428e-01 -2.41641775e-01 -4.34358306e-02
3.33700716e-01 1.63624868e-01 -7.54392922e-01 3.60081434e-01
-3.43174219e-01 -2.63919145e-01 4.96785522e-01 7.10310936e-01
5.84724545e-01 -6.79165065e-01 -6.51209474e-01 4.20479737e-02
-1.06707640e-01 -1.36194885e+00 -1.34151131e-01 9.20215175e-02
-1.21235287e+00 -9.45140719e-01 -7.54823804e-01 -3.91719788e-01
3.31002980e-01 1.99834660e-01 1.36264837e+00 -1.72554597e-01
-2.78170198e-01 4.62516248e-01 -1.55573279e-01 -7.38470331e-02
3.15958895e-02 2.41124809e-01 1.08636409e-01 -8.76315892e-01
7.57371858e-02 -7.78942347e-01 -1.15518260e+00 1.60485163e-01
-4.85431314e-01 2.78959066e-01 7.19446987e-02 5.06172180e-01
6.83143437e-01 -1.18585122e+00 -1.66663781e-01 -3.88548642e-01
-2.12638855e-01 -4.21777874e-01 -7.67180204e-01 -6.41856253e-01
-5.29514909e-01 -1.14144221e-01 -1.81958657e-02 1.24489352e-01
-7.59987354e-01 2.16962263e-01 -5.09079695e-01 -7.12220609e-01
3.15763894e-03 1.57072425e-01 7.08089843e-02 -4.20513451e-01
1.11564136e+00 -9.32924077e-03 -1.83899254e-02 -1.76952660e-01
3.54009807e-01 2.86150306e-01 5.52634299e-01 -3.28294039e-01
7.08167970e-01 1.01556456e+00 4.46767546e-02 -7.64197290e-01
-6.31550491e-01 -5.95181465e-01 -6.52140379e-01 -1.88226730e-01
1.02949190e+00 -1.59203243e+00 -5.74834406e-01 5.71988106e-01
-1.01468539e+00 -9.62541282e-01 -2.88913310e-01 6.69423878e-01
-8.50322366e-01 5.31801224e-01 -9.14676487e-01 -3.39758217e-01
-3.16310436e-01 -1.18724573e+00 1.70983398e+00 2.05833942e-01
-1.66353524e-01 -1.14109600e+00 6.64677322e-01 3.60280544e-01
1.86660185e-01 5.16777754e-01 -1.17545947e-01 3.00741822e-01
-1.11169493e+00 -2.58638198e-03 -3.06923121e-01 1.01002164e-01
-3.85392830e-02 -1.27957597e-01 -1.23536909e+00 -5.87922037e-01
-4.84429270e-01 -7.10206687e-01 9.06282425e-01 5.99874437e-01
4.38244045e-01 2.08717391e-01 -1.60656407e-01 1.57977796e+00
1.82761478e+00 -1.38398007e-01 9.59394991e-01 7.75521755e-01
1.25880206e+00 2.59724464e-02 8.42700899e-01 6.66724682e-01
8.45487893e-01 6.56834960e-01 9.46578801e-01 -2.81637102e-01
-2.54092991e-01 -3.90783846e-01 1.96521252e-01 5.70801079e-01
-1.61612347e-01 -3.45249623e-02 -1.18017817e+00 7.08673000e-01
-1.70735788e+00 -5.66630065e-01 -5.46503589e-02 2.12318635e+00
5.26071608e-01 8.45972598e-02 -2.05334231e-01 -2.48249888e-01
-9.61894449e-03 4.53981191e-01 -7.91308105e-01 -3.00551832e-01
-2.38603368e-01 8.67300406e-02 1.13539159e+00 1.30993629e+00
-8.13664854e-01 1.21322417e+00 5.79099846e+00 2.81823874e-01
-1.03174758e+00 1.61925361e-01 4.92108434e-01 -6.27379775e-01
-5.51302791e-01 1.30975932e-01 -8.77121508e-01 5.56263700e-02
8.15587521e-01 6.89346492e-02 3.78644258e-01 1.03885782e+00
1.47062644e-01 -6.93754733e-01 -1.04699838e+00 1.34197462e+00
2.02568933e-01 -1.63419700e+00 -3.37139487e-01 3.52975488e-01
1.11885405e+00 1.10246503e+00 -3.89569312e-01 -5.03694899e-02
4.52507436e-01 -8.97917747e-01 8.25384915e-01 2.10514873e-01
1.22893906e+00 -6.05095327e-01 5.77713847e-01 1.07098117e-01
-1.37225282e+00 3.00280750e-01 -6.98124349e-01 -5.15476346e-01
4.61491138e-01 4.16339099e-01 -8.90206337e-01 6.33966088e-01
8.28930140e-01 1.36879361e+00 -4.99438256e-01 1.11776900e+00
-3.50430250e-01 -1.62040755e-01 -6.85873926e-01 3.74297053e-01
1.56392425e-01 -1.22838669e-01 4.86794859e-01 9.43921924e-01
6.79310501e-01 -2.17495874e-01 7.12920576e-02 3.55301082e-01
-5.43674119e-02 -5.94756790e-02 -1.14355969e+00 6.57869041e-01
4.85043436e-01 8.77893269e-01 -3.78731191e-01 -4.90660816e-01
-2.63114840e-01 1.25003803e+00 4.63251919e-01 1.90938622e-01
-7.01364040e-01 -3.92358527e-02 1.28934252e+00 2.26624519e-01
1.18537791e-01 -6.24886453e-01 -4.52255040e-01 -1.43439293e+00
-1.36435861e-02 -6.30285740e-01 1.36888787e-01 -1.22150409e+00
-6.39102876e-01 5.27584672e-01 -4.96195182e-02 -1.36795282e+00
-6.23687923e-01 -4.39364284e-01 -3.70836593e-02 9.05628383e-01
-1.59882593e+00 -9.73872483e-01 -1.21420002e+00 5.38666368e-01
6.29443526e-01 3.47279966e-01 5.60753405e-01 3.71396124e-01
1.26729786e-01 1.70617402e-01 7.83631131e-02 -1.51169762e-01
8.68897140e-01 -1.34799600e+00 1.13521707e+00 8.81203234e-01
4.94750664e-02 1.76431105e-01 7.61761546e-01 -8.82458448e-01
-1.53669822e+00 -8.48640621e-01 5.16919076e-01 -7.65635371e-01
5.76336861e-01 -4.40926015e-01 -4.93977964e-01 1.01543725e+00
6.62595183e-02 2.18404785e-01 8.06454569e-02 -1.51411489e-01
-2.49196202e-01 -1.05414435e-01 -1.14452589e+00 4.88842309e-01
1.56143403e+00 -6.40679121e-01 -1.65024981e-01 3.47359955e-01
7.70521045e-01 -1.43436730e+00 -8.83342624e-01 6.34877741e-01
7.94955075e-01 -1.50699282e+00 1.13470340e+00 2.27742046e-01
8.22425246e-01 -4.34503108e-01 -6.00656331e-01 -1.11706340e+00
-3.35742952e-03 -4.28030223e-01 -1.38534293e-01 4.58419234e-01
2.67819911e-01 -6.73348367e-01 1.27904224e+00 2.40406901e-01
-4.08659637e-01 -5.73867917e-01 -1.24807465e+00 -5.66888213e-01
-1.20879963e-01 -7.21393704e-01 1.71536878e-01 5.47290742e-01
-3.28967482e-01 1.44455969e-01 -4.55449939e-01 2.49321997e-01
8.37446749e-01 -2.39443660e-01 1.42265952e+00 -6.17333412e-01
-4.23356533e-01 7.72264898e-02 -1.07411432e+00 -1.65542507e+00
-1.88516229e-01 -5.43173552e-01 1.13714293e-01 -2.08331656e+00
-2.58483320e-01 -1.73780024e-01 4.30906087e-01 1.90035269e-01
2.09893748e-01 9.12532508e-01 2.07137801e-02 2.07864165e-01
-4.70151812e-01 6.01785183e-01 1.21984732e+00 2.39739656e-01
-1.39314562e-01 -6.13817811e-01 -2.85075493e-02 6.92354739e-01
6.47047222e-01 -2.25432098e-01 -5.98423660e-01 -9.29758549e-01
5.50757766e-01 2.31934533e-01 3.54914874e-01 -1.56796372e+00
8.24095458e-02 5.18155545e-02 5.71733475e-01 -6.42569423e-01
1.01655412e+00 -2.44500011e-01 2.44491473e-01 4.87286538e-01
3.04663181e-01 3.52011621e-01 3.78690183e-01 4.39883530e-01
-3.01281791e-02 2.72409648e-01 6.17360890e-01 -2.98662424e-01
-1.26796055e+00 5.30632555e-01 2.97107399e-02 2.93906420e-01
7.47977614e-01 -7.10515797e-01 -6.16479397e-01 -7.07085311e-01
-6.29905522e-01 2.29628742e-01 1.30100429e+00 2.56509453e-01
9.61556375e-01 -1.10585701e+00 -6.23676002e-01 3.59492421e-01
3.55717629e-01 7.11379588e-01 4.37415063e-01 7.10042179e-01
-1.53588223e+00 1.90182775e-01 -1.50625393e-01 -1.19163203e+00
-1.14777148e+00 -1.81249708e-01 6.26946092e-01 1.81284875e-01
-9.54413533e-01 1.22735298e+00 3.63326371e-01 -6.35999441e-01
1.51051208e-01 -6.91283762e-01 4.78279799e-01 -2.28077441e-01
2.40087777e-01 4.41053212e-01 7.54956007e-02 -5.65359473e-01
-4.84681040e-01 1.04519272e+00 3.37090015e-01 -5.40641904e-01
1.14756334e+00 -4.59251851e-01 2.44252875e-01 4.02323872e-01
1.43115675e+00 1.29958227e-01 -1.94369316e+00 1.32301241e-01
-5.28172255e-01 -9.82127726e-01 2.24340498e-03 -3.23461473e-01
-7.88208723e-01 7.48578548e-01 6.56061351e-01 -6.46259785e-01
6.43622220e-01 8.89739618e-02 6.00636959e-01 4.49433327e-01
7.64191389e-01 -7.69263625e-01 9.53628942e-02 8.80620956e-01
8.16757739e-01 -1.50142252e+00 4.57772106e-01 -3.42554063e-01
-3.93448412e-01 7.54645109e-01 8.04382801e-01 -4.75043416e-01
4.01331127e-01 5.20705342e-01 4.81894225e-01 -2.19226971e-01
-6.03649914e-01 -2.19541371e-01 -5.25916331e-02 6.32898867e-01
1.95433959e-01 -2.44067609e-01 2.59681940e-01 -8.07701290e-01
-4.82659459e-01 4.76456992e-02 7.26317942e-01 1.09256530e+00
-4.49429750e-01 -8.30309331e-01 -2.68959790e-01 2.06569910e-01
5.60704153e-03 -3.30206782e-01 4.58414033e-02 9.55309927e-01
-1.52899414e-01 5.96684873e-01 3.86527151e-01 -5.28653085e-01
1.78222179e-01 -4.80623841e-01 7.52222657e-01 -7.91499257e-01
-9.12610292e-02 1.12771377e-01 4.90910619e-01 -1.41174805e+00
-3.00959527e-01 -6.62551224e-01 -1.38935530e+00 -6.28603280e-01
3.68160695e-01 -4.07855332e-01 9.00891304e-01 6.65642500e-01
4.34363842e-01 2.20012814e-01 1.69098839e-01 -1.79267812e+00
1.96401641e-01 -1.03952193e+00 -4.17744517e-01 1.00333303e-01
6.63507879e-01 -6.43274844e-01 -7.20056474e-01 -3.78685862e-01] | [8.315435409545898, -2.3638577461242676] |
b4c8cb71-c5f3-448d-ba19-96a47a3cf860 | document-enhancement-system-using-auto | null | null | https://openreview.net/forum?id=S1Mnzp9qLB | https://openreview.net/pdf?id=S1Mnzp9qLB | Document Enhancement System Using Auto-encoders | The conversion of scanned documents to digital forms is performed using an Optical Character Recognition (OCR) software. This work focuses on improving the quality of scanned documents in order to improve the OCR output. We create an end-to-end document enhancement pipeline which takes in a set of noisy documents and produces clean ones. Deep neural network based denoising auto-encoders are trained to improve the OCR quality. We train a blind model that works on different noise levels of scanned text documents. Results are shown for blurring and watermark noise removal from noisy scanned documents. | ['Nigel P. Duffy', 'Hamid Motahari', 'Sridhar V. Dasaratha', 'Sunil R. Tiyyagura', 'Mehrdad J. Gangeh'] | 2019-09-14 | null | null | null | neurips-workshop-document-intelligen-2019-12 | ['document-enhancement'] | ['computer-vision'] | [ 9.89584923e-01 -3.80333662e-01 5.32435060e-01 -5.80018103e-01
-5.94511867e-01 -5.50983369e-01 5.01892626e-01 -1.66329816e-01
-5.82661510e-01 1.45224094e-01 4.79030639e-01 -1.61010906e-01
7.41507765e-03 -6.88938618e-01 -5.74201345e-01 -2.66226888e-01
1.79528415e-01 -1.26545608e-01 1.54300332e-01 -2.38889009e-01
7.53188729e-01 5.90977490e-01 -1.03764880e+00 9.84378695e-01
6.25106275e-01 6.11967862e-01 3.50997657e-01 1.81708086e+00
-1.46717966e-01 7.30806649e-01 -1.10116410e+00 -6.43615603e-01
5.47871768e-01 -2.46367708e-01 -6.02519929e-01 5.10897458e-01
8.47188652e-01 -1.04083931e+00 -7.65907228e-01 1.46328151e+00
5.95129728e-01 -1.81504115e-01 4.72098440e-01 -3.52140248e-01
-1.81443453e+00 7.30103672e-01 -4.54687357e-01 2.10930064e-01
3.24017853e-01 1.08703367e-01 4.33036685e-01 -7.95018375e-01
7.42690623e-01 1.33818567e+00 7.02216506e-01 7.04251587e-01
-1.09547651e+00 -2.80985594e-01 -4.65873718e-01 1.13298781e-01
-8.45265567e-01 -5.11218786e-01 5.03022552e-01 -1.15733184e-02
1.09527957e+00 2.48501122e-01 3.84503275e-01 8.70933712e-01
4.38705742e-01 8.44684780e-01 8.14550638e-01 -8.34136009e-01
2.28597540e-02 -1.90600276e-01 3.16684306e-01 3.55203986e-01
3.36958766e-01 1.08584259e-02 -6.98717058e-01 4.75459158e-01
6.94476068e-01 -3.36700156e-02 -3.72443259e-01 4.48656648e-01
-8.35268497e-01 4.46414888e-01 1.80431172e-01 4.84715760e-01
-3.33525717e-01 3.32754314e-01 4.06902343e-01 7.47156143e-01
1.55586839e-01 4.70436662e-01 -4.23856005e-02 -1.33582130e-01
-1.61324883e+00 -2.38463953e-01 4.08580393e-01 7.96218276e-01
2.86140472e-01 1.98557541e-01 -3.83252025e-01 9.31618273e-01
4.66993988e-01 5.99464655e-01 8.22304964e-01 -8.43775868e-01
4.29749489e-01 2.98139632e-01 5.04728146e-02 -9.36014712e-01
1.96788460e-01 -1.41969502e-01 -6.66318655e-01 6.69267297e-01
-1.35198966e-01 4.14893702e-02 -1.78009593e+00 5.73578954e-01
-3.61690432e-01 -3.13356012e-01 3.95266086e-01 9.93695438e-01
5.51226795e-01 9.69214916e-01 -4.48904604e-01 3.85886043e-01
1.21471930e+00 -1.08136761e+00 -1.42733276e+00 -4.22971666e-01
1.60987210e-02 -1.50325930e+00 9.63411629e-01 8.96321476e-01
-1.44403410e+00 -7.80635118e-01 -1.54130828e+00 -7.47704148e-01
-6.04065180e-01 6.38756990e-01 -9.58672166e-02 9.99208927e-01
-1.37809026e+00 8.46514106e-01 -7.43978620e-01 -1.13648795e-01
6.77609146e-01 3.30816239e-01 -4.24824715e-01 -3.46935034e-01
-9.00246918e-01 9.06947315e-01 2.79868245e-01 4.50980544e-01
-6.79190338e-01 -3.96483988e-01 -6.48902535e-01 1.55435011e-01
-2.13717282e-01 2.51275096e-02 1.30679274e+00 -1.17136073e+00
-1.84510183e+00 6.27817690e-01 1.93259209e-01 -5.63762426e-01
6.98612750e-01 -8.64471555e-01 -7.33511090e-01 5.41921198e-01
-4.29877728e-01 4.90355611e-01 1.55190265e+00 -1.09117055e+00
-4.20386136e-01 -1.90239400e-01 -7.30837464e-01 8.01981241e-03
-5.04489362e-01 3.24212313e-01 -7.95974493e-01 -1.18606830e+00
1.08764373e-01 -1.56145215e-01 2.02854410e-01 7.29198977e-02
-5.02503157e-01 6.31403208e-01 1.38473153e+00 -1.53833675e+00
1.17639709e+00 -2.20824552e+00 -3.24026674e-01 2.60252208e-01
-3.42210710e-01 7.45895922e-01 -6.95887506e-01 3.28961521e-01
2.11132690e-02 2.18027368e-01 -3.91407371e-01 -4.67770517e-01
-5.64817041e-02 -1.83932036e-01 -2.54582107e-01 4.03496265e-01
7.52968490e-01 8.69060636e-01 -4.13044602e-01 -9.78976637e-02
3.41933876e-01 8.59582007e-01 -3.12684566e-01 1.45349398e-01
-6.63632378e-02 -4.04090941e-01 2.61786312e-01 5.77553272e-01
1.25829208e+00 4.10717428e-01 -1.19549215e-01 -2.56774127e-01
-1.08144013e-02 9.74372998e-02 -1.25660253e+00 1.60061300e+00
-1.78375080e-01 1.37752342e+00 3.24553102e-01 -1.92762688e-01
1.07559812e+00 1.31148949e-01 -3.27879876e-01 -7.18862772e-01
2.93973118e-01 2.48722434e-01 -8.34502652e-02 -6.11332357e-01
1.20525849e+00 2.13192344e-01 5.77857494e-01 4.04460818e-01
6.32036924e-02 -3.85606796e-01 4.61013734e-01 1.74726218e-01
1.24639845e+00 7.66173080e-02 -2.94215977e-01 2.23997429e-01
4.36086595e-01 -1.35186747e-01 -2.79730320e-01 9.22724724e-01
-6.95584193e-02 1.15174711e+00 7.09101111e-02 -1.43954992e-01
-1.64949059e+00 -7.67357707e-01 9.18586254e-02 7.65968144e-01
-2.02333286e-01 1.64187197e-02 -1.05572498e+00 -2.71848202e-01
-2.27597833e-01 7.64886320e-01 -5.35816491e-01 -3.33522439e-01
-6.64066672e-01 -4.96672451e-01 9.01611328e-01 4.36430633e-01
5.92375278e-01 -1.10657728e+00 -4.82611269e-01 2.97650009e-01
3.55585068e-01 -1.03604853e+00 -8.17887366e-01 4.37582374e-01
-8.44268560e-01 -7.73813188e-01 -1.18424237e+00 -1.17275989e+00
9.04012084e-01 1.90645486e-01 5.32598376e-01 1.95550889e-01
-5.68727374e-01 1.90605000e-01 -5.96949697e-01 -2.90282816e-01
-1.04711521e+00 -1.98593333e-01 -3.42689425e-01 6.02030531e-02
5.87603033e-01 -5.93623891e-02 -5.48422813e-01 -1.69162124e-01
-1.80664766e+00 -3.53356481e-01 1.09200358e+00 5.79721868e-01
1.82200715e-01 5.21641910e-01 -2.05742508e-01 -5.86341798e-01
1.21476197e+00 4.22585398e-01 -7.41673589e-01 4.01844114e-01
-5.57550490e-01 3.57075512e-01 6.81217253e-01 -6.16193891e-01
-1.37617218e+00 8.18181261e-02 -1.70981631e-01 -8.10179561e-02
-1.99456736e-01 3.32228065e-01 1.24450117e-01 1.69136841e-02
7.43153572e-01 3.94505203e-01 -1.55622467e-01 -8.15577149e-01
3.62514794e-01 1.36443770e+00 1.19550371e+00 3.21755022e-01
1.08114552e+00 3.47875237e-01 -4.40138698e-01 -9.05161202e-01
-1.90392360e-01 -3.56222928e-01 -7.83085346e-01 -1.15312628e-01
6.00604773e-01 -5.83891690e-01 6.96895272e-02 1.16128302e+00
-1.44673479e+00 -3.42254251e-01 -1.49098542e-02 2.66816258e-01
7.95471892e-02 8.12061191e-01 -1.13498724e+00 -7.98480332e-01
-8.42634380e-01 -1.05084622e+00 1.09463930e+00 3.85359347e-01
6.83979616e-02 -4.94545102e-01 1.09013163e-01 9.79186147e-02
5.99301219e-01 -4.08350945e-01 5.86552024e-01 -3.49894255e-01
-4.02140588e-01 -6.56866789e-01 -4.46744174e-01 1.04085326e+00
2.77711570e-01 5.73163569e-01 -1.01470339e+00 -2.29619205e-01
-7.54403546e-02 -1.27391852e-02 1.28040755e+00 4.44292217e-01
9.07971442e-01 -3.02302897e-01 2.39384606e-01 6.09763145e-01
1.66202557e+00 3.57416838e-01 1.60808527e+00 6.54427707e-01
4.05091584e-01 3.25731963e-01 5.96983396e-02 1.53012648e-01
-5.29809296e-01 7.69495293e-02 8.99481624e-02 -3.46286409e-02
-6.76464558e-01 -4.52021472e-02 7.54880846e-01 5.35670459e-01
4.26239133e-01 -6.83950543e-01 -6.37641788e-01 4.45236534e-01
-1.08257627e+00 -9.81776118e-01 -3.28104049e-01 1.68360806e+00
1.16786242e+00 2.58508414e-01 -5.49849331e-01 4.56090063e-01
1.10948431e+00 1.39243484e-01 -2.51889348e-01 -9.95119572e-01
-3.18614095e-01 6.16237402e-01 8.17463636e-01 7.03073502e-01
-1.16899335e+00 9.59168673e-01 7.04461288e+00 4.54116434e-01
-1.28362656e+00 -5.05856574e-01 4.35247153e-01 -6.47687986e-02
5.07049542e-03 -4.24178690e-01 -5.20418525e-01 4.80316192e-01
1.02149928e+00 4.46818382e-01 6.64150536e-01 5.62648475e-01
3.54010195e-01 -2.18208209e-01 -8.05527329e-01 8.01413774e-01
2.62620598e-01 -1.27268887e+00 2.93696135e-01 -2.36906245e-01
8.05971205e-01 -2.10733160e-01 3.92448694e-01 -2.90569544e-01
3.47905934e-01 -1.16613615e+00 9.47181106e-01 8.45179021e-01
8.52749109e-01 -6.67743742e-01 9.72003400e-01 -3.33944708e-01
-4.81864870e-01 -4.79029417e-02 -5.57927191e-01 3.89440119e-01
1.37101531e-01 7.34967172e-01 -9.01566327e-01 4.64576436e-03
7.74777710e-01 4.60441619e-01 -9.27006483e-01 1.15271819e+00
-2.81635165e-01 3.72278631e-01 -1.32542863e-01 1.27162725e-01
2.30493426e-01 -1.35083139e-01 2.88668126e-01 1.87273705e+00
4.22428399e-01 -3.36056091e-02 -8.12246144e-01 8.95606756e-01
-5.59578776e-01 -3.53935242e-01 -2.33231578e-02 -6.96813822e-01
9.55282971e-02 1.28355479e+00 -7.67470360e-01 -2.87720203e-01
-1.11193053e-01 1.90367031e+00 -3.38151962e-01 3.58333051e-01
-3.73419374e-01 -1.27318716e+00 1.50349587e-01 -2.25081384e-01
8.25887144e-01 -2.85128593e-01 -4.21730399e-01 -9.29975033e-01
1.90545887e-01 -1.25078845e+00 -1.86829329e-01 -1.21123731e+00
-1.00909793e+00 4.71650511e-01 -9.62020278e-01 -9.56958354e-01
3.16466779e-01 -1.08485746e+00 -5.90812624e-01 1.06244838e+00
-1.67033017e+00 -7.43167341e-01 -3.82772923e-01 7.46707395e-02
8.58504832e-01 -2.35293359e-01 4.14215326e-01 2.76286095e-01
-3.20899427e-01 4.70828682e-01 6.77440584e-01 5.79397142e-01
9.30410385e-01 -1.38802624e+00 1.06016743e+00 1.59651995e+00
4.24134806e-02 6.80410981e-01 7.43132472e-01 -1.02512550e+00
-1.40459347e+00 -1.03054667e+00 6.36955976e-01 -1.87450219e-02
3.32146823e-01 -1.73385486e-01 -1.14181840e+00 3.15274656e-01
7.79500961e-01 -2.96127349e-01 3.40100586e-01 -1.01883090e+00
-3.22933614e-01 1.09645024e-01 -1.34517539e+00 4.57820535e-01
3.00574422e-01 -7.37013578e-01 -8.06898773e-01 -1.16755433e-01
4.17438984e-01 -4.43192959e-01 -4.09265161e-01 -2.54557550e-01
7.21350253e-01 -8.29833567e-01 8.54261696e-01 -5.56293607e-01
1.03424060e+00 -5.99085867e-01 -6.80352896e-02 -1.25734866e+00
-2.10365757e-01 -8.14221323e-01 -7.21752495e-02 1.11615527e+00
4.17223454e-01 2.22520426e-01 6.88445330e-01 4.66730267e-01
6.39794245e-02 1.28966004e-01 -3.31109911e-01 -4.16427612e-01
-1.87935367e-01 -4.73477691e-01 4.06845450e-01 4.03815269e-01
-3.17866713e-01 -1.26343459e-01 -5.62506199e-01 3.47399533e-01
4.65092003e-01 -5.87360263e-01 1.93653077e-01 -6.19298697e-01
-6.76637590e-02 -3.02167684e-01 -3.38661015e-01 -9.15962160e-01
-5.34171879e-01 -4.31967854e-01 3.99668157e-01 -1.84117639e+00
-6.53669536e-02 6.53796256e-01 -9.22957212e-02 2.01076344e-01
-1.73036724e-01 6.32821679e-01 3.52336675e-01 1.20765418e-02
-2.59681582e-01 2.23141000e-01 9.97918725e-01 -5.93570530e-01
-1.83503240e-01 -3.93982530e-01 -4.18109924e-01 4.25426364e-01
8.26410174e-01 -4.08502787e-01 2.05193356e-01 -1.18133450e+00
2.50924140e-01 -6.63188875e-01 3.32137883e-01 -1.10613215e+00
2.12803975e-01 4.07010883e-01 1.21838999e+00 -9.36045170e-01
-3.80625320e-03 -9.17053580e-01 -4.01903361e-01 6.78267241e-01
-9.13905025e-01 6.17220514e-02 2.44312927e-01 4.82861638e-01
-1.86659351e-01 -1.04640377e+00 9.67086375e-01 -2.62008328e-02
-5.95926583e-01 -4.01427895e-01 -8.08166921e-01 -6.37399435e-01
2.93948293e-01 -6.33529127e-01 -4.19317424e-01 -5.71319997e-01
-5.26876032e-01 -2.99955696e-01 5.29315591e-01 3.33648920e-01
1.08768916e+00 -1.02248955e+00 -8.32047045e-01 3.51204842e-01
-4.33310777e-01 -3.33711058e-01 5.51729910e-02 2.28760261e-02
-1.46116149e+00 3.54051620e-01 -4.51047748e-01 -2.84987018e-02
-1.50156891e+00 3.59840930e-01 4.55405593e-01 -6.80667385e-02
-3.69577765e-01 9.94433224e-01 -1.02948976e+00 6.78536370e-02
4.38441008e-01 -5.58451653e-01 -1.83406129e-01 -2.63522238e-01
1.34006727e+00 7.15205848e-01 4.78894442e-01 -2.78167814e-01
9.44208726e-03 4.99374270e-01 -5.39583266e-01 -4.65376079e-01
1.64663219e+00 -1.90194204e-01 -2.22247019e-01 -4.66968834e-01
1.45126510e+00 2.56477088e-01 -1.31204093e+00 4.95942775e-03
3.25915366e-01 -5.07250011e-01 6.11252904e-01 -1.34707415e+00
-1.11431396e+00 8.84478271e-01 1.28796518e+00 1.24773405e-01
1.42378545e+00 -8.28060925e-01 9.68365729e-01 7.88532615e-01
-5.64657271e-01 -1.74887669e+00 3.33533674e-01 5.70618033e-01
9.81273711e-01 -1.17122090e+00 3.12111706e-01 -3.23900357e-02
-4.61668462e-01 2.07067251e+00 6.09167479e-02 -3.11265916e-01
8.52969438e-02 7.35418439e-01 4.73558038e-01 -3.18618603e-02
-1.90715462e-01 4.94396649e-02 3.88058215e-01 9.07016635e-01
4.86963958e-01 -5.70202529e-01 -4.44611460e-02 3.86307091e-01
-1.76393822e-01 2.80676633e-01 1.07658315e+00 1.19978690e+00
-6.34951115e-01 -1.17394459e+00 -1.07537496e+00 4.60939258e-01
-8.21599185e-01 -4.84480739e-01 -8.38002205e-01 8.50065500e-02
-1.10626286e-02 1.13579178e+00 5.40503487e-02 -2.00901955e-01
2.50365436e-01 1.43773094e-01 3.81636590e-01 -4.58721370e-01
-8.41332018e-01 2.27070689e-01 -2.86686033e-01 -2.61430919e-01
-1.50021821e-01 -2.24345863e-01 -1.00937521e+00 -2.73915678e-01
-4.15126979e-01 -5.08701682e-01 1.09360468e+00 5.09059250e-01
2.59370357e-01 7.45410562e-01 3.44590276e-01 -7.02619314e-01
-6.71785593e-01 -1.15721118e+00 -4.81220067e-01 4.49742198e-01
8.71169865e-01 5.77169418e-01 -2.97628224e-01 9.43887174e-01] | [11.818465232849121, 2.4665098190307617] |
04ed567e-4c19-41ba-98f0-d102fb09cb92 | visualization-of-emergency-department | 1907.11039 | null | https://arxiv.org/abs/1907.11039v1 | https://arxiv.org/pdf/1907.11039v1.pdf | Visualization of Emergency Department Clinical Data for Interpretable Patient Phenotyping | Visual summarization of clinical data collected on patients contained within the electronic health record (EHR) may enable precise and rapid triage at the time of patient presentation to an emergency department (ED). The triage process is critical in the appropriate allocation of resources and in anticipating eventual patient disposition, typically admission to the hospital or discharge home. EHR data are high-dimensional and complex, but offer the opportunity to discover and characterize underlying data-driven patient phenotypes. These phenotypes will enable improved, personalized therapeutic decision making and prognostication. In this work, we focus on the challenge of two-dimensional patient projections. A low dimensional embedding offers visual interpretability lost in higher dimensions. While linear dimensionality reduction techniques such as principal component analysis are often used towards this aim, they are insufficient to describe the variance of patient data. In this work, we employ the newly-described non-linear embedding technique called uniform manifold approximation and projection (UMAP). UMAP seeks to capture both local and global structures in high-dimensional data. We then use Gaussian mixture models to identify clusters in the embedded data and use the adjusted Rand index (ARI) to establish stability in the discovery of these clusters. This technique is applied to five common clinical chief complaints from a real-world ED EHR dataset, describing the emergent properties of discovered clusters. We observe clinically-relevant cluster attributes, suggesting that visual embeddings of EHR data using non-linear dimensionality reduction is a promising approach to reveal data-driven patient phenotypes. In the five chief complaints, we find between 2 and 6 clusters, with the peak mean pairwise ARI between subsequent training iterations to range from 0.35 to 0.74. | ['Bobak J. Mortazavi', 'R. Andrew Taylor', 'Nathan C. Hurley', 'Adrian D. Haimovich'] | 2019-07-05 | null | null | null | null | ['patient-phenotyping'] | ['medical'] | [ 6.84704185e-02 -2.58357860e-02 7.21492320e-02 -4.54263389e-02
-6.37669861e-01 -6.25520289e-01 3.30627292e-01 8.54793787e-01
-1.50191069e-01 2.26412326e-01 7.26590633e-01 -4.26476836e-01
-6.73968375e-01 -3.69507134e-01 1.03746086e-01 -8.56791496e-01
-6.91047013e-01 8.48228574e-01 -6.22165203e-01 2.27145031e-01
-1.60105839e-01 8.41719031e-01 -1.21571672e+00 5.52338243e-01
5.06876349e-01 6.64896250e-01 -4.55616228e-02 8.90774846e-01
-5.29738255e-02 3.40755761e-01 -3.30109268e-01 1.12797581e-01
2.09701315e-01 -3.89078826e-01 -3.94015461e-01 1.69261560e-01
-2.30580822e-01 -1.24384258e-02 -2.52121925e-01 5.75626731e-01
8.51813853e-01 -1.37062937e-01 9.21706617e-01 -1.09099960e+00
-4.72241282e-01 9.59668979e-02 -3.84843558e-01 2.78648794e-01
3.76243889e-01 2.13877335e-01 7.49955773e-01 -8.93376708e-01
7.41603196e-01 1.02000022e+00 6.98345244e-01 4.75755900e-01
-1.66707885e+00 -5.89339528e-03 -2.27683604e-01 3.59933414e-02
-1.46149111e+00 -2.69342571e-01 6.93334639e-01 -9.34469998e-01
5.72337151e-01 5.65573454e-01 8.52112293e-01 9.00868297e-01
2.47757390e-01 4.51282352e-01 9.97312248e-01 -1.03638127e-01
4.60128605e-01 2.00021774e-01 1.44082367e-01 3.77625346e-01
3.56169462e-01 -1.49424061e-01 -1.69916078e-01 -1.01670361e+00
4.28107888e-01 7.81076849e-01 -4.38535482e-01 -6.53887570e-01
-1.45830059e+00 8.52846086e-01 2.01579928e-01 2.18420044e-01
-8.65117967e-01 -3.33238810e-01 4.20135856e-01 5.37895039e-02
2.23413199e-01 5.44727802e-01 -3.19636554e-01 -4.02212083e-01
-7.23200619e-01 -8.17853883e-02 5.41712582e-01 4.17517394e-01
2.62856305e-01 -2.59885997e-01 -1.87782586e-01 5.20598114e-01
1.78217545e-01 2.57945687e-01 6.35902524e-01 -7.68728375e-01
1.72087073e-01 1.05960214e+00 4.40981947e-02 -1.03154683e+00
-9.09788847e-01 -1.57921389e-01 -1.14585972e+00 -3.47716063e-02
9.45517346e-02 -1.18410900e-01 -6.65624499e-01 1.32446313e+00
6.05068445e-01 1.38084486e-01 2.88658619e-01 6.35832071e-01
5.67258358e-01 3.08146089e-01 2.66051292e-01 -3.74280334e-01
1.71279418e+00 1.50578069e-02 -6.31366849e-01 4.20396000e-01
1.08906960e+00 -4.72183764e-01 9.63839769e-01 2.51919925e-01
-6.92861438e-01 -1.14591882e-01 -6.29695058e-01 4.43876058e-01
-1.89090043e-01 1.16878875e-01 1.54216498e-01 4.99404043e-01
-7.75460243e-01 5.65195322e-01 -1.09011233e+00 -4.82386053e-01
6.28437102e-01 2.49280512e-01 -6.50998712e-01 -3.63072664e-01
-6.13000631e-01 4.09314543e-01 1.78599745e-01 -3.34329680e-02
-3.21441591e-01 -9.95276511e-01 -6.29825234e-01 4.56622653e-02
-1.50369480e-01 -9.67252970e-01 4.51234907e-01 -5.39117008e-02
-8.21386099e-01 6.79484010e-01 -2.32388198e-01 -2.65578210e-01
3.11350286e-01 -4.16071415e-02 -5.28351843e-01 3.77198637e-01
-2.21442223e-01 2.52543241e-01 6.73518479e-01 -1.19675124e+00
-2.86793739e-01 -7.49545813e-01 -6.91039205e-01 1.52955249e-01
-5.24404109e-01 -1.24371886e-01 -1.50210327e-02 -4.75499779e-01
2.43848607e-01 -1.10783207e+00 -4.22700405e-01 -1.00968219e-01
-3.66243392e-01 1.05707988e-01 1.10352898e+00 -6.86654806e-01
1.50049031e+00 -2.41065192e+00 2.26949960e-01 3.87046397e-01
9.22940731e-01 2.25550443e-01 1.27402008e-01 7.25598693e-01
-3.87384921e-01 1.69859394e-01 -4.15529996e-01 -3.55224758e-01
-1.83957741e-01 1.47182584e-01 -1.60266295e-01 7.13028789e-01
3.38722378e-01 9.92439747e-01 -7.80364633e-01 -3.46078485e-01
2.33526155e-01 8.20233166e-01 -5.66484392e-01 4.56090957e-01
3.39165837e-01 6.35414481e-01 -3.47236514e-01 4.91475850e-01
4.19323206e-01 -6.05851233e-01 3.86026710e-01 -2.41883665e-01
1.79175004e-01 -5.39001152e-02 -9.98504281e-01 1.23956132e+00
1.21931776e-01 5.05355954e-01 -6.83740079e-02 -8.20792496e-01
7.39915550e-01 4.55297768e-01 1.17537594e+00 -2.31046140e-01
2.70339027e-02 -2.23759860e-01 2.83424139e-01 -6.94291413e-01
1.90171808e-01 -1.32003069e-01 2.91374829e-02 6.54768527e-01
-5.85428774e-01 4.90198106e-01 -1.66314721e-01 2.83210248e-01
1.44010031e+00 -5.90155363e-01 3.95963788e-01 -1.88810855e-01
2.66440846e-02 1.44775972e-01 3.78985822e-01 1.26614526e-01
-3.87995929e-01 8.21918249e-01 6.74959481e-01 -6.09320104e-01
-1.26087153e+00 -1.13208497e+00 -4.83578533e-01 2.97339827e-01
-3.51046354e-01 -6.48049176e-01 -3.24411720e-01 -4.30026025e-01
2.48540506e-01 5.44639647e-01 -8.04225147e-01 -3.87428612e-01
-2.35186979e-01 -1.20023715e+00 2.97279090e-01 3.98267955e-01
-3.71061981e-01 -7.10691333e-01 -1.00732768e+00 3.06189656e-01
5.25703095e-03 -8.43037665e-01 -2.86235929e-01 1.39936805e-01
-1.13162053e+00 -1.28950274e+00 -7.52861023e-01 -3.37237060e-01
9.07848060e-01 -1.13149881e-01 7.03336656e-01 -3.13746870e-01
-9.96587873e-01 9.03442204e-01 -2.46851578e-01 -1.47376344e-01
-4.79230136e-01 -4.28016156e-01 4.97576147e-01 2.45356709e-01
6.34721100e-01 -5.29754519e-01 -1.03781855e+00 1.42669111e-01
-9.30774510e-01 -1.31002426e-01 5.13404071e-01 8.60275567e-01
7.43247032e-01 -7.12956563e-02 2.16489941e-01 -7.03313351e-01
1.08122849e+00 -8.48551333e-01 2.13913545e-02 1.26079038e-01
-8.30434680e-01 1.41550928e-01 6.34490430e-01 -5.18052042e-01
-3.10864806e-01 6.72905147e-02 3.80362302e-01 -8.42769802e-01
-5.53471804e-01 5.99055052e-01 1.71798632e-01 4.99186337e-01
6.88528478e-01 1.60778254e-01 2.79071867e-01 -7.03741133e-01
2.81498164e-01 8.46511722e-01 2.48206586e-01 -1.13156222e-01
3.94195378e-01 7.50308633e-01 2.31885001e-01 -9.37336445e-01
-3.35874334e-02 -8.71048272e-01 -7.83919275e-01 1.69794798e-01
9.52205181e-01 -7.14491248e-01 -8.87815356e-01 -4.97137278e-01
-6.30674005e-01 -6.22343719e-02 -5.36099494e-01 5.55121720e-01
-4.62502629e-01 4.88515943e-01 -4.50300694e-01 -7.57350504e-01
-3.52038354e-01 -9.05753314e-01 1.08455384e+00 -3.65705550e-01
-8.12954366e-01 -1.09862006e+00 5.90003192e-01 3.40192914e-02
1.25403672e-01 8.13853800e-01 1.39395440e+00 -8.97460222e-01
8.95434804e-03 -4.35793221e-01 -9.21989456e-02 -4.83509339e-02
5.44466972e-01 -1.29172532e-02 -7.70278633e-01 -4.27064806e-01
-1.28164077e-02 3.12042385e-01 3.96128595e-01 5.97277939e-01
9.83207464e-01 -3.44923049e-01 -5.62230349e-01 6.37510836e-01
1.03616774e+00 2.49446064e-01 3.48104239e-01 1.69056613e-04
7.93308198e-01 8.46949577e-01 3.17706585e-01 9.56069231e-01
4.58561838e-01 5.67522407e-01 2.10010007e-01 -1.33514181e-01
1.82792082e-01 -8.23770389e-02 7.45980293e-02 9.35912907e-01
1.28510535e-01 1.14664339e-01 -1.38335967e+00 6.02261186e-01
-1.74111223e+00 -9.14936543e-01 -2.71841049e-01 2.46025801e+00
3.87548625e-01 -4.56042707e-01 4.41887170e-01 1.63699448e-01
5.77382088e-01 -3.32507432e-01 -4.97021705e-01 -3.78472209e-01
-9.72348079e-02 -2.54126072e-01 1.81830376e-01 5.31874225e-02
-8.63655508e-01 1.92790985e-01 6.23114204e+00 7.54923820e-02
-1.11532974e+00 -1.04982756e-01 7.64969230e-01 -3.07036906e-01
-2.95590788e-01 -2.13161215e-01 -3.74314129e-01 4.85935777e-01
1.32078993e+00 -1.45724401e-01 2.87432700e-01 6.47896469e-01
5.94663143e-01 2.24222988e-01 -1.22174764e+00 1.40687335e+00
-1.08575881e-01 -1.28240657e+00 -7.99882226e-03 6.34421527e-01
5.08722603e-01 -1.40755162e-01 2.76473582e-01 -3.04021358e-01
3.87618542e-02 -1.16635323e+00 -1.25191242e-01 6.24517679e-01
1.13414133e+00 -7.75572836e-01 4.58479613e-01 2.04687685e-01
-1.08532381e+00 -3.75927746e-01 -2.62480587e-01 3.18201125e-01
3.82740200e-01 6.14439428e-01 -1.46223474e+00 2.79710293e-01
6.31798744e-01 7.50118315e-01 -3.93084079e-01 1.00996625e+00
5.47850490e-01 4.50048029e-01 -2.89051980e-01 4.10516262e-01
-9.67719220e-03 -4.03376192e-01 7.37438917e-01 1.04445624e+00
5.18907309e-01 3.75753880e-01 5.94338737e-02 7.52577007e-01
3.66028190e-01 2.28602812e-01 -7.79049456e-01 -5.27286351e-01
2.42655262e-01 1.32008648e+00 -6.97603285e-01 -3.60670030e-01
-1.03032738e-01 9.95727718e-01 1.11950628e-01 3.64568353e-01
-3.61289829e-01 -5.90626709e-02 1.08959162e+00 5.16464233e-01
2.21695721e-01 -2.64857322e-01 -2.69348383e-01 -9.64747787e-01
-1.42622411e-01 -9.37153339e-01 6.62053168e-01 -2.91134506e-01
-1.29514539e+00 5.91710031e-01 -8.41078013e-02 -1.62843549e+00
-4.71573412e-01 -5.60800076e-01 -4.74046707e-01 8.37487876e-01
-1.01738036e+00 -7.27121890e-01 -2.81513184e-01 6.02794588e-01
-7.68039823e-02 -1.65608555e-01 1.44129539e+00 2.33573601e-01
-6.78326607e-01 2.27278605e-01 6.29656017e-01 -1.43004730e-02
6.21312857e-01 -1.24065411e+00 9.44317207e-02 3.12943339e-01
1.21741248e-02 9.10919189e-01 6.28547370e-01 -7.11347044e-01
-1.52887642e+00 -1.11320662e+00 7.06105113e-01 -6.23498797e-01
5.06055892e-01 -2.06471682e-01 -1.10904312e+00 4.69208956e-01
-2.64256924e-01 1.86904907e-01 1.56430435e+00 -1.49011299e-01
-2.17071667e-01 3.14088017e-01 -1.14994752e+00 6.92232966e-01
6.17048264e-01 -5.65123439e-01 -4.51720566e-01 1.97545499e-01
3.87988806e-01 1.40163004e-01 -1.38770413e+00 3.59373510e-01
4.51915264e-01 -7.62504876e-01 9.77362692e-01 -1.12838590e+00
4.08457443e-02 -3.38327020e-01 -1.78398088e-01 -1.36278200e+00
-6.04695976e-01 -7.08271146e-01 -3.73322636e-01 7.45653927e-01
1.87317267e-01 -6.75700545e-01 7.99082398e-01 9.67356443e-01
1.28966928e-01 -1.08614111e+00 -7.49193728e-01 -3.11033726e-01
-2.67879695e-01 -2.95024455e-01 4.49226528e-01 1.12268102e+00
3.31067204e-01 -2.39794441e-02 -1.70819998e-01 2.60437161e-01
5.92099547e-01 1.68950856e-01 7.00725019e-01 -1.52061653e+00
-3.09320748e-01 -4.06863004e-01 -8.46659780e-01 -3.61407220e-01
-4.55500335e-01 -9.36175108e-01 -5.37538111e-01 -1.55864358e+00
3.66612226e-01 -6.09415948e-01 -4.41894084e-01 2.15823874e-01
-2.71856368e-01 -3.09464857e-02 4.50141318e-02 5.90045750e-01
-2.67449468e-01 5.22732556e-01 7.11753190e-01 6.70367703e-02
-7.05456555e-01 -1.46185875e-01 -6.68861747e-01 4.10339206e-01
7.96704888e-01 -6.31639063e-01 -3.49841744e-01 1.51477352e-01
-1.64611377e-02 5.41570783e-02 3.01400334e-01 -6.57145679e-01
1.48676513e-02 7.57679716e-02 4.49648380e-01 -5.57685196e-01
4.72817093e-01 -9.96510029e-01 5.36526501e-01 6.93231106e-01
-1.64228156e-01 5.63070416e-01 2.58978426e-01 8.20805192e-01
-1.74824938e-01 4.14821416e-01 4.18435097e-01 1.63670346e-01
-2.33320743e-01 3.64749968e-01 -3.99427086e-01 -2.19749175e-02
1.27736771e+00 -3.72154504e-01 -8.47120136e-02 -3.95645678e-01
-1.15981364e+00 2.66810171e-02 5.69167435e-01 2.40486950e-01
9.14543450e-01 -1.46829164e+00 -7.47491002e-01 3.87198269e-01
4.79189575e-01 -1.16679959e-01 4.84314799e-01 1.42574978e+00
-4.79964852e-01 3.63867939e-01 -5.23768552e-02 -9.57169294e-01
-1.61773276e+00 8.97712588e-01 -6.09170645e-02 -1.91309467e-01
-1.00750709e+00 2.51443177e-01 2.85430759e-01 -1.68368056e-01
-6.25579208e-02 -1.39545754e-01 -3.37883353e-01 3.18293005e-01
7.73669839e-01 5.13220549e-01 -9.74661633e-02 -7.93785512e-01
-3.95484686e-01 5.74742734e-01 9.42636058e-02 1.83868974e-01
1.55143762e+00 -2.06562340e-01 -5.24439551e-02 6.38999164e-01
1.56265056e+00 -3.79071049e-02 -1.01261604e+00 4.40367050e-02
1.93129629e-01 -3.68744701e-01 -7.31944218e-02 -2.30213076e-01
-5.43883383e-01 1.03834856e+00 9.80078161e-01 4.25540954e-01
1.00647891e+00 1.40513808e-01 5.06992280e-01 1.48223981e-01
-2.23222613e-01 -5.68949997e-01 -1.22593962e-01 -1.68099701e-01
5.93960404e-01 -1.01594841e+00 -5.19151501e-02 -5.83009496e-02
-8.86363864e-01 1.04234254e+00 -1.37040675e-01 1.29133463e-01
7.45882690e-01 1.08954653e-01 2.15720743e-01 -6.21992171e-01
-8.45530629e-01 1.05509616e-01 3.70036274e-01 5.99906802e-01
3.26014459e-01 4.46838915e-01 3.22755761e-02 3.85645241e-01
-3.00407894e-02 -4.03318435e-01 3.50615829e-01 9.42508519e-01
-1.67655423e-01 -9.62667525e-01 -6.44127488e-01 7.47798324e-01
-3.26659232e-01 -8.81336778e-02 -4.44641739e-01 4.65618432e-01
-1.96394086e-01 5.91148496e-01 3.23588818e-01 -4.47324097e-01
3.33781987e-01 5.06240189e-01 1.80795901e-02 -5.07172108e-01
-1.75460294e-01 2.84725815e-01 -2.53979564e-01 -6.13792479e-01
-8.52944776e-02 -9.55296755e-01 -1.25219870e+00 -1.30951181e-01
2.14040145e-01 1.16930865e-01 6.23698413e-01 3.50047886e-01
1.22150993e+00 4.82388854e-01 5.80568075e-01 -5.70658684e-01
-4.18713450e-01 -6.03377700e-01 -7.22667396e-01 8.51768792e-01
5.70678651e-01 -3.55174631e-01 -3.81493717e-01 2.27027044e-01] | [7.095866680145264, 5.424148082733154] |
71459493-0ee6-4bd5-bcec-b9ddbe941b63 | the-bayesian-low-rank-determinantal-point | 1608.04245 | null | http://arxiv.org/abs/1608.04245v2 | http://arxiv.org/pdf/1608.04245v2.pdf | The Bayesian Low-Rank Determinantal Point Process Mixture Model | Determinantal point processes (DPPs) are an elegant model for encoding
probabilities over subsets, such as shopping baskets, of a ground set, such as
an item catalog. They are useful for a number of machine learning tasks,
including product recommendation. DPPs are parametrized by a positive
semi-definite kernel matrix. Recent work has shown that using a low-rank
factorization of this kernel provides remarkable scalability improvements that
open the door to training on large-scale datasets and computing online
recommendations, both of which are infeasible with standard DPP models that use
a full-rank kernel. In this paper we present a low-rank DPP mixture model that
allows us to represent the latent structure present in observed subsets as a
mixture of a number of component low-rank DPPs, where each component DPP is
responsible for representing a portion of the observed data. The mixture model
allows us to effectively address the capacity constraints of the low-rank DPP
model. We present an efficient and scalable Markov Chain Monte Carlo (MCMC)
learning algorithm for our model that uses Gibbs sampling and stochastic
gradient Hamiltonian Monte Carlo (SGHMC). Using an evaluation on several
real-world product recommendation datasets, we show that our low-rank DPP
mixture model provides substantially better predictive performance than is
possible with a single low-rank or full-rank DPP, and significantly better
performance than several other competing recommendation methods in many cases. | ['Mike Gartrell', 'Noam Koenigstein', 'Ulrich Paquet'] | 2016-08-15 | null | null | null | null | ['product-recommendation'] | ['miscellaneous'] | [ 8.61665830e-02 -2.13735640e-01 -6.22864068e-01 -1.08958229e-01
-1.00386107e+00 -6.87444627e-01 7.81505942e-01 8.65971204e-03
9.07480344e-02 4.94447827e-01 4.43423122e-01 -5.31355202e-01
-3.44097942e-01 -7.78709710e-01 -8.62219155e-01 -8.10178816e-01
-1.29080296e-01 1.07668328e+00 2.29374290e-01 6.91361874e-02
2.24293649e-01 2.43186623e-01 -1.57640851e+00 4.76322502e-01
6.35449529e-01 6.26975119e-01 6.51401952e-02 5.84173739e-01
1.96398139e-01 5.22691846e-01 1.94766179e-01 -4.81440187e-01
2.40223154e-01 -5.89662977e-02 -5.44118464e-01 1.04634397e-01
2.03857109e-01 -1.91921562e-01 -2.00216696e-01 7.97194481e-01
1.02266051e-01 3.21840018e-01 1.28798592e+00 -1.03439224e+00
-4.04462218e-01 6.62793398e-01 -7.19591558e-01 6.75469125e-03
3.61279905e-01 -2.02305302e-01 1.47668004e+00 -9.23093319e-01
4.04444546e-01 1.17060721e+00 7.67118931e-01 -2.27764361e-02
-1.94523311e+00 -3.51655096e-01 -4.44078818e-02 -2.48022571e-01
-1.42332101e+00 -2.36951634e-01 4.38475788e-01 -6.72486484e-01
8.25411439e-01 2.37863183e-01 6.92275703e-01 9.86104012e-01
3.41066301e-01 9.52040017e-01 1.10285246e+00 -3.04205298e-01
5.89324832e-01 1.72384847e-02 7.32573330e-01 7.48010874e-01
6.71727419e-01 1.06129376e-02 -7.25584209e-01 -1.15767086e+00
9.70021248e-01 2.03984007e-01 1.16982711e-02 -8.21666002e-01
-1.04506898e+00 1.32017970e+00 -4.14964646e-01 -2.63830841e-01
-5.38636684e-01 3.16895247e-01 7.44100064e-02 -1.69312447e-01
3.96285295e-01 3.37396674e-02 -3.33787769e-01 -1.61100343e-01
-1.15115690e+00 5.07733941e-01 1.30834401e+00 7.49069035e-01
7.24134386e-01 -3.03622991e-01 -2.52391189e-01 7.85519063e-01
8.69750559e-01 6.45784318e-01 6.89068437e-02 -1.05707240e+00
3.47534508e-01 1.70067593e-01 5.90729058e-01 -6.74699187e-01
-1.37734428e-01 -4.04391259e-01 -8.15412521e-01 -1.32212996e-01
6.00433230e-01 1.72666311e-01 -6.96311057e-01 1.61385798e+00
3.14362556e-01 3.48106116e-01 -6.81292415e-02 6.49535954e-01
1.05417199e-01 1.10619068e+00 -1.64424479e-02 -3.06141376e-01
1.40465629e+00 -8.14535797e-01 -2.10727453e-01 6.21250831e-02
6.94384694e-01 -8.68793011e-01 8.99886966e-01 7.52789855e-01
-1.04743743e+00 -1.72110796e-01 -9.89422381e-01 7.67651126e-02
1.44613340e-01 3.13736290e-01 1.24541759e+00 1.07349932e+00
-9.59396899e-01 6.49500370e-01 -9.58814263e-01 -1.59669906e-01
1.31701544e-01 3.27013463e-01 -4.85582799e-02 -4.67391759e-01
-9.27807808e-01 3.49206835e-01 -6.71420172e-02 -2.34171152e-01
-7.81508088e-01 -7.55805731e-01 -5.50725520e-01 2.35269710e-01
1.17336154e-01 -7.55061090e-01 1.23891950e+00 -1.97455093e-01
-1.56887555e+00 2.87317485e-01 -3.18602145e-01 -4.34729487e-01
1.73426822e-01 -2.15713575e-01 -2.68217802e-01 -7.53391311e-02
-9.98029187e-02 3.43428433e-01 9.43918526e-01 -1.17543972e+00
-3.75669479e-01 -1.40841380e-01 -2.01357052e-01 1.83389723e-01
1.64718419e-01 -7.75538236e-02 -5.63792288e-01 -4.29474592e-01
3.45130295e-01 -1.39716780e+00 -4.86513287e-01 -5.21671116e-01
-3.18659693e-01 -2.01557010e-01 1.20817102e-01 -5.11558533e-01
1.17141151e+00 -2.12143683e+00 1.32481739e-01 6.90672636e-01
-3.45253944e-03 -4.66991141e-02 -5.22570722e-02 7.62231708e-01
2.42981285e-01 -2.33339705e-02 8.95456597e-03 -5.04814446e-01
4.35599089e-01 3.38528156e-01 -6.24836564e-01 6.66475415e-01
-1.56185612e-01 6.85023665e-01 -6.94059193e-01 -9.31513384e-02
4.36767712e-02 4.19934332e-01 -9.29921627e-01 -3.93016070e-01
-4.20120507e-01 2.38862541e-02 -3.91371906e-01 2.56349534e-01
7.33770788e-01 -6.58189416e-01 4.09571409e-01 -9.73728150e-02
1.86927289e-01 4.75904524e-01 -1.83065736e+00 1.46017087e+00
-9.02182460e-02 3.09430026e-02 -1.19942039e-01 -4.78211224e-01
3.97119403e-01 2.06651285e-01 4.97222453e-01 5.35472669e-02
-3.40037853e-01 3.10037494e-01 -1.50077492e-01 3.59000951e-01
8.04717720e-01 -3.02462608e-01 -2.26277739e-01 8.49322975e-01
4.52075973e-02 2.46888682e-01 3.76692623e-01 5.08275092e-01
1.08008516e+00 5.02720289e-02 1.27307758e-01 -5.49158514e-01
1.53387263e-01 -9.87488031e-02 4.42348689e-01 1.19793952e+00
2.66443402e-01 5.66432178e-01 5.08088470e-01 -2.52745271e-01
-1.24877870e+00 -1.32712615e+00 -3.18859965e-01 9.61161137e-01
-5.77001497e-02 -7.64158905e-01 -3.57895881e-01 -2.36507192e-01
2.49971554e-01 7.96437860e-01 -2.22948998e-01 -3.38760428e-02
-1.41591668e-01 -1.44123137e+00 2.21531272e-01 3.05047661e-01
6.38216436e-02 -4.03379291e-01 8.18799436e-02 4.21360552e-01
4.87669744e-02 -8.68499875e-01 -6.53246701e-01 2.11088568e-01
-1.11587000e+00 -8.67267489e-01 -6.60169780e-01 -4.60669100e-01
4.50258046e-01 4.67199445e-01 1.08447075e+00 -5.70703208e-01
-5.46026789e-02 6.83437765e-01 -1.98493257e-01 1.43526480e-01
-5.59729457e-01 4.23433026e-03 4.04201031e-01 1.04801849e-01
4.34117734e-01 -5.36261737e-01 -5.88079214e-01 4.18284684e-01
-8.61048222e-01 2.28087559e-01 5.79121411e-01 8.10598671e-01
6.86612964e-01 1.83393881e-01 2.43776962e-02 -9.80577946e-01
6.62373543e-01 -6.65569603e-01 -7.36024261e-01 1.44319400e-01
-7.27115273e-01 4.35163230e-01 3.05503219e-01 -5.73382258e-01
-1.06076384e+00 7.78493956e-02 -6.59915507e-02 -7.12800547e-02
4.10324931e-02 6.97043717e-01 1.62342966e-01 1.50765017e-01
4.69716877e-01 1.56583130e-01 -1.43747032e-01 -7.25243092e-01
7.14106858e-01 4.32788223e-01 -7.90579543e-02 -9.83700097e-01
4.61659908e-01 5.69497228e-01 3.36377114e-01 -7.11870372e-01
-8.31749320e-01 -7.52691567e-01 -2.53580779e-01 2.52682775e-01
5.60608804e-01 -1.06188667e+00 -6.36680067e-01 2.41118133e-01
-7.08222985e-01 -3.73028755e-01 -4.46522444e-01 9.58115399e-01
-8.63850117e-01 4.97814983e-01 -1.18502414e+00 -9.60917115e-01
-6.26227632e-02 -8.68313789e-01 1.14549041e+00 -1.16686374e-01
-7.04838037e-02 -8.77292812e-01 5.65217137e-01 4.12322432e-01
1.86428130e-02 -4.46876466e-01 1.16036320e+00 -6.12799883e-01
-6.66887343e-01 -3.57454151e-01 5.13403676e-02 1.78886101e-01
-4.66043860e-01 2.96863243e-02 -4.61965084e-01 -4.18674022e-01
-1.42316267e-01 -7.86783919e-02 1.06488228e+00 6.79950416e-01
5.96211255e-01 -1.76122844e-01 -4.89109814e-01 3.05893451e-01
1.45427120e+00 -2.80980915e-01 5.92075765e-01 -1.95831254e-01
6.09879315e-01 1.60054773e-01 3.60908449e-01 7.44445801e-01
3.68817359e-01 8.07238519e-01 -1.77669927e-01 1.98894903e-01
1.40221626e-01 -6.49880111e-01 6.11358404e-01 9.64618146e-01
-1.56823814e-01 -1.89422309e-01 -6.71830058e-01 2.16259152e-01
-2.31109762e+00 -1.11636937e+00 -5.66686273e-01 2.62814999e+00
6.50938392e-01 1.50829509e-01 3.30316663e-01 -7.66923055e-02
5.98408401e-01 -4.49388996e-02 -4.38985616e-01 -2.51303345e-01
-6.28428683e-02 2.51134485e-01 8.81123126e-01 6.26874685e-01
-1.12320530e+00 6.27992690e-01 7.27507830e+00 1.18618715e+00
-3.28678250e-01 2.76860744e-01 4.01248753e-01 -1.45148933e-01
-4.48918164e-01 3.54568541e-01 -1.25980818e+00 5.02333641e-01
1.23572671e+00 2.04283223e-01 5.60888648e-01 8.24651122e-01
8.41403157e-02 -4.47227806e-01 -1.13848770e+00 9.76573825e-01
-5.16887121e-02 -1.39848602e+00 1.86243713e-01 8.11953008e-01
1.10333574e+00 1.32312700e-01 2.91489691e-01 4.39536184e-01
1.04174960e+00 -6.77552879e-01 5.30503929e-01 5.31024158e-01
4.99587804e-01 -7.30506659e-01 4.12279546e-01 6.57473445e-01
-1.06711268e+00 -2.22186022e-03 -7.08488584e-01 -8.48804414e-03
4.74921316e-01 1.05835211e+00 -4.66731042e-01 3.71371597e-01
2.43921906e-01 4.98835087e-01 -6.17811689e-03 1.05723989e+00
1.08292580e-01 1.21876419e+00 -7.64288008e-01 1.73244670e-01
6.28439933e-02 -7.36313641e-01 5.24648488e-01 1.07797110e+00
5.16398430e-01 1.19538143e-01 2.17069387e-01 6.35198414e-01
1.12674087e-01 2.14277923e-01 -3.26418430e-01 -1.66555345e-01
1.30476296e-01 1.02216542e+00 -7.97124028e-01 -2.73766309e-01
-5.36378980e-01 9.39984679e-01 2.58911818e-01 3.86868566e-01
-7.08283484e-01 4.98213470e-01 6.72807813e-01 2.05841556e-01
7.09624708e-01 -4.85410362e-01 5.15656546e-02 -1.61369610e+00
-3.96637946e-01 -7.29365885e-01 2.58953631e-01 -4.42883253e-01
-1.67545760e+00 -2.32540160e-01 1.57100171e-01 -9.08396304e-01
-2.85855114e-01 -6.98660195e-01 -1.97758704e-01 9.55850303e-01
-1.16781282e+00 -9.23988044e-01 6.26738310e-01 4.36606199e-01
1.80859834e-01 -7.16676861e-02 1.05167162e+00 1.62043467e-01
-4.73579526e-01 2.53674120e-01 9.93935287e-01 -4.96957064e-01
3.75475585e-01 -1.23863029e+00 2.48344153e-01 5.38865745e-01
6.11970961e-01 9.90600169e-01 7.82660902e-01 -8.17061603e-01
-1.77790856e+00 -7.02344060e-01 6.18986964e-01 -5.26875436e-01
6.79239929e-01 -4.55004662e-01 -5.38884699e-01 7.25018680e-01
-4.12907094e-01 -1.71089754e-01 1.29572845e+00 7.08835900e-01
-4.94364917e-01 1.93736598e-01 -9.07573879e-01 5.64811826e-01
7.28860080e-01 -4.47654516e-01 -3.93438488e-01 4.90819782e-01
2.20745817e-01 -2.13535186e-02 -9.18069005e-01 1.30816191e-01
7.56389081e-01 -8.76223326e-01 1.04980624e+00 -6.38314903e-01
2.44963780e-01 -2.98955470e-01 -4.39510643e-01 -1.02949715e+00
-9.40695524e-01 -7.32010484e-01 -7.40331292e-01 8.87355447e-01
4.74733084e-01 -4.69508082e-01 9.69178140e-01 8.74518633e-01
1.63239747e-01 -6.16638422e-01 -7.18194902e-01 -7.27020383e-01
5.43672554e-02 -6.16295815e-01 3.66869837e-01 4.97778982e-01
1.06336335e-02 5.04968941e-01 -6.17720902e-01 3.31914127e-01
1.02400231e+00 4.41229671e-01 6.61321580e-01 -1.44786131e+00
-9.84252810e-01 -1.57093093e-01 -2.05776349e-01 -1.66691589e+00
-8.01892430e-02 -9.96516764e-01 -1.69011518e-01 -1.50430632e+00
8.67661715e-01 -7.08722711e-01 -1.04108617e-01 -2.56087836e-02
1.08561143e-01 2.50269711e-01 4.07650648e-03 7.04889655e-01
-6.80803299e-01 3.66679370e-01 8.50974321e-01 1.86545074e-01
-2.02182487e-01 3.00975531e-01 -5.77174187e-01 7.74504662e-01
4.85569596e-01 -5.74973941e-01 -3.82108688e-01 1.94791958e-01
8.40492308e-01 1.48475483e-01 1.25711292e-01 -7.56962776e-01
7.54150972e-02 -9.38805938e-02 2.72109121e-01 -8.18877876e-01
5.27172506e-01 -3.78075242e-01 6.85770333e-01 2.52016813e-01
-3.30748469e-01 -4.41769511e-01 -1.41377270e-01 1.11476791e+00
2.38307759e-01 -5.59295654e-01 4.79712129e-01 -1.39344007e-01
-1.88300803e-01 2.72827595e-01 -6.52959585e-01 -2.57747769e-01
8.12177420e-01 8.14848319e-02 -6.18263371e-02 -4.98432100e-01
-8.56476903e-01 -2.39969149e-01 5.27735889e-01 -7.37029761e-02
2.34890386e-01 -1.44953263e+00 -6.24236465e-01 8.59670117e-02
-4.95192371e-02 -4.82539922e-01 3.32502037e-01 8.23114097e-01
-2.89607167e-01 6.61753535e-01 3.55907321e-01 -5.33460736e-01
-8.75573575e-01 5.79981565e-01 -1.61905184e-01 -9.22958791e-01
-5.49628079e-01 7.02624738e-01 2.82070100e-01 -2.63842016e-01
-1.81660846e-01 -2.87678301e-01 6.97345063e-02 -8.87134746e-02
5.50719559e-01 4.67877954e-01 -2.50080168e-01 -7.13080525e-01
1.05935456e-02 3.62844646e-01 -3.09574872e-01 -5.26892066e-01
1.15317333e+00 -1.22017264e-01 3.98672819e-02 6.87276781e-01
8.61399651e-01 2.99199075e-01 -1.38743830e+00 -2.65272409e-01
-2.08609745e-01 -3.13490719e-01 1.52861625e-01 -3.47277880e-01
-4.31497961e-01 5.85324585e-01 2.98073858e-01 3.16478580e-01
3.05257708e-01 2.30054796e-01 7.40273833e-01 3.26066315e-01
7.99473464e-01 -8.90092254e-01 -1.14300102e-01 5.20223320e-01
2.47637510e-01 -8.92145932e-01 1.70363769e-01 -6.72833920e-01
-6.73213780e-01 7.80844033e-01 -3.29553872e-01 -4.41682279e-01
1.02057099e+00 3.56275551e-02 -8.85941029e-01 -1.53100803e-01
-9.34479296e-01 -1.37358114e-01 4.15356606e-01 2.74075925e-01
2.03036860e-01 3.94834787e-01 -4.50625896e-01 9.26850319e-01
9.22259912e-02 9.98568162e-02 5.30861914e-01 8.00624967e-01
-5.57212055e-01 -1.48200190e+00 -3.55998933e-01 9.80858505e-01
-4.08832610e-01 -4.14349496e-01 2.93104410e-01 1.69854879e-01
-1.75589263e-01 9.69599485e-01 -7.08465744e-03 -1.22579604e-01
-1.35837287e-01 4.50851202e-01 7.64243186e-01 -8.25103462e-01
-1.62521929e-01 4.59889650e-01 2.41006911e-01 -3.17282200e-01
-2.00831041e-01 -1.30357373e+00 -7.58389711e-01 -5.83543420e-01
-4.52172250e-01 3.96185905e-01 7.13946640e-01 8.67712855e-01
5.31127751e-01 -1.83168486e-01 4.30031538e-01 -8.81143391e-01
-1.05160439e+00 -7.60852396e-01 -1.31195092e+00 2.32719064e-01
-1.08990788e-01 -8.20228875e-01 -4.51808155e-01 8.93571507e-03] | [7.46917724609375, 4.404806137084961] |
37a536ec-a0e7-4097-aa95-3079261df966 | open-relation-modeling-learning-to-define | 2108.09241 | null | https://arxiv.org/abs/2108.09241v2 | https://arxiv.org/pdf/2108.09241v2.pdf | Open Relation Modeling: Learning to Define Relations between Entities | Relations between entities can be represented by different instances, e.g., a sentence containing both entities or a fact in a Knowledge Graph (KG). However, these instances may not well capture the general relations between entities, may be difficult to understand by humans, even may not be found due to the incompleteness of the knowledge source. In this paper, we introduce the Open Relation Modeling problem - given two entities, generate a coherent sentence describing the relation between them. To solve this problem, we propose to teach machines to generate definition-like relation descriptions by letting them learn from defining entities. Specifically, we fine-tune Pre-trained Language Models (PLMs) to produce definitions conditioned on extracted entity pairs. To help PLMs reason between entities and provide additional relational knowledge to PLMs for open relation modeling, we incorporate reasoning paths in KGs and include a reasoning path selection mechanism. Experimental results show that our model can generate concise but informative relation descriptions that capture the representative characteristics of entities. | ['Wen-mei Hwu', 'JinJun Xiong', 'Kevin Chen-Chuan Chang', 'Jie Huang'] | 2021-08-20 | null | https://aclanthology.org/2022.findings-acl.26 | https://aclanthology.org/2022.findings-acl.26.pdf | findings-acl-2022-5 | ['open-relation-modeling'] | ['natural-language-processing'] | [ 1.04082115e-01 9.76740241e-01 -2.72950172e-01 -4.98157173e-01
-4.03701901e-01 -6.73647046e-01 4.17697728e-01 6.22518718e-01
1.56375989e-01 9.42638278e-01 2.68123657e-01 -3.68878722e-01
-2.28341788e-01 -1.55711675e+00 -8.64868760e-01 1.33559287e-01
-1.86725318e-01 8.53593886e-01 2.18415409e-01 -3.20313245e-01
-2.31589481e-01 2.32253075e-01 -1.14738452e+00 4.53144103e-01
1.29798186e+00 5.36594808e-01 1.89890444e-01 2.21081838e-01
-6.97315097e-01 1.27028799e+00 -5.57373583e-01 -9.94252622e-01
-2.52670825e-01 -4.45459843e-01 -1.41051459e+00 4.14393693e-02
9.03998390e-02 -1.83491006e-01 -4.07131493e-01 9.39993262e-01
-6.63395226e-02 2.02457696e-01 8.91642928e-01 -1.30194628e+00
-9.10377443e-01 1.42203832e+00 -2.67879575e-01 -5.52646853e-02
7.70289779e-01 -3.33266705e-01 1.50476813e+00 -6.24154389e-01
1.00738418e+00 1.27881742e+00 3.47977757e-01 6.59117579e-01
-1.24594128e+00 -3.81891429e-01 3.19842368e-01 5.40361822e-01
-1.54661238e+00 -2.40825817e-01 5.70364058e-01 -1.51163459e-01
1.16478539e+00 2.68563986e-01 6.08249009e-01 7.59848237e-01
-2.12752715e-01 7.54385352e-01 3.96771342e-01 -5.31450808e-01
-1.19330525e-01 5.42625487e-01 4.34145868e-01 7.54023552e-01
7.60729432e-01 -4.44278538e-01 -4.08346742e-01 -5.60697950e-02
6.90838337e-01 -3.26355994e-01 -4.58409846e-01 -4.03389633e-01
-1.06098115e+00 7.14499831e-01 6.34769678e-01 1.96792185e-01
-3.66360456e-01 -1.67988405e-01 1.63171187e-01 2.32097343e-01
1.52132139e-01 8.07252705e-01 -6.92625105e-01 2.47507527e-01
-3.71124268e-01 4.07161176e-01 1.28308773e+00 1.49076486e+00
9.63854134e-01 -5.14063299e-01 -1.01590075e-01 7.13209748e-01
4.36983317e-01 2.71447361e-01 1.59172386e-01 -6.55865133e-01
7.17997968e-01 1.02111793e+00 8.03892240e-02 -1.12565875e+00
-3.18653405e-01 -3.49734902e-01 -6.66706324e-01 -7.45312154e-01
7.68759623e-02 -2.37360865e-01 -5.00486374e-01 1.83008909e+00
4.29261327e-01 9.50805917e-02 5.84023595e-01 5.16154408e-01
1.52877462e+00 7.66793251e-01 3.20279032e-01 -3.16335261e-01
1.56244385e+00 -9.86269653e-01 -8.51599991e-01 -4.64019239e-01
1.07510114e+00 -2.05028594e-01 6.04789853e-01 -2.06893042e-01
-9.60850179e-01 -3.25855225e-01 -9.15898561e-01 -1.88253850e-01
-5.38758874e-01 1.07855164e-01 7.31430888e-01 1.06631540e-01
-4.97680217e-01 5.30621111e-01 -5.58357120e-01 -3.44096929e-01
2.56279975e-01 1.69727340e-01 -5.31263113e-01 -3.25642437e-01
-1.75247085e+00 1.17631114e+00 1.06331122e+00 -2.14764792e-02
-4.92631376e-01 -7.02244222e-01 -1.52588129e+00 3.95976961e-01
8.53027821e-01 -1.16235864e+00 1.14312530e+00 -5.41747987e-01
-8.83151889e-01 6.71417177e-01 -2.76603281e-01 -5.18325448e-01
-1.38961017e-01 -8.23471770e-02 -6.64349616e-01 1.63231686e-01
2.96757042e-01 6.57344401e-01 -8.22585449e-03 -1.52394009e+00
-5.09850800e-01 -1.13938861e-01 7.83375323e-01 3.44227195e-01
-9.55082625e-02 -7.53353760e-02 -4.29775625e-01 -2.05505982e-01
3.43811721e-01 -6.78856969e-01 -3.14548641e-01 -5.09845078e-01
-1.01084232e+00 -4.25330013e-01 2.53206968e-01 -4.70920324e-01
1.46449852e+00 -1.74317861e+00 -4.88304980e-02 2.77961969e-01
4.19841856e-01 2.28349835e-01 -3.51404808e-02 6.75230443e-01
-1.58250719e-01 4.01221156e-01 -3.86609547e-02 9.99764428e-02
8.99217129e-02 5.77048659e-01 -4.16643500e-01 -3.83853108e-01
5.52157104e-01 1.08372200e+00 -1.17408073e+00 -7.78954983e-01
-1.80927068e-01 8.96657333e-02 -3.10294807e-01 3.87377620e-01
-6.99300587e-01 8.02583843e-02 -7.14483023e-01 1.96497902e-01
5.58614135e-01 -5.88840485e-01 5.40137589e-01 -5.00282884e-01
5.31352043e-01 8.36763740e-01 -1.47499168e+00 1.28438735e+00
-6.24508083e-01 3.17016780e-01 -5.79072416e-01 -8.26608777e-01
9.52286303e-01 4.13480461e-01 2.23140744e-03 -7.57883936e-02
-2.12285429e-01 1.60434749e-02 1.17360100e-01 -7.89707303e-01
5.54312885e-01 -3.30038875e-01 2.58040745e-02 4.51329172e-01
3.81006360e-01 -1.33279130e-01 7.53059030e-01 6.97289467e-01
1.05453551e+00 5.55144018e-03 6.04512513e-01 3.04180175e-01
6.51677370e-01 2.46637017e-01 7.67622709e-01 6.09935224e-01
4.38167363e-01 1.68253392e-01 8.32299232e-01 -1.58591717e-01
-5.10992348e-01 -1.05344200e+00 2.30267271e-03 5.31367838e-01
3.48653436e-01 -1.16251874e+00 -4.04856682e-01 -9.63214695e-01
-2.30031595e-01 1.02349138e+00 -2.60551542e-01 -3.23638797e-01
-4.70016718e-01 -3.96155626e-01 5.48693955e-01 5.92336297e-01
4.71185625e-01 -1.00694132e+00 -2.82159179e-01 1.55462518e-01
-6.11443162e-01 -1.50226831e+00 1.21974610e-02 9.87185389e-02
-6.35577321e-01 -1.25725341e+00 4.51899953e-02 -1.06757271e+00
1.05071449e+00 -8.02441463e-02 1.67417037e+00 8.32046047e-02
2.01087803e-01 5.67633547e-02 -5.26656747e-01 -3.19688171e-01
-5.95843315e-01 2.15999529e-01 -1.65681109e-01 -3.18358451e-01
6.45227790e-01 -4.72953916e-01 8.57202783e-02 8.08578953e-02
-9.14834797e-01 5.36682487e-01 4.83094126e-01 6.08840942e-01
5.21058857e-01 6.48639202e-01 6.03747189e-01 -1.49679124e+00
6.73408687e-01 -6.87713087e-01 -1.34884462e-01 9.14521456e-01
-4.85528052e-01 4.91581351e-01 7.67800033e-01 -3.21763754e-01
-1.27506995e+00 -1.86697334e-01 8.56840312e-02 -3.92358489e-02
-1.72227889e-01 1.17854393e+00 -6.03768706e-01 3.18790495e-01
5.51778138e-01 9.81447622e-02 -6.10284865e-01 -1.05285622e-01
6.71649098e-01 2.84195185e-01 1.67053998e-01 -9.65851307e-01
1.06474757e+00 -4.40546721e-02 8.17893073e-02 -3.84155184e-01
-1.43012953e+00 -1.57974914e-01 -6.46043599e-01 3.07038367e-01
5.38287699e-01 -9.83973265e-01 -4.22436982e-01 -3.84258181e-01
-1.46890998e+00 -1.63393561e-02 -3.25994343e-01 3.97521526e-01
-3.06688040e-01 1.92679450e-01 -6.73825443e-01 -5.84784865e-01
-2.18591228e-01 -7.39696443e-01 5.33044517e-01 6.04400814e-01
-5.18419683e-01 -1.16517198e+00 4.26643677e-02 2.51708686e-01
-8.12218189e-02 8.11326355e-02 1.40488052e+00 -1.12900734e+00
-7.38894939e-01 -1.06718525e-01 -2.80404270e-01 4.52118255e-02
3.86902094e-01 1.36820093e-01 -4.83726501e-01 3.93744081e-01
-6.36568666e-01 -4.46357906e-01 4.41606969e-01 -8.57424885e-02
7.19982386e-01 -6.70357525e-01 -6.79841578e-01 2.02415675e-01
1.20312488e+00 2.96974182e-02 4.41571206e-01 1.55819431e-01
7.97276974e-01 1.06523514e+00 6.65931225e-01 2.04897523e-01
8.85597169e-01 4.09641087e-01 -1.45874158e-01 2.05499917e-01
-8.00533295e-02 -9.45749104e-01 -5.91602214e-02 7.86548615e-01
6.00584410e-02 -1.29689530e-01 -8.41044247e-01 6.60096884e-01
-1.77066302e+00 -1.02429605e+00 -2.48655766e-01 1.72631371e+00
1.45728600e+00 1.69021457e-01 -3.40425491e-01 -1.75363034e-01
5.87910593e-01 -1.06997222e-01 -3.31096739e-01 -1.07814193e-01
-2.42597893e-01 7.33750612e-02 1.16955191e-02 6.61178231e-01
-7.27281690e-01 1.25428510e+00 5.18630552e+00 7.38433242e-01
-4.43435162e-01 -2.99055308e-01 1.59987867e-01 3.72824758e-01
-8.87446642e-01 5.64285100e-01 -1.06464338e+00 -1.10408634e-01
7.79522479e-01 -7.91541815e-01 5.45164496e-02 7.09417105e-01
-1.75674304e-01 1.70469787e-02 -1.48005295e+00 6.35247588e-01
3.61338556e-02 -1.27876353e+00 6.12387538e-01 -2.33539835e-01
6.83972359e-01 -6.35081768e-01 -5.97157121e-01 7.23357618e-01
6.95427120e-01 -1.03361380e+00 3.65847349e-01 4.16315019e-01
3.05351347e-01 -7.30720043e-01 7.41061509e-01 5.64529061e-01
-1.42836297e+00 2.26880386e-01 -5.84328353e-01 -1.53840547e-02
2.26360694e-01 5.07541001e-01 -1.15194464e+00 1.18160284e+00
9.90051329e-02 7.34674394e-01 -5.86429954e-01 8.64153504e-01
-1.04743171e+00 3.39868993e-01 -2.46417359e-01 -2.27736369e-01
-7.52002746e-02 -9.60949957e-02 2.99014390e-01 1.03739023e+00
9.21500176e-02 6.30718112e-01 1.72439724e-01 1.24412918e+00
-3.28669429e-01 1.17248632e-01 -7.29547679e-01 -2.92161465e-01
7.38896310e-01 1.27423978e+00 -4.98386860e-01 -5.39043307e-01
-5.57502091e-01 7.01651037e-01 8.01179945e-01 5.28223276e-01
-6.78559661e-01 -6.39145494e-01 5.25978982e-01 4.33930121e-02
5.18675596e-02 -8.13137461e-03 -3.82896475e-02 -1.53642952e+00
1.20281368e-01 -6.21985614e-01 5.80242336e-01 -1.18701184e+00
-1.19980848e+00 6.91180766e-01 3.01713288e-01 -9.47638035e-01
-4.88292247e-01 -2.78560728e-01 -5.25042713e-01 9.09235775e-01
-1.46835732e+00 -1.14985144e+00 -1.05218805e-01 3.25653762e-01
2.85344929e-01 2.73830861e-01 1.00913727e+00 1.26720369e-01
-5.60183227e-01 3.94921899e-01 -6.67921484e-01 5.30877829e-01
4.69674557e-01 -1.40007639e+00 3.46798420e-01 8.98418009e-01
6.43306375e-01 1.33658218e+00 8.77624691e-01 -8.47823918e-01
-1.17394316e+00 -1.16256499e+00 1.52162516e+00 -3.57439041e-01
7.27280200e-01 -5.55380769e-02 -1.12013376e+00 1.31097758e+00
1.00065209e-01 2.62049139e-02 1.09387243e+00 5.58896542e-01
-5.70776165e-01 3.96672189e-02 -8.75734687e-01 7.48723507e-01
1.18157136e+00 -5.43101847e-01 -1.18934608e+00 3.29015613e-01
1.01397538e+00 -5.65608144e-01 -1.00667644e+00 6.01867616e-01
8.14075023e-02 -3.70618969e-01 8.63273799e-01 -1.18713045e+00
7.87637889e-01 -5.17122209e-01 4.77876812e-02 -1.61161327e+00
-2.19331861e-01 -2.32323378e-01 -6.51849985e-01 1.69979477e+00
1.11838460e+00 -3.27211738e-01 6.68846250e-01 1.02320123e+00
1.65657118e-01 -8.84867966e-01 -3.21944803e-01 -6.62505627e-01
-2.12146133e-01 -2.95869559e-01 8.18332374e-01 1.19655144e+00
7.52564311e-01 1.02370870e+00 6.85464963e-02 6.73972368e-01
3.61153513e-01 6.71428919e-01 6.00876689e-01 -1.29167581e+00
-3.97366613e-01 -6.85454607e-02 -2.29990438e-01 -1.19880056e+00
6.58058941e-01 -9.98146772e-01 -1.81544140e-01 -2.25798655e+00
3.18489581e-01 -7.60300875e-01 2.17895284e-01 7.23519027e-01
-4.88431871e-01 -5.23457944e-01 3.01684923e-02 -7.23119602e-02
-6.76839471e-01 5.47146082e-01 1.40344477e+00 -3.29522967e-01
-1.00831263e-01 5.96441738e-02 -1.03042638e+00 6.70714259e-01
4.15321082e-01 -3.79006267e-01 -9.27102149e-01 -3.24012339e-01
6.78451180e-01 5.03073037e-01 1.21303931e-01 -6.44774079e-01
3.62002313e-01 -4.38006073e-01 1.53588578e-01 -3.10449064e-01
1.50249645e-01 -8.54850352e-01 3.48663092e-01 1.35042056e-01
-6.17426753e-01 -4.12858158e-01 4.12581190e-02 3.17342609e-01
-5.09340942e-01 -5.74536026e-01 1.74664855e-01 -2.55597264e-01
-7.67751336e-01 2.97220081e-01 1.02230281e-01 3.61399293e-01
8.87797773e-01 1.30135909e-01 -5.76881111e-01 -3.35054994e-01
-9.27279711e-01 6.92697525e-01 2.30182216e-01 2.98120618e-01
7.42367208e-01 -1.33248496e+00 -5.78297734e-01 -1.76203415e-01
5.19011915e-01 5.90678751e-01 2.55867928e-01 3.27084929e-01
-3.91027004e-01 5.60645938e-01 1.66469082e-01 1.45791084e-01
-1.30850852e+00 7.16061592e-01 3.79790813e-01 -6.39061868e-01
-4.39975262e-01 9.84633923e-01 2.70012021e-01 -7.67255366e-01
1.53920501e-01 -4.87657130e-01 -7.64293969e-01 -4.53754663e-02
6.53673291e-01 -3.22863817e-01 -2.40268543e-01 -5.44895351e-01
-3.38865489e-01 2.02923253e-01 -2.43570507e-01 1.96567670e-01
1.08196020e+00 -2.05554485e-01 -2.51663387e-01 4.59026664e-01
7.61907876e-01 1.39897183e-01 -5.60220838e-01 -7.09342360e-01
3.11207265e-01 -2.36937016e-01 -5.25107026e-01 -7.57414043e-01
-7.30499864e-01 4.86144602e-01 -5.95549107e-01 4.70689505e-01
8.36684346e-01 5.72154045e-01 8.40942204e-01 7.55449891e-01
4.66715991e-01 -4.89491791e-01 -3.30559462e-01 6.60367310e-01
6.57734871e-01 -1.02912259e+00 1.85220074e-02 -1.32422209e+00
-8.23594630e-01 1.24028206e+00 1.10929060e+00 3.95192951e-01
3.75220329e-01 1.43661454e-01 -1.07492872e-01 -3.52706343e-01
-1.20843899e+00 -4.95747805e-01 3.07816207e-01 6.70629621e-01
6.55563891e-01 2.11847410e-01 -3.01130742e-01 9.59635854e-01
-3.95748079e-01 -1.70287743e-01 7.26565182e-01 8.03952932e-01
-4.03694421e-01 -1.32525754e+00 1.43627599e-01 4.71530974e-01
-1.99739233e-01 -4.25536811e-01 -5.43690026e-01 4.96601403e-01
1.60189897e-01 9.61553454e-01 -1.56620964e-01 -3.72982085e-01
5.29770792e-01 6.07886054e-02 6.00245237e-01 -1.13462961e+00
-2.50109345e-01 -7.34794199e-01 8.39116037e-01 -6.43626228e-02
-4.19622362e-01 -3.72286052e-01 -1.65555632e+00 -3.17388415e-01
-5.24247825e-01 6.38442695e-01 -2.60712989e-02 1.33127975e+00
2.51780838e-01 6.96512878e-01 2.85519242e-01 1.68570429e-01
-2.69392550e-01 -9.27300096e-01 -7.76410162e-01 4.27832395e-01
-7.02863559e-02 -5.98722816e-01 -1.95818208e-02 2.32925117e-01] | [9.180152893066406, 8.202777862548828] |
f1f3ad30-0979-4403-9d87-883c97586f45 | lightweight-adapter-tuning-for-multilingual | 2106.01463 | null | https://arxiv.org/abs/2106.01463v2 | https://arxiv.org/pdf/2106.01463v2.pdf | Lightweight Adapter Tuning for Multilingual Speech Translation | Adapter modules were recently introduced as an efficient alternative to fine-tuning in NLP. Adapter tuning consists in freezing pretrained parameters of a model and injecting lightweight modules between layers, resulting in the addition of only a small number of task-specific trainable parameters. While adapter tuning was investigated for multilingual neural machine translation, this paper proposes a comprehensive analysis of adapters for multilingual speech translation (ST). Starting from different pre-trained models (a multilingual ST trained on parallel data or a multilingual BART (mBART) trained on non-parallel multilingual data), we show that adapters can be used to: (a) efficiently specialize ST to specific language pairs with a low extra cost in terms of parameters, and (b) transfer from an automatic speech recognition (ASR) task and an mBART pre-trained model to a multilingual ST task. Experiments show that adapter tuning offer competitive results to full fine-tuning, while being much more parameter-efficient. | ['Laurent Besacier', 'Didier Schwab', 'Jiatao Gu', 'Changhan Wang', 'Juan Pino', 'Hang Le'] | 2021-06-02 | null | https://aclanthology.org/2021.acl-short.103 | https://aclanthology.org/2021.acl-short.103.pdf | acl-2021-5 | ['speech-to-text-translation'] | ['natural-language-processing'] | [ 2.16802403e-01 3.16849858e-01 -1.84656084e-01 -4.80643839e-01
-1.33684576e+00 -8.24130535e-01 7.68303275e-01 -2.41064966e-01
-6.37659132e-01 7.75490642e-01 1.81335568e-01 -8.50064278e-01
4.20536011e-01 -3.79573405e-01 -1.18089032e+00 -3.58188510e-01
3.10026646e-01 1.09325433e+00 -7.32551217e-02 -4.15975600e-01
-4.85027462e-01 4.27119642e-01 -9.41366971e-01 5.21501243e-01
9.69626069e-01 6.52328312e-01 4.64609742e-01 5.75465322e-01
-2.68862844e-01 2.32150182e-01 -4.29716289e-01 -8.02301824e-01
2.69970268e-01 -4.42616373e-01 -1.14291561e+00 -2.88718879e-01
4.95907545e-01 2.24413216e-01 1.28602553e-02 7.62417793e-01
6.35666966e-01 -1.65085524e-01 4.57308322e-01 -7.43566990e-01
-6.99060500e-01 1.23149514e+00 7.09019750e-02 6.72714859e-02
1.82828069e-01 1.71706110e-01 8.42000067e-01 -9.85285103e-01
6.17701411e-01 1.37988615e+00 7.08348036e-01 7.11558044e-01
-1.59509158e+00 -4.95782614e-01 -6.38904190e-03 5.87518141e-03
-1.19409657e+00 -9.72361505e-01 2.03155249e-01 -2.74696171e-01
1.62568855e+00 2.64896095e-01 2.69758046e-01 1.44993877e+00
7.00517893e-02 7.03603029e-01 1.22053254e+00 -5.76923072e-01
-1.19004250e-01 5.52035809e-01 -2.49574155e-01 5.64808249e-01
-3.83806437e-01 1.74507007e-01 -4.24435556e-01 -6.69582514e-03
6.80450499e-01 -6.14262581e-01 -1.76524594e-01 -2.28325412e-01
-1.71803117e+00 7.66606092e-01 4.49803799e-01 5.57691097e-01
-1.70620233e-01 9.49516892e-03 6.05125606e-01 1.01098084e+00
4.49612826e-01 5.10527194e-01 -9.28289592e-01 -1.00332692e-01
-8.44565809e-01 -3.30401361e-01 9.19354558e-01 1.03256905e+00
9.41725612e-01 2.01211527e-01 -7.91207477e-02 1.24203956e+00
-7.12771937e-02 8.69514644e-01 8.43011141e-01 -4.56181437e-01
1.00552237e+00 2.70656615e-01 -2.96517968e-01 -1.00314736e-01
-3.14271688e-01 -6.13543153e-01 -7.24947333e-01 -3.00658315e-01
1.89740434e-01 -2.28698999e-01 -7.68790066e-01 1.94515467e+00
8.32208805e-03 -6.96300194e-02 3.71328890e-01 7.00681388e-01
4.20680434e-01 9.53101575e-01 -3.63308527e-02 -1.19760364e-01
1.19712985e+00 -1.38053250e+00 -2.73491204e-01 -5.23970783e-01
9.87012863e-01 -1.14550304e+00 1.42247570e+00 8.50701109e-02
-1.42190456e+00 -6.81836128e-01 -8.56550336e-01 -2.95350164e-01
-5.36308527e-01 2.69961178e-01 4.24120069e-01 6.85959160e-01
-1.54513729e+00 4.08287615e-01 -7.62163818e-01 -6.04461730e-01
2.13759094e-02 6.99689984e-01 -5.96649230e-01 6.86691329e-02
-1.25543892e+00 1.38550353e+00 4.80243355e-01 -5.40622771e-02
-1.11574519e+00 -6.08864486e-01 -9.55021799e-01 6.29413128e-02
1.43590480e-01 -1.19482040e+00 1.54126644e+00 -1.45789087e+00
-2.11434865e+00 9.63756979e-01 -1.85331494e-01 -6.18107259e-01
4.56404805e-01 -1.31594807e-01 -4.09067810e-01 -1.29281074e-01
-8.52678088e-04 8.00214112e-01 8.99828076e-01 -7.86108792e-01
-4.38959330e-01 -1.03395440e-01 6.46904260e-02 2.78969318e-01
-3.28456223e-01 4.49596733e-01 -3.64562243e-01 -6.76256061e-01
-3.99796516e-01 -1.07264209e+00 -1.45978272e-01 -6.63565695e-01
-2.07538977e-01 -4.25217971e-02 3.32132995e-01 -8.25965881e-01
9.42954719e-01 -1.67391789e+00 6.98776841e-01 -1.23655565e-01
-5.00237048e-01 6.44258678e-01 -7.73610115e-01 5.33085108e-01
-1.58627838e-01 2.36039814e-02 -5.47177017e-01 -5.28766692e-01
5.54192327e-02 5.22415400e-01 -4.12276775e-01 1.73810318e-01
4.07193094e-01 1.25223315e+00 -7.70563483e-01 -2.56133467e-01
1.01894520e-01 5.04603744e-01 -5.15969515e-01 2.89415240e-01
-3.10692549e-01 6.77485824e-01 6.93822373e-03 2.89200962e-01
3.24445397e-01 8.58135447e-02 1.57148913e-01 -7.28080571e-02
-7.96886459e-02 9.56695855e-01 -5.76777339e-01 2.08750415e+00
-1.32607520e+00 3.98622334e-01 2.68683374e-01 -1.03816664e+00
7.63822794e-01 7.07900226e-01 -1.15406707e-01 -8.62883210e-01
8.08687508e-02 8.11540902e-01 -1.14907019e-01 -3.74167025e-01
2.65392482e-01 -2.97498405e-01 -2.42165551e-01 8.30643296e-01
7.02868760e-01 -9.85069498e-02 1.51958615e-01 -1.60743445e-01
9.26091850e-01 3.30211759e-01 1.87746406e-01 -3.79514486e-01
8.84827316e-01 -1.48104444e-01 1.95215315e-01 4.24534321e-01
3.82277399e-01 4.17583048e-01 4.05363506e-04 -3.93426031e-01
-1.22750080e+00 -1.16700077e+00 1.09135702e-01 1.39988041e+00
-5.63043594e-01 -3.62644583e-01 -1.14177716e+00 -9.00154352e-01
-4.02281433e-01 6.77851856e-01 -3.02158266e-01 -6.85868114e-02
-1.16542172e+00 -6.49478972e-01 8.44765663e-01 3.19301069e-01
2.96478659e-01 -1.08678412e+00 2.74908021e-02 2.11333111e-01
-4.91714180e-01 -1.29313588e+00 -9.40810382e-01 5.42757034e-01
-1.00799012e+00 -3.64397645e-01 -8.54864955e-01 -1.04994893e+00
5.20012796e-01 -7.81523809e-02 1.34245563e+00 -4.05783892e-01
3.72444004e-01 6.33940026e-02 -7.50660449e-02 -5.66030368e-02
-1.14954925e+00 9.18641925e-01 2.44410366e-01 2.10337024e-02
5.24026528e-02 -7.18469024e-01 1.06301615e-02 4.32442397e-01
-8.02529991e-01 1.42706811e-01 9.94304478e-01 8.94894898e-01
3.68565410e-01 -9.10176814e-01 7.54087567e-01 -8.37749362e-01
5.71107745e-01 -2.66148657e-01 -6.35358274e-01 3.64105612e-01
-6.08014345e-01 5.11643946e-01 9.63287354e-01 -5.64879060e-01
-9.78656590e-01 1.21681385e-01 -4.77212280e-01 -3.38157177e-01
-7.68165439e-02 4.90807891e-01 -2.69167155e-01 -1.47933736e-01
8.25421691e-01 3.12812924e-01 -1.02543477e-02 -8.49553049e-01
7.13792562e-01 8.96334887e-01 5.38333356e-01 -7.74952531e-01
8.05166781e-01 -6.45011365e-02 -3.94818664e-01 -5.46117663e-01
-6.71104968e-01 -3.26103032e-01 -9.74213600e-01 3.46166581e-01
7.32447982e-01 -9.90607977e-01 -2.25537121e-02 1.69441029e-01
-1.32998049e+00 -8.22209001e-01 -2.67072350e-01 5.23004055e-01
-8.81626189e-01 1.45741478e-01 -8.29309762e-01 -4.95387353e-02
-5.24437666e-01 -1.36269748e+00 1.27437520e+00 -4.26339775e-01
-2.46008068e-01 -1.31135833e+00 3.81945401e-01 6.16972268e-01
6.79960072e-01 -5.30227005e-01 1.03624022e+00 -7.25501955e-01
-4.19681370e-01 5.44901006e-02 -1.28314883e-01 6.39046669e-01
-5.69389609e-04 -4.88353610e-01 -1.18382120e+00 -4.42175686e-01
-8.27428028e-02 -3.23808551e-01 6.55474246e-01 8.77127200e-02
3.00991923e-01 -7.14916885e-01 -1.62132487e-01 8.04013371e-01
1.06940162e+00 -1.16797648e-01 3.23259175e-01 3.18308264e-01
6.50808036e-01 4.58418816e-01 8.39277282e-02 -4.20846075e-01
4.63740289e-01 1.12987578e+00 8.34348947e-02 -2.58687019e-01
-3.28494340e-01 -2.81994522e-01 1.07014406e+00 1.54708624e+00
6.71914667e-02 6.25558197e-02 -8.88238072e-01 7.01907635e-01
-1.78492296e+00 -5.01796901e-01 2.17619032e-01 2.39651370e+00
1.32375121e+00 -2.13012367e-01 3.08848232e-01 -4.14287597e-01
6.42598987e-01 -8.55632350e-02 -1.49617746e-01 -1.15439343e+00
-2.88138539e-01 5.81962526e-01 4.67581302e-01 9.13219094e-01
-8.28103304e-01 1.57368791e+00 6.44844627e+00 9.59487677e-01
-1.46800017e+00 7.72294283e-01 2.13822693e-01 -8.30684602e-02
-4.49857742e-01 2.02968329e-01 -7.82274127e-01 2.01556221e-01
1.80996430e+00 1.25288004e-02 8.21102560e-01 5.56094050e-01
4.44910303e-03 6.68773293e-01 -1.51861989e+00 7.70388603e-01
1.73567664e-02 -1.23302448e+00 4.97227281e-01 -1.08511709e-01
7.66771376e-01 7.08872497e-01 -1.02803959e-02 6.95740640e-01
3.74472439e-01 -1.04527724e+00 7.79935181e-01 8.64553824e-02
9.53412175e-01 -6.96379185e-01 6.72077835e-01 4.75204378e-01
-8.33725810e-01 1.65435180e-01 -4.53171909e-01 6.04715869e-02
2.27743059e-01 2.59099752e-01 -1.07602787e+00 7.67181873e-01
5.27805984e-01 4.29109633e-01 -4.51858640e-01 5.14751196e-01
-3.26794505e-01 5.87851644e-01 -2.89041370e-01 3.37042719e-01
5.37983775e-01 -2.73696572e-01 5.44018447e-01 1.66715896e+00
4.26871777e-01 -7.20161915e-01 -1.16701787e-02 6.80671334e-01
-2.61948079e-01 4.38970268e-01 -5.06685615e-01 -1.22159332e-01
6.89579025e-02 1.24770844e+00 -5.01198545e-02 -4.15910065e-01
-3.58870089e-01 1.32441247e+00 7.52128661e-01 3.61113548e-01
-6.08467579e-01 -2.62414813e-02 5.55108488e-01 1.49952306e-03
3.10770899e-01 -2.06891119e-01 -6.34955317e-02 -1.43216181e+00
-2.53271647e-02 -1.18992543e+00 1.29961818e-01 -7.81179965e-01
-1.07617831e+00 1.19098425e+00 -2.80091792e-01 -9.57452238e-01
-7.30672121e-01 -5.97329676e-01 -2.42166504e-01 1.27874577e+00
-1.62556648e+00 -1.57254839e+00 4.15916026e-01 7.17039824e-01
6.80059254e-01 -4.93197560e-01 1.25267828e+00 5.76068401e-01
-5.85602641e-01 8.61354649e-01 7.86883458e-02 3.80909182e-02
9.51413453e-01 -1.18856633e+00 8.50983739e-01 7.91547894e-01
4.11160320e-01 6.91015124e-01 4.30872053e-01 -1.86637938e-01
-1.19991660e+00 -1.28123701e+00 1.61098897e+00 -5.81060052e-01
8.45953524e-01 -6.94983840e-01 -7.45497227e-01 1.06795323e+00
6.81210518e-01 -3.00663471e-01 5.25757134e-01 1.60717383e-01
-5.95288396e-01 -2.17819616e-01 -8.12762201e-01 7.41304994e-01
1.01524985e+00 -9.22384560e-01 -6.79030538e-01 3.72643799e-01
9.66450930e-01 -3.58056664e-01 -1.07596195e+00 2.60901839e-01
3.28905791e-01 -5.67198694e-01 9.87828553e-01 -7.18393981e-01
1.38505697e-01 -6.18632771e-02 -1.20642073e-01 -1.82298625e+00
-1.36376247e-01 -1.07100523e+00 2.32446387e-01 1.25017071e+00
1.05790222e+00 -9.45511341e-01 6.24297671e-02 -8.89979228e-02
-6.29738927e-01 -4.35988933e-01 -1.19655538e+00 -1.11664593e+00
5.91385841e-01 -3.06992918e-01 7.59860754e-01 9.95495439e-01
6.35211775e-03 1.00964165e+00 -2.83319652e-01 1.52155444e-01
1.10836722e-01 -6.47436408e-03 8.65617514e-01 -9.37859714e-01
-5.94778299e-01 -6.10645533e-01 5.67164831e-02 -9.67637181e-01
5.15517116e-01 -1.59470093e+00 -3.21520939e-02 -1.18172836e+00
9.34348404e-02 -5.01574636e-01 -1.54206097e-01 7.72615254e-01
6.78253472e-02 3.17617118e-01 6.81620315e-02 3.23344916e-01
-2.31509864e-01 5.37730932e-01 9.76220667e-01 -2.62184799e-01
-4.36248295e-02 9.53152627e-02 -4.15419579e-01 3.64244819e-01
7.51029909e-01 -5.58533490e-01 -3.12413871e-01 -9.82708335e-01
3.56363058e-01 1.28096178e-01 5.64459935e-02 -7.84646034e-01
-5.00806570e-02 1.81383416e-01 -2.12156802e-01 1.23066403e-01
2.85274446e-01 -6.90834880e-01 1.98128030e-01 3.15720409e-01
-3.54952067e-01 2.24378258e-01 4.50352699e-01 -2.60159560e-02
-4.23290014e-01 -2.61322618e-01 8.08350563e-01 -1.27026454e-01
-2.48757958e-01 1.40146419e-01 -3.61361295e-01 -1.64984033e-01
5.12435555e-01 6.70546293e-02 -2.77848661e-01 -1.61208123e-01
-8.50913942e-01 -7.84155279e-02 4.14787382e-01 7.07528234e-01
-1.06195519e-02 -1.28995466e+00 -8.85406494e-01 4.40172553e-01
1.58091843e-01 -3.15472960e-01 -1.00575082e-01 1.07673383e+00
-3.49604905e-01 1.04883111e+00 -2.29141220e-01 -7.05234766e-01
-1.00996494e+00 6.02044880e-01 5.06213665e-01 -5.70302248e-01
-3.61364186e-01 7.91161418e-01 3.95595580e-01 -1.46538460e+00
-1.63574647e-02 -4.36699212e-01 2.74455696e-01 -1.94007784e-01
1.77367017e-01 -1.88998040e-02 5.45707524e-01 -8.87032092e-01
-3.07292432e-01 6.57045245e-01 -7.01578036e-02 -4.86993849e-01
1.19420791e+00 -2.52190739e-01 -1.94726303e-01 4.99764085e-01
1.20824528e+00 9.81483757e-02 -8.34165990e-01 -6.46899819e-01
6.88839331e-02 2.62457699e-01 -1.46556022e-02 -1.03113306e+00
-9.49345529e-01 1.12784934e+00 3.22295994e-01 -2.61797339e-01
1.04465997e+00 9.40299556e-02 1.02172458e+00 5.91224372e-01
5.27519941e-01 -9.65411782e-01 -2.93246835e-01 8.18085551e-01
1.11141860e+00 -1.07136834e+00 -7.78687119e-01 -1.39084578e-01
-6.26650035e-01 1.02698457e+00 1.33242488e-01 7.29002878e-02
2.43823230e-01 4.56216991e-01 4.68662590e-01 3.57597351e-01
-8.75596166e-01 -1.98217452e-01 3.56412858e-01 6.08059645e-01
6.45182788e-01 1.42729759e-01 -5.76413125e-02 3.44568878e-01
-5.11753440e-01 -1.10218026e-01 1.22664332e-01 3.53626788e-01
-6.93479776e-02 -1.66985047e+00 -2.29494110e-01 -6.03651069e-02
-4.70905334e-01 -5.64677000e-01 -5.18320978e-01 6.07430220e-01
-4.41981032e-02 7.54961967e-01 -9.87498090e-02 -2.19946384e-01
2.95918614e-01 4.11837727e-01 6.90720379e-01 -8.09205294e-01
-1.25528824e+00 5.04791103e-02 4.56931382e-01 -6.78743899e-01
-2.25646138e-01 -6.11468792e-01 -6.79552138e-01 -8.11635405e-02
-1.47861779e-01 2.61305183e-01 1.01456404e+00 1.12555337e+00
4.92520839e-01 3.60703766e-01 4.67781752e-01 -8.85344565e-01
-6.81021690e-01 -1.21964002e+00 1.18616574e-01 -2.45610089e-03
4.09584999e-01 -5.39420024e-02 -2.12171152e-01 1.88376814e-01] | [11.579437255859375, 10.265050888061523] |
08bd2b49-61d3-4aa4-b294-7748247b6141 | gender-classification-from-iris-texture | 1905.00372 | null | http://arxiv.org/abs/1905.00372v1 | http://arxiv.org/pdf/1905.00372v1.pdf | Gender Classification from Iris Texture Images Using a New Set of Binary Statistical Image Features | Soft biometric information such as gender can contribute to many applications
like as identification and security. This paper explores the use of a Binary
Statistical Features (BSIF) algorithm for classifying gender from iris texture
images captured with NIR sensors. It uses the same pipeline for iris
recognition systems consisting of iris segmentation, normalisation and then
classification. Experiments show that applying BSIF is not straightforward
since it can create artificial textures causing misclassification. In order to
overcome this limitation, a new set of filters was trained from eye images and
different sized filters with padding bands were tested on a subject-disjoint
database. A Modified-BSIF (MBSIF) method was implemented. The latter achieved
better gender classification results (94.6\% and 91.33\% for the left and right
eye respectively). These results are competitive with the state of the art in
gender classification. In an additional contribution, a novel gender labelled
database was created and it will be available upon request. | ['Juan Tapia', 'Claudia Arellano'] | 2019-05-01 | null | null | null | null | ['iris-segmentation'] | ['medical'] | [ 4.94096905e-01 2.82455474e-01 3.60472687e-02 -7.43942797e-01
-1.86734945e-02 -3.23565304e-01 6.09970987e-01 1.49298366e-02
-5.67864597e-01 6.62776172e-01 -9.72743258e-02 -2.50981480e-01
-3.15352082e-01 -4.82025504e-01 -2.39395857e-01 -9.90545511e-01
3.93935889e-01 5.03431082e-01 -5.62206749e-03 4.91819195e-02
8.44274163e-01 5.92873037e-01 -2.24800348e+00 4.68955517e-01
8.02259624e-01 8.61334562e-01 -3.22749168e-01 8.05289209e-01
-1.20178908e-01 2.71553159e-01 -7.96307445e-01 -4.32150453e-01
4.16072965e-01 -5.35083711e-01 -8.24574888e-01 1.70301452e-01
8.92509222e-01 -1.14748962e-01 3.93302381e-01 9.47032511e-01
7.27316797e-01 6.32690936e-02 6.69873893e-01 -7.07045019e-01
-2.71166295e-01 2.12693602e-01 -9.19324100e-01 6.49036169e-02
5.08993208e-01 2.54767835e-02 1.42506585e-01 -3.88786286e-01
3.68712604e-01 1.31409585e+00 6.96024716e-01 8.24445665e-01
-1.23116243e+00 -6.92584217e-01 -5.90081811e-01 6.14358857e-02
-1.22662139e+00 -4.77821678e-01 1.91294640e-01 -5.00177026e-01
9.64355886e-01 1.02032578e+00 5.81821442e-01 6.74139857e-01
7.11342394e-02 3.75818223e-01 2.24724770e+00 -8.05095732e-01
-1.76720291e-01 6.00217581e-01 2.31540933e-01 5.09909868e-01
2.07448393e-01 5.65275848e-01 -5.18223524e-01 9.12436545e-02
3.87965888e-01 -3.46077234e-01 3.03504527e-01 1.93385094e-01
-8.18750739e-01 4.74194705e-01 -1.38034567e-01 3.87633026e-01
-7.71402270e-02 -4.90224421e-01 3.48226756e-01 4.34551358e-01
4.44691449e-01 5.21098673e-01 -3.84211749e-01 -3.05242270e-01
-1.03515840e+00 3.58908996e-02 7.10111976e-01 3.29546839e-01
4.70618427e-01 -2.89339781e-01 -4.54979032e-01 9.95804906e-01
4.10976529e-01 7.73653030e-01 3.46414149e-01 -3.52626681e-01
-3.41413915e-01 7.84569681e-01 -7.74781406e-02 -7.93780446e-01
-6.50561512e-01 -2.42705584e-01 -6.08185530e-01 3.64377260e-01
7.59512067e-01 1.02832891e-01 -1.56606138e+00 9.60836649e-01
4.18574214e-01 1.91729024e-01 -2.52349019e-01 1.03275430e+00
1.29046595e+00 5.75963184e-02 3.41217406e-02 3.03933099e-02
1.45592690e+00 -2.82143831e-01 -6.49817705e-01 2.30405852e-01
1.42742023e-01 -1.42005777e+00 7.21159697e-01 7.02206850e-01
-8.37001503e-01 -6.26864314e-01 -6.43131077e-01 1.79652438e-01
-5.86918473e-01 5.40324688e-01 6.39828026e-01 1.75121439e+00
-1.07674420e+00 3.21985960e-01 -7.14701712e-01 -8.23874772e-01
2.47331396e-01 1.06029725e+00 -3.25837731e-01 3.41531843e-01
-4.13222879e-01 7.47846961e-01 1.65388957e-01 7.05175772e-02
1.28111213e-01 -2.61082649e-01 -8.44142973e-01 -7.00960457e-01
-2.43903562e-01 -3.08940977e-01 8.62240374e-01 -1.10317314e+00
-1.92530835e+00 1.73574841e+00 -5.07250547e-01 -2.02110529e-01
4.27243918e-01 6.06704317e-02 -5.38524091e-01 -2.37481073e-02
-1.33356318e-01 4.90916222e-01 1.09004152e+00 -8.76801312e-01
-7.42951632e-01 -8.55384290e-01 -4.18592632e-01 -1.63064316e-01
-5.84253073e-02 5.85136950e-01 -1.92419589e-01 -5.47006488e-01
3.00211042e-01 -9.95388091e-01 3.19664478e-01 -7.11721420e-01
-5.95878601e-01 -2.55541235e-01 5.24718761e-01 -1.03597355e+00
1.11776721e+00 -2.00465274e+00 -9.30268764e-02 7.67561078e-01
1.36867240e-01 6.04500711e-01 2.45019659e-01 2.12283339e-02
-2.06918061e-01 -3.33986878e-02 -4.22884598e-02 -2.99713939e-01
-1.34568978e-02 2.34866485e-01 1.40776217e-01 7.01248348e-01
-1.04070567e-02 4.26647425e-01 -3.21778238e-01 -2.86554068e-01
6.48208201e-01 5.22569478e-01 -2.37387940e-01 -6.18499657e-03
3.72025728e-01 8.89121592e-01 1.94873944e-01 1.02081430e+00
1.25903833e+00 5.14139295e-01 -2.37539589e-01 -2.11488470e-01
-4.00131196e-01 -2.04923138e-01 -1.46487319e+00 1.03765333e+00
-8.89915675e-02 5.83898067e-01 -6.14833720e-02 -9.72538948e-01
1.27673376e+00 9.78830084e-02 2.84407735e-01 -7.26798296e-01
4.10506099e-01 4.65630829e-01 1.82728201e-01 -8.96496117e-01
3.77779424e-01 -2.20958233e-01 3.51653516e-01 2.32596636e-01
1.82095289e-01 3.88854034e-02 3.20813656e-01 -6.38704836e-01
2.63151884e-01 2.59500146e-01 -3.93509530e-02 -4.70880538e-01
1.06521225e+00 -7.58909360e-02 3.18991512e-01 8.40769529e-01
-3.38088483e-01 5.61893880e-01 3.07696104e-01 -8.23213100e-01
-8.27914655e-01 -7.07814455e-01 -8.83848906e-01 8.14889312e-01
-1.03238322e-01 -6.78835884e-02 -9.05868709e-01 -4.21596557e-01
1.41281515e-01 -4.81521897e-03 -8.12338293e-01 2.89444625e-01
-2.49812514e-01 -1.37939155e+00 4.17651623e-01 1.77267548e-02
5.56083143e-01 -8.17131460e-01 -6.42814636e-01 -3.55903238e-01
2.72394776e-01 -6.20793104e-01 1.16314292e-01 -2.12900877e-01
-6.99868321e-01 -1.40930963e+00 -7.71071315e-01 -5.92825413e-01
7.95630157e-01 -3.89124125e-01 9.86931860e-01 2.01469302e-01
-8.32814217e-01 1.74774066e-01 -2.06334531e-01 -1.00256383e+00
-2.96683639e-01 -9.76016861e-04 9.27086249e-02 5.41545510e-01
1.21919215e+00 1.90383606e-02 -7.43404508e-01 4.16903079e-01
-5.88463724e-01 -2.15970814e-01 5.35071373e-01 9.57649767e-01
4.71806228e-01 -3.27909924e-02 -3.65295596e-02 -1.12139308e+00
3.67722183e-01 2.02916875e-01 -7.12510228e-01 -7.37276301e-02
-7.98793733e-01 -1.46538347e-01 -2.28875689e-03 -9.32748318e-02
-1.03103995e+00 5.92603274e-02 -2.07367420e-01 2.48939335e-01
-7.50030041e-01 5.88120334e-02 3.93858254e-01 -6.78731918e-01
7.99995482e-01 -3.10519300e-02 5.37359357e-01 -7.64397919e-01
-3.75647277e-01 1.39366293e+00 4.12736863e-01 -4.18114811e-01
3.95919263e-01 4.39218342e-01 3.12350482e-01 -8.82578611e-01
-4.11924094e-01 -8.28746676e-01 -6.80437624e-01 -2.43363246e-01
8.71973574e-01 -4.90558535e-01 -1.13439906e+00 1.13798916e+00
-3.84503245e-01 -1.89241823e-02 -7.33230636e-02 7.23513603e-01
-1.35555804e-01 2.07294956e-01 -4.22735691e-01 -1.14284849e+00
-2.91837841e-01 -1.06236851e+00 1.11643088e+00 9.21486795e-01
-2.63990611e-01 -6.97158873e-01 -7.42176780e-04 7.11321831e-01
4.75524038e-01 3.85656297e-01 4.18643057e-01 -4.62946773e-01
1.08521149e-01 -2.89581060e-01 -3.34458917e-01 2.55638629e-01
2.64304757e-01 4.58563447e-01 -1.46741641e+00 -2.30716035e-01
-2.38023505e-01 1.78429168e-02 8.77599835e-01 6.52688980e-01
1.13755488e+00 1.13084152e-01 -1.68565646e-01 8.49094510e-01
1.42807007e+00 3.77378613e-01 9.87065911e-01 6.41207814e-01
3.74104172e-01 8.60007346e-01 4.25400883e-01 4.05222625e-01
1.10350013e-01 6.08430862e-01 4.47892472e-02 -6.33225799e-01
-3.02495003e-01 5.22253275e-01 -2.95649707e-01 -2.78467294e-02
-9.00953472e-01 3.61658394e-01 -1.23789847e+00 3.54951292e-01
-1.36645448e+00 -8.11180174e-01 -5.80430984e-01 2.51997900e+00
8.52809072e-01 -4.47538272e-02 5.48986673e-01 5.30237257e-01
6.56575799e-01 -4.10614640e-01 -1.31845504e-01 -1.09421039e+00
-2.71504402e-01 1.14108682e+00 6.94545090e-01 4.99274671e-01
-1.48751676e+00 6.88317835e-01 6.00406837e+00 6.31823957e-01
-1.43352723e+00 -1.58316255e-01 7.86403835e-01 2.59804353e-02
4.49909955e-01 -2.94147521e-01 -9.05328333e-01 6.09694421e-01
9.69143033e-01 5.71170628e-01 3.88470471e-01 2.90126558e-02
1.44239128e-01 -8.13365519e-01 -3.59990001e-01 1.40417695e+00
1.80525944e-01 -7.69298851e-01 -4.19965595e-01 1.29140690e-01
6.33009791e-01 -2.99747020e-01 3.39152277e-01 -1.45641789e-01
-4.46980238e-01 -1.38660860e+00 1.23452738e-01 9.66848731e-01
1.20877182e+00 -9.25652564e-01 1.30724394e+00 -2.64481664e-01
-6.49154603e-01 -2.13660505e-02 -2.33052090e-01 -3.37064147e-01
-5.27017951e-01 4.67167646e-01 -8.48450482e-01 4.55489337e-01
1.14910638e+00 5.79606354e-01 -1.15373492e+00 1.36868131e+00
1.25738129e-01 5.73794663e-01 -4.32910293e-01 4.07022610e-02
-2.89258540e-01 -3.30699176e-01 4.58388954e-01 1.19802153e+00
2.35676244e-01 -1.02350235e-01 -2.60318696e-01 4.13077205e-01
6.78131282e-01 2.91636318e-01 -1.69774294e-01 1.55127198e-02
-1.50002092e-01 1.31093013e+00 -1.01204693e+00 -2.02289537e-01
-3.08823019e-01 7.33251095e-01 -6.24318004e-01 1.23221450e-01
-2.72098809e-01 -5.28686821e-01 6.15015388e-01 2.90897489e-01
-4.55177240e-02 3.59201580e-01 -8.03725123e-01 -8.81817818e-01
-2.40582749e-01 -1.03235185e+00 5.00241160e-01 -3.71288031e-01
-1.04728591e+00 4.38425273e-01 -1.38940901e-01 -9.94132578e-01
-1.38937831e-01 -1.02040637e+00 -3.03113669e-01 1.42277122e+00
-1.15638626e+00 -1.38122225e+00 -5.62650859e-01 4.20514524e-01
-5.83426096e-02 -6.48240030e-01 1.06527460e+00 2.45515198e-01
-7.80261874e-01 9.55219865e-01 -7.62844160e-02 6.48737103e-02
1.08163929e+00 -1.56611454e+00 -1.68949783e-01 7.83767402e-01
-2.14622915e-01 6.99204445e-01 9.14095938e-01 -4.34230298e-01
-1.16331089e+00 -4.91806984e-01 1.10781527e+00 -5.28402627e-01
-2.26117633e-02 -2.35164333e-02 -4.05282199e-01 8.36483762e-02
3.26378912e-01 -1.70599639e-01 1.13289082e+00 2.09497094e-01
-1.22237122e-02 -2.31508985e-01 -1.65774131e+00 1.46526143e-01
5.75448036e-01 -4.51885492e-01 -3.31257075e-01 1.74827650e-01
-6.65173411e-01 -9.70105529e-01 -9.88590360e-01 8.00909877e-01
8.71074736e-01 -1.36655343e+00 8.25809062e-01 -3.38840604e-01
1.86110109e-01 -4.27633226e-01 2.93093175e-01 -6.87712073e-01
1.81961283e-01 -6.60004735e-01 6.81670010e-02 1.14221215e+00
2.25868165e-01 -1.00613928e+00 9.03090537e-01 7.65334368e-01
2.39097893e-01 -6.84488595e-01 -7.73376763e-01 -5.29513180e-01
-3.28008205e-01 7.18969330e-02 6.75346136e-01 6.76925182e-01
-2.62892365e-01 -2.00256586e-01 -1.24069557e-01 6.29599690e-02
7.57519126e-01 2.12691635e-01 8.41770947e-01 -1.57520366e+00
-1.04395457e-01 -6.07906163e-01 -1.04210794e+00 5.86308911e-03
-1.57234490e-01 -5.71872890e-01 -5.07984698e-01 -1.00060570e+00
1.67355031e-01 -4.95814681e-01 -4.66075808e-01 5.59999228e-01
-8.78484547e-02 1.25006545e+00 -2.33824372e-01 3.28411981e-02
1.46681145e-01 -4.69510257e-01 1.01130521e+00 -7.82664791e-02
-3.74756634e-01 4.26410794e-01 -5.01509726e-01 3.92378122e-01
1.14425766e+00 -1.26418009e-01 9.89633333e-03 1.92262158e-01
1.37912497e-01 -5.07089913e-01 3.63216370e-01 -9.21401322e-01
-2.12777201e-02 1.17012024e-01 8.30712557e-01 -5.11425078e-01
7.44679123e-02 -4.09190983e-01 2.02009469e-01 5.80670297e-01
1.79212496e-01 -2.57549614e-01 5.40500939e-01 -6.79603815e-02
-3.21813613e-01 -1.16301298e-01 9.04735625e-01 1.23805940e-01
-6.71923637e-01 -1.86330080e-01 -2.13055372e-01 -5.44272482e-01
1.01942337e+00 -9.29226518e-01 -3.42538893e-01 1.83245867e-01
-9.83387589e-01 -3.58014137e-01 6.42884195e-01 3.80736411e-01
2.39991173e-01 -6.42009735e-01 -7.40067005e-01 1.08910811e+00
2.29359180e-01 -5.61734140e-01 2.20398545e-01 1.42922843e+00
-8.00945103e-01 4.48987663e-01 -6.24655426e-01 -9.22351003e-01
-2.07517695e+00 7.84032941e-02 5.75916588e-01 3.46751094e-01
-1.62305340e-01 1.25034297e+00 -3.57103795e-01 -6.53639078e-01
2.14902699e-01 -1.60372481e-01 -8.55466068e-01 2.29980603e-01
6.19495511e-01 4.71214205e-01 5.58500707e-01 -8.46018076e-01
-3.85888100e-01 1.05059898e+00 -5.71496822e-02 2.59408858e-02
1.12679696e+00 -1.36860922e-01 -7.95107365e-01 1.17323518e-01
5.93868852e-01 2.52693951e-01 -7.25535274e-01 1.66517824e-01
1.02867261e-01 -1.06864023e+00 3.25974263e-02 -1.32864904e+00
-1.06765139e+00 6.25987351e-01 1.59163010e+00 1.52659774e-01
1.56797898e+00 -3.18480909e-01 2.62390614e-01 -1.43169045e-01
1.67872280e-01 -1.40931940e+00 -8.12445402e-01 2.19270959e-01
3.75333488e-01 -1.46704102e+00 -7.64530199e-03 -5.61755478e-01
-1.84253186e-01 1.36039650e+00 3.77077132e-01 2.17804492e-01
5.90372384e-01 9.45225731e-02 4.93973613e-01 -3.16222012e-01
1.06276974e-01 -5.39354205e-01 8.16065133e-01 7.99989700e-01
9.01878595e-01 3.68808597e-01 -1.02054524e+00 2.00553462e-01
-4.56287295e-01 1.10167526e-01 4.08763915e-01 7.86310911e-01
2.59547792e-02 -1.55658400e+00 -7.94523358e-01 9.72634256e-01
-1.01465523e+00 8.60559717e-02 -4.94917631e-01 3.48937988e-01
6.97402775e-01 1.20078230e+00 5.08622304e-02 -5.40604293e-01
1.33467749e-01 6.83488622e-02 8.31786811e-01 -4.20582891e-01
-9.60259795e-01 2.18595326e-01 1.39628753e-01 -6.30493104e-01
-9.29604113e-01 -8.24710250e-01 -5.33059716e-01 -5.58940351e-01
-1.52593568e-01 -1.75953418e-01 9.58103120e-01 7.53916800e-01
2.41087079e-01 3.12478960e-01 4.36359853e-01 -5.26879609e-01
-1.60012562e-02 -1.17254305e+00 -9.38309491e-01 5.89297056e-01
4.33315992e-01 -7.58447051e-01 -2.57565677e-01 3.19579750e-01] | [3.7431323528289795, -3.6314902305603027] |
f675a13e-9b10-4c66-bcab-9b9aca772071 | online-fairness-aware-learning-with | 2108.06231 | null | https://arxiv.org/abs/2108.06231v1 | https://arxiv.org/pdf/2108.06231v1.pdf | Online Fairness-Aware Learning with Imbalanced Data Streams | Data-driven learning algorithms are employed in many online applications, in which data become available over time, like network monitoring, stock price prediction, job applications, etc. The underlying data distribution might evolve over time calling for model adaptation as new instances arrive and old instances become obsolete. In such dynamic environments, the so-called data streams, fairness-aware learning cannot be considered as a one-off requirement, but rather it should comprise a continual requirement over the stream. Recent fairness-aware stream classifiers ignore the problem of class imbalance, which manifests in many real-life applications, and mitigate discrimination mainly because they "reject" minority instances at large due to their inability to effectively learn all classes. In this work, we propose \ours, an online fairness-aware approach that maintains a valid and fair classifier over the stream. \ours~is an online boosting approach that changes the training distribution in an online fashion by monitoring stream's class imbalance and tweaks its decision boundary to mitigate discriminatory outcomes over the stream. Experiments on 8 real-world and 1 synthetic datasets from different domains with varying class imbalance demonstrate the superiority of our method over state-of-the-art fairness-aware stream approaches with a range (relative) increase [11.2\%-14.2\%] in balanced accuracy, [22.6\%-31.8\%] in gmean, [42.5\%-49.6\%] in recall, [14.3\%-25.7\%] in kappa and [89.4\%-96.6\%] in statistical parity (fairness). | ['Eirini Ntoutsi', 'Wenbin Zhang', 'Vasileios Iosifidis'] | 2021-08-13 | null | null | null | null | ['stock-price-prediction'] | ['time-series'] | [ 1.64864659e-01 -2.99687505e-01 -5.90906382e-01 -6.82434380e-01
-5.45206368e-01 -4.16925043e-01 4.69336629e-01 6.63206279e-01
-6.13382518e-01 1.14684117e+00 -3.45261723e-01 -3.74682367e-01
-2.32716992e-01 -9.23952103e-01 -4.70009536e-01 -6.67026520e-01
-3.30011278e-01 6.58732474e-01 3.34135801e-01 -2.78914958e-01
4.34870452e-01 2.18928471e-01 -1.72078955e+00 3.35375220e-01
9.82947886e-01 1.38894558e+00 -7.06504345e-01 6.92152083e-01
-3.77660066e-01 1.04323995e+00 -1.00336850e+00 -6.04117334e-01
5.26243925e-01 -3.71746808e-01 -3.34011942e-01 -2.25793108e-01
2.33090341e-01 -3.03999782e-01 1.34957016e-01 8.61744046e-01
4.23173845e-01 1.18508533e-01 7.53901541e-01 -1.83481371e+00
-5.67227826e-02 7.28318453e-01 -1.18752265e+00 4.89071012e-01
-1.64319590e-01 8.28529745e-02 7.56985724e-01 -2.32665792e-01
1.16487309e-01 8.91017318e-01 5.82137048e-01 6.05764985e-01
-1.22632039e+00 -1.19749224e+00 2.66141862e-01 2.71554030e-02
-1.02091396e+00 -5.33186555e-01 5.52334130e-01 -4.92048681e-01
6.18192673e-01 7.24744856e-01 3.67673337e-01 4.83975381e-01
3.76811415e-01 4.44440812e-01 1.28873158e+00 -2.93665469e-01
5.66662729e-01 2.73354858e-01 2.02262625e-01 1.53604774e-02
5.90734184e-01 2.10622892e-01 -8.35520089e-01 -5.94268560e-01
1.12195201e-01 7.90812373e-02 7.02331215e-02 -1.12155892e-01
-5.63408673e-01 8.51810932e-01 -2.29657665e-02 -1.17697656e-01
-3.66177052e-01 -6.92792162e-02 8.09141159e-01 6.34570301e-01
1.12928629e+00 1.16287768e-01 -5.58610916e-01 -4.15485293e-01
-1.08539867e+00 6.19796097e-01 9.47892725e-01 7.34941423e-01
3.84567589e-01 2.23944694e-01 -1.25770271e-01 8.52870464e-01
-2.08716709e-02 4.66537654e-01 6.62672520e-01 -8.88395965e-01
4.41711605e-01 5.23224711e-01 1.81546062e-01 -7.41055667e-01
-1.85008049e-01 -8.41674626e-01 -1.09820104e+00 5.30435026e-01
6.71528459e-01 -5.84860258e-02 -5.03686249e-01 1.92108607e+00
5.00498116e-01 -9.66309085e-02 -1.78413823e-01 7.55900323e-01
1.06816992e-01 3.40400308e-01 3.94648582e-01 -8.21635664e-01
9.08305407e-01 -3.41070175e-01 -5.21851599e-01 8.49443376e-02
2.58100897e-01 -7.07065642e-01 6.94417059e-01 8.37557554e-01
-1.16269541e+00 -3.32462758e-01 -8.85242403e-01 7.45552719e-01
-2.23968908e-01 -8.60591292e-01 4.86697882e-01 1.05701637e+00
-5.35611570e-01 6.65220022e-01 -2.96379685e-01 1.49542801e-02
6.68256164e-01 3.23152483e-01 3.60350944e-02 -5.09385893e-04
-1.01274240e+00 3.29113305e-01 1.63085744e-01 -3.68425071e-01
-6.95306540e-01 -1.09709978e+00 -2.04721496e-01 -2.93490998e-02
3.98400515e-01 -4.22641814e-01 1.30634642e+00 -1.40440094e+00
-1.01158535e+00 7.75150418e-01 -5.75855449e-02 -6.37442172e-01
1.07469189e+00 -5.70061095e-02 -4.53689158e-01 -3.97049636e-01
3.48303467e-02 1.16167977e-01 7.42902100e-01 -1.39323246e+00
-1.10805941e+00 -5.46375751e-01 -1.45005763e-01 1.84016451e-01
-1.11441083e-01 4.67560738e-02 4.23080266e-01 -8.21782589e-01
-2.33072698e-01 -5.73761821e-01 -2.18532681e-01 -8.55091959e-02
3.04928105e-02 -1.77214831e-01 8.18487406e-01 -3.40340078e-01
1.60498428e+00 -1.86625934e+00 -6.55582488e-01 4.22609717e-01
3.12246174e-01 2.91090935e-01 3.09621722e-01 2.63741046e-01
-6.24797009e-02 2.72384107e-01 -4.13717717e-01 -1.67879865e-01
-1.04358727e-02 1.28889456e-01 -3.47310781e-01 5.07725477e-01
-1.07244216e-01 1.89870700e-01 -8.15687895e-01 -1.73662215e-01
5.02885617e-02 -2.11914226e-01 -5.50603390e-01 3.20613503e-01
-6.81925267e-02 7.56075606e-02 -5.05085103e-02 5.52335858e-01
9.81422365e-01 1.69024393e-01 7.35944137e-02 5.92525899e-01
-2.59991027e-02 -9.55922157e-02 -1.32680488e+00 8.50464523e-01
-2.64581472e-01 3.20185512e-01 9.35085863e-02 -1.30928659e+00
1.06447780e+00 2.61976629e-01 6.46498561e-01 -8.44443440e-01
6.26574904e-02 3.22492957e-01 2.24061236e-01 -7.77062029e-02
5.87848008e-01 -5.87086260e-01 -9.39455330e-02 7.32037783e-01
-5.75194597e-01 1.82134286e-01 1.65039092e-01 1.99605033e-01
1.08672035e+00 -4.81103331e-01 4.04002130e-01 -5.09416997e-01
4.75418597e-01 -8.55381149e-05 8.76135290e-01 9.63660777e-01
-6.47134006e-01 3.04384589e-01 6.55544758e-01 -7.34264970e-01
-9.63053823e-01 -1.11159015e+00 -1.80262178e-01 1.50713825e+00
7.15218410e-02 -8.81350413e-02 -6.05429292e-01 -6.66840613e-01
4.32354450e-01 7.70689547e-01 -4.71548468e-01 -2.86711156e-01
-3.82572502e-01 -1.24911118e+00 4.04207081e-01 2.90238649e-01
4.09881502e-01 -7.72274613e-01 -9.40370798e-01 2.53834903e-01
5.21431305e-02 -6.14117205e-01 -2.29057789e-01 3.39326113e-01
-9.74541843e-01 -1.03302133e+00 -2.07367390e-01 1.34626940e-01
5.19777477e-01 8.18763748e-02 1.49955738e+00 1.48523107e-01
-4.06812549e-01 -2.90030446e-02 -3.22466731e-01 -1.11186039e+00
-4.82518911e-01 -1.39176752e-02 2.67989725e-01 9.75050554e-02
5.72378397e-01 -5.35797179e-01 -8.25281203e-01 3.79489273e-01
-9.84283090e-01 -4.54543173e-01 -8.64441320e-03 8.61355603e-01
4.87762690e-01 2.98622698e-01 1.34782040e+00 -1.41718757e+00
7.86810875e-01 -8.98878276e-01 -4.42138225e-01 8.90989602e-02
-1.24945629e+00 -4.13648844e-01 9.72879529e-01 -4.24052179e-01
-9.47854578e-01 -4.74208862e-01 3.35756272e-01 -3.07293296e-01
-4.66638505e-02 9.49081853e-02 6.18648306e-02 4.29682553e-01
8.31245959e-01 -3.45718898e-02 1.31255895e-01 -2.44637683e-01
-5.51610403e-02 1.03342283e+00 2.92602062e-01 -7.94010460e-01
6.68768406e-01 3.48929971e-01 -1.07073039e-01 -2.61135101e-01
-6.51563346e-01 -3.70159566e-01 -2.92838663e-02 -3.55920404e-01
9.43572521e-02 -9.25101757e-01 -8.61041486e-01 5.47493994e-01
-4.89816934e-01 -4.28775042e-01 -6.94871008e-01 1.60356283e-01
-3.05370539e-01 6.43773079e-02 -3.58993679e-01 -1.36592937e+00
-7.50025570e-01 -5.96958578e-01 4.58025634e-01 4.48750645e-01
-4.81100112e-01 -6.18844926e-01 -4.57616821e-02 3.70068103e-01
7.34364927e-01 4.23993915e-01 8.80538523e-01 -9.50045764e-01
-3.74599956e-02 -4.14436817e-01 -2.85307080e-01 5.12205482e-01
7.30861872e-02 1.40633345e-01 -1.09248054e+00 -5.08934021e-01
-1.06434599e-01 -2.48943180e-01 4.11150217e-01 2.41976038e-01
1.55482149e+00 -3.91232520e-01 2.16184352e-02 2.02398390e-01
1.33278811e+00 6.18194818e-01 3.05996746e-01 2.07823023e-01
1.54879719e-01 6.99129224e-01 7.51427472e-01 1.22045922e+00
2.80527681e-01 3.76425654e-01 6.13638043e-01 2.38477483e-01
2.40577400e-01 8.03512931e-02 1.37464821e-01 5.87896466e-01
7.52579421e-03 -3.77154648e-01 -8.21311772e-01 4.51468229e-01
-1.80066895e+00 -9.96025681e-01 -2.39375740e-01 2.80824471e+00
1.12823772e+00 5.90693057e-01 4.68058079e-01 5.07052422e-01
7.79527128e-01 -1.05633929e-01 -7.83734441e-01 -9.21793818e-01
1.82545036e-01 3.31735253e-01 6.95214629e-01 1.87798858e-01
-9.70747352e-01 3.58574420e-01 5.13316917e+00 9.42561805e-01
-1.28850055e+00 6.60337731e-02 1.37820804e+00 -4.65162605e-01
-1.49426103e-01 -7.31540546e-02 -3.88161808e-01 9.61050928e-01
1.14625251e+00 -7.21971810e-01 2.12059557e-01 8.46608639e-01
2.78487206e-01 -5.99848211e-01 -1.01114941e+00 1.10694706e+00
-1.91001236e-01 -1.00937080e+00 -2.05100209e-01 -1.58302195e-03
6.98927104e-01 -1.30285457e-01 -1.59901250e-02 4.10066187e-01
3.29340577e-01 -9.24535275e-01 8.87427032e-01 6.03872716e-01
9.21714246e-01 -1.15364349e+00 8.16551983e-01 6.81561291e-01
-7.99518764e-01 -4.90997825e-03 -2.72925943e-01 -4.55208063e-01
-1.88031599e-01 1.29161155e+00 -4.70538110e-01 3.33222955e-01
9.86455262e-01 5.65198995e-02 -2.75040269e-01 9.59048390e-01
4.17684913e-01 7.34342754e-01 -2.43340805e-01 -4.95438650e-02
-2.90645301e-01 3.66415642e-02 4.19290006e-01 1.07962656e+00
-2.13828515e-02 2.46563345e-01 2.09480360e-01 3.48959774e-01
-3.11233371e-01 2.85292059e-01 -3.01250726e-01 4.63371694e-01
5.94703496e-01 9.76812601e-01 -6.24863446e-01 -5.31012774e-01
-3.93339582e-02 3.93058717e-01 -8.23990852e-02 2.09070623e-01
-8.53244007e-01 -6.43205285e-01 9.11804676e-01 3.99692625e-01
-2.49822453e-01 3.03934246e-01 -7.38140821e-01 -9.51377869e-01
1.27234131e-01 -1.06094372e+00 6.85668647e-01 -1.36461649e-02
-1.78726637e+00 4.31084186e-01 -6.18457869e-02 -1.31335139e+00
-3.56231004e-01 -1.35200590e-01 -6.06153309e-01 9.28526044e-01
-1.56183064e+00 -3.32490355e-01 -3.48320842e-01 5.34983158e-01
4.28264022e-01 -3.84421796e-01 5.69439888e-01 4.88534749e-01
-4.65906203e-01 9.53425229e-01 3.71000975e-01 -1.43431678e-01
1.03672624e+00 -1.16690767e+00 -8.63895267e-02 6.83113933e-01
-2.69167662e-01 3.39075744e-01 7.48445630e-01 -5.18870831e-01
-7.61952460e-01 -1.18347585e+00 8.92423391e-01 -2.56341934e-01
3.56484920e-01 -3.48415434e-01 -1.17405653e+00 3.03874221e-02
-5.74567467e-02 4.56626058e-01 9.81139839e-01 1.25349954e-01
-5.57472408e-01 -8.27763855e-01 -1.68719244e+00 2.86179900e-01
8.22356343e-01 -1.26476541e-01 -1.83769725e-02 -1.64292082e-02
2.81459808e-01 -3.53762776e-01 -9.01860893e-01 5.04958272e-01
7.77539790e-01 -1.32980609e+00 4.46292043e-01 -7.75508761e-01
3.92484516e-01 1.93075761e-02 -2.25478902e-01 -1.20988023e+00
-8.90980195e-03 -6.29574418e-01 -1.21939220e-01 1.38821030e+00
2.72519052e-01 -1.01037574e+00 8.90496731e-01 9.68038082e-01
2.14745402e-01 -7.43236065e-01 -1.22885931e+00 -5.76948345e-01
2.72691935e-01 -6.54413283e-01 7.47384012e-01 1.21247351e+00
-1.14890538e-01 -1.54110059e-01 -3.21104437e-01 -3.33515257e-01
1.00239825e+00 2.01929301e-01 8.30324352e-01 -1.34825742e+00
-2.77981192e-01 -5.98017514e-01 -2.01574430e-01 -1.35256141e-01
-4.30359133e-02 -6.91423535e-01 -2.36291736e-01 -7.30269909e-01
4.20298040e-01 -1.02233934e+00 -8.11384082e-01 4.13235813e-01
-2.94006258e-01 2.25124657e-01 3.05567473e-01 3.20151538e-01
-5.42075098e-01 2.66752660e-01 5.01629233e-01 1.30810076e-03
-1.84153482e-01 3.10728729e-01 -9.58049417e-01 4.04946655e-01
1.18284285e+00 -8.63217711e-01 -4.85728085e-01 -2.67364131e-03
1.61108166e-01 1.09891459e-01 1.46908671e-01 -9.54146743e-01
-3.28665450e-02 -5.38102686e-01 4.67483461e-01 -2.09849760e-01
-3.68402630e-01 -7.08943665e-01 2.95654535e-01 6.87082648e-01
-4.20262605e-01 6.36676401e-02 1.01238295e-01 6.87645733e-01
-2.03269988e-01 7.89351240e-02 9.49426889e-01 -4.20189910e-02
-3.10809582e-01 2.99277574e-01 -1.09309264e-01 5.85052788e-01
1.33913064e+00 -2.03093469e-01 -5.64668655e-01 -4.96702790e-01
-2.74219960e-01 3.78114849e-01 3.12678397e-01 4.48641717e-01
1.79299027e-01 -9.96664286e-01 -9.05267656e-01 2.22507805e-01
1.36679351e-01 3.13133709e-02 2.82691658e-01 5.30963898e-01
-4.28798616e-01 -1.94086626e-01 -2.40262061e-01 -4.68178093e-01
-1.22280443e+00 2.82969683e-01 3.04573387e-01 -3.11744034e-01
-1.51360715e-02 7.41706789e-01 6.28473237e-02 -1.34654582e-01
3.53657514e-01 6.94447383e-02 2.21319616e-01 3.27618837e-01
5.04814029e-01 7.19146192e-01 3.13343406e-01 -1.50185436e-01
-4.25573558e-01 4.64972816e-02 -1.38785735e-01 5.79720773e-02
1.08031917e+00 -1.24830581e-01 -2.26405054e-01 6.97410762e-01
7.50953019e-01 -1.01281144e-01 -1.11657572e+00 -1.75142199e-01
-6.71679992e-03 -8.31509650e-01 -2.21223190e-01 -1.02263451e+00
-1.12942088e+00 6.78433895e-01 7.72960901e-01 5.37779212e-01
1.39663279e+00 -5.24915397e-01 6.08870268e-01 -3.40630621e-01
6.35278523e-01 -1.33841741e+00 -3.56661826e-01 1.38669744e-01
4.18412834e-01 -1.39221823e+00 2.39606872e-01 2.64881202e-03
-6.21275663e-01 8.96659493e-01 7.79228806e-01 -7.41542578e-02
7.77286649e-01 3.42185855e-01 1.35596424e-01 2.72194594e-01
-1.30492616e+00 3.40593010e-01 -1.63057119e-01 3.52007926e-01
5.59038877e-01 5.48303843e-01 -8.01908135e-01 7.65584648e-01
-9.71662328e-02 1.12248257e-01 5.80814779e-01 1.06679678e+00
-3.15885693e-01 -1.25076878e+00 -3.91875058e-01 1.02765810e+00
-8.93244386e-01 1.61832526e-01 -4.30399366e-02 4.93388295e-01
4.45858926e-01 1.15107298e+00 4.89534765e-01 -1.62234187e-01
4.37115371e-01 2.73944736e-01 -7.27966800e-02 -2.69771665e-01
-1.01035118e+00 -1.38992993e-02 -7.83730205e-03 -3.69248629e-01
-3.89478058e-01 -7.59980679e-01 -1.11208057e+00 -9.16413367e-01
-3.13757323e-02 3.48955035e-01 6.56265676e-01 5.92044771e-01
2.04560697e-01 5.32382667e-01 1.09664536e+00 -2.14433923e-01
-9.40689504e-01 -5.48235774e-01 -9.42045331e-01 6.39552474e-01
2.02129424e-01 -4.72646981e-01 -5.63873827e-01 -8.31552148e-02] | [8.861295700073242, 4.269344806671143] |
ed1eb385-3124-47b4-ab0c-c2f4d7e1363e | does-the-brain-represent-words-an-evaluation | 1806.00591 | null | http://arxiv.org/abs/1806.00591v1 | http://arxiv.org/pdf/1806.00591v1.pdf | Does the brain represent words? An evaluation of brain decoding studies of language understanding | Language decoding studies have identified word representations which can be
used to predict brain activity in response to novel words and sentences
(Anderson et al., 2016; Pereira et al., 2018). The unspoken assumption of these
studies is that, during processing, linguistic information is transformed into
some shared semantic space, and those semantic representations are then used
for a variety of linguistic and non-linguistic tasks. We claim that current
studies vastly underdetermine the content of these representations, the
algorithms which the brain deploys to produce and consume them, and the
computational tasks which they are designed to solve. We illustrate this
indeterminacy with an extension of the sentence-decoding experiment of Pereira
et al. (2018), showing how standard evaluations fail to distinguish between
language processing models which deploy different mechanisms and which are
optimized to solve very different tasks. We conclude by suggesting changes to
the brain decoding paradigm which can support stronger claims of neural
representation. | ['Anna Ivanova', 'Jon Gauthier'] | 2018-06-02 | null | null | null | null | ['brain-decoding', 'brain-decoding'] | ['medical', 'miscellaneous'] | [ 5.27961314e-01 5.99263832e-02 1.89163629e-02 -4.04699355e-01
-2.26833478e-01 -8.60939264e-01 9.82649505e-01 4.76611584e-01
-6.57793283e-01 3.14929485e-01 8.08309019e-01 -7.50973940e-01
6.85510337e-02 -6.21775150e-01 -3.96094739e-01 -2.10062206e-01
2.42928341e-01 2.65935808e-01 -4.83463332e-02 -3.83071184e-01
8.60949039e-01 3.44597876e-01 -1.59011495e+00 3.90677392e-01
7.72926748e-01 4.42526072e-01 5.78770578e-01 6.84905291e-01
-1.98090583e-01 6.19461715e-01 -5.66934407e-01 -2.88862765e-01
-1.48230597e-01 -6.17701888e-01 -9.31581378e-01 -3.31528246e-01
1.82999343e-01 -5.93476482e-02 -2.67413378e-01 1.02472210e+00
3.90271991e-01 2.59661973e-01 7.71372378e-01 -5.11648595e-01
-9.68928099e-01 6.07678294e-01 -1.84429232e-02 7.23496437e-01
5.77956498e-01 3.39928061e-01 9.30220783e-01 -9.89157319e-01
5.35560608e-01 1.29865015e+00 4.24628735e-01 6.75490260e-01
-1.36265361e+00 -5.15794396e-01 2.27869511e-01 1.51278093e-01
-1.07001507e+00 -1.05600119e+00 3.49810183e-01 -6.60981297e-01
1.29678130e+00 2.37232044e-01 8.59327197e-01 1.36615050e+00
5.87236047e-01 2.73546487e-01 1.52190030e+00 -4.47690666e-01
2.38970593e-01 -6.75927028e-02 3.50433946e-01 3.87600780e-01
6.15102291e-01 1.50358737e-01 -9.71962392e-01 2.43134554e-02
4.23835993e-01 -3.52807313e-01 -2.56490558e-01 2.53012836e-01
-1.27972615e+00 7.67514706e-01 1.11790001e-01 9.96333480e-01
-4.81201291e-01 1.90577209e-02 6.31906807e-01 2.64816314e-01
5.24062037e-01 7.83740640e-01 -2.60970533e-01 -2.07900688e-01
-1.03561163e+00 4.63527404e-02 5.47143459e-01 2.51377553e-01
3.31697911e-01 6.71421140e-02 -1.72418863e-01 8.67804468e-01
7.65006125e-01 3.47104877e-01 1.00889444e+00 -5.32122254e-01
3.31872612e-01 5.91330491e-02 -1.51576445e-01 -1.19302404e+00
-5.28327882e-01 -2.69251019e-01 -2.33577546e-02 -2.39073962e-01
4.43082839e-01 -2.60828640e-02 -5.45031607e-01 2.02651572e+00
-3.13457102e-01 -1.31290346e-01 1.24799830e-04 9.50311303e-01
5.21808684e-01 5.86492479e-01 8.06994259e-01 -2.50982255e-01
1.50728714e+00 -4.74104464e-01 -5.43711960e-01 -1.08700407e+00
8.80591393e-01 -5.77172279e-01 1.21773195e+00 2.24764794e-01
-1.31711137e+00 -5.23527503e-01 -1.28721213e+00 -4.18973356e-01
-3.24194729e-01 -3.89896423e-01 1.03275704e+00 7.16023386e-01
-1.36077404e+00 4.93175149e-01 -7.01473892e-01 -7.27649629e-01
4.53891635e-01 -1.15669742e-01 -1.10665753e-01 2.68321596e-02
-1.30036294e+00 1.63686216e+00 5.22499382e-01 2.31395170e-01
-8.78157377e-01 -2.23602295e-01 -7.50788987e-01 7.95967281e-02
-1.77423730e-02 -7.06337452e-01 1.12229621e+00 -1.48770404e+00
-1.24413753e+00 1.50435770e+00 -5.39321303e-01 -2.61362791e-01
-3.31847250e-01 4.67005856e-02 -5.81437051e-01 1.67517632e-01
2.47905135e-01 3.80919158e-01 6.88305080e-01 -9.74137306e-01
-1.35370880e-01 -6.96642518e-01 -3.40888686e-02 3.81532103e-01
-1.69443563e-01 5.11500120e-01 2.95621604e-01 -8.52497220e-01
4.37299520e-01 -6.09201074e-01 1.78403318e-01 -2.04495728e-01
1.10441856e-01 -3.48842025e-01 -1.38220340e-01 -9.16565239e-01
1.15172291e+00 -2.12205982e+00 3.72113138e-01 -1.03977785e-01
3.96525145e-01 6.84101880e-02 -3.51253629e-01 5.90667307e-01
-2.69389838e-01 7.19445765e-01 -4.51606512e-01 -3.77806164e-02
2.67358631e-01 -1.11454790e-02 -5.41447043e-01 7.12888896e-01
1.46312147e-01 1.22414088e+00 -9.61756587e-01 -7.60560334e-02
-2.44833052e-01 3.88862520e-01 -3.45470548e-01 -1.64358199e-01
-9.16546285e-02 3.33505303e-01 -2.55307019e-01 3.37991059e-01
4.16440904e-01 4.20838548e-03 4.63452399e-01 9.17066336e-02
-2.68425554e-01 1.11322033e+00 -6.30470455e-01 1.72035575e+00
-3.70885074e-01 1.20218790e+00 8.63225684e-02 -1.33873177e+00
7.46276855e-01 3.22059631e-01 -2.00321198e-01 -8.74419153e-01
5.95179141e-01 2.33053535e-01 7.20027149e-01 -6.00145817e-01
2.94237196e-01 -5.26984751e-01 -1.00869000e-01 1.04296172e+00
3.35080288e-02 -1.45963445e-01 1.11800894e-01 9.29669663e-02
1.00151622e+00 6.24866076e-02 3.47887248e-01 -7.20987380e-01
4.78906065e-01 -1.36535376e-01 1.99081585e-01 7.68225610e-01
-3.98863733e-01 1.75389588e-01 5.92203259e-01 -2.68670082e-01
-8.28944325e-01 -1.14690244e+00 -3.90587300e-01 1.46571589e+00
-1.67326465e-01 -2.87585676e-01 -7.78915644e-01 -5.75028136e-02
-3.37251067e-01 1.43673754e+00 -6.24944448e-01 -5.83697081e-01
-3.68747979e-01 -7.16368377e-01 6.35367453e-01 2.38212466e-01
-1.12589538e-01 -1.34816027e+00 -1.17828429e+00 2.82596380e-01
-1.04894996e-01 -8.27273369e-01 -1.93719178e-01 1.27213120e-01
-1.01931822e+00 -9.12332177e-01 9.98521820e-02 -7.32133985e-01
7.72476494e-01 5.17324150e-01 1.04684234e+00 4.07364845e-01
-1.46353573e-01 5.81224084e-01 -3.04765016e-01 -3.80975932e-01
-5.70730269e-01 -8.79500210e-02 -7.47419000e-02 -2.17087820e-01
5.81510305e-01 -5.11197090e-01 -4.31598306e-01 -3.74650896e-01
-1.01136184e+00 2.22097889e-01 5.36195874e-01 5.86419821e-01
1.59305647e-01 -4.48885292e-01 8.01243603e-01 -7.69619644e-01
1.22339308e+00 -9.61296797e-01 -8.66296887e-02 1.69556096e-01
-6.34495914e-01 1.18836947e-01 4.20593262e-01 -2.62564480e-01
-9.52450395e-01 -6.90678835e-01 -1.08137757e-01 5.12087226e-01
-2.28723645e-01 6.96238279e-01 2.29296714e-01 1.50778905e-01
7.21219420e-01 5.55951476e-01 1.26791358e-01 -1.11483052e-01
2.92154461e-01 4.43872660e-01 2.05751732e-01 -9.60266173e-01
5.25957644e-01 3.02654535e-01 -5.18298924e-01 -7.22243011e-01
-8.31033349e-01 1.20913044e-01 -7.55042017e-01 -1.93005145e-01
1.08211708e+00 -9.77448761e-01 -5.22881567e-01 1.91122726e-01
-1.36738789e+00 -3.03451002e-01 -1.21840447e-01 4.92785573e-01
-6.07284427e-01 1.69404387e-01 -4.83397573e-01 -7.85943866e-01
-1.75701499e-01 -1.03299391e+00 8.15182269e-01 -1.73829924e-02
-8.19143891e-01 -1.14414430e+00 -1.00638568e-01 3.51911247e-01
6.50354981e-01 -5.70135117e-02 1.18982100e+00 -8.84062171e-01
-5.03639095e-02 2.34179974e-01 -1.71402425e-01 -1.05799809e-01
-2.05186456e-01 -3.15762728e-01 -9.65129495e-01 -1.51924640e-01
7.50914872e-01 -4.08318907e-01 9.49214280e-01 1.20886162e-01
1.10634351e+00 -1.67737290e-01 -1.84352249e-01 4.42573458e-01
1.28958941e+00 1.54601976e-01 5.26309311e-01 1.86842903e-01
1.27907246e-01 1.07524109e+00 2.98344456e-02 -1.72408238e-01
5.22564709e-01 3.75364870e-01 -4.61281650e-02 6.42504334e-01
1.13126457e-01 -1.73325732e-01 9.02688086e-01 9.98888969e-01
2.69574910e-01 -2.69091249e-01 -1.04208040e+00 6.47648215e-01
-1.50520849e+00 -1.15633452e+00 -1.52858317e-01 2.11735082e+00
9.29288745e-01 1.50109291e-01 -3.37288409e-01 -1.59157276e-01
6.38712406e-01 5.28634012e-01 -3.97145778e-01 -1.04710555e+00
-1.60598874e-01 6.08123839e-01 1.69663697e-01 5.65634727e-01
-3.98691475e-01 1.25623512e+00 6.93704176e+00 3.27284634e-01
-1.15152371e+00 4.52000409e-01 4.77237433e-01 -1.20575530e-02
-6.58511817e-01 1.05397470e-01 -1.67976499e-01 4.26935613e-01
1.65174913e+00 -6.14417553e-01 9.95355189e-01 1.52944848e-01
3.18495750e-01 -4.51254070e-01 -1.18975627e+00 6.97284639e-01
5.71989536e-01 -1.06953204e+00 -2.48253513e-02 -2.92959601e-01
9.83053744e-02 1.17415346e-01 -1.89595446e-02 1.29175678e-01
-8.67413078e-03 -1.47877491e+00 1.02160168e+00 6.52349353e-01
4.32726592e-01 -5.59470691e-02 2.45399743e-01 3.28168899e-01
-6.34357274e-01 2.82698162e-02 -4.23042178e-01 -6.40592217e-01
2.51755387e-01 8.54301080e-02 -4.04043466e-01 -2.87218273e-01
2.56594539e-01 4.79323864e-01 -8.56894553e-01 3.69648576e-01
-7.02990174e-01 8.02518308e-01 1.02888659e-01 -2.92670131e-01
3.33434679e-02 -5.21178506e-02 5.79433501e-01 1.12666607e+00
2.09202185e-01 1.29916281e-01 -2.98608750e-01 1.22232318e+00
1.92813128e-01 4.21909481e-01 -7.51282811e-01 -5.48174977e-01
7.76555181e-01 1.06301618e+00 -1.04623270e+00 -3.06594342e-01
-4.62748706e-01 9.09382999e-01 4.62942153e-01 3.28635275e-01
-5.84458649e-01 8.27320442e-02 6.84240460e-01 2.99526248e-02
-1.00500949e-01 -6.04659438e-01 -7.37073421e-01 -1.28179991e+00
-8.92265290e-02 -7.25086987e-01 -1.14121102e-01 -1.03827858e+00
-1.17635059e+00 2.75860041e-01 1.69968624e-02 -1.00751474e-01
-9.40155536e-02 -8.49114656e-01 -5.10371268e-01 1.35246539e+00
-1.21024680e+00 -6.90835536e-01 1.81709915e-01 3.42539549e-01
7.24372685e-01 6.01409152e-02 1.02707016e+00 -1.98936850e-01
-5.26466966e-01 5.78489155e-02 -1.95441097e-01 -3.86727192e-02
4.50852782e-01 -6.60509109e-01 5.31882703e-01 8.69242311e-01
3.92164379e-01 1.10244906e+00 6.64710283e-01 -6.40255749e-01
-1.62802410e+00 -4.73926038e-01 1.32943594e+00 -7.84360766e-01
6.51840031e-01 -7.19277203e-01 -1.03948522e+00 7.60929525e-01
5.06847262e-01 -3.14785331e-01 1.00954890e+00 4.42244820e-02
-4.21526730e-01 4.92365062e-01 -9.78695273e-01 9.33082879e-01
1.38330567e+00 -9.10224140e-01 -1.20219636e+00 3.90342683e-01
6.11146688e-01 3.86639498e-02 -6.20499611e-01 -2.49346942e-01
5.57504177e-01 -8.67722750e-01 8.07362199e-01 -9.91496980e-01
4.15575355e-01 -1.18939579e-02 -2.00958923e-01 -1.46865451e+00
-5.54024696e-01 -9.65387523e-02 2.56940752e-01 1.00370252e+00
4.98969257e-01 -1.09966099e+00 -5.76566458e-02 5.72505057e-01
-2.67561525e-01 -4.30703789e-01 -9.87844229e-01 -2.34745458e-01
4.26153958e-01 -7.17837751e-01 2.62890011e-01 1.08257878e+00
1.33815274e-01 6.21657014e-01 4.46433902e-01 -1.28446624e-01
1.70056805e-01 -4.38135087e-01 1.56421334e-01 -1.32607353e+00
-1.42524615e-02 -9.32457268e-01 -6.81789964e-02 -5.59452951e-01
6.56142175e-01 -1.51358378e+00 1.88150797e-02 -1.62200522e+00
1.12845920e-01 -7.81284943e-02 -3.67234915e-01 4.13016558e-01
-1.37347117e-01 -7.13131856e-03 4.35261548e-01 2.93000340e-01
-2.46062368e-01 2.36112699e-01 8.06819558e-01 1.89475223e-01
9.89459008e-02 -8.41421247e-01 -1.59674263e+00 6.88940167e-01
9.76263165e-01 -4.42812502e-01 -3.97150815e-01 -9.05804813e-01
6.52264774e-01 -3.61686170e-01 4.71637219e-01 -6.35743439e-01
9.78502259e-02 -3.48894715e-01 5.14089167e-01 2.24982843e-01
-2.30718348e-02 -3.80602896e-01 -7.06082880e-02 6.84650242e-01
-6.27649605e-01 5.16168237e-01 5.47419727e-01 1.96145460e-01
1.38980046e-01 -6.65725470e-01 7.12321937e-01 -3.24038714e-01
-7.60048926e-01 -2.84430236e-01 -8.26963067e-01 4.52401906e-01
9.08781469e-01 -4.51659501e-01 -4.97775763e-01 -1.94233254e-01
-6.50963068e-01 -7.02947378e-02 3.33228201e-01 6.07475936e-01
6.43551469e-01 -1.15667462e+00 -8.02638412e-01 1.34688050e-01
-6.14215024e-02 -5.62692583e-01 5.16375862e-02 9.98391509e-01
-5.44911981e-01 5.30058503e-01 -2.61555821e-01 -9.31074959e-04
-4.20122474e-01 5.27620614e-01 2.56566674e-01 1.74403518e-01
-4.90832299e-01 6.71238840e-01 2.12269932e-01 6.85514417e-03
-4.71106708e-01 -6.92422092e-02 -2.82630295e-01 6.02162838e-01
5.16198993e-01 -7.73754790e-02 -1.53000638e-01 -9.46420431e-01
-4.28286046e-01 2.16890991e-01 5.53615689e-02 -2.49046549e-01
1.31972861e+00 -2.29580775e-01 -6.91214204e-01 9.21980858e-01
8.22622418e-01 1.68089226e-01 -7.26767004e-01 7.33910948e-02
2.07939342e-01 -2.92106450e-01 5.04986346e-02 -7.72784054e-01
-5.01995623e-01 1.07011867e+00 2.19023943e-01 2.04170227e-01
8.69016826e-01 2.81644259e-02 3.83379579e-01 6.95909485e-02
3.03338856e-01 -1.35124516e+00 -2.16091976e-01 6.89363837e-01
1.24902725e+00 -7.67100930e-01 -5.95365558e-03 1.08196512e-01
-4.65165198e-01 1.30089414e+00 5.18121839e-01 -2.68925160e-01
3.22404176e-01 -1.01336345e-01 -3.19079638e-01 -3.52635801e-01
-1.10211170e+00 -2.37659186e-01 1.38011754e-01 6.04292333e-01
1.03643274e+00 1.42453060e-01 -1.17025554e+00 7.79181778e-01
-5.60249627e-01 -4.75603074e-01 5.03103137e-01 7.10368514e-01
-8.00485015e-01 -8.87582839e-01 -6.58594072e-01 7.05462456e-01
-3.62086624e-01 -5.48695385e-01 -5.28323948e-01 5.58016062e-01
1.24188691e-01 1.04365730e+00 3.00123572e-01 -1.84501093e-02
4.30992246e-02 7.26397693e-01 5.79398215e-01 -1.31166649e+00
-7.56725013e-01 -2.91467994e-01 1.82714894e-01 -4.41483736e-01
-2.30221823e-01 -1.08970737e+00 -1.31942201e+00 -1.57804757e-01
8.31405967e-02 -2.93094981e-02 6.67505801e-01 1.28811383e+00
3.68396878e-01 6.68426752e-01 1.25623085e-02 -6.25874043e-01
-4.68768716e-01 -1.07886529e+00 -2.94661105e-01 3.42515290e-01
-3.33977230e-02 -6.84571087e-01 -3.86367321e-01 1.39369890e-02] | [10.262152671813965, 8.52951717376709] |
7974689f-4558-4ebb-8abc-ca084bdff234 | blemish-aware-and-progressive-face-retouching | null | null | http://openaccess.thecvf.com//content/CVPR2023/html/Xie_Blemish-Aware_and_Progressive_Face_Retouching_With_Limited_Paired_Data_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Xie_Blemish-Aware_and_Progressive_Face_Retouching_With_Limited_Paired_Data_CVPR_2023_paper.pdf | Blemish-Aware and Progressive Face Retouching With Limited Paired Data | Face retouching aims to remove facial blemishes, while at the same time maintaining the textual details of a given input image. The main challenge lies in distinguishing blemishes from the facial characteristics, such as moles. Training an image-to-image translation network with pixel-wise supervision suffers from the problem of expensive paired training data, since professional retouching needs specialized experience and is time-consuming. In this paper, we propose a Blemish-aware and Progressive Face Retouching model, which is referred to as BPFRe. Our framework can be partitioned into two manageable stages to perform progressive blemish removal. Specifically, an encoder-decoder-based module learns to coarsely remove the blemishes at the first stage, and the resulting intermediate features are injected into a generator to enrich local detail at the second stage. We find that explicitly suppressing the blemishes can contribute to an effective collaboration among the components. Toward this end, we incorporate an attention module, which learns to infer a blemish-aware map and further determine the corresponding weights, which are then used to refine the intermediate features transferred from the encoder to the decoder, and from the decoder to the generator. Therefore, BPFRe is able to deliver significant performance gains on a wide range of face retouching tasks. It is worth noting that we reduce the dependence of BPFRe on paired training samples by imposing effective regularization on unpaired ones. | ['Hau San Wong', 'Zhiwen Yu', 'Si Wu', 'Zhen Xu', 'Wen Xue', 'Lianxin Xie'] | 2023-01-01 | null | null | null | cvpr-2023-1 | ['image-to-image-translation', 'image-to-image-translation'] | ['computer-vision', 'miscellaneous'] | [ 5.04628658e-01 2.72192091e-01 3.07719670e-02 -2.42102563e-01
-7.78288305e-01 -3.22427005e-01 4.83867288e-01 -2.22912982e-01
-2.44128346e-01 5.92195809e-01 2.55373746e-01 5.01315445e-02
2.90104479e-01 -7.29560971e-01 -8.70236278e-01 -8.67188036e-01
5.01535296e-01 -8.39402899e-03 -1.71907485e-01 -1.66475266e-01
6.08499981e-02 4.38021004e-01 -1.53387666e+00 1.98754638e-01
1.00005722e+00 1.02471697e+00 9.35124829e-02 4.62784648e-01
1.60791297e-02 5.79030454e-01 -3.67195398e-01 -8.27025950e-01
2.73386329e-01 -7.47854292e-01 -4.55852181e-01 3.42609584e-01
8.35656941e-01 -6.11712039e-01 -4.31490272e-01 1.15604472e+00
5.36189854e-01 -5.55525236e-02 4.60414737e-01 -1.00265336e+00
-7.02970088e-01 4.30673510e-01 -9.75686252e-01 6.13252558e-02
1.60237625e-02 3.87006283e-01 7.47509181e-01 -1.11338937e+00
4.89382535e-01 1.31441164e+00 4.83643621e-01 9.05246794e-01
-1.44573092e+00 -9.84868288e-01 -2.72070747e-02 1.71452925e-01
-1.46574128e+00 -9.64700699e-01 8.64667892e-01 -3.30411196e-01
3.67936879e-01 2.19234332e-01 6.13905132e-01 1.00297678e+00
1.08975329e-01 5.80014706e-01 8.62058043e-01 -2.31944546e-01
-1.02761947e-02 5.69225587e-02 -2.04834744e-01 9.22828972e-01
1.17492795e-01 9.51434672e-02 -5.76369524e-01 1.80534795e-01
8.99015427e-01 1.70832366e-01 -5.03994882e-01 -8.76588672e-02
-7.64819860e-01 6.73833370e-01 7.02468097e-01 1.69763863e-01
-2.87267089e-01 1.11336157e-01 4.42983694e-02 2.47854054e-01
6.95509493e-01 3.17202061e-01 -2.74453938e-01 2.27149591e-01
-1.14915764e+00 -1.33831546e-01 2.62183994e-01 5.88909268e-01
1.19993472e+00 -2.43465737e-01 -5.09498417e-01 1.03231764e+00
2.14350656e-01 3.11165959e-01 2.48780861e-01 -8.04441690e-01
4.03942436e-01 4.71900731e-01 -1.61003485e-01 -8.85179102e-01
3.70527953e-02 -4.19878513e-01 -9.88823771e-01 2.93127686e-01
2.47997805e-01 -2.16851145e-01 -1.24238741e+00 1.95056295e+00
3.53379250e-01 4.35471833e-01 -2.40758792e-01 9.33304131e-01
7.50760853e-01 5.86747587e-01 1.51990131e-01 -3.62420440e-01
1.34100890e+00 -1.13946915e+00 -6.60455704e-01 -4.16466206e-01
1.81017339e-01 -6.83376491e-01 1.03834653e+00 5.05329296e-02
-1.32557511e+00 -4.42723006e-01 -9.42314446e-01 -4.59770948e-01
1.27603278e-01 1.82512090e-01 1.73414573e-01 3.73193115e-01
-1.27952051e+00 4.86811697e-01 -6.26835048e-01 -2.24878147e-01
8.89229357e-01 3.77894312e-01 -5.09567499e-01 -2.38785833e-01
-1.00063729e+00 7.49670386e-01 1.49266854e-01 3.27043116e-01
-1.03282142e+00 -8.51948559e-01 -1.03530419e+00 2.21408293e-01
3.91849190e-01 -8.57484698e-01 9.95094180e-01 -1.40171516e+00
-1.67764235e+00 9.35463488e-01 -5.26479840e-01 -9.35710594e-03
6.20673716e-01 9.93123576e-02 -2.48067170e-01 1.36993811e-01
-9.95322540e-02 9.29856181e-01 1.56929266e+00 -1.24620438e+00
-5.57252884e-01 -3.08816463e-01 -8.76538903e-02 2.18913019e-01
-5.96113443e-01 -1.29192531e-01 -1.05412376e+00 -7.55857766e-01
-5.24564162e-02 -8.25163126e-01 1.22813419e-01 1.91769451e-01
-4.19606388e-01 -2.20418051e-02 5.26665866e-01 -9.29687083e-01
1.08572602e+00 -2.31760573e+00 4.29601073e-01 1.11610621e-01
7.40940452e-01 3.35988462e-01 -5.93228161e-01 2.06315666e-02
-2.57701337e-01 5.49672991e-02 -3.37329537e-01 -5.27856469e-01
-3.97126794e-01 -1.46774665e-01 -2.26792037e-01 4.68460560e-01
6.03946447e-01 1.21797848e+00 -7.82448947e-01 -3.35288525e-01
5.80623299e-02 6.40652001e-01 -6.75074041e-01 3.06904823e-01
-1.64709985e-01 5.60359180e-01 -1.45816594e-01 3.59166026e-01
9.34012651e-01 -2.60005563e-01 -2.50194985e-02 -4.29544896e-01
2.36129574e-02 1.09027930e-01 -5.49974740e-01 1.62760854e+00
-5.34842789e-01 6.64033353e-01 1.93610951e-01 -6.25356495e-01
7.60395110e-01 2.01038614e-01 2.57594734e-01 -7.28997707e-01
2.97338456e-01 1.58237681e-01 -1.79959685e-01 -4.60662514e-01
3.52762491e-01 -4.70942259e-01 2.82682240e-01 4.12173122e-01
1.45496503e-01 1.59463137e-01 -5.15012033e-02 1.25552475e-01
1.06550848e+00 -2.11986467e-01 7.71693559e-03 -5.45624606e-02
6.27092719e-01 -5.73538184e-01 5.99219501e-01 6.17026947e-02
-1.05105847e-01 8.07713091e-01 5.03863633e-01 -1.70111537e-01
-1.03532243e+00 -1.00855207e+00 6.68723509e-02 9.86123383e-01
2.91765422e-01 -2.83417761e-01 -1.01835048e+00 -8.58977735e-01
-9.95252430e-02 3.63273978e-01 -8.57963324e-01 -5.95243275e-01
-3.57153505e-01 -5.42480588e-01 4.03761178e-01 2.22272605e-01
6.35692000e-01 -9.83560205e-01 1.69809200e-02 -8.18617418e-02
-2.42848545e-01 -7.57857263e-01 -1.02872515e+00 -2.52574772e-01
-6.92582250e-01 -8.47767651e-01 -9.34397161e-01 -7.86729693e-01
1.15816379e+00 4.89953250e-01 8.39524150e-01 4.00477171e-01
-2.78621584e-01 -7.07873628e-02 -9.51484442e-02 -9.00943950e-02
-3.59857529e-01 1.71619728e-01 -4.04908031e-01 5.55758953e-01
1.71582177e-01 -5.52391708e-01 -8.09438229e-01 2.37019792e-01
-9.72143292e-01 3.34063977e-01 7.93534338e-01 9.48574960e-01
4.55451071e-01 1.10447846e-01 4.43848908e-01 -9.11816657e-01
4.29624736e-01 -2.53904223e-01 -3.80415708e-01 2.81064510e-01
-4.37666476e-01 1.61162436e-01 4.36487794e-01 -3.91554236e-01
-1.10072529e+00 6.27916828e-02 -2.32988879e-01 -5.89122832e-01
1.50182694e-01 1.58466265e-01 -5.97614050e-01 -2.75365859e-01
5.12582242e-01 2.56947041e-01 1.10592127e-01 -5.12683809e-01
4.71361339e-01 6.18095219e-01 5.81927896e-01 -1.78042665e-01
1.06754494e+00 4.39321846e-01 -3.96401912e-01 -6.33738697e-01
-9.06257510e-01 -3.77930664e-02 -3.41605037e-01 -2.46474519e-01
7.08817005e-01 -9.98440027e-01 -6.04228497e-01 6.33485854e-01
-1.06317687e+00 -4.98802751e-01 -2.81036139e-01 1.24106347e-03
-2.61580676e-01 2.63051242e-01 -7.00746953e-01 -3.84616315e-01
-4.45032090e-01 -1.09429932e+00 1.10099328e+00 5.03974676e-01
9.97979492e-02 -6.49167538e-01 -2.87918607e-03 4.77103442e-01
2.14088380e-01 -1.37051001e-01 8.78081203e-01 -4.55377065e-02
-7.38884926e-01 -4.66871858e-02 -5.77562749e-01 3.77065361e-01
3.01930755e-01 -5.94565459e-02 -1.23511052e+00 -5.73653758e-01
-1.53421341e-02 -3.23657453e-01 1.27986634e+00 1.86103672e-01
1.31031764e+00 -4.92846757e-01 -1.66115880e-01 8.77570450e-01
1.09342682e+00 -1.63026005e-01 1.02118337e+00 -7.09107444e-02
9.27777827e-01 7.14045823e-01 2.50900954e-01 1.17694147e-01
3.95220906e-01 6.10783100e-01 3.99958193e-01 -3.51644248e-01
-6.62159324e-01 -5.73718011e-01 5.27679920e-01 6.35980785e-01
1.93128034e-01 -2.72409201e-01 -4.74565178e-01 4.22766536e-01
-1.55565190e+00 -9.10238385e-01 3.45748574e-01 2.21026063e+00
1.18390954e+00 -2.13682175e-01 -1.36512876e-01 -1.75010245e-02
1.04350471e+00 2.27878302e-01 -7.20983744e-01 -4.23641130e-02
5.92523292e-02 2.22287163e-01 2.87530541e-01 7.14177608e-01
-8.97976398e-01 1.24488330e+00 5.20548820e+00 1.03105557e+00
-1.36235464e+00 2.48490274e-01 9.27392185e-01 -3.07989240e-01
-5.95223010e-01 -1.16404392e-01 -6.61186159e-01 6.81753278e-01
6.05587065e-01 -1.50923552e-02 9.31094348e-01 2.97924250e-01
1.92883119e-01 9.30907652e-02 -1.06335390e+00 1.21420217e+00
3.54197115e-01 -1.20064700e+00 1.95033535e-01 5.33521688e-03
8.93931985e-01 -2.87721336e-01 2.87225395e-01 -1.24425357e-02
5.74226230e-02 -1.16383827e+00 6.64527655e-01 5.75987518e-01
1.37385893e+00 -8.34643722e-01 3.84735823e-01 -4.36590128e-02
-1.05215764e+00 3.10737863e-02 -3.87529314e-01 3.80376667e-01
-2.70146858e-02 8.11215937e-01 -6.55257523e-01 2.96035677e-01
4.50464427e-01 8.00284386e-01 -5.82241297e-01 1.00670576e+00
-5.52672267e-01 2.42184073e-01 -6.49924129e-02 4.46760774e-01
-1.87684968e-01 -1.12428531e-01 3.48526835e-01 8.56972456e-01
4.69016194e-01 4.08211630e-03 -2.79576123e-01 1.01689672e+00
-6.10498428e-01 1.93642769e-02 -2.42329717e-01 -6.39849305e-02
4.19846117e-01 1.50037432e+00 -4.72105145e-01 -2.02074841e-01
-3.54883343e-01 1.42666817e+00 5.40406346e-01 5.24920821e-01
-7.45541692e-01 -3.59142095e-01 8.94274771e-01 2.94733375e-01
3.91380101e-01 1.65283829e-01 -3.91303271e-01 -1.19876134e+00
2.57899687e-02 -9.01690662e-01 2.07457453e-01 -6.87766612e-01
-1.24117064e+00 7.44271934e-01 -6.12370074e-01 -9.36318576e-01
9.71373022e-02 -1.78088024e-01 -6.64478958e-01 1.13622892e+00
-1.73355103e+00 -1.27288687e+00 -7.03945458e-01 6.91686273e-01
4.30751920e-01 1.00999326e-01 3.69538069e-01 5.04308581e-01
-1.06639993e+00 1.06787741e+00 -2.95602530e-01 1.29806742e-01
1.03594649e+00 -8.26076865e-01 3.49104345e-01 1.01237178e+00
1.97510701e-02 5.60069799e-01 5.28975725e-01 -6.73798740e-01
-1.07211637e+00 -1.45301151e+00 9.65802252e-01 -2.60630161e-01
3.40890348e-01 -4.13274437e-01 -1.06280684e+00 5.53458452e-01
6.05853088e-02 3.43104042e-02 3.54217917e-01 -1.69686064e-01
-6.46847308e-01 -4.11391407e-01 -9.95894849e-01 8.32749665e-01
1.04206586e+00 -7.60384262e-01 -2.60691792e-01 -9.77753475e-02
5.99103749e-01 -2.12337777e-01 -5.73742211e-01 2.77148515e-01
4.83386219e-01 -9.13241684e-01 7.69123495e-01 -2.23302081e-01
7.08846927e-01 -2.27342844e-01 4.88075048e-01 -1.40813088e+00
-4.47517514e-01 -8.85469496e-01 -1.87043417e-02 1.52473879e+00
4.45595354e-01 -5.50362349e-01 8.29814911e-01 7.31234252e-01
7.31637180e-02 -4.49279159e-01 -7.85023332e-01 -2.27986008e-01
5.92766004e-03 3.77246132e-03 7.00307667e-01 8.42863500e-01
-1.25659838e-01 5.66556096e-01 -6.18029475e-01 7.90455565e-02
4.97653604e-01 -9.47334059e-03 7.16236949e-01 -1.02540886e+00
-2.08682880e-01 -4.87767726e-01 -1.33280486e-01 -1.12804759e+00
1.97969988e-01 -9.48188066e-01 3.15143228e-01 -1.09501481e+00
5.89360774e-01 -3.12720418e-01 -2.30086282e-01 6.41238868e-01
-6.63651645e-01 8.14954460e-01 1.20468140e-01 2.73585886e-01
-2.81618655e-01 8.29604387e-01 1.62937653e+00 -2.50080794e-01
-2.69022107e-01 -1.50476187e-01 -1.17933786e+00 5.47719061e-01
6.50673568e-01 -3.73979956e-01 -4.68421727e-01 -6.80713594e-01
1.37950107e-01 -4.63394001e-02 5.24253607e-01 -7.04044640e-01
1.08398624e-01 -4.11902927e-02 4.31782424e-01 -3.82460356e-02
3.43691319e-01 -6.15275860e-01 1.65870354e-01 2.60522634e-01
-4.20909137e-01 -3.74395818e-01 2.07707524e-01 5.99763870e-01
-9.94178206e-02 -4.80830409e-02 1.18344164e+00 2.04679936e-01
-3.63340974e-01 8.02049816e-01 -7.82650039e-02 -1.17570587e-01
9.97422934e-01 -6.66524842e-02 -4.46844190e-01 -4.04022127e-01
-5.56910157e-01 5.20690382e-02 9.08236980e-01 3.86530250e-01
6.13169789e-01 -1.36435032e+00 -8.05211306e-01 6.90611959e-01
8.05246234e-02 -5.11545166e-02 5.97883582e-01 9.12725568e-01
-3.92747343e-01 -1.45491119e-02 -2.32580364e-01 -2.95481592e-01
-1.31582248e+00 6.36070788e-01 4.21396464e-01 -2.09141821e-02
-6.59299433e-01 1.15536559e+00 7.78982997e-01 1.92649350e-01
2.24383518e-01 1.87574383e-02 -1.13557763e-01 1.29033625e-01
8.50677848e-01 1.83284804e-01 1.35222614e-01 -7.04624891e-01
-1.00401707e-01 7.08283186e-01 -4.41990256e-01 -9.58892629e-02
1.16857207e+00 -4.18784350e-01 -3.02796751e-01 -1.05473116e-01
1.41963482e+00 1.09619349e-01 -1.83376217e+00 -3.28223377e-01
-3.67866755e-01 -6.40514255e-01 1.89770818e-01 -7.83843577e-01
-1.66617048e+00 9.77585971e-01 5.43762445e-01 -2.65941292e-01
1.57797849e+00 8.84998366e-02 1.09473073e+00 -1.56267002e-01
8.23549703e-02 -6.84011877e-01 2.23158479e-01 1.78372234e-01
9.36760485e-01 -1.04282260e+00 -3.69075239e-01 -5.84627569e-01
-5.68877935e-01 8.50614965e-01 6.35937393e-01 4.11872715e-02
4.34507757e-01 1.40536606e-01 1.28482571e-02 -5.64752221e-02
-7.65535355e-01 -2.49910429e-01 4.47363824e-01 4.46990222e-01
2.05259398e-01 -1.88725203e-01 -6.14649765e-02 4.28960472e-01
-1.09312110e-01 1.21274859e-01 1.87204957e-01 2.40662426e-01
-4.01994199e-01 -1.07793725e+00 -2.93682814e-01 4.87153471e-01
-3.69598508e-01 -2.30528459e-01 -4.67847437e-01 2.36343592e-01
3.54951531e-01 8.76905859e-01 7.67299309e-02 -5.62595606e-01
1.97192520e-01 -9.69276354e-02 5.87893665e-01 -7.64610410e-01
-5.23164570e-01 1.86289206e-01 -2.83127993e-01 -6.21554554e-01
-7.70789012e-02 -3.16440076e-01 -8.23684096e-01 -6.67027593e-01
-2.32888892e-01 -2.97800750e-02 1.62875339e-01 8.44999969e-01
5.75135648e-01 5.55954695e-01 8.40108335e-01 -9.01750565e-01
-1.51266426e-01 -8.90237987e-01 -5.51661015e-01 5.66862166e-01
6.19793296e-01 -5.39893568e-01 -2.81732023e-01 1.95375472e-01] | [12.684867858886719, -0.11263102293014526] |
952059a8-3ab0-4e49-bf23-d193b9655bb0 | dasvdd-deep-autoencoding-support-vector-data | 2106.05410 | null | https://arxiv.org/abs/2106.05410v2 | https://arxiv.org/pdf/2106.05410v2.pdf | DASVDD: Deep Autoencoding Support Vector Data Descriptor for Anomaly Detection | Semi-supervised anomaly detection aims to detect anomalies from normal samples using a model that is trained on normal data. With recent advancements in deep learning, researchers have designed efficient deep anomaly detection methods. Existing works commonly use neural networks to map the data into a more informative representation and then apply an anomaly detection algorithm. In this paper, we propose a method, DASVDD, that jointly learns the parameters of an autoencoder while minimizing the volume of an enclosing hyper-sphere on its latent representation. We propose an anomaly score which is a combination of autoencoder's reconstruction error and the distance from the center of the enclosing hypersphere in the latent representation. Minimizing this anomaly score aids us in learning the underlying distribution of the normal class during training. Including the reconstruction error in the anomaly score ensures that DASVDD does not suffer from the common hypersphere collapse issue since the DASVDD model does not converge to the trivial solution of mapping all inputs to a constant point in the latent representation. Experimental evaluations on several benchmark datasets show that the proposed method outperforms the commonly used state-of-the-art anomaly detection algorithms while maintaining robust performance across different anomaly classes. | ['Narges Armanfard', 'Hadi Hojjati'] | 2021-06-09 | null | null | null | null | ['supervised-anomaly-detection'] | ['computer-vision'] | [-1.89999446e-01 3.37656811e-02 5.61014712e-01 -4.00929451e-01
-1.17216108e-03 -1.12685606e-01 5.19450128e-01 3.78534377e-01
-2.51852870e-01 1.10137269e-01 -1.59053847e-01 -1.53612480e-01
-2.74365153e-02 -9.40634012e-01 -6.54194772e-01 -9.91469145e-01
-1.48099899e-01 7.42881000e-01 2.04054397e-02 -5.82111739e-02
3.11223805e-01 7.79298007e-01 -1.49051690e+00 -1.03439786e-01
1.05627286e+00 1.22133243e+00 -4.82076645e-01 3.66305172e-01
-3.51686358e-01 3.52902949e-01 -8.66089880e-01 -1.65454537e-01
6.34370744e-01 -3.90964419e-01 -2.89536148e-01 -6.41446700e-03
5.77808440e-01 -4.65812117e-01 -2.12664604e-01 1.42093992e+00
8.11065808e-02 3.99356663e-01 1.07474160e+00 -1.63027215e+00
-5.64964592e-01 1.01463951e-01 -5.77649713e-01 4.02058750e-01
-4.16301973e-02 -2.38534600e-01 9.05932248e-01 -9.12318885e-01
1.37289584e-01 1.09507680e+00 8.24094534e-01 4.29714501e-01
-1.17013299e+00 -3.89686555e-01 6.92954883e-02 1.24314442e-01
-1.43941832e+00 1.09260432e-01 9.23998892e-01 -7.04474866e-01
8.90734136e-01 1.86053216e-01 6.12114012e-01 7.92360544e-01
3.30184132e-01 5.52629113e-01 4.58631873e-01 -4.18285787e-01
5.34323990e-01 -7.61452504e-03 5.92048541e-02 8.13219488e-01
5.17797232e-01 -8.41680393e-02 6.84081092e-02 -6.03242338e-01
7.50733972e-01 4.66873139e-01 -1.48049220e-01 -7.48726189e-01
-6.94760859e-01 1.14293540e+00 3.09050053e-01 3.04990202e-01
-5.16687334e-01 -2.78131515e-01 4.21746135e-01 3.63346666e-01
5.92596769e-01 4.58682418e-01 -9.41045731e-02 1.68510661e-01
-6.87342823e-01 2.15360001e-01 7.20435202e-01 3.09832573e-01
6.13070786e-01 5.44015944e-01 1.61786273e-01 7.92975545e-01
5.28187156e-01 3.04561704e-01 8.75173569e-01 -5.35766721e-01
1.47934958e-01 1.19407499e+00 -7.92418793e-02 -1.24568152e+00
-2.69462407e-01 -3.15127820e-01 -9.46887016e-01 5.07262826e-01
5.44290543e-01 -1.55661613e-01 -8.23794127e-01 1.35560858e+00
5.69162369e-01 6.06957018e-01 2.59415567e-01 9.51404512e-01
2.76677072e-01 7.89133310e-01 -2.49175236e-01 2.01847836e-01
7.69458055e-01 -6.42361343e-01 -5.12502432e-01 -1.53237954e-01
6.80733621e-01 -3.61137033e-01 8.80214810e-01 4.47961003e-01
-6.33051336e-01 -2.08529547e-01 -1.38723195e+00 3.93224746e-01
-5.20439088e-01 4.88030687e-02 3.28079045e-01 3.54292810e-01
-6.76980078e-01 8.10444832e-01 -1.11148989e+00 -2.63251245e-01
2.09296331e-01 2.05326840e-01 -4.77810383e-01 2.89408356e-01
-8.92341733e-01 7.76192248e-01 7.08130896e-01 -1.49680182e-01
-7.10510969e-01 -5.08900106e-01 -9.58334148e-01 1.15657248e-01
-6.17796881e-03 -1.02589376e-01 8.57087910e-01 -1.06593847e+00
-1.25514686e+00 6.37780488e-01 3.26422513e-01 -8.20870578e-01
3.71985793e-01 -4.16344315e-01 -7.07810819e-01 1.19181938e-01
-1.62570670e-01 9.51985046e-02 1.28017151e+00 -8.86537015e-01
-4.42353278e-01 -6.80132806e-01 -6.08608663e-01 4.01829481e-02
-7.66935289e-01 -4.20219243e-01 9.71716717e-02 -7.38455951e-01
6.69665158e-01 -7.55153418e-01 -9.79325250e-02 5.14176860e-02
-4.04957950e-01 -5.61367691e-01 1.44730461e+00 -3.80746454e-01
1.15550745e+00 -2.36358523e+00 -8.49453732e-02 7.72473395e-01
3.35175693e-01 4.51966166e-01 1.46652699e-01 1.56217724e-01
-4.43368077e-01 -3.73780370e-01 -6.98094368e-01 -2.82605648e-01
-7.59657845e-02 4.19655144e-01 -6.77289903e-01 8.54261339e-01
1.95734188e-01 2.49282897e-01 -5.98215997e-01 -3.49504501e-02
3.12636942e-01 4.84546602e-01 -4.85275358e-01 5.80316305e-01
-1.47585526e-01 1.11183010e-01 -3.69959474e-01 4.43614393e-01
7.40794480e-01 2.81201042e-02 -4.98387218e-01 3.99943709e-01
5.70576563e-02 -1.10592470e-01 -1.38459957e+00 1.30319715e+00
1.99573133e-02 6.75380349e-01 -2.36434191e-01 -1.47353745e+00
1.48742366e+00 1.41914040e-01 7.07932651e-01 -2.95021176e-01
1.51078165e-01 3.34096819e-01 8.25936440e-03 -3.09004605e-01
3.77768278e-02 -4.32598218e-02 3.45349431e-01 6.56762123e-01
1.82236254e-01 2.57219464e-01 -1.35835454e-01 -7.03081861e-02
1.03112924e+00 -3.47553283e-01 1.98279724e-01 -3.30238938e-01
7.61295736e-01 -2.93138564e-01 5.73139369e-01 6.45097494e-01
-2.01835498e-01 6.38091803e-01 6.41861558e-01 -1.07199073e+00
-1.14614081e+00 -1.29753113e+00 -4.44408238e-01 8.53523254e-01
-1.83204040e-01 -2.02687517e-01 -9.97845054e-01 -9.29238319e-01
2.06988439e-01 1.04790986e+00 -7.11668253e-01 -6.72830641e-01
-4.65449065e-01 -8.71758342e-01 7.01325059e-01 5.18487930e-01
4.51861590e-01 -1.03100932e+00 -6.45030737e-01 5.10582179e-02
1.26456365e-01 -6.89326704e-01 -1.26796439e-01 1.57373846e-01
-1.04120338e+00 -1.06315219e+00 -5.40961027e-01 -6.26173437e-01
1.14353299e+00 -2.78348356e-01 7.13412881e-01 7.58507177e-02
-3.75667691e-01 3.26472223e-01 -1.96264401e-01 -7.44024336e-01
-4.24130321e-01 -4.92969036e-01 4.58555907e-01 5.12096107e-01
1.02261281e+00 -6.23705566e-01 -3.99864018e-01 2.72608072e-01
-1.03562701e+00 -6.50637567e-01 1.44317493e-01 7.17633188e-01
7.79014885e-01 2.25206330e-01 4.10261780e-01 -4.49306428e-01
5.27844131e-01 -8.00996959e-01 -9.24643934e-01 -1.12470180e-01
-8.11382532e-01 4.08594310e-01 8.21022689e-01 -2.96745479e-01
-6.54216528e-01 -1.48058489e-01 -1.40640393e-01 -8.92445147e-01
-4.82086957e-01 1.60375431e-01 6.30368292e-02 1.01806276e-01
8.59455466e-01 4.03618306e-01 3.23745757e-01 -6.15060806e-01
-1.30089432e-01 5.48168719e-01 6.12774491e-01 -3.32808882e-01
8.87648404e-01 5.80695212e-01 1.90098047e-01 -8.92595291e-01
-7.58327246e-01 -5.54788470e-01 -7.58466840e-01 2.26298347e-02
6.73780560e-01 -4.47911710e-01 -2.81594396e-01 5.91611087e-01
-7.49448240e-01 7.22285509e-02 -6.76170111e-01 5.08074820e-01
-3.32287222e-01 5.51572025e-01 -2.81243771e-01 -7.43682086e-01
-5.56811333e-01 -7.38922060e-01 7.84154356e-01 1.11983143e-01
-1.51575178e-01 -1.17954123e+00 6.38835251e-01 -3.49601895e-01
3.42737079e-01 5.87968528e-01 1.08309507e+00 -1.49500930e+00
2.42804531e-02 -7.70446777e-01 1.35862514e-01 8.77426684e-01
4.67854738e-02 8.28464702e-02 -8.63724411e-01 -3.45432252e-01
4.25744236e-01 8.35855827e-02 9.26813245e-01 2.82894522e-01
1.49151742e+00 -5.52518964e-01 -9.85167101e-02 9.07471359e-01
1.31168115e+00 3.57358843e-01 4.89279628e-01 5.84016562e-01
7.07196295e-01 3.79409790e-01 2.85412908e-01 6.46316469e-01
-1.16743088e-01 2.56247073e-01 7.51058698e-01 1.58335403e-01
5.56483924e-01 -2.46325172e-02 4.24608111e-01 5.83603680e-01
2.13355348e-01 -1.82066988e-02 -1.24764788e+00 3.43293518e-01
-1.81043661e+00 -8.09079409e-01 4.33871523e-02 2.45972586e+00
2.67945349e-01 2.69967139e-01 8.03523213e-02 4.46513116e-01
6.05677068e-01 -7.21143112e-02 -7.93215990e-01 -8.43960166e-01
1.28213165e-03 -5.62076010e-02 1.19549269e-02 2.77988881e-01
-1.33426321e+00 5.66451013e-01 5.72979021e+00 2.90841877e-01
-1.21168590e+00 -3.63504022e-01 2.05667526e-01 1.28750235e-01
7.76996240e-02 -3.63248080e-01 -6.51461542e-01 4.07839715e-01
8.70995700e-01 -6.86995313e-02 1.04497485e-01 1.35634995e+00
-3.27937961e-01 2.35915378e-01 -1.25940502e+00 8.82529795e-01
4.13298637e-01 -7.88590252e-01 3.29563558e-01 1.19652286e-01
6.43561363e-01 1.64458081e-01 2.33916685e-01 2.80567974e-01
-2.48392392e-02 -9.98202860e-01 2.17400640e-01 6.19874179e-01
2.99462050e-01 -9.98334765e-01 8.88500452e-01 3.20413083e-01
-8.89909089e-01 -2.70718634e-01 -6.79163098e-01 1.33535769e-02
-4.80272263e-01 6.01971805e-01 -1.13132322e+00 -8.41819942e-02
7.37776458e-01 6.45286739e-01 -6.76841080e-01 1.12647069e+00
8.06608722e-02 5.74088514e-01 -6.38240159e-01 2.77122021e-01
4.26444799e-01 -7.64021933e-01 1.18179810e+00 9.58302677e-01
5.55400729e-01 -1.22344099e-01 2.20289737e-01 1.06739402e+00
1.70482576e-01 2.58137643e-01 -9.63343859e-01 -6.79401234e-02
2.52540469e-01 9.33100760e-01 -5.74770272e-01 -1.61290675e-01
-3.76741856e-01 9.41054225e-01 2.14254349e-01 2.68448055e-01
-4.84460145e-01 -6.91018105e-01 9.30356085e-01 1.12482645e-01
2.69974381e-01 4.76468652e-02 -2.13667467e-01 -1.05820835e+00
1.47599131e-01 -7.16064036e-01 8.69268537e-01 -3.34477633e-01
-1.40965962e+00 7.33613610e-01 -5.34764677e-02 -1.34719539e+00
-5.43302178e-01 -8.50840747e-01 -1.19280267e+00 5.98993063e-01
-8.62308323e-01 -4.50031430e-01 -3.49625111e-01 9.06252980e-01
2.16289967e-01 -8.58184814e-01 1.17773163e+00 -2.41902731e-02
-8.51338983e-01 6.72671914e-01 5.55869758e-01 5.78266680e-01
4.20561701e-01 -1.67788744e+00 3.95382106e-01 9.80691016e-01
3.30749065e-01 5.05165398e-01 7.60800779e-01 -6.04083717e-01
-7.88557947e-01 -1.12207031e+00 2.50414342e-01 -3.52402061e-01
6.16174161e-01 -1.55440673e-01 -1.51669884e+00 8.65663290e-01
-2.72494674e-01 4.46680039e-01 6.98296428e-01 -4.99537564e-04
-4.22000796e-01 -1.66693524e-01 -1.38478744e+00 4.10190374e-01
2.32878760e-01 -2.95782477e-01 -9.16525185e-01 1.69548169e-01
3.07765782e-01 -3.01562905e-01 -6.22817636e-01 4.45078790e-01
1.43981650e-01 -9.43891168e-01 7.80810118e-01 -8.14739883e-01
1.11902669e-01 -2.68577397e-01 -7.17804283e-02 -1.43474376e+00
-5.86485863e-02 -9.99595970e-02 -8.60314608e-01 7.85666049e-01
-5.67157380e-02 -8.34852934e-01 9.47710156e-01 4.45381820e-01
-1.81298420e-01 -9.95962024e-01 -1.07366931e+00 -6.57837033e-01
8.38630646e-02 -2.52675563e-01 5.85372567e-01 1.01252353e+00
-2.50118047e-01 -1.39424115e-01 7.07107484e-02 5.58670402e-01
9.05144095e-01 -8.94207135e-02 6.05312884e-01 -1.92245555e+00
1.51872575e-01 -4.56464946e-01 -1.00094640e+00 -3.70899886e-01
2.36709639e-01 -8.38524222e-01 -1.72208041e-01 -1.10748911e+00
-3.52139890e-01 -9.10141841e-02 -6.35173678e-01 6.12177551e-01
1.28970519e-01 -1.41286880e-01 -4.63528365e-01 1.82418257e-01
-3.35787833e-01 8.26163352e-01 5.92772424e-01 5.94940968e-02
-4.72120404e-01 1.16780266e-01 -2.89641589e-01 1.48462784e+00
9.68587875e-01 -6.88616693e-01 -2.37734929e-01 -1.65997952e-01
-4.76410203e-02 -4.70884889e-01 3.60637635e-01 -1.37602580e+00
1.45062700e-01 8.24663788e-02 8.21991742e-01 -6.96824014e-01
1.65451497e-01 -1.05159044e+00 -3.97574902e-01 4.20556754e-01
-2.08956227e-01 2.49287769e-01 2.45495901e-01 7.67720878e-01
-1.80401161e-01 -4.97153223e-01 9.99764979e-01 2.60650843e-01
-4.69279617e-01 3.97342801e-01 -3.77549320e-01 6.95056915e-02
1.25238359e+00 -2.87385672e-01 7.11076409e-02 -2.09595263e-01
-6.78064406e-01 1.55289620e-01 3.16065937e-01 6.07785642e-01
8.81049275e-01 -1.50125551e+00 -6.17725611e-01 9.70968246e-01
2.29194716e-01 1.81178778e-01 -3.78547646e-02 4.92605150e-01
-8.33193839e-01 2.33968094e-01 -5.08907437e-01 -8.86645913e-01
-9.59970772e-01 3.16019744e-01 8.08110416e-01 -8.88310466e-03
-1.22752047e+00 7.45570362e-01 9.09580290e-02 -5.58927238e-01
5.56927919e-01 -1.29396871e-01 -3.48314911e-01 -8.42056423e-02
6.64864719e-01 5.00792861e-01 1.68079853e-01 -7.22368181e-01
-2.04715893e-01 2.67026991e-01 -3.21170956e-01 1.31792143e-01
1.35298920e+00 2.82431990e-01 -1.48630098e-01 5.19033432e-01
1.30807984e+00 -3.51385087e-01 -1.28709745e+00 -3.38794053e-01
1.16781406e-01 -3.86813819e-01 2.46720314e-01 -3.79133701e-01
-1.16130769e+00 8.89550269e-01 1.04776633e+00 4.34974879e-01
1.03527582e+00 -3.14762801e-01 7.47683227e-01 9.21206415e-01
-3.70542079e-01 -1.31228209e+00 3.26219410e-01 6.52078509e-01
1.00372243e+00 -1.16043031e+00 -1.51408657e-01 3.91433507e-01
-6.10870242e-01 1.38418305e+00 1.07769179e+00 -7.20537305e-01
8.45897913e-01 3.25775631e-02 1.07427239e-01 -3.98504257e-01
-2.95948476e-01 4.40598160e-01 5.87270141e-01 4.44091827e-01
-8.37431401e-02 -1.34646386e-01 3.70317012e-01 4.49313313e-01
-3.38153332e-01 -8.26044917e-01 3.71622384e-01 7.00422823e-01
-7.54051745e-01 -6.46057367e-01 -6.74998999e-01 7.25724816e-01
-5.33668756e-01 2.30563506e-01 -3.88409227e-01 5.99017024e-01
7.59712458e-02 3.48698527e-01 8.17067981e-01 -9.54170749e-02
2.50901699e-01 6.03541255e-01 -2.79257689e-02 -4.51683372e-01
5.82900122e-02 -2.08885923e-01 -8.80926192e-01 -5.54052889e-01
2.65147656e-01 -6.81431174e-01 -1.55949330e+00 -1.57899596e-02
-2.37799704e-01 4.64071155e-01 7.13868976e-01 1.04795372e+00
2.53499418e-01 2.30386391e-01 5.89650393e-01 -4.59788024e-01
-8.85142148e-01 -9.23871100e-01 -8.48081470e-01 5.29778481e-01
7.56416321e-01 -7.28928745e-01 -8.10231507e-01 -4.16687071e-01] | [7.614740371704102, 2.3569207191467285] |
349481fe-83b5-4e35-bfed-a8ff73e55c83 | are-deep-policy-gradient-algorithms-truly | 1811.02553 | null | https://arxiv.org/abs/1811.02553v4 | https://arxiv.org/pdf/1811.02553v4.pdf | A Closer Look at Deep Policy Gradients | We study how the behavior of deep policy gradient algorithms reflects the conceptual framework motivating their development. To this end, we propose a fine-grained analysis of state-of-the-art methods based on key elements of this framework: gradient estimation, value prediction, and optimization landscapes. Our results show that the behavior of deep policy gradient algorithms often deviates from what their motivating framework would predict: the surrogate objective does not match the true reward landscape, learned value estimators fail to fit the true value function, and gradient estimates poorly correlate with the "true" gradient. The mismatch between predicted and empirical behavior we uncover highlights our poor understanding of current methods, and indicates the need to move beyond current benchmark-centric evaluation methods. | ['Firdaus Janoos', 'Shibani Santurkar', 'Aleksander Madry', 'Andrew Ilyas', 'Larry Rudolph', 'Logan Engstrom', 'Dimitris Tsipras'] | 2018-11-06 | null | https://openreview.net/forum?id=ryxdEkHtPS | https://openreview.net/pdf?id=ryxdEkHtPS | iclr-2020-1 | ['value-prediction'] | ['computer-code'] | [-1.16826624e-01 -2.31606618e-01 -9.55472052e-01 -3.87575150e-01
-6.38467669e-01 -8.54231656e-01 9.09518003e-01 1.20494790e-01
-7.13445961e-01 1.13125992e+00 5.63125014e-01 -6.86169386e-01
-2.03818783e-01 -4.98685151e-01 -7.99193382e-01 -3.92529249e-01
-7.28387907e-02 2.90605754e-01 7.51975998e-02 -3.68773431e-01
6.28149509e-01 2.45727688e-01 -1.27640164e+00 1.37756556e-01
7.77138829e-01 1.13308537e+00 -2.67801404e-01 6.07528627e-01
-3.80360223e-02 9.29750323e-01 -6.13148510e-01 -3.03544223e-01
3.52409989e-01 -4.79141653e-01 -5.06528199e-01 -6.24326944e-01
4.84055251e-01 -7.43703187e-01 -3.55369449e-01 1.15293419e+00
3.19815040e-01 1.48920730e-01 7.78742790e-01 -1.41460335e+00
-7.34285414e-01 5.26772439e-01 -3.68970543e-01 4.20360774e-01
7.44191781e-02 7.43183374e-01 1.22334135e+00 -2.86482215e-01
8.26328695e-01 1.03300726e+00 7.99466252e-01 6.07500732e-01
-1.42351639e+00 -3.68679166e-01 4.49536383e-01 -8.02978948e-02
-6.89744711e-01 -4.59842801e-01 4.48269606e-01 -6.02649510e-01
9.94468749e-01 -5.33848070e-02 9.88829315e-01 1.41618168e+00
3.47428232e-01 6.77149236e-01 1.40386117e+00 3.52881029e-02
5.68486512e-01 1.12126864e-01 6.86755180e-02 5.23007691e-01
5.05966008e-01 8.60302567e-01 -4.92847204e-01 -4.25294787e-01
7.05833912e-01 -1.60417169e-01 -3.14069211e-01 -6.69893205e-01
-1.01693285e+00 7.15475976e-01 3.94254357e-01 2.21276984e-01
-4.53613132e-01 7.95319378e-01 4.51926291e-01 3.57075959e-01
4.50429589e-01 8.32430720e-01 -5.61052382e-01 -9.90222514e-01
-1.08786249e+00 8.42234850e-01 8.66252244e-01 4.95992124e-01
6.24911070e-01 1.62308007e-01 -5.98676383e-01 3.89219522e-01
4.80600297e-02 3.93341422e-01 5.89371264e-01 -1.38747644e+00
3.39795381e-01 3.80583793e-01 8.14072788e-01 -6.84616685e-01
-3.73723149e-01 -8.36667955e-01 5.84647479e-03 6.14149868e-01
9.78118718e-01 -4.89102215e-01 -7.09552348e-01 1.93025720e+00
-7.83691853e-02 -1.76400781e-01 -9.48270336e-02 9.15058911e-01
5.88561259e-02 3.11382443e-01 1.61667183e-01 -3.28149274e-02
5.39819479e-01 -9.07070637e-01 -3.24100196e-01 -4.49591696e-01
5.44168055e-01 -3.08044165e-01 1.62946999e+00 1.19310468e-01
-1.28506827e+00 -1.77413091e-01 -1.04968882e+00 4.25846100e-01
-1.06712520e-01 -3.50738131e-02 6.99520051e-01 6.37654960e-01
-1.19914484e+00 1.11020505e+00 -8.26890111e-01 -1.33517981e-01
5.44814229e-01 4.19473089e-03 2.59073824e-01 3.05902421e-01
-6.99164569e-01 1.08960402e+00 2.54275858e-01 -2.28858635e-01
-1.19179666e+00 -8.98700297e-01 -2.81245440e-01 2.53413826e-01
4.28614885e-01 -5.60800135e-01 1.81242406e+00 -1.09932721e+00
-1.75072908e+00 4.88948107e-01 5.21506276e-03 -7.51715958e-01
1.02519190e+00 -4.47626173e-01 2.86492519e-03 -2.50400126e-01
-2.01196641e-01 3.79547536e-01 7.02997208e-01 -1.26161397e+00
-7.19398737e-01 -2.10882440e-01 6.56785220e-02 2.38692820e-01
-2.12755010e-01 -4.43762213e-01 2.00035661e-01 -3.61014545e-01
-4.47210550e-01 -6.62890792e-01 -2.83071518e-01 -9.96864289e-02
-2.76211321e-01 -2.72310153e-02 3.57120007e-01 -1.35352984e-01
1.55083072e+00 -1.79097056e+00 -1.75456017e-01 2.66078860e-01
4.52487051e-01 -4.05562818e-02 -1.34412333e-01 5.72910666e-01
3.91792029e-01 2.82433629e-01 6.38636202e-02 -1.16306101e-03
5.30525386e-01 -2.24741884e-02 -6.20196998e-01 6.61820769e-01
-1.29979253e-01 1.10173583e+00 -1.31022739e+00 -2.20995899e-02
-3.65410820e-02 -9.67064314e-03 -7.09836245e-01 2.22612783e-01
-6.24848723e-01 2.10921898e-01 -4.92355436e-01 5.34034908e-01
1.73230082e-01 -2.83433139e-01 3.00931275e-01 1.20299719e-01
-4.94052708e-01 6.43291295e-01 -6.49808824e-01 1.25983286e+00
-5.75132817e-02 7.47465670e-01 -6.39547855e-02 -8.92659605e-01
7.65247703e-01 -1.87235236e-01 5.46386421e-01 -8.30892622e-01
-1.24179833e-02 5.30904233e-01 1.97899088e-01 -1.45025238e-01
5.39886951e-01 -2.20078658e-02 3.24245840e-01 6.39515579e-01
-1.35568246e-01 -1.95811108e-01 1.84018329e-01 -1.75808012e-01
1.23089659e+00 5.81196904e-01 1.62827015e-01 -5.08974373e-01
-1.95828602e-01 4.28624272e-01 4.96990174e-01 1.39453030e+00
-6.15969837e-01 6.30917549e-02 1.10049689e+00 -5.69324076e-01
-1.19134355e+00 -1.39952052e+00 4.51499149e-02 1.36972725e+00
1.73854269e-02 -2.69796103e-01 -7.76820302e-01 -8.97573888e-01
5.12494326e-01 6.73774242e-01 -9.34428096e-01 -2.69237429e-01
-3.80846083e-01 -6.01925373e-01 5.97224414e-01 5.81908166e-01
3.87916774e-01 -9.16444004e-01 -1.12004340e+00 4.34465528e-01
2.74381012e-01 -7.23273635e-01 -2.84811258e-01 7.66291888e-03
-9.70464289e-01 -1.16074145e+00 -6.24585092e-01 -5.54861911e-02
1.74511209e-01 -2.17912033e-01 1.65990961e+00 -2.66408846e-02
2.47312486e-01 4.93220955e-01 -9.61125121e-02 -3.86044055e-01
-4.49873209e-01 1.79289520e-01 6.12842031e-02 -5.14657855e-01
1.92914292e-01 -5.57779312e-01 -1.02166021e+00 2.36075893e-01
-4.35946912e-01 -4.72656041e-01 5.35143495e-01 9.37160790e-01
3.89635056e-01 -4.96931732e-01 7.43391454e-01 -6.73635781e-01
1.45711827e+00 -5.71408689e-01 -9.26592231e-01 3.43342870e-01
-1.31658328e+00 4.02452111e-01 5.92456520e-01 -5.61742127e-01
-9.77950335e-01 -4.71632779e-01 1.21282004e-01 -2.08556101e-01
1.84709355e-01 3.38859797e-01 6.44146919e-01 3.75199355e-02
1.00590241e+00 1.41968831e-01 1.71670094e-02 -3.68781865e-01
4.45271611e-01 2.31910497e-01 5.51313221e-01 -1.08076465e+00
3.17046553e-01 4.74232256e-01 -1.90511525e-01 -1.27638951e-01
-7.50714004e-01 1.06791869e-01 1.08374253e-01 -2.72887081e-01
4.12716448e-01 -4.89422590e-01 -9.77029085e-01 3.27441007e-01
-8.67762208e-01 -1.14958835e+00 -7.63817608e-01 5.11648178e-01
-8.86756718e-01 1.09047466e-03 -6.03220642e-01 -1.01870286e+00
-2.83260256e-01 -1.32490492e+00 7.73287117e-01 3.55209857e-01
-3.43655437e-01 -1.16768146e+00 5.17020822e-01 -2.81925887e-01
9.46160376e-01 3.18545967e-01 1.04953289e+00 -4.01780516e-01
-2.92248070e-01 2.34867305e-01 -2.98237115e-01 2.64569581e-01
-9.31114629e-02 1.29061878e-01 -7.23212004e-01 -3.56553584e-01
-1.10699140e-01 -4.77775842e-01 8.49595129e-01 7.26463675e-01
8.67249131e-01 -2.87300378e-01 -1.79737642e-01 7.88361847e-01
1.48721993e+00 1.98802561e-01 2.71765947e-01 8.44819546e-01
1.88458443e-01 2.99654573e-01 3.50341171e-01 7.02859581e-01
1.44970879e-01 4.13981378e-01 5.13702631e-01 3.03214103e-01
2.06923455e-01 -6.55179083e-01 5.20152032e-01 -3.79955396e-02
-3.01432051e-03 -1.78002596e-01 -9.50532377e-01 4.10731375e-01
-2.07043338e+00 -1.11593068e+00 2.99683332e-01 2.33449912e+00
9.06407177e-01 6.37407064e-01 6.23481929e-01 -5.20284414e-01
3.45427096e-01 3.58994126e-01 -1.35805583e+00 -6.71097934e-01
2.01723322e-01 5.00330068e-02 8.36218476e-01 4.81354803e-01
-6.95290267e-01 1.09659588e+00 8.49373722e+00 5.81947982e-01
-1.18765223e+00 -1.10631380e-02 8.15216660e-01 -3.48143220e-01
-4.99605030e-01 1.49379492e-01 -5.76868773e-01 5.21497905e-01
9.73921716e-01 -4.63416338e-01 8.84793639e-01 1.03218925e+00
3.45686734e-01 -1.26073644e-01 -1.26959002e+00 7.03458846e-01
-6.32281661e-01 -1.62536895e+00 -3.36488336e-01 3.68973881e-01
8.43817592e-01 5.02882123e-01 4.51092362e-01 5.77569067e-01
1.06507015e+00 -1.25541377e+00 9.22837675e-01 5.40096283e-01
6.61558449e-01 -4.28999156e-01 3.47740024e-01 2.15860635e-01
-5.89862525e-01 -4.60898787e-01 -2.42496029e-01 -4.78687614e-01
-3.63104045e-02 3.73392612e-01 -8.25789452e-01 -3.62457544e-01
6.14504755e-01 5.84606469e-01 -3.50047171e-01 9.58394766e-01
-2.24084377e-01 7.74923027e-01 -2.68382132e-01 -4.76211250e-01
7.96709001e-01 -3.11837882e-01 6.24394000e-01 1.12483037e+00
8.41317698e-02 -5.99170685e-01 7.01414049e-02 1.20629859e+00
-1.57795370e-01 3.82780470e-02 -6.04459047e-01 -4.16194111e-01
4.96018291e-01 6.98674023e-01 -4.69741613e-01 -3.58058423e-01
-1.74764380e-01 4.87252831e-01 6.89649880e-01 7.42163241e-01
-8.18687081e-01 -3.70118394e-02 1.14686346e+00 -3.86095382e-02
5.88842183e-02 -2.64943153e-01 -6.38016403e-01 -1.13304162e+00
2.44508103e-01 -9.73598361e-01 6.06081560e-02 -5.41890502e-01
-1.27020812e+00 2.84189254e-01 5.72876371e-02 -8.62145662e-01
-7.26645827e-01 -6.89262986e-01 -8.06614459e-01 6.19077802e-01
-1.48911452e+00 -2.12401986e-01 -4.11896072e-02 2.69115478e-01
1.84705973e-01 -1.85402364e-01 3.53490651e-01 -2.57910162e-01
-2.80209124e-01 6.69186950e-01 7.31587768e-01 -1.50013968e-01
4.85528648e-01 -1.41299558e+00 6.79951370e-01 4.51115191e-01
-1.16695896e-01 5.32988727e-01 9.92703795e-01 -6.61986709e-01
-1.39590418e+00 -4.95833009e-01 1.51419220e-02 -6.31102085e-01
9.80845749e-01 7.91155100e-02 -4.71900672e-01 4.74328488e-01
2.67110132e-02 -4.74855155e-02 1.31076038e-01 2.92381793e-01
-3.26600283e-01 -3.91949490e-02 -1.12207961e+00 9.44457471e-01
1.23350060e+00 -4.64612365e-01 -2.50471085e-01 9.49502513e-02
4.64388520e-01 -4.66019720e-01 -5.72837949e-01 3.40167612e-01
9.90346253e-01 -1.50188792e+00 7.91127980e-01 -1.10668826e+00
5.53199828e-01 2.62198359e-01 -2.77886569e-01 -1.79678178e+00
-6.25875443e-02 -7.04369664e-01 -4.29379404e-01 5.67783475e-01
4.67626780e-01 -8.99063110e-01 1.12393856e+00 7.37367511e-01
-3.84723134e-02 -1.24620402e+00 -8.29915524e-01 -9.87917721e-01
5.07072210e-01 -4.49716419e-01 7.68605888e-01 7.33688772e-01
6.49952069e-02 -3.35463472e-02 -6.79281726e-02 -4.96599048e-01
6.74080372e-01 1.67082414e-01 6.36301458e-01 -8.34293783e-01
-6.24311984e-01 -1.35533309e+00 -1.48338914e-01 -1.19411147e+00
1.80270955e-01 -4.35716897e-01 -1.36411011e-01 -1.41766798e+00
3.21736746e-02 -4.92944896e-01 -5.23677886e-01 2.51742095e-01
-1.73535123e-01 -2.23195881e-01 2.14898348e-01 1.65479332e-01
-5.64969003e-01 6.68063462e-01 1.13590240e+00 7.69395009e-02
-5.27473807e-01 -5.31801842e-02 -8.60624135e-01 6.86670542e-01
1.04157102e+00 -4.64503795e-01 -5.56468964e-01 -4.24490064e-01
4.73103732e-01 1.36868358e-02 3.02646965e-01 -9.69120681e-01
-7.67178237e-02 -7.32252896e-01 4.29778486e-01 3.36735658e-02
3.50087173e-02 -4.54632401e-01 -3.88000041e-01 6.16517663e-01
-7.89652228e-01 5.07530391e-01 1.41912445e-01 5.41638374e-01
1.62660599e-01 -1.07884437e-01 5.99775791e-01 -2.04783544e-01
-5.33610225e-01 1.28877863e-01 -2.94468880e-01 7.26910532e-01
6.34681940e-01 -1.40771493e-01 -7.19110429e-01 -5.86548030e-01
-2.71853119e-01 3.23455960e-01 8.22119594e-01 1.43914819e-01
3.93778980e-01 -1.12932599e+00 -5.28385162e-01 -2.19475955e-01
-1.16762146e-01 -7.33202875e-01 -3.28672558e-01 6.27160847e-01
-3.56408715e-01 2.74786085e-01 -2.05484197e-01 -3.08704466e-01
-3.12914044e-01 2.19763517e-01 8.72417748e-01 -5.60729265e-01
-4.00647789e-01 4.83775973e-01 -2.01351494e-02 -3.15636307e-01
3.65675986e-01 -6.44593656e-01 2.36315414e-01 -3.01497519e-01
2.96500802e-01 5.65088153e-01 -3.74461085e-01 1.40909016e-01
-1.06604345e-01 1.03664175e-01 1.01990819e-01 -6.07761204e-01
1.26216829e+00 1.66724563e-01 3.00462782e-01 4.74440068e-01
1.03101671e+00 -1.14711009e-01 -2.05652070e+00 3.18035185e-02
2.10797623e-01 -5.88766694e-01 2.16623455e-01 -1.14906991e+00
-7.36952901e-01 6.88206911e-01 7.50676453e-01 3.33333641e-01
6.56238258e-01 -4.44862723e-01 6.32333636e-01 6.51171744e-01
2.97610581e-01 -1.59394741e+00 -5.69093265e-02 5.16681612e-01
6.17304444e-01 -1.20348191e+00 1.20174527e-01 5.37225962e-01
-7.34474123e-01 9.42696512e-01 7.41577685e-01 -3.64708722e-01
5.58582544e-01 1.58557415e-01 7.76253715e-02 -1.15896069e-01
-1.05029070e+00 -2.35787526e-01 2.33418345e-02 5.20929635e-01
3.54931653e-01 1.39284715e-01 -3.98034692e-01 3.01473767e-01
-2.68293262e-01 2.30014339e-01 3.34036887e-01 1.05383587e+00
-5.66255271e-01 -9.91457760e-01 -8.49619210e-02 6.84588075e-01
-6.39288425e-01 -1.68902040e-01 -4.80550379e-01 1.01136684e+00
-5.45459569e-01 7.76303649e-01 5.29619828e-02 -4.14135218e-01
3.95259172e-01 9.28382482e-03 5.25812984e-01 -8.72449130e-02
-7.76753008e-01 -2.97278106e-01 1.59843087e-01 -9.77758884e-01
1.51285604e-01 -3.80564064e-01 -9.69314873e-01 -4.41538483e-01
2.62060165e-01 2.70590603e-01 8.11520636e-01 8.50475192e-01
4.72041547e-01 1.32623658e-01 4.70945388e-01 -8.39787245e-01
-1.50472009e+00 -6.17291331e-01 -2.62843102e-01 4.40554976e-01
6.45309985e-01 -7.51671612e-01 -4.02352512e-01 -5.86245596e-01] | [4.100866317749023, 2.283757209777832] |
36d15a46-6723-4b2f-878d-126feb488235 | sdxl-improving-latent-diffusion-models-for | 2307.01952 | null | https://arxiv.org/abs/2307.01952v1 | https://arxiv.org/pdf/2307.01952v1.pdf | SDXL: Improving Latent Diffusion Models for High-Resolution Image Synthesis | We present SDXL, a latent diffusion model for text-to-image synthesis. Compared to previous versions of Stable Diffusion, SDXL leverages a three times larger UNet backbone: The increase of model parameters is mainly due to more attention blocks and a larger cross-attention context as SDXL uses a second text encoder. We design multiple novel conditioning schemes and train SDXL on multiple aspect ratios. We also introduce a refinement model which is used to improve the visual fidelity of samples generated by SDXL using a post-hoc image-to-image technique. We demonstrate that SDXL shows drastically improved performance compared the previous versions of Stable Diffusion and achieves results competitive with those of black-box state-of-the-art image generators. In the spirit of promoting open research and fostering transparency in large model training and evaluation, we provide access to code and model weights at https://github.com/Stability-AI/generative-models | ['Robin Rombach', 'Joe Penna', 'Jonas Müller', 'Tim Dockhorn', 'Andreas Blattmann', 'Kyle Lacey', 'Zion English', 'Dustin Podell'] | 2023-07-04 | null | null | null | null | ['image-generation'] | ['computer-vision'] | [ 2.41221040e-01 3.36138606e-01 -1.50375128e-01 -2.28732809e-01
-9.78814185e-01 -5.18370926e-01 1.05231762e+00 -3.78034413e-01
-1.78299010e-01 5.97426057e-01 6.61359549e-01 -3.06146175e-01
3.08366627e-01 -7.42163718e-01 -9.22103465e-01 -5.38143098e-01
3.14205199e-01 5.21249592e-01 9.18336436e-02 -9.01548117e-02
9.48620737e-02 2.77006716e-01 -1.10290205e+00 3.83342475e-01
6.27880871e-01 5.98907769e-01 2.79566586e-01 1.02056229e+00
9.88402218e-02 1.08554614e+00 -5.13458908e-01 -4.58110601e-01
3.87020826e-01 -7.20263779e-01 -6.47038043e-01 1.47991613e-01
6.53983533e-01 -6.11423552e-01 -5.97856581e-01 7.79843271e-01
7.68609881e-01 -7.21684992e-02 6.75090611e-01 -1.29969084e+00
-1.46737027e+00 7.73715436e-01 -6.15783513e-01 5.85201196e-02
1.16234317e-01 7.25191474e-01 1.09666383e+00 -9.64456141e-01
9.39542770e-01 1.44119465e+00 6.01586282e-01 9.01283503e-01
-1.63389158e+00 -7.10207820e-01 1.13747958e-02 -2.83586413e-01
-1.24258554e+00 -8.65985215e-01 4.77896273e-01 -3.74111682e-01
1.02569151e+00 -1.98000576e-02 5.88120639e-01 1.74753940e+00
1.38992444e-01 9.30769145e-01 1.02666986e+00 -4.02433425e-01
1.45676300e-01 1.37910858e-01 -4.40342069e-01 7.89060295e-01
4.62062880e-02 2.63630062e-01 -4.98938799e-01 8.84351060e-02
1.08340275e+00 -1.58357292e-01 -2.46413752e-01 -3.22980911e-01
-1.15894103e+00 9.71226871e-01 5.39758325e-01 4.89311554e-02
-2.99352318e-01 7.56767809e-01 2.17161123e-02 1.89966679e-01
5.67888796e-01 5.44528544e-01 -1.44655824e-01 -2.39011109e-01
-1.21009326e+00 4.19464558e-01 6.04114115e-01 1.28411710e+00
5.51504850e-01 5.02663136e-01 -5.38699150e-01 6.65964484e-01
4.16459769e-01 7.66418934e-01 4.96109009e-01 -1.36384368e+00
3.88725996e-01 1.74572587e-01 -1.01556130e-01 -4.95046765e-01
1.58168197e-01 -5.37793994e-01 -7.68766284e-01 4.96768057e-01
-1.17287757e-02 -3.06286603e-01 -1.29076564e+00 1.65843785e+00
-3.91871706e-02 4.62474041e-02 -8.71057883e-02 6.21645212e-01
5.33815622e-01 8.86591971e-01 -5.74177736e-03 1.73765242e-01
8.17332864e-01 -1.39383066e+00 -5.10084808e-01 -3.29977930e-01
4.90163505e-01 -7.95179844e-01 1.32377446e+00 2.75646746e-01
-1.49279439e+00 -5.15740216e-01 -9.85474288e-01 -4.91006762e-01
-5.28957024e-02 1.02609694e-01 4.95605856e-01 4.88593280e-01
-1.44988310e+00 5.65268278e-01 -8.99064183e-01 -2.03269541e-01
7.63412952e-01 7.18222186e-02 -1.06826447e-01 -9.88505036e-02
-7.62560844e-01 7.09924936e-01 -1.37798458e-01 -2.96872169e-01
-1.37552178e+00 -9.20809865e-01 -8.82735491e-01 -5.01427017e-02
3.65284234e-02 -1.14105022e+00 1.44101489e+00 -1.05421138e+00
-1.68590736e+00 7.87927687e-01 -1.08275935e-01 -5.88985324e-01
8.45275402e-01 -2.34392852e-01 -6.21240325e-02 7.01541528e-02
2.28781253e-01 1.25061512e+00 1.17315495e+00 -1.41804171e+00
-2.48146877e-01 1.66274548e-01 -1.54939264e-01 7.84053840e-03
-1.15870416e-01 -1.38494268e-01 -6.87493265e-01 -9.46480691e-01
-5.23855865e-01 -1.03793156e+00 -2.96887577e-01 3.24361622e-01
-4.31755215e-01 2.01673865e-01 8.99193108e-01 -5.74159145e-01
1.12205565e+00 -2.09060359e+00 3.70485395e-01 -1.83880311e-02
5.74428618e-01 2.07303151e-01 -4.99217629e-01 5.37445307e-01
-2.46549211e-03 4.05437797e-01 -3.95819873e-01 -9.53227878e-01
1.88549548e-01 4.23305929e-02 -5.28792262e-01 1.59382597e-01
3.02134007e-01 1.22786283e+00 -6.21848285e-01 -4.01940346e-01
5.41605689e-02 7.45334327e-01 -8.68942738e-01 1.01533055e-01
-5.61366916e-01 3.87768388e-01 -1.81998044e-01 3.45628262e-01
4.27890271e-01 -5.09027779e-01 -2.09852085e-01 -1.43029960e-02
-2.01808885e-02 2.38404185e-01 -8.16142082e-01 1.97408164e+00
-7.00745165e-01 9.74893689e-01 -1.25108063e-01 -1.96707428e-01
6.14729524e-01 2.66485035e-01 1.68823570e-01 -6.80590153e-01
9.76656601e-02 1.13372371e-01 -2.46329531e-01 -6.98887035e-02
5.61974525e-01 1.56153971e-02 3.35692406e-01 8.28322351e-01
3.01249683e-01 -5.54779828e-01 3.10381591e-01 6.95077300e-01
1.08216655e+00 4.29233193e-01 -2.45293528e-01 -1.52921349e-01
-2.07832888e-01 -2.17897519e-01 1.45812079e-01 7.73636639e-01
1.91010579e-01 9.65322614e-01 5.18303990e-01 -1.27066076e-01
-1.64177835e+00 -1.24688959e+00 -4.31789691e-03 6.92540407e-01
-2.30036199e-01 -6.89499855e-01 -8.61036479e-01 -5.55625618e-01
3.76053452e-02 1.01984620e+00 -8.83868992e-01 -1.50727183e-01
-3.03934783e-01 -5.73314250e-01 8.57208788e-01 6.31895721e-01
4.67907965e-01 -1.10949981e+00 -3.36895436e-01 5.51304780e-02
2.72507668e-01 -9.35489237e-01 -8.88821423e-01 7.00437184e-03
-7.55768538e-01 -4.84644234e-01 -1.01728415e+00 -6.54847980e-01
6.97047293e-01 5.33360243e-02 1.25327396e+00 8.10327157e-02
-3.94262642e-01 1.92110598e-01 -1.99304566e-01 -3.82030040e-01
-9.06739831e-01 2.16288909e-01 -3.18669617e-01 -2.24068552e-01
-2.15979993e-01 -5.86047709e-01 -8.79695833e-01 7.03066215e-02
-1.15189278e+00 5.06377935e-01 7.26408422e-01 7.62007415e-01
4.25474972e-01 -3.79091859e-01 1.39701769e-01 -9.93094146e-01
8.64297688e-01 -3.56926084e-01 -6.55107021e-01 -1.09352164e-01
-9.75327253e-01 4.52848583e-01 3.59120369e-01 -4.64820385e-01
-1.07614684e+00 -1.90282881e-01 -3.04487824e-01 -7.09765851e-01
1.39978543e-01 7.11617842e-02 1.72380894e-01 1.08830720e-01
8.92278850e-01 1.18498862e-01 5.02006263e-02 -2.41751432e-01
9.61680055e-01 4.54601496e-01 2.35877082e-01 -5.11729121e-01
8.61513257e-01 3.59960079e-01 -2.80437142e-01 -5.14386356e-01
-5.38603365e-01 1.78466737e-01 -2.51963377e-01 8.52198228e-02
9.17426527e-01 -1.10565042e+00 -3.09294462e-01 5.82275927e-01
-1.11206579e+00 -1.20962870e+00 -7.83031344e-01 8.05375800e-02
-6.44746423e-01 -3.18779163e-02 -9.83061016e-01 -3.83488148e-01
-4.64339435e-01 -1.30490923e+00 1.32002020e+00 -1.09775424e-01
-3.82042706e-01 -1.06920516e+00 3.31351906e-01 1.81321278e-01
8.77076268e-01 1.88977942e-02 6.53571069e-01 -9.32410806e-02
-8.51320267e-01 1.31020650e-01 -3.61202985e-01 5.56237221e-01
-9.72425118e-02 3.79914820e-01 -9.77189541e-01 -3.85205060e-01
-3.13495725e-01 -4.96118754e-01 1.10912263e+00 5.92133641e-01
8.29815090e-01 -4.11674351e-01 -3.48331064e-01 9.51327682e-01
1.34450722e+00 -8.56177658e-02 8.26599896e-01 1.60316899e-01
9.03979838e-01 1.00664333e-01 -1.52316406e-01 3.78177643e-01
3.17884624e-01 5.75707197e-01 1.60501339e-02 -4.84479219e-01
-8.29925358e-01 -5.79347432e-01 5.47733128e-01 7.49381781e-01
2.05139086e-01 -5.77775061e-01 -7.88399756e-01 5.69190800e-01
-1.71901810e+00 -1.13335991e+00 -5.54381385e-02 1.81777668e+00
1.06741679e+00 2.46293530e-01 3.01599037e-02 -2.99265981e-01
3.93922955e-01 3.72071743e-01 -6.11275196e-01 -3.78859371e-01
-2.93957770e-01 3.48573983e-01 6.71766043e-01 9.31552112e-01
-6.52884245e-01 1.24631691e+00 7.19908667e+00 9.63225067e-01
-1.16600537e+00 2.87303656e-01 9.27708745e-01 -5.46771288e-01
-7.45250106e-01 6.26108348e-02 -8.50942016e-01 3.97298485e-01
9.35583115e-01 -2.04650015e-01 5.43250144e-01 7.17876792e-01
2.58871257e-01 1.24007940e-01 -9.88731325e-01 7.58272469e-01
1.58077285e-01 -1.71387708e+00 3.07887048e-01 3.12765568e-01
1.19516861e+00 4.34548318e-01 4.51586545e-01 1.44099712e-01
1.01277506e+00 -1.08150554e+00 1.09891272e+00 4.58603919e-01
1.13506520e+00 -4.55939859e-01 8.06549415e-02 -2.08358429e-02
-7.23803520e-01 6.17629513e-02 -2.79130101e-01 2.74365067e-01
3.25846285e-01 4.49547678e-01 -6.36954486e-01 1.10240523e-02
4.49530303e-01 8.16987336e-01 -7.05411315e-01 6.43897355e-01
-5.48760116e-01 8.29299927e-01 -1.45610765e-01 2.70341069e-01
2.50243932e-01 2.08025966e-02 5.15606761e-01 1.31981826e+00
3.78818870e-01 -3.58356208e-01 -2.49467105e-01 1.47622526e+00
-4.42971289e-01 -2.03112274e-01 -7.77064085e-01 -8.35043341e-02
3.30014020e-01 1.10985684e+00 -3.94600272e-01 -5.18237293e-01
-2.92878717e-01 1.43671060e+00 3.13332170e-01 5.27388394e-01
-1.02071559e+00 -3.03252727e-01 3.99822623e-01 3.29541683e-01
5.03394663e-01 -3.64627749e-01 -3.59001547e-01 -1.17123890e+00
-2.61295229e-01 -9.89609063e-01 -1.74053028e-01 -1.25422144e+00
-1.17443466e+00 8.81706178e-01 -5.92073649e-02 -7.63805926e-01
-3.35620731e-01 -3.16281199e-01 -5.80286026e-01 1.05814898e+00
-1.30723011e+00 -1.54825830e+00 -2.70111769e-01 4.30096418e-01
7.50551343e-01 -1.36656269e-01 7.22296715e-01 1.57512426e-01
-4.92403179e-01 6.47299528e-01 1.19078860e-01 4.45148535e-02
9.08650994e-01 -1.24918091e+00 1.24402559e+00 1.05797637e+00
3.49903703e-01 6.64328039e-01 7.31882632e-01 -7.28497446e-01
-1.16874695e+00 -1.18320143e+00 5.38196743e-01 -7.73089707e-01
7.60468602e-01 -5.06169796e-01 -4.53534126e-01 1.12635374e+00
8.75039577e-01 -2.05898717e-01 3.70723218e-01 -1.75724834e-01
-4.72831011e-01 6.65585622e-02 -7.83538818e-01 9.47447181e-01
1.16659236e+00 -5.60445189e-01 -8.11632872e-02 1.99897334e-01
1.05831361e+00 -3.75682920e-01 -6.25110805e-01 -9.27236378e-02
5.01246691e-01 -9.42245662e-01 8.30192983e-01 -3.06529731e-01
9.56678092e-01 -8.72424394e-02 -2.50766445e-02 -1.46474040e+00
-5.71280479e-01 -1.02205145e+00 -1.00709856e-01 1.19227123e+00
8.05908024e-01 -4.65076655e-01 7.87166655e-01 5.59984446e-01
-2.05026314e-01 -6.64995551e-01 -4.83454585e-01 -7.15695381e-01
2.74066418e-01 -3.68428379e-01 5.02364933e-01 6.91973686e-01
-3.98796737e-01 7.21628904e-01 -5.82868397e-01 -1.95302784e-01
5.82761347e-01 -2.19507039e-01 9.67596412e-01 -5.93930900e-01
-7.45042622e-01 -6.62328660e-01 -1.18962140e-03 -1.37641466e+00
-1.48851678e-01 -8.62445474e-01 -4.03835960e-02 -1.64634812e+00
1.64605185e-01 -4.61772293e-01 1.18957140e-01 5.84298611e-01
-4.31392007e-02 6.17387235e-01 4.66423124e-01 3.25667351e-01
-4.70421851e-01 8.28288913e-01 1.51660061e+00 -2.02047780e-01
-1.74492672e-01 -3.53392124e-01 -9.31981683e-01 2.76764274e-01
7.03569770e-01 -5.75018525e-01 -7.16195643e-01 -9.51113880e-01
2.84072012e-01 -1.31217882e-01 3.97968739e-01 -1.11433363e+00
1.65103003e-01 7.82800093e-02 3.89967859e-01 -1.50547683e-01
4.39913958e-01 -3.67840469e-01 3.99657100e-01 3.22237045e-01
-6.52037203e-01 2.98545897e-01 2.89645314e-01 4.04724658e-01
4.34361175e-02 -1.49320224e-02 8.85418773e-01 -1.66704074e-01
-2.45091960e-01 5.03272474e-01 -3.06206524e-01 2.44056225e-01
8.81747067e-01 2.96010193e-03 -5.47039390e-01 -7.99894810e-01
-5.11587143e-01 -1.33439563e-02 8.58606935e-01 5.29430807e-01
5.88528872e-01 -1.29007685e+00 -9.07342315e-01 3.12347144e-01
-2.12372877e-02 4.05406542e-02 3.40367481e-02 5.84994912e-01
-7.06942081e-01 2.00281039e-01 -5.85011989e-02 -3.65135729e-01
-1.01043713e+00 4.90327060e-01 2.38683283e-01 -1.74480274e-01
-7.48296380e-01 1.07655990e+00 3.89520198e-01 -5.71478829e-02
2.26575490e-02 -2.86994278e-01 6.82872653e-01 -3.53850782e-01
4.07676935e-01 1.61767960e-01 -2.75357813e-01 -3.97625178e-01
1.00439325e-01 3.59589756e-01 -2.68969178e-01 -7.15602815e-01
1.35433328e+00 -9.41155702e-02 2.26083577e-01 2.19393969e-01
1.17827082e+00 3.22833240e-01 -1.85493469e+00 -3.24343070e-02
-6.54819429e-01 -3.47635806e-01 3.16346765e-01 -9.52719331e-01
-1.28753185e+00 9.15062904e-01 4.64138776e-01 -8.57233852e-02
8.10591698e-01 -7.83485845e-02 8.57810557e-01 -8.01511854e-02
-3.10291070e-02 -8.69430065e-01 3.30365598e-01 2.40504250e-01
1.20779085e+00 -9.85617340e-01 9.77436528e-02 2.71937307e-02
-9.78802145e-01 7.08600521e-01 4.48714346e-01 -3.11030388e-01
5.35129249e-01 7.50669003e-01 2.47867033e-01 -1.14707835e-01
-1.19355440e+00 4.08117361e-02 8.72472748e-02 5.70434809e-01
5.60479999e-01 -2.49508947e-01 2.50149101e-01 -1.28381494e-02
-2.75743514e-01 1.84552893e-01 5.26400328e-01 7.46030331e-01
-8.91440362e-02 -1.35416043e+00 -5.29420981e-03 3.22268248e-01
-3.20873916e-01 -6.39696836e-01 -1.77957982e-01 6.85225487e-01
-1.46339834e-01 7.60585487e-01 1.17979579e-01 -1.68266252e-01
7.87760466e-02 -9.75881740e-02 6.06499851e-01 -5.90661108e-01
-4.51879025e-01 1.83808193e-01 1.26570500e-02 -5.83815217e-01
-1.44358441e-01 -4.63757575e-01 -1.00141621e+00 -6.75181270e-01
-2.15492606e-01 -1.67215839e-01 7.15082884e-01 4.64383513e-01
9.03227687e-01 7.40658343e-01 3.96953821e-01 -1.05339158e+00
-2.88245291e-01 -9.75855708e-01 -3.08512181e-01 4.34969544e-01
3.48612666e-01 -2.64949530e-01 -2.47839049e-01 4.14613068e-01] | [11.346586227416992, -0.22454309463500977] |
f9f7d280-4b0b-4705-886b-7122b8ab7518 | autonomous-systems-autonomous-systems-indoor | 2304.08893 | null | https://arxiv.org/abs/2304.08893v1 | https://arxiv.org/pdf/2304.08893v1.pdf | Autonomous Systems: Autonomous Systems: Indoor Drone Navigation | Drones are a promising technology for autonomous data collection and indoor sensing. In situations when human-controlled UAVs may not be practical or dependable, such as in uncharted or dangerous locations, the usage of autonomous UAVs offers flexibility, cost savings, and reduced risk. The system creates a simulated quadcopter capable of autonomously travelling in an indoor environment using the gazebo simulation tool and the ros navigation system framework known as Navigaation2. While Nav2 has successfully shown the functioning of autonomous navigation in terrestrial robots and vehicles, the same hasn't been accomplished with unmanned aerial vehicles and still has to be done. The goal is to use the slam toolbox for ROS and the Nav2 navigation system framework to construct a simulated drone that can move autonomously in an indoor (gps-less) environment. | ['Manoj Kumar Rajagopal', 'Naren M', 'Santosh Narayan', 'Aswin Iyer'] | 2023-04-18 | null | null | null | null | ['drone-navigation', 'autonomous-navigation'] | ['computer-vision', 'computer-vision'] | [-1.66665599e-01 -1.23466335e-01 6.74750030e-01 -2.99011409e-01
3.60708058e-01 -9.40921366e-01 6.01638496e-01 -1.70528844e-01
-5.84326744e-01 8.94252658e-01 -7.44373381e-01 -7.02078640e-01
-2.85936564e-01 -1.03884780e+00 -2.73135900e-01 -3.84715170e-01
-4.51318115e-01 3.44084799e-01 4.64176923e-01 -1.00496900e+00
-3.54242437e-02 8.84967625e-01 -2.02474928e+00 -8.34089577e-01
6.16093934e-01 6.16682053e-01 4.52852309e-01 1.02519107e+00
5.71537256e-01 2.05473229e-01 -3.98797810e-01 4.35734540e-01
7.73955524e-01 -1.98153630e-01 -3.23644459e-01 -2.88521469e-01
2.50379652e-01 -2.75588393e-01 -1.32291451e-01 7.03954816e-01
3.68638188e-01 4.19716716e-01 2.11489230e-01 -1.54946673e+00
3.22251707e-01 -2.39893109e-01 3.50753129e-01 -8.15588385e-02
7.79386282e-01 6.78340718e-02 2.61130124e-01 -2.75805533e-01
8.09422731e-01 7.39092648e-01 6.48716092e-01 -1.36622667e-01
-7.99394310e-01 -2.98823476e-01 -1.81797981e-01 -3.63990545e-01
-1.66294265e+00 -1.96817860e-01 -8.12411606e-02 -3.29325765e-01
8.72987688e-01 4.46829289e-01 1.14287102e+00 5.64213336e-01
8.35190415e-01 -2.60624811e-02 7.10335672e-01 -2.83649415e-01
4.67923701e-01 1.42154977e-01 -3.96896005e-01 8.78064275e-01
7.58279741e-01 3.02179694e-01 -1.07474878e-01 7.57087022e-02
6.15939856e-01 1.42698705e-01 -5.98058462e-01 -9.28860843e-01
-1.57881904e+00 5.73014796e-01 8.00567627e-01 2.31323510e-01
-4.53265816e-01 1.58749938e-01 2.48698294e-02 6.45107865e-01
-2.19691828e-01 8.60165358e-01 -3.58853042e-01 -3.92917484e-01
-9.30570900e-01 7.93423474e-01 9.75513041e-01 1.42758298e+00
8.17471862e-01 3.14700574e-01 7.47102261e-01 -9.60657652e-03
2.94447690e-01 7.35247850e-01 1.69217199e-01 -1.15453529e+00
-1.15629986e-01 7.83568561e-01 7.41268337e-01 -1.22377956e+00
-7.38369226e-01 -1.80636808e-01 -2.16338843e-01 9.49537516e-01
1.84116766e-01 -6.78842247e-01 -7.33334839e-01 8.85546207e-01
5.81193864e-01 -2.68093914e-01 3.29725266e-01 1.09155953e+00
4.20896381e-01 5.21807671e-01 -4.61063504e-01 2.07871571e-01
9.85552311e-01 -4.83040839e-01 -5.97489417e-01 -3.87012720e-01
9.56966639e-01 -5.99414110e-01 5.38473487e-01 4.68413293e-01
-4.31962490e-01 -3.38036984e-01 -1.49294901e+00 1.30326346e-01
-9.60660279e-01 -7.43781328e-02 5.07817686e-01 7.20963955e-01
-1.35012496e+00 3.34225833e-01 -1.10936952e+00 -8.18887115e-01
-2.56875128e-01 4.66114968e-01 -5.99451184e-01 -1.10302605e-01
-1.22816527e+00 1.22166812e+00 3.10660124e-01 9.73634347e-02
-1.05341399e+00 -2.82343954e-01 -1.18422580e+00 -2.04796821e-01
2.48013720e-01 -7.46764183e-01 9.82242346e-01 -7.03853741e-02
-1.47202504e+00 2.03352317e-01 1.80311888e-01 -7.49599159e-01
6.43541992e-01 -1.44598931e-01 -3.58161271e-01 -1.24237649e-01
4.20235604e-01 7.88454115e-01 2.82736093e-01 -8.99393559e-01
-1.05258191e+00 -2.19528690e-01 5.95412850e-01 4.87370819e-01
5.10246873e-01 -4.14330333e-01 9.82691199e-02 3.19779478e-02
4.37056571e-01 -1.58970344e+00 -5.60159385e-01 3.16191055e-02
-1.24335662e-02 4.79177654e-01 1.22066426e+00 -5.62194362e-02
5.30517757e-01 -1.87043953e+00 -1.57226250e-01 1.34714067e-01
-3.59650403e-01 2.17433244e-01 3.52966398e-01 9.54195261e-01
3.84788334e-01 -1.20121896e-01 -1.07765943e-02 1.92179874e-01
4.88176569e-02 2.99299091e-01 -1.28468469e-01 5.67949116e-01
-6.13490522e-01 3.44648033e-01 -1.06742442e+00 -3.25661860e-02
6.00519300e-01 5.73545396e-01 -3.32390666e-01 -1.60758957e-01
1.43132182e-02 5.81031859e-01 -4.31953311e-01 7.94040263e-01
6.01583719e-01 6.23770177e-01 8.39782655e-02 4.74148929e-01
-1.18513966e+00 2.30061620e-01 -1.41818428e+00 1.79326606e+00
-6.34610474e-01 1.05793178e+00 5.62347054e-01 -2.23842993e-01
1.03482926e+00 1.09634265e-01 9.76364166e-02 -3.97964388e-01
2.00832129e-01 3.82297903e-01 -9.67469215e-02 -3.17754209e-01
1.15365958e+00 7.90661350e-02 -1.91501111e-01 -2.01337058e-02
-2.09086508e-01 -9.62352812e-01 1.98742196e-01 -1.20336385e-02
1.31394362e+00 6.17448747e-01 3.22248995e-01 -5.39378703e-01
4.23425049e-01 9.48355854e-01 1.85068220e-01 4.64487165e-01
-1.50350943e-01 2.46207595e-01 -9.00351182e-02 -5.95962286e-01
-6.41219318e-01 -7.23427355e-01 -2.45071277e-01 6.35118961e-01
6.09626830e-01 -5.21578074e-01 -5.41167080e-01 -6.63432330e-02
2.24709585e-02 7.77844012e-01 -1.11669332e-01 3.21551323e-01
-1.70935616e-02 -1.26070976e-01 4.13801491e-01 -3.81900102e-01
6.82587445e-01 -4.62699503e-01 -1.69881439e+00 3.29941541e-01
2.83311576e-01 -1.10944223e+00 5.44244051e-01 2.70485669e-01
-4.23965722e-01 -1.01935172e+00 -2.19287604e-01 -4.54757422e-01
4.77678448e-01 9.91135716e-01 4.05095786e-01 -1.40945120e-02
-3.75897676e-01 8.17404926e-01 -6.73392355e-01 -7.41376817e-01
1.20216504e-01 6.23283349e-02 5.21210790e-01 -6.62111998e-01
-1.19801804e-01 -5.00085831e-01 -7.35231876e-01 8.48467290e-01
-5.98493874e-01 -2.58024573e-01 6.40120059e-02 2.04511225e-01
4.02221501e-01 6.18356168e-01 2.03292131e-01 -2.57869065e-01
3.82975340e-01 -4.82374936e-01 -1.20148849e+00 -3.68325740e-01
-6.12163484e-01 -6.89245462e-01 6.32366002e-01 2.94348300e-01
-4.29988742e-01 5.85469186e-01 6.26763403e-02 1.49701819e-01
-4.68069971e-01 5.81624806e-01 -6.96662292e-02 -7.99303472e-01
5.76208711e-01 -1.22919334e-02 1.28107056e-01 1.47305712e-01
5.80452941e-02 7.76892781e-01 3.57528895e-01 1.57307208e-01
1.03159368e+00 7.11568236e-01 3.68016809e-01 -1.32665503e+00
9.37933251e-02 -2.51752287e-01 -7.46137083e-01 -3.44216228e-01
6.73982382e-01 -1.05536139e+00 -6.70261919e-01 -2.17911284e-02
-6.43657267e-01 -5.27456462e-01 -1.40387788e-01 5.97183645e-01
-4.96006876e-01 8.28686059e-02 3.92688423e-01 -8.91480267e-01
1.34354368e-01 -1.47231722e+00 7.28650570e-01 5.36144078e-01
-4.33984637e-01 -7.98695505e-01 2.01190218e-01 -1.10221229e-01
6.47376657e-01 7.07844675e-01 -1.54270986e-02 -1.91372290e-01
-1.00736654e+00 -8.03201795e-01 3.26473802e-01 -1.85795441e-01
2.05385283e-01 1.09319463e-02 -4.88686889e-01 -5.02811611e-01
-2.52120942e-01 1.19371533e-01 8.21627118e-03 8.11040178e-02
-1.31216019e-01 1.11400850e-01 -6.06924474e-01 6.62165403e-01
1.45232105e+00 7.13357329e-01 6.27605438e-01 8.55121255e-01
1.81978106e-01 5.10679722e-01 1.46594739e+00 1.20492056e-01
6.37021184e-01 7.23265290e-01 1.25612938e+00 2.37771824e-01
7.27581918e-01 -1.20791778e-01 2.79588342e-01 -8.49971324e-02
-1.84880614e-01 -4.68808353e-01 -1.03726709e+00 5.95618784e-01
-1.76453745e+00 -5.76799810e-01 -6.85324848e-01 2.44106817e+00
-4.93542492e-01 6.58055693e-02 -2.83351809e-01 2.51664706e-02
2.11056739e-01 -1.23641603e-02 -5.20233959e-02 -6.62523925e-01
2.80814201e-01 -2.24190965e-01 1.13158882e+00 6.91835880e-01
-1.03117108e+00 1.07584083e+00 5.91274691e+00 -2.35797986e-01
-1.15376580e+00 -1.98020667e-01 -5.77629089e-01 2.35442266e-01
9.36843380e-02 7.11374342e-01 -7.90534139e-01 2.23828807e-01
1.25828433e+00 -6.77478090e-02 4.49578702e-01 1.12427270e+00
5.33097804e-01 -7.52691150e-01 -4.49432641e-01 5.08466303e-01
-3.16039830e-01 -1.17822278e+00 -5.20275235e-01 4.62667376e-01
3.92276198e-01 3.26081872e-01 -2.63336718e-01 1.95767343e-01
3.97615433e-01 -7.33728886e-01 6.69755399e-01 4.67138469e-01
5.11109650e-01 -8.93256307e-01 8.52066219e-01 7.46814847e-01
-1.07972276e+00 -3.97609696e-02 -6.21087909e-01 -6.52037442e-01
4.48441625e-01 6.98796287e-02 -1.52161002e+00 8.28263998e-01
8.84065688e-01 3.87393445e-01 -3.76000434e-01 1.36556041e+00
-3.47699761e-01 -8.69241729e-02 -7.63202667e-01 -3.01631480e-01
6.88863575e-01 -6.62942588e-01 9.58323538e-01 5.18868983e-01
1.06666183e+00 2.92323120e-02 8.70164484e-02 3.03643137e-01
7.23721445e-01 -1.69700861e-01 -1.48176408e+00 2.23485529e-01
3.67755920e-01 1.50741494e+00 -7.14764714e-01 1.49114728e-01
6.29315823e-02 9.24072683e-01 -5.65463662e-01 2.05281034e-01
-6.18318677e-01 -1.02911937e+00 9.45136011e-01 4.70003188e-01
2.10203379e-01 -1.15606356e+00 4.02561069e-01 -6.36541724e-01
-5.59448183e-01 -4.58187222e-01 -2.76958346e-01 -1.40754998e+00
1.60660699e-01 7.31344104e-01 -7.76543394e-02 -1.76920879e+00
-5.99286556e-01 -6.54285014e-01 -3.84036511e-01 5.97374022e-01
-1.17820859e+00 -1.11768377e+00 -7.72491634e-01 4.75124598e-01
2.81786323e-01 -1.48458377e-01 1.15540779e+00 -1.38288125e-01
-3.82906497e-02 -4.30566758e-01 -3.42806987e-02 -5.40220022e-01
6.60165370e-01 -1.30398238e+00 3.45057696e-01 1.06125522e+00
-2.56670147e-01 7.55952001e-01 1.29700470e+00 -8.09052587e-01
-1.83217871e+00 -9.70519841e-01 5.69245219e-01 -4.39526647e-01
6.04909062e-01 -4.51740652e-01 -2.15682074e-01 9.85792577e-01
1.14106663e-01 -7.30761662e-02 3.02502394e-01 -5.47682106e-01
6.66044354e-01 -1.67103142e-01 -1.30466068e+00 9.57721174e-01
8.52413774e-01 -4.79997927e-03 -2.14515239e-01 4.25205559e-01
7.61550903e-01 -7.28931069e-01 -7.89448678e-01 4.16304290e-01
7.17119396e-01 -1.27602851e+00 7.69141674e-01 1.13973409e-01
-5.54509878e-01 -9.86513734e-01 -3.73415887e-01 -1.62149036e+00
-7.60014215e-03 -1.02922535e+00 7.38810062e-01 8.75023186e-01
3.44179869e-01 -1.09203362e+00 5.36682069e-01 6.76411569e-01
-5.59864759e-01 -1.83543786e-02 -1.13419604e+00 -8.27725410e-01
-6.08608305e-01 -4.12968636e-01 7.68076956e-01 5.27210951e-01
-5.27312644e-02 -8.60142112e-02 -1.90994710e-01 9.25563157e-01
1.79935724e-01 -7.98597410e-02 1.48789644e+00 -1.54950631e+00
5.71696520e-01 3.35177273e-01 -1.08915448e+00 -5.34717798e-01
-1.20793469e-01 -3.94702584e-01 2.22858801e-01 -1.91088653e+00
-1.41022074e+00 -7.34961987e-01 5.44476569e-01 3.94616812e-01
8.12891901e-01 6.78727925e-02 -8.37608054e-02 1.00300491e-01
-3.61133784e-01 2.57970721e-01 1.00209916e+00 3.56403589e-01
-5.02349854e-01 3.83456260e-01 -2.39476427e-01 6.88558996e-01
6.91381156e-01 -3.11503977e-01 -7.05002904e-01 2.31498107e-02
6.06983840e-01 2.80659109e-01 5.16513288e-01 -1.81176043e+00
6.31445169e-01 2.20933706e-02 1.62129939e-01 -6.85496092e-01
6.78321719e-01 -1.55166483e+00 5.78641236e-01 6.94196045e-01
6.21728361e-01 4.67539608e-01 2.99115241e-01 5.34298897e-01
-1.00190572e-01 -2.55681008e-01 3.88681948e-01 -2.65450537e-01
-9.29271042e-01 2.74014124e-03 -1.26060009e+00 -6.87140465e-01
1.52457416e+00 -5.43518066e-01 -3.93697917e-01 -6.72149777e-01
-5.40756404e-01 5.15033484e-01 1.34627712e+00 2.49119714e-01
6.17486715e-01 -7.35716641e-01 1.10239733e-03 5.62460840e-01
7.62867630e-02 1.87841982e-01 1.48063123e-01 6.55948281e-01
-1.54570854e+00 7.41360009e-01 -6.43761337e-01 -6.30602539e-01
-1.01109219e+00 2.12699041e-01 3.96465719e-01 4.88755792e-01
-5.89914620e-01 4.52482611e-01 -4.03749555e-01 -9.53250527e-01
-4.19820875e-01 -4.68918920e-01 -1.03578657e-01 -9.36382264e-02
3.40588152e-01 5.49204826e-01 3.08094144e-01 -8.19072723e-01
-6.00499213e-01 3.82269830e-01 5.71540296e-01 -4.48225498e-01
1.29505050e+00 -6.80351794e-01 -4.60954010e-02 3.14173251e-01
4.11394447e-01 2.89712459e-01 -1.06895995e+00 9.65183735e-01
-1.51716486e-01 -6.38492107e-01 3.27611327e-01 -4.20423716e-01
-2.70337939e-01 4.31645632e-01 8.78866553e-01 7.02485681e-01
6.92369401e-01 -8.90776157e-01 5.50392926e-01 9.34650004e-01
1.37796247e+00 -8.51033330e-01 -1.01149714e+00 7.42142856e-01
6.20147765e-01 -9.35989380e-01 2.29367793e-01 -3.02708536e-01
-6.16955042e-01 1.14465988e+00 3.01073879e-01 -3.18357378e-01
5.59184432e-01 2.03597263e-01 3.08973759e-01 -2.73723811e-01
-4.37874496e-01 -4.90805745e-01 -3.90787214e-01 1.10715306e+00
6.37905523e-02 -1.85521282e-02 -8.85445178e-02 -5.07208332e-02
-8.52782667e-01 2.65078302e-02 1.24156117e+00 1.58567476e+00
-7.50364244e-01 -8.55936527e-01 -5.63232422e-01 -1.74939662e-01
7.39089325e-02 3.26905757e-01 -2.31407940e-01 1.52337646e+00
3.63999605e-01 1.07679844e+00 -4.13790457e-02 -5.21439373e-01
6.07384264e-01 -1.03454992e-01 2.09825300e-03 -5.96555829e-01
-4.62908655e-01 -3.91611606e-01 3.61579359e-01 -6.70389831e-01
-2.07158938e-01 -6.48860693e-01 -1.39591348e+00 -1.36241287e-01
-1.79747671e-01 3.20777446e-01 1.51542854e+00 5.91276467e-01
6.00803018e-01 2.96557814e-01 4.00947779e-01 -1.28410149e+00
2.88246155e-01 -4.10725683e-01 -8.91295254e-01 -9.76599813e-01
4.03113037e-01 -8.86945426e-01 -3.74537587e-01 -5.15626192e-01] | [7.275373458862305, -1.8777130842208862] |
13cbb61a-309e-4cb2-a1f1-b0c14ea5e790 | fairness-in-representation-for-multilingual | null | null | https://openreview.net/forum?id=-llS6TiOew | https://openreview.net/pdf?id=-llS6TiOew | Fairness in Representation for Multilingual NLP: Insights from Controlled Experiments on Conditional Language Modeling | We perform systematically and fairly controlled experiments with the 6-layer Transformer to investigate whether languages which have been traditionally considered morphologically rich (AR and RU) and poor (ZH) are equally hard to conditional-language-model. We evaluate through statistical comparisons across 30 possible language directions from the 6 languages of the United Nations Parallel Corpus on 3 representation levels --- character, byte, and word. Results show that performance is relative to the representation granularity of each of the languages, not to the language as a whole. By eliminating statistically significant performance disparity on the character and byte levels, we show that performance disparity is not a necessary condition. The disparity that mirrors the morphological complexity hierarchy is a byproduct of word segmentation. Evidence from data statistics, along with the fact that word segmentation is qualitatively indeterminate, renders a decades-long debate on morphological complexity (unless it is being intentionally modeled in a word-based, meaning-driven context) irrelevant in the context of computing. The intent of our work is to help effect more objectivity and adequacy in evaluation as well as fairness and inclusivity in experimental setup in the area of language and computing so to uphold diversity in ML and AI research. Multilinguality is real and relevant in computing not due to canonical, structural linguistic concepts such as morphology or "words" in our minds, but rather standards related to internationalization and localization, such as character encoding --- something which has thus far been sorely overlooked in our discourse and curricula.
| ['Ada Wan'] | 2021-09-29 | null | null | null | iclr-2022-4 | ['multilingual-nlp'] | ['natural-language-processing'] | [ 6.90364093e-02 -1.68884933e-01 -2.94901222e-01 -2.92685270e-01
-5.32549739e-01 -7.14521110e-01 6.89557433e-01 5.13428807e-01
-9.97276843e-01 5.47917128e-01 8.09226751e-01 -1.13244724e+00
-2.01132208e-01 -7.05818057e-01 -4.28183109e-01 -3.84750456e-01
1.14749305e-01 4.24558938e-01 -1.80610031e-01 -4.11908031e-01
7.28131890e-01 5.98178327e-01 -1.41299701e+00 1.61144257e-01
1.03890753e+00 1.42468482e-01 9.08153132e-02 5.16613960e-01
-4.72506791e-01 6.57025039e-01 -5.98178506e-01 -5.31967461e-01
8.08341056e-02 -4.37609881e-01 -1.01173592e+00 -2.25179698e-02
3.85830849e-01 1.80108801e-01 3.55649620e-01 1.21111763e+00
2.65374571e-01 -2.36705095e-01 7.69389868e-01 -4.91098970e-01
-8.04336369e-01 9.32390630e-01 -3.49128485e-01 2.82449603e-01
4.19963300e-01 4.04720217e-01 1.21114302e+00 -6.10076368e-01
6.65385842e-01 1.59688568e+00 4.42578912e-01 1.96230143e-01
-1.17376554e+00 -4.05637860e-01 3.00677627e-01 -2.43901640e-01
-1.58891201e+00 -6.33988917e-01 3.23756427e-01 -6.99739695e-01
1.07763410e+00 5.02007961e-01 7.59632766e-01 5.87462783e-01
3.83279651e-01 2.61085421e-01 1.41255462e+00 -8.81803095e-01
2.72465199e-02 3.32401514e-01 1.61553040e-01 5.42993307e-01
8.76759112e-01 -6.44911453e-02 -3.15094948e-01 1.48093313e-01
7.34518588e-01 -7.89833128e-01 -1.94827020e-01 2.29553267e-01
-1.26969588e+00 8.36911678e-01 -9.14449543e-02 1.03457355e+00
-2.89969090e-02 -9.69137102e-02 5.76378405e-01 4.30459827e-01
1.58354580e-01 6.77258015e-01 -4.33701575e-01 -3.41167480e-01
-9.21669066e-01 2.63945639e-01 8.88597012e-01 5.28756976e-01
2.96353638e-01 4.82424200e-01 3.47374648e-01 6.95968390e-01
5.05746603e-01 4.57949698e-01 6.44838691e-01 -8.12269032e-01
4.73524392e-01 5.73056281e-01 -2.14882329e-01 -1.01094639e+00
-3.89259309e-01 -4.45161432e-01 -4.19203907e-01 2.23882005e-01
8.34598482e-01 1.04095772e-01 -6.62918508e-01 1.98188353e+00
-4.79250681e-03 -8.46028566e-01 6.10722415e-02 7.15643704e-01
4.84519392e-01 6.42048240e-01 6.03266537e-01 -3.62911314e-01
1.66760480e+00 -1.75518945e-01 -4.41084415e-01 -2.54078418e-01
1.16010737e+00 -9.52758670e-01 1.64904034e+00 5.28240025e-01
-1.36332178e+00 -4.20467436e-01 -1.27187085e+00 -3.32461566e-01
-5.98886251e-01 -1.83312654e-01 8.07279348e-01 1.22946692e+00
-1.10473835e+00 3.87513489e-01 -6.95067704e-01 -4.50261146e-01
-4.53128368e-02 1.52267024e-01 -2.41575733e-01 2.38778397e-01
-9.75105643e-01 1.37309301e+00 6.09640360e-01 5.77656226e-03
-1.88671142e-01 -1.94655165e-01 -8.66512179e-01 -1.32648245e-01
3.59442458e-02 -3.34219307e-01 7.96723843e-01 -1.22077692e+00
-1.00099921e+00 1.18981397e+00 2.79651675e-02 -1.55451775e-01
3.30738455e-01 9.25142765e-02 -6.31155550e-01 -3.16831172e-01
8.53777975e-02 5.16104758e-01 2.17647329e-01 -1.35949206e+00
-6.33898973e-01 -5.33287525e-01 8.10797960e-02 4.63013262e-01
-8.90975744e-02 5.82724988e-01 1.06880479e-01 -6.33492768e-01
2.72693545e-01 -6.62668169e-01 -1.30818948e-01 -5.55747151e-01
-6.38192073e-02 -3.20640653e-01 -6.01410829e-02 -8.97462964e-01
1.56530058e+00 -2.01776123e+00 -1.57946482e-01 3.07913333e-01
-5.94947003e-02 2.96360794e-02 1.98098551e-02 4.57533628e-01
-2.60783583e-01 7.94080377e-01 -1.82137519e-01 1.04561776e-01
2.91337460e-01 3.61061633e-01 -8.63160416e-02 8.28315496e-01
2.42658675e-01 6.26825929e-01 -6.85723484e-01 -7.21081138e-01
5.08668087e-02 2.66467959e-01 -6.71012521e-01 -6.43088520e-01
2.36770641e-02 7.50416145e-02 -1.88194782e-01 7.07517803e-01
3.74738604e-01 2.77137756e-01 4.14793849e-01 1.42987534e-01
-7.77271092e-01 8.08048248e-01 -1.13918602e+00 1.48829901e+00
-4.77873832e-01 4.81662333e-01 -3.61805037e-02 -8.84557843e-01
5.72822392e-01 1.38794497e-01 -1.79015204e-01 -1.03542197e+00
2.81466305e-01 6.63975716e-01 9.44610775e-01 -2.57611752e-01
9.89841700e-01 -4.16180581e-01 -3.34644496e-01 3.05358231e-01
-1.79464728e-01 -4.87078875e-01 5.48427582e-01 -1.30739240e-02
4.89462882e-01 2.52139956e-01 3.36459488e-01 -1.11945879e+00
4.82763678e-01 1.75248846e-01 6.59037292e-01 1.90228835e-01
2.73979641e-02 3.01538646e-01 5.35108984e-01 -7.34257624e-02
-1.31774759e+00 -1.02974212e+00 -6.05378926e-01 1.05797374e+00
-1.91783011e-01 -3.07789594e-01 -6.83902025e-01 -1.65253761e-03
-1.61029190e-01 1.22084558e+00 -2.25698709e-01 1.82270765e-01
-8.52614343e-01 -9.32194173e-01 6.95272207e-01 2.61828750e-01
-3.31940651e-02 -9.50776339e-01 -8.32917929e-01 2.74699777e-01
3.31458837e-01 -8.57210457e-01 -1.12304889e-01 2.23200262e-01
-8.16554606e-01 -5.95443726e-01 -3.63839269e-01 -8.17709029e-01
3.99737716e-01 -2.82699257e-01 1.24288297e+00 3.09769988e-01
-5.72138885e-03 2.30949610e-01 -2.12068602e-01 -5.42950273e-01
-8.45415115e-01 -7.96169564e-02 -2.76605096e-02 -5.83326638e-01
4.51618731e-01 -3.79181355e-01 -1.59519479e-01 -2.92038918e-01
-1.10604370e+00 -2.30070487e-01 4.59316730e-01 4.28312987e-01
1.48134857e-01 2.37771738e-02 4.37365919e-01 -1.02537310e+00
7.48509049e-01 -3.21127832e-01 -4.63082433e-01 2.53250927e-01
-7.41738558e-01 5.58445230e-02 2.98531592e-01 -2.06852615e-01
-9.93312180e-01 -5.58752477e-01 -3.54837596e-01 6.23864889e-01
-2.47589588e-01 8.03016841e-01 -3.99476916e-01 1.97114557e-01
7.42243409e-01 -5.24124391e-02 -4.32568118e-02 -3.79971236e-01
4.01644081e-01 6.25077844e-01 5.97891390e-01 -1.10520923e+00
5.99513650e-01 1.38248622e-01 -2.26546407e-01 -1.27793384e+00
-2.36549124e-01 -2.23831534e-01 -5.74090362e-01 1.15501218e-01
9.95665491e-01 -9.78725612e-01 -4.24613297e-01 1.50098829e-02
-1.12928414e+00 -1.68877229e-01 -3.54171544e-01 6.30355000e-01
-2.89857298e-01 5.01451254e-01 -7.91644275e-01 -9.44757938e-01
-7.53869414e-02 -1.25142539e+00 5.84399223e-01 -2.15303265e-02
-8.26834679e-01 -1.33357143e+00 -1.80155799e-01 2.58854717e-01
3.16462338e-01 1.98414341e-01 1.65492213e+00 -7.07365453e-01
-2.00912029e-01 1.48262307e-01 -1.71066508e-01 3.18358958e-01
-4.60380800e-02 4.18765903e-01 -7.30661988e-01 -1.27335221e-01
2.24704161e-01 -3.18690270e-01 4.54126924e-01 4.96034399e-02
4.31318283e-01 -3.17315459e-01 3.24666977e-01 2.75117040e-01
1.84904385e+00 3.68149370e-01 7.28744864e-01 5.61498106e-01
5.18383622e-01 1.07210803e+00 1.31353423e-01 8.36348981e-02
6.16592526e-01 2.13333160e-01 -1.70676962e-01 -1.70978829e-01
-1.63992196e-01 -3.12873013e-02 4.83384788e-01 1.51182973e+00
-6.50436208e-02 -1.75623730e-01 -1.38975108e+00 7.94359446e-01
-1.09443450e+00 -7.55013347e-01 -4.36371416e-01 2.41996789e+00
1.11840594e+00 5.02847433e-01 8.70549381e-02 3.37357581e-01
5.09186685e-01 1.76352784e-01 2.96650857e-01 -1.31649387e+00
-3.71794701e-01 4.61985558e-01 4.70470995e-01 8.68877470e-01
-4.51754749e-01 1.19080865e+00 6.32537651e+00 9.28574800e-01
-1.14652205e+00 -7.82595724e-02 7.02420831e-01 2.47096002e-01
-9.02714133e-01 2.27508515e-01 -6.20271325e-01 3.01091313e-01
1.05876207e+00 -4.22010750e-01 2.18508884e-01 3.16534340e-01
2.68331558e-01 -4.69965845e-01 -1.16012549e+00 4.40889686e-01
-1.77217480e-02 -8.59223008e-01 2.72306442e-01 2.50860393e-01
4.64538157e-01 -3.99706364e-02 2.51494378e-01 1.93800464e-01
3.69513094e-01 -1.42442691e+00 1.46836817e+00 -7.98369059e-04
9.15301681e-01 -8.35131884e-01 5.55778623e-01 3.12917739e-01
-8.93962145e-01 2.96685129e-01 -4.06848818e-01 -5.29713511e-01
-4.42366628e-03 3.09385300e-01 -3.44370425e-01 4.28676188e-01
9.13558826e-02 -1.32833168e-01 -6.90075576e-01 6.49425864e-01
1.18297888e-02 8.45952928e-01 -3.12762380e-01 -4.39656638e-02
5.55909216e-01 -3.40672910e-01 4.72868711e-01 1.79454398e+00
4.49772254e-02 7.60080367e-02 -1.88407768e-02 8.35013151e-01
5.28363943e-01 7.44223475e-01 -5.58466434e-01 -4.20213610e-01
5.00406623e-01 8.39721680e-01 -1.14974058e+00 -3.98805775e-02
-7.11553633e-01 3.67416501e-01 2.90789697e-02 1.77437812e-01
-3.14407498e-01 -2.15670049e-01 4.67579931e-01 2.43504271e-01
-1.58043087e-01 -5.96598625e-01 -9.32053089e-01 -9.88779306e-01
-1.21860251e-01 -1.23834205e+00 1.87236831e-01 -2.06204370e-01
-1.07835090e+00 4.81764078e-01 7.25467354e-02 -5.58103979e-01
-3.37767899e-01 -9.57485795e-01 -2.88381726e-01 1.30510366e+00
-1.20237780e+00 -1.01633286e+00 5.27948081e-01 2.66480029e-01
2.22393408e-01 1.41872251e-02 8.72447610e-01 1.25531644e-01
-2.90058553e-01 6.22922957e-01 -8.67660567e-02 2.58024305e-01
2.18540668e-01 -1.26192212e+00 4.74772125e-01 9.61880922e-01
4.00352180e-01 1.03756225e+00 7.48185992e-01 -5.75083375e-01
-1.26223910e+00 -3.49726498e-01 1.48994172e+00 -6.45145416e-01
9.58246469e-01 -2.35573873e-01 -1.01818502e+00 3.32264215e-01
4.17260319e-01 -6.95721745e-01 7.42903531e-01 4.24492449e-01
-3.85599732e-01 2.21527651e-01 -9.70242321e-01 8.72959077e-01
8.67805839e-01 -6.98008060e-01 -9.24946368e-01 -4.03323956e-02
7.96011269e-01 -5.69054261e-02 -7.86553562e-01 2.20123261e-01
5.67921281e-01 -8.97339880e-01 5.62800825e-01 -5.32430887e-01
3.09851021e-01 -2.27956753e-02 -4.92688924e-01 -1.03029060e+00
-3.22687149e-01 -3.28487009e-01 1.05963695e+00 1.44141209e+00
7.48686433e-01 -7.09707081e-01 5.56649506e-01 5.73160827e-01
-4.27209914e-01 -4.48145479e-01 -9.66145277e-01 -6.59962714e-01
1.03014064e+00 -8.67358387e-01 6.06381655e-01 1.29557443e+00
2.36443430e-01 5.47066987e-01 5.48129916e-01 -1.66676357e-01
2.65165955e-01 -2.17415765e-01 3.77014220e-01 -1.12826109e+00
-3.14630479e-01 -1.12095165e+00 -4.85197157e-01 -3.49555075e-01
2.77633537e-02 -1.04688478e+00 -1.61042422e-01 -1.39601469e+00
-3.45360823e-02 -5.91837406e-01 4.21809331e-02 1.10183386e-02
9.52684134e-02 -5.77492118e-02 4.90057230e-01 5.37405461e-02
1.67101547e-02 -4.54549305e-02 1.16320658e+00 2.78293103e-01
-3.17710452e-02 -5.35330951e-01 -1.15457296e+00 1.09612572e+00
7.64124811e-01 -7.28094056e-02 -3.77764225e-01 -8.03444386e-01
6.96201921e-01 -3.97607461e-02 -4.60207686e-02 -9.64467824e-01
-3.57127078e-02 -4.15536791e-01 3.21769387e-01 -1.06487416e-01
-3.37630026e-02 -7.10048676e-01 1.03418753e-01 5.31571388e-01
-2.50813961e-01 6.44441485e-01 3.79532903e-01 -3.00482541e-01
-4.33852784e-02 -5.15619159e-01 6.30183518e-01 -4.42969978e-01
-5.85594118e-01 -3.03426832e-01 -4.90164131e-01 4.74766612e-01
5.25740981e-01 -6.08302236e-01 -3.13877672e-01 1.04862489e-02
-2.81890959e-01 -4.30636704e-02 9.74946201e-01 1.63423821e-01
1.62209570e-01 -8.85404468e-01 -1.06346488e+00 1.22613430e-01
2.75686122e-02 -2.78890699e-01 -1.96649969e-01 4.88782763e-01
-1.03047407e+00 5.42827785e-01 -1.87564313e-01 -1.26240447e-01
-8.04008424e-01 4.53158051e-01 1.04377769e-01 -1.42764404e-01
-3.55239153e-01 8.26486468e-01 1.53937787e-01 -2.79494494e-01
2.79084921e-01 -5.29721320e-01 -2.64938146e-01 2.25893334e-01
1.36912942e-01 2.40605995e-01 6.73687551e-03 -1.04683614e+00
-3.20347518e-01 4.01448697e-01 4.28876579e-02 -5.83486855e-01
8.75836313e-01 -7.86507726e-02 -3.81855756e-01 8.56370151e-01
8.91557395e-01 6.72498107e-01 -3.30355704e-01 2.35659018e-01
5.99246979e-01 -3.72433275e-01 -7.60665759e-02 -8.73605072e-01
-2.74682879e-01 9.22144711e-01 2.73534566e-01 2.55062103e-01
6.42405927e-01 -3.10855955e-01 3.15531582e-01 -8.53753090e-02
4.08868104e-01 -1.47012579e+00 -5.71559966e-01 5.37646890e-01
8.14342260e-01 -7.18362272e-01 2.46810764e-01 -2.42154658e-01
-6.11162484e-01 7.13665664e-01 4.46605563e-01 1.04090571e-01
3.08684498e-01 5.53196788e-01 9.34045836e-02 -1.61828753e-02
-7.59334207e-01 -4.03953105e-01 6.70463890e-02 3.88800204e-01
1.24928713e+00 4.30689216e-01 -1.28574800e+00 3.26173574e-01
-8.53807271e-01 -5.76783597e-01 5.31362355e-01 8.56903911e-01
-6.64257586e-01 -1.25983500e+00 -6.41006649e-01 2.53824651e-01
-9.28724587e-01 -5.35914063e-01 -3.16355169e-01 1.36113346e+00
5.71393549e-01 8.21219444e-01 4.39949125e-01 -1.00662112e-01
1.12276105e-02 2.29782149e-01 6.12431824e-01 -7.91968763e-01
-8.70668113e-01 3.35184664e-01 5.80733240e-01 -3.19482610e-02
-2.15285450e-01 -8.45112205e-01 -1.35747719e+00 -7.42302120e-01
-2.34834284e-01 3.26843232e-01 9.11726594e-01 9.29794610e-01
-4.17809486e-01 1.44436002e-01 -8.31600353e-02 -3.89192641e-01
-7.47969747e-01 -8.56128812e-01 -7.90082037e-01 1.27641320e-01
-1.36143595e-01 -1.66189969e-01 -4.32846576e-01 -1.08715676e-01] | [10.563170433044434, 9.958029747009277] |
c70d0ea5-2f28-4b42-909a-ff139aeadd3e | fedabc-targeting-fair-competition-in | 2302.07450 | null | https://arxiv.org/abs/2302.07450v1 | https://arxiv.org/pdf/2302.07450v1.pdf | FedABC: Targeting Fair Competition in Personalized Federated Learning | Federated learning aims to collaboratively train models without accessing their client's local private data. The data may be Non-IID for different clients and thus resulting in poor performance. Recently, personalized federated learning (PFL) has achieved great success in handling Non-IID data by enforcing regularization in local optimization or improving the model aggregation scheme on the server. However, most of the PFL approaches do not take into account the unfair competition issue caused by the imbalanced data distribution and lack of positive samples for some classes in each client. To address this issue, we propose a novel and generic PFL framework termed Federated Averaging via Binary Classification, dubbed FedABC. In particular, we adopt the ``one-vs-all'' training strategy in each client to alleviate the unfair competition between classes by constructing a personalized binary classification problem for each class. This may aggravate the class imbalance challenge and thus a novel personalized binary classification loss that incorporates both the under-sampling and hard sample mining strategies is designed. Extensive experiments are conducted on two popular datasets under different settings, and the results demonstrate that our FedABC can significantly outperform the existing counterparts. | ['DaCheng Tao', 'Yonggang Wen', 'Kehua Su', 'Han Hu', 'Yong Luo', 'Li Shen', 'Dui Wang'] | 2023-02-15 | null | null | null | null | ['personalized-federated-learning'] | ['methodology'] | [-1.94858953e-01 -3.63985360e-01 -3.12709481e-01 -9.27683532e-01
-1.06650686e+00 -2.26482064e-01 4.49748747e-02 1.48445240e-03
-1.20296150e-01 9.45864260e-01 -2.93629128e-03 -1.25662491e-01
-2.44181484e-01 -8.01586151e-01 -6.34432912e-01 -1.11988366e+00
2.76385307e-01 4.47924078e-01 -1.05295248e-01 2.33627617e-01
-2.79482850e-03 2.16705203e-01 -1.48200715e+00 4.17402625e-01
1.32480526e+00 1.45199740e+00 -1.87627390e-01 1.02877446e-01
-3.44321847e-01 8.80098701e-01 -6.65096700e-01 -7.33462453e-01
5.25538266e-01 -1.89393729e-01 -4.65492398e-01 1.06222078e-01
4.20759559e-01 -5.54635704e-01 -1.31431535e-01 1.15325606e+00
7.27545917e-01 4.83837388e-02 2.73102134e-01 -1.60289979e+00
-3.96240324e-01 7.05042958e-01 -7.74950147e-01 9.89641696e-02
-3.44492495e-01 1.86745197e-01 1.05252612e+00 -7.31556177e-01
5.93619049e-02 1.11058080e+00 7.21091568e-01 4.38706160e-01
-1.10881281e+00 -9.09172654e-01 5.34271538e-01 3.81973267e-01
-1.26490724e+00 -4.24353570e-01 8.15290809e-01 -2.44969890e-01
3.08847129e-01 5.95433652e-01 2.12444231e-01 6.73743248e-01
-1.52246028e-01 1.06265378e+00 1.01579857e+00 -6.42464608e-02
4.50530797e-01 3.91244173e-01 2.92110234e-01 2.07334220e-01
2.44514868e-01 -2.26929262e-01 -4.07389998e-01 -7.55666852e-01
1.62349746e-01 5.83664536e-01 -1.80962831e-01 -5.86198390e-01
-5.20796657e-01 8.11928213e-01 3.10700595e-01 -1.43358067e-01
-4.85800952e-01 -2.48190597e-01 7.43477404e-01 3.54315609e-01
8.64369392e-01 -2.55873233e-01 -8.50529313e-01 1.68508783e-01
-7.93402493e-01 3.76849085e-01 7.44811356e-01 6.18701279e-01
1.01402974e+00 -2.19334960e-01 -3.52329344e-01 1.06151962e+00
2.86302045e-02 1.25741348e-01 7.31310844e-01 -8.98084939e-01
8.97297680e-01 8.66413236e-01 6.48210570e-02 -9.91731048e-01
4.74633649e-02 -8.70558739e-01 -9.58135366e-01 1.80821884e-02
2.73895591e-01 -3.57080609e-01 -2.26695612e-01 1.87362337e+00
9.15127754e-01 1.38105273e-01 1.69756487e-02 9.25587237e-01
2.89813489e-01 2.74490744e-01 7.31298923e-02 -4.26058710e-01
9.00392771e-01 -1.22649550e+00 -7.04648674e-01 1.19107626e-01
7.36506581e-01 -5.63237727e-01 9.02428210e-01 4.53274339e-01
-8.36683452e-01 -1.18547967e-02 -7.09320366e-01 3.49883288e-01
-7.20287338e-02 -4.81932238e-02 7.49635994e-01 7.70685136e-01
-3.74471664e-01 5.80933571e-01 -5.71757078e-01 1.88068241e-01
1.04375005e+00 3.51243585e-01 -2.79537767e-01 -4.13607001e-01
-9.07322228e-01 2.70161688e-01 -1.09391555e-01 -4.57619727e-02
-6.25086367e-01 -9.76502120e-01 -3.25961620e-01 1.35232329e-01
4.96565729e-01 -5.54170310e-01 1.38482988e+00 -1.22322345e+00
-1.05622077e+00 4.84024704e-01 -9.44220051e-02 -4.03444946e-01
9.80663598e-01 -2.77065802e-02 -2.64774382e-01 -4.15437698e-01
2.42357887e-02 -2.31644616e-01 6.05187774e-01 -1.13943231e+00
-9.83309090e-01 -1.03575635e+00 -1.30329534e-01 2.23109096e-01
-6.34989321e-01 -9.03862901e-03 -1.59433499e-01 -6.95837379e-01
1.50402367e-01 -3.70436072e-01 -4.07705605e-01 4.18316647e-02
-3.52732599e-01 -4.29933518e-01 1.13696313e+00 -3.98025751e-01
1.30067897e+00 -2.05355287e+00 -4.69875723e-01 6.03886694e-02
2.94544995e-01 2.61622310e-01 -6.93220124e-02 2.18237802e-01
6.01854287e-02 -6.52413294e-02 -1.45861104e-01 -6.86434150e-01
1.29364640e-01 3.83708090e-01 -2.44236335e-01 6.67056262e-01
-2.53276646e-01 4.16629225e-01 -6.03932202e-01 -6.65622830e-01
-2.29151815e-01 2.95402288e-01 -8.75052094e-01 4.62006301e-01
-2.95155704e-01 3.47219974e-01 -7.23747551e-01 8.28467429e-01
1.30563211e+00 -3.48174453e-01 2.63397634e-01 -1.33731022e-01
5.01863100e-02 2.93776006e-01 -1.35537267e+00 1.24469852e+00
-4.05181855e-01 -1.44084036e-01 5.63879073e-01 -1.32252717e+00
7.50584364e-01 2.88830698e-01 7.87027180e-01 -6.77679420e-01
3.68199088e-02 3.79530936e-01 -3.17978740e-01 -6.50087357e-01
1.23688420e-02 -2.77321696e-01 1.72440350e-01 7.78939545e-01
-2.56111562e-01 6.55472279e-01 -2.25239396e-01 4.83718216e-02
1.01544392e+00 -2.65101939e-01 -1.10316202e-01 -7.55639225e-02
5.67522287e-01 -2.50048637e-01 1.43785942e+00 7.78210700e-01
-6.46654546e-01 5.12642801e-01 4.52042401e-01 -5.72602153e-01
-6.40488565e-01 -6.20429635e-01 -1.56721652e-01 1.50181961e+00
1.55199662e-01 -1.45053193e-01 -6.57096565e-01 -1.09836566e+00
3.44970137e-01 4.37523127e-01 -3.67522210e-01 -1.39305502e-01
-2.94979334e-01 -1.24195480e+00 2.76327103e-01 2.77845085e-01
4.50063765e-01 -5.43515682e-01 -1.46771580e-01 3.53602797e-01
-5.11171937e-01 -6.17249012e-01 -6.73550010e-01 2.73069799e-01
-9.23815668e-01 -1.12090051e+00 -4.91238624e-01 -4.24393475e-01
6.02403820e-01 5.43799520e-01 8.69559228e-01 -1.32931083e-01
-3.20694655e-01 -1.00151986e-01 -2.54263133e-01 -4.18891639e-01
7.14168772e-02 9.31472555e-02 5.22973053e-02 6.89454317e-01
5.36144555e-01 -6.87503159e-01 -9.76266325e-01 5.29276669e-01
-9.28079963e-01 -3.04173857e-01 2.62827575e-01 1.04519522e+00
4.81751859e-01 3.61337692e-01 1.00870585e+00 -1.11100423e+00
5.05049229e-01 -9.03779030e-01 -3.19382310e-01 4.72486079e-01
-9.00320172e-01 -2.55214781e-01 9.08142507e-01 -3.80303800e-01
-1.23547518e+00 -1.45705491e-01 3.95574607e-02 -5.28781533e-01
4.63211127e-02 1.74742743e-01 -7.38475859e-01 -7.84172490e-02
3.41837466e-01 6.21911138e-02 3.05106431e-01 -1.05036914e+00
-6.19725473e-02 1.14105105e+00 2.09431306e-01 -6.19956315e-01
4.95740712e-01 4.70801264e-01 -3.74226093e-01 1.30893394e-01
-9.94488895e-01 -4.41730380e-01 5.19191585e-02 7.53132403e-02
3.54005173e-02 -1.00076759e+00 -7.75048971e-01 7.02820480e-01
-8.68082225e-01 2.50713788e-02 -2.99585223e-01 3.99160385e-01
-3.85179341e-01 4.46285427e-01 -5.51482499e-01 -1.03430128e+00
-8.09368312e-01 -1.16133726e+00 5.84207356e-01 4.04058188e-01
3.03205073e-01 -5.81985533e-01 3.02759744e-02 7.74018288e-01
6.64534152e-01 1.40501270e-02 9.26891088e-01 -1.09047246e+00
-4.31421340e-01 -5.24522364e-01 -3.88624281e-01 6.37280524e-01
2.57845044e-01 -1.98015153e-01 -1.02433729e+00 -4.71280158e-01
2.93629557e-01 -5.56726336e-01 4.79206473e-01 -4.98328395e-02
1.64747572e+00 -8.70574832e-01 -1.76915869e-01 6.31590366e-01
1.51666927e+00 -1.40894413e-01 9.85705629e-02 2.17923820e-01
5.18854499e-01 7.82495618e-01 6.99169874e-01 1.09040654e+00
6.52784884e-01 6.27934754e-01 8.11248541e-01 6.18768968e-02
3.26162547e-01 -8.33031833e-02 3.31884399e-02 5.72770119e-01
5.04253924e-01 -4.46978174e-02 -5.80386698e-01 5.01861334e-01
-2.10507488e+00 -7.40589261e-01 -5.56941815e-02 2.42487907e+00
1.08273423e+00 -4.17428941e-01 1.05125055e-01 1.21934548e-01
9.20388877e-01 1.17888777e-02 -9.24553752e-01 -2.62082756e-01
-3.23807806e-01 -1.29080653e-01 4.86303777e-01 3.23624685e-02
-1.06441784e+00 3.16567093e-01 5.03277922e+00 1.16941798e+00
-1.21094930e+00 4.28705484e-01 1.10242534e+00 -3.83317351e-01
-2.85529375e-01 -3.66463140e-02 -8.14933598e-01 9.38047826e-01
5.42405903e-01 -3.12540352e-01 3.96519214e-01 1.21533167e+00
2.07264572e-01 1.28087133e-01 -7.24723279e-01 1.04104328e+00
-1.92193270e-01 -1.11295927e+00 -1.47190198e-01 1.12337545e-01
8.51964712e-01 1.89760968e-01 1.53943049e-02 3.61937791e-01
4.01693553e-01 -6.77469969e-01 5.07949769e-01 5.35495996e-01
3.75718206e-01 -1.01197457e+00 6.49306595e-01 7.69398868e-01
-7.38728642e-01 -5.52894652e-01 -4.70865190e-01 8.61429721e-02
-2.84808040e-01 1.02670372e+00 -2.65996605e-01 6.56971335e-01
1.01022935e+00 2.77639925e-01 -4.08553571e-01 1.37251103e+00
5.93932390e-01 5.79251289e-01 -4.50450152e-01 3.40688109e-01
-1.74737275e-02 -2.26670265e-01 3.18850100e-01 6.51574075e-01
4.67899404e-02 -1.28077775e-01 2.62560636e-01 5.00844240e-01
-2.53691435e-01 5.94520926e-01 -8.40213448e-02 4.84391183e-01
5.53954542e-01 1.23921561e+00 -1.83945000e-02 -3.28321934e-01
-4.92783815e-01 6.12335205e-01 5.44819415e-01 2.02498049e-01
-7.31760144e-01 -3.78479987e-01 1.05047309e+00 -1.46875709e-01
3.42184961e-01 4.98298407e-01 -5.14780819e-01 -1.39115775e+00
3.95654172e-01 -1.12988091e+00 8.57142568e-01 -5.27576767e-02
-1.98697901e+00 3.76855522e-01 -6.30287409e-01 -1.08818913e+00
4.90304008e-02 7.54112154e-02 -8.22642326e-01 9.30437863e-01
-1.55251682e+00 -8.54060650e-01 -3.70730847e-01 8.47641289e-01
2.50855744e-01 -4.48636971e-02 5.99275947e-01 9.30628598e-01
-1.12988269e+00 1.13165390e+00 6.32350564e-01 -2.23463118e-01
9.52999532e-01 -8.83407176e-01 -3.02946389e-01 4.70390558e-01
-5.14248848e-01 3.82001311e-01 4.89072502e-01 -2.78151691e-01
-1.27711546e+00 -1.41693950e+00 8.50798070e-01 4.58343215e-02
2.63363719e-01 -1.76952228e-01 -1.20347798e+00 4.83077407e-01
-2.69689500e-01 5.85999429e-01 1.15769458e+00 7.12625235e-02
-6.33000135e-01 -9.03066933e-01 -1.79157817e+00 2.06920132e-01
8.73350859e-01 -1.55384466e-01 4.23379876e-02 7.13233471e-01
6.76453769e-01 -4.42990698e-02 -8.41461957e-01 6.10950708e-01
4.12395298e-01 -1.21298480e+00 6.57266796e-01 -8.87923241e-01
1.45173550e-01 -8.86383578e-02 -4.72854227e-01 -1.00718045e+00
-6.71666712e-02 -4.86083865e-01 -3.10171664e-01 1.71858776e+00
8.83467197e-02 -1.08161867e+00 1.15768838e+00 1.15840125e+00
1.88940316e-01 -1.15275073e+00 -1.17471886e+00 -6.33512080e-01
1.34620860e-01 -2.82663733e-01 1.26687717e+00 1.05779600e+00
-1.58936292e-01 -2.07243085e-01 -4.60447013e-01 8.73922482e-02
1.03435314e+00 5.53845167e-01 1.01042402e+00 -1.16596103e+00
-4.57121283e-01 -3.01486671e-01 -6.26092106e-02 -7.64365017e-01
9.71538126e-02 -6.90156579e-01 -1.78863451e-01 -8.71167541e-01
4.52864349e-01 -1.22052610e+00 -5.96466601e-01 7.17105031e-01
-3.46414715e-01 1.32357493e-01 5.40741682e-02 6.16572201e-01
-8.99344444e-01 8.45415890e-01 9.15329695e-01 -6.13628961e-02
-1.91434752e-02 5.37501276e-01 -1.07024431e+00 3.46360177e-01
7.97490001e-01 -6.26131058e-01 -1.80276915e-01 -3.87838811e-01
-1.12024166e-01 7.97916874e-02 1.72201902e-01 -8.14818919e-01
3.46749455e-01 -4.04757947e-01 1.64083034e-01 -6.57578051e-01
-6.82626516e-02 -1.05567670e+00 1.84625790e-01 2.70215005e-01
-1.65051892e-01 -3.14287305e-01 -2.74150431e-01 7.70891666e-01
-2.27967218e-01 -5.83704887e-03 9.09476936e-01 -2.01132316e-02
1.89519711e-02 6.85536027e-01 2.27878690e-01 4.02805693e-02
1.16496944e+00 7.43656456e-02 -4.76201355e-01 -2.37818569e-01
-4.37351167e-01 7.51618862e-01 5.24919748e-01 1.08628973e-01
-3.46211880e-03 -1.35727489e+00 -6.57338738e-01 3.14541310e-01
1.19819202e-01 3.37909371e-01 8.81970823e-01 9.89899576e-01
-6.61712214e-02 1.89654052e-01 1.47912517e-01 -3.90395135e-01
-1.15824497e+00 6.25745475e-01 5.94917417e-01 -5.02294421e-01
-3.32490176e-01 8.25042486e-01 2.20249772e-01 -9.19129193e-01
6.68351769e-01 4.58505452e-01 1.85017481e-01 1.58781201e-01
6.39760137e-01 6.92530334e-01 5.30342340e-01 -3.48050326e-01
-3.53871733e-01 4.31507379e-02 -4.63443965e-01 5.06494224e-01
1.49555004e+00 -4.24369544e-01 -4.08953726e-01 2.23795816e-01
1.40371406e+00 -3.04952953e-02 -1.50363827e+00 -7.81818032e-01
-3.20494622e-01 -8.75861645e-01 -1.47023527e-02 -8.40360463e-01
-1.64767337e+00 5.29540181e-01 5.82681119e-01 1.73036873e-01
1.30185091e+00 -3.54404211e-01 1.11194336e+00 -1.63394250e-02
5.66698253e-01 -1.11386943e+00 -3.44029844e-01 2.15722546e-01
4.19256032e-01 -1.39996886e+00 -1.42987341e-01 -1.94946289e-01
-5.78594804e-01 7.21529961e-01 8.79708946e-01 1.41297877e-01
6.74507439e-01 1.60946362e-02 2.65474230e-01 1.97157621e-01
-1.13181496e+00 4.67920423e-01 -3.03701490e-01 2.13366225e-01
7.70032331e-02 1.99762672e-01 -5.57332337e-01 1.19275582e+00
1.78010955e-01 1.03872649e-01 -1.30946143e-02 1.15674162e+00
-9.17474478e-02 -1.30832982e+00 -5.96917391e-01 8.55008602e-01
-8.53824019e-01 2.76094854e-01 -9.03180242e-02 -1.46192806e-02
2.91124135e-01 9.55139756e-01 -2.48820940e-03 -2.67194688e-01
1.79333493e-01 2.36100674e-01 -2.03947108e-02 -3.36327225e-01
-8.85411441e-01 1.93471536e-02 -3.72771055e-01 -5.88527083e-01
-1.07207395e-01 -6.01172924e-01 -8.93103302e-01 -5.69626033e-01
-5.20080149e-01 4.47573334e-01 7.42501259e-01 7.33929992e-01
8.25681567e-01 9.57728252e-02 1.51347661e+00 -5.84783792e-01
-1.48093367e+00 -7.66613126e-01 -8.82893443e-01 4.30988133e-01
3.68964434e-01 -5.12089550e-01 -7.70142317e-01 -5.11388838e-01] | [5.8370747566223145, 6.304182529449463] |
4214ba75-d1e6-454f-9162-af43cb845592 | unsupervised-object-level-video-summarization | 1801.00543 | null | http://arxiv.org/abs/1801.00543v2 | http://arxiv.org/pdf/1801.00543v2.pdf | Unsupervised Object-Level Video Summarization with Online Motion Auto-Encoder | Unsupervised video summarization plays an important role on digesting,
browsing, and searching the ever-growing videos every day, and the underlying
fine-grained semantic and motion information (i.e., objects of interest and
their key motions) in online videos has been barely touched. In this paper, we
investigate a pioneer research direction towards the fine-grained unsupervised
object-level video summarization. It can be distinguished from existing
pipelines in two aspects: extracting key motions of participated objects, and
learning to summarize in an unsupervised and online manner. To achieve this
goal, we propose a novel online motion Auto-Encoder (online motion-AE)
framework that functions on the super-segmented object motion clips.
Comprehensive experiments on a newly-collected surveillance dataset and public
datasets have demonstrated the effectiveness of our proposed method. | ['Yu-jia Zhang', 'Min Tan', 'Dingwen Zhang', 'Xiaodan Liang', 'Eric P. Xing'] | 2018-01-02 | null | null | null | null | ['unsupervised-video-summarization'] | ['computer-vision'] | [ 2.87551105e-01 -3.50098401e-01 -6.08001411e-01 -2.81154811e-01
-4.64031994e-01 -4.03744012e-01 5.02894461e-01 -1.00101260e-02
-2.06628680e-01 3.84845197e-01 9.25333858e-01 1.99343845e-01
5.79090677e-02 -3.56967866e-01 -6.17071629e-01 -5.97070813e-01
-8.32621455e-02 -2.21531048e-01 7.90189028e-01 1.57475069e-01
3.76937509e-01 1.58515051e-01 -1.70251298e+00 4.29647505e-01
8.89978290e-01 8.91735077e-01 3.99846166e-01 8.96725774e-01
-5.86193390e-02 1.31472373e+00 -5.46562731e-01 -3.82788241e-01
9.18246992e-03 -7.18329608e-01 -9.57132638e-01 6.75788224e-01
6.46272838e-01 -1.04301643e+00 -7.98501611e-01 1.06442201e+00
1.47388473e-01 3.03921640e-01 2.71936178e-01 -1.11339259e+00
-4.04026031e-01 5.43041587e-01 -5.57892442e-01 5.61966240e-01
4.87727404e-01 2.14092210e-01 1.02725184e+00 -5.54407001e-01
9.21591282e-01 1.05106330e+00 3.40399474e-01 4.50709254e-01
-4.46679264e-01 -1.83853433e-01 4.37197179e-01 6.01472735e-01
-1.07990968e+00 -5.44954956e-01 8.17285359e-01 -4.00935024e-01
7.24096775e-01 4.04909849e-01 8.26427817e-01 1.11862540e+00
2.48101298e-02 1.37434149e+00 2.73965657e-01 3.84455509e-02
1.87668473e-01 -3.38655859e-01 4.68535610e-02 6.38547301e-01
3.91503483e-01 -5.16731083e-01 -7.55901396e-01 -3.46360840e-02
6.75424933e-01 3.71628314e-01 -3.63661319e-01 -2.80210495e-01
-1.44755268e+00 4.32128906e-01 -4.93256934e-02 3.56283218e-01
-5.75020492e-01 1.99693501e-01 7.90168047e-01 -1.01652667e-02
4.98566568e-01 -4.79815230e-02 -2.28436500e-01 -4.64477688e-01
-1.29430032e+00 3.23724836e-01 6.29704595e-01 1.24318492e+00
5.54965615e-01 6.87008277e-02 -4.29511756e-01 5.37346065e-01
6.45198599e-02 3.00079077e-01 5.86866558e-01 -1.32162762e+00
4.94798392e-01 6.15198255e-01 2.62274832e-01 -1.21834111e+00
-6.87963516e-02 -1.76497623e-01 -9.20637310e-01 -6.36132121e-01
-3.78380716e-02 -2.72289459e-02 -6.64850056e-01 1.34493315e+00
4.94584799e-01 4.70404565e-01 -1.05460122e-01 1.12968898e+00
1.16282511e+00 9.93189931e-01 1.51090786e-01 -6.60361409e-01
1.28846300e+00 -1.43006778e+00 -9.75464940e-01 7.00341463e-02
3.35104346e-01 -5.14495730e-01 6.01619601e-01 2.06947729e-01
-1.30503917e+00 -8.13286185e-01 -8.56933475e-01 -2.64050066e-01
1.42843604e-01 6.07267097e-02 5.73147655e-01 7.60860443e-02
-8.96854579e-01 4.92611647e-01 -1.06474662e+00 -4.94139433e-01
6.75862432e-01 -1.43313289e-01 -3.69484633e-01 -1.77323788e-01
-9.03655708e-01 1.81863636e-01 7.94657588e-01 5.42735718e-02
-1.12645853e+00 -4.26307619e-01 -8.82776916e-01 1.78486779e-01
9.34758782e-01 -8.84283721e-01 1.36460054e+00 -1.04122710e+00
-1.33562112e+00 3.86693716e-01 -3.41453969e-01 -6.61047697e-01
3.95600945e-01 -6.17891550e-01 -3.13150078e-01 7.30551422e-01
1.61552161e-01 7.06876099e-01 1.19997764e+00 -9.11650598e-01
-1.28212595e+00 -1.94412664e-01 1.12449870e-01 2.97537565e-01
-7.84333527e-01 1.84983656e-01 -9.62561190e-01 -1.13666523e+00
-1.21539503e-01 -6.20790899e-01 -2.72712648e-01 -2.62046158e-01
-2.99802512e-01 -2.34510690e-01 1.12855601e+00 -1.14662778e+00
1.69854355e+00 -2.21989942e+00 3.99946839e-01 -4.89030123e-01
3.79507184e-01 4.52826768e-01 4.78810817e-02 4.81485724e-01
3.72437835e-01 -1.50458455e-01 -2.90598094e-01 -1.51853740e-01
-6.87618852e-02 -1.06420957e-01 -3.65360200e-01 1.95107907e-01
8.75748247e-02 9.98525083e-01 -1.21213627e+00 -8.45010102e-01
3.25376689e-02 8.62828642e-02 -8.13184559e-01 3.43726724e-01
-2.98348248e-01 5.30584931e-01 -8.17158103e-01 8.84552717e-01
4.25310493e-01 -2.46097922e-01 9.31930766e-02 -5.65376103e-01
-2.21023113e-01 -8.81022140e-02 -9.31749821e-01 1.94640207e+00
2.26802915e-01 7.46231318e-01 -6.37114421e-02 -1.02001333e+00
2.95586199e-01 2.67386377e-01 9.94687319e-01 -4.07896847e-01
-7.88138807e-02 -8.08781534e-02 -5.19482374e-01 -9.47384059e-01
1.10760880e+00 4.60036099e-01 -1.39889985e-01 3.01924855e-01
2.42065161e-01 3.12175035e-01 6.27925515e-01 4.92194861e-01
1.19787836e+00 4.25014853e-01 5.99823534e-01 8.32402036e-02
4.78460759e-01 4.29247350e-01 7.48304784e-01 6.39751911e-01
-4.59878832e-01 5.33121645e-01 2.62176931e-01 -4.05835211e-01
-1.02792561e+00 -6.39518142e-01 4.25668359e-01 1.18541777e+00
4.33249056e-01 -7.62285709e-01 -1.00892675e+00 -6.65571511e-01
-3.20035726e-01 1.83922738e-01 -2.17741609e-01 -1.59952611e-01
-7.20764458e-01 -3.24225724e-01 3.62697303e-01 5.94529927e-01
7.87477374e-01 -1.10095417e+00 -9.31760609e-01 3.64254594e-01
-7.36210108e-01 -1.49625194e+00 -8.69159520e-01 -6.81294322e-01
-1.07357061e+00 -1.05706084e+00 -9.59466815e-01 -9.18946028e-01
5.18793821e-01 9.41167831e-01 9.71813560e-01 -1.13003924e-01
-2.13354453e-01 5.50986230e-01 -8.37171912e-01 -2.01584175e-01
-1.06673263e-01 1.41054928e-01 -4.42873053e-02 3.16891432e-01
3.18859071e-01 -3.76224369e-01 -9.26151454e-01 1.39541671e-01
-1.34745193e+00 3.47102225e-01 7.01363206e-01 3.73124391e-01
5.09140670e-01 2.26790786e-01 6.08445048e-01 -6.57556415e-01
3.38416785e-01 -6.16052866e-01 -3.35608661e-01 2.53287405e-01
1.30349815e-01 -2.74994344e-01 5.72668254e-01 -2.64438301e-01
-1.27051270e+00 8.47424939e-02 3.22429389e-02 -5.31162918e-01
-3.09410870e-01 5.43202698e-01 -2.74097204e-01 3.93025160e-01
4.48990893e-03 6.70605779e-01 -4.25260991e-01 -5.82655847e-01
5.27459979e-01 7.47199714e-01 9.88712311e-01 -1.75385073e-01
6.92623138e-01 8.64506304e-01 -3.80616695e-01 -1.05436480e+00
-9.17410374e-01 -9.46900487e-01 -6.55395210e-01 -3.18034530e-01
1.13103831e+00 -1.05745053e+00 -5.17335773e-01 6.12681627e-01
-1.17215931e+00 -1.52379554e-02 -3.52013856e-01 5.10584593e-01
-6.28065228e-01 1.00833344e+00 -5.72700441e-01 -5.17822444e-01
-4.80504215e-01 -8.41162384e-01 1.19797194e+00 3.85083973e-01
-1.36238575e-01 -6.25790179e-01 5.01894206e-02 5.76633930e-01
1.71234459e-01 2.47969896e-01 4.05072361e-01 -3.93277019e-01
-9.76322055e-01 -2.30770856e-02 -1.32819444e-01 3.14429790e-01
2.27141112e-01 1.38497397e-01 -4.25097793e-01 -2.60480106e-01
1.20819792e-01 -7.16606528e-02 1.26025176e+00 5.82449794e-01
1.37436080e+00 -6.77441120e-01 -3.03788096e-01 5.71062028e-01
1.08582449e+00 2.30168700e-01 5.08004844e-01 2.09450975e-01
9.28596199e-01 5.81211209e-01 1.10584891e+00 8.32780719e-01
5.53383946e-01 5.29855609e-01 4.48183864e-01 4.19329047e-01
-1.81257114e-01 -4.15613979e-01 5.35346866e-01 1.15200794e+00
-4.01760429e-01 -2.23678723e-01 -3.92440557e-01 8.32611978e-01
-2.26424980e+00 -1.62584341e+00 8.31000134e-02 1.70768797e+00
5.88961482e-01 -7.84111917e-02 2.93666214e-01 -1.55469865e-01
7.83405840e-01 7.28268862e-01 -7.25014210e-01 3.10395688e-01
-5.70724672e-03 -4.15562689e-01 2.08924636e-01 -1.18890181e-02
-1.48725748e+00 9.94016767e-01 5.29936886e+00 8.91679287e-01
-8.19708169e-01 1.34977818e-01 4.10255492e-01 -1.53175861e-01
-6.45869300e-02 -1.87622588e-02 -7.13679671e-01 6.93233788e-01
5.84837735e-01 -2.58873522e-01 1.77728146e-01 9.64641690e-01
5.56053042e-01 -2.13188171e-01 -9.08542633e-01 1.21264207e+00
3.82619232e-01 -1.66084874e+00 4.49582040e-01 -2.73796618e-01
1.02893066e+00 -2.44451612e-01 -2.58068889e-01 2.57069409e-01
-1.81964651e-01 -3.82787377e-01 9.76817250e-01 5.93894303e-01
4.56833601e-01 -6.27101004e-01 6.19254291e-01 4.57849234e-01
-1.50248027e+00 -1.60053864e-01 -2.58794785e-01 1.31274819e-01
4.72692043e-01 4.09756243e-01 -3.71116489e-01 9.50457156e-01
9.70353782e-01 1.29793882e+00 -3.79789531e-01 1.17338502e+00
3.75246070e-02 7.91539133e-01 1.21911779e-01 2.02859670e-01
3.45398098e-01 -7.75977671e-02 8.71156394e-01 1.39641094e+00
3.04206312e-01 5.34976065e-01 3.14524591e-01 2.94071823e-01
-7.24543482e-02 2.55760271e-03 -3.24072152e-01 -4.79926825e-01
2.02308252e-01 1.30923450e+00 -8.30109358e-01 -7.59408295e-01
-5.43400884e-01 1.26768327e+00 -1.74989954e-01 3.79420608e-01
-1.03783906e+00 -3.27530742e-01 4.72274244e-01 -1.09436717e-02
7.59950995e-01 -4.30987269e-01 4.37633842e-01 -1.65717113e+00
2.65016019e-01 -9.18864846e-01 5.40491223e-01 -6.85445607e-01
-8.82279515e-01 2.68133730e-01 1.41164824e-01 -1.51693666e+00
-3.25227886e-01 -6.41240329e-02 -5.73966146e-01 -6.86694980e-02
-1.27139103e+00 -1.11178565e+00 -7.42669761e-01 6.14804685e-01
1.33838308e+00 -3.55195142e-02 1.45800859e-01 3.62008721e-01
-6.55922413e-01 1.14512503e-01 6.93650991e-02 1.56166077e-01
5.49136460e-01 -6.84298456e-01 3.59358937e-01 1.25481689e+00
1.76620096e-01 4.49264586e-01 6.08749509e-01 -8.82791340e-01
-1.84644639e+00 -1.48107278e+00 5.13985932e-01 -1.62459835e-01
5.76697826e-01 4.16981280e-02 -7.89549768e-01 7.22907901e-01
3.04764122e-01 -1.60650998e-01 5.03997684e-01 -6.33248985e-01
1.28413931e-01 -5.70438243e-02 -7.14588523e-01 7.02632546e-01
1.53935170e+00 -3.31581026e-01 -8.36106598e-01 2.70684391e-01
9.88788486e-01 -2.58229226e-01 -6.30750537e-01 4.34257001e-01
5.87460101e-01 -9.72940564e-01 8.98357272e-01 -7.21304238e-01
7.21740544e-01 -4.51485127e-01 -4.56787683e-02 -1.01639509e+00
-3.19901824e-01 -8.61960113e-01 -8.81805420e-01 1.31874955e+00
-5.50873637e-01 -4.18111980e-02 8.22426558e-01 -6.03281558e-02
-3.83962899e-01 -5.90417981e-01 -4.25448269e-01 -4.97626752e-01
-9.96994674e-01 -3.52211714e-01 4.18371379e-01 7.56337166e-01
-6.53067455e-02 9.65876952e-02 -8.13506961e-01 2.62502342e-01
7.49862850e-01 4.29279864e-01 9.87848997e-01 -8.23197842e-01
-1.95385501e-01 -4.33101982e-01 -6.75284207e-01 -1.55307245e+00
4.25214879e-02 -4.10904199e-01 2.86768880e-02 -1.78803694e+00
8.14899087e-01 5.05550861e-01 -1.68332696e-01 5.68288639e-02
-4.15551901e-01 8.03384185e-02 1.68841287e-01 5.54874718e-01
-1.52956355e+00 5.86732328e-01 1.12072778e+00 -3.14842999e-01
-1.36669368e-01 -1.29636079e-01 -5.41779459e-01 9.22732592e-01
5.49504519e-01 -2.45183587e-01 -5.18546939e-01 -5.05725861e-01
-1.21450014e-01 2.45344386e-01 4.21647966e-01 -1.01589274e+00
4.97469306e-01 -4.48064893e-01 1.81000844e-01 -1.13858414e+00
6.10693134e-02 -6.39921308e-01 1.03203356e-01 5.26447892e-01
-1.75890908e-01 -8.97217840e-02 -2.76242830e-02 9.30204391e-01
-6.57323360e-01 -6.60479218e-02 3.27758372e-01 -1.37942001e-01
-1.48853791e+00 6.55334949e-01 -3.14939827e-01 7.84827098e-02
1.14023757e+00 -4.05583799e-01 -1.56121418e-01 -5.36171496e-01
-4.12146240e-01 2.80966669e-01 5.00538766e-01 6.48358464e-01
7.80445278e-01 -1.26372147e+00 -7.93545067e-01 -1.16214387e-01
1.45398065e-01 1.25201106e-01 7.27719128e-01 8.73397112e-01
-6.97086155e-01 6.44799531e-01 -2.59433657e-01 -5.94923139e-01
-1.36255169e+00 6.43758953e-01 -2.77535021e-01 -9.61861461e-02
-6.93820834e-01 4.77990478e-01 4.04260576e-01 5.18160224e-01
3.01984310e-01 -5.00888944e-01 -4.49747443e-01 1.36281863e-01
9.70049202e-01 7.34464169e-01 -3.91045809e-01 -7.97570109e-01
-1.72534868e-01 4.64734018e-01 -2.09736273e-01 2.48267829e-01
1.42165697e+00 -4.88380790e-01 -1.70753270e-01 2.83390939e-01
1.09436035e+00 6.04467932e-03 -1.60173321e+00 -9.65878367e-02
1.17383763e-01 -5.76299489e-01 -2.90409386e-01 4.04364765e-02
-9.49543715e-01 5.35903156e-01 1.06252968e-01 2.79122770e-01
1.35524392e+00 1.08542569e-01 1.35709095e+00 4.31435943e-01
4.80403364e-01 -1.22313905e+00 3.49987924e-01 2.73171365e-01
5.98966420e-01 -1.15922225e+00 1.57435939e-01 -4.74551141e-01
-6.29102230e-01 1.02679312e+00 4.55810636e-01 -2.32943259e-02
1.42935812e-01 -2.47570634e-01 -4.69206214e-01 -1.07168138e-01
-7.39290595e-01 -2.05338985e-01 4.24390346e-01 3.71736765e-01
1.74256340e-01 -2.03426585e-01 -3.03265095e-01 6.87654793e-01
2.86137968e-01 3.21246743e-01 4.84320492e-01 1.17421985e+00
-8.89384031e-01 -4.32694942e-01 -2.31403694e-01 3.61947745e-01
-6.87601447e-01 1.23965353e-01 -2.55442113e-01 4.08914566e-01
-5.12449071e-02 8.98989916e-01 -1.40327006e-03 -2.69409984e-01
1.74326584e-01 -2.67292023e-01 2.99880177e-01 -4.79005545e-01
-1.17601439e-01 3.13344337e-02 -9.33405086e-02 -8.61924827e-01
-9.78570282e-01 -6.91720307e-01 -1.11225224e+00 -1.05915345e-01
2.89456099e-02 1.71212047e-01 3.35750133e-01 9.16767955e-01
4.19319510e-01 5.93764901e-01 6.57734811e-01 -1.11374319e+00
-3.81132483e-01 -8.05166423e-01 -2.57110506e-01 5.28492153e-01
4.13131952e-01 -3.15480471e-01 -8.17989707e-02 7.84494758e-01] | [10.351653099060059, 0.4733276963233948] |
2307fa24-4846-457d-baaa-e59de5208728 | diffcollage-parallel-generation-of-large | 2303.17076 | null | https://arxiv.org/abs/2303.17076v1 | https://arxiv.org/pdf/2303.17076v1.pdf | DiffCollage: Parallel Generation of Large Content with Diffusion Models | We present DiffCollage, a compositional diffusion model that can generate large content by leveraging diffusion models trained on generating pieces of the large content. Our approach is based on a factor graph representation where each factor node represents a portion of the content and a variable node represents their overlap. This representation allows us to aggregate intermediate outputs from diffusion models defined on individual nodes to generate content of arbitrary size and shape in parallel without resorting to an autoregressive generation procedure. We apply DiffCollage to various tasks, including infinite image generation, panorama image generation, and long-duration text-guided motion generation. Extensive experimental results with a comparison to strong autoregressive baselines verify the effectiveness of our approach. | ['Ming-Yu Liu', 'Yongxin Chen', 'Xun Huang', 'Jiaming Song', 'Qinsheng Zhang'] | 2023-03-30 | null | http://openaccess.thecvf.com//content/CVPR2023/html/Zhang_DiffCollage_Parallel_Generation_of_Large_Content_With_Diffusion_Models_CVPR_2023_paper.html | http://openaccess.thecvf.com//content/CVPR2023/papers/Zhang_DiffCollage_Parallel_Generation_of_Large_Content_With_Diffusion_Models_CVPR_2023_paper.pdf | cvpr-2023-1 | ['infinite-image-generation'] | ['computer-vision'] | [ 3.42744470e-01 5.62317312e-01 8.24411213e-03 2.41953194e-01
-9.08408284e-01 -6.38324797e-01 1.30776358e+00 -2.80850857e-01
7.69001767e-02 7.20704496e-01 8.36742997e-01 -4.16269869e-01
4.21605855e-01 -1.14284313e+00 -8.53380203e-01 -6.76271081e-01
6.12585172e-02 5.23827672e-01 2.24468663e-01 -3.74925822e-01
6.86479732e-02 2.21563607e-01 -9.42768693e-01 4.64496404e-01
7.08839715e-01 5.51912844e-01 3.84952784e-01 1.26801527e+00
-1.48004159e-01 1.28187275e+00 -8.72092843e-01 -5.31579673e-01
2.47919425e-01 -1.00089037e+00 -6.09386146e-01 3.98535579e-01
4.81407821e-01 -6.16019666e-01 -6.19432390e-01 5.82009733e-01
4.41268653e-01 2.06097543e-01 1.12014508e+00 -1.30368459e+00
-1.15415335e+00 1.14307487e+00 -5.65598249e-01 4.89204116e-02
4.62278664e-01 3.41803104e-01 8.85038555e-01 -8.78675997e-01
1.27877283e+00 1.41036403e+00 4.99217331e-01 6.45843923e-01
-1.46697521e+00 -2.38844484e-01 1.51327655e-01 -4.65028584e-01
-1.05227721e+00 -4.07206208e-01 6.50703669e-01 -4.29276913e-01
8.80036831e-01 -3.66175696e-02 7.87526846e-01 1.42728698e+00
4.43032146e-01 9.74635363e-01 8.27372253e-01 -3.95709455e-01
1.82471797e-01 -3.63076538e-01 -3.79787505e-01 9.11382675e-01
-7.39892274e-02 1.23668417e-01 -3.52152824e-01 -3.76069188e-01
1.25081348e+00 -3.47693920e-01 -1.66539907e-01 -1.72985628e-01
-1.50901937e+00 1.02538884e+00 3.95351797e-01 7.90165067e-02
-7.39332497e-01 8.14641833e-01 5.02227508e-02 1.18236437e-01
6.61010623e-01 2.79170185e-01 1.59150034e-01 2.08814647e-02
-1.22390151e+00 7.48565018e-01 8.05060983e-01 1.21833372e+00
5.30548394e-01 6.16544724e-01 -7.00785220e-01 5.48283994e-01
1.27673149e-01 7.40759969e-01 5.84005356e-01 -1.18305135e+00
4.55612510e-01 1.48200691e-01 1.20738849e-01 -8.53719115e-01
5.95017560e-02 -2.80751381e-02 -1.03385639e+00 2.17247188e-01
2.46421471e-01 -5.71721435e-01 -1.34667242e+00 1.65701699e+00
1.99723125e-01 3.11997414e-01 1.06702335e-01 4.52572018e-01
6.08566344e-01 1.25719178e+00 3.79579701e-02 -5.14536388e-02
1.03819716e+00 -1.44972575e+00 -5.93128502e-01 1.77172069e-02
2.50551194e-01 -8.68995428e-01 8.22148323e-01 2.43322045e-01
-1.59309709e+00 -5.22288680e-01 -8.07848215e-01 -1.40876263e-01
-1.31098703e-01 -1.49784967e-01 4.43363309e-01 4.21674073e-01
-1.55312061e+00 5.18813729e-01 -5.90629339e-01 1.10407382e-01
2.76011199e-01 -8.11195970e-02 -1.46825060e-01 -2.63338219e-02
-1.02269363e+00 4.26954687e-01 2.60965556e-01 -1.78352028e-01
-1.22035599e+00 -6.37530148e-01 -9.72936094e-01 -6.69846386e-02
3.18353362e-02 -1.52792096e+00 1.42558181e+00 -9.08753216e-01
-1.71115065e+00 2.68906862e-01 -2.32135102e-01 -7.56275117e-01
8.04381847e-01 -4.70805913e-02 -1.41081661e-01 3.09664696e-01
1.68203056e-01 1.13797534e+00 1.37769890e+00 -1.26900423e+00
-3.47861916e-01 2.42316291e-01 -6.69691116e-02 2.41811112e-01
8.15820396e-02 -1.65621310e-01 -6.38116658e-01 -1.34922981e+00
-2.95801401e-01 -1.21403956e+00 -8.01304758e-01 -2.26889327e-01
-6.98485017e-01 9.21782628e-02 7.66183376e-01 -7.03408241e-01
1.24175322e+00 -1.66364324e+00 5.62818885e-01 2.67878592e-01
3.05628389e-01 -2.42107660e-01 -6.60968602e-01 8.28613400e-01
2.03889802e-01 3.88974726e-01 -4.21003342e-01 -4.13945347e-01
2.08310806e-03 -3.85249853e-02 -8.35808516e-01 -1.18721381e-01
2.70218968e-01 1.42406857e+00 -1.00067770e+00 -5.35897672e-01
-2.88819466e-02 5.91140330e-01 -7.28266120e-01 3.21941853e-01
-8.22334290e-01 2.52517313e-01 -3.40909123e-01 2.51256347e-01
3.57723743e-01 -5.93974471e-01 1.88507549e-02 7.70095661e-02
2.23015308e-01 -3.58163081e-02 -9.44913626e-01 1.77023959e+00
-5.29261410e-01 6.39877319e-01 -1.78278804e-01 -1.58091769e-01
6.23165190e-01 3.43123555e-01 2.88496315e-01 -2.62541771e-01
-7.12103620e-02 -2.63148788e-02 -4.08596098e-01 -8.19367692e-02
1.23490238e+00 -5.86761422e-02 -2.27475554e-01 1.18181682e+00
2.17058398e-02 -7.91995704e-01 4.58739281e-01 9.44407046e-01
1.08360839e+00 2.87026852e-01 -1.04711093e-04 5.86165302e-02
1.35960639e-01 -1.07206311e-02 -2.21494734e-01 1.05882800e+00
4.14671779e-01 9.62724268e-01 5.80122828e-01 -1.91852510e-01
-1.56881285e+00 -1.32524264e+00 6.14864945e-01 8.87183011e-01
-1.82885244e-01 -6.27386987e-01 -9.36750829e-01 -7.12636590e-01
-2.35977232e-01 7.78358042e-01 -7.72757649e-01 1.35160625e-01
-7.29036391e-01 -6.92356825e-01 5.97438157e-01 5.66565633e-01
2.10796818e-01 -1.29504335e+00 -3.19798738e-01 3.41942728e-01
-4.00645286e-01 -1.06598604e+00 -9.01639700e-01 -5.10390937e-01
-7.51748681e-01 -5.22122562e-01 -1.50640655e+00 -7.94675410e-01
7.64690101e-01 3.89140397e-01 1.44366002e+00 3.61771546e-02
-2.66830008e-02 5.17121851e-01 -2.11330339e-01 -1.08526133e-01
-1.23305929e+00 2.11314753e-01 -4.87591475e-01 -6.50363192e-02
-6.66976750e-01 -5.10832906e-01 -7.39184260e-01 -3.97074260e-02
-1.40314424e+00 4.47093815e-01 5.51773965e-01 8.08526158e-01
4.37083155e-01 -1.95031643e-01 5.10983407e-01 -1.09894490e+00
1.29174054e+00 -7.09856808e-01 -3.91270131e-01 6.67386428e-02
-3.66917938e-01 2.56269753e-01 4.97576803e-01 -7.81373918e-01
-1.25430536e+00 -1.79438308e-01 2.64332891e-02 -4.28067118e-01
2.67436564e-01 4.41715032e-01 4.10110742e-01 4.84993905e-01
8.57494056e-01 5.61707854e-01 7.98477009e-02 5.65844513e-02
1.26264858e+00 2.44806558e-01 6.13878429e-01 -5.49048841e-01
9.62518156e-01 6.03881180e-01 -3.98574397e-02 -7.41588116e-01
-2.53532320e-01 2.74685442e-01 -3.69573683e-01 -2.07688957e-01
9.87029374e-01 -1.15513051e+00 2.89910510e-02 7.49951661e-01
-1.30960298e+00 -9.58609760e-01 -5.83361447e-01 -7.66617805e-02
-8.31767440e-01 4.23099160e-01 -1.15664947e+00 -5.85468531e-01
-6.24903440e-01 -9.72621620e-01 1.26024878e+00 -4.41036609e-05
-4.29135501e-01 -1.07363391e+00 2.81415164e-01 1.37007400e-01
6.68743372e-01 2.54272610e-01 7.76027203e-01 -1.14765354e-01
-1.01247358e+00 -2.34050691e-01 -7.14600459e-02 3.18583511e-02
-5.69221936e-02 2.73506403e-01 -4.64066833e-01 -5.07152379e-02
-5.92270672e-01 -4.84789073e-01 1.08585346e+00 5.95808744e-01
7.49959409e-01 -5.68637550e-01 -2.25841522e-01 2.86096722e-01
1.08944857e+00 -1.38177112e-01 8.32645059e-01 -4.95257005e-02
8.63332212e-01 3.30522954e-01 -1.97828151e-02 5.45023739e-01
5.05478978e-01 4.24253315e-01 9.51249152e-02 -2.04954892e-01
-7.29940057e-01 -8.44269454e-01 5.68458915e-01 8.46301198e-01
-9.57945883e-02 -8.63517463e-01 -5.91690898e-01 7.28512466e-01
-1.67147231e+00 -1.28193939e+00 -2.81837344e-01 1.63440430e+00
6.97724044e-01 -7.26520941e-02 2.54494399e-01 -3.05313557e-01
6.63303852e-01 6.06020868e-01 -3.44158173e-01 -3.23327750e-01
-2.97975123e-01 2.47007877e-01 2.87522763e-01 8.05317760e-01
-8.31231713e-01 1.32473838e+00 7.89114094e+00 9.34111714e-01
-8.20422769e-01 -7.65096247e-02 8.91311407e-01 -1.08172119e-01
-9.76490498e-01 -9.07567888e-02 -4.96775627e-01 4.05672729e-01
9.16850150e-01 -3.10077369e-01 4.32254672e-01 6.17238700e-01
2.72885859e-01 -8.09438750e-02 -5.71651638e-01 5.09128273e-01
2.52675533e-01 -1.89909303e+00 7.46285856e-01 1.34069368e-01
1.49266994e+00 -4.17880975e-02 2.39871919e-01 1.22457400e-01
1.39773428e+00 -1.01109350e+00 9.97860014e-01 6.19654775e-01
6.97248459e-01 -4.80415195e-01 1.40044019e-01 2.92387635e-01
-1.13571966e+00 1.59008056e-01 -1.76149026e-01 2.94099957e-01
9.70697939e-01 5.10501564e-01 -9.05693829e-01 3.67613405e-01
-2.03810609e-03 4.44305092e-01 -3.40211868e-01 5.19236028e-01
-4.77336943e-01 5.72803617e-01 -1.52114540e-01 9.82488543e-02
3.72411907e-01 -3.71674716e-01 5.99480450e-01 1.33413756e+00
7.72866249e-01 -7.55765736e-02 2.91385911e-02 1.06809938e+00
-3.44048291e-01 -8.61181412e-03 -8.36775601e-01 -1.48534656e-01
2.74467498e-01 1.14358687e+00 -7.95112371e-01 -9.40457523e-01
-2.17435494e-01 1.48949420e+00 2.49398306e-01 7.05192149e-01
-9.51058090e-01 -1.58942342e-01 2.30296880e-01 2.29135975e-01
6.53184354e-01 -4.39839214e-01 -7.80277839e-03 -1.16992509e+00
-2.99476624e-01 -1.02663362e+00 3.00571442e-01 -1.33681989e+00
-1.14809728e+00 1.01599181e+00 3.48148681e-02 -9.27068770e-01
-9.86571431e-01 8.46541747e-02 -7.60157704e-01 1.02609944e+00
-1.09131348e+00 -1.54510105e+00 -2.23296478e-01 6.69411778e-01
7.84760714e-01 -1.65410072e-01 5.44225156e-01 -3.38237971e-01
-1.88021898e-01 1.96448401e-01 -1.46091223e-01 1.62095428e-02
3.72803152e-01 -1.20293117e+00 1.21423829e+00 9.46825147e-01
3.72060120e-01 3.18413198e-01 6.93713963e-01 -1.03514016e+00
-1.13616014e+00 -1.17451191e+00 8.43579769e-01 -4.92581367e-01
8.20316136e-01 -2.71488577e-01 -5.50049901e-01 8.10626566e-01
9.02122080e-01 -3.19573462e-01 4.04099137e-01 -5.44757247e-01
-4.35960054e-01 5.41681945e-01 -6.56399727e-01 1.09872901e+00
1.13687325e+00 -3.44694912e-01 -2.46486127e-01 2.54606068e-01
9.95370686e-01 -4.64051366e-01 -7.84450054e-01 -4.90292348e-02
3.98907125e-01 -8.31755042e-01 9.97880638e-01 -5.04528880e-01
1.28995025e+00 -4.39008437e-02 1.36698410e-01 -1.57870948e+00
-4.01232392e-01 -1.15245497e+00 -4.67463225e-01 9.19554174e-01
6.65748417e-01 -1.94842055e-01 8.76753271e-01 5.77240407e-01
6.30835257e-03 -2.94436395e-01 -3.72147918e-01 -4.94022757e-01
2.81765133e-01 -2.87849039e-01 7.10416436e-01 6.04560018e-01
-3.08244258e-01 5.23863912e-01 -7.31058538e-01 -1.85060576e-01
4.64987755e-01 2.06708312e-01 1.16588342e+00 -6.06016994e-01
-6.08646512e-01 -6.19949996e-01 2.31667131e-01 -1.54397190e+00
1.43822655e-01 -9.24825132e-01 3.10904300e-03 -1.95442677e+00
6.63571134e-02 -2.93447971e-01 3.86523068e-01 1.85687155e-01
-3.82798702e-01 5.48648775e-01 5.65162420e-01 3.78873765e-01
-1.89309567e-01 7.56004989e-01 1.70159340e+00 -2.61538386e-01
-2.62383908e-01 -2.50152528e-01 -6.88078284e-01 4.84148383e-01
5.67344010e-01 -2.67397135e-01 -7.70881414e-01 -5.90100825e-01
3.93226892e-01 5.26576936e-01 2.23505229e-01 -8.44564438e-01
7.98195004e-02 -2.34241784e-01 3.89018446e-01 -4.04483140e-01
3.45677912e-01 -1.85915053e-01 5.18105030e-01 3.42388034e-01
-5.69809020e-01 4.47811246e-01 -2.29019090e-03 8.05840075e-01
-3.87458391e-02 -6.40789233e-03 3.76073122e-01 -4.04186726e-01
-3.94779533e-01 5.02259672e-01 -8.67544770e-01 3.23289819e-02
9.23256338e-01 1.22548893e-01 -3.80554646e-01 -1.02741575e+00
-7.27010369e-01 -1.65684953e-01 5.64525425e-01 5.38529813e-01
7.79467165e-01 -1.58043039e+00 -1.07827842e+00 7.47128427e-02
-2.90646404e-01 1.88809466e-02 2.13008925e-01 2.76308268e-01
-6.84011996e-01 1.29192635e-01 9.51951519e-02 -2.73058176e-01
-8.75781775e-01 6.48367941e-01 -6.47824407e-02 -7.76042104e-01
-6.95345283e-01 7.18441069e-01 3.55179220e-01 3.11610270e-02
-4.15678322e-01 -2.34894395e-01 1.04016043e-01 -2.16775723e-02
6.91661537e-01 3.18069339e-01 -5.99050522e-01 -6.53303683e-01
4.82639790e-01 2.95035452e-01 4.85682599e-02 -8.70434701e-01
1.00623703e+00 -2.41843402e-01 -2.83205416e-03 1.28162116e-01
8.42564464e-01 3.12917858e-01 -1.51569533e+00 -1.87329445e-02
-4.81420875e-01 -1.75357580e-01 -7.17187524e-02 -5.63554764e-01
-1.19902921e+00 5.43850005e-01 -1.64206609e-01 2.86722958e-01
7.73833156e-01 7.25429431e-02 1.14885247e+00 -9.14702471e-03
1.90208092e-01 -8.09982002e-01 5.95226288e-01 3.79459143e-01
1.23642170e+00 -7.42444336e-01 -1.03790589e-01 -3.07831764e-01
-1.01633537e+00 9.27456379e-01 1.99853808e-01 -3.68274540e-01
4.61208731e-01 4.64529872e-01 1.90441266e-01 4.06356938e-02
-1.10872662e+00 -4.61701676e-02 2.91951448e-01 5.96078455e-01
2.87371099e-01 -8.05513561e-03 -1.69443980e-01 9.60995704e-02
-3.66715491e-01 4.09918539e-02 8.71442080e-01 6.91172361e-01
-1.97538227e-01 -1.15656471e+00 -3.57698798e-01 2.76785851e-01
-3.54939640e-01 -4.61129308e-01 -4.75352973e-01 6.51019096e-01
-4.15971845e-01 8.13460290e-01 2.92789608e-01 -9.13640559e-02
-1.49377286e-01 2.28573456e-02 4.38881248e-01 -3.62791240e-01
-5.29819131e-01 5.06717384e-01 1.85345307e-01 -4.26992297e-01
-2.77347416e-01 -4.78159726e-01 -1.07461762e+00 -3.26052010e-01
4.66675982e-02 -1.42674074e-02 4.62224782e-01 3.17987889e-01
5.44883430e-01 5.29902160e-01 5.32057583e-01 -1.10384059e+00
-4.09324855e-01 -1.10669029e+00 -2.78702378e-01 6.36042356e-01
3.11108232e-01 8.05931762e-02 -1.21161096e-01 6.55076444e-01] | [11.133367538452148, -0.2509036660194397] |
e76b88ec-5366-482a-acfc-1d4de5b9f566 | mining-both-commonality-and-specificity-from | 2303.02677 | null | https://arxiv.org/abs/2303.02677v1 | https://arxiv.org/pdf/2303.02677v1.pdf | Mining both Commonality and Specificity from Multiple Documents for Multi-Document Summarization | The multi-document summarization task requires the designed summarizer to generate a short text that covers the important information of original documents and satisfies content diversity. This paper proposes a multi-document summarization approach based on hierarchical clustering of documents. It utilizes the constructed class tree of documents to extract both the sentences reflecting the commonality of all documents and the sentences reflecting the specificity of some subclasses of these documents for generating a summary, so as to satisfy the coverage and diversity requirements of multi-document summarization. Comparative experiments with different variant approaches on DUC'2002-2004 datasets prove the effectiveness of mining both the commonality and specificity of documents for multi-document summarization. Experiments on DUC'2004 and Multi-News datasets show that our approach achieves competitive performance compared to the state-of-the-art unsupervised and supervised approaches. | ['Bing Ma'] | 2023-03-05 | null | null | null | null | ['multi-document-summarization', 'document-summarization'] | ['natural-language-processing', 'natural-language-processing'] | [ 2.98059613e-01 1.60992965e-01 -3.73856783e-01 -1.90161660e-01
-1.13398123e+00 -7.16224194e-01 5.82873821e-01 7.82343745e-01
-1.07067507e-02 1.07289124e+00 1.08506310e+00 1.87984228e-01
-3.68917555e-01 -5.26137471e-01 -2.01113924e-01 -4.71785128e-01
2.67038997e-02 5.22547662e-01 4.05069083e-01 -2.19611466e-01
9.12378192e-01 8.53374824e-02 -1.72415709e+00 9.27410305e-01
1.41206503e+00 2.57887512e-01 2.85334140e-01 1.00086951e+00
-5.54709852e-01 6.45990551e-01 -1.26160502e+00 3.67712788e-02
-1.87593251e-01 -1.00649202e+00 -9.03875470e-01 4.33523893e-01
5.50175667e-01 -8.65704119e-02 9.84862447e-02 9.36508894e-01
5.47248006e-01 2.21438020e-01 9.76670146e-01 -7.09244609e-01
-4.36474860e-01 9.10220563e-01 -8.23221862e-01 4.15848225e-01
6.64667606e-01 -7.12525249e-01 1.37248206e+00 -3.05836231e-01
6.99089170e-01 9.57455218e-01 3.12576741e-01 3.60965639e-01
-7.07012117e-01 -9.26957950e-02 1.61343515e-01 -1.88182876e-01
-8.52328002e-01 -4.21591699e-01 6.08461261e-01 -1.06873170e-01
1.23943281e+00 8.06577981e-01 5.10851622e-01 6.10158563e-01
5.40386498e-01 8.55471134e-01 3.94511640e-01 -6.07873023e-01
3.25147092e-01 1.83665797e-01 8.46656501e-01 4.51443225e-01
1.00075853e+00 -1.01301420e+00 -4.80328619e-01 -5.41791916e-01
-1.66115358e-01 -2.24830762e-01 -2.52326638e-01 2.21172988e-01
-9.78450596e-01 7.99304783e-01 -5.95493317e-01 8.16361904e-01
-5.40082753e-01 -3.51056755e-01 9.11033094e-01 1.65603593e-01
6.01449609e-01 6.17134452e-01 -1.69626549e-01 4.05851752e-02
-1.52713454e+00 3.99370283e-01 1.09798634e+00 1.29779232e+00
4.18960392e-01 1.24664987e-02 -5.38993359e-01 8.53424072e-01
-1.11590452e-01 2.01231316e-01 9.58520055e-01 -8.66439164e-01
8.82089078e-01 1.01825261e+00 7.04361945e-02 -1.05102921e+00
-3.88883770e-01 -5.78256130e-01 -1.06430304e+00 -7.65889108e-01
-5.58451653e-01 -3.44672710e-01 -4.71349180e-01 1.13371634e+00
1.32398903e-01 -6.00753307e-01 5.81062376e-01 1.23246975e-01
1.34531808e+00 1.16871321e+00 -4.79339242e-01 -1.08915305e+00
1.19633973e+00 -9.88708198e-01 -8.10936689e-01 4.21393104e-02
5.33087313e-01 -7.19967246e-01 5.11214495e-01 2.76831716e-01
-1.27721381e+00 -5.51627278e-01 -1.24039817e+00 1.46725863e-01
4.20693308e-02 4.99057502e-01 2.66386509e-01 6.06584132e-01
-7.80206382e-01 4.33848202e-01 -2.76337087e-01 -6.32169724e-01
8.34839642e-02 8.22906643e-02 -2.06901461e-01 1.87710285e-01
-7.15627611e-01 5.25585651e-01 9.74075437e-01 -5.85586548e-01
-3.39731127e-01 -4.90198761e-01 -5.62404752e-01 3.95849645e-01
2.69658625e-01 -1.03267157e+00 1.13459146e+00 -5.82597673e-01
-1.02161694e+00 4.33640629e-01 -4.02857065e-01 -4.83196527e-01
3.23946446e-01 -2.15787292e-01 -3.13149214e-01 7.37639904e-01
5.22347629e-01 1.42286286e-01 4.86114353e-01 -1.32072246e+00
-1.26851654e+00 -2.48223305e-01 -3.26927125e-01 4.81226414e-01
-5.48315525e-01 7.63285086e-02 -3.82639706e-01 -4.90657270e-01
6.08333275e-02 -4.62738454e-01 -1.12411477e-01 -1.43812275e+00
-1.03431916e+00 -5.06491065e-01 1.00902236e+00 -6.65910602e-01
1.88071883e+00 -1.77879548e+00 1.19516648e-01 1.37987003e-01
1.24738261e-01 1.31366983e-01 -1.78376824e-01 1.20838094e+00
1.76960513e-01 2.39737198e-01 -2.55058587e-01 -2.97846168e-01
-2.16332093e-01 4.38958704e-02 -5.75316846e-01 1.52747065e-01
-2.44251937e-01 4.12744075e-01 -7.78827488e-01 -9.34813440e-01
-1.62276134e-01 -3.94131750e-01 -4.12564337e-01 1.15116470e-01
-3.33710581e-01 6.00399747e-02 -7.98098862e-01 3.32172722e-01
3.18890661e-01 9.38300341e-02 3.21672678e-01 -1.01638421e-01
-3.68802577e-01 2.12112203e-01 -1.17650962e+00 1.31649995e+00
8.99411514e-02 4.42575634e-01 -3.42576116e-01 -1.20790327e+00
9.73719120e-01 3.49131674e-01 7.15129554e-01 -1.55033976e-01
8.64648446e-02 2.40404397e-01 -3.31555039e-01 -5.99950314e-01
1.35786545e+00 7.93689042e-02 -5.30920506e-01 7.61287808e-01
9.90125313e-02 -4.02136743e-01 1.14260662e+00 1.02306998e+00
1.05413151e+00 -5.91698587e-01 8.30648780e-01 -5.29597402e-01
6.65349424e-01 2.75844663e-01 4.73499805e-01 1.11037052e+00
3.72191072e-01 7.01569021e-01 7.05440998e-01 5.56055121e-02
-1.16404319e+00 -5.04877269e-01 2.77825505e-01 8.29322517e-01
5.18036447e-03 -9.59960401e-01 -8.17572296e-01 -7.38879681e-01
-1.37070224e-01 1.09612346e+00 -4.39198494e-01 -3.09025198e-02
-4.02776301e-01 -8.43254924e-01 4.26787436e-01 1.25738680e-01
6.12369895e-01 -7.49606073e-01 -7.02223659e-01 3.38451028e-01
-5.48921764e-01 -8.85204673e-01 -5.39648950e-01 -4.24591498e-03
-7.53913879e-01 -1.15172946e+00 -6.51484013e-01 -8.93509567e-01
6.37538850e-01 6.33782685e-01 7.23718107e-01 -4.21834774e-02
5.61392382e-02 4.85776275e-01 -8.60904396e-01 -5.30272245e-01
-9.47712481e-01 6.46035314e-01 -3.37003395e-02 -2.07415715e-01
-9.97084901e-02 -4.25570905e-01 -3.36952098e-02 -2.10454345e-01
-1.04723275e+00 6.87412992e-02 4.50341731e-01 4.85389888e-01
3.97889644e-01 7.51452684e-01 1.18681967e+00 -1.08871627e+00
1.47425842e+00 -5.28463662e-01 5.38124703e-02 7.64185071e-01
-3.86489868e-01 1.07727394e-01 8.62883329e-01 -6.62077814e-02
-1.26012981e+00 -3.47941041e-01 2.07326010e-01 4.97482896e-01
2.11756434e-02 7.72407651e-01 -1.01246245e-01 7.96070278e-01
5.76119006e-01 7.93558002e-01 -3.74752373e-01 -3.88970226e-01
2.56059021e-01 1.17434943e+00 6.37227833e-01 -3.72201294e-01
2.19218433e-01 3.50450844e-01 -2.02795640e-01 -1.30576456e+00
-1.04844701e+00 -1.13743186e+00 -6.09607458e-01 -2.13948205e-01
5.84629655e-01 -6.77535713e-01 -4.81062457e-02 4.81665134e-03
-1.32440424e+00 6.43436313e-01 -4.12286490e-01 2.97414452e-01
-5.19111514e-01 1.03648818e+00 -2.44855061e-01 -7.73802817e-01
-1.22523665e+00 -5.45633316e-01 1.01960039e+00 4.04470980e-01
-7.57175505e-01 -6.13721907e-01 4.27981198e-01 2.80299693e-01
-1.91477478e-01 3.98995638e-01 1.11686635e+00 -1.28226447e+00
1.08567104e-01 -4.81849670e-01 2.12108612e-01 3.86360079e-01
4.69804496e-01 3.17762673e-01 -1.48678735e-01 -3.31380010e-01
1.43406078e-01 -5.77398725e-02 1.36431825e+00 6.36648774e-01
6.54342353e-01 -8.02627206e-01 -4.34423447e-01 -1.76499888e-01
1.24859583e+00 2.60243148e-01 4.06777054e-01 2.28221670e-01
3.47663492e-01 7.35132337e-01 7.31573582e-01 8.01520288e-01
3.95986557e-01 1.70221608e-02 -2.23885179e-01 4.65366632e-01
6.33974448e-02 -2.65197717e-02 2.55051434e-01 1.38040996e+00
2.14078262e-01 -8.02713692e-01 -4.45523888e-01 8.79905164e-01
-2.12115216e+00 -1.55262101e+00 -5.68139374e-01 1.64532983e+00
9.60625231e-01 8.19851309e-02 4.24487293e-01 3.81617308e-01
9.12828267e-01 4.40887362e-01 -1.32084191e-01 -1.02436244e+00
-4.00324106e-01 -3.25649112e-01 9.95895639e-02 2.47350782e-01
-9.28233266e-01 7.01483130e-01 6.44898319e+00 9.79247510e-01
-3.47032189e-01 -2.79282510e-01 1.76549122e-01 -2.60425091e-01
-6.42237067e-01 -5.61562032e-02 -1.21430087e+00 3.37476909e-01
9.83048379e-01 -8.98854375e-01 -4.66096878e-01 7.31776118e-01
3.78389210e-01 -6.52506173e-01 -7.64357328e-01 4.58128989e-01
7.57974744e-01 -1.62180889e+00 6.65338397e-01 -2.94532686e-01
1.49126625e+00 -2.74426043e-01 -5.23190856e-01 1.13058932e-01
2.77959496e-01 -2.68191695e-01 6.99410558e-01 3.99117798e-01
4.42456484e-01 -9.52993512e-01 8.78714442e-01 8.80121231e-01
-1.00367796e+00 -4.22552526e-02 -5.93642414e-01 2.34136358e-01
1.67116821e-01 8.48116934e-01 -7.19372213e-01 1.30028045e+00
3.86192739e-01 8.33352566e-01 -5.51468015e-01 9.02631879e-01
9.80706606e-03 4.88095820e-01 2.79771592e-02 -4.56437975e-01
4.75511402e-01 -2.33011797e-01 9.76214111e-01 1.71787190e+00
3.83613825e-01 1.94052055e-01 2.31850833e-01 1.54551014e-01
3.84420231e-02 5.22743702e-01 -4.45376813e-01 -2.57224739e-01
5.00959873e-01 1.08896363e+00 -8.16272557e-01 -9.52636361e-01
-3.33185978e-02 7.98378527e-01 -9.27707646e-03 4.90297601e-02
-2.69016892e-01 -8.90173614e-01 -2.25089416e-01 -1.78851768e-01
5.48545182e-01 -9.45931599e-02 -5.40644586e-01 -1.04987013e+00
1.08702488e-01 -1.00842524e+00 7.50467598e-01 -3.55161130e-01
-9.00310338e-01 6.08495235e-01 3.32015097e-01 -1.20399404e+00
-4.51182425e-01 4.04345572e-01 -1.08302832e+00 2.91005462e-01
-1.00360274e+00 -9.10565794e-01 -2.68953554e-02 2.33677268e-01
1.01026690e+00 -5.70001245e-01 7.57146358e-01 -3.69235665e-01
-5.38634181e-01 4.00255620e-02 7.68868923e-01 -3.57688040e-01
4.94233459e-01 -1.23761845e+00 1.99151337e-01 1.01528680e+00
2.17982326e-02 5.72445273e-01 9.69299853e-01 -1.08972108e+00
-9.05025482e-01 -1.06711996e+00 1.33349442e+00 -7.35931620e-02
2.25581124e-01 2.28460625e-01 -7.17795491e-01 4.72460151e-01
6.79617107e-01 -1.31935525e+00 1.10918558e+00 -5.67311160e-02
-1.53063126e-02 -7.54786134e-02 -9.39742386e-01 4.22151238e-01
5.13821483e-01 1.69303626e-01 -1.33038306e+00 3.61440510e-01
6.44372463e-01 3.64905186e-02 -5.39048254e-01 1.57197610e-01
4.12756234e-01 -9.59036767e-01 4.53547060e-01 -5.73288083e-01
8.51435363e-01 -1.76545545e-01 -8.77972916e-02 -1.47447872e+00
-2.35624954e-01 -5.57473242e-01 -1.56476840e-01 1.71715653e+00
3.21876585e-01 -2.46085525e-01 3.84516358e-01 -1.30778119e-01
-4.94591445e-01 -4.65468735e-01 -6.71590507e-01 -6.09156132e-01
-1.30555838e-01 2.68344373e-01 4.18053448e-01 6.61236107e-01
4.86993253e-01 8.11914623e-01 -3.85331064e-01 -2.24968359e-01
6.18738651e-01 6.72475517e-01 8.89903188e-01 -1.29493153e+00
8.54242668e-02 -5.35072803e-01 3.95778492e-02 -7.22039521e-01
2.70287246e-01 -8.09503376e-01 -6.36482611e-02 -2.49943614e+00
7.80508757e-01 3.91976982e-01 2.39145830e-01 -8.38258583e-03
-2.50655770e-01 -3.70600104e-01 8.26554671e-02 4.51167166e-01
-1.18734872e+00 5.08207858e-01 1.00224829e+00 -2.88244009e-01
-7.11129367e-01 1.91231728e-01 -1.29131567e+00 6.51651502e-01
8.43610585e-01 -5.44790030e-01 -5.30499399e-01 -5.78652620e-02
6.36286438e-02 1.48970157e-01 -5.28719127e-01 -9.62391436e-01
4.95396763e-01 -1.81667373e-01 1.47982761e-01 -1.49723697e+00
-3.05970639e-01 -2.26270631e-01 -2.95666140e-02 5.17874956e-01
-6.95106030e-01 5.19901374e-03 8.72068033e-02 6.54818475e-01
-4.33790356e-01 -7.34349012e-01 5.39490819e-01 -3.34738582e-01
-3.72094065e-01 -2.37718359e-01 -7.10249364e-01 2.48771846e-01
9.07268882e-01 -4.28414792e-01 -5.08220732e-01 -4.32606459e-01
-1.57119855e-01 4.43032026e-01 1.28764421e-01 3.18039715e-01
6.42878234e-01 -7.95407891e-01 -1.30993712e+00 -1.76306799e-01
6.75294846e-02 -7.75883794e-02 3.31893414e-01 3.99674416e-01
-5.17171562e-01 8.06038797e-01 -1.01215109e-01 -2.31789365e-01
-1.70285439e+00 1.76989049e-01 -3.06849152e-01 -7.75759578e-01
-7.41139948e-01 2.71634340e-01 -1.77390128e-01 1.45802468e-01
4.43049930e-02 -1.55381277e-01 -7.52564788e-01 6.27583563e-01
7.99648583e-01 6.80850625e-01 1.49367884e-01 -4.60178375e-01
-1.64294153e-01 3.58079821e-01 -5.02044022e-01 -6.09790497e-02
1.46012461e+00 -4.02157545e-01 -5.45041323e-01 4.47817564e-01
1.10972619e+00 4.07055795e-01 -2.40734324e-01 -4.49868515e-02
3.10554683e-01 -1.10149914e-02 -2.02826500e-01 -5.51058352e-01
-2.73939222e-01 2.38347873e-01 -6.24409020e-01 6.88709617e-01
1.02707839e+00 1.77893847e-01 6.88636601e-01 6.17297113e-01
3.17128114e-02 -1.49730790e+00 2.33734533e-01 5.10581672e-01
1.18573904e+00 -7.36559391e-01 7.04764307e-01 -2.36813679e-01
-9.98669028e-01 1.20017791e+00 2.45293066e-01 -1.42169997e-01
9.06347781e-02 -2.52636727e-02 -5.20696938e-01 -2.28684396e-01
-9.30848956e-01 2.26893704e-02 4.63283449e-01 2.35720992e-01
2.74823517e-01 -6.16865652e-03 -1.16438270e+00 8.31023812e-01
-5.20800531e-01 -4.57549155e-01 1.02810693e+00 1.03386283e+00
-1.32957017e+00 -7.65874863e-01 -2.90108353e-01 9.14859593e-01
-5.63858390e-01 1.21801719e-01 -9.82971370e-01 6.01444542e-01
-3.68975878e-01 1.44029331e+00 -1.76633149e-02 -5.49488515e-02
4.61791724e-01 -4.64061322e-03 3.28091085e-01 -1.22036600e+00
-8.59654009e-01 3.33043993e-01 5.76460600e-01 1.86804906e-01
-6.30627275e-01 -8.45544517e-01 -1.38622487e+00 -2.23459512e-01
-4.89258528e-01 9.19128895e-01 4.75730449e-01 9.78100002e-01
3.60721439e-01 6.92782283e-01 9.22410429e-01 -5.12504220e-01
-7.93899059e-01 -1.22035182e+00 -9.15988624e-01 1.36855766e-01
3.76970947e-01 1.37010962e-01 -2.94999182e-01 1.79160342e-01] | [12.580982208251953, 9.573553085327148] |
63ed4671-2ac3-4d1a-b17e-1c8ed0ddfc97 | clear-contrastive-learning-for-sentence | 2012.15466 | null | https://arxiv.org/abs/2012.15466v1 | https://arxiv.org/pdf/2012.15466v1.pdf | CLEAR: Contrastive Learning for Sentence Representation | Pre-trained language models have proven their unique powers in capturing implicit language features. However, most pre-training approaches focus on the word-level training objective, while sentence-level objectives are rarely studied. In this paper, we propose Contrastive LEArning for sentence Representation (CLEAR), which employs multiple sentence-level augmentation strategies in order to learn a noise-invariant sentence representation. These augmentations include word and span deletion, reordering, and substitution. Furthermore, we investigate the key reasons that make contrastive learning effective through numerous experiments. We observe that different sentence augmentations during pre-training lead to different performance improvements on various downstream tasks. Our approach is shown to outperform multiple existing methods on both SentEval and GLUE benchmarks. | ['Hao Ma', 'Fei Sun', 'Madian Khabsa', 'Jiatao Gu', 'Sinong Wang', 'Zhuofeng Wu'] | 2020-12-31 | null | null | null | null | ['linguistic-acceptability'] | ['natural-language-processing'] | [ 5.65086067e-01 -8.36101621e-02 -3.78710896e-01 -5.55818141e-01
-1.02720737e+00 -4.56898212e-01 6.31298065e-01 5.33339202e-01
-7.56097257e-01 8.37240934e-01 6.74046159e-01 -6.31860614e-01
3.42682689e-01 -4.85589504e-01 -4.99002457e-01 -2.49230400e-01
3.50292623e-02 1.18464734e-02 -6.69924170e-02 -7.23992407e-01
3.45217049e-01 1.38555065e-01 -9.04614925e-01 5.70451677e-01
9.39020336e-01 4.33542192e-01 2.54898995e-01 6.89144135e-01
-5.45301080e-01 7.93168485e-01 -7.67280400e-01 -5.49028516e-01
6.61744401e-02 -4.24457282e-01 -8.10253263e-01 -8.34119394e-02
4.41440672e-01 -3.40403430e-02 -1.90778524e-01 8.39264095e-01
4.83360142e-01 2.72310138e-01 4.96317208e-01 -6.64111555e-01
-7.44579077e-01 1.10639513e+00 -5.07915914e-01 6.11898303e-01
3.84053349e-01 6.23994917e-02 1.42773604e+00 -1.10395908e+00
4.50622141e-01 1.27144408e+00 6.99613750e-01 6.65299594e-01
-1.49500477e+00 -4.91708845e-01 4.39515591e-01 4.22160883e-05
-9.98507798e-01 -6.69966221e-01 9.16824043e-01 -8.17785934e-02
1.40350878e+00 1.63808003e-01 1.93039570e-02 1.30873978e+00
2.02758104e-01 1.11173940e+00 1.11804450e+00 -8.64465475e-01
-1.24553092e-01 -6.30250871e-02 6.04803920e-01 5.02501488e-01
2.04254434e-01 -1.25969216e-01 -5.60027182e-01 -4.06053662e-03
4.16827917e-01 -1.63068473e-01 -1.30336121e-01 1.37889504e-01
-9.39887822e-01 9.77411866e-01 2.21703649e-01 6.11834526e-01
-2.02375397e-01 8.88273865e-02 9.39839602e-01 6.36463881e-01
8.24654579e-01 7.25574970e-01 -9.13987100e-01 -2.34446436e-01
-8.88479352e-01 -2.67242882e-02 5.87058246e-01 6.35985136e-01
5.07785857e-01 2.81733245e-01 -7.26924837e-01 9.60325897e-01
7.36355409e-02 2.39067793e-01 7.37059116e-01 -4.36971933e-01
9.56153810e-01 5.44638038e-01 -3.78060967e-01 -7.11558938e-01
-2.94858098e-01 -8.23078811e-01 -1.00561440e+00 -3.36206377e-01
5.68988398e-02 -2.17516527e-01 -9.13698316e-01 1.90440464e+00
-2.80506223e-01 9.65816155e-02 3.01152617e-01 3.80862564e-01
9.47254956e-01 5.59477150e-01 3.83402884e-01 -2.51696140e-01
1.12382007e+00 -1.16165590e+00 -7.51591027e-01 -5.80494463e-01
1.21197367e+00 -7.36061275e-01 1.53557944e+00 4.38559651e-02
-1.25586295e+00 -5.77959538e-01 -9.33405936e-01 -3.25302690e-01
-2.11590543e-01 1.13072649e-01 6.46146297e-01 5.71536362e-01
-8.99496198e-01 4.28503633e-01 -7.00757146e-01 -1.02721535e-01
2.16131166e-01 1.36969268e-01 -2.58938879e-01 -4.23116386e-02
-1.41531205e+00 9.47639287e-01 3.93370271e-01 -1.41283512e-01
-3.84194404e-01 -7.08343387e-01 -1.21387565e+00 2.67743707e-01
3.34019333e-01 -7.76425779e-01 1.40891850e+00 -9.03614223e-01
-1.52760363e+00 8.23658884e-01 -4.95723486e-01 -7.14642644e-01
1.41848132e-01 -4.42484796e-01 -2.66250342e-01 -1.92232832e-01
4.19384129e-02 4.23525214e-01 6.84320748e-01 -1.00510585e+00
-2.01825470e-01 -4.44355495e-02 2.24672049e-01 3.07505667e-01
-6.19905591e-01 3.83773088e-01 -2.07202673e-01 -1.20932007e+00
-2.16223955e-01 -4.10556257e-01 -5.05915523e-01 -6.25388324e-01
-4.97409850e-01 -3.65944475e-01 3.61306459e-01 -7.63327062e-01
1.58872938e+00 -2.11983776e+00 1.21305123e-01 -1.02998190e-01
-8.24588910e-02 5.66988766e-01 -6.38377547e-01 6.60888672e-01
-1.54693410e-01 4.48098421e-01 -4.46019679e-01 -9.41291749e-01
-1.35932192e-01 3.16724390e-01 -4.60061312e-01 -9.27980021e-02
5.75492501e-01 1.16223204e+00 -8.99893463e-01 -4.29638147e-01
4.84344065e-02 3.94140303e-01 -5.68318725e-01 1.29437909e-01
-1.73842236e-01 3.29151213e-01 -2.15379030e-01 3.59199017e-01
3.96240950e-01 -4.03640382e-02 2.86571264e-01 1.61806658e-01
7.22581968e-02 1.28183401e+00 -5.73106885e-01 1.86478865e+00
-8.23722184e-01 5.07567286e-01 -1.12242401e-01 -1.06561112e+00
8.68069649e-01 2.45318681e-01 -7.05107972e-02 -7.46050298e-01
1.40845209e-01 7.35581368e-02 2.03563794e-01 -2.23149553e-01
7.69886494e-01 -3.47594142e-01 -2.17400521e-01 4.32082355e-01
8.27861950e-02 -4.38940413e-02 4.77577418e-01 4.66406286e-01
1.20010686e+00 -1.02462687e-01 4.39413697e-01 -2.08955020e-01
6.80295527e-01 -3.62443089e-01 4.77557689e-01 8.17510188e-01
1.84523240e-01 6.02990985e-01 6.01870477e-01 1.24443313e-02
-7.13295937e-01 -8.40876341e-01 9.25730690e-02 1.42920911e+00
-3.31116080e-01 -7.49608815e-01 -5.73989034e-01 -1.05066466e+00
-1.11170426e-01 9.92678046e-01 -3.99065942e-01 -2.87215054e-01
-8.17033112e-01 -6.32231236e-01 7.50420749e-01 7.98962831e-01
2.82663852e-01 -1.17048502e+00 1.12292571e-02 2.24851951e-01
-1.65187329e-01 -1.14787662e+00 -7.52129316e-01 3.09219807e-01
-1.15096450e+00 -7.35557020e-01 -3.44689727e-01 -9.55248654e-01
7.16721356e-01 3.38205606e-01 1.46503663e+00 3.73383850e-01
7.26895556e-02 9.42693129e-02 -5.33017337e-01 -2.09208593e-01
-5.06547570e-01 6.98200583e-01 -7.03191906e-02 -3.39988768e-01
2.82580823e-01 -5.45641780e-01 -1.46574765e-01 -4.01433676e-01
-9.38041866e-01 -1.01861767e-01 7.77477920e-01 1.06374657e+00
2.56443769e-01 -4.87161428e-01 8.45702946e-01 -1.27293432e+00
1.16876328e+00 -3.01840454e-01 4.71443683e-02 2.32965872e-01
-4.89219666e-01 3.35451126e-01 7.50159562e-01 -2.03820303e-01
-1.04324889e+00 -3.63329649e-01 -5.83171666e-01 7.18438551e-02
-1.72939464e-01 1.00620854e+00 -1.28031805e-01 2.91003704e-01
5.43101549e-01 4.20384645e-01 -6.21473789e-02 -7.56996691e-01
3.21522981e-01 5.54051161e-01 3.08032632e-01 -6.30576849e-01
8.99574041e-01 -8.13680440e-02 -3.00796092e-01 -9.06149328e-01
-1.34322059e+00 -3.78719777e-01 -7.80127108e-01 5.18151700e-01
4.10308212e-01 -9.53852415e-01 -1.33858100e-01 1.32167965e-01
-1.33097541e+00 -2.17378303e-01 -5.15404940e-01 2.64571130e-01
-6.25799969e-02 6.75324321e-01 -9.08514857e-01 -7.47141182e-01
-7.16076612e-01 -8.68636668e-01 1.01073670e+00 -1.78566590e-01
-4.30218488e-01 -1.26708496e+00 2.03536674e-01 3.33590090e-01
5.37341237e-01 -2.66493112e-01 1.05631316e+00 -9.61863399e-01
-1.20407157e-01 -8.69358405e-02 -4.57242653e-02 8.38745415e-01
2.34958261e-01 -2.65471816e-01 -6.63439572e-01 -4.30630356e-01
-1.78994134e-01 -4.03953463e-01 1.28318298e+00 1.67844877e-01
1.29135907e+00 -4.51184988e-01 3.50049473e-02 5.57361901e-01
1.20830607e+00 -3.11292768e-01 5.60308456e-01 3.53137851e-01
5.36767125e-01 3.95840377e-01 3.76075536e-01 8.00988674e-02
3.11461419e-01 5.32510936e-01 6.44130930e-02 -2.25864798e-01
-2.49737754e-01 -3.00655186e-01 6.32725954e-01 1.32492065e+00
1.67369440e-01 -3.86676699e-01 -7.25327969e-01 5.78083754e-01
-1.69297349e+00 -8.63918960e-01 -1.28427565e-01 1.89257491e+00
1.18742704e+00 6.41762316e-01 -8.42736810e-02 1.98393106e-01
3.66556436e-01 4.07682747e-01 -2.07064256e-01 -8.37053537e-01
-3.79869193e-01 7.34622121e-01 2.81474590e-01 6.40901268e-01
-1.13542080e+00 1.27696192e+00 6.74267244e+00 7.49704480e-01
-1.04044580e+00 2.12191954e-01 6.47259712e-01 -8.81054997e-02
-7.44175553e-01 -1.27337668e-02 -9.04256523e-01 3.36219341e-01
9.08350468e-01 -2.17879787e-01 -5.83015033e-04 5.52593589e-01
1.97325349e-01 2.48952106e-01 -1.04642391e+00 6.37606919e-01
2.57690102e-01 -1.35674262e+00 2.64932543e-01 -3.47599119e-01
6.69845462e-01 1.33445412e-01 3.74648534e-02 8.30788374e-01
1.97049439e-01 -9.95699286e-01 3.43771696e-01 2.25383624e-01
6.87091470e-01 -7.50413001e-01 1.01310611e+00 3.80974025e-01
-9.88799036e-01 1.36750534e-01 -1.44366145e-01 -6.25594914e-01
3.49287540e-01 5.65901458e-01 -8.88089359e-01 6.44641638e-01
1.07629530e-01 8.07088673e-01 -8.31130147e-01 6.77180111e-01
-6.88372910e-01 1.20398474e+00 -3.40622067e-02 -1.33771867e-01
4.13008809e-01 -1.95498988e-02 5.27943969e-01 1.71466911e+00
-1.08704597e-01 -2.14355424e-01 2.27286428e-01 5.01686633e-01
-5.08442044e-01 4.01193857e-01 -5.52362859e-01 -1.87739193e-01
3.37603331e-01 9.97429371e-01 -3.15311521e-01 -3.63336951e-01
-6.20034456e-01 9.27326262e-01 6.73659801e-01 2.99083382e-01
-4.39191103e-01 -3.97502214e-01 8.93941760e-01 -3.56642678e-02
2.50095963e-01 -6.12949431e-01 -5.49388409e-01 -1.41924834e+00
1.15113959e-01 -9.24783170e-01 3.28981042e-01 -2.36862004e-01
-1.45985460e+00 6.21778429e-01 -2.16745570e-01 -6.91089392e-01
-4.15013820e-01 -4.96470213e-01 -8.60946774e-01 1.06082416e+00
-1.90238965e+00 -1.08517230e+00 3.97838861e-01 1.68715760e-01
1.04676056e+00 -2.41836652e-01 8.75162542e-01 2.65950471e-01
-7.91795194e-01 1.02755558e+00 9.99552086e-02 4.34766412e-01
6.07534528e-01 -1.25971365e+00 9.19927597e-01 1.01632869e+00
5.51085472e-01 9.51277554e-01 5.65221906e-01 -5.39544523e-01
-1.11700499e+00 -1.07197213e+00 1.49393976e+00 -4.09718245e-01
7.39999354e-01 -5.74807167e-01 -1.11337435e+00 8.20755541e-01
4.81175661e-01 -1.49092838e-01 8.53854060e-01 6.07885838e-01
-4.78326082e-01 7.95443058e-02 -7.03566968e-01 6.83918357e-01
1.11935258e+00 -7.38312721e-01 -1.07438588e+00 2.03328267e-01
1.03324664e+00 -1.45596743e-01 -4.61208522e-01 4.87474024e-01
-1.48363598e-02 -4.69923139e-01 9.88368213e-01 -1.31476426e+00
6.97162986e-01 2.86027372e-01 -3.55313197e-02 -1.51259434e+00
-3.76205027e-01 -5.37721276e-01 -2.46585354e-01 1.58254683e+00
7.63873398e-01 -6.62745118e-01 6.11861646e-01 7.53981024e-02
-4.12823260e-01 -9.70268488e-01 -7.17001677e-01 -9.73708391e-01
3.97318155e-01 -4.63320881e-01 3.32255840e-01 1.04798996e+00
6.30482957e-02 9.65012610e-01 -2.76288122e-01 -2.14208573e-01
2.62527555e-01 7.79723004e-03 5.49915910e-01 -8.63137245e-01
-3.19518059e-01 -5.96341968e-01 -1.82973936e-01 -1.36101234e+00
6.40640378e-01 -1.26659393e+00 -1.04296077e-02 -1.55300021e+00
2.79407948e-01 -2.15479478e-01 -5.62107265e-01 5.86911082e-01
-6.89380109e-01 1.17430938e-02 2.57231176e-01 -2.07004771e-01
-5.60795546e-01 7.85189331e-01 9.52743649e-01 -1.65525794e-01
-2.00918466e-01 -1.00546449e-01 -9.84964669e-01 6.50138378e-01
1.12226510e+00 -4.28811491e-01 -2.23410130e-01 -9.12795901e-01
3.10775489e-01 -3.54548365e-01 -5.05297594e-02 -5.88562906e-01
-5.66218384e-02 -2.96992753e-02 -7.36110508e-02 -5.42926669e-01
2.15906724e-01 -2.68401027e-01 -8.76716971e-01 4.87379789e-01
-8.56797516e-01 2.37510890e-01 3.27989161e-01 3.48145604e-01
-4.13753122e-01 -4.91863489e-01 4.48300451e-01 -1.94296360e-01
-5.58336020e-01 8.62895027e-02 -4.92360204e-01 4.73619044e-01
4.22764421e-01 1.14986636e-01 -3.06645017e-02 -2.83919066e-01
-4.63965297e-01 2.36599922e-01 7.21987113e-02 5.40723681e-01
6.77848577e-01 -1.09183264e+00 -1.10431814e+00 2.57776707e-01
7.79486028e-03 -9.57851410e-02 -1.03139773e-01 6.87306464e-01
-1.41889796e-01 5.42763531e-01 2.89375097e-01 -2.52909064e-01
-1.51596105e+00 3.36430758e-01 -3.33984047e-02 -7.57201433e-01
-4.50787127e-01 1.05738902e+00 -3.13551724e-02 -7.57586181e-01
2.63226748e-01 -5.30105233e-01 -2.48872533e-01 -7.21961185e-02
3.79027069e-01 9.48791951e-02 2.75750220e-01 -4.83247191e-01
-3.30788553e-01 1.50613233e-01 -5.12844741e-01 9.43487659e-02
1.44050229e+00 -9.02223960e-02 -1.30374998e-01 4.92236972e-01
1.26105356e+00 2.46373177e-01 -6.94619417e-01 -6.59314752e-01
5.11544228e-01 -2.55611479e-01 5.64792566e-02 -7.72041500e-01
-7.45245516e-01 1.07187486e+00 -6.90264106e-02 1.50081992e-01
1.09492218e+00 -1.11259229e-01 1.10921407e+00 6.24540389e-01
6.67529404e-02 -9.44358647e-01 1.43869713e-01 1.19206774e+00
8.67615283e-01 -1.31605709e+00 -9.76908654e-02 -6.39735103e-01
-6.14498794e-01 9.11330640e-01 6.38824642e-01 -8.95206183e-02
3.88082504e-01 3.85657042e-01 -6.31812494e-03 1.40028954e-01
-1.05994797e+00 -2.93106228e-01 4.11018282e-01 1.51069686e-01
1.12337828e+00 -1.00747794e-01 -7.09977746e-01 7.38939345e-01
-3.10476035e-01 -2.35118642e-01 3.23319495e-01 1.06372404e+00
-3.98026705e-01 -1.62222433e+00 1.39919976e-02 5.89700699e-01
-7.14547575e-01 -8.28358710e-01 -5.38323581e-01 5.86130321e-01
-2.89389908e-01 9.00341451e-01 3.17947045e-02 -1.91739738e-01
4.12232071e-01 3.59965682e-01 4.28897530e-01 -1.27698576e+00
-9.94881570e-01 -2.82985806e-01 5.67491770e-01 -1.79134861e-01
-2.55266160e-01 -5.78693807e-01 -1.27610087e+00 -3.25874165e-02
-2.79018879e-01 3.27248961e-01 4.18577194e-01 1.16514289e+00
4.58942890e-01 8.19840789e-01 5.79239607e-01 -5.22615910e-01
-8.95715296e-01 -1.33832073e+00 -1.44313186e-01 4.90148962e-01
4.38162416e-01 -2.61203140e-01 -3.77019525e-01 -1.51058406e-01] | [10.91950798034668, 8.751779556274414] |
0835b737-05bc-4593-b481-2866fddac7c6 | position-offset-label-prediction-for | null | null | https://aclanthology.org/2022.coling-1.480 | https://aclanthology.org/2022.coling-1.480.pdf | Position Offset Label Prediction for Grammatical Error Correction | We introduce a novel position offset label prediction subtask to the encoder-decoder architecture for grammatical error correction (GEC) task. To keep the meaning of the input sentence unchanged, only a few words should be inserted or deleted during correction, and most of tokens in the erroneous sentence appear in the paired correct sentence with limited position movement. Inspired by this observation, we design an auxiliary task to predict position offset label (POL) of tokens, which is naturally capable of integrating different correction editing operations into a unified framework. Based on the predicted POL, we further propose a new copy mechanism (P-copy) to replace the vanilla copy module. Experimental results on Chinese, English and Japanese datasets demonstrate that our proposed POL-Pc framework obviously improves the performance of baseline models. Moreover, our model yields consistent performance gain over various data augmentation methods. Especially, after incorporating synthetic data, our model achieves a 38.95 F-0.5 score on Chinese GEC dataset, which outperforms the previous state-of-the-art by a wide margin of 1.98 points. | ['Yunfang Wu', 'Xu sun', 'Jingsong Yu', 'Xiuyu Wu'] | null | null | null | null | coling-2022-10 | ['grammatical-error-correction'] | ['natural-language-processing'] | [ 3.55180740e-01 2.27846295e-01 -2.48797052e-02 -5.29963315e-01
-9.79457319e-01 -1.80112924e-02 1.46743685e-01 2.75124222e-01
-5.94945371e-01 8.66413951e-01 3.91119003e-01 -3.40649277e-01
6.99303150e-01 -3.46736580e-01 -1.06400418e+00 -3.92009676e-01
5.52841961e-01 2.49823779e-02 1.35654256e-01 -2.90405929e-01
3.61562997e-01 -4.24811482e-01 -1.04308844e+00 3.32907706e-01
1.59380889e+00 4.06099737e-01 6.21178806e-01 3.98716241e-01
-2.01169014e-01 6.20658219e-01 -7.02869058e-01 -7.14939892e-01
-2.55840600e-01 -6.81762040e-01 -6.03949487e-01 -3.05026203e-01
1.32768467e-01 -2.95364499e-01 -7.91078880e-02 1.30581510e+00
4.46836472e-01 -5.25249615e-02 3.99073303e-01 -6.17291451e-01
-1.26091182e+00 1.07435679e+00 -6.04348958e-01 -1.22439498e-02
2.19068930e-01 -1.48442522e-01 9.80545878e-01 -1.35976291e+00
5.38128853e-01 1.05510020e+00 6.22682750e-01 8.85278165e-01
-7.83787549e-01 -5.71299314e-01 4.99980181e-01 2.25960657e-01
-1.29801726e+00 -4.14748788e-01 5.02376914e-01 7.81931952e-02
1.23776007e+00 2.39890784e-01 2.57826924e-01 8.47831190e-01
2.42205709e-01 8.88832927e-01 6.49866819e-01 -7.93797255e-01
-1.90241367e-01 1.54485591e-02 6.88584223e-02 6.81604743e-01
3.02903861e-01 -2.49630883e-01 -4.40925717e-01 3.37939918e-01
4.74041328e-02 -1.80711195e-01 -4.66881245e-01 3.43426228e-01
-1.16286421e+00 3.85257155e-01 2.38642752e-01 8.84375498e-02
-7.12109031e-03 5.59652805e-01 5.55059910e-01 3.07391405e-01
6.90988660e-01 1.32561803e-01 -6.32426798e-01 -4.76248473e-01
-6.74597561e-01 3.20627868e-01 2.79532582e-01 1.52259314e+00
6.32426322e-01 8.56486559e-02 -5.85397124e-01 1.05684185e+00
4.51020658e-01 7.03621745e-01 5.05311191e-01 -5.44017673e-01
1.15388358e+00 6.04281783e-01 1.17593750e-01 -6.70084715e-01
-1.49047717e-01 -7.64657438e-01 -9.46044981e-01 -5.25571764e-01
-4.87110810e-03 -1.10891469e-01 -6.96718693e-01 1.75724268e+00
6.58938363e-02 9.30873007e-02 -1.26070872e-01 8.05755734e-01
6.43718779e-01 6.62690222e-01 2.08290845e-01 -3.48236591e-01
1.13176799e+00 -1.22812378e+00 -1.07354629e+00 -3.77529323e-01
1.19944894e+00 -8.90538931e-01 1.40257943e+00 1.22491196e-01
-1.05025840e+00 -5.30450642e-01 -1.04416001e+00 -5.46725214e-01
1.16108887e-01 7.58810997e-01 2.64766335e-01 6.54865742e-01
-7.90997922e-01 6.50168777e-01 -8.44075620e-01 9.54341888e-02
1.47348791e-01 -9.87841710e-02 -8.36337954e-02 -2.58032650e-01
-1.14961672e+00 8.30922067e-01 4.07124519e-01 3.61432970e-01
-4.09557670e-01 -7.46039629e-01 -8.57787549e-01 1.14251204e-01
1.35990366e-01 -3.41925353e-01 1.44279528e+00 -6.07829452e-01
-1.30054808e+00 6.41962171e-01 -8.44342887e-01 -3.18682939e-01
6.15560770e-01 -4.36139375e-01 -5.59891105e-01 -5.23216069e-01
1.96406186e-01 5.97420275e-01 4.39690173e-01 -1.09131861e+00
-8.10571790e-01 -8.71034339e-02 -4.39566702e-01 2.07134351e-01
-1.02419600e-01 3.45125973e-01 -8.10230434e-01 -1.15979326e+00
1.42175943e-01 -8.25443625e-01 -8.27440247e-02 -2.96576083e-01
-5.27895570e-01 -2.19120115e-01 3.82536709e-01 -1.13530815e+00
1.95789349e+00 -2.13791585e+00 1.58736959e-01 -1.44899309e-01
9.11535099e-02 5.08461118e-01 -2.61773705e-01 3.17005396e-01
-9.10989121e-02 4.19443727e-01 -5.59094310e-01 -9.97769892e-01
2.75138784e-02 1.51715115e-01 -3.83590281e-01 1.83819190e-01
3.21883321e-01 1.10714328e+00 -1.03383386e+00 -2.58021414e-01
-3.02544802e-01 7.73412436e-02 -8.71402800e-01 3.27695794e-02
-2.30296180e-01 4.25090134e-01 -1.14158183e-01 6.43095255e-01
1.15048718e+00 -3.66592705e-02 1.76574022e-01 4.62827951e-01
-4.27607186e-02 8.67675781e-01 -8.64609480e-01 1.81660390e+00
-3.99316907e-01 3.01504105e-01 -4.20459598e-01 -5.84266782e-01
1.17355561e+00 2.30299398e-01 -3.20142329e-01 -7.88113773e-01
-7.28699705e-03 5.32700300e-01 7.83174559e-02 -2.00438142e-01
1.12769961e+00 1.66084543e-01 -2.16064259e-01 5.64330578e-01
-9.45692286e-02 3.18707913e-01 1.51656106e-01 2.15567634e-01
9.14672315e-01 1.27504975e-01 1.29495248e-01 4.33722846e-02
6.79796040e-01 -2.98517019e-01 1.21969914e+00 6.74274266e-01
-3.11860651e-01 7.58763611e-01 6.60834610e-01 2.30051540e-02
-1.05542052e+00 -6.04937315e-01 1.34790331e-01 8.67600560e-01
1.63080066e-01 -7.76346326e-01 -1.13280988e+00 -8.25988829e-01
-1.99847981e-01 9.80374694e-01 -4.48173702e-01 -4.33154553e-01
-8.74395847e-01 -8.14562917e-01 8.00912797e-01 7.30212510e-01
5.13233602e-01 -1.00988555e+00 2.95973092e-01 4.99684781e-01
-5.29250264e-01 -9.30668592e-01 -1.05899525e+00 -1.66070014e-01
-8.93024206e-01 -7.16515601e-01 -4.84604388e-01 -1.08773184e+00
8.96852255e-01 2.80372560e-01 8.08442593e-01 6.96787059e-01
3.76112163e-01 -4.09314930e-01 -8.53325367e-01 -3.57043028e-01
-6.52121425e-01 1.60827994e-01 -6.95847571e-02 -2.78418690e-01
4.81521279e-01 -2.78974831e-01 -2.71519899e-01 -3.35827060e-02
-4.45339292e-01 3.91268581e-01 4.59001124e-01 1.05159783e+00
7.78096855e-01 -5.05419731e-01 7.76975632e-01 -1.27461743e+00
6.72035277e-01 -3.03227425e-01 -2.58519739e-01 6.38953686e-01
-9.75824058e-01 1.35008857e-01 7.96248734e-01 -1.43930584e-01
-1.28161454e+00 -1.04542568e-01 -4.37103063e-01 -2.35032383e-02
3.42328370e-01 6.40651166e-01 -5.22186041e-01 2.09001631e-01
1.46204829e-01 5.21041512e-01 -2.26428583e-01 -9.15693462e-01
3.86112362e-01 9.21936870e-01 7.11245358e-01 -4.55311358e-01
4.74558830e-01 -2.63738066e-01 -5.35735190e-01 -1.85640603e-02
-7.91959465e-01 -2.51125195e-03 -7.24631250e-01 -4.86917272e-02
4.84659851e-01 -1.20986843e+00 -3.57520670e-01 8.57112527e-01
-1.77249455e+00 -1.49369352e-02 1.32957920e-01 3.18123370e-01
-4.73895967e-02 6.43712521e-01 -9.15716648e-01 -5.41996181e-01
-5.51986575e-01 -1.15784717e+00 1.03265333e+00 1.60205469e-01
3.36019248e-02 -6.72617853e-01 -1.79908887e-01 2.78388023e-01
3.49999994e-01 -2.51267254e-01 1.16898060e+00 -4.55436468e-01
-6.98476851e-01 -1.57138690e-01 -2.85778791e-01 6.10812545e-01
8.55716392e-02 -8.01823363e-02 -6.46689713e-01 -2.43686661e-01
-2.96922177e-01 2.68121138e-02 1.18979692e+00 -2.74237812e-01
1.59811926e+00 -2.10327074e-01 -1.50764078e-01 6.58546329e-01
1.15198970e+00 5.42013794e-02 8.74624491e-01 4.39648703e-02
8.75488520e-01 1.11367725e-01 8.34596038e-01 3.67887080e-01
7.84981728e-01 6.45815372e-01 3.71843904e-01 3.83411050e-01
-3.08493137e-01 -8.65870893e-01 4.64145631e-01 1.68808377e+00
-5.43432795e-02 -5.37982821e-01 -5.99676967e-01 5.55741310e-01
-1.83593416e+00 -5.44804454e-01 -6.57965362e-01 2.20219755e+00
1.26264238e+00 1.98965222e-01 -6.64254367e-01 9.61989313e-02
8.98230255e-01 3.74936173e-03 -2.74569452e-01 -6.44166172e-01
-1.86863452e-01 2.05161482e-01 5.60625017e-01 5.31006932e-01
-8.16202164e-01 1.21464729e+00 5.96075869e+00 1.04413295e+00
-8.59900117e-01 3.48118067e-01 5.38876891e-01 1.93807155e-01
-7.62214124e-01 1.86178729e-01 -1.17135787e+00 1.07641077e+00
8.43496323e-01 -1.33923933e-01 3.94531459e-01 6.92254543e-01
4.94982541e-01 7.79723525e-02 -8.78741086e-01 7.28886485e-01
3.42987686e-01 -1.20664811e+00 1.07196137e-01 -2.69929022e-01
8.85228753e-01 -1.92938387e-01 -1.18106969e-01 6.70052409e-01
2.08098572e-02 -7.50501513e-01 9.07074690e-01 6.50546968e-01
1.10366035e+00 -6.90174162e-01 8.93720269e-01 4.42741543e-01
-9.20552731e-01 2.29666159e-01 -8.03059816e-01 -2.94011414e-01
3.79349887e-01 7.02530324e-01 -5.33907235e-01 6.69535816e-01
3.58689278e-01 1.07740355e+00 -8.40176761e-01 8.60751271e-01
-9.94112372e-01 9.76597071e-01 2.10485429e-01 -2.42456347e-01
-1.08574480e-01 -1.00337312e-01 3.02555144e-01 1.38685858e+00
5.85229218e-01 4.44846340e-02 -3.65428746e-01 8.10781598e-01
-6.75651789e-01 2.16177434e-01 5.18444739e-02 2.00467616e-01
1.10610211e+00 8.57459307e-01 2.87887119e-02 -3.11830789e-01
-4.11291331e-01 1.45916891e+00 8.00456107e-01 1.39599815e-01
-1.11212063e+00 -7.98075616e-01 6.28340960e-01 -3.00086856e-01
3.74780864e-01 4.88539562e-02 -6.16233945e-01 -1.39547467e+00
6.16908550e-01 -8.36549044e-01 -1.43950358e-01 -7.16491163e-01
-8.89477074e-01 3.70532334e-01 -6.73590362e-01 -1.55783129e+00
5.19219488e-02 -3.23022515e-01 -8.94006133e-01 1.21761715e+00
-1.76683760e+00 -1.04883218e+00 -1.85305446e-01 -5.10182045e-02
4.81821269e-01 -5.73879071e-02 8.04424882e-01 6.03425980e-01
-1.01927662e+00 1.40720570e+00 5.32596886e-01 2.20950276e-01
1.00398326e+00 -1.25390673e+00 9.51471686e-01 1.54274881e+00
-1.43395975e-01 8.12791407e-01 2.91303128e-01 -9.18080389e-01
-9.88913774e-01 -1.65297496e+00 1.94155979e+00 -3.47633749e-01
1.84214443e-01 -5.36172032e-01 -1.19771028e+00 8.21346998e-01
9.21840668e-02 -9.20599103e-02 3.39170426e-01 3.47028137e-03
-3.28444630e-01 -5.74860200e-02 -7.93164790e-01 5.80516636e-01
1.22369003e+00 -4.32234317e-01 -6.81719661e-01 2.15749040e-01
1.32120013e+00 -9.47803438e-01 -4.91003424e-01 2.79684216e-01
2.80405744e-03 -5.24718463e-01 3.82275343e-01 -6.42051756e-01
8.85221124e-01 -5.55321038e-01 -7.31667057e-02 -1.58256412e+00
-4.62231487e-01 -4.90091920e-01 -2.43873462e-01 1.64516485e+00
7.68839002e-01 -3.94478887e-01 4.03690010e-01 5.15086889e-01
-9.51957464e-01 -8.09195220e-01 -9.00746584e-01 -6.19690359e-01
2.38476619e-01 -5.49703360e-01 8.95179629e-01 7.43771136e-01
1.24047749e-01 6.37956858e-02 -7.84517825e-01 7.87742212e-02
2.27678895e-01 -3.23246628e-01 5.45970857e-01 -5.95997691e-01
-1.92964956e-01 -7.64721036e-02 1.00933008e-01 -1.70459604e+00
1.74646989e-01 -1.21028984e+00 3.71438235e-01 -1.39445329e+00
2.51922786e-01 -6.36852682e-01 -2.05505237e-01 6.63866997e-01
-9.80488420e-01 1.20478980e-02 2.49702871e-01 1.06366314e-01
-6.26228034e-01 1.06374836e+00 1.20212793e+00 4.96625807e-03
2.73437984e-02 -1.82164133e-01 -9.70123351e-01 4.01269138e-01
6.57336831e-01 -7.58054912e-01 1.26886830e-01 -1.24809158e+00
3.51864040e-01 -2.69631445e-01 -7.92619810e-02 -5.67042112e-01
2.42703557e-01 -2.58963127e-02 5.81040457e-02 -5.96013248e-01
-2.60493249e-01 -2.51133174e-01 -2.77416557e-01 4.55048412e-01
-5.27659237e-01 3.20717812e-01 -6.19266555e-02 6.18436038e-01
-2.58355498e-01 -5.01115918e-01 5.57398558e-01 1.50573343e-01
-6.36296332e-01 2.12714374e-02 -2.49498963e-01 2.13765576e-01
6.03511870e-01 1.24834202e-01 -6.20111406e-01 1.79045573e-02
-2.21813798e-01 5.30446649e-01 3.48480940e-01 6.37387931e-01
5.93646228e-01 -1.47937799e+00 -9.16610479e-01 3.45539808e-01
4.63880867e-01 8.58944282e-02 3.34175676e-01 8.20667505e-01
-6.30605936e-01 4.45355952e-01 3.28684866e-01 -1.81488708e-01
-1.14989901e+00 1.55905798e-01 2.16833368e-01 -1.91001609e-01
-6.12825334e-01 1.03724146e+00 -2.94351935e-01 -8.70479763e-01
1.20215684e-01 -6.15574121e-01 5.41813672e-02 -4.59536999e-01
7.90690064e-01 2.83069104e-01 4.74023223e-01 -3.97244751e-01
-3.18623006e-01 2.30817199e-01 -5.18377125e-01 3.27867031e-01
1.15347219e+00 -4.46442693e-01 -3.94821435e-01 1.03548683e-01
9.42391276e-01 3.82301986e-01 -1.05258811e+00 -4.33135360e-01
8.19586217e-02 -5.59376717e-01 -1.99117243e-01 -1.10141218e+00
-9.75301445e-01 9.87368107e-01 1.14643663e-01 -7.69740045e-01
1.12357128e+00 -3.24441105e-01 1.01085150e+00 2.94331461e-01
2.90923059e-01 -1.24291122e+00 -3.29176158e-01 1.09224033e+00
7.79012859e-01 -1.09974396e+00 -3.84004802e-01 -6.33947551e-01
-8.99106085e-01 8.21733117e-01 9.73823071e-01 5.28036542e-02
2.33310983e-01 -3.48711088e-02 -3.37313302e-02 4.28965986e-01
-9.82555330e-01 2.25602254e-01 7.16012716e-02 3.70505184e-01
8.78171086e-01 2.11756885e-01 -1.18783426e+00 1.16884029e+00
-1.82305425e-01 -1.06759153e-01 9.48143303e-01 6.27365470e-01
-5.10826647e-01 -1.50847709e+00 1.65826321e-01 4.18269485e-01
-4.79021579e-01 -6.48989320e-01 -9.54178497e-02 4.38635975e-01
2.99847841e-01 8.58704805e-01 1.19999513e-01 -5.34122944e-01
4.10257638e-01 7.94643909e-02 1.32654905e-01 -7.62686551e-01
-6.57215059e-01 -1.36301994e-01 1.10218696e-01 -3.13736945e-01
1.28122801e-02 -7.66054511e-01 -1.58230734e+00 -4.10790235e-01
-5.50534248e-01 2.09054604e-01 5.29940128e-01 9.19168353e-01
6.54736638e-01 7.28639543e-01 5.97880900e-01 -6.81062043e-02
-7.43198991e-01 -1.50574815e+00 -5.38804770e-01 2.59830445e-01
1.06740855e-01 -2.95653880e-01 -9.68661830e-02 6.37927502e-02] | [11.040520668029785, 10.727279663085938] |
923a07a5-1dad-41d7-ba5e-62aee3bb3bda | learning-inter-superpoint-affinity-for-weakly | 2210.05534 | null | https://arxiv.org/abs/2210.05534v1 | https://arxiv.org/pdf/2210.05534v1.pdf | Learning Inter-Superpoint Affinity for Weakly Supervised 3D Instance Segmentation | Due to the few annotated labels of 3D point clouds, how to learn discriminative features of point clouds to segment object instances is a challenging problem. In this paper, we propose a simple yet effective 3D instance segmentation framework that can achieve good performance by annotating only one point for each instance. Specifically, to tackle extremely few labels for instance segmentation, we first oversegment the point cloud into superpoints in an unsupervised manner and extend the point-level annotations to the superpoint level. Then, based on the superpoint graph, we propose an inter-superpoint affinity mining module that considers the semantic and spatial relations to adaptively learn inter-superpoint affinity to generate high-quality pseudo labels via semantic-aware random walk. Finally, we propose a volume-aware instance refinement module to segment high-quality instances by applying volume constraints of objects in clustering on the superpoint graph. Extensive experiments on the ScanNet-v2 and S3DIS datasets demonstrate that our method achieves state-of-the-art performance in the weakly supervised point cloud instance segmentation task, and even outperforms some fully supervised methods. | ['Jin Xie', 'Le Hui', 'Linghua Tang'] | 2022-10-11 | null | null | null | null | ['3d-instance-segmentation-1'] | ['computer-vision'] | [-6.21291548e-02 4.04991806e-01 -4.33114350e-01 -6.44735694e-01
-1.02085674e+00 -5.50699055e-01 3.42802644e-01 3.31927001e-01
1.92293003e-02 9.85414386e-02 -7.14234769e-01 -1.39581844e-01
-3.04660052e-01 -8.63733351e-01 -1.02134252e+00 -4.45006728e-01
-2.52752513e-01 1.30516458e+00 9.20107126e-01 2.61891961e-01
3.80383730e-01 1.00086021e+00 -1.66649079e+00 -5.41250966e-02
1.12402093e+00 9.64786232e-01 4.45446342e-01 1.74877539e-01
-7.97496080e-01 7.77391419e-02 -1.91449970e-01 -1.41656240e-02
6.05095446e-01 1.43674806e-01 -1.04449093e+00 6.09829187e-01
4.95066762e-01 5.28090335e-02 3.31314355e-01 1.05646229e+00
2.54535004e-02 5.64243831e-02 7.93581605e-01 -1.26118076e+00
-1.04314722e-01 3.38745773e-01 -1.02365398e+00 -1.01521583e-02
-8.57584849e-02 -1.01427287e-02 1.09591711e+00 -1.03342855e+00
5.45702517e-01 1.13996434e+00 5.34023821e-01 1.56467736e-01
-1.17953980e+00 -7.11753249e-01 5.10360062e-01 -4.10485901e-02
-1.65056562e+00 1.95327550e-01 1.13243461e+00 -5.15399873e-01
7.49589503e-01 1.94634378e-01 8.13941121e-01 1.84026092e-01
-5.08880436e-01 7.82270849e-01 8.95278990e-01 -3.29552367e-02
4.48687971e-01 -3.07433773e-02 2.67095804e-01 7.42227852e-01
4.39538853e-03 -4.39555347e-01 -1.53999403e-01 -2.98195332e-01
9.95191097e-01 3.00153971e-01 6.46939427e-02 -8.65086377e-01
-1.22434747e+00 6.77614331e-01 8.19339514e-01 1.01063080e-01
-3.54695827e-01 2.12806463e-02 5.84505349e-02 -2.02355668e-01
8.30104828e-01 3.27026337e-01 -7.85929978e-01 3.32134426e-01
-1.16041410e+00 1.32307604e-01 5.79097867e-01 1.48250175e+00
1.48884845e+00 -6.72044814e-01 -2.51246616e-02 8.52032185e-01
4.49760228e-01 4.31884050e-01 -3.63563538e-01 -1.11586452e+00
4.65435833e-01 1.26597619e+00 6.05145171e-02 -7.99655318e-01
-4.79006827e-01 -5.89653254e-01 -5.50659835e-01 2.15353966e-01
7.49483854e-02 4.92554873e-01 -1.37126553e+00 9.38560605e-01
1.03568995e+00 8.02511632e-01 -5.35909355e-01 1.12467289e+00
7.36221492e-01 5.97106516e-01 1.29394799e-01 -7.95488879e-02
1.15896142e+00 -8.92158151e-01 7.05593359e-03 -2.40391672e-01
6.48329794e-01 -4.11439985e-01 1.12647152e+00 1.25394031e-01
-9.35092390e-01 -6.02058232e-01 -8.71873736e-01 1.06347622e-02
-8.67070630e-02 -7.80161005e-03 5.57798564e-01 1.97292477e-01
-6.45049691e-01 6.10548615e-01 -1.07819128e+00 -8.26889277e-02
8.62452626e-01 5.24538815e-01 -7.89075494e-02 -1.61517069e-01
-5.73959053e-01 2.64098257e-01 6.00853622e-01 -1.81688562e-01
-7.12542772e-01 -1.11593914e+00 -7.14758754e-01 -1.91714734e-01
7.04965413e-01 -6.91816092e-01 8.71815205e-01 -5.19920707e-01
-1.08077896e+00 1.20251083e+00 -2.05797613e-01 -4.49301787e-02
3.60105574e-01 4.37575066e-03 1.31617472e-01 4.15518850e-01
5.11836767e-01 8.97581518e-01 7.49446094e-01 -1.98474014e+00
-1.11685371e+00 -8.50926816e-01 -2.00736034e-03 2.44863123e-01
2.39555195e-01 -5.63326240e-01 -1.16616392e+00 -1.82704315e-01
9.81210291e-01 -1.00976896e+00 -6.36117518e-01 -1.04328632e-01
-6.88910782e-01 -7.77023494e-01 1.00703669e+00 -2.93954965e-02
7.31452227e-01 -2.10541916e+00 1.90826401e-01 8.00397635e-01
3.95383120e-01 -2.03929827e-01 1.00656338e-01 -8.81305933e-02
2.14671016e-01 2.22090214e-01 -7.48780489e-01 -5.75547576e-01
1.97250005e-02 5.54007947e-01 -2.12431978e-02 4.51757282e-01
4.02447015e-01 8.27760458e-01 -1.08039677e+00 -8.75845075e-01
5.26214600e-01 2.88927913e-01 -6.04134619e-01 2.63039261e-01
-6.54128969e-01 6.08666956e-01 -9.26336169e-01 9.28852022e-01
1.04422581e+00 -6.18924379e-01 -3.14971864e-01 -2.47337207e-01
-2.15287596e-01 6.48560002e-02 -1.20516682e+00 2.12499928e+00
-3.13977629e-01 -1.37004182e-01 -5.04975468e-02 -8.57132494e-01
1.01867461e+00 -1.91337034e-01 9.22279775e-01 -1.55021787e-01
-2.19728023e-01 2.88371861e-01 -5.87042630e-01 -3.64598334e-01
1.75329477e-01 -5.30109145e-02 7.25768879e-02 1.83968201e-01
1.00477263e-02 -8.31487954e-01 -1.42077625e-01 2.78268278e-01
8.94955099e-01 1.93328500e-01 -2.29116380e-01 -2.01613024e-01
5.51521719e-01 4.89352226e-01 6.18775964e-01 5.89077413e-01
1.99585110e-01 1.01550460e+00 2.39966854e-01 -3.35283488e-01
-1.02234638e+00 -1.19600737e+00 -4.44366038e-01 7.46305883e-01
8.51082206e-01 -3.29700321e-01 -8.59046280e-01 -1.06551886e+00
2.22664848e-01 5.67051649e-01 -4.52255100e-01 2.40418404e-01
-4.07973826e-01 -5.51187515e-01 -9.30367112e-02 5.03135204e-01
3.43358785e-01 -6.33734405e-01 -2.52217233e-01 4.11634147e-02
2.85536684e-02 -1.31302798e+00 -2.54745543e-01 2.58833438e-01
-1.16593850e+00 -1.05716324e+00 -4.28327948e-01 -8.83391023e-01
1.07514775e+00 4.23824459e-01 1.26713169e+00 3.63971382e-01
6.11178838e-02 2.47635052e-01 -5.01703858e-01 -2.26172030e-01
1.78550765e-01 4.49586987e-01 -3.01897824e-01 -1.29368454e-01
5.87352395e-01 -5.25349498e-01 -6.53591931e-01 5.57521641e-01
-7.41744876e-01 -7.90720154e-03 6.81439757e-01 3.01973015e-01
1.59448719e+00 2.39232540e-01 1.83604676e-02 -1.23434794e+00
-2.55304784e-01 -5.52001297e-01 -7.08270133e-01 7.00347871e-02
-4.56463248e-01 -1.05140001e-01 9.54982862e-02 -1.10151395e-01
-8.01474094e-01 6.28997684e-01 -2.13040575e-01 -9.11869347e-01
-5.03849089e-01 2.27285758e-01 -4.79047984e-01 -3.35372776e-01
2.88752884e-01 -5.52204885e-02 -4.42494273e-01 -6.30802870e-01
4.92216319e-01 5.64707577e-01 3.93602759e-01 -7.49378920e-01
1.33162069e+00 8.31413627e-01 1.55812189e-01 -7.70070732e-01
-1.17424500e+00 -1.05665052e+00 -1.22697663e+00 -2.27797747e-01
1.10589993e+00 -1.00219107e+00 -4.60976064e-01 -2.52089947e-02
-1.14350045e+00 -3.49717081e-01 -4.19806510e-01 3.29298764e-01
-5.16826928e-01 2.01207280e-01 -3.33222121e-01 -5.82814157e-01
1.23876128e-02 -1.22770381e+00 1.78884220e+00 3.69867757e-02
2.50869811e-01 -7.17191696e-01 -4.20494266e-02 5.95457494e-01
-4.50463891e-01 2.85111934e-01 8.64758432e-01 -6.45688057e-01
-1.30610478e+00 4.27721854e-04 -4.15640682e-01 1.87835276e-01
2.90578790e-02 -1.05707645e-01 -7.72456884e-01 8.15417734e-04
-5.58222979e-02 4.28094976e-02 7.23772585e-01 2.60158032e-01
1.73832226e+00 1.29525587e-01 -7.61094034e-01 9.65511560e-01
1.34128392e+00 -2.94188440e-01 3.36671114e-01 1.02654718e-01
1.33162379e+00 6.18598700e-01 1.05470192e+00 3.84843022e-01
6.37418866e-01 4.94707793e-01 8.12435687e-01 -2.21689358e-01
1.47530049e-01 -3.32946062e-01 -5.65376878e-01 7.49602973e-01
-2.60776192e-01 1.28615618e-01 -1.12337172e+00 6.19964540e-01
-1.82907760e+00 -3.97792876e-01 -6.58883810e-01 1.99150312e+00
5.34027696e-01 4.28869545e-01 1.44509077e-01 9.01470054e-03
7.99411535e-01 -2.59575639e-02 -6.92685366e-01 4.34461147e-01
3.00926387e-01 2.00608209e-01 8.05062473e-01 4.34316188e-01
-1.09957147e+00 1.20412183e+00 4.88455820e+00 9.51148331e-01
-6.56768680e-01 2.78171152e-01 6.62436008e-01 1.25007316e-01
-4.10867393e-01 1.94303095e-01 -8.74877632e-01 4.77338433e-01
2.30395496e-01 3.90501827e-01 3.85588706e-02 1.14125502e+00
1.31312460e-01 1.28756523e-01 -1.06387353e+00 1.04379237e+00
-2.02549279e-01 -1.22196329e+00 1.37873366e-01 1.89344138e-01
9.19696689e-01 2.92871147e-01 -2.66889066e-01 -2.89535727e-02
4.16993722e-02 -7.16782212e-01 6.19342148e-01 2.82390296e-01
6.83309793e-01 -1.03583467e+00 4.54541326e-01 6.67125702e-01
-1.48329341e+00 2.42783278e-01 -6.02218091e-01 4.09766883e-01
1.64588168e-01 1.00662971e+00 -8.77623677e-01 6.91651642e-01
1.03003931e+00 7.91239023e-01 -4.77657378e-01 1.20154881e+00
-3.54462594e-01 5.43836117e-01 -6.45016670e-01 3.29447031e-01
4.27023709e-01 -5.52496612e-01 5.56815565e-01 9.89872813e-01
2.59911329e-01 5.83262086e-01 6.39065027e-01 1.11452711e+00
-6.60093874e-02 5.29736243e-02 -3.08584213e-01 5.33781469e-01
7.16115475e-01 1.46695518e+00 -1.30850840e+00 -2.40258425e-01
-2.72654653e-01 8.00605118e-01 4.11379576e-01 1.63930237e-01
-7.64942944e-01 -8.81504193e-02 6.41288877e-01 5.36769688e-01
4.33043957e-01 -4.43623841e-01 -6.04845405e-01 -7.78398871e-01
1.12913832e-01 -7.97936320e-02 2.48289853e-01 -7.16947913e-01
-1.38791287e+00 3.35235626e-01 1.59452632e-01 -1.45923328e+00
1.96133465e-01 -2.58387566e-01 -5.65748632e-01 6.11713707e-01
-1.74404633e+00 -1.33800042e+00 -6.02565765e-01 5.50303638e-01
6.60847068e-01 4.01411861e-01 7.35322908e-02 2.81591564e-01
-2.03464136e-01 2.79078614e-02 -3.38439316e-01 1.71791168e-03
1.57715663e-01 -1.46654189e+00 4.43681180e-01 3.62522185e-01
2.49425471e-01 3.79524946e-01 2.78736860e-01 -8.65018368e-01
-1.09520853e+00 -1.56754553e+00 3.82059425e-01 -7.12499917e-01
3.30325514e-01 -5.25255680e-01 -1.36046100e+00 5.57330132e-01
-4.68063504e-01 3.69966537e-01 3.90210003e-01 4.19138595e-02
-7.64757395e-02 -7.81628564e-02 -1.20443821e+00 1.48591265e-01
1.65121281e+00 -3.47009420e-01 -7.08925784e-01 7.54009724e-01
1.17021585e+00 -6.91462338e-01 -1.04831123e+00 8.19134235e-01
-2.89534301e-01 -7.41961718e-01 1.28889954e+00 -2.83483416e-01
2.64782786e-01 -8.69625330e-01 1.62898436e-01 -9.85149205e-01
-3.93875122e-01 -1.77665025e-01 4.11044732e-02 1.36237395e+00
5.12611151e-01 -2.79417694e-01 1.29140484e+00 4.93569016e-01
-6.33216679e-01 -7.38041818e-01 -9.84649479e-01 -6.91416264e-01
-1.50663018e-01 -7.13496745e-01 9.47551012e-01 1.08859181e+00
-6.19474590e-01 1.77216917e-01 3.96111786e-01 9.24206913e-01
1.16413701e+00 5.52706897e-01 9.94772136e-01 -1.64158249e+00
1.65420622e-02 -1.40267760e-01 -5.81388354e-01 -1.42129433e+00
4.05308753e-01 -1.14057291e+00 2.04161167e-01 -1.64421690e+00
8.16325545e-02 -1.25607336e+00 -2.02837095e-01 3.84011984e-01
-2.61343122e-01 4.09139395e-01 -9.83089581e-02 5.79808712e-01
-8.97233605e-01 4.00575429e-01 1.27813637e+00 -2.59859890e-01
-4.49722826e-01 2.23529011e-01 -2.56987393e-01 7.83401191e-01
4.50304210e-01 -6.09233499e-01 -4.42346692e-01 -5.73094487e-01
1.20451264e-02 -1.67086959e-01 4.26694155e-01 -1.01894760e+00
3.88269186e-01 -2.96983570e-01 3.78041685e-01 -1.33626139e+00
3.37898165e-01 -1.15255940e+00 7.78560266e-02 -8.16610381e-02
-5.71457148e-02 -4.64426637e-01 -1.16088338e-01 7.39026785e-01
-2.37783436e-02 -1.99972600e-01 7.97709346e-01 -3.33430231e-01
-7.36888111e-01 9.67064559e-01 5.49272060e-01 2.34038867e-02
1.37148643e+00 -3.89429569e-01 2.38219157e-01 4.75684732e-01
-6.79370940e-01 8.25087190e-01 9.29990470e-01 3.25014740e-01
7.62125015e-01 -1.06865764e+00 -4.52524334e-01 1.81883037e-01
4.29018408e-01 1.29760683e+00 2.95980841e-01 6.11130297e-01
-5.55544496e-01 4.38838638e-02 3.54377747e-01 -1.59856462e+00
-9.93424058e-01 4.74703610e-01 1.58312500e-01 2.72238165e-01
-9.02323067e-01 1.00427008e+00 4.12282139e-01 -7.19417274e-01
6.26358762e-02 -3.90312701e-01 -8.90169367e-02 -1.69411495e-01
-6.61057085e-02 2.09012151e-01 2.22225145e-01 -6.47109807e-01
-6.29240274e-01 1.17638505e+00 4.72871587e-02 2.97207713e-01
1.49140358e+00 -7.95003250e-02 -2.62051642e-01 5.03202021e-01
1.14805961e+00 -1.86504513e-01 -1.47233260e+00 -1.28492489e-01
2.17442244e-01 -7.18979239e-01 1.53791040e-01 -3.42557073e-01
-1.36049390e+00 6.81703329e-01 4.06705379e-01 2.86823094e-01
8.10462832e-01 9.10429955e-01 6.94489598e-01 2.21039191e-01
5.46835780e-01 -9.63867188e-01 -1.47381425e-01 2.20590547e-01
4.15759176e-01 -1.34563947e+00 9.79803279e-02 -1.14278221e+00
-4.03881162e-01 7.79763579e-01 8.41582298e-01 -3.43046874e-01
8.31935227e-01 3.03762928e-02 -1.10140622e-01 -7.25035846e-01
-2.54077137e-01 -3.60418886e-01 5.09500682e-01 5.60562909e-01
-1.29476160e-01 6.87420517e-02 5.55682629e-02 2.49961749e-01
-1.09290600e-01 -3.15997720e-01 -8.42638388e-02 7.47774363e-01
-6.87784195e-01 -9.30796504e-01 -5.66410661e-01 7.47580409e-01
-1.08264270e-03 3.54306459e-01 -3.87281805e-01 7.81352520e-01
3.78492266e-01 6.35244191e-01 5.56545079e-01 -4.29671824e-01
5.78172266e-01 -2.31759742e-01 2.77005494e-01 -1.09290671e+00
-2.64228970e-01 2.04782665e-01 -4.31834131e-01 -6.75473034e-01
-6.47601485e-01 -6.54024959e-01 -1.95907581e+00 1.39689624e-01
-4.85950649e-01 4.67486680e-01 7.48612761e-01 1.08355916e+00
4.73140270e-01 3.04295540e-01 8.78673851e-01 -1.32581317e+00
2.91511472e-02 -5.47276199e-01 -7.38479972e-01 5.87134898e-01
5.38117290e-02 -8.06295633e-01 -5.75771809e-01 -1.30983576e-01] | [8.021469116210938, -3.159083604812622] |
8c6a9aa0-dece-44f3-8555-52cb499a8ac5 | sprt-based-efficient-best-arm-identification | 2207.11158 | null | https://arxiv.org/abs/2207.11158v3 | https://arxiv.org/pdf/2207.11158v3.pdf | SPRT-based Efficient Best Arm Identification in Stochastic Bandits | This paper investigates the best arm identification (BAI) problem in stochastic multi-armed bandits in the fixed confidence setting. The general class of the exponential family of bandits is considered. The existing algorithms for the exponential family of bandits face computational challenges. To mitigate these challenges, the BAI problem is viewed and analyzed as a sequential composite hypothesis testing task, and a framework is proposed that adopts the likelihood ratio-based tests known to be effective for sequential testing. Based on this test statistic, a BAI algorithm is designed that leverages the canonical sequential probability ratio tests for arm selection and is amenable to tractable analysis for the exponential family of bandits. This algorithm has two key features: (1) its sample complexity is asymptotically optimal, and (2) it is guaranteed to be $\delta-$PAC. Existing efficient approaches focus on the Gaussian setting and require Thompson sampling for the arm deemed the best and the challenger arm. Additionally, this paper analytically quantifies the computational expense of identifying the challenger in an existing approach. Finally, numerical experiments are provided to support the analysis. | ['Ali Tajer', 'Arpan Mukherjee'] | 2022-07-22 | null | null | null | null | ['thompson-sampling'] | ['methodology'] | [ 3.25691968e-01 -2.98448980e-01 -9.58813250e-01 -1.35832027e-01
-1.36052549e+00 -9.30933297e-01 1.43274188e-01 -1.19095884e-01
-1.33288339e-01 9.64355767e-01 -2.72622287e-01 -1.02427804e+00
-8.50601435e-01 -6.67464435e-01 -7.49659419e-01 -8.42489719e-01
-2.38198578e-01 7.75025725e-01 -1.76010579e-01 2.27008134e-01
2.64874399e-01 3.12463135e-01 -1.02456963e+00 -2.25472122e-01
9.29210246e-01 1.67602432e+00 -1.73141167e-01 7.11914062e-01
3.97263408e-01 3.98573637e-01 -3.12376618e-01 -6.20923281e-01
7.41256952e-01 -8.45907152e-01 -6.30610645e-01 2.96522737e-01
2.30442062e-01 -6.49247169e-01 -7.89619535e-02 1.17737591e+00
4.91414934e-01 1.59090951e-01 4.93005574e-01 -1.34312701e+00
-3.69612753e-01 9.17335629e-01 -9.18621540e-01 4.84498858e-01
1.67226747e-01 -2.00773641e-01 1.52923918e+00 -4.74002391e-01
-2.98652083e-01 1.31658494e+00 4.32661921e-01 -1.56416029e-01
-1.28999484e+00 -9.09875393e-01 1.75754279e-01 -1.92898829e-02
-1.23707116e+00 -3.02221388e-01 4.81544048e-01 -3.25780749e-01
1.56595111e-01 6.88090861e-01 8.34050119e-01 8.00085604e-01
-2.71763951e-01 1.24305856e+00 1.36629784e+00 -7.21236765e-01
5.59912682e-01 8.29545185e-02 5.57550550e-01 3.24132621e-01
6.56142414e-01 4.24396753e-01 -3.52365613e-01 -7.97298193e-01
8.67562413e-01 1.71374660e-02 -1.28896907e-01 -3.77049565e-01
-7.40587652e-01 1.27959168e+00 -1.23597220e-01 -2.90509313e-01
-5.86875975e-01 1.17035769e-01 1.26317173e-01 3.23889464e-01
6.67385221e-01 1.09998710e-01 -3.48781884e-01 -1.04766609e-02
-1.09146249e+00 5.09698987e-01 9.52172101e-01 1.01386535e+00
2.64557004e-01 1.52140250e-02 -6.96637213e-01 6.64514840e-01
3.49370420e-01 7.93204546e-01 -1.09355822e-02 -6.86966002e-01
6.86469078e-01 -5.35059944e-02 7.40250945e-01 -6.24876142e-01
2.05154456e-02 -9.15637136e-01 -4.69430655e-01 -9.50073972e-02
4.40869600e-01 -1.67199329e-01 -6.52533948e-01 1.55535150e+00
5.65314829e-01 2.36993492e-01 -2.24624857e-01 6.30825102e-01
-1.60180151e-01 3.60321999e-01 -3.46037775e-01 -6.99027300e-01
1.38928747e+00 -9.82745290e-01 -7.14834273e-01 -3.61125648e-01
1.55270547e-01 -5.72972357e-01 6.21782780e-01 4.56606120e-01
-1.38012946e+00 4.67461087e-02 -1.16946638e+00 9.15664017e-01
1.97043270e-01 -1.01077631e-01 8.50556374e-01 1.40780771e+00
-8.44787836e-01 1.28304631e-01 -6.83559239e-01 9.05921385e-02
5.80546498e-01 4.36762542e-01 4.15565878e-01 -1.48608506e-01
-8.18265378e-01 3.57507706e-01 3.87816727e-01 7.29020685e-02
-8.36173832e-01 -3.86092335e-01 -5.72712362e-01 2.07002893e-01
6.09873593e-01 -6.96935773e-01 1.77091742e+00 -1.08698726e+00
-1.69119656e+00 2.18937322e-01 -3.34715903e-01 -8.53748500e-01
7.24349916e-01 -1.85492322e-01 -4.85216267e-02 3.15208405e-01
4.13102180e-01 -4.75600004e-01 9.52155411e-01 -8.60518992e-01
-1.23009169e+00 -5.82774639e-01 8.22471976e-02 -4.46883552e-02
-2.67506301e-01 2.31855243e-01 -2.48502657e-01 -9.23029006e-01
2.73579597e-01 -1.13839829e+00 -4.15418297e-01 -8.07709217e-01
-6.46491289e-01 -4.34665084e-02 1.82840630e-01 -4.55438554e-01
1.58965182e+00 -1.95743406e+00 -3.68627101e-01 9.31820869e-01
-2.23884135e-01 -2.04024926e-01 1.01025462e-01 3.12549949e-01
-2.73470394e-02 1.03870302e-01 7.50000626e-02 -5.79276159e-02
1.68854252e-01 -2.88371086e-01 -6.97177649e-01 8.92601669e-01
-2.73454994e-01 7.93755293e-01 -5.77886224e-01 -2.89265066e-01
-1.30752757e-01 -3.78029108e-01 -5.65295219e-01 1.70237467e-01
-5.68613820e-02 2.25283563e-01 -8.82267237e-01 9.11584318e-01
7.84430265e-01 -6.66290581e-01 2.41009146e-01 1.88798741e-01
-1.47238970e-01 4.90410589e-02 -1.39361703e+00 6.68269753e-01
-7.52639249e-02 -4.43583801e-02 1.45542592e-01 -1.44437194e+00
5.44516206e-01 3.22358072e-01 3.28101933e-01 -4.81837720e-01
2.54633456e-01 5.48895001e-01 -1.89239066e-02 -1.12100035e-01
9.21797752e-02 -1.60452291e-01 -1.13559425e-01 7.72683918e-01
-3.69217604e-01 1.53017744e-01 3.06183010e-01 -1.15372948e-01
9.80421364e-01 -3.43122363e-01 7.21438944e-01 -4.03770834e-01
9.09032598e-02 -1.75455675e-01 4.84465271e-01 1.65478611e+00
-1.84838519e-01 1.32872477e-01 3.23979437e-01 -2.04572842e-01
-7.26450682e-01 -1.05923581e+00 -2.68007576e-01 1.22549379e+00
9.94269997e-02 3.88544619e-01 -4.13551211e-01 -3.11344862e-01
3.46284151e-01 8.53365242e-01 -6.66657865e-01 2.65172571e-01
-1.45745069e-01 -9.50144708e-01 1.55867159e-01 5.12484550e-01
4.25398529e-01 -2.36185819e-01 -6.34108245e-01 4.65719908e-01
-3.70608605e-02 -8.98691177e-01 -5.97127855e-01 2.39948988e-01
-8.57453763e-01 -1.10724199e+00 -9.20600653e-01 -5.12426972e-01
1.80022135e-01 7.70838916e-01 8.24275494e-01 -4.52960640e-01
2.13780984e-01 6.47947192e-01 -4.49394971e-01 -7.60962784e-01
-5.44021614e-02 -2.14470565e-01 -2.48708855e-02 6.14743471e-01
2.49950215e-01 -3.48259747e-01 -7.90748894e-01 6.23582900e-01
-7.60133803e-01 -3.76439303e-01 5.81184208e-01 1.12569761e+00
6.53652132e-01 9.44573358e-02 7.77005553e-01 -5.21488428e-01
6.69681489e-01 -7.23130405e-01 -1.11860454e+00 5.18694162e-01
-6.02462828e-01 -2.76966274e-01 1.45142779e-01 -6.20460391e-01
-7.92195976e-01 -2.83974886e-01 5.31279981e-01 -1.56077519e-01
4.28968728e-01 8.38508666e-01 1.54606193e-01 -1.15007125e-01
1.75154120e-01 2.91206509e-01 3.80275548e-02 -3.58448803e-01
-9.21310559e-02 7.59724498e-01 1.87363848e-01 -8.24663281e-01
5.07269084e-01 6.78947151e-01 2.52013188e-02 -4.83782619e-01
-1.36235142e+00 -6.01587594e-01 3.30590039e-01 2.16030832e-02
2.37648636e-01 -8.13040733e-01 -1.11685002e+00 6.10774830e-02
-6.89220786e-01 -2.11999819e-01 -7.13798329e-02 9.35145676e-01
-1.05226707e+00 3.25368434e-01 -4.00827497e-01 -1.81381488e+00
-3.94477874e-01 -1.19933748e+00 7.47430503e-01 2.66279340e-01
-9.80548412e-02 -6.54374957e-01 -6.02492020e-02 4.04086173e-01
1.92869619e-01 1.03398934e-01 9.34476376e-01 -1.06789005e+00
-6.01102412e-01 -7.06089079e-01 -2.09709868e-01 1.38072371e-01
1.31753702e-02 -3.39860737e-01 -3.85893553e-01 -6.47317827e-01
2.25407034e-01 -2.95294017e-01 5.27553201e-01 1.21955156e+00
1.29165089e+00 -6.32893324e-01 -4.77767050e-01 3.44902456e-01
1.21301746e+00 6.81040049e-01 2.70701110e-01 7.68675804e-01
-3.14504743e-01 3.07221293e-01 9.79826272e-01 1.01884115e+00
1.22868836e-01 6.38296306e-01 4.66427326e-01 2.30938911e-01
7.45756507e-01 -4.41626832e-02 1.33203015e-01 2.37425998e-01
1.48999408e-01 -3.95500839e-01 -6.57635808e-01 7.11232126e-01
-2.13577557e+00 -1.06962991e+00 2.12923601e-01 2.74044919e+00
7.84957826e-01 1.56531453e-01 6.75239086e-01 2.29135111e-01
1.09222126e+00 -2.30742753e-01 -7.81230450e-01 -1.34373680e-01
-3.28925490e-01 4.35010284e-01 9.16025281e-01 2.94145972e-01
-1.29640782e+00 3.48013341e-01 7.05369186e+00 1.18038297e+00
-4.04238939e-01 2.19074577e-01 1.09510994e+00 -2.90889829e-01
8.16462412e-02 1.01384185e-01 -1.12316179e+00 5.45463562e-01
1.00785351e+00 -5.56342363e-01 3.96067083e-01 9.51529324e-01
3.25339228e-01 -2.18092740e-01 -1.12483537e+00 1.12378085e+00
-2.28485212e-01 -1.15319312e+00 -1.90246031e-01 3.76888216e-01
8.46284091e-01 -1.74226075e-01 7.41082430e-01 8.55757520e-02
8.38324189e-01 -8.68590891e-01 1.09356928e+00 1.25932336e-01
6.57547474e-01 -1.03105807e+00 6.46130025e-01 1.95532575e-01
-7.94684410e-01 -7.54213870e-01 -6.85722977e-02 -7.35791549e-02
1.97221145e-01 6.54014528e-01 -6.08911216e-01 5.41332364e-01
7.18225896e-01 7.71798342e-02 6.01610765e-02 1.71050417e+00
1.17734909e-01 1.04421306e+00 -6.18186057e-01 -1.80350453e-01
5.39257050e-01 -5.18429816e-01 5.99316001e-01 8.19992065e-01
8.75638664e-01 -2.07541678e-02 5.45052230e-01 3.39723766e-01
2.27722615e-01 2.09283873e-01 -9.71536487e-02 -1.39543951e-01
8.08316588e-01 5.34997225e-01 -7.47090220e-01 -3.65579277e-01
-4.16061282e-01 4.42656010e-01 -6.53417110e-02 4.29626614e-01
-9.30428863e-01 -6.72816932e-02 4.23766881e-01 -1.93765193e-01
8.75497639e-01 1.98662430e-01 -5.37718117e-01 -6.38953388e-01
1.25769511e-01 -1.20414090e+00 9.03407872e-01 -9.46860984e-02
-1.39589596e+00 7.45494440e-02 2.26366013e-01 -1.15944970e+00
-2.53878623e-01 -4.69125539e-01 -4.23615038e-01 7.55562782e-01
-1.26528609e+00 -8.24675381e-01 4.12001789e-01 4.75096762e-01
4.65275198e-01 -2.17863470e-01 5.56646764e-01 8.39692876e-02
-7.60309696e-01 1.01517677e+00 8.03416193e-01 -2.88239658e-01
8.42099264e-02 -1.01355255e+00 1.01037435e-02 8.06813300e-01
-6.19684942e-02 5.22127509e-01 9.62723076e-01 -5.82373977e-01
-1.42827141e+00 -8.93138468e-01 1.39507353e-01 -5.13109863e-02
1.07281470e+00 1.08115807e-01 -9.24690738e-02 8.72916043e-01
-2.54426986e-01 -2.13688895e-01 1.10372865e+00 2.46058732e-01
-3.72904807e-01 -1.82531744e-01 -9.96373117e-01 6.21646941e-01
7.75125444e-01 -6.58419132e-02 -1.82250798e-01 6.92826211e-01
4.04949725e-01 -1.90065548e-01 -6.09521866e-01 4.17356968e-01
9.60675955e-01 -8.39393377e-01 9.39679623e-01 -8.90899897e-01
-7.75495023e-02 1.93100557e-01 -5.02255440e-01 -9.75697160e-01
-4.07248050e-01 -1.16910970e+00 -4.44912702e-01 8.45051885e-01
5.59778154e-01 -1.04449666e+00 7.43562460e-01 3.77691776e-01
3.95554125e-01 -6.34949267e-01 -1.42497730e+00 -1.27601218e+00
-3.01739760e-02 -4.97756124e-01 6.85005367e-01 5.23729265e-01
-1.66727938e-02 -4.00476530e-02 -6.39371037e-01 3.78696263e-01
1.02418149e+00 6.51131272e-01 7.07717240e-01 -1.11505795e+00
-1.05170405e+00 -6.11081779e-01 -8.75916984e-03 -1.50292492e+00
-8.78184512e-02 -4.38985854e-01 -3.46672870e-02 -9.81685579e-01
5.43275535e-01 -5.08923054e-01 -6.70210481e-01 7.65733868e-02
-1.48592040e-01 -4.52325307e-02 -1.56215385e-01 4.86046858e-02
-6.19872510e-01 2.91734338e-01 7.13726044e-01 -2.06942290e-01
-2.48485789e-01 1.03586698e+00 -1.05240512e+00 4.27680105e-01
7.95793176e-01 -4.99779969e-01 -3.77765805e-01 8.22843388e-02
3.76764327e-01 6.50677562e-01 3.71066511e-01 -3.80331665e-01
1.10996112e-01 -4.22405183e-01 1.85882882e-03 -9.50590491e-01
2.40503281e-01 -8.15169811e-01 1.33919865e-01 6.16190434e-01
-3.37339401e-01 -4.31829831e-03 -2.43588798e-02 9.81861770e-01
4.00338359e-02 -4.52897638e-01 6.34655237e-01 1.04651175e-01
2.81770110e-01 3.76883805e-01 -2.86410064e-01 3.63005809e-02
1.14067507e+00 -2.81182349e-01 1.25661939e-02 -8.92939746e-01
-4.48596030e-01 3.72088850e-01 -2.53525794e-01 -2.87439346e-01
2.24802896e-01 -1.28526509e+00 -8.76230061e-01 -1.42740220e-01
-3.46124433e-02 -4.66088980e-01 8.04585740e-02 1.18997145e+00
6.34755343e-02 7.60545909e-01 4.91389126e-01 -4.87028450e-01
-1.08674848e+00 5.93719900e-01 4.37857248e-02 -5.94840765e-01
-4.44456898e-02 9.81461227e-01 1.42949477e-01 3.76774192e-01
2.88722008e-01 9.11748335e-02 1.86398923e-01 1.13900244e-01
9.25670326e-01 6.95108235e-01 1.63261639e-03 -1.80808365e-01
-8.09839815e-02 1.95353366e-02 -2.92320818e-01 -4.80392694e-01
1.37556267e+00 -3.27240825e-01 -2.22593784e-01 2.87761152e-01
8.61495137e-01 1.11245506e-01 -1.09689939e+00 -5.84250510e-01
2.24616170e-01 -7.16159761e-01 1.62961155e-01 -5.40288091e-01
-5.93548417e-01 1.69507951e-01 4.23113793e-01 8.90282750e-01
1.01050746e+00 -6.21511862e-02 4.52439547e-01 1.38661698e-01
6.26399934e-01 -9.62520421e-01 -2.28188396e-01 3.48836452e-01
6.27703965e-01 -1.08625782e+00 1.13516442e-01 -2.92568326e-01
-1.29343346e-01 8.77007604e-01 -9.62225124e-02 -3.23790908e-02
6.09676480e-01 -2.41274741e-02 -6.62794828e-01 1.03546232e-01
-3.69785756e-01 -1.98233724e-01 1.29834041e-01 1.19490117e-01
-2.98032742e-02 4.29541439e-01 -5.87807715e-01 1.01415110e+00
-3.14385772e-01 -1.37120932e-01 2.55336672e-01 1.01353848e+00
-3.86528194e-01 -7.28727162e-01 -8.28649640e-01 9.15938556e-01
-9.65889871e-01 -2.19313070e-01 2.20067389e-02 5.01509964e-01
-8.75150979e-01 1.42342830e+00 1.59442723e-02 1.01389341e-01
1.56610981e-01 -3.19079868e-02 4.95638490e-01 -2.76262969e-01
-2.86106803e-02 5.74301898e-01 5.15552834e-02 -2.88638860e-01
-3.65984857e-01 -8.88222635e-01 -2.11894512e-01 -1.39309034e-01
-8.38749826e-01 5.68484604e-01 4.68001276e-01 9.92712021e-01
6.52780086e-02 7.85629079e-02 1.09503579e+00 -7.18014419e-01
-1.66670036e+00 -7.89530516e-01 -1.00931478e+00 -1.89745933e-01
4.30518448e-01 -8.83403838e-01 -6.14075363e-01 -6.31045520e-01] | [4.557054042816162, 3.309588670730591] |
42cf4dd9-ab0b-4925-bf51-91d5b670a490 | which-clustering-do-you-want-inducing-your | 1401.5389 | null | http://arxiv.org/abs/1401.5389v1 | http://arxiv.org/pdf/1401.5389v1.pdf | Which Clustering Do You Want? Inducing Your Ideal Clustering with Minimal Feedback | While traditional research on text clustering has largely focused on grouping
documents by topic, it is conceivable that a user may want to cluster documents
along other dimensions, such as the authors mood, gender, age, or sentiment.
Without knowing the users intention, a clustering algorithm will only group
documents along the most prominent dimension, which may not be the one the user
desires. To address the problem of clustering documents along the user-desired
dimension, previous work has focused on learning a similarity metric from data
manually annotated with the users intention or having a human construct a
feature space in an interactive manner during the clustering process. With the
goal of reducing reliance on human knowledge for fine-tuning the similarity
function or selecting the relevant features required by these approaches, we
propose a novel active clustering algorithm, which allows a user to easily
select the dimension along which she wants to cluster the documents by
inspecting only a small number of words. We demonstrate the viability of our
algorithm on a variety of commonly-used sentiment datasets. | ['Sajib Dasgupta', 'Vincent Ng'] | 2014-01-16 | null | null | null | null | ['text-clustering'] | ['natural-language-processing'] | [ 6.77251071e-02 -7.14885518e-02 -1.45532459e-01 -6.18570805e-01
-5.60187459e-01 -8.06736887e-01 4.97705281e-01 8.93439174e-01
-6.15186155e-01 -7.34144002e-02 3.35439652e-01 -4.57612634e-01
-2.46217951e-01 -7.50595689e-01 2.03271642e-01 -6.44361734e-01
1.52022168e-01 6.99620187e-01 -4.10346836e-02 6.88324869e-02
8.04548979e-01 2.10807800e-01 -1.64432764e+00 1.29590869e-01
9.26740587e-01 6.64245307e-01 1.52904794e-01 4.12198067e-01
-4.71482873e-01 3.40578228e-01 -5.32111764e-01 -2.33832926e-01
1.87467501e-01 -5.26261508e-01 -7.06341088e-01 4.66598362e-01
1.66197063e-03 -6.80100173e-02 5.60881734e-01 9.99601722e-01
1.88323915e-01 3.79956633e-01 8.89276445e-01 -1.09755158e+00
-2.03004003e-01 7.66036749e-01 -4.99124229e-01 -1.04523160e-01
3.49891216e-01 -2.68878758e-01 1.35043240e+00 -6.66745245e-01
5.52132607e-01 9.99266982e-01 3.98107380e-01 1.71982944e-01
-1.34734130e+00 -4.79364932e-01 3.80130053e-01 -1.34959802e-01
-1.38026488e+00 -3.06842655e-01 1.06722867e+00 -7.43916988e-01
4.06575203e-01 5.15568137e-01 3.61424893e-01 4.76031661e-01
-2.47170642e-01 4.82749939e-01 7.53663182e-01 -5.98079979e-01
6.96220398e-01 6.33598685e-01 5.41098237e-01 3.90512228e-01
2.25345001e-01 -6.90993071e-01 -2.99637705e-01 -6.06426477e-01
6.59709275e-02 1.48041755e-01 -1.08093031e-01 -8.11292291e-01
-1.18480825e+00 1.20404255e+00 2.59880535e-03 3.65631253e-01
-2.89582014e-01 -4.85141188e-01 3.93304706e-01 2.71292090e-01
5.29641926e-01 1.01660252e+00 -5.68782628e-01 -1.56188682e-01
-1.06566477e+00 7.33052790e-02 9.93322134e-01 6.30180180e-01
1.08453631e+00 -6.69283926e-01 2.40819931e-01 7.30614066e-01
4.72594976e-01 -6.31585792e-02 5.93549013e-01 -7.69664586e-01
1.69320256e-01 1.14564514e+00 3.95176172e-01 -1.17053354e+00
-4.58470702e-01 -9.94124450e-03 -3.01522404e-01 2.56355554e-02
5.18377841e-01 -3.22841108e-01 -6.59215689e-01 1.40154326e+00
4.86229539e-01 -6.92624271e-01 3.43901701e-02 7.39051342e-01
4.43922132e-01 4.07471240e-01 -2.47010328e-02 -5.43836713e-01
1.11458278e+00 -4.72886503e-01 -4.68148082e-01 -1.51865277e-02
8.15285921e-01 -8.01258087e-01 1.27694499e+00 4.82311070e-01
-6.93832695e-01 -3.34979296e-01 -7.41512656e-01 1.37026221e-01
-4.76329803e-01 -6.31344330e-04 5.92683256e-01 7.54365981e-01
-8.43041360e-01 3.96748960e-01 -8.91079485e-01 -7.68158853e-01
-2.13538796e-01 5.13448536e-01 -1.29276261e-01 2.68350065e-01
-9.07696962e-01 5.48739791e-01 2.35153526e-01 -3.94909263e-01
1.92465372e-02 -2.96978503e-01 -5.94296396e-01 1.80190921e-01
4.34814751e-01 -4.25774932e-01 1.05762196e+00 -1.25527692e+00
-1.20969915e+00 6.60332322e-01 -1.97737008e-01 5.09937368e-02
2.66561121e-01 3.25610414e-02 -2.08925396e-01 1.00736450e-02
2.44797751e-01 4.55564201e-01 7.70733058e-01 -1.35872936e+00
-7.18268156e-01 -5.90116680e-01 9.08528864e-02 6.59817457e-01
-9.13414299e-01 1.61779687e-01 -7.24968433e-01 -3.70167762e-01
4.61230785e-01 -1.12975216e+00 -4.09310758e-01 -4.23744798e-01
-3.86822432e-01 -5.49567640e-01 8.93030405e-01 -2.53428727e-01
1.23755741e+00 -2.27103567e+00 7.36950561e-02 8.26169252e-01
3.21960360e-01 -2.57059783e-01 3.20921779e-01 5.64357519e-01
1.36094555e-01 3.43976855e-01 5.97909018e-02 -3.53798985e-01
-1.29593639e-02 -1.54474318e-01 -7.72874579e-02 4.76686716e-01
-3.36756349e-01 2.10201249e-01 -8.83892417e-01 -7.44510114e-01
-9.97995213e-02 1.95945397e-01 -6.64935827e-01 1.88839957e-01
-9.76822972e-02 1.89285055e-01 -5.91803074e-01 2.23151982e-01
1.72632709e-01 -4.21374589e-01 5.04340112e-01 2.50243377e-02
-2.31932506e-01 4.07372653e-01 -1.36831081e+00 1.27770174e+00
-3.17914814e-01 4.18058991e-01 5.51413968e-02 -8.61268938e-01
9.85150278e-01 1.65090263e-01 1.02247763e+00 -2.02733994e-01
3.01705778e-01 -2.09131584e-01 1.21804059e-01 -1.85846329e-01
7.73965120e-01 1.79700032e-01 -2.80539095e-01 1.13290799e+00
-3.68057162e-01 2.02038303e-01 4.15997207e-01 3.63100290e-01
1.03623164e+00 -5.71263254e-01 2.22319067e-01 -4.90306616e-01
2.38508299e-01 2.93083638e-01 2.95517623e-01 6.26221955e-01
-5.85005470e-02 3.95037293e-01 5.41986763e-01 -5.10271899e-02
-7.34682858e-01 -8.25870514e-01 -7.39474744e-02 1.47185731e+00
1.72617719e-01 -8.23736191e-01 -6.80985868e-01 -7.87942350e-01
-9.70657617e-02 7.20428765e-01 -7.01230943e-01 1.10222004e-01
-9.85741392e-02 -7.11937189e-01 -1.89568803e-01 6.49438947e-02
-2.88093425e-02 -6.37068212e-01 -6.47778451e-01 9.57129672e-02
-2.21855462e-01 -5.83213449e-01 -5.19892514e-01 4.33302909e-01
-6.65653706e-01 -8.64117205e-01 -1.32732570e-01 -6.79695129e-01
1.22523665e+00 3.87118697e-01 9.54229653e-01 1.58326611e-01
3.98878232e-02 5.67475617e-01 -6.72436357e-01 -3.30160171e-01
-1.81858629e-01 4.62071300e-01 -5.78644052e-02 2.39212260e-01
8.30289006e-01 -3.52556646e-01 -6.12408757e-01 3.25477749e-01
-7.81013727e-01 -6.02311976e-02 1.13713987e-01 2.99867332e-01
1.64348215e-01 7.47802436e-01 3.50292236e-01 -1.26877177e+00
9.49585557e-01 -5.45257688e-01 -2.35981211e-01 7.88853839e-02
-8.84225905e-01 3.33530232e-02 6.64738774e-01 -5.61799586e-01
-6.54700518e-01 5.54432631e-01 2.66476929e-01 1.16569959e-01
-1.90307394e-01 7.02423871e-01 -2.66537637e-01 2.83322036e-01
6.90618277e-01 5.19606331e-03 -1.41269177e-01 -3.01349103e-01
4.37052310e-01 1.00805295e+00 9.66878906e-02 -4.14282292e-01
6.70529246e-01 3.38522941e-01 -5.94740272e-01 -7.91097343e-01
-6.38828993e-01 -1.04341841e+00 -9.56252456e-01 -1.09726526e-01
7.29093194e-01 -6.23994470e-01 -6.21710598e-01 -7.91564658e-02
-5.86419940e-01 -2.38036141e-02 -1.47779770e-02 4.55899119e-01
-2.73645371e-01 3.27054560e-01 -1.32099092e-01 -5.50197363e-01
-1.87468305e-01 -1.16172290e+00 7.68203795e-01 2.30742786e-02
-1.09480143e+00 -1.05590451e+00 4.10848893e-02 3.48012507e-01
2.49734804e-01 -3.15037556e-03 1.14483058e+00 -1.29826939e+00
-1.11026227e-01 -5.85093021e-01 2.14262962e-01 -1.60743028e-01
6.49248600e-01 1.78616330e-01 -3.99633020e-01 -3.68807495e-01
4.16874737e-02 -1.49574041e-01 5.00658035e-01 8.82522762e-02
9.93089855e-01 -2.24041387e-01 -5.40358901e-01 1.56513438e-01
1.25787795e+00 5.14463603e-01 1.11504607e-02 3.61959606e-01
6.68859184e-01 8.52827370e-01 5.79654217e-01 5.23225009e-01
6.42038643e-01 3.74039412e-01 -8.54605157e-03 -2.21621051e-01
8.63183737e-01 7.82851353e-02 6.79833144e-02 6.25766695e-01
3.07855904e-01 -2.61613250e-01 -1.10513198e+00 5.87212205e-01
-1.77497745e+00 -6.86164498e-01 2.63463289e-01 2.22319126e+00
8.38181913e-01 4.02822793e-01 3.66860956e-01 5.28202415e-01
8.01966548e-01 -8.89447853e-02 -5.07104933e-01 -3.70513260e-01
4.68588293e-01 -2.76876748e-01 3.76692235e-01 6.00323498e-01
-1.13041890e+00 8.24837804e-01 5.64425516e+00 2.17938900e-01
-1.28184605e+00 -4.84867066e-01 6.06985271e-01 -9.21947882e-02
-3.73978108e-01 2.86415696e-01 -8.68747115e-01 4.56639647e-01
5.70999503e-01 -3.14420342e-01 3.20281386e-01 8.88523638e-01
5.21772563e-01 -3.58647525e-01 -1.26673090e+00 7.19603598e-01
2.60553718e-01 -6.82662427e-01 -7.24378526e-02 2.40866780e-01
5.13461411e-01 -5.27024209e-01 -2.47435942e-02 2.98137888e-02
6.92072093e-01 -7.47349620e-01 2.84310013e-01 1.56633794e-01
3.08532059e-01 -9.70097661e-01 1.74207613e-01 5.39670229e-01
-8.31270933e-01 -2.14212969e-01 5.15731126e-02 -1.70956954e-01
-1.76688880e-01 5.86643457e-01 -1.15905547e+00 -6.01502173e-02
5.44665575e-01 4.44748908e-01 -7.66463697e-01 6.49363935e-01
2.75995076e-01 7.29484558e-01 -4.22381550e-01 -2.74541616e-01
1.69304639e-01 -2.30481297e-01 9.63483825e-02 1.15316105e+00
2.89977998e-01 1.29278302e-01 5.58755696e-01 2.51905411e-01
1.25694633e-01 6.46170020e-01 -3.92168730e-01 -3.61079603e-01
6.85170531e-01 1.55625713e+00 -1.52631652e+00 -1.97801724e-01
-3.82451773e-01 8.51464033e-01 1.08094782e-01 2.06223235e-01
-2.32230335e-01 -7.47945964e-01 4.53817517e-01 5.14888406e-01
2.36708418e-01 -4.83716279e-01 -6.34005368e-01 -9.73918557e-01
-2.08575517e-01 -8.01936388e-01 4.79254216e-01 -4.34471041e-01
-1.00318956e+00 3.94064724e-01 -2.63142049e-01 -9.82128680e-01
-4.33033645e-01 -2.20789775e-01 -5.27583599e-01 5.96419752e-01
-6.39174581e-01 -6.48765862e-01 -1.83600500e-01 7.05366492e-01
3.45517516e-01 -1.47209451e-01 7.28292227e-01 2.84091309e-02
-3.80093068e-01 3.31319690e-01 2.67764419e-01 2.79950708e-01
9.10588920e-01 -1.35448658e+00 8.96003246e-02 5.78926861e-01
2.27110878e-01 1.08406651e+00 1.00950861e+00 -6.03688061e-01
-1.09329188e+00 -6.94090724e-01 1.18567562e+00 -4.43972617e-01
6.88777387e-01 -6.00482106e-01 -6.02504790e-01 6.08970225e-01
2.20376235e-02 -6.71224117e-01 1.31212842e+00 6.03761673e-01
-1.34591728e-01 -3.05770151e-02 -1.04579878e+00 7.79192746e-01
4.49176013e-01 -4.54893410e-01 -4.66487437e-01 3.05128276e-01
3.30789417e-01 3.01805064e-02 -7.71061122e-01 5.05788699e-02
2.84642190e-01 -7.43239462e-01 5.65710306e-01 -4.12553489e-01
1.67548448e-01 -4.74547267e-01 -1.01281710e-01 -1.28146887e+00
-4.55523461e-01 -3.76180261e-01 3.03349793e-01 1.46431208e+00
5.92407405e-01 -2.78652400e-01 9.07066822e-01 1.21490824e+00
3.17989647e-01 -5.89968741e-01 -2.51604855e-01 -1.57269225e-01
-2.31045097e-01 -3.70503992e-01 5.24874091e-01 1.22539020e+00
5.30238807e-01 7.71080196e-01 -3.53812017e-02 2.08398208e-01
4.36682731e-01 4.99072522e-01 9.19809401e-01 -1.76189327e+00
-6.81091566e-04 -4.40210789e-01 -1.99966073e-01 -8.03975582e-01
-6.31479034e-03 -7.09323943e-01 2.97042161e-01 -1.58100367e+00
1.72410920e-01 -6.96626425e-01 -2.62797266e-01 3.84744555e-01
-2.64711052e-01 -6.69106990e-02 -2.06323247e-02 4.59216505e-01
-9.39939737e-01 -9.98584628e-02 7.12159395e-01 1.42215550e-01
-7.52392590e-01 1.91822261e-01 -1.31818485e+00 8.55533779e-01
7.02716589e-01 -5.45503855e-01 -8.53392124e-01 -6.62622452e-02
4.61050808e-01 -1.97354987e-01 -3.03895146e-01 -6.45771027e-01
5.86910427e-01 -4.09909785e-01 5.14548898e-01 -5.31787872e-01
5.86124621e-02 -9.17490244e-01 3.91319878e-02 2.80204359e-02
-7.37508476e-01 1.13302022e-01 -3.81348103e-01 3.55998158e-01
-5.94953820e-02 -2.47699782e-01 7.21882939e-01 -2.18539968e-01
-4.24955785e-01 7.79687464e-02 -8.08918238e-01 -8.20178017e-02
1.10121036e+00 -3.90151829e-01 2.59870172e-01 -4.92054909e-01
-7.76575327e-01 2.72790194e-01 1.00293684e+00 4.82939065e-01
1.94265291e-01 -8.50302577e-01 -2.53918618e-01 7.54617527e-02
2.39089057e-01 -1.69254184e-01 -3.64301234e-01 3.33152622e-01
-7.72933364e-02 1.16296366e-01 1.90141290e-01 -5.22217572e-01
-1.61738765e+00 6.46708846e-01 -1.25482753e-01 -1.20888576e-01
-3.80229533e-01 5.28634131e-01 -2.80359741e-02 -2.80075431e-01
2.74576336e-01 1.29680589e-01 -6.90142691e-01 8.17611873e-01
3.08252990e-01 1.56398371e-01 2.07391411e-01 -6.73594415e-01
-3.63623530e-01 3.08190525e-01 -4.08538908e-01 -3.26766372e-01
1.31879461e+00 -4.95482087e-01 -9.88347083e-02 6.97156131e-01
1.20067358e+00 2.50807852e-01 -9.40763116e-01 -2.76794791e-01
4.46126908e-01 -4.23850358e-01 1.90382451e-02 -4.36580151e-01
-8.17853093e-01 4.45661038e-01 3.65469784e-01 7.50928342e-01
1.12747467e+00 3.09787765e-02 4.76592302e-01 5.28476119e-01
1.92872852e-01 -1.35392797e+00 2.76508689e-01 4.04542774e-01
1.60491943e-01 -1.23171270e+00 1.62744194e-01 -2.10734591e-01
-8.93595874e-01 9.82115805e-01 5.35872638e-01 1.12060741e-01
9.66634512e-01 2.95578718e-01 5.10861814e-01 -2.84764171e-01
-6.13889992e-01 -1.40971512e-01 2.08320543e-01 1.34469867e-01
7.64770985e-01 7.05594644e-02 -5.11786938e-01 3.35614651e-01
-6.83317125e-01 -5.35653830e-01 6.46000206e-01 1.08555257e+00
-7.32045352e-01 -1.24847758e+00 -4.09134895e-01 8.17478955e-01
-5.20067513e-01 5.35738692e-02 -8.92008543e-01 3.26034933e-01
4.01956737e-02 1.21633554e+00 4.32945251e-01 -3.80710840e-01
3.60710807e-02 7.81126022e-02 -6.65764958e-02 -1.05024648e+00
-5.74805021e-01 1.83677405e-01 4.03780974e-02 -4.51113135e-02
-3.01373839e-01 -1.07089674e+00 -1.34279847e+00 -2.02507287e-01
-3.54426950e-01 6.48976088e-01 7.44430840e-01 1.02207911e+00
1.69100434e-01 -1.65710062e-01 9.92483497e-01 -6.01658881e-01
-1.23588583e-02 -7.17142463e-01 -5.94291389e-01 6.59146547e-01
2.08338588e-01 -3.31694037e-01 -4.39442039e-01 2.84487844e-01] | [10.42912483215332, 7.2179694175720215] |
8b980edd-f6d3-4df0-8984-eacee5bff446 | dpm-clustering-sensitive-data-through | 2307.02969 | null | https://arxiv.org/abs/2307.02969v1 | https://arxiv.org/pdf/2307.02969v1.pdf | DPM: Clustering Sensitive Data through Separation | Privacy-preserving clustering groups data points in an unsupervised manner whilst ensuring that sensitive information remains protected. Previous privacy-preserving clustering focused on identifying concentration of point clouds. In this paper, we take another path and focus on identifying appropriate separators that split a data set. We introduce the novel differentially private clustering algorithm DPM that searches for accurate data point separators in a differentially private manner. DPM addresses two key challenges for finding accurate separators: identifying separators that are large gaps between clusters instead of small gaps within a cluster and, to efficiently spend the privacy budget, prioritising separators that split the data into large subparts. Using the differentially private Exponential Mechanism, DPM randomly chooses cluster separators with provably high utility: For a data set $D$, if there is a wide low-density separator in the central $60\%$ quantile, DPM finds that separator with probability $1 - \exp(-\sqrt{|D|})$. Our experimental evaluation demonstrates that DPM achieves significant improvements in terms of the clustering metric inertia. With the inertia results of the non-private KMeans++ as a baseline, for $\varepsilon = 1$ and $\delta=10^{-5}$ DPM improves upon the difference to the baseline by up to $50\%$ for a synthetic data set and by up to $62\%$ for a real-world data set compared to a state-of-the-art clustering algorithm by Chang and Kamath. | ['Esfandiar Mohammadi', 'Florian Thaeter', 'Marcel Gehrke', 'Tanya Braun', 'Johannes Liebenow', 'Yara Schütt'] | 2023-07-06 | null | null | null | null | ['clustering'] | ['methodology'] | [-5.08322567e-03 7.53075629e-02 -9.90750194e-02 -3.28389853e-01
-1.19093335e+00 -1.02395570e+00 6.10317700e-02 6.62910223e-01
-5.33148468e-01 5.19664466e-01 -3.02003384e-01 -3.08516622e-01
-2.57953912e-01 -1.06414580e+00 -8.21256936e-01 -1.01524758e+00
-4.66437697e-01 7.01464593e-01 2.30713516e-01 4.02696371e-01
3.14584911e-01 5.97795904e-01 -1.43898678e+00 -3.00093321e-03
8.15196991e-01 8.83455217e-01 -3.89977634e-01 3.29197437e-01
1.98365435e-01 -2.08116651e-01 -5.85069895e-01 -5.33393562e-01
9.49829459e-01 -8.96465406e-02 -7.07293868e-01 -1.00713328e-01
1.26308039e-01 -3.46737429e-02 -1.07819341e-01 1.28722656e+00
3.30860078e-01 2.86110770e-02 7.07461298e-01 -1.68678188e+00
-2.84120381e-01 2.98995674e-01 -1.07761884e+00 -7.98250586e-02
6.04834929e-02 1.27368569e-01 8.40059519e-01 -2.22310066e-01
4.03556019e-01 7.45961308e-01 7.27184594e-01 2.75361896e-01
-1.75460708e+00 -1.14780796e+00 -2.18777746e-01 -3.66560817e-01
-1.97177982e+00 -1.73087195e-01 3.75628412e-01 -3.96101207e-01
5.28411925e-01 8.61870885e-01 2.33293310e-01 2.54497290e-01
-7.52218515e-02 3.22698832e-01 8.94753635e-01 -7.05592707e-02
7.08190441e-01 2.03554943e-01 4.64382283e-02 1.65701538e-01
5.86494207e-01 -6.38932884e-02 -1.91500872e-01 -9.16706622e-01
2.35215530e-01 3.65638614e-01 -2.83639848e-01 -5.24248958e-01
-9.16409969e-01 9.41520631e-01 4.03584629e-01 -6.38511404e-02
9.05174669e-03 3.09342146e-01 2.36715034e-01 9.11529269e-03
3.66255909e-01 3.89684319e-01 -2.63479948e-01 -1.16524305e-02
-1.17353737e+00 4.41600531e-01 6.87765598e-01 1.27724171e+00
1.02948833e+00 -8.77094150e-01 -4.00894359e-02 3.79570723e-01
3.37804183e-02 5.59887350e-01 -6.17456473e-02 -1.09372032e+00
5.99572301e-01 8.12282324e-01 3.58417034e-01 -1.21055448e+00
6.03540381e-03 1.14014030e-01 -8.95272553e-01 3.25522691e-01
4.55583543e-01 -3.48590165e-01 -6.24074161e-01 1.71014535e+00
5.37724733e-01 -8.05932272e-04 1.11440353e-01 6.73488259e-01
1.92221850e-02 7.90584028e-01 -3.60519104e-02 -2.48634309e-01
1.38784897e+00 -1.33473009e-01 -1.17322281e-01 1.18010677e-01
5.34575880e-01 -4.56634432e-01 8.11859608e-01 1.74534306e-01
-9.01144803e-01 1.01436265e-01 -8.77216697e-01 1.10812284e-01
-4.37042058e-01 -2.57918358e-01 3.93935859e-01 9.14035261e-01
-1.10030961e+00 4.91749525e-01 -9.44507658e-01 -3.01631927e-01
8.29827845e-01 6.71271265e-01 -3.94341797e-01 -1.44185454e-01
-7.77373552e-01 -5.55457100e-02 2.61640251e-01 -5.19348145e-01
-4.58887517e-01 -9.35692728e-01 -5.99572301e-01 1.36526912e-01
1.75480396e-01 -4.96589005e-01 5.31526804e-01 -3.11225384e-01
-6.18982911e-01 9.92835879e-01 -1.52434394e-01 -8.35570335e-01
6.30928218e-01 3.48459989e-01 -1.75883651e-01 3.78457487e-01
3.28893632e-01 6.85426831e-01 4.50363100e-01 -1.74366355e+00
-1.10216880e+00 -8.57438803e-01 -3.38268399e-01 1.97423212e-02
-2.48513490e-01 5.43791652e-02 -5.79571366e-01 -4.29870307e-01
2.02232212e-01 -1.25818312e+00 -5.85563362e-01 3.14761817e-01
-5.62584698e-01 1.53495073e-01 1.00774467e+00 -1.76215574e-01
1.26543927e+00 -2.51187038e+00 -3.32114756e-01 1.05477011e+00
4.86230999e-01 4.90748361e-02 4.21036631e-01 4.11746234e-01
6.83159605e-02 5.71846008e-01 -8.63612592e-01 -5.21669686e-01
1.10449351e-01 1.88660517e-01 -2.83138752e-01 1.05371261e+00
-3.89259756e-01 2.28550240e-01 -7.77734101e-01 -2.70276427e-01
7.18916133e-02 2.10559070e-01 -5.02101183e-01 -3.03736236e-02
-3.57922874e-02 1.13266908e-01 -3.20822537e-01 4.38858092e-01
1.45276403e+00 -2.44757235e-01 1.85949668e-01 3.59327793e-01
-2.12907910e-01 -3.46493185e-01 -1.52579999e+00 1.30985725e+00
2.29877129e-01 2.63347477e-01 4.82005447e-01 -6.00592077e-01
1.04341960e+00 1.92979008e-01 8.29974532e-01 -6.41485378e-02
3.48977298e-02 3.19536358e-01 -4.85576004e-01 1.84402734e-01
5.31448424e-01 -2.28296787e-01 -4.17812020e-01 6.28900170e-01
-7.03187227e-01 -4.81900014e-02 -3.47644597e-01 2.19998941e-01
1.50602973e+00 -8.33664358e-01 -9.09545645e-02 -5.64811468e-01
1.11527815e-01 4.94375825e-02 7.40477383e-01 6.01434112e-01
-3.84647161e-01 1.06132627e+00 5.09540677e-01 4.82367985e-02
-7.54703462e-01 -1.13462174e+00 -2.81467259e-01 5.63693345e-01
6.36144042e-01 -4.42352295e-01 -9.58756685e-01 -6.30309463e-01
6.10462129e-01 7.42215812e-01 -6.27775967e-01 -1.05370775e-01
-5.86340800e-02 -8.22940469e-01 6.73110425e-01 1.41229883e-01
5.91815293e-01 -6.03055656e-01 -7.96755433e-01 -2.89066941e-01
6.16336614e-02 -6.04290962e-01 -6.35945559e-01 3.46999258e-01
-6.85365796e-01 -1.05463302e+00 -4.13197041e-01 -5.08146346e-01
1.08553684e+00 5.51291347e-01 5.16232848e-01 -3.18468362e-02
-2.31015906e-01 1.03042506e-01 -1.98634699e-01 -3.64227504e-01
-4.91596200e-02 -4.08911175e-04 -7.68050775e-02 6.46895915e-02
8.68685842e-01 -6.87780321e-01 -9.61641431e-01 4.84353155e-01
-9.80771780e-01 -6.56103551e-01 8.71634409e-02 2.16658339e-01
9.21163380e-01 7.01623023e-01 1.67027801e-01 -9.81895387e-01
3.12178403e-01 -9.14658904e-01 -9.99135852e-01 4.07220721e-02
-8.73809338e-01 -2.37545505e-01 6.08198225e-01 -5.40517317e-03
-3.78728479e-01 3.64063263e-01 2.96926647e-01 -7.38354087e-01
-3.07879537e-01 -8.90298486e-02 -4.00697380e-01 -1.16074257e-01
5.47352791e-01 1.21001221e-01 3.32068443e-01 -3.76424581e-01
3.14686626e-01 7.99483836e-01 6.33510351e-01 -6.53772891e-01
8.99095237e-01 1.10632944e+00 1.35814577e-01 -4.57845271e-01
2.94793129e-01 -7.86650896e-01 -3.97748619e-01 4.39677328e-01
7.05068231e-01 -8.59101951e-01 -1.28282940e+00 1.60448194e-01
-6.28940821e-01 6.36765808e-02 -5.86654544e-01 6.55557588e-02
-4.16536152e-01 5.49243987e-01 -2.44868115e-01 -1.04398954e+00
-3.40837300e-01 -1.18889201e+00 1.06722069e+00 2.59338245e-02
-2.27289200e-01 -6.21644676e-01 -5.17389551e-02 2.65047044e-01
3.32046710e-02 9.61760044e-01 7.38302588e-01 -7.74952114e-01
-6.55828238e-01 -5.80329597e-01 -1.41938880e-01 1.16759855e-02
1.39100283e-01 5.60700931e-02 -8.41836035e-01 -7.22734392e-01
-1.12843655e-01 2.24599838e-01 5.26599765e-01 2.21947059e-01
1.51410043e+00 -7.43437946e-01 -8.49040806e-01 8.47052455e-01
1.60652637e+00 5.29341042e-01 8.09624195e-01 2.10631534e-01
3.37054193e-01 5.11379600e-01 5.26322484e-01 7.48351097e-01
3.91934067e-01 3.15971345e-01 5.16734898e-01 -1.53368175e-01
6.74148500e-01 -2.40359426e-01 -1.40901715e-01 -9.94885936e-02
5.85152805e-01 -3.98787826e-01 -1.08261847e+00 1.03379929e+00
-1.79022002e+00 -7.21111953e-01 -2.20075399e-01 2.87166524e+00
7.93694794e-01 -9.34177078e-03 3.27274203e-01 9.46472362e-02
9.91223276e-01 -6.88312622e-03 -5.75875759e-01 -2.44975090e-01
-3.70034315e-02 1.89478979e-01 1.37498450e+00 4.22476500e-01
-1.13637114e+00 5.89518845e-01 5.10909224e+00 9.64393973e-01
-7.09542096e-01 -8.88570547e-02 8.82283807e-01 -3.80938023e-01
-4.05549198e-01 3.62432122e-01 -6.38367116e-01 1.01462162e+00
9.89392996e-01 -3.75822902e-01 3.64066690e-01 8.68216038e-01
3.04291025e-02 -4.56260026e-01 -1.16869855e+00 1.09336901e+00
-3.26533765e-01 -1.31086481e+00 -3.67172003e-01 8.21231425e-01
7.07981825e-01 -3.19165476e-02 3.50501567e-01 -3.75095218e-01
8.65839720e-01 -9.78244424e-01 4.65506077e-01 1.07935714e-02
9.19413447e-01 -1.33395505e+00 4.91366208e-01 4.37038958e-01
-1.14676476e+00 -1.71387494e-01 -3.29947978e-01 3.63895893e-01
-1.00298248e-01 7.72896707e-01 -8.41468394e-01 5.72617292e-01
1.25597990e+00 9.00432095e-02 -2.79519498e-01 1.00737524e+00
2.91342556e-01 5.61989486e-01 -1.01415801e+00 3.23726535e-01
1.02266766e-01 -4.19235379e-01 6.36050403e-01 1.14147496e+00
3.62672180e-01 5.40951192e-01 2.81712562e-02 8.65361452e-01
-2.99181551e-01 -2.10771640e-03 -6.47587061e-01 2.39519790e-01
1.12134957e+00 9.44352865e-01 -8.45272243e-01 3.87253352e-02
1.72392264e-01 9.22424912e-01 -7.13646039e-02 2.28334829e-01
-6.92731619e-01 -6.98474169e-01 1.25873542e+00 4.57791537e-01
6.07978225e-01 -1.54864803e-01 -7.62864292e-01 -5.57124794e-01
4.23394032e-02 -6.72668099e-01 8.95037830e-01 -1.38246268e-01
-1.33939528e+00 2.34734088e-01 8.90472233e-02 -1.33113301e+00
1.85742825e-01 3.82317901e-02 -5.00851870e-01 8.81220758e-01
-8.81845176e-01 -8.24718595e-01 -7.39071667e-02 6.78712666e-01
-4.51250225e-01 2.23262250e-01 8.31880093e-01 1.93477929e-01
-2.64179170e-01 9.94124234e-01 8.00485730e-01 9.84434038e-02
5.49258649e-01 -1.09697247e+00 3.20325017e-01 9.91764665e-01
-1.95864588e-01 8.88499558e-01 6.46232069e-01 -7.19343245e-01
-1.13478196e+00 -1.23765838e+00 6.61259949e-01 -4.09984708e-01
1.14537187e-01 -8.15159678e-01 -9.21203971e-01 5.91477633e-01
-9.06713381e-02 2.42019176e-01 1.02626824e+00 -3.69929016e-01
-3.05457830e-01 -3.27489883e-01 -2.36968327e+00 4.12540704e-01
8.38797987e-01 -2.36550987e-01 -1.03457570e-02 1.94327801e-01
6.47147357e-01 -1.81315184e-01 -1.14534485e+00 1.90584123e-01
6.01188950e-02 -1.01559865e+00 7.72758782e-01 -9.31184962e-02
-1.10064030e-01 -7.76565969e-01 -5.13285697e-01 -8.48488033e-01
-1.83677897e-01 -1.04368556e+00 2.98943132e-01 1.53056049e+00
3.58697921e-01 -8.47423971e-01 1.19865477e+00 1.37996328e+00
4.76744920e-01 -5.44464707e-01 -1.43699884e+00 -6.05118930e-01
3.48530203e-01 -1.52121872e-01 1.13370097e+00 1.05593705e+00
2.57228017e-02 -4.31166291e-01 1.29950881e-01 6.91302478e-01
1.02905178e+00 1.94886833e-01 9.53552902e-01 -1.10149550e+00
-3.68953198e-02 -2.14369953e-01 -5.74567080e-01 -6.64798200e-01
-1.25214219e-01 -8.43631685e-01 -8.40451568e-02 -1.12674677e+00
2.08695740e-01 -1.05413353e+00 -4.75088283e-02 6.60794258e-01
1.38552338e-01 1.32082731e-01 8.95187035e-02 3.25603873e-01
-4.32198375e-01 2.01929227e-01 2.49292076e-01 -4.84119989e-02
-2.22024351e-01 1.15446702e-01 -1.06430757e+00 2.43999153e-01
7.06943154e-01 -7.26757348e-01 -4.03991848e-01 -6.55826554e-02
-2.30351850e-01 -1.02844477e-01 2.86201656e-01 -9.63597238e-01
3.69809061e-01 -2.72063464e-01 1.48795813e-01 -8.34774077e-01
1.92305818e-01 -1.42509687e+00 8.33239257e-01 3.85997087e-01
-1.18253857e-01 5.21298163e-02 3.56127709e-01 8.73000741e-01
1.34705141e-01 1.47293925e-01 1.09216738e+00 -6.94481563e-03
-9.66658220e-02 5.11111677e-01 -1.07785679e-01 -1.95228029e-02
1.79030323e+00 -6.30779386e-01 -3.95405233e-01 -2.83083171e-01
-2.53776342e-01 6.64370298e-01 1.34687531e+00 -5.86589053e-03
3.28249097e-01 -9.31005538e-01 -4.49523360e-01 3.05235416e-01
2.86652952e-01 6.33373082e-01 1.46052614e-01 3.83239120e-01
-6.59027100e-01 4.56464514e-02 1.91917241e-01 -6.10002220e-01
-1.41140795e+00 7.34269083e-01 2.63156116e-01 2.30141714e-01
-6.00679815e-01 1.03115475e+00 8.35242793e-02 -2.71530867e-01
3.30671519e-01 -2.55339414e-01 7.53403723e-01 -1.31842792e-01
4.17532444e-01 6.33740962e-01 2.44151447e-02 -7.14645386e-01
-6.83901072e-01 2.66391307e-01 2.31828149e-02 -3.57407928e-01
1.35597515e+00 -1.19640484e-01 -3.19855362e-01 -1.02287352e-01
1.68513429e+00 2.95193166e-01 -1.40952003e+00 1.63484812e-01
-1.15361452e-01 -1.09638393e+00 -3.90815794e-01 -4.32798862e-01
-1.24594724e+00 4.35610741e-01 8.13546240e-01 4.85687166e-01
1.05039084e+00 1.28676638e-01 8.04759979e-01 -2.82767773e-01
6.95425808e-01 -9.37787831e-01 -6.44284964e-01 -2.46085331e-01
3.87866378e-01 -9.99732196e-01 3.19384009e-01 -4.23203707e-01
-6.19190872e-01 4.59919930e-01 4.05789971e-01 -1.16497867e-01
9.57331240e-01 3.35184634e-01 -1.87513262e-01 -2.94299126e-01
-2.88710117e-01 3.24690819e-01 -2.95913637e-01 6.47803068e-01
-3.74914259e-01 3.79011989e-01 -3.08331430e-01 5.23895383e-01
-3.42213660e-01 -2.27929384e-01 2.14425415e-01 1.28920364e+00
-2.99957424e-01 -1.10893297e+00 -6.57392204e-01 5.03153145e-01
-5.57445467e-01 2.02227265e-01 -5.34892082e-01 6.42244041e-01
3.32505286e-01 1.07774580e+00 4.92324591e-01 -3.70001853e-01
1.59199536e-01 -1.11725695e-01 -2.49890342e-01 -3.26254785e-01
-8.40336144e-01 -9.53683555e-02 -4.35928255e-01 -4.33878899e-01
-7.57446140e-02 -1.06789362e+00 -1.70028770e+00 -7.68943012e-01
-9.57113057e-02 5.32046735e-01 6.02871895e-01 2.21709415e-01
9.96070087e-01 -5.72004497e-01 1.00030887e+00 -3.40235651e-01
-4.55046684e-01 -1.91770926e-01 -9.12226260e-01 5.03070056e-01
2.72974312e-01 -3.07208776e-01 -7.27545202e-01 -1.45975664e-01] | [6.074069499969482, 6.6207594871521] |
1df4eabd-85d0-4b26-b57b-5083e49a42af | a-feasible-framework-for-arbitrary-shaped | 1912.04561 | null | https://arxiv.org/abs/1912.04561v2 | https://arxiv.org/pdf/1912.04561v2.pdf | A Feasible Framework for Arbitrary-Shaped Scene Text Recognition | Deep learning based methods have achieved surprising progress in Scene Text Recognition (STR), one of classic problems in computer vision. In this paper, we propose a feasible framework for multi-lingual arbitrary-shaped STR, including instance segmentation based text detection and language model based attention mechanism for text recognition. Our STR algorithm not only recognizes Latin and Non-Latin characters, but also supports arbitrary-shaped text recognition. Our method wins the championship on Scene Text Spotting Task (Latin Only, Latin and Chinese) of ICDAR2019 Robust Reading Challenge on ArbitraryShaped Text Competition. Code is available at https://github.com/zhang0jhon/AttentionOCR. | ['Wei Wang', 'Qingjie Liu', 'Di Huang', 'Yunhong Wang', 'Jinjin Zhang'] | 2019-12-10 | null | null | null | null | ['text-spotting'] | ['computer-vision'] | [ 0.21448414 -0.34704524 -0.08425187 -0.3341143 -1.0419716 -0.7428026
0.7701414 0.15851323 -0.6041505 0.16657433 0.41549432 -0.6941401
0.40079 -0.47817153 -0.69268703 -0.30145866 0.9657074 0.7557792
0.09508628 -0.18067384 0.6440099 0.30716074 -0.93252355 0.6336561
0.87089777 0.6321896 0.40957326 1.1325845 -0.51021445 0.77061695
-0.4779123 -0.5485365 0.13066302 -0.3196173 -0.97336686 0.25072873
1.112648 -0.4022772 -0.7910202 1.0494179 0.69007134 -0.06489685
0.8059326 -0.795424 -1.1910908 1.1753545 -0.90887487 0.34532624
0.33163917 0.16191009 1.1479392 -1.1781394 0.43857405 1.2534362
0.5220956 0.57806486 -0.51855695 -0.22932203 0.10135373 0.33253625
-1.3466114 -0.5972414 0.52919644 -0.39237058 1.2126814 0.58146286
0.21375613 1.1615627 0.19661808 1.7001643 0.7564227 -0.7402043
-0.14440306 -0.21552433 0.44851258 0.8213035 0.13892817 -0.79519
-0.41672966 0.439429 0.6238095 -0.21463782 -0.2490267 0.27822605
-1.5593579 0.762719 0.04961586 0.3695827 0.18927449 0.01780409
0.74842924 0.17804757 0.25217372 0.32049942 -0.3995712 -0.26427558
-1.1868039 -0.13490748 0.4196492 1.2655438 0.13186707 0.34658727
-0.6006015 1.2533605 0.4092527 0.9710771 0.9587038 -0.1356655
0.9727624 0.61808306 -0.25027946 -0.58564496 -0.38408363 -0.09676702
-1.1158947 -0.37351418 0.467227 -0.04041649 -1.4774723 0.75664675
-0.02892186 -0.46178457 -0.14433976 0.76685476 1.2521968 0.84698254
-0.25539932 0.41326278 1.4136716 -1.3945371 -0.59470075 -0.2690132
0.87910384 -1.2021224 1.5068052 0.3418998 -0.96192867 -0.41427913
-0.8446187 -0.8619807 -0.53721696 0.9152174 0.10631234 0.65579396
-1.1550905 -0.18999778 -0.746155 -0.89940894 0.59246475 0.18162072
-0.07827657 -0.07177961 -0.7378989 0.57867694 0.44751674 0.18895304
-0.47174507 0.06524944 -0.72075605 -0.02056075 0.22575517 -0.36277404
1.2069558 -1.1131382 -1.4246227 1.4773774 -0.3104755 -0.42250046
0.9042977 -0.39944983 -0.43325222 0.05537247 0.09924893 0.6881257
0.9771956 -0.6804431 -0.4146656 -0.40265054 -0.46303722 0.4328706
-0.18121506 0.4168723 -0.82158107 -1.0026461 0.2297135 -0.5077783
0.15596169 0.07509621 -1.1127733 -0.73254144 0.8476995 -0.9989418
1.0185286 -1.9598328 -0.2536708 -0.14658368 0.14799063 0.27866095
-0.16886199 0.4587314 0.3073263 0.341621 -0.2774881 -0.48133042
0.23310065 -0.15436305 -0.6061308 0.5507669 -0.03192814 1.3790479
-0.3562379 -0.71872634 0.41290972 0.1189547 -0.0301718 -0.33415908
-0.51653516 -0.00750522 -0.55043805 0.8825193 0.75058085 -0.32768795
-0.1686787 0.16669907 -0.3016264 0.13392404 -0.90986985 1.6464449
-0.02415977 1.394595 -0.17984083 -0.9098821 0.8014151 0.13367386
0.0900792 -0.7989482 0.5814841 0.09085926 -0.18130453 -0.6504285
0.9828683 0.55661 -0.05709627 0.4935817 -0.14659043 -0.15952532
0.29747695 0.41476694 0.7892711 0.1176217 0.1672011 -0.4174093
0.6327505 0.25801933 -0.0463555 1.1735076 -0.31197324 1.044845
0.35849938 -0.4185516 -1.4611257 -0.7204761 -0.35102376 1.3535124
-0.01887808 -0.3718399 -0.9012075 -0.6887134 -0.20578991 0.747281
-0.58015347 0.3843292 -0.81277174 -0.623229 1.2430311 0.5499789
0.9385607 -1.2119179 -0.26435357 -0.16231675 -0.3885579 -1.298676
-1.0474224 0.00998298 -0.5835073 -0.8118617 -1.1632473 -1.3302213
0.69446117 0.28519645 0.7634084 0.01464495 -0.54624254 0.6956701
-0.5274979 -0.52700746 -0.2902085 0.1788724 -0.4428388 0.15269151
0.5263643 0.46355432 -0.2318013 0.08769951 -0.8267967 0.41358393
0.4651703 0.7639994 0.64459753 -0.33226106 0.04846826 -0.6821413
0.62332076 -0.09802669 -0.5865972 0.7258011 -0.18665525 -0.20676267
0.74535555 -0.26601955 -0.87860763 0.09025584 -0.27262655 -0.0637214
-0.567443 0.47020575 -0.03472912 0.2653474 0.44647676 1.0914409
-0.659686 -0.4794263 0.26665896 1.2855407 0.4905189 -0.46542272
0.48244035 0.4905016 -0.5330075 -1.5823517 -0.67263985 -0.5915505
-1.032235 -0.0388173 1.2106268 -0.8463707 -0.45357344 1.0919739
-1.1925666 -0.74682987 0.10535709 0.2450432 -0.43505672 0.7076888
-0.6885663 -0.7284237 -0.67256844 -0.86318225 1.5361398 0.22304986
0.0991234 -0.97977275 -0.08694645 0.7988653 0.18405369 -0.3363396
0.8092712 -0.9574659 -0.5131414 -0.17596601 -0.51661617 -0.03867859
-0.22634214 0.3819486 -0.82827455 -0.27584124 -0.33908296 -0.40016842
1.1780404 0.48706198 1.4050987 -0.13595682 0.02163884 0.7336865
1.2621562 0.05273242 0.6684335 0.47744435 1.2681155 0.15293638
0.06314818 0.43901166 0.6971384 0.55494964 -0.11509103 -0.20513846
-0.4594391 -0.1820782 0.38555428 0.8842016 0.46499324 -0.8722426
-1.5452474 0.569442 -1.9141078 -0.92690736 -0.78188324 1.8465121
0.60609174 0.01943978 -0.03485261 -0.02103797 0.96063775 0.12909159
-0.73564506 -0.71370435 -0.788572 0.07048605 0.5789798 0.45236242
-1.448009 1.6445317 5.4967294 1.1446557 -1.3709824 -0.06340846
0.84687114 0.19125119 -0.1635685 -0.47137275 -1.0618204 0.3634626
0.4687244 -0.31752825 0.3038928 0.5888948 0.24949265 -0.07242665
-0.75615245 1.2095819 0.7237924 -1.3845794 0.5108264 -0.3530624
0.72772235 0.7286211 0.15471376 0.28014138 0.17160389 -1.2206066
0.93004674 0.19134793 0.9939986 -0.30755755 0.3306504 0.3147558
-1.120051 0.20635076 -0.5723209 0.3245581 -0.26440442 0.07422232
-0.82520574 0.292385 0.6691512 0.9948953 -1.0511143 1.2281346
-0.1617611 0.9694458 -0.39496547 -0.45677698 0.57527655 -0.15699331
0.66284454 1.7026291 0.07825263 -0.0698576 0.1559067 0.6725432
-0.52410823 0.66638196 -0.4791099 -0.27085975 0.16003865 1.078211
-1.436433 -0.5703976 -0.41940174 1.3563595 0.23445821 0.5233844
-0.72038853 -0.5724624 -0.1695646 -0.31846443 0.45308468 -0.5153711
-0.92178303 -1.720142 0.17040417 -0.99948525 0.564432 -0.9556669
-1.0762079 0.5371132 -0.6575949 -0.8907597 0.5742208 -0.90955126
-0.80891716 0.67155683 -1.0571947 -1.5724454 -0.18855152 0.98922426
1.5550325 -0.4270711 0.415429 0.0321279 -1.025191 1.0500215
0.92336696 0.9114854 0.7266878 -1.4239991 0.93608135 1.0740417
0.7369908 0.20061602 0.28100744 -0.7499142 -1.7032468 -1.1713917
1.0142734 -0.717756 0.80437875 -0.7045887 -0.709293 0.96957356
0.50478786 -0.5551062 0.20567842 -0.2588119 -0.41293508 0.4609218
-0.8397684 1.0634772 0.53942347 -0.44444507 -0.44322264 0.8138288
0.3380546 -0.5850242 -0.29484427 -0.10271529 0.36755803 -0.5110388
0.56355494 -0.46133417 0.59071153 -0.03527902 -0.18913811 -0.5147728
-0.11780661 -0.5592805 0.3335618 1.0672555 0.5448795 -0.5513164
0.8190017 0.27066994 -0.5001897 -0.4109983 -1.0561963 -0.62067974
0.72079724 -0.4861885 -0.03054055 1.2713599 -0.07068813 0.3448506
-0.39802846 -0.08062486 0.52433884 0.06379465 0.63376784 -0.6752264
-0.190139 -1.191048 -0.24664484 -1.6041116 0.03431638 -1.2812171
0.07308056 -1.842778 0.42244864 0.04870522 0.1634046 0.7414244
-0.12555084 0.30351672 0.3746073 0.3185512 -0.9707709 0.57589185
1.1733106 -0.6281128 0.14600588 -0.14020373 -0.35266614 0.5816532
1.2926668 -0.13712823 0.02967863 -1.1939641 0.13244224 -0.20788331
0.20999837 -0.6847547 0.49582574 -0.12482158 0.5977011 -1.055737
-0.08882467 -0.25550184 -0.8477431 0.22081167 -0.7776791 0.12864098
0.40340826 0.17770107 0.03220611 -0.28222257 0.78749424 -0.06625737
-0.65243596 0.34477404 -0.86433256 0.41052943 0.5731579 -0.48685342
-0.84295696 -0.26953962 -0.5097125 0.4436168 0.412269 0.7286836
0.6940346 -0.7641877 -1.3255547 0.1071491 0.21694557 -0.18106723
0.22376467 0.88596934 -1.0252184 0.8496546 0.02890616 -0.7030294
-1.4968957 0.23904221 0.4316512 -0.06637053 -0.90075564 1.0457201
0.15669338 -0.61954904 0.44002497 -0.2213283 -0.11213673 -0.367699
0.4186556 0.3842688 -0.0378176 -0.7782672 -0.2668707 0.8765017
-0.4982193 -0.07284395 0.91965324 -0.3343226 -0.16417693 0.6000961
0.85361034 0.09173255 -0.8143422 -0.317877 0.252817 -0.11432525
0.15909697 -1.1463146 -0.7025359 1.2157 0.43858162 0.07681643
0.7529293 -0.06437833 0.9211073 0.77636206 -0.15616196 -1.378765
-0.034798 1.1612294 1.0415483 -1.6392792 -0.1691289 0.11055577
-0.8100131 1.454134 0.6267324 -0.03758782 0.38262013 0.18684654
0.19639394 -0.14817277 -0.58578503 -0.30301782 0.5565068 0.328236
0.7646082 0.03021046 -0.11271103 0.10881018 -0.25985885 -0.59066117
0.8307561 0.77902603 -0.6937253 -0.79673207 -0.49870268 0.7753933
-0.5286856 -0.7401955 -0.9980667 0.4978048 -0.36535272 0.86986893
0.11483642 0.0759354 0.05499812 0.0606452 0.40367705 -0.54067177
-0.4968194 0.37209538 -0.12911303 0.06271745 0.2806347 -0.7987829
-1.3111906 -0.3651601 -0.10530014 -0.26415312 0.77446556 0.84301144
0.30555475 0.25206947 0.39437294 -0.40716282 -0.28070247 -1.0708855
-0.5163253 0.14505976 0.3069373 0.27405974 -0.11124635 0.42890579] | [11.972616195678711, 2.2877774238586426] |
5f372c3f-7dd8-4dda-abdc-51c7b15ba303 | psycholinguistic-features-for-deceptive-role | null | null | https://aclanthology.org/N16-1047 | https://aclanthology.org/N16-1047.pdf | Psycholinguistic Features for Deceptive Role Detection in Werewolf | null | ['Eyal Amir', 'Roxana Girju', 'Codruta Girlea'] | 2016-06-01 | null | null | null | naacl-2016-6 | ['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] | [-7.2285027503967285, 3.659602642059326] |
68da04b3-8711-4872-b19c-bb346a58aaca | streetsurf-extending-multi-view-implicit | 2306.04988 | null | https://arxiv.org/abs/2306.04988v1 | https://arxiv.org/pdf/2306.04988v1.pdf | StreetSurf: Extending Multi-view Implicit Surface Reconstruction to Street Views | We present a novel multi-view implicit surface reconstruction technique, termed StreetSurf, that is readily applicable to street view images in widely-used autonomous driving datasets, such as Waymo-perception sequences, without necessarily requiring LiDAR data. As neural rendering research expands rapidly, its integration into street views has started to draw interests. Existing approaches on street views either mainly focus on novel view synthesis with little exploration of the scene geometry, or rely heavily on dense LiDAR data when investigating reconstruction. Neither of them investigates multi-view implicit surface reconstruction, especially under settings without LiDAR data. Our method extends prior object-centric neural surface reconstruction techniques to address the unique challenges posed by the unbounded street views that are captured with non-object-centric, long and narrow camera trajectories. We delimit the unbounded space into three parts, close-range, distant-view and sky, with aligned cuboid boundaries, and adapt cuboid/hyper-cuboid hash-grids along with road-surface initialization scheme for finer and disentangled representation. To further address the geometric errors arising from textureless regions and insufficient viewing angles, we adopt geometric priors that are estimated using general purpose monocular models. Coupled with our implementation of efficient and fine-grained multi-stage ray marching strategy, we achieve state of the art reconstruction quality in both geometry and appearance within only one to two hours of training time with a single RTX3090 GPU for each street view sequence. Furthermore, we demonstrate that the reconstructed implicit surfaces have rich potential for various downstream tasks, including ray tracing and LiDAR simulation. | ['Yikang Li', 'Dongliang Wang', 'Chenjing Ding', 'Chiyu Wang', 'Botian Shi', 'Yeqi Bai', 'Xinyang Li', 'Nianchen Deng', 'Jianfei Guo'] | 2023-06-08 | null | null | null | null | ['neural-rendering', 'novel-view-synthesis'] | ['computer-vision', 'computer-vision'] | [ 2.56807417e-01 -5.61501905e-02 2.38807052e-01 -3.01288694e-01
-8.28674674e-01 -6.75035954e-01 8.46863091e-01 -4.68627065e-01
-2.23336563e-01 7.04723239e-01 -3.24997976e-02 -5.24639010e-01
1.43394753e-01 -1.10855782e+00 -8.78103971e-01 -4.28037912e-01
3.37918252e-01 8.80263925e-01 4.23438966e-01 -3.32737535e-01
2.52020091e-01 7.09374547e-01 -2.10881114e+00 3.91620323e-02
9.34725225e-01 7.73674548e-01 1.98181182e-01 7.90328801e-01
-2.03376412e-01 1.78791389e-01 2.96419878e-02 -2.90826201e-01
5.44536293e-01 8.89025778e-02 -2.48711422e-01 2.48162657e-01
7.13098943e-01 -6.20774448e-01 -1.87154144e-01 6.99385226e-01
4.43281263e-01 1.15730166e-01 5.40844619e-01 -1.06007302e+00
4.55848575e-02 -5.30357122e-01 -9.32926655e-01 -3.36234421e-02
2.43448153e-01 4.04909581e-01 6.43429279e-01 -1.30773783e+00
7.95088887e-01 1.14610243e+00 7.16752470e-01 3.08966458e-01
-1.25895548e+00 -7.49299943e-01 2.97698200e-01 -6.41774014e-02
-1.32458830e+00 -4.72405404e-01 8.21857393e-01 -5.43918610e-01
9.36748505e-01 3.86754990e-01 7.27916896e-01 9.57694411e-01
1.30800158e-01 3.20252091e-01 1.46419847e+00 3.35600995e-03
2.54508048e-01 -1.78828891e-02 -1.62549391e-01 7.95300961e-01
2.87979543e-01 4.65547293e-01 -6.61547780e-01 -7.08708838e-02
1.03075731e+00 1.97173562e-03 -2.24711776e-01 -8.75307858e-01
-1.20003247e+00 8.55739355e-01 1.79625452e-01 -6.20428145e-01
-1.68946341e-01 2.57649630e-01 1.29030019e-01 -1.46309063e-01
8.21732819e-01 -6.22603446e-02 -3.96700978e-01 -1.06642403e-01
-8.16341102e-01 5.64608097e-01 5.29212475e-01 1.28418148e+00
1.25887775e+00 1.34716630e-01 5.10019124e-01 6.39786839e-01
3.75725448e-01 9.23331797e-01 -4.98143554e-01 -1.43388510e+00
7.15239644e-01 4.31048423e-01 2.54543960e-01 -6.09471917e-01
-2.82931536e-01 -2.53550917e-01 -6.56148970e-01 6.86570048e-01
1.84114128e-01 6.15058765e-02 -1.01560509e+00 1.26697052e+00
8.75055730e-01 3.92994106e-01 -1.50466397e-01 9.19335425e-01
7.34898746e-01 5.03584206e-01 -5.19614339e-01 7.45206997e-02
1.45748842e+00 -9.03064072e-01 -1.01085678e-01 -3.00457180e-01
5.08648157e-01 -6.48297071e-01 1.18757522e+00 5.30315161e-01
-1.17047155e+00 -2.51044214e-01 -1.09097362e+00 -4.45152819e-01
-2.56061167e-01 -1.52744085e-01 6.61436915e-01 6.68340683e-01
-9.27802503e-01 2.42137119e-01 -8.15537333e-01 -4.37779352e-02
4.26169634e-01 1.94163248e-01 -1.80251762e-01 -5.27670801e-01
-6.93259954e-01 7.35474586e-01 -2.85601467e-01 -2.45355051e-02
-8.65431666e-01 -1.07538462e+00 -8.44295263e-01 -3.17154944e-01
6.09521747e-01 -1.26059484e+00 9.58050251e-01 -3.23221087e-01
-1.51183140e+00 9.34790730e-01 -3.91481519e-01 -1.47343650e-01
6.84444487e-01 -3.49006444e-01 -5.95089532e-02 6.96834102e-02
2.21258223e-01 7.06693947e-01 5.98752558e-01 -1.74484444e+00
-4.06225771e-01 -5.96116960e-01 8.40010494e-02 6.95089400e-01
5.22632182e-01 -5.19603133e-01 -5.08373976e-01 -2.10742831e-01
4.36390430e-01 -1.03797007e+00 -4.47307378e-01 4.03207362e-01
-2.93692827e-01 3.29677373e-01 9.13454711e-01 -2.91809767e-01
5.17355740e-01 -1.86557102e+00 3.72070968e-02 1.43652424e-01
1.57286301e-01 -1.57798722e-01 2.84529835e-01 4.39936131e-01
3.06002021e-01 6.46623001e-02 -5.10609746e-01 -6.09076858e-01
-1.71936423e-01 5.14885128e-01 -4.90235656e-01 6.80289686e-01
-1.14248142e-01 5.17404974e-01 -6.85172796e-01 -3.04857254e-01
5.06369531e-01 7.06516981e-01 -9.38611388e-01 2.07770631e-01
-4.02011216e-01 8.91659737e-01 -4.54282850e-01 5.48616588e-01
1.26639295e+00 -1.17417842e-01 -2.17227429e-01 -1.03082672e-01
-6.78206444e-01 3.98251891e-01 -1.17505956e+00 2.04094434e+00
-7.64085233e-01 4.64878142e-01 4.78698492e-01 -3.46586972e-01
7.06145644e-01 -1.14552118e-01 3.39720398e-01 -7.24442422e-01
-1.32340550e-01 1.67834103e-01 -5.13221264e-01 -2.36463726e-01
7.10495353e-01 -3.31815660e-01 2.46551782e-01 3.56764972e-01
-5.81275702e-01 -8.08092713e-01 -2.62817681e-01 1.24553896e-01
7.66525149e-01 7.11085379e-01 -5.14998883e-02 -1.80535495e-01
2.75104135e-01 1.87827483e-01 4.77624089e-01 2.94333071e-01
3.92326057e-01 1.10272348e+00 6.02554157e-02 -4.14489269e-01
-1.30666804e+00 -1.39679813e+00 -4.24091280e-01 5.76053500e-01
4.74783182e-01 -2.23152235e-01 -4.95672852e-01 -1.92748621e-01
6.09249733e-02 8.58062625e-01 -2.52139151e-01 3.95948082e-01
-8.66317630e-01 -5.72524607e-01 2.68063158e-01 2.77818114e-01
5.79997063e-01 -5.90057790e-01 -1.17263842e+00 -2.46080700e-02
-1.09500282e-01 -1.25319874e+00 -1.24308676e-01 -1.98823698e-02
-1.10527980e+00 -1.17582726e+00 -5.65126836e-01 5.39655611e-02
5.58534920e-01 8.25297058e-01 1.40475595e+00 -6.09150417e-02
-1.25307456e-01 6.33787930e-01 1.00329839e-01 -2.33803749e-01
1.62098855e-02 -2.57812560e-01 -1.35501221e-01 -2.32068338e-02
-7.82328397e-02 -9.39488173e-01 -9.45413172e-01 5.57046294e-01
-5.55132091e-01 8.02818060e-01 2.73612678e-01 6.06085539e-01
8.23128343e-01 -5.11493146e-01 -7.41620734e-02 -9.73523974e-01
-1.77282825e-01 -7.26473927e-01 -8.91773164e-01 -2.98503578e-01
-6.28553152e-01 -1.66787148e-01 2.77263224e-01 4.22861092e-02
-1.29981792e+00 1.75413899e-02 -2.29655951e-01 -7.12388813e-01
-2.23395437e-01 -2.34689079e-02 -7.23974705e-02 -1.78436890e-01
5.02435625e-01 2.69133896e-01 -1.29404053e-01 -3.29185963e-01
6.87815785e-01 2.26437360e-01 2.88909733e-01 -7.82137454e-01
9.33866024e-01 1.24380457e+00 3.14637870e-01 -1.14120805e+00
-5.58088958e-01 -4.27059114e-01 -7.05459356e-01 -4.02109981e-01
1.03313458e+00 -1.15845788e+00 -7.29509354e-01 2.40423411e-01
-1.19499755e+00 -4.16338950e-01 -1.87037349e-01 4.31077689e-01
-9.09223855e-01 4.11028773e-01 -1.48972809e-01 -1.06291783e+00
-3.83293740e-02 -1.34551179e+00 1.62357187e+00 -1.89322159e-01
9.55395848e-02 -7.55989134e-01 1.16285428e-01 7.00427055e-01
1.86439693e-01 4.24019963e-01 8.28437209e-01 3.63036782e-01
-1.54144454e+00 5.66585422e-01 -3.20204407e-01 -1.82388678e-01
-2.48455793e-01 -1.20323502e-01 -1.26844907e+00 -2.04744160e-01
4.90014628e-02 -2.54753262e-01 7.68368304e-01 3.68213058e-01
9.64739442e-01 1.46256641e-01 -4.25888240e-01 1.40059114e+00
1.49516034e+00 5.19140288e-02 6.19443476e-01 3.66433829e-01
1.00170302e+00 8.05168688e-01 6.22547865e-01 4.65633452e-01
9.80886042e-01 7.41659462e-01 9.05968428e-01 -2.64589071e-01
-1.22900993e-01 -3.81510645e-01 4.88137603e-02 6.73566639e-01
-3.20705056e-01 -1.88866019e-01 -9.54366505e-01 3.76359761e-01
-1.41017079e+00 -6.74245954e-01 -6.02766991e-01 2.31203389e+00
3.04863542e-01 4.28091139e-02 -1.57023907e-01 -5.55905923e-02
1.67537287e-01 3.79761696e-01 -8.59886110e-01 -2.39223883e-01
-5.85077405e-02 1.71147451e-01 8.66878867e-01 7.46490598e-01
-5.24628341e-01 1.00486231e+00 5.61041164e+00 5.51147461e-01
-9.69138682e-01 3.00151050e-01 5.47414422e-01 -1.83334276e-01
-1.05552590e+00 3.89718443e-01 -9.22704279e-01 9.28324014e-02
6.79361045e-01 3.96373004e-01 4.85239267e-01 6.00274026e-01
2.69528180e-01 -6.42333150e-01 -9.16223466e-01 1.15918589e+00
-6.99541718e-02 -1.54598379e+00 -5.61956912e-02 3.87827903e-01
7.75698185e-01 6.45833135e-01 8.23970884e-02 -2.63516046e-02
4.35798883e-01 -1.02898598e+00 8.84770155e-01 3.69644463e-01
1.23595631e+00 -6.48853302e-01 -6.84129521e-02 7.74049759e-01
-1.42709625e+00 3.37070197e-01 -2.02799648e-01 -1.40814796e-01
5.62691569e-01 5.05530477e-01 -6.37290120e-01 8.57864022e-01
7.67436206e-01 6.37699902e-01 -2.09175825e-01 5.67301691e-01
2.11335137e-01 2.79631525e-01 -6.70082450e-01 5.18447757e-01
1.97482690e-01 -7.38661706e-01 7.29294956e-01 6.56100988e-01
4.37747598e-01 4.03030217e-01 5.34868166e-02 1.04825187e+00
3.50989401e-01 -3.30623448e-01 -1.16872668e+00 7.28246033e-01
2.78498083e-01 1.03356373e+00 -8.19720209e-01 -1.75962344e-01
-6.73941076e-01 6.57297373e-01 2.43010089e-01 5.53690493e-01
-7.81151235e-01 4.01102036e-01 8.62871110e-01 8.60723138e-01
3.76693100e-01 -7.42492557e-01 -8.40839982e-01 -1.28559804e+00
5.36777861e-02 -5.05571067e-01 -1.57534257e-01 -1.05889726e+00
-8.27847540e-01 4.45472091e-01 4.37966734e-01 -1.28176379e+00
-5.89818545e-02 -4.21012819e-01 -3.57740521e-01 1.08479822e+00
-1.96515858e+00 -1.28327000e+00 -5.00569165e-01 6.60485089e-01
8.02196980e-01 2.38337293e-01 5.62573731e-01 2.58061856e-01
-6.49384558e-02 -9.92172733e-02 -2.00436842e-02 -6.03369892e-01
2.52126336e-01 -8.92053664e-01 8.37521136e-01 7.18869269e-01
-6.55715019e-02 5.00174582e-01 5.63450098e-01 -7.70637870e-01
-1.80589974e+00 -1.09880364e+00 1.75210178e-01 -6.48940265e-01
1.81157604e-01 -7.36771405e-01 -8.21964860e-01 6.81166053e-01
2.89725661e-02 2.18672097e-01 2.45053172e-01 2.63268780e-02
-3.01937848e-01 1.74683277e-02 -1.07530892e+00 7.10295022e-01
1.54108369e+00 -4.56225902e-01 -9.33818072e-02 2.14914262e-01
5.00882983e-01 -9.00798321e-01 -4.41470295e-01 6.23957157e-01
5.36045790e-01 -1.55173004e+00 1.30919397e+00 -4.05259095e-02
3.68776053e-01 -4.19873208e-01 -4.17010337e-01 -1.03349555e+00
7.18502924e-02 -6.56902850e-01 6.94148466e-02 7.95756578e-01
7.20310435e-02 -8.44230115e-01 1.14034033e+00 3.19197416e-01
-6.75807178e-01 -1.04565716e+00 -1.08459294e+00 -3.65632266e-01
2.62167342e-02 -7.83696055e-01 5.75899422e-01 6.19954228e-01
-9.71001208e-01 3.60748768e-01 -2.98764676e-01 4.52233642e-01
8.80557835e-01 4.22352433e-01 1.30121183e+00 -1.28936923e+00
-2.22351849e-01 -1.08206607e-01 2.10978955e-01 -1.65894973e+00
-1.38389394e-01 -7.27659285e-01 1.23347845e-02 -1.43415427e+00
-2.18522653e-01 -1.00073028e+00 6.87364459e-01 -3.32868636e-01
1.71850145e-01 3.11294883e-01 -3.92088033e-02 1.30481854e-01
-1.07410878e-01 8.42782557e-01 1.40759301e+00 3.82998884e-01
4.02264223e-02 -5.44901900e-02 -2.43131191e-01 1.07168818e+00
3.11855316e-01 -3.16906303e-01 -8.23286831e-01 -9.11177993e-01
5.59873164e-01 5.13326705e-01 6.08150601e-01 -8.38834643e-01
8.97345599e-03 -3.99135828e-01 1.37995422e-01 -1.35055423e+00
1.14444685e+00 -8.69933307e-01 5.16443193e-01 2.54639201e-02
2.91488171e-01 3.93941879e-01 1.45492584e-01 7.66454041e-01
1.88407883e-01 1.48520648e-01 7.64412045e-01 -4.93963093e-01
-4.67246473e-01 6.18904233e-01 -1.59831211e-01 3.07159990e-01
1.07105458e+00 -7.99698174e-01 -1.66692644e-01 -2.65465707e-01
-4.08267707e-01 3.00442010e-01 1.08356059e+00 1.36282742e-01
8.72850895e-01 -1.16847074e+00 -5.94575524e-01 5.55802345e-01
1.32328704e-01 8.81989002e-01 4.62105423e-01 6.45914435e-01
-8.87739658e-01 2.94397622e-01 1.18632421e-01 -1.10937619e+00
-1.07217824e+00 1.54390097e-01 2.73184091e-01 -5.94245978e-02
-1.23854363e+00 7.36898839e-01 9.39658523e-01 -7.71993101e-01
-3.60087961e-01 -4.39506829e-01 1.24247499e-01 -2.89826155e-01
7.28630573e-02 5.77585459e-01 1.66838497e-01 -8.37992132e-01
-1.26523167e-01 1.24782658e+00 3.06767881e-01 -4.24039334e-01
1.25563252e+00 -2.75154412e-01 2.81163931e-01 6.10684037e-01
8.33461165e-01 2.90743232e-01 -1.79516292e+00 5.61858825e-02
-5.85469007e-01 -6.81972504e-01 1.63186416e-01 -3.70708108e-01
-7.64445364e-01 1.33671546e+00 3.78627270e-01 -2.75553346e-01
7.64335990e-01 -2.28168458e-01 9.32552099e-01 1.01538785e-01
8.99738252e-01 -6.70070648e-01 -3.03837568e-01 6.83889925e-01
7.66235650e-01 -1.19445384e+00 3.42818290e-01 -9.41559970e-01
-5.09146333e-01 8.92661214e-01 6.07491255e-01 -3.03642035e-01
6.31101012e-01 3.26641530e-01 -1.26093611e-01 -5.02908051e-01
-7.92914867e-01 -3.58114913e-02 1.33436337e-01 6.56250894e-01
-3.95118073e-02 -1.11974128e-01 3.30043405e-01 -1.95408300e-01
-4.60804105e-01 -2.50741780e-01 5.62698960e-01 7.70094991e-01
-3.50163430e-01 -8.69110286e-01 -3.92164052e-01 3.93527746e-01
5.51181212e-02 -2.29977831e-01 1.97002113e-01 9.38552976e-01
2.62236714e-01 4.45329964e-01 4.18487459e-01 -1.40058383e-01
3.50623935e-01 -2.24670961e-01 6.24555886e-01 -8.32168698e-01
-2.54685253e-01 7.70174935e-02 3.30704272e-01 -7.61513412e-01
-3.13542962e-01 -8.89899671e-01 -1.33205831e+00 -4.58063990e-01
-1.76255256e-01 -2.35591263e-01 8.58461380e-01 9.51710463e-01
5.18352687e-01 2.16451347e-01 5.32000661e-01 -1.42183912e+00
-1.67310461e-01 -4.78938162e-01 -1.79839700e-01 -1.66426718e-01
5.44118702e-01 -1.18936872e+00 -2.24001259e-01 -2.74565816e-01] | [8.977422714233398, -2.963477611541748] |
79d7c6e0-aeb3-47a5-add2-17eefa1ef7fe | d2-decentralized-training-over-decentralized | null | null | https://icml.cc/Conferences/2018/Schedule?showEvent=2485 | http://proceedings.mlr.press/v80/tang18a/tang18a.pdf | $D^2$: Decentralized Training over Decentralized Data |
While training a machine learning model using multiple workers, each of which collects data from its own data source, it would be useful when the data collected from different workers are unique and different. Ironically, recent analysis of decentralized parallel stochastic gradient descent (D-PSGD) relies on the assumption that the data hosted on different workers are not too different. In this paper, we ask the question: Can we design a decentralized parallel stochastic gradient descent algorithm that is less sensitive to the data variance across workers? In this paper, we present D$^2$, a novel decentralized parallel stochastic gradient descent algorithm designed for large data variance \xr{among workers} (imprecisely, “decentralized” data). The core of D$^2$ is a variance reduction extension of D-PSGD. It improves the convergence rate from $O\left({\sigma \over \sqrt{nT}} + {(n\zeta^2)^{\frac{1}{3}} \over T^{2/3}}\right)$ to $O\left({\sigma \over \sqrt{nT}}\right)$ where $\zeta^{2}$ denotes the variance among data on different workers. As a result, D$^2$ is robust to data variance among workers. We empirically evaluated D$^2$ on image classification tasks, where each worker has access to only the data of a limited set of labels, and find that D$^2$ significantly outperforms D-PSGD.
| ['Ce Zhang', 'Ming Yan', 'Hanlin Tang', 'Ji Liu', 'Xiangru Lian'] | 2018-07-01 | null | null | null | icml-2018-7 | ['multi-view-subspace-clustering'] | ['computer-vision'] | [-3.18120629e-01 -2.06137687e-01 1.32803336e-01 -5.23264945e-01
-8.27985227e-01 -4.55773175e-01 2.04841290e-02 2.06672117e-01
-1.01232433e+00 6.95014060e-01 -4.03568864e-01 -6.34729147e-01
-2.99219042e-01 -7.79187560e-01 -6.15075469e-01 -8.73969555e-01
-2.79576987e-01 5.27172446e-01 2.14861080e-01 -1.78155914e-01
4.21764106e-02 3.65518540e-01 -1.37664115e+00 -7.21385479e-02
4.32685733e-01 1.27926290e+00 4.13384736e-02 6.58105671e-01
-1.79650873e-01 7.56795585e-01 -9.00473714e-01 -4.18729395e-01
7.40543842e-01 -6.47403538e-01 -5.68964660e-01 -1.65836006e-01
2.72996515e-01 -1.81722611e-01 5.49130067e-02 1.08888638e+00
7.22283661e-01 1.08175881e-01 5.81330895e-01 -1.17070067e+00
-2.16516078e-01 5.82023799e-01 -1.08039582e+00 4.92334872e-01
-3.02653879e-01 4.00086865e-03 7.01630831e-01 -7.27796912e-01
3.91181350e-01 8.62696588e-01 8.14222276e-01 3.77359182e-01
-1.29221880e+00 -8.65971267e-01 4.83013988e-01 -5.22928536e-01
-1.37285864e+00 -1.22159325e-01 5.78772724e-01 -5.83839417e-01
7.88921118e-01 3.07941467e-01 9.85239521e-02 4.36869860e-01
1.64929822e-01 5.97229183e-01 1.28053236e+00 -4.49231207e-01
4.97626096e-01 3.09377402e-01 3.51314515e-01 8.75159621e-01
4.28255320e-01 -1.52240783e-01 -7.39647150e-01 -5.36017299e-01
4.78088379e-01 2.63280123e-01 4.08562422e-02 -5.39730117e-02
-6.89710081e-01 8.95824790e-01 9.39347818e-02 5.79884239e-02
-1.73863128e-01 4.14313972e-01 5.06420553e-01 6.80147707e-01
7.39402950e-01 7.25912750e-02 -9.51386273e-01 -1.90305531e-01
-9.06481981e-01 1.41869187e-01 8.92988622e-01 9.90133226e-01
1.28002894e+00 1.19966380e-01 -1.47446170e-01 7.64207184e-01
1.99736595e-01 1.05998755e+00 6.45702422e-01 -1.05291700e+00
7.23737180e-01 5.60722113e-01 1.29037112e-01 -8.69934082e-01
-5.70985556e-01 -4.14984912e-01 -9.98400271e-01 3.04126740e-01
6.53916180e-01 -1.01818883e+00 -3.53359014e-01 1.86196721e+00
3.73000711e-01 -4.33735222e-01 -2.98958093e-01 6.24826670e-01
2.23327830e-01 2.21975788e-01 -2.18181670e-01 -3.42642695e-01
1.08454382e+00 -5.92273295e-01 -4.57710177e-02 -2.49212563e-01
8.74085426e-01 -6.45662844e-01 1.19336414e+00 3.50061595e-01
-1.03764248e+00 -4.12177026e-01 -7.72044480e-01 1.09853828e-02
-1.94418982e-01 1.99044451e-01 6.62703574e-01 1.05545735e+00
-1.15151298e+00 5.67505836e-01 -9.06304181e-01 1.43623263e-01
3.21805477e-01 6.70750022e-01 -7.74474889e-02 1.52731210e-01
-7.24319696e-01 1.92975551e-01 -3.41909707e-01 -9.03846547e-02
-7.29669869e-01 -3.89843583e-01 -4.18619186e-01 -5.35679981e-02
4.97255176e-01 -5.08331597e-01 1.22553802e+00 -5.93968093e-01
-1.37012231e+00 7.29990065e-01 -5.40799201e-01 -3.25893164e-01
6.89988017e-01 5.02108037e-02 1.72316343e-01 -2.05919787e-01
4.01868194e-01 3.16766575e-02 7.91110516e-01 -1.07408011e+00
-1.01244938e+00 -9.42540467e-01 -2.03070670e-01 1.47569273e-02
-6.08888507e-01 8.18158463e-02 -3.43465835e-01 -3.75856668e-01
1.41218871e-01 -1.12788725e+00 -2.76887804e-01 -4.02551085e-01
-1.93221256e-01 -3.60973567e-01 6.21888936e-01 -2.40280911e-01
1.27293134e+00 -2.16178608e+00 -3.22596729e-01 6.31028295e-01
3.62322360e-01 9.23770070e-02 4.24231082e-01 3.33565772e-01
3.79215181e-01 2.54350156e-01 -5.27626984e-02 -6.54462457e-01
4.89299707e-02 1.40683278e-01 -1.43091500e-01 5.86219788e-01
-4.62629855e-01 2.78874278e-01 -7.38426924e-01 -2.91990906e-01
-4.60589647e-01 5.42916879e-02 -3.31324935e-01 -3.02279085e-01
-8.74946415e-02 5.47786839e-02 -7.79057801e-01 3.22620988e-01
7.64155567e-01 -6.84615552e-01 3.85008574e-01 4.99220192e-01
-1.45603955e-01 -2.16510631e-02 -1.65484905e+00 1.34220910e+00
-4.15510416e-01 2.87874192e-01 4.92944151e-01 -9.84692931e-01
9.96364772e-01 1.15057580e-01 8.09216678e-01 -6.69803739e-01
1.80476725e-01 2.98329562e-01 -3.34375173e-01 -1.42654076e-01
3.28328341e-01 -1.04898065e-01 -1.77070349e-01 1.27533031e+00
-2.11244956e-01 -2.15330467e-01 3.71629655e-01 2.11962685e-01
1.34345400e+00 -4.50708538e-01 -2.05873296e-01 -4.77682561e-01
2.84637719e-01 -1.72702760e-01 8.93304825e-01 1.30041146e+00
-2.98866540e-01 2.86355525e-01 9.33836937e-01 -5.61882794e-01
-9.97210443e-01 -9.39338684e-01 1.64588556e-01 1.72032285e+00
9.71245542e-02 -3.36894631e-01 -8.68384600e-01 -9.34249461e-01
2.31343523e-01 1.96183950e-01 -5.57552755e-01 1.97233379e-01
-3.19989204e-01 -1.08892989e+00 6.73181355e-01 3.59702468e-01
7.10319996e-01 -7.37056732e-01 -7.35183358e-01 4.08453792e-02
9.01518613e-02 -5.48618674e-01 -5.93873024e-01 6.03766620e-01
-8.26780319e-01 -9.41133797e-01 -5.69164276e-01 -4.28647906e-01
1.05349171e+00 5.73184490e-01 1.08392012e+00 4.80385423e-02
-2.67083138e-01 3.45849901e-01 -2.37435117e-01 -1.12976587e+00
3.72981206e-02 2.93391228e-01 4.30066347e-01 -1.11450814e-01
6.61484301e-01 -5.10187209e-01 -7.59681880e-01 5.55758357e-01
-9.30828631e-01 -5.38895071e-01 2.80544907e-01 7.91073918e-01
7.62146771e-01 2.09048435e-01 4.56962228e-01 -1.23636401e+00
8.29624832e-01 -3.46926361e-01 -9.77653503e-01 2.53832489e-01
-9.62463975e-01 1.87557489e-01 9.13853347e-01 -2.21053272e-01
-7.91831851e-01 -3.47468965e-02 1.14924341e-01 -2.62451082e-01
2.12452546e-01 2.31529206e-01 4.86084640e-01 2.65084356e-01
1.12811255e+00 1.73144236e-01 1.54223815e-01 -6.06753230e-01
7.88406804e-02 7.51401544e-01 5.57733215e-02 -8.63901138e-01
4.32939768e-01 9.35591757e-01 1.66692153e-01 -4.55651343e-01
-8.14170957e-01 -3.22477430e-01 -1.35956630e-01 2.65931249e-01
3.96744728e-01 -9.73459244e-01 -1.12673914e+00 4.71639365e-01
-6.09032631e-01 -6.85556650e-01 -5.90144455e-01 5.03410816e-01
-4.66894954e-01 2.00941592e-01 -5.62732279e-01 -1.02100003e+00
-6.20324850e-01 -1.24490476e+00 7.68722653e-01 2.98184901e-01
5.87339653e-03 -8.22429717e-01 -9.86159742e-02 4.08798069e-01
3.73329371e-01 -1.73896119e-01 6.73715770e-01 -8.63685668e-01
-1.01024844e-01 -3.88521284e-01 -2.07659289e-01 8.58589709e-01
2.74832755e-01 -2.41685390e-01 -7.10350335e-01 -7.08009541e-01
4.13079083e-01 -3.64866883e-01 7.06979513e-01 2.44659349e-01
1.24325538e+00 -3.17974001e-01 -2.66802877e-01 3.38982880e-01
1.30629170e+00 5.91219254e-02 -6.69210330e-02 9.08578634e-02
2.14653701e-01 4.05036509e-01 2.36375168e-01 1.01948059e+00
3.05021733e-01 1.71268597e-01 1.02524847e-01 -1.24479681e-01
2.38428131e-01 1.51825398e-02 2.78308392e-01 4.02325302e-01
8.15636013e-03 -8.79637673e-02 -1.01804841e+00 5.20750821e-01
-1.71105468e+00 -6.59904003e-01 -1.40141934e-01 2.29720306e+00
9.92516637e-01 1.03308454e-01 2.11390033e-01 -2.28431020e-02
6.28377140e-01 -1.87057897e-01 -7.51686513e-01 -5.40916204e-01
-9.04584303e-02 5.64561903e-01 9.32824612e-01 2.68442154e-01
-7.24660575e-01 6.68922663e-01 5.31481266e+00 1.02950227e+00
-1.13700902e+00 2.57897884e-01 8.46805751e-01 -6.51523292e-01
-9.98399928e-02 1.07322194e-01 -1.23431301e+00 8.83749127e-01
6.87951207e-01 -2.74592400e-01 4.49869692e-01 1.12110245e+00
2.47310057e-01 -6.43919945e-01 -1.00184727e+00 1.00955534e+00
-8.23064521e-02 -1.11366987e+00 -4.33920711e-01 3.63512129e-01
1.22306681e+00 4.48473215e-01 4.20699090e-01 1.71536788e-01
1.17057216e+00 -6.72739029e-01 5.21747053e-01 1.73760429e-01
6.01156950e-01 -7.65131712e-01 7.00270772e-01 8.37483227e-01
-8.59442711e-01 -3.46805692e-01 -3.75222266e-01 -2.65299946e-01
-1.29689887e-01 1.30302370e+00 -6.69485688e-01 4.88653362e-01
1.31742084e+00 -2.48729840e-01 -3.55064273e-01 2.20772907e-01
-5.51987663e-02 3.74591589e-01 -7.53303587e-01 -2.41671294e-01
5.03261499e-02 -1.93216190e-01 1.00825943e-01 1.03467155e+00
3.50711524e-01 1.60386160e-01 3.77641618e-01 2.77163535e-01
-3.87701869e-01 1.43499374e-01 -5.03085971e-01 3.12456489e-01
4.70062733e-01 9.32637513e-01 -8.10198724e-01 -3.91983747e-01
-3.92107725e-01 7.24767506e-01 5.98069549e-01 4.21082169e-01
-3.54306519e-01 -7.48025179e-01 6.96810901e-01 1.10020265e-01
4.13578302e-01 -4.60220903e-01 -5.52806139e-01 -8.69049072e-01
4.49094206e-01 -6.69989467e-01 6.44540310e-01 -1.33905396e-01
-1.60834670e+00 4.33873922e-01 -1.05337195e-01 -7.18870759e-01
-3.97544503e-01 -5.40415823e-01 -1.21149272e-01 1.09801757e+00
-9.35367167e-01 -3.86951685e-01 -1.18467554e-01 8.96632791e-01
2.22304594e-02 -4.84336793e-01 6.05634630e-01 1.13297030e-01
-5.07275283e-01 9.43429649e-01 8.18907917e-01 3.43624592e-01
6.97268784e-01 -1.04053283e+00 2.46546324e-02 6.36002541e-01
8.81522596e-02 8.35903168e-01 5.26695967e-01 -5.68348765e-01
-1.12105227e+00 -8.31911087e-01 8.96560609e-01 -3.41330409e-01
3.99117827e-01 -3.04717839e-01 -4.86233860e-01 6.07999921e-01
-2.42770523e-01 3.06852788e-01 9.38788652e-01 8.46307576e-02
-3.72582167e-01 -6.73242748e-01 -1.42717695e+00 4.34593707e-01
8.65341306e-01 -5.41049063e-01 1.38049021e-01 3.34868848e-01
2.69576877e-01 -3.06166410e-01 -8.61321092e-01 1.93220213e-01
2.10557580e-01 -1.33174932e+00 4.53328699e-01 -4.17857409e-01
-1.95653588e-02 -5.90276644e-02 -2.99384475e-01 -1.06650460e+00
1.60709590e-01 -8.31523836e-01 2.60645717e-01 9.62756455e-01
6.23383880e-01 -9.93585408e-01 1.06896746e+00 1.15932500e+00
3.03543299e-01 -6.68324292e-01 -1.16848147e+00 -9.18139517e-01
4.74789709e-01 -3.66311193e-01 4.25684810e-01 7.07620978e-01
-1.24643654e-01 -2.69713365e-02 -5.02495281e-02 -1.23890482e-01
5.35447717e-01 3.38078439e-02 9.78828788e-01 -9.64495957e-01
-6.77985191e-01 -3.15863341e-01 1.77571326e-01 -1.06525457e+00
-1.52946085e-01 -6.53244853e-01 -5.72630614e-02 -1.13489223e+00
1.59538567e-01 -9.65783715e-01 -4.18656290e-01 7.61730492e-01
-4.54972573e-02 1.71678960e-01 1.33047864e-01 4.47247386e-01
-6.02372110e-01 1.99913785e-01 7.20682025e-01 7.03590438e-02
-4.51426804e-01 3.00917774e-01 -8.29430699e-01 6.97509170e-01
8.52240920e-01 -8.20703447e-01 -5.43069363e-01 -9.49051976e-01
7.07282126e-01 -8.64715800e-02 4.28272188e-02 -7.07756519e-01
5.41283607e-01 -2.18196720e-01 2.76232630e-01 -3.07008296e-01
7.38563165e-02 -5.18188119e-01 -6.28087670e-02 3.55104536e-01
-2.06411868e-01 5.44392645e-01 -7.21175447e-02 6.09119356e-01
6.86435997e-02 -2.93826818e-01 6.01490915e-01 -5.92180789e-01
5.50745800e-02 2.71359205e-01 -2.96156555e-01 2.95019567e-01
9.36372280e-01 1.56458333e-01 -4.01694179e-01 -3.09380502e-01
-5.27215600e-01 3.48870009e-01 3.40583205e-01 -2.35074416e-01
6.70873001e-02 -7.16747820e-01 -6.34547412e-01 2.30342522e-01
-3.38170290e-01 6.18783832e-01 2.35240325e-01 7.14565575e-01
-3.47789019e-01 1.41103864e-01 3.71589124e-01 -5.47117114e-01
-1.03515840e+00 3.68259817e-01 1.96458459e-01 -5.22262156e-01
-1.23203672e-01 1.38465595e+00 -2.77762711e-01 -4.22586799e-01
2.26198450e-01 -8.13196748e-02 6.77293062e-01 1.48054361e-01
3.30361485e-01 9.13013101e-01 2.64404774e-01 -1.18010305e-02
-3.28455120e-01 2.32526660e-01 -2.65835732e-01 -5.96155941e-01
1.18955588e+00 -2.89885342e-01 -4.19458300e-01 5.47971010e-01
1.31957531e+00 1.58408403e-01 -1.54329717e+00 -6.36776924e-01
-1.32531285e-01 -3.17246020e-01 -3.34625900e-01 -5.79561353e-01
-1.26194441e+00 7.43261158e-01 6.11078024e-01 5.43640018e-01
9.53522682e-01 -6.11267984e-02 5.67979395e-01 5.35872579e-01
7.69569397e-01 -1.70466745e+00 2.43565783e-01 6.35391176e-01
1.51960254e-01 -1.26112294e+00 3.16055685e-01 1.71964198e-01
-7.94671297e-01 7.25783229e-01 5.09109020e-01 -1.12220883e-01
1.04004967e+00 2.97668874e-01 3.04058075e-01 -2.14520290e-01
-5.74640810e-01 7.31879622e-02 -4.84423667e-01 2.13759497e-01
3.04220587e-01 -7.32188746e-02 -6.02735043e-01 6.73598051e-01
-4.23326582e-01 2.20878959e-01 2.96905071e-01 1.36940396e+00
-2.73613960e-01 -1.29283130e+00 -3.21804613e-01 6.83483720e-01
-8.17080498e-01 1.57844514e-01 -2.35362828e-01 5.57810485e-01
4.51595694e-01 1.29077232e+00 1.27592325e-01 -1.49292946e-01
1.54024005e-01 1.72292978e-01 7.40095302e-02 -5.50482571e-01
-9.43902731e-01 -1.68288704e-02 -3.11561108e-01 -2.22741380e-01
-5.98238893e-02 -7.18043983e-01 -1.60546076e+00 -5.63267410e-01
8.68702680e-02 5.44276059e-01 1.09166682e+00 7.85149515e-01
6.38982594e-01 4.93994765e-02 1.17509997e+00 -4.61833298e-01
-1.22746229e+00 -6.72736943e-01 -1.05264342e+00 3.60425442e-01
6.37397990e-02 -2.98978716e-01 -5.39308727e-01 -5.56517281e-02] | [6.343105792999268, 4.760767936706543] |
859dc5f6-b49c-4d14-9e1b-382d755b5d87 | carlane-a-lane-detection-benchmark-for | 2206.08083 | null | https://arxiv.org/abs/2206.08083v3 | https://arxiv.org/pdf/2206.08083v3.pdf | CARLANE: A Lane Detection Benchmark for Unsupervised Domain Adaptation from Simulation to multiple Real-World Domains | Unsupervised Domain Adaptation demonstrates great potential to mitigate domain shifts by transferring models from labeled source domains to unlabeled target domains. While Unsupervised Domain Adaptation has been applied to a wide variety of complex vision tasks, only few works focus on lane detection for autonomous driving. This can be attributed to the lack of publicly available datasets. To facilitate research in these directions, we propose CARLANE, a 3-way sim-to-real domain adaptation benchmark for 2D lane detection. CARLANE encompasses the single-target datasets MoLane and TuLane and the multi-target dataset MuLane. These datasets are built from three different domains, which cover diverse scenes and contain a total of 163K unique images, 118K of which are annotated. In addition we evaluate and report systematic baselines, including our own method, which builds upon Prototypical Cross-domain Self-supervised Learning. We find that false positive and false negative rates of the evaluated domain adaptation methods are high compared to those of fully supervised baselines. This affirms the need for benchmarks such as CARLANE to further strengthen research in Unsupervised Domain Adaptation for lane detection. CARLANE, all evaluated models and the corresponding implementations are publicly available at https://carlanebenchmark.github.io. | ['Johann Haselberger', 'Bonifaz Stuhr', 'Julian Gebele'] | 2022-06-16 | null | null | null | null | ['lane-detection', '2d-semantic-segmentation', 'unsupervised-pre-training'] | ['computer-vision', 'computer-vision', 'methodology'] | [ 8.32542330e-02 -1.44736797e-01 -6.67840600e-01 -7.31275916e-01
-9.01509583e-01 -8.32716405e-01 7.85502970e-01 -3.00162464e-01
-4.86253053e-01 9.02423739e-01 1.54326215e-01 -3.33490402e-01
3.02308321e-01 -4.51993018e-01 -5.89279652e-01 -6.58716977e-01
7.57499486e-02 5.78711987e-01 6.23133123e-01 -3.44173938e-01
1.88788548e-01 3.74742717e-01 -1.62189412e+00 1.49853542e-01
9.38652992e-01 7.68239021e-01 -1.46264145e-02 6.11171007e-01
6.19273111e-02 5.93346894e-01 -3.94503951e-01 -2.78360397e-01
5.51305890e-01 -1.83600232e-01 -5.52795291e-01 1.67156294e-01
9.18794036e-01 -4.19127882e-01 -4.47347283e-01 9.60894763e-01
7.20263481e-01 1.73557669e-01 7.94591665e-01 -1.62965953e+00
-5.14813185e-01 2.64930185e-02 -5.36540031e-01 3.88321191e-01
-8.79720896e-02 6.58902586e-01 6.62021816e-01 -9.53259528e-01
8.81364107e-01 1.22785854e+00 7.77501702e-01 4.75800276e-01
-1.20882714e+00 -1.07110763e+00 2.90287975e-02 3.41681361e-01
-1.15746224e+00 -7.96404302e-01 8.22764516e-01 -7.21847594e-01
8.62233460e-01 -3.51299554e-01 2.43503395e-02 1.51576042e+00
1.66786924e-01 7.61616051e-01 1.45361793e+00 -2.45066181e-01
1.82124332e-01 4.08394158e-01 3.50072235e-01 3.32064569e-01
3.21747184e-01 6.94265664e-01 -4.81409341e-01 1.72474518e-01
4.03681129e-01 -4.86669630e-01 2.65669376e-01 -9.62114990e-01
-1.30775845e+00 9.25930440e-01 1.25619620e-01 -3.73347402e-01
1.60539925e-01 -3.96659702e-01 8.33179772e-01 4.10993576e-01
3.25731307e-01 4.48548198e-01 -5.73418260e-01 -1.73482686e-01
-4.18588430e-01 5.06385922e-01 7.10259199e-01 1.40920472e+00
1.00006461e+00 3.12382337e-02 1.00739144e-01 9.84644353e-01
4.23813164e-02 9.64212656e-01 2.81637520e-01 -1.22208607e+00
6.27093196e-01 4.36089545e-01 1.51150659e-01 -7.99859226e-01
-4.64846969e-01 -8.32754299e-02 -5.20860791e-01 4.91033822e-01
6.97686255e-01 -3.39965701e-01 -9.37819004e-01 1.46135199e+00
2.57597327e-01 1.02715671e-01 3.86612296e-01 1.02969396e+00
1.07875752e+00 3.62947375e-01 2.45440230e-01 3.64461035e-01
1.05221879e+00 -1.23206687e+00 -4.44389373e-01 -8.78018439e-01
7.55731463e-01 -8.12198341e-01 1.17883837e+00 2.09009007e-01
-5.04891813e-01 -8.37164521e-01 -1.13906598e+00 -1.83099359e-01
-6.90790832e-01 6.39715046e-02 4.65689152e-01 5.01576602e-01
-6.34697497e-01 -1.69042856e-01 -2.81956136e-01 -7.52813935e-01
5.31245172e-01 -2.85110101e-02 -5.21632731e-01 -2.61196285e-01
-1.27965415e+00 1.11334229e+00 5.93032241e-01 -3.98171753e-01
-1.08434403e+00 -6.59203053e-01 -1.02624953e+00 -6.43218637e-01
4.22481656e-01 -1.97405770e-01 1.52362323e+00 -7.82377422e-01
-1.21365154e+00 1.36253166e+00 -1.16895199e-01 -8.31366658e-01
5.66472709e-01 -2.61864483e-01 -9.32264626e-01 -9.55134183e-02
5.46137512e-01 1.08558142e+00 6.80814385e-01 -1.28385031e+00
-1.06139410e+00 -1.66350245e-01 -2.03208849e-01 1.76353872e-01
-6.47956971e-03 -2.70712167e-01 -3.44926208e-01 -4.44479316e-01
-3.65063697e-01 -1.14954925e+00 -1.71200141e-01 -2.02090859e-01
-2.79635489e-01 -7.74393603e-02 1.15604341e+00 -3.83892268e-01
7.89343596e-01 -2.37094784e+00 -4.28637743e-01 -1.61310986e-01
1.04714662e-01 3.97302061e-01 -3.86283815e-01 3.37640554e-01
-1.14779375e-01 -3.57019931e-01 -3.46314132e-01 -9.02331695e-02
-1.14098461e-02 3.57284784e-01 -5.32103062e-01 5.27584732e-01
4.98520672e-01 8.93052638e-01 -9.10604417e-01 -4.82393861e-01
4.47167546e-01 -1.70446694e-01 -3.11256766e-01 6.69788942e-02
-1.49379760e-01 6.36226714e-01 -1.84229344e-01 6.82649851e-01
9.84825850e-01 -5.17956493e-03 -1.67757154e-01 2.31782775e-02
-1.40936211e-01 4.17852968e-01 -9.31540489e-01 1.59157395e+00
-4.05773893e-02 1.17291522e+00 -1.90048233e-01 -1.03770804e+00
1.27803016e+00 -2.82889664e-01 3.04300427e-01 -1.15241146e+00
-1.15651011e-01 1.74618125e-01 -1.27877668e-01 -4.52636778e-01
6.99041128e-01 -4.13052477e-02 -5.93854606e-01 -4.80476208e-02
4.41296063e-02 -2.49860555e-01 2.81515241e-01 1.78641289e-01
1.02812088e+00 1.64356261e-01 4.12889630e-01 -3.24290454e-01
5.69543719e-01 8.55121791e-01 6.62834764e-01 7.74588346e-01
-1.04109609e+00 3.88042808e-01 3.39093477e-01 -3.43796015e-01
-1.22303414e+00 -1.47940576e+00 -3.51503819e-01 1.23954070e+00
4.66060936e-01 7.50647038e-02 -4.65099990e-01 -9.11005437e-01
4.73570079e-01 7.84584284e-01 -4.63905632e-01 -2.75756419e-01
-3.33259910e-01 -2.93920606e-01 8.22347462e-01 6.18553638e-01
9.87241983e-01 -9.90769804e-01 -4.54305977e-01 -1.02956733e-02
-1.22535720e-01 -1.72575188e+00 -3.47993404e-01 2.27483556e-01
-6.56005859e-01 -1.14903140e+00 -3.71734649e-01 -1.02951503e+00
2.35093161e-01 5.60867310e-01 1.47800016e+00 -8.18073511e-01
-5.60796559e-02 5.55217743e-01 -3.56779039e-01 -7.79149175e-01
-6.64297342e-01 3.12502444e-01 2.01911122e-01 -3.94126952e-01
1.07664776e+00 -2.07404673e-01 -3.77280295e-01 8.27231884e-01
-3.65520954e-01 -5.40713733e-03 7.67600417e-01 7.57383466e-01
6.06356621e-01 -3.69086057e-01 9.11206007e-01 -1.03773737e+00
4.87401158e-01 -6.08554304e-01 -9.24796879e-01 -3.52564543e-01
-7.55004883e-01 -1.36902675e-01 5.20524740e-01 -3.89846832e-01
-1.24394953e+00 2.82585919e-01 1.48795992e-01 -3.47320497e-01
-9.38157082e-01 4.51404005e-02 -1.61767319e-01 -4.41771448e-02
1.11027992e+00 1.97688475e-01 3.44157189e-01 -1.92013249e-01
5.02858460e-01 8.26537848e-01 1.05413306e+00 -3.35983276e-01
1.03054833e+00 5.73347270e-01 -3.18443477e-01 -9.86608267e-01
-7.18702137e-01 -9.01100874e-01 -8.75956118e-01 -1.58087909e-01
8.23328435e-01 -1.32271683e+00 -6.18547536e-02 6.04983687e-01
-6.95092082e-01 -8.21325421e-01 -3.16444099e-01 4.01482671e-01
-8.23610187e-01 2.63613582e-01 7.99553745e-05 -3.36626559e-01
3.07604492e-01 -9.54304874e-01 9.96554732e-01 1.11940578e-01
-1.16387628e-01 -1.02909684e+00 3.07713926e-01 5.34734786e-01
2.31716067e-01 2.97708452e-01 4.12053734e-01 -7.72457063e-01
-3.20397824e-01 -1.06453411e-01 -5.33871472e-01 6.04107559e-01
1.20394677e-01 -3.21560442e-01 -1.11580133e+00 -3.00921947e-01
-5.68926573e-01 -5.77406645e-01 1.01651001e+00 2.21588016e-01
7.51953781e-01 2.82660961e-01 -4.52539802e-01 5.10822654e-01
1.06427157e+00 2.40402460e-01 6.44577980e-01 8.63490522e-01
6.62935674e-01 6.57368720e-01 1.15447533e+00 2.75914997e-01
7.78726280e-01 6.19566917e-01 2.39328727e-01 -3.36748481e-01
-4.87304807e-01 -2.50312597e-01 5.07333636e-01 2.90229499e-01
4.85581279e-01 -2.10411586e-02 -1.42761421e+00 9.62122023e-01
-1.69791949e+00 -9.95885253e-01 -3.45581800e-01 2.11933875e+00
5.72360814e-01 4.39415455e-01 6.11532807e-01 -1.49384752e-01
5.83486140e-01 2.02351242e-01 -9.80062068e-01 -2.94947296e-01
-4.00987774e-01 -3.73629898e-01 1.02328372e+00 3.39049011e-01
-1.55409098e+00 1.34823310e+00 6.50274277e+00 5.48509121e-01
-1.09384489e+00 3.79244946e-02 4.12879497e-01 1.25084043e-01
2.84239322e-01 -2.88441293e-02 -1.17054999e+00 4.22485113e-01
1.16470790e+00 -2.85193384e-01 6.54307902e-02 1.17970777e+00
3.22532058e-01 -1.99663252e-01 -1.04983425e+00 8.75660837e-01
-9.96522326e-03 -1.06354058e+00 -3.06855917e-01 8.18836764e-02
9.35606897e-01 5.78995585e-01 2.34389022e-01 8.11424434e-01
6.43368900e-01 -7.56926954e-01 4.34422970e-01 1.55088782e-01
9.54617620e-01 -8.06227744e-01 5.76166809e-01 2.99572855e-01
-9.44887817e-01 -1.60972446e-01 -4.12335783e-01 -1.30590275e-01
3.58111225e-02 3.97507578e-01 -1.27786040e+00 1.22061506e-01
8.24474812e-01 1.18420434e+00 -8.66662383e-01 9.13425505e-01
-1.35091364e-01 8.49136710e-01 -1.35489911e-01 3.35602343e-01
4.01444256e-01 1.08992122e-02 6.39815271e-01 1.60764968e+00
-1.96573794e-01 -2.53053993e-01 4.27748412e-01 5.39157748e-01
7.60846734e-02 -1.31406561e-01 -1.28083110e+00 1.65768579e-01
6.61837995e-01 1.01408958e+00 -2.34353870e-01 -4.80739385e-01
-7.72812665e-01 7.40700543e-01 1.24137573e-01 5.38610160e-01
-1.00340414e+00 -4.19039786e-01 1.24814928e+00 3.29046071e-01
3.13110858e-01 -5.10433614e-01 -5.66446841e-01 -8.88765752e-01
-1.84308261e-01 -9.67859328e-01 4.65502232e-01 -6.20506644e-01
-1.66001022e+00 1.45036086e-01 4.08598989e-01 -1.64348042e+00
-2.77295172e-01 -8.08560789e-01 -2.44150028e-01 6.36514962e-01
-2.14060855e+00 -9.42051351e-01 -7.68446088e-01 7.36352742e-01
8.75471830e-01 -6.41979575e-01 5.46436191e-01 3.57569486e-01
-4.50242788e-01 7.43377805e-01 3.32547218e-01 3.91960979e-01
1.66968024e+00 -1.03402984e+00 8.97982180e-01 8.60446990e-01
-3.24663967e-01 9.28043500e-02 6.66313291e-01 -5.33466816e-01
-1.10195780e+00 -1.59580052e+00 5.40187180e-01 -9.11203146e-01
7.85769582e-01 -3.97482246e-01 -9.31253970e-01 8.12397063e-01
2.21064612e-01 1.80093154e-01 4.37732249e-01 -3.49678919e-02
-5.04338682e-01 -2.82467037e-01 -1.05202281e+00 5.39183199e-01
1.23286164e+00 -3.07558298e-01 -6.17085218e-01 2.48566672e-01
5.29537737e-01 -5.48504174e-01 -7.80031800e-01 4.72606838e-01
3.47427040e-01 -1.18539476e+00 1.01627004e+00 -5.43161809e-01
4.93973762e-01 -3.70687962e-01 -1.45135164e-01 -1.39298797e+00
-3.72210920e-01 -6.96607679e-03 9.34612602e-02 1.16712940e+00
4.14951146e-01 -1.04798663e+00 8.44375134e-01 1.59185156e-01
-3.84739518e-01 -3.53957824e-02 -7.51004040e-01 -1.28763580e+00
4.95247096e-01 -5.24702609e-01 3.61513048e-01 1.02176452e+00
-3.72777969e-01 5.43624043e-01 -2.09274068e-01 2.22882926e-01
8.94494534e-01 -3.80694084e-02 1.49764025e+00 -1.39151680e+00
2.21747413e-01 -2.81051755e-01 -5.64551651e-01 -1.27265012e+00
5.41783631e-01 -8.70980024e-01 3.08106989e-01 -1.17798662e+00
6.40307143e-02 -6.02604687e-01 -6.63220063e-02 5.41957200e-01
1.33109108e-01 2.32976481e-01 -8.99464861e-02 3.34789634e-01
-7.90274143e-01 3.75003099e-01 1.12694299e+00 -3.23962063e-01
-2.35447124e-01 -2.87609041e-01 -7.63983190e-01 7.40432382e-01
1.28209972e+00 -2.03115016e-01 -5.31625092e-01 -4.43659127e-01
-5.37614107e-01 -3.97927970e-01 6.10289931e-01 -1.32375741e+00
5.98643683e-02 -4.49624956e-01 2.19316587e-01 -7.70268679e-01
6.09647594e-02 -7.24365771e-01 -4.51828361e-01 6.57268390e-02
-2.46736854e-01 -4.61946838e-02 7.38246500e-01 6.58850253e-01
-3.18887651e-01 3.61479402e-01 1.14626598e+00 1.35821834e-01
-2.06829786e+00 4.65986617e-02 -3.31846446e-01 5.99715352e-01
1.31746686e+00 -6.06847227e-01 -7.55079806e-01 -3.06668550e-01
-3.84695500e-01 6.17007077e-01 6.84267461e-01 9.60357726e-01
4.51846182e-01 -1.43442428e+00 -8.21256995e-01 3.12459737e-01
9.58357215e-01 3.80217433e-02 1.18654193e-02 6.91893816e-01
-2.24577039e-01 6.96205378e-01 -6.31509542e-01 -9.98234689e-01
-1.07715356e+00 6.27992451e-01 2.48022750e-01 8.76027718e-02
-6.42580211e-01 4.79924291e-01 3.95942688e-01 -8.80082369e-01
-2.73468774e-02 -1.39061645e-01 -4.55803126e-02 -1.22128472e-01
3.64568204e-01 4.21845227e-01 -3.37649137e-02 -8.93735349e-01
-5.77520728e-01 4.33347762e-01 -3.14788893e-02 1.00321226e-01
8.67458284e-01 -3.33308697e-01 5.78048944e-01 4.86762166e-01
1.08107531e+00 -1.70752898e-01 -1.71435213e+00 -3.39085042e-01
1.95853129e-01 -4.39368337e-01 -1.79288253e-01 -1.05775690e+00
-5.67726731e-01 7.83938706e-01 9.99622762e-01 -2.60966718e-01
1.06187260e+00 1.87759586e-02 6.02863550e-01 4.67187434e-01
4.21156615e-01 -1.35886836e+00 4.28716913e-02 1.01169097e+00
5.96410215e-01 -2.04023838e+00 -2.94676661e-01 -4.01252985e-01
-1.19543648e+00 6.26443028e-01 1.10578418e+00 -2.33315572e-01
3.07082802e-01 3.36984932e-01 5.07010102e-01 7.47718215e-02
-6.34382308e-01 -4.66911227e-01 1.52165264e-01 1.30801272e+00
2.65860170e-01 1.64961711e-01 1.33503312e-02 3.38975519e-01
-2.08923966e-01 1.24381386e-01 5.28944373e-01 7.67887712e-01
-5.87546289e-01 -9.56303835e-01 -4.69559938e-01 2.86041141e-01
3.34040880e-01 2.15247780e-01 -5.95254421e-01 1.28661191e+00
1.96335524e-01 9.34372425e-01 1.01367496e-01 -4.58665997e-01
7.56214321e-01 1.45811409e-01 -4.65377467e-03 -5.40554583e-01
8.05282742e-02 -4.26908821e-01 4.15256977e-01 -5.70915639e-01
-2.69506425e-01 -9.12078142e-01 -1.21379697e+00 -3.01943988e-01
3.55421573e-01 -2.77430326e-01 3.77524048e-01 5.78144610e-01
6.43505871e-01 3.20219785e-01 3.88520956e-01 -8.29919875e-01
-5.73078752e-01 -7.88964748e-01 -3.66899282e-01 7.05736458e-01
4.96024698e-01 -1.17381990e+00 -2.33185545e-01 2.89235651e-01] | [8.19431209564209, -1.6727945804595947] |
df335934-cc96-4f49-aff7-aaa356affa25 | expobench-benchmarking-surrogate-based | 2106.04618 | null | https://arxiv.org/abs/2106.04618v2 | https://arxiv.org/pdf/2106.04618v2.pdf | EXPObench: Benchmarking Surrogate-based Optimisation Algorithms on Expensive Black-box Functions | Surrogate algorithms such as Bayesian optimisation are especially designed for black-box optimisation problems with expensive objectives, such as hyperparameter tuning or simulation-based optimisation. In the literature, these algorithms are usually evaluated with synthetic benchmarks which are well established but have no expensive objective, and only on one or two real-life applications which vary wildly between papers. There is a clear lack of standardisation when it comes to benchmarking surrogate algorithms on real-life, expensive, black-box objective functions. This makes it very difficult to draw conclusions on the effect of algorithmic contributions and to give substantial advice on which method to use when. A new benchmark library, EXPObench, provides first steps towards such a standardisation. The library is used to provide an extensive comparison of six different surrogate algorithms on four expensive optimisation problems from different real-life applications. This has led to new insights regarding the relative importance of exploration, the evaluation time of the objective, and the used model. We also provide rules of thumb for which surrogate algorithm to use in which situation. A further contribution is that we make the algorithms and benchmark problem instances publicly available, contributing to more uniform analysis of surrogate algorithms. Most importantly, we include the performance of the six algorithms on all evaluated problem instances. This results in a unique new dataset that lowers the bar for researching new methods as the number of expensive evaluations required for comparison is significantly reduced. | ['Mathijs de Weerdt', 'Sicco Verwer', 'Rickard Karlsson', 'Arthur Guijt', 'Laurens Bliek'] | 2021-06-08 | null | null | null | null | ['bayesian-optimisation'] | ['methodology'] | [ 3.58447462e-01 -1.55805618e-01 -9.58128348e-02 -3.22993606e-01
-1.07988417e+00 -6.16131604e-01 7.85182834e-01 3.59269053e-01
-6.67685986e-01 1.18142724e+00 7.27494359e-02 -3.69388342e-01
-7.85585165e-01 -6.33522689e-01 -5.04259646e-01 -1.02711177e+00
-2.08102524e-01 7.53205001e-01 -8.71159360e-02 -3.79421748e-02
4.54170078e-01 2.35702142e-01 -1.71505392e+00 -2.37078950e-01
6.65986240e-01 9.46313500e-01 -1.59652546e-01 7.09673405e-01
3.44190389e-01 -7.50037283e-02 -7.91549146e-01 -4.33051050e-01
3.49226445e-01 -3.74026716e-01 -6.23972535e-01 -2.45628655e-01
-1.80522725e-01 1.08949542e-01 3.26077670e-01 6.85905933e-01
1.05457306e+00 3.63426924e-01 6.30546749e-01 -1.23303699e+00
1.89742055e-02 4.72174019e-01 -3.54539663e-01 1.71951458e-01
3.66819292e-01 5.91929436e-01 8.15497637e-01 -2.91144311e-01
4.16619688e-01 1.20249641e+00 8.03576827e-01 2.19587728e-01
-1.52915609e+00 -2.51447320e-01 1.34074427e-02 9.10986774e-03
-1.23398364e+00 -6.17399216e-01 3.29487473e-01 -2.18312860e-01
9.25305545e-01 8.70459497e-01 6.95569932e-01 1.12069726e+00
-3.09476312e-02 3.83625269e-01 1.55052483e+00 -3.09882402e-01
6.40557408e-01 1.54612705e-01 -2.18664601e-01 -7.58478194e-02
5.23580372e-01 8.16391528e-01 -1.67418867e-01 -4.06911224e-01
1.76717669e-01 -5.15666544e-01 -4.06569511e-01 -5.31575441e-01
-1.31870174e+00 8.82633328e-01 9.36343521e-02 1.43912554e-01
-2.60075063e-01 1.96516052e-01 7.23418891e-01 2.60555029e-01
3.99821460e-01 8.28584313e-01 -5.63208163e-01 -7.48825073e-01
-9.16658223e-01 6.49054885e-01 1.16237664e+00 4.46924448e-01
2.93787926e-01 -1.86532401e-02 -3.34913701e-01 8.18001688e-01
5.41355312e-01 4.17310119e-01 4.21066612e-01 -1.03478467e+00
2.87717432e-01 1.56208843e-01 4.39392924e-01 -9.12804127e-01
-4.14274186e-01 -7.12387562e-01 -5.83114862e-01 5.37695944e-01
6.72042847e-01 -2.39788726e-01 -6.40804589e-01 1.54240191e+00
3.60111147e-01 -1.57134905e-01 3.12094390e-02 9.22178566e-01
5.38426638e-01 4.88848805e-01 -9.09338370e-02 -2.51165450e-01
1.10098910e+00 -7.24123180e-01 -3.30454528e-01 -1.59118280e-01
6.62573814e-01 -9.45531011e-01 1.13381934e+00 5.67153037e-01
-1.09653199e+00 -9.97646227e-02 -1.12990808e+00 4.21053529e-01
-5.95475018e-01 -4.88224655e-01 7.51273811e-01 1.02114379e+00
-7.33757377e-01 8.48972201e-01 -6.82146132e-01 -1.83757082e-01
1.64776489e-01 6.10111833e-01 9.52953100e-03 6.05559051e-02
-1.13615978e+00 1.19177818e+00 5.71594000e-01 1.61814272e-01
-6.73757374e-01 -8.73166203e-01 -6.83202624e-01 -1.48938239e-01
4.33019936e-01 -8.11924577e-01 1.14785826e+00 -9.26271558e-01
-1.61023521e+00 6.72232568e-01 2.47389957e-01 -4.67804700e-01
9.69436884e-01 1.50421143e-01 -3.73378068e-01 -5.19791722e-01
-1.36326009e-03 2.73399174e-01 5.83148956e-01 -1.06959224e+00
-3.79587591e-01 -1.06861740e-01 1.35712745e-02 2.75571615e-01
3.20127010e-01 2.34766528e-01 -2.27013230e-01 -7.73747981e-01
-4.50992405e-01 -7.82268584e-01 -4.30503100e-01 -3.93592179e-01
-6.07435629e-02 7.83137903e-02 4.01514828e-01 -2.46763781e-01
1.42856872e+00 -1.97489464e+00 1.56972140e-01 2.49394849e-01
-2.53188938e-01 1.26405835e-01 -8.92449021e-02 5.87654233e-01
-5.68692014e-02 1.74051166e-01 -7.05193341e-01 -1.15181886e-01
4.55075622e-01 3.62955034e-01 -1.58806413e-01 7.23019242e-01
4.09641415e-02 8.04835856e-01 -1.09909856e+00 -3.12119544e-01
4.41144764e-01 4.85661238e-01 -3.24867100e-01 -2.42588427e-02
-1.73555151e-01 4.96474773e-01 -2.61028320e-01 5.42891502e-01
3.60937774e-01 -1.66553818e-02 -2.12137371e-01 2.71877646e-01
-1.06652431e-01 3.58930409e-01 -1.51591945e+00 1.37465000e+00
-6.14542723e-01 4.75076288e-01 7.76857836e-03 -1.04223526e+00
7.93090463e-01 2.66125023e-01 3.56418431e-01 -6.49324119e-01
2.16105744e-01 4.54567432e-01 2.86608577e-01 -2.30904296e-02
1.11505888e-01 -1.88166320e-01 -1.10592954e-01 5.64858496e-01
-3.54894906e-01 -5.77435493e-01 6.79950416e-01 -5.04823804e-01
1.05086279e+00 3.41486007e-01 3.12430263e-01 -5.61189950e-01
4.79530275e-01 2.61211488e-02 5.11903524e-01 7.14464009e-01
5.03872484e-02 7.66539514e-01 5.22744477e-01 -5.08513987e-01
-9.18020248e-01 -7.51218200e-01 -6.65931225e-01 8.49706233e-01
-1.63422331e-01 -4.14520055e-01 -4.94055390e-01 -3.13444525e-01
1.59902155e-01 9.26994622e-01 -7.41806328e-01 -1.22292079e-01
-4.40037698e-01 -1.58274639e+00 4.22787160e-01 2.96885997e-01
1.20195508e-01 -9.51137543e-01 -1.11091924e+00 3.75719160e-01
1.95771888e-01 -5.54895341e-01 -9.95486304e-02 3.75313312e-01
-1.12534845e+00 -1.13544512e+00 -7.43796527e-01 5.98051585e-02
4.36981142e-01 -5.09057105e-01 1.46144855e+00 -3.04318462e-02
-2.17883259e-01 2.91838944e-01 -1.15486212e-01 -6.34039521e-01
-4.64139670e-01 1.58279613e-02 -1.19153827e-01 -2.58787215e-01
1.75235122e-01 -5.06522000e-01 -4.69336718e-01 6.26396537e-01
-8.26014519e-01 -4.26920533e-01 5.66502333e-01 1.10886538e+00
5.65335512e-01 1.50330048e-02 3.94218922e-01 -7.61348546e-01
1.06284595e+00 -6.44687355e-01 -1.06215537e+00 2.40467906e-01
-1.12887383e+00 3.93730968e-01 3.81043643e-01 -2.70381093e-01
-7.28980303e-01 -3.89977515e-01 -8.62556994e-02 4.72046956e-02
-2.98998263e-02 6.33021891e-01 4.50508706e-02 -3.03780437e-02
7.78161585e-01 -1.89884081e-01 6.63430477e-03 -5.24010122e-01
1.31808639e-01 5.23510218e-01 3.27424020e-01 -9.17356253e-01
5.74960589e-01 1.18393891e-01 2.55436212e-01 -6.19625270e-01
-4.38699931e-01 -2.61687517e-01 -1.43595457e-01 7.20946863e-02
2.04768375e-01 -3.98969710e-01 -7.78554201e-01 1.17210865e-01
-6.47123218e-01 -6.79783642e-01 -6.88887775e-01 4.73236531e-01
-9.57999468e-01 2.22516656e-01 1.40595227e-01 -9.06432807e-01
-1.94898620e-01 -1.49581027e+00 9.44001973e-01 9.04296264e-02
-6.67994380e-01 -1.37765801e+00 4.22788471e-01 2.32374042e-01
6.81818426e-01 8.25877726e-01 6.36302292e-01 -4.90155816e-01
-1.67290628e-01 -3.78828675e-01 3.95922689e-03 2.00842857e-01
-1.16079316e-01 3.06580514e-01 -9.95800138e-01 -5.27877510e-01
9.89950374e-02 -1.37640834e-01 5.43646276e-01 5.28903067e-01
1.07264137e+00 -2.36175045e-01 -3.23053986e-01 7.37113714e-01
1.26776922e+00 1.03795692e-01 5.47799349e-01 8.63506317e-01
-3.66240367e-02 8.10529768e-01 8.59233320e-01 4.28072423e-01
-7.74607584e-02 1.04395676e+00 4.38755810e-01 -1.40129682e-02
2.02276155e-01 3.65141422e-01 2.46369898e-01 4.96827900e-01
-2.09710032e-01 -3.32918614e-01 -1.08078063e+00 5.76663196e-01
-1.89373088e+00 -6.42072856e-01 -1.88198537e-01 2.78150845e+00
9.23910141e-01 2.94000596e-01 3.07895124e-01 4.96110618e-01
5.58919251e-01 7.24368095e-02 -4.45497394e-01 -8.22470605e-01
-4.57547456e-02 3.31494987e-01 7.14676678e-01 5.29321671e-01
-9.32583630e-01 8.46649427e-03 7.00324345e+00 8.80172729e-01
-9.59484756e-01 9.89673957e-02 6.50740445e-01 -5.46154618e-01
-1.38655722e-01 1.95027560e-01 -4.47611153e-01 8.79806101e-01
1.47753716e+00 -4.70127523e-01 6.43066227e-01 5.12151539e-01
4.59132582e-01 -5.10556459e-01 -1.31778932e+00 8.64765048e-01
-3.95648599e-01 -1.10907006e+00 -6.12451792e-01 1.80512130e-01
6.64480329e-01 1.89700097e-01 3.47774513e-02 1.91397667e-01
3.97075057e-01 -1.48557544e+00 5.41000605e-01 3.51763517e-01
5.37297606e-01 -8.72667730e-01 1.06889784e+00 1.21759579e-01
-5.64557731e-01 -2.83801444e-02 -5.25806509e-02 -2.14432254e-01
2.64082700e-01 6.17900014e-01 -6.53699934e-01 6.63862765e-01
8.00294399e-01 3.02933484e-01 -3.74213248e-01 1.57768667e+00
-1.45483285e-01 6.93479300e-01 -7.15017676e-01 -2.58341998e-01
3.91754091e-01 -2.62965769e-01 8.63007486e-01 1.15792978e+00
1.57922730e-01 -3.40519726e-01 -3.44901174e-01 7.84335732e-01
5.62954307e-01 1.33202642e-01 -4.85896885e-01 -1.53452203e-01
3.77387077e-01 1.03450906e+00 -7.90470958e-01 1.51339173e-01
-2.25506485e-01 2.40306988e-01 -2.86290050e-01 2.86739856e-01
-7.14881301e-01 -5.70125639e-01 6.85101807e-01 1.10646203e-01
2.72156391e-02 2.70492639e-02 -5.74651718e-01 -6.26910269e-01
7.18363076e-02 -1.11026430e+00 5.93380988e-01 -4.99122024e-01
-1.13950586e+00 2.68768251e-01 5.83750129e-01 -8.92728746e-01
-5.98586917e-01 -6.10560477e-01 -5.68121910e-01 1.18372297e+00
-1.20039833e+00 -2.63824046e-01 -1.33391678e-01 -8.99120495e-02
1.65146247e-01 3.20775248e-02 8.87819648e-01 2.51934975e-01
-6.19931281e-01 4.64131027e-01 6.27216280e-01 -3.90165925e-01
6.98152840e-01 -1.32389557e+00 4.91378456e-01 4.98826146e-01
-5.68517670e-02 5.06968260e-01 1.16598213e+00 -3.47100049e-01
-1.31354189e+00 -6.49984956e-01 3.04525942e-01 -9.00807440e-01
7.18920767e-01 -2.24078774e-01 -6.41669333e-01 1.83979362e-01
5.53047955e-02 -1.10374235e-01 6.58125639e-01 1.94417804e-01
3.83788079e-01 -1.83466107e-01 -1.31220734e+00 4.47834760e-01
6.92246914e-01 1.08436190e-01 -5.01457691e-01 3.01546514e-01
1.99754268e-01 -5.45176804e-01 -1.14494503e+00 6.90919459e-01
6.30165696e-01 -1.05681849e+00 9.94897485e-01 -4.60012674e-01
1.80413857e-01 -3.26706916e-01 -1.17970839e-01 -1.45119202e+00
1.96858481e-01 -8.86033893e-01 -1.22675195e-01 1.21675169e+00
3.76208007e-01 -1.19923151e+00 4.69126701e-01 8.53388369e-01
2.07848549e-01 -1.28140771e+00 -1.13871872e+00 -9.64474738e-01
1.30567312e-01 -4.94733781e-01 9.58076954e-01 6.51737571e-01
-5.48060119e-01 1.12353802e-01 5.47740757e-02 -3.37780148e-01
6.95860028e-01 7.82920793e-02 8.76885176e-01 -1.31507981e+00
-5.38487554e-01 -8.51524234e-01 -4.89192337e-01 -3.22819531e-01
-2.51227736e-01 -5.97930253e-01 -1.11961797e-01 -1.15623581e+00
-1.74362406e-01 -7.56989956e-01 -3.65219206e-01 1.39583826e-01
-1.07680410e-01 2.68204600e-01 -1.35065243e-01 -1.36134801e-02
-1.93622991e-01 5.06845295e-01 9.58819032e-01 -3.78354974e-02
-2.36616313e-01 3.84090185e-01 -6.77096307e-01 4.81790900e-01
7.73925722e-01 -7.01074898e-01 -4.59066510e-01 -2.09145546e-01
6.57685816e-01 -2.35778213e-01 3.91750902e-01 -8.59147549e-01
-1.50294453e-01 -2.35013321e-01 2.63850003e-01 -1.73344612e-01
1.88787252e-01 -7.42945254e-01 6.15797341e-01 4.98841405e-01
-1.98778331e-01 2.44451642e-01 5.16625524e-01 4.67526764e-01
-1.04312710e-01 -7.05907524e-01 8.42715979e-01 -1.33340314e-01
-3.65040302e-01 -8.89438167e-02 8.33766982e-02 5.26009738e-01
1.22435141e+00 -5.06718755e-01 -1.93062574e-02 -3.39359581e-01
-5.97570419e-01 4.89867628e-01 8.49760950e-01 3.09071630e-01
-8.74311998e-02 -1.05971289e+00 -9.59692419e-01 5.64976186e-02
1.33451879e-01 1.76703930e-01 -2.82805532e-01 1.11888671e+00
-6.45262420e-01 5.20088017e-01 -8.86736661e-02 -6.20913386e-01
-1.10876286e+00 5.04990458e-01 4.13557112e-01 -3.13551277e-01
-3.41101408e-01 5.80383897e-01 -2.36737296e-01 -5.17634094e-01
3.17918599e-01 -2.31716082e-01 2.99174756e-01 1.65423229e-01
1.82590514e-01 8.68653417e-01 3.27563286e-01 -4.47913706e-01
-4.54411507e-01 4.53403175e-01 3.91534150e-01 -3.40717286e-01
1.47464097e+00 7.38775507e-02 -1.78885341e-01 5.96282899e-01
8.70963216e-01 -1.57360435e-01 -1.21428239e+00 4.13008571e-01
1.30663157e-01 -6.99321568e-01 2.12544441e-01 -1.11040306e+00
-7.33608484e-01 7.27900624e-01 6.47752583e-01 2.77650982e-01
1.15518129e+00 -5.12736738e-01 1.14298858e-01 2.07626715e-01
4.28631514e-01 -1.10256743e+00 -6.55169964e-01 1.71924829e-01
1.14451563e+00 -1.18069863e+00 5.43479621e-01 1.19325452e-01
-2.67740548e-01 9.24473524e-01 1.06444299e-01 4.36225459e-02
3.33093017e-01 2.70854115e-01 -3.59461531e-02 -7.77845010e-02
-7.74857044e-01 6.35561869e-02 3.93187016e-01 6.57921672e-01
3.60334158e-01 -2.73587424e-02 -6.95508420e-01 1.86740741e-01
-3.92239839e-01 -7.50612393e-02 2.19124734e-01 9.42413330e-01
1.75768673e-01 -1.52546191e+00 -6.40749872e-01 5.43024361e-01
-6.02894425e-01 4.97362800e-02 -2.96744496e-01 1.06231952e+00
-1.95066735e-01 8.94516110e-01 -2.86848307e-01 2.26839632e-01
3.21613312e-01 -2.97917128e-02 6.33502960e-01 -3.99166673e-01
-8.97936583e-01 -2.85999864e-01 3.35293710e-01 -5.99419236e-01
-3.66747916e-01 -8.54735970e-01 -5.86932242e-01 -4.83947963e-01
-4.45745260e-01 5.49228132e-01 9.12629604e-01 7.89957166e-01
2.68128008e-01 5.33960760e-01 2.49167696e-01 -1.05458081e+00
-9.57467377e-01 -7.45168030e-01 -2.62906134e-01 1.39277577e-01
1.58295631e-01 -9.78616059e-01 -5.38817823e-01 -5.39219439e-01] | [6.274319648742676, 3.7876124382019043] |
d9304b26-467e-4a80-b23c-bfb02fad8ffe | graph-attention-recurrent-neural-networks-for | 2103.10760 | null | https://arxiv.org/abs/2103.10760v2 | https://arxiv.org/pdf/2103.10760v2.pdf | Graph Attention Recurrent Neural Networks for Correlated Time Series Forecasting -- Full version | We consider a setting where multiple entities inter-act with each other over time and the time-varying statuses of the entities are represented as multiple correlated time series. For example, speed sensors are deployed in different locations in a road network, where the speed of a specific location across time is captured by the corresponding sensor as a time series, resulting in multiple speed time series from different locations, which are often correlated. To enable accurate forecasting on correlated time series, we proposes graph attention recurrent neural networks.First, we build a graph among different entities by taking into account spatial proximity and employ a multi-head attention mechanism to derive adaptive weight matrices for the graph to capture the correlations among vertices (e.g., speeds at different locations) at different timestamps. Second, we employ recurrent neural networks to take into account temporal dependency while taking into account the adaptive weight matrices learned from the first step to consider the correlations among time series.Experiments on a large real-world speed time series data set suggest that the proposed method is effective and outperforms the state-of-the-art in most settings. This manuscript provides a full version of a workshop paper [1]. | ['Bin Yang', 'Chenjuan Guo', 'Razvan-Gabriel Cirstea'] | 2021-03-19 | null | null | null | null | ['correlated-time-series-forecasting'] | ['time-series'] | [-2.73867130e-01 -2.23178208e-01 -1.29589915e-01 -3.10672611e-01
-5.38128018e-02 -3.57337683e-01 3.38006824e-01 6.03319287e-01
-3.61139059e-01 4.55452025e-01 3.63794684e-01 -2.59153903e-01
-4.53285486e-01 -1.14523780e+00 -8.70892107e-01 -3.77606988e-01
-6.73492491e-01 2.24429399e-01 1.06364138e-01 -3.69933754e-01
-5.15722297e-02 4.78568107e-01 -1.17415261e+00 -4.34726655e-01
8.03508162e-01 8.86889458e-01 1.83947742e-01 7.06276655e-01
-5.54063432e-02 6.65585876e-01 -7.37619340e-01 -3.95635664e-01
-2.20659319e-02 1.47308469e-01 -2.08973721e-01 -1.11074738e-01
3.22025344e-02 -6.81456029e-02 -8.95809472e-01 7.78746784e-01
2.01853633e-01 4.87606198e-01 3.39097291e-01 -1.64260137e+00
-8.81055951e-01 9.53096330e-01 -8.42747390e-01 5.92880428e-01
5.67232957e-03 -1.78491846e-01 1.09396040e+00 -3.53199989e-01
1.55310899e-01 9.60403919e-01 6.71103537e-01 -8.63820761e-02
-9.74709749e-01 -7.17610896e-01 9.58369732e-01 4.90435570e-01
-1.46954596e+00 5.97882271e-02 1.14575875e+00 -2.34386757e-01
1.12109745e+00 2.06319705e-01 8.68296385e-01 1.05237067e+00
4.09920424e-01 5.34392118e-01 2.43082866e-02 3.02570939e-01
-7.70562887e-02 -3.96705270e-01 3.40724051e-01 2.98069447e-01
3.77164096e-01 -3.20930809e-01 -3.04269552e-01 -9.74542201e-02
4.45992798e-01 7.86097884e-01 -1.29385635e-01 1.25440732e-01
-1.25907850e+00 5.72827220e-01 9.70948994e-01 4.56651092e-01
-8.94661427e-01 4.08349752e-01 3.19825977e-01 1.90504029e-01
7.76264906e-01 8.10447522e-03 -4.38591808e-01 -5.38194589e-02
-4.30046201e-01 5.30068539e-02 4.85428661e-01 9.50329304e-01
6.21905625e-01 3.04641098e-01 -3.84185128e-02 3.95929724e-01
2.78053164e-01 8.76514733e-01 2.79086679e-01 -4.64011639e-01
1.04951024e+00 6.22941792e-01 2.71687686e-01 -1.80454063e+00
-9.30802703e-01 -3.64946008e-01 -1.18806541e+00 -6.32627666e-01
3.08037400e-01 -5.69674671e-01 -6.82687163e-01 2.14229202e+00
3.32714587e-01 1.16758847e+00 -1.88650295e-01 7.71570146e-01
5.25348425e-01 1.00383997e+00 2.59877414e-01 -3.52318972e-01
9.65886652e-01 -5.94142079e-01 -9.22403276e-01 -1.40407635e-02
5.87204099e-01 -1.58963248e-01 3.20497572e-01 -1.96531564e-01
-7.89308965e-01 -4.90313202e-01 -8.21123302e-01 2.53924072e-01
-9.74196613e-01 -7.09755048e-02 5.44491112e-01 -4.74907719e-02
-7.98909724e-01 7.69544005e-01 -9.53307390e-01 -3.18944514e-01
-1.58520520e-01 1.82073146e-01 -1.13065653e-01 3.20331842e-01
-1.74964464e+00 5.87337255e-01 1.26419300e-02 5.65783083e-01
-2.51812518e-01 -5.69322824e-01 -7.48957992e-01 2.19136000e-01
2.74445027e-01 -4.76063132e-01 6.84525013e-01 -5.21957457e-01
-9.39470053e-01 2.75453907e-02 -3.06644201e-01 -5.06066203e-01
2.09136933e-01 -1.12116158e-01 -1.26501131e+00 -3.15010965e-01
1.91742659e-01 -1.86426193e-01 5.58174372e-01 -7.53929913e-01
-6.91170394e-01 -3.11951041e-01 1.50495633e-01 7.52235055e-02
-3.28431964e-01 -2.58257329e-01 -6.56546772e-01 -6.10578358e-01
-1.44470826e-01 -9.63218153e-01 -3.40218872e-01 -6.29121602e-01
-5.76549351e-01 -3.57662320e-01 8.80998909e-01 -6.26245201e-01
1.90941918e+00 -2.14282227e+00 3.26582849e-01 4.71172601e-01
5.12322545e-01 1.23648643e-01 -2.34724686e-01 7.63844609e-01
-6.53666705e-02 1.13249265e-01 1.06906377e-01 -4.83661622e-01
-1.40306219e-01 2.88860053e-01 -2.97792912e-01 5.64378262e-01
1.42246649e-01 9.29346144e-01 -1.23604596e+00 7.47416243e-02
2.78608292e-01 7.15631068e-01 1.45341143e-01 9.10983160e-02
1.43224625e-02 2.63243198e-01 -6.32695198e-01 1.99053615e-01
4.56943989e-01 -5.83390057e-01 4.55304310e-02 -2.63523161e-01
-2.14514494e-01 7.05787688e-02 -1.17145288e+00 1.20961428e+00
-5.72925091e-01 7.28586018e-01 -3.19775015e-01 -9.93083119e-01
8.07215214e-01 4.32678998e-01 7.40022719e-01 -6.94505572e-01
1.14999272e-01 -1.39157951e-01 1.79320052e-02 -5.33822954e-01
6.47474647e-01 6.11978590e-01 -2.34125331e-01 3.30989122e-01
-4.59830493e-01 6.03141367e-01 2.27333874e-01 2.80398041e-01
1.22603667e+00 -5.37805676e-01 -7.52282217e-02 3.88176262e-01
2.69332677e-01 -3.80269915e-01 6.85952425e-01 5.26144505e-01
-1.46923706e-01 1.89259514e-01 3.87460679e-01 -7.29476213e-01
-9.87316549e-01 -7.09666908e-01 4.85602140e-01 9.17213678e-01
3.04152399e-01 -4.66642231e-01 -4.18234915e-02 -4.03992295e-01
2.90751815e-01 5.22465587e-01 -1.13341880e+00 -7.67025724e-02
-6.18585706e-01 -7.77631104e-01 2.45686471e-01 6.95899546e-01
2.92856514e-01 -6.82221174e-01 -4.50416833e-01 6.53673470e-01
-3.03346276e-01 -1.14705491e+00 -7.02916920e-01 -9.98004749e-02
-5.88279068e-01 -9.19290423e-01 -6.83297813e-01 -3.11697245e-01
5.78101158e-01 7.32100546e-01 9.21253204e-01 1.53366819e-01
4.63214889e-02 3.69735986e-01 -2.88841873e-01 -5.04557550e-01
4.50022131e-01 2.67777592e-01 1.15907170e-01 6.49455845e-01
2.52749115e-01 -7.54305124e-01 -6.26204669e-01 1.96217000e-01
-6.74777985e-01 -7.34331682e-02 1.55078396e-01 4.82428819e-01
5.64028025e-01 2.00051323e-01 6.11429870e-01 -6.75046384e-01
7.14522243e-01 -1.29862559e+00 -5.80912650e-01 2.69550294e-01
-4.41645443e-01 -2.14892924e-01 8.87165546e-01 -8.03743243e-01
-4.90897715e-01 -3.62947345e-01 4.43200022e-01 -7.36247003e-01
1.97775438e-01 1.03348696e+00 1.16058409e-01 4.47491884e-01
6.46534190e-02 2.79785823e-02 -5.36079109e-01 -1.26245067e-01
3.18299055e-01 3.40717971e-01 3.67080659e-01 -1.24034576e-01
8.79307687e-01 4.18731809e-01 5.66779152e-02 -8.28825057e-01
-5.69313705e-01 -5.71257234e-01 -3.44108045e-01 -3.71794254e-01
6.27535760e-01 -1.00751424e+00 -7.28593946e-01 6.28227055e-01
-1.26689100e+00 -3.58071893e-01 1.00583944e-03 6.67275846e-01
9.65602472e-02 -8.34213048e-02 -6.49825990e-01 -8.59766483e-01
-2.07558095e-01 -6.22623503e-01 9.81784999e-01 5.11484802e-01
-1.87847823e-01 -1.55887771e+00 7.46131092e-02 -2.86768377e-01
6.22497678e-01 5.65520346e-01 6.16694450e-01 -5.96053064e-01
-4.88523930e-01 -5.57955265e-01 -3.15490782e-01 -5.14165342e-01
4.56118882e-01 4.20917183e-01 -3.63675207e-01 -1.07427217e-01
-6.24337733e-01 6.57233357e-01 5.56842089e-01 6.05353534e-01
1.13103580e+00 -2.98752964e-01 -6.66952848e-01 5.28143167e-01
1.17623329e+00 2.94259727e-01 2.33502984e-01 1.13960363e-01
1.34922218e+00 5.52841842e-01 4.40035164e-01 6.38185263e-01
1.20018578e+00 6.27488852e-01 5.07452905e-01 -1.54217094e-01
3.21081012e-01 -3.63528073e-01 8.50655660e-02 1.16347158e+00
-3.63207161e-01 -7.79987037e-01 -9.49902356e-01 8.97603095e-01
-2.33572292e+00 -1.34221673e+00 -5.60094416e-01 2.31559944e+00
-1.47559524e-01 6.56181648e-02 2.95134842e-01 -1.03938490e-01
9.13086772e-01 7.46653199e-01 -9.89193499e-01 -6.49578422e-02
-1.29885927e-01 -5.20174444e-01 1.06270123e+00 5.48873663e-01
-9.49082434e-01 6.73430443e-01 5.48848629e+00 9.29935873e-02
-1.37741554e+00 -2.27217413e-02 3.08061123e-01 -2.45414212e-01
-4.64954674e-01 -2.38164335e-01 -4.42400873e-01 9.00748968e-01
1.43245637e+00 -6.09237790e-01 7.09473431e-01 3.89710188e-01
6.18764281e-01 3.27966541e-01 -6.06295466e-01 7.57418573e-01
-1.78602323e-01 -1.12229669e+00 -1.54669419e-01 3.71768773e-02
6.07065737e-01 8.18963289e-01 1.06680676e-01 2.39380524e-01
4.78293180e-01 -6.75929189e-01 4.10965323e-01 8.87398958e-01
2.51639485e-01 -8.35577548e-01 5.83593190e-01 1.84564412e-01
-2.11714125e+00 -1.10905863e-01 -1.50656655e-01 -1.58664569e-01
5.42047381e-01 8.42001140e-01 -4.18702453e-01 7.77934551e-01
8.13340843e-01 1.40540373e+00 -2.67688870e-01 9.58176970e-01
-1.10617258e-01 5.64595520e-01 -5.06855488e-01 -3.94232005e-01
2.91528046e-01 -3.96425992e-01 6.46656692e-01 8.57142687e-01
5.66816926e-01 1.17734730e-01 4.39548492e-03 5.39801955e-01
-2.07018644e-01 -1.06939547e-01 -9.57793474e-01 -7.26298690e-02
8.67931426e-01 1.16166687e+00 -4.84826386e-01 -5.16670048e-01
-6.20496213e-01 6.86370671e-01 5.31704664e-01 1.04186714e+00
-1.26102722e+00 -6.72617793e-01 9.41103160e-01 3.62819433e-03
2.68134385e-01 -5.64224303e-01 1.44268602e-01 -1.15877926e+00
1.53487802e-01 1.50299426e-02 4.67527866e-01 -7.48601735e-01
-1.37496150e+00 6.80031717e-01 -1.86982393e-01 -1.42733598e+00
-1.56885102e-01 3.42494175e-02 -1.04979992e+00 9.11599517e-01
-1.47778964e+00 -9.99909222e-01 -3.80045682e-01 7.22847223e-01
2.88074553e-01 1.32495627e-01 4.65549946e-01 4.74154115e-01
-1.12906122e+00 2.94223130e-01 1.38753176e-01 4.11351621e-01
3.23994786e-01 -1.14276803e+00 1.09911335e+00 1.00870180e+00
2.23012879e-01 6.43355072e-01 5.48665285e-01 -8.57866585e-01
-1.53965962e+00 -1.49187064e+00 1.17024863e+00 -1.73395798e-01
1.34845078e+00 -1.33948117e-01 -1.09189498e+00 1.01795185e+00
2.51986086e-02 3.01947951e-01 5.74939251e-01 3.13667566e-01
-3.25429827e-01 -3.00008774e-01 -6.84458673e-01 7.18195796e-01
1.02446699e+00 -6.22504771e-01 -2.49231577e-01 3.11355472e-01
9.47233677e-01 -4.52074766e-01 -9.78805006e-01 1.74598083e-01
3.96667540e-01 -3.39624822e-01 7.43307769e-01 -5.66265941e-01
5.89341782e-02 -4.04844910e-01 7.49683306e-02 -1.88997209e+00
-7.00412810e-01 -5.84821463e-01 -5.46277940e-01 1.06921506e+00
6.35689437e-01 -1.19371629e+00 4.32951748e-01 9.77143288e-01
1.24827921e-01 -6.26691520e-01 -9.49573636e-01 -5.51324248e-01
-4.11534905e-01 -5.31628072e-01 1.20934260e+00 1.12742567e+00
2.62352731e-02 4.38562155e-01 -7.26114929e-01 7.43145108e-01
4.73454803e-01 1.27843648e-01 6.80240333e-01 -1.26666594e+00
3.78855616e-02 -5.04657805e-01 -4.97078925e-01 -8.74756515e-01
1.63316160e-01 -5.28276742e-01 -3.47247571e-01 -1.63470709e+00
-4.17029530e-01 -4.68483746e-01 -7.58292258e-01 3.08664411e-01
-4.43846494e-01 -3.14074278e-01 8.86956751e-02 1.44794192e-02
-6.44531906e-01 5.34933746e-01 8.71664703e-01 -3.87015313e-01
-3.53812665e-01 2.48508349e-01 -5.21505296e-01 3.90679687e-01
7.84326613e-01 -4.18722451e-01 -6.36575580e-01 -7.52074540e-01
4.57017273e-01 4.19854492e-01 9.04058814e-02 -8.75687659e-01
5.99345863e-01 -2.31279045e-01 1.49718434e-01 -7.21836686e-01
4.56091344e-01 -1.23558652e+00 6.09496951e-01 2.14565217e-01
-1.90226540e-01 8.37808549e-01 1.91854343e-01 1.18691111e+00
-1.94376543e-01 6.52474225e-01 -1.53810143e-01 2.85641611e-01
-7.10033000e-01 1.01124215e+00 -3.07124317e-01 -7.49847740e-02
1.03242028e+00 -3.92381139e-02 -4.55859423e-01 -7.41652012e-01
-6.04323328e-01 9.31366503e-01 1.02187842e-01 8.16198111e-01
4.16090488e-01 -1.46681213e+00 -5.62147856e-01 -3.90716642e-02
2.02778742e-01 -3.08834929e-02 7.50332952e-01 9.32394385e-01
7.46629015e-02 1.99518532e-01 8.35239664e-02 -4.00919586e-01
-9.82662022e-01 7.32791424e-01 2.43945107e-01 -3.91577005e-01
-5.12784362e-01 5.17868042e-01 -3.94476414e-01 -1.77118674e-01
1.57879978e-01 -6.80720448e-01 -5.90589881e-01 4.03160453e-01
5.79838872e-01 5.05111933e-01 -8.93546641e-02 -9.05271173e-01
-4.01956260e-01 9.05284405e-01 1.52733594e-01 2.34448448e-01
1.55758905e+00 -5.24037898e-01 2.04647556e-01 1.12981164e+00
1.32228100e+00 -3.18440534e-02 -1.14013863e+00 -5.18394768e-01
-1.32813454e-01 -1.97059959e-01 7.56618306e-02 -2.72320777e-01
-1.67693913e+00 5.22794068e-01 1.62505150e-01 9.58833992e-01
1.05622602e+00 -3.40201706e-01 1.13300753e+00 8.01458359e-02
4.87106234e-01 -1.08586431e+00 -5.31377077e-01 6.91044629e-01
6.25327528e-01 -1.13629735e+00 -2.90489525e-01 -1.19938254e-01
-5.87045193e-01 8.65240037e-01 3.45305115e-01 -2.60634750e-01
1.04938626e+00 -1.02643177e-01 -2.23457161e-02 -2.50775486e-01
-9.93338943e-01 -3.54234040e-01 2.74843425e-01 3.67959112e-01
2.75746435e-01 6.59448028e-01 2.29200944e-02 3.88790131e-01
-4.62477980e-03 -3.19794685e-01 5.11686206e-01 5.39373040e-01
1.40538454e-01 -6.44792974e-01 -5.14010489e-02 4.95442271e-01
-1.84175909e-01 2.54680365e-01 -4.46148887e-02 5.59069514e-01
-1.96986303e-01 1.19143188e+00 5.45253932e-01 -8.98695946e-01
7.10223675e-01 -3.65806073e-01 -2.19852060e-01 -1.62243530e-01
-5.29167831e-01 -3.82222921e-01 6.30257204e-02 -8.50865126e-01
-3.23889434e-01 -7.22670972e-01 -1.31455529e+00 -7.80013323e-01
-8.63986909e-02 1.35117292e-01 6.06469095e-01 8.06866407e-01
8.72389376e-01 1.02712834e+00 1.05902076e+00 -1.06260729e+00
2.31997922e-01 -8.68492246e-01 -7.32832670e-01 6.00295484e-01
7.68352807e-01 -8.27631414e-01 -4.09662485e-01 -4.29373384e-01] | [6.661051273345947, 2.6201932430267334] |
b6157fb4-7df4-4b18-87bd-aac6fefa8581 | unsupervised-pcfg-induction-for-grounded | null | null | https://aclanthology.org/D12-1040 | https://aclanthology.org/D12-1040.pdf | Unsupervised PCFG Induction for Grounded Language Learning with Highly Ambiguous Supervision | null | ['Raymond Mooney', 'Joohyun Kim'] | 2012-07-01 | null | null | null | emnlp-2012-7 | ['grounded-language-learning'] | ['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] | [-7.135559558868408, 3.5461292266845703] |
acaffd22-8264-454a-8967-272c27d78953 | tabmixer-excavating-label-distribution | 2210.13852 | null | https://arxiv.org/abs/2210.13852v1 | https://arxiv.org/pdf/2210.13852v1.pdf | TabMixer: Excavating Label Distribution Learning with Small-scale Features | Label distribution learning (LDL) differs from multi-label learning which aims at representing the polysemy of instances by transforming single-label values into descriptive degrees. Unfortunately, the feature space of the label distribution dataset is affected by human factors and the inductive bias of the feature extractor causing uncertainty in the feature space. Especially, for datasets with small-scale feature spaces (the feature space dimension $\approx$ the label space), the existing LDL algorithms do not perform well. To address this issue, we seek to model the uncertainty augmentation of the feature space to alleviate the problem in LDL tasks. Specifically, we start with augmenting each feature value in the feature vector of a sample into a vector (sampling on a Gaussian distribution function). Which, the variance parameter of the Gaussian distribution function is learned by using a sub-network, and the mean parameter is filled by this feature value. Then, each feature vector is augmented to a matrix which is fed into a mixer with local attention (\textit{TabMixer}) to extract the latent feature. Finally, the latent feature is squeezed to yield an accurate label distribution via a squeezed network. Extensive experiments verify that our proposed algorithm can be competitive compared to other LDL algorithms on several benchmarks. | ['Xiuyi Jia', 'Zhuoran Zheng', 'Weiyi Cong'] | 2022-10-25 | null | null | null | null | ['multi-label-learning'] | ['methodology'] | [ 3.30514610e-01 2.21490622e-01 -4.55357641e-01 -7.66914725e-01
-7.10171580e-01 -5.37766516e-01 2.88764626e-01 1.76154040e-02
-2.03155905e-01 7.92917013e-01 2.52192199e-01 -5.40532209e-02
-2.33946919e-01 -5.84712505e-01 -7.04518855e-01 -1.12058449e+00
2.89846271e-01 6.38503730e-01 -5.78661680e-01 3.40491384e-01
5.99850342e-02 2.27591828e-01 -1.45273006e+00 1.75897762e-01
8.32455635e-01 1.28500688e+00 -1.47066057e-01 6.67521432e-02
-4.92399931e-01 7.22405434e-01 -6.63312495e-01 -2.06723645e-01
3.66597533e-01 -5.05033731e-01 -7.35872984e-01 2.75444537e-01
4.31611538e-01 -2.66500533e-01 -2.16277286e-01 1.56313276e+00
3.33941996e-01 2.52286673e-01 1.10053515e+00 -1.62991178e+00
-9.23292935e-01 9.20964360e-01 -8.19811642e-01 -1.39890701e-01
1.46541931e-02 -1.43071845e-01 1.19390726e+00 -9.99965370e-01
5.29196560e-01 1.54428697e+00 4.53102261e-01 5.03968537e-01
-1.47393811e+00 -9.39769566e-01 2.76622355e-01 -1.90727964e-01
-1.47243226e+00 -1.42218813e-01 9.98524666e-01 -5.20444870e-01
4.64340210e-01 1.40093192e-01 3.99633944e-01 1.11275876e+00
2.42981106e-01 9.14833248e-01 1.17080700e+00 -8.48445371e-02
2.59560317e-01 3.63905609e-01 3.93072933e-01 6.23461604e-01
2.02055484e-01 -7.75308162e-02 -3.09921265e-01 -2.68757075e-01
6.25281751e-01 4.88483816e-01 1.20395226e-02 -6.23855591e-01
-1.16020298e+00 9.81986403e-01 4.64781165e-01 1.08274154e-01
-2.78246760e-01 2.44657502e-01 2.93729872e-01 4.13793862e-01
5.42216361e-01 4.68086928e-01 -6.13691568e-01 2.14174151e-01
-8.59798551e-01 3.82850468e-01 5.98511577e-01 1.03485966e+00
1.19334173e+00 -1.73107654e-01 -4.58897412e-01 7.83345282e-01
6.77886009e-01 4.82204646e-01 6.80881381e-01 -8.16082418e-01
5.39226830e-01 8.55874479e-01 -3.39256003e-02 -9.85664666e-01
-6.62581742e-01 -5.98858714e-01 -1.06892765e+00 -1.07229635e-01
5.15612721e-01 -2.77300686e-01 -8.22691798e-01 1.97478390e+00
4.20528650e-01 5.11647761e-02 -4.52565067e-02 8.51131141e-01
5.83794475e-01 7.67295539e-01 8.44689831e-02 -4.53605115e-01
1.39679801e+00 -8.38258982e-01 -8.54656458e-01 -2.93286562e-01
6.63624108e-01 -3.84188294e-01 1.11950052e+00 3.90040785e-01
-5.66756189e-01 -4.38770622e-01 -1.01556766e+00 -2.46511120e-02
-3.77122164e-01 2.04236940e-01 5.64234138e-01 3.66833508e-01
-4.66608256e-01 5.58703601e-01 -4.45693493e-01 1.16239525e-01
6.88290179e-01 4.21987832e-01 -4.72535878e-01 -1.28291860e-01
-1.30224228e+00 4.91550118e-01 7.82493532e-01 8.64195004e-02
-7.15973735e-01 -7.72908330e-01 -1.15912104e+00 1.77755579e-01
4.81141537e-01 -5.00439286e-01 9.07511413e-01 -8.03188741e-01
-1.24032760e+00 6.13961756e-01 2.13289931e-02 -4.23745774e-02
2.90879130e-01 1.10034365e-02 -3.40506673e-01 -2.48867601e-01
3.48067284e-01 6.90399289e-01 1.12725258e+00 -1.25337923e+00
-6.13928735e-01 -4.93135810e-01 -7.09040165e-02 2.08292931e-01
-3.26023132e-01 -1.39142498e-01 -1.16687138e-02 -5.68171620e-01
4.40625966e-01 -8.69303942e-01 -1.33247942e-01 -2.27960110e-01
-7.71478415e-01 -5.28482020e-01 6.29859507e-01 -3.28889668e-01
1.39899683e+00 -2.45238256e+00 1.88757598e-01 3.83942127e-01
5.26150823e-01 -3.47804815e-01 -9.33887586e-02 1.57253101e-01
-2.57431865e-01 1.42768905e-01 -2.06000865e-01 -4.98278201e-01
4.43568081e-01 2.61054188e-01 -3.10000539e-01 7.95284688e-01
1.87439561e-01 8.33751261e-01 -1.01668537e+00 -5.22462189e-01
-4.58895825e-02 2.38579780e-01 -5.82695782e-01 2.32586816e-01
-3.87512803e-01 3.69915158e-01 -6.12178326e-01 7.25079894e-01
9.52032804e-01 -4.78982180e-01 7.99885467e-02 -2.77387202e-01
2.63635427e-01 1.24508733e-04 -1.62839258e+00 1.68445230e+00
-2.31215715e-01 3.19133475e-02 -3.04258823e-01 -8.96150231e-01
1.08993840e+00 8.92865006e-03 6.35302842e-01 -1.96694836e-01
3.79071951e-01 5.77507056e-02 -5.59863402e-04 -4.31828707e-01
3.32466155e-01 -4.49506134e-01 -3.55263144e-01 6.30087435e-01
3.47510070e-01 4.42722887e-02 1.11032508e-01 8.15790445e-02
8.29674125e-01 2.27398854e-02 1.37429193e-01 -3.42013061e-01
4.54755962e-01 -6.50811434e-01 8.67293775e-01 5.17094791e-01
-2.11124778e-01 4.32596594e-01 1.00306106e+00 -5.06127715e-01
-8.94064128e-01 -1.12999701e+00 -3.94200712e-01 1.21382225e+00
1.07552253e-01 -5.46867967e-01 -6.06953681e-01 -1.32054758e+00
3.50126266e-01 7.20572770e-01 -9.21532094e-01 -3.82644773e-01
-1.54575244e-01 -9.12426412e-01 2.97790021e-01 5.77592909e-01
2.69826204e-01 -1.06169713e+00 -1.51917047e-03 1.93076193e-01
-4.36177105e-02 -4.52052951e-01 -7.15155005e-01 6.25993967e-01
-5.19981623e-01 -9.07534242e-01 -3.05850744e-01 -6.96136713e-01
1.00026023e+00 -3.05328548e-01 7.86442399e-01 -3.71457875e-01
-8.05883631e-02 -2.16767699e-01 6.80017564e-03 -1.23324916e-01
-2.75073707e-01 1.37246549e-01 1.90256983e-01 3.53234082e-01
7.60622859e-01 -5.32107353e-01 -5.38781404e-01 8.39899778e-02
-1.14288294e+00 -2.02957138e-01 4.97555017e-01 1.23105609e+00
6.73465431e-01 3.05647850e-01 8.46446335e-01 -1.22229385e+00
6.76838696e-01 -8.59277129e-01 -3.93526822e-01 1.18039893e-02
-8.40024352e-01 4.41826075e-01 5.69286108e-01 -6.65938556e-01
-7.78274238e-01 2.65117258e-01 5.85671179e-02 -4.57929254e-01
-1.26698941e-01 5.24207711e-01 -5.70867777e-01 5.39976954e-01
5.25493503e-01 7.45312944e-02 3.04011703e-01 -5.18653214e-01
6.47679925e-01 8.09136748e-01 2.26126313e-01 -5.35631657e-01
6.78525507e-01 1.35343730e-01 4.08712812e-02 -1.22948065e-01
-1.27846479e+00 -4.10798609e-01 -6.98512793e-01 3.60173099e-02
6.57082975e-01 -6.92909181e-01 -6.76331043e-01 4.76167440e-01
-7.15053082e-01 1.43754676e-01 -5.87908924e-01 4.76466835e-01
-5.94406486e-01 8.93959329e-02 -6.11758232e-01 -5.78294933e-01
-1.01967126e-01 -1.28947067e+00 1.13803244e+00 3.66734296e-01
-1.61611587e-01 -9.37730491e-01 1.35667965e-01 1.50276572e-01
1.84117302e-01 2.12746799e-01 1.31529248e+00 -1.05168188e+00
-1.52803257e-01 -2.75720775e-01 -3.39280874e-01 4.91707593e-01
2.87890226e-01 -2.80552328e-01 -1.00406754e+00 -3.61777693e-01
1.68987975e-01 -5.93087912e-01 8.03019583e-01 3.75090778e-01
1.45483136e+00 -4.13483322e-01 -4.30649549e-01 5.67048669e-01
1.28027689e+00 -3.30613069e-02 1.23932242e-01 8.23575631e-02
8.32275689e-01 4.39601541e-01 6.03900373e-01 6.72798574e-01
4.31078613e-01 2.27568567e-01 5.21530211e-01 1.34771034e-01
1.42105699e-01 -3.49953860e-01 8.69254544e-02 6.00961983e-01
6.40927434e-01 -8.66293088e-02 -5.40972054e-01 2.02641532e-01
-1.83267391e+00 -6.17346287e-01 2.76751369e-01 2.06335211e+00
1.09687138e+00 9.50881094e-02 8.65104720e-02 7.83332363e-02
7.84188330e-01 2.68665999e-01 -1.05551076e+00 -4.75211442e-02
-6.30466118e-02 -2.59013206e-01 3.15209687e-01 3.71783376e-01
-1.19160628e+00 7.19405949e-01 5.85166407e+00 8.90690327e-01
-9.81775701e-01 -1.27417192e-01 6.56276584e-01 -9.42837596e-02
-5.74780703e-01 -1.19236737e-01 -1.02830017e+00 7.78971374e-01
6.05703890e-01 -8.40792656e-02 5.01713157e-01 8.85200918e-01
-3.91081393e-01 2.32807174e-01 -1.35581732e+00 1.11996853e+00
1.85718462e-01 -6.98358536e-01 9.96000096e-02 2.75545359e-01
6.47186458e-01 -3.41344714e-01 4.84699577e-01 8.42656255e-01
2.09535345e-01 -1.14270067e+00 5.61081767e-01 6.27470493e-01
9.44547892e-01 -9.49277461e-01 7.61730373e-01 4.46668506e-01
-9.68730032e-01 -4.42322522e-01 -7.26362348e-01 -7.27902027e-03
-1.12160258e-01 9.16734695e-01 -6.77951455e-01 2.10564882e-01
2.13825896e-01 7.00993538e-01 -6.34782970e-01 6.55560195e-01
-4.66038212e-02 3.41746420e-01 -2.70103544e-01 -2.71111019e-02
2.67610013e-01 -4.78046685e-01 1.95680827e-01 9.02499378e-01
3.86763215e-01 -2.07094029e-01 4.44252223e-01 1.10963523e+00
-5.20176530e-01 3.40994328e-01 -6.64153814e-01 -2.37632424e-01
5.26752651e-01 1.51985300e+00 -6.11801445e-01 -4.62736487e-01
-3.58008742e-01 8.10169756e-01 6.65755093e-01 3.01440686e-01
-8.66236389e-01 -4.61291313e-01 5.28488338e-01 -2.42426708e-01
7.60201141e-02 3.81117225e-01 -1.83112800e-01 -1.17410588e+00
-6.24447688e-02 -8.36576939e-01 5.16978145e-01 -4.06599373e-01
-1.73079085e+00 5.10093391e-01 -1.25458449e-01 -1.11790097e+00
-4.38169628e-01 -4.44533378e-01 -1.70330107e-02 9.57386374e-01
-1.13032758e+00 -1.11765039e+00 5.05523831e-02 4.98926967e-01
2.64879644e-01 -3.00800472e-01 8.61322105e-01 3.25052947e-01
-7.28768528e-01 8.28782260e-01 3.68247867e-01 1.04514584e-02
9.92231190e-01 -1.62267160e+00 7.85453990e-03 2.02157021e-01
-3.04374984e-03 7.13752806e-01 5.74101567e-01 -7.19957650e-01
-1.22299898e+00 -1.32094133e+00 8.29068184e-01 -5.11170268e-01
6.76296353e-01 -6.37747705e-01 -8.94109249e-01 9.00584042e-01
-9.34500471e-02 4.58373249e-01 1.00686336e+00 1.72467157e-01
-5.79361022e-01 -1.23933204e-01 -1.47429931e+00 2.44844750e-01
8.52482855e-01 -5.93794763e-01 -6.23478651e-01 3.16271275e-01
8.04511845e-01 -4.69978660e-01 -1.00701654e+00 3.21378082e-01
5.81399739e-01 -4.93105322e-01 7.50852346e-01 -9.46535230e-01
3.72873008e-01 -3.27267706e-01 -2.83416539e-01 -1.71902525e+00
-7.15389490e-01 -2.33868927e-01 -3.71413708e-01 1.49565911e+00
3.82269770e-01 -6.66376173e-01 6.43837035e-01 7.40603626e-01
7.06167668e-02 -1.00351179e+00 -7.84033000e-01 -5.19651771e-01
3.11566710e-01 -1.05030894e-01 1.10158694e+00 1.11260331e+00
1.23093367e-01 4.99365538e-01 -4.83035207e-01 4.70135361e-03
7.11748898e-01 2.63232231e-01 3.08659405e-01 -1.39846098e+00
-1.46448120e-01 -3.24800432e-01 -2.67564297e-01 -1.01363933e+00
5.96543252e-01 -1.24156630e+00 8.21648464e-02 -1.16956973e+00
5.00191867e-01 -5.48643112e-01 -7.27287233e-01 4.78813142e-01
-4.13614690e-01 3.63660716e-02 4.21627089e-02 1.97135713e-02
-7.19690144e-01 6.73829675e-01 1.25618958e+00 -3.92231256e-01
-2.02376515e-01 -3.74029800e-02 -1.08660483e+00 7.72576928e-01
5.96191525e-01 -7.65866756e-01 -5.13804734e-01 -2.89875939e-02
4.41618413e-01 -4.75387387e-02 -4.37507629e-02 -6.26766205e-01
7.83113092e-02 -2.37629786e-01 6.45367920e-01 -6.86462998e-01
2.34094277e-01 -1.04760206e+00 9.18006226e-02 2.14550599e-01
-7.04766870e-01 -2.39037395e-01 -2.56486237e-01 5.65543771e-01
-9.11336541e-02 -3.97906065e-01 6.04866147e-01 8.13182294e-02
-1.19315110e-01 6.19687557e-01 -1.36842772e-01 2.29043588e-01
9.51677918e-01 1.91279516e-01 -4.33768898e-01 6.55678213e-02
-7.54655898e-01 4.85477030e-01 2.69248158e-01 4.18951958e-01
3.54134172e-01 -1.99036026e+00 -5.38824797e-01 7.38468528e-01
3.18813443e-01 2.95070082e-01 3.80650848e-01 5.59184849e-01
8.31365138e-02 1.59987152e-01 -7.64181688e-02 -5.85931897e-01
-6.96620822e-01 9.22552466e-01 3.33077639e-01 -2.51063287e-01
-4.56322938e-01 9.09955204e-01 3.77082586e-01 -7.02273965e-01
3.97426426e-01 -4.86920118e-01 -3.42295974e-01 5.89468420e-01
6.30830765e-01 1.80834651e-01 -2.53752619e-01 -5.49938560e-01
-3.16873282e-01 4.45584059e-01 -2.92460114e-01 6.79497942e-02
1.29082561e+00 -2.74818331e-01 -3.91501904e-01 7.97887325e-01
1.65089560e+00 -1.06647521e-01 -1.32133365e+00 -5.67152441e-01
-2.48724386e-01 -4.94103640e-01 -1.17402591e-01 -7.90349841e-01
-9.88628566e-01 6.14958644e-01 5.56817293e-01 3.43868911e-01
7.91267037e-01 1.48056880e-01 5.07506132e-01 3.70798916e-01
1.05328903e-01 -1.05532622e+00 1.35415226e-01 5.28788626e-01
7.63186812e-01 -1.33656657e+00 -3.20611566e-01 -3.11121996e-02
-6.33965433e-01 9.62313235e-01 7.90888906e-01 -6.07969165e-02
9.70487952e-01 2.90477812e-01 1.41808659e-01 -3.54197085e-01
-4.79081571e-01 1.45406321e-01 2.96089441e-01 3.24197143e-01
2.79306531e-01 1.65670007e-01 -4.29114923e-02 1.13007951e+00
-2.82844573e-01 -2.10002244e-01 2.63462991e-01 5.30417740e-01
-4.23356205e-01 -9.45157647e-01 -3.02526087e-01 8.21755350e-01
-3.64107311e-01 -8.85965824e-02 -7.56554082e-02 4.14849728e-01
4.55211699e-01 6.59311056e-01 6.93214685e-02 -3.38883162e-01
1.50025159e-01 6.20919704e-01 2.88338095e-01 -8.17894936e-01
-2.37566844e-01 1.73944607e-01 -2.79906452e-01 -4.66869891e-01
7.47377872e-02 -8.17970455e-01 -1.43182945e+00 2.52691004e-02
-3.82400274e-01 2.42778748e-01 5.75014889e-01 9.83397722e-01
9.71997306e-02 6.81417406e-01 9.86910939e-01 -3.87010753e-01
-9.98885870e-01 -1.31472290e+00 -1.23548174e+00 6.80840731e-01
4.24901664e-01 -9.76954758e-01 -7.36975908e-01 -2.07255200e-01] | [9.526296615600586, 3.889653205871582] |
475ab367-5177-40ea-9d01-abaeca827ec2 | discrete-time-risk-sensitive-portfolio | 2201.02828 | null | https://arxiv.org/abs/2201.02828v1 | https://arxiv.org/pdf/2201.02828v1.pdf | Discrete-time risk sensitive portfolio optimization with proportional transaction costs | In this paper we consider a discrete-time risk sensitive portfolio optimization over a long time horizon with proportional transaction costs. We show that within the log-return i.i.d. framework the solution to a suitable Bellman equation exists under minimal assumptions and can be used to characterize the optimal strategies for both risk-averse and risk-seeking cases. Moreover, using numerical examples, we show how a Bellman equation analysis can be used to construct or refine optimal trading strategies in the presence of transaction costs. | ['Łukasz Stettner', 'Marcin Pitera'] | 2022-01-08 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-3.06755632e-01 2.57387310e-01 -1.85887173e-01 -1.98146105e-01
-7.71154404e-01 -9.85011816e-01 7.30733946e-02 1.24670900e-01
-4.95613128e-01 8.92034411e-01 -3.60145628e-01 -5.41490316e-01
-7.34910607e-01 -7.85454512e-01 -3.86219025e-01 -7.35396743e-01
-5.44652760e-01 4.59535956e-01 -3.85430001e-04 -9.53527391e-02
5.36805034e-01 6.34336054e-01 -6.31555259e-01 -3.33659321e-01
6.03461504e-01 1.55847502e+00 -4.14656788e-01 5.45087576e-01
2.48811156e-01 6.53620780e-01 -3.95238072e-01 -5.83088815e-01
1.10807443e+00 -2.35382304e-01 -4.73739892e-01 -6.89147040e-02
-2.70616710e-01 -6.31417274e-01 1.44296423e-01 1.23780990e+00
3.28481525e-01 3.06566268e-01 6.04423583e-01 -1.14841902e+00
-2.88326219e-02 4.27903593e-01 -4.34045464e-01 4.05694962e-01
-8.60476643e-02 -2.05943987e-01 1.23455155e+00 -5.62588573e-01
2.44244039e-01 9.28024292e-01 5.00206470e-01 4.07584190e-01
-1.12683284e+00 -2.32949704e-01 3.38735916e-02 -4.74985480e-01
-7.32542634e-01 3.13890912e-02 3.03104430e-01 -5.70037425e-01
8.22718561e-01 5.01679718e-01 6.39010549e-01 1.91985592e-01
7.59198785e-01 4.91583228e-01 1.18152809e+00 -2.09847942e-01
2.90284842e-01 8.06123838e-02 3.83057930e-02 1.49550229e-01
6.71162188e-01 7.10248411e-01 2.71647811e-01 -5.89317501e-01
8.97142529e-01 2.39628613e-01 1.27946185e-02 -4.49085236e-01
-7.10465968e-01 1.15793335e+00 1.81008093e-02 1.19973019e-01
-5.84176958e-01 -2.39583347e-02 1.61200225e-01 1.04910803e+00
5.22037387e-01 5.77392519e-01 -2.93109864e-01 -2.25948561e-02
-5.96394539e-01 5.12513876e-01 1.28917205e+00 6.49078310e-01
-4.20260504e-02 1.53000608e-01 -2.92705595e-01 2.45152295e-01
2.53786892e-01 5.20198345e-01 -2.69430168e-02 -1.35971189e+00
8.16359282e-01 -9.11862552e-02 1.02945065e+00 -5.76321602e-01
-2.61977643e-01 -6.03364706e-01 -2.43453875e-01 6.58901632e-01
6.38902962e-01 -5.14321864e-01 -1.23216053e-02 1.44910038e+00
-1.09654374e-01 -2.41506949e-01 3.48666489e-01 4.26280022e-01
-8.47129822e-01 6.50815904e-01 -6.32005453e-01 -1.14053893e+00
7.88525343e-01 -6.42437637e-01 -7.71788001e-01 1.97356865e-01
4.11690563e-01 -4.58789200e-01 3.91233951e-01 7.49284208e-01
-1.39651346e+00 4.61300910e-01 -9.70674753e-01 6.95041656e-01
1.96710661e-01 -5.66778183e-01 1.78169549e-01 9.51720536e-01
-1.02600026e+00 1.09011531e+00 -6.42916024e-01 3.64984244e-01
2.33876362e-01 3.79467219e-01 4.23129290e-01 4.78969872e-01
-1.02637231e+00 9.13563490e-01 2.37704307e-01 3.06887776e-01
-9.93199289e-01 -6.73142374e-01 -2.47846946e-01 1.82165474e-01
1.00344765e+00 -3.05683136e-01 1.65019143e+00 -8.26760530e-01
-1.53379631e+00 3.25924873e-01 6.76502347e-01 -9.24768567e-01
1.18679380e+00 -2.08067715e-01 1.04914121e-01 2.95515537e-01
-9.48296860e-03 -7.12235093e-01 5.58933020e-01 -5.78954816e-01
-7.73381770e-01 -3.91696036e-01 2.20387593e-01 1.89833656e-01
-2.28404123e-02 3.23958457e-01 6.87882066e-01 -1.08072996e+00
-1.74845293e-01 -1.00294459e+00 -5.40451169e-01 -9.77158397e-02
2.44243275e-02 3.11852306e-01 1.65205821e-01 -6.86373591e-01
1.09997773e+00 -1.87274992e+00 -1.51489824e-01 4.57076609e-01
-3.11160982e-01 -2.93693721e-01 4.34983462e-01 7.02437937e-01
-1.46335527e-01 3.26326162e-01 -4.45766896e-01 6.15701266e-02
4.25593257e-01 -2.01079920e-01 -7.08968103e-01 5.89818895e-01
-2.66046584e-01 7.29862273e-01 -5.03212035e-01 -1.33136250e-02
-2.06735715e-01 -3.34973633e-01 -2.65748799e-01 2.53850251e-01
-7.37466738e-02 -1.30382285e-01 -7.17674136e-01 4.81819212e-01
5.72206140e-01 1.27777547e-01 2.07499802e-01 6.17412269e-01
-5.03731370e-01 -2.49390259e-01 -1.33644223e+00 5.75641513e-01
-5.50158441e-01 3.16563576e-01 4.72924829e-01 -1.13189816e+00
5.68114817e-01 4.16457385e-01 4.64245617e-01 -4.03470039e-01
2.38330901e-01 5.17688334e-01 -2.44280681e-01 -3.56792286e-02
1.98304966e-01 -1.15549290e+00 -3.15873802e-01 1.04703367e+00
-5.94712794e-01 1.38000205e-01 -1.25711188e-01 -1.50731757e-01
8.12195122e-01 -4.73824233e-01 4.69933540e-01 -5.95339358e-01
1.39215946e-01 -2.94767976e-01 6.18510127e-01 6.76056623e-01
-8.88578519e-02 7.63362572e-02 1.17371070e+00 -1.07347488e-01
-7.10750997e-01 -9.75243330e-01 -3.19220483e-01 5.49692690e-01
-2.77970105e-01 5.36741853e-01 -5.01070917e-01 -5.59970737e-01
5.00053406e-01 9.47669685e-01 -6.13670468e-01 9.04367492e-02
-3.18043709e-01 -1.00559926e+00 -1.87211502e-02 3.78089160e-01
3.48630697e-01 -7.66204238e-01 -1.22549164e+00 5.16593099e-01
5.19963801e-01 -4.26810324e-01 -7.10552096e-01 2.43612364e-01
-9.35034037e-01 -1.01572812e+00 -1.07331073e+00 -4.57144417e-02
3.76588672e-01 -8.34387727e-03 7.22519457e-01 -4.54130769e-01
5.73086813e-02 5.86117327e-01 5.21949753e-02 -5.49650311e-01
-1.94593519e-01 -6.82402194e-01 -1.59203455e-01 4.57985818e-01
-3.02253753e-01 -2.17593405e-02 -5.85622132e-01 4.14370716e-01
-7.37554014e-01 -6.19136691e-01 1.07371606e-01 7.75168657e-01
6.02942884e-01 5.29329956e-01 7.45698035e-01 -8.09087038e-01
9.36195493e-01 -4.21810448e-01 -1.66677797e+00 8.31656158e-01
-1.26274872e+00 3.21529865e-01 2.89207518e-01 -2.45437697e-01
-1.50474143e+00 -2.44059429e-01 5.11660635e-01 -1.33468628e-01
8.32695603e-01 5.84905863e-01 -3.25003773e-01 -3.89269978e-01
-1.52245373e-01 -1.75098091e-01 1.78246349e-01 -7.38468230e-01
4.52077426e-02 4.21614796e-01 7.41972551e-02 -5.92529178e-01
6.23853981e-01 3.70125026e-01 4.54979211e-01 -1.51948109e-01
-6.98249280e-01 -1.84187070e-02 -2.37947494e-01 -9.42291468e-02
6.22900665e-01 -3.24517787e-01 -1.09258628e+00 5.75478256e-01
-7.27737844e-01 -3.77222419e-01 -7.07776725e-01 4.73400295e-01
-1.21576941e+00 1.19333893e-01 -7.67759621e-01 -1.55140436e+00
-2.40950137e-01 -6.66933954e-01 1.52025938e-01 1.96354315e-01
2.34946251e-01 -1.48378360e+00 3.23744357e-01 1.24181382e-01
3.57621700e-01 6.98348403e-01 4.87842113e-01 -8.14149618e-01
-8.42219234e-01 -5.89002252e-01 1.93942770e-01 4.93467122e-01
-6.95932806e-02 -1.63147062e-01 -3.65437299e-01 -4.49229896e-01
1.05975235e+00 7.82579854e-02 4.61194962e-01 5.53802371e-01
5.52970529e-01 -1.11258268e+00 -5.86342402e-02 5.65981448e-01
1.68627191e+00 8.02621961e-01 1.50863111e-01 6.59276843e-01
-4.21606079e-02 1.23327374e+00 1.37998021e+00 7.17866182e-01
-8.30941200e-02 4.74608839e-01 2.97044516e-01 6.55722022e-01
1.37445772e+00 -2.52065826e-02 3.00316751e-01 -5.50752431e-02
1.89390555e-01 5.54245077e-02 -7.75940955e-01 4.44431126e-01
-1.94452727e+00 -1.21897018e+00 1.59813091e-01 2.76375484e+00
7.91216850e-01 2.37762377e-01 6.82115972e-01 -3.50904316e-02
6.79958761e-01 -1.98700368e-01 -8.13936830e-01 -7.40157664e-01
-1.71693891e-01 8.19982365e-02 1.20942926e+00 8.00290167e-01
-7.54051268e-01 1.11004181e-01 7.24716425e+00 5.87047935e-01
-7.87870467e-01 1.07973613e-01 9.10119414e-01 -5.00057280e-01
-5.29587984e-01 1.48291171e-01 -6.08990848e-01 6.02812231e-01
1.05352712e+00 -1.08149409e+00 2.00395107e-01 7.41443574e-01
1.42931059e-01 -1.76186323e-01 -8.50301325e-01 5.53054273e-01
-6.49817228e-01 -1.17309618e+00 -8.58876109e-01 7.40033865e-01
7.41068304e-01 -6.28142893e-01 3.54601383e-01 -1.85105547e-01
3.55782449e-01 -8.66787672e-01 7.70139337e-01 7.89142907e-01
2.16462702e-01 -1.32145524e+00 8.88412476e-01 3.15536648e-01
-9.55136657e-01 -4.07479137e-01 -1.02963142e-01 -1.21043578e-01
5.39952338e-01 4.31799412e-01 -2.71822870e-01 6.30833030e-01
1.46953955e-01 2.84170777e-01 3.97018641e-02 1.30940461e+00
1.30342424e-01 1.52980730e-01 -4.78314459e-01 -3.40339392e-02
3.59734684e-01 -8.26671302e-01 4.75517273e-01 6.20471954e-01
6.18706703e-01 3.22660148e-01 -1.87415928e-01 8.90775263e-01
4.40804332e-01 8.00009668e-02 -5.21253347e-01 -1.96043983e-01
9.89041254e-02 7.74519563e-01 -9.22380507e-01 -9.66858566e-02
-3.18660438e-01 8.46679270e-01 -3.02388906e-01 3.52281719e-01
-4.01406407e-01 -6.46240234e-01 4.45046932e-01 -1.15703791e-02
3.98654878e-01 -1.94134098e-02 -5.02839208e-01 -1.02298856e+00
4.69238222e-01 -4.34761524e-01 1.05618811e+00 6.84465468e-02
-1.13151240e+00 1.32649571e-01 3.30537736e-01 -1.04537594e+00
-6.73775971e-01 -6.40756190e-01 -9.06668425e-01 1.00520122e+00
-1.26203156e+00 -9.81399715e-02 7.69858062e-01 3.54875922e-01
-9.99798700e-02 -2.84792393e-01 2.52980739e-01 -3.56478900e-01
-7.59213090e-01 2.53271312e-01 9.07042563e-01 -2.08058849e-01
-1.51691064e-01 -1.67103493e+00 2.38781452e-01 9.75767255e-01
-1.80371448e-01 4.34809566e-01 7.49741793e-01 -7.54257798e-01
-1.12208188e+00 -6.79123044e-01 6.45210803e-01 -9.36869457e-02
1.13366616e+00 1.30428717e-01 -8.60716283e-01 8.01323831e-01
1.60140440e-01 -2.27575619e-02 5.15460253e-01 -3.69454086e-01
9.23546255e-02 -4.35374469e-01 -1.54353321e+00 2.48493597e-01
5.21166921e-01 -3.09514016e-01 -5.86174548e-01 2.38695771e-01
5.93460739e-01 -6.03029467e-02 -1.12562358e+00 1.87860459e-01
6.70032382e-01 -8.13408196e-01 8.62743318e-01 -5.96645951e-01
-5.32730818e-01 3.26699913e-01 -1.19154423e-01 -1.04793942e+00
9.74274427e-02 -1.66913342e+00 1.20230913e-02 9.83624816e-01
2.68326402e-01 -1.40112245e+00 3.84265602e-01 1.19398057e+00
5.66008866e-01 -7.42569268e-01 -1.48361063e+00 -1.55808842e+00
7.00698137e-01 -1.93076227e-02 4.44599509e-01 3.98208648e-01
3.07384193e-01 -4.75220054e-01 -4.38278139e-01 -2.89762229e-01
1.32229757e+00 4.89134729e-01 -1.60114825e-01 -8.76318395e-01
-5.75563729e-01 -6.54787064e-01 2.84505486e-01 -5.27238786e-01
1.71532094e-01 -4.62119758e-01 -4.41021174e-02 -7.90022612e-01
-8.74842703e-02 -4.80960548e-01 -6.76280856e-01 -1.04101226e-01
7.96974227e-02 -4.47183311e-01 5.83701134e-01 1.57437906e-01
-1.12368390e-01 4.73305762e-01 7.59053826e-01 1.82962686e-01
-3.30877863e-02 1.00414455e+00 -7.00749159e-01 5.15361726e-01
1.15680242e+00 -6.14511371e-01 -5.02207875e-01 3.04135621e-01
5.77931285e-01 9.45276916e-01 2.87776023e-01 -1.17671236e-01
-1.81283146e-01 -9.41178739e-01 -5.95779598e-01 -5.78235090e-01
7.67199695e-03 -8.59728158e-01 4.69256192e-01 1.16359520e+00
-4.94699299e-01 2.12190479e-01 -3.39765579e-01 7.97034383e-01
-7.10234568e-02 -1.01636541e+00 8.79809797e-01 -1.97681040e-01
2.25172997e-01 3.86784673e-01 -3.06408107e-01 4.53515679e-01
1.53016198e+00 1.85780257e-01 -9.01367739e-02 -6.58839881e-01
-7.74170041e-01 5.26211202e-01 4.84564096e-01 -1.95248052e-01
4.57779437e-01 -1.26322818e+00 -4.15889949e-01 -1.51803985e-01
-4.46719348e-01 -5.06062925e-01 1.02962270e-01 8.05830538e-01
-5.71181178e-01 6.87396646e-01 -1.69040859e-01 2.00768843e-01
-8.27904642e-01 6.99467421e-01 9.03431237e-01 -5.16760111e-01
-2.79577196e-01 5.94225764e-01 2.35181555e-01 4.61424530e-01
1.64224118e-01 -3.91388535e-01 1.57259762e-01 5.06037593e-01
6.97768748e-01 9.38470423e-01 -2.79624522e-01 -5.61158732e-02
-5.12336753e-02 4.90610719e-01 1.33706197e-01 -1.09280157e+00
1.44091415e+00 -3.96305859e-01 -9.37583521e-02 7.01827049e-01
9.89929140e-01 -1.38165370e-01 -1.62408686e+00 -4.04791050e-02
7.62530088e-01 -8.54667068e-01 -2.22774759e-01 -4.46512878e-01
-1.17464876e+00 8.46751273e-01 1.88969836e-01 8.57069433e-01
8.75251055e-01 -4.88553941e-01 5.33993065e-01 3.22671533e-01
6.45489156e-01 -1.31021750e+00 -2.81445384e-01 -3.11493175e-03
1.18539834e+00 -8.66650939e-01 -1.08300574e-01 -1.58877611e-01
-7.56855130e-01 1.13178265e+00 -1.28612727e-01 -5.83699942e-01
1.17480826e+00 4.06428128e-01 -1.98860243e-01 1.68168575e-01
-8.48145545e-01 2.12974519e-01 1.01157673e-01 2.20738813e-01
-1.01639122e-01 3.92486066e-01 -6.93405449e-01 8.72109413e-01
-5.50633110e-02 -2.00165018e-01 9.91168082e-01 1.27609456e+00
-4.90590602e-01 -1.09377623e+00 -3.43826443e-01 4.29249346e-01
-1.01425314e+00 1.69315606e-01 -4.26894337e-01 6.22673690e-01
-7.38171279e-01 6.55390441e-01 3.35141551e-03 5.03636599e-01
5.80374956e-01 8.40172842e-02 5.47370970e-01 -4.78670239e-01
-6.83045745e-01 3.49349529e-01 -5.48352934e-02 -4.17003483e-01
-9.83961746e-02 -1.10748208e+00 -9.15509760e-01 -3.47913712e-01
-3.01749825e-01 3.38631481e-01 1.49568960e-01 6.60194397e-01
-1.89191788e-01 -2.01059252e-01 1.29490387e+00 -4.14573625e-02
-1.82047153e+00 -4.51424599e-01 -1.32216704e+00 -2.40718573e-01
5.88777542e-01 -4.18461740e-01 -8.10945809e-01 -5.44385910e-01] | [4.898488521575928, 3.919424057006836] |
4e2fd861-2c79-427c-8bba-58a01839b39c | assessing-demographic-bias-transfer-from | 2205.10049 | null | https://arxiv.org/abs/2205.10049v1 | https://arxiv.org/pdf/2205.10049v1.pdf | Assessing Demographic Bias Transfer from Dataset to Model: A Case Study in Facial Expression Recognition | The increasing amount of applications of Artificial Intelligence (AI) has led researchers to study the social impact of these technologies and evaluate their fairness. Unfortunately, current fairness metrics are hard to apply in multi-class multi-demographic classification problems, such as Facial Expression Recognition (FER). We propose a new set of metrics to approach these problems. Of the three metrics proposed, two focus on the representational and stereotypical bias of the dataset, and the third one on the residual bias of the trained model. These metrics combined can potentially be used to study and compare diverse bias mitigation methods. We demonstrate the usefulness of the metrics by applying them to a FER problem based on the popular Affectnet dataset. Like many other datasets for FER, Affectnet is a large Internet-sourced dataset with 291,651 labeled images. Obtaining images from the Internet raises some concerns over the fairness of any system trained on this data and its ability to generalize properly to diverse populations. We first analyze the dataset and some variants, finding substantial racial bias and gender stereotypes. We then extract several subsets with different demographic properties and train a model on each one, observing the amount of residual bias in the different setups. We also provide a second analysis on a different dataset, FER+. | ['Mikel Galar', 'Daniel Paternain', 'Iris Dominguez-Catena'] | 2022-05-20 | null | null | null | null | ['facial-expression-recognition'] | ['computer-vision'] | [ 2.15624705e-01 1.94499224e-01 -3.53639275e-01 -8.18106174e-01
-1.13712654e-01 -3.62189591e-01 7.02144146e-01 1.18367203e-01
-7.64608204e-01 8.58565867e-01 2.51319230e-01 -9.64820385e-03
6.39609620e-03 -6.14786625e-01 -2.68408775e-01 -5.49526095e-01
2.06179589e-01 3.36463541e-01 -2.85421193e-01 -3.79763722e-01
4.89536405e-01 4.89015698e-01 -1.85200381e+00 1.98761165e-01
8.88513327e-01 1.07959664e+00 -6.85488164e-01 3.23078513e-01
4.90932465e-02 8.11808944e-01 -8.54416490e-01 -1.10382509e+00
3.87930870e-01 -4.07647908e-01 -9.35997725e-01 -2.38051265e-01
1.01725030e+00 -2.60087013e-01 1.56879216e-01 1.15440238e+00
6.88872576e-01 -1.16129711e-01 8.24376225e-01 -1.77697837e+00
-4.35339749e-01 5.12195528e-01 -8.70651841e-01 1.56497881e-01
2.83906698e-01 -6.72733337e-02 8.65125597e-01 -5.02251685e-01
7.68309712e-01 1.76541734e+00 6.32983208e-01 7.97863007e-01
-1.15719295e+00 -1.18659353e+00 -8.01209211e-02 7.93791935e-02
-1.14182353e+00 -7.42608070e-01 4.49012488e-01 -5.82124293e-01
1.30870461e-01 3.86146784e-01 3.95563483e-01 1.24403119e+00
-2.23177522e-01 4.23883438e-01 1.79342318e+00 -2.58544415e-01
8.56424198e-02 5.99756360e-01 2.94722080e-01 6.28984690e-01
4.51212168e-01 5.99896722e-02 -5.75867414e-01 -4.97209668e-01
2.37367496e-01 -2.73392111e-01 5.23847267e-02 -2.59002447e-01
-7.14245081e-01 1.04663157e+00 3.80891442e-01 1.77527294e-01
-1.45538837e-01 -5.99635765e-04 5.51770270e-01 4.33566123e-01
8.58954370e-01 5.82237005e-01 -1.88515157e-01 -1.20421387e-01
-9.66363072e-01 5.57514191e-01 7.68321037e-01 4.11494136e-01
8.51127625e-01 -8.35354030e-02 -3.46839368e-01 1.13582063e+00
-2.85919700e-02 5.96770287e-01 3.85191470e-01 -1.16631436e+00
1.58953026e-01 6.30437613e-01 -7.28494674e-02 -1.35922003e+00
-4.13671970e-01 8.07808638e-02 -7.44187951e-01 4.97831225e-01
7.91705608e-01 -4.21994686e-01 -5.80510020e-01 2.06258154e+00
3.06267917e-01 -3.19664538e-01 -3.98191929e-01 8.60184729e-01
8.32627058e-01 1.40913948e-01 4.24451232e-01 -9.10020899e-03
1.14389753e+00 -4.83183622e-01 -3.03055048e-01 -6.38354197e-02
5.71424961e-01 -6.34177566e-01 8.57037783e-01 1.78011715e-01
-9.07413125e-01 -1.37119472e-01 -7.91256309e-01 1.31773561e-01
-5.47481894e-01 -8.97499546e-02 7.21520782e-01 1.25657296e+00
-1.02357447e+00 6.29128397e-01 7.98021108e-02 -5.60120821e-01
8.94327402e-01 4.13727880e-01 -4.41129953e-01 -1.41839415e-01
-1.07282650e+00 1.16972446e+00 -1.84930235e-01 -5.51432788e-01
-4.00613844e-01 -7.79370785e-01 -5.65130293e-01 -5.62743433e-02
1.14958912e-01 -4.28320825e-01 8.90343189e-01 -1.87658679e+00
-1.07925224e+00 1.59758496e+00 -1.74049869e-01 -1.21469527e-01
7.64970779e-01 2.24279493e-01 -3.79271716e-01 -1.69158816e-01
3.05687904e-01 8.07642579e-01 8.12039912e-01 -1.28822339e+00
-5.39574862e-01 -7.23113179e-01 1.71442866e-01 -8.09891224e-02
-5.79397202e-01 5.75780928e-01 3.04684907e-01 -4.96004105e-01
-5.48935831e-01 -1.05457461e+00 9.77811404e-03 1.03587046e-01
-2.62215376e-01 -1.90889701e-01 4.51859802e-01 -5.12206197e-01
1.13431573e+00 -2.07276392e+00 1.06215224e-01 3.52054507e-01
2.62164772e-01 3.08308482e-01 -2.40053013e-01 -6.88517913e-02
-2.96885610e-01 6.18003726e-01 -1.91620380e-01 -2.66127050e-01
-5.64758517e-02 3.83752622e-02 7.05209449e-02 5.35797954e-01
3.07192117e-01 4.86564517e-01 -6.84433937e-01 -6.56940103e-01
-1.85385704e-01 2.08085895e-01 -7.64435887e-01 2.06794068e-02
2.78144658e-01 3.40095639e-01 -2.64596194e-01 6.55616164e-01
8.60282302e-01 2.95314074e-01 1.01478308e-01 -4.49947901e-02
3.97979170e-02 -1.43424854e-01 -8.38463664e-01 9.88978863e-01
-2.34994382e-01 7.10128069e-01 7.41277635e-02 -9.67911720e-01
1.02388239e+00 -8.61271843e-02 5.76490879e-01 -9.13781166e-01
5.01101851e-01 2.43646950e-01 2.64019161e-01 -3.41055393e-01
6.57390237e-01 -5.33465862e-01 -3.15952539e-01 6.32718384e-01
1.69191152e-01 6.17727600e-02 3.10721159e-01 1.75089747e-01
8.06795716e-01 -3.54939610e-01 2.85088420e-01 -6.08771443e-01
4.75354433e-01 -1.45041540e-01 4.67503726e-01 6.59400702e-01
-7.67603695e-01 5.38728476e-01 8.95215631e-01 -5.98115325e-01
-9.73344445e-01 -6.51563585e-01 -2.75645733e-01 1.56930566e+00
2.39056703e-02 -1.41161889e-01 -7.85746336e-01 -9.04617608e-01
4.55556989e-01 6.22666717e-01 -1.08084738e+00 -3.03699136e-01
-1.44387066e-01 -1.11149418e+00 8.64683092e-01 -1.44013949e-02
4.18374658e-01 -8.56100082e-01 -5.38704038e-01 -4.49649751e-01
-1.91864774e-01 -8.89398754e-01 3.08730751e-02 -4.29414272e-01
-4.04202044e-01 -1.30005181e+00 -6.24645472e-01 -7.12610036e-02
4.50030059e-01 -5.88081516e-02 1.60307848e+00 3.82480562e-01
-2.73228049e-01 3.69474649e-01 -2.46255308e-01 -9.64566708e-01
-5.89704514e-01 6.10502623e-02 1.55356705e-01 4.08621311e-01
6.72198653e-01 -3.58600199e-01 -3.80851150e-01 5.15701652e-01
-7.91876197e-01 -2.43833885e-01 3.31324399e-01 6.26462102e-01
-1.47405311e-01 -4.71721292e-01 7.61731982e-01 -1.20823431e+00
6.87312841e-01 -7.66331792e-01 -1.07168391e-01 1.48049116e-01
-6.77395761e-01 -1.58078104e-01 2.84957767e-01 -4.14896160e-01
-1.10295367e+00 -4.38074201e-01 7.41321519e-02 -1.50767446e-01
-1.61242664e-01 1.18553057e-01 -1.23512194e-01 -2.25785941e-01
9.38040018e-01 -6.63951993e-01 2.63469845e-01 -2.35713199e-01
2.67933071e-01 8.25070977e-01 1.90646410e-01 -7.95642018e-01
4.44755018e-01 5.10007083e-01 1.20963261e-01 -8.06132197e-01
-1.00796318e+00 -8.72863270e-03 -4.33713287e-01 -5.61813951e-01
5.30500054e-01 -6.64217710e-01 -8.81892204e-01 4.90447581e-01
-7.96663225e-01 -8.57111812e-02 -2.56473780e-01 9.11532119e-02
-3.18608552e-01 -6.32708818e-02 -2.72216171e-01 -9.31016922e-01
-2.03168228e-01 -8.40315223e-01 9.33549523e-01 2.61851937e-01
-6.41637146e-01 -7.35903144e-01 8.80713314e-02 6.00658476e-01
5.83510280e-01 5.64423621e-01 9.21154678e-01 -6.92970693e-01
2.68730521e-01 -9.31894034e-02 -5.68254173e-01 3.11594427e-01
4.57064398e-02 2.19251022e-01 -1.25378823e+00 -9.94216576e-02
-2.63333946e-01 -7.74467707e-01 1.04912210e+00 2.19922811e-01
1.34678149e+00 -2.79204965e-01 -4.38173413e-02 2.72510409e-01
1.31801009e+00 -1.42218500e-01 6.62542284e-01 3.47972095e-01
7.15628088e-01 1.12400723e+00 5.02001286e-01 5.18477678e-01
5.18334508e-01 5.63164532e-01 4.57608044e-01 -2.89466500e-01
3.89353335e-02 1.53672975e-02 2.20474407e-01 9.57638100e-02
-4.88536835e-01 2.43764417e-03 -1.01290679e+00 3.31968784e-01
-1.64401329e+00 -1.10284424e+00 6.74407259e-02 2.19015741e+00
7.22485125e-01 -2.80139446e-01 6.80895627e-01 7.02914223e-02
9.74273384e-01 3.15423906e-01 -4.49885726e-01 -1.02495384e+00
-4.24898446e-01 3.40170681e-01 4.59798396e-01 2.35100418e-01
-1.08590364e+00 7.34843969e-01 6.98058701e+00 7.94926047e-01
-1.29718864e+00 -2.18029991e-02 1.41062593e+00 -4.06578183e-01
-2.32556060e-01 -2.52294719e-01 -3.70610118e-01 4.48765039e-01
1.03830469e+00 -4.17134345e-01 3.98352623e-01 7.56060719e-01
3.63362171e-02 -3.08803260e-01 -1.08843219e+00 1.04608166e+00
4.14645553e-01 -7.73956418e-01 1.52028203e-01 4.18711454e-02
7.45767117e-01 -1.76352829e-01 1.86346501e-01 2.80116618e-01
3.35230172e-01 -1.49327779e+00 7.43003249e-01 3.73008639e-01
9.71984863e-01 -8.55280161e-01 7.04756141e-01 -1.10862732e-01
-1.52923733e-01 -2.85671532e-01 -4.02475059e-01 -4.82299745e-01
-3.15688074e-01 5.58664322e-01 -3.80539149e-01 2.73940980e-01
9.80224431e-01 6.57821715e-01 -9.71301436e-01 7.37940371e-01
2.77382135e-01 4.48068440e-01 -1.75223798e-01 -1.73865125e-01
-1.85394466e-01 -2.40115866e-01 3.21717978e-01 1.06179976e+00
2.11103231e-01 -1.94522977e-01 -3.11891556e-01 7.72738695e-01
-4.32338774e-01 4.26143885e-01 -7.19428897e-01 -9.18954238e-02
2.73397565e-01 1.51137710e+00 -2.96199471e-01 -3.27721089e-01
-4.35517967e-01 5.79934359e-01 5.54583013e-01 1.03420272e-01
-6.92055821e-01 2.17647925e-02 1.05834579e+00 1.47656858e-01
-4.03264731e-01 4.25307244e-01 -4.83905196e-01 -1.04552317e+00
-2.28957653e-01 -1.34513617e+00 5.29823184e-01 -6.24522448e-01
-1.65194011e+00 3.47856224e-01 1.67746991e-01 -7.19551563e-01
-6.08478226e-02 -6.06983662e-01 -3.54842603e-01 8.64627063e-01
-1.39657271e+00 -1.06497574e+00 -4.48228210e-01 4.80193079e-01
-1.38428271e-01 -3.88107777e-01 7.64849484e-01 5.04088163e-01
-7.47854471e-01 7.68868685e-01 -3.01620603e-01 1.58045650e-01
1.24789894e+00 -1.06916809e+00 -1.92420691e-01 3.80551338e-01
-2.46013656e-01 3.39755386e-01 6.25310361e-01 -2.74640828e-01
-7.08726823e-01 -8.87880981e-01 7.75434256e-01 -6.78555369e-01
4.14355665e-01 -1.62388265e-01 -5.54255962e-01 5.57674646e-01
1.58523500e-01 -1.38891637e-01 9.45430636e-01 3.09152424e-01
-7.15869248e-01 -2.20898002e-01 -1.61162829e+00 5.61728835e-01
1.19514537e+00 -2.41369188e-01 -2.17504099e-01 -4.01336886e-02
4.63306233e-02 -3.59249488e-02 -8.48663330e-01 5.34597099e-01
9.75464046e-01 -1.63415122e+00 6.31094694e-01 -1.05713272e+00
1.07016385e+00 4.04088438e-01 -2.17207074e-01 -1.56617737e+00
-3.10922861e-01 -1.70158148e-01 4.98551071e-01 1.60271144e+00
4.42262530e-01 -7.57586300e-01 8.35839093e-01 1.08631110e+00
6.55461431e-01 -6.82774901e-01 -7.67225742e-01 -4.82416391e-01
6.24688923e-01 -6.42097741e-02 8.51293445e-01 1.38268471e+00
-2.17549056e-01 4.04588610e-01 -5.48458636e-01 -4.83661294e-01
5.82428336e-01 7.67269731e-02 1.07049966e+00 -1.67115116e+00
2.62361705e-01 -8.05209994e-01 -6.02395177e-01 2.62185298e-02
6.36801124e-01 -8.86306405e-01 -4.47558999e-01 -6.70846701e-01
7.12380469e-01 -7.70983875e-01 -1.19634219e-01 4.49379921e-01
-3.07683289e-01 5.55060148e-01 4.27076399e-01 5.56300655e-02
-2.82896578e-01 2.93884158e-01 1.01381302e+00 -1.80793375e-01
2.54293531e-01 -2.15923652e-01 -1.24381542e+00 9.15876865e-01
8.38502645e-01 -5.16885757e-01 -1.06786989e-01 -1.23123139e-01
2.48368070e-01 -4.76286829e-01 4.48858649e-01 -7.25296736e-01
-3.85584354e-01 -4.21473175e-01 5.21718264e-01 3.13982725e-01
2.39990637e-01 -6.43571198e-01 -1.08619168e-01 1.53823763e-01
-4.54285800e-01 3.05715613e-02 4.49435264e-02 4.28591594e-02
-1.52796984e-01 -8.84000435e-02 1.12689638e+00 -1.68772772e-01
-4.98316646e-01 3.36926162e-01 -9.70163941e-02 5.23581862e-01
9.58962679e-01 -7.93906525e-02 -7.30971694e-01 -5.46115935e-01
-3.05262059e-01 8.80041867e-02 7.97912657e-01 6.88243330e-01
1.14602763e-02 -1.29258323e+00 -1.05930495e+00 6.50885627e-02
3.68514717e-01 -7.87478209e-01 -3.91860195e-02 7.42743850e-01
-1.72218621e-01 -1.96028560e-01 -8.91938984e-01 -3.68662357e-01
-1.64320540e+00 3.00430000e-01 5.87321758e-01 5.52214682e-03
3.82615447e-01 6.71273351e-01 2.30797261e-01 -5.29184580e-01
-1.35067999e-01 4.03909683e-01 -4.96129751e-01 5.46068847e-01
4.35898989e-01 6.79692388e-01 -2.01905817e-01 -1.19948721e+00
-4.45034087e-01 3.61763597e-01 2.54316568e-01 -1.66449711e-01
1.27745259e+00 -8.05989802e-02 -5.10096669e-01 3.62306476e-01
1.23446989e+00 1.78053275e-01 -5.75779021e-01 4.37267870e-02
1.19667761e-01 -9.29066122e-01 -2.34777689e-01 -8.21793079e-01
-1.33350301e+00 6.91265821e-01 6.98439240e-01 3.31758440e-01
1.08379173e+00 -2.79874146e-01 6.07873872e-02 -2.55297050e-02
3.20715606e-01 -1.12300754e+00 -7.91462958e-02 3.10381025e-01
8.41488361e-01 -1.47972679e+00 1.37036577e-01 -4.62658763e-01
-8.58665109e-01 7.53467619e-01 7.57640362e-01 -9.15445574e-03
6.31617069e-01 -6.05113171e-02 3.77179325e-01 -2.18605638e-01
-4.74011838e-01 -1.51624128e-01 2.37724274e-01 5.72803736e-01
6.65087461e-01 2.08300561e-01 -6.86619401e-01 4.82553959e-01
-6.24483287e-01 1.03961386e-01 7.05278277e-01 5.53903937e-01
-9.80062131e-03 -1.11940634e+00 -3.67693841e-01 8.62065077e-01
-8.17035437e-01 1.88316688e-01 -1.14087737e+00 7.25495100e-01
3.74570251e-01 9.94903922e-01 1.88537821e-01 -5.10377645e-01
3.81343037e-01 1.45533010e-01 5.42761564e-01 -3.63818675e-01
-7.78975129e-01 -8.74439120e-01 5.75626373e-01 -7.74964809e-01
-9.20834959e-01 -9.22432661e-01 -4.84897494e-01 -9.16056633e-01
2.53019761e-03 -1.49824172e-01 6.66492939e-01 7.45851040e-01
2.88227141e-01 -6.19308613e-02 8.16480756e-01 -8.48529339e-01
-2.57541299e-01 -9.48130608e-01 -7.11679697e-01 1.29462576e+00
1.44812793e-01 -8.93284798e-01 -4.49231148e-01 -3.21547627e-01] | [13.022562026977539, 1.361531376838684] |
cd4c5987-502c-4cba-9267-890f17f00b05 | on-diffusion-modeling-for-anomaly-detection | 2305.18593 | null | https://arxiv.org/abs/2305.18593v1 | https://arxiv.org/pdf/2305.18593v1.pdf | On Diffusion Modeling for Anomaly Detection | Known for their impressive performance in generative modeling, diffusion models are attractive candidates for density-based anomaly detection. This paper investigates different variations of diffusion modeling for unsupervised and semi-supervised anomaly detection. In particular, we find that Denoising Diffusion Probability Models (DDPM) are performant on anomaly detection benchmarks yet computationally expensive. By simplifying DDPM in application to anomaly detection, we are naturally led to an alternative approach called Diffusion Time Probabilistic Model (DTPM). DTPM estimates the posterior distribution over diffusion time for a given input, enabling the identification of anomalies due to their higher posterior density at larger timesteps. We derive an analytical form for this posterior density and leverage a deep neural network to improve inference efficiency. Through empirical evaluations on the ADBench benchmark, we demonstrate that all diffusion-based anomaly detection methods perform competitively. Notably, DTPM achieves orders of magnitude faster inference time than DDPM, while outperforming it on this benchmark. These results establish diffusion-based anomaly detection as an interpretable and scalable alternative to traditional methods and recent deep-learning techniques. | ['Siamak Ravanbakhsh', 'Yashar Hezaveh', 'Vineet Jain', 'Victor Livernoche'] | 2023-05-29 | null | null | null | null | ['supervised-anomaly-detection', 'semi-supervised-anomaly-detection'] | ['computer-vision', 'computer-vision'] | [-2.23599762e-01 -1.67918339e-01 8.29847455e-02 -1.94490626e-01
-6.80685461e-01 -3.84433001e-01 1.04515338e+00 5.90776384e-01
-2.69378036e-01 1.64118335e-01 2.81220436e-01 -4.94907618e-01
-1.40945956e-01 -9.66435254e-01 -5.11744916e-01 -6.15351439e-01
-5.07524908e-01 8.68437409e-01 2.71566838e-01 2.74605453e-01
4.12835449e-01 6.78367376e-01 -1.11762464e+00 -1.43435284e-01
8.83517742e-01 1.15095770e+00 -7.34122455e-01 9.10321295e-01
-3.99129063e-01 9.12494838e-01 -7.85281599e-01 -3.96376282e-01
1.62666291e-01 -1.44843385e-01 -6.07144773e-01 7.20242485e-02
3.57954681e-01 -8.34288955e-01 -6.29968226e-01 9.26311255e-01
2.46510357e-01 2.05923796e-01 1.06617200e+00 -1.51963496e+00
-6.85657561e-01 4.99702662e-01 -9.09246027e-01 1.06919253e+00
8.30612928e-02 1.93608820e-01 9.90830600e-01 -8.78095746e-01
2.54677802e-01 1.28428006e+00 8.14865112e-01 3.20686579e-01
-1.42722845e+00 -3.22035104e-01 3.29906225e-01 2.10004047e-01
-1.30391431e+00 -3.42577428e-01 6.79387987e-01 -4.65400487e-01
1.17350709e+00 1.54147670e-01 3.84359121e-01 1.47916639e+00
4.24208611e-01 1.21787584e+00 7.38009632e-01 6.40772358e-02
5.98522365e-01 -5.57144463e-01 1.40596747e-01 6.69951677e-01
3.15311104e-01 -3.12056635e-02 -6.88667715e-01 -8.17811906e-01
6.35461271e-01 3.95355076e-01 1.17882915e-01 -1.66556895e-01
-8.91094148e-01 8.26688766e-01 -9.94041860e-02 5.79601852e-03
-6.01594746e-01 5.86489499e-01 6.72521710e-01 5.46117246e-01
1.15458012e+00 1.08950801e-01 -1.12063631e-01 -6.23717666e-01
-1.07000160e+00 3.37687701e-01 9.01450098e-01 5.03408074e-01
3.72697413e-01 5.94071090e-01 -4.42740142e-01 5.69887877e-01
6.94854617e-01 4.97005135e-01 3.52015376e-01 -9.43483949e-01
1.52226642e-01 4.88204181e-01 -1.84250381e-02 -9.47384119e-01
-1.46368191e-01 -4.21653360e-01 -1.02809227e+00 1.17195204e-01
5.64251482e-01 -1.44525081e-01 -1.06746340e+00 1.34445822e+00
2.90083140e-01 6.78079128e-01 7.36087039e-02 3.55684698e-01
8.90262052e-02 7.93404400e-01 1.72437578e-01 4.28412594e-02
8.16444337e-01 -5.43699086e-01 -4.23972309e-01 -1.88031211e-01
7.51334190e-01 -4.42222208e-01 6.78041279e-01 6.42925680e-01
-8.51360023e-01 1.69800341e-01 -6.71278179e-01 3.34145635e-01
-1.27374098e-01 -4.63936627e-01 6.72505915e-01 6.71995342e-01
-1.10120785e+00 9.04183507e-01 -1.60910416e+00 -4.03798968e-01
7.49259591e-01 -1.97717454e-02 9.16769579e-02 6.64076656e-02
-6.57473981e-01 3.85057181e-01 4.15454432e-02 -2.27698341e-01
-1.37987220e+00 -8.08527172e-01 -8.19969237e-01 1.00717098e-01
4.07391712e-02 -3.53436649e-01 1.32973897e+00 -4.81346875e-01
-1.23380256e+00 6.97106779e-01 -2.59728730e-01 -1.25874603e+00
6.14427924e-01 -4.42108333e-01 -5.88510096e-01 2.54133105e-01
9.89273340e-02 1.61969319e-01 1.39040494e+00 -7.81005621e-01
-5.72506070e-01 -2.76487470e-01 -4.66921508e-01 -2.73964822e-01
-6.13693595e-01 -4.04028979e-04 -2.37480193e-01 -9.16535735e-01
2.43848339e-01 -5.55690467e-01 -3.34848970e-01 1.90949231e-01
-6.45505846e-01 -4.67807949e-01 1.33215392e+00 -5.34742773e-01
1.36954677e+00 -2.20510721e+00 -1.05479784e-01 7.19542027e-01
8.02265644e-01 8.77740011e-02 2.08104432e-01 4.64871347e-01
2.72506505e-01 3.29694524e-02 -7.27202058e-01 -9.79599059e-01
3.11435163e-01 6.53372526e-01 -7.27323174e-01 7.98837841e-01
5.10154784e-01 7.94186950e-01 -9.80391979e-01 -2.42846772e-01
6.44229427e-02 2.71942109e-01 -6.45337820e-01 1.03479140e-01
-3.70227277e-01 5.39583206e-01 -4.45154607e-01 1.00892413e+00
5.64936638e-01 -3.38564485e-01 -3.19392622e-01 5.66135824e-01
2.60495007e-01 -1.57274101e-02 -9.43184972e-01 1.44480681e+00
-2.13201523e-01 9.05234933e-01 -1.42926827e-01 -9.44393754e-01
8.95407856e-01 1.81190223e-01 6.56026900e-01 -3.43571693e-01
-1.79248139e-01 2.59730756e-01 -1.54739663e-01 -2.36477312e-02
4.62775707e-01 2.20339015e-01 8.04663599e-02 9.97766078e-01
6.67966157e-02 1.91943288e-01 2.63502568e-01 5.66871047e-01
1.84880650e+00 -2.01660767e-01 -1.24154016e-01 -2.91193783e-01
9.24332514e-02 -2.86579072e-01 3.33885252e-01 1.23618186e+00
-3.53574783e-01 5.22655547e-01 7.93201625e-01 -6.14616394e-01
-1.02847016e+00 -1.65669525e+00 -2.11765617e-01 9.11475182e-01
-3.26769710e-01 -6.45586789e-01 -5.02360404e-01 -8.62897813e-01
2.83718169e-01 8.96770775e-01 -5.60555696e-01 -3.12976092e-01
-3.76449883e-01 -1.19986272e+00 1.00844765e+00 8.36767912e-01
3.84852618e-01 -7.39808977e-01 -2.61007369e-01 3.62650305e-01
1.80715114e-01 -8.47904325e-01 -2.03286573e-01 7.06671700e-02
-1.12037599e+00 -6.84229434e-01 -5.82968235e-01 8.62460807e-02
5.32082796e-01 -3.39917988e-01 1.19754148e+00 2.35963054e-02
-2.39710793e-01 1.02806067e+00 -1.67915240e-01 -6.42922521e-01
-7.74163067e-01 -1.10801078e-01 2.99192131e-01 2.24128783e-01
8.31666052e-01 -1.03323436e+00 -7.60091722e-01 5.85786775e-02
-1.10628557e+00 -7.83132732e-01 4.07573313e-01 4.59266812e-01
6.06361866e-01 6.02262057e-02 3.93845528e-01 -6.64884508e-01
9.65244949e-01 -1.06597733e+00 -7.16457188e-01 -4.33632255e-01
-1.04396033e+00 1.81208462e-01 3.00384313e-01 -3.49276185e-01
-8.70432079e-01 -4.65520203e-01 -4.26260054e-01 -7.75209665e-01
-3.92570794e-01 4.11247969e-01 5.16761303e-01 1.37251794e-01
6.24286234e-01 5.12477338e-01 5.98683320e-02 -5.40471375e-01
1.14298634e-01 2.58574665e-01 5.81011236e-01 -8.64549994e-01
8.82706642e-01 9.85839248e-01 6.24426566e-02 -9.24470842e-01
-7.22959220e-01 -5.14355838e-01 -2.32262149e-01 7.01978728e-02
6.88496649e-01 -7.35584676e-01 -3.25681597e-01 8.08877528e-01
-9.34794068e-01 -3.38330925e-01 -6.40905619e-01 1.61767259e-01
-3.06681246e-01 8.86676550e-01 -1.08946621e+00 -1.07179046e+00
-4.65971947e-01 -6.74896121e-01 1.31766462e+00 -2.42638558e-01
-5.53243339e-01 -1.47411454e+00 4.41443920e-01 -3.33431125e-01
7.82180786e-01 2.63346583e-01 8.56768191e-01 -1.22205162e+00
-5.57398200e-01 -5.32747328e-01 -1.91793799e-01 4.88812327e-01
-1.69640303e-01 2.95900553e-01 -9.86486077e-01 -1.89466491e-01
2.58321967e-02 2.17375174e-01 1.11350358e+00 4.15405273e-01
1.50343370e+00 -2.99130470e-01 -3.53590667e-01 6.85508490e-01
1.04463065e+00 -2.56243050e-01 6.34786963e-01 4.57521230e-01
8.26093316e-01 -4.11110818e-02 9.57256779e-02 8.97649825e-01
2.70331115e-01 1.07824832e-01 5.50029814e-01 2.16696978e-01
2.24757478e-01 -1.42541692e-01 6.28925800e-01 5.85199356e-01
8.98672417e-02 -5.14960110e-01 -1.34720886e+00 8.16667974e-01
-1.88748431e+00 -1.12703955e+00 -3.68317068e-01 2.20385122e+00
4.78617907e-01 4.92949456e-01 3.29147100e-01 2.90392101e-01
3.51385176e-01 3.00396562e-01 -7.52585411e-01 -5.37546277e-01
-1.48878798e-01 2.68765092e-01 4.12510663e-01 2.33205497e-01
-1.29013431e+00 5.34747779e-01 7.23858547e+00 8.54610622e-01
-6.59273922e-01 2.10465476e-01 7.75386930e-01 -1.00526862e-01
-2.67576188e-01 -2.25111350e-01 -6.58862114e-01 6.89927876e-01
1.40177679e+00 -2.32925028e-01 2.34626811e-02 9.27845597e-01
-5.85006513e-02 -1.28727853e-01 -1.18861389e+00 7.67811596e-01
-1.92977294e-01 -1.21289372e+00 2.74940848e-01 3.72366935e-01
8.36638629e-01 6.42397523e-01 3.09577614e-01 2.48204246e-01
6.23215735e-01 -8.73809934e-01 4.05913949e-01 6.94449663e-01
2.13049911e-02 -9.17859435e-01 6.70754671e-01 2.62232870e-01
-8.91865313e-01 -1.67341039e-01 -3.88989389e-01 -6.12777285e-02
1.86746687e-01 1.13902140e+00 -6.31525338e-01 1.56619206e-01
8.13936114e-01 8.61588240e-01 -5.02857149e-01 1.06500745e+00
-5.64393736e-02 1.26877296e+00 -8.13290477e-01 4.51160312e-01
4.65246797e-01 -1.70442507e-01 1.22123790e+00 1.31689394e+00
5.87627590e-01 -5.70609629e-01 6.44879341e-02 1.13432574e+00
-6.20862395e-02 -2.83267587e-01 -8.24169934e-01 -4.60017323e-01
4.06978667e-01 9.82046247e-01 -8.37548852e-01 -4.36576486e-01
-4.04440761e-01 1.30976939e+00 1.89129218e-01 4.68449980e-01
-7.28692234e-01 4.49498324e-03 1.04012299e+00 8.99529681e-02
3.71639460e-01 -4.72753435e-01 -4.23127748e-02 -1.16652799e+00
-1.09953053e-01 -6.71361744e-01 7.12149620e-01 -1.54576555e-01
-2.03618336e+00 3.31799090e-01 6.91925585e-02 -9.56316531e-01
-5.23869812e-01 -5.32758534e-01 -1.14863431e+00 6.93510175e-01
-1.24643743e+00 -7.24961340e-01 3.22098099e-02 5.40537596e-01
5.47585785e-01 -1.90982640e-01 7.68269181e-01 1.96498722e-01
-7.66983628e-01 3.86488289e-01 5.16347945e-01 2.63726324e-01
3.08241308e-01 -1.86954045e+00 1.26968575e+00 1.38494051e+00
5.30286729e-01 4.23251629e-01 8.04722428e-01 -8.94218206e-01
-1.30483949e+00 -1.21288633e+00 2.21056461e-01 -1.01150894e+00
1.18428946e+00 -4.36369218e-02 -1.31293750e+00 8.45357120e-01
2.54261270e-02 1.03894137e-01 6.61394358e-01 2.02577233e-01
-3.76744360e-01 2.63812453e-01 -1.12208951e+00 5.45854509e-01
9.23317432e-01 -6.84982240e-01 -3.53960693e-01 5.05381882e-01
3.58423799e-01 -3.38628262e-01 -9.94704723e-01 1.83480918e-01
-3.23754549e-02 -1.12025833e+00 9.84335423e-01 -5.91987908e-01
4.06945705e-01 -1.17634609e-01 -1.76286586e-02 -1.19940102e+00
-9.89623293e-02 -9.61882055e-01 -1.34303379e+00 1.18416703e+00
2.61292070e-01 -8.73948693e-01 8.53994906e-01 5.33343434e-01
-1.48085177e-01 -6.94619000e-01 -1.07235515e+00 -9.98034775e-01
1.20793603e-01 -1.05169678e+00 4.29608166e-01 8.52813303e-01
-3.66018355e-01 -2.23494574e-01 -1.81906074e-01 5.06070197e-01
1.00151443e+00 -2.19918758e-01 6.20025516e-01 -1.50068820e+00
-3.54086876e-01 -6.62975192e-01 -7.86726117e-01 -1.18556273e+00
1.74584702e-01 -6.93714917e-01 -2.79095858e-01 -1.40980744e+00
-2.04247490e-01 -1.13903530e-01 -5.06218135e-01 2.29735717e-01
-8.25942904e-02 2.48878896e-01 -7.42709339e-01 4.63240087e-01
-7.50039101e-01 5.98373771e-01 4.33758765e-01 -6.22416064e-02
7.12602027e-03 1.93768963e-01 -3.14792395e-01 1.03090024e+00
8.78414989e-01 -6.32796288e-01 -3.08791816e-01 -5.22935510e-01
2.40472034e-01 -5.29578269e-01 5.54069996e-01 -1.14613032e+00
3.01090807e-01 2.11837754e-01 4.49752390e-01 -7.37646699e-01
2.43790388e-01 -4.33604538e-01 -4.63003784e-01 3.82494211e-01
-4.00417633e-02 6.40620708e-01 3.61655802e-01 1.33434331e+00
-1.17039092e-01 2.31511220e-02 4.48211312e-01 1.44542783e-01
-9.37629342e-01 6.49366856e-01 -1.05706525e+00 3.26149851e-01
7.56725132e-01 -8.15723240e-02 -2.39742354e-01 -7.78382480e-01
-7.83658624e-01 9.35320258e-02 3.76769036e-01 1.32950529e-01
7.11939275e-01 -1.14123714e+00 -7.13766634e-01 2.50360727e-01
1.48346156e-01 8.76050815e-02 1.20988943e-01 1.18569434e+00
-6.50332868e-01 -1.52888909e-01 1.67512700e-01 -9.71727014e-01
-6.95158184e-01 3.48434389e-01 2.58857429e-01 -3.16336840e-01
-1.05023038e+00 1.08221745e+00 -1.09272465e-01 8.45460873e-03
3.97672266e-01 -3.36973876e-01 4.05837446e-01 -2.48556614e-01
5.72403848e-01 6.06715024e-01 -1.23347156e-02 -2.09688574e-01
-2.32939348e-01 -4.01312932e-02 -4.26800519e-01 -1.25608593e-01
1.10633898e+00 -6.74624294e-02 -1.49155080e-01 5.61385512e-01
7.31806338e-01 -5.03143109e-02 -1.37073386e+00 -5.17403066e-01
5.00766098e-01 -4.05666828e-01 2.21494794e-01 -3.00942242e-01
-9.75687742e-01 8.87145340e-01 5.92916250e-01 7.64726996e-01
9.12638009e-01 2.91112643e-02 9.71181273e-01 6.00198388e-01
-1.89670935e-01 -9.89841938e-01 2.22952619e-01 6.15985215e-01
4.36692685e-01 -1.15025425e+00 -6.01372868e-02 9.01385695e-02
-3.28255743e-01 8.71973693e-01 5.28067827e-01 -4.25375760e-01
7.59576082e-01 4.72881019e-01 -4.60232235e-02 -5.88427126e-01
-7.68604934e-01 7.15926141e-02 3.21275800e-01 5.71421981e-01
9.94338244e-02 -2.69345909e-01 4.39273596e-01 6.26115203e-02
1.03081666e-01 -6.68532550e-01 2.87482738e-01 1.07590365e+00
-4.20226514e-01 -9.09105957e-01 -1.83862850e-01 9.68610227e-01
-7.46353447e-01 -2.30758324e-01 -3.12364966e-01 4.41170990e-01
-5.71520746e-01 7.60273874e-01 7.72295058e-01 7.55906850e-02
-9.42437127e-02 3.75506282e-01 1.73909619e-01 -4.38659817e-01
-1.62584156e-01 -1.54793352e-01 -1.46801591e-01 -8.19293678e-01
2.86444053e-02 -7.40726948e-01 -1.06886292e+00 -7.21246898e-01
3.86778004e-02 1.10669740e-01 4.88269210e-01 1.18189359e+00
5.81977725e-01 4.73299831e-01 9.00745541e-02 -5.59455931e-01
-8.20147514e-01 -9.08040762e-01 -6.56357110e-01 3.75148803e-01
3.93088788e-01 -5.20544291e-01 -6.66959524e-01 -3.48679096e-01] | [7.619528293609619, 2.4099254608154297] |
f76289ac-4a79-434c-8152-7fe245172e05 | preventing-errors-in-person-detection-a-part | 2307.04533 | null | https://arxiv.org/abs/2307.04533v1 | https://arxiv.org/pdf/2307.04533v1.pdf | Preventing Errors in Person Detection: A Part-Based Self-Monitoring Framework | The ability to detect learned objects regardless of their appearance is crucial for autonomous systems in real-world applications. Especially for detecting humans, which is often a fundamental task in safety-critical applications, it is vital to prevent errors. To address this challenge, we propose a self-monitoring framework that allows for the perception system to perform plausibility checks at runtime. We show that by incorporating an additional component for detecting human body parts, we are able to significantly reduce the number of missed human detections by factors of up to 9 when compared to a baseline setup, which was trained only on holistic person objects. Additionally, we found that training a model jointly on humans and their body parts leads to a substantial reduction in false positive detections by up to 50% compared to training on humans alone. We performed comprehensive experiments on the publicly available datasets DensePose and Pascal VOC in order to demonstrate the effectiveness of our framework. Code is available at https://github.com/ FraunhoferIKS/smf-object-detection. | ['Stephan Günnemann', 'Karsten Roscher', 'Andrea Matic', 'Franziska Schwaiger'] | 2023-07-10 | null | null | null | null | ['object-detection', 'human-detection'] | ['computer-vision', 'computer-vision'] | [ 2.15198755e-01 2.40701124e-01 4.28149134e-01 -2.32598275e-01
-4.45146233e-01 -3.63295019e-01 4.53151584e-01 3.76605541e-01
-6.77886426e-01 3.44941586e-01 -2.75405586e-01 2.80577373e-02
4.08762962e-01 -6.23360395e-01 -8.51428390e-01 -3.33543330e-01
-1.50306374e-01 3.89692396e-01 7.74535120e-01 -6.14837669e-02
-8.29258040e-02 4.56642538e-01 -2.11730146e+00 1.65343642e-01
4.31170017e-01 9.48612750e-01 1.62189052e-01 8.41642559e-01
6.79950416e-01 5.21374762e-01 -6.28170609e-01 -3.61783981e-01
5.23112237e-01 -2.61630952e-01 -4.36649770e-01 2.58740306e-01
8.24877024e-01 -5.80992281e-01 -2.23304972e-01 9.86051559e-01
3.96010190e-01 1.87693626e-01 2.97497571e-01 -1.27743423e+00
1.50893286e-01 2.35284436e-02 -5.88156760e-01 1.48908868e-01
4.58364248e-01 4.76806670e-01 7.70921767e-01 -6.05974615e-01
2.60807872e-01 1.18733907e+00 5.79038620e-01 6.85021281e-01
-1.16447890e+00 -5.11423647e-01 1.90266117e-01 2.24619493e-01
-1.28074408e+00 -7.25116730e-01 3.19474280e-01 -4.38824087e-01
8.53582501e-01 2.32428744e-01 6.37655318e-01 1.02600789e+00
-6.65250793e-02 6.98775709e-01 9.17238653e-01 -5.40870130e-01
1.07477278e-01 2.70557523e-01 1.07006721e-01 9.11193967e-01
7.69681215e-01 2.62703687e-01 -4.75242168e-01 5.81089109e-02
6.35448337e-01 -5.57427183e-02 -5.68891726e-02 -5.69780469e-01
-1.28836215e+00 4.52307343e-01 4.98719126e-01 -3.95544134e-02
-4.48837280e-01 2.84523785e-01 3.22081000e-01 -1.46575779e-01
3.96931209e-02 2.41763070e-01 -1.15390807e-01 -8.63274559e-02
-5.82968950e-01 3.23610067e-01 5.97357929e-01 6.90228760e-01
3.39845061e-01 -2.70582527e-01 -1.72807977e-01 5.27012348e-01
2.75485069e-02 6.48671567e-01 4.81807068e-02 -9.37985778e-01
1.00485004e-01 6.88872278e-01 5.73124766e-01 -8.59237254e-01
-4.95405793e-01 -4.62686777e-01 -6.06273711e-01 5.63746154e-01
4.44245398e-01 -1.30759761e-01 -8.15968335e-01 1.63922679e+00
8.56878996e-01 -9.59173143e-02 -2.56830722e-01 1.15701699e+00
3.88253778e-01 2.13062346e-01 5.60609326e-02 1.11936435e-01
1.62660098e+00 -7.88104355e-01 -4.06249642e-01 -5.98215163e-01
3.54461730e-01 -6.84725165e-01 1.00715804e+00 6.26752138e-01
-9.07973886e-01 -8.00880969e-01 -1.16221809e+00 4.89999279e-02
-7.59227425e-02 4.76211011e-01 4.86445516e-01 7.00402558e-01
-7.22726047e-01 5.04225314e-01 -1.04995358e+00 -7.29017198e-01
3.40197831e-01 3.96680117e-01 -5.23421764e-01 9.75060165e-02
-6.74914002e-01 1.01129711e+00 5.02598107e-01 5.76036721e-02
-9.86275852e-01 -2.52094775e-01 -9.63600993e-01 -3.43869850e-02
7.91931748e-01 -7.90012538e-01 1.49894428e+00 -7.10638463e-01
-1.04650795e+00 8.67846906e-01 -9.54781100e-02 -8.08290541e-01
1.04527426e+00 -6.24690175e-01 -1.40257791e-01 2.04693317e-01
3.69100645e-02 7.74968207e-01 8.42134953e-01 -9.71617877e-01
-1.01921129e+00 -3.21545064e-01 2.59197623e-01 1.22805856e-01
-2.23705694e-01 2.34767143e-02 -5.27992070e-01 -3.04502219e-01
-2.14997828e-01 -1.22246277e+00 -1.23429276e-01 3.89689684e-01
-3.88236970e-01 -6.49425834e-02 6.76202357e-01 -6.49641633e-01
6.04066432e-01 -2.11785531e+00 -1.44824609e-01 -2.57821288e-03
2.44638219e-01 4.87806410e-01 9.91943032e-02 1.44761726e-01
2.61025637e-01 -4.30333853e-01 -1.77964792e-01 -5.53908288e-01
-1.22003406e-01 -6.57054409e-03 4.50537838e-02 5.98821223e-01
2.66440481e-01 4.89866763e-01 -8.18777204e-01 -3.62302214e-01
6.16972864e-01 4.31981206e-01 -2.59304971e-01 2.04423085e-01
-1.90710917e-01 4.27893579e-01 3.09588928e-02 5.76484442e-01
3.50983441e-01 -2.75007576e-01 1.31067693e-01 -5.25171012e-02
7.06080124e-02 2.06600502e-01 -1.38394117e+00 1.18846202e+00
-2.29487374e-01 4.35691357e-01 1.55288145e-01 -5.89352190e-01
4.58480835e-01 2.71729469e-01 2.08950430e-01 -5.93677998e-01
2.30733335e-01 -2.76587885e-02 1.98862612e-01 -8.67232457e-02
4.66142267e-01 3.73199023e-02 6.17537415e-03 1.55922100e-01
-4.44222502e-02 3.32890391e-01 4.37998861e-01 2.45035380e-01
1.36919379e+00 1.91332325e-02 3.50855500e-01 9.59903151e-02
5.04514635e-01 1.24556914e-01 5.84741235e-01 7.22060859e-01
-5.00314832e-01 4.42945957e-01 2.10779861e-01 -5.31164229e-01
-9.40243661e-01 -1.04855847e+00 7.59616047e-02 9.14881349e-01
4.24575806e-01 -4.86601979e-01 -9.21588004e-01 -7.68143237e-01
1.18550621e-01 7.02475369e-01 -5.79013467e-01 -1.21965125e-01
-3.79519552e-01 -4.05755579e-01 3.99983734e-01 7.09563851e-01
5.79386175e-01 -9.75238442e-01 -1.46331334e+00 3.21562178e-02
-1.75173417e-01 -1.47193503e+00 -1.93155870e-01 -9.57081914e-02
-5.21539688e-01 -1.43304515e+00 -4.77630496e-01 -2.74512976e-01
6.49613380e-01 3.70749921e-01 8.46440673e-01 3.31202269e-01
-9.62392271e-01 6.67889416e-01 -1.88698769e-01 -6.38515353e-01
-3.86204809e-01 -2.01843813e-01 3.00935984e-01 -6.79931790e-02
2.02328399e-01 -1.69081837e-01 -8.44931424e-01 5.13953745e-01
-4.87840056e-01 1.62415534e-01 6.39123797e-01 4.45705652e-01
4.76037472e-01 1.73980016e-02 2.02352419e-01 -6.70258582e-01
1.37634128e-01 4.93704400e-04 -1.04951906e+00 1.07894115e-01
-3.36396426e-01 -1.79347277e-01 1.73273966e-01 -2.49788344e-01
-9.31532741e-01 5.90056360e-01 -2.33651940e-02 -2.35269904e-01
-5.04325688e-01 -3.38094771e-01 7.19053596e-02 -1.25163764e-01
7.66301215e-01 2.22915667e-04 1.11709662e-01 -2.57278383e-01
2.73337394e-01 4.14542973e-01 7.78187633e-01 -2.91940063e-01
7.40329146e-01 6.64714336e-01 1.67639986e-01 -7.99923122e-01
-9.16656852e-01 -8.70979726e-01 -6.32049143e-01 -5.16320467e-01
8.61446202e-01 -1.00574470e+00 -9.61571574e-01 3.26371044e-01
-1.05421531e+00 -3.47016990e-01 -1.25460446e-01 3.48028600e-01
-4.60036218e-01 4.28033769e-01 -4.53223646e-01 -1.10909021e+00
-3.37974519e-01 -7.71821737e-01 1.13704455e+00 1.90684155e-01
-3.12101692e-01 -3.74787837e-01 -1.57884881e-01 5.13502061e-01
1.38603151e-01 4.50987846e-01 1.14553176e-01 -4.58515435e-01
-5.43501854e-01 -5.16620636e-01 -1.04413018e-01 3.67502689e-01
9.36246440e-02 -2.53821045e-01 -1.02905703e+00 -4.69775945e-01
-2.82749206e-01 -3.01654994e-01 9.43305612e-01 9.15201157e-02
7.28207409e-01 -1.28459126e-01 -4.94843602e-01 -4.85026510e-03
9.77949977e-01 -3.06573033e-01 4.44176406e-01 2.14538604e-01
5.54899812e-01 6.66837871e-01 8.70524704e-01 4.74197924e-01
3.54009420e-01 8.35432649e-01 5.75770736e-01 -9.34570283e-02
-3.86991560e-01 -2.48284042e-02 3.67327213e-01 -1.06753521e-01
-2.70397812e-01 -6.94528818e-02 -8.77122879e-01 6.26080215e-01
-1.98899925e+00 -1.10244358e+00 -2.20801756e-01 2.72657776e+00
3.99489909e-01 5.57585716e-01 6.04837000e-01 2.72636920e-01
8.18553150e-01 -3.51824015e-01 -5.16192317e-01 2.10336059e-01
4.08734471e-01 -2.85186805e-02 6.12672329e-01 2.69708693e-01
-1.38212335e+00 7.78689861e-01 5.67292881e+00 2.85992384e-01
-9.05445516e-01 1.37497216e-01 2.36493051e-01 -3.74223143e-01
6.94499850e-01 -2.69453794e-01 -9.85905409e-01 3.99727225e-01
8.38311017e-01 6.12097569e-02 1.86491281e-01 1.01280904e+00
7.94893876e-02 -5.04246712e-01 -1.19185841e+00 8.87725949e-01
6.25967905e-02 -6.10402822e-01 -4.60338145e-01 5.07225879e-02
3.49950910e-01 1.53146358e-02 -1.84911445e-01 1.75133422e-01
2.02869952e-01 -7.14666545e-01 9.11192119e-01 2.52202630e-01
4.07325357e-01 -6.27431750e-01 6.83956742e-01 6.67576849e-01
-1.14457500e+00 -1.67135149e-02 -2.98235267e-01 -2.57595241e-01
2.12644875e-01 6.84801579e-01 -1.15764129e+00 3.22456688e-01
8.91249359e-01 2.41803303e-01 -7.96265185e-01 1.35605621e+00
-5.59839964e-01 4.37482923e-01 -5.94821274e-01 1.96475402e-01
-3.82473022e-01 5.16738653e-01 6.56767905e-01 1.09294379e+00
2.12265849e-01 -9.77619551e-03 3.19893986e-01 6.80570245e-01
1.10785656e-01 -2.56898612e-01 -3.16088915e-01 2.34898537e-01
2.77707577e-01 1.44439352e+00 -8.50805104e-01 -4.23453093e-01
-1.71457499e-01 1.29352129e+00 3.45274866e-01 -1.27283618e-01
-1.08035946e+00 -3.36550981e-01 5.34826100e-01 6.55896783e-01
4.72897202e-01 -5.16467929e-01 1.77871466e-01 -9.43479061e-01
4.20036286e-01 -7.89784431e-01 4.12574887e-01 -4.52545375e-01
-8.74755144e-01 4.35909063e-01 -1.51668727e-01 -1.02059233e+00
-2.38319248e-01 -7.54583180e-01 -3.22183043e-01 5.60831666e-01
-1.12287831e+00 -1.23713851e+00 -7.93663800e-01 4.23267335e-01
3.90700668e-01 4.00054872e-01 5.86831689e-01 2.22778574e-01
-5.44612110e-01 5.26929557e-01 -6.50996208e-01 1.49317339e-01
8.34340096e-01 -1.19072568e+00 5.94668210e-01 1.20477164e+00
1.91420078e-01 5.63200653e-01 9.96916831e-01 -7.30744183e-01
-1.14798820e+00 -1.03593612e+00 5.36559522e-01 -6.55714989e-01
9.87520888e-02 -4.56684649e-01 -7.69979060e-01 6.88784301e-01
-2.72434317e-02 2.49097392e-01 3.21001440e-01 1.36371076e-01
-3.27319056e-01 -1.43882081e-01 -1.17521226e+00 3.67735177e-01
1.12154555e+00 -4.15088125e-02 -5.48830688e-01 3.63400578e-01
3.93869162e-01 -5.84688425e-01 -4.47708428e-01 7.21959114e-01
7.51579344e-01 -1.37276053e+00 9.82949138e-01 -1.18352644e-01
-2.77032126e-02 -7.62750447e-01 3.42380442e-02 -7.73574591e-01
-2.09141508e-01 -2.11226344e-01 -4.22520399e-01 9.24822628e-01
4.60915342e-02 -7.08885431e-01 5.97406626e-01 5.90116739e-01
1.86753929e-01 -1.38143331e-01 -8.44964862e-01 -9.57935810e-01
-7.23698020e-01 -3.34903955e-01 7.00602308e-03 2.70453691e-01
-2.79381305e-01 2.42952481e-01 -4.16442603e-01 7.46040940e-01
8.28612268e-01 -1.16517827e-01 1.21108043e+00 -1.15314221e+00
-6.68925166e-01 -1.26571998e-01 -8.39752197e-01 -7.52108634e-01
-3.65174651e-01 -2.60303319e-01 4.41990525e-01 -1.58777630e+00
3.27830970e-01 -1.60034731e-01 -3.90869439e-01 7.47119009e-01
-2.69914716e-01 5.36684096e-01 3.32880646e-01 1.37803555e-01
-9.21501637e-01 1.83557600e-01 6.51017308e-01 6.78808764e-02
1.21119142e-01 2.19847396e-01 -4.39995676e-01 9.01399732e-01
8.59229863e-01 -4.71398026e-01 -1.23349510e-01 -1.94596857e-01
-3.14204209e-02 -2.56457448e-01 1.04746997e+00 -1.88622999e+00
2.21139774e-01 6.40356988e-02 6.53628528e-01 -5.05858004e-01
6.87498987e-01 -6.73904061e-01 4.56843264e-02 9.16375399e-01
-7.54166916e-02 -1.80728167e-01 5.04358768e-01 7.46047735e-01
1.26544699e-01 1.36693930e-02 8.28820467e-01 -1.60279602e-01
-7.83573985e-01 -7.81762227e-02 -1.98642716e-01 -3.41349065e-01
1.44127131e+00 -1.00637916e-02 -2.44163513e-01 -3.51783276e-01
-5.43065131e-01 2.79754072e-01 7.68541098e-01 4.25132960e-01
4.14340377e-01 -8.89158547e-01 -5.43156743e-01 4.60478030e-02
4.18118060e-01 -9.56824124e-02 2.05723733e-01 9.39767599e-01
-4.58439887e-01 2.72736222e-01 -2.20084846e-01 -7.09771872e-01
-1.83076072e+00 6.43244267e-01 4.43346769e-01 1.00162707e-01
-6.34542465e-01 7.64571786e-01 1.93847880e-01 -5.12278937e-02
4.99391109e-01 -2.34202191e-01 2.24402159e-01 -3.57752860e-01
7.61443496e-01 5.43402195e-01 1.61038369e-01 -5.85146725e-01
-6.81246519e-01 3.22465450e-01 -5.80317155e-02 -2.04313025e-01
8.31829190e-01 4.54451218e-02 2.65475273e-01 1.85732320e-01
3.94746333e-01 1.36400446e-01 -1.48568654e+00 3.75106260e-02
-9.67381522e-02 -4.78879303e-01 -1.29758850e-01 -9.78652298e-01
-6.08364582e-01 7.53899753e-01 8.76387596e-01 4.29758914e-02
9.25413370e-01 1.52016610e-01 5.92398524e-01 5.19786358e-01
7.32399583e-01 -9.98659849e-01 2.02779487e-01 3.66974548e-02
8.42829525e-01 -1.63401806e+00 2.27468237e-01 -6.15149736e-01
-5.21419644e-01 7.74829566e-01 8.98171723e-01 -7.18431920e-02
9.74834561e-02 1.29998729e-01 1.37073519e-02 -1.52643681e-01
-5.58924913e-01 -6.48241282e-01 4.00976241e-01 6.33894861e-01
1.85510129e-01 1.86986551e-01 4.06436361e-02 2.53136039e-01
-4.73646037e-02 -8.67686197e-02 2.88697869e-01 1.12889767e+00
-8.08969140e-01 -1.02805293e+00 -5.99715650e-01 2.77101278e-01
-3.92457306e-01 3.01099330e-01 -4.69410658e-01 7.54977047e-01
3.10975075e-01 1.09951353e+00 -2.47521568e-02 -2.57690102e-01
5.27253568e-01 -5.51921576e-02 7.82932103e-01 -7.83564687e-01
-3.23818415e-01 -1.76332727e-01 2.66065449e-01 -9.51853633e-01
-3.35231274e-01 -8.30751419e-01 -1.26212835e+00 -8.49007592e-02
-2.48683512e-01 -3.08979541e-01 6.61840558e-01 6.84102893e-01
4.30525929e-01 5.94794989e-01 -1.52333844e-02 -8.37421775e-01
-6.54193938e-01 -8.04914415e-01 -1.99422628e-01 5.42262495e-01
3.23958963e-01 -8.81222308e-01 -1.58866853e-01 2.03070685e-01] | [7.907919406890869, -0.011560487560927868] |
05163292-42b1-4164-b6e8-403f16f6c80b | weakly-supervised-arrhythmia-detection-based | 2012.05641 | null | https://arxiv.org/abs/2012.05641v1 | https://arxiv.org/pdf/2012.05641v1.pdf | Weakly Supervised Arrhythmia Detection Based on Deep Convolutional Neural Network | Supervised deep learning has been widely used in the studies of automatic ECG classification, which largely benefits from sufficient annotation of large datasets. However, most of the existing large ECG datasets are roughly annotated, so the classification model trained on them can only detect the existence of abnormalities in a whole recording, but cannot determine their exact occurrence time. In addition, it may take huge time and economic cost to construct a fine-annotated ECG dataset. Therefore, this study proposes weakly supervised deep learning models for detecting abnormal ECG events and their occurrence time. The available supervision information for the models is limited to the event types in an ECG record, excluding the specific occurring time of each event. By leverage of feature locality of deep convolution neural network, the models first make predictions based on the local features, and then aggregate the local predictions to infer the existence of each event during the whole record. Through training, the local predictions are expected to reflect the specific occurring time of each event. To test their potentials, we apply the models for detecting cardiac rhythmic and morphological arrhythmias by using the AFDB and MITDB datasets, respectively. The results show that the models achieve beat-level accuracies of 99.09% in detecting atrial fibrillation, and 99.13% in detecting morphological arrhythmias, which are comparable to that of fully supervised learning models, demonstrating their effectiveness. The local prediction maps revealed by this method are also helpful to analyze and diagnose the decision logic of record-level classification models. | ['Henggui Zhang', 'Yongfeng Yuan', 'Runnan He', 'Qince Li', 'Kuanquan Wang', 'Yang Liu'] | 2020-12-10 | null | null | null | null | ['arrhythmia-detection', 'ecg-classification'] | ['medical', 'medical'] | [ 1.83974072e-01 -1.10622339e-01 -1.52194396e-01 -4.77145582e-01
-6.80710614e-01 -4.95685577e-01 -2.89784789e-01 4.03718710e-01
-4.92325239e-02 7.16555238e-01 -3.94063406e-02 -3.77785772e-01
-2.23435447e-01 -7.94654667e-01 -3.75004113e-01 -7.79909968e-01
-4.31202114e-01 2.72222757e-01 -2.02282578e-01 3.41360539e-01
-9.73519012e-02 3.34997565e-01 -9.64444518e-01 6.27821505e-01
8.63740504e-01 1.48884833e+00 1.45857865e-02 6.29892886e-01
2.50371665e-01 9.82040048e-01 -9.74109113e-01 1.17023639e-01
-1.29668117e-01 -7.66569793e-01 -5.05941749e-01 -6.86052516e-02
-2.78959781e-01 -4.55427229e-01 -3.67906511e-01 7.06446350e-01
9.40022588e-01 -5.81626594e-01 4.61932838e-01 -7.85406530e-01
-7.30507597e-02 8.27991307e-01 -2.78887838e-01 7.14041710e-01
1.32745028e-01 6.93027228e-02 9.42578733e-01 -6.00983799e-01
1.98430866e-01 2.36780062e-01 1.05705965e+00 1.39543042e-01
-8.39279413e-01 -6.94327593e-01 -1.79904640e-01 2.10515797e-01
-1.53526580e+00 -1.56721219e-01 9.87147272e-01 -5.07424891e-01
6.04834497e-01 2.20680058e-01 7.76223421e-01 9.01839137e-01
2.81283826e-01 5.64673901e-01 7.83823431e-01 -1.77520037e-01
7.02004805e-02 -2.29156360e-01 1.42675072e-01 6.54583097e-01
-7.52566680e-02 3.40678506e-02 -4.11469758e-01 -3.97005200e-01
8.62964153e-01 4.76168603e-01 -4.26605105e-01 2.81620562e-01
-1.54870963e+00 4.03738946e-01 4.40486312e-01 2.80556440e-01
-6.21557832e-01 -3.09754848e-01 7.38106310e-01 2.67302275e-01
4.10435855e-01 4.13447499e-01 -8.69870961e-01 -2.33660772e-01
-9.85611856e-01 -5.20542450e-02 5.53462684e-01 5.22881448e-01
5.21805048e-01 7.69688711e-02 -4.56706941e-01 6.08057380e-01
1.02515951e-01 2.50699133e-01 6.13057315e-01 -5.98849177e-01
4.07488823e-01 9.65655446e-01 -6.83896467e-02 -1.04366505e+00
-7.40568221e-01 -9.61790442e-01 -1.28428125e+00 -4.14682299e-01
3.13138664e-01 -4.18873906e-01 -5.89633703e-01 1.42106855e+00
-1.94023661e-02 4.70844060e-01 4.19964157e-02 8.25527966e-01
8.54555011e-01 5.01104176e-01 -5.14990501e-02 -6.32225513e-01
1.28626812e+00 -3.32790524e-01 -8.82896006e-01 5.65839410e-02
1.01185262e+00 -2.32118636e-01 6.92811608e-01 3.60119522e-01
-7.82865644e-01 -7.28016675e-01 -1.00832319e+00 4.81026411e-01
1.95563272e-01 7.09111154e-01 6.24626756e-01 4.16714549e-01
-4.82037842e-01 7.76953220e-01 -9.10902083e-01 6.58915788e-02
8.78081203e-01 2.95299321e-01 6.28605559e-02 1.28217474e-01
-1.48459780e+00 5.25221050e-01 3.56716722e-01 6.20230258e-01
-1.04258025e+00 -6.64176047e-01 -5.42316437e-01 2.73251623e-01
-8.94229710e-02 -4.23771590e-01 9.23875093e-01 -8.05870831e-01
-7.49405921e-01 7.86974072e-01 -2.15882763e-01 -7.95353949e-01
5.31183660e-01 1.67794153e-02 -7.46265531e-01 1.59872085e-01
1.93580583e-01 2.83530522e-02 6.59704626e-01 -6.17256641e-01
-7.01866269e-01 -4.30499643e-01 -8.08433294e-02 -1.73875261e-02
-3.56610447e-01 7.45640602e-03 -3.04613709e-01 -7.62461960e-01
2.07441241e-01 -4.62419122e-01 -2.20008671e-01 -2.87433147e-01
-4.62197691e-01 -1.34306923e-01 7.12418675e-01 -8.12712967e-01
1.76206541e+00 -2.56448483e+00 -2.65984565e-01 2.11558148e-01
6.47873402e-01 1.71058416e-01 3.88252884e-01 9.72583666e-02
-6.91793561e-02 3.73090245e-02 -3.63339394e-01 2.33745843e-01
-5.54982662e-01 1.98498726e-01 -6.36861145e-01 4.02437776e-01
2.41990387e-01 9.88609195e-01 -6.84696913e-01 -5.08554459e-01
3.24970372e-02 2.80858576e-01 -1.34619653e-01 3.24615628e-01
1.10352166e-01 6.63642466e-01 -6.01826787e-01 5.83131373e-01
2.44853437e-01 -4.57144827e-01 4.01783377e-01 -4.47203636e-01
2.02624917e-01 3.91620487e-01 -8.34701300e-01 1.47236121e+00
-8.81440565e-02 4.32465017e-01 -6.03810549e-01 -1.41555524e+00
1.06629956e+00 8.66298616e-01 7.54232109e-01 -5.01549900e-01
8.52558613e-02 3.10920954e-01 4.16446120e-01 -5.47631979e-01
-5.11150360e-01 -7.87507892e-02 -7.95981437e-02 4.83425081e-01
-1.44157901e-01 5.54076076e-01 -4.66075949e-02 -4.68968637e-02
1.28166366e+00 -4.85423245e-02 3.84255409e-01 -9.49044749e-02
3.87667894e-01 -2.12035790e-01 1.00831378e+00 9.09992993e-01
-2.31740192e-01 7.58931279e-01 7.43063092e-01 -1.12166345e+00
-6.01891816e-01 -1.02441788e+00 -5.86707234e-01 4.74425644e-01
1.55478403e-01 -6.15508497e-01 -5.48932552e-01 -9.77053285e-01
-2.80284017e-01 1.99291427e-02 -5.96024513e-01 -4.30954784e-01
-5.83891928e-01 -1.23495746e+00 9.96059895e-01 9.44546521e-01
5.42173624e-01 -1.50810957e+00 -1.08374858e+00 5.35214901e-01
-6.13206089e-01 -9.02899802e-01 -5.54685555e-02 4.05445218e-01
-1.16561234e+00 -1.16113973e+00 -4.91571486e-01 -8.11346769e-01
5.64698517e-01 -4.88580137e-01 1.14490569e+00 3.61771613e-01
-5.73489785e-01 -3.39818567e-01 -2.83621192e-01 -5.81040502e-01
-1.09938927e-01 -9.23352242e-02 7.92791843e-02 3.29226792e-01
3.29254270e-01 -6.20180726e-01 -8.07515800e-01 3.78596634e-01
-3.54348153e-01 -9.04218033e-02 6.47625387e-01 9.26671386e-01
8.11707616e-01 3.00718784e-01 1.15013433e+00 -1.01804638e+00
5.61441481e-01 -3.93438995e-01 -8.58259350e-02 6.89737201e-02
-5.87696850e-01 -5.02651036e-01 8.18982065e-01 -3.79257053e-01
-5.87481439e-01 2.51982007e-02 -2.44977340e-01 -3.83185983e-01
-3.31638992e-01 8.28834653e-01 -2.07461119e-01 5.25922239e-01
6.78065658e-01 5.11103809e-01 -3.65981162e-01 -1.56377196e-01
-5.54508209e-01 7.70922303e-01 4.97966111e-01 -3.81888360e-01
2.23838434e-01 3.05815518e-01 -2.64228433e-01 -3.61109376e-01
-1.19928920e+00 -3.70794684e-01 -6.99482858e-01 -1.32694215e-01
8.31015885e-01 -1.04000854e+00 -5.59446454e-01 5.72546959e-01
-1.06346118e+00 -1.89742580e-01 -4.79370862e-01 4.16733384e-01
-3.32358181e-01 3.05021435e-01 -8.16068053e-01 -6.38574362e-01
-5.58390975e-01 -7.64264941e-01 1.00299406e+00 -5.97196296e-02
-4.39722359e-01 -9.61174607e-01 -2.12952375e-01 1.24433979e-01
2.34755084e-01 5.12842953e-01 1.00586963e+00 -7.98811972e-01
-2.50943959e-01 -6.11162186e-01 -1.15837581e-01 3.16756368e-01
4.81315225e-01 -1.92305043e-01 -1.12544763e+00 -1.29700080e-01
1.92052558e-01 -1.12274252e-01 6.33414567e-01 6.30660653e-01
1.66447997e+00 -1.39190882e-01 -4.27497417e-01 7.43167520e-01
8.96384954e-01 4.48305845e-01 6.72703862e-01 -7.84689784e-02
7.08641529e-01 6.81822374e-02 5.79758883e-01 7.44259775e-01
2.08086699e-01 3.49848896e-01 1.67862356e-01 -5.78141272e-01
2.19778851e-01 2.54830462e-03 8.07933584e-02 7.98229635e-01
-3.98112804e-01 1.51117459e-01 -1.13361955e+00 5.19759536e-01
-1.93044758e+00 -8.70392859e-01 -2.30251297e-01 2.26311779e+00
9.71868992e-01 3.44421953e-01 -4.13832106e-02 7.19207883e-01
7.15823591e-01 -3.47014107e-02 -7.64825404e-01 -6.90591931e-02
-2.87565887e-01 2.53477335e-01 -1.01010077e-01 -1.61558002e-01
-1.25743854e+00 2.75702268e-01 6.11567545e+00 4.82890964e-01
-1.37900043e+00 4.68825437e-02 1.04513860e+00 5.02160862e-02
3.28492194e-01 -2.09340781e-01 -4.59519207e-01 6.63207769e-01
1.08597827e+00 1.95050593e-02 3.20583209e-02 5.12844622e-01
4.24056321e-01 4.56757694e-01 -1.26191854e+00 1.07481658e+00
-1.30829573e-01 -1.39668190e+00 -8.91999304e-02 -5.56626320e-02
4.65231299e-01 7.88171496e-03 -3.44171584e-01 3.71485740e-01
-5.10371804e-01 -1.15870762e+00 3.31963688e-01 6.23801529e-01
1.03693426e+00 -7.62631416e-01 1.25428331e+00 6.15939736e-01
-1.20238590e+00 -3.97417873e-01 -1.98396087e-01 -4.17320132e-01
-3.44450772e-02 1.02968359e+00 -7.89706945e-01 5.25248945e-01
9.16893065e-01 1.17805874e+00 -4.39807236e-01 9.50582385e-01
-3.26652765e-01 1.35020471e+00 -1.26447618e-01 4.47503954e-01
-4.02405113e-01 1.45990193e-01 1.99795574e-01 1.01138246e+00
2.69582957e-01 1.36793226e-01 4.63302821e-01 8.10695350e-01
-1.69901773e-01 1.16986342e-01 -3.97796303e-01 -1.03698121e-02
4.00543958e-01 1.21285784e+00 -5.70168912e-01 -5.38348734e-01
-2.53430307e-01 6.79875016e-01 1.39058560e-01 3.58844966e-01
-1.00413132e+00 -6.95376873e-01 8.51007551e-02 6.65155575e-02
6.32438660e-02 3.70489687e-01 -7.66811192e-01 -1.16676736e+00
3.57348531e-01 -8.51208925e-01 6.69974744e-01 -5.23302436e-01
-1.26851964e+00 7.65214384e-01 -6.98891044e-01 -1.61521053e+00
-2.14329064e-01 -2.10873738e-01 -1.03800738e+00 1.00104237e+00
-1.17025125e+00 -8.06202352e-01 -3.62323731e-01 6.11165166e-01
3.03891987e-01 -1.07924305e-01 1.27576220e+00 5.19602776e-01
-7.02985227e-01 6.76829457e-01 -4.46273237e-01 7.98456907e-01
5.50071597e-01 -1.17461860e+00 9.19014961e-03 7.42401898e-01
2.29324415e-01 6.15894556e-01 1.70004904e-01 -4.92513031e-01
-8.89569283e-01 -1.24928713e+00 1.04149067e+00 -4.40680861e-01
3.73763412e-01 -1.26995862e-01 -1.08619571e+00 5.14939368e-01
-3.49517077e-01 6.50007129e-01 8.66477311e-01 3.40219289e-01
-7.38042742e-02 -4.11434889e-01 -8.27686906e-01 6.36556447e-02
8.64482462e-01 -7.08790541e-01 -6.36472821e-01 3.97951007e-01
3.38921994e-01 -4.50200230e-01 -1.20954633e+00 9.50217128e-01
6.86796188e-01 -8.03664327e-01 7.51019120e-01 -6.45854354e-01
6.41123652e-01 -3.57199967e-01 2.37596542e-01 -1.03422773e+00
-3.10476780e-01 -3.59397352e-01 -4.96852130e-01 1.05889726e+00
6.31240189e-01 -5.72118640e-01 5.74795306e-01 9.39061940e-02
-3.22011054e-01 -1.06256318e+00 -6.47271216e-01 -2.53243923e-01
-4.05942559e-01 -5.49832463e-01 5.03896117e-01 1.20907509e+00
1.52738854e-01 2.62540013e-01 -3.42212856e-01 5.42187274e-01
3.76395553e-01 6.31040871e-01 1.39035195e-01 -1.37917447e+00
-4.19300765e-01 -1.15082465e-01 -6.29460335e-01 -5.92553616e-01
-1.79902047e-01 -9.74975288e-01 -8.12856853e-02 -1.47510934e+00
2.83755839e-01 -6.26391232e-01 -1.11959958e+00 7.65766382e-01
-4.74196970e-01 4.88380671e-01 -4.30313170e-01 4.67884809e-01
-4.55648422e-01 5.83430156e-02 1.03690183e+00 -1.12698518e-01
-3.38004351e-01 4.77967352e-01 -6.23417437e-01 1.01117778e+00
8.81412029e-01 -3.96552980e-01 -3.00122589e-01 -2.43690670e-01
1.95400745e-01 4.82001185e-01 4.03733373e-01 -1.15304029e+00
1.58379413e-02 3.06631774e-01 9.39864635e-01 -5.77132106e-01
-1.42706215e-01 -5.62868178e-01 -1.79597422e-01 5.74122131e-01
-4.88599598e-01 -1.95617259e-01 1.22984290e-01 6.37858152e-01
-6.44829512e-01 3.00142467e-01 4.33291912e-01 -9.31522716e-03
-3.37609828e-01 6.65119112e-01 -2.67566681e-01 2.03610286e-01
9.18720126e-01 -1.76770180e-01 1.14445426e-01 -9.94516164e-02
-1.07037485e+00 5.95892034e-02 -2.42040783e-01 2.08966061e-01
7.25709975e-01 -1.15837789e+00 -8.30336273e-01 4.41582441e-01
3.72034073e-01 2.61501193e-01 5.02168536e-01 1.14832294e+00
-4.68947262e-01 1.61354825e-01 -1.85532004e-01 -1.04729843e+00
-1.08598471e+00 4.11754280e-01 5.53849995e-01 -3.01658213e-01
-1.03347456e+00 5.69585621e-01 4.16715831e-01 -1.33860081e-01
2.56545633e-01 -4.65802073e-01 -3.99827778e-01 -1.79958735e-02
7.00125456e-01 -2.93162116e-03 3.42367470e-01 -9.06629115e-02
-5.16141415e-01 3.81229967e-01 1.01541756e-02 6.35478675e-01
1.22701955e+00 2.15967432e-01 -1.82920784e-01 5.89286804e-01
6.83043480e-01 -1.72973141e-01 -1.07225108e+00 -1.77133277e-01
-4.87959199e-02 -7.81964627e-04 -2.42300350e-02 -1.03445649e+00
-1.31561804e+00 1.38357484e+00 7.74067879e-01 1.58781365e-01
1.51309419e+00 -1.03536911e-01 8.35936129e-01 1.69705912e-01
2.61723191e-01 -5.75574756e-01 -3.78919244e-01 1.68503314e-01
3.94489497e-01 -8.76206994e-01 -1.95307612e-01 -1.93322003e-01
-8.13689113e-01 1.23790252e+00 4.06701326e-01 1.38765515e-03
6.80959940e-01 4.41882432e-01 3.02189499e-01 -2.77522445e-01
-6.68416560e-01 2.90164918e-01 2.38539577e-01 4.49268997e-01
5.33970475e-01 3.20548296e-01 -1.86478227e-01 1.27404666e+00
9.70713198e-02 2.94057250e-01 2.33507559e-01 7.21861243e-01
-1.13507904e-01 -7.68904865e-01 6.17420413e-02 1.02482915e+00
-8.53135169e-01 -2.64366895e-01 -1.71664491e-01 7.09129646e-02
4.79452610e-01 8.77811432e-01 6.69299662e-02 -4.65294600e-01
1.75386339e-01 2.56766737e-01 1.15512490e-01 -6.51150286e-01
-6.76476419e-01 1.88601494e-01 3.68735716e-02 -3.34427536e-01
-1.62350520e-01 -5.16366482e-01 -1.59552085e+00 4.84917909e-01
-3.18668634e-01 2.13973448e-01 1.42424868e-03 1.18462312e+00
5.48482060e-01 1.02845156e+00 7.67024219e-01 -2.73754746e-01
-4.44198400e-01 -1.10112417e+00 -8.13604534e-01 4.73225296e-01
4.91255939e-01 -2.27778479e-01 -2.34708756e-01 4.39249307e-01] | [14.282023429870605, 3.2607674598693848] |
bda7e8d5-baef-40ad-a506-c6afe636a1a6 | deep-signal-recovery-with-one-bit | 1812.00797 | null | http://arxiv.org/abs/1812.00797v1 | http://arxiv.org/pdf/1812.00797v1.pdf | Deep Signal Recovery with One-Bit Quantization | Machine learning, and more specifically deep learning, have shown remarkable
performance in sensing, communications, and inference. In this paper, we
consider the application of the deep unfolding technique in the problem of
signal reconstruction from its one-bit noisy measurements. Namely, we propose a
model-based machine learning method and unfold the iterations of an inference
optimization algorithm into the layers of a deep neural network for one-bit
signal recovery. The resulting network, which we refer to as DeepRec, can
efficiently handle the recovery of high-dimensional signals from acquired
one-bit noisy measurements. The proposed method results in an improvement in
accuracy and computational efficiency with respect to the original framework as
shown through numerical analysis. | ['Naveed Naimipour', 'Shahin Khobahi', 'Yonina C. Eldar', 'Mojtaba Soltanalian'] | 2018-11-30 | null | null | null | null | ['inference-optimization'] | ['audio'] | [ 5.57905018e-01 -1.55266419e-01 4.76389192e-02 -1.47427887e-01
-1.04476023e+00 -8.58096629e-02 3.12489301e-01 -2.88934380e-01
-2.08409801e-01 7.87194967e-01 1.53694645e-01 -6.45405948e-01
-3.72236729e-01 -4.79725242e-01 -8.03614676e-01 -9.37727332e-01
-2.90222883e-01 1.10384319e-02 -7.70016432e-01 2.59256512e-01
7.83462171e-03 5.26146293e-01 -7.05557287e-01 -1.80502430e-01
3.40483516e-01 1.59802139e+00 -3.35836202e-01 7.00158775e-01
5.24925411e-01 8.32123458e-01 -3.92575830e-01 -1.94895387e-01
2.80209392e-01 -4.29940194e-01 -4.55111504e-01 -9.66020375e-02
-7.05670640e-02 -5.87729394e-01 -7.24537015e-01 1.23154473e+00
8.03273559e-01 -6.77995756e-02 3.44488055e-01 -5.08140862e-01
-1.29435286e-01 9.95504677e-01 -4.31552708e-01 2.04430059e-01
2.50451356e-01 -2.10039511e-01 7.88011491e-01 -1.15049899e+00
-7.27441981e-02 9.19246018e-01 1.01345038e+00 1.69124827e-01
-1.16392422e+00 -6.01890981e-01 -7.08421290e-01 3.87618423e-01
-1.56184769e+00 -1.05122173e+00 1.13695800e+00 -2.47365177e-01
5.89908063e-01 1.70343667e-02 4.24685299e-01 1.05562031e+00
1.90947384e-01 8.33657026e-01 9.77737427e-01 -6.21734500e-01
3.55057240e-01 -5.24404466e-01 4.97296751e-02 5.51967323e-01
3.38834852e-01 5.70218682e-01 -4.38982904e-01 -2.03254044e-01
6.27571642e-01 1.62801638e-01 -4.22798932e-01 -4.22500074e-02
-1.22129142e+00 7.66165197e-01 4.67448056e-01 3.04223269e-01
-8.42802942e-01 6.57264292e-01 3.07142466e-01 5.59084535e-01
4.07829791e-01 4.37001288e-01 -1.03531219e-01 -1.10987052e-01
-1.27545691e+00 3.69686857e-02 1.00474620e+00 3.65706295e-01
5.77329695e-01 6.89047039e-01 2.78667789e-02 4.55573201e-01
4.26573604e-01 6.97736681e-01 2.01592818e-01 -8.96281958e-01
2.88323373e-01 -2.28514403e-01 2.03229100e-01 -1.21026433e+00
-5.68427205e-01 -9.23560917e-01 -1.63518894e+00 -3.72275859e-01
4.84679779e-03 -7.41851389e-01 -5.75609028e-01 1.62632632e+00
8.27987045e-02 7.40711927e-01 4.78614271e-01 9.01504159e-01
3.72828215e-01 7.22805858e-01 -6.41185045e-01 -4.51532483e-01
8.88291478e-01 -5.37508845e-01 -9.52259660e-01 -2.24947438e-01
3.15980166e-02 -5.96785665e-01 1.24836802e-01 6.83471262e-01
-1.27619374e+00 -5.61415553e-01 -1.52871776e+00 2.64124602e-01
1.15539715e-01 1.71542183e-01 5.49726486e-01 6.53612316e-01
-8.28837812e-01 7.22388923e-01 -9.35882926e-01 2.84623176e-01
4.45348471e-01 4.66590166e-01 3.03155165e-02 -2.13065118e-01
-1.25495291e+00 5.59339941e-01 3.81446779e-01 7.41207778e-01
-1.28950989e+00 -3.94907624e-01 -8.33223879e-01 1.97682604e-01
3.03792566e-01 -6.37373745e-01 1.30714250e+00 -6.00360870e-01
-1.84843957e+00 3.31141353e-01 -3.01561773e-01 -1.05811143e+00
-1.92740578e-02 -2.93025464e-01 -6.28871024e-01 2.05378279e-01
-4.22468007e-01 -3.22593570e-01 1.45434260e+00 -7.78678000e-01
-1.02623552e-01 -5.38484395e-01 -1.86854705e-01 -5.04394650e-01
7.82738924e-02 -3.83519679e-01 -7.12115988e-02 -6.06015563e-01
5.28546870e-01 -6.87626064e-01 -4.91545320e-01 -4.03401673e-01
-4.72473413e-01 3.81282628e-01 5.52789569e-01 -9.37527895e-01
1.19568217e+00 -2.23982525e+00 4.29668695e-01 4.81967896e-01
3.55995685e-01 3.89740407e-01 1.27088264e-01 5.76370239e-01
-5.47273122e-02 -4.07963097e-01 -3.26447397e-01 -5.92797816e-01
-8.35656077e-02 1.94204673e-01 -5.84158361e-01 9.95295227e-01
-2.66991127e-02 1.02383924e+00 -7.98880279e-01 1.34273618e-01
1.83116481e-01 4.99759495e-01 -3.11274379e-01 3.92505884e-01
-7.83740953e-02 8.51078808e-01 -3.37224722e-01 6.67949736e-01
7.43244112e-01 -4.42576677e-01 3.35497171e-01 -4.57697958e-01
4.83919345e-02 4.13781792e-01 -9.93283272e-01 2.00649166e+00
-5.51777899e-01 8.10469866e-01 5.56488574e-01 -1.74391019e+00
8.68486702e-01 7.00613558e-01 5.77693641e-01 -8.50503266e-01
4.76788372e-01 3.71805280e-01 -4.88947742e-02 -5.66464841e-01
2.79794693e-01 -4.06012595e-01 -2.43829742e-01 6.63172245e-01
-2.09909026e-02 6.30118474e-02 -3.49388659e-01 1.19760465e-02
1.11960173e+00 -3.96813929e-01 4.83798683e-01 4.11766581e-02
6.01967454e-01 -7.34901130e-01 3.95598650e-01 1.07533836e+00
2.96268705e-02 3.07744145e-02 2.03861594e-01 -3.77311409e-01
-1.10117233e+00 -8.57976198e-01 -1.99010909e-01 6.02480650e-01
-1.13317624e-01 -1.31943794e-02 -5.66600263e-01 1.11216418e-01
3.78794223e-02 3.43890011e-01 -1.69761166e-01 -2.95286864e-01
-6.48305237e-01 -8.61138463e-01 8.43304396e-01 2.90399849e-01
6.74279153e-01 -7.77548909e-01 -5.94623804e-01 3.64022523e-01
-8.38323459e-02 -1.42475951e+00 9.81857777e-02 3.60836446e-01
-7.56906927e-01 -7.13423908e-01 -7.15706706e-01 -5.16737878e-01
2.76576638e-01 -1.35645881e-01 7.81722784e-01 -2.94460148e-01
-6.50841668e-02 3.88839483e-01 -2.83799797e-01 -2.76890665e-01
-2.88938463e-01 -7.75426030e-02 3.36357892e-01 6.30290866e-01
-9.77316406e-03 -1.07087934e+00 -5.33232450e-01 -2.89744914e-01
-8.44059348e-01 -2.14336693e-01 1.00519693e+00 1.09443486e+00
5.06906986e-01 5.87522127e-02 8.00273538e-01 -6.66732073e-01
7.15984344e-01 -6.14556491e-01 -7.36949801e-01 -9.51043814e-02
-4.20666486e-01 4.46953684e-01 7.26482809e-01 -4.54905368e-02
-5.58916628e-01 -2.29199119e-02 -3.80967021e-01 -4.74995285e-01
3.60057533e-01 1.01964617e+00 9.69091617e-03 -4.56996620e-01
3.55901361e-01 6.44425213e-01 -2.16175113e-02 -6.53912246e-01
2.36180007e-01 9.43456292e-01 8.47330093e-01 -5.87387979e-01
7.41681218e-01 3.77200186e-01 4.14132148e-01 -8.18841994e-01
-1.12254393e+00 -2.90468723e-01 -4.20502156e-01 9.39381719e-02
3.90750676e-01 -9.78958666e-01 -9.32231009e-01 6.67163193e-01
-1.17576849e+00 -1.87395439e-01 -1.65724337e-01 7.19134688e-01
-6.28443837e-01 5.48943162e-01 -6.70366466e-01 -1.11258507e+00
-6.84357345e-01 -8.30155969e-01 9.64207053e-01 1.07637038e-02
2.95270950e-01 -9.18848276e-01 2.18719900e-01 -1.13691978e-01
5.74032187e-01 3.77046108e-01 6.43800557e-01 -3.18467021e-01
-6.68440163e-01 -6.27539098e-01 -2.54282951e-01 6.38234198e-01
-2.60232300e-01 -8.91950846e-01 -9.52720463e-01 -6.66947067e-01
5.87592065e-01 -3.75779808e-01 8.51630390e-01 5.84802508e-01
1.24426877e+00 -5.71572483e-01 -1.13439672e-01 1.12276912e+00
1.63549602e+00 -7.84905627e-02 5.93914092e-01 -2.87307471e-01
3.73935163e-01 -2.46385008e-01 7.93032523e-04 9.25037622e-01
-1.34076044e-01 5.09104311e-01 4.67225701e-01 6.86426386e-02
9.98371169e-02 -6.15286864e-02 2.25905433e-01 1.31185699e+00
2.41730779e-01 -5.69280870e-02 -5.29870689e-01 4.23366517e-01
-1.73429811e+00 -9.23946381e-01 3.50950062e-01 2.15271401e+00
6.54012322e-01 -9.07456204e-02 -2.87467271e-01 5.83368897e-01
2.44486466e-01 4.35231447e-01 -7.83316195e-01 -2.47683570e-01
2.55255271e-02 7.03254521e-01 4.79980588e-01 5.61217070e-01
-1.00736821e+00 3.46093744e-01 7.00478983e+00 5.59984446e-01
-1.34311461e+00 2.10330904e-01 4.35268164e-01 1.20756276e-01
1.01949848e-01 -1.95209250e-01 -3.43867123e-01 4.49165523e-01
1.39765275e+00 -4.49801702e-03 9.45373118e-01 3.39430839e-01
3.92551988e-01 2.45637968e-01 -1.18499100e+00 1.47354615e+00
2.87816711e-02 -1.37274373e+00 -1.96195856e-01 2.47107092e-02
7.00075984e-01 8.50631073e-02 2.36460268e-01 1.45286828e-01
-1.27474144e-01 -1.21720195e+00 4.85208750e-01 8.93118322e-01
8.12448025e-01 -7.65652716e-01 8.31567466e-01 5.24309397e-01
-8.24601531e-01 -3.77098799e-01 -2.99399346e-01 -5.93437910e-01
4.29340959e-01 1.18057644e+00 -7.99849987e-01 7.77706027e-01
1.72897607e-01 9.44064379e-01 2.66584277e-01 1.08962393e+00
-3.56737934e-02 9.60920811e-01 -2.91459620e-01 -8.48069489e-02
2.82876670e-01 -2.24825785e-01 6.72580063e-01 1.09112000e+00
6.06323242e-01 2.52128422e-01 2.43140295e-01 7.55422413e-01
-3.17208290e-01 -5.98558724e-01 -5.19892097e-01 -2.12907314e-01
5.04238844e-01 8.11053395e-01 7.67660141e-02 -3.00461620e-01
-1.76159382e-01 6.52666509e-01 6.46025017e-02 5.50434887e-01
-5.58105648e-01 -4.64950293e-01 4.13561434e-01 -2.21068561e-01
6.66751087e-01 -5.18778205e-01 -4.06965762e-01 -1.15284896e+00
1.37586966e-01 -9.97078419e-01 -5.64309508e-02 -4.37719911e-01
-1.22675717e+00 3.17137569e-01 -3.98272485e-01 -1.01647878e+00
-6.49925768e-01 -5.45510948e-01 -4.07440633e-01 7.71770120e-01
-1.68325257e+00 -8.38546872e-01 -1.82423424e-02 7.65693963e-01
4.69785854e-02 -1.58395439e-01 9.41408098e-01 6.36945128e-01
-6.52590334e-01 6.53258085e-01 6.02185249e-01 3.97307426e-01
-1.94756072e-02 -8.98492634e-01 2.08325267e-01 1.27155542e+00
2.70084202e-01 5.80498219e-01 7.75403559e-01 -1.11790806e-01
-2.06210709e+00 -9.71897244e-01 4.85612392e-01 3.23043078e-01
6.44037724e-01 -9.49397758e-02 -4.30293858e-01 7.29790866e-01
1.33867785e-01 1.37994632e-01 6.24245167e-01 -5.26084676e-02
-3.07756633e-01 -3.07524085e-01 -1.05780876e+00 2.16705073e-02
5.10809243e-01 -8.45016658e-01 -5.16961038e-01 2.39129320e-01
4.76906538e-01 -5.22441387e-01 -9.11401629e-01 5.12874305e-01
7.50118434e-01 -7.46193230e-01 1.24273086e+00 -4.03975934e-01
1.55524164e-01 -4.69295122e-02 -6.12842381e-01 -1.16614842e+00
-2.11705357e-01 -1.22869980e+00 -9.06991065e-01 3.98003578e-01
2.88780242e-01 -5.72464347e-01 6.46359503e-01 -4.61382419e-02
-1.86816454e-01 -8.20260584e-01 -1.15073276e+00 -4.28911120e-01
-2.09683388e-01 -7.46389806e-01 6.23414636e-01 5.62049747e-01
1.00870663e-02 5.33779323e-01 -1.12204981e+00 6.25978231e-01
1.13413167e+00 3.34353983e-01 6.85335577e-01 -1.01698136e+00
-9.62823212e-01 -1.93002716e-01 -4.06161606e-01 -1.79435039e+00
7.63698965e-02 -7.52630234e-01 2.88882703e-01 -1.02513707e+00
4.23407853e-02 -2.75358528e-01 -5.65673590e-01 -1.61217675e-02
1.99516505e-01 3.17960054e-01 -6.58836439e-02 7.02953711e-02
-6.88268900e-01 5.91449738e-01 8.80619466e-01 -3.17515135e-01
1.13335080e-01 4.96057749e-01 -5.75927079e-01 4.06955868e-01
6.63545012e-01 -3.07691455e-01 1.29827335e-01 -7.19220281e-01
3.65995169e-01 6.74114704e-01 4.92662549e-01 -1.39824712e+00
3.94440353e-01 5.32652795e-01 3.49692643e-01 -5.43079376e-01
6.92538917e-01 -8.22517514e-01 1.51018426e-01 7.81931698e-01
-3.37047875e-01 -4.08774793e-01 -6.14342876e-02 7.53870130e-01
-3.09456408e-01 -1.74409136e-01 6.60849273e-01 1.05342180e-01
-3.91121209e-01 2.61030525e-01 -3.12052608e-01 -2.39449948e-01
3.23136806e-01 3.20452839e-01 2.09833547e-01 -6.85501933e-01
-1.05224288e+00 -1.98287636e-01 -3.45002204e-01 -3.82409632e-01
7.84769475e-01 -1.38672221e+00 -8.32038462e-01 5.68253577e-01
-3.75186801e-01 -2.86059529e-01 1.29680693e-01 1.11594701e+00
-1.19705692e-01 7.76209235e-01 2.35246807e-01 -6.14886582e-01
-5.67364573e-01 4.89231288e-01 5.51501930e-01 -5.91724396e-01
-5.45194745e-01 7.23886073e-01 -5.89352012e-01 -9.52109620e-02
3.38730633e-01 -4.52139139e-01 2.82226622e-01 -4.66892868e-01
7.49168694e-01 3.76667708e-01 3.99767198e-02 -4.07259375e-01
-9.72078592e-02 5.63852191e-01 3.08288455e-01 -2.27908179e-01
1.31245792e+00 -1.56562090e-01 -3.46273363e-01 4.45447028e-01
1.65838909e+00 9.89656057e-03 -1.11151195e+00 -7.98648655e-01
8.12079012e-02 -2.03659117e-01 4.28007543e-01 -6.39231861e-01
-1.15222669e+00 9.58040059e-01 6.32336378e-01 3.89918506e-01
1.32993460e+00 -1.72228977e-01 1.30023277e+00 8.91224742e-01
5.91569543e-01 -4.65120941e-01 -1.30014747e-01 6.44722462e-01
4.59979981e-01 -1.05304635e+00 -1.42610250e-02 3.10737997e-01
3.54460627e-01 9.99805033e-01 -5.27981341e-01 -3.64466161e-01
1.03294098e+00 3.34642261e-01 -2.98898697e-01 -2.10636422e-01
-5.08662164e-01 3.65679301e-02 2.46711731e-01 3.83826554e-01
6.10758550e-02 3.38065833e-01 4.95011173e-03 7.16418147e-01
-1.37933612e-01 3.24669808e-01 2.76457965e-01 5.98658800e-01
-5.34624994e-01 -7.45775044e-01 -5.07340491e-01 3.89861315e-01
-6.62709117e-01 -1.65527672e-01 -8.78126081e-03 6.62608221e-02
-2.38694966e-01 1.03113508e+00 -2.60734022e-01 -5.12440443e-01
-2.00740136e-02 -5.88284386e-03 5.75531423e-01 -1.33312747e-01
2.57320721e-02 7.34231621e-02 1.77975241e-02 -5.33631384e-01
-5.51983058e-01 -5.53803504e-01 -8.57498169e-01 -3.36893767e-01
-1.65413454e-01 2.26392508e-01 8.01313460e-01 1.30307186e+00
4.75463271e-01 5.54015398e-01 1.06609070e+00 -1.01798058e+00
-1.15278172e+00 -1.05016863e+00 -6.91574574e-01 -8.55971575e-02
1.14945495e+00 -2.23198846e-01 -3.00120085e-01 -2.52083689e-01] | [6.515263557434082, 1.4637624025344849] |
7240f75d-f2d8-44ca-978c-66e3574e019a | sltunet-a-simple-unified-model-for-sign-1 | 2305.01778 | null | https://arxiv.org/abs/2305.01778v1 | https://arxiv.org/pdf/2305.01778v1.pdf | SLTUNET: A Simple Unified Model for Sign Language Translation | Despite recent successes with neural models for sign language translation (SLT), translation quality still lags behind spoken languages because of the data scarcity and modality gap between sign video and text. To address both problems, we investigate strategies for cross-modality representation sharing for SLT. We propose SLTUNET, a simple unified neural model designed to support multiple SLTrelated tasks jointly, such as sign-to-gloss, gloss-to-text and sign-to-text translation. Jointly modeling different tasks endows SLTUNET with the capability to explore the cross-task relatedness that could help narrow the modality gap. In addition, this allows us to leverage the knowledge from external resources, such as abundant parallel data used for spoken-language machine translation (MT). We show in experiments that SLTUNET achieves competitive and even state-of-the-art performance on PHOENIX-2014T and CSL-Daily when augmented with MT data and equipped with a set of optimization techniques. We further use the DGS Corpus for end-to-end SLT for the first time. It covers broader domains with a significantly larger vocabulary, which is more challenging and which we consider to allow for a more realistic assessment of the current state of SLT than the former two. Still, SLTUNET obtains improved results on the DGS Corpus. Code is available at https://github.com/bzhangGo/sltunet. | ['Rico Sennrich', 'Mathias Müller', 'Biao Zhang'] | 2023-05-02 | sltunet-a-simple-unified-model-for-sign | https://openreview.net/forum?id=EBS4C77p_5S | https://openreview.net/pdf?id=EBS4C77p_5S | international-conference-on-learning-2 | ['sign-language-translation'] | ['computer-vision'] | [ 1.89239189e-01 -2.13024750e-01 -4.81231391e-01 -2.96601623e-01
-1.28138375e+00 -6.23834252e-01 8.52253973e-01 -6.04331195e-01
-5.69263101e-01 5.34662902e-01 7.85827398e-01 -3.75652313e-01
2.81834066e-01 -1.08672395e-01 -6.87094808e-01 -3.48903567e-01
2.43308112e-01 6.20415628e-01 -1.38930082e-01 -2.92518705e-01
-3.21613640e-01 -1.50098011e-01 -1.35017586e+00 6.01649642e-01
8.36797297e-01 7.60457456e-01 3.69297415e-01 4.50299650e-01
-1.10012144e-01 5.80513418e-01 -9.45390090e-02 -5.26220918e-01
3.78312945e-01 -4.75507528e-01 -6.90435052e-01 -1.43110946e-01
1.02810037e+00 -4.83182788e-01 -5.08790910e-01 5.64524472e-01
9.74970520e-01 -1.56184779e-02 5.63315988e-01 -1.28573167e+00
-8.20046246e-01 6.51771367e-01 -2.81822294e-01 -1.24929264e-01
3.55471760e-01 5.98419070e-01 1.23589551e+00 -1.25921571e+00
1.03263223e+00 1.34641790e+00 3.83331656e-01 9.52051520e-01
-9.21476126e-01 -7.42559910e-01 1.74540862e-01 2.24453688e-01
-1.10023618e+00 -9.05079186e-01 2.72645712e-01 -1.32482633e-01
1.13096762e+00 2.13338614e-01 4.94603932e-01 1.66095710e+00
-2.55986094e-01 1.44101000e+00 1.26730835e+00 -4.51977342e-01
-2.08487302e-01 -2.69866139e-01 -1.73556566e-01 6.27627552e-01
-2.16671228e-01 -2.36624852e-02 -1.16856194e+00 8.85170251e-02
5.27465284e-01 -4.16844398e-01 -5.56947351e-01 -1.33080438e-01
-1.76370370e+00 4.31462198e-01 1.65669903e-01 4.32595968e-01
-2.34853789e-01 2.22340345e-01 5.35488248e-01 7.42388487e-01
4.21773463e-01 2.12382689e-01 -7.34889925e-01 -4.78712022e-01
-8.58158827e-01 1.79249793e-01 8.97528172e-01 1.00828087e+00
8.67177024e-02 3.06938048e-02 -4.32642132e-01 9.64055240e-01
3.66633177e-01 1.04173434e+00 4.94996369e-01 -7.91379929e-01
1.19572520e+00 1.00184470e-01 -1.64002001e-01 -1.33603901e-01
-2.00937167e-01 -2.64898539e-01 -6.86477184e-01 -3.63545828e-02
7.51877189e-01 -2.00486392e-01 -1.17324734e+00 2.15933156e+00
4.37527560e-02 8.52808654e-02 2.67450809e-01 1.22924614e+00
9.89046335e-01 5.76974928e-01 2.81275865e-02 3.01860366e-02
1.45121062e+00 -1.18559623e+00 -6.41140580e-01 -4.12666202e-01
8.62505496e-01 -9.20055747e-01 1.44576299e+00 9.22723189e-02
-1.09084177e+00 -1.49415568e-01 -6.27501667e-01 -2.77815789e-01
-2.26395786e-01 3.83035272e-01 4.88267064e-01 1.97424725e-01
-1.21131086e+00 1.40265688e-01 -9.55548406e-01 -9.84866619e-01
4.09200132e-01 1.08446024e-01 -5.32190740e-01 -4.48049247e-01
-1.17278719e+00 1.17494571e+00 -4.66323365e-03 1.56430677e-01
-5.21943271e-01 -4.84043121e-01 -7.52307653e-01 -3.30740571e-01
4.32634264e-01 -1.01722121e+00 1.47307086e+00 -9.46278274e-01
-1.69699919e+00 9.89409268e-01 -4.16894197e-01 -2.18453571e-01
7.90370286e-01 -2.88472444e-01 -3.88598800e-01 -1.44710476e-02
5.73989889e-03 9.49179232e-01 6.72072768e-01 -7.21343815e-01
-5.45729637e-01 -2.69901454e-01 -2.14642182e-01 5.64495564e-01
-2.71384984e-01 3.68161231e-01 -8.81590128e-01 -5.93183637e-01
1.22849259e-03 -1.17246962e+00 2.34171569e-01 1.76159114e-01
-2.56782711e-01 -3.25856239e-01 6.12098098e-01 -1.00986266e+00
9.06181514e-01 -1.96231437e+00 6.45872653e-01 -1.73209995e-01
-4.03085090e-02 3.19472641e-01 -8.48857462e-01 6.96292400e-01
3.13299447e-01 2.46523581e-02 -2.13099360e-01 -8.56291592e-01
2.17797175e-01 4.07771081e-01 -2.37945914e-01 2.65155822e-01
1.56826392e-01 1.28988409e+00 -6.90009654e-01 -4.04914796e-01
-9.02973190e-02 4.13055778e-01 -3.67612898e-01 8.46661031e-02
-4.65618849e-01 8.21210623e-01 -3.51277143e-01 9.25787687e-01
2.84932971e-01 -2.24257126e-01 2.51983106e-01 -1.08073048e-01
-1.15743294e-01 6.62422717e-01 -6.28570855e-01 2.22006702e+00
-5.88181555e-01 8.88879478e-01 7.09476769e-02 -5.29777169e-01
3.85000616e-01 5.51702559e-01 4.14237529e-01 -1.09642112e+00
1.79260671e-01 6.04394794e-01 -1.47825368e-02 -6.83624566e-01
3.78431737e-01 -1.29942119e-01 -8.71023536e-02 7.70027161e-01
2.66082108e-01 6.34166673e-02 2.26902932e-01 1.95400745e-01
1.02014673e+00 4.68098879e-01 -1.39030665e-01 3.11725643e-02
4.06145528e-02 -1.94151532e-02 1.69633925e-01 5.87419212e-01
-5.61848283e-02 4.98781353e-01 3.25299613e-02 3.66402231e-02
-9.45561588e-01 -1.15174353e+00 1.75098866e-01 1.36836326e+00
-1.52923197e-01 -2.72561073e-01 -4.38285798e-01 -7.44301379e-01
-3.97283956e-03 6.58980668e-01 -2.53823429e-01 7.15007186e-02
-8.29795480e-01 -4.39004779e-01 8.80360305e-01 4.41489637e-01
4.52011377e-01 -1.03394878e+00 -2.89188057e-01 -9.52404365e-02
-7.92972386e-01 -1.50222290e+00 -8.92103314e-01 -1.64509684e-01
-7.44224727e-01 -7.10323095e-01 -1.11079311e+00 -8.44418466e-01
2.95988053e-01 2.06644937e-01 1.16556108e+00 -1.87098667e-01
1.32896245e-01 6.57719910e-01 -5.31301022e-01 -2.09546804e-01
-5.64460278e-01 3.77332062e-01 1.36307105e-01 -1.50152653e-01
3.15731168e-01 -4.95196015e-01 -3.54865342e-01 3.09976965e-01
-8.02147388e-01 4.28547442e-01 7.92905986e-01 8.63296270e-01
2.17818022e-01 -1.00994599e+00 4.82240617e-01 -1.39738083e-01
6.45166695e-01 -3.15047860e-01 -2.43978903e-01 5.11836469e-01
-3.11980665e-01 2.08726108e-01 1.80775166e-01 -7.36939430e-01
-8.52848887e-01 -4.28701252e-01 -9.65698734e-02 -4.64157254e-01
1.21908776e-01 8.26093376e-01 -1.18433729e-01 3.82521823e-02
3.44419271e-01 5.09432733e-01 2.66460180e-01 -6.82172120e-01
6.74826503e-01 8.73283863e-01 5.43906212e-01 -6.47688866e-01
5.85657954e-01 1.58751711e-01 -3.44255239e-01 -6.14873469e-01
-6.48038805e-01 -2.39120439e-01 -3.76903772e-01 -2.30209351e-01
6.41613424e-01 -1.15011537e+00 -6.33036077e-01 6.15637541e-01
-1.11831033e+00 -8.00104916e-01 -1.98716801e-02 6.26910806e-01
-6.10269666e-01 2.94812530e-01 -7.27564514e-01 -4.94892299e-01
-5.99420488e-01 -1.10255992e+00 1.27717280e+00 -2.36653581e-01
-2.24987537e-01 -8.20193231e-01 1.02882899e-01 8.50828767e-01
5.72534084e-01 -1.98241621e-01 6.35875046e-01 -5.70228279e-01
-7.40555406e-01 3.19540501e-02 -3.09971273e-01 4.04294223e-01
3.35784554e-02 -3.79691273e-01 -7.81175077e-01 -6.01599395e-01
-5.04483819e-01 -7.54579723e-01 1.00372851e+00 1.34197608e-01
4.77071315e-01 -3.41302037e-01 -5.88148609e-02 5.95438719e-01
1.04635644e+00 -2.85823613e-01 5.06749690e-01 2.31360853e-01
8.55313182e-01 5.02685428e-01 4.99517411e-01 1.66486710e-01
9.55585957e-01 1.05889463e+00 4.74011227e-02 -8.12293962e-02
-7.20899761e-01 -4.15574133e-01 8.63582194e-01 1.38204253e+00
-2.58178145e-01 -6.06318712e-01 -9.83569324e-01 7.38292992e-01
-2.01331592e+00 -7.56213784e-01 1.99356005e-01 2.07178378e+00
9.85277176e-01 -2.98242807e-01 2.14203209e-01 -4.11711931e-01
2.67897069e-01 2.35734120e-01 -5.13756096e-01 -1.38872966e-01
-4.00808960e-01 1.22800462e-01 5.50572157e-01 3.64433110e-01
-8.03398788e-01 1.35168159e+00 5.49386787e+00 8.90722156e-01
-1.42863727e+00 4.60148692e-01 3.99964750e-02 -5.44168055e-01
-2.85693318e-01 -1.59089252e-01 -7.41514027e-01 4.12405223e-01
8.49625587e-01 7.70691782e-02 7.99966812e-01 6.84919059e-02
3.18202972e-01 1.22007236e-01 -1.24734211e+00 9.53196466e-01
2.25533202e-01 -1.14710367e+00 1.49391398e-01 1.46178856e-01
7.03881323e-01 1.03890097e+00 1.84914082e-01 5.00096738e-01
3.12257349e-01 -8.40540349e-01 9.72927094e-01 3.64030600e-01
1.30606425e+00 -1.00983165e-01 4.70828623e-01 3.35529923e-01
-1.11521971e+00 1.18106417e-01 2.13491201e-01 1.00538120e-01
4.24307138e-01 1.99205466e-02 -8.06703985e-01 5.66556156e-01
3.39094520e-01 8.96481991e-01 -3.01391035e-01 7.54440308e-01
-3.52758527e-01 7.02978194e-01 -6.76092923e-01 -4.29821275e-02
4.31236595e-01 -8.22322816e-02 8.01712751e-01 1.19196069e+00
6.47266805e-01 -3.30777317e-01 1.65116042e-01 5.88801503e-01
-3.43520284e-01 3.31070662e-01 -6.62461698e-01 -3.34924519e-01
4.15946573e-01 7.31348217e-01 -2.77159084e-02 -3.03155214e-01
-6.51195347e-01 1.31124234e+00 3.90397012e-01 7.29355454e-01
-7.15259492e-01 2.64772564e-01 8.41596603e-01 -1.65851526e-02
2.74241030e-01 -5.00578463e-01 -8.17907676e-02 -1.74285305e+00
4.03283298e-01 -1.23383236e+00 3.60950261e-01 -8.77658844e-01
-1.34716570e+00 6.01036668e-01 -5.56865446e-02 -1.40768778e+00
-4.94367182e-01 -6.99593782e-01 -1.73721567e-01 9.28239703e-01
-1.63291347e+00 -1.86971915e+00 3.63713428e-02 7.94266999e-01
7.13722765e-01 -3.13233793e-01 6.54713571e-01 4.26814556e-01
-3.37284893e-01 9.84319329e-01 1.36969566e-01 1.87045336e-01
1.20220327e+00 -7.00614095e-01 6.31074965e-01 8.89359713e-01
4.03006077e-01 4.19861317e-01 5.70244849e-01 -6.62668943e-01
-1.70576298e+00 -9.51581299e-01 1.34649348e+00 -6.94882274e-01
9.32842791e-01 -4.03815061e-01 -5.02703011e-01 7.78900981e-01
4.18219447e-01 -2.33549133e-01 4.19584423e-01 1.00796126e-01
-6.72200918e-01 2.27211237e-01 -6.81042373e-01 1.06597400e+00
1.55943668e+00 -8.40070009e-01 -4.01028544e-01 3.64343315e-01
7.44516015e-01 -5.71611941e-01 -8.20811272e-01 2.98882902e-01
9.60829318e-01 -3.82911772e-01 7.21802294e-01 -6.36524975e-01
5.54027736e-01 -8.64658877e-02 -5.17934501e-01 -1.37450135e+00
1.99476317e-01 -7.72569537e-01 -6.03777952e-02 1.01330686e+00
7.12665081e-01 -7.88969278e-01 4.16048527e-01 4.62962121e-01
-2.07697302e-01 -7.70448029e-01 -1.33905196e+00 -9.91820574e-01
2.43627876e-01 -5.91573656e-01 3.84544045e-01 9.58298624e-01
2.90839076e-02 4.97676671e-01 -8.10886085e-01 -9.29314196e-02
6.00737095e-01 1.35545552e-01 9.36373949e-01 -8.10947955e-01
-4.51203585e-01 -6.00901484e-01 1.60854310e-02 -1.37046278e+00
3.10629725e-01 -1.28661096e+00 3.99591289e-02 -1.72901297e+00
1.34138659e-01 -1.54937655e-01 -8.73666629e-02 9.79533672e-01
7.41791949e-02 3.18808526e-01 7.01239765e-01 4.17064637e-01
-5.98985553e-01 7.98130035e-01 1.62246597e+00 -1.28298149e-01
-1.24494880e-01 -2.42873400e-01 -4.98054534e-01 2.77595729e-01
7.27530420e-01 -2.11295635e-01 -1.68820307e-01 -1.14117801e+00
-4.78599370e-02 2.11377725e-01 3.55146080e-01 -5.82754433e-01
1.65829733e-01 -3.87730338e-02 -3.36225599e-01 -3.54233503e-01
4.87659514e-01 -6.91567481e-01 -1.41640827e-01 1.47616223e-01
-4.97621894e-01 -5.44968247e-02 1.66481137e-01 2.62858361e-01
-2.10808247e-01 3.18044990e-01 3.67688626e-01 7.91480914e-02
-6.86632574e-01 3.28217268e-01 -3.61173630e-01 2.47076049e-01
3.03797007e-01 -6.97943643e-02 -6.13961577e-01 -7.19325602e-01
-5.32514274e-01 5.00923514e-01 4.21620250e-01 8.68100762e-01
4.63346720e-01 -1.52939320e+00 -1.16814566e+00 9.24319252e-02
5.62217712e-01 -4.26387668e-01 9.23759341e-02 1.32434583e+00
-1.84056964e-02 5.58305442e-01 -1.69251606e-01 -5.45953989e-01
-1.39847696e+00 -1.53639734e-01 1.85887992e-01 -3.26516449e-01
-6.31994784e-01 7.76275694e-01 -1.32351995e-01 -7.77358830e-01
3.05385053e-01 -5.29383659e-01 2.49726996e-01 -7.37066045e-02
2.42771208e-01 7.77778849e-02 -1.07069433e-01 -8.54070544e-01
-4.10520494e-01 5.58679581e-01 6.93565458e-02 -5.99682391e-01
1.18940294e+00 -3.26764077e-01 8.92872214e-02 4.84363765e-01
1.13347077e+00 -1.38457388e-01 -1.05550027e+00 -7.80891001e-01
-1.49256900e-01 -2.74462014e-01 3.78440917e-02 -1.41918468e+00
-9.96318638e-01 8.39498043e-01 5.31677067e-01 -7.77278662e-01
1.09193671e+00 2.27541074e-01 1.25805056e+00 5.59383571e-01
5.40968835e-01 -9.33860242e-01 -9.54590831e-03 8.39698911e-01
1.15834904e+00 -1.52090645e+00 -4.09352094e-01 -1.25253936e-02
-8.43224585e-01 7.48383582e-01 3.89353663e-01 4.32240248e-01
9.84966308e-02 1.66859239e-01 4.72927183e-01 1.38689885e-02
-1.03998935e+00 -5.14377475e-01 6.69280410e-01 3.65083098e-01
5.03475308e-01 1.21931069e-01 -2.33823150e-01 2.99455971e-01
-5.21967225e-02 3.01636428e-01 3.61120738e-02 8.20938826e-01
4.58353236e-02 -1.32004368e+00 -2.06635132e-01 2.10330382e-01
-2.13990778e-01 -4.79655951e-01 -4.46791470e-01 7.73803473e-01
-8.73156562e-02 8.74172390e-01 -1.80921212e-01 -3.94103020e-01
4.04640198e-01 3.03460121e-01 8.39510083e-01 -2.96117604e-01
-4.70540285e-01 2.85120547e-01 6.06323540e-01 -6.72614276e-01
-5.19121885e-01 -8.33038867e-01 -9.03629839e-01 -2.67437786e-01
-1.92428520e-03 -4.47410762e-01 7.42043495e-01 1.22355902e+00
4.68050271e-01 1.74649820e-01 -6.34562457e-03 -8.80052805e-01
-5.98156035e-01 -1.24778020e+00 -1.72200501e-01 3.52344215e-01
3.06602597e-01 -4.30848062e-01 -1.65144548e-01 -9.73084010e-03] | [9.252911567687988, -6.570659160614014] |
1da18deb-9c01-4ff8-bfa6-3917eb074311 | ditto-nerf-diffusion-based-iterative-text-to | 2304.02827 | null | https://arxiv.org/abs/2304.02827v1 | https://arxiv.org/pdf/2304.02827v1.pdf | DITTO-NeRF: Diffusion-based Iterative Text To Omni-directional 3D Model | The increasing demand for high-quality 3D content creation has motivated the development of automated methods for creating 3D object models from a single image and/or from a text prompt. However, the reconstructed 3D objects using state-of-the-art image-to-3D methods still exhibit low correspondence to the given image and low multi-view consistency. Recent state-of-the-art text-to-3D methods are also limited, yielding 3D samples with low diversity per prompt with long synthesis time. To address these challenges, we propose DITTO-NeRF, a novel pipeline to generate a high-quality 3D NeRF model from a text prompt or a single image. Our DITTO-NeRF consists of constructing high-quality partial 3D object for limited in-boundary (IB) angles using the given or text-generated 2D image from the frontal view and then iteratively reconstructing the remaining 3D NeRF using inpainting latent diffusion model. We propose progressive 3D object reconstruction schemes in terms of scales (low to high resolution), angles (IB angles initially to outer-boundary (OB) later), and masks (object to background boundary) in our DITTO-NeRF so that high-quality information on IB can be propagated into OB. Our DITTO-NeRF outperforms state-of-the-art methods in terms of fidelity and diversity qualitatively and quantitatively with much faster training times than prior arts on image/text-to-3D such as DreamFusion, and NeuralLift-360. | ['Se Young Chun', 'Gwanghyun Kim', 'Hayeon Kim', 'Hoigi Seo'] | 2023-04-06 | null | null | null | null | ['3d-object-reconstruction', 'image-to-3d', 'object-reconstruction', 'text-to-3d'] | ['computer-vision', 'computer-vision', 'computer-vision', 'computer-vision'] | [ 1.92506388e-01 1.72577858e-01 2.23874643e-01 -1.81944072e-01
-1.15108442e+00 -5.02879500e-01 7.49615967e-01 -4.66571599e-01
3.04787718e-02 4.39227700e-01 2.84063220e-01 8.87474194e-02
1.99760824e-01 -5.38676560e-01 -9.52377141e-01 -3.99363220e-01
4.20544416e-01 9.58354831e-01 5.32454073e-01 -6.13064878e-03
5.13220616e-02 9.37876165e-01 -1.44380426e+00 4.17855233e-01
6.52052402e-01 1.08463275e+00 5.21366417e-01 7.85128534e-01
-2.42670760e-01 5.11090040e-01 -2.75903672e-01 -3.39916199e-01
6.77030563e-01 -6.54145300e-01 -5.79432189e-01 5.59516907e-01
1.05304050e+00 -9.65731084e-01 -4.16028678e-01 7.71145046e-01
4.67993349e-01 -1.16635421e-02 6.86892629e-01 -9.07168031e-01
-6.19264781e-01 -1.21495418e-01 -6.58272505e-01 -7.60563761e-02
5.99753022e-01 4.09477413e-01 4.72526401e-01 -1.39064538e+00
1.35589099e+00 1.45074975e+00 7.30716944e-01 5.35659790e-01
-1.52580976e+00 -3.63934547e-01 2.80898269e-02 -2.04606324e-01
-1.11397636e+00 -6.40085876e-01 8.92396927e-01 -5.68911374e-01
1.02421451e+00 1.60065088e-02 8.31736445e-01 1.00111222e+00
2.09200397e-01 6.04966223e-01 1.21806943e+00 -4.20894444e-01
1.29687026e-01 -5.50700426e-02 -4.62309361e-01 8.34288597e-01
-1.82044774e-01 2.53946453e-01 -8.33134890e-01 -5.66801280e-02
1.49879801e+00 -7.45439082e-02 -3.14983785e-01 -3.89609754e-01
-1.37547362e+00 3.89303714e-01 3.03128213e-01 -1.11325845e-01
-5.67223072e-01 1.58391923e-01 -1.27950385e-01 1.61143631e-01
8.78278732e-01 3.15374017e-01 -3.37210178e-01 -1.62726194e-02
-1.28762031e+00 5.45858562e-01 4.47742105e-01 1.24155033e+00
8.06036055e-01 2.51172155e-01 -2.90785469e-02 7.58688211e-01
2.33784437e-01 6.27315223e-01 6.65520364e-03 -1.52955163e+00
3.74514371e-01 3.13100338e-01 3.80502194e-01 -6.20924652e-01
-2.34951228e-01 -2.13990167e-01 -6.90753758e-01 5.09242296e-01
2.08203822e-01 1.88308671e-01 -1.37750769e+00 1.17540598e+00
8.29929829e-01 -1.23230889e-01 -1.63981810e-01 1.13298774e+00
9.84719634e-01 8.03269267e-01 -7.49553263e-01 -4.11747664e-01
9.43921447e-01 -1.01064003e+00 -7.17823744e-01 -2.93927997e-01
1.39573351e-01 -1.03870237e+00 9.61297572e-01 5.17754316e-01
-1.86619985e+00 -5.69211543e-01 -6.85516834e-01 -6.10971808e-01
2.72492379e-01 -2.38365963e-01 1.26144201e-01 7.61793181e-02
-1.15131998e+00 5.43837965e-01 -7.81358838e-01 -1.64815217e-01
7.06294715e-01 -4.63214368e-02 -6.92012072e-01 -6.67226732e-01
-5.70818424e-01 1.01921272e+00 8.92053694e-02 -2.52718985e-01
-1.29711080e+00 -1.26305091e+00 -7.57431686e-01 -3.19911778e-01
3.51234585e-01 -1.09657550e+00 1.26121759e+00 -6.93470895e-01
-1.57322097e+00 1.04501712e+00 -9.81218219e-02 -1.36291578e-01
8.10238242e-01 -1.15839392e-01 2.52521355e-02 5.86557329e-01
1.92727432e-01 1.06289780e+00 1.10312402e+00 -1.79137528e+00
-3.89761955e-01 -3.20283651e-01 -1.90886185e-01 4.19133067e-01
2.51728624e-01 -6.99175671e-02 -7.80117691e-01 -7.45202243e-01
4.67209250e-01 -6.30551159e-01 -8.18714797e-02 8.64707351e-01
-2.32979208e-01 3.34638566e-01 1.17861474e+00 -9.57404852e-01
4.92624581e-01 -1.95032620e+00 3.15423250e-01 -4.18044955e-01
3.09371471e-01 9.62741897e-02 -2.59515017e-01 2.58544147e-01
2.64403149e-02 -5.02275601e-02 -4.04545009e-01 -9.88153040e-01
-2.71944970e-01 2.06247628e-01 -1.63195819e-01 5.69541097e-01
1.17443442e-01 9.35334027e-01 -7.95925260e-01 -5.84278822e-01
6.17929161e-01 7.30799437e-01 -8.26828122e-01 6.09176099e-01
-6.17128372e-01 6.64512753e-01 -1.76228955e-01 8.38239968e-01
8.88450861e-01 -3.13582271e-01 -3.05350214e-01 -6.53798819e-01
-2.71721691e-01 7.46242404e-02 -1.09650230e+00 2.15032434e+00
-3.92940611e-01 5.55123746e-01 4.90151554e-01 -5.02163112e-01
7.24202693e-01 2.72490591e-01 6.71514571e-01 -6.46211684e-01
-8.50222036e-02 3.47623706e-01 -3.81444305e-01 -4.45887148e-01
5.65244079e-01 -4.26686645e-01 3.60531121e-01 3.74238104e-01
3.46078873e-01 -1.14405680e+00 4.18145992e-02 3.64075422e-01
7.64843225e-01 5.53990781e-01 -8.81318003e-02 1.72950402e-01
-9.79200527e-02 9.71828327e-02 2.79236138e-01 5.35048485e-01
2.28695706e-01 1.55137646e+00 2.06218570e-01 -3.68680775e-01
-1.75997639e+00 -1.12941587e+00 -2.38505527e-01 -9.83360969e-03
1.33297160e-01 -2.79661000e-01 -6.94613397e-01 -6.72015786e-01
-1.09519407e-01 7.78412223e-01 -5.35203159e-01 2.84851581e-01
-5.11421502e-01 -1.38493404e-01 2.36847118e-01 1.97536394e-01
5.48800111e-01 -9.04772997e-01 -3.58265489e-01 3.16923380e-01
-3.19542229e-01 -1.50617611e+00 -8.09678376e-01 5.22937402e-02
-1.17622995e+00 -7.63886869e-01 -1.06049740e+00 -4.70671445e-01
8.03430259e-01 4.76412833e-01 1.39905953e+00 -2.83528030e-01
-2.39860922e-01 4.79124814e-01 -2.26061270e-01 -9.82141271e-02
-7.65871584e-01 -5.28309584e-01 -3.81927788e-02 -6.57439679e-02
-7.02413142e-01 -8.05746377e-01 -7.26599216e-01 5.24246573e-01
-1.24497592e+00 7.36224890e-01 5.25339603e-01 7.31850088e-01
1.02506793e+00 -1.13110803e-01 4.93313856e-02 -4.28556770e-01
3.60412784e-02 -9.94998962e-02 -6.39083624e-01 -7.13039488e-02
-4.37749416e-01 -6.62948564e-02 4.62737650e-01 -6.40755832e-01
-1.27517891e+00 1.32808164e-01 -2.46384576e-01 -1.26863205e+00
-1.14159688e-01 -2.63079796e-02 -1.48475423e-01 -1.13412820e-01
7.54448533e-01 4.16164845e-01 1.90453008e-01 -5.98763347e-01
5.11156738e-01 2.87702173e-01 5.99325597e-01 -4.00858700e-01
7.92293191e-01 8.56240213e-01 4.78619225e-02 -8.64083946e-01
-1.02625954e+00 -1.95768073e-01 -7.12093711e-01 -4.00231659e-01
9.87259209e-01 -9.31203127e-01 -1.17615774e-01 6.75054848e-01
-1.43359983e+00 -7.14835942e-01 -6.89139605e-01 3.37300241e-01
-8.57437789e-01 3.23060215e-01 -7.02999353e-01 -5.67650676e-01
-3.48229259e-01 -1.23410332e+00 1.70301712e+00 -7.40848854e-02
-6.76607415e-02 -7.41427660e-01 -2.03079179e-01 7.43510008e-01
2.50947446e-01 2.65676618e-01 7.73290932e-01 2.73981035e-01
-1.18360829e+00 1.13596961e-01 -2.88717508e-01 5.72824895e-01
1.84385963e-02 -1.93491012e-01 -7.50752985e-01 -7.80041143e-02
1.97997913e-01 -4.95447516e-01 4.65253085e-01 7.63616860e-01
7.37625897e-01 -3.21124017e-01 -1.78460389e-01 9.92167652e-01
1.39001179e+00 -9.12072882e-02 7.08939731e-01 6.67439476e-02
8.77532125e-01 4.48821962e-01 6.15184724e-01 2.69448251e-01
2.60856241e-01 1.08165205e+00 6.27200603e-01 -3.12422842e-01
-8.64597142e-01 -4.51136738e-01 4.13959622e-01 6.50777757e-01
5.50488830e-02 -3.39233786e-01 -4.71997440e-01 5.97453058e-01
-1.38528061e+00 -9.06173050e-01 -2.40259141e-01 1.95564938e+00
1.01922357e+00 -6.17916649e-03 -2.95008868e-02 -2.35362262e-01
4.13046122e-01 2.47426152e-01 -6.67422771e-01 3.36848409e-03
-3.50713044e-01 -7.87891075e-02 2.44295269e-01 8.39019120e-01
-3.81703436e-01 9.52676475e-01 6.11894798e+00 9.78025615e-01
-1.07434475e+00 3.37320715e-01 7.76469707e-01 -3.83509755e-01
-6.01400614e-01 6.56247139e-02 -9.57229733e-01 1.88168496e-01
4.55467463e-01 2.44517028e-01 4.86871690e-01 6.46011233e-01
3.72223198e-01 -4.15126055e-01 -1.08420014e+00 1.35291088e+00
3.86673033e-01 -1.90385735e+00 2.95594692e-01 1.59923121e-01
1.26717484e+00 1.66788563e-01 -2.24474400e-01 -2.24629402e-01
1.74256936e-01 -7.82016754e-01 1.25133181e+00 7.64058650e-01
1.28981328e+00 -3.82298648e-01 2.48437189e-02 6.05554461e-01
-8.23270380e-01 4.95358169e-01 -1.21277668e-01 5.62329173e-01
8.87376010e-01 9.51564968e-01 -7.67859817e-01 6.96454823e-01
8.45685482e-01 7.70868719e-01 -2.91224271e-01 6.98372006e-01
-7.25079188e-03 1.18840106e-01 -5.95833659e-01 6.45312309e-01
1.28880188e-01 -1.45411491e-01 7.76984334e-01 8.38497877e-01
6.64132118e-01 4.00725991e-01 1.21346995e-01 1.29116940e+00
-2.23471388e-01 -3.82186353e-01 -6.27459466e-01 1.61899358e-01
1.16594963e-01 1.20406699e+00 -8.33922207e-01 -4.33300763e-01
-2.76855946e-01 1.44015479e+00 1.17244743e-01 3.00867498e-01
-5.99186957e-01 2.06598580e-01 4.05748665e-01 8.41557801e-01
4.98249739e-01 -3.94598126e-01 -1.49330854e-01 -1.19238102e+00
2.38081113e-01 -7.98860312e-01 -1.45673856e-01 -1.65247965e+00
-1.31132531e+00 7.91430831e-01 2.67740697e-01 -1.31064939e+00
-1.83604673e-01 -3.15142989e-01 -9.68516171e-02 9.08002853e-01
-1.42871749e+00 -1.34142447e+00 -4.50628459e-01 6.26771867e-01
1.03371847e+00 1.73173323e-01 5.53367078e-01 2.07865402e-01
2.33763114e-01 4.45999168e-02 -4.65672463e-02 -4.35927957e-01
7.02548802e-01 -7.47506618e-01 5.93312681e-01 7.60700524e-01
2.05646604e-02 -2.81801950e-02 4.97027308e-01 -9.69902635e-01
-1.54172361e+00 -9.80570674e-01 5.07769525e-01 -7.41228402e-01
6.36832090e-03 -4.75932479e-01 -8.07717383e-01 4.99782443e-01
1.19973093e-01 5.18074870e-01 -2.44460478e-01 -6.70895636e-01
-1.19445235e-01 9.93227512e-02 -1.47192764e+00 6.51739657e-01
1.34299862e+00 -5.40109575e-01 -3.54439497e-01 3.49024653e-01
8.95119250e-01 -1.00947452e+00 -8.51509988e-01 2.16456577e-01
5.56504905e-01 -1.36774409e+00 1.21136320e+00 1.00253969e-01
8.44906390e-01 -4.83925819e-01 -2.31991276e-01 -1.34622812e+00
-1.12193458e-01 -1.01665282e+00 -3.39318246e-01 9.99483049e-01
3.85116264e-02 -1.06429242e-01 8.25456738e-01 5.10982275e-01
-6.28703058e-01 -8.09152305e-01 -9.92727518e-01 -6.84118092e-01
-1.78060621e-01 -5.42980611e-01 3.25226128e-01 9.43413317e-01
-8.59285772e-01 2.29479238e-01 -4.69660014e-01 5.90896495e-02
9.15663123e-01 1.77501053e-01 1.01933289e+00 -1.00759554e+00
-2.48466685e-01 -1.17770091e-01 -7.81322420e-02 -1.66272187e+00
-2.85090566e-01 -8.22718203e-01 -5.03448071e-03 -1.85320079e+00
-1.20251132e-02 -4.03792083e-01 7.86567032e-01 1.69589296e-01
2.52444297e-01 5.26540339e-01 1.87694505e-01 4.03391391e-01
-1.95327982e-01 9.08292294e-01 2.02971578e+00 9.91574600e-02
-1.54646128e-01 -5.47592819e-01 -2.59106189e-01 8.41732025e-01
7.47420713e-02 -6.96501851e-01 -2.64424413e-01 -5.87824404e-01
6.38934597e-02 6.00794613e-01 6.21282160e-01 -7.55484223e-01
-5.38100116e-02 -1.94179252e-01 8.01068842e-01 -1.11655068e+00
1.00557363e+00 -8.70606244e-01 6.27794087e-01 -9.69991237e-02
-5.75714819e-02 -5.32240607e-02 1.27573386e-01 4.45433021e-01
-3.41822654e-02 -1.43230915e-01 1.03661489e+00 -5.16470194e-01
-3.19156677e-01 6.96235538e-01 5.08107757e-03 2.55624920e-01
6.95559800e-01 -5.61961353e-01 -3.55908796e-02 -5.15349150e-01
-7.20204115e-01 -1.84653223e-01 1.06700873e+00 2.87408739e-01
1.02681363e+00 -1.44933116e+00 -8.79444420e-01 5.22984266e-01
-1.19053185e-01 7.54836917e-01 5.94044328e-01 7.57795870e-01
-7.93139696e-01 2.16989458e-01 -9.90233421e-02 -9.90882754e-01
-1.15432680e+00 2.94118524e-01 5.57335138e-01 -1.46300942e-01
-1.22203898e+00 7.71350443e-01 4.72882628e-01 -5.69225371e-01
1.71678886e-02 -3.21609646e-01 6.50841117e-01 -1.95316747e-01
3.75942826e-01 1.02199480e-01 2.55626798e-01 -6.77157342e-01
3.94555405e-02 8.87034893e-01 -1.41779155e-01 -6.48619175e-01
1.66874897e+00 -1.30055740e-01 2.40004823e-01 1.79759085e-01
1.17137551e+00 -3.48301716e-02 -2.07549596e+00 -1.71500966e-01
-9.04773057e-01 -9.97282386e-01 4.13262367e-01 -7.67174065e-01
-1.12992513e+00 6.73825860e-01 4.94638741e-01 -2.00667650e-01
9.62996781e-01 2.50198245e-01 1.04598272e+00 -9.48043689e-02
3.61407042e-01 -8.53277445e-01 4.58072215e-01 3.41556430e-01
1.36541879e+00 -1.13813257e+00 3.61320674e-01 -4.40284163e-01
-5.93218029e-01 1.03169453e+00 5.26294053e-01 -6.60554394e-02
5.70542157e-01 3.89172435e-01 -5.94442114e-02 -3.45628709e-01
-7.01446831e-01 1.38536543e-01 4.40927953e-01 8.16223800e-01
7.66862258e-02 -6.01888001e-01 3.77590805e-01 2.80102305e-02
4.21698391e-02 4.50179279e-02 6.09041333e-01 7.68353760e-01
2.43462715e-02 -9.96518493e-01 -6.06647670e-01 2.56226391e-01
-1.45762384e-01 -2.18177274e-01 -2.02367172e-01 7.83097565e-01
8.53140876e-02 5.40218353e-01 1.42191127e-01 8.89659822e-02
4.99222577e-01 -1.99183911e-01 1.14265549e+00 -6.24123096e-01
-3.84523213e-01 6.08937383e-01 1.98822245e-01 -7.71242797e-01
-5.02433658e-01 -7.24489450e-01 -1.19546175e+00 -2.47661382e-01
-3.44476074e-01 -4.26723659e-01 8.01029444e-01 7.12768018e-01
5.93743205e-01 3.03556800e-01 5.29731214e-01 -1.89500725e+00
-2.88309325e-02 -8.27873588e-01 -5.31440854e-01 4.34487790e-01
4.67445701e-01 -6.24498069e-01 -5.12710452e-01 3.58181894e-01] | [9.236065864562988, -3.1199660301208496] |
418f672d-81af-4781-8531-d26bb8325b33 | treeflow-probabilistic-programming-and | 2211.05220 | null | https://arxiv.org/abs/2211.05220v1 | https://arxiv.org/pdf/2211.05220v1.pdf | TreeFlow: probabilistic programming and automatic differentiation for phylogenetics | Probabilistic programming frameworks are powerful tools for statistical modelling and inference. They are not immediately generalisable to phylogenetic problems due to the particular computational properties of the phylogenetic tree object. TreeFlow is a software library for probabilistic programming and automatic differentiation with phylogenetic trees. It implements inference algorithms for phylogenetic tree times and model parameters given a tree topology. We demonstrate how TreeFlow can be used to quickly implement and assess new models. We also show that it provides reasonable performance for gradient-based inference algorithms compared to specialized computational libraries for phylogenetics. | ['Alexei Drummond', 'Frederick A Matsen IV', 'Marc A Suchard', 'Hassan Nasif', 'Xiang Ji', 'Mathieu Fourment', 'Christiaan Swanepoel'] | 2022-11-09 | null | null | null | null | ['probabilistic-programming'] | ['methodology'] | [ 2.02129215e-01 -5.04377067e-01 -1.92141309e-02 -6.09186530e-01
-4.03965473e-01 -9.26937342e-01 5.14091969e-01 2.70241827e-01
-4.28975672e-01 7.82103479e-01 -3.01396191e-01 -8.65503728e-01
-4.20350820e-01 -7.71282375e-01 -1.49425000e-01 -1.03709745e+00
-4.26064372e-01 9.74897623e-01 5.36904335e-01 2.78816164e-01
6.33427441e-01 7.48980999e-01 -1.58260667e+00 -1.35893151e-01
6.03735209e-01 8.68291333e-02 4.45017815e-01 1.30585003e+00
-2.86834061e-01 4.04164106e-01 -4.98974711e-01 -3.80548418e-01
-3.82680029e-01 -3.14397186e-01 -8.42792392e-01 -6.14772320e-01
-1.98973566e-02 -2.03452572e-01 6.11162931e-02 9.48396444e-01
5.64529717e-01 -1.74247310e-01 7.67710805e-01 -1.30911744e+00
-7.14049190e-02 8.95427227e-01 -3.51182520e-01 4.80365723e-01
3.38889688e-01 2.87526786e-01 9.60014045e-01 -5.55965841e-01
6.65168822e-01 1.74404454e+00 1.27645421e+00 5.16052321e-02
-1.65946388e+00 -1.90864712e-01 -3.30805898e-01 1.34914935e-01
-1.29048431e+00 -2.88818002e-01 1.32718742e-01 -7.52738297e-01
1.09604990e+00 5.81803143e-01 4.39739913e-01 5.93312263e-01
4.48161542e-01 5.67930281e-01 1.25194573e+00 -1.31868184e-01
4.44378406e-01 -3.26487392e-01 3.65042359e-01 7.46865988e-01
-1.58473432e-01 2.72890478e-01 -5.36110520e-01 -9.23053324e-01
6.00583673e-01 -4.57334429e-01 1.30640954e-01 -5.37811518e-02
-1.11624229e+00 9.76115406e-01 -1.33420184e-01 8.44950452e-02
-3.18431437e-01 6.50867641e-01 5.45379341e-01 -2.87247121e-01
3.13871145e-01 1.85393214e-01 -6.39359653e-01 -7.03633666e-01
-9.71254408e-01 6.09990120e-01 9.98147070e-01 8.90021503e-01
4.58523184e-01 7.47846588e-02 3.02590191e-01 9.66880739e-01
7.50948310e-01 2.74235249e-01 1.84696555e-01 -1.47193754e+00
-6.59614205e-01 -5.52713796e-02 -2.05814257e-01 -6.75492167e-01
-5.31109333e-01 -8.78085271e-02 -2.78189480e-01 1.48695663e-01
5.40849566e-01 2.90966988e-01 -7.00675249e-01 1.49800754e+00
7.59597361e-01 -6.19660765e-02 -4.20128942e-01 7.29831904e-02
6.30166769e-01 7.08764195e-01 4.05025363e-01 -3.66026014e-01
1.30600345e+00 -2.15117916e-01 -4.77111757e-01 1.47000268e-01
5.22395194e-01 -8.41055036e-01 5.87696493e-01 4.28987086e-01
-8.09759557e-01 -5.41434363e-02 -6.80307984e-01 -1.65324137e-02
-4.63122725e-01 -5.07726133e-01 1.26448131e+00 1.05221689e+00
-1.30910671e+00 1.10896885e+00 -1.18507588e+00 -4.90259826e-01
2.62658060e-01 1.56590074e-01 -4.09106761e-02 1.18666708e-01
-5.73863328e-01 9.68988001e-01 8.28033030e-01 2.92905122e-01
-8.04249287e-01 -6.63661778e-01 -5.11159897e-01 -1.87147841e-01
-1.34973988e-01 -6.73604548e-01 1.36211193e+00 -6.01821095e-02
-1.61740208e+00 1.15676224e+00 -2.58084774e-01 -2.62607694e-01
3.67644131e-01 2.27405399e-01 2.00853184e-01 -5.52579947e-02
6.78715259e-02 4.87350613e-01 2.27250934e-01 -7.27916956e-01
-5.66071153e-01 -5.58015406e-01 -4.68254387e-01 -1.75548583e-01
4.33803171e-01 7.75576591e-01 -3.11586231e-01 -5.60043335e-01
3.34106624e-01 -8.99359226e-01 -4.92130518e-01 4.65280980e-01
-3.23221892e-01 -4.52563375e-01 3.54797423e-01 -8.95324945e-01
9.59381759e-01 -1.79260707e+00 3.57797027e-01 -7.57898316e-02
-1.28235981e-01 -2.15752035e-01 1.37559563e-01 5.36740899e-01
4.02416587e-02 4.28525090e-01 -9.52400982e-01 9.09253955e-02
4.29414771e-02 7.23321438e-01 4.79700081e-02 7.72928119e-01
-7.09422380e-02 2.48639405e-01 -1.28320169e+00 -8.52182686e-01
4.37560856e-01 4.89155412e-01 -6.72984838e-01 3.14613819e-01
-2.73981601e-01 2.52846658e-01 -1.82176337e-01 7.98668563e-01
9.66458201e-01 1.14845350e-01 7.66576886e-01 4.38439071e-01
-6.23376429e-01 3.75151157e-01 -1.00052714e+00 1.43428266e+00
-8.29086825e-02 5.75333416e-01 1.84797123e-01 -5.62270761e-01
9.64735746e-01 -8.69678631e-02 2.17120454e-01 6.83479726e-01
1.77049674e-02 2.34567985e-01 2.60410279e-01 -2.38952994e-01
1.87108755e-01 -3.47715437e-01 2.30605513e-01 3.05952758e-01
3.47561777e-01 -9.44261193e-01 4.66674417e-01 7.67794624e-02
1.08183479e+00 8.57004881e-01 6.16978228e-01 -7.41151214e-01
3.40933114e-01 5.90294637e-02 7.87809849e-01 8.51417303e-01
-1.18970729e-01 1.03990264e-01 5.22486508e-01 -4.29980397e-01
-9.99190152e-01 -1.61190653e+00 -1.06254649e+00 1.61090529e+00
-5.02646327e-01 -6.33346915e-01 -4.21917677e-01 -4.63907570e-02
2.34409168e-01 1.04443204e+00 -3.71572226e-01 3.40929478e-01
-1.68222547e-01 -1.38127923e+00 8.75748396e-01 2.69433737e-01
-3.11493635e-01 -1.03792655e+00 -7.14177191e-01 5.11819005e-01
3.02469507e-02 -5.17982364e-01 1.27412096e-01 5.98057270e-01
-1.25421798e+00 -1.11526299e+00 -3.34518790e-01 -4.10595894e-01
1.51547059e-01 -2.65837759e-01 9.86592233e-01 8.94429907e-03
-5.67395508e-01 -4.29651774e-02 -3.04085854e-03 -3.30500096e-01
-8.83755386e-01 -2.12478101e-01 7.51386806e-02 -8.91511917e-01
2.88359255e-01 -1.00775540e+00 -3.68391648e-02 3.94393772e-01
-6.91061616e-01 -3.75804640e-02 -2.96482027e-01 1.11984742e+00
5.24035871e-01 2.14102700e-01 -1.17416486e-01 -7.88531601e-01
4.10509169e-01 -6.42126620e-01 -1.15569901e+00 3.66911978e-01
-3.17213893e-01 2.59343415e-01 2.72189140e-01 -2.60653719e-02
-9.96967375e-01 2.28395164e-02 -6.14989936e-01 3.55162233e-01
1.71556268e-02 7.40365446e-01 1.31584883e-01 7.05032051e-02
3.42602015e-01 1.32487983e-01 -4.79561277e-02 -7.77122617e-01
2.57259548e-01 7.34447658e-01 7.25796521e-01 -1.04670453e+00
2.00853646e-01 3.59792620e-01 3.87465924e-01 -1.03493655e+00
-1.96984619e-01 -2.73804635e-01 -9.56095278e-01 -2.37560704e-01
5.81470251e-01 -3.66776735e-01 -1.14463794e+00 7.16825187e-01
-1.17462897e+00 -1.72436655e-01 2.60636985e-01 3.92041385e-01
-8.14091384e-01 7.37010717e-01 -8.01365018e-01 -1.23454130e+00
-1.71793103e-01 -1.06399918e+00 8.99048150e-01 1.07360326e-01
-3.21363747e-01 -1.21785474e+00 6.63741231e-01 -6.52555525e-02
1.69669867e-01 5.09836435e-01 9.88760412e-01 -3.88236582e-01
-3.28103527e-02 4.92000319e-02 7.02135190e-02 -3.48940939e-02
-4.55728769e-02 1.57823086e+00 -9.91142929e-01 4.92397025e-02
-6.45759255e-02 -3.55643150e-03 6.49805129e-01 5.58231652e-01
1.40463519e+00 -1.52487323e-01 -6.38633251e-01 9.16697919e-01
1.28915393e+00 2.96974391e-01 3.44878465e-01 2.45422542e-01
3.98766279e-01 1.10380626e+00 4.62075502e-01 6.65980279e-01
3.93955201e-01 6.15293682e-01 1.48005500e-01 6.72254920e-01
4.04608607e-01 -1.74554661e-01 1.53603747e-01 9.06639755e-01
-9.09236446e-02 9.24352091e-03 -1.62908483e+00 5.48776567e-01
-1.66213131e+00 -1.14013517e+00 -8.48318517e-01 2.16154075e+00
1.31405044e+00 -1.65599436e-01 2.67760545e-01 -1.85611509e-02
9.35331285e-01 -1.26303554e-01 -6.24893486e-01 -8.65431905e-01
-2.02715829e-01 3.42630982e-01 4.07743782e-01 5.67421377e-01
-6.46250844e-01 8.78927052e-01 9.00831699e+00 7.51668990e-01
-3.80065560e-01 3.74047160e-02 2.58853495e-01 1.66354671e-01
-4.17746931e-01 4.78607029e-01 -6.69737637e-01 6.15203559e-01
1.40289450e+00 -3.73013377e-01 3.51137042e-01 9.56177592e-01
2.95524970e-02 -2.93764353e-01 -1.14618301e+00 5.89860857e-01
-7.12838709e-01 -1.23786032e+00 -4.26651120e-01 9.82696116e-02
-3.10422317e-03 5.14545918e-01 -3.52914333e-01 -1.02399901e-01
1.22578645e+00 -9.47951138e-01 6.16456270e-01 2.65453190e-01
3.88600856e-01 -8.41802299e-01 6.13646507e-01 2.69834757e-01
-7.03582108e-01 3.06384355e-01 -7.47340143e-01 -1.26724154e-01
4.48798954e-01 8.02366614e-01 -1.28881097e+00 3.04213017e-01
1.14101899e+00 4.34762686e-01 -6.92666769e-01 1.46741807e+00
-3.67183179e-01 8.43494713e-01 -9.33722675e-01 -1.07002899e-01
4.51658331e-02 -3.37159157e-01 6.25795186e-01 1.44854343e+00
2.52148628e-01 -2.62290448e-01 -9.72697809e-02 8.56331229e-01
8.48833978e-01 -1.00888215e-01 -4.87781763e-01 -2.26838335e-01
9.75143552e-01 1.09563053e+00 -1.11053729e+00 -3.27090502e-01
1.89478904e-01 5.29956698e-01 1.75764173e-01 -1.87589943e-01
-7.41269648e-01 -3.61563742e-01 7.81987011e-01 -3.98078740e-01
3.51232141e-01 -4.91400391e-01 -2.93940932e-01 -7.80411601e-01
-7.04728842e-01 -7.78291523e-01 8.16715300e-01 -1.02546632e+00
-1.45187664e+00 2.74962962e-01 4.42507714e-01 -2.24024832e-01
-6.50295496e-01 -1.03606224e+00 -5.61231792e-01 9.51749921e-01
-6.84646487e-01 -6.84754193e-01 2.70434022e-01 -2.09991960e-03
9.45607126e-02 4.41328824e-01 1.10340762e+00 -4.19023901e-01
-7.48878121e-01 7.80620724e-02 7.43119419e-01 -4.72852409e-01
1.87613070e-01 -1.67245054e+00 8.68838489e-01 8.29003036e-01
-2.14116246e-01 9.95858729e-01 1.33108377e+00 -7.03585327e-01
-1.49061060e+00 -3.48303437e-01 6.97611392e-01 -5.76988280e-01
1.20282042e+00 -4.26270396e-01 -9.53891218e-01 6.47539794e-01
-1.75939761e-02 -4.06734943e-01 1.05529201e+00 5.77982128e-01
-5.16710043e-01 4.09278929e-01 -1.33282351e+00 5.92942953e-01
1.09816170e+00 -6.40122771e-01 -6.17392898e-01 5.87898493e-01
4.15770769e-01 -8.22785348e-02 -1.52732658e+00 2.45344743e-01
1.01665902e+00 -1.07226276e+00 9.53264415e-01 -5.58480799e-01
3.65039259e-02 -7.41294086e-01 -3.33504438e-01 -8.66050243e-01
-4.15174156e-01 -8.24911714e-01 1.47879437e-01 1.39192128e+00
1.88342795e-01 -6.53906822e-01 3.28475177e-01 2.16981173e-01
1.26287162e-01 -1.32754117e-01 -1.37205291e+00 -8.19663405e-01
4.42404926e-01 -6.15351498e-01 5.87922156e-01 9.65627074e-01
-4.12390269e-02 4.19256613e-02 -4.79409769e-02 2.62797385e-01
1.10988700e+00 1.03065044e-01 7.19219446e-01 -1.37888908e+00
-7.16485977e-01 -7.27306604e-01 -6.30905211e-01 -4.15030777e-01
-2.32336149e-02 -8.17459822e-01 5.94010711e-01 -1.21161664e+00
2.98265040e-01 -5.92244446e-01 8.25232491e-02 4.74968642e-01
-4.77361381e-01 -1.35069981e-01 -2.21619859e-01 1.35544792e-01
1.52800024e-01 2.39006460e-01 2.44284064e-01 3.85841787e-01
2.24286035e-01 3.28950174e-02 -2.47765943e-01 8.57187927e-01
7.70123482e-01 -1.08023715e+00 8.45454261e-02 -2.00238720e-01
2.17616141e-01 9.86116529e-02 2.33163744e-01 -5.87175012e-01
-3.01094670e-02 -3.86976838e-01 7.30292127e-02 -1.08781338e+00
1.18268915e-01 -1.51590407e-01 9.07844961e-01 9.16549981e-01
-5.79315871e-02 4.19886708e-01 3.22068423e-01 4.11025941e-01
2.60123998e-01 -7.82944381e-01 8.92407298e-01 -1.54398054e-01
-4.85060662e-01 -4.00021560e-02 -9.90440845e-01 -3.45712036e-01
7.08447874e-01 -1.65322974e-01 -3.87983829e-01 1.57089934e-01
-6.33107960e-01 2.58350313e-01 8.42024803e-01 -6.25632107e-02
2.53563404e-01 -6.62744880e-01 -9.75144863e-01 -4.44062978e-01
2.85495594e-02 -3.15864772e-01 1.51552677e-01 6.80607617e-01
-1.39718831e+00 7.44266212e-02 -5.19797146e-01 -1.01861989e+00
-1.62757099e+00 8.12109590e-01 2.20186204e-01 3.56084155e-03
-3.37800711e-01 1.12019551e+00 -1.07707649e-01 -1.11796498e+00
6.65293681e-03 -6.96751801e-03 2.65208542e-01 -1.72706351e-01
6.67843521e-01 6.76311731e-01 -5.99094868e-01 -4.85596776e-01
-9.21493232e-01 3.41746211e-01 9.99035537e-02 -3.31882089e-01
1.59652340e+00 -3.03304531e-02 -9.93310809e-01 1.03243446e+00
7.70024955e-01 -2.25628793e-01 -9.96619701e-01 1.53011814e-01
6.93362176e-01 -4.87997830e-01 -1.40199825e-01 -7.52659798e-01
-1.95808589e-01 1.03522730e+00 2.88200974e-01 1.84834808e-01
7.55525947e-01 -3.21948268e-02 -2.62756437e-01 2.25494206e-01
4.10946757e-01 -6.84643924e-01 -1.00392675e+00 4.53507572e-01
6.76251829e-01 -5.91119349e-01 4.73673522e-01 -5.13798416e-01
1.58718809e-01 1.37619472e+00 2.23245621e-02 2.71219909e-01
6.16231382e-01 5.60984075e-01 -2.85632700e-01 -1.72518864e-01
-1.02063644e+00 -8.47840905e-02 -2.70311087e-01 1.04442143e+00
7.94040561e-01 2.35473305e-01 -3.66672903e-01 -1.85207799e-01
-5.90077996e-01 -1.67601794e-01 5.35382628e-01 1.03732646e+00
-4.42342997e-01 -1.26822031e+00 -5.87181032e-01 2.18936950e-01
-5.14304996e-01 -3.55548114e-01 -4.00470495e-01 5.92805862e-01
-1.76965445e-01 7.63801813e-01 1.25966325e-01 -4.43309397e-02
-5.24368703e-01 2.35246509e-01 7.54263461e-01 -5.90116739e-01
-6.09821916e-01 7.25976676e-02 4.79416400e-01 -4.00679678e-01
-4.24577236e-01 -1.16441286e+00 -8.82659197e-01 -8.95241022e-01
-5.11739373e-01 3.81069928e-01 1.18960416e+00 4.59984660e-01
-1.58058740e-02 1.57585338e-01 3.05847853e-01 -6.78587139e-01
-4.99262244e-01 -9.57619488e-01 -6.14133298e-01 -5.60684621e-01
-3.21549028e-01 -5.78257143e-01 -4.53372985e-01 -4.62946482e-02] | [4.9459452629089355, 5.0653204917907715] |
535a1123-d0e4-4a71-a4a2-2946a5175d12 | instant-neural-graphics-primitives-with-a | 2201.05989 | null | https://arxiv.org/abs/2201.05989v2 | https://arxiv.org/pdf/2201.05989v2.pdf | Instant Neural Graphics Primitives with a Multiresolution Hash Encoding | Neural graphics primitives, parameterized by fully connected neural networks, can be costly to train and evaluate. We reduce this cost with a versatile new input encoding that permits the use of a smaller network without sacrificing quality, thus significantly reducing the number of floating point and memory access operations: a small neural network is augmented by a multiresolution hash table of trainable feature vectors whose values are optimized through stochastic gradient descent. The multiresolution structure allows the network to disambiguate hash collisions, making for a simple architecture that is trivial to parallelize on modern GPUs. We leverage this parallelism by implementing the whole system using fully-fused CUDA kernels with a focus on minimizing wasted bandwidth and compute operations. We achieve a combined speedup of several orders of magnitude, enabling training of high-quality neural graphics primitives in a matter of seconds, and rendering in tens of milliseconds at a resolution of ${1920\!\times\!1080}$. | ['Alexander Keller', 'Christoph Schied', 'Alex Evans', 'Thomas Müller'] | 2022-01-16 | null | null | null | null | ['neural-radiance-caching'] | ['computer-vision'] | [-6.88100010e-02 -2.32529957e-02 7.00097084e-02 -2.26888761e-01
-6.84196591e-01 -3.94576937e-01 4.06264335e-01 6.04176462e-01
-1.00757062e+00 5.26807666e-01 -2.45208055e-01 -5.64951122e-01
2.82760531e-01 -1.40029216e+00 -9.37956154e-01 -6.19949520e-01
-2.51850247e-01 3.95529419e-01 5.68762183e-01 -9.10661444e-02
2.61519939e-01 8.45454276e-01 -2.19800282e+00 3.12295526e-01
4.19087678e-01 1.37874508e+00 -5.62581234e-03 8.14884841e-01
-1.14231952e-01 4.50043976e-01 -4.77561802e-01 -3.83308381e-01
5.08671463e-01 2.37836510e-01 -3.66119951e-01 -3.82450521e-01
7.83585608e-01 -6.02737725e-01 -1.99418157e-01 7.71567643e-01
6.60574317e-01 3.14988554e-01 2.35792443e-01 -9.39895093e-01
-3.18087841e-04 3.41103166e-01 -4.51971114e-01 1.45195141e-01
-7.38165155e-02 1.99601352e-01 9.81499076e-01 -6.46403730e-01
3.28998476e-01 9.68595922e-01 8.57307792e-01 1.69830397e-01
-1.30763113e+00 -7.91569233e-01 -1.59931853e-01 -6.95844740e-02
-1.42150402e+00 -4.05468613e-01 1.33832857e-01 -1.15168206e-01
1.45772946e+00 4.17004406e-01 8.95657063e-01 5.27381003e-01
1.70588076e-01 2.84403026e-01 5.82060874e-01 -2.20920786e-01
5.35009682e-01 -2.39126951e-01 3.27468753e-01 1.15852928e+00
4.80852932e-01 2.68357657e-02 -3.73971045e-01 -3.61243010e-01
1.20810592e+00 -1.97892282e-02 -8.67026374e-02 -8.02431032e-02
-1.11556005e+00 1.09220290e+00 4.41984802e-01 -1.96022928e-01
-3.09312642e-01 5.31155109e-01 7.06571102e-01 1.56740785e-01
8.97288416e-03 4.01722461e-01 -4.20428455e-01 -3.29084963e-01
-9.16417480e-01 4.26405340e-01 1.06669831e+00 7.29662359e-01
1.10424352e+00 2.89679676e-01 1.83439925e-01 5.80791831e-01
-4.60236371e-02 4.74714220e-01 2.21256614e-01 -1.14914489e+00
9.03926045e-02 4.90454555e-01 -1.68842196e-01 -1.04876888e+00
-5.96031308e-01 -2.83950329e-01 -1.02527499e+00 7.68819451e-01
3.71225685e-01 -3.18266511e-01 -7.08407879e-01 1.37350011e+00
6.25668049e-01 1.07476719e-01 -3.49621564e-01 6.26483858e-01
4.23432946e-01 8.54342699e-01 -8.75861198e-02 1.45388797e-01
1.49503410e+00 -9.07755733e-01 1.48649558e-01 2.36671157e-02
7.15925276e-01 -7.31752336e-01 1.28494918e+00 5.19702196e-01
-1.27499688e+00 -4.89770591e-01 -1.42050958e+00 -4.70539421e-01
-2.55322367e-01 -2.56932765e-01 1.26408267e+00 6.50834322e-01
-1.33637071e+00 9.88288701e-01 -1.30592430e+00 1.76854923e-01
4.11363214e-01 9.87908959e-01 -2.68234223e-01 2.55071402e-01
-7.08086610e-01 4.47080910e-01 6.00760400e-01 -2.46132910e-01
-3.48538369e-01 -9.84008253e-01 -9.16419625e-01 3.79576683e-01
3.94769311e-02 -9.04255152e-01 9.74130034e-01 -5.75542688e-01
-1.55745137e+00 5.63330352e-01 2.67142639e-03 -6.89993203e-01
1.47538364e-01 -1.52437806e-01 8.75679925e-02 1.25134796e-01
-2.05346778e-01 8.13548565e-01 7.65159249e-01 -5.54115593e-01
-8.03533852e-01 -2.85023570e-01 1.47849228e-03 8.38211998e-02
-5.91939092e-01 -6.94807991e-02 -6.32782400e-01 -5.66680908e-01
4.92866188e-02 -9.83258009e-01 -3.80803883e-01 3.33779454e-01
-1.80326015e-01 1.01364870e-02 6.16792142e-01 1.24298986e-02
9.62012827e-01 -1.94814467e+00 -3.36199403e-01 2.87244946e-01
4.55858111e-01 2.18444690e-01 1.20130688e-01 -5.33800758e-02
2.20281154e-01 -1.03307493e-01 -1.55605108e-01 -3.46119195e-01
-7.97042251e-02 3.63301456e-01 -2.91944206e-01 3.60006690e-01
5.96383363e-02 6.77773178e-01 -6.03107929e-01 -2.25539461e-01
1.95237964e-01 8.19878221e-01 -1.00812566e+00 5.03103733e-02
-1.49175540e-01 -3.04071996e-02 -3.95404756e-01 4.01913881e-01
6.01622999e-01 -5.00613213e-01 -7.75727555e-02 -2.03737348e-01
-4.92597938e-01 5.84803164e-01 -1.28998411e+00 1.64474678e+00
-5.48854589e-01 5.06362498e-01 -1.92828954e-03 -6.54774427e-01
7.96760440e-01 -1.14342965e-01 2.80022860e-01 -6.40533924e-01
3.59991908e-01 7.08689094e-02 -2.31543198e-01 3.34843546e-01
8.69975150e-01 -3.04203760e-03 -2.24476159e-01 7.99244523e-01
-7.74518475e-02 -8.16395059e-02 1.04977272e-01 -1.55025184e-01
1.16144359e+00 -6.96768635e-04 -3.48461345e-02 -4.12112415e-01
3.09142768e-02 8.58428925e-02 4.65218753e-01 7.82507300e-01
3.40837061e-01 4.23620969e-01 5.85048556e-01 -1.06041384e+00
-1.37590718e+00 -9.47317064e-01 -2.49955580e-01 1.65391600e+00
-1.39825478e-01 -6.33950889e-01 -8.42935801e-01 9.86458212e-02
5.86981438e-02 3.89510155e-01 -4.45725352e-01 4.32156511e-02
-1.03293753e+00 -9.21570837e-01 7.77101040e-01 7.32926548e-01
5.19963980e-01 -8.46890271e-01 -1.24094975e+00 2.88061500e-01
7.21615136e-01 -8.85775089e-01 -1.19290374e-01 5.62207937e-01
-1.17455649e+00 -6.29626691e-01 -3.33042830e-01 -6.64877713e-01
5.38909197e-01 6.11380488e-02 1.29895139e+00 4.46860015e-01
-4.94512111e-01 -3.89864385e-01 2.54000127e-01 -3.37773487e-02
-1.04807891e-01 3.19268376e-01 9.28041264e-02 -4.75998253e-01
1.82367697e-01 -9.88263965e-01 -7.71452725e-01 -7.82727972e-02
-8.99794042e-01 4.15047139e-01 4.39119041e-01 9.36492622e-01
9.04858172e-01 -6.46406412e-02 -2.39908352e-01 -8.94906104e-01
2.37096697e-01 -1.17514506e-01 -1.26513660e+00 -3.17618012e-01
-4.91924107e-01 4.55462724e-01 9.33082521e-01 -4.57436621e-01
-6.27942801e-01 1.76717669e-01 -5.40897548e-01 -3.02946955e-01
-2.77030710e-02 1.34767503e-01 3.32188278e-01 -5.62385917e-01
7.49770224e-01 3.55997086e-02 2.53552534e-02 -2.77856469e-01
4.78347033e-01 -2.82013621e-02 7.50280440e-01 -9.74910736e-01
4.69899356e-01 5.02146780e-01 2.55324930e-01 -7.32792079e-01
-3.84485543e-01 6.79391623e-02 -3.77529949e-01 3.25923622e-01
6.95427716e-01 -8.98665965e-01 -1.33821857e+00 3.72328907e-01
-1.05629563e+00 -5.84205210e-01 -3.81055295e-01 2.44620040e-01
-3.80418122e-01 3.63257676e-02 -1.04489672e+00 -2.43285492e-01
-6.69877172e-01 -1.43331194e+00 1.05790734e+00 1.98186383e-01
-2.02564240e-01 -5.75384736e-01 -1.15409613e-01 -1.68459833e-01
5.66721559e-01 3.51259232e-01 1.09400487e+00 -3.61854196e-01
-1.03532541e+00 -1.32613078e-01 -6.36809409e-01 1.63404159e-02
-4.29427713e-01 7.35411942e-02 -8.83364379e-01 -3.00678283e-01
2.99093593e-02 -3.40270430e-01 8.50001812e-01 1.21465862e-01
1.32467937e+00 -3.37052345e-01 -9.20025781e-02 1.23271894e+00
1.50861895e+00 -8.55952650e-02 6.77187681e-01 3.94930899e-01
8.97347987e-01 5.81459962e-02 1.40255513e-02 7.84054816e-01
3.32167536e-01 5.17380178e-01 4.44270223e-01 -1.72765285e-01
8.68580416e-02 9.67225805e-02 -1.72061965e-01 8.64002347e-01
-2.46319652e-01 1.24507755e-01 -1.05760539e+00 9.17389467e-02
-1.50906336e+00 -5.49797058e-01 1.19728528e-01 2.55959177e+00
8.42481017e-01 4.29130912e-01 -1.66347134e-03 6.78279251e-02
3.19526047e-01 2.52973616e-01 -5.12217641e-01 -6.33357644e-01
9.55428407e-02 1.01738918e+00 8.96386087e-01 5.76866329e-01
-1.04868770e+00 1.12843168e+00 6.45747328e+00 8.48138034e-01
-1.57455146e+00 -6.82346225e-02 9.49347913e-01 -3.93876076e-01
-2.41235986e-01 -9.11226943e-02 -1.00950825e+00 2.61775672e-01
1.23638344e+00 -1.82571970e-02 5.20875514e-01 1.14524293e+00
-3.85772794e-01 -2.13807374e-01 -1.01480281e+00 1.18474412e+00
-2.80949533e-01 -1.65861082e+00 1.53085545e-01 2.74729699e-01
4.98322099e-01 5.35111129e-01 1.82392821e-01 1.79274634e-01
4.70239699e-01 -1.11524796e+00 6.28183484e-01 8.24523252e-03
9.72824633e-01 -1.29237223e+00 2.49021128e-01 2.47989148e-01
-1.14936233e+00 4.40673344e-02 -7.33333707e-01 -3.69030058e-01
1.86430402e-02 5.19696951e-01 -5.92964768e-01 -1.54716998e-01
8.05524409e-01 -3.18098702e-02 -3.60418320e-01 7.05557644e-01
2.07035348e-01 2.96163291e-01 -9.78006244e-01 -1.85684621e-01
3.00235301e-01 -2.12767031e-02 8.27571228e-02 1.20925558e+00
4.05501813e-01 2.20813289e-01 -2.99083011e-04 6.63662672e-01
-2.92808026e-01 5.75592667e-02 -3.84625614e-01 2.15064406e-01
5.66214979e-01 1.30882585e+00 -1.18755031e+00 -4.78783339e-01
-3.33971083e-01 1.08664978e+00 6.40156269e-01 -1.89889014e-01
-7.33646870e-01 -7.05899596e-01 8.52918804e-01 5.79562113e-02
5.77270687e-01 -6.75788939e-01 -7.47913599e-01 -1.08161259e+00
-1.25298262e-01 -6.89265013e-01 9.88748074e-02 -2.84619302e-01
-6.25223100e-01 8.41279566e-01 -4.45961118e-01 -7.06614256e-01
-4.38778311e-01 -7.64099896e-01 -3.87697399e-01 7.52700627e-01
-1.08600771e+00 -7.47396410e-01 -3.16652507e-01 5.45601726e-01
-1.52030727e-03 4.19534044e-03 1.12246931e+00 4.52246189e-01
-4.25340027e-01 8.59309733e-01 -1.43790692e-01 -5.42509975e-03
2.62071192e-01 -1.04964840e+00 1.16068578e+00 4.14305091e-01
8.43911320e-02 8.81408393e-01 4.90009010e-01 -3.53693306e-01
-1.80529213e+00 -9.01999891e-01 5.39229691e-01 -7.95579851e-02
6.27604842e-01 -6.03101432e-01 -1.18819439e+00 4.31522876e-01
-2.00443551e-01 3.53156209e-01 7.62214959e-01 2.09208027e-01
-6.51722789e-01 -7.05439895e-02 -9.92704272e-01 7.94349253e-01
8.94801199e-01 -5.11970222e-01 1.11121863e-01 3.28341931e-01
7.70196021e-01 -8.66847038e-01 -1.00781989e+00 1.76120862e-01
7.55887985e-01 -1.13445914e+00 1.26739609e+00 -3.36571336e-01
1.24413617e-01 -1.21127822e-01 -1.45368174e-01 -4.56499279e-01
-3.46758246e-01 -6.29874825e-01 -8.67635086e-02 4.73411709e-01
1.97269022e-01 -7.51909494e-01 1.17916918e+00 8.00091505e-01
-1.61334246e-01 -1.01093137e+00 -9.30541813e-01 -2.97969222e-01
4.88525704e-02 -6.94406867e-01 7.37771451e-01 7.16460228e-01
-3.00629884e-01 2.29091227e-01 -9.29852650e-02 2.26603001e-01
4.82989818e-01 9.90400091e-02 7.21719801e-01 -1.13636053e+00
-7.49622762e-01 -6.45049691e-01 -6.39982641e-01 -1.22244895e+00
3.26530216e-03 -6.64239228e-01 -1.94704145e-01 -7.19732463e-01
-1.32327184e-01 -9.12197709e-01 -1.85879514e-01 7.30952501e-01
1.03864513e-01 7.46977329e-01 1.74690690e-02 1.06525362e-01
-4.59935367e-01 2.36986563e-01 8.02986979e-01 4.04720277e-01
-1.15685880e-01 -3.40535700e-01 -6.56715572e-01 9.60584223e-01
5.49991488e-01 -4.10022676e-01 -2.68077433e-01 -9.89869595e-01
3.69122177e-01 1.16496503e-01 3.72950196e-01 -1.41676116e+00
4.93312418e-01 1.81830078e-01 5.96067429e-01 -4.93344009e-01
6.97520971e-01 -3.70057434e-01 3.83844934e-02 4.19545293e-01
-1.49274319e-01 5.79568267e-01 7.70563722e-01 1.28363639e-01
1.12470500e-01 -2.89966371e-02 9.00884390e-01 -5.33377975e-02
-4.63623762e-01 5.16888440e-01 -1.98039323e-01 -2.31064215e-01
7.41751492e-01 -1.94902733e-01 -1.29035652e-01 -1.04581565e-01
-2.66856015e-01 -2.26796567e-01 8.21420252e-01 -1.43020228e-01
3.56867760e-01 -1.19835472e+00 -2.79587358e-01 4.51879144e-01
-4.69329357e-01 3.72038037e-01 2.38537893e-01 1.84030935e-01
-1.21851659e+00 2.98809558e-01 -5.32622337e-01 -5.67699969e-01
-1.16800952e+00 3.55397940e-01 1.04667507e-01 -2.94520825e-01
-8.84566963e-01 1.24392784e+00 1.60109028e-01 -1.09087206e-01
1.53372183e-01 -2.81517208e-01 3.22391480e-01 -2.12230444e-01
9.05469060e-01 4.60694879e-01 4.60231513e-01 -4.34166282e-01
-6.09787889e-02 4.03257132e-01 -2.01980084e-01 -1.38569772e-01
1.30289972e+00 5.09279490e-01 -3.69731456e-01 4.90904637e-02
1.44344676e+00 -8.72084349e-02 -1.47648692e+00 -1.27960285e-02
-2.33924910e-01 -1.48621887e-01 3.53179544e-01 -1.51846692e-01
-1.03081405e+00 9.46758747e-01 5.35881281e-01 -1.05903938e-01
1.02776074e+00 -4.45467412e-01 1.37927663e+00 8.33421946e-01
6.68729663e-01 -7.85525918e-01 -2.39825308e-01 8.51468205e-01
2.28015274e-01 -8.51941586e-01 1.24619476e-01 -2.90905863e-01
-3.21101733e-02 1.38582802e+00 3.44643801e-01 -6.98171079e-01
5.62287033e-01 9.12857771e-01 -1.73929244e-01 -4.39194143e-02
-6.66596055e-01 1.59443766e-01 1.44147038e-01 2.37745911e-01
2.76149243e-01 3.46611813e-03 4.94651459e-02 2.07298711e-01
-6.23721659e-01 -1.59413770e-01 7.06575364e-02 1.15491807e+00
-6.53143346e-01 -1.30593264e+00 -9.83613580e-02 5.98964334e-01
-4.99680489e-01 -5.56516230e-01 5.26047230e-01 7.34421551e-01
-6.03363961e-02 1.95821717e-01 7.40769088e-01 -4.10791457e-01
7.29813874e-02 -1.14049539e-01 4.85280544e-01 -5.45403600e-01
-9.13621664e-01 -3.86913829e-02 -9.72775146e-02 -9.07187521e-01
4.17857468e-01 -1.93244338e-01 -1.57610857e+00 -8.20133328e-01
1.08888566e-01 -8.18758160e-02 9.80904639e-01 5.86112797e-01
7.78910697e-01 3.26766580e-01 2.44919568e-01 -1.33512890e+00
-3.87906194e-01 -1.11241594e-01 -2.43619263e-01 -4.94687334e-02
1.56422555e-01 -4.68165368e-01 6.91916198e-02 -1.88338995e-01] | [8.50973129272461, 2.9681661128997803] |
b9b3c1d9-b79c-4515-ae96-a5c4d4a392fa | safe-motion-planning-with-environment | 2305.06004 | null | https://arxiv.org/abs/2305.06004v1 | https://arxiv.org/pdf/2305.06004v1.pdf | Safe motion planning with environment uncertainty | We present an approach for safe motion planning under robot state and environment (obstacle and landmark location) uncertainties. To this end, we first develop an approach that accounts for the landmark uncertainties during robot localization. Existing planning approaches assume that the landmark locations are well known or are known with little uncertainty. However, this might not be true in practice. Noisy sensors and imperfect motions compound to the errors originating from the estimate of environment features. Moreover, possible occlusions and dynamic objects in the environment render imperfect landmark estimation. Consequently, not considering this uncertainty can wrongly localize the robot, leading to inefficient plans. Our approach thus incorporates the landmark uncertainty within the Bayes filter estimation framework. We also analyze the effect of considering this uncertainty and delineate the conditions under which it can be ignored. Second, we extend the state-of-the-art by computing an exact expression for the collision probability under Gaussian distributed robot motion, perception and obstacle location uncertainties. We formulate the collision probability process as a quadratic form in random variables. Under Gaussian distribution assumptions, an exact expression for collision probability is thus obtained which is computable in real-time. In contrast, existing approaches approximate the collision probability using upper-bounds that can lead to overly conservative estimate and thereby suboptimal plans. We demonstrate and evaluate our approach using a theoretical example and simulations. We also present a comparison of our approach to different state-of-the-art methods. | ['Marco Baglietto', 'Fulvio Mastrogiovanni', 'Antony Thomas'] | 2023-05-10 | null | null | null | null | ['motion-planning'] | ['robots'] | [ 9.39433351e-02 4.28793550e-01 4.90594096e-02 -2.01751500e-01
-1.04247308e+00 -4.85692352e-01 5.79831183e-01 4.32825744e-01
-8.11728120e-01 1.01153004e+00 -1.59858137e-01 -1.93779439e-01
-4.05393153e-01 -1.03060210e+00 -9.97683823e-01 -7.20340133e-01
-2.33214974e-01 8.54402184e-01 5.84519446e-01 -8.01676586e-02
3.29546601e-01 5.60006201e-01 -1.34816432e+00 -7.07322896e-01
7.57861078e-01 7.58485377e-01 6.19594753e-01 7.21522748e-01
-7.55157694e-02 3.95786196e-01 -6.23525441e-01 1.16647109e-01
2.46072754e-01 5.12411408e-02 -6.88708663e-01 -2.81982720e-02
-4.21452403e-01 -4.12227273e-01 -1.84867561e-01 1.24779451e+00
2.88036168e-01 3.05906683e-01 8.57869208e-01 -1.33084095e+00
1.79752052e-01 3.60721499e-01 -2.89812744e-01 -2.18397513e-01
4.45545912e-01 -2.72393506e-02 3.39333653e-01 -6.63355589e-01
4.76542592e-01 1.35623240e+00 7.86608994e-01 1.68155730e-01
-1.01501310e+00 -4.14141156e-02 2.32730508e-01 6.47531226e-02
-1.75250220e+00 -3.67376328e-01 4.03047860e-01 -2.62818575e-01
7.68331885e-01 -4.68617380e-02 5.90081997e-02 7.00465202e-01
6.93356216e-01 4.43940967e-01 7.49882340e-01 -4.85950291e-01
6.36754155e-01 -1.31856263e-01 -8.88242349e-02 4.76477325e-01
7.43485749e-01 3.58209431e-01 -2.28780508e-02 -2.00752273e-01
5.48502147e-01 -1.22217610e-01 -6.91710785e-02 -6.13081396e-01
-1.16591716e+00 6.64231360e-01 3.61805379e-01 1.46296293e-01
-7.27781594e-01 5.25106788e-01 1.49653763e-01 -1.42930806e-01
-1.18968561e-01 3.54310013e-02 -1.40489623e-01 -2.15739980e-01
-3.83111507e-01 4.29198295e-01 8.67560089e-01 1.23996902e+00
9.52900350e-01 -1.57075897e-01 2.71755964e-01 3.54696453e-01
6.35709047e-01 7.20146418e-01 1.03824371e-02 -1.14843428e+00
4.65306640e-01 1.88686047e-02 8.96163285e-01 -1.00283861e+00
-8.14201713e-01 -1.69297367e-01 -4.93453771e-01 5.16196668e-01
7.27432668e-01 -2.14386374e-01 -9.15071547e-01 1.56232905e+00
4.13690925e-01 -4.70448472e-02 4.28139180e-01 6.61204696e-01
-3.69117483e-02 5.19533515e-01 -1.24575719e-02 -3.40371311e-01
1.09199798e+00 -5.17858088e-01 -1.13131070e+00 -4.68920559e-01
5.91023445e-01 -7.88595855e-01 4.85123008e-01 3.74753058e-01
-7.87988484e-01 -2.73326248e-01 -1.11591899e+00 3.17119539e-01
-1.91981882e-01 -1.01487122e-01 2.69486219e-01 5.92530370e-01
-8.72709334e-01 5.76263011e-01 -1.21959472e+00 -5.00688076e-01
-1.75794974e-01 3.73499811e-01 -1.53942809e-01 -2.55798727e-01
-9.35718596e-01 1.43105996e+00 4.40221637e-01 3.72995257e-01
-9.89045620e-01 8.76465961e-02 -9.93555486e-01 -2.31519714e-01
8.90854180e-01 -5.00326037e-01 1.57062030e+00 -1.23805769e-01
-1.58146572e+00 -1.42377615e-01 -4.07593518e-01 -5.65655351e-01
6.19758427e-01 -3.63751620e-01 5.09955734e-02 -1.16350390e-01
2.88604498e-01 1.23240046e-01 2.72184819e-01 -1.76696718e+00
-7.42407262e-01 -2.74544120e-01 3.01019877e-01 2.94943094e-01
6.12900972e-01 -5.27179062e-01 -3.72573704e-01 4.80200909e-03
6.98413312e-01 -1.05747044e+00 -1.01769257e+00 -1.70274526e-01
-4.76758152e-01 9.94391441e-02 1.97337836e-01 -2.24859223e-01
8.96250427e-01 -1.95617247e+00 -7.87781328e-02 4.98228461e-01
-3.45960528e-01 -4.20754403e-01 1.90745771e-01 7.76257217e-01
7.10051298e-01 -1.20776203e-02 -3.37277025e-01 -6.08723402e-01
2.29572803e-01 8.05953622e-01 -2.34091118e-01 9.36328411e-01
-1.60104573e-01 4.20476019e-01 -1.13663459e+00 -4.17312086e-01
5.89825273e-01 3.93486291e-01 -2.72813112e-01 -9.53130350e-02
-6.11199848e-02 3.35225642e-01 -7.71603763e-01 4.88688469e-01
6.85486317e-01 3.99133116e-01 2.41398752e-01 1.01996571e-01
-5.21039367e-01 2.17391849e-01 -1.64366412e+00 1.48871779e+00
-5.59610307e-01 1.84324354e-01 3.14759374e-01 -7.95183361e-01
8.43174100e-01 2.18795851e-01 4.09201652e-01 -3.18215102e-01
3.22738171e-01 5.53560197e-01 -2.32586831e-01 -2.66249806e-01
8.87927830e-01 -2.50438571e-01 -3.86466533e-01 3.73873599e-02
-4.46263492e-01 -4.86577243e-01 -1.26048148e-01 -1.66536108e-01
1.33088744e+00 2.95554221e-01 6.09619677e-01 -1.90110818e-01
3.28611553e-01 2.29426280e-01 5.85022748e-01 9.89077330e-01
-4.05566752e-01 3.44921559e-01 2.13863090e-01 -1.34527087e-01
-7.52235889e-01 -1.14276016e+00 -9.98739749e-02 3.11820507e-01
7.85240114e-01 -1.80118665e-01 -5.98918855e-01 -2.00340077e-01
-8.70253593e-02 9.11941409e-01 -5.30035138e-01 4.79227901e-02
-5.49570680e-01 -6.57757819e-01 3.18244815e-01 3.41712266e-01
1.44580036e-01 -5.18173993e-01 -1.07012773e+00 6.05240524e-01
-2.43219525e-01 -1.24253535e+00 2.54398197e-01 4.03733552e-01
-6.28079474e-01 -1.15759730e+00 -3.75340939e-01 -2.20037267e-01
8.25868070e-01 3.14173043e-01 6.12806261e-01 -4.53193933e-02
7.39114434e-02 7.39093959e-01 -4.53728527e-01 -5.77065110e-01
-5.49773157e-01 -3.83890837e-01 3.41055304e-01 -4.79380608e-01
4.75601070e-02 -3.37009281e-01 -3.38413000e-01 4.23002243e-01
-6.75264716e-01 -4.50413197e-01 6.47100329e-01 5.38374603e-01
8.09096158e-01 6.92755759e-01 3.58151764e-01 -4.16470110e-01
5.08051634e-01 -6.31089211e-01 -9.81300712e-01 2.28953874e-03
-4.97967541e-01 3.22519928e-01 2.02101305e-01 -2.40075484e-01
-1.12981367e+00 5.24623990e-01 -1.07692806e-02 -8.71120393e-02
-3.61819893e-01 7.06197679e-01 -3.11512023e-01 -2.20474616e-01
4.92205530e-01 3.17834169e-02 -1.24405533e-01 -2.31527984e-01
2.09117427e-01 4.12193805e-01 6.91640019e-01 -8.16536546e-01
9.17910099e-01 8.21773469e-01 5.58140397e-01 -7.22175896e-01
-3.70135933e-01 -5.02963722e-01 -6.71776056e-01 -2.73651004e-01
6.52024865e-01 -6.92474127e-01 -7.98506916e-01 1.84624746e-01
-1.43733275e+00 -3.02205741e-01 -2.13947728e-01 7.13367105e-01
-1.00090826e+00 5.98551095e-01 -1.25327229e-01 -1.72849035e+00
5.09266853e-01 -1.44777584e+00 1.03827810e+00 -6.86804438e-03
-1.84811756e-01 -7.77537227e-01 3.06200031e-02 -3.98862571e-01
2.88877934e-01 5.59897840e-01 4.28330898e-01 -4.33098674e-01
-8.50518286e-01 -4.30033684e-01 4.27793488e-02 -2.63910681e-01
7.13117868e-02 -2.27701873e-01 -5.41004777e-01 3.33635807e-02
9.78261232e-02 2.99932748e-01 4.25460070e-01 4.98769134e-01
2.29385704e-01 -1.24264695e-01 -7.99275815e-01 -8.64867792e-02
1.70705426e+00 4.83702034e-01 6.26304924e-01 8.32618773e-01
2.90222168e-01 9.27558005e-01 1.36177480e+00 8.07105005e-01
7.97059178e-01 8.30991030e-01 1.01021516e+00 5.08072674e-01
3.54847431e-01 -1.45805478e-01 3.05934280e-01 2.77747452e-01
-4.13032621e-02 -4.58143830e-01 -1.16660428e+00 8.60241175e-01
-2.12791109e+00 -6.75015748e-01 -3.94218326e-01 2.52903533e+00
2.73595452e-01 1.84956193e-01 -2.55318016e-01 2.77745724e-01
6.42191947e-01 -2.22721651e-01 -3.17777097e-01 -1.63267121e-01
2.24118590e-01 -4.27457392e-01 1.34684956e+00 1.12978256e+00
-1.01221180e+00 7.15008378e-01 6.35430527e+00 5.44930577e-01
-5.09397268e-01 1.14855804e-01 -1.68029696e-01 4.53871936e-01
-2.61491179e-01 3.05612832e-01 -9.16957140e-01 2.05006659e-01
9.09407258e-01 -1.76358074e-01 2.48837173e-01 1.02335310e+00
2.44222254e-01 -8.70615602e-01 -8.33161712e-01 5.39589942e-01
-3.34189653e-01 -6.76463664e-01 -6.56557977e-01 3.13919514e-01
3.86169314e-01 1.20417222e-01 -2.39414245e-01 1.38296217e-01
7.64126420e-01 -7.97589719e-01 1.16739786e+00 5.96887231e-01
1.38532028e-01 -8.81743610e-01 1.11569929e+00 7.76500881e-01
-1.22324038e+00 -2.22342312e-01 -5.64508080e-01 -2.90686816e-01
9.44033384e-01 6.08764231e-01 -1.09956431e+00 1.08787882e+00
3.65630001e-01 2.46139020e-02 8.21356922e-02 1.38835013e+00
-4.31908995e-01 -5.25856465e-02 -8.56660426e-01 -1.99145138e-01
4.24517006e-01 2.95007695e-02 8.39719415e-01 7.69871473e-01
7.21444011e-01 1.43354043e-01 4.81783003e-01 5.20857155e-01
7.92374551e-01 -2.03664556e-01 -6.62428141e-01 5.20689070e-01
8.58118474e-01 5.36436737e-01 -8.99894953e-01 -2.15420388e-02
-2.29595989e-01 6.43299043e-01 9.73468423e-02 4.93554175e-01
-6.77897573e-01 -2.57303476e-01 4.64173764e-01 -1.58972234e-01
1.87594563e-01 -7.92838752e-01 -3.58034909e-01 -4.81998533e-01
5.14033213e-02 -2.01591685e-01 -8.03631991e-02 -4.03222024e-01
-7.81060159e-01 5.53870976e-01 5.22537231e-01 -1.37798727e+00
-6.46213770e-01 -5.38313389e-01 -1.76835969e-01 6.35877371e-01
-1.65052617e+00 -7.29843497e-01 -6.39548078e-02 1.08968690e-01
4.45087016e-01 3.89713466e-01 8.41428220e-01 6.17269687e-02
-1.76978096e-01 -8.57679993e-02 2.13081986e-01 -3.88694614e-01
4.32996690e-01 -1.03443503e+00 2.15564504e-01 9.88374293e-01
-3.95269364e-01 6.12305939e-01 1.45374441e+00 -1.02854359e+00
-1.44133270e+00 -9.42405403e-01 6.97353423e-01 -4.24065083e-01
6.00206137e-01 4.81851026e-02 -6.20707870e-01 7.28957236e-01
-3.21238577e-01 -1.11473963e-01 6.93846270e-02 -1.50191307e-01
1.50075361e-01 4.52665776e-01 -1.35999584e+00 5.90313613e-01
7.42687166e-01 -2.68752296e-02 -5.34934700e-01 2.38517538e-01
5.95343649e-01 -5.97531378e-01 -6.50254071e-01 8.38508070e-01
5.98554015e-01 -8.23671937e-01 8.60860050e-01 2.86958784e-01
-3.68147343e-01 -7.24296749e-01 -6.60495698e-01 -1.20089114e+00
2.83734575e-02 -6.14886642e-01 2.74481863e-01 9.36004639e-01
2.44749293e-01 -7.35511720e-01 5.75599194e-01 8.51514697e-01
-3.57136548e-01 -4.29140449e-01 -1.49978244e+00 -1.20808613e+00
-7.17684031e-02 -9.70993817e-01 5.99287510e-01 2.38207385e-01
-4.05510925e-02 -3.40834796e-01 -2.37127051e-01 9.56655920e-01
8.56154859e-01 -3.79477710e-01 8.86767864e-01 -1.02363443e+00
1.19272828e-01 -1.32037520e-01 -5.40719748e-01 -1.05155373e+00
2.68485367e-01 7.29203001e-02 1.16125333e+00 -1.86439955e+00
-3.26332539e-01 -1.06875634e+00 1.45058766e-01 7.87784085e-02
1.21180542e-01 -3.73341620e-01 4.24824320e-02 1.92122161e-01
-4.80231851e-01 6.37326300e-01 8.33948374e-01 1.39926121e-01
-1.74275652e-01 3.30694109e-01 -6.82464615e-02 1.15973878e+00
7.89194822e-01 -6.35077536e-01 -2.93267190e-01 -4.37797964e-01
2.55580157e-01 2.07723692e-01 3.25553834e-01 -1.29506958e+00
5.58272302e-01 -3.92228484e-01 -2.37984300e-01 -7.83982992e-01
6.87175691e-01 -1.37282741e+00 3.59255582e-01 6.17993236e-01
1.10987552e-01 -1.72204822e-02 2.59647816e-01 1.15305626e+00
8.16691015e-03 -9.16651309e-01 3.14417362e-01 -1.81654364e-01
-8.07311893e-01 -1.72626942e-01 -8.94862354e-01 -6.93976343e-01
1.15835285e+00 -2.42390245e-01 -7.71787241e-02 -7.04618216e-01
-6.98256552e-01 3.00846964e-01 6.30846322e-01 1.16457663e-01
6.84934616e-01 -1.08780801e+00 -3.10381740e-01 -2.57434756e-01
-3.47487554e-02 3.98115307e-01 1.70046434e-01 8.20462763e-01
-4.98344302e-01 4.82073575e-01 2.07085431e-01 -4.99493659e-01
-7.66530573e-01 6.16945744e-01 1.66923001e-01 -1.91252932e-01
-1.09115228e-01 4.29061532e-01 3.34497616e-02 -5.25503576e-01
3.50390524e-01 -4.89003062e-01 -1.35708258e-01 -3.31442773e-01
4.37502116e-01 6.11978650e-01 -2.49973103e-01 -9.45915520e-01
-5.75792432e-01 8.01937461e-01 5.30948520e-01 -5.10357857e-01
8.42046857e-01 -9.37139928e-01 8.38181674e-02 6.46806836e-01
5.14767885e-01 2.40204886e-01 -1.21071458e+00 -9.20120925e-02
3.95795941e-01 -3.66744846e-01 -1.00492269e-01 -3.72265548e-01
-1.54700279e-01 5.05853653e-01 4.66917336e-01 1.06484093e-01
6.33225083e-01 -2.80436091e-02 3.69809538e-01 6.81399047e-01
1.36344004e+00 -1.03248334e+00 -4.07681465e-01 8.24479878e-01
7.87905097e-01 -1.22366512e+00 2.20508441e-01 -9.62849379e-01
-4.01992321e-01 9.63078082e-01 2.67183125e-01 -1.80481628e-01
5.68378627e-01 6.23004138e-01 9.51641873e-02 1.95050254e-01
-3.61615837e-01 -5.88095725e-01 -2.34946176e-01 8.68721724e-01
-1.70999527e-01 1.49374813e-01 -3.83713484e-01 5.45311749e-01
-1.84115455e-01 -1.27055705e-01 8.56544197e-01 1.40465701e+00
-8.83177459e-01 -9.69925880e-01 -9.25326645e-01 -2.33088344e-01
-4.00289506e-01 2.55772561e-01 2.62955785e-01 1.07476425e+00
-3.02781612e-02 1.31657290e+00 -5.38754649e-02 -2.56473005e-01
5.09404004e-01 -2.17376649e-01 3.02460849e-01 -5.64760447e-01
2.57943541e-01 1.99145824e-01 3.61048490e-01 -7.19224215e-01
-4.22706217e-01 -7.15549529e-01 -1.85637045e+00 1.44035801e-01
-4.96995538e-01 1.59724489e-01 1.06435871e+00 1.12813115e+00
5.29288985e-02 5.62884152e-01 2.21255660e-01 -1.31365216e+00
-8.33176553e-01 -7.55457401e-01 -4.39213902e-01 -4.20755595e-01
4.30710912e-01 -1.24675369e+00 -5.37342072e-01 -2.74210840e-01] | [5.062306880950928, 1.5190446376800537] |
d5a628f5-a344-4b5b-8f36-5a5ff5867563 | an-attention-and-prediction-guided-visual | 2104.13018 | null | https://arxiv.org/abs/2104.13018v7 | https://arxiv.org/pdf/2104.13018v7.pdf | Attention and Prediction Guided Motion Detection for Low-Contrast Small Moving Targets | Small target motion detection within complex natural environments is an extremely challenging task for autonomous robots. Surprisingly, the visual systems of insects have evolved to be highly efficient in detecting mates and tracking prey, even though targets occupy as small as a few degrees of their visual fields. The excellent sensitivity to small target motion relies on a class of specialized neurons called small target motion detectors (STMDs). However, existing STMD-based models are heavily dependent on visual contrast and perform poorly in complex natural environments where small targets generally exhibit extremely low contrast against neighbouring backgrounds. In this paper, we develop an attention and prediction guided visual system to overcome this limitation. The developed visual system comprises three main subsystems, namely, an attention module, an STMD-based neural network, and a prediction module. The attention module searches for potential small targets in the predicted areas of the input image and enhances their contrast against complex background. The STMD-based neural network receives the contrast-enhanced image and discriminates small moving targets from background false positives. The prediction module foresees future positions of the detected targets and generates a prediction map for the attention module. The three subsystems are connected in a recurrent architecture allowing information to be processed sequentially to activate specific areas for small target detection. Extensive experiments on synthetic and real-world datasets demonstrate the effectiveness and superiority of the proposed visual system for detecting small, low-contrast moving targets against complex natural environments. | ['Cheng Hu', 'Shigang Yue', 'Jigen Peng', 'Huatian Wang', 'Jiannan Zhao', 'Hongxin Wang'] | 2021-04-27 | null | null | null | null | ['motion-detection'] | ['computer-vision'] | [ 5.57992101e-01 -4.04729247e-01 -4.78563793e-02 5.91995604e-02
2.16856599e-01 -4.23348099e-01 5.57882845e-01 -7.54624754e-02
-6.72880173e-01 3.87883544e-01 -3.63791615e-01 1.08911499e-01
3.65287215e-01 -7.83905029e-01 -4.10825819e-01 -9.24179494e-01
-2.59656936e-01 1.37097999e-01 1.20571470e+00 -8.56026635e-02
5.71358383e-01 6.13725781e-01 -1.85937071e+00 3.36681992e-01
3.70246112e-01 9.60836053e-01 9.34084177e-01 8.10541153e-01
1.56337142e-01 6.22844398e-01 -8.06157947e-01 2.26840764e-01
3.35177511e-01 -5.41396201e-01 -2.47166790e-02 -6.39180616e-02
1.23272024e-01 -4.34762090e-02 -1.38668358e-01 1.10171592e+00
5.20176828e-01 1.16725504e-01 6.83283567e-01 -1.10392547e+00
-5.65939486e-01 1.19278252e-01 -7.79255092e-01 1.13090348e+00
2.73333639e-01 5.92897892e-01 5.04599273e-01 -8.86874437e-01
3.86173517e-01 1.33447099e+00 3.93474638e-01 8.02494228e-01
-1.08883965e+00 -4.67327505e-01 2.49655232e-01 4.13085520e-01
-1.13167644e+00 -4.75922376e-01 6.37174904e-01 -4.18078899e-01
1.29870117e+00 2.02343509e-01 6.74360633e-01 1.00693476e+00
6.83680058e-01 4.78678197e-01 8.45587850e-01 -2.16199651e-01
2.65185028e-01 1.79774612e-01 -7.44153559e-02 4.11345512e-01
4.83313113e-01 5.35282731e-01 -4.04245317e-01 1.35165036e-01
8.52190673e-01 -9.45625640e-03 -3.79371405e-01 -3.68847251e-01
-1.10567117e+00 7.20111012e-01 8.35644066e-01 4.55429822e-01
-5.25039256e-01 -1.59813657e-01 -8.79826676e-03 4.68791425e-02
2.74897297e-03 6.42324686e-02 2.68403627e-02 2.64925569e-01
-5.58470726e-01 -1.73696309e-01 6.05033994e-01 4.65052277e-01
4.43616211e-01 4.23148513e-01 -1.08948417e-01 8.18346560e-01
4.12025660e-01 9.27011430e-01 8.11335683e-01 -4.95234430e-01
4.01076466e-01 7.02902079e-01 2.36784756e-01 -1.61658120e+00
-4.79678959e-01 -5.33668458e-01 -8.09056044e-01 7.42955506e-01
2.74259925e-01 9.72675011e-02 -9.19261754e-01 1.75837982e+00
4.14246023e-01 -5.14857508e-02 2.05393702e-01 1.17761040e+00
8.41464400e-01 9.47960198e-01 2.10115775e-01 -3.69868636e-01
1.13361585e+00 -8.64963293e-01 -2.93566167e-01 -1.06579328e+00
2.51194220e-02 -4.63437200e-01 7.20286608e-01 -9.03124884e-02
-9.92283762e-01 -8.34480107e-01 -1.02460706e+00 5.06275654e-01
-4.64675575e-01 3.90127599e-01 1.62396431e-01 3.72924477e-01
-1.04621410e+00 2.30785578e-01 -7.59922862e-01 -7.06463218e-01
2.41662636e-01 4.77832586e-01 -1.53521553e-01 2.77771801e-01
-9.34054315e-01 1.11767578e+00 4.33338374e-01 4.61846530e-01
-1.27970707e+00 2.38979816e-01 -9.71692145e-01 1.96302459e-01
8.76753479e-02 -4.75265712e-01 9.54802096e-01 -1.36138475e+00
-1.27383399e+00 9.08687294e-01 -2.62473553e-01 -5.15625834e-01
2.18733311e-01 2.08949670e-01 -2.97539830e-01 1.11218795e-01
3.16329807e-01 7.80211687e-01 1.17132735e+00 -1.01141167e+00
-1.11198401e+00 -3.87931377e-01 -2.70214558e-01 6.77094281e-01
-1.60572141e-01 9.41199884e-02 -2.86674559e-01 -6.33442521e-01
3.04662101e-02 -6.95645332e-01 -4.19622391e-01 6.69980347e-02
2.07551923e-02 -3.52882668e-02 1.06006932e+00 -4.11679372e-02
9.93787527e-01 -2.19374108e+00 1.34350613e-01 -1.44671366e-01
-1.08105533e-01 5.46625733e-01 -2.54252315e-01 -3.84643935e-02
1.92114100e-01 -5.22555411e-01 -1.95272371e-01 2.56576955e-01
-4.70306545e-01 -6.32835329e-02 -2.24066630e-01 4.47416037e-01
4.12355125e-01 6.04673445e-01 -8.83707047e-01 -4.29469556e-01
3.82393688e-01 3.57849270e-01 -2.11769804e-01 2.87848413e-01
9.06409975e-03 3.33222389e-01 -3.52044970e-01 8.02325904e-01
5.63080370e-01 -2.10194811e-01 -4.67439415e-03 6.37629330e-02
-5.89645028e-01 -5.04607968e-02 -1.14335597e+00 7.79755712e-01
8.40953514e-02 8.71314704e-01 3.04186702e-01 -1.02252662e+00
1.20674610e+00 -1.86305329e-01 -6.99296743e-02 -9.12802696e-01
3.26306283e-01 1.35930613e-01 3.86191338e-01 -4.48921889e-01
3.51300031e-01 2.26236554e-03 1.35321736e-01 1.84510022e-01
-2.56838799e-01 2.49344975e-01 2.44329959e-01 -1.33938104e-01
9.56742525e-01 -1.39702009e-02 5.20423055e-01 -5.33489920e-02
7.38784015e-01 2.09855467e-01 8.08778226e-01 8.22356045e-01
-5.65589249e-01 2.41978601e-01 -8.64869505e-02 -6.33396804e-01
-6.86569631e-01 -1.02870548e+00 5.74404486e-02 1.24494016e+00
8.87443542e-01 1.76272541e-01 -3.55660975e-01 -3.15943271e-01
-1.34828866e-01 3.46616447e-01 -7.66490281e-01 -4.31422681e-01
-7.68911898e-01 -9.66321170e-01 6.98589534e-02 6.17181420e-01
6.68652952e-01 -1.78555882e+00 -1.79015148e+00 2.36713171e-01
2.03062907e-01 -6.42707825e-01 -3.01028281e-01 4.56780344e-01
-6.13744736e-01 -1.01576102e+00 -8.96728814e-01 -1.26698244e+00
6.89249575e-01 8.74091446e-01 8.31373513e-01 -1.39710993e-01
-4.95131254e-01 1.05271406e-01 -1.29797280e-01 -5.15441895e-01
-2.20703423e-01 -4.79115874e-01 2.22912997e-01 6.73248693e-02
4.12158996e-01 -3.20526391e-01 -6.23119950e-01 6.97395861e-01
-5.29657304e-01 -1.66638233e-02 9.90492404e-01 8.45495522e-01
5.19445837e-01 -3.61927189e-02 2.76410341e-01 -1.60987526e-01
3.72312874e-01 -5.23103297e-01 -9.91348028e-01 1.60299331e-01
1.71110183e-01 -6.91170320e-02 7.64523566e-01 -1.01133454e+00
-9.79799926e-01 2.54461616e-01 3.47496837e-01 -3.12460929e-01
-2.38263741e-01 4.98095378e-02 -6.57268837e-02 -3.48898917e-01
7.70229936e-01 6.86926246e-01 -3.07960123e-01 -2.72074547e-02
-2.50678480e-01 4.68260765e-01 6.52621150e-01 2.64540881e-01
7.52808034e-01 4.43464220e-01 1.15045466e-01 -1.03155947e+00
-2.28458777e-01 -3.65179062e-01 -4.45329815e-01 -5.88652611e-01
8.25188994e-01 -6.59043312e-01 -7.57351160e-01 6.12946928e-01
-9.66516972e-01 -1.64784074e-01 -1.20048821e-01 5.26674449e-01
-3.40500772e-01 2.09363997e-01 -4.55085427e-01 -1.01688397e+00
-4.69468981e-01 -1.01468432e+00 8.29114318e-01 1.05137098e+00
7.17160776e-02 -6.08123243e-01 1.33740470e-01 -2.69567728e-01
6.47833347e-01 5.00214398e-02 4.32278484e-01 -4.67211455e-01
-6.25252068e-01 -1.78643182e-01 -2.58772463e-01 -2.49314487e-01
2.05368772e-01 3.12108751e-02 -8.40125024e-01 -3.99225205e-01
1.96411639e-01 -2.69306153e-01 1.12956858e+00 8.07201266e-01
2.89796650e-01 3.38959731e-02 -8.41378868e-01 3.97170424e-01
1.24068618e+00 8.80684733e-01 1.49993569e-01 4.79359597e-01
2.19388932e-01 5.87858379e-01 8.89608979e-01 4.39444482e-01
-2.13441163e-01 6.44677520e-01 7.82315135e-01 2.30995030e-03
9.32358950e-03 1.44748896e-01 5.91782153e-01 1.55794039e-01
-2.33334359e-02 -3.64769787e-01 -7.39017427e-01 6.10203981e-01
-1.68215084e+00 -1.30021751e+00 -3.64063643e-02 2.43108058e+00
2.30555087e-01 5.23469210e-01 1.32099926e-01 -1.24420747e-01
1.06615913e+00 8.00095648e-02 -9.06288385e-01 -2.05049321e-01
-5.79130709e-01 -2.67742157e-01 3.35993528e-01 2.89684623e-01
-1.30572224e+00 1.00636148e+00 6.22597790e+00 3.95595670e-01
-1.40633476e+00 -3.02600473e-01 3.91323715e-01 5.55048175e-02
4.70805526e-01 -2.85998166e-01 -1.17149508e+00 5.72946668e-01
5.79518795e-01 -5.73769286e-02 1.10971719e-01 1.05673873e+00
1.37434348e-01 -5.81281543e-01 -6.65313721e-01 8.95126581e-01
1.96490765e-01 -9.27280068e-01 9.53671783e-02 -1.80674806e-01
2.55686969e-01 2.31127683e-02 4.41956192e-01 1.15216404e-01
2.12687001e-01 -8.80893528e-01 6.25914097e-01 3.98746550e-01
2.55646169e-01 -5.24740160e-01 6.66315734e-01 6.09090328e-01
-1.45040488e+00 -5.23827374e-01 -9.77526069e-01 -2.94535875e-01
-6.56408891e-02 -6.56131804e-02 -7.42592931e-01 -2.24841848e-01
1.01765847e+00 7.22853661e-01 -7.18080878e-01 1.41283143e+00
-5.60755283e-03 1.53978676e-01 -5.25998950e-01 -5.15792727e-01
3.56012493e-01 -3.76922488e-02 9.93191361e-01 1.13319421e+00
3.14498752e-01 1.89949706e-01 -5.99691905e-02 8.27031195e-01
4.94605362e-01 -1.02668934e-01 -8.97457361e-01 1.04563862e-01
4.06924844e-01 1.33252084e+00 -1.08738875e+00 -3.18318188e-01
-2.82039255e-01 1.07791924e+00 3.43890727e-01 2.05931440e-01
-7.78336465e-01 -4.42547530e-01 3.32525671e-01 -1.97953150e-01
7.44791090e-01 8.05866253e-03 9.23528299e-02 -1.02845895e+00
-1.91720188e-01 -6.69559777e-01 6.22011900e-01 -8.18128467e-01
-9.07055199e-01 9.59717870e-01 -3.37191850e-01 -1.41821849e+00
-1.79119617e-01 -7.41109192e-01 -9.65888441e-01 9.27638948e-01
-1.07856476e+00 -7.98916340e-01 -5.12955308e-01 6.45091176e-01
8.86219919e-01 -2.78821260e-01 5.15601158e-01 -2.48781770e-01
-6.85662091e-01 1.97381154e-01 3.32366377e-02 -7.14597702e-02
4.50312644e-01 -8.18109632e-01 2.58976936e-01 1.15997398e+00
-5.09427339e-02 2.07300395e-01 7.40015447e-01 -9.00642812e-01
-1.18546760e+00 -8.86978924e-01 4.88653898e-01 -1.74694136e-01
3.17630559e-01 -2.97702730e-01 -9.25359905e-01 2.19244644e-01
1.26951382e-01 7.24030882e-02 2.88299620e-01 -6.99211538e-01
1.11416683e-01 -1.28004402e-01 -1.05942225e+00 5.86358249e-01
8.11776042e-01 2.20434532e-01 -7.78868139e-01 -3.97949107e-02
9.93918329e-02 -1.13929644e-01 9.00873989e-02 4.85446125e-01
6.23268425e-01 -1.12913764e+00 1.01870382e+00 -4.16602641e-01
-1.23561375e-01 -7.49058902e-01 -9.04905424e-02 -1.10962582e+00
-8.46653163e-01 -8.97144228e-02 1.78840861e-01 8.52074385e-01
2.65533835e-01 -4.37537402e-01 7.99652874e-01 -4.69235480e-02
1.59646422e-02 -1.85903326e-01 -9.36313331e-01 -5.60005486e-01
-5.60840130e-01 1.67815655e-01 -2.97023863e-01 4.34121460e-01
-4.16242331e-02 5.18698812e-01 -3.91067386e-01 4.76347953e-01
7.09478498e-01 5.05173445e-01 5.02708733e-01 -1.18561590e+00
-2.43572339e-01 -5.19863069e-01 -7.67449975e-01 -1.00842762e+00
-2.69720435e-01 -3.61089230e-01 4.33758765e-01 -1.44557297e+00
2.49442354e-01 9.14532021e-02 -3.84018600e-01 4.42925304e-01
-3.11625063e-01 5.41938305e-01 1.69015184e-01 4.06722575e-01
-7.44437933e-01 4.65356052e-01 9.08174753e-01 -2.94913381e-01
-5.07842839e-01 3.17087293e-01 -1.98459342e-01 7.89037049e-01
6.68612957e-01 -3.84781063e-01 -1.88364461e-01 -3.73903900e-01
-3.58529031e-01 -4.41535674e-02 3.36198658e-01 -1.52785373e+00
5.14352024e-01 -3.52459818e-01 1.09296131e+00 -7.34170377e-01
5.00788093e-01 -7.68011630e-01 2.75259018e-02 1.16832840e+00
-1.06148548e-01 1.29420131e-01 4.99475896e-01 5.97526670e-01
-9.36651900e-02 -1.05663911e-01 1.23001802e+00 -3.65767479e-01
-1.25074506e+00 -1.46201864e-01 -9.50043797e-01 -2.11488515e-01
1.43386626e+00 -6.55390441e-01 -4.12083149e-01 -2.18961939e-01
-6.50697529e-01 2.08134606e-01 3.79405677e-01 4.80985612e-01
9.77779031e-01 -1.09550428e+00 -4.17389512e-01 6.54179275e-01
-1.23729035e-01 -4.13336992e-01 2.36995101e-01 7.05608845e-01
-5.16002476e-01 6.50981069e-01 -8.94913673e-01 -8.89660776e-01
-1.57335424e+00 9.34601128e-01 5.35706520e-01 1.66884691e-01
-2.26096973e-01 1.16686261e+00 7.74993360e-01 2.78100252e-01
2.85569817e-01 -2.21768484e-01 -8.12322915e-01 -2.05761299e-01
9.17647123e-01 1.67954758e-01 -5.51815391e-01 -1.11939633e+00
-6.81374192e-01 9.19922411e-01 7.79299140e-02 7.91364685e-02
1.24693942e+00 -1.87236562e-01 1.76168010e-01 4.10831600e-01
3.80603105e-01 -2.52619654e-01 -1.49961746e+00 2.23978469e-03
-9.76157561e-02 -4.99686331e-01 -2.81800002e-01 -5.73209763e-01
-8.73178482e-01 1.18660569e+00 8.85321319e-01 3.55757684e-01
1.49546576e+00 -1.55486204e-02 1.24160662e-01 2.94971883e-01
3.47149134e-01 -7.36436307e-01 3.23203176e-01 5.39317489e-01
7.95386732e-01 -1.17584014e+00 -3.30430478e-01 -6.38738871e-02
-7.81439960e-01 1.09862387e+00 1.18677139e+00 -2.66081184e-01
1.95141807e-01 3.65273207e-01 1.04313977e-01 8.96415189e-02
-1.01682866e+00 -4.29355145e-01 3.03880274e-01 1.03630662e+00
-1.82821658e-02 -3.49251628e-01 1.02388458e-02 3.29952240e-01
1.73666775e-01 -4.24738377e-01 2.71389365e-01 1.06258953e+00
-1.14840722e+00 -3.75288427e-01 -8.02340329e-01 5.10117784e-02
-2.95019031e-01 3.33514735e-02 -7.79498100e-01 6.51221097e-01
1.32193580e-01 9.69421685e-01 3.90624046e-01 -4.82191771e-01
2.31461674e-01 -5.07732570e-01 2.87252367e-01 -4.41849887e-01
-5.52612782e-01 2.61604458e-01 -3.76732290e-01 -5.55244744e-01
-5.31619430e-01 -5.03195107e-01 -1.11911714e+00 2.12328151e-01
-2.60415196e-01 2.03033924e-01 3.51860583e-01 5.00964642e-01
1.96278349e-01 1.99867994e-01 5.34920573e-01 -1.19745982e+00
-3.03435475e-01 -9.39725399e-01 -3.50721359e-01 1.14094220e-01
3.38185281e-01 -8.64932656e-01 -4.66686130e-01 -1.24106556e-01] | [8.405806541442871, -0.8109832406044006] |
0f8f5182-4a42-486c-b52d-e92635cfed7a | bytecover3-accurate-cover-song-identification | 2303.11692 | null | https://arxiv.org/abs/2303.11692v1 | https://arxiv.org/pdf/2303.11692v1.pdf | ByteCover3: Accurate Cover Song Identification on Short Queries | Deep learning based methods have become a paradigm for cover song identification (CSI) in recent years, where the ByteCover systems have achieved state-of-the-art results on all the mainstream datasets of CSI. However, with the burgeon of short videos, many real-world applications require matching short music excerpts to full-length music tracks in the database, which is still under-explored and waiting for an industrial-level solution. In this paper, we upgrade the previous ByteCover systems to ByteCover3 that utilizes local features to further improve the identification performance of short music queries. ByteCover3 is designed with a local alignment loss (LAL) module and a two-stage feature retrieval pipeline, allowing the system to perform CSI in a more precise and efficient way. We evaluated ByteCover3 on multiple datasets with different benchmark settings, where ByteCover3 beat all the compared methods including its previous versions. | ['Zejun Ma', 'Bilei Zhu', 'Huidong Liang', 'Xia Liang', 'Zijie Wang', 'Xingjian Du'] | 2023-03-21 | null | null | null | null | ['cover-song-identification'] | ['music'] | [ 1.23269431e-01 -8.32197070e-01 -2.21663088e-01 -1.03610501e-01
-1.32635379e+00 -7.62281179e-01 3.47487837e-01 -4.32569087e-02
-4.23339754e-01 3.40089738e-01 1.44533977e-01 1.92953125e-01
-3.13771546e-01 -6.70562506e-01 -7.81184256e-01 -5.28525591e-01
-1.88518599e-01 6.13687456e-01 1.52968541e-01 -3.89768444e-02
1.88338220e-01 4.71987501e-02 -1.86743033e+00 5.56907117e-01
3.34350973e-01 1.30786204e+00 -4.34849737e-03 6.70361280e-01
9.30072740e-02 4.46821988e-01 -6.87357605e-01 -5.94128311e-01
5.83700895e-01 -3.48325402e-01 -5.49358368e-01 -6.03354871e-01
7.94337451e-01 -2.71289915e-01 -4.51124638e-01 8.25470626e-01
9.91039336e-01 -1.38019130e-01 2.23411024e-01 -1.10769737e+00
-5.01533039e-02 1.08166754e+00 -5.56227326e-01 8.28091353e-02
4.69882905e-01 1.20177649e-01 1.43701267e+00 -6.89199209e-01
3.13528508e-01 1.06228375e+00 1.11984825e+00 -4.22928762e-03
-1.08499897e+00 -1.12646687e+00 -2.02346474e-01 5.69471359e-01
-1.68315649e+00 -4.08536822e-01 4.31764245e-01 -2.19802618e-01
9.69877183e-01 6.63085699e-01 7.55421102e-01 9.52841222e-01
-2.40956396e-01 1.02156270e+00 7.40114033e-01 -2.40658522e-01
-1.19872443e-01 -2.08305418e-01 3.98265980e-02 2.87377805e-01
2.25548074e-01 7.71204755e-02 -8.62160444e-01 -1.54769033e-01
4.85954314e-01 -7.27308393e-02 -2.88737118e-01 -3.78574692e-02
-1.46673393e+00 5.75796962e-01 2.51310557e-01 3.48485887e-01
-2.90244430e-01 4.02223319e-01 7.22656429e-01 5.06937921e-01
2.71528572e-01 6.39359295e-01 -2.62524039e-01 -6.72948122e-01
-1.28679609e+00 5.69540977e-01 1.01761663e+00 7.70054638e-01
3.92939508e-01 -1.17844708e-01 -5.62039196e-01 9.10298228e-01
-2.54849643e-01 4.70305830e-01 5.36328673e-01 -7.41400480e-01
5.69320202e-01 4.67762589e-01 -2.00938091e-01 -1.03974199e+00
-7.10476935e-02 -1.22349918e+00 -6.50376856e-01 -2.53028750e-01
2.55083531e-01 2.50287950e-01 -3.05975348e-01 1.48025644e+00
2.61982884e-02 5.54285109e-01 -2.21703187e-01 1.15380883e+00
6.87004209e-01 4.87360328e-01 -5.36900222e-01 1.03423916e-01
9.62402046e-01 -1.05718374e+00 -3.51531893e-01 7.50750154e-02
4.23071802e-01 -1.05329525e+00 1.18601537e+00 8.55143487e-01
-7.83740699e-01 -7.16552556e-01 -1.29414976e+00 1.56961292e-01
-7.89763704e-02 1.53410733e-01 6.41957343e-01 5.27436495e-01
-7.25879133e-01 8.83925498e-01 -5.37072182e-01 -7.03584328e-02
4.61655438e-01 5.39651811e-01 -3.48184764e-01 -3.32824364e-02
-1.18387985e+00 1.38260320e-01 4.47017491e-01 7.80249834e-02
-8.87812257e-01 -8.22435617e-01 -3.95436913e-01 2.34091654e-01
5.30630589e-01 -4.45671052e-01 1.29230142e+00 -7.98688352e-01
-1.26773381e+00 8.63644779e-01 3.80142331e-01 -7.92703986e-01
5.26344359e-01 -7.10923493e-01 -4.45250779e-01 -2.42388144e-01
-1.10512190e-01 3.12376291e-01 9.52363908e-01 -6.77799284e-01
-6.82334721e-01 -1.66407704e-01 4.05039871e-03 1.61846399e-01
-6.04167521e-01 -4.09706421e-02 -9.17228520e-01 -8.11946452e-01
-2.01502919e-01 -1.17033482e+00 3.24355423e-01 -3.45919192e-01
-6.13623083e-01 -7.42067322e-02 6.55769050e-01 -4.80536222e-01
1.71562874e+00 -2.49339700e+00 2.76463389e-01 1.09484032e-01
6.07532449e-02 6.63034558e-01 -2.32695967e-01 7.58665442e-01
1.50827557e-01 -2.11289436e-01 -6.07714541e-02 -5.56963325e-01
4.17095691e-01 -1.74492896e-01 -4.32994574e-01 4.58483368e-01
-3.88232559e-01 8.08452487e-01 -5.58527827e-01 -2.69592851e-01
6.49619615e-03 2.51807719e-01 -6.51343226e-01 3.59283894e-01
-2.20772669e-01 1.18877292e-01 -8.08345377e-02 7.39257634e-01
6.48770332e-01 -5.44983037e-02 -2.54394673e-02 -2.83157021e-01
-1.02387212e-01 4.88707930e-01 -1.31551874e+00 2.02831745e+00
-3.60733151e-01 6.52646065e-01 -2.31734395e-01 -7.78536856e-01
7.37901151e-01 7.83092901e-02 7.21589744e-01 -5.38638055e-01
9.77493450e-03 4.85984683e-01 7.17719272e-02 -1.32754996e-01
3.74118298e-01 3.89079809e-01 -2.44405016e-01 6.07868075e-01
-1.68128818e-01 2.57289290e-01 9.18438360e-02 -2.70710625e-02
1.12350988e+00 1.02873631e-01 6.22774027e-02 1.49377167e-01
5.53355098e-01 -2.09267046e-02 5.29002488e-01 1.03850842e+00
2.23894298e-01 7.50870168e-01 3.25665660e-02 -3.55991304e-01
-7.48164356e-01 -8.73979390e-01 -9.82965976e-02 1.12202871e+00
9.25294608e-02 -8.95634711e-01 -8.35438788e-01 -5.90818584e-01
3.46084028e-01 2.72193581e-01 -1.05759308e-01 -9.90296230e-02
-6.38313413e-01 -5.44265449e-01 1.06407332e+00 3.44229758e-01
7.03675389e-01 -1.23229456e+00 -5.82335114e-01 4.32833672e-01
-2.42108345e-01 -9.41771984e-01 -7.46878862e-01 2.37530507e-02
-3.93693805e-01 -9.66577232e-01 -6.42516613e-01 -5.11320531e-01
-3.47551316e-01 3.65441382e-01 1.22184420e+00 4.53216024e-02
-5.47686040e-01 7.25246593e-02 -6.11219466e-01 -5.25100529e-01
-1.43441513e-01 6.71909869e-01 1.71588272e-01 -2.46733874e-02
3.90112370e-01 -8.01599205e-01 -8.08317244e-01 3.38971585e-01
-8.62267554e-01 -5.67005835e-02 7.43950725e-01 8.78561556e-01
8.96695733e-01 -1.40493393e-01 6.36643589e-01 -5.74575245e-01
3.71262789e-01 -2.80544192e-01 -6.53506517e-01 5.77742532e-02
-7.05330968e-01 -1.61128119e-01 4.62783903e-01 -4.24267620e-01
-2.49274552e-01 9.20747779e-03 -4.61236477e-01 -6.81540072e-01
1.56489819e-01 7.76973963e-01 -1.35439515e-01 -1.08189382e-01
4.65756416e-01 4.65983987e-01 -1.98784694e-01 -1.03416061e+00
-6.50074659e-03 1.07823789e+00 8.59524786e-01 -4.00645405e-01
8.42359066e-01 1.52844548e-01 -1.35314077e-01 -4.34834719e-01
-1.19537985e+00 -8.02595496e-01 -4.48324263e-01 -4.66126800e-02
2.79646695e-01 -1.02879846e+00 -9.01787758e-01 7.27625549e-01
-6.88797712e-01 -1.37298644e-01 -2.76764542e-01 4.74884361e-01
-5.47298193e-01 3.45437527e-01 -4.77802962e-01 -4.99968916e-01
-8.77647877e-01 -1.13687885e+00 1.28463197e+00 -1.62372872e-01
-7.87159055e-02 -2.76677430e-01 4.12411004e-01 3.93765569e-01
5.00414789e-01 1.37270093e-01 5.76975465e-01 -7.75092602e-01
-8.27876151e-01 -5.69930971e-01 -6.47186339e-02 5.12309015e-01
-6.21929727e-02 -4.63314027e-01 -1.40271962e+00 -5.66370130e-01
-3.52912754e-01 -4.23230171e-01 1.18930781e+00 1.14152551e-01
1.56447864e+00 -2.18885735e-01 -1.48402020e-01 1.33120513e+00
1.28930962e+00 -7.38779679e-02 6.69892311e-01 6.62033260e-01
6.79648101e-01 -1.05973005e-01 9.16210830e-01 5.25844991e-01
9.87905338e-02 1.27083671e+00 6.62752271e-01 7.55921602e-02
-2.54638195e-01 -2.93669075e-01 5.65034032e-01 1.08750689e+00
2.00121731e-01 -2.74597019e-01 -6.33800924e-01 3.92204314e-01
-1.93062615e+00 -1.04479992e+00 1.42607629e-01 2.61244559e+00
8.94380391e-01 1.41858399e-01 2.50571787e-01 6.33119702e-01
4.40585136e-01 2.98004091e-01 -8.27085614e-01 -7.15106949e-02
-2.58313686e-01 4.47243661e-01 4.66542214e-01 -2.87510097e-01
-1.52926719e+00 8.89184296e-01 5.98036909e+00 1.32315493e+00
-1.33376944e+00 3.19122374e-02 2.77186096e-01 -3.73062611e-01
4.29206565e-02 -1.39654249e-01 -9.80765760e-01 4.44333732e-01
7.83121526e-01 -9.96942744e-02 7.81949103e-01 8.00816655e-01
-3.32066447e-01 1.85943604e-01 -1.18770945e+00 1.58432949e+00
8.16282332e-02 -1.57613420e+00 -5.42970486e-02 1.44721910e-01
5.39155781e-01 3.58277619e-01 1.97321966e-01 5.80192626e-01
-5.01975179e-01 -1.00956690e+00 8.11923742e-01 4.89135653e-01
1.27281094e+00 -8.98682058e-01 8.90574753e-01 3.04324418e-01
-1.39547253e+00 -2.21223027e-01 -2.07580552e-01 -2.90765855e-02
-9.34996456e-02 6.60937250e-01 -6.73040926e-01 7.91437626e-01
8.46980691e-01 1.06346548e+00 -6.02733850e-01 1.51735914e+00
3.53358626e-01 9.18029964e-01 -3.49338144e-01 1.29820615e-01
8.91720355e-02 5.11058886e-03 8.26285601e-01 1.29189920e+00
5.17734230e-01 -6.61408067e-01 3.74504864e-01 5.01829803e-01
-3.68272901e-01 2.70822734e-01 -2.25410193e-01 -4.03629065e-01
5.34842849e-01 1.21191502e+00 -1.84857085e-01 -6.34011477e-02
-1.93149894e-01 1.08567572e+00 1.22018583e-01 -2.27910906e-01
-9.61291969e-01 -4.82951969e-01 1.02884543e+00 2.03990892e-01
4.70616221e-01 -1.25927979e-03 1.43393636e-01 -1.15027463e+00
1.00412816e-01 -1.50195611e+00 5.22230446e-01 -2.77495891e-01
-1.22988498e+00 5.52951276e-01 -2.18536258e-01 -1.63076282e+00
-2.90838510e-01 -2.70037860e-01 -4.59649324e-01 6.68132901e-01
-1.27726626e+00 -1.16323602e+00 -3.39431435e-01 5.27864397e-01
5.20849586e-01 -5.29213071e-01 1.07993984e+00 8.88445854e-01
-5.32671213e-01 1.18966043e+00 4.89621133e-01 2.02474013e-01
1.16689003e+00 -1.04160178e+00 6.94206357e-01 4.20980126e-01
9.26166534e-01 4.38263685e-01 5.20483792e-01 -3.30766678e-01
-1.99998772e+00 -1.06455183e+00 4.62564766e-01 -1.10427774e-01
5.13145506e-01 -5.79924464e-01 -6.32326484e-01 3.92782688e-01
-2.55085435e-03 -1.72470465e-01 7.67165840e-01 2.39008293e-01
-5.51197410e-01 -7.64511526e-01 -5.83065152e-01 1.30728796e-01
1.31644416e+00 -6.67075455e-01 -2.69251913e-02 1.85191214e-01
7.38919497e-01 -6.31637394e-01 -9.01714146e-01 5.14848650e-01
1.17867172e+00 -1.07556248e+00 1.13570201e+00 -1.42424554e-01
-7.77933300e-02 -4.37206656e-01 -3.43978137e-01 -9.80693460e-01
-2.54709721e-01 -8.03987563e-01 -2.94738263e-01 1.40434849e+00
2.40246609e-01 -3.09668422e-01 6.57532156e-01 -5.49326360e-01
-2.61837602e-01 -7.92519152e-01 -9.37313139e-01 -1.04521167e+00
-4.53109175e-01 -7.56284177e-01 1.05051768e+00 6.30295277e-01
-2.69715816e-01 7.64699280e-02 -7.80175626e-01 -2.61898011e-01
5.40665329e-01 8.26347291e-01 1.28282869e+00 -1.34604573e+00
-9.34978604e-01 -4.88557130e-01 -7.96281636e-01 -1.03419769e+00
7.43735954e-02 -1.14253068e+00 -1.03050560e-01 -1.08461893e+00
4.83330131e-01 -5.55650294e-01 -7.37227499e-01 4.14664775e-01
-1.57682225e-02 8.07401121e-01 3.53684127e-01 5.39152801e-01
-8.80226254e-01 4.19643551e-01 8.23992133e-01 -3.84435594e-01
-1.50532246e-01 2.95444578e-01 -6.79523230e-01 3.96513432e-01
6.37147784e-01 -5.37410975e-01 -2.51598179e-01 -3.89772177e-01
4.37445402e-01 -1.87062398e-01 2.02745914e-01 -1.58628035e+00
2.32587948e-01 3.96229863e-01 2.50697166e-01 -8.51774514e-01
5.44791460e-01 -6.44336760e-01 4.91314083e-01 3.57497185e-01
-3.38425845e-01 -5.82697950e-02 2.38631144e-01 4.92709577e-01
-5.11746407e-01 1.54183373e-01 4.58405554e-01 3.56532991e-01
-5.70043504e-01 5.33607662e-01 2.09913298e-01 8.76298249e-02
5.49588799e-01 -7.09932074e-02 -1.00724921e-01 -4.96531129e-01
-3.47147405e-01 -3.95609178e-02 3.06515902e-01 5.47903121e-01
4.07778651e-01 -1.27774358e+00 -8.64944875e-01 1.78404838e-01
3.20817351e-01 -2.73799419e-01 1.22532889e-01 9.35380816e-01
-4.47934657e-01 5.88406861e-01 -8.68853182e-03 -7.92822182e-01
-1.46874475e+00 3.37679058e-01 2.32543781e-01 -4.04783279e-01
-8.38706911e-01 8.65464807e-01 -1.27426162e-01 -3.20746630e-01
7.55252779e-01 -2.46298332e-02 9.83897410e-03 -5.16404808e-02
6.46427810e-01 3.24693143e-01 2.57468849e-01 -5.69708467e-01
-3.12794566e-01 5.81398964e-01 -9.19985846e-02 1.06056988e-01
1.29076838e+00 1.61619529e-01 5.22046275e-02 6.61406100e-01
1.15817642e+00 1.56330436e-01 -9.40339923e-01 -3.62620473e-01
9.61310416e-02 -7.13666201e-01 1.65807158e-01 -8.68094563e-01
-1.10821939e+00 8.22154462e-01 8.87029469e-01 -2.40308642e-02
1.33863091e+00 -1.59685493e-01 1.35515308e+00 5.00524104e-01
6.32102787e-01 -8.03173542e-01 -9.65309739e-02 4.92800593e-01
8.72189522e-01 -1.04041147e+00 8.19445252e-02 -6.40133023e-02
-3.30459028e-01 8.95303309e-01 2.56246090e-01 -5.92537522e-02
5.46076477e-01 4.20618683e-01 -9.24977511e-02 5.27409092e-02
-5.95085680e-01 -2.70070404e-01 6.34275079e-01 1.76515505e-01
4.53229040e-01 9.84326452e-02 -2.30150983e-01 8.96663547e-01
-4.99802709e-01 -8.84547085e-02 -1.95668876e-01 4.84339178e-01
-3.23637456e-01 -1.52767956e+00 -3.35150689e-01 5.54750025e-01
-7.16799080e-01 -2.95733362e-01 -5.47273576e-01 6.63217783e-01
2.58542418e-01 6.80651844e-01 -8.19133371e-02 -9.04024005e-01
3.12026978e-01 -1.39050528e-01 3.85910034e-01 -5.05224884e-01
-1.14267743e+00 3.20085049e-01 9.41944420e-02 -8.29417825e-01
-2.21456155e-01 -7.05089867e-01 -9.03211653e-01 -3.58415335e-01
-4.84475583e-01 1.56866387e-01 7.72057652e-01 7.01644838e-01
6.12077713e-01 3.92706573e-01 6.79218411e-01 -8.11378002e-01
-8.27875972e-01 -1.03618431e+00 -6.16460621e-01 3.13658208e-01
2.13026479e-01 -5.29145241e-01 -1.10345237e-01 -2.75917828e-01] | [15.682795524597168, 5.226885795593262] |
f63fef42-20e6-4077-a5e4-60f649080460 | merging-datasets-for-aggressive-text | null | null | https://aclanthology.org/W18-4416 | https://aclanthology.org/W18-4416.pdf | Merging Datasets for Aggressive Text Identification | This paper presents the approach of the team {``}groutar{''} to the shared task on Aggression Identification, considering the test sets in English, both from Facebook and general Social Media. This experiment aims to test the effect of merging new datasets in the performance of classification models. We followed a standard machine learning approach with training, validation, and testing phases, and considered features such as part-of-speech, frequencies of insults, punctuation, sentiment, and capitalization. In terms of algorithms, we experimented with Boosted Logistic Regression, Multi-Layer Perceptron, Parallel Random Forest and eXtreme Gradient Boosting. One question appearing was how to merge datasets using different classification systems (e.g. aggression vs. toxicity). Other issue concerns the possibility to generalize models and apply them to data from different social networks. Regarding these, we merged two datasets, and the results showed that training with similar data is an advantage in the classification of social networks data. However, adding data from different platforms, allowed slightly better results in both Facebook and Social Media, indicating that more generalized models can be an advantage. | ["S{\\'e}rgio Nunes", "Jos{\\'e} Ferreira", 'Luiz Pires', 'Paula Fortuna', 'Guilherme Routar'] | 2018-08-01 | null | null | null | coling-2018-8 | ['aggression-identification'] | ['natural-language-processing'] | [-3.04170609e-01 2.14436620e-01 -1.40565950e-02 -3.93652231e-01
-1.12134591e-01 -2.28919953e-01 6.13468945e-01 8.29894781e-01
-1.02670717e+00 9.02059257e-01 1.71115056e-01 -1.40961990e-01
-6.38198793e-01 -8.71692359e-01 -2.71964431e-01 -3.65343332e-01
-2.19268844e-01 6.30184472e-01 3.16207349e-01 -6.78713679e-01
4.74665374e-01 5.96585095e-01 -1.73514497e+00 6.03955984e-01
5.27813971e-01 3.97784233e-01 -1.42694309e-01 4.27951187e-01
-2.75862247e-01 8.93094420e-01 -8.22423160e-01 -1.01647377e+00
-2.33322233e-02 -1.16988190e-01 -1.11938798e+00 -2.58137614e-01
7.79997706e-02 -9.92398337e-02 1.94784269e-01 7.79596150e-01
5.88019013e-01 -6.22285232e-02 6.46512151e-01 -1.28136337e+00
-7.74156582e-03 8.83685231e-01 -2.26235747e-01 2.37234160e-01
8.37178469e-01 4.63566594e-02 6.73757255e-01 -2.29239091e-01
7.59437323e-01 1.32543755e+00 8.20429921e-01 5.48403144e-01
-1.24573970e+00 -7.35215187e-01 3.61629315e-02 4.67777282e-01
-1.01910162e+00 -1.40421718e-01 4.42493409e-01 -8.80932808e-01
7.91971684e-01 4.35651958e-01 6.75454021e-01 1.58486235e+00
-2.80604720e-01 1.56189695e-01 1.69974792e+00 -3.74805629e-01
2.85410583e-01 1.08791578e+00 5.21755755e-01 2.35687569e-01
3.50130171e-01 -5.22143096e-02 -4.77840602e-01 -4.52190280e-01
-1.28976703e-01 -3.33901495e-01 -1.38362190e-02 1.85711086e-01
-3.45186919e-01 9.79286671e-01 8.79677534e-02 9.19915795e-01
-2.34688357e-01 -4.96976882e-01 8.55243206e-01 4.40099984e-01
6.26050830e-01 4.72048819e-01 -5.36609173e-01 -4.72194999e-01
-6.51593268e-01 2.64577001e-01 1.08013475e+00 -4.79808971e-02
6.66512132e-01 -3.25905204e-01 1.52643159e-01 1.03153801e+00
2.16662526e-01 1.09850932e-02 9.14348662e-01 -4.69711185e-01
5.52956581e-01 7.93168664e-01 -3.81460309e-01 -9.19858813e-01
-9.12679017e-01 -4.62708831e-01 -2.49820024e-01 4.59072828e-01
7.09447086e-01 -4.56416219e-01 -1.46948338e-01 1.69136357e+00
3.14371973e-01 -1.29771501e-01 -7.00965598e-02 6.19441032e-01
7.63088226e-01 -3.52117326e-03 2.95661718e-01 -2.61303335e-01
9.13103342e-01 -5.21410823e-01 -5.07534266e-01 -5.82725815e-02
1.15756214e+00 -4.92147386e-01 7.69896388e-01 8.43028367e-01
-9.56984222e-01 -3.76142651e-01 -8.21321130e-01 2.91575462e-01
-9.92693186e-01 -2.80185223e-01 5.65676987e-01 1.27495539e+00
-8.55483651e-01 1.08920896e+00 -3.75601679e-01 -6.35945559e-01
6.71305060e-02 4.14034247e-01 -5.79902470e-01 3.77534360e-01
-1.21354759e+00 1.16773152e+00 3.67617875e-01 -3.16988260e-01
-1.99398488e-01 -2.29031220e-01 -4.27565783e-01 -5.26243448e-02
-4.81950352e-03 -2.61729240e-01 6.99768126e-01 -1.14968431e+00
-1.56310368e+00 1.21331441e+00 4.33839947e-01 -3.61663878e-01
9.24421549e-01 -3.17224637e-02 -4.31322366e-01 -3.53434145e-01
8.14712122e-02 1.84788838e-01 3.75400066e-01 -9.00598586e-01
-4.72647130e-01 -7.79944301e-01 1.75245866e-01 -1.16440959e-01
-5.58545053e-01 6.85517967e-01 3.68781567e-01 -1.93810076e-01
-4.12723154e-01 -9.31545615e-01 -1.03042319e-01 -6.70066893e-01
-1.84122637e-01 -1.65636420e-01 1.87967539e-01 -9.34774399e-01
1.42580414e+00 -2.17607498e+00 3.43855947e-01 3.48828524e-01
-1.35596350e-01 3.25908691e-01 3.84406686e-01 6.31759405e-01
-5.00964165e-01 3.99412870e-01 4.47966121e-02 -6.00872934e-01
-6.25279322e-02 1.75423801e-01 2.21099555e-01 3.22732985e-01
-1.72500774e-01 4.68230154e-03 -5.81088006e-01 -4.84049112e-01
-1.83577314e-02 1.88997716e-01 -6.08056486e-01 -1.71288371e-01
2.30430335e-01 5.45236230e-01 -1.03613585e-01 1.08202688e-01
4.04090524e-01 4.90956426e-01 8.94932225e-02 4.76708978e-01
-2.57257015e-01 3.41176242e-01 -1.08728659e+00 1.11232340e+00
-4.54778016e-01 4.91422415e-01 6.65735826e-02 -9.86755610e-01
1.13933051e+00 1.91549599e-01 3.91990036e-01 -5.84242702e-01
5.04125774e-01 3.97783518e-01 3.86890590e-01 -7.92868555e-01
2.07715720e-01 -1.40534446e-01 7.87045509e-02 2.99054414e-01
1.28239572e-01 3.17106955e-02 4.72335696e-01 -1.41028360e-01
9.62988198e-01 -1.47282586e-01 1.28680781e-01 1.00994315e-02
6.93894446e-01 -2.08490223e-01 2.11083472e-01 5.24385273e-01
-4.01066579e-02 2.37522408e-01 1.00286543e+00 -1.63067624e-01
-6.59112811e-01 -5.72411418e-01 -3.31705004e-01 1.23942542e+00
-5.67729175e-01 -6.02908254e-01 -9.30676103e-01 -7.58854389e-01
-6.01392686e-02 8.25463712e-01 -6.61669493e-01 -2.12566942e-01
-2.76965439e-01 -1.21933854e+00 6.89811826e-01 -3.00360262e-01
1.54560223e-01 -8.27001512e-01 -3.70804191e-01 4.51232828e-02
7.10543469e-02 -8.56808186e-01 7.37895906e-01 2.94339389e-01
-7.99933016e-01 -1.02794397e+00 -2.22793147e-01 -1.87123701e-01
-3.27530988e-02 -4.15916145e-01 7.91337729e-01 3.45761865e-01
-1.06059544e-01 1.50870338e-01 -7.96964526e-01 -7.21774161e-01
-7.62687743e-01 3.98300081e-01 7.77391195e-02 1.37591928e-01
4.26268727e-01 -8.74217987e-01 1.96881652e-01 2.41782814e-01
-7.97584653e-01 -3.12249064e-01 3.65017027e-01 4.34897751e-01
-4.22473162e-01 -1.32796273e-01 4.83814895e-01 -1.19772196e+00
1.01834893e+00 -7.46977150e-01 -3.11899763e-02 -1.91867203e-01
-6.43590391e-01 -4.35618073e-01 4.41710621e-01 -4.30876881e-01
-6.76907480e-01 -3.79144132e-01 -7.71011710e-01 2.75236279e-01
-7.18084037e-01 5.44060051e-01 4.92820889e-02 -1.13068379e-01
1.18122876e+00 -2.85038114e-01 2.56139725e-01 -6.35160208e-01
-2.47231677e-01 1.04599452e+00 -4.51892823e-01 -3.64900291e-01
4.50583786e-01 9.07083899e-02 -6.98566735e-02 -9.29024696e-01
-3.54720443e-01 -3.66988540e-01 -7.26999104e-01 -4.46493357e-01
8.45444679e-01 -5.49647510e-01 -8.77942741e-01 6.95264280e-01
-1.00401270e+00 -3.63771021e-01 -2.21623659e-01 7.61445642e-01
-3.22245091e-01 3.45718503e-01 -5.77385187e-01 -9.53349769e-01
-3.44346394e-03 -9.18973327e-01 3.57900828e-01 1.02037087e-01
-6.03991568e-01 -9.68097866e-01 5.83027482e-01 7.10234523e-01
3.80135626e-01 2.26232976e-01 8.76657426e-01 -1.47423816e+00
5.13591588e-01 -4.00277227e-01 1.86140433e-01 6.55212164e-01
-1.59947321e-01 1.85965493e-01 -1.15110052e+00 -2.82732155e-02
6.92150043e-03 -4.98695582e-01 6.72094524e-01 -3.47488731e-01
9.02885854e-01 -1.96937278e-01 -5.59016950e-02 -1.99304018e-02
1.03565300e+00 2.13183194e-01 7.30471432e-01 8.33872139e-01
1.87121868e-01 1.11963451e+00 3.70837212e-01 5.70753098e-01
1.74581930e-01 9.17702794e-01 3.19808304e-01 3.08792200e-02
9.48594436e-02 1.01881541e-01 5.92634737e-01 4.00373787e-01
-3.67046595e-01 1.56421244e-01 -9.56487238e-01 1.32225886e-01
-1.58906901e+00 -1.00060499e+00 -5.19228876e-01 2.19587541e+00
4.73980248e-01 5.88583052e-01 9.24224973e-01 7.00447977e-01
7.09472001e-01 -1.36731192e-01 1.48984820e-01 -1.31128538e+00
-1.10447310e-01 2.91850835e-01 2.39146665e-01 5.05578101e-01
-6.31176829e-01 3.59865278e-01 5.58262300e+00 7.77662814e-01
-1.32730234e+00 3.48852873e-01 6.72058344e-01 -3.93062621e-01
2.20549211e-01 -1.97060332e-01 -7.44354129e-01 8.73951018e-01
1.39351487e+00 1.06153227e-01 4.50832307e-01 7.14783669e-01
1.79459780e-01 -2.95277685e-01 -8.08954239e-01 5.49013138e-01
2.21113920e-01 -5.73811471e-01 -3.56820464e-01 3.00585747e-01
4.64412905e-02 1.64050594e-01 -1.37254164e-01 6.28853798e-01
1.41582668e-01 -8.76438081e-01 6.58660889e-01 5.73035121e-01
7.51299113e-02 -4.26196754e-01 1.12159610e+00 6.40857279e-01
-6.02635704e-02 -4.78690594e-01 2.78530084e-02 -6.44398987e-01
-1.81697369e-01 4.04436022e-01 -1.02427971e+00 6.61654890e-01
1.04111993e+00 1.87776402e-01 -1.03761446e+00 1.11196041e+00
-7.53911436e-02 6.89877212e-01 -5.09721339e-01 -5.98707020e-01
-3.02529382e-03 -3.61265749e-01 4.17392701e-01 1.31018949e+00
7.01511651e-02 -4.03094351e-01 -2.75509119e-01 2.64961064e-01
5.66478729e-01 8.50709617e-01 -4.81512308e-01 2.35520884e-01
-1.48483261e-01 1.25148559e+00 -5.97654521e-01 -1.08299423e-02
-4.93428320e-01 5.05143523e-01 6.40042365e-01 -1.23322502e-01
-8.31969678e-01 -1.45860165e-01 3.22038740e-01 6.95818663e-01
-3.21962118e-01 1.31602794e-01 -3.06959182e-01 -6.86088920e-01
-2.05498606e-01 -8.95477474e-01 6.61189973e-01 -6.57981277e-01
-1.29525185e+00 6.03449166e-01 3.32814187e-01 -9.05503333e-01
-2.00364098e-01 -8.56388271e-01 -5.88129580e-01 7.19004154e-01
-7.76307940e-01 -8.69975090e-01 -1.53526306e-01 4.69412059e-01
-6.51505515e-02 -3.48563641e-01 6.89593196e-01 6.95551753e-01
-7.95423031e-01 4.23378050e-01 -3.28549072e-02 1.32678255e-01
7.57464290e-01 -1.14558244e+00 -3.14239979e-01 1.21174484e-01
3.36326356e-03 1.15177691e-01 7.41334498e-01 -5.89249253e-01
-3.02555114e-01 -5.20292342e-01 8.74243319e-01 -4.09780920e-01
9.18648660e-01 -3.41076970e-01 -8.23778152e-01 3.63481402e-01
1.75085336e-01 -7.56020486e-01 1.18115640e+00 7.16957450e-01
-1.96756199e-01 -8.28585923e-02 -1.32905972e+00 4.27082568e-01
9.68132794e-01 -2.90963113e-01 -4.12896037e-01 5.08098066e-01
-2.04376411e-02 -1.17129311e-01 -8.97936046e-01 2.95688570e-01
4.77698773e-01 -1.64742577e+00 5.73536992e-01 -1.03895450e+00
5.47562003e-01 4.67910290e-01 1.41950354e-01 -1.49637246e+00
-1.85867071e-01 -1.14966892e-01 6.88364089e-01 1.55384922e+00
8.04705083e-01 -1.00026131e+00 7.72471249e-01 7.40625501e-01
8.35063457e-02 -5.77736795e-01 -1.27154779e+00 -5.61284065e-01
5.11089921e-01 -6.07760787e-01 3.70017141e-01 1.06503654e+00
6.47996306e-01 5.12448490e-01 -1.75829887e-01 -2.47604147e-01
-6.86812326e-02 -5.79307318e-01 1.13639295e+00 -1.65616179e+00
-5.92573166e-01 -6.96259618e-01 -9.37483490e-01 2.50329942e-01
4.61603224e-01 -9.79657531e-01 -8.96402478e-01 -1.10839689e+00
9.74976569e-02 -4.57811892e-01 9.05263945e-02 3.59367311e-01
1.08460346e-02 1.59521416e-01 3.15365702e-01 -3.05648804e-01
-2.43740603e-01 7.12467879e-02 5.57882607e-01 1.57767072e-01
-2.27939874e-01 3.34551603e-01 -5.18261790e-01 8.73515308e-01
1.01185930e+00 -7.39206195e-01 -3.43811465e-03 -3.25524225e-03
5.83870590e-01 -1.15776211e-01 2.20651507e-01 -1.19776273e+00
1.41803890e-01 5.57917729e-02 9.00136456e-02 1.50027424e-01
6.32446945e-01 -7.03182697e-01 3.54721278e-01 6.22761548e-01
-1.87609732e-01 -2.11643074e-02 2.00898692e-01 -2.16514058e-03
-1.28773168e-01 -1.05923104e+00 5.96602380e-01 -2.58180380e-01
-1.77488904e-02 -2.87996322e-01 -6.30631745e-01 -2.25561678e-01
1.14482176e+00 -3.31367314e-01 -3.07715535e-01 -3.75506461e-01
-1.30501425e+00 -1.33960322e-01 3.82704943e-01 3.70675951e-01
-7.27808252e-02 -6.02159798e-01 -9.78312790e-01 -3.61281596e-02
-5.55877090e-02 -7.94285536e-01 3.15108329e-01 1.13492811e+00
-5.07471085e-01 5.66040426e-02 -4.67956424e-01 -2.81382233e-01
-1.81655705e+00 6.20631874e-01 4.73058254e-01 -4.31194395e-01
3.15972328e-01 5.62105179e-01 -6.05165422e-01 -7.05704987e-01
9.98154208e-02 6.12224638e-02 -8.32399130e-01 6.66898966e-01
1.81954309e-01 8.58550906e-01 5.25693417e-01 -8.25972617e-01
-1.18763313e-01 3.19834501e-01 -1.89550236e-01 -3.57591242e-01
1.37297630e+00 2.15721354e-01 -3.73940676e-01 7.12559879e-01
1.20051622e+00 5.07950783e-01 -1.56743377e-02 2.26336017e-01
2.46144593e-01 -4.84967232e-01 -2.65245318e-01 -7.24116683e-01
-6.02358878e-01 7.34204769e-01 7.35099912e-01 8.28133345e-01
7.87275434e-01 -1.12353927e-02 -2.18504384e-01 2.76756167e-01
4.44036692e-01 -1.25235713e+00 -1.86548695e-01 4.41594750e-01
8.92112792e-01 -9.89745319e-01 1.11429520e-01 -5.44600248e-01
-5.44587910e-01 1.14211345e+00 6.32411659e-01 1.53978512e-01
7.50292242e-01 9.59360972e-03 -1.31226480e-01 -1.06581256e-01
-5.66405773e-01 -3.13771397e-01 -2.31568530e-01 5.25141716e-01
3.90377551e-01 -5.67772193e-03 -1.13423526e+00 9.17136371e-01
-7.15690792e-01 9.17123854e-02 7.70143688e-01 5.73883474e-01
-1.81354761e-01 -1.36264181e+00 -6.44776165e-01 6.94716811e-01
-8.26724410e-01 2.51707047e-01 -1.08562219e+00 9.90291655e-01
7.69413829e-01 1.04938090e+00 -4.15992998e-02 -6.83022976e-01
6.45820022e-01 3.31175119e-01 3.47070664e-01 -6.57524288e-01
-1.46007133e+00 -6.43034756e-01 8.08097661e-01 -1.23386927e-01
-2.60400534e-01 -1.12418842e+00 -5.66720963e-01 -4.54545289e-01
-4.55044538e-01 4.05828804e-01 1.11659431e+00 9.76180434e-01
-2.96379756e-02 2.59701818e-01 6.57544315e-01 -5.63417792e-01
-5.03280222e-01 -1.37690973e+00 -5.33177674e-01 7.76241362e-01
-1.98142141e-01 -6.89470112e-01 -7.70040572e-01 -5.22261977e-01] | [8.816535949707031, 10.64580249786377] |
890ffa7c-2b27-4d16-90e4-f9fc4f111855 | passive-defense-against-3d-adversarial-point | 2205.08738 | null | https://arxiv.org/abs/2205.08738v3 | https://arxiv.org/pdf/2205.08738v3.pdf | 3D-VFD: A Victim-free Detector against 3D Adversarial Point Clouds | 3D deep models consuming point clouds have achieved sound application effects in computer vision. However, recent studies have shown they are vulnerable to 3D adversarial point clouds. In this paper, we regard these malicious point clouds as 3D steganography examples and present a new perspective, 3D steganalysis, to counter such examples. Specifically, we propose 3D-VFD, a victim-free detector against 3D adversarial point clouds. Its core idea is to capture the discrepancies between residual geometric feature distributions of benign point clouds and adversarial point clouds and map these point clouds to a lower dimensional space where we can efficiently distinguish them. Unlike existing detection techniques against 3D adversarial point clouds, 3D-VFD does not rely on the victim 3D deep model's outputs for discrimination. Extensive experiments demonstrate that 3D-VFD achieves state-of-the-art detection and can effectively detect 3D adversarial attacks based on point adding and point perturbation while keeping fast detection speed. | ['Xiaohua Xie', 'Yi Zhou', 'Zixuan Chen', 'Huajun Zhou', 'Jiahao Zhu'] | 2022-05-18 | null | null | null | null | ['steganalysis'] | ['computer-vision'] | [ 2.39513323e-01 2.76531249e-01 2.75699288e-01 4.80001360e-01
-7.32339442e-01 -1.17452908e+00 8.92991245e-01 -3.14089417e-01
-3.16324793e-02 -4.24314700e-02 -6.90416694e-01 -5.74523926e-01
5.76458454e-01 -1.05392635e+00 -1.14087641e+00 -8.03683043e-01
-1.39181197e-01 5.00004768e-01 5.51637053e-01 -5.61996400e-01
4.92535025e-01 1.28785253e+00 -1.07552290e+00 -1.40152171e-01
4.63341743e-01 6.72046840e-01 -3.53780210e-01 1.01653504e+00
-4.69972491e-02 1.76055897e-02 -1.07881200e+00 -6.20853782e-01
1.12458897e+00 1.80291533e-02 -1.31069258e-01 -6.02931976e-02
7.39525437e-01 -4.64011222e-01 -8.32059801e-01 1.71857977e+00
4.59899187e-01 -4.77347255e-01 6.69432461e-01 -1.94521511e+00
-8.61497998e-01 -3.30548808e-02 -7.56978333e-01 -4.75474820e-02
4.40545022e-01 7.44352221e-01 7.25716352e-02 -6.13609493e-01
3.18216294e-01 1.87766910e+00 9.04092133e-01 9.10721898e-01
-9.81430531e-01 -1.40311563e+00 -3.16960365e-01 -2.98850268e-01
-1.48825741e+00 2.50872434e-03 1.01934242e+00 -4.52044904e-01
6.80200040e-01 6.63790226e-01 5.41448832e-01 1.33992183e+00
4.19708490e-01 4.35768157e-01 9.82800961e-01 -2.19266385e-01
-1.18040979e-01 9.68499202e-03 -4.41928029e-01 5.76405466e-01
5.64304709e-01 8.20613027e-01 1.62488177e-01 -7.92584538e-01
9.57680225e-01 2.18825251e-01 -3.70094180e-01 -5.50695777e-01
-1.23832595e+00 1.03517127e+00 8.26372802e-01 -1.48138732e-01
-9.45263132e-02 4.59826589e-01 2.18425959e-01 6.13515675e-01
2.99891710e-01 3.93695444e-01 2.94126794e-02 4.77243155e-01
-1.14941597e-01 4.25329328e-01 6.71290815e-01 1.31040061e+00
4.03252989e-01 1.55704960e-01 2.73588240e-01 -1.04167230e-01
5.46647668e-01 1.50876009e+00 -2.02167138e-01 -8.22290301e-01
4.41927671e-01 3.63433838e-01 -1.92226712e-02 -1.69669402e+00
2.58053601e-01 6.01972751e-02 -1.06937122e+00 1.07705820e+00
6.30461946e-02 8.75820518e-02 -9.64342237e-01 1.30400634e+00
6.72525346e-01 6.48974419e-01 3.64565074e-01 9.01954651e-01
6.88160002e-01 4.66538072e-01 -2.91481614e-01 3.78394634e-01
9.98410404e-01 -1.76287189e-01 1.63942099e-01 -1.06109649e-01
5.31176865e-01 -7.12275028e-01 4.54629064e-01 -3.28443907e-02
-9.59532917e-01 -4.00392532e-01 -1.27064729e+00 4.13446099e-01
-4.60998684e-01 -8.26452374e-01 2.28581488e-01 1.20663702e+00
-8.96108568e-01 4.25456762e-01 -8.50159645e-01 -8.70513637e-03
8.33445966e-01 6.96142435e-01 -5.59091985e-01 5.51909357e-02
-1.25735199e+00 8.80623519e-01 8.03377405e-02 -2.53829896e-01
-1.06705713e+00 -6.53188944e-01 -7.82702863e-01 -4.08587664e-01
1.86844058e-02 -9.09920037e-01 8.83860648e-01 -5.16467988e-01
-1.00777137e+00 1.53157747e+00 3.86911631e-01 -7.43043065e-01
8.89655232e-01 1.18931472e-01 -3.86685878e-01 2.67609239e-01
8.87449980e-02 5.76704323e-01 1.38620341e+00 -1.63303602e+00
-3.65293026e-01 -8.64063323e-01 3.54478694e-02 -3.69106382e-02
3.67681772e-01 -6.03915192e-02 -1.86974764e-01 -5.37223637e-01
4.76069123e-01 -1.43323588e+00 -4.43626940e-01 6.59128726e-01
-8.81687641e-01 1.94058120e-01 1.34179974e+00 3.19645315e-01
-1.29145205e-01 -2.46244931e+00 -3.12905878e-01 4.43356335e-01
7.42419302e-01 8.05937290e-01 -9.55120921e-02 2.62840003e-01
-1.36266276e-01 5.78393757e-01 -9.69757289e-02 -2.69581936e-02
1.56413019e-01 2.22690359e-01 -8.76482844e-01 1.32435513e+00
2.56536812e-01 1.04926467e+00 -1.12845552e+00 -1.69988468e-01
6.74337804e-01 6.00881934e-01 -4.18101072e-01 -1.93123832e-01
1.19972490e-01 4.51638252e-01 -7.71975815e-01 6.57618046e-01
1.71773398e+00 1.38824612e-01 -7.23287702e-01 1.85911387e-01
4.15188521e-01 -1.55109346e-01 -7.27995932e-01 8.57158899e-01
2.49090493e-01 6.38536215e-01 1.01147786e-01 -5.57705879e-01
1.25689232e+00 1.95572779e-01 3.62732083e-01 -1.56799242e-01
3.36951375e-01 2.84754038e-01 -1.43032819e-01 -3.55600715e-02
2.09904015e-01 -8.27307813e-04 -4.86084759e-01 1.18498221e-01
-5.01482964e-01 -8.97148728e-01 -1.11174524e+00 1.12959504e-01
1.44424438e+00 -5.39275467e-01 8.65534917e-02 2.61151940e-01
7.44525969e-01 1.75882608e-01 2.28486866e-01 1.34461451e+00
-4.51086760e-01 6.76341355e-01 1.43504485e-01 -2.31643945e-01
-1.24975812e+00 -1.54736662e+00 7.46436566e-02 -9.23513174e-02
8.80530238e-01 2.48213992e-01 -3.82031828e-01 -1.14274526e+00
7.85546005e-01 5.62748611e-01 -5.70982456e-01 -7.11358607e-01
-5.97725987e-01 4.39507216e-02 1.31819308e+00 -8.76028985e-02
6.20048463e-01 -5.76100349e-01 -4.57113892e-01 -1.93641961e-01
5.22032022e-01 -1.06834304e+00 -3.19899380e-01 -4.43979532e-01
-8.12917352e-01 -1.37171066e+00 -5.79711974e-01 -7.34892070e-01
7.29613364e-01 1.00918579e+00 1.03672457e+00 4.06610340e-01
1.27529353e-01 4.83024031e-01 -4.51921940e-01 -1.09338474e+00
-1.22877574e+00 -4.65361118e-01 4.63692993e-01 -3.75967443e-01
8.54418278e-01 -5.98352492e-01 -5.03749669e-01 5.17656744e-01
-8.52139652e-01 -5.36049843e-01 3.98393393e-01 3.45160306e-01
4.96255308e-01 -7.57960305e-02 -2.86748260e-01 -7.27724731e-01
1.59431979e-01 -4.40232158e-01 -8.81756425e-01 -2.94612139e-01
-9.10501778e-02 -4.38403547e-01 3.78538847e-01 -7.32179224e-01
2.05481321e-01 6.28002211e-02 -3.46777231e-01 -1.41382504e+00
-3.50425482e-01 -6.42682910e-01 -2.53815264e-01 -1.24667037e+00
7.66631007e-01 4.45959359e-01 3.75687391e-01 -5.55546433e-02
3.78325552e-01 6.15755737e-01 9.00992572e-01 3.82882357e-02
2.08834982e+00 1.06438458e+00 4.20001566e-01 -7.55681753e-01
-1.18398607e-01 -3.53258163e-01 -4.97816354e-01 -1.62415117e-01
6.93422198e-01 -9.74178672e-01 -1.04104710e+00 8.19452345e-01
-1.64054656e+00 3.55554163e-01 -1.69377461e-01 1.90228969e-01
-5.78836620e-01 7.71740496e-01 -2.03526422e-01 -7.22147167e-01
-2.11446613e-01 -1.27684569e+00 1.37918663e+00 -3.98560762e-01
2.27380261e-01 -8.25523317e-01 -1.01946026e-01 -1.38997421e-01
3.55550721e-02 1.16829956e+00 4.55315530e-01 -9.23204541e-01
-8.55288029e-01 -9.25434351e-01 -1.23918476e-02 2.68133461e-01
1.56490132e-01 -6.80155233e-02 -8.17256868e-01 -5.52429855e-01
5.96635163e-01 1.94426551e-01 4.78560060e-01 -5.16587161e-02
9.01389480e-01 -4.45882559e-01 -6.57963336e-01 1.13187039e+00
1.20670474e+00 1.46253869e-01 7.54331112e-01 5.31138301e-01
1.02028763e+00 -7.58481696e-02 5.17424464e-01 3.11576016e-02
-1.53694287e-01 4.52565670e-01 1.38532090e+00 -1.07479900e-01
1.46124855e-01 -3.86922359e-01 2.40017831e-01 2.13000461e-01
3.88546079e-01 -3.89876723e-01 -9.14228380e-01 2.36195728e-01
-1.19681478e+00 -8.91726494e-01 -5.23471773e-01 1.99217403e+00
1.15977354e-01 5.43002725e-01 -1.40553877e-01 2.08029404e-01
1.11463857e+00 4.40489441e-01 -7.93021381e-01 -2.22018600e-01
-2.25431278e-01 -4.60483283e-02 1.34190893e+00 3.19710374e-01
-1.30856919e+00 1.09262669e+00 5.70438290e+00 5.93907475e-01
-9.84750509e-01 6.24526888e-02 -2.62250029e-03 1.72460973e-01
-2.77491957e-01 -2.16811091e-01 -6.11416459e-01 5.33474565e-01
2.39410818e-01 -3.39901865e-01 -1.73267797e-02 1.04007113e+00
-3.56575787e-01 8.05010200e-01 -8.78906846e-01 1.19853163e+00
5.23370542e-02 -1.42959666e+00 4.17854398e-01 7.37810194e-01
4.94174123e-01 3.67917687e-01 5.65603077e-01 1.43644810e-02
8.37303400e-01 -8.10142338e-01 4.29480523e-01 -1.59780181e-03
9.04071629e-01 -8.46537471e-01 5.44754505e-01 6.89669907e-01
-7.12068200e-01 2.70771116e-01 -8.32633495e-01 3.89900893e-01
-7.69643206e-03 7.93291330e-02 -1.03513229e+00 4.89015758e-01
7.32348919e-01 5.00590205e-01 -1.71701580e-01 7.98222423e-01
-2.79566497e-01 2.40276560e-01 -6.16686106e-01 4.49764691e-02
6.11262739e-01 1.55534729e-01 1.76907945e+00 7.34759033e-01
5.65112770e-01 2.90297747e-01 7.75613487e-02 1.00765717e+00
-8.85384753e-02 -6.78832114e-01 -1.59179378e+00 2.46322066e-01
8.59838247e-01 4.72873598e-01 -3.34936589e-01 -5.03661484e-02
-6.82922974e-02 1.11804187e+00 -3.62723529e-01 -6.98302239e-02
-8.79484892e-01 -3.89780611e-01 1.50569332e+00 6.89434644e-04
6.03769362e-01 -5.18543005e-01 -1.30125567e-01 -9.79939640e-01
-3.96327943e-01 -9.22370017e-01 -2.32451528e-01 -7.30701625e-01
-1.52148342e+00 3.66718471e-01 -3.10018629e-01 -1.92591953e+00
-7.83160478e-02 -8.46723497e-01 -7.00119674e-01 9.63019848e-01
-1.23556101e+00 -1.27633297e+00 -1.62999764e-01 1.05837715e+00
-6.76252320e-02 -3.98328751e-01 7.44330585e-01 -2.72431582e-01
2.58719802e-01 7.11262941e-01 1.26465037e-01 5.45962095e-01
3.05488318e-01 -1.10481131e+00 1.60344625e+00 9.37125862e-01
1.03537902e-01 3.32431763e-01 8.55142474e-01 -9.24550593e-01
-1.81264007e+00 -1.19279289e+00 2.34108120e-01 -1.02328300e+00
5.80791235e-01 -4.81281817e-01 -9.76947606e-01 6.36914730e-01
-3.71186882e-01 1.66763887e-01 3.25519413e-01 -9.52372074e-01
-7.50075102e-01 3.55491340e-01 -1.90443492e+00 7.05003858e-01
1.16075242e+00 -6.00235343e-01 -7.32485294e-01 6.35899961e-01
1.18980575e+00 -9.18397546e-01 -5.96612036e-01 4.33843285e-01
1.86314266e-02 -9.81919825e-01 1.75062716e+00 -4.37269449e-01
-2.72444133e-02 -5.83282292e-01 -3.17638844e-01 -1.05057287e+00
-3.02205890e-01 -1.02720451e+00 -2.73919821e-01 6.32966578e-01
-2.05622956e-01 -1.08630478e+00 1.12047040e+00 3.49703692e-02
-6.99704736e-02 -2.44664978e-02 -1.29191625e+00 -1.03728247e+00
3.32768023e-01 -4.83739018e-01 1.15213823e+00 1.10129428e+00
-8.80348206e-01 -5.51303387e-01 -2.10013345e-01 1.38785958e+00
1.32556975e+00 -1.24467626e-01 1.48985243e+00 -1.29986453e+00
2.14381367e-01 -4.84808952e-01 -1.45932722e+00 -1.16898274e+00
2.19047904e-01 -8.99527907e-01 -1.85943112e-01 -6.80978060e-01
-6.43083334e-01 -5.40212810e-01 1.15808845e-01 6.42037392e-02
-8.97270367e-02 8.02826405e-01 6.49867415e-01 6.26928449e-01
-4.11792211e-02 3.54089618e-01 1.27350020e+00 -5.33803761e-01
1.07647024e-01 5.52043617e-01 -6.27822876e-01 7.33645439e-01
6.44520998e-01 -9.37163413e-01 -2.28884220e-02 -4.83569056e-01
-1.08659059e-01 -1.06011786e-01 1.08968008e+00 -1.19928670e+00
1.50820747e-01 -4.76498045e-02 4.10366207e-01 -1.00888038e+00
5.99788070e-01 -1.37132370e+00 1.79416701e-01 1.10670459e+00
3.33404660e-01 -1.10621052e-02 2.12428972e-01 9.04940069e-01
7.88937584e-02 1.19361110e-01 1.08145773e+00 -3.50047916e-01
-5.32939434e-01 8.19780469e-01 -8.67297500e-03 -2.05854908e-01
1.53535128e+00 -7.22323000e-01 -5.39485991e-01 -3.02486002e-01
-3.77955526e-01 9.77482125e-02 1.14638889e+00 5.63562036e-01
1.07582927e+00 -1.48811162e+00 -8.29606295e-01 7.05663264e-01
2.46856317e-01 2.60860324e-01 -2.16500578e-03 1.18000329e-01
-1.01420343e+00 1.02106825e-01 -8.82459134e-02 -1.29636514e+00
-1.53514922e+00 1.12203610e+00 6.41750693e-01 4.32707161e-01
-8.77138495e-01 1.11923540e+00 2.63150692e-01 -7.40543723e-01
-6.49348125e-02 7.69865811e-02 3.86646718e-01 -7.40921319e-01
4.19814318e-01 9.21458080e-02 -3.20302844e-01 -1.17261779e+00
-5.62885940e-01 7.85893440e-01 -5.39452136e-02 2.52624780e-01
9.06929493e-01 1.95766483e-02 9.66608599e-02 -1.00699745e-01
1.38155234e+00 1.80057958e-01 -1.13860369e+00 -1.37689188e-01
-4.31098163e-01 -1.25581932e+00 -2.05736279e-01 6.12190887e-02
-1.00477087e+00 7.47092426e-01 8.61208916e-01 6.31459057e-01
6.63797855e-01 1.97953463e-01 1.01354992e+00 3.31545800e-01
7.09159315e-01 2.03146294e-01 -2.32608423e-01 5.31213164e-01
7.88262606e-01 -1.26030385e+00 -7.73857236e-02 -6.95034564e-01
-1.54168293e-01 7.98156381e-01 4.10084367e-01 -1.04844654e+00
7.07063138e-01 6.65635020e-02 3.53393286e-01 -3.91852051e-01
-8.93619657e-02 9.63123515e-02 8.98849666e-02 1.34235072e+00
-1.00251567e+00 -5.96552752e-02 7.25573659e-01 -3.85966331e-01
-6.41252697e-01 -6.02095187e-01 3.00486237e-01 8.61626446e-01
-4.14885730e-01 -8.92880082e-01 -1.07643723e+00 -4.68252599e-02
-3.99322480e-01 1.55869350e-01 -8.40833843e-01 1.16209507e+00
1.33332191e-02 7.44094551e-01 1.60651937e-01 -9.76404846e-01
5.76762497e-01 -6.77522004e-01 4.08862680e-01 -3.36599022e-01
-6.53937221e-01 -4.10866499e-01 -7.60602891e-01 -5.83287954e-01
-1.44947350e-01 -4.78742033e-01 -6.94806993e-01 -9.81594145e-01
-3.10906321e-01 -8.64843503e-02 7.34114468e-01 4.50989664e-01
5.71711600e-01 -1.06937036e-01 1.39141369e+00 -1.34214270e+00
-1.02274156e+00 -3.57672930e-01 -5.25483906e-01 5.63238263e-01
1.08584929e+00 -5.03308594e-01 -1.08575952e+00 -6.84903741e-01] | [7.702873229980469, -4.467500686645508] |
7235d322-a13a-4726-b53b-1807af840006 | discrete-graph-structure-learning-for-1 | 2101.06861 | null | https://arxiv.org/abs/2101.06861v3 | https://arxiv.org/pdf/2101.06861v3.pdf | Discrete Graph Structure Learning for Forecasting Multiple Time Series | Time series forecasting is an extensively studied subject in statistics, economics, and computer science. Exploration of the correlation and causation among the variables in a multivariate time series shows promise in enhancing the performance of a time series model. When using deep neural networks as forecasting models, we hypothesize that exploiting the pairwise information among multiple (multivariate) time series also improves their forecast. If an explicit graph structure is known, graph neural networks (GNNs) have been demonstrated as powerful tools to exploit the structure. In this work, we propose learning the structure simultaneously with the GNN if the graph is unknown. We cast the problem as learning a probabilistic graph model through optimizing the mean performance over the graph distribution. The distribution is parameterized by a neural network so that discrete graphs can be sampled differentiably through reparameterization. Empirical evaluations show that our method is simpler, more efficient, and better performing than a recently proposed bilevel learning approach for graph structure learning, as well as a broad array of forecasting models, either deep or non-deep learning based, and graph or non-graph based. | ['Jinbo Bi', 'Jie Chen', 'Chao Shang'] | 2021-01-18 | discrete-graph-structure-learning-for | https://openreview.net/forum?id=WEHSlH5mOk | https://openreview.net/pdf?id=WEHSlH5mOk | iclr-2021-1 | ['graph-structure-learning'] | ['graphs'] | [-1.82941154e-01 3.00563514e-01 -2.95439720e-01 -3.73382211e-01
-1.15044661e-01 -7.11126089e-01 7.84806192e-01 3.14643800e-01
1.72587439e-01 5.34794629e-01 3.28876346e-01 -6.78962231e-01
-3.94949675e-01 -1.01652551e+00 -9.88172948e-01 -6.10642910e-01
-9.24631119e-01 7.36271322e-01 -4.34632450e-01 -1.08984992e-01
3.27699520e-02 6.08769178e-01 -1.04019094e+00 -3.03097337e-01
7.06812024e-01 8.89027953e-01 -1.38500303e-01 7.89736390e-01
-1.36182949e-01 9.45380628e-01 -3.50927085e-01 -4.72717792e-01
3.18855822e-01 -2.79323459e-01 -3.91388446e-01 6.94886670e-02
3.62457901e-01 5.91717288e-02 -7.65119553e-01 1.00968766e+00
1.41129583e-01 2.83577085e-01 8.24022710e-01 -1.56808293e+00
-7.81485856e-01 1.01838577e+00 -6.74585998e-01 2.30688900e-01
-3.47310714e-02 -1.11861743e-01 1.35555732e+00 -2.65204579e-01
4.13639098e-01 1.25935066e+00 1.04328239e+00 -2.26201460e-01
-1.55292428e+00 -7.11058319e-01 2.12281764e-01 2.54346073e-01
-1.19948018e+00 1.44655704e-02 1.38677919e+00 -6.95806265e-01
8.51968586e-01 -1.41340613e-01 8.42764974e-01 9.82218504e-01
7.48329997e-01 5.48480511e-01 9.68890548e-01 -2.57055581e-01
2.86524504e-01 -3.94757390e-01 2.32119203e-01 1.01470220e+00
2.90250123e-01 2.54170895e-01 -2.91832566e-01 -3.47363323e-01
7.31219828e-01 3.54062766e-01 -7.30150566e-02 -4.29565072e-01
-9.78650153e-01 1.26477766e+00 6.68228507e-01 2.95738667e-01
-6.63479626e-01 5.95089614e-01 3.25642675e-01 5.20803094e-01
1.05675304e+00 5.08941352e-01 -3.25056940e-01 2.10362181e-01
-8.76717150e-01 3.43468301e-02 1.25446272e+00 6.08691990e-01
9.48005021e-01 5.55174470e-01 1.32770032e-01 3.28251004e-01
2.46922970e-01 4.81036276e-01 5.26715219e-02 -7.22502649e-01
5.04920483e-01 5.58652341e-01 -4.03706729e-01 -1.77356064e+00
-9.57487166e-01 -7.17888236e-01 -1.41212964e+00 -1.59740314e-01
5.15272915e-01 -4.73064005e-01 -9.48649764e-01 1.83167350e+00
9.26679298e-02 8.02038491e-01 -3.07928860e-01 4.80750322e-01
5.52224278e-01 8.12627971e-01 -1.23488888e-01 -2.23615050e-01
7.02808738e-01 -6.22718751e-01 -6.32476687e-01 -8.16836730e-02
7.27735400e-01 -3.01891983e-01 6.19530678e-01 2.42360517e-01
-5.50197065e-01 -3.14699739e-01 -8.66882801e-01 2.08244413e-01
-5.01409352e-01 -3.37265760e-01 1.07951760e+00 6.21698678e-01
-1.40259099e+00 9.77374971e-01 -9.33118284e-01 -1.58918664e-01
3.90951663e-01 3.90969634e-01 -1.93672374e-01 -1.16189897e-01
-1.26605558e+00 6.85503423e-01 3.36867034e-01 6.20436743e-02
-6.18619025e-01 -6.40364110e-01 -9.73273277e-01 3.79735440e-01
2.73438871e-01 -8.30299914e-01 7.01131284e-01 -9.49146807e-01
-1.26403058e+00 3.24337870e-01 1.42482370e-01 -8.39864969e-01
1.47050351e-01 5.08631408e-01 -4.28528339e-01 -6.49517328e-02
-2.67909259e-01 1.18299767e-01 1.07352948e+00 -9.08301830e-01
-1.17659397e-01 -4.87712353e-01 1.52366668e-01 -1.69348821e-01
-2.45620891e-01 -3.71138752e-01 1.74466357e-01 -7.61251688e-01
2.91696768e-02 -1.06474984e+00 -3.70616078e-01 -3.74503940e-01
-5.18501103e-01 -4.66518521e-01 6.75593734e-01 -9.89388585e-01
1.21846259e+00 -1.72056568e+00 1.54381692e-01 8.15643311e-01
6.79765463e-01 -2.06100896e-01 -3.66074473e-01 8.51446211e-01
-8.20207745e-02 1.88933671e-01 -1.57183424e-01 -3.61430228e-01
6.75539151e-02 3.98064077e-01 -2.38235727e-01 8.02879214e-01
-1.09858122e-02 1.15691507e+00 -9.32239592e-01 8.26843530e-02
7.63331577e-02 6.18789613e-01 -3.22533876e-01 7.68805966e-02
-3.58640969e-01 3.77529114e-01 -2.97071069e-01 3.63178909e-01
5.66920400e-01 -8.16578388e-01 5.80113649e-01 -7.74511620e-02
4.54493791e-01 -1.01019256e-02 -9.77138817e-01 1.27343154e+00
-3.95390868e-01 8.47799301e-01 -2.95199335e-01 -1.62154758e+00
1.10969102e+00 1.13663122e-01 5.56321204e-01 -5.45553684e-01
1.46408260e-01 -1.09810628e-01 2.52119362e-01 -2.71538198e-01
2.96408176e-01 -1.12835132e-01 1.44046575e-01 5.23248076e-01
2.22751424e-01 1.06528081e-01 1.59114286e-01 2.49332592e-01
1.15654099e+00 -2.39123508e-01 2.85067201e-01 -3.81952256e-01
2.28127073e-02 -1.90072209e-01 1.82849690e-01 7.18971014e-01
2.59844482e-01 1.67835370e-01 1.01159000e+00 -6.72408283e-01
-1.08415425e+00 -8.14123154e-01 3.62736583e-01 9.77105319e-01
-4.07661229e-01 -3.91845286e-01 -3.32171589e-01 -5.66328108e-01
3.77864033e-01 7.90784121e-01 -8.77166331e-01 -1.50484636e-01
-5.66281199e-01 -7.14436769e-01 1.68373421e-01 6.22349024e-01
-6.90373257e-02 -7.59375215e-01 3.80291224e-01 2.36513123e-01
5.76272234e-02 -9.61879015e-01 -6.35388255e-01 9.14258584e-02
-1.04232287e+00 -1.07077599e+00 -4.22566146e-01 -5.62765121e-01
5.79916358e-01 1.95758924e-01 1.31596470e+00 -1.17124461e-01
2.74164006e-02 6.86406434e-01 4.20458755e-03 -4.40263778e-01
-4.19870228e-01 6.43693283e-02 -9.62715670e-02 3.28659207e-01
1.10872664e-01 -1.02343810e+00 -4.00620639e-01 -1.59263358e-01
-7.85106182e-01 -1.21354043e-01 2.46193290e-01 7.77091444e-01
4.21451002e-01 2.47428581e-01 6.19176626e-01 -1.01892745e+00
9.22263205e-01 -1.02181137e+00 -8.68379772e-01 2.66269445e-01
-9.05784845e-01 3.50206852e-01 9.21528459e-01 -5.59598625e-01
-2.98064530e-01 -2.54037440e-01 4.10930574e-01 -7.75425375e-01
1.90220565e-01 1.48949158e+00 4.07217354e-01 -1.89138174e-01
3.60486150e-01 6.38119206e-02 3.79229158e-01 -2.28353590e-01
4.91853237e-01 -1.52766585e-01 7.49052390e-02 -3.54022056e-01
6.96296334e-01 3.39662969e-01 8.09128165e-01 -6.70937300e-01
-6.91206217e-01 -2.76752651e-01 -5.36988258e-01 -3.67961168e-01
6.35461926e-01 -8.03286314e-01 -8.21792066e-01 2.30810732e-01
-1.02094448e+00 -4.43193346e-01 -1.05534270e-02 6.52264595e-01
-3.47072452e-01 2.29157865e-01 -4.26787525e-01 -8.07983696e-01
-1.95245415e-01 -6.22028530e-01 8.41611862e-01 -6.01404719e-03
-1.94983147e-02 -2.07802510e+00 3.01481426e-01 -1.48443773e-01
4.03479397e-01 6.70873702e-01 1.18451321e+00 -9.78532851e-01
-6.05447352e-01 -5.02666473e-01 -2.94725835e-01 1.35023698e-01
1.02826536e-01 5.85666597e-02 -5.05945981e-01 -4.36422944e-01
-2.42831782e-01 1.52205840e-01 8.15871477e-01 9.20466125e-01
1.23597705e+00 -7.59840012e-01 -2.45729953e-01 9.31262255e-01
1.38409710e+00 6.72314391e-02 1.99012488e-01 4.27189469e-03
1.24759936e+00 7.65228808e-01 -3.78426343e-01 4.82951939e-01
9.67721820e-01 1.16166778e-01 5.58196664e-01 1.57665126e-02
2.15282083e-01 -5.01594067e-01 2.54606664e-01 1.18545663e+00
-1.99223235e-01 -5.71575165e-01 -1.36079121e+00 2.97753096e-01
-2.04778433e+00 -1.06932104e+00 -2.29013115e-01 1.96260905e+00
1.83000907e-01 -2.01733857e-02 2.85925299e-01 -2.30894491e-01
7.71443725e-01 4.97132689e-01 -6.90556049e-01 -3.29508632e-01
-3.13292474e-01 1.97063744e-01 8.11154485e-01 5.24851918e-01
-1.08454633e+00 4.91723120e-01 6.62277508e+00 5.08975923e-01
-1.17182624e+00 -2.67499298e-01 8.90428662e-01 4.40028876e-01
-6.95310533e-01 1.07772099e-02 -3.44208896e-01 3.18799317e-01
1.37874913e+00 -5.65012157e-01 7.87547827e-01 5.89742541e-01
2.10612983e-01 4.62210655e-01 -1.19870937e+00 8.79825830e-01
1.64936736e-01 -1.71508181e+00 1.65657848e-02 3.92255664e-01
9.61117804e-01 3.30462992e-01 1.71514690e-01 4.61165160e-01
6.45455122e-01 -1.16042972e+00 2.93092251e-01 9.30536509e-01
3.18767130e-01 -9.01853800e-01 7.31050670e-01 3.98947060e-01
-1.32501173e+00 -1.34870410e-01 -3.35008502e-01 -4.25784051e-01
3.54437530e-02 7.91959822e-01 -1.04681838e+00 7.91278899e-01
2.90758252e-01 1.30087936e+00 -6.02774322e-01 9.28955317e-01
1.25476215e-02 1.08017361e+00 -3.48456562e-01 -2.16079101e-01
3.03587675e-01 -7.82430112e-01 6.32861257e-01 9.26652074e-01
4.11288708e-01 -8.99094939e-02 6.21957362e-01 8.63622487e-01
-3.49050015e-01 6.36056066e-02 -1.10624349e+00 -6.99007630e-01
1.25712216e-01 1.17178786e+00 -8.66450131e-01 -2.87511200e-01
-4.76806432e-01 3.39107990e-01 5.52192628e-01 7.19478965e-01
-4.93896216e-01 -1.64001390e-01 3.09707850e-01 6.51499480e-02
3.32469761e-01 -6.87595844e-01 -2.30755568e-01 -1.12787735e+00
-1.98029160e-01 -6.56883538e-01 6.88186169e-01 -6.69070899e-01
-1.91697133e+00 5.13677597e-01 4.92482036e-02 -1.07787097e+00
-6.67910933e-01 -6.34071648e-01 -7.62243927e-01 9.33269262e-01
-1.37473834e+00 -1.35992658e+00 4.40528430e-02 5.61522365e-01
-6.96380362e-02 -3.43473524e-01 5.51186681e-01 -1.54082194e-01
-3.77183646e-01 2.79069215e-01 4.43774343e-01 1.75454274e-01
1.09760232e-01 -1.45340967e+00 6.43567204e-01 7.86625862e-01
4.72894132e-01 3.46217632e-01 6.30418479e-01 -8.45633209e-01
-1.87809062e+00 -1.20061791e+00 7.60959387e-01 -2.16159523e-01
1.49457693e+00 -5.13623238e-01 -1.00200570e+00 8.90256345e-01
2.01836541e-01 8.55888650e-02 6.10971630e-01 5.76683223e-01
-5.44987857e-01 -1.79478064e-01 -7.68721521e-01 4.40126002e-01
7.70985901e-01 -7.32799590e-01 -1.57045648e-01 6.70751452e-01
8.42541873e-01 -1.41444206e-01 -1.08813584e+00 2.18096614e-01
3.01962078e-01 -6.06127322e-01 8.14862788e-01 -8.46655488e-01
3.86582464e-01 1.09923333e-01 -1.64500087e-01 -1.72586215e+00
-5.29706419e-01 -9.29518461e-01 -7.11321712e-01 9.16224003e-01
3.93023342e-01 -1.07007778e+00 8.06210399e-01 4.99708116e-01
5.41761816e-02 -6.14951015e-01 -6.85539007e-01 -9.33607519e-01
1.29739195e-01 -4.16405588e-01 7.21896946e-01 1.37555587e+00
-2.36227632e-01 5.31275153e-01 -5.56755185e-01 4.64344710e-01
8.76166284e-01 3.42458725e-01 7.45948195e-01 -1.63257468e+00
-4.25242305e-01 -7.20505476e-01 -6.29019260e-01 -7.20693827e-01
6.69860125e-01 -1.37393034e+00 -5.35326898e-01 -1.64565170e+00
-9.64898467e-02 -1.37978971e-01 -3.92205030e-01 2.26694629e-01
-3.01233903e-02 -3.83308172e-01 1.05234101e-01 6.68674037e-02
-2.37844974e-01 5.11224806e-01 1.24586737e+00 -2.79130191e-01
-7.63069093e-02 2.99592048e-01 -5.43708324e-01 5.15758932e-01
8.40636790e-01 -3.72712880e-01 -6.42086446e-01 -2.37426683e-01
5.45862496e-01 6.09821916e-01 4.35174286e-01 -6.03600204e-01
4.37477231e-01 -1.44817233e-01 2.19576299e-01 -5.97000480e-01
4.69373465e-02 -8.06176543e-01 3.50621194e-01 2.35363126e-01
-3.42358559e-01 6.75805628e-01 4.04169112e-02 1.09776473e+00
-1.76737428e-01 3.19966495e-01 2.01466754e-01 4.34362255e-02
-4.85500216e-01 9.89584506e-01 -5.04474305e-02 9.07212123e-02
5.92571080e-01 -2.46494599e-02 -2.31827885e-01 -1.10613859e+00
-7.97844112e-01 3.00757259e-01 -5.96831106e-02 3.86987478e-01
4.24839079e-01 -1.33491647e+00 -7.29830682e-01 8.76916647e-02
-1.96234748e-01 -4.75876778e-01 1.44986629e-01 9.28516150e-01
-2.37620130e-01 2.56957978e-01 1.31223248e-02 -6.81238174e-01
-7.24396944e-01 8.11842382e-01 2.49008477e-01 -5.60320735e-01
-5.75169742e-01 6.00831747e-01 1.09207630e-01 -5.82490742e-01
1.32911310e-01 -4.90965039e-01 -2.44327649e-01 2.15081990e-01
6.83433264e-02 4.90031242e-01 -1.60893619e-01 -5.83346248e-01
2.27687438e-03 3.44030321e-01 2.94875473e-01 2.52330750e-01
1.70707202e+00 -9.21097249e-02 -4.58761424e-01 8.55401933e-01
1.57692957e+00 -2.84394234e-01 -1.06022573e+00 -4.43096966e-01
2.36759171e-01 -2.27061242e-01 1.49711579e-01 -3.89119238e-01
-1.22343671e+00 7.35741675e-01 6.52797446e-02 1.09640634e+00
9.21694696e-01 -5.35522513e-02 5.09421468e-01 4.07470047e-01
2.84459025e-01 -5.80607235e-01 -2.31037401e-02 7.68636107e-01
8.83887112e-01 -1.32432079e+00 1.09873913e-01 -5.18219471e-02
-3.89728248e-01 1.45071816e+00 2.16122326e-02 -7.31690884e-01
1.44582105e+00 3.67019661e-02 -2.93055087e-01 -5.05274534e-01
-8.23769748e-01 -8.73054788e-02 9.67397749e-01 5.77100694e-01
3.50558430e-01 5.23586512e-01 1.60784528e-01 2.40482062e-01
-4.38995123e-01 -3.39470059e-01 5.41425407e-01 3.14908177e-01
-1.12304604e-02 -7.12818503e-01 -7.56703466e-02 9.94005620e-01
-2.61160851e-01 -1.83409631e-01 -2.44811431e-01 7.70969629e-01
-5.48884630e-01 8.79154980e-01 2.66178012e-01 -5.61453104e-01
1.37814274e-02 -4.97642569e-02 2.67708331e-01 -3.62033188e-01
-2.87451923e-01 -8.61739591e-02 2.22849205e-01 -5.68729818e-01
-5.07595778e-01 -6.83188140e-01 -7.50493288e-01 -6.90689981e-01
-2.92998195e-01 -1.69095490e-02 7.57969797e-01 1.07675171e+00
5.34500778e-01 5.93617737e-01 8.85165155e-01 -9.61601019e-01
-4.53693807e-01 -7.08828509e-01 -8.54530871e-01 2.22979710e-01
5.34094393e-01 -6.29965842e-01 -5.40094554e-01 -2.02441096e-01] | [6.7967705726623535, 2.946995258331299] |
6099f91b-130c-46df-994a-7d6877808770 | a-deep-insight-into-measuring-face-image | 2110.11111 | null | https://arxiv.org/abs/2110.11111v2 | https://arxiv.org/pdf/2110.11111v2.pdf | A Deep Insight into Measuring Face Image Utility with General and Face-specific Image Quality Metrics | Quality scores provide a measure to evaluate the utility of biometric samples for biometric recognition. Biometric recognition systems require high-quality samples to achieve optimal performance. This paper focuses on face images and the measurement of face image utility with general and face-specific image quality metrics. While face-specific metrics rely on features of aligned face images, general image quality metrics can be used on the global image and relate to human perceptions. In this paper, we analyze the gap between the general image quality metrics and the face image quality metrics. Our contribution lies in a thorough examination of how different the image quality assessment algorithms relate to the utility for the face recognition task. The results of image quality assessment algorithms are further compared with those of dedicated face image quality assessment algorithms. In total, 25 different quality metrics are evaluated on three face image databases, BioSecure, LFW, and VGGFace2 using three open-source face recognition solutions, SphereFace, ArcFace, and FaceNet. Our results reveal a clear correlation between learned image metrics to face image utility even without being specifically trained as a face utility measure. Individual handcrafted features lack general stability and perform significantly worse than general face-specific quality metrics. We additionally provide a visual insight into the image areas contributing to the quality score of a selected set of quality assessment methods. | ['Naser Damer', 'Olaf Henniger', 'Cong Chen', 'Biying Fu'] | 2021-10-21 | null | null | null | null | ['face-image-quality', 'face-image-quality-assessment'] | ['computer-vision', 'computer-vision'] | [ 1.24391474e-01 -9.53630731e-02 6.54251799e-02 -8.74166727e-01
-6.77462161e-01 -5.36130190e-01 5.39277732e-01 -3.90854061e-01
-2.52399385e-01 3.93402517e-01 1.66025668e-01 1.84389710e-01
-4.86317247e-01 -7.76190877e-01 -5.49266189e-02 -5.89783609e-01
-5.18224835e-02 3.21514726e-01 -6.23052776e-01 -1.99208930e-01
2.31975719e-01 9.76857543e-01 -1.90908146e+00 3.57376009e-01
4.65384960e-01 1.46349847e+00 -5.72866321e-01 7.12727726e-01
8.76315534e-02 7.22149536e-02 -7.95190275e-01 -8.04406047e-01
4.34439957e-01 -4.03418779e-01 -7.36014843e-01 1.81981593e-01
1.15566754e+00 -4.39552397e-01 -6.57106340e-02 9.05930400e-01
9.51397061e-01 -4.11274165e-01 7.56957471e-01 -1.49304795e+00
-6.58854365e-01 -1.78453684e-01 -1.34461656e-01 3.81943434e-01
8.03263664e-01 7.49699891e-01 9.00926590e-01 -1.26531518e+00
6.76655471e-01 1.66984618e+00 7.14141011e-01 6.85020328e-01
-1.23480034e+00 -7.09777951e-01 -6.01700842e-01 2.91806519e-01
-1.39715779e+00 -1.15259027e+00 3.18380952e-01 -5.17413497e-01
8.98526311e-01 3.07513356e-01 5.02532125e-01 7.67142296e-01
1.33208066e-01 8.68473649e-02 1.42798364e+00 -2.57272631e-01
-8.00142437e-02 1.37009129e-01 9.47392639e-03 9.84398961e-01
2.56860077e-01 5.00169516e-01 -7.86730051e-01 -2.60390460e-01
7.68055916e-01 -2.11644158e-01 -1.57417938e-01 -8.72446671e-02
-7.61343658e-01 6.40390694e-01 2.77252823e-01 5.61153352e-01
-3.55973214e-01 -9.29918140e-02 1.51432887e-01 7.17871308e-01
2.90573478e-01 3.10068995e-01 -2.02901214e-01 -1.08254962e-01
-1.17197335e+00 -3.01081967e-02 9.26146269e-01 4.68505442e-01
1.01384938e+00 2.28766337e-01 -5.36489010e-01 9.90763903e-01
4.20394331e-01 9.03716803e-01 2.65078098e-01 -1.13148379e+00
-1.92829579e-01 6.39186323e-01 -2.28793681e-01 -1.27307725e+00
-4.32473600e-01 -1.12931125e-01 -4.93867874e-01 7.34563172e-01
6.12127721e-01 1.99089661e-01 -6.95326388e-01 1.39274085e+00
1.74860314e-01 -2.27837786e-01 -2.02963814e-01 6.71903312e-01
1.27261984e+00 -4.28221859e-02 -1.65646657e-01 -1.89179569e-01
1.56616056e+00 -3.58830065e-01 -6.23662055e-01 2.24560812e-01
-1.28188804e-01 -9.62930739e-01 1.23150706e+00 3.09823424e-01
-1.24034953e+00 -8.73281240e-01 -1.03344607e+00 3.16761523e-01
-3.45833331e-01 1.26898602e-01 3.49312246e-01 1.67296779e+00
-1.65722930e+00 5.57408512e-01 -2.35385165e-01 -5.49311280e-01
8.08569610e-01 6.12621725e-01 -1.04161906e+00 -8.64776000e-02
-6.02910101e-01 1.20660973e+00 -1.74169287e-01 -2.14474007e-01
-1.05648720e+00 -7.50640750e-01 -7.63794303e-01 9.54982489e-02
-1.41091883e-01 -3.98286998e-01 8.94715130e-01 -1.14815807e+00
-1.38694894e+00 1.34754527e+00 -3.23505193e-01 1.19166523e-01
2.62346327e-01 2.99711019e-01 -6.72192752e-01 5.85305035e-01
-8.77187178e-02 6.66878879e-01 1.05717659e+00 -1.09594917e+00
-2.48049125e-01 -5.90703070e-01 -1.08623259e-01 -1.77330360e-01
-4.80448902e-01 4.47031885e-01 -2.90195972e-01 -3.87654454e-01
-3.91076028e-01 -3.83220673e-01 6.38276279e-01 5.37130773e-01
3.50512058e-01 -2.40394682e-01 7.61540234e-01 -6.39602244e-01
9.48824883e-01 -2.03516173e+00 -2.51332164e-01 5.03372729e-01
1.61745608e-01 3.40134561e-01 -7.89758086e-01 1.68966636e-01
-1.07561834e-01 2.87750810e-01 -9.67084989e-03 -8.87969807e-02
-1.92980990e-02 7.35118836e-02 5.01149535e-01 6.03748441e-01
6.25709653e-01 1.12671793e+00 -6.18434906e-01 -6.20539963e-01
1.29021674e-01 8.38923037e-01 -6.21495187e-01 1.70519099e-01
5.47807634e-01 1.61244750e-01 1.37385324e-01 1.18208647e+00
8.70118082e-01 2.23166887e-02 7.01607913e-02 -6.22258902e-01
3.87300700e-01 -2.18647167e-01 -1.02248168e+00 1.16845584e+00
-3.81009966e-01 5.43234944e-01 3.90774935e-01 -5.34986258e-01
9.90892410e-01 4.17263061e-01 6.22456312e-01 -8.62507343e-01
1.23485319e-01 1.73745558e-01 1.83596343e-01 -4.18625355e-01
1.65247135e-02 -3.45344216e-01 6.28915668e-01 6.82830453e-01
8.30550492e-01 -1.17702827e-01 1.81015626e-01 -2.01790884e-01
9.90804672e-01 -5.61561696e-02 4.43549395e-01 -5.05109012e-01
8.36465716e-01 -6.90194964e-01 2.00487405e-01 3.84254277e-01
-5.74542224e-01 8.07804763e-01 4.37173933e-01 -4.74974781e-01
-8.64121437e-01 -1.13768530e+00 -6.00430965e-01 8.37915778e-01
-3.83405864e-01 -5.87628722e-01 -1.11268795e+00 -8.25307965e-01
2.57426262e-01 -1.93792105e-01 -8.68727386e-01 -1.60777539e-01
-1.13785662e-01 -7.33528674e-01 8.84193420e-01 1.75291389e-01
5.36958933e-01 -1.19207740e+00 -4.96577024e-01 -1.73745379e-01
4.98757288e-02 -8.05334687e-01 -5.89752853e-01 -6.30929172e-01
-6.18262589e-01 -1.45604062e+00 -5.24088860e-01 -2.11493194e-01
6.17479026e-01 1.21950246e-01 1.44917035e+00 6.30560279e-01
-8.47179830e-01 1.09554422e+00 -1.53123409e-01 -2.16269404e-01
-3.82537901e-01 -4.17690396e-01 4.27545577e-01 2.85795957e-01
6.03003621e-01 -2.21435323e-01 -8.75067353e-01 7.57496357e-01
-7.94992745e-01 -9.33905840e-01 5.51932693e-01 1.08982682e+00
2.18258813e-01 -2.07143173e-01 5.82733870e-01 -3.22471410e-01
7.92649984e-01 -5.14963530e-02 -1.36321008e-01 4.07333791e-01
-9.75866258e-01 -1.68716863e-01 -9.01016071e-02 8.09585005e-02
-8.97867560e-01 -2.65483260e-01 -3.26162666e-01 -1.64138868e-01
-1.25481337e-01 1.58467352e-01 -4.63434577e-01 -7.66333103e-01
1.07061601e+00 -1.00741498e-01 8.40951025e-01 -2.37588152e-01
5.35272732e-02 6.64763153e-01 4.79460150e-01 -6.29500687e-01
6.72533989e-01 4.60979819e-01 1.48606850e-02 -1.05235231e+00
-1.46750510e-01 -2.78185397e-01 -6.63961232e-01 -6.78542554e-01
4.12278146e-01 -5.23300111e-01 -1.36378992e+00 5.12993932e-01
-7.66882539e-01 1.83586419e-01 -2.80460536e-01 9.52098221e-02
-5.22124648e-01 5.17356098e-01 -4.15609449e-01 -9.13707793e-01
-5.92317760e-01 -1.37077081e+00 1.34935188e+00 1.99265406e-01
-2.06902713e-01 -7.52602994e-01 -1.29982546e-01 4.38559026e-01
7.91684508e-01 2.33245969e-01 3.70268375e-01 -6.29289895e-02
-2.32897192e-01 -2.96254396e-01 -5.19989550e-01 6.23170197e-01
6.53280437e-01 3.11557740e-01 -1.56581807e+00 -5.13410926e-01
-3.84701081e-02 -3.57691973e-01 6.01557255e-01 2.95510083e-01
8.35226715e-01 -5.28894722e-01 1.84680432e-01 6.73844039e-01
1.38787723e+00 2.87911817e-02 1.02642155e+00 -9.95914787e-02
1.80921555e-01 9.33374286e-01 4.26574469e-01 3.68120313e-01
1.07295047e-02 8.94074380e-01 2.79202819e-01 -2.78487429e-02
-3.97739410e-01 1.60426825e-01 5.17030418e-01 1.13329194e-01
-5.63621819e-01 2.80298352e-01 -8.48164558e-01 -8.66390616e-02
-1.13704872e+00 -1.27875936e+00 4.42537218e-01 2.17289305e+00
7.39070296e-01 -4.09908026e-01 5.53776860e-01 2.52455920e-01
4.88200188e-01 -2.04953447e-01 -3.64393950e-01 -2.83567816e-01
-3.22860569e-01 7.11563647e-01 -5.15820049e-02 4.65500504e-01
-7.41649032e-01 7.26637959e-01 8.03220749e+00 7.30766356e-01
-1.10706413e+00 2.02333093e-01 8.72877240e-01 -3.27150285e-01
-1.04216121e-01 -4.88366365e-01 -7.24457622e-01 2.57247865e-01
1.17652035e+00 -1.97519720e-01 5.91756344e-01 6.33388281e-01
8.12522843e-02 1.36881411e-01 -1.23563766e+00 1.54300141e+00
4.40178692e-01 -1.13476288e+00 2.40275636e-01 4.65561748e-01
3.68296087e-01 -3.09340626e-01 4.89989191e-01 -2.53473490e-01
-2.36712545e-01 -1.75674522e+00 4.63194698e-01 9.15722728e-01
1.53887999e+00 -6.93059087e-01 7.45605707e-01 -6.55482054e-01
-1.05286193e+00 -6.95367530e-02 -3.91366810e-01 2.08906829e-01
-4.29369450e-01 5.12827873e-01 -6.58678949e-01 3.22997242e-01
9.57269490e-01 3.97040606e-01 -9.10194397e-01 6.64757550e-01
3.15822780e-01 2.49005929e-01 -1.69679254e-01 3.43704969e-01
-2.71393180e-01 -1.18802987e-01 2.15431258e-01 1.31024683e+00
4.09323692e-01 -5.16540231e-03 -4.89131987e-01 8.36990774e-01
-1.28923774e-01 3.04838181e-01 -6.38213098e-01 -1.70345500e-01
3.30583394e-01 1.66549969e+00 -3.52697641e-01 -1.27277166e-01
-6.21881247e-01 7.94654012e-01 -8.86792764e-02 1.09062262e-01
-2.41544440e-01 -1.14272945e-01 1.24558866e+00 1.97500721e-01
-1.67777866e-01 9.42608640e-02 -2.93641955e-01 -9.15413797e-01
2.50879545e-02 -1.44981778e+00 4.19529796e-01 -4.58404571e-01
-1.48563671e+00 8.60633492e-01 -1.98932052e-01 -1.01039183e+00
-2.17965230e-01 -1.12955737e+00 -5.24406552e-01 1.14621174e+00
-1.22975159e+00 -1.15153766e+00 -7.88571298e-01 7.15584219e-01
1.84248332e-02 -7.98736274e-01 1.10778928e+00 3.84968281e-01
-2.05082074e-01 1.21382093e+00 -4.20184046e-01 5.18844910e-02
8.45502198e-01 -9.69359636e-01 4.47559148e-01 4.73720461e-01
1.08886123e-01 8.27584445e-01 2.40770474e-01 -2.58321404e-01
-1.45752573e+00 -5.68839490e-01 4.38249797e-01 -8.85877967e-01
-7.36924559e-02 6.30209967e-02 -5.88885486e-01 2.39545312e-02
2.62443930e-01 2.20727965e-01 1.09493756e+00 -1.17107527e-02
-6.73709869e-01 -5.27517974e-01 -1.97753036e+00 1.48362994e-01
1.28605998e+00 -8.90331268e-01 -1.64785385e-01 -1.26308473e-02
-3.08481276e-01 2.96249509e-01 -1.35408151e+00 4.93807048e-01
1.19916141e+00 -1.34311426e+00 1.16860926e+00 -4.37659681e-01
1.51200056e-01 -2.32705489e-01 -2.64776081e-01 -1.11952126e+00
-5.17765582e-01 -3.63904148e-01 6.21701069e-02 1.21962380e+00
2.75477499e-01 -7.42243946e-01 7.89432645e-01 5.74923217e-01
4.95234758e-01 -5.60206175e-01 -1.07861257e+00 -7.87603617e-01
-4.18286845e-02 -1.26557842e-01 1.08926868e+00 7.06507266e-01
-3.33489967e-03 -1.67576335e-02 2.28421367e-03 -2.56127000e-01
8.67157459e-01 -2.83167124e-01 8.21737945e-01 -1.39265060e+00
-5.40569872e-02 -9.11579311e-01 -1.15528286e+00 2.50269696e-02
-3.61749046e-02 -8.83678377e-01 -4.38444585e-01 -1.10101640e+00
3.66404623e-01 -2.50058472e-02 -2.71155417e-01 6.02029026e-01
-1.19642802e-01 1.01114035e+00 2.92849898e-01 -2.77693830e-02
-1.94936115e-02 1.86841309e-01 1.12875652e+00 -4.07593757e-01
2.17721671e-01 -2.56633341e-01 -8.75250340e-01 3.06851208e-01
7.81129062e-01 1.80490866e-01 -1.74734935e-01 -1.51901878e-02
-3.38418745e-02 -2.52531081e-01 3.58303517e-01 -1.22717452e+00
-1.24251083e-01 6.43210635e-02 8.68243277e-01 2.37637758e-02
3.69811833e-01 -7.88439035e-01 2.87678868e-01 5.66343486e-01
2.06878200e-01 1.12519376e-01 2.39238501e-01 -1.02864958e-01
-3.46488982e-01 -1.86823308e-01 1.32600439e+00 -2.57939577e-01
-6.06769681e-01 5.89444339e-01 6.98573738e-02 -2.48052493e-01
7.06181347e-01 -6.98628545e-01 -3.77833724e-01 -4.30473268e-01
-8.66547763e-01 -2.35261410e-01 6.29202902e-01 5.03265738e-01
1.04507017e+00 -1.60587251e+00 -1.03493929e+00 6.65685475e-01
4.02015150e-01 -1.14232790e+00 7.46580437e-02 7.22138822e-01
-5.06951988e-01 3.39625984e-01 -9.93629217e-01 -7.56517768e-01
-1.81387889e+00 2.48548597e-01 6.99025929e-01 2.61347920e-01
7.05553442e-02 7.00892031e-01 8.80687162e-02 -4.97424543e-01
2.02717427e-02 2.36025617e-01 -2.16291115e-01 3.75676423e-01
1.09595013e+00 5.45384705e-01 4.19887215e-01 -1.13794315e+00
-5.70753574e-01 9.01968658e-01 2.34967664e-01 -2.04342529e-01
1.14299178e+00 -1.54040735e-02 -3.67427945e-01 -2.08422810e-01
1.31061769e+00 -2.20082819e-01 -8.76802981e-01 1.69290472e-02
-5.40926643e-02 -9.19966936e-01 -1.73373640e-01 -1.06938529e+00
-1.47093844e+00 8.60919893e-01 1.39548111e+00 -2.98441220e-02
1.41531384e+00 -1.46779776e-01 -1.44130727e-02 1.64123967e-01
5.70165455e-01 -1.01324975e+00 4.12688822e-01 1.81932703e-01
1.32799530e+00 -1.38020384e+00 7.50427879e-03 -3.16957712e-01
-4.23259526e-01 1.25870895e+00 5.94042718e-01 3.03559095e-01
8.01564395e-01 1.79309234e-01 2.61670798e-01 -4.79975253e-01
-3.01252902e-01 -3.75030100e-01 7.73327351e-01 1.13044739e+00
7.56335676e-01 -5.63987456e-02 -3.03074569e-01 2.60038197e-01
-3.64619315e-01 1.03597261e-01 -1.87722947e-02 6.07677639e-01
-4.04631704e-01 -1.31373727e+00 -5.70933163e-01 7.88732111e-01
-6.37636721e-01 9.96898934e-02 -7.04016030e-01 4.65459585e-01
2.27765143e-01 1.16341436e+00 -1.05158582e-01 -5.36643803e-01
4.13596451e-01 3.56216073e-01 9.62599277e-01 -5.00193000e-01
-8.93741369e-01 -3.66139263e-01 -2.28617918e-02 -9.76786256e-01
-6.39692068e-01 -6.12518668e-01 -5.77113092e-01 -7.29805171e-01
-2.01230004e-01 -1.16937347e-01 7.81954706e-01 4.83367234e-01
4.27512705e-01 9.51019954e-03 6.31315410e-01 -9.13992047e-01
-2.66711473e-01 -1.16063213e+00 -6.26926184e-01 5.28952420e-01
2.95708895e-01 -7.50128388e-01 -2.29898766e-01 1.04460925e-01] | [13.062430381774902, 0.8888477683067322] |
19946a13-f062-4032-be77-59ca0387bd8b | magnetic-resonance-fingerprinting-using | 1812.08155 | null | http://arxiv.org/abs/1812.08155v1 | http://arxiv.org/pdf/1812.08155v1.pdf | Magnetic Resonance Fingerprinting using Recurrent Neural Networks | Magnetic Resonance Fingerprinting (MRF) is a new approach to quantitative
magnetic resonance imaging that allows simultaneous measurement of multiple
tissue properties in a single, time-efficient acquisition. Standard MRF
reconstructs parametric maps using dictionary matching and lacks scalability
due to computational inefficiency. We propose to perform MRF map reconstruction
using a recurrent neural network, which exploits the time-dependent information
of the MRF signal evolution. We evaluate our method on multiparametric
synthetic signals and compare it to existing MRF map reconstruction approaches,
including those based on neural networks. Our method achieves state-of-the-art
estimates of T1 and T2 values. In addition, the reconstruction time is
significantly reduced compared to dictionary-matching based approaches. | ['Claudia Prieto', 'Nicolo Fuin', 'Ilkay Oksuz', 'Rene M. Botnar', 'James Clough', 'Gastao Cruz', 'Aurelien Bustin', 'Julia A. Schnabel', 'Andrew P. King'] | 2018-12-19 | null | null | null | null | ['magnetic-resonance-fingerprinting'] | ['medical'] | [ 5.63024163e-01 -1.52409732e-01 -2.32640564e-01 -3.68980914e-01
-1.08172691e+00 -3.34746629e-01 1.85860783e-01 1.56715557e-01
-4.69876319e-01 6.69038355e-01 2.41268054e-01 -9.06021819e-02
-4.65473741e-01 -5.63797414e-01 -6.29075885e-01 -6.70347810e-01
-4.32047635e-01 8.01011622e-01 4.11495179e-01 4.23619859e-02
6.94351196e-02 7.15044498e-01 -8.43048155e-01 1.07298419e-01
5.91584921e-01 9.50156629e-01 5.20275772e-01 4.15817589e-01
2.75221378e-01 1.02603209e+00 -4.67099473e-02 3.05183053e-01
1.10835157e-01 -3.59835863e-01 -7.54589021e-01 -3.93951058e-01
4.72754031e-01 -6.33013368e-01 -8.68186057e-01 9.94254768e-01
8.92646253e-01 2.24320710e-01 6.47804439e-01 -5.32692492e-01
-3.76346856e-01 8.33368897e-01 -5.03647089e-01 7.06804752e-01
3.32846902e-02 -3.29310387e-01 2.87971556e-01 -9.60390270e-01
8.71143222e-01 6.24201238e-01 1.09489799e+00 4.04185206e-01
-1.68217123e+00 -3.28622580e-01 -5.74118495e-01 -3.74969393e-02
-1.26348460e+00 -4.68416542e-01 8.02221358e-01 -5.33525467e-01
7.65155017e-01 5.72166853e-02 5.82922816e-01 7.23742366e-01
6.67669475e-01 4.94891584e-01 1.45459938e+00 -3.11852276e-01
-1.23464689e-01 -4.57074255e-01 5.36595024e-02 9.23504949e-01
-1.08299814e-01 4.78615552e-01 -4.28961754e-01 -4.25780267e-01
1.40551591e+00 -8.19619745e-02 -3.64859879e-01 -5.18796265e-01
-1.91409683e+00 7.55244315e-01 -6.48657307e-02 6.25296474e-01
-7.25544453e-01 4.97629553e-01 5.56387901e-01 1.84994891e-01
2.34835565e-01 5.28233647e-01 -1.22232707e-02 -1.54477032e-02
-1.42423356e+00 6.81870207e-02 3.41492593e-01 2.26513177e-01
2.76669264e-01 3.40647370e-01 -3.39269727e-01 1.03324068e+00
6.07723594e-02 6.52854443e-01 9.17869449e-01 -1.56103706e+00
-1.79269373e-01 -3.64777684e-01 3.68191972e-02 -1.12752497e+00
-9.70266759e-01 -4.63355899e-01 -8.41499269e-01 -1.49345070e-01
4.66070265e-01 9.18892920e-02 -6.83163762e-01 1.79555357e+00
2.87948489e-01 4.65191245e-01 -4.38445210e-01 1.12931323e+00
8.33782792e-01 1.90953836e-01 -2.48684168e-01 -5.41768610e-01
1.06002593e+00 -6.62934601e-01 -8.65543187e-01 2.32148081e-01
2.99014688e-01 -5.23186326e-01 5.83362937e-01 2.47392103e-01
-1.20704949e+00 -2.58050889e-01 -9.03819025e-01 2.29271203e-01
1.84591234e-01 -3.33378054e-02 7.98538089e-01 6.78853095e-01
-1.12220430e+00 1.02447975e+00 -1.15600848e+00 2.84271985e-01
7.44002610e-02 6.10171199e-01 -4.89554524e-01 2.94907261e-02
-1.39652359e+00 9.63241994e-01 2.67327935e-01 8.70976150e-02
-1.16178381e+00 -1.32298779e+00 -7.76615143e-01 -5.19273698e-01
1.94190949e-01 -4.61020380e-01 1.24192309e+00 -8.87566432e-02
-1.66293144e+00 7.83319652e-01 -1.73973605e-01 -6.19862378e-01
4.71126914e-01 3.50531429e-01 -5.01634777e-01 8.62681985e-01
3.76957089e-01 6.09010279e-01 7.36090958e-01 -7.64286578e-01
1.77984133e-01 -3.30536664e-01 -4.46382493e-01 -1.23678379e-01
2.67973453e-01 8.35847333e-02 8.14298317e-02 -8.40157211e-01
6.80178285e-01 -8.86800885e-01 -4.93726432e-01 1.14440685e-02
-2.34573841e-01 5.70931971e-01 2.12176323e-01 -9.96918797e-01
9.70354080e-01 -1.77548194e+00 -4.60596615e-03 4.00824755e-01
6.60153925e-01 -1.52631640e-01 -1.02495298e-01 -1.44734338e-01
-3.39868724e-01 -2.45628744e-01 -4.39051688e-01 1.38646901e-01
-2.91246772e-01 -1.13267675e-01 -1.27391025e-01 9.01281476e-01
-3.39868516e-01 1.09901619e+00 -1.09653628e+00 -5.55130005e-01
2.91476578e-01 5.92155457e-01 -2.16180250e-01 4.09757011e-02
3.79708588e-01 9.45804000e-01 -2.94596434e-01 6.42729640e-01
3.72065961e-01 -4.07292128e-01 5.88915765e-01 -6.74696147e-01
-9.88804847e-02 -3.97173241e-02 -8.07677567e-01 2.09357905e+00
-4.54260617e-01 5.31495750e-01 7.71346912e-02 -1.33573031e+00
9.12673712e-01 7.28759348e-01 1.39535284e+00 -1.06306827e+00
1.26166090e-01 6.37150645e-01 -8.45740587e-02 -3.84470850e-01
1.30903855e-01 -6.68172956e-01 2.42637157e-01 9.14368927e-01
9.71942171e-02 -5.37983440e-02 2.24691350e-02 -7.11995587e-02
1.15790653e+00 1.83014467e-01 1.07305318e-01 -6.00386322e-01
3.45491797e-01 -3.24352413e-01 3.71899009e-01 1.05010438e+00
-4.46691066e-01 6.80759013e-01 9.15163010e-02 -6.57644987e-01
-1.18219900e+00 -1.32895398e+00 -6.91849649e-01 7.11512208e-01
2.29355022e-02 9.92013216e-02 -5.29251218e-01 -2.46641651e-01
6.38786033e-02 1.62498519e-01 -9.82627332e-01 1.88775826e-03
-1.00669301e+00 -1.09729803e+00 6.13169909e-01 4.52721059e-01
1.63368672e-01 -8.93722534e-01 -6.90212309e-01 8.13949883e-01
-4.86955404e-01 -1.27665508e+00 -6.04960203e-01 1.58817679e-01
-1.36598849e+00 -7.94690073e-01 -1.18751097e+00 -5.76347113e-01
4.51437116e-01 1.91395506e-01 1.09155273e+00 -4.27639097e-01
-3.76992226e-01 3.49632055e-01 1.50134355e-01 1.57060355e-01
-5.29035091e-01 -7.75197595e-02 2.17429951e-01 2.09236257e-02
-2.66377538e-01 -1.06233597e+00 -7.89637864e-01 5.03315747e-01
-7.31690168e-01 -5.79349063e-02 5.31597972e-01 1.07127261e+00
1.38840020e+00 -3.61340940e-01 8.15101743e-01 -8.81120265e-01
4.98350114e-01 -3.36360246e-01 -4.65947807e-01 2.78643221e-01
-6.91633642e-01 2.90424615e-01 3.66741478e-01 -7.14992702e-01
-7.89212584e-01 -1.91217549e-02 1.56641528e-01 -6.31359339e-01
2.83940732e-01 6.11339808e-01 7.01438785e-01 -7.51529217e-01
7.48688579e-01 6.30309165e-01 4.80303407e-01 -2.92937249e-01
4.31829095e-02 8.16102251e-02 1.13386786e+00 -7.41625667e-01
2.40991533e-01 7.17770100e-01 4.87503827e-01 -6.02616966e-01
-4.89360124e-01 -2.82229424e-01 -7.21266985e-01 -5.43851316e-01
4.53210026e-01 -4.89447385e-01 -7.99322784e-01 2.22879231e-01
-8.19015920e-01 -3.93894315e-01 -3.21863294e-01 1.01981306e+00
-1.05764544e+00 5.94830215e-01 -9.98448014e-01 -3.63886625e-01
-6.05142891e-01 -1.32916546e+00 9.28947568e-01 -3.25274467e-01
-1.05814695e-01 -1.17347765e+00 4.51287776e-01 1.26992583e-01
8.49379122e-01 7.39689946e-01 7.86444664e-01 -2.50669360e-01
-2.50074595e-01 -9.22664478e-02 -7.69263133e-02 -7.44050145e-02
4.63548042e-02 -7.95871139e-01 -5.58219135e-01 -3.16332370e-01
3.62311065e-01 -1.78729966e-01 7.05793083e-01 1.20047081e+00
1.20925426e+00 1.44471988e-01 -2.11527422e-01 7.78436482e-01
1.36155951e+00 3.84511977e-01 5.60056746e-01 1.86886236e-01
5.40630698e-01 4.57357734e-01 2.26597205e-01 2.88908303e-01
8.91170874e-02 8.10202718e-01 -1.35889634e-01 -1.63380370e-01
-3.46789718e-01 -1.65826112e-01 -8.13207105e-02 1.12927628e+00
-2.05366626e-01 6.12160981e-01 -1.01903355e+00 5.24268806e-01
-1.63564372e+00 -9.19441938e-01 -7.12632807e-03 2.22283316e+00
9.22611058e-01 -1.88451350e-01 1.91528663e-01 8.48400965e-03
8.25535417e-01 1.62575170e-01 -7.19782293e-01 1.85314208e-01
-7.11901337e-02 4.19554889e-01 6.28082991e-01 5.97222745e-01
-8.53732944e-01 3.28419894e-01 8.45923901e+00 6.37491703e-01
-1.50519979e+00 6.32106125e-01 5.36580145e-01 -1.45324573e-01
-4.55657184e-01 -4.57064748e-01 -9.53047536e-03 2.40800694e-01
1.26634073e+00 -1.50275528e-01 7.84269869e-01 2.78623998e-01
3.50385398e-01 1.46655058e-02 -7.73172438e-01 1.16425645e+00
3.09495851e-02 -1.66842532e+00 -3.40835690e-01 -8.55207294e-02
3.57922703e-01 4.41362977e-01 6.31015673e-02 -2.18869582e-01
-1.38155550e-01 -1.06001759e+00 5.66226721e-01 8.76706779e-01
1.31954622e+00 -5.14566004e-01 5.14710546e-01 7.23303016e-03
-9.41995919e-01 3.59435141e-01 -2.94239372e-01 5.49921095e-01
4.28493500e-01 1.00307071e+00 -8.35735381e-01 4.25759941e-01
4.54012781e-01 5.87435246e-01 -1.64815068e-01 9.12289739e-01
2.40636989e-01 4.09356207e-01 -3.26155782e-01 4.40269798e-01
-8.58009309e-02 -2.31159896e-01 6.91533804e-01 9.71754432e-01
2.74463594e-01 3.39064971e-02 1.61147311e-01 9.44276392e-01
1.56696707e-01 1.99472740e-01 -5.15169442e-01 -4.62806672e-02
4.48703140e-01 1.11824358e+00 -9.56727564e-01 -2.93884039e-01
-1.45076782e-01 5.60050011e-01 1.00800842e-01 1.24740355e-01
-6.44993067e-01 -3.16712260e-01 -3.33020419e-01 5.07856488e-01
5.37557751e-02 -2.56269276e-01 5.04433885e-02 -1.14485145e+00
-1.73711970e-01 -7.02624500e-01 1.51348054e-01 -7.34459996e-01
-1.06176949e+00 6.14805818e-01 1.35696620e-01 -9.39153671e-01
-5.66971481e-01 -3.76755446e-01 -2.53351331e-02 8.35735261e-01
-1.43144333e+00 -8.42286110e-01 -1.12683075e-02 5.33075988e-01
-1.53731495e-01 -1.56006649e-01 9.62936163e-01 5.91660500e-01
-1.16220713e-01 4.17976052e-01 3.05262983e-01 1.43314078e-01
6.14428163e-01 -1.24071658e+00 1.07377678e-01 4.73026752e-01
-9.08775702e-02 6.55853450e-01 4.31381762e-01 -6.81437969e-01
-1.38228154e+00 -7.06380010e-01 5.58469772e-01 -1.86402008e-01
7.96063423e-01 7.01250955e-02 -9.83444214e-01 5.30491352e-01
-4.05294985e-01 6.68546379e-01 8.32341969e-01 -3.02120000e-01
-1.29105031e-01 -1.55312270e-01 -1.35725749e+00 5.95229752e-02
6.46968961e-01 -8.83901417e-01 -5.32545805e-01 2.80310422e-01
4.46816951e-01 -6.72143221e-01 -1.64939821e+00 4.65087563e-01
1.21516073e+00 -8.32794487e-01 1.19191980e+00 -2.38630846e-01
1.03435382e-01 -1.60097957e-01 -2.10446492e-01 -1.03199756e+00
-5.05964994e-01 -7.60264277e-01 -3.11287671e-01 2.52785146e-01
1.84238985e-01 -7.34344482e-01 8.59433949e-01 4.12720561e-01
-1.45954892e-01 -7.96391845e-01 -1.24341238e+00 -7.64407218e-01
8.81133825e-02 -3.05619389e-01 5.02965748e-01 1.19329154e+00
-8.72020870e-02 -8.74554813e-02 -6.08412921e-01 -1.28856003e-01
1.16902637e+00 2.62463510e-01 -1.46015152e-01 -1.16153347e+00
-5.95647395e-01 -2.64055073e-01 -1.59992009e-01 -6.37745559e-01
2.30212227e-01 -1.29640234e+00 -3.78112718e-02 -1.04522491e+00
4.01850194e-01 -5.50867558e-01 -7.36289740e-01 2.16717780e-01
3.11808378e-01 6.58574224e-01 -5.21097928e-02 4.76389676e-01
-1.99616745e-01 1.57870010e-01 1.76136613e+00 -1.80915192e-01
-8.85772631e-02 -4.37268257e-01 -2.58915961e-01 4.05448765e-01
5.57543635e-01 -7.08777606e-01 -3.45471889e-01 -2.21240997e-01
3.13691311e-02 1.13938248e+00 3.34578216e-01 -1.02702439e+00
2.37321407e-01 4.33821492e-02 5.76154649e-01 -5.15759826e-01
2.27199271e-01 -4.30023283e-01 5.84826767e-01 7.36092865e-01
-4.83190864e-01 1.64360717e-01 4.80489954e-02 2.56963760e-01
-2.81627029e-01 -3.96100909e-01 9.90506709e-01 -3.75895143e-01
-2.76499957e-01 6.30251527e-01 -4.73855108e-01 -1.04279350e-02
3.19916546e-01 -1.04513533e-01 1.00990050e-01 -2.77134895e-01
-1.05806911e+00 -3.91351670e-01 -5.55734448e-02 -1.38679922e-01
6.93138003e-01 -1.57409096e+00 -7.38186002e-01 4.12664674e-02
-3.42451364e-01 -5.79710782e-01 6.45355344e-01 1.46315384e+00
-6.91738188e-01 6.03673577e-01 -5.60546458e-01 -7.60537267e-01
-5.06680787e-01 4.50967252e-01 1.10610712e+00 -7.11665154e-01
-8.26466024e-01 2.64201999e-01 -1.28814746e-02 -7.62604296e-01
-3.54727477e-01 -6.95958808e-02 -1.47991970e-01 -1.02308497e-01
7.45248497e-01 4.87902820e-01 2.08393127e-01 -5.90108037e-01
-3.94553661e-01 6.98874652e-01 1.54414913e-02 -4.81858879e-01
1.35772300e+00 -2.80631203e-02 -2.50038683e-01 5.43347001e-01
1.17952573e+00 -2.57615715e-01 -9.10046220e-01 -5.72954595e-01
-1.44672930e-01 -3.19878399e-01 7.60779262e-01 -9.22499478e-01
-1.26991415e+00 7.60298908e-01 9.33759391e-01 -2.37965837e-01
9.25311804e-01 -2.22505257e-01 1.01984084e+00 2.09380969e-01
7.05215096e-01 -8.75139296e-01 1.86027698e-02 1.53986484e-01
8.40051293e-01 -1.00256574e+00 9.94127616e-02 -3.00813287e-01
-1.16673246e-01 1.28671658e+00 -1.18215457e-01 -7.68303573e-02
7.05691159e-01 5.07544279e-01 -1.28248502e-02 -3.54846418e-01
-2.09442288e-01 5.23159385e-01 3.63258243e-01 7.19705105e-01
4.71516639e-01 2.47520611e-01 -3.50998610e-01 3.81852746e-01
3.17027643e-02 4.18001980e-01 3.72802109e-01 7.08012283e-01
-2.37221077e-01 -1.00995207e+00 -3.76172751e-01 7.12583482e-01
-7.67364740e-01 -1.32185429e-01 4.28975224e-01 5.34484029e-01
-4.30240452e-01 2.06137046e-01 -1.15990035e-01 -8.19403604e-02
2.15598166e-01 -1.23883978e-01 1.09823692e+00 -1.49245486e-01
-4.07777548e-01 4.09105122e-01 -2.32449733e-02 -8.91119361e-01
-7.12294579e-01 -8.92846525e-01 -1.37576401e+00 -1.01882093e-01
3.55248600e-02 3.67756560e-02 6.81899726e-01 8.26971471e-01
1.49062708e-01 5.22122920e-01 7.56512344e-01 -8.60879242e-01
-4.24119860e-01 -7.76664793e-01 -1.04830801e+00 1.67805836e-01
4.22597080e-01 -9.15590465e-01 4.11461554e-02 -2.66515255e-01] | [13.488677024841309, -2.4192278385162354] |
d5c0d21c-daad-4943-bdf0-4905b903f8db | learning-with-latent-group-sparsity-via-heat | 2201.08326 | null | https://arxiv.org/abs/2201.08326v1 | https://arxiv.org/pdf/2201.08326v1.pdf | Learning with latent group sparsity via heat flow dynamics on networks | Group or cluster structure on explanatory variables in machine learning problems is a very general phenomenon, which has attracted broad interest from practitioners and theoreticians alike. In this work we contribute an approach to learning under such group structure, that does not require prior information on the group identities. Our paradigm is motivated by the Laplacian geometry of an underlying network with a related community structure, and proceeds by directly incorporating this into a penalty that is effectively computed via a heat flow-based local network dynamics. In fact, we demonstrate a procedure to construct such a network based on the available data. Notably, we dispense with computationally intensive pre-processing involving clustering of variables, spectral or otherwise. Our technique is underpinned by rigorous theorems that guarantee its effective performance and provide bounds on its sample complexity. In particular, in a wide range of settings, it provably suffices to run the heat flow dynamics for time that is only logarithmic in the problem dimensions. We explore in detail the interfaces of our approach with key statistical physics models in network science, such as the Gaussian Free Field and the Stochastic Block Model. We validate our approach by successful applications to real-world data from a wide array of application domains, including computer science, genetics, climatology and economics. Our work raises the possibility of applying similar diffusion-based techniques to classical learning tasks, exploiting the interplay between geometric, dynamical and stochastic structures underlying the data. | ['Soumendu Sundar Mukherjee', 'Subhroshekhar Ghosh'] | 2022-01-20 | null | null | null | null | ['stochastic-block-model'] | ['graphs'] | [ 4.53786612e-01 3.32105517e-01 -1.37320319e-02 9.88085493e-02
-2.55684137e-01 -7.33771503e-01 7.60686278e-01 5.72822571e-01
-3.86852652e-01 7.86382854e-01 -2.30760500e-01 -4.59339917e-01
-8.27821255e-01 -9.70755398e-01 -5.56175888e-01 -1.25972521e+00
-5.25016844e-01 4.69089955e-01 2.38640774e-02 -2.49134302e-01
3.15414935e-01 4.35582638e-01 -1.05067599e+00 -4.76831406e-01
9.57535148e-01 6.25596106e-01 -1.07913002e-01 7.25175917e-01
-2.72936411e-02 4.72386509e-01 -2.60451045e-02 -3.30162406e-01
2.38330796e-01 -5.95712602e-01 -9.44418430e-01 3.13576818e-01
6.16371483e-02 6.14880621e-01 -1.99377015e-01 1.09238708e+00
3.86516124e-01 3.51761609e-01 9.63456035e-01 -1.22305357e+00
-4.47105229e-01 6.72272742e-01 -8.20943058e-01 3.33804786e-01
-1.31576926e-01 -8.66750777e-02 1.21000469e+00 -4.65407223e-01
6.62902772e-01 1.12923801e+00 7.89150536e-01 1.73042744e-01
-1.93637073e+00 -2.75302857e-01 5.75938970e-02 5.18409461e-02
-1.32253301e+00 -7.85312578e-02 9.75001812e-01 -8.95449042e-01
2.78627962e-01 1.85336620e-01 5.95827401e-01 7.62615681e-01
9.67221111e-02 3.14435393e-01 1.13442647e+00 -6.99749827e-01
5.87264538e-01 -5.81507720e-02 2.20640302e-01 8.52916718e-01
3.86547178e-01 -4.94322479e-02 -4.27422851e-01 -3.72274727e-01
7.96859622e-01 1.46856427e-01 -2.80553997e-01 -7.39416838e-01
-1.04666436e+00 1.41146922e+00 4.28429484e-01 3.75620693e-01
-2.41365954e-01 1.07979394e-01 2.05817521e-01 4.51557934e-01
9.23228264e-01 1.82355046e-01 -2.91906327e-01 2.67472804e-01
-7.47538686e-01 1.35661155e-01 1.14948344e+00 5.73616087e-01
1.01946592e+00 -3.24740887e-01 4.73271191e-01 5.12899518e-01
2.49264508e-01 3.85235608e-01 -1.28361911e-01 -1.03016090e+00
1.98481515e-01 2.73195237e-01 -1.14692353e-01 -1.37350452e+00
-5.05434096e-01 -5.97427249e-01 -1.39634490e+00 1.64100081e-01
7.97811329e-01 -2.64829248e-01 -3.00105900e-01 2.04650259e+00
5.30712306e-01 3.31067264e-01 -2.52678543e-01 5.98679781e-01
-8.33982974e-02 3.64709914e-01 -1.58446282e-01 -4.19259727e-01
9.20171380e-01 -5.65428495e-01 -4.29792494e-01 2.15981022e-01
7.93273389e-01 -3.57599497e-01 6.64321423e-01 4.87472475e-01
-9.67973173e-01 -1.20469339e-01 -7.36029923e-01 2.31980160e-01
-4.17770237e-01 -4.12723720e-01 7.76437759e-01 6.43756032e-01
-1.39874113e+00 1.09727585e+00 -8.91406536e-01 -8.71103048e-01
4.03902352e-01 4.65664566e-01 -2.19976783e-01 1.01274617e-01
-9.54440236e-01 4.60769266e-01 2.35408358e-02 1.93299085e-01
-3.63614827e-01 -5.34233272e-01 -6.86972141e-01 -7.72493556e-02
5.05364120e-01 -9.20439661e-01 6.93721294e-01 -1.00321341e+00
-1.18988383e+00 6.89981341e-01 -1.07660405e-01 -4.20582414e-01
7.35366464e-01 1.84117928e-01 2.01861367e-01 2.35585168e-01
-2.04789285e-02 -2.05709729e-02 9.33066189e-01 -8.83947670e-01
-9.79808718e-02 -6.07799649e-01 8.30485821e-02 -1.31672785e-01
-4.45816547e-01 -3.35133523e-01 2.88184453e-02 -5.80205023e-01
1.56087592e-01 -1.18544638e+00 -6.83469832e-01 3.10243398e-01
-3.79837364e-01 -2.50168473e-01 6.09237969e-01 -2.70017199e-02
8.41471791e-01 -1.96784306e+00 6.05061173e-01 7.03572810e-01
7.19791234e-01 -2.62562960e-01 -6.79659247e-02 7.30214298e-01
-7.27873072e-02 4.27069843e-01 -7.42493272e-01 -1.75058752e-01
-9.29002836e-02 1.74489051e-01 -1.72969386e-01 1.10053957e+00
2.87054360e-01 7.48492181e-01 -1.10492587e+00 -3.59188408e-01
1.11481259e-02 4.40951437e-01 -8.30132723e-01 -2.59832352e-01
-6.24510534e-02 6.08604491e-01 -6.67173743e-01 1.26006261e-01
5.83737731e-01 -7.00782597e-01 3.94643754e-01 3.38822335e-01
6.04620203e-02 2.83681229e-02 -1.21014512e+00 1.55860507e+00
-4.02302235e-01 6.70500517e-01 6.14636600e-01 -1.72819781e+00
5.10639191e-01 2.40235366e-02 6.89958334e-01 1.52458251e-01
9.70466062e-02 1.24952525e-01 1.33743033e-01 -3.65296990e-01
-8.55546221e-02 -2.41020262e-01 1.04427159e-01 6.87455714e-01
-5.38086854e-02 4.01809849e-02 2.52123654e-01 4.25499171e-01
1.38449419e+00 -2.64725417e-01 2.83681989e-01 -8.94739509e-01
6.00878179e-01 -1.96051732e-01 2.23303512e-01 8.62958968e-01
-4.12630960e-02 9.90335345e-02 9.52248216e-01 -4.91382144e-02
-1.04985237e+00 -1.00699461e+00 -2.81859815e-01 9.41315353e-01
-1.51310926e-02 -3.53089541e-01 -8.13231409e-01 -5.38835049e-01
1.51839614e-01 -8.15796629e-02 -1.01508188e+00 -9.08900052e-02
-3.52337509e-01 -1.11329126e+00 2.57174104e-01 -2.41334438e-02
6.77979141e-02 -7.07012057e-01 -7.57755116e-02 2.07383469e-01
1.21099643e-01 -7.65130043e-01 -2.87694663e-01 3.42859745e-01
-1.12465322e+00 -1.29697406e+00 -6.23659551e-01 -4.84171003e-01
7.29927182e-01 1.25098959e-01 1.07350326e+00 3.74089271e-01
-4.64145243e-01 5.66235423e-01 -9.59416404e-02 -5.71033098e-02
-3.44516605e-01 4.43205595e-01 1.39165431e-01 3.00797999e-01
1.53975368e-01 -9.82593834e-01 -5.67391992e-01 6.42302260e-02
-1.14923286e+00 -1.19844578e-01 4.72658098e-01 7.93762743e-01
1.99099794e-01 3.73691559e-01 4.62754130e-01 -1.09735262e+00
4.77739006e-01 -1.03482318e+00 -7.51625717e-01 -5.87004125e-02
-5.58193505e-01 3.26928169e-01 5.81887603e-01 -3.35871607e-01
-5.00971198e-01 1.66551694e-02 4.02923405e-01 -5.13287447e-02
2.18144491e-01 8.03644955e-01 7.52571374e-02 -3.44692647e-01
4.88481700e-01 -2.03829501e-02 3.87223005e-01 -4.10007894e-01
4.65134263e-01 2.51051992e-01 2.60675222e-01 -6.03860259e-01
1.29979742e+00 8.84741426e-01 7.54649282e-01 -1.24007714e+00
-5.92886567e-01 -4.33737248e-01 -9.15200472e-01 -1.81237847e-01
7.20230460e-01 -4.81377810e-01 -9.97669518e-01 2.85409570e-01
-1.03705132e+00 -3.91896814e-01 -1.04384035e-01 5.32808065e-01
-6.34482086e-01 3.82486165e-01 -6.35580540e-01 -1.10538864e+00
1.33012012e-01 -6.26471043e-01 6.80778742e-01 -1.31960630e-01
7.83208013e-02 -1.86560929e+00 4.38661516e-01 1.16090260e-01
4.89479631e-01 5.40912628e-01 1.08569336e+00 -5.50529122e-01
-6.87048256e-01 1.45864695e-01 -1.06902935e-01 2.63452251e-02
1.32015049e-01 1.85824454e-01 -6.47614598e-01 -2.53514439e-01
1.73628449e-01 2.32063252e-02 1.19429529e+00 4.96962905e-01
1.17100477e+00 -2.79034078e-01 -4.10384983e-01 5.95737994e-01
1.60867715e+00 -5.59442699e-01 -6.51106685e-02 -1.33494481e-01
8.57661188e-01 1.09218872e+00 -2.14953739e-02 5.20284057e-01
2.13274565e-02 5.21463990e-01 2.96837211e-01 -2.23609731e-01
3.38423550e-01 -6.47995993e-02 2.04865158e-01 1.11251366e+00
-2.57560551e-01 -1.67612955e-01 -8.76570642e-01 4.56204951e-01
-1.98598516e+00 -1.09492826e+00 -6.04739547e-01 2.24256372e+00
9.26351011e-01 -3.99825536e-02 2.69259185e-01 2.10014433e-01
9.07875299e-01 1.29428014e-01 -6.48589253e-01 -1.95636898e-01
-2.51371682e-01 4.40522492e-01 7.79399395e-01 8.97700906e-01
-1.13899338e+00 6.31865561e-01 6.01241684e+00 8.84744048e-01
-9.05510664e-01 1.62171781e-01 6.67541087e-01 -1.02796704e-02
-3.55435967e-01 1.14798062e-01 -3.62613976e-01 3.34985316e-01
8.73353720e-01 -3.83765072e-01 6.72066510e-01 5.79095721e-01
5.79089642e-01 -1.45045891e-01 -1.11875737e+00 5.96623838e-01
-1.78443447e-01 -1.31028473e+00 -4.85436827e-01 6.88627183e-01
8.40500772e-01 1.85617898e-02 6.70803115e-02 -2.19509378e-01
6.80256367e-01 -1.06836414e+00 8.90975073e-02 3.79825413e-01
5.85416198e-01 -6.99457526e-01 3.43152672e-01 3.01972717e-01
-1.10827518e+00 1.17005654e-01 -2.41962716e-01 -3.91239524e-01
-7.25698173e-02 1.08043778e+00 -3.45061213e-01 5.61545312e-01
3.15291524e-01 1.06254160e+00 -4.52409327e-01 8.96593750e-01
1.10879652e-01 8.71343017e-01 -6.30110860e-01 -1.65893640e-02
1.78758070e-01 -7.98900962e-01 5.81423879e-01 1.09328294e+00
7.80536383e-02 1.40764387e-04 6.72358423e-02 8.54457021e-01
-1.66324168e-01 2.49183059e-01 -9.05954301e-01 -2.38382779e-02
2.06866190e-01 1.42997038e+00 -1.14559627e+00 -4.37835790e-03
-3.88170779e-01 7.23380446e-01 4.48437780e-01 5.26731849e-01
-4.53486025e-01 -2.68021673e-01 6.03349388e-01 2.69496083e-01
4.00369257e-01 -3.93580884e-01 -2.92335302e-01 -1.21021104e+00
-3.82866524e-02 -4.25356954e-01 2.55789906e-01 -5.84545620e-02
-1.84913075e+00 2.23486245e-01 -4.90212925e-02 -7.33668745e-01
-1.90182671e-01 -5.49388111e-01 -7.27418482e-01 8.93691421e-01
-1.48140419e+00 -6.79701626e-01 -3.84067148e-02 7.29913414e-01
3.77347022e-02 2.31203794e-01 5.50966322e-01 2.39174679e-01
-7.30869412e-01 -1.33740520e-02 6.78077996e-01 -4.83138449e-02
5.20464718e-01 -1.62788677e+00 2.36399531e-01 7.16293812e-01
1.30189568e-01 5.77428222e-01 8.67211640e-01 -4.76471007e-01
-1.45403862e+00 -1.00815320e+00 8.69880617e-01 -5.50203919e-01
1.45404136e+00 -8.43562067e-01 -9.92007673e-01 3.88682634e-01
1.01953126e-01 2.17644826e-01 6.58169627e-01 3.49451929e-01
-5.17895184e-02 2.04200745e-02 -8.91170144e-01 3.38718206e-01
1.33379221e+00 -5.76728165e-01 -9.01001245e-02 7.31390536e-01
4.16441888e-01 4.02884305e-01 -7.24040270e-01 5.32319583e-02
1.29427806e-01 -6.75543547e-01 8.11735809e-01 -7.53860474e-01
5.19982219e-01 -8.09910372e-02 1.30908698e-01 -1.19215882e+00
-2.71284372e-01 -1.14769852e+00 8.72604176e-03 1.09187591e+00
3.85317326e-01 -8.81133378e-01 7.81527340e-01 4.21789885e-01
5.73190033e-01 -9.17963803e-01 -9.07324255e-01 -8.03904891e-01
5.88455796e-01 -2.67368317e-01 -1.63787320e-01 1.23530066e+00
3.07490751e-02 4.48001325e-01 -9.54631567e-02 -4.49806191e-02
1.03785467e+00 -1.40479848e-01 6.04372740e-01 -1.85889721e+00
-5.32651424e-01 -7.40736246e-01 -4.44881141e-01 -8.98274958e-01
3.46727371e-01 -1.12587047e+00 -1.80698186e-01 -1.00576353e+00
3.33035380e-01 -7.73884475e-01 -1.73927337e-01 -1.13496678e-02
-1.44770846e-01 2.48413220e-01 -2.10565720e-02 3.94204408e-01
-4.70670789e-01 4.70552176e-01 8.41729403e-01 1.33550480e-01
-2.54808456e-01 1.89993918e-01 -6.23139977e-01 8.31331015e-01
6.86922252e-01 -5.67729771e-01 -4.60205197e-01 -5.78721911e-02
5.33397257e-01 -1.04353972e-01 6.21220052e-01 -6.29199386e-01
4.87087727e-01 -2.29407087e-01 -6.63940236e-02 6.10660017e-02
2.88195107e-02 -6.83720112e-01 -1.36472180e-01 5.49791574e-01
-6.53653502e-01 4.39928845e-02 -3.17087442e-01 1.05102277e+00
2.23089978e-01 -2.52265453e-01 8.07457089e-01 1.51122048e-01
7.68871754e-02 5.06304801e-01 -4.05710965e-01 3.76727879e-01
9.56940114e-01 2.09441960e-01 -9.95529369e-02 -5.05704403e-01
-8.71446192e-01 1.40634403e-01 4.90086555e-01 -1.71614021e-01
1.23836122e-01 -1.03693354e+00 -8.72845411e-01 7.31198192e-02
-4.00678873e-01 -2.45710477e-01 8.03719684e-02 1.34484887e+00
-3.53275299e-01 2.70795673e-01 3.42442214e-01 -7.50740409e-01
-1.06385255e+00 6.75027907e-01 1.92828029e-01 -3.92596841e-01
-5.06066084e-01 5.31001270e-01 3.36442590e-01 -1.90987915e-01
-4.10012603e-02 -8.77091661e-02 6.87793642e-02 2.24178270e-01
2.15991497e-01 5.07686138e-01 -2.04772413e-01 -4.07687753e-01
-2.25698501e-01 6.77132249e-01 1.93595752e-01 -4.35255229e-01
1.49731660e+00 -3.81767243e-01 -4.65569496e-01 6.94056869e-01
1.34313381e+00 2.48296380e-01 -1.27599072e+00 -6.14443898e-01
3.44944000e-01 -1.08352140e-01 -7.16168806e-03 -7.63619766e-02
-1.18783820e+00 9.12690520e-01 9.63219851e-02 1.09029818e+00
9.96420145e-01 4.25433904e-01 2.18369156e-01 4.26790565e-01
2.09326908e-01 -8.71862352e-01 -1.00321032e-01 4.02577430e-01
5.02167284e-01 -1.22977531e+00 -2.11579520e-02 -5.47126710e-01
7.81264603e-02 1.09877884e+00 -8.95051733e-02 -4.89865929e-01
1.15387750e+00 8.16715062e-02 -3.67498726e-01 -2.65642345e-01
-7.73707867e-01 -3.38300377e-01 1.42244071e-01 4.53083336e-01
3.86948526e-01 -6.41703829e-02 -5.73240519e-01 3.27214636e-02
-1.57195851e-01 -3.34250808e-01 6.42586350e-01 6.78300202e-01
-3.72318059e-01 -1.32630134e+00 -1.77687883e-01 4.99604315e-01
-5.90397716e-01 -2.80458689e-01 -5.72456479e-01 8.21639419e-01
-7.63753280e-02 9.49588895e-01 9.02441442e-02 1.39313534e-01
-4.36954170e-01 -8.20575804e-02 3.36877644e-01 -5.69650054e-01
-2.38315791e-01 1.76169455e-01 -4.20300186e-01 -2.57504106e-01
-9.26773906e-01 -1.02672994e+00 -8.62334907e-01 -8.07779372e-01
-3.71299565e-01 3.10398161e-01 6.55158341e-01 1.20253956e+00
2.51306683e-01 2.37828821e-01 8.28503489e-01 -8.26731682e-01
-4.46622103e-01 -8.09726954e-01 -9.19123411e-01 1.93740487e-01
6.47646010e-01 -6.18348241e-01 -1.03631639e+00 1.13338925e-01] | [6.950779438018799, 5.138308048248291] |
7bacaedd-8019-4531-a25c-eac4d08d36ee | acquiring-background-knowledge-to-improve | 1709.05467 | null | http://arxiv.org/abs/1709.05467v1 | http://arxiv.org/pdf/1709.05467v1.pdf | Acquiring Background Knowledge to Improve Moral Value Prediction | In this paper, we address the problem of detecting expressions of moral
values in tweets using content analysis. This is a particularly challenging
problem because moral values are often only implicitly signaled in language,
and tweets contain little contextual information due to length constraints. To
address these obstacles, we present a novel approach to automatically acquire
background knowledge from an external knowledge base to enrich input texts and
thus improve moral value prediction. By combining basic text features with
background knowledge, our overall context-aware framework achieves performance
comparable to a single human annotator. To the best of our knowledge, this is
the first attempt to incorporate background knowledge for the prediction of
implicit psychological variables in the area of computational social science. | ['Marlon Mooijman', 'Heng Ji', 'Morteza Dehghani', 'Joe Hoover', 'Ying Lin'] | 2017-09-16 | null | null | null | null | ['value-prediction'] | ['computer-code'] | [ 4.12270635e-01 5.40238202e-01 -6.84257567e-01 -5.82189560e-01
-1.25161380e-01 -2.13762388e-01 4.48721468e-01 8.52202237e-01
-9.58300829e-01 9.90430057e-01 3.82180333e-01 -7.60674104e-02
6.99402243e-02 -1.01960063e+00 -1.59591958e-01 -2.89029926e-01
2.68188536e-01 4.31234896e-01 8.08558017e-02 -5.48610091e-01
6.62913203e-01 1.63731426e-01 -1.46169269e+00 3.29209477e-01
9.43031192e-01 6.47663534e-01 -3.97027463e-01 4.63013530e-01
-2.16538906e-01 1.24323845e+00 -4.59740132e-01 -1.09945750e+00
-2.63617694e-01 -4.56149638e-01 -1.14769530e+00 -4.91341978e-01
2.58141845e-01 -4.77049828e-01 1.59945086e-01 1.00681567e+00
3.90525490e-01 2.21624985e-01 7.05464303e-01 -1.19699299e+00
-6.99723601e-01 9.65042114e-01 -2.87516773e-01 1.88705668e-01
6.46703660e-01 -2.70732015e-01 1.32777202e+00 -4.10820454e-01
7.74556339e-01 1.16898537e+00 7.46342063e-01 7.07917452e-01
-1.28989136e+00 -5.53327203e-01 1.00068145e-01 3.76238406e-01
-9.78675485e-01 -3.65835816e-01 1.03643715e+00 -6.62063956e-01
8.02277923e-01 2.54109859e-01 4.73955780e-01 1.32052565e+00
-5.18691540e-01 6.24361992e-01 1.33796489e+00 -5.84706485e-01
3.22312653e-01 5.58695853e-01 5.73903978e-01 5.56154549e-01
3.41891468e-01 -3.55438352e-01 -5.94621301e-01 -5.53337514e-01
1.48174182e-01 -4.89219069e-01 2.08056837e-01 1.87987406e-02
-1.02295208e+00 1.04982650e+00 2.50302977e-03 4.83160198e-01
-4.02334392e-01 1.43983439e-01 5.93408227e-01 2.47570917e-01
7.77025640e-01 7.42879272e-01 -3.78685743e-01 -6.39046848e-01
-8.61450076e-01 4.04450834e-01 1.19986784e+00 4.96417224e-01
7.46119022e-01 -4.13270652e-01 -1.48319528e-01 8.03679407e-01
-6.35376200e-03 5.08008063e-01 2.44244620e-01 -1.23790669e+00
1.77154720e-01 5.42909563e-01 3.58361900e-01 -1.53159785e+00
-5.48630118e-01 -1.35409519e-01 -1.41303912e-01 -1.64436176e-01
7.64021277e-01 -3.54286432e-01 4.07976434e-02 1.80226064e+00
5.20946205e-01 5.84839806e-02 3.05487178e-02 7.53356636e-01
1.02852499e+00 2.16806307e-01 5.52119493e-01 -3.37869704e-01
1.41007507e+00 -4.19842869e-01 -1.04560184e+00 -2.10819319e-01
1.08046269e+00 -5.78514695e-01 7.65432060e-01 2.08521724e-01
-9.11962450e-01 6.91847876e-02 -4.90772843e-01 -8.80096033e-02
-4.42228019e-01 -1.65128455e-01 9.59696591e-01 7.74041116e-01
-6.25009120e-01 6.35770619e-01 -2.26752266e-01 -6.49629235e-01
4.51115221e-01 2.90455312e-01 -2.92374700e-01 2.49060258e-01
-1.63102126e+00 1.14451826e+00 5.30075073e-01 -4.41190451e-01
1.41763166e-01 -4.55706507e-01 -7.00752556e-01 -1.89317986e-01
6.71760678e-01 -3.43137771e-01 1.14993823e+00 -1.41643703e+00
-1.44368136e+00 1.37365484e+00 -2.39233375e-01 -3.82041991e-01
4.97792661e-01 -2.66121089e-01 -1.48527816e-01 9.59164724e-02
2.65495867e-01 3.25671822e-01 3.59919399e-01 -1.00445580e+00
-4.32420909e-01 -2.09542975e-01 3.33331287e-01 5.22653162e-02
-9.73215282e-01 6.42088473e-01 1.47099644e-01 -6.08391821e-01
-6.31328285e-01 -8.13171744e-01 -1.39593363e-01 -3.95700336e-01
1.76773980e-01 -5.00616372e-01 3.52565289e-01 -6.82591736e-01
1.42126882e+00 -1.90054584e+00 -5.96991703e-02 3.16361248e-01
2.20684826e-01 3.02085847e-01 3.02995145e-01 3.68643194e-01
3.70607466e-01 3.02072048e-01 -2.12971512e-02 6.16866387e-02
8.83219317e-02 4.18102354e-01 -3.02880462e-02 3.20636690e-01
-6.19277917e-03 9.05900061e-01 -1.45316792e+00 -1.08236051e+00
-1.87477276e-01 1.27592966e-01 -8.70420694e-01 8.00574385e-03
-1.73071846e-01 2.45543301e-01 -6.05455279e-01 3.74589086e-01
4.21425015e-01 -5.15649058e-02 4.12119478e-01 2.36757576e-01
-2.06062689e-01 4.09711510e-01 -7.70562828e-01 1.05208063e+00
-2.05341429e-01 7.55560935e-01 6.96323514e-02 -1.30640435e+00
1.07351613e+00 2.01071039e-01 6.32465839e-01 -6.97452128e-01
6.10176682e-01 1.62390009e-01 -1.96777254e-01 -6.84320748e-01
8.05611074e-01 -5.92391253e-01 -3.20718795e-01 5.22728443e-01
-2.91658103e-01 1.00096755e-01 2.90072441e-01 1.41917497e-01
9.23709571e-01 1.11272493e-02 5.20569086e-01 -2.58023292e-01
7.10448444e-01 1.61602125e-01 8.80826116e-01 7.33826101e-01
-5.79721987e-01 -1.08653098e-01 1.02497756e+00 -4.72932667e-01
-9.11077619e-01 -4.03287202e-01 -2.88246214e-01 1.72642779e+00
1.08579621e-02 -5.95493019e-01 -8.04346263e-01 -6.73402011e-01
1.53771877e-01 8.88653815e-01 -7.17840314e-01 7.70706264e-03
-5.45018733e-01 -6.57470286e-01 7.20061183e-01 3.91705275e-01
2.47351557e-01 -1.10983336e+00 -5.99388540e-01 2.54542321e-01
-5.93516588e-01 -1.56458354e+00 1.73333138e-01 -1.15287729e-01
-7.61134028e-01 -1.02836883e+00 -2.03932837e-01 -5.78334153e-01
5.13052106e-01 -3.11622173e-01 1.18001795e+00 4.81888384e-01
5.93187623e-02 3.01957548e-01 -6.06965959e-01 -6.50721669e-01
-5.18804669e-01 1.60371736e-01 -6.11389130e-02 -7.21540153e-02
1.06162226e+00 -3.80630583e-01 -1.98199674e-01 6.64666668e-02
-6.43552542e-01 8.59489813e-02 1.08273394e-01 7.45197356e-01
-1.06990427e-01 -1.55689821e-01 6.67720497e-01 -1.23452377e+00
8.62446785e-01 -7.05462515e-01 -2.09040955e-01 -1.06759757e-01
-6.33499622e-01 -7.29702413e-02 5.19538462e-01 -3.12687516e-01
-1.10917103e+00 -1.26376554e-01 -2.06513137e-01 4.81428951e-01
-4.27045017e-01 7.18736291e-01 2.73432016e-01 5.52370399e-02
7.57540047e-01 -4.32012677e-01 -1.62303187e-02 -4.04480487e-01
2.21706912e-01 9.64793324e-01 3.77644777e-01 -9.27585542e-01
4.15136248e-01 3.32243919e-01 1.77145600e-01 -1.03992915e+00
-1.31079519e+00 -5.84062636e-01 -8.83750975e-01 -5.49754024e-01
8.72315228e-01 -7.06092834e-01 -1.02914190e+00 1.47067159e-01
-1.02881837e+00 -3.27326775e-01 3.39616984e-02 5.95571458e-01
-5.14889777e-01 5.98228812e-01 -6.30381763e-01 -9.92568731e-01
-1.35885194e-01 -4.06472206e-01 4.63325322e-01 6.88804314e-02
-8.49084139e-01 -1.17009091e+00 1.86571136e-01 9.14915860e-01
2.20932603e-01 5.59979498e-01 6.28844619e-01 -8.33074152e-01
3.26063126e-01 -2.44056419e-01 -1.56409889e-01 1.73119321e-01
-1.77398503e-01 3.98582220e-03 -9.19112504e-01 2.03159258e-01
-3.34447891e-01 -7.42793024e-01 6.64643645e-01 -1.63911924e-01
9.27908301e-01 -4.62343395e-01 -1.08112305e-01 1.64248142e-02
1.04752016e+00 -2.18615144e-01 4.65766132e-01 8.63217473e-01
5.08822083e-01 1.06559622e+00 8.83708358e-01 9.23544347e-01
6.80507362e-01 6.41949058e-01 1.21042028e-01 1.17622748e-01
5.49382746e-01 7.50036761e-02 2.90228575e-01 4.50590998e-01
-5.03624320e-01 1.69340983e-01 -1.23994160e+00 6.40779972e-01
-2.11916590e+00 -1.50351393e+00 -4.52740043e-01 1.67526317e+00
1.31511092e+00 1.15607113e-01 3.05183649e-01 1.77549213e-01
8.66473198e-01 -1.69581957e-02 -1.57655459e-02 -1.07191551e+00
7.48528093e-02 4.71978262e-02 4.85768855e-01 5.32261610e-01
-1.20431900e+00 1.11651945e+00 6.65939808e+00 5.17124295e-01
-7.01768517e-01 1.45785242e-01 3.81513655e-01 -1.88616827e-01
-3.98219466e-01 3.44527327e-02 -6.25512838e-01 4.36365843e-01
9.66218174e-01 -2.00349525e-01 2.77204752e-01 9.85581100e-01
2.97930390e-01 -1.89682022e-01 -1.02444673e+00 6.86487496e-01
1.18807308e-01 -7.77027488e-01 -5.24596572e-01 1.27827138e-01
6.40497863e-01 -2.25084469e-01 -2.18534335e-01 3.69273633e-01
3.37783694e-01 -8.81102979e-01 5.89684010e-01 6.02005780e-01
3.78077298e-01 -8.23962688e-01 1.05463696e+00 4.33798045e-01
-4.25731361e-01 -5.96207678e-02 -3.32851231e-01 -7.71755397e-01
5.14948852e-02 7.93125510e-01 -8.13949347e-01 2.95882057e-02
5.60331225e-01 7.11793303e-01 -4.79829252e-01 4.98691738e-01
-2.03127384e-01 8.51887107e-01 -1.77779138e-01 -5.40731907e-01
2.83285052e-01 9.92821455e-02 4.56162542e-01 1.69084346e+00
-1.25227213e-01 3.74307483e-01 1.31543234e-01 8.19397748e-01
-3.17500114e-01 6.97114587e-01 -6.99733496e-01 -2.62916952e-01
2.29876488e-01 1.37503386e+00 -6.58532739e-01 -4.05710667e-01
-7.25947738e-01 6.76292658e-01 6.50777280e-01 1.74360909e-02
-7.27992654e-01 -3.78576010e-01 5.73992670e-01 1.91698760e-01
-1.27647698e-01 -4.07914147e-02 -4.85081881e-01 -1.23179889e+00
-2.77791768e-01 -7.09809065e-01 5.33436775e-01 -3.83283347e-01
-1.35458016e+00 -1.33926526e-01 7.64003173e-02 -4.80703771e-01
-1.46487311e-01 -7.46587396e-01 -2.55919904e-01 4.78547513e-01
-1.57423103e+00 -9.33636546e-01 -4.16495293e-01 3.64169776e-01
-2.45932147e-01 3.29028443e-03 8.19834769e-01 1.92294493e-01
-3.32195640e-01 5.94083786e-01 4.72863801e-02 4.39824313e-01
1.03423846e+00 -1.14249361e+00 -1.53218284e-01 2.84065694e-01
-4.27602559e-01 5.91594994e-01 1.00446510e+00 -7.87398100e-01
-8.74346852e-01 -7.61699319e-01 1.44663060e+00 -7.40244508e-01
9.18776453e-01 -3.20934147e-01 -8.11206877e-01 4.91400301e-01
4.59655449e-02 -3.71857017e-01 1.51537704e+00 6.76684260e-01
-3.25273782e-01 2.18015790e-01 -1.13872492e+00 6.15203559e-01
9.22049344e-01 -5.40899277e-01 -1.07554984e+00 -3.59921181e-03
3.63817692e-01 -1.01155534e-01 -1.15901029e+00 2.00053334e-01
5.95722914e-01 -8.00181985e-01 7.67192543e-01 -6.93788409e-01
7.63102114e-01 2.41734624e-01 1.54635869e-02 -8.27141225e-01
-3.79729539e-01 -3.95428330e-01 6.99853599e-02 1.67730784e+00
2.77713925e-01 -4.52784836e-01 9.30165827e-01 1.26312149e+00
4.22164738e-01 -3.91105711e-01 -6.22394621e-01 -4.12420750e-01
3.80054414e-01 -4.83211786e-01 5.10266185e-01 1.47810590e+00
6.97872400e-01 3.36700410e-01 -6.88283741e-01 -4.27406698e-01
6.12829566e-01 2.70174555e-02 7.70165443e-01 -1.68277609e+00
-4.16073687e-02 -5.00174761e-01 -4.40233409e-01 -2.56257981e-01
8.63644779e-01 -9.34359908e-01 -1.17317557e-01 -1.43386686e+00
5.04854858e-01 -4.53247041e-01 -2.61860073e-01 5.56901217e-01
-3.63893032e-01 3.21428329e-01 9.87019092e-02 -9.73218009e-02
-8.18509102e-01 2.79691637e-01 8.98555875e-01 2.21277207e-01
-1.66657388e-01 -4.50026900e-01 -1.06151962e+00 1.01112568e+00
1.36447060e+00 -6.99787796e-01 -7.99259394e-02 -4.42508794e-02
9.99534249e-01 -2.62652308e-01 3.99948955e-01 -4.26697761e-01
-2.77206358e-02 -9.42433655e-01 7.85097629e-02 -1.91344440e-01
5.33669032e-02 -6.93607211e-01 -3.53676856e-01 2.54929751e-01
-7.18264282e-01 -4.68225390e-01 5.73643763e-03 3.54361981e-01
1.70402415e-02 -7.37261415e-01 7.04636812e-01 -4.02062058e-01
-7.24428236e-01 -3.55668396e-01 -7.20771134e-01 4.15385813e-01
1.17527795e+00 -1.20425902e-01 -4.90202874e-01 -4.33775067e-01
-4.35266167e-01 1.73819616e-01 6.94104970e-01 2.58454621e-01
3.53417873e-01 -1.29854465e+00 -8.13533723e-01 -3.96995276e-01
1.98198691e-01 -5.70770621e-01 -1.72700524e-01 9.70888793e-01
-3.86264354e-01 2.42863148e-01 -3.27432483e-01 -8.29709396e-02
-1.60044873e+00 6.89233005e-01 -1.40954688e-01 -7.92610794e-02
-4.72859085e-01 5.27508914e-01 -2.14078411e-01 -3.79594177e-01
2.02276576e-02 3.40778738e-01 -9.28737044e-01 6.72261000e-01
4.31350708e-01 5.83476663e-01 -3.53960425e-01 -1.03851545e+00
-4.88201648e-01 3.31934273e-01 2.23228738e-01 -1.12283692e-01
1.54342914e+00 -1.95789248e-01 -6.35520756e-01 5.25494099e-01
1.14514625e+00 3.10950905e-01 -3.55779350e-01 -3.94421726e-01
5.23420155e-01 -7.02873111e-01 8.91151428e-02 -5.91422617e-01
-5.50525248e-01 6.15458965e-01 -1.64180681e-01 2.01706156e-01
7.87773788e-01 -1.83545500e-02 6.73290431e-01 7.42519975e-01
1.70084521e-01 -2.09051299e+00 2.22493559e-01 7.17487097e-01
3.72962445e-01 -1.37306631e+00 1.72467306e-01 -2.97522604e-01
-1.03938580e+00 1.16457582e+00 7.63337612e-01 1.93210602e-01
4.25172329e-01 8.94788746e-03 1.20731644e-01 -1.97727501e-01
-7.72976995e-01 -4.75893646e-01 5.36925644e-02 4.68411922e-01
8.91484201e-01 4.36279289e-02 -1.02001667e+00 7.14598715e-01
-4.35890466e-01 2.73798555e-01 9.17040169e-01 1.04875028e+00
-6.67857289e-01 -1.15491009e+00 -4.94334459e-01 4.27220851e-01
-1.23710096e+00 -1.73709393e-01 -8.99328053e-01 5.06711662e-01
2.34012604e-01 1.14277077e+00 -1.84435248e-01 -2.36425892e-01
1.09216623e-01 2.59271324e-01 1.36226192e-01 -5.05191743e-01
-7.56990075e-01 -2.50996023e-01 8.07326019e-01 -3.53469193e-01
-9.80053663e-01 -8.59129548e-01 -1.31993067e+00 -8.66991222e-01
-1.21440664e-01 2.91320413e-01 3.64102483e-01 9.47254539e-01
8.88203923e-03 8.24857652e-02 4.51877147e-01 -3.51201296e-01
-2.45240867e-01 -6.14834070e-01 -4.32576686e-01 9.15396631e-01
7.38182068e-02 -6.91540718e-01 -2.73483574e-01 4.51499037e-02] | [9.221970558166504, 10.080038070678711] |
dc7a1e62-e93a-4bc1-b718-1d4c8f93dff0 | structurenet-hierarchical-graph-networks-for | 1908.00575 | null | https://arxiv.org/abs/1908.00575v1 | https://arxiv.org/pdf/1908.00575v1.pdf | StructureNet: Hierarchical Graph Networks for 3D Shape Generation | The ability to generate novel, diverse, and realistic 3D shapes along with associated part semantics and structure is central to many applications requiring high-quality 3D assets or large volumes of realistic training data. A key challenge towards this goal is how to accommodate diverse shape variations, including both continuous deformations of parts as well as structural or discrete alterations which add to, remove from, or modify the shape constituents and compositional structure. Such object structure can typically be organized into a hierarchy of constituent object parts and relationships, represented as a hierarchy of n-ary graphs. We introduce StructureNet, a hierarchical graph network which (i) can directly encode shapes represented as such n-ary graphs; (ii) can be robustly trained on large and complex shape families; and (iii) can be used to generate a great diversity of realistic structured shape geometries. Technically, we accomplish this by drawing inspiration from recent advances in graph neural networks to propose an order-invariant encoding of n-ary graphs, considering jointly both part geometry and inter-part relations during network training. We extensively evaluate the quality of the learned latent spaces for various shape families and show significant advantages over baseline and competing methods. The learned latent spaces enable several structure-aware geometry processing applications, including shape generation and interpolation, shape editing, or shape structure discovery directly from un-annotated images, point clouds, or partial scans. | ['Paul Guerrero', 'Peter Wonka', 'Li Yi', 'Leonidas J. Guibas', 'Niloy Mitra', 'Kaichun Mo', 'Hao Su'] | 2019-08-01 | null | null | null | null | ['3d-shape-generation'] | ['computer-vision'] | [ 3.72402787e-01 3.86134446e-01 2.79245287e-01 -3.89200956e-01
-4.25815523e-01 -9.08796310e-01 6.88586593e-01 2.79099226e-01
2.98136741e-01 4.87106711e-01 3.68237615e-01 -9.99115035e-02
-2.41407558e-01 -1.31552708e+00 -9.06405985e-01 -4.56729144e-01
-1.04529686e-01 1.08093703e+00 2.24388123e-01 -3.54069680e-01
5.07641509e-02 1.47772133e+00 -1.42151189e+00 9.41440538e-02
7.80276060e-01 8.98241758e-01 -9.46483389e-02 3.80613983e-01
-3.60417843e-01 -1.78481698e-01 -2.39207000e-01 -3.70239049e-01
4.90559459e-01 -4.55746911e-02 -5.99575460e-01 4.32914764e-01
5.95840573e-01 -8.41954127e-02 -1.06702380e-01 7.39854276e-01
4.77102637e-01 1.54772028e-01 9.14468169e-01 -8.47137570e-01
-9.03883100e-01 4.18560565e-01 -2.42595166e-01 -5.08277237e-01
1.53999239e-01 2.79111534e-01 7.93665290e-01 -1.09680605e+00
9.38966274e-01 1.41419351e+00 5.68477750e-01 4.94390666e-01
-1.59106231e+00 -2.18111187e-01 9.84928235e-02 -4.99802470e-01
-1.34706247e+00 -3.19934696e-01 1.20283425e+00 -5.06987691e-01
8.79052520e-01 2.60372043e-01 9.27746654e-01 7.92559862e-01
1.84517294e-01 3.70251536e-01 4.78582948e-01 -1.99888200e-01
3.37738663e-01 -4.60046142e-01 -4.59920406e-01 7.76268423e-01
2.91045517e-01 -1.63714379e-01 -9.22342688e-02 -3.66552025e-01
1.34201670e+00 7.92993680e-02 -1.93631798e-01 -9.77171957e-01
-1.30098283e+00 7.53252447e-01 8.00642550e-01 7.62054101e-02
-4.64494824e-01 3.36259037e-01 9.18034911e-02 -2.15117987e-02
4.87722099e-01 7.53046155e-01 -4.37139094e-01 3.60767156e-01
-6.29650295e-01 5.61550975e-01 5.48156619e-01 1.26887119e+00
8.90652061e-01 5.29908299e-01 -1.67679325e-01 8.96453857e-01
1.99797988e-01 5.60222745e-01 5.85486321e-03 -9.80113149e-01
4.09463614e-01 1.05457878e+00 -9.06429440e-02 -1.02662098e+00
-3.56663018e-01 -3.77888709e-01 -1.25904477e+00 2.16447741e-01
1.96311716e-03 2.57329792e-01 -1.30149353e+00 1.62143958e+00
5.73073030e-01 -1.15549229e-01 -3.18305731e-01 4.47762489e-01
1.03568435e+00 4.30449426e-01 -1.79735631e-01 6.71451315e-02
1.05438435e+00 -4.14147884e-01 -2.44442895e-01 -1.17855808e-02
3.06758761e-01 -6.45185351e-01 9.13983762e-01 2.31506452e-02
-1.60036194e+00 -4.80559796e-01 -7.84649670e-01 -1.51021406e-01
-3.49700391e-01 -3.55516374e-01 6.79681361e-01 3.09361011e-01
-1.25073814e+00 7.35879302e-01 -6.96397662e-01 1.43650338e-01
8.07755828e-01 4.82015073e-01 -5.59974670e-01 -1.87690139e-01
-7.67618358e-01 2.60594547e-01 2.56000370e-01 2.26907134e-01
-6.04239941e-01 -8.63460720e-01 -1.32904160e+00 7.92313963e-02
4.10687804e-01 -1.27669084e+00 7.95326710e-01 -4.32783455e-01
-1.30401373e+00 8.34744751e-01 1.35329276e-01 5.47494143e-02
2.86549181e-01 2.61218965e-01 1.77625827e-02 -6.46504387e-02
-1.15481138e-01 8.32014620e-01 9.47552443e-01 -1.43073177e+00
2.53093392e-01 -5.78032434e-01 -2.21827556e-03 2.16340154e-01
8.78520533e-02 -4.47165459e-01 -2.90256023e-01 -9.65269268e-01
6.46836460e-01 -8.39796305e-01 -5.36048472e-01 4.12308127e-01
-5.57223320e-01 -4.51907441e-02 9.41504359e-01 -4.70180392e-01
5.34007847e-01 -1.90589583e+00 4.43373293e-01 5.18551052e-01
3.92997831e-01 1.65898561e-01 -3.38516802e-01 5.18603146e-01
-9.87234339e-02 4.03560787e-01 -6.23700202e-01 -2.21449569e-01
1.39317170e-01 4.70081180e-01 -1.70089692e-01 7.40467459e-02
4.82120812e-01 1.43521976e+00 -7.91020215e-01 -2.15055317e-01
3.02671432e-01 6.82579458e-01 -8.79977524e-01 1.77483380e-01
-6.31797552e-01 6.60228789e-01 -5.11384964e-01 7.64240324e-01
6.60798371e-01 -3.04608434e-01 -4.49976586e-02 -5.11053383e-01
1.43906653e-01 1.44226655e-01 -1.24312389e+00 1.67264295e+00
-3.95240635e-01 9.39180255e-02 2.08517864e-01 -8.52693200e-01
1.25615144e+00 1.52334392e-01 6.77672088e-01 -3.48420531e-01
-1.36839123e-02 2.21481875e-01 -1.42244339e-01 -5.77133633e-02
4.31017190e-01 -2.43685812e-01 -1.64030958e-03 4.16119605e-01
-6.32285103e-02 -1.04289806e+00 -8.24805573e-02 1.50691971e-01
9.85351562e-01 1.12975709e-01 1.93947196e-01 -1.38957769e-01
3.41957390e-01 -3.73066306e-01 3.87569129e-01 2.77241915e-01
5.09849966e-01 8.35671127e-01 3.00733000e-01 -6.76719606e-01
-1.34685600e+00 -1.51378214e+00 -4.46814336e-02 5.45881331e-01
-7.08098784e-02 -5.11068441e-02 -3.58682871e-01 -5.16756773e-01
2.79262036e-01 6.24571025e-01 -5.36969423e-01 -2.15906173e-01
-9.35156465e-01 -2.60310799e-01 1.97894961e-01 6.47428215e-01
1.26381010e-01 -1.40449727e+00 -1.75955519e-01 3.07128638e-01
2.72245377e-01 -1.00524688e+00 -7.40280747e-01 -6.16319105e-02
-1.13540435e+00 -1.01108599e+00 -5.05657434e-01 -7.76930392e-01
1.08328283e+00 2.68266369e-02 1.34861672e+00 3.06023985e-01
-3.66238445e-01 5.49851358e-01 -1.31434619e-01 -2.04854906e-01
-6.51241958e-01 -5.89783974e-02 -1.23582833e-01 4.75715734e-02
-6.01586044e-01 -1.22297263e+00 -4.73615587e-01 1.77950665e-01
-1.31816268e+00 2.63768852e-01 4.61481780e-01 6.51301563e-01
1.14237177e+00 -1.37077838e-01 5.24954498e-01 -8.87553751e-01
4.77561474e-01 -6.20960779e-02 -6.58060193e-01 2.29023159e-01
-2.49612927e-01 3.80647928e-01 5.56053579e-01 -3.47983479e-01
-8.03555608e-01 1.73427537e-01 -2.54345924e-01 -6.12234056e-01
-1.87570959e-01 4.03657824e-01 -5.58604717e-01 -2.05922842e-01
6.36040986e-01 1.58983290e-01 -1.16178570e-02 -3.77901584e-01
9.63277757e-01 7.96199068e-02 5.48304200e-01 -1.03836358e+00
1.10214913e+00 5.26548922e-01 4.86775219e-01 -9.50743079e-01
-3.51800799e-01 4.03718315e-02 -1.20859933e+00 2.79575977e-02
7.34164834e-01 -4.80940074e-01 -2.84856498e-01 4.43902194e-01
-1.24089718e+00 -5.44923306e-01 -7.57146418e-01 -6.86428919e-02
-6.33839130e-01 3.38983029e-01 -4.44589972e-01 -3.72347444e-01
-5.26783884e-01 -1.03022552e+00 1.50289035e+00 -1.29353538e-01
-1.49184138e-01 -1.12517464e+00 -1.95627913e-01 -1.48616713e-02
4.32348669e-01 8.82289410e-01 1.48740840e+00 -1.38735026e-01
-8.83542597e-01 -1.74440578e-01 -4.37380187e-02 2.64396489e-01
4.10968482e-01 2.36250624e-01 -3.53354871e-01 -3.24881881e-01
-4.03828382e-01 -2.33267203e-01 4.69659030e-01 4.00922596e-01
1.42767811e+00 -3.91184300e-01 -1.96735308e-01 8.35835755e-01
1.15964127e+00 -1.43950164e-01 6.17425442e-01 -4.19579744e-01
1.20333052e+00 6.63411021e-01 -1.82561874e-01 3.17494899e-01
2.54969984e-01 7.99891949e-01 6.68121338e-01 -6.72369078e-02
-3.24364990e-01 -4.72219110e-01 -1.99442625e-01 1.01419222e+00
-2.06302509e-01 -8.33023638e-02 -9.58389342e-01 5.05608678e-01
-1.40686536e+00 -6.66875184e-01 -2.09352195e-01 2.15415764e+00
6.66849017e-01 8.01545605e-02 -1.39662355e-01 -1.87365100e-01
6.06017888e-01 2.06964180e-01 -8.28278005e-01 -3.85140896e-01
-1.57108635e-01 5.85989952e-01 2.52594024e-01 2.35992432e-01
-6.48729742e-01 8.88624549e-01 6.18155575e+00 5.99062204e-01
-1.04132712e+00 -3.05111229e-01 5.44972420e-01 1.21431574e-01
-1.09662259e+00 -1.06238276e-01 -3.11079353e-01 -5.68616912e-02
3.23885381e-01 -9.68673080e-02 4.50904042e-01 6.81884110e-01
-3.33636105e-02 5.29280722e-01 -1.21296120e+00 9.79903162e-01
-1.23406827e-01 -1.80685627e+00 7.12785959e-01 1.22240499e-01
1.03124189e+00 -1.75961033e-01 -1.50304794e-01 5.14985435e-02
4.70486373e-01 -1.22442627e+00 7.40318060e-01 5.77361643e-01
1.24903667e+00 -6.32631660e-01 1.04939945e-01 2.96937168e-01
-1.58737051e+00 3.84320050e-01 -4.08236116e-01 3.20163697e-01
3.96540970e-01 6.28853381e-01 -6.79222107e-01 8.78414810e-01
2.58220226e-01 5.99169552e-01 -3.85901123e-01 9.00598466e-01
-1.48444429e-01 2.86223590e-01 -5.23787796e-01 3.26071292e-01
2.23022718e-02 -4.05622184e-01 6.27807617e-01 6.34451687e-01
4.55367088e-01 2.65226930e-01 2.34278247e-01 1.41108096e+00
-3.32408339e-01 -4.31591470e-04 -8.68111372e-01 -1.84058681e-01
6.54840410e-01 1.16872215e+00 -9.55697417e-01 -1.89269394e-01
-1.46271273e-01 6.68676317e-01 4.30075198e-01 4.18951094e-01
-4.55257267e-01 -3.02497726e-02 5.79561651e-01 5.25227726e-01
5.29454052e-01 -7.25887597e-01 -4.79808033e-01 -8.78466070e-01
1.42674968e-01 -5.23469388e-01 5.47081716e-02 -9.53265965e-01
-1.43195641e+00 7.02207148e-01 7.81236365e-02 -1.07246232e+00
-1.32725298e-01 -5.67717314e-01 -5.29012024e-01 8.24431956e-01
-1.02336538e+00 -1.63612175e+00 -3.91724408e-01 4.63529676e-01
4.04247344e-01 -1.19376428e-01 9.13540065e-01 -2.07448974e-02
6.01422563e-02 3.59110832e-01 -2.06864923e-01 8.45658630e-02
1.29446268e-01 -1.14465356e+00 9.27833796e-01 4.28687036e-01
2.85020232e-01 4.21799839e-01 1.80744603e-01 -9.72150028e-01
-1.44683993e+00 -1.45788312e+00 3.60893607e-01 -5.79570055e-01
1.44741997e-01 -5.71073055e-01 -1.13874626e+00 6.56539381e-01
-5.42452812e-01 4.24492061e-01 2.47414649e-01 -1.64703757e-01
-4.96787608e-01 2.35043734e-01 -1.26004231e+00 6.78595662e-01
1.55567348e+00 -3.68255556e-01 -4.56564695e-01 3.13619137e-01
9.52309012e-01 -6.74694061e-01 -1.07737303e+00 8.84798050e-01
4.73147810e-01 -8.10299456e-01 1.32913315e+00 -7.58290827e-01
4.29874748e-01 -1.22630104e-01 -1.99946791e-01 -1.36082017e+00
-4.88213092e-01 -5.71280181e-01 -1.85356736e-01 8.18377614e-01
3.12363744e-01 -5.22221923e-01 9.21273708e-01 7.62050748e-01
-5.75098276e-01 -1.13476872e+00 -8.14673424e-01 -6.46583974e-01
2.55697012e-01 -4.22413200e-01 1.18062449e+00 9.65956807e-01
-9.14145708e-01 1.36921600e-01 4.37072478e-02 3.66705880e-02
5.47108352e-01 4.20399368e-01 9.20949399e-01 -1.53210211e+00
-1.04587674e-01 -6.08449340e-01 -5.31641066e-01 -1.11655354e+00
7.94827044e-02 -1.26323390e+00 -1.03193276e-01 -2.03924108e+00
-3.43476266e-01 -7.48928368e-01 3.42273057e-01 6.08985364e-01
1.61251321e-01 1.91525221e-01 5.26267570e-04 1.05716601e-01
-9.41733643e-02 1.08786345e+00 1.92670381e+00 -1.39901444e-01
-3.21202785e-01 6.23084158e-02 -5.94922006e-01 7.25499988e-01
4.28144217e-01 -1.44628659e-01 -5.09549677e-01 -4.91135359e-01
4.50705171e-01 2.25352928e-01 4.75176185e-01 -6.36963010e-01
-1.82139114e-01 -3.29580158e-01 3.84235829e-01 -7.20059097e-01
4.61625308e-01 -8.22026730e-01 6.11531198e-01 5.59484586e-02
-4.56731580e-02 1.37883767e-01 1.85803115e-01 5.41789353e-01
3.14075276e-02 1.66370302e-01 6.76613569e-01 -5.09730756e-01
-2.73491144e-01 1.01992881e+00 2.70550877e-01 7.64359534e-02
7.60829449e-01 -6.17841899e-01 4.20454703e-02 -3.55613023e-01
-9.22107637e-01 -1.00346114e-02 7.82010019e-01 4.94439989e-01
9.04273152e-01 -1.87353206e+00 -7.61286139e-01 5.85993767e-01
1.86276749e-01 9.58426416e-01 3.51986200e-01 2.24918529e-01
-5.44975340e-01 2.44402904e-02 -2.53741592e-01 -7.22161472e-01
-9.28193331e-01 3.08725268e-01 3.79754066e-01 -1.57121420e-01
-8.37701738e-01 7.76742697e-01 4.95286733e-01 -1.01634908e+00
-2.01582029e-01 -6.04132116e-01 1.14040546e-01 -3.98452938e-01
-1.00642532e-01 5.75054809e-02 2.49613717e-01 -8.55585039e-01
-5.15521318e-02 8.48620534e-01 3.70046914e-01 1.68094233e-01
1.70126534e+00 3.17120522e-01 -3.54718417e-01 2.37590119e-01
1.05316234e+00 1.24468662e-01 -1.31399310e+00 -2.52617031e-01
-3.39136869e-01 -4.30263460e-01 -3.12681645e-01 -4.58857715e-01
-1.16057479e+00 7.07051396e-01 -1.86903998e-02 1.90380171e-01
7.53675997e-01 3.74194950e-01 8.78653705e-01 2.97215730e-01
7.15063274e-01 -6.32990301e-01 3.69569004e-01 5.57893395e-01
1.69999576e+00 -7.31374323e-01 1.45098612e-01 -8.03734362e-01
-9.96289626e-02 1.20697796e+00 3.69174838e-01 -3.53285968e-01
9.15296912e-01 1.38042659e-01 -4.01089907e-01 -7.33844757e-01
-3.84946585e-01 3.04575078e-02 9.08653557e-01 7.40182936e-01
2.66355306e-01 2.21155286e-01 1.98179975e-01 2.02735230e-01
-4.01008755e-01 -4.42421526e-01 2.49246746e-01 6.47340715e-01
-1.36927083e-01 -1.36776531e+00 -2.82538235e-01 7.54376352e-01
2.08289385e-01 1.33766174e-01 -4.25473928e-01 6.95742786e-01
1.96069255e-01 7.32537434e-02 8.33349526e-02 -1.51897117e-01
6.22517765e-01 2.57649515e-02 7.21510828e-01 -1.02209580e+00
-1.73863664e-01 -6.46808967e-02 -1.50644124e-01 -5.15854478e-01
-2.48026609e-01 -6.40066922e-01 -1.29793620e+00 -2.21923739e-02
-8.17256346e-02 -2.82440841e-01 5.82973778e-01 7.34116971e-01
6.63629115e-01 5.98661125e-01 4.28783625e-01 -1.47494495e+00
-3.12214285e-01 -6.82753980e-01 -4.87881452e-01 8.61355245e-01
-1.02872783e-02 -7.76845515e-01 -1.42970785e-01 5.89208305e-02] | [8.823758125305176, -3.6563656330108643] |
7b258e57-c60d-4a40-b64c-5422eedf9a95 | dimensionality-reduction-as-probabilistic | 2304.07658 | null | https://arxiv.org/abs/2304.07658v2 | https://arxiv.org/pdf/2304.07658v2.pdf | Dimensionality Reduction as Probabilistic Inference | Dimensionality reduction (DR) algorithms compress high-dimensional data into a lower dimensional representation while preserving important features of the data. DR is a critical step in many analysis pipelines as it enables visualisation, noise reduction and efficient downstream processing of the data. In this work, we introduce the ProbDR variational framework, which interprets a wide range of classical DR algorithms as probabilistic inference algorithms in this framework. ProbDR encompasses PCA, CMDS, LLE, LE, MVU, diffusion maps, kPCA, Isomap, (t-)SNE, and UMAP. In our framework, a low-dimensional latent variable is used to construct a covariance, precision, or a graph Laplacian matrix, which can be used as part of a generative model for the data. Inference is done by optimizing an evidence lower bound. We demonstrate the internal consistency of our framework and show that it enables the use of probabilistic programming languages (PPLs) for DR. Additionally, we illustrate that the framework facilitates reasoning about unseen data and argue that our generative models approximate Gaussian processes (GPs) on manifolds. By providing a unified view of DR, our framework facilitates communication, reasoning about uncertainties, model composition, and extensions, particularly when domain knowledge is present. | ['Neil D. Lawrence', 'Vidhi Lalchand', 'Francisco Vargas', 'Aditya Ravuri'] | 2023-04-15 | null | null | null | null | ['probabilistic-programming'] | ['methodology'] | [-2.34838858e-01 3.15807819e-01 3.01534086e-01 -4.24256772e-02
-6.45427883e-01 -8.62543046e-01 1.18198895e+00 -3.44088897e-02
1.17141694e-01 2.68310249e-01 4.44605887e-01 -4.32111561e-01
-7.16995120e-01 -8.70357037e-01 -5.95473945e-01 -9.34186220e-01
-2.45950613e-02 8.42650890e-01 5.80317639e-02 2.82524824e-01
1.87180668e-01 9.53782439e-01 -1.32097757e+00 -1.85155362e-01
5.36914229e-01 7.52244592e-01 -4.33273911e-02 7.91273832e-01
-1.29834741e-01 4.85104263e-01 1.24154519e-02 -6.58287346e-01
1.93200991e-01 2.15630070e-03 -6.44757092e-01 -2.02446356e-01
-9.27015394e-02 -2.99310964e-02 -4.60124016e-01 1.15600610e+00
2.14797541e-01 3.34024757e-01 1.26083136e+00 -1.63812375e+00
-7.17698097e-01 4.88857538e-01 -4.34439003e-01 -1.27697825e-01
1.40263036e-01 1.09786168e-01 7.93017685e-01 -1.06587338e+00
7.76842475e-01 1.74961007e+00 6.79953694e-01 1.77175775e-01
-1.70712686e+00 2.98706945e-02 -1.47996739e-01 -1.50912851e-01
-1.44856632e+00 -3.00328523e-01 7.32838750e-01 -8.88599932e-01
7.80453205e-01 4.34963673e-01 5.13642669e-01 1.02537656e+00
2.68566996e-01 7.64409840e-01 7.33819723e-01 -2.49532580e-01
6.77857757e-01 6.49238899e-02 2.49176398e-01 5.32298625e-01
3.44282597e-01 3.75739783e-02 -6.10167503e-01 -6.18758440e-01
5.67153335e-01 -2.06434205e-02 -4.18910980e-02 -9.44956541e-01
-1.03708148e+00 9.40222919e-01 -2.94027813e-02 -2.55879730e-01
-4.25143361e-01 2.27511883e-01 1.63964599e-01 -2.22877622e-01
5.41643023e-01 1.79156110e-01 -1.49824485e-01 -4.24876809e-01
-8.42203617e-01 2.68319547e-01 1.02784669e+00 1.01106131e+00
7.15669215e-01 -3.08782309e-01 -7.21792057e-02 4.05881643e-01
8.68388593e-01 6.50121748e-01 8.15783888e-02 -1.42882752e+00
1.68803751e-01 2.60865808e-01 8.74617547e-02 -1.13112283e+00
-4.48071808e-01 -1.93131045e-02 -8.24597776e-01 1.74442127e-01
4.34824936e-02 -8.64502043e-02 -6.67559564e-01 1.61154473e+00
5.43515801e-01 3.61905307e-01 2.85514444e-01 6.69274747e-01
3.51193786e-01 6.16239667e-01 7.84520954e-02 -5.21207415e-02
1.08309090e+00 -3.15209746e-01 -6.93593442e-01 1.21496908e-01
4.62484568e-01 -4.43659991e-01 7.87438691e-01 6.72472119e-01
-9.66666996e-01 -2.28839278e-01 -7.54182279e-01 -4.52637672e-01
-4.63536203e-01 -1.15509465e-01 6.01204276e-01 6.49286926e-01
-1.35093307e+00 8.39547694e-01 -1.50044239e+00 -3.78428251e-01
3.22846949e-01 6.02945238e-02 -2.89931804e-01 7.46316835e-02
-7.25224733e-01 8.10506165e-01 2.32947320e-01 1.39525190e-01
-7.87219942e-01 -7.92186201e-01 -1.06862700e+00 4.58956696e-02
-5.45170307e-02 -9.06035781e-01 7.64140189e-01 2.50841320e-01
-1.44655645e+00 6.40698791e-01 -2.11961031e-01 -5.80004811e-01
6.79735780e-01 -3.46859366e-01 -1.21097475e-01 1.39870793e-01
-2.28716075e-01 6.04659021e-01 1.03070879e+00 -1.11799765e+00
-1.33453801e-01 -4.66226608e-01 -3.70463848e-01 3.91033962e-02
8.71752575e-02 -2.65497357e-01 -6.41182363e-01 -3.37110907e-01
4.57498968e-01 -1.07577431e+00 -5.43837361e-02 -5.39324246e-02
-7.60465026e-01 -1.22566491e-01 1.10782266e+00 -7.49114275e-01
1.24111438e+00 -2.42924786e+00 8.90987754e-01 5.69627523e-01
4.54927385e-01 -3.17146212e-01 3.38639766e-01 5.34674466e-01
6.52541667e-02 4.67941016e-01 -5.93491137e-01 -7.66069233e-01
7.26970851e-01 3.03236306e-01 -4.51759458e-01 7.33176827e-01
3.56654048e-01 9.66817677e-01 -7.95202553e-01 -3.27188641e-01
5.51177740e-01 8.82753551e-01 -4.30326462e-01 8.00962672e-02
-2.22530231e-01 3.76831234e-01 -3.45525116e-01 4.83076185e-01
7.53084838e-01 -1.32105231e-01 1.15846083e-01 -4.11404788e-01
-1.77462593e-01 -3.47468555e-02 -1.52132916e+00 1.78925979e+00
-8.72662291e-02 5.78976333e-01 1.76281974e-01 -5.44463992e-01
7.85589516e-01 -1.57784477e-01 3.78145814e-01 1.13895141e-01
-1.01700470e-01 -1.30104825e-01 -3.77025396e-01 -3.59378248e-01
4.79449362e-01 2.02372774e-01 9.11469478e-03 6.85283184e-01
-2.73450976e-03 -5.13641357e-01 1.84942439e-01 5.17797351e-01
1.20941484e+00 6.28611088e-01 -9.14764404e-02 -3.37164700e-01
2.63892353e-01 3.85927930e-02 1.74318612e-01 5.54488957e-01
7.13070435e-03 6.65070772e-01 9.56080198e-01 1.47943348e-02
-1.29959977e+00 -1.57882464e+00 -4.76611406e-01 5.10872126e-01
-2.10061342e-01 -6.00780189e-01 -7.27298975e-01 -2.74343520e-01
2.07321316e-01 1.20008433e+00 -6.81698740e-01 -2.77351201e-01
-1.38931394e-01 -1.03777611e+00 4.04848933e-01 3.60796720e-01
-3.38912793e-02 -4.30962861e-01 -2.83953875e-01 -9.62233096e-02
2.56384790e-01 -7.37871647e-01 7.43406372e-06 4.08305526e-02
-9.28740740e-01 -1.00417399e+00 -3.15449625e-01 -5.19568799e-03
5.77021837e-01 -4.39068414e-02 8.71437609e-01 -7.32994258e-01
-3.00594628e-01 8.92015755e-01 6.74485564e-02 -2.83736885e-01
-5.84167421e-01 -3.13406765e-01 3.45637262e-01 -8.69932547e-02
4.46160972e-01 -7.77965128e-01 -3.73486370e-01 1.02699986e-02
-9.95119572e-01 -1.14911400e-01 2.76153535e-01 5.37425458e-01
1.04313493e+00 2.60556489e-01 -2.93602020e-01 -7.22170055e-01
9.14843678e-01 -8.81399333e-01 -8.63330722e-01 1.99013531e-01
-7.33994067e-01 5.98390341e-01 8.29220116e-02 -2.82318175e-01
-1.07714152e+00 -1.56679121e-03 2.25656301e-01 -1.02724195e+00
-7.39762187e-02 5.16771197e-01 -3.48934978e-01 2.97677368e-01
6.12514436e-01 1.01927564e-01 3.29498976e-01 -7.34977782e-01
1.04521549e+00 4.88741457e-01 5.87071538e-01 -7.99833119e-01
7.57206082e-01 9.51671004e-01 5.75005472e-01 -7.27345288e-01
-4.26050007e-01 -2.97240883e-01 -1.01608169e+00 3.75445411e-02
1.01065874e+00 -6.89121246e-01 -7.00577974e-01 1.37488991e-01
-8.96198213e-01 -2.06900343e-01 -4.19720054e-01 5.72642148e-01
-7.05031991e-01 4.50362772e-01 -4.95240867e-01 -1.07117915e+00
-8.60835686e-02 -1.28469169e+00 1.26015294e+00 -2.66227908e-02
-2.82101005e-01 -1.35207927e+00 2.08378255e-01 -1.28998265e-01
2.73510963e-01 4.23914790e-01 1.05973911e+00 -7.22911775e-01
-5.09283364e-01 -1.77922145e-01 -1.68515518e-01 2.77175784e-01
-1.64669529e-01 6.26316905e-01 -1.00975513e+00 2.69855689e-02
5.31717986e-02 2.11391240e-01 6.26532853e-01 6.26356125e-01
1.12568605e+00 -6.35785162e-02 -4.79843378e-01 8.90058041e-01
1.10771811e+00 -3.29573423e-01 4.90424544e-01 7.93227255e-02
8.51784289e-01 7.49498248e-01 2.40840882e-01 3.95319670e-01
5.77260733e-01 2.64145523e-01 5.15446842e-01 4.01967913e-01
7.62398764e-02 -2.99187303e-01 2.69497693e-01 6.05717063e-01
9.06848982e-02 3.71427648e-02 -1.16463900e+00 2.05456778e-01
-2.07551336e+00 -8.61945927e-01 -6.14825726e-01 2.29249024e+00
4.50327307e-01 -1.35243200e-02 -2.00736318e-02 -8.79705697e-02
5.71851373e-01 -1.12618975e-01 -7.08138824e-01 -3.31646264e-01
-8.17240402e-02 -1.28421471e-01 4.99889612e-01 6.17964685e-01
-1.11849201e+00 7.01833129e-01 6.85358810e+00 5.28766036e-01
-5.49737155e-01 5.81078678e-02 4.04207379e-01 -7.93476254e-02
-4.82626468e-01 7.82636777e-02 -7.59756684e-01 3.26283187e-01
1.24708676e+00 -3.76594812e-01 6.84222221e-01 1.02064431e+00
3.58164012e-01 -8.62926990e-02 -1.40855193e+00 9.72370207e-01
-9.00743455e-02 -1.36264420e+00 -6.02715351e-02 4.30054218e-01
4.76397514e-01 1.72807500e-01 1.78732842e-01 7.96108618e-02
9.17014241e-01 -8.82071972e-01 6.47467792e-01 1.25198698e+00
5.67575574e-01 -8.28705490e-01 4.11188215e-01 2.95622200e-01
-7.35346675e-01 2.96636730e-01 -4.45174009e-01 3.71200264e-01
3.69054556e-01 8.19589019e-01 -6.62017226e-01 6.07983053e-01
6.48417294e-01 7.95868278e-01 -4.60320950e-01 7.20144510e-01
-2.97450662e-01 4.58613902e-01 -7.64318287e-01 4.57298249e-01
-1.34271942e-02 -7.98130751e-01 9.07443225e-01 1.09129953e+00
5.13578594e-01 -1.28264472e-01 -1.48782477e-01 1.38825488e+00
1.98383838e-01 -3.51367593e-01 -6.14728868e-01 -2.67330647e-01
5.29980898e-01 1.27850926e+00 -6.12344205e-01 -1.60834044e-01
-1.11406781e-01 9.13780510e-01 2.41645694e-01 5.29683650e-01
-4.91916895e-01 -4.84412909e-02 9.78136182e-01 -2.10202292e-01
2.04984263e-01 -6.18010879e-01 -3.18383694e-01 -1.05266309e+00
-8.42231289e-02 -4.11947429e-01 4.40541089e-01 -1.04444909e+00
-1.52338505e+00 3.98126394e-02 3.94153655e-01 -8.87421012e-01
-5.64571738e-01 -7.72901952e-01 -4.68665838e-01 1.16891038e+00
-1.05467391e+00 -9.33615148e-01 -9.63258073e-02 3.58590007e-01
-3.02562327e-03 -3.68271489e-03 7.48991787e-01 -1.83280185e-01
-8.47523153e-01 -2.27292687e-01 5.78042865e-01 -1.81417301e-01
3.10258120e-01 -1.70152462e+00 5.29710352e-01 1.01623309e+00
1.01613499e-01 9.72371399e-01 9.42366123e-01 -9.65351284e-01
-2.04915285e+00 -1.06894863e+00 1.71212822e-01 -8.45994949e-01
1.21847379e+00 -3.92538726e-01 -9.73662198e-01 8.47316384e-01
-3.89193535e-01 4.31753695e-02 7.66730726e-01 5.99599965e-02
-1.58317193e-01 2.84531564e-01 -1.18639672e+00 5.55988014e-01
8.57868731e-01 -7.07360625e-01 -6.07858598e-01 3.64505142e-01
7.09987223e-01 -4.84924078e-01 -1.22732675e+00 -2.16812976e-02
4.15973753e-01 -8.34643602e-01 1.16403806e+00 -6.31246924e-01
9.88763124e-02 -5.55965364e-01 -3.86712015e-01 -1.27903569e+00
-3.15142632e-01 -8.58929515e-01 -9.69659269e-01 1.29829037e+00
2.11213022e-01 -4.68217969e-01 5.18781245e-01 1.29316485e+00
-9.30063650e-02 -2.40422398e-01 -8.49415660e-01 -6.30860329e-01
2.51145184e-01 -1.14558136e+00 8.26651514e-01 7.51384974e-01
-1.24343596e-01 -1.83350906e-01 -4.05048691e-02 6.54902577e-01
1.13183153e+00 -1.51533619e-01 8.06882381e-01 -1.52968478e+00
-3.28635395e-01 -5.77470541e-01 -4.71393824e-01 -7.05493450e-01
1.09679595e-01 -1.01483476e+00 -3.40981811e-01 -1.68649578e+00
-3.31351161e-02 -4.11652565e-01 2.25564972e-01 6.64751232e-02
1.33816823e-01 -1.57777667e-01 -3.27369906e-02 6.32066727e-01
-3.75993341e-01 5.73934793e-01 6.62483513e-01 1.40390724e-01
-3.60264122e-01 -2.48863950e-01 -4.58999962e-01 8.69156003e-01
5.57659030e-01 -4.21477109e-01 -4.08487201e-01 -2.70787150e-01
4.89280343e-01 -3.64676982e-01 5.32136738e-01 -6.67497516e-01
4.34420466e-01 -2.69401558e-02 4.75620925e-01 -7.83317924e-01
5.07721663e-01 -6.14291072e-01 7.18590438e-01 7.81818479e-02
-5.09221330e-02 -4.79034968e-02 1.62651315e-01 8.71282935e-01
1.00979835e-01 -1.32287338e-01 3.45872581e-01 1.04286134e-01
-4.82509643e-01 2.43667215e-01 -8.76251161e-02 -1.23351872e-01
9.50515389e-01 7.65354410e-02 -2.76385397e-01 -2.54496038e-01
-1.25386143e+00 4.24797207e-01 6.79149508e-01 2.36727208e-01
4.42345887e-01 -1.22161472e+00 -5.23755670e-01 1.09709516e-01
-8.07931647e-02 3.46690953e-01 3.01770240e-01 9.99442220e-01
-5.46198845e-01 2.77679384e-01 3.59422445e-01 -9.29827571e-01
-7.83797026e-01 6.91834033e-01 2.46747583e-01 1.10287160e-01
-8.00500691e-01 6.15964949e-01 -3.14487480e-02 -4.44158435e-01
1.11705005e-01 -3.05309504e-01 1.19608395e-01 9.34218243e-02
5.63596964e-01 8.55087101e-01 -1.69989556e-01 -6.42931104e-01
-2.51880795e-01 3.67298871e-01 3.45389247e-01 -5.19074440e-01
1.56368291e+00 -6.23139143e-01 -4.85619962e-01 7.88089395e-01
9.45861638e-01 1.03238095e-02 -1.64554131e+00 -9.78395715e-02
1.91440254e-01 -1.96474656e-01 4.91510242e-01 -4.55427229e-01
-5.59506118e-01 1.09212136e+00 3.43099385e-01 4.76583838e-01
7.56492317e-01 4.01823431e-01 2.15425212e-02 2.68879712e-01
4.01700810e-02 -1.04289103e+00 -5.53185046e-01 2.36823335e-01
1.02032936e+00 -8.34287941e-01 2.61045396e-01 -3.69536877e-01
-7.12870538e-01 1.27294898e+00 -5.55548370e-02 7.99669325e-02
1.06679261e+00 3.31090063e-01 -4.72705096e-01 -4.92357075e-01
-5.99697530e-01 -8.60575587e-02 3.76880884e-01 7.56308377e-01
-7.60149732e-02 8.22431371e-02 1.99910581e-01 5.91851056e-01
-4.98862147e-01 -3.46933037e-01 4.79431301e-01 8.42126787e-01
-1.50775239e-01 -1.00934958e+00 -5.94963849e-01 4.29149151e-01
-1.61378399e-01 -8.38738233e-02 -1.70321628e-01 7.43502498e-01
-1.41181141e-01 6.13209188e-01 2.58581102e-01 -1.96485266e-01
1.42962009e-01 5.30911922e-01 1.60233930e-01 -3.86299729e-01
2.41144344e-01 3.19742322e-01 -1.02787711e-01 -7.25161076e-01
-1.84740752e-01 -1.16643918e+00 -1.22914088e+00 -4.58021373e-01
1.16978839e-01 6.72508031e-02 1.19717979e+00 8.93684328e-01
7.25336254e-01 3.00568283e-01 1.39295310e-01 -8.66628528e-01
-5.33130050e-01 -8.30696464e-01 -7.34788358e-01 1.89464450e-01
9.65658277e-02 -8.02588999e-01 -7.20576763e-01 2.07016662e-01] | [6.90895938873291, 3.791161298751831] |
34cb9680-e218-4eb0-94b8-d86da62612d3 | a-wearable-eeg-system-for-closed-loop | 2212.11273 | null | https://arxiv.org/abs/2212.11273v1 | https://arxiv.org/pdf/2212.11273v1.pdf | A Wearable EEG System for Closed-Loop Neuromodulation of High-Frequency Sleep-Related Oscillations | In healthy sleepers, cortical alpha oscillations are present during the transition from wakefulness to sleep, and dissipate at sleep onset. For individuals with insomnia, alpha power is elevated during the wake-sleep transition and can persist throughout the night. Neuromodulation techniques using phase-locked stimulation have been put forth as alternatives to drugs for improving slow-wave sleep quality. Due to technical limitations, this approach has not been tested on faster frequency alpha oscillations. Here we examine the feasibility of using an endpoint-corrected version of the Hilbert Transform (ecHT) algorithm implemented on-device to measure alpha phase and deliver phase-locked auditory stimulation to modulate alpha and promote sleep initiation. First, the ecHT algorithm is implemented on a tabletop electroencephalogram (EEG) device and used to measure the timing of the auditory evoked response and its delivery at precise phases of the alpha oscillation. Secondly, a pilot at-home study tests feasibility to use a wearable version of the neuromodulation device for real-time phase-locked stimulation in the alpha (8-12 Hz) frequency range. Auditory stimulation was delivered at the intended phases of alpha with high precision, and alpha oscillations were affected differently by stimuli delivered at opposing phases. Our wearable system was capable of measuring sleep micro- and macro-events present in the EEG that were appropriate for clinical sleep scoring. Sleep onset latencies were reduced for a subset of subjects displaying sleep onset insomnia symptoms in the stimulation condition. This study demonstrates the feasibility of closed-loop tracking and neuromodulation of alpha oscillations using a wearable EEG device. Preliminary results suggest that this approach could be used to accelerate sleep initiation in individuals with objective insomnia symptoms. | ['David Wang', 'Ryan Yost', 'Heather Read', 'Ryan Neely', 'Scott Bressler'] | 2022-12-21 | null | null | null | null | ['sleep-quality-prediction'] | ['medical'] | [ 1.47978231e-01 -3.40467006e-01 2.37807304e-01 -1.59626435e-02
-2.96034217e-01 -4.44968343e-01 -1.14981458e-01 3.20370942e-01
-7.53924608e-01 7.98469126e-01 3.55146199e-01 -1.71518072e-01
-8.38747546e-02 -2.92954355e-01 -6.57059550e-02 -6.55266047e-01
-5.45149922e-01 1.26713216e-01 3.72981690e-02 -1.42711371e-01
1.83883458e-01 8.48851949e-02 -1.68081617e+00 3.70855480e-01
7.12917686e-01 8.09244215e-01 3.90135080e-01 5.39402068e-01
4.02552754e-01 -2.82722354e-01 -1.08398890e+00 6.59244180e-01
2.84450855e-02 -7.46510386e-01 -2.02393010e-01 -3.93843859e-01
-3.81194919e-01 -1.38077974e-01 1.67100266e-01 6.76078081e-01
1.11417770e+00 -1.16087653e-01 5.18538840e-02 -1.22272313e+00
1.58495530e-01 2.16247529e-01 -3.66565913e-01 1.01385427e+00
8.82978857e-01 3.14588845e-01 2.87788272e-01 -4.51808631e-01
-1.91926301e-01 3.04421782e-01 7.16779828e-01 5.16579390e-01
-1.66289806e+00 -9.61669207e-01 -7.31057644e-01 5.25732577e-01
-1.54685855e+00 -7.09347785e-01 5.72133064e-01 -3.23653638e-01
1.71165109e+00 1.76458791e-01 1.48623896e+00 9.26534832e-01
1.07240558e+00 -2.39258543e-01 1.42480206e+00 -3.35079879e-01
6.15856111e-01 8.56367052e-02 8.50259513e-02 -1.33103430e-01
2.83058792e-01 2.55686820e-01 -9.97699022e-01 -2.36124218e-01
2.10482895e-01 -2.79745787e-01 -7.30424821e-01 5.69657564e-01
-9.41310406e-01 4.51856107e-01 9.19135213e-02 7.06871510e-01
-7.48260260e-01 -4.07958701e-02 3.11363339e-01 1.76216632e-01
3.93687755e-01 6.88657105e-01 -2.24560305e-01 -1.04607809e+00
-1.30162525e+00 -1.30687058e-01 4.73978221e-01 2.25271210e-01
3.40556383e-01 -1.87835872e-01 -2.60204703e-01 5.96285224e-01
3.47176611e-01 6.47656143e-01 1.02130699e+00 -6.45886123e-01
-1.85523570e-01 3.62153232e-01 3.14375877e-01 -4.04496342e-01
-1.27495897e+00 -1.24283500e-01 -2.24785075e-01 1.74538732e-01
1.00277416e-01 -2.15104163e-01 -4.15022194e-01 1.44733119e+00
-1.24102928e-01 1.69565976e-01 -3.12949210e-01 8.62201691e-01
4.69498694e-01 5.06269872e-01 2.40462750e-01 -8.44546199e-01
1.83740926e+00 8.57313499e-02 -9.54976380e-01 -3.65428209e-01
2.58382291e-01 -6.54015243e-01 1.40082526e+00 4.72070456e-01
-1.18717051e+00 -4.63823438e-01 -1.33836126e+00 3.00678641e-01
2.23486535e-02 -1.67934895e-01 -7.32495561e-02 1.12286365e+00
-1.38905036e+00 2.72269607e-01 -1.42970490e+00 -5.44133067e-01
1.01847924e-01 9.51166034e-01 -1.11638226e-01 6.81411147e-01
-9.89012361e-01 9.90257263e-01 -1.53905705e-01 1.88184991e-01
-2.03925788e-01 -9.73824978e-01 -2.83807188e-01 2.78338552e-01
-7.24964559e-01 -1.03536320e+00 1.16522980e+00 -5.56451023e-01
-1.58621776e+00 6.86479926e-01 -4.45974082e-01 -4.06337470e-01
-5.02761543e-01 -7.80229792e-02 -7.40682304e-01 4.30255592e-01
3.71060848e-01 2.55162716e-01 6.64292157e-01 -3.92066836e-01
-3.41408432e-01 -7.50420213e-01 -4.18543398e-01 1.23869807e-01
-1.16636589e-01 2.09016249e-01 7.15353191e-01 -5.18267639e-02
-1.25709906e-01 -7.55328715e-01 1.96272641e-01 -5.48985958e-01
1.88955233e-01 4.66547385e-02 5.21878421e-01 -2.84698069e-01
1.33708143e+00 -2.45286751e+00 -6.23750150e-01 2.29847908e-01
1.29841045e-01 7.21332431e-02 2.04218090e-01 6.71733737e-01
-2.63316214e-01 -8.12904462e-02 -3.73320468e-02 -6.68416098e-02
-7.70037770e-02 -3.53163742e-02 -8.56802892e-03 9.02840674e-01
-2.68946499e-01 7.48067737e-01 -8.01908135e-01 2.09715337e-01
9.64066982e-02 7.18221724e-01 -3.61265838e-01 2.09108442e-01
4.81270462e-01 4.71761793e-01 4.01190728e-01 4.19075251e-01
4.80292320e-01 1.33021101e-01 -5.39086945e-02 -1.99448049e-01
-6.05049610e-01 1.02324581e+00 -5.17836571e-01 1.56194210e+00
-4.43062067e-01 7.49590874e-01 6.80512413e-02 -4.48246390e-01
8.02488744e-01 5.76740146e-01 4.92111892e-01 -1.43457210e+00
2.69047081e-01 2.50746280e-01 5.59517682e-01 -6.59167051e-01
1.41982272e-01 -8.63737106e-01 3.31982642e-01 5.64723253e-01
-2.05509171e-01 -5.10294810e-02 1.23361751e-01 -3.56472760e-01
1.41882384e+00 -2.68666208e-01 5.09281456e-01 -7.48733997e-01
-1.30856102e-02 -3.31392050e-01 5.09596914e-02 4.09377307e-01
-3.27370942e-01 4.01821464e-01 5.97093180e-02 -6.32449985e-02
-4.78589028e-01 -1.16151464e+00 -5.00131130e-01 7.83224761e-01
7.90372416e-02 -7.89440453e-01 -5.32268345e-01 3.03624660e-01
-2.35721990e-01 9.41844404e-01 -5.17917395e-01 -6.52114451e-01
-3.65006886e-02 -1.01210916e+00 3.64957869e-01 3.55418831e-01
2.21178949e-01 -1.21742761e+00 -1.45899940e+00 5.72668552e-01
-2.64750212e-01 -5.60174644e-01 -3.14226359e-01 7.72263885e-01
-7.38653481e-01 -7.46825874e-01 -3.09436500e-01 -4.15269405e-01
4.13182437e-01 5.91750890e-02 7.84647286e-01 -2.52067804e-01
-4.47549731e-01 7.27820098e-01 -2.10938096e-01 -6.96795583e-01
1.30871013e-01 -3.47766817e-01 4.91543144e-01 -2.65640855e-01
8.69233191e-01 -1.49785936e+00 -1.12002718e+00 2.30071351e-01
-3.15618008e-01 -3.82286549e-01 4.73102093e-01 2.18686238e-01
3.75314981e-01 -1.62503198e-01 8.06166828e-01 5.38572259e-02
1.23441303e+00 -3.75144303e-01 -1.28745720e-01 -4.56859320e-01
-7.69995272e-01 -4.02033538e-01 5.48494816e-01 -4.90839094e-01
-5.47031045e-01 -4.83069152e-01 -2.29918957e-01 2.66382992e-01
-1.96128964e-01 3.14100236e-01 1.11684069e-01 2.04372674e-01
1.08446491e+00 2.72555947e-01 2.28142351e-01 5.05287796e-02
-4.38766330e-01 1.00776422e+00 5.84034801e-01 1.58338211e-02
1.66826285e-02 4.33069080e-01 -3.45693082e-02 -1.14122617e+00
-2.24785417e-01 -7.18776286e-01 -1.14755668e-01 -3.31769027e-02
8.62673759e-01 -1.00886858e+00 -9.94714856e-01 7.68565852e-03
-7.35010445e-01 -5.87405741e-01 -3.17436069e-01 1.03414416e+00
-6.52214229e-01 -1.65180266e-01 -1.55372053e-01 -8.50026190e-01
-9.78013992e-01 -8.56437147e-01 8.26336920e-01 4.58076507e-01
-9.37180698e-01 -5.11988938e-01 6.07077360e-01 -1.84689328e-01
6.28766477e-01 -1.37551278e-01 5.92435002e-01 -2.33425677e-01
3.00967276e-01 -1.24823265e-01 5.06077826e-01 -1.68761432e-01
4.89979506e-01 -4.09405559e-01 -1.08441675e+00 -2.21380517e-01
6.92528427e-01 6.65080696e-02 -2.55223382e-02 9.17266667e-01
3.64126861e-01 -1.31504193e-01 -3.51779789e-01 4.71764177e-01
1.15756786e+00 7.17548192e-01 9.52533245e-01 4.82705891e-01
-2.80097127e-01 6.60661189e-03 2.07546756e-01 5.03766060e-01
1.06563650e-01 6.27748251e-01 -6.85797110e-02 2.13041946e-01
3.41584231e-03 2.42742911e-01 6.62360847e-01 5.36229193e-01
-1.12982215e-02 2.18590349e-01 -6.08917236e-01 5.01099527e-01
-8.71061504e-01 -9.49377596e-01 -5.01628697e-01 2.60631371e+00
8.55730832e-01 1.82524636e-01 5.32672942e-01 2.25212231e-01
2.16814309e-01 -6.83648527e-01 -1.42234042e-01 -8.91779006e-01
2.63143092e-01 1.11091340e+00 1.58052653e-01 3.81931186e-01
-1.20866977e-01 -6.00108244e-02 6.96475649e+00 -6.67481944e-02
-1.52708876e+00 4.35942829e-01 -1.42170310e-01 -7.89738178e-01
-7.65506085e-03 5.00908345e-02 -6.53561473e-01 1.10359824e+00
1.98907804e+00 -5.73825300e-01 6.08693004e-01 2.80493289e-01
1.29556227e+00 -7.76690125e-01 -1.07855618e+00 1.12209630e+00
-1.53281406e-01 -8.87817025e-01 -7.70105124e-01 2.07634360e-01
-1.00010231e-01 8.22272301e-02 -1.79976448e-01 -2.81362645e-02
-1.17406905e+00 -9.34916735e-01 4.27684844e-01 6.51373446e-01
1.12122595e+00 -8.14957559e-01 6.01150095e-01 1.86018437e-01
-1.11028910e+00 -2.64605492e-01 -1.57362238e-01 -5.81207514e-01
3.19252640e-01 6.05900764e-01 -1.11699581e+00 -2.95794219e-01
9.26519334e-01 2.98975497e-01 -5.89199662e-01 1.36972046e+00
-2.21104845e-01 1.08088195e+00 -8.03617656e-01 -1.55656323e-01
-5.77992499e-01 -9.69176069e-02 4.85214889e-01 6.99934304e-01
6.33808374e-01 2.50564396e-01 -6.27721846e-01 1.09536564e+00
6.16296172e-01 -3.46840262e-01 -4.89511490e-01 -2.05344316e-02
4.68167871e-01 1.18664968e+00 -9.35603321e-01 2.54376173e-01
-3.50015163e-01 5.65879703e-01 -5.80726862e-01 2.94129342e-01
-5.61787903e-01 -6.20149791e-01 4.80885744e-01 6.44800663e-01
-2.52445310e-01 -1.28989115e-01 -6.89842999e-01 -4.88004029e-01
1.52160034e-01 -5.91871679e-01 -2.77154818e-02 -1.08595359e+00
-6.11001611e-01 6.08311355e-01 6.80053607e-02 -1.19470429e+00
-2.45823681e-01 3.73419351e-03 -1.11305988e+00 1.10465050e+00
-7.95874238e-01 -4.24141914e-01 -2.62563586e-01 5.95564187e-01
2.46979624e-01 4.14136052e-01 1.17726719e+00 2.56628603e-01
-3.57214846e-02 3.01550686e-01 -3.15609723e-01 -8.16263378e-01
6.67371869e-01 -1.14647424e+00 -3.32975686e-01 5.96032202e-01
-2.03941777e-01 1.06097329e+00 1.09402263e+00 -6.39480293e-01
-1.04267895e+00 -5.10629714e-01 1.05549800e+00 -2.47149378e-01
4.53103662e-01 -6.20717168e-01 -5.96085310e-01 1.76886007e-01
4.62019414e-01 -6.54206038e-01 1.30732226e+00 -1.53024614e-01
5.41228056e-01 -5.24298251e-01 -1.19938147e+00 3.94982517e-01
4.07345474e-01 -5.82627118e-01 -8.24121177e-01 1.92856759e-01
2.37512261e-01 -1.01752896e-02 -7.97583222e-01 1.88519046e-01
8.79777431e-01 -1.16608155e+00 3.76138866e-01 4.45972055e-01
-1.36894360e-01 -3.52148712e-01 4.69788104e-01 -1.50448799e+00
-2.77619869e-01 -1.20777202e+00 4.43152823e-02 9.32137609e-01
1.66026443e-01 -9.58092153e-01 3.34294468e-01 4.44264084e-01
-5.21698296e-01 -4.87601995e-01 -1.18371761e+00 -5.72328389e-01
-5.38442135e-01 -1.33953914e-01 1.86937004e-01 -1.92880258e-01
1.08919954e+00 7.44211376e-01 3.06185395e-01 1.34794965e-01
-4.85659484e-03 -2.13046789e-01 8.31416026e-02 -8.02020848e-01
-2.79342532e-01 -2.38541797e-01 -7.11700261e-01 -3.14645618e-01
-3.37729126e-01 -7.90575743e-01 1.29091769e-01 -1.64846277e+00
-7.65850544e-02 3.76144290e-01 -1.97574049e-01 4.99722421e-01
2.63735857e-02 5.26503742e-01 -3.71315390e-01 -5.92559725e-02
3.14484239e-01 4.49322045e-01 4.64905739e-01 1.62087768e-01
-1.06402445e+00 2.63915032e-01 -7.48948097e-01 2.79527098e-01
9.01202619e-01 -7.70833671e-01 -8.98346245e-01 2.47665629e-01
3.75689149e-01 -6.25291690e-02 2.41738349e-01 -1.32934904e+00
4.19149876e-01 5.42937398e-01 4.53179240e-01 -5.00417709e-01
4.67815727e-01 -6.99068248e-01 4.40005451e-01 6.88120365e-01
2.35622376e-01 4.87034708e-01 6.40000939e-01 1.70355245e-01
3.10640097e-01 6.41727820e-02 6.58606291e-01 1.35226458e-01
4.42336500e-02 -4.37614411e-01 -1.25742829e+00 -1.20791852e-01
8.68621469e-01 -7.87732005e-01 -4.57686961e-01 -2.85910070e-01
-9.90780950e-01 -2.50901818e-01 3.99759054e-01 -3.42032164e-02
6.30808055e-01 -9.52813148e-01 -1.43422812e-01 7.30286956e-01
-7.32954824e-03 -5.32787919e-01 3.17466676e-01 1.84448385e+00
-4.01495993e-01 5.20660937e-01 -6.70461774e-01 -9.34780359e-01
-1.18067777e+00 3.30213994e-01 5.13373375e-01 3.36257398e-01
-9.28431451e-01 7.22375929e-01 -1.85821414e-01 9.74240959e-01
-5.09286597e-02 -4.22984809e-01 -2.40740851e-01 1.16018392e-02
9.86673176e-01 2.02204823e-01 6.82945430e-01 -5.61142415e-02
-8.13660681e-01 2.78054237e-01 4.47014838e-01 -3.81375253e-01
1.17798603e+00 -4.46214825e-01 -4.74056035e-01 9.91596580e-01
8.77613723e-01 2.26720855e-01 -7.02974439e-01 9.69906986e-01
-2.65193403e-01 -1.76107898e-01 6.86447695e-02 -1.08434856e+00
-4.72179294e-01 6.75480664e-01 1.43291783e+00 4.08162355e-01
1.74617517e+00 -1.19574949e-01 6.82703674e-01 -6.44336194e-02
6.02831721e-01 -1.00059521e+00 -1.49287537e-01 -3.91355693e-01
5.78439713e-01 2.04397039e-03 1.03154428e-01 2.59775341e-01
-9.61669981e-02 1.05014908e+00 1.28534317e-01 -1.23890035e-01
9.10174608e-01 5.78982890e-01 9.88082960e-03 -4.71691012e-01
-7.29197919e-01 -1.70537382e-01 -1.94975827e-02 9.90483582e-01
5.72132468e-01 1.18220106e-01 -1.15361619e+00 8.29806030e-01
-7.57913470e-01 4.29396719e-01 9.00910497e-01 8.62256765e-01
-4.83082861e-01 -9.07360017e-01 -5.22684753e-01 8.85459721e-01
-4.97358799e-01 -1.41606539e-01 -5.89045659e-02 4.59174991e-01
2.94350892e-01 1.50662613e+00 4.03223306e-01 -3.24356139e-01
5.93761206e-01 4.45938170e-01 7.91096449e-01 -8.17891002e-01
-8.16938519e-01 4.96571869e-01 -2.19895095e-02 -7.51040816e-01
-5.18565416e-01 -7.12015450e-01 -1.47848463e+00 1.67304695e-01
3.65246087e-02 4.82807517e-01 8.06078732e-01 1.04542279e+00
7.18717754e-01 5.83016634e-01 2.73716509e-01 -8.74772727e-01
5.28939068e-03 -1.22113931e+00 -1.18359137e+00 -1.55926466e-01
5.69456458e-01 -6.21700644e-01 -6.46003544e-01 1.23791432e-03] | [13.484414100646973, 3.42234206199646] |
ce71c654-9db3-477e-bf20-53e8c06ba4b4 | federated-learning-enables-big-data-for-rare | 2204.10836 | null | https://arxiv.org/abs/2204.10836v2 | https://arxiv.org/pdf/2204.10836v2.pdf | Federated Learning Enables Big Data for Rare Cancer Boundary Detection | Although machine learning (ML) has shown promise in numerous domains, there are concerns about generalizability to out-of-sample data. This is currently addressed by centrally sharing ample, and importantly diverse, data from multiple sites. However, such centralization is challenging to scale (or even not feasible) due to various limitations. Federated ML (FL) provides an alternative to train accurate and generalizable ML models, by only sharing numerical model updates. Here we present findings from the largest FL study to-date, involving data from 71 healthcare institutions across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, utilizing the largest dataset of such patients ever used in the literature (25,256 MRI scans from 6,314 patients). We demonstrate a 33% improvement over a publicly trained model to delineate the surgically targetable tumor, and 23% improvement over the tumor's entire extent. We anticipate our study to: 1) enable more studies in healthcare informed by large and diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further quantitative analyses for glioblastoma via performance optimization of our consensus model for eventual public release, and 3) demonstrate the effectiveness of FL at such scale and task complexity as a paradigm shift for multi-site collaborations, alleviating the need for data sharing. | ['Carmen Balana', 'Spyridon Bakas', 'Jason Martin', 'Jill S Barnholtz-Sloan', 'Bjoern Menze', 'Prashant Shah', 'Charles Apgar', 'Lisa Cimino', 'Cynthia Price', 'Sailaja Marella', 'Brian Bialecki', 'Kendall Schmidt', 'Deepak Kattil Veettil', 'James Gimpel', 'Michael A Boss', 'Fabio Y Moraes', 'Danielle Cutler', 'Anh Tran', 'Ricky Hu', 'Jacob J Peoples', 'Mohammad Hamghalam', 'Amber L Simpson', 'Farouk Dako', 'Mayowa Soneye', 'Adeleye Dorcas', "Mustapha Shu'aibu", 'Babatunde Osobu', 'Olubunmi Odafe-Oyibotha', 'Dotun Oyekunle', 'Godwin Ogbole', 'Sergio A Velastin', 'Eduardo López', 'Elvis Ríos', 'Franco Vera', 'Cristobal Mendoza', 'Esteban Torche', 'Pamela Guevara', 'Sebastian Quevedo', 'Francis Loayza', 'Heydy Franco-Maldonado', 'Enrique Pelaez', 'Tahsin Kurc', 'David Payne', 'Kartik M Mani', 'Sampurna Shrestha', 'Prateek Prasanna', 'Joel Saltz', 'Hassan F Shaykh', 'Bertrand Pouymayou', 'Andrea Bink', 'Michael Weller', 'Tobias Weiss', 'Hongwei Li', 'Aikaterini Kotrotsou', 'John M Buatti', 'Bryan Allen', 'Heba Ismael', 'Yusung Kim', 'Stephen Baek', 'Joseph Choi', 'Akshitkumar Mistry', 'Lola B Chambless', 'Silky Chotai', 'Kaiwen Xu', 'Karthik Ramadass', 'Bennett Landman', 'Martin Lepage', 'David Fortin', 'Martin Vallières', 'Ann-Christin Hau', 'Olivier Keunen', 'Simone P Niclou', 'Saba Adabi', 'Tzu-Chi Tseng', 'Sonam Sharma', 'Yading Yuan', 'Stefan Klein', 'Pim J French', 'Martin J van den Bent', 'Arnaud JPE Vincent', 'Hendrikus J Dubbink', 'Joost W Schouten', 'Renske Gahrmann', 'Georgios Kapsas', 'Maarten MJ Wijnenga', 'Fatih Incekara', 'Ahmed Alafandi', 'Sebastian R van der Voort', 'Marion Smits', 'Neeraj Kumar', 'Jonathan Chen', 'Ipsa Yadav', 'Rohan Bareja', 'Ruchika Verma', 'Pallavi Tiwari', 'Ning Wen', 'Laila Poisson', 'Bing Luo', 'Sung Soo Ahn', 'Jong Hee Chang', 'Seung-Koo Lee', 'Yoon Seong Choi', 'Dimitrios M Kardamakis', 'Christina Kalogeropoulou', 'Konstantinos Moustakas', 'Evangelia I Zacharaki', 'Ilias Haliassos', 'Sotiris Alexiou', 'Panagiotis Tsiganos', 'Vasileios Panagiotopoulos', 'Peter Zampakis', 'Matthew Williams', 'Luke V M Dixon', 'Ben Glocker', 'Konstantinos Kamnitsas', 'Jacob Mandel', 'Fanny Morón', 'Snehal Prabhudesai', 'Joonsang Lee', 'Sevcan Turk', 'Toshio Moritani', 'Ota Yoshiaki', 'Nicholas Wang', 'Arvind Rao', 'J Rajiv Bapuraj', 'Ashok Srinivasan', 'Murat AK', 'Linmin Pei', 'Rivka R Colen', 'Kristin Waite', 'Vachan Vadmal', 'Andrew E Sloan', 'Chaitra Badve', 'David Ryan Ormond', 'John Thompson', 'Rourke Haas', 'Eric Fu', 'Derrick Murcia', 'Roland Wiest', 'Raphael Meier', 'Piotr Radojewski', 'Johannes Slotboom', 'Richard McKinley', 'Yvonne W Lui', 'Matthew Lee', 'Rajan Jain', 'Marie Metz', 'Ivan Ezhov', 'Florian Kofler', 'Benedikt Wiestler', 'Daniel Marcus', 'Pamela Lamontagne', 'Diego R Lucio', 'David Menotti', 'Flávia Sprenger', 'Bernardo C A Teixeira', 'Samuel B Martins', 'Alexandre Xavier Falcão', 'Marc Agzarian', 'Chee Chong', 'Sargam Bhardwaj', 'Minh-Son To', 'Chencai Wang', 'Akifumi Hagiwara', 'Talia Oughourlian', 'Catalina Raymond', 'Timothy F Cloughesy', 'Benjamin M Ellingson', 'Benjamin C Wagner', 'James Holcomb', 'Divya Reddy', 'Marco C Pinho', 'Chandan Ganesh Bangalore Yogananda', 'Joseph A Maldjian', 'Joaquim Farinhas', 'J Ross Mitchell', 'Michael A Vogelbaum', 'Václav Vybíhal', 'Marek Dostál', 'Tereza Kopřivová', 'Miloš Keřkovský', 'Petr Matula', 'Jan Michálek', 'Filip Lux', 'Michal Kozubek', 'Qi Dou', 'Pheng Ann Heng', 'Cheng Chen', 'Tiffany Y So', 'Meirui Jiang', 'Archana Venkataraman', 'Craig K Jones', 'Haris I Sair', 'Adam P Dicker', 'Adam E Flanders', 'Joshua D Palmer', 'Joseph Lombardo', 'Gregory S Alexander', 'Spencer Liem', 'Gaurav Shukla', 'Stephan Meckel', 'Andreas Haunschmidt', 'Georg Necker', 'Josef Pichler', 'Jhon Gómez', 'Jose Bernal', 'Ana Abello', 'William Escobar', 'Maria Trujillo', 'Alejandro Herrera-Trujillo', 'Jonas Teuwen', 'Regina G H Beets-Tan', 'Sean Benson', 'Mark Muzi', 'Abhishek Mahajan', 'Ho Sung Kim', 'Jihye Yun', 'Ji Eun Park', 'Mauricio Reyes', 'Mohamed Qayati', 'Mahmoud Mekhaimar', 'Shady Gamal', 'Ahmed Abbassy', 'Michael Hill', 'Kenneth Kolodziej', 'Aly Abayazeed', 'Carmen Dragos', 'Haris Shuaib', 'Marc Modat', 'Alysha Chelliah', 'Thomas C Booth', 'Subha Madhavan', 'Anas Belouali', 'Camelia Bencheqroun', 'Anousheh Sayah', 'Krithika Bhuvaneshwar', 'Yuriy Gusev', 'Reid C Thompson', 'Johannes Trenkler', 'Josep Puig', 'Jaume Capellades', 'Ken Chang', 'Raymond Y Huang', 'Kavi Fatania', 'Russell Frood', 'Stuart Currie', 'Robert Jeraj', 'Matthew Larson', 'John Garrett', 'Jitender Saini', 'Umang Pandey', 'Manali Jadhav', 'Madhura Ingalhalikar', 'Soonmee Cha', 'Javier Villanueva-Meyer', 'Jeffrey Rudie', 'Evan Calabrese', 'Wolfgang Wick', 'Martin Bendszus', 'Maximilian Zenk', 'Klaus Maier-Hein', 'Felix Sahm', 'Chandrakanth J Preetha', 'Gianluca Brugnara', 'Philipp Vollmuth', 'Suyash Mohan', 'Michel Bilello', 'Satyam Ghodasara', 'Chiharu Sako', 'Christos Davatzikos', 'Deepthi Karkada', 'Alexey Gruzdev', 'Patrick Foley', 'G Anthony Reina', 'Shih-han Wang', 'Micah Sheller', 'Brandon Edwards', 'Ujjwal Baid', 'Sarthak Pati'] | 2022-04-22 | null | null | null | null | ['boundary-detection'] | ['computer-vision'] | [ 1.96317539e-01 2.87374616e-01 -5.16920745e-01 -2.63304204e-01
-1.81028509e+00 -3.95583600e-01 1.36614978e-01 4.80315328e-01
-7.52725482e-01 9.88485634e-01 6.89977765e-01 -7.08129168e-01
-1.78589836e-01 -3.49142075e-01 -4.77464139e-01 -7.34108269e-01
-2.05377176e-01 6.73583746e-01 -2.40764558e-01 4.15977418e-01
-2.62728403e-03 7.16735542e-01 -6.82305932e-01 4.10542816e-01
8.25325966e-01 7.05073893e-01 4.39885288e-01 3.80950600e-01
1.61902264e-01 5.90902507e-01 -3.39461952e-01 1.58818960e-01
2.11734965e-01 -1.12640791e-01 -8.41425657e-01 5.94735667e-02
4.94270265e-01 -4.58550572e-01 -7.01391473e-02 5.30724049e-01
7.55650818e-01 -3.44257325e-01 5.85620463e-01 -8.87525678e-01
-4.33736980e-01 5.56177557e-01 -5.61471224e-01 2.01677874e-01
-8.65403190e-02 6.53519630e-01 8.76907885e-01 -5.05700707e-01
1.07886720e+00 4.70303118e-01 1.05782008e+00 5.43473125e-01
-1.13669074e+00 -8.27174664e-01 7.53436284e-03 -2.04069749e-01
-1.37810612e+00 -4.59965199e-01 6.11065179e-02 -7.77617514e-01
9.81881618e-01 3.00915331e-01 7.77985156e-01 1.14603209e+00
5.04520059e-01 4.88958240e-01 1.16816354e+00 -1.23126619e-01
2.87592113e-01 -2.99176484e-01 9.49147344e-02 7.75804937e-01
3.89104545e-01 -3.71401995e-01 -3.96672279e-01 -8.39193583e-01
6.42171800e-01 1.83073506e-01 -5.36814034e-01 -1.62408739e-01
-1.53148401e+00 8.33886802e-01 3.04298460e-01 1.12100489e-01
-5.22537291e-01 -1.56997740e-02 4.65969175e-01 -9.31912065e-02
5.92862368e-01 2.15372488e-01 -5.29087782e-01 -2.06818104e-01
-1.11817467e+00 4.24812734e-02 6.53357923e-01 5.64474404e-01
1.07476920e-01 -4.72177804e-01 1.48312524e-01 6.93636835e-01
1.65969059e-01 3.36907744e-01 7.71556318e-01 -9.15189981e-01
4.38743562e-01 4.89863634e-01 -5.85314855e-02 -5.34065366e-01
-1.19074464e+00 -6.69504523e-01 -9.37047005e-01 6.97256029e-02
6.37919784e-01 -5.14533281e-01 -7.27880955e-01 1.82002294e+00
1.52368873e-01 -4.29461673e-02 -1.16397068e-01 7.23729789e-01
6.37087524e-01 -2.45300516e-01 1.50443539e-01 -2.45884165e-01
1.43681300e+00 -5.92682362e-01 -3.77988815e-01 -1.34443641e-01
1.41362691e+00 -4.99834418e-01 8.48459899e-01 5.82198143e-01
-8.45774531e-01 3.64024758e-01 -5.02200365e-01 1.35575131e-01
-5.94791137e-02 9.60903466e-02 1.00905359e+00 5.88051796e-01
-1.27556562e+00 4.81427163e-02 -1.46054327e+00 -8.28091443e-01
1.14728642e+00 4.53456968e-01 -8.36229265e-01 -3.63503285e-02
-6.13660812e-01 1.08082771e+00 -3.19193788e-02 -1.10167433e-02
-9.74434853e-01 -1.29334021e+00 -6.14561200e-01 -2.99624652e-01
1.16405740e-01 -1.06054580e+00 1.15572941e+00 -7.10662901e-01
-7.10625827e-01 9.16584969e-01 -2.74713516e-01 -6.51442349e-01
6.91967785e-01 2.56428719e-01 -2.43751165e-02 9.69768092e-02
3.18986952e-01 6.41254425e-01 6.79789782e-02 -7.92788565e-01
-5.59294760e-01 -7.32596755e-01 -7.85789549e-01 3.54649276e-01
-3.63995194e-01 4.43318561e-02 6.33800775e-02 -4.78646129e-01
2.22796664e-01 -9.78677690e-01 -7.80929208e-01 1.98037207e-01
-8.31254050e-02 2.02440411e-01 4.58966017e-01 -1.18595660e+00
8.70720983e-01 -1.89731538e+00 -2.41459399e-01 6.53200075e-02
6.81732416e-01 -2.61872947e-01 1.13299221e-01 2.36553028e-01
-1.67681426e-01 3.33896160e-01 -4.17876601e-01 -4.08686876e-01
-5.27677059e-01 -1.77596763e-01 2.55979121e-01 9.96068776e-01
-6.67490363e-02 1.08528662e+00 -6.66809022e-01 -5.67766905e-01
-1.95667788e-01 1.28833920e-01 -6.90586269e-01 -3.50979716e-01
2.24457636e-01 4.88912970e-01 -4.06615168e-01 7.74398267e-01
3.84253651e-01 -5.59098005e-01 4.65122908e-01 2.21446343e-02
-1.43644601e-01 -9.25197378e-02 -7.39384770e-01 1.91913569e+00
-3.83120477e-01 5.28805435e-01 6.02611423e-01 -7.03201294e-01
5.66932261e-01 3.95397484e-01 1.10488677e+00 -1.85071200e-01
8.95056129e-02 4.30288166e-01 4.03147668e-01 -4.82777119e-01
-3.79629135e-02 -2.95289040e-01 1.01855017e-01 6.80894852e-01
-2.26761371e-01 -1.47483885e-01 -2.96760976e-01 2.52076030e-01
1.76539040e+00 -4.59363133e-01 3.95133495e-01 -4.25520271e-01
-1.85313612e-01 3.61513942e-01 6.47615910e-01 9.25147951e-01
-6.75944448e-01 8.09602618e-01 5.30416727e-01 -2.20873564e-01
-9.39746678e-01 -1.00235569e+00 -7.46612787e-01 5.27330279e-01
-6.73890531e-01 -1.60813987e-01 -4.46040660e-01 -7.13217735e-01
3.09936762e-01 4.45897907e-01 -6.30840600e-01 1.68080285e-01
-2.53171861e-01 -1.22495162e+00 7.63208687e-01 5.10141075e-01
8.70069340e-02 -5.37074804e-01 -7.28852570e-01 3.46482575e-01
-4.46172029e-01 -1.04809368e+00 -4.29389924e-01 3.29903215e-01
-1.03169584e+00 -1.17140567e+00 -1.05485451e+00 -5.17901301e-01
8.07627499e-01 -5.33999875e-02 5.65240026e-01 -1.42318383e-01
-7.58524835e-01 4.59035367e-01 3.28420140e-02 -5.48058927e-01
-3.06976557e-01 1.36520803e-01 8.72308612e-02 -3.91459614e-01
2.80944616e-01 -2.40534127e-01 -7.23215401e-01 9.89649594e-02
-6.49976611e-01 2.81608343e-01 7.86012292e-01 1.02994382e+00
5.51800370e-01 -4.19756353e-01 9.24634695e-01 -1.04198241e+00
6.39609814e-01 -8.23301017e-01 -3.12937319e-01 1.12739421e-01
-6.62543416e-01 -3.52096945e-01 2.45216310e-01 -1.22499451e-01
-9.55280900e-01 -5.90255624e-03 1.00732461e-01 6.11774437e-02
-1.85329571e-01 9.10970986e-01 5.27148128e-01 -2.35298187e-01
8.87118280e-01 -1.29534781e-01 6.78367317e-01 -2.44511023e-01
-7.70018473e-02 8.99347007e-01 2.45530725e-01 -5.73516607e-01
1.83643684e-01 8.32558990e-01 1.56561300e-01 -7.07519174e-01
-5.04851758e-01 -6.78677320e-01 -4.32838142e-01 -6.13282435e-03
7.06087232e-01 -1.08457303e+00 -5.80868661e-01 5.58861494e-01
-6.37301803e-01 -7.44030654e-01 -1.97368249e-01 8.97637546e-01
-5.16968489e-01 9.95993689e-02 -7.85364211e-01 -4.61459547e-01
-6.02083802e-01 -1.47077572e+00 1.10452628e+00 -1.13032460e-01
-5.23664355e-01 -9.72217500e-01 1.54752046e-01 7.00930834e-01
7.08062291e-01 5.00552118e-01 9.01507437e-01 -8.86299312e-01
-4.45796192e-01 -2.79978931e-01 -2.68173128e-01 -8.18188041e-02
5.61838508e-01 -6.51155412e-02 -7.99019158e-01 -3.80001128e-01
-2.13036776e-01 -5.74729443e-01 7.26524651e-01 7.81835914e-01
1.25657737e+00 1.09236456e-01 -7.16699064e-01 5.39557099e-01
1.21439123e+00 -1.63442511e-02 2.23698035e-01 4.19114530e-01
4.20821220e-01 2.18202487e-01 1.25693396e-01 6.31064832e-01
6.61219776e-01 3.99193674e-01 1.95916176e-01 -3.55555952e-01
-2.09143817e-01 1.70097828e-01 -1.15207762e-01 6.55322373e-01
1.93594098e-02 2.80534029e-01 -1.62620723e+00 6.28657818e-01
-1.69919968e+00 -6.29992425e-01 -3.55796819e-03 2.35864329e+00
9.52294588e-01 1.23003066e-01 5.69936186e-02 -5.63126624e-01
4.63618129e-01 -2.19297349e-01 -7.30871141e-01 -3.41352560e-02
-1.71777651e-01 1.52661338e-01 8.15965414e-01 2.73380280e-01
-7.73824513e-01 4.19286847e-01 7.07916784e+00 5.96696377e-01
-1.18535042e+00 5.33191919e-01 1.18666160e+00 -6.14376247e-01
-1.63690060e-01 -8.48684460e-03 -5.83150029e-01 1.47420630e-01
9.62035239e-01 -4.09036726e-01 1.88225001e-01 6.44174278e-01
7.26109564e-01 -3.87392759e-01 -1.02562237e+00 8.69900823e-01
7.42636397e-02 -1.69272614e+00 -4.42799270e-01 6.11812472e-01
7.57862210e-01 7.47182786e-01 -4.22101356e-02 3.94930504e-03
3.38754714e-01 -1.34726429e+00 3.59683394e-01 7.99988806e-01
9.41682994e-01 -4.59559590e-01 8.83524895e-01 3.85947466e-01
-6.31606519e-01 -3.53557318e-02 5.25000645e-03 8.20404440e-02
1.18134074e-01 8.01853776e-01 -1.41834068e+00 5.13462305e-01
5.48346698e-01 5.68578839e-01 -6.26857519e-01 1.23152077e+00
5.82849562e-01 8.13740373e-01 -5.54298460e-01 1.29464179e-01
1.49274677e-01 1.21092334e-01 2.21725434e-01 1.04149890e+00
2.27298573e-01 1.23036504e-01 2.21949704e-02 5.57962477e-01
-2.15468593e-02 3.05267215e-01 -6.43896580e-01 -3.32718045e-02
4.32527274e-01 1.53967321e+00 -7.68991172e-01 5.38932793e-02
-6.70756459e-01 2.91664064e-01 6.64583147e-01 2.14751437e-01
-4.94092703e-01 4.70922329e-02 4.61357594e-01 2.01382086e-01
-4.13282067e-01 -3.04260075e-01 -7.00777233e-01 -9.59637880e-01
-9.82935876e-02 -1.09308493e+00 8.39340270e-01 -5.81696093e-01
-1.38033521e+00 2.43103936e-01 -2.81922907e-01 -8.65268171e-01
-2.98968423e-02 -5.89455962e-01 -2.49175519e-01 1.13013625e+00
-1.23263371e+00 -1.09629500e+00 -2.96748698e-01 4.37018752e-01
1.02498099e-01 -6.07773103e-02 1.08348703e+00 3.61322835e-02
-6.67639375e-01 7.01069474e-01 1.93235576e-01 -3.33504677e-02
9.89744961e-01 -7.79128373e-01 -2.67210960e-01 2.87514299e-01
-4.42229539e-01 6.82035565e-01 6.46848604e-02 -8.10013831e-01
-1.34701443e+00 -9.82068181e-01 5.45944214e-01 -6.65572643e-01
9.66525853e-01 -1.73243716e-01 -6.27255559e-01 1.10219550e+00
-6.33482337e-02 -6.90519577e-03 1.12274802e+00 3.99418086e-01
1.28633112e-01 2.57682614e-02 -1.38962770e+00 6.10369444e-01
9.09805298e-01 -3.42264086e-01 -2.99853057e-01 7.06794202e-01
3.64190578e-01 -4.94872868e-01 -1.43367767e+00 5.38572550e-01
4.55722421e-01 -6.30104542e-01 4.56641972e-01 -6.29960895e-01
2.42288381e-01 1.32685855e-01 -1.90444797e-01 -1.20290375e+00
-2.40885302e-01 -4.35504317e-01 3.69375706e-01 8.12583387e-01
8.83028626e-01 -1.15462387e+00 1.05564368e+00 1.11638546e+00
-4.47719723e-01 -1.26160371e+00 -1.17442358e+00 -4.02631670e-01
5.99840999e-01 -5.53050876e-01 3.56582463e-01 1.06689847e+00
3.99487346e-01 -4.67972308e-01 3.35808188e-01 1.45134926e-01
5.69656134e-01 -1.27293929e-01 5.47567546e-01 -8.65067482e-01
-2.94049591e-01 -6.48787081e-01 -4.59055930e-01 -2.71838367e-01
1.24257267e-01 -1.28989387e+00 -2.76054829e-01 -1.71228862e+00
8.94582987e-01 -8.84159088e-01 -2.32767284e-01 8.99894536e-01
-4.33568433e-02 -4.41971831e-02 -2.09112301e-01 3.44406724e-01
-2.12522596e-01 2.11225167e-01 1.29346442e+00 -5.13388123e-03
8.08408111e-02 -3.15214366e-01 -1.07993591e+00 6.14792109e-01
5.94211996e-01 -3.91310513e-01 -1.60414547e-01 -3.77802461e-01
-3.42000276e-01 2.42200896e-01 4.30887461e-01 -8.92133057e-01
4.14128155e-01 -4.20751393e-01 5.97820461e-01 -2.53792435e-01
2.89448965e-02 -5.73603630e-01 4.27163154e-01 6.75410390e-01
-2.61896640e-01 7.33907893e-02 3.67895663e-01 2.45150790e-01
8.87312889e-02 -2.45240033e-02 5.42754769e-01 -3.46094579e-01
-2.13883236e-01 5.21022022e-01 -4.69326407e-01 2.10634381e-01
1.25285006e+00 -5.80103584e-02 -7.51839578e-01 -2.83896208e-01
-8.54485869e-01 1.98469996e-01 5.82057655e-01 -4.67228815e-02
3.07040960e-01 -1.02329445e+00 -1.14544499e+00 -5.59266508e-02
1.50138780e-01 2.63460338e-01 3.92107576e-01 1.46953940e+00
-6.23580575e-01 6.47656798e-01 5.13234250e-02 -5.95682085e-01
-9.45424080e-01 6.08510263e-02 5.57862222e-01 -4.04104233e-01
-9.93292689e-01 6.45594478e-01 2.13531017e-01 -5.88206530e-01
1.67638391e-01 -3.35235506e-01 4.52415407e-01 -3.51660177e-02
4.11975563e-01 8.78435597e-02 4.84726518e-01 -2.86674947e-01
-6.19000614e-01 -1.29103258e-01 -2.94987142e-01 -3.39901954e-01
1.46585894e+00 1.55775890e-01 -7.95840919e-02 2.85228193e-01
1.14532065e+00 6.18811287e-02 -1.08355629e+00 6.17348440e-02
-1.33393586e-01 -1.76961590e-02 2.45973513e-01 -8.99048865e-01
-9.23115730e-01 1.68106273e-01 5.71220994e-01 -4.02865559e-01
9.86810982e-01 9.25416723e-02 5.40739536e-01 9.74658430e-02
6.50982618e-01 -7.18201697e-01 -3.69675577e-01 1.24589540e-01
7.99872041e-01 -1.13291752e+00 7.28289783e-02 7.77376490e-03
-6.55687571e-01 8.39225233e-01 4.34875131e-01 2.58176297e-01
8.17998111e-01 6.83714390e-01 2.75258899e-01 -2.66859859e-01
-1.05721283e+00 3.03182244e-01 -1.98384255e-01 5.71283281e-01
6.13701642e-01 3.18699419e-01 -5.59144795e-01 8.47977698e-01
-1.94128275e-01 4.11625117e-01 7.08805919e-01 1.16943526e+00
-1.75208881e-01 -9.81488705e-01 -3.70512664e-01 1.32194281e+00
-4.97908533e-01 -3.45006794e-01 -1.13174766e-01 8.88285100e-01
-1.48656860e-01 6.67845190e-01 -5.55102415e-02 2.39729702e-01
9.42001343e-02 1.57730579e-01 4.06185627e-01 -7.01166093e-01
-4.89280432e-01 1.91931240e-02 2.28763297e-01 -4.54673320e-01
-5.38901538e-02 -1.23764586e+00 -1.33495319e+00 -8.40564892e-02
-2.28663668e-01 -1.63823083e-01 9.21468198e-01 9.34824824e-01
6.48033082e-01 5.58281779e-01 2.37810209e-01 -5.05187750e-01
-7.46377647e-01 -8.71478021e-01 -6.07356668e-01 1.65373534e-01
1.69101998e-01 -5.13738453e-01 -4.10534561e-01 -2.07887292e-01] | [14.943219184875488, -2.666996717453003] |
26037ead-6d1f-4a2e-9c8a-ae7eacdfe7a6 | photoelectric-factor-prediction-using | 2206.08950 | null | https://arxiv.org/abs/2206.08950v1 | https://arxiv.org/pdf/2206.08950v1.pdf | Photoelectric Factor Prediction Using Automated Learning and Uncertainty Quantification | The photoelectric factor (PEF) is an important well logging tool to distinguish between different types of reservoir rocks because PEF measurement is sensitive to elements with high atomic number. Furthermore, the ratio of rock minerals could be determined by combining PEF log with other well logs. However, PEF log could be missing in some cases such as in old well logs and wells drilled with barite-based mud. Therefore, developing models for estimating missing PEF log is essential in those circumstances. In this work, we developed various machine learning models to predict PEF values using the following well logs as inputs: bulk density (RHOB), neutron porosity (NPHI), gamma ray (GR), compressional and shear velocity. The predictions of PEF values using adaptive-network-fuzzy inference system (ANFIS) and artificial neural network (ANN) models have errors of about 16% and 14% average absolute percentage error (AAPE) in the testing dataset, respectively. Thus, a different approach was proposed that is based on the concept of automated machine learning. It works by automatically searching for the optimal model type and optimizes its hyperparameters for the dataset under investigation. This approach selected a Gaussian process regression (GPR) model for accurate estimation of PEF values. The developed GPR model decreases the AAPE of the predicted PEF values in the testing dataset to about 10% AAPE. This error could be further decreased to about 2% by modeling the potential noise in the measurements using the GPR model. | ['Abdulazeez Abdulraheem', 'Salaheldin Elkatatny', 'Ahmed Farid Ibrahim', 'Khalid L. Alsamadony'] | 2022-06-17 | null | null | null | null | ['gpr', 'gpr'] | ['computer-vision', 'miscellaneous'] | [-2.02675536e-01 1.21980309e-01 2.77946115e-01 -2.19344541e-01
-2.56375432e-01 5.91343492e-02 3.97371233e-01 3.57100427e-01
-4.55561638e-01 1.04483223e+00 -1.70283571e-01 -1.59091458e-01
-4.67820108e-01 -1.41892898e+00 -5.15038311e-01 -7.66398489e-01
-4.41654474e-02 9.90935504e-01 5.73009253e-01 -1.26278639e-01
7.81467915e-01 6.94448113e-01 -1.66914809e+00 3.26189362e-02
1.00550532e+00 8.24216366e-01 5.59227586e-01 3.08162272e-01
-1.40483126e-01 7.98494518e-01 -5.52197754e-01 2.58851022e-01
6.73546642e-03 -1.44225225e-01 -4.67292517e-01 -5.50739706e-01
-6.79371715e-01 -4.46174532e-01 2.51573384e-01 9.68156576e-01
2.06936464e-01 1.34318754e-01 1.40181911e+00 -6.90148294e-01
-1.94985583e-01 8.41621399e-01 -5.07044673e-01 1.22807980e-01
1.02807255e-02 -1.36691704e-01 2.95761287e-01 -1.10534942e+00
-1.56751871e-01 1.09132075e+00 4.98425364e-01 1.25897110e-01
-8.18770587e-01 -6.89750075e-01 -7.77321339e-01 4.31143194e-01
-1.41584587e+00 3.84126268e-02 5.08995652e-01 -7.30056882e-01
1.03893828e+00 8.90868008e-02 9.44455504e-01 2.16010943e-01
4.26452339e-01 -1.62308328e-02 1.46949637e+00 -7.54521489e-01
4.29615617e-01 1.48221821e-01 2.20795915e-01 3.51025492e-01
7.45960295e-01 3.77371162e-01 -1.22516617e-01 -4.63865548e-02
8.83248866e-01 -2.91733354e-01 -1.66802838e-01 6.78723812e-01
-4.62652743e-01 8.30515385e-01 -7.98240826e-02 5.33879101e-01
-4.67619956e-01 2.72571854e-03 4.50721532e-02 1.28594667e-01
1.55649692e-01 5.46479166e-01 -3.24411124e-01 -3.14495087e-01
-8.94987941e-01 2.95687348e-01 1.17121112e+00 2.77968526e-01
8.84432018e-01 3.98957461e-01 2.97689587e-01 9.21207368e-01
8.31106961e-01 8.91562581e-01 6.88154101e-01 -5.07005334e-01
1.68152347e-01 4.93961006e-01 2.13573501e-01 -7.49741435e-01
-2.12122411e-01 2.96212226e-01 -5.83089232e-01 3.32101822e-01
3.78323317e-01 -1.85868517e-02 -9.09481108e-01 8.72254193e-01
6.09404966e-02 -7.36621916e-02 1.90701589e-01 5.49718797e-01
7.85340011e-01 1.01284122e+00 1.29576430e-01 -2.39440590e-01
1.19311500e+00 -2.44261846e-01 -6.81329250e-01 -6.89032525e-02
1.88434254e-02 -4.80648667e-01 9.36376214e-01 4.72669870e-01
-8.12041223e-01 -2.74933845e-01 -1.24319983e+00 6.43191338e-01
-2.05597922e-01 9.45549011e-02 2.14720637e-01 7.81023383e-01
-5.51939547e-01 1.11688888e+00 -1.22980189e+00 3.13553102e-02
-8.03071707e-02 4.48821723e-01 -8.64851102e-02 4.18460757e-01
-1.27743435e+00 1.16039538e+00 7.07949340e-01 5.10379255e-01
-8.36071312e-01 -5.96636176e-01 -5.31253219e-01 1.98243573e-01
-4.66472954e-02 6.37923703e-02 8.19848776e-01 -8.63822550e-02
-1.77021837e+00 1.22251280e-01 1.49640486e-01 -3.71596426e-01
3.57243598e-01 -2.63334900e-01 -3.80823672e-01 3.29007149e-01
-1.34830758e-01 -1.89763501e-01 4.86833006e-01 -1.23000431e+00
-3.05524528e-01 -1.99450284e-01 -6.85793579e-01 -3.33435655e-01
-1.11879654e-01 -1.01454154e-01 4.24136966e-01 -1.66153818e-01
3.88543874e-01 -4.42420393e-01 -2.43281260e-01 -6.36845589e-01
-7.55945742e-02 -1.91005975e-01 7.25938261e-01 -1.04063547e+00
9.76144910e-01 -1.71413994e+00 -5.37993908e-01 1.11160469e+00
-2.00157881e-01 2.68548787e-01 7.42552280e-01 3.72566402e-01
1.09269157e-01 2.27503970e-01 -5.85692286e-01 5.20520031e-01
-3.02553505e-01 3.31749529e-01 2.02900812e-01 2.53900588e-01
1.60564914e-01 2.00480565e-01 -7.13752270e-01 -5.26616514e-01
3.39881599e-01 4.27731693e-01 -6.50541782e-02 2.09891260e-01
2.11628079e-02 1.35441437e-01 -5.69952130e-01 8.54594052e-01
9.36959505e-01 1.94860831e-01 -5.89074604e-02 -2.72271633e-01
-3.74969363e-01 -5.60279079e-02 -1.43023944e+00 4.83993173e-01
-3.27231437e-01 8.66235271e-02 -8.79231170e-02 -1.16704297e+00
1.81695592e+00 3.76950413e-01 5.34449220e-01 -6.50298595e-01
2.76800036e-01 8.50524127e-01 1.55570701e-01 -9.05490637e-01
5.06397486e-01 -6.39711738e-01 4.22774762e-01 2.44438365e-01
-2.26618022e-01 -2.13331237e-01 2.45734498e-01 -3.97330850e-01
6.70852840e-01 -1.40973693e-02 1.38612077e-01 -7.74804235e-01
6.44165993e-01 -1.50957510e-01 3.61888468e-01 5.17606556e-01
3.30168784e-01 2.78640807e-01 3.33726674e-01 -1.43792599e-01
-1.22496402e+00 -9.72191036e-01 -5.39470375e-01 1.13716289e-01
1.88335180e-01 1.85547203e-01 -5.40916324e-01 2.22834066e-01
3.32627967e-02 6.87028527e-01 -2.62489200e-01 -9.86881554e-02
-6.75341666e-01 -1.40222049e+00 5.53789914e-01 7.51550257e-01
7.55709767e-01 -1.22598243e+00 -7.61742651e-01 5.72579086e-01
8.12125579e-02 -6.52880430e-01 7.40404844e-01 5.08210897e-01
-1.28269184e+00 -1.25951695e+00 -4.83972609e-01 -3.32082808e-01
4.14330363e-01 -5.92008293e-01 8.44300091e-01 2.05304325e-01
1.73672304e-01 -1.18796833e-01 -5.31598985e-01 -6.51011527e-01
-9.33196127e-01 -3.03981632e-01 -1.93865195e-01 -3.45987380e-01
5.99405110e-01 -5.20620108e-01 -4.00332630e-01 3.01159799e-01
-7.04101145e-01 -4.47335720e-01 5.24469495e-01 7.18919694e-01
6.13853455e-01 5.64467430e-01 8.11521351e-01 -8.89813066e-01
7.01938570e-01 -7.04100609e-01 -6.91960216e-01 2.27042705e-01
-1.05901766e+00 2.25068524e-01 6.27383471e-01 -2.30993658e-01
-1.28650737e+00 -5.30321538e-01 -3.72704744e-01 6.52363617e-03
-1.16655119e-01 8.84558678e-01 -1.67970851e-01 -6.26766905e-02
7.54047632e-01 7.34130070e-02 2.29087830e-01 -5.88048458e-01
-8.68841231e-01 9.07201290e-01 6.02653205e-01 -8.80594969e-01
6.41727626e-01 1.35600626e-01 3.87439162e-01 -9.85664070e-01
-1.86838657e-01 -1.66317701e-01 -3.25943530e-01 -4.19310629e-01
6.25033498e-01 -6.78926945e-01 -8.85497272e-01 8.07966471e-01
-6.30657971e-01 -2.78555691e-01 -1.08324245e-01 1.07690227e+00
-2.79000342e-01 5.57669103e-01 -7.02485204e-01 -1.28865039e+00
-6.62272990e-01 -9.87907112e-01 2.84848630e-01 5.69612265e-01
-1.46719608e-02 -8.84272158e-01 -4.75807525e-02 3.89449894e-02
5.07096767e-01 4.16551858e-01 1.00167418e+00 -7.26625085e-01
-1.98800355e-01 -2.91459113e-01 7.28344843e-02 6.43170357e-01
-1.47266919e-02 3.93220693e-01 -7.29101300e-01 2.60714114e-01
4.62253869e-01 2.55080499e-02 8.17324340e-01 5.20793855e-01
8.39706242e-01 8.78203940e-03 -1.49732932e-01 2.94935286e-01
1.96122146e+00 9.04404223e-01 1.16790450e+00 9.09143388e-01
2.43209660e-01 2.19229385e-01 8.70615423e-01 6.75915062e-01
-1.11948937e-01 1.54483151e-02 4.06606704e-01 5.04547894e-01
4.36568618e-01 -9.63749364e-02 3.90134275e-01 5.29104590e-01
-1.08283162e+00 4.48908471e-02 -1.04701984e+00 4.89600867e-01
-1.29824984e+00 -1.18039000e+00 -5.75663030e-01 2.33375812e+00
7.65419006e-01 2.23970398e-01 -6.91674277e-02 9.57942307e-01
8.24814916e-01 -5.36664546e-01 -3.12398765e-02 -5.84847689e-01
-6.62824363e-02 8.15232098e-01 7.60443449e-01 5.00544667e-01
-5.20436943e-01 2.81493753e-01 5.53175926e+00 6.01361275e-01
-1.18165696e+00 -3.05857331e-01 2.98800111e-01 5.80575764e-01
-3.54007304e-01 2.58517236e-01 -8.34031820e-01 8.60217988e-01
1.16767955e+00 1.82546466e-03 2.11477518e-01 6.59214497e-01
4.47816938e-01 -9.45821047e-01 -5.40055990e-01 6.14769638e-01
-3.65666837e-01 -8.28644991e-01 -2.69711226e-01 4.80616353e-02
2.46183768e-01 -2.67893553e-01 -5.87892592e-01 3.51433009e-02
3.81089491e-03 -1.00671124e+00 5.61740816e-01 1.01401806e+00
6.60312712e-01 -6.77204072e-01 1.35503614e+00 3.77402514e-01
-1.03011978e+00 -1.53122053e-01 -5.69549263e-01 -2.34521180e-01
3.69297624e-01 8.57599914e-01 -1.07314181e+00 5.34381747e-01
9.83397424e-01 2.22570062e-01 -3.71217936e-01 1.17175937e+00
-1.91065669e-01 8.60527039e-01 -1.14818037e+00 -3.64624977e-01
-2.18576044e-01 -7.79630899e-01 2.52603740e-01 8.64928424e-01
8.91110301e-01 1.43542573e-01 -5.32051325e-01 8.89623284e-01
6.20118201e-01 1.92824826e-01 -1.48216803e-02 5.78270219e-02
9.30764735e-01 6.74469054e-01 -8.15513432e-01 -3.30041498e-01
-3.34010608e-02 -9.05533582e-02 -3.34946275e-01 3.02964146e-03
-6.90395653e-01 -5.57604551e-01 -2.40327194e-01 7.67022967e-01
2.02900246e-02 1.27809703e-01 -4.70261306e-01 -6.86905563e-01
-2.13121139e-02 -2.37100214e-01 1.23508468e-01 -5.65256298e-01
-1.13996649e+00 3.26141745e-01 5.93470454e-01 -9.56082642e-01
-3.42289567e-01 -7.37317324e-01 -9.17510688e-01 1.07657862e+00
-1.34948158e+00 -5.88379085e-01 -5.52944243e-01 2.91712850e-01
-8.53210911e-02 -2.98670828e-01 6.57607138e-01 3.29579264e-01
-3.46739054e-01 2.30310217e-01 5.45701921e-01 2.78389901e-02
2.62292176e-01 -1.16294074e+00 -4.68319625e-01 5.35135686e-01
-6.68145716e-01 1.57396700e-02 1.00366676e+00 -9.67765212e-01
-8.16748977e-01 -6.35326564e-01 7.26792336e-01 3.85456324e-01
5.91545761e-01 3.71918529e-01 -1.34564257e+00 2.99372464e-01
-5.01471698e-01 -1.95790231e-01 3.95610869e-01 -3.96364480e-01
3.56992513e-01 -7.17873126e-02 -1.55820429e+00 -8.42348933e-02
1.05273779e-02 -2.46392358e-02 -6.35239720e-01 -1.71611860e-01
-2.90394366e-01 -3.05878967e-01 -1.66243684e+00 8.04739594e-01
7.70045996e-01 -1.05366683e+00 9.16508496e-01 1.18552677e-01
4.96432334e-01 -3.47467691e-01 -6.97110817e-02 -9.03883040e-01
-8.11298862e-02 2.97846198e-01 -9.33264792e-02 1.29464793e+00
5.03387570e-01 -6.83219731e-01 8.72210681e-01 7.82168508e-01
-2.12513432e-01 -6.86633587e-01 -8.31878841e-01 -7.18010604e-01
7.94404671e-02 -1.45654455e-01 6.82836890e-01 4.79455113e-01
-1.01607688e-01 -1.66478261e-01 -2.10756406e-01 1.47958338e-01
5.17898440e-01 -4.42682683e-01 3.43528181e-01 -1.65488100e+00
-3.86079818e-01 -9.46496353e-02 -4.51230913e-01 -1.76007420e-01
-1.15093999e-01 -4.46721762e-01 8.20879266e-02 -1.68696535e+00
-9.12851319e-02 -8.24290514e-01 5.51296212e-03 2.99233854e-01
3.28634083e-02 -3.98542434e-02 -7.26153851e-02 5.21806896e-01
7.62234569e-01 5.54105997e-01 1.08735764e+00 -4.85520577e-03
-5.43307602e-01 4.09784496e-01 -1.33038079e-02 8.18645954e-01
9.70012128e-01 -5.81913054e-01 -1.21576674e-01 -5.47870062e-02
4.28696930e-01 3.19631428e-01 2.57220685e-01 -1.27771854e+00
-5.93833067e-02 -3.72736514e-01 5.43234468e-01 -8.38258445e-01
1.02875613e-01 -1.10275161e+00 7.58619606e-01 7.48571515e-01
4.37409133e-01 -3.29598576e-01 -4.55789194e-02 3.64335835e-01
-5.11904955e-01 -1.11668611e+00 8.62567246e-01 -3.95900637e-01
-6.16530836e-01 -2.64534265e-01 -6.45166755e-01 -6.26185358e-01
9.41639662e-01 -9.24728751e-01 -2.97848880e-01 4.41564573e-03
-7.62448013e-01 -1.33644029e-01 1.67464077e-01 -4.96919721e-01
7.30004132e-01 -7.73089707e-01 -6.51751578e-01 3.03848416e-01
-3.90790105e-01 4.13981199e-01 3.29689294e-01 8.04393172e-01
-1.32031095e+00 -1.29163966e-01 -4.58427519e-01 -3.91584814e-01
-8.06021154e-01 -1.04097754e-01 5.42688847e-01 -3.46796751e-01
-3.54836762e-01 3.30333710e-01 -6.95001125e-01 1.10299550e-01
-5.41901708e-01 -3.05806309e-01 -8.60618591e-01 6.69201016e-02
3.82027209e-01 1.02119839e+00 2.90221721e-01 -4.55200285e-01
-1.51257053e-01 6.47535861e-01 2.93801844e-01 -5.84462322e-02
1.69695234e+00 2.35795155e-01 -4.05340701e-01 4.70256984e-01
7.04793632e-01 -1.92873050e-02 -8.62261355e-01 2.95635283e-01
3.11100364e-01 -2.88656503e-01 -1.49970964e-01 -6.33824944e-01
-9.06474650e-01 5.09752572e-01 4.82424647e-01 9.38809961e-02
1.11451006e+00 -5.17713428e-02 4.97411668e-01 5.17656982e-01
2.36658603e-01 -1.48552847e+00 -2.50286788e-01 5.15289068e-01
6.56245053e-01 -9.51609850e-01 2.66611964e-01 -3.88262242e-01
-3.04748863e-01 1.71335268e+00 7.33434916e-01 -4.10726786e-01
1.35065353e+00 7.86627233e-01 -2.59891123e-01 -2.31507391e-01
-2.16954663e-01 3.02228987e-01 -2.90029466e-01 5.55119812e-01
4.03870642e-01 1.16393276e-01 -8.73641491e-01 9.80597258e-01
-5.14681935e-01 3.48877490e-01 8.87020171e-01 1.13036311e+00
-1.13396084e+00 -9.26380694e-01 -9.71203625e-01 1.02503073e+00
-5.93010843e-01 1.67825654e-01 4.20365214e-01 9.09660816e-01
-4.19900492e-02 8.56360912e-01 3.76723081e-01 -3.52638006e-01
3.36779058e-01 1.80912782e-02 2.80021459e-01 -3.02052289e-01
-2.86785573e-01 2.56537080e-01 8.56890753e-02 2.50552356e-01
-8.73909593e-01 -4.71275300e-01 -1.75866222e+00 -3.74528706e-01
-6.30971849e-01 6.57882154e-01 8.02762151e-01 1.19155431e+00
-6.49936259e-01 1.72236010e-01 5.13609171e-01 -6.43842995e-01
-5.68489909e-01 -1.53593767e+00 -1.32526243e+00 6.16661720e-02
-3.69176120e-01 -1.03174078e+00 -7.98302352e-01 4.52646986e-02] | [6.297726631164551, 3.2250866889953613] |
92ddbde4-dab7-496b-b957-c11215279805 | mwe-as-wsd-solving-multiword-expression | 2303.06623 | null | https://arxiv.org/abs/2303.06623v1 | https://arxiv.org/pdf/2303.06623v1.pdf | MWE as WSD: Solving Multiword Expression Identification with Word Sense Disambiguation | Recent work in word sense disambiguation (WSD) utilizes encodings of the sense gloss (definition text), in addition to the input words and context, to improve performance. In this work we demonstrate that this approach can be adapted for use in multiword expression (MWE) identification by training a Bi-encoder model which uses gloss and context information to filter MWE candidates produced from a simple rule-based extraction pipeline. We achieve state-of-the-art results in MWE identification on the DiMSUM dataset, and competitive results on the PARSEME 1.1 English dataset using this method. Our model also retains most of its ability to perform WSD, demonstrating that a single model can successfully be applied to both of these tasks. Additionally, we experiment with applying Poly-encoder models to MWE identification and WSD, introducing a modified Poly-encoder architecture which outperforms the standard Poly-encoder on these tasks. | ['Jacob Hoffman', 'Joshua Tanner'] | 2023-03-12 | null | null | null | null | ['word-sense-disambiguation'] | ['natural-language-processing'] | [ 6.23597503e-01 7.01746196e-02 -1.54372126e-01 -4.27585602e-01
-1.15699232e+00 -8.15939903e-01 6.40005350e-01 3.57248515e-01
-1.14326465e+00 6.40087903e-01 4.86971498e-01 -4.01779741e-01
1.58014074e-02 -5.16023457e-01 -2.37334192e-01 -1.00760534e-01
4.34906036e-02 5.97462416e-01 2.09918007e-01 -7.87048042e-01
6.13719858e-02 1.36654347e-01 -1.55946445e+00 7.41867483e-01
5.70476770e-01 8.60350072e-01 3.78984809e-01 6.90601170e-01
-5.99854589e-01 5.98664403e-01 -8.21515858e-01 -5.07259786e-01
2.70236321e-02 6.27889931e-02 -1.22478950e+00 -4.65107530e-01
6.49574995e-01 1.41150886e-02 2.32384931e-02 7.72331059e-01
7.51606584e-01 1.02002338e-01 5.02098918e-01 -5.25110304e-01
-4.34761584e-01 9.55309987e-01 -1.80734470e-01 2.77412266e-01
6.02625549e-01 -3.31057727e-01 1.66841578e+00 -9.58089650e-01
9.73633707e-01 1.30892718e+00 7.22114325e-01 6.74153924e-01
-1.28700495e+00 -5.68568289e-01 -7.62170032e-02 1.02969415e-01
-1.44455957e+00 -6.31218791e-01 2.67733485e-01 5.26902489e-02
2.10300565e+00 2.63668537e-01 4.23989445e-01 1.20243883e+00
-1.08052395e-01 1.18121386e+00 1.07574975e+00 -9.46568608e-01
-3.27833071e-02 -1.93981737e-01 3.59864980e-01 7.86593497e-01
1.02325834e-01 -1.27909467e-01 -7.63014913e-01 -2.89397985e-01
-9.88829043e-03 -7.15146840e-01 -1.72206953e-01 2.24145085e-01
-9.26864207e-01 8.02137315e-01 -9.83169004e-02 7.00589180e-01
-1.79891467e-01 1.73815116e-01 6.93876088e-01 4.07160103e-01
6.00445986e-01 1.03698373e+00 -9.61692214e-01 -3.22622538e-01
-1.23514223e+00 3.40262532e-01 9.32144463e-01 8.22478831e-01
5.42075515e-01 -1.63076594e-01 -4.75991189e-01 1.16283464e+00
-7.92783648e-02 3.24974805e-01 8.75727713e-01 -7.86220491e-01
3.33775312e-01 5.43421745e-01 1.75484106e-01 -5.01633286e-01
-6.64059520e-01 -3.96098197e-01 2.22601946e-02 2.31169644e-04
-2.35163383e-02 -2.04353616e-01 -1.06959355e+00 1.91669607e+00
2.62258202e-02 1.75848067e-01 2.89764583e-01 4.61139828e-01
7.96887159e-01 4.14558172e-01 4.70835835e-01 -1.29258230e-01
1.59967577e+00 -4.86419648e-01 -7.52604842e-01 -6.05789185e-01
1.16898739e+00 -9.88039553e-01 9.71299827e-01 3.71988326e-01
-1.04329264e+00 -3.11653554e-01 -1.27319098e+00 -2.87843674e-01
-9.05531883e-01 3.80748749e-01 7.40309954e-01 5.57901025e-01
-1.02379346e+00 7.16781914e-01 -7.11597025e-01 -6.02130294e-01
2.36372009e-01 4.63295728e-01 -3.52051258e-01 -5.55351451e-02
-1.80307007e+00 1.33127046e+00 8.29479873e-01 -5.06493866e-01
-1.58439368e-01 -7.07062483e-01 -1.06755209e+00 -5.55721745e-02
2.46053740e-01 -7.98163891e-01 1.48890686e+00 -6.39294088e-01
-1.20837653e+00 1.19839454e+00 -3.55099767e-01 -6.53983355e-01
-2.37597227e-01 -5.30332148e-01 -8.47010016e-01 5.60914259e-03
3.04841161e-01 6.00135505e-01 4.99888450e-01 -8.63089859e-01
-1.00071633e+00 -2.54478455e-01 6.38150936e-03 2.65675604e-01
-4.41276342e-01 5.48707247e-01 -2.44113013e-01 -7.30711281e-01
-5.19010484e-01 -8.17936182e-01 -3.54331210e-02 -6.29624128e-01
-2.55965292e-01 -6.97879851e-01 6.06838286e-01 -6.59426272e-01
1.89035344e+00 -1.89032257e+00 5.07174013e-03 1.65190324e-01
-6.87415227e-02 7.97768116e-01 -4.86343235e-01 5.89879811e-01
-1.92306399e-01 4.95584458e-01 -3.31555665e-01 -8.67116868e-01
3.29946280e-01 6.72933757e-01 -1.50842801e-01 -2.45580435e-01
6.65730357e-01 8.76090169e-01 -1.17764628e+00 -4.81440037e-01
4.40893061e-02 3.32746416e-01 -2.85347193e-01 1.71143800e-01
-2.62002856e-01 -5.13519228e-01 -2.90404111e-01 4.81788337e-01
2.72071838e-01 1.82404168e-04 6.23798668e-01 -2.38609046e-01
-8.09159800e-02 9.00551975e-01 -1.22321332e+00 2.04254508e+00
-1.07724798e+00 5.41067243e-01 2.38693114e-02 -7.71906674e-01
5.47460496e-01 3.82957548e-01 4.24395740e-01 -6.51818216e-01
1.07527308e-01 6.86358690e-01 -9.00132582e-02 -2.75985181e-01
9.07925546e-01 -2.17271179e-01 -4.23477948e-01 3.51956099e-01
7.76810348e-01 -1.72102213e-01 6.64223611e-01 3.28604549e-01
1.42022836e+00 2.15856999e-01 6.98473811e-01 -3.73827279e-01
4.20044273e-01 1.50508583e-01 4.28489119e-01 7.59826183e-01
3.57568949e-01 2.22135425e-01 -4.97696847e-02 -1.85538143e-01
-7.32375622e-01 -8.12392056e-01 -3.56374145e-01 1.50147378e+00
-2.59103686e-01 -1.18307269e+00 -5.73076963e-01 -8.06094885e-01
7.75939524e-02 1.25612473e+00 -6.15794659e-01 -1.88726131e-02
-8.50942433e-01 -9.11227703e-01 1.07298112e+00 5.10472357e-01
-1.29598211e-02 -8.58223855e-01 -6.99702144e-01 7.85389364e-01
-1.24604084e-01 -1.20545125e+00 -3.20464283e-01 8.63325059e-01
-1.01780951e-01 -9.59855914e-01 -1.20611191e-01 -7.18998373e-01
-2.20238075e-01 -1.50194764e-01 1.60938799e+00 2.03662962e-02
-5.24724901e-01 3.00636590e-01 -5.34905970e-01 -7.09977984e-01
-6.13206625e-01 5.94944656e-01 3.97054888e-02 -5.23693383e-01
1.09479582e+00 -3.40895146e-01 -1.68241173e-01 -1.95473820e-01
-1.12328899e+00 -2.67568499e-01 4.77299392e-01 1.15837836e+00
4.37470585e-01 -3.02884728e-01 5.31879067e-01 -1.21166635e+00
1.01112187e+00 -3.32974255e-01 -1.50578260e-01 4.17074561e-01
-9.32000339e-01 5.25027692e-01 3.43724996e-01 -8.43146741e-02
-1.06068277e+00 1.05981946e-01 -7.08856881e-01 -5.03412858e-02
-1.33782059e-01 6.62505448e-01 -5.64829148e-02 7.46300593e-02
7.31633306e-01 -1.22811571e-01 -3.61239821e-01 -6.76011562e-01
7.71801412e-01 9.85593259e-01 4.89321619e-01 -6.27325833e-01
2.32663706e-01 -4.33243141e-02 -1.04817919e-01 -5.35950303e-01
-1.25434625e+00 -8.32482517e-01 -5.82908571e-01 4.99744654e-01
8.61066461e-01 -9.32577550e-01 -1.65097833e-01 5.00961915e-02
-1.36244428e+00 2.17494406e-02 -2.95460910e-01 1.04750879e-01
-2.18094394e-01 3.69847149e-01 -6.02776349e-01 -7.36562610e-01
-6.31623507e-01 -8.08934450e-01 1.56760573e+00 -2.85635646e-02
-8.69036555e-01 -1.21444058e+00 1.84414655e-01 -2.68304646e-02
4.90071714e-01 -5.01765683e-02 1.02068162e+00 -1.18399858e+00
2.32900321e-01 -4.30802330e-02 7.76853487e-02 3.19269776e-01
1.24885611e-01 -3.80216807e-01 -1.05668020e+00 -5.34387380e-02
-5.15467823e-01 -7.18345821e-01 1.42023158e+00 -1.11674450e-01
7.50206411e-01 -1.28471598e-01 -5.43263674e-01 4.92817074e-01
1.50742567e+00 -8.66607428e-02 2.29656622e-01 5.12835562e-01
3.86584222e-01 3.01716268e-01 6.45395696e-01 3.64899278e-01
4.13751841e-01 7.35477209e-01 1.11593589e-01 -9.66451019e-02
-3.62228543e-01 -5.90724349e-02 3.52391422e-01 8.10859501e-01
2.02213973e-01 -4.62469250e-01 -9.80642438e-01 7.98199534e-01
-1.61711383e+00 -8.29753458e-01 1.34650335e-01 1.75175667e+00
1.46182895e+00 1.80825189e-01 -2.24178240e-01 1.03633352e-01
1.89067468e-01 4.60676670e-01 -6.67513385e-02 -1.05210280e+00
-5.59322059e-01 1.34100318e+00 7.80228138e-01 6.74385428e-01
-1.35322893e+00 1.51068389e+00 6.98409986e+00 1.27672350e+00
-7.74677455e-01 1.56885654e-01 3.51636857e-02 -2.19702516e-02
-4.76188570e-01 -1.15496822e-01 -1.11221814e+00 1.88630983e-01
1.13220632e+00 -4.02828231e-02 3.53546709e-01 6.74743593e-01
-4.62202489e-01 -3.95898856e-02 -1.05125797e+00 9.12861466e-01
2.28859752e-01 -1.27243793e+00 2.34356020e-02 -4.46087211e-01
3.88343960e-01 3.54554355e-01 -3.98683071e-01 5.11552453e-01
8.08752596e-01 -1.02246940e+00 3.98655951e-01 1.82336852e-01
1.11128676e+00 -8.31868589e-01 9.72469509e-01 -2.59496551e-02
-1.25022936e+00 3.26858647e-02 -2.87741363e-01 1.71515465e-01
1.46037266e-01 5.81949532e-01 -1.11991882e+00 8.49960327e-01
3.42999309e-01 6.65862322e-01 -6.07487082e-01 3.56562138e-01
-5.17478228e-01 7.01860785e-01 -5.01047611e-01 -2.09251240e-01
3.47190619e-01 4.00551379e-01 7.01721013e-01 2.17721677e+00
2.73136079e-01 -6.73906580e-02 1.98960081e-01 5.05627394e-01
-6.76727518e-02 3.18805426e-01 -3.40249717e-01 -2.05983445e-01
6.09087706e-01 1.29903805e+00 -1.44385859e-01 -6.32661521e-01
-3.80662233e-01 1.21547329e+00 7.16563523e-01 1.44143058e-02
-1.41188666e-01 -7.42461324e-01 1.10873139e+00 -4.31004703e-01
5.67824244e-01 -2.32317775e-01 -1.29898399e-01 -1.15938628e+00
-2.99212754e-01 -9.03672218e-01 8.30483615e-01 -6.26425564e-01
-1.51458108e+00 7.62647092e-01 -7.62552768e-02 -6.87786102e-01
-6.61280513e-01 -1.16990566e+00 -2.85003364e-01 1.09308434e+00
-1.86779249e+00 -1.25551212e+00 3.47748250e-01 1.99318901e-01
4.55182016e-01 -8.39336663e-02 1.67002380e+00 3.59161139e-01
-2.73848444e-01 7.20908046e-01 -6.69452474e-02 2.04307705e-01
9.35434163e-01 -1.93652487e+00 5.91810942e-01 7.57121384e-01
7.12982237e-01 9.73521709e-01 5.79036474e-01 -5.30763566e-01
-1.16957414e+00 -1.02216494e+00 1.59673166e+00 -6.09719038e-01
9.11743343e-01 -4.47395653e-01 -7.31499612e-01 4.77354854e-01
6.53382540e-01 -2.34098613e-01 1.22566545e+00 6.95571721e-01
-5.18494964e-01 2.91112870e-01 -8.59983027e-01 4.30209428e-01
1.42217422e+00 -7.52375960e-01 -1.21775138e+00 2.29311176e-02
8.12334120e-01 -5.46149254e-01 -1.04328215e+00 3.31716120e-01
5.74455738e-01 -3.81653845e-01 9.56126690e-01 -1.04947531e+00
3.08518976e-01 -1.19204670e-01 -4.13396090e-01 -1.89951062e+00
-2.79989034e-01 -7.08749831e-01 -2.85991192e-01 1.45874333e+00
6.98009312e-01 -2.32450217e-01 4.61983234e-02 2.72813022e-01
-2.66482979e-01 -8.19316149e-01 -1.09289479e+00 -6.64235473e-01
2.04847053e-01 -7.49025702e-01 7.60162354e-01 9.70617414e-01
3.73210013e-01 9.35787618e-01 4.40568198e-03 -1.61185935e-01
1.49231076e-01 2.12677658e-01 2.79710386e-02 -1.06829214e+00
-3.55715275e-01 -5.48660874e-01 -3.58904243e-01 -1.00691140e+00
6.31502569e-01 -1.29898036e+00 4.22541834e-02 -1.56729364e+00
-1.93005860e-01 -3.34005386e-01 -7.94798970e-01 9.25357163e-01
-5.98723054e-01 9.28714797e-02 1.30868107e-01 -3.64625484e-01
-6.52535558e-01 2.18980998e-01 4.76089239e-01 -1.20220765e-01
1.90443814e-01 -6.81820333e-01 -6.88283861e-01 5.14320731e-01
6.35308683e-01 -5.63372552e-01 -1.22760102e-01 -7.56907761e-01
4.87672895e-01 -3.88903677e-01 -2.52626121e-01 -6.78029954e-01
2.43595857e-02 9.01926085e-02 2.94382334e-01 -3.18318963e-01
2.94447243e-01 -3.81498873e-01 -5.24531841e-01 5.29960357e-02
-5.81776381e-01 2.13172421e-01 4.90969867e-01 1.93553686e-01
-3.05739790e-01 -4.96761113e-01 4.52317595e-01 -3.55022103e-01
-1.19051325e+00 -1.41694024e-01 -6.41240418e-01 5.47139347e-01
2.84394562e-01 1.87672570e-01 -1.71636060e-01 8.39600265e-02
-6.25654638e-01 2.26855054e-01 1.95797175e-01 5.94873071e-01
3.32564145e-01 -1.18739712e+00 -8.03139865e-01 6.74559474e-02
4.96132553e-01 -3.92619997e-01 -4.68366921e-01 2.92421699e-01
-1.58812031e-02 3.90690625e-01 9.31512937e-02 -1.24440296e-02
-1.45309913e+00 3.77781898e-01 2.65630126e-01 -7.66288579e-01
-1.62572581e-02 1.16973567e+00 -6.19801521e-01 -3.81970942e-01
-8.43365639e-02 -4.59061474e-01 -3.54851574e-01 3.10734838e-01
7.25296259e-01 -4.71156696e-03 5.19383371e-01 -4.93415475e-01
-6.96684718e-01 3.43205154e-01 -1.45984992e-01 -2.44479239e-01
1.39419496e+00 -1.58176303e-01 -6.45033121e-02 3.03586423e-01
1.35108936e+00 2.81929582e-01 -3.27798396e-01 -4.17822659e-01
7.82403111e-01 -1.65479720e-01 3.60342711e-01 -1.22911429e+00
-4.29105163e-01 7.82517970e-01 4.89956051e-01 -2.58224364e-02
1.06983554e+00 2.95357276e-02 1.20422959e+00 6.57115459e-01
3.63975376e-01 -1.57070172e+00 -3.59150380e-01 9.41787124e-01
5.10598719e-01 -1.19226372e+00 -2.00244248e-01 -2.97085553e-01
-5.05593181e-01 1.32112217e+00 1.62424713e-01 1.02524424e-03
5.61466038e-01 7.44104326e-01 9.33209881e-02 -2.52923846e-01
-9.99303877e-01 -9.53163803e-01 4.40598309e-01 4.29407805e-01
9.42760527e-01 1.91889763e-01 -8.46840858e-01 1.00616419e+00
-3.15400332e-01 2.55726911e-02 1.05432980e-01 1.07355297e+00
-3.78290594e-01 -1.85027611e+00 3.48435938e-01 5.88187695e-01
-9.12052572e-01 -8.45484614e-01 -7.01887250e-01 6.10948741e-01
5.46977401e-01 9.43128526e-01 -2.98030786e-02 -4.51900393e-01
3.63314122e-01 7.07029164e-01 5.94630718e-01 -1.03223872e+00
-9.06573355e-01 -1.32700950e-01 1.18329990e+00 -7.63071597e-01
-6.29024923e-01 -6.82552814e-01 -1.32260942e+00 2.64112175e-01
-3.54470462e-01 1.16031341e-01 5.82411349e-01 1.12318742e+00
4.56904471e-01 4.74399805e-01 2.57876903e-01 -2.16060027e-01
-6.27764344e-01 -1.14624774e+00 -5.19122422e-01 7.59659469e-01
1.15100473e-01 -7.14458048e-01 6.45301957e-03 -1.00592099e-01] | [10.356470108032227, 9.4617280960083] |
420d6065-f31e-4ba6-a44b-e96ada343249 | gafx-a-general-audio-feature-extractor | 2207.09145 | null | https://arxiv.org/abs/2207.09145v1 | https://arxiv.org/pdf/2207.09145v1.pdf | GAFX: A General Audio Feature eXtractor | Most machine learning models for audio tasks are dealing with a handcrafted feature, the spectrogram. However, it is still unknown whether the spectrogram could be replaced with deep learning based features. In this paper, we answer this question by comparing the different learnable neural networks extracting features with a successful spectrogram model and proposed a General Audio Feature eXtractor (GAFX) based on a dual U-Net (GAFX-U), ResNet (GAFX-R), and Attention (GAFX-A) modules. We design experiments to evaluate this model on the music genre classification task on the GTZAN dataset and perform a detailed ablation study of different configurations of our framework and our model GAFX-U, following the Audio Spectrogram Transformer (AST) classifier achieves competitive performance. | ['Xiaohu Zhu', 'Hanhaodi Zhang', 'Zhaoyang Bu'] | 2022-07-19 | null | null | null | null | ['genre-classification'] | ['computer-vision'] | [ 2.95067549e-01 -3.21468413e-02 3.34743142e-01 -2.02149227e-01
-9.37199235e-01 -6.44474864e-01 5.04862309e-01 -2.03349382e-01
-2.49639079e-01 5.66826701e-01 5.22943139e-01 -1.48803145e-01
-3.68311465e-01 -5.96135914e-01 -6.96588635e-01 -5.14042199e-01
-2.42602900e-01 8.69025569e-03 -5.40917516e-02 -2.43344635e-01
1.34044498e-01 2.06469044e-01 -1.74030256e+00 5.48534334e-01
1.87548965e-01 1.34719133e+00 4.58839573e-02 9.06337082e-01
3.12681437e-01 9.09134686e-01 -7.90836155e-01 -1.39445856e-01
5.41729391e-01 -4.49534655e-01 -1.13440764e+00 -3.96478862e-01
6.11666739e-01 -3.51762891e-01 -2.67379194e-01 6.01613462e-01
6.69080317e-01 4.73164976e-01 3.85189444e-01 -1.28597140e+00
-6.36809886e-01 1.35690343e+00 -7.09918439e-02 3.96404058e-01
4.04884517e-01 1.15776613e-01 1.54516268e+00 -7.24755228e-01
4.05811161e-01 1.14154351e+00 1.02419019e+00 3.19401711e-01
-1.05901468e+00 -6.28116369e-01 7.15330392e-02 5.32944679e-01
-1.27275348e+00 -4.26618904e-01 7.79906273e-01 -3.54010940e-01
1.34016442e+00 4.89380151e-01 6.81443095e-01 1.32189274e+00
1.80931166e-01 7.65004635e-01 9.32194650e-01 -5.67157209e-01
1.11007772e-01 -2.18909934e-01 1.10163122e-01 4.05595690e-01
-5.35828769e-01 3.16072583e-01 -8.68453205e-01 -1.39894873e-01
7.10765421e-01 -2.23144948e-01 -6.01458311e-01 1.09334983e-01
-1.03037655e+00 7.10066080e-01 4.88184184e-01 6.18270099e-01
-2.50072747e-01 4.87328947e-01 5.67366779e-01 8.60898137e-01
4.92032111e-01 1.07006264e+00 -8.79515946e-01 -5.21745503e-01
-9.73363757e-01 3.37135315e-01 5.55659771e-01 3.51438314e-01
5.37439644e-01 6.18777215e-01 -4.12621051e-01 9.57232416e-01
-2.13582695e-01 -5.46498224e-02 8.95889103e-01 -9.17797327e-01
5.54394862e-03 1.90249979e-01 -3.74355704e-01 -6.22003913e-01
-6.21732473e-01 -8.37608218e-01 -5.12831986e-01 1.17875822e-01
3.63565356e-01 -3.60114574e-01 -5.82193792e-01 1.80918205e+00
-1.02539062e-01 6.58617973e-01 -3.53543758e-01 1.01943672e+00
9.13473308e-01 5.40478051e-01 -1.89781934e-01 3.10500294e-01
1.06812894e+00 -1.07301879e+00 -4.20000046e-01 2.24991918e-01
3.76848996e-01 -8.78297091e-01 1.45329010e+00 6.72395408e-01
-9.65974033e-01 -9.69529688e-01 -1.06848419e+00 -2.71001697e-01
-3.50107640e-01 4.54411149e-01 5.85159898e-01 4.35357451e-01
-1.31988454e+00 1.35077620e+00 -5.22305906e-01 -1.89587817e-01
1.44077554e-01 5.06192565e-01 -4.44310188e-01 6.60135269e-01
-1.23662508e+00 6.19230330e-01 5.13639390e-01 -1.62935629e-01
-1.11105311e+00 -9.25310493e-01 -6.76405668e-01 7.13109493e-01
2.61549860e-01 -8.85334313e-01 1.54410374e+00 -1.15376246e+00
-2.14636493e+00 3.97846490e-01 3.04064423e-01 -8.17441583e-01
1.45591497e-01 -6.89122140e-01 -2.85823405e-01 -3.00899521e-03
-1.92191467e-01 4.54784751e-01 1.28581846e+00 -7.46962368e-01
-6.64383590e-01 7.95692031e-04 3.06184232e-01 -1.79167360e-01
-3.24117571e-01 -1.24858571e-02 5.44966877e-01 -1.10942757e+00
-2.55891919e-01 -9.45466280e-01 2.71823674e-01 -5.14789104e-01
-3.88806641e-01 -3.00086081e-01 6.89468086e-01 -5.99885881e-01
1.25419426e+00 -2.35009170e+00 2.18482777e-01 3.14909918e-03
5.91366775e-02 3.28298688e-01 -5.39371789e-01 5.25389433e-01
-6.85294688e-01 1.72020327e-02 -1.25535235e-01 -3.20044100e-01
8.37308392e-02 -1.95323661e-01 -7.65243649e-01 1.17958955e-01
3.45059365e-01 6.90322042e-01 -7.70446002e-01 4.04853314e-01
1.66460037e-01 4.78272051e-01 -9.27677095e-01 1.27348125e-01
-1.95138454e-01 2.48380601e-01 -1.32359505e-01 4.68278110e-01
2.35987395e-01 8.76650438e-02 -2.00144127e-02 -2.44305611e-01
-1.89051032e-01 7.98908174e-01 -9.86714721e-01 1.91216850e+00
-7.74580419e-01 7.88719773e-01 -1.36326686e-01 -7.80528545e-01
8.61073613e-01 6.81260109e-01 6.96661651e-01 -3.08827817e-01
2.99701303e-01 8.55614096e-02 2.80640751e-01 -2.22343206e-01
4.17143643e-01 -2.71037996e-01 1.32472776e-02 5.67120016e-01
9.84171391e-01 1.28755912e-01 -1.30699992e-01 -7.66206980e-02
1.58682227e+00 4.04455453e-01 2.64658213e-01 -3.19355607e-01
4.32415962e-01 -5.43834567e-01 1.15174986e-01 8.63343060e-01
-7.24355057e-02 9.54186559e-01 3.13557386e-01 -6.36668384e-01
-9.05324817e-01 -8.55577350e-01 -1.60791367e-01 1.57066894e+00
-5.84575832e-01 -1.04041076e+00 -8.08297932e-01 -6.42405808e-01
-4.25892835e-03 6.40642345e-01 -8.03917050e-01 -3.81354868e-01
-4.36723948e-01 -3.34752947e-01 7.77759314e-01 4.76991713e-01
3.48548502e-01 -1.52238190e+00 -6.81335330e-01 5.31518579e-01
-3.42067815e-02 -8.41048181e-01 -3.36336434e-01 6.31806374e-01
-4.70657825e-01 -9.32435572e-01 -3.57151657e-01 -4.03724998e-01
-4.63620573e-01 5.43622673e-02 1.17175543e+00 -2.46838436e-01
-2.04531178e-01 4.65128422e-01 -6.42292261e-01 -6.27370715e-01
-9.74426121e-02 6.34715378e-01 1.21669263e-01 2.30356649e-01
1.65085405e-01 -1.18021297e+00 -4.15625721e-01 -1.64634734e-01
-7.88255751e-01 -2.58987486e-01 1.97695464e-01 8.07945192e-01
3.04388911e-01 -2.30698660e-01 9.35009897e-01 -5.92948556e-01
8.30446780e-01 -3.15631419e-01 -8.53718370e-02 -2.01220602e-01
-1.66949809e-01 3.80030125e-02 9.57847834e-01 -3.49798977e-01
-6.84818864e-01 7.05850646e-02 -7.11115539e-01 -7.70016074e-01
-1.59844562e-01 5.12438536e-01 7.77012557e-02 2.33245984e-01
7.38628566e-01 3.51737030e-02 -3.72653872e-01 -9.75941658e-01
3.40137005e-01 6.90390408e-01 6.08384669e-01 -5.13465822e-01
5.89410305e-01 1.55621976e-01 -2.80376375e-01 -7.43457437e-01
-1.10445583e+00 -2.96881616e-01 -4.26780939e-01 -1.63531691e-01
6.31173432e-01 -8.33870828e-01 -7.45997250e-01 3.05516511e-01
-9.47684646e-01 -2.34954670e-01 -7.87618637e-01 6.69758737e-01
-8.64697397e-01 -1.06025159e-01 -7.11951613e-01 -7.10932732e-01
-5.14624119e-01 -9.11969960e-01 1.25469196e+00 1.74646415e-02
-5.00583231e-01 -7.54212797e-01 3.85378748e-01 -5.67048788e-02
6.02074087e-01 1.72957510e-01 9.63271976e-01 -9.23182309e-01
-2.89601028e-01 4.17493917e-02 8.40438679e-02 5.48491776e-01
2.69036412e-01 -7.34758154e-02 -1.88132071e+00 -2.35953957e-01
1.37856752e-01 -4.90065277e-01 1.25974596e+00 5.73959172e-01
1.55516160e+00 -3.44991326e-01 4.08064634e-01 8.39071631e-01
1.06981897e+00 1.94508627e-01 6.12961173e-01 3.83999497e-01
4.60392296e-01 -1.26498818e-01 5.66245690e-02 6.57068253e-01
-7.39675686e-02 7.98610210e-01 4.86926228e-01 2.05119923e-01
-3.84767860e-01 -3.58285159e-01 6.08134210e-01 1.02198255e+00
-5.32651007e-01 -2.07997322e-01 -4.87117290e-01 3.38602543e-01
-1.79355681e+00 -9.56784844e-01 4.47382629e-01 1.90682626e+00
7.03251302e-01 7.20307752e-02 3.60241055e-01 8.83558571e-01
3.78704786e-01 9.09110159e-02 -1.71043858e-01 -5.74274540e-01
1.49809215e-02 9.83294904e-01 -2.16042951e-01 3.13178509e-01
-1.35082603e+00 8.88780773e-01 6.58498716e+00 9.58851814e-01
-1.51579785e+00 3.20251822e-01 2.20366687e-01 -3.80175173e-01
-4.18366976e-02 -3.37015353e-02 -3.16117227e-01 2.43839070e-01
1.15003467e+00 1.57466471e-01 8.96688461e-01 8.81160796e-01
-1.21635884e-01 4.56276536e-01 -1.36018264e+00 1.04688966e+00
-1.26652256e-01 -1.30288112e+00 2.94052511e-01 -1.64837658e-01
2.81709164e-01 1.54769853e-01 8.19502547e-02 7.55855143e-01
2.12968588e-01 -1.05255306e+00 8.83629382e-01 5.76290965e-01
7.51831532e-01 -7.07605481e-01 5.14603972e-01 -2.28492171e-01
-1.15813625e+00 -3.49941790e-01 -2.61419505e-01 -4.74895090e-01
-1.69539139e-01 4.58658427e-01 -9.93189514e-01 8.36569548e-01
9.69350874e-01 9.43651795e-01 -5.72430551e-01 1.02734792e+00
-1.17253046e-02 1.17970467e+00 -2.86143214e-01 3.32971483e-01
3.80578130e-01 7.78428912e-02 9.60498631e-01 1.08097780e+00
6.12625241e-01 -4.27113861e-01 -1.12348571e-01 6.70098245e-01
-2.12227896e-01 2.13978827e-01 -2.92221934e-01 -1.68091878e-01
1.55554652e-01 1.46281064e+00 -4.86133963e-01 -1.42225400e-01
-9.66784805e-02 6.43898964e-01 3.82625371e-01 1.24804072e-01
-7.06611514e-01 -5.07495224e-01 7.94294477e-01 2.60919482e-01
3.54744285e-01 1.75420240e-01 4.43600453e-02 -1.22038937e+00
-1.33814290e-01 -1.03285766e+00 4.53763425e-01 -1.02088118e+00
-1.29197669e+00 1.04275179e+00 -3.91850531e-01 -1.35417020e+00
-4.90345061e-01 -5.98613501e-01 -6.98655128e-01 6.41633749e-01
-1.27059376e+00 -1.15205467e+00 4.32742536e-02 7.42349923e-01
5.48845887e-01 -5.95253408e-01 1.09537411e+00 3.99349481e-01
-2.83916086e-01 6.07849598e-01 -1.18473262e-01 5.91142513e-02
6.75842285e-01 -1.49416697e+00 3.87918651e-01 3.20499569e-01
8.09112012e-01 4.74309534e-01 5.01226604e-01 -7.39591494e-02
-9.80991960e-01 -1.14765692e+00 5.82220554e-01 -4.17563736e-01
8.41682196e-01 -3.70438099e-01 -9.03965652e-01 8.96298647e-01
4.59744126e-01 -5.58521338e-02 7.22930312e-01 4.87767458e-01
-4.79091287e-01 -1.31599098e-01 -6.77156985e-01 1.93274245e-01
1.08265960e+00 -8.91753852e-01 -5.74919283e-01 3.36004868e-02
6.98555112e-01 -1.18589789e-01 -9.81602609e-01 4.34519291e-01
7.02547550e-01 -1.07784009e+00 8.66449654e-01 -7.92688966e-01
4.90686148e-01 -1.11686543e-01 -3.02151740e-01 -1.75619042e+00
-4.94107872e-01 -8.35690260e-01 -1.51343584e-01 1.30977535e+00
1.94451600e-01 -4.80330527e-01 1.44545197e-01 -2.69058824e-01
-7.00184166e-01 -6.93872929e-01 -1.35304797e+00 -7.87542820e-01
-1.41134961e-02 -6.86167240e-01 8.51239085e-01 1.17326105e+00
1.16245553e-01 8.03513765e-01 -6.22084618e-01 -3.52853060e-01
-1.82795107e-01 2.08870500e-01 6.74268544e-01 -1.42188001e+00
-8.20996404e-01 -6.31129444e-01 -6.77656651e-01 -4.53760922e-01
2.26146832e-01 -1.08531356e+00 -3.05070698e-01 -1.06309128e+00
-8.10359195e-02 -8.72478336e-02 -7.61303484e-01 7.94294596e-01
2.06023697e-02 3.58872920e-01 5.09556949e-01 -1.66944221e-01
-3.88736874e-01 6.92796886e-01 9.54813361e-01 -2.06656724e-01
-3.31829011e-01 7.34805465e-02 -5.77103317e-01 8.35233748e-01
8.31654251e-01 -4.38713670e-01 -3.70754570e-01 -1.81041658e-01
3.04418743e-01 5.07630780e-02 4.21262085e-01 -1.44938934e+00
-2.95052052e-01 4.74956453e-01 3.91376078e-01 -1.87272668e-01
6.34296417e-01 -5.76408684e-01 5.95553331e-02 8.83710235e-02
-5.07031858e-01 -7.46040940e-02 3.32559794e-01 6.32766932e-02
-4.75936979e-01 -2.05349341e-01 4.37851042e-01 -1.46345387e-03
-2.62703210e-01 7.23705292e-02 -3.70430708e-01 -2.10539922e-01
3.55802655e-01 9.69369933e-02 -2.14387089e-01 -6.36948466e-01
-1.17816007e+00 -6.09934807e-01 -2.58335024e-01 7.39088118e-01
3.79491657e-01 -1.41217899e+00 -8.26306760e-01 3.30947071e-01
-2.00736582e-01 -5.63024640e-01 2.15449885e-01 4.51397061e-01
-1.06185630e-01 4.40812886e-01 -3.79624307e-01 -3.92226785e-01
-1.00638282e+00 4.74733412e-01 4.87137675e-01 -4.03614640e-01
-6.34676814e-01 9.03855979e-01 8.91558081e-02 -4.31832254e-01
1.73457205e-01 -6.74340725e-01 -2.88505286e-01 1.87802181e-01
5.36706865e-01 4.17742491e-01 4.99362141e-01 -4.45700347e-01
-8.90827179e-02 3.20021212e-01 3.02520879e-02 2.10199901e-03
1.75185215e+00 1.95344031e-01 2.40992963e-01 7.14962423e-01
1.06305313e+00 1.18583560e-01 -1.05197990e+00 -1.31181493e-01
-8.23836327e-02 -1.56599119e-01 2.58145422e-01 -9.55750704e-01
-1.25555313e+00 8.97870183e-01 7.85612047e-01 6.45633638e-01
1.36321759e+00 -1.53839737e-01 7.36856163e-01 3.79954696e-01
2.95052707e-01 -8.23566914e-01 8.98879394e-03 7.67511249e-01
1.23555374e+00 -6.11967564e-01 -1.40313059e-01 1.54810902e-02
-4.25154239e-01 1.30753505e+00 2.96096087e-01 -6.05378151e-01
9.93404448e-01 2.78025955e-01 6.50929660e-02 6.71704262e-02
-9.37704444e-01 -4.10322696e-01 5.79990864e-01 2.24396780e-01
7.50491500e-01 -9.52379555e-02 8.57357308e-02 1.24428523e+00
-1.07802033e+00 1.49216250e-01 4.67011392e-01 8.34506929e-01
-2.08498359e-01 -1.10554266e+00 -2.17558265e-01 4.90622580e-01
-7.88065255e-01 -2.91891456e-01 -5.83464086e-01 6.58991277e-01
3.98126096e-01 8.31253886e-01 9.53945816e-02 -9.80335474e-01
1.96975172e-01 5.37044406e-01 6.56490564e-01 -6.49566233e-01
-1.50213873e+00 3.13904911e-01 1.18164215e-02 -5.66992939e-01
-4.14458960e-01 -4.39963162e-01 -8.65305305e-01 5.22846021e-02
-3.75067949e-01 1.11755386e-01 5.36081851e-01 9.15920496e-01
4.20813292e-01 1.19981694e+00 6.91213310e-01 -1.08695006e+00
-4.18619603e-01 -1.37736440e+00 -8.07024837e-01 3.24701965e-01
5.70059538e-01 -6.84216857e-01 -2.45390996e-01 1.20653935e-01] | [15.594429969787598, 5.24717378616333] |
8592102e-2020-4db8-9012-ecc86bea5323 | bloom-library-multimodal-datasets-in-300 | 2210.14712 | null | https://arxiv.org/abs/2210.14712v1 | https://arxiv.org/pdf/2210.14712v1.pdf | Bloom Library: Multimodal Datasets in 300+ Languages for a Variety of Downstream Tasks | We present Bloom Library, a linguistically diverse set of multimodal and multilingual datasets for language modeling, image captioning, visual storytelling, and speech synthesis/recognition. These datasets represent either the most, or among the most, multilingual datasets for each of the included downstream tasks. In total, the initial release of the Bloom Library datasets covers 363 languages across 32 language families. We train downstream task models for various languages represented in the data, showing the viability of the data for future work in low-resource, multimodal NLP and establishing the first known baselines for these downstream tasks in certain languages (e.g., Bisu [bzi], with an estimated population of 700 users). Some of these first-of-their-kind baselines are comparable to state-of-the-art performance for higher-resourced languages. The Bloom Library datasets are released under Creative Commons licenses on the Hugging Face datasets hub to catalyze more linguistically diverse research in the included downstream tasks. | ['Daniel Whitenack', 'Abraham Owodunni', 'Anna Filighera', 'Jacob Mansdorfer', 'Joshua Nemecek', 'Colin Leong'] | 2022-10-26 | null | null | null | null | ['visual-storytelling'] | ['natural-language-processing'] | [-6.65846542e-02 1.11910403e-01 -6.08565271e-01 -2.14193210e-01
-1.26393235e+00 -9.05433834e-01 1.05081427e+00 -2.66066402e-01
-3.52170885e-01 6.81234181e-01 8.69115233e-01 -4.53705549e-01
5.38645089e-01 -2.23952815e-01 -7.69069135e-01 -3.82077366e-01
1.22546710e-01 8.98840785e-01 -3.38574558e-01 -3.87511700e-01
-9.76822227e-02 2.91439950e-01 -1.44764483e+00 9.27394807e-01
4.86720324e-01 7.36257136e-01 1.80835828e-01 5.88904142e-01
-2.78364033e-01 6.82493865e-01 -2.32433185e-01 -9.37714875e-01
-2.31110692e-01 -2.11521074e-01 -8.60507369e-01 4.50766087e-02
7.69031882e-01 -2.44087517e-01 -3.88959855e-01 4.08247292e-01
8.44667733e-01 -2.23212972e-01 7.09133923e-01 -1.22356725e+00
-1.33467388e+00 9.09070492e-01 -5.33737183e-01 -1.63560167e-01
7.89161265e-01 3.69347006e-01 1.09551835e+00 -1.36468947e+00
1.09349859e+00 1.83861005e+00 3.41637999e-01 7.06996918e-01
-1.18403256e+00 -7.96096444e-01 6.99909031e-02 -4.08214629e-02
-1.49225950e+00 -1.20689058e+00 2.67298490e-01 -5.24373829e-01
1.10453308e+00 7.05814511e-02 2.65698344e-01 1.89227664e+00
-5.96554935e-01 1.25206041e+00 1.19482398e+00 -5.28110981e-01
-6.07614338e-01 3.59686017e-01 -3.95172775e-01 9.04419839e-01
-1.76843375e-01 -1.50991082e-01 -1.04523158e+00 2.95981970e-02
4.46417183e-01 -7.77427435e-01 -2.63512403e-01 1.62294745e-01
-1.68966329e+00 7.32774258e-01 -1.06315650e-01 1.79728389e-01
1.17867447e-01 6.46022195e-03 5.74226499e-01 2.24741042e-01
5.96423566e-01 4.49643582e-02 -4.64335710e-01 -3.11336160e-01
-8.11555922e-01 1.99209303e-01 7.37173975e-01 1.35807395e+00
4.80473340e-01 1.12359202e-03 -4.24905926e-01 1.17980003e+00
5.52332282e-01 7.36766338e-01 1.98072970e-01 -8.30028713e-01
1.01737118e+00 2.97024727e-01 -1.56145528e-01 -3.38369220e-01
-3.00662726e-01 2.30908021e-01 -4.01329368e-01 -3.40550870e-01
6.69433475e-01 -2.51177996e-01 -8.30097020e-01 2.02012300e+00
-3.18025500e-02 -5.59851117e-02 2.52691597e-01 5.87305903e-01
1.55914676e+00 8.91729772e-01 2.87085533e-01 1.91392917e-02
1.41763699e+00 -9.89014268e-01 -3.92858684e-01 -2.55181700e-01
6.09500408e-01 -1.21559238e+00 1.47703803e+00 2.54440635e-01
-1.24472964e+00 -5.20137787e-01 -5.97293913e-01 -6.65004015e-01
-5.17049193e-01 6.15032375e-01 8.79350901e-01 6.39216840e-01
-1.70115912e+00 -3.69020700e-01 -2.99392074e-01 -8.82052779e-01
2.30225235e-01 -1.53988777e-02 -8.80495012e-01 -2.82553613e-01
-1.19403231e+00 1.03074574e+00 2.25581333e-01 -7.69390836e-02
-1.23921442e+00 -5.47173738e-01 -1.11791289e+00 -5.47452450e-01
1.26580462e-01 -3.43029141e-01 1.22444189e+00 -9.08906877e-01
-1.11554635e+00 1.78593373e+00 -3.51651669e-01 -1.50070386e-02
5.95876992e-01 -9.13751051e-02 -6.39705300e-01 -1.06545009e-01
2.46360466e-01 1.44478381e+00 4.42032576e-01 -1.32357252e+00
-5.23918390e-01 -1.78562984e-01 -1.03264742e-01 4.35320735e-01
-1.95487306e-01 6.15625024e-01 -9.85538006e-01 -5.56751072e-01
-6.73124194e-01 -8.76366556e-01 3.16155016e-01 -1.70970351e-01
-5.97627342e-01 -3.46840978e-01 5.39354622e-01 -1.02244532e+00
9.61076438e-01 -2.12811041e+00 2.07610086e-01 -3.62571687e-01
-1.14064105e-01 -4.64407727e-02 -7.37036288e-01 8.97027731e-01
-4.42617089e-02 4.23658937e-01 1.03068165e-01 -9.73368943e-01
1.29434854e-01 -8.05585980e-02 -2.34998822e-01 4.85573977e-01
3.28850478e-01 1.18284309e+00 -6.35010481e-01 -5.49620509e-01
3.69182117e-02 6.41578853e-01 -2.15060696e-01 1.23094350e-01
-3.37805420e-01 8.59823108e-01 -3.77146751e-02 1.02443290e+00
4.14449513e-01 1.28460109e-01 1.50029168e-01 -2.07937583e-01
-4.65948522e-01 1.78114086e-01 -5.58980703e-01 2.08386469e+00
-6.10835552e-01 1.06196284e+00 3.90986115e-01 -4.26845819e-01
6.95695817e-01 6.31342828e-01 1.61136821e-01 -6.37986779e-01
6.55450970e-02 3.26017767e-01 -1.15271389e-01 -5.78622818e-01
4.96250927e-01 1.73656225e-01 -4.61476654e-01 3.64337325e-01
3.68909478e-01 -1.60454333e-01 4.75941718e-01 2.91297704e-01
3.48005921e-01 3.31480116e-01 7.95147195e-02 -1.58952266e-01
4.95769411e-01 -1.43897146e-01 -1.26707219e-02 7.03964889e-01
-2.58342952e-01 6.84427738e-01 4.89515543e-01 -5.00471815e-02
-1.19430411e+00 -1.13232732e+00 -2.67821461e-01 1.59232128e+00
-5.47995806e-01 -4.25753057e-01 -6.29015982e-01 -1.71020150e-01
-5.11958748e-02 7.36971796e-01 -3.66638452e-01 2.89242327e-01
-2.84553915e-01 -5.91237783e-01 1.31370294e+00 3.20247769e-01
3.38775486e-01 -1.31216347e+00 1.76829413e-01 -3.41372229e-02
-4.58380073e-01 -1.57856572e+00 -4.76729631e-01 -5.13268709e-01
-5.76674826e-02 -8.02935123e-01 -1.20193589e+00 -1.10551012e+00
1.63447946e-01 -1.92794502e-01 1.51283574e+00 -3.32231134e-01
1.94522412e-03 8.02881122e-01 -1.24590240e-01 -3.02987933e-01
-6.52597070e-01 1.11490496e-01 -1.81714892e-02 -1.81318130e-02
2.92672604e-01 -1.55495424e-02 -2.53965333e-02 -4.25410122e-02
-4.47886348e-01 5.03423035e-01 4.64722604e-01 5.45480072e-01
3.51900101e-01 -8.91653597e-01 5.87226629e-01 -6.21031404e-01
7.03038871e-01 -5.69866478e-01 -4.49359328e-01 6.29840672e-01
-1.87324621e-02 -3.73477489e-01 2.94040702e-02 -4.65065509e-01
-1.34944522e+00 -4.08027731e-02 -2.20389739e-01 -6.45844489e-02
-4.40799415e-01 5.41673005e-01 -2.96785295e-01 2.01544687e-01
1.55008376e-01 1.62393600e-01 -2.41440654e-01 -5.79307795e-01
1.20045638e+00 9.12060022e-01 9.21686709e-01 -9.90207136e-01
3.15943390e-01 3.16769332e-01 -2.25190818e-01 -1.13757849e+00
-4.66235280e-01 -1.41483709e-01 -6.50899351e-01 -3.15791309e-01
1.02562690e+00 -1.44619274e+00 -8.80014598e-01 7.13480234e-01
-1.40401793e+00 -5.46973586e-01 2.56311595e-01 1.55001670e-01
-5.18033206e-01 6.69311285e-02 -7.56074548e-01 -1.05401289e+00
-5.77028766e-02 -1.52655101e+00 1.60288250e+00 -2.36581676e-02
-3.62585872e-01 -9.44787621e-01 -2.46833847e-03 8.08829665e-01
1.01674438e-01 2.32784823e-01 1.11633730e+00 -5.11059105e-01
-3.39685947e-01 2.52263188e-01 -5.23058474e-01 1.91585292e-04
-2.10588157e-01 3.72649461e-01 -1.09497285e+00 -2.87478805e-01
-8.14339101e-01 -1.09275436e+00 9.92177546e-01 2.64431179e-01
6.17895961e-01 -9.94646922e-02 -2.97143370e-01 5.21256506e-01
9.10410047e-01 4.08598781e-02 3.56150299e-01 6.83429092e-02
8.77175152e-01 9.95357513e-01 2.14185581e-01 2.45711252e-01
1.00236022e+00 8.10037851e-01 -3.49946357e-02 -3.67834419e-01
-5.06301939e-01 -5.54674327e-01 7.93595672e-01 7.83691525e-01
4.08308208e-02 -4.77192044e-01 -1.36368752e+00 8.61922562e-01
-1.73827410e+00 -6.64111733e-01 -2.01040894e-01 2.04543281e+00
1.03886521e+00 -4.47229415e-01 3.48475397e-01 -7.48010635e-01
6.46236300e-01 3.28967601e-01 -1.10072397e-01 -5.35130918e-01
-7.70751655e-01 1.94622204e-02 2.64416307e-01 6.67947590e-01
-1.14574623e+00 1.63627911e+00 6.82441330e+00 8.57049406e-01
-1.05582035e+00 4.31748301e-01 1.05111170e+00 -3.41850162e-01
-4.59784150e-01 -6.45562336e-02 -1.09893906e+00 3.38935971e-01
1.31160259e+00 5.41178472e-02 8.58954012e-01 4.16887611e-01
2.73407847e-01 -3.45310606e-02 -1.24074709e+00 1.26193559e+00
5.40662408e-01 -1.40225363e+00 1.27183214e-01 -7.07811490e-02
8.17684770e-01 5.55197001e-01 5.00218570e-01 4.26944137e-01
2.13845938e-01 -1.49322641e+00 1.16904581e+00 4.60833341e-01
1.63584352e+00 -7.96880603e-01 1.01781324e-01 1.01794101e-01
-1.09286571e+00 1.90339223e-01 4.10368778e-02 2.54274786e-01
5.33922195e-01 1.11688543e-02 -6.95933223e-01 1.55453160e-01
6.75173879e-01 6.75596237e-01 -8.54048491e-01 5.10525703e-01
-1.81105256e-01 5.72203040e-01 -1.06730901e-01 1.36739433e-01
4.31805611e-01 -8.77785459e-02 4.88730609e-01 1.71172309e+00
2.83661932e-01 -4.37678546e-01 3.65595549e-01 6.74549460e-01
-5.15897989e-01 4.16043162e-01 -1.03150213e+00 -7.10630119e-01
5.51992536e-01 1.17657232e+00 -3.15926224e-01 -1.56199411e-01
-9.14546609e-01 9.32794213e-01 5.34581900e-01 5.32394350e-01
-6.89147055e-01 1.85168356e-01 4.30021346e-01 -1.56054586e-01
-3.26468885e-01 -3.11436445e-01 -1.79172419e-02 -1.20792878e+00
-3.15858275e-01 -1.26087248e+00 4.18886811e-01 -1.13069844e+00
-1.50222826e+00 6.61624312e-01 2.33395621e-01 -4.52760249e-01
-5.15722752e-01 -9.20614064e-01 -5.93206892e-03 1.07713616e+00
-1.42838418e+00 -2.23720241e+00 8.02895874e-02 9.11307037e-01
7.31941164e-01 -7.46774316e-01 8.68042707e-01 4.71475840e-01
-6.17645681e-01 7.72437572e-01 -6.70201704e-02 1.79367989e-01
1.19935822e+00 -9.34817493e-01 5.15710294e-01 6.55045211e-01
3.99605274e-01 5.36704004e-01 2.33236134e-01 -6.91831768e-01
-1.39032221e+00 -7.94309497e-01 1.38309741e+00 -7.32691407e-01
1.03078139e+00 -1.06818724e+00 -4.42773730e-01 1.25089216e+00
9.10920084e-01 -4.92250353e-01 6.55830860e-01 1.45357400e-01
-4.54737544e-01 2.82932818e-01 -8.80719900e-01 9.92999077e-01
1.26800537e+00 -9.63838100e-01 -1.99137986e-01 6.38213217e-01
7.65351951e-01 -4.70936209e-01 -6.94630623e-01 2.33072400e-01
7.23933995e-01 -8.08543622e-01 9.72308218e-01 -6.85876310e-01
6.65678918e-01 1.59566998e-01 -3.85007590e-01 -1.07554603e+00
2.46380623e-02 -9.98616219e-01 2.60619670e-01 2.07686210e+00
9.84650910e-01 -4.42718297e-01 2.98280954e-01 3.55719417e-01
-1.50105909e-01 -5.00627279e-01 -1.02403700e+00 -3.03883433e-01
3.31811190e-01 -7.77170300e-01 3.81043434e-01 9.52764690e-01
-2.21464127e-01 5.32536328e-01 -6.14467204e-01 -1.82374343e-01
3.25175077e-01 -3.14241767e-01 8.89453053e-01 -7.97769368e-01
-1.37622699e-01 -7.47997522e-01 -6.45527896e-03 -8.38563621e-01
7.13907063e-01 -1.30083108e+00 -3.23731810e-01 -1.64654195e+00
3.47637355e-01 -2.51600295e-01 3.67392659e-01 8.33884597e-01
1.04598477e-01 4.34616029e-01 2.53638268e-01 9.09034312e-02
-4.01109278e-01 3.43795061e-01 1.15447462e+00 -2.27806896e-01
9.38723162e-02 -4.14396226e-01 -9.32327509e-01 6.35831714e-01
3.34248155e-01 1.77955270e-01 -5.30797243e-01 -9.41459537e-01
2.29289547e-01 -9.21608955e-02 1.23976275e-01 -3.24491829e-01
-1.65551066e-01 -8.81703719e-02 3.22981536e-01 -6.09621763e-01
6.35130405e-01 -3.24296296e-01 -9.42425281e-02 -1.49255618e-01
-4.43946570e-01 4.88238931e-01 6.49243057e-01 -1.71166509e-01
-1.76096305e-01 2.09482834e-01 5.24041474e-01 -2.30110720e-01
-1.07420325e+00 3.35383296e-01 -4.66112405e-01 4.78298217e-01
7.15351939e-01 9.28471908e-02 -7.22187519e-01 -8.65348577e-01
-5.84267914e-01 2.71410704e-01 3.62404495e-01 8.99291396e-01
4.12507504e-01 -1.44763505e+00 -1.35150933e+00 5.96732087e-02
6.26931012e-01 -4.49048430e-01 -1.36731025e-02 8.98321986e-01
-4.06516761e-01 6.72502995e-01 -1.40490204e-01 -4.24100459e-01
-1.14685893e+00 3.01355243e-01 2.45221145e-02 1.79461706e-02
-1.71049640e-01 1.19535947e+00 2.27451250e-01 -7.06736803e-01
2.63227850e-01 7.68180639e-02 -2.76677430e-01 3.43824178e-01
4.36429769e-01 1.80619672e-01 -3.27898175e-01 -1.51531875e+00
-5.44398248e-01 2.94645429e-01 1.30071968e-01 -8.78385842e-01
1.05945945e+00 -3.42999339e-01 -3.20997387e-01 7.42938876e-01
1.20130813e+00 4.54503626e-01 -7.87219644e-01 7.68696144e-02
9.84814465e-02 -1.53200969e-01 -8.27106014e-02 -1.20058370e+00
-7.44912565e-01 1.00616169e+00 4.02484000e-01 -4.11035478e-01
9.57915604e-01 5.07727861e-01 5.26403189e-01 3.47877443e-02
2.85142213e-01 -8.56109619e-01 -2.85228103e-01 7.61852801e-01
1.42899418e+00 -1.31130648e+00 -4.04078275e-01 -1.80054858e-01
-1.18256831e+00 7.88482785e-01 7.07758248e-01 5.69413960e-01
1.88573152e-01 3.87503505e-01 3.57136488e-01 4.13594134e-02
-8.01355898e-01 -3.96532059e-01 5.56810081e-01 8.99699092e-01
1.20353556e+00 2.16934294e-01 -2.41465479e-01 5.53704381e-01
-4.13181275e-01 -2.85911888e-01 1.99249998e-01 3.85493010e-01
2.17321113e-01 -1.17371321e+00 -3.30076069e-01 -1.28942132e-01
-3.59475464e-01 -7.10457981e-01 -9.20348108e-01 1.15362418e+00
2.59207666e-01 1.23046851e+00 8.60098228e-02 -1.71271581e-02
2.42847100e-01 3.22005779e-01 7.04580545e-01 -6.44905686e-01
-3.62558842e-01 2.60370016e-01 8.21983397e-01 -4.39851642e-01
-2.59745270e-01 -8.13995242e-01 -8.14783931e-01 -3.34506035e-01
4.22968239e-01 -5.49112260e-01 7.02887356e-01 7.52839327e-01
3.14584792e-01 -9.70289707e-02 1.35937296e-02 -1.11342585e+00
2.20829606e-01 -1.08293891e+00 -2.94064879e-01 3.24474871e-01
1.66742459e-01 -4.58991349e-01 -1.68577340e-02 8.87974165e-03] | [11.21796703338623, 1.5863239765167236] |
5730c260-d6ea-41e5-9f8d-452ed7fa8760 | perceptual-kalman-filters-online-state | 2306.02400 | null | https://arxiv.org/abs/2306.02400v1 | https://arxiv.org/pdf/2306.02400v1.pdf | Perceptual Kalman Filters: Online State Estimation under a Perfect Perceptual-Quality Constraint | Many practical settings call for the reconstruction of temporal signals from corrupted or missing data. Classic examples include decoding, tracking, signal enhancement and denoising. Since the reconstructed signals are ultimately viewed by humans, it is desirable to achieve reconstructions that are pleasing to human perception. Mathematically, perfect perceptual-quality is achieved when the distribution of restored signals is the same as that of natural signals, a requirement which has been heavily researched in static estimation settings (i.e. when a whole signal is processed at once). Here, we study the problem of optimal causal filtering under a perfect perceptual-quality constraint, which is a task of fundamentally different nature. Specifically, we analyze a Gaussian Markov signal observed through a linear noisy transformation. In the absence of perceptual constraints, the Kalman filter is known to be optimal in the MSE sense for this setting. Here, we show that adding the perfect perceptual quality constraint (i.e. the requirement of temporal consistency), introduces a fundamental dilemma whereby the filter may have to "knowingly" ignore new information revealed by the observations in order to conform to its past decisions. This often comes at the cost of a significant increase in the MSE (beyond that encountered in static settings). Our analysis goes beyond the classic innovation process of the Kalman filter, and introduces the novel concept of an unutilized information process. Using this tool, we present a recursive formula for perceptual filters, and demonstrate the qualitative effects of perfect perceptual-quality estimation on a video reconstruction problem. | ['Ron Meir', 'Tomer Michaeli', 'Dror Freirich'] | 2023-06-04 | null | null | null | null | ['video-reconstruction'] | ['computer-vision'] | [ 6.79555953e-01 6.19302131e-02 3.20045948e-01 4.76911850e-02
-3.96544188e-01 -5.83541930e-01 4.82678175e-01 -4.04526778e-02
-4.62725222e-01 6.75932825e-01 2.17075408e-01 -2.30359361e-01
-4.39849287e-01 -4.58827406e-01 -7.77240038e-01 -1.00735962e+00
-1.35457456e-01 -3.26450586e-01 1.32696226e-01 -7.37680346e-02
1.43106014e-01 3.60875517e-01 -1.60852027e+00 -1.44511655e-01
7.34780729e-01 1.02001143e+00 5.05156457e-01 1.06184483e+00
3.50236863e-01 6.22800469e-01 -5.59660017e-01 -3.84877115e-01
3.53696555e-01 -6.57043278e-01 -3.69574964e-01 4.85026896e-01
2.82024685e-02 -4.57435846e-01 -3.52292001e-01 1.44883668e+00
2.38104135e-01 2.43114948e-01 4.12043899e-01 -9.39487040e-01
-5.19716024e-01 3.43522817e-01 -1.18533641e-01 1.59631073e-01
6.07756317e-01 1.22486919e-01 8.14876437e-01 -6.12943232e-01
4.91798133e-01 1.10456026e+00 5.50902903e-01 3.65984648e-01
-1.54927635e+00 -2.03783602e-01 1.40514746e-01 1.85138926e-01
-1.12625742e+00 -8.31084907e-01 6.32399321e-01 -5.41609704e-01
3.11616361e-01 4.20265973e-01 5.52240670e-01 1.13822973e+00
4.89615768e-01 3.54479641e-01 1.10835838e+00 -4.46846753e-01
3.26517254e-01 5.13868369e-02 3.20507027e-02 1.92015156e-01
2.35698894e-01 5.37582457e-01 -5.88323891e-01 1.26746565e-01
6.82932794e-01 -1.54915720e-01 -8.21655810e-01 -2.79041052e-01
-1.23788750e+00 3.35462272e-01 -1.21981084e-01 5.07007241e-01
-6.70452595e-01 2.10541263e-01 8.94042253e-02 8.90463114e-01
3.42463315e-01 4.16349173e-01 -2.95378655e-01 -1.79655686e-01
-8.84658158e-01 6.61909059e-02 8.86589348e-01 7.44685054e-01
2.69479364e-01 3.11437815e-01 -2.15173536e-03 3.44032973e-01
2.84101754e-01 7.20826089e-01 1.77398145e-01 -1.23420966e+00
-2.47529875e-02 -3.08804393e-01 6.35878980e-01 -1.11439049e+00
-1.98921517e-01 -9.96240437e-01 -1.01217639e+00 4.66923237e-01
8.64953518e-01 -9.65986028e-02 -6.15750730e-01 2.07667851e+00
1.32333398e-01 2.83300996e-01 1.81370333e-01 9.25061584e-01
-8.41834396e-02 7.45937467e-01 -3.14943522e-01 -9.83884454e-01
1.07711804e+00 -6.71206191e-02 -1.30337489e+00 -8.73327181e-02
-1.68696880e-01 -9.67396379e-01 8.65535080e-01 1.02933848e+00
-1.22646773e+00 -5.80883563e-01 -1.30057466e+00 3.28966260e-01
1.85182363e-01 -2.34767750e-01 6.76138252e-02 6.81270003e-01
-1.10360539e+00 8.83292437e-01 -6.89090848e-01 -2.84264773e-01
-2.95553923e-01 -3.72417681e-02 -2.37082750e-01 3.15587036e-02
-9.94373262e-01 8.99543881e-01 9.73941609e-02 4.31960285e-01
-9.84168410e-01 -5.02213776e-01 -5.13783514e-01 1.78126618e-01
7.13022411e-01 -5.73153675e-01 1.42149818e+00 -1.33571458e+00
-1.44628441e+00 2.82055110e-01 -3.33402455e-01 -4.61980343e-01
9.25616622e-01 -3.74615967e-01 -5.53072095e-01 1.96722478e-01
-2.10005611e-01 -1.56490102e-01 1.57377911e+00 -1.41842067e+00
-3.39446306e-01 -1.25686601e-01 6.44679070e-02 2.94198859e-02
1.19992211e-01 -1.92141205e-01 -2.79177606e-01 -8.09722245e-01
6.30909979e-01 -8.99454474e-01 -3.09680343e-01 7.96529055e-02
-1.97849482e-01 4.34589148e-01 3.63621652e-01 -9.01613772e-01
1.22335637e+00 -2.50040054e+00 1.99035957e-01 2.81587213e-01
3.20886374e-02 -1.29166394e-01 9.66114476e-02 5.29833019e-01
-2.61193365e-01 -1.44261613e-01 -2.74549186e-01 -3.92233819e-01
3.77321281e-02 1.14910036e-01 -4.91106212e-01 8.12514603e-01
3.05478014e-02 2.92289048e-01 -9.54408169e-01 -9.20723379e-02
2.81571150e-01 4.30270761e-01 -5.58396935e-01 1.69405311e-01
-5.64882606e-02 7.26029038e-01 -1.58818215e-01 1.24660514e-01
6.06484175e-01 -1.24435082e-01 2.57475406e-01 -3.91655058e-01
-2.69046992e-01 -1.49112821e-01 -1.66450286e+00 1.55891502e+00
-3.96027416e-01 8.08201373e-01 5.19936860e-01 -8.82778227e-01
5.03159761e-01 6.70618653e-01 3.92438382e-01 -6.57613397e-01
3.63632590e-02 1.20708540e-01 1.31588832e-01 -4.82557327e-01
5.39893806e-01 -4.98308927e-01 1.24407634e-01 2.09772825e-01
-1.31174803e-01 -3.58016342e-02 -2.72314269e-02 1.84840143e-01
1.20448303e+00 2.19116407e-03 4.53776896e-01 -2.49123201e-01
2.60081470e-01 -4.13765758e-01 7.18206942e-01 1.04765391e+00
-9.49009061e-02 4.89064097e-01 4.43179190e-01 1.60193563e-01
-1.08691442e+00 -1.20311964e+00 -9.19109583e-02 5.46652019e-01
1.84886292e-01 -2.22680390e-01 -5.94429731e-01 7.76795894e-02
-2.58012205e-01 7.45681226e-01 -4.41368729e-01 -1.45740747e-01
-3.12011272e-01 -4.19204772e-01 -1.71732903e-02 -2.49527860e-02
3.81872177e-01 -5.05865276e-01 -9.28687692e-01 6.55059934e-01
-1.22445375e-01 -1.11442995e+00 -3.44173074e-01 1.31490618e-01
-8.64374876e-01 -8.52238953e-01 -6.97269499e-01 -6.55532852e-02
4.85799223e-01 3.10611457e-01 7.92944252e-01 1.85689814e-02
1.97226524e-01 8.04932892e-01 -4.00608480e-01 -6.88432306e-02
-6.27136290e-01 -8.00988555e-01 2.28700727e-01 4.93398607e-01
-3.05557370e-01 -7.65878201e-01 -5.05610645e-01 2.52816767e-01
-1.18387365e+00 4.66165952e-02 5.39258718e-01 8.49574566e-01
2.82814056e-01 6.88677967e-01 3.89529318e-01 -3.84131759e-01
5.74595094e-01 -2.45734379e-01 -7.25090027e-01 4.16157693e-02
-4.78986502e-01 1.08039878e-01 7.05202222e-01 -6.72762692e-01
-1.29436398e+00 1.31137846e-02 -2.60774866e-02 -3.49608719e-01
-1.92381367e-02 4.85837400e-01 -3.75447005e-01 3.07230428e-02
5.42915285e-01 3.94099712e-01 1.97009579e-03 -5.36921501e-01
4.01559383e-01 3.74312580e-01 8.09720457e-01 -4.12337542e-01
8.34364235e-01 5.74323356e-01 2.88612187e-01 -1.14066315e+00
-5.60967982e-01 -3.41316700e-01 -5.80999255e-01 -5.89110136e-01
5.59884906e-01 -6.32579803e-01 -8.26819241e-01 4.26568180e-01
-1.26958919e+00 -1.32953063e-01 -6.04517579e-01 8.10741484e-01
-8.26963007e-01 6.19223416e-01 -4.37559813e-01 -1.36985981e+00
4.40596968e-01 -1.00861347e+00 5.07928729e-01 -5.40459156e-02
-4.32665646e-02 -8.62074852e-01 -2.92064488e-01 -2.27029964e-01
4.16262984e-01 2.03140005e-01 4.78028864e-01 -2.64519025e-02
-7.83175886e-01 -2.21619494e-02 9.91183147e-02 6.32486463e-01
6.01169802e-02 -1.81203857e-01 -9.12544787e-01 -3.05251658e-01
7.41744995e-01 4.90139455e-01 7.31582820e-01 6.57127798e-01
5.79483688e-01 -4.79578733e-01 8.21579918e-02 3.88094515e-01
1.60672367e+00 5.46436071e-01 6.19585097e-01 5.87204657e-03
1.50979728e-01 8.30472827e-01 5.51665902e-01 6.14955544e-01
-2.44789615e-01 7.04647899e-01 4.32071954e-01 1.50636077e-01
-3.44856232e-02 -5.11486977e-02 3.84292543e-01 8.71582627e-01
-2.91413274e-02 -4.34558839e-01 -3.85669380e-01 5.07614136e-01
-1.79533362e+00 -1.15792310e+00 -1.11235358e-01 2.63342714e+00
6.30433142e-01 3.11842471e-01 -2.76868016e-01 5.54683447e-01
7.10736692e-01 7.63235986e-02 -5.42698503e-01 -2.70084411e-01
-2.24703193e-01 -9.96194929e-02 6.65906072e-01 6.30002022e-01
-8.47284913e-01 1.63062289e-01 6.39108229e+00 6.13202453e-01
-8.23361993e-01 3.10061842e-01 8.02359506e-02 1.73187628e-01
-2.02282265e-01 2.96946347e-01 -1.89646870e-01 5.76502085e-01
1.07106805e+00 -4.96487260e-01 5.95649242e-01 1.59500435e-01
8.41715813e-01 -4.69408929e-01 -1.16178775e+00 9.34525192e-01
-2.16155961e-01 -7.34485924e-01 -4.21312451e-01 2.60063440e-01
3.09727997e-01 -6.83794439e-01 2.12526679e-01 -3.61838400e-01
1.17294490e-01 -7.33769536e-01 1.08684874e+00 1.05766189e+00
5.15392601e-01 -4.26461697e-01 4.33041692e-01 4.63473767e-01
-1.02289677e+00 -3.57461691e-01 -1.08046360e-01 -4.34266359e-01
7.09861338e-01 1.09010077e+00 -2.44661614e-01 7.18356073e-01
3.65583003e-01 6.40394151e-01 1.00757726e-01 1.36743557e+00
-2.27570340e-01 5.65962136e-01 -2.99004078e-01 3.17958474e-01
-3.87013927e-02 -4.60815012e-01 1.11843705e+00 1.09934521e+00
6.98494315e-01 2.88352847e-01 -1.53146833e-01 6.26949966e-01
3.20723414e-01 -1.32402092e-01 -5.81417501e-01 1.85489669e-01
3.23934168e-01 6.04434371e-01 -5.93449414e-01 -2.63691008e-01
-3.88865143e-01 1.01827121e+00 -5.25171041e-01 5.49528301e-01
-5.58869660e-01 -9.96441171e-02 5.48998773e-01 -6.85336208e-03
1.95799604e-01 -5.17360091e-01 1.88326910e-02 -1.05591047e+00
1.76609114e-01 -8.53268027e-01 8.53014216e-02 -8.82397890e-01
-1.14246809e+00 1.88086823e-01 5.42931317e-04 -1.54587376e+00
-2.10672677e-01 -2.66449958e-01 -2.27959529e-01 8.08701932e-01
-1.28873074e+00 -3.99238378e-01 9.35774818e-02 4.92559075e-01
5.21732092e-01 3.98778498e-01 3.69263977e-01 4.36922669e-01
-1.67141080e-01 2.92238016e-02 5.13084054e-01 -4.28377986e-01
5.65720677e-01 -1.20665622e+00 -3.92792113e-02 1.38242972e+00
3.44783552e-02 6.35491848e-01 1.62056088e+00 -6.79648638e-01
-1.66130698e+00 -6.66694462e-01 6.71924710e-01 -5.26116677e-02
7.31820464e-01 -9.23668370e-02 -1.00135696e+00 4.90702331e-01
8.21276680e-02 -3.82993639e-01 2.82658488e-01 -2.51350492e-01
-5.64751728e-03 -2.00531065e-01 -9.68525946e-01 5.64200103e-01
9.78733540e-01 -4.66355294e-01 -7.33750582e-01 1.43701360e-01
7.97693133e-01 -1.21467948e-01 -7.42525220e-01 2.53702760e-01
7.68307388e-01 -1.12414813e+00 9.68364418e-01 -2.34580845e-01
-1.85576100e-02 -5.07970095e-01 -4.92300689e-01 -1.40293777e+00
-2.88365543e-01 -1.04255211e+00 7.22752437e-02 9.72406387e-01
1.45995945e-01 -7.20672488e-01 2.38238081e-01 4.33672249e-01
-7.51233175e-02 2.79524233e-02 -1.18150449e+00 -1.16326332e+00
-3.72523636e-01 -9.22981203e-01 2.32336447e-01 7.61215746e-01
-1.38063103e-01 -1.63632650e-02 -8.54129314e-01 5.76769471e-01
1.01315665e+00 -2.30828151e-01 4.96036381e-01 -1.09742498e+00
-8.12756002e-01 -2.13196307e-01 -3.78258735e-01 -1.44087470e+00
-2.70740151e-01 -1.29601672e-01 4.19622123e-01 -1.17339921e+00
-2.18260363e-01 -1.31225720e-01 -1.66213498e-01 -4.33307111e-01
1.66684389e-01 -1.07454769e-01 4.59908992e-01 3.09779197e-01
-1.56334713e-01 4.43043828e-01 1.25094736e+00 1.08475946e-01
-1.75733447e-01 4.03518975e-01 -4.34783489e-01 7.81216323e-01
3.77866119e-01 -4.46469098e-01 -4.99363512e-01 -2.57115453e-01
5.51918983e-01 5.78129947e-01 6.72860563e-01 -1.08990157e+00
5.09288847e-01 -7.02227429e-02 1.67416900e-01 -4.66110796e-01
6.82530463e-01 -1.25966167e+00 8.70902061e-01 5.39536715e-01
-2.76540637e-01 1.95474792e-02 -1.66351333e-01 1.06422126e+00
-8.03014711e-02 -4.64008451e-01 7.43510306e-01 -1.50177643e-01
-6.48451269e-01 -2.45217517e-01 -6.91881597e-01 -4.44219589e-01
7.85627782e-01 -3.30463380e-01 -5.14360145e-02 -8.72591257e-01
-1.29257190e+00 -1.96312621e-01 5.11476398e-01 1.94969233e-02
6.56640410e-01 -8.84381533e-01 -5.69557250e-01 -5.71046211e-03
-3.27436894e-01 -5.40504992e-01 4.85661745e-01 1.17557490e+00
-6.16875570e-03 1.83016717e-01 1.14454046e-01 -6.06593132e-01
-1.12416482e+00 8.01015198e-01 3.17095339e-01 4.66541164e-02
-7.29848623e-01 3.48339885e-01 2.04251960e-01 5.19512713e-01
2.29197174e-01 -4.95817512e-01 3.97148589e-03 1.10876195e-01
6.08102024e-01 3.35545510e-01 3.78938317e-02 -5.16702175e-01
4.84107845e-02 5.27209938e-01 4.49012578e-01 -6.77140474e-01
1.01138771e+00 -7.53827631e-01 9.29094404e-02 8.38199914e-01
8.06849897e-01 3.31970155e-01 -1.51070964e+00 -3.77215624e-01
4.04870994e-02 -7.70873487e-01 1.94334388e-01 -6.79703712e-01
-7.00060785e-01 7.34158039e-01 7.04376459e-01 7.57774949e-01
1.48937702e+00 -2.64762193e-01 2.91224092e-01 1.59677386e-01
4.72961724e-01 -9.16526139e-01 5.73015353e-03 2.48171091e-01
9.65911329e-01 -8.08852553e-01 2.90021440e-03 -3.04014981e-01
-2.43676484e-01 1.02140558e+00 -2.79244453e-01 2.57035950e-04
7.85092652e-01 2.89331913e-01 -2.30686679e-01 1.91789582e-01
-7.26037085e-01 -3.56730103e-01 1.48840576e-01 5.28148055e-01
1.02903947e-01 1.28918499e-01 -4.35683638e-01 3.56217414e-01
-9.13078710e-02 -2.33230507e-03 9.42686975e-01 9.10897017e-01
-5.90517104e-01 -8.28417242e-01 -8.25016618e-01 2.58678377e-01
-6.41152978e-01 -6.92084059e-02 2.18913510e-01 5.11333168e-01
-6.93507493e-02 1.48533130e+00 -1.19159192e-01 -1.48805618e-01
4.87269372e-01 -9.29181278e-02 4.78190899e-01 -2.81622857e-01
-1.07984535e-01 5.09060502e-01 -1.13219144e-02 -6.60370529e-01
-7.49853551e-01 -8.47436488e-01 -6.04815185e-01 -2.76313037e-01
-3.32090080e-01 1.73349649e-01 8.06161165e-01 1.03668392e+00
-8.51101428e-02 7.91281879e-01 5.95101118e-01 -7.00297177e-01
-7.63164401e-01 -6.72592521e-01 -9.30498183e-01 2.39438191e-01
8.56271982e-01 -5.68013906e-01 -8.49033952e-01 4.00546670e-01] | [11.438981056213379, -2.0996625423431396] |
e1979097-3054-4e4b-b132-293f8a273b5f | from-solution-synthesis-to-student-attempt | 2205.01265 | null | https://arxiv.org/abs/2205.01265v3 | https://arxiv.org/pdf/2205.01265v3.pdf | From {Solution Synthesis} to {Student Attempt Synthesis} for Block-Based Visual Programming Tasks | Block-based visual programming environments are increasingly used to introduce computing concepts to beginners. Given that programming tasks are open-ended and conceptual, novice students often struggle when learning in these environments. AI-driven programming tutors hold great promise in automatically assisting struggling students, and need several components to realize this potential. We investigate the crucial component of student modeling, in particular, the ability to automatically infer students' misconceptions for predicting (synthesizing) their behavior. We introduce a novel benchmark, StudentSyn, centered around the following challenge: For a given student, synthesize the student's attempt on a new target task after observing the student's attempt on a fixed reference task. This challenge is akin to that of program synthesis; however, instead of synthesizing a {solution} (i.e., program an expert would write), the goal here is to synthesize a {student attempt} (i.e., program that a given student would write). We first show that human experts (TutorSS) can achieve high performance on the benchmark, whereas simple baselines perform poorly. Then, we develop two neuro/symbolic techniques (NeurSS and SymSS) in a quest to close this gap with TutorSS. | ['Nikitas Theodoropoulos', 'Adish Singla'] | 2022-05-03 | null | null | null | null | ['program-synthesis', 'misconceptions'] | ['computer-code', 'miscellaneous'] | [ 2.74935156e-01 3.16114187e-01 -4.37969826e-02 -4.22957778e-01
-4.03069079e-01 -7.72696197e-01 5.18050015e-01 6.84650302e-01
-1.45297199e-01 1.57173783e-01 -2.69288033e-01 -1.06212831e+00
1.00124761e-01 -9.37951505e-01 -7.92246163e-01 -2.60807633e-01
2.38059476e-01 4.78299439e-01 2.65231252e-01 -3.85635763e-01
5.36221147e-01 5.62867165e-01 -1.79901326e+00 3.67445856e-01
1.27702749e+00 1.61140740e-01 1.46585569e-01 1.10369849e+00
-2.62847751e-01 1.54746819e+00 -8.17547500e-01 -3.28717262e-01
-3.56702715e-01 -6.32631600e-01 -1.18887734e+00 -3.73821110e-01
6.40222847e-01 -1.64767653e-01 -5.17512485e-02 1.25267065e+00
8.06762502e-02 4.65112031e-01 7.58180380e-01 -1.74380183e+00
-5.12853324e-01 9.69156325e-01 -2.17774913e-01 3.86037976e-02
6.64912760e-01 5.30888438e-01 9.15100813e-01 -5.82530677e-01
4.30849373e-01 9.98464108e-01 5.92961013e-01 9.52364445e-01
-1.59553659e+00 -4.34957385e-01 2.53447086e-01 1.05722316e-01
-9.71869767e-01 -2.62825906e-01 7.30892837e-01 -8.67117822e-01
6.78688049e-01 4.08331543e-01 9.54923153e-01 9.62564349e-01
-1.16271779e-01 1.27641261e+00 9.49207366e-01 -5.53251207e-01
1.98424131e-01 4.89266247e-01 6.36527300e-01 8.06924164e-01
-1.63713604e-01 2.35939085e-01 -4.21119839e-01 5.49300648e-02
4.38725054e-01 -9.62270871e-02 -3.43223214e-01 -6.88190877e-01
-9.69345927e-01 8.02827954e-01 2.63766915e-01 3.28357905e-01
1.62660375e-01 1.80271894e-01 2.08486125e-01 6.01971745e-01
-4.28519905e-01 8.81008983e-01 -3.17278266e-01 -5.33094168e-01
-1.02587950e+00 8.22704613e-01 1.18568730e+00 1.21054888e+00
3.24861288e-01 4.32198346e-01 -2.35396907e-01 4.53011811e-01
2.77950972e-01 1.11415453e-01 4.79693860e-01 -1.04546189e+00
2.11204827e-01 9.35570538e-01 -1.93310097e-01 -8.01419139e-01
-1.37550935e-01 -1.75936580e-01 -4.14212704e-01 7.95868278e-01
7.98681557e-01 -1.94792956e-01 -9.57067609e-01 1.76446402e+00
1.52597710e-01 4.07303989e-01 1.43661335e-01 5.72964072e-01
1.08103812e+00 8.48575056e-01 3.74711037e-01 7.72246951e-03
1.26711261e+00 -1.04759872e+00 -1.09428734e-01 -4.14362341e-01
1.07504821e+00 -4.03237879e-01 1.18466926e+00 8.48323643e-01
-1.58273315e+00 -7.66507804e-01 -9.95699406e-01 -7.59194326e-03
-2.36352846e-01 -3.96627456e-01 4.23807800e-01 6.66055143e-01
-1.05944824e+00 7.59829462e-01 -8.38628709e-01 2.21947189e-02
1.89914390e-01 1.97648719e-01 8.06770474e-02 -1.13458604e-01
-7.44656801e-01 9.79736269e-01 3.64336699e-01 -5.22604823e-01
-1.18388569e+00 -1.31859779e+00 -9.70319986e-01 7.06776798e-01
2.53287822e-01 -3.54060203e-01 2.00575280e+00 -1.06972468e+00
-1.55539763e+00 1.03281641e+00 2.38986351e-02 -1.78821832e-01
6.31648600e-01 2.27029577e-01 2.46796474e-01 -2.49091968e-01
-4.38229330e-02 6.64594650e-01 3.25798005e-01 -1.28958976e+00
-5.97578943e-01 -1.39180228e-01 3.21162999e-01 2.62577146e-01
-2.55815715e-01 1.53021798e-01 -1.66601926e-01 -5.11738241e-01
2.78786957e-01 -8.95852029e-01 -1.88004136e-01 1.88424718e-02
-2.27193192e-01 -6.96719944e-01 4.83498096e-01 -1.15814894e-01
1.36975527e+00 -1.91606486e+00 3.84195536e-01 4.39692467e-01
6.51105165e-01 5.86893737e-01 2.05247793e-02 3.10548097e-01
-5.01895130e-01 2.63501346e-01 -1.69964045e-01 1.28088472e-02
2.49005675e-01 -1.96988150e-01 -3.79047513e-01 -1.13667391e-01
5.51132336e-02 7.98827529e-01 -1.14904952e+00 -4.39317673e-01
2.98391581e-01 -2.44421847e-02 -8.46327662e-01 6.91124976e-01
-4.99402851e-01 2.51839072e-01 -4.28599775e-01 4.69560593e-01
3.05691957e-01 -2.19016939e-01 9.58100259e-02 5.97274125e-01
-9.33854058e-02 4.08688396e-01 -1.20372558e+00 1.35542047e+00
-5.54919481e-01 8.43266606e-01 -1.31075354e-02 -1.31095111e+00
9.00888741e-01 2.68363684e-01 -1.41681328e-01 -4.04659539e-01
-2.35428642e-02 -9.37374756e-02 5.56251287e-01 -4.84705687e-01
2.53660172e-01 4.99091856e-02 1.16230249e-01 8.41906309e-01
2.23117732e-02 -6.74452305e-01 2.40341425e-01 5.60879767e-01
1.35724378e+00 3.55589032e-01 2.20925301e-01 -2.88212448e-01
4.83684957e-01 4.35981333e-01 2.38914102e-01 1.13064361e+00
-2.01228678e-01 2.73086965e-01 9.79655802e-01 -6.22355461e-01
-9.51525986e-01 -9.47870791e-01 3.23328823e-01 1.51513469e+00
-2.86187440e-01 -5.98185182e-01 -8.16003144e-01 -5.84073544e-01
-6.43652827e-02 1.23556590e+00 -4.19048250e-01 -4.51072961e-01
-8.56157422e-01 -7.75079057e-02 5.60990810e-01 6.49237216e-01
1.39622053e-03 -1.34948516e+00 -1.12537491e+00 3.12986165e-01
1.74933508e-01 -5.69292724e-01 -2.15953901e-01 5.08526504e-01
-7.40225255e-01 -9.83019769e-01 -4.42264616e-01 -1.14885271e+00
9.81671453e-01 1.03446633e-01 1.61263466e+00 9.29548800e-01
-3.05750519e-01 4.90721166e-01 -4.06702841e-03 -6.42296553e-01
-7.89626062e-01 -1.18300781e-01 -2.34497249e-01 -7.95197725e-01
4.98438925e-01 -6.80151999e-01 -1.37439296e-01 2.60886643e-02
-6.13049746e-01 6.94448471e-01 1.82698071e-01 6.31756902e-01
-6.14694618e-02 -7.40460828e-02 6.35015368e-02 -1.20584440e+00
9.05912042e-01 -3.91787946e-01 -8.37746441e-01 4.74462777e-01
-7.47466981e-01 2.19713941e-01 9.01216447e-01 -7.34906673e-01
-6.74342155e-01 1.11225009e-01 -5.54774851e-02 -3.35117489e-01
-5.15425742e-01 8.58920395e-01 1.36874676e-01 -9.98811424e-02
1.10349262e+00 4.32089508e-01 -1.22995064e-01 -5.00159012e-03
-7.43396208e-02 1.46709472e-01 7.74265349e-01 -1.47090364e+00
9.97418046e-01 -5.93641639e-01 -2.01328591e-01 -4.66354907e-01
-6.87692463e-01 -6.50478154e-02 -3.94768655e-01 -3.84961605e-01
4.18179065e-01 -8.12210619e-01 -1.61143506e+00 3.66921365e-01
-1.17514837e+00 -1.00649500e+00 -1.73213854e-01 -5.56413643e-03
-5.32410622e-01 -2.96388447e-01 -4.67128247e-01 -8.28904033e-01
1.57707512e-01 -1.70420992e+00 2.49667913e-01 6.09986365e-01
-7.89526641e-01 -1.02237463e+00 6.98611587e-02 3.89345974e-01
3.98444563e-01 2.19387621e-01 1.48112452e+00 -1.17733097e+00
-3.99599820e-01 1.21199027e-01 8.71409941e-03 9.66843739e-02
-5.33694267e-01 4.80120450e-01 -8.22275937e-01 5.18878922e-02
2.32749544e-02 -8.02422881e-01 1.32565007e-01 -6.65651336e-02
1.32601225e+00 -2.58831203e-01 -2.75708109e-01 4.99934614e-01
8.97031844e-01 4.80559617e-01 3.16498578e-01 3.14211130e-01
8.13606799e-01 7.30600357e-01 1.46499455e-01 -5.79638518e-02
5.64675868e-01 5.28894365e-01 1.49339154e-01 3.95830572e-01
2.31024131e-01 -3.85508776e-01 4.03322101e-01 5.67085564e-01
9.76006985e-02 -2.26731505e-02 -1.81177700e+00 5.99754572e-01
-1.81856525e+00 -8.75029206e-01 -3.95272434e-01 1.96885920e+00
1.29164970e+00 3.89701754e-01 1.71380326e-01 2.11478263e-01
2.74358511e-01 -9.78126675e-02 -4.47221965e-01 -8.81826162e-01
7.91393876e-01 6.71809196e-01 -3.20548832e-01 7.00259328e-01
-5.45734525e-01 1.10874522e+00 5.24863338e+00 5.26005328e-01
-1.09209549e+00 -4.87475067e-01 4.61143076e-01 3.74199271e-01
-5.15750289e-01 -4.57058363e-02 -9.64532971e-01 1.94878280e-01
1.33837640e+00 -4.91036952e-01 6.14836752e-01 1.15965104e+00
-2.89056599e-01 -1.48486674e-01 -1.97232020e+00 5.31547189e-01
4.61770520e-02 -1.25448251e+00 -6.36883676e-02 -4.33147699e-01
7.76288271e-01 -7.76648223e-01 8.00475180e-02 1.04742718e+00
1.03928745e+00 -1.44983077e+00 7.64849722e-01 1.93711713e-01
2.15615794e-01 -8.53061140e-01 1.74770609e-01 8.94400179e-01
-8.22959244e-01 2.53793299e-02 1.11883663e-01 -4.18546975e-01
-6.48308933e-01 -2.14983046e-01 -1.07791400e+00 -2.11931005e-01
4.65894550e-01 3.02963912e-01 -7.83832967e-01 1.01995540e+00
-7.85728157e-01 5.63452065e-01 -5.57922535e-02 -5.00976980e-01
1.94624782e-01 5.39199114e-02 3.64829123e-01 9.51291442e-01
2.94028342e-01 6.94856286e-01 4.06869978e-01 1.47657669e+00
5.32617085e-02 -2.13244498e-01 -5.93887568e-01 -2.45837927e-01
5.57245910e-01 1.04052734e+00 -5.37902355e-01 -4.70982581e-01
-2.07635388e-01 6.24148250e-01 6.29025459e-01 4.44737703e-01
-5.06775975e-01 -5.87428868e-01 5.71815550e-01 1.34854391e-01
-9.72229391e-02 -9.19402242e-02 -6.35319173e-01 -9.21399593e-01
-2.96067655e-01 -1.59526420e+00 6.11168034e-02 -1.24884772e+00
-5.26932001e-01 1.73414141e-01 2.00921688e-02 -9.15435433e-01
-4.99423474e-01 -6.61337078e-01 -1.36015284e+00 1.17114210e+00
-9.61919904e-01 -5.39766073e-01 -7.05818713e-01 4.37406212e-01
5.74034691e-01 -9.71808359e-02 9.57201481e-01 -3.61279756e-01
-5.34409761e-01 8.33861470e-01 -6.26034856e-01 2.33643681e-01
3.63864332e-01 -1.70263326e+00 6.50190711e-01 8.72318149e-01
1.93368882e-01 8.87665093e-01 1.10909343e+00 -4.65034753e-01
-1.58112466e+00 -6.94307566e-01 9.29352880e-01 -6.83844984e-01
6.70207322e-01 -1.91811115e-01 -1.22943258e+00 9.66892838e-01
-8.71311314e-03 -8.66208225e-02 5.00078082e-01 1.12015292e-01
-3.54292870e-01 3.58336806e-01 -8.54666770e-01 1.04325521e+00
6.75791442e-01 -4.86681908e-01 -1.14385283e+00 3.35712641e-01
6.25536680e-01 -1.04550767e+00 -5.89711845e-01 7.06467777e-02
4.65771735e-01 -9.28163528e-01 1.02281427e+00 -1.11525536e+00
1.08780694e+00 -1.43293917e-01 4.80461240e-01 -1.62808478e+00
3.22770402e-02 -5.90138376e-01 -1.88507691e-01 1.09879160e+00
2.46641323e-01 -4.72765528e-02 1.14613402e+00 1.32983780e+00
-3.77658516e-01 -8.63569081e-01 -1.02767767e-02 -3.17895859e-01
4.00701761e-01 -6.00803614e-01 3.30552757e-01 1.27350855e+00
4.60721701e-01 2.52982616e-01 2.79511034e-01 1.67214230e-01
4.45289433e-01 1.31115302e-01 1.11972213e+00 -1.62670469e+00
-3.27540249e-01 -8.71889114e-01 -6.29185215e-02 -1.01572299e+00
3.68322283e-01 -1.10302830e+00 3.17526340e-01 -1.09251130e+00
-2.39969287e-02 -6.89827919e-01 9.51124504e-02 6.27741337e-01
-3.97799700e-01 -3.94444793e-01 3.67571205e-01 -1.65420830e-01
-3.94840837e-01 -1.37711227e-01 1.17841375e+00 -1.36124253e-01
-4.41520452e-01 4.72609639e-01 -7.68441141e-01 1.13188016e+00
4.18656707e-01 -6.29302561e-01 -5.71318924e-01 -4.26810086e-01
5.79152763e-01 6.52851284e-01 5.32424688e-01 -1.25968587e+00
1.05035973e+00 -6.52632475e-01 3.68015587e-01 -1.42510086e-01
-2.28004009e-01 -7.91785955e-01 -2.10375860e-01 6.55767500e-01
-9.15056765e-01 4.04273152e-01 5.14619946e-01 -1.23669162e-01
-2.23490953e-01 -1.02345574e+00 8.52061689e-01 -3.99018586e-01
-4.75082457e-01 -1.39179096e-01 -7.12371588e-01 3.93254429e-01
1.23867643e+00 -2.23937124e-01 -4.59554285e-01 -4.75361735e-01
-9.05731916e-01 7.23590553e-01 3.09509039e-01 1.18050776e-01
7.30622172e-01 -9.69488144e-01 -4.83315736e-01 3.12957019e-01
7.05429018e-02 4.86448139e-01 -8.92245844e-02 5.76162279e-01
-9.64543462e-01 2.21297875e-01 -4.80197281e-01 -6.48143589e-01
-1.61401510e+00 4.66294408e-01 2.45673776e-01 -3.78768474e-01
-3.17986727e-01 1.33515024e+00 1.00671299e-01 -9.46144104e-01
7.51205683e-01 -4.54106569e-01 -3.07117492e-01 1.10548757e-01
7.65101373e-01 8.77903104e-02 -9.05448012e-03 2.47379094e-01
6.87171146e-02 1.17740311e-01 -4.04470682e-01 -4.90227761e-03
1.46613383e+00 6.24741077e-01 1.54732466e-01 5.67541182e-01
6.76755607e-01 -2.00985134e-01 -8.25090706e-01 -2.79948086e-01
2.98808187e-01 -8.85190517e-02 -2.29307100e-01 -1.02310348e+00
-5.82266271e-01 1.36932766e+00 1.24736227e-01 3.31184804e-01
6.51406944e-01 -4.52835917e-01 9.29057151e-02 7.87391961e-01
5.76009713e-02 -4.88389641e-01 2.49582723e-01 8.32286000e-01
6.68022811e-01 -1.40436220e+00 2.83179265e-02 -2.00487956e-01
-5.59230030e-01 1.48221421e+00 1.50156713e+00 -6.25242442e-02
8.16655681e-02 3.01367581e-01 -5.53133041e-02 -2.74349481e-01
-1.13540590e+00 3.79606098e-01 3.34397823e-01 4.04813319e-01
8.41507435e-01 -1.71471499e-02 4.25111473e-01 5.83442986e-01
-6.21329308e-01 -4.41804938e-02 9.88714039e-01 1.33690858e+00
-5.40918529e-01 -9.33580399e-01 -4.65172410e-01 2.94229954e-01
-1.24670617e-01 -4.14881051e-01 -4.22627330e-01 7.65514076e-01
-1.88133270e-01 4.69885409e-01 -5.87810948e-02 -2.70664424e-01
4.41772401e-01 6.35409832e-01 5.32841146e-01 -1.19065452e+00
-1.29454601e+00 -7.12081075e-01 -2.75234580e-01 -3.41521621e-01
1.39193967e-01 -5.51293314e-01 -1.20567501e+00 -6.78783894e-01
2.62261838e-01 3.54438603e-01 4.95439708e-01 9.29256022e-01
-3.51968408e-01 7.92535424e-01 1.68543637e-01 -5.43493629e-01
-9.09732103e-01 -6.53030217e-01 1.61220983e-01 1.54433712e-01
3.68366927e-01 -2.13365242e-01 -4.10931915e-01 1.85906813e-01] | [9.413805961608887, 7.371584415435791] |
67f32829-5b32-403f-bf8a-d0391e0d153b | enhancing-building-semantic-segmentation | 2307.04101 | null | https://arxiv.org/abs/2307.04101v1 | https://arxiv.org/pdf/2307.04101v1.pdf | Enhancing Building Semantic Segmentation Accuracy with Super Resolution and Deep Learning: Investigating the Impact of Spatial Resolution on Various Datasets | The development of remote sensing and deep learning techniques has enabled building semantic segmentation with high accuracy and efficiency. Despite their success in different tasks, the discussions on the impact of spatial resolution on deep learning based building semantic segmentation are quite inadequate, which makes choosing a higher cost-effective data source a big challenge. To address the issue mentioned above, in this study, we create remote sensing images among three study areas into multiple spatial resolutions by super-resolution and down-sampling. After that, two representative deep learning architectures: UNet and FPN, are selected for model training and testing. The experimental results obtained from three cities with two deep learning models indicate that the spatial resolution greatly influences building segmentation results, and with a better cost-effectiveness around 0.3m, which we believe will be an important insight for data selection and preparation. | ['Ryosuke Shibasaki', 'Jinyue Yan', 'Xiaoya Song', 'Dou Huang', 'Haoran Zhang', 'Xiaodan Shi', 'Zhiling Guo'] | 2023-07-09 | null | null | null | null | ['semantic-segmentation', 'super-resolution'] | ['computer-vision', 'computer-vision'] | [ 2.81106611e-03 -2.58959532e-01 1.93761359e-03 -3.51719767e-01
-7.07173586e-01 -2.98819765e-02 3.55682880e-01 -1.12326875e-01
-3.92128050e-01 8.56789351e-01 1.44435510e-01 -3.75105321e-01
-4.20105785e-01 -1.65562093e+00 -3.78309697e-01 -8.23623300e-01
-1.53061554e-01 3.49278837e-01 2.08140194e-01 -2.87477434e-01
1.68845296e-01 6.17072165e-01 -1.69322443e+00 3.34184378e-01
1.25915456e+00 7.55815864e-01 6.73613071e-01 2.88081527e-01
-4.50410753e-01 4.33906466e-01 -3.95271927e-01 1.52566686e-01
4.10303563e-01 -1.40301198e-01 -1.09125006e+00 1.84451893e-01
2.31368363e-01 -5.99276960e-01 -8.17615315e-02 1.04889476e+00
6.89000726e-01 1.77026629e-01 4.10847127e-01 -5.92536390e-01
-6.69016719e-01 8.34114611e-01 -6.87025726e-01 2.76073992e-01
-4.28264469e-01 -8.54628086e-02 8.92319620e-01 -7.16630638e-01
-4.95097637e-02 1.06561017e+00 9.51804399e-01 2.49849319e-01
-9.83030856e-01 -7.80860901e-01 1.15995228e-01 2.81511545e-02
-1.83814168e+00 -1.08136572e-01 7.38203168e-01 -5.69173932e-01
6.78396583e-01 2.86798865e-01 8.74673665e-01 3.39965135e-01
-1.69031322e-01 5.60835242e-01 1.36005998e+00 -1.55775815e-01
1.84399113e-01 -1.05491772e-01 -9.71403793e-02 3.40471506e-01
3.63649964e-01 2.05426943e-02 1.58196718e-01 2.78670609e-01
1.36418593e+00 2.38999367e-01 -2.08649561e-01 3.97489816e-01
-7.46116817e-01 1.13533640e+00 1.22361887e+00 7.67730892e-01
-5.24823427e-01 1.19969934e-01 -4.44374122e-02 -1.77451789e-01
5.91819465e-01 2.49674484e-01 -4.63555485e-01 3.54947686e-01
-1.08849394e+00 1.96392193e-01 -2.87999064e-02 5.43633997e-01
1.03822184e+00 3.12938273e-01 2.08638236e-01 1.22459352e+00
3.44361722e-01 5.60573280e-01 2.88315713e-01 -8.71136725e-01
4.49531555e-01 7.86889911e-01 1.72298789e-01 -9.87616062e-01
-6.89088941e-01 -5.21769583e-01 -1.14538014e+00 2.67358720e-01
2.34680176e-01 -1.98144004e-01 -1.21862304e+00 1.24246192e+00
1.28136516e-01 -9.51486547e-03 -1.31955460e-01 1.04563594e+00
1.03043401e+00 7.90157020e-01 5.24641037e-01 2.73447752e-01
1.25726843e+00 -7.01996148e-01 -3.50408882e-01 -4.85241950e-01
4.93312776e-01 -4.85360414e-01 1.28727031e+00 8.28606039e-02
-5.67102492e-01 -7.40144730e-01 -9.60433304e-01 1.12962484e-01
-5.66437125e-01 2.17105776e-01 1.00623727e+00 7.37240672e-01
-1.05594540e+00 6.65316880e-01 -7.00570166e-01 -3.83412033e-01
8.34589839e-01 5.54741919e-02 -1.29741710e-02 2.44782474e-02
-1.41779876e+00 4.80126143e-01 7.16009438e-01 4.24847782e-01
-6.97035849e-01 -6.14780724e-01 -4.29632813e-01 2.27947924e-02
-9.55387652e-02 -5.90188801e-01 8.30470145e-01 -8.80085886e-01
-9.79752481e-01 9.01047111e-01 3.76265734e-01 -4.01654005e-01
4.69513297e-01 -2.79223680e-01 -4.28365707e-01 -3.22102979e-02
5.05617678e-01 7.64037967e-01 3.65466177e-01 -1.46835983e+00
-1.07929623e+00 -5.58003724e-01 2.85811961e-01 3.10502648e-01
1.65583640e-02 -7.98186958e-02 -1.77608266e-01 -4.84993339e-01
5.46843827e-01 -5.47812223e-01 -6.99688792e-01 -5.56264162e-01
-2.14297414e-01 -9.30470824e-02 5.94447613e-01 -1.03009582e+00
1.17176652e+00 -1.90308344e+00 -4.02555645e-01 1.47286922e-01
2.14225724e-01 3.71149004e-01 1.01032473e-01 1.69776514e-01
-2.86967121e-02 7.86524653e-01 -5.69012523e-01 2.85610229e-01
-4.16955620e-01 2.37457499e-01 -1.44980848e-02 1.79919481e-01
8.19149390e-02 7.64794827e-01 -6.44279182e-01 -6.18817866e-01
5.40212512e-01 7.04491735e-01 -3.50839615e-01 7.73656219e-02
-5.99396899e-02 7.17405498e-01 -1.04322398e+00 4.85498250e-01
1.01416099e+00 -3.20134968e-01 -1.35079145e-01 -5.36574870e-02
-4.76043075e-01 2.77886361e-01 -1.19639933e+00 1.31291783e+00
-5.35941541e-01 5.57981789e-01 1.44584803e-02 -1.04546976e+00
1.13531840e+00 8.73514786e-02 5.55262923e-01 -1.21273899e+00
-4.42365594e-02 1.51552767e-01 -2.56688118e-01 -6.04851305e-01
6.40554786e-01 -4.31930274e-01 8.23709592e-02 2.27230638e-01
-6.21545911e-01 -1.73531502e-01 -1.32723168e-01 -3.66421759e-01
3.96009028e-01 -3.07142567e-02 -1.58479124e-01 -4.34724301e-01
3.77704084e-01 2.65275955e-01 5.30550480e-01 6.98394895e-01
-8.02525282e-02 8.58661473e-01 -9.42329317e-02 -8.84187877e-01
-1.21032047e+00 -8.27877343e-01 -4.03186083e-01 1.08188617e+00
1.19160503e-01 2.66863912e-01 -1.03215790e+00 -2.60873884e-01
-2.39158034e-01 5.81859410e-01 -4.51285690e-01 3.53145182e-01
-7.12841868e-01 -1.35053158e+00 5.74982703e-01 7.21417010e-01
1.43314362e+00 -1.18717611e+00 -7.70526171e-01 3.56964648e-01
-4.87274170e-01 -8.62915158e-01 5.08934975e-01 -1.30344585e-01
-1.14259887e+00 -9.04735923e-01 -6.93824172e-01 -7.22383380e-01
4.58315700e-01 6.88150167e-01 1.15712571e+00 3.70839298e-01
-4.50548865e-02 -1.86784670e-01 -4.44264948e-01 -2.22329780e-01
-1.49667442e-01 5.12293518e-01 -4.90933657e-01 -2.52157152e-01
3.08510482e-01 -7.18417943e-01 -8.88898551e-01 2.25385442e-01
-8.65307033e-01 4.46274042e-01 5.81601024e-01 2.99266666e-01
6.83133781e-01 8.26376975e-01 5.86470485e-01 -8.00725460e-01
3.27392340e-01 -3.81242037e-01 -5.82203746e-01 5.17156646e-02
-6.07231438e-01 -3.03431571e-01 3.85950536e-01 3.92808467e-01
-1.32358754e+00 -1.56475097e-01 -6.67192817e-01 2.80462891e-01
-6.42513216e-01 7.85689056e-01 -2.60505140e-01 2.30737850e-01
7.22719371e-01 9.56194773e-02 -6.12651169e-01 -7.81828403e-01
2.38986939e-01 7.74043322e-01 8.84455144e-02 -4.52128232e-01
7.59461880e-01 7.37479806e-01 -4.31970865e-01 -1.19825232e+00
-1.03491282e+00 -2.24898770e-01 -7.94851005e-01 -9.99002904e-02
1.42428744e+00 -1.15547705e+00 -2.06992067e-02 6.57068074e-01
-8.56177270e-01 -5.33676445e-01 -1.89273149e-01 4.00074899e-01
-1.37779683e-01 -4.90392335e-02 -4.00419682e-01 -7.91608512e-01
-4.97302324e-01 -1.01998878e+00 9.03921247e-01 5.83114147e-01
2.24410608e-01 -1.00455451e+00 -2.19918296e-01 7.84002602e-01
7.83704102e-01 3.91244799e-01 9.14441407e-01 -1.10504642e-01
-9.71488118e-01 1.07287273e-01 -7.85960972e-01 3.93078655e-01
2.36099198e-01 -4.24715392e-02 -1.17508674e+00 -2.84783244e-02
-9.91692990e-02 -1.81988273e-02 1.02067780e+00 8.58263314e-01
1.41947913e+00 2.20620707e-02 -2.11895168e-01 6.63869023e-01
1.98154867e+00 2.47362167e-01 9.27844286e-01 8.97073269e-01
9.12566245e-01 5.86922228e-01 5.86599469e-01 4.16205257e-01
3.93457144e-01 2.33619526e-01 7.60585427e-01 -5.98376632e-01
-5.85459135e-02 -1.05608940e-01 -4.05305773e-01 5.40446222e-01
-7.46765971e-01 1.83771253e-01 -1.43266606e+00 8.17189515e-01
-1.53644669e+00 -1.00975823e+00 -4.43232000e-01 2.00887752e+00
5.37794769e-01 -8.50640908e-02 3.56598087e-02 3.16155851e-01
7.50510514e-01 5.95358968e-01 -2.73555517e-01 1.99684590e-01
-2.52906084e-01 3.46624285e-01 7.71714509e-01 4.99831557e-01
-1.42245424e+00 1.20272505e+00 6.41869545e+00 9.05058444e-01
-1.45853043e+00 1.48830771e-01 9.28666174e-01 3.52198422e-01
-4.34232444e-01 -2.54136354e-01 -7.63198495e-01 3.55033964e-01
7.49765277e-01 1.86201558e-01 3.36699575e-01 7.77543724e-01
7.61734843e-01 -2.09209442e-01 -2.25957051e-01 8.50477695e-01
-7.68568754e-01 -1.46287608e+00 1.11511432e-01 1.36512965e-01
9.19499338e-01 5.37579954e-01 -7.43504614e-02 2.33165011e-01
5.83499551e-01 -1.32887685e+00 4.81077999e-01 2.68553108e-01
5.53467095e-01 -9.16949153e-01 7.55903304e-01 4.00372922e-01
-1.61556113e+00 -2.17006505e-01 -7.07785308e-01 -3.62552762e-01
7.37157241e-02 7.49533892e-01 -5.19163191e-01 7.23715186e-01
1.11000121e+00 4.96308476e-01 -5.09199262e-01 9.03717399e-01
-3.02765757e-01 9.09236372e-01 -3.48719358e-01 4.07355905e-01
6.22305274e-01 -5.07224739e-01 -7.28481635e-02 1.09064794e+00
4.45229888e-01 2.75261134e-01 2.94505358e-01 1.02956951e+00
2.39898711e-01 8.35846514e-02 -5.23930848e-01 1.61310539e-01
6.54511392e-01 1.18044686e+00 -1.08681953e+00 -1.21655144e-01
-2.64748186e-01 4.22432572e-01 4.24905196e-02 4.52104479e-01
-1.01364303e+00 -2.28987448e-02 6.96599782e-01 4.08026546e-01
2.81817913e-01 -2.92065948e-01 -7.92834520e-01 -7.59928465e-01
-2.94998974e-01 -5.71054637e-01 2.17586830e-01 -7.98291147e-01
-8.90266955e-01 4.49796081e-01 8.49974528e-02 -8.89089882e-01
4.17658955e-01 -2.62642920e-01 -6.03536963e-01 1.00578630e+00
-1.91831565e+00 -1.41312766e+00 -7.35987067e-01 3.88176292e-01
5.36076069e-01 1.88408017e-01 5.93811572e-01 6.96980596e-01
-6.91785991e-01 -1.26877829e-01 1.32293358e-01 4.60886627e-01
-2.25322172e-01 -1.00495553e+00 4.89403486e-01 7.79081702e-01
-1.16753668e-01 1.66240379e-01 3.77090603e-01 -3.91623497e-01
-4.60346043e-01 -1.52703655e+00 7.04515994e-01 -5.05640823e-03
3.27026188e-01 2.24268049e-01 -1.19000876e+00 5.47566652e-01
-1.36582116e-02 -4.60825980e-01 5.11869729e-01 8.16152059e-03
7.11734444e-02 -2.66488671e-01 -1.16446614e+00 4.80393469e-01
1.08228588e+00 -5.13404369e-01 -4.33938712e-01 2.28385448e-01
5.76323569e-01 -2.55299471e-02 -9.52629447e-01 6.48499429e-01
2.40462482e-01 -1.38054240e+00 1.17149222e+00 -2.38618448e-01
5.93469441e-01 -2.71191746e-01 -3.58613193e-01 -1.13894308e+00
-7.82808959e-01 3.55537325e-01 9.68794346e-01 1.28047228e+00
3.16740900e-01 -6.71883523e-01 1.12312210e+00 4.30486679e-01
-1.76950768e-01 -4.96499002e-01 -8.83446813e-01 -3.85398597e-01
5.89050472e-01 -5.85654438e-01 1.21800101e+00 1.21816993e+00
-1.10055912e+00 3.12546462e-01 -1.38643518e-01 5.05789280e-01
4.04403239e-01 3.55559170e-01 6.37547910e-01 -1.52213359e+00
4.73886937e-01 -7.19893217e-01 1.49671715e-02 -1.00993705e+00
-1.82798088e-01 -4.55440462e-01 -1.38316557e-01 -2.32337952e+00
9.55002680e-02 -1.02431321e+00 -2.70662755e-01 5.03483653e-01
-2.46691838e-01 3.47252071e-01 -2.79580027e-01 3.46350431e-01
-4.79428656e-02 5.73142171e-01 1.19192874e+00 -3.33314568e-01
-1.46524698e-01 1.14263736e-01 -7.33469307e-01 1.03375936e+00
1.34841406e+00 -2.62470067e-01 -3.43919218e-01 -9.99894202e-01
1.81319982e-01 -2.24079490e-01 3.03161144e-01 -1.25582492e+00
-3.36582631e-01 -4.46646541e-01 4.36977744e-01 -1.00522137e+00
1.33401230e-01 -9.98098910e-01 3.83993328e-01 3.88852537e-01
3.59553397e-02 -3.05641592e-01 2.21838221e-01 2.87262145e-02
-2.09020495e-01 -2.44028926e-01 1.03430450e+00 -5.42910695e-01
-1.30091536e+00 5.98150373e-01 -3.16811979e-01 -4.13782261e-02
6.17752612e-01 -5.32462001e-01 -5.49546182e-02 -5.33276163e-02
-6.78175747e-01 1.43069640e-01 3.78424972e-01 1.25937313e-01
5.37554324e-01 -1.33122540e+00 -9.55734432e-01 2.48667032e-01
-1.68988064e-01 6.22060180e-01 7.25362897e-01 3.44187647e-01
-9.64216709e-01 2.63825476e-01 -2.72817940e-01 -6.17971063e-01
-8.84648383e-01 -2.60534622e-02 7.22150147e-01 -3.45525354e-01
-6.95198894e-01 8.90429914e-01 2.91520000e-01 -7.63803005e-01
-3.22956085e-01 -3.53098929e-01 -6.22880399e-01 1.69335052e-01
3.19904834e-01 6.06220722e-01 4.10320610e-02 -8.17218602e-01
-1.26833111e-01 8.18212628e-01 5.07325590e-01 1.47011980e-01
1.66463578e+00 -3.62059802e-01 -1.53359920e-01 3.02916855e-01
7.50013471e-01 -2.71362454e-01 -1.31237888e+00 -2.29118794e-01
5.83602302e-02 -6.56858027e-01 4.66901064e-01 -5.83416462e-01
-1.58150268e+00 1.10321629e+00 9.03625309e-01 6.99095368e-01
1.18835759e+00 1.50992684e-02 7.89419472e-01 1.41122758e-01
4.26759899e-01 -1.31332219e+00 -3.54008377e-01 5.45555592e-01
8.07393014e-01 -1.56745052e+00 1.34994984e-01 -3.76183867e-01
-2.59985507e-01 7.40328968e-01 6.69438303e-01 -9.24012065e-03
7.43914723e-01 -7.96075761e-02 2.94002295e-01 -4.11167860e-01
1.48531616e-01 -5.55825174e-01 -1.22946419e-01 7.56138742e-01
6.60629928e-01 4.98568386e-01 -1.15195185e-01 2.42593810e-01
-3.61148447e-01 5.21323234e-02 5.16630635e-02 6.36296332e-01
-9.84425664e-01 -6.85219169e-01 -6.34483337e-01 3.80321831e-01
-4.15764749e-01 -1.75920740e-01 8.43639299e-02 9.96289730e-01
4.32782412e-01 8.91877651e-01 2.42908567e-01 -4.53577846e-01
2.42992535e-01 -3.13975006e-01 1.09309658e-01 -5.55191696e-01
-4.40217435e-01 2.58328803e-02 8.42041969e-02 -2.32272983e-01
-8.20246756e-01 -2.35972002e-01 -1.49009359e+00 -6.61675453e-01
-3.86785269e-01 1.91950321e-01 5.90148151e-01 1.09152198e+00
7.90470764e-02 6.45791888e-01 6.81650579e-01 -1.00223410e+00
-1.12476408e-01 -1.02961671e+00 -8.73673022e-01 1.07510552e-01
-1.16800964e-02 -5.86479068e-01 -1.82908267e-01 -1.57130331e-01] | [9.33131217956543, -1.3891724348068237] |
0e20fde9-acb7-424e-918a-bcdf593badcd | exactly-optimal-quickest-change-detection-of | 2303.13778 | null | https://arxiv.org/abs/2303.13778v1 | https://arxiv.org/pdf/2303.13778v1.pdf | Exactly Optimal Quickest Change Detection of Markov Chains | This paper establishes that an exactly optimal rule for Bayesian Quickest Change Detection (QCD) of Markov chains is a threshold test on the no change posterior. We also provide a computationally efficient scalar filter for the no change posterior whose effort is independent of the dimension of the chains. We establish that an (undesirable) weak practical super-martingale phenomenon can be exhibited by the no change posterior when the before and after chains are too close in a relative entropy rate sense. The proposed detector is examined in simulation studies. | ['Aaron McFadyen', 'Jasmin James', 'Caitlin Tompkins', 'Justin M. Kennedy', 'Jason J. Ford'] | 2023-03-24 | null | null | null | null | ['change-detection'] | ['computer-vision'] | [ 1.90718725e-01 -5.11226878e-02 9.88212451e-02 -1.06679529e-01
-7.04443872e-01 -6.71895027e-01 7.03972161e-01 2.01054234e-02
-6.30885124e-01 1.13339257e+00 -2.39299744e-01 -6.29645586e-01
8.74099061e-02 -4.77474719e-01 -6.52874708e-01 -1.13147533e+00
-3.06261808e-01 4.72130150e-01 7.72246361e-01 5.35987392e-02
3.31955850e-01 6.26556277e-01 -9.51893508e-01 -3.57452899e-01
4.08975780e-01 9.93403673e-01 -1.37813002e-01 1.44720590e+00
4.67717230e-01 4.06257033e-01 -5.35787344e-01 -3.16634238e-01
6.12889051e-01 -9.08458173e-01 -1.86156720e-01 -2.57572919e-01
-8.96508545e-02 -4.18418527e-01 -6.61459938e-02 1.83147573e+00
5.26593387e-01 -1.32568851e-01 8.67802680e-01 -7.85687387e-01
1.16669253e-01 8.47208142e-01 -5.38432181e-01 8.48488271e-01
1.75647125e-01 3.65677506e-01 6.66557312e-01 -6.44525528e-01
3.21308166e-01 1.44322610e+00 7.59972334e-01 2.47646987e-01
-1.27970827e+00 -8.46648276e-01 -3.59578341e-01 -1.51594698e-01
-1.58174753e+00 -3.66147995e-01 4.27341402e-01 -3.37387741e-01
4.75583196e-01 2.61750847e-01 6.25950813e-01 1.03530800e+00
1.27202237e+00 2.48835593e-01 1.60548162e+00 -5.55442691e-01
7.06274927e-01 3.13318446e-02 4.51178074e-01 5.24864256e-01
7.35588372e-01 7.26530969e-01 -1.82496220e-01 -4.98961806e-01
8.08833659e-01 -2.56290495e-01 8.53945687e-02 -1.96346771e-02
-9.60770369e-01 7.71454394e-01 -7.20204055e-01 1.55706510e-01
-2.29987919e-01 7.22533345e-01 1.65942967e-01 5.76178670e-01
3.07605267e-01 1.12521641e-01 -3.46092433e-01 -4.62598175e-01
-7.10546613e-01 2.94253945e-01 1.06235325e+00 1.21698749e+00
1.45813286e-01 -6.92928163e-03 -3.01662475e-01 -4.30962086e-01
5.10041654e-01 1.26855409e+00 -3.05419862e-01 -8.33533227e-01
1.93592623e-01 -7.39625216e-01 9.04343188e-01 -3.56535852e-01
-3.45442146e-01 -6.91921771e-01 -7.17925310e-01 1.13133177e-01
7.74773657e-01 -6.09932542e-01 -6.49921536e-01 1.54398525e+00
2.94943392e-01 2.03420743e-01 -1.79429069e-01 5.03813207e-01
-2.31101021e-01 5.82065940e-01 -2.83444464e-01 -1.09341729e+00
1.01038885e+00 1.33429810e-01 -1.21723044e+00 2.87712991e-01
2.49148518e-01 -7.85328209e-01 4.96120006e-01 5.75277150e-01
-1.17392528e+00 -1.20446563e-01 -1.31805718e+00 8.63988459e-01
2.62223512e-01 -7.43061483e-01 1.60043254e-01 1.21260524e+00
-8.84451568e-01 7.72282004e-01 -1.00020802e+00 -1.39696747e-01
-4.07865271e-02 9.74989831e-02 6.20785713e-01 3.14312994e-01
-1.29591537e+00 7.60758042e-01 3.65931153e-01 4.79179882e-02
-1.25698936e+00 -3.76383305e-01 -6.69642016e-02 -1.40404165e-01
2.58759886e-01 -1.97683588e-01 1.40132403e+00 -4.98533875e-01
-1.64373219e+00 3.33774209e-01 1.79116756e-01 -6.74031019e-01
9.99273658e-01 -1.72936261e-01 -6.42644644e-01 2.30563715e-01
-1.02861039e-01 -4.45822686e-01 1.34909678e+00 -9.39066947e-01
-6.91967726e-01 -2.03202739e-01 -7.62719810e-01 -2.67329663e-01
4.58296567e-01 2.30005294e-01 2.82093942e-01 -4.48360026e-01
2.00335845e-01 -1.00452638e+00 -5.92218637e-01 -5.16315818e-01
-5.53006232e-01 -3.94924097e-02 4.13729787e-01 -5.65486729e-01
1.22717476e+00 -1.88305485e+00 -6.26752257e-01 6.53892934e-01
-1.57956600e-01 -2.75246918e-01 4.59836930e-01 3.95639122e-01
1.37919292e-01 1.56264767e-01 -1.55403063e-01 1.67565107e-01
9.40045938e-02 -5.16722724e-02 -4.92330402e-01 1.37228668e+00
-1.67332500e-01 3.86727974e-02 -1.10366261e+00 -5.35899460e-01
-9.11280215e-02 1.66224819e-02 -2.66429752e-01 4.55154032e-02
4.55545634e-02 4.34847057e-01 -4.85414535e-01 3.62467438e-01
1.04112113e+00 -4.83032130e-02 1.81728378e-01 2.49233156e-01
-3.96435887e-01 3.90308872e-02 -1.43651021e+00 6.47579789e-01
2.67386258e-01 4.72094059e-01 1.45893008e-01 -6.82539403e-01
7.31226265e-01 4.82735038e-01 3.31645161e-01 -4.87949640e-01
6.38165772e-01 2.15658411e-01 1.36847049e-01 -1.06733359e-01
2.11231709e-01 -9.92074668e-01 -4.76862162e-01 3.91282886e-01
-2.52980173e-01 -2.07662135e-01 9.50903371e-02 1.39242753e-01
1.32894194e+00 -2.63310552e-01 6.55727088e-01 -8.94928694e-01
7.09118247e-02 -4.81465697e-01 6.98263764e-01 1.64475501e+00
-6.14003539e-01 -1.18030589e-02 5.43022573e-01 5.74582852e-02
-1.18068552e+00 -1.49536049e+00 -4.56472993e-01 4.58281100e-01
4.40366745e-01 2.29365304e-02 -5.15515685e-01 -4.28161949e-01
9.07082984e-04 9.13070142e-01 -4.39088613e-01 4.47697379e-02
-4.95550811e-01 -1.07108772e+00 4.61810052e-01 2.69414544e-01
3.61404568e-01 -3.33840668e-01 -6.51113391e-01 5.48328340e-01
4.73717391e-01 -1.03143656e+00 -4.42505985e-01 4.45035279e-01
-1.03517318e+00 -8.49482775e-01 -3.80986899e-01 -8.58227089e-02
3.61093730e-01 -2.20022753e-01 7.57557213e-01 -6.73557580e-01
2.61104912e-01 3.88386875e-01 -1.59045726e-01 -4.77977157e-01
-1.01775718e+00 -7.77841449e-01 4.55309838e-01 8.11591558e-03
2.55688488e-01 -3.91925365e-01 -4.33341950e-01 3.22081566e-01
-5.05180657e-01 -6.28073394e-01 3.63342732e-01 4.15460229e-01
4.58543628e-01 5.15429914e-01 3.34391534e-01 -3.88065845e-01
7.45065391e-01 -4.26260948e-01 -1.31481028e+00 -2.42600907e-02
-9.20638621e-01 4.06430662e-01 3.40865791e-01 -3.95263731e-01
-1.24104047e+00 -1.36389837e-01 4.41155098e-02 -1.42819239e-02
1.37924761e-01 8.22765753e-03 -2.27584336e-02 -5.14981002e-02
4.41482902e-01 1.26234084e-01 -3.35392743e-01 -6.52225375e-01
2.44104207e-01 3.91591012e-01 6.07710540e-01 -5.08207619e-01
8.62086356e-01 7.92797804e-01 6.24345362e-01 -9.04314756e-01
-5.36198854e-01 -3.72167736e-01 -5.14028609e-01 -4.15317833e-01
6.95789576e-01 -5.71195245e-01 -9.85736430e-01 3.70475173e-01
-1.14016259e+00 -1.66688919e-01 -3.67500812e-01 9.93964911e-01
-7.53476501e-01 8.15088272e-01 -6.87859237e-01 -1.51909256e+00
2.84077525e-02 -6.17502451e-01 4.10243660e-01 1.88123047e-01
-6.65010363e-02 -1.00055563e+00 4.35580045e-01 -6.00036979e-01
6.49666116e-02 3.92989963e-01 4.24012303e-01 -5.31423032e-01
-5.10667264e-01 -4.55378026e-01 1.06954850e-01 3.27248722e-01
-3.20873231e-01 3.08799714e-01 -3.45278084e-01 -6.41596377e-01
5.60520411e-01 4.11155492e-01 7.69006848e-01 8.13957930e-01
3.15466136e-01 -6.16641164e-01 -4.94244725e-01 2.04556972e-01
1.49912453e+00 6.71526909e-01 4.43955898e-01 2.16981559e-03
6.15796186e-02 -1.01066753e-01 7.83980608e-01 1.04378510e+00
-5.37151277e-01 3.91214401e-01 4.79895547e-02 5.13645291e-01
2.44377032e-01 -1.43358082e-01 8.99113894e-01 9.42514837e-01
5.73325083e-02 -3.69768322e-01 -6.79481566e-01 4.36349809e-01
-1.47112417e+00 -1.36417150e+00 -5.90907931e-01 2.47847509e+00
1.12586462e+00 7.85396934e-01 1.10232957e-01 1.01546764e-01
1.17434609e+00 -3.20519954e-01 -4.55870956e-01 -7.30580449e-01
-6.28982037e-02 2.38462135e-01 1.31340003e+00 6.52842104e-01
-1.01129282e+00 3.71465713e-01 8.48883724e+00 9.57464337e-01
-3.87317181e-01 6.09723628e-01 1.47996992e-01 -2.59621777e-02
6.88118264e-02 3.37200522e-01 -1.31723285e+00 1.04457748e+00
1.26140141e+00 -3.04659963e-01 -1.37934610e-01 5.33926129e-01
5.09925067e-01 -4.82058316e-01 -1.08039522e+00 6.67977273e-01
-4.33501095e-01 -9.12557244e-01 -5.10823727e-01 3.74765396e-01
8.17156434e-01 -2.99155056e-01 1.31363207e-02 -8.76818374e-02
8.73663843e-01 -4.69861090e-01 7.75473893e-01 8.93456995e-01
6.00994110e-01 -1.16529310e+00 9.36287761e-01 4.28080559e-01
-9.81645882e-01 -2.14143954e-02 -3.48710209e-01 -2.64497310e-01
5.52209616e-01 1.24868917e+00 -8.33939612e-01 -1.28678372e-02
3.21601570e-01 7.17611760e-02 -3.23555456e-03 1.19653296e+00
-1.27886638e-01 1.23991978e+00 -7.74550438e-01 -4.25915062e-01
1.90209851e-01 -3.85296226e-01 1.26858699e+00 1.50427735e+00
5.81330657e-01 1.78423479e-01 5.08152917e-02 5.56178033e-01
3.60595047e-01 -3.91395211e-01 -3.10507953e-01 3.22073624e-02
6.43552363e-01 2.97802210e-01 -1.15751696e+00 -4.64282691e-01
-1.66935787e-01 6.30833745e-01 -7.56180227e-01 3.46400380e-01
-8.53880227e-01 -4.87150788e-01 3.04255694e-01 -4.46221977e-02
7.51941681e-01 -1.97840616e-01 -2.96817988e-01 -8.16963315e-01
-1.92153946e-01 -5.06395996e-01 5.55727363e-01 -2.42980756e-02
-1.29133165e+00 -1.86075106e-01 3.17531705e-01 -1.10112047e+00
-2.91781783e-01 -2.70457625e-01 -4.09166694e-01 5.50073743e-01
-7.40161300e-01 -1.24076933e-01 6.98182940e-01 5.70759177e-01
-3.91926244e-02 3.55038010e-02 1.97784051e-01 -1.83686838e-01
-2.80747145e-01 4.50584412e-01 1.03059196e+00 -6.04538731e-02
1.75124943e-01 -1.16334736e+00 1.80990294e-01 1.36630607e+00
-5.46489596e-01 2.99761027e-01 1.83875585e+00 -1.16251814e+00
-1.29254234e+00 -7.51334250e-01 5.86921096e-01 -3.50694507e-01
1.02601075e+00 -3.02260518e-01 -3.52206498e-01 5.10511637e-01
-2.69835517e-02 -2.81120151e-01 2.22185045e-01 -1.00950710e-01
4.31476384e-02 -1.07062303e-01 -1.24668562e+00 4.23012584e-01
8.25965226e-01 -2.93685377e-01 -7.00130045e-01 4.50293660e-01
4.22259331e-01 1.59712613e-01 -8.63943219e-01 3.27775508e-01
5.98508954e-01 -9.27828252e-01 5.09584129e-01 -4.37651664e-01
-4.70783651e-01 -3.93144757e-01 -3.58937055e-01 -8.37358057e-01
-3.11647117e-01 -1.59143913e+00 -4.26092327e-01 8.57938647e-01
4.62606624e-02 -6.83169365e-01 3.80053669e-01 1.47091135e-01
2.40633488e-01 4.32918966e-03 -1.50873256e+00 -1.63670731e+00
3.59584531e-03 -5.18041551e-01 3.59533429e-02 4.84469920e-01
2.11012453e-01 2.84237683e-01 -6.24709368e-01 5.03809214e-01
1.31825936e+00 -9.46334451e-02 2.61166066e-01 -1.28897452e+00
-6.30877852e-01 -3.11077714e-01 -4.27646428e-01 -1.27811420e+00
-3.87065411e-01 -3.11640143e-01 5.41308880e-01 -6.69944763e-01
5.94657063e-01 -8.48356411e-02 -5.86683571e-01 -5.54231226e-01
8.56573060e-02 -3.21840495e-01 -6.45260187e-03 6.49238825e-02
-7.07011461e-01 5.33754468e-01 1.09108174e+00 3.41483772e-01
-6.21123128e-02 4.31657195e-01 -1.12216748e-01 7.44123459e-01
7.77319074e-01 -1.13230848e+00 5.39058410e-02 6.95085645e-01
4.66356069e-01 6.12537026e-01 2.24274710e-01 -1.18608928e+00
-6.98386431e-02 -2.66848028e-01 2.30972394e-02 -1.22626710e+00
-4.62450571e-02 -6.57549977e-01 3.85905057e-01 1.25357878e+00
-3.75507295e-01 3.75037156e-02 -2.69204140e-01 1.18459785e+00
5.75028598e-01 -5.11901975e-01 1.40147531e+00 5.71625382e-02
-3.15337107e-02 8.48936588e-02 -1.01458478e+00 6.44171759e-02
1.12487197e+00 8.58572721e-02 -1.59111049e-03 -7.30229795e-01
-7.50081182e-01 -1.46667719e-01 2.18195468e-01 -4.84031200e-01
3.14768583e-01 -1.21345413e+00 -6.64837003e-01 -8.18929598e-02
-4.81449634e-01 -8.77151549e-01 -1.18565798e-01 1.06508815e+00
-4.65566337e-01 5.24202764e-01 -1.88579969e-02 -6.49039149e-01
-1.12222075e+00 7.65893042e-01 2.87458122e-01 -3.78990620e-01
-5.70281148e-01 8.87881219e-01 -1.85919598e-01 5.44835448e-01
1.71056345e-01 -4.72673297e-01 5.26473224e-01 -2.38081455e-01
8.57184410e-01 9.25702333e-01 -4.01072621e-01 -1.65094450e-01
-4.15206611e-01 1.87779278e-01 -7.13553354e-02 -8.09066653e-01
7.71546066e-01 -5.11731267e-01 5.85974455e-02 9.39058542e-01
9.72885847e-01 2.54445910e-01 -1.73894358e+00 -1.67698458e-01
2.98004001e-01 -4.55257118e-01 -6.10133074e-03 -3.08177054e-01
-1.92494690e-01 7.83354521e-01 1.10321355e+00 6.33159220e-01
5.80832839e-01 5.11047244e-02 5.73219538e-01 5.33603966e-01
6.42451644e-01 -1.64351797e+00 -8.92155021e-02 5.54395020e-01
5.33577919e-01 -7.20835924e-01 2.42865309e-01 -6.56834692e-02
-1.70356363e-01 9.52089608e-01 -2.88927585e-01 -2.98109353e-01
1.13654315e+00 7.61361122e-01 -4.59043652e-01 -7.96369240e-02
-8.14070761e-01 -1.08816825e-01 -3.81835103e-01 5.77526033e-01
-9.49651003e-02 5.08515954e-01 -8.79881620e-01 2.99894214e-01
-1.18860304e-02 -1.89936534e-01 9.18319583e-01 9.74068999e-01
-9.76025283e-01 -4.72712487e-01 -7.61936605e-01 2.89544046e-01
-7.51339376e-01 -7.37729222e-02 -1.82490945e-02 7.79436767e-01
-3.58632833e-01 1.26926017e+00 1.70560032e-01 -9.79460403e-02
-9.60317776e-02 1.94487646e-02 5.87174058e-01 -1.70931239e-02
1.05335630e-01 5.04897773e-01 2.70934641e-01 -4.99631941e-01
-2.25267485e-01 -1.11151195e+00 -9.39932287e-01 -6.97180808e-01
-5.10527015e-01 3.61611187e-01 4.40146297e-01 1.17914307e+00
-2.82489300e-01 3.83690223e-02 9.83682752e-01 -2.68829107e-01
-1.31863260e+00 -9.89910007e-01 -1.24892044e+00 -1.34374559e-01
4.78775203e-01 -5.99171698e-01 -9.37642813e-01 -2.50162892e-02] | [6.908048629760742, 4.0222063064575195] |
770de7dc-6ecc-4de0-843b-612b4c973ba1 | iterated-learning-for-emergent-systematicity-1 | 2105.01119 | null | https://arxiv.org/abs/2105.01119v1 | https://arxiv.org/pdf/2105.01119v1.pdf | Iterated learning for emergent systematicity in VQA | Although neural module networks have an architectural bias towards compositionality, they require gold standard layouts to generalize systematically in practice. When instead learning layouts and modules jointly, compositionality does not arise automatically and an explicit pressure is necessary for the emergence of layouts exhibiting the right structure. We propose to address this problem using iterated learning, a cognitive science theory of the emergence of compositional languages in nature that has primarily been applied to simple referential games in machine learning. Considering the layouts of module networks as samples from an emergent language, we use iterated learning to encourage the development of structure within this language. We show that the resulting layouts support systematic generalization in neural agents solving the more complex task of visual question-answering. Our regularized iterated learning method can outperform baselines without iterated learning on SHAPES-SyGeT (SHAPES Systematic Generalization Test), a new split of the SHAPES dataset we introduce to evaluate systematic generalization, and on CLOSURE, an extension of CLEVR also designed to test systematic generalization. We demonstrate superior performance in recovering ground-truth compositional program structure with limited supervision on both SHAPES-SyGeT and CLEVR. | ['Aaron Courville', 'Eeshan Dhekane', 'Yuchen Lu', 'Max Schwarzer', 'Ankit Vani'] | 2021-05-03 | iterated-learning-for-emergent-systematicity | https://openreview.net/forum?id=Pd_oMxH8IlF | https://openreview.net/pdf?id=Pd_oMxH8IlF | iclr-2021-1 | ['systematic-generalization'] | ['reasoning'] | [ 1.89339295e-01 5.65944552e-01 -1.36727570e-02 -1.31695643e-01
-1.89968973e-01 -9.70452249e-01 7.08810568e-01 1.61359515e-02
-1.87095478e-01 2.03727990e-01 1.81701273e-01 -7.20673680e-01
-1.69061124e-01 -8.74757588e-01 -1.25111723e+00 -3.86164337e-01
-2.66509920e-01 4.95647252e-01 2.08553210e-01 -2.63731033e-01
1.80805489e-01 8.75893831e-02 -1.49547517e+00 1.01951587e+00
9.36664879e-01 3.45162392e-01 4.90294456e-01 7.16825485e-01
-3.60799342e-01 1.13077223e+00 -1.28692001e-01 -2.94911683e-01
3.90055031e-01 -6.27793193e-01 -1.02935207e+00 1.78238317e-01
1.01486242e+00 1.19061451e-02 -7.03358278e-02 8.44309986e-01
1.53990626e-01 -1.15312658e-01 9.79234278e-01 -1.18817413e+00
-1.17427850e+00 1.22749114e+00 -4.22617137e-01 1.12377167e-01
2.03459710e-01 6.66369259e-01 1.54346859e+00 -8.69202018e-01
1.04351163e+00 1.24715328e+00 8.88374269e-01 7.66359329e-01
-2.00676632e+00 -2.63535291e-01 3.57503653e-01 -1.35420769e-01
-1.18572021e+00 -5.03297091e-01 7.52247453e-01 -8.80414248e-01
1.14691675e+00 -9.95865315e-02 5.81824958e-01 1.14150977e+00
-1.42169923e-01 9.34066117e-01 1.11127794e+00 -3.50026459e-01
3.63116175e-01 3.60214040e-02 -1.32017717e-01 1.22011936e+00
1.85750067e-01 3.29813063e-01 -5.05027175e-01 8.79205111e-03
8.40596139e-01 -2.51759499e-01 -1.77566156e-01 -1.04898906e+00
-1.06049824e+00 6.79220676e-01 9.23017025e-01 3.85659754e-01
-5.53947836e-02 2.53655404e-01 3.28714073e-01 6.51429117e-01
2.19221450e-02 9.97890174e-01 -3.57418031e-01 3.50133926e-01
-5.99957764e-01 3.46795797e-01 6.85839117e-01 1.08776712e+00
8.82661760e-01 9.83624086e-02 3.19163166e-02 6.00157142e-01
2.34519422e-01 3.36597353e-01 2.57669926e-01 -1.03205252e+00
4.44816917e-01 1.25559604e+00 -5.32402158e-01 -6.21387959e-01
-3.94858450e-01 -5.93532979e-01 -7.89878726e-01 3.35713029e-01
6.28783762e-01 -1.22504801e-01 -6.19034886e-01 2.05768824e+00
-1.56731963e-01 -2.38237858e-01 1.14190653e-01 4.99412954e-01
6.88809037e-01 5.56130886e-01 -3.22678052e-02 4.12847757e-01
9.85189140e-01 -8.69261861e-01 1.80097863e-01 -4.60501701e-01
7.74056792e-01 -9.36974213e-02 1.79872203e+00 3.28546315e-01
-1.03894889e+00 -6.63209558e-01 -1.26724148e+00 -7.02990666e-02
-3.79162341e-01 -1.36840299e-01 6.44256890e-01 5.20710289e-01
-1.21823800e+00 6.27701640e-01 -4.58407164e-01 -4.34216022e-01
7.61011720e-01 1.18047066e-01 -2.61665374e-01 2.81108946e-01
-4.61063445e-01 7.74249196e-01 3.15138340e-01 -3.60827744e-01
-1.08209848e+00 -9.50177073e-01 -8.89434755e-01 3.44668441e-02
3.66952360e-01 -1.01670170e+00 1.36538935e+00 -1.57975674e+00
-9.95926499e-01 1.28281224e+00 4.36928496e-02 -5.23800015e-01
2.50098854e-01 2.66087383e-01 2.16239527e-01 -1.23206593e-01
-7.36550465e-02 9.42355931e-01 8.86133194e-01 -1.53026211e+00
-2.62675464e-01 -1.51004449e-01 4.27020162e-01 5.97688891e-02
-2.94725507e-01 -6.78124905e-01 2.63898820e-01 -4.22034502e-01
5.10577746e-02 -7.44741678e-01 -9.72986147e-02 5.06579392e-02
-2.62786925e-01 -3.17704350e-01 3.26637834e-01 -3.44731748e-01
8.86675835e-01 -2.16327548e+00 7.36112595e-01 2.26560444e-01
4.17113155e-01 -4.73888218e-01 -4.87821937e-01 3.57260644e-01
-3.25178504e-01 3.31489056e-01 -4.25827861e-01 -1.82089105e-01
1.83701172e-01 1.92086369e-01 -7.27283061e-01 1.89978704e-01
3.32707703e-01 1.43055034e+00 -8.95293713e-01 -1.26489908e-01
-3.38105202e-01 -3.02680016e-01 -1.05263031e+00 1.51069880e-01
-9.07738268e-01 -1.30377999e-02 4.00885463e-01 2.87954092e-01
2.36013159e-01 -5.15262723e-01 4.08745080e-01 4.37850952e-02
-1.17292985e-01 2.55998254e-01 -7.97721803e-01 2.07604814e+00
-6.07235014e-01 9.21901882e-01 -7.12840930e-02 -9.11936998e-01
6.94648147e-01 6.72868500e-03 -1.14883527e-01 -9.14159954e-01
-3.54220122e-01 2.86501069e-02 5.37480533e-01 -6.97639644e-01
3.55755627e-01 -2.44307622e-01 -1.61478221e-01 1.05525362e+00
3.03109229e-01 -2.23787293e-01 1.09385036e-01 4.60239768e-01
1.35579181e+00 2.24890471e-01 9.15515795e-02 -7.18456149e-01
1.44727185e-01 2.77862281e-01 6.94563985e-02 8.60567749e-01
2.06654832e-01 2.85120040e-01 7.18217671e-01 -6.77080333e-01
-1.54589820e+00 -1.61315954e+00 2.77356714e-01 1.53359032e+00
-2.48879090e-01 -4.02313977e-01 -6.48208737e-01 -7.65910268e-01
-3.13074365e-02 7.17872620e-01 -8.76563072e-01 -2.11039096e-01
-5.66201389e-01 -3.89519691e-01 6.15068257e-01 3.92552078e-01
3.58819783e-01 -1.44458413e+00 -8.78652036e-01 -2.31892884e-01
1.22995526e-01 -7.04993308e-01 -4.31829482e-01 2.97338784e-01
-1.10389209e+00 -1.25783813e+00 -4.25281346e-01 -1.15559793e+00
1.10052955e+00 9.40281004e-02 1.73329592e+00 4.74751949e-01
-4.74279761e-01 7.71465778e-01 1.11011826e-01 -9.87110659e-02
-7.76283324e-01 5.25557935e-01 -3.21510971e-01 -4.30569440e-01
1.61644910e-02 -9.53280330e-01 -3.90101165e-01 2.28417113e-01
-1.15997076e+00 4.54503328e-01 7.91658461e-01 8.45373690e-01
-1.76825479e-01 -9.04235467e-02 5.90778768e-01 -1.13283896e+00
8.76207590e-01 -4.39716667e-01 -7.13177383e-01 4.45848435e-01
-3.72072011e-01 6.24274731e-01 7.47072935e-01 -5.66130102e-01
-9.02264774e-01 -6.10205764e-03 4.07436728e-01 -8.17785859e-02
-2.83692986e-01 6.64353907e-01 -3.07254437e-02 1.41867593e-01
1.52060473e+00 3.74574184e-01 2.31150091e-01 -2.63199329e-01
8.35542500e-01 1.23072855e-01 5.73308229e-01 -1.13789451e+00
1.04530442e+00 3.20976347e-01 -1.30117819e-01 -9.18970048e-01
-3.65414858e-01 2.32965015e-02 -7.92279422e-01 8.12907796e-03
5.82084715e-01 -9.01247799e-01 -6.01623833e-01 1.79330647e-01
-1.19747853e+00 -1.10223961e+00 -4.18028086e-01 -5.56713283e-01
-6.39746130e-01 1.29447579e-01 -4.59150285e-01 -5.72984815e-01
-1.09502070e-01 -8.23833644e-01 8.23719144e-01 -8.72532055e-02
-4.96434867e-01 -1.12706423e+00 4.49540675e-01 -1.90515622e-01
5.60364902e-01 9.84645076e-03 1.88310075e+00 -3.41925263e-01
-1.16628718e+00 5.60806096e-01 -1.79880828e-01 4.84518297e-02
3.40527184e-02 -5.36568463e-02 -8.70582402e-01 -3.07962209e-01
-5.30336678e-01 -6.48180783e-01 1.03203535e+00 -2.07419060e-02
9.21581089e-01 -3.52504224e-01 -1.34333028e-02 6.18572593e-01
1.27435637e+00 -1.65860131e-01 5.37132800e-01 3.90009165e-01
3.89408618e-01 8.38219225e-01 -5.41631937e-01 -3.80922332e-02
5.63860714e-01 2.74135321e-01 3.66224974e-01 2.84673087e-02
-2.46931881e-01 -7.25852668e-01 6.21550143e-01 7.19335616e-01
1.55004323e-01 4.98893827e-01 -1.09303021e+00 6.45429134e-01
-1.76303422e+00 -1.03065836e+00 2.50498027e-01 1.96750224e+00
9.31148827e-01 5.92869706e-02 3.52160305e-01 -3.37408990e-01
3.31357151e-01 1.06515631e-01 -6.31370604e-01 -3.14665794e-01
-4.44936663e-01 2.94244945e-01 -4.91442308e-02 5.44258177e-01
-8.46531630e-01 9.31024611e-01 6.34679270e+00 4.90016639e-01
-8.86708260e-01 2.31916960e-02 5.69825292e-01 3.13936025e-02
-7.51693308e-01 6.72622472e-02 -2.21320570e-01 -1.17453903e-01
5.01312613e-01 -3.30432281e-02 1.09433925e+00 6.67112052e-01
-3.11873466e-01 -1.96252242e-02 -1.71314740e+00 8.10180247e-01
-3.25574204e-02 -1.69366217e+00 1.86519831e-01 -3.67978401e-03
1.13174701e+00 -8.73175543e-03 4.20344263e-01 7.04653919e-01
4.31317180e-01 -1.35811496e+00 8.54271293e-01 4.52033162e-01
5.54497778e-01 -3.06729019e-01 -8.29298720e-02 5.96275330e-01
-1.04430175e+00 -2.39906952e-01 -4.48475361e-01 -2.72449583e-01
-4.40858305e-01 -9.15635191e-03 -1.12636125e+00 1.27124845e-03
3.61714572e-01 5.40381193e-01 -1.33592856e+00 9.46990848e-01
-4.83967274e-01 3.89984637e-01 -6.09297305e-02 -3.41784060e-01
6.23735674e-02 1.50130048e-01 4.27001923e-01 9.58488643e-01
5.09640053e-02 -4.45224941e-01 -1.50598675e-01 1.73365617e+00
-2.52431095e-01 -7.48850703e-02 -1.23593771e+00 -2.89751470e-01
1.45854175e-01 1.06284988e+00 -7.25539625e-01 -1.06368065e-01
-3.62105310e-01 7.08189309e-01 8.56120169e-01 6.02426708e-01
-4.44931298e-01 1.41753972e-01 2.24601910e-01 4.00323391e-01
3.61721903e-01 -3.43817651e-01 -5.70642114e-01 -1.09237540e+00
-1.02993481e-01 -1.29033327e+00 2.61555582e-01 -7.73041725e-01
-1.44643056e+00 5.46332359e-01 -1.74935523e-03 -8.04163158e-01
-2.58550048e-01 -9.43270445e-01 -8.93218517e-01 5.77242970e-01
-6.23833418e-01 -1.28013980e+00 -1.08275287e-01 4.53468889e-01
5.60119152e-01 -6.72584474e-01 7.88628340e-01 -2.74789065e-01
-3.54733676e-01 5.26306748e-01 -3.09428960e-01 1.90150216e-01
3.14919978e-01 -1.64568663e+00 7.57449150e-01 8.07120919e-01
6.96885705e-01 1.07989633e+00 5.79118907e-01 -5.28809726e-01
-1.44552815e+00 -1.00934970e+00 3.06125611e-01 -7.39373386e-01
7.50063479e-01 -9.20642018e-01 -7.86630273e-01 8.31687033e-01
7.40527809e-01 -4.30949122e-01 6.09777749e-01 5.61940670e-01
-1.02376771e+00 2.14159146e-01 -6.11782908e-01 1.04025042e+00
1.61972344e+00 -1.03616428e+00 -1.05263376e+00 -1.78764816e-02
7.72807956e-01 9.22851413e-02 -3.22123259e-01 2.69779384e-01
4.91087824e-01 -1.01201487e+00 8.70682657e-01 -9.48853076e-01
9.19795096e-01 -4.20022815e-01 -1.08035170e-01 -1.49714661e+00
-3.74315679e-01 -5.76706350e-01 -9.16823074e-02 8.24675858e-01
9.54993069e-01 -4.43674535e-01 9.09159958e-01 2.11680487e-01
-7.43343905e-02 -6.20423853e-01 -6.35327876e-01 -6.14056468e-01
5.85182905e-01 -4.54340070e-01 4.75403070e-01 8.19167793e-01
3.64629984e-01 8.70556056e-01 3.28863829e-01 -3.73712443e-02
7.81015873e-01 4.73150045e-01 1.09342480e+00 -1.03777599e+00
-9.48785305e-01 -1.03137732e+00 -4.85930741e-02 -1.18129206e+00
3.85107100e-01 -1.54423654e+00 -9.48018283e-02 -1.21789575e+00
5.58428049e-01 -1.10362716e-01 2.75490224e-01 3.48196864e-01
1.42483130e-01 -1.79961115e-01 2.60751635e-01 7.16395676e-02
-9.11488473e-01 4.78865236e-01 1.14193022e+00 -6.53999031e-01
-2.34221727e-01 -3.06392074e-01 -1.01980865e+00 6.64137840e-01
6.46448553e-01 3.56124504e-03 -7.27338791e-01 -8.34858358e-01
1.09027123e+00 -6.34572327e-01 8.18091750e-01 -8.77608538e-01
4.69644606e-01 1.44551873e-01 3.16838086e-01 -3.32874328e-01
-1.22969903e-01 -6.76621974e-01 1.38679333e-03 4.60332155e-01
-6.52975678e-01 3.52544814e-01 5.52416027e-01 3.64078701e-01
3.51469070e-01 -3.71266127e-01 3.36380959e-01 -5.50614595e-01
-8.77550125e-01 5.54280588e-03 -5.57052553e-01 5.16443789e-01
6.14114523e-01 -5.64502850e-02 -8.56787622e-01 -3.24981958e-01
-8.36795092e-01 1.90899968e-01 8.20373416e-01 4.88161832e-01
8.18734884e-01 -1.34526813e+00 -3.98317486e-01 2.70578831e-01
3.86444688e-01 1.13076285e-01 1.51922464e-01 5.50045967e-01
-6.03027523e-01 1.77258924e-01 -7.84774721e-02 -7.95390606e-01
-8.73377562e-01 6.44675732e-01 5.88492036e-01 -1.43930763e-01
-4.51307148e-01 1.00765693e+00 6.96926177e-01 -1.00886798e+00
2.68928170e-01 -6.28638744e-01 2.07119420e-01 -1.43898353e-01
2.74589539e-01 -9.57471281e-02 -2.87686437e-01 -1.44215122e-01
1.21576995e-01 3.45289648e-01 1.96815759e-01 -2.86077976e-01
1.36312962e+00 1.88759357e-01 -6.28565013e-01 7.59100735e-01
8.73161495e-01 -3.28929946e-02 -1.30690217e+00 -3.44541430e-01
5.88969827e-01 -1.22963317e-01 -5.71730971e-01 -7.96886325e-01
-6.23216808e-01 1.03428578e+00 3.46328795e-01 2.86979884e-01
7.23334670e-01 3.09770495e-01 4.51480858e-02 8.69034231e-01
4.80486155e-01 -7.68057466e-01 7.43077278e-01 7.52181113e-01
1.29042983e+00 -9.38513994e-01 -2.46842816e-01 1.83164954e-01
-3.46819073e-01 1.11042821e+00 9.85441983e-01 -3.37473035e-01
4.65124965e-01 1.96690768e-01 -2.53843069e-01 -4.57872838e-01
-1.21271372e+00 -1.70594543e-01 5.24765730e-01 7.77591109e-01
4.35515642e-01 -7.33544352e-03 6.32522106e-01 5.66194594e-01
-4.42904234e-01 -4.25723493e-01 2.44168967e-01 7.35705376e-01
-5.70386469e-01 -9.06122625e-01 -2.69088298e-01 4.01255518e-01
2.85578370e-01 -3.51577401e-01 -8.40088606e-01 7.62988150e-01
4.45043772e-01 3.21847707e-01 2.37027749e-01 -3.95081669e-01
2.70610631e-01 4.69276696e-01 1.10788107e+00 -9.72287714e-01
-5.73989034e-01 -3.58930200e-01 1.00229902e-03 -2.57249683e-01
-1.55289739e-01 -7.51389742e-01 -1.02285099e+00 -3.00235450e-01
3.21820885e-01 -2.84228772e-01 7.56931379e-02 7.85430491e-01
2.99333751e-01 7.22750783e-01 1.39781177e-01 -7.38593578e-01
-4.43822175e-01 -8.61432076e-01 -2.38408834e-01 6.73869431e-01
3.04700136e-01 -4.12452966e-01 -2.24185288e-01 1.15108132e-01] | [9.541594505310059, 6.968441963195801] |
518c9b75-f362-40e8-b09a-67b4b57c8980 | model-based-image-adjustment-for-a-successful | 2103.03062 | null | https://arxiv.org/abs/2103.03062v1 | https://arxiv.org/pdf/2103.03062v1.pdf | Model-based image adjustment for a successful pansharpening | A new model-based image adjustment for the enhancement of multi-resolution image fusion or pansharpening is proposed. Such image adjustment is needed for most pansharpening methods using panchromatic band and/or intensity image (calculated as a weighted sum of multispectral bands) as an input. Due various reasons, e.g. calibration inaccuracies, usage of different sensors, input images for pansharpening: low resolution multispectral image or more precisely the calculated intensity image and high resolution panchromatic image may differ in values of their physical properties, e.g. radiances or reflectances depending on the processing level. But the same objects/classes in both images should exhibit similar values or more generally similar statistics. Similarity definition will depend on a particular application. For a successful fusion of data from two sensors the energy balance between radiances/reflectances of both sensors should hold. A virtual band is introduced to compensate for total energy disbalance in different sensors. Its estimation consists of several steps: first, weights for individual spectral bands are estimated in a low resolution scale, where both multispectral and panchromatic images (low pass filtered version) are available, then, the estimated virtual band is up-sampled to a high scale and, finally, high resolution panchromatic band is corrected by subtracting virtual band. This corrected panchromatic band is used instead of original panchromatic image in the following pansharpening. It is shown, for example, that the performance quality of component substitution based methods can be increased significantly. | ['Gintautas Palubinskas'] | 2021-03-04 | null | null | null | null | ['pansharpening'] | ['computer-vision'] | [ 9.11729455e-01 -7.50422120e-01 -8.89104530e-02 -1.14520214e-01
-6.68279707e-01 -6.87578022e-01 3.73865664e-01 5.35280518e-02
-3.88458192e-01 7.60239780e-01 -1.87308207e-01 3.67978401e-02
-5.08433521e-01 -1.33537149e+00 -1.96314320e-01 -1.15850222e+00
3.94607991e-01 1.14725977e-01 6.67703748e-02 -4.58043665e-01
-1.38244286e-01 1.02971971e+00 -1.80318499e+00 1.67787418e-01
8.73095393e-01 9.15358543e-01 4.20993090e-01 9.15134609e-01
-5.61525337e-02 1.66366056e-01 -3.27650279e-01 7.05361590e-02
6.54284358e-01 -6.92386806e-01 -3.52492064e-01 5.54123819e-01
3.26979965e-01 -9.39293113e-03 3.17011118e-01 1.68063104e+00
1.36105046e-01 2.18653589e-01 5.33194602e-01 -6.47635996e-01
-3.54784995e-01 4.09737229e-01 -1.27019250e+00 3.84440683e-02
-8.17551650e-03 -6.28509372e-02 7.55953193e-01 -4.42255974e-01
1.43650874e-01 9.48106289e-01 4.04305696e-01 -2.07683761e-02
-1.56625187e+00 -4.43479985e-01 -3.14601243e-01 7.06823766e-02
-1.20321727e+00 1.12138733e-01 1.01882267e+00 -2.14148521e-01
3.57435495e-01 8.11074495e-01 1.08091891e+00 1.53522730e-01
1.79895282e-01 -1.95110753e-01 1.71540487e+00 -6.74714744e-01
-1.01289518e-01 2.13304609e-01 2.44753316e-01 7.99581409e-02
5.93686640e-01 2.87613451e-01 5.18828891e-02 -1.48420647e-01
5.63459933e-01 1.47751316e-01 -6.41288102e-01 -5.69232218e-02
-9.39285576e-01 6.28070891e-01 6.67960763e-01 8.28252316e-01
-7.96062410e-01 -4.32740033e-01 -1.06449075e-01 5.26268780e-01
4.18928683e-01 2.97629833e-01 -4.67885315e-01 4.98484135e-01
-9.85234618e-01 9.88674015e-02 1.10999815e-01 4.61698100e-02
1.44359136e+00 -9.93682817e-02 4.22072649e-01 1.16133344e+00
3.81682031e-02 8.72160912e-01 4.89870340e-01 -6.64825439e-01
1.49771258e-01 5.49485981e-01 3.19288522e-01 -1.05459106e+00
-4.63851303e-01 -3.06529254e-01 -1.12976253e+00 5.45248508e-01
1.29494846e-01 -2.61280127e-02 -8.42500806e-01 1.24148905e+00
4.53508705e-01 -2.04025388e-01 3.62881988e-01 1.07969940e+00
3.33006024e-01 1.12068403e+00 -7.18113482e-02 -7.52719522e-01
1.53360701e+00 -7.33986557e-01 -6.48334742e-01 -1.43097565e-01
-2.68496126e-02 -1.15597963e+00 6.37512088e-01 4.95739102e-01
-8.90236735e-01 -8.57278645e-01 -1.07773471e+00 3.85881990e-01
-6.55380726e-01 2.52539277e-01 3.37875813e-01 8.59300554e-01
-6.41943395e-01 6.50863171e-01 -4.06755477e-01 -2.30965123e-01
-3.72026682e-01 -6.25710711e-02 -2.33312115e-01 1.24833755e-01
-1.01366115e+00 8.51792455e-01 8.34005177e-01 2.12784246e-01
1.92498252e-01 -6.84535980e-01 -6.34541392e-01 -6.49100170e-02
-1.36287957e-01 -2.03121424e-01 6.12197459e-01 -1.55048358e+00
-1.42914557e+00 1.01585519e+00 2.04145730e-01 -2.76117623e-01
2.84129828e-01 1.92914784e-01 -1.07387125e+00 2.80854285e-01
-7.64435753e-02 -1.27033502e-01 9.91995156e-01 -1.44048393e+00
-8.44029546e-01 -5.48564672e-01 -1.75666615e-01 4.25945848e-01
-3.66573073e-02 1.92957819e-01 -9.76968836e-03 -5.09073257e-01
6.21828675e-01 -5.87857425e-01 -2.43390366e-01 -3.26042265e-01
7.85767883e-02 5.09791017e-01 8.62871230e-01 -9.69272316e-01
6.01790190e-01 -2.21518087e+00 -4.44128476e-02 6.17027342e-01
-3.44423175e-01 2.97947377e-01 -2.31616810e-01 1.02159893e-02
-6.57232523e-01 -1.30374610e-01 -6.98771715e-01 5.08046448e-01
-6.58948421e-01 -1.38654197e-02 -8.27772096e-02 6.70847893e-01
-4.84205782e-02 2.26125315e-01 -6.82939589e-01 -3.26610118e-01
6.84154510e-01 6.82629704e-01 1.91715509e-01 -5.76579235e-02
3.81045192e-02 4.52174634e-01 -2.06660181e-01 4.43947434e-01
1.46020913e+00 4.65745807e-01 1.03357717e-01 -7.20396817e-01
-6.25775456e-01 -4.94181633e-01 -1.35520530e+00 1.01951098e+00
-6.89349413e-01 3.84534270e-01 5.07013738e-01 -1.15088129e+00
1.21180534e+00 4.69029218e-01 8.13782394e-01 -7.74098992e-01
-9.82754156e-02 2.70303667e-01 -3.19120228e-01 -1.78680703e-01
9.08365846e-01 -3.89040798e-01 4.84251708e-01 1.54428676e-01
-5.86803615e-01 -9.71842587e-01 2.43145451e-01 -5.28431535e-01
-7.90335536e-02 1.59083419e-02 5.98463237e-01 -9.20929760e-02
1.06900454e+00 3.83991152e-01 3.88836116e-01 3.39251429e-01
1.93022832e-01 5.52698672e-01 -2.52019446e-02 -2.63375252e-01
-1.17966568e+00 -7.72622108e-01 -5.33445716e-01 7.32590199e-01
2.73569524e-01 3.24099422e-01 -5.38545072e-01 1.80763990e-01
-1.63010240e-01 6.06997371e-01 -4.04660553e-01 8.18640441e-02
-4.31535661e-01 -1.69039452e+00 -7.02202395e-02 -2.00000599e-01
9.60312843e-01 -8.09800744e-01 -8.39622140e-01 3.83577347e-01
-1.99294746e-01 -9.80361760e-01 2.48181000e-01 -1.36594683e-01
-1.19788468e+00 -1.19838190e+00 -7.91194499e-01 -1.87542930e-01
5.01709402e-01 8.44814062e-01 8.40790212e-01 -1.98348109e-02
-1.93945155e-01 3.37431192e-01 -5.06731331e-01 -2.62870789e-01
-6.30714774e-01 -6.27318919e-01 -3.79710168e-01 4.11418051e-01
1.50890037e-01 -5.36035895e-01 -4.28077638e-01 2.21552372e-01
-1.19768977e+00 2.64477491e-01 5.83347857e-01 5.25742769e-01
1.00116026e+00 1.07847869e+00 -1.36000430e-02 -5.53256750e-01
3.86386067e-01 -9.27119181e-02 -1.23143387e+00 2.66206205e-01
-4.82952446e-01 -4.85180885e-01 6.29320621e-01 -2.07962543e-01
-1.51909220e+00 -7.44440928e-02 -1.52092008e-02 1.22698717e-01
-3.42261881e-01 4.83114243e-01 -1.74905017e-01 -3.25815558e-01
8.52370799e-01 3.45302820e-01 -2.92279005e-01 -6.35544240e-01
2.48232797e-01 7.71477461e-01 8.64427090e-01 -1.54005766e-01
9.85323846e-01 7.17785299e-01 2.96062827e-01 -1.22470677e+00
-5.39331257e-01 -5.38548648e-01 -5.57227373e-01 -4.30936158e-01
1.04516947e+00 -8.85125697e-01 -4.58368137e-02 7.44736195e-01
-8.35218668e-01 1.20507054e-01 -2.41633907e-01 6.65771365e-01
-2.17462674e-01 5.72148144e-01 -2.33297408e-01 -1.08603036e+00
-3.93134713e-01 -9.13668394e-01 5.57190597e-01 6.51820362e-01
4.67548817e-01 -8.40970874e-01 2.57552952e-01 5.52447557e-01
4.41554904e-01 5.84721088e-01 7.25520253e-01 2.68424779e-01
-1.10532299e-01 -3.43639731e-01 -6.03297651e-01 6.87086463e-01
6.31653786e-01 4.46146965e-01 -9.03795600e-01 -1.77377418e-01
4.63121831e-01 2.04133660e-01 9.19631541e-01 6.56539619e-01
9.99624550e-01 4.18197028e-02 -6.01095930e-02 5.92938662e-01
2.08907080e+00 1.23101152e-01 6.23951495e-01 5.88045478e-01
3.86933208e-01 6.14384294e-01 8.37127805e-01 1.00427628e-01
-2.92749345e-01 6.96547329e-01 4.98941213e-01 -5.18492222e-01
-1.09766044e-01 5.00844598e-01 2.26928085e-01 3.01296562e-01
-7.43122876e-01 2.86368817e-01 -4.80695158e-01 4.68427807e-01
-1.53926528e+00 -1.15269423e+00 -7.59959638e-01 2.69898748e+00
8.76189590e-01 -3.07063371e-01 -2.02851053e-02 6.29204571e-01
1.13010871e+00 4.54874843e-01 -1.73866570e-01 -3.78656238e-01
-6.06068730e-01 4.24558222e-01 9.86950159e-01 8.28587711e-01
-1.01179159e+00 3.36156994e-01 5.19281292e+00 5.58735192e-01
-1.33484566e+00 1.31891847e-01 1.92360401e-01 3.11302960e-01
-2.68044055e-01 9.48553160e-02 -1.85432523e-01 2.01782092e-01
6.90327823e-01 -3.40819359e-01 6.06792450e-01 4.51759279e-01
3.50424737e-01 -7.43864536e-01 -1.07731655e-01 1.10566247e+00
-2.75350690e-01 -7.84744620e-01 -8.49048495e-02 6.93971943e-03
9.65446889e-01 1.73912048e-02 -1.51570171e-01 -5.26489377e-01
1.68986231e-01 -3.95699799e-01 4.13082629e-01 6.43002152e-01
5.13137996e-01 -8.82622778e-01 7.68688381e-01 3.34850967e-01
-1.45732141e+00 1.83811206e-02 -7.29772687e-01 5.88992648e-02
2.01802626e-01 1.22770250e+00 7.11167082e-02 1.29614139e+00
6.78362727e-01 4.51738387e-01 -2.50197411e-01 9.29437160e-01
-2.65971243e-01 2.56105125e-01 -4.75624621e-01 5.11851370e-01
1.67230636e-01 -1.08362579e+00 5.76482713e-01 1.11674368e+00
4.91524905e-01 3.60651970e-01 -1.48799136e-01 7.61549711e-01
4.78952199e-01 3.39048892e-01 -2.80496240e-01 1.26154065e-01
1.97416171e-02 1.60123706e+00 -5.62093318e-01 -6.48973823e-01
-4.59585398e-01 7.83332407e-01 -5.78358471e-01 6.35030746e-01
-7.16859877e-01 -1.04368672e-01 5.94710827e-01 -2.12394610e-01
-1.84699222e-02 8.40844959e-02 -4.17452812e-01 -1.19093215e+00
-2.72176355e-01 -7.06415355e-01 4.92962301e-01 -1.04601038e+00
-8.13785315e-01 5.17367601e-01 1.77703932e-01 -1.20036089e+00
7.64300302e-02 -4.75446463e-01 -3.63563865e-01 1.52980137e+00
-1.82733428e+00 -9.59235728e-01 -7.14218915e-01 7.83879876e-01
1.94576159e-01 2.17424527e-01 8.65461588e-01 8.74131545e-02
-3.79649550e-01 -3.55210394e-01 3.51676047e-01 -3.71116847e-01
3.26478362e-01 -1.06227338e+00 -2.42062584e-01 1.11850572e+00
-4.17556837e-02 -1.35237589e-01 1.01849020e+00 -4.58330691e-01
-9.12770748e-01 -9.72763836e-01 4.99073476e-01 2.61241108e-01
4.83069658e-01 5.30115306e-01 -9.59856570e-01 3.65137726e-01
2.55983561e-01 -1.89579904e-01 4.82112318e-01 -6.70122206e-01
-7.09840581e-02 -5.04888594e-01 -1.51495457e+00 3.12121987e-01
1.08136550e-01 -3.14492524e-01 -4.82994407e-01 4.64384347e-01
1.09390356e-01 -9.46608335e-02 -1.09489727e+00 5.56966066e-01
6.31603301e-01 -1.41214669e+00 1.17143905e+00 5.05461209e-02
1.62890643e-01 -5.91946900e-01 -2.22723305e-01 -1.36263120e+00
-6.11879587e-01 -3.08300499e-02 7.31633544e-01 8.22541356e-01
2.93062717e-01 -8.06342423e-01 2.98495352e-01 2.05145478e-01
4.37910587e-01 1.70424446e-01 -5.25521159e-01 -5.63813329e-01
-1.72106192e-01 -3.04940253e-01 7.82334805e-01 1.11572230e+00
-4.78057176e-01 -3.34285311e-02 -3.09193254e-01 8.57261539e-01
1.02256548e+00 7.22525060e-01 5.92275977e-01 -1.50976598e+00
-3.30419689e-01 -5.98610520e-01 -6.79513812e-02 -2.52020001e-01
-2.82164812e-01 -3.66158217e-01 -4.31024671e-01 -1.60647285e+00
-2.84398682e-02 -2.75070250e-01 -2.84888834e-01 1.71834841e-01
-8.34719539e-02 7.48539805e-01 1.80624068e-01 3.05683196e-01
7.83239245e-01 -1.91937312e-02 1.13861191e+00 -2.81564981e-01
-6.05592012e-01 2.66486675e-01 -3.09082419e-01 8.30972672e-01
9.17764843e-01 -1.63822696e-01 -5.29998131e-02 -8.59527066e-02
1.94351271e-01 3.10783595e-01 4.48790282e-01 -1.23681617e+00
-2.75077045e-01 -6.03282571e-01 3.78009826e-01 -6.89814508e-01
5.12986898e-01 -1.49457502e+00 9.62100744e-01 5.56563735e-01
2.78261095e-01 -3.98508608e-01 1.92558736e-01 8.45190436e-02
-3.25374812e-01 -4.31667596e-01 1.38408506e+00 -3.09570104e-01
-7.89870143e-01 -1.17375039e-01 -2.81843692e-01 -9.75010574e-01
7.76982188e-01 -7.06455231e-01 -1.02853850e-01 -2.35507935e-01
-5.99484026e-01 -4.05776531e-01 7.27219284e-01 -4.67431918e-02
2.12641284e-01 -1.18059349e+00 -8.56241047e-01 2.06056997e-01
7.25399703e-02 -3.66827369e-01 5.97023129e-01 9.05344665e-01
-8.01912367e-01 -6.36523515e-02 -6.06004775e-01 -2.63592809e-01
-1.63167775e+00 4.78363365e-01 8.00871611e-01 -2.89878547e-01
-4.19258505e-01 3.01109135e-01 -1.24170609e-01 -2.72957623e-01
-6.89976454e-01 -2.47681081e-01 -5.13501406e-01 3.68126363e-01
7.78717458e-01 3.99515450e-01 2.07674742e-01 -1.14582300e+00
-1.86997473e-01 1.27156210e+00 5.17559648e-01 -1.95726201e-01
1.43051648e+00 -2.38175407e-01 -7.99789011e-01 2.33733058e-01
8.96666884e-01 4.20643598e-01 -9.08351243e-01 -2.63357610e-01
-6.20301187e-01 -7.73329735e-01 4.40469235e-01 -6.31160557e-01
-1.30138063e+00 4.20808434e-01 8.43832791e-01 6.94041014e-01
1.99123645e+00 -3.75980377e-01 2.33042717e-01 2.04708171e-03
1.68749213e-01 -1.14504421e+00 -7.93495357e-01 -1.08476326e-01
8.59633505e-01 -9.95304167e-01 4.83284622e-01 -6.33774698e-01
-2.55837381e-01 1.50610423e+00 6.07266575e-02 1.33841455e-01
4.46650505e-01 -7.29670525e-02 3.74044001e-01 -8.08962435e-02
7.97673091e-02 -3.18964750e-01 3.14825982e-01 6.22442365e-01
4.25156236e-01 3.59642953e-01 -7.51304090e-01 -2.98012018e-01
-3.64594638e-01 -1.21469021e-01 6.67415559e-01 4.05894756e-01
-7.40456164e-01 -8.57131839e-01 -1.39581406e+00 3.04997623e-01
-2.63580620e-01 4.71602120e-02 -9.06789228e-02 5.51573634e-01
3.17508638e-01 1.05034471e+00 1.46606728e-01 1.28037542e-01
4.48188454e-01 -1.47032931e-01 4.26190525e-01 -2.56566089e-02
-3.99278402e-01 3.79671544e-01 -9.56686288e-02 -2.33660802e-01
-1.25205278e+00 -6.19490206e-01 -8.31568718e-01 -4.16854799e-01
-5.43250203e-01 3.39496285e-01 1.04095531e+00 6.04648650e-01
-5.54870307e-01 2.24979699e-01 8.64769995e-01 -9.79624867e-01
-8.02570507e-02 -8.99532080e-01 -1.27744031e+00 4.06624615e-01
4.13220257e-01 -2.48764813e-01 -5.39318323e-01 8.73792917e-02] | [10.07407283782959, -2.0995922088623047] |
54519b26-2ad5-4cfa-8df9-06169cac7d70 | on-the-possibility-of-rewarding-structure | 1910.04023 | null | https://arxiv.org/abs/1910.04023v4 | https://arxiv.org/pdf/1910.04023v4.pdf | On the Possibility of Rewarding Structure Learning Agents: Mutual Information on Linguistic Random Sets | We present a first attempt to elucidate a theoretical and empirical approach to design the reward provided by a natural language environment to some structure learning agent. To this end, we revisit the Information Theory of unsupervised induction of phrase-structure grammars to characterize the behavior of simulated actions modeled as set-valued random variables (random sets of linguistic samples) constituting semantic structures. Our results showed empirical evidence of that simulated semantic structures (Open Information Extraction triplets) can be distinguished from randomly constructed ones by observing the Mutual Information among their constituents. This suggests the possibility of rewarding structure learning agents without using pretrained structural analyzers (oracle actors/experts). | ['J. Anibal Arias-Aguilar', 'Mauricio Carrasco-Ruíz', 'Ignacio Arroyo-Fernández'] | 2019-10-09 | null | null | null | null | ['open-information-extraction'] | ['natural-language-processing'] | [ 4.86592680e-01 1.23572159e+00 -1.68814912e-01 -5.64584255e-01
-4.98062313e-01 -9.39227164e-01 9.44817126e-01 4.86997105e-02
-5.93133688e-01 7.41264284e-01 1.85646594e-01 -6.15658522e-01
-2.22726256e-01 -1.00156701e+00 -8.75194669e-01 -5.76709092e-01
-2.63951808e-01 1.20226228e+00 2.05024719e-01 -4.20484617e-02
3.14112097e-01 5.05424857e-01 -1.58395076e+00 1.73286587e-01
8.45150411e-01 1.83397412e-01 3.58929217e-01 7.79553592e-01
-4.57660794e-01 1.41540349e+00 -6.40798926e-01 -6.46846831e-01
3.61255705e-01 -7.60140777e-01 -1.26421344e+00 4.03130025e-01
-3.12379122e-01 -1.84199646e-01 1.11951604e-01 1.62079298e+00
-1.40237227e-01 1.11508854e-01 9.12781119e-01 -1.24388790e+00
-5.92097580e-01 1.54346871e+00 2.37684160e-01 9.36995745e-02
6.22205496e-01 4.11610126e-01 1.45111418e+00 -1.96238801e-01
9.72269595e-01 1.66854024e+00 -1.50255904e-01 6.99360251e-01
-1.42001641e+00 -1.46603107e-01 7.88378492e-02 2.42435355e-02
-9.38919187e-01 -5.28093100e-01 7.50056982e-01 -2.74631381e-01
1.30503988e+00 4.46470939e-02 4.42839235e-01 1.19459772e+00
-9.64792445e-02 1.06977570e+00 1.17645168e+00 -1.04747021e+00
4.81660366e-01 4.67552364e-01 2.49060169e-01 1.04763019e+00
4.72096622e-01 7.61402428e-01 -2.43993059e-01 -3.02409112e-01
7.03991175e-01 -8.24521065e-01 4.16692406e-01 -6.16948962e-01
-1.08758259e+00 8.69859457e-01 -1.51996076e-01 5.70550919e-01
-4.28015113e-01 1.73180625e-01 1.63242683e-01 7.20197022e-01
-2.86593586e-01 1.05524361e+00 -5.62020659e-01 -3.39143947e-02
4.14336920e-02 2.21194252e-01 1.05711591e+00 1.10923159e+00
8.83018374e-01 6.28977492e-02 9.81421545e-02 2.81238496e-01
6.75644994e-01 6.91325605e-01 5.07964492e-01 -1.33611023e+00
4.03678179e-01 6.16626918e-01 2.28779510e-01 -5.07121503e-01
-4.10016388e-01 -8.21362138e-02 4.77942936e-02 1.32801414e-01
5.97091198e-01 -4.47121918e-01 -4.24497664e-01 1.87553990e+00
1.97132438e-01 -3.08570683e-01 5.53289175e-01 4.40420121e-01
2.08754957e-01 3.74727339e-01 3.99392307e-01 -4.01678056e-01
1.12448204e+00 -6.45192444e-01 -6.52308106e-01 -3.36146533e-01
1.02488077e+00 -8.58200639e-02 1.19030726e+00 1.29429340e-01
-1.15184414e+00 -1.03617989e-01 -7.33857572e-01 4.46295649e-01
-5.18874265e-02 -4.29564327e-01 1.11281359e+00 8.83640707e-01
-1.12540102e+00 5.24353623e-01 -9.45835650e-01 -8.95595625e-02
1.92053840e-01 2.68080384e-01 -2.61638671e-01 4.67254698e-01
-1.18196809e+00 7.28786886e-01 7.70140588e-01 -3.20712209e-01
-1.39680994e+00 2.92252004e-01 -9.87561166e-01 1.35618478e-01
7.65279233e-01 -5.23176670e-01 1.67846191e+00 -1.76909769e+00
-1.73348641e+00 1.24643445e+00 -8.82499143e-02 -7.20133126e-01
3.89441140e-02 2.55885571e-01 -2.04598650e-01 3.33269775e-01
1.52050778e-01 3.72602552e-01 7.30911195e-01 -1.59808183e+00
-6.21842861e-01 -2.30066314e-01 3.66098195e-01 3.54247123e-01
2.22613245e-01 5.03845453e-01 5.34193456e-01 -3.11545521e-01
-8.43353942e-02 -8.20508480e-01 -5.91839015e-01 -8.64343047e-01
-6.59193039e-01 -3.72382104e-01 -3.39394659e-01 -1.50428282e-03
7.52122879e-01 -1.84122789e+00 7.08485693e-02 5.14082670e-01
1.02640860e-01 -9.18842033e-02 -4.46371704e-01 4.38468635e-01
-2.40383029e-01 4.32861745e-01 -1.96722120e-01 1.67662218e-01
3.89647275e-01 4.79903221e-01 -8.64247903e-02 1.06343530e-01
2.08858728e-01 9.70291317e-01 -1.13339663e+00 -5.55281281e-01
-7.68975616e-02 -3.75708044e-01 -6.25899732e-01 5.56259394e-01
-7.69854367e-01 2.92356580e-01 -1.08924317e+00 4.81275767e-01
-1.28275707e-01 -1.94791630e-01 7.41825879e-01 6.26841307e-01
1.96927935e-01 8.03682506e-01 -1.18043494e+00 1.19066179e+00
-4.27611135e-02 2.24397793e-01 -1.84863970e-01 -1.14880538e+00
7.03582525e-01 4.58933920e-01 8.69099796e-02 -6.53554738e-01
4.63931352e-01 5.90040758e-02 6.48600042e-01 -7.43336797e-01
1.63761020e-01 -5.71406782e-01 -2.56998509e-01 1.08048546e+00
3.46268654e-01 -3.66723612e-02 1.01223253e-01 3.30618173e-01
1.24207771e+00 4.22692932e-02 4.50034469e-01 -4.61677432e-01
4.71063286e-01 1.01766497e-01 4.73669797e-01 1.21916175e+00
-2.72253513e-01 -2.19568297e-01 8.04753125e-01 -4.16369468e-01
-1.06198800e+00 -1.07350516e+00 1.42304897e-01 1.31917608e+00
9.50265769e-03 -2.91874647e-01 -1.30349791e+00 -9.30261791e-01
-5.08402050e-01 1.46873856e+00 -3.69635016e-01 -4.42626514e-02
-5.61734736e-01 -4.60232884e-01 6.76778793e-01 2.94535816e-01
8.96936432e-02 -1.93694711e+00 -8.93004060e-01 4.14087802e-01
3.87358665e-02 -1.00853252e+00 1.54348891e-02 6.68187320e-01
-9.49853122e-01 -1.14552462e+00 3.09726655e-01 -7.38639474e-01
7.97222912e-01 -3.93876940e-01 1.43076599e+00 1.02590986e-01
2.99281329e-01 8.90613377e-01 -4.01077598e-01 -5.75248837e-01
-1.27483106e+00 -3.96769773e-03 9.71103683e-02 -7.25420266e-02
6.62594140e-01 -5.79723656e-01 7.91616738e-02 1.80319279e-01
-9.32331622e-01 -4.31510732e-02 5.84608436e-01 6.73116505e-01
3.02807361e-01 2.67486721e-01 1.67977214e-01 -1.27685869e+00
9.69057262e-01 -3.22339982e-01 -1.00964558e+00 5.93884885e-01
-5.98005235e-01 9.95101213e-01 6.17558300e-01 -2.11098745e-01
-1.51063371e+00 2.34473273e-01 3.02508920e-01 2.38323465e-01
-7.40618467e-01 4.37140524e-01 -6.48233950e-01 3.79366249e-01
8.90858769e-01 3.78855497e-01 -4.18812223e-03 -2.05844432e-01
5.59895992e-01 3.39502603e-01 2.19350368e-01 -1.38996375e+00
7.86419511e-01 2.71290019e-02 -7.85768330e-02 -4.71252799e-01
-6.94309056e-01 8.79576206e-02 -6.58763707e-01 -3.53793353e-02
6.93788290e-01 -4.55813408e-01 -9.55967069e-01 4.54211161e-02
-1.00520384e+00 -7.04831481e-01 -6.47182524e-01 2.42475986e-01
-1.20085800e+00 2.25186273e-01 -3.83708149e-01 -1.29282606e+00
9.07879323e-02 -1.11797750e+00 7.54805565e-01 5.99061251e-02
-3.33736032e-01 -1.00500071e+00 2.46062696e-01 2.43416294e-01
-1.70727104e-01 -2.91143984e-01 1.16081619e+00 -1.53307986e+00
-8.79321158e-01 1.53264627e-01 3.46403748e-01 1.42499372e-01
-9.70505327e-02 9.72719789e-02 -8.91625047e-01 2.53053337e-01
3.27111244e-01 -4.64308530e-01 1.47529110e-01 2.97340780e-01
6.50021136e-01 -6.25673354e-01 -2.27419376e-01 8.39742273e-02
1.19817758e+00 7.63054073e-01 4.94005531e-01 5.42629302e-01
2.51240551e-01 1.10913217e+00 4.88066673e-01 2.15677470e-01
2.24931657e-01 7.34408796e-02 1.05837628e-01 5.54197907e-01
4.59936708e-01 -7.11580873e-01 4.59815830e-01 3.31355006e-01
-8.92137364e-02 -2.56813854e-01 -1.05943561e+00 5.33537686e-01
-1.57893908e+00 -1.12577677e+00 2.64579475e-01 1.91291726e+00
9.66830552e-01 4.18919712e-01 2.37633035e-01 -5.61810844e-02
8.93783092e-01 -9.91088301e-02 -3.35407019e-01 -4.54299361e-01
-2.10736036e-01 2.76858598e-01 5.52647650e-01 7.53050804e-01
-7.70900965e-01 1.32153368e+00 7.64249372e+00 4.29832578e-01
-9.13100839e-02 -2.53704250e-01 5.40494621e-01 4.86237496e-01
-7.69551098e-01 5.37974596e-01 -5.90895653e-01 1.88275382e-01
1.47481596e+00 -3.55256319e-01 9.02083993e-01 8.55804682e-01
1.08357205e-03 -2.46849582e-01 -1.50591624e+00 2.29585573e-01
-4.33958948e-01 -1.02343488e+00 2.32825756e-01 1.86450690e-01
3.95575881e-01 -8.73116702e-02 -2.99199849e-01 2.44348392e-01
1.49982738e+00 -8.85563195e-01 8.77597272e-01 4.13372934e-01
1.69098735e-01 -6.71610236e-01 4.21662807e-01 8.16687584e-01
-6.01313114e-01 -3.56257766e-01 -3.21530163e-01 -1.75664425e-01
-1.42047729e-03 7.08965659e-02 -1.15828431e+00 1.49916351e-01
-1.19003087e-01 1.44509211e-01 -3.86280358e-01 3.48255873e-01
-7.90147960e-01 9.16387200e-01 -2.69933075e-01 -4.46663201e-01
3.31365108e-01 -5.12726665e-01 7.13709652e-01 8.20841670e-01
-3.16460013e-01 4.09025580e-01 1.96024746e-01 1.09758127e+00
-1.38897924e-02 4.03157361e-02 -9.41884577e-01 -3.69034261e-01
4.48783994e-01 7.53348529e-01 -9.36670184e-01 -6.92716479e-01
-3.03511381e-01 3.79519373e-01 3.38625789e-01 1.87571019e-01
-4.01086062e-01 -4.52999100e-02 3.28952163e-01 -7.42765069e-02
-7.00304704e-03 7.68969133e-02 -1.64018080e-01 -1.30747652e+00
-2.09339038e-01 -1.37406063e+00 5.19781947e-01 -7.17574358e-01
-1.18278384e+00 5.28808951e-01 2.53907561e-01 -5.24136484e-01
-7.74680078e-01 -5.18194199e-01 -3.63858163e-01 5.74515939e-01
-9.80433226e-01 -5.90742648e-01 6.21508479e-01 5.21128833e-01
3.01716357e-01 -6.43927455e-01 1.07459319e+00 -6.80545986e-01
-3.37315559e-01 4.01917905e-01 -2.48687506e-01 5.59143305e-01
-1.73473954e-01 -1.56785786e+00 6.40147626e-01 9.08714354e-01
8.59594822e-01 8.10033023e-01 9.94543433e-01 -6.83465302e-01
-1.36500525e+00 -4.99038398e-01 9.05725956e-01 -8.20776463e-01
8.99089515e-01 -1.44947067e-01 -8.08511138e-01 1.16911268e+00
2.86873221e-01 -3.66376966e-01 6.49565339e-01 9.48558375e-02
-3.39944750e-01 4.53687370e-01 -1.31434345e+00 6.08034909e-01
1.45313227e+00 -6.59354270e-01 -1.36353672e+00 2.19520718e-01
8.88169408e-01 7.59117380e-02 -5.48659384e-01 1.58695072e-01
7.44054615e-02 -9.60505724e-01 6.08149171e-01 -1.27774835e+00
3.23553562e-01 -1.16239168e-01 -2.26579547e-01 -1.14868212e+00
-2.88283885e-01 -8.52597535e-01 1.35577351e-01 9.79712248e-01
7.46633530e-01 -8.71646047e-01 8.62150371e-01 1.22734523e+00
2.26634219e-01 1.10419318e-01 -4.85628515e-01 -8.23491156e-01
1.41375020e-01 -5.86919069e-01 6.20127261e-01 8.23140800e-01
3.88910592e-01 7.16803849e-01 3.04749727e-01 1.13186099e-01
8.60993743e-01 2.12780073e-01 4.37525004e-01 -1.20705938e+00
-8.08767676e-01 -3.68367344e-01 -2.44983822e-01 -8.08240056e-01
8.01231742e-01 -1.11297143e+00 3.03898513e-01 -9.99690354e-01
2.44143397e-01 -5.55531204e-01 -2.68924326e-01 4.27845925e-01
-4.81037311e-02 -9.58068252e-01 1.17101960e-01 3.88876125e-02
-8.55592847e-01 3.05298895e-01 1.21488464e+00 1.68292075e-02
-1.64695397e-01 1.09258287e-01 -9.78197694e-01 1.02650785e+00
7.91644692e-01 -8.83720994e-01 -8.15273702e-01 -3.00405055e-01
6.33794367e-01 4.55194950e-01 1.86014205e-01 -3.79903316e-01
4.96515859e-05 -6.35942400e-01 -3.10114771e-01 3.03862970e-02
-5.64340591e-01 -8.62511814e-01 1.14647143e-01 3.20566922e-01
-1.02084589e+00 -6.55728206e-02 -2.92251617e-01 4.86588985e-01
-1.13836288e-01 -1.07472587e+00 4.42365289e-01 -7.52232432e-01
-4.57177192e-01 -1.92844540e-01 -1.04005194e+00 3.64442319e-01
9.63900745e-01 7.87364319e-03 1.26623958e-01 -2.29346380e-01
-9.03545320e-01 -1.23237759e-01 5.81679344e-01 -1.31585807e-01
2.17093766e-01 -6.63156807e-01 -3.83856624e-01 1.48054212e-01
9.00408402e-02 4.87939082e-02 -4.56981242e-01 -3.80753651e-02
-3.00092757e-01 5.43749630e-01 -2.66353190e-01 -1.34787709e-01
-8.82230759e-01 9.13146079e-01 5.31068087e-01 -5.38661182e-01
-3.99149090e-01 7.28280425e-01 3.09606642e-01 -5.63438296e-01
2.06630155e-01 -2.60215878e-01 -2.42275566e-01 -3.01284343e-01
3.96739274e-01 -2.35252142e-01 -3.42848003e-01 -3.09162140e-01
-6.75459346e-03 -1.33771017e-01 5.50222322e-02 -5.38069546e-01
1.36041701e+00 -5.88302687e-02 -1.85804039e-01 4.02813137e-01
8.04714262e-01 4.06232774e-02 -1.03582478e+00 -5.52681565e-01
8.64306629e-01 -2.01961502e-01 -5.14308274e-01 -7.00164855e-01
-6.71402752e-01 3.84717375e-01 6.98761344e-02 6.43118143e-01
7.54655242e-01 4.43920434e-01 2.81230155e-02 1.12803698e+00
1.02907372e+00 -1.08170354e+00 3.06151062e-02 4.95598376e-01
4.35681283e-01 -9.82228458e-01 -4.67453837e-01 -2.60102659e-01
-9.85160232e-01 8.82084727e-01 2.85797894e-01 -3.90733808e-01
4.05183822e-01 5.66183805e-01 -1.30632073e-01 -5.38970232e-01
-1.08474958e+00 -6.01526976e-01 -2.90827513e-01 1.09899735e+00
2.17364028e-01 5.41692555e-01 -9.24523398e-02 6.65236592e-01
-4.51593786e-01 -4.91868407e-01 6.69987381e-01 9.78667140e-01
-6.82617188e-01 -1.46769929e+00 -3.14611137e-01 3.40531170e-01
-5.61445117e-01 -3.30458343e-01 -7.94796884e-01 5.87474644e-01
-3.30910623e-01 9.11005795e-01 -9.76644456e-02 7.45869661e-03
2.25705817e-01 4.26070839e-01 7.09806919e-01 -1.00150287e+00
-5.57602227e-01 -1.17926016e-01 5.86281419e-01 -5.28127432e-01
-7.66702712e-01 -1.07562518e+00 -1.41035581e+00 3.16535056e-01
-1.58519268e-01 4.22779769e-01 4.26981926e-01 1.14424014e+00
-1.45129427e-01 3.00446242e-01 8.46212387e-01 -1.85017377e-01
-8.18196654e-01 -7.13188291e-01 -5.70099592e-01 5.82459152e-01
-2.37253010e-01 -5.70469677e-01 -5.58172464e-01 3.67627472e-01] | [4.10689640045166, 1.3673323392868042] |
0a94d5a0-7a9a-4d0d-a210-c2f30636f60a | multi-task-learning-network-for-emotion | 2003.01478 | null | https://arxiv.org/abs/2003.01478v2 | https://arxiv.org/pdf/2003.01478v2.pdf | Multi-Task Learning with Auxiliary Speaker Identification for Conversational Emotion Recognition | Conversational emotion recognition (CER) has attracted increasing interests in the natural language processing (NLP) community. Different from the vanilla emotion recognition, effective speaker-sensitive utterance representation is one major challenge for CER. In this paper, we exploit speaker identification (SI) as an auxiliary task to enhance the utterance representation in conversations. By this method, we can learn better speaker-aware contextual representations from the additional SI corpus. Experiments on two benchmark datasets demonstrate that the proposed architecture is highly effective for CER, obtaining new state-of-the-art results on two datasets. | ['Yijiang Liu', 'Meishan Zhang', 'Donghong Ji', 'Jingye Li'] | 2020-03-03 | null | null | null | null | ['emotion-recognition-in-conversation'] | ['natural-language-processing'] | [ 3.73764634e-01 -6.27123564e-02 1.06342316e-01 -7.96700001e-01
-1.18966687e+00 -3.19013417e-01 6.32375062e-01 -2.43296325e-01
-4.71609592e-01 6.84839904e-01 7.21342385e-01 2.63335072e-02
3.77249151e-01 -8.09633285e-02 -1.24546401e-01 -7.84301996e-01
6.56138733e-02 -5.80709949e-02 -2.44685322e-01 -3.81151438e-01
1.61112696e-01 3.21359158e-01 -1.61567509e+00 4.98508424e-01
9.57626224e-01 1.25793898e+00 -1.27884760e-01 5.62419236e-01
-6.24989033e-01 8.47699881e-01 -8.14018190e-01 -2.83879340e-01
-4.30904865e-01 -5.80498874e-01 -7.89562166e-01 2.74547011e-01
-8.76002610e-02 3.07132423e-01 1.12537973e-01 8.54071081e-01
5.87504506e-01 5.18382549e-01 5.86927474e-01 -1.17889380e+00
-3.15222681e-01 7.80577302e-01 -4.70427424e-01 2.72079527e-01
4.95997816e-01 -3.92028153e-01 9.60739076e-01 -1.23120344e+00
2.16981515e-01 1.55342352e+00 9.41831768e-02 8.86912405e-01
-7.72250891e-01 -7.09496796e-01 5.80582321e-01 5.99213362e-01
-1.37134671e+00 -8.05495262e-01 1.30126584e+00 -4.86176573e-02
9.40433979e-01 5.60827374e-01 4.22139876e-02 1.41570115e+00
-2.07180753e-01 1.30150974e+00 9.61723804e-01 -4.86648917e-01
3.75452012e-01 4.73822266e-01 3.48995745e-01 1.16613932e-01
-7.92374969e-01 -2.03038245e-01 -7.02764392e-01 -1.49814010e-01
3.31903584e-02 -7.01031834e-02 -4.63755369e-01 6.94753677e-02
-8.84260893e-01 8.98694813e-01 1.70388818e-01 6.40029669e-01
-6.46909237e-01 -2.09632158e-01 8.36620390e-01 2.68388301e-01
8.97603035e-01 2.82240987e-01 -4.45614964e-01 -6.78495884e-01
-5.24929821e-01 -2.74555594e-01 8.15342188e-01 4.60464060e-01
5.56669354e-01 3.71894091e-01 -1.63329288e-01 1.34638989e+00
7.81648979e-02 3.64470273e-01 8.43200445e-01 -4.56601292e-01
3.95670056e-01 6.97039723e-01 -2.20855288e-02 -9.67517316e-01
-3.06772381e-01 -3.49129438e-01 -1.12884569e+00 -4.87269819e-01
-2.66181976e-01 -4.25843716e-01 -4.18719769e-01 1.78599501e+00
3.44017029e-01 6.33917153e-01 6.43732071e-01 8.60121489e-01
1.09952497e+00 1.16254807e+00 2.61976838e-01 -4.99643654e-01
1.54017365e+00 -1.29094219e+00 -1.09980190e+00 -5.92556119e-01
3.98364484e-01 -5.07868409e-01 9.85247910e-01 4.06104952e-01
-5.46918511e-01 -2.55867809e-01 -8.90718341e-01 1.33225083e-01
-4.28440183e-01 2.41352290e-01 5.73864996e-01 8.54954958e-01
-8.11771214e-01 -2.42342755e-01 -6.12717927e-01 -2.53363997e-01
1.54848173e-01 2.66056024e-02 -3.96336615e-01 6.16282299e-02
-1.53005278e+00 7.94941425e-01 5.12078144e-02 6.58165812e-01
-6.80389345e-01 -2.47431383e-01 -1.22321522e+00 3.72617900e-01
4.90743160e-01 2.90781166e-02 1.29943216e+00 -1.30511844e+00
-2.04188871e+00 6.31128788e-01 -8.48529518e-01 -4.00925457e-01
1.74575672e-02 -2.01561555e-01 -9.78664458e-01 1.54418454e-01
-3.55811775e-01 2.90032417e-01 1.09008479e+00 -1.18189430e+00
-3.66424561e-01 -5.07918537e-01 -3.37201804e-01 3.80732566e-01
-6.15939438e-01 6.45864010e-01 -3.31264406e-01 -4.59105521e-01
8.29058606e-03 -6.69974804e-01 -2.74745017e-01 -6.72850788e-01
-2.54022241e-01 -7.49806285e-01 9.82619464e-01 -7.35970914e-01
1.17657232e+00 -2.37758589e+00 2.32879668e-01 8.13360810e-02
-2.43042290e-01 3.33393008e-01 -2.91473657e-01 2.94289470e-01
-1.84066758e-01 4.98505868e-02 -4.37621832e-01 -8.09041321e-01
1.54574767e-01 9.57391188e-02 -4.19995189e-01 1.63664013e-01
5.54363132e-01 8.28454196e-01 -4.88477945e-01 -2.89461672e-01
3.75304550e-01 8.42850268e-01 -3.17739695e-01 5.19481003e-01
1.69982195e-01 7.29239166e-01 -4.66605157e-01 5.01639605e-01
6.58261180e-01 1.86140895e-01 1.89059377e-01 1.47887528e-01
-3.32475128e-03 2.95857131e-01 -1.14239419e+00 1.62386727e+00
-8.04073393e-01 6.34481668e-01 4.42546487e-01 -1.41511476e+00
1.37066078e+00 5.91403306e-01 1.91468909e-01 -5.86543560e-01
3.64516556e-01 -1.08519197e-01 -2.35011056e-01 -5.96468151e-01
4.85134542e-01 -3.34962368e-01 -5.36311328e-01 2.92077065e-01
1.63660087e-02 5.37214167e-02 -2.36362383e-01 1.20106600e-01
7.05682874e-01 -6.21579289e-01 5.19164503e-01 -1.55590326e-01
1.37417376e+00 -6.05523050e-01 8.25013280e-01 3.10936302e-01
-6.76925361e-01 1.67151943e-01 3.50225389e-01 -1.20548904e-01
-1.19931854e-01 -7.33258426e-01 -2.23791953e-02 1.46060610e+00
-4.39176075e-02 -2.62970060e-01 -8.53012621e-01 -8.36823285e-01
-5.33289552e-01 8.85934234e-01 -6.90026045e-01 -3.52351397e-01
-4.90578949e-01 -4.54048991e-01 4.63762373e-01 6.11692488e-01
5.22189677e-01 -1.42011023e+00 -1.48185506e-01 2.75371253e-01
-6.77636981e-01 -1.35364580e+00 -6.76219463e-01 5.77699989e-02
-3.90003204e-01 -5.73196530e-01 -6.63403630e-01 -9.42425311e-01
4.07164067e-01 3.23277652e-01 9.73880231e-01 -3.67919803e-01
-1.84703083e-03 5.43760121e-01 -6.58309042e-01 -5.92925787e-01
-4.05496925e-01 1.56588349e-02 8.21930319e-02 6.55310631e-01
7.89704144e-01 -2.87968874e-01 -3.09300274e-01 4.22703683e-01
-7.77548671e-01 -3.23025405e-01 3.30747545e-01 7.95407832e-01
2.07866833e-01 -8.12275857e-02 1.34728706e+00 -6.53441191e-01
9.13973808e-01 -3.86464328e-01 -7.95912519e-02 3.02821547e-01
3.11415326e-02 -1.33086704e-02 5.71932256e-01 -4.03227448e-01
-1.87949121e+00 5.49918562e-02 -5.75801432e-01 -6.91945031e-02
-5.76995969e-01 7.53990769e-01 -7.19527960e-01 1.89698607e-01
3.19648623e-01 5.09865582e-01 -1.00045457e-01 -4.69649881e-01
3.44019890e-01 1.33385968e+00 5.14571548e-01 -5.24802744e-01
5.53238355e-02 2.77734458e-01 -7.58439183e-01 -1.27554536e+00
-8.71551335e-01 -8.86619866e-01 -3.41276050e-01 -3.16760212e-01
8.17161620e-01 -9.38979864e-01 -8.47421527e-01 3.59535992e-01
-1.24596727e+00 2.76564717e-01 5.51410727e-02 6.30526304e-01
-2.04745919e-01 4.46464151e-01 -4.72264528e-01 -1.49964476e+00
-6.78188920e-01 -1.28438890e+00 1.12753153e+00 3.63491058e-01
-1.17296971e-01 -8.42356145e-01 2.99892388e-02 5.94236016e-01
6.12888396e-01 -1.45110488e-01 4.60738480e-01 -1.02485192e+00
8.37031901e-02 -3.52775335e-01 -4.02434058e-02 8.55477512e-01
1.36650383e-01 -3.36568326e-01 -1.47069979e+00 9.44309235e-02
4.77824181e-01 -4.11917239e-01 9.73709643e-01 5.99239618e-02
1.34535074e+00 -2.60202616e-01 1.28892027e-02 4.49664965e-02
7.73702264e-01 2.24650428e-01 6.04847312e-01 -1.43902153e-01
3.46775889e-01 9.12635207e-01 6.94158792e-01 5.44519663e-01
4.36978072e-01 5.31446457e-01 1.54808402e-01 1.01406395e-01
5.20435393e-01 1.08693048e-01 8.00095260e-01 1.18510842e+00
2.54346430e-01 -1.61781579e-01 -7.92195499e-01 5.10655880e-01
-1.92773354e+00 -8.92305553e-01 2.70005465e-01 1.74204636e+00
8.05555165e-01 -3.70627820e-01 -1.30109847e-01 6.36709481e-02
1.02244198e+00 5.69941461e-01 -5.30843318e-01 -8.20314169e-01
-1.38519451e-01 -1.33680468e-02 -4.37647402e-01 4.00514573e-01
-1.16963100e+00 1.13659751e+00 5.65854645e+00 7.72055447e-01
-1.26997292e+00 1.50018677e-01 7.29631841e-01 2.07675621e-01
-2.55733561e-02 -5.66282511e-01 -7.94929981e-01 2.88828135e-01
1.27092016e+00 -3.76191616e-01 3.32953125e-01 1.18666685e+00
2.73708820e-01 1.86572805e-01 -9.72232163e-01 1.34865415e+00
7.12229073e-01 -8.90250921e-01 -1.29949227e-01 -4.06028628e-01
5.25955260e-01 -1.89334288e-01 -9.84657332e-02 1.10912657e+00
-1.21805169e-01 -8.78583789e-01 2.19220668e-01 2.78693140e-01
4.62265134e-01 -1.08454883e+00 1.11060226e+00 4.34302926e-01
-1.32656813e+00 -5.12959398e-02 -2.52575070e-01 -8.59526079e-03
4.44374472e-01 5.20494580e-01 -7.47593820e-01 5.28455436e-01
5.86760521e-01 5.55682182e-01 -2.78716713e-01 5.29170334e-01
-3.14645946e-01 8.99095774e-01 -1.73696145e-01 -3.73252779e-01
2.28312418e-01 -1.32306173e-01 5.71472168e-01 1.45966721e+00
6.80728927e-02 2.84435719e-01 1.15530089e-01 4.10163909e-01
-4.30535406e-01 5.55619717e-01 -3.94733071e-01 -2.76847512e-01
3.55112910e-01 1.47945583e+00 -1.79142654e-01 -3.84322912e-01
-4.38426822e-01 1.02491426e+00 4.37633604e-01 5.01312971e-01
-7.42381275e-01 -2.72555590e-01 9.01175082e-01 -9.38333392e-01
2.94765353e-01 -8.92170146e-02 1.20232843e-01 -1.40899467e+00
1.85534153e-02 -1.00807893e+00 4.86150891e-01 -4.28731143e-01
-1.39590502e+00 8.64246786e-01 -4.29619104e-01 -9.40049529e-01
-5.16199350e-01 -4.06097084e-01 -9.10749674e-01 7.32368469e-01
-1.74875474e+00 -9.52299178e-01 -2.05570415e-01 7.12086499e-01
1.00159967e+00 -1.46663830e-01 1.21044528e+00 2.03175768e-01
-8.30396712e-01 7.50281870e-01 -1.30541099e-03 1.41043872e-01
8.14059913e-01 -1.01718533e+00 5.56029426e-03 6.83152318e-01
1.63460284e-01 5.28001904e-01 5.92793405e-01 -1.88589375e-02
-1.36344552e+00 -9.55808997e-01 9.56186593e-01 -6.70748502e-02
5.24813950e-01 -5.74121237e-01 -1.05257821e+00 6.15482807e-01
5.20046711e-01 -3.51498574e-01 1.17279661e+00 4.95496780e-01
-4.26248014e-01 -2.09777772e-01 -1.14875901e+00 4.44345325e-01
5.07363975e-01 -8.24481487e-01 -7.07903028e-01 -5.04722036e-02
8.95625591e-01 -4.86927852e-02 -5.64143717e-01 3.40523750e-01
2.45465308e-01 -5.81846297e-01 8.05979609e-01 -6.39596641e-01
5.88822626e-02 5.21573573e-02 -3.35431844e-01 -1.70847631e+00
2.83433288e-01 -3.95833462e-01 -6.17954023e-02 1.74111843e+00
4.82528687e-01 -8.06482017e-01 5.52863121e-01 8.11956048e-01
-2.36220494e-01 -4.92978513e-01 -1.15835035e+00 -5.51088572e-01
-1.23244405e-01 -7.39621818e-01 7.69292414e-01 1.04941964e+00
5.24233699e-01 8.34021926e-01 -5.43838501e-01 1.47754729e-01
1.86749548e-01 1.05392471e-01 7.19687700e-01 -1.09973776e+00
1.19863681e-01 -3.96901786e-01 -3.74227017e-01 -1.13812256e+00
9.66964960e-01 -6.27267718e-01 5.37140965e-01 -1.24114931e+00
3.31649706e-02 2.42038444e-01 -6.24933481e-01 2.24432886e-01
-4.77009237e-01 -1.98870793e-01 5.74882366e-02 -4.33687061e-01
-9.31862950e-01 1.38182604e+00 7.37109363e-01 -3.15424353e-01
-2.89756209e-01 -3.62188518e-02 -9.54220414e-01 7.13511825e-01
9.59955990e-01 -1.32004574e-01 -1.76175594e-01 -1.80738587e-02
-2.21751377e-01 1.39823571e-01 -1.32599741e-01 -6.68653905e-01
1.35978356e-01 -9.76450667e-02 -1.20600447e-01 -6.71907365e-01
8.05667996e-01 -8.81818354e-01 -5.89645982e-01 -1.34862095e-01
-6.60340548e-01 -4.47455049e-01 3.78143251e-01 6.69359982e-01
-8.01771760e-01 -1.87558964e-01 5.71178377e-01 1.59216076e-01
-1.10829127e+00 1.30238160e-01 -6.17316246e-01 2.51425803e-02
8.13379407e-01 4.47736025e-01 -2.93910682e-01 -7.44413316e-01
-7.23412991e-01 4.99388963e-01 -3.61723363e-01 8.57640982e-01
8.41393113e-01 -1.22803450e+00 -9.11320210e-01 1.45691335e-01
4.41586554e-01 -2.22524330e-01 8.25806558e-01 7.38422990e-01
3.66180867e-01 4.79359806e-01 2.90792286e-01 -5.96458197e-01
-1.52711678e+00 4.76368099e-01 1.82640761e-01 -2.23767340e-01
-1.77320525e-01 9.80680883e-01 4.13112015e-01 -8.25163364e-01
4.69281912e-01 -4.49778512e-02 -7.40865469e-01 1.82061508e-01
1.13266361e+00 1.16406366e-01 6.56713769e-02 -9.31843281e-01
-6.79136992e-01 -3.37764644e-03 -2.81470954e-01 -2.54038185e-01
1.33513141e+00 -5.71528792e-01 -8.97822827e-02 8.10625732e-01
1.41853690e+00 -2.22286321e-02 -7.84705937e-01 -5.60634375e-01
2.28913113e-01 -2.06968293e-01 2.23529801e-01 -7.30586171e-01
-8.49977195e-01 1.11246920e+00 4.70501930e-01 2.05243751e-01
1.18119228e+00 1.78177580e-01 8.08705866e-01 6.48930490e-01
1.19367205e-01 -1.17148709e+00 4.67103049e-02 6.42988980e-01
1.46126902e+00 -1.57613814e+00 -6.35390043e-01 -5.05293369e-01
-1.27223742e+00 7.78904736e-01 5.93543887e-01 2.13671282e-01
6.73453152e-01 4.62886244e-02 4.67429280e-01 9.18107182e-02
-8.37522626e-01 -1.33332267e-01 2.74203569e-01 3.95493507e-01
6.06616497e-01 3.29550862e-01 -2.46790811e-01 1.30470729e+00
-2.53694411e-02 -3.84902418e-01 1.85626194e-01 7.08194673e-01
-4.21391904e-01 -1.05470634e+00 -3.53828758e-01 1.18877150e-01
-4.27039713e-01 -1.33811623e-01 -7.23989964e-01 1.03614792e-01
-7.87824035e-01 1.65703976e+00 -1.13411844e-02 -2.60491431e-01
2.69652009e-01 6.36323333e-01 -1.74461886e-01 -5.81595004e-01
-6.25288129e-01 9.95024964e-02 4.89016056e-01 -7.15171337e-01
-6.80299461e-01 -6.19493544e-01 -1.24794555e+00 1.23029269e-01
-3.82453322e-01 7.05293179e-01 1.02943635e+00 1.17024958e+00
5.94023883e-01 6.27073228e-01 1.23048127e+00 -7.92383075e-01
-4.88630712e-01 -1.27232349e+00 -6.31686091e-01 3.92276704e-01
3.13583910e-01 -5.16672194e-01 -7.82862186e-01 -1.53932720e-01] | [13.188431739807129, 5.939605236053467] |
6b7ee39e-5a44-4c99-887a-db092c46eb49 | defix-detecting-and-fixing-failure-scenarios | 2210.16567 | null | https://arxiv.org/abs/2210.16567v1 | https://arxiv.org/pdf/2210.16567v1.pdf | DeFIX: Detecting and Fixing Failure Scenarios with Reinforcement Learning in Imitation Learning Based Autonomous Driving | Safely navigating through an urban environment without violating any traffic rules is a crucial performance target for reliable autonomous driving. In this paper, we present a Reinforcement Learning (RL) based methodology to DEtect and FIX (DeFIX) failures of an Imitation Learning (IL) agent by extracting infraction spots and re-constructing mini-scenarios on these infraction areas to train an RL agent for fixing the shortcomings of the IL approach. DeFIX is a continuous learning framework, where extraction of failure scenarios and training of RL agents are executed in an infinite loop. After each new policy is trained and added to the library of policies, a policy classifier method effectively decides on which policy to activate at each step during the evaluation. It is demonstrated that even with only one RL agent trained on failure scenario of an IL agent, DeFIX method is either competitive or does outperform state-of-the-art IL and RL based autonomous urban driving benchmarks. We trained and validated our approach on the most challenging map (Town05) of CARLA simulator which involves complex, realistic, and adversarial driving scenarios. The source code is publicly available at https://github.com/data-and-decision-lab/DeFIX | ['Nazim Kemal Ure', 'Ferhat Yurdakul', 'Halil Durmus', 'Feyza Eksen', 'Resul Dagdanov'] | 2022-10-29 | null | null | null | null | ['carla-map-leaderboard'] | ['robots'] | [-4.07181717e-02 2.78438449e-01 -1.08024783e-01 -2.70069122e-01
-6.82473600e-01 -6.44351304e-01 7.94155598e-01 -3.39636300e-03
-7.21858263e-01 9.64718878e-01 -5.39083838e-01 -8.12114656e-01
-5.35720102e-02 -8.09696913e-01 -1.19383264e+00 -6.92166865e-01
-3.59474212e-01 7.73337185e-01 6.07956707e-01 -4.97129560e-01
2.52216429e-01 5.97235501e-01 -1.93249631e+00 -1.74359769e-01
9.16504443e-01 6.33157790e-01 1.07802652e-01 9.28140700e-01
4.30483758e-01 9.70590234e-01 -6.26678526e-01 5.91628850e-02
5.55543125e-01 2.13143267e-02 -4.88879889e-01 -2.64364541e-01
1.22643411e-01 -3.38744432e-01 -5.21754563e-01 8.47598553e-01
5.09132028e-01 1.68439165e-01 5.02253771e-01 -1.82603037e+00
3.29987675e-01 5.27064025e-01 -1.64998963e-01 1.66022554e-01
4.24110591e-02 9.64636445e-01 3.18620116e-01 -5.20215571e-01
4.47423875e-01 1.08946407e+00 4.24881369e-01 5.65342784e-01
-8.88730347e-01 -8.02323639e-01 1.34663612e-01 4.11842406e-01
-1.24094594e+00 -6.11172140e-01 5.78212440e-01 -4.34751600e-01
1.12936234e+00 -4.02230886e-04 5.18622041e-01 1.22246170e+00
5.13327241e-01 8.34346414e-01 1.16874528e+00 -6.20137788e-02
7.47712553e-01 1.11768954e-01 -2.22095028e-01 8.85442674e-01
3.86075750e-02 1.14839804e+00 -1.02371350e-01 5.82203716e-02
2.10418552e-01 -5.83461225e-01 3.40783268e-01 -4.16101128e-01
-1.15599167e+00 7.40666270e-01 5.05310297e-01 -1.68287486e-01
-5.77840865e-01 5.21706104e-01 6.16061866e-01 4.46506798e-01
-3.33425850e-02 4.17278975e-01 -2.59098828e-01 -5.29171228e-01
-4.83840168e-01 5.09127319e-01 7.18086720e-01 8.73708248e-01
9.36721265e-01 3.33902359e-01 -1.63950339e-01 3.71138960e-01
1.03472389e-01 7.51084447e-01 1.12617843e-01 -1.24581683e+00
4.41834152e-01 4.75867927e-01 4.20780778e-01 -6.33491516e-01
-6.38871610e-01 -2.44535223e-01 -3.78621191e-01 1.18380964e+00
2.51423657e-01 -6.28592730e-01 -8.64145100e-01 1.65786827e+00
5.20539165e-01 6.30425096e-01 5.31853855e-01 6.75148427e-01
2.72893280e-01 6.39188111e-01 1.17983788e-01 1.88770875e-01
9.37562764e-01 -8.62181067e-01 -3.01799834e-01 -7.36269712e-01
8.13330472e-01 -4.12787497e-01 7.63705611e-01 4.66136634e-01
-8.17576170e-01 -7.01494992e-01 -1.43647301e+00 6.07990742e-01
-6.51544869e-01 8.68301168e-02 2.84024566e-01 3.01324993e-01
-1.05554318e+00 3.62532407e-01 -7.48753428e-01 -1.90759569e-01
3.79860282e-01 5.35656631e-01 -3.09462100e-01 -1.04290135e-01
-1.41882324e+00 1.31146252e+00 5.52361012e-01 6.24248274e-02
-1.75008440e+00 -5.16551852e-01 -9.05168116e-01 -4.79547858e-01
6.40225172e-01 -2.12261602e-01 1.38049805e+00 -7.80588269e-01
-1.79206777e+00 5.89324355e-01 2.24754766e-01 -9.77965713e-01
9.31101263e-01 -1.73522368e-01 -4.47914690e-01 -1.99661657e-01
2.20140576e-01 9.93906558e-01 1.11346161e+00 -1.57482946e+00
-1.07397878e+00 2.39142440e-02 2.50158429e-01 2.29513511e-01
7.38316655e-01 -4.53734457e-01 -1.92387044e-01 -2.79145315e-02
-4.92092282e-01 -1.20772302e+00 -2.31984347e-01 -4.43366379e-01
-3.86824071e-01 -3.29696000e-01 9.53949094e-01 -3.49379390e-01
9.20511544e-01 -2.08918405e+00 2.73967478e-02 3.82066548e-01
-8.50923434e-02 5.88352203e-01 -3.20185274e-01 5.32695711e-01
3.30334127e-01 -2.75296420e-01 -3.85535181e-01 -1.86208919e-01
2.39548936e-01 6.09842241e-01 -4.17926550e-01 5.65560699e-01
2.29938567e-01 9.50259149e-01 -1.11424005e+00 -3.38914871e-01
8.94528449e-01 1.52278990e-01 -3.72781575e-01 4.00958300e-01
-3.52197140e-01 7.60688126e-01 -5.73665380e-01 4.76312310e-01
6.03569031e-01 5.16881406e-01 -2.60709703e-01 4.78699833e-01
-3.67662758e-01 1.44296400e-02 -1.07770002e+00 1.31621695e+00
-7.59413540e-01 6.83943450e-01 -9.47499722e-02 -1.00865304e+00
1.05414534e+00 6.16565999e-03 3.80803734e-01 -1.14207923e+00
2.51931220e-01 4.34240609e-01 7.95777366e-02 -7.68508315e-01
2.46095583e-01 3.95420492e-01 -5.98513722e-01 1.23132013e-01
-1.49367765e-01 -2.55489260e-01 3.31907988e-01 1.42197516e-02
1.51679671e+00 4.66150999e-01 -5.00852391e-02 -1.21940427e-01
9.47890639e-01 3.21953893e-01 5.20542145e-01 1.10356688e+00
-7.43325472e-01 -1.61261007e-01 5.54612458e-01 -6.33315921e-01
-1.01282191e+00 -1.15171051e+00 1.11731969e-01 1.02290547e+00
4.96186435e-01 -4.83337007e-02 -7.06406116e-01 -9.38454092e-01
3.66421431e-01 1.30562842e+00 -7.11662829e-01 -4.95355844e-01
-9.28981066e-01 -2.57815629e-01 8.95827234e-01 1.56876162e-01
7.09099710e-01 -1.57229316e+00 -1.31868446e+00 4.08596158e-01
8.25309455e-02 -1.11341238e+00 6.25204965e-02 2.69328773e-01
-2.85213739e-02 -1.28929722e+00 1.04967497e-01 -6.59724891e-01
5.70907950e-01 -6.52206838e-02 1.03085470e+00 8.53709877e-02
-7.44887814e-02 1.06371127e-01 -1.81472853e-01 -6.07504547e-01
-1.19871843e+00 1.39794931e-01 6.82453066e-02 -1.30468562e-01
2.68294543e-01 -2.71062791e-01 -5.47926784e-01 6.79414988e-01
-4.56785619e-01 -7.03232596e-03 7.04854906e-01 6.01188064e-01
4.29419011e-01 1.95664033e-01 8.37484062e-01 -4.48614508e-01
5.59350669e-01 -6.56144023e-01 -1.25873423e+00 -5.78843653e-02
-5.10549247e-01 3.32186192e-01 8.55219781e-01 -2.14985758e-01
-7.72784352e-01 2.73423970e-01 -2.45362818e-01 -4.06537771e-01
-5.67682922e-01 -7.16685951e-02 8.17080885e-02 -5.63559309e-02
8.04885268e-01 2.25083917e-01 3.28499347e-01 3.11625153e-01
3.22668582e-01 5.26510477e-01 6.61691964e-01 -5.24658144e-01
9.78074849e-01 2.77644008e-01 6.01341836e-02 -4.01823044e-01
-2.75440097e-01 -2.26413131e-01 -4.41395938e-01 -8.65817308e-01
7.49877870e-01 -7.90505469e-01 -1.02475989e+00 5.54322839e-01
-7.00311601e-01 -1.12174439e+00 -2.95804918e-01 2.39024311e-01
-9.76497173e-01 -2.81404167e-01 8.04460943e-02 -8.76867354e-01
8.63404199e-02 -1.52813768e+00 9.67739940e-01 2.60740966e-01
1.42847642e-01 -7.86103785e-01 3.72314095e-01 8.79682973e-02
5.13836324e-01 5.13974845e-01 4.41507399e-01 -4.09544766e-01
-4.81206656e-01 -2.53195852e-01 2.94305056e-01 2.11709917e-01
-2.77661473e-01 2.06955262e-02 -9.31186855e-01 -5.14722347e-01
-5.65383136e-01 -5.95608234e-01 6.86816871e-01 2.10012749e-01
8.62512290e-01 -2.67149270e-01 -4.60388541e-01 4.18178409e-01
1.29911852e+00 4.82403100e-01 7.52626181e-01 8.18981647e-01
2.62503535e-01 6.66913927e-01 1.09103596e+00 2.26023182e-01
5.28430581e-01 5.58570683e-01 9.61958289e-01 8.49683955e-03
1.07520148e-01 -3.87228608e-01 8.50945532e-01 -1.11798465e-01
3.60428065e-01 -2.51597971e-01 -1.00743163e+00 5.27191579e-01
-2.08373046e+00 -8.84591281e-01 1.84439912e-01 2.33001924e+00
4.66778696e-01 7.87728667e-01 1.86054260e-01 1.95061505e-01
4.40031141e-01 4.13441807e-02 -9.58803594e-01 -7.18264401e-01
8.42810422e-02 -7.85624832e-02 1.02594578e+00 9.69287813e-01
-1.24805617e+00 1.45498955e+00 5.16519403e+00 7.53042042e-01
-1.20155060e+00 -3.13097388e-02 4.15160567e-01 2.14822754e-01
1.80440113e-01 6.21778667e-02 -7.57561624e-01 4.74021465e-01
1.16966569e+00 1.11787528e-01 7.40833521e-01 7.75821924e-01
5.80330253e-01 -4.94727850e-01 -9.48802114e-01 4.96906847e-01
-4.08120900e-01 -1.13144088e+00 -5.87552786e-01 -6.81760386e-02
5.25006473e-01 6.12378955e-01 8.72773528e-02 1.01910734e+00
8.19123983e-01 -1.06279504e+00 9.21831369e-01 4.33986604e-01
5.88893175e-01 -1.09821057e+00 8.44748199e-01 6.28249586e-01
-8.79449546e-01 -4.48215306e-01 -1.82924882e-01 3.65710258e-02
4.97036502e-02 -1.30917728e-01 -1.24210823e+00 4.88286912e-01
5.93359411e-01 5.21872938e-01 -6.78309798e-01 8.33184898e-01
-6.76982880e-01 7.38164365e-01 -4.07537103e-01 -6.72613084e-02
7.61529863e-01 4.54089083e-02 8.52426112e-01 9.84543443e-01
-4.25093509e-02 -3.11745882e-01 3.35928798e-01 6.74560547e-01
4.04061288e-01 -4.59130287e-01 -9.76892292e-01 5.77943087e-01
6.18318141e-01 1.15788198e+00 -3.68588477e-01 -2.97518194e-01
-7.09295496e-02 5.12979686e-01 4.65174258e-01 3.60743910e-01
-1.40219533e+00 -2.40276948e-01 8.13222945e-01 5.59168262e-03
3.06990236e-01 -1.89462692e-01 -3.30212303e-02 -4.58161801e-01
-9.57470164e-02 -7.74429440e-01 1.01493999e-01 -5.53835630e-01
-6.81098104e-01 6.43259645e-01 1.66894212e-01 -1.38064849e+00
-5.87414145e-01 -2.99546182e-01 -8.26138139e-01 5.57979822e-01
-1.87049246e+00 -8.71788800e-01 -3.59077185e-01 5.33530056e-01
6.45881653e-01 -5.07572114e-01 4.78732646e-01 -6.32991258e-04
-7.27649927e-01 4.81906056e-01 -9.63480994e-02 -2.58678079e-01
5.19813001e-01 -1.24059057e+00 5.34225285e-01 9.82944846e-01
-4.67831224e-01 -3.35547417e-01 1.00863004e+00 -6.19745731e-01
-1.43323839e+00 -1.52573931e+00 2.94077307e-01 -3.98313016e-01
4.71716046e-01 -4.66506898e-01 -6.69466257e-01 6.59760833e-01
2.85235524e-01 5.24482317e-02 -1.97250560e-01 -6.46184146e-01
1.49911702e-01 -2.24835292e-01 -1.28199041e+00 6.24767721e-01
6.87852085e-01 3.46949473e-02 -2.93971926e-01 2.60462850e-01
4.86526728e-01 -5.76174080e-01 -4.13371384e-01 5.64123809e-01
2.54006207e-01 -9.93428648e-01 7.99398363e-01 -3.11027229e-01
-6.47199824e-02 -7.44575024e-01 1.70818061e-01 -1.40982318e+00
-1.98640698e-03 -8.86561930e-01 -2.79888455e-02 7.11622298e-01
6.04750812e-01 -1.00780082e+00 6.41453922e-01 9.95294973e-02
-3.86162877e-01 -7.78963149e-01 -1.33535051e+00 -8.14869165e-01
2.55116254e-01 -6.05785668e-01 5.77851653e-01 3.01142544e-01
-2.72693127e-01 3.28157283e-02 -3.26788545e-01 6.40961111e-01
6.90315247e-01 -4.82490838e-01 1.22016108e+00 -6.74103975e-01
9.70095024e-03 -4.30502445e-01 -4.46857750e-01 -6.32468402e-01
5.98415911e-01 -7.63032198e-01 5.76168776e-01 -1.24865556e+00
-7.55598068e-01 -9.90451813e-01 -2.25619495e-01 6.83851302e-01
2.81330142e-02 -1.93342611e-01 7.95009211e-02 2.31099203e-02
-8.03674400e-01 5.61847448e-01 9.15416718e-01 -3.79195660e-01
-3.94127935e-01 3.50843132e-01 -4.62147258e-02 4.95912820e-01
1.52786231e+00 -4.54994798e-01 -5.32159090e-01 -1.53358638e-01
6.08620606e-02 1.36199832e-01 6.38284266e-01 -1.58413303e+00
2.96565026e-01 -1.94675013e-01 -4.56820987e-03 -6.42051101e-01
9.01499242e-02 -7.69644976e-01 -6.26200140e-02 1.02662218e+00
-2.63919830e-01 3.01111639e-01 6.80991888e-01 4.72925246e-01
7.55308196e-02 -1.90009032e-05 8.85270536e-01 1.50632903e-01
-1.33048379e+00 7.75832981e-02 -7.93651700e-01 7.49608949e-02
1.77314270e+00 -6.22321442e-02 -3.30774009e-01 -1.67365998e-01
-3.91045958e-01 1.05527079e+00 3.29367429e-01 8.08825672e-01
7.41256416e-01 -9.25740480e-01 -1.00167978e+00 5.63644767e-01
2.55998701e-01 -1.08603993e-02 1.65932834e-01 7.59838283e-01
-6.46442056e-01 2.23833978e-01 -4.43111032e-01 -7.88446367e-01
-8.50610793e-01 7.53337264e-01 8.31991613e-01 -2.78929949e-01
-6.79673374e-01 3.73645931e-01 -8.91236365e-02 -8.11600745e-01
2.76962459e-01 -1.48165733e-01 -3.28827262e-01 -5.16390026e-01
1.87046796e-01 4.02127653e-01 6.44735172e-02 -7.87083030e-01
-4.86834317e-01 2.28714541e-01 1.69129312e-01 -3.54148567e-01
8.69117379e-01 2.00767405e-02 5.37104964e-01 2.28349507e-01
8.07668507e-01 -3.80975157e-01 -1.95691562e+00 2.07892075e-01
-9.97402668e-02 -5.42378016e-02 -6.27159402e-02 -1.06606162e+00
-8.32303345e-01 5.23225963e-01 7.72394657e-01 9.99672897e-03
7.87700117e-01 -1.92292735e-01 5.13480246e-01 3.74809265e-01
6.31409228e-01 -1.36009729e+00 -7.83616975e-02 6.55016065e-01
1.05400813e+00 -1.50601459e+00 -4.18330640e-01 1.81018338e-01
-8.64459217e-01 7.73001254e-01 1.06559038e+00 -5.31892121e-01
5.53610384e-01 4.42417413e-01 1.98447317e-01 -1.37279391e-01
-9.79722381e-01 -2.93914437e-01 -3.05261791e-01 7.28308856e-01
-6.73053563e-01 4.13749009e-01 9.70464349e-02 -2.28845894e-01
-4.38912481e-01 -1.10897854e-01 5.27715981e-01 1.21283436e+00
-6.16025090e-01 -7.71242559e-01 -2.44334057e-01 1.58084437e-01
1.16540633e-01 4.87012714e-01 -5.71830496e-02 1.08397436e+00
4.94596541e-01 1.17528856e+00 4.98912185e-02 -6.82581484e-01
5.93022466e-01 -1.46646708e-01 1.40131146e-01 -6.41886219e-02
-6.26281500e-01 -4.89774704e-01 2.76157975e-01 -1.02987719e+00
9.60469060e-03 -6.98339999e-01 -1.58535969e+00 -3.13948572e-01
1.33628383e-01 5.49629033e-02 7.89830863e-01 9.14228141e-01
5.28293967e-01 7.01746225e-01 1.02925026e+00 -9.65568781e-01
-6.55329943e-01 -6.92102194e-01 1.06623784e-01 1.79391980e-01
6.28192425e-01 -1.00778091e+00 -4.49584007e-01 -4.42545027e-01] | [5.093986511230469, 1.2593127489089966] |
e53008f7-01fd-4ac4-a0c4-b0978f5a0022 | transformer-based-models-and-hardware | 2304.10891 | null | https://arxiv.org/abs/2304.10891v1 | https://arxiv.org/pdf/2304.10891v1.pdf | Transformer-based models and hardware acceleration analysis in autonomous driving: A survey | Transformer architectures have exhibited promising performance in various autonomous driving applications in recent years. On the other hand, its dedicated hardware acceleration on portable computational platforms has become the next critical step for practical deployment in real autonomous vehicles. This survey paper provides a comprehensive overview, benchmark, and analysis of Transformer-based models specifically tailored for autonomous driving tasks such as lane detection, segmentation, tracking, planning, and decision-making. We review different architectures for organizing Transformer inputs and outputs, such as encoder-decoder and encoder-only structures, and explore their respective advantages and disadvantages. Furthermore, we discuss Transformer-related operators and their hardware acceleration schemes in depth, taking into account key factors such as quantization and runtime. We specifically illustrate the operator level comparison between layers from convolutional neural network, Swin-Transformer, and Transformer with 4D encoder. The paper also highlights the challenges, trends, and current insights in Transformer-based models, addressing their hardware deployment and acceleration issues within the context of long-term autonomous driving applications. | ['Xi Chen', 'Zheng Liu', 'Juan Zhong'] | 2023-04-21 | null | null | null | null | ['lane-detection'] | ['computer-vision'] | [ 1.39260516e-01 6.85486523e-03 -4.04741406e-01 -5.73953211e-01
-3.69668871e-01 -4.66042727e-01 4.77479249e-01 -2.41619244e-01
-3.67720127e-01 2.13890880e-01 -2.40226865e-01 -8.52544665e-01
2.03514516e-01 -7.85437942e-01 -7.81369984e-01 -5.81884086e-01
-6.94988817e-02 2.80495018e-01 5.35957634e-01 -5.19751728e-01
1.08819149e-01 6.64669394e-01 -2.26754117e+00 1.19506262e-01
8.63839090e-01 1.59655404e+00 4.50656861e-01 6.89552367e-01
4.21419628e-02 9.08736408e-01 -4.14352417e-01 -6.30131662e-01
3.91396403e-01 1.87059477e-01 -2.34579742e-01 -7.44811893e-02
5.57445645e-01 -4.22531098e-01 -7.56845474e-01 9.28645074e-01
4.41402018e-01 -3.02676231e-01 3.89113724e-01 -1.69294548e+00
-1.73132583e-01 3.65787983e-01 -1.05885543e-01 3.20944905e-01
-3.69341016e-01 3.24333906e-01 7.14849591e-01 -9.47543859e-01
1.97031826e-01 1.18840218e+00 9.89171088e-01 2.62830943e-01
-5.85602641e-01 -8.24477375e-01 3.13125970e-03 9.08262253e-01
-1.32977295e+00 -6.92085207e-01 4.58014220e-01 -4.56101567e-01
1.66607177e+00 -1.15265235e-01 8.84263575e-01 6.18193865e-01
9.93529379e-01 1.04027259e+00 5.92321396e-01 5.66489026e-02
1.09948792e-01 7.37297982e-02 1.00682549e-01 9.03019190e-01
3.10175806e-01 4.22900915e-01 -6.57023787e-01 5.22168219e-01
4.50050741e-01 -3.77637357e-01 3.77914608e-01 -4.80342448e-01
-9.52180445e-01 7.85134852e-01 3.81338447e-01 -2.51515269e-01
-1.89253286e-01 5.45248568e-01 9.75580931e-01 2.40297154e-01
-7.82507583e-02 -1.63926348e-01 -3.74779761e-01 -4.26513284e-01
-8.55968654e-01 3.87255758e-01 3.18708479e-01 1.75984538e+00
8.72947633e-01 6.03471816e-01 -1.20008789e-01 5.24105430e-01
2.21843034e-01 7.36370564e-01 4.16068941e-01 -1.02301037e+00
6.04634702e-01 3.71028572e-01 -3.20303142e-01 -5.24408221e-01
-4.44221616e-01 -6.62067354e-01 -7.59629607e-01 4.84652013e-01
-3.37261736e-01 9.04107019e-02 -9.40036595e-01 1.07568336e+00
-1.34609463e-02 1.81735512e-02 1.72048271e-01 6.25735760e-01
1.04149938e+00 3.59815300e-01 4.00005765e-02 2.40248099e-01
1.58980417e+00 -1.30977201e+00 -9.72236037e-01 -7.13719964e-01
7.30997086e-01 -6.77744508e-01 5.55723071e-01 1.67244151e-01
-1.06317580e+00 -9.97658432e-01 -1.55422068e+00 -7.95629799e-01
-6.93275690e-01 6.23538077e-01 6.79760158e-01 7.21206367e-01
-1.24620807e+00 3.26166928e-01 -1.13704598e+00 -2.99202353e-01
4.78489459e-01 6.44136310e-01 6.04983307e-02 2.32708916e-01
-1.25810134e+00 1.22814167e+00 2.99480766e-01 3.96451175e-01
-1.18698168e+00 -4.39846516e-01 -1.02376032e+00 6.56188838e-03
3.58966785e-03 -4.57923651e-01 1.87945652e+00 -2.53615290e-01
-1.53702700e+00 6.42738879e-01 -5.12381732e-01 -1.21192658e+00
4.32836354e-01 -7.70337582e-02 -5.47845662e-01 -2.09951416e-01
2.61250436e-01 1.12695932e+00 7.24616826e-01 -4.03875262e-01
-1.46941030e+00 -1.08990394e-01 1.20726377e-01 2.33706981e-01
-1.21698640e-01 -2.24899217e-01 -5.79567969e-01 -3.28858681e-02
-9.45302099e-02 -1.01420355e+00 -2.30932370e-01 2.10833639e-01
3.43808271e-02 -2.34137222e-01 1.50084364e+00 -2.17282757e-01
1.20716751e+00 -2.35791492e+00 -2.95417279e-01 -2.31301859e-01
3.72531235e-01 3.63507032e-01 2.33993828e-01 2.06403747e-01
2.93527067e-01 -4.71153140e-01 1.15110569e-01 -3.33010793e-01
2.44336724e-01 5.15656292e-01 -3.79596144e-01 3.51140916e-01
3.13416898e-01 1.15841424e+00 -7.78321922e-01 -5.62748313e-01
7.80601263e-01 4.28065240e-01 -3.46956909e-01 -1.58266842e-01
1.38382107e-01 -5.12429401e-02 -3.20268929e-01 7.50599682e-01
7.21301854e-01 9.33137834e-02 -1.09801359e-01 -6.09738290e-01
-6.30405009e-01 6.85562313e-01 -6.72622740e-01 1.30666089e+00
-6.11269355e-01 1.71158504e+00 1.92990750e-01 -9.47975218e-01
8.62584889e-01 2.02812895e-01 8.07603821e-02 -1.19441581e+00
3.30806166e-01 5.67997992e-01 6.03983477e-02 -3.19111675e-01
1.16941106e+00 2.16784671e-01 -1.71848908e-01 -3.46181750e-01
1.32619500e-01 -1.67529866e-01 5.07068455e-01 2.21465975e-02
8.56970847e-01 -5.18563166e-02 1.37661204e-01 -2.80264348e-01
5.66614687e-01 2.55053163e-01 5.34370542e-01 4.72956657e-01
-6.55377090e-01 5.92384208e-03 4.44702566e-01 -4.98862535e-01
-1.14267254e+00 -9.19264257e-01 -4.07626122e-01 9.26242769e-01
4.49546665e-01 -6.49653077e-01 -6.24007523e-01 -1.97054625e-01
1.11277997e-01 6.06474936e-01 -4.05677795e-01 -3.07042420e-01
-6.38858855e-01 -3.37551445e-01 8.91602755e-01 1.02156484e+00
1.22611666e+00 -4.64426547e-01 -1.45901275e+00 3.75565171e-01
-1.23842619e-01 -1.64157689e+00 5.18744402e-02 7.29783714e-01
-8.44576061e-01 -7.55158842e-01 -2.83597857e-02 -1.03602481e+00
1.78350415e-02 5.68877518e-01 9.49135184e-01 -3.53870094e-01
-1.51862860e-01 -9.50011238e-02 1.75643161e-01 -6.04173005e-01
-1.32364362e-01 1.94805667e-01 1.33281257e-02 -4.26170081e-01
5.98264873e-01 -2.32487649e-01 -5.63365042e-01 5.32896698e-01
-3.63960505e-01 2.08394304e-01 8.69639397e-01 6.90190554e-01
5.35948217e-01 2.28837386e-01 1.20809495e-01 -2.71147937e-01
3.62088501e-01 -4.51813973e-02 -9.91400063e-01 -8.29026997e-02
-6.50547087e-01 -2.96245818e-03 5.77708483e-01 2.03304708e-01
-7.84961820e-01 2.86084354e-01 -5.34091473e-01 -5.17952323e-01
2.38677755e-01 6.32234812e-02 -2.35887960e-01 -4.16089207e-01
4.98521268e-01 2.77056366e-01 1.38409212e-01 -9.34444070e-02
2.47427121e-01 9.19665039e-01 6.88575447e-01 1.56135885e-02
5.75138569e-01 4.76100057e-01 2.01576814e-01 -8.22068751e-01
-3.21065158e-01 -3.31665069e-01 -5.71140647e-01 -5.28413117e-01
7.84009099e-01 -1.33885431e+00 -8.04333687e-01 5.67360759e-01
-1.09891546e+00 -3.13450098e-01 -2.25985795e-01 4.46573853e-01
-8.42617512e-01 -1.29541457e-01 -6.87293053e-01 -4.01740879e-01
-2.42302462e-01 -2.10991168e+00 1.40369046e+00 1.50675118e-01
2.56349206e-01 -7.36103952e-01 -6.27101839e-01 3.57620776e-01
6.77488863e-01 -1.01351701e-01 6.04403377e-01 -4.30674367e-02
-1.02591574e+00 -1.66109785e-01 -3.58850777e-01 2.77231842e-01
-2.28520408e-01 -2.98805572e-02 -1.23598409e+00 -1.21304654e-01
-3.87102813e-01 -7.65190795e-02 7.10630059e-01 3.94075990e-01
9.61234689e-01 1.04610346e-01 -7.39765584e-01 8.33242297e-01
1.13919473e+00 4.38687384e-01 6.12010658e-01 4.60103452e-01
4.14679676e-01 2.55398214e-01 9.19696569e-01 1.01533189e-01
5.63336730e-01 6.98025525e-01 6.81432426e-01 4.90271440e-03
-3.76511276e-01 -2.83575803e-01 6.15967870e-01 8.37123632e-01
1.48247972e-01 -2.14586973e-01 -8.34257007e-01 5.73410153e-01
-1.65321338e+00 -6.91299140e-01 -3.15871507e-01 1.81822217e+00
-3.36950123e-02 6.50698662e-01 -6.35794252e-02 4.11927491e-01
4.92730230e-01 5.93732037e-02 -8.20891380e-01 -9.04398680e-01
-8.13313127e-02 -4.59793070e-03 1.38764691e+00 3.28730226e-01
-1.08967710e+00 1.20018589e+00 7.54421139e+00 1.02587199e+00
-1.37057710e+00 1.94152832e-01 4.07611668e-01 6.12142682e-02
1.10752635e-01 -1.79953352e-01 -1.35397518e+00 2.63864785e-01
1.42871130e+00 -1.69056356e-01 7.48800784e-02 1.21468353e+00
3.17420572e-01 -2.80948542e-02 -1.08780885e+00 9.27564919e-01
-2.03030422e-01 -1.53027964e+00 -2.66604833e-02 2.16881350e-01
3.85141134e-01 6.65139198e-01 4.14289951e-01 4.65724915e-01
8.79205484e-03 -8.41331840e-01 1.28368747e+00 -1.40133247e-01
1.24232304e+00 -9.82757568e-01 9.23534036e-01 2.39803679e-02
-1.95039475e+00 -3.28679740e-01 -4.74318147e-01 -2.89714545e-01
3.70586634e-01 4.57199007e-01 -7.52541125e-01 3.12924027e-01
1.02948272e+00 1.09396946e+00 -5.99234700e-01 7.17119277e-01
-5.39992861e-02 2.53063500e-01 -3.12682092e-01 -4.08398472e-02
6.44438803e-01 -1.00780517e-01 3.25964063e-01 1.35083675e+00
3.77210706e-01 -2.82726407e-01 -2.69057781e-01 4.15901273e-01
3.73570323e-01 -5.64144731e-01 -7.60036707e-01 1.45205513e-01
6.30785048e-01 1.24291468e+00 -5.51936805e-01 -6.63349509e-01
-7.96450257e-01 5.78634501e-01 8.20243731e-02 5.65522835e-02
-1.40369558e+00 -8.30611289e-01 1.31542706e+00 2.40367055e-01
4.94185537e-01 -5.81802189e-01 -5.20541608e-01 -5.08913994e-01
4.67444137e-02 -5.71428418e-01 -1.82955578e-01 -7.62852907e-01
-1.39142722e-01 6.65279508e-01 4.51498516e-02 -1.70551479e+00
-2.66387612e-01 -1.11315966e+00 -1.35504678e-01 4.58651781e-01
-1.94449306e+00 -8.04043591e-01 -5.38009048e-01 3.20756137e-01
8.17739785e-01 -3.73775899e-01 1.71005532e-01 7.21063375e-01
-6.42732978e-01 8.63732040e-01 3.27533752e-01 -8.18838775e-02
2.85148889e-01 -7.70157456e-01 9.48340297e-01 8.67306530e-01
-6.13105536e-01 3.01402301e-01 8.28508258e-01 -1.32339209e-01
-1.90821123e+00 -1.57672131e+00 9.21144664e-01 -7.41562527e-03
6.46559000e-01 -5.24276137e-01 -3.49122673e-01 1.00553226e+00
3.54000419e-01 -1.06179267e-01 1.06536441e-01 -5.40893316e-01
6.59271032e-02 -6.36464179e-01 -9.10144925e-01 6.19956553e-01
1.29648459e+00 -5.10848522e-01 -6.49871351e-03 2.49069721e-01
5.13243377e-01 -1.01545393e+00 -6.37968242e-01 4.42101747e-01
7.29429543e-01 -1.28199899e+00 7.74624467e-01 1.77095324e-01
2.52831757e-01 -5.94463289e-01 -1.46774247e-01 -8.21457803e-01
-3.89662862e-01 -6.13067746e-01 -7.09794760e-02 6.70780301e-01
2.48499349e-01 -9.28957641e-01 8.97408962e-01 -1.63437314e-02
-1.05132341e+00 -7.69021511e-01 -1.21931112e+00 -8.57474148e-01
-1.16293579e-01 -7.96142101e-01 7.31561363e-01 -1.19072109e-01
-5.72667941e-02 4.63158637e-01 -7.18756616e-02 1.93750963e-01
5.13418734e-01 -1.71821073e-01 7.45979548e-01 -7.87665188e-01
2.98147678e-01 -6.47314668e-01 -1.25336897e+00 -1.83720267e+00
-1.14102922e-02 -7.63194919e-01 3.15762222e-01 -1.50833225e+00
-3.65064025e-01 -4.18059021e-01 2.45561451e-01 2.29514465e-01
4.32229191e-01 3.42058003e-01 2.13940185e-03 -4.45897132e-02
-5.51423728e-01 6.01571858e-01 1.02033305e+00 -4.27063584e-01
3.77830476e-01 -1.93439528e-01 -3.46679389e-01 4.79891032e-01
9.87018704e-01 1.03443541e-01 -7.37778485e-01 -8.26479912e-01
8.21559429e-02 -1.34564161e-01 3.45583260e-01 -1.61518395e+00
6.41271889e-01 2.30566159e-01 8.68797898e-02 -1.22956789e+00
6.15628481e-01 -9.70385432e-01 1.46106228e-01 7.25869954e-01
-1.72098080e-04 7.91194797e-01 6.36688411e-01 3.70277852e-01
-5.60021877e-01 1.67259261e-01 8.36080849e-01 2.70802140e-01
-1.53640902e+00 2.65460938e-01 -1.27393389e+00 -2.98889279e-01
1.24483812e+00 -8.77554715e-01 -3.46857101e-01 -9.80085433e-02
-2.77063638e-01 5.61725318e-01 1.21953867e-01 5.83708644e-01
6.29376352e-01 -1.34594941e+00 -3.07162523e-01 5.92721939e-01
4.15372193e-01 1.51929911e-02 1.14545740e-01 8.91851604e-01
-8.35231960e-01 1.29642451e+00 -5.50049245e-01 -1.14167333e+00
-1.26442981e+00 3.72376770e-01 6.80309057e-01 7.10931942e-02
-5.62761247e-01 6.50769651e-01 2.88800895e-01 -1.90273404e-01
2.69707620e-01 -8.09361577e-01 -5.55161871e-02 -2.73092896e-01
1.71861663e-01 6.03457212e-01 7.62081742e-01 -8.59288156e-01
-4.88529295e-01 6.91326380e-01 7.72044137e-02 7.78592974e-02
4.65503216e-01 -3.29552293e-01 2.53040999e-01 2.02810198e-01
1.20044124e+00 -7.02548027e-01 -1.29761028e+00 5.82475513e-02
-2.56735265e-01 -3.43521498e-02 4.24240947e-01 -1.91902205e-01
-1.38309503e+00 1.36205018e+00 9.47511792e-01 -1.50433019e-01
1.19850683e+00 -2.82591671e-01 1.10165513e+00 4.66864139e-01
6.15729392e-01 -1.29204822e+00 -5.32727778e-01 1.01549745e+00
4.24858063e-01 -9.02932167e-01 -2.45326921e-01 -5.53274155e-01
-3.98224026e-01 1.12888098e+00 7.55290031e-01 9.14979428e-02
6.95093334e-01 8.20154607e-01 5.75407408e-02 -2.74894256e-02
-1.06040657e+00 -3.16738069e-01 -9.65577662e-02 7.91254640e-01
4.13945377e-01 2.29152977e-01 -1.37344062e-01 -1.32778913e-01
-7.94829428e-01 -4.30334955e-02 3.87514055e-01 1.28884149e+00
-6.25809789e-01 -7.41561413e-01 -3.82673591e-02 5.45534253e-01
9.17709097e-02 -1.27696143e-02 1.10098727e-01 8.36218059e-01
4.90118414e-01 9.86891508e-01 4.53631341e-01 -8.93912911e-01
4.77576464e-01 -1.64667487e-01 2.90078491e-01 5.98298199e-02
-4.67505425e-01 -2.46286526e-01 3.67515951e-01 -8.08655322e-01
-9.78786424e-02 -7.03596652e-01 -1.19350278e+00 -6.50717378e-01
-2.56814718e-01 3.57629098e-02 9.61433589e-01 7.46753514e-01
7.26427913e-01 9.65942621e-01 2.75746495e-01 -8.95777881e-01
-2.39731267e-01 -5.73261321e-01 -2.82221735e-01 -7.10330725e-01
6.45595908e-01 -9.68427062e-01 -7.57184252e-02 3.86719219e-02] | [8.10883903503418, -1.2386804819107056] |
de9bb5fa-1a68-4064-8bb5-04a6f2f985b5 | dare-slam-degeneracy-aware-and-resilient-loop | 2102.05117 | null | https://arxiv.org/abs/2102.05117v1 | https://arxiv.org/pdf/2102.05117v1.pdf | DARE-SLAM: Degeneracy-Aware and Resilient Loop Closing in Perceptually-Degraded Environments | Enabling fully autonomous robots capable of navigating and exploring large-scale, unknown and complex environments has been at the core of robotics research for several decades. A key requirement in autonomous exploration is building accurate and consistent maps of the unknown environment that can be used for reliable navigation. Loop closure detection, the ability to assert that a robot has returned to a previously visited location, is crucial for consistent mapping as it reduces the drift caused by error accumulation in the estimated robot trajectory. Moreover, in multi-robot systems, loop closures enable merging local maps obtained by a team of robots into a consistent global map of the environment. In this paper, we present a degeneracy-aware and drift-resilient loop closing method to improve place recognition and resolve 3D location ambiguities for simultaneous localization and mapping (SLAM) in GPS-denied, large-scale and perceptually-degraded environments. More specifically, we focus on SLAM in subterranean environments (e.g., lava tubes, caves, and mines) that represent examples of complex and ambiguous environments where current methods have inadequate performance. | ['Ali-akbar Agha-mohammadi', 'Curtis Padgett', 'Sally Wood', 'Matteo Palieri', 'Kamak Ebadi'] | 2021-02-09 | null | null | null | null | ['loop-closure-detection'] | ['computer-vision'] | [ 1.49297580e-01 -3.33829343e-01 2.91053474e-01 -4.43050057e-01
-5.93361259e-01 -9.49719667e-01 3.45034748e-01 4.84246999e-01
-5.85728168e-01 1.05649936e+00 -2.45395303e-01 -2.64128834e-01
-5.03920436e-01 -5.73425114e-01 -8.08559000e-01 -4.44161773e-01
-5.49586058e-01 8.27203214e-01 5.42881310e-01 -5.03837645e-01
5.08681059e-01 8.47991109e-01 -1.29937732e+00 -6.76399171e-01
8.39011788e-01 5.62434196e-01 8.52275848e-01 4.18102652e-01
2.57496059e-01 1.89958885e-01 -2.16089204e-01 4.69348490e-01
2.91580886e-01 7.03638420e-02 -7.14578986e-01 -1.13845535e-01
2.68324882e-01 -1.23973452e-01 -4.82596941e-02 1.03138280e+00
1.94503590e-01 3.39060813e-01 2.32180685e-01 -1.35986221e+00
1.82361037e-01 3.83365192e-02 -3.23867619e-01 9.72611457e-02
4.08884078e-01 -6.64481223e-02 3.77834409e-01 -6.82552636e-01
5.39913177e-01 1.09382570e+00 8.90940189e-01 -2.33213231e-01
-1.15018094e+00 -4.28244710e-01 1.35180324e-01 1.00579113e-01
-1.73250437e+00 -6.38442516e-01 2.59508282e-01 -3.49181920e-01
8.24629307e-01 -3.70151363e-02 4.96121973e-01 6.16429448e-01
7.08875299e-01 -1.37341246e-01 6.90370619e-01 -1.82719350e-01
4.80951399e-01 -1.03413232e-01 -5.63256085e-01 6.59370780e-01
8.78522277e-01 -1.74847737e-01 -8.47655594e-01 -1.43305525e-01
7.33276486e-01 1.16407886e-01 -2.81699151e-01 -9.86841500e-01
-1.34876120e+00 6.08314455e-01 7.32236624e-01 7.13005811e-02
-6.33315265e-01 1.58919334e-01 -4.34615053e-02 1.66537359e-01
-2.60685921e-01 7.52639711e-01 -1.81954548e-01 -4.83118325e-01
-4.71385986e-01 3.55486542e-01 7.59140313e-01 1.24969757e+00
1.10748386e+00 -7.65117537e-03 8.78120363e-01 4.16668981e-01
2.65995979e-01 8.76437962e-01 3.25740218e-01 -1.16227245e+00
4.35678363e-01 5.88251531e-01 7.96595871e-01 -1.21399438e+00
-7.87291825e-01 -4.00034159e-01 -4.56144959e-01 4.74584222e-01
1.35993972e-01 5.13276011e-02 -6.43639982e-01 1.60745895e+00
4.98239726e-01 -2.58084565e-01 4.47071970e-01 8.89033794e-01
-1.79683760e-01 1.37571365e-01 -3.85022402e-01 1.18846297e-01
1.07402587e+00 -6.15693212e-01 -5.24704278e-01 -1.18381608e+00
4.73990709e-01 -5.62068284e-01 4.13276762e-01 1.23339094e-01
-2.71507233e-01 -1.43506944e-01 -1.35300994e+00 1.31106228e-01
-3.68415684e-01 -3.10791969e-01 4.96165067e-01 1.04920805e-01
-1.22166753e+00 3.72859508e-01 -1.17610669e+00 -7.31030166e-01
-3.49612758e-02 3.11881512e-01 -8.30886781e-01 -2.41631150e-01
-9.27929342e-01 1.43762398e+00 4.58366632e-01 4.71410036e-01
-7.88004398e-01 -4.19336697e-03 -1.18198872e+00 -4.40015405e-01
3.76639873e-01 -1.97838575e-01 1.05872095e+00 -2.05085665e-01
-7.81148434e-01 4.54086274e-01 -5.27849913e-01 -5.85460424e-01
4.37959671e-01 -4.14041013e-01 -3.06511134e-01 -2.02367514e-01
8.13116729e-01 5.65940082e-01 3.73205096e-01 -1.35944843e+00
-9.56668317e-01 -7.07300365e-01 -3.09002548e-01 7.51859426e-01
6.51144207e-01 -6.99795008e-01 -1.39613166e-01 2.00421646e-01
1.34800839e+00 -1.31044436e+00 -5.97293913e-01 -5.48410676e-02
3.83289270e-02 3.46027285e-01 7.48029232e-01 -5.39442360e-01
5.53409398e-01 -2.18852401e+00 -5.24563249e-03 3.13828945e-01
-2.10938066e-01 -4.45222855e-01 2.12936193e-01 8.37524712e-01
6.63282812e-01 -2.11272568e-01 -3.38463664e-01 -2.99418718e-01
-2.16227219e-01 7.24352062e-01 -2.99270809e-01 8.40030432e-01
-2.01779380e-01 4.43470061e-01 -1.10800946e+00 -1.00124553e-02
2.65745163e-01 1.99844062e-01 -5.35528548e-02 -3.47599298e-01
2.38356486e-01 8.58914852e-01 -3.96992922e-01 7.93160617e-01
7.31610119e-01 1.96061164e-01 2.81120330e-01 3.97124231e-01
-8.42047691e-01 3.90874624e-01 -1.57079268e+00 2.14059138e+00
-3.86683851e-01 8.63246202e-01 4.81148183e-01 -4.15698349e-01
1.32607722e+00 -1.71187073e-01 1.05717674e-01 -9.87469792e-01
-2.08719268e-01 9.12465215e-01 -1.86046511e-01 -2.59616792e-01
1.40196967e+00 6.12733960e-02 -2.60911494e-01 6.45560026e-02
-5.26243031e-01 -4.82541829e-01 -5.05522788e-02 -3.34731936e-02
1.27731133e+00 6.27920181e-02 4.76735830e-01 -2.70214230e-01
1.70048699e-01 6.12017572e-01 7.95101702e-01 8.89162064e-01
-4.01867211e-01 3.32528889e-01 -5.82773723e-02 -3.73765945e-01
-9.96154249e-01 -1.15172982e+00 -2.62229770e-01 5.65940976e-01
1.10690618e+00 7.34896511e-02 2.27797359e-01 8.49136040e-02
4.14400220e-01 3.31632078e-01 -2.94595540e-01 -5.49459681e-02
-6.53701186e-01 -3.54229242e-01 2.42075607e-01 2.53764838e-01
6.27317071e-01 -5.77893198e-01 -1.27592552e+00 6.22463286e-01
-4.83783036e-01 -1.07498825e+00 3.14283550e-01 5.85600793e-01
-8.60721290e-01 -1.03692174e+00 -4.52412665e-02 -9.37320292e-01
8.42266083e-01 8.57370377e-01 4.34973836e-01 -3.81298363e-01
-4.29011211e-02 2.28330210e-01 -4.17697698e-01 -2.38132253e-01
-1.98644608e-01 -1.82402954e-01 6.34406626e-01 -4.40930724e-01
-4.87750806e-02 -6.33436143e-01 -2.76486576e-01 7.59147704e-01
-1.50978327e-01 -1.33283630e-01 6.22876525e-01 4.90376443e-01
8.23393404e-01 2.32576698e-01 5.37479579e-01 -1.52452603e-01
2.74975508e-01 -6.31935239e-01 -9.08952236e-01 -8.46980959e-02
-5.18294334e-01 4.77036126e-02 1.12271711e-01 -1.11456409e-01
-6.00364029e-01 3.52808028e-01 2.91469127e-01 1.94320142e-01
-1.13870583e-01 5.99002659e-01 1.84442803e-01 -4.87397462e-01
7.90763736e-01 3.68628323e-01 1.01711802e-01 -3.90119910e-01
1.44336671e-01 7.07680821e-01 1.16751814e+00 -2.36119494e-01
6.40187562e-01 9.35312212e-01 3.20825338e-01 -6.96811378e-01
-9.39619616e-02 -7.35480130e-01 -8.54345322e-01 -6.08103164e-02
2.63493299e-01 -1.17051780e+00 -3.63143086e-01 3.54305416e-01
-9.05309916e-01 -3.76581550e-01 8.97847861e-02 6.76750362e-01
-5.35125911e-01 1.65994987e-01 2.81062454e-01 -1.08779466e+00
3.43866274e-02 -1.10636008e+00 9.02046621e-01 5.50436616e-01
-4.11251754e-01 -7.24618435e-01 4.19916391e-01 -1.49453953e-01
4.96967375e-01 5.81976414e-01 1.61294550e-01 -2.55092144e-01
-6.74762487e-01 -3.95015985e-01 3.99501473e-02 -5.43883979e-01
4.66459244e-01 -6.55441642e-01 -5.96818626e-01 -7.29623914e-01
4.14419472e-02 6.99782791e-03 3.95865381e-01 7.01794997e-02
-2.37754956e-01 -3.58965062e-02 -8.09338629e-01 4.32447910e-01
1.52635443e+00 5.54310143e-01 2.35247418e-01 1.08850944e+00
2.31793046e-01 5.48807621e-01 1.11441648e+00 4.26024169e-01
8.84318709e-01 5.44779122e-01 9.42313492e-01 3.77998143e-01
3.92104954e-01 -3.88234049e-01 4.00759876e-02 8.07055160e-02
4.01490897e-01 -5.65319248e-02 -1.07195759e+00 1.10144341e+00
-2.21368623e+00 -3.45826834e-01 -2.15233192e-01 2.43457747e+00
3.03542633e-02 1.22667797e-01 -6.28006160e-01 -1.36535376e-01
8.08460712e-01 -9.04835686e-02 -9.37758505e-01 -3.01177870e-03
-1.16658628e-01 -6.20889306e-01 1.18917537e+00 7.79711604e-01
-1.01054370e+00 9.18990314e-01 5.30367708e+00 -7.37768710e-02
-1.03181207e+00 -8.44632834e-02 -5.02410531e-01 2.01888099e-01
1.72674172e-02 3.86079848e-01 -6.43202841e-01 1.36590511e-01
7.40756094e-01 9.93379876e-02 3.96476984e-01 1.00963104e+00
2.09016114e-01 -1.19264627e+00 -8.53574753e-01 9.88792181e-01
-2.08407789e-01 -1.07310593e+00 -7.51347899e-01 1.43041700e-01
7.48124957e-01 6.10735595e-01 -7.63782486e-02 -1.02974087e-01
3.92511159e-01 -6.23701036e-01 1.21572268e+00 3.53467077e-01
5.48807085e-01 -6.15055799e-01 9.61402714e-01 5.98977447e-01
-1.30345321e+00 -2.18690768e-01 -4.04201210e-01 -4.43738639e-01
6.75827205e-01 3.42606336e-01 -1.44692445e+00 6.76880538e-01
9.26288307e-01 5.15931129e-01 -4.95355338e-01 1.45629883e+00
-3.78004760e-01 -3.94651800e-01 -8.30178380e-01 -6.85052499e-02
5.03198564e-01 -1.92987412e-01 7.95051515e-01 4.10218716e-01
6.46245122e-01 -1.27601087e-01 3.39781702e-01 6.18372440e-01
5.24678409e-01 -3.81421924e-01 -8.85891378e-01 2.70253897e-01
1.10892701e+00 1.00514507e+00 -7.98752606e-01 5.79325967e-02
1.55863851e-01 8.32678199e-01 2.33534411e-01 1.72403157e-01
-3.37660581e-01 -6.21008694e-01 7.84954071e-01 -1.94212019e-01
2.34765802e-02 -1.10391688e+00 -3.43268752e-01 -5.86969674e-01
3.42142671e-01 -1.75866604e-01 -1.30283087e-01 -8.23041320e-01
-2.77812719e-01 3.86478961e-01 -3.91391069e-01 -1.48234832e+00
-5.45004487e-01 5.16125700e-03 -1.28205433e-01 9.98365700e-01
-1.67591667e+00 -6.80535078e-01 -7.11719155e-01 1.62948012e-01
3.35596532e-01 1.47883803e-01 8.04644883e-01 -8.58557895e-02
1.76152959e-01 -3.40981513e-01 5.28391302e-01 -3.86226803e-01
8.22078884e-01 -1.09622169e+00 6.80475414e-01 1.12826169e+00
-1.91398457e-01 7.25445986e-01 1.14761174e+00 -1.26208603e+00
-1.68772054e+00 -9.64569628e-01 8.42178822e-01 -3.74740452e-01
5.31195462e-01 -3.23568016e-01 -9.33334768e-01 8.12126994e-01
-7.37242818e-01 -2.31596768e-01 -5.04151992e-02 -2.09530666e-02
6.47032540e-03 -3.86887081e-02 -1.23815584e+00 4.76861924e-01
9.40142393e-01 -3.43419582e-01 -3.61087650e-01 6.68349639e-02
4.84318793e-01 -8.84022474e-01 -4.76648003e-01 4.22854811e-01
8.11293423e-01 -9.38373744e-01 6.22753918e-01 2.65523672e-01
-7.02365577e-01 -7.83803940e-01 -5.39586186e-01 -1.31058621e+00
-2.21270949e-01 -5.71251631e-01 4.81636733e-01 8.23791385e-01
1.24607436e-01 -1.03882170e+00 6.18765891e-01 4.85836118e-01
-3.56396049e-01 -2.86431406e-02 -1.40805960e+00 -1.09196055e+00
-5.44789910e-01 -2.08372355e-01 4.78380024e-01 4.93786752e-01
-3.13432366e-02 1.85999218e-02 -2.02230275e-01 1.03344691e+00
6.70410335e-01 4.75781932e-02 1.03953052e+00 -1.31636059e+00
5.77811360e-01 -5.83567098e-03 -8.53522956e-01 -9.38132584e-01
-1.40435830e-01 -3.79399687e-01 9.20586705e-01 -1.88195074e+00
-4.74011987e-01 -1.15450335e+00 3.21179628e-01 4.19047028e-01
3.90437007e-01 2.01439741e-03 -1.26470119e-01 7.04839349e-01
-7.54671454e-01 4.60008830e-01 5.65413058e-01 1.37770891e-01
-5.14519811e-01 -5.86476875e-03 -3.54516119e-01 5.95174432e-01
8.29008639e-01 -6.13017082e-01 -1.97898880e-01 -7.74992704e-01
5.79387605e-01 1.31828219e-01 2.73985714e-01 -1.49073291e+00
8.07578146e-01 -2.65973538e-01 1.59856528e-01 -7.86983788e-01
6.31399751e-01 -8.64781201e-01 5.51005483e-01 7.51318932e-01
2.24269822e-01 2.39589781e-01 2.19988823e-01 8.76728535e-01
-2.99311697e-01 -3.66518617e-01 5.52411854e-01 -2.11165234e-01
-1.61180341e+00 -4.28472683e-02 -4.15400535e-01 -2.75503665e-01
1.03482640e+00 -3.77498686e-01 -4.29787934e-01 -3.31850529e-01
-4.24086869e-01 8.28651786e-01 1.13973629e+00 5.48899055e-01
7.28780270e-01 -9.29079115e-01 -4.01852965e-01 4.89250749e-01
4.14901078e-01 5.33999562e-01 2.38579884e-01 8.77520263e-01
-9.64692771e-01 5.64602137e-01 -4.98791635e-01 -9.58568394e-01
-9.61999714e-01 -3.26838382e-02 1.88224822e-01 4.51650649e-01
-6.14777148e-01 6.61954880e-01 -3.27083200e-01 -6.90021694e-01
7.05336183e-02 -2.84867316e-01 1.77795127e-01 -3.01912874e-01
5.30136466e-01 4.79238182e-01 2.49575168e-01 -8.95390630e-01
-9.14175689e-01 5.71559191e-01 3.05559993e-01 -4.27909583e-01
1.20306408e+00 -1.03813124e+00 -1.97981954e-01 6.25188529e-01
6.58918262e-01 5.17210998e-02 -1.44974446e+00 -1.10899322e-01
2.46365085e-01 -5.85994124e-01 -8.38218108e-02 -4.71303225e-01
3.19209248e-02 4.07044619e-01 7.53893375e-01 -1.38982749e-02
4.76278216e-01 7.23698586e-02 3.62617552e-01 8.88723135e-01
1.35212362e+00 -9.76850629e-01 -3.75374496e-01 9.00208056e-01
7.99645066e-01 -1.32841802e+00 1.76225398e-02 1.47812888e-01
-5.65230429e-01 1.09350169e+00 4.06911135e-01 1.70493856e-01
-7.77267739e-02 -3.88894603e-02 9.17987153e-02 -2.41928995e-02
-1.79684967e-01 -1.44868299e-01 -4.84059066e-01 9.28416312e-01
-5.81706703e-01 1.86201304e-01 3.00523609e-01 -1.25545546e-01
-5.13038218e-01 -5.25237620e-01 8.46705198e-01 1.61346567e+00
-1.30212677e+00 -4.57950175e-01 -8.38980317e-01 -8.37339833e-02
4.97415572e-01 3.78075451e-01 8.60752724e-03 7.94032753e-01
-1.65377744e-02 1.08396804e+00 4.05438066e-01 -3.12568903e-01
3.44374806e-01 -1.15942448e-01 3.12505633e-01 -4.56253439e-01
2.84198761e-01 -1.50319323e-01 2.13342741e-01 -6.53457820e-01
-1.42325595e-01 -9.45783854e-01 -1.70276237e+00 -7.68336281e-02
-2.66468763e-01 3.80505890e-01 1.45912242e+00 9.09825325e-01
6.15277290e-01 1.91749185e-01 4.12438750e-01 -1.05978119e+00
-2.64211059e-01 -8.12923431e-01 -7.97139883e-01 -3.60266805e-01
7.89511800e-01 -1.10436428e+00 -2.50924408e-01 -4.85568643e-01] | [7.231740951538086, -2.028764009475708] |
9aadc27f-1ec3-4349-be7e-5f9d25fcbbd7 | distilling-knowledge-from-language-models-for | 2210.05991 | null | https://arxiv.org/abs/2210.05991v2 | https://arxiv.org/pdf/2210.05991v2.pdf | Text-Derived Knowledge Helps Vision: A Simple Cross-modal Distillation for Video-based Action Anticipation | Anticipating future actions in a video is useful for many autonomous and assistive technologies. Most prior action anticipation work treat this as a vision modality problem, where the models learn the task information primarily from the video features in the action anticipation datasets. However, knowledge about action sequences can also be obtained from external textual data. In this work, we show how knowledge in pretrained language models can be adapted and distilled into vision-based action anticipation models. We show that a simple distillation technique can achieve effective knowledge transfer and provide consistent gains on a strong vision model (Anticipative Vision Transformer) for two action anticipation datasets (3.5% relative gain on EGTEA-GAZE+ and 7.2% relative gain on EPIC-KITCHEN 55), giving a new state-of-the-art result. | ['Niranjan Balasubramanian', 'Minh Hoai', 'Tanvi Aggarwal', 'Sayontan Ghosh'] | 2022-10-12 | null | null | null | null | ['action-anticipation'] | ['computer-vision'] | [ 4.98831600e-01 4.81823027e-01 -5.46113789e-01 -4.12451148e-01
-4.31922257e-01 -3.66563231e-01 1.00408089e+00 -4.84489024e-01
-5.26498377e-01 3.56929183e-01 6.66557014e-01 -3.45529243e-02
3.66234392e-01 -1.51856616e-01 -9.36725855e-01 -3.61600518e-01
-7.85245374e-02 2.28029281e-01 2.44147152e-01 -1.96363330e-01
8.06635395e-02 3.21568251e-02 -1.53022397e+00 8.84918571e-01
3.09923947e-01 9.77158129e-01 4.41184729e-01 1.01962125e+00
1.72572315e-01 1.67921913e+00 1.89867020e-01 -2.42733836e-01
3.27279896e-01 -1.07130788e-01 -1.01066875e+00 3.29935431e-01
6.17031813e-01 -8.66821647e-01 -9.18542087e-01 5.88514209e-01
1.03595898e-01 2.92311516e-02 4.87464994e-01 -1.51927686e+00
-7.70621955e-01 7.47838378e-01 -5.85365236e-01 1.33773625e-01
5.72545409e-01 9.81941104e-01 1.00367391e+00 -6.87920332e-01
7.02937365e-01 1.37612975e+00 4.86238241e-01 9.06633794e-01
-7.72317350e-01 -3.25347245e-01 5.34906685e-01 8.91261935e-01
-7.15340018e-01 -8.26136112e-01 5.00417113e-01 -5.19321859e-01
1.64490807e+00 -7.08708167e-02 9.16598558e-01 1.45394361e+00
1.97024107e-01 1.59383774e+00 7.09842205e-01 -3.91717046e-01
-1.14042006e-01 -3.25014353e-01 1.14526637e-01 7.39999354e-01
-1.96033642e-01 3.48604262e-01 -1.12628710e+00 5.60812056e-01
4.75036770e-01 7.73671046e-02 -1.60281748e-01 -3.29560488e-01
-1.28898239e+00 5.08766174e-01 5.39295316e-01 -1.64917767e-01
-9.08059180e-01 9.26488280e-01 5.26312292e-01 9.91733521e-02
2.63461024e-01 1.48670956e-01 -5.63234270e-01 -7.88858831e-01
-4.80301738e-01 8.82766768e-02 5.77134490e-01 1.09474909e+00
5.52142620e-01 1.50376245e-01 -3.82482767e-01 6.08674809e-02
6.04396343e-01 7.42568254e-01 3.95631552e-01 -1.52268136e+00
6.37172103e-01 3.65253359e-01 2.76831657e-01 -5.23133695e-01
-1.86345071e-01 3.62398088e-01 -2.78444827e-01 1.45807981e-01
2.97351092e-01 -3.78611118e-01 -1.32129800e+00 1.72261965e+00
-2.22507026e-02 4.71975863e-01 4.58203703e-01 6.75224900e-01
7.90729940e-01 4.50888902e-01 5.68623483e-01 -2.36540392e-01
1.21260262e+00 -1.41940105e+00 -7.23689854e-01 -1.13448036e+00
8.70857120e-01 -4.34941858e-01 1.04479134e+00 3.17350239e-01
-1.12090600e+00 -6.23856425e-01 -7.26595938e-01 -2.24689350e-01
5.84922358e-02 2.10547939e-01 1.06917858e+00 3.27813327e-02
-1.30759776e+00 2.92493343e-01 -1.24212146e+00 -5.94780803e-01
7.60750711e-01 3.20202440e-01 -5.68814814e-01 -3.91809613e-01
-1.00913465e+00 9.96245027e-01 4.18154061e-01 7.65197054e-02
-1.21550918e+00 -7.97317684e-01 -1.01470959e+00 -3.29684764e-02
6.64951026e-01 -8.58896554e-01 1.80567360e+00 -1.34331667e+00
-1.52681124e+00 8.20661962e-01 -1.91588178e-01 -1.08789194e+00
5.28848529e-01 -9.99165177e-01 -2.49229729e-01 3.21690023e-01
-1.79857254e-01 1.33417213e+00 9.70111370e-01 -7.64662027e-01
-7.38100648e-01 -1.65901825e-01 5.53528547e-01 4.09156352e-01
-3.34798127e-01 -8.36976711e-03 -6.26980722e-01 -1.18060239e-01
-4.03091520e-01 -1.19789803e+00 -2.09904358e-01 3.07143003e-01
-1.07167810e-01 -1.97965294e-01 1.01441050e+00 -6.91103160e-01
8.19924891e-01 -1.99038351e+00 1.59305602e-01 -5.33638000e-01
1.64694160e-01 5.00888348e-01 -5.76883554e-01 4.12510157e-01
-1.18965760e-01 -1.69449911e-01 2.17865169e-01 -3.45710397e-01
2.14703009e-02 3.83278161e-01 -5.45123756e-01 1.53275087e-01
4.01805192e-01 1.48121035e+00 -1.06663752e+00 -3.31797272e-01
5.11797309e-01 6.22136652e-01 -6.43289208e-01 2.51589447e-01
-5.54566383e-01 2.65526325e-01 -4.22010392e-01 5.58709919e-01
6.31626472e-02 -3.27976376e-01 1.92486331e-01 -3.17632705e-01
7.79981762e-02 6.29814342e-02 -4.09949332e-01 2.05895019e+00
-3.58042270e-01 1.16928709e+00 -3.97771955e-01 -8.11369181e-01
3.58552486e-01 3.30900162e-01 6.30327582e-01 -6.97769821e-01
-1.16001360e-01 -5.39138138e-01 5.03533073e-02 -1.04747057e+00
5.61311781e-01 1.52654290e-01 1.40689686e-01 2.57540166e-01
2.00194083e-02 -2.35400908e-02 8.07270631e-02 3.20295036e-01
1.32158291e+00 6.20810270e-01 2.84891069e-01 3.42852443e-01
1.74646825e-01 3.64456952e-01 4.36194271e-01 7.08403170e-01
-4.76211190e-01 2.40636706e-01 2.35996425e-01 -6.48277998e-01
-6.54250085e-01 -8.91364217e-01 7.14386106e-01 1.21128452e+00
-1.32690921e-01 -6.79974735e-01 -5.28309286e-01 -7.24404335e-01
-2.09637195e-01 1.03227973e+00 -6.15993381e-01 -5.13679326e-01
-6.12019718e-01 -1.18027650e-01 1.87983111e-01 1.24337530e+00
8.73104572e-01 -1.33676147e+00 -1.13597393e+00 -1.81686878e-01
-4.02412921e-01 -1.56016004e+00 -4.19566065e-01 -2.05214426e-01
-7.69848585e-01 -1.00915158e+00 -2.84877926e-01 -3.73501897e-01
3.29087764e-01 4.79317486e-01 1.24387097e+00 -7.22886473e-02
-1.36361867e-01 1.20970058e+00 -4.60139662e-01 -6.92107558e-01
-3.39795679e-01 -3.91947687e-01 2.07893997e-01 -7.60154426e-02
8.02134871e-01 -3.15746754e-01 -6.09262466e-01 -2.41964966e-01
-5.80007553e-01 5.88668108e-01 7.14892030e-01 6.64186001e-01
3.81910235e-01 -4.80340153e-01 9.46494043e-02 -4.90581930e-01
9.92138609e-02 -2.46729314e-01 -3.02840531e-01 2.56587565e-01
-4.45900053e-01 9.24272165e-02 5.67885600e-02 -5.76494455e-01
-1.35347080e+00 7.61487484e-01 2.68121153e-01 -8.55793059e-01
-2.51576573e-01 5.99506259e-01 -6.57038987e-02 1.58779979e-01
5.21384060e-01 2.10030705e-01 8.17466974e-02 -1.07934870e-01
8.70237410e-01 4.76118445e-01 8.10618222e-01 -1.48636982e-01
4.36625838e-01 6.42044306e-01 -1.02030441e-01 -6.66848123e-01
-1.06147254e+00 -4.31161493e-01 -7.45856524e-01 -4.71685767e-01
1.06793618e+00 -1.39801192e+00 -9.64476049e-01 6.02222621e-01
-1.19448745e+00 -1.05180800e+00 -4.97040570e-01 5.89923978e-01
-1.20361769e+00 3.92705172e-01 -4.23619926e-01 -7.65933275e-01
-3.44267279e-01 -9.79183614e-01 1.08206332e+00 2.09251031e-01
-4.64734644e-01 -8.83619010e-01 -1.01411246e-01 5.74272871e-01
1.06331080e-01 -4.89896797e-02 3.99301738e-01 -2.88865507e-01
-9.29700375e-01 -1.95813194e-01 -2.72649884e-01 3.43504518e-01
1.00572310e-01 4.65141088e-02 -1.13261402e+00 -9.55383927e-02
-2.94858456e-01 -7.41643965e-01 1.27611041e+00 5.32261848e-01
9.38931406e-01 -3.31941873e-01 -5.30010581e-01 3.46913844e-01
1.06181681e+00 9.60348248e-02 9.04494703e-01 2.37048969e-01
1.00305486e+00 4.18372601e-01 1.10168779e+00 5.55992901e-01
6.45844162e-01 6.21574700e-01 7.56698549e-01 3.12299967e-01
-4.63517010e-01 -3.31007659e-01 1.09833610e+00 1.78114116e-01
-5.05323231e-01 -3.50902617e-01 -1.22223854e+00 7.75253832e-01
-2.37766862e+00 -1.31721902e+00 8.86525586e-03 1.68223810e+00
7.18339860e-01 1.34488836e-01 -2.44717719e-03 -2.93854505e-01
2.57368803e-01 3.45930129e-01 -8.70613992e-01 -2.77788818e-01
1.92097113e-01 -2.41838381e-01 4.77451682e-01 4.17213678e-01
-1.22749269e+00 1.40009725e+00 7.12112331e+00 3.65716785e-01
-9.78435993e-01 -1.02970064e-01 2.25482136e-01 -3.47552210e-01
-2.45228652e-02 1.50868937e-01 -7.22969234e-01 1.28368989e-01
1.17707920e+00 -2.09892362e-01 2.26534322e-01 8.71882796e-01
3.71724606e-01 -3.20584863e-01 -1.58720767e+00 1.22097743e+00
1.14405438e-01 -1.48313320e+00 7.32583553e-02 -9.45631713e-02
4.93867308e-01 4.72630709e-01 2.13953614e-01 5.54699421e-01
5.95353425e-01 -9.64841366e-01 6.13522053e-01 9.30911720e-01
7.47970939e-01 -3.12076449e-01 2.89424151e-01 3.40151459e-01
-1.01697278e+00 -4.36289489e-01 -3.80796827e-02 -6.03946209e-01
4.17139351e-01 4.10448797e-02 -9.70139444e-01 2.42983580e-01
7.02143908e-01 1.53762591e+00 -4.71083045e-01 6.68670118e-01
-5.40633619e-01 6.65233135e-01 -8.10520425e-02 2.25400493e-01
3.57913733e-01 3.19224477e-01 4.96211141e-01 9.20878649e-01
5.12292683e-02 2.36854270e-01 1.77098170e-01 2.97846228e-01
-7.56666902e-03 -6.36991560e-01 -9.70762253e-01 -5.10781348e-01
-2.27962602e-02 9.91090357e-01 -1.83022730e-02 -5.71487546e-01
-7.62518644e-01 1.14444232e+00 3.19317728e-01 4.59265739e-01
-9.86761808e-01 3.62281203e-01 1.01584482e+00 -7.97346085e-02
7.21347034e-01 -3.75984967e-01 1.20382473e-01 -1.15239823e+00
-9.10522193e-02 -7.62012422e-01 3.28719258e-01 -1.40127659e+00
-8.13077271e-01 2.47433841e-01 2.10255817e-01 -1.29874480e+00
-1.05260265e+00 -8.23757648e-01 -4.83887464e-01 2.72667944e-01
-1.48052371e+00 -1.60940778e+00 -5.63996911e-01 6.17979944e-01
9.84817624e-01 -2.10511863e-01 7.86198974e-01 -2.32933506e-01
-2.89016038e-01 2.79043049e-01 -4.44538176e-01 1.05027013e-01
7.05169559e-01 -1.01979172e+00 7.09202468e-01 1.04995465e+00
3.62769037e-01 7.84266740e-02 8.85547876e-01 -6.53693020e-01
-1.90342176e+00 -1.22336555e+00 8.37487161e-01 -9.92012501e-01
8.25947046e-01 1.04041629e-01 -4.69239295e-01 1.46649420e+00
6.60929382e-01 8.93117636e-02 3.11917901e-01 1.26763741e-02
-5.13898671e-01 1.11136518e-01 -6.72656298e-01 9.16077971e-01
1.60305893e+00 -6.87958598e-01 -7.14473426e-01 5.24608076e-01
9.49875355e-01 -5.03031373e-01 -5.75616479e-01 3.73259276e-01
7.64424622e-01 -7.36780107e-01 1.15940654e+00 -1.14865112e+00
9.06759977e-01 -6.79451823e-02 -2.36262754e-01 -1.07314956e+00
-2.32469276e-01 -7.97899365e-01 -8.22621644e-01 8.87515604e-01
2.92535685e-02 -1.53576106e-01 7.00737357e-01 1.09471428e+00
-1.54318094e-01 -4.40529436e-01 -4.35246676e-01 -7.41257191e-01
-3.96204561e-01 -7.95255005e-01 1.38715714e-01 4.96430039e-01
1.56233296e-01 4.19353634e-01 -6.55967534e-01 -6.52275234e-02
3.46888900e-01 -5.52398153e-02 9.30340886e-01 -8.19356918e-01
-3.32066685e-01 -4.57036421e-02 -5.97314060e-01 -1.36468542e+00
5.31314254e-01 -5.27704656e-01 2.98587292e-01 -1.77132201e+00
2.58032143e-01 3.46732050e-01 -2.65435964e-01 1.13499272e+00
-3.65104638e-02 4.76197116e-02 6.30506694e-01 5.41566499e-03
-9.71575439e-01 6.50409579e-01 1.24031615e+00 -3.77638429e-01
-1.14549272e-01 -1.61261663e-01 -7.08915174e-01 1.10247326e+00
6.17401898e-01 -5.68160079e-02 -8.86623442e-01 -8.69568169e-01
1.39563426e-01 5.71255572e-02 7.37799883e-01 -1.06727040e+00
4.06224757e-01 -5.03486693e-01 1.50702506e-01 -5.37442207e-01
8.19772661e-01 -7.57775366e-01 -1.21838421e-01 5.73249102e-01
-4.86993253e-01 -5.77228852e-02 5.77836275e-01 9.08654153e-01
6.84039295e-03 2.92156279e-01 1.79883197e-01 -1.28232121e-01
-1.85627246e+00 3.51651698e-01 -4.94662791e-01 -5.64841777e-02
1.09890068e+00 -3.16569179e-01 -6.07213616e-01 -7.74313033e-01
-7.39778936e-01 4.85490412e-01 2.54465967e-01 7.43393004e-01
8.02930355e-01 -1.31651855e+00 -7.19985723e-01 -1.07022651e-01
3.72915566e-01 -2.54845321e-01 4.48976040e-01 1.07459402e+00
4.60679494e-02 4.87160176e-01 -6.24984264e-01 -7.53390431e-01
-1.51936758e+00 7.29605913e-01 -8.14771932e-03 -1.57498494e-02
-8.85762751e-01 1.00615811e+00 3.98413777e-01 3.46194863e-01
2.94071347e-01 -3.68691891e-01 -1.81883395e-01 -1.40329033e-01
7.86592305e-01 6.50245175e-02 -5.35636187e-01 -6.04133308e-01
-5.48249006e-01 3.66005301e-01 -2.73456335e-01 -1.72569051e-01
1.34622276e+00 -2.98362225e-01 3.55880082e-01 3.10295314e-01
8.01090837e-01 -6.83440924e-01 -2.05486250e+00 -2.74405479e-01
-1.13883883e-01 -4.02199984e-01 2.08819404e-01 -8.72012317e-01
-1.07371736e+00 9.85811412e-01 6.29919112e-01 -2.99087256e-01
1.24349833e+00 1.40142009e-01 6.33910060e-01 9.48169827e-01
2.23945141e-01 -1.08514977e+00 3.66471857e-01 6.53888583e-01
1.18530047e+00 -1.63231611e+00 -4.05556597e-02 -2.72334695e-01
-1.26203167e+00 9.26736116e-01 8.16897929e-01 5.04140109e-02
4.88949746e-01 1.96149245e-01 9.07175019e-02 -2.50573456e-01
-1.31818569e+00 -5.69140077e-01 2.24763900e-01 9.48511422e-01
1.88886091e-01 6.78709745e-02 2.80689657e-01 4.88322884e-01
1.46768451e-01 6.07356310e-01 5.65571487e-01 9.13446724e-01
-3.00235063e-01 -7.93609262e-01 3.56948197e-01 2.87957549e-01
3.28094773e-02 -2.87297159e-01 -4.30247605e-01 8.31881583e-01
-1.89926609e-01 1.01695633e+00 2.67071843e-01 -6.77231967e-01
1.66062459e-01 2.32289836e-01 7.37107038e-01 -5.19583762e-01
-1.28941223e-01 -2.81088620e-01 5.87530375e-01 -1.26745343e+00
-1.00149381e+00 -8.45365107e-01 -1.43173230e+00 -1.30196035e-01
8.48022103e-02 -7.04578876e-01 2.38713339e-01 1.26818085e+00
6.37698233e-01 5.61863422e-01 1.27062827e-01 -9.68944311e-01
-3.99198145e-01 -8.00644279e-01 -1.18141035e-02 3.06078821e-01
3.34802777e-01 -6.59929037e-01 -9.15597007e-02 7.16678143e-01] | [8.230287551879883, 0.5417092442512512] |
a01c24a5-0086-4703-b61a-17d6215ca9ae | self-supervised-learning-of-pretext-invariant | 1912.01991 | null | https://arxiv.org/abs/1912.01991v1 | https://arxiv.org/pdf/1912.01991v1.pdf | Self-Supervised Learning of Pretext-Invariant Representations | The goal of self-supervised learning from images is to construct image representations that are semantically meaningful via pretext tasks that do not require semantic annotations for a large training set of images. Many pretext tasks lead to representations that are covariant with image transformations. We argue that, instead, semantic representations ought to be invariant under such transformations. Specifically, we develop Pretext-Invariant Representation Learning (PIRL, pronounced as "pearl") that learns invariant representations based on pretext tasks. We use PIRL with a commonly used pretext task that involves solving jigsaw puzzles. We find that PIRL substantially improves the semantic quality of the learned image representations. Our approach sets a new state-of-the-art in self-supervised learning from images on several popular benchmarks for self-supervised learning. Despite being unsupervised, PIRL outperforms supervised pre-training in learning image representations for object detection. Altogether, our results demonstrate the potential of self-supervised learning of image representations with good invariance properties. | ['Laurens van der Maaten', 'Ishan Misra'] | 2019-12-04 | self-supervised-learning-of-pretext-invariant-1 | http://openaccess.thecvf.com/content_CVPR_2020/html/Misra_Self-Supervised_Learning_of_Pretext-Invariant_Representations_CVPR_2020_paper.html | http://openaccess.thecvf.com/content_CVPR_2020/papers/Misra_Self-Supervised_Learning_of_Pretext-Invariant_Representations_CVPR_2020_paper.pdf | cvpr-2020-6 | ['self-supervised-image-classification'] | ['computer-vision'] | [ 7.63815641e-01 3.88628393e-01 -2.68545926e-01 -5.24684727e-01
-5.25245249e-01 -5.43074310e-01 9.72309232e-01 1.04160786e-01
-3.35624874e-01 2.45343283e-01 4.18633968e-01 2.77369060e-02
-1.89355373e-01 -9.06419873e-01 -1.18346024e+00 -5.44534087e-01
1.55117765e-01 4.45394397e-01 2.42543846e-01 -4.80615675e-01
3.24178845e-01 2.53579825e-01 -1.63982046e+00 8.00231934e-01
4.35277134e-01 8.65583777e-01 2.04937354e-01 5.64782500e-01
-1.52353019e-01 1.30446219e+00 -3.26866120e-01 -2.54948735e-01
4.08357024e-01 -4.82984632e-01 -1.43510163e+00 2.85765737e-01
7.91834116e-01 -4.33825217e-02 -4.40284699e-01 1.13049567e+00
-1.27628714e-01 3.05507362e-01 1.18069792e+00 -1.49537635e+00
-1.26458156e+00 5.41845739e-01 -3.45663369e-01 1.54365972e-01
1.95350558e-01 2.44868159e-01 1.33186650e+00 -6.48279488e-01
8.70240927e-01 1.53032529e+00 5.65532148e-01 5.45706093e-01
-1.60539603e+00 -3.89241397e-01 -2.79760268e-02 2.42774814e-01
-9.03191805e-01 -3.64191979e-01 7.80562222e-01 -5.32550931e-01
9.65013504e-01 1.57040104e-01 2.30572775e-01 1.24631834e+00
1.93632650e-03 9.79468942e-01 1.20459461e+00 -5.95347166e-01
9.35974047e-02 -3.12762111e-02 2.81145781e-01 8.26973915e-01
2.25402817e-01 5.70550188e-02 -5.11816621e-01 1.60107076e-01
1.01397467e+00 1.07675254e-01 2.08007589e-01 -9.73334074e-01
-1.39026403e+00 9.52062726e-01 8.34785819e-01 3.24364483e-01
-1.32287204e-01 3.82677674e-01 6.53717399e-01 6.17304921e-01
3.02443475e-01 9.11510050e-01 -4.85998958e-01 3.44587058e-01
-4.78983223e-01 -1.10042691e-02 4.71096158e-01 1.06458747e+00
1.09020913e+00 5.96365705e-02 -1.81348681e-01 7.87758768e-01
-2.42213942e-02 3.41148049e-01 7.83700287e-01 -1.14201963e+00
1.58182591e-01 8.17500234e-01 -3.32677752e-01 -8.92712593e-01
-2.49249682e-01 -2.77878791e-01 -7.66951978e-01 4.95251834e-01
4.71088380e-01 3.98848385e-01 -9.64110434e-01 1.76624727e+00
-1.97001114e-01 6.66036010e-02 4.34287995e-01 6.84200048e-01
9.07222331e-01 4.57986981e-01 2.61114955e-01 2.49629036e-01
1.30001998e+00 -1.25258243e+00 -2.10906491e-01 -3.09286475e-01
6.81914985e-01 -5.97960055e-01 1.47574449e+00 1.92710727e-01
-7.55762398e-01 -7.53563046e-01 -1.05436409e+00 -2.60508388e-01
-5.47238410e-01 -4.07820046e-02 9.99648988e-01 2.89968282e-01
-8.70637238e-01 7.75713146e-01 -6.84060216e-01 -7.88673818e-01
8.05913866e-01 2.82298416e-01 -7.42264628e-01 -1.34126991e-02
-5.33417046e-01 8.99304152e-01 6.70553744e-01 -8.25149536e-01
-1.14456415e+00 -8.68025661e-01 -1.12596595e+00 5.13509009e-03
2.43623868e-01 -6.66575015e-01 1.19759977e+00 -1.69869316e+00
-1.26017296e+00 1.46811295e+00 2.21794501e-01 -7.24227548e-01
1.98324755e-01 -1.44841060e-01 -5.75942211e-02 4.93818551e-01
5.39875329e-01 1.09271121e+00 1.28386033e+00 -1.41714978e+00
-2.42214411e-01 -2.37885684e-01 3.16922873e-01 -8.41950253e-02
-4.64207023e-01 -1.16646670e-01 -1.05370961e-01 -8.00312340e-01
1.15845099e-01 -8.77865970e-01 -1.57448947e-01 8.25561434e-02
-3.34946722e-01 -4.10929501e-01 8.29254925e-01 -1.41887516e-01
2.31916949e-01 -2.13670278e+00 8.83785859e-02 4.08674106e-02
9.46609676e-02 9.08628106e-02 -7.74314821e-01 1.36749133e-01
-5.89097857e-01 -1.05219474e-02 -2.85155475e-01 -1.98487893e-01
1.04101889e-01 4.62412834e-01 -6.12589002e-01 4.52714652e-01
4.12354112e-01 1.24891019e+00 -1.10165560e+00 -5.53342223e-01
4.28049147e-01 8.01675916e-02 -7.09142506e-01 3.65322173e-01
-4.85107034e-01 4.13025349e-01 -3.16615134e-01 4.69355613e-01
3.22902203e-01 -5.78730166e-01 -8.56981874e-02 -3.32683027e-01
1.97332114e-01 2.70918816e-01 -6.57060206e-01 2.23249865e+00
-5.08926570e-01 6.92440450e-01 -4.87438649e-01 -1.56933892e+00
9.57030475e-01 -9.17864814e-02 6.51264787e-01 -9.03259277e-01
1.37717992e-01 -7.40969405e-02 -3.01691115e-01 -5.07516146e-01
3.08401227e-01 -1.44064307e-01 -5.39413514e-03 7.50658929e-01
9.49090958e-01 -4.24750328e-01 2.74869949e-01 4.28592086e-01
1.11475694e+00 4.16883528e-01 4.44683105e-01 -3.97645175e-01
3.71894956e-01 1.79612145e-01 2.16412500e-01 8.78788590e-01
3.38604786e-02 7.06590116e-01 4.87443417e-01 -6.93525970e-01
-1.23081994e+00 -1.21378267e+00 -6.63053840e-02 1.47718394e+00
8.74700844e-02 -6.29395902e-01 -6.55862153e-01 -1.07762194e+00
4.34909873e-02 5.59989274e-01 -9.73298192e-01 -3.05824190e-01
-2.71482259e-01 -3.52926910e-01 4.99744207e-01 5.42965829e-01
5.21064103e-01 -1.54537916e+00 -4.98611003e-01 -7.44318292e-02
1.34740621e-01 -1.29869270e+00 -1.99940011e-01 4.11939025e-01
-9.18920338e-01 -1.59418643e+00 -5.05401790e-01 -1.13848841e+00
1.16492581e+00 6.31600142e-01 1.34715474e+00 2.72974670e-01
-6.21031225e-01 9.80094790e-01 -5.64777792e-01 -3.66498739e-01
-6.09576523e-01 7.95692131e-02 -3.06826970e-03 4.84637320e-02
2.92903155e-01 -7.71596611e-01 -3.09614837e-01 1.84266016e-01
-1.21607828e+00 1.65225327e-01 7.32620060e-01 8.84038866e-01
6.56153083e-01 -1.40176997e-01 4.90228742e-01 -1.01079524e+00
1.63536787e-01 -3.44612837e-01 -4.47944671e-01 3.18310589e-01
-4.72226977e-01 7.03601837e-01 7.46319532e-01 -3.67249966e-01
-9.88238692e-01 4.31858718e-01 1.35890126e-01 -3.69116277e-01
-5.27922213e-01 -1.64296493e-01 8.00323859e-02 -3.04607570e-01
1.19438303e+00 2.45972916e-01 1.99100927e-01 -3.72277081e-01
8.15205097e-01 2.51935214e-01 9.31423843e-01 -9.20386374e-01
1.02768421e+00 8.29663932e-01 1.61272705e-01 -7.74384797e-01
-1.30648613e+00 -6.14882886e-01 -1.03776765e+00 1.77371904e-01
1.01327443e+00 -7.82660127e-01 -5.75762451e-01 1.70136318e-02
-9.74939287e-01 -4.29570884e-01 -8.57375443e-01 1.78041041e-01
-1.26050615e+00 5.75622022e-01 -2.70293713e-01 -2.22209379e-01
-2.40393996e-01 -8.26161981e-01 1.03755379e+00 -3.48969772e-02
-3.90364647e-01 -1.13424027e+00 1.62580669e-01 4.85854119e-01
2.52739936e-01 2.11484343e-01 1.00298810e+00 -6.97431982e-01
-4.47221875e-01 2.40552008e-01 -5.49828410e-01 6.74857616e-01
4.04976845e-01 -5.29657841e-01 -1.09533274e+00 -3.34590673e-01
-1.95876628e-01 -9.67033327e-01 1.29019475e+00 8.17347914e-02
1.47083080e+00 -4.48943853e-01 1.09030316e-02 8.81008327e-01
1.46591055e+00 -4.71223027e-01 7.36554086e-01 6.43485248e-01
6.97470903e-01 6.68851674e-01 3.67598504e-01 1.64145544e-01
6.37711138e-02 4.26977813e-01 6.56726718e-01 -2.47857496e-01
-4.95723516e-01 -5.94322145e-01 4.84544754e-01 3.82349700e-01
-1.30506843e-01 3.92943799e-01 -5.83822668e-01 5.57366371e-01
-1.96713150e+00 -9.51941907e-01 -1.43432617e-02 1.89723933e+00
8.69183004e-01 -1.97476316e-02 7.57181868e-02 -5.73433153e-02
3.87573808e-01 2.09609494e-01 -5.42181969e-01 -2.55479604e-01
-2.83767819e-01 6.65852606e-01 6.90717578e-01 3.17408033e-02
-1.57400298e+00 1.38432276e+00 6.85737896e+00 6.26767397e-01
-7.64999628e-01 1.82640895e-01 4.03227717e-01 5.16019821e-01
-1.85865730e-01 1.59786433e-01 -3.28890860e-01 1.38646010e-02
6.24930322e-01 8.07284266e-02 3.36542904e-01 1.21401167e+00
-3.46357375e-01 2.11726978e-01 -1.45489323e+00 1.11894834e+00
4.70315754e-01 -1.63720763e+00 5.97541273e-01 -2.48256922e-01
1.06648386e+00 -3.89684588e-02 1.72783285e-01 3.17264438e-01
6.94820583e-01 -1.21665895e+00 5.57311475e-01 3.88480663e-01
6.08038008e-01 -4.68729168e-01 3.20638984e-01 -1.49432700e-02
-9.09591615e-01 -6.78815320e-02 -8.83562982e-01 -7.36463740e-02
-3.65348548e-01 5.45165390e-02 -7.55927801e-01 3.98694187e-01
5.27775168e-01 1.30165756e+00 -1.05487788e+00 7.63572633e-01
-6.32041276e-01 5.22992194e-01 1.02237873e-01 4.21035469e-01
4.64998960e-01 -2.92907394e-02 2.01912224e-01 1.35175312e+00
-7.84129426e-02 -1.51309773e-01 2.29537845e-01 1.14248908e+00
-3.23396951e-01 6.26113936e-02 -1.05352759e+00 -8.70575830e-02
-2.01146886e-01 1.21332717e+00 -8.29953671e-01 -4.32702363e-01
-4.17472124e-01 1.26122510e+00 4.20901626e-01 2.89077520e-01
-4.15375441e-01 -8.92207623e-02 7.40091264e-01 7.50171300e-03
2.27804735e-01 -1.94136217e-01 -2.69092023e-01 -1.23806715e+00
-3.37880313e-01 -9.54009712e-01 5.59076309e-01 -1.05074656e+00
-1.59770000e+00 2.16025949e-01 -2.90101767e-02 -1.19996393e+00
-2.29937479e-01 -1.14285731e+00 -6.20716095e-01 2.10957956e-02
-1.64336824e+00 -1.65992343e+00 -3.96286607e-01 1.03868151e+00
9.71885562e-01 -6.02260768e-01 1.14622593e+00 -2.22469911e-01
5.82585484e-02 4.22304928e-01 -4.70176376e-02 4.51650113e-01
7.70625055e-01 -1.52126503e+00 3.97656143e-01 5.92682421e-01
9.42225277e-01 6.20268941e-01 4.45551306e-01 -2.59866923e-01
-1.47297633e+00 -1.17054677e+00 2.97101676e-01 -6.98815405e-01
7.53462374e-01 -2.43612245e-01 -8.13540161e-01 1.06471694e+00
2.81871676e-01 4.10803348e-01 5.25095701e-01 1.83875889e-01
-1.16336453e+00 8.06906968e-02 -7.47452140e-01 5.04835606e-01
1.24640048e+00 -8.35938752e-01 -1.08941841e+00 8.33980560e-01
6.88197553e-01 5.95927127e-02 -6.48410738e-01 1.98922440e-01
3.49968523e-01 -7.01949716e-01 1.53132308e+00 -1.21827149e+00
5.61805487e-01 -1.30555466e-01 -1.77657709e-01 -1.32732761e+00
-5.34926176e-01 -5.02391279e-01 3.08024734e-01 9.61199164e-01
1.08683184e-01 -2.79626459e-01 8.06877255e-01 2.79587179e-01
-7.93010294e-02 1.97883882e-02 -5.20773768e-01 -1.27493036e+00
2.24063158e-01 -3.42515349e-01 3.15524936e-01 1.17781103e+00
1.26814377e-02 5.64043701e-01 -2.05952048e-01 -1.55791432e-01
9.65532959e-01 1.54325485e-01 1.07024157e+00 -1.43651295e+00
-1.93760768e-01 -3.23789179e-01 -7.54680336e-01 -8.59963477e-01
8.08459818e-01 -1.52096021e+00 -9.20746177e-02 -1.47343886e+00
6.04017198e-01 -3.11006129e-01 -4.75855589e-01 1.00858819e+00
1.84396341e-01 5.32308221e-01 2.15735748e-01 3.82781684e-01
-1.24917042e+00 5.38108706e-01 1.22447503e+00 -6.78141177e-01
1.25819534e-01 -4.18597728e-01 -7.66245484e-01 1.00090933e+00
9.08549249e-01 -6.00037277e-01 -5.71192503e-01 -4.20974195e-01
1.16914690e-01 -7.49008119e-01 7.52426803e-01 -9.56092596e-01
2.17235163e-02 -2.96591192e-01 4.35687661e-01 -5.56246564e-02
1.40643910e-01 -7.40517199e-01 -3.69016737e-01 3.20643455e-01
-7.31180131e-01 -2.20105469e-01 4.34119143e-02 3.90112251e-01
-2.15500861e-01 -3.88666242e-01 9.72222388e-01 -4.83757794e-01
-1.22244537e+00 1.35233298e-01 -6.32938221e-02 2.88854063e-01
7.95517027e-01 -1.66583627e-01 -4.03875202e-01 -4.10145402e-01
-6.77955508e-01 -1.27640918e-01 5.77625513e-01 7.80725658e-01
7.15692878e-01 -1.36100507e+00 -6.78712726e-01 2.36399263e-01
7.41164088e-01 -2.14426875e-01 -9.53372493e-02 3.46904248e-01
-4.71673667e-01 3.89408648e-01 -8.48340988e-01 -6.58879459e-01
-1.24731410e+00 8.77822042e-01 6.46254495e-02 -1.36620149e-01
-1.02353024e+00 6.17922485e-01 6.60508752e-01 -6.98547721e-01
1.15310535e-01 -4.02791589e-01 -2.32768565e-01 -1.69824913e-01
4.28603590e-01 -1.17682584e-03 -7.49158412e-02 -7.69567311e-01
-2.53828969e-02 6.00925148e-01 -3.59993055e-02 7.79323801e-02
1.70082104e+00 2.89906442e-01 -3.22930366e-01 1.17095344e-01
1.45486796e+00 -5.37572980e-01 -1.36167526e+00 -4.98678207e-01
2.82006294e-01 -4.78400469e-01 -1.75437406e-01 -5.08244455e-01
-1.05670202e+00 7.60582507e-01 4.36909407e-01 -1.35868222e-01
8.73144269e-01 4.47457135e-01 4.53363985e-01 1.07065523e+00
4.71505672e-01 -1.16282368e+00 1.00841892e+00 7.66494453e-01
1.20579290e+00 -1.63059843e+00 8.75683501e-02 -2.96607703e-01
-7.34804213e-01 1.33487344e+00 6.31332636e-01 -6.22333765e-01
3.73329580e-01 -1.55961975e-01 -1.43229574e-01 -3.93032879e-01
-3.95242631e-01 -6.43350840e-01 5.26970923e-01 1.11552536e+00
2.46038750e-01 -4.92025949e-02 1.85094804e-01 4.92539346e-01
-3.36483777e-01 -3.36271942e-01 3.37871760e-01 7.57276118e-01
-4.15027440e-01 -1.24034584e+00 -3.34313005e-01 2.85670638e-01
-1.39739871e-01 -2.11939700e-02 -7.23824024e-01 8.30951989e-01
8.06927457e-02 7.69654095e-01 7.87905529e-02 -2.51951694e-01
1.79442450e-01 1.62101597e-01 8.00035357e-01 -9.87926841e-01
-4.27340150e-01 -5.14966846e-01 -1.85171947e-01 -7.83912838e-01
-7.53481269e-01 -3.73142183e-01 -1.34397304e+00 2.75611103e-01
2.16063097e-01 -1.15500204e-01 4.96680260e-01 1.02651966e+00
2.06790775e-01 4.98492926e-01 5.13384938e-01 -8.48273039e-01
-5.14763415e-01 -8.55741918e-01 -4.19359565e-01 1.29435706e+00
2.25958794e-01 -7.59184301e-01 -1.85010895e-01 5.49716115e-01] | [9.656173706054688, 2.1529197692871094] |
24dd34c7-e0ae-468d-a7bf-c5ff4844e9a5 | structured-pruning-for-multi-task-deep-neural | 2304.06840 | null | https://arxiv.org/abs/2304.06840v1 | https://arxiv.org/pdf/2304.06840v1.pdf | Structured Pruning for Multi-Task Deep Neural Networks | Although multi-task deep neural network (DNN) models have computation and storage benefits over individual single-task DNN models, they can be further optimized via model compression. Numerous structured pruning methods are already developed that can readily achieve speedups in single-task models, but the pruning of multi-task networks has not yet been extensively studied. In this work, we investigate the effectiveness of structured pruning on multi-task models. We use an existing single-task filter pruning criterion and also introduce an MTL-based filter pruning criterion for estimating the filter importance scores. We prune the model using an iterative pruning strategy with both pruning methods. We show that, with careful hyper-parameter tuning, architectures obtained from different pruning methods do not have significant differences in their performances across tasks when the number of parameters is similar. We also show that iterative structure pruning may not be the best way to achieve a well-performing pruned model because, at extreme pruning levels, there is a high drop in performance across all tasks. But when the same models are randomly initialized and re-trained, they show better results. | ['Hui Guan', 'Lijun Zhang', 'Siddhant Garg'] | 2023-04-13 | null | null | null | null | ['model-compression'] | ['methodology'] | [ 1.96808100e-01 -7.42319003e-02 7.10686296e-02 -5.21952629e-01
-4.99110550e-01 -1.16599761e-01 3.06677729e-01 6.83478266e-02
-8.80498588e-01 6.77960575e-01 -3.85204777e-02 -3.11994284e-01
-3.50169212e-01 -4.84687537e-01 -5.90911746e-01 -4.46272641e-01
1.72121063e-01 6.41172945e-01 7.64669001e-01 5.95881753e-02
-1.13504101e-02 2.52626210e-01 -1.71490943e+00 6.35171771e-01
8.43092918e-01 9.01737750e-01 8.57337415e-01 7.85363495e-01
-2.51011312e-01 6.58365667e-01 -1.02009463e+00 -7.18209445e-01
5.20693302e-01 1.10012502e-03 -9.24080372e-01 -4.51235712e-01
7.31098473e-01 -2.32546329e-01 -1.14672650e-02 1.13154757e+00
4.15803820e-01 1.47425458e-01 3.30093533e-01 -7.27912903e-01
4.88021933e-02 8.65133584e-01 -1.91122547e-01 5.59941411e-01
-4.66140598e-01 -2.27505341e-01 1.02712905e+00 -6.78486645e-01
4.29997116e-01 1.45490205e+00 8.43809128e-01 5.57595432e-01
-1.50597572e+00 -7.54453361e-01 4.91940349e-01 9.23212022e-02
-1.01530969e+00 -5.04165590e-01 3.23572636e-01 -1.20851688e-01
1.55274105e+00 2.46869158e-02 7.03315198e-01 8.81266177e-01
3.32719982e-01 7.15074837e-01 7.14644909e-01 -5.33219755e-01
9.11303610e-02 8.30447748e-02 4.47807580e-01 6.33889198e-01
7.46604621e-01 -1.60220489e-01 -7.21348107e-01 -2.06472531e-01
4.05659050e-01 1.47481766e-02 4.83590476e-02 9.72259697e-03
-8.44219863e-01 8.04461479e-01 1.52876358e-02 4.80527371e-01
-3.10591221e-01 3.68765771e-01 7.43225396e-01 6.97007716e-01
7.30434299e-01 4.54541892e-01 -1.02260566e+00 -2.11347669e-01
-1.55095983e+00 4.95212108e-01 9.12522793e-01 8.16873550e-01
6.03842199e-01 2.07925916e-01 -1.83782980e-01 1.18935299e+00
-8.78862143e-02 -4.23743613e-02 6.39934838e-01 -9.84176278e-01
6.27307355e-01 3.63752365e-01 -3.37343127e-01 -4.91020679e-01
-5.73045075e-01 -5.53393364e-01 -7.15223312e-01 4.11432534e-01
5.08325398e-01 -4.99695837e-01 -1.04823434e+00 1.65918136e+00
-1.40207812e-01 -2.05277368e-01 5.39258383e-02 3.80901486e-01
6.53948903e-01 3.58187228e-01 4.93357927e-01 -1.08357139e-01
1.48820603e+00 -1.10180748e+00 -6.52695000e-01 -5.88341892e-01
9.10176694e-01 -8.56821060e-01 8.10282648e-01 7.17745483e-01
-1.31953752e+00 -6.23593986e-01 -1.15354478e+00 -2.91110992e-01
-3.73008013e-01 4.61927116e-01 6.67618155e-01 7.40082860e-01
-1.13723505e+00 9.05582309e-01 -1.00412166e+00 -2.47322366e-01
7.29462564e-01 5.45372665e-01 -3.58728804e-02 -9.06880274e-02
-9.89283442e-01 1.49515402e+00 8.13870966e-01 -2.56293207e-01
-9.36578870e-01 -9.41292703e-01 -2.16764688e-01 4.54241753e-01
2.12262794e-01 -7.43745506e-01 1.54947686e+00 -8.19965661e-01
-1.20755482e+00 6.47769570e-01 -4.23272401e-01 -9.78989482e-01
3.04341584e-01 -5.32160461e-01 -1.93213925e-01 -1.23077340e-01
-2.91551679e-01 9.52384472e-01 7.82957971e-01 -7.72135913e-01
-9.40181613e-01 -1.83580175e-01 1.73266992e-01 3.13082159e-01
-5.32017231e-01 4.33901280e-01 -2.32430592e-01 -7.98507273e-01
-1.29378131e-02 -6.20044291e-01 -3.96897733e-01 -1.07125878e-01
-1.64596200e-01 -3.53214771e-01 8.51642728e-01 -4.73100513e-01
1.40940797e+00 -1.92431080e+00 -1.33656502e-01 -8.00037012e-02
5.31502187e-01 6.77784026e-01 -2.70864308e-01 -3.99298742e-02
8.71868953e-02 3.99841547e-01 1.08956657e-01 -7.57592499e-01
-3.25502634e-01 4.06739086e-01 6.36866465e-02 -8.72054845e-02
1.71692774e-01 5.36069512e-01 -4.98473018e-01 -5.95484138e-01
-1.10804394e-01 3.39056492e-01 -7.53797829e-01 -3.44609082e-01
-3.55254024e-01 -3.33757669e-01 -2.53349364e-01 3.09779555e-01
6.02940857e-01 -1.29158780e-01 2.53465116e-01 -1.25983879e-01
8.45817700e-02 7.79367745e-01 -1.00531638e+00 1.53084052e+00
-5.05754530e-01 1.06975102e+00 2.08537787e-01 -1.10673869e+00
8.57521653e-01 2.62908608e-01 2.89791107e-01 -6.40548825e-01
-1.25845596e-01 3.03207636e-01 6.74837053e-01 -3.47027509e-03
6.07831836e-01 -1.94052681e-01 3.73628944e-01 4.50230807e-01
4.76884991e-01 9.05579180e-02 6.75158501e-01 6.94998773e-03
1.40590644e+00 -4.11437184e-01 4.18020450e-02 -5.94135463e-01
7.93180019e-02 1.14003785e-01 6.40276730e-01 9.08385932e-01
-1.54886898e-02 4.26880300e-01 6.81088388e-01 -6.61022186e-01
-1.26506293e+00 -5.87532103e-01 -3.98543030e-01 1.60980117e+00
-3.89042050e-01 -7.46904492e-01 -8.32703710e-01 -5.70656359e-01
-1.60446428e-02 5.80251813e-01 -3.26263100e-01 -6.14450015e-02
-7.39606321e-01 -1.21080971e+00 7.79988945e-01 4.46034670e-01
5.18522322e-01 -9.21733260e-01 -1.13011742e+00 4.76344734e-01
-1.31358644e-02 -1.23135376e+00 5.15550487e-02 1.00947607e+00
-1.65933895e+00 -9.49349284e-01 -7.54178047e-01 -7.55794466e-01
4.15354788e-01 2.82108992e-01 1.35734010e+00 2.43961468e-01
-5.77807128e-02 -3.81447583e-01 -1.58465996e-01 -8.74965668e-01
-3.15721840e-01 6.46086633e-01 -2.60372926e-02 -6.78863704e-01
6.02093160e-01 -5.67319572e-01 -3.33797425e-01 2.11531430e-01
-6.93622053e-01 1.92724392e-01 8.03179204e-01 7.84662724e-01
6.36989653e-01 2.49436900e-01 5.41626573e-01 -1.10862660e+00
1.07964790e+00 3.22094187e-02 -5.44406414e-01 3.82581562e-01
-7.70395815e-01 4.36707586e-01 7.18083382e-01 -8.26638043e-01
-1.09634042e+00 1.34844603e-02 -1.57960176e-01 -5.73376536e-01
3.00771720e-03 3.12143832e-01 2.95542806e-01 -1.64629146e-01
7.45938420e-01 -1.54857635e-01 -2.58319080e-01 -9.65253949e-01
4.79702763e-02 1.16123877e-01 -7.91558176e-02 -5.27296484e-01
1.77402660e-01 1.23726830e-01 -8.57910290e-02 -6.65190578e-01
-9.41804349e-01 -2.40942448e-01 -6.58908427e-01 1.45024076e-01
5.81799388e-01 -8.98430169e-01 -2.68998474e-01 2.32894212e-01
-1.51946199e+00 -7.17268646e-01 -2.52070218e-01 5.45778573e-01
-2.40189761e-01 1.66023031e-01 -6.63770676e-01 -6.68219090e-01
-5.36318481e-01 -1.24533820e+00 7.88718104e-01 -5.51485382e-02
-2.84945130e-01 -9.13869977e-01 -4.60411049e-02 -1.48858711e-01
5.88634372e-01 -4.88441616e-01 1.30027640e+00 -1.14664209e+00
-3.53523999e-01 1.51410028e-01 -3.65393341e-01 5.58215976e-01
-2.92086303e-01 -2.94558797e-02 -9.66466367e-01 -1.36524215e-01
1.28320679e-01 -3.23973238e-01 1.49289918e+00 8.61504257e-01
1.26709008e+00 -2.01556101e-01 -5.65894186e-01 6.39520526e-01
1.33299947e+00 2.07050294e-01 4.99953896e-01 4.16464001e-01
3.11280996e-01 5.08649826e-01 2.49967739e-01 1.50830165e-01
-1.52177125e-01 5.52317262e-01 2.69101173e-01 1.33102894e-01
-3.81809175e-01 2.17776746e-01 4.08159465e-01 8.32224190e-01
-2.23208055e-01 -1.38805494e-01 -8.91871452e-01 3.94264847e-01
-1.65851104e+00 -9.49713171e-01 -1.43416718e-01 2.04354572e+00
9.56434786e-01 7.79710233e-01 -4.08547930e-02 1.14304595e-01
4.92542058e-01 1.08179368e-01 -5.19133627e-01 -7.96256065e-01
-2.26667166e-01 4.90357846e-01 6.45398319e-01 4.05102044e-01
-1.03886676e+00 1.06819427e+00 7.52837467e+00 1.03185546e+00
-7.55862057e-01 5.91110229e-01 6.49390519e-01 -8.46257389e-01
1.18891463e-01 -3.54968496e-02 -1.57174385e+00 3.31033856e-01
1.28630388e+00 -2.07357004e-01 2.07593009e-01 1.12447083e+00
1.22280538e-01 -2.80947596e-01 -1.06434691e+00 8.93448234e-01
-2.50439495e-01 -1.35139441e+00 2.10276768e-01 1.05827451e-01
7.60165572e-01 4.89252120e-01 -2.96065956e-01 5.74490666e-01
5.44077754e-01 -8.59148622e-01 8.05177808e-01 3.14501196e-01
4.83062029e-01 -7.42045879e-01 6.50652826e-01 4.89321619e-01
-1.06338954e+00 -3.22858751e-01 -1.11782444e+00 -2.89493054e-02
1.06262460e-01 1.02284276e+00 -4.59324628e-01 5.81870899e-02
9.68484998e-01 2.28372842e-01 -7.90812075e-01 1.40940678e+00
1.49480000e-01 5.40272593e-01 -7.24869668e-01 -1.47935525e-01
1.40977889e-01 2.87961215e-02 4.71646637e-01 1.57989621e+00
2.77558506e-01 -2.61941463e-01 -9.70438197e-02 6.77941203e-01
-1.73267618e-01 -5.15796542e-02 -4.72895682e-01 3.08351785e-01
4.34726387e-01 1.01340568e+00 -7.56500244e-01 -5.53753018e-01
-4.99986231e-01 5.00980496e-01 7.78552592e-01 3.37884277e-01
-6.87604249e-01 -3.66865635e-01 9.91306782e-01 8.70382413e-02
4.97645050e-01 -2.31783599e-01 -6.16641045e-01 -9.39507008e-01
2.22088816e-03 -7.17954218e-01 3.06298912e-01 -5.08826673e-01
-9.01570737e-01 8.49897265e-01 3.08441430e-01 -6.65293753e-01
-1.86818868e-01 -8.76158059e-01 -6.04665101e-01 8.05290639e-01
-1.46295905e+00 -5.66660106e-01 7.68684074e-02 3.23718607e-01
1.04491508e+00 -2.75051802e-01 7.94119418e-01 4.37042981e-01
-6.36527061e-01 7.70575404e-01 1.69736460e-01 -1.39471695e-01
5.61049342e-01 -9.55184340e-01 4.19287533e-01 1.03122211e+00
2.94924706e-01 5.70937276e-01 6.94441378e-01 -6.43884957e-01
-6.55700028e-01 -1.19714057e+00 1.02561307e+00 -1.48269609e-01
3.13601971e-01 -4.71818864e-01 -1.04114354e+00 8.10224652e-01
1.48101702e-01 -3.42507094e-01 5.09987652e-01 5.27631462e-01
-1.54708222e-01 -1.95645094e-01 -9.51134264e-01 5.97299933e-01
1.28041804e+00 -1.01155482e-01 -6.81714058e-01 4.61272180e-01
8.42524648e-01 -2.51383066e-01 -7.26161301e-01 4.05901253e-01
5.84789336e-01 -1.01370585e+00 8.79007995e-01 -6.78568959e-01
2.57989019e-01 3.81714672e-01 3.97963403e-03 -1.27698457e+00
-3.52616400e-01 -3.71975899e-01 -3.65213066e-01 8.50876272e-01
8.24248850e-01 -6.29277349e-01 1.05501211e+00 6.42561436e-01
-2.97843009e-01 -1.03628719e+00 -1.06471550e+00 -1.07344019e+00
1.96438864e-01 -6.89466357e-01 3.71264309e-01 2.11504251e-01
-3.68955225e-01 4.32719797e-01 -1.12839408e-01 -3.20360124e-01
4.68938112e-01 -3.96917731e-01 3.89691383e-01 -1.82630849e+00
-2.20569208e-01 -9.99011457e-01 5.95589876e-02 -1.27537572e+00
4.44537476e-02 -8.34049106e-01 -9.52871218e-02 -1.52417338e+00
1.81068286e-01 -5.72426915e-01 -2.22697780e-01 7.19083130e-01
-5.55907674e-02 -2.32148260e-01 3.94652903e-01 3.70005429e-01
-5.93517721e-01 2.35580131e-01 9.81095195e-01 1.07608713e-01
-3.21913093e-01 2.75913805e-01 -5.10247767e-01 8.82484734e-01
9.18882906e-01 -1.01460266e+00 -5.74827194e-01 -1.05499852e+00
2.67674655e-01 -4.56810802e-01 1.63597733e-01 -1.44986343e+00
4.89099085e-01 2.71025926e-01 4.22555834e-01 -7.32384026e-01
5.65901697e-01 -7.18611300e-01 3.39655541e-02 6.48018062e-01
-4.94194984e-01 3.69382322e-01 5.64309478e-01 5.74113548e-01
-1.42304212e-01 -9.50132906e-01 1.02251458e+00 -3.37408334e-01
-5.20585179e-01 1.03737660e-01 -6.13719404e-01 1.45808116e-01
5.22606015e-01 -4.89216119e-01 -2.91692197e-01 1.17602862e-01
-7.31495023e-01 1.23950534e-01 -2.07241829e-02 2.25487784e-01
4.60208625e-01 -9.09691513e-01 -6.96571767e-01 5.35383746e-02
-4.11220759e-01 1.98432088e-01 -2.01892164e-02 6.99872971e-01
-4.33615535e-01 7.55178928e-01 -1.63527995e-01 -3.86885166e-01
-1.65180779e+00 1.64917544e-01 3.79129499e-01 -1.02098477e+00
-5.90789795e-01 1.02374411e+00 9.93530750e-02 -1.22377053e-01
5.89197636e-01 -7.73303509e-01 -8.88298824e-02 2.93148130e-01
5.45448124e-01 4.76673692e-01 5.97927988e-01 -1.18715046e-02
-6.29652366e-02 2.93824404e-01 -5.50950170e-01 2.20395904e-02
1.47840977e+00 2.44014472e-01 2.81245172e-01 1.96035624e-01
9.00768816e-01 -5.34997165e-01 -1.18322694e+00 -2.28044137e-01
2.69947112e-01 -1.57237321e-01 3.20222795e-01 -6.32891655e-01
-1.32030547e+00 1.11286139e+00 4.28313226e-01 2.86007404e-01
1.17400634e+00 -2.53869236e-01 7.69617736e-01 8.51173401e-01
2.38813132e-01 -1.28011644e+00 -3.20979767e-02 1.03204834e+00
5.63208878e-01 -8.75652015e-01 2.69086331e-01 -1.78585693e-01
-4.55326796e-01 1.18895698e+00 8.15919697e-01 -1.43420979e-01
8.28978837e-01 6.51725709e-01 -3.85375917e-01 -2.47307763e-01
-1.30019021e+00 -3.58964622e-01 2.54120324e-02 5.54227948e-01
3.73267710e-01 -2.20440269e-01 -3.85719568e-01 6.81247413e-01
-3.93425703e-01 -6.55947253e-02 1.81648493e-01 6.30785942e-01
-8.40816379e-01 -1.15297997e+00 1.06848264e-02 1.14929032e+00
-8.98074687e-01 -5.52179813e-01 -8.94070342e-02 9.10237134e-01
1.78633288e-01 7.04670250e-01 1.92202002e-01 -3.70778382e-01
2.83708066e-01 5.89437008e-01 4.82039601e-01 -9.23328817e-01
-1.06347215e+00 -3.36305350e-02 3.86215687e-01 -3.13338012e-01
-2.57502615e-01 -4.66047138e-01 -9.57225621e-01 -5.09371817e-01
-5.79666257e-01 -1.81640521e-01 7.48919368e-01 8.80250275e-01
4.63341922e-01 6.71566427e-01 -3.35497946e-01 -8.67630005e-01
-6.03067875e-01 -1.16273057e+00 -3.07857990e-01 -1.55368103e-02
2.93720551e-02 -7.97384441e-01 -2.14191645e-01 -1.00629553e-01] | [8.608660697937012, 3.3013458251953125] |
dddc1c6b-cc3a-4713-9f89-766c000a2177 | training-neural-networks-based-on-imperialist | 1704.04095 | null | http://arxiv.org/abs/1704.04095v1 | http://arxiv.org/pdf/1704.04095v1.pdf | Training Neural Networks Based on Imperialist Competitive Algorithm for Predicting Earthquake Intensity | In this study we determined neural network weights and biases by Imperialist
Competitive Algorithm (ICA) in order to train network for predicting earthquake
intensity in Richter. For this reason, we used dependent parameters like
earthquake occurrence time, epicenter's latitude and longitude in degree, focal
depth in kilometer, and the seismological center distance from epicenter and
earthquake focal center in kilometer which has been provided by Berkeley data
base. The studied neural network has two hidden layer: its first layer has 16
neurons and the second layer has 24 neurons. By using ICA algorithm, average
error for testing data is 0.0007 with a variance equal to 0.318. The earthquake
prediction error in Richter by MSE criteria for ICA algorithm is 0.101, but by
using GA, the MSE value is 0.115. | ['Mohsen Moradi'] | 2017-02-13 | null | null | null | null | ['earthquake-prediction'] | ['computer-vision'] | [-3.38432461e-01 -2.38749627e-02 -8.16345401e-03 3.80113982e-02
-1.84306890e-01 -3.83330017e-01 2.39600509e-01 -1.65272817e-01
-9.88919556e-01 1.35816026e+00 4.56265695e-02 -4.23961490e-01
-3.22807580e-01 -1.03087735e+00 -6.72999263e-01 -9.43733513e-01
-3.61466736e-01 4.38147098e-01 9.92305428e-02 -3.88681203e-01
7.45357692e-01 6.69286907e-01 -8.80731404e-01 -1.77039564e-01
5.53407609e-01 9.91675079e-01 -1.32871745e-02 6.19858801e-01
4.48297679e-01 9.27610636e-01 -9.04094338e-01 1.30633339e-02
3.01457912e-01 -4.83579963e-01 -6.36747599e-01 -6.32911146e-01
-7.86611557e-01 -1.33533612e-01 -2.75339484e-01 7.83901155e-01
5.36775768e-01 1.57101393e-01 1.10526454e+00 -1.05300450e+00
-5.59178829e-01 9.52740312e-01 -4.66600329e-01 6.55195951e-01
-3.95257711e-01 -3.12456846e-01 3.73224169e-01 -5.74772537e-01
4.36000019e-01 6.45119429e-01 8.32446098e-01 3.80527645e-01
-7.49258995e-01 -8.55701447e-01 -5.56353092e-01 2.47810781e-01
-1.70010245e+00 1.63653076e-01 8.34426165e-01 -4.44087535e-01
1.03994119e+00 1.61306322e-01 8.77186894e-01 7.48637736e-01
6.93330109e-01 3.66062298e-02 9.38394070e-01 -4.32339698e-01
5.80425680e-01 1.00954287e-01 -1.07129507e-01 1.92512140e-01
7.38405704e-01 9.92278680e-02 1.75458223e-01 -3.29451039e-02
1.10727501e+00 -2.46976212e-01 2.87624653e-02 7.23727882e-01
-1.04533768e+00 1.25886524e+00 5.99495351e-01 8.47634435e-01
-5.25679231e-01 4.29785758e-01 2.90173531e-01 3.42212081e-01
1.28344744e-01 9.15035009e-01 -5.39818048e-01 1.00395851e-01
-6.65892839e-01 1.39198408e-01 8.19743574e-01 1.72798246e-01
4.60258335e-01 5.68763614e-01 5.31409442e-01 5.57578743e-01
-5.22090169e-03 5.94675303e-01 8.83816063e-01 -7.01162457e-01
4.42690343e-01 5.22947848e-01 1.84529781e-01 -1.26779175e+00
-7.58779645e-01 -7.58971155e-01 -1.18432057e+00 5.24722375e-02
4.69809085e-01 -9.55807388e-01 -6.28478706e-01 1.49351585e+00
-2.13776395e-01 3.30552965e-01 3.85692596e-01 1.07865262e+00
4.69875455e-01 8.46694469e-01 4.33862209e-02 -2.23081321e-01
1.11355817e+00 -5.50391436e-01 -3.13887060e-01 -1.77142039e-01
6.87182605e-01 -4.04724687e-01 3.19276154e-01 4.50404763e-01
-8.56034160e-01 -3.22537065e-01 -1.26661897e+00 6.27065420e-01
-6.08919382e-01 3.38279068e-01 3.01750243e-01 4.53599125e-01
-5.83371460e-01 5.41505873e-01 -5.32279849e-01 -6.46785647e-02
-2.44857207e-01 3.65852296e-01 -2.67428488e-01 7.74367809e-01
-1.85193002e+00 1.15598631e+00 9.08723950e-01 1.58659011e-01
-6.78032041e-01 -1.70382753e-01 -4.47212815e-01 3.00246537e-01
-1.81734011e-01 -2.92869002e-01 6.12473488e-01 -7.34988570e-01
-1.22399259e+00 2.57659733e-01 6.52845204e-01 -8.57991278e-01
3.73951286e-01 2.16768190e-01 -7.80052066e-01 1.31701678e-01
-1.73249692e-01 4.29576755e-01 2.43120223e-01 -1.04393983e+00
-5.34555912e-01 -3.67229342e-01 -4.49309379e-01 2.35912632e-02
-5.73763132e-01 -1.73736155e-01 1.06948063e-01 -8.21530342e-01
5.36025286e-01 -8.65065157e-01 -5.63632131e-01 -1.13266337e+00
-4.96242911e-01 -5.96573055e-02 4.64359432e-01 -7.82238841e-01
1.52368021e+00 -2.01283336e+00 -2.20460489e-01 8.56452346e-01
-5.19997217e-02 -1.26276195e-01 3.87053400e-01 3.42628151e-01
-2.16285840e-01 3.11265439e-01 -3.90838712e-01 7.51406550e-01
-3.84205371e-01 1.58331662e-01 -1.14446826e-01 3.99130344e-01
1.04878835e-01 3.76273155e-01 -4.97849494e-01 -4.66745287e-01
-4.93879393e-02 5.58232963e-01 -4.27589566e-01 -1.40644088e-02
5.44777870e-01 5.75398132e-02 -7.34903395e-01 3.46821457e-01
5.02789855e-01 -3.57168727e-02 -2.38374785e-01 1.36787845e-02
-4.14795190e-01 -2.83179969e-01 -1.45837128e+00 7.70369291e-01
-3.81108195e-01 7.69671798e-01 -5.69462657e-01 -1.14526939e+00
1.60433364e+00 7.30012357e-01 2.60132164e-01 -5.33260167e-01
6.46401882e-01 4.43523556e-01 2.32278541e-01 -6.78279877e-01
4.01548803e-01 -9.19877067e-02 -1.06755830e-01 4.89606649e-01
-3.02132428e-01 -2.94938460e-02 3.98536503e-01 -2.24062905e-01
9.09085751e-01 -5.41680455e-01 9.69296508e-03 -3.60640138e-01
4.51438308e-01 1.92419603e-01 6.41096950e-01 7.12733626e-01
2.85841376e-01 5.66278279e-01 8.29547703e-01 -7.07700014e-01
-1.52607906e+00 -4.84901756e-01 -5.31734288e-01 4.65071976e-01
-1.33152187e-01 4.21416312e-01 -7.96355903e-01 3.53324711e-02
-1.79627314e-01 7.01224864e-01 -9.60001469e-01 -2.56797433e-01
-7.82250404e-01 -1.13031006e+00 9.17320371e-01 6.71679139e-01
1.01396978e+00 -1.15094030e+00 -9.44569170e-01 4.37407941e-01
6.24098070e-02 -6.89924419e-01 2.70563781e-01 5.22644997e-01
-1.07278502e+00 -8.97428453e-01 -1.01261353e+00 -7.69743741e-01
6.92351937e-01 -6.76675677e-01 8.24682057e-01 1.05603650e-01
-2.37814374e-02 -4.50138986e-01 -2.32318670e-01 -5.18510580e-01
-1.63722262e-01 4.64766711e-01 -1.27762675e-01 -2.56645113e-01
1.22617111e-01 -5.41702807e-01 -6.97360218e-01 3.26094389e-01
-7.58914471e-01 -2.11007163e-01 6.69808626e-01 6.74792945e-01
4.35927287e-02 5.01426160e-01 1.08543241e+00 -4.27915812e-01
6.68487370e-01 -9.75828767e-01 -7.30759323e-01 4.68819365e-02
-6.34365857e-01 -1.29071027e-01 6.86644733e-01 -3.86437178e-01
-8.58894229e-01 -6.56027496e-02 5.62686622e-02 2.44951889e-01
-7.78310299e-02 9.09592867e-01 4.91699129e-01 2.95091599e-01
8.74478400e-01 1.69763565e-01 -5.96108198e-01 -2.99193919e-01
-4.42840636e-01 6.15861356e-01 7.73525834e-01 -1.70351610e-01
4.95308280e-01 -4.60020937e-02 1.44775525e-01 -5.25288343e-01
-9.40761864e-02 9.32986736e-02 -3.58559847e-01 -2.48328418e-01
1.04884219e+00 -7.52934694e-01 -7.99151242e-01 5.04421294e-01
-1.04258502e+00 -1.02178007e-01 2.05148637e-01 1.15932345e+00
-1.13947392e-01 -4.20203298e-01 -7.70868003e-01 -1.11196959e+00
-4.69295532e-01 -7.53549099e-01 -1.44905940e-01 4.19696152e-01
-1.30484298e-01 -1.12222970e+00 8.16967711e-02 4.70290775e-04
4.83234674e-01 6.83681250e-01 7.07596481e-01 -7.96387076e-01
7.63229951e-02 -5.11896431e-01 -1.46629706e-01 4.94188815e-01
-3.08891565e-01 1.43069029e-01 -5.86327493e-01 1.02245502e-01
2.24959195e-01 -3.72847281e-02 7.13500619e-01 7.30702043e-01
7.76752174e-01 -4.25666749e-01 -8.43427181e-02 4.11593169e-01
1.85002398e+00 1.00476480e+00 7.00571775e-01 1.06519008e+00
3.58838707e-01 3.56424242e-01 2.38137677e-01 5.75800300e-01
1.79635845e-02 -1.37535423e-01 2.63013929e-01 -1.76547229e-01
5.55174232e-01 2.08141506e-01 -1.27107590e-01 8.08378637e-01
-6.31420493e-01 -2.97557950e-01 -1.32795441e+00 7.32448220e-01
-1.50723040e+00 -1.07058918e+00 -3.15583557e-01 1.91804123e+00
7.49696374e-01 4.89446282e-01 -1.03422080e-03 6.67140186e-01
7.63014555e-01 -2.75690705e-01 -2.70627618e-01 -6.97630763e-01
-5.47582448e-01 -1.47875145e-01 1.04337728e+00 3.60434562e-01
-6.19115770e-01 4.50605273e-01 5.81030178e+00 5.86293280e-01
-1.37521124e+00 -1.28049538e-01 1.04217398e+00 1.11375302e-01
-4.68703778e-03 -2.17386559e-01 -6.47309780e-01 8.00706148e-01
1.02150583e+00 2.63273716e-04 4.06789601e-01 8.30635607e-01
1.07907079e-01 -4.93279278e-01 -2.66243547e-01 5.71557403e-01
-9.78922844e-02 -1.31799495e+00 -2.51930594e-01 -1.57579824e-01
1.14313626e+00 1.03774399e-01 -5.29307052e-02 1.06935747e-01
3.31113279e-01 -1.12749112e+00 4.37825441e-01 6.72774136e-01
4.32245791e-01 -1.43253493e+00 1.45983541e+00 2.45068312e-01
-9.02015448e-01 -4.45259869e-01 -4.75035757e-01 -4.73151565e-01
-1.96113050e-01 7.54711747e-01 -9.82646406e-01 2.12870583e-01
8.73577774e-01 -2.01938231e-03 -2.70517796e-01 9.29004252e-01
6.02054410e-02 1.03364599e+00 -6.67888641e-01 -4.71270710e-01
5.84781051e-01 -4.75120962e-01 1.86620831e-01 1.04988313e+00
6.49761677e-01 3.19555372e-01 -5.72604656e-01 4.03119862e-01
6.14782982e-02 2.12639183e-01 -4.49061394e-01 3.09044063e-01
1.13402581e+00 8.19956064e-01 -9.72927868e-01 -3.87750715e-01
3.15152824e-01 1.78384587e-01 -1.54259771e-01 5.52010655e-01
-9.69023645e-01 -1.07312977e+00 -3.38122100e-01 -1.16864346e-01
6.36887625e-02 9.44792554e-02 -9.71586823e-01 -4.26279694e-01
-4.38815087e-01 -2.17198476e-01 2.99604207e-01 -9.46486950e-01
-7.35792339e-01 1.01131773e+00 1.26262337e-01 -8.90360951e-01
-4.30264652e-01 -5.78921199e-01 -9.88040507e-01 9.15873051e-01
-6.75533295e-01 -4.95470345e-01 1.82494204e-02 5.87981224e-01
3.21929865e-02 -7.23549426e-01 5.99839091e-01 3.24617565e-01
-1.05483210e+00 4.10477966e-01 4.85143036e-01 5.93977630e-01
2.55236980e-02 -9.18922722e-01 2.58548446e-02 4.36531574e-01
-5.09576678e-01 3.02435458e-01 8.67399633e-01 -5.63157260e-01
-7.10759640e-01 -7.32982993e-01 1.29511368e+00 4.52891216e-02
6.24363661e-01 1.85036764e-01 -7.74565339e-01 5.27592719e-01
2.13063210e-01 -4.16290939e-01 3.73817533e-01 -1.21778883e-01
2.72437483e-01 -9.53484029e-02 -1.32140481e+00 7.08348691e-01
-5.55853397e-02 -3.04797366e-02 -4.58429605e-01 2.35257104e-01
3.25576603e-01 -3.95095199e-02 -1.27811503e+00 5.07201672e-01
5.82205951e-01 -7.99791813e-01 8.71988356e-01 -1.54853404e-01
4.41602856e-01 -1.98112354e-01 4.05684747e-02 -1.04639971e+00
-4.09442246e-01 -5.23233749e-02 5.75815678e-01 9.32732224e-01
9.64845121e-01 -1.15509939e+00 8.56773973e-01 3.83461654e-01
-1.61837973e-02 -1.02696538e+00 -1.07167602e+00 -5.89392662e-01
2.34628245e-01 -1.61481395e-01 7.44856119e-01 1.22732115e+00
-1.54334620e-01 1.84294790e-01 -3.85431707e-01 -3.90698239e-02
2.82756239e-01 -4.04437244e-01 1.04549684e-01 -1.25936055e+00
7.31560439e-02 -4.56369013e-01 -1.03992112e-01 9.00372565e-02
-2.31849506e-01 -4.08832818e-01 -4.33112830e-01 -1.22468555e+00
-9.41166654e-02 -6.84563696e-01 -6.90656006e-01 5.95804751e-01
2.71064728e-01 5.94149113e-01 -1.88611686e-01 1.04578033e-01
4.47499365e-01 6.88081980e-02 8.64407837e-01 8.85240510e-02
-3.84531647e-01 -1.33218408e-01 -1.42794803e-01 9.71034408e-01
1.74232435e+00 -8.52961302e-01 -3.23491484e-01 -4.62417245e-01
6.77136719e-01 4.51442510e-01 8.86377841e-02 -1.40220380e+00
3.63303095e-01 -3.89001846e-01 8.58202934e-01 -6.29381299e-01
9.46844816e-02 -7.29425251e-01 5.98175764e-01 9.95777726e-01
-3.24281067e-01 5.02685010e-01 -2.47726031e-02 -2.70315856e-02
-2.16953531e-01 -7.03712702e-01 3.92407030e-01 1.40052568e-02
-2.65818059e-01 -3.02497178e-01 -8.62272978e-01 -2.13536233e-01
9.15773034e-01 -5.45590520e-01 -1.04062095e-01 -1.46457598e-01
-5.19429982e-01 -1.76902860e-01 2.99117789e-02 -7.97979310e-02
4.70003039e-01 -1.34165561e+00 -1.01779938e+00 6.02732273e-03
-5.29204190e-01 -4.94315356e-01 1.48756161e-01 9.49688137e-01
-1.19595683e+00 3.28384399e-01 -6.26092732e-01 3.05312909e-02
-6.19991481e-01 1.49410948e-01 4.37043995e-01 -9.48166102e-02
-1.29101709e-01 8.83090854e-01 -3.57175827e-01 -8.39877352e-02
-4.32580076e-02 -1.92085057e-01 -7.56897748e-01 1.04336083e-01
1.30537018e-01 9.81035352e-01 -2.99440753e-02 -4.32505190e-01
-1.81558669e-01 6.78793788e-01 4.97844666e-01 -4.12141651e-01
1.40128374e+00 3.38115513e-01 -1.79124340e-01 4.73840892e-01
1.28165996e+00 -2.91421115e-01 -7.05187142e-01 3.98732185e-01
1.04780599e-01 8.59652236e-02 2.67416716e-01 -9.78767335e-01
-1.26929510e+00 3.41543496e-01 6.72648668e-01 4.37571347e-01
1.17947912e+00 -4.63196307e-01 7.45404303e-01 8.96648049e-01
-1.93577874e-02 -1.52056408e+00 -4.73602563e-01 7.41190791e-01
1.04895830e+00 -8.33579302e-01 -1.97973661e-02 3.97557586e-01
-5.50354421e-01 1.19391501e+00 4.66980666e-01 -7.22951472e-01
8.24080765e-01 4.18331444e-01 1.90438583e-01 6.22666106e-02
-6.50308311e-01 5.32504797e-01 -3.94017063e-02 1.29251778e-01
3.76745135e-01 1.27371490e-01 -8.46238196e-01 8.60582650e-01
-7.16059804e-01 -1.57408759e-01 6.18405163e-01 8.84964108e-01
-7.20199645e-01 -2.67831713e-01 -9.61729467e-01 2.99647838e-01
-6.13485634e-01 -1.64130151e-01 -5.88581897e-02 1.16767633e+00
3.05652976e-01 7.60610282e-01 5.41811407e-01 -5.69422007e-01
-3.33682559e-02 -3.56705278e-01 -1.09376021e-01 3.80458623e-01
-8.02510381e-01 -8.98881480e-02 7.84723759e-02 3.06236327e-01
6.18112907e-02 -3.27375889e-01 -1.71980608e+00 -4.55586135e-01
-5.83281338e-01 1.14163446e+00 1.02508235e+00 6.60085678e-01
-2.14175925e-01 4.27534014e-01 9.21027064e-01 -5.69580376e-01
-9.03809145e-02 -1.43405890e+00 -8.09210598e-01 -1.21759847e-02
-1.75303787e-01 -3.17481726e-01 -7.77646244e-01 2.63627544e-02] | [6.367530822753906, 3.0615179538726807] |
caa30eea-e92e-416b-85d9-f5ec594d88b6 | toward-moire-free-and-detail-preserving | 2305.08585 | null | https://arxiv.org/abs/2305.08585v1 | https://arxiv.org/pdf/2305.08585v1.pdf | Toward Moiré-Free and Detail-Preserving Demosaicking | 3D convolutions are commonly employed by demosaicking neural models, in the same way as solving other image restoration problems. Counter-intuitively, we show that 3D convolutions implicitly impede the RGB color spectra from exchanging complementary information, resulting in spectral-inconsistent inference of the local spatial high frequency components. As a consequence, shallow 3D convolution networks suffer the Moir\'e artifacts, but deep 3D convolutions cause over-smoothness. We analyze the fundamental difference between demosaicking and other problems that predict lost pixels between available ones (e.g., super-resolution reconstruction), and present the underlying reasons for the confliction between Moir\'e-free and detail-preserving. From the new perspective, our work decouples the common standard convolution procedure to spectral and spatial feature aggregations, which allow strengthening global communication in the spectral dimension while respecting local contrast in the spatial dimension. We apply our demosaicking model to two tasks: Joint Demosaicking-Denoising and Independently Demosaicking. In both applications, our model substantially alleviates artifacts such as Moir\'e and over-smoothness at similar or lower computational cost to currently top-performing models, as validated by diverse evaluations. Source code will be released along with paper publication. | ['Zitong An', 'Haoyuan Shi', 'Bo Zhao', 'Yan Niu', 'Xuanchen Li'] | 2023-05-15 | null | null | null | null | ['demosaicking'] | ['computer-vision'] | [ 3.17301422e-01 -2.68445939e-01 4.06443834e-01 -1.49996772e-01
-5.53880036e-01 -3.96141350e-01 3.44988316e-01 -2.99685031e-01
-3.51472229e-01 6.91642582e-01 4.78076398e-01 -1.42886981e-01
-3.74210298e-01 -8.98534119e-01 -7.24643409e-01 -1.05214107e+00
-8.58920589e-02 -2.55286932e-01 -1.65180907e-01 -3.15063745e-01
5.10583967e-02 6.73733175e-01 -1.66582263e+00 4.00231659e-01
1.15118957e+00 1.19071174e+00 2.39704847e-01 5.99085569e-01
2.16857437e-02 5.10609925e-01 -2.36135006e-01 -2.08940923e-01
5.87635577e-01 -3.91479015e-01 -5.14829636e-01 2.50420362e-01
7.60402501e-01 -5.80647171e-01 -5.54230928e-01 1.49964142e+00
3.95231307e-01 1.18192337e-01 4.17313725e-01 -7.17332244e-01
-1.11265004e+00 2.64994025e-01 -6.36534750e-01 6.77539334e-02
8.46896414e-03 2.10217491e-01 7.38539755e-01 -1.11539054e+00
3.50236326e-01 1.20900500e+00 1.15702486e+00 2.82160521e-01
-1.62527251e+00 -4.96804446e-01 1.18827835e-01 -2.76973397e-02
-1.44593835e+00 -6.11090899e-01 9.41969991e-01 -2.99463212e-01
7.57170200e-01 4.49422568e-01 7.16848135e-01 9.25359428e-01
2.79503822e-01 2.70099640e-01 1.31058967e+00 -2.82337070e-01
1.57585144e-01 -3.44028592e-01 -1.75790787e-01 2.70711750e-01
3.01869452e-01 2.57806212e-01 -7.62440383e-01 -6.06460273e-02
1.23236763e+00 7.03705177e-02 -6.87923014e-01 -1.98199570e-01
-1.19855905e+00 4.96622473e-01 5.50860226e-01 1.83878139e-01
-5.69723129e-01 3.06879997e-01 -1.17850803e-01 4.55134630e-01
9.36338365e-01 3.30508530e-01 -4.89929497e-01 4.78222519e-01
-9.77867007e-01 4.03423727e-01 2.10698992e-01 6.34777367e-01
9.11197960e-01 2.51560271e-01 1.02736957e-01 7.34091282e-01
4.01378542e-01 5.08003712e-01 -4.55183685e-02 -1.42363584e+00
2.64299899e-01 2.03396723e-01 3.92524719e-01 -9.73543584e-01
-3.94570917e-01 -7.79971242e-01 -1.38731337e+00 7.19767690e-01
3.99332196e-01 -8.19808394e-02 -7.75249302e-01 1.84166908e+00
9.90516469e-02 3.43519121e-01 -4.27854434e-03 1.44743371e+00
6.99872077e-01 6.45137608e-01 -5.33207916e-02 -3.07779402e-01
1.39926922e+00 -3.98799300e-01 -7.52361596e-01 -2.90833861e-01
4.92805708e-03 -8.44462574e-01 9.42399025e-01 4.50351000e-01
-1.31164408e+00 -5.06498456e-01 -1.01090252e+00 -2.56579936e-01
1.00161480e-02 9.91928391e-03 9.78956580e-01 6.55845284e-01
-1.52455211e+00 9.60685015e-01 -7.85303891e-01 -2.57538974e-01
4.56362069e-01 7.98078254e-02 -3.23958516e-01 -1.67047799e-01
-9.84383583e-01 7.07068205e-01 -2.63203103e-02 5.21754920e-01
-5.23844957e-01 -1.17074347e+00 -6.46953762e-01 1.09379329e-01
2.58460641e-02 -8.91315162e-01 8.42957139e-01 -1.24923265e+00
-1.24837112e+00 8.95336330e-01 5.60735054e-02 -2.72310197e-01
4.80522901e-01 -2.86866218e-01 -5.93734741e-01 1.21355377e-01
-6.70090094e-02 3.76284331e-01 1.02543020e+00 -1.56983685e+00
-2.30936125e-01 -5.53056657e-01 -1.27461433e-01 4.05585974e-01
-1.68297455e-01 -1.52662307e-01 -2.84143060e-01 -9.62318361e-01
6.65071011e-01 -3.58037144e-01 -2.80882448e-01 5.23602426e-01
-2.19757274e-01 6.41871572e-01 3.39288384e-01 -8.88049960e-01
7.74076164e-01 -2.29868031e+00 1.30727679e-01 5.99865727e-02
5.44018447e-01 5.03417179e-02 -3.04777950e-01 3.33803564e-01
-3.05163234e-01 4.17432524e-02 -5.45656562e-01 -4.74628270e-01
-2.47180760e-01 1.78437699e-02 -3.53390634e-01 8.22972715e-01
3.12662184e-01 6.62837803e-01 -7.90252149e-01 -8.64901468e-02
3.70657176e-01 9.66227114e-01 -7.03676105e-01 -1.24638043e-01
-4.74725701e-02 7.25623786e-01 -1.42978907e-01 6.45261467e-01
1.45689726e+00 -2.38221124e-01 2.01025456e-01 -6.48580611e-01
-5.46820462e-01 2.18895406e-01 -1.29058087e+00 1.91628349e+00
-4.54177022e-01 6.98244333e-01 7.17153907e-01 -9.15748775e-01
6.14099920e-01 1.51081145e-01 4.59982842e-01 -9.89858329e-01
-2.21105814e-01 2.73926198e-01 -4.14968371e-01 -2.74515003e-01
6.81251407e-01 -5.07974207e-01 3.81152332e-01 2.60086656e-02
-4.29264344e-02 -7.57787079e-02 -4.18956339e-01 -3.44311073e-02
8.69893968e-01 2.96659827e-01 5.25559625e-03 -5.93350589e-01
1.97914511e-01 -7.26060942e-02 4.89397258e-01 8.91463518e-01
-3.72681208e-02 9.87170041e-01 2.61222482e-01 -4.99658257e-01
-1.12651396e+00 -1.26631570e+00 -2.81703770e-01 6.86364651e-01
3.74626726e-01 -1.55193120e-01 -6.57605648e-01 6.26572594e-02
6.34400472e-02 4.12954658e-01 -5.87947190e-01 -7.32230246e-02
-5.56884110e-01 -1.08485544e+00 3.19927037e-01 4.46416661e-02
7.82306671e-01 -5.70829630e-01 -4.30957437e-01 1.89412326e-01
-2.47462556e-01 -1.02413452e+00 -2.76881039e-01 2.91735947e-01
-1.17937028e+00 -8.50616813e-01 -8.47247779e-01 -4.35066879e-01
4.97187167e-01 7.58103848e-01 1.34134972e+00 3.04897994e-01
-9.46015865e-02 2.30420217e-01 -2.09661439e-01 4.47958596e-02
-2.00377420e-01 -5.14216363e-01 3.59577239e-02 1.86529711e-01
-9.89324898e-02 -1.19805598e+00 -9.90508080e-01 1.11287512e-01
-1.09248936e+00 4.27394211e-01 5.51262319e-01 7.52372622e-01
4.69283998e-01 1.17376946e-01 2.58501559e-01 -4.35712188e-01
4.31083322e-01 -3.51196736e-01 -5.97247183e-01 3.34659666e-02
-5.45136452e-01 -4.00974117e-02 7.39253342e-01 -2.45125309e-01
-1.28409100e+00 -5.40149696e-02 -3.05144548e-01 -5.23286879e-01
-1.79173693e-01 3.36441815e-01 -2.20231012e-01 -4.44181412e-01
6.89635158e-01 3.16726357e-01 1.57914571e-02 -9.34255600e-01
2.17363805e-01 2.52105266e-01 7.48190165e-01 -3.00716043e-01
1.00389969e+00 9.17956412e-01 6.08360991e-02 -1.12082052e+00
-6.06308818e-01 -1.68057933e-01 -3.22336882e-01 -2.79491276e-01
8.02218020e-01 -1.25365078e+00 -8.60474825e-01 8.85096014e-01
-1.15882671e+00 -3.33379686e-01 -4.28866237e-01 4.31428462e-01
-2.68766820e-01 6.40492260e-01 -7.99027979e-01 -6.62802756e-01
-7.92902112e-02 -8.92104566e-01 1.08759165e+00 2.09125191e-01
4.02188689e-01 -7.26268589e-01 -1.62499592e-01 1.99433863e-01
6.85432017e-01 8.54544863e-02 9.46228862e-01 4.10269052e-01
-8.77585173e-01 4.04065698e-01 -7.22301543e-01 4.52403069e-01
-9.98958275e-02 -3.41010451e-01 -1.33454561e+00 -1.97097957e-01
3.44350338e-01 1.17100000e-01 1.16989756e+00 8.04701328e-01
1.11070585e+00 -2.36839905e-01 2.73481220e-01 1.23918390e+00
1.72561932e+00 -3.71220917e-01 9.40499783e-01 2.20930651e-01
6.45422816e-01 4.96961385e-01 4.84477282e-02 5.34144580e-01
9.29207876e-02 7.22533703e-01 8.14009428e-01 -5.89049041e-01
-5.87059498e-01 -7.20165595e-02 1.97257295e-01 5.59881568e-01
-3.65569085e-01 -7.21127465e-02 -5.42836905e-01 3.89305055e-01
-1.66361010e+00 -8.30007613e-01 -4.56581295e-01 2.17972136e+00
7.94577360e-01 -3.09121341e-01 -2.35243216e-01 6.00703768e-02
6.32157326e-01 4.93035197e-01 -4.43183899e-01 9.83263329e-02
-6.42050028e-01 3.41628969e-01 9.19962108e-01 7.13827848e-01
-9.71909881e-01 6.50893211e-01 6.04042482e+00 6.99731171e-01
-1.05216420e+00 2.33499497e-01 6.35206342e-01 -1.80656120e-01
-7.26321757e-01 -1.74618866e-02 -7.27121830e-02 3.62361550e-01
4.32427257e-01 3.52781683e-01 8.24744761e-01 1.81410164e-01
5.60470819e-01 -1.89213797e-01 -7.28315711e-01 1.27385986e+00
-1.50396556e-01 -1.61514962e+00 1.41451791e-01 1.36473939e-01
6.06104255e-01 3.19324769e-02 1.18062884e-01 -3.17611367e-01
2.37396076e-01 -1.10321188e+00 1.07766223e+00 5.99406421e-01
9.84069407e-01 -5.66780448e-01 4.29681569e-01 2.81699486e-02
-1.11059368e+00 8.50072429e-02 -3.49518746e-01 -3.32468659e-01
2.44193181e-01 1.24479508e+00 2.27083147e-01 8.23252797e-01
1.00843751e+00 6.67643607e-01 -2.31376678e-01 1.00981379e+00
-3.18621472e-02 2.47488782e-01 -1.36971995e-01 6.78797424e-01
4.88631129e-02 -7.86446512e-01 7.28269637e-01 9.82923746e-01
6.03235126e-01 2.75634766e-01 -2.21413389e-01 1.36326492e+00
-3.37786488e-02 -3.95364016e-01 -3.34487826e-01 4.07203734e-01
2.82936573e-01 9.07651126e-01 -5.44970274e-01 4.03622985e-02
-5.26604712e-01 1.35706317e+00 -8.42244774e-02 9.12438750e-01
-5.79483509e-01 -6.32231608e-02 1.35674298e+00 2.36871019e-01
2.48493925e-01 -3.92922729e-01 -6.45563662e-01 -1.38173151e+00
2.56076217e-01 -7.58027494e-01 -1.86098889e-01 -1.04378676e+00
-1.44015992e+00 2.36374453e-01 -4.17950869e-01 -1.23905301e+00
5.00433564e-01 -7.80894339e-01 -3.74426872e-01 1.23963463e+00
-1.92926192e+00 -1.06902277e+00 -4.09576595e-01 7.55596519e-01
1.43237129e-01 5.35458982e-01 6.71404362e-01 6.96978092e-01
-3.25536877e-01 6.71938360e-02 2.25752056e-01 -1.11206762e-01
5.93997777e-01 -1.02427959e+00 5.33285141e-01 1.15736008e+00
-1.69147000e-01 5.62404215e-01 9.11758542e-01 -5.91471314e-01
-1.68812561e+00 -9.04379070e-01 6.25514567e-01 -3.30432877e-02
4.36908901e-01 -3.61843348e-01 -1.00148034e+00 4.13757086e-01
6.55493066e-02 2.18724132e-01 2.51961619e-01 -8.56825709e-02
-3.93697888e-01 -1.49026722e-01 -1.20513558e+00 6.14573896e-01
1.41812778e+00 -7.14058876e-01 -4.00704332e-02 1.24541558e-01
7.58540869e-01 -3.37177038e-01 -8.47280502e-01 2.40035936e-01
4.81181294e-01 -1.59833920e+00 1.52569389e+00 -3.20113569e-01
6.42466605e-01 -4.83325779e-01 -5.21272540e-01 -1.22988832e+00
-5.17129481e-01 -6.52434111e-01 7.04766214e-02 9.72312689e-01
1.31894976e-01 -7.19854295e-01 6.02114260e-01 2.20245600e-01
-3.99127692e-01 -2.19707817e-01 -1.14963174e+00 -6.11318529e-01
9.50958058e-02 -5.70604622e-01 4.61283237e-01 1.25816107e+00
-6.09779000e-01 -1.62228227e-01 -5.98734796e-01 7.75080502e-01
1.09683764e+00 4.68139760e-02 1.58601239e-01 -1.33129489e+00
-2.74518102e-01 -4.64979291e-01 -5.53354174e-02 -1.28444350e+00
-1.55860186e-01 -7.20520914e-01 3.85272130e-02 -1.39390206e+00
6.03266954e-02 -5.41719019e-01 -1.57731563e-01 1.05442464e-01
-5.61081655e-02 5.76961160e-01 1.77980542e-01 3.29178244e-01
-1.28916860e-01 5.31129956e-01 1.34504437e+00 1.80853065e-02
-1.08311065e-01 -3.26923549e-01 -9.28486705e-01 6.87605619e-01
5.62009990e-01 -1.92083821e-01 -1.23787396e-01 -9.53518212e-01
6.27106786e-01 2.18424290e-01 1.04644585e+00 -9.76311803e-01
-1.42483618e-02 -1.54622376e-01 7.04107285e-01 -1.77409008e-01
4.28252637e-01 -9.39324915e-01 4.91859823e-01 2.12982237e-01
2.18248367e-02 -2.68809170e-01 1.83098555e-01 4.68133658e-01
-1.23993978e-01 1.80220619e-01 9.66129005e-01 -3.72195303e-01
-7.69890964e-01 2.04165041e-01 -4.43071097e-01 -2.01815411e-01
2.78206378e-01 -3.77702653e-01 -4.57623124e-01 -3.82679641e-01
-6.16428435e-01 -2.57189751e-01 9.45298314e-01 -8.93591642e-02
6.90105975e-01 -1.23079658e+00 -8.09610307e-01 4.49107766e-01
-2.59486198e-01 -1.61448438e-02 8.42736483e-01 1.03999400e+00
-6.59919620e-01 -2.15306148e-01 -2.94649273e-01 -5.01305223e-01
-8.34526420e-01 2.86243081e-01 7.10464418e-01 9.20431018e-02
-1.04306901e+00 8.57857287e-01 3.45579773e-01 -3.69856775e-01
1.33208558e-01 -4.39737141e-01 3.13721687e-01 -1.10605039e-01
5.53222120e-01 2.94441432e-01 3.55039597e-01 -3.33972901e-01
-3.89074117e-01 6.79649472e-01 3.36175084e-01 -7.49720261e-02
1.64714265e+00 -5.25590003e-01 -3.76085013e-01 7.79618919e-02
1.02547562e+00 2.11857304e-01 -1.70369494e+00 -2.13785663e-01
-5.19086599e-01 -7.01074302e-01 5.48427522e-01 -7.91585505e-01
-1.39820480e+00 7.50314236e-01 7.56088614e-01 2.73405045e-01
1.58436811e+00 -1.55767843e-01 6.29466116e-01 4.50588390e-02
1.76594168e-01 -8.23995054e-01 -2.71209449e-01 3.07833165e-01
9.02683079e-01 -1.00483429e+00 1.92447320e-01 -5.91639340e-01
-1.24409541e-01 8.62290800e-01 1.56346574e-01 -1.68570310e-01
5.55728495e-01 2.40351930e-01 -2.86316723e-02 -2.68379569e-01
-3.00186336e-01 -1.64637327e-01 3.95098403e-02 7.76697218e-01
3.53745282e-01 -8.93128105e-03 3.16943019e-03 4.19141263e-01
1.62619054e-01 -1.18361157e-03 5.07894695e-01 5.86386979e-01
-2.77349591e-01 -7.10129857e-01 -7.27733195e-01 1.29627138e-01
-3.24386179e-01 -6.11857891e-01 -1.70650810e-01 6.25725508e-01
3.24722379e-01 8.68531048e-01 2.95038372e-01 -2.18165293e-01
2.87044615e-01 -1.93246916e-01 5.66969514e-01 -9.32326168e-02
-5.49860418e-01 2.79657394e-01 7.41840228e-02 -7.26163030e-01
-6.77180231e-01 -4.41451222e-01 -8.44652951e-01 -7.65037417e-01
7.32327485e-03 -2.97501117e-01 4.99667585e-01 5.61134696e-01
4.56707239e-01 5.86614609e-01 4.16426837e-01 -1.35866046e+00
-4.11327660e-01 -6.14208639e-01 -9.71084952e-01 3.87577683e-01
8.60442758e-01 -3.87875885e-01 -6.63932860e-01 -6.69792891e-02] | [11.161294937133789, -2.361258029937744] |
a2a0cbe7-af6d-4b06-8f80-8d54599afba1 | contextual-dictionary-lookup-for-knowledge | 2306.07719 | null | https://arxiv.org/abs/2306.07719v1 | https://arxiv.org/pdf/2306.07719v1.pdf | Contextual Dictionary Lookup for Knowledge Graph Completion | Knowledge graph completion (KGC) aims to solve the incompleteness of knowledge graphs (KGs) by predicting missing links from known triples, numbers of knowledge graph embedding (KGE) models have been proposed to perform KGC by learning embeddings. Nevertheless, most existing embedding models map each relation into a unique vector, overlooking the specific fine-grained semantics of them under different entities. Additionally, the few available fine-grained semantic models rely on clustering algorithms, resulting in limited performance and applicability due to the cumbersome two-stage training process. In this paper, we present a novel method utilizing contextual dictionary lookup, enabling conventional embedding models to learn fine-grained semantics of relations in an end-to-end manner. More specifically, we represent each relation using a dictionary that contains multiple latent semantics. The composition of a given entity and the dictionary's central semantics serves as the context for generating a lookup, thus determining the fine-grained semantics of the relation adaptively. The proposed loss function optimizes both the central and fine-grained semantics simultaneously to ensure their semantic consistency. Besides, we introduce two metrics to assess the validity and accuracy of the dictionary lookup operation. We extend several KGE models with the method, resulting in substantial performance improvements on widely-used benchmark datasets. | ['Yuren Zhou', 'Zibin Zheng', 'Chuan Chen', 'Yining Wang', 'YouMing Liu', 'Delai Qiu', 'Jining Wang'] | 2023-06-13 | null | null | null | null | ['graph-embedding', 'knowledge-graph-embedding', 'knowledge-graph-completion', 'knowledge-graphs'] | ['graphs', 'graphs', 'knowledge-base', 'knowledge-base'] | [-1.89425141e-01 1.97806090e-01 -5.48173130e-01 -3.09891582e-01
-2.46470317e-01 -4.07377601e-01 5.69864810e-01 6.48310542e-01
-1.86548293e-01 6.69718444e-01 4.18156624e-01 -3.25815938e-02
-4.56927508e-01 -1.25020099e+00 -7.24760890e-01 -6.05387926e-01
4.43452820e-02 7.14921594e-01 1.39283732e-01 -2.35016912e-01
-1.50056943e-01 1.10887781e-01 -1.49916589e+00 1.12712577e-01
1.06250274e+00 9.57533479e-01 2.19560981e-01 4.12822440e-02
-4.75488096e-01 6.62561059e-01 -2.52013385e-01 -8.15198004e-01
-3.06378747e-03 -1.99258119e-01 -7.52206564e-01 -2.53044128e-01
2.56567866e-01 8.99210721e-02 -6.84361100e-01 1.21077108e+00
1.76277369e-01 1.19328566e-01 6.18920445e-01 -1.38203382e+00
-1.10870016e+00 8.49976122e-01 -1.82479620e-01 -1.27032802e-01
3.04781824e-01 -3.71128440e-01 1.51457512e+00 -1.01608920e+00
6.86569870e-01 1.04456675e+00 6.29212379e-01 2.01848403e-01
-1.15083873e+00 -4.47189212e-01 2.19388634e-01 6.21627510e-01
-1.72084999e+00 -1.36323273e-01 9.28411067e-01 -3.00967127e-01
9.06294882e-01 -5.63931055e-02 7.31326163e-01 8.09258819e-01
-1.62071049e-01 3.83094609e-01 7.94061005e-01 -3.98489088e-01
3.39160830e-01 2.82185495e-01 2.08269507e-01 8.51225555e-01
7.13745296e-01 -2.37012401e-01 -6.75798535e-01 -2.20596626e-01
5.40113807e-01 4.76409383e-02 -2.29582205e-01 -7.72256851e-01
-1.08276224e+00 8.11033547e-01 5.68246305e-01 2.55485028e-01
-3.50315154e-01 1.93692550e-01 5.07281303e-01 2.63217598e-01
1.15477704e-01 8.10734928e-02 -4.80675817e-01 5.59558198e-02
-5.15487254e-01 1.93103105e-01 9.10557926e-01 1.21312964e+00
1.14681172e+00 -2.08160728e-01 -4.14619892e-04 7.25839078e-01
4.93888825e-01 2.07292169e-01 4.24635738e-01 -4.26529378e-01
6.31081104e-01 1.13398302e+00 -7.02998741e-03 -1.65219271e+00
-2.27535754e-01 -4.69687760e-01 -8.67406607e-01 -5.26219010e-01
9.20283720e-02 3.47023547e-01 -5.57147264e-01 1.72061646e+00
7.19490707e-01 3.38977426e-01 1.83034882e-01 8.94113958e-01
8.40032279e-01 4.59447175e-01 3.75995010e-01 -1.43906223e-02
1.41312873e+00 -8.45452547e-01 -8.37047756e-01 -1.51691148e-02
7.52833366e-01 -4.84674871e-01 1.04520917e+00 -2.26860624e-02
-5.60844600e-01 -4.54416156e-01 -1.20403826e+00 -3.12680513e-01
-7.63440013e-01 9.46588665e-02 9.60810602e-01 4.79372323e-01
-7.97661364e-01 5.19178391e-01 -6.55900836e-01 -3.52765739e-01
6.80569038e-02 1.49874210e-01 -5.68134010e-01 -2.86649585e-01
-1.62190711e+00 7.57269084e-01 9.81386602e-01 7.18174726e-02
-6.29317999e-01 -7.48139083e-01 -1.12005925e+00 4.00900185e-01
7.69013405e-01 -8.11266541e-01 2.99560785e-01 -1.79742381e-01
-9.81009960e-01 6.34804606e-01 -5.99640869e-02 -3.68289500e-01
-2.99698859e-03 5.95137291e-02 -6.45393550e-01 6.87562972e-02
3.68412793e-01 2.31380716e-01 6.11188829e-01 -1.39834976e+00
-6.02810681e-01 -5.67687988e-01 2.88639128e-01 2.72068828e-01
-6.61083758e-01 -5.40588737e-01 -7.83077180e-01 -8.40362191e-01
2.76410341e-01 -5.93677104e-01 1.76668674e-01 -2.23166317e-01
-3.75639439e-01 -3.81841928e-01 6.55094981e-01 -5.90250432e-01
1.61563909e+00 -2.04764676e+00 3.74780923e-01 3.67736757e-01
4.58875239e-01 5.67346551e-02 3.11916880e-02 6.91412508e-01
9.72670317e-02 -1.19849574e-03 -1.31587163e-01 -2.20743611e-01
3.79560411e-01 5.45929611e-01 -2.75466561e-01 2.54945278e-01
4.21995521e-02 1.07605505e+00 -1.16920269e+00 -6.19978487e-01
1.41977355e-01 4.50981379e-01 -4.98998731e-01 2.24145204e-02
-3.07426125e-01 -1.10756002e-01 -5.46998978e-01 8.22618783e-01
5.36575019e-01 -5.25456667e-01 7.20519006e-01 -7.36364245e-01
4.59125578e-01 5.65252751e-02 -1.45194161e+00 1.83376670e+00
-4.78064001e-01 -9.82475132e-02 -1.46366820e-01 -1.22820652e+00
1.00548303e+00 1.49902090e-01 4.62855190e-01 -4.62548643e-01
-8.72114822e-02 3.22396934e-01 -4.65093076e-01 -2.73808002e-01
7.27423131e-01 -1.46553531e-01 -2.43150264e-01 2.81240880e-01
3.48843306e-01 3.03484380e-01 1.96222976e-01 6.20657027e-01
1.02700770e+00 2.33673379e-02 3.38286191e-01 -7.91107416e-02
5.22193551e-01 8.50262120e-02 6.95816576e-01 4.31816190e-01
-2.39134412e-02 1.83225214e-01 3.61192673e-01 -4.27418172e-01
-8.25288177e-01 -1.07451546e+00 8.08447078e-02 8.94408345e-01
6.47809267e-01 -1.01619840e+00 -3.63500565e-01 -8.85432363e-01
4.27054167e-01 5.74636579e-01 -6.32098675e-01 -4.99711305e-01
-2.16360778e-01 -5.91661572e-01 5.03966033e-01 5.11859894e-01
3.78906131e-01 -7.90222228e-01 1.95944440e-02 3.23912203e-01
-3.78584355e-01 -1.32894480e+00 -4.28713292e-01 1.41995670e-02
-6.71901822e-01 -1.42059469e+00 -1.38689250e-01 -9.05056417e-01
7.01667786e-01 2.17461392e-01 1.08615375e+00 1.56137660e-01
-1.20434999e-01 3.71040642e-01 -6.25192881e-01 3.58975500e-01
-9.81365591e-02 8.46480578e-02 1.63424075e-01 1.80997610e-01
5.65002143e-01 -7.29478121e-01 -4.67208117e-01 1.45150244e-01
-1.04923677e+00 -6.22162111e-02 5.15078187e-01 9.38892305e-01
8.29453588e-01 6.46664202e-01 6.48955226e-01 -1.11642611e+00
6.75972402e-01 -7.24150360e-01 -4.90798652e-01 7.00146079e-01
-1.13431954e+00 3.33472311e-01 7.71430671e-01 -1.35262802e-01
-9.49697614e-01 -2.37021327e-01 2.22435474e-01 -6.26989782e-01
1.66008979e-01 1.02871275e+00 -4.50812638e-01 1.26335341e-02
3.54234517e-01 4.68449920e-01 -2.34433368e-01 -5.07352114e-01
7.98350751e-01 3.44115347e-01 5.26692688e-01 -1.08065426e+00
9.11533952e-01 2.77615219e-01 4.81152460e-02 -3.61006379e-01
-8.15788448e-01 -5.80734909e-01 -4.82132494e-01 6.00445494e-02
7.00957119e-01 -1.03732312e+00 -6.87977016e-01 9.27665159e-02
-9.50673163e-01 2.29799449e-01 -2.89800018e-01 3.92685503e-01
-3.22241247e-01 5.35571039e-01 -4.88966107e-01 -2.82408416e-01
-1.62938386e-01 -8.94920766e-01 9.97427225e-01 7.07759485e-02
4.94178534e-02 -1.23878503e+00 6.60603568e-02 4.99845088e-01
1.41099587e-01 2.00976521e-01 1.54795992e+00 -5.41764855e-01
-7.30689168e-01 -2.11841345e-01 -4.37685728e-01 1.20648108e-01
2.11103871e-01 -3.95186275e-01 -5.07910669e-01 -2.66533464e-01
-4.59099799e-01 -3.29296350e-01 6.54932737e-01 -1.57083422e-01
9.22572374e-01 -3.84032160e-01 -5.19015253e-01 5.31838775e-01
1.72116327e+00 -7.57121518e-02 4.27953541e-01 3.59392107e-01
1.20003331e+00 4.32811707e-01 6.53566003e-01 3.40039164e-01
9.86280084e-01 6.84076250e-01 2.94055790e-01 4.45228130e-01
-2.13464290e-01 -9.51866627e-01 -3.68502960e-02 1.12544537e+00
7.98078775e-02 1.60642620e-02 -7.23567188e-01 7.64412642e-01
-2.05587840e+00 -7.34546304e-01 1.35996357e-01 2.08225369e+00
1.03170562e+00 -5.49674220e-03 -2.78258801e-01 1.06572814e-01
7.45339453e-01 2.05594957e-01 -4.14667457e-01 1.24155484e-01
-1.21788777e-01 1.17962562e-01 4.46265578e-01 4.47965860e-01
-8.40853930e-01 1.15131140e+00 5.02683401e+00 1.07462656e+00
-6.99548304e-01 1.87861890e-01 -6.81406409e-02 3.15798908e-01
-8.87789905e-01 4.68400091e-01 -8.66778553e-01 4.65799749e-01
5.35554349e-01 -2.72044122e-01 5.50577283e-01 7.44339585e-01
-2.47910991e-01 1.98250890e-01 -1.03875387e+00 9.88301098e-01
-1.11009754e-01 -1.46375453e+00 4.13380861e-01 -4.86827940e-02
7.78638363e-01 -4.67060804e-01 -3.33270013e-01 6.32916212e-01
2.63791025e-01 -7.18776047e-01 5.80508649e-01 4.65328783e-01
9.23623204e-01 -6.88554168e-01 6.19620085e-01 3.80276731e-04
-1.63849103e+00 5.65422624e-02 -4.32199329e-01 1.66299000e-01
9.82809812e-03 7.43805528e-01 -7.40384400e-01 1.20642507e+00
5.90398371e-01 7.69578815e-01 -4.30590898e-01 4.26858753e-01
-4.09224659e-01 3.76321584e-01 -1.42482251e-01 1.48837343e-01
1.44052684e-01 -3.53863657e-01 3.08040977e-01 8.60597312e-01
1.52901143e-01 8.78059044e-02 3.60827476e-01 8.12922597e-01
-4.82254207e-01 2.53999561e-01 -3.99716198e-01 -3.92405689e-01
1.07448912e+00 1.16262329e+00 -5.80020428e-01 -3.30741942e-01
-6.20539010e-01 8.70633483e-01 7.85170197e-01 4.32257533e-01
-7.32999504e-01 -5.29359579e-01 6.42071068e-01 4.60216887e-02
3.79497856e-01 -9.65387002e-02 -2.11008027e-01 -1.38121665e+00
3.25637579e-01 -6.68549418e-01 7.30833650e-01 -4.37106401e-01
-1.33975387e+00 1.97548091e-01 6.32219091e-02 -8.30280244e-01
9.87419561e-02 -3.52286100e-01 -2.29116917e-01 7.87305713e-01
-1.65220177e+00 -1.38549554e+00 -4.21315879e-01 7.27428854e-01
1.01699665e-01 -2.46042460e-01 9.61872578e-01 5.85775077e-01
-4.82079744e-01 7.45310068e-01 1.32753327e-01 1.33645967e-01
5.80687404e-01 -1.33428490e+00 7.29777217e-02 6.11474633e-01
2.32337862e-01 9.13525343e-01 5.00153542e-01 -9.41538334e-01
-1.65547943e+00 -1.25104225e+00 1.25025761e+00 -1.75366163e-01
9.01126266e-01 -1.78687036e-01 -9.97590184e-01 7.50960410e-01
-3.06984693e-01 2.88666546e-01 6.91363215e-01 6.49224937e-01
-7.56492615e-01 -3.97687435e-01 -1.00952685e+00 4.39129084e-01
1.13101578e+00 -8.26280892e-01 -6.33343339e-01 8.69524404e-02
9.30665374e-01 -1.80435106e-01 -1.32164478e+00 4.72618282e-01
4.92794544e-01 -5.79173863e-01 9.42163527e-01 -6.39942706e-01
3.57231766e-01 -6.51056588e-01 -4.54620093e-01 -1.16489112e+00
-4.85553652e-01 -8.05685893e-02 -7.07928240e-01 1.28991890e+00
2.29524970e-01 -6.05238318e-01 9.27104890e-01 4.26724404e-01
-4.00544442e-02 -8.92044365e-01 -8.11172426e-01 -6.39161587e-01
-4.26395148e-01 -1.83768764e-01 9.78405595e-01 1.38097477e+00
1.53094858e-01 4.83186215e-01 -3.59126091e-01 4.75445658e-01
7.85174668e-01 4.44940090e-01 5.60228109e-01 -1.26381242e+00
-3.19783151e-01 -2.26916954e-01 -8.02226543e-01 -7.39073634e-01
3.50994945e-01 -1.33422160e+00 -3.96788120e-01 -1.65999603e+00
2.42518798e-01 -8.75045180e-01 -4.51255441e-01 6.20937884e-01
-3.66032183e-01 -5.74015491e-02 1.78086460e-02 3.66928577e-01
-6.76591754e-01 9.52346504e-01 1.01719189e+00 -4.72060353e-01
6.42960295e-02 -5.73933125e-01 -9.41392899e-01 2.50774205e-01
5.05006850e-01 -4.82440561e-01 -9.15667713e-01 -4.93990481e-01
5.44237256e-01 -3.48671600e-02 4.68484044e-01 -7.02160716e-01
5.42679608e-01 -9.30451453e-02 1.09147608e-01 -3.38231057e-01
2.01288059e-01 -1.05426502e+00 5.00128448e-01 6.37366176e-02
-1.10204719e-01 -2.56561130e-01 -1.45525202e-01 1.17565453e+00
-5.53846180e-01 -1.62575692e-01 3.05075556e-01 -1.33817941e-02
-1.12547469e+00 4.93796349e-01 4.86227751e-01 1.64435431e-01
9.83131647e-01 -1.00474119e-01 -1.28180876e-01 -1.50250681e-02
-9.43300128e-01 4.05168384e-01 5.07008612e-01 5.39522648e-01
6.67250395e-01 -1.81644988e+00 -3.28271508e-01 1.85191885e-01
5.61682284e-01 1.74848706e-01 2.87420422e-01 5.94261229e-01
-3.15153718e-01 3.91785204e-01 7.72262961e-02 -1.80563942e-01
-8.15901458e-01 7.57415652e-01 7.71874096e-03 -4.71062809e-01
-5.64779401e-01 5.68268657e-01 -3.03897932e-02 -5.08053839e-01
2.71743417e-01 -8.46620947e-02 -1.34350613e-01 1.40452132e-01
1.51211530e-01 3.71494591e-01 2.61999100e-01 -6.41120374e-01
-4.41652715e-01 4.55419153e-01 -1.92457885e-01 3.54533821e-01
1.14210725e+00 -2.04556152e-01 -2.80672610e-01 2.14792907e-01
1.18841195e+00 -7.59411156e-02 -7.37045228e-01 -6.71131313e-01
1.13178164e-01 -5.54486871e-01 8.53622481e-02 -6.55897617e-01
-1.02062285e+00 3.51835996e-01 1.46107540e-01 2.78319465e-03
9.50982094e-01 1.47343159e-01 8.65576267e-01 3.78411949e-01
6.81197226e-01 -1.15085816e+00 -1.55661091e-01 3.20669770e-01
3.95466805e-01 -1.08477890e+00 1.34566203e-02 -7.07903743e-01
-6.34629488e-01 8.75851929e-01 5.22494376e-01 2.27069750e-01
8.14184487e-01 -2.21441269e-01 -2.76243597e-01 -6.49817884e-01
-4.87448007e-01 -2.33840063e-01 2.91597992e-01 5.87376714e-01
7.60551542e-02 3.56371284e-01 -4.40103441e-01 9.69131052e-01
-7.01768249e-02 -2.15451196e-02 4.51317355e-02 7.60740995e-01
-2.98249602e-01 -1.25035334e+00 3.75500023e-02 4.22361165e-01
9.65857971e-03 -7.20389485e-02 -1.42963290e-01 6.98370516e-01
2.86343455e-01 8.04640353e-01 -3.38352799e-01 -5.76285481e-01
4.54367638e-01 1.06034182e-01 3.51914436e-01 -6.15650475e-01
7.53514022e-02 -3.87603849e-01 7.33884946e-02 -4.68455911e-01
-2.44666681e-01 -2.62629360e-01 -1.43706667e+00 -4.66970086e-01
-3.52874279e-01 4.89821851e-01 2.90976018e-01 8.26402664e-01
5.83159685e-01 3.56255710e-01 4.82576698e-01 -1.84565455e-01
-6.05389714e-01 -6.00397289e-01 -9.64178741e-01 8.42304528e-01
-1.19113259e-01 -1.08472002e+00 -5.14728352e-02 -1.04796337e-02] | [8.771209716796875, 7.8449296951293945] |
3a087759-77a2-4bf4-bfd0-097d496c3cad | reasoning-about-goals-steps-and-temporal | 2009.07690 | null | https://arxiv.org/abs/2009.07690v2 | https://arxiv.org/pdf/2009.07690v2.pdf | Reasoning about Goals, Steps, and Temporal Ordering with WikiHow | We propose a suite of reasoning tasks on two types of relations between procedural events: goal-step relations ("learn poses" is a step in the larger goal of "doing yoga") and step-step temporal relations ("buy a yoga mat" typically precedes "learn poses"). We introduce a dataset targeting these two relations based on wikiHow, a website of instructional how-to articles. Our human-validated test set serves as a reliable benchmark for commonsense inference, with a gap of about 10% to 20% between the performance of state-of-the-art transformer models and human performance. Our automatically-generated training set allows models to effectively transfer to out-of-domain tasks requiring knowledge of procedural events, with greatly improved performances on SWAG, Snips, and the Story Cloze Test in zero- and few-shot settings. | ['Chris Callison-Burch', 'Li Zhang', 'Qing Lyu'] | 2020-09-16 | null | https://aclanthology.org/2020.emnlp-main.374 | https://aclanthology.org/2020.emnlp-main.374.pdf | emnlp-2020-11 | ['cloze-test'] | ['natural-language-processing'] | [ 2.70272762e-01 4.25748944e-01 -2.31473714e-01 -2.31385469e-01
-9.34760034e-01 -6.22878015e-01 9.51715946e-01 4.69002694e-01
-2.12787867e-01 7.84200907e-01 4.87998277e-01 -4.74129707e-01
-2.02843100e-01 -1.07998407e+00 -9.72406328e-01 -2.52550840e-01
2.55613446e-01 7.16624856e-01 5.32480836e-01 -7.87987411e-01
1.73904955e-01 -7.35337660e-02 -1.68270910e+00 5.70647180e-01
9.74215865e-01 5.87649226e-01 1.48358747e-01 5.24452448e-01
-2.76399031e-02 1.83561373e+00 -5.34121752e-01 -8.30185175e-01
-2.88579792e-01 -7.73303390e-01 -1.33975852e+00 -4.88339096e-01
7.27681994e-01 -2.75756210e-01 -3.55170548e-01 1.01836562e+00
2.07915217e-01 5.35351396e-01 6.24140084e-01 -9.84549463e-01
-8.13328743e-01 1.38183606e+00 1.61864571e-02 6.81584954e-01
9.12110209e-01 3.17235261e-01 1.39369273e+00 -4.57841486e-01
1.00844383e+00 9.69219267e-01 7.12448180e-01 3.10909420e-01
-1.27634823e+00 -6.49347425e-01 -2.74136774e-02 7.59930253e-01
-9.64775085e-01 -5.08807778e-01 7.11276472e-01 -5.41143775e-01
1.44534087e+00 1.96063787e-01 8.85383189e-01 1.42856634e+00
1.20852090e-01 1.01880240e+00 1.16129220e+00 -5.38883686e-01
4.00579385e-02 -2.23606408e-01 3.24824363e-01 1.02569020e+00
-3.31739411e-02 -6.14415817e-02 -8.88211131e-01 3.36479485e-01
6.23305678e-01 -4.86668825e-01 -2.03690544e-01 -9.05391872e-02
-1.18616593e+00 8.44642401e-01 7.42602646e-02 2.85066605e-01
-1.51079565e-01 3.64877105e-01 5.05079269e-01 4.06885237e-01
3.29588860e-01 8.58669281e-01 -1.54059142e-01 -8.16318572e-01
-8.44714701e-01 6.81588590e-01 9.89002585e-01 1.06464374e+00
2.12727338e-01 3.73734534e-02 -2.89243430e-01 6.69178545e-01
-2.35511214e-01 1.91244185e-01 4.53380316e-01 -1.15550721e+00
7.34392464e-01 3.79144341e-01 1.81899115e-01 -7.98778474e-01
-3.29117447e-01 -3.40024650e-01 -9.36045349e-02 -1.27166152e-01
8.21706831e-01 1.69247434e-01 -6.12950206e-01 1.86511707e+00
1.66171491e-01 5.12215674e-01 -7.51606300e-02 5.76493979e-01
1.31989944e+00 5.06364763e-01 2.84697354e-01 -2.67573982e-01
1.56020153e+00 -9.25617635e-01 -8.61080170e-01 -4.43405420e-01
8.70391011e-01 -3.82967710e-01 1.75680411e+00 4.81148899e-01
-1.59954035e+00 -3.46622676e-01 -9.13765013e-01 -5.70296168e-01
-4.62717593e-01 -1.62167534e-01 8.44801664e-01 1.16294101e-01
-6.38007462e-01 6.76593304e-01 -7.62613237e-01 -3.87459219e-01
4.19424683e-01 -3.60873073e-01 -2.55044550e-01 -2.05548018e-01
-1.45239413e+00 1.47500062e+00 4.36945170e-01 -5.07578552e-01
-1.31544971e+00 -1.34677291e+00 -1.18071914e+00 3.37763220e-01
8.23622823e-01 -6.53303206e-01 1.65944457e+00 -4.01636720e-01
-1.64688587e+00 1.57863092e+00 7.22636059e-02 -5.75445056e-01
5.15484035e-01 -6.93831861e-01 -3.57316375e-01 1.11469798e-01
3.06659937e-01 1.18285134e-01 3.45203727e-01 -6.47671878e-01
-4.91327792e-01 -2.21213438e-02 5.93893349e-01 3.29526812e-01
1.19148552e-01 2.58151680e-01 1.29866853e-01 -4.49814945e-01
-2.49101724e-02 -6.77241743e-01 3.64678204e-01 -6.68593824e-01
-4.52159375e-01 -6.49902701e-01 2.43911952e-01 -6.44996226e-01
1.41915047e+00 -1.83461285e+00 3.16558212e-01 -4.38096613e-01
1.16487570e-01 -2.65671790e-01 9.74309146e-02 4.74741191e-01
-3.61588031e-01 -1.99784026e-01 1.23319872e-01 -6.63668811e-02
3.02548975e-01 2.44631261e-01 -6.45014167e-01 1.72811151e-01
-4.96309400e-02 9.89961088e-01 -1.52310133e+00 -5.61359346e-01
1.34704530e-01 -2.45943531e-01 -6.43051505e-01 3.57419342e-01
-6.18866980e-01 -1.09768488e-01 -3.16462032e-02 3.27395171e-01
-1.87898293e-01 -3.72211635e-01 9.92697403e-02 -1.46919593e-01
3.66629325e-02 1.44590080e+00 -7.25345194e-01 1.93883073e+00
-4.99480426e-01 8.29918444e-01 -5.45203507e-01 -7.35172689e-01
3.19194168e-01 4.25209463e-01 -6.15317449e-02 -6.34358883e-01
2.46023163e-01 -3.33747655e-01 1.71359599e-01 -7.76673555e-01
5.47581196e-01 -4.32694852e-01 -3.01029533e-01 4.97235864e-01
5.61551332e-01 -8.27114880e-01 8.38310063e-01 4.33854520e-01
1.37500000e+00 6.34720504e-01 6.54627681e-01 -4.08008844e-01
6.65495321e-02 3.57767731e-01 2.92832196e-01 8.16169202e-01
-3.06560920e-04 2.63069451e-01 8.80911827e-01 -3.90961617e-01
-7.77064264e-01 -1.25513661e+00 -9.42218229e-02 1.62596583e+00
2.45364383e-03 -8.21675539e-01 -7.29250193e-01 -5.61986983e-01
-4.25702214e-01 1.79706657e+00 -9.34804976e-01 -2.54367232e-01
-6.08084857e-01 -2.04676092e-01 7.46144772e-01 5.80811143e-01
4.66826856e-01 -1.23408437e+00 -8.60704362e-01 2.00526252e-01
-5.47670841e-01 -1.39784431e+00 -1.47704393e-01 3.57859373e-01
-5.49767435e-01 -1.30401707e+00 2.05451623e-01 -4.90582347e-01
8.94375145e-02 -1.58858687e-01 1.93552184e+00 -1.15074188e-01
-5.56508377e-02 2.76608735e-01 -3.23288381e-01 -5.80603778e-01
-4.84986633e-01 -1.11584917e-01 -2.39931613e-01 -8.15386474e-01
4.53937739e-01 -9.94296074e-01 -1.25295207e-01 -1.11178502e-01
-5.26576400e-01 6.06465936e-01 -1.17770275e-02 7.82894313e-01
3.49982798e-01 1.79165304e-01 1.54366463e-01 -1.25690067e+00
6.04839861e-01 -5.78043044e-01 -5.11149049e-01 2.30166689e-01
-3.59554708e-01 4.68081385e-02 6.46981657e-01 -5.00079751e-01
-1.19630516e+00 -5.72284460e-01 1.63602337e-01 -2.96492547e-01
-1.40791669e-01 7.22149968e-01 1.45139536e-02 5.29199541e-01
1.00800335e+00 3.85296464e-01 -4.88202214e-01 -1.24051161e-01
7.08777130e-01 -3.76431286e-01 8.95249844e-01 -1.32309389e+00
6.64271057e-01 3.41162086e-02 -4.53070626e-02 -3.86586756e-01
-1.76705253e+00 -8.83624256e-02 -2.53265411e-01 -1.26564473e-01
9.54754949e-01 -1.04255927e+00 -9.38454211e-01 1.84007183e-01
-1.07833946e+00 -8.72452617e-01 -5.81787646e-01 3.02082568e-01
-9.67892468e-01 -1.76967710e-01 -1.14046478e+00 -3.29839706e-01
-5.10104597e-02 -9.38235581e-01 6.59465194e-01 2.00270951e-01
-6.53788388e-01 -1.10682964e+00 1.30217761e-01 7.68105567e-01
1.26710320e-02 2.58314669e-01 1.42257333e+00 -9.15583789e-01
-4.03807759e-01 2.04545170e-01 3.42694446e-02 8.40491876e-02
-1.30042329e-01 -1.79261975e-02 -8.15523446e-01 4.07406330e-01
1.10875681e-01 -1.00652277e+00 5.79110146e-01 8.31088722e-02
1.00694191e+00 -3.16783577e-01 -1.39291614e-01 2.63328284e-01
1.03796482e+00 9.02978033e-02 8.73273015e-01 4.39823955e-01
5.87403119e-01 1.19285099e-01 6.06074274e-01 2.75548846e-01
8.34093153e-01 6.66995108e-01 1.58478498e-01 6.12286150e-01
-4.68658358e-01 -6.97012186e-01 2.59115338e-01 7.03112364e-01
-3.87467265e-01 -1.25952318e-01 -1.09539652e+00 9.17669594e-01
-1.80489874e+00 -1.68369102e+00 -1.66465849e-01 1.75716722e+00
1.81456399e+00 4.86832261e-01 -6.27523735e-02 -4.66965251e-02
2.49888957e-01 3.62127692e-01 -3.64817269e-02 -3.24086100e-01
1.41480982e-01 7.93961167e-01 -1.46829173e-01 6.20647490e-01
-9.97501135e-01 1.37502491e+00 6.81505013e+00 1.09123755e+00
-5.65741718e-01 2.89067626e-01 2.19226480e-01 -2.93522745e-01
-2.95714170e-01 3.46633419e-02 -6.98349059e-01 4.20923948e-01
1.06153989e+00 -3.37064177e-01 6.12509310e-01 8.59993160e-01
-8.35854113e-02 -2.66949803e-01 -1.78432059e+00 7.26855159e-01
1.69344366e-01 -1.56262922e+00 -6.08229786e-02 -6.47625983e-01
8.76093209e-01 -1.84389263e-01 -1.04324445e-01 9.70132113e-01
1.07399058e+00 -1.15457022e+00 1.01904440e+00 3.27018172e-01
6.28141522e-01 -4.88041043e-01 2.14900509e-01 3.84376258e-01
-9.14044201e-01 3.61137986e-01 1.22502195e-02 -6.02472901e-01
-1.03317788e-02 4.57022786e-01 -8.58408213e-01 2.55026042e-01
4.73775864e-01 8.06445837e-01 -3.73772174e-01 5.46257615e-01
-1.17596388e+00 1.02112675e+00 -2.30952963e-01 -8.72332454e-02
1.45584002e-01 -9.46170390e-02 6.70983374e-01 1.19013321e+00
1.41524542e-02 6.10403717e-01 5.27065061e-02 1.17232728e+00
-2.22082227e-01 -3.03881586e-01 -7.02911735e-01 -8.16875547e-02
6.89117610e-01 1.04464722e+00 -4.82492149e-01 -5.88929713e-01
-1.77308530e-01 6.64392591e-01 7.04481959e-01 1.81190252e-01
-1.31651413e+00 -2.37989768e-01 6.18431151e-01 3.03625822e-01
3.43147457e-01 -5.31534106e-02 -3.63842815e-01 -1.13975871e+00
-3.38118792e-01 -1.03785491e+00 5.36895096e-01 -1.41166675e+00
-1.02560973e+00 2.82417936e-03 4.43758190e-01 -8.08330715e-01
-5.75426996e-01 -3.32234025e-01 -8.62299919e-01 4.55482811e-01
-1.05699265e+00 -1.11140954e+00 -3.36008430e-01 5.98645329e-01
6.21032894e-01 1.27910748e-01 1.02473330e+00 -5.56614660e-02
-2.62376964e-01 4.27409112e-01 -5.52851975e-01 2.98375458e-01
4.91867989e-01 -1.38468564e+00 1.42826572e-01 9.60453451e-01
3.13890189e-01 6.41275465e-01 1.27104938e+00 -6.06182158e-01
-1.23687768e+00 -7.94529796e-01 9.80701685e-01 -1.08259988e+00
1.16389620e+00 -1.60203248e-01 -8.65814865e-01 1.44638979e+00
4.63232338e-01 -1.74843311e-01 7.89860368e-01 7.55352080e-01
-9.20933545e-01 2.45877206e-01 -8.66844237e-01 9.80996132e-01
1.52902973e+00 -8.71450126e-01 -1.55164528e+00 7.03494012e-01
9.31263208e-01 -1.17215395e+00 -9.63604391e-01 5.32749705e-02
2.25033954e-01 -9.45618331e-01 9.32484090e-01 -1.11173558e+00
1.55656159e+00 -1.00035006e-02 -2.57766619e-02 -1.68457782e+00
-4.17511255e-01 -5.83058596e-01 -4.39610749e-01 1.12790680e+00
3.33326370e-01 -6.08468950e-02 4.42675710e-01 5.78095436e-01
-4.48477596e-01 -5.95427036e-01 -7.74595499e-01 -6.73124075e-01
3.78315970e-02 -7.20916331e-01 2.72855580e-01 1.38267684e+00
6.70152962e-01 7.90832341e-01 1.15343526e-01 -1.73938453e-01
5.30732155e-01 3.11052710e-01 5.53209543e-01 -9.89629269e-01
-6.02446556e-01 -5.88878572e-01 -4.03112099e-02 -7.92611361e-01
4.43429470e-01 -1.11564589e+00 9.78506207e-02 -1.55244684e+00
4.89866376e-01 -7.02877566e-02 -2.66601183e-02 8.81274760e-01
-2.86819935e-01 -7.69271851e-02 1.02595605e-01 -3.26256484e-01
-8.31903219e-01 3.13197792e-01 1.39304805e+00 -1.62023127e-01
2.09834846e-03 -5.84623635e-01 -6.66476011e-01 8.93895447e-01
3.32760721e-01 -3.45004261e-01 -7.43928552e-01 -4.05010700e-01
6.40478849e-01 4.56512630e-01 3.76349509e-01 -1.10856724e+00
1.80063114e-01 -7.59299576e-01 -2.63921209e-02 -1.55832037e-01
4.75758523e-01 -3.46368670e-01 1.06940590e-01 1.32478565e-01
-8.52517724e-01 -7.89575502e-02 4.50202346e-01 6.42617494e-02
-1.39837876e-01 -2.64522672e-01 5.88254511e-01 -3.23455125e-01
-7.39725769e-01 -4.59946692e-01 -2.74716824e-01 8.44011068e-01
9.64733660e-01 -7.46273771e-02 -8.57925951e-01 -4.87901896e-01
-7.33120501e-01 -2.52599455e-02 2.11343709e-02 2.77802408e-01
4.40006524e-01 -1.17941403e+00 -7.37134457e-01 -4.07435954e-01
2.05037430e-01 1.53301448e-01 1.21036001e-01 8.50758016e-01
-5.16322434e-01 2.79634297e-01 -2.30701715e-01 -2.42568076e-01
-1.18668735e+00 5.13578176e-01 1.85093895e-01 -7.25863457e-01
-7.75590241e-01 1.21210682e+00 -9.97538865e-03 -1.95504084e-01
-1.55083546e-02 -7.67483771e-01 -6.69528395e-02 3.02696973e-02
5.90212226e-01 2.28156000e-01 -1.45978844e-02 5.46516851e-03
-1.91152766e-01 -4.73256968e-02 -1.51429310e-01 2.56727915e-02
1.31780255e+00 3.72032344e-01 -1.22492962e-01 9.25905883e-01
6.10794902e-01 2.22326696e-01 -8.17173660e-01 -1.71913937e-01
-2.04462305e-01 -1.59730703e-01 -5.76009601e-02 -1.20081151e+00
-4.05863285e-01 5.87702334e-01 -3.70948523e-01 3.04603189e-01
7.35832393e-01 2.71945238e-01 6.32804871e-01 5.16784906e-01
5.45280457e-01 -1.00810611e+00 3.09441388e-01 1.02021599e+00
9.07314360e-01 -9.59743083e-01 1.02579311e-01 -5.48117936e-01
-6.79154217e-01 7.56752670e-01 8.54009569e-01 -1.57542616e-01
1.69754192e-01 3.18640232e-01 -5.30175805e-01 -5.34916759e-01
-1.14170158e+00 -2.53242165e-01 2.51374215e-01 3.73192668e-01
7.35966682e-01 3.08074892e-01 -1.96406350e-01 9.22545373e-01
-8.50706637e-01 2.72352189e-01 5.98338008e-01 7.98962712e-01
-4.56189007e-01 -4.16884869e-01 -9.32555944e-02 3.69881600e-01
-2.83324569e-01 -6.55527472e-01 -1.49186745e-01 1.08993316e+00
-3.91507559e-02 7.61140764e-01 7.28777125e-02 -1.96002826e-01
3.26451451e-01 4.39603716e-01 1.16994464e+00 -8.06817949e-01
-7.16630697e-01 -6.50127411e-01 8.91630411e-01 -8.62063766e-01
-2.72068560e-01 -5.74329019e-01 -1.41725910e+00 -6.05631948e-01
4.08723690e-02 1.04541425e-02 -1.71550121e-02 1.45151734e+00
-1.65984765e-01 9.10190940e-01 -3.16408336e-01 -3.09700996e-01
-4.28783745e-01 -1.08995974e+00 -2.78331965e-01 9.68597651e-01
-2.24346921e-01 -8.51974249e-01 -6.74454942e-02 3.22354943e-01] | [10.759053230285645, 8.464929580688477] |
5b11672c-28e7-4607-9554-bddc79db9e03 | 3d-siamese-voxel-to-bev-tracker-for-sparse | 2111.04426 | null | https://arxiv.org/abs/2111.04426v2 | https://arxiv.org/pdf/2111.04426v2.pdf | 3D Siamese Voxel-to-BEV Tracker for Sparse Point Clouds | 3D object tracking in point clouds is still a challenging problem due to the sparsity of LiDAR points in dynamic environments. In this work, we propose a Siamese voxel-to-BEV tracker, which can significantly improve the tracking performance in sparse 3D point clouds. Specifically, it consists of a Siamese shape-aware feature learning network and a voxel-to-BEV target localization network. The Siamese shape-aware feature learning network can capture 3D shape information of the object to learn the discriminative features of the object so that the potential target from the background in sparse point clouds can be identified. To this end, we first perform template feature embedding to embed the template's feature into the potential target and then generate a dense 3D shape to characterize the shape information of the potential target. For localizing the tracked target, the voxel-to-BEV target localization network regresses the target's 2D center and the $z$-axis center from the dense bird's eye view (BEV) feature map in an anchor-free manner. Concretely, we compress the voxelized point cloud along $z$-axis through max pooling to obtain a dense BEV feature map, where the regression of the 2D center and the $z$-axis center can be performed more effectively. Extensive evaluation on the KITTI and nuScenes datasets shows that our method significantly outperforms the current state-of-the-art methods by a large margin. | ['Jian Yang', 'Jin Xie', 'Mingmei Cheng', 'Lingpeng Wang', 'Le Hui'] | 2021-11-08 | null | http://proceedings.neurips.cc/paper/2021/hash/f0fcf351df4eb6786e9bb6fc4e2dee02-Abstract.html | http://proceedings.neurips.cc/paper/2021/file/f0fcf351df4eb6786e9bb6fc4e2dee02-Paper.pdf | neurips-2021-12 | ['3d-object-tracking'] | ['computer-vision'] | [-3.59251618e-01 -3.31644505e-01 -2.12601572e-01 -2.36213535e-01
-7.71674514e-01 -4.99697685e-01 4.88691926e-01 -1.57212108e-01
-4.23102736e-01 1.31958127e-01 -2.84908950e-01 1.12573773e-01
-5.90430573e-02 -6.55005217e-01 -9.21439707e-01 -6.88492298e-01
-2.43206322e-01 7.98110306e-01 5.13304770e-01 8.82244557e-02
1.78217754e-01 1.12918437e+00 -1.46612465e+00 -4.24130738e-01
3.76558036e-01 1.35562313e+00 1.26809433e-01 1.99819848e-01
-1.78224623e-01 -8.12171772e-02 -4.45965022e-01 1.84038341e-01
7.10098088e-01 3.32120717e-01 1.67616785e-01 -6.02423996e-02
1.02781045e+00 -1.69067383e-01 -5.23771048e-01 1.17245114e+00
3.01262259e-01 5.14882691e-02 7.32169092e-01 -1.46738744e+00
-2.82520384e-01 -2.21785665e-01 -9.52881992e-01 2.31194720e-01
5.00248857e-02 3.60094607e-01 7.92250574e-01 -1.27342522e+00
7.31337309e-01 1.50521016e+00 7.45594919e-01 3.90385032e-01
-1.13599181e+00 -1.27466905e+00 3.41474503e-01 -6.81504458e-02
-1.82791424e+00 -1.03119589e-01 1.00861061e+00 -6.79279327e-01
8.06931257e-01 -4.69050230e-03 8.74604464e-01 6.67699695e-01
2.07663849e-01 6.32115304e-01 6.96799517e-01 1.89126894e-01
-1.44563630e-01 1.38651049e-02 -1.52717561e-01 8.75853598e-01
2.92823017e-01 6.18147850e-01 -5.13203621e-01 -2.16557309e-01
8.23146105e-01 4.86971170e-01 -9.06580836e-02 -1.11115503e+00
-1.32793117e+00 9.06189978e-01 1.10020292e+00 -1.16699852e-01
-2.48030290e-01 4.36900169e-01 -8.29084683e-03 -8.54512379e-02
5.17453730e-01 -4.79506478e-02 -4.82765108e-01 2.63778448e-01
-9.37594175e-01 5.12857556e-01 4.19997901e-01 1.25852025e+00
9.71292317e-01 1.22807182e-01 -2.15894565e-01 2.53338665e-01
9.29789007e-01 1.17864716e+00 -7.55883381e-02 -8.53426516e-01
5.48759222e-01 8.89006674e-01 2.64197797e-01 -9.97087836e-01
-2.32280880e-01 -4.98518169e-01 -5.16387403e-01 6.36759818e-01
3.05334687e-01 1.15942828e-01 -9.74956274e-01 1.47347355e+00
1.00398529e+00 7.00421035e-01 -3.16444606e-01 1.14562702e+00
7.93477774e-01 6.49189591e-01 -2.25928441e-01 1.07582383e-01
1.19279456e+00 -6.40398264e-01 -1.14840582e-01 -2.91308761e-01
3.77524406e-01 -6.44664586e-01 5.27566910e-01 -2.28105649e-01
-7.15746284e-01 -5.94589889e-01 -1.01883221e+00 1.79133743e-01
-2.51166284e-01 2.18375191e-01 2.51650542e-01 2.82769531e-01
-6.41519547e-01 3.38204980e-01 -1.12888455e+00 -1.14137240e-01
7.82104373e-01 5.23852170e-01 -3.36113125e-01 -9.49921012e-02
-5.74295223e-01 7.93252051e-01 1.35735527e-01 5.78243621e-02
-1.08248806e+00 -1.12934399e+00 -1.12645245e+00 -1.82341132e-02
5.43796241e-01 -7.47983217e-01 8.27424705e-01 2.61515309e-03
-1.19764185e+00 8.98256421e-01 -3.25334460e-01 -3.58429402e-01
4.23761278e-01 -1.73964992e-01 -6.08052760e-02 -1.67496786e-01
4.61659133e-01 7.68763065e-01 1.11875188e+00 -1.19157207e+00
-8.90958130e-01 -9.85085547e-01 -5.38547039e-01 1.96892783e-01
3.22127432e-01 -2.29778752e-01 -6.34769797e-01 -3.89890641e-01
6.23902857e-01 -1.15056086e+00 -6.50697574e-02 7.05215573e-01
-2.06182763e-01 -4.87809688e-01 1.38061988e+00 -9.26931202e-02
4.55020458e-01 -2.44829059e+00 -4.63782810e-02 2.54283875e-01
4.78146672e-01 7.50102326e-02 -2.34276652e-01 -3.00881732e-02
1.50777295e-01 -2.81064957e-01 1.45866856e-01 -3.64655495e-01
5.13765216e-02 -1.87715255e-02 -3.40604454e-01 1.00879419e+00
3.59452069e-01 1.07690406e+00 -7.82565832e-01 -4.90372717e-01
4.21045750e-01 8.10026407e-01 -3.82066458e-01 2.06964254e-01
-3.20926875e-01 3.12896311e-01 -9.56489444e-01 9.18204367e-01
9.69364941e-01 -8.50858465e-02 -6.94464862e-01 -2.60049194e-01
-4.44845080e-01 -1.18242763e-01 -1.18857646e+00 1.65319347e+00
-9.85168591e-02 4.52283621e-01 2.53787905e-01 -4.68002766e-01
1.40933239e+00 -2.07580283e-01 9.20042396e-01 -4.13531303e-01
1.35869816e-01 7.44896680e-02 -6.07957393e-02 2.03077421e-01
2.26634771e-01 -7.36147091e-02 1.03065753e-02 -3.60449404e-02
-1.06208853e-01 -5.26056647e-01 -4.39999402e-01 -9.05439779e-02
8.54553103e-01 3.04876089e-01 -1.17996581e-01 2.78513394e-02
4.91557300e-01 2.26414263e-01 8.08758974e-01 4.75877523e-01
-4.70986485e-01 3.14839363e-01 -1.31145611e-01 -4.47643220e-01
-8.41974974e-01 -1.50375009e+00 -2.13556722e-01 4.85846967e-01
4.49180812e-01 -1.45831779e-01 -1.46063149e-01 -7.83397198e-01
8.16908896e-01 2.78755963e-01 -4.86702472e-01 -1.56950325e-01
-7.21997082e-01 1.31483406e-01 -2.35745944e-02 5.16078055e-01
2.61946499e-01 -6.38110876e-01 -7.04977632e-01 4.78365533e-02
3.97410750e-01 -1.07248521e+00 -7.62169063e-01 1.56121597e-01
-9.61503685e-01 -1.07727134e+00 -4.76981461e-01 -6.65503860e-01
5.47248840e-01 6.52178764e-01 5.85373461e-01 -1.95312560e-01
-3.00547838e-01 3.36644679e-01 6.69727549e-02 -7.70332098e-01
3.24955285e-01 -2.63661504e-01 3.43363136e-01 -1.48544803e-01
7.37777352e-01 -2.89368272e-01 -6.35937393e-01 5.71766555e-01
-1.87589690e-01 -4.77347255e-01 3.72001976e-01 5.49876153e-01
1.32481706e+00 -4.40785177e-02 -1.52843624e-01 -2.59862304e-01
-1.33574530e-01 -3.42392385e-01 -1.53394401e+00 -2.66358942e-01
-4.03457910e-01 -1.60735801e-01 1.31953195e-01 -5.82972586e-01
-2.55073845e-01 7.92388916e-01 1.73665956e-01 -1.50927758e+00
7.81210363e-02 8.27072114e-02 -2.25711167e-01 -6.00026846e-01
4.08645511e-01 2.25946754e-01 9.64171737e-02 -4.96047616e-01
3.13936561e-01 2.40592495e-01 5.53273797e-01 -3.03536713e-01
1.78081906e+00 7.71298647e-01 3.79944533e-01 -6.90997064e-01
-8.96631062e-01 -8.29933822e-01 -8.74125957e-01 -3.08561414e-01
8.65103543e-01 -1.27304304e+00 -9.94438589e-01 -2.03112438e-02
-1.19199216e+00 -3.68283093e-02 -6.07778311e-01 7.91116238e-01
-4.76120442e-01 8.43305290e-02 -4.26081419e-02 -8.06744516e-01
-2.86145031e-01 -1.14073241e+00 1.68118954e+00 3.41747969e-01
2.61858821e-01 -4.95988876e-01 2.40507737e-01 2.39218082e-02
-1.05944439e-03 3.86369318e-01 3.62505466e-01 -4.56479251e-01
-1.25576115e+00 -6.30681157e-01 -4.69486296e-01 1.87061690e-02
-2.04231560e-01 8.65337625e-03 -7.07240283e-01 -5.92464387e-01
1.70409560e-01 1.61794662e-01 6.73996270e-01 6.79714799e-01
7.97253013e-01 2.15916499e-01 -8.67263615e-01 1.13192511e+00
1.38498271e+00 4.56661470e-02 -5.80379739e-02 1.91975459e-01
8.74738753e-01 2.13030338e-01 1.16798866e+00 3.08377951e-01
3.52162123e-01 8.32336247e-01 9.64106858e-01 1.66335776e-01
-1.37359828e-01 -6.11059070e-01 2.62570113e-01 5.07852674e-01
2.34082952e-01 3.56236339e-01 -7.96649218e-01 5.05693614e-01
-1.58090413e+00 -5.68743587e-01 -9.54708979e-02 2.31941199e+00
1.51155397e-01 2.43104473e-01 7.57304952e-02 -5.23897588e-01
6.31226718e-01 3.01859736e-01 -9.17210698e-01 4.37915444e-01
2.59646714e-01 8.49933773e-02 7.16444910e-01 3.01345557e-01
-1.14900804e+00 1.22463512e+00 4.87184572e+00 6.69318557e-01
-1.23977983e+00 4.06947508e-02 -2.97466516e-01 -3.97232503e-01
1.36577234e-01 -6.18177420e-03 -1.65008461e+00 4.71613377e-01
5.21082640e-01 -2.55097806e-01 6.37009069e-02 1.23590457e+00
5.48309758e-02 2.77027845e-01 -1.05741131e+00 1.26263952e+00
1.38780307e-02 -1.31251287e+00 -1.91497698e-01 4.47722167e-01
3.43352437e-01 7.36670554e-01 2.47503638e-01 3.78120333e-01
3.22380632e-01 -8.55475664e-01 7.24656999e-01 3.93253595e-01
8.01784873e-01 -6.81186140e-01 3.50522906e-01 6.86173618e-01
-1.83921123e+00 1.10554710e-01 -8.15605223e-01 2.92583585e-01
6.71664998e-02 3.64097834e-01 -9.30396736e-01 2.53686696e-01
1.04316175e+00 9.68540549e-01 -3.96168768e-01 1.40909123e+00
-1.50333464e-01 1.48773924e-01 -8.30947518e-01 2.79982015e-03
3.58212978e-01 -3.41416985e-01 1.05682194e+00 7.47280777e-01
7.08721936e-01 -1.38612147e-02 5.13284802e-01 1.18720102e+00
-3.63470912e-02 -8.51039365e-02 -9.44689512e-01 2.20677435e-01
6.89428329e-01 1.27990949e+00 -4.97063994e-01 -6.76347269e-03
-2.75173485e-01 4.98684704e-01 2.89085120e-01 1.84395045e-01
-8.53945136e-01 -1.93183035e-01 1.14667964e+00 3.13385606e-01
9.93336201e-01 -5.89899421e-01 1.17118984e-01 -1.12320507e+00
7.86242783e-02 -3.31184089e-01 1.90555155e-01 -8.17818224e-01
-1.25064492e+00 4.72142071e-01 -1.71389461e-01 -1.72378361e+00
2.19241176e-02 -7.36759365e-01 -6.03031635e-01 1.13984644e+00
-1.59695196e+00 -1.34063780e+00 -4.92567450e-01 8.57254207e-01
2.82702088e-01 -1.72541410e-01 3.09223354e-01 3.39241587e-02
-1.69402942e-01 3.61368448e-01 -6.07425570e-02 2.77332753e-01
4.85234976e-01 -8.69616568e-01 4.71389472e-01 5.32334924e-01
4.69416142e-01 4.08482075e-01 3.76969248e-01 -1.09978557e+00
-1.80401444e+00 -1.53250575e+00 5.25797367e-01 -8.53813052e-01
8.02066565e-01 -6.22558177e-01 -7.96359658e-01 8.69246244e-01
-5.78571975e-01 7.48227835e-01 8.94267038e-02 -9.05123800e-02
-5.49148023e-01 -3.89437258e-01 -1.16197085e+00 4.34197038e-01
1.12803042e+00 -4.90294755e-01 -7.09688246e-01 3.56223792e-01
9.30561304e-01 -8.17923784e-01 -8.72751176e-01 5.94745278e-01
4.28942263e-01 -3.94253194e-01 1.54721987e+00 -4.38686311e-01
-2.46622518e-01 -8.20665061e-01 -2.03969359e-01 -1.17594182e+00
-4.62011069e-01 -2.82903194e-01 -3.52269292e-01 8.01789939e-01
-9.01340507e-03 -4.50704128e-01 1.31526756e+00 1.16805494e-01
-2.15937525e-01 -5.64456582e-01 -1.53947544e+00 -1.01259124e+00
4.16171215e-02 -4.84736621e-01 6.32732332e-01 6.22690320e-01
-7.26008832e-01 2.78353631e-01 1.85502414e-03 6.55571103e-01
1.25464237e+00 4.90057766e-01 1.21366525e+00 -1.67449296e+00
2.94994622e-01 -3.41088146e-01 -8.30358326e-01 -1.45323706e+00
4.08610582e-01 -1.16072381e+00 1.36448339e-01 -1.11268651e+00
-1.29344299e-01 -5.80750406e-01 -1.46854356e-01 2.23660082e-01
8.97925347e-02 1.56546786e-01 4.66544539e-01 2.59960473e-01
-4.30294335e-01 8.86037827e-01 1.49835563e+00 -4.84175324e-01
-1.22770764e-01 3.93678784e-01 -1.43700272e-01 6.55444086e-01
2.46661142e-01 -9.22168553e-01 8.05696249e-02 -3.72885793e-01
-3.51164401e-01 1.36014417e-01 7.11804569e-01 -1.01004744e+00
5.23720086e-01 -1.86986730e-01 7.51663387e-01 -1.57955956e+00
8.86395693e-01 -1.33857965e+00 2.31925566e-02 5.64230025e-01
1.96516499e-01 -9.71864909e-02 2.00900212e-01 8.07100773e-01
-1.04893215e-01 8.69712159e-02 8.87085676e-01 -5.07428264e-03
-5.11767268e-01 1.23159528e+00 2.93339312e-01 -9.14695635e-02
1.34249043e+00 -4.29843903e-01 -9.82859060e-02 7.83241838e-02
-3.87905419e-01 5.90338111e-01 6.30822122e-01 5.34075439e-01
9.89093244e-01 -1.63977861e+00 -6.80118084e-01 6.01091385e-01
2.40289584e-01 5.22330403e-01 5.84234744e-02 7.85843253e-01
-1.76940814e-01 4.66239363e-01 -2.32483037e-02 -1.44549930e+00
-1.35918844e+00 8.62413526e-01 4.13856804e-01 1.02902249e-01
-9.89937663e-01 9.53001797e-01 3.96673173e-01 -5.56843042e-01
4.15447593e-01 -4.83347267e-01 1.78750381e-02 -2.27308914e-01
3.90640914e-01 1.63476244e-01 -3.66908967e-01 -1.07301700e+00
-6.86152041e-01 1.38129985e+00 9.24626291e-02 1.53504461e-01
1.36674047e+00 1.17457062e-01 1.63789526e-01 2.62885988e-01
1.37959170e+00 6.31197840e-02 -1.67375362e+00 -5.21711707e-01
-7.99564496e-02 -9.72530723e-01 1.98884770e-01 -1.85804650e-01
-1.37164021e+00 9.86597061e-01 9.96818423e-01 -2.66533464e-01
4.34635818e-01 3.58523667e-01 6.32184505e-01 3.01317573e-01
4.70818698e-01 -5.58716655e-01 -1.59924943e-02 6.79638624e-01
7.77405381e-01 -1.31064165e+00 7.42653161e-02 -4.95630533e-01
-2.19129995e-01 9.01925325e-01 7.68811464e-01 -5.90171397e-01
9.93888736e-01 2.69975126e-01 1.24653526e-01 -6.14774048e-01
-3.68527383e-01 -2.73569643e-01 7.18352079e-01 8.90695572e-01
-2.81764120e-01 -4.31540608e-02 4.74622995e-01 2.46297657e-01
-2.42515489e-01 -1.69901803e-01 -2.84422427e-01 7.38092840e-01
-6.25941753e-01 -6.80731177e-01 -7.12521076e-01 4.44159746e-01
1.50809988e-01 1.62972227e-01 -2.60143220e-01 1.00037241e+00
2.03190684e-01 3.61686736e-01 3.43939096e-01 -4.11402822e-01
6.53164983e-01 -1.01746760e-01 5.03690243e-01 -8.41792285e-01
-4.04794782e-01 4.64130193e-01 -6.10381186e-01 -8.99113357e-01
-5.48298024e-02 -8.34609926e-01 -1.41241646e+00 -7.03640357e-02
-2.91403681e-01 2.27796682e-03 9.50238168e-01 5.74863911e-01
4.06694591e-01 1.18665643e-01 6.57090962e-01 -1.33566976e+00
-7.12980509e-01 -5.36600530e-01 -6.40732944e-01 1.31950155e-01
4.63456929e-01 -1.21791148e+00 -4.52543378e-01 -4.87067044e-01] | [6.6672682762146, -2.391969680786133] |
2d9251ce-8758-411d-81ee-6268b74df0b0 | gd-vdm-generated-depth-for-better-diffusion | 2306.11173 | null | https://arxiv.org/abs/2306.11173v1 | https://arxiv.org/pdf/2306.11173v1.pdf | GD-VDM: Generated Depth for better Diffusion-based Video Generation | The field of generative models has recently witnessed significant progress, with diffusion models showing remarkable performance in image generation. In light of this success, there is a growing interest in exploring the application of diffusion models to other modalities. One such challenge is the generation of coherent videos of complex scenes, which poses several technical difficulties, such as capturing temporal dependencies and generating long, high-resolution videos. This paper proposes GD-VDM, a novel diffusion model for video generation, demonstrating promising results. GD-VDM is based on a two-phase generation process involving generating depth videos followed by a novel diffusion Vid2Vid model that generates a coherent real-world video. We evaluated GD-VDM on the Cityscapes dataset and found that it generates more diverse and complex scenes compared to natural baselines, demonstrating the efficacy of our approach. | ['Ethan Fetaya', 'Lior Bracha', 'Idan Achituve', 'Ariel Lapid'] | 2023-06-19 | null | null | null | null | ['video-generation'] | ['computer-vision'] | [ 2.52953768e-01 7.08200857e-02 3.60084146e-01 -1.24066779e-02
-8.20203900e-01 -5.57152748e-01 1.27191436e+00 -4.64175105e-01
-1.20160311e-01 6.77033126e-01 6.33525670e-01 1.64882302e-01
1.05259024e-01 -8.75749409e-01 -5.05587339e-01 -6.49912238e-01
-7.58555755e-02 4.48601276e-01 2.50538945e-01 -2.10536554e-01
-2.06486583e-02 2.59409070e-01 -1.60196626e+00 5.23230195e-01
5.73845088e-01 4.87051278e-01 3.04174483e-01 1.08051205e+00
2.32131600e-01 9.67612624e-01 -7.67775834e-01 -6.41691208e-01
1.57206833e-01 -9.69417870e-01 -6.36561930e-01 3.56763333e-01
5.51816285e-01 -5.59393048e-01 -4.92901772e-01 6.26522720e-01
7.56261826e-01 2.16867730e-01 8.25709045e-01 -1.47514772e+00
-6.80531323e-01 3.88745099e-01 -5.67391098e-01 2.80589581e-01
8.19643795e-01 3.98999274e-01 7.62154639e-01 -7.07326949e-01
1.33300555e+00 1.38140082e+00 4.20190662e-01 8.33143950e-01
-1.41071391e+00 -3.18358392e-01 -7.09263459e-02 1.14704564e-01
-1.28178954e+00 -5.13726652e-01 5.03345013e-01 -5.70218801e-01
1.16530645e+00 7.59606883e-02 9.72919345e-01 1.76687205e+00
2.29009330e-01 8.18234444e-01 1.00170207e+00 -2.39119217e-01
2.35696912e-01 -1.89345732e-01 -7.00410128e-01 4.21947777e-01
1.57394975e-01 3.27830523e-01 -7.67512560e-01 6.20061271e-02
9.23891664e-01 -3.33181441e-01 -2.92387098e-01 -1.54145181e-01
-1.33170986e+00 8.49848270e-01 3.46244387e-02 2.87528455e-01
-6.80516183e-01 2.27211311e-01 1.50563985e-01 8.42704773e-02
6.32039607e-01 1.65457651e-01 2.12100208e-01 -4.07108366e-01
-1.26759696e+00 7.74782419e-01 5.98274589e-01 9.84177828e-01
2.53134370e-01 2.29496539e-01 -4.66054082e-01 6.60035133e-01
3.49177063e-01 7.13593841e-01 3.29190761e-01 -1.15584981e+00
4.53670114e-01 -3.29808658e-03 3.27579156e-02 -1.02761483e+00
-7.56400153e-02 7.75068849e-02 -9.69779491e-01 3.90776694e-01
1.84008405e-01 -4.24646378e-01 -9.90933299e-01 1.70523489e+00
2.80688226e-01 3.35315377e-01 2.93913901e-01 7.67241716e-01
9.30113554e-01 1.00280142e+00 1.14676796e-01 -1.58824548e-01
7.94941008e-01 -9.48497355e-01 -5.67571938e-01 -2.10621040e-02
1.63704783e-01 -9.13445354e-01 5.23667216e-01 4.22146171e-01
-1.46169686e+00 -6.27028883e-01 -6.96594238e-01 -6.99307621e-02
1.42720323e-02 -2.12726429e-01 6.08152926e-01 5.00715375e-01
-1.45436704e+00 3.98680240e-01 -6.03502691e-01 -5.97690940e-01
4.30470586e-01 -1.15137719e-01 -5.30311823e-01 -3.24806482e-01
-9.61531878e-01 5.80429196e-01 3.33372980e-01 -4.43004584e-03
-1.24960327e+00 -5.07089972e-01 -1.07761073e+00 -2.30134279e-01
-2.80411486e-02 -1.18291640e+00 1.24546182e+00 -7.60214567e-01
-1.45819116e+00 6.87117875e-01 -1.60008788e-01 -4.96352077e-01
1.01811790e+00 -1.58964604e-01 -3.92053097e-01 2.30192110e-01
2.24889651e-01 1.28533733e+00 1.13128233e+00 -1.28881335e+00
-5.99858761e-01 9.67376530e-02 3.75261009e-02 3.44756275e-01
1.02226757e-01 -2.69295663e-01 -6.59190416e-01 -8.37355018e-01
-3.26068193e-01 -1.15318596e+00 -3.34847629e-01 -3.87784421e-01
-3.71693611e-01 -3.27310748e-02 8.63760412e-01 -5.89357257e-01
1.03769374e+00 -2.00998545e+00 3.73200625e-01 -3.71568538e-02
2.06621423e-01 2.42694408e-01 -4.94725019e-01 6.91514254e-01
1.51925804e-02 2.69452125e-01 -1.38451800e-01 -6.19415820e-01
-1.60133436e-01 1.33463576e-01 -4.71451223e-01 -7.27892071e-02
4.48030293e-01 1.04819989e+00 -1.10435724e+00 -5.40203214e-01
2.58513957e-01 8.72909248e-01 -7.37791777e-01 3.58265489e-01
-3.78555208e-01 6.86428010e-01 -8.42332467e-02 4.74277020e-01
5.32099009e-01 -1.76758662e-01 3.04213003e-03 7.80429412e-03
-9.37632471e-02 -1.50857002e-01 -1.11146557e+00 1.86280489e+00
-2.10228235e-01 1.05108583e+00 -4.05584216e-01 -3.51646006e-01
6.14870667e-01 4.65575099e-01 5.14638305e-01 -8.00247014e-01
-1.13712855e-01 -3.60215008e-02 -1.89590856e-01 -7.20179081e-01
7.76950836e-01 -2.41893113e-01 1.10196628e-01 5.37583947e-01
2.08275408e-01 -7.35814631e-01 7.35167563e-01 7.09078550e-01
1.11343849e+00 5.26514649e-01 -1.53595090e-01 2.69242138e-01
4.51507233e-02 9.60872099e-02 1.33684367e-01 7.20568478e-01
1.00407548e-01 1.25541377e+00 3.31216663e-01 -1.33429900e-01
-1.19748402e+00 -1.29913759e+00 4.31309164e-01 3.29268605e-01
-9.15261954e-02 -6.97597086e-01 -8.77291143e-01 -5.65840244e-01
-3.78116280e-01 6.91800833e-01 -6.33190870e-01 7.65135586e-02
-5.13937533e-01 -8.92266452e-01 5.38095713e-01 3.91853660e-01
6.30137384e-01 -1.35288560e+00 -6.27174497e-01 3.76892269e-01
-6.41064167e-01 -1.29121995e+00 -2.56113142e-01 -4.67650205e-01
-7.49265790e-01 -8.83446038e-01 -1.22293663e+00 -5.70931494e-01
5.48511267e-01 4.45413619e-01 1.45216048e+00 -9.49991345e-02
-4.48425114e-01 6.69205785e-01 -4.36573237e-01 -3.22591066e-01
-7.81109691e-01 -5.52231036e-02 -2.12236688e-01 -5.92060871e-02
-1.58089429e-01 -3.93830240e-01 -6.73269987e-01 8.17173272e-02
-1.36061990e+00 4.26884264e-01 4.82455373e-01 5.63431799e-01
5.75076103e-01 1.90304339e-01 4.26870078e-01 -7.93456316e-01
9.11506236e-01 -6.15604103e-01 -4.49793607e-01 -4.33751270e-02
-3.50796908e-01 -1.43969297e-01 2.38396555e-01 -4.02210027e-01
-1.49247694e+00 -6.38668314e-02 -2.93688834e-01 -2.87997127e-01
-1.45712897e-01 4.12204802e-01 1.34114057e-01 4.05151188e-01
6.68993652e-01 2.74053127e-01 -7.13236630e-02 -7.47351050e-02
6.08913958e-01 2.54220605e-01 6.10530674e-01 -4.12916243e-01
9.12791610e-01 6.31821156e-01 -9.24554653e-03 -1.19372797e+00
-3.33250821e-01 -3.02642193e-02 -4.20569688e-01 -5.32960415e-01
1.10622978e+00 -1.09588683e+00 -1.83639601e-01 8.06466341e-01
-1.21498919e+00 -6.91211045e-01 -4.16102201e-01 3.87543082e-01
-7.63710260e-01 3.20932657e-01 -6.62506402e-01 -5.21149993e-01
-7.67985657e-02 -1.01647389e+00 1.28074968e+00 2.69485980e-01
-4.67726886e-01 -1.07835722e+00 5.94540775e-01 2.83986837e-01
5.05409479e-01 5.43451369e-01 5.13949990e-01 1.39405712e-01
-9.58665490e-01 -4.59000329e-03 -6.75966814e-02 2.21263304e-01
1.62756182e-02 4.86622542e-01 -8.92591417e-01 -1.31255433e-01
-3.42366338e-01 -3.78750056e-01 9.15590167e-01 5.31998336e-01
5.10820866e-01 6.87613115e-02 -2.36567169e-01 5.34129143e-01
1.31797469e+00 1.85953826e-01 1.07611358e+00 2.21936464e-01
5.83088100e-01 4.51833069e-01 4.36144710e-01 4.18013453e-01
5.39359272e-01 6.73484027e-01 2.55481124e-01 -2.43417636e-01
-7.66992271e-01 -5.01459539e-01 4.57836568e-01 7.13759840e-01
-4.87474471e-01 -9.04698193e-01 -7.81178951e-01 6.57169521e-01
-1.82197416e+00 -1.66966641e+00 -3.66194695e-01 1.87049782e+00
4.08517063e-01 -5.95778748e-02 2.08933502e-01 8.29478502e-02
4.33819205e-01 3.89531612e-01 -2.66585052e-01 -1.66569859e-01
-5.04853010e-01 1.05330475e-01 -1.78649694e-01 3.38937014e-01
-8.92457664e-01 9.15559828e-01 7.57423592e+00 5.22649467e-01
-1.12223959e+00 -5.68156056e-02 7.57124841e-01 -2.70890087e-01
-4.61538196e-01 -1.88471675e-01 -7.22478271e-01 4.08030957e-01
7.78730392e-01 -2.49712527e-01 2.49780372e-01 5.79566360e-01
5.17919600e-01 -4.39634323e-01 -8.38185489e-01 1.06797290e+00
3.35584909e-01 -1.60551298e+00 3.83093208e-01 2.32383013e-01
1.33295453e+00 5.47666326e-02 2.02817768e-01 1.86829939e-01
6.43119752e-01 -9.21076775e-01 9.33030427e-01 5.59334636e-01
8.07050884e-01 -4.91944611e-01 4.57265109e-01 2.50949055e-01
-1.13146293e+00 2.10090816e-01 -2.53341138e-01 9.25934836e-02
9.02903080e-01 5.73437870e-01 -8.80772889e-01 5.36673725e-01
7.37902343e-01 9.65822458e-01 -5.69777429e-01 1.14250290e+00
-3.61631483e-01 4.68459636e-01 -6.69887215e-02 3.60065073e-01
1.76661417e-01 -2.03307748e-01 4.94901836e-01 1.44999027e+00
8.14056695e-01 2.58533396e-02 -9.14683565e-02 8.15306544e-01
3.69958468e-02 -1.60441905e-01 -1.04713643e+00 -2.20536366e-01
6.56367168e-02 1.16785014e+00 -7.69937396e-01 -5.78536034e-01
-3.25868964e-01 1.36711216e+00 1.28582075e-01 4.78991210e-01
-1.10670412e+00 1.56016484e-01 5.15998244e-01 1.57314405e-01
4.10269350e-01 -5.36795735e-01 2.61581212e-01 -1.23495686e+00
-1.75733507e-01 -9.38747048e-01 3.02450180e-01 -1.21955431e+00
-1.06746817e+00 9.36613023e-01 2.73052871e-01 -1.22613776e+00
-8.18050742e-01 6.43598363e-02 -4.83772039e-01 7.67526686e-01
-1.35279024e+00 -1.25328302e+00 -7.16525376e-01 7.22352803e-01
9.41862047e-01 -4.01202552e-02 6.44417942e-01 2.90197939e-01
-3.00391614e-01 -2.66142543e-02 -2.57164538e-01 -2.50316858e-01
6.60545409e-01 -1.04197359e+00 8.46538365e-01 1.08216107e+00
5.50446212e-01 1.30151525e-01 7.59558439e-01 -8.73470128e-01
-9.89357293e-01 -1.21092916e+00 9.47600782e-01 -5.21256030e-01
2.72422045e-01 -2.02277049e-01 -6.09542489e-01 5.42840481e-01
7.63145506e-01 -3.06517601e-01 4.78420109e-01 -5.57641208e-01
-1.42236874e-01 2.51022577e-01 -9.44871545e-01 7.27153361e-01
1.36891615e+00 -3.34196955e-01 -1.76520258e-01 1.43343925e-01
4.04141217e-01 -4.91778314e-01 -5.65367997e-01 1.56247750e-01
5.50583959e-01 -1.49765372e+00 1.00277400e+00 -1.40024096e-01
8.80190134e-01 -2.99374282e-01 -1.85797475e-02 -1.60639930e+00
-3.76940429e-01 -9.13072050e-01 -1.29286304e-01 1.33424544e+00
2.45175108e-01 -1.19347006e-01 5.94182611e-01 4.69395697e-01
1.14452511e-01 -2.10509986e-01 -4.74393278e-01 -7.25233018e-01
-2.29074016e-01 -4.68259186e-01 4.82015252e-01 7.17101216e-01
-4.79322910e-01 4.64553416e-01 -6.51733279e-01 -1.48950249e-01
5.52447557e-01 -6.66261464e-02 1.20777214e+00 -9.66020226e-01
-4.72818941e-01 -3.97351652e-01 -4.16547060e-01 -1.06935382e+00
-9.08469707e-02 -5.34999847e-01 6.18259609e-02 -2.07896209e+00
1.98096693e-01 -1.43070504e-01 3.15443307e-01 -7.58364424e-02
-1.35332644e-01 6.54921591e-01 4.26579267e-01 2.03847945e-01
-4.78078216e-01 5.25089502e-01 1.52525556e+00 -1.02883235e-01
-2.21944317e-01 -2.44113758e-01 -4.99810338e-01 5.29932737e-01
5.55167854e-01 -3.54431301e-01 -8.97386789e-01 -6.75166130e-01
3.69129598e-01 2.09155083e-01 4.20714527e-01 -1.25039327e+00
-9.24552679e-02 -1.28815010e-01 6.14009976e-01 -5.55769145e-01
5.52679956e-01 -3.64496499e-01 7.92496800e-01 4.18564290e-01
-1.00318760e-01 5.31268895e-01 1.48278281e-01 5.21983683e-01
-3.99544924e-01 1.01621732e-01 5.54942429e-01 -2.55422771e-01
-9.68252242e-01 3.32543224e-01 -5.74163377e-01 1.92464277e-01
1.11196160e+00 -2.82642394e-01 -3.18309456e-01 -8.54975939e-01
-7.97414541e-01 -2.51558155e-01 5.82471788e-01 7.95239866e-01
7.57227540e-01 -1.37463176e+00 -1.16290677e+00 2.68860638e-01
-4.81061190e-02 1.08496562e-01 5.36039889e-01 5.37466824e-01
-6.12813890e-01 1.93579584e-01 -1.95126772e-01 -7.19019055e-01
-1.34450281e+00 3.98277134e-01 2.72374582e-02 -3.27941746e-01
-7.59043694e-01 8.27179849e-01 3.80594343e-01 2.06403881e-01
-2.28505805e-01 -1.30631208e-01 -1.71509907e-01 2.24977240e-01
6.80172026e-01 3.54800582e-01 -1.89039245e-01 -9.12685692e-01
2.51490716e-02 4.11675185e-01 1.84336901e-01 -7.78230727e-01
1.42012739e+00 -2.07037210e-01 3.03617209e-01 1.61651924e-01
8.52713406e-01 7.62828290e-02 -1.58928275e+00 1.75609842e-01
-3.48400533e-01 -6.30213678e-01 -1.79018825e-01 -6.79953337e-01
-1.11655915e+00 7.75208890e-01 3.77856135e-01 3.07350665e-01
1.09393322e+00 -2.09913636e-03 8.95213485e-01 -2.04296991e-01
4.71908718e-01 -7.89117754e-01 5.26841938e-01 4.41668123e-01
1.14913285e+00 -1.14851737e+00 -9.65391770e-02 -2.50336647e-01
-9.02198732e-01 8.97165060e-01 4.64073956e-01 -1.05605042e-02
3.23970228e-01 2.27805823e-01 2.90321440e-01 -1.76728353e-01
-1.00130498e+00 -2.88422674e-01 3.00596088e-01 1.12111223e+00
5.20428658e-01 -1.32122383e-01 -1.24940999e-01 -1.43781200e-01
-2.29917899e-01 2.20573038e-01 7.50789046e-01 7.97996938e-01
4.14992049e-02 -1.21034753e+00 -2.57655501e-01 4.78745624e-02
-1.87968984e-01 -7.17630014e-02 -3.26654643e-01 8.32387626e-01
1.31578296e-01 1.12603498e+00 9.37300995e-02 -2.28334144e-01
1.93986207e-01 -1.93641931e-01 7.00487316e-01 -5.13382494e-01
-3.34673285e-01 3.69148642e-01 1.79310963e-01 -6.37204945e-01
-8.18226218e-01 -1.00794363e+00 -9.07054484e-01 -3.43271375e-01
5.55932596e-02 -9.15897191e-02 6.88224673e-01 6.03591502e-01
5.21366596e-01 5.79055130e-01 3.94740909e-01 -1.22712016e+00
9.17068571e-02 -8.99241507e-01 -3.19648743e-01 7.64555275e-01
2.16121987e-01 -4.56987321e-01 -5.96437752e-02 6.54880106e-01] | [11.004794120788574, -0.29501327872276306] |
0675a295-3b9d-490b-b460-d2b01e0f95e5 | deepprog-a-transformer-based-framework-for | 2104.03642 | null | https://arxiv.org/abs/2104.03642v3 | https://arxiv.org/pdf/2104.03642v3.pdf | CLIMAT: Clinically-Inspired Multi-Agent Transformers for Knee Osteoarthritis Trajectory Forecasting | In medical applications, deep learning methods are built to automate diagnostic tasks. However, a clinically relevant question that practitioners usually face, is how to predict the future trajectory of a disease (prognosis). Current methods for such a problem often require domain knowledge, and are complicated to apply. In this paper, we formulate the prognosis prediction problem as a one-to-many forecasting problem from multimodal data. Inspired by a clinical decision-making process with two agents -- a radiologist and a general practitioner, we model a prognosis prediction problem with two transformer-based components that share information between each other. The first block in this model aims to analyze the imaging data, and the second block leverages the internal representations of the first one as inputs, also fusing them with auxiliary patient data. We show the effectiveness of our method in predicting the development of structural knee osteoarthritis changes over time. Our results show that the proposed method outperforms the state-of-the-art baselines in terms of various performance metrics. In addition, we empirically show that the existence of the multi-agent transformers with depths of 2 is sufficient to achieve good performances. Our code is publicly available at \url{https://github.com/MIPT-Oulu/CLIMAT}. | ['Aleksei Tiulpin', 'Matthew B. Blaschko', 'Simo Saarakkala', 'Huy Hoang Nguyen'] | 2021-04-08 | null | null | null | null | ['disease-trajectory-forecasting'] | ['medical'] | [-1.01964444e-01 1.73546538e-01 -2.79815495e-01 -2.91420668e-01
-1.01927054e+00 -1.40781552e-01 3.50571662e-01 3.38880479e-01
-8.34073722e-02 6.96617842e-01 4.48303878e-01 -3.51537436e-01
-4.67267722e-01 -6.00367844e-01 -5.04121184e-01 -9.27457333e-01
-3.86589706e-01 9.20066237e-01 1.74587011e-01 6.55896664e-02
-1.22319221e-01 2.62295991e-01 -9.91707087e-01 5.30484557e-01
6.94017947e-01 1.37854648e+00 1.23854473e-01 4.86450791e-01
4.79426473e-01 1.42372692e+00 9.44241043e-03 -3.26095611e-01
-6.94791079e-02 -2.91243464e-01 -9.55050409e-01 5.00218831e-02
-2.57856101e-01 -4.46758032e-01 -4.35077280e-01 5.92138290e-01
6.13948166e-01 -2.81255364e-01 8.21173787e-01 -1.14450514e+00
-2.31543601e-01 4.42247182e-01 -4.52112764e-01 2.92200655e-01
-2.39098631e-02 1.56480119e-01 9.35525417e-01 -6.09844804e-01
5.28944373e-01 8.59392166e-01 6.28988504e-01 3.78606528e-01
-1.02297199e+00 -3.35144103e-01 1.11875430e-01 4.21269625e-01
-1.02571487e+00 -2.46510670e-01 8.44115615e-01 -7.88472235e-01
4.32906151e-01 1.13713861e-01 7.06407607e-01 1.30572152e+00
6.16087794e-01 8.90159190e-01 1.01905298e+00 8.06137845e-02
2.61150718e-01 -1.67545706e-01 2.50651270e-01 6.64666176e-01
-8.57208520e-02 1.32014364e-01 -4.75751072e-01 -3.97528112e-01
6.28678679e-01 2.45084241e-01 -3.04527760e-01 -3.03032100e-01
-1.41333616e+00 8.29928577e-01 5.52479863e-01 2.91052073e-01
-8.70448530e-01 1.99889451e-01 5.03848076e-01 2.73899317e-01
5.10851324e-01 2.03149214e-01 -3.53135198e-01 1.34122847e-02
-7.93588698e-01 2.45022923e-01 5.70974588e-01 2.88319111e-01
9.41524561e-03 -3.95706505e-01 -1.57980263e-01 7.82171607e-01
3.59262556e-01 2.06734911e-01 7.86861241e-01 -8.00475240e-01
2.23074108e-02 5.26109815e-01 1.52050048e-01 -8.60151649e-01
-7.50782490e-01 -8.90647173e-01 -1.05923676e+00 -1.12875327e-01
3.47117603e-01 -1.91047594e-01 -8.04231405e-01 1.63052201e+00
5.16003668e-01 4.37549621e-01 -1.34529456e-01 1.10447192e+00
5.96800923e-01 3.74937654e-01 1.39642566e-01 -2.05488905e-01
1.67580330e+00 -9.95361447e-01 -5.16929984e-01 -2.77144819e-01
8.64195585e-01 -5.04124582e-01 6.76297724e-01 6.04847491e-01
-1.09177589e+00 -1.85967103e-01 -6.63422763e-01 9.42065492e-02
8.68723094e-02 4.19349432e-01 6.46964908e-01 -1.01446301e-01
-9.58369374e-01 7.04300046e-01 -1.26680028e+00 -2.68017113e-01
3.66071373e-01 3.98473293e-01 -2.57202119e-01 -1.28632158e-01
-1.25022507e+00 1.04856014e+00 -7.17732534e-02 2.20496655e-01
-1.19279432e+00 -7.76410878e-01 -3.53659838e-01 -2.62164567e-02
2.73863167e-01 -1.19887531e+00 1.49473858e+00 -7.17687547e-01
-1.07454240e+00 8.18942010e-01 -4.23007347e-02 -5.34599960e-01
6.73816144e-01 -3.72725457e-01 -2.54176378e-01 2.13842779e-01
1.26987755e-01 2.16280982e-01 6.01438761e-01 -1.14449573e+00
-6.92251444e-01 -6.87250793e-01 -7.01490492e-02 1.64038062e-01
-2.91627914e-01 -1.77347600e-01 -2.75119185e-01 -6.72320247e-01
7.93530606e-03 -1.02295792e+00 -5.03199041e-01 6.32765666e-02
-5.60736418e-01 -2.66378611e-01 4.82867062e-01 -8.25694740e-01
1.19542229e+00 -1.95372999e+00 4.23466623e-01 1.37820959e-01
6.10956132e-01 -1.42905578e-01 1.27728418e-01 5.63909650e-01
-2.43600860e-01 -2.93610901e-01 -9.05333981e-02 -4.19500798e-01
-2.64754415e-01 1.17159970e-01 -1.62849188e-01 6.01978421e-01
-1.71502046e-02 8.89591992e-01 -7.97784448e-01 -5.42271316e-01
-1.05593339e-01 5.17124951e-01 -4.28672433e-01 4.33668882e-01
-1.00553803e-01 8.14950109e-01 -8.96863759e-01 6.83551788e-01
6.29874170e-02 -1.06619728e+00 4.64003801e-01 -3.25684488e-01
2.59613335e-01 3.28056753e-01 -7.49764264e-01 1.66340208e+00
-2.97663420e-01 -4.73062554e-03 -2.84364522e-02 -1.46538043e+00
4.76442277e-01 6.44306660e-01 1.08351719e+00 -4.49604869e-01
2.79214621e-01 4.51945186e-01 2.93727249e-01 -7.31254041e-01
-2.82660723e-01 -3.62849414e-01 1.28193814e-02 6.21258557e-01
-1.10223100e-01 3.04924339e-01 -5.62620461e-02 2.26764351e-01
1.56765282e+00 -1.62605837e-01 1.79053679e-01 -7.73943216e-02
4.84403998e-01 9.03115645e-02 6.61698043e-01 5.26713431e-01
-8.86346698e-02 3.58559638e-01 7.98663259e-01 -6.03789568e-01
-8.85290623e-01 -9.49540615e-01 -5.96963838e-02 6.36128724e-01
-3.36921364e-02 -8.33793059e-02 -5.16549826e-01 -6.19294405e-01
4.46239188e-02 2.97473669e-01 -8.09567094e-01 -2.12371111e-01
-4.96786147e-01 -9.37473476e-01 2.34050810e-01 6.17200792e-01
-4.34247926e-02 -8.45194519e-01 -7.65963733e-01 3.89713883e-01
-4.01154816e-01 -1.03191710e+00 -2.07117066e-01 3.75623885e-03
-1.13268411e+00 -1.19082057e+00 -9.62912679e-01 -5.93961716e-01
5.04096508e-01 -2.06325158e-01 1.04741776e+00 2.49744833e-01
-2.82080501e-01 5.59696734e-01 -3.11041296e-01 -2.20635131e-01
-3.89196664e-01 7.33161196e-02 8.56072381e-02 2.15678200e-01
-1.53603479e-01 -8.46597850e-01 -1.23520911e+00 1.17434166e-01
-7.34272957e-01 3.57712746e-01 8.71754229e-01 1.01713216e+00
7.18281865e-01 -1.18316770e-01 5.69717467e-01 -8.36660802e-01
4.56373900e-01 -8.69785607e-01 -9.03213024e-02 2.14586899e-01
-5.71658969e-01 -2.85537969e-02 4.29589659e-01 -4.02781248e-01
-7.40830064e-01 1.32317036e-01 -9.41214338e-02 -4.91249442e-01
-1.23354636e-01 1.04897237e+00 2.67320007e-01 3.67904902e-01
2.84114659e-01 2.30881482e-01 2.64594525e-01 -5.16993582e-01
-3.84173915e-02 4.40443784e-01 5.11038184e-01 -5.35245657e-01
3.36053282e-01 5.69259763e-01 1.40573651e-01 -2.87428468e-01
-1.09091437e+00 -3.57481003e-01 -2.94682056e-01 -4.61613268e-01
6.75954163e-01 -9.54359174e-01 -8.31235230e-01 5.61354518e-01
-9.21521008e-01 -4.74165589e-01 -2.37962812e-01 6.32307708e-01
-6.68776393e-01 2.22955234e-02 -9.98017907e-01 -3.95362973e-01
-5.65639675e-01 -1.32512188e+00 1.14653742e+00 -1.51667088e-01
-9.77439433e-02 -1.15391254e+00 3.01792800e-01 5.01936853e-01
3.17463696e-01 3.54564697e-01 1.24633408e+00 -8.29883277e-01
-3.77307743e-01 -1.76613390e-01 2.86371857e-02 5.28863147e-02
1.66908219e-01 -5.00510454e-01 -6.99400127e-01 -1.87052786e-01
3.00200433e-01 -4.01302755e-01 7.57138252e-01 6.87626362e-01
1.38224411e+00 -1.78645059e-01 -6.63429379e-01 2.51136661e-01
1.10704529e+00 1.30770430e-01 5.26811779e-01 2.68255323e-01
5.14587820e-01 6.95711017e-01 5.93135178e-01 5.88856459e-01
8.75228405e-01 5.90532482e-01 6.90122068e-01 -1.29753903e-01
-8.93455371e-02 1.15044519e-01 3.76486033e-02 1.07551610e+00
-2.21853435e-01 -1.87101021e-01 -1.42890131e+00 7.40470767e-01
-2.23151636e+00 -6.04272723e-01 -3.90780047e-02 1.83967078e+00
8.93331528e-01 3.71319018e-02 1.19239591e-01 1.40134960e-01
3.18083912e-01 -1.49168655e-01 -7.22303092e-01 2.40497291e-01
3.42248648e-01 -1.13893613e-01 7.18976185e-02 2.10774213e-01
-1.07726455e+00 4.64511991e-01 5.60391712e+00 5.08843482e-01
-1.31861866e+00 3.34204227e-01 1.02193630e+00 -1.65115952e-01
-7.45309293e-02 -1.35871932e-01 -3.14156294e-01 4.81713891e-01
1.03180683e+00 -8.95401761e-02 9.44887623e-02 4.43311065e-01
5.18262446e-01 -7.04584718e-02 -1.41659570e+00 6.53137624e-01
-2.78633714e-01 -1.26924074e+00 -1.20480187e-01 1.77726418e-01
2.97376335e-01 2.06577316e-01 2.08377510e-01 3.22719775e-02
2.95057386e-01 -1.05788648e+00 4.53482687e-01 9.77056563e-01
4.05643046e-01 -2.46307313e-01 8.07761729e-01 4.58829820e-01
-9.50044513e-01 -1.96092546e-01 2.26291180e-01 9.20398161e-02
2.94966370e-01 8.58181059e-01 -6.98561549e-01 7.08223999e-01
6.18057013e-01 9.43998635e-01 -3.53555292e-01 1.13130891e+00
-1.75038710e-01 8.08974743e-01 -1.28384337e-01 3.97191763e-01
1.51625127e-01 -2.07596016e-03 2.63468504e-01 7.11140990e-01
4.52672660e-01 2.99226999e-01 2.30061218e-01 5.28743148e-01
7.34504685e-02 2.07876489e-02 -3.29919219e-01 1.79102588e-02
1.54758275e-01 1.24745524e+00 -5.60844660e-01 -4.19632196e-01
-3.42077881e-01 7.25488067e-01 3.01619828e-01 2.65128195e-01
-9.62084472e-01 3.17191184e-01 4.58073348e-01 4.17722017e-01
1.55710578e-01 1.04871340e-01 -2.68950969e-01 -1.11842966e+00
7.24103227e-02 -1.10064816e+00 6.62973166e-01 -7.67880738e-01
-1.47282481e+00 6.00433946e-01 -3.34963173e-01 -1.36459565e+00
-4.24833298e-01 -6.25492156e-01 -4.58019674e-01 4.26836282e-01
-1.43201578e+00 -1.53297448e+00 -2.29424223e-01 6.55136287e-01
3.02770466e-01 -1.23858638e-03 8.65798771e-01 5.29040873e-01
-5.17941535e-01 8.78482014e-02 5.87807670e-02 2.23318517e-01
5.48905015e-01 -1.07672501e+00 -1.92464128e-01 4.27208155e-01
-2.46566623e-01 2.97429562e-01 5.74736238e-01 -6.62463129e-01
-1.40507531e+00 -9.31680560e-01 8.15381527e-01 -1.83561370e-01
1.04076934e+00 1.51132941e-01 -7.93205738e-01 7.73771942e-01
-8.27355459e-02 2.59287328e-01 7.69531667e-01 -9.02266894e-03
-4.08922359e-02 -1.64253116e-01 -7.66975939e-01 3.45871180e-01
7.53602147e-01 -3.17869365e-01 -4.15092409e-01 5.93378723e-01
3.67729783e-01 -3.87806177e-01 -1.42170179e+00 6.26159072e-01
7.28110671e-01 -7.24439323e-01 1.05477870e+00 -8.89107227e-01
1.03199959e+00 -1.24600446e-02 -2.02368945e-02 -1.35752702e+00
-2.60653734e-01 -1.09995566e-01 -3.58794510e-01 5.65159619e-01
3.90016168e-01 -6.21395528e-01 6.44229949e-01 4.62101519e-01
-9.15089846e-02 -1.52331805e+00 -9.91704881e-01 -5.11871815e-01
1.91016495e-01 -2.58970678e-01 2.40318447e-01 9.99942005e-01
1.04446419e-01 3.50156933e-01 -4.97767866e-01 4.20591354e-01
6.00442410e-01 3.13022524e-01 2.25592449e-01 -1.40498042e+00
-5.96843243e-01 -3.25486779e-01 -2.85473108e-01 -7.42877781e-01
-7.03780875e-02 -7.70026267e-01 -1.78642601e-01 -1.77033412e+00
5.92098296e-01 -5.50548673e-01 -6.74932182e-01 6.43396139e-01
-1.39231384e-01 -6.40585274e-02 -5.29682823e-02 6.01333678e-01
-4.60835457e-01 6.13715827e-01 1.25848007e+00 -1.72090322e-01
2.00409710e-01 1.91587180e-01 -6.07561886e-01 7.74551451e-01
7.83446550e-01 -6.31447971e-01 -4.49669361e-01 -5.58632731e-01
2.35502630e-01 6.47574484e-01 7.95676589e-01 -9.19335842e-01
4.35278445e-01 -1.75713692e-02 1.46238178e-01 -3.64950031e-01
5.11639595e-01 -7.01688647e-01 3.28396827e-01 9.54824150e-01
-3.03174376e-01 1.53993830e-01 -1.89793423e-01 5.16569674e-01
-3.44096184e-01 7.85466284e-02 5.61620593e-01 -1.44788653e-01
-2.41166547e-01 5.54426789e-01 -3.49139720e-01 -4.79150191e-02
1.03802872e+00 4.54274058e-01 -3.38571399e-01 -4.37903464e-01
-1.14854789e+00 4.84731376e-01 -8.57992321e-02 2.21546724e-01
6.03198111e-01 -1.22751462e+00 -9.60262358e-01 -3.53453428e-01
-8.03716108e-02 -4.72013727e-02 5.47827542e-01 1.63777506e+00
-2.41652414e-01 3.09883058e-01 -4.35235314e-02 -7.62588739e-01
-1.12480605e+00 4.58672732e-01 6.91551864e-01 -9.67566431e-01
-7.40315020e-01 6.88248813e-01 3.99082243e-01 -2.13580206e-01
2.82489240e-01 -1.83792442e-01 -2.86332905e-01 9.97082591e-02
3.78064871e-01 1.33846492e-01 1.85075477e-01 -3.09930891e-01
-4.96447980e-01 4.00227308e-01 -3.64596635e-01 -5.26251793e-02
1.77399337e+00 -1.71239749e-02 -1.19581901e-01 5.47292888e-01
9.12867129e-01 -5.64607680e-01 -7.29946554e-01 -4.07575935e-01
2.56025279e-03 1.89423338e-01 2.98650473e-01 -1.03339219e+00
-1.34953749e+00 9.44747746e-01 7.67051578e-01 1.92324802e-01
1.26571286e+00 3.35046023e-01 9.09498334e-01 4.71095592e-02
2.87284017e-01 -6.83440208e-01 2.33884528e-01 1.87889680e-01
1.01554453e+00 -1.25686681e+00 7.87049308e-02 -2.50359267e-01
-7.91324854e-01 8.84041548e-01 3.51249695e-01 -1.00502238e-01
9.51326370e-01 2.30573922e-01 1.65925607e-01 -5.51711321e-01
-1.22049737e+00 4.44141738e-02 3.98239493e-01 1.17189981e-01
4.68672454e-01 2.26443708e-01 -3.83899808e-01 9.82872248e-01
8.97919945e-03 4.18753624e-01 3.01764786e-01 8.88075650e-01
-1.51114305e-02 -1.17490494e+00 -1.58761621e-01 7.20606387e-01
-6.43548310e-01 -8.40441417e-03 -6.91067800e-03 5.37361622e-01
-1.19159244e-01 6.03795230e-01 -1.52107939e-01 -4.50140804e-01
2.04479039e-01 -2.93775219e-02 3.16706270e-01 -3.60942721e-01
-4.88280624e-01 3.09223950e-01 9.31478217e-02 -7.99965858e-01
-4.85838294e-01 -7.93830514e-01 -1.11530519e+00 -1.27609059e-01
1.07140027e-01 1.11221254e-03 3.75668854e-01 9.96147513e-01
5.08278012e-01 9.61029291e-01 5.52494824e-01 -5.37796855e-01
-7.43158102e-01 -9.21964347e-01 -5.35128534e-01 1.81613594e-01
4.49615926e-01 -8.72627378e-01 -1.17550924e-01 1.06852546e-01] | [14.846692085266113, -1.9904801845550537] |
fea10599-c3f9-41b8-9fa0-1b843fd41409 | olkavs-an-open-large-scale-korean-audio | 2301.06375 | null | https://arxiv.org/abs/2301.06375v1 | https://arxiv.org/pdf/2301.06375v1.pdf | OLKAVS: An Open Large-Scale Korean Audio-Visual Speech Dataset | Inspired by humans comprehending speech in a multi-modal manner, various audio-visual datasets have been constructed. However, most existing datasets focus on English, induce dependencies with various prediction models during dataset preparation, and have only a small number of multi-view videos. To mitigate the limitations, we recently developed the Open Large-scale Korean Audio-Visual Speech (OLKAVS) dataset, which is the largest among publicly available audio-visual speech datasets. The dataset contains 1,150 hours of transcribed audio from 1,107 Korean speakers in a studio setup with nine different viewpoints and various noise situations. We also provide the pre-trained baseline models for two tasks, audio-visual speech recognition and lip reading. We conducted experiments based on the models to verify the effectiveness of multi-modal and multi-view training over uni-modal and frontal-view-only training. We expect the OLKAVS dataset to facilitate multi-modal research in broader areas such as Korean speech recognition, speaker recognition, pronunciation level classification, and mouth motion analysis. | ['Hyung-Min Park', 'Rae-Hong Park', 'Jun Hwan Ahn', 'Seung-Hyun Lee', 'Kwanghee Choi', 'Jung-Wook Hwang', 'Jeongkyun Park'] | 2023-01-16 | null | null | null | null | ['speaker-recognition', 'audio-visual-speech-recognition'] | ['speech', 'speech'] | [-2.59118021e-01 -3.59494060e-01 -3.60569000e-01 -5.79049230e-01
-1.30453277e+00 -4.67936069e-01 3.95099491e-01 -4.34174567e-01
-4.28342745e-02 1.75560579e-01 8.19438577e-01 -1.94500551e-01
4.83014822e-01 3.38908359e-02 -6.24919474e-01 -6.18293703e-01
2.94997424e-01 4.98951524e-02 5.78885563e-02 1.34950101e-01
8.58580917e-02 2.32936606e-01 -1.82477820e+00 9.19466853e-01
3.64098221e-01 9.57177341e-01 3.54086816e-01 9.77741659e-01
3.18505675e-01 5.95888317e-01 -5.09064198e-01 -3.94589305e-01
-2.59338737e-01 -1.41110256e-01 -6.40266001e-01 5.24718583e-01
6.40382290e-01 -5.60639441e-01 -3.91960680e-01 7.92768657e-01
1.05084634e+00 9.21719596e-02 4.75325108e-01 -1.52907145e+00
-1.01879072e+00 4.96365100e-01 -3.00053060e-01 2.06608281e-01
8.09763253e-01 4.35045004e-01 9.44363475e-01 -1.45150459e+00
4.82727617e-01 1.50403082e+00 2.92610139e-01 8.49644959e-01
-8.12301874e-01 -7.01632798e-01 3.37236643e-01 6.06783152e-01
-1.32716203e+00 -1.30156779e+00 1.00860810e+00 -3.41335088e-01
1.09460878e+00 2.97917217e-01 6.42557561e-01 1.78472996e+00
-2.96427459e-01 1.18642557e+00 1.00929618e+00 -3.19265395e-01
-1.15612894e-01 3.43663484e-01 -1.55666605e-01 3.39744210e-01
-5.58268905e-01 7.01192245e-02 -9.49830532e-01 -1.00529743e-02
1.50564402e-01 -3.49860281e-01 -6.36780500e-01 -3.03575337e-01
-1.57503629e+00 7.85300195e-01 -1.67010725e-01 5.76143973e-02
9.59735960e-02 -4.92533565e-01 7.65078247e-01 2.34324723e-01
3.69161636e-01 -4.28699911e-01 -7.21829116e-01 -2.86262721e-01
-7.76320457e-01 -2.34101206e-01 4.62527007e-01 1.14852464e+00
2.09903806e-01 3.83012772e-01 1.51207149e-01 1.36565447e+00
7.88851321e-01 8.50357056e-01 8.09275866e-01 -8.93527865e-01
7.81534016e-01 7.50284493e-02 -2.24638015e-01 -5.64667702e-01
-3.59620005e-01 2.74061263e-01 -6.59114897e-01 -2.93036759e-01
1.87489793e-01 9.06605572e-02 -7.98571587e-01 1.82137573e+00
3.45559895e-01 5.73622249e-02 3.97798151e-01 8.97536874e-01
1.53208137e+00 8.76272261e-01 -6.53868541e-02 -6.20986104e-01
1.43287981e+00 -1.23678243e+00 -9.34268713e-01 -6.40099943e-02
4.00498807e-01 -1.06425571e+00 1.65153027e+00 7.03637779e-01
-1.06543541e+00 -8.69610786e-01 -6.94680095e-01 -2.55405426e-01
-2.01458842e-01 7.06987917e-01 3.53227317e-01 7.22570837e-01
-1.09913886e+00 -4.14679080e-01 -6.30446374e-01 -2.90708572e-01
1.42030850e-01 -1.54630505e-02 -6.97596848e-01 -5.26752733e-02
-1.12096214e+00 4.65126276e-01 -4.47811484e-02 1.77629575e-01
-1.49888206e+00 -4.08778131e-01 -1.08931363e+00 -2.80904919e-01
2.86918491e-01 -3.60714853e-01 1.42387807e+00 -9.56242442e-01
-1.71271467e+00 1.02151978e+00 -6.59559548e-01 4.78590764e-02
2.69397974e-01 -5.15181236e-02 -9.66479599e-01 2.87865132e-01
1.20768580e-03 9.19012368e-01 1.13360417e+00 -1.20811188e+00
-2.55453318e-01 -6.06374264e-01 -1.58316508e-01 2.25117519e-01
-4.16990638e-01 3.33980441e-01 -5.77445805e-01 -8.78539741e-01
-9.43110138e-03 -8.94318700e-01 5.99818110e-01 -2.16206208e-01
-5.42429745e-01 -3.88297707e-01 8.83041739e-01 -9.09632564e-01
9.69588578e-01 -2.57665563e+00 1.87271282e-01 -5.37010610e-01
-1.55126289e-01 1.98455662e-01 -3.45233858e-01 3.12146395e-01
-2.62753546e-01 7.51198158e-02 3.34735543e-01 -6.67251706e-01
-1.28243446e-01 -5.30148819e-02 -3.94487053e-01 3.84912372e-01
-2.43352860e-01 4.61769521e-01 -5.32743573e-01 -7.57737935e-01
1.56328306e-01 6.50121987e-01 -6.72302186e-01 3.49219203e-01
9.72506218e-03 3.52790713e-01 -2.09805652e-01 1.21599042e+00
6.10299587e-01 -1.28608271e-01 -1.42156810e-01 -4.43479776e-01
8.81971940e-02 4.02772069e-01 -1.00793397e+00 1.89561605e+00
-6.57089233e-01 9.98390913e-01 4.70777988e-01 -7.96836972e-01
5.83141029e-01 8.95196319e-01 2.91490227e-01 -4.94619548e-01
3.04494482e-02 6.99114725e-02 -2.16167256e-01 -1.05980563e+00
4.64884341e-01 1.89472735e-02 2.30561137e-01 1.46251723e-01
2.28241757e-01 -2.14245301e-02 -5.72080612e-02 2.11499408e-01
3.04071456e-01 -2.06871003e-01 1.82851497e-02 2.45315880e-01
8.86682212e-01 -4.41137850e-01 5.50713420e-01 9.76505131e-02
-7.50802934e-01 9.08628404e-01 3.03301066e-01 -6.22798502e-02
-6.31017983e-01 -1.24325418e+00 -1.41125217e-01 1.38325059e+00
-1.41031787e-01 -6.00707054e-01 -7.13483691e-01 -5.27413487e-01
-3.67054999e-01 5.63290954e-01 -2.22434923e-01 1.20359518e-01
-2.46140316e-01 -3.23247939e-01 7.34126389e-01 6.75256968e-01
3.59762847e-01 -1.24806190e+00 1.21644340e-01 -2.46223047e-01
-8.25398266e-01 -1.60320985e+00 -9.49267089e-01 -3.54584038e-01
-5.66793501e-01 -1.06550825e+00 -8.39047253e-01 -1.19052970e+00
3.53812456e-01 4.40640152e-01 7.64551699e-01 -4.80822861e-01
-9.05307010e-02 8.48804951e-01 -4.12579209e-01 -3.80868495e-01
-5.33513129e-01 -2.69288719e-01 5.79557598e-01 2.12447569e-01
3.19483131e-01 -2.39314362e-01 -3.68607968e-01 6.51208401e-01
-4.61900681e-01 1.25676412e-02 3.79309326e-01 8.74216616e-01
5.56095958e-01 -3.14155161e-01 8.20742488e-01 3.39526236e-02
5.82892656e-01 -1.08791120e-01 -3.36751193e-01 3.22839439e-01
-6.30903095e-02 -5.20567179e-01 3.69859308e-01 -9.30915475e-01
-1.14583075e+00 7.01854238e-03 -5.23761749e-01 -7.40025520e-01
-5.00803113e-01 2.44005531e-01 -8.35351646e-01 3.36760759e-01
1.80447489e-01 6.11693859e-01 2.01771840e-01 -4.92392212e-01
3.41115564e-01 1.44271445e+00 4.69009817e-01 -2.12957039e-01
2.17629880e-01 3.30230504e-01 -7.42093623e-01 -1.48841608e+00
-4.74215537e-01 -5.03300667e-01 -5.72176933e-01 -3.93292755e-01
1.15576041e+00 -1.58639228e+00 -1.02248466e+00 9.08411324e-01
-1.22081745e+00 -2.03159168e-01 3.06614071e-01 8.55995834e-01
-6.15557194e-01 5.53702652e-01 -5.78194141e-01 -8.84740710e-01
-6.51510805e-02 -1.73921287e+00 1.31119788e+00 -2.41932962e-02
3.12888324e-02 -7.09700882e-01 -1.08485222e-01 1.16885769e+00
8.74853730e-02 -6.59003079e-01 7.17685759e-01 -5.89192033e-01
-2.80953974e-01 2.11917982e-01 1.70154676e-01 6.43995464e-01
2.03980178e-01 2.68360913e-01 -1.53816164e+00 -3.79767478e-01
-3.14952470e-02 -9.73927796e-01 7.74364233e-01 6.71518862e-01
1.38554060e+00 -8.74882266e-02 -2.11956814e-01 4.25833285e-01
4.82031912e-01 4.22150582e-01 4.54000980e-01 -1.09903060e-01
6.09717071e-01 7.26143718e-01 6.84366763e-01 3.81393969e-01
8.03547621e-01 9.05061424e-01 3.91140848e-01 1.43704742e-01
-1.76765352e-01 -3.70744884e-01 9.95166123e-01 1.65447998e+00
2.29820177e-01 -4.35307562e-01 -5.99256456e-01 7.73220301e-01
-1.13493073e+00 -1.16316199e+00 2.14812353e-01 1.91142178e+00
7.26891279e-01 -1.54056549e-01 4.85691279e-01 2.09388822e-01
7.14368284e-01 4.81981099e-01 -4.85278517e-01 -3.56514275e-01
-2.00737163e-01 -5.09848893e-01 -2.84881085e-01 5.08430779e-01
-1.23542809e+00 8.70778382e-01 6.28910732e+00 6.81550145e-01
-1.50819564e+00 1.25856057e-01 6.25864565e-01 -3.83510649e-01
-2.85557032e-01 -3.53141218e-01 -1.02624738e+00 2.56879836e-01
1.06863427e+00 2.68022925e-01 3.91240507e-01 1.02573717e+00
5.46645343e-01 2.34894156e-01 -1.05265164e+00 1.49643826e+00
7.26463199e-01 -9.86849368e-01 1.53220147e-01 -9.85353440e-03
1.93668112e-01 2.82458097e-01 2.62067109e-01 3.36952060e-01
-4.40808356e-01 -1.00729930e+00 8.24703872e-01 1.22444063e-01
9.65501487e-01 -4.62776333e-01 1.67715922e-01 4.67119008e-01
-1.39631236e+00 -1.34987518e-01 -1.66877568e-01 4.76014405e-01
3.84984732e-01 -5.52183278e-02 -8.44662368e-01 3.39467674e-01
1.01217890e+00 8.58529747e-01 -5.47541797e-01 5.28670669e-01
-1.36297266e-03 7.00220346e-01 -6.50834888e-02 -1.07452776e-02
-2.29815677e-01 2.86520511e-01 5.96165121e-01 9.95472312e-01
2.34374791e-01 -1.15088537e-01 1.18383234e-02 3.37361902e-01
-2.86447853e-02 2.84166813e-01 -6.77161455e-01 -3.44961375e-01
5.99864900e-01 1.17210877e+00 -2.71286488e-01 -2.13767469e-01
-9.89282668e-01 5.70288241e-01 -9.48703662e-02 4.77136284e-01
-1.03556180e+00 -5.55712217e-03 8.04237664e-01 3.75935175e-02
3.91003460e-01 -2.68393934e-01 1.98854208e-01 -1.58892763e+00
1.40217975e-01 -1.33677268e+00 3.04577529e-01 -1.38230073e+00
-1.15053391e+00 8.33272755e-01 -1.20335026e-03 -1.46351302e+00
-4.10898626e-01 -9.26014721e-01 -3.18447828e-01 4.72403765e-01
-1.44883263e+00 -1.38242614e+00 -2.38621041e-01 1.07073569e+00
1.38214922e+00 -6.52650952e-01 8.29754353e-01 4.88153785e-01
-7.14005530e-01 8.66294920e-01 -2.72825778e-01 1.96509436e-01
1.13065052e+00 -5.92709243e-01 1.44689260e-02 4.51143801e-01
4.19571280e-01 5.50818205e-01 3.01480174e-01 -2.63668239e-01
-1.64086223e+00 -7.63206482e-01 9.13564563e-01 -4.22544956e-01
4.77470040e-01 -5.71366191e-01 -7.49813139e-01 9.13271487e-01
6.46327198e-01 1.14181772e-01 1.02922189e+00 -5.11258245e-02
-3.47245544e-01 -1.89612195e-01 -7.10753679e-01 3.65912825e-01
8.97088349e-01 -1.03164649e+00 -6.64196432e-01 2.09047645e-01
8.66498828e-01 -4.34923798e-01 -8.71984959e-01 3.96160632e-01
7.60011375e-01 -9.18267429e-01 1.21250975e+00 -7.33136475e-01
2.72371858e-01 -1.16813868e-01 -6.33072317e-01 -1.38951504e+00
2.74030030e-01 -4.54612017e-01 1.20244041e-01 1.67335451e+00
4.54677165e-01 -4.08546805e-01 3.76729161e-01 -3.40648852e-02
-3.74106735e-01 -4.27949578e-01 -1.19268358e+00 -5.65686405e-01
-1.82525918e-01 -7.48094082e-01 3.49189788e-01 9.97722507e-01
4.95248139e-02 5.07697880e-01 -7.12803602e-01 3.50108057e-01
3.12360287e-01 9.09266323e-02 1.00829303e+00 -9.95078743e-01
-2.01435670e-01 -2.33346540e-02 -1.92152560e-01 -1.29437137e+00
6.56665027e-01 -7.11598754e-01 -1.54626340e-01 -1.23176575e+00
2.32732520e-01 2.64558762e-01 1.57260031e-01 4.07827049e-01
1.34219229e-01 -1.42187372e-01 -8.92614853e-03 1.86069205e-01
-5.47512174e-01 8.11713755e-01 1.49220908e+00 -3.94200623e-01
-2.02525318e-01 1.94728777e-01 -4.85696942e-01 9.04963374e-01
3.64719242e-01 -6.48626462e-02 -7.46782720e-01 -5.41542768e-01
-3.21401745e-01 5.76011360e-01 4.24716532e-01 -4.48079705e-01
2.14376181e-01 -2.00669199e-01 3.42989296e-01 -1.05150533e+00
1.02755821e+00 -5.19851089e-01 -2.26979271e-01 -3.32624055e-02
-4.48579311e-01 2.62741566e-01 3.00222248e-01 4.48202521e-01
-7.44005263e-01 2.74183303e-01 7.49479473e-01 -2.45946757e-02
-8.10454011e-01 2.56895900e-01 -4.79136914e-01 3.64365816e-01
9.65694368e-01 -1.03515722e-01 -4.29851711e-01 -7.93850005e-01
-1.19940746e+00 7.38134012e-02 1.39466003e-01 8.90261233e-01
1.06076717e+00 -1.48814487e+00 -5.72110951e-01 5.58715224e-01
3.85240734e-01 -2.76085407e-01 6.71826482e-01 9.80018735e-01
-3.02699991e-02 6.60058916e-01 -3.24207246e-02 -9.62762058e-01
-1.90636396e+00 6.47511542e-01 2.63055623e-01 5.04454672e-01
-4.97980446e-01 8.98213804e-01 3.59218091e-01 -5.66421092e-01
6.93065524e-01 -4.08104867e-01 -3.42672884e-01 3.04627508e-01
6.79349482e-01 3.23007613e-01 -4.17525396e-02 -1.30231667e+00
-6.05315804e-01 5.12280226e-01 -8.14793408e-02 -3.36405843e-01
1.08049512e+00 -5.56021988e-01 2.95833707e-01 8.14618111e-01
1.36542845e+00 4.48504359e-01 -9.87133980e-01 -7.47890305e-03
-7.11869597e-01 -4.43530321e-01 6.26350492e-02 -5.06666601e-01
-1.10068595e+00 1.47358871e+00 6.04480982e-01 1.07037388e-02
1.28869975e+00 2.45498985e-01 7.07627952e-01 1.81395471e-01
1.39542699e-01 -1.16581917e+00 6.85191810e-01 3.56153667e-01
1.31501579e+00 -1.50208700e+00 -3.84661049e-01 -4.14138734e-01
-1.32340431e+00 1.12819111e+00 8.93509924e-01 8.13929260e-01
8.56130898e-01 1.42267913e-01 6.37631297e-01 1.47075489e-01
-1.06878912e+00 -4.39794511e-02 2.41880745e-01 8.71527016e-01
6.36612773e-01 -8.82600807e-03 4.62352097e-01 8.44460130e-01
-3.55092525e-01 -2.50779986e-01 4.48759615e-01 4.64049220e-01
-1.84001490e-01 -8.74210715e-01 -4.33398157e-01 -1.30143076e-01
-4.92989451e-01 -1.01269774e-01 -5.25432646e-01 8.41139436e-01
-3.38287413e-01 1.33949888e+00 -1.41573861e-01 -5.58104396e-01
3.52602929e-01 3.59381616e-01 3.88718665e-01 -3.12998056e-01
-8.48837271e-02 6.03042841e-01 9.79115665e-02 -4.13481921e-01
-6.35219634e-01 -9.68269050e-01 -1.03210974e+00 -3.39377970e-02
-3.40923637e-01 -1.43495589e-01 5.80604136e-01 8.18251610e-01
3.99528712e-01 3.65057737e-01 7.82129526e-01 -9.12323296e-01
-6.54644668e-01 -1.29351258e+00 -5.26255250e-01 2.34090298e-01
6.34629786e-01 -8.01901519e-01 -7.15614378e-01 2.90883005e-01] | [14.283489227294922, 5.070558071136475] |
52cd30c3-3ea8-4697-a881-7df75eb8f7e6 | deep-transport-network-for-unsupervised-video | null | null | http://openaccess.thecvf.com//content/ICCV2021/html/Zhang_Deep_Transport_Network_for_Unsupervised_Video_Object_Segmentation_ICCV_2021_paper.html | http://openaccess.thecvf.com//content/ICCV2021/papers/Zhang_Deep_Transport_Network_for_Unsupervised_Video_Object_Segmentation_ICCV_2021_paper.pdf | Deep Transport Network for Unsupervised Video Object Segmentation | The popular unsupervised video object segmentation methods fuse the RGB frame and optical flow via a two-stream network. However, they cannot handle the distracting noises in each input modality, which may vastly deteriorate the model performance. We propose to establish the correspondence between the input modalities while suppressing the distracting signals via optimal structural matching. Given a video frame, we extract the dense local features from the RGB image and optical flow, and treat them as two complex structured representations. The Wasserstein distance is then employed to compute the global optimal flows to transport the features in one modality to the other, where the magnitude of each flow measures the extent of the alignment between two local features. To plug the structural matching into a two-stream network for end-to-end training, we factorize the input cost matrix into small spatial blocks and design a differentiable long-short Sinkhorn module consisting of a long-distant Sinkhorn layer and a short-distant Sinkhorn layer. We integrate the module into a dedicated two-stream network and dub our model TransportNet. Our experiments show that aligning motion-appearance yields the state-of-the-art results on the popular video object segmentation datasets. | ['Bo Liu', 'Qingshan Liu', 'Dong Liu', 'Zicheng Zhao', 'Kaihua Zhang'] | 2021-01-01 | null | null | null | iccv-2021-1 | ['unsupervised-video-object-segmentation'] | ['computer-vision'] | [ 1.76917717e-01 -2.24980295e-01 -1.84241682e-01 -5.22939622e-01
-5.47375739e-01 -5.27043283e-01 2.64903247e-01 -4.52820390e-01
-6.29197717e-01 3.20484847e-01 -4.25450280e-02 4.71950844e-02
-5.68360835e-02 -6.20831370e-01 -8.64065170e-01 -8.14355612e-01
2.08454773e-01 -1.28164485e-01 4.95798886e-01 1.97104484e-01
1.73473909e-01 2.85417378e-01 -1.41335905e+00 2.32095331e-01
1.00475121e+00 1.43955004e+00 2.47625366e-01 6.65913165e-01
-3.02007407e-01 1.04361558e+00 -1.41932502e-01 -2.52071679e-01
5.38983881e-01 -7.03971207e-01 -9.53191102e-01 3.80068272e-01
8.03293586e-01 -5.49670041e-01 -6.59865499e-01 1.18915606e+00
2.08517775e-01 4.15929109e-01 3.28449309e-01 -1.47026134e+00
-6.12397552e-01 2.05254301e-01 -6.75311029e-01 4.00269330e-01
1.15676790e-01 5.92700601e-01 8.94787848e-01 -8.07873070e-01
7.29185343e-01 1.35439456e+00 4.62348670e-01 5.23768187e-01
-1.20586288e+00 -4.11698967e-01 4.54235196e-01 1.52203858e-01
-1.02150011e+00 -5.12396276e-01 9.70566869e-01 -5.31449258e-01
4.72226322e-01 7.62992678e-03 8.17659676e-01 8.30554187e-01
-1.54591739e-01 1.12836981e+00 6.93628252e-01 2.27787957e-01
-1.14018805e-02 -2.56876022e-01 1.40097747e-02 9.23693478e-01
-6.01464920e-02 1.71247404e-02 -6.43725812e-01 4.29360598e-01
1.07962656e+00 3.29418570e-01 -4.53472465e-01 -4.99400437e-01
-1.23164475e+00 6.02212369e-01 8.64391863e-01 1.18675418e-01
-1.68945149e-01 4.17081684e-01 2.80092180e-01 2.39802882e-01
3.93701226e-01 8.49624351e-02 -4.05244499e-01 -7.96389803e-02
-1.05847728e+00 6.28075451e-02 4.26230043e-01 9.52143252e-01
1.17067873e+00 -2.32701376e-02 -1.69942230e-01 5.64131558e-01
6.27157569e-01 5.05921304e-01 1.88558921e-01 -1.47453189e+00
8.04077506e-01 6.02385700e-01 -3.82625088e-02 -9.85265732e-01
-1.79695904e-01 -1.13326788e-01 -7.71574199e-01 7.64485449e-02
7.14890718e-01 -1.17574714e-01 -9.59173858e-01 1.74820185e+00
4.61260855e-01 5.41880488e-01 -1.83195859e-01 1.50113571e+00
6.23707712e-01 8.00687015e-01 1.13030396e-01 -1.46091878e-01
9.56854463e-01 -1.45391285e+00 -6.02496207e-01 -2.82797933e-01
4.19223011e-01 -6.66510999e-01 9.63238418e-01 4.20380384e-02
-1.44322979e+00 -6.99096262e-01 -9.02216613e-01 -6.72422111e-01
-2.03974590e-01 -8.07745904e-02 5.08077323e-01 1.59838885e-01
-8.00583959e-01 9.11767840e-01 -1.21421373e+00 9.11681261e-03
7.53931105e-01 2.70989329e-01 -4.73837912e-01 -2.71997809e-01
-1.00325835e+00 4.81747121e-01 1.56354889e-01 6.32999122e-01
-8.44050705e-01 -8.10510755e-01 -1.08202755e+00 -1.21501558e-01
3.61330718e-01 -9.15876031e-01 7.71239877e-01 -1.25816834e+00
-1.58611047e+00 6.72258198e-01 -3.04309934e-01 6.28309473e-02
8.34503531e-01 -1.99455470e-01 2.83028539e-02 4.52954292e-01
2.58693963e-01 9.34273779e-01 1.03047574e+00 -1.03753960e+00
-8.62494469e-01 -3.12329859e-01 -4.57052403e-04 2.60985523e-01
-2.55962640e-01 -3.11584249e-02 -8.83660495e-01 -6.21418774e-01
2.89866984e-01 -6.60578668e-01 -1.40283123e-01 6.27468526e-01
-2.90084809e-01 -3.91774885e-02 1.00925124e+00 -5.82813323e-01
9.48248923e-01 -2.28607893e+00 5.23673832e-01 3.88721786e-02
2.19338074e-01 1.99985370e-01 -4.45460498e-01 -3.06467205e-01
7.13708401e-02 -3.44540924e-02 -4.68049228e-01 -5.17665744e-01
-2.51212388e-01 3.55596393e-01 -3.49406786e-02 7.86258340e-01
5.89911997e-01 1.11644244e+00 -1.34089625e+00 -5.52777231e-01
3.32968831e-01 4.94827956e-01 -7.21186459e-01 5.68210959e-01
-1.46194488e-01 6.40052736e-01 -6.57385886e-01 5.54063618e-01
8.71642828e-01 -1.73098817e-01 -3.46856892e-01 -4.24091905e-01
-1.49230525e-01 2.33103216e-01 -1.17272460e+00 2.27461648e+00
-4.61468361e-02 6.13109231e-01 2.50713289e-01 -1.02069116e+00
5.64874530e-01 -6.85389489e-02 8.02668035e-01 -6.59406126e-01
2.53655106e-01 2.56400615e-01 -1.28549397e-01 -9.90447760e-01
1.36237472e-01 -1.74196735e-02 2.68840283e-01 2.74965018e-01
3.33309382e-01 -4.01844941e-02 4.05839026e-01 1.32342890e-01
8.52904022e-01 4.39678967e-01 -4.89803612e-01 -9.01471004e-02
7.48982131e-01 -3.99454176e-01 7.66672790e-01 3.72212708e-01
-5.30541003e-01 1.03893554e+00 6.85419321e-01 -4.89630818e-01
-7.80210495e-01 -1.23953247e+00 2.69998938e-01 1.05740309e+00
6.70098245e-01 -8.14291015e-02 -6.57475770e-01 -1.00955915e+00
6.11337572e-02 2.11179927e-02 -6.70668900e-01 -2.73683846e-01
-6.89227045e-01 -3.93662095e-01 4.41567987e-01 6.88028872e-01
8.56017172e-01 -9.03243005e-01 -7.15617418e-01 -1.89080838e-05
-4.65616137e-01 -1.26720679e+00 -1.07074809e+00 9.22103226e-02
-8.21609616e-01 -1.14243662e+00 -8.06765974e-01 -8.70226562e-01
7.52722561e-01 4.94247258e-01 7.20146716e-01 1.55450031e-01
-1.93753075e-02 1.63737491e-01 -8.93960968e-02 3.02193463e-01
2.25750983e-01 1.12193264e-01 -3.28033715e-01 5.34419358e-01
6.76365718e-02 -4.29185987e-01 -9.94966745e-01 3.43629390e-01
-1.22258186e+00 4.34399992e-02 1.80218488e-01 7.21049547e-01
5.17373919e-01 -6.12335861e-01 1.32330328e-01 -5.08539021e-01
1.33996055e-01 -3.04542929e-01 -5.96451819e-01 8.73988941e-02
-1.77300453e-01 1.95908338e-01 4.79253948e-01 -4.01789278e-01
-8.81597817e-01 4.31059718e-01 -4.99735922e-02 -9.57889378e-01
5.40303476e-02 2.29891613e-01 -4.10523027e-01 -1.38625637e-01
9.30659398e-02 4.45717461e-02 9.27747712e-02 -3.13237995e-01
8.08165729e-01 3.88979435e-01 7.81310499e-01 -4.43381041e-01
9.14441168e-01 7.62820542e-01 -1.90036431e-01 -5.38853884e-01
-1.02801907e+00 -5.97631574e-01 -8.34500611e-01 -4.35011089e-01
1.25845933e+00 -6.90814435e-01 -7.32233763e-01 7.70963132e-01
-1.29957640e+00 -3.95071924e-01 -4.99014467e-01 3.96783233e-01
-6.16274059e-01 2.81048089e-01 -6.32637918e-01 -4.74675268e-01
-4.19739336e-02 -1.46048963e+00 1.14260530e+00 5.70692599e-01
3.30844283e-01 -8.55603099e-01 -1.01928920e-01 3.20053220e-01
2.41195038e-01 1.23060241e-01 5.27404964e-01 -8.78784359e-02
-9.45818901e-01 1.70026690e-01 -6.74763501e-01 7.30160415e-01
1.25198737e-01 1.48874283e-01 -1.03375411e+00 -1.09779105e-01
6.47469386e-02 -3.56107503e-01 1.27424085e+00 4.00603265e-01
1.16640270e+00 -6.87036663e-02 -7.28651956e-02 1.31503439e+00
1.40357268e+00 -4.45270129e-02 6.50626242e-01 1.56882957e-01
1.31442058e+00 6.11057401e-01 3.36959720e-01 -6.14619143e-02
4.31736112e-01 2.54298866e-01 5.77816725e-01 -1.74694121e-01
-2.53841221e-01 -2.95734972e-01 5.52780807e-01 8.49809706e-01
-1.05655670e-01 1.75183024e-02 -5.42435467e-01 6.48336112e-01
-2.00435495e+00 -8.52832913e-01 -5.53109050e-02 1.91037512e+00
8.35965097e-01 6.66960850e-02 6.70541823e-02 -1.58165738e-01
5.77627718e-01 4.66731966e-01 -6.73158884e-01 6.09938130e-02
-2.37260178e-01 -2.18137726e-01 5.51170051e-01 6.70194626e-01
-1.19859052e+00 9.49359775e-01 5.20350122e+00 5.15648305e-01
-1.48935831e+00 1.31004393e-01 7.11116672e-01 -3.78609836e-01
-3.33534360e-01 1.48326356e-03 -2.88024783e-01 6.60482764e-01
5.17160475e-01 2.24566162e-01 5.45737565e-01 4.65743691e-01
3.92262727e-01 -4.55059931e-02 -1.34724724e+00 1.00230968e+00
-7.35824257e-02 -1.36201227e+00 -1.02857156e-02 -2.10727885e-01
8.09218466e-01 2.35849544e-01 5.74273989e-02 -1.37560919e-01
-1.81555718e-01 -7.54440308e-01 1.07876027e+00 7.39922583e-01
5.80607712e-01 -4.32378083e-01 4.76847708e-01 6.38912991e-02
-1.31489837e+00 -2.69011967e-02 -2.53444791e-01 1.65438816e-01
4.31464612e-01 3.51065546e-01 2.51273662e-01 5.36288440e-01
8.25301409e-01 1.26500833e+00 -3.84871542e-01 1.13701677e+00
-1.32458150e-01 2.94189930e-01 -4.53732222e-01 4.09187943e-01
5.59008002e-01 -6.45010114e-01 5.60462952e-01 1.08146119e+00
7.72227049e-02 5.06376587e-02 2.59741843e-01 1.08893013e+00
-3.41238678e-01 -3.54972154e-01 -3.19919109e-01 -6.53360188e-02
9.07344464e-03 1.23951614e+00 -7.41090775e-01 -3.92446458e-01
-6.93794131e-01 1.26311374e+00 3.23376328e-01 7.08724678e-01
-8.06972146e-01 -5.00871420e-01 9.05909956e-01 -8.98008198e-02
3.91736120e-01 -2.22781524e-01 -3.56393754e-01 -1.50072598e+00
2.87724644e-01 -5.16102135e-01 3.54900867e-01 -8.59806061e-01
-1.44806123e+00 3.39656532e-01 -3.21731448e-01 -1.30685246e+00
3.13720070e-02 -6.53959990e-01 -5.85564017e-01 8.84525955e-01
-1.81361616e+00 -1.03179669e+00 -5.59931815e-01 8.28618050e-01
5.38093209e-01 3.00653160e-01 7.69721344e-02 5.75446963e-01
-8.63079309e-01 3.07743758e-01 -9.40512046e-02 5.53258657e-01
7.05920041e-01 -1.07903039e+00 1.01350769e-01 1.13676596e+00
-7.08500817e-02 4.14800197e-01 1.82804346e-01 -3.01135033e-01
-1.40881240e+00 -1.27640688e+00 5.79073608e-01 -2.13793337e-01
8.34684014e-01 -3.90701771e-01 -1.15800393e+00 6.17879212e-01
1.48101151e-01 9.08732355e-01 4.53137904e-02 -5.88430464e-01
-5.08581340e-01 -3.50569993e-01 -9.16922867e-01 4.00881648e-01
1.33383811e+00 -6.95415497e-01 -4.58671004e-01 -1.29921939e-02
8.91576111e-01 -4.18358743e-01 -7.29675531e-01 1.06402948e-01
5.22777617e-01 -9.17023599e-01 8.52173984e-01 -8.39021027e-01
8.74271512e-01 -6.33787155e-01 4.14591059e-02 -1.11094034e+00
-1.82704121e-01 -8.18086982e-01 -9.25996229e-02 1.16424894e+00
2.80248493e-01 -3.89955014e-01 7.76541829e-01 7.35014737e-01
-1.75162852e-01 -7.85230100e-01 -9.26914096e-01 -4.47054088e-01
4.58445475e-02 -5.02681851e-01 2.83946037e-01 8.08960617e-01
-1.39556274e-01 3.72368693e-01 -1.79652184e-01 1.96767552e-03
6.44872367e-01 1.07859373e-01 5.31984270e-01 -8.90108109e-01
2.31425464e-02 -6.17653787e-01 -5.15420079e-01 -1.54351699e+00
1.75826877e-01 -8.82351637e-01 3.91437262e-01 -1.39734066e+00
7.83225074e-02 -2.93930978e-01 -4.65845495e-01 4.04823720e-01
-4.45295036e-01 3.21829110e-01 4.91655499e-01 2.34520376e-01
-7.30379105e-01 8.27336848e-01 1.67572832e+00 -3.74909341e-01
-2.23081350e-01 -3.11130613e-01 -5.06584883e-01 8.20258439e-01
2.63909847e-01 -4.24149334e-01 -5.03758848e-01 -1.01832151e+00
-8.56991634e-02 -6.91026226e-02 4.73867476e-01 -7.72401810e-01
4.93283212e-01 -3.06073189e-01 4.74690765e-01 -3.25930029e-01
1.33620128e-01 -9.25153732e-01 -5.15924037e-01 2.38895029e-01
-4.98520076e-01 -8.95894691e-02 -1.23472005e-01 6.43296182e-01
-3.91952485e-01 -1.00703910e-03 9.48313892e-01 2.30509136e-02
-6.00810051e-01 7.59725213e-01 3.15051526e-02 3.91960144e-01
9.01606441e-01 -2.47517735e-01 -2.74869919e-01 -1.98801998e-02
-6.90329671e-01 5.79489350e-01 3.90175879e-01 5.22997081e-01
7.46766269e-01 -1.30462325e+00 -3.88266116e-01 5.37824571e-01
-2.54470617e-01 5.52034676e-01 3.19369465e-01 1.18813193e+00
-6.79888725e-01 9.65272821e-03 -3.52996230e-01 -9.61838901e-01
-6.71223342e-01 4.52543378e-01 7.64804959e-01 1.49324551e-01
-5.96536279e-01 1.14619982e+00 3.45492810e-01 -1.28749207e-01
4.26394373e-01 -5.76061666e-01 9.10006091e-02 1.57181025e-01
4.60269332e-01 5.40561557e-01 -2.52247691e-01 -7.34747946e-01
-4.53226149e-01 8.91566575e-01 1.78617194e-01 -1.35125861e-01
1.32009125e+00 -4.31772262e-01 -3.56013298e-01 4.63281900e-01
1.84197199e+00 -4.91727889e-01 -2.08421898e+00 -1.87775522e-01
-2.41705239e-01 -7.40935028e-01 7.07489625e-02 -1.52885646e-01
-1.93795443e+00 1.13421881e+00 4.49135721e-01 1.32812247e-01
1.28323030e+00 -1.19651910e-02 1.04281020e+00 6.54618815e-02
-1.37085542e-01 -9.81275022e-01 1.41617134e-01 5.09017348e-01
4.74638551e-01 -1.24642801e+00 -2.79213130e-01 -5.42566121e-01
-5.04620790e-01 1.15238953e+00 7.12297678e-01 -3.47737581e-01
6.89257622e-01 9.07253847e-02 2.28927702e-01 -1.35061741e-02
-5.40750146e-01 -3.27689797e-01 3.50596696e-01 5.00438988e-01
1.33405149e-01 -4.45192039e-01 -3.89575809e-02 3.66752982e-01
3.43603134e-01 1.99101731e-01 2.28514135e-01 7.48488128e-01
-2.25873008e-01 -7.03996420e-01 -7.47311264e-02 1.53591394e-01
-2.14038312e-01 1.14027500e-01 -1.51528046e-01 5.67797840e-01
3.87951285e-01 7.61579335e-01 4.41191971e-01 -2.79148579e-01
4.69561249e-01 -1.41698673e-01 5.51494479e-01 -1.75472006e-01
-5.05481422e-01 3.38977307e-01 -3.99112076e-01 -1.09865010e+00
-7.77896225e-01 -6.15025222e-01 -1.48433471e+00 -1.91430360e-01
-1.09327078e-01 -2.17637375e-01 2.80057043e-01 1.11439824e+00
1.73096821e-01 5.75877309e-01 7.35838830e-01 -1.15794039e+00
-2.91575164e-01 -6.66030765e-01 -4.32344884e-01 7.18165457e-01
7.70461082e-01 -5.15950799e-01 -4.00126666e-01 4.99622881e-01] | [9.181207656860352, -0.24208083748817444] |
db2bf765-aa2c-4987-a347-866119b607b0 | understanding-bloom-an-empirical-study-on | 2211.14865 | null | https://arxiv.org/abs/2211.14865v2 | https://arxiv.org/pdf/2211.14865v2.pdf | Understanding BLOOM: An empirical study on diverse NLP tasks | We view the landscape of large language models (LLMs) through the lens of the recently released BLOOM model to understand the performance of BLOOM and other decoder-only LLMs compared to BERT-style encoder-only models. We achieve this by evaluating the smaller BLOOM model variants (\textit{350m/560m} and \textit{1b3/1b7}) on several NLP benchmark datasets and popular leaderboards. We make the following observations: (1) BLOOM performance does not scale with parameter size, unlike other LLMs like GPT and BERT. Experiments fine-tuning BLOOM models show that the 560m variant performs similarly to or better than the 1b7 variant, (2) Zero-shot cross-lingual and multi-lingual fine-tuning experiments show that BLOOM is at par or worse than monolingual GPT-2 models, and (3) Toxicity analysis of prompt-based text generation using the RealToxicityPrompts dataset shows that the text generated by BLOOM is at least 17\% less toxic than GPT-2 and GPT-3 models. | ['Preethi Raghavan', 'SaiKrishna Rallabandi', 'Parag Pravin Dakle'] | 2022-11-27 | null | null | null | null | ['few-shot-text-classification'] | ['natural-language-processing'] | [-1.50550857e-01 3.69956106e-01 -2.66962796e-01 1.23482354e-01
-1.36128318e+00 -8.41643155e-01 1.12327468e+00 1.88780665e-01
-4.57263887e-01 9.09281731e-01 6.84671283e-01 -6.95000768e-01
-1.00850008e-01 -5.53597093e-01 -1.00634372e+00 -3.62500936e-01
1.21056847e-02 9.71102953e-01 1.07349895e-01 -6.51789606e-01
-2.14579776e-02 -1.32441431e-01 -9.48151410e-01 5.78964472e-01
1.04827321e+00 4.08649087e-01 4.75876331e-01 8.08330953e-01
-2.82024503e-01 1.02509892e+00 -6.85955524e-01 -7.40037620e-01
3.21131051e-01 -4.42048460e-01 -7.85011530e-01 -5.49989820e-01
3.96117955e-01 -2.27276891e-01 -2.67462820e-01 6.85440183e-01
8.84420216e-01 -2.53547907e-01 6.65133774e-01 -1.14189219e+00
-6.61890149e-01 1.38515699e+00 -4.75714028e-01 1.02228768e-01
4.76832688e-01 5.08257806e-01 1.09248316e+00 -5.48959255e-01
7.39557207e-01 1.60872936e+00 9.47520971e-01 5.31056166e-01
-1.34504509e+00 -7.77272165e-01 1.05620787e-01 -2.93372780e-01
-1.44158947e+00 -5.86790681e-01 -3.33978474e-01 -3.91953856e-01
1.41091061e+00 3.44147980e-01 5.49914956e-01 1.59433222e+00
6.27619028e-01 1.01756299e+00 1.30078900e+00 -4.40174639e-01
-5.30425273e-02 2.59805471e-01 -1.73795804e-01 5.36528885e-01
5.32543004e-01 3.51388961e-01 -9.62386549e-01 -2.35791519e-01
2.34423995e-01 -9.54612613e-01 -2.14364171e-01 5.61895013e-01
-1.33331430e+00 8.08548391e-01 -1.73340946e-01 3.01566750e-01
-2.69282401e-01 5.39986968e-01 3.97995234e-01 4.70885307e-01
6.38710737e-01 7.75341630e-01 -5.24823129e-01 -6.54807925e-01
-1.25773525e+00 5.57965457e-01 1.08337760e+00 1.48224056e+00
5.10673881e-01 1.72221169e-01 -6.03486359e-01 7.88597763e-01
3.29981863e-01 8.07575166e-01 6.49981320e-01 -8.49097490e-01
8.35756183e-01 5.03738299e-02 1.05217889e-01 -2.48862758e-01
-5.23066521e-01 -4.40627217e-01 -4.01055127e-01 -2.65111148e-01
4.94089752e-01 -5.23221433e-01 -6.71336710e-01 1.73595905e+00
-3.32345188e-01 -3.92235905e-01 2.32357562e-01 2.20362812e-01
9.90097642e-01 1.17357922e+00 2.65607864e-01 -2.42803767e-01
1.15214622e+00 -9.05536830e-01 -6.08765841e-01 -4.52378988e-01
1.26735544e+00 -9.64172602e-01 1.26379287e+00 4.85595047e-01
-1.30880630e+00 -2.98734456e-01 -8.61830890e-01 -1.55015007e-01
-2.84467727e-01 -5.06974868e-02 5.70141196e-01 9.38110113e-01
-1.46729422e+00 2.08041549e-01 -4.49099481e-01 -6.23905063e-01
-1.31534278e-01 2.15834647e-01 -1.18232407e-01 -6.53481558e-02
-1.43356073e+00 1.17511117e+00 6.14549696e-01 -3.47791314e-01
-1.29250801e+00 -8.69995534e-01 -6.94832563e-01 -2.60401275e-02
-6.74739154e-03 -6.65395617e-01 1.56539631e+00 -4.90139276e-01
-1.56198514e+00 5.78713477e-01 -2.63100207e-01 -7.83077121e-01
7.45521545e-01 -3.02010685e-01 -3.20865035e-01 -3.04490209e-01
1.99427426e-01 1.08956468e+00 7.40819424e-02 -1.11861134e+00
-7.05039918e-01 2.71680743e-01 1.10628881e-01 4.37856227e-01
-2.23923773e-01 2.82243341e-01 -2.34259725e-01 -4.47338223e-01
-6.16276145e-01 -7.86531687e-01 -2.81639230e-02 -5.84401429e-01
-7.50039458e-01 -1.96489915e-01 3.75070065e-01 -6.78257346e-01
1.42027748e+00 -1.58149147e+00 -2.97016323e-01 -1.78358808e-01
-6.46016449e-02 1.81030005e-01 -6.34978116e-01 1.17042220e+00
2.09733680e-01 5.13901532e-01 1.18755080e-01 -5.75866342e-01
3.23185086e-01 1.96494102e-01 -2.82641888e-01 2.07180511e-02
-2.34655440e-01 1.19605219e+00 -8.63239527e-01 -5.27492285e-01
-1.02109477e-01 1.01806551e-01 -7.21463978e-01 -1.56971663e-01
-7.73754537e-01 2.77346104e-01 -1.51734948e-01 3.61028224e-01
3.17443579e-01 -2.54441351e-01 7.03980029e-02 3.05211842e-01
-5.07687628e-01 7.90599346e-01 -6.04402900e-01 1.72263551e+00
-6.50925457e-01 7.89487600e-01 -2.47049376e-01 -1.32499546e-01
5.17802238e-01 6.35459661e-01 1.91206336e-01 -1.03653300e+00
-1.29321888e-01 3.97213519e-01 2.25918189e-01 -4.52863663e-01
8.66668701e-01 -6.27760291e-02 -1.79680929e-01 8.37482989e-01
3.46066244e-03 -5.02516687e-01 7.60839641e-01 5.35329401e-01
1.16003466e+00 9.88614112e-02 9.01239738e-02 -6.78152144e-01
-5.75403869e-02 8.76088142e-02 3.46564680e-01 1.15060747e+00
1.85630292e-01 4.03278381e-01 6.92998230e-01 6.56740814e-02
-1.13036275e+00 -8.75272632e-01 -3.97484601e-02 1.26899374e+00
-2.15991452e-01 -1.09519947e+00 -9.50345635e-01 -3.74071032e-01
-1.69402421e-01 1.38495481e+00 -3.91667068e-01 -5.77109260e-03
-4.27777410e-01 -1.18919301e+00 1.32894385e+00 2.43999809e-01
5.38203835e-01 -6.61466837e-01 -2.05259889e-01 3.05848837e-01
-5.27810633e-01 -1.02833021e+00 -6.49910212e-01 2.92355180e-01
-5.25489151e-01 -4.95681435e-01 -6.93732023e-01 -4.80542511e-01
1.21031314e-01 -2.58186013e-01 1.35768342e+00 -2.46975541e-01
4.26534921e-01 1.53491274e-01 -5.69545209e-01 -6.52529657e-01
-1.13851833e+00 4.96458232e-01 -2.24774573e-02 -6.42815709e-01
1.73507333e-01 -2.11570427e-01 -1.37613356e-01 4.16400563e-03
-5.88193715e-01 4.81214285e-01 7.95611143e-01 7.33652890e-01
2.03547701e-01 -2.19189361e-01 5.66366374e-01 -1.02289271e+00
1.06106138e+00 -4.90804106e-01 -5.44764876e-01 4.55776185e-01
-7.87924528e-01 2.05704182e-01 5.92240691e-01 -2.53371418e-01
-1.23725355e+00 -4.55337852e-01 -4.65189606e-01 3.16954970e-01
3.31106931e-01 7.36327469e-01 -8.07093233e-02 3.01695615e-01
1.00883806e+00 4.67817336e-01 -3.78401548e-01 -5.02358913e-01
5.66457391e-01 7.02439010e-01 2.40289986e-01 -9.43316936e-01
5.38792133e-01 3.75875495e-02 -4.98091251e-01 -8.41907442e-01
-5.70594013e-01 8.92386436e-02 -1.32517904e-01 -1.03137428e-02
8.37523341e-01 -1.19625115e+00 -5.49045324e-01 4.81272787e-01
-1.37975121e+00 -1.10756934e+00 -3.24429423e-01 3.52079183e-01
-6.53305590e-01 2.20718980e-01 -1.05663240e+00 -7.22935736e-01
-5.05437195e-01 -1.26211703e+00 1.26819658e+00 -1.49079904e-01
-6.13228500e-01 -1.17014968e+00 3.61697525e-01 2.79753387e-01
3.73549581e-01 -2.76325434e-01 1.19096529e+00 -8.13337505e-01
-4.00121033e-01 1.10576443e-01 -5.82251102e-02 1.28555164e-01
-2.01545417e-01 2.52226312e-02 -7.83638000e-01 -3.36713612e-01
-2.68997401e-01 -5.25835335e-01 6.61749840e-01 5.23470044e-01
3.62893879e-01 -6.18933022e-01 -3.23612750e-01 5.68745673e-01
1.18874907e+00 2.41281241e-01 6.55966640e-01 4.21808749e-01
4.13856983e-01 4.46678400e-01 1.98401466e-01 4.07081008e-01
7.48189628e-01 4.66880679e-01 6.41490892e-02 1.63978696e-01
-2.83618480e-01 -8.02850902e-01 1.13211465e+00 1.13036549e+00
3.48803729e-01 -9.62240338e-01 -1.13435817e+00 4.87891048e-01
-1.55167425e+00 -7.75505185e-01 -5.65410137e-01 2.02282143e+00
1.29098737e+00 2.72541285e-01 5.67936487e-02 -1.85142413e-01
2.90398508e-01 1.44059375e-01 -4.09004139e-03 -8.23330462e-01
-5.29339194e-01 3.67266759e-02 7.38007784e-01 6.83144033e-01
-4.86604929e-01 1.28665948e+00 7.76649666e+00 1.26495552e+00
-9.03258562e-01 4.41124022e-01 5.89545250e-01 -5.13770103e-01
-6.72560036e-01 1.56773865e-01 -1.30611169e+00 7.42597759e-01
1.66740978e+00 -5.77963769e-01 4.75864023e-01 2.79359400e-01
4.82200623e-01 -2.25586176e-01 -1.40818739e+00 5.48154891e-01
-2.37336904e-02 -1.39508486e+00 2.82440484e-01 1.67138413e-01
1.13415074e+00 5.74700058e-01 1.01281941e-01 6.60746515e-01
1.22028577e+00 -1.31746984e+00 1.31446803e+00 2.04283357e-01
8.95018280e-01 -4.62195426e-01 4.04490292e-01 7.80940533e-01
-8.21029544e-01 -2.78117694e-02 -4.02620226e-01 5.18922023e-02
3.56015384e-01 5.47595322e-01 -1.02646780e+00 5.51994145e-01
5.05882382e-01 4.78563130e-01 -6.08852446e-01 8.84389937e-01
-2.71205127e-01 8.54056358e-01 -5.36956429e-01 -3.59162875e-02
5.69560111e-01 2.40044609e-01 5.72217822e-01 1.56548548e+00
4.80499595e-01 -2.40633905e-01 -1.31610811e-01 9.18905675e-01
-2.08863571e-01 1.73654526e-01 -5.34968615e-01 -3.29894006e-01
6.04181230e-01 8.52778196e-01 -2.04823792e-01 -4.42752898e-01
-3.67479980e-01 6.68994486e-01 -1.46088889e-04 3.31673831e-01
-7.38438785e-01 1.07689634e-01 3.37574929e-01 4.59238812e-02
-7.18154237e-02 -8.89296755e-02 -2.67405838e-01 -8.36406648e-01
-4.27148700e-01 -1.31958258e+00 2.47928053e-01 -1.13902831e+00
-1.03289747e+00 7.16207922e-01 1.74883127e-01 -8.93281877e-01
-5.95778346e-01 -4.53916579e-01 -3.06867033e-01 1.07277703e+00
-1.25976920e+00 -1.22582138e+00 2.47118950e-01 2.21824005e-01
7.02467561e-01 -1.41826794e-01 6.85311317e-01 2.38105148e-01
-4.78000909e-01 8.76464307e-01 5.42168438e-01 -2.00000539e-01
7.50832260e-01 -1.20111609e+00 9.88575697e-01 8.18768501e-01
1.23030216e-01 5.44670880e-01 8.20730925e-01 -9.28299725e-01
-1.25996149e+00 -1.16658473e+00 1.39196801e+00 -8.51264536e-01
6.12421870e-01 -5.99981070e-01 -3.41931999e-01 9.35701311e-01
7.58360386e-01 -9.52509284e-01 3.29978466e-01 8.53829160e-02
-2.69185275e-01 9.08798575e-02 -8.15369785e-01 9.29002285e-01
9.61813390e-01 -4.60355490e-01 -2.93994248e-01 7.51333535e-01
9.64901328e-01 -4.62627172e-01 -9.10903513e-01 4.52787317e-02
4.56488758e-01 -8.95547688e-01 4.98745710e-01 -3.48311841e-01
6.85939312e-01 2.09458843e-01 -2.71293133e-01 -1.62283027e+00
-1.11423790e-01 -1.14862144e+00 1.97148129e-01 1.29657090e+00
1.01376867e+00 -7.63742805e-01 4.28037673e-01 2.47035980e-01
-4.47920114e-01 -5.09440482e-01 -8.78552258e-01 -1.17791748e+00
8.25734437e-01 -6.10675752e-01 6.72486603e-01 6.31929755e-01
1.36308506e-01 3.12911719e-01 -4.00457472e-01 -3.41884851e-01
2.99887747e-01 -5.39765716e-01 7.09306479e-01 -7.36041069e-01
-5.55248857e-01 -5.97716689e-01 3.69550169e-01 -1.44840360e+00
-1.91253945e-02 -1.23158133e+00 1.14769883e-01 -1.76957345e+00
3.74777287e-01 -5.87819874e-01 2.11670756e-01 5.78808546e-01
8.93676132e-02 4.56901267e-02 2.86527634e-01 2.76355475e-01
-3.18852723e-01 5.65715194e-01 1.19710851e+00 -1.49216965e-01
4.91683930e-02 -4.14506942e-01 -1.03211188e+00 4.97418433e-01
4.72546667e-01 -5.35756052e-01 -5.97111464e-01 -9.43864644e-01
8.64527345e-01 2.16483742e-01 -3.52058448e-02 -8.61939013e-01
3.40357721e-01 3.17341387e-02 -6.77041709e-02 -4.64142948e-01
2.41045043e-01 -1.16756573e-01 3.15464318e-01 5.34396529e-01
-7.45442033e-01 3.61135244e-01 4.99481529e-01 1.14494726e-01
1.55758560e-01 -4.18306917e-01 4.35770363e-01 -2.96042591e-01
-3.80970329e-01 7.74627998e-02 -9.28924203e-01 5.71096122e-01
3.70764881e-01 -1.16414808e-01 -9.12622929e-01 -6.81903362e-01
-6.27828911e-02 4.53938007e-01 3.95291179e-01 4.66733426e-01
-4.35152650e-02 -1.06916165e+00 -1.20871937e+00 -1.47518143e-01
1.64857075e-01 -2.73262411e-01 -3.86379361e-02 1.09661019e+00
-7.37930715e-01 9.61959422e-01 2.93506235e-01 -5.04016519e-01
-8.16881120e-01 8.05476159e-02 5.60266137e-01 -6.97095513e-01
-3.09200227e-01 1.12313724e+00 6.44133508e-01 -4.58738953e-01
5.63386939e-02 -3.81443620e-01 3.45703185e-01 -1.52304366e-01
3.38093460e-01 4.38069969e-01 6.22139610e-02 -3.84893119e-01
-4.09473591e-02 -3.59413065e-02 -2.94972271e-01 -6.19343698e-01
1.01789021e+00 -6.69914037e-02 -4.31982242e-02 4.86105680e-01
8.01232517e-01 1.78433970e-01 -9.99939322e-01 4.55367938e-02
1.14246346e-01 1.20073855e-01 -1.12752780e-01 -1.40692234e+00
-5.23631096e-01 7.71366179e-01 1.31349310e-01 -3.34627032e-02
7.04284430e-01 -1.39909178e-01 9.27816153e-01 3.43916684e-01
5.56501031e-01 -1.25342953e+00 -2.32755989e-01 8.81345391e-01
9.52305317e-01 -6.51289761e-01 -1.64187104e-01 1.24444894e-01
-7.51703739e-01 5.75158477e-01 4.93359208e-01 5.18799007e-01
2.50598699e-01 7.08721876e-01 3.39284576e-02 -2.92679742e-02
-1.52234280e+00 -3.31543460e-02 -1.60078034e-01 4.70162004e-01
8.04686785e-01 2.56482400e-02 -5.30947149e-01 5.98256588e-01
-9.68481481e-01 -3.56607661e-02 6.60361171e-01 5.82322657e-01
-3.11104149e-01 -1.31245732e+00 -2.59548783e-01 4.48302031e-01
-4.76389468e-01 -8.68537903e-01 -4.95843142e-01 1.06661046e+00
1.33331940e-01 1.28943527e+00 -1.56363443e-01 -4.61679280e-01
1.34167448e-01 1.84610367e-01 5.23261309e-01 -7.06755936e-01
-1.15906990e+00 2.61194110e-01 6.75600410e-01 -2.70390123e-01
2.20196992e-01 -7.70487189e-01 -8.81657839e-01 -8.17647099e-01
-2.88114250e-01 3.82329434e-01 5.42491794e-01 8.57314944e-01
2.89197832e-01 3.17166984e-01 1.15538396e-01 -4.60896403e-01
-6.74173832e-01 -1.37261057e+00 -5.36795616e-01 -9.89876762e-02
-1.85328107e-02 -1.00387283e-01 -2.25330874e-01 -2.01225784e-02] | [11.556554794311523, 8.88492202758789] |
0f5c2495-7d7e-4d78-bfef-d6b1545490ca | reasoning-structural-relation-for-occlusion | 2112.10087 | null | https://arxiv.org/abs/2112.10087v1 | https://arxiv.org/pdf/2112.10087v1.pdf | Reasoning Structural Relation for Occlusion-Robust Facial Landmark Localization | In facial landmark localization tasks, various occlusions heavily degrade the localization accuracy due to the partial observability of facial features. This paper proposes a structural relation network (SRN) for occlusion-robust landmark localization. Unlike most existing methods that simply exploit the shape constraint, the proposed SRN aims to capture the structural relations among different facial components. These relations can be considered a more powerful shape constraint against occlusion. To achieve this, a hierarchical structural relation module (HSRM) is designed to hierarchically reason the structural relations that represent both long- and short-distance spatial dependencies. Compared with existing network architectures, HSRM can efficiently model the spatial relations by leveraging its geometry-aware network architecture, which reduces the semantic ambiguity caused by occlusion. Moreover, the SRN augments the training data by synthesizing occluded faces. To further extend our SRN for occluded video data, we formulate the occluded face synthesis as a Markov decision process (MDP). Specifically, it plans the movement of the dynamic occlusion based on an accumulated reward associated with the performance degradation of the pre-trained SRN. This procedure augments hard samples for robust facial landmark tracking. Extensive experimental results indicate that the proposed method achieves outstanding performance on occluded and masked faces. Code is available at https://github.com/zhuccly/SRN. | ['Weiqin Tong', 'Songmin Dai', 'Jide Li', 'Xiaoqiang Li', 'Congcong Zhu'] | 2021-12-19 | null | null | null | null | ['face-alignment', 'landmark-tracking'] | ['computer-vision', 'computer-vision'] | [-7.09584579e-02 1.78758577e-01 -4.64064389e-01 -6.26662016e-01
-4.50896710e-01 -1.32713288e-01 3.83265942e-01 -5.86412549e-01
1.60039570e-02 2.43314341e-01 1.61132976e-01 -5.74402250e-02
-1.78266913e-01 -5.09899974e-01 -6.39117718e-01 -7.23926187e-01
-5.04921265e-02 1.79283731e-02 -1.55793661e-02 -1.30265660e-03
1.68072395e-02 8.81034136e-01 -1.45540595e+00 5.92222735e-02
7.06434190e-01 1.30929732e+00 1.83058128e-01 -5.39004393e-02
2.39228681e-02 6.51965678e-01 -3.22128087e-01 -1.16408907e-01
3.12838495e-01 -1.36352628e-01 -4.57929730e-01 2.87793934e-01
5.46691179e-01 -4.31198627e-01 -5.56821585e-01 1.07438159e+00
4.99694943e-01 2.31091008e-01 3.61998260e-01 -1.45345652e+00
-6.78519070e-01 3.75497788e-01 -7.53282011e-01 7.45907351e-02
3.14501315e-01 2.52863262e-02 8.96716893e-01 -1.29093683e+00
5.58794081e-01 1.61319506e+00 4.16219741e-01 6.48242593e-01
-9.97067750e-01 -9.73634839e-01 7.09141493e-01 2.57821858e-01
-1.85588682e+00 -8.99613500e-01 1.09742832e+00 -1.90930143e-01
4.49935138e-01 1.76697150e-01 4.15099978e-01 9.57794845e-01
-7.66786933e-02 8.76012623e-01 7.04722285e-01 -1.81206897e-01
6.31545559e-02 -3.31444770e-01 -2.17678070e-01 9.49890792e-01
-7.94529170e-03 2.49746904e-01 -7.13354707e-01 4.30741273e-02
1.12986004e+00 3.03445365e-02 -2.76212722e-01 -4.74176735e-01
-5.65447688e-01 5.19634426e-01 7.71724045e-01 1.77784637e-01
-2.04818010e-01 2.83799112e-01 8.27132538e-02 5.38577177e-02
5.27842283e-01 -1.29337177e-01 -4.13913131e-01 4.61073995e-01
-8.11859369e-01 -2.31658161e-01 2.74696440e-01 1.10940731e+00
9.38165963e-01 2.79557854e-01 -2.55658597e-01 8.08975279e-01
8.90132010e-01 3.60615522e-01 1.23072885e-01 -1.11368537e+00
3.93227339e-01 7.03993917e-01 -6.90269768e-02 -1.31908238e+00
-4.54725415e-01 -5.21023631e-01 -7.63421178e-01 1.06614359e-01
1.77941605e-01 9.32472348e-02 -1.00310946e+00 1.89340699e+00
7.10082471e-01 5.45661628e-01 -2.29952678e-01 9.40178275e-01
9.75499213e-01 4.08841133e-01 1.32029906e-01 -2.65034407e-01
1.25947928e+00 -9.88530338e-01 -8.60108554e-01 -2.25911811e-01
3.55551958e-01 -7.81198204e-01 7.72767305e-01 1.14392899e-02
-8.47963452e-01 -7.04454780e-01 -8.18688333e-01 -1.40085816e-01
9.21125617e-03 6.02368176e-01 6.40210986e-01 4.40134108e-01
-1.19660664e+00 2.73854196e-01 -9.88936067e-01 -8.24338496e-02
8.36885929e-01 7.49855638e-01 -4.06343400e-01 -2.42717162e-01
-9.09276903e-01 4.45563406e-01 -1.44154364e-02 9.78903472e-01
-9.78864372e-01 -4.86301661e-01 -1.15412235e+00 1.26907304e-01
5.43400466e-01 -2.87231326e-01 9.26080525e-01 -8.56035054e-01
-1.59479773e+00 5.64597249e-01 -5.63867748e-01 2.79554185e-02
3.78414363e-01 -1.72829911e-01 -3.32738400e-01 1.84097856e-01
9.60629135e-02 9.52087641e-01 1.06077027e+00 -1.32383227e+00
-4.27759051e-01 -4.47460443e-01 -3.31745632e-02 2.13819534e-01
-3.41426849e-01 1.51759401e-01 -8.42989862e-01 -7.56282032e-01
5.72643638e-01 -7.42904246e-01 -2.17240617e-01 4.10392731e-01
-4.07560468e-01 -4.05465096e-01 1.00191796e+00 -5.92696667e-01
1.08337712e+00 -2.28656030e+00 1.46939665e-01 3.20570558e-01
7.14363605e-02 2.42017746e-01 -6.26501143e-01 2.58810129e-02
-1.64797679e-01 -6.41014725e-02 1.27950579e-01 -7.72055447e-01
-4.09007678e-03 3.03775787e-01 -3.10191810e-01 7.47471094e-01
2.81201571e-01 1.08043969e+00 -5.81830382e-01 -6.13347888e-01
1.36175796e-01 8.70414674e-01 -4.45716172e-01 1.05413362e-01
-1.93265349e-01 7.11988747e-01 -6.19187117e-01 1.02618551e+00
1.01484978e+00 -1.75184026e-01 1.63576439e-01 -4.69415873e-01
6.35625347e-02 7.08798924e-03 -1.13623548e+00 1.84559226e+00
-3.44956279e-01 4.13208425e-01 2.24198014e-01 -6.68387115e-01
1.10333550e+00 2.62215197e-01 5.69731355e-01 -8.55057061e-01
3.02177221e-01 5.65577066e-03 -1.22986093e-01 -4.13002014e-01
-2.17905212e-02 4.31807846e-01 3.70703518e-01 1.15583166e-01
-7.16517940e-02 5.54424226e-01 -1.84730843e-01 -8.87777880e-02
8.30639362e-01 4.84841615e-01 1.12218410e-01 -2.34489068e-01
9.22476470e-01 -7.33384490e-01 1.11705601e+00 1.93623081e-01
-3.79995257e-01 2.91917801e-01 4.36516196e-01 -7.40014553e-01
-4.34544116e-01 -9.33579087e-01 -2.16571331e-01 1.08141983e+00
4.02764410e-01 -3.87190223e-01 -6.90912902e-01 -7.06495047e-01
-1.78569511e-01 6.07461669e-02 -7.33677387e-01 -1.78281277e-01
-7.75449514e-01 -4.26598787e-01 3.85963738e-01 5.60588121e-01
6.99893296e-01 -1.14708984e+00 -1.81097329e-01 -7.99422115e-02
-1.81226522e-01 -1.20272112e+00 -7.01829672e-01 -2.89493293e-01
-6.86278820e-01 -1.00232160e+00 -3.09839129e-01 -1.03005493e+00
1.14397967e+00 3.55990201e-01 5.91206014e-01 5.44826508e-01
-2.83944905e-01 1.23764716e-01 -1.77916378e-01 7.51358643e-02
1.98674843e-01 2.95130946e-02 1.65642798e-01 4.79399353e-01
2.89422750e-01 -7.00013280e-01 -7.59573042e-01 7.02086031e-01
-7.33521760e-01 -1.09340601e-01 6.40382648e-01 7.18475580e-01
7.99623251e-01 1.65219679e-01 4.25136536e-01 -4.27830756e-01
-3.38948369e-02 -2.14307979e-01 -8.40934694e-01 3.18240225e-01
-2.38517672e-01 6.86268462e-03 2.75701433e-01 -5.10906696e-01
-1.16602790e+00 4.40330535e-01 -2.03028634e-01 -7.70485699e-01
-1.29784554e-01 1.24286205e-01 -8.20343673e-01 -5.26037037e-01
1.30076170e-01 1.63359717e-01 -6.95351586e-02 -4.42144454e-01
3.58451307e-01 2.79181659e-01 2.90670812e-01 -7.09404945e-01
9.61393118e-01 7.41119623e-01 2.52735585e-01 -6.04190528e-01
-1.01207781e+00 -1.45511523e-01 -7.51433671e-01 -3.18902493e-01
7.09672034e-01 -1.00231981e+00 -1.04710829e+00 2.64616787e-01
-1.20847201e+00 -3.55467886e-01 5.76103106e-02 3.13351244e-01
-3.57074648e-01 2.32568011e-01 -6.05498672e-01 -8.87050927e-01
-3.61817852e-02 -1.34534657e+00 1.28862131e+00 3.39321792e-01
8.26088339e-02 -8.17829430e-01 -4.36826348e-01 1.79267094e-01
2.14065328e-01 2.10432127e-01 5.58571815e-01 -1.63010404e-01
-1.06587791e+00 -1.78935528e-02 -3.51720273e-01 1.33322209e-01
3.61173451e-01 2.08214354e-02 -9.90505219e-01 -2.98782915e-01
5.89456037e-02 -1.48490712e-01 7.30590284e-01 4.75325763e-01
1.28580630e+00 -3.89966935e-01 -4.65218574e-01 9.33853328e-01
1.09784722e+00 2.56933391e-01 5.85398853e-01 1.75125390e-01
1.05363500e+00 8.09441388e-01 7.28376091e-01 2.35915601e-01
3.94957334e-01 8.29337060e-01 7.01921165e-01 -1.88538745e-01
-2.68773377e-01 -4.69403327e-01 3.17249656e-01 5.05811989e-01
5.17130457e-02 -8.47025029e-03 -6.06722653e-01 1.07155889e-01
-1.94066823e+00 -6.29766762e-01 1.55577093e-01 1.84637415e+00
5.05194426e-01 -1.46176592e-01 -3.33952069e-01 -1.96652904e-01
7.98323274e-01 4.00707513e-01 -5.92803180e-01 2.17289060e-01
-1.23114459e-01 -2.27495674e-02 1.49723351e-01 7.45778739e-01
-1.02921510e+00 1.33131611e+00 5.40724945e+00 1.08431077e+00
-1.08157098e+00 2.30790451e-01 7.14449167e-01 -2.44325381e-02
-1.51624113e-01 -6.23750165e-02 -1.11457682e+00 3.88779670e-01
2.69238740e-01 4.19272274e-01 3.99043500e-01 9.09999609e-01
4.20887977e-01 9.49792191e-02 -9.94281828e-01 1.00559926e+00
1.27123341e-01 -1.06910729e+00 1.23331189e-01 2.67049819e-01
6.45227075e-01 -2.95830131e-01 4.22066540e-01 -2.38826927e-02
5.98389171e-02 -1.13183308e+00 6.71461403e-01 4.48349267e-01
9.00613427e-01 -7.55902588e-01 4.94354576e-01 1.02926465e-02
-1.77654803e+00 -1.65795416e-01 -4.49858099e-01 9.94697362e-02
9.21607465e-02 3.52423131e-01 -3.66418034e-01 4.78592545e-01
5.31045496e-01 7.64265001e-01 -6.35352015e-01 8.18038166e-01
-8.21988165e-01 1.68386251e-01 -3.07373345e-01 4.83648658e-01
7.82581139e-03 -2.16718808e-01 3.08642626e-01 5.65924883e-01
1.58407152e-01 1.13621525e-01 2.64726639e-01 7.43882239e-01
-1.56757399e-01 8.94741341e-02 -3.91781449e-01 4.48016465e-01
7.85044789e-01 1.41594923e+00 -8.27313423e-01 8.14988315e-02
-3.67893219e-01 8.85969758e-01 6.14234030e-01 6.52129531e-01
-9.37238753e-01 1.40849054e-01 7.13112712e-01 -1.35041559e-02
3.84325206e-01 -2.05983371e-01 -8.22121724e-02 -9.66018379e-01
3.12593520e-01 -7.02005208e-01 1.84793487e-01 -5.77868581e-01
-9.41407084e-01 9.51645732e-01 -2.80798018e-01 -1.18100739e+00
2.09609061e-01 -4.06901151e-01 -4.41963464e-01 7.65384436e-01
-1.73690021e+00 -1.49401963e+00 -3.78046900e-01 6.88362956e-01
5.21734059e-01 -6.11700639e-02 5.49378693e-01 5.31467557e-01
-1.14459479e+00 8.15428972e-01 -5.18383086e-01 2.97102690e-01
6.26317561e-01 -6.08446538e-01 1.87865451e-01 9.29864109e-01
2.61419088e-01 7.89193571e-01 2.42115483e-01 -7.16427624e-01
-1.17827702e+00 -1.46650839e+00 7.08493292e-01 -3.07618022e-01
4.73486543e-01 -5.28255761e-01 -9.02511835e-01 8.28282118e-01
-2.05683425e-01 6.20341480e-01 4.53720540e-01 -1.14348888e-01
-6.88898861e-01 -4.40265954e-01 -9.50135410e-01 6.61581218e-01
1.53019369e+00 -5.90721428e-01 -7.71588013e-02 3.23065102e-01
7.45148063e-01 -5.22073805e-01 -7.13782191e-01 6.15007341e-01
5.54412663e-01 -7.88897395e-01 9.80852306e-01 -2.87183046e-01
2.05682904e-01 -6.39040530e-01 -2.50734806e-01 -7.59080827e-01
-3.30379814e-01 -7.49940336e-01 -3.76442313e-01 1.39995146e+00
5.63297048e-02 -5.62699139e-01 1.00946963e+00 6.56208038e-01
-4.45869491e-02 -1.09327185e+00 -1.23370028e+00 -8.71266723e-01
-4.21090961e-01 -2.71635115e-01 8.13263953e-01 7.73173273e-01
-5.19711673e-01 1.59544632e-01 -4.65400994e-01 7.17375457e-01
6.47809267e-01 3.29626873e-02 6.35717213e-01 -1.12767160e+00
1.10561848e-01 -3.35666269e-01 -4.48687941e-01 -1.33377790e+00
6.42295480e-01 -7.11109102e-01 9.54859555e-02 -1.21850634e+00
-1.45003602e-01 -7.76785910e-01 -2.37546742e-01 7.17826009e-01
-1.28894478e-01 4.00799930e-01 8.15913677e-02 3.79479617e-01
-8.01615477e-01 8.95594299e-01 1.35938919e+00 9.89933759e-02
-2.86152124e-01 2.32823528e-02 -5.19312024e-01 8.97392690e-01
7.49908686e-01 -4.11192954e-01 -5.90067387e-01 -6.34792447e-01
-2.71373917e-03 6.19065715e-04 4.30822909e-01 -8.25613916e-01
4.52780515e-01 -1.64864078e-01 6.46510303e-01 -6.60914004e-01
5.65037012e-01 -1.03148735e+00 2.61560410e-01 3.72325897e-01
-2.05660805e-01 -7.88421035e-02 1.15071766e-01 5.84266365e-01
-2.69190848e-01 1.78631157e-01 6.52043164e-01 1.94082871e-01
-6.41700387e-01 9.14035141e-01 1.31569386e-01 -3.98831308e-01
1.02280080e+00 -1.25518352e-01 -1.81627333e-01 -1.81323677e-01
-7.97398090e-01 3.79681140e-01 2.58783132e-01 6.31851792e-01
7.78673828e-01 -1.54190218e+00 -3.07187587e-01 4.55601424e-01
-7.63740689e-02 2.74278253e-01 4.46442604e-01 8.92288387e-01
-2.58906633e-01 4.01910394e-01 -1.04123764e-01 -5.96074820e-01
-1.27122676e+00 6.40671968e-01 4.57693934e-01 2.60800286e-03
-4.94339377e-01 1.12563539e+00 6.51759624e-01 -2.48919949e-01
5.24854720e-01 -1.04080327e-01 -3.48903537e-01 2.44349334e-02
6.17739975e-01 1.28308490e-01 -1.63518414e-01 -9.86181140e-01
-5.22703409e-01 8.74346733e-01 -1.60186142e-01 1.94612980e-01
1.15579724e+00 -3.84447336e-01 -3.86481732e-01 -1.49310708e-01
1.26911092e+00 2.20139548e-02 -1.72744107e+00 -4.59996253e-01
-1.40003815e-01 -5.93665719e-01 -5.09016449e-03 -2.42275313e-01
-1.55077779e+00 7.29403198e-01 5.36594093e-01 -4.20847386e-01
1.35806620e+00 2.11433265e-02 4.15332973e-01 1.79105297e-01
4.67108548e-01 -7.50278890e-01 3.18350941e-01 2.66204268e-01
1.03592110e+00 -1.15913999e+00 2.45949645e-02 -1.00567067e+00
-2.21816465e-01 1.00770557e+00 1.09413052e+00 1.44930258e-02
7.72994936e-01 1.67205870e-01 4.83976975e-02 -2.97708273e-01
-4.29352999e-01 -2.41316959e-01 4.79053915e-01 5.21420896e-01
8.51822048e-02 -2.10821569e-01 -6.73248768e-02 4.65330601e-01
6.78220391e-02 -1.83532864e-01 -1.72356158e-01 7.09893942e-01
-2.36745968e-01 -1.16345108e+00 -3.17034781e-01 3.00255381e-02
-1.62481561e-01 -2.28051618e-02 -1.55986398e-01 6.78368509e-01
4.87806886e-01 8.80573213e-01 3.45823076e-03 -3.51993054e-01
1.80283472e-01 -3.35901529e-01 3.21610630e-01 -5.84115446e-01
1.89222284e-02 3.49528342e-01 -2.71929979e-01 -1.05830586e+00
-4.90868896e-01 -5.42515397e-01 -1.25528502e+00 -2.73214653e-02
-3.36984605e-01 -6.81590959e-02 3.96858990e-01 9.19192433e-01
5.12539148e-01 5.74260771e-01 8.17167938e-01 -9.30982530e-01
-1.70002028e-01 -7.06765175e-01 -3.76959503e-01 -5.06028980e-02
4.11455661e-01 -1.04554951e+00 -2.01631069e-01 -8.02846923e-02] | [13.400725364685059, 0.44058793783187866] |
da8fc742-db99-449c-a733-a2716c877056 | cs-um6p-at-semeval-2021-task-1-a-deep | null | null | https://aclanthology.org/2021.semeval-1.73 | https://aclanthology.org/2021.semeval-1.73.pdf | CS-UM6P at SemEval-2021 Task 1: A Deep Learning Model-based Pre-trained Transformer Encoder for Lexical Complexity | Lexical Complexity Prediction (LCP) involves assigning a difficulty score to a particular word or expression, in a text intended for a target audience. In this paper, we introduce a new deep learning-based system for this challenging task. The proposed system consists of a deep learning model, based on pre-trained transformer encoder, for word and Multi-Word Expression (MWE) complexity prediction. First, on top of the encoder{'}s contextualized word embedding, our model employs an attention layer on the input context and the complex word or MWE. Then, the attention output is concatenated with the pooled output of the encoder and passed to a regression module. We investigate both single-task and joint training on both Sub-Tasks data using multiple pre-trained transformer-based encoders. The obtained results are very promising and show the effectiveness of fine-tuning pre-trained transformers for LCP task. | ['Ismail Berrada', 'Kabil Essefar', 'Abdellah El Mekki', 'Abdelkader El Mahdaouy', 'Nabil El Mamoun'] | 2021-08-01 | null | null | null | semeval-2021 | ['lexical-complexity-prediction'] | ['natural-language-processing'] | [ 3.42396051e-01 2.59132981e-01 -2.14967858e-02 -5.33397317e-01
-1.05251288e+00 -1.70905128e-01 3.68847430e-01 3.54413897e-01
-7.40945399e-01 3.00971448e-01 4.40789431e-01 -4.89610583e-01
2.25993574e-01 -7.18165874e-01 -6.54861093e-01 -2.27507815e-01
3.75177294e-01 2.57307678e-01 2.35271510e-02 -1.33005947e-01
1.60731375e-01 -3.51143535e-03 -1.31239069e+00 5.80255032e-01
8.24716330e-01 1.22425365e+00 6.20583832e-01 7.78160572e-01
-3.46356004e-01 1.00491643e+00 -3.88564974e-01 -8.21545064e-01
-2.25796327e-01 -2.53086001e-01 -7.48406112e-01 -1.99742436e-01
1.32044002e-01 -4.92226966e-02 -8.27246681e-02 8.25958848e-01
6.06109798e-01 9.82281566e-02 7.05012679e-01 -5.97951055e-01
-5.24298668e-01 7.79699385e-01 -1.82851329e-01 2.20954731e-01
3.03840220e-01 3.56007405e-02 1.65150285e+00 -1.13323116e+00
3.94772500e-01 1.12701356e+00 5.72717965e-01 2.59949178e-01
-9.77003872e-01 -6.93712056e-01 3.80388647e-01 5.34550369e-01
-1.10486424e+00 -3.70970517e-01 8.43412638e-01 -4.14767087e-01
1.67426395e+00 -8.87745768e-02 6.44710600e-01 1.20705664e+00
4.68110204e-01 8.24924111e-01 8.45390320e-01 -5.98341465e-01
1.08040854e-01 3.04803461e-01 2.61245281e-01 8.18057775e-01
-3.17636490e-01 -1.22855425e-01 -4.88204211e-01 1.95604321e-02
5.34212291e-02 -4.08514589e-01 -1.78319328e-02 1.98427394e-01
-6.90807819e-01 1.14353597e+00 2.77422041e-01 3.70640665e-01
-5.00186026e-01 2.40757335e-02 5.69718897e-01 3.69268119e-01
7.73522198e-01 3.97035092e-01 -7.45607615e-01 -2.85030931e-01
-7.59762824e-01 5.64311594e-02 7.96349585e-01 7.44391084e-01
4.86758113e-01 -2.18081430e-01 -5.45968652e-01 1.09381199e+00
3.70760292e-01 1.17954232e-01 8.09049249e-01 -1.62717074e-01
8.36431265e-01 6.65309846e-01 -3.11624259e-01 -7.07791269e-01
-5.10907769e-01 -5.88858128e-01 -4.22254682e-01 -1.30800724e-01
2.70396937e-02 -2.80325890e-01 -6.67777956e-01 1.84896755e+00
1.19411252e-01 -8.34006444e-02 1.41645238e-01 5.97532928e-01
7.38334656e-01 1.07748568e+00 6.36765480e-01 -2.08366334e-01
1.69189358e+00 -1.27868724e+00 -7.94383585e-01 -3.96862268e-01
9.90237534e-01 -9.59266484e-01 1.49012542e+00 3.70180309e-01
-1.22605145e+00 -6.12533271e-01 -1.13844395e+00 -5.28858066e-01
-4.87048119e-01 4.46088463e-01 1.26203820e-02 2.85183817e-01
-7.17502356e-01 3.83542925e-01 -4.25371736e-01 -8.08014870e-02
3.35166156e-01 1.94612503e-01 -1.36141196e-01 -4.21980470e-02
-1.48764086e+00 1.30732381e+00 2.02872843e-01 -7.94904400e-03
-6.44936025e-01 -6.10425711e-01 -1.04307234e+00 5.51109612e-01
1.00787543e-01 -7.23060966e-01 1.30167377e+00 -1.12216353e+00
-1.85531962e+00 7.26713121e-01 -2.63589203e-01 -2.47787908e-01
6.91984966e-02 -4.43531871e-01 -3.10246080e-01 -2.94968307e-01
-3.47893382e-03 3.16321164e-01 9.52920675e-01 -5.43883920e-01
-7.00155854e-01 -1.77640185e-01 1.65784582e-01 3.92639071e-01
-6.53300226e-01 1.72302127e-01 -4.26982135e-01 -8.14491928e-01
-5.85608304e-01 -5.80396056e-01 -1.20898411e-01 -3.72766584e-01
-3.79326522e-01 -7.99324214e-01 4.81351882e-01 -1.04868448e+00
1.77632141e+00 -2.18779659e+00 6.11755967e-01 6.95515424e-02
8.13340917e-02 4.46181089e-01 -5.97104847e-01 4.97131318e-01
-1.59362197e-01 2.81227380e-02 -2.34730579e-02 -6.62522316e-01
1.73130676e-01 -4.95141745e-02 -1.03746029e-03 3.54416855e-02
6.60733640e-01 9.74606395e-01 -7.06611037e-01 -3.98708522e-01
9.29539725e-02 4.43905592e-01 -7.35291839e-01 5.89262247e-01
-4.27697480e-01 -5.94216697e-02 -4.00084168e-01 1.94759622e-01
3.58485430e-01 -3.16056646e-02 2.75910378e-01 -3.46636027e-01
-2.29458138e-01 7.76092887e-01 -7.22187519e-01 1.58294582e+00
-1.37472987e+00 7.00706959e-01 -2.65257120e-01 -1.05841863e+00
8.32081258e-01 3.32861245e-01 5.28422277e-03 -9.95265722e-01
4.16802645e-01 1.48966298e-01 1.16561703e-01 -7.43657053e-01
4.79257911e-01 -3.93955261e-01 -2.95238167e-01 5.51588595e-01
6.09453559e-01 1.19341381e-01 2.00760514e-02 -1.46822795e-01
1.24980044e+00 -7.92282447e-02 4.49366331e-01 -1.31277665e-01
7.64308095e-01 -6.01302922e-01 1.81706339e-01 1.70913130e-01
3.80987972e-02 1.32016569e-01 7.49579728e-01 -4.19218719e-01
-1.27431905e+00 -5.48342168e-01 -6.21881289e-03 1.72830331e+00
-3.32136065e-01 -8.56739879e-01 -7.72098362e-01 -6.65474474e-01
-2.12260708e-01 8.56445253e-01 -6.89134419e-01 -3.76718104e-01
-5.33723950e-01 -5.49316764e-01 4.19450521e-01 6.35499358e-01
8.95999596e-02 -1.23498762e+00 -4.83791977e-01 5.85410476e-01
-2.85668284e-01 -1.25704193e+00 -5.70967078e-01 5.31881988e-01
-4.22276735e-01 -7.03473151e-01 -5.06606579e-01 -1.10271180e+00
2.30506837e-01 -2.79870808e-01 1.10970426e+00 -1.66802723e-02
-9.52382535e-02 -1.39863536e-01 -5.34558475e-01 -4.88348216e-01
-2.53187895e-01 5.70539534e-01 -3.25217754e-01 1.65800825e-01
5.62144876e-01 -3.94483000e-01 -3.47896993e-01 -1.61472544e-01
-5.07580936e-01 1.88929141e-01 8.75432789e-01 9.72190022e-01
4.25467283e-01 -1.15371995e-01 4.62474406e-01 -8.58965695e-01
1.02981997e+00 -4.47952360e-01 -4.28373992e-01 2.84830630e-01
-3.35791528e-01 4.57248688e-01 6.50737584e-01 -5.31634808e-01
-9.44779575e-01 9.47332829e-02 -7.63965964e-01 -6.90113232e-02
3.82590383e-01 8.34573984e-01 -4.55675870e-01 3.41450423e-01
3.93398106e-01 4.24587876e-02 -3.28843296e-01 -5.10950923e-01
4.81122226e-01 8.66987705e-01 -4.28235345e-02 -4.30403501e-01
3.79867733e-01 -4.13785368e-01 -4.81341988e-01 -7.59761453e-01
-1.13152683e+00 -2.92278737e-01 -5.16208529e-01 -4.52759564e-02
1.06707180e+00 -1.05552256e+00 -4.52892065e-01 2.69873857e-01
-1.52035892e+00 -4.97323781e-01 -1.88226908e-01 3.99356663e-01
-3.99730057e-01 2.58015655e-02 -5.31673908e-01 -7.08142161e-01
-6.33659542e-01 -1.17315304e+00 1.18232143e+00 -7.57305920e-02
-3.69653255e-01 -1.00256097e+00 2.20445246e-01 3.34768832e-01
5.04355907e-01 -1.51824981e-01 1.48400748e+00 -8.85409296e-01
-1.03308819e-01 -3.17326665e-01 -3.32484990e-01 6.90008759e-01
-1.72877923e-01 -1.10508487e-01 -1.00643528e+00 2.45623570e-02
2.28044186e-02 -7.04682231e-01 8.75487626e-01 2.71533638e-01
1.31068921e+00 -4.51462358e-01 -2.32800450e-02 4.72555161e-01
1.44618905e+00 -1.97733402e-01 5.00739038e-01 2.22369567e-01
5.95569551e-01 5.98945796e-01 4.37060356e-01 4.73836720e-01
6.67109847e-01 7.67617583e-01 9.17165205e-02 -9.04838294e-02
-1.08828180e-01 -1.75579190e-01 5.99925339e-01 1.32799983e+00
6.34668171e-02 -3.55077744e-01 -7.13855386e-01 3.67609948e-01
-1.55833733e+00 -6.95740402e-01 1.59030050e-01 1.86426258e+00
1.02669537e+00 4.32475537e-01 -1.00088805e-01 1.34315193e-01
3.49387616e-01 3.11560392e-01 -4.15183246e-01 -1.12420738e+00
1.17494300e-01 7.93061793e-01 8.03615078e-02 8.80546212e-01
-9.61697459e-01 1.23291445e+00 5.36110067e+00 1.11721551e+00
-1.13713884e+00 4.98348236e-01 5.38157403e-01 -4.70705144e-02
-3.99810970e-01 -2.42959723e-01 -8.31624925e-01 3.52037638e-01
1.27330363e+00 -3.30888033e-01 4.27561969e-01 7.74508417e-01
1.11651681e-02 7.29785487e-02 -1.20861423e+00 7.50494361e-01
1.30637020e-01 -7.56813645e-01 1.26858562e-01 -1.12003490e-01
2.34496012e-01 -2.86364350e-02 2.11286321e-01 7.97642112e-01
-1.31443944e-02 -1.11856484e+00 6.96149170e-01 3.41600269e-01
8.40406597e-01 -9.77142334e-01 9.34140384e-01 3.22426677e-01
-1.39020169e+00 -2.20436260e-01 -4.61246729e-01 -2.55967289e-01
2.81540155e-02 6.11715138e-01 -5.49148500e-01 1.97866961e-01
2.68673569e-01 6.26399994e-01 -5.28252482e-01 3.14368695e-01
-6.36436462e-01 6.89023614e-01 -1.47508487e-01 -5.33933640e-01
4.28628087e-01 1.72036752e-01 1.27644584e-01 1.75770903e+00
2.02693522e-01 1.64961293e-01 -1.06236361e-01 5.77578664e-01
-2.85802066e-01 7.48169780e-01 -4.09947604e-01 -1.61145896e-01
2.12917447e-01 1.50088334e+00 -3.22196144e-03 -5.44999063e-01
-7.34885514e-01 1.04900813e+00 1.09642816e+00 7.13271871e-02
-9.24048543e-01 -6.71214640e-01 6.81903481e-01 -9.43685398e-02
6.11369014e-01 1.86492950e-02 -3.33641291e-01 -1.16560543e+00
1.33558840e-01 -8.05175364e-01 3.08910698e-01 -6.40913785e-01
-1.20036864e+00 9.29025829e-01 -4.72031176e-01 -8.84386539e-01
-1.99484065e-01 -9.04281020e-01 -8.74244213e-01 1.25414228e+00
-1.78142977e+00 -1.22622311e+00 3.95297185e-02 3.87945414e-01
9.28178728e-01 -2.72256732e-01 1.05418646e+00 3.17965299e-01
-7.24700570e-01 9.25978482e-01 -1.52622208e-01 -7.35995844e-02
5.22228122e-01 -1.25631070e+00 4.06994909e-01 4.24004644e-01
2.45728306e-02 2.46258944e-01 4.17257398e-01 -4.29240525e-01
-1.03077388e+00 -1.32837987e+00 1.77217102e+00 -4.04708236e-01
8.36190343e-01 -7.18743205e-01 -8.83572638e-01 5.73076904e-01
4.27268863e-01 -3.38360965e-01 1.03368723e+00 5.07916510e-01
-4.33616638e-01 -3.17951851e-02 -6.04677439e-01 3.58350009e-01
6.64625049e-01 -8.16781402e-01 -8.24601710e-01 3.13226044e-01
9.86567438e-01 -2.15759486e-01 -1.00118256e+00 1.44965395e-01
7.35139430e-01 -5.56630313e-01 8.01163852e-01 -8.11436534e-01
9.59658146e-01 4.22706276e-01 -2.29656637e-01 -1.59132457e+00
-7.03989625e-01 1.25520641e-03 -1.25693128e-01 1.03861964e+00
6.88811362e-01 -9.33304355e-02 4.33251590e-01 3.92174333e-01
-3.29559326e-01 -1.31199408e+00 -9.95613933e-01 -2.38980994e-01
3.02298576e-01 -6.15047634e-01 2.71554381e-01 6.37024343e-01
4.59703237e-01 1.21638048e+00 -2.97874004e-01 1.89715140e-02
2.32464939e-01 -2.41226450e-01 3.32360119e-01 -8.53384793e-01
-3.88019323e-01 -6.56284750e-01 -1.86422065e-01 -9.64104295e-01
5.12887716e-01 -1.05855894e+00 -5.22325337e-02 -1.27267551e+00
1.11148864e-01 4.22695689e-02 -4.26409304e-01 4.44022715e-01
-6.53901696e-01 -4.69189733e-02 1.93558112e-01 -3.39211226e-01
-5.20126224e-01 9.06186998e-01 1.01135349e+00 -1.52927384e-01
-5.91670685e-02 -9.65051055e-02 -6.29087090e-01 6.16182506e-01
7.45253205e-01 -3.99537861e-01 -2.31476620e-01 -6.97119832e-01
3.66405398e-01 -9.70528573e-02 -5.21863848e-02 -7.74907887e-01
1.32462502e-01 1.96513027e-01 3.26228648e-01 -3.97410184e-01
2.99635649e-01 -5.77101886e-01 -3.90073031e-01 5.28673351e-01
-6.50068760e-01 1.61479548e-01 3.18296134e-01 1.81624576e-01
-2.68151045e-01 -4.61283475e-01 7.76502669e-01 4.72548716e-02
-4.22943771e-01 1.60425439e-01 -6.51480377e-01 8.90711471e-02
6.81473136e-01 3.00820798e-01 1.30625144e-01 -1.04912445e-01
-9.15473759e-01 2.39540190e-01 -2.30518177e-01 4.81646955e-01
7.15699673e-01 -1.40110183e+00 -7.69242227e-01 2.83947051e-01
3.52017283e-01 -2.43253380e-01 1.68607116e-01 4.89491165e-01
-1.33229956e-01 4.03018206e-01 -1.20046183e-01 -9.38900560e-02
-1.17489290e+00 6.91808879e-01 2.93935150e-01 -8.32863212e-01
-3.62667739e-01 1.09977257e+00 3.22588414e-01 -3.79964620e-01
3.81321698e-01 -6.81161225e-01 -4.78459924e-01 2.61315107e-01
5.88095903e-01 1.64902322e-02 2.54902631e-01 -6.47889912e-01
-1.92499787e-01 6.86620295e-01 -2.95636266e-01 -1.60325289e-01
1.41795218e+00 2.61929911e-02 1.39632553e-01 5.21260142e-01
1.67311776e+00 -2.69292921e-01 -8.58415008e-01 -5.89875400e-01
7.67770484e-02 4.49581183e-02 2.66220152e-01 -7.60083318e-01
-8.70113194e-01 1.34188747e+00 4.04715240e-01 -2.61593312e-01
1.17985833e+00 1.07150100e-01 9.35779810e-01 3.17487061e-01
-1.87985301e-01 -1.43885887e+00 4.02932048e-01 8.95769596e-01
1.03094554e+00 -1.04813671e+00 -2.73081541e-01 -2.40638584e-01
-6.79153442e-01 1.15568447e+00 6.58069670e-01 -1.52146697e-01
8.39263260e-01 2.27699906e-01 -3.41322459e-02 -1.05088070e-01
-1.37302625e+00 -2.86370099e-01 3.71673465e-01 1.92597672e-01
8.38408172e-01 1.06947655e-02 -6.15647197e-01 1.27617204e+00
-3.32944661e-01 -4.16776314e-02 1.40468329e-01 6.07932627e-01
-5.48272491e-01 -1.16273546e+00 2.54931450e-01 4.67545837e-01
-3.87947828e-01 -5.00382304e-01 -2.86187112e-01 2.59421736e-01
2.92836010e-01 6.37543619e-01 1.55387446e-01 -6.18404567e-01
7.59007633e-01 4.74661529e-01 3.49307299e-01 -7.81255484e-01
-9.80719864e-01 9.62786842e-03 3.74533415e-01 -5.84280610e-01
-1.89677137e-03 -4.79121298e-01 -8.25005412e-01 2.08624918e-02
-3.95651817e-01 1.13533184e-01 7.47676432e-01 1.15768981e+00
1.27113253e-01 8.05336833e-01 7.84329951e-01 -6.81995213e-01
-5.42904258e-01 -1.38010836e+00 -2.52309144e-01 4.36731845e-01
9.49255452e-02 -4.61183876e-01 -3.35084163e-02 2.72213574e-02] | [10.826109886169434, 8.632277488708496] |
34ab7a78-c716-4bd3-913a-9935a793629f | emergence-of-self-reproducing-metabolisms-as | 2103.08245 | null | https://arxiv.org/abs/2103.08245v3 | https://arxiv.org/pdf/2103.08245v3.pdf | Emergence of Self-Reproducing Metabolisms as Recursive Algorithms in an Artificial Chemistry | One of the main goals of Artificial Life is to research the conditions for the emergence of life, not necessarily as it is, but as it could be. Artificial Chemistries are one of the most important tools for this purpose because they provide us with a basic framework to investigate under which conditions metabolisms capable of reproducing themselves, and ultimately, of evolving, can emerge. While there have been successful attempts at producing examples of emergent self-reproducing metabolisms, the set of rules involved remain too complex to shed much light on the underlying principles at work. In this paper, we hypothesize that the key property needed for self-reproducing metabolisms to emerge is the existence of an auto-catalyzed subset of Turing-complete reactions. We validate this hypothesis with a minimalistic Artificial Chemistry with conservation laws, which is based on a Turing-complete rewriting system called Combinatory Logic. Our experiments show that a single run of this chemistry, starting from a tabula rasa state, discovers -- with no external intervention -- a wide range of emergent structures including ones that self-reproduce in each cycle. All of these structures take the form of recursive algorithms that acquire basic constituents from the environment and decompose them in a process that is remarkably similar to biological metabolisms. | ['Tomas Mikolov', 'Germán Kruszewski'] | 2021-03-15 | null | null | null | null | ['artificial-life'] | ['miscellaneous'] | [ 4.09051389e-01 2.17557564e-01 2.49759972e-01 2.37430513e-01
6.10681295e-01 -9.62511837e-01 9.46874321e-01 2.04508066e-01
6.25391826e-02 9.60293114e-01 -2.25912333e-01 -4.96101737e-01
-1.20338239e-02 -1.03498900e+00 -7.70410717e-01 -1.15839219e+00
-3.35169017e-01 4.95438725e-01 2.66249567e-01 -7.99698114e-01
4.79273833e-02 6.63828075e-01 -1.96375823e+00 -9.39462520e-03
1.00385201e+00 2.05290079e-01 4.03074950e-01 7.69420564e-01
-2.70871334e-02 8.22683513e-01 -3.34402889e-01 -1.92667916e-01
3.86231124e-01 -1.36456716e+00 -7.70898223e-01 1.13936454e-01
-7.04434514e-01 5.00944436e-01 -1.30437657e-01 7.54220426e-01
-1.18145272e-01 2.43535768e-02 9.84866917e-01 -8.07070971e-01
-5.16679049e-01 6.28273845e-01 2.53612008e-02 -1.54164722e-02
3.15105617e-01 5.38863242e-01 7.49253154e-01 -3.38009745e-01
9.02897954e-01 1.07944572e+00 1.87170625e-01 8.29425395e-01
-1.40818477e+00 -9.44052935e-02 -3.78385305e-01 -1.69493809e-01
-1.19093299e+00 -7.70336807e-01 2.44245067e-01 -3.97496700e-01
1.11609399e+00 5.53240001e-01 1.34126961e+00 9.82438326e-01
7.25675404e-01 9.96393487e-02 1.39077151e+00 -6.47378683e-01
5.31607330e-01 1.58243611e-01 -4.33079004e-01 8.96021903e-01
9.65167105e-01 2.67188966e-01 -5.49777448e-01 6.86919913e-02
7.49499738e-01 -3.11671406e-01 -3.63347709e-01 -3.94047439e-01
-1.27306890e+00 3.40630680e-01 -5.14975116e-02 7.48433530e-01
-4.64702845e-01 1.62636682e-01 -1.01327650e-01 3.56849372e-01
-4.71738633e-03 9.95517433e-01 -5.56658804e-01 -3.06251459e-02
-2.65996248e-01 7.15661515e-03 1.31120646e+00 5.79903603e-01
6.90121651e-01 -1.18894801e-02 4.30837989e-01 2.27956712e-01
2.36121476e-01 3.81729126e-01 5.66313386e-01 -6.66266680e-01
-5.16106606e-01 9.49236333e-01 -4.07924876e-02 -5.65331280e-01
-2.96944857e-01 -3.59427750e-01 -6.91290557e-01 3.13125551e-01
5.79143584e-01 -2.27037981e-01 -6.16532981e-01 1.85566545e+00
4.72768575e-01 -2.29495481e-01 4.56454396e-01 5.85039318e-01
1.03559986e-01 1.05808949e+00 -2.07477763e-01 -7.59369075e-01
1.46023846e+00 -5.38065076e-01 -4.73085731e-01 2.81562507e-01
5.00383139e-01 -4.62476999e-01 8.35622132e-01 3.83304715e-01
-1.26809478e+00 1.10631716e-02 -1.21259904e+00 4.77479547e-01
-7.61826754e-01 -6.65533006e-01 1.12905431e+00 8.24756205e-01
-1.17691529e+00 8.13605249e-01 -8.70135725e-01 -1.09212089e+00
-4.49546278e-01 2.01939717e-01 -2.61845529e-01 4.26458448e-01
-1.05448163e+00 1.11632705e+00 5.21573067e-01 1.36136645e-02
-1.13782871e+00 -1.40491933e-01 -3.30494493e-01 -2.99206171e-02
4.62767005e-01 -1.12169385e+00 1.12635958e+00 -1.06182885e+00
-1.69522965e+00 8.85912836e-01 -2.65485704e-01 -2.37279490e-01
2.87144303e-01 6.83598876e-01 -1.84860647e-01 8.47895816e-02
-3.07383329e-01 4.52455521e-01 4.56246704e-01 -1.33762658e+00
-3.56944680e-01 -2.88886845e-01 -2.90596727e-02 6.63785711e-02
-2.19221503e-01 -1.24270292e-02 9.97612625e-02 -1.63788944e-01
3.42940204e-02 -1.10741019e+00 -3.55924755e-01 -3.46299082e-01
-3.60933602e-01 -3.23813289e-01 1.50081083e-01 1.26389757e-01
7.70234525e-01 -1.69547200e+00 6.07893169e-01 1.87001884e-01
1.99171722e-01 -3.32385954e-03 7.03368858e-02 1.12034190e+00
-1.91272155e-03 4.94826078e-01 -4.36113983e-01 3.10336709e-01
-1.88739702e-01 3.55152786e-01 -1.65307477e-01 4.04737413e-01
2.86770642e-01 8.01844895e-01 -1.08059168e+00 -2.01376408e-01
-7.95734152e-02 3.59095901e-01 -4.00313616e-01 4.19018827e-02
-7.54656613e-01 6.35412693e-01 -3.58561873e-01 5.07672250e-01
-1.21732935e-01 -3.48424047e-01 5.90856671e-01 4.68428046e-01
-7.55270481e-01 1.15201533e-01 -8.46478224e-01 1.31714678e+00
-1.05692968e-01 4.76042062e-01 5.81353568e-02 -8.06042552e-01
7.85391152e-01 2.92336345e-01 6.36515498e-01 -4.57306266e-01
2.88591653e-01 4.82027143e-01 5.94767511e-01 -5.05618215e-01
2.31962621e-01 -4.78710830e-01 1.50630921e-01 7.48831034e-01
-1.15673728e-01 -3.43555927e-01 7.83247232e-01 9.74319875e-02
1.37331712e+00 2.09249526e-01 5.14663517e-01 -7.11362064e-01
6.76856160e-01 2.93690801e-01 5.94956934e-01 7.01803625e-01
3.33644710e-02 2.32793048e-01 4.90918636e-01 -6.16224110e-01
-1.37101102e+00 -8.78886878e-01 -7.99362883e-02 6.44478679e-01
3.15235138e-01 -5.11662543e-01 -1.06606591e+00 2.36117139e-01
-4.12742227e-01 6.47393823e-01 -6.37523353e-01 -4.64154065e-01
-3.31788659e-01 -1.06539893e+00 5.01224637e-01 -4.09339279e-01
2.74023026e-01 -1.42889690e+00 -1.13076675e+00 3.97504687e-01
1.49851637e-02 -6.15392029e-01 1.26050293e-01 6.18647754e-01
-8.29121530e-01 -1.30324757e+00 -3.03096563e-01 -5.12439311e-01
7.70240068e-01 2.48579830e-01 9.49388146e-01 5.13475239e-01
-7.72323549e-01 2.30264962e-01 -3.11879039e-01 -5.22504985e-01
-1.16548884e+00 -1.25127798e-02 3.20361942e-01 -8.48578438e-02
-7.07035884e-02 -9.03904259e-01 -4.50305343e-01 1.01834089e-01
-1.15558064e+00 2.37640426e-01 6.23065412e-01 5.40510118e-01
3.36395919e-01 3.96046579e-01 4.23918039e-01 -5.82324982e-01
6.14818692e-01 -4.95824724e-01 -5.41011035e-01 5.94227135e-01
-7.65916467e-01 4.35170293e-01 1.00911415e+00 -3.44636530e-01
-7.05319464e-01 1.98123395e-01 2.39534870e-01 4.40980941e-01
-2.79712915e-01 3.05423915e-01 -2.83799052e-01 1.26654208e-01
8.28790903e-01 8.30758095e-01 1.95446104e-01 -2.41623625e-01
3.36756855e-01 3.69022101e-01 3.01443607e-01 -9.00257051e-01
9.07115877e-01 4.87366647e-01 4.98967856e-01 -1.28850830e+00
-1.42733827e-01 5.57783730e-02 -4.38662648e-01 -3.39397311e-01
1.04480028e+00 -3.64366859e-01 -8.82685244e-01 4.30960953e-01
-1.01149321e+00 -4.43755418e-01 -6.79275155e-01 1.10516824e-01
-7.39122570e-01 1.34803921e-01 -5.07898927e-01 -1.06431723e+00
-8.95447508e-02 -9.21133280e-01 4.61491019e-01 4.57748294e-01
-2.57577151e-01 -7.64088571e-01 6.08200192e-01 -1.59891322e-01
5.23989141e-01 5.03290355e-01 1.20160711e+00 -4.69386160e-01
-8.13792944e-01 1.91077188e-01 5.37711978e-01 -2.79028773e-01
3.87977272e-01 7.06974149e-01 -5.84536254e-01 -4.50011790e-02
1.30579159e-01 2.41456404e-02 6.67061806e-01 -9.84833464e-02
3.21599782e-01 -2.98169672e-01 -4.04987931e-01 5.05778849e-01
1.39507234e+00 6.57587111e-01 7.04940081e-01 4.13518190e-01
1.05514146e-01 9.58512187e-01 -2.11726964e-01 2.08993211e-01
1.59699634e-01 2.90249854e-01 4.59385812e-01 1.05505295e-01
1.20921262e-01 -1.12247661e-01 7.00306118e-01 1.14867806e+00
-5.22207618e-01 -3.12537163e-01 -8.06768656e-01 3.39584887e-01
-1.70012653e+00 -1.13448954e+00 -3.45555633e-01 2.25169373e+00
1.09355593e+00 -8.94784480e-02 3.48227292e-01 1.20073840e-01
6.49124265e-01 -3.20319712e-01 -6.59160674e-01 -6.98794365e-01
-4.90838557e-01 3.31416547e-01 2.41051033e-01 4.05691475e-01
-4.31908220e-01 9.45901573e-01 7.17177582e+00 2.82153748e-02
-1.02280009e+00 -2.87366301e-01 3.28868538e-01 1.45268187e-01
-4.93931293e-01 6.19727075e-01 -3.24174643e-01 2.93197393e-01
1.28254128e+00 -4.47030365e-01 1.05801606e+00 3.69911313e-01
4.90263253e-01 -2.56146669e-01 -1.06601536e+00 3.30847263e-01
-1.75586373e-01 -1.29021227e+00 1.37256876e-01 4.03981894e-01
6.97644174e-01 -2.22972929e-01 -5.02247453e-01 -8.50336030e-02
6.77045941e-01 -1.20066762e+00 8.44216585e-01 6.07592881e-01
4.19141740e-01 -4.56235439e-01 -3.09555363e-02 6.65772140e-01
-8.55892360e-01 -1.06555998e-01 -4.54073548e-01 -4.30840880e-01
1.37875592e-02 6.49114966e-01 -6.96589828e-01 3.12149704e-01
4.00624067e-01 2.26082236e-01 -5.39930284e-01 1.01356673e+00
-2.29955345e-01 4.39115286e-01 -4.51613843e-01 -6.59250736e-01
-7.61870444e-02 -6.31747544e-01 6.74287915e-01 1.03418577e+00
4.38902229e-01 4.11129296e-01 -4.39284623e-01 1.13900399e+00
1.97446436e-01 7.83471465e-02 -1.09352398e+00 -5.41788816e-01
2.84095466e-01 1.32394469e+00 -1.28427315e+00 -3.07578743e-01
8.76897648e-02 8.17124844e-01 1.80617813e-02 1.14937782e-01
-5.78767061e-01 -2.89588064e-01 7.38220930e-01 1.17634729e-01
1.48340091e-01 -4.73941177e-01 -6.89643919e-02 -1.38907123e+00
-4.12575185e-01 -1.06328404e+00 -6.94955327e-03 -6.21372759e-01
-7.96281099e-01 4.05584008e-01 -2.72514164e-01 -4.50043380e-01
-2.86465108e-01 -3.62471253e-01 -6.67704344e-01 4.30282354e-01
-1.02967107e+00 -6.56245291e-01 -1.97793871e-01 4.20422584e-01
3.63489300e-01 -4.23755608e-02 9.40754652e-01 -5.10132909e-01
-8.12211096e-01 -7.43375272e-02 3.44706714e-01 -5.77355742e-01
1.76825181e-01 -1.20804381e+00 2.71040410e-01 1.01357615e+00
3.24184448e-01 9.59246218e-01 1.15408278e+00 -6.64061546e-01
-2.16540074e+00 -5.97917259e-01 9.37795579e-01 -4.56946492e-01
4.95674253e-01 -7.20058739e-01 -5.24683297e-01 2.18525127e-01
4.54590142e-01 -6.22903466e-01 5.18067360e-01 -1.96879715e-01
-7.80171007e-02 1.64286536e-03 -9.64940548e-01 1.12723315e+00
1.39110029e+00 -1.00209512e-01 -4.21689600e-01 4.45160061e-01
7.93289185e-01 1.93582803e-01 -7.93442011e-01 1.85657397e-01
6.72460198e-01 -1.22522509e+00 6.75212920e-01 -7.48516262e-01
5.37899315e-01 -6.94113135e-01 1.92638665e-01 -1.19571888e+00
-4.54106629e-01 -1.27078128e+00 1.07286219e-02 9.50264037e-01
6.05923295e-01 -9.84243989e-01 1.85179129e-01 3.47882569e-01
-3.19446065e-02 -5.25329232e-01 -5.58648407e-01 -1.10626388e+00
1.73944026e-01 4.07711625e-01 7.58197963e-01 9.52431798e-01
4.87975478e-01 4.94780898e-01 1.18123479e-01 -3.55498761e-01
5.03781259e-01 1.05773866e-01 6.52704537e-01 -1.18550122e+00
-5.16432524e-01 -7.08993733e-01 -2.08478227e-01 -4.33864266e-01
-9.82356220e-02 -9.19912219e-01 2.89327532e-01 -1.34434533e+00
2.11754814e-01 -2.24075660e-01 -1.04174890e-01 3.29240412e-01
2.49733239e-01 -1.66077048e-01 5.77331632e-02 5.01574755e-01
-3.47025603e-01 3.59511942e-01 1.20140707e+00 2.83710331e-01
-3.26780975e-01 -4.45230633e-01 -1.00096178e+00 4.29734141e-01
9.38183546e-01 -5.68544269e-01 -5.09430289e-01 1.41092747e-01
7.55179167e-01 -8.46931059e-03 -1.08357400e-01 -1.04320240e+00
2.21228659e-01 -6.05215788e-01 3.69748697e-02 2.12805659e-01
-1.00825150e-02 -6.09976232e-01 8.50957572e-01 1.00077355e+00
-2.36452013e-01 2.99673468e-01 -1.64372653e-01 5.39405584e-01
4.48838562e-01 -4.13544714e-01 6.94078326e-01 -6.97825372e-01
-1.25926390e-01 -7.09453449e-02 -1.28805852e+00 -1.85321927e-01
1.34862256e+00 -3.16992283e-01 -4.31271374e-01 -7.02357814e-02
-4.31060761e-01 -1.94464758e-01 1.39666867e+00 1.01741299e-01
5.85570112e-02 -7.28970706e-01 -6.03347778e-01 1.09447747e-01
-4.88346405e-02 -1.71074733e-01 -5.63691378e-01 6.39213204e-01
-1.05596459e+00 5.10988355e-01 -3.74322206e-01 -4.24615413e-01
-7.01761425e-01 8.31315339e-01 6.12545073e-01 9.84437112e-03
-4.92337734e-01 3.98435801e-01 3.30808640e-01 -1.67203307e-01
-3.60719889e-01 -2.22965702e-01 1.58215657e-01 -2.67972231e-01
3.62254918e-01 2.40427675e-03 -1.78883418e-01 -3.54757875e-01
-2.92679131e-01 3.08320820e-01 4.61141527e-01 -2.25608453e-01
1.54797673e+00 -3.17171700e-02 -9.87739325e-01 7.19136059e-01
3.96886438e-01 1.88114028e-02 -8.20143223e-01 5.71597397e-01
1.28794194e-03 -2.10892465e-02 -6.46134734e-01 -6.74677670e-01
-5.70279300e-01 4.54469532e-01 -1.27189994e-01 9.28437710e-01
1.02541530e+00 4.00082059e-02 3.80413681e-01 7.76924014e-01
5.70496321e-01 -9.05489624e-01 -6.85060769e-02 3.76337558e-01
6.13154113e-01 -4.86289501e-01 -1.66578174e-01 -2.17764825e-01
7.97704421e-03 1.21648955e+00 2.00351611e-01 -6.27951920e-02
2.37849802e-01 2.92884320e-01 -4.23074126e-01 -2.65681446e-01
-1.49259245e+00 -4.55249012e-01 -4.78443146e-01 6.83523118e-01
5.49043298e-01 5.36992215e-02 -8.41058910e-01 4.95363809e-02
-2.08228752e-01 -2.81020671e-01 9.65909183e-01 9.96798575e-01
-7.87451684e-01 -1.33973157e+00 -3.74271542e-01 -2.89037870e-03
-2.08553538e-01 1.47036865e-01 -1.22089565e+00 6.82827413e-01
1.86155677e-01 1.11716008e+00 -3.22776467e-01 -2.98106432e-01
-8.00564885e-02 3.93059999e-01 9.15357530e-01 -5.18927932e-01
-8.04509163e-01 -2.39804626e-01 2.16648608e-01 -1.57647908e-01
-3.69445384e-01 -9.81135309e-01 -1.31490874e+00 -6.58046365e-01
-4.11630183e-01 5.30570626e-01 8.76475513e-01 8.24351132e-01
3.05547684e-01 4.37173814e-01 7.67398179e-01 -5.79200745e-01
-3.20617437e-01 -4.37118948e-01 -6.09705627e-01 2.83139259e-01
-6.56110495e-02 -3.49035114e-01 -4.94596064e-01 3.72536749e-01] | [5.5981245040893555, 4.175902843475342] |
0e4cf482-6f05-4753-9ccb-c3ab5f0089ab | risk-sharing-measuring-variability-and | 2302.04034 | null | https://arxiv.org/abs/2302.04034v1 | https://arxiv.org/pdf/2302.04034v1.pdf | Risk sharing, measuring variability, and distortion riskmetrics | We address the problem of sharing risk among agents with preferences modelled by a general class of comonotonic additive and law-based functionals that need not be either monotone or convex. Such functionals are called distortion riskmetrics, which include many statistical measures of risk and variability used in portfolio optimization and insurance. The set of Pareto-optimal allocations is characterized under various settings of general or comonotonic risk sharing problems. We solve explicitly Pareto-optimal allocations among agents using the Gini deviation, the mean-median deviation, or the inter-quantile difference as the relevant variability measures. The latter is of particular interest, as optimal allocations are not comonotonic in the presence of inter-quantile difference agents; instead, the optimal allocation features a mixture of pairwise counter-monotonic structures, showing some patterns of extremal negative dependence. | ['Ruodu Wang', 'Liyuan Lin', 'Jean-Gabriel Lauzier'] | 2023-02-08 | null | null | null | null | ['portfolio-optimization'] | ['time-series'] | [-4.65576261e-01 2.21731439e-01 -4.81715471e-01 -2.90964574e-01
-6.02607310e-01 -1.00195181e+00 6.13442242e-01 7.79981613e-02
-5.17099977e-01 9.13767040e-01 5.49807191e-01 -2.70789146e-01
-1.05778635e+00 -8.00957382e-01 -1.45909891e-01 -9.42491353e-01
-4.23289001e-01 5.85874438e-01 -1.29596740e-01 -1.89415842e-01
2.70331830e-01 5.68691015e-01 -1.25020623e+00 -8.89761448e-02
1.15641642e+00 1.09900522e+00 -4.89516377e-01 3.16917360e-01
1.35246307e-01 4.31684554e-01 -6.77662373e-01 -8.09928894e-01
5.89937329e-01 -2.70194143e-01 -1.86405346e-01 -2.86399841e-01
-2.63698816e-01 -1.63476825e-01 3.19523126e-01 1.42802906e+00
2.17232332e-01 1.42267302e-01 1.37048519e+00 -1.45166290e+00
-8.40700626e-01 8.04504931e-01 -3.25743169e-01 3.10186952e-01
-7.33624175e-02 -1.10179946e-01 1.29440105e+00 -3.06352973e-01
1.01556741e-01 1.31363654e+00 3.32854748e-01 4.46647465e-01
-1.24517488e+00 -1.15684040e-01 -1.64228119e-02 -2.76305407e-01
-1.03305888e+00 -2.42425911e-02 2.30106816e-01 -8.22322071e-01
4.11308646e-01 7.10303068e-01 4.35238123e-01 6.34562731e-01
6.16641402e-01 3.89508880e-03 1.17218316e+00 -1.64553955e-01
7.12330639e-01 2.19903216e-01 -1.13526471e-01 1.64790332e-01
9.34356034e-01 5.45927048e-01 1.24375135e-01 -6.49406612e-01
6.06216729e-01 1.33318156e-01 -2.55742997e-01 -6.67642236e-01
-7.56085515e-01 1.09920442e+00 6.15802361e-03 3.03523570e-01
-5.49417436e-01 2.08842624e-02 1.29047930e-01 6.08707845e-01
7.32935369e-01 3.27813447e-01 -2.55874991e-01 1.68960065e-01
-3.19022357e-01 5.23431361e-01 8.16609561e-01 6.17780030e-01
2.81356454e-01 5.42139895e-02 -4.77621049e-01 5.18568933e-01
4.26415354e-01 5.42612553e-01 1.14758328e-01 -1.10304606e+00
7.04395890e-01 2.95808911e-01 7.82216609e-01 -6.78579152e-01
-4.80838358e-01 -4.76073772e-01 -7.10308194e-01 7.85166383e-01
9.01594222e-01 -5.08715868e-01 -2.58840993e-02 2.10340714e+00
6.92501441e-02 -4.52416837e-01 1.40923962e-01 7.70445406e-01
-2.39458054e-01 2.46717885e-01 3.11369151e-02 -9.39218521e-01
1.05279648e+00 -3.58652264e-01 -7.02324271e-01 4.25934792e-01
5.32470882e-01 -4.37711924e-01 6.51626945e-01 1.73205450e-01
-1.34436381e+00 2.88066149e-01 -5.28685391e-01 6.37790501e-01
-1.28869027e-01 -7.66466618e-01 8.06710348e-02 1.14965272e+00
-9.08661425e-01 8.02078962e-01 -8.93241838e-02 3.42928082e-01
3.99367094e-01 6.08964078e-03 5.93974888e-02 3.48535031e-01
-9.04094577e-01 9.06415761e-01 5.46152964e-02 -3.69861454e-01
-7.92258382e-01 -9.96828496e-01 -5.74864626e-01 4.50570613e-01
4.79637057e-01 -5.35260260e-01 1.10296845e+00 -9.71143126e-01
-1.36386442e+00 6.84545696e-01 6.38357997e-01 -4.28883553e-01
1.05514836e+00 -1.72610339e-02 -2.59734154e-01 -1.82981670e-01
1.73556861e-02 -5.26465595e-01 5.32484472e-01 -9.70986784e-01
-2.94363886e-01 -5.61957479e-01 4.51434255e-01 3.88945788e-01
-4.10437817e-03 4.37736452e-01 8.39869499e-01 -9.91109550e-01
-4.81346399e-01 -7.71167338e-01 -3.67540985e-01 -1.71591923e-01
-4.08528149e-01 -2.60157794e-01 -1.87407613e-01 -3.73601764e-01
1.15587795e+00 -2.21348739e+00 -2.37394813e-02 4.43013251e-01
-2.83017218e-01 -3.86882842e-01 2.38855835e-02 4.09126014e-01
-1.89660639e-01 3.95421505e-01 -5.48291385e-01 4.69088256e-02
8.19239974e-01 -3.18678133e-02 -9.65217650e-02 9.32886779e-01
-1.82094604e-01 7.33543873e-01 -5.70352435e-01 1.51271924e-01
-1.38585269e-01 -7.06014261e-02 -4.85625029e-01 1.02655090e-01
-2.83602506e-01 2.01480404e-01 -4.47572052e-01 2.08862945e-01
5.74739039e-01 1.85681313e-01 2.60361254e-01 4.40541744e-01
-4.08249468e-01 1.59377635e-01 -1.32709110e+00 7.54053771e-01
-2.24627942e-01 -4.73865122e-03 4.93774936e-02 -1.16375327e+00
6.69534981e-01 4.58569497e-01 6.26665890e-01 -3.54081810e-01
4.22119588e-01 3.18162411e-01 2.45277956e-01 -2.25888535e-01
2.49126807e-01 -9.08376098e-01 -3.29386562e-01 1.00296628e+00
-3.07198495e-01 1.52563006e-01 2.11766716e-02 -1.92286313e-01
5.61151981e-01 -5.46624601e-01 6.79665983e-01 -1.29970193e+00
3.70973200e-01 -5.55297136e-01 7.39385009e-01 3.69075209e-01
-1.39223978e-01 4.87721741e-01 1.22384083e+00 4.26496975e-02
-9.69282031e-01 -1.44589758e+00 -2.05509827e-01 9.20190036e-01
5.29279783e-02 4.21808094e-01 -7.45234370e-01 -5.31056404e-01
4.56947505e-01 9.86867309e-01 -7.24113941e-01 -1.79426402e-01
-3.23966369e-02 -1.17866004e+00 9.26343128e-02 2.02593327e-01
-1.03804665e-02 -7.63772964e-01 -6.06351256e-01 -3.88746941e-03
3.24014246e-01 -2.57738739e-01 -9.81721163e-01 -1.63035020e-01
-5.89677334e-01 -1.21211267e+00 -1.11479008e+00 9.79980081e-02
1.88912079e-01 -2.27135926e-01 1.08386028e+00 -3.69208336e-01
1.29787087e-01 6.17885470e-01 1.53543010e-01 -7.08004296e-01
-1.87957183e-01 -7.61387706e-01 2.85589278e-01 5.19126415e-01
-1.30835474e-01 -4.11810637e-01 -6.63668156e-01 5.20932078e-01
-9.04677153e-01 -9.81604040e-01 -2.15855971e-01 4.13564831e-01
4.08584833e-01 5.48816696e-02 1.02163327e+00 -5.28714836e-01
9.82909441e-01 -8.17855656e-01 -1.09959638e+00 4.89094377e-01
-6.66110873e-01 1.08601183e-01 4.19482857e-01 -2.19105899e-01
-1.29373980e+00 -9.07172859e-01 6.14119947e-01 1.65171057e-01
2.14734256e-01 4.79342461e-01 -7.09347367e-01 1.84028670e-01
3.30714732e-01 -4.65509057e-01 -5.15823811e-02 -6.60529912e-01
2.24132910e-01 4.58554476e-01 1.98904797e-01 -6.16338551e-01
5.50446808e-01 6.33324146e-01 1.40411422e-01 -3.94590914e-01
-7.01839209e-01 -6.15928546e-02 -8.55471268e-02 3.63945886e-02
1.05429864e+00 -3.89941841e-01 -9.34348047e-01 4.10225183e-01
-8.18572521e-01 -2.60719985e-01 -9.56011295e-01 6.58271849e-01
-1.11018991e+00 2.33531147e-01 -3.34671468e-01 -1.50286293e+00
-1.09742666e-02 -9.14693654e-01 2.06835091e-01 2.45042086e-01
5.57698235e-02 -1.40328264e+00 3.24820220e-01 -5.47618186e-03
4.63889748e-01 4.49061483e-01 1.23356605e+00 -8.76367331e-01
-2.82661855e-01 -6.30872790e-03 3.69429030e-02 2.70096004e-01
1.03513740e-01 -5.98588549e-02 -5.52207351e-01 -1.60288185e-01
2.94645160e-01 2.77033985e-01 5.22766829e-01 7.79495597e-01
7.20317543e-01 -9.03866887e-01 1.79921389e-01 3.95371675e-01
1.43448067e+00 6.66737080e-01 2.67817289e-01 1.51537627e-01
3.21100503e-02 1.12189794e+00 4.39145178e-01 7.83551514e-01
3.96302521e-01 7.27734208e-01 6.60057306e-01 7.32129276e-01
9.10147965e-01 8.93356800e-02 4.91452008e-01 9.92845073e-02
-3.83114964e-01 -2.40375742e-01 -5.77071011e-01 4.01977539e-01
-1.84567773e+00 -1.39930284e+00 2.64162961e-02 2.77255177e+00
5.19811809e-01 -6.20014593e-02 8.95933509e-01 -7.45510980e-02
8.26150894e-01 3.37989368e-02 -4.66050863e-01 -9.85871255e-01
-2.97522068e-01 -1.00820854e-01 7.08524883e-01 7.61867881e-01
-8.29861939e-01 -7.60720000e-02 6.48667526e+00 5.93792021e-01
-4.64456588e-01 2.30061039e-01 7.11764634e-01 -3.37215722e-01
-8.11649561e-01 -1.88503176e-01 -3.98103863e-01 8.47144783e-01
8.13884139e-01 -1.02489495e+00 1.58103570e-01 6.64548457e-01
3.02131593e-01 -5.42991124e-02 -1.01989245e+00 5.73152840e-01
-4.58907247e-01 -1.07834721e+00 -5.53454936e-01 6.74613237e-01
1.16484010e+00 -4.14868444e-01 4.28373903e-01 -3.23948741e-01
8.75839770e-01 -1.17028916e+00 1.06613135e+00 9.41490352e-01
5.58412194e-01 -1.27998173e+00 7.54478335e-01 3.92423868e-01
-7.08144665e-01 -4.26145911e-01 -4.42361116e-01 -2.65553445e-01
5.41161180e-01 9.09887850e-01 3.52042198e-01 7.08862841e-01
4.53180254e-01 2.82223076e-01 6.60101846e-02 1.35671735e+00
1.34902999e-01 4.41830121e-02 -2.35524297e-01 1.51076123e-01
1.68384582e-01 -9.37944531e-01 6.04979753e-01 6.67645812e-01
7.59593010e-01 1.95112184e-01 -1.25326827e-01 8.96934986e-01
5.39672002e-02 1.42957866e-01 -7.36692965e-01 1.26416728e-01
4.39949989e-01 8.35842729e-01 -4.59650397e-01 -4.62526008e-02
-4.68314379e-01 2.37799287e-01 -1.31723240e-01 3.20060134e-01
-6.57985330e-01 1.93767343e-02 1.38255453e+00 -8.76828749e-03
6.80113956e-02 5.14222891e-04 -6.23438895e-01 -1.11171496e+00
7.36290589e-02 -4.69160825e-01 9.69393969e-01 -3.20116505e-02
-1.78009987e+00 1.54036298e-01 3.99411947e-01 -1.05232441e+00
-2.74625570e-01 -5.71774423e-01 -1.07814670e+00 1.09682727e+00
-1.19082224e+00 -2.26021066e-01 5.38076997e-01 5.09169579e-01
-1.06449254e-01 -4.10062045e-01 4.74465758e-01 -7.68087879e-02
-4.07340586e-01 4.45116788e-01 6.80796683e-01 -4.91481036e-01
1.25522390e-01 -1.61275661e+00 -2.59949148e-01 5.42352915e-01
-9.84705836e-02 2.93660462e-01 7.49896407e-01 -4.40442920e-01
-5.43052793e-01 -9.47921872e-01 7.30136991e-01 -4.70982283e-01
9.15526807e-01 2.01699194e-02 -6.93722248e-01 5.91639340e-01
2.58935273e-01 -2.03518853e-01 7.03047335e-01 -1.99051365e-01
-3.10595036e-01 -3.98183987e-02 -1.55890298e+00 6.04923487e-01
1.12199104e+00 -2.57046431e-01 -5.17706454e-01 4.13464963e-01
6.22972965e-01 5.04982710e-01 -9.07363772e-01 3.17505926e-01
4.74168390e-01 -1.45669234e+00 9.62167561e-01 -7.99336731e-01
3.43198404e-02 1.89202905e-01 -4.80189592e-01 -1.58488023e+00
-4.17833626e-01 -8.08600008e-01 3.99265438e-01 1.17473340e+00
3.42597395e-01 -1.18775439e+00 3.05716187e-01 6.29322529e-01
1.51494265e-01 -6.70654178e-01 -1.41412055e+00 -1.35185528e+00
8.31633329e-01 -1.81233957e-01 9.64252353e-01 8.46071720e-01
3.64078075e-01 -2.28171572e-01 -2.06871077e-01 -1.72717705e-01
1.13311076e+00 1.27348125e-01 -4.52547707e-02 -1.27030921e+00
-3.37113470e-01 -1.01251602e+00 -1.29894629e-01 -1.36553243e-01
2.16821238e-01 -9.08047616e-01 -2.38480672e-01 -1.00081432e+00
1.11376800e-01 -4.86874968e-01 -4.48327065e-01 -2.27367356e-01
1.71630532e-01 -1.10567115e-01 3.95518333e-01 -2.33392250e-02
-2.03019023e-01 7.11380303e-01 8.72941375e-01 1.57494955e-02
-1.84521034e-01 6.10094965e-01 -9.47428167e-01 1.02407336e+00
9.33965385e-01 -3.63382846e-01 -5.67179203e-01 3.64140384e-02
4.62750167e-01 3.95304590e-01 3.64715219e-01 -3.73812199e-01
-2.65114576e-01 -1.12197256e+00 -3.58802408e-01 -1.89082921e-01
-3.93252037e-02 -7.77069271e-01 6.32863760e-01 6.01407886e-01
-4.90294129e-01 3.83822501e-01 -5.08524239e-01 5.62523067e-01
1.29016921e-01 -6.17481649e-01 1.16400969e+00 -1.83625557e-02
2.54130065e-01 3.80478919e-01 -6.42291188e-01 4.78529423e-01
1.51168609e+00 8.64306986e-02 -5.54060042e-01 -6.60242379e-01
-6.67756438e-01 7.77167007e-02 5.86093843e-01 1.81473583e-01
1.11752450e-01 -1.58121300e+00 -8.13037932e-01 -4.20865029e-01
-7.70732388e-02 -4.46838468e-01 2.85327494e-01 7.42659688e-01
-7.81029239e-02 5.09177387e-01 -2.37890407e-01 1.78745136e-01
-8.58654916e-01 8.55475724e-01 8.86314511e-01 -3.92546356e-01
-1.17661007e-01 6.49868727e-01 9.63526964e-01 -1.75260589e-01
-5.15171811e-02 -1.53123483e-01 -3.38115662e-01 4.90123600e-01
6.41745210e-01 9.99177337e-01 -3.14278543e-01 -6.08667970e-01
-4.21095788e-01 4.49943632e-01 4.59209591e-01 -5.66019297e-01
1.32711291e+00 -2.66949147e-01 -2.36980826e-01 4.43360925e-01
9.15148795e-01 1.15822330e-01 -1.33012760e+00 -4.24182899e-02
6.13714039e-01 -4.29159522e-01 -6.03031635e-01 -7.48784423e-01
-9.94207561e-01 5.98740697e-01 1.71420023e-01 9.37731564e-01
7.02511728e-01 1.11592382e-01 -6.48207068e-02 -3.00353914e-01
5.39953709e-01 -1.07127404e+00 1.26699370e-03 3.56590867e-01
1.39414036e+00 -6.64264381e-01 -3.32438499e-01 -2.36248702e-01
-7.66594887e-01 5.73876143e-01 -3.60623486e-02 -5.60413301e-01
1.31516159e+00 3.50880265e-01 -1.31353363e-01 1.18509017e-01
-7.17167497e-01 -3.17994863e-01 4.83425200e-01 8.59440267e-01
4.60351221e-02 7.11528480e-01 -9.88936663e-01 1.10607815e+00
-3.42971087e-01 -5.79338253e-01 6.41516149e-01 4.91532385e-01
-3.62505913e-01 -8.59232128e-01 -5.30294478e-01 5.07017314e-01
-8.00875366e-01 1.38222709e-01 -1.50433734e-01 4.19987500e-01
-1.01628579e-01 9.13295150e-01 4.08697486e-01 2.66367376e-01
5.02831757e-01 -6.61233217e-02 3.40338409e-01 -5.23395479e-01
-4.77005929e-01 -5.05980179e-02 -5.74561842e-02 -3.18727463e-01
-3.73712897e-01 -8.82456779e-01 -6.06308818e-01 -5.47989488e-01
2.61691690e-04 1.22416459e-01 1.38144806e-01 1.02084851e+00
-1.75085649e-01 3.64991389e-02 8.57873499e-01 -4.56493437e-01
-1.36002803e+00 -7.18494117e-01 -1.33398950e+00 3.71371329e-01
3.70394319e-01 -7.49223650e-01 -6.95960402e-01 -5.53451002e-01] | [4.968071460723877, 3.8927197456359863] |
73d22a27-89c8-4b40-9cc5-589bdc1f0c3f | graph-convolutional-networks-for-event | null | null | https://aclanthology.org/2021.naacl-main.273 | https://aclanthology.org/2021.naacl-main.273.pdf | Graph Convolutional Networks for Event Causality Identification with Rich Document-level Structures | We study the problem of Event Causality Identification (ECI) to detect causal relation between event mention pairs in text. Although deep learning models have recently shown state-of-the-art performance for ECI, they are limited to the intra-sentence setting where event mention pairs are presented in the same sentences. This work addresses this issue by developing a novel deep learning model for document-level ECI (DECI) to accept inter-sentence event mention pairs. As such, we propose a graph-based model that constructs interaction graphs to capture relevant connections between important objects for DECI in input documents. Such interaction graphs are then consumed by graph convolutional networks to learn document context-augmented representations for causality prediction between events. Various information sources are introduced to enrich the interaction graphs for DECI, featuring discourse, syntax, and semantic information. Our extensive experiments show that the proposed model achieves state-of-the-art performance on two benchmark datasets. | ['Thien Huu Nguyen', 'Minh Tran Phu'] | 2021-06-01 | null | null | null | naacl-2021-4 | ['event-causality-identification'] | ['natural-language-processing'] | [ 3.04328889e-01 5.28080881e-01 -2.56514311e-01 -4.64541316e-01
-4.16400582e-01 -4.77908701e-01 1.01537108e+00 9.31110620e-01
-6.64389208e-02 6.19773686e-01 1.02156436e+00 -5.48486352e-01
-3.52682084e-01 -1.06135893e+00 -9.00086701e-01 1.16069555e-01
-5.86926997e-01 5.02256572e-01 2.90733248e-01 -1.68868899e-01
2.61873990e-01 2.70130068e-01 -1.02656472e+00 8.99935246e-01
7.80978024e-01 5.34973145e-01 -1.44299760e-01 8.07504117e-01
-4.72825795e-01 1.71421969e+00 -8.59854162e-01 -5.49860358e-01
-6.24224484e-01 -6.20662928e-01 -1.34943926e+00 -3.30247521e-01
8.70983005e-02 -2.34061062e-01 -8.11336339e-01 7.54945993e-01
1.94362327e-01 1.15021266e-01 8.03836763e-01 -1.34920406e+00
-8.70271504e-01 1.55619276e+00 -3.69785309e-01 7.07009971e-01
7.82658339e-01 -4.38658357e-01 1.56787097e+00 -7.50770390e-01
8.57708693e-01 1.75081778e+00 1.96703956e-01 9.97275710e-02
-6.03903711e-01 -4.86246824e-01 7.03537285e-01 8.84802997e-01
-7.40322292e-01 7.20941499e-02 1.03144014e+00 -2.14898691e-01
1.55667877e+00 2.19639778e-01 3.49883467e-01 1.26375818e+00
3.83768708e-01 1.18131530e+00 1.39839038e-01 -3.38864893e-01
-4.11297344e-02 -7.11236238e-01 7.41706133e-01 6.13958359e-01
1.36211172e-01 -2.60935754e-01 -6.71326399e-01 3.25887911e-02
4.70605731e-01 -2.20605433e-01 -2.44468406e-01 3.80006939e-01
-1.51253796e+00 9.99373138e-01 7.63185322e-01 4.86691147e-01
-4.98626947e-01 4.15864199e-01 8.64547729e-01 1.96254298e-01
4.02525514e-01 1.69658124e-01 -3.25720996e-01 1.60751820e-01
-2.34569296e-01 3.23857427e-01 8.26116264e-01 9.44098949e-01
7.12776557e-02 -3.24799389e-01 -7.14858234e-01 5.45409739e-01
3.67528290e-01 -1.51472107e-01 2.89703637e-01 -2.26484001e-01
1.00797176e+00 1.27283216e+00 -1.78504214e-01 -1.58808994e+00
-8.26052010e-01 -3.61283392e-01 -9.97855663e-01 -5.02366602e-01
1.78758547e-01 -2.17242464e-01 -5.02213299e-01 1.71711218e+00
3.61445993e-01 6.01912796e-01 2.35569105e-01 8.39683115e-01
1.50524902e+00 9.01679695e-01 3.85969400e-01 -2.48461664e-01
1.21761906e+00 -8.96419764e-01 -1.04126203e+00 -3.76367718e-01
8.40068519e-01 -4.73804623e-01 8.52649987e-01 -7.35808685e-02
-1.03766263e+00 -3.83103222e-01 -9.01841164e-01 -3.21755499e-01
-3.55895430e-01 -5.35917692e-02 7.00244665e-01 -2.12589920e-01
-7.19772160e-01 4.57733124e-01 -4.50588226e-01 -3.59691590e-01
3.08428138e-01 1.85978696e-01 -1.02903523e-01 1.72584653e-01
-2.02927017e+00 6.09897554e-01 7.95464814e-01 2.81378806e-01
-8.62012565e-01 -7.87450790e-01 -1.10802722e+00 4.78106350e-01
6.02979481e-01 -5.46669364e-01 1.13187289e+00 -6.94477379e-01
-9.34485197e-01 6.86258376e-01 -2.26036981e-01 -6.74582720e-01
1.56577140e-01 -3.35190922e-01 -8.33353341e-01 2.41470501e-01
1.51770249e-01 -3.29058873e-03 3.57360214e-01 -9.96581078e-01
-6.32613778e-01 -3.06944102e-01 4.05582190e-01 2.30825186e-01
-2.59328961e-01 6.30252600e-01 -4.63385493e-01 -8.11966181e-01
-1.33160844e-01 -3.44083309e-01 9.35156178e-03 -6.25471175e-01
-8.59093368e-01 -1.04932201e+00 9.69950676e-01 -7.32046604e-01
1.64924705e+00 -1.84899485e+00 1.06674634e-01 -1.32644013e-01
4.63628620e-01 2.42608190e-02 -2.99857706e-01 6.99107170e-01
-5.46385467e-01 2.81608313e-01 -1.43147809e-02 -1.24587409e-01
-1.86787266e-02 1.19568937e-01 -5.15344799e-01 -6.08637324e-03
6.73482716e-01 1.14596117e+00 -1.48369408e+00 -6.53596342e-01
5.57334116e-03 3.11200619e-01 -2.85680264e-01 5.04084647e-01
-5.59273660e-01 2.67167330e-01 -4.87190872e-01 1.27929114e-02
1.80144385e-01 -7.03571320e-01 6.32065117e-01 -2.35528529e-01
3.14281374e-01 9.62504625e-01 -1.12283111e+00 1.41512167e+00
-3.08373928e-01 5.92022836e-01 -5.34627140e-01 -1.39725745e+00
8.49780738e-01 6.37976348e-01 2.74490744e-01 -5.48957288e-01
8.22531655e-02 -1.60364732e-01 7.36050680e-02 -4.84451294e-01
4.41558391e-01 3.82798404e-01 -3.06877971e-01 3.95414025e-01
1.34080514e-01 4.78957504e-01 4.96275991e-01 8.56355190e-01
1.21388113e+00 -1.93051174e-01 3.89005899e-01 -3.26859429e-02
7.38334000e-01 -1.43044636e-01 4.68832463e-01 7.34889805e-01
6.81654364e-02 4.48557049e-01 1.15539491e+00 -4.35859919e-01
-5.24421871e-01 -9.09059286e-01 5.03002882e-01 9.80251312e-01
1.56852290e-01 -5.70012271e-01 -4.66055244e-01 -1.15760958e+00
-3.05667430e-01 1.10171139e+00 -8.62228811e-01 -1.85158893e-01
-1.04193199e+00 -6.89377964e-01 3.59744161e-01 9.71349537e-01
5.24827302e-01 -1.36319017e+00 -1.47684321e-01 5.67827284e-01
-6.69248402e-01 -1.12372005e+00 -5.40764093e-01 3.44208367e-02
-4.35247272e-01 -1.46299875e+00 -2.05680188e-02 -1.00606835e+00
3.03372651e-01 -5.73721938e-02 1.63961697e+00 9.85811725e-02
-3.14300209e-02 3.06138188e-01 -6.03674889e-01 -4.84065443e-01
-5.61969519e-01 -4.32029031e-02 -4.39038903e-01 7.20492676e-02
3.85667294e-01 -3.43458176e-01 -6.49219036e-01 -1.75034069e-02
-9.16023433e-01 1.06315715e-02 6.46841675e-02 7.54357040e-01
4.56407994e-01 2.91489989e-01 9.85150635e-01 -1.08777511e+00
8.79901409e-01 -8.65312040e-01 -2.09482744e-01 5.28107107e-01
-2.33583733e-01 9.02115107e-02 8.81745338e-01 -1.92961887e-01
-1.49072552e+00 -5.35353720e-01 9.13859606e-02 -8.70909821e-03
-9.94953737e-02 1.30647242e+00 -4.24025923e-01 8.89734924e-01
5.13247728e-01 -2.94295818e-01 -7.12791443e-01 -1.60716176e-02
5.67779183e-01 3.19314152e-01 7.00942576e-01 -4.22948003e-01
9.88945067e-02 1.08903863e-01 1.49210515e-02 -3.72120380e-01
-1.25499725e+00 -3.67346764e-01 -6.21794760e-01 -3.00274819e-01
1.06579816e+00 -7.37369537e-01 -8.23435485e-01 2.15964600e-01
-1.74987316e+00 -3.42976362e-01 2.83639673e-02 3.48101169e-01
-2.71013319e-01 3.51736695e-01 -6.42065585e-01 -5.28869987e-01
-5.33724368e-01 -6.61429882e-01 1.01491916e+00 1.46630481e-01
-4.34962124e-01 -1.36628723e+00 7.03930035e-02 1.11049198e-01
-3.84422213e-01 5.08755445e-01 1.36069810e+00 -1.04181564e+00
-4.17267084e-01 -1.51288286e-01 -4.79915708e-01 -3.67583752e-01
1.83011100e-01 9.93511453e-02 -5.23324430e-01 1.72357082e-01
-4.37655360e-01 -2.09191918e-01 9.66145754e-01 4.22899246e-01
1.42220736e+00 -5.68911135e-01 -6.79723740e-01 1.75183956e-02
9.99577820e-01 3.17079931e-01 7.08982170e-01 1.45169437e-01
1.01697648e+00 5.55046499e-01 5.60540497e-01 4.57212299e-01
8.20271492e-01 3.18992168e-01 6.99517429e-01 -2.66139582e-02
-4.09921259e-01 -4.32120949e-01 7.50464201e-02 6.77857399e-01
3.54467742e-02 -1.00672996e+00 -1.13475168e+00 1.00663900e+00
-2.25711608e+00 -1.07702816e+00 -8.54581177e-01 1.60188198e+00
8.31871271e-01 3.03630501e-01 -3.04260496e-02 1.97095200e-01
8.96922529e-01 4.06397611e-01 -2.61272758e-01 -3.76156062e-01
-1.39745429e-01 -4.73219156e-02 -3.84287983e-02 5.36785841e-01
-1.36181545e+00 8.88580143e-01 5.49234533e+00 3.57624561e-01
-7.10365951e-01 -5.38906502e-03 7.72877514e-01 3.42611700e-01
-5.47557414e-01 -1.45507023e-01 -6.79427862e-01 3.30056518e-01
1.01323903e+00 -3.43626976e-01 -1.00197762e-01 4.71071929e-01
2.52560228e-01 3.06326151e-01 -1.29221499e+00 5.05943716e-01
-3.37849557e-03 -1.87779891e+00 3.66656095e-01 -5.02030313e-01
5.95718920e-01 -2.40475044e-01 -3.88959527e-01 2.71061122e-01
5.98789632e-01 -9.47514534e-01 4.94960994e-01 4.96048123e-01
4.65508789e-01 -8.50308239e-01 7.56407797e-01 -1.52629510e-01
-1.48698080e+00 -1.75287440e-01 -1.29114792e-01 -2.69228131e-01
2.79690385e-01 7.25648701e-01 -1.11333609e+00 1.12798452e+00
5.66267192e-01 1.11112177e+00 -5.80228567e-01 5.22257328e-01
-8.43818665e-01 1.07887852e+00 4.33426350e-03 -4.13235337e-01
2.22382322e-01 3.43789101e-01 5.61364591e-01 1.55521834e+00
-1.42016873e-01 2.54545689e-01 3.76118794e-02 9.47385013e-01
-4.09538537e-01 5.28177358e-02 -5.77989221e-01 -2.58577496e-01
4.87986892e-01 9.28298295e-01 -8.11107159e-01 -5.49722791e-01
-5.58562100e-01 9.94624317e-01 7.37823009e-01 2.80461997e-01
-1.01901805e+00 -4.69971418e-01 3.57353777e-01 -1.56250373e-01
2.01516032e-01 -1.33964360e-01 -2.80794054e-01 -1.12635398e+00
-1.19969040e-01 -4.63811070e-01 1.23237979e+00 -7.47182846e-01
-1.54306030e+00 3.77053589e-01 3.32579762e-01 -6.12749875e-01
-4.26587433e-01 -5.28197527e-01 -1.32443166e+00 5.12723744e-01
-1.41106641e+00 -1.25678611e+00 -2.86768079e-01 6.99368298e-01
6.15326464e-01 7.06651658e-02 6.69264317e-01 3.10713500e-01
-6.42364562e-01 5.38339853e-01 -2.55742133e-01 7.12652743e-01
4.30207819e-01 -1.42440140e+00 8.53204072e-01 1.16345751e+00
3.12302679e-01 5.13339162e-01 4.41126347e-01 -1.14157665e+00
-1.18061674e+00 -1.58480108e+00 1.57476389e+00 -3.79168749e-01
1.08063102e+00 -2.66876996e-01 -9.75240231e-01 1.06036556e+00
5.52051187e-01 -1.31409749e-01 5.23022890e-01 4.81468350e-01
-2.83094257e-01 2.20649242e-01 -4.28893656e-01 7.47090161e-01
1.37475908e+00 -4.53288555e-01 -1.01464558e+00 6.28448784e-01
1.33730793e+00 -3.49174351e-01 -8.16731751e-01 3.98101598e-01
-2.01200441e-01 -4.56727266e-01 1.12262189e+00 -1.16318130e+00
1.16699970e+00 -1.96738675e-01 2.86521047e-01 -1.27816880e+00
-4.98631477e-01 -5.30649185e-01 -5.95026314e-01 1.59889448e+00
6.97914898e-01 -2.42848262e-01 2.54690796e-01 3.30375105e-01
-4.12598342e-01 -4.37952280e-01 -5.62929571e-01 -5.14566243e-01
-1.50773348e-02 -5.91262102e-01 7.29971170e-01 1.18014014e+00
5.17458975e-01 1.05142915e+00 -1.34964241e-02 5.13955832e-01
3.91330630e-01 4.34693813e-01 2.41455570e-01 -1.23569882e+00
-6.87220767e-02 -4.14150566e-01 -1.93654418e-01 -5.74112415e-01
6.91534042e-01 -1.15983820e+00 -1.85603142e-01 -2.30980802e+00
3.71202379e-01 1.28221408e-01 -4.03020710e-01 4.78284925e-01
-9.01131451e-01 -4.59314197e-01 -7.29582235e-02 -1.98212579e-01
-7.21857548e-01 5.27153254e-01 1.09222806e+00 -4.05940384e-01
-1.95875764e-01 -4.69421297e-01 -4.71029431e-01 6.90083504e-01
7.37147152e-01 -2.66965121e-01 -7.86597013e-01 -6.41273320e-01
5.44714332e-01 2.97936201e-01 6.30272865e-01 -5.96210420e-01
3.05061132e-01 -2.91807920e-01 2.20191583e-01 -8.74241412e-01
-4.48772997e-01 -4.08506960e-01 -1.75379351e-01 6.39306381e-02
-9.90778625e-01 -7.91526064e-02 1.54292107e-01 8.97748470e-01
-4.70645159e-01 -9.35582444e-02 1.22341342e-01 -2.92318165e-02
-8.83136451e-01 3.13130230e-01 -2.94193238e-01 4.42488790e-01
7.05577195e-01 4.25954670e-01 -6.00861847e-01 -4.49739665e-01
-6.86427653e-01 3.25066954e-01 -5.81849396e-01 6.87439024e-01
7.01860845e-01 -1.33162463e+00 -9.85945404e-01 -3.78630877e-01
2.26171374e-01 2.34042034e-01 2.00029850e-01 5.02642334e-01
-6.35073856e-02 4.12683815e-01 3.53460550e-01 -1.75339893e-01
-1.21138287e+00 9.53158677e-01 9.90076289e-02 -9.44193661e-01
-8.42176437e-01 1.07548428e+00 6.06786549e-01 -2.74699748e-01
3.08093160e-01 -7.12385237e-01 -7.43087053e-01 5.74659929e-03
6.58774555e-01 7.44274929e-02 5.99597627e-03 -2.25170970e-01
-5.73839962e-01 -2.08135426e-01 -2.37439528e-01 2.25164890e-01
1.22373950e+00 -7.09558204e-02 -4.52296257e-01 3.10129076e-01
1.10573459e+00 -4.56154108e-01 -8.04893553e-01 -2.67164737e-01
3.70533347e-01 1.42172596e-03 3.55976704e-03 -9.31683540e-01
-8.80135834e-01 8.66709173e-01 -1.22227974e-01 4.33541745e-01
1.00857508e+00 4.03938562e-01 8.18486691e-01 3.36477041e-01
-2.36365631e-01 -8.93710196e-01 4.33642387e-01 9.36255276e-01
1.50782490e+00 -1.10540140e+00 -2.84699827e-01 -7.21252441e-01
-6.03810012e-01 1.28418648e+00 6.90064371e-01 -1.37404770e-01
6.12533629e-01 1.66066721e-01 -5.13870895e-01 -6.34321272e-01
-8.47823918e-01 -1.57566518e-01 5.19947886e-01 4.71498519e-01
9.52723026e-01 1.44098952e-01 -5.93669415e-01 9.95692313e-01
1.54006198e-01 -9.63315964e-02 4.89720851e-01 6.59347951e-01
1.06297731e-02 -7.72718430e-01 -8.50680768e-02 3.38512033e-01
-4.04264927e-01 -3.51447493e-01 -7.04809844e-01 7.23916948e-01
-3.95210832e-01 1.35796893e+00 2.49438107e-01 -1.34470925e-01
4.15998816e-01 1.10833853e-01 2.45293692e-01 -5.47798753e-01
-7.28921592e-01 -2.28312731e-01 6.34818077e-01 -5.79864919e-01
-4.91000175e-01 -5.12058079e-01 -1.92295468e+00 -8.69951621e-02
3.62681113e-02 -1.82378236e-02 -3.37674208e-02 1.05616176e+00
4.82098222e-01 1.38893759e+00 4.60331827e-01 -1.82024077e-01
8.93675089e-02 -8.95435274e-01 -2.27772638e-01 7.73265541e-01
2.10625574e-01 -6.03944778e-01 -1.64995342e-02 3.14313740e-01] | [9.078768730163574, 9.064620971679688] |
a5ed1d4c-5c73-4e0c-b8f4-fccc35b47c2d | neural-guided-program-synthesis-of | null | null | https://aclanthology.org/2022.pandl-1.10 | https://aclanthology.org/2022.pandl-1.10.pdf | Neural-Guided Program Synthesis of Information Extraction Rules Using Self-Supervision | We propose a neural-based approach for rule synthesis designed to help bridge the gap between the interpretability, precision and maintainability exhibited by rule-based information extraction systems with the scalability and convenience of statistical information extraction systems. This is achieved by avoiding placing the burden of learning another specialized language on domain experts and instead asking them to provide a small set of examples in the form of highlighted spans of text. We introduce a transformer-based architecture that drives a rule synthesis system that leverages a self-supervised approach for pre-training a large-scale language model complemented by an analysis of different loss functions and aggregation mechanisms for variable length sequences of user-annotated spans of text. The results are encouraging and point to different desirable properties, such as speed and quality, depending on the choice of loss and aggregation method. | ['Marco A. Valenzuela-Escárcega', 'Gus Hahn-Powell', 'Robert Vacareanu', 'Enrique Noriega-Atala'] | null | null | null | null | pandl-coling-2022-10 | ['program-synthesis'] | ['computer-code'] | [ 4.03536230e-01 4.07282412e-01 -2.53794253e-01 -6.24462962e-01
-8.62268090e-01 -6.96728170e-01 7.61994421e-01 5.45297444e-01
-4.57008690e-01 8.22954357e-01 -5.31132296e-02 -7.86194861e-01
-4.94210660e-01 -7.96907604e-01 -5.55572033e-01 -1.73263222e-01
-2.31775999e-01 5.15344322e-01 1.73186630e-01 -2.96850711e-01
2.28717197e-02 6.03673220e-01 -1.67699468e+00 5.58544576e-01
9.05436218e-01 1.21494257e+00 -1.10746861e-01 5.95331252e-01
-2.97423720e-01 9.09827471e-01 -6.02556825e-01 -5.94900072e-01
3.38173896e-01 -3.30439478e-01 -5.90113997e-01 1.10925429e-01
3.01017612e-01 5.62271625e-02 -1.01385482e-01 7.05588222e-01
2.10594416e-01 2.13972037e-03 5.55988252e-01 -1.09666288e+00
-3.86566013e-01 7.18450546e-01 -1.27022386e-01 1.77983060e-01
3.54706526e-01 2.14910582e-01 1.38297260e+00 -7.69216716e-01
6.82392597e-01 1.11500359e+00 7.30396926e-01 4.85305965e-01
-1.35782707e+00 -3.92482132e-01 2.13540569e-01 1.70338377e-02
-1.22662711e+00 -6.51789963e-01 5.21878600e-01 -3.98926824e-01
1.32522523e+00 2.45738730e-01 5.88630140e-01 9.65629518e-01
2.49095514e-01 6.95518911e-01 5.29902637e-01 -6.72320366e-01
4.54540700e-01 5.27871013e-01 -6.08258173e-02 9.38722372e-01
4.05095845e-01 7.57251680e-02 -6.51431441e-01 -2.19495296e-01
2.01329887e-01 -2.84640759e-01 -2.73810420e-03 -5.44629812e-01
-7.93852150e-01 8.25632274e-01 -1.30538434e-01 3.50062907e-01
-3.54835659e-01 -1.54209495e-01 7.77226090e-01 5.08244872e-01
4.45817769e-01 8.94416869e-01 -9.76625621e-01 -1.73776727e-02
-1.26450372e+00 3.16252917e-01 9.39157486e-01 9.87946093e-01
5.41341603e-01 1.24742940e-01 -2.53728986e-01 5.93441427e-01
1.70322224e-01 1.73296705e-01 3.70706081e-01 -7.08493173e-01
6.27036870e-01 9.61587548e-01 5.11904657e-02 -7.58034527e-01
-3.64317119e-01 -5.25567710e-01 -4.35157508e-01 2.99724072e-01
5.41138709e-01 -4.58032161e-01 -7.10099161e-01 1.68156123e+00
1.58741847e-01 -7.76630282e-01 -4.48534153e-02 3.78350496e-01
2.12435991e-01 5.47764897e-01 1.28682405e-01 -2.84468830e-01
1.34510434e+00 -5.49918532e-01 -4.25035357e-01 -3.04780841e-01
7.33494103e-01 -2.64891028e-01 1.40921450e+00 5.99525094e-01
-1.07459867e+00 -2.17683434e-01 -1.20770276e+00 -1.25308529e-01
-6.91933751e-01 2.36672848e-01 4.90819454e-01 5.79685867e-01
-7.70474017e-01 7.51276970e-01 -6.69190407e-01 -5.32101952e-02
5.49387157e-01 3.82114440e-01 -3.58523279e-01 2.26785064e-01
-1.17452836e+00 1.01776588e+00 6.94489717e-01 -9.02914852e-02
-4.94826138e-01 -1.02229083e+00 -8.54286909e-01 3.93093735e-01
5.41504920e-01 -5.74993968e-01 1.20168436e+00 -1.09657562e+00
-1.23226643e+00 6.33089781e-01 -3.72731462e-02 -7.83158302e-01
5.67195296e-01 -2.99624145e-01 -4.72241372e-01 -8.89656972e-03
-1.29655778e-01 3.02363485e-01 6.21546149e-01 -8.83215308e-01
-8.13950360e-01 -1.95094869e-01 1.29580861e-02 -1.49082392e-01
-4.20504838e-01 1.05320983e-01 5.97833022e-02 -5.24945617e-01
-5.00006378e-01 -3.84031057e-01 -2.79492021e-01 4.46602777e-02
-5.27887702e-01 -1.88874155e-01 7.48767853e-01 -6.98535740e-01
1.43317676e+00 -1.92319798e+00 -2.66758174e-01 5.29653311e-01
-6.95254188e-03 4.29848850e-01 -1.38020992e-01 6.02773666e-01
-9.35618281e-02 3.14644367e-01 -3.36425990e-01 -3.20558175e-02
1.58332020e-01 -9.92429331e-02 -5.02241373e-01 8.26724619e-03
7.69927323e-01 8.31063271e-01 -8.96154881e-01 -5.26074886e-01
1.76776312e-02 2.10037440e-01 -4.89573658e-01 2.51550674e-01
-8.58269989e-01 -2.87158713e-02 -3.51459295e-01 3.95445526e-01
1.77305434e-02 -3.57631892e-01 4.31245863e-01 4.58338810e-03
-6.35548010e-02 9.16266859e-01 -9.91234541e-01 1.55786312e+00
-6.64762557e-01 5.80764413e-01 -1.56430513e-01 -8.68991792e-01
1.08070350e+00 3.03354084e-01 2.75498956e-01 -4.76132900e-01
2.72651780e-02 1.15417711e-01 2.72461213e-02 -6.08047187e-01
3.61997634e-01 -3.28835487e-01 -9.69459563e-02 6.66148484e-01
1.07757121e-01 -1.42896637e-01 7.37268209e-01 2.02907160e-01
1.41088963e+00 2.45673195e-01 5.36977530e-01 -2.94034839e-01
5.64538181e-01 2.31882229e-01 4.47316647e-01 5.83927751e-01
3.32244068e-01 1.30915910e-01 6.81871831e-01 -7.02535272e-01
-1.34215212e+00 -6.95297718e-01 -7.91463479e-02 1.13923168e+00
-7.86149919e-01 -6.41932964e-01 -5.38070381e-01 -9.44403708e-01
1.43291935e-01 1.07456946e+00 -4.33982849e-01 -2.71474034e-01
-4.84849334e-01 -5.26178479e-01 7.37154543e-01 5.24391294e-01
3.81284505e-02 -1.03839135e+00 -8.63379896e-01 3.13344806e-01
1.26102537e-01 -1.12142694e+00 -3.63895535e-01 7.19488800e-01
-8.47445428e-01 -1.04643393e+00 1.50038287e-01 -4.59622145e-01
5.84358156e-01 -4.84238267e-01 1.34802985e+00 4.62905802e-02
-2.40642175e-01 -6.66027293e-02 -1.26788870e-01 -5.59777498e-01
-6.06246412e-01 4.12415713e-01 -1.11140482e-01 -2.31442183e-01
4.32899147e-01 -7.56805420e-01 -1.40284553e-01 -1.15836054e-01
-1.00315630e+00 -1.05835192e-01 6.60019040e-01 7.91441381e-01
2.53206074e-01 9.60293859e-02 5.49868822e-01 -1.06241691e+00
9.84925091e-01 -3.25050771e-01 -8.05382311e-01 6.00299299e-01
-1.08079028e+00 6.28826320e-01 1.11259699e+00 -4.06971931e-01
-9.46451306e-01 2.74935693e-01 5.62846288e-02 -7.45144859e-02
-9.56551954e-02 7.06580043e-01 -3.39310616e-01 4.19114232e-01
9.36747611e-01 2.30908588e-01 1.23579092e-01 -3.14261973e-01
5.84153235e-01 6.25915706e-01 4.25484300e-01 -6.44834757e-01
7.85863400e-01 9.05936211e-02 -1.43724978e-02 -4.98457551e-01
-8.20954680e-01 6.93030655e-02 -5.13216615e-01 1.97818920e-01
2.89647520e-01 -6.18520796e-01 -6.29410625e-01 -1.22576356e-01
-9.41986620e-01 -4.74742025e-01 -7.54162550e-01 -3.77768353e-02
-5.75615525e-01 1.83180109e-01 -4.23066020e-01 -9.01141047e-01
-5.72233438e-01 -5.81941485e-01 7.95982480e-01 -1.24331139e-01
-6.78462029e-01 -9.71537411e-01 -2.86368355e-02 -6.97189718e-02
2.88704723e-01 8.39222744e-02 1.41299701e+00 -1.17909408e+00
-5.11654377e-01 -4.08599436e-01 -1.62050650e-01 4.19570655e-01
1.65836975e-01 1.16611771e-01 -7.68554032e-01 4.36988054e-03
-4.07343298e-01 -5.22127330e-01 4.50484306e-01 3.01659275e-02
1.11744058e+00 -8.12913179e-01 -1.33940801e-01 3.65159690e-01
1.30424166e+00 2.02019438e-01 3.58080357e-01 4.20529097e-01
1.49777353e-01 7.11749017e-01 3.88029933e-01 4.61789131e-01
1.08443245e-01 5.14885724e-01 -4.33345288e-02 1.17784254e-01
9.38338041e-02 -6.67741299e-01 1.79725558e-01 3.78721766e-02
3.58610719e-01 -2.44171172e-01 -9.89483595e-01 5.78128874e-01
-1.60976589e+00 -9.59200978e-01 5.56152642e-01 2.09062743e+00
1.34101808e+00 6.21150315e-01 4.18460995e-01 5.10939538e-01
3.19872290e-01 -7.73330126e-03 -5.77520013e-01 -7.56956279e-01
-8.17645574e-04 3.24245393e-01 4.00225729e-01 4.44568992e-01
-9.43217814e-01 7.90650368e-01 6.92440033e+00 8.52671266e-01
-1.11217344e+00 -3.83082956e-01 5.30061245e-01 -3.06208640e-01
-4.80745941e-01 -4.95633967e-02 -8.35727155e-01 4.29901838e-01
1.41308522e+00 -3.55885237e-01 5.07332146e-01 7.51352012e-01
3.93030375e-01 9.71511230e-02 -1.46212590e+00 3.79177272e-01
-1.99566916e-01 -1.46511114e+00 2.24442348e-01 -1.22972019e-01
2.11337507e-01 -2.55223781e-01 -1.90370038e-01 1.28356472e-01
5.40717781e-01 -1.15316665e+00 8.84745896e-01 5.48259437e-01
8.27136636e-01 -6.34777963e-01 3.53064954e-01 5.33633530e-01
-7.94370234e-01 -4.78168756e-01 1.89520270e-02 -1.27644107e-01
-7.14664608e-02 8.67791176e-01 -1.13433206e+00 3.33853453e-01
3.26526970e-01 3.55105877e-01 -4.53486919e-01 6.13732934e-01
-3.04731220e-01 5.45221746e-01 -4.51686174e-01 -2.07189918e-01
9.74966064e-02 7.10451156e-02 1.98216006e-01 1.57996333e+00
1.49304733e-01 -1.58298567e-01 -3.85090075e-02 1.04488158e+00
3.66146713e-02 1.97775945e-01 -8.89423788e-01 -2.70998091e-01
8.11651289e-01 1.11955309e+00 -3.34470809e-01 -4.16915804e-01
-4.06492651e-01 2.43610576e-01 4.72345352e-01 1.96395680e-01
-3.40345323e-01 -7.63745725e-01 2.98262328e-01 3.34109604e-01
7.70389020e-01 -1.89162567e-01 -6.60058498e-01 -9.96304274e-01
3.66639435e-01 -1.27530801e+00 4.94433075e-01 -5.77451050e-01
-1.01286638e+00 6.55966520e-01 3.43943313e-02 -9.66688454e-01
-7.37003148e-01 -5.23306549e-01 -4.15783137e-01 8.00397038e-01
-1.18376207e+00 -9.21360016e-01 4.35116202e-01 1.83931381e-01
4.27290082e-01 -4.43779588e-01 7.22992361e-01 7.48321861e-02
-4.05177206e-01 8.15726280e-01 -1.43855870e-01 2.97518253e-01
1.45460978e-01 -1.24691212e+00 4.82973158e-01 9.49567258e-01
3.15128416e-01 7.86267042e-01 8.80425215e-01 -3.30306321e-01
-1.24876034e+00 -1.10859144e+00 1.37628770e+00 -5.98927140e-01
7.23924816e-01 -4.22130942e-01 -8.40734363e-01 6.08237028e-01
1.85711868e-02 -1.30372480e-01 5.85304022e-01 3.59404713e-01
-6.28150642e-01 -4.36347634e-01 -1.22382832e+00 4.25496101e-01
9.51814592e-01 -6.19554877e-01 -6.88145995e-01 5.48465364e-02
5.07312119e-01 -3.14257517e-02 -8.19501042e-01 2.92962164e-01
5.60574770e-01 -6.45914495e-01 7.55325317e-01 -1.13597870e+00
6.77013814e-01 -2.04090327e-01 -2.40793708e-03 -1.15567219e+00
-1.75518990e-01 -6.44524038e-01 -3.93365026e-01 1.15963376e+00
1.09141898e+00 -4.35137004e-01 8.43835533e-01 9.28058624e-01
1.41303182e-01 -1.06079555e+00 -6.78616941e-01 -7.54396677e-01
-2.10654028e-02 -4.43116695e-01 7.00966179e-01 5.16717136e-01
4.68307644e-01 6.95264876e-01 -1.29765004e-01 -2.16430619e-01
2.56796807e-01 2.99466718e-02 4.95298505e-01 -1.27165246e+00
-1.92371026e-01 -4.09547031e-01 -8.61752555e-02 -6.97825789e-01
1.03210516e-01 -8.88676107e-01 1.90686107e-01 -1.21151376e+00
-1.12508409e-01 -4.29846585e-01 -2.38991246e-01 9.51399803e-01
1.91855565e-01 -3.42745245e-01 1.46547304e-02 -1.75000384e-01
-6.26277924e-01 2.73333907e-01 7.44642437e-01 -1.02515168e-01
-3.83750200e-01 -7.26261884e-02 -1.05450606e+00 6.14418805e-01
6.65120244e-01 -4.61239308e-01 -6.90371573e-01 -2.55366325e-01
7.20483601e-01 5.87005429e-02 1.56580120e-01 -9.18064594e-01
1.16654299e-01 -3.39661807e-01 3.06180686e-01 -2.99697578e-01
-1.27628096e-03 -8.33668709e-01 -2.57657226e-02 2.85113007e-01
-1.00591230e+00 1.88447163e-01 2.69691616e-01 3.93283367e-01
-4.63063158e-02 -1.87190473e-01 6.66158319e-01 -1.06707312e-01
-2.55205959e-01 1.32675111e-01 -5.57913184e-01 2.36971244e-01
6.86350942e-01 -5.02622165e-02 -7.08062127e-02 -2.77963161e-01
-5.23017585e-01 1.33264512e-01 2.75140673e-01 2.78874993e-01
4.30643380e-01 -8.76195073e-01 -6.02588415e-01 4.42024171e-01
1.12901300e-01 -1.72663499e-02 -5.09422958e-01 3.77390862e-01
-1.68411478e-01 6.15489662e-01 -8.20411965e-02 -1.22803472e-01
-1.01984358e+00 6.58949196e-01 3.69001389e-01 -7.81466782e-01
-6.39139533e-01 4.78505522e-01 -4.50265020e-01 -2.75968403e-01
3.88872325e-01 -4.46337134e-01 7.03357607e-02 -6.47474751e-02
6.01978540e-01 4.48192423e-03 5.26675761e-01 1.80190474e-01
-4.47083116e-01 -2.17949599e-01 -2.05338791e-01 -8.51241499e-02
1.36500204e+00 2.20442936e-01 1.55259579e-01 3.49714190e-01
7.24081099e-01 -1.37999160e-02 -1.23131442e+00 -3.93284947e-01
6.57821178e-01 -9.23238993e-02 -1.17600143e-01 -1.17630172e+00
-8.27276289e-01 7.18585551e-01 1.39300078e-01 3.01597267e-01
1.04817474e+00 -9.87943187e-02 7.41853058e-01 8.62537980e-01
1.32411405e-01 -1.20878232e+00 -1.07717372e-01 6.02804482e-01
7.87340522e-01 -7.50208080e-01 6.89670518e-02 -1.35266289e-01
-5.07607698e-01 1.17429662e+00 2.98489571e-01 -2.54743882e-02
5.18347263e-01 7.46473372e-01 4.90799583e-02 -1.58982202e-01
-1.42707705e+00 -6.46319017e-02 3.30044329e-01 6.78676724e-01
6.02709115e-01 -2.67828614e-01 -3.12971622e-01 7.32823312e-01
-4.09514636e-01 2.77286828e-01 1.54535070e-01 9.65737283e-01
-4.97192383e-01 -1.38083434e+00 5.65488636e-03 7.43119299e-01
-4.97694343e-01 -2.56333709e-01 -5.29251933e-01 8.37631881e-01
9.04902443e-02 8.26886594e-01 -1.96462438e-01 -3.59071434e-01
3.04903001e-01 6.33528411e-01 3.45498621e-01 -7.49860823e-01
-5.94910324e-01 -3.40089023e-01 7.21158624e-01 -2.81647235e-01
-2.35785376e-02 -6.67540908e-01 -9.71033692e-01 -5.23517951e-02
-1.12017930e-01 4.00223255e-01 4.92402166e-01 1.24853969e+00
5.35024285e-01 2.43281618e-01 5.68821013e-01 -1.66143924e-01
-9.41816568e-01 -8.55957806e-01 -3.69876057e-01 3.50393325e-01
4.61401731e-01 -3.51538718e-01 -1.20535322e-01 1.03801131e-01] | [9.752248764038086, 7.863102912902832] |
0730a86a-c878-4768-8ddf-84bafb82c0fb | improving-the-morphological-analysis-of | null | null | https://aclanthology.org/W16-3715 | https://aclanthology.org/W16-3715.pdf | Improving the Morphological Analysis of Classical Sanskrit | The paper describes a new tagset for the morphological disambiguation of Sanskrit, and compares the accuracy of two machine learning methods (Conditional Random Fields, deep recurrent neural networks) for this task, with a special focus on how to model the lexicographic information. It reports a significant improvement over previously published results. | ['Oliver Hellwig'] | 2016-12-01 | null | null | null | ws-2016-12 | ['morphological-disambiguation'] | ['natural-language-processing'] | [ 8.85512009e-02 -4.80349511e-02 -3.50390106e-01 -1.84833854e-01
-5.72399139e-01 -6.87251985e-01 8.05681467e-01 4.38603848e-01
-9.33431208e-01 8.93074274e-01 9.23903942e-01 -7.28394687e-01
-4.14941549e-01 -6.01006150e-01 9.31622460e-02 -6.25276625e-01
-2.42338359e-01 1.13883162e+00 -5.52978925e-02 -5.70276320e-01
7.60632932e-01 3.92190874e-01 -9.76307750e-01 5.09571731e-01
8.64319921e-01 7.53503799e-01 4.57411677e-01 4.77802157e-01
-4.96403694e-01 1.13573182e+00 -7.83063293e-01 -6.47363544e-01
-1.69616938e-01 7.69959167e-02 -1.60717070e+00 -3.24380487e-01
4.97542858e-01 2.86768347e-01 -3.90944183e-01 1.20638859e+00
6.08486950e-01 3.74866635e-01 8.13518226e-01 -8.16549659e-02
-1.27430212e+00 1.39631438e+00 1.55126542e-01 8.42709124e-01
4.53519881e-01 -6.91049993e-01 1.50291097e+00 -2.55178988e-01
1.15320408e+00 9.03419256e-01 1.04603219e+00 4.57210004e-01
-7.80040085e-01 -2.55149275e-01 -1.09598696e-01 2.58666843e-01
-1.39669991e+00 -1.89461797e-01 3.93403977e-01 -5.00855446e-01
1.64460075e+00 5.14473915e-02 3.60004783e-01 1.11072600e+00
5.78084111e-01 8.59260023e-01 1.45641971e+00 -6.81772649e-01
-2.72452850e-02 -4.68291081e-02 7.16256738e-01 6.20773137e-01
-5.07605635e-03 2.01878622e-01 -3.42173904e-01 -2.40757570e-01
5.76450288e-01 -2.13960603e-01 -4.75182496e-02 3.14699322e-01
-1.21556211e+00 9.66265559e-01 3.33275735e-01 1.27338684e+00
-5.54661691e-01 -3.92688096e-01 4.83136147e-01 3.35675389e-01
5.59321284e-01 1.05002427e+00 -1.00502908e+00 8.15787390e-02
-1.01372886e+00 -2.29238905e-02 1.06830561e+00 7.32480943e-01
2.58755803e-01 1.56736135e-01 -5.64486943e-02 8.48251998e-01
2.21595764e-01 4.61074829e-01 1.04846430e+00 -2.62711465e-01
4.07016754e-01 2.08507657e-01 -1.96999446e-01 -1.11877859e+00
-7.59573102e-01 -2.83468157e-01 -5.78096688e-01 -7.39524424e-01
4.37191986e-02 -1.90095782e-01 -1.18047488e+00 8.74653399e-01
-2.75398433e-01 -1.33977771e-01 3.71777445e-01 5.97858191e-01
1.23782182e+00 8.00869524e-01 5.95689356e-01 -3.52246404e-01
1.43872416e+00 -5.97512007e-01 -1.14000499e+00 -1.60448685e-01
6.09607160e-01 -1.06963623e+00 3.32933336e-01 3.83672744e-01
-7.03082681e-01 -3.19268167e-01 -6.91641331e-01 -3.62493023e-02
-9.09825623e-01 -1.75555740e-02 1.09252203e+00 8.07206273e-01
-1.36965525e+00 8.90955150e-01 -7.72383928e-01 -7.70349026e-01
-2.20681205e-01 6.93379462e-01 -4.09634054e-01 4.93145764e-01
-1.75984848e+00 1.20144522e+00 9.79869246e-01 3.67698446e-02
-3.02395701e-01 -2.80754138e-02 -9.46715891e-01 -2.77551323e-01
2.34282032e-01 -1.43092901e-01 1.20331573e+00 -8.00181568e-01
-1.54010010e+00 1.20954287e+00 2.07471982e-01 -7.49480665e-01
-3.71975034e-01 -2.15847358e-01 -1.05118179e+00 -3.32355976e-01
1.80489451e-01 2.78630525e-01 3.50283623e-01 -7.82255352e-01
-8.21042836e-01 -2.91015029e-01 -3.49271983e-01 1.86704278e-01
-1.52109876e-01 8.12774420e-01 3.37883495e-02 -1.23760033e+00
3.48786622e-01 -7.94146538e-01 -6.16511643e-01 -1.67113554e+00
-4.11674649e-01 -6.98305964e-01 7.72310570e-02 -1.25836957e+00
1.39644098e+00 -2.09507728e+00 2.29004204e-01 8.34930837e-02
2.03012582e-02 3.24653089e-01 -5.28655620e-03 5.79215169e-01
-1.59344733e-01 5.10639846e-01 4.44286726e-02 3.17122452e-02
-8.27011093e-02 5.30592322e-01 -4.23995733e-01 1.94576383e-01
-1.03758268e-01 9.79410589e-01 -1.00500810e+00 -6.23802364e-01
4.84536216e-02 4.64768797e-01 -2.94294767e-02 -1.60158128e-01
-9.37271416e-02 1.16895340e-01 -3.21081489e-01 7.22788870e-01
1.94805235e-01 -1.14096582e-01 1.05479622e+00 -1.19638249e-01
-3.30409139e-01 1.22501433e+00 -9.15161490e-01 1.47066426e+00
-3.57459605e-01 4.89534974e-01 -3.15286726e-01 -7.06779301e-01
8.67039502e-01 5.33107519e-01 3.65464032e-01 -7.42967010e-01
4.48511899e-01 4.30584252e-01 -2.37529144e-01 -3.52972656e-01
1.29702330e+00 -1.91534340e-01 -5.73983669e-01 4.60434943e-01
6.14874005e-01 4.12928164e-01 -3.52401882e-02 7.27174655e-02
8.84682059e-01 3.16997766e-02 7.67472386e-01 -5.55716753e-01
1.57548606e-01 9.21304971e-02 6.60381198e-01 6.61273062e-01
4.00706641e-02 1.24768242e-01 1.45088673e-01 -9.93276715e-01
-7.53906369e-01 -6.49929702e-01 -3.15407366e-01 1.62492192e+00
-6.00682246e-03 -6.06538415e-01 -5.73071420e-01 -1.18466389e+00
-2.40758032e-01 7.85990775e-01 -7.78600693e-01 4.30394083e-01
-1.07025075e+00 -1.32135773e+00 7.15529442e-01 5.84499359e-01
2.15118676e-01 -1.57889175e+00 -1.31515890e-01 4.42498893e-01
-2.84697950e-01 -9.76533294e-01 -1.41861275e-01 1.11794007e+00
-8.80137980e-01 -8.56557965e-01 -2.93822646e-01 -1.51113439e+00
2.88279414e-01 -1.74858123e-01 1.41902673e+00 1.00401528e-01
2.56105959e-01 -1.82822809e-01 -7.69325137e-01 1.73475564e-01
-5.04521310e-01 6.48382962e-01 7.41492286e-02 -4.58390623e-01
1.00118661e+00 -3.31439972e-01 1.52631372e-01 -1.41229657e-02
-9.87536490e-01 -5.68976164e-01 6.01300657e-01 1.02147424e+00
4.11426693e-01 2.65366793e-01 -4.81477194e-02 -1.35753953e+00
8.24458480e-01 -5.22182226e-01 -3.56472075e-01 2.02769041e-01
-9.63998616e-01 5.97744346e-01 2.86958247e-01 -3.54530632e-01
-1.04772449e+00 1.38118088e-01 -4.79475796e-01 6.09548509e-01
-3.03917885e-01 9.15830433e-01 1.38465062e-01 -7.32614473e-02
7.18443871e-01 1.08178362e-01 -9.36630189e-01 -1.13700557e+00
3.80277485e-01 9.87825155e-01 6.07438803e-01 -3.43405902e-01
5.23374937e-02 -5.71285933e-02 -5.41690886e-01 -5.17700076e-01
-7.92190552e-01 -6.49448156e-01 -1.25634396e+00 1.55833364e-01
1.02881241e+00 -4.33577538e-01 -1.46016940e-01 2.48847723e-01
-1.05897939e+00 -6.33689538e-02 -1.17688604e-01 6.99053288e-01
-3.51800829e-01 3.63873154e-01 -1.34539127e+00 -4.96724278e-01
-4.82187539e-01 -7.44512141e-01 7.82033622e-01 7.09203407e-02
-3.43636394e-01 -1.57082689e+00 5.41713059e-01 1.46588176e-01
2.80699402e-01 -1.76760212e-01 1.10367429e+00 -1.67354524e+00
-7.41111189e-02 6.27040640e-02 2.05510575e-02 -1.16380759e-01
7.83802345e-02 -2.71168649e-01 -6.96295023e-01 1.76238015e-01
-5.65518364e-02 -1.20768279e-01 1.16905797e+00 4.29126322e-01
2.57608294e-01 -4.26580518e-01 -4.81889248e-01 4.29753780e-01
1.45045733e+00 6.37601435e-01 5.50368428e-01 7.68867135e-01
6.85195804e-01 1.88739419e-01 3.11490446e-01 -9.46690049e-03
3.41981411e-01 5.91697633e-01 -1.62145212e-01 2.07691610e-01
-1.35613844e-01 -1.92409992e-01 1.95472926e-01 1.53031182e+00
-2.85021782e-01 -3.07664961e-01 -1.27006602e+00 6.97049439e-01
-1.76059401e+00 -7.36745417e-01 -2.03964673e-02 1.67045152e+00
9.50922906e-01 -7.37806857e-02 -9.11224931e-02 1.05509862e-01
1.00114548e+00 5.62052011e-01 3.68498713e-01 -7.48148382e-01
-4.53538626e-01 6.09594464e-01 8.83220971e-01 6.82820201e-01
-1.62984931e+00 1.79164410e+00 8.39988995e+00 9.66221452e-01
-9.53432202e-01 3.05211693e-01 2.39229932e-01 5.73079467e-01
-1.06959619e-01 3.87418866e-02 -1.11548555e+00 2.43251726e-01
1.39893389e+00 6.30888760e-01 2.88197845e-01 6.10221267e-01
-5.01667082e-01 1.18159071e-01 -5.98040104e-01 5.43157637e-01
5.12927592e-01 -1.37032127e+00 2.69490629e-01 1.28085539e-01
7.25659728e-01 4.21962053e-01 -4.58626375e-02 4.17391896e-01
9.53368962e-01 -1.00105751e+00 5.13226151e-01 4.23804700e-01
5.27161896e-01 -8.52432370e-01 1.41401517e+00 -1.21294126e-01
-8.96828532e-01 4.43539023e-02 -2.13245362e-01 -4.34944220e-02
7.10684434e-02 3.82559776e-01 -1.30851209e+00 5.64682961e-01
6.64609671e-01 9.76434529e-01 -6.96857154e-01 8.35993469e-01
-5.59318244e-01 7.06775427e-01 7.78887123e-02 -6.40792012e-01
5.77002347e-01 2.06856593e-01 7.15413034e-01 1.80026996e+00
-7.08524277e-03 -5.03800362e-02 5.39352298e-01 9.08271074e-02
5.73202372e-02 7.66980171e-01 -2.48445511e-01 -2.90648550e-01
2.74402827e-01 9.94120896e-01 -1.24759042e+00 -3.10750693e-01
1.44863218e-01 9.40761626e-01 4.93487328e-01 -3.43056954e-02
-1.51326612e-01 -4.78944093e-01 -2.48317905e-02 -2.07599595e-01
5.36814988e-01 -3.63197923e-01 -3.94708306e-01 -1.26361740e+00
-4.83801365e-01 -9.16443825e-01 1.08539963e+00 -3.95186156e-01
-1.48770356e+00 1.33199966e+00 -3.30605119e-01 -7.72211492e-01
-4.23419535e-01 -9.21342492e-01 2.44206816e-01 6.71849847e-01
-1.23999560e+00 -1.03567290e+00 8.43625963e-01 3.08874369e-01
2.87635386e-01 -6.01687372e-01 1.36039615e+00 3.86277020e-01
-1.69541225e-01 3.86804521e-01 3.79186332e-01 8.59046817e-01
6.65013373e-01 -1.74150968e+00 9.05631542e-01 7.55766630e-01
8.71296823e-01 6.93745792e-01 4.32764530e-01 -1.08320618e+00
-8.66503298e-01 -6.56406760e-01 2.03197169e+00 -7.69345462e-01
9.44033802e-01 -1.01529799e-01 -5.47635376e-01 9.25465524e-01
6.88176692e-01 -5.85379601e-01 9.38782215e-01 8.36488485e-01
-4.37032610e-01 4.51988399e-01 -9.04798925e-01 2.00297579e-01
9.00241077e-01 -6.10246181e-01 -1.35854828e+00 5.00344217e-01
5.62812865e-01 -2.93494463e-01 -1.01675248e+00 4.04414922e-01
1.99837580e-01 -3.52597475e-01 6.06887519e-01 -1.04439819e+00
3.40987332e-02 -1.80644527e-01 -2.71561384e-01 -1.45517933e+00
-9.29430604e-01 -7.98059464e-01 1.72758386e-01 1.03757012e+00
7.87062764e-01 -2.75264591e-01 5.63975036e-01 -1.49420202e-01
2.61130948e-02 -2.65751481e-01 -1.01635253e+00 -5.18343151e-01
5.89260533e-02 -4.01321113e-01 5.61415255e-01 1.54932058e+00
5.53717077e-01 7.62063980e-01 -5.16661406e-01 -5.19096106e-02
-2.14583337e-01 2.77539283e-01 -4.63963449e-01 -1.38492918e+00
-1.58974066e-01 -5.27237177e-01 -7.71266460e-01 -8.89041305e-01
4.73929703e-01 -1.26437926e+00 5.96791431e-02 -1.52635586e+00
1.45178795e-01 -5.19730806e-01 -7.87714422e-01 8.31664324e-01
-4.61793430e-02 2.62015998e-01 -7.92061165e-03 2.50280112e-01
-1.06368148e+00 -3.98824140e-02 6.16218925e-01 -2.24708945e-01
-3.54421586e-01 -1.65943146e-01 -8.03580165e-01 4.95798022e-01
7.51759291e-01 -8.17587137e-01 6.09325290e-01 -5.77259839e-01
9.62296486e-01 -9.15689487e-03 -4.01149690e-01 -7.33034134e-01
2.78624088e-01 2.28606220e-02 3.61032456e-01 -6.19354188e-01
-4.13853705e-01 -5.45954764e-01 2.08560172e-02 3.41622740e-01
-6.61018014e-01 6.24645174e-01 -3.99950668e-02 3.01256150e-01
-4.40856606e-01 -6.10118568e-01 3.70502442e-01 -5.27724385e-01
-1.14783096e+00 3.78426863e-03 -9.76792216e-01 -3.19953375e-02
3.72652143e-01 3.56723040e-01 5.76107688e-02 1.11969613e-01
-1.28982878e+00 -1.42976373e-01 -5.59326319e-04 5.33483207e-01
2.82046407e-01 -1.39272296e+00 -7.05175221e-01 1.84596911e-01
-1.92276254e-01 -7.64584899e-01 -2.72559077e-01 3.84869665e-01
-7.25347877e-01 8.48202407e-01 -3.62208545e-01 1.34292739e-02
-1.24645197e+00 8.51289511e-01 3.53546828e-01 -8.30401361e-01
-5.69037318e-01 9.08461571e-01 -8.45994473e-01 -7.64203727e-01
1.87302336e-01 -1.22971639e-01 -1.31411791e+00 3.41811866e-01
4.43332285e-01 1.39071032e-01 5.25304973e-01 -1.06729400e+00
-4.63243753e-01 3.29954237e-01 -6.18150294e-01 -1.25211328e-01
1.34353614e+00 8.87903571e-03 -5.49704373e-01 5.22411823e-01
7.74217129e-01 4.35645700e-01 -8.52956846e-02 -5.78957260e-01
8.89709830e-01 -1.38197346e-02 3.53523225e-01 -1.27103162e+00
-9.84171450e-01 4.12495106e-01 5.18387079e-01 6.66625738e-01
9.31482852e-01 1.42177597e-01 8.94503653e-01 6.40031874e-01
5.18909454e-01 -1.45848048e+00 -7.04325616e-01 1.38681090e+00
1.57894611e-01 -8.08431208e-01 -2.87648682e-02 -2.81091362e-01
-6.98739290e-01 1.39823091e+00 -1.24119863e-01 -3.78204197e-01
9.92637634e-01 3.23612988e-01 4.14856553e-01 -4.60138172e-01
-4.21152979e-01 -3.92012417e-01 5.15366197e-01 6.26149237e-01
8.78946304e-01 3.30312580e-01 -8.20516825e-01 7.72959948e-01
-7.08513498e-01 -4.20455992e-01 3.49896938e-01 1.04130483e+00
-4.09483880e-01 -1.38858616e+00 -1.58229902e-01 4.56548542e-01
-1.27015805e+00 -9.00152624e-01 -9.05823171e-01 4.69458073e-01
3.68306488e-01 7.22934783e-01 5.93443811e-02 -7.16656029e-01
-1.69260845e-01 3.01534325e-01 1.30152225e-01 -8.69137585e-01
-1.31615460e+00 2.31835693e-01 5.99825680e-01 -2.31172919e-01
-8.32057476e-01 -9.86433268e-01 -9.82487500e-01 -2.26136371e-02
-6.30019069e-01 7.74318576e-01 5.96062660e-01 1.11503029e+00
1.66145246e-02 2.07217976e-01 3.98159444e-01 -4.63076711e-01
-7.63421059e-02 -1.30272973e+00 -8.44976366e-01 1.33362815e-01
-1.36969015e-02 -3.02340597e-01 -1.99752748e-01 7.54909515e-02] | [10.177136421203613, 9.903135299682617] |
503ea157-3e5f-4743-9fcb-19a13ae43596 | recurrent-chunking-mechanisms-for-long-text | 2005.08056 | null | https://arxiv.org/abs/2005.08056v2 | https://arxiv.org/pdf/2005.08056v2.pdf | Recurrent Chunking Mechanisms for Long-Text Machine Reading Comprehension | In this paper, we study machine reading comprehension (MRC) on long texts, where a model takes as inputs a lengthy document and a question and then extracts a text span from the document as an answer. State-of-the-art models tend to use a pretrained transformer model (e.g., BERT) to encode the joint contextual information of document and question. However, these transformer-based models can only take a fixed-length (e.g., 512) text as its input. To deal with even longer text inputs, previous approaches usually chunk them into equally-spaced segments and predict answers based on each segment independently without considering the information from other segments. As a result, they may form segments that fail to cover the correct answer span or retain insufficient contexts around it, which significantly degrades the performance. Moreover, they are less capable of answering questions that need cross-segment information. We propose to let a model learn to chunk in a more flexible way via reinforcement learning: a model can decide the next segment that it wants to process in either direction. We also employ recurrent mechanisms to enable information to flow across segments. Experiments on three MRC datasets -- CoQA, QuAC, and TriviaQA -- demonstrate the effectiveness of our proposed recurrent chunking mechanisms: we can obtain segments that are more likely to contain complete answers and at the same time provide sufficient contexts around the ground truth answers for better predictions. | ['Hongyu Gong', 'Yelong Shen', 'Jianshu Chen', 'Dian Yu', 'Dong Yu'] | 2020-05-16 | recurrent-chunking-mechanisms-for-long-text-1 | https://aclanthology.org/2020.acl-main.603 | https://aclanthology.org/2020.acl-main.603.pdf | acl-2020-6 | ['triviaqa'] | ['miscellaneous'] | [ 4.64146256e-01 3.53688836e-01 -2.49402702e-01 -3.86685312e-01
-1.07486546e+00 -7.15126932e-01 1.55167669e-01 5.57476223e-01
-5.64837158e-01 7.54080951e-01 3.78840655e-01 -6.49611115e-01
-8.60032998e-03 -1.13710880e+00 -9.24925089e-01 -2.53196001e-01
5.28016925e-01 5.52677691e-01 6.15581810e-01 -2.80453622e-01
4.92775947e-01 -2.85506874e-01 -1.50033367e+00 7.88649082e-01
1.50265634e+00 7.70177722e-01 6.65551662e-01 8.29986155e-01
-6.56073570e-01 1.05283594e+00 -7.18831897e-01 -4.99136388e-01
-1.17380790e-01 -8.07176054e-01 -1.43254161e+00 -1.39689356e-01
2.88885266e-01 -4.67657566e-01 -2.55834132e-01 6.78094387e-01
1.97822198e-01 1.94534615e-01 4.58501339e-01 -6.34576023e-01
-6.23036683e-01 9.90012228e-01 -2.58919567e-01 2.23746866e-01
8.63323867e-01 5.17528094e-02 1.31530404e+00 -6.21582270e-01
4.68459249e-01 1.10326922e+00 3.49043161e-01 6.42312407e-01
-9.65481043e-01 -2.91475147e-01 3.53697479e-01 5.35569549e-01
-7.67260671e-01 -2.52891630e-01 6.41455710e-01 3.08895279e-02
9.93200541e-01 5.07707059e-01 4.34232414e-01 8.77344608e-01
2.56563395e-01 1.23121941e+00 9.84869599e-01 -5.78413606e-01
2.13023841e-01 -2.07761884e-01 4.13418025e-01 5.50612926e-01
-3.07548404e-01 -2.79599637e-01 -4.75316525e-01 9.43431109e-02
2.41813719e-01 -5.60882613e-02 -5.86562157e-01 9.83912125e-02
-1.18206775e+00 9.70021248e-01 5.63007236e-01 3.42076004e-01
-4.46249992e-01 -3.84177238e-01 2.83521146e-01 6.48308098e-01
1.74420312e-01 4.87815261e-01 -5.10577559e-01 -1.99389398e-01
-1.00747108e+00 3.75733316e-01 1.04804587e+00 8.57505262e-01
8.26619208e-01 -5.26415348e-01 -5.33528984e-01 7.79999018e-01
2.09571072e-03 3.17150950e-01 5.83062768e-01 -8.02453935e-01
1.09201705e+00 7.76706219e-01 1.82402015e-01 -7.08385587e-01
-2.55257040e-01 -2.86903381e-01 -7.19239950e-01 -4.31681991e-01
5.83843827e-01 -1.28920853e-01 -9.25502062e-01 1.68471122e+00
2.77960509e-01 -1.69819877e-01 3.56156290e-01 9.00895238e-01
8.81582379e-01 1.01340222e+00 -6.63268492e-02 -2.37868235e-01
1.37189126e+00 -1.31562662e+00 -5.86791575e-01 -4.38199103e-01
7.58179843e-01 -7.92261541e-01 1.37256718e+00 3.37791979e-01
-1.30892825e+00 -7.12299645e-01 -8.12397122e-01 -3.08090866e-01
-6.78794160e-02 1.21868543e-01 -1.13821387e-01 2.35089853e-01
-8.39122474e-01 3.76419723e-01 -3.98440510e-01 -4.41924445e-02
-9.15860815e-04 4.87039387e-02 -3.22419256e-02 -4.70352232e-01
-1.34176099e+00 7.92574883e-01 4.60566521e-01 -2.12377328e-02
-6.22060299e-01 -4.79294777e-01 -8.05716515e-01 4.36986923e-01
5.77930868e-01 -7.01901555e-01 1.60847735e+00 -1.10269511e+00
-1.51046002e+00 3.10425818e-01 -6.26466632e-01 -5.56033611e-01
3.50509912e-01 -3.54074866e-01 -1.06711991e-01 5.34857810e-01
-1.27848536e-02 8.20702910e-01 8.15926671e-01 -8.62067282e-01
-6.78017318e-01 -3.12313288e-01 5.36049664e-01 4.18985218e-01
-1.81192040e-01 -2.48012185e-01 -3.65085512e-01 -5.15823603e-01
2.34494999e-01 -7.50715256e-01 -1.58216078e-02 -5.27994454e-01
-4.97222483e-01 -5.39730191e-01 5.18861294e-01 -1.01812446e+00
1.63656223e+00 -1.81534958e+00 3.81762207e-01 -1.29245341e-01
1.93923578e-01 3.86459172e-01 -4.31162834e-01 6.11203492e-01
3.55699360e-01 7.40347132e-02 -3.29950452e-01 -2.44360775e-01
-1.88328281e-01 3.47677082e-01 -5.83956242e-01 -5.98260686e-02
7.81716257e-02 1.10472667e+00 -7.96105146e-01 -5.56665301e-01
-2.32465297e-01 -1.23964339e-01 -7.20204175e-01 6.16756976e-01
-9.06787157e-01 2.57487833e-01 -5.55506289e-01 2.61305213e-01
5.81512690e-01 -2.41567552e-01 -1.12416064e-02 1.73301548e-01
8.14733207e-02 7.11327553e-01 -8.77676547e-01 1.51826060e+00
-7.00176954e-01 4.37253088e-01 -7.64632877e-03 -1.03684998e+00
9.87196982e-01 2.51993835e-01 -1.33917272e-01 -9.95119572e-01
-1.90806121e-01 2.57401198e-01 1.17037401e-01 -7.74138749e-01
8.58618915e-01 3.34732272e-02 -8.72939453e-02 7.63805687e-01
-2.29219601e-01 4.40633968e-02 3.24804634e-01 3.24893266e-01
1.14311826e+00 -1.37776345e-01 1.36540338e-01 1.82649776e-01
8.48445594e-01 -6.76506944e-03 4.67797756e-01 9.36619937e-01
4.52607870e-02 6.34866118e-01 6.49692476e-01 -1.03198312e-01
-9.35735762e-01 -8.05560112e-01 3.06325644e-01 1.35851645e+00
2.10006371e-01 -5.03711998e-01 -9.54462528e-01 -1.12930536e+00
-2.14169398e-01 9.43481684e-01 -5.29042959e-01 -2.02650294e-01
-9.22008932e-01 4.64898795e-02 4.62921888e-01 5.39865494e-01
6.27128959e-01 -1.39964354e+00 -9.50646043e-01 4.85746831e-01
-8.28865409e-01 -8.36846292e-01 -7.47716427e-01 1.13790214e-01
-9.14449275e-01 -1.02432990e+00 -8.58023405e-01 -1.02934146e+00
6.13433957e-01 2.71260381e-01 1.30337071e+00 4.87924695e-01
4.38026190e-01 1.06183462e-01 -8.82402778e-01 -1.14818715e-01
-6.00176096e-01 6.54914677e-01 -6.50603950e-01 -1.64829880e-01
3.81874174e-01 -1.18638307e-01 -6.76866472e-01 5.23537099e-01
-1.06637621e+00 2.12688357e-01 7.25923896e-01 1.04514813e+00
5.73023856e-01 -2.10280865e-01 1.10295892e+00 -1.08380067e+00
1.02897990e+00 -6.38205111e-01 -2.54290432e-01 8.07915807e-01
-4.91977960e-01 3.70495647e-01 1.10628986e+00 -4.21916276e-01
-1.07954812e+00 -2.86180437e-01 -3.61335099e-01 -1.13098815e-01
-1.71087414e-01 8.62756610e-01 -4.64334339e-02 5.86204529e-01
6.08613014e-01 5.69062650e-01 -6.68530837e-02 -4.82801914e-01
3.34687054e-01 9.18095231e-01 4.90438402e-01 -6.43410146e-01
2.99967349e-01 -2.07173109e-01 -6.68347836e-01 -4.88688201e-01
-1.09851074e+00 -5.90041220e-01 -4.48125988e-01 -3.62829603e-02
7.12500393e-01 -4.55086410e-01 -4.83012497e-01 2.71012098e-01
-1.33382535e+00 -4.50871944e-01 -2.36215279e-01 2.32100815e-01
-4.05579209e-01 4.65956450e-01 -7.47490406e-01 -5.68353057e-01
-5.91565609e-01 -1.15714633e+00 8.08498323e-01 5.04034817e-01
-2.80062973e-01 -8.29627395e-01 -1.84750352e-02 6.57915056e-01
4.82799143e-01 -3.36440057e-01 1.29157579e+00 -7.41534531e-01
-4.54894900e-01 1.85534898e-02 -1.29617065e-01 2.63672590e-01
-6.92816302e-02 -2.70893961e-01 -6.80878043e-01 -2.41105020e-01
2.66691577e-03 -7.51445889e-01 1.10459220e+00 -2.37289164e-02
1.29563975e+00 -7.01264083e-01 1.85006429e-02 6.47476166e-02
1.08028984e+00 1.00958355e-01 7.69397199e-01 1.38204560e-01
2.88754344e-01 7.90906847e-01 8.23271215e-01 1.77119866e-01
7.13651478e-01 4.26023811e-01 4.21344876e-01 3.21895152e-01
-3.80603387e-03 -6.05023921e-01 3.20042133e-01 1.32037878e+00
3.34633559e-01 -5.97646117e-01 -7.89334536e-01 7.48151422e-01
-1.79585779e+00 -8.49764109e-01 -5.69070391e-02 2.09249544e+00
1.20538056e+00 2.26452127e-01 -7.23191863e-03 3.17578763e-01
5.27257681e-01 1.63403481e-01 -7.24897742e-01 -6.74389482e-01
3.29803340e-02 3.71919513e-01 -4.18222323e-02 6.98826551e-01
-6.02459490e-01 9.85850096e-01 5.49061632e+00 8.00790608e-01
-1.00953496e+00 -1.37412816e-01 5.38302839e-01 1.82110578e-01
-7.38083243e-01 9.62922126e-02 -7.25488484e-01 4.71065968e-01
1.11252332e+00 1.30740227e-02 4.09525961e-01 3.87160569e-01
-7.03515485e-02 -5.44111133e-01 -1.07343793e+00 4.49880391e-01
9.85909551e-02 -1.27431118e+00 2.00255007e-01 -3.93088222e-01
6.27418935e-01 -2.57154614e-01 -1.46250322e-01 6.94392920e-01
1.20349778e-02 -1.10164106e+00 7.76715934e-01 6.08624637e-01
5.30261040e-01 -7.60134697e-01 8.11244667e-01 1.10651481e+00
-9.54973996e-01 -2.61318266e-01 -6.44633830e-01 -1.72426879e-01
1.17685072e-01 3.34253550e-01 -8.52680206e-01 6.81085229e-01
3.74701232e-01 3.86929959e-01 -6.55465901e-01 1.16606474e+00
-5.80059469e-01 9.69691515e-01 -2.00463593e-01 -4.30344522e-01
3.42910230e-01 -5.73727710e-04 2.35802948e-01 1.02548444e+00
3.11121285e-01 3.58816653e-01 2.30878845e-01 5.79864740e-01
-2.98689812e-01 2.90635258e-01 -1.51059613e-01 2.11020589e-01
6.37372851e-01 7.36374319e-01 -4.21352327e-01 -4.06632036e-01
-5.83002448e-01 1.03804922e+00 6.79507554e-01 3.47347081e-01
-6.17118716e-01 -7.05227256e-01 1.52637176e-02 6.20524101e-02
5.42728305e-01 3.10125258e-02 -3.45958889e-01 -1.23510611e+00
3.66968215e-01 -1.27457631e+00 6.54331684e-01 -7.37996995e-01
-9.73889768e-01 6.77795649e-01 -1.94876254e-01 -9.84784722e-01
-5.25234997e-01 -8.09368119e-02 -7.54705310e-01 9.88542795e-01
-1.62990081e+00 -7.08267450e-01 -2.20652714e-01 5.71555078e-01
1.06049931e+00 2.43992180e-01 6.98513091e-01 -8.73842090e-02
-3.77343386e-01 7.44376659e-01 3.04256845e-03 2.91967001e-02
5.80361068e-01 -1.32221043e+00 2.08875552e-01 6.71745837e-01
2.69996881e-01 5.08920908e-01 5.62051296e-01 -5.55794358e-01
-1.23352718e+00 -8.04806232e-01 1.20917952e+00 -2.28950158e-01
2.89755493e-01 -1.10658772e-01 -1.48942482e+00 5.69521427e-01
5.08130133e-01 -4.28884536e-01 5.11576653e-01 7.30582848e-02
-3.77635747e-01 -5.72022386e-02 -9.39389169e-01 4.06782478e-01
6.07314289e-01 -6.09874845e-01 -1.22241163e+00 1.59479696e-02
1.00212991e+00 -5.05504370e-01 -6.63455009e-01 2.48874336e-01
3.45094562e-01 -1.06808877e+00 5.32908201e-01 -5.50589979e-01
8.27574432e-01 -7.10282177e-02 1.23343900e-01 -1.45348680e+00
6.00323379e-02 -3.80334646e-01 -3.66476536e-01 1.04566431e+00
6.74389005e-01 -3.88684511e-01 7.97425032e-01 3.46491277e-01
-1.85003296e-01 -1.26040936e+00 -9.93004978e-01 -4.22006428e-01
4.42365974e-01 -2.47782439e-01 8.59300673e-01 4.45187390e-01
3.11336517e-01 6.62185490e-01 -1.31183892e-01 -5.82780205e-02
1.47869006e-01 6.50796831e-01 6.68798387e-01 -8.84650469e-01
-3.69229913e-01 -2.91300207e-01 4.45030600e-01 -1.92701912e+00
2.57544041e-01 -7.70631135e-01 3.63268673e-01 -1.73721147e+00
8.12971517e-02 -5.39025962e-01 -2.66590342e-02 3.75382960e-01
-6.07313037e-01 -3.26285243e-01 2.28971243e-01 2.55893975e-01
-8.80402565e-01 6.12916946e-01 1.59564567e+00 -1.80522040e-01
-2.55716860e-01 3.20897311e-01 -6.44235015e-01 3.95831794e-01
1.03596520e+00 -3.35578114e-01 -6.44541860e-01 -7.20842421e-01
3.36155444e-01 7.96425045e-01 4.06921469e-02 -7.60713041e-01
4.84157801e-01 -2.01921493e-01 2.47833833e-01 -8.20908368e-01
-1.95103958e-02 -4.55807924e-01 -5.00134647e-01 4.17445958e-01
-9.06861842e-01 2.18778610e-01 -1.99523917e-03 5.63774705e-01
-4.68849421e-01 -8.15186918e-01 5.05101144e-01 -2.00876147e-01
-2.72714764e-01 -6.00768290e-02 -4.69290257e-01 4.47404951e-01
7.36182034e-01 -4.75926325e-02 -5.37899137e-01 -6.57878995e-01
-4.62740332e-01 6.92454875e-01 3.39738995e-01 4.28384721e-01
8.81106257e-01 -8.42347443e-01 -8.73198092e-01 9.85223800e-02
6.08067587e-02 3.96113455e-01 4.15835559e-01 6.01503134e-01
-4.46399122e-01 6.93486929e-01 3.04320715e-02 -4.39912349e-01
-1.24051940e+00 5.73631346e-01 1.71713516e-01 -8.28839898e-01
-5.61471522e-01 6.40511811e-01 4.60999757e-02 -5.65510035e-01
3.11102599e-01 -6.66369915e-01 -5.76186657e-01 1.59082308e-01
6.88124180e-01 2.00222999e-01 1.63296267e-01 -3.29691201e-01
1.65455952e-01 4.33303952e-01 -3.61890554e-01 -1.74566507e-02
9.45233941e-01 -3.02664429e-01 -4.85363416e-02 2.49528974e-01
9.84008074e-01 1.81417674e-01 -1.18824732e+00 -5.98804474e-01
1.27491713e-01 -2.62640417e-01 -2.94953287e-01 -9.36562896e-01
-7.60249794e-01 1.13240731e+00 -9.18383896e-02 4.40447778e-01
1.17680621e+00 1.77910566e-01 1.25697052e+00 5.50734818e-01
1.79994151e-01 -9.96788085e-01 3.04195076e-01 8.35832536e-01
9.12951052e-01 -9.18019593e-01 -3.80149096e-01 -1.87134340e-01
-6.36485219e-01 1.34705937e+00 8.41706276e-01 2.60092199e-01
8.36516023e-02 -2.17277601e-01 -2.13394072e-02 2.08092570e-01
-1.04528451e+00 -1.72356933e-01 3.10885727e-01 2.16852739e-01
3.50785941e-01 7.43970424e-02 -4.22183752e-01 7.61069298e-01
-4.48870361e-01 -9.04830694e-02 6.55676961e-01 1.01475799e+00
-8.97916734e-01 -1.18490684e+00 -4.10690218e-01 6.96754038e-01
-3.63929212e-01 -7.43367299e-02 -3.30364108e-01 2.76363909e-01
-1.65855363e-01 1.29243875e+00 -6.85769692e-02 -3.37704688e-01
3.63363206e-01 2.70975173e-01 3.36412996e-01 -6.45113409e-01
-9.09620345e-01 -2.49131769e-01 2.22086702e-02 -3.49923879e-01
-1.61675543e-01 -6.05513752e-01 -1.41108334e+00 -3.29852283e-01
-4.64515865e-01 5.74276328e-01 1.07143171e-01 1.12843275e+00
2.87043512e-01 4.43836063e-01 5.16007781e-01 -8.88371542e-02
-9.64367867e-01 -1.17659044e+00 -4.22719680e-02 2.12265730e-01
5.49737871e-01 -6.49158508e-02 -1.79606900e-01 -1.23248294e-01] | [11.410008430480957, 8.123052597045898] |
9dcc7233-615c-4594-89ee-16440eabda6f | scaling-up-discourse-quality-annotation-for | null | null | https://aclanthology.org/2022.lrec-1.353 | https://aclanthology.org/2022.lrec-1.353.pdf | Scaling up Discourse Quality Annotation for Political Science | The empirical quantification of the quality of a contribution to a political discussion is at the heart of deliberative theory, the subdiscipline of political science which investigates decision-making in deliberative democracy. Existing annotation on deliberative quality is time-consuming and carried out by experts, typically resulting in small datasets which also suffer from strong class imbalance. Scaling up such annotations with automatic tools is desirable, but very challenging. We take up this challenge and explore different strategies to improve the prediction of deliberative quality dimensions (justification, common good, interactivity, respect) in a standard dataset. Our results show that simple data augmentation techniques successfully alleviate data imbalance. Classifiers based on linguistic features (textual complexity and sentiment/polarity) and classifiers integrating argument quality annotations (from the argument mining community in NLP) were consistently outperformed by transformer-based models, with or without data augmentation. | ['Gabriella Lapesa', 'Neele Falk'] | null | null | null | null | lrec-2022-6 | ['argument-mining'] | ['natural-language-processing'] | [ 1.65172875e-01 7.42870808e-01 -5.57180703e-01 -4.04347479e-01
-9.99330759e-01 -9.66437876e-01 1.19418526e+00 8.68981600e-01
-4.34157848e-01 7.96677768e-01 1.10649633e+00 -1.09274483e+00
-3.84378523e-01 -8.93896759e-01 -3.42194647e-01 -2.95724183e-01
4.98920858e-01 7.48471200e-01 -3.22611481e-01 -3.36502582e-01
5.89911640e-01 -2.14373335e-01 -1.41753447e+00 6.65554345e-01
1.46473992e+00 6.70016527e-01 -6.74333572e-01 4.06744391e-01
-5.52007318e-01 1.02607286e+00 -4.82691586e-01 -9.82521713e-01
1.52538791e-01 -4.71366972e-01 -1.33114278e+00 -3.28597546e-01
4.32712972e-01 1.66649461e-01 4.13684726e-01 8.95797968e-01
6.17700636e-01 -5.41659057e-01 6.76086366e-01 -9.72445905e-01
-3.61629337e-01 1.20787513e+00 -2.11557165e-01 2.10913196e-01
3.93205583e-01 1.02855392e-01 1.46727347e+00 -4.27118063e-01
1.06770837e+00 1.33688056e+00 8.94380152e-01 1.99298002e-03
-1.45573223e+00 -3.05500239e-01 1.07595570e-01 8.97699967e-02
-1.84771821e-01 -6.05216444e-01 6.77199543e-01 -8.08686674e-01
7.04974055e-01 4.96346027e-01 1.01448488e+00 1.07831097e+00
-2.30719119e-01 5.56081891e-01 2.09257174e+00 -6.54386878e-01
9.77314413e-02 2.22832635e-01 5.48168600e-01 4.94958431e-01
7.56908178e-01 -2.34041706e-01 -3.11392069e-01 -5.54877281e-01
-1.05124526e-01 -3.71121347e-01 -1.61893919e-01 -1.04153089e-01
-1.27938700e+00 1.25886035e+00 1.92053646e-01 5.01041234e-01
-4.33552295e-01 -3.24809015e-01 6.42795861e-01 7.69026935e-01
8.41140509e-01 9.66330290e-01 -9.16189253e-01 -5.36549807e-01
-8.60640705e-01 4.89588857e-01 1.31348968e+00 2.43652642e-01
5.83391130e-01 -5.30918598e-01 -3.89246285e-01 7.85064876e-01
1.27915964e-01 3.95777613e-01 2.01558307e-01 -1.04687762e+00
8.40129435e-01 1.33482039e+00 2.07108453e-01 -8.38525951e-01
-5.93788564e-01 -5.38446903e-01 -3.06869477e-01 2.88542420e-01
9.28156435e-01 -4.52953964e-01 -5.84880650e-01 1.51435518e+00
6.01945937e-01 -1.02420521e+00 7.75321573e-02 7.32941866e-01
9.71658111e-01 2.67590284e-01 5.21416187e-01 -2.64376879e-01
1.54600286e+00 -5.51856875e-01 -7.60696828e-01 -1.41301304e-01
1.00360763e+00 -1.01945853e+00 1.30134654e+00 2.84748048e-01
-1.05927801e+00 1.20722875e-01 -7.62359798e-01 -1.63968354e-01
-3.74862939e-01 -4.81007025e-02 9.64657426e-01 8.80287826e-01
-4.52652842e-01 6.94509625e-01 -2.69876540e-01 -2.52171397e-01
6.19600356e-01 -6.19477518e-02 -2.74243146e-01 2.87585884e-01
-1.02479792e+00 1.29232895e+00 -4.41157669e-02 -3.34296584e-01
7.69286156e-02 -7.72304535e-01 -7.35027075e-01 -1.92112684e-01
5.57429790e-01 -7.67103136e-01 1.16501474e+00 -1.33196807e+00
-1.21661305e+00 1.22218680e+00 2.42736131e-01 -3.76076877e-01
8.06872189e-01 -2.04213545e-01 -1.05671950e-01 -4.74207401e-01
4.61025000e-01 8.09010044e-02 2.53533870e-01 -1.05500054e+00
-6.40553594e-01 -4.92523879e-01 3.24712843e-01 1.67337671e-01
-2.89109409e-01 2.65542448e-01 4.81533498e-01 -2.53915519e-01
1.22033253e-01 -7.59110987e-01 -1.78559676e-01 -4.33328241e-01
-3.19201380e-01 -6.22405767e-01 4.97967184e-01 -9.01664257e-01
1.31351101e+00 -1.47191250e+00 4.59350012e-02 2.44370922e-01
2.54854500e-01 3.86135697e-01 2.44037345e-01 5.84751248e-01
5.02087064e-02 7.77357221e-01 1.06421016e-01 3.65688652e-01
2.59949923e-01 3.29884067e-02 -2.49546349e-01 3.87952268e-01
7.94215873e-02 9.85654652e-01 -7.91939437e-01 -6.78252101e-01
-5.81666529e-02 5.45592271e-02 -8.89786899e-01 -3.31681043e-01
-2.80382544e-01 3.31314623e-01 -6.58946514e-01 7.88623273e-01
3.09314936e-01 -4.06492144e-01 7.96192586e-01 -3.11934888e-01
-6.34243131e-01 1.35150898e+00 -8.70301843e-01 1.30444574e+00
-5.60966730e-01 7.04263389e-01 2.30377585e-01 -1.04157937e+00
7.37757087e-01 2.57012010e-01 1.02182746e-01 -1.05824077e+00
4.83229905e-01 5.98068058e-01 5.12116849e-01 -6.39785588e-01
2.74781466e-01 -1.27501845e-01 -3.03334802e-01 7.28809774e-01
-3.14139724e-01 -1.58859506e-01 3.60302180e-01 2.82059312e-02
1.06384325e+00 7.94254020e-02 6.04915440e-01 -6.76784456e-01
2.81466991e-01 5.62618613e-01 8.44361067e-01 4.39056128e-01
-8.06287155e-02 4.22647335e-02 9.72171962e-01 -7.83529341e-01
-1.51551259e+00 -1.51658714e-01 -2.82678962e-01 1.02918494e+00
-4.68268752e-01 -6.39613807e-01 -6.13191187e-01 -9.47068989e-01
1.10513851e-01 5.61392725e-01 -7.62299955e-01 5.52362978e-01
-6.88382745e-01 -9.24273908e-01 2.81780213e-01 2.77677421e-02
4.66842830e-01 -8.46027255e-01 -7.98817694e-01 4.37935829e-01
-6.35125399e-01 -9.06532407e-01 3.69915277e-01 3.16365749e-01
-8.43254924e-01 -1.64147019e+00 -1.63325429e-01 -1.94548368e-01
3.91565442e-01 -2.27449745e-01 1.41042697e+00 2.87734330e-01
-7.40574896e-02 -1.70874178e-01 -3.84202659e-01 -6.90015376e-01
-7.06562877e-01 2.34765604e-01 -2.74347544e-01 -5.77077746e-01
2.68270969e-01 -6.18761420e-01 -5.59598744e-01 -1.92439894e-03
-4.94890213e-01 3.45670342e-01 5.21313846e-01 9.78793561e-01
-3.88374999e-02 -4.77608114e-01 5.36893070e-01 -1.43296301e+00
1.06663656e+00 -3.78026336e-01 -4.56769615e-01 6.03682585e-02
-1.10953557e+00 1.66423067e-01 1.93597004e-01 -2.25262880e-01
-1.05902326e+00 -8.72636855e-01 -2.56193995e-01 9.24092591e-01
8.79521742e-02 9.34332907e-01 -1.54177636e-01 9.10712779e-02
1.16154838e+00 -6.98349774e-01 1.22447647e-01 -4.07031119e-01
4.82681930e-01 6.42997622e-01 -1.39011547e-01 -8.76300335e-01
6.82091951e-01 4.59585100e-01 -1.62855268e-01 -3.57240528e-01
-1.43052542e+00 -1.34588495e-01 -2.80344546e-01 -1.52813971e-01
4.48830605e-01 -8.30697298e-01 -7.11515546e-01 -1.18816808e-01
-9.37270105e-01 -2.44740337e-01 -4.05937076e-01 5.00834703e-01
-2.95289397e-01 1.91679612e-01 -3.03815216e-01 -8.02948534e-01
-6.84869826e-01 -7.74346828e-01 5.45657456e-01 2.14790493e-01
-7.74013817e-01 -9.83468592e-01 3.28145891e-01 1.29318094e+00
4.81801182e-01 7.41287291e-01 1.13066661e+00 -8.52239072e-01
-1.34177834e-01 5.53877130e-02 -2.62175292e-01 1.57417864e-01
-1.21152587e-01 1.22019447e-01 -7.54346490e-01 2.80826420e-01
-1.13847464e-01 -6.46859527e-01 7.31978357e-01 2.73992449e-01
3.23389292e-01 -1.03275228e+00 -1.42321765e-01 7.55764917e-02
1.24306977e+00 -3.91501307e-01 4.17402446e-01 1.07187784e+00
1.61002904e-01 1.15517986e+00 7.39679515e-01 4.35754597e-01
5.96694410e-01 4.33183610e-01 5.32864183e-02 -7.21648782e-02
-1.15702473e-01 -2.24605173e-01 6.19210973e-02 6.42831266e-01
-4.65725154e-01 5.17202243e-02 -1.40302706e+00 8.05929244e-01
-1.78797591e+00 -1.12716258e+00 -7.93932140e-01 1.59107304e+00
1.17536008e+00 4.14712816e-01 1.69014394e-01 6.96478784e-01
2.41132468e-01 7.35954046e-02 2.62265354e-01 -8.01713645e-01
-3.81623447e-01 2.25479081e-01 3.56626511e-01 4.82658088e-01
-9.50632215e-01 7.72280097e-01 5.93742275e+00 3.08672786e-01
-7.93235123e-01 1.19598627e-01 7.90956974e-01 1.60885960e-01
-7.88719654e-01 5.89423060e-01 -5.75225890e-01 4.52338219e-01
9.22350526e-01 -1.67336136e-01 5.35669699e-02 6.45079255e-01
3.34233612e-01 -3.01676303e-01 -9.21630621e-01 2.92227805e-01
-3.17442000e-01 -1.70208168e+00 -2.53716022e-01 3.97508055e-01
9.90474403e-01 5.86083606e-02 -6.49239570e-02 3.34370792e-01
9.73569691e-01 -9.48088586e-01 8.71595204e-01 1.66000232e-01
4.00229335e-01 -4.37564343e-01 8.45381856e-01 5.68263054e-01
-2.83343881e-01 -3.42499584e-01 1.38623253e-01 -5.64234257e-01
2.03555692e-02 9.10913825e-01 -8.24464798e-01 4.03645754e-01
6.00982070e-01 3.48032206e-01 -5.58395147e-01 6.40000284e-01
-4.23651695e-01 9.70218956e-01 -1.70583159e-01 -1.77891538e-01
3.83994877e-01 -2.14274913e-01 5.33151209e-01 1.09541762e+00
5.82391247e-02 1.52735054e-01 8.57615471e-03 7.04063892e-01
-2.53126889e-01 5.47325611e-01 -4.55529600e-01 -3.31590950e-01
4.13383514e-01 1.51787984e+00 -4.97262567e-01 -4.07242417e-01
-2.77776480e-01 -7.49295354e-02 4.35462922e-01 -1.54511452e-01
-4.72712874e-01 2.29018167e-01 5.19735456e-01 3.80373210e-01
-1.81474164e-01 1.37479216e-01 -8.79396617e-01 -9.45040882e-01
1.76872879e-01 -1.31483042e+00 4.69668746e-01 -5.48315011e-02
-1.17128670e+00 8.68863761e-02 -3.31907004e-01 -9.13785279e-01
-1.61342040e-01 -3.94301385e-01 -6.50412679e-01 6.70019925e-01
-1.65768290e+00 -1.42804539e+00 -3.71195316e-01 -4.23125289e-02
1.99285001e-01 8.21562558e-02 7.67250478e-01 2.76814401e-01
-1.10787988e-01 1.29767090e-01 -2.12625384e-01 1.52750194e-01
7.29997039e-01 -1.32055402e+00 1.68355361e-01 3.16098869e-01
-3.85668203e-02 4.98815507e-01 9.73632634e-01 -6.30076349e-01
-1.14489090e+00 -6.11093104e-01 1.56652415e+00 -7.25514293e-01
8.62068594e-01 -4.21673554e-04 -6.01127923e-01 1.91495776e-01
4.49102312e-01 -5.31397343e-01 1.05624008e+00 8.98068368e-01
-5.70185006e-01 2.85215467e-01 -1.15193117e+00 4.98240143e-01
1.03384125e+00 -4.02975738e-01 -1.09157848e+00 4.71733451e-01
2.21077889e-01 -1.82839513e-01 -9.23402369e-01 3.75698984e-01
5.81519842e-01 -8.96837652e-01 6.59651697e-01 -7.90667236e-01
7.40634203e-01 3.98511142e-02 1.36302069e-01 -1.22474778e+00
-2.23725662e-01 -5.72132766e-01 4.22184914e-01 1.43825901e+00
7.84811914e-01 -4.99957234e-01 6.62098527e-01 5.63113093e-01
9.25858796e-04 -6.29125297e-01 -7.74268091e-01 -2.06452772e-01
3.87311548e-01 -1.33570045e-01 5.78917563e-01 1.44443548e+00
3.89926553e-01 7.24033177e-01 9.89620667e-03 -5.07671177e-01
5.14707327e-01 5.18034875e-01 1.03870296e+00 -1.95048761e+00
2.33218193e-01 -7.48454452e-01 8.61813873e-02 -1.26784578e-01
1.33212328e-01 -1.00755763e+00 -5.66064239e-01 -2.17323852e+00
3.31726342e-01 -7.29388595e-01 3.99712265e-01 4.14833337e-01
-1.41880631e-01 -8.46091360e-02 6.53146505e-02 2.66871721e-01
-2.56825298e-01 2.67877966e-01 1.22412813e+00 -2.49089941e-01
-2.33985305e-01 -1.85965493e-01 -1.33405900e+00 1.16767895e+00
7.80329287e-01 -4.64251578e-01 9.28492323e-02 -3.95399958e-01
1.22916985e+00 -1.13649182e-02 2.73650289e-01 -5.92097640e-01
6.55487413e-03 -4.07055259e-01 1.69511616e-01 -3.84503156e-01
-1.17802963e-01 -5.14235616e-01 1.04080275e-01 4.17082280e-01
-6.12318635e-01 9.54257548e-02 -1.83377147e-01 1.85951605e-01
-1.77083790e-01 -1.23528026e-01 4.90881503e-01 -1.62477627e-01
1.34560198e-01 -2.76963562e-01 -1.27581909e-01 5.37074208e-01
5.89408338e-01 1.06051844e-02 -9.25244689e-01 -1.44611880e-01
-5.63581288e-01 6.89179674e-02 5.23072898e-01 8.01279396e-02
-1.22861549e-01 -1.15961885e+00 -1.27765000e+00 -5.16853094e-01
9.34189856e-02 -3.97732139e-01 -2.20358297e-01 1.20294333e+00
-4.70215619e-01 5.29060125e-01 -1.43403336e-01 -1.80447429e-01
-1.55689597e+00 4.06638868e-02 -5.49695306e-02 -8.22728395e-01
-4.21279401e-01 4.26309526e-01 -5.84582329e-01 -5.86239100e-01
-2.11825445e-01 -2.68533051e-01 -7.11543560e-01 8.59881222e-01
1.13666818e-01 4.30953979e-01 8.31601694e-02 -5.25996268e-01
-1.79278180e-01 2.58800387e-01 1.29162565e-01 -3.33467722e-01
1.71812689e+00 1.19407438e-01 -3.56492609e-01 2.33418524e-01
7.27990031e-01 5.03509939e-01 -4.50864971e-01 -3.05722177e-01
5.58292747e-01 -4.51314688e-01 1.19807564e-01 -1.20582497e+00
-3.94187152e-01 6.20787501e-01 2.07373813e-01 4.75377947e-01
4.54088897e-01 1.37477247e-02 2.91195512e-01 3.85500759e-01
7.67854601e-02 -1.68002892e+00 -2.22034827e-01 5.51166534e-01
9.71202970e-01 -1.44003701e+00 3.83186221e-01 -3.68768722e-01
-6.45204186e-01 9.48507190e-01 1.40325025e-01 2.29202837e-01
5.57736218e-01 1.27105534e-01 1.98778361e-01 -5.95871508e-01
-1.00358498e+00 -3.20932955e-01 3.72129619e-01 2.68002748e-02
9.89523947e-01 3.49574327e-01 -1.51903570e+00 6.46996140e-01
-6.68028474e-01 -2.53571123e-01 4.20932502e-01 9.12613750e-01
-4.71960068e-01 -1.38788319e+00 -5.09521067e-01 6.88505530e-01
-9.63381171e-01 -2.15364531e-01 -8.55264366e-01 8.75482917e-01
3.81760925e-01 1.11374331e+00 -2.64513552e-01 5.14377542e-02
2.70494878e-01 2.08142862e-01 3.23456615e-01 -3.86848330e-01
-1.41361666e+00 2.09875964e-03 1.27814043e+00 -3.77992541e-01
-9.08047974e-01 -9.73105907e-01 -7.75228441e-01 -5.86980104e-01
-3.26916009e-01 5.38339794e-01 1.04659927e+00 9.69619155e-01
1.64758548e-01 2.86937863e-01 3.72510672e-01 -4.04290445e-02
-6.75561011e-01 -1.32744396e+00 1.54464275e-01 6.88977897e-01
8.95245001e-02 -5.20056486e-01 -4.47623819e-01 -1.30372748e-01] | [9.354303359985352, 9.712018966674805] |
c6b56c7f-42b6-4e52-bd05-a0b701571e9f | measuring-hidden-bias-within-face-recognition | 2110.09839 | null | https://arxiv.org/abs/2110.09839v1 | https://arxiv.org/pdf/2110.09839v1.pdf | Measuring Hidden Bias within Face Recognition via Racial Phenotypes | Recent work reports disparate performance for intersectional racial groups across face recognition tasks: face verification and identification. However, the definition of those racial groups has a significant impact on the underlying findings of such racial bias analysis. Previous studies define these groups based on either demographic information (e.g. African, Asian etc.) or skin tone (e.g. lighter or darker skins). The use of such sensitive or broad group definitions has disadvantages for bias investigation and subsequent counter-bias solutions design. By contrast, this study introduces an alternative racial bias analysis methodology via facial phenotype attributes for face recognition. We use the set of observable characteristics of an individual face where a race-related facial phenotype is hence specific to the human face and correlated to the racial profile of the subject. We propose categorical test cases to investigate the individual influence of those attributes on bias within face recognition tasks. We compare our phenotype-based grouping methodology with previous grouping strategies and show that phenotype-based groupings uncover hidden bias without reliance upon any potentially protected attributes or ill-defined grouping strategies. Furthermore, we contribute corresponding phenotype attribute category labels for two face recognition tasks: RFW for face verification and VGGFace2 (test set) for face identification. | ['Toby P. Breckon', 'Noura Al Moubayed', 'Furkan Tektas', 'Seyma Yucer'] | 2021-10-19 | null | null | null | null | ['face-identification'] | ['computer-vision'] | [ 4.17415857e-01 -4.78302911e-02 -4.77851301e-01 -8.30078304e-01
-1.94672152e-01 -6.26685381e-01 7.30920613e-01 7.55289719e-02
-2.48487443e-01 6.20486617e-01 2.23866209e-01 -6.57666564e-01
-4.32723492e-01 -6.27870917e-01 -1.80827752e-01 -8.58784616e-01
-1.55206725e-01 1.05683334e-01 -6.78569317e-01 4.27477844e-02
6.01885021e-01 7.87474573e-01 -1.85549068e+00 1.90788001e-01
7.32867718e-01 9.67779696e-01 -6.14710391e-01 3.76324862e-01
1.76744595e-01 1.76242709e-01 -6.07864976e-01 -5.22053480e-01
4.23909843e-01 -5.33764064e-01 -5.65805852e-01 -1.02773204e-01
1.16673970e+00 -2.02423036e-01 4.07616496e-01 1.08977008e+00
8.24619412e-01 -3.28610331e-01 1.09310102e+00 -1.50426400e+00
-9.09066558e-01 3.35835755e-01 -1.05447316e+00 1.40120968e-01
3.54871571e-01 1.66736543e-01 6.07992351e-01 -7.33029783e-01
5.37433982e-01 1.52399588e+00 7.91749716e-01 9.85798717e-01
-1.48696136e+00 -1.26651013e+00 2.30400175e-01 -9.16882306e-02
-1.62351918e+00 -8.68961871e-01 8.69465888e-01 -8.27064753e-01
1.91390693e-01 5.33968866e-01 2.51074374e-01 1.13785684e+00
1.64756589e-02 -1.79496512e-01 1.84432244e+00 -3.60706210e-01
1.54718459e-01 3.81147683e-01 4.32373583e-01 7.04339266e-01
5.83001077e-01 5.91832340e-01 -6.30481601e-01 -7.34622598e-01
3.86887938e-01 -3.04174870e-01 -6.22576065e-02 -1.12906314e-01
-8.68141234e-01 7.01023102e-01 7.34298751e-02 1.20759852e-01
-2.50718385e-01 -7.35823438e-02 3.87850374e-01 4.21500444e-01
6.55126393e-01 3.61969888e-01 -1.34011239e-01 3.81082535e-01
-1.00063610e+00 1.40258670e-01 4.48216110e-01 4.03995395e-01
8.79030585e-01 1.10724382e-01 -3.80255103e-01 1.21308112e+00
4.63099450e-01 6.30696237e-01 2.42873982e-01 -7.33726203e-01
3.84404138e-02 5.14214039e-01 -6.47841347e-03 -1.06456459e+00
-3.99536192e-01 -5.75494021e-02 -6.61031485e-01 7.96790242e-01
6.78257108e-01 -2.25751117e-01 -1.07824802e+00 2.16074300e+00
5.66820979e-01 -1.61161169e-01 -2.04102799e-01 8.92793477e-01
7.93520033e-01 -2.45064184e-01 5.86378694e-01 -3.10165405e-01
1.74586952e+00 3.20030414e-02 -4.11678642e-01 1.51662752e-01
3.75017822e-01 -7.28967428e-01 8.52340937e-01 1.74862608e-01
-7.09629714e-01 -4.91279423e-01 -9.46350455e-01 2.83801109e-01
-5.31351507e-01 1.05743445e-01 6.22739851e-01 1.82739627e+00
-1.14348829e+00 2.54348099e-01 6.45304024e-02 -5.80680728e-01
8.12644601e-01 6.31002128e-01 -6.36015594e-01 1.34078249e-01
-8.20801616e-01 8.23262632e-01 -2.98319280e-01 3.42454910e-02
-8.15764368e-01 -1.07132769e+00 -6.85912490e-01 -5.54791987e-01
2.56296955e-02 -7.55519807e-01 4.77192223e-01 -1.48647976e+00
-1.04535079e+00 1.66795623e+00 -5.98652720e-01 6.11188188e-02
6.03068531e-01 3.04238051e-01 -7.22328544e-01 -1.07411416e-02
2.49980949e-02 6.22421265e-01 1.33824718e+00 -1.48319495e+00
-2.64918864e-01 -8.90712619e-01 -1.62168846e-01 -2.82960176e-01
-1.53204948e-01 8.18508208e-01 7.65356362e-01 -6.68801010e-01
-3.55956964e-02 -9.35309231e-01 2.10672662e-01 2.27658674e-01
-2.93465942e-01 -2.93847919e-01 8.25589240e-01 -6.39706254e-01
1.13315558e+00 -2.23432326e+00 -4.00282025e-01 5.62110662e-01
4.22688544e-01 9.02220085e-02 -2.00828895e-01 2.35755723e-02
-5.76962173e-01 8.11147034e-01 -2.30614007e-01 -1.66650370e-01
-2.65186839e-02 -1.11076020e-01 -2.59239197e-01 1.01899827e+00
5.10168374e-01 4.57062006e-01 -4.19409424e-01 -6.54917836e-01
-1.47724673e-01 4.80029315e-01 -6.02267385e-01 -7.81445354e-02
4.49935287e-01 3.57473522e-01 -1.37862414e-01 1.03125489e+00
1.11375499e+00 6.07848704e-01 1.27888054e-01 -2.44385913e-01
-2.04029754e-01 5.69058955e-02 -9.33856368e-01 6.59456551e-01
-2.12494090e-01 4.65935111e-01 1.57793388e-01 -5.63442409e-01
1.36173475e+00 2.64015287e-01 3.47952515e-01 -2.40168348e-01
2.05068797e-01 2.37159446e-01 5.69793224e-01 -2.65324056e-01
2.34753624e-01 -5.70216000e-01 3.19319427e-01 6.79468036e-01
-1.61439344e-01 3.22267473e-01 -1.55285269e-01 -2.99703091e-01
4.59446579e-01 -1.85210228e-01 4.91617024e-01 -9.17819440e-01
7.00528085e-01 -3.29133570e-01 6.91880822e-01 6.78205669e-01
-7.88593709e-01 7.41516948e-01 6.80139363e-01 -3.56456697e-01
-8.33342433e-01 -9.66928661e-01 -7.46641338e-01 1.11016524e+00
-2.36301452e-01 8.47487338e-03 -5.68103194e-01 -8.39703441e-01
5.76729417e-01 3.84129703e-01 -1.38536346e+00 -3.17069829e-01
-3.19366843e-01 -1.10313976e+00 9.42636728e-01 3.00538570e-01
1.60333335e-01 -7.32098997e-01 -3.51481974e-01 -6.76305950e-01
5.86154997e-01 -4.19959605e-01 -2.73154557e-01 -3.28492582e-01
-6.10948682e-01 -1.42629540e+00 -5.22354603e-01 -4.45167184e-01
9.26880360e-01 2.11978499e-02 1.00065303e+00 4.60720360e-01
-5.14015555e-01 4.64906514e-01 -1.55402526e-01 -7.76668072e-01
-3.89119238e-01 -4.10576224e-01 4.08194929e-01 7.39786685e-01
6.92004442e-01 -4.37848061e-01 -7.04716265e-01 4.87862140e-01
-4.74395007e-01 -4.57607090e-01 2.36074865e-01 5.83465099e-01
-7.30132684e-02 -4.12797064e-01 9.14584994e-01 -1.08368802e+00
6.80811763e-01 -6.88169539e-01 -1.83203042e-01 3.02165240e-01
-9.28584039e-01 -3.23481739e-01 5.68276905e-02 -5.84139287e-01
-1.20767379e+00 -4.17020977e-01 2.07970247e-01 -2.07779016e-02
-6.51183367e-01 7.93302257e-04 -3.45298588e-01 -4.31930989e-01
7.50252247e-01 -2.41582036e-01 3.60581428e-01 -2.15595827e-01
-5.60480207e-02 7.49859989e-01 2.24509209e-01 -9.78782535e-01
7.70018697e-01 6.29018843e-01 2.90372103e-01 -7.86944151e-01
-2.47336239e-01 -2.42260635e-01 -8.73997390e-01 -4.73679692e-01
8.17978323e-01 -6.71613634e-01 -9.64585543e-01 3.70315045e-01
-7.01863706e-01 2.81764716e-02 9.93290171e-02 4.44417864e-01
-2.10642979e-01 2.71804512e-01 1.82516929e-02 -1.32384193e+00
-2.19475046e-01 -8.02949786e-01 1.03179121e+00 1.76845063e-02
-8.00328076e-01 -8.93377841e-01 -1.06479444e-01 4.14779246e-01
3.70360821e-01 5.96192896e-01 1.19966638e+00 -6.60894513e-01
1.45136714e-01 -2.19355058e-02 -5.10007083e-01 2.57866085e-01
4.14072841e-01 4.76481587e-01 -1.50037920e+00 -1.37696832e-01
-1.13521270e-01 -2.40037933e-01 7.69286990e-01 2.75714129e-01
1.15032852e+00 -1.05850480e-01 -2.34987661e-01 6.39431238e-01
1.35701978e+00 3.66766512e-01 4.64524180e-01 -4.23814729e-02
6.63670659e-01 1.29099667e+00 2.61565357e-01 2.83044815e-01
-4.16342840e-02 4.37949330e-01 1.65331081e-01 -1.27305210e-01
-2.99672931e-01 5.65768182e-02 3.47085625e-01 -2.89494634e-01
-4.59217757e-01 2.72324830e-01 -1.01040316e+00 4.01759177e-01
-1.12183583e+00 -9.78696525e-01 -2.86585212e-01 2.34891915e+00
6.60093844e-01 -4.90989178e-01 4.14112598e-01 8.04384202e-02
1.10177290e+00 3.03282022e-01 -4.44248825e-01 -7.33745217e-01
-3.86844635e-01 2.28910729e-01 3.34715515e-01 4.89563644e-01
-8.48977327e-01 3.70133579e-01 7.06376982e+00 4.52128142e-01
-1.20717192e+00 -1.60070919e-02 1.02890241e+00 -1.86063260e-01
-3.92004699e-01 -9.26005170e-02 -7.87651479e-01 4.60441232e-01
6.41227961e-01 -1.74537286e-01 2.42763832e-01 3.88733506e-01
3.25055927e-01 -1.72368437e-01 -1.26249552e+00 1.00688100e+00
3.35539490e-01 -7.11820483e-01 9.37661976e-02 4.34596300e-01
6.15751803e-01 -5.87358892e-01 5.81804574e-01 -2.27759913e-01
1.40306905e-01 -1.47098053e+00 6.63844466e-01 2.63343632e-01
1.14669263e+00 -5.25190592e-01 4.23791796e-01 -4.17115152e-01
-6.49479926e-01 -2.86807418e-01 -2.07530990e-01 -1.87983006e-01
-3.05240989e-01 5.12381852e-01 -7.56466925e-01 3.46949250e-01
5.55412292e-01 2.24583328e-01 -8.92876506e-01 6.28754377e-01
2.02884495e-01 7.99435914e-01 -5.63390404e-02 2.10805997e-01
-2.00673506e-01 -3.13278019e-01 4.50546652e-01 1.06182325e+00
3.43684316e-01 3.32762413e-02 -1.93191320e-01 1.11113966e+00
1.05546251e-01 3.35823774e-01 -8.09180498e-01 1.23840183e-01
6.00786746e-01 1.26690137e+00 -5.90633154e-01 -1.57408282e-01
-4.42081034e-01 3.64464045e-01 3.86471301e-02 3.94436568e-01
-4.67214584e-01 6.43096119e-02 1.39628673e+00 3.02374095e-01
-3.54936659e-01 1.49166510e-01 -8.92708659e-01 -6.42732441e-01
-1.13513872e-01 -1.10941386e+00 6.87107980e-01 -3.49051595e-01
-1.60319471e+00 1.97083086e-01 1.35150760e-01 -7.84155309e-01
1.10058717e-01 -7.63979495e-01 -6.57914340e-01 1.51462603e+00
-1.36025035e+00 -1.42368567e+00 -3.73295933e-01 5.49043000e-01
-1.95460975e-01 -5.64607143e-01 7.28870749e-01 2.51121819e-01
-7.25721657e-01 9.19967532e-01 -5.74636996e-01 3.06790233e-01
1.26483643e+00 -1.03224063e+00 -7.27168918e-02 6.15015686e-01
-2.64373958e-01 1.07906115e+00 4.61025387e-01 -7.27679729e-01
-1.03612018e+00 -8.56168926e-01 9.37995672e-01 -7.76597679e-01
3.09618324e-01 -4.18193579e-01 -6.29426897e-01 3.64646196e-01
9.45368558e-02 -1.25482097e-01 1.33085537e+00 6.78871036e-01
-1.08235931e+00 -2.83703595e-01 -1.66142690e+00 5.80362380e-01
1.32335126e+00 -7.16684520e-01 -3.21650624e-01 -3.58789265e-02
-1.10788614e-01 4.43408072e-01 -9.89034355e-01 5.85072577e-01
9.20810640e-01 -1.35985553e+00 9.22018230e-01 -9.68809605e-01
3.77580941e-01 -3.86235148e-01 -1.89484105e-01 -1.09228885e+00
-4.59998906e-01 -4.55653101e-01 6.20622158e-01 1.71695852e+00
4.77823824e-01 -1.16901600e+00 6.89697981e-01 9.15410817e-01
2.47590154e-01 -4.32388932e-01 -8.18449795e-01 -6.37219965e-01
4.97649252e-01 -1.14888407e-01 1.06997073e+00 1.38595760e+00
-2.15854809e-01 -1.76003084e-01 -1.30031735e-01 2.66374946e-01
8.34307790e-01 2.33716682e-01 8.51855099e-01 -1.48574746e+00
2.84601718e-01 -8.33604038e-01 -4.77794021e-01 2.91515529e-01
6.52234793e-01 -8.37609828e-01 -3.79018843e-01 -6.07268929e-01
4.13160563e-01 -9.18710411e-01 -2.53199399e-01 4.14056152e-01
-4.02480841e-01 5.82125247e-01 1.07962430e-01 -9.03475508e-02
4.89445984e-01 -9.92786810e-02 8.77466142e-01 -1.27908111e-01
6.73942566e-02 -2.08343536e-01 -1.26249325e+00 4.84834790e-01
9.54208374e-01 -2.88876384e-01 -3.99831206e-01 3.67599167e-02
3.70175540e-02 -4.41119105e-01 7.89618909e-01 -6.12613559e-01
-2.43115291e-01 -4.52770084e-01 6.31239593e-01 1.00075521e-01
9.51413065e-02 -7.00383365e-01 1.25492290e-01 4.39545989e-01
-3.08505207e-01 1.53338283e-01 2.19972320e-02 1.50066003e-01
-2.97196675e-02 -2.01689467e-01 1.03744209e+00 4.42046374e-02
-5.39354742e-01 2.67227739e-01 -2.83918917e-01 -7.26119131e-02
1.12846267e+00 -6.85452878e-01 -6.87688529e-01 -1.07169777e-01
-5.45936942e-01 -2.96124518e-01 8.21462333e-01 5.11718273e-01
2.56487042e-01 -1.26196229e+00 -1.10011613e+00 6.15148008e-01
3.74608546e-01 -9.57737327e-01 7.45087713e-02 1.02333117e+00
-3.24314982e-02 1.25038568e-02 -6.65305197e-01 -5.03266692e-01
-2.01333785e+00 6.56583488e-01 5.52987993e-01 6.05664015e-01
1.68648437e-01 8.40305448e-01 5.22248328e-01 -3.22116047e-01
-1.91258028e-01 1.50558069e-01 -2.99157083e-01 5.85768759e-01
5.56407452e-01 6.32801771e-01 -1.04056932e-01 -1.03165460e+00
-6.55688643e-01 8.21540296e-01 1.26097232e-01 -1.63923539e-02
7.93787181e-01 -4.90318425e-02 -5.72110355e-01 2.73821324e-01
1.19930851e+00 4.51099664e-01 -6.85691476e-01 3.19713295e-01
1.79384574e-02 -9.30972517e-01 -2.90768027e-01 -8.93954515e-01
-9.15166259e-01 6.27805650e-01 9.47174191e-01 3.69226299e-02
1.00639975e+00 -1.49348408e-01 -3.44239563e-01 -3.42628211e-01
2.65376419e-01 -8.63950431e-01 -5.35584688e-01 -1.38055652e-01
9.77652252e-01 -1.41765821e+00 2.13433400e-01 -8.24784517e-01
-3.47369313e-01 7.28780687e-01 6.64832532e-01 1.34875551e-01
8.96758676e-01 7.20586628e-02 5.98271787e-01 -2.46469915e-01
-5.32642245e-01 -1.78922057e-01 6.17884040e-01 1.12312782e+00
8.39299858e-01 2.85476357e-01 -7.60684967e-01 3.39040399e-01
-5.07954895e-01 -3.19537610e-01 1.98692128e-01 5.80465019e-01
-2.49431320e-02 -1.12542331e+00 -8.93793285e-01 8.13798249e-01
-6.33872449e-01 1.20913843e-02 -9.68767762e-01 7.81982601e-01
5.85051358e-01 1.02189398e+00 3.82975817e-01 -3.17983985e-01
9.28044021e-02 5.24261832e-01 7.61714756e-01 -5.20576835e-01
-9.68215823e-01 -5.48081815e-01 4.30757731e-01 -3.27963501e-01
-7.74637043e-01 -1.08188033e+00 -4.72017467e-01 -6.43420935e-01
2.11942326e-02 -1.31014273e-01 5.54785669e-01 5.65509975e-01
3.14297140e-01 -2.01283887e-01 6.48856759e-01 -4.19470459e-01
-3.55397105e-01 -9.97615814e-01 -7.48066187e-01 7.06894279e-01
5.19870698e-01 -9.81361628e-01 -3.59066308e-01 6.11719787e-02] | [13.03630542755127, 1.2332451343536377] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.