{"id":"W2607390531","doi":"10.1007/s40687-018-0148-y","title":"Deep relaxation: partial differential equations for optimizing deep neural networks","year":2018,"lang":"en","type":"preprint","venue":"Research in the Mathematical Sciences","topic":"Model Reduction and Neural Networks","field":"Physics and Astronomy","cited_by":21,"is_retracted":false,"has_abstract":false,"ca_institutions":"McGill University","funders":"Air Force Office of Scientific Research; Army Research Office; Office of Naval Research; Simons Foundation; U.S. Department of Energy","keywords":"Curse of dimensionality; Partial differential equation; Artificial neural network; Nonlinear system; Computer science; Homogenization (climate); Relaxation (psychology); Stochastic gradient descent; Applied mathematics; Context (archaeology); Stochastic partial differential equation; Mathematical optimization; Mathematics; Artificial intelligence; Mathematical analysis; Physics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001919445,0.001791524,0.001840168,0.0006437359,0.0004175044,0.001458203,0.001784317,0.002641653,0.004655503],"category_scores_gemma":[0.009021468,0.001136774,0.001161193,0.0009419404,0.001464139,0.001985222,0.002449033,0.003845104,0.0008099427],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001502272,"about_ca_system_score_gemma":0.002072696,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00705741,"about_ca_topic_score_gemma":0.007685985,"domain_scores_codex":[0.999486,0.00023986,0.00002577371,0.00009084033,0.0001054915,0.0000521675],"domain_scores_gemma":[0.9982898,0.001235143,0.00008587306,0.0001158898,0.0001964994,0.00007677121],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000084579,0.00007186407,0.0003431292,0.0002696688,0.00009468731,0.00006891816,0.0000719699,0.8382052,0.00106911,0.1104489,0.007949331,0.04132273],"study_design_scores_gemma":[0.000009210083,0.000007592997,0.00002187247,0.00001119661,0.000005413393,0.000004827521,0.00000468943,0.9739674,0.0001561979,0.02498124,0.0008266374,0.000003659383],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003726071,0.0008468027,0.9911516,0.0006728459,0.0001643217,0.00003981631,0.0001341646,0.0002479333,0.003016378],"genre_scores_gemma":[0.3259158,0.001836,0.642779,0.001360245,0.0006379096,0.0008320085,0.00114286,0.001263194,0.02423303],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.00705741,"threshold_uncertainty_score":0.01557422,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2019696259217802,"score_gpt":0.422924077972178,"score_spread":0.2209544520503978,"validation_status":"score_only:v0-immature-baseline","note":"Baseline scores from an immature model (maturity gate not passed). Scores rank; they never assert a category."}}