{"id":"W2971055146","doi":"","title":"Fast Convergence of Natural Gradient Descent for Over-Parameterized Neural Networks","year":2019,"lang":"en","type":"article","venue":"arXiv (Cornell University)","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":30,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Initialization; Jacobian matrix and determinant; Gradient descent; Maxima and minima; Parameterized complexity; Convergence (economics); Artificial neural network; Applied mathematics; Mathematics; Computer science; Mathematical optimization; Algorithm; Mathematical analysis; Artificial intelligence","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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0001160574,0.0001565507,0.0002193454,0.0001438047,0.00005972925,0.00002845638,0.0008399122,0.00006280425,0.00002117215],"category_scores_gemma":[0.00002802358,0.0001676728,0.0001438989,0.0005820592,0.00007773385,0.0003814872,0.0002248721,0.0001000113,0.000005317826],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00007714796,"about_ca_system_score_gemma":0.00002536238,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001739543,"about_ca_topic_score_gemma":0.00000225846,"domain_scores_codex":[0.9989135,0.00004145097,0.0001844094,0.0004929361,0.00007144838,0.0002962081],"domain_scores_gemma":[0.9988788,0.0001440426,0.0001911125,0.0005414121,0.0001552149,0.00008943759],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0001549685,0.0001176542,0.01121683,0.00004178064,0.00005275336,0.0000123827,0.0001633136,0.5006101,0.0009284229,0.4857311,0.0001850578,0.0007856544],"study_design_scores_gemma":[0.0007407904,0.0001949506,0.001450069,0.00001880557,0.00001455296,0.000002821483,0.00001720817,0.994463,0.000883524,0.002006943,0.00002493814,0.0001823937],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3205087,0.00001601225,0.6783341,0.00001414745,0.0005830848,0.0003724842,0.000002816591,0.0001152522,0.00005350623],"genre_scores_gemma":[0.9857384,0.00001238055,0.01389382,0.00007407898,0.00001282523,0.000001674049,0.000005247383,0.000009492313,0.0002520929],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.6652297,"threshold_uncertainty_score":0.6837497,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03222558805984906,"score_gpt":0.1798108982464483,"score_spread":0.1475853101865993,"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."}}