{"id":"W3102590949","doi":"10.1088/1742-5468/ac3ae7","title":"Hausdorff dimension, heavy tails, and generalization in neural networks*","year":2021,"lang":"en","type":"article","venue":"Journal of Statistical Mechanics Theory and Experiment","topic":"Markov Chains and Monte Carlo Methods","field":"Mathematics","cited_by":20,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Agence Nationale de la Recherche","keywords":"Generalization; Stochastic gradient descent; Applied mathematics; Artificial neural network; Hausdorff dimension; Mathematics; Hausdorff distance; Computer science; Range (aeronautics); Dimension (graph theory); Metric (unit); Artificial intelligence; Pure mathematics; Mathematical analysis","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":[0.006092908,0.0008843944,0.001204869,0.001849459,0.0009572602,0.00152486,0.001308575,0.001465836,0.001254829],"category_scores_gemma":[0.03053104,0.0005079071,0.001031338,0.0008349335,0.004499766,0.004454482,0.003727882,0.002891259,0.0001563427],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002013976,"about_ca_system_score_gemma":0.0008872726,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002757943,"about_ca_topic_score_gemma":0.001377584,"domain_scores_codex":[0.9983857,0.0006153917,0.0001313361,0.0003672081,0.0003327291,0.000167618],"domain_scores_gemma":[0.9651509,0.02465038,0.00345485,0.003408431,0.001919831,0.001415658],"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":[0.0002289678,0.0001054648,0.01126138,0.0002621841,0.0001532825,0.0002969067,0.0005513538,0.5300042,0.0043255,0.4284207,0.00167558,0.02271441],"study_design_scores_gemma":[0.000008402135,0.00006411431,0.001796221,0.00003814158,0.00001219131,0.00005780309,0.00004237944,0.780853,0.001066928,0.2156555,0.0003729783,0.00003229532],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.3492411,0.002639916,0.6402736,0.001893746,0.0001046207,0.00005595562,0.0003176987,0.0004810815,0.00499236],"genre_scores_gemma":[0.9746663,0.0008287664,0.022903,0.0002224897,0.00008477767,0.00007856138,0.0002175983,0.00007293997,0.000925437],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.006092908,"threshold_uncertainty_score":0.03222275,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02945931311743106,"score_gpt":0.3379642316986228,"score_spread":0.3085049185811917,"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."}}