{"id":"W3101581426","doi":"","title":"Deep learning versus kernel learning: an empirical study of loss landscape geometry and the time evolution of the Neural Tangent Kernel","year":2020,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Neural Networks and Applications","field":"Computer Science","cited_by":20,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Initialization; Artificial neural network; Kernel (algebra); Chaotic; Artificial intelligence; Tangent; Computer science; Linearization; Deep learning; Nonlinear system; Machine learning; Geometry; Algorithm; Mathematics; Physics; Discrete mathematics","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":[0.0003234592,0.0001311292,0.0002221693,0.00005326177,0.0003740665,0.0002674657,0.0005594307,0.00005024881,0.000001616146],"category_scores_gemma":[0.00009826265,0.00007748987,0.00004789261,0.0007955338,0.00009571833,0.001283137,0.0002189032,0.0003083967,0.00000441963],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002078527,"about_ca_system_score_gemma":0.00003366771,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003970473,"about_ca_topic_score_gemma":0.000002093103,"domain_scores_codex":[0.9984418,0.0002528685,0.0005427082,0.0001706808,0.0004302297,0.0001617597],"domain_scores_gemma":[0.9987576,0.0001211951,0.0006370895,0.0002181525,0.0001945291,0.00007143486],"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":[0.0003760999,0.0001523963,0.06254462,0.000388393,0.00005018894,9.887814e-7,0.03475111,0.8638301,0.0002921046,0.001568656,0.0002206647,0.03582463],"study_design_scores_gemma":[0.001163141,0.000274061,0.01136638,0.00001935436,0.00001832289,0.00001100549,0.002940659,0.9838576,0.00001845254,0.000009546917,0.0002369212,0.00008459167],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.9826525,0.0003301722,0.01491739,0.001121482,0.0001679958,0.0005533391,8.83238e-7,0.0001027644,0.0001534111],"genre_scores_gemma":[0.9997357,0.000003618812,0.00003424985,0.00009086803,0.00007597698,0.00002798834,0.00000377256,0.000005893364,0.0000218964],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.1200274,"threshold_uncertainty_score":0.3159945,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02115437571371863,"score_gpt":0.263866539708851,"score_spread":0.2427121639951324,"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."}}