{"id":"W2943463079","doi":"10.48550/arxiv.1904.13262","title":"Implicit Regularization of Discrete Gradient Dynamics in Linear Neural Networks","year":2019,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Sparse and Compressive Sensing Techniques","field":"Engineering","cited_by":27,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Initialization; Regularization (linguistics); Parameterized complexity; Artificial neural network; Convergence (economics); Computer science; Applied mathematics; Rank (graph theory); Gradient descent; Algorithm; Mathematical optimization; Mathematics; 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":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.000060081,0.0002218446,0.0003195164,0.0002240566,0.00001779487,0.00001243687,0.0003250663,0.0002941821,0.000004028865],"category_scores_gemma":[0.000005883821,0.0002721884,0.0001209034,0.0002854204,0.00004307587,0.00007279919,0.0003167894,0.0004635144,0.00000174683],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002619191,"about_ca_system_score_gemma":0.00001399907,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001445584,"about_ca_topic_score_gemma":0.0001214445,"domain_scores_codex":[0.9991634,0.0000401593,0.000197324,0.0003438711,0.00004009623,0.0002151895],"domain_scores_gemma":[0.999184,0.00002826576,0.0001066283,0.0005793458,0.00005854132,0.00004328542],"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":[0.00001620744,0.00001159854,0.005638744,0.00004759465,0.00003411144,0.00003042297,0.00002434206,0.9833769,0.00005024731,0.01059209,0.00004621162,0.0001315675],"study_design_scores_gemma":[0.0001593922,0.00002146795,0.001424199,0.0001642374,0.00003993728,0.000001365177,0.00002855848,0.9931106,0.0002190255,0.004572224,0.00001354497,0.0002454634],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5049847,0.00005963784,0.4926819,0.00000593251,0.0004345736,0.0002340702,0.00001485212,0.0002369241,0.001347328],"genre_scores_gemma":[0.9991773,0.0002135254,0.0002739786,0.000007902232,0.00003937456,4.183921e-7,0.0001036909,0.00003765471,0.0001461723],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4941925,"threshold_uncertainty_score":0.9999731,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02612952777006198,"score_gpt":0.1747604210850382,"score_spread":0.1486308933149762,"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."}}