{"id":"W2963096987","doi":"","title":"A closer look at memorization in deep networks","year":2017,"lang":"en","type":"article","venue":"Jagiellonian University Repository (Jagiellonian University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":654,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal; Canadian Institute for Advanced Research; McGill University; Université de Montréal; Concordia University","funders":"Samsung; Institut de Valorisation des Données; Natural Sciences and Engineering Research Council of Canada; Samsung Advanced Institute of Technology; Canadian Institute for Advanced Research","keywords":"Memorization; Deep neural networks; Computer science; Artificial intelligence; Generalization; Deep learning; Robustness (evolution); Regularization (linguistics); Artificial neural network; Machine learning; Adversarial system; Noise (video); Mathematics; Cognitive psychology; Psychology","routes":{"ca_aff":true,"ca_fund":true,"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","sts"],"consensus_categories":[],"category_scores_codex":[0.0003477087,0.0004873408,0.0005179047,0.0008825139,0.003034165,0.0004611238,0.004593667,0.0004761601,0.00004889975],"category_scores_gemma":[0.00008742735,0.0006495372,0.0002763145,0.000903806,0.0004811232,0.003126668,0.002926979,0.0008673564,0.0000830725],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00155285,"about_ca_system_score_gemma":0.0002695573,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0009550848,"about_ca_topic_score_gemma":0.001130631,"domain_scores_codex":[0.9965017,0.0004792547,0.000325517,0.001324973,0.0004888386,0.0008797212],"domain_scores_gemma":[0.9961329,0.0001676582,0.0006726148,0.002416907,0.0002032151,0.0004067252],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"observational","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0009061918,0.0006276394,0.6465755,0.0001327912,0.0004045678,0.03109387,0.006032169,0.1917568,0.002257756,0.1068334,0.001385177,0.01199421],"study_design_scores_gemma":[0.008372591,0.0004116449,0.1742723,0.0003506275,0.0002612247,0.000308381,0.002038054,0.6825202,0.001366098,0.0003510062,0.1267395,0.003008367],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1847321,0.0001055203,0.567003,0.00208427,0.002855883,0.0008323239,0.000005019169,0.0008661632,0.2415157],"genre_scores_gemma":[0.9380847,0.00008944096,0.007425959,0.00008910522,0.000200077,2.765397e-7,0.000008133653,0.00003944557,0.05406284],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7533526,"threshold_uncertainty_score":0.9995956,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.006018646023883206,"score_gpt":0.1883984567121577,"score_spread":0.1823798106882745,"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."}}