{"id":"W2150394007","doi":"","title":"From PAC-Bayes Bounds to KL Regularization","year":2009,"lang":"en","type":"article","venue":"","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université Laval","funders":"","keywords":"Coordinate descent; Regularization (linguistics); Mathematics; Boosting (machine learning); Upper and lower bounds; Convex function; Algorithm; Proximal gradient methods for learning; Bayes' theorem; Regular polygon; Elastic net regularization; Mathematical optimization; Applied mathematics; Kullback–Leibler divergence; Computer science; Convex optimization; Artificial intelligence; Regression; Convex combination; Statistics; Bayesian probability; 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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00007998468,0.00008991687,0.00008725088,0.0001252415,0.00006705132,0.0001993944,0.0005034961,0.00004558695,0.00005425071],"category_scores_gemma":[0.00006983694,0.00008376847,0.000025412,0.0004877561,0.000009874011,0.0003290295,0.00007114716,0.00003607112,0.00005552772],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00003389745,"about_ca_system_score_gemma":0.00002138049,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002720641,"about_ca_topic_score_gemma":0.000001781168,"domain_scores_codex":[0.9991685,0.0000172871,0.0001506397,0.0003026238,0.000207529,0.0001534054],"domain_scores_gemma":[0.9992816,0.00003013847,0.0000370011,0.0004774305,0.00008244249,0.00009142022],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.000003110944,0.00005409124,0.00007090838,6.310755e-7,0.00000487852,0.00000235004,0.0008038064,0.0007991591,0.00207515,0.9361187,0.01761253,0.04245464],"study_design_scores_gemma":[0.0002885305,0.0004239051,0.004925429,0.00003318897,0.000006858875,0.000005041632,0.0000286281,0.254373,0.04228358,0.6925828,0.004584362,0.0004646633],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0006388806,0.00001031157,0.9850278,0.004522943,0.0001553802,0.0001551716,0.00000127202,0.0008662159,0.008621968],"genre_scores_gemma":[0.2406065,0.000001327148,0.7558082,0.002413848,0.0000604913,0.000008081223,0.000009201379,0.000004440303,0.001087989],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2535738,"threshold_uncertainty_score":0.3415979,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.00881585654239763,"score_gpt":0.2416711849258008,"score_spread":0.2328553283834032,"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."}}