{"id":"W3034736247","doi":"10.24963/ijcai.2020/374","title":"SVRG for Policy Evaluation with Fewer Gradient Evaluations","year":2020,"lang":"en","type":"preprint","venue":"","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal; McGill University","funders":"","keywords":"Reinforcement learning; Variance (accounting); Computer science; Convergence (economics); Function (biology); Artificial intelligence; Computation; Mathematical optimization; Gradient method; Bellman equation; Work (physics); Scale (ratio); Value (mathematics); Machine learning; Mathematics; Algorithm","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":[0.003096676,0.001530396,0.00196429,0.0008120543,0.0005259131,0.001218796,0.001641932,0.002093042,0.007032037],"category_scores_gemma":[0.01223341,0.0008382494,0.0009058607,0.0008668607,0.0009092833,0.001356269,0.001482611,0.00319374,0.002237434],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001127383,"about_ca_system_score_gemma":0.002615141,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.006099036,"about_ca_topic_score_gemma":0.007915528,"domain_scores_codex":[0.9980518,0.0008668919,0.0001306321,0.00032305,0.0004680623,0.0001596248],"domain_scores_gemma":[0.9964558,0.002075089,0.0002142847,0.0006209229,0.0005219123,0.000111817],"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":[0.0002390386,0.0001058713,0.0005597429,0.0001393299,0.00006755027,0.0001073433,0.0000625173,0.8236985,0.002400226,0.02433358,0.008846845,0.1394395],"study_design_scores_gemma":[0.00001129794,0.00001399239,0.0000365192,0.000005443932,0.000002475505,0.00000881018,0.000002741002,0.9944584,0.0003972092,0.00431317,0.0007465582,0.000003444879],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.005423009,0.0002345918,0.9894923,0.0002654476,0.00009121124,0.00008398367,0.000123515,0.002721303,0.001564651],"genre_scores_gemma":[0.2474214,0.000199942,0.7440842,0.0004773531,0.00009592639,0.0005331503,0.0009007421,0.001130418,0.00515677],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007032037,"threshold_uncertainty_score":0.02352452,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08629240404755965,"score_gpt":0.3720355181123799,"score_spread":0.2857431140648203,"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."}}