{"id":"W3085735333","doi":"10.48550/arxiv.2009.06799","title":"The Importance of Pessimism in Fixed-Dataset Policy Optimization","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Bandit Algorithms Research","field":"Decision Sciences","cited_by":23,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University; Université de Montréal","funders":"","keywords":"Pessimism; Core (optical fiber); Computer science; Value (mathematics); Order (exchange); Mathematical optimization; Artificial intelligence; Econometrics; Economics; Machine learning; 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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.03015423,0.002144508,0.00304116,0.001581107,0.001813267,0.006150447,0.003438239,0.0034806,0.002662442],"category_scores_gemma":[0.1769186,0.001398151,0.001558751,0.001836894,0.005512182,0.009918544,0.004494096,0.00773227,0.0005458486],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.004095358,"about_ca_system_score_gemma":0.004684683,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002179769,"about_ca_topic_score_gemma":0.001490548,"domain_scores_codex":[0.9819201,0.01026066,0.0008279578,0.002669407,0.002621215,0.00170071],"domain_scores_gemma":[0.8095816,0.1627936,0.009587283,0.01068469,0.004388583,0.002964249],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0007496413,0.0001909836,0.003578562,0.0003101531,0.0001950698,0.0001558998,0.000266731,0.8427249,0.001010186,0.1280823,0.002744354,0.01999124],"study_design_scores_gemma":[0.00005678304,0.0001427076,0.0002931963,0.00007293744,0.00002454392,0.00007347978,0.00005626893,0.8660347,0.0008589808,0.131857,0.0005060238,0.00002334852],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06355803,0.002045398,0.9233038,0.00436783,0.0002037221,0.0001118895,0.0001927447,0.0005529092,0.005663601],"genre_scores_gemma":[0.8746093,0.001240715,0.1202018,0.001242505,0.0003298449,0.0002614721,0.0002323471,0.0003372264,0.00154484],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.03015423,"threshold_uncertainty_score":0.1594727,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2439048591061609,"score_gpt":0.3377463759698552,"score_spread":0.0938415168636943,"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."}}