{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.001335809,0.0002308516,0.0004267151,0.0005222058,0.0001687769,0.0001254339,0.002780929,0.0002140509,0.000141249],"category_scores_gemma":[0.003770969,0.0001850875,0.0001474131,0.002438261,0.0003952414,0.00031987,0.002293832,0.0007396169,0.00006465655],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002534435,"about_ca_system_score_gemma":0.000650418,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001902328,"about_ca_topic_score_gemma":0.0002734138,"domain_scores_codex":[0.9970718,0.0003805919,0.0005861251,0.001045062,0.0005388095,0.0003775868],"domain_scores_gemma":[0.9959764,0.001331521,0.000636453,0.001519197,0.0003680579,0.000168351],"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":[0.00008847005,0.00003103741,0.004165369,0.00001221951,0.00001707535,0.0001265536,0.00007372678,0.9885399,0.000004813996,0.003167623,0.002941163,0.0008320907],"study_design_scores_gemma":[0.0005474791,0.00003953739,0.003281377,0.00003819322,0.00001300593,0.000001567573,0.0004404069,0.8942457,0.0001254152,0.09705177,0.003947516,0.0002680532],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02542161,0.0002536943,0.966964,0.002610211,0.000364754,0.0008095918,0.002039257,0.00005472175,0.001482155],"genre_scores_gemma":[0.9965786,0.0008418409,0.00102744,0.000102002,0.0001195468,0.000001857417,0.000274765,0.00002144374,0.001032494],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.971157,"threshold_uncertainty_score":0.7547651,"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."}}