{"id":"W4308364487","doi":"10.1109/tnnls.2022.3217189","title":"Monotonic Quantile Network for Worst-Case Offline Reinforcement Learning","year":2022,"lang":"en","type":"article","venue":"IEEE Transactions on Neural Networks and Learning Systems","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":16,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"","keywords":"Quantile; Reinforcement learning; Quantile regression; CVAR; Computer science; Quantile function; Monotonic function; Function (biology); Bellman equation; Q-learning; Mathematical optimization; Offline learning; Artificial intelligence; Expected shortfall; Machine learning; Econometrics; Online learning; Mathematics; Cumulative distribution function; Risk management; Statistics; Economics; Probability density function","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":["metaepi_narrow","sts"],"consensus_categories":[],"category_scores_codex":[0.001287982,0.0003604175,0.0004289364,0.000187996,0.003584122,0.000429545,0.0004672117,0.0001065846,0.00003011114],"category_scores_gemma":[0.00001659521,0.0003730181,0.0002014447,0.0006450524,0.0000497436,0.0003167907,0.00002971254,0.001881494,0.000005264619],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.000133928,"about_ca_system_score_gemma":0.0000366072,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001231775,"about_ca_topic_score_gemma":0.000006481243,"domain_scores_codex":[0.9967082,0.0006818013,0.0006758041,0.0006716871,0.0004569324,0.0008055364],"domain_scores_gemma":[0.9980834,0.0007984457,0.0004025091,0.0004431213,0.00008393309,0.000188569],"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":[0.00006829351,0.00002565886,0.00006863742,0.00003964239,0.00007075047,0.00007583574,0.0003244463,0.9926523,0.00001022417,0.0006282402,0.0004457658,0.005590223],"study_design_scores_gemma":[0.000775132,0.001500507,0.000008194828,0.0000465318,0.00004143143,0.0006360969,0.0003481389,0.9777313,0.00000594584,0.000003482736,0.01851405,0.0003891396],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.005306731,0.0004408435,0.9895608,0.0001478066,0.002988646,0.0009429316,0.000001432863,0.0004773885,0.0001334325],"genre_scores_gemma":[0.9917547,0.00007924741,0.001270106,0.0001718882,0.0002919007,0.0004698637,0.00001194574,0.00005755334,0.005892831],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9882907,"threshold_uncertainty_score":0.9998721,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02027412986570837,"score_gpt":0.2429601647000125,"score_spread":0.2226860348343041,"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."}}