{"id":"W4392182143","doi":"10.1109/icra57147.2024.10611316","title":"Safety Optimized Reinforcement Learning via Multi-Objective Policy Optimization","year":2024,"lang":"en","type":"preprint","venue":"","topic":"Software Reliability and Analysis Research","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Victoria","funders":"","keywords":"Reinforcement learning; Computer science; Reinforcement; Artificial intelligence; Psychology; Social psychology","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"],"consensus_categories":[],"category_scores_codex":[0.001244583,0.0003961167,0.0005543964,0.0008544285,0.0002566256,0.0008072414,0.001455863,0.0003723662,0.0002415411],"category_scores_gemma":[0.0007405761,0.0003446563,0.0004688311,0.00126464,0.00008396208,0.0002088445,0.007196515,0.001731216,0.0002696133],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0008007261,"about_ca_system_score_gemma":0.0009615772,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001650878,"about_ca_topic_score_gemma":0.00001173982,"domain_scores_codex":[0.996306,0.0003765658,0.0006734182,0.001267639,0.0008383089,0.0005380702],"domain_scores_gemma":[0.9977104,0.0003070558,0.0001796364,0.001135025,0.0004583362,0.0002095685],"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":[0.00001630544,0.00004869801,0.00001645894,0.0001320075,0.0002127484,0.000008766689,0.0006535934,0.9800077,0.000007944742,0.002766324,0.00006891481,0.01606053],"study_design_scores_gemma":[0.0003797114,0.00004705873,0.00001858222,0.0001137281,0.00003257698,0.000003366395,0.00003585761,0.9962987,0.0001758657,0.002176339,0.0003589824,0.0003592253],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0000221617,0.0002333889,0.9883779,0.002204661,0.0003073012,0.0005930716,0.000002264857,0.000945052,0.007314191],"genre_scores_gemma":[0.1189589,0.00121464,0.8612181,0.0002671598,0.0003404658,0.0002064192,0.0001494886,0.00005738817,0.01758749],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.1271598,"threshold_uncertainty_score":0.9999005,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02094009631972392,"score_gpt":0.3142482420144251,"score_spread":0.2933081456947012,"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."}}