{"id":"W4388092895","doi":"10.48550/arxiv.2310.18811","title":"Hierarchical Framework for Interpretable and Probabilistic Model-Based Safe Reinforcement Learning","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Software Reliability and Analysis Research","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Science Foundation Ireland; European Commission; Canadian Institute of Steel Construction","keywords":"Reinforcement learning; Interpretability; Probabilistic logic; Computer science; Artificial intelligence; Machine learning; Black box; Domain (mathematical analysis); Focus (optics)","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0007944762,0.0002773044,0.0004177307,0.0003476819,0.0002978747,0.0002527736,0.001346564,0.0003487121,0.00001261433],"category_scores_gemma":[0.001142142,0.0003010665,0.0002833991,0.0006104573,0.0002077105,0.0001792427,0.002211607,0.001147762,0.00002529573],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002428377,"about_ca_system_score_gemma":0.0004286141,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006179361,"about_ca_topic_score_gemma":0.00001173937,"domain_scores_codex":[0.9975604,0.0001775197,0.0002652371,0.00131856,0.0001780299,0.0005002383],"domain_scores_gemma":[0.9969559,0.001415299,0.0001458924,0.00101329,0.0002283889,0.0002412208],"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":[0.00004492104,0.00003390742,0.0005642044,0.0002940061,0.00006055298,0.00001594703,0.0001243884,0.8213322,0.000002273704,0.1768092,0.00003301358,0.0006854013],"study_design_scores_gemma":[0.0001772182,0.00007237819,0.00004240077,0.0001669206,0.00003864501,2.468219e-7,0.00001336227,0.6808752,0.000008626348,0.3183132,0.00008307144,0.0002087003],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01784325,0.00002661865,0.9805732,0.0003933652,0.0001238273,0.0005617549,0.000006692431,0.0003333979,0.0001379316],"genre_scores_gemma":[0.9661595,0.00008805553,0.03172769,0.00006676478,0.00003505708,0.00001419663,0.00002692683,0.0000217829,0.001859977],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9488454,"threshold_uncertainty_score":0.9999442,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.09495309765304245,"score_gpt":0.2452769170882211,"score_spread":0.1503238194351786,"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."}}