{"id":"W3120394327","doi":"10.1007/s10515-021-00313-x","title":"Faults in deep reinforcement learning programs: a taxonomy and a detection approach","year":2021,"lang":"en","type":"article","venue":"Automated Software Engineering","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":38,"is_retracted":false,"has_abstract":false,"ca_institutions":"Polytechnique Montréal","funders":"Fonds de Recherche du Québec-Société et Culture; Canadian Network for Research and Innovation in Machining Technology, Natural Sciences and Engineering Research Council of Canada","keywords":"Reinforcement learning; Computer science; Categorization; Artificial intelligence; Fault detection and isolation; Software; Machine learning; Software engineering; Programming language","routes":{"ca_aff":true,"ca_fund":true,"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":[0.0002110985,0.0001947393,0.0001911775,0.0001911119,0.00008413102,0.0001793689,0.0002175217,0.0001023567,0.000002177179],"category_scores_gemma":[0.0003572913,0.0002226182,0.00003776576,0.0007347932,0.00001319195,0.0004359383,0.0002534089,0.0003510063,0.000007650413],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001291047,"about_ca_system_score_gemma":0.00004099965,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001642172,"about_ca_topic_score_gemma":0.00000296422,"domain_scores_codex":[0.9986405,0.00003952716,0.0003059131,0.0003853496,0.0002279794,0.0004007521],"domain_scores_gemma":[0.999404,0.00008277166,0.00007715905,0.0002831702,0.00006535836,0.00008759039],"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":[0.000001190173,0.00001221172,0.002134752,0.00008569149,0.00001968221,0.00003456164,0.0004831199,0.9830033,0.0002236038,0.0001630102,0.000001421474,0.01383751],"study_design_scores_gemma":[0.0003720941,0.00005631621,0.00275767,0.00008155998,0.000004969716,0.00006679403,0.00006961809,0.9947466,0.0006674416,0.000001995088,0.0009373189,0.0002376695],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01661594,0.000139928,0.9799865,0.000008905316,0.0001369234,0.0002674763,2.980257e-8,0.002736699,0.0001075309],"genre_scores_gemma":[0.7958848,0.00001465701,0.2038474,0.00001588424,0.00001610411,0.0001223858,0.00001131688,0.00001878522,0.00006864298],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7792689,"threshold_uncertainty_score":0.9078106,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01206261324535972,"score_gpt":0.2026977882953615,"score_spread":0.1906351750500018,"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."}}