{"id":"W7015655229","doi":"","title":"Towards alignment of Reinforcement Learning agents; for consideration of safety, robustness and fairness.","year":2024,"lang":"en","type":"dissertation","venue":"eScholarship@McGill (McGill)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Robustness (evolution); Reinforcement learning; Control theory (sociology); Control (management)","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":[0.001414176,0.0006448035,0.000936622,0.0005407542,0.0005372026,0.0001724467,0.0008286829,0.0005148104,0.00005351495],"category_scores_gemma":[0.0007074304,0.0006809856,0.0003272096,0.0004906335,0.00008136617,0.000787576,0.0004351816,0.0007894693,0.00001171615],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0004314109,"about_ca_system_score_gemma":0.0001588522,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00009948739,"about_ca_topic_score_gemma":0.00004894149,"domain_scores_codex":[0.9953908,0.0002133494,0.00167962,0.000985781,0.001188835,0.000541579],"domain_scores_gemma":[0.9966893,0.0003384304,0.001329283,0.0007312399,0.0007308658,0.0001809124],"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":[0.0001133965,0.0000427575,0.000008364585,0.002418007,0.0003440995,0.00001006376,0.0001081453,0.5761154,0.004505429,0.3780934,0.00001105911,0.03822992],"study_design_scores_gemma":[0.004378482,0.002685746,0.0005490477,0.005347478,0.001190543,0.00004973425,0.001658347,0.6062022,0.3014296,0.02266906,0.05074287,0.003096855],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5734562,0.00275838,0.2084902,0.0004073019,0.0158,0.01432921,0.0006869858,0.001921623,0.1821501],"genre_scores_gemma":[0.9772069,0.0004512531,0.01158997,0.00005459391,0.00002908197,0.000151432,0.0006769245,0.0001171747,0.009722662],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4037507,"threshold_uncertainty_score":0.9995641,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02355615140994175,"score_gpt":0.2650591302153393,"score_spread":0.2415029788053975,"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."}}