{"id":"W2904453761","doi":"10.48550/arxiv.1812.02900","title":"Off-Policy Deep Reinforcement Learning without Exploration","year":2018,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":280,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Reinforcement learning; Computer science; Extrapolation; Artificial intelligence; Reinforcement; Class (philosophy); Uncorrelated; Space (punctuation); Action (physics); Control (management); Machine learning; Mathematics; Engineering","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.00175459,0.0008779161,0.0009195449,0.0002358588,0.0002485543,0.000637893,0.001275634,0.0009281942,0.001795976],"category_scores_gemma":[0.006565266,0.0003839192,0.0002757785,0.0002647619,0.001603134,0.001156481,0.001097471,0.001711086,0.0003052453],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0009173156,"about_ca_system_score_gemma":0.001365539,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004081399,"about_ca_topic_score_gemma":0.00296958,"domain_scores_codex":[0.9994866,0.0001809631,0.00002546252,0.0001173282,0.0001037463,0.0000858753],"domain_scores_gemma":[0.9972153,0.001867331,0.0002265053,0.0003140177,0.0002433279,0.000133557],"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":[0.0001961833,0.0001025265,0.0009824656,0.00007172533,0.00002939377,0.00005945359,0.00005794657,0.92636,0.001664713,0.01860477,0.001220653,0.05065031],"study_design_scores_gemma":[0.00001389427,0.00003694152,0.00005309807,0.00000492369,0.000002481471,0.000007197447,0.000003602458,0.9935582,0.0004494046,0.005643531,0.0002238565,0.000002925314],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.06402618,0.0002984046,0.9297512,0.0004241476,0.00007065608,0.00008364939,0.00005854042,0.000870413,0.004416904],"genre_scores_gemma":[0.9364135,0.0001042177,0.06045766,0.0002028105,0.0000226822,0.0001246775,0.00007287462,0.00005439977,0.002547248],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.004081399,"threshold_uncertainty_score":0.009279311,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.0781399192839883,"score_gpt":0.2175502889724059,"score_spread":0.1394103696884176,"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."}}