{"id":"W4315588467","doi":"10.48550/arxiv.2301.02952","title":"Learning Symbolic Representations for Reinforcement Learning of Non-Markovian Behavior","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada; Government of Canada; Canadian Institute for Advanced Research; Agencia Nacional de Investigación y Desarrollo; Microsoft Research","keywords":"Reinforcement learning; Computer science; Artificial intelligence; Abstraction; Action (physics); Representation (politics); Function (biology); Markov process; State (computer science); Temporal difference learning; Markov decision process; Machine learning; Theoretical computer science; Algorithm; Mathematics","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":[0.0005925968,0.0003959961,0.0005305415,0.0006477762,0.0004114739,0.0001580362,0.00196141,0.0003228233,0.0000256508],"category_scores_gemma":[0.0003830305,0.0005072811,0.0004311918,0.0008590049,0.0001280926,0.0003842206,0.002585221,0.001079476,0.0000769646],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002416091,"about_ca_system_score_gemma":0.0002739921,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002683365,"about_ca_topic_score_gemma":0.000009432861,"domain_scores_codex":[0.9972922,0.0001629311,0.0005710211,0.001149879,0.000245505,0.0005784939],"domain_scores_gemma":[0.9967412,0.000472171,0.0009225452,0.001274545,0.0004105968,0.000178917],"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":[0.00001751212,0.00002867177,0.01280584,0.000187994,0.0001233822,0.00004716912,0.000836084,0.9701473,0.0001064967,0.01522267,0.0001362041,0.0003406935],"study_design_scores_gemma":[0.0006074395,0.0002723919,0.004244685,0.0001910827,0.0001786605,0.00000165347,0.0004103258,0.9923964,0.0002153471,0.0006222909,0.0003925301,0.0004672504],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04229413,0.000007071922,0.9534913,0.00005665769,0.0007133769,0.001024958,0.00000353464,0.0004721362,0.001936886],"genre_scores_gemma":[0.9670489,0.000124682,0.005496534,0.0000171872,0.00007489214,0.00002326791,0.0001230289,0.00005645539,0.02703508],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9479947,"threshold_uncertainty_score":0.9997379,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08623908006050819,"score_gpt":0.2431459818392111,"score_spread":0.1569069017787029,"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."}}