{"id":"W3202097587","doi":"","title":"Learning One Representation to Optimize All Rewards","year":2021,"lang":"en","type":"article","venue":"Neural Information Processing Systems","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"Université de Montréal","funders":"","keywords":"Computer science; Representation (politics); Markov decision process; Artificial intelligence; Reinforcement learning; A priori and a posteriori; Machine learning; Markov process; Temporal difference learning; Mathematics","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":["scholarly_communication"],"consensus_categories":[],"category_scores_codex":[0.000397404,0.0001480859,0.0002017328,0.0001816481,0.0002576611,0.001948632,0.0004399693,0.00007754767,0.000007063779],"category_scores_gemma":[0.0004330518,0.0001554516,0.00004651492,0.0008205595,0.00001381978,0.004895332,0.0002262669,0.0002521466,0.0002211588],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001050128,"about_ca_system_score_gemma":0.0001295639,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002523971,"about_ca_topic_score_gemma":3.940029e-7,"domain_scores_codex":[0.9979658,0.0001277282,0.0006694891,0.0002315237,0.0006987865,0.0003067169],"domain_scores_gemma":[0.998405,0.00005686645,0.0003905897,0.0003806896,0.0006373291,0.0001294847],"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":[0.000003673543,0.000005698917,0.0001528693,0.0001302303,0.000008938043,0.000003240719,0.003529894,0.9776489,0.0004119393,0.0009129422,0.0007323239,0.01645931],"study_design_scores_gemma":[0.000253623,0.00006866658,0.000220492,0.0001263446,0.000006606404,0.00005485972,0.0006038564,0.9804281,0.001510239,0.00001030592,0.01652518,0.0001917013],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.003236144,0.00006014124,0.9854096,0.00132387,0.0007224199,0.0002518636,4.422879e-7,0.0004982012,0.008497358],"genre_scores_gemma":[0.9584229,0.000009677467,0.03746135,0.001298405,0.0001205913,0.00004992551,0.00006446659,0.00001295815,0.002559744],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9551867,"threshold_uncertainty_score":0.9990875,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.04197470961659992,"score_gpt":0.2918375720424113,"score_spread":0.2498628624258114,"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."}}