{"id":"W3176342773","doi":"","title":"OptiDICE: Offline Policy Optimization via Stationary Distribution Correction Estimation","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Hyperparameter; Reinforcement learning; Benchmark (surveying); Computer science; Bootstrapping (finance); Set (abstract data type); Mathematical optimization; Online and offline; Offline learning; Artificial intelligence; Machine learning; Mathematics; Online learning; Econometrics","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":[0.001798568,0.001678811,0.002206263,0.000717259,0.0005251712,0.00171601,0.002236865,0.001862959,0.00386],"category_scores_gemma":[0.00962144,0.0008725745,0.0007147885,0.0006974193,0.001233178,0.001919568,0.002155328,0.003588654,0.001327472],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001209379,"about_ca_system_score_gemma":0.003723854,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.007563427,"about_ca_topic_score_gemma":0.00774251,"domain_scores_codex":[0.9990023,0.0002736774,0.00006097256,0.0003127572,0.0002291611,0.0001212233],"domain_scores_gemma":[0.9967577,0.002105755,0.0002338406,0.0004702031,0.000284882,0.0001475519],"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":[0.0001826908,0.0002033911,0.000984475,0.0001347996,0.000073348,0.00007816788,0.00006683471,0.8017306,0.001419073,0.008775121,0.006385382,0.1799661],"study_design_scores_gemma":[0.00002853064,0.00003036918,0.00006413044,0.00001003327,0.000004547257,0.00001827215,0.000005647179,0.9944111,0.0007021969,0.003680689,0.001037419,0.000007041594],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009484437,0.000466906,0.9818772,0.0003432123,0.0001504633,0.0001323361,0.0001363219,0.003983814,0.003425349],"genre_scores_gemma":[0.44395,0.0004302789,0.5447258,0.0007849704,0.0001694097,0.0004527809,0.0008356369,0.001064369,0.007586744],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.007563427,"threshold_uncertainty_score":0.01503879,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03488079122941772,"score_gpt":0.2006629496360864,"score_spread":0.1657821584066687,"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."}}