{"id":"W4378771708","doi":"10.48550/arxiv.2305.18246","title":"Provable and Practical: Efficient Exploration in Reinforcement Learning via Langevin Monte Carlo","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":5,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Alberta Machine Intelligence Institute; Canadian Institute for Advanced Research","keywords":"Reinforcement learning; Markov chain Monte Carlo; Computer science; Monte Carlo method; Markov decision process; Posterior probability; Mathematical optimization; Scalability; Gaussian; Regret; Artificial intelligence; Algorithm; Markov process; Mathematics; Machine learning; Bayesian probability; Physics","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":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002710675,0.001277844,0.001706479,0.0004941478,0.0006991493,0.001206109,0.0019689,0.001651197,0.003888207],"category_scores_gemma":[0.01336358,0.0008625507,0.0007604792,0.0005932983,0.002491447,0.001887361,0.002653808,0.003186923,0.00063717],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001632279,"about_ca_system_score_gemma":0.002564151,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.004791046,"about_ca_topic_score_gemma":0.00502431,"domain_scores_codex":[0.9988386,0.0005289766,0.00004342297,0.0001645516,0.0002939918,0.0001304953],"domain_scores_gemma":[0.9939002,0.00484415,0.0003062268,0.0003687497,0.0003289818,0.0002516461],"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":[0.00007136999,0.00003504075,0.0004777941,0.00006407389,0.00002757027,0.00006752389,0.00005661618,0.943332,0.0005021087,0.03933984,0.00117638,0.01484966],"study_design_scores_gemma":[0.00001004328,0.000007615263,0.00001737395,0.000005151615,0.000002110562,0.000005358237,0.000002095864,0.9880355,0.0001050744,0.01164158,0.0001655304,0.000002576656],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.009180003,0.0002579501,0.9866738,0.0004445423,0.00003966222,0.00004738339,0.00004248069,0.0004657373,0.002848494],"genre_scores_gemma":[0.7254192,0.000400815,0.2687368,0.0004707277,0.0001019265,0.0004769962,0.0001805512,0.000384333,0.00382852],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004791046,"threshold_uncertainty_score":0.01433557,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1289487721838383,"score_gpt":0.2268522974340941,"score_spread":0.09790352525025586,"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."}}