{"id":"W3209211538","doi":"10.48550/arxiv.2110.15572","title":"Understanding the Effect of Stochasticity in Policy Optimization","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Reinforcement Learning in Robotics","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Alberta","funders":"","keywords":"Mathematical optimization; Oracle; Convergence (economics); Initialization; Stochastic optimization; Rate of convergence; Computer science; Optimization problem; Mathematics; Key (lock); Economics","routes":{"ca_aff":true,"ca_fund":false,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":false},"retraction":null,"screen":null,"direct_labels":[{"model":"gemma","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"empirical","about_ca_system":false,"about_ca_topic":false,"confidence":"low","status":"direct model label, unvalidated"},{"model":"gpt","categories":[],"domain":null,"study_design":"simulation_or_modeling","genre":"methods","about_ca_system":false,"about_ca_topic":false,"confidence":"low","status":"direct model label, unvalidated"}],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.005600668,0.0009795551,0.001409535,0.000598921,0.0007120927,0.001791808,0.001330852,0.001697462,0.002037908],"category_scores_gemma":[0.04462805,0.0008724524,0.000809209,0.000443693,0.002656396,0.003778361,0.002781656,0.003200765,0.0002599935],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001836776,"about_ca_system_score_gemma":0.002619234,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.005166444,"about_ca_topic_score_gemma":0.002690119,"domain_scores_codex":[0.997121,0.001365734,0.0001366518,0.0004807828,0.0005824161,0.0003132767],"domain_scores_gemma":[0.9671645,0.0269282,0.002443626,0.001639745,0.001236712,0.000587158],"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":[0.00005518796,0.00002988608,0.000872486,0.00005539138,0.00003657284,0.00005095181,0.00006509164,0.9529535,0.001141604,0.03842221,0.0001864653,0.006130619],"study_design_scores_gemma":[0.000005067217,0.0000371301,0.0001685212,0.000009274041,0.000005408788,0.00001249775,0.000007767802,0.9833205,0.0005500547,0.01573917,0.0001384275,0.000006051764],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04455686,0.0003393774,0.9514996,0.0007837349,0.00004705379,0.00003990608,0.00003306425,0.0001463486,0.002553959],"genre_scores_gemma":[0.9160882,0.0005920406,0.08081082,0.0003779064,0.0000978362,0.00009241257,0.00007479238,0.0001454944,0.001720404],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.005600668,"threshold_uncertainty_score":0.02961946,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.106162173705461,"score_gpt":0.209392311833225,"score_spread":0.1032301381277639,"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."}}