{"id":"W3034426742","doi":"","title":"On the Global Convergence Rates of Softmax Policy Gradient Methods","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Stochastic Gradient Optimization Techniques","field":"Computer Science","cited_by":60,"is_retracted":false,"has_abstract":true,"ca_institutions":"Google (Canada); University of Alberta","funders":"","keywords":"Softmax function; Mathematics; Applied mathematics; Initialization; Rate of convergence; Bounded function; Entropy (arrow of time); Mathematical optimization; Computer science; Mathematical analysis; Physics; Artificial neural network; Artificial intelligence","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":[0.01399367,0.002029287,0.00178829,0.002040979,0.001016796,0.002657219,0.001903821,0.001870007,0.006353655],"category_scores_gemma":[0.08193297,0.0007869434,0.00140508,0.001180713,0.003803023,0.005344062,0.004316058,0.006235855,0.001529197],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001790141,"about_ca_system_score_gemma":0.002122564,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002244779,"about_ca_topic_score_gemma":0.0018338,"domain_scores_codex":[0.995894,0.00230904,0.0001768768,0.0005068696,0.000778091,0.0003351392],"domain_scores_gemma":[0.9517057,0.0409589,0.001542606,0.0023645,0.0026977,0.000730632],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"simulation_or_modeling","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0005591377,0.00012656,0.003173841,0.0006390379,0.0001542176,0.0001800132,0.0004527468,0.5007182,0.003348097,0.4093667,0.005316202,0.07596518],"study_design_scores_gemma":[0.00002673542,0.00009842859,0.0004148478,0.0001621654,0.00002571008,0.00005563247,0.00004126317,0.908459,0.001994324,0.08715637,0.001539153,0.00002628927],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.0154223,0.002615289,0.9706143,0.001347542,0.0001752303,0.00008704802,0.0001034892,0.000492844,0.009142016],"genre_scores_gemma":[0.5800518,0.006189335,0.3935958,0.001592224,0.0007418306,0.0009044636,0.000558943,0.002073812,0.01429178],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.01399367,"threshold_uncertainty_score":0.07400644,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1321246909531475,"score_gpt":0.2731406912845342,"score_spread":0.1410160003313867,"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."}}