{"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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.0003567304,0.0003020602,0.0003461068,0.0001790823,0.0001160491,0.00006247321,0.002896332,0.0001708908,0.00002898299],"category_scores_gemma":[0.0005064196,0.0002690158,0.0002268458,0.001488569,0.0002419018,0.0001246894,0.002165176,0.0003534818,0.00002546409],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002986196,"about_ca_system_score_gemma":0.0003453701,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001993108,"about_ca_topic_score_gemma":0.00000340089,"domain_scores_codex":[0.9980834,0.0003454876,0.0002599978,0.0008954284,0.0001351267,0.000280528],"domain_scores_gemma":[0.9975074,0.0004341888,0.0004105871,0.001268922,0.0002214426,0.000157437],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"theoretical_or_conceptual","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.00001358124,0.00004310112,0.0003152688,0.00002647558,0.00005432014,0.0000167864,0.000160422,0.05896636,0.00001554106,0.9397418,0.0003832518,0.0002630985],"study_design_scores_gemma":[0.0001038137,0.00008669919,0.0002976469,0.00005290578,0.00002445584,0.000001568499,0.0000239668,0.449756,0.001350105,0.5480401,0.00006169121,0.0002010802],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":"theoretical_or_conceptual","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.006197568,0.0000277909,0.9888895,0.001235838,0.0004379664,0.0004204079,0.00002766654,0.0003030064,0.002460226],"genre_scores_gemma":[0.9350761,0.00007126485,0.06426056,0.0004450962,0.00003370449,0.000002808984,0.000004855512,0.00001120441,0.00009437412],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.9288785,"threshold_uncertainty_score":0.9999762,"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."}}