{"id":"W3082716386","doi":"10.48550/arxiv.2008.13773","title":"Beyond variance reduction: Understanding the true impact of baselines on policy optimization","year":2020,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Bandit Algorithms Research","field":"Decision Sciences","cited_by":4,"is_retracted":false,"has_abstract":true,"ca_institutions":"McGill University","funders":"","keywords":"Reduction (mathematics); Variance reduction; Variance (accounting); Econometrics; Economics; Computer science; Environmental science; Mathematics","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":[0.0155857,0.001204108,0.002504814,0.001275849,0.00129881,0.004991753,0.001745436,0.002840595,0.002992947],"category_scores_gemma":[0.1038685,0.0008008158,0.0009432449,0.001014351,0.003670829,0.009314548,0.003539331,0.005237567,0.0005456461],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002755211,"about_ca_system_score_gemma":0.002284216,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002880643,"about_ca_topic_score_gemma":0.002088369,"domain_scores_codex":[0.9933408,0.00354901,0.0002596252,0.001292864,0.001038922,0.0005187542],"domain_scores_gemma":[0.955556,0.03437899,0.002853082,0.004786415,0.001450331,0.0009751238],"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":[0.0004263203,0.0002105921,0.005571587,0.0002294295,0.0001930882,0.0001313406,0.0003135288,0.5190145,0.001748954,0.4115176,0.002733516,0.0579096],"study_design_scores_gemma":[0.00002707813,0.0001398348,0.001051722,0.00005045616,0.00002214545,0.00003463512,0.00004695369,0.7515382,0.0006915109,0.2455266,0.0008485661,0.00002228443],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1395593,0.002738589,0.8348233,0.005917937,0.0001913269,0.0001343928,0.0002710983,0.0008208053,0.01554333],"genre_scores_gemma":[0.9458205,0.0006817067,0.05048161,0.0006529715,0.0001235322,0.0001278141,0.0001572553,0.0002323058,0.001722276],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.0155857,"threshold_uncertainty_score":0.08242607,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3346032301983551,"score_gpt":0.3524518645244382,"score_spread":0.01784863432608302,"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."}}