{"id":"W3210467483","doi":"10.48550/arxiv.2111.00870","title":"Statistical Consequences of Dueling Bandits","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Advanced Bandit Algorithms Research","field":"Decision Sciences","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Toronto","funders":"","keywords":"Regret; Computer science; Thompson sampling; Operations research; Management science; Machine learning; Mathematics; 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":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","insufficient_payload"],"consensus_categories":[],"category_scores_codex":[0.001532624,0.00027749,0.0007250941,0.0005264662,0.0001474305,0.0001830917,0.001767817,0.0003055529,0.001605549],"category_scores_gemma":[0.003709927,0.0002657836,0.0002224146,0.001179316,0.001067071,0.0003027714,0.00188997,0.0007902421,0.0001355662],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001528158,"about_ca_system_score_gemma":0.0008892489,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0001599608,"about_ca_topic_score_gemma":0.00006417343,"domain_scores_codex":[0.9961589,0.0005760414,0.0006100498,0.001429501,0.0007894792,0.0004360189],"domain_scores_gemma":[0.9938175,0.002905384,0.0004717215,0.001246892,0.001301786,0.0002567072],"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":[0.0002069472,0.0002209048,0.02343346,0.0001536036,0.0002160774,0.002894128,0.0003714737,0.9477423,0.0005452814,0.01596809,0.0008415589,0.007406122],"study_design_scores_gemma":[0.001781101,0.0003185458,0.02317275,0.0005678477,0.0002160737,0.00005037325,0.005495337,0.4606523,0.01195462,0.4929009,0.001315621,0.00157454],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5042049,0.0001166955,0.4936008,0.00004774182,0.0004559444,0.0001805609,0.0001527195,0.00003427521,0.001206348],"genre_scores_gemma":[0.994461,0.00028042,0.003013221,0.00001711743,0.00006543287,6.861396e-7,0.00003670908,0.00001722567,0.002108217],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.4905876,"threshold_uncertainty_score":0.9999794,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3207690795247978,"score_gpt":0.3360774280993675,"score_spread":0.01530834857456964,"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."}}