{"id":"W3172823396","doi":"10.48550/arxiv.2106.01336","title":"Improved Rates for Differentially Private Stochastic Convex Optimization with Heavy-Tailed Data","year":2021,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":13,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"","keywords":"Bounded function; Convex optimization; Lipschitz continuity; Mathematics; Regular polygon; Convex function; Combinatorics; Proper convex function; Matching (statistics); Function (biology); Applied mathematics; Mathematical optimization; Subderivative; Statistics; Pure mathematics; Mathematical analysis","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","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0003055153,0.0004862584,0.0005653908,0.0002662991,0.0002252518,0.0005561073,0.04096368,0.0004262701,0.00002298164],"category_scores_gemma":[0.003234553,0.0005044041,0.0001020021,0.0006813668,0.0002184617,0.001212007,0.1789051,0.0006332379,0.000004004815],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0001964856,"about_ca_system_score_gemma":0.0004771562,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00006096332,"about_ca_topic_score_gemma":0.00007514007,"domain_scores_codex":[0.9964389,0.0001205092,0.0002988366,0.002456847,0.000138128,0.0005467476],"domain_scores_gemma":[0.9803846,0.0002883218,0.0004807004,0.01835455,0.0003597093,0.0001320809],"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":[0.000459285,0.0004417788,0.0008072464,0.0006067821,0.0009822722,0.0002600774,0.00009524512,0.9754063,0.0003545313,0.0144954,0.004705482,0.001385611],"study_design_scores_gemma":[0.001028119,0.0001048922,0.00007076513,0.000198479,0.0001409151,0.000004666435,0.00003608556,0.9765185,0.0004605796,0.02082406,0.00002930897,0.0005836221],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02733016,0.0000617644,0.968777,0.001040197,0.0005509849,0.00103981,0.0002683954,0.0009066864,0.00002502464],"genre_scores_gemma":[0.6987797,0.00009044275,0.2995699,0.00005041931,0.00004054609,0.000008147332,0.001334396,0.00003810697,0.00008834548],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.6714495,"threshold_uncertainty_score":0.9997408,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.08526170636481022,"score_gpt":0.2270890065065224,"score_spread":0.1418273001417122,"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."}}