{"id":"W4388514504","doi":"10.48550/arxiv.2311.03683","title":"Preventing Arbitrarily High Confidence on Far-Away Data in Point-Estimated Discriminative Neural Networks","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":0,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Government of Canada; Canadian Institute for Advanced Research","keywords":"Discriminative model; Computer science; Artificial neural network; Training set; Class (philosophy); Machine learning; Test data; Artificial intelligence; Logit; De facto; Point (geometry); Set (abstract data type); Data set; Mathematics","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow","open_science","research_integrity"],"consensus_categories":["open_science"],"category_scores_codex":[0.001011265,0.0005647115,0.0006102969,0.0005242422,0.0002547688,0.0002962485,0.006419835,0.0003905542,0.000020051],"category_scores_gemma":[0.0006177624,0.0006585479,0.0001343625,0.001323352,0.0001972598,0.001247181,0.01160332,0.002564968,0.00005293065],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0003095075,"about_ca_system_score_gemma":0.0001869255,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001910111,"about_ca_topic_score_gemma":0.0003683714,"domain_scores_codex":[0.9954013,0.0006395953,0.0004497583,0.002533775,0.0002242118,0.0007513437],"domain_scores_gemma":[0.9951571,0.0008874125,0.0005996833,0.003061365,0.0001236479,0.0001707543],"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":[0.0000427978,0.00006110876,0.004092259,0.00005585532,0.0000503234,0.001105177,0.0002367338,0.8788544,0.000001766774,0.1146608,0.00009152133,0.0007472794],"study_design_scores_gemma":[0.0005451922,0.00005382781,0.01602499,0.0005159653,0.00005642313,0.000004390611,0.0001195226,0.9523442,0.000005681894,0.02972621,0.000008540997,0.0005949904],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.06149385,0.00002910552,0.9345579,0.0004686528,0.001727839,0.0005275542,0.00003245859,0.0006591642,0.000503451],"genre_scores_gemma":[0.9941422,0.00006505375,0.004623802,0.0001585087,0.0001292793,0.000002277427,0.0002059739,0.00005375246,0.0006191233],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.9326484,"threshold_uncertainty_score":0.9997361,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1618378683921594,"score_gpt":0.2594925872341089,"score_spread":0.09765471884194954,"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."}}