{"id":"W4285223007","doi":"10.18653/v1/2022.acl-short.59","title":"Region-dependent temperature scaling for certainty calibration and application to class-imbalanced token classification","year":2022,"lang":"en","type":"article","venue":"","topic":"Anomaly Detection Techniques and Applications","field":"Computer Science","cited_by":3,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Guelph; National Research Council Canada; Vector Institute","funders":"","keywords":"Interpretability; Certainty; Calibration; Context (archaeology); Computer science; Metric (unit); Scaling; Class (philosophy); Range (aeronautics); Algorithm; Artificial intelligence; Mathematics; Statistics; Engineering","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":[0.006414087,0.001442566,0.00159625,0.00222821,0.001285029,0.002859822,0.002984189,0.001884902,0.003004309],"category_scores_gemma":[0.04403494,0.0004760984,0.001033833,0.002526534,0.002206909,0.005259255,0.004453391,0.003694978,0.001303681],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.002252739,"about_ca_system_score_gemma":0.001601984,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00356688,"about_ca_topic_score_gemma":0.003593946,"domain_scores_codex":[0.9943457,0.00120027,0.0004138567,0.001776061,0.001930311,0.0003337812],"domain_scores_gemma":[0.986827,0.005052122,0.001793744,0.003200083,0.002727563,0.0003993871],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.0007219869,0.0002184956,0.008487977,0.0003471137,0.0002059653,0.0002250974,0.0009065362,0.3061042,0.03103394,0.04742224,0.00872336,0.595603],"study_design_scores_gemma":[0.00003766749,0.0001485029,0.002635149,0.00008203018,0.00005948753,0.0003172521,0.000157282,0.9054527,0.02547061,0.05779662,0.007753163,0.00008955439],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01468761,0.0007286489,0.9801834,0.000304691,0.0001773999,0.00009756488,0.000202509,0.002171413,0.00144682],"genre_scores_gemma":[0.4235468,0.0004609757,0.5714567,0.0003089567,0.0003363334,0.0002594468,0.0006384524,0.001096722,0.001895598],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.006414087,"threshold_uncertainty_score":0.03392136,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.01434903487237385,"score_gpt":0.2557438075222285,"score_spread":0.2413947726498546,"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."}}