{"id":"W4312292505","doi":"10.1145/3522664.3528617","title":"Identification of out-of-distribution cases of CNN using class-based surprise adequacy","year":2022,"lang":"en","type":"article","venue":"","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Polytechnique Montréal","funders":"","keywords":"Surprise; Computer science; Identification (biology); Class (philosophy); Artificial intelligence; Calibration; Pattern recognition (psychology); Distribution (mathematics); Machine learning; Mathematics; Statistics","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":[0.003894232,0.0008397208,0.0007962063,0.001157701,0.0005810789,0.001588778,0.00128589,0.001684555,0.002021568],"category_scores_gemma":[0.02925762,0.0004623554,0.0005878144,0.0004149881,0.002021947,0.002401457,0.002211967,0.0020094,0.0005994504],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001159135,"about_ca_system_score_gemma":0.0007085017,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001476701,"about_ca_topic_score_gemma":0.001667095,"domain_scores_codex":[0.9975874,0.0004616608,0.000135908,0.0006739278,0.0008273945,0.0003136903],"domain_scores_gemma":[0.9884189,0.005555223,0.001465934,0.00274869,0.001451081,0.0003602305],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"not_applicable","study_design_scores_codex":[0.001962883,0.0003324819,0.2205445,0.0003994602,0.0004135985,0.006411241,0.001195298,0.298576,0.08304105,0.06302679,0.01888541,0.3052113],"study_design_scores_gemma":[0.0000217697,0.00008907531,0.01849383,0.00006295692,0.00004369549,0.001950377,0.0001776656,0.9088392,0.03245079,0.03534365,0.002487203,0.00003983865],"study_design_candidate":"not_applicable","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5861405,0.0004400203,0.4003001,0.001468497,0.0002072602,0.0001060841,0.0004197424,0.002545035,0.008372793],"genre_scores_gemma":[0.9882264,0.00004381139,0.01053252,0.0001828383,0.00002473227,0.00002145468,0.0002309637,0.0001344405,0.0006029138],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.003894232,"threshold_uncertainty_score":0.02059495,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.03799936767009324,"score_gpt":0.3108427062098721,"score_spread":0.2728433385397789,"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."}}