{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0008203041,0.0000772946,0.0001800238,0.000127142,0.0001167171,0.0000146621,0.0005730392,0.00002786314,0.00006426327],"category_scores_gemma":[0.0003403399,0.00008383199,0.00007861479,0.0005861294,0.00006436074,0.0002463064,0.0003180742,0.0001267271,0.00000103747],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00009802485,"about_ca_system_score_gemma":0.0001669156,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002347294,"about_ca_topic_score_gemma":0.000004932528,"domain_scores_codex":[0.9984582,0.0002287932,0.0005025316,0.0002143327,0.0004747339,0.0001214009],"domain_scores_gemma":[0.9984543,0.000263498,0.0006102546,0.0004501763,0.0001958967,0.00002589308],"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":[0.00004235792,0.0001980243,0.005219114,0.00008617512,0.00001960436,0.000007830039,0.0004951544,0.8948441,0.05370944,0.03943954,0.00009233688,0.005846385],"study_design_scores_gemma":[0.0002782861,0.00007616085,0.00110718,0.00001118574,0.00001614405,0.000004936595,0.0001538061,0.9390731,0.05884239,0.0002645933,0.00008392689,0.00008827836],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.2725926,0.00001889699,0.7268376,0.00006743983,0.0002991989,0.000086002,0.00001851418,0.00003142464,0.00004824321],"genre_scores_gemma":[0.9870034,4.076878e-7,0.01288189,0.000009681291,0.000013611,0.000003951711,0.00002507748,0.000005453415,0.00005648801],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.7144108,"threshold_uncertainty_score":0.3418569,"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."}}