{"id":"W4319451761","doi":"10.1145/3583564","title":"Finding Deviated Behaviors of the Compressed DNN Models for Image Classifications","year":2023,"lang":"en","type":"article","venue":"ACM Transactions on Software Engineering and Methodology","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada; National Natural Science Foundation of China; Hong Kong University of Science and Technology; University of Waterloo; Cisco Systems","keywords":"Computer science; Artificial intelligence; Task (project management); Image (mathematics); Machine learning; Markov chain; Fitness function; Artificial neural network; Pattern recognition (psychology); Data mining; Genetic algorithm","routes":{"ca_aff":true,"ca_fund":true,"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":[0.001920741,0.001872991,0.0007753744,0.0008672229,0.0004735337,0.001082146,0.002042452,0.001506445,0.001778218],"category_scores_gemma":[0.01302045,0.0006530997,0.0009605799,0.0005008052,0.0008988621,0.00299729,0.001231957,0.002681914,0.0006526902],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001627481,"about_ca_system_score_gemma":0.001621623,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.01219413,"about_ca_topic_score_gemma":0.01361183,"domain_scores_codex":[0.998982,0.0002171599,0.00007367652,0.0002965774,0.000303504,0.0001269201],"domain_scores_gemma":[0.9952484,0.002854984,0.0003775793,0.0007676631,0.0005959811,0.0001555686],"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":[0.0008454837,0.0004503467,0.0164009,0.0003377381,0.0002512703,0.0006519511,0.0002847099,0.6445417,0.02904478,0.00429448,0.00778387,0.2951128],"study_design_scores_gemma":[0.00001501948,0.00005316416,0.0005731389,0.0000107483,0.00001362051,0.00004566833,0.00002218479,0.9900184,0.00758114,0.001340596,0.0003167194,0.000009547821],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.6567275,0.001578112,0.3192862,0.00151006,0.0002449499,0.0002209958,0.0009859035,0.01452264,0.004923607],"genre_scores_gemma":[0.9081795,0.0002649821,0.08761168,0.0004213331,0.00002962248,0.0001272593,0.001608888,0.0003777068,0.001378959],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.01219413,"threshold_uncertainty_score":0.02424628,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1374384112985051,"score_gpt":0.3485295355613078,"score_spread":0.2110911242628027,"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."}}