{"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":"codex-gemma-dda1882f352a","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.0005877779,0.0001157839,0.0001909476,0.0001971866,0.0002178378,0.00001859915,0.000569166,0.0000955467,0.000002299466],"category_scores_gemma":[0.001115393,0.0001005958,0.00008847838,0.0005560675,0.00004971308,0.0001498641,0.00003583164,0.0002477525,0.000001147836],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00002183742,"about_ca_system_score_gemma":0.00003147515,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00001278858,"about_ca_topic_score_gemma":0.000001717767,"domain_scores_codex":[0.9990917,0.0001579284,0.0001994288,0.0002517765,0.00009412884,0.0002049913],"domain_scores_gemma":[0.9958057,0.003495512,0.00006756735,0.0005296271,0.00006100351,0.00004061085],"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":[0.000005827789,0.00001550551,0.0000499385,0.00003745665,0.00002801601,6.598966e-7,0.0006059855,0.9748384,0.005618596,0.003666882,0.00001668274,0.015116],"study_design_scores_gemma":[0.0004262817,0.00005153558,0.002779264,0.00004520805,0.00005809423,0.00001221674,0.00007062645,0.9848506,0.008170486,0.003035307,0.000322937,0.0001774431],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.01331661,0.00001794176,0.9851058,0.0003780554,0.0005685728,0.0001902165,0.00001576772,0.0004046488,0.000002418664],"genre_scores_gemma":[0.2668053,0.00001119701,0.7330029,0.00001814534,0.00001130713,0.00006936484,0.000003013313,0.00001512866,0.00006369971],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.2534887,"threshold_uncertainty_score":0.4102176,"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."}}