{"id":"W4382317858","doi":"10.1609/aaai.v37i8.26150","title":"DisGUIDE: Disagreement-Guided Data-Free Model Extraction","year":2023,"lang":"en","type":"article","venue":"Proceedings of the AAAI Conference on Artificial Intelligence","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":19,"is_retracted":false,"has_abstract":true,"ca_institutions":"York University","funders":"University of Waterloo; Purdue University; National Science Foundation","keywords":"Computer science; Generator (circuit theory); Data mining; Machine learning; Data extraction; Scheme (mathematics); Artificial intelligence; Power (physics)","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":[0.004237098,0.001538763,0.001556282,0.0009120248,0.0008346073,0.00153209,0.003226745,0.002260797,0.002598885],"category_scores_gemma":[0.01801972,0.0007238963,0.001220517,0.0007327554,0.002108631,0.005979931,0.006597233,0.003989219,0.002170681],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.001075177,"about_ca_system_score_gemma":0.001437646,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001514139,"about_ca_topic_score_gemma":0.001842738,"domain_scores_codex":[0.9961174,0.00147593,0.0002141633,0.0007099751,0.001208317,0.0002741595],"domain_scores_gemma":[0.9910782,0.003121204,0.0005071017,0.004454737,0.0006394839,0.0001992739],"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":[0.001365578,0.0005255935,0.007789437,0.000321503,0.0004279864,0.0007446003,0.0007671892,0.4440579,0.03698174,0.04685837,0.03270234,0.4274577],"study_design_scores_gemma":[0.00004954473,0.0001026879,0.0002097025,0.00001132537,0.00001678566,0.000196753,0.0000419223,0.9614763,0.01458965,0.02041041,0.002874415,0.00002043231],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.04628615,0.0004504414,0.9371069,0.0009602673,0.0001294133,0.0001654214,0.0003999056,0.01224896,0.002252445],"genre_scores_gemma":[0.7090938,0.0002160613,0.2808828,0.001200257,0.000132895,0.0002593341,0.002205244,0.001189493,0.004820131],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.004237098,"threshold_uncertainty_score":0.02240819,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.2242784790751523,"score_gpt":0.3819567872458058,"score_spread":0.1576783081706535,"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."}}