{"id":"W4401990434","doi":"10.1109/dsn-w60302.2024.00021","title":"Unlearning Backdoor Attacks Through Gradient-Based Model Pruning","year":2024,"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":"","funders":"Queensland University of Technology; Canadian Institute for Advanced Research","keywords":"Backdoor; Computer science; Pruning; Artificial intelligence; Computer security","routes":{"ca_aff":false,"ca_fund":true,"ca_venue":false,"about_ca":false,"invisible_to_affiliation_only":true},"retraction":null,"screen":null,"direct_labels":[],"prediction":{"model_version":"metacan-v3-hybrid-931329e0061c","candidate_categories":[],"consensus_categories":[],"category_scores_codex":[0.002255636,0.001598328,0.001417725,0.0008600606,0.0005092929,0.001188497,0.001914006,0.001471288,0.001418212],"category_scores_gemma":[0.01278308,0.0006171191,0.0008897829,0.0004268763,0.001646498,0.002625628,0.002820111,0.003279907,0.0004297401],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0007717715,"about_ca_system_score_gemma":0.00139978,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.001455485,"about_ca_topic_score_gemma":0.002232809,"domain_scores_codex":[0.9985763,0.0004429803,0.00007122155,0.0002453777,0.0004721312,0.0001920418],"domain_scores_gemma":[0.9944758,0.003527779,0.0005618801,0.0008477197,0.0004086326,0.0001781729],"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":[0.0001208683,0.00009922318,0.0009653599,0.00008516863,0.00006317026,0.0001498211,0.0000630018,0.9113208,0.004437301,0.0206993,0.001640825,0.0603552],"study_design_scores_gemma":[0.000004121137,0.00002177692,0.00004372466,0.000008645136,0.000006246366,0.00003370223,0.000004191082,0.9913974,0.001115041,0.007179014,0.0001824206,0.000003764181],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.02305901,0.0004029157,0.9734313,0.0003708186,0.0000430353,0.00006882284,0.00004601646,0.0008898398,0.001688128],"genre_scores_gemma":[0.87071,0.0004195072,0.1255031,0.000422608,0.00009306975,0.0001373941,0.0001864153,0.0001945528,0.002333371],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.002255636,"threshold_uncertainty_score":0.01192909,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.02581618714497178,"score_gpt":0.2994003539497049,"score_spread":0.2735841668047332,"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."}}