{"id":"W3201293175","doi":"10.2478/popets-2022-0023","title":"Disparate Vulnerability to Membership Inference Attacks","year":2021,"lang":"en","type":"preprint","venue":"Proceedings on Privacy Enhancing Technologies","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":18,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of Toronto","funders":"Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung; National Science Foundation","keywords":"Vulnerability (computing); Inference; Generalization; Computer science; Population; Differential (mechanical device); Computer security; Vulnerability assessment; Data mining; Machine learning; Artificial intelligence; Mathematics; Psychology; Engineering; Demography; Social psychology","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":["metaresearch","metaepi_narrow","scholarly_communication","open_science","research_integrity"],"consensus_categories":["metaepi_narrow","open_science","research_integrity"],"category_scores_codex":[0.001959294,0.001480492,0.001547444,0.001383429,0.0004725021,0.002718348,0.09424496,0.002096321,0.00002646273],"category_scores_gemma":[0.223572,0.001461685,0.0003387284,0.002838162,0.0004794719,0.001530395,0.4645405,0.005148049,0.0001698532],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00106896,"about_ca_system_score_gemma":0.0004925622,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00005461321,"about_ca_topic_score_gemma":0.00002559118,"domain_scores_codex":[0.9904399,0.00006760078,0.001417188,0.004748678,0.001448959,0.001877703],"domain_scores_gemma":[0.9769966,0.0007302848,0.0009784707,0.02000094,0.001056041,0.0002376491],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"theoretical_or_conceptual","study_design_scores_codex":[0.0002745706,0.00238371,0.02477493,0.007871193,0.001454515,0.0005154665,0.005838114,0.001070199,0.06162269,0.1314076,0.3536526,0.4091344],"study_design_scores_gemma":[0.0002498786,0.0003082717,0.0006667359,0.002215743,0.00004501513,0.00002465814,0.0007144081,0.01287687,0.4573042,0.5217418,0.001847045,0.002005385],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"empirical","genre_gemma":"empirical","genre_scores_codex":[0.5901886,0.0006833511,0.2462588,0.1175008,0.002053012,0.002283314,0.00005152515,0.03801433,0.00296629],"genre_scores_gemma":[0.6408367,0.0003148075,0.3575373,0.0003464447,0.00005162898,0.0007807302,0.00001828556,0.00006874169,0.00004537773],"genre_candidate":"empirical","genre_consensus":"empirical","teacher_disagreement_score":0.407129,"threshold_uncertainty_score":0.9997944,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.05369393369691378,"score_gpt":0.3268907542991967,"score_spread":0.273196820602283,"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."}}