{"id":"W4294335582","doi":"10.1109/satml54575.2023.00012","title":"Exploiting Fairness to Enhance Sensitive Attributes Reconstruction","year":2023,"lang":"en","type":"preprint","venue":"","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":9,"is_retracted":false,"has_abstract":true,"ca_institutions":"École de Technologie Supérieure","funders":"Agence Nationale de la Recherche","keywords":"Adversary; Computer science; Set (abstract data type); Black box; Range (aeronautics); Machine learning; Artificial intelligence; Training set; Adversary model; Computer security","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","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0008130134,0.0003795981,0.0004425347,0.0004414851,0.0001641546,0.000499877,0.02865899,0.0004419668,0.00001133641],"category_scores_gemma":[0.02873841,0.0003981309,0.0001092899,0.0008618434,0.00009009837,0.0006084788,0.3802494,0.0009281877,0.0006448895],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002785639,"about_ca_system_score_gemma":0.0001413711,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.0002543987,"about_ca_topic_score_gemma":0.0000668059,"domain_scores_codex":[0.9965873,0.000111701,0.0004931759,0.001774837,0.0004369952,0.0005959821],"domain_scores_gemma":[0.9864818,0.000460771,0.000265204,0.01237333,0.0003088765,0.0001100575],"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":[0.00001011026,0.00004539928,0.0006145252,0.0002698806,0.0001757164,0.0001904278,0.0004494016,0.0006814261,0.003674484,0.008545795,0.4027438,0.582599],"study_design_scores_gemma":[0.00006838891,0.00004141148,0.0006952561,0.0009052539,0.0000111658,0.00006371696,0.0003909498,0.114107,0.2502384,0.6320556,0.0004287051,0.000994108],"study_design_candidate":"theoretical_or_conceptual","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"methods","genre_scores_codex":[0.02050879,0.00001979755,0.9166591,0.05106496,0.003233646,0.0004709568,0.00008320266,0.007100723,0.0008588131],"genre_scores_gemma":[0.1292343,0.00007542118,0.8695709,0.0002522687,0.0001851998,0.0001754983,0.00005331029,0.00003918482,0.0004139671],"genre_candidate":"methods","genre_consensus":"methods","teacher_disagreement_score":0.6235098,"threshold_uncertainty_score":0.9998471,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07078224612534335,"score_gpt":0.3186410416197341,"score_spread":0.2478587954943908,"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."}}