{"id":"W4383468997","doi":"10.48550/arxiv.2307.01610","title":"Overconfidence is a Dangerous Thing: Mitigating Membership Inference Attacks by Enforcing Less Confident Prediction","year":2023,"lang":"en","type":"preprint","venue":"arXiv (Cornell University)","topic":"Adversarial Robustness in Machine Learning","field":"Computer Science","cited_by":1,"is_retracted":false,"has_abstract":true,"ca_institutions":"","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Overconfidence effect; Computer science; HAMP; Exploit; Inference; Machine learning; Training set; Artificial intelligence; Benchmark (surveying); Entropy (arrow of time); Data mining; 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":"codex-gemma-dda1882f352a","candidate_categories":["metaepi_narrow"],"consensus_categories":[],"category_scores_codex":[0.001005878,0.0006437661,0.0006153064,0.0003827818,0.0005691325,0.0005885874,0.003401723,0.0006383595,0.00005995632],"category_scores_gemma":[0.0005635464,0.0008154045,0.0002889342,0.0009775924,0.0002216753,0.001349051,0.004533316,0.002224854,0.0001262535],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0006312129,"about_ca_system_score_gemma":0.0004210579,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.002747586,"about_ca_topic_score_gemma":0.0001060234,"domain_scores_codex":[0.9956389,0.0004155039,0.0005531429,0.002163051,0.0004388291,0.0007905554],"domain_scores_gemma":[0.9959108,0.0008632627,0.0008534049,0.001752141,0.0003124383,0.0003079167],"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":[0.00002404653,0.00004300294,0.01967631,0.0003278562,0.0001921097,0.0003876202,0.005249956,0.9045707,0.0001939092,0.06641015,0.001869892,0.001054432],"study_design_scores_gemma":[0.0005257606,0.0000539401,0.0007122492,0.0009036508,0.0001071911,0.000009954203,0.001115063,0.9770632,0.0004524187,0.01791286,0.0003091101,0.0008346145],"study_design_candidate":"simulation_or_modeling","study_design_consensus":"simulation_or_modeling","genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.1083541,0.00006935718,0.8861955,0.0003072129,0.00174494,0.0004608793,0.00004781295,0.001213713,0.001606551],"genre_scores_gemma":[0.9929614,0.0001450324,0.003642234,0.0002664739,0.0001988648,0.000005247426,0.00005759248,0.00006171048,0.002661465],"genre_candidate":"empirical","genre_consensus":null,"teacher_disagreement_score":0.8846073,"threshold_uncertainty_score":0.9994297,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.1314094578956852,"score_gpt":0.2432273903084682,"score_spread":0.111817932412783,"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."}}