{"id":"W4404663741","doi":"10.1613/jair.1.15985","title":"Differentially Private Neural Tangent Kernels (DP-NTK) for Privacy-Preserving Data Generation","year":2024,"lang":"en","type":"article","venue":"Journal of Artificial Intelligence Research","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":2,"is_retracted":false,"has_abstract":true,"ca_institutions":"Vector Institute; University of British Columbia","funders":"Novo Nordisk; Natural Sciences and Engineering Research Council of Canada; Mitacs; Canadian Institute for Advanced Research","keywords":"Computer science; Generator (circuit theory); Metric (unit); Benchmark (surveying); Synthetic data; Tangent; Artificial intelligence; Differential privacy; Data mining; Quality (philosophy); Machine learning; Pattern recognition (psychology); Mathematics; 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":"codex-gemma-dda1882f352a","candidate_categories":["metaresearch","scholarly_communication","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.007860595,0.0002586914,0.000379774,0.001021443,0.0004045192,0.002397303,0.05224336,0.0002042562,0.00007426927],"category_scores_gemma":[0.04254904,0.0002151317,0.0001770998,0.001570684,0.0002663446,0.003208365,0.07342359,0.001352069,0.00006529779],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.0002851143,"about_ca_system_score_gemma":0.0005192416,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00003603883,"about_ca_topic_score_gemma":0.00004133441,"domain_scores_codex":[0.9945415,0.0004203944,0.001293421,0.0009421547,0.001849493,0.0009529961],"domain_scores_gemma":[0.9883217,0.001397027,0.0002900488,0.008666562,0.001078124,0.0002465092],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"design_other","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.00008543671,0.0002288509,0.00004846668,0.0002028365,0.0001672766,0.0002339856,0.0004017464,0.0003806452,0.1031988,0.03370816,0.1534978,0.707846],"study_design_scores_gemma":[0.00002810647,0.0002706943,0.00001891927,0.0001454901,0.00001244578,0.00005454151,0.00006206986,0.6721396,0.07245838,0.2482651,0.006388881,0.0001557204],"study_design_candidate":"design_other","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.05017557,0.001987857,0.9073972,0.03724366,0.002345261,0.0005511678,0.00005587794,0.0001986902,0.00004472392],"genre_scores_gemma":[0.7274624,0.0009088889,0.2694826,0.00005478163,0.001897958,0.00003058371,0.00003818924,0.00005081757,0.00007377166],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.7076902,"threshold_uncertainty_score":0.9986383,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.3973325884084702,"score_gpt":0.4591663783214747,"score_spread":0.06183378991300453,"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."}}