{"id":"W4377293783","doi":"10.1038/s41467-023-38569-4","title":"Decentralized federated learning through proxy model sharing","year":2023,"lang":"en","type":"article","venue":"Nature Communications","topic":"Privacy-Preserving Technologies in Data","field":"Computer Science","cited_by":143,"is_retracted":false,"has_abstract":true,"ca_institutions":"Toronto Metropolitan University; Vector Institute; Carleton University; University of Waterloo","funders":"Natural Sciences and Engineering Research Council of Canada","keywords":"Computer science; Federated learning; Proxy (statistics); Differential privacy; Overhead (engineering); Data sharing; Private information retrieval; Upload; Computer security; Machine learning; Artificial intelligence; Data mining; World Wide Web","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","open_science"],"consensus_categories":["open_science"],"category_scores_codex":[0.0004208914,0.0001443767,0.0001488432,0.0001376386,0.0008559444,0.0003345368,0.0503168,0.0003228632,0.000003476964],"category_scores_gemma":[0.01253332,0.0001456513,0.00005163938,0.001868067,0.0001047626,0.0008588696,0.1300412,0.001811897,0.0001141515],"about_ca_system_candidate":false,"about_ca_system_consensus":false,"about_ca_system_score_codex":0.00008613068,"about_ca_system_score_gemma":0.00007958207,"about_ca_topic_candidate":false,"about_ca_topic_consensus":false,"about_ca_topic_score_codex":0.00002002692,"about_ca_topic_score_gemma":0.00004440137,"domain_scores_codex":[0.9986117,0.00009995075,0.0002340868,0.000421605,0.0002506798,0.0003820112],"domain_scores_gemma":[0.9848102,0.0002771981,0.0001071719,0.01462023,0.000143926,0.00004128226],"domain_codex":null,"domain_gemma":null,"domain_candidate":null,"domain_consensus":null,"study_design_codex":"not_applicable","study_design_gemma":"simulation_or_modeling","study_design_scores_codex":[0.000007128167,0.000168124,0.00448127,0.00002978661,0.0001061572,0.00001040695,0.001058471,0.00383037,0.003949667,0.4691436,0.5045486,0.01266644],"study_design_scores_gemma":[0.0001397204,0.000005941156,0.0001971435,0.00002458656,0.000003805119,0.000003199232,0.00003546959,0.8278613,0.0005019695,0.1529539,0.01813491,0.0001380324],"study_design_candidate":"simulation_or_modeling","study_design_consensus":null,"genre_codex":"methods","genre_gemma":"empirical","genre_scores_codex":[0.01047742,0.005430609,0.5976284,0.3447084,0.0005252017,0.0009461261,0.00003212046,0.02023666,0.02001517],"genre_scores_gemma":[0.5463899,0.001365342,0.4515516,0.0002899904,0.000006289918,0.00006635569,0.0001236141,0.00001577726,0.0001910881],"genre_candidate":"methods","genre_consensus":null,"teacher_disagreement_score":0.8240309,"threshold_uncertainty_score":0.9957845,"prediction_status":"machine_predicted_unvalidated"},"machine_scores":{"provisional":true,"baseline":true,"maturity_gate_passed":false,"score_opus":0.07609968010346246,"score_gpt":0.3558427230844551,"score_spread":0.2797430429809926,"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."}}