Efficient privacy-preserving federated learning under dishonest-majority setting  

作　　者：Yinbin MIAO Da KUANG Xinghua LI Tao LENG Ximeng LIU Jianfeng MA 

机构地区：[1]School of Cyber Engineering,Xidian University,Xi’an 710071,China [2]Intelligent Policing Key Laboratory of Sichuan Province,Sichuan Police College,Luzhou 646000,China [3]Institute of Information Engineering,Chinese Academy of Sciences,Beijing 100085,China [4]School of Cyber Security,University of Chinese Academy of Sciences,Beijing 100049,China [5]Key Laboratory of Information Security of Network Systems,Fuzhou University,Fuzhou 350108,China

出　　处：《Science China(Information Sciences)》2024年第5期319-320,共2页中国科学（信息科学）（英文版）

基　　金：supported by National Natural Science Foundation of China(Grant Nos.62072361,62125205,U23A20303);Key Research and Development Program of Shaanxi(Grant No.2022GY-019);Shaanxi Fundamental Science Research Project for Mathematics and Physics(Grant No.22JSY019);Opening Project of Intelligent Policing Key Laboratory of Sichuan Province(Grant No.ZNJW2023KFMS002);Open Fund of Key Laboratory of Computing Power Network and Information Security(Grant No.2023ZD020).

摘　　要：Federated learning(FL)is an emerging distributed learning paradigm that solves the problem of isolated data by jointly learning the global model through distributed clients.However,recent studies have shown that FL may not always guarantee sufficient privacy preservation,this is mainly because the model parameters(e.g.,weights or gradients)may leak sensitive information to malicious adversaries.Thus,many privacy-preserving FL(PPFL)solutions have attracted much attention from both academic and industrial fields.However,there are still some issues to be solved.

关 键 词：PRESERVING jointly GUARANTEE 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TP309[自动化与计算机技术—控制科学与工程]