A Privacy-Preserving Scheme for Multi-Party Vertical Federated Learning  

作　　者：FAN Mochan ZHANG Zhipeng LI Difei ZHANG Qiming YAO Haidong 

机构地区：[1]School of Information&Communication Engineering,University of Electronic Science and Technology of China,Chengdu 611731,China [2]ZTE Corporation,Shenzhen 518057,China [3]State Key Laboratory of Mobile Network and Mobile Multimedia Technology,Shenzhen 518055,China

出　　处：《ZTE Communications》2024年第4期89-96,共8页中兴通讯技术（英文版）

基　　金：supported in part by ZTE Industry-University-Institute Cooperation Funds under Grant No. 202211FKY00112;Open Research Projects of Zhejiang Lab under Grant No. 2022QA0AB02;Natural Science Foundation of Sichuan Province under Grant No. 2022NSFSC0913

摘　　要：As an important branch of federated learning,vertical federated learning(VFL)enables multiple institutions to train on the same user samples,bringing considerable industry benefits.However,VFL needs to exchange user features among multiple institutions,which raises concerns about privacy leakage.Moreover,existing multi-party VFL privacy-preserving schemes suffer from issues such as poor reli-ability and high communication overhead.To address these issues,we propose a privacy protection scheme for four institutional VFLs,named FVFL.A hierarchical framework is first introduced to support federated training among four institutions.We also design a verifiable repli-cated secret sharing(RSS)protocol(32)-sharing and combine it with homomorphic encryption to ensure the reliability of FVFL while ensuring the privacy of features and intermediate results of the four institutions.Our theoretical analysis proves the reliability and security of the pro-posed FVFL.Extended experiments verify that the proposed scheme achieves excellent performance with a low communication overhead.

关 键 词：vertical federated learning privacy protection replicated secret sharing 

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]