联邦学习与攻防对抗综述  

Review of Federal Learning and Offensive-Defensive Confrontation

作　　者：杨丽 朱凌波 于越明 苗银宾 YANG Li;ZHU Lingbo;YU Yueming;MIAO Yinbin(School of Cyber Engineering,Xidian University,Xi’an 710126,China)

机构地区：[1]西安电子科技大学网络与信息安全学院,西安710126

出　　处：《信息网络安全》2023年第12期69-90,共22页Netinfo Security

基　　金：国家自然科学基金[62072361];陕西省重点研发计划[2022GY-019];陕西省数理基础科学研究项目[22JSY019]。

摘　　要：随着机器学习技术的不断发展,个人隐私问题被广泛重视。由于用户数据被发送至中心节点导致集中学习受到相当程度的制约,所以联邦学习作为一个数据不出本地便可以完成模型训练的框架应运而生。但联邦学习机制依旧会受到各种攻击的影响而导致安全性和隐私性降低。文章先从联邦学习的基本定义入手,再对机密性和完整性两个方面进行重点分析、总结联邦学习中的威胁和防御手段,最后结合这些问题来讨论该领域在未来的发展方向。With the continuous development of machine learning technology,personal privacy issues have attracted widespread attention.Centralized learning is subject to a considerable degree of constraints due to the fact that user data is sent to the central node.Therefore,federal learning as a data can be completed locally.The framework of model training came into being.However,the federated learning mechanism will still be affected by various attacks and reduce the security and privacy.This paper started with the basic definition of federal learning,and then analyzed and summarized the threats and defense means in federal learning from two aspects of confidentiality and integrity.Finally,through these problems,the future development direction of this field was discussed.

关 键 词：联邦学习 机密性 完整性 防御手段 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]