报告题目：Differentially Private Distributed Algorithms for Stochastic Aggregative Games

报告人：王继民 北京科技大学副教授

报告时间：2021年12月5日14:30-15:30

报告地点：腾讯会议ID：454-598-552   金沙集团wwW3354CC305室

报告摘要：

Due to the privacy issues caused by information exchange between the players, the design of privacy-preserving distributed algorithms for stochastic aggregative games is urgent. In this paper, we propose three differentially private distributed algorithms to seek the equilibrium in stochastic aggregative games. The input perturbation method and output perturbation method are both given, where each player protects sensitive information by adding time-varying random noises. For the case of output perturbation, the convergence of the algorithm is given by using mini-batch methods, and the mean square convergence error is reduced. For the case of input perturbation, a differentially private distributed stochastic approximation-type algorithm is proposed. Under suitable communication graph conditions, the algorithm can achieve almost sure convergence and ( ϵ, δ)-differential privacy. The convergence rate of the algorithm is also rigorously presented for the first time, where the optimal convergence rate O(1/k) in mean square sense is obtained. When the privacy budget of each iteration is very small, a lot of privacy noise is added to the algorithm, which slows down the convergence rate of the algorithm. Then, a differentially private distributed algorithm is designed by using mini-batch methods, which reduces the influence of added privacy noise on the performance of the algorithm, and improves the convergence rate of the algorithm. Specifically, when the batch sizes and the number of consensus times at each iteration grow at a suitable rate, an exponential rate of convergence can be achieved with the same level of privacy.

报告人简介：

王继民，北京科技大学自动化学院副教授。2018年博士毕业于山东大学数学学院，随后在中国科学院数学与系统科学研究院从事博士后研究（2018-2020），2020年12月加入北京科技大学自动化学院任现职。研究方向为控制系统的隐私保护与安全，随机系统和网络化控制系统等，发表控制理论方向权威期刊Automatica、IEEE Transactions on Cybernatics、SCIENCE CHINA Information Sciences等学术论文16篇，他引100余次，申请中国发明专利两项。2017年5月至2018年5月在澳大利亚纽卡斯尔大学访问，曾获得山东大学优秀博士论文（2019）、许国志博士后奖励基金（2018）和Asian Journal of Control优秀审稿人（2020）等荣誉。主持科技部外专项目、中国博士后基金和中央高校基本科研业务费等3项，参与国家重点研发计划和国家自然科学基金等多项。