2021年秋季先进机器人与人工智能系列学术讲座(第184期)


我校机器人与信息自动化研究所 天津市智能机器人技术重点实验室

Institute of Robotics and Automatic Information System

Tianjin Key Laboratory of Intelligent Robotics

2021年秋季先进机器人与人工智能系列学术讲座(第184期)

Seminar Series:Advanced Robotics & Artificial Intelligence

报告时间:2021年10月20日(周三),上午10:00-11:00

报告地点:皇冠体育官网南楼105室

报告题目:Differentially Private Distributed Optimization over Networks

报告人:朱善迎 研究员

研究领域: 网络系统的分布式估计和优化、多智能体协同控制、微能源网的能量管理

单位:上海交通大学

报告摘要:

Distributed optimization has received much attention recently due to its wide applications in sensor fusion, resource allocation, and machine learning. Common features of these examples are that there is no centralized center involved and the resources, such as sensing, communication, and computation, are usually scattered throughout the network, which necessitate completely distributed algorithms. However, the vulnerability nature of wireless communications makes the messages transmitted between agents at risk of being intercepted by attackers, which will cause issues of information disclosure. In this talk, I will introduce a distributed method that can ensure convergence but is still guaranteed to achieve differential privacy. A tradeoff between differential privacy and convergence accuracy is characterized. Application to economic dispatch problems in smart grids will also be discussed.


报告人简介:

朱善迎,上海交通大学自动化系研究员,博士生导师。博士毕业于上海交通大学自动化系。2013年至2015年,在新加坡南洋理工大学以及伯克利教育联盟(BEARS)开展博士后研究工作。主要研究领域为网络系统的分布式估计和优化、多智能体协同控制、微能源网的能量管理等。主持国家自然科学基金优青/面上项目、国家重点研发计划课题等9项,参与国家自然科学基金重大/重点项目等8项。发表论文70余篇,合作出版英文专著一部。曾担任多个国际会议的TPC/IPC成员、Invited Session/Local Arrangement/Publicity/Track Co-Chair等。现为IEEE工业信息学技术委员会委员,美国《数学评论》评论员,中国自动化学会青工委委员以及TCCT 多自主体控制学组委员等。





Baidu
sogou