报告题目：
Privacy-Preserving Distributed ADMM WithEvent-Triggered Communication
报告摘要：
Distributed optimization is a hot issue in current research, which is widely used in largescalemachine learning, smart grid, wireless sensor networks and other fields. Communicationbetween the agents is required in the distributed optimization algorithms, which on the onehand will bring about the privacy leakage problem and on the other hand, the communication inlarge scale network always suffers from excessive energy consumption, limited communicationbandwidth and some other problems. Therefore, the privacy preserving and communication efficiencyhave become important indicators to measure the performance of the algorithm. However,the existing privacy-preserving algorithms often face the problems of highcommunicationcost and high computation burden in a single iteration. We propose a communication-efficient and privacy preserving algorithm termed as PC-DQM. In PC-DQM, an event-triggered mechanism is designed to schedule the communication instants for reducing communication cost. Simultaneously, for privacy preservation, a Hessian matrix with perturbed noise is introduced to quadratically approximate the objective function, which results in a closed form of primal vector update and then avoids solving a subproblem at each iteration with possible high computation cost. We theoretically show that PC-DQM can protect privacy but without losing accuracy. In addition, we rigorously prove that PC-DQM converges linearly to the exact optimal solution for strongly convex and smooth objective functions. Finally, numerical simulation is presented to illustrate the effectiveness and efficiency of our algorithm.
个人介绍：
张振，东南大学计算机学院在读二年级博士生。研究方向为分布式机器学习算法以及算法安全。