*主持人：李韬 教授

*讲座内容简介：
In this talk, we investigate several algorithms for constrained stochastic optimization problems over time-varying random network, where the agents are to collectively minimize a sum of locally known cost functions subject to individual constraint set. Our distributed algorithms are mainly divided into two categories, dual domain and primal dual domain. On the one hand, asymptotic properties of a distributed algorithm based on dual averaging of gradients are considered. Not only almost sure (a.s.) convergence and the rate of a.s. convergence, but also asymptotic normality and asymptotic efficiency of the algorithm are obtained. On the other hand, asymptotic properties of a primal-dual distributed algorithm for stochastic optimization problem with equality and inequality constraints are investigated. A.s. consensus and a.s. convergence of the algorithm are proven under mild conditions.

*主讲人简介：
陈性敏，大连理工大学数学科学学院，副教授、硕士生导师。主要研究方向为系统辨识与控制、随机优化与分布式优化、非参数统计与统计学习。主持完成国家自然科学基金项目一项，参与国家自然科学基金项目三项。