报告题目：ADMM for Separable Convex and Nonconvex Optimization

报告人： 白建超  副教授

报告时间：2019年11月29日（周五）8:30—10:00

报告地点：威廉希尔学术报告厅

报告摘要: In this talk, we show our recent work on a stochastic Alternating Direction Method of Multipliers (ADMM) for solving the two-block separable convex optimization problem subject to linearly equality constraints as well as a deterministic ADMM for solving the separable but nonconvex case. By the aid of variational analysis and by constructing a potential function, preliminary convergence results of these two methods are presented. We exam the numerical performance of our methods for solving the graph guided fused lasso model in big-data applications and the large-scale signal recovery problem. Several further questions are discussed finally.

报告人简介: 白建超，西北工业大学副教授，先后师从桂林电子科技大学段雪峰教授、西安交通大学李继成教授和Louisiana State University张洪超教授。2019年3月起在西北工业大学应用数学系工作，主要从事非线性最优化方法的理论与应用研究，近期在设计大数据处理中的非凸随机优化算法。现以第一作者或通讯作者在Computational Optimization and Applications、Optimization Letters、Journal of Computational and Applied Mathematics、Applied Numerical Mathematics等SCI期刊发表论文12篇，主持一项中央高校基本科研业务费项目。

   欢迎广大师生参加！