best365网页版登录学术报告

Deterministic and Stochastic ADMM for Structured Convex Optimization

白建超 副教授

（西北工业大学数学与统计学院）

报告时间：2021年4月21日下午4:00-5:00

报告地点：沙河校区J1-110

报告摘要：In this talk, our recent researches on a deterministic alternating direction method of multipliers (ADMM) and two stochastic ADMMs are presented for solving the muti-block structured convex optimization problems whose objective functions are possibly nonsmooth. The main convergence results of these methods are showed concisely. Performance of the methods is verified by testing the latent variable Gaussian graphical model selection problem in statistical learning and the graph-guided fused lasso problem in supervised learning, respectively. Finally, several further questions are shared and discussed.

报告人简介：白建超，博士(后)，西北工业大学数学与统计学院副教授，硕士生导师。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路易斯安那州立大学张洪超教授。主要从事机器学习和统计学习等领域的大规模优化方法、理论与应用研究。在Computational Optimization and Applications、IEEE Transactions on Medical Imaging等期刊上发表SCI论文20余篇。现主持国家自然科学基金项目、中国博士后科学基金项目、中央高校基本科研业务费项目各1项。

邀请人： 谢家新