报告题目：A Class of Line Search Type Methods for Nonsmooth Convex Regularized Minimization

主 讲 人：周 伟 军

单 位：长沙理工大学

时 间：11月25日15:00

腾 讯 ID：948 791 387

摘 要:

This paper presents a class of line search type methods for solving the regularized optimization model whose objective function is the sum of a smooth function and a nonsmooth convex regularized term. This problem has many applications such as in compressive sensing and sparse reconstruction. Three special cases of this class of methods are proposed and their convergence theorems are established. They are generalizations of some existing BB gradient methods and PRP type nonlinear conjugate gradient methods for smooth unconstrained optimization problems. The proposed methods are applied to sparse reconstruction and some numerical results are reported to show their efficiency.

简 介：

周伟军，长沙理工大学教授。2000、2003和2006年在湖南大学分别获学术、硕士和博士学位，2007年在日本国立弘前大学访问一年，2008-2010年在香港理工大学进行博士后工作。主要从事数值优化研究，在Math. Comput.、SIAM J. Optim.等期刊发表论文20余篇，主持完成国家自科项目2项，获省自然科学奖二等奖1项。