报告题目：Proximity-gradient-subgradient Algorithms for a Class of Nonsmooth and Nonconvex Fractional Optimization Problems

报告人：李洽副教授，中山大学

报告时间：2023年11月24日（周五）14:30-15:30

报告地点：腾讯会议：284-281-668

报告摘要：In this talk, we consider a class of nonsmooth and nonconvex fractional optimization problems. We propose proximity-gradient-subgradient algorithms (PGSA) for solving these problems. Moreover, the nonmonotone linesearch scheme and extrapolation technique are incorporated into PGSA, respectively. We show the subsequential convergence of the proposed algorithms under mild conditions, while the global sequential convergence of them are analyzed by assuming the KL property of some auxiliary functions. In particular, we show that PGSA for the sparse generalized eigenvalue problem (SGEP) exhibits a linear convergence rate. Finally, numerical experiments on SGEP, L1/L2 and L1/SK sparse recovery are conducted to demonstrate the efficiency of the proposed algorithms.

报告人简介：李洽，中山大学数据科学与计算机学院副教授，博士生导师，现任中山大学计算机学院数据科学系副主任，广东省计算数学学会常务理事，广东省计算科学重点实验室成员。曾在《SIAM Journal on Optimization》、《Applied and Computational Harmonic Analysis》、《Inverse Problems》等知名国际学术刊物发表十多篇论文，其中，发表在《Inverse Problems》上的论文被该杂志评为2017年“Highlight”。主持3项国家级科研项目，研究方向主要包括最优化理论与算法及在机器学习、大数据分析、图像处理等领域中的应用。

邀请人：向道红