报告人：王少新（曲阜师范大学）

时间：2023年11月17日 16:00-

地点：理科楼LA108

摘要：Kernel logistic regression (KLR) is a powerful classification method and find its popularity in many fields. However, in many practical applications, the indefinite similarity measure or indefinite kernel function can capture the domain specific structure in the data, which often violates the Mercer’s Theorem. Considering the sparsity and the indefinite kernel function, we introduce the L1-norm regularized indefinite kernel logistic regression model, and propose a DC optimization framework based algorithm to solve our problem. The convergence analysis is also given. Extensive numerical experiments are employed to illustrate the efficiency of the proposed model.

简介：王少新，博士，曲阜师范大学统计与数据科学学院副教授，硕士研究生导师。主要研究兴趣为高维复杂数据分析、统计学习理论及应用。已在Statistics in Medicine、Computational Statistics、Linear Algebra and its Applications、Journal of Computational and Applied Mathematics等期刊发表SCI/SSCI论文10余篇。目前主持国家自然科学基金青年基金1项，山东省自然科学基金青年基金1项。

邀请人：夏小超

欢迎广大师生积极参与！