报告题目:Local search approximation algorithms for the spherical $k$-means problem
报告人:张冬梅
报告时间:2019年5月11日下午17:00
报告地点:数计学院4号楼229会议室
报告摘要:
In this talk, we study the spherical $k$-means problem (SKMP) which is one of the most well-studied clustering problems. In the SKMP, we are given an $n$- client set $\\mathcal{D}$ in $d$-dimensional unit sphere $\\mathbb{S}^d$, and an integer $k \\le n$. The goal is to open a center subset $F \\subset \\mathbb{S}^d$ with $ |F| \\le k$ that minimizes the sum of cosine dissimilarity measure for each client in $ \\mathcal{D} $ to the nearest open center. We give a $(2 (4+\\sqrt{7}) + \\varepsilon)$-approximation algorithm for this problem using local search scheme. (Joint work with Yukun Cheng, Min Li, Yishui Wang, and Dachuan Xu)
报告人简介
张冬梅,山东大学计算机应用技术专业博士,现为山东建筑大学计算机科学与技术学院教授。研究方向为组合优化、机器学习、数据挖掘、信息检索等。承担国家自然科学基金、山东省自然科学基金、山东省高校科技计划项目、山东省信息产业厅、济南市科技局等项目10余项。曾获得山东省科学技术进步奖三等奖、山东省计算机应用优秀成果奖二等奖等。在Journal of Global Optimization、Journal of Combinatorial Optimization、Optimization Letters、Asia-Pacific Journal of Operational Research、Journal of Industrial and Management Optimization、运筹学学报等发表论文四十余篇。主编和参编《操作系统》,《计算机引论》等教材四部。