报告题目:Compressive Sensing for Graph Clustering
报告人:Ming-Jun Lai（美国佐治亚大学数学系）
时间：2019年12月12日15:00-16:00
地点：沙河主楼E404
摘要: This talk is based on a joint work with Daneil Mckenzie. We will explainhow to phrase the cut improvement for graphs as a sparse recovery problem,whence one can use algorithms originally developed for compressivesensing (such as SubspacePursuit to solve it. We show that thisapproach to cut improvement is fast, both in theory and practice and moreoverenjoys statistical guarantees of success when applied to graphs drawn fromprobabilistic models such as the Stochastic Block Model. We then proposenew methods for local clustering and semi-supervised clustering, which enjoysimilar guarantees of success and speed. Finally, we demonstrate ourapproach with extensive numerical benchmarking.
报告人简介：来明骏博士是美国佐治亚大学数学系的终身教授，研究领域有数值分析和逼近论，发表了100多篇期刊论文和大约33篇会议论文和一本多元样条函数的专箸。其中在“SIAM”杂志上有13篇关于数值分析、优化、图像科学、数学分析的论文，获得五次美国国家科学基金会的科研资助，每次3至4年的经费。应邀在40多所国际著名大学给学术讲座，包括哈佛大学、剑桥大学（英格兰）、范德比尔特大学、密歇根大学, UCLA大学等等。他目前是国际著名杂志“Appliedand Computational Harmonic Analysis”和“Journal of Numerical Mathematics”的副主编。他和80位学者学生有合作论文。他指导出19位博士。还在指导3位博士学生。
邀请人：谢家新
