时 间：2025年10月17日 10:00 - 11:00
地 点：普陀校区理科大楼A1514室
报告人：冯龙南开大学教授
主持人：刘玉坤华东师范大学教授
摘 要：
High-dimensional data analysis has garnered significant attention in recent years. However, many existing methodologies rely heavily on the assumption of Gaussian or sub-Gaussian distributions, which can lead to sub-optimal performance when the underlying data exhibit heavy tails. To address this limitation, we adopt a robust framework based on spatial-sign methods under the assumption of elliptically symmetric distributions. This approach offers enhanced robustness and efficiency in heavy-tailed scenarios. In this talk, I will present the application of spatial-sign methods to a variety of high-dimensional statistical problems, including hypothesis testing, principal component analysis, linear and quadratic discriminant analysis, graphical models, and canonical correlation analysis. The proposed methods provide a unified and robust toolkit for high-dimensional inference beyond traditional distributional assumptions.
报告人简介：
冯龙现任南开大学统计与数据科学学院教授、博士生导师。入选教育部青年人才计划、南开大学百名青年学科带头人。曾获得教育部学术新人奖，南开大学优秀博士论文奖。主要从事高维数据分析方面的研究，在统计学国际顶尖杂志JRSSB, JASA、Biometrika、Annals of Statistics、JOE、JBES等发表50余篇论文。主持一项天津市杰出青年基金、国家自然科学基金面上项目和青年项目。担任Statistical Theory and Related Field副主编，中国现场统计研究会贝叶斯统计分会副理事长。