报告题目：An ensemble method for multiple change-point detection in moderately high-dimensional data

报告人：Prof Yuehua Wu

邀请人：黄磊

讲座时间：2026年5月28日（星期四）上午9:00

讲座地点： 犀浦校区三号教学楼 X30423

报告摘要：Change-point detection (CPD) has been studied extensively due to its wide range of applications across various fields. However, CPD remains a challenging task for complex data characterized by high dimensionality, correlations, outliers, or heavy-tailed distributions. In this talk, we present an integrated change-point detection method called PCA-uCPD, which utilizes principal component analysis (PCA) to project the original data series into uncorrelated principal components (PCs). We then apply existing univariate change-point detection methods to these PCs, followed by a refining technique to obtain the final change-point estimates for the original sequences. This method features a flexible architecture capable of handling complex data. We provide theoretical justifications to guarantee the feasibility of the method under specific conditions and conduct simulations to assess its performance across various data-generating scenarios. Finally, the efficacy of PCA-uCPD is demonstrated through applications to both genetic and financial datasets.

主讲人简介：吴月华，加拿大约克大学数学与统计系教授。她师从世界著名统计学家C. R. Rao，于1989年获得美国匹兹堡大学统计学博士学位。目前，她从事高维数据分析、模型选择、变点分析、时空建模、环境统计、统计金融和贝叶斯方法等多领域研究， 是国际统计学会的当选会员。她在Proceeding of National Academy Science, USA, Biometrika, Journal of Economics等期刊上发表了150余篇学术论文，也一直承担加拿大国家自然科学基金科研项目。另外，她目前是Entropy Section Board Member和Entropy特刊“用于对高维和复杂数据进行建模的统计方法：第三期”的客座编辑，也是Springer Nature 系列丛书“数学奇迹：本着 CR Rao 精神的文本和专著”编辑委员会副主编。