English  |   学校主页  |   收藏本站



【主 题】 Powerful Graph-Based Change-Point Tests for High-Dimensional Data

【报告人】 Yuehua Amy Wu , 教授

             York University

【时 间】 2019年3月1日(星期五)10:00-11:00

【地 点】 上海财经大学统计与管理学院大楼1208会议室

摘 要】Change-point detection in high-dimensional time series is central to many applications in science and engineering, including neuroscience, signal processing, evolving network analysis, image analysis, and text analysis. In this talk, we present a change-point test for high-dimensional data where we estimate the change-point via a ratio cut. The change-point test makes use of a Bayesian-type statistic based on the shortest Hamiltonian path. We show that the change-point test is consistent. In addition, we provide a probabilistic error bound on the change-point estimate. We demonstrate through simulation studies that the test is powerful against alternatives that consider a shift in mean or variance and is able to accurately capture the location of the change-point if it exists. A real data example using cell division data is provided. We will also briefly introduce other graph-based change-point tests in this talk.

Joint work with X. Shi and C. R. Rao

嘉宾简介】Dr. Yuehua Amy Wu is the Professor of Statistics, York University, Canada. Her research interesting are M-estimation, Multiple change-point analysis, and High-dimensional statistics. et al.