时 间：2025年4月21日（周一）10：30 – 11：30
地 点：理科大楼A1514室
报告人：张晨 清华大学副教授
主持人：唐炎林 华东师范大学教授
摘 要：
This talk explores machine learning to address a problem of Partially Observable Online Change Detection (POOCD), where only a subset of variables can be observed to online monitor a complex process for change-point detection. POOCD is challenging because the learner not only needs to detect on-the-fly whether a change occurs based on partially observed historical data, but also needs to cleverly choose a subset of informative variables to observe in the next learning round, to maximize the overall sequential change detection performance. In this talk, we present a series of methods for POOCD, targeting on different data types such as high-dimensional vector data, functional data, dynamic network data, and image data. Their applications to real case datasets in manufacturing, meteorology and social science are also discussed.
报告人简介：
Chen Zhang is an Associate Professor in Industrial Engineering, Tsinghua University. She received her B.Eng. degree in Electronic Science and Technology from Tianjin University in 2012, and her Ph.D. degree in Industrial Systems Engineering from National University of Singapore in 2017. Her research interests include developing statistical and machine learning methods for complex data modeling and monitoring. Her research has been published in IISE Transactions, Journal of Quality Technology, Technometrics, IEEE TASE etc. She has received several best paper awards from ASQ, IISE, and INFORMS.