报告名称：A Model-Based Monitoring Framework for Tensor Count Data in Passenger Flow Surveillance
报告人：吴纯杰 教授 上海财经大学
报告时间：2025年11月2日 上午8:30--9:30
报告地点：伍卓群楼第2报告厅
校内联系人：李聪 li_cong@jlu.edu.cn
报告摘要：
The increasing prevalence of tensor count data across various applications, including passenger flow data in urban rail transit systems, necessitates the development of advanced monitoring techniques. Traditional online monitoring methods often fail to accommodate the unique characteristics of count data or are designed exclusively for vectorized data, making them unsuitable for general-order tensor count processes. Our proposed method addresses these limitations by introducing a new Tensor Poisson Log-Normal (TPLN) model. To address the estimation difficulties arising from the multi-dimensional latent variables in the TPLN model, we developed an efficient variational Gaussian approximation (VGA) approach for Phase I modeling. In Phase II surveillance, we formulated an online parameter estimation algorithm based on the Laplace approximation method to meet the real-time computation requirements. Additionally, we designed an exponentially weighted likelihoodbased monitoring statistic to identify anomalies in online monitoring. The effectiveness and superiority of our method is validated through comprehensive simulations and an application to real-time passenger flow surveillance in the Hong Kong Mass Transit Railway (MTR).
个人简介：
吴纯杰，南开大学统计学博士、上海财经大学统计与数据科学学院讲席教授、教务处副处长、博士生导师，宝钢优秀教师奖获得者。中国现场统计研究会可靠性工程分会常务理事、大数据统计分会常务理事和上海统计学会理事等。主要研究领域为应用统计和政府统计，在JASA、AoAS、Tech、IISE和《中国科学:数学》等期刊发表高质量论文50多篇，主持国家自然科学基金项目3项和国家统计局重大项目1项和上海自然科学基金项目1项；国家级一流本科课程和上海市精品课程、课程思政示范课程、示范教学团队《数理统计》负责人，国家级一流统计学专业建设点负责人。
