前沿论坛

题目：Changepoint Detection in Complex Models: Cross-Fitting Is Needed

报告人：邹长亮（南开大学）

邀请人：徐平峰（东北师范大学）

时间：2024年12月4日 10：00~11：00

地点：腾讯会议（#腾讯会议： 349-630-640）

报告内容简介：  Changepoint detection is commonly approached by minimizing the sum of in-sample losses to quantify the model's overall fit across distinct data segments. However, we observe that flexible modelling techniques, particularly those involving hyperparameter tuning or model selection, often lead to inaccurate changepoint estimation due to biases that distort the target of in-sample loss minimization. To mitigate this issue, we propose a novel cross-fitting methodology that incorporates out-of-sample loss evaluations using independent samples separate from those used for model fitting. This approach ensures consistent changepoint estimation, relying solely on the models' predictive accuracy across nearly homogeneous data segments. Extensive numerical experiments demonstrate that our proposed cross-fitting strategy significantly enhances the reliability and adaptability of changepoint detection in complex scenarios.

报告人简介：邹长亮，南开大学统计与数据科学学院教授、统计研究院院长。2008年博士毕业于南开大学，随后留校任教。主要从事统计学及其与数据科学领域的交叉研究和应用。研究兴趣包括：高维数据统计推断、大规模数据流分析、变点和异常点检测等，在统计学和机器学习相关领域的顶尖杂志《Annals of Statistics》《Biometrika》《Journal of the American Statistical Association》《Journal of Machine Learning Research》上发表论文二十余篇，入选爱思唯尔“中国高被引学者”。主持国家基金委优青、杰青、重点项目、重大项目课题和科技部重点研发计划课题等。任教育部科技委委员、全国应用统计专业硕士教学指导委员会委员、中国现场统计研究会副理事长等。