《科学大讲堂 第249期》Prof. Yingying Fan：Model-X knockoffs框架及其稳健性

以下内容根据公开信息整理，并经大模型处理生成，可能存在疏漏或误差，请以实际信息为准。

  • 题目: Model-X knockoffs框架及其稳健性
  • 主讲人：Prof. Yingying Fan @ 南加州大学
  • 时间：2026年6月26日 11:00-12:00
  • 地点：理学院一楼1142报告厅

主讲人简介

Yingying Fan currently serves as Associate Dean of the PhD Program and Centennial Chair Professor at the University of Southern California (USC) Marshall School of Business, and Professor in the Department of Data Science and Operations Management. Her research interests cover statistics, data science, machine learning, economics, big data and business applications. She received the honor of the Medallion Lecture awarded by the Institute of Mathematical Statistics (IMS), and is an elected Fellow of both IMS and the American Statistical Association (ASA). She has also earned a host of prestigious academic awards, including the Guy Bronze Medal of the Royal Statistical Society (RSS), the ASA Noether Young Scholar Award and the National Science Foundation (NSF) CAREER Award.

Professor Fan has authored over 100 papers in top international journals and conferences across statistics, econometrics, machine learning and interdisciplinary fields. Her publications appear in leading academic venues: core statistics journals The Annals of Statistics (AOS), Journal of the American Statistical Association (JASA), Journal of the Royal Statistical Society, Series B (JRSSB) and Biometrika; top econometrics journals Journal of Econometrics and Journal of Financial Econometrics; the leading machine learning journal Journal of Machine Learning Research; the top computer science conference NeurIPS; as well as the multidisciplinary journal PNAS. For academic services, she has worked as Associate Editor for numerous renowned international journals: The Annals of Statistics (2022–2024), Information and Inference (2022–2024), Journal of Business & Economic Statistics (2018–2024), Journal of Econometrics (2015–2018), Journal of the American Statistical Association (2014–present), Journal of Multivariate Analysis (2013–2016) and The Econometrics Journal (2012–2024). Currently, she holds the position of Co-Editor-in-Chief of Journal of Business and Economic Statistics (JBES) and Statistics Surveys.

讲座简介

We investigate the robustness of the model-X knockoffs framework with respect to the misspecified or estimated feature distribution. We achieve such a goal by theoretically studying the feature selection performance of a practically implemented knockoffs algorithm, which we name as the approximate knockoffs (ARK) procedure, under the measures of the false discovery rate (FDR) and k-familywise error rate (k-FWER). The approximate knockoffs procedure differs from the model-X knockoffs procedure only in that the former uses the misspecified or estimated feature distribution. A key technique in our theoretical analyses is to couple the approximate knockoffs procedure with the model-X knockoffs procedure so that random variables in these two procedures can be close in realizations. We prove that if such coupled model-X knockoffs procedure exists, the approximate knockoffs procedure can achieve the asymptotic FDR or k-FWER control at the target level. We showcase three specific constructions of such coupled model-X knockoff variables, verifying their existence and justifying the robustness of the model-X knockoffs framework. Additionally, we formally connect our concept of knockoff variable coupling to a type of Wasserstein distance.

海报链接