报告题目：Why Both Asymptotic Theory and Applications Matter in the Age of AI

主讲人：郭绍俊副教授（中国人民大学）

时间：2026年10月17日（周六）16:00 p.m.

地点：北院卓远楼305会议室

主办单位：统计与数学学院

摘要：

Asymptotic theory plays a central role in statistical inference, particularly in the analysis of real-world data. However, some asymptotic frameworks fall short in practical settings. In this talk, we first review several historical breakthroughs in statistical research and then present an example from nonparametric regression and regression discontinuity designs (RDD). We show that Wilks’ theorem is applicable only when an undersmoothed bandwidth is selected. To address this limitation, we propose a novel approach, Robust Empirical Likelihood, designed specifically for nonparametric regression and RDD. We prove that the Wilks phenomenon holds under more relaxed bandwidth conditions in nonparametric regression and in both sharp and fuzzy RDD settings. Extensive simulation studies confirm the effectiveness of the proposed method. Ultimately, this example serves as a wake-up call, highlighting the urgent need to develop asymptotic frameworks that effectively support empirical applications in the age of AI.

主讲人简介：

郭绍俊，现为中国人民大学统计与大数据研究院长聘副教授。2003年本科毕业于山东师范大学，2008年获得中国科学院数学与系统科学研究院理学博士学位。2008-2016年任中国科学院数学与系统科学研究院助理研究员。2009-2010年赴美国普林斯顿大学运筹与金融工程系博士后研究，主要研究方向为高维数据分析。2014-2016年赴英国伦敦经济学院统计系做博士后研究，主要研究大维时间序列建模。目前主要研究方向有：统计学习；非参数及半参数统计建模；高维生存分析及函数型数据分析等。研究成果发表于JRSS B, JASA, Biometrika, Bernoulli, Journal of Econometrics, JBES等期刊。