报告题目：Adaptive Multi-Prior Lasso for High-Dimensional Generalized Linear Models
报告人：张庆昭 教授
报告时间：2025年9月25日（周四）晚上19:00—20:00
报告地点：藕舫楼724会议室
腾讯会议号：931-714-126
主持人：卫雨婷 讲师
报告简介：
Incorporating external information into high-dimensional genetic modeling has been shown to substantially improve statistical performance, both theoretically and empirically. Such information, referred to as prior information or priors, has become increasingly available from multiple sources with advances in genetic research, yet its reliability can vary considerably. However, existing methods often incorporate these priors without assessing their quality, which may lead to unsatisfactory or even misleading results. To systematically and effectively leverage these priors, we propose the adaptive Multi-Prior Lasso -- a novel method that simultaneously identifies reliable prior sources and integrates them to enhance the current data modeling. Built within the high-dimensional generalized linear model framework, the proposed method assigns data-driven adaptive weights to each prior, selectively emphasizing more reliable sources while down-weighting less credible ones. We establish theoretical guarantees under regularity conditions and demonstrate the method’s superior performance in estimation, prediction, and variable selection through extensive simulations. Application to breast cancer genetic data further highlights its practical utility.
报告人简介：
张庆昭，厦门大学经济学院统计学与数据科学系教授，博士生导师，国际统计学会 elected member。2013年从中国科学院数学与系统科学院取得概率论与数理统计博士学位，2013-2016年先后在中国科学院大学和耶鲁大学进行博士后研究。主要研究方向为复杂数据分析、概率图模型、统计机器学习等。近年来在JASA、JBES、JMLR、JoE、Biometrics、Statistica Sinica、Statistics in Medicine、《统计研究》等期刊发表论文数十篇。目前担任国际统计学杂志 Annals of the Institute of Statistical Mathematics 的 Associate Editor，以及中国现场统计学会高维数据分析学会理事、统计交叉科学研究分会常务理事、全国工业统计学理事、中国青年统计学家协会理事等。主持国家自然科学基金、教育部人文社会科学基金和全国统计科学研究重点项目等。
欢迎广大师生踊跃参加！
数学与统计学院
江苏省应用数学（南京信息工程大学）中心
江苏省系统建模与数据分析国际合作联合实验室
江苏省统计科学研究基地
2025年9月24日
