题目: Model-free Variable Selection in High Dimension via Constrained Kernel Regression 

报告人：李长城 教授(大连理工大学)

 Abstract： We propose a model-free variable selection approach, namely constrained kernel regression. Instead of relying on model-based loss functions, the proposed constrained kernel regression is developed based on conditional independence relationship measured by conditional distance covariance/correlation. The conditional distance covariance/correlation is approximated by the kernel method. And the constrained kernel regression coefficient vector is defined to be the vector satisfying the conditional independence constraints. By varying the tuning parameter in the conditional independence constraints, the proposed approach provides a solution path. And we prove that the proposed approach can consistently identify the true important predictor set under high-dimensional model-free settings with appropriate tuning parameters. The advantage of the proposed method is further shown by various numerical studies. More specifically, the proposed method outperforms existing model-based in the presence of model misspecification and has better or comparable performance with existing methods when their models are correctly specified.

报告人简介：李长城，大连理工大学永利澳门官网入口教授。本科就读于北京大学永利澳门官网入口，获得统计学学士学位；博士阶段师从美国宾夕法尼亚州州立大学统计系李润泽教授，进行高维统计领域的学习，获得统计学博士学位。研究兴趣主要包括高维统计推断及高维因果推断。在高维统计的理论、应用以及计算方面进行了一系列研究，文章发表于一流学术期刊Journal of American Statistical Association、Journal of Econometrics、Annals of Applied Statistics、Statistica Sinica等，入选国家级青年人才计划。

时间： 2024年4月23日 19：00-20：00

地点：腾讯会议号：162-841-812

联系人：胡涛