报告时间：2021年5月10日, 19：00-20：00

报告地点：腾讯会议：947 864 925

报告人：刘一鸣博士（新加坡南洋理工大学）

报告摘要：In high dimensional regimes, we propose a new test to evaluate the conditional mean dependence of a response variable and the corresponding covariates. Inspired by the nonparametric method and the idea of the extreme-type statistic, we construct a novel testing statistic. Due to the idea of nonparametric estimation, the proposed method is applicable to the various structures of dependence between the response variables and the covariates. Theoretically, we find the limiting null distribution ofthe new proposed extreme type statistic under a mild mixing condition. Moreover, to make the testing more powerful in any underlying structures, we find a more general test statistic and prove its asymptotic properties. The power analysis of both methods is also considered. In the real data analysis, we also propose a new way to conduct the feature screening based on our result. The performance of our estimators and other methods are conducted through extensive simulations.

报告人介绍：刘一鸣，现为新加坡南洋理工大学博士后。2020年7月博士毕业于新加坡南洋理工大学数理学院。目前主要的研究方向包括：机器学习，高维统计推断的理论研究和应用，随机矩阵等。在硕博期间已在Statistica Sinica，Computational statistics & data analysis，Science China Mathematics等SCI期刊发表论文若干篇。

邀请人：马学俊