随机复杂结构与数据科学重点实验室
学术报告


浏览次数:

 
Speaker:

王学钦,中山大学数学学院

Inviter: 孙六全 研究员
Title:
A two sample test for high-dimensional covariance matrices with Random Projection
Time & Venue:

2019.7.1 14:00 N702

Abstract:

We propose a novel two sample test for high-dimensional covariance matrices with random projection. The test is applicable even when the number of covariates is much larger than the sample size, and therefore enjoy wide scope of applicability in practice. Meanwhile, our test is nonparametric and model-free, which make the proposed test robust to model misspecification. Under some mild conditions, its asymptotic properties are derived without specifying an explicit relationship between the number of covariates and the sample size. Furthermore, the asymptotic null distribution of the proposed testing statistics is approximated by a permutation procedure. Numerical studies are carried out to evaluate the finite-sample performance of the proposed test. The results indicate that our test is competitive with other tests in a wide range of settings. Specially, it is very powerful when there are only few large or many diagonal disturbances between two covariance matrices.

Affiliation:  

学术报告中国科学院数学与系统科学研究院应用数学研究所
地址 北京市海淀区中关村东路55号 思源楼6-7层 南楼5-6、8层 100190
?2000-2013 京ICP备05058656号