Gaussian approximation for high-dimensional data: Recent progress
Date: Thursday, 21 October 2021
Time: 4-5 pm
Speaker: Prof Yuta Koike (University of Tokyo)
Abstract:
We review recent progress in Gaussian approximation of a sum of high-dimensional independent random vectors (and related statistics) on hyper-rectangles, where the dimension can be much larger than the sample size. Such approximation is useful for justifying bootstrap approximation of maximum statistics in high-dimensional data and is therefore important for uniform inference in high-dimensional models.