Gaussian Approximations for Maxima of Random Vectors under (2+ι)-th Moments

05/27/2019

∙

We derive a Gaussian approximation result for the maximum of a sum of random vectors under (2+ι)-th moments. Our main theorem is abstract and nonasymptotic, and can be applied to a variety of statistical learning problems. The proof uses the Lindeberg telescopic sum device along with some other newly developed technical results.

READ FULL TEXT

Gaussian Approximations for Maxima of Random Vectors under (2+ι)-th Moments

Strong Gaussian Approximation for the Sum of Random Vectors

Semi-analytical approximations to statistical moments of sigmoid and softmax mappings of normal variables

Expectation Propagation on the Maximum of Correlated Normal Variables

Multi-Normex Distributions for the Sum of Random Vectors. Rates of Convergence

Maximum interpoint distance of high-dimensional random vectors

On a formula for moments of the multivariate normal distribution generalizing Stein's lemma and Isserlis theorem

Random Matrix Theory Proves that Deep Learning Representations of GAN-data Behave as Gaussian Mixtures

Gaussian Approximations for Maxima of Random Vectors under (2+ι)-th Moments

Related Research

Strong Gaussian Approximation for the Sum of Random Vectors

Semi-analytical approximations to statistical moments of sigmoid and softmax mappings of normal variables

Expectation Propagation on the Maximum of Correlated Normal Variables

Multi-Normex Distributions for the Sum of Random Vectors. Rates of Convergence

Maximum interpoint distance of high-dimensional random vectors

On a formula for moments of the multivariate normal distribution generalizing Stein's lemma and Isserlis theorem

Random Matrix Theory Proves that Deep Learning Representations of GAN-data Behave as Gaussian Mixtures