We give a nearly-optimal algorithm for testing uniformity of distributions
supported on $\-1,1\^n$, which makes $O (n/\varepsilon^2)$
queries to a subcube conditional sampling oracle (Bhattacharyya and Chakraborty
(2018)). The key technical component is a natural notion of random restriction
for distributions on $\-1,1\^n$, and a quantitative analysis of how such a
restriction affects the mean vector of the distribution. Along the way, we
consider the problem of mean testing with independent samples and provide a
nearly-optimal algorithm.
Description
[1911.07357] Random Restrictions of High-Dimensional Distributions and Uniformity Testing with Subcube Conditioning
%0 Journal Article
%1 canonne2019random
%A Canonne, Clément L.
%A Chen, Xi
%A Kamath, Gautam
%A Levi, Amit
%A Waingarten, Erik
%D 2019
%K bounds learning probability readings stats theory
%T Random Restrictions of High-Dimensional Distributions and Uniformity
Testing with Subcube Conditioning
%U http://arxiv.org/abs/1911.07357
%X We give a nearly-optimal algorithm for testing uniformity of distributions
supported on $\-1,1\^n$, which makes $O (n/\varepsilon^2)$
queries to a subcube conditional sampling oracle (Bhattacharyya and Chakraborty
(2018)). The key technical component is a natural notion of random restriction
for distributions on $\-1,1\^n$, and a quantitative analysis of how such a
restriction affects the mean vector of the distribution. Along the way, we
consider the problem of mean testing with independent samples and provide a
nearly-optimal algorithm.
@article{canonne2019random,
abstract = {We give a nearly-optimal algorithm for testing uniformity of distributions
supported on $\{-1,1\}^n$, which makes $\tilde O (\sqrt{n}/\varepsilon^2)$
queries to a subcube conditional sampling oracle (Bhattacharyya and Chakraborty
(2018)). The key technical component is a natural notion of random restriction
for distributions on $\{-1,1\}^n$, and a quantitative analysis of how such a
restriction affects the mean vector of the distribution. Along the way, we
consider the problem of mean testing with independent samples and provide a
nearly-optimal algorithm.},
added-at = {2019-11-19T18:18:46.000+0100},
author = {Canonne, Clément L. and Chen, Xi and Kamath, Gautam and Levi, Amit and Waingarten, Erik},
biburl = {https://www.bibsonomy.org/bibtex/2d50f230d5d45969ce0f9a711b86c7d7a/kirk86},
description = {[1911.07357] Random Restrictions of High-Dimensional Distributions and Uniformity Testing with Subcube Conditioning},
interhash = {e5ef0b840fce08bca697d44f69b33e03},
intrahash = {d50f230d5d45969ce0f9a711b86c7d7a},
keywords = {bounds learning probability readings stats theory},
note = {cite arxiv:1911.07357Comment: 40 pages},
timestamp = {2019-11-19T18:18:46.000+0100},
title = {Random Restrictions of High-Dimensional Distributions and Uniformity
Testing with Subcube Conditioning},
url = {http://arxiv.org/abs/1911.07357},
year = 2019
}