Entropy and Learning of Lipschitz Functions under Log-Concave Measures

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Bizeul, Pierre, Klartag, Boaz
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866916947500728320
author Bizeul, Pierre
Klartag, Boaz
author_facet Bizeul, Pierre
Klartag, Boaz
contents We study regression of $1$-Lipschitz functions under a log-concave measure $μ$ on $\mathbb{R}^d$. We focus on the high-dimensional regime where the sample size $n$ is subexponential in $d$, in which distribution-free estimators are ineffective. We analyze two polynomial-based procedures: the projection estimator, which relies on knowledge of an orthogonal polynomial basis of $μ$, and the least-squares estimator over low-degree polynomials, which requires no knowledge of $μ$ whatsoever. Their risk is governed by the rate of polynomial approximation of Lipschitz functions in $L^2(μ)$. When this rate matches the Gaussian one, we show that both estimators achieve minimax bounds over a wide range of parameters. A key ingredient is sharp entropy estimates for the class of $1$-Lipschitz functions in $L^2(μ)$, which are new even in the Gaussian setting.
format Preprint
id arxiv_https___arxiv_org_abs_2509_10355
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Entropy and Learning of Lipschitz Functions under Log-Concave Measures
Bizeul, Pierre
Klartag, Boaz
Probability
Functional Analysis
We study regression of $1$-Lipschitz functions under a log-concave measure $μ$ on $\mathbb{R}^d$. We focus on the high-dimensional regime where the sample size $n$ is subexponential in $d$, in which distribution-free estimators are ineffective. We analyze two polynomial-based procedures: the projection estimator, which relies on knowledge of an orthogonal polynomial basis of $μ$, and the least-squares estimator over low-degree polynomials, which requires no knowledge of $μ$ whatsoever. Their risk is governed by the rate of polynomial approximation of Lipschitz functions in $L^2(μ)$. When this rate matches the Gaussian one, we show that both estimators achieve minimax bounds over a wide range of parameters. A key ingredient is sharp entropy estimates for the class of $1$-Lipschitz functions in $L^2(μ)$, which are new even in the Gaussian setting.
title Entropy and Learning of Lipschitz Functions under Log-Concave Measures
topic Probability
Functional Analysis
url https://arxiv.org/abs/2509.10355