Optimal convex $M$-estimation via score matching

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Feng, Oliver Y., Kao, Yu-Chun, Xu, Min, Samworth, Richard J.
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866913862581747712
author Feng, Oliver Y.
Kao, Yu-Chun
Xu, Min
Samworth, Richard J.
author_facet Feng, Oliver Y.
Kao, Yu-Chun
Xu, Min
Samworth, Richard J.
contents In the context of linear regression, we construct a data-driven convex loss function with respect to which empirical risk minimisation yields optimal asymptotic variance in the downstream estimation of the regression coefficients. At the population level, the negative derivative of the optimal convex loss is the best decreasing approximation of the derivative of the log-density of the noise distribution. This motivates a fitting process via a nonparametric extension of score matching, corresponding to a log-concave projection of the noise distribution with respect to the Fisher divergence. At the sample level, our semiparametric estimator is computationally efficient, and we prove that it attains the minimal asymptotic covariance among all convex $M$-estimators. As an example of a non-log-concave setting, the optimal convex loss function for Cauchy errors is Huber-like, and our procedure yields asymptotic efficiency greater than $0.87$ relative to the maximum likelihood estimator of the regression coefficients that uses oracle knowledge of this error distribution. In this sense, we provide robustness and facilitate computation without sacrificing much statistical efficiency. Numerical experiments using our accompanying R package 'asm' confirm the practical merits of our proposal.
format Preprint
id arxiv_https___arxiv_org_abs_2403_16688
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Optimal convex $M$-estimation via score matching
Feng, Oliver Y.
Kao, Yu-Chun
Xu, Min
Samworth, Richard J.
Statistics Theory
Methodology
Machine Learning
In the context of linear regression, we construct a data-driven convex loss function with respect to which empirical risk minimisation yields optimal asymptotic variance in the downstream estimation of the regression coefficients. At the population level, the negative derivative of the optimal convex loss is the best decreasing approximation of the derivative of the log-density of the noise distribution. This motivates a fitting process via a nonparametric extension of score matching, corresponding to a log-concave projection of the noise distribution with respect to the Fisher divergence. At the sample level, our semiparametric estimator is computationally efficient, and we prove that it attains the minimal asymptotic covariance among all convex $M$-estimators. As an example of a non-log-concave setting, the optimal convex loss function for Cauchy errors is Huber-like, and our procedure yields asymptotic efficiency greater than $0.87$ relative to the maximum likelihood estimator of the regression coefficients that uses oracle knowledge of this error distribution. In this sense, we provide robustness and facilitate computation without sacrificing much statistical efficiency. Numerical experiments using our accompanying R package 'asm' confirm the practical merits of our proposal.
title Optimal convex $M$-estimation via score matching
topic Statistics Theory
Methodology
Machine Learning
url https://arxiv.org/abs/2403.16688