Empirical Risk Minimization with Relative Entropy Regularization

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Perlaza, Samir M., Bisson, Gaetan, Esnaola, Iñaki, Jean-Marie, Alain, Rini, Stefano
Format: Preprint
Published: 2022
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910400422871040
author Perlaza, Samir M.
Bisson, Gaetan
Esnaola, Iñaki
Jean-Marie, Alain
Rini, Stefano
author_facet Perlaza, Samir M.
Bisson, Gaetan
Esnaola, Iñaki
Jean-Marie, Alain
Rini, Stefano
contents The empirical risk minimization (ERM) problem with relative entropy regularization (ERM-RER) is investigated under the assumption that the reference measure is a $σ$-finite measure, and not necessarily a probability measure. Under this assumption, which leads to a generalization of the ERM-RER problem allowing a larger degree of flexibility for incorporating prior knowledge, numerous relevant properties are stated. Among these properties, the solution to this problem, if it exists, is shown to be a unique probability measure, mutually absolutely continuous with the reference measure. Such a solution exhibits a probably-approximately-correct guarantee for the ERM problem independently of whether the latter possesses a solution. For a fixed dataset and under a specific condition, the empirical risk is shown to be a sub-Gaussian random variable when the models are sampled from the solution to the ERM-RER problem. The generalization capabilities of the solution to the ERM-RER problem (the Gibbs algorithm) are studied via the sensitivity of the expected empirical risk to deviations from such a solution towards alternative probability measures. Finally, an interesting connection between sensitivity, generalization error, and lautum information is established.
format Preprint
id arxiv_https___arxiv_org_abs_2211_06617
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle Empirical Risk Minimization with Relative Entropy Regularization
Perlaza, Samir M.
Bisson, Gaetan
Esnaola, Iñaki
Jean-Marie, Alain
Rini, Stefano
Statistics Theory
Information Theory
Machine Learning
The empirical risk minimization (ERM) problem with relative entropy regularization (ERM-RER) is investigated under the assumption that the reference measure is a $σ$-finite measure, and not necessarily a probability measure. Under this assumption, which leads to a generalization of the ERM-RER problem allowing a larger degree of flexibility for incorporating prior knowledge, numerous relevant properties are stated. Among these properties, the solution to this problem, if it exists, is shown to be a unique probability measure, mutually absolutely continuous with the reference measure. Such a solution exhibits a probably-approximately-correct guarantee for the ERM problem independently of whether the latter possesses a solution. For a fixed dataset and under a specific condition, the empirical risk is shown to be a sub-Gaussian random variable when the models are sampled from the solution to the ERM-RER problem. The generalization capabilities of the solution to the ERM-RER problem (the Gibbs algorithm) are studied via the sensitivity of the expected empirical risk to deviations from such a solution towards alternative probability measures. Finally, an interesting connection between sensitivity, generalization error, and lautum information is established.
title Empirical Risk Minimization with Relative Entropy Regularization
topic Statistics Theory
Information Theory
Machine Learning
url https://arxiv.org/abs/2211.06617