New Insights into Learning with Correntropy Based Regression

TL;DR

This paper provides new theoretical insights into correntropy-based regression, showing its robustness, unifying different regression targets, and analyzing its convergence properties under certain noise conditions.

Contribution

It demonstrates that correntropy regression can be derived from minimum distance estimation, unifies approaches to mean, median, and mode regression, and analyzes its error bounds and convergence rates.

Findings

Correntropy regression is a form of minimum distance estimation with robustness properties.

It unifies the estimation of mean, median, and mode functions under certain conditions.

Error bounds and exponential convergence rates are established under conditional moment assumptions.

Abstract

Stemming from information-theoretic learning, the correntropy criterion and its applications to machine learning tasks have been extensively explored and studied. Its application to regression problems leads to the robustness enhanced regression paradigm -- namely, correntropy based regression. Having drawn a great variety of successful real-world applications, its theoretical properties have also been investigated recently in a series of studies from a statistical learning viewpoint. The resulting big picture is that correntropy based regression regresses towards the conditional mode function or the conditional mean function robustly under certain conditions. Continuing this trend and going further, in the present study, we report some new insights into this problem. First, we show that under the additive noise regression model, such a regression paradigm can be deduced from minimum distance estimation, implying that the resulting estimator is essentially a minimum distance estimator and thus possesses robustness properties. Second, we show that the regression paradigm, in fact, provides a unified approach to regression problems in that it approaches the conditional mean, the conditional mode, as well as the conditional median functions under certain conditions. Third, we present some new results when it is utilized to learn the conditional mean function by developing its error bounds and exponential convergence rates under conditional $(1+\epsilon)$-moment assumptions. The saturation effect on the established convergence rates, which was observed under $(1+\epsilon)$-moment assumptions, still occurs, indicating the inherent bias of the regression estimator. These novel insights deepen our understanding of correntropy based regression, help cement the theoretic correntropy framework, and also enable us to investigate learning schemes induced by general bounded nonconvex loss functions.