Bayesian Global-Local Regularization

TL;DR

This paper introduces a unified Bayesian framework for global-local regularization that adaptively shrinks parameters in high-dimensional models, achieving near-minimax risk and connecting classical and modern techniques.

Contribution

It develops a novel empirical Bayes approach with order constraints that generalizes Stein's estimator and provides theoretical guarantees for sparse, ordered models.

Findings

Achieves near-minimax risk in high-dimensional sparse models.

Unifies classical regularization and Bayesian hierarchical methods.

Demonstrates flexibility in polynomial regression applications.

Abstract

We propose a unified framework for global-local regularization that bridges the gap between classical techniques -- such as ridge regression and the nonnegative garotte -- and modern Bayesian hierarchical modeling. By estimating local regularization strengths via marginal likelihood under order constraints, our approach generalizes Stein's positive-part estimator and provides a principled mechanism for adaptive shrinkage in high-dimensional settings. We establish that this isotonic empirical Bayes estimator achieves near-minimax risk (up to logarithmic factors) over sparse ordered model classes, constituting a significant advance in high-dimensional statistical inference. Applications to orthogonal polynomial regression demonstrate the methodology's flexibility, while our theoretical results clarify the connections between empirical Bayes, shape-constrained estimation, and degrees-of-freedom adjustments.