Sharp regret–Hellinger bounds for Gaussian empirical Bayes via polynomial approximation

Jiafeng Chen and Yihong Wu · 2026

Read Sharp regret–Hellinger bounds for Gaussian empirical Bayes via polynomial approximation by Jiafeng Chen and Yihong Wu (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-02070.txt.

Description

Jiafeng Chen and Yihong Wu sharpen a classic result in empirical Bayes theory for the normal means model. The problem is to bound the regret of a learned Bayes rule by the Hellinger distance between estimated and true marginal densities. A 2009 result by Jiang and Zhang did this only after regularising the rule and with an extra cubic log factor, via a delicate recursive proof.

Chen and Wu use polynomial approximation and Bernstein-type bounds for weighted norms to bound the unregularised regret directly. The proof is simpler and gives sharper, sometimes optimal, bounds. For priors on a bounded range they prove the sharp form, and the method extends to priors with exponential tails. Heavy tails still need regularisation.

Catalog record

CREATORJiafeng Chen and Yihong Wu
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE56 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.02070

Browse the full library · About the Tower of Babel