Equivalence of Informations Characterizes Bregman Divergences

Bregman divergences form a class of distance-like comparison functions which plays fundamental roles in optimization, statistics, and information theory. One important property of Bregman divergences is that they generate agreement between two useful formulations of information content (in the sense...

Full description

Saved in:
Bibliographic Details
Main Author: Philip S. Chodrow
Format: Article
Language:English
Published: MDPI AG 2025-07-01
Series:Entropy
Subjects:
Online Access:https://www.mdpi.com/1099-4300/27/7/766
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850078060281331712
author Philip S. Chodrow
author_facet Philip S. Chodrow
author_sort Philip S. Chodrow
collection DOAJ
description Bregman divergences form a class of distance-like comparison functions which plays fundamental roles in optimization, statistics, and information theory. One important property of Bregman divergences is that they generate agreement between two useful formulations of information content (in the sense of variability or non-uniformity) in weighted collections of vectors. The first of these is the Jensen gap information, which measures the difference between the mean value of a strictly convex function evaluated on a weighted set of vectors and the value of that function evaluated at the centroid of that collection. The second of these is the divergence information, which measures the mean divergence of the vectors in the collection from their centroid. In this brief note, we prove that the agreement between Jensen gap and divergence informations in fact characterizes the class of Bregman divergences; they are the only divergences that generate this agreement for arbitrary weighted sets of data vectors.
format Article
id doaj-art-d21fab4ccb734dfb9e859cfd037bea65
institution DOAJ
issn 1099-4300
language English
publishDate 2025-07-01
publisher MDPI AG
record_format Article
series Entropy
spelling doaj-art-d21fab4ccb734dfb9e859cfd037bea652025-08-20T02:45:38ZengMDPI AGEntropy1099-43002025-07-0127776610.3390/e27070766Equivalence of Informations Characterizes Bregman DivergencesPhilip S. Chodrow0Department of Computer Science, Middlebury College, Middlebury, VT 05753, USABregman divergences form a class of distance-like comparison functions which plays fundamental roles in optimization, statistics, and information theory. One important property of Bregman divergences is that they generate agreement between two useful formulations of information content (in the sense of variability or non-uniformity) in weighted collections of vectors. The first of these is the Jensen gap information, which measures the difference between the mean value of a strictly convex function evaluated on a weighted set of vectors and the value of that function evaluated at the centroid of that collection. The second of these is the divergence information, which measures the mean divergence of the vectors in the collection from their centroid. In this brief note, we prove that the agreement between Jensen gap and divergence informations in fact characterizes the class of Bregman divergences; they are the only divergences that generate this agreement for arbitrary weighted sets of data vectors.https://www.mdpi.com/1099-4300/27/7/766Bregman divergencemutual informationcharacterization theoremconvexity
spellingShingle Philip S. Chodrow
Equivalence of Informations Characterizes Bregman Divergences
Entropy
Bregman divergence
mutual information
characterization theorem
convexity
title Equivalence of Informations Characterizes Bregman Divergences
title_full Equivalence of Informations Characterizes Bregman Divergences
title_fullStr Equivalence of Informations Characterizes Bregman Divergences
title_full_unstemmed Equivalence of Informations Characterizes Bregman Divergences
title_short Equivalence of Informations Characterizes Bregman Divergences
title_sort equivalence of informations characterizes bregman divergences
topic Bregman divergence
mutual information
characterization theorem
convexity
url https://www.mdpi.com/1099-4300/27/7/766
work_keys_str_mv AT philipschodrow equivalenceofinformationscharacterizesbregmandivergences