Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence

The non-parametric version of Amari’s dually affine Information Geometry provides a practical calculus to perform computations of interest in statistical machine learning. The method uses the notion of a statistical bundle, a mathematical structure that includes both probability densities and random...

Full description

Saved in:
Bibliographic Details
Main Author: Giovanni Pistone
Format: Article
Language:English
Published: MDPI AG 2025-03-01
Series:Stats
Subjects:
Online Access:https://www.mdpi.com/2571-905X/8/2/25
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850164830267244544
author Giovanni Pistone
author_facet Giovanni Pistone
author_sort Giovanni Pistone
collection DOAJ
description The non-parametric version of Amari’s dually affine Information Geometry provides a practical calculus to perform computations of interest in statistical machine learning. The method uses the notion of a statistical bundle, a mathematical structure that includes both probability densities and random variables to capture the spirit of Fisherian statistics. We focus on computations involving a constrained minimization of the Kullback–Leibler divergence. We show how to obtain neat and principled versions of known computations in applications such as mean-field approximation, adversarial generative models, and variational Bayes.
format Article
id doaj-art-e8163ac4cdf44529a792e70e5f6c4e4f
institution OA Journals
issn 2571-905X
language English
publishDate 2025-03-01
publisher MDPI AG
record_format Article
series Stats
spelling doaj-art-e8163ac4cdf44529a792e70e5f6c4e4f2025-08-20T02:21:53ZengMDPI AGStats2571-905X2025-03-01822510.3390/stats8020025Affine Calculus for Constrained Minima of the Kullback–Leibler DivergenceGiovanni Pistone0De Castro Statistics, Collegio Carlo Alberto, 10122 Torino, ItalyThe non-parametric version of Amari’s dually affine Information Geometry provides a practical calculus to perform computations of interest in statistical machine learning. The method uses the notion of a statistical bundle, a mathematical structure that includes both probability densities and random variables to capture the spirit of Fisherian statistics. We focus on computations involving a constrained minimization of the Kullback–Leibler divergence. We show how to obtain neat and principled versions of known computations in applications such as mean-field approximation, adversarial generative models, and variational Bayes.https://www.mdpi.com/2571-905X/8/2/25information geometryKullback–Leibler divergencestatistical bundlenatural gradient
spellingShingle Giovanni Pistone
Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
Stats
information geometry
Kullback–Leibler divergence
statistical bundle
natural gradient
title Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
title_full Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
title_fullStr Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
title_full_unstemmed Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
title_short Affine Calculus for Constrained Minima of the Kullback–Leibler Divergence
title_sort affine calculus for constrained minima of the kullback leibler divergence
topic information geometry
Kullback–Leibler divergence
statistical bundle
natural gradient
url https://www.mdpi.com/2571-905X/8/2/25
work_keys_str_mv AT giovannipistone affinecalculusforconstrainedminimaofthekullbackleiblerdivergence