Reconsidering Dependency Networks from an Information Geometry Perspective

Kazuya Takabatake [email protected], Shotaro Akaho [email protected] · 2026

Read Reconsidering Dependency Networks from an Information Geometry Perspective by Kazuya Takabatake [email protected], Shotaro Akaho [email protected] (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-2604-01117.txt.

Description

Takabatake and Akaho study pseudo-Gibbs sampling in dependency networks through an information-geometric lens, treating every sampling move as a projection onto the manifold of full conditionals. They define the full conditional divergence and give an upper bound pinning down the stationary distribution among distributions. Structure and parameter learning turn into optimization problems splitting into subproblems per node, and the learned model distribution tends to the true distribution as training samples grow without bound. Tests confirm the upper bound is tight in practice.

Catalog record

CREATORKazuya Takabatake [email protected], Shotaro Akaho [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE50022
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.01117

Browse the full library · About the Tower of Babel