Distributed Parameter Estimation in Probabilistic Graphical Models

TL;DR

This paper establishes theoretical foundations for distributed parameter estimation in undirected probabilistic graphical models, ensuring global consistency through a general condition on composite likelihood decompositions.

Contribution

It introduces a novel condition on composite likelihood decompositions that guarantees the consistency of distributed estimators in probabilistic graphical models.

Findings

Global consistency of distributed estimators is guaranteed under the new condition.

Theoretical results apply to a broad class of undirected probabilistic graphical models.

Abstract

This paper presents foundational theoretical results on distributed parameter estimation for undirected probabilistic graphical models. It introduces a general condition on composite likelihood decompositions of these models which guarantees the global consistency of distributed estimators, provided the local estimators are consistent.