A Localization Approach to Improve Iterative Proportional Scaling in Gaussian Graphical Models

TL;DR

This paper presents a localized iterative proportional scaling method for Gaussian graphical models, reducing computational costs by exploiting model structure, with numerical experiments demonstrating its efficiency.

Contribution

It introduces a novel localization technique for iterative proportional scaling in Gaussian graphical models, improving computational efficiency.

Findings

Reduced computational cost through localization

Effective in decomposable Gaussian models

Numerical experiments show competitive performance

Abstract

We discuss an efficient implementation of the iterative proportional scaling procedure in the multivariate Gaussian graphical models. We show that the computational cost can be reduced by localization of the update procedure in each iterative step by using the structure of a decomposable model obtained by triangulation of the graph associated with the model. Some numerical experiments demonstrate the competitive performance of the proposed algorithm.