A visual introduction to Gaussian Belief Propagation

TL;DR

This paper visually introduces Gaussian Belief Propagation (GBP), an efficient message-passing algorithm for probabilistic inference on complex graphs, emphasizing its scalability and suitability for future machine learning systems.

Contribution

It provides a visual explanation of GBP and highlights its advantages as a scalable, hardware-friendly inference method for complex probabilistic models.

Findings

GBP updates rely only on local information

GBP converges independently of message schedule

GBP is suitable for scalable distributed inference

Abstract

In this article, we present a visual introduction to Gaussian Belief Propagation (GBP), an approximate probabilistic inference algorithm that operates by passing messages between the nodes of arbitrarily structured factor graphs. A special case of loopy belief propagation, GBP updates rely only on local information and will converge independently of the message schedule. Our key argument is that, given recent trends in computing hardware, GBP has the right computational properties to act as a scalable distributed probabilistic inference framework for future machine learning systems.