α Belief Propagation as Fully Factorized Approximation

2019/08/23 by Dong Liu, Nima N. Moghadam, Liu, Dong +7
Computer Science · #Bayesian Modeling and Causal Inference #Distributed Sensor Networks and Detection Algorithms #Error Correcting Code Techniques #FOS: Computer and information sciences #Gaussian Processes and Bayesian Inference #Machine Learning (cs.LG) #Machine Learning (stat.ML)

paper · pdf · doi:10.48550/arxiv.1908.08906

openalex publication_date 2019/08/23 · openalex created_date 2022/07/28 · openalex updated_date 2026/07/28

Abstract

Belief propagation (BP) can do exact inference in loop-free graphs, but its performance could be poor in graphs with loops, and the understanding of its solution is limited. This work gives an interpretable belief propagation rule that is actually minimization of a localized α-divergence. We term this algorithm as α belief propagation (α-BP). The performance of α-BP is tested in MAP (maximum a posterior) inference problems, where α-BP can outperform (loopy) BP by a significant margin even in fully-connected graphs.

Related

• α Belief Propagation for Approximate Inference
• Accuracy Bounds for Belief Propagation
• Self-Guided Belief Propagation -- A Homotopy Continuation Method
• The Linearization of Belief Propagation on Pairwise Markov Networks
• Belief propagation for networks with loops: The neighborhoods-intersections-based method