Slice Sampling Particle Belief Propagation

TL;DR

This paper introduces a slice sampling based particle belief propagation algorithm that improves inference in continuous label Markov random fields by eliminating the need for a proposal distribution, leading to better convergence.

Contribution

The paper presents a novel slice sampling approach for particle belief propagation, removing the dependence on proposal distributions for improved inference efficiency.

Findings

Superior convergence in image denoising example

Validated on complex relational 2D feature tracking

Outperforms traditional MCMC methods

Abstract

Inference in continuous label Markov random fields is a challenging task. We use particle belief propagation (PBP) for solving the inference problem in continuous label space. Sampling particles from the belief distribution is typically done by using Metropolis-Hastings Markov chain Monte Carlo methods which involves sampling from a proposal distribution. This proposal distribution has to be carefully designed depending on the particular model and input data to achieve fast convergence. We propose to avoid dependence on a proposal distribution by introducing a slice sampling based PBP algorithm. The proposed approach shows superior convergence performance on an image denoising toy example. Our findings are validated on a challenging relational 2D feature tracking application.