Message Scheduling for Performant, Many-Core Belief Propagation

TL;DR

This paper introduces a novel randomized message scheduling method for belief propagation on GPUs, significantly improving performance by effectively balancing speed and convergence in parallel inference tasks.

Contribution

The paper proposes a new randomized message scheduling approach, RnBP, that outperforms existing methods for belief propagation on many-core GPU systems.

Findings

RnBP outperforms existing scheduling methods.

Tradeoff between speed and convergence is demonstrated.

Effective utilization of parallelism improves BP performance.

Abstract

Belief Propagation (BP) is a message-passing algorithm for approximate inference over Probabilistic Graphical Models (PGMs), finding many applications such as computer vision, error-correcting codes, and protein-folding. While general, the convergence and speed of the algorithm has limited its practical use on difficult inference problems. As an algorithm that is highly amenable to parallelization, many-core Graphical Processing Units (GPUs) could significantly improve BP performance. Improving BP through many-core systems is non-trivial: the scheduling of messages in the algorithm strongly affects performance. We present a study of message scheduling for BP on GPUs. We demonstrate that BP exhibits a tradeoff between speed and convergence based on parallelism and show that existing message schedulings are not able to utilize this tradeoff. To this end, we present a novel randomized message scheduling approach, Randomized BP (RnBP), which outperforms existing methods on the GPU.