AIJul 4, 2012
Sufficient conditions for convergence of Loopy Belief Propagation
arXiv:1207.1405v169 citations
Originality Incremental advance
AI Analysis
This addresses convergence guarantees for a widely used approximate inference algorithm in probabilistic graphical models, but is incremental as it builds on existing theory.
The paper derived new sufficient conditions for the convergence of Loopy Belief Propagation to a unique fixed point, improving upon prior results, with conditions appearing sharp for binary variables with (anti-)ferromagnetic interactions.
We derive novel sufficient conditions for convergence of Loopy Belief Propagation (also known as the Sum-Product algorithm) to a unique fixed point. Our results improve upon previously known conditions. For binary variables with (anti-)ferromagnetic interactions, our conditions seem to be sharp.