Optimizing Reweighted Belief Propagation for Distributed Likelihood Fusion Problems
For researchers working on distributed inference, this work provides an analytical understanding and optimization of convergence for reweighted BP in cyclic graphs, though it is incremental in nature.
The paper investigates reweighted belief propagation for distributed likelihood fusion problems with short cycles, deriving convergence conditions and optimizing speed. It achieves significantly faster convergence compared to standard belief consensus.
Belief propagation (BP) is a powerful tool to solve distributed inference problems, though it is limited by short cycles in the corresponding factor graph. Such cycles may lead to incorrect solutions or oscillatory behavior. Only for certain types of problems are convergence properties understood. We extend this knowledge by investigating the use of reweighted BP for distributed likelihood fusion problems, which are characterized by equality constraints along possibly short cycles. Through a linear formulation of BP, we are able to analytically derive convergence conditions for certain types of graphs and optimize the convergence speed. We compare with standard belief consensus and observe significantly faster convergence.