AI RESEARCH
Avoiding Non-Integrable Beliefs in Expectation Propagation
arXiv CS.LG
•
ArXi:2604.04264v1 Announce Type: cross Expectation Propagation (EP) is a widely used iterative message-passing algorithm that decomposes a global inference problem into multiple local ones. It approximates marginal distributions as ``beliefs'' using intermediate functions called ``messages''. It has been shown that the stationary points of EP are the same as corresponding constrained Bethe Free Energy (BFE) optimization problem. Therefore, EP is an iterative method of optimizing the constrained.