Reference: Suermondt, H. J. & Cooper, G. F. Updating Probabilities in Multiply Connected Belief Networks. March, 1988.
Abstract: This paper focuses on probability updates in multiply connected belief networks. Pearl has designed the method of conditioning, which enables us to apply his alogrithm for belief updates in singly connected networks to multiply connected belief newtoks by selecting a cutset for the newtork and instantiating these cutset nodes. We discuss conditions that need to be satisfied by the cutset nodes. We present a heuristic algorithm for finding a small cutset that satisfies these conditions.
Notes: 9 pages.