Dynamic Planar-Cuts: Efficient Computation of Min-Marginals for Outer-Planar MRFs

This paper presents an algorithm for Dynamic MAP inference and the computation of min-marginals in boolean outer-planar MRFs. Our goal is to efficiently solve an instance of the MAP problem given that we have already solved a related instance. As an example of Dynamic MAP inference, we consider the problem of computing minmarginals. The theoretical contribution of this paper is that we prove that we can extend the MAP inference algorithm of Schraudolph & Kamenetsky (2008) to efficiently compute exact min-marginals for all variables in the same time complexity as the MAP algorithm, which is a speed-up of O(n) over a näıve extension. We call our proposed algorithm – Dynamic Planar-Cuts. As a practical contribution, our implementation achieves a 25-fold speedup (for a 1000 node network) over the näıve approach.