Tree-based reparameterization for approximate inference on loopy graphs

We develop a tree-based reparameterization framework that provides a new conceptual view of a large class of iterative algorithms for computing approximate marginals in graphs with cycles. It includes belief propagation (BP), which can be reformulated as a very local form of reparameterization. More generally, we consider algorithms that perform exact computations over spanning trees of the full graph. On the practical side, we find that such tree reparameterization (TRP) algorithms have convergence properties superior to BP. The reparameterization perspective also provides a number of theoretical insights into approximate inference, including a new characterization of fixed points; and an invariance intrinsic to TRP/BP. These two properties enable us to analyze and bound the error between the TRP/BP approximations and the actual marginals. While our results arise naturally from the TRP perspective, most of them apply in an algorithm-independent manner to any local minimum of the Bethe free energy. Our results also have natural extensions to more structured approximations [e.g., 1, 2].

[1]  Yair Weiss,et al.  Correctness of Local Probability Propagation in Graphical Models with Loops , 2000, Neural Computation.

[2]  G. Fournet.  Theory of Cooperative Phenomena , 1952 .

[3]  R. Kikuchi.  A Theory of Cooperative Phenomena , 1951 .

[4]  Yee Whye Teh,et al.  Belief Optimization for Binary Networks: A Stable Alternative to Loopy Belief Propagation , 2001, UAI.

[5]  Benjamin Van Roy,et al.  An Analysis of Turbo Decoding with Gaussian Densities , 1999, NIPS.

[6]  William T. Freeman,et al.  Correctness of Belief Propagation in Gaussian Graphical Models of Arbitrary Topology , 1999, Neural Computation.

[7]  W. Freeman,et al.  Generalized Belief Propagation , 2000, NIPS.

[8]  John B. Anderson,et al.  Tailbiting MAP Decoders , 1998, IEEE J. Sel. Areas Commun..

[9]  Thomas J. Richardson,et al.  The geometry of turbo-decoding dynamics , 2000, IEEE Trans. Inf. Theory.

[10]  Michael I. Jordan.  Graphical Models , 1998 .

[11]  Martin J. Wainwright,et al.  Stochastic processes on graphs with cycles: geometric and variational approaches , 2002 .

[12]  Tom Minka,et al.  A family of algorithms for approximate Bayesian inference , 2001 .

[13]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems - networks of plausible inference , 1991, Morgan Kaufmann series in representation and reasoning.

[14]  William T. Freeman,et al.  On the optimality of solutions of the max-product belief-propagation algorithm in arbitrary graphs , 2001, IEEE Trans. Inf. Theory.

[15]  A. Yuille.  A Double-Loop Algorithm to Minimize the Bethe and Kikuchi Free Energies , 2001 .

[16]  Brendan J. Frey,et al.  Iterative Decoding of Compound Codes by Probability Propagation in Graphical Models , 1998, IEEE J. Sel. Areas Commun..

[17]  Judea Pearl,et al.  Probabilistic reasoning in intelligent systems , 1988 .