Convergent message passing algorithms - a unifying view

Message-passing algorithms have emerged as powerful techniques for approximate inference in graphical models. When these algorithms converge, they can be shown to find local (or sometimes even global) optima of variational formulations to the inference problem. But many of the most popular algorithms are not guaranteed to converge. This has lead to recent interest in convergent message-passing algorithms. In this paper, we present a unified view of convergent message-passing algorithms. We present a simple derivation of an abstract algorithm, tree-consistency bound optimization (TCBO) that is provably convergent in both its sum and max product forms. We then show that many of the existing convergent algorithms are instances of our TCBO algorithm, and obtain novel convergent algorithms "for free" by exchanging maximizations and summations in existing algorithms. In particular, we show that Wainwright's non-convergent sum-product algorithm for tree based variational bounds, is actually convergent with the right update order for the case where trees are monotonic chains.

[1]  Michael I. Jordan,et al.  Loopy Belief Propagation for Approximate Inference: An Empirical Study , 1999, UAI.

[2]  Alan L. Yuille,et al.  CCCP Algorithms to Minimize the Bethe and Kikuchi Free Energies: Convergent Alternatives to Belief Propagation , 2002, Neural Computation.

[3]  Shun-ichi Amari,et al.  Stochastic Reasoning, Free Energy, and Information Geometry , 2004, Neural Computation.

[4]  William T. Freeman,et al.  Constructing free-energy approximations and generalized belief propagation algorithms , 2005, IEEE Transactions on Information Theory.

[5]  Martin J. Wainwright,et al.  MAP estimation via agreement on trees: message-passing and linear programming , 2005, IEEE Transactions on Information Theory.

[6]  Martin J. Wainwright,et al.  A new class of upper bounds on the log partition function , 2002, IEEE Transactions on Information Theory.

[7]  Tom Heskes,et al.  Convexity Arguments for Efficient Minimization of the Bethe and Kikuchi Free Energies , 2006, J. Artif. Intell. Res..

[8]  Vladimir Kolmogorov,et al.  Convergent Tree-Reweighted Message Passing for Energy Minimization , 2006, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[9]  Tommi S. Jaakkola,et al.  Fixing Max-Product: Convergent Message Passing Algorithms for MAP LP-Relaxations , 2007, NIPS.

[10]  Tommi S. Jaakkola,et al.  Convergent Propagation Algorithms via Oriented Trees , 2007, UAI.

[11]  Tomás Werner,et al.  A Linear Programming Approach to Max-Sum Problem: A Review , 2007, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[12]  Tamir Hazan,et al.  Convergent Message-Passing Algorithms for Inference over General Graphs with Convex Free Energies , 2008, UAI.

[13]  Michael I. Jordan,et al.  Graphical Models, Exponential Families, and Variational Inference , 2008, Found. Trends Mach. Learn..

[14]  Tommi S. Jaakkola,et al.  Tree Block Coordinate Descent for MAP in Graphical Models , 2009, AISTATS.

[15]  Tamir Hazan,et al.  Norm-Product Belief Propagation: Primal-Dual Message-Passing for Approximate Inference , 2009, IEEE Transactions on Information Theory.