Solving CLQG Influence Diagrams Using Arc-Reversal Operations in a Strong Junction Tree

This paper presents an architecture for solving conditional linear-quadratic Gaussian (CLQG) influence diagrams (IDs) by Lazy Propagation (LP). A strong junction tree (SJT) is used to guide the elimination order whereas the marginalization operation is based on arc-reversal (AR). The use of AR for marginalization simplifies the implementation and gives the architecture a number of advantages. The key benefits of using LP in combination with AR to solve CLQG IDs are illustrated by examples and in experiments. The results of a preliminary performance evaluation are promising.

[1]  Dennis Nilsson,et al.  Probabilities of Future Decisions , 2000 .

[2]  Robert G. Cowell,et al.  Local Propagation in Conditional Gaussian Bayesian Networks , 2005, J. Mach. Learn. Res..

[3]  Frank Jensen,et al.  From Influence Diagrams to junction Trees , 1994, UAI.

[4]  Rina Dechter,et al.  A New Perspective on Algorithms for Optimizing Policies under Uncertainty , 2000, AIPS.

[5]  Anders L. Madsen,et al.  Lazy Evaluation of Symmetric Bayesian Decision Problems , 1999, UAI.

[6]  Anders L. Madsen,et al.  Solving linear-quadratic conditional Gaussian influence diagrams , 2005, Int. J. Approx. Reason..

[7]  Prakash P. Shenoy.  Inference in Hybrid Bayesian Networks Using Mixtures of Gaussians , 2006, UAI.

[8]  Ross D. Shachter.  Evaluating Influence Diagrams , 1986, Oper. Res..

[9]  C. Robert Kenley.  INFLUENCE DIAGRAM MODELS WITH CONTINUOUS VARIABLES , 1986 .

[10]  Prakash P. Shenoy,et al.  Valuation-Based Systems for Bayesian Decision Analysis , 1992, Oper. Res..

[11]  A.L. Madsen,et al.  Variations over the message computation algorithm of lazy propagation , 2005, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics).

[12]  C. Robert Kenley,et al.  Gaussian influence diagrams , 1989 .

[13]  Thomas D. Nielsen.  Decomposition of influence diagrams , 2001, J. Appl. Non Class. Logics.

[14]  Ross D. Shachter,et al.  Decision Making Using Probabilistic Inference Methods , 1992, UAI.

[15]  Ronald A. Howard,et al.  Influence Diagrams , 2005, Decis. Anal..

[16]  Scott M. Olmsted.  On representing and solving decision problems , 1983 .

[17]  Prakash P. Shenoy,et al.  Decision making with hybrid influence diagrams using mixtures of truncated exponentials , 2004, Eur. J. Oper. Res..

[18]  Ross D. Shachter,et al.  Symbolic Probabilistic Inference in Belief Networks , 1990, AAAI.