Abstract

I present a parallel algorithm for exact probabilistic inference in Bayesian networks. For polytree networks with n variables, the worstcase time complexity is O(logn) on a CREW PRAM (concurrent-read, exclusive-write parallel random-access machine) with n processors, for any constant number of evidence variables. For arbitrary networks, the time complexity is O(r 3w log n) for n processors, or O(w log n) for r 3w n processors, where r is the maximum range of any variable, and w is the induced width (the maximum clique size), after moralizing and triangulating the network. 1 INTRODUCTION Two key breakthroughs make representation of and reasoning with probabilities practical, and have led to a proliferation of related research within the artificial intelligence community. Bayesian networks exploit conditional independence to represent joint probability distributions compactly, and associated inference algorithms evaluate arbitrary conditional probabilities implied by the network r...

Citations

7321	Introduction to Algorithms - Cormen, Leiserson, et al. - 2001
5666	Probabilistic reasoning in intelligent systems - Pearl - 1988
2959	Artificial Intelligence: a Modern Approach - Russell, Norvig - 1995
1081	Local Computations with Probabilities on graphical structures and their applications to expert systems (with discussion - Lauritzen, Spiegelhalter
595	An introduction to parallel algorithms - Jájá - 1992
496	The computational complexity of probabilistic inference using bayesian belief networks - Cooper - 1990
240	Bayesian updating in causal probabilistic networks by local computations - Jensen, Lauritzen, et al. - 1990
220	Approximating probabilistic inference in Bayesian belief networks is NP-hard - Dagum, Luby - 1993
177	Bayesian analysis in expert systems - Spiegelhalter, Dawid, et al. - 1993
165	Probabilistic reasoning in expert systems: Theory and algorithms - Neapolitan - 1990
117	Parallel tree contraction and its application - Miller, Reif - 1985
103	Probabilistic inference and influence diagram - Shachter - 1988
83	A Simple Parallel Tree Contraction Algorithm - Abrahamson, Dadoun, et al.
42	Global conditioning for probabilistic inference in belief networks - Shachter, Andersen
24	Evidence absorption and propagation through evidence reversal - Shachter - 1989
17	Parallel implementations of probabilistic inference - KOZLOV, SINGH - 1996
16	Logarithmic-time updates and queries in probabilistic networks - Delcher, Grove, et al. - 1995
15	A parallel lauritzen-spiegelhalter algorithm for probabilistic inference - Kozlov, Singh - 1994
12	Directed reduction algorithms and decomposable graphs - Shachter, Andersen, et al. - 1990
6	Distributed inference in Bayesian networks - D'iez, Mira - 1994
4	Parallelizing probabilistic inference: Some early explorations - D'Ambrosio, Fountain, et al. - 1992
3	Decomposable probabilistic influence diagrams - Chyu - 1991
3	Lectures on parallel computation - Gibbons, Spirakis - 1993
2	Treewidth: Computations and Approximations - Klocks - 1994