Over the last 20 years or so, Bayesian networks (BNs) [Pe88, Ne90, RN95, CDLS99] have become the key method for representation and reasoning under uncertainty in AI. BNs not only provide a natural and compact way to encode exponentially sized joint probability distributions, but also provide a basis for efficient probabilistic inference. Although there exists polynomial time inference algorithm for specific classes of Bayesian networks, i.e., trees and singly connected networks, in general both exact belief update and belief revision are NPhard [Co90, Sh94]. Furthermore, approximations of them are also NPhard [DL93b, AH98]. Given the NPhard complexity results, one of the major challenges in applying BNs into real-world applications is the design of efficient approximate inference algorithms working under real-time constraints for very large probabilistic models. Researchers have developed various kinds of exact and approximate Bayesian network inference algorithms. Some of them are particularly designed for real-time inference. In this paper, we attempt to present a review to BN inference algorithms in general, and real-time inference algorithms in particular to provide a framework to understand the differences and relationships between these algorithms.

Chavez, R. Martin, Cooper, Gregory F.

In recent years, researchers in decision analysis and artificial intelligence (Al) have used Bayesian belief networks to build models of expert opinion. Using standard methods drawn from the theory of computational complexity, workers in the field have shown that the problem of probabilistic inference in belief networks is difficult and almost certainly intractable. K N ET, a software environment for constructing knowledge-based systems within the axiomatic framework of decision theory, contains a randomized approximation scheme for probabilistic inference. The algorithm can, in many circumstances, perform efficient approximate inference in large and richly interconnected models of medical diagnosis. Unlike previously described stochastic algorithms for probabilistic inference, the randomized approximation scheme computes a priori bounds on running time by analyzing the structure and contents of the belief network. In this article, we describe a randomized algorithm for probabilistic inference and analyze its performance mathematically. Then, we devote the major portion of the paper to a discussion of the algorithm's empirical behavior. The results indicate that the generation of good trials (that is, trials whose distribution closely matches the true distribution), rather than the computation of numerous mediocre trials, dominates the performance of stochastic simulation. Key words: probabilistic inference, belief networks, stochastic simulation, computational complexity theory, randomized algorithms.

An important subclass of hybrid Bayesian networks are those that represent Conditional Linear Gaussian (CLG) distributions --- a distribution with a multivariate Gaussian component for each instantiation of the discrete variables. In this paper we explore the problem of inference in CLGs. We show that inference in CLGs can be significantly harder than inference in Bayes Nets. In particular, we prove that even if the CLG is restricted to an extremely simple structure of a polytree in which every continuous node has at most one discrete ancestor, the inference task is NP-hard.To deal with the often prohibitive computational cost of the exact inference algorithm for CLGs, we explore several approximate inference algorithms. These algorithms try to find a small subset of Gaussians which are a good approximation to the full mixture distribution. We consider two Monte Carlo approaches and a novel approach that enumerates mixture components in order of prior probability. We compare these methods on a variety of problems and show that our novel algorithm is very promising for large, hybrid diagnosis problems.

Kozlov, Alexander V., Singh, Jaswinder Pal

Although probabilistic inference in a general Bayesian belief network is an NP-hard problem, computation time for inference can be reduced in most practical cases by exploiting domain knowledge and by making approximations in the knowledge representation. In this paper we introduce the property of similarity of states and a new method for approximate knowledge representation and inference which is based on this property. We define two or more states of a node to be similar when the ratio of their probabilities, the likelihood ratio, does not depend on the instantiations of the other nodes in the network. We show that the similarity of states exposes redundancies in the joint probability distribution which can be exploited to reduce the computation time of probabilistic inference in networks with multiple similar states, and that the computational complexity in the networks with exponentially many similar states might be polynomial. We demonstrate our ideas on the example of a BN2O network -- a two layer network often used in diagnostic problems -- by reducing it to a very close network with multiple similar states. We show that the answers to practical queries converge very fast to the answers obtained with the original network. The maximum error is as low as 5% for models that require only 10% of the computation time needed by the original BN2O model.

MAP is the problem of finding a most probable instantiation of a set of nvariables in a Bayesian network, given some evidence. MAP appears to be a significantly harder problem than the related problems of computing the probability of evidence Pr, or MPE a special case of MAP. Because of the complexity of MAP, and the lack of viable algorithms to approximate it,MAP computations are generally avoided by practitioners. This paper investigates the complexity of MAP. We show that MAP is complete for NP. We also provide negative complexity results for elimination based algorithms. It turns out that MAP remains hard even when MPE, and Pr are easy. We show that MAP is NPcomplete when the networks are restricted to polytrees, and even then can not be effectively approximated. Because there is no approximation algorithm with guaranteed results, we investigate best effort approximations. We introduce a generic MAP approximation framework. As one instantiation of it, we implement local search coupled with belief propagation BP to approximate MAP. We show how to extract approximate evidence retraction information from belief propagation which allows us to perform efficient local search. This allows MAP approximation even on networks that are too complex to even exactly solve the easier problems of computing Pr or MPE. Experimental results indicate that using BP and local search provides accurate MAP estimates in many cases.