Showing papers by "David Poole published in 1994"

A simple approach to Bayesian network computations

Abstract: The general problem of computing poste rior probabilities in Bayesian networks is NP hard Cooper However e cient algorithms are often possible for particular applications by exploiting problem struc tures It is well understood that the key to the materialization of such a possibil ity is to make use of conditional indepen dence and work with factorizations of joint probabilities rather than joint probabilities themselves Di erent exact approaches can be characterized in terms of their choices of factorizations We propose a new approach which adopts a straightforward way for fac torizing joint probabilities In comparison with the clique tree propagation approach our approach is very simple It allows the pruning of irrelevant variables it accommo dates changes to the knowledge base more easily it is easier to implement More importantly it can be adapted to utilize both intercausal independence and condi tional independence in one uniform frame work On the other hand clique tree prop agation is better in terms of facilitating pre computations

Representing diagnosis knowledge

Abstract: This paper considers therepresentation problem: namely how to go from an abstract problem to a formal representation of the problem. We consider this for two conceptions of logic-based diagnosis, namely abductive and consistency-based diagnosis. We show how to represent diagnostic problems that can be conceptualised causally in each of the frameworks, and show that both representations of the same problems give the same answers. This is a local transformation that allows for an expressive (albeit propositional) language for giving the constraints on what symptoms and causes can coexist, including non-strict causation. This non-strict causation can be represented in each frameworkwithout adding special reasoning constructs to either framework. This is presented as a starting point for a study of the representation problem in diagnosis, rather than as an end in itself.

Intercausal independence and heterogeneous factorization

Abstract: It is well known that conditional independence can be used to factorize a joint probability into a multiplication of conditional probabilities. This paper proposes a constructive definition of intercausal independence, which can be used to further factorize a conditional probability. An inference algorithm is developed, which makes use of both conditional independence and intercausal independence to reduce inference complexity in Bayesian networks.

Solving asymmetric decision problems with influence diagrams

Abstract: While influence diagrams have many advantages as a representation framework for Bayesian decision problems, they have a serious drawback in handling asymmetric decision problems. To be represented in an influence diagram, an asymmetric decision problem must be symmetrized. A considerable amount of unnecessary computation may be involved when a symmetrized influence diagram is evaluated by conventional algorithms. In this paper we present an approach for avoiding such unnecessary computation in influence diagram evaluation.

Intercausal independence and heterogeneous actorization

Abstract: It is well known that conditional independence can be used to factorize a joint probability into a multiplication of conditional probabilities. This paper proposes a constructive definition of intercausal independence, which can be used to further factorize a conditional probability. An inference algorithm is developed, which makes use of both conditional independence and intercausal independence to reduce inference complexity in Bayesian networks.

Minimizing decision table sizes in influence diagrams: dimension shrinking

Abstract: One goal in evaluating an influence diagram is to compute an optimal decision table for each decision node. More often than not, one is able to shrink the sizes of some of the optimal decision tables without any loss of information. This paper investigates when the opportunities for such shrinkings arise and how can we detect them as early as possible so as to to avoid unnecessary computations. One type of shrinking, namely dimension shrinking, is studied. A relationship between dimension shrinking and what we call lonely arcs is established, which enables us to make use of the opportunities for dimension shrinking by means of pruning lonely arcs at a preprocessing stage.