Reverend Bayes on Inference Engines: A Distributed Hierarchical Approach

REVEREND BAYES ON INFERENCE ENGINES: A DISTRIBUTED HIERARCHICAL APPROACH(*)(**) Judea Pearl Cognitive Systems Laboratory School of Engineering and Applied Science University of California, Los Angeles 90024 ABSTRACT This paper presents generalizations of Bayes likelihood-ratio updating rule which facilitate an asynchronous propagation of the impacts of new beliefs and/or new evidence in hierarchically organized inference structures with multi-hypotheses variables. The computational scheme proposed specifies a set of belief parameters, communication messages and updating rules which guarantee that the diffusion of updated beliefs is accomplished in a single pass and complies with the tenets of Bayes calculus. Introduction This paper addresses the issue ofefficiently propagating the impact of new evidence and beliefs through a complex network of hierarchically organized inference rules. Such networks find wide applications in expert-systems Cl], [2],[3],speech recognition [4], situation assessment [5], the modelling of reading comprehension [6] and judicial reasoning [7]. Many AI researchers have accepted the myth that a respectable computational model of inexact reasoning must distort, modify or ignore at least some principles of probability calculus.