Relevance-Based Incremental Belief Updating in Bayesian Networks

Relevance reasoning in Bayesian networks can be used to improve eeciency of belief updating algorithms by identifying and pruning those parts of a network that are irrelevant for the computation. Relevance reasoning is based on the graphical property of d{separation and other simple and eecient techniques, the computational complexity of which is usually negligible when compared to the complexity of belief updating in general. This paper describes a belief updating technique based on relevance reasoning that is applicable in practical systems in which observations and model revisions are interleaved with belief updating. Our technique invalidates the posterior beliefs of those nodes that depend probabilisti-1 cally on the new evidence or the revised part of the model and focuses the subsequent belief updating on the invalidated beliefs rather than on all beliefs. Very often observations and model updating invalidate only a small fraction of the beliefs and our scheme can then lead to substantial savings in computation. We report results of empirical tests for incremental belief updating when the evidence gathering is interleaved with reasoning. These tests demonstrate practical signiicance of our approach.