Simplifying Bayesian Belief Nets while Preserving MPE or MPGE Ordering

Book One

Volume

Proceedings of the AAAI Conference on Artificial Intelligence, 12

Student Abstracts

The abstraction of probability inference is a process of searching for a representation in which only the desirable properties of the solutions are preserved. Simplification is one of such abstraction which reduces the size of large databases and speeds transmission and processing of probabilistic information[Sy and Sher, 94].