Some Practical Issues in Modeling Diagnostic Systems with Multiply Sectioned Bayesian Networks
Yang Xiang, Kristian G. Olesen, Finn Verner Jensen · 1999
Multiply Sectioned Bayesian Networks (MSBNs) provide a distributed framework for diagnosis of large systems based on probabilistic knowledge. To ensure exact inference, the partition of a large system into subsystems and the representation of subsystems must follow a set of technical constraints. How to satisfy these goals for a given system may not be obvious to a practitioner. In this paper, we address three practical modeling issues. Introduction Bayesian networks (BNs) (Pea88; Nea90; Jen96) provide a normative formalism for diagnosis based on probabilistic domain knowledge. In the past decade, researchers have studied how to model diagnostic problems using BNs (Hec90; DGH92; HBR95; ?; KP97), and many algorithms have been proposed to perform inference in BNs (Pea88; Sha96; CGH97; Jen96). Most of these methods are based on a centralized BN representation of the system to be diagnosed. Multiply Sectioned Bayesian Networks (MSBNs) provide a distributed framework for diagnosis of larg...