Abstract

Abstract Bayesian networks provide a modeling language and associated inference algorithm for stochastic domains. They have been successfully applied in a variety of medium-scale applications. However, when faced with a large complex domain, the task of modeling using Bayesian networks begins to resemble the task of pro gramming using logical circuits. In this paper, we de scribe an object-oriented Bayesian network (OOBN) lan guage, which allows complex domains to be described in terms of inter-related objects. We use a Bayesian net work fragment to describe the probabilistic relations be tween the attributes of an object. These attributes can themsel ves be objects, providing a natural framework for encoding part-of hierarchies. Classes are used to pro vide a reusable probabilistic model which can be applied to multiple similar objects. Classes also support inher itance of model fragments from a class to a subclass, allowing the common aspects of related classes to be defined only once. Our language has clear declarative semantics: an OOBN can be interpreted as a stochas tic functional program, so that it uniquely specifies a probabilistic model. We provide an inference algorithm for OOBNs, and show that much of the structural infor mation encoded by an OOBN-particularly the encap sulation of variables within an object and the reuse of model fragments in different contexts--can also be used to speed up the inference process.