Possibilistic Networks with Local Structure

A recent topic in probabilistic network learning is to exploit local network structure, i.e. to capture regularities in the conditional probability distributions, and to learn networks with local structure from data. In this paper we apply this idea to possibilistic networks, i.e. we try to capture regularities in conditional possibility distributions, and present a modification of the learning algorithm for Bayesian networks with local structure suggested in [7]. The idea underlying this modification is to exploit the decision graph structure that is used to represent the regularities not only to capture a larger set of regularities than decision trees can, but also to improve the learning process.