Answering queries in hybrid Bayesian networks using importance sampling

In this paper we propose an algorithm for answering queries in hybrid Bayesian networks where the underlying probability distribution is of class MTE (mixture of truncated exponentials). The algorithm is based on importance sampling simulation. We show how, like existing importance sampling algorithms for discrete networks, it is able to provide answers to multiple queries simultaneously using a single sample. The behaviour of the new algorithm is experimentally tested and compared with previous methods existing in the literature.

► An algorithm for answering queries in hybrid Bayesian networks is proposed.
► The underlying probability distribution is of class MTE.
► The algorithm is based on importance sampling simulation.
► Answers to multiple queries simultaneously using a single sample can be carried out.
► The performance of the algorithm is tested and compared with previous methods.