Symbolic Probabilitistic Inference in Large BN2O Networks

TL;DR

This paper introduces symbolic probabilistic inference techniques for large BN2O networks, exploiting network structure to improve efficiency over previous methods and proposing a new approximation approach with promising initial results.

Contribution

It presents a novel application of the SPI language to large BN2O networks, enhancing inference efficiency and introducing a new approximation method.

Findings

Significant structural properties can be exploited for faster inference.

Symbolic techniques reduce computation for cause posterior marginals.

Preliminary experiments show promising results for the new approximation method.

Abstract

A BN2O network is a two level belief net in which the parent interactions are modeled using the noisy-or interaction model. In this paper we discuss application of the SPI local expression language to efficient inference in large BN2O networks. In particular, we show that there is significant structure, which can be exploited to improve over the Quickscore result. We further describe how symbolic techniques can provide information which can significantly reduce the computation required for computing all cause posterior marginals. Finally, we present a novel approximation technique with preliminary experimental results.