Symbolic Probabilitistic Inference in Large BN2O Networks
This work addresses computational bottlenecks in belief network inference for AI and probabilistic modeling, but appears incremental as it builds on existing methods like SPI and Quickscore.
The paper tackles efficient probabilistic inference in large BN2O networks by applying the SPI local expression language, showing it exploits structure to improve over Quickscore and reduces computation for cause posterior marginals, with preliminary results from a novel approximation technique.
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.