Multiplicative Factorization of Noisy-Max

TL;DR

This paper introduces a new representation for noisy-max that enables efficient exact inference in large Bayesian networks, demonstrated on complex medical networks where previous methods struggled.

Contribution

We propose a novel multiplicative factorization of noisy-max that improves inference efficiency in general Bayesian networks.

Findings

Successfully computed queries in QMR-DT and CPCS networks.

Outperformed previous exact inference methods on large medical networks.

Demonstrated practical applicability of the new representation.

Abstract

The noisy-or and its generalization noisy-max have been utilized to reduce the complexity of knowledge acquisition. In this paper, we present a new representation of noisy-max that allows for efficient inference in general Bayesian networks. Empirical studies show that our method is capable of computing queries in well-known large medical networks, QMR-DT and CPCS, for which no previous exact inference method has been shown to perform well.