Parameter Adjustment in Bayes Networks. The generalized noisy OR-gate

TL;DR

This paper introduces a new parameter learning model for Bayesian networks using Gaussian distributions, generalizes the noisy OR-gate for multivalued variables, and develops an efficient inference algorithm.

Contribution

It presents a novel Gaussian-based parameter update method, extends the noisy OR-gate to multivalued variables, and offers an efficient inference algorithm for complex networks.

Findings

Efficient probability computation proportional to parent count.

Successful generalization of noisy OR-gate for multivalued variables.

Effective parameter learning in Bayesian networks with loops.

Abstract

Spiegelhalter and Lauritzen [15] studied sequential learning in Bayesian networks and proposed three models for the representation of conditional probabilities. A forth model, shown here, assumes that the parameter distribution is given by a product of Gaussian functions and updates them from the _ and _r messages of evidence propagation. We also generalize the noisy OR-gate for multivalued variables, develop the algorithm to compute probability in time proportional to the number of parents (even in networks with loops) and apply the learning model to this gate.