Probabilistic Networks: New Models and New Methods

David J C MacKay

In this paper I describe the implementation of a probabilistic model in BUGS. BUGS is a language for a program which performs `Bayesian inference Using Gibbs Sampling'. It is possible to implement surprisingly complex regression models in this environment. I demonstrate the inference of an unknown interpolant with an unknown input-dependent noise level.

postscript.


David MacKay's: home page, publications. bibtex file.