Computing Upper and Lower Bounds on Likelihoods in Intractable Networks

TL;DR

This paper introduces deterministic methods to compute bounds on marginal probabilities in large intractable networks, aiding approximate inference when exact calculations are infeasible.

Contribution

It presents novel techniques for bounding likelihoods in sigmoid and noisy-OR networks, improving approximate inference in large-scale probabilistic models.

Findings

Bounds are tight as shown by numerical experiments

Techniques are effective for large networks where exact inference is impossible

Applicable to sigmoid and noisy-OR network structures

Abstract

We present deterministic techniques for computing upper and lower bounds on marginal probabilities in sigmoid and noisy-OR networks. These techniques become useful when the size of the network (or clique size) precludes exact computations. We illustrate the tightness of the bounds by numerical experiments.