Conditioning Methods for Exact and Approximate Inference in Causal Networks

TL;DR

This paper introduces two new algorithms for exact and approximate inference in causal networks, improving efficiency and offering a trade-off between accuracy and computation time, supported by experimental results.

Contribution

It proposes dynamic conditioning for efficient exact inference and B-conditioning for adjustable approximate inference in causal networks.

Findings

Dynamic conditioning has linear complexity on certain networks.

B-conditioning allows control over approximation quality versus computation time.

Experimental results demonstrate the effectiveness of both algorithms.

Abstract

We present two algorithms for exact and approximate inference in causal networks. The first algorithm, dynamic conditioning, is a refinement of cutset conditioning that has linear complexity on some networks for which cutset conditioning is exponential. The second algorithm, B-conditioning, is an algorithm for approximate inference that allows one to trade-off the quality of approximations with the computation time. We also present some experimental results illustrating the properties of the proposed algorithms.