From Influence Diagrams to Junction Trees

TL;DR

This paper introduces a novel method for solving decision problems using influence diagrams by transforming them into junction trees through specialized triangulation and message passing algorithms, enabling efficient computation of utilities and policies.

Contribution

It presents a new approach that combines influence diagrams with junction tree algorithms, enhancing decision analysis efficiency.

Findings

Efficient computation of expected utilities.

Optimal decision policies derived from junction trees.

Improved triangulation techniques for influence diagrams.

Abstract

We present an approach to the solution of decision problems formulated as influence diagrams. This approach involves a special triangulation of the underlying graph, the construction of a junction tree with special properties, and a message passing algorithm operating on the junction tree for computation of expected utilities and optimal decision policies.