Practical Modeling of Bayesian Decision Problems -- Exploiting Deterministic Relations

Thewidespreaduse of influence diagramsto represent andsolve Bayesiandecision problemsis still limited by the inflexibility andrather restrictive semanticsof influence diagrams. In this paper, wepropose a number of extensions and adjustmentsto the definition of influence diagramsin order to makethe practical use of influencediagramsmoreflexible andless restrictive. In particular, wedescribe howdeterministic relations can be exploitedto increasethe flexibility andefficiency of representing and solving Bayesiandecision problems. Theissues addressed in this paper were motivated by the construction of a decision support systemfor mission managementof unmannedunderwater vehicles.

• Type: Conference
• Year: 2001
• Where: FLAIRS
• Authors: Anders L. Madsen, Kristian G. Olesen, Søren L. Dittmer
• Keywords: Artificial Intelligence, FLAIRS 2001, Influence Diagramsin Order, Influencediagramsmoreflexible Andless Restrictive, Restrictive Semanticsof Influence