Information Fusion and Decision Making for Utility-based Agents

摘要

We describe a model of utility-based agents that synthesizes information and determines the best way to make decisions in multi-agent systems. In this model the beliefs axe decomposed into two levels. The higher level represents the possible information about the possible worlds. The lower level subsequently estimates a number function for the belief. We also present a decision theory for this kind of beliefs. The Dempster-Shafer theory is used to describe the beliefs in this decision theory. To describe utilities we introduce a multi-set valued mapping to represent the relationship between action utilities and state partial ignorance.