Bayesian Networks Cognitive Decision Aids Components for Battlefield Management

The focus of this proposal is to develop a suite of decision aids components to accomplish planning/replanning, learning, and inference. These components can then be assembled into applications to accommodate different needs and perspectives at different levels of the command structure. The particular components to be developed include (1) Influence Diagram component, (2) Bayesian Networks Inference component, (3) Bayesian Networks Learning component, and (4) Genetic Algorithm component. The Genetic Algorithm component and the Influence Diagram component allow users to generate and rank a set of alternative decisions, and enable users to evaluate the effects of changing events on a selected strategy. The Bayesian Networks Inference component allows users to maintain their situation awareness by propagating new information to all decision points which that information can affect. Adaptation of the decision aid process as new information becomes available is accomplished in the Bayesian Networks Learning component. Due the modular representation of knowledge of the Bayesian network structures and the ability of the Genetic algorithms to handle diverse optimization problems, the proposed components are scalable and sufficiently flexible to accommodate the needs of different decision aid applications. The project team consists of Scientific Systems Company, Inc. (SSCI) as the prime contractor, and the Boeing Company as the sub-contractor.