Scalable Inference Algorithms for the BMD Decision Architecture.

"IET proposes to develop a suite of hybrid, exact+approximate Bayesian network solution algorithms that can robustly and dynamically trade-off accuracy of the solution algorithm against the solution's time and memory requirements. These algorithms willenable scaling of the Missile Defense Agency's Decision Architecture to simulate and experiment with the expected solution complexity of fully modeled Decision Architecture Bayesian networks. The result of this Phase I effort will be a suite of hybridinference algorithms that combine the strengths of exact and approximate algorithms to dynamically optimize solution accuracy to Bayesian Network queries against the time and memory required to execute the hybrid solution algorithm computations. There areseveral commercial markets that routinely give rise to Bayesian Networks whose reduction in solution time is likely to be a selling point for a hybrid solution algorithm suite including of medical diagnosis, raw resource exploration and pharmaceutical drugdiscovery."