SOFTWARE TOOLKIT FOR BASIC & APPLIED VISION RESEARCH

Not Available One of the most difficult problems in decision support is real-time decision-making under uncertainty. The information age has vastly increased the number of data sources while at the same time compressing the time available for making decisions. More the data sources, greater the uncertainty in the accuracy of the decision since any one of the data sources could be erroneous. Thus there is a greater risk that the commander (user) would make a wrong decision. In fact, assessing the risk involved in the decisionmakeing process is an important component of decision support. We adapt a technique, called Value-at-Risk (VAR), developed by the financial services community for risk-based decision-making. We propose a new approach for real-time computation and interactive display of VAR. The approach uses a Monte Carlo method for training a neural network using historical patterns of risk factors. We choose a particular scenario of the information warfare problem as our initial application domain. We propose to demonstrate a software prototype of this decision support system at the end of Phase I. Further, we propose a four-month option effort for demonstration the approach to potential DOD and private sector clients, and thereafter refining the approach and implementation to facilitate commercialization.BENEFITS: A software product/service for risk-based decision support aimed at information securtiy applications is expected.