Evolutionary Programming Techniques for Modeling of C2 Processes.

摘要

This report contains a summary of the work done on this effort over the past twelve months. Two projects were performed, one of which investigates the basic ideas of evolutionary programming (EP) techniques, the other is an attempt to apply EP to an air campaign simulation. The first project provided evidence that EP can find the ideal solution in a complex space. It also seemed to show that a faster convergence to a good solution may be obtained by keeping the various parameters involved in EP within certain limits. Although there was insufficient time to complete the second project, it did show that it is possible to use neural networks (NNs) to control a two-dimensional entity in two dimensional space, and that even randomly initialized (untrained) NNs can exhibit interesting behavior.