High Speed Artificial Neural Network Defense

HIGH-SPEED ARTIFICIAL NEURAL NETS ARE APPLIED IN MULTIPLE LAYERS AS PART OF A HYBRID ARCHITECTURE TO THE DECISION AIDING FUNCTIONS OF THE BATTLE MANAGEMENT SEGMENT OF A STRATEGIC DEFENSE COMMAND AND CONTROL SYSTEM. THE PROJECT WILL INCLUDE DEMONSTRATION OF TWO EARLY PROTOTYPES IN PHASE I. THESE SERVE TO FACILITATE DEFINITION OF REQUIREMENTS AND PRELIMINARY DESIGN OF A PROTOTYPE PHASE II TECHNOLOGY DEMONSTRATOR. THE ARTIFICIAL NEURAL NET TECHNOLOGY USED IS CAPABLE OF EXTREMLY HIGH-SPEED DECISION GENERATION. OUR APPROACH SEPARATES THE MECHANICS OF LEARNING FROM THE MECHANICS OF GENERATING DECISIONS. THE COMPUTATIONAL COMPLEXITY OF THE APPROACH IS MINIMAL. WE HAVE DEMONSTRATED GENERATION OF COMPLEX TACTICAL DECISIONS IN APPROXIMATELY 10 MILLISECONDS USING A PERSONAL COMPUTER IN OUR EARLIER WORK IN THIS AREA. USE OF THE "LEARNING" FEATURES OF OUR PROPOSED DECISION AIDING SYSTEM WILL REQUIRE NO SPECIAL SKILL OR TRAINING ON THE PART OF THE HUMAN WHO TAKES ON THE TASK OF TEACHING THE SYSTEM. THE PRIMARY OBJECTIVE OF OUR PROPOSED RESEARCH WILL BE TO DEMONSTRATE THE USEFULNESS OF HIGH-SPEED ARTIFICIAL NEURAL NETS, EMBEDDED IN A MULTI-LOOP ARCHITECTURE TO SUPPORT BATTLE MANAGEMENT DECISION AIDING IN A STRATEGIC DEFENSE COMMAND AND CONTROL ENVIRONMENT.