INTELLIGENT TDA SYSTEMS USING NEURAL NETWORKS

WE PROPOSE TO DEVELOP AN INTELLIGENT TACTICAL DECISION AIDS (TDA) SYSTEM NEURAL NETWORK BY DEVELOPING A PROTOTYPE EXPERT SYSTEM, AND CONVERTING IT TO A NEURAL NETWORK REPRESENTATION USING AN AUTOMATIC TRANSLATOR. EXPERT SYSTEMS TECHNOLOGY IS REQUIRED TO REPRESENT A HIGH DEGREE OF INTELLIGENCE AND THE TRANSLATION IS REQUIRED TO ACHIEVE DRAMATICALLY IMPROVED PROCESSING TIME AND OTHER BENEFITS OF NEURAL NETWORKS. THIS PROTOTYPE WILL ABSOLUTELY DEMONSTRATE THE FEASIBLITY OF THE CONCEPT AND CLEAR THE WAY FOR DEVELOPMENT OF A COMPLETE SYSTEM IN PHASE II. WHILE EXPERT SYSTEMS ARE AMONG THE MOST COMMON AND SUCCESSFUL ARTIFICIAL INTELLIGENCE SYSTEMS, THEIR SLOWNESS AND FRAGILITY TEND TO LIMIT THEIR APPLICABILITY. HOWEVER, BY TRANSLATING A RULE AND FRAME BASED SYSTEM INTO A NEURAL NETWORK REPRESENTATION, PERFORMANCE IS DRAMATICALLY IMPROVED THROUGH PARALLELIZATION AND WE DERIVE THE COMBINED BENEFITS OF EXPERT SYSTEMS AND NEURAL NETWORKS. THE RESULTING SYSTEM IS ADAPTABLE, FAULT TOLERANT AND EASY TO UNDERSTAND BECAUSE OF ITS EXPERT SYSTEM ORIGIN. THE NEURAL NETWORK REPRESENTATION CAN BE IMPLEMENTED ON PARALLEL PROCESSING MACHINES OR NEURAL NETWORK HARDWARE.