MODULARLY EXPANDABLE ARTIFICIAL NEURAL NETWORK COMPUTER FOR DEVELOPMENT AND APPLICATION OF LARGE SCALE ANNS

SUCCESSFUL APPLICATION OF ARTIFICIAL NEURAL NETWORKS (ANNS) TO ROBOTIC CONTROL, COMPUTER VISION, ADVANCED SIGNAL PROCESSING AND SPEECH RECOGNITION, REQUIRE TWO ORDERS OF MAGNITUDE INCREASE IN EXECUTION SPEED AND STORAGE SIZE OVER COURRENT ANN SIMULATION EQUIPMENT. IMPORTANT ALSO, IS THE NEED FOR EQUIPMENT THAT IS APPROPRIATE FOR PROLIFERATED APPLICATION OF ANNS TO A DIVERSE FIELD OF COMPUTERS AND ENVIRONMENTS. THE OPPORTUNITY NOW EXISTS TO DESIGN AND BUILD A LARGE SCALE ARTIFICIAL NEURAL NETWORK COMPUTER CAPABLE OF ACHIEVING SPEEDS OF 10(11) INTERCONNECTIONS/SECOND AND A STORAGE CAPACITY OF GIGA-BYTES. THIS COMPUTER WILL BE MODULARLY EXPANDABLE AND OFFER EASY INTEGRATION WITH EXISTING COMPUTER SYSTEMS FOR THE APPLICATIONS OF ARTIFICIAL NEURAL NETWORKS TO A HOST OF REAL WORLD ENVIRONMENTS. UNDER THE PHASE I EFFORT, QUARTEC WILL CONDUCT A PROOF-OF-CONCEPT STUDY LEADING TO THE DEVELOPMENT OF THE MODULARLY EXPANDABLE ARTIFICIAL NEURAL NETWORK (MEANN) COMPUTER. THE MEANN COMPUTER BRINGS TOGETHER SEVERAL STATE-OF-THE-ART TECHNOLOGIES TO MAKE POSSIBLE A POWERFUL ARTIFICIAL NEURAL NETWORK DEVELOPMENT AND APPLICATION PLATFORM.