ALL OPTICAL, PARALLEL, CONTINUOUS ARTIFICIAL NEURAL SYSTEM

ARTIFICIAL NEURAL SYSTEMS (ANS) ARE EXTREMELY PROMISING AS AWAY OF ACHIEVING "TRUE INTELLIGENCE". THE JAPANESE HAVE CHALLENGED THE U.S. TO A MULTIBILLION DOLLAR 10 YEAR COOPERATIVE EFFORT IN THIS FIELD WHICH THEY CALL THE SIXTH GENERATION COMPUTER. THE ESSENCE IS TO USE MANY SIMPLE PROCESSORS MULTIPLY AND COMPLEXLY INTERCONNECTED. OPTICS OFFERS A THREE DIMENSIONALITY FOR INTERCONNECTS UNAPPROACHABLE BY ELECTRONICS. IN PARTICULAR, THE TECHNIQUEPROPOSED HERE CAN MAKE 10(12) WEIGHTED INTERCONNECTS IN PARALLEL. IN ADDITION, IT ALLOWS CONTINUOUS TIME OPERATION, FAST RESPONSE, LOW POWER CONSUMPTION, LOW COST, AND ALL OPTICAL. THE PHASE I EFFORT IS AN EXPLORATION OF THE POTENTIALS AND LIMITATIONS OF THE ALL OPTICAL ANS. ISSUES SUCH AS ACCURACY, SPEED, STABILITY, SIZE, AND ABILITY TO ACCOMMODATE VARIOUS ANS PARADIGMS WILL BE STUDIED. IN PHASE II A GENERIC ANS WILL BE DESIGNED USING COMMERCIAL PARTS AND ITS PERFORMANCE PROPERTIES WILL BE PREDICTED. THE CONTRACT WILL BE WITH A NODAL SYSTEMS CORPORATION, A COMPANY CREATED SOLELY TO EXPLOIT THE OPTICALANS. ONE OF ITS PRINCIPALS, DR. H. J. CAUFIELD, IS WELL KNOWN IN ALL FIELDS OF OPTICAL COMPUTING. A SUBCONTRACT TO THE CENTER FOR APPLIED OPTICS OF UAH PROVIDES FURTHER ACADEMIC SUPPORT.