HARDWARE NEURAL NETWORK IMPLEMENTATION FOR THE RECOGNITION OF CHEMICAL AND BIOLOGICAL WARFARE AGENTS

RECENT DEVELOPMENTS IN THE FIELD OF ARTIFICIAL NEURAL NETWORKS HAVE DEMONSTRATED THE UTILITY OF THESE SYSTEMS FOR A VARIETY OF PURPOSES INCLUDING PATTERN RECOGNITION, ASSOCIATIVE MEMORY, AND CONTROL. TO DATE, THE MAJORITY OF THESE APPLICATIONS HAVE BEEN IMPLEMENTED OF CONVENTIONAL SEQUENTIAL PROCESSORS OR FAST FLOATING POINT PROCESSORS. DURING THE PAST TWO YEARS, SIGNIFICANT PROGRESS HAS ALSO BEEN MADE IN THE DEVELOPMENT OF BOTH ANALOG AND DIGITAL NEURAL DEVICES. IN FACT THE STATE OF THE ART IS SUFFICIENTLY ADVANCED, THAT THESE DEVICES ARE AVAILABLE AS COMMERCIAL, OFF THE SHELF, PRODUCTS. THE UNITED STATES ARMY CHEMICAL RESEARCH, DEVELOPMENT, AND ENGINEERING CENTER (CRDEC) HAS BEEN EXPLORING A VARIETY OF TECHNOLOGIES CAPABLE OF DETECTING SUBTOXIC CONCENTRATIONS OF CHEMICAL AND BIOLOGICAL THREAT AGENTS. THE CHEMICAL BIOLOGICAL MASS SPECTROMETER (CBMS) PROGRAM IS ONE OF THE KEY PROJECTS IN ACHIEVING THE ARMY'S GOAL OF IMPROVING THE SURVIVABILITY OF U.S. FORCES TO CB ATTACK. THE CURRENT CBMS DESIGN INCORPORATES A NEURAL NETWORK, DEVELOPED BY TAU CORPORATION, FOR THE RECOGNITION OFCHEMICAL AND BIOLOGICAL WARFARE AGENTS FROM THEIR MASS SPECTRA; THIS NETWORK IS IMPLEMENTED IN SOFTWARE AN EXECUTES ON A CONVENTIONAL SEQUENTIAL PROCESSOR. IN THIS PROPOSED EFFORT, TAU, TOGETHER WITH OUR SUBCONTRACTOR TELEDYNE CME, WILL DEFINE A BOARD WHICH WILL INCORPORATE ARTIFICIAL NEURAL NETWORKS IMPLEMENTED IN HARDWARE. THIS BOARD WILL BE DESIGNED TO INTEGRATE SEAMLESSLY INTO THE EXISTING CBMS HARDWARE ARCHITECTURE. ANTICIPATED BENEFITS/POTENTIAL COMMERICIAL APPLICATIONS OF THE RESEARCH OR DEVELOPMENT. THE PROPOSED RESEARCH AND DEVELOPMENT WILL RESULT IN THE DEVELOPMENT OF AN ENHANCED CAPABILITY FOR THE CBMS