REALIZATION OF A NEURONAL HARDWARE WITH D IGITAL SIGNAL PROCESSOR ANDPROGRAMMABLE GATE ARRAYS

In this paper we describe how the processing speed of a radial basis n eural network can be performed by the use of field programmable gate a rrays (FPGA), The calculation of the very time-consuming exponential f unction is taken by an optimized CORDIC-processor. We determine the nu mber of the necessary FPGAs and do a processing speed comparison betwe en FPGA and DSP referring to an application in speech recognition.