FPGA Implementation of Neural Network Digital Systems

Abstract

The utilization of the FPGA (Field Programmable Gate Array) for neural network execution gives adaptability and flexibility in programmable frameworks. For Artificial Neural Networks (ANNs) in a real-time application, regular chip configuration suffers the limitations of schedule and cost. For neutral network digital systems that are comparably low precision, FPGAs have higher speed and a more modest size as compared to microcontrollers. In expansion, ANN digital systems based on FPGAs have reasonably been accomplished with classification applications. The programmability of reconfigurable FPGAs yields the accessibility of quick specific reason equipment for wide applications. Its programmability could set the conditions to investigate new neural network calculations and issues of a scale that would not be doable with a customary processor. The objective of this work is to understand the implementation of neural network digital systems utilizing FPGAs.