A Modularization Hardware Implementation Approach for Artificial Neural Network

Hardware implementation has been proven to be an effective way to take full advantage of the parallel and distributed computation ability of artificial neural network. To simplify the hardware implementation process of different kinds of neural networks, a modularization and digitization implementation method based on FPGA is proposed. Firstly, some commonly used artificial neural network structures are divided into several functional modules, which are then digitized with HDL. Finally, the hardware implementation of an expected neural network can be achieved by combining those related modules with ease in FPGA. The modularization construction and hardware implementation process of a discrete Hopfield neural network is taken as an example to validate the feasibility and effectiveness of the method.