An FPGA-Based General Purpose Neural Network Chip With On-Chip Learning

In this paper, a description of a general purpose neural network chip with on-chip learning is given. The design is implemented using Xilinx Vertex II XCV 1000 Field Programmable Gate Array (FPGA). An XOR gate simulation was used as a testing application. Results and comparison of both software and hardware implementations are listed. A second testing application in noise cancellation and voice recognition is currently under development