This paper presents a method to implement a multi-layer feedforward neural network (back propagation - BP network) using a field programmable gate array. First, the algorithm is simulated in theory using relevant software, and the hardware structure of the feedforward neural network is constructed on this basis. The Sigmoid activation function is mainly implemented by a lookup table, and a specific hardware solution to the XOR problem is presented. Finally, the Quartus II simulation results of the BP network solving the XOR problem are presented, indicating the feasibility of the solution.
You Might Like
Recommended ContentMore
Open source project More
Popular Components
Searched by Users
Just Take a LookMore
Trending Downloads
Trending ArticlesMore