Implementation of a digital neuron with nonlinear activation function using piecewise linear approximation technique
Amit Mishra, Zaheeruddin, Krishna Raj · 2007
In this paper the authors propose a simplified digital hardware implementation of a neuron with nonlinear function using piecewise linear approximation technique. The sigmoidal function is selected as an activation function in this design .The proposed digital neuron could be used as a basic building block to make a general architecture of a neural network. The paper has incorporated the description of the main building block needed and problems faced during construction of neuron architecture. It also includes the preparation of inputs in a specified format required for piecewise linear approximation technique. Design methodology of activation function is discussed in this paper, which shows also the simulation results as well as FPGA implementation of the design.