Implementation of digital neuron cell using 8-bit activation function

This paper presents the development of the neuron through digital component. The brain generated signals are in the form of spikes similar to electrical pulses. The most important property of neural networks is ability of learning and in artificial neural networks the knowledge (learning information) is represented in the form of weights of the connections between the neurons. Artificial neural networks simplify the behavior of the human brain so artificial neural network is applicable in different fields such as automation, medical, robotics, electronics, security, transport, military, aviation, etc. To deal with the problem of implementation here top down method is used. Which is nothing but to divide a complex design in easier designs or modules, each module is redefined with greater details or divided in more subsystems. Here sigmoid function is used as activation function and this nonlinear function is calculated by using linear piecewise technique. And further approximations have been taken on account of reducing the input output functions.