Implementation of artificial neural networks using current mode analogue circuit techniques
P.K. Houselander, John T. Taylor · 1989
The logistic function is a close approximation to the biological neuron's dendritic input (graded response) versus the output activation (number of action potentials/second). It also has some useful mathematical characteristics and the resulting output is a unipolar signal. An artificial neural network is constructed by cascading these simple neurons in structures consisting of one or more layers.