Θ(1) Time Neural Network Minimum Distance Classifier and its Application to Optical Character Recognition Problem
Abdelwahed Namir, Mohammed Mestari, K. Akodadi, Antoine Badi · 2008
We propose a special neural network model, NNC, (Neural Network Classifier), which with a classification problem of l classes C1,C2,...,Cl, classifies an unknown vector to one class using a minimum distance classification technique. The NNC consists of three kinds of neurons, linear, quasi-linear and threshold-logic neurons, distributed over 12 layers. Thus, its processing time is constant and 12 times that of a single neuron. NNC architecture is regular and simple, it can be easily implemented using VLSI technology. High performance of NNC is demonstrated by applying it to an optical character recognition problem.