OPTOELECTRONIC IMPLEMENTATION OF ARTIFICIALNEURAL NETWORK: PERCEPTRON LEARNING RULE AND MCATEGORYCLASSIFIER

Abstract

Single neuron perceptron is designed as a classifier of two different classes using the hardlimiter activation function (i.e. in the absence of light, and presence of light). An example is designed and tested so that the proposed circuit learned different categories and then used as a
classifier for two different classes because of the use of single neuron. Additional electronic circuits were used for computation processes. The Computer simulation results indicate stable solution that compares with theoretical results. Single layer perceptron M-category classifier is designed as a classifier for more than two classes. An example is designed and tested for the verification. The example learns after (5) iterations. Computer simulation results indicate stable solution that compares favorably with theoretical results.