COMPARISON BETWEEN THE NEURAL NETWORKS MLP AND SOM FOR THE TASK OF THE STANDARDS RECOGNITION

Abstract

Artificial neural networks can be applied indifferent areas because of their ability to learn fromsamples. This article presents results of the performanceanalysis of a neural network supervised by Multiple LayersPerceptrons and another Neural Network Self-OrganizingMaps Non-supervised. This analysis aims to identify theperformance when a neural network is trained and aftercompare the results. The results were obtained from asimulator tool implemented in MATLAB ®. These studydoing part of a broader search of the behavior of neuralnetworks, which have different architectures and subjectedto different rates of input parameters will result in differentapplications.

Index Terms - Artificial Neural Networks, Multi LayerPerceptron, Self-Organizing Maps and Test Error.