Estudo comparativo de análise e classificação de base de dados utilizando redes neurais mlp e máquinas de vetores de suporte svm

Abstract

This work aims to show the results of the use of two neural network implementation techniques to analyze and classify a database (named Car Evaluation) for automotive vehicles. In particular, this database has infor mation about several characteristics that can be used as evaluation criteria (price, comfort, technology) for the quality of a vehicle. In the first part of this work, we will present the database and its peculiarities, as well as the con cepts related to support vector machine (SVM) and multi-layer perceptron (MLP) techniques. In the second part, the implementation of these techniques is performed, verifying the performance and analyzing the results of each of them regarding the dynamics and precision in the data classification. In the last part, the final considerations of the techniques are presented in front of the obtained results.