Automatic Classification Using Neural Networks

• Autores: Gamal A.M. Al-Shawadfi, Hindi A. Al-Hindi
• Localización: International journal of the computer, the internet and management, ISSN 0858-7027, Vol. 11, Nº. 3 (SEP-DIC), 2003, págs. 72-82
• Idioma: inglés
• Texto completo no disponible (Saber más ...)
• Resumen

  • This paper proposes an artificial neural network (ANN) to perform linear and nonlinear classification of objects into several classes. The theoretical and practical aspects of the proposed approach are introduced, and its validity was evaluated by the rate of correct classifications. A Matlab macro program was written for automatic classification using an artificial neural network for linear and nonlinear classification problems. The network is designed, trained and tested with different sample sizes. The results were compared to those obtained from Fisher discriminant function. In contrast with the classical classification procedures, ANNs do not require any pre assumptions about types of data, distributions or the variance covariance matrices. The numerical results illustrate the capabilities of ANNs in solving linear and nonlinear classification problems.