Dynamic Optimal Training for Competitive Neural Networks

Authors

  • Mohammed Madiafi Information Processing Laboratory, Ben M'Sik Faculty of Sciences, Hassan II Mohammedia-Casablanca University, B.P. 7955 Av. Cdt Driss El Harti, 20800 Casablanca
  • Abdelaziz Bouroumi Information Processing Laboratory, Ben M'Sik Faculty of Sciences, Hassan II Mohammedia-Casablanca University, B.P. 7955 Av. Cdt Driss El Harti, 20800 Casablanca

Keywords:

Competitive neural networks, unsupervised learning, clustering, pattern classification, image compression

Abstract

This paper introduces an unsupervised learning algorithm for optimal training of competitive neural networks. The learning rule of this algorithm is rived from the minimization of a new objective criterion using the gradient descent technique. Its learning rate and competition difficulty are dynamically adjusted throughout iterations. Numerical results that illustrate the performance of this algorithm in unsupervised pattern classification and image compression are also presented, discussed, and compared to those provided by other well-known algorithms for several examples of real test data.

Downloads

Download data is not yet available.

Author Biographies

Mohammed Madiafi, Information Processing Laboratory, Ben M'Sik Faculty of Sciences, Hassan II Mohammedia-Casablanca University, B.P. 7955 Av. Cdt Driss El Harti, 20800 Casablanca

Received the B.S. degree in electrical engineering from Hassan II Mohammedia-Casablanca University, Casablanca, Morocco, in 2006 and the M.S. degree in information processing from Hassan II Mohammedia-Casablanca University, Casablanca, Morocco, in 2008. He is Phd student at the Modeling and Instrumentation Laboratory, Casablanca, Morocco. His current research interests include artificial neural networks, fuzzy and intelligent systems, evolutionary algorithms, pattern classification and recognition, unsupervised learning, and data compression.

Abdelaziz Bouroumi, Information Processing Laboratory, Ben M'Sik Faculty of Sciences, Hassan II Mohammedia-Casablanca University, B.P. 7955 Av. Cdt Driss El Harti, 20800 Casablanca

Received the "Doctorat d'état" degree from Mohammed V-Agdal University, Rabat, Morocco, in 2000. He is a Professor of information processing and computer science at the Hassan II Mohammedia-Casablanca University. His current research interests include fuzzy and intelligent systems, artificial neural networks, evolutionary algorithms, pattern classification and recognition, unsupervised learning, collaborative learning and e-learning.

Downloads

Published

2014-06-24

How to Cite

Madiafi, M., & Bouroumi, A. (2014). Dynamic Optimal Training for Competitive Neural Networks. Computing and Informatics, 33(2), 237–258. Retrieved from https://www.cai.sk/ojs/index.php/cai/article/view/544