Author(s)

F.J. Lopez Aligue, I. Acevedo Sotoca, H. Gonzalez Velasco, M.A. Jaramillo Moran & M. Garcia Valle

Abstract

A neural model for pattern recognition F. J. Lopez Aligue, I. Acevedo Sotoca, H. Gonzalez Velasco, M.A. Jaramillo Moran, M. Garcia Valle Departamento de Electronica e Ing. Electromecdnica, Facultad de Ciencias, Universidad de Extremadura, 06007, Badajoz, Spain A neural network model is presented which conjugates the properties of the neuron used and its relationship with the specific connectivity diagram being simulated. The new paradigm is based on the similarity that exists between the configuration of the input to the neuron, which is in its turn a part of the total input to the layer containing that neuron, and the arrangement and value of the weights. Together with this description, a semi-Hebbian learning strategy adapted to this model is presented. The results of a practical implementation to a graphic character recognition process are encouraging for planned further work. 1 Introduction The design of neural networks for pattern recognition purposes is one of the most signific

Keywords