Multi-Valued Neuron with Sigmoid Activation Function for Pattern ClassificationReportar como inadecuado




Multi-Valued Neuron with Sigmoid Activation Function for Pattern Classification - Descarga este documento en PDF. Documentación en PDF para descargar gratis. Disponible también para leer online.

Multi-Valued Neuron MVN was proposed forpattern classification. It operates with complex-valued inputs, outputs, andweights, and its learning algorithm is based on error-correcting rule. Theactivation function of MVN is not differentiable. Therefore, we can not applybackpropagation when constructing multilayer structures. In this paper, we proposea new neuron model, MVN-sig, to simulate the mechanism of MVN withdifferentiable activation function. We expect MVN-sig to achieve higherperformance than MVN. We run several classification benchmark datasets to comparethe performance of MVN-sig with that of MVN. The experimental results show agood potential to develop a multilayer networks based on MVN-sig.

KEYWORDS

Pattern Classification; Multi-Valued Neuron MVN; Differentiable Activation Function; Backpropagation

Cite this paper

Wu, S. , Chiou, Y. and Lee, S. 2014 Multi-Valued Neuron with Sigmoid Activation Function for Pattern Classification. Journal of Computer and Communications, 2, 172-181. doi: 10.4236-jcc.2014.24023.





Autor: Shen-Fu Wu, Yu-Shu Chiou, Shie-Jue Lee

Fuente: http://www.scirp.org/



DESCARGAR PDF




Documentos relacionados