Pruning RBF networks with QLP decomposition

Authors:
Edwirde Luiz Silva;Paulo Lisboa;Andrés González Carmona
Affiliations:
Departamento de Matemática e Estatística, Universidade Estadual da Paraíba - UEPB, Paraíba, Brasil;School of Computing and Mathematical Sciences, Liverpool John Moores University, Liverpool, England;Universidad de Granada, Departamento de Estadística e Investigación Operativa, España
Venue:
NN'07 Proceedings of the 8th Conference on 8th WSEAS International Conference on Neural Networks - Volume 8
Year:
2007

Citing 4
Cited 1

NETLAB: algorithms for pattern recognition

NETLAB: algorithms for pattern recognition
Pattern Recognition and Machine Learning (Information Science and Statistics)

Pattern Recognition and Machine Learning (Information Science and Statistics)
Comparison of Wiener filter solution by SVD with decompositions QR and QLP

AIKED'07 Proceedings of the 6th Conference on 6th WSEAS Int. Conf. on Artificial Intelligence, Knowledge Engineering and Data Bases - Volume 6
An efficient sequential learning algorithm for growing and pruning RBF (GAP-RBF) networks

IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics

Regression with radial basis function artificial neural networks using QLP decomposition to prune hidden nodes with different functional form

NN'07 Proceedings of the 8th Conference on 8th WSEAS International Conference on Neural Networks - Volume 8

Quantified Score

Hi-index	0.00

Visualization

Abstract

The radial basis function (RBF) network is the main practical alternative to the multi-layer perceptron for non-linear modeling. This paper describes a methodology to adjust predictions models, calculated from experimental data using regression with Gaussian basis functions reduced by QLP decomposition. After introducing the concepts of linear basis function models and matrix design reduced by QLP decomposition, the method is applied to RBF networks with different choices of the hidden basis function. The QLP method is effective for reducing the network size by pruning hidden nodes, resulting in a parsimonious model which accurate out-of-sample prediction for a sinusoidal test function. Simulation results showed that Gaussian basis functions produced the best results for this bench mark problem.