Letter: Improving constructive training of RBF networks through selective pruning and model selection

Authors:
Adriano L. I. Oliveira;Bruno J. M. Melo;Silvio R. L. Meira
Affiliations:
Polytechnic School, University of Pernambuco, Rua Benfica, 455, Madalena, Recife - PE 50.750-410, Brazil and Center of Informatics, Federal University of Pernambuco, P.O. Box 7851, Cidade Universi ...;Polytechnic School, University of Pernambuco, Rua Benfica, 455, Madalena, Recife - PE 50.750-410, Brazil;Center of Informatics, Federal University of Pernambuco, P.O. Box 7851, Cidade Universitaria, Recife - PE 50.732-970, Brazil
Venue:
Neurocomputing
Year:
2005

Citing 1
Cited 5

Improving RBF-DDA Performance on Optical Character Recognition through Parameter Selection

ICPR '04 Proceedings of the Pattern Recognition, 17th International Conference on (ICPR'04) Volume 4 - Volume 04

Letters: Improved pruning strategy for radial basis function networks with dynamic decay adjustment

Neurocomputing
Letters: Novelty detection with constructive probabilistic neural networks

Neurocomputing
Visual RBF network design based on Star Coordinates

Advances in Engineering Software
A comparative study on support vector machine and constructive RBF neural network for prediction of success of dental implants

CIARP'05 Proceedings of the 10th Iberoamerican Congress conference on Progress in Pattern Recognition, Image Analysis and Applications
A comparative study on machine learning techniques for prediction of success of dental implants

MICAI'05 Proceedings of the 4th Mexican international conference on Advances in Artificial Intelligence

Quantified Score

Hi-index	0.01

Visualization

Abstract

This letter proposes a constructive training method for radial basis function networks. The proposed method is an extension of the dynamic decay adjustment (DDA) algorithm, a fast constructive algorithm for classification problems. The proposed method, which is based on selective pruning and DDA model selection, aims to improve the generalization performance of DDA without generating larger networks. Simulations using four image recognition datasets from the UCI repository demonstrate the validity of the proposed method.