Title :
The study of linear radial basis function network
Author_Institution :
Lab. of Neural Network, Chinese Acad. of Sci., Beijing, China
Abstract :
A special kind of RBF network, linear radial basis function network (LRBFN), was proposed. For LRBFN, the weight of a neuron has a relatively clear relationship with the distance between its center and the discriminant surface. Aimed to classify two kinds of samples, a two-stages training algorithm for LRBFN was given. In the first stage, the distance matrix of all samples is used to construct the discriminant surface; in the second stage, the neuron centers for LRBFN are reduced according to the weights in the previous stage. The experimental results show that LRBFN acts very well to solve classification problems.
Keywords :
learning (artificial intelligence); matrix algebra; pattern classification; radial basis function networks; discriminant surface construction; distance matrix; linear RBF network; linear radial basis function network; sample classification; training algorithm; Electronic mail; Neural networks; Neurons; Pattern recognition; Radial basis function networks;
Conference_Titel :
Intelligent Control and Automation, 2004. WCICA 2004. Fifth World Congress on
Print_ISBN :
0-7803-8273-0
DOI :
10.1109/WCICA.2004.1341924
