Title :
Separable recursive training algorithms for feedforward neural networks
Author :
Asirvadam, Vijanth S. ; McLoone, Seán F. ; Irwin, George W.
Author_Institution :
Sch. of Electr. & Electron. Eng., Queen´´s Univ., Belfast, UK
fDate :
6/24/1905 12:00:00 AM
Abstract :
Novel separable recursive training strategies are derived for the training of feedforward neural networks. These hybrid algorithms combine nonlinear recursive optimization of hidden-layer nonlinear weights with recursive least-squares optimization of linear output-layer weights in one integrated routine. Experimental results for two benchmark problems demonstrate the superiority of the new hybrid training schemes compared to conventional counterparts
Keywords :
feedforward neural nets; iterative methods; learning (artificial intelligence); least squares approximations; optimisation; benchmark problems; feedforward neural networks; hidden layer nonlinear weights; hybrid algorithms; integrated routine; linear output-layer weights; nonlinear recursive optimization; recursive least-squares optimization; separable recursive training algorithms; Continuous-stirred tank reactor; Equations; Feedforward neural networks; Intelligent control; Intelligent networks; Intelligent systems; Least squares methods; Neural networks; Neurons; Vectors;
Conference_Titel :
Neural Networks, 2002. IJCNN '02. Proceedings of the 2002 International Joint Conference on
Conference_Location :
Honolulu, HI
Print_ISBN :
0-7803-7278-6
DOI :
10.1109/IJCNN.2002.1007667
