Computational & Technology Resources
an online resource for computational,
engineering & technology publications
Civil-Comp Proceedings
ISSN 1759-3433
CCP: 38
ADVANCES IN COMPUTATIONAL STRUCTURES TECHNOLOGY
Edited by: B.H.V. Topping
Paper II.5

A Neural Network Trained by Genetic Algorithm

W.M. Jenkins

Department of Civil Engineering, University of Leeds, Leeds, United Kingdom

Full Bibliographic Reference for this paper
W.M. Jenkins, "A Neural Network Trained by Genetic Algorithm", in B.H.V. Topping, (Editor), "Advances in Computational Structures Technology", Civil-Comp Press, Edinburgh, UK, pp 77-84, 1996. doi:10.4203/ccp.38.2.5
Abstract
Genetic algorithm (GA) - based method of training a neural network is proposed as an alternative to the usual "back-propagation" (BP) algorithm. Considerable changes are needed in the usual design of the GA due to the intrinsic need to provide the algorithm with specified discrete values of the variables (weights). In order to give the GA freedom to select weights from a more-or-less unlimited range of values, the decoding of the binary strings is accompanied by a progressive "shift" of the centre of the range of positive/negative values provided for selection by the algorithm.

The proposed method is applied to a simple beam bending moment situation and to an approximate analysis of a structural grillage. The results obtain by using the GA with shift, are compared with those from a Conventional BP training and with the "exact" results.

purchase the full-text of this paper (price £20)

go to the previous paper
go to the next paper
return to the table of contents
return to the book description
purchase this book (price £82 +P&P)