Abstract
A system that uses a Genetic Algorithm to specify the structure of a backpropagation type neural net is described. The GA can achieve significant improvements in performance over both random and fully connected nets. The efficacy of the evaluation procedure is demonstrated by showing that the system is able to eliminate a small penalty flag despite the high noise levels.
Cite
CITATION STYLE
Hancock, P. J. B., & Smith, L. S. (1991). GANNET: Genetic design of a neural net recognition for face. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 496 LNCS, pp. 292–296). Springer Verlag. https://doi.org/10.1007/BFb0029766
Register to see more suggestions
Mendeley helps you to discover research relevant for your work.
