An application of genetic algorithms to evolve Hopfield type optimum network architectures for object extraction

Genetic Algorithms (GAS) have heen used l o evolve I iop j ie ld type oplimuni neural network archileciures for object background classijicatioii. Each chromosome of the GA represenis an archiieciure. The initial population is set randomly. The eirerqy t i d l l f : a i the converged state of each network is taken as its fitness. The best chromosome of I h e j i n a l generotion is taken io be the optimum network configuraiion. Tlie evolved networks are fo,iintl l o have less (compared lo t h e corresponding fixet’ fully connected version) connectivity f o r providing coniparnble oulp.uts.

[1]  Sankar K. Pal,et al.  Object Background Classification Using Hopfield Type Neural Network , 1992, Int. J. Pattern Recognit. Artif. Intell..

[2]  L. Darrell Whitley,et al.  Genetic algorithms and neural networks: optimizing connections and connectivity , 1990, Parallel Comput..

[3]  J J Hopfield,et al.  Neurons with graded response have collective computational properties like those of two-state neurons. , 1984, Proceedings of the National Academy of Sciences of the United States of America.