Lamarckian evolution of associative memory

Author

Imada, Akira ; Araki, Keijiro

Author_Institution

Grad. Sch. of Inf. Sci., Nara Inst. of Sci. & Technol., Ikoma, Japan

fYear

1996

Firstpage

676

Lastpage

680

Abstract

There has been a lot of research which applies evolutionary techniques to layered neural networks. However, their application to Hopfield neural networks remain few so far. We apply genetic algorithms to a fully connected Hopfield associative memory model. In an earlier paper, we reported that random weight matrices were evolved to store a number of patterns only by means of a simple genetic algorithm (A. Imada and K. Araki, 1995). We propose that the storage capacity can be enlarged by incorporating Lamarckian inheritance to the genetic algorithm

Keywords

Hopfield neural nets; content-addressable storage; genetic algorithms; inheritance; Hopfield neural networks; Lamarckian evolution; Lamarckian inheritance; associative memory; evolutionary techniques; fully connected Hopfield associative memory model; genetic algorithms; layered neural networks; random weight matrices; simple genetic algorithm; storage capacity; Artificial neural networks; Associative memory; Computer simulation; Genetic algorithms; Hopfield neural networks; Information science; Neural networks; Neurons; Organisms; Transfer functions;

fLanguage

English

Publisher

ieee

Conference_Titel

Evolutionary Computation, 1996., Proceedings of IEEE International Conference on

Conference_Location

Nagoya

Print_ISBN

0-7803-2902-3

Type

conf

DOI

10.1109/ICEC.1996.542682

Filename

542682