Title :
Maintaining chaos in an associative chaotic neural network exhibiting intermittency
Author :
Adachi, Masaharu
Author_Institution :
Dept. of Electron. Eng., Tokyo Denki Univ., Japan
Abstract :
The paper presents an attempt to maintain chaos in an associative chaotic neural network, which exhibits intermittency without control. The network to be controlled is composed of 16 chaotic neurons with synaptic weights that are determined by a conventional auto-associative matrix to store three orthogonal patterns. The network shows intermittency with certain parameter values without control. In this paper, the network with the intermittency is controlled in order to maintain chaos in the network. The control is applied only when the state vector comes to the neighborhood of the point just before it falls into the laminar phase. Perturbations are applied so that the state vector may not fall into the laminar phase. An example of maintaining chaos with the control is shown in the paper.
Keywords :
chaos; content-addressable storage; neural nets; search problems; associative chaotic neural network; autoassociative matrix; chaos control method; chaotic neurons; intermittency control; laminar phase; orthogonal patterns; search problems; state vector; synaptic weights; Associative memory; Bifurcation; Chaos; Control systems; Educational institutions; Intelligent networks; Maintenance engineering; Neural networks; Neurons; Size control;
Conference_Titel :
Circuits and Systems, 2004. ISCAS '04. Proceedings of the 2004 International Symposium on
Print_ISBN :
0-7803-8251-X
DOI :
10.1109/ISCAS.2004.1329121