Dynamical Associative Memory based on Synchronization in Chaotic Neural Network : Proposal for the Scenario via Spatio-Temporal Symmetry Recovering Bifurcation

概要

論文の詳細を見る
In this paper, we have investigated dynamical bases for the phenomena of "dynamical associative memory" in a chaotic neural network. By the term dynamical associative memory, we mean a set of the memories, for instance, {Ai| i=1,2,...,M} , which have been stored in the neural network are retritved successively and irregularly. The phenomena are interesting, because they provide a possibility of linking the mamories stored in a neural network in general as already pointed out by some researchers, but also they may provide very typical systems in which the system symmetry works very nicely for acquring a new function of information processings (=dynamical associative memory). After summarizing our experimental results on a chaotic neural network for dynamical associative memory obtained by using the orthogonal stored patterns, where the important findings are the phenomena of chaotic self-switching between stored memories, we propose a theoretical frame for understanding the phenomena of chaotic self-switching through the notion : Spatio-Temporal Symmetry Recovering Bifurcation
社団法人電子情報通信学会の論文
1997-02-06