抄録
The purpose of this paper is to introduce the idea of an autonomous decentralized system to a pictorial pattern recognition based on an autoassociative memory and to examine its possibility.
Synergetic computing proposed by H. Haken is one useful method of the construction of the autoassociative memory. Since prototype pictures are treated as the vector data in this method, the two-dimensional informations would have been lost. So, we may be able to reform this system.
One of the examples that the patterns are formatted by the two-dimensional interaction is found in the chemical reactions like a B-Z reaction. The mathematical models of these have been studied by use of the reaction-diffusion equation.
In this paper we apply the differential equation (the reaction-diffusion equation) which describes the pattern formation from the viewpoint of “autonomous decentralization” concept to make an autoassociative memory. Here we compose the system so that it recalls the memories by self-organization of the best inner structure, by which we propose one of the examples that show the composition of the autonomous decentralized system.
