ON THE OPTIMUM DESIGN OF CLUSTER STRUCTURES BY USING A GENETIC ALGORITHM

In discussing self-organizing neural networks, to some extent a large- scale network is assumed in order to achieve generality and adaptabili ty. This paper discusses an optimal structurization method for a nonli near network, based on a self-organizing algorithm with a two-layer st ructure. The basic structure of the network combines a self-organizing layer and a single-layer perceptron network. In the learning stage, b oth the self-organizing algorithm and the supervised learning algorith m are applied for each datum. Because of this structure, the network a chieves highly precise signal processing based on learning, i.e., self -organization and supervised learning. A previous paper used, this kin d of network in the estimation df spectra. However, among problems tha t remained were the long processing time required in the learning stag e due to the formation of unnecessary cluster nodes, and the fact that unnecessary nodes sometimes degrade estimation performance. From this perspective, it seems important in achieving a high-speed and highly precise system, to optimize cluster structure by eliminating unnecessa ry nodes. This paper presents a method for optimal network design base d on a genetic algorithm that can attain a smaller scale network with higher precision than any conventional network. It is shown that perfo rmance is better than for a conventional network. The network is appli ed ti, the spectra estimation problem to demonstrate its effectiveness . (C) 1998 Scripta Technica, Electron Comm Jpn Pt 2, 81(3): 53-63, 199 8.