In this paper, a new method which can alter the values of the parameters in neural networks is proposed in order to enhance the representation abilities of the networks. As an example, a fuzzy reference network is used to modify the parameters in this article, even though any kind of networks such as radial basis function networks and neural networks can be adopted to realize varying parameters. From simulations, it is shown that the network using the proposed method is better than the conventional neural networks in terms of representation abilities of the networks.
|Number of pages
|Research Reports on Information Science and Electrical Engineering of Kyushu University
|Published - 1998 Sept 1
ASJC Scopus subject areas
- General Computer Science
- Electrical and Electronic Engineering