Abstract
The architecture and learning scheme of a new fuzzy logic system implemented in the framework of a neural network is proposed. The proposed network can construct its rules and optimize its membership functions by data training. Both error back propagation and least-squares estimation are applied to the learning scheme. The convergence of training is expected to be faster because the least-squares estimation is applied to the estimation of the consequence parameters of the system and back propagation is applied only to the estimation of premise parameters. Due to the proposed architecture, even a high-order fuzzy system can be implemented with this learning scheme. In our simulation, the proposed network is employed for modeling a nonlinear function, an operator's control of a chemical plant, and stock prices.
Get full access to this article
View all access options for this article.