Fuzzy neural networks for learning fuzzy IF-THEN rules

This study is dedicated to developing a fuzzy neural network with linguisti c teaching signals. The proposed network, which can be applied either as a fuzzy expert system or a fuzzy controller, is able to process and learn the numerical information as well as the linguistic information. The network c onsists of two parts : (1) initial weights generation and (2) error back-pr opagation (EBP)-type learning algorithm. In the first part, a genetic algor ithm (GA) generates the initial weights for a fuzzy neural network in order to prevent the network getting stuck to the local minimum. The second part employs the EBP-type learning algorithm for fine-tuning. In addition, the unimportant weights are eliminated during the training process. The simulat ed results do not only indicate that the proposed network can accurately le arn the relations of fuzzy inputs and fuzzy outputs, but also show that the initial weights from the GA can coverage better and weight elimination rea lly can reduce the training error. Moreover, real-world problem results sho w that the proposed network is able to learn the fuzzy IF-THEN rules captur ed from the retailing experts regarding the promotion effect on the sales.