Neural Network Based Fuzzy Systems Design
Yaochu Jin · Studies in fuzziness and soft computing · 2003
Fuzzy rules are able to represent knowledge that is understandable to human beings. Traditional fuzzy rules are usually generated from expert knowledge and human heuristics. This gives rise to two main drawbacks of traditional fuzzy systems for modeling and control. First, the fuzzy rules are very simple and the performance of the fuzzy system is low. In most cases, fuzzy memberships are determined heuristically and therefore, the knowledge represented by the fuzzy rules may be shallow. Second, it is difficult to efficiently extract fuzzy rules for high-dimensional systems due to the limitation of human thinking. In particular, traditional fuzzy systems are lack of learning capability, whereas learning is one of the most important features of intelligent systems. These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.