| Summary: | 碩士 === 國立臺南大學 === 資訊教育研究所碩士班 === 93 === In recent years, there has been an increasing interest in the fusion of neural networks and fuzzy logic. Most of the existing fuzzy neural networks (FNN) models have been proposed to implement different types of single-stage fuzzy reasoning mechanisms. Single-stage fuzzy reasoning, however, is only the most basic among a human being’s various types of reasoning mechanisms. Syllogistic fuzzy reasoning is essential to effectively build up a large scale system that is more complicated and relatively closed to human being’s reasoning mechanisms. The cascaded fuzzy neural network (CFNN) model combined with syllogistic fuzzy reasoning and neural networks successfully. This model can learn something efficiently. However, the rule selection is not regular and do not have a specified rule yet. They used genetic algorithms to establish the rules. So, in this paper we proposed a new method that is applied the back-propagation method to update weights and selects the rule with the maximum absolute weight. We also called this method as “select the maximum effect factor method”. Based on research results, our methods are more efficient and accurate than other methods. Keywords：Cascaded fuzzy neural network (CFNN)、Syllogistic fuzzy reasoning、select the maximum effect factor method
|
|---|
