Study on Fuzzy Classifier Based on Genetic Algorithm Optimization

Most of the fuzzy classifiers are created by fuzzy rules based on transcendent knowledge or expert's knowledge. In mountains of the existing data, it is difficult to obtain transcendent knowledge and then more difficult to obtain fuzzy rules. To solve this problem, a new way for creating fuzzy classifier based on GA for a classification problem of quantitative attribute is proposed in this paper, which consists of a set of fuzzy rules generated according to the attribute of the classified objects, then choosing the optimal fuzzy rules subset forming the classifier by the genetic algorithm to reduce the number of rules and improve the classification accuracy. The result of data simulation was applied to the iris with better effects.