Attributes regrouping in fuzzy rule based classification systems

In fuzzy rule based classification systems, a high number of predictive attributes leads to an explosion of the number of generated rules and can affect the learning algorithm precision. Thus, the increase of the number of features can degrade the predictive capacity of the fuzzy rule based classification systems. In this article, we propose a supervised learning method by automatic generation of fuzzy classification rules, entitled SIFCO. This method is adapted to the representation and the prediction of high-dimensional pattern classification problems. This characteristic is obtained by studying the attributes regrouping by correlation research among the training set elements. This approach, checked experimentally, guarantees an important reduction of rules number without altering too much good classification rates. Several experiences were carried out on various data in order to compare SIFCO with other rules based learning methods.