Evolutionary approaches to the learning of fuzzy rule-based classification systems
Óscar Cordón, Francisco Herrera, María José del Jesús · 1999
Introduction The construction of a Classification System has been tackled many times using fuzzy rules as knowledge representation tool ([4], [7] and [12]). The resulting systems are called Fuzzy Rule-Based Classification Systems (FRBCSs) and their success is fundamentally due to two reasons. On the one hand, the use of fuzzy logic makes possible the treatment of imprecise, uncertain or incomplete information, very common in real classification problems. On the other hand, rules represent the knowledge in a comprehensible form for those who will use the Classification System, making possible the use of this kind of systems as a tool in decision making processes. All in all, the fuzzy rules allow us to work in a transparent way in a feasible computer environment with the opaque classification schemes often used by human beings for these kinds of tasks [81]. The design of an FRBCS by means of a supervised learning process, which describes with higher possible precision the clas