A Constructive Method for Building Fuzzy Rule-Based Systems
Ignacio Requena, Armando J. Blanco, Miguel Delgado‐Rodríguez · International Journal of Uncertainty Fuzziness and Knowledge-Based Systems · 2003
This paper proposes a new method for identifying unknown systems with Fuzzy Rule-Based Systems (FRBSs). The method employs different methodologies from the discipline of Soft Computing (Artificial Neural Networks, Fuzzy Clustering) and follows a three-stage process. Firstly, the structure of the FRBS rules is determined using a feature selection process. A fuzzy clustering procedure is then used to establish the number of fuzzy rules. In the third step, the fuzzy membership functions are constructed for the linguistic labels. Finally, the empirical performance of the algorithm is studied by applying it to a number of classification and approximation problems.