Genetic learning of accurate TS models based on local fuzzy prototyping.
Rafael Alcalá, Jesús Alcalá‐Fdez, Jorge Casillas, Óscar Cordón, Francisco Herrera · 2003
This work presents the use of local fuzzy prototypes as a first approximation to obtain accurate local semantics-based Takagi-Sugeno rules. A two-stage evolutionary algorithm considering the interaction between input and output variables has been developed. Firstly, it performs a local identification of prototypes, and then, a postprocessing stage is considered to refine them. The proposal has been tested with a real-world problem achieving good results.