HYBRID METHOD OF BUILDING FUZZY SYSTEMS BASED ON ISLAND MODEL
We present a hybrid method for identifying the structure and parameters of fuzzy systems based on the island model. The following algorithms perform structure identification: bee colony algorithm for rules generating, rule base generation algorithm through exception of inefficient rules, the rule base generation algorithm with a given structure. Fuzzy systems parameters are optimized by adaptive evolutionary strategy and the bee colony algorithm for parameters identification and by least squares method. The interaction of these algorithms is based on the island model. The performance analysis based on real dataset shows the effectiveness of the proposed method.
Keywords: fuzzy system, structure generation, island model, bee colony algorithms, evolution strategies, least squares method
