Fuzzy systems modeling in practice (Q5947595)

When applying fuzzy systems modeling, four problems may arise: generalization ability of the fuzzy model, the appearance of rules without conclusion in the extracted rule-base, noise effects on the modeling process and the influence of the amount of the used training data on the learning performance. The authors analyse and discuss these topics by modeling a linear functional relation using a basic learning algorithm. Depending on the characteristics of the learning algorithm and the structure of the fuzzy model, some problems can be eliminated or attenuated. Further, the importance of pre-processing of data is discussed.