Data-Driven Fuzzy Modeling: Transparency and Complexity Issues

Recently, the interest in data-driven approaches to the modeling of nonlinear processes has increased. Techniques based on fuzzy sets and rule-based systems have proven suitable mainly because of their potential to yield transparent models that are at the same time reasonably accurate. Many of the data-driven fuzzy modeling algorithms, however, aim primarily at good numerical approximation, while little attention is paid to the complexity and interpretability of the resulting model. This paper deals with the issues of complexity and transparency in rule-based fuzzy models obtained through various data-driven identification algorithms. They include grid partition approaches and tree construction algorithms, fuzzy clustering and nonlinear parameter-optimization methods. Different rule base reduction techniques are addressed as well.