Hierarchical genetic fuzzy systems (Q1602475)

Fuzzy rule-based systems use inference rules to express expert knowledge, but unfortunately, in a number of applications, the amount of knowledge required may not be easily available, and the tuning of the model parameters can be very time consuming. In such cases, evolutionary computation approaches may be helpful, and here the authors propose a class of hierarchical genetic fuzzy systems based on a hierarchical evolutionary technique scheme, combined with a least squares method. After the introduction and the motivation, the hierarchical evolutionary process with the coding scheme, the evolutionary algorithm and the genetic operators are described. Then the authors explain how the least-squares method is used to obtain the consequent parameters as well as the pruning process. Simulation results are displayed, and lastly one discusses the computational cost necessary to use this technique.