This paper discusses new method for extracting fuzzy rules directly from numerical input-output data. The method was first developed for pattern classification and then was extended for function approximation. The fuzzy rules with variable fuzzy regions are defined by activation hyperboxes wich show the existence region of data for a class and inhibition hyperboxes which inhibit the existence of data for that class. These rules are extracted from numerical data by recursively resolving overlaps between two classes. For pattern classification, the method is applied to recognition of numerals of license plates and its performance is compare with neural networks. For function approximation, the approximation accuracy of the fuzzy system is compared with that of neural networks using an operation learning application of a water purification plant.
展开▼
