Automatic training of generalized min-max classifiers

Among fuzzy classifiers, min-max networks have the advantage to be trained in a constructive way, by a simple learning procedure. The classification strategy of Simpson´s min-max classifier (1992) consists in covering the training data with hyperboxes constrained to have their boundary surfaces parallel to the coordinate axes of the chosen reference system. In order to obtain a more accurate data coverage, it is possible to adopt a new classification model which allows to arrange the hyperboxes orientation along any direction of the data space. The training algorithm is based on the ARC/PARC technique, which already yields better performances with respect to the original Simpson´s algorithm. Although the most important feature of a classifier is its generalization capability, the effectiveness of a training procedure is strictly related to its automation degree. A low automation degree can be a serious drawback for a classification system, since it can prevent an unskilled user from successfully generate an acceptable model. From this point of view, a learning procedure should not depend on any critical parameter. The automation degree of the new classification system is evaluated in the paper