Rule base normalization in Takagi-Sugeno ensemble

The paper shows a method for obtaining one fuzzy rule base from the ensemble of neuro-fuzzy Takagi-Sugeno systems. Ensembles are one of methods for improving classification accuracy. In case of such an ensemble we obtain a set of classifiers with separate rule bases. All these fuzzy rules cannot be treated as one set of rules. The paper proposes a method for normalizing each rule base during learning. Thanks to this, all rule bases have similar overall activation levels and we can treat fuzzy rules coming from different systems as rules from the same (single) system.