Compact and transparent fuzzy models and classifiers through iterative complexity reduction

In our previous work we showed that genetic algorithms (GAS) provide a powe rful tool to increase the accuracy of fuzzy models for both systems modelin g and classification. In addition to these results, we explore the GA to En d redundancy in the fuzzy model for the purpose of model reduction. An aggr egated similarity measure is applied to search for redundancy in the rule b ase description. As a result, we propose an iterative fuzzy identification technique starting with data-based fuzzy clustering with an overestimated n umber of local models. The GA is then applied to find redundancy among the local models with a criterion based on maximal accuracy and maximal set sim ilarity. After the reduction steps, the GA is applied with another criterio n searching for minimal set similarity and maximal accuracy. This results i n an automatic identification scheme with fuzzy clustering, rule base simpl ification and constrained genetic optimization with low-human intervention. The proposed modeling approach is then demonstrated for a system identific ation and a classification problem. Results are compared to other approache s in the literature. Attractive models with respect to compactness, transpa rency and accuracy, are the result of this symbiosis.