Towards a coevolutionary approach for interval type-2 fuzzy modeling

This paper presents a particular case for a possible approach to coevolutionary interval type-2 fuzzy modeling. Some principles derived from a well-founded coevolutionary method for type one fuzzy systems are used and other rising questions about the application of coevolution to type-2 fuzzy systems are proposed. A benchmark problem related to design interval type-2 fuzzy time forecasters for time series prediction is treated by means of cooperative coevolution. Statistical results show that interval type-2 fuzzy systems working as evolvable individuals could be appropriated for evolutionary fuzzy modeling, since better and more stable solutions are found in comparison to type-1 fuzzy systems.