On the cooperation of fuzzy neural networks via a coevolutionary approach

This paper describes an effort on the characterization of cooperative fuzzy neural networks (CFNNs). CFNNs encompass any conceptual or architectural aggregate in which two or more fuzzy neural networks (FNNs) work cooperatively for the accomplishment of high-level objectives. In such a context, the behavior of an FNN is influenced, by some means, by the behavior of its peers, and the performance of the whole group should contribute as complementary guidance for the training of the individual FNNs. A coevolutionary approach is presented as an auxiliary mechanism for the design and implementation of CFNNs. Implementation issues are described as a means to attest the applicability of the proposal