NOW COMES THE TIME TO DEFUZZIFY NEURO-FUZZY MODELS

Fuzzy models present a singular Janus-face: On the one hand, they are knowledge-based software environments constructed from a collection of linguistic IF-THEN rules, and on the other hand, they realize nonline ar mappings which have interesting mathematical properties like ''low- order interpolation'' and ''universal function approximation''. Neuro- fuzzy basically provides fuzzy models with the capacity, based on the available data, to compensate for the missing human knowledge by an au tomatic self-tuning of the structure and the parameters. A first conse quence of this hybridization between the architectural and representat ional aspect of fuzzy models and the learning mechanisms of neural net works has been to progressively increase and fuzzify the contrast betw een the two Janus faces: readability or performance. (C) 1997 Elsevier Science B.V.