Spectral techniques and soft computing

Soft Computing denotes a set of paradigma related to cognitive modelling, which in the last years have been intensively studied, since important synergy effects among members of this set have been disclosed. Because of this, Soft Computing has emerged as an environment to effectively work with real world complex problems. Fuzzy Logic, Genetic Algorithms and Neural Networks are possibly the best known representatives of Soft Computing. In this paper we show how Spectral Techniques may help to further study these subjects or to improve their performance. The name Spectral Techniques comprises Methods and Applications based on Abstract Harmonic Analysis. A preliminary version of this work was presented at the 5th. International Workshop on Spectral techniques, Beijing, 1994. The author gladly acknowledges support received then from the Volkswagen Foundation, Germany