Normal forms in BL-algebra and their contribution to universal approximation of functions

This paper continues the investigation of approximating properties of generalized normal forms in fuzzy logic. The problem is formalized and solved algebraically. Normal forms are considered in two variants: infinite and finite. It is proved that infinite normal forms are universal representation formulas whereas finite normal forms are universal approximation formulas for any L-valued function where L is a support set of any complete BL-algebra. The estimation of the quality of approximation is suggested. Moreover, functions which can be precisely represented by the discrete normal forms are considered.

[1]  Irina Perfilieva,et al.  Fuzzy Logic Normal Forms for Control Law Representation , 1999 .

[2]  J. Buckley Sugeno type controllers are universal controllers , 1993 .

[3]  Irina Perfilieva,et al.  Normal forms for fuzzy logic functions and their approximation ability , 2001, Fuzzy Sets Syst..

[4]  V. Novák,et al.  Mathematical Principles of Fuzzy Logic , 1999 .

[5]  Petr Hájek,et al.  Metamathematics of Fuzzy Logic , 1998, Trends in Logic.

[6]  Robert Babuska,et al.  Fuzzy algorithms for control , 1999 .

[7]  Juan Luis Castro,et al.  Fuzzy systems with defuzzification are universal approximators , 1996, IEEE Trans. Syst. Man Cybern. Part B.

[8]  Vladik Kreinovich,et al.  Normal Forms for fuzzy Logic - an Application of Kolmogorov's Theorem , 1996, Int. J. Uncertain. Fuzziness Knowl. Based Syst..

[9]  B. Kosko Fuzzy systems as universal approximators , 1992, [1992 Proceedings] IEEE International Conference on Fuzzy Systems.