Approximation theory of fuzzy systems based upon genuine many-valued implications -- MIMO cases (Q1855479)

It has already constructively been proved [ibid. 130, 147-157 (2002; Zbl 1017.93056), see the review above] that the SISO fuzzy systems based upon genuine many-valued implications are universal approximators (they are called Boolean type fuzzy systems in this paper). These results are extended to multi-input-multi-output (MIMO) fuzzy systems based on genuine many-valued implications. In Section 2, the authors review the inference methods of multi-rules fuzzy systems for MIMO fuzzy systems. They discuss the approximation properties of MIMO fuzzy systems based on R-implications in Section 3 and those of MIMO fuzzy systems based on S- and QL-implications in Section 4. In Section 5, they illustratively design two fuzzy systems based on genuine many-valued implications for two given functions with a required approximation accuracy and compare the optimal fuzzy rules to those conjunctive fuzzy systems. Conclusions are made in the last section. One appendix for the proofs of the propositions in the paper is included.