I. Introduction
FUZZY approximation often times relies on using dense rule bases and large numbers of antecedent variables and linguistic terms to attain high precision. Such practice, however, often leads to undesirable computational delay, storage and retrieval problem. One approach to alleviate this “curse of dimensionality” situation is to adopt sparse fuzzy rules, and then apply interpolation to extract conclusion for observation falling into regions not covered by the antecedents [1], [2]. A number of interpolation techniques have been proposed in this regards, ranging from point-by-point extraction [1], to solid cutting techniques [3], and to the conservation of relative fuzziness [4], [5]. It is worth noting that fuzzy interpolation is applicable not only for sparse rules, but also to dense ones during their buildup phase.