Comparative Analysis of Fuzzy Rule Interpolation Techniques Across Various Scenarios Using a Set of Benchmarks

Abstract

This paper presents a set of benchmarks to evaluate the performance of Fuzzy Rule Interpolation (FRI) methods under various challenging conditions. FRI methods are widely used for handling sparse fuzzy rule bases and reducing decision complexity. Despite lacking overlap with the antecedents of any rule in the rule bases, FRI can still produce a conclusion. To unify the requirements of FRI methods, several conditions have been proposed. Among these, the convex and normal fuzzy set condition and the Piece-wise linearity condition are the most common. In this paper, we introduce new benchmark scenarios for testing FRI methods. These benchmarks aim to serve as a reference for evaluating and comparing the accuracy and effectiveness of FRI methods. By using these benchmarks, researchers can compare different FRI methods and identify areas for improvement in the field of fuzzy inference.