Optimization of interval type-2 fuzzy logic systems using tabu search algorithms

Abstract

In this paper, a new method to optimise interval type-2 fuzzy logic systems (IT2FLSs) using two forms of tabu search (TS) is presented. TS algorithms search for the best configuration of the interval type-2 fuzzy logic system (IT2FLS) rule-base membership functions parameters. Directed tabu search (DTS), which uses pattern search to control TS moves, and short-term tabu search (STS) have been applied with IT2FLS to a classification problem of two benchmark datasets. The main objective is to better improve the configuration of (IT2FLSs) using an intelligent directed search rather than a random-based search. The comparison results show that DTS outperformed the default TS by a noticeable difference. This observation reveals the importance of using guided search moves rather than using a randomized search direction especially with complex large intelligent systems as in (IT2FLSs).