Fuzzy System Based on Class Association Rules

Abstract

Fuzzy system has been proved to be a universal approximator, yet the curse of dimensionality is still the unsolved problem for it. Class association rules (CARs) are interesting and frequent patterns derived from data through adapted Apriori algorithm. Using CARs to build the fuzzy rule base of fuzzy system can solve the curse of dimensionality problem effectively. Thus, a novel fuzzy system based on CARs is proposed in this paper. The process of how to build the fuzzy system and its whole execution are presented in detail. Furthermore, comparative experiments are also made to prove the effectiveness of the presented strategy.