An evolutionary method to generate fuzzy rule-based system for classification problems with semantics order of language.

Hoàng Văn Thông, Nguyễn Cát Hồ, Nguyễn Văn Long


In this paper, we propose a method to design fuzzy rule-based systems for classification problems with reference to the idea proposed in [17] to generate initial rule-based system. The generation of rules in this method is based on evolutionary multi-objective optimization [10-15] and hedge algebra methods [1-8] and proposed a rule evaluation measure bases on the improvement of the rule evaluation measure proposed in [17]. The linguistic terms used to generate fuzzy rules are designed based on hedge algebra and the generic algorithms based on optimizing the fuzzy parameters of hedge algebra. The proposed method is tested on 9 typical problems published in the UCI [18] with high performance classification while ensuring the rule-based system easy understanding.


  • There are currently no refbacks.

Journal of Computer Science and Cybernetics ISSN: 1813-9663

Published by Vietnam Academy of Science and Technology