Cooperative fuzzy rulebase construction based on a novel fuzzy decision tree

Abstract

Fuzzy Inference Systems (FIS) are much considerable due to their interpretability and uncertainty factors. Hence, Fuzzy Rule-Based Classifier Systems (FRBCS) are widely investigated in aspects of construction and parameter learning. Also, decision trees are recursive structures which are not only simple and accurate, but also are fast in classification due to partitioning the feature space in a multi-stage process. Combination of fuzzy reasoning and decision trees gathers capabilities of both systems in an integrated one. In this paper, a novel fuzzy decision tree (FDT) is proposed for extracting fuzzy rules which are both accurate and cooperative due to dependency structure of decision tree. Furthermore, a weighting method is used to emphasize the corporation of the rules. Finally, the proposed method is compared with a well-known rule construction method named SRC on 8 UCI datasets. Experiments show a significant improvement on classification performance of the proposed method in comparison with SRC.