Search

Fuzzy clustering is superior to crisp clustering when the boundaries among the clusters are vague and ambiguous. However, the main limitation of both fuzzy and crisp clustering algorithms is their sensitivity to the number of potential clusters and/or their initial positions. Moreover, the comprehensibility of obtained clusters is not expertized, whereupon in data-mining applications, the discovered knowledge is not understandable for human users. To overcome these restrictions, a novel fuzzy rule-based clustering algorithm (FRBC) is proposed in this paper. Like fuzzy rule-based classifiers, the FRBC employs a supervised classification approach to do the unsupervised cluster analysis. It tries to automatically explore the potential clusters in the data patterns and identify them with some interpretable fuzzy rules. Simultaneous classification of data patterns with these fuzzy rules can reveal the actual boundaries of the clusters. To illustrate the capability of FRBC to explore the clusters in data, the experimental results on some benchmark datasets are obtained and compared with other fuzzy clustering algorithms. The clusters specified by fuzzy rules are human understandable with acceptable accuracy. [ABSTRACT FROM AUTHOR]

This paper considers the automatic design of fuzzy-rule-based classification systems from labeled data. The performance of classifiers and the interpretability of generated rules are of major importance in these systems. In past research, some genetic-based algorithms have been used for the rule learning process. These genetic fuzzy systems have utilized different approaches to encode rules. In this paper, we have proposed a novel steady-state genetic algorithm to extract a compact set of good fuzzy rules from numerical data (SGERD). The selection mechanism of this algorithm is nonrandom, and only the best individuals can survive. Our approach is very simple and fast, and can be applied to high-dimensional problems with numerical attributes. To select the rules having high generalization capabilities, our algorithm makes use of some rule- and data-dependent parameters. We have also proposed an enhancing function that modifies the rule evaluation measures in order to assess the candidate rules more effectively before their selection. Experiments on some well-known data sets are performed to show the performance of SGERD. [ABSTRACT FROM AUTHOR]