期刊名称：International Journal of Electronics, Communication and Soft Computing Science and Engineering
印刷版ISSN：2277-9477
出版年度：2014
卷号：3
期号：Special
出版社：IJECSCSE
摘要：Fuzzy clustering is superior to crisp clustering when theboundaries among the clusters are vague and ambiguous.However, the main limitation of both fuzzy and crisp clusteringalgorithms is their sensitivity to the number of potential clustersand/or their initial positions. Moreover, the comprehensibility ofobtained clusters is not expertized, whereupon in dataapplications, the discovered knowledge is not understandable forhuman users. To overcome these restrictions, a novel fuzzy rulebased clustering algorithm (FRBC) is proposed. Like fuzzy rulebased classifiers, the FRBC employs a supervised classificationapproach to do the unsupervised cluster analysis. It tries toautomatically explore the potential clusters in the data patternsand identify them with some interpretable fuzzy rules.Simultaneous classification of data patterns with these fuzzy rulescan reveal the actual boundaries of the clusters.hard clustering, fuzzy clustering that allows overlap betweenclusters is able to provide a more accurate and natural descriptionof the underlying structure of real-world data. The same asmeans, most existing studies on fuzzy clustering, including thewell-known fuzzy c-means (FCM) and some recently proposedapproaches are deals with vector-based data, of which each objectis represented as a vector in some feature space.shown fuzzy c-means (FCM) clustering to be a powerful tool topartition samples into different categories. However, the objectivefunction of FCM is based only on the sum of distances of samplesto their cluster centers, which is equal to the trace of the withincluster scatter matrix. Hence a clustering algorithm based on bothwithin and between-cluster scatter matrices, extended from lineardiscriminant analysis (LDA), and its application to anunsupervised feature extraction (FE). LDA methods comprisebetween and within- cluster scatter matrices modified from thebetween- and within-class scatter matrices of LDA.
关键词：FCM Clustering; Fuzzy Rule Based Clustering;Linear Discriminant Analysis (LDA).
