z Shaeiri - Department of Electrical and Computer Engineering, Babol Nooshirvani University of Technology, Babol, Iran
r Ghaderi - Department of Electrical and Computer Engineering, Shahid Beheshti University of Technology, Tehran, Iran

Recognizing genes with distinctive expression levels can help in prevention, diagnosis and treatment of the diseases at the genomic level. In this paper, fast Global k-means (fast GKM) is developed forclustering the gene expression datasets. Fast GKM is a significant improvement of the k-meansclustering method. It is an incremental clustering method which starts with one cluster. Iteratively new clusters are added. Since in each epoch, all data points are examined for the next cluster center, it is believed that fast GKM attains a near global solution. In the gene expression clustering problem, geneswith significant differential expression levels, across the output disease classes, are important for the accurate classification of samples. Thus, a fuzzy entropy measure which is designated based on maximum within class and minimum between class relevance is exerted in to the search procedure ofthe fast GKM. As a result, the search procedure of the proposed method is conducted in such a way toprovide clusters which assembles the most discriminative genes closer to their centers. Therefore, capacity of the fast GKM which is its ability to find global clusters is managed in a profitable way. Todemonstrate the usefulness of the proposed method, three published microarray datasets are used: Leukemia, Prostate, and Colon. Classification results are found robust and accurate using three public classification methods: K-NN, SVM, and Naïve Bayesian

Gene Expression Data Clustering,Global K-means Algorithm,Fuzzy Entropy Measure