A framework for density weighted kernel fuzzy c-means on gene expression data

Clustering techniques have been widely used for gene expression data analysis. However, noise, high dimension and redundancies are serious issues, making the traditional clustering algorithms sensitive to the choice of parameters and initialization. Therefore, the results lack stability and reliability. In this paper, we propose a novel clustering method, which utilizes the density information in the feature space. A cluster center initialization method is also presented which can highly improve the clustering accuracy. Finally, we give an investigation to the parameters selection in Gaussian kernel. Experiments show that our proposed method has better performance than the traditional ones.