Abstract: Rough set was applied to clustering method in view of soft kernel of support vector clustering(SVC). The kernel function was modified through introducing upper and lower boundary. During clustering, the algorithm can not only deal with boundary points and find soft clusters with arbitrary shapes, but also control softness of boundary region by interactively adjusting parameters. SVC based on rough set solved classification of uncertain boundary region without extra cost. The experimental results indicate that the method can deal with soft boundary effictively, proving its correctness.

Key words: computer software, clustering, support vector clustering(SVC), rough set, rough based SVC, Lagrange function