Ramp-based twin support vector clustering

Traditional plane-based clustering methods measure the within-cluster or between-cluster scatter by linear, quadratic or some other unbounded functions, which are sensitive to the samples far from the cluster center. This paper introduces the ramp functions into plane-based clustering and proposes a ramp-based twin support vector clustering (RampTWSVC). RampTWSVC is very robust to the samples far from the cluster center, because its within-cluster and between-cluster scatters are measured by the bounded ramp functions. Thus, it is easier to find the intrinsic clusters than other plane-based clustering methods. The nonconvex programming problem in RampTWSVC is solved efficiently through an alternating iteration algorithm, and its local solution can be obtained in a finite number of iterations theoretically. In addition, its nonlinear manifold clustering formation is also proposed via a kernel trick. Experimental results on several benchmark datasets show the better performance of our RampTWSVC compared with other plane-based clustering methods.