Stochastic gradient descent support vector clustering

Support-based clustering method has recently drawn plenty of attention because of its applications in solving the difficult and diverse clustering or outlier detection problem. Support-based clustering method undergoes two phases: finding the domain of novelty and doing cluster assignment. To find the domain of novelty, the training time given by the current solvers is typically quadratic in the size of the training dataset. It impedes the use of support-based clustering method for the large-scale datasets. In this paper, we propose applying Stochastic Gradient Descent framework to the first phase of support-based clustering for finding the domain of novelty in form of a half-space. The experiment established of the large-scale datasets shows that the proposed method offers comparable cluster solution quality to the baseline while being able to run 200 times faster.