A novel distributed clustering algorithm based on OCSVM

In this paper, aiming to accelerate the clustering method of Support Vector Machine for large-scale dataset, we present a novel method for clustering inspired by the OCSVM and the Multi-Agent framework, in which the data are divided to different agents, and the global clustering result can be generalized from the agents. Moreover, according to the One-Class Support Vector Machine theory, this paper conducts a study on the setting of parameter involved in the clustering algorithm. Lastly, the experimental results indicate that the clustering method we proposed in this paper is more efficient for large dataset.