An improved clustering algorithm based support vectors
Donghui Chen,Zhijing Liu,Zonghu Wang
2011-01-01
Journal of Computational Information Systems
Abstract:Support vector clustering (SVC) is an important clustering algorithm based on support vector machine (SVM) and kernel methods. SVC algorithm performed better than the other traditional clustering methods, such as a global optimum, treatment of data sets of arbitrary shape, no need for specifying the number of clusters, fewer parameters, and easy treatment of high dimensional data. SV clustering consists of two phases, training based support vector machine and labeling clusters. The former requires pricy computation by calculating Lagrange multipliers. The latter phase needs to consume much more computing time because of the calculation of the adjacency matrix and random sampling which incurs errors. In this paper, we propose an entropy-based algorithm to reduce the time of Lagrange multipliers. In the labeling piece, the idea is to cluster SVs firstly, and then construct a classifier based on labeled SVs, finally label other data using the classifier. Theoretical analysis and empirical evidence suggest that the improved clustering algorithm overcomes two bottlenecks in SVC problem, and when compared with common clustering methods, it does a good job in performance and efficiency. © 2011 Binary Information Press December, 2011.