Extended Kernel Self-Organizing Map Clustering Algorithm

Abstract

The self-Organizing Map allows to visualize the underlying structure of high dimensional data. However, the original relies on the use of Euclidean distances which often becomes a serious drawback for number of real problems. Donald and others map the data in input space into a high 2-dimension feature space, here SOM algorithm are performed. However, its disadvantage lies in lack of direct descriptions about the clustering’s center and result .In this paper, we extend of SOM, a novel kernel SOM algorithm is proposed from energy function. The idea of kernel Self-Organizing Map is applied to kernel trick. The inner product of the mapping value of the original data in feature space is replaced by a kernel function, the winner neuron and weights of each neuron can be initialized and updated by kernel Euclidean norm in the feature space. This trick resolve the non-liners can’t clustering in the input space and can’t direct descriptions about the clustering’s center and result. In this paper, some data are applied to test KSOM and SOM algorithm,The result of the experiments show KSOM algorithm has better performance than SOM.