On some hierarchical clustering algorithms using kernel functions
Yuki Endo, Hideyuki Haruyama, Takayoshi OKUBO · 2005
Kernel functions are remarked in the field of clustering or pattern recognition. Using the functions, good results of clustering can be obtained for data with hard nonlinear distribution. But there is the problem that the calculation cost increases using the kernel functions. In this paper, new hierarchical clustering algorithms using kernel functions are proposed. Moreover, the availability of proposed algorithms is discussed through some numerical examples. The proposed methods do not include the kernel functions positively so that the methods can be introduced into the existing resources easily.