Survey on Spectral Clustering Algorithms

Spectral clustering algorithms are newly developing technique in recent years.Unlike the traditional clustering algorithms,these apply spectral graph theory to solve the clustering of non-convex sphere of sample spaces,so that they can be converged to global optimal solution.In this paper,the clustering principle based on graph theory is first introduced,and then spectral clustering algorithms are categorized according to rules of graph partition,and typical algorithms are studied emphatically,as well as their advantages and disadvantages are presented in detail.Finally,some valuable directions for further research are proposed.