谱聚类是从图论中演化出来的算法，后来在聚类中得到了广泛的应用。它的主要思想是把所有的数据看做空间中的点，这些点之间可以用边连接起来。距离较远的两个点之间的边权重值较低，而距离较近的两个点之间的边权重值较高，通过对所有数据点组成的图进行切图，让切图后不同的子图间边权重和尽可能的低，而子图内的边权重和尽可能的高，从而达到聚类的目的。(Spectral clustering is an algorithm evolved from graph theory, and has been widely used in clustering. Its main idea is to look at all the data as points in space, which can be joined by edges. The edge weight between two point distance value is lower, and the edge weight between two points near the higher values of map, through the composition of all data points, so that different sub graph edge weights and as low as possible, and edge weights and sub graph as far as possible, so as to achieve the purpose of clustering.)