Rapid spectrum embedding and clustering method based on graph learning

Abstract

The invention discloses a rapid spectrum embedding and clustering method based on graph learning. According to the method, a bipartite graph is constructed through a neighbor method, wherein the bipartite graph comprises the relationship between original data points and anchor points and the relationship between the anchor points, and the low-dimensional representation of data is obtained throughthe rapid spectrum embedding of the bipartite graph; a self-adaptive neighbor graph structure is updated, the original bipartite graph structure is corrected by using the self-adaptive neighbor graphstructure, and a better bipartite graph structure of the data points and the anchor points is obtained by means of learning; and an adaptive neighbor graph structure with c connected domains finally is obtained through iterative updating; and a clustering result is obtained while an optimal graph structure is obtained. According to the algorithm provided by the invention, relatively good clustering precision is obtained on a plurality of reference data sets.

Description

technical field [0001] The invention relates to the technical fields of data mining and pattern recognition, in particular to a fast spectrum embedding clustering method based on graph learning. Background technique [0002] At present, some scholars have proposed an adaptive nearest neighbor spectral embedding clustering method. This method mainly has two steps. The first step is to perform spectral embedding on the data, and the second step is adaptive nearest neighbor clustering. The detailed process of this method is as follows: [0003] 1) Spectral embedding of the data [0004] The clustering result can be regarded as the mapping of the original data, and the mapping function is: [0005] Y=X T W+1b T (1) [0006] The objective function of spectral clustering is: [0007] [0008] s.t.F T F=I. (2) [0009] in, or L=I-D -1 W plus the linear embedding regular term, we can get [0010] [0011] s.t.F T F=I. (3) [0012] Bring formula (1) into formula (...

AI Technical Summary

Problems solved by technology

[0028] In the prior art, the self-adaptive nearest neighbor spectrum embedding clustering method is used to construct a bipartite graph through the relationship between the original data point and the anchor point to embed the label matrix, without considering the connection between the anchor points, and the information reflected by the graph structure is not enough whole

In the pursuit of high accuracy, the computational complexity is not well considered, so the application on data with high dimensionality or large amount of data does not perform well

Images