Enhanced anchor graph semi-supervised classification method

Abstract

The invention discloses an enhanced anchor graph semi-supervised classification method. The method comprises the steps of S100, preparing a data set; S200, extracting m anchor points from the data setprepared in the step S100 by adopting an anchor point extraction method to obtain an anchor point set U; S300, according to the anchor point set U obtained in the step S200, establishing a relationship between an anchor point in the anchor point set U or a node in the marked data set X1 and other nodes in the data set by using an anchor point-based probabilistic neighbor method to obtain a relationship matrix Z *; S400, constructing a graph structure by taking the anchor points in the anchor point set U obtained in the step S200 and the nodes in the marked data set X1 as nodes, and performinglabel propagation by utilizing an extended label propagation method; and S500, performing label propagation according to the relation matrix Z * obtained in the step S300 and the label matrix F * obtained in the step S400 to obtain a final classification result. The time complexity and the space complexity in the semi-supervised classification process can be reduced, and the efficiency is improved.

Description

technical field [0001] The present disclosure relates to data classification methods, in particular, to an enhanced semi-supervised classification method with anchor graphs (Enhanced Semi-supervised Classification with Anchor Graph, ESCAG). Background technique [0002] Graph-based semi-supervised learning provides an effective paradigm for modeling manifold structures in high-dimensional spaces that may exist in massive data sources, and it has been shown to be effective in propagating a limited number of initial labels to a large number of future datasets. Labeled data, requiring a low number of labeled samples, thus meeting the needs of many emerging applications, such as image annotation and information retrieval. However, most of the current graph-based semi-supervised learning methods focus on the accuracy of classification, and less research has been done on reducing the complexity of the method. As the number of data samples increases, the graph-based semi-supervise...

AI Technical Summary

Problems solved by technology

However, the anchor points of this method are more sensitive to noise; while associating nodes with a fixed number of anchor points, the adaptability is poor

Images