Multi-label classification method based on flow pattern matrix completion

Abstract

The invention discloses a multi-label classification method based on flow pattern matrix completion, and the method gives consideration to an inter-feature flow pattern hypothesis in a process of multi-label learning carried out through employing matrix completion, and can improve the multi-label classification algorithm recognition efficiency more remarkably. Compared with the prior art, the method provided by the invention is greatly improved at three multi-label learning indexes (hamming loss, mean precision and coverage).

Description

technical field [0001] The invention belongs to the technical field of label classification, and in particular relates to the design of a multi-label classification method based on flow pattern matrix completion. Background technique [0002] Multi-label classification problems are ubiquitous, and they are more practical in practical applications than ordinary two-class classification problems. For example, when classifying and labeling pictures on the Internet, since there are often multiple targets on the picture (the same picture may belong to multiple categories at the same time), it is difficult to classify it into two categories using ordinary two-class classification methods , and it is more appropriate to classify as multi-label. For example, there are 5 types of targets in the picture collection, then the label of a picture with 3 targets may be (0,1,0,1,1), indicating that this The pictures belong to the second, fourth and fifth categories. [0003] Matrix comple...

AI Technical Summary

Problems solved by technology

[0004] The purpose of the present invention is to solve the problem that the matrix completion technology ignores the nature of the smoothness assumption between features in the prior art, and proposes a multi-label classification method based on streamline matrix completion

Images