Semi-supervised clustering method and semi-supervised clustering system based on nonnegative matrix factorization

Abstract

The invention discloses a semi-supervised clustering method based on nonnegative matrix factorization, which comprises the steps of carrying out nonnegative matrix factorization projection on an original data matrix, and acquiring a low-dimension approximate matrix, which has both neighborhood preserving and similarity preserving, of original data; carrying out clustering on the low-dimension approximate matrix of the original data by using an algorithm receiving parameter K to acquire a clustering result; and evaluating the clustering result by using two types of evaluation standards of precision and mutual information. The semi-supervised clustering method disclosed by the invention is based on nonnegative matrix factorization, not only considers neighborhood preserving of the original data, but also considers the consistency of similarity in an original space and a low-dimension manifold subspace, so that the clustering performance is enabled to be greatly improved when prior information is great in amount, and the clustering performance can still be well preserved when the prior information is little. The invention further discloses a semi-supervised clustering system based on nonnegative matrix factorization.

Description

technical field [0001] The invention relates to the technical field of cluster analysis, in particular to a semi-supervised clustering method and system based on non-negative matrix decomposition. Background technique [0002] In recent years, non-negative matrix factorization technology has played a very important role in pattern recognition and artificial intelligence. There has been research showing evidence based on partial representations of the human brain, both psychologically and physiologically. Non-negative matrix representations have inherent advantages in learning partial representations similar to faces, images, and documents. Also, in many problems such as information retrieval, computer vision, and pattern recognition, the data are characterized by high dimensionality, making learning directly from examples infeasible. Researchers expect to decompose high-dimensional data matrix to obtain low-dimensional representation after high-dimensional matrix decomposi...

AI Technical Summary

Problems solved by technology

This method forces the representation of the projection space to have the same type of label as the data in the original space. The disadvantage of this method is that when there are few known labeled data, the method degenerates into NMF, which cannot effectively use the internal structure of the original data. Class performance is not improved

In addition, the CNMF method uses not a constraint pair, but a hard mark, and this information is generally difficult to obtain

Images