Feature selection method and device based on adaptive graph structure constraint subspace learning

Abstract

The invention discloses a feature selection method and device based on adaptive graph structure constraint subspace learning, and the method comprises the following steps: selecting a plurality of features from an original high-dimensional feature space to form a subspace through the subspace learning based on matrix decomposition; through adaptive graph structure learning, constraining the obtained subspace having the same structural features as the original high-dimensional feature space. The invention further provides a feature selection system based on adaptive graph structure constraint subspace learning, terminal equipment and a computer readable storage medium. Adaptive graph structure learning is applied to feature selection, and it is ensured that a feature subspace obtained through feature selection has a similar data structure with an original feature space. According to the method and device, the subspace learning process based on matrix decomposition and the adaptive graph structure learning process are jointly optimized, and the performance of the processing method is improved.

Description

technical field [0001] The invention belongs to the field of machine learning, and in particular relates to a feature selection method and equipment based on self-adaptive graph structure constrained subspace learning. Background technique [0002] With the development of various sensor technologies, the ability of modern sensor systems to collect data is getting stronger and stronger. At the same time, a large amount of high-dimensional data is also generated. These high-dimensional data usually contain some redundant data, irrelevant data and noise data, and these data often have adverse effects on subsequent learning algorithms. At present, feature selection methods are widely used to preprocess these data. Feature selection is based on selecting a subset of features to approximate all features for dimensionality reduction. Feature selection can be supervised, unsupervised or semi-supervised. In practical applications, it is usually difficult to obtain labeled data, s...

AI Technical Summary

Problems solved by technology

[0006] The purpose of the present invention is to provide a feature selection method and equipment based on adaptive graph structure constrained subspace learning to solve the problem of inaccurate processing and unstable performance of the unsupervised feature selection method in the prior art, which ensures While maintaining the subspace of the original high-dimensional feature space, the learned subspace maintains the structural characteristics of the original high-dimensional feature space

Images