Interactive representation-based domain adaptive subspace learning method

Abstract

A domain adaptive subspace learning method based on interactive representation belongs to the technical field of image classification. In the method, two interactive representation models are established based on low-rank constraint on a source domain and a target domain, so as to better coordinate the distribution difference between data; then a distance constraint is designed to simulate a subspace relation between a source domain and a target domain, and the problem that the two domains are inconsistent is solved; further, a combined framework is developed, and additional classification discrimination is obtained by using regression based on labels, so that the adaptability and robustness of the model are promoted. Compared with other methods, the method is higher in accuracy and more stable in performance.

Description

technical field [0001] The invention belongs to the technical field of image classification, and in particular relates to a domain adaptive subspace learning method based on interactive representation. Background technique [0002] In most machine learning methods, the source and target domains are usually considered to have similar distributions. However, in the real world, the source and target domains are affected by light, angle, background, so the problem of domain inconsistency arises, that is, the source and target domains cannot have similar distributions. However, when classifying images, domain inconsistency will lead to a decrease in classification accuracy, and traditional domain adaptive methods only consider local information but not global information. Contents of the invention [0003] The present invention overcomes the deficiencies of the above-mentioned technologies, and provides a domain-adaptive subspace learning method based on interactive representa...

AI Technical Summary

Problems solved by technology

However, in the real world, the source and target domains are affected by light, angle, background, so the problem of domain inconsistency occurs, that is, the source and target domains cannot have similar distributions

However, when classifying images, domain inconsistency will lead to a decrease in classification accuracy, and traditional domain adaptive methods only consider local information but not global information.

Images