Multi-label image classification method based on robust feature space joint learning

Abstract

The invention discloses a multi-label image classification method based on robust feature space joint learning. The method comprises the following steps: firstly, preparing multi-label data; introducing a feature selection and nuclear norm low-rank representation method into the feature space and the tag space, so that a robust multi-tag classification model jointly learned by the tag and the feature space is constructed; mapping a q-dimensional label space into an r-dimensional label space by using a robust low-rank projection matrix V, so that V represents robust low-rank projection learned in the label space; introducing an augmented Lagrange multiplier method to solve a matrix P and a matrix V in a target function of the multi-label classification model; taking the multi-label data as a training data sample to train the proposed multi-label classification model to obtain an optimal robust low-rank coefficient matrix P and an optimal robust low-rank projection matrix V; and completing training of a multi-label classification model based on the optimal robust low-rank coefficient matrix P and the robust low-rank projection matrix V, and performing multi-label image classification by using the multi-label classification model.

Description

technical field [0001] The invention relates to the technical field of computer image processing, in particular to a multi-label image classification method for robust feature space joint learning. Background technique [0002] In recent years, multi-label image classification is one of the popular technologies in the field of machine learning, and has been widely used to solve various practical application problems. Unlike traditional supervised learning with only one mutually exclusive label, multi-label classification includes multiple different Tag of. Due to the ubiquitous correlation between the feature space and the label space of multi-label classification, the complexity of the training space increases with the increase of the number of samples, and there is also label correlation in the multi-label, these factors will negatively affect the classification performance. influences. [0003] However, most of the current multi-label image classification methods only s...

AI Technical Summary

Problems solved by technology

Due to the ubiquitous correlation between the feature space and the label space of multi-label classification, the complexity of the training space increases with the increase of the number of samples, and there is also label correlation in the multi-label, these factors will negatively affect the classification performance. influences

[0003] However, most of the current multi-label image classification methods only study the technology related to the feature space, while ignoring the research on the label output space.

Therefore, most multi-label classifiers will be deeply affected by the label quality

[0004] In addition, the high-dimensional original data set generally has noise caused by data corruption and missing, which greatly reduces the accuracy of classification results.

At present, most classification techniques only focus on feature space noise, and there are very few studies on label space noise.

Therefore, the anti-noise function of these techniques in multi-label classification is not significant

Images