A Multi-label Classification Algorithm Combining Feature Screening and Label Correlation.

Multi-label classification is a hot topic in the field of data mining. It has important applications in text classification, image, video annotation, music emotion classification, and other fields. In the past, most papers only used label correlation or feature screening to improve the accuracy of the multi-label classification and paid one-sided attention to feature screening while ignoring the correlation between labels. Therefore, in this paper, a multi-classification algorithm (MIRD) combining feature screening and label correlation is proposed, which not only combines the correlation between labels and features and design thresholds to screen features, but also uses association rules to update the label set to realize the correlation between labels, making full use of the correlation for multi-label classification. Finally, the proposed algorithm is compared with other multi-label algorithms, and the results show that it can achieve better results from most data sets, which proves that the proposed algorithm is better than the comparison algorithm. [ABSTRACT FROM AUTHOR]