Multi-label classification review and opportunities

作     者：Weng, Wei Li, Yu-Wen Liu, Jing-Hua Wu, Shun-Xiang Chen, Chin-Ling 

作者机构：College of Computer and Information Engineering Xiamen University of Technology Xiamen361024 China School of Instrument Science and Engineering Southeast University Nanjing210096 China College of Computer Science and Technology Huaqiao University Xiamen361021 China Department of Automation Xiamen University Xiamen361005 China School of Computer and Information Engineering Xiamen University of Technology Xiamen361005 China School of Information Engineering Changchun Sci-Tech University Changchun130600 China Department of Computer Science and Information Engineering Chaoyang University of Technology Taichung41349 Taiwan 

出 版 物：《Journal of Network Intelligence》 (J. Network Intell.)

年 卷 期：2021年第6卷第2期

页      面：255-275页

核心收录：

基　　金：Acknowledgment. This research was supported by the Natural Science Foundation of Jiangsu Province of China (BK20200364)  the Natural Science Foundation of the Fujian Province of China (2018J01572)  and the National Natural Science Foundation of China (62001111) 

主　　题：Classification (of information) 

摘      要：Multi-label classification originated from text classification and has became one of the most widely studied machine learning frameworks. After nearly twenty years of development, many multi-label classification models have been produced. In this paper, the representative algorithms are introduced and reviewed. On the other hand, in recent years, the volume of multi-label data has become larger, and the features and labels have become higher dimensions, so the multi-label algorithms have also generated new trends. The main research directions focus on label specific feature, label correlation, and dimension compression. This paper discusses these studies in detail. Different from other review works, this paper discusses the evolution and improvement process of various algorithms and summarizes the main research directions in recent years. The content summarized in this paper can provide a more comprehensive perspective for related researchers to understand the main research contents in this direction, and inspire new research means and methods. © 2021 Global Research Online. All rights reserved.