Semi-Supervised Dual Relation Learning for Multi-Label Classification

被引:0
|
作者
Wang, Lichen [1 ]
Liu, Yunyu [1 ]
Di, Hang [1 ]
Qin, Can [1 ]
Sun, Gan [1 ]
Fu, Yun [1 ,2 ]
机构
[1] Northeastern Univ, Dept Elect & Comp Engn, Boston, MA 02115 USA
[2] Northeastern Univ, Khoury Coll Comp Sci, Boston, MA 02115 USA
关键词
Semantics; Feature extraction; Correlation; Training; Tensors; Task analysis; Sun; Label relation learning; semi-supervised learning; multi-label learning; image retrieval; image annotation; GRAPH;
D O I
10.1109/TIP.2021.3122003
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
In a real-world scenario, an object could contain multiple tags instead of a single categorical label. To this end, multi-label learning (MLL) emerged. In MLL, the feature distributions are long-tailed and the complex semantic label relation and the long-tailed training samples are the main challenges. Semi-supervised learning is a potential solution. While, existing methods are mainly designed for single class scenario while ignoring the latent label relations. In addition, they cannot well handle the distribution shift commonly existing across source and target domains. To this end, a Semi-supervised Dual Relation Learning (SDRL) framework for multi-label classification is proposed. SDRL utilizes a few labeled samples as well as large scale unlabeled samples in the training stage. It jointly explores the inter-instance feature-level relation and the intra-instance label-level relation even from the unlabeled samples. In our model, a dual-classifier structure is deployed to obtain domain invariant representations. The prediction results from the classifiers are further compared and the most confident predictions are extracted as pseudo labels. A trainable label relation tensor is designed to explicitly explore the pairwise latent label relations and refine the predicted labels. SDRL is able to effectively and efficiently explore the feature-label relation as well as the label-label relation knowledge without any extra semantic knowledge. We evaluated SDRL in general and zero-shot multi-label classification tasks and we concluded that SDRL is superior to other SOTA baselines. Furthermore, extensive ablation studies have been done which reveal the effectiveness of each component in our framework.
引用
收藏
页码:9125 / 9135
页数:11
相关论文
共 50 条
  • [1] Semi-supervised dual relation learning for multi-label classification
    Wang, Lichen
    Liu, Yunyu
    Di, Hang
    Qin, Can
    Sun, Gan
    Fu, Yun
    [J]. IEEE Transactions on Image Processing, 2021, 30 : 9125 - 9135
  • [2] Dual Relation Semi-Supervised Multi-Label Learning
    Wang, Lichen
    Liu, Yunyu
    Qin, Can
    Sun, Gan
    Fu, Yun
    [J]. THIRTY-FOURTH AAAI CONFERENCE ON ARTIFICIAL INTELLIGENCE, THE THIRTY-SECOND INNOVATIVE APPLICATIONS OF ARTIFICIAL INTELLIGENCE CONFERENCE AND THE TENTH AAAI SYMPOSIUM ON EDUCATIONAL ADVANCES IN ARTIFICIAL INTELLIGENCE, 2020, 34 : 6227 - 6234
  • [3] A survey of multi-label classification based on supervised and semi-supervised learning
    Han, Meng
    Wu, Hongxin
    Chen, Zhiqiang
    Li, Muhang
    Zhang, Xilong
    [J]. INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS, 2023, 14 (03) : 697 - 724
  • [4] A survey of multi-label classification based on supervised and semi-supervised learning
    Meng Han
    Hongxin Wu
    Zhiqiang Chen
    Muhang Li
    Xilong Zhang
    [J]. International Journal of Machine Learning and Cybernetics, 2023, 14 : 697 - 724
  • [5] Using Semi-Supervised Learning in Multi-label Classification Problems
    Santos, Araken M.
    Canuto, Anne M. P.
    [J]. 2012 INTERNATIONAL JOINT CONFERENCE ON NEURAL NETWORKS (IJCNN), 2012,
  • [6] Discrete Semi-supervised Multi-label Learning for Image Classification
    Xie, Liang
    He, Lang
    Shu, Haohao
    Hu, Shengyuan
    [J]. ADVANCES IN MULTIMEDIA INFORMATION PROCESSING, PT I, 2018, 11164 : 808 - 818
  • [7] Applying semi-supervised learning in hierarchical multi-label classification
    Santos, Araken
    Canuto, Anne
    [J]. EXPERT SYSTEMS WITH APPLICATIONS, 2014, 41 (14) : 6075 - 6085
  • [8] Multi-label learning vector quantization for semi-supervised classification
    Chen, Ning
    Ribeiro, Bernardete
    Tang, Chaosheng
    Chen, An
    [J]. INTELLIGENT DATA ANALYSIS, 2019, 23 (04) : 839 - 853
  • [9] Robust Multi-Label Semi-Supervised Classification
    Li, Sheng
    Fu, Yun
    [J]. 2017 IEEE INTERNATIONAL CONFERENCE ON BIG DATA (BIG DATA), 2017, : 27 - 36
  • [10] Semi-Supervised Partial Multi-Label Learning
    Xie, Ming-Kun
    Huang, Sheng-Jun
    [J]. 20TH IEEE INTERNATIONAL CONFERENCE ON DATA MINING (ICDM 2020), 2020, : 691 - 700