Abstract
Identified the specific features from the instances belong to a certain class label is meaningful, the “purified” feature representation contains such label information can be shared with other feature learning. Besides, it is essential to distinguish the sample association relationship behind the multi-label da-tasets, which is conducive to improve the performance of the algorithm. However, most algorithms aim to capture the mapping between instances and labels, while ignoring the information about instance relations and label cor-relation hidden in the data structure. Motivated by these issues, we leverage the deep network to learn the special feature representations without aban-doning overlapped features. Meanwhile, the Euclidean metric matrices are leveraged to construct the diagonal matrix for the diffusion function, it en-sures that the results of model training by similar instance features are con-sistent. Further, considering the contributions of these feature representation are different and have influences on the final prediction results, thus the self-attention mechanism is introduced to fusion the other label specific in-stance features to build the new joint feature representation, which derive dynamic weights for multi-label prediction. Finally, experimental results of the real data sets show promising availabilities of our approach.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Dou, Z., Cui, H., Wang, B.: Learning global and local consistent representations for unsupervised image retrieval via deep graph diffusion networks. CoRR, abs/2001.01284 (2020)
Hoffmann, M., Noé, F.: Generating valid euclidean distance matrices. CoRR, abs/1910.03131 (2019)
Huang, J., Li, G., Wang, S., Xue, Z., Huang, Q.: Multi-label classification by exploiting local positive and negative pair-wise label correlation. Neurocomputing 257, 164–174 (2017)
Jiang, Y.-G., Zu-xuan, W., Wang, J., Xue, X., Chang, S.-F.: Exploiting feature and class relationships in video categorization with regularized deep neural networks. IEEE Trans. Pattern Anal. Mach. Intell. 40(2), 352–364 (2018)
Jing, L., Shen, C., Yang, L., Yu, J., Ng, M.K.: Multi-label classification by semi-supervised singular value decomposition. IEEE Trans. Image Process. 26(10), 4612–4625 (2017)
Kang, L., Wu, L., Yang, Y.H.: A novel unsupervised approach for multi-level image clustering from unordered image collection. Front. Comput. Sci. 7(1), 69–829 (2013)
Ke, T., Jing, L., Lv, H., Zhang, L., Yaping, H.: Global and local learning from posi-tive and unlabeled examples. Appl. Intell. 48(8), 2373–2392 (2018)
Hu, Y.-H.L.J.-H., Zhou, Y.J.Z.-H.: Towards discovering what patterns trigger what labels. In: Proceedings of the Twenty-Sixth AAAI Conference on Artificial Intelligence, 22–26 July 2012, Toronto, Ontario, Canada (2012)
Nam, J.: Learning Label Structures with Neural Networks for Multi-label Classification. PhD thesis, Darmstadt University of Technology, Germany (2019)
Seyedi, S.A., Lotfi, A., Moradi, P., Qader, N.N.: Dynamic graph-based label propagation for density peaks clustering. Expert Syst. Appl. 115, 314–328 (2019)
Xiao, T., Yin-he, W., Qin-ruo, W.: A rotation and scale invariance face recognition method based on complex network and image contour. In: 12th International Conference on Control Automation Robotics & Vision, ICARCV2012, Guangzhou, China, 5–7 December 2012, pp. 371–376 (2012)
Fei, W., et al.: Weakly semi-supervised deep learning for multi-label image annotation. IEEE Trans. Big Data 1(3), 109–122 (2015)
Guo-qiang, W., Zheng, R., Tian, Y., Liu, D.: Joint ranking SVM and binary relevance with robust low-rank learning for multi-label classification. Neural Netw. 122, 24–39 (2020)
Yan, Z., Liu, W., Wen, S., Yang, Y.: Multi-label image classification by feature attention network. IEEE Access 7, 98005–98013 (2019)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Lian, Sm., Liu, Jw., Luo, Xl. (2020). Partially Disentangled Latent Relations for Multi-label Deep Learning. In: Yang, H., Pasupa, K., Leung, A.CS., Kwok, J.T., Chan, J.H., King, I. (eds) Neural Information Processing. ICONIP 2020. Lecture Notes in Computer Science(), vol 12533. Springer, Cham. https://doi.org/10.1007/978-3-030-63833-7_48
Download citation
DOI: https://doi.org/10.1007/978-3-030-63833-7_48
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-63832-0
Online ISBN: 978-3-030-63833-7
eBook Packages: Computer ScienceComputer Science (R0)