Improvement of E-MIMLSVM+ Algorithm Based on Semi-Supervised Learning
The MIMLSVM algorithm is to transform the MIML learning problem into a single-instance multi-label learning problem, which is used as a bridge to degenerate into a single-instance single-label learning. However, this degradation algorithm is relatively easy to understand, but in the degradation process will lose some information, affecting the classification effect. By using multi-tasking learning, E-MIMLSVM+ is used to combine tag relevance to improve the algorithm MIMLSVM+. In order to make full use of the unlabeled samples to improve the classification accuracy, the paper improves MIMLSVM algorithm by using the semi-supervised learning method. Experimental results show that the proposed method can achieve higher classification accuracy.
KeywordsSVM MIML Semi-supervised learning Multitask learning
The work was sponsored by the Institute of computer vision, image processing and pattern recognition, Zhejiang Sci-Tech University.
- 1.Zhou, Z. H., Zhang, M. L., Huang, S. J., Li, Y. F. (2008). MIML: A framework for learning with ambiguous objects. CORR abs/0808.3231.Google Scholar
- 7.Zhou, Z. H., & Zhang, M. L. (2006). Multi-instance multi-label learning with application to scene classification. Advances in Neural Information Processing Systems, 176(1), 1609–1616.Google Scholar
- 8.Zhang, M. L., Zhou, Z. H. (2008). M3MIML: A maximum margin method for multi-instance multi-label learning. Proceedings of the 8th IEEE International Conference on Data Mining(ICDM’08), Pisa, Italy, 208: 688–697.Google Scholar
- 9.Li, Y. X., Ji, S. W., Kumar, S., Ye, J. P., & Zhou, Z. H. (2012). Drosophila gene expression pattern annotation through multi-instance multi-label learning. Transactions on Computational Biology and Bioinformatics, 9(1), 1445–1450.Google Scholar
- 10.Blum, A., & Chawla, S. (2001). Learning from labeled and unlabeled data using graph mincuts. In Proceedings of the 18th International Conference on Machine Learning (ICML), 19–26, Williamston, MA.Google Scholar