MLTSVM: A novel twin support vector machine to multi-label learning
作者：
Highlights：
• The first nonparallel hyperplane SVM (MLTSVM) classifier applied in multi-label learning is proposed.
• The multi-label information of the dataset can be effectively captured by multiple nonparallel hyperplanes.
• The ambiguity of the predicting procedure is avoided by the effective decision function.
• An efficient SOR algorithm is applied to solve the proposed MLTSVM.
• Experimental results confirm the feasibility and superiority of the proposed MLTSVM.
摘要
Highlights•The first nonparallel hyperplane SVM (MLTSVM) classifier applied in multi-label learning is proposed.•The multi-label information of the dataset can be effectively captured by multiple nonparallel hyperplanes.•The ambiguity of the predicting procedure is avoided by the effective decision function.•An efficient SOR algorithm is applied to solve the proposed MLTSVM.•Experimental results confirm the feasibility and superiority of the proposed MLTSVM.
论文关键词：Multi-label classification,Support vector machines,Twin support vector machines,Quadratic programming,Successive overrelaxation
论文评审过程：Received 7 February 2015, Revised 6 September 2015, Accepted 9 October 2015, Available online 3 November 2015, Version of Record 24 December 2015.
论文官网地址：https://doi.org/10.1016/j.patcog.2015.10.008
