An efficient weighted Lagrangian twin support vector machine for imbalanced data classification
作者：
Highlights：
• We propose an efficient WLTSVM for imbalanced data classification.
• A graph based under-sampling strategy is introduced, which is robustness to outliers.
• The weight biases are embedded in our WLTSVM formulations.
• One Lagrangian training algorithm is presented and its convergence is proven.
• Experimental results show its feasibility and efficiency.
摘要
Highlights•We propose an efficient WLTSVM for imbalanced data classification.•A graph based under-sampling strategy is introduced, which is robustness to outliers.•The weight biases are embedded in our WLTSVM formulations.•One Lagrangian training algorithm is presented and its convergence is proven.•Experimental results show its feasibility and efficiency.
论文关键词：Imbalanced data classification,Twin support vector machine,Weighted twin support vector machine,Lagrangian functions,Quadratic cost functions
论文评审过程：Received 10 January 2013, Revised 2 March 2014, Accepted 13 March 2014, Available online 25 March 2014.
论文官网地址：https://doi.org/10.1016/j.patcog.2014.03.008
