Robust multiclass least squares support vector classifier with optimal error distribution
作者：
Highlights：
• By minimizing the mean and variance of the modeling errors class-wisely, a robust least squares support vector classifier (RLSSVC) with optimal error distribution for the binary classification is proposed.
• The theoretical analysis indicates that the variance of the modeling errors of RLSSVC is smaller than that of RLSSVR for the binary classification problems.
• By introducing the multiple error adjusting factors, RLSSVC is then generalized for solving the multiclass classification problems, which inherits the superiority of RLSSVC in solving the binary classification problems.
• The robustness analysis delivers that RLSSVC assigns the smaller weights for the training instances with the larger errors, while the larger weights for the training instances with the smaller errors.
• The performance of RLSSVC is further improved by introducing the metric learning and kernel trick.
摘要
•By minimizing the mean and variance of the modeling errors class-wisely, a robust least squares support vector classifier (RLSSVC) with optimal error distribution for the binary classification is proposed.•The theoretical analysis indicates that the variance of the modeling errors of RLSSVC is smaller than that of RLSSVR for the binary classification problems.•By introducing the multiple error adjusting factors, RLSSVC is then generalized for solving the multiclass classification problems, which inherits the superiority of RLSSVC in solving the binary classification problems.•The robustness analysis delivers that RLSSVC assigns the smaller weights for the training instances with the larger errors, while the larger weights for the training instances with the smaller errors.•The performance of RLSSVC is further improved by introducing the metric learning and kernel trick.
论文关键词：Outliers,Robust least squares support vector classifier,Error distribution,Multiclass classification
论文评审过程：Received 22 July 2020, Revised 7 November 2020, Accepted 2 December 2020, Available online 4 December 2020, Version of Record 21 January 2021.
论文官网地址：https://doi.org/10.1016/j.knosys.2020.106652
