A Non-singular Twin Support Vector Machine
Due to high efficiency, twin support vector machine (TWSVM) is suitable for large-scale classification problems. However, there is a singularity in solving the quadratic programming problems (QPPs). In order to overcome it, a new method to solve the QPPs is proposed in this paper, named non-singular twin support vector machine (NSTWSVM). We introduce a nonzero term to the result of the problem. Compared to the TWSVM, it does not need extra parameters. In addition, the successive overrelaxation technique is adopted to solve the QPPs in the NSTWSVM algorithm to speed up the training procedure. Experimental results show the effectiveness of the proposed method in both computation time and accuracy.
KeywordsTwin support vector machine Singularity Classification
This work was supported in part by the National Natural Science Foundation of China under Grants (61472307, 51405387), the Key Research Project of Shaanxi Province (2018GY-018) and the Foundation of Education Department of Shaanxi Province (17JK0713).
- 4.Chen, S., Wu, X.: Improved projection twin support vector machine. ACTA Electron. Sinca 45(2), 408–416 (2017). https://doi.org/10.3969/j.issn.0372-2112.2017.02.020CrossRefGoogle Scholar