A novel hybrid classification framework using SVM and differential evolution

作者：Yu, Xiaobing^*; Wang, Xuming

来源：Soft Computing, 2017, 21(14): 4029-4044.

DOI：10.1007/s00500-016-2054-9

• Classification
• SVM
• Differential evolution
• Hybrid model

The kernel parameters setting for SVM in a training process impacts on the classification accuracy. The hybrid framework based on SVM and ensemble differential evolution is proposed to enhance the classification accuracy. The ensemble differential evolution is utilized to optimize the kernel parameters. The ensemble differential evolution algorithm employs two trial vector generation strategies and two control parameter settings. Ten real-world datasets using the proposed hybrid framework are tested. Compared with the SVM variants, the proposed hybrid model improves the classification accuracy of SVM.

• 出版日期2017-7
• 单位南京信息工程大学