RBF-SVM and its Application on Network Security Risk Evaluation
| Autor: | Cong-Cong Li, Xiao-dong Yu, Hui-sheng Gao, Ai-ling Guo |
|---|---|
| Rok vydání: | 2008 |
| Předmět: |
Computer Science::Machine Learning
business.industry Computer science Network security ComputingMethodologies_IMAGEPROCESSINGANDCOMPUTERVISION Machine learning computer.software_genre Support vector machine Kernel (linear algebra) ComputingMethodologies_PATTERNRECOGNITION Computer Science::Sound Kernel (statistics) Pattern recognition (psychology) Radial basis function kernel Data mining Artificial intelligence business computer |
| Zdroj: | 2008 4th International Conference on Wireless Communications, Networking and Mobile Computing. |
| DOI: | 10.1109/wicom.2008.1110 |
| Popis: | Support vector machine is a novel machine learning method in recent years, the SVM with RBF is widely used in pattern recognition because of its good learning properties. If Support vector machine is applied into risk assessment, it will get better assessment results. But the performance of RBF-SVM is influenced greatly by the parameter of C and sigma. The principle of SVM and the essence of kernel function are introduced in this paper,This paper analyses the influence of the parameters of C and sigma to the performance of RBF-SVM, and then the picture of the changing curve of the C and sigma affect the number of SV and wrong recognition rate are presented. The result indicate that we can get the best assessment model if choose the appropriate RBF parameter. AT last, through risk evaluation with SVMs under different kernel functions, the superiority performance of RBF-SVM is validated. Simultaneously, the best learning performance of RBF-SVM assessment model is received. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|