| 一种高效的特征选择机制应用于入侵检测 |
| |
| 引用本文： | 张永,曹东侠.一种高效的特征选择机制应用于入侵检测[J].甘肃科学学报,2011,23(3):120-123. |
| |
| 作者姓名： | 张永 曹东侠 |
| |
| 作者单位： | 兰州理工大学计算机与通信学院,甘肃兰州,730050 |
| |
| 基金项目： | 甘肃省自然科学基金项目(0809RJZA015) |
| |
| 摘 要： | 基于数据集的wrapper型特征选择方法,通过去除原特征集中的冗余特征,不仅降低了对计算资源的耗用,而且在保持检测精度的前提下,提升了检测速度,有效改善了检测性能.在特征选择具体过程中,采用全局和局部搜索性能好的遗传算法和禁忌搜索混合搜索策略,针对网络数据的异构性,采用改进了的基于核函数的SVM作为评价特征子集分类性能...
|
| 关 键 词： | 特征选择 遗传算法 禁忌搜索 SVM 入侵检测 |
An Efficient Strategy of Feature Selecting Applied to Intrusive Detection |
| |
| ZHANG Yong,CAO Dong-xia.An Efficient Strategy of Feature Selecting Applied to Intrusive Detection[J].Journal of Gansu Sciences,2011,23(3):120-123. |
| |
| Authors: | ZHANG Yong CAO Dong-xia |
| |
| Institution: | ZHANG Yong,CAO Dong-xia(Institute of Computer and Communications,Lanzhou University of Science and Technology,Lanzhou 730050,China) |
| |
| Abstract: | An efficient strategy of feature selecting is applied to intrusive detection to reduce the cost of computation and to quicken the detecting velocity.During the process,the text adopts a hybrid search mechanism which combines genetic algorithm and taboo search organically,by means of improved SVM as evaluation criterion.The best-quality attributes can be obtained when the algorithm terminates.Experiments show that the selected features are superior to the original ones in detecting performance. |
| |
| Keywords: | feature selecting genetic algorithm taboo search SVM intrusive detection |
| 本文献已被 CNKI 维普 万方数据 等数据库收录！ |
|
