摘要
在对潜在网络威胁检测的研究过程中,采用当前算法进行潜在网络威胁检测时入侵特征的重复性对特征分类参数造成干扰,导致入侵检测效率低的问题。提出采用过往入侵特征分析算法的潜在网络威胁检测方法。上述方法首先将潜在网络威胁检测的正确率定义为网络威胁检测的优化目标函数,以SVM为分类器,利用过往入侵特征分析算法通过潜在网络威胁入侵样本训练SVM分类器,将过往入侵特征集与SVM参数作为约束条件建立潜在网络威胁入侵特征分类模型,然后加入粒子群算法依据分类结果在过往的潜在网络威胁入侵特征空间进行全局搜索,组建最优潜在网络威胁检测模型,进而精确地完成了潜在网络威胁检测。仿真结果证明,采用过往入侵特征分析算法的潜在网络威胁检测方法检测精确度高,适应性强。
A potential network threat detection method was proposed based on past intrusion feature analysis algo- rithm. In this method, firstly, the correct rate of potential network threat detection is defined as the optimization ob- jective function of network threat detection, a SVM is taken as classifier, and the past intrusion feature analysis algo- rithm is applied. Through the intrusion samples of potential threat network, the SVM classifier is trained. The past intrusion feature set and the SVM parameters are as constraint condition to establish the classification model of poten- tial network threat intrusion feature, and then the particle swarm algorithm is added. Based on classification results, global search in the past potential network threat intrusion feature space is carried out to form the optimal potential network threat detection model, and accurately complete the potential network threat detection. The simulation results show that the potential network threat detection method based on past intrusion feature algorithm has high accuracy and strong adaptability.
出处
《计算机仿真》
CSCD
北大核心
2015年第9期331-334,共4页
Computer Simulation
基金
深圳信息职业技术学院校级科研项目(LG2014029)
关键词
入侵检测
特征选择
粒子群
Intrusion detection
Feature selection
Particle swarm
