Due to a large amount of raw data,and high dimension and redundancy in intrusion detection,resulting in the problem of low recognition,long operation and bad performance for traditional intrusion detection algorithm in the face of massive data detection,a meth-od of combining compressed sensing and least square support vector machine to apply to intrusion detection system is put forward. Inno-vation as follows:Introducing compressive sampling to extract the feature from the original data,the high dimensional data is transformed into a low one on the premise of retaining the main feature for original data;Using least square support vector machine to directly train and classify data in the observation domain,and the kernel function is constructed by the combination of kernel function. Simulation show that using compressed sensing to extract the feature can significantly reserve the original feature. Moreover,least square support vector ma-chine can accelerate the speed of classifying without losing accuracy. This method can greatly reduce the training time,and effectively im-prove the accuracy of detection.%由于入侵检测中具有原始数据量大、维度较高、冗余度较大等特点,导致传统的入侵检测算法面对海量数据时检测识别度低,运行时间长,性能较差.为此,文中提出了一种将压缩感知和最小二乘支持向量机应用于入侵检测系统的方法.其创新点主要在于:引入压缩采样技术提取原始数据特征,在保留原数据主要特征的前提下,将高维数据转化为低维数据;利用最小二乘支持向量机直接在观测域中训练和分类数据,且核函数通过组合核函数构建.仿真结果表明,运用压缩感知进行特征提取能够极大保留原始特征,而最小二乘支持向量机能够在不损失精度的前提下加速分类.该方法能够较大地减少训练时间,并可以有效提高检测精度.
展开▼
