论文部分内容阅读
为改善网络安全防护水平,提出一种基于偏最小二乘(PLS)法和核向量机(CVM)的组合式异常入侵检测方法.首先,采用PLS算法提取网络数据的主成分,构建特征集;然后,利用CVM构建特征集的异常入侵检测模型,进而完成异常入侵检测和判定.仿真实验结果表明,所提出的方法具有CVM的大规模数据快速处理能力,而且检测性能与L1-SVM和L2-SVM大致相当,尤其主成分数为1 538时能保持相对较高的检测水平,验证了将其用于异常入侵检测的有效性和可行性.
In order to improve the level of network security protection, a combined anomaly detection method based on partial least squares (PLS) and kernel vector machine (CVM) is proposed.Firstly, PLS algorithm is used to extract the principal components of network data and construct feature set. Then, the CVM is used to construct the anomaly detection model of the feature set, and then the anomaly detection and decision are completed.The simulation results show that the proposed method has the capability of large-scale data processing of CVM, and the detection performance is similar to that of L1- SVM and L2- SVM is roughly the same, especially when the main component is 1 538, it can maintain a relatively high level of detection, which verifies the validity and feasibility of using it for anomaly detection.