文件名称:UnsupervisedAnomalyDetectionBasedOnPrincipalCompon
介绍说明--下载内容均来自于网络,请自行研究使用
入侵检测系统在训练过程中需要大量有标识的监督数据进行学习,不利于其应用和推广.为了解决该问题,提出了一种基于主成分分析的无监督异常检测方法,在最小均方误差原则下学习样本的主要特征,经过压缩和还原的互逆过程后能最大限度地复制样本信息,从而根据均方误差的差异检测出异常信息.构建的仿真系统经过实验证明,基于主成分分析的无监督异常检测方法能够在无需专家前期参与的情况下检测出入侵,实验结果验证了其有效性.-Intrusion Detection System in the training process requires a large logo of Jiandushuoju learning, negative effect on their application and promotion. In order to solve the problem, a principal component analysis based on unsupervised anomaly detection method, the principle of the minimum mean square error The main characteristics of the sample under study, after compression and the reciprocal reduction procedure to copy the sample information as possible to the mean square error of the difference according to detect anomalies. Construction of the simulation system has been proved, not based on principal component analysis anomaly detection methods to monitor without the participation of experts in early detection of invasive cases, experimental results show its effectiveness.
相关搜索: 入侵检测
(系统自动生成,下载前可以参看下载内容)
下载文件列表
UnsupervisedAnomalyDetectionBasedOnPrincipalComponentsAnalysis.pdf