基于PCA与FCM的入侵检测样本数据压缩方法

A compression approach to intrusion detection of sample data according to PCA and FCM

采用行列双向压缩的数据处理策略,提出了一种基于主成分分析与模糊C-均值聚类算法的入侵检测样本数据压缩方法。该方法首先采用主成分分析法对数据冗余特征进行压缩,然后采用模糊C-均值聚类算法对冗余样本进行压缩,由此可挖掘入侵检测样本数据中的关键特征和关键样本。通过KDD CUP99数据集测试证明:数据双向压缩可减少入侵检测分类器的计算量,进而可提高其实时检测性能和检测推断的准确性。

Using the processing principle of dual-directional data compression, this paper presents a compression approach to the intrusion detection of sample data according to PCA and FCM to enhance the effectiveness and accuracy of intrusion detection. Firstly, PCA is used to remove the redundant features. Then, FCM is adopted to eliminate the redundant instances to discover the key features and instances from the intrusion detection sample data. Finally, an illustrative experiment on KDD CUP99 data test set is made to show that the method of dual data compression can reduce the number of train- ing classifiers and the computation of intrusion detection and further improve the efficiency of real time detection and the accuracy of intrusion detection.