Random subspace PCA based intrusion detection classifier ensemble

To Solve the problem of low accuracy and high false alarm, a construction method of Bagging ensemble based on random subspace PCA (Principle Component Analysis) was proposed. To create a training data for a base classifier, the feature set is randomly split into several subsets and PCA is applied to each subset. all principal components are retained to keep the variety information in the data; To increase the diversity of classifiers in the ensemble, random sampling with replacement is used to choose non-empty sample subset of each class; To avoid the performance deterioration problem caused by sample imbalance, we also adopt balance strategy in sampling. The novel method is applied to MIT KDD 99 dataset and the results demonstrate that better performance can be achieved in comparison with SVM-Bagging ensemble.