Abstract: A novel Ensemble Intrusion Detection System is proposed in this study. In this system, Principle Component Analysis (PCA) and Independent Component Analysis (ICA) feature extraction approaches are used to construct two Support Vector Machine (SVM) classifiers. Then the results are combined to pursue higher performance. Because the costs of false positive error and false negative error are asymmetric in IDS, we introduce Pareto-Optimal Approach to obtain the optimal weight for the ensemble system. Experiments on the data set KDD Cup 1999 Data show that the proposed system outperforms standard SVM, PCA SVM and ICA SVM.