Affiliation:
1. Hebei Normal University
2. Shijiazhuang Institute of Railway Technology
Abstract
A support vector machine (SVM) model combined Laplacian Eigenmaps (LE) with Cross Validation (CV) is proposed for intrusion detection. In the proposed model, a classifier is adopted to estimate whether an action is an attack or not. Maximum Likelihood Estimation (MLE) is used to estimate the intrinsic dimensions, and LE is used as a preprocessor of SVM to reduce the dimensions of feature vectors then training time is shortened. In order to improve the performance of SVM, CV is used to optimize the parameters of SVM in RBF kernel function. Compared with other detection algorithms, the experimental results show that the proposed model has the advantages: shorter training time, higher accuracy rate and lower false positive rate.
Publisher
Trans Tech Publications, Ltd.
Reference12 articles.
1. Om H, Kundu A. A hybrid system for reducing the false alarm rate of anomaly intrusion detection system[C]/Recent Advances in Information Technology (RAIT), 2012 1st International Conference on. IEEE, 2012: 131-136.
2. Yanxue Z, Dongmei Z, Jinxing L. Approach to Forecasting Multi-step Attack Based on Fuzzy Hidden Markov Model[J]. Journal of Applied Sciences, 2013, 13(22).
3. De Castro L N, Timmis J. Artificial immune systems: a new computational intelligence approach[M]. Springer, (2002).
4. Hashem S H. Efficiency of Svm and Pca to Enhance Intrusion Detection System[J]. Journal of Asian Scientific Research, 2013, 3(4): 381-395.
5. Eskin E. Anomaly detection over noisy data using learned probability distributions[J]. (2000).