A graphical feature generation approach for intrusion detection
Department of Instrumentation Science and Technology, National University of Defense Technology, 410073 Changsha, P.R. China
a Corresponding author: email@example.com
In order to develop a novel effective and efficient intrusion detection system, a novel hybrid method based on a graphical features-based k-nearest neighbor approach, namely GFNN, is proposed in this paper. In GFNN, k-means clustering algorithm is used to extract cluster centre of each class in the given dataset. Then, the distance between a specific data sample and each cluster centre is calculated, and a radar chart is plotted based on the new data composed of distance based features. The sub-barycentre based features for each sample are extracted from the radar chart. As a result, our proposed approach transforms the original multi-dimensional feature space into 5-dimensional sub-barycentre feature space. The experimental results of 10-fold cross-validation based on the KDDcup99 dataset show that the GFNN not only performs better than or similar to several other approaches in terms of classification accuracy, precision, and recall. It also provides high computational efficiency for the time of classifier training and testing.
Key words: Intrusion detection / Machine learning / Graphical feature / Radar chart / k-means / k-nearest neighbors
© Owned by the authors, published by EDP Sciences, 2016
This is an Open Access article distributed under the terms of the Creative Commons Attribution License 4.0, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.