Automated Feature Weighting for Network Anomaly Detection

A number of network features is used to describe normal and intrusive traffic patterns. However the choice of features is dependent on which pattern to be detected. In order to identify which network features are more important for a particular network pattern, we propose an automated feature weighting method based on a fuzzy subspace approach to vector quantization modeling that can assign a weight to each feature when network models are trained. The proposed method not only increases the detection rate but also reduces false alarm rate as presented in our experiments.