An Effective Ensemble Automatic Feature Selection Method for Network Intrusion Detection

The mass of redundant and irrelevant data in network traffic brings serious challenges to intrusion detection, feature selection can effectively remove meaningless information from the data. Most current filtered embedded methods use a fixed threshold or ratio determine number features subset, which requires priori knowledge. In contrast, wrapped are computationally complex time-consuming; meanwhile, individual have bias evaluating features. This work designs an ensemble-based automatic method called EAFS. Firstly, we calculate importance ranks based on methods, then add subsets sequentially by evaluate subset performance comprehensively designing NSOM obtain with largest value. When searching for higher accuracy is retained lower computational complexity calculating when full set used. Finally, obtained ensembled, comparing experimental results three large-scale public datasets, described this study help classification, also compared other discover that our outperforms recent terms performance.