Numerical Analysis for Relevant Features in Intrusion Detection (NARFid)
José A. González · 2009
Abstract : This thesis evaluates the usefulness of good feature subsets for the general classification task of identifying cyber attacks and network services. The generality of the selected features elucidates the relevance or irrelevance for the classification task of intrusion detection. Additionally, the work provides an extension to assessing features by inter-class separability(Bhattacharyya Coefficient) for multiple class problems, which intends to select the best-performing features for all of the classes.