Improving Effectiveness of Intrusion Detection by Correlation Feature Selection

In this paper, the authors propose a new feature selection procedure for intrusion detection, which is based on filter method used in machine learning. They focus on Correlation Feature Selection (CFS) and transform the problem of feature selection by means of CFS measure into a mixed 0-1 linear programming problem with a number of constraints and variables that is linear in the number of full set features. The mixed 0-1 linear programming problem can then be solved by using branch-and-bound algorithm. This feature selection algorithm was compared experimentally with the best-first-CFS and the genetic-algorithm-CFS methods regarding the feature selection capabilities. Classification accuracies obtained after the feature selection by means of the C4.5 and the BayesNet over the KDD CUP’99 dataset were also tested. Experiments show that the authors’ method outperforms the best-first-CFS and the genetic-algorithm-CFS methods by removing much more redundant features while keeping the classification accuracies or getting better performances.