An Improved Charged System Search for Feature Selection of IDS

Abstract

Charged system search (CSS) algorithm which is inspired by the coulomb’s law and laws of motion has been proved to be competitive with existing evolutionary algorithms, such as Particle Swarm Optimization (PSO) algorithm. The shortcomings of existing CSS, algorithm is small convergence precision and readily captured in a local optimum value at the next evolution stage. The paper introduces an enhanced Charged System Search algorithm with Levy Flight (CSSLF), by looking at the information of the, best solutions into the exploration strategy for the feature subset selection. The Support Vector Machine has been applied as a classifier for assessment of the features selected. The experimental results on KDD-NSL dataset reveals that CSSLF discover better solutions than CSS in terms of larger detection rate, nominal false alarm rate and enhanced accuracy than the existing approaches.