A RF-PSO Based Hybrid Feature Selection Model in Intrusion Detection System
Hengxun Li, Wei Guo, Guoying Wu, Yanxia Li · 2018
Feature selection is an indispensable part in intrusion detection system to reduce irrelevant and redundant features and save the system from slow training and testing process. In general, wrapper and filter methods have been proposed for feature selection, but wrapper methods demand heavy computational resource for training and cross validation while filter methods lack the capability to minimize generalization error. In this paper we propose a hybrid feature selection model based on random forest and particle swarm optimization which makes use of both an independent measure and a learning algorithm to evaluate feature subsets. It uses the independent measure to decide the best subsets for a given cardinality and uses the learning algorithm to select the final best subset among the best subsets across different cardinalities. We utilized the KDD1999 dataset to evaluate the TPR and FPR of the proposed model, and compare them with CFS and SVM algorithms, the experimental evaluations demonstrate the capability of the proposed model.