A decision tree-based rule formation with combined PSO-GA algorithm for intrusion detection system

摘要

Intrusion detection is the method of analysing and monitoring the actions in the internet to recognise the ciphers of security issues. Nowadays, the existing intrusion detection algorithms concentrate on the issues of feature selection, because some of the features are redundant and irrelevant that yields lengthy detection procedures. This paper proposes a combined particle swarm optimisation with genetic algorithm (CPSO-GA) approach to improve the intrusion detection accuracy. Initially, the dataset is loaded and pre-processed to remove the noisy and redundant information. Then, the necessary features are selected based on the proposed CPSO-GA. The decision rules are formulated for the selected features to improve the attacker prediction. If any new type of attacker established, the dynamic features are analysed, because, the static features are not altered for any instances. The proposed approach achieves higher intrusion detection rate and lesser error percentage than the existing feature selection algorithms and decision tree classifiers.