Application of Elephant Herd Optimization Algorithm Based on Levy Flight Strategy in Intrusion Detection

Abstract

Intrusion Detection System (IDS) is designed to detect attacks against computer systems and networks. The IDS needs to process huge amounts of data, where irrelevant or redundant feature data can lead to degraded detection accuracy. In order to improve the accuracy of intrusion detection, feature selection can be used to delete irrelevant feature data. Considering that Elephant Herd Optimization (EHO) algorithm has a good classification ability in reducing feature redundancy, it can be applied to network intrusion detection. But the EHO algorithm converges too quickly in the searching process and it is then easy to fall into the local optimum, which constrains the classification performance of the algorithm. In this paper, an improved Elephant Herd Optimization based on Levy Flight strategy (LFEHO) is proposed, which overcomes the defects of easy precocity and low convergence accuracy of the original EHO algorithm. The classification performance is improved under the premise of ensuring the accuracy rate. This paper uses KDD CUP 99 as a simulation data set to test, and the experimental results show that the classification accuracy is improved. Compared with the basic EHO algorithm, classical Particle Swarm Optimization (PSO) algorithm and Moth-Flame Optimization (MFO) algorithm, the proposed LFEHO algorithm is superior to other algorithms in the optimization accuracy and convergence performance.