Improved Wild Horse Optimizer with Cooperative Multi-Strategies for Network Intrusion Detection

Due to the problems of redundant features and high dimensionality in network intrusion detection, The original Wild Horse Optimiser (WHO) suffers from a lack of exploratory power and gets into local optimality when applied to network intrusion detection. Therefore, an improved wild horse optimizer with collaborative multi-strategy (WHOCMS) is proposed to solve these problems. Sine chaos dynamic opposition-based learning strategy is used to initialize the population and lay the foundation for global search for the optimal solution; adaptive weighted average strategy is used to prevent individuals from moving to the edge of the optimal solution and improve the exploration ability of the algorithm. The WHOCMS algorithm is tested compared with the original WHO algorithm and other classical algorithms in six benchmark functions, the results show that the WHOCMS algorithm is highly competitive in convergence speed and accuracy. Applying the WHOCMS algorithm into the network intrusion detection UNSW-NB 15 dataset significantly reduces the number of data dimensions. Compared with the original WHO algorithm, the accuracy is improved by 3.71% and the intrusion detection time is reduced by 27.85%, which effectively improves the performance of network intrusion detection.