一种基于ISSA-GBDT的入侵检测模型
An Intrusion Detection Model Based on ISSA-GBDT
肖香梅 1林志兴 1余建1
作者信息
- 1. 三明学院网络技术中心,福建三明 365004
- 折叠
摘要
针对机器学习算法的超参数选择不科学会影响网络入侵检测模型检测率的问题,提出一种融合改进麻雀搜索算法(ISSA)优化梯度提升决策树(GBDT)的入侵检测模型.ISSA采用佳点集初始化种群,丰富算法种群的多样性;加入透镜成像反向学习策略,增强算法跳出局部最优的能力;引入鲸鱼优化算法(WOA)对麻雀个体的位置进行扰动,使算法具有更强的全局搜索能力和更高的收敛精度.通过ISSA寻优得到GBDT的最优超参数,建立ISSA-GBDT入侵检测模型.实验结果表明,该模型具有更高的检测精度.
Abstract
Network intrusion detection based on Gradient Boosting Decision Tree(GBDT)has the problem that the unscientific hyper parameter selection of GBDT affects the model detection rate.Therefore,an intrusion detection model that integrates Improved Sparrow Search Algorithm(ISSA)to optimize GBDT is proposed.ISSA utilizes the good point set to initialize the population,enriching the diversity of the algorithm population.It incorporates a lens imaging reverse learning strategy to enhance the ability of the algorithm to escape local optimum.Additionally,the Whale Optimization Algorithm(WOA)is introduced to perturb the position of the sparrow individuals,providing the algorithm with stronger global search capability and higher convergence accuracy.The optimal hyper parameters of GBDT are obtained by ISSA optimization,and the ISSA-GBDT intrusion detection model is established.Experimental results show that the model has higher detection accuracy.
关键词
入侵检测/麻雀搜索算法/梯度提升决策树/超参数优化Key words
intrusion detection/sparrow search algorithm/gradient boosted decision trees/hyper parameter optimization引用本文复制引用
出版年
2024
NETL
NSTL
万方数据