期刊论文

中文

系统仿真学报

1004-731X

2014

26

7

1517-1521

10.16182/j.cnki.joss.2014.07.034

湖南省科技计划资助项目（2014FJ3003） 湖南省教育厅一般项目（12C0436）

本校为第一机构

网络环境中,系统受到多种入侵行为威胁,具有变化快、种类多、随机性强的特点,提出一种基于AdaBoost的网络入侵智能检测算法,利用神经网络建立入侵检测学习器,采用AdaBoost集成学习方法对基学习器迭代训练,生成最终的入侵检测模型,仿真实验表明,该方法相对传统的BP神经网络算法具有较大的优势,降低了随机样本对模型的影响,增加了神经网络的泛化能力,使最终的入侵检测模型更加贴近真实的网络入侵样本,减小了小样本集导致的模型精度大幅下降的问题,对识别、检测各类入侵行为,对维护网络安全,具有十分重要的意义。

In network environment, system is affected by numerous invasion threat which has the characteristic of fast change, variety and strong random, The AdaBoost network intrusion intelligent detection algorithm was proposed. Intrusion detection learning device was built by using neural network, which was trained by using AdaBoost integrated study method, and final intrusion detection model was generated. Simulation results show, this method has many advantages, and it can reduce the influence of random samples of the model and increase the generalization ability of neural network, which can make th...