基于模糊神经网络集成的入侵检测模型

A integrated fuzzy-neural network intrusion detection model

日益严峻的网络安全形势和网络协议本身的缺陷,使传统的防火墙防御的方式无法胜任。为提高对网络入侵防御能力,提出了模糊神经网络集成的入侵检测模型:首先抓取网络中的数据流,使用模糊数学的方法对数据记录入侵特征预处理。然后用集成的模糊神经网络模块接收预处理模块导入的训练数据和测试数据,通过反复训练学习,把各子树中节点的权值收敛到确定值。训练完成后,模型用于检测网络中的数据。响应模块接收模糊神经网络模块处理结果做出相应的响应。实验使用KDDCUP99网络入侵检测数据集对模型进行评测,并与单一神经网络模型相比较。结果表明模糊神经网络集成的方法检测结果比较稳定,在整体上比单一神经网络的误报率、漏报率和错报率有所降低,准确率和数据集泛化能力明显提高。

With increasing serious situation of network security and network defects in the agreement itself,it is incompetent to use the traditional way of a firewall.A fuzzy neural network model of integrated intrusion detection is proposed to improve the ability of intrusion prevention in a network.First,the data stream is obtained from the network,and the fuzzy approach is used to perform data pre-processing on characteristics of invasion.Then,the training and testing data is received by the integrated fuzzy neural network module from the data pre-processing module.Through repeated training and learning,the weights of nodes in the sub-trees converge to determine values.When training is completed,the model is used to detect the network data.The response module receives the results of the fuzzy neural network module and makes the appropriate response.In the experiment,the network intrusion detection datasets,a part of KDDCUP99,are used to evaluate the integrated fuzzy neural network,and are compared to a single neural network model.On the whole,the result shows that the fuzzy neural network ensemble method results are more stable.It slightly reduces the false alarm rate,false negative rate and false positive rate and significantly improves on accuracy and ability of datasets generalization.