基于模糊积分的多神经网络融合模型的研究

为了进一步提高网络入侵检测系统的检测性能,将模糊积分理论和神经网络技术应用到网络入侵检测中,提出了基于模糊积分的多神经网络融合模型MNNF。它的基本思想是按照TCP/IP属性集的类别不同将TCP/IP数据集分成三个不同属性集的子数据集,在不同属性集上训练形成不同的子神经网络,然后用模糊积分将多个子神经网络对TCP/IP数据的检测结果进行非线性融合形成最优判断。实验结果表明,MNNF模型应用在网络入侵检测中可以得到比单个神经网络更好的入侵检测性能。

Research on model of combining multiple neural networks by fuzzy integral

The model of Multiple Neural Networks by Fuzzy(MNNF) integral presented in this paper is an effective method to improve the detection performance of network intrusion detection system.The basic idea of MNNF is to divide TCP/IP dataset into three sub-datasets according to different attributes of TCP/IP data,train on different sub-datasets and construct different sub-neural networks separately,and detect TCP/IP data by different sub-neural networks,and then nonlinearly combine the results from multiple sub-neural networks by fuzzy integral.The experiment results show that this technique is superior to the single neural networks for network intrusion detection in terms of classification accuracy.