基函数神经网络入侵检测系统的实现

Implementation of Intrusion Detection System Based on Basis Function Neural Network

提出一种新型的基函数神经网络用于入侵检测技术中,其中每个神经元的活跃函数各不相同,彼此正交,在更高层次上完成对生物神经系统的模拟,它即可以用于异常检测以检测出新的攻击,也可以用于误用检测以检测出已恬的攻击及其变种。根据所用基函数神经网络的基本结构和训练方法,在Windows环境下进行了基于网络的入侵检测实验,结果表明,运用基函数神经网络检测入侵,可提高入侵检测系统的准确检测率。

This paper describes a new model of designing intrusion detection system based on basis function neural networks,which contains different and orthogonal active function,simulates biology neural system on higher level.It applies in not only anomaly detection to detect new attacks,but also in misuse detection to detect known attacks and its aberrance.Basic structure and training method of basis function neural networks is explained.An experiment based on Windows platform is given,the experimental result shows that the approach improves the accuracy of detection.