改进后的BP神经网络在入侵检测中的应用

Application of Improved BP Neural Network in Intrusion Detection

随着网络攻击事件的频发入侵检测技术日益重要,由于传统的入侵检测技术易出现误报、漏报率高且不易识别未知网络攻击等问题,结合BP神经网络良好的自适应性、非线性映射能力和知识归纳等特征,针对入侵检测提出了一种改进算法。通过对BP网络中各个连接的权、阈值有针对性地配备了动态可变的学习率,使权、阈值的修改更有针对性,满足了权、阈值多变的要求,克服BP网络收敛速度慢和易陷入局部极小值的缺点。最后采用Kdd CUP99入侵检测数据集在Matlab 8.0上进行了仿真实验。实验结果表明,该算法具有较高的检测率和较低的误报率、漏报率,能够达到预期的入侵检测效果。

With the increasing intrusion detection technology of network attack events,due to the traditional intrusion detection technology prone to false positives,high false negative rate,and difficult to identify unknown network attacks and other issues,combined with BP neural network good adaptability,non-linear mapping ability and knowledge induction,this paper proposes an improved algorithm for intrusion detection.Through the right of each connection in the BP network,the threshold is targeted with the dynamic variable learning rate,which makes the modification of the right and the threshold more specific,satisfies the requirement of changing the threshold and the threshold,and overcomes the convergence speed of the BP network slow and easy to fall into the local minimum of the shortcomings.Finally,the simulation experiment is carried out in Matlab 8.0 using KddCUP99 intrusion detection data set.The experimental results show that the algorithm has high detection rate and low false alarm rate and false negative rate,and can achieve the expected intrusion detection effect.