基于LM-BP神经网络的入侵检测系统的研究

针对传统BP神经网络存在收敛速度慢、易陷入局部最小点、计算量大的不足，提出一种结合Levenberg—Marquardt优化算法的BP神经网络，并应用在一个误用入侵检测系统。实验结果表明，新系统结合了异常检测和误用检测两者的优势，快速检测新型入侵，降低误警率和漏警率。

The Research of Intrusion Detection System Based on LM-BP Neural Network

In order to solve the low speed in constringency, getting into local minimum easily and the heavy calculating problem of traditional BP NN, an optimized Levenberg-Marquardt(LM)arithmetic is presented. The arithmetic constructed IDS with optimized BP NN based on a successful misuse IDS. For taking advantage of the anomaly IDS and misuse IDS, experiments prove the new IDS has lower rate of false alert and omitting alert, it could recognise unknown attacks rapidly.