入侵检测的进化神经网络研究

Research on Evolutionary Neural Network of Intrusion Detection

针对传统入侵检测系统漏报率和误报率高的问题,提出基于遗传神经网络的入侵检测系统。该系统将遗传算法和BP算法有机结合,利用遗传算法较强的宏观搜索能力和全局寻优特点,优化BP网络的初始权值和阈值,并在此基础上进行神经网络学习,从而建立入侵检测系统的优化分析模型。实验结果表明,该算法可以有效地运用于入侵检测系统中。

Aiming at the problem of the higher rate of losing alarm and false alarm in traditional intrusion detection system, a novel intrusion detection system based on genetic neural network is proposed, which combines genetic algorithm with BP algorithm to optimize the initial values of weights and bias of BP network by using the characteristic of strong macro search ability and optimization in overall situation. On basis of this, the neural network study is conducted, and the optimized analysis model of intrusion detection system is set up. Experimental results show this algorithm can be applied to this intrusion detection system effectively.