遗传禁忌算法优化BP网络用于入侵检测

Genetic tabu algorithm for BP network intrusion detection

针对入侵检测系统存在的高漏报率和误报率,提出一种基于遗传禁忌神经网络的入侵检测模型。该模型基于遗传禁忌算法的全局搜索和BP网络局部精确搜索的特性,将遗传禁忌算法和BP算法有机结合,利用遗传禁忌算法优化BP网络初始权重,同时引入小生境技术改进遗传禁忌算法。实验表明,改进的遗传禁忌算法优化BP网络用于入侵检测能提高入侵检测的效率,降低误警率,可在一定程度上提高入侵检测系统的准确率。

For high omission rate and false alarm rate in intrusion detection system,this paper proposed a tabu-based genetic neural network intrusion detection model. The model was based on genetic tabu algorithm of global search and BP global network of local search precision features, combined genetic tabu algorithm and BP algorithm, and used genetic tabu algorithm initial weights of BP network, at the same time, used niching technology to improve genetic tabu algorithm. Experiments show that the improved genetic tabu algorithm optimizing the BP network for intrusion detection can improve the efficiency of intrusion detection, lower false positive rate, improve the accuracy in the intrusion detection system to some extent.