Abstract: Considering current intrusion detection system with high misinformation rate and low detection rate, this paper applies fuzzy Self- Organizing Feature Map(SOFM) neural network to intrusion detection. After being trained, the fuzzy SOFM network can become a stable nerve network system and identify a network normal and abnormal behavior effectively. Experimental results show that using the KDD99 databases, intrusion detection rate is more than 92% when the misinformation rate is below 3%.

Key words: intrusion detection, neural network, fuzzy technology, Self-Organizing Feature Map(SOFM)

摘要： 针对目前入侵检测系统误报率过高、检测率不高和对未知入侵检测能力有限的缺陷，提出一种基于模糊SOFM的网络入侵检测方法，经训练后可形成一个稳定的神经网络系统，有效地识别网络正常行为和异常行为。采用KDD99数据集对系统进行实验，结果表明，系统在保持误报率低于3%的情况下，入侵检测率最高可以达到92%以上。

关键词: 入侵检测, 神经网络, 模糊技术, 自组织特征映射

CLC Number: 

• TP393