摘要
提出将层次自组织特征映射神经网络算法应用于网络异常检测,算法自顶向下逐层生成神经网络结构并细化聚类,将神经元的组织和连接方式从平面扩展到层次与平面连接相结合,大大加速了获胜神经元的搜寻过程。基于此种算法,设计并实现了网络异常检测系统中的数据分析器HSOMDA,在DARPA 1999数据集上的实验表明其具有较高的检测性能和时间性能。
This paper presents an application of a hierarchical self-organizing map (HSOM) algorithm for network anomaly detection. By growing in a top-down approach and extending the connections of neurons from horizontal dimension to both horizontal and hierarchical dimension, HSOM significantly accelerates the searching of winning neurons. Based on HSOM, a data analyzer in network anomaly detection system HSOMDA is designed and implemented. The results of experiments on DARPA 1999 dataset demonstrate its effectiveness and efficiency in anomaly detection.
出处
《计算机应用与软件》
CSCD
北大核心
2006年第5期3-4,8,共3页
Computer Applications and Software
基金
国家自然科学基金项目(60403033)。
关键词
异常检测
聚类
层次自组织特征映射
Anomaly detection Clustering Hierarchical self-organizing map