摘要
在众多的入侵检测系统中,当训练集合发生变化的时候,都需要重新进行训练学习.在现实世界中,由于数据的统计模型可能是未知的或是时间相关的,因此要求学习系统的很多参数只能随着部分输入数据进行增量式地调整更新.文章提出了有导师学习的ESOM模型理论背景并将它应用到入侵检测的在线识别领域中,同时给出相应的实验结果和分析结论.
In an increasingly computerized world, there is an overwhelming demand for systems with the ability of on-line, selfadaptive learning to process many real world data analysis. In many IDS, when the train datas are changed, IDS will be required to learn again. In real world, as in the eases when the statistical models of data are unknown or time-dependent, and the parameters of the learning system need to be updated incrementally while only a partial glimpse of incoming data is available. This paper presents some theoretical background for the supervised evolving self-organizing map (ESOM) model and further apply it in solving on-line intrusion detection problems. Some results are reported along with analysis and concluding remarks.
出处
《小型微型计算机系统》
CSCD
北大核心
2005年第9期1510-1513,共4页
Journal of Chinese Computer Systems