摘要： 介绍并改进了一种基于动态模型的,用于入侵检测系统的机器学习算法。该算法抽取出IP分组和TCP连接的23个属性域,经过训练得到规则集。结合改进后的模型,使用该规则集对1999 DARPA入侵检测离线评估数据集进行了测试,取得了较好的结果。
关键词:
入侵检测系统,
机器学习,
TCP/IP,
动态模型
Abstract: An intrusion detection system based on a new nonstationary machine learning technique is introduced and improved. It extracts 23 attributes from IP packet and incoming TCP connections, and induces a set of rules after training. With the modified model, the IDS in the 1999 DARPA intrusion detection offline evaluation data set is tested, and a better result is got.
中图分类号:
郭敬宇. 一种用于入侵检测的改进的动态机器学习模型[J]. , 2004, 30(14): 103-.
GUO Jingyu. An Improved Nonstationary Machine Learning Model for Intrusion Detection[J]. , 2004, 30(14): 103-.