基于机器学习的入侵检测方法实验与分析

Experiments and analysis for intrusion detection method based on machine learning

入侵检测系统(IDS)是保障信息安全的重要手段。分析了机器学习应用于网络连接级的异常检测模型的过程,然后建立了异常检测系统原型,以验证此方法用于IDS的可能性及所能达到的性能。实验以DARPA网络数据为例,对数据的特征进行了分析、选取及构造,并针对多种情况进行了测试。实验结果表明,该IDS系统具有很好的检测性能。最后对结果进行了分析,并得出了几个有用的结论。

Intrusion detection system(IDS)is a very important instrument in the domain of information security.Machine learning was used to construct an IDS model on network connection level in order to test the feasibility and its performance.Ex- perimental results using a set of benchmark data from DARPA have shown that the performances of this anomaly detection system are fairly high.At the end,some useful conclusions were drawn according to the results.