Intrusion Detection Algorithm Based On Machine Learning

The development of the Internet brings people both convenience and network security problems.Traditional network security technologies such as firewall,data encryption,and user authentication system are all passive defense techniques which have a good performance for known attacks,but not for unknown attacks.Different from these traditional technologies,Intrusion Detection System(IDS),is an innovative and proactive network security technology,which has become a hot topic in research in recent years.With the era of big data coming,Machine Learning(ML)technology develops rapidly.IDS based on machine learning is proved practical.Therefore,we do our research about intrusion detection algorithm based on machine learning in this paper.Intrusion data has some features that will affect the performance of IDS.We summary these features as multicomponent,data imbalance,time-varying and unknown attacks.Traditional IDS based on ML usually only think about one single ML algorithm,which has limited performance.How to design a framework which can combine different algorithms scientifically is a noteworthy topic.The works of this paper include:Firstly,we propose a new IDS framework called HMLD which conclude MH-DE module,MH-ML module and MEM module.HMLD can use different modelling methods for different attacks so that it can improve the detecting performance;Secondly,we propose the Clussifilter algorithm,and integrate it into the HMLD framework;Thirdly,we research the KDDCUP99 dataset and on the basis of optimizing and integrating KDDCUP99,we build new datasets KDD-I and KDD-II,which is more suitable for the IDS project.In this paper,we use KDDCUP99 dataset to evaluate the performance of HMLD.The experimental results show that HMLD can reach 96.70%accuracy in KDDCUP99 which is better than prior works.