期刊名称：International Journal of Computer Science & Technology
印刷版ISSN：2229-4333
电子版ISSN：0976-8491
出版年度：2013
卷号：4
期号：3
页码：192-195
语种：English
出版社：Ayushmaan Technologies
摘要：During the last two decades the research community is extensively advising the use of Machine Learning systems for Network Security (anomaly detection). This paper concentrates on the design procedure of machine learning systems, explaining the basic terminology, also specifying the procedures on creating the training and test datasets, on choosing among several performance measures available. Experiments comparing the state of the art machine learning algorithms, by taking ROC as the performance measure are conducted and results were given. The algorithms compared are Adaboost, Bagging, KNN, SVM and MLP. For the experiments KDD 99 Intrusion Dataset and Email spam database are used. Before concluding, several kinds of attacks aimed at machine learning systems are specified.
