Support Vector Machines for TCP traffic classification

摘要: Support Vector Machines (SVM) represent one of the most promising Machine Learning (ML) tools that can be applied to problem traffic classification in IP networks. In case SVMs, there are still open questions need addressed before they generally classifiers. Having being designed essentially as techniques for binary classification, their generalization multi-class problems is under research. Furthermore, performance highly susceptible correct optimization working parameters. this paper we describe an approach based on SVM. We apply approaches solving with SVMs task statistical and a simple algorithm allows classifier perform correctly little training few hundred samples. The accuracy proposed then evaluated over three sets traces, coming from different topological points Internet. Although results relatively preliminary, confirm SVM-based classifiers very effective at discriminating generated by applications, even reduced set sizes.