A Survey on Feature Selection Techniques for Internet Traffic Classification

Feature selection technique has a great importance in Internet traffic classification. Machine learning (ML) algorithms have been generally applied in novel traffic classification. In this paper we provide an overview of three major approaches to classify different categories of Internet traffic: Port based approach, Payload based approach, Statistical-based approach. This paper also explain feature selection algorithms, which are classified into 3 methods: Filter method, Wrapper method, Embedded Method along with their benefits and limitations and also provides an overview of some of the feature selection technique present in literature. The aim of the survey gives a brief idea about feature selection techniques which can be applied to many machine learning algorithms to avoid problems like class imbalance, concept drift, low efficiency, and low classification rate etc.