A new approach for internet traffic classification: Artificial Bee Colony algorithm-OSELM

Abstract

IP traffic classification is significant for Internet service providers and other private and public organizations, for example in various tasks such as bandwidth scheduling, network error detection, Internet service quality analysis, pricing for users who use specific Internet applications, tracking Internet traffic data and security for specific government agencies. Online sequential extreme learning machine is a method of online learning solving the problem of observing data of different sizes. However, the input weights and OSELM hidden layer biases are randomly determined. This results in an incorrect classification result. In the proposed method, the data are classified using anonline sequential extreme learning machine algorithm. Certainsoftware based on artificial bee colony algorithm (ABC -OSELM) was developed to select the parameters of the sequential fast online machine learning algorithm. The simulation results show that the proposed method has achieved a 7% improvement in accuracy criteria compared to the base paper.