Internet traffic classification based on associative classifiers

Traffic classification plays a significant role on efficient traffic management, traffic inspection, service class mapping, inspection of security and errors and network charging. Supervised machine learning algorithms were previously applied to internet traffic classification. However, they suffer from understandability and interpretability issues. To improve the understandability of the classification process for the network administrators and human experts, we propose the application of the associative classifiers (AC) to the Internet traffic classification. In this paper, three associative classification algorithms (CBA, CMAR, and CPAR) have applied to the traffic statistics-based classification problem. Our conducted experiments on real network dataset show that AC has the potential to become an excellent tool for analyzing Internet traffic.