A New Structure for Internet Video Traffic Classification Using Machine Learning

Network classification is the key of network supervision, QoS (Quality of Service) guarantee, and network security. This paper devises a novel chain structure for fine-grained classification of network video streams with less memory, lower latency and higher accuracy. The paper classifies the Internet videos from the perspective of QoS. It applies modified Chi-squared algorithm to discretize data, and uses a novel structure to boost classification performance. The proposed scheme is compared with different Ensemble Learning methods and classification structures. The experimental results suggest that the proposed method is better than existing methods for video traffic classification.