A Natural Language-Inspired Multi-label Video Streaming Traffic Classification Method Based on Deep Neural Networks

Yan Shi, Dezhi Feng, and Subir Biswas Electrical and Computer Engineering, Michigan State University, East Lansing, MI Abstract: This paper presents a deep-learning based traffic might not scale well and need updates to work under the new classification method for identifying multiple streaming video traffic conditions. Growth in video streaming traffic is arguably sources at the same time within an encrypted tunnel. The work the most significant recent change in network traffic, yet there defines a novel feature inspired by Natural Language are only a limited number of researches targeting video Processing (NLP) that allows existing NLP techniques to help streaming protocols [7]-[9]. The feature extraction method is (where multiple types of network traffic occur at the same time) described, and a large dataset containing video streaming and is left out of the existing research as well but happens quite often web traffic is created to verify its effectiveness. Results are in real-world situations. The targeted traffic type needs to be obtained by applying several NLP methods to show that the extended to cover these changes. We also show the ability to learning using deep learning methods. The trend has prompted achieve zero-shot learning with the proposed method.