A Research of Network Applications Classification Based on Deep Learning
Nowadays, the huge traffic generated by a growing number of network applications occupies enormous network bandwidth and increases the burden of network management. The ability to identify and categorize network applications accurately is crucial for learning network traffic conditions, finding people’s online behavior and accelerating the development of the Internet. The prior traffic classification methods often have unstable recognition rate and high computational complexity, which affects the network traffic management and application categories monitoring. Therefore, this paper proposes a method of using the deep learning technology to classify network applications. First, we propose a network application classification model based on Deep Belief Network (DBN). Then we construct a DBN-based model suitable for network applications classification with the Tensorflow framework. Finally, the classification performances of this DBN-based model and the BP-based model are compared on the real data sets. The experimental results show that the applications classification model based on DBN has higher classification accuracy for P2P applications.
KeywordsDeep learning Deep belief network Network applications classification
This work was supported by a grant from the Key Research and Development Program of Zhejiang (No. 2017C03058), Zhejiang Provincial Key Laboratory of New Network Standards and Technologies (NNST) (No. 2013E10012).
- 2.Yu, K., Jia, L., Chen, Y., et al.: The yesterday, today and tomorrow of deep learning. J. Comput. Res. Dev. 50(9), 1799–1804 (2013)Google Scholar
- 5.Internet assigned numbers authority [EB/OL] (2008). http://www.iana.org
- 8.Wang, D., Zhang, L., Yuan, Z., et al.: Characterizing application behaviors for classifying P2P traffic. In: International Conference on Computing, Networking and Communications, pp. 21–25. IEEE (2014)Google Scholar
- 10.Wang, X., Li, Y.: The Introduction and Improvement of EDA, p. 225. Xidian University Press, Xi’an (2005)Google Scholar
- 11.Le, Q.V.: Building high-level features using large scale unsupervised learning. IEEE (2013)Google Scholar
- 12.Oravec, M., Podhradsky, P.: Medical image compression by backpropagation neural network and discrete orthogonal transforms. WIT Trans. Biomed. Health 4 (1970)Google Scholar