Tree-RNN: Tree structural recurrent neural network for network traffic classification

作者：

Highlights：

• We divide a large classification into small classifications with the tree structure.

• A specific classifier is set for each small classification after division.

• Tree-RNN can complement each other in performance with multiple classifiers.

• The specific divided rules are introduced for any number of traffic classes.

• The cosine is used to judge the similarity between classes.

摘要

•We divide a large classification into small classifications with the tree structure.•A specific classifier is set for each small classification after division.•Tree-RNN can complement each other in performance with multiple classifiers.•The specific divided rules are introduced for any number of traffic classes.•The cosine is used to judge the similarity between classes.

论文关键词：Network traffic classification,Deep learning,Recurrent neural network,Tree structure,End-to-end

论文评审过程：Received 7 August 2020, Revised 31 October 2020, Accepted 21 November 2020, Available online 2 December 2020, Version of Record 5 December 2020.

论文官网地址：https://doi.org/10.1016/j.eswa.2020.114363