An Effective Encrypted Traffic Classification Method Based on Pruning Convolutional Neural Networks for Cloud Platform
Zilong Han, Erxia Li, Shichao Li, Xiaoyong Li, Chaoqun Kang, Ruiwen Deng, Yali Gao · 2021 2nd International Conference on Electronics, Communications and Information Technology (CECIT) · 2021
With the rapid development of cloud computing technology, the traffic of cloud platform also increases rapidly. Most of these traffic adopt encryption technology, the original traffic classification method is no longer effective. In this paper, we propose an efficient encrypted traffic classification method based on pruning convolution neural network: PCNN. This method does not need to extract features manually, but automatically extracts high-level features through CNN, which makes the model have generalization ability and reduces the cost. Then the pruning technology is adopted to prune the model and retain the parameters that are highly important to the model, which greatly reduces the size and computation of the model and will not affect the performance of the model. We used the public network dataset ISCXVPN2016 to verify our proposed method, the experiments show that the F1 score of our proposed approach is 93% which is improved by 3%, and the precision is 94%, which is improved by 5%.