YaTC

This repository contains the code for the paper: <br> Yet Another Traffic Classifier: A Masked Autoencoder Based Traffic Transformer with Multi-Level Flow Representation <br> In Thirty-Seventh AAAI Conference on Artificial Intelligence (AAAI 2023).

Overview

<img src="YaTC.png">

The training strategy of YaTC is divided into two stages: pre-training stage and fine-tuning stage.

The Pre-trained Model

Path: ./output_dir/pretrained-model.pth

Pre-trained model Link

Datasets Information

Path: ./data

Datasets Link

Number of class

• ISCXVPN2016_MFR: 7
• ISCXTor2016_MFR: 8
• USTC-TFC2016_MFR: 20
• CICIoT2022_MFR: 10

Dependency

python=3.8
timm=0.3.2
torch=1.9.0
numpy=1.19.5
scikit-learn=0.24.2

Code Information

• data_process.py: process traffic pcap files of flows to MFR matires
• models_YaTC: the code of the classifier model
• pre-train.py: the code of pre-training stage
• fine-tune.py: the code of fine-tuning stage

Pre-training

python pre-train.py --batch_size 128 --blr 1e-3 --steps 150000 --mask_ratio 0.9

Fine-tuning

python fine-tune.py --blr 2e-3 --epochs 200 --data_path ./data/ISCXVPN2016_MFR --nb_classes 7

Contact-Info

Link to our laboratory: SJTU-NSSL

Ruijie Zhao <br> Email: ruijiezhao@sjtu.edu.cn

Mingwei Zhan <br> Email: mw.zhan@sjtu.edu.cn

Reference

R. Zhao, M. Zhan, X. Deng, Y. Wang, Y. Wang, G. Gui, and Z. Xue, ``Yet Another Traffic Classifier: A Masked Autoencoder Based Traffic Transformer with Multi-Level Flow Representation,'' in AAAI Conference on Artificial Intelligence (AAAI'23), Washington, United States, Feb. 7--14, 2023, pp. 1--8.