xgr19 / SAM-for-Traffic-Classification

Self-attentive deep learning method for online traffic classification and its interpretability (CN21 & NetAI20)
GNU General Public License v3.0
35 stars 14 forks source link
deep-learning network-traffic-classification self-attention

SAM-for-Traffic-Classification

Paper title: Self-attentive deep learning method for online traffic classification and its interpretability

Accepted by Elsevier Computer Networks (https://doi.org/10.1016/j.comnet.2021.108267)

NetAI20 version https://github.com/xgr19/SAM-for-Traffic-Classification/tree/SAM-before-NetAI

Run the files as follow:

  1. python3 preprocess.py
  2. python3 tool.py
  3. python3 train.py

The dataset is available at http://mawi.wide.ad.jp/mawi/samplepoint-G/2020/202006101400.html

Other datasets (the code of this repo is reuseful): ISCX:https://www.unb.ca/cic/datasets/vpn.html UNIBS:http://netweb.ing.unibs.it/~ntw/tools/traces/

We have new repositories about DL algorithms on traffic classification: xgr19/Loong, xgr19/Soter, xgr19/SOTERIA, xgr19/Mousika! Especially, Mousika is about using P4 to reach nanosecond-level classification.