/EEG-adversarial-robustness

Primary LanguageJupyter Notebook

Code for IEEE MLSP 2022 "Adversarial Training for the Adversarial Robustness of EEG-based Brain-Computer Interfaces"

Supplementary materials

Result of DeepCNN:

Accuracy of DeepCNN on UFGSM, TLM-UAP, and SAP examples generated to fool undefended models, and models trained with HBaR, TRADES, MART, PGD, and PGD with HBaR on MI, P300, and ERN (from left to right).

Result of ShallowCNN:

Accuracy of ShallowCNN on UFGSM, TLM-UAP, and SAP examples generated to fool undefended models, and models trained with HBaR, TRADES, MART, PGD, and PGD with HBaR on MI, P300, and ERN (from left to right).

Citing this work

If you use this code in your work, please cite the accompanying paper:

@inproceedings{li2022adversarial,
  title={Adversarial Training for the Adversarial Robustness of EEG-Based Brain-Computer Interfaces},
  author={Li, Yunhuan and Yu, Xi and Yu, Shujian and Chen, Badong},
  booktitle={2022 IEEE 32nd International Workshop on Machine Learning for Signal Processing (MLSP)},
  pages={1--6},
  year={2022},
  organization={IEEE}
}

Requirements