Abstract

Action recognition has been heavily employed in many applications such as autonomous vehicles, surveillance, etc, where its robustness is a primary concern. With the recent uprising of adversarial attack which automatically and strategically look to computing data pertubation in order to fool well-trained classifiers, this project looks into the vulnerability of existing classifiers against adversarial attack and how to improve their resistence and robustness.

Resources

  1. He Wang, Yunfeng Diao, Zichang Tan, Guodong Guo.  Defending Black-box Skeleton-based Human Activity Classifiers.   The AAAI Conference on Artificial Intelligence (AAAI). 2023 Conference
     Paper   Video   Code   BibTex
    @inproceedings{wang2023defending_1,
  author = {He Wang and Yunfeng Diao and Zichang Tan and Guodong Guo},
  title = {Defending Black-box Skeleton-based Human Activity Classifiers},
  booktitle = {The AAAI Conference on Artificial Intelligence (AAAI)},
  year = {2023}
}

  2. He Wang, Yunfeng Diao.  Defending Black-box Classifiers by Bayesian Boundary Correction.   arxiv. 2023 Preprint
     Paper   BibTex
    @misc{wang2023defending,
  author = {He Wang and Yunfeng Diao},
  title = {Defending Black-box Classifiers by Bayesian Boundary Correction},
  series = {arxiv},
  year = {2023}
}

  3. Yunfeng Diao, He Wang, Tianjia Shao, Yongliang Yang, Kun Zhou, David Hogg.  Understanding the Vulnerability of Skeleton-based Human Activity Recognition via Black-box Attack.   arxiv. 2022 Preprint
     Paper   BibTex
    @misc{diao2022understanding,
  author = {Yunfeng Diao and He Wang and Tianjia Shao and Yongliang Yang and Kun Zhou and David Hogg},
  title = {Understanding the Vulnerability of Skeleton-based Human Activity Recognition via Black-box Attack},
  series = {arxiv},
  year = {2022}
}

  4. Yufeng Diao, Tianjia Shao, Yongliang Yang, Kun Zhou, He Wang.  BASAR: Black-box attack on skeletal action recognition.   The Conference on Computer Vision and Pattern Recognition (CVPR). 2021 Conference
     Paper   Supplement   Video   Code   Errata   BibTex
    @inproceedings{diao2021basar,
  author = {Yufeng Diao and Tianjia Shao and Yongliang Yang and Kun Zhou and He Wang},
  title = {BASAR: Black-box attack on skeletal action recognition},
  booktitle = {The Conference on Computer Vision and Pattern Recognition (CVPR)},
  pages = {7597--7607},
  year = {2021}
}

  5. He Wang, Feixiang He, Zhexi Peng, Tianjia Shao, Yongliang Yang, Kun Zhou, David Hogg.  Understanding the robustness of skeleton-based action recognition under adversarial attack.   The Conference on Computer Vision and Pattern Recognition (CVPR). 2021 Conference
     Paper   Supplement   Video   Code   BibTex
    @inproceedings{wang2021understanding,
  author = {He Wang and Feixiang He and Zhexi Peng and Tianjia Shao and Yongliang Yang and Kun Zhou and David Hogg},
  title = {Understanding the robustness of skeleton-based action recognition under adversarial attack},
  booktitle = {The Conference on Computer Vision and Pattern Recognition (CVPR)},
  pages = {14656--14665},
  year = {2021}
}

Acknowledgement

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under grant agreement No 899739 CrowdDNA, EPSRC (EP/R031193/1), NSF China (No. 61772462, No. U1736217), RCUK grant CAMERA (EP/M023281/1, EP/T014865/1) and the 100 Talents Program of Zhejiang University.