BruSLiAttack.github.io

README

This is for releasing the source code of the paper BruSLeAttack: Query-Efficient Score-Based Sparse Adversarial Attack

Poster: Poster

Artifacts: Artifacts.

These clean and perturbed samples & artifacts can be tested on Google Cloud Vision (Test them for free!!!)

Cite our research:

@inproceedings{Vo2024,
    title = {BruSLeAttack: A Query-Efficient Score-Based Black-Box Sparse Adversarial Attack},
    year = {2024},
    booktitle = {International Conference on Learning Representations (ICLR)},
    author = Viet Quoc Vo and Ehsan Abbasnejad and Damith C. Ranasinghe},
}

TODO

• Add the testing code.