Adversarial Attacks Leverage Interference Between Features in Superposition

Imperial College London
Under Review
Code arXiv

Code, along with guidance on how reproduce the main results of the paper, will be published upon publication.

Citation

If you found the paper useful, please consider citing:

Copied! 
@article{stevinson2025adversarialsuperposition,
  title={Adversarial Attacks Leverage Interference Between Features in Superposition},
  author={Stevinson, Edward and Prieto, Lucas and Barsbey, Melih and Birdal, Tolga},
  year={2025},
  archivePrefix={arXiv},
  primaryClass={cs.CV}
}

Contact

Please contact e.stevinson22@imperial.ac.uk for any inquiries related to this work.