@misc{indiciaeef7f2bf24b70,
  title = {Robustness of deep learning classification to adversarial input on GPUs: asynchronous parallel accumulation is a source of vulnerability},
  author = {Sanjif Shanmugavelu and Mathieu Taillefumier and Christopher Culver and Vijay Ganesh and Oscar Hernandez and Ada Sedova},
  year = {2025},
  url = {https://arxiv.org/abs/2503.17173},
  note = {Source identifier: 2503.17173}
}