@misc{indiciae081d55b013cc,
  title = {Gradient Concentration, Not Weight Saliency, Explains Representation-Level Class Unlearning},
  author = {Billel Habbati and Alessio Merlo and Luca Verderame and Meriem Guerar},
  year = {2026},
  url = {https://arxiv.org/abs/2607.21353},
  note = {Source identifier: 2607.21353}
}