@misc{indiciae5c8ecb59755a,
  title = {Dataset Distillation Efficiently Encodes Low-Dimensional Representations from Gradient-Based Learning of Non-Linear Tasks},
  author = {Yuri Kinoshita and Naoki Nishikawa and Taro Toyoizumi},
  year = {2026},
  url = {https://arxiv.org/abs/2603.14830},
  note = {Source identifier: 2603.14830}
}