TY  - RPRT
TI  - Dataset Distillation Efficiently Encodes Low-Dimensional Representations from Gradient-Based Learning of Non-Linear Tasks
AU  - Yuri Kinoshita
AU  - Naoki Nishikawa
AU  - Taro Toyoizumi
PY  - 2026
UR  - https://arxiv.org/abs/2603.14830
ID  - 2603.14830
ER  -