SRe2L
SRe2L copied to clipboard
(NeurIPS 2023 spotlight) Large-scale Dataset Distillation/Condensation, 50 IPC (Images Per Class) achieves the highest 60.8% on original ImageNet-1K val set.
Large-scale Dataset Distillation
This is a collection of our work targeted at large-scale dataset distillation.
SCDD : Self-supervised Compression Method for Dataset Distillation .
CDA : Dataset Distillation in Large Data Era, arXiv:2311.18838.
SRe2L (
@NeurIPS'23 spotlight): Squeeze, Recover and Relabel: Dataset Condensation at ImageNet Scale From A New Perspective.
Citation
@article{yin2023dataset,
title={Dataset Distillation in Large Data Era},
author={Yin, Zeyuan and Shen, Zhiqiang},
journal={arXiv preprint arXiv:2311.18838},
year={2023}
}
@inproceedings{yin2023squeeze,
title={Squeeze, Recover and Relabel: Dataset Condensation at ImageNet Scale From A New Perspective},
author={Yin, Zeyuan and Xing, Eric and Shen, Zhiqiang},
booktitle={Thirty-seventh Conference on Neural Information Processing Systems},
year={2023},
}