Chengcheng Guo, Bo Zhao, Yanbing Bai

arXiv preprint

Comprehensive library and empirical study of coreset selection methods for deep learning, finding that random selection remains a strong baseline across many settings.

https://arxiv.org/abs/2204.08499