1 paper · 1 filter
Qing Zhou, Bingxuan Zhao, Tao Yang +3
Dynamic data pruning accelerates deep learning by selectively omitting less informative samples during training. While per-sample loss is a common importance metric, obtaining it c…