1 paper
Artem Artemev, Rui Xia, Benjamin M. Boyd +4
Many machine learning techniques rely on approximating a loss function's curvature, but this is notoriously hard to do at the scale of modern deep networks. Surprisingly, no previo…