1 paper
Jianhao Xu, Zhuang Yang
The existing optimizers for deep neural networks (DNNs) typically rely on either the ℓ2 norm or the ℓ∞ norm, resulting in optimizers that do not adapt well to subs…