An argument in favor of strong scaling for deep neural networks with small datasets

Open in new window