Learning Loss for Test-Time Augmentation

Ildoo Kim, Younghoon Kim, Sungwoong Kim

Data augmentation has been actively studied for robust neural networks. Most of the recent data augmentation methods focus on augmenting datasets during the training phase. At the…