Skip to content

Latest commit

 

History

History
4 lines (3 loc) · 457 Bytes

File metadata and controls

4 lines (3 loc) · 457 Bytes

FeatureSpace-InstanceSpaceAugmentation

For training the supervised model, I perform data augmentation in both instance space and feature space. The instance space augmentation for the supervised model consists of some image transformation based on our domain knowledge, as well as smart augmentation. On the other hand, I implement Mixup as our feature space augmentaion for the supervised model. I intergate smart augmentation and Mixup in our system.