Better Robustness by More Coverage: Adversarial Training with Mixup Augmentation for Robust Fine-tuning.
Chenglei SiZhengyan ZhangFanchao QiZhiyuan LiuYasheng WangQun LiuMaosong SunPublished in: CoRR (2020)
Keyphrases
- fine tuning
- fine tuned
- high robustness
- fine tune
- viable alternative
- multi agent
- digital image watermarking
- training samples
- real time
- multiscale
- computationally efficient
- robust estimation
- semi fragile watermarking
- database
- greater robustness
- geometric attacks
- training algorithm
- partial occlusion
- supervised learning
- training set
- video sequences
- feature selection
- neural network