Gradient-Free Adversarial Training Against Image Corruption for Learning-based Steering.
Yu ShenLaura Y. ZhengManli ShuWeizi LiTom GoldsteinMing C. LinPublished in: NeurIPS (2021)
Keyphrases
- learning algorithm
- supervised learning
- input image
- image features
- learning process
- image segmentation
- online learning
- image data
- multiscale
- image content
- image analysis
- reinforcement learning
- edge detection
- image collections
- image classification
- low level
- learning stage
- piecewise constant
- test images
- grey level
- gradient direction
- gradient field
- training process
- image structure
- image regions
- feature points
- level set
- similarity measure