Improving Generalization in Visual Reinforcement Learning via Conflict-aware Gradient Agreement Augmentation.
Siao LiuZhaoyu ChenYang LiuYuzheng WangDingkang YangZhile ZhaoZiqing ZhouXie YiWei LiWenqiang ZhangZhongxue GanPublished in: ICCV (2023)
Keyphrases
- reinforcement learning
- visual information
- state space
- policy gradient
- computer vision
- learning process
- low level
- visual features
- function approximation
- learning algorithm
- image gradient
- visual perception
- model free
- conflict resolution
- markov decision processes
- edge detection
- high level
- neural network
- resolving conflicts
- robotic control
- gradient information
- visual analysis
- human vision
- temporal difference
- learning problems