DiffStitch: Boosting Offline Reinforcement Learning with Diffusion-based Trajectory Stitching.
Guanghe LiYixiang ShanZhengbang ZhuTing LongWeinan ZhangPublished in: CoRR (2024)
Keyphrases
- reinforcement learning
- learning algorithm
- anisotropic diffusion
- function approximation
- image stitching
- diffusion process
- optimal policy
- model free
- markov decision processes
- temporal difference
- state space
- real time
- weak learners
- ensemble learning
- feature selection
- ensemble methods
- multi agent
- weak classifiers
- transfer learning
- action selection
- learning process
- boosting algorithms
- trajectory data
- diffusion processes
- combining multiple
- machine learning
- long run
- reinforcement learning algorithms
- edge detection
- tensor field
- optical flow