Optimal Conservative Offline RL with General Function Approximation via Augmented Lagrangian.
Paria RashidinejadHanlin ZhuKunhe YangStuart RussellJiantao JiaoPublished in: ICLR (2023)
Keyphrases
- function approximation
- reinforcement learning
- augmented lagrangian
- model free
- learning tasks
- dynamic programming
- temporal difference
- radial basis function
- function approximators
- convex optimization
- reinforcement learning algorithms
- neural network
- state space
- temporal difference methods
- td learning
- constrained optimization
- image denoising
- transfer learning
- learning algorithm