Offline-Boosted Actor-Critic: Adaptively Blending Optimal Historical Behaviors in Deep Off-Policy RL.
Yu LuoTianying JiFuchun SunJianwei ZhangHuazhe XuXianyuan ZhanPublished in: CoRR (2024)
Keyphrases
- actor critic
- optimal control
- reinforcement learning
- approximate dynamic programming
- policy gradient
- temporal difference
- dynamic programming
- average reward
- reinforcement learning algorithms
- neuro fuzzy
- function approximation
- gradient method
- optimal solution
- control problems
- policy iteration
- long run
- objective function
- control policy
- optimal policy
- neural network