Mildly Conservative Q-Learning for Offline Reinforcement Learning.
Jiafei LyuXiaoteng MaXiu LiZongqing LuPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- function approximation
- reinforcement learning algorithms
- model free
- state space
- markov decision processes
- temporal difference learning
- action selection
- optimal policy
- policy iteration
- learning algorithm
- temporal difference
- real time
- continuous state and action spaces
- control problems
- transfer learning
- optimal control
- stochastic approximation
- multi agent reinforcement learning
- reinforcement learning methods
- relational reinforcement learning
- state action space
- dynamic programming
- multi agent
- average reward
- markov decision process
- supervised learning
- learning process
- continuous state spaces
- policy search
- hierarchical reinforcement learning
- machine learning
- continuous state
- exploration strategy
- cooperative