A Regularized Implicit Policy for Offline Reinforcement Learning.
Shentao YangZhendong WangHuangjie ZhengYihao FengMingyuan ZhouPublished in: CoRR (2022)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- action selection
- markov decision process
- actor critic
- function approximation
- action space
- state space
- policy gradient
- partially observable
- control policies
- policy evaluation
- approximate dynamic programming
- partially observable environments
- reinforcement learning algorithms
- reward function
- function approximators
- reinforcement learning problems
- control policy
- markov decision processes
- temporal difference
- policy iteration
- model free
- least squares
- state action
- learning algorithm
- state and action spaces
- markov decision problems
- continuous state spaces
- approximate policy iteration
- learning problems
- continuous state
- real time
- average reward
- finite state
- agent learns
- control problems
- robotic control
- partially observable domains
- model free reinforcement learning
- inverse reinforcement learning
- decision problems
- machine learning
- asymptotically optimal
- infinite horizon
- transfer learning
- learning process
- multi agent