Anytime-Competitive Reinforcement Learning with Policy Prior.
Jianyi YangPengfei LiTongxin LiAdam WiermanShaolei RenPublished in: NeurIPS (2023)
Keyphrases
- reinforcement learning
- optimal policy
- policy search
- markov decision process
- markov decision processes
- action selection
- partially observable environments
- function approximators
- actor critic
- approximate dynamic programming
- policy evaluation
- markov decision problems
- function approximation
- policy iteration
- action space
- partially observable
- decision problems
- control policy
- reinforcement learning problems
- state space
- state action
- reward function
- policy gradient
- reinforcement learning algorithms
- state and action spaces
- prior information
- anytime algorithms
- infinite horizon
- machine learning
- prior knowledge
- control policies
- state dependent
- dynamic programming
- continuous state spaces
- temporal difference
- policy gradient methods
- model free reinforcement learning
- finite state
- partially observable markov decision processes
- control problems
- average cost
- continuous state
- rl algorithms
- transfer learning
- reinforcement learning methods
- average reward
- optimal control
- model free