Sample-efficient Learning of Infinite-horizon Average-reward MDPs with General Function Approximation.
Jianliang HeHan ZhongZhuoran YangPublished in: ICLR (2024)
Keyphrases
- function approximation
- average reward
- infinite horizon
- markov decision processes
- reinforcement learning
- optimal policy
- efficient learning
- policy iteration
- long run
- model free
- finite horizon
- temporal difference
- total reward
- reinforcement learning algorithms
- stochastic games
- policy evaluation
- discount factor
- average cost
- markov decision process
- td learning
- actor critic
- partially observable
- state space
- discounted reward
- learning algorithm
- finite state
- optimal control
- decision problems
- dynamic programming
- policy gradient
- learning tasks
- markov decision problems
- function approximators
- state action
- radial basis function
- partially observable markov decision processes
- sufficient conditions
- dec pomdps
- multi agent
- action space
- supervised learning
- linear programming
- markov chain
- multistage
- fixed point