Whittle index based Q-learning for restless bandits with average reward.
Konstantin E. AvrachenkovVivek S. BorkarPublished in: Autom. (2022)
Keyphrases
- average reward
- optimal policy
- reinforcement learning
- discounted reward
- policy iteration
- state action
- model free
- markov decision processes
- optimal control
- hierarchical reinforcement learning
- long run
- reinforcement learning algorithms
- sample path
- stochastic games
- actor critic
- semi markov decision processes
- dynamic programming
- state space
- optimality criterion
- td learning
- total reward
- function approximation
- stochastic systems
- rl algorithms
- semi markov
- reward function
- infinite horizon
- decision problems
- markov decision process
- initial state
- markov chain
- partially observable markov decision processes
- state and action spaces
- markov decision problems
- temporal difference
- finite state
- fixed point
- least squares
- policy gradient
- reinforcement learning methods
- computational complexity
- multi agent