Optimistic Q-learning for average reward and episodic reinforcement learning.
Priyank AgrawalShipra AgrawalPublished in: CoRR (2024)
Keyphrases
- average reward
- reinforcement learning
- optimal policy
- model free
- markov decision processes
- policy iteration
- discounted reward
- state action
- reinforcement learning algorithms
- state space
- function approximation
- rl algorithms
- semi markov decision processes
- stochastic games
- actor critic
- optimality criterion
- learning algorithm
- total reward
- long run
- temporal difference learning
- machine learning
- state and action spaces
- hierarchical reinforcement learning
- multi agent
- temporal difference
- reinforcement learning methods
- policy gradient
- policy evaluation
- action selection
- supervised learning
- dynamic programming
- function approximators
- markov chain
- sufficient conditions
- approximate dynamic programming
- reward function
- td learning
- decision problems
- finite state
- fixed point