Full Gradient Deep Reinforcement Learning for Average-Reward Criterion.
Tejas PagareVivek S. BorkarKonstantin AvrachenkovPublished in: L4DC (2023)
Keyphrases
- average reward
- reinforcement learning
- optimality criterion
- policy gradient
- markov decision processes
- optimal policy
- model free
- policy iteration
- actor critic
- long run
- function approximation
- state action
- rl algorithms
- state and action spaces
- stochastic games
- semi markov decision processes
- reinforcement learning algorithms
- discounted reward
- temporal difference
- state space
- hierarchical reinforcement learning
- total reward
- partially observable markov decision processes
- machine learning
- learning algorithm
- reward function
- dynamic programming
- reinforcement learning methods
- multi agent
- state abstraction
- evaluation function
- td learning
- least squares