Critic-Actor for Average Reward MDPs with Function Approximation: A Finite-Time Analysis.
Prashansa PandaShalabh BhatnagarPublished in: CoRR (2024)
Keyphrases
- function approximation
- reinforcement learning
- average reward
- markov decision processes
- actor critic
- model free
- policy iteration
- optimal policy
- policy gradient
- reinforcement learning algorithms
- state and action spaces
- temporal difference
- td learning
- radial basis function
- temporal difference learning
- long run
- policy evaluation
- monte carlo
- neural network
- semi markov decision processes
- approximate dynamic programming
- action space
- learning tasks
- least squares
- machine learning