Reinforcement Learning for Infinite-Horizon Average-Reward MDPs with Multinomial Logistic Function Approximation.
Jaehyun ParkDabeen LeePublished in: CoRR (2024)
Keyphrases
- function approximation
- average reward
- infinite horizon
- reinforcement learning
- markov decision processes
- optimal policy
- finite horizon
- state space
- long run
- policy iteration
- model free
- total reward
- stochastic games
- discounted reward
- optimal control
- decision problems
- temporal difference
- partially observable
- finite state
- markov decision process
- state action
- average cost
- reinforcement learning algorithms
- temporal difference learning
- dynamic programming
- probabilistic model
- state and action spaces
- function approximators
- learning algorithm
- markov decision problems
- multi agent
- multistage
- td learning
- machine learning
- action space
- rl algorithms
- supervised learning
- initial state
- policy gradient
- transfer learning
- reinforcement learning methods
- approximate dynamic programming
- partially observable markov decision processes
- state variables
- policy evaluation
- actor critic