Off-Policy Average Reward Actor-Critic with Deterministic Policy Search.
Naman SaxenaSubhojyoti KhastigirShishir KolathayaShalabh BhatnagarPublished in: CoRR (2023)
Keyphrases
- policy gradient
- average reward
- actor critic
- markov decision processes
- optimal policy
- reinforcement learning
- long run
- partially observable markov decision processes
- stochastic games
- model free
- reinforcement learning algorithms
- function approximation
- markov chain
- state action
- policy iteration
- reward function
- state space
- approximation methods
- gradient method
- optimal control
- multi agent
- variance reduction
- belief state
- neural network
- single agent
- dynamical systems
- dynamic programming
- temporal difference
- infinite horizon
- decision problems
- learning algorithm