Variance-constrained actor-critic algorithms for discounted and average reward MDPs.
Prashanth L. A.Mohammad GhavamzadehPublished in: Mach. Learn. (2016)
Keyphrases
- average reward
- markov decision processes
- policy iteration
- optimal policy
- model free
- actor critic
- long run
- reinforcement learning
- semi markov decision processes
- discounted reward
- stochastic games
- partially observable markov decision processes
- markov chain
- finite state
- state and action spaces
- learning algorithm
- infinite horizon
- markov decision process
- computational complexity
- rl algorithms
- policy gradient
- state space
- fixed point
- dynamic programming
- state action
- average cost
- markov decision problems
- approximate dynamic programming
- search algorithm
- least squares