Geometric Policy Iteration for Markov Decision Processes.
Yue WuJesús A. De LoeraPublished in: CoRR (2022)
Keyphrases
- policy iteration
- markov decision processes
- sample path
- optimal policy
- finite state
- state space
- approximate dynamic programming
- reinforcement learning
- factored mdps
- policy evaluation
- transition matrices
- average reward
- dynamic programming
- average cost
- markov decision process
- markov decision problems
- temporal difference
- fixed point
- reinforcement learning algorithms
- least squares
- infinite horizon
- finite horizon
- action space
- decision processes
- planning under uncertainty
- discounted reward
- actor critic
- search algorithm
- linear programming
- random walk
- np hard
- multistage
- markov games
- stochastic games