The complexity of synchronizing Markov decision processes.
Laurent DoyenThierry MassartMahsa ShirmohammadiPublished in: J. Comput. Syst. Sci. (2019)
Keyphrases
- markov decision processes
- finite state
- optimal policy
- state space
- transition matrices
- reinforcement learning
- policy iteration
- decision theoretic planning
- dynamic programming
- decision processes
- action sets
- reachability analysis
- partially observable
- reinforcement learning algorithms
- model based reinforcement learning
- risk sensitive
- finite horizon
- average cost
- average reward
- decision problems
- policy evaluation
- state abstraction
- markov decision process
- factored mdps
- real time dynamic programming
- stationary policies
- state and action spaces
- reward function
- least squares