Transience in Countable MDPs.
Stefan KieferRichard MayrMahsa ShirmohammadiPatrick TotzkePublished in: CoRR (2020)
Keyphrases
- average cost
- markov decision processes
- long run
- reinforcement learning
- factored mdps
- finite state
- optimal policy
- finite horizon
- state space
- markov chain
- dynamic programming
- finite number
- linear programming
- policy iteration
- markov decision process
- average reward
- decision theoretic planning
- approximate dynamic programming
- infinite horizon
- markov decision problems
- planning under uncertainty
- linear program
- multistage
- reward function
- stationary policies
- probabilistic planning
- semi markov decision processes
- initial state
- total cost
- action sets
- state and action spaces
- decision diagrams
- data sets
- real time dynamic programming
- policy search
- partially observable
- reinforcement learning algorithms
- transition probabilities
- optimal control
- multi agent