On the Episode Duration Distribution in Fixed-Policy Markov Decision Processes.
Itamar ArelAndrew S. DavisPublished in: FLAIRS Conference (2010)
Keyphrases
- markov decision processes
- optimal policy
- policy iteration
- markov decision process
- finite horizon
- infinite horizon
- average reward
- state and action spaces
- average cost
- action space
- reward function
- partially observable
- finite state
- state space
- reinforcement learning
- decision processes
- decision problems
- policy evaluation
- dynamic programming
- expected reward
- decision theoretic planning
- markov decision problems
- discounted reward
- total reward
- reinforcement learning algorithms
- transition matrices
- reachability analysis
- long run
- action sets
- policy iteration algorithm
- state dependent
- partially observable markov decision processes
- continuous state spaces
- factored mdps
- planning under uncertainty
- multistage
- stationary policies
- control policies
- model free
- real time dynamic programming
- initial state
- expected cost
- semi markov decision processes
- model based reinforcement learning
- objective function