Efficient Policies for Stationary Possibilistic Markov Decision Processes.
Nahla Ben AmorZeineb El KhalfiHélène FargierRégis SabaddinPublished in: ECSQARU (2017)
Keyphrases
- markov decision processes
- optimal policy
- stationary policies
- markov decision process
- decision processes
- state space
- average cost
- reinforcement learning
- dynamic programming
- reward function
- finite state
- reachability analysis
- transition matrices
- infinite horizon
- factored mdps
- planning under uncertainty
- policy iteration
- average reward
- linear program
- non stationary
- risk sensitive
- decentralized control
- partially observable
- reinforcement learning algorithms
- decision problems
- state and action spaces
- total reward
- finite horizon
- partially observable markov decision processes
- decision theoretic planning
- markov decision problems
- macro actions
- action sets
- long run
- action space
- discounted reward
- model based reinforcement learning
- sufficient conditions
- policy iteration algorithm