Markov Decision Processes with Discounted Rewards: New Action Elimination Procedure.
Abdellatif SemmouriMostafa JourhmaneBahaa Eddine ElbaghazaouiPublished in: CBI (2021)
Keyphrases
- markov decision processes
- discounted reward
- action space
- decision theoretic planning
- optimal policy
- finite state
- average reward
- finite horizon
- reinforcement learning
- state space
- dynamic programming
- infinite horizon
- transition matrices
- policy iteration
- action sets
- partially observable
- reachability analysis
- average cost
- expected reward
- planning under uncertainty
- risk sensitive
- factored mdps
- decision processes
- reinforcement learning algorithms
- decentralized control
- state and action spaces
- total reward
- model based reinforcement learning
- sequential decision making under uncertainty
- initial state
- reward function
- semi markov decision processes
- state abstraction
- long run
- decision problems