Multi-Objective Approaches to Markov Decision Processes with Uncertain Transition Parameters.
Dimitri ScheftelowitschPeter BuchholzVahid HashemiHolger HermannsPublished in: CoRR (2017)
Keyphrases
- markov decision processes
- multi objective
- optimal policy
- finite state
- state space
- reinforcement learning
- transition matrices
- dynamic programming
- finite horizon
- factored mdps
- planning under uncertainty
- reachability analysis
- average cost
- policy iteration
- risk sensitive
- reinforcement learning algorithms
- action sets
- decision processes
- markov decision process
- decision theoretic planning
- infinite horizon
- model based reinforcement learning
- machine learning
- semi markov decision processes
- real valued
- state and action spaces
- average reward
- partially observable
- reward function
- long run
- multi objective optimization