Trading Performance for Stability in Markov Decision Processes
Tomás BrázdilKrishnendu ChatterjeeVojtech ForejtAntonín KuceraPublished in: CoRR (2013)
Keyphrases
- markov decision processes
- finite state
- state space
- reinforcement learning
- optimal policy
- transition matrices
- reinforcement learning algorithms
- dynamic programming
- average cost
- reachability analysis
- partially observable
- decision theoretic planning
- policy iteration
- infinite horizon
- action space
- decision processes
- finite horizon
- risk sensitive
- average reward
- planning under uncertainty
- markov decision process
- factored mdps
- action sets
- model based reinforcement learning
- reward function
- multi agent
- state abstraction
- state and action spaces
- semi markov decision processes