Trading performance for stability in Markov decision processes.
Tomás BrázdilKrishnendu ChatterjeeVojtech ForejtAntonín KuceraPublished in: J. Comput. Syst. Sci. (2017)
Keyphrases
- markov decision processes
- state space
- dynamic programming
- optimal policy
- finite state
- reinforcement learning
- transition matrices
- factored mdps
- average reward
- decision processes
- policy iteration
- reachability analysis
- reward function
- planning under uncertainty
- decision theoretic planning
- model based reinforcement learning
- reinforcement learning algorithms
- infinite horizon
- finite horizon
- risk sensitive
- action sets
- average cost
- action space
- partially observable
- least squares
- real time dynamic programming
- machine learning
- state and action spaces
- supply chain
- decision diagrams
- long run