A Theory of Regularized Markov Decision Processes.
Matthieu GeistBruno ScherrerOlivier PietquinPublished in: ICML (2019)
Keyphrases
- markov decision processes
- finite state
- optimal policy
- state space
- transition matrices
- policy iteration
- finite horizon
- planning under uncertainty
- reinforcement learning
- decision theoretic planning
- dynamic programming
- partially observable
- infinite horizon
- reinforcement learning algorithms
- average reward
- least squares
- decision processes
- action space
- markov decision process
- average cost
- reachability analysis
- factored mdps
- action sets
- objective function
- reward function
- markov decision problems
- risk sensitive
- heuristic search
- semi markov decision processes
- model based reinforcement learning