Entropic Regularization of Markov Decision Processes.
Boris BelousovJan PetersPublished in: Entropy (2019)
Keyphrases
- markov decision processes
- finite state
- state space
- reinforcement learning
- optimal policy
- dynamic programming
- reinforcement learning algorithms
- transition matrices
- regularization parameter
- partially observable
- policy iteration
- reward function
- decision theoretic planning
- planning under uncertainty
- decision processes
- finite horizon
- risk sensitive
- factored mdps
- reachability analysis
- model based reinforcement learning
- markov decision process
- average cost
- state abstraction
- state and action spaces
- discounted reward
- semi markov decision processes
- average reward
- probabilistic planning
- decision theoretic
- action sets
- infinite horizon