Safe Policy Improvement in Constrained Markov Decision Processes.
Luigi BerducciRadu GrosuPublished in: ISoLA (1) (2022)
Keyphrases
- markov decision processes
- optimal policy
- policy iteration
- markov decision process
- average reward
- finite horizon
- state space
- infinite horizon
- average cost
- partially observable
- state and action spaces
- reward function
- finite state
- action space
- reinforcement learning
- dynamic programming
- decision processes
- partially observable markov decision processes
- decision problems
- discounted reward
- reachability analysis
- markov decision problems
- total reward
- policy iteration algorithm
- action sets
- long run
- multistage
- factored mdps
- risk sensitive
- continuous state spaces
- decision theoretic planning
- planning under uncertainty
- stationary policies
- transition matrices
- reinforcement learning algorithms
- policy evaluation
- continuous state
- expected reward
- control policies
- model based reinforcement learning
- state dependent
- machine learning
- sufficient conditions
- least squares
- search space
- semi markov decision processes