On Skolem-Hardness and Saturation Points in Markov Decision Processes.
Jakob PiribauerChristel BaierPublished in: ICALP (2020)
Keyphrases
- markov decision processes
- optimal policy
- state space
- dynamic programming
- finite state
- transition matrices
- policy iteration
- reinforcement learning
- reachability analysis
- decision theoretic planning
- planning under uncertainty
- factored mdps
- decision processes
- reinforcement learning algorithms
- infinite horizon
- average reward
- action sets
- computational complexity
- average cost
- first order logic
- markov decision process
- state abstraction
- action space
- finite horizon
- partially observable
- reward function
- model based reinforcement learning
- np hard
- real valued
- discounted reward
- least squares
- real time dynamic programming
- stochastic shortest path