Max-Plus Linear Approximations for Deterministic Continuous-State Markov Decision Processes.
Eloïse BerthierFrancis R. BachPublished in: IEEE Control. Syst. Lett. (2020)
Keyphrases
- markov decision processes
- continuous state
- finite state
- action space
- reinforcement learning
- partially observable markov decision processes
- continuous state spaces
- optimal policy
- state space
- control policies
- dynamic programming
- finite horizon
- policy iteration
- state action
- reinforcement learning algorithms
- markov decision process
- robot navigation
- function approximators
- partially observable
- average reward
- stochastic processes
- real valued
- planning problems
- reward function
- multi agent
- markov chain
- machine learning
- stochastic games
- average cost
- infinite horizon
- decision problems