The Linear Program approach in multi-chain Markov Decision Processes revisited.
Eitan AltmanFlos SpieksmaPublished in: Math. Methods Oper. Res. (1995)
Keyphrases
- markov decision processes
- linear program
- dynamic programming
- linear programming
- average cost
- stationary policies
- semi infinite
- factored mdps
- optimal policy
- approximate dynamic programming
- state space
- transition matrices
- optimal solution
- finite state
- policy iteration
- finite horizon
- reinforcement learning
- stochastic programming
- decision processes
- objective function
- column generation
- planning under uncertainty
- decision theoretic planning
- np hard
- reachability analysis
- model based reinforcement learning
- action sets
- partially observable
- action space
- markov decision process
- mixed integer linear program
- reward function
- infinite horizon
- markov decision problems
- special case
- multistage
- decision problems
- average reward