A linear programming based approach for composite-action Markov decision processes.
Zhicong ZhangShuai LiXiaohui YanLiangwei ZhangPublished in: RAIRO Oper. Res. (2019)
Keyphrases
- markov decision processes
- linear programming
- dynamic programming
- action space
- decision theoretic planning
- discounted reward
- policy iteration
- optimal policy
- state space
- action sets
- finite state
- linear program
- markov decision problems
- reinforcement learning
- average reward
- average cost
- np hard
- reachability analysis
- risk sensitive
- planning under uncertainty
- transition matrices
- state and action spaces
- finite horizon
- factored mdps
- expected reward
- optimal solution
- model based reinforcement learning
- reward function
- markov decision process
- integer programming
- infinite horizon
- partially observable
- stochastic games
- decision processes
- learning algorithm