A Linearly Relaxed Approximate Linear Program for Markov Decision Processes.
Chandrashekar LakshminarayananShalabh BhatnagarCsaba SzepesváriPublished in: IEEE Trans. Autom. Control. (2018)
Keyphrases
- linear program
- markov decision processes
- factored mdps
- dynamic programming
- optimal solution
- linear programming
- average cost
- approximate dynamic programming
- policy evaluation
- state space
- finite state
- optimal policy
- policy iteration
- stationary policies
- stochastic programming
- semi infinite
- transition matrices
- reinforcement learning
- objective function
- partially observable
- column generation
- planning under uncertainty
- action space
- action sets
- decision processes
- model based reinforcement learning
- average reward
- decision theoretic planning
- np hard
- finite horizon
- exact solution
- markov decision process
- approximate solutions
- infinite horizon
- markov decision problems
- reward function
- machine learning
- lower bound
- search space
- knapsack problem
- optimal control