Singularly perturbed linear programs and Markov decision processes.
Konstantin AvrachenkovJerzy A. FilarVladimir GaitsgoryAndrew StillmanPublished in: Oper. Res. Lett. (2016)
Keyphrases
- linear program
- markov decision processes
- dynamic programming
- linear programming
- state space
- stochastic programming
- reinforcement learning
- optimal policy
- average cost
- policy iteration
- transition matrices
- optimal solution
- finite state
- finite horizon
- objective function
- factored mdps
- decision theoretic planning
- simplex method
- planning under uncertainty
- reinforcement learning algorithms
- decision processes
- average reward
- markov decision process
- original data
- extreme points
- np hard
- model based reinforcement learning
- partially observable
- reachability analysis
- action space
- action sets
- markov games
- infinite horizon
- markov decision problems
- stochastic games