Variance Reduced Value Iteration and Faster Algorithms for Solving Markov Decision Processes.
Aaron SidfordMengdi WangXian WuYinyu YePublished in: CoRR (2017)
Keyphrases
- markov decision processes
- policy iteration
- stochastic shortest path
- factored mdps
- transition matrices
- reachability analysis
- finite state
- state space
- optimal policy
- reinforcement learning
- reinforcement learning algorithms
- semi markov decision processes
- dynamic programming
- decision theoretic planning
- partially observable
- infinite horizon
- policy evaluation
- partially observable markov decision processes
- average reward
- markov decision process
- policy iteration algorithm
- planning under uncertainty
- action sets
- markov decision problems
- continuous state spaces
- state and action spaces
- average cost
- learning algorithm
- finite horizon
- fixed point
- least squares