A Tale of Two Efficient Value Iteration Algorithms for Solving Linear MDPs with Large Action Space.
Zhaozhuo XuZhao SongAnshumali ShrivastavaPublished in: AISTATS (2023)
Keyphrases
- markov decision processes
- state space
- action space
- policy iteration
- stochastic shortest path
- factored mdps
- continuous state spaces
- reinforcement learning
- algebraic decision diagrams
- computational complexity
- markov decision process
- planning under uncertainty
- markov decision problems
- partially observable markov decision processes
- heuristic search
- function approximators
- control policies
- orders of magnitude
- optimal policy
- sufficient conditions
- linear programming