Improved Algorithms for Misspecified Linear Markov Decision Processes.
Daniel VialAdvait ParulekarSanjay ShakkottaiR. SrikantPublished in: AISTATS (2022)
Keyphrases
- markov decision processes
- policy iteration
- optimal policy
- reinforcement learning
- factored mdps
- state space
- learning algorithm
- finite state
- finite horizon
- computational complexity
- dynamic programming
- reachability analysis
- action sets
- decision theoretic planning
- planning under uncertainty
- average reward
- average cost
- model free
- convergence rate