Partial Policy Iteration for L1-Robust Markov Decision Processes.
Chin Pang HoMarek PetrikWolfram WiesemannPublished in: CoRR (2020)
Keyphrases
- markov decision processes
- policy iteration
- optimal policy
- reinforcement learning
- factored mdps
- sample path
- state space
- finite state
- infinite horizon
- transition matrices
- fixed point
- average reward
- markov decision process
- policy evaluation
- dynamic programming
- least squares
- model free
- approximate dynamic programming
- reinforcement learning algorithms
- finite horizon
- planning under uncertainty
- markov decision problems
- decision processes
- average cost
- state and action spaces
- stochastic games
- machine learning
- discounted reward
- markov games
- decision problems
- random walk
- partially observable
- policy iteration algorithm