Provably efficient representation selection in Low-rank Markov Decision Processes: from online to offline RL.
Weitong ZhangJiafan HeDongruo ZhouAmy ZhangQuanquan GuPublished in: UAI (2023)
Keyphrases
- markov decision processes
- low rank
- reinforcement learning
- decision theoretic planning
- optimal policy
- finite state
- state space
- policy iteration
- reinforcement learning algorithms
- state and action spaces
- missing data
- matrix factorization
- action space
- linear combination
- convex optimization
- dynamic programming
- transition matrices
- partially observable
- markov decision process
- matrix completion
- kernel matrix
- infinite horizon
- singular value decomposition
- average reward
- average cost
- low rank matrix
- reward function
- model free
- function approximation
- high order
- semi supervised
- machine learning