Low-Rank MDPs with Continuous Action Spaces.
Andrew BennettNathan KallusMiruna OprescuPublished in: CoRR (2023)
Keyphrases
- action space
- low rank
- markov decision processes
- state space
- state and action spaces
- reinforcement learning
- matrix factorization
- continuous state spaces
- linear combination
- missing data
- rank minimization
- real valued
- continuous state
- convex optimization
- low rank matrix
- singular value decomposition
- high dimensional data
- semi supervised
- high order
- markov decision process
- stochastic processes
- optimal policy
- markov decision problems
- kernel matrix
- average cost
- action selection
- policy iteration
- data analysis
- reinforcement learning algorithms
- data sets
- dynamic programming
- infinite horizon
- average reward
- single agent
- partially observable
- multi agent
- least squares
- active learning