Reinforcement Learning in Linear MDPs: Constant Regret and Representation Selection.
Matteo PapiniAndrea TirinzoniAldo PacchianoMarcello RestelliAlessandro LazaricMatteo PirottaPublished in: CoRR (2021)
Keyphrases
- reinforcement learning
- markov decision processes
- reward function
- state space
- optimal policy
- online learning
- decision theoretic planning
- markov decision process
- machine learning
- function approximation
- total reward
- model based reinforcement learning
- lower bound
- function approximators
- continuous state and action spaces
- action sets
- reinforcement learning algorithms
- temporal difference
- optimal control
- loss function
- e learning
- partially observable
- action space
- model free
- average reward
- policy evaluation
- approximate dynamic programming
- policy search
- factored mdps
- worst case
- learning algorithm