Reinforcement Learning in Parametric MDPs with Exponential Families.
Sayak Ray ChowdhuryAditya GopalanOdalric-Ambrym MaillardPublished in: AISTATS (2021)
Keyphrases
- reinforcement learning
- exponential family
- markov decision processes
- maximum likelihood
- state space
- graphical models
- log likelihood
- state and action spaces
- closed form
- statistical models
- density estimation
- markov decision process
- model free
- optimal policy
- mixture model
- missing values
- order statistics
- policy iteration
- dynamic programming
- continuous state and action spaces
- reward function
- learning algorithm
- variational methods
- markov decision problems
- action space
- function approximators
- bayesian networks
- probability density function
- information theoretic
- multiscale
- markov chain monte carlo
- maximum a posteriori
- missing data
- generative model