Provably Efficient Exploration in Constrained Reinforcement Learning: Posterior Sampling Is All You Need.
Danil ProvodinPratik GajaneMykola PechenizkiyMaurits KapteinPublished in: CoRR (2023)
Keyphrases
- reinforcement learning
- markov chain monte carlo
- metropolis hastings
- function approximation
- probability distribution
- posterior distribution
- random sampling
- learning algorithm
- optimal policy
- monte carlo
- reinforcement learning algorithms
- model free
- robotic control
- sampling algorithm
- state space
- parameter estimation
- sampling strategies
- multi agent reinforcement learning
- sampling methods
- gaussian process
- optimal control
- parameter space
- transfer learning
- markov chain
- multi agent
- probabilistic model
- dynamic programming
- reinforcement learning methods
- posterior probability
- proposal distribution
- policy search