Continuous Action Reinforcement Learning From a Mixture of Interpretable Experts.
Riad AkrourDavide TateoJan PetersPublished in: IEEE Trans. Pattern Anal. Mach. Intell. (2022)
Keyphrases
- continuous action
- policy search
- reinforcement learning
- continuous state
- action space
- state space
- partially observable markov decision processes
- robot navigation
- reinforcement learning algorithms
- function approximation
- model free
- optimal policy
- finite state
- dynamic programming
- optimal control
- learning algorithm
- action selection
- temporal difference
- reward function
- markov decision processes
- policy gradient
- state dependent
- control policies
- bayesian networks
- learning problems
- expectation maximization
- markov decision process
- heuristic search
- single agent