Safe reinforcement learning for probabilistic reachability and safety specifications: A Lyapunov-based approach.
Subin HuhInsoon YangPublished in: CoRR (2020)
Keyphrases
- reinforcement learning
- state space
- probabilistic model
- adaptive control
- stability analysis
- dynamical systems
- sufficient conditions
- generative model
- high level
- learning algorithm
- uncertain data
- transitive closure
- function approximation
- class of nonlinear systems
- control scheme
- reinforcement learning algorithms
- control theory
- closed loop
- markov decision processes
- optimal policy
- control system
- dynamic model
- model free
- nonlinear systems
- supervised learning
- search space
- action space
- delay insensitive
- multi agent