Learning a Low-dimensional Representation of a Safe Region for Safe Reinforcement Learning on Dynamical Systems.
Zhehua ZhouOzgur S. OguzMarion LeiboldMartin BussPublished in: CoRR (2020)
Keyphrases
- dynamical systems
- reinforcement learning
- partially observable
- reinforcement learning methods
- learning algorithm
- predictive state representations
- learning process
- state space
- hidden state
- qualitative simulation
- adaptive control
- differential equations
- dynamic systems
- learning problems
- supervised learning
- function approximation
- control theory
- learning tasks
- optimal policy
- connectionist networks
- dynamical models
- multi agent
- dynamical behavior