Learning a Low-Dimensional Representation of a Safe Region for Safe Reinforcement Learning on Dynamical Systems.
Zhehua ZhouOzgur S. OguzMarion LeiboldMartin BussPublished in: IEEE Trans. Neural Networks Learn. Syst. (2023)
Keyphrases
- dynamical systems
- reinforcement learning
- reinforcement learning methods
- learning process
- predictive state representations
- hidden state
- learning algorithm
- state space
- partially observable
- connectionist networks
- qualitative simulation
- action selection
- dynamic systems
- linear systems
- function approximation
- learning problems
- policy search
- markov decision processes
- policy search methods