Safe reinforcement learning for dynamical systems using barrier certificates.
Qingye ZhaoYi ZhangXuandong LiPublished in: Connect. Sci. (2022)
Keyphrases
- dynamical systems
- reinforcement learning
- state space
- reinforcement learning methods
- partially observable
- hidden state
- differential equations
- dynamic systems
- policy search methods
- reinforcement learning algorithms
- nonlinear dynamical systems
- phase space
- function approximation
- dynamical behavior
- temporal difference
- optimal policy
- control theory
- markov decision processes
- predictive state representations
- dynamic programming
- nonlinear dynamics
- qualitative simulation
- linear systems
- optimal control
- action space
- markov decision problems
- continuous state
- partially observable markov decision processes
- particle filter
- video sequences
- learning problems
- discrete dynamical systems
- multi agent