Learning by Doing: Controlling a Dynamical System using Causality, Control, and Reinforcement Learning.
Sebastian WeichwaldSøren Wengel MogensenTabitha Edith LeeDominik BaumannOliver KroemerIsabelle GuyonSebastian TrimpeJonas PetersNiklas PfisterPublished in: CoRR (2022)
Keyphrases
- dynamical systems
- reinforcement learning
- state space
- policy search methods
- differential equations
- discrete event
- partially observable
- reinforcement learning methods
- nonlinear dynamical systems
- hidden state
- dynamical behavior
- optimal control
- linear quadratic
- dynamic systems
- linear systems
- function approximation
- phase space
- control system
- nonlinear dynamics
- partially observable markov decision processes
- adaptive control
- control method
- dynamic programming
- action selection
- markov decision processes
- immune network
- machine learning
- search space
- lyapunov exponents