Learning Dynamical Systems via Koopman Operator Regression in Reproducing Kernel Hilbert Spaces

In light of recent machine learning progress, it is appealing to ask if the properties of dynamical systems can be estimated ( learned) from empirical data.