We present a supervised learning method to learn the propagator map of a dynamical system from partial and noisy observations. In our computationally cheap and easy-to-implement framework, a neural network consisting of random feature maps is trained sequentially by incoming observations within a data assimilation procedure. By employing Takens’s embedding theorem, the network is trained on delay coordinates. We show that the combination of random feature maps and data assimilation, called RAFDA, outperforms standard random feature maps for which the dynamics is learned using batch data.
Skip Nav Destination
Research Article| October 12 2021
Combining machine learning and data assimilation to forecast dynamical systems from noisy partial observations
Georg A. Gottwald ;
Georg A. Gottwald, Sebastian Reich; Combining machine learning and data assimilation to forecast dynamical systems from noisy partial observations. Chaos 1 October 2021; 31 (10): 101103. https://doi.org/10.1063/5.0066080
Download citation file:
Don't already have an account? Register
You could not be signed in. Please check your credentials and make sure you have an active account and try again.
Could not validate captcha. Please try again.