11 December, 10 o’clock, Salle de réunion SCAI, Jussieu.
Batiment Esclangon 1er étage
Participer à la réunion Zoom
https://us02web.zoom.us/j/89174956656
Geosciences have long-standing experience in modeling, forecasting, or estimating complex dynamical systems like the atmosphere or the ocean. Most of these models came from physical laws and are described by PDE. Usually, sparse and noisy observations of such systems are available. The first need to produce a forecast is to estimate initial conditions. This is usually done via Data Assimilation (DA), a set of methods that optimally combines a dynamical model and observations, focusing on system state estimation. In variational formalism, it’s a PDE-constrained optimization problem that requires adjoint modeling to calculate gradients. This field is very close to Machine Learning (ML) in the sense that both learn from data.
ML algorithms have demonstrated impressive results of spatiotemporal forecasting, but to do so it needs dense data which is rarely the case in earth sciences. Also, tools provided by the deep learning community based on automatic differentiation are particularly suitable for variational DA, avoiding explicit adjoint modeling.
What motivates this discussion is that physics-based model is often
incomplete and machine learning can provide a learnable class of model
while data assimilation can provide dense data.
The a similar talk can be found here, and an early conference paper can be found here.