AI4OAC: workshop on AI for Ocean, Atmosphere and Climate Dynamics

This workshop is co-organized in the framework of EUR Isblue (Interdisciplinary Graduate School for the Blue Planet, Brest), LEFE/MANU project IA-OAC, ANR project Melody and ANR AI Chair OceaniX (PI: R. Fablet, IMT Atlantique), which both aim at developing AI frameworks for Ocean-Atmosphere-Climate Science. Broadly speaking, the availability of large amounts of simulation and observation data and the emergence of artificial intelligence technologies (big data architectures, GPUs, prolific learning) open up new opportunities to explore open questions in ocean, atmospheric and climate sciences through a data-centric paradigm, rather than a "classical" physical paradigm, one of the central scientific challenges being precisely the ability to unify these two paradigms. The workshop aims to explore, review, discuss and advance these challenges.

The program will combine keynotes (see below invited speakers) and working groups. Any participant will have the opportunity to propose a theme for a working group during an initial open forum session. Two types of working groups will be encouraged: data-challenge-like groups and more prospective working groups investigating AI-based formulations for ocean/atmosphere/climate challenges.

Feel free to contact for additional information: ronan.fablet AT imt-atlantique.fr

List of keynote speakers

• Prof. Mihai Datcu, DLR, "AI and Earth Observation"  
• Prof. Anne Sabourin, TelecomParisTech, "PCA for Multivariate Extremes"
• Dr. Samy Bengio, Google Research, "Towards Better Understanding Generalization in Deep Learning"
• Dr. Olivier Pannekoucke, CERFACS, "NN generator for PDEs"
• Dr. Gilles Tissot, INRIA Rennes, "Koopman Operator and RKHS"
• Dr. Nicolas Papadakis, CNRS/IMB, "Optimal transport for image assimilation"

Organizing committee

• Prof. Ronan Fablet, IMT Atlantique/Lab-STICC, Brest, France
• Dr. Bertrand Chapron, Ifremer, LOPS, Brest, France
• Dr. Laurent Debreu, INRIA GRA, Grenoble, France
• Prof. Patrick Gallinari, Sorbonne Univ., Paris, France
• Dr. Julien Le Sommer, CNRS, IGE, Grenoble, France
• Dr. Philippe Naveau, CNRS, LSCE, Paris, France
• Prof. Sylvie Thiria, Sorbonne Univ., Paris, France

List of working groups' themes (preliminary)

The workshop will involve both methodological and topical working groups. Below is a preliminary list of the envisioned working groups, which may be updated during the open forum session the first day of the workshop:

• Learning-based approaches for reduced-order (stochastic) modeling
• Learning & Space-time extremes
• Generative modeling for OAC processes
• Automated NN generation from PDEs
• Dynamical Systems, Optimal Transport & Physics-informed Learning
• Learning closures for ocean dynamics
• Learning-based separation of wave and eddy processes
• (End-to-end) learning and space-time interpolation of geophysical dynamics
• Learning & biogeochemical dynamics

Each workshop participant will have the opportunity to participate to at least one methodological working group and one topical working group.