The NuTS (Numérique en Terre Solide) network is organizing a thematic event focused on the intersection of artificial intelligence and Earth sciences. This two-day gathering will be held at the Institut de Physique du Globe de Paris from Monday noon, July 7, to early afternoon on Tuesday, July 8. The event features two half-day sessions with invited speakers designed to foster interdisciplinary dialogue: one emphasizing scientific challenges in the geosciences community, and the other highlighting methodological advances in AI. A convivial gathering between the sessions will provide opportunities for informal exchange and networking. The primary goal is to stimulate new collaborations between researchers in Earth sciences and artificial intelligence, encouraging mutual understanding and joint innovation. The organizers also intend this event to be the first in a recurring series that strengthens ties between these communities. Participation in the event is free, with venue and refreshments covered by the organizers; however, attendees are expected to use their own research funds for travel and accommodation. Attendees will have the opportunity to present their work in form of a poster during the poster session.

The event will take place in the main amphitheater of the institut de physique du globe de Paris, 1 Rue Jussieu 75005 Paris.

Monday, July 7: AI in Earth Sciences

13:30 — Welcoming coffee

13:55 — Introductory speech

14h 

• Quentin Bletery — AI-driven earthquake and tsunami early warning
• Charles Le Losq — From lava to eruptions: How AI can help advance volcanological research.
• Jannes Münchmeyer — What can deep learning earthquake catalogs actually tell us?
  
• Catherine Pothier — Unsupervised data mining technique to extract patterns corresponding to aseismic SLOW slip events: application to the Izmit section of the north Anatolian fault
  

15h30

• Spotlight poster presentations 

• Coffee break

16h30

• Antoine Lucas — Can AI be leveraged to revisit unsolved problems in critical zone research?
• Giuseppe Costantino — Signal extraction from noisy geospatial data: detecting transient deformations through spatiotemporal deep learning
  
• Joachim Rimpot — Self-Supervised learning for the exploration of large seismological datasets
  
• Andrea Tomasi — Towards the prediction of elastic (seismic) anisotropy in the upper mantle using supervised deep-learning

18h Cocktail and buffet

Tuesday, July 8 : Session AI 

8h30 Coffee and croissants 

9h 

• Christophe Rigotti — Bridging local and global explanations of tree ensemble predictions using co-clustering of SHAP values
• Sylvain Lobry  — Foundation models for remote sensing: which one(s)?

10h30 Coffee break 

11h

• Erwan Allys — Unsupervised modeling and separation of seismic signals with scattering transforms
• Stéphanie Allassonnière — Taking advantage of geometry in Variational Autoencoders for data representation and augmentation 
  

12h30 Closing speech