Primary Convener:
Hamed Alemohammad, Clark University

Convener:
David Bell, Universities Space Research Association Moffett Field
Wenwen Li, Arizona State Univ
Ankur Kumar, University of Alabama in Huntsville

Early Career Convener:
Rufai Balogun, Clark University

Chair:
David Bell, Universities Space Research Association Moffett Field
Wenwen Li, Arizona State Univ
Michael K. Barker, NASA Goddard Space Flight Center
Ankur Kumar, University of Alabama in Huntsville

Foundation models (FMs) are transforming scientific discovery across disciplines. These generalized AI models are trained through self-supervised methods for a plethora of downstream applications and eliminate the need for extensive labeled datasets. This session will explore their growing role in Earth, Space, and Planetary sciences. Topics of interest include but are not limited to:1) Pretraining strategies and training data curation2) Intermediate learning strategies to enhance FMs’ domain adaptability3) Development of multimodal, multiscale, and spatiotemporal-aware FMs4) Emerging strategies for for fine-tuning FMs5) Cross-domain FMs that bridge Earth, Space, and Planetary sciences6) Interpretability, uncertainty quantification, and trustworthiness of FM outputs7) Key science applications of FMsThe session also encourages discussion on broad community involvement toward development of open FMs for science that are accessible for all. Our goal is to foster discussion about how FMs are shaping the next generation of scientific tools and workflows.

Index Terms
1906 Computational models, algorithms
1920 Emerging informatics technologies
1942 Machine learning

Cross-Listed:
P - Planetary Sciences
SH - SPA-Solar and Heliospheric Physics
A - Atmospheric Sciences
OS - Ocean Sciences

Suggested Itineraries:
Machine Learning and AI

Neighborhoods:
1. Science Nexus