Abstract
This presentation provides an overview of recent advancements in Geospatial Foundation Models (FMs) for Earth observation through satellite remote sensing data.
Scaling Machine Learning with Supercomputing on Large Remote Sensing Datasets
The rapid proliferation of data in the information age has increased the complexity of data-driven challenges across various fields of science and engineering. This shift has sparked a transformation in Machine Learning (ML), moving towards unsupervised and self-supervised representation learning, as well as multimodal approaches. Significant advancements have been made in deep learning algorithms for both image and language-based tasks, including applications in Earth Science. These innovations leverage the synergies between self-supervised learning and the growing availability of supercomputing resources, leading to the development of Foundation Models (FMs). This presentation provides an overview of recent advancements in Geospatial Foundation Models (FMs) for Earth observation through satellite remote sensing data. It highlights the ongoing activities of the AI4EO use case within the European Pilot for Exascale (EUPEX) project and explores the challenges and opportunities in interdisciplinary research at the intersection of machine learning, supercomputing, and remote sensing.
