BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Drupal//recurring_events_ical//2.0//EN
BEGIN:VEVENT
UID:d9e531b4-8fe8-40bd-be8a-95a42f9f14c8@support.access-ci.org
DTSTAMP:20260305T130447Z
DTSTART:20260306T180000Z
DTEND:20260306T213000Z
SUMMARY:GeoAI on HPC: from single-task paradigm to multiple-task Geoscience
  Foundation Models
DESCRIPTION:This workshop explores the evolving landscape of Geospatial Art
 ificial Intelligence (GeoAI) in High-Performance Computing (HPC) environme
 nts. We'll start with the foundational concepts of GeoAI, explaining commo
 n tasks and how AI is applied to geospatial data like satellite imagery. 
 Next, we'll introduce the practical tools that facilitate this work and se
 t up a working environment on HPC Clusters. We'll examine the TorchGeo lib
 rary, which offers a comprehensive toolkit for training models on geospati
 al datasets within PyTorch. We'll also cover TerraTorch, a fine-tuning and
  benchmarking toolkit that extends capabilities for Geospatial Foundation 
 Models. Finally, we'll present several case studies to demonstrate the tr
 ansition from traditional, single-task GeoAI models—such as a model trai
 ned to identify specific land cover types—to the latest advancements in 
 Geoscience Foundation Models. These larger, more flexible models can perfo
 rm specific geoscience tasks with significantly less new data than require
 d for traditional models. We will also introduce three methods for perform
 ing these tasks with HPC, with a main focus on two of them through online 
 experiments: interactive jobs via OoD and batch jobs.
URL:https://support.access-ci.org/events/8885
END:VEVENT
END:VCALENDAR