Faster Climate Sciences - from DSLs to AI
DescriptionAs machine learning applications continue to succeed, the industry is shifting towards optimizing AI workloads. In this paper, we make three predictions about how this will impact high-performance computing (HPC) technology. We then identify challenges and opportunities for future weather and climate simulations based on these predictions. We explore various implementation options, from purely data-driven modeling to hybrid approaches that combine data-driven and model-driven simulations. Finally, we propose a strategy for massive data compression using a deep learning model overfitted to climate data. These approaches will enable weather and climate scientists to confidently embrace the future of AI-based accelerator architectures.
TimeTuesday, June 2717:00 - 17:30 CEST
Climate, Weather and Earth Sciences