Tung Nguyen, doctoral student in computer science, UCLA

Recent data-driven and machine learning-based approaches aim to directly solve a forecasting or forward projection task by learning a data-driven functional mapping using deep neural networks. However, these networks are trained using curated and homogeneous climate datasets for specific spatio-temporal tasks, and therefore lack the generality of physics-based, computationally intensive numerical models currently in use. for weather and climate modeling. We develop and demonstrate ClimaX, a flexible and generalizable deep learning model for weather and climate science that can be trained using heterogeneous datasets spanning different variables, spatio-temporal coverage, and physical underpinnings. ClimaX extends the Transformer architecture with new coding and aggregation blocks that enable efficient use of available compute and data while maintaining general utility. ClimaX is pre-trained with a self-supervised learning objective on CMIP6-derived climate datasets. The pre-trained ClimaX can then be fine-tuned to meet a wide range of climate and weather tasks, including those that involve atmospheric variables and spatio-temporal scales invisible during pre-training. Compared to existing data-driven benchmarks, we show that this generality in ClimaX results in superior performance on weather forecast and climate projection benchmarks, even when pretraining at higher resolutions and computational budgets. lower.

Biography :
Tung Nguyen is a second-year doctoral student in computer science at UCLA. His research interests lie at the intersection of decision making, sequence modeling, and uncertainty quantification. He also wants to anchor his research in applications to sustainability, particularly climate change.

Please take the opportunity to connect and share this video with your friends and family if you find it useful.