Time Series Foundation Models (TSFMs) for Carbon-Aware Research and Sustainability
This in-person workshop will introduce the UCLA research community to the emerging class of Time Series Foundation Models (TSFMs) and their application to sustainability challenges such as electricity grid carbon intensity forecasting. Participants will gain hands-on experience with TSFM frameworks, open-source datasets (e.g., ElectricityMaps), and computational tools designed to enable accurate, generalizable forecasting without requiring extensive region-specific data. The workshop will emphasize reproducible workflows, integration with high-performance computing, and the potential for interdisciplinary applications across energy, environment, and policy domains.
Target Audience: Faculty, researchers, and graduate students in computer science, engineering, environmental science, data science, and policy who are interested in applying AI-driven time-series forecasting to sustainability and decarbonization problems.
Learning Outcomes:
- Understand the principles of TSFMs and their advantages over traditional forecasting methods.
- Learn how to access and utilize global carbon intensity datasets for research.
- Gain practical skills in fine-tuning and deploying TSFMs for forecasting tasks.
- Explore interdisciplinary applications of TSFMs, including energy demand forecasting, workload scheduling, and environmental policy support.
- Build capacity to apply AI tools in advancing sustainable and carbon-aware research.
This workshop will be hosted by IDRE Fellow, Dr. Kang Yang.