Anuj Karpatne’s project, “Lake-GPT: Building a Foundational Model for Aquatic Science,” is one of the first 35 projects to receive compute time support through the National Artificial Intelligence Research Resources (NAIRR) pilot program, marking an important milestone in connecting U.S. researchers and educators to the computational, data, and training resources needed to advance artificial intelligence (AI).
The NAIRR pilot awards are a collaborative effort led by the National Science Foundation in collaboration with other U.S. federal agencies and are a result of President Joe Biden’s landmark Executive Order on the Safe, Secure and Trusted Development and Use of AI, providing researchers and students with access to critical AI resources and data.
Kalpatne, an associate professor in the Department of Computer Science and core faculty in the Sanghani Center for Artificial Intelligence and Data Analytics, was one of 10 awardees invited to speak at a White House event hosted by the Office of Science, Technology and Opportunity Policy on Frontiers in AI Research on May 6 to announce the launch of the NAIRR pilot program. Kalpatne is also one of two awardees invited to give a longer talk about the NAIRR project at the AI for National Competitiveness Expo hosted in Washington, D.C. by the Special Competitive Research Project.
The second group of NAIRR pilot awards, announced in late May, include Debswapna Bhattacharya, associate professor of computer science, and Xuan Wang, assistant professor of computer science and core faculty member at the Sanghani Center.
All NAIRR pilot awardees receive six months of support and access to advanced computing systems funded by the National Science Foundation or supported by the Department of Energy for AI research. Karpatne is using the AI capabilities of Oak Ridge National Laboratory’s Summit supercomputer to model water quality in lakes and reservoirs across the U.S.; Bhattacharya is using graphics processing units (GPUs) from the Texas Advanced Computing Center to power the underlying AI models for pandemic forecasting; and Wang is using the Pittsburgh Supercomputing Center’s Neocortex as an allocated resource.
Construction of new basic models in hydrosphere science
Karpatne’s research explores how advances in large-scale foundational models such as ChatGPT can spur similar breakthroughs at the forefront of AI research, enabling a scientific revolution in areas where the nature of data is quite different from that of industry data, including internet-scale text and image data.
In the NAIRR project, Karpatne and his team are leveraging new advances in the emerging field of ecological knowledge-guided machine learning to create a new class of foundational models in aquatic sciences, called Lake-GPTs, to model a range of processes relevant to water quality in lakes and reservoirs.
This is a clear departure from the mainstream practice of building “black box” AI models that rely solely on observations contained in the data, but also leverage the wealth of scientific knowledge available in many domains in various formats, including aquatic sciences, such as the laws of conservation of mass and energy.
“Water supply and quality crises are deepening in many parts of the world as climate change and rising water demand continue to put added pressure on already fragile water resources,” Carpatone said. “Understanding how to best respond to the current crisis requires exploring scenarios of change in both water use and climate trajectories that can guide further actions to meet societal needs while protecting precious water resources.”
Carpatone is collaborating with lake scientists and ecologists from Virginia Tech and other universities.
- Keiran Carey, professor of freshwater ecosystem sciences and the Roger Moore and Mojdeh Khatam Moore faculty fellow in the Department of Biological Sciences;
- Mary Lofton, postdoctoral researcher in the Department of Biological Sciences
- Paul Hanson, distinguished research professor at the University of Wisconsin-Madison Center for Limnology and principal investigator of the National Science Foundation Macrosystems Biology grant supporting this research.
- Bennett McAfee, a doctoral student under Hanson’s supervision
- Abhilash Neog, a PhD student supervised by Mr. Karpatne
- Sepideh Fatemi Kolasgan, a doctoral student supervised by Karpatne
- Alka Do is a PhD graduate from Karpatne’s group and is now a Distinguished Staff Fellow at Oak Ridge National Laboratory.
“We hope that our research will not only enable new models in aquatic sciences to predict water quality in lakes and reservoirs, but also uncover new properties of lake processes and interactions at the macrosystem scale, advancing scientific knowledge,” Karpatne said.
