The US Department of Energy’s Oak Ridge National Laboratory took yet another step toward leading the AI revolution by hosting its third workshop on Artificial Intelligence for Robust Engineering and Science, or AIRES.
Robust engineering refers to the process of designing, building, and controlling systems to avoid or mitigate failures. The introductory AIRES workshop in January 2020 explored these foundations, and the follow-up workshop in January 2021 focused on the foundations of AI for constructing, deploying, and assuring the robustness of digital twins.
This year’s 3-day event was held on April 26–28 at ORNL in a virtual + in-person format and garnered 150 participants. Organizers built on the success of the first two workshops by exploring and developing the foundations of AI in knowledge-informed modeling and prediction, deployment considerations and related topics via three primary thrusts:
- “Knowledge-informed AI,” which focused on the technical challenges associated with developing robust digital models;
- “Assurance,” which focused on the technical challenges associated with assuring the robustness of digital twins and
- “Codesign Ecosystem,” which focuses on the practical challenges associated with digital twins.
Opening remarks were delivered by Doug Kothe, director of DOE’s Exascale Computing Project and associate laboratory director for computing at ORNL. Shortly thereafter, a panel discussion titled “Digital Twins and Artificial Intelligence - Overview of Needs and Challenges” featured researchers from industry, academia and the national labs.
The conference keynote, titled “Knowledge-Guided Machine Learning: A New Framework for Accelerating Scientific Discovery,” was presented by Vipin Kumar, a Regents Professor at the University of Minnesota and William Norris Endowed Chair in the Department of Computer Science and Engineering.
He noted that early efforts to merge scientific knowledge and machine learning have led to promising results, but much of this work is in isolated domains and directed toward specific applications. Although this research is still in its early stages, expectations are rising and community crosspollination is necessary to advance the state of the art.
“Knowledge-guided machine learning techniques are fundamentally more powerful than standard machine learning approaches and traditional mechanistic models used by the scientific community to address environmental problems,” said Kumar.
Talks throughout the event addressed the conference theme and various thrusts, and the conference concluded with five breakout sessions and summary presentations aimed at addressing natural systems, trusted AI for digital twins, digital twins for engineered systems, uncertainty quantification for digital twins and edge deployment for speed and power efficiency.
Pradeep Ramuhalli and David Womble cochaired the conference. “Artificial intelligence and digital twins, together with DOE’s [high-performance computing] capabilities have the potential to transform DOE science and engineering,” said Womble. “This conference is a step toward realizing that potential.”
UT-Battelle manages ORNL for the Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. The Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit energy.gov/science.