English
Arabic
Chinese (Simplified)
Dutch
Esperanto
French
German
Greek
Italian
Japanese
Korean
Portuguese
Russian
Spanish
Urdu
Funding for the project, titled “DnC2S: Decision and Control of Complex Systems: A Data‐Driven Framework,” was awarded via the Department of Energy’s Office of Science.
ORNL is partnering with DOE’s Pacific Northwest National Laboratory, Arizona State University and the University of California, Santa Barbara, to develop a more unified theoretical foundation based on probabilistic graphical models, or PGMs.
PGMs show promise to enhance a range of data-intensive tasks including many in machine learning, such as uncertainty quantification and reinforcement learning. However, numerous fundamental questions remain before PGMs can be applied to decision and control systems both in the private sector and across the DOE research landscape.
“The plan is to develop a data-driven approach to model construction for complex systems such as scientific user facilities, wind turbine farms and the electrical grid,” said ORNL’s Frank Liu, principal investigator and distinguished scientist. “We need to define the constraints and uncertainties inherent in these systems in order to use AI to operate them more efficiently.”
In recognition of AI’s potential, ORNL has made a significant internal research investment toward an AI initiative aimed at accelerating time to solution for researchers across the scientific spectrum and equipping cross-cutting research teams with leading-edge data capabilities to tackle the most complex scientific and national security challenges.
Through the initiative, ORNL researchers have shown that machine learning algorithms can be used to extract information from signals with low signal-to-noise ratios. They have also developed algorithms capable of accelerating modeling and simulation with very little training data. Plus, ORNL has designed novel biomimetic neuromorphic devices capable of detecting epileptic seizures.
The initiative harnesses the laboratory’s unique suite of expertise, compute capabilities and facilities to accelerate the adoption of AI in scientific fields of interest to DOE.
This same collection of assets will be critical in ensuring the success of Liu’s project. “The diverse expertise and infrastructures at both ORNL and PNNL provide basic building blocks that we can rely on to better understand the foundational math underpinning these complex systems,” Liu said.
Specifically, research within the project is divided among four interdependent focus areas – model construction, uncertainty quantification, decision and control and continual learning – that each have unique desired outcomes. Yet, taken together, they ensure the success of the overall effort.
A fifth cross-cutting area, known as application demonstration, will validate results from the individual efforts and apply these findings, via proof-of-concept demonstrations, to a range of applications including building energy efficiency, reactor power control and science user facility control.
“AI has become an intrinsic part of science and engineering, but a lot of research remains to be done,” said Liu. “This project will build on the foundational work of the lab’s AI initiative and develop new capabilities to strengthen the use of AI in decision-making and control processes for complex systems.”
The award was part of $16 million total invested by DOE to develop the potential of artificial intelligence and machine learning across the physical sciences and the management of complex systems.
ORNL is managed by UT-Battelle for the U.S. Department of Energy’s Office of Science, the single largest supporter of basic research in the physical sciences in the United States. DOE’s Office of Science is working to address some of the most pressing challenges of our time. For more information, please visit https://www.energy.gov/science.
