Argonne National Lab Receives Grant for Interdisciplinary AI Research
Argonne National Laboratory has been awarded nearly $3 million in Department of Energy funding for two projects aimed at advancing artificial intelligence and machine learning technology.
The projects will explore how AI can be used to manage large data sets or create better outcomes where little data is available, Argonne said Monday.
One of the projects is focused on the creation of surrogate models that can lower the costs of running complex simulations.
Romit Maulik, an Argonne computational scientist who has been working on modeling strategies, said the award will enable significant reductions in time-to-solutions and costs.
“For example, instead of using a massive machine to simulate the climate, we could run many smaller cheaper simulations,” he explained.
Maulik is carrying out the project in collaboration with Los Alamos National Laboratory, the Illinois Institute of Technology in Chicago and Johns Hopkins University.
The second project is aimed at using machine learning-accelerated simulations to improve forecasting, data assimilation and prediction of the frequency of extreme events.
Argonne said that current modeling technology is not accurate enough to predict the frequency of events like cascading blackouts and cold snaps.
Mihai Anitescu, an Argonne mathematician working on the project, explained that current machine learning techniques are not fit to analyze extreme events with little past data.
The grant money was awarded by DOE’s Office Advanced Scientific Computing Research. The office awarded a total of five interdisciplinary projects aimed at leveraging AI for work in DOE’s national laboratories.
The projects were selected through a DOE funding opportunity announcement titled “Data-intensive Scientific Machine Learning and Analysis.”
Category: Federal Civilian
Tags: Argonne National Laboratory artificial intelligence Department of Energy federal civilian machine learning Mihai Anitescu Romit Maulik