Autonomous MultiScale: Embedded Machine Learning for Smart Simulations
Peer-Timo Bremer | 22-SI-004
The goal of the project is to enable a new class of multiscale simulations by developing an autonomous framework to efficiently and effectively couple different computing scales through advanced machine learning models. If successful, it has the potential to transform many computational science applications and enable coupled simulations at unprecedented scale and fidelity.
Publications, Presentations, and Patents
S. Fridovich-Keil, B. R. Bartoldson, J. Diffenderfer, B. Kailkhura, and P.-T. Bremer, March 2022, “Models Out of Line: A Fourier Lens on Distribution Shift Robustness.” ICML 2022 Workshop on Principles of Distribution Shift, Baltimore, MD. 2022. https://icml.cc/Conferences/2022/ScheduleMultitrack?event=20543.
S. Fridovich-Keil, B. R. Bartoldson, J. Diffenderfer, B. Kailkhura, and P.-T. Bremer, 2022 “Models Out of Line: A Fourier Lens on Distribution Shift Robustness." NeurIPS 2022. New Orleans, LA. 2022. https://paperswithcode.com/paper/models-out-of-line-a-fourier-lens-on. Accepted.
P.T. Bremer. "Multi-Modal Data Representations: Challenges and Opportunities." Presentation at 1st AI@DOE Workshop. 2021.
P.-T. Bremer. "Cognitive Simulations Connecting Simulations and Experiments." Presentation, Southern Cal Edison. 2022.
P.-T. Bremer. "Autonomous MultiScale (AMS) Embedded Machine Learning for Smart Simulations." 2022 WPD/WSC CogSim/ML Workshop, LLNL. 2022.
J. Gaffney. "Autonomous MultiScale (AMS) Embedded Machine Learning for Smart Simulations." Poster at External Review Committee presentation for PLS. 2022.