DENAS: Deep Neuroevolution at Scale
Samson Jacobs | 21-ERD-026
We are developing new methods to identify architectures for neural networks at high-performance computing scale and training neural networks to address scientific applications at these larger scales. This research advances the capabilities of neural networks to address many high-priority national security and scientific initiatives such as nuclear stockpile stewardship, inertial confinement fusion, and modeling for drug design, including countermeasures to COVID-19.
Publications, Presentations, and Patents
Sam Ade Jacobs, Tuan Tran, Derek Mariscal, Tim Moon, Blagoje Djordjevic, Michael Wyatt, Brian Van Essen, and Tammy Ma. "Learning Augmentation from Data: A Case Study in Scientific Diagnostic Simulations." 63rd Annual Meeting of the APS Division of Plasma Physics, 2021.
Sam Ade Jacobs. "Enabling Rapid COVID-19 Small Molecule Drug Design Through Scalable Deep Learning of Generative Models." Invited Panelist Talk, Richard Tapia Diversity in Computing Conference (online), 2021.
Sam Ade Jacobs, "Learning to Learn at HPC Scale." Invited Talk, LLNL COMP Scholar Program, 2021.