Communications of the ACM
Scientists se Machine Learning to Speed Discovery of Metallic Glass
Scientists at the U.S. Department of Energy's SLAC National Accelerator Laboratory, the U.S. National Institute of Standards and Technology, and Northwestern University integrated machine learning with their experiments to discover and improve metallic gl
Credit: Yvonne Tang/SLAC National Accelerator Laboratory
Researchers from the U.S. Department of Energy's SLAC National Accelerator Laboratory, the U.S. National Institute of Standards and Technology, and Northwestern University used artificial intelligence (AI) to discover and improve metallic glass faster and less expensively.
They used SLAC's Stanford Synchrotron Radiation Lightsource (SSRL), integrating machine learning with experiments, to rapidly produce and screen hundreds of sample materials concurrently to find new metallic glass compounds 200 times faster than previously was possible.
The team mined about 50 years' worth of materials data with machine learning algorithms, and from the obtained knowledge created two sets of sample alloys. Both sets were scanned by an SSRL x-ray beam, the data fed into a database, and new machine-learning results generated and applied to the preparation of new samples that also underwent scanning and machine learning.
By the third round, the success rate for finding metallic glass had climbed from one out of 300 or 400 samples tested to one out of two or three samples tested.
From SLAC National Accelerator Laboratory
View Full Article
Abstracts Copyright © 2018 Information Inc., Bethesda, Maryland, USA
No entries found