logo
ResearchBunny Logo
Scaling deep learning for materials discovery

Engineering and Technology

Scaling deep learning for materials discovery

A. Merchant, S. Batzner, et al.

This groundbreaking research by Amil Merchant, Simon Batzner, Samuel S. Schoenholz, Muratahan Aykol, Gowoon Cheon, and Ekin Dogus Cubuk shows how scaled graph neural networks can revolutionize materials discovery by uncovering 2.2 million new stable structures from a dataset of 48,000 crystals. This includes complex materials with unique elemental combinations never found before!

00:00
00:00
Playback language: English
Abstract
This paper demonstrates that scaled graph neural networks (GNNs) can significantly improve the efficiency of materials discovery. By training GNNs on a dataset of 48,000 stable crystals and employing active learning, the researchers discovered 2.2 million new stable structures, representing an order-of-magnitude increase. These discoveries include materials with more than four unique elements, previously challenging to find. The resulting large dataset also enabled the training of highly accurate learned interatomic potentials, useful for molecular dynamics simulations and predicting properties like ionic conductivity.
Publisher
Nature
Published On
Nov 29, 2023
Authors
Amil Merchant, Simon Batzner, Samuel S. Schoenholz, Muratahan Aykol, Gowoon Cheon, Ekin Dogus Cubuk
Tags
graph neural networks
materials discovery
active learning
stable structures
interatomic potentials
ionic conductivity
machine learning

Related Publications

Explore these studies to deepen your understanding of the subject.

Accelerating the discovery of acceptor materials for organic solar cells by deep learning
Chemistry

Accelerating the discovery of acceptor materials for organic solar cells by deep learning

J. Sun, D. Li, et al.

Deep-learning-based image segmentation integrated with optical microscopy for automatically searching for two-dimensional materials
Engineering and Technology

Deep-learning-based image segmentation integrated with optical microscopy for automatically searching for two-dimensional materials

S. Masubuchi, E. Watanabe, et al.

Bias-Free Multiobjective Active Learning for Materials Design and Discovery
Chemistry

Bias-Free Multiobjective Active Learning for Materials Design and Discovery

K. M. Jablonka, G. M. Jothiappan, et al.

A general and transferable deep learning framework for predicting phase formation in materials
Engineering and Technology

A general and transferable deep learning framework for predicting phase formation in materials

S. Feng, H. Fu, et al.

Listen, Learn & Level Up
Over 10,000 hours of research content in 25+ fields, available in 12+ languages.
No more digging through PDFs, just hit play and absorb the world's latest research in your language, on your time.
listen to research audio papers with researchbunny