skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Learning atoms for materials discovery

Abstract

Exciting advances have been made in artificial intelligence (AI) during recent decades. Among them, applications of machine learning (ML) and deep learning techniques brought human-competitive performances in various tasks of fields, including image recognition, speech recognition, and natural language understanding. Even in Go, the ancient game of profound complexity, the AI player has already beat human world champions convincingly with and without learning from the human. In this work, we show that our unsupervised machines (Atom2Vec) can learn the basic properties of atoms by themselves from the extensive database of known compounds and materials. These learned properties are represented in terms of high-dimensional vectors, and clustering of atoms in vector space classifies them into meaningful groups consistent with human knowledge. Furthermore, we use the atom vectors as basic input units for neural networks and other ML models designed and trained to predict materials properties, which demonstrate significant accuracy.

Authors:
 [1];  [1];  [1];  [2];  [2];  [3]
  1. Stanford Univ., Stanford, CA (United States)
  2. Temple Univ., Philadelphia, PA (United States)
  3. Stanford Univ., Stanford, CA (United States); SLAC National Accelerator Lab., Menlo Park, CA (United States)
Publication Date:
Research Org.:
SLAC National Accelerator Lab., Menlo Park, CA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1463356
Alternate Identifier(s):
OSTI ID: 1457210
Grant/Contract Number:  
AC02-76SF00515; SC0012575
Resource Type:
Journal Article: Accepted Manuscript
Journal Name:
Proceedings of the National Academy of Sciences of the United States of America
Additional Journal Information:
Journal Volume: 115; Journal Issue: 28; Journal ID: ISSN 0027-8424
Publisher:
National Academy of Sciences, Washington, DC (United States)
Country of Publication:
United States
Language:
English
Subject:
74 ATOMIC AND MOLECULAR PHYSICS; atomism; machine learning; materials discovery

Citation Formats

Zhou, Quan, Tang, Peizhe, Liu, Shenxiu, Pan, Jinbo, Yan, Qimin, and Zhang, Shou -Cheng. Learning atoms for materials discovery. United States: N. p., 2018. Web. doi:10.1073/pnas.1801181115.
Zhou, Quan, Tang, Peizhe, Liu, Shenxiu, Pan, Jinbo, Yan, Qimin, & Zhang, Shou -Cheng. Learning atoms for materials discovery. United States. doi:10.1073/pnas.1801181115.
Zhou, Quan, Tang, Peizhe, Liu, Shenxiu, Pan, Jinbo, Yan, Qimin, and Zhang, Shou -Cheng. Tue . "Learning atoms for materials discovery". United States. doi:10.1073/pnas.1801181115. https://www.osti.gov/servlets/purl/1463356.
@article{osti_1463356,
title = {Learning atoms for materials discovery},
author = {Zhou, Quan and Tang, Peizhe and Liu, Shenxiu and Pan, Jinbo and Yan, Qimin and Zhang, Shou -Cheng},
abstractNote = {Exciting advances have been made in artificial intelligence (AI) during recent decades. Among them, applications of machine learning (ML) and deep learning techniques brought human-competitive performances in various tasks of fields, including image recognition, speech recognition, and natural language understanding. Even in Go, the ancient game of profound complexity, the AI player has already beat human world champions convincingly with and without learning from the human. In this work, we show that our unsupervised machines (Atom2Vec) can learn the basic properties of atoms by themselves from the extensive database of known compounds and materials. These learned properties are represented in terms of high-dimensional vectors, and clustering of atoms in vector space classifies them into meaningful groups consistent with human knowledge. Furthermore, we use the atom vectors as basic input units for neural networks and other ML models designed and trained to predict materials properties, which demonstrate significant accuracy.},
doi = {10.1073/pnas.1801181115},
journal = {Proceedings of the National Academy of Sciences of the United States of America},
number = 28,
volume = 115,
place = {United States},
year = {Tue Jun 26 00:00:00 EDT 2018},
month = {Tue Jun 26 00:00:00 EDT 2018}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Citation Metrics:
Cited by: 1 work
Citation information provided by
Web of Science

Save / Share:

Works referenced in this record:

Towards the computational design of solid catalysts
journal, April 2009

  • Nørskov, J.; Bligaard, T.; Rossmeisl, J.
  • Nature Chemistry, Vol. 1, Issue 1, p. 37-46
  • DOI: 10.1038/nchem.121