Benchmarking graph neural networks for materials chemistry

Abstract Graph neural networks (GNNs) have received intense interest as a rapidly expanding class of machine learning models remarkably well-suited for materials applications. To date, a number of successful GNNs have been proposed and demonstrated for systems ranging from crystal stability to elect...

Descripción completa

Guardado en:

Detalles Bibliográficos
Autores principales:	Victor Fung, Jiaxin Zhang, Eric Juarez, Bobby G. Sumpter
Formato:	article
Lenguaje:	EN
Publicado:	Nature Portfolio 2021
Materias:	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765
Acceso en línea:	https://doaj.org/article/86b6e1b70bf24df7bb74a248da2c8e25
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

id	oai:doaj.org-article:86b6e1b70bf24df7bb74a248da2c8e25
record_format	dspace
spelling	oai:doaj.org-article:86b6e1b70bf24df7bb74a248da2c8e252021-12-02T15:57:19ZBenchmarking graph neural networks for materials chemistry10.1038/s41524-021-00554-02057-3960https://doaj.org/article/86b6e1b70bf24df7bb74a248da2c8e252021-06-01T00:00:00Zhttps://doi.org/10.1038/s41524-021-00554-0https://doaj.org/toc/2057-3960Abstract Graph neural networks (GNNs) have received intense interest as a rapidly expanding class of machine learning models remarkably well-suited for materials applications. To date, a number of successful GNNs have been proposed and demonstrated for systems ranging from crystal stability to electronic property prediction and to surface chemistry and heterogeneous catalysis. However, a consistent benchmark of these models remains lacking, hindering the development and consistent evaluation of new models in the materials field. Here, we present a workflow and testing platform, MatDeepLearn, for quickly and reproducibly assessing and comparing GNNs and other machine learning models. We use this platform to optimize and evaluate a selection of top performing GNNs on several representative datasets in computational materials chemistry. From our investigations we note the importance of hyperparameter selection and find roughly similar performances for the top models once optimized. We identify several strengths in GNNs over conventional models in cases with compositionally diverse datasets and in its overall flexibility with respect to inputs, due to learned rather than defined representations. Meanwhile several weaknesses of GNNs are also observed including high data requirements, and suggestions for further improvement for applications in materials chemistry are discussed.Victor FungJiaxin ZhangEric JuarezBobby G. SumpterNature PortfolioarticleMaterials of engineering and construction. Mechanics of materialsTA401-492Computer softwareQA76.75-76.765ENnpj Computational Materials, Vol 7, Iss 1, Pp 1-8 (2021)
institution	DOAJ
collection	DOAJ
language	EN
topic	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765
spellingShingle	Materials of engineering and construction. Mechanics of materials TA401-492 Computer software QA76.75-76.765 Victor Fung Jiaxin Zhang Eric Juarez Bobby G. Sumpter Benchmarking graph neural networks for materials chemistry
description	Abstract Graph neural networks (GNNs) have received intense interest as a rapidly expanding class of machine learning models remarkably well-suited for materials applications. To date, a number of successful GNNs have been proposed and demonstrated for systems ranging from crystal stability to electronic property prediction and to surface chemistry and heterogeneous catalysis. However, a consistent benchmark of these models remains lacking, hindering the development and consistent evaluation of new models in the materials field. Here, we present a workflow and testing platform, MatDeepLearn, for quickly and reproducibly assessing and comparing GNNs and other machine learning models. We use this platform to optimize and evaluate a selection of top performing GNNs on several representative datasets in computational materials chemistry. From our investigations we note the importance of hyperparameter selection and find roughly similar performances for the top models once optimized. We identify several strengths in GNNs over conventional models in cases with compositionally diverse datasets and in its overall flexibility with respect to inputs, due to learned rather than defined representations. Meanwhile several weaknesses of GNNs are also observed including high data requirements, and suggestions for further improvement for applications in materials chemistry are discussed.
format	article
author	Victor Fung Jiaxin Zhang Eric Juarez Bobby G. Sumpter
author_facet	Victor Fung Jiaxin Zhang Eric Juarez Bobby G. Sumpter
author_sort	Victor Fung
title	Benchmarking graph neural networks for materials chemistry
title_short	Benchmarking graph neural networks for materials chemistry
title_full	Benchmarking graph neural networks for materials chemistry
title_fullStr	Benchmarking graph neural networks for materials chemistry
title_full_unstemmed	Benchmarking graph neural networks for materials chemistry
title_sort	benchmarking graph neural networks for materials chemistry
publisher	Nature Portfolio
publishDate	2021
url	https://doaj.org/article/86b6e1b70bf24df7bb74a248da2c8e25
work_keys_str_mv	AT victorfung benchmarkinggraphneuralnetworksformaterialschemistry AT jiaxinzhang benchmarkinggraphneuralnetworksformaterialschemistry AT ericjuarez benchmarkinggraphneuralnetworksformaterialschemistry AT bobbygsumpter benchmarkinggraphneuralnetworksformaterialschemistry
_version_	1718385338279264256

Benchmarking graph neural networks for materials chemistry

Ejemplares similares