Learning Representations of Molecules and Materials with Atomistic Neural Networks

Kristof T. Schütt; Alexandre Tkatchenko; Klaus Robert Müller

doi:10.1007/978-3-030-40245-7_11

Learning Representations of Molecules and Materials with Atomistic Neural Networks

Kristof T. Schütt
, Alexandre Tkatchenko
, Klaus Robert Müller^*

^*Corresponding author for this work

Department of Artificial Intelligence

Research output: Chapter in Book/Report/Conference proceeding › Chapter

9 Citations (Scopus)

Abstract

Deep learning has been shown to learn efficient representations for structured data such as images, text, or audio. In this chapter, we present neural network architectures that are able to learn efficient representations of molecules and materials. In particular, the continuous-filter convolutional network SchNet accurately predicts chemical properties across compositional and configurational space on a variety of datasets. Beyond that, we analyze the obtained representations to find evidence that their spatial and chemical properties agree with chemical intuition.

Original language	English
Title of host publication	Lecture Notes in Physics
Publisher	Springer
Pages	215-230
Number of pages	16
DOIs	https://doi.org/10.1007/978-3-030-40245-7_11
Publication status	Published - 2020

Publication series

Name	Lecture Notes in Physics
Volume	968
ISSN (Print)	0075-8450
ISSN (Electronic)	1616-6361

Bibliographical note

Publisher Copyright:
© 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.

ASJC Scopus subject areas

Physics and Astronomy (miscellaneous)

Access to Document

10.1007/978-3-030-40245-7_11

Cite this

@inbook{d28d884e53d742e7a4981ff47ad1e3f3,

title = "Learning Representations of Molecules and Materials with Atomistic Neural Networks",

abstract = "Deep learning has been shown to learn efficient representations for structured data such as images, text, or audio. In this chapter, we present neural network architectures that are able to learn efficient representations of molecules and materials. In particular, the continuous-filter convolutional network SchNet accurately predicts chemical properties across compositional and configurational space on a variety of datasets. Beyond that, we analyze the obtained representations to find evidence that their spatial and chemical properties agree with chemical intuition.",

author = "Sch{\"u}tt, \{Kristof T.\} and Alexandre Tkatchenko and M{\"u}ller, \{Klaus Robert\}",

note = "Publisher Copyright: {\textcopyright} 2020, The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland AG.",

year = "2020",

doi = "10.1007/978-3-030-40245-7\_11",

language = "English",

series = "Lecture Notes in Physics",

publisher = "Springer",

pages = "215--230",

booktitle = "Lecture Notes in Physics",

}

TY - CHAP

T1 - Learning Representations of Molecules and Materials with Atomistic Neural Networks

AU - Schütt, Kristof T.

AU - Tkatchenko, Alexandre

AU - Müller, Klaus Robert

PY - 2020

Y1 - 2020

N2 - Deep learning has been shown to learn efficient representations for structured data such as images, text, or audio. In this chapter, we present neural network architectures that are able to learn efficient representations of molecules and materials. In particular, the continuous-filter convolutional network SchNet accurately predicts chemical properties across compositional and configurational space on a variety of datasets. Beyond that, we analyze the obtained representations to find evidence that their spatial and chemical properties agree with chemical intuition.

AB - Deep learning has been shown to learn efficient representations for structured data such as images, text, or audio. In this chapter, we present neural network architectures that are able to learn efficient representations of molecules and materials. In particular, the continuous-filter convolutional network SchNet accurately predicts chemical properties across compositional and configurational space on a variety of datasets. Beyond that, we analyze the obtained representations to find evidence that their spatial and chemical properties agree with chemical intuition.

UR - https://www.scopus.com/pages/publications/85086083702

U2 - 10.1007/978-3-030-40245-7_11

DO - 10.1007/978-3-030-40245-7_11

M3 - Chapter

AN - SCOPUS:85086083702

T3 - Lecture Notes in Physics

SP - 215

EP - 230

BT - Lecture Notes in Physics

PB - Springer

ER -

Learning Representations of Molecules and Materials with Atomistic Neural Networks

Abstract

Publication series

Bibliographical note

ASJC Scopus subject areas

Access to Document

Other files and links

Fingerprint

Cite this