SchNet: A continuous-filter convolutional neural network for modeling quantum interactions

K. T. Schütt, P. J. Kindermans, H. E. Sauceda, S. Chmiela, A. Tkatchenko, Klaus Muller

Research output: Contribution to journalConference article

32 Citations (Scopus)

Abstract

Deep learning has the potential to revolutionize quantum chemistry as it is ideally suited to learn representations for structured data and speed up the exploration of chemical space. While convolutional neural networks have proven to be the first choice for images, audio and video data, the atoms in molecules are not restricted to a grid. Instead, their precise locations contain essential physical information, that would get lost if discretized. Thus, we propose to use continuous-filter convolutional layers to be able to model local correlations without requiring the data to lie on a grid. We apply those layers in SchNet: a novel deep learning architecture modeling quantum interactions in molecules. We obtain a joint model for the total energy and interatomic forces that follows fundamental quantum-chemical principles. Our architecture achieves state-of-the-art performance for benchmarks of equilibrium molecules and molecular dynamics trajectories. Finally, we introduce a more challenging benchmark with chemical and structural variations that suggests the path for further work.

Original languageEnglish
Pages (from-to)992-1002
Number of pages11
JournalAdvances in Neural Information Processing Systems
Volume2017-December
Publication statusPublished - 2017 Jan 1
Event31st Annual Conference on Neural Information Processing Systems, NIPS 2017 - Long Beach, United States
Duration: 2017 Dec 42017 Dec 9

Fingerprint

Neural networks
Molecules
Quantum chemistry
Molecular dynamics
Trajectories
Atoms
Deep learning

ASJC Scopus subject areas

  • Computer Networks and Communications
  • Information Systems
  • Signal Processing

Cite this

Schütt, K. T., Kindermans, P. J., Sauceda, H. E., Chmiela, S., Tkatchenko, A., & Muller, K. (2017). SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. Advances in Neural Information Processing Systems, 2017-December, 992-1002.

SchNet : A continuous-filter convolutional neural network for modeling quantum interactions. / Schütt, K. T.; Kindermans, P. J.; Sauceda, H. E.; Chmiela, S.; Tkatchenko, A.; Muller, Klaus.

In: Advances in Neural Information Processing Systems, Vol. 2017-December, 01.01.2017, p. 992-1002.

Research output: Contribution to journalConference article

Schütt, KT, Kindermans, PJ, Sauceda, HE, Chmiela, S, Tkatchenko, A & Muller, K 2017, 'SchNet: A continuous-filter convolutional neural network for modeling quantum interactions', Advances in Neural Information Processing Systems, vol. 2017-December, pp. 992-1002.
Schütt KT, Kindermans PJ, Sauceda HE, Chmiela S, Tkatchenko A, Muller K. SchNet: A continuous-filter convolutional neural network for modeling quantum interactions. Advances in Neural Information Processing Systems. 2017 Jan 1;2017-December:992-1002.
Schütt, K. T. ; Kindermans, P. J. ; Sauceda, H. E. ; Chmiela, S. ; Tkatchenko, A. ; Muller, Klaus. / SchNet : A continuous-filter convolutional neural network for modeling quantum interactions. In: Advances in Neural Information Processing Systems. 2017 ; Vol. 2017-December. pp. 992-1002.
@article{2188902d0e9044c5a951900ca85a12ad,
title = "SchNet: A continuous-filter convolutional neural network for modeling quantum interactions",
abstract = "Deep learning has the potential to revolutionize quantum chemistry as it is ideally suited to learn representations for structured data and speed up the exploration of chemical space. While convolutional neural networks have proven to be the first choice for images, audio and video data, the atoms in molecules are not restricted to a grid. Instead, their precise locations contain essential physical information, that would get lost if discretized. Thus, we propose to use continuous-filter convolutional layers to be able to model local correlations without requiring the data to lie on a grid. We apply those layers in SchNet: a novel deep learning architecture modeling quantum interactions in molecules. We obtain a joint model for the total energy and interatomic forces that follows fundamental quantum-chemical principles. Our architecture achieves state-of-the-art performance for benchmarks of equilibrium molecules and molecular dynamics trajectories. Finally, we introduce a more challenging benchmark with chemical and structural variations that suggests the path for further work.",
author = "Sch{\"u}tt, {K. T.} and Kindermans, {P. J.} and Sauceda, {H. E.} and S. Chmiela and A. Tkatchenko and Klaus Muller",
year = "2017",
month = "1",
day = "1",
language = "English",
volume = "2017-December",
pages = "992--1002",
journal = "Advances in Neural Information Processing Systems",
issn = "1049-5258",

}

TY - JOUR

T1 - SchNet

T2 - A continuous-filter convolutional neural network for modeling quantum interactions

AU - Schütt, K. T.

AU - Kindermans, P. J.

AU - Sauceda, H. E.

AU - Chmiela, S.

AU - Tkatchenko, A.

AU - Muller, Klaus

PY - 2017/1/1

Y1 - 2017/1/1

N2 - Deep learning has the potential to revolutionize quantum chemistry as it is ideally suited to learn representations for structured data and speed up the exploration of chemical space. While convolutional neural networks have proven to be the first choice for images, audio and video data, the atoms in molecules are not restricted to a grid. Instead, their precise locations contain essential physical information, that would get lost if discretized. Thus, we propose to use continuous-filter convolutional layers to be able to model local correlations without requiring the data to lie on a grid. We apply those layers in SchNet: a novel deep learning architecture modeling quantum interactions in molecules. We obtain a joint model for the total energy and interatomic forces that follows fundamental quantum-chemical principles. Our architecture achieves state-of-the-art performance for benchmarks of equilibrium molecules and molecular dynamics trajectories. Finally, we introduce a more challenging benchmark with chemical and structural variations that suggests the path for further work.

AB - Deep learning has the potential to revolutionize quantum chemistry as it is ideally suited to learn representations for structured data and speed up the exploration of chemical space. While convolutional neural networks have proven to be the first choice for images, audio and video data, the atoms in molecules are not restricted to a grid. Instead, their precise locations contain essential physical information, that would get lost if discretized. Thus, we propose to use continuous-filter convolutional layers to be able to model local correlations without requiring the data to lie on a grid. We apply those layers in SchNet: a novel deep learning architecture modeling quantum interactions in molecules. We obtain a joint model for the total energy and interatomic forces that follows fundamental quantum-chemical principles. Our architecture achieves state-of-the-art performance for benchmarks of equilibrium molecules and molecular dynamics trajectories. Finally, we introduce a more challenging benchmark with chemical and structural variations that suggests the path for further work.

UR - http://www.scopus.com/inward/record.url?scp=85044257334&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85044257334&partnerID=8YFLogxK

M3 - Conference article

AN - SCOPUS:85044257334

VL - 2017-December

SP - 992

EP - 1002

JO - Advances in Neural Information Processing Systems

JF - Advances in Neural Information Processing Systems

SN - 1049-5258

ER -