Feature optimization for atomistic machine learning yields a data-driven construction of the periodic table of the elements-Reference-Cited by-同舟云学术

Feature optimization for atomistic machine learning yields a data-driven construction of the periodic table of the elements

Published:2018 Issue:47 Volume:20 Page:29661-29668
ISSN:1463-9076
Container-title:Physical Chemistry Chemical Physics
language:en
Short-container-title:Phys. Chem. Chem. Phys.

Author:

Willatt Michael J.¹²³⁴⁵,Musil Félix¹²³⁴⁵,Ceriotti Michele¹²³⁴⁵^ORCID

Affiliation:

1. National Center for Computational Design and Discovery of Novel Materials (MARVEL)

2. Laboratory of Computational Science and Modelling

3. Institute of Materials

4. Ecole Polytechnique Federale de Lausanne

5. Lausanne

Abstract

By representing elements as points in a low-dimensional chemical space it is possible to improve the performance of a machine-learning model for a chemically-diverse dataset. The resulting coordinates are reminiscent of the main groups of the periodic table.

Funder

H2020 European Research Council

Schweizerischer Nationalfonds zur Förderung der Wissenschaftlichen Forschung

Publisher

Royal Society of Chemistry (RSC)

Subject

Physical and Theoretical Chemistry,General Physics and Astronomy

Link

http://pubs.rsc.org/en/content/articlepdf/2018/CP/C8CP05921G

Reference48 articles.

1. Generalized Neural-Network Representation of High-Dimensional Potential-Energy Surfaces

2. Gaussian Approximation Potentials: The Accuracy of Quantum Mechanics, without the Electrons