Committee neural network potentials control generalization errors and enable active learning

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216208%3A11320%2F20%3A10415721" target="_blank" >RIV/00216208:11320/20:10415721 - isvavai.cz</a>

Výsledek na webu

<a href="https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=YrqNwhks_L" target="_blank" >https://verso.is.cuni.cz/pub/verso.fpl?fname=obd_publikace_handle&handle=YrqNwhks_L</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1063/5.0016004" target="_blank" >10.1063/5.0016004</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Committee neural network potentials control generalization errors and enable active learning

Popis výsledku v původním jazyce

It is well known in the field of machine learning that committee models improve accuracy, provide generalization error estimates, and enable active learning strategies. In this work, we adapt these concepts to interatomic potentials based on artificial neural networks. Instead of a single model, multiple models that share the same atomic environment descriptors yield an average that outperforms its individual members as well as a measure of the generalization error in the form of the committee disagreement. We not only use this disagreement to identify the most relevant configurations to build up the model&apos;s training set in an active learning procedure but also monitor and bias it during simulations to control the generalization error. This facilitates the adaptive development of committee neural network potentials and their training sets while keeping the number of ab initio calculations to a minimum. To illustrate the benefits of this methodology, we apply it to the development of a committee model for water in the condensed phase. Starting from a single reference ab initio simulation, we use active learning to expand into new state points and to describe the quantum nature of the nuclei. The final model, trained on 814 reference calculations, yields excellent results under a range of conditions, from liquid water at ambient and elevated temperatures and pressures to different phases of ice, and the air-water interface-all including nuclear quantum effects. This approach to committee models will enable the systematic development of robust machine learning models for a broad range of systems.

Název v anglickém jazyce

Committee neural network potentials control generalization errors and enable active learning

Popis výsledku anglicky

It is well known in the field of machine learning that committee models improve accuracy, provide generalization error estimates, and enable active learning strategies. In this work, we adapt these concepts to interatomic potentials based on artificial neural networks. Instead of a single model, multiple models that share the same atomic environment descriptors yield an average that outperforms its individual members as well as a measure of the generalization error in the form of the committee disagreement. We not only use this disagreement to identify the most relevant configurations to build up the model&apos;s training set in an active learning procedure but also monitor and bias it during simulations to control the generalization error. This facilitates the adaptive development of committee neural network potentials and their training sets while keeping the number of ab initio calculations to a minimum. To illustrate the benefits of this methodology, we apply it to the development of a committee model for water in the condensed phase. Starting from a single reference ab initio simulation, we use active learning to expand into new state points and to describe the quantum nature of the nuclei. The final model, trained on 814 reference calculations, yields excellent results under a range of conditions, from liquid water at ambient and elevated temperatures and pressures to different phases of ice, and the air-water interface-all including nuclear quantum effects. This approach to committee models will enable the systematic development of robust machine learning models for a broad range of systems.

Druh

J_imp - Článek v periodiku v databázi Web of Science

CEP obor

—

OECD FORD obor

10403 - Physical chemistry

Projekt

—

Návaznosti

S - Specificky vyzkum na vysokych skolach<br>I - Institucionalni podpora na dlouhodoby koncepcni rozvoj vyzkumne organizace

Rok uplatnění

2020

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název periodika

Journal of Chemical Physics

ISSN

0021-9606

e-ISSN

—

Svazek periodika

153

Číslo periodika v rámci svazku

10

Stát vydavatele periodika

US - Spojené státy americké

Počet stran výsledku

13

Strana od-do

104105

Kód UT WoS článku

000570950400002

EID výsledku v databázi Scopus

2-s2.0-85091054914