傳統(tǒng)的材料計(jì)算盡管能夠提供關(guān)于假設(shè)材料物理性質(zhì)的準(zhǔn)確信息，但仍具有一定的局限性。與傳統(tǒng)方法不同，材料信息學(xué)（MI）方法首先將原始數(shù)據(jù)描述轉(zhuǎn)換為可用于數(shù)學(xué)推理和推斷的適當(dāng)表示。

Fig. 2 Overview of proposed SCANN architecture.

近年來(lái)，許多基于深度學(xué)習(xí)（DL）的MI方法被開(kāi)發(fā)出來(lái)，以應(yīng)對(duì)材料表示方面的挑戰(zhàn)并預(yù)測(cè)其物理特性。然而，目前在材料研究中使用的DL模型在提供用來(lái)解釋預(yù)測(cè)和理解材料構(gòu)效關(guān)系的有效信息方面表現(xiàn)不足。

Fig. 3 Visualizations of structure–property relationships for molecules in QM9 dataset.

來(lái)自日本科學(xué)與技術(shù)高級(jí)研究所的Tien-Sinh Vu等，提出了一種可詮釋DL架構(gòu)，該架構(gòu)結(jié)合了注意力機(jī)制來(lái)預(yù)測(cè)材料特性并深入理解其構(gòu)效關(guān)系。

Fig. 4 Correspondence between obtained GA scores of carbon, nitrogen, and oxygen atomic sites and molecular orbitals of molecular?structures in QM9 dataset.

作者使用兩個(gè)著名的數(shù)據(jù)集（QM9和Materials Project數(shù)據(jù)集），以及三個(gè)內(nèi)部開(kāi)發(fā)的計(jì)算材料數(shù)據(jù)集，對(duì)所提出的架構(gòu)進(jìn)行了評(píng)估。訓(xùn)練–測(cè)試–分割驗(yàn)證證實(shí)了使用DL架構(gòu)導(dǎo)出的模型具有強(qiáng)大的預(yù)測(cè)能力，可與當(dāng)前最先進(jìn)的模型相媲美。

Fig. 5 Visualizations of structure–property relationships for fullerene molecules.

此外，基于第一性原理計(jì)算的比較驗(yàn)證表明，在解釋與物理性質(zhì)有關(guān)的構(gòu)效關(guān)系時(shí)，原子的局部結(jié)構(gòu)對(duì)材料結(jié)構(gòu)表示的注意程度至關(guān)重要。這些性質(zhì)包括分子軌道能量和晶體形成能。

Fig. 6 Visualization of relationship between the adsorption energy and the deformation of a graphene flake with a platinum atom adsorbed on a graphene flake.

通過(guò)預(yù)測(cè)材料性質(zhì)并明確識(shí)別相應(yīng)結(jié)構(gòu)中的關(guān)鍵特征，本工作所提出的架構(gòu)在加速材料設(shè)計(jì)方面顯示出了巨大的潛力。該文近期發(fā)布于npj Computational Materials 9: 215 (2023).

Editorial Summary

A central challenge in the field of materials science involves the use of both experience and theory to explore the compositions and structures of materials with specific properties and subsequently validating them through experimentation. Traditionally, materials have been characterized based on their elemental compositions and structures. Researchers have primarily relied on their knowledge and experience to predict certain properties of hypothetical materials with specific compositions and structures. Traditional material calculations can provide accurate information on the physical properties of hypothetical materials, but they still have certain limitations. Unlike traditional approaches, materials informatics (MI) approaches initially involve the conversion of primitive data descriptions into appropriate representations that can be used for mathematical reasoning and inference. Recently, various deep learning (DL)-based MI approaches have been developed to address the challenges associated with material representation and to predict physical properties.

However, DL models currently employed in materials research exhibit certain limitations in delivering meaningful information for interpreting predictions and comprehending the relationships between structure and material properties.?

Tien-Sinh Vu et al. from Japan Advanced Institute of Science and Technology, proposed an interpretable DL architecture that incorporates the attention mechanism to predict material properties and gained insights into their structure–property relationships. The proposed architecture was evaluated using two well-known datasets (the QM9 and the Materials Project datasets), and three in-house-developed computational materials datasets. Train–test–split validations confirmed that the models derived using the proposed DL architecture exhibit strong predictive capabilities, which are comparable to those of current state-of-the-art models. Furthermore, comparative validations, based on first-principles calculations, indicated that the degree of attention of the atoms’ local structures to the representation of the material structure is critical when interpreting structure–property relationships with respect to physical properties. These properties encompass molecular orbital energies and the formation energies of crystals. The proposed architecture shows great potential in accelerating material design by predicting material properties and explicitly identifying crucial features within the corresponding structures.?This article was recently published in npj Computational Materials 9: 215 (2023).

原文Abstract及其翻譯

Abstract

Deep learning (DL) models currently employed in materials research exhibit certain limitations in delivering meaningful information for interpreting predictions and comprehending the relationships between structure and material properties. To address these limitations, we propose an interpretable DL architecture that incorporates the attention mechanism to predict material properties and gain insights into their structure–property relationships.

The proposed architecture is evaluated using two well-known datasets (the QM9 and the Materials Project datasets), and three in-house-developed computational materials datasets. Train–test–split validations confirm that the models derived using the proposed DL architecture exhibit strong predictive capabilities, which are comparable to those of current state-of-the-art models. Furthermore, comparative validations, based on first-principles calculations, indicate that the degree of attention of the atoms’ local structures to the representation of the material structure is critical when interpreting structure–property relationships with respect to physical properties. These properties encompass molecular orbital energies and the formation energies of crystals. The proposed architecture shows great potential in accelerating material design by predicting material properties and explicitly identifying crucial features within the corresponding structures.

摘要

目前，在材料研究中使用的深度學(xué)習(xí)（DL）模型在提供用來(lái)解釋預(yù)測(cè)和理解材料構(gòu)效關(guān)系的有效信息方面表現(xiàn)出一定的局限性。為了解決這些局限，我們提出了一種可詮釋的DL架構(gòu)，該架構(gòu)結(jié)合了注意力機(jī)制來(lái)預(yù)測(cè)材料特性并深入理解其構(gòu)效關(guān)系。我們使用兩個(gè)著名的數(shù)據(jù)集（QM9和Materials Project數(shù)據(jù)集），以及三個(gè)內(nèi)部開(kāi)發(fā)的計(jì)算材料數(shù)據(jù)集，對(duì)所提出的架構(gòu)進(jìn)行了評(píng)估。訓(xùn)練–測(cè)試–分割驗(yàn)證證實(shí)了使用DL架構(gòu)導(dǎo)出的模型具有強(qiáng)大的預(yù)測(cè)能力，可與當(dāng)前最先進(jìn)的模型相媲美。此外，基于第一性原理計(jì)算的比較驗(yàn)證表明，在解釋與物理性質(zhì)有關(guān)的構(gòu)效關(guān)系時(shí)，原子的局部結(jié)構(gòu)對(duì)材料結(jié)構(gòu)演示的注意程度至關(guān)重要。這些性質(zhì)包括分子軌道能量和晶體形成能。通過(guò)預(yù)測(cè)材料性質(zhì)并明確識(shí)別相應(yīng)結(jié)構(gòu)中的關(guān)鍵特征，本工作所提出的架構(gòu)在加速材料設(shè)計(jì)方面顯示出了巨大的潛力。