Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties.

  title={Crystal Graph Convolutional Neural Networks for an Accurate and Interpretable Prediction of Material Properties.},
  author={Tian Xie and Jeffrey C. Grossman},
  journal={Physical review letters},
  volume={120 14},
  • T. Xie, J. Grossman
  • Published 27 October 2017
  • Computer Science, Materials Science
  • Physical review letters
The use of machine learning methods for accelerating the design of crystalline materials usually requires manually constructed feature vectors or complex transformation of atom coordinates to input the crystal structure, which either constrains the model to certain crystal types or makes it difficult to provide chemical insights. [] Key Method Our method provides a highly accurate prediction of density functional theory calculated properties for eight different properties of crystals with various structure…

Figures and Tables from this paper

Topological representations of crystalline compounds for the machine-learning prediction of materials properties

This work introduces an algebraic topology-based method, called atom-specific persistent homology (ASPH), as a unique representation of crystal structures that outperforms previous work such as Voronoi tessellations and Coulomb matrix method using the same ML algorithm and datasets.

Developing an improved crystal graph convolutional neural network framework for accelerated materials discovery

An improved variant of the CGCNN model (iCGCNN) is developed that outperforms the original by incorporating information of the Voronoi tessellated crystal structure, explicit 3-body correlations of neighboring constituent atoms, and an optimized chemical representation of interatomic bonds in the crystal graphs.

Orbital Graph Convolutional Neural Network for Material Property Prediction

The OGCNN model with high predictive accuracy can be used to discover new materials among the immense phase and compound spaces of materials and significantly outperforms other conventional regression machine learning algorithms where different crystal featurization methods have been used.

Atom table convolutional neural networks for an accurate prediction of compounds properties

This work develops an atom table convolutional neural networks that only requires the component information to directly learn the experimental properties from the features constructed by itself, which is valuable for high throughput screening and helpful to understand the underlying physics.

Atomistic graph networks for experimental materials property prediction

This work shows how material descriptors can be learned from the structures present in large scale datasets of material simulations; and how they can be used to improve the prediction of an experimental property, the energy of formation of a solid.

Crystal graph convolutional neural networks for per-site property prediction

This work presents a persite crystal graph convolutional neural network that predicts a wide array of atom-level properties such as Bader charge, magnetic moment, or site-projected band centers.

Prediction of Material Properties using Crystal Graph Convolutional Neural Networks

The proposed model employs GCNs in the domain of materials science while incorporating a relatively more sophisticated pooling mechanism that takes into account the structure of the crystals being featurized, with better results for all the attributes tested as compared to traditional first-principles methods used for computing the same.

Study of crystal properties based on attention mechanism and crystal graph convolutional neural network

An attention mechanism-based crystal graph convolutional neural network is proposed, which builds a machine learning model by inputting crystallographic information files and target properties to learn the local chemical environment and node normalization is added to reduce the risk of overfitting.

Toward machine learning for microscopic mechanisms: A formula search for crystal structure stability based on atomic properties

A combinatorial machine-learning approach to obtain physical formulas based on simple and easily accessible ingredients, such as atomic properties, to predict the energetic stability of semiconducting binary compounds with respect to zinc blende and rocksalt crystal structures is proposed.

Machine learning approaches for feature engineering of the crystal structure: Application to the prediction of the formation energy of cubic compounds

An approach for efficient and scalable feature engineering of the crystal structure in compounds of different chemical compositions is presented and it is shown that the top 25 features identified by the proposed framework serve as good regressors for the formation energy of the crystalline substance for both model building strategies.



Universal fragment descriptors for predicting properties of inorganic crystals

Data from the AFLOW repository for ab initio calculations is combined with Quantitative Materials Structure-Property Relationship models to predict important properties: metal/insulator classification, band gap energy, bulk/shear moduli, Debye temperature and heat capacities.

How to represent crystal structures for machine learning: Towards fast prediction of electronic properties

It is found that conventional representations of the input data, such as the Coulomb matrix, are not suitable for the training of learning machines in the case of periodic solids and proposes a novel crystal structure representation for which learning and competitive prediction accuracies become possible within an unrestricted class of spd systems of arbitrary unit-cell size.

Representation of compounds for machine-learning prediction of physical properties

The effect of the descriptor sets on the efficiency of Bayesian optimization in addition to the accuracy of the kernel ridge regression models are examined, which exhibit good predictive performances.

A Statistical Learning Framework for Materials Science: Application to Elastic Moduli of k-nary Inorganic Polycrystalline Compounds

This work presents a SL framework that addresses challenges in materials science applications, where datasets are diverse but of modest size, and extreme values are often of interest.

Materials Cartography: Representing and Mining Material Space Using Structural and Electronic Fingerprints

The issue of scientific discovery in materials databases is addressed by introducing novel analytical approaches based on structural and electronic materials fingerprints, which contribute to the emerging field of materials informati...

Charting the complete elastic properties of inorganic crystalline compounds

The largest database of calculated elastic properties for inorganic compounds to date is presented and the different ways it can be accessed and analyzed in efforts related to materials discovery and design are described.

Crystal structure representations for machine learning models of formation energies

We introduce and evaluate a set of feature vector representations of crystal structures for machine learning (ML) models of formation energies of solids. ML models of atomization energies of organi

The Inorganic Crystal Structure Database (ICSD)—Present and Future

An overview of the product portfolio and the current activities of the Inorganic Crystal Structure Database is given.

Big data of materials science: critical role of the descriptor.

A trustful prediction of new promising materials, identification of anomalies, and scientific advancement are doubtful when the scientific connection between the descriptor and the actuating mechanisms is unclear.

The Open Quantum Materials Database (OQMD): assessing the accuracy of DFT formation energies

The Open Quantum Materials Database (OQMD) is a high-throughput database currently consisting of nearly 300,000 density functional theory (DFT) total energy calculations of compounds from the