Atomistic graph networks for experimental materials property prediction

TL;DR

This paper introduces atomistic graph networks that learn material descriptors from large simulation datasets, significantly improving experimental property predictions, especially with limited data or unseen chemical spaces.

Contribution

It presents a novel approach using graph neural networks to learn descriptors from simulations, enhancing experimental property prediction accuracy over existing composition-based methods.

Findings

Learned descriptors outperform composition-based methods.

Approach improves with less training data.

Method generalizes well to unseen chemical spaces.

Abstract

Machine Learning (ML) has the potential to accelerate discovery of new materials and shed light on useful properties of existing materials. A key difficulty when applying ML in Materials Science is that experimental datasets of material properties tend to be small. In this work we show how material descriptors can be learned from the structures present in large scale datasets of material simulations; and how these descriptors can be used to improve the prediction of an experimental property, the energy of formation of a solid. The material descriptors are learned by training a Graph Neural Network to regress simulated formation energies from a material's atomistic structure. Using these learned features for experimental property predictions outperforms existing methods that are based solely on chemical composition. Moreover, we find that the advantage of our approach increases as the generalization requirements of the task are made more stringent, for example when limiting the amount of training data or when generalizing to unseen chemical spaces.