Band-gap regression with architecture-optimized message-passing neural networks

TL;DR

This paper develops an optimized message-passing neural network architecture for predicting material band gaps, demonstrating superior performance and uncertainty quantification over existing models through neural architecture search and ensemble methods.

Contribution

It introduces a neural architecture search for MPNNs tailored to band gap prediction, improving accuracy and robustness over prior models.

Findings

Ensemble models outperform individual neural networks in band gap prediction.

Neural architecture search identifies optimal MPNN configurations for this task.

Uncertainty quantification methods show ensemble approach is superior.

Abstract

Graph-based neural networks and, specifically, message-passing neural networks (MPNNs) have shown great potential in predicting physical properties of solids. In this work, we train an MPNN to first classify materials through density functional theory data from the AFLOW database as being metallic or semiconducting/insulating. We then perform a neural-architecture search to explore the model architecture and hyperparameter space of MPNNs to predict the band gaps of the materials identified as non-metals. The parameters in the search include the number of message-passing steps, latent size, and activation-function, among others. The top-performing models from the search are pooled into an ensemble that significantly outperforms existing models from the literature. Uncertainty quantification is evaluated with Monte-Carlo Dropout and ensembling, with the ensemble method proving superior. The domain of applicability of the ensemble model is analyzed with respect to the crystal systems, the inclusion of a Hubbard parameter in the density functional calculations, and the atomic species building up the materials.