Multi-Task Multi-Fidelity Learning of Properties for Energetic Materials

TL;DR

This paper demonstrates that multi-task neural networks trained on multi-modal experimental and computational data can effectively predict multiple properties of energetic materials, especially benefiting properties with limited data, enabling large-scale screening.

Contribution

The study introduces a multi-task learning framework utilizing multi-modal data for energetic materials, improving prediction accuracy over single-task models and facilitating large-scale property screening.

Findings

Multi-task neural networks outperform single-task models.

Multi-modal data enhances prediction accuracy, especially for data-scarce properties.

Models are based on simple molecular descriptors and applicable to large-scale screening.

Abstract

Data science and artificial intelligence are playing an increasingly important role in the physical sciences. Unfortunately, in the field of energetic materials data scarcity limits the accuracy and even applicability of ML tools. To address data limitations, we compiled multi-modal data: both experimental and computational results for several properties. We find that multi-task neural networks can learn from multi-modal data and outperform single-task models trained for specific properties. As expected, the improvement is more significant for data-scarce properties. These models are trained using descriptors built from simple molecular information and can be readily applied for large-scale materials screening to explore multiple properties simultaneously. This approach is widely applicable to fields outside energetic materials.