About this Abstract |
| Meeting |
2025 TMS Annual Meeting & Exhibition
|
| Symposium
|
AI/Data Informatics: Computational Model Development, Verification, Validation, and Uncertainty Quantification
|
| Presentation Title |
Self-Supervised Learning (SSL) for Crystal Property Prediction via Structure Denoising |
| Author(s) |
Alexander New, Nam Q Le, Michael Pekala, Christopher Stiles |
| On-Site Speaker (Planned) |
Alexander New |
| Abstract Scope |
Accurate prediction of the properties of crystalline materials is crucial for targeted discovery, and this prediction is increasingly done with data-driven models. However, for many properties of interest, the number of materials for which a specific property has been determined is much smaller than the number of known materials. To overcome this disparity, we propose a novel self-supervised learning (SSL) strategy for material property prediction. Our approach, crystal denoising self-supervised learning (CDSSL), pretrains predictive models (e.g., crystal graph convolutional neural networks) with a pretext task based on recovering valid material structures when given perturbed versions of these structures. Pretraining is performed using a large dataset of crystal structures without calculated properties. After pretraining, the model is finetuned on a property prediction task with labeled data. We demonstrate that CDSSL models out-perform models trained without SSL, across material types, properties, and dataset sizes. |
| Proceedings Inclusion? |
Planned: |
| Keywords |
Machine Learning, Computational Materials Science & Engineering, Other |