Achieving Robust Extrapolation in Materials Property Prediction via Decoupled Transfer Learning
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866908843291705344 |
|---|---|
| author | Sugiura, Tasuku Mizoguchi, Teruyasu |
| author_facet | Sugiura, Tasuku Mizoguchi, Teruyasu |
| contents | Machine learning has revolutionized materials property prediction, yet fails catastrophically when extrapolating beyond training distributions-precisely the capability required for discovering unprecedented materials. Graph neural networks (GNNs) exhibit this collapse because end-to-end training fundamentally couples learned representations to target property distributions, preventing genuine extrapolation. We demonstrate that decoupled transfer learning-separating pretrained GNN feature extractors from simple regressors-overcomes this barrier. Pretrained features provide transferable structural knowledge, while simple regressors enable smooth extrapolation by maintaining learned trends beyond training boundaries. Benchmarked on layered intercalation compounds through four rigorous extrapolation scenarios and a temporal Materials Project split, our framework achieves 68% error reduction (RMSE: 0.881 vs. 2.778 eV/atom) versus end-to-end GNNs for extrapolation. Failure analysis reveals extrapolation succeeds for continuous chemical space but fails for discontinuous space, establishing clear design principles. Validated on Fermi energy prediction, this framework is immediately deployable using existing pretrained models, requiring no architectural innovations-transforming ML-driven materials discovery. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_18054 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Achieving Robust Extrapolation in Materials Property Prediction via Decoupled Transfer Learning Sugiura, Tasuku Mizoguchi, Teruyasu Materials Science Machine learning has revolutionized materials property prediction, yet fails catastrophically when extrapolating beyond training distributions-precisely the capability required for discovering unprecedented materials. Graph neural networks (GNNs) exhibit this collapse because end-to-end training fundamentally couples learned representations to target property distributions, preventing genuine extrapolation. We demonstrate that decoupled transfer learning-separating pretrained GNN feature extractors from simple regressors-overcomes this barrier. Pretrained features provide transferable structural knowledge, while simple regressors enable smooth extrapolation by maintaining learned trends beyond training boundaries. Benchmarked on layered intercalation compounds through four rigorous extrapolation scenarios and a temporal Materials Project split, our framework achieves 68% error reduction (RMSE: 0.881 vs. 2.778 eV/atom) versus end-to-end GNNs for extrapolation. Failure analysis reveals extrapolation succeeds for continuous chemical space but fails for discontinuous space, establishing clear design principles. Validated on Fermi energy prediction, this framework is immediately deployable using existing pretrained models, requiring no architectural innovations-transforming ML-driven materials discovery. |
| title | Achieving Robust Extrapolation in Materials Property Prediction via Decoupled Transfer Learning |
| topic | Materials Science |
| url | https://arxiv.org/abs/2602.18054 |