Domain-informed explainable boosting machines for trustworthy lateral spread predictions
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866912972253691904 |
|---|---|
| author | Hsiao, Cheng-Hsi Kumar, Krishna Rathje, Ellen M. |
| author_facet | Hsiao, Cheng-Hsi Kumar, Krishna Rathje, Ellen M. |
| contents | Explainable Boosting Machines (EBMs) provide transparent predictions through additive shape functions, enabling direct inspection of feature contributions. However, EBMs can learn non-physical relationships that reduce their reliability in natural hazard applications. This study presents a domain-informed framework to improve the physical consistency of EBMs for lateral spreading prediction. Our approach modifies learned shape functions based on domain knowledge. These modifications correct non-physical behavior while maintaining data-driven patterns. We apply the method to the 2011 Christchurch earthquake dataset and correct non-physical trends observed in the original EBM. The resulting model produces more physically consistent global and local explanations, with an acceptable tradeoff in accuracy (4--5\%). |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_17175 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Domain-informed explainable boosting machines for trustworthy lateral spread predictions Hsiao, Cheng-Hsi Kumar, Krishna Rathje, Ellen M. Machine Learning Geophysics Explainable Boosting Machines (EBMs) provide transparent predictions through additive shape functions, enabling direct inspection of feature contributions. However, EBMs can learn non-physical relationships that reduce their reliability in natural hazard applications. This study presents a domain-informed framework to improve the physical consistency of EBMs for lateral spreading prediction. Our approach modifies learned shape functions based on domain knowledge. These modifications correct non-physical behavior while maintaining data-driven patterns. We apply the method to the 2011 Christchurch earthquake dataset and correct non-physical trends observed in the original EBM. The resulting model produces more physically consistent global and local explanations, with an acceptable tradeoff in accuracy (4--5\%). |
| title | Domain-informed explainable boosting machines for trustworthy lateral spread predictions |
| topic | Machine Learning Geophysics |
| url | https://arxiv.org/abs/2603.17175 |