A mine-cavity study found that geometric prediction accuracy varied by ground history
Source check: Source checked. Not independent scientific confirmation.
Finding: On held-out cavity scans from one mine, a 3D U-Net achieved a mean overlap score of 0.601 versus 0.491 for the planning baseline. The ranking reversed in previously unexcavated ground: 0.525 for the learned model versus 0.592 for the baseline.
Technology: Ensemble of fitted three-dimensional U-Nets
Publication: Preprint · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- The U-Net ensemble used recorded field inputs to estimate cavity geometry. Its predictions were compared with recorded cavity scans.
- Test and result
- The study randomly held out historical cavities and compared volume overlap against surveyed boundaries and the site’s geometric baseline.
- Limitation / still unknown
- The data came from one mine and roughly 90% of training-domain examples involved neighboring filled excavations. The subgroup scores do not establish prospective or cross-site reliability.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://arxiv.org/abs/2610.36303v1 · https://arxiv.org/html/2610.36303v1
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: linked separately.
Review date: 21 Nov 2026. Review scope: §4.1, Tables 2–4 and aggregate test-split provenance.
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.