Geocfm improves sampling of mineral sites from geophysical images
GeoCFM: Positive-Only Conditional Flow Matching for Mineral Occurrence Sampling
Machine Learning
Summary
Finding minerals underground is hard because we only know the sites where minerals are found, not where they are absent. The authors treat this as a problem of learning where mineral deposits might occur given geographic images, rather than guessing a score for each location. They introduce GeoCFM, a method that generates likely mineral deposit locations conditioned on geophysical data, better matching observed mineral sites and accounting for uncertainty. This approach was tested on synthetic and real datasets and outperformed previous methods.
What this means in practice
- •For mineral exploration teams: Generate probable mineral deposit locations from complex geo-image data to better guide exploration efforts where labels only exist for found sites.
- •For environmental survey teams: Estimate spatial distributions of scarce natural resources using few positive samples and noisy background data from geophysical sensors.
Authors
Moshe Eliasof, Eldad Haber
Abstract
Critical mineral discovery is a positive-only problem: deposits are observed as sparse locations, while unlabeled regions are not reliable negatives, and similar geophysical signatures can arise from different subsurface states. We therefore model mineral targeting as learning a conditional spatial distribution over occurrence locations, $π(p\mid d)$, given geo-images $d$, rather than predicting a deterministic per-pixel score map. We introduce GeoCFM, a conditional flow-matching model that generates mineral occurrence point sets conditioned on multi-channel geo-images; GeoCFM learns a point-wise transport field in $\mathbb{R}^2$, using UNet features with point-conditioned velocity prediction to bridge dense rasters and sparse supervision without pseudo-negatives. On a synthetic magnetics--geochemistry benchmark with latent activation and on USGS Earth MRI data with a spatially disjoint tile split, GeoCFM improves geometric agreement with observed occurrences over score-map and non-conditional baselines, while representing epistemic uncertainty through conditional sampling.