149 6.2. Validation Results Using this criterion, 49 out of 60 deposits were successfully identified (82%). 7. RESULTS The generated prospectivity map reveals spatial patterns consistent with known porphyry mineralization systems. High-similarity zones (similarity ≥ 0.80) are spatially concentrated around known deposits, indicating that the model successfully captured geological signatures associated with porphyry systems. These zones are not randomly distributed but instead exhibit coherent spatial continuity, suggesting control by regional geological structures and magmatic features. The model demonstrated strong predictive capability by successfully identifying 49 out of 60 withheld deposits (82%), as described in Section 6. Importantly, the detected deposits are predominantly located within well-defined, high-confidence anomaly zones, reinforcing the reliability of the latent-space similarity approach. Beyond reproducing known mineralization patterns, the prospectivity map also delineates previously unrecognized targets. A total of 30 high-prospectivity zones (similarity ≥ 0.90) were identified across the study area (see Figure 8). These zones represent areas with strong resemblance to the learned porphyry signature and constitute priority targets for further exploration. 8. DISCUSSION 8.1. Strengths of the Approach The proposed methodology offers several advantages over conventional mineral prospectivity mapping techniques. By using an unsupervised framework, it avoids the need for explicit negative labeling,
RkJQdWJsaXNoZXIy MTM0Mzk2