Figure 3 –False positive mineralization rates (>0.5% Cu) and (b) missed mineralization rates (>0.5% Cu) for 2021–2022 across model configurations and the site Kriging model. Method 1—DL trained on assay data only—produced performance virtually indistinguishable from that of the site Kriging model, confirming that the DL architecture itself does not confer performance gains in the absence of supplementary data. Method 3—incorporating geological logging as both input feature and training target—reconciled 1.67 times more mineralization at an equivalent false positive rate, demonstrating that the performance differential is attributable specifically to the integration of geological logging data. The model guided a 2,200 m underground expansion drill program across three target zones (Zones 6, 16, and 65), each located at a minimum of 10 m from previously known mineralization, in volumes where the DL model predicted economic copper grades and the site Kriging model predicted waste. A minimum threshold of 60% differential in contained Cu between the DL and Kriging predictions was applied as the target classification criterion. A successful outcome was defined as verification of 0.24 kT in-situ Cu to the Measured resource classification per target. Figure 4 – Plan view of the three target mineralization zones identified by the DL model as economic ore in volumes classified as waste by the site Kriging model. Twelve drill holes were completed, ranging from 100 m to 296 m in length, drilled from existing underground infrastructure. All 11 holes intersected high-grade Cu at grades exceeding three times the economic cut-off. Results by target zone are summarized in Table 2. Table 2 – Chilean IOCG drilling results by target zone Target Outcome Notable Finding Zone 6 All holes hit HG Historical drilling oriented perpendicular to mineralization; HG smoothed in compositing
RkJQdWJsaXNoZXIy MTM0Mzk2