Discrimination of Pb-Zn deposit types using the trace element data of galena based on deep learning

• DOI: 10.1016/j.oregeorev.2024.106133
• Publication Date: 2024-06-24T16:34:56Z
• Authors: Shuaibing Li; Zhixin Zhao; Tengfeng Miao; Xiaolong Li; Siqi Yu; Yimeng Zhang; Junhao Wei; Kun Zou
• Keywords: Trace element

Different types of ore deposits exhibit distinct metal sources, physicochemical conditions, and ore-forming processes. Galena, a key sulfide in Pb-Zn deposits, possesses trace elements that may be utilized for classifying deposit types. Presently, research on based galena is sparse, there lack robust methods distinguishing using these elements. In this study, we demonstrate deep learning algorithm, the galena, can effectively The model training process UMAP visualization, evaluation was conducted multiple statistical metrics, confusion matrices, ROC curves. Finally, by dissecting 'black box' with established comprehensive set analysis discriminating types: discrimination-visualization-evaluation-dissection. A dataset comprising 828 LA-ICP-MS analyses from 34 worldwide curated peer-reviewed sources. It includes data 7 (Se, Ag, Cd, Sn, Sb, Tl, Bi) across carbonate replacement (CRD), epithermal, mississippi valley type (MVT), sedimentary exhalative (SEDEX), skarn, volcanogenic massive (VMS), vein-type deposits. Initially, selected algorithm three visualization (PCA, t-SNE UMAP), its ability to balance global local structures visualizing internals model. Subsequently, developed (1D-CNN) classification. Various metrics visual indicate exceptional performance our model, achieving an overall accuracy 99.18 % test set. SHAP used analyze importance different Elements such as Bi were found particularly significant types, confirming influence within galena. Therefore, conclude algorithms identify offering new insights into study