Intelligent Recognition of Ore-Forming Anomalies Based on Multisource Data Fusion: A Case Study of the Daqiao Mining Area, Gansu Province, China

Novel prediction methods using artificial intelligence have been developed to improve the identification, discovery, and utilization of new types of mineral resources at new depths or using new technologies. However, most artificial intelligence methods require large training data sets that are often unavailable for mineralization prediction models, leading to inaccuracies. To address this issue, we developed a semi-supervised machine-learning method to identify metallogenic anomalies using the density-based spatial clustering of applications with noise method and autoencoder. The outputs of this method show irregularity in distributions inferred from geological, geochemical, and hyperspectral remote sensing data that match known mineralization locations. We focus on the Daqiao mining area of Gansu Province in China to show that the model predictions are highly consistent with known deposits of the Yinmahe and Daqiao gold mines, and two new prospecting areas have been highlighted for further field confirmation. The accuracy of this semi-supervised learning method was verified by an interdisciplinary intelligent analysis, showing that this method could have wide-reaching applications for improving regional geological surveys. Plain Language Summary Mineral resources are irreplaceable and necessary for modernization. Novel prediction methods using artificial intelligence have been developed to improve the prediction of mineral resources. However, the training data sets are still a big problem for the artificial intelligence methods used in this application. To solve this issue, we developed a semi-supervised machine-learning method to identify metallogenic anomalies using the DBSCAN and autoencoder. The accuracy of this method was verified by an interdisciplinary intelligent analysis, showing that this method could have wide-reaching applications for improving regional geological surveys.