Deep Learning in Mineral Prospectivity Mapping: From Algorithmic Advances to Exploration

Mineral exploration involves significant uncertainty and risk, which poses major challenges for effective decision-making and investment management. In recent years, deep learning techniques have gained increasing attention in mineral prospectivity mapping due to their strong capability to model complex nonlinear relationships and integrate multi-source geoscientific data. This study presents a systematic analytical review of deep learning-based methods applied to mineral prospectivity mapping, with the aim of identifying methodological trends, classifying algorithmic approaches, and evaluating their implications for exploration management. Peer-reviewed articles published in internationally indexed journals were selected and analyzed based on data types, model architectures, learning strategies, evaluation metrics, and interpretability. The reviewed methods were categorized into convolutional neural network-based models, anomaly detection approaches using autoencoders and generative adversarial networks, graph-based deep learning models, and interpretable deep learning frameworks. The results indicate that deep learning methods can significantly improve targeting efficiency and uncertainty reduction; however, their effectiveness strongly depends on data availability, geological complexity, and model interpretability. This review emphasizes the importance of context-driven model selection and provides insights to support risk-aware and informed decision-making in mineral exploration.

کلیدواژه ها:

Mineral Prospectivity Mapping ، Deep Learning ، Exploration Risk ، Decision Making