LEVERAGING NEURO-SYMBOLIC AI FOR ENHANCED MINERAL PREDICTION IN DIVERSE COPPER DEPOSITS

hile machine learning (ML) and deep learning (DL) have found increasing applications in geoscience, particularly in mineral prediction, the use of Neuro-Symbolic AI (NSAI) remains less common despite its advantages. Building on the foundation of Symbolic reasoning and DL, the innovative application of NSAI is explored to predict mineralization systems in various copper deposits. Geochemical data from different copper deposits are collected and labeled, with a Logical Neural Network employed to formulate rules and facts. These rules, derived from economic geology textbooks, and facts, based on real copper deposit data, are embedded within the neural network as a logic layer. This approach bridges symbolic reasoning and neural network capabilities, ensuring the incorporation of domain-specific knowledge into the predictive model.In our research, traditional ML and DL constraints are overcome by integrating expert insights into the model, thereby enhancing its interpretability and operational efficiency. The synergy between DL’s computational power and symbolic reasoning’s precision allows for the creation of models that not only predict but also explain mineralization patterns in copper deposits. Notable progress has been made in developing a practical framework that accurately identifies key geochemical signatures specific to various types of copper deposits. By incorporating rules from economic geology and real-world data, our model demonstrates improved prediction accuracy and explanatory power.

Our findings indicate that NSAI has the potential to revolutionize mineral prediction, providing geoscientists with a powerful tool to enhance the understanding and exploitation of mineral resources. This research underscores the capability of NSAI to substantially improve mineral prediction accuracy, thereby supporting more informed decision-making processes and promoting sustainable resource exploitation. By employing this innovative approach, a more profound comprehension of mineralization systems is realized, contributing significantly to the advancement of geoscience.

Keywords: Neuro-Symbolic AI; Mineral prediction; Logical Neural Networks; Copper deposits; Deep learning