| Titre : |
Application of artificial intelligence to the prospecting of Pb-Zn deposits in Algeria and metallogenic implications |
| Type de document : |
document électronique |
| Auteurs : |
Selma Remidi, Auteur ; Abdelhak Boutaleb, Directeur de thèse ; Salah Eddine Tachi, Directeur de thèse |
| Editeur : |
[S.l.] : [s.n.] |
| Année de publication : |
2026 |
| Importance : |
1 fichier PDF (9.9 Mo) |
| Note générale : |
Mode d'accès : accès au texte intégral par intranet.
Thèse de Doctorat : Génie Minier : Alger, Ecole Nationale Polytechnique : 2026
Bibliogr. p. 127-156 |
| Langues : |
Anglais (eng) |
| Mots-clés : |
MPM
Pb-Zn
Artificial intelligence
Polymetallic
Mineral prospectivity
Machine Learning
Deep Learning |
| Index. décimale : |
D000226 |
| Résumé : |
In the current global context of increasing demand for base metals, mineral exploration has become a high-risk and capital-intensive strategic priority for Algeria’s economic diversification. The primary objective of this study is to reduce exploration uncertainty by identifying and mapping undiscovered Lead–Zinc (Pb–Zn) mineral potential in Northeast Algeria. This research represents the first comprehensive application of multiple predictive modeling approaches for Mineral Prospectivity Mapping (MPM) in the region, capitalizing on its complex tectono-sedimentary and magmatic framework to support informed mining investment decisions.
To achieve this objective, a robust multi-criteria GIS-based framework was developed to compare two distinct modeling paradigms. Knowledge-driven approaches, including the Analytic Hierarchy Process (AHP) and Fuzzy Logic, were employed to translate expert geological knowledge and metallogenic concepts (Source–Drain–Trap) into spatial predictions.
In parallel, data-driven models were implemented using advanced machine learning algorithms, namely Random Forest (RF), Light Gradient Boosting Machine (LightGBM), and Convolutional Neural Networks (CNN). These models were further enhanced through a Stacking ensemble strategy, integrating multiple learners to better capture complex and non-linear geological relationships.
All models were rigorously validated using a comprehensive set of statistical performance metrics. The results indicate that expert-based models, particularly Fuzzy Logic, remain reliable for geological interpretation, achieving an accuracy of 78.27%. However, the data-driven Stacking ensemble model outperformed all other approaches, attaining an exceptional accuracy of 97.67% and an Area Under the Curve (AUC) of 0.983. The final MPM outputs successfully delineate high-prospectivity zones, notably along the coastal magmatic belt and major structural corridors of the External Domain. This study provides a robust decision-support framework for mineral exploration, significantly improving target prioritization and resource management strategies in Algeria. |
Application of artificial intelligence to the prospecting of Pb-Zn deposits in Algeria and metallogenic implications [document électronique] / Selma Remidi, Auteur ; Abdelhak Boutaleb, Directeur de thèse ; Salah Eddine Tachi, Directeur de thèse . - [S.l.] : [s.n.], 2026 . - 1 fichier PDF (9.9 Mo).
Mode d'accès : accès au texte intégral par intranet.
Thèse de Doctorat : Génie Minier : Alger, Ecole Nationale Polytechnique : 2026
Bibliogr. p. 127-156 Langues : Anglais ( eng)
| Mots-clés : |
MPM
Pb-Zn
Artificial intelligence
Polymetallic
Mineral prospectivity
Machine Learning
Deep Learning |
| Index. décimale : |
D000226 |
| Résumé : |
In the current global context of increasing demand for base metals, mineral exploration has become a high-risk and capital-intensive strategic priority for Algeria’s economic diversification. The primary objective of this study is to reduce exploration uncertainty by identifying and mapping undiscovered Lead–Zinc (Pb–Zn) mineral potential in Northeast Algeria. This research represents the first comprehensive application of multiple predictive modeling approaches for Mineral Prospectivity Mapping (MPM) in the region, capitalizing on its complex tectono-sedimentary and magmatic framework to support informed mining investment decisions.
To achieve this objective, a robust multi-criteria GIS-based framework was developed to compare two distinct modeling paradigms. Knowledge-driven approaches, including the Analytic Hierarchy Process (AHP) and Fuzzy Logic, were employed to translate expert geological knowledge and metallogenic concepts (Source–Drain–Trap) into spatial predictions.
In parallel, data-driven models were implemented using advanced machine learning algorithms, namely Random Forest (RF), Light Gradient Boosting Machine (LightGBM), and Convolutional Neural Networks (CNN). These models were further enhanced through a Stacking ensemble strategy, integrating multiple learners to better capture complex and non-linear geological relationships.
All models were rigorously validated using a comprehensive set of statistical performance metrics. The results indicate that expert-based models, particularly Fuzzy Logic, remain reliable for geological interpretation, achieving an accuracy of 78.27%. However, the data-driven Stacking ensemble model outperformed all other approaches, attaining an exceptional accuracy of 97.67% and an Area Under the Curve (AUC) of 0.983. The final MPM outputs successfully delineate high-prospectivity zones, notably along the coastal magmatic belt and major structural corridors of the External Domain. This study provides a robust decision-support framework for mineral exploration, significantly improving target prioritization and resource management strategies in Algeria. |
|
D000100 
Ajouter le résultat dans votre panier Faire une suggestion Affiner la rechercheApplication of artificial intelligence to the prospecting of Pb-Zn deposits in Algeria and metallogenic implications (2026)
