Presentation – Machine Learning in Geoscience — Is It Any Good?

SGA STUDENT CHAPTER, ERLANGEN
LECTURE SERIES - MINERAL DEPOSITS

Machine Learning in Geoscience — Is It Any Good?
Speaker: Dr. Marc Fassbender (CEO, Quminex)

Monday, July 14th 2025 at 5:15PM (UTC+1)

Location: Übungsraum Mineralogie, GeoZentrum Nordbayern.

Join the talk also via

Link: https://fau.zoom-x.de/j/69752718441
Meeting ID: 697 5271 8441

Dinner at a restaurant after the talk. Please contact the SGA Chapter if you want to join the dinner.
Contact: sga.erlangen@gmail.com

 

Abstract:
Machine learning (ML) and artificial intelligence (AI) have evolved from buzzwords into powerful, widely adopted tools across both industry and academia. In geoscience, however, their application presents unique challenges. Geochemical data—crucial in mineral exploration—are increasingly voluminous, straining traditional interpretation techniques such as scatterplots, discrimination diagrams, and multi-element plots. These conventional methods often neglect the compositional nature of geochemical data, leading to misleading or incorrect conclusions.

In recent years, the field has seen a growing shift toward statistical and ML-based approaches, enabled by advances in software that make these tools more accessible to geoscientists. While this shift offers great potential for deriving quantitative insights, it has also introduced risks of misapplication, underlining the need for a deeper understanding of both data properties and methodological limitations.
This talk explores the value of multivariate analysis, emphasizes the importance of treating geochemical data as compositional, and illustrates best practices to avoid common pitfalls. It also discusses the evolving skill set required in the modern geoscience workforce to effectively leverage these tools for responsible and insightful data analysis.

Marc Fassbender, CEO of Quminex graduated with a Ph.D. from the University of Ottawa, Canada. He possesses expert knowledge of the host rocks associated with actively forming Volcanogenic Massive Sulfide (VMS) deposits on the modern seafloor and their ancient analogues, utilizing petrology, geochemistry, and machine learning techniques. Marc earned both his B.Sc. and M.Sc. in geology from the University of Erlangen-Nuremberg, Germany, where his research focused on Heavy Rare Earth Elements (HREE) enrichment in olivine and the petrogenetic evolution of the Vergenoeg F-Fe-REE deposit in South Africa. He has extensive experience in economic geology, geochemistry, and the application of machine learning in geoscience. Marc specializes in integrating datasets through machine learning to address complex geoscientific challenges.