Research
As a researcher, I work on artificial intelligence and its application in practice. This is where I collect my papers, publications and articles.
Focus areas
My work sits at the intersection of machine learning and practical application.
Precise detection, classification and localization of objects and defects in images, for example to optimize inspection processes.
From classical learning methods to deep neural networks: robust models that deliver reliable results even with little annotated data.
Large language models and retrieval-augmented generation: making knowledge from documents accessible and generating well-founded answers.
Capturing, linking and using knowledge across distributed systems: from the data source to a consistent knowledge base.
AI that people enjoy using: understandable results, intuitive interaction and trust through transparency.
Publications
Marvin Woller, Jan-Vincent Mock, Barbara Essendorfer
Computer vision for preprocessing STANAG documents: a model based on DocLayout-YOLO detects layout elements on document pages, enabling efficient, automated preparation for downstream analysis.
Jan-Vincent Mock, Marvin Woller, Barbara Essendorfer
Semantic search across heterogeneous ISR data in coalition shared data systems: retrieval-augmented generation connects distributed data holdings and makes them searchable in natural language.
Marvin Woller, Matthias Grundmann, Hannes Hartenstein
The topology of the Bitcoin P2P network is deliberately hidden. This preprint uses iterative algorithms based on time-ordered INV announcements to show that the network topology can still be inferred, validated in simulation and with real-world Bitcoin data.
Whether you have questions about a publication, ideas for joint research or want to bring current methods into your company: I look forward to hearing from you.
Get in touch