Research Group – Applied Machine Learning Lab

Rafet Sifa is a Professor of Applied Machine Learning at the University of Bonn, a principal investigator at the Lamarr Institute for Machine Learning and Artificial Intelligence, and head of the Hybrid Intelligence Department at the Fraunhofer Institute for Intelligent Analysis and Information Systems (IAIS) in Germany. His work focuses on building AI systems that are not only mathematically sound but also ready for deployment, bringing together data-driven learning, domain expertise, and optimization. At the Lamarr Institute, he contributes to the development of foundation models tailored to Europe’s linguistic, cultural, and regulatory landscape, with the goal of supporting a responsible and sovereign AI ecosystem. His research interests include foundation models, hybrid representation learning, as well as classical- and quantum-optimization, with a particular emphasis on making systems robust, data-efficient, and reliable even under noisy data, scarce labels, or strict operational constraints.

He has authored more than 150 scientific publications and two books, with contributions ranging from theoretical advances to real-world applications. Together with his teams, he develops systems that understand language, analyze complex data, and support decision-making in domains such as medicine, finance, law, digital forensics, marketing, and scientific research. In financial auditing, for instance, he has been leading the development of end-to-end systems for document understanding, numerical and logical consistency checking, and compliance verification, changing how large audit firms approach routine work and freeing human experts to focus on judgment rather than monotonous tasks. His parallel research in legal AI takes this further, building systems that can reason over national laws and legal codes to answer legal questions and assess whether documents meet specific requirements. In rare disease research, his hybrid learning and patient-matching methods help connect patients with overlapping phenotypes across fragmented data sources, contributing to earlier and more personalized diagnoses. In ophthalmology, his work on AI-based diabetic retinopathy screening supports a move from reactive treatment to earlier, population-scale detection. In radiology, he develops systems deployed in multiple clinics that link free-text clinical reports, imaging data, and other relevant expert information, enabling decision support at a scale that would not be feasible manually. In surgical data science, his teams work on systems that recognize instruments and anatomical structures and detect critical events in the operating room, laying the groundwork for future intelligent surgical assistance. Finally, in behavioral analytics, he has developed machine learning methods to interpret complex user telemetry and temporal interaction data for behavioral profiling, (in)activity prediction, anomaly detection, and adaptive interventions, which are used by numerous companies to inform decision-making in digital platforms, marketing, and forensic investigations alike. Parts of this work have received Best Paper Awards at both renowned machine learning and application-focused conferences and have supported collaborations between academia and industry.

A central theme in his research is a hybrid view of AI that combines learning, simulation, and optimization so that theoretical insight and practical reliability reinforce each other. He leads interdisciplinary research groups at the University of Bonn and Fraunhofer IAIS, working with colleagues from computer science, medicine, the natural sciences, economics, and the humanities on topics such as large language models, hybrid decision support, medical imaging, classical computer vision, and quantum-inspired methods. Beyond research, he is actively involved in teaching and supervision. He teaches graduate courses in machine learning, data mining, and explainable AI, and he aims to create a group culture in which diversity and inclusion are taken seriously and researchers feel encouraged to pursue unconventional ideas. He supervises doctoral and postdoctoral researchers with an emphasis on intellectual independence and the freedom to question established approaches. In addition, he co-organizes international workshops and summer schools, advises companies, and gives invited talks and keynotes to stakeholders in healthcare, insurance, law, public administration, and other sectors that seek to integrate AI into their practice.