Research Network

Kristian Kersting's research bridges statistical relational learning, deep probabilistic programming, and neuro-symbolic AI. His work on lifted inference and sum-product networks has advanced efficient reasoning in high-dimensional structured domains. Recent focus areas include physics-informed machine learning and trustworthy AI.

Mira Mezini's research focuses on next-generation software engineering methods, including AI-assisted code generation, modular software architectures, and reactive programming. Her group develops tools that use large language models to detect bugs, suggest refactorings, and generate test cases at enterprise scale.

Iryna Gurevych leads one of Europe's largest NLP research groups. Her work on argument mining, stance detection, and scholarly document analysis has been widely adopted. Recent focus includes multilingual fact verification, low-resource NLP for European languages, and AI-assisted scientific writing.

Jan Peters is a leading figure in robot learning, focusing on how robots can acquire complex motor skills through reinforcement learning and imitation. His group has demonstrated learning of table tennis, dexterous manipulation, and locomotion. Close collaboration with the MyoMod team on prosthetic control.

Stefan Roth's group works on visual inference, combining generative and discriminative approaches to computer vision. Research areas include image restoration, optical flow, and scene understanding. Recent work on diffusion models for industrial inspection has direct commercial applications.

Visvanathan Ramesh works on the systems side of AI — designing architectures that deploy deep learning models efficiently on edge devices and in industrial settings. His group has developed frameworks for real-time visual analytics in manufacturing, with direct technology transfer to PanocularAI.

Anna Rohrbach researches the intersection of vision and language — teaching AI to describe, reason about, and answer questions about visual content. Her work on video grounding and referring expressions has been foundational for multimodal AI assistants.

Gemma Roig bridges computational neuroscience and computer vision, studying how biological visual systems can inform better artificial ones. Her work on neural predictivity benchmarks has influenced the design of vision transformers. She collaborates closely with the dianovi team on medical imaging.