Marcus Rohrbach

Marcus Rohrbach is a Professor for 'Multimodal Reliable AI' at Technische Universität Darmstadt (TU Darmstadt), focusing on the intersection of computer vision, computational linguistics, and machine learning. He specializes in integrating symbolic AI approaches into deep neural networks to create trustworthy systems for human-AI interaction aligned with European values.

• Member of European AI initiatives ELLIS and CLAIRE
• Previously worked at UC Berkeley (2014-2017) and Meta AI/Fundamental AI Research (2017-2022)
• Developed models for describing unknown visual objects

Research Focus:
Advancing reliable AI through multimodal learning, reducing forgetting in continuous learning, and improving robustness with symbolic reasoning. His work enables systems to interpret and describe novel visual elements beyond training data limitations.

Awards:

• Visual Question Answering Challenge (2016, 2018)
• VizWiz Challenge (2018)

He contributes to AI research by sharing models, software, and datasets with the global community, impacting both academia and industry.