Wojciech Samek

Wojciech Samek is a Professor in the Department of Electrical Engineering and Computer Science at the Technical University of Berlin and Head of the AI Department at Fraunhofer Heinrich Hertz Institute (HHI), Germany. He holds a joint appointment, bridging academic research and industrial application in artificial intelligence. He is a Fellow at BIFOLD and ELLIS Unit Berlin, and a Principal Investigator in several DFG projects including DeSBi and BIOQIC.

• PhD (Dr. rer. nat.) with distinction, Technical University of Berlin, 2014
• Studies in Computer Science, Humboldt University, Heriot-Watt University, University of Edinburgh

His research centers on Explainable AI (XAI), Trustworthy Deep Learning, and Efficient AI. He pioneered Layer-wise Relevance Propagation (LRP), a foundational method for interpreting deep neural networks. His work spans model interpretation, robustness against adversarial attacks, neural network compression, federated learning, and applications in healthcare, communications, and multimedia.

The recent articles highlight a strong trend toward extending explainability beyond deep models, with research on concept-level explanations, unsupervised learning interpretability, and XAI-driven model improvement. There is a consistent focus on robustness, privacy, and efficiency in distributed learning settings, particularly for federated and edge AI. Applications in medical AI and neuroscience are prominent, emphasizing safety and interpretability in high-stakes domains.

• Best Paper Award, Pattern Recognition (2020)
• Digital Signal Processing Best Paper Prize (2022)
• Fellow, BIFOLD - Berlin Institute for the Foundation of Learning and Data
• Member, Germany's Platform for Artificial Intelligence
• Senior Editor, IEEE TNNLS; Associate Editor, Pattern Recognition, Digital Signal Processing, PLoS ONE
• Area Chair, NeurIPS, ICML, NAACL
• Program Chair, IEEE MLSP 2023
• Contributor to ISO/IEC MPEG-17 NNC standard

Prof. Samek advises a large group of PhD students and leads a vibrant research team at Fraunhofer HHI and TU Berlin. His group has secured significant funding through DFG, BIFOLD, and industrial collaborations. He has co-authored over 200 peer-reviewed papers, many of which are ESI Hot or Highly Cited. He is deeply involved in organizing workshops and tutorials on XAI, federated learning, and neural compression at top venues like NeurIPS, ICML, CVPR, and IEEE conferences.

His lab develops open-source tools such as the LRP Toolbox, Keras Explanation Toolbox, DeepCABAC, and Quantus, promoting reproducibility and adoption of interpretable and efficient AI methods. The team is actively working on next-generation AI that is not only accurate but also transparent, robust, and trustworthy.