NEWS

New Whitepaper on XAI
Prof. Dr. Wojciech Samek is co-author of the new white paper “Explainable AI”, published today by the Plattform Lernende Systeme. Together with colleagues from the “Technological Enablers and Data Science” working group of the Platform Lernende Systeme, they explored different application areas, guided by the question: Who should be told what, how and for what purpose?

SIGMOD 2025 Conference Contributions
BIFOLD researchers will present multiple contributions at the SIGMOD/PODS 2025 conference, scheduled for June 22–27, 2025, in Berlin.
EVENTS
BLISS AI SPEAKER SERIES #21: Progress in AI Safety and Security
Dr. Adel Bibi, Senior Researcher at University of Oxford, will discuss "Progress in AI Safety and Security". The talk will last approximately 45 minutes. After the talk, seize the opportunity to connect with fellow AI enthusiasts to share ideas and questions while enjoying free drinks and pizza. Door close by 7.15pm, so please come early!
Lunch Talk: Robin Geyer - Measuring Orthogonality in Representations of Generative Models
In unsupervised representation learning, models aim to distill essential features from high-dimensional data into lower-dimensional learned representations, guided by inductive biases. Understanding the characteristics that make a good representation remains a topic of ongoing research.
*Data collected from 2020 onwards.
OPPORTUNITIES
BIFOLD welcomes job applications from interested parties year-round.

PUBLICATIONS
Patrick Kahardipraja, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin
The Atlas of In-Context Learning: How Attention Heads Shape In-Context Retrieval Augmentation
Qianli Wang, Van Bach Nguyen, Nils Feldhus, Luis Felipe Villa-Arenas, Christin Seifert, Sebastian Möller, Vera Schmitt
Truth or Twist? Optimal Model Selection for Reliable Label Flipping Evaluation in LLM-based Counterfactuals
Qianli Wang, Mingyang Wang, Nils Feldhus, Simon Ostermann, Yuan Cao, Hinrich Schütze, Sebastian Möller, Vera Schmitt
Through a Compressed Lens: Investigating the Impact of Quantization on LLM Explainability and Interpretability

GRADUATE SCHOOL
Based on a highly competitive application process, the BIFOLD Graduate School offers an innovative fast-track PhD Program for students holding a bachelor’s degree, as well as a PhD Program for students with a master’s degree.