I am a PhD student at the University of Tübingen and the Max Planck Research School for Intelligent Systems (IMPRS-IS), working in the lab of Wieland Brendel. The last two years, I interned at different teams of Google Brain. Earlier, I’ve completed my Bachelors and Masters from the University of Göttingen, spent some time doing research at Volkswagen and BMW, and conducted research in the lab of Matthias Bethge.

I started doing research on object detection and segmentation but quickly moved on to understanding failure cases of machine learning models in computer vision tasks through the lense of adversarial perturbations and common corruptions. Currently, I am most interested in understanding why neural networks learn what they learn and what they learn, and building tools for achieving this.

Current research interests: self-supervised learning, adversarial robustness, disentanglement, XAI

Latest Publications

NeurIPS 2023 (Spotlight)

Scale Alone Does not Improve Mechanistic Interpretability in Vision Models

Zimmermann*, R. S., Klein*, T. and Brendel, W.

Preprint

Don’t trust your eyes: on the (un)reliability of feature visualizations

Geirhos*, R., Zimmermann*, R. S., Bilodeau*, B., Brendel, W., Kim, B

Preprint

Sensitivity of Slot-Based Object-Centric Models to their Number of Slots

Zimmermann, R. S., van Steenkiste, S., Sajjadi, M. S. M., Kipf, T., Greff, K.

More publications