Banner Banner
Icon

November 27, 2025

NeurIPS 2025 Conference Contributions

Eighteen research contributions from BIFOLD teams across core areas of machine learning and AI

The Thirty-Ninth Annual Conference on Neural Information Processing Systems (NeurIPS) will take place across two locations in 2025: from November 30 to December 5 in Mexico City, and from December 2 to 7 at the San Diego Convention Center. Since its founding in 1987, NeurIPS has evolved into one of the world’s leading interdisciplinary forums for advances in machine learning and computational neuroscience. The multi-track program spans invited talks, peer-reviewed oral and poster presentations, demonstrations, symposia, tutorials, and an industry-focused expo, complemented by a rich ecosystem of topical workshops that foster interactive discussion and early-stage idea exchange.

This year, 18 contributions originate from BIFOLD researchers or were developed in close collaboration with BIFOLD teams — underscoring the institute’s strong presence in cutting-edge machine learning research. In addition, Farnoush Rezaei Jafari from the Machine Learning Group has been selected as a Top Reviewer for NeurIPS 2025, recognizing her outstanding commitment to the scientific review process.

Below is an overview of BIFOLD’s contributions:

ImageNet-trained CNNs are not biased towards texture: Revisiting feature reliance through controlled suppression. Tom Burgert, Oliver Stoll, Paolo Rota, Begüm Demir.

  • Code
  • Link
  • Th, Dec 4, 20:00 CET -- San Diego Poster Session 3
  • Th, Dec 4, 19:40 CET -- Oral 3B Deep Learning 2

Beyond Scalars: Concept-Based Alignment Analysis in Vision Transformers. Johanna Vielhaben, Dilyara Bareeva, Jim Berend, Wojciech Samek, Nils Strodthoff.

  • Paper
  • Th, Dec 4, 01:30 CET -- San Diego Poster Session 2

Capturing Polysemanticity with PRISM: A Multi-Concept Feature Description Framework. Laura Kopf, Nils Feldhus, Kirill Bykov, Philine Lou Bommer, Anna Hedström, Marina M.-C. Höhne, Oliver Eberle.

  • Paper
  • We, Dec 3, 20:00 CET -- San Diego Poster Session 1

Fractional Diffusion Bridge Models. Gabriel Nobis, Maximilian Springenberg, Arina Belova, Rembert Daems, Christoph Knochenhauer, Manfred Opper, Tolga Birdal, and Wojciech Samek.

  • Paper
  • We, Dec 3, 20:00 CET -- San Diego Poster Session 1

Manipulating Feature Visualizations with Gradient Slingshots. Dilyara Bareeva, Marina M.-C. Höhne, Alexander Warnecke, Lukas Pirch, Klaus-Robert Müller, Konrad Rieck, Sebastian Lapuschkin, Kirill Bykov.

  • Paper
  • We, Dec 3, 20:00 CET -- Mexico City Poster Session 1

Path Gradients after Flow Matching. Lorenz Vaitl, Leon Klein.

  • Paper
  • Sa, Dec 6, 01:30 CET -- San Diego Poster Session 6

Sampling 3D Molecular Conformers with Diffusion Transformers*. J. Thorben Frank, Winfried Ripken, Gregor Lied, Klaus-Robert Müller, Oliver T. Unke, Stefan Chmiela.

  • Code
  • Paper
  • Th, Dec 4, 20:00 CET -- San Diego Poster Session 3

*Poster NeurIPS 2025 poster  + 2 Workshops (ML4PS and AI4Science)


Smoothed Differentiation Efficiently Mitigates Shattered Gradients in Explanations. Adrian Hill, Neal McKee, Johannes Maeß, Stefan Blücher, Klaus-Robert Müller. 

  • Code
  • Paper
  • Th, Dec 4, 01:30 CET -- San Diego Poster Session 2

The Atlas of In-Context Learning: How Attention Heads Shape In-Context Retrieval Augmentation. Patrick Kahardipraja, Reduan Achtibat, Thomas Wiegand, Wojciech Samek, Sebastian Lapuschkin.

  • Paper
  • Th, Dec 4, 20:00 CET -- Mexico City Poster Session 3

Correcting misinterpretations of additive models. Clark, B., Wilming, R., Schulz, H., Zhumagambetov, R., Panknin, D., & Haufe, S.

  • Paper 
  • Fr, Dec 5, 01:30 CET -- San Diego Poster Session 4

Minimizing False-Positive Attributions in Explanations of Non-Linear Models. Gjølbye, A., Haufe, S., & Hansen, L. K.. 

  • Paper
  • Sa, Dec 6, 01:30 CET -- San Diego Poster Session 6

DMRG Quantum Chemistry Dataset for Multi-Reference Machine Learning. Stefan Gugler, Nina Glaser.

From Prediction to Proposal in Catalysis: robust evaluation, LRP explanations, and relevance-guided candidate generation***. Parastoo Semnani, Mihail Bogojeski, Florian Bley, Zizheng Zhang, Qiong Wu, Thomas Kneib, Jan Herrmann, Christoph Weisser, Florina Patcas, Klaus-Robert Müller.

Iterative Inference in a Chess-Playing Neural Network. Elias Sandmann, Sebastian Lapuschkin, Wojciech Samek

RelP: Faithful and Efficient Circuit Discovery in Language Models via Relevance Patching. Farnoush Rezaei Jafari, Oliver Eberle, Ashkan Khakzar, Neel Nanda.

Towards Mechanistic Defenses Against Typographic Attacks in CLIP. Lorenz Hufe, Constantin Venhoff, Maximilian Dreyer, Sebastian Lapuschkin, Wojciech Samek. 

Sampling 3D Molecular Conformers with Diffusion Transformers*. J. Thorben Frank, Winfried Ripken, Gregor Lied, Klaus-Robert Müller, Oliver T. Unke, Stefan Chmiela.

*Poster NeurIPS 2025 poster  + 2 Workshops (ML4PS and AI4Science)

Spectral Uncertainty Decomposition for Sparse Bayesian Learning**. Ismail Huseynov and Stefan Haufe.