Home | Publications | KSS+25

Graph Networks Struggle With Variable Scale

MCML Authors

Link to Profile Daniel Cremers

Daniel Cremers

Prof. Dr.

Director

Computer Vision & Artificial Intelligence

Abstract

Standard graph neural networks assign vastly different latent embeddings to graphs describing the same object at different resolution scales. This precludes consistency in applications and prevents generalization between scales as would fundamentally be needed e.g. in AI4Science. We uncover the underlying obstruction, investigate its origin and show how to overcome it by modifying the message passing paradigm.

inproceedings KSS+25

ICBINB @ICLR 2025

Workshop I Can't Believe It's Not Better: Challenges in Applied Deep Learning at the 13th International Conference on Learning Representations. Singapore, Apr 24-28, 2025.

Authors

C. Koke • Y. Shen • A. Saroha • M. Eisenberger • B. Rieck • M. M. Bronstein • D. Cremers

Links

URL

Research Area

 B1 | Computer Vision

BibTeXKey: KSS+25

#p-cremers #p-wang-xi
Back to Top