08.05.2025

Christian Koke Wins ICLR MLMP 2025 Best Paper Award

Our Junior Member Honored for Work on Multiscale Graph Networks

MCML Junior Member Christian Koke and his co-authors have received the ICLR 2025 MLMP Best Paper Award for their paper “On Incorporating Scale into Graph Networks”. The award includes 2,000 GPU-hours from Nebius and honors their outstanding contribution to multiscale graph network research.

Congratulations from us!

Check out the full paper:

C. Koke, Y. Shen, A. Saroha, M. Eisenberger, B. Rieck, M. M. B. Michael M. Bronstein and D. Cremers.
On Incorporating Scale into Graph Networks.
MLMP @ICLR 2025 - Workshop on Machine Learning Multiscale Processes at the 13th International Conference on Learning Representations (ICLR 2025). Singapore, Apr 24-28, 2025. Best Paper Award. URL

Abstract

Standard graph neural networks assign vastly different latent embeddings to graphs describing the same physical system at different resolution scales. This precludes consistency in applications and prevents generalization between scales as would fundamentally be needed in many scientific applications. We uncover the underlying obstruction, investigate its origin and show how to overcome it.

MCML Authors