Modeling complex systems with graph neural ODEs

Event: GeMMS (2024), Eindhoven, NL

Date: 2024/06/27

Abstract

Combining neural differential equations (NDEs) with relational inductive bias from an underlying graph is a promising modelling approach for dynamical systems from various domains. However, the robustness of this approach across dynamics, graphs, and data quality remains poorly understood. Here we take a first step in this direction by assessing the ability of graph NDEs to generalize to unseen initial conditions on the training graph and new graphs from the same random graph ensemble.

Poster

If you use these materials, please cite:

@misc{laber2026_neuralODE,
  title = {When do neural ordinary differential equations generalize on complex networks},
  author = {Laber, Moritz and Klein, Brennan and Eliassi-Rad, Tina},
  year = {2026},
  archiveprefix = {arXiv}
  eprint = {2602.08980},
  doi = {10.48550/arXiv.2602.08980}
}