Manifold GCN: Diffusion-based Convolutional Neural Network for Manifold-valued Graphs
International Journal of Computer Vision · 2026-06 · 134 (6)
DOI: 10.1007/s11263-026-02899-9 ↗Kurz gefasst
Fachartikel in der Zeitschrift International Journal of Computer Vision, online erschienen am 10.06.2026. Verfasst von Martin Hanik und 2 weiteren Autor:innen. Die Arbeit nennt 111 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Deutsche Forschungsgemeinschaft, BIFOLD - Berlin Institute for the Foundations of Learning and Data.
Abstract (Originalsprache)
We propose two graph neural network layers for graphs with features in a Riemannian manifold. First, based on a manifold-valued graph diffusion equation, we construct a diffusion layer that can be applied to an arbitrary number of nodes and graph connectivity patterns. Second, we model a tangent multilayer perceptron by transferring ideas from the vector neuron framework to our general setting. Both layers are equivariant under node permutations and the feature manifold’s isometries. These properties have led to a beneficial inductive bias in many deep-learning tasks. Furthermore, they enable novel, more flexible feature designs. Numerical examples on synthetic data and an Alzheimer’s classification application on triangle meshes of the right hippocampus demonstrate the usefulness of our new layers: While they apply to a much broader class of problems, they outperform task-specific state-of-the-art networks.
Förderung
- Deutsche Forschungsgemeinschaft
- BIFOLD - Berlin Institute for the Foundations of Learning and Data
Zitieren
APA
Hanik, M., Steidl, G., & Tycowicz, C. V. (2026). Manifold GCN: Diffusion-based Convolutional Neural Network for Manifold-valued Graphs. International Journal of Computer Vision, 134(6). https://doi.org/10.1007/s11263-026-02899-9
Vancouver
Hanik M, Steidl G, Tycowicz CV. Manifold GCN: Diffusion-based Convolutional Neural Network for Manifold-valued Graphs. International Journal of Computer Vision. 2026;134(6). doi:10.1007/s11263-026-02899-9
BibTeX
@article{Hanik2026,
author = {Hanik, Martin and Steidl, Gabriele and Tycowicz, Christoph von},
title = {Manifold GCN: Diffusion-based Convolutional Neural Network for Manifold-valued Graphs},
journal = {International Journal of Computer Vision},
year = {2026},
volume = {134},
number = {6},
doi = {10.1007/s11263-026-02899-9}
}
EUVIMED