Full-Spectrum Graph Neural Networks: Expressive and Scalable
Xiaohan Wang, Deyu Bo, Longlong Li, Kelin Xia · 2026
Read Full-Spectrum Graph Neural Networks: Expressive and Scalable by Xiaohan Wang, Deyu Bo, Longlong Li, Kelin Xia (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2605-05759.txt.
Introduction
Spectral graph neural networks filter signals by a graph's eigenvalues, and their expressiveness is capped by a classic combinatorial test. Xiaohan Wang, Deyu Bo, Longlong Li, and Kelin Xia lift the whole construction one order up. Their full-spectrum networks move signals from nodes to pairs of nodes, and filter over pairs of eigenvalues instead of single values.
Classical spectral networks reappear as a diagonal special case. The authors prove the construction breaks the old expressiveness ceiling while staying scalable. The result reframes spectral filtering as a two-dimensional problem with strictly more room to maneuver.
Catalog record
| CREATOR | Xiaohan Wang, Deyu Bo, Longlong Li, Kelin Xia |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 139 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2605.05759