Depth-Adaptive Graph Neural Networks via Learnable Bakry-Émery Curvature DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06
Asela Hevapathige, Ahad N. Zehmakan, Qing Wang · 2025
Read Depth-Adaptive Graph Neural Networks via Learnable Bakry-Émery Curvature DOI: XXXXXXX.XXXXXXX Conference: Make sure to enter the correct conference title from your rights confirmation email; June 03–05, 2018; Woodstock, NY ISBN: 978-1-4503-XXXX-X/2018/06 by Asela Hevapathige, Ahad N. Zehmakan, Qing Wang (2025) 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-2503-01079.txt.
Description
Asela Hevapathige, Ahad Zehmakan, and Qing Wang present a 2025 paper on graph neural networks. Such networks learn from linked data by passing messages between points. Their depth is usually fixed for all points, which wastes effort on some and starves others of context.
The method ties depth to a learnable form of curvature, a measure drawn from diffusion geometry. Points with high curvature need few layers, while flat regions gain from deeper passing. A fast approximation keeps the curvature math workable on large graphs. Theory links curvature to how distinct a point's features can stay during learning. Tests on standard benchmarks show steady gains across varied graph learning tasks.
Catalog record
| CREATOR | Asela Hevapathige, Ahad N. Zehmakan, Qing Wang |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 59 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-01079