Graph Augmentation for Cross Graph Domain Generalization
Guanzi Chen †, Jiying Zhang †, Yang Li · 2025
Read Graph Augmentation for Cross Graph Domain Generalization by Guanzi Chen †, Jiying Zhang †, Yang Li (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-2502-18188.txt.
Description
Guanzi Chen, Jiying Zhang, and Yang Li tackled a transfer problem in 2025: teaching a graph network on one network and using it on another with a different weave. Labels are plentiful on the source graph and absent on the target. Rather than retrain the model, they reshape the training data itself.
Their recipe drops weak edges that may be noise, forcing the network to lean on the stable skeleton beneath. It also adds edges between nodes whose features cluster together, building shared structure across graphs. Together, the cuts and adds push the network to keep what travels well. On citation network tests with shifted structure, the method beat the usual augmentation tricks. The authors argue data shaping is an underused lever for graph transfer.
Catalog record
| CREATOR | Guanzi Chen †, Jiying Zhang †, Yang Li |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 41 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2502-18188