Evolving graph attention networks for dynamic link prediction
Yucai Jiang, Rongying Shan, Gang Fu, Zhuolin Li, Jingjing Sun, Zhongyun Bao, Feifei Wei · 2026
Read Evolving graph attention networks for dynamic link prediction by Yucai Jiang, Rongying Shan, Gang Fu, Zhuolin Li, Jingjing Sun, Zhongyun Bao, Feifei Wei (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/plos-pone-0349685.txt.
Description
Yucai Jiang and co-authors built a learning system for networks that never sit still, such as social platforms where users and connections appear and vanish daily. Most graph learning tools assume a fixed structure, so they miss how relationships shift over time. Their 2026 study proposes a network that pays selective attention to neighbors and then updates that attention as the graph changes.
A memory component carries the attention weights forward, letting the model adjust which neighbors matter as ties form and fade. In tests on standard benchmark datasets, the system beat leading rival methods at predicting future links. The authors present it as a step toward tools that treat change itself as part of the data, useful wherever networks grow and rewire constantly.
Catalog record
| CREATOR | Yucai Jiang, Rongying Shan, Gang Fu, Zhuolin Li, Jingjing Sun, Zhongyun Bao, Feifei Wei |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 31521 |
| LICENSE | CC_BY |
| RIGHTS | CC BY |
| SOURCE | PLOS |