Personalized Federated Learning via Learning Dynamic Graphs
Ziran Zhou, Guanyu Gao, Xiaohu Wu, Yan Lyu · 2025
Read Personalized Federated Learning via Learning Dynamic Graphs by Ziran Zhou, Guanyu Gao, Xiaohu Wu, Yan Lyu (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-2503-05474.txt.
Description
Ziran Zhou, Guanyu Gao, Xiaohu Wu, and Yan Lyu personalize federated learning by learning how clients relate in a 2025 study. In federated learning each client keeps its data and trains a local model, but clients' data differ, so one shared model fits many of them poorly. Most personalization work tweaks the global model afterward and ignores how local models are combined in the first place.
The authors argue the missing piece is the graph formed by clients, and that it should be learned and allowed to change. Aggregation then follows that dynamic graph, so similar clients influence each other most. Tailored models follow from better mixing.
Catalog record
| CREATOR | Ziran Zhou, Guanyu Gao, Xiaohu Wu, Yan Lyu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-05474