FedABC : Attention-Based Client Selection for Federated Learning with Long-Term View
Wenxuan Ye 1 2 , Xueli An 1 , Junfan Wang 3 , Xueqiang Yan 3 , Georg Carle 2 · 2025
Read FedABC : Attention-Based Client Selection for Federated Learning with Long-Term View by Wenxuan Ye 1 2 , Xueli An 1 , Junfan Wang 3 , Xueqiang Yan 3 , Georg Carle 2 (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-2507-20871.txt.
Description
Wenxuan Ye and four co-authors propose in 2025 FedABC, a method to pick which devices join each round of federated learning with a long-term view. In federated learning, phones and edge nodes train a shared model without sending raw data. Uneven local data slows training, and eager joining drains bandwidth.
Attention-style scoring ranks clients by model likeness and unique value to the shared model. An optimization layer steers picks as the model's needs evolve. A later-is-better rule raises joining rates in late rounds, when fine gains matter most. Simulations on image data back the design. The method suits 6G networks where native AI support is a stated goal.
Catalog record
| CREATOR | Wenxuan Ye 1 2 , Xueli An 1 , Junfan Wang 3 , Xueqiang Yan 3 , Georg Carle 2 |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 30 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-20871