Learning to Collaborate via Structures: Cluster-Guided Item Alignment for Federated Recommendation
Yuchun Tu, Zhiwei Li, Bingli Sun, Yixuan Li, Xiao Song · 2026
Read Learning to Collaborate via Structures: Cluster-Guided Item Alignment for Federated Recommendation by Yuchun Tu, Zhiwei Li, Bingli Sun, Yixuan Li, Xiao Song (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/arxiv-2602-21957.txt.
Description
Yuchun Tu and four co-authors rethink how recommendation systems learn across many devices without pooling private user data. Standard federated methods sync full item vectors between server and devices, assuming exact coordinate alignment is what makes collaboration work. The authors argue instead that the relative relationships among items matter more than shared coordinates.
Their framework, CGFedRec, has the server learn item clusters and send only compact cluster labels downstream, so devices need not hold global item vectors at all. Local vectors stay free to vary, capturing personal taste while global structure keeps them consistent. Experiments show much lower communication costs with recommendation accuracy that beats rivals across several data sets.
Catalog record
| CREATOR | Yuchun Tu, Zhiwei Li, Bingli Sun, Yixuan Li, Xiao Song |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 45 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-21957