One-Shot Clustering for Federated Learning Under Clustering-Agnostic Assumption
Maciej Krzysztof Zuziak [email protected], Roberto Pellungrini [email protected], Salvatore Rinzivillo [email protected] · 2025
Read One-Shot Clustering for Federated Learning Under Clustering-Agnostic Assumption by Maciej Krzysztof Zuziak [email protected], Roberto Pellungrini [email protected], Salvatore Rinzivillo [email protected] (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-2509-01587.txt.
Description
Federated learning trains one model from many clients without moving their private data, but a single model fits diverse clients poorly. Maciej Krzysztof Zuziak, Roberto Pellungrini, and Salvatore Rinzivillo work on clustered federated learning, where clients are grouped so each cohort gets a more personal model. The hard part is knowing when and how to form those groups.
Their algorithm, OCFL, is clustering-agnostic, meaning it does not assume a set number or shape of groups in advance. It computes cosine distances between clients' gradients and watches a temperature measure that signals when the shared model begins to converge. That signal picks the earliest suitable moment to cluster, in a single shot. The authors present the method as a practical answer to a subfield whose basic settings still differ from standard federated learning.
Catalog record
| CREATOR | Maciej Krzysztof Zuziak [email protected], Roberto Pellungrini [email protected], Salvatore Rinzivillo [email protected] |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 104 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-01587