Power to the Clients: Federated Learning in a Dictatorship Setting
Mohammadsajad Alipour, Mohammad Mohammadi Amiri · 2025
Read Power to the Clients: Federated Learning in a Dictatorship Setting by Mohammadsajad Alipour, Mohammad Mohammadi Amiri (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-2510-22149.txt.
Description
Mohammadsajad Alipour and Mohammad Mohammadi Amiri imagine the worst coworker in federated learning, the scheme where many clients train one shared model without pooling their private data. Their invention is the dictator client: a malicious participant who can wipe every other client's contribution out of the server model while keeping its own fully intact.
The pair design concrete attacks that grant that power and analyze what happens to training in each case. They then complicate the picture with several dictators at once, cooperating, ignoring each other, or banding together with plans to betray the alliance later. For every arrangement they track the effect on whether the global model still converges. The exercise maps a sharp vulnerability class so that defenses can be designed against defined attacks rather than vague fears.
Catalog record
| CREATOR | Mohammadsajad Alipour, Mohammad Mohammadi Amiri |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 62 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-22149