Robust Distributed Estimation: Extending Gossip Algorithms to Ranking and Trimmed Means
Anna van Elst, Igor Colin, Stephan Clémençon · 2025
Read Robust Distributed Estimation: Extending Gossip Algorithms to Ranking and Trimmed Means by Anna van Elst, Igor Colin, Stephan Clémençon (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-2505-17836.txt.
Description
Anna van Elst, Igor Colin, and Stephan Clémençon extend a family of chat-based algorithms to survive bad actors in the network. Gossip algorithms let each node in a network swap numbers with its neighbors until all agree, with no central control. Their 2025 paper notes that plain averaging breaks down when some nodes are corrupt or faulty, since one wild value can drag the mean far off.
The authors build GoRank, a gossip method that estimates ranks rather than raw values. On top of it they build GoTrim, which estimates a trimmed mean that discards extreme values. They prove both methods settle at a steady rate as rounds pass, and they test how much corruption GoTrim can withstand before it breaks. Trials on varied network shapes, data patterns, and attack schemes back up the theory.
Catalog record
| CREATOR | Anna van Elst, Igor Colin, Stephan Clémençon |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 87 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-17836