eris : Enhancing Privacy and Communication Efficiency in Serverless Federated Learning
Dario Fenoglio, Pasquale Polverino, Jacopo Quizi, Martin Gjoreski, Marc Langheinrich · 2026
Read eris : Enhancing Privacy and Communication Efficiency in Serverless Federated Learning by Dario Fenoglio, Pasquale Polverino, Jacopo Quizi, Martin Gjoreski, Marc Langheinrich (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-2602-08617.txt.
Description
Scaling federated learning to billion-parameter models forces a painful trade among privacy, scalability, and utility. Dario Fenoglio and five co-authors introduce ERIS, a frame built on federated shard aggregation. Their arXiv paper splits each client update into separate shards that are combined across several client-side aggregators.
Sharding removes the central aggregation bottleneck, limits what any single observer can see, and preserves the central update after the shards are put back together. Optional distributed shifted compression cuts payloads and exposed coordinates further. Theory shows convergence is preserved and leakage is bounded by the visible share, shrinking with more aggregators. Tests match FedAvg utility while cutting bottlenecks and resisting inference attacks. Readers get a cryptography-light path to large-scale federated learning.
Catalog record
| CREATOR | Dario Fenoglio, Pasquale Polverino, Jacopo Quizi, Martin Gjoreski, Marc Langheinrich |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 147 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602-08617