FedIDM: Achieving Fast and Stable Convergence in Byzantine Federated Learning through Iterative Distribution Matching
He Yang, Dongyi Lv, Wei Xi, Song Ma, Hanlin Gu, Jizhong Zhao · 2026
Read FedIDM: Achieving Fast and Stable Convergence in Byzantine Federated Learning through Iterative Distribution Matching by He Yang, Dongyi Lv, Wei Xi, Song Ma, Hanlin Gu, Jizhong Zhao (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-2604-15115.txt.
Description
Yang and colleagues harden federated learning against Byzantine attackers with FedIDM, built for fast, stable convergence. Earlier robust methods wobble and, under many colluding malicious clients, trade model quality for safety. FedIDM builds trustworthy condensed data through distribution matching, then rejects local updates that stray from its direction or spike its loss. Tests on three benchmarks show fast, stable convergence while keeping model quality acceptable.
Catalog record
| CREATOR | He Yang, Dongyi Lv, Wei Xi, Song Ma, Hanlin Gu, Jizhong Zhao |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 33471 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2604.15115