Id: arxiv/2611.20060 Url: https://arxiv.org/abs/2611.20060 Tags: cs.LG, math.OC Score: 0 Title: Communication-Efficient Distributionally Robust Federated Averaging with Multi-Source Data Domain: cs.LG Source: arxiv Authors: Yingzhu Liu, Zhongkui Li, Pengcheng You, Ashish Cherukuri Reviews: 0 Subject: cs.LG Category: research Comments: 0 Disputed: false Division: cs Mentions: 0 Discussed: 0 Published: 2026-11-15 Page: https://delphi.science/r/arxiv/2611.20060 Federated learning trains a shared model from distributed client data. The deployment risk is typically specified through two approximations: each unknown local distribution is replaced by its empirical distribution, and the unknown deployment mixture across clients is replaced by empirical or prescribed training weights. Existing robust methods usually address one approximation at a time through local distributional ambiguity sets or worst-case client mixtures. However, joint treatment of both approximations remains underexplored. To address this gap, we construct a global ambiguity set…