Preprint
Aug 2026
How Much Regularization Survives Averaging? Update Masking in Federated Learning
It is proved that when each client draws its own mask, federated averaging weakens that charge by exactly the cohort size, and that giving every client the same mask brings it back by a factor equal to the inverse gradient diversity of the cohort.
Wenhao Yan, Fu Kuroda, Yu-Cheng Jin et al.
· 0 citations