FedGO weights each client's prediction by the estimated density ratio of that client's data using GAN discriminators, and proves this weighting makes the ensemble at least as good as the best single model under convex loss.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Provably Near-Optimal Federated Ensemble Distillation with Negligible Overhead
FedGO weights each client's prediction by the estimated density ratio of that client's data using GAN discriminators, and proves this weighting makes the ensemble at least as good as the best single model under convex loss.