FedPF frames privacy and fairness in federated learning as a zero-sum game, shows privacy reduces bias-detection power under finite samples, and cuts discrimination up to 42.9% while retaining competitive accuracy.
Counter- factual fairness,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.LG 1years
2025 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
FedPF: Accurate Target Privacy Preserving Federated Learning Balancing Fairness and Utility
FedPF frames privacy and fairness in federated learning as a zero-sum game, shows privacy reduces bias-detection power under finite samples, and cuts discrimination up to 42.9% while retaining competitive accuracy.