The author explores the interplay between global and local fairness in federated learning using information decomposition, highlighting trade-offs and fundamental limits.
Information-theoretic perspective reveals trade-offs between global and local fairness in Federated Learning.