Maverick Matters: Client Contribution and Selection in Federated Learning

Federated learning (FL) enables collaborative learning between parties, called clients, without sharing the original and potentially sensitive data. To ensure fast convergence in the presence of such heterogeneous clients, it is imperative to timely select clients who can effectively contribute to learning. A realistic but overlooked case of...

Multiplayer boycotts in convex games

We extend the notion of boycotts between players in cooperative games to boycotts between coalitions. We prove that convex games offer a proper setting for studying these games. Boycotts have a heterogeneous effect. Individual players that are targeted by many-on-one boycotts suffer most, while non-participating players may actually benefit...