FEDQV: Leveraging Quadratic Voting in FL

June 10, 2024

MLEDGE has already started producing relevant papers in top CS venues. This month, our colleague Tianyue Chu is presenting the paper “FedQV: Leveraging Quadratic Voting in Federated Learning“ at the SIGMETRICS conference (ranked A* according to CORE2023) in Venice, Italy. This paper relates to the FedSecure research component of the project.

Federated Learning (FL) permits different parties to collaboratively train a global model without disclosing their respective local labels. This paper introduces FEDQV, a novel aggregation algorithm built upon the quadratic voting scheme. Its theoretical analysis establishes that FEDQV is a truthful mechanism in which bidding according to one’s true valuation is a dominant strategy that achieves a convergence rate matching that of state-of-the-art methods. Furthermore, its empirical analysis using multiple real-world datasets shows that combining FEDQV with unequal voting “budgets” according to a reputation score increases its performance benefits even further. Finally, it shows that FEDQV can enhance the robustness of FL against poisoning attacks and, combined with Byzantine-robust privacy-preserving mechanisms,  against privacy attacks.