Strategic Incentivization for Locally Differentially Private Federated Learning

2025-08-09

AI Summary

In federated learning, local differential privacy (LDP) safeguards client data privacy but degrades global model accuracy due to gradient perturbation, posing a fundamental privacy–utility trade-off. This work is the first to formalize this trade-off as a strategic game between clients and the server. We propose a dynamic token-based incentive mechanism wherein the server allocates tokens to clients according to their contribution quality—measured by gradient utility—thereby encouraging clients to adaptively reduce LDP noise magnitude. Our approach integrates LDP-compliant gradient perturbation, game-theoretic modeling of client-server interactions, and token-driven access control for privacy-utility coordination. Experiments demonstrate that, under strict LDP guarantees (ε ≤ 2), our mechanism accelerates model convergence by 37% and improves final test accuracy by 4.2–6.8 percentage points, achieving dynamic co-optimization of privacy preservation and model utility.

Technology Category

• Machine Learning: Privacy
• Game Theory and Economic Paradigms: Mechanism Design
• Multiagent Systems: Mechanism Design

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Privacy-enhancing technologies
• Economics, Online Markets and Human Computation: Fairness, privacy, and diversity in economic environments

Abstract

In Federated Learning (FL), multiple clients jointly train a machine learning model by sharing gradient information, instead of raw data, with a server over multiple rounds. To address the possibility of information leakage in spite of sharing only the gradients, Local Differential Privacy (LDP) is often used. In LDP, clients add a selective amount of noise to the gradients before sending the same to the server. Although such noise addition protects the privacy of clients, it leads to a degradation in global model accuracy. In this paper, we model this privacy-accuracy trade-off as a game, where the sever incentivizes the clients to add a lower degree of noise for achieving higher accuracy, while the clients attempt to preserve their privacy at the cost of a potential loss in accuracy. A token based incentivization mechanism is introduced in which the quantum of tokens credited to a client in an FL round is a function of the degree of perturbation of its gradients. The client can later access a newly updated global model only after acquiring enough tokens, which are to be deducted from its balance. We identify the players, their actions and payoff, and perform a strategic analysis of the game. Extensive experiments were carried out to study the impact of different parameters.

Problem

• Balancing privacy and accuracy in federated learning with LDP
• Incentivizing clients to reduce noise for better model performance
• Strategic token-based mechanism for privacy-accuracy trade-off management

Innovation

• Token-based incentivization for LDP in FL
• Game theory models privacy-accuracy trade-off
• Clients earn tokens for lower noise gradients