Social Welfare Maximization for Federated Learning with Network Effects

TL;DR

This work addresses social welfare optimization in federated learning when client participation exhibits network effects and data heterogeneity. It introduces a Model Trading and Sharing (MTS) framework that lets clients either participate to train or purchase trained FL models, and develops the socially efficient SEMTS mechanism that uses only model-customer payments to align individual incentives with welfare. The network-effects analysis shows $\epsilon$ depends on participation $K$, data sizes $D_i$, and heterogeneity parameters $\sigma^2$ and $\gamma^2$, yielding non-monotonic welfare effects; SEMTS is shown to maximize social welfare under incomplete information and heterogeneous clients. Experiments on a 20-node FL hardware prototype demonstrate substantial welfare gains over baselines (up to 148.86% on MNIST and 300% on CIFAR-10), validating the practical impact of network-effect aware incentive design in FL.

Abstract

A proper mechanism design can help federated learning (FL) to achieve good social welfare by coordinating self-interested clients through the learning process. However, existing mechanisms neglect the network effects of client participation, leading to suboptimal incentives and social welfare. This paper addresses this gap by exploring network effects in FL incentive mechanism design. We establish a theoretical model to analyze FL model performance and quantify the impact of network effects on heterogeneous client participation. Our analysis reveals the non-monotonic nature of FL network effects. To leverage such effects, we propose a model trading and sharing (MTS) framework that allows clients to obtain FL models through participation or purchase. To tackle heterogeneous clients' strategic behaviors, we further design a socially efficient model trading and sharing (SEMTS) mechanism. Our mechanism achieves social welfare maximization solely through customer payments, without additional incentive costs. Experimental results on an FL hardware prototype demonstrate up to 148.86% improvement in social welfare compared to existing mechanisms.

Figures (6)

• Figure 1: Network Effects of Client Participation in FL.
• Figure 2: The Model Trading and Sharing (MTS) Framework.
• Figure 3: Network Effects of Type-j Client Participation.
• Figure 4: Hardware Prototype of FL.
• Figure 5: Social Efficiency of Different Mechanisms under the MTS Framework.