Federated Learning by Utility-Constrained Stochastic Aggregation for Improving Rational Participation

TL;DR

This paper introduces FedUCA, a framework that incentivizes rational clients to participate in federated learning by balancing their utility thresholds, thereby improving overall model performance.

Contribution

The paper proposes a novel utility-constrained stochastic aggregation method to sustain client participation in federated learning with heterogeneous, rational agents.

Findings

FedUCA increases client retention rates.

Global model performance improves with FedUCA.

Framework effectively balances client utility and participation.

Abstract

Federated Learning (FL) algorithms implicitly assume that clients passively comply with server-side orchestration by sharing local model updates upon server request. However, this overlooks an important aspect in real-world cross-silo environments: clients are often rational agents who may prioritize their utilities such as local model performance over that of the global model. In settings with significant statistical heterogeneity, rational clients may opt out of the federation if the perceived benefits of collaboration fail to meet their local utility thresholds. Such attrition degrades the global model performance and can lead to the collapse of the federated training process. In this work, we introduce FedUCA, (Federated Learning by Utility-Constrained Stochastic Aggregation for Improving Rational Participation), a framework that formalizes the server's role as an optimizer seeking to maximize global model performance by sustaining client participation. We substantiate our framework through extensive experiments on standard datasets demonstrating that by prioritizing participation feasibility, FedUCA achieves significantly higher client retention and, consequently, a superior global model performance.