Lightweight and Robust Federated Data Valuation

2025-09-29

AI Summary

Federated learning suffers from poor robustness under non-IID data and malicious clients, while existing Shapley-value-based contribution evaluation methods incur prohibitive computational overhead and lack scalability. To address this, we propose FedIF—a lightweight, trajectory-based influence estimation framework for client contribution assessment. FedIF efficiently approximates each client’s contribution to the global model by analyzing gradient trajectories on a public validation set, incorporating local weight normalization and influence score smoothing. We theoretically establish a tighter bound on the global loss variation under noise compared to prior approaches. Extensive experiments on CIFAR-10 and Fashion-MNIST demonstrate that FedIF significantly outperforms Shapley-based methods in robustness against label noise, gradient perturbations, and adversarial attacks. Moreover, it reduces model aggregation overhead by up to 450×, enabling scalable and reliable federated learning in heterogeneous and hostile environments.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Adversarial Attacks & Robustness
• Multiagent Systems: Adversarial Agents

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Security and Privacy: Large-scale security measurements
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Federated learning (FL) faces persistent robustness challenges due to non-IID data distributions and adversarial client behavior. A promising mitigation strategy is contribution evaluation, which enables adaptive aggregation by quantifying each client's utility to the global model. However, state-of-the-art Shapley-value-based approaches incur high computational overhead due to repeated model reweighting and inference, which limits their scalability. We propose FedIF, a novel FL aggregation framework that leverages trajectory-based influence estimation to efficiently compute client contributions. FedIF adapts decentralized FL by introducing normalized and smoothed influence scores computed from lightweight gradient operations on client updates and a public validation set. Theoretical analysis demonstrates that FedIF yields a tighter bound on one-step global loss change under noisy conditions. Extensive experiments on CIFAR-10 and Fashion-MNIST show that FedIF achieves robustness comparable to or exceeding SV-based methods in the presence of label noise, gradient noise, and adversarial samples, while reducing aggregation overhead by up to 450x. Ablation studies confirm the effectiveness of FedIF's design choices, including local weight normalization and influence smoothing. Our results establish FedIF as a practical, theoretically grounded, and scalable alternative to Shapley-value-based approaches for efficient and robust FL in real-world deployments.

Problem

• Addresses computational inefficiency in federated learning client evaluation methods
• Mitigates robustness challenges from non-IID data and adversarial clients
• Provides scalable alternative to Shapley-value-based contribution assessment

Innovation

• Uses trajectory-based influence estimation for client contributions
• Computes normalized smoothed scores from lightweight gradient operations
• Reduces aggregation overhead significantly while maintaining robustness