Federated Domain Generalization with Data-free On-server Gradient Matching

2025-01-24

AI Summary

To address weak cross-domain generalization in Federated Domain Generalization (FDG) caused by distributed heterogeneous domain data—and the reliance of conventional methods on client-side data uploads or additional communication overhead—this paper proposes a data-free, zero-communication, incremental server-side gradient-matching framework. Our method implicitly aligns gradient directions across domains at the server without requiring raw data or feature transmission from clients. Its core innovation is the first gradient inner-product maximization mechanism for learning domain-invariant gradient directions, enabling communication-free aggregation of domain-specific features. The approach is orthogonal to mainstream FL and FDG algorithms, offering plug-and-play compatibility. Extensive experiments on four federated learning benchmarks (MNIST, EMNIST, CIFAR-10, CIFAR-100) and three FDG benchmarks (PACS, VLCS, OfficeHome) demonstrate consistent and significant improvements over state-of-the-art methods.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Learning & Optimization for CV
• Application Domains: Humanities & Computational Social Science

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Vertical and domain-specific search
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing

Abstract

Domain Generalization (DG) aims to learn from multiple known source domains a model that can generalize well to unknown target domains. One of the key approaches in DG is training an encoder which generates domain-invariant representations. However, this approach is not applicable in Federated Domain Generalization (FDG), where data from various domains are distributed across different clients. In this paper, we introduce a novel approach, dubbed Federated Learning via On-server Matching Gradient (FedOMG), which can emph{efficiently leverage domain information from distributed domains}. Specifically, we utilize the local gradients as information about the distributed models to find an invariant gradient direction across all domains through gradient inner product maximization. The advantages are two-fold: 1) FedOMG can aggregate the characteristics of distributed models on the centralized server without incurring any additional communication cost, and 2) FedOMG is orthogonal to many existing FL/FDG methods, allowing for additional performance improvements by being seamlessly integrated with them. Extensive experimental evaluations on various settings to demonstrate the robustness of FedOMG compared to other FL/FDG baselines. Our method outperforms recent SOTA baselines on four FL benchmark datasets (MNIST, EMNIST, CIFAR-10, and CIFAR-100), and three FDG benchmark datasets (PACS, VLCS, and OfficeHome).

Problem

• Federated Learning
• Domain Generalization
• Data Distribution

Innovation

• Federated Domain Generalization
• Data Integration
• Performance Enhancement