FedeCouple: Fine-Grained Balancing of Global-Generalization and Local-Adaptability in Federated Learning

2025-11-12

AI Summary

In federated learning, the imbalance between global generalization and local adaptation weakens coupling between feature extractors and classifiers, degrading model performance. To address this, we propose FedeCouple—a personalized federated learning framework that jointly optimizes global and local feature representations for fine-grained coordination. It introduces non-transmissive anchor-guided feature space refinement to preserve data privacy and minimize communication overhead, and incorporates dynamic knowledge distillation to enhance classifier generalization. Theoretically, FedeCouple supports convergence analysis under non-convex objectives. Extensive experiments on five image classification benchmarks demonstrate consistent superiority over nine state-of-the-art baselines, achieving up to a 4.3% absolute accuracy gain. Results validate FedeCouple’s effectiveness, stability, scalability, and security—particularly in heterogeneous, privacy-sensitive settings.

Technology Category

• Machine Learning: Distributed Machine Learning & Federated Learning
• Computer Vision: Learning & Optimization for CV
• Search and Optimization: Learning to Search

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Systems and Infrastructure for Web, Mobile and WoT: Federated Web and WoT systems, including distributed, federated and edge-based data processing
• Security and Privacy: Security and privacy of machine learning and AI applications

Abstract

In privacy-preserving mobile network transmission scenarios with heterogeneous client data, personalized federated learning methods that decouple feature extractors and classifiers have demonstrated notable advantages in enhancing learning capability. However, many existing approaches primarily focus on feature space consistency and classification personalization during local training, often neglecting the local adaptability of the extractor and the global generalization of the classifier. This oversight results in insufficient coordination and weak coupling between the components, ultimately degrading the overall model performance. To address this challenge, we propose FedeCouple, a federated learning method that balances global generalization and local adaptability at a fine-grained level. Our approach jointly learns global and local feature representations while employing dynamic knowledge distillation to enhance the generalization of personalized classifiers. We further introduce anchors to refine the feature space; their strict locality and non-transmission inherently preserve privacy and reduce communication overhead. Furthermore, we provide a theoretical analysis proving that FedeCouple converges for nonconvex objectives, with iterates approaching a stationary point as the number of communication rounds increases. Extensive experiments conducted on five image-classification datasets demonstrate that FedeCouple consistently outperforms nine baseline methods in effectiveness, stability, scalability, and security. Notably, in experiments evaluating effectiveness, FedeCouple surpasses the best baseline by a significant margin of 4.3%.

Problem

• Balancing global generalization with local adaptability in federated learning systems
• Addressing insufficient coordination between feature extractors and classifiers in personalized FL
• Improving model performance in privacy-preserving scenarios with heterogeneous client data

Innovation

• Balances global generalization and local adaptability
• Employs dynamic knowledge distillation for classifiers
• Uses local anchors to refine feature space