Beyond Personalization: Federated Recommendation with Calibration via Low-rank Decomposition

2025-06-11

AI Summary

In federated recommendation, global item embedding aggregation induces user embedding drift, degrading personalization performance. To address this, we propose PFedCLR—a novel framework that (i) theoretically models this drift phenomenon for the first time; (ii) introduces a dual-purpose calibration mechanism featuring a low-rank buffer matrix to jointly optimize local user embeddings and global item embeddings, thereby correcting bias while preserving user-specific characteristics; and (iii) adopts a training-upload separation protocol to ensure end-to-end privacy compliance without additional noise. By integrating low-rank decomposition, embedding calibration, and client-level personalization modeling, PFedCLR achieves strict privacy guarantees without sacrificing utility. Extensive experiments on multiple benchmark datasets demonstrate that PFedCLR significantly outperforms state-of-the-art methods in recommendation accuracy, communication efficiency, and rigorous privacy protection.

Technology Category

• Machine Learning: Privacy
• Computer Vision: Bias, Fairness & Privacy
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Privacy-enhancing technologies

Abstract

Federated recommendation (FR) is a promising paradigm to protect user privacy in recommender systems. Distinct from general federated scenarios, FR inherently needs to preserve client-specific parameters, i.e., user embeddings, for privacy and personalization. However, we empirically find that globally aggregated item embeddings can induce skew in user embeddings, resulting in suboptimal performance. To this end, we theoretically analyze the user embedding skew issue and propose Personalized Federated recommendation with Calibration via Low-Rank decomposition (PFedCLR). Specifically, PFedCLR introduces an integrated dual-function mechanism, implemented with a buffer matrix, to jointly calibrate local user embedding and personalize global item embeddings. To ensure efficiency, we employ a low-rank decomposition of the buffer matrix to reduce the model overhead. Furthermore, for privacy, we train and upload the local model before personalization, preventing the server from accessing sensitive information. Extensive experiments demonstrate that PFedCLR effectively mitigates user embedding skew and achieves a desirable trade-off among performance, efficiency, and privacy, outperforming state-of-the-art (SOTA) methods.

Problem

• Mitigates user embedding skew in federated recommendation systems
• Calibrates local user and global item embeddings jointly
• Ensures privacy and efficiency via low-rank decomposition

Innovation

• Low-rank decomposition for efficient calibration
• Dual-function mechanism with buffer matrix
• Privacy-preserving local model training