Abstract:Collecting and training over sensitive personal data raise severe privacy concerns in personalized recommendation systems, and federated learning can potentially alleviate the problem by training models over decentralized user <a class="link-external link-http" href="http://data.However" rel="external noopener nofollow">this http URL</a>, a theoretically private solution in both the training and serving stages of federated recommendation is essential but still <a class="link-external link-http" href="http://lacking.Furthermore" rel="external noopener nofollow">this http URL</a>, naively applying differential privacy (DP) to the two stages in federated recommendation would fail to achieve a satisfactory trade-off between privacy and utility due to the high-dimensional characteristics of model gradients and hidden <a class="link-external link-http" href="http://representations.In" rel="external noopener nofollow">this http URL</a> this work, we propose a federated news recommendation method for achieving a better utility in model training and online serving under a DP <a class="link-external link-http" href="http://guarantee.We" rel="external noopener nofollow">this http URL</a> first clarify the DP definition over behavior data for each round in the life-circle of federated recommendation <a class="link-external link-http" href="http://systems.Next" rel="external noopener nofollow">this http URL</a>, we propose a privacy-preserving online serving mechanism under this definition based on the idea of decomposing user embeddings with public basic vectors and perturbing the lower-dimensional combination coefficients. We apply a random behavior padding mechanism to reduce the required noise intensity for better utility. Besides, we design a federated recommendation model training method, which can generate effective and public basic vectors for serving while providing DP for training participants. We avoid the dimension-dependent noise for large models via label permutation and differentially private attention modules. Experiments on real-world news recommendation datasets validate that our method achieves superior utility under a DP guarantee in both training and serving of federated news recommendations.

FedRec: Privacy-Preserving News Recommendation with Federated Learning

Efficient-FedRec: Efficient Federated Learning Framework for Privacy-Preserving News Recommendation

Randomization is all you need: A privacy-preserving federated learning framework for news recommendation

Uni-FedRec: A Unified Privacy-Preserving News Recommendation Framework for Model Training and Online Serving

PrivateRec: Differentially Private Model Training and Online Serving for Federated News Recommendation

PrivateRec: Differentially Private Training and Serving for Federated News Recommendation

Privacy-enhanced Lossless Federated Recommendation with Client-Selection

A trusted recommendation scheme for privacy protection based on federated learning

User Consented Federated Recommender System Against Personalized Attribute Inference Attack

A Federated Learning Approach for Privacy Protection in Context-Aware Recommender Systems.

Fast-adapting and privacy-preserving federated recommender system

A split-federated learning and edge-cloud based efficient and privacy-preserving large-scale item recommendation model

PDC-FRS: Privacy-preserving Data Contribution for Federated Recommender System

FedDeepFM: A Factorization Machine-Based Neural Network for Recommendation in Federated Learning

FedRec++: Lossless Federated Recommendation with Explicit Feedback

PerFedRec++: Enhancing Personalized Federated Recommendation with Self-Supervised Pre-Training

FedCL: Federated Contrastive Learning for Privacy-Preserving Recommendation

Decentralized Federated Recommendation with Privacy-Aware Structured Client-Level Graph

Federated News Recommendation with Fine-grained Interpolation and Dynamic Clustering

FedRec+: Enhancing Privacy and Addressing Heterogeneity in Federated Recommendation Systems

UA-FedRec: Untargeted Attack on Federated News Recommendation