Decentralized Matrix Factorization with Heterogeneous Differential Privacy

TL;DR

This paper introduces HDPMF, a novel decentralized matrix factorization method that provides heterogeneous differential privacy guarantees, balancing privacy and accuracy in untrusted recommender systems.

Contribution

It is the first to achieve heterogeneous differential privacy in decentralized matrix factorization for untrusted recommenders, using a novel rescaling scheme.

Findings

HDPMF offers better privacy-accuracy trade-offs.

It performs well in high-dimensional models.

It is effective on sparse datasets.

Abstract

Conventional matrix factorization relies on centralized collection of users' data for recommendation, which might introduce an increased risk of privacy leakage especially when the recommender is untrusted. Existing differentially private matrix factorization methods either assume the recommender is trusted, or can only provide a uniform level of privacy protection for all users and items with untrusted recommender. In this paper, we propose a novel Heterogeneous Differentially Private Matrix Factorization algorithm (denoted as HDPMF) for untrusted recommender. To the best of our knowledge, we are the first to achieve heterogeneous differential privacy for decentralized matrix factorization in untrusted recommender scenario. Specifically, our framework uses modified stretching mechanism with an innovative rescaling scheme to achieve better trade off between privacy and accuracy. Meanwhile, by allocating privacy budget properly, we can capture homogeneous privacy preference within a user/item but heterogeneous privacy preference across different users/items. Theoretical analysis confirms that HDPMF renders rigorous privacy guarantee, and exhaustive experiments demonstrate its superiority especially in strong privacy guarantee, high dimension model and sparse dataset scenario.