Applying differential privacy to matrix factorization

Recommender systems are increasingly becoming an inte-gral part of on-line services. As the recommendations rely on personal user information, there is an inherent loss of pri-vacy resulting from the use of such systems. While several works studied privacy-enhanced neighborhood-based recom-mendations, little attention has been paid to privacy pre-serving latent factor models, like those represented by ma-trix factorization techniques. In this paper, we address the problem of privacy preserving matrix factorization by utiliz-ing differential privacy, a rigorous and provable privacy pre-serving method. We propose and study several approaches for applying differential privacy to matrix factorization, and evaluate the privacy-accuracy trade-offs offered by each ap-proach. We show that input perturbation yields the best recommendation accuracy, while guaranteeing a solid level of privacy protection.