# Privacy and Fairness in Recommender Systems via Adversarial Training of User Representations

arXiv: https://arxiv.org/abs/1807.03521 · HTML: https://mlsanity.com/paper/1807.03521/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 10 Jul 2018.

## Built on by (most cited first, 8 of 8)

- [Contrastive Learning for Fair Representations](https://mlsanity.com/paper/2109.10645/) (2021)
- [A Comprehensive Survey on Trustworthy Recommender Systems](https://mlsanity.com/paper/2209.10117/) (2022)
- [Fairness in Recommendation: Foundations, Methods and Applications](https://mlsanity.com/paper/2205.13619/) (2022)
- [Semi-supervised Adversarial Learning for Complementary Item Recommendation](https://mlsanity.com/paper/2303.05812/) (2023)
- [Collaborative Image Understanding](https://mlsanity.com/paper/2210.11907/) (2022)
- [Consumer-side Fairness in Recommender Systems: A Systematic Survey of Methods and Evaluation](https://mlsanity.com/paper/2305.09330/) (2023)
- [Providing Previously Unseen Users Fair Recommendations Using Variational Autoencoders](https://mlsanity.com/paper/2308.15230/) (2023)
- [Exploring How Fair Model Representations Relate to Fair Recommendations](https://mlsanity.com/paper/2603.24396/) (2026)

## Abstract (arXiv)

Latent factor models for recommender systems represent users and items as low dimensional vectors. Privacy risks of such systems have previously been studied mostly in the context of recovery of personal information in the form of usage records from the training data. However, the user representations themselves may be used together with external data to recover private user information such as gender and age. In this paper we show that user vectors calculated by a common recommender system can be exploited in this way. We propose the privacy-adversarial framework to eliminate such leakage of private information, and study the trade-off between recommender performance and leakage both theoretically and empirically using a benchmark dataset. An advantage of the proposed method is that it also helps guarantee fairness of results, since all implicit knowledge of a set of attributes is scrubbed from the representations used by the model, and thus can't enter into the decision making. We discuss further applications of this method towards the generation of deeper and more insightful recommendations.