CoChat
Sign Up
Cited by 6Privacy-Preserving Technologies in DataRecommender Systems and TechniquesEthics and Social Impacts of AI

Fairness-aware Differentially Private Collaborative Filtering

Open access
Yang, Zhenhuan, Ge, Yingqiang, Su, Congzhe, Wang, Dingxian, Zhao, Xiaoting, Ying, Yiming·2023·Companion Proceedings of the ACM Web Conference 2023
DOI 10.1145/3543873.3587577
Open this paper in CoChat — free accountOpen PDF

Abstract

Recently, there has been an increasing adoption of differential privacy guided algorithms for privacy-preserving machine learning tasks. However, the use of such algorithms comes with trade-offs in terms of algorithmic fairness, which has been widely acknowledged. Specifically, we have empirically observed that the classical collaborative filtering method, trained by differentially private stochastic gradient descent (DP-SGD), results in a disparate impact on user groups with respect to different user engagement levels. This, in turn, causes the original unfair model to become even more biased against inactive users. To address the above issues, we propose DP-Fair, a two-stage framework for collaborative filtering based algorithms. Specifically, it combines differential privacy mechanisms with fairness constraints to protect user privacy while ensuring fair recommendations. The experimental results, based on Amazon datasets, and user history logs collected from Etsy, one of the largest e-commerce platforms, demonstrate that our proposed method exhibits superior performance in terms of both overall accuracy and user group fairness on both shallow and deep recommendation models compared to vanilla DP-SGD.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1145/3543873.3587577

Cites

3 of 3 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • In a centralized setup, the system is a trusted server responsible for maintaining data privacy.

    “In a centralized setup, the system is a trusted server responsible for maintaining data privacy [99, 123-126].”

    From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [125]

    The Introduction and Section 2.2 describe protecting user data with differential privacy and DP-SGD. They do not describe a centralized setup or identify a trusted server as responsible for maintaining privacy. (LLM, conf=0.75)

    Experimental — accuracy not yet measured

  • To achieve privacy protection, centralized RS typically uses non-cryptographic techniques due to their ease of application and low computational complexity.

    “To achieve privacy protection, centralized RS typically uses non-cryptographic techniques due to their ease of application and low computational complexity [8, 123-125].”

    From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [125]

    The Introduction says differential privacy is a popular way to prevent privacy leakage in collaborative filtering, and that DP-SGD is often adopted for its scalability. It does not say that centralized recommender systems typically favor non-cryptographic techniques because they are easy to apply or have low computational complexity. (LLM, conf=0.83)

    Experimental — accuracy not yet measured

  • Yang et al. [125] proposed a solution that balanced fairness and privacy, albeit at the cost of accuracy.

    “They proposed a solution that balanced fairness and privacy, albeit at the cost of accuracy.”

    From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [125]

    The Abstract and Section 3 describe DP-Fair as combining privacy protection with fairness constraints. Section 4.2.1 says accuracy-related utility declines as privacy constraints tighten, but it also reports that DP-Fair improves overall utility over DP-SGD; the claimed accuracy cost therefore needs a comparator and does not describe a loss relative to DP-SGD. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Yang et al. [125] noted that formal privacy guarantees couldn't ensure fairness, leading to biased recommendations.

    “Yang et al. [125] noted that formal privacy guarantees couldn't ensure fairness, leading to biased recommendations.”

    From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [125]

  • Laplace, Gaussian, and Exponential mechanisms are commonly applied for noise addition to satisfy DP.

    “Usually, Laplace [14, 15, 171, 172, 176, 177], Gaussian [101, 123, 125] & Exponential [126] mechanisms are commonly applied for noise addition to satisfy the DP.”

    From A Deep Dive into Fairness, Bias, Threats, and Privacy in Recommender Systems: Insights and Future Research (2024) · cites this paper as [125]

    Citation [126] is attached specifically to the Exponential mechanism. Section 2.2 describes Gaussian noise addition in DP-SGD, but our copy does not discuss an Exponential mechanism or its use for noise addition (searched: “exponential mechanism” and “exponential noise addition”). (LLM, conf=0.76)

    Experimental — accuracy not yet measured

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 27

  1. Martin Abadi, Andy Chu, Ian Goodfellow, H Brendan McMahan, Ilya Mironov, Kunal Talwar, and Li Zhang. 2016. Deep learning with diferential privacy. In Proceedings of the 2016 ACM SIGSAC conference on computer and communications security. 308-318.
  2. Eugene Bagdasaryan, Omid Poursaeed, and Vitaly Shmatikov. 2019. Diferential privacy has disparate impact on model accuracy. Advances in Neural Information Processing Systems 32 (2019), 15479-15488.
  3. Arnaud Berlioz, Arik Friedman, Mohamed Ali Kaafar, Roksana Boreli, and Shlomo Berkovsky. 2015. Applying diferential privacy to matrix factorization. In Pro- ceedings of the 9th ACM Conference on Recommender Systems. 107-114.
  4. Zhiqi Bu, Jinshuo Dong, Qi Long, and Weijie J Su. 2020. Deep learning with gaussian diferential privacy. Harvard data science review 2020, 23 (2020), 10- 1162.
  5. Cynthia Dwork, Frank McSherry, Kobbi Nissim, and Adam Smith. 2006. Cali- brating noise to sensitivity in private data analysis. In Theory of cryptography conference. Springer, 265-284.
  6. Cynthia Dwork, Aaron Roth, et al. 2014. The algorithmic foundations of difer- ential privacy. Found. Trends Theor. Comput. Sci. 9, 3-4 (2014), 211-407.
  7. Zuohui Fu, Yikun Xian, Ruoyuan Gao, Jieyu Zhao, Qiaoying Huang, Yingqiang Ge, Shuyuan Xu, Shijie Geng, Chirag Shah, Yongfeng Zhang, et al. 2020. Fairness- aware explainable recommendation over knowledge graphs. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. 69-78.
  8. Yingqiang Ge, Shuchang Liu, Zuohui Fu, Juntao Tan, Zelong Li, Shuyuan Xu, Yunqi Li, Yikun Xian, and Yongfeng Zhang. 2022. A survey on trustworthy recommender systems. arXiv preprint arXiv:2207.12515 (2022). · arXiv
  9. Yingqiang Ge, Shuchang Liu, Ruoyuan Gao, Yikun Xian, Yunqi Li, Xiangyu Zhao, Changhua Pei, Fei Sun, Junfeng Ge, Wenwu Ou, and Yongfeng Zhang. 2021. Towards Long-term Fairness in Recommendation. In Proceedings of the 14th ACM International Conference on Web Search and Data Mining. 445-453.
  10. Yingqiang Ge, Shuya Zhao, Honglu Zhou, Changhua Pei, Fei Sun, Wenwu Ou, and Yongfeng Zhang. 2020. Understanding Echo Chambers in E-Commerce Rec- ommender Systems. In Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval (Virtual Event, China) (SI- GIR '20). Association for Computing Machinery, New York, NY, USA, 2261-2270. https://doi.org/10.1145/3397271.3401431 · DOI
  11. Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative fltering. In proceedings of the 25th international conference on world wide web. 507-517.
  12. Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. 2017. Neural collaborative fltering. In Proceedings of the 26th international conference on world wide web. 173-182.
  13. Jingyu Hua, Chang Xia, and Sheng Zhong. 2015. Diferentially Private Matrix Factorization. In IJCAI.
  14. Toshihiro Kamishima, Shotaro Akaho, and Jun Sakuma. 2011. Fairness-aware learning through regularization approach. In 2011 IEEE 11th International Confer- ence on Data Mining Workshops. IEEE, 643-650.
  15. Yunqi Li, Hanxiong Chen, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2021. User-oriented Fairness in Recommendation. In Proceedings of the Web Conference WWW '23 Companion, April 30-May 04, 2023, Austin, TX, USA 2021. 624-632.
  16. Yunqi Li, Yingqiang Ge, and Yongfeng Zhang. 2021. Tutorial on Fairness of Machine Learning in Recommender Systems. In Proceedings of the 30th CIKM.
  17. Shuchang Liu, Fei Sun, Yingqiang Ge, Changhua Pei, and Yongfeng Zhang. 2021. Variation Control and Evaluation for Generative Slate Recommenda- tions. In Proceedings of the Web Conference 2021 (Ljubljana, Slovenia) (WWW '21). Association for Computing Machinery, New York, NY, USA, 436-448. https://doi.org/10.1145/3442381.3449864 · DOI
  18. Ziqi Liu, Yu-Xiang Wang, and Alexander Smola. 2015. Fast diferentially private matrix factorization. In Proceedings of the 9th ACM Conference on Recommender Systems. 171-178.
  19. Kiwan Maeng, Haiyu Lu, Luca Melis, John Nguyen, Mike Rabbat, and Carole-Jean Wu. 2022. Towards fair federated recommendation learning: Characterizing the inter-dependence of system and data heterogeneity. In Proceedings of the 16th ACM Conference on Recommender Systems. 156-167.
  20. H Brendan McMahan, Galen Andrew, Ulfar Erlingsson, Steve Chien, Ilya Mironov, Nicolas Papernot, and Peter Kairouz. 2018. A general approach to adding difer- ential privacy to iterative training procedures. arXiv preprint arXiv:1812.06210 (2018). · arXiv
  21. Stefen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme. 2012. BPR: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618 (2012). · arXiv
  22. Hyejin Shin, Sungwook Kim, Junbum Shin, and Xiaokui Xiao. 2018. Privacy enhanced matrix factorization for recommendation with local diferential privacy. IEEE Transactions on Knowledge and Data Engineering 30, 9 (2018), 1770-1782.
  23. Zhenhuan Yang, Shu Hu, Yunwen Lei, Kush R Vashney, Siwei Lyu, and Yiming Ying. 2022. Diferentially private sgda for minimax problems. In Uncertainty in Artifcial Intelligence. PMLR, 2192-2202.
  24. Zhenhuan Yang, Yan Lok Ko, Kush R Varshney, and Yiming Ying. 2022. Minimax AUC Fairness: Efcient Algorithm with Provable Convergence. arXiv preprint arXiv:2208.10451 (2022). · arXiv
  25. Zhenhuan Yang, Yunwen Lei, Siwei Lyu, and Yiming Ying. 2021. Stability and diferential privacy of stochastic gradient descent for pairwise learning with non- smooth loss. In International Conference on Artifcial Intelligence and Statistics. PMLR, 2026-2034.
  26. Meike Zehlike, Ke Yang, and Julia Stoyanovich. 2021. Fairness in ranking: A survey. arXiv preprint arXiv:2103.14000 (2021). · arXiv
  27. Zhitao Zhu, Shijing Si, Jianzong Wang, and Jing Xiao. 2022. Cali3F: Calibrated Fast Fair Federated Recommendation System. In 2022 International Joint Conference on Neural Networks (IJCNN). IEEE, 1-8.

Cited by

Sorted by citations

Authors
Yang, Z., Ge, Y., Su, C., Wang, D., Zhao, X., & Ying, Y.
Affiliations
  • Etsy (United States) (US)
  • Rutgers, The State University of New Jersey (US)
  • University at Albany, State University of New York (US)
ORCID
Venue
Companion Proceedings of the ACM Web Conference 2023
Publisher
ACM
Published
2023-04-30
Pages
927-931
DOI
10.1145/3543873.3587577
OpenAlex
W4327811240
Cited by
6
References
3
Citations by year
2026: 2 · 2025: 1 · 2024: 2 · 2023: 1
FWCI
1.06
Type
conference paper
Language
en
Open access
Licence
www.acm.org/publications/policies/copyright_policy#Background (vor)
Locations
  • International World Wide Web Conference (WWW) (publishedVersion)
  • arXiv (Cornell University) (submittedVersion, cc-by)
  • arXiv (Cornell University) (cc-by)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › Privacy-Preserving Technologies in Data
  • Physical Sciences › Computer Science › Information Systems › Recommender Systems and Techniques
  • Social Sciences › Safety Research › Ethics and Social Impacts of AI
Keywords
differential privacy, collaborative filtering, algorithmic fairness, DP-SGD, fairness constraints, user engagement, recommender systems, privacy-preserving machine learning

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review