Skip to main navigation Skip to search Skip to main content

Meta-Learning and Targeted Differential Privacy to Improve the Accuracy-Privacy Trade-off in Recommendations

Research output: Chapter in Book/Report/Conference proceedingConference paperpeer-review

Abstract

Balancing differential privacy (DP) with recommendation accuracy is a key challenge in privacy-preserving recommender systems, since DP-noise degrades accuracy. We address this trade-off at both the data and model levels. At the data level, we apply DP only to the most stereotypical user data likely to reveal sensitive attributes, such as gender or age, to reduce unnecessary perturbation; we refer to this as targeted DP. At the model level, we use meta-learning to improve robustness to remaining DP-noise. This achieves a better trade-off between accuracy and privacy than standard approaches: Meta-learning improves accuracy and targeted DP leads to lower empirical privacy risk compared to uniformly applied DP and full DP baselines. Overall, our findings show that selectively applying DP at the data level together with meta-learning at the model level can effectively balance recommendation accuracy and user privacy.

Original languageEnglish
Title of host publicationUMAP 2026 - Proceedings of the 34th ACM International Conference on User Modeling, Adaptation and Personalization
PublisherAssociation for Computing Machinery (ACM)
Pages517-520
Number of pages4
ISBN (Electronic)9798400723117
DOIs
Publication statusPublished - 7 Jun 2026
Event34th ACM International Conference on User Modeling, Adaptation and Personalization, UMAP 2026 - Gothenburg, Sweden
Duration: 8 Jun 202611 Jun 2026

Publication series

NameUMAP 2026 - Proceedings of the 34th ACM International Conference on User Modeling, Adaptation and Personalization

Conference

Conference34th ACM International Conference on User Modeling, Adaptation and Personalization, UMAP 2026
Country/TerritorySweden
CityGothenburg
Period8/06/2611/06/26

Keywords

  • Accuracy-Privacy Trade-off
  • Differential Privacy
  • Meta-Learning
  • Recommender Systems
  • Stereotypical User Preferences

ASJC Scopus subject areas

  • Artificial Intelligence
  • Human-Computer Interaction
  • Safety, Risk, Reliability and Quality
  • Media Technology
  • Modelling and Simulation

Fields of Expertise

  • Information, Communication & Computing

Cite this