!!Projects per year
Abstrakti
Uncertainty-aware user modeling is crucial for de-signing AI systems that adapt to users in real-time while addressing privacy concerns. This paper pro-poses a novel framework for privacy-preserving probabilistic user modeling that integrates un-certainty quantification and differential privacy (DP). Building on neural processes (NPs), a scalable latent variable probabilistic model, we enable meta-learning for user behaviour prediction under privacy constraints. By employing differentially private stochastic gradient descent (DP-SGD), our method achieves rigorous privacy guarantees while preserving predictive accuracy. Unlike prior work, which primarily addresses privacy-preserving learning for convex or smooth functions, we establish theoretical guarantees for non-convex objectives, focusing on the utility-privacy trade-offs inherent in uncertainty-aware models. Through extensive experiments, we demonstrate that our approach achieves competitive accuracy under stringent privacy budgets. Our results showcase the potential of privacy-preserving probabilistic user models to enable trustworthy AI systems in real-world interactive applications.
| Alkuperäiskieli | Englanti |
|---|---|
| Sivut | 3979-3989 |
| Sivumäärä | 11 |
| Julkaisu | Proceedings of Machine Learning Research |
| Vuosikerta | 286 |
| Tila | Julkaistu - 2025 |
| OKM-julkaisutyyppi | A4 Artikkeli konferenssijulkaisussa |
| Tapahtuma | Conference on Uncertainty in Artificial Intelligence - Rio de Janeiro, Brasilia Kesto: 21 heinäk. 2025 → 25 heinäk. 2025 Konferenssinumero: 41 |
Rahoitus
This work was supported by the Research Council of Finland Flagship programme: Finnish Center for Artificial Intelligence FCAI and decisions 358958, 359567. Amir Sonee, Haripriya Harikumar, and Samuel Kaski were supported by the UKRI Turing AI World-Leading Researcher Fellowship, [EP/W002973/1].
Sormenjälki
Sukella tutkimusaiheisiin 'Privacy-Preserving Neural Processes for Probabilistic User Modeling'. Ne muodostavat yhdessä ainutlaatuisen sormenjäljen.-
PRIMUS: Secure Use of Data in Health
Kaski, S. (Vastuullinen johtaja), Ulukir, B. (Projektin jäsen), Hedman, A. (Projektin jäsen), Zhu, L. (Projektin jäsen), Mäkinen, L. (Projektin jäsen), Garg, N. (Projektin jäsen), Loría, J. (Projektin jäsen) & Ou, Y. (Projektin jäsen)
01/01/2024 → 31/12/2026
Projekti: RCF Academy Project targeted call
-
AIM4REAL: AIM4REAL - Artificial Intelligence for Personalised Medicine for Real
Kaski, S. (Vastuullinen johtaja), Kortia, A. (Projektin jäsen), Loría, J. (Projektin jäsen), Garg, N. (Projektin jäsen), Türkseven, D. (Projektin jäsen) & Mäkinen, L. (Projektin jäsen)
01/01/2024 → 31/12/2025
Projekti: RCF Academy Project targeted call
-
-: Finnish Center for Artificial Intelligence
Kaski, S. (Vastuullinen johtaja)
01/01/2019 → 31/12/2022
Projekti: Academy of Finland: Other research funding
Siteeraa tätä
- APA
- Author
- BIBTEX
- Harvard
- Standard
- RIS
- Vancouver