Academic Journal
Differentially Private Gradient Flow based on the Sliced Wasserstein Distance
| العنوان: | Differentially Private Gradient Flow based on the Sliced Wasserstein Distance |
|---|---|
| المؤلفون: | Sebag, Ilana, Sreenivas, Muni, Franceschi, Jean-Yves, Rakotomamonjy, Alain, Gartrell, Mike, Atif, Jamal, Allauzen, Alexandre |
| المساهمون: | Criteo AI Lab, Criteo Paris, Machine Intelligence and Learning Systems (MILES), Laboratoire d'analyse et modélisation de systèmes pour l'aide à la décision (LAMSADE), Université Paris Dauphine-PSL, Université Paris Sciences et Lettres (PSL)-Université Paris Sciences et Lettres (PSL)-Centre National de la Recherche Scientifique (CNRS)-Université Paris Dauphine-PSL, Université Paris Sciences et Lettres (PSL)-Université Paris Sciences et Lettres (PSL)-Centre National de la Recherche Scientifique (CNRS), Sigma Nova, Ecole Superieure de Physique et de Chimie Industrielles de la Ville de Paris (ESPCI Paris), Université Paris Sciences et Lettres (PSL) |
| المصدر: | ISSN: 2835-8856 ; Transactions on Machine Learning Research Journal ; https://hal.science/hal-04664174 ; Transactions on Machine Learning Research Journal, 2025. |
| بيانات النشر: | CCSD OpenReview.net, 2022 |
| سنة النشر: | 2025 |
| المجموعة: | Université Paris-Dauphine: HAL |
| مصطلحات موضوعية: | Machine leaning, Differential privacy, Generative AI, Privacy, [STAT.ML]Statistics [stat]/Machine Learning [stat.ML] |
| الوصف: | International audience ; Safeguarding privacy in sensitive training data is paramount, particularly in the context of generative modeling. This can be achieved through either differentially private stochastic gradient descent or a differentially private metric for training models or generators. In this paper, we introduce a novel differentially private generative modeling approach based on a gradient flow in the space of probability measures. To this end, we define the gradient flow of the Gaussian-smoothed Sliced Wasserstein Distance, including the associated stochastic differential equation (SDE). By discretizing and defining a numerical scheme for solving this SDE, we demonstrate the link between smoothing and differential privacy based on a Gaussian mechanism, due to a specific form of the SDE's drift term. We then analyze the differential privacy guarantee of our gradient flow, which accounts for both the smoothing and the Wiener process introduced by the SDE itself. Experiments show that our proposed model can generate higher-fidelity data at a low privacy budget compared to a generator-based model, offering a promising alternative. |
| نوع الوثيقة: | article in journal/newspaper |
| اللغة: | English |
| الاتاحة: | https://hal.science/hal-04664174 https://hal.science/hal-04664174v2/document https://hal.science/hal-04664174v2/file/DP_SW_GF%20%286%29.pdf |
| Rights: | info:eu-repo/semantics/OpenAccess |
| رقم الانضمام: | edsbas.AA0F1833 |
| قاعدة البيانات: | BASE |
| الوصف غير متاح. |