DP-SGD with weight clipping - CRISTAL-MAGNET Accéder directement au contenu
Communication Dans Un Congrès Année : 2023

DP-SGD with weight clipping

Résumé

Recently, due to the popularity of deep neural networks and other methods whose training typically relies on the optimization of an objective function, and due to concerns for data privacy, there is a lot of interest in differentially private gradient descent methods. To achieve differential privacy guarantees with a minimum amount of noise, it is important to be able to bound precisely the sensitivity of the information which the participants will observe. In this study, we present a novel approach that mitigates the bias arising from traditional gradient clipping. By leveraging a public upper bound of the Lipschitz value of the current model and its current location within the search domain, we can achieve refined noise level adjustments. We present a new algorithm with improved differential privacy guarantees and a systematic empirical evaluation, showing that our new approach outperforms existing approaches also in practice.
Fichier principal
Vignette du fichier
main.pdf (661.61 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04614505 , version 1 (17-06-2024)

Identifiants

Citer

Antoine Barczewski, Jan Ramon. DP-SGD with weight clipping. CAp (Conférence sur l'Apprentissage automatique) 2024, SSFAM (Société Savante Française d'Apprentissage Machine); AFRIF (Association Française pour la Reconnaissance et l'Interprétation des Formes), Jul 2024, Lille (France), France. ⟨10.48550/arXiv.2310.18001⟩. ⟨hal-04614505⟩
0 Consultations
0 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More