Semi-Universal Adversarial Perturbations - Archive ouverte HAL Accéder directement au contenu
Proceedings/Recueil Des Communications Année : 2023

Semi-Universal Adversarial Perturbations

Jordan Frecon
  • Fonction : Auteur correspondant
  • PersonId : 1090292
  • IdRef : 196503922

Connectez-vous pour contacter l'auteur
Paul Viallard
Emilie Morvant
Gilles Gasso
Amaury Habrard

Résumé

This paper introduces semi-universal perturbations that bridge the gap between specific and universal adversarial perturbations. The original idea is to craft a specific perturbation by choosing it among a set of $L$ universal perturbations. We propose to jointly learn the perturbations of this set to maximize the chances to attack each example by allowing it to choose its own perturbation. To do so, we derive an algorithm, with convergence guarantees under Lipschitz continuity assumptions. Semi-universal perturbations offer a better flexibility, interpretability and diversity, confirmed by our experiments. Additionally, we provide a generalization bound on the abilities of the perturbations to attack new examples.

Mots clés

Fichier principal
Vignette du fichier
2022_Frecon_J_hal_suap.pdf (550.66 Ko) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03615461 , version 1 (21-03-2022)
hal-03615461 , version 2 (07-06-2023)
hal-03615461 , version 3 (07-06-2024)

Identifiants

  • HAL Id : hal-03615461 , version 1

Citer

Jordan Frecon, Paul Viallard, Emilie Morvant, Gilles Gasso, Amaury Habrard, et al.. Semi-Universal Adversarial Perturbations. 2023. ⟨hal-03615461v1⟩
133 Consultations
174 Téléchargements

Partager

Gmail Mastodon Facebook X LinkedIn More