Fool SHAP with Stealthily Biased Sampling - Archive ouverte HAL
Communication Dans Un Congrès Année : 2023

Fool SHAP with Stealthily Biased Sampling

Gabriel Laberge
Ulrich Aïvodji
  • Fonction : Auteur
  • PersonId : 1128931
Satoshi Hara
  • Fonction : Auteur
  • PersonId : 1291329
Mario Marchand
  • Fonction : Auteur
  • PersonId : 1291330
Foutse Khomh
  • Fonction : Auteur
  • PersonId : 1291331

Résumé

SHAP explanations aim at identifying which features contribute the most to the difference in model prediction at a specific input versus a background distribution. Recent studies have shown that they can be manipulated by malicious adversaries to produce arbitrary desired explanations. However, existing attacks focus solely on altering the black-box model itself. In this paper, we propose a complementary family of attacks that leave the model intact and manipulate SHAP explanations using stealthily biased sampling of the data points used to approximate expectations w.r.t the background distribution. In the context of fairness audit, we show that our attack can reduce the importance of a sensitive feature when explaining the difference in outcomes between groups while remaining undetected. More precisely, experiments performed on real-world datasets showed that our attack could yield up to a 90% relative decrease in amplitude of the sensitive feature attribution. These results highlight the manipulability of SHAP explanations and encourage auditors to treat them with skepticism.
Fichier principal
Vignette du fichier
Stealth_Fool_SHAP.pdf (1.04 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04232192 , version 1 (07-10-2023)

Identifiants

  • HAL Id : hal-04232192 , version 1

Citer

Gabriel Laberge, Ulrich Aïvodji, Satoshi Hara, Mario Marchand, Foutse Khomh. Fool SHAP with Stealthily Biased Sampling. International Conference on Learning Representations (ICLR), May 2023, Kigali, Rwanda. ⟨hal-04232192⟩
83 Consultations
66 Téléchargements

Partager

More