Quantifying fairness of federated learning LPPM models - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Quantifying fairness of federated learning LPPM models

Résumé

Despite the great potential offered by Artificial Intelligence in the context of smart mobility, it comes with the greater challenge of preserving the privacy of users. Federated Learning (FL) has gained popularity as a privacy-friendly approach, however, an equally important aspect rarely addressed in the literature, is its fairness. In this work we audit a FL-based privacy-preserving model. We use Entropy to determine similarity within the system's input data and compare its value against that of the output to detect unfair treatment.
Fichier principal
Vignette du fichier
Assessing_the_fairness_of_EDEN (1).pdf (468.08 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-03703623 , version 1 (24-06-2022)

Identifiants

Citer

Amina Ben Salem, Besma Khalfoun, Sonia Ben Mokhtar, Afra Mashhadi. Quantifying fairness of federated learning LPPM models. MobiSys '22: The 20th Annual International Conference on Mobile Systems, Applications and Services, Jun 2022, Portland, France. pp.569-570, ⟨10.1145/3498361.3538788⟩. ⟨hal-03703623⟩
45 Consultations
146 Téléchargements

Altmetric

Partager

Gmail Facebook X LinkedIn More