Recent work has proposed novel data augmentation methods to improve the adversarial robustness of deep neural networks. In this paper, we re-evaluate such methods through the lens of different metrics that characterize the augmented manifold, finding contradictory evidence. Our extensive empirical analysis involving 5 data augmentation methods, all tested with an increasing probability of augmentation, shows that: (i) novel data augmentation methods proposed to improve adversarial robustness only improve it when combined with classical augmentations (like image flipping and rotation), and even worsen adversarial robustness if used in isolation; and (ii) adversarial robustness is significantly affected by the augmentation probability, conversely to what is claimed in recent work. We conclude by discussing how to rethink the development and evaluation of novel data augmentation methods for adversarial robustness. Our open-source code is available at https://github.com/eghbalz/rethink_da_for_ar

Rethinking data augmentation for adversarial robustness

Pintor, Maura;Biggio, Battista;
2024-01-01

Abstract

Recent work has proposed novel data augmentation methods to improve the adversarial robustness of deep neural networks. In this paper, we re-evaluate such methods through the lens of different metrics that characterize the augmented manifold, finding contradictory evidence. Our extensive empirical analysis involving 5 data augmentation methods, all tested with an increasing probability of augmentation, shows that: (i) novel data augmentation methods proposed to improve adversarial robustness only improve it when combined with classical augmentations (like image flipping and rotation), and even worsen adversarial robustness if used in isolation; and (ii) adversarial robustness is significantly affected by the augmentation probability, conversely to what is claimed in recent work. We conclude by discussing how to rethink the development and evaluation of novel data augmentation methods for adversarial robustness. Our open-source code is available at https://github.com/eghbalz/rethink_da_for_ar
File in questo prodotto:
File Dimensione Formato  
1-s2.0-S0020025523014238-main.pdf

Solo gestori archivio

Tipologia: versione editoriale
Dimensione 1.74 MB
Formato Adobe PDF
1.74 MB Adobe PDF   Visualizza/Apri   Richiedi una copia
_INS__Rethinking_Data_Augmentation-3.pdf

accesso aperto

Tipologia: versione pre-print
Dimensione 1.55 MB
Formato Adobe PDF
1.55 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: https://hdl.handle.net/11584/382423
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact