Skip to Main content Skip to Navigation

L 1-norm double backpropagation adversarial defense

Ismaïla Seck 1, 2 Gaëlle Loosli 1 Stephane Canu 3
1 LIMOS - Laboratoire d'Informatique, de Modélisation et d'Optimisation des Systèmes
2 LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
3 DocApp - LITIS - Equipe Apprentissage
LITIS - Laboratoire d'Informatique, de Traitement de l'Information et des Systèmes
Abstract : Adversarial examples are a challenging open problem for deep neural networks. We propose in this paper to add a penalization term that forces the decision function to be at in some regions of the input space, such that it becomes, at least locally, less sensitive to attacks. Our proposition is theoretically motivated and shows on a rst set of carefully conducted experiments that it behaves as expected when used alone, and seems promising when coupled with adversarial training.
Document type :
Conference papers
Complete list of metadatas

Cited literature [7 references]  Display  Hide  Download
https://hal.uca.fr/hal-02049020
Contributor : Gaelle Loosli <>
Submitted on : Tuesday, March 5, 2019 - 8:44:47 AM
Last modification on : Wednesday, March 4, 2020 - 12:28:03 PM
Document(s) archivé(s) le : Thursday, June 6, 2019 - 12:26:26 PM

Files

Vignette du document
main.pdf
Files produced by the author(s)

Identifiers

  • HAL Id : hal-02049020, version 1
  • ARXIV : 1903.01715

Collections

PRES_CLERMONT | CNRS | LIMOS | LITIS | INSA-ROUEN | COMUE-NORMANDIE | UNIROUEN | UNILEHAVRE | INSA-GROUPE

Citation

Ismaïla Seck, Gaëlle Loosli, Stephane Canu. L 1-norm double backpropagation adversarial defense. ESANN - European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning, Apr 2019, Bruges, France. ⟨hal-02049020⟩

Export

BibTeX TEI DC DCterms EndNote

Share

Metrics

Record views

120

Files downloads

56

Contact

Informations

CCSD