Smooth adversarial examples

Autor: Hanwei Zhang, Yannis Avrithis, Teddy Furon, Laurent Amsaleg

Publikováno v: EURASIP Journal on Information Security, Vol 2020, Iss 1, Pp 1-12 (2020)

Abstract This paper investigates the visual quality of the adversarial examples. Recent papers propose to smooth the perturbations to get rid of high frequency artifacts. In this work, smoothing has a different meaning as it perceptually shapes the p

Externí odkaz: https://doaj.org/article/a57438bb77614f47b5968f003709067f

Patch Replacement: A Transformation-based Method to Improve Robustness against Adversarial Attacks

Autor: Teddy Furon, Hanwei Zhang, Yannis Avrithis, Laurent Amsaleg

Publikováno v: ACM Multimedia, Trustworthy AI Workshop
Trustworthy AI 2021-1st International Workshop on Trustworthy AI for Multimedia Computing
Trustworthy AI 2021-1st International Workshop on Trustworthy AI for Multimedia Computing, Oct 2021, Virtual, China. pp.1-10, ⟨10.1145/3475731.3484955⟩
Trustworthy AI @ ACM Multimedia

International audience; Deep Neural Networks (DNNs) are robust against intra-class variability of images, pose variations and random noise, but vulnerable to imperceptible adversarial perturbations that are well-crafted precisely to mislead. While ra

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::233a0cc4eb4ee96b470421820742f8f4
https://hal.archives-ouvertes.fr/hal-03363999

Attaques et défenses de réseaux de neurones profonds : le cas de la classification d’images

Autor: Hanwei ZHANG, Teddy FURON, Laurent AMSALEG, Yannis AVRITHIS

L’apprentissage automatique utilisant des réseaux neuronaux profonds appliqués à la reconnaissance d’images fonctionne extrêmement bien. Néanmoins, il est possible de modifier intentionnellement et très légèrement les images, modification

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_________::aa88c7beb17055033bad2264e63a21c8
https://doi.org/10.51926/iste.9026.ch2

Asymmetric metric learning for knowledge transfer

Autor: Mateusz Budnik, Yannis Avrithis

Publikováno v: CVPR 2021-IEEE/CVF Conference on Computer Vision and Pattern Recognition
CVPR 2021-IEEE/CVF Conference on Computer Vision and Pattern Recognition, Jun 2021, Virtual, France. pp.8224-8234, ⟨10.1109/CVPR46437.2021.00813⟩
CVPR

International audience; Knowledge transfer from large teacher models to smaller student models has recently been studied for metric learning, focusing on fine-grained classification. In this work, focusing on instance-level image retrieval, we study

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::d350298c50cd619bd49ed09e14591266
https://inria.hal.science/hal-03529901

Local Propagation for Few-Shot Learning

Autor: Yannis Avrithis, Yann Lifchitz, Sylvaine Picard

Publikováno v: ICPR
ICPR 2020-25th International Conference on Pattern Recognition
ICPR 2020-25th International Conference on Pattern Recognition, Jan 2021, Virtual, Italy. pp.1-8
ICPR 2020-25th International Conference on Pattern Recognition, Jan 2021, Virtual, Italy

The challenge in few-shot learning is that available data is not enough to capture the underlying distribution. To mitigate this, two emerging directions are (a) using local image representations, essentially multiplying the amount of data by a const

Externí odkaz: https://explore.openaire.eu/search/publication?articleId=doi_dedup___::c5bcf9bf42ccf5c9527383f78737c4ba
https://doi.org/10.1109/icpr48806.2021.9412178