Probabilistic Jacobian-Based Saliency Maps Attacks

Probabilistic Jacobian-Based Saliency Maps Attacks

Neural network classifiers (NNCs) are known to be vulnerable to malicious adversarial perturbations of inputs including those modifying a small fraction of the input features named sparse or <inline-formula><math display="inline"><semantics><msub><mi>L</mi&...

Full description

Bibliographic Details
Main Authors: Théo Combey, António Loison, Maxime Faucher, Hatem Hajri
Format: Article
Language:English
Published: MDPI AG 2020-11-01
Series:Machine Learning and Knowledge Extraction
Subjects:
Jacobian-based Saliency Map
adversarial attacks
deep neural network classifiers
MNIST
CIFAR-10
GTSRB
Online Access:https://www.mdpi.com/2504-4990/2/4/30

Internet

https://www.mdpi.com/2504-4990/2/4/30

Similar Items