Maximal Jacobian-based Saliency Map Attack

Abstract

The Jacobian-based Saliency Map Attack is a family of adversarial attack methods for fooling classification models, such as deep neural networks for image classification tasks. By saturating a few pixels in a given image to their maximum or minimum values, JSMA can cause the model to misclassify the resulting adversarial image as a specified erroneous target class. We propose two variants of JSMA, one which removes the requirement to specify a target class, and another that additionally does not need to specify whether to only increase or decrease pixel intensities. Our experiments highlight the competitive speeds and qualities of these variants when applied to datasets of hand-written digits and natural scenes.

arXiv commentExtended version of extended abstract for MAIS 2018

References in corpus (5)

• Explaining and Harnessing Adversarial Examples
• Fashion-MNIST: a Novel Image Dataset for Benchmarking Machine Learning Algorithms
• Intriguing properties of neural networks
• Ensemble Adversarial Training: Attacks and Defenses
• Defensive Distillation is Not Robust to Adversarial Examples

Cited by in corpus (9)

• Launching Adversarial Attacks against Network Intrusion Detection Systems for IoT
• Towards Understanding and Boosting Adversarial Transferability from a Distribution Perspective
• Detecting Adversarial Examples by Input Transformations, Defense Perturbations, and Voting
• Get your Foes Fooled: Proximal Gradient Split Learning for Defense against Model Inversion Attacks on IoMT data
• Breaking certified defenses: Semantic adversarial examples with spoofed robustness certificates
• Generating Image Adversarial Examples by Embedding Digital Watermarks
• Sparse Adversarial Video Attacks with Spatial Transformations
• Class-Aware Robust Adversarial Training for Object Detection
• Fast Approximate Spectral Normalization for Robust Deep Neural Networks