Towards Imperceptible and Robust Adversarial Example Attacks against Neural Networks

TL;DR

This paper introduces a new adversarial attack method that produces imperceptible and robust examples by considering human perception and noise tolerance, challenging the security of neural networks.

Contribution

It proposes a novel adversarial attack technique that accounts for human perceptual metrics and enhances robustness against physical-world noise.

Findings

Effective in generating imperceptible adversarial examples

Demonstrates increased robustness in physical environments

Outperforms previous attack methods in stealth and durability

Abstract

Machine learning systems based on deep neural networks, being able to produce state-of-the-art results on various perception tasks, have gained mainstream adoption in many applications. However, they are shown to be vulnerable to adversarial example attack, which generates malicious output by adding slight perturbations to the input. Previous adversarial example crafting methods, however, use simple metrics to evaluate the distances between the original examples and the adversarial ones, which could be easily detected by human eyes. In addition, these attacks are often not robust due to the inevitable noises and deviation in the physical world. In this work, we present a new adversarial example attack crafting method, which takes the human perceptual system into consideration and maximizes the noise tolerance of the crafted adversarial example. Experimental results demonstrate the efficacy of the proposed technique.