EAD: Elastic-Net Attacks to Deep Neural Networks via Adversarial Examples

TL;DR

This paper introduces EAD, a new method for generating adversarial examples for deep neural networks using elastic-net regularization, emphasizing $L_1$ distortion to produce sparse, effective attacks with improved transferability.

Contribution

The paper formulates adversarial attacks as an elastic-net regularized optimization problem, incorporating $L_1$ distortion, and demonstrates its effectiveness and advantages over existing methods.

Findings

EAD produces adversarial examples with small $L_1$ distortion.

EAD achieves attack performance comparable to state-of-the-art methods.

EAD improves transferability and complements adversarial training.

Abstract

Recent studies have highlighted the vulnerability of deep neural networks (DNNs) to adversarial examples - a visually indistinguishable adversarial image can easily be crafted to cause a well-trained model to misclassify. Existing methods for crafting adversarial examples are based on $L_2$ and $L_\infty$ distortion metrics. However, despite the fact that $L_1$ distortion accounts for the total variation and encourages sparsity in the perturbation, little has been developed for crafting $L_1$-based adversarial examples. In this paper, we formulate the process of attacking DNNs via adversarial examples as an elastic-net regularized optimization problem. Our elastic-net attacks to DNNs (EAD) feature $L_1$-oriented adversarial examples and include the state-of-the-art $L_2$ attack as a special case. Experimental results on MNIST, CIFAR10 and ImageNet show that EAD can yield a distinct set of adversarial examples with small $L_1$ distortion and attains similar attack performance to the state-of-the-art methods in different attack scenarios. More importantly, EAD leads to improved attack transferability and complements adversarial training for DNNs, suggesting novel insights on leveraging $L_1$ distortion in adversarial machine learning and security implications of DNNs.