Data-Free Adversarial Perturbations for Practical Black-Box Attack

TL;DR

This paper introduces a novel data-free approach for generating adversarial perturbations that can effectively fool black-box neural network models without requiring access to training data, highlighting persistent vulnerabilities.

Contribution

The authors propose a data-free adversarial attack method that outperforms existing universal perturbation techniques in black-box scenarios without training data.

Findings

High fooling rates on target models

Outperforms other universal adversarial perturbation methods

Models remain vulnerable without training data access

Abstract

Neural networks are vulnerable to adversarial examples, which are malicious inputs crafted to fool pre-trained models. Adversarial examples often exhibit black-box attacking transferability, which allows that adversarial examples crafted for one model can fool another model. However, existing black-box attack methods require samples from the training data distribution to improve the transferability of adversarial examples across different models. Because of the data dependence, the fooling ability of adversarial perturbations is only applicable when training data are accessible. In this paper, we present a data-free method for crafting adversarial perturbations that can fool a target model without any knowledge about the training data distribution. In the practical setting of a black-box attack scenario where attackers do not have access to target models and training data, our method achieves high fooling rates on target models and outperforms other universal adversarial perturbation methods. Our method empirically shows that current deep learning models are still at risk even when the attackers do not have access to training data.