WaveAttack: Asymmetric Frequency Obfuscation-based Backdoor Attacks Against Deep Neural Networks

TL;DR

WaveAttack introduces a novel frequency-based backdoor attack leveraging wavelet transforms and asymmetric obfuscation to improve stealthiness and effectiveness against deep neural networks, outperforming existing methods in image fidelity.

Contribution

This paper presents WaveAttack, a new backdoor attack method using frequency domain features and asymmetric obfuscation to enhance stealth and attack success.

Findings

Achieves higher stealthiness and effectiveness than state-of-the-art methods.

Improves image fidelity metrics significantly, with up to 28.27% PSNR increase.

Reduces Inception Score (IS) by 70.59%.

Abstract

Due to the popularity of Artificial Intelligence (AI) technology, numerous backdoor attacks are designed by adversaries to mislead deep neural network predictions by manipulating training samples and training processes. Although backdoor attacks are effective in various real scenarios, they still suffer from the problems of both low fidelity of poisoned samples and non-negligible transfer in latent space, which make them easily detectable by existing backdoor detection algorithms. To overcome the weakness, this paper proposes a novel frequency-based backdoor attack method named WaveAttack, which obtains image high-frequency features through Discrete Wavelet Transform (DWT) to generate backdoor triggers. Furthermore, we introduce an asymmetric frequency obfuscation method, which can add an adaptive residual in the training and inference stage to improve the impact of triggers and further enhance the effectiveness of WaveAttack. Comprehensive experimental results show that WaveAttack not only achieves higher stealthiness and effectiveness, but also outperforms state-of-the-art (SOTA) backdoor attack methods in the fidelity of images by up to 28.27\% improvement in PSNR, 1.61\% improvement in SSIM, and 70.59\% reduction in IS.