Diffusion Models for Imperceptible and Transferable Adversarial Attack
Top Cited Papers
- 15 October 2024
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 47 (2) , 961-977
Abstract
Many existing adversarial attacks generate $L_{p}$ -norm perturbations on image RGB space. Despite some achievements in transferability and attack success rate, the crafted adversarial examples are easily perceived by human eyes. Towards visual imperceptibility, some recent works explore unrestricted attacks without $L_{p}$ -norm constraints, yet lacking transferability of attacking black-box models. In this work, we propose a novel imperceptible and transferable attack by leveraging both the generative and discriminative power of diffusion models. Specifically, instead of direct manipulation in pixel space, we craft perturbations in the latent space of diffusion models. Combined with well-designed content-preserving structures, we can generate human-insensitive perturbations embedded with semantic clues. For better transferability, we further “deceive” the diffusion model which can be viewed as an implicit recognition surrogate, by distracting its attention away from the target regions. To our knowledge, our proposed method, DiffAttack, is the first that introduces diffusion models into the adversarial attack field. Extensive experiments conducted across diverse model architectures (CNNs, Transformers, and MLPs), datasets (ImageNet, CUB-200, and Standford Cars), and defense mechanisms underscore the superiority of our attack over existing methods such as iterative attacks, GAN-based attacks, and ensemble attacks. Furthermore, we provide a comprehensive discussion on future research avenues in diffusion-based adversarial attacks, aiming to chart a course for this burgeoning field.
Keywords
All Related Versions
Funding Information
- National Natural Science Foundation of China (62125102, 623B2013)
- National Key Research and Development Program of China (2022ZD0160401)
- Beijing Natural Science Foundation (JL23005)
- Fundamental Research Funds for the Central Universities
This publication has 45 references indexed in Scilit:
- Dense reinforcement learning for safety validation of autonomous vehiclesNature, 2023
- TFormer: A throughout fusion transformer for multi-modal skin lesion diagnosisComputers in Biology and Medicine, 2023
- Frequency Domain Model Augmentation for Adversarial Attack2022
- Dermoscopic image retrieval based on rotation-invariance deep hashingMedical Image Analysis, 2022
- SemanticAdv: Generating Adversarial Examples via Attribute-Conditioned Image Editing2020
- Patch-Wise Attack for Fooling Deep Neural Network2020
- Transferable Adversarial Perturbations2018
- Adversarial Attacks and Defences Competition2018
- Rethinking the Inception Architecture for Computer Vision2016
- Matching Local Self-Similarities across Images and Videos2007