Home > Research > Publications & Outputs > Concealed Backdoor Attack on Diffusion Models f...

Research

Electronic data

  • conceal

    Accepted author manuscript, 921 KB, PDF document

    Available under license: CC BY: Creative Commons Attribution 4.0 International License

Links

Text available via DOI:

Keywords

View graph of relations

Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise

Research output: Contribution to Journal/Magazine › Journal article › peer-review

E-pub ahead of print

Standard

Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise. / Li, J.; Tan, Y.-A.; Fan, S. et al.
In: IEEE Transactions on Consumer Electronics, 11.12.2024.

Research output: Contribution to Journal/Magazine › Journal article › peer-review

Harvard

Li, J, Tan, Y-A, Fan, S, Li, F, Liu, X, Liu, R, Li, Y & Meng, W 2024, 'Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise', IEEE Transactions on Consumer Electronics. https://doi.org/10.1109/TCE.2024.3516352

APA

Li, J., Tan, Y.-A., Fan, S., Li, F., Liu, X., Liu, R., Li, Y., & Meng, W. (2024). Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise. IEEE Transactions on Consumer Electronics. Advance online publication. https://doi.org/10.1109/TCE.2024.3516352

Vancouver

Li J, Tan YA, Fan S, Li F, Liu X, Liu R et al. Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise. IEEE Transactions on Consumer Electronics. 2024 Dec 11. Epub 2024 Dec 11. doi: 10.1109/TCE.2024.3516352

Author

Li, J. ; Tan, Y.-A. ; Fan, S. et al. / Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise. In: IEEE Transactions on Consumer Electronics. 2024.

Bibtex

@article{4ab1d549dea041f8aa4ffa82b4996927,
title = "Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise",
abstract = "Edge AI-driven diffusion models (DMs) are increasingly integrated into consumer devices for high-quality data generation and content creation. This paper introduces InvisibleDiffusion, a novel backdoor attack framework for diffusion models in consumer electronics, designed to remain undetected by utilizing a non-standard Gaussian distribution as a concealed trigger. Unlike previous backdoor methods, InvisibleDiffusion does not rely on obvious visual triggers, enhancing its stealthiness. Extensive experiments demonstrate that InvisibleDiffusion achieves high attack efficacy against DDPM and DDIM models on CIFAR-10 and CelebA datasets, while maintaining the functional integrity of the models. Our code is available for reproducibility at https://anonymous.4open.science/r/b2hoaWNhbnRzZWV0aGF0bm9vb29vb29vb29v. ",
keywords = "Consumer devices, Edge AI, Generative artificial intelligence, Security in deep learning, Gaussian distribution, Gaussian noise (electronic), Backdoors, Diffusion model, Driven diffusion, Gaussians, High quality data, Smart devices",
author = "J. Li and Y.-A. Tan and S. Fan and F. Li and X. Liu and R. Liu and Y. Li and W. Meng",
year = "2024",
month = dec,
day = "11",
doi = "10.1109/TCE.2024.3516352",
language = "English",
journal = "IEEE Transactions on Consumer Electronics",
issn = "0098-3063",
publisher = "Institute of Electrical and Electronics Engineers Inc.",

}

RIS

TY - JOUR

T1 - Concealed Backdoor Attack on Diffusion Models for Smart Devices with Non-standard Gaussian Distribution Noise

AU - Li, J.

AU - Tan, Y.-A.

AU - Fan, S.

AU - Li, F.

AU - Liu, X.

AU - Liu, R.

AU - Li, Y.

AU - Meng, W.

PY - 2024/12/11

Y1 - 2024/12/11

N2 - Edge AI-driven diffusion models (DMs) are increasingly integrated into consumer devices for high-quality data generation and content creation. This paper introduces InvisibleDiffusion, a novel backdoor attack framework for diffusion models in consumer electronics, designed to remain undetected by utilizing a non-standard Gaussian distribution as a concealed trigger. Unlike previous backdoor methods, InvisibleDiffusion does not rely on obvious visual triggers, enhancing its stealthiness. Extensive experiments demonstrate that InvisibleDiffusion achieves high attack efficacy against DDPM and DDIM models on CIFAR-10 and CelebA datasets, while maintaining the functional integrity of the models. Our code is available for reproducibility at https://anonymous.4open.science/r/b2hoaWNhbnRzZWV0aGF0bm9vb29vb29vb29v.

AB - Edge AI-driven diffusion models (DMs) are increasingly integrated into consumer devices for high-quality data generation and content creation. This paper introduces InvisibleDiffusion, a novel backdoor attack framework for diffusion models in consumer electronics, designed to remain undetected by utilizing a non-standard Gaussian distribution as a concealed trigger. Unlike previous backdoor methods, InvisibleDiffusion does not rely on obvious visual triggers, enhancing its stealthiness. Extensive experiments demonstrate that InvisibleDiffusion achieves high attack efficacy against DDPM and DDIM models on CIFAR-10 and CelebA datasets, while maintaining the functional integrity of the models. Our code is available for reproducibility at https://anonymous.4open.science/r/b2hoaWNhbnRzZWV0aGF0bm9vb29vb29vb29v.

KW - Consumer devices

KW - Edge AI

KW - Generative artificial intelligence

KW - Security in deep learning

KW - Gaussian distribution

KW - Gaussian noise (electronic)

KW - Backdoors

KW - Diffusion model

KW - Driven diffusion

KW - Gaussians

KW - High quality data

KW - Smart devices

U2 - 10.1109/TCE.2024.3516352

DO - 10.1109/TCE.2024.3516352

M3 - Journal article

JO - IEEE Transactions on Consumer Electronics

JF - IEEE Transactions on Consumer Electronics

SN - 0098-3063

ER -