EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Ditto: An Adaptable and Highly Robust Invisible Backdoor Attack Towards Deep Neural Networks

Wenhao Zhang, Lianheng Zou, Yingying Xiong, Peng Shi, Xiao He

Electronics · 08.04.2026 · 15 (8) · 1551

DOI: 10.3390/electronics15081551 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Electronics, online erschienen am 08.04.2026. Verfasst von Wenhao Zhang und 4 weiteren Autor:innen. Die Arbeit nennt 46 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Key R&D Program of China.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

With the widespread application of deep neural networks across various fields, issues related to model security have become increasingly prevalent. Backdoor attacks, as a covert method of attack, can implant malicious behavior during the model training process, causing the model to perform predetermined tasks under specific trigger conditions. However, current backdoor attacks struggle to achieve a good balance between stealthiness and attack success rate, and there is an issue in which certain data transformation operations can negatively impact attack performance. To address these issues, this paper proposes a specialized backdoor attack method called Ditto. It first uses a boundary detection algorithm and a padding algorithm to determine the trigger’s insertion position. The trigger is then dynamically generated using a generative adversarial network, taking into account the texture features of the images. Subsequently, the trigger is applied to the images, and its level of stealthiness is adjusted. Compared to existing popular backdoor attack methods, the experimental results ensure a high level of stealthiness while also maintaining a high attack success rate and a high accuracy for clean data. Furthermore, our attack method exhibits considerable robustness and adaptability, demonstrating effective resistance against baseline backdoor defense techniques.

Förderung

  • National Key R&D Program of China

Zitieren

APA

Zhang, W., Zou, L., Xiong, Y., Shi, P., & He, X. (2026). Ditto: An Adaptable and Highly Robust Invisible Backdoor Attack Towards Deep Neural Networks. Electronics, 15(8), 1551. https://doi.org/10.3390/electronics15081551

Vancouver

Zhang W, Zou L, Xiong Y, Shi P, He X. Ditto: An Adaptable and Highly Robust Invisible Backdoor Attack Towards Deep Neural Networks. Electronics. 2026;15(8):1551. doi:10.3390/electronics15081551

BibTeX
@article{Zhang2026,
  author = {Zhang, Wenhao and Zou, Lianheng and Xiong, Yingying and Shi, Peng and He, Xiao},
  title = {Ditto: An Adaptable and Highly Robust Invisible Backdoor Attack Towards Deep Neural Networks},
  journal = {Electronics},
  year = {2026},
  volume = {15},
  number = {8},
  pages = {1551},
  doi = {10.3390/electronics15081551}
}
Kostenlos starten →