Watermarking for Model Ownership Verification:Invisible at Deployment, Activated by Updates
ACM Transactions on Privacy and Security · 31.08.2026 · 29 (3) · 1-28
DOI: 10.1145/3817059 ↗Kurz gefasst
Fachartikel in der Zeitschrift ACM Transactions on Privacy and Security, online erschienen am 11.06.2026. Verfasst von Hewang Nie und 5 weiteren Autor:innen. Die Arbeit nennt 50 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China, Natural Science Foundation of Guangxi Province, Guangxi “Bagui Scholar” Project.
Abstract (Originalsprache)
Deep neural networks for image classification require protection against unauthorized use and redistribution. Existing watermarking methods suffer from a critical vulnerability: watermarks are always active and detectable, allowing adversaries to identify and remove them before deployment. We propose DormMark, a novel framework for image classification models that introduces delayed-activation watermarks which remain dormant and hidden under deployment-time black-box query auditing during initial deployment, but automatically activate upon fine-tuning. Our approach employs a three-stage training paradigm: (1) embedding watermarks using triggered samples, (2) masking to suppress watermark functionality while preserving its latent presence, and (3) activation through standard fine-tuning without owner intervention. This mechanism exploits neural networks’ forgetting-remembering behaviors during continued training, creating a fragile equilibrium that behaves similarly to a clean model under deployment-time black-box auditing but reliably manifests ownership indicators after modification. We consider a private-key black-box verification setting in which the owner keeps the concrete trigger instances secret. Experiments across multiple architectures (VGG19, ResNet-18/56, DenseNet-121, WideResNet-34) and datasets (CIFAR-10, CIFAR-100, GTSRB) demonstrate 100% watermark success rates, high imperceptibility (PSNR > 38 dB, SSIM = 0.99), negligible accuracy loss (< 0.04%), and robustness against 80% parameter pruning. DormMark represents a paradigm shift from static to conditionally-activated ownership verification, providing a more robust framework for intellectual property protection.
Förderung
- National Natural Science Foundation of China
- Natural Science Foundation of Guangxi Province
- Guangxi “Bagui Scholar” Project
Zitieren
APA
Nie, H., Xiao, J., Lu, Z., Shen, R., Yu, C., & Tang, Z. (2026). Watermarking for Model Ownership Verification:Invisible at Deployment, Activated by Updates. ACM Transactions on Privacy and Security, 29(3), 1-28. https://doi.org/10.1145/3817059
Vancouver
Nie H, Xiao J, Lu Z, Shen R, Yu C, Tang Z. Watermarking for Model Ownership Verification:Invisible at Deployment, Activated by Updates. ACM Transactions on Privacy and Security. 2026;29(3):1-28. doi:10.1145/3817059
BibTeX
@article{Nie2026,
author = {Nie, Hewang and Xiao, Jue and Lu, Zhi and Shen, Renfei and Yu, Chunqiang and Tang, Zhenjun},
title = {Watermarking for Model Ownership Verification:Invisible at Deployment, Activated by Updates},
journal = {ACM Transactions on Privacy and Security},
year = {2026},
volume = {29},
number = {3},
pages = {1--28},
doi = {10.1145/3817059}
}
EUVIMED