EUVIMEDEUVIMED entdecken
Crossref

Class-feature Watermark: A Resilient Black-box Watermark Against Model Extraction Attacks

Yaxin Xiao, Qingqing Ye, Zi Liang, Haoyang Li, RongHua Li, Huadi Zheng, Haibo Hu

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (42) · 35903-35912

DOI: 10.1609/aaai.v40i42.40905 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Yaxin Xiao und 6 weiteren Autor:innen. Die Arbeit wurde bisher 2-mal zitiert (Crossref).

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Machine learning models constitute valuable intellectual property, yet remain vulnerable to model extraction attacks (MEA), where adversaries replicate their functionality through black-box queries. Model watermarking counters MEAs by embedding forensic markers for ownership verification. Current black-box watermarks prioritize MEA survival through representation entanglement, yet inadequately explore resilience against sequential MEAs and removal attacks. Our study reveals that this risk is underestimated because existing removal methods are weakened by entanglement. To address this gap, we propose Watermark Removal attacK (WRK), which circumvents entanglement constraints by exploiting decision boundaries shaped by prevailing sample-level watermark artifacts. WRK effectively reduces watermark success rates by ≥88.79% across existing watermarking benchmarks. For robust protection, we propose Class-Feature Watermarks (CFW), which improve resilience by leveraging class-level artifacts. CFW constructs a synthetic class using out-of-domain samples, eliminating vulnerable decision boundaries between original domain samples and their artifact-modified counterparts (watermark samples). CFW concurrently optimizes both MEA transferability and post-MEA stability. Experiments across multiple domains show that CFW consistently outperforms prior methods in resilience, maintaining a watermark success rate of ≥70.15% in extracted models even under the combined MEA and WRK distortion, while preserving the utility of protected models.

Zitieren

APA

Xiao, Y., Ye, Q., Liang, Z., Li, H., Li, R., Zheng, H., & Hu, H. (2026). Class-feature Watermark: A Resilient Black-box Watermark Against Model Extraction Attacks. Proceedings of the AAAI Conference on Artificial Intelligence, 40(42), 35903-35912. https://doi.org/10.1609/aaai.v40i42.40905

Vancouver

Xiao Y, Ye Q, Liang Z, Li H, Li R, Zheng H, et al. Class-feature Watermark: A Resilient Black-box Watermark Against Model Extraction Attacks. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(42):35903-35912. doi:10.1609/aaai.v40i42.40905

BibTeX
@article{Xiao2026,
  author = {Xiao, Yaxin and Ye, Qingqing and Liang, Zi and Li, Haoyang and Li, RongHua and Zheng, Huadi and Hu, Haibo},
  title = {Class-feature Watermark: A Resilient Black-box Watermark Against Model Extraction Attacks},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {42},
  pages = {35903--35912},
  doi = {10.1609/aaai.v40i42.40905}
}
Kostenlos starten →