English

Performance-lossless Black-box Model Watermarking

Cryptography and Security 2024-04-16 v2

Abstract

With the development of deep learning, high-value and high-cost models have become valuable assets, and related intellectual property protection technologies have become a hot topic. However, existing model watermarking work in black-box scenarios mainly originates from training-based backdoor methods, which probably degrade primary task performance. To address this, we propose a branch backdoor-based model watermarking protocol to protect model intellectual property, where a construction based on a message authentication scheme is adopted as the branch indicator after a comparative analysis with secure cryptographic technologies primitives. We prove the lossless performance of the protocol by reduction. In addition, we analyze the potential threats to the protocol and provide a secure and feasible watermarking instance for language models.

Keywords

Cite

@article{arxiv.2312.06488,
  title  = {Performance-lossless Black-box Model Watermarking},
  author = {Na Zhao and Kejiang Chen and Weiming Zhang and Nenghai Yu},
  journal= {arXiv preprint arXiv:2312.06488},
  year   = {2024}
}
R2 v1 2026-06-28T13:47:16.698Z