跳到主要导航 跳到搜索 跳到主要内容

Fragile Model Watermark for integrity protection: Leveraging boundary volatility and sensitive sample-pairing

  • Zhen Zhe Gao
  • , Zhenjun Tang
  • , Zhaoxia Yin*
  • , Baoyuan Wu
  • , Yue Lu
  • *此作品的通讯作者
  • East China Normal University
  • Guangxi Normal University
  • The Chinese University of Hong Kong, Shenzhen

科研成果: 书/报告/会议事项章节 › 会议稿件 › 同行评审

摘要

Neural networks have increasingly influenced people's lives. Ensuring the faithful deployment of neural networks as designed by their model owners is crucial, as they may be susceptible to various malicious or unintentional modifications, such as backdooring and poisoning attacks. Fragile model watermarks aim to prevent unexpected tampering that could lead DNN models to make incorrect decisions. They ensure the detection of any tampering with the model as sensitively as possible. However, prior watermarking methods suffered from inefficient sample generation and insufficient sensitivity, limiting their practical applicability. Our approach employs a sample-pairing technique, placing the model boundaries between pairs of samples, while simultaneously maximizing logits. This ensures that the model's decision results of sensitive samples change as much as possible and the Top-1 labels easily alter regardless of the direction it moves. Experimental evaluations conducted across multiple models and datasets demonstrate the superior sensitivity and generation efficiency of our method compared to the current approaches.

源语言英语
主期刊名2024 IEEE International Conference on Multimedia and Expo, ICME 2024
出版商IEEE Computer Society
ISBN(电子版)9798350390155
DOI
出版状态已出版 - 2024
活动2024 IEEE International Conference on Multimedia and Expo, ICME 2024 - Niagra Falls, 加拿大
期限: 15 7月 2024 → 19 7月 2024

丛书

姓名Proceedings - IEEE International Conference on Multimedia and Expo
ISSN(印刷版)1945-7871
ISSN(电子版)1945-788X

会议

会议2024 IEEE International Conference on Multimedia and Expo, ICME 2024
国家/地区加拿大
市Niagra Falls
时期15/07/24 → 19/07/24

学术指纹

探究 'Fragile Model Watermark for integrity protection: Leveraging boundary volatility and sensitive sample-pairing' 的科研主题。它们共同构成独一无二的学术指纹。

引用此