Jove
Visualize
联系我们
JoVE
x logofacebook logolinkedin logoyoutube logo
关于 JoVE
概览领导团队博客JoVE 帮助中心
作者
出版流程编辑委员会范围与政策同行评审常见问题投稿
图书馆员
用户评价订阅访问资源图书馆顾问委员会常见问题
研究
JoVE JournalMethods CollectionsJoVE Encyclopedia of Experiments存档
教育
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab Manual教师资源中心教师网站
使用条款与条件
隐私政策
政策

相关概念视频

您也可能阅读

相关文章

通过共同作者、期刊和引用图与本文相关的文章。

排序
Same author

[Diagnostic value of parameters of glucose metabolism as screening tests for insulinoma].

Zhonghua yi xue za zhi·2010
Same author

Cooperative binding of MgATP and MgADP in the trimeric P(II) protein GlnK2 from Archaeoglobus fulgidus.

Journal of molecular biology·2010
Same author

Autophagy plays an important role in sunitinib-mediated cell death in H9c2 cardiac muscle cells.

Toxicology and applied pharmacology·2010
Same author

Design and synthesis of novel benzimidazole derivatives as inhibitors of hepatitis B virus.

Bioorganic & medicinal chemistry·2010
Same author

Evaluation of genetic susceptibility loci for obesity in Chinese women.

American journal of epidemiology·2010
Same author

Up-regulation of death receptor 4 and 5 by celastrol enhances the anti-cancer activity of TRAIL/Apo-2L.

Cancer letters·2010

相关实验视频

Updated: May 24, 2025

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
03:31

Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

Published on: December 15, 2023

451

隐形的DNN水印对抗模型提取攻击

Zuping Xi, Zuomin Qu, Wei Lu

    IEEE transactions on cybernetics
    |March 3, 2025
    PubMed
    概括
    此摘要是机器生成的。

    这项研究引入了一个新的协作深度神经网络 (DNN) 水标框架. 它通过将水标和模型任务联系起来,有效地保护DNN模型免受提取攻击,确保所有权验证.

    更多相关视频

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
    04:48

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

    Published on: November 30, 2022

    2.6K
    A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
    04:23

    A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

    Published on: April 21, 2023

    1.7K

    相关实验视频

    Last Updated: May 24, 2025

    Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications
    03:31

    Author Spotlight: Enhancement of Salient Object Detection for Smart Grid Applications

    Published on: December 15, 2023

    451
    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography
    04:48

    Application of Deep Learning-Based Medical Image Segmentation via Orbital Computed Tomography

    Published on: November 30, 2022

    2.6K
    A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images
    04:23

    A Swin Transformer-Based Model for Thyroid Nodule Detection in Ultrasound Images

    Published on: April 21, 2023

    1.7K

    科学领域:

    • 计算机科学 计算机科学
    • 人工智能的人工智能

    背景情况:

    • 深度神经网络 (DNN) 是有价值的商业资产,需要对未经授权的使用进行保护.
    • 目前的DNN水印方法由于独立的水印和模型任务,容易受到模型提取攻击.

    研究的目的:

    • 提出一个新的协作DNN水印框架,以防范模型提取攻击.
    • 通过改进水印技术,提高DNN模型的安全性和稳定性.

    主要方法:

    • 开发了一个协作框架,其中水标生成和嵌入相互连接.
    • 使用不可察觉的触发器样本,注入目标标签信息以进行隐形和指导.
    • 实现了特征合,以将触发样本特征与任务分配样本对齐.

    主要成果:

    • 拟议的框架成功地防御了模型提取攻击.
    • 触发器样本在被盗模型中被识别为任务分配样本,使所有权验证成为可能.
    • 在CIFAR10,CIFAR100和ImageNet上的实验表明了卓越的性能.

    结论:

    • 协作DNN水印框架提供了有效的保护,防止模型提取.
    • 功能合机制确保了强大的水标嵌入和所有权验证.
    • 这种方法显著提高了有价值的DNN模型的安全性.