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

Data Validation01:15

Data Validation

164
Method validation is a crucial process in analytical chemistry designed to confirm that a given method consistently produces reliable and high-quality results. This process is essential when a method is applied to different sample matrices or when procedural modifications are made, ensuring that the results meet acceptable standards across various applications.
Key parameters for method validation include:
164

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

A framework for assessing carbon effect of land consolidation with life cycle assessment: A case study in China.

Journal of environmental management·2020
Same author

Precise control of the interlayer twist angle in large scale MoS<sub>2</sub> homostructures.

Nature communications·2020
Same author

Atomic-Precision Repair of a Few-Layer 2H-MoTe<sub>2</sub> Thin Film by Phase Transition and Recrystallization Induced by a Heterophase Interface.

Advanced materials (Deerfield Beach, Fla.)·2020
Same author

A single molecular sensor for selective and differential colorimetric/ratiometric detection of Cu<sup>2+</sup> and Pd<sup>2+</sup> in 100% aqueous solution.

Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy·2020
Same author

Effect of extracellular polymer substances on the tetracycline removal during coagulation process.

Bioresource technology·2020
Same author

Identification and Comparison of Tannins in Gall of Rhus chinensis Mill. and Gall of Quercus infectoria Oliv. by High-Performance Liquid Chromatography-Electrospray Mass Spectrometry.

Journal of chromatographic science·2020

相关实验视频

Updated: Jul 10, 2025

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
05:47

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems

Published on: June 13, 2025

236

在下游任务中学习一个强大的基础模型来抵御清洁标签数据中毒攻击.

Ting Zhou1, Hanshu Yan2, Bo Han3

  • 1Shandong University, Jinan, China.

Neural networks : the official journal of the International Neural Network Society
|November 19, 2023
PubMed
概括

本研究介绍了在转移学习中对数据中毒攻击的防御策略. 该方法通过调整特征距离来增强基础模型,显著提高了对恶意操纵的稳定性.

关键词:
清洁标签中毒攻击的攻击.坚固的基础模型.转移学习转移学习

更多相关视频

Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K
Label-Free Quantitative Proteomics Workflow for Discovery-Driven Host-Pathogen Interactions
05:37

Label-Free Quantitative Proteomics Workflow for Discovery-Driven Host-Pathogen Interactions

Published on: October 20, 2020

6.8K

相关实验视频

Last Updated: Jul 10, 2025

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems
05:47

Evidence-based Knowledge Synthesis and Hypothesis Validation: Navigating Biomedical Knowledge Bases via Explainable AI and Agentic Systems

Published on: June 13, 2025

236
Deep Neural Networks for Image-Based Dietary Assessment
13:19

Deep Neural Networks for Image-Based Dietary Assessment

Published on: March 13, 2021

9.2K
Label-Free Quantitative Proteomics Workflow for Discovery-Driven Host-Pathogen Interactions
05:37

Label-Free Quantitative Proteomics Workflow for Discovery-Driven Host-Pathogen Interactions

Published on: October 20, 2020

6.8K

科学领域:

  • 机器学习 机器学习
  • 人工智能的人工智能
  • 网络安全 网络安全

背景情况:

  • 转移学习利用预先训练的模型来完成下游任务,从而为数据中毒创造了脆弱性.
  • 攻击者将恶意数据注入重新训练集中,以操纵模型行为.
  • 现有的防御系统在与复杂的攻击作斗争,例如清洁标签中毒.

研究的目的:

  • 在转移学习中,制定强有力的防御策略来抵御数据中毒攻击.
  • 提高下游应用中使用的基础模型的安全性.
  • 降低机器学习管道中对抗性操纵的成功率.

主要方法:

  • 提出了一项防务战略,重点是预训练强大的基础模型.
  • 实施技术以减少对抗特征的距离.
  • 实施技术以增加类间特征距离.

主要成果:

  • 拟议的战略显著降低了数据中毒攻击的成功率.
  • 在最先进的清洁标签中毒攻击中表现出卓越的防御性能.
  • 在转移学习场景中验证的有效性.

结论:

  • 开发的防御策略为转移学习中的数据中毒提供了有效的保护.
  • 调整特征距离是构建强大的基础模型的可行方法.
  • 这项工作有助于在现实应用中确保机器学习模型的安全.