Jove
Visualize
Contact Us
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Related Experiment Videos

Backdoor Training Paradigm in Generative Adversarial Networks.

Huangji Wang1, Fan Cheng1

  • 1Department of Computer Science and Engineering, Shanghai Jiao Tong University, Shanghai 200240, China.

Entropy (Basel, Switzerland)
|March 28, 2025
PubMed
Summary
This summary is machine-generated.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same journal

Research on a Regional Availability Evaluation Model for Road-Area High-Entropy Energy Based on Synergy Factors.

Entropy (Basel, Switzerland)·2026
Same journal

Atmospheric Turbulence Channel Modeling and Performance Analysis of a CO-ZP-OFDM Coherent Optical Communication System for UAV Air-to-Ground Scenarios.

Entropy (Basel, Switzerland)·2026
Same journal

Information Geometry and Asymptotic Theory for SMML Estimators.

Entropy (Basel, Switzerland)·2026
Same journal

Correlation Entropy and Power-Law Kinetics.

Entropy (Basel, Switzerland)·2026
Same journal

Research on the Contagion of Systemic Financial Risk Under the Impact of Climate Risks-From the Perspective of Complex Networks and Machine Learning.

Entropy (Basel, Switzerland)·2026
Same journal

The Statistical-Mechanical Meaning of the Wave Function of Quantum Mechanics.

Entropy (Basel, Switzerland)·2026
See all related articles

Researchers developed a new method to embed backdoor triggers in generative models, enhancing their security and robustness against attacks. This unified approach works across different generative model types, improving machine learning safety.

Area of Science:

  • Machine Learning Security
  • Generative Models
  • Cybersecurity

Background:

  • Backdoor attacks are a significant threat in machine learning, often introduced via training injection mechanisms.
  • Existing methods for embedding backdoor triggers in generative models lack a unified approach.

Purpose of the Study:

  • To identify a unifying pattern in existing backdoor injection methods for generative models.
  • To propose a novel, generalized paradigm for backdoor training injection in diverse generative models.

Main Methods:

  • Developed a unified loss function design for backdoor injection.
  • Applied the proposed paradigm to Generative Adversarial Networks (GANs) and Diffusion Models.
  • Conducted experiments to validate effectiveness and generalizability.
Keywords:
GANbackdoor attackdiffusion modelgenerative modelparadigm

Related Experiment Videos

Main Results:

  • The proposed method successfully embedded backdoor triggers in GANs.
  • Demonstrated the effectiveness and generalizability of the paradigm across different generative models.
  • Showcased enhanced model security and robustness through successful backdoor embedding.

Conclusions:

  • The novel paradigm offers a unified and effective approach for backdoor injection in generative models.
  • This work contributes a new methodological framework for improving the safety and reliability of generative AI.
  • The findings advance the field of machine learning security, particularly for generative applications.