Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Related Concept Videos

You might also read

Related Articles

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

Sort by

Same author

Pixel Latency Measurements of Event Cameras.

IEEE transactions on instrumentation and measurement·2026

Same author

Relating human and AI-based detection limits in scanning electron microscopy dimensional metrology.

Journal of micro/nanolithography, MEMS, and MOEMS : JM3·2026

Same author

AI driven 3D subcellular RPE map discovers cell state transitions in establishment of apical-basal polarity.

NPJ artificial intelligence·2026

Same author

Enabling global image data sharing in the life sciences.

Nature methods·2025

Same author

Memory-efficient semantic segmentation of large microscopy images using graph-based neural networks.

Microscopy (Oxford, England)·2023

Same author

SEQUIN is an R/Shiny framework for rapid and reproducible analysis of RNA-seq data.

Cell reports methods·2023

Same journal

Non-Thermal Plasma Accelerates Astrocyte Regrowth and Neurite Regeneration Following Physical Trauma In Vitro.

Applied sciences (Basel, Switzerland)·2026

Same journal

The Basic Reproduction Number for Petri Net Models: A Next-Generation Matrix Approach.

Applied sciences (Basel, Switzerland)·2026

Same journal

Tetracenomycin Aglycones Primarily Inhibit Cell Growth and Proliferation in Mammalian Cancer Cell Lines.

Applied sciences (Basel, Switzerland)·2026

Same journal

Neuroengineering Frontiers: A Selective Review of Neural Interfaces, Brain-Machine Interactions, and Artificial Intelligence in Neurodegenerative Diseases.

Applied sciences (Basel, Switzerland)·2026

Same journal

Engineering Brain Injury In Vitro: Human iPSC-Based Organoids in Microfluidic Systems.

Applied sciences (Basel, Switzerland)·2026

Same journal

Comparing Virtual Reality and Robotic Training Effects on Balance Ability and Confidence in Older Adults.

Applied sciences (Basel, Switzerland)·2026

See all related articles

Search research articles

Related Experiment Video

Updated: Oct 24, 2025

Designing and Implementing Nervous System Simulations on LEGO Robots

Designing and Implementing Nervous System Simulations on LEGO Robots

Published on: May 25, 2013

Designing Trojan Detectors in Neural Networks Using Interactive Simulations.

Peter Bajcsy¹, Nicholas J Schaub², Michael Majurski¹

¹Information Technology Laboratory, National Institute of Standards and Technology (NIST), 100 Bureau Drive. Gaithersburg, MD 20899.

Applied Sciences (Basel, Switzerland)

|August 13, 2021

Summary

This summary is machine-generated.

This study introduces a novel method for detecting neural network (NN) trojans using interactive simulations and Kullback-Liebler (KL) divergence measurements. The developed trojan detector effectively distinguishes between clean and compromised NNs.

Keywords:

neural network models security trojan attacks

More Related Videos

Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks

Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks

Published on: March 2, 2015

Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms

Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms

Published on: November 1, 2019

Related Experiment Videos

Last Updated: Oct 24, 2025

Designing and Implementing Nervous System Simulations on LEGO Robots

Designing and Implementing Nervous System Simulations on LEGO Robots

Published on: May 25, 2013

Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks

Closed-loop Neuro-robotic Experiments to Test Computational Properties of Neuronal Networks

Published on: March 2, 2015

Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms

Statistical Modelling of Cortical Connectivity Using Non-invasive Electroencephalograms

Published on: November 1, 2019

Area of Science:

Computer Science
Artificial Intelligence
Machine Learning Security

Background:

Neural networks (NNs) are vulnerable to trojans, which are malicious inputs causing unintended misclassifications.
Understanding trojan encodings in NNs is crucial for developing effective defense mechanisms.

Purpose of the Study:

To investigate the encodings of various trojan types within fully connected layers of NNs.
To develop and evaluate a trojan detection method for NNs.

Main Methods:

Simulated nine types of trojan embeddings in dot patterns using TensorFlow Playground.
Devised measurements of NN states, including modified Kullback-Liebler (KL) divergence for model inefficiency.
Designed a trojan detector based on KL divergence measurements per NN layer and class label.

Main Results:

The KL divergence measurement effectively discriminates between NNs with and without trojans.
The robustness of the trojan detector was demonstrated across different NN architectures, datasets, and trojan types.

Conclusions:

Interactive simulations and KL divergence measurements provide a viable approach for designing NN trojan detectors.
The developed trojan detector shows robustness and can be used for classifying NN models with or without trojans.