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 Concept Videos

Cluster Sampling Method01:20

Cluster Sampling Method

Appropriate sampling methods ensure that samples are drawn without bias and accurately represent the population. Because measuring the entire population in a study is not practical, researchers use samples to represent the population of interest.
To choose a cluster sample, divide the population into clusters (groups) and then randomly select some of the clusters. All the members from these clusters are in the cluster sample. For example, if you randomly sample four departments from your...

You might also read

Related Articles

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

Sort by
Same author

[Subcellular re-distribution of F-actin in MCF7 cells upon inducible expression of CDKN2A/p16<sup>INK4a</sup>].

Xi bao yu fen zi mian yi xue za zhi = Chinese journal of cellular and molecular immunology·2017
Same author

Availability, Price, and Affordability of Selected Essential Medicines for Chronic Diseases in 11 Countries of the Asia Pacific Region: A Secondary Analysis.

Asia-Pacific journal of public health·2017
Same author

Immunization With a Subunit Hepatitis C Virus Vaccine Elicits Pan-Genotypic Neutralizing Antibodies and Intrahepatic T-Cell Responses in Nonhuman Primates.

The Journal of infectious diseases·2017
Same author

Immunogold scanning electron microscopy can reveal the polysaccharide architecture of xylem cell walls.

Journal of experimental botany·2017
Same author

Competition between Hydrogen Bonds and Coordination Bonds Steered by the Surface Molecular Coverage.

ACS nano·2017
Same author

Intrathecal administration of human bone marrow mesenchymal stem cells genetically modified with human proenkephalin gene decrease nociceptive pain in neuropathic rats.

Molecular pain·2017

Related Experiment Video

Updated: Jun 17, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

SAKCL: a deep neural network test data selection method based on self-attention and K-means clustering.

Tingting Huo1, Qiang Sun1, Rui Ding2

  • 1The School of Computer and Information Technology, Mudanjiang Normal University, Mudanjiang, China.

Scientific Reports
|June 15, 2026
PubMed
Summary

A new method called SAKCL uses self-attention and K-means clustering to select better test data for deep neural networks (DNNs). This improves defect detection and model accuracy in critical applications.

Keywords:
Deep neural networksDefect detectionK-means clusteringSelf-attention mechanismTest data selection

More Related Videos

Artificial Intelligence-Based System for Detecting Attention Levels in Students
06:37

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

Related Experiment Videos

Last Updated: Jun 17, 2026

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments
03:31

End-To-End Deep Neural Network for Salient Object Detection in Complex Environments

Published on: December 15, 2023

Artificial Intelligence-Based System for Detecting Attention Levels in Students
06:37

Artificial Intelligence-Based System for Detecting Attention Levels in Students

Published on: December 15, 2023

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Software Engineering

Background:

  • Deep neural networks (DNNs), like traditional software, can have defects impacting safety-critical systems.
  • Effective defect detection in DNNs relies heavily on the quality of test datasets.

Purpose of the Study:

  • To introduce a novel test data selection method for enhancing DNN quality assurance.
  • To improve the balance between fault detection capability and diversity in test datasets.

Main Methods:

  • A coordinated approach combining a self-attention mechanism and K-means clustering (SAKCL).
  • Self-attention highlights informative features and reduces redundancy.
  • K-means clustering organizes data based on refined features to capture structural relationships.

Main Results:

  • SAKCL consistently outperforms existing methods across benchmark datasets and DNN architectures.
  • Achieved average improvements: >12% in fault-revealing test cases (FDR), >6.5% in diversity (KMNC), and +3.319% in retraining accuracy (ΔAcc).
  • Statistical analysis confirms significant and reliable improvements (p < 0.01, Cliff's δ > 0.8).

Conclusions:

  • SAKCL offers a practical and scalable solution for effective test data selection in DNNs.
  • The method enhances DNN quality assurance by improving defect detection and model robustness.