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

AVC: Selecting discriminative features on basis of AUC by maximizing variable complementarity.

Lei Sun1, Jun Wang1, Jinmao Wei2

  • 1Institute of Big Data, College of Computer and Control Engineering, Nankai University, 38 Tongyan Road, Tianjin, 300350, China.

BMC Bioinformatics
|April 1, 2017
PubMed
Summary

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

A unified framework combining linear and 3D molecular features for robust drug-protein interaction prediction.

Cell reports methods·2026
Same author

Accurate prediction of drug-protein interactions by maintaining the original topological relationships among embeddings.

BMC biology·2025
Same author

PCLT-PPI: Predicting Multi-Type Interactions Between Proteins Based on Point Cloud Structure and Local Topology Preservation.

IEEE journal of biomedical and health informatics·2025
Same author

Dual-Regularized Feature Selection for Class-Specific and Global Feature Associations.

Entropy (Basel, Switzerland)·2025
Same author

Precise detection of cell-type-specific domains in spatial transcriptomics.

Cell reports methods·2024
Same author

Predicting Drug-Protein Interactions by Self-Adaptively Adjusting the Topological Structure of the Heterogeneous Network.

IEEE journal of biomedical and health informatics·2023

This study introduces a new ROC-based method for selecting disease-related genes. It effectively identifies complementary features, improving classification accuracy with fewer selected genes.

Area of Science:

  • Bioinformatics
  • Machine Learning
  • Genomics

Background:

  • Receiver Operator Characteristic (ROC) curves are vital for evaluating classification performance in biomedicine, especially with imbalanced or cost-sensitive data.
  • Existing ROC-based feature selection methods effectively evaluate individual genes but struggle to reduce feature redundancy, hindering optimal subset selection.
  • Feature redundancy is a critical challenge in machine learning for identifying true disease-related gene subsets.

Purpose of the Study:

  • To develop a novel ROC-based feature selection approach that addresses the limitations of existing methods by assessing feature complementarity.
  • To improve the identification of disease-related genes by reducing redundancy and selecting optimal feature subsets.
  • To enhance classification performance in biomedical data analysis.
Keywords:
AUCFeature complementarityFeature selectionROC curve

Related Experiment Videos

Main Methods:

  • Propose a novel method to assess feature complementarity by measuring distances between misclassified instances and their nearest misses across feature dimensions.
  • Develop a new filter feature selection approach based on ROC analysis, utilizing an efficient heuristic search strategy.
  • Evaluate the approach on diverse microarray datasets to select optimal features with high complementarities.

Main Results:

  • The proposed approach effectively identifies complementary features, leading to improved classification performance.
  • Classifiers built on the selected feature subsets achieve minimal balanced error rates.
  • The method successfully selects a small number of significant features, reducing data complexity.

Conclusions:

  • The novel ROC-based feature selection approach outperforms existing methods in selecting fewer, more informative features.
  • This method significantly enhances classification performance compared to other ROC-based techniques.
  • The approach provides an effective strategy for identifying disease-related genes and improving machine learning models in biomedicine.