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

Mixed group ranks: preference and confidence in classifier combination.

Ofer Melnik1, Yehuda Vardi, Cun-Hui Zhang

  • 1DIMACS CORE, Rutgers University, Piscataway, NJ 08854-8018, USA. melnik@dimacs.rutgers.edu

IEEE Transactions on Pattern Analysis and Machine Intelligence
|January 12, 2005
PubMed
Summary

Classifier combination enhances performance by integrating multiple classifier outputs. A new method, Mixed Group Ranks (MGR), balances classifier preference and confidence for improved results in large-scale biometrics and face recognition studies.

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

Comments on "A Scale-Free Approach for False Discovery Rate Control in Generalized Linear Models".

Journal of the American Statistical Association·2024
Same author

Finding a resveratrol analogue as potential anticancer agent with apoptosis and cycle arrest.

Journal of pharmacological sciences·2020
Same author

Finding an efficient tetramethylated hydroxydiethylene of resveratrol analogue for potential anticancer agent.

BMC chemistry·2020
Same author

A near-infrared fluorescent probe for direct and selective detection of cysteine over homocysteine and glutathione.

Spectrochimica acta. Part A, Molecular and biomolecular spectroscopy·2018
Same author

Lasso adjustments of treatment effect estimates in randomized experiments.

Proceedings of the National Academy of Sciences of the United States of America·2016
Same author

Estimation and Selection via Absolute Penalized Convex Minimization And Its Multistage Adaptive Applications.

Journal of machine learning research : JMLR·2013

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Pattern Recognition

Background:

  • Classifier combination can improve performance by integrating multiple classifier outputs.
  • Domains with numerous classes, like biometrics, require robust combination strategies.
  • Existing methods include Borda Count, Logistic Regression, and Highest Rank, representing extremes of a continuum.

Purpose of the Study:

  • To present an axiomatic framework for desirable mathematical properties of rank-based classifier combination functions.
  • To introduce Mixed Group Ranks (MGR) as a novel combination function generalizing existing methods.
  • To demonstrate the effectiveness of MGR in large-scale classification tasks.

Main Methods:

  • Developed an axiomatic framework defining properties for classifier combination functions.

Related Experiment Videos

  • Introduced Mixed Group Ranks (MGR) as a new function balancing classifier preference and confidence.
  • Conducted experiments using large datasets from the FERET face recognition study.

    • Main Results:

    • The proposed framework encompasses existing combination rules as specific cases.
    • Mixed Group Ranks (MGR) effectively balances classifier preference and confidence.
    • Experimental results on the FERET dataset demonstrate MGR's efficacy in large-class biometrics.

    Conclusions:

    • The axiomatic framework provides a theoretical basis for classifier combination.
    • Mixed Group Ranks (MGR) is a powerful and effective approach for combining rank-based classifiers.
    • MGR shows significant promise for applications in large-scale biometrics and face recognition.