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

Learning vector quantization with training data selection.

Carlos E Pedreira1

  • 1Federal University of Rio de Janeiro-UFRJ, Rio de Janeiro, Brazil. carlosp@centroin.com.br

IEEE Transactions on Pattern Analysis and Machine Intelligence
|January 13, 2006
PubMed
Summary

This study introduces a novel method for training Learning Vector Quantization (LVQ) models by strategically selecting data points for prototype updates. This approach aims to improve classification accuracy by guiding prototypes to optimal positions and reducing errors.

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

T-cell immune profile in blood of systemic mastocytosis: Association with disease features.

Allergy·2024
Same author

Altered innate immune profile in blood of systemic mastocytosis patients.

Clinical and translational allergy·2022
Same author

Immunophenotypic Analysis of Acute Megakaryoblastic Leukemia: A EuroFlow Study.

Cancers·2022
Same author

Expert-independent classification of mature B-cell neoplasms using standardized flow cytometry: a multicentric study.

Blood advances·2021
Same author

Automated identification of leukocyte subsets improves standardization of database-guided expert-supervised diagnostic orientation in acute leukemia: a EuroFlow study.

Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc·2020
Same author

Maturation-associated gene expression profiles during normal human bone marrow erythropoiesis.

Cell death discovery·2019

Area of Science:

  • Machine Learning
  • Data Science

Background:

  • Learning Vector Quantization (LVQ) is a machine learning algorithm used for classification.
  • Traditional LVQ methods can suffer from inefficient prototype convergence, leading to suboptimal classification performance.

Purpose of the Study:

  • To develop an improved LVQ training method that enhances prototype convergence and reduces misclassification errors.
  • To introduce a data selection strategy for more effective LVQ prototype updates.

Main Methods:

  • Proposes a method to select a subset of training data points for updating LVQ prototypes.
  • Identifies data points at risk of misclassification by other prototypes for inclusion in the update set.
  • Utilizes a weighted norm, rather than Euclidean distance, to assign differential importance to input attributes.

Related Experiment Videos

Main Results:

  • The proposed method aims to guide LVQ prototypes to more effective locations in the feature space.
  • The technique is expected to diminish misclassification errors compared to standard LVQ approaches.
  • The methodology was validated through controlled experiments and on real-world web datasets.

Conclusions:

  • The data point selection strategy combined with a weighted norm offers a promising approach for improving LVQ performance.
  • This method enhances the efficiency and accuracy of LVQ models in classification tasks.
  • Further research can explore variations in the weighted norm and data selection criteria.