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

Oriented principal component analysis for large margin classifiers.

S Bermejo1, J Cabestany

  • 1Department of Electronic Engineering, Universitat Politècnica de Catalunya (UPC), Barcelona, Spain. sbermejo@eel.upc.es

Neural Networks : the Official Journal of the International Neural Network Society
|January 5, 2002
PubMed
Summary
This summary is machine-generated.

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

Clinical guide to perioperative management for videothoracoscopy lung resection (Section of Cardiac, Vascular and Thoracic Anesthesia, SEDAR; Spanish Society of Thoracic Surgery, SECT; Spanish Society of Physiotherapy).

Revista espanola de anestesiologia y reanimacion·2022
Same author

Neutrophilic dermatosis of the dorsal hands.

Revista clinica espanola·2021
Same author

Clinical guide to perioperative management for videothoracoscopy lung resection (Section of Cardiac, Vascular and Thoracic Anesthesia, SEDAR; Spanish Society of Thoracic Surgery, SECT; Spanish Society of Physiotherapy).

Revista espanola de anestesiologia y reanimacion·2021
Same author

Recommendations for the evaluation and management of the anticipated and non-anticipated difficult airway of the Societat Catalana d'Anestesiologia, Reanimació i Terapèutica del Dolor, based on the adaptation of clinical practice guidelines and expert consensus.

Revista espanola de anestesiologia y reanimacion·2020
Same author

Neutrophilic dermatosis of the dorsal hands.

Revista clinica espanola·2020
Same author

Estimating bradykinesia severity in Parkinson's disease by analysing gait through a waist-worn sensor.

Computers in biology and medicine·2017

This study introduces a novel global learning algorithm for large margin classifiers, incorporating Principal Component Analysis (PCA) for feature selection. This method enhances classifier generalization by optimizing feature extraction and margin control simultaneously.

Area of Science:

  • Machine Learning
  • Pattern Recognition
  • Computational Statistics

Background:

  • Large margin classifiers, including Multilayer Perceptrons (MLPs), aim for high-confidence class assignments.
  • Regularization and feature extraction are crucial for improving the generalization capabilities of these classifiers.
  • Optimal feature subsets are problem- and classifier-dependent, necessitating adaptive learning algorithms.

Purpose of the Study:

  • To develop a global learning algorithm for large margin classifiers that integrates feature extraction techniques.
  • To enhance classifier generalization by controlling the margin and penalizing classifiers with overly large margins.
  • To investigate the efficacy of Principal Component Analysis (PCA) as a regularization term for margin control.

Main Methods:

Related Experiment Videos

  • Optimization of a cost function incorporating margin error and regularization terms.
  • Inclusion of a Principal Component Analysis (PCA) term to control classifier margin.
  • Proposal of a constrained, separate training approach for feature extractors and classifiers.

Main Results:

  • Demonstration that a PCA term can effectively control the margin of a large margin classifier.
  • The proposed separate training strategy allows flexibility and incorporation of heuristics.
  • Experimental results indicate the potential of the developed method for improved performance.

Conclusions:

  • The integration of PCA within a global learning framework offers an effective approach to margin control in large margin classifiers.
  • The flexible, constrained search algorithm facilitates enhanced performance through heuristic integration.
  • The proposed method shows promise for improving the generalization properties of large margin classifiers in various classification tasks.