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

Semisupervised learning of classifiers: theory, algorithms, and their application to human-computer interaction.

Ira Cohen1, Fabio G Cozman, Nicu Sebe

  • 1Hewlett-Packard Labs, 1501 Page Mill Rd., Palo Alto, CA 94304, USA. ira.cohen@hp.com

IEEE Transactions on Pattern Analysis and Machine Intelligence
|December 3, 2004
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

OmniCharacter++: Towards Comprehensive Benchmark for Realistic Role-Playing Agents.

IEEE transactions on pattern analysis and machine intelligence·2026
Same author

Sharpness-Aware Fine-Tuning for OOD Detection.

IEEE transactions on image processing : a publication of the IEEE Signal Processing Society·2026
Same author

Stable-Hair V2: Real-World Hair Transfer via Multiple-View Diffusion Model.

IEEE transactions on visualization and computer graphics·2026
Same author

A Unified Masked Jigsaw Puzzle Framework for Vision and Language Models.

IEEE transactions on pattern analysis and machine intelligence·2025
Same author

H<sub>2</sub>OT: Hierarchical Hourglass Tokenizer for Efficient Video Pose Transformers.

IEEE transactions on pattern analysis and machine intelligence·2025
Same author

SeCoV2: Semantic Connectivity-Driven Pseudo-Labeling for Robust Cross-Domain Semantic Segmentation.

IEEE transactions on pattern analysis and machine intelligence·2025

This study analyzes using unlabeled data for training probabilistic classifiers. Unlabeled data can improve classification performance, but only under specific conditions; otherwise, it can be detrimental.

Area of Science:

  • Computer Science
  • Artificial Intelligence
  • Machine Learning

Background:

  • Automatic classification is fundamental to pattern recognition and human-computer interaction.
  • Training classifiers with both labeled and unlabeled data is an active research area.

Purpose of the Study:

  • To analyze the conditions under which unlabeled data improves classification performance.
  • To investigate the impact of violating these conditions.
  • To propose a new structure learning algorithm for Bayesian networks that leverages unlabeled data.

Main Methods:

  • Theoretical analysis of probabilistic classifier training with labeled and unlabeled data.
  • Development of a new structure learning algorithm for Bayesian networks.
  • Empirical evaluation in facial expression recognition and face detection applications.

Related Experiment Videos

Main Results:

  • Identified specific conditions where unlabeled data enhances classification performance.
  • Demonstrated that violating these conditions can degrade performance.
  • Showcased improved classification through the proposed Bayesian network algorithm.

Conclusions:

  • Unlabeled data can be a valuable resource for improving classification, but its use must be carefully considered.
  • The proposed algorithm effectively utilizes unlabeled data in Bayesian networks for enhanced pattern recognition.
  • The findings have direct applications in human-computer interaction and facial analysis.