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 multiview face subspaces and facial pose estimation using independent component analysis.

Stan Z Li1, XiaoGuang Lu, Xinwen Hou

  • 1Microsoft Research Asia, Beijing 100080, China. szli@microsoft.com

IEEE Transactions on Image Processing : a Publication of the IEEE Signal Processing Society
|June 24, 2005
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

CHIP/STUB1 suppresses colorectal tumorigenesis through ubiquitin-mediated degradation of LDHA and inhibition of glycolysis.

Cellular oncology (Dordrecht, Netherlands)·2026
Same author

Towards the construction of a virtual yeast.

Nature·2026
Same author

MoE-Enhanced Explainable Deep Manifold Transformation for Complex Data Embedding and Visualization.

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

An Integrated Rotation-Programmed and Gravity-Driven Microfluidic Platform for Direct Bacterial Detection from Whole Blood.

Analytical chemistry·2026
Same author

Efficient Inference for Large Reasoning Models: A Survey.

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

3D-Photothermal Ultrafast PCR via rGO/PDMS-Based Microfluidic Chip and Its Application in Rapid Identification of Gram-Negative Bacteria in Whole Blood.

ACS applied materials & interfaces·2026

Independent Component Analysis (ICA) and its variants like ISA and TICA effectively learn face view representations from mixed data without labels. These methods outperform Principal Component Analysis (PCA) in unsupervised multiview face recognition.

Area of Science:

  • Computer Vision
  • Machine Learning
  • Pattern Recognition

Background:

  • Unsupervised learning of object representations from multiview data is challenging.
  • Principal Component Analysis (PCA) struggles with mixed-view data due to reliance on second-order statistics.
  • Higher-order statistics are crucial for characterizing object views.

Purpose of the Study:

  • To present an Independent Component Analysis (ICA) based approach for learning view-specific face representations.
  • To evaluate ICA variants (ISA, TICA) for unsupervised multiview face analysis.
  • To investigate the emergent properties of view subspaces learned by ISA.

Main Methods:

  • Utilizing Independent Component Analysis (ICA) and its variants: Independent Subspace Analysis (ISA) and Topographic Independent Component Analysis (TICA).

Related Experiment Videos

  • Applying these methods to learn basis components from unlabeled, mixed-view face datasets.
  • Analyzing the learned subspaces to understand view characterization.

    • Main Results:

    • ICA, TICA, and ISA successfully learn view-specific basis components unsupervisedly.
    • These methods demonstrate superior performance over PCA for multiview face representation.
    • Investigated ISA results revealed emergent view subspaces with underlying formation reasons.

    Conclusions:

    • ICA-based methods are effective for unsupervised learning of view-specific representations from multiview data.
    • Higher-order statistics utilized by ICA variants are key to robust object view characterization.
    • The study provides insights into the unsupervised formation of view subspaces in face recognition.