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

Discriminative components of data.

Jaakko Peltonen1, Samuel Kaski

  • 1Neural Networks Research Center, Helsinki University of Technology, Helsinki, FIN-02015 HUT Finland. jaakko.peltonen@hut.fi

IEEE Transactions on Neural Networks
|March 1, 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

PI-RADS v2 and Adverse Prostate Cancer Outcomes: A Cross-cohort Replication Study Across Three Centers.

European urology oncology·2026
Same author

Misspecification-robust likelihood-free inference in high dimensions.

Computational statistics·2025
Same author

Towards modeling evolving longitudinal health trajectories with a transformer-based deep learning model.

Annals of epidemiology·2025
Same author

VitroBert: modeling DILI by pretraining BERT on in vitro data.

Journal of cheminformatics·2025
Same author

E-GuARD: expert-guided augmentation for the robust detection of compounds interfering with biological assays.

Journal of cheminformatics·2025
Same author

Molecular property prediction using pretrained-BERT and Bayesian active learning: a data-efficient approach to drug design.

Journal of cheminformatics·2025
Same journal

Universal perceptron and DNA-like learning algorithm for binary neural networks: LSBF and PBF implementations.

IEEE transactions on neural networks·2013
Same journal

Guest editorial: special section on white box nonlinear prediction models.

IEEE transactions on neural networks·2011
Same journal

Data-based fault-tolerant control of high-speed trains with traction/braking notch nonlinearities and actuator failures.

IEEE transactions on neural networks·2011
Same journal

Guest editorial: special section on data-based control, modeling, and optimization.

IEEE transactions on neural networks·2011
Same journal

Neural network-based multiple robot simultaneous localization and mapping.

IEEE transactions on neural networks·2011
Same journal

Data-driven model-free adaptive control for a class of MIMO nonlinear discrete-time systems.

IEEE transactions on neural networks·2011
See all related articles

A new probabilistic model generalizes linear discriminant analysis (LDA) to find informative components for data classes. This method enhances data exploration and dimensionality reduction, outperforming existing techniques.

Area of Science:

  • Machine Learning
  • Statistical Modeling
  • Data Science

Background:

  • Classical linear discriminant analysis (LDA) is a standard technique for finding informative components.
  • Existing methods may not fully capture class-relevant information or optimize class distribution predictability.

Purpose of the Study:

  • To introduce a novel probabilistic model that generalizes LDA.
  • To identify components that maximize class distribution predictability, enhancing data relevance.

Main Methods:

  • Developed a simple probabilistic model to generalize linear discriminant analysis (LDA).
  • Components were designed to maximize class distribution predictability, asymptotically equivalent to maximizing mutual information and finding principal components in Fisher metrics.
  • Utilized Fisher metrics to measure class-relevant distances.

Related Experiment Videos

Main Results:

  • The proposed components effectively capture information relevant to data classes.
  • Empirical experiments demonstrated superior performance compared to classical methods and a Renyi entropy-based alternative.
  • The method achieved comparable computational cost to existing approaches.

Conclusions:

  • The novel probabilistic model offers an effective generalization of LDA for identifying informative components.
  • The approach is valuable for data exploration, visualization, and dimensionality reduction.
  • This method provides a robust and computationally efficient alternative for class-discriminative analysis.