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

Stochastic organization of output codes in multiclass learning problems.

W Utschick1, W Weichselberger

  • 1Institute for Network Theory and Signal Processing, Munich University of Technology, D-80290 Munich, Germany.

Neural Computation
|May 22, 2001
PubMed
Summary

This study introduces a new algorithm for multiclass learning problems, optimizing output codes before classifier training. The novel method enhances decision rules, outperforming existing one per class coding and error-correcting output coding approaches.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same journal

A Model-Free Reinforcement Learning Implementation of Decision Making Under Uncertainty by Sequential Sampling.

Neural computation·2026
Same journal

DROP: Distributional and Regular Optimism and Pessimism for Reinforcement Learning.

Neural computation·2026
Same journal

Hierarchical Active Inference Using Successor Representations.

Neural computation·2026
Same journal

W-Kernel and Its Principal Space for Frequentist Evaluation of Bayesian Estimators.

Neural computation·2026
Same journal

A Hidden Markov Model-Inspired Sequence Classification Method for Hyperdimensional Computing.

Neural computation·2026
Same journal

Sparse Graphical Modeling for Electrophysiological Phase-Based Connectivity Using Circular Statistics.

Neural computation·2026

Area of Science:

  • Machine Learning
  • Computer Science
  • Artificial Intelligence

Background:

  • Standard multiclass learning relies on predefined decomposition schemes like one per class (OPC) and error-correcting output coding (ECOC).
  • These methods decompose problems before classifier training, with code impact assessed post-learning.
  • Existing approaches may not yield optimal decision rules for complex multiclass problems.

Purpose of the Study:

  • To develop a novel algorithm for designing optimal output codes in multiclass learning.
  • To improve the performance of multiclass classifiers by optimizing the decomposition process.
  • To address the limitations of prior decomposition methods in influencing classifier learning.

Main Methods:

  • A new algorithm for output code design in multiclass learning problems is presented.

Related Experiment Videos

  • The algorithm employs a maximum-likelihood objective function.
  • It integrates the expectation-maximization (EM) algorithm to minimize an augmented objective function.

    • Main Results:

    • The proposed method optimizes the decomposition of multiclass problems into binary (two-class) problems.
    • Experimental results demonstrate the effectiveness of the optimized output codes.
    • The optimized codes show potential gains over traditional OPC and ECOC methods.

    Conclusions:

    • The novel algorithm offers a superior approach to output code design for multiclass learning.
    • Optimized output codes lead to improved classifier performance compared to standard methods.
    • This work provides a more effective strategy for decomposing complex multiclass learning tasks.