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

A new discriminative kernel from probabilistic models.

Koji Tsuda1, Motoaki Kawanabe, Gunnar Rätsch

  • 1AIST Computational Biology Research Center, Koto-ku, Tokyo, 135-0064, Japan. koji.tsuda@aist.go.jp

Neural Computation
|October 25, 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

Towards robust foundation models for digital pathology.

Nature communications·2026
Same author

Beyond attention heatmaps: How to get better explanations for multiple instance learning models in histopathology.

Medical image analysis·2026
Same author

Representation learning for multi-modal spatially resolved transcriptomics data.

Bioinformatics (Oxford, England)·2026
Same author

Estimation of Physiological Metrics from Resting ECGs Using Deep Learning in the UK Biobank, Including submaximal exercise derived V̇O <sub>2</sub> max, Body Fat Percentage, and Grip Strength.

medRxiv : the preprint server for health sciences·2026
Same author

AI-based discovery of functional boundaries in the human brain from intraoperative electrophysiology.

medRxiv : the preprint server for health sciences·2026
Same author

Modeling attention and binding in the brain through bidirectional recurrent gating.

Nature communications·2026
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
See all related articles

Researchers developed a new TOP kernel, derived from posterior log-odds, outperforming the existing Fisher kernel for machine learning tasks like DNA and protein analysis. This kernel offers a novel approach to feature extraction from probabilistic models.

Area of Science:

  • Machine Learning
  • Bioinformatics
  • Computational Biology

Background:

  • Kernel functions are crucial for machine learning, enabling algorithms like Support Vector Machines (SVMs) to analyze complex data.
  • The Fisher kernel, proposed by Jaakkola and Haussler (1999), constructs kernels from probabilistic models using marginal log-likelihood, showing success in biological data analysis.
  • There is a need for advanced kernel methods that can capture more nuanced information from probabilistic models for improved discriminative power.

Purpose of the Study:

  • To introduce a novel kernel function, the TOP kernel, derived from tangent vectors of posterior log-odds.
  • To develop a theoretical framework for analyzing feature extractors from probabilistic models.
  • To evaluate the performance of the TOP kernel against the established Fisher kernel in experimental settings.

Related Experiment Videos

Main Methods:

  • The TOP kernel is derived from tangent vectors of posterior log-odds, contrasting with the Fisher kernel's reliance on marginal log-likelihood.
  • A theoretical framework for feature extraction from probabilistic models was developed to analyze the properties of the TOP kernel.
  • The TOP kernel was implemented and compared with the Fisher kernel using discriminative classifiers, likely Support Vector Machines, on relevant datasets.

Main Results:

  • The TOP kernel demonstrated favorable performance compared to the Fisher kernel in experimental evaluations.
  • The theoretical framework provided insights into the nature of feature extraction from probabilistic models.
  • The study confirms the potential of the TOP kernel as a powerful tool in machine learning applications, particularly in bioinformatics.

Conclusions:

  • The TOP kernel represents a significant advancement in kernel function construction from probabilistic models.
  • The proposed method offers a complementary and potentially superior alternative to the Fisher kernel for tasks involving biological sequence analysis.
  • Further research can explore the broader applicability of the TOP kernel and the theoretical framework across different domains.