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 comparative study on kernel-based probabilistic neural networks for speaker verification.

K K Yiu1, M W Mak, S Y Kung

  • 1Center for Multimedia Signal Processing, Department of Electronic and Information Engineering, The Hong Kong Polytechnic University, Hung Hom, Hong Kong, R.O.C. Michael.KKYiu@polyu.edu.hk

International Journal of Neural Systems
|November 9, 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

Adding learning to cellular genetic algorithms for training recurrent neural networks.

IEEE transactions on neural networks·2008
Same author

Probabilistic principal component subspaces: a hierarchical finite mixture model for data visualization.

IEEE transactions on neural networks·2008
Same author

Estimation of elliptical basis function parameters by the EM algorithm with application to speaker verification.

IEEE transactions on neural networks·2008
Same author

Multi-metric and multi-substructure biclustering analysis for gene expression data.

Proceedings. IEEE Computational Systems Bioinformatics Conference·2006
Same author

Accurate detection of aneuploidies in array CGH and gene expression microarray data.

Bioinformatics (Oxford, England)·2004
Same author

Analytical model of a three-dimensional integral image recording system that uses circular- and hexagonal-based spherical surface microlenses.

Journal of the Optical Society of America. A, Optics, image science, and vision·2001
Same journal

Latent Space Projections and Atlases, a Cautionary Tale in Deep Neuroimaging using Autoencoders.

International journal of neural systems·2026
Same journal

Transformer-Based Anomaly Detection for Neurodegenerative Screening in MRI Images.

International journal of neural systems·2026
Same journal

Discrete Wavelet Convolution for Learnable Time-Frequency Representation with Application to Seizure Prediction.

International journal of neural systems·2026
Same journal

Automatic Seizure Detection using Hierarchical Spectral-Temporal Feature Learning with an Imbalance-Aware Transformer.

International journal of neural systems·2026
Same journal

Pyramid Vision Transformer-Enhanced Conformer Network for Epileptic Seizure Recognition Using MultiChannel EEG Signals.

International journal of neural systems·2026
Same journal

A Time-Frequency Decoupled Contrastive Learning Framework for Electroencephalography-Based Parkinson's Disease Diagnosis.

International journal of neural systems·2026
See all related articles

Probabilistic decision-based neural networks (PDBNNs) and Gaussian mixture models (GMMs) show superior performance in speaker verification compared to elliptical basis function networks (EBFNs). PDBNNs offer more predictable performance due to their globally supervised learning approach.

Area of Science:

  • Speech processing
  • Machine learning
  • Biometrics

Background:

  • Speaker verification systems are crucial for security and access control.
  • Evaluating different machine learning models is essential for improving accuracy.
  • Kernel-based probabilistic neural networks offer a promising approach for pattern recognition tasks.

Purpose of the Study:

  • To compare the effectiveness of probabilistic decision-based neural networks (PDBNNs), Gaussian mixture models (GMMs), and elliptical basis function networks (EBFNs) for speaker verification.
  • To investigate modifications to PDBNN training algorithms for enhanced speaker verification performance.
  • To analyze the impact of different threshold determination methods on system predictability.

Main Methods:

  • Experimental evaluation of PDBNNs, GMMs, and EBFNs using the YOHO corpus with 138 speakers.

Related Experiment Videos

  • Modification of the PDBNN original training algorithm for speaker verification.
  • Comparison of equal error rates (EER) and analysis of false acceptance rates (FAR) variations.

    • Main Results:

    • PDBNNs and GMMs achieved a lower equal error rate (0.33%) compared to EBFNs (0.48%).
    • Globally supervised learning in PDBNNs resulted in lower and less variable false acceptance rates.
    • Ad-hoc threshold determination in EBFNs and GMMs led to significant variations in error rates.

    Conclusions:

    • PDBNN- and GMM-based speaker models demonstrate superior performance over EBFN models.
    • The globally supervised learning of PDBNNs enhances system predictability by stabilizing decision thresholds.
    • PDBNNs present a robust and predictable solution for speaker verification applications.