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

Deterministic annealing EM algorithm.

N Ueda1, R Nakano

  • 1NITT Communication Science Laboratories, Hikaridai, Seika-cho, Soraku-gun, Kyoto 619-02, Japan.

Neural Networks : the Official Journal of the International Neural Network Society
|March 29, 2003
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

Effects of an anti-angiogenic agent, TNP-470, on the growth of oral squamous cell carcinomas.

Oral oncology·2000
Same author

Clinical usefulness of telomerase assay for the detection of lymph node metastasis in patients with oral malignancy.

Archives of pathology & laboratory medicine·2000
Same author

[A case of malignant fibrous histiocytoma of mesocolon successfully resected after combined chemotherapy with epirubicin, CDDP and vincristine].

Gan to kagaku ryoho. Cancer & chemotherapy·2000
Same author

Tubular cell damage in acute renal failure-apoptosis, necrosis, or both.

Nephrology, dialysis, transplantation : official publication of the European Dialysis and Transplant Association - European Renal Association·2000
Same author

Antiplatelet drugs attenuate progression of carotid intima-media thickness in subjects with type 2 diabetes.

Thrombosis research·2000
Same author

Fatty acid cyclooxygenase induction and prostaglandin D synthesis in a human megakaryoblastic cell line CMK differentiated by phorbol ester.

Advances in experimental medicine and biology·2000
Same journal

Exploiting audio-visual modalities in videos: Object detection via multi-stage bilateral coupling network.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Reliability-aware modality completion with cross-modal distillation for federated learning with missing modalities.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

IGFD-Net: Illumination-guided frequency decoupling for polarization image fusion.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Multiple-Strategies dung beetle optimizer and its applications in engineering optimization and bankruptcy prediction.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Aggregating global-scale pixel-wise forgery cues within a graph.

Neural networks : the official journal of the International Neural Network Society·2026
Same journal

Finite-Time intermittent control for secure synchronization of Neutral-Type stochastic delayed neural networks under aperiodic DoS attacks.

Neural networks : the official journal of the International Neural Network Society·2026
See all related articles

This study introduces a deterministic annealing EM (DAEM) algorithm to improve maximum likelihood estimation. The DAEM algorithm enhances accuracy and efficiency, overcoming local maxima issues common in conventional methods.

Area of Science:

  • Machine Learning
  • Statistical Modeling
  • Computational Statistics

Background:

  • Conventional Expectation-Maximization (EM) algorithms often converge to local maxima, limiting estimation accuracy.
  • Maximum likelihood estimation (MLE) is crucial for parameter estimation in statistical models.
  • Probabilistic neural networks (PNNs) require robust density estimation for effective training.

Purpose of the Study:

  • To present a novel Deterministic Annealing EM (DAEM) algorithm for MLE problems.
  • To address and overcome the local maxima problem inherent in traditional EM algorithms.
  • To demonstrate the DAEM algorithm's effectiveness in training PNNs for probability density estimation.

Main Methods:

  • Derivation of a new posterior distribution parameterized by 'temperature' using maximum entropy principle.

Related Experiment Videos

  • Reformulation of the EM process as a thermodynamic free energy minimization problem via statistical mechanics analogy.
  • Deterministic minimization at each temperature step for efficient search compared to simulated annealing.

    • Main Results:

    • The DAEM algorithm provides better estimates, independent of initial parameter values.
    • DAEM demonstrates superior efficiency in search execution compared to simulated annealing.
    • Successful application of DAEM to train PNNs for accurate probability density estimation.

    Conclusions:

    • The DAEM algorithm offers a robust solution to the local maxima problem in MLE.
    • DAEM provides a more efficient and accurate estimation method, especially for complex models like PNNs.
    • This approach enhances the reliability and performance of statistical modeling and machine learning tasks.