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 fast learning algorithm for deep belief nets.

Geoffrey E Hinton1, Simon Osindero, Yee-Whye Teh

  • 1Department of Computer Science, University of Toronto, Canada. hinton@cs.toronto.edu

Neural Computation
|June 13, 2006
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

Adaptive Mixtures of Local Experts.

Neural computation·2019
Same author

Modeling natural images using gated MRFs.

IEEE transactions on pattern analysis and machine intelligence·2013
Same author

Machine learning for neuroscience.

Neural systems & circuits·2012
Same author

Unsupervised discovery of nonlinear structure using contrastive backpropagation.

Cognitive science·2011
Same author

Comparing classification methods for longitudinal fMRI studies.

Neural computation·2010
Same author

Learning to represent spatial transformations with factored higher-order Boltzmann machines.

Neural computation·2010
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 complementary priors to simplify inference in deep belief networks. This enables a fast, greedy learning algorithm for deep directed belief networks, improving generative models and digit classification accuracy.

Area of Science:

  • Artificial Intelligence
  • Machine Learning
  • Deep Learning

Background:

  • Inference in densely connected belief networks with many hidden layers is challenging due to explaining-away effects.
  • Existing methods struggle with the complexity of deep, layered network architectures.

Purpose of the Study:

  • To introduce complementary priors as a method to eliminate explaining-away effects.
  • To develop a fast, greedy algorithm for learning deep directed belief networks.
  • To enhance generative modeling and digit classification performance.

Main Methods:

  • Utilized complementary priors to derive a fast, greedy learning algorithm.
  • Trained deep directed belief networks layer by layer, with the top two layers forming an undirected associative memory.

Related Experiment Videos

  • Employed a contrastive wake-sleep algorithm for fine-tuning network weights.

    • Main Results:

    • A three-hidden-layer network achieved a highly effective generative model of handwritten digit images and labels.
    • The generative model outperformed leading discriminative learning algorithms in digit classification.
    • Identified low-dimensional digit manifolds as ravines in the free-energy landscape of the associative memory.

    Conclusions:

    • Complementary priors offer a significant advancement for learning deep belief networks.
    • The developed greedy algorithm and fine-tuning process enable superior generative modeling.
    • The approach provides a novel way to understand and visualize learned representations in deep networks.