Refinements of Universal Approximation Results for Deep Belief Networks and Restricted Boltzmann Machines

DOI10.1162/NECO_a_00113zbMath1274.62128arXiv1005.1593WikidataQ47323275 ScholiaQ47323275MaRDI QIDQ3015455

Guido Montúfar, Nihat Ay

Publication date: 13 July 2011

Published in: Neural Computation (Search for Journal in Brave)

Full work available at URL: https://arxiv.org/abs/1005.1593

Mathematics Subject Classification ID

68T05: Learning and adaptive systems in artificial intelligence

62E17: Approximations to statistical distributions (nonasymptotic)

Related Items

Mixture decompositions of exponential families using a decomposition of their sample spaces, When Does a Mixture of Products Contain a Product of Mixtures?, Dimension of Marginals of Kronecker Product Models, An Infinite Restricted Boltzmann Machine, Universal Approximation Depth and Errors of Narrow Belief Networks with Discrete Units, Training restricted Boltzmann machines: an introduction, Exploiting layerwise convexity of rectifier networks with sign constrained weights, Hierarchical models as marginals of hierarchical models

Cites Work

• Representational Power of Restricted Boltzmann Machines and Deep Belief Networks
• Deep, Narrow Sigmoid Belief Networks Are Universal Approximators
• Deep Belief Networks Are Compact Universal Approximators
• A Fast Learning Algorithm for Deep Belief Nets
• The measure of the critical values of differentiable maps