|
gptkbp:instanceOf
|
gptkb:model
gptkb:convolutional_neural_network
generative model
|
|
gptkbp:advantage
|
feature extraction
unsupervised pretraining
generative capabilities
|
|
gptkbp:architecture
|
layered
|
|
gptkbp:canBe
|
speech recognition
image recognition
dimensionality reduction
data generation
|
|
gptkbp:canBeFineTunedBy
|
backpropagation
|
|
gptkbp:category
|
gptkb:artificial_intelligence
gptkb:machine_learning
deep learning
|
|
gptkbp:citation
|
high
|
|
gptkbp:composedOf
|
stacked Restricted Boltzmann Machines
|
|
gptkbp:developedBy
|
gptkb:Geoffrey_Hinton
gptkb:Simon_Osindero
gptkb:Yee-Whye_Teh
|
|
gptkbp:hasHiddenLayers
|
multiple hidden layers
|
|
gptkbp:hasInputLayer
|
visible layer
|
|
gptkbp:hasOutputLayer
|
output layer
|
|
https://www.w3.org/2000/01/rdf-schema#label
|
Deep Belief Networks
|
|
gptkbp:implementedIn
|
gptkb:Python
gptkb:TensorFlow
gptkb:MATLAB
gptkb:PyTorch
|
|
gptkbp:influenced
|
deep learning research
development of deep architectures
|
|
gptkbp:introducedIn
|
2006
|
|
gptkbp:limitation
|
vanishing gradient problem
training complexity
difficulty scaling to very deep networks
|
|
gptkbp:notablePublication
|
gptkb:A_Fast_Learning_Algorithm_for_Deep_Belief_Nets
|
|
gptkbp:publishedIn
|
gptkb:Science_(journal)
|
|
gptkbp:relatedTo
|
gptkb:Deep_Neural_Networks
Boltzmann machine
Deep Learning
|
|
gptkbp:trainer
|
greedy layer-wise pretraining
|
|
gptkbp:type
|
unsupervised
supervised (after pretraining)
|
|
gptkbp:usedFor
|
gptkb:dictionary
dimensionality reduction
unsupervised learning
feature learning
generative modeling
|
|
gptkbp:bfsParent
|
gptkb:Representation_Learning
gptkb:Restricted_Boltzmann_Machines
|
|
gptkbp:bfsLayer
|
8
|