Restricted Boltzmann Machines

GPTKB entity

Statements (50)
Predicate Object
gptkbp:instanceOf gptkb:convolutional_neural_network
gptkb:Probabilistic_Graphical_Model
gptkbp:activatedBy gptkb:Sigmoid
gptkbp:canBe gptkb:Speech_Recognition
gptkb:Collaborative_Filtering
Image Recognition
Topic Modeling
Anomaly Detection
Data Generation
Data Reconstruction
Pretraining Neural Networks
gptkbp:canBeTrainedBy gptkb:Persistent_Contrastive_Divergence
Stochastic Gradient Descent
gptkbp:developedBy gptkb:Geoffrey_Hinton
gptkbp:energyFunction Yes
gptkbp:extendsTo gptkb:Gaussian-Bernoulli_RBM
gptkb:Replicated_Softmax_RBM
gptkbp:field gptkb:Machine_Learning
gptkb:artificial_intelligence
Deep Learning
gptkbp:hasConnectionsBetween Visible and Hidden Layers
gptkbp:hasHiddenLayer Yes
gptkbp:hasNoConnectionsBetween Units in Same Layer
gptkbp:hasVisibleLayer Yes
https://www.w3.org/2000/01/rdf-schema#label Restricted Boltzmann Machines
gptkbp:input Binary
gptkbp:introducedIn 1986
gptkbp:limitation Approximate Inference Required
Difficult to Train for Large Datasets
Limited Scalability
Sensitive to Hyperparameters
gptkbp:output Binary
gptkbp:parameter Weights
Biases
gptkbp:popularizedBy 2006
gptkbp:relatedTo gptkb:Hopfield_Network
gptkb:Boltzmann_Machine
gptkb:Markov_Random_Field
gptkb:Deep_Neural_Network
Boltzmann machine
gptkbp:stackable Yes
gptkbp:trainingAlgorithm gptkb:Contrastive_Divergence
gptkbp:usedFor gptkb:Collaborative_Filtering
Classification
Dimensionality Reduction
Feature Learning
Pretraining Deep Networks
gptkbp:usedIn gptkb:Deep_Belief_Networks
gptkbp:bfsParent gptkb:A_Fast_Learning_Algorithm_for_Deep_Belief_Nets_(Hinton_et_al.,_2006)
gptkbp:bfsLayer 7