C# 클래스 Accord.Neuro.Networks.DeepBeliefNetwork

상속: ActivationNetwork
파일 보기 프로젝트 열기: accord-net/framework 1 사용 예제들

공개 메소드들

메소드 설명
Compute ( double input ) : double[]

Computes the network's outputs for a given input.

Compute ( double input, int layerIndex ) : double[]

Computes the network's outputs for a given input.

CreateGaussianBernoulli ( int inputsCount ) : DeepBeliefNetwork

Creates a Gaussian-Bernoulli network.

CreateMixedNetwork ( IStochasticFunction visible, IStochasticFunction hidden, int inputsCount ) : DeepBeliefNetwork

Creates a Mixed-Bernoulli network.

DeepBeliefNetwork ( IStochasticFunction function, int inputsCount ) : System

Creates a new DeepBeliefNetwork.

DeepBeliefNetwork ( int inputsCount ) : System

Creates a new DeepBeliefNetwork.

GenerateInput ( double output ) : double[]

Samples an input vector from the network given an output vector.

GenerateOutput ( double input ) : double[]

Samples an output vector from the network given an input vector.

GenerateOutput ( double input, int layerIndex ) : double[]

Samples an output vector from the network given an input vector.

Load ( Stream stream ) : DeepBeliefNetwork

Loads a network from a stream.

Load ( string path ) : DeepBeliefNetwork

Loads a network from a file.

Pop ( ) : void

Removes the last layer from the network.

Push ( RestrictedBoltzmannMachine network ) : void

Stacks a new Boltzmann Machine at the end of this network.

Push ( int neurons ) : void

Inserts a new layer at the end of this network.

Push ( int neurons, IStochasticFunction function ) : void

Inserts a new layer at the end of this network.

Push ( int neurons, IStochasticFunction visibleFunction, IStochasticFunction hiddenFunction ) : void

Inserts a new layer at the end of this network.

Reconstruct ( double output ) : double[]

Reconstructs a input vector for a given output.

Reconstruct ( double output, int layerIndex ) : double[]

Reconstructs a input vector using the output vector of a given layer.

Save ( Stream stream ) : void

Saves the network to a stream.

Save ( string path ) : void

Saves the network to a stream.

UpdateVisibleWeights ( ) : void

Updates the weights of the visible layers by copying the reverse of the weights in the hidden layers.

메소드 상세

Compute() 공개 메소드

Computes the network's outputs for a given input.
public Compute ( double input ) : double[]
input double The input vector.
리턴 double[]

Compute() 공개 메소드

Computes the network's outputs for a given input.
public Compute ( double input, int layerIndex ) : double[]
input double The input vector.
layerIndex int The index of the layer.
리턴 double[]

CreateGaussianBernoulli() 공개 정적인 메소드

Creates a Gaussian-Bernoulli network.
public static CreateGaussianBernoulli ( int inputsCount ) : DeepBeliefNetwork
inputsCount int The number of inputs for the network.
리턴 DeepBeliefNetwork

CreateMixedNetwork() 공개 정적인 메소드

Creates a Mixed-Bernoulli network.
public static CreateMixedNetwork ( IStochasticFunction visible, IStochasticFunction hidden, int inputsCount ) : DeepBeliefNetwork
visible IStochasticFunction The to be used in the first visible layer.
hidden IStochasticFunction The to be used in all other layers.
inputsCount int The number of inputs for the network.
리턴 DeepBeliefNetwork

DeepBeliefNetwork() 공개 메소드

Creates a new DeepBeliefNetwork.
public DeepBeliefNetwork ( IStochasticFunction function, int inputsCount ) : System
function IStochasticFunction The activation function to be used in the network neurons.
inputsCount int The number of inputs for the network.
리턴 System

DeepBeliefNetwork() 공개 메소드

Creates a new DeepBeliefNetwork.
public DeepBeliefNetwork ( int inputsCount ) : System
inputsCount int The number of inputs for the network.
리턴 System

GenerateInput() 공개 메소드

Samples an input vector from the network given an output vector.
public GenerateInput ( double output ) : double[]
output double An output vector.
리턴 double[]

GenerateOutput() 공개 메소드

Samples an output vector from the network given an input vector.
public GenerateOutput ( double input ) : double[]
input double An input vector.
리턴 double[]

GenerateOutput() 공개 메소드

Samples an output vector from the network given an input vector.
public GenerateOutput ( double input, int layerIndex ) : double[]
input double An input vector.
layerIndex int The index of the layer.
리턴 double[]

Load() 공개 정적인 메소드

Loads a network from a stream.
public static Load ( Stream stream ) : DeepBeliefNetwork
stream Stream The network from which the machine is to be deserialized.
리턴 DeepBeliefNetwork

Load() 공개 정적인 메소드

Loads a network from a file.
public static Load ( string path ) : DeepBeliefNetwork
path string The path to the file from which the network is to be deserialized.
리턴 DeepBeliefNetwork

Pop() 공개 메소드

Removes the last layer from the network.
public Pop ( ) : void
리턴 void

Push() 공개 메소드

Stacks a new Boltzmann Machine at the end of this network.
public Push ( RestrictedBoltzmannMachine network ) : void
network RestrictedBoltzmannMachine The machine to be added to the network.
리턴 void

Push() 공개 메소드

Inserts a new layer at the end of this network.
public Push ( int neurons ) : void
neurons int The number of neurons in the new layer.
리턴 void

Push() 공개 메소드

Inserts a new layer at the end of this network.
public Push ( int neurons, IStochasticFunction function ) : void
neurons int The number of neurons in the new layer.
function IStochasticFunction The activation function which should be used by the neurons.
리턴 void

Push() 공개 메소드

Inserts a new layer at the end of this network.
public Push ( int neurons, IStochasticFunction visibleFunction, IStochasticFunction hiddenFunction ) : void
neurons int The number of neurons in the layer.
visibleFunction IStochasticFunction The activation function which should be used by the visible neurons.
hiddenFunction IStochasticFunction The activation function which should be used by the hidden neurons.
리턴 void

Reconstruct() 공개 메소드

Reconstructs a input vector for a given output.
public Reconstruct ( double output ) : double[]
output double The output vector.
리턴 double[]

Reconstruct() 공개 메소드

Reconstructs a input vector using the output vector of a given layer.
public Reconstruct ( double output, int layerIndex ) : double[]
output double The output vector.
layerIndex int The index of the layer.
리턴 double[]

Save() 공개 메소드

Saves the network to a stream.
public Save ( Stream stream ) : void
stream Stream The stream to which the network is to be serialized.
리턴 void

Save() 공개 메소드

Saves the network to a stream.
public Save ( string path ) : void
path string The file path to which the network is to be serialized.
리턴 void

UpdateVisibleWeights() 공개 메소드

Updates the weights of the visible layers by copying the reverse of the weights in the hidden layers.
public UpdateVisibleWeights ( ) : void
리턴 void