C# Class Encog.Neural.Flat.Train.Prop.TrainFlatNetworkSCG

Train a network using scaled conjugate gradient.

Inheritance: TrainFlatNetworkProp

显示文件 Open project: encog/encog-silverlight-core Class Usage Examples

Public Methods

Method	Description
CalculateGradients ( ) : void	Calculate the gradients. They are normalized as well.
InitOthers ( ) : void	Not needed for this training type.
Iteration ( ) : void	Perform one iteration.
TrainFlatNetworkSCG ( FlatNetwork network, IMLDataSet training ) : System	Construct the training object.
UpdateWeight ( double gradients, double lastGradient, int index ) : double	Update the weights.

Private Methods

Method	Description
Init ( ) : void	Calculate the starting set of gradients.

Method Details

CalculateGradients() public method

Calculate the gradients. They are normalized as well.

public CalculateGradients ( ) : void
return	void

InitOthers() public method

Not needed for this training type.

public InitOthers ( ) : void
return	void

Iteration() public method

Perform one iteration.

public Iteration ( ) : void
return	void

TrainFlatNetworkSCG() public method

Construct the training object.

public TrainFlatNetworkSCG ( FlatNetwork network, IMLDataSet training ) : System
network	FlatNetwork	The network to train.
training	IMLDataSet	The training data to use.
return	System

UpdateWeight() public method

Update the weights.

public UpdateWeight ( double gradients, double lastGradient, int index ) : double
gradients	double	The current gradients.
lastGradient	double	The last gradients.
index	int	The weight index being updated.
return	double