C# Class Encog.Neural.Networks.Training.Competitive.CompetitiveTraining

This class implements competitive training, which would be used in a winner-take-all neural network, such as the self organizing map (SOM). This is an unsupervised training method, no ideal data is needed on the training set. If ideal data is provided, it will be ignored. Training is done by looping over all of the training elements and calculating a "best matching unit" (BMU). This BMU output neuron is then adjusted to better "learn" this pattern. Additionally, this training may be applied to othr "nearby" output neurons. The degree to which nearby neurons are update is defined by the neighborhood function. A neighborhood function is required to determine the degree to which neighboring neurons (to the winning neuron) are updated by each training iteration. Because this is unsupervised training, calculating an error to measure progress by is difficult. The error is defined to be the "worst", or longest, Euclidean distance of any of the BMU's. This value should be minimized, as learning progresses. Because only the BMU neuron and its close neighbors are updated, you can end up with some output neurons that learn nothing. By default these neurons are forced to win patterns that are not represented well. This spreads out the workload among all output neurons. This feature is used by default, but can be disabled by setting the "forceWinner" property.

Inheritance: BasicTraining, ILearningRate

Show file Open project: encog/encog-silverlight-core Class Usage Examples

Public Methods

Method	Description
AutoDecay ( ) : void	Should be called each iteration if autodecay is desired.
CompetitiveTraining ( BasicNetwork network, double learningRate, INeuralDataSet training, INeighborhoodFunction neighborhood ) : log4net	Create an instance of competitive training.
Decay ( double d ) : void	Called to decay the learning rate and radius by the specified amount.
Decay ( double decayRate, double decayRadius ) : void	Decay the learning rate and radius by the specified amount.
Iteration ( ) : void	Perform one training iteration.
SetAutoDecay ( int plannedIterations, double startRate, double endRate, double startRadius, double endRadius ) : void	Setup autodecay. This will decrease the radius and learning rate from the start values to the end values.
SetParams ( double rate, double radius ) : void	Set the learning rate and radius.
ToString ( ) : String	Returns this object as a string.
TrainPattern ( INeuralData pattern ) : void	Train the specified pattern. Find a winning neuron and adjust all neurons according to the neighborhood function.

Private Methods

Method	Description
ApplyCorrection ( ) : void	Loop over the synapses to be trained and apply any corrections that were determined by this training iteration.
CopyInputPattern ( ISynapse synapse, int outputNeuron, INeuralData input ) : void	Copy the specified input pattern to the weight matrix. This causes an output neuron to learn this pattern "exactly". This is useful when a winner is to be forced.
DetermineNewWeight ( double weight, double input, int currentNeuron, int bmu ) : double	Determine the weight adjustment for a single neuron during a training iteration.
ForceWinners ( ISynapse synapse, int won, INeuralData leastRepresented ) : bool	Force any neurons that did not win to off-load patterns from overworked neurons.
Train ( int bmu, ISynapse synapse, INeuralData input ) : void	Train for the specified synapse and BMU.
TrainPattern ( ISynapse synapse, INeuralData input, int current, int bmu ) : void	Train for the specified pattern.

Method Details

AutoDecay() public method

Should be called each iteration if autodecay is desired.

public AutoDecay ( ) : void
return	void

CompetitiveTraining() public method

Create an instance of competitive training.

public CompetitiveTraining ( BasicNetwork network, double learningRate, INeuralDataSet training, INeighborhoodFunction neighborhood ) : log4net
network	BasicNetwork	The network to train.
learningRate	double	The learning rate, how much to apply per iteration.
training	INeuralDataSet	The training set (unsupervised).
neighborhood	INeighborhoodFunction	The neighborhood function to use.
return	log4net

Decay() public method

Called to decay the learning rate and radius by the specified amount.

public Decay ( double d ) : void
d	double	The percent to decay by.
return	void

Decay() public method

Decay the learning rate and radius by the specified amount.

public Decay ( double decayRate, double decayRadius ) : void
decayRate	double	The percent to decay the learning rate by.
decayRadius	double	The percent to decay the radius by.
return	void

Iteration() public method

Perform one training iteration.

public Iteration ( ) : void
return	void

SetAutoDecay() public method

Setup autodecay. This will decrease the radius and learning rate from the start values to the end values.

public SetAutoDecay ( int plannedIterations, double startRate, double endRate, double startRadius, double endRadius ) : void
plannedIterations	int	The number of iterations that are planned. /// This allows the decay rate to be determined.
startRate	double	The starting learning rate.
endRate	double	The ending learning rate.
startRadius	double	The starting radius.
endRadius	double	The ending radius.
return	void

SetParams() public method

Set the learning rate and radius.

public SetParams ( double rate, double radius ) : void
rate	double	The new learning rate.
radius	double	The new radius.
return	void

ToString() public method

Returns this object as a string.

public ToString ( ) : String
return	String

TrainPattern() public method

Train the specified pattern. Find a winning neuron and adjust all neurons according to the neighborhood function.

public TrainPattern ( INeuralData pattern ) : void
pattern	INeuralData	The pattern to train.
return	void