C# Class Encog.Neural.Networks.Training.Strategy.ResetStrategy

The reset strategy will reset the weights if the neural network fails to fall below a specified error by a specified number of cycles. This can be useful to throw out initially "bad/hard" random initializations of the weight matrix.

Inheritance: IStrategy

Mostra file Open project: encog/encog-silverlight-core

Public Methods

Method	Description
Init ( ITrain train ) : void	Initialize this strategy.
PostIteration ( ) : void	Called just after a training iteration.
PreIteration ( ) : void	Called just before a training iteration.
ResetStrategy ( double required, int cycles ) : System	Construct a reset strategy. The error rate must fall below the required rate in the specified number of cycles, or the neural network will be reset to random weights and thresholds.

Method Details

Init() public method

Initialize this strategy.

public Init ( ITrain train ) : void
train	ITrain	The training algorithm.
return	void

PostIteration() public method

Called just after a training iteration.

public PostIteration ( ) : void
return	void

PreIteration() public method

Called just before a training iteration.

public PreIteration ( ) : void
return	void

ResetStrategy() public method

Construct a reset strategy. The error rate must fall below the required rate in the specified number of cycles, or the neural network will be reset to random weights and thresholds.

public ResetStrategy ( double required, int cycles ) : System
required	double	The required error rate.
cycles	int	The number of cycles to reach that rate.
return	System