de.htwdd.rosenkoenig.neuro.net.training
Class TeachingAlgorithm

java.lang.Object
  de.htwdd.rosenkoenig.neuro.net.training.TeachingAlgorithm

Direct Known Subclasses:

Backpropagation

public abstract class TeachingAlgorithm
extends java.lang.Object

Base class of algorithms to train feed forward neural nets. All training parameters (e.g. learning rate, momentum, ...) can be accessed via the property trainer.

Field Summary

protected Trainer

trainer All training parameters can be accessed via this property.

Constructor Summary

TeachingAlgorithm()

Method Summary

Trainer	getTrainer() Getter of the property trainer
void	setTrainer(Trainer trainer) Setter of the property trainer
abstract void	teachPatterns(java.util.List<Pattern> patterns) Concrete teaching algorithms have to implement this method.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

trainer

protected Trainer trainer

All training parameters can be accessed via this property.

Constructor Detail

TeachingAlgorithm

public TeachingAlgorithm()

Method Detail

getTrainer

public Trainer getTrainer()

Getter of the property trainer

Returns:

Returns the trainer.

setTrainer

public void setTrainer(Trainer trainer)

Setter of the property trainer

Parameters:

trainer - The trainer to set.

teachPatterns

public abstract void teachPatterns(java.util.List<Pattern> patterns)

Concrete teaching algorithms have to implement this method. teachPatterns should run one training cycle on the provided pattern set and make use of the teaching parameters provided by trainer.

Parameters:

patterns - set of training patterns