org.encog.neural.freeform.training
public class FreeformBackPropagation extends FreeformPropagationTraining implements Serializable
FLAT_SPOT_CONST
Constructor and Description |
---|
FreeformBackPropagation(FreeformNetwork theNetwork,
MLDataSet theTraining,
double theLearningRate,
double theMomentum)
Construct a back propagation trainer.
|
Modifier and Type | Method and Description |
---|---|
protected void |
learnConnection(FreeformConnection connection)
Learn for a single connection.
|
TrainingContinuation |
pause()
Pause the training to continue later.
|
void |
resume(TrainingContinuation state)
Resume training.
|
canContinue, finishTraining, getBatchSize, getError, getImplementationType, getIteration, getMethod, getTraining, isFixFlatSopt, iteration, iteration, learn, processBatches, processPureBatch, setBatchSize, setError, setFixFlatSopt, setIteration
addStrategy, getStrategies, isTrainingDone, postIteration, preIteration, setTraining
public FreeformBackPropagation(FreeformNetwork theNetwork, MLDataSet theTraining, double theLearningRate, double theMomentum)
theNetwork
- The network to train.theTraining
- The training data to use. The coefficient for how much of the gradient is applied to each weight.theLearningRate
- The learning rate. The coefficient for how much of the previous delta is applied to each weight.
In theory, prevents local minima stall.theMomentum
- The momentum.protected void learnConnection(FreeformConnection connection)
learnConnection
in class FreeformPropagationTraining
connection
- The connection to learn from.public TrainingContinuation pause()
public void resume(TrainingContinuation state)
Copyright © 2014. All Rights Reserved.