org.encog.neural.networks.training.concurrent.jobs
Class BPROPJob

java.lang.Object
  org.encog.neural.networks.training.concurrent.jobs.TrainingJob
      org.encog.neural.networks.training.concurrent.jobs.BPROPJob

public class BPROPJob
extends TrainingJob

A training definition for BPROP training.

Constructor Summary

BPROPJob(BasicNetwork network, NeuralDataSet training, boolean loadToMemory, double learningRate, double momentum)
Construct a job definition for RPROP.

BPROPJob(BasicNetwork network, NeuralDataSet training, boolean loadToMemory, double learningRate, double momentum, double localRatio, int globalRatio, double segmentationRatio, int iterationsPer)
Construct a job definition for RPROP.

Method Summary

void	createTrainer(OpenCLTrainingProfile profile, boolean singleThreaded) Create a trainer to use.
double	getLearningRate()
double	getMomentum()
void	setLearningRate(double learningRate)
void	setMomentum(double momentum)

Methods inherited from class org.encog.neural.networks.training.concurrent.jobs.TrainingJob

getError, getGlobalRatio, getIterationsPer, getLocalRatio, getNetwork, getSegmentationRatio, getStrategies, getTrain, getTraining, isLoadToMemory, setError, setGlobalRatio, setIterationsPer, setLoadToMemory, setLocalRatio, setNetwork, setSegmentationRatio, setTrain, setTraining, shouldContinue

Methods inherited from class java.lang.Object

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

Constructor Detail

BPROPJob

public BPROPJob(BasicNetwork network,
                NeuralDataSet training,
                boolean loadToMemory,
                double learningRate,
                double momentum)

Construct a job definition for RPROP. For more information on backprop, see the Backpropagation class. Use OpenCLratio of 1.0 and process one iteration per cycle.

Parameters:

network - The network to use.

training - The training data to use.

loadToMemory - Should binary data be loaded to memory?

learningRate - THe learning rate to use.

momentum - The momentum to use.

BPROPJob

public BPROPJob(BasicNetwork network,
                NeuralDataSet training,
                boolean loadToMemory,
                double learningRate,
                double momentum,
                double localRatio,
                int globalRatio,
                double segmentationRatio,
                int iterationsPer)

Construct a job definition for RPROP. For more information on backprop, see the Backpropagation class.

Parameters:

network - The network to use.

training - The training data to use.

loadToMemory - Should binary data be loaded to memory?

learningRate - THe learning rate to use.

momentum - The momentum to use.

localRatio - The local ratio, used if this job is performed by an OpenCL Device.

globalRatio - The global ratio, used if this job is performed by an OpenCL Device.

segmentationRatio - The segmentation ratio, used if this job is performed by an OpenCL Device.

iterationsPer - How many iterations to process per cycle.

Method Detail

createTrainer

public void createTrainer(OpenCLTrainingProfile profile,
                          boolean singleThreaded)

Create a trainer to use.

Specified by:

createTrainer in class TrainingJob

Parameters:

profile - The OpenCL training profile to use.

getLearningRate

public double getLearningRate()

Returns:

the learningRate

getMomentum

public double getMomentum()

Returns:

the momentum

setLearningRate

public void setLearningRate(double learningRate)

Parameters:

learningRate - the learningRate to set

setMomentum

public void setMomentum(double momentum)

Parameters:

momentum - the momentum to set