net.sourceforge.cilib.neuralnetwork.generic.trainingstrategies
Class FFNN_GD_TrainingStrategy
java.lang.Object
net.sourceforge.cilib.neuralnetwork.generic.trainingstrategies.FFNN_GD_TrainingStrategy
- All Implemented Interfaces:
- Initializable, TrainingStrategy, Observer
public class FFNN_GD_TrainingStrategy
- extends Object
- implements TrainingStrategy, Observer
- Author:
- stefanv
This is an Application in the Generic Framework - it does not form part of the framework.
It implements multi layer Gradient Decent (GD) training for Feedforward NN (FFNN).
To use the class, call the invokeTrainer() method with the last presented
pattern as an argument.
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
FFNN_GD_TrainingStrategy
public FFNN_GD_TrainingStrategy()
initialize
public void initialize()
- Specified by:
initialize
in interface Initializable
updateFanInWeights
public void updateFanInWeights(NeuronConfig n,
Type delta)
invokeTrainer
public void invokeTrainer(Object args)
- Specified by:
invokeTrainer
in interface TrainingStrategy
validate
public void validate()
- Specified by:
validate
in interface Observer
preEpochActions
public void preEpochActions(Object args)
- Specified by:
preEpochActions
in interface TrainingStrategy
postEpochActions
public void postEpochActions(Object args)
- Specified by:
postEpochActions
in interface TrainingStrategy
setDelta
public void setDelta(ErrorSignal delta)
setLearningRate
public void setLearningRate(Double learningRate)
setMomentum
public void setMomentum(Double momentum)
setTopology
public void setTopology(NeuralNetworkTopology topo)
- Specified by:
setTopology
in interface TrainingStrategy
Copyright © 2009 CIRG. All Rights Reserved.