org.neuroph.nnet.learning
Class CompetitiveLearning
java.lang.Object
org.neuroph.core.learning.LearningRule
org.neuroph.core.learning.IterativeLearning
org.neuroph.core.learning.UnsupervisedLearning
org.neuroph.nnet.learning.CompetitiveLearning
- All Implemented Interfaces:
- Serializable
public class CompetitiveLearning
- extends UnsupervisedLearning
Competitive learning rule.
- Author:
- Zoran Sevarac
- See Also:
- Serialized Form
Methods inherited from class org.neuroph.core.learning.IterativeLearning |
afterEpoch, beforeEpoch, doOneLearningIteration, getCurrentIteration, getLearningRate, isPausedLearning, learn, learn, onStart, pause, resume, setLearningRate, setMaxIterations |
Methods inherited from class java.lang.Object |
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait |
CompetitiveLearning
public CompetitiveLearning()
- Creates new instance of CompetitiveLearning
doLearningEpoch
public void doLearningEpoch(DataSet trainingSet)
- This method does one learning epoch for the unsupervised learning rules.
It iterates through the training set and trains network weights for each
element. Stops learning after one epoch.
- Overrides:
doLearningEpoch
in class UnsupervisedLearning
- Parameters:
trainingSet
- training set for training network
updateNetworkWeights
protected void updateNetworkWeights()
- Adjusts weights for the winning neuron
- Specified by:
updateNetworkWeights
in class UnsupervisedLearning
Copyright © 2012. All Rights Reserved.