经过调试发现Neuroph 有问题，有问题，自己用代码训练NullPointerException，beforeEpoch

来源：互联网发布：springmvc源码包编辑：程序博客网时间：2024/04/29 23:57

经过调试发现Neuroph有问题，自己用代码训练NullPointerException，beforeEpoch

BackPropagation dp = new BackPropagation();

dp.setNeuralNetwork(neuralNetwork);

neuralNetwork.setLearningRule(dp);

dp.doOneLearningIteration(trainingSet);---》内部机制问题

Exception in thread"main" java.lang.NullPointerException

atorg.neuroph.core.learning.SupervisedLearning.beforeEpoch(SupervisedLearning.java:142)

atorg.neuroph.core.learning.IterativeLearning.doOneLearningIteration(IterativeLearning.java:242)

atGoodTrainingAI.traintest(GoodTrainingAI.java:49)

atGoodTrainingAI.main(GoodTrainingAI.java:20)

自定义beforeEpoch里面没有调用onstart，导致ErrorFunction没有初始化

也没有setDataSet

只能自己继承Backpropagation，在重写BeforeEpoch(){ 调用super.onStart(),super.beforeEpoch()},

注意在MyLearningRule dp调用之前，需要dp.setDataSet(dataSet)；

MyLearningRule.java

public class MyLearningRule extends BackPropagation {

protected void beforeEpoch() {

// TODO Auto-generatedmethod stub

super.onStart();

super.beforeEpoch();

}

Main.java调用

BackPropagation dp = new BackPropagation();

//MyLearningRuledp = new MyLearningRule();

dp.setTrainingSet(trainingSet);

dp.setLearningRate(0.5);

dp.setNeuralNetwork(neuralNetwork);

neuralNetwork.setLearningRule(dp);

intepoch = 1 ;

doubletotalNetError = 1;

{

dp.doOneLearningIteration(trainingSet);

totalNetError= dp.getTotalNetworkError() ;

BackPropagation dp = new BackPropagation();

dp.setNeuralNetwork(neuralNetwork);

neuralNetwork.setLearningRule(dp);

dp.doOneLearningIteration(trainingSet);---》内部机制问题

Exception in thread"main" java.lang.NullPointerException

atorg.neuroph.core.learning.SupervisedLearning.beforeEpoch(SupervisedLearning.java:142)

atorg.neuroph.core.learning.IterativeLearning.doOneLearningIteration(IterativeLearning.java:242)

atGoodTrainingAI.traintest(GoodTrainingAI.java:49)

atGoodTrainingAI.main(GoodTrainingAI.java:20)

自定义beforeEpoch里面没有调用onstart，导致ErrorFunction没有初始化

也没有setDataSet

只能自己继承Backpropagation，在重写BeforeEpoch(){ 调用super.onStart(),super.beforeEpoch()},

注意在MyLearningRule dp调用之前，需要dp.setDataSet(dataSet)；

MyLearningRule.java

public class MyLearningRule extends BackPropagation {

protected void beforeEpoch() {

// TODO Auto-generatedmethod stub

super.onStart();

super.beforeEpoch();

}

Main.java调用

BackPropagation dp = new BackPropagation();

//MyLearningRuledp = new MyLearningRule();

dp.setTrainingSet(trainingSet);

dp.setLearningRate(0.5);

dp.setNeuralNetwork(neuralNetwork);

neuralNetwork.setLearningRule(dp);

intepoch = 1 ;

doubletotalNetError = 1;

{

dp.doOneLearningIteration(trainingSet);

totalNetError= dp.getTotalNetworkError() ;

0 0