Java Train类代码示例

import org.encog.neural.networks.training.Train; //导入依赖的package包/类
@Test
public void trainAndStore() {
    BasicMLDataSet dataSet = getData();

    // Create network
    BasicNetwork network = getNetwork();

    // Train
    System.out.println("Training network...");
    Train train = new ResilientPropagation(network, dataSet);
    for (int i = 0; i < TRAIN_ITERATIONS; i++) {
        train.iteration();
    }
    System.out.println("Training finished, error: " + train.getError());

    // Save to file
    System.out.println("Saving to file...");
    saveToFile(network);
    System.out.println("Done");
}
 

import org.encog.neural.networks.training.Train; //导入依赖的package包/类
private Train getTrain (NeuralDataSet trainingSet, BasicNetwork network) {

		//final Train train =
			//new ManhattanPropagation(network, trainingSet,
			//0.001);
		
		// Train the neural network, we use resilient propagation
		final ResilientPropagation train = new ResilientPropagation(network, trainingSet);
		train.setThreadCount(0);
		// Reset if improve is less than 1% over 5 cycles
		train.addStrategy(new RequiredImprovementStrategy(DEFAULT_SELECTION_LIMIT));
		
		
		return train;
	}
 

import org.encog.neural.networks.training.Train; //导入依赖的package包/类
public static void test(double[][] inputValues, double[][] outputValues)
{
    NeuralDataSet trainingSet = new BasicNeuralDataSet(inputValues, outputValues);
    BasicNetwork network = new BasicNetwork();
    network.addLayer(new BasicLayer(new ActivationSigmoid(), false, 4));
    network.addLayer(new BasicLayer(new ActivationSigmoid(), false, 1000));
    network.addLayer(new BasicLayer(new ActivationLinear(), false, 1));
    network.getStructure().finalizeStructure();
    network.reset();
    final Train train = new ResilientPropagation(network, trainingSet);
    int epoch = 1;
    do
    {
        train.iteration();
        System.out.println("Epoch #" + epoch + " Error:" + train.getError());
        epoch++;
    }
    while(epoch < 10000);
    System.out.println("Neural Network Results:");
    for(MLDataPair pair : trainingSet)
    {
        final MLData output = network.compute(pair.getInput());
        System.out.println(pair.getInput().getData(0) + "," + pair.getInput().getData(1) + ", actual="
            + output.getData(0) + ",ideal=" + pair.getIdeal().getData(0));

    }
}
 

import org.encog.neural.networks.training.Train; //导入依赖的package包/类
private void trainOnModelStructureChange(double[][] x, double[][] y, Class<?> activation, BasicNetwork network, Train train) throws InstantiationException, IllegalAccessException {
	
       NeuralDataSet trainingSet = new BasicNeuralDataSet(x, y);
	
       if (network == null && train == null) {
       	network = buildNetwork(trainingSet, DEFAULT_CHANGE_HIDDEN_NO, x[0].length, activation);
   		train = getTrain(trainingSet, network);
       }
	
	int epoch = 1;
	long time = System.currentTimeMillis();
	double error[];
	double[] previousNetworkBestError = null;	
	do {
		train.iteration();
		error = getError(trainingSet, network); 
		logger.debug("Epoch #" + epoch + " Error:" + print(error));
		epoch++;
		if (isBetter(error,bestQuality) && 
				(bestEverError == null || 0.2 <= betterPercentage(error,bestEverError))) {
			bestEverNetwork = (BasicNetwork)network.clone();
			bestEverError = error;
			logger.debug("Epoch #" + epoch + " Best Ever Error:" + print(error));
			
               previousNetworkBestError = error;
			network = buildNetwork(trainingSet, currentNoOfHiddenNode + DEFAULT_CHANGE_HIDDEN_NO, x[0].length, activation);
			train = getTrain(trainingSet, network);
			time = System.currentTimeMillis();
			continue;
		}		
		
		// Change on hidden layer, retrain a new ANN
		if(System.currentTimeMillis() - time > DEFAULT_STRUCTURE_SELECTION_TIME_LIMIT ) {
			if (previousNetworkBestError != null && previousNetworkBestError.equals(bestEverError)) {					
				currentNoOfHiddenNode -= DEFAULT_CHANGE_HIDDEN_NO;					
				break;
			}
			
			previousNetworkBestError = error;
			
			network = buildNetwork(trainingSet, currentNoOfHiddenNode + DEFAULT_CHANGE_HIDDEN_NO, x[0].length, activation);
			train = getTrain(trainingSet, network);
			time = System.currentTimeMillis();
		}
		
	} while (true);
	

	logger.debug("Training Time: "+ (System.currentTimeMillis() - time));
	logger.debug("Current number of hidden node: "+ currentNoOfHiddenNode);
	calculateError(trainingSet);
}