Neuroph感知机自我学习实现记忆逻辑与

1.创建增加学习算法的感知机

	/**
	 * @author Ragty
	 * @param  增加学习算法的感知机(记忆逻辑与)
	 * @serialData 2018.4.22
	 * @param inputNeuralCount
	 */
	public void creatPerceptron(int inputNeuralCount){
		
		//设置类型为感知机
		this.setNetworkType(NeuralNetworkType.PERCEPTRON);
		
		//建立输入神经元，表示输入刺激
		NeuronProperties inputNeuronProperties = new NeuronProperties();
		inputNeuronProperties.setProperty("neuronType",InputNeuron.class);
		
		//建立输入层
		Layer inputLayer = LayerFactory.createLayer(inputNeuralCount, inputNeuronProperties);
		this.addLayer(inputLayer);
		inputLayer.addNeuron(new BiasNeuron());
		
		//建立输出神经元（传输函数为step）
		NeuronProperties outputNeuronProperties = new NeuronProperties();
		outputNeuronProperties.setProperty("transferFunction", TransferFunctionType.STEP);
		
		//建立输出层
		Layer outputLayer = LayerFactory.createLayer(1, outputNeuronProperties);
		this.addLayer(outputLayer);
		
		//输入层输出层全连接
		ConnectionFactory.fullConnect(inputLayer, outputLayer);
		NeuralNetworkFactory.setDefaultIO(this);
		
		//设置感知机学习算法
		this.setLearningRule(new perceptronLearningRule());
	}

2.创建学习算法

public class perceptronLearningRule extends SupervisedLearning implements Serializable{

	private static final long serialVersionUID = 1L;

	public perceptronLearningRule() {

    }
	
	/**
	 * @author Ragty
	 * @param  迭代计算权值
	 * @serialData 2018.4.22
	 */
	@Override
    protected void updateNetworkWeights(double[] outputError) {
        int i = 0;
        for (Neuron neuron : neuralNetwork.getOutputNeurons()) {
            neuron.setError(outputError[i]); 
            double neuronError = neuron.getError();
            // 根据所有的神经元输入 迭代学习
            for (Connection connection : neuron.getInputConnections()) {
                // 神经元的一个输入
                double input = connection.getInput();
                // 计算权值的变更
                double weightChange =  neuronError * input;
                // 更新权值
                Weight weight = connection.getWeight();
                weight.weightChange = weightChange;                
                weight.value += weightChange;
            }

            i++;
        }
    }

	

}

3.训练数据并测试

public class AndPerceptron implements LearningEventListener{

	public static void main(String[] args) {
		new AndPerceptron().run();
	}
	
	public void run(){
		
	   //给出学习的训练数据(用于训练神经网络)
	   //数据集有两个输入，一个输出
	   //dataSetRow的构造函数接受两个参数，第一个为输入向量，第二个为期望值
	   DataSet trainningSet = new DataSet(2,1);	
	   trainningSet.addRow(new DataSetRow(new double[]{0,0},new double[]{0}));
	   trainningSet.addRow(new DataSetRow(new double[]{0,1},new double[]{0}));
	   trainningSet.addRow(new DataSetRow(new double[]{1,0},new double[]{0}));
	   trainningSet.addRow(new DataSetRow(new double[]{1,1},new double[]{1}));
	   
	   //创建一个只有两个输入节点的感知机
	   simplePerceptron andPerceptron = new simplePerceptron(2);
	   
	   //给学习过程增加事件监听器（监督训练）
	   perceptronLearningRule learningRule = (perceptronLearningRule) andPerceptron.getLearningRule();
	   learningRule.addListener(this);	
	   
	   //使用训练数据训练感知机（进行学习）
	   System.out.println("训练开始");
	   andPerceptron.learn(trainningSet);
	   
	   //测试感知机是否能正确输出
	   System.out.println("测试输出");
	   testNeuralNetwork(andPerceptron, trainningSet);
	}
	
	
	/**
	 * @author Ragty
	 * @param  训练之后对网络测试(测试感知机)
	 * @serialData 2018.4.22
	 * @param neuralNetwork
	 * @param data
	 */
	public static void testNeuralNetwork(NeuralNetwork neuralNetwork, DataSet testSet){
		
		for(DataSetRow testSetRow : testSet.getRows()){
			neuralNetwork.setInput(testSetRow.getInput());
			neuralNetwork.calculate();
			double[] networkOutput = neuralNetwork.getOutput();
			
			System.out.println("Input:"+Arrays.toString(testSetRow.getInput()));
			System.out.println("Output:"+Arrays.toString(networkOutput));
		}
		
	}

	
	//监督训练过程
	@Override
	public void handleLearningEvent(LearningEvent event) {
		// TODO Auto-generated method stub
		//所有迭代学习算法的基类, 它为它的所有子类提供迭代学习过程
		IterativeLearning bp = (IterativeLearning) event.getSource();
		System.out.println("iterate:"+bp.getCurrentIteration());
		System.out.println(Arrays.toString(bp.getNeuralNetwork().getWeights()));
	}
	
	
	
}

4.学习结果