345449.vhj5l3oj7.asia

Test_Hopfield.class

2000-06-20T16:00:00Z

Hopfield.class

2000-06-19T16:00:00Z

Test_Hopfield.java

2000-06-19T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */

public class Test_Hopfield {
 static float [] data [] = {
 {1, 1, 1, -1, -1, -1, -1, -1, -1, -1},
 {-1, -1, -1, 1, 1, 1, -1, -1, -1, -1},
 {-1, -1, -1, -1, -1, -1, -1, 1, 1, 1}
 };
 static Hopfield test;
 public static void main(String[] args) {
 test = new Hopfield(10);
 test.addTrainingData(data[0]);
 test.addTrainingData(data[1]);
 test.addTrainingData(data[2]);
 test.train();
 helper(test, "pattern 0", data[0]);
 helper(test, "pattern 1", data[1]);
 helper(test, "pattern 2", data[2]);
 }
 private static void helper(Hopfield test, String s, float [] test_data) {
 float [] dd = new float[10];
 for (int i=0; i<10; i++) {
 dd[i] = test_data[i];
 }
 int index = (int)(9.0f * (float)Math.random());
 if (dd[index] < 0.0f) dd[index] = 1.0f;
 else dd[index] = -1.0f;
 float [] rr = test.recall(dd, 5);
 System.out.println(s);
 for (int i=0; i<10; i++) System.out.print(pp(rr[i]) + " ");
 System.out.println();
 }
 private static int pp(float x) {
 if (x > 0.1f) return 1;
 return 0;
 }
}

misc.dfPackage

2000-06-19T16:00:00Z

Hopfield.java

2000-06-18T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */

import java.util.*;

public class Hopfield {

public Hopfield(int numInputs) {
 this.numInputs = numInputs;
 weights = new float[numInputs][numInputs];
 inputCells = new float[numInputs];
 tempStorage = new float[numInputs];
 }
 public void addTrainingData(float [] data) {
 trainingData.addElement(data);
 }
 public void train() {
 for (int j=1; j<numInputs; j++) {
 for (int i=0; i<j; i++) {
 for (int n=0; n<trainingData.size(); n++) {
 float [] data = (float [])trainingData.elementAt(n);
 float temp1 = adjustInput(data[i]) * adjustInput(data[j]);
 float temp = truncate(temp1 + weights[j][i]);
 weights[i][j] = weights[j][i] = temp;
 }
 }
 }
 for (int i=0; i<numInputs; i++) {
 tempStorage[i] = 0.0f;
 for (int j=0; j<i; j++) {
 tempStorage[i] += weights[i][j];
 }
 }
 }
 public float [] recall(float [] pattern, int numIterations) {
 for (int i=0; i<numInputs; i++) inputCells[i] = pattern[i];
 for (int ii = 0; ii<numIterations; ii++) {
 for (int i=0; i<numInputs; i++) {
 if (deltaEnergy(i) > 0.0f) {
 inputCells[i] = 1.0f;
 } else {
 inputCells[i] = 0.0f;
 }
 }
 }
 return inputCells;
 }
 private float adjustInput(float x) {
 if (x < 0.1f) return -1.0f;
 return 1.0f;
 }
 private float truncate(float x) {
 //return Math.round(x);
 int i = (int)x;
 return (float)i;
 }
 private float deltaEnergy(int index) {
 float temp = 0.0f;
 for (int j=0; j<numInputs; j++) {
 temp += weights[index][j] * inputCells[j];
 }
 return 2.0f * temp - tempStorage[index];
 }
 int numInputs;
 Vector trainingData = new Vector();
 float [][] weights;
 float [] tempStorage;
 float [] inputCells;
}

Neural_1H.java

2000-06-18T16:00:00Z

/**
 * Title: Neural_1H
 * Description: A one hidden layer back propagtion neural network model
 * Copyright: Copyright (c) 2000 Mark Watson. Artistic Open Source License
 * @author Mark Watson
 * @version 1.2
 */

import java.util.*;
import java.io.*;

class Neural_1H implements Serializable {

 protected int numInputs;
 protected int numHidden;
 protected int numOutputs;

 protected int numTraining;

 public float inputs[];
 protected float hidden[];
 public float outputs[];

 protected float W1[][];
 protected float W2[][];

 protected float output_errors[];
 protected float hidden_errors[];

 transient protected Vector inputTraining = new Vector();
 transient protected Vector outputTraining = new Vector();

 Neural_1H(int num_in, int num_hidden, int num_output) {
	numInputs = num_in;
	numHidden = num_hidden;
	numOutputs = num_output;
	inputs = new float[numInputs];
	hidden = new float[numHidden];
	outputs = new float[numOutputs];
	W1 = new float[numInputs][numHidden];
	W2 = new float[numHidden][numOutputs];
	randomizeWeights();

	output_errors = new float[numOutputs];
	hidden_errors = new float[numHidden];
 }
 public void addTrainingExample(float [] inputs, float [] outputs) {
	if (inputs.length != numInputs || outputs.length != numOutputs) {
	 System.out.println("addTrainingExample(): array size is wrong");
	 return;
	}
	inputTraining.addElement(inputs);
	outputTraining.addElement(outputs);
 }

 public static Neural_1H Factory(String serialized_file_name) {
	Neural_1H nn = null;
	try {
	 InputStream ins = ClassLoader.getSystemResourceAsStream(serialized_file_name);
	 if (ins==null) {
		System.out.println("CachedExamples(): failed to open 'cache.dat' in JAR file");
		System.exit(1);
	 } else {
		ObjectInputStream p = new ObjectInputStream(ins);
		nn = (Neural_1H)p.readObject();
 nn.inputTraining = new Vector();
 nn.outputTraining = new Vector();
		ins.close();
	 }
	} catch (Exception e) {
	 e.printStackTrace();
	 return null;
	}
	return nn;
 }

 public void save(String file_name) {
	try {
	 FileOutputStream ostream = new FileOutputStream(file_name);
	 ObjectOutputStream p = new ObjectOutputStream(ostream);
	 p.writeObject(this);
	 p.flush();
	 ostream.close();
	} catch (Exception e) {
	 e.printStackTrace();
	}
 }

 public void randomizeWeights() {
 // Randomize weights here:
 for (int ii=0; ii<numInputs; ii++)
 for (int hh=0; hh<numHidden; hh++)
 W1[ii][hh] =
 0.1f * (float)Math.random() - 0.05f;
 for (int hh=0; hh<numHidden; hh++)
 for (int oo=0; oo<numOutputs; oo++)
 W2[hh][oo] =
 0.1f * (float)Math.random() - 0.05f;
 }

 public float [] recall(float [] in) {
 for (int i=0; i<numInputs; i++) inputs[i] = in[i];
 forwardPass();
 float [] ret = new float[numOutputs];
 for (int i=0; i<numOutputs; i++) ret[i] = outputs[i];
 return ret;
 }

 public void forwardPass() {
 int i, h, o;
 for (h=0; h<numHidden; h++) {
 hidden[h] = 0.0f;
 }
 for (i=0; i<numInputs; i++) {
 for (h=0; h<numHidden; h++) {
 hidden[h] +=
 inputs[i] * W1[i][h];
 }
 }
 for (o=0; o<numOutputs; o++)
 outputs[o] = 0.0f;
 for (h=0; h<numHidden; h++) {
 for (o=0; o<numOutputs; o++) {
 outputs[o] +=
 sigmoid(hidden[h]) * W2[h][o];
 }
 }
 for (o=0; o<numOutputs; o++)
 outputs[o] = sigmoid(outputs[o]);
 }

 public float train() {
 return train(inputTraining, outputTraining);
 }

 // for debug graphics: train only one example at a time:
 private int current_example = 0;
 public float train(Vector v_ins, Vector v_outs) {
 int i, h, o;
 float error = 0.0f;
 int num_cases = v_ins.size();
 //for (int example=0; example<num_cases; example++) {
 // zero out error arrays:
 for (h=0; h<numHidden; h++)
 hidden_errors[h] = 0.0f;
 for (o=0; o<numOutputs; o++)
 output_errors[o] = 0.0f;
 // copy the input values:
 for (i=0; i<numInputs; i++) {
 inputs[i] = ((float [])v_ins.elementAt(current_example))[i];
 }
 // copy the ouytput values:
 float [] outs = (float [])v_outs.elementAt(current_example);

 // perform a forward pass through the network:

 forwardPass();

 for (o=0; o<numOutputs; o++) {
 output_errors[o] =
 (outs[o] -
 outputs[o])
 *sigmoidP(outputs[o]);
 }
 for (h=0; h<numHidden; h++) {
 hidden_errors[h] = 0.0f;
 for (o=0; o<numOutputs; o++) {
 hidden_errors[h] +=
 output_errors[o]*W2[h][o];
 }
 }
 for (h=0; h<numHidden; h++) {
 hidden_errors[h] =
 hidden_errors[h]*sigmoidP(hidden[h]);
 }
 // update the hidden to output weights:
 for (o=0; o<numOutputs; o++) {
 for (h=0; h<numHidden; h++) {
 W2[h][o] +=
 0.5 * output_errors[o] * hidden[h];
 }
 }
 // update the input to hidden weights:
 for (h=0; h<numHidden; h++) {
 for (i=0; i<numInputs; i++) {
 W1[i][h] +=
 0.5 * hidden_errors[h] * inputs[i];
 }
 }
 for (o=0; o<numOutputs; o++)
 error += Math.abs(output_errors[o]);
 //}
 current_example++;
 if (current_example >= num_cases) current_example = 0;
 return error;
 }

 protected float sigmoid(float x) {
 return
 (float)((1.0f/(1.0f+Math.exp((double)(-x))))-0.5f);
 }

 protected float sigmoidP(float x) {
 double z = sigmoid(x) + 0.5f;
 return (float)(z * (1.0f - z));
 }

}

Graph.java

2000-06-17T16:00:00Z

import javax.swing.*;
import java.awt.*;
/**
 * Title: Graph
 * Description: Simple program to produce a graph for the book
 * Copyright: Copyright (c) Mark Watson
 * Company: 
 * @author Mark Watson
 * @version 1.0
 */

public class Graph extends JFrame {
 GraphPanel jPanel1;
 float [] data1;
 float [] data2;
 public Graph() {
 try {
 int size = 500;
 data1 = new float[size];
 data2 = new float[size];
 float xmin = -5;
 float xmax = 5;
 for (int i=0; i<size; i++) {
 float x = i;
 x = xmin + x * (xmax - xmin) / (float)size;
 data1[i] = sigmoid(x);
 data2[i] = sigmoidP(x);
 }
 jbInit();
 }
 catch(Exception e) {
 e.printStackTrace();
 }
 }

protected float sigmoid(float x) {
    return
     (float)((1.0f/(1.0f+Math.exp((double)(-x))))-0.5f);
  }

protected float sigmoidP(float x) {
    double z = sigmoid(x) + 0.5f;
    return (float)(z * (1.0f - z));
  }

public static void main(String[] args) {
    Graph untitled11 = new Graph();
  }

private void jbInit() throws Exception {
    jPanel1 = new GraphPanel(data1, data2);
    jPanel1.setBackground(Color.white);
    this.setDefaultCloseOperation(3);
    this.getContentPane().add(jPanel1, BorderLayout.CENTER);
    setSize(550, 300);
    jPanel1.setVisible(true);
    setVisible(true);
  }
}

GraphPanel.java

2000-06-17T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */
import javax.swing.*;
import java.awt.*;

public class GraphPanel extends java.awt.Canvas { // JPanel {

public GraphPanel(float [] data1, float []data2) {
 super();
 this.data1 = data1;
 this.data2 = data2;
 }
 Color black = new Color(0, 0, 0);
 float [] data1;
 float [] data2;
 public void paint(Graphics g) {
 if (data1 == null || data2 == null) return;
 int width = this.getWidth();
 int height = this.getHeight();
 System.out.println("height="+height);
 float min = 99999999.9f;
 float max = -min;
 int maxindex = 0;
 float maxval = 0.0f;
 for (int i=0; i<data1.length; i++) { // assume length of data1 and data2 are the same
 if (min > data1[i]) min = data1[i];
 if (max < data1[i]) max = data1[i];
 if (min > data2[i]) min = data2[i];
 if (max < data2[i]) max = data2[i];
 }
 System.out.println("min=" + min +", max=" + max);
 g.setColor(Color.red);
 for (int i=0; i<data1.length - 1; i++) {
 float y1 = height - 5 - 0.95f *height * ((data1[i] - min) / (max - min));
 float y2 = height - 5 - 0.95f *height * ((data1[i+1] - min) / (max - min));
 //System.out.println("data["+i+"]="+data[i]+", y1="+y1+", y2="+y2);
 g.drawLine(i+20, (int)y1, i+21, (int)y2);
 y1 = height - 5 - 0.95f *height * ((data2[i] - min) / (max - min));
 y2 = height - 5 - 0.95f *height * ((data2[i+1] - min) / (max - min));
 //System.out.println("data["+i+"]="+data[i]+", y1="+y1+", y2="+y2);
 g.drawLine(i+20, (int)y1, i+21, (int)y2);
 }
 float yzero = height - 5 - 0.95f *height * ((0.0f - min) / (max - min));
 g.setColor(black);
 g.drawLine(20, (int)yzero, data2.length + 19, (int)yzero);
 g.drawLine(width / 2, height/2 - 118, width/2, height/2 + 118);
 g.drawString("Sigmoid", width / 2 - 100, 3 * height / 4 - 10);
 g.drawString("SigmoidP", width / 3 - 5, 1 * height / 4 + 10);
 g.drawString("-5", 4, (int)yzero);
 g.drawString("5", width - 19, (int)yzero);
 g.drawString("0.5", width/2 - 7, 12);
 g.drawString("-0.5", width/2 - 9, height - 3);
 }
}

Test_1H.java

2000-05-09T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */

public class Test_1H {

static float [] in1 = { -0.4f, -0.4f, +0.4f };
  static float [] in2 = { -0.4f, +0.4f, -0.4f };
  static float [] in3 = { +0.4f, -0.4f, -0.4f };

static float [] out1= { +0.4f, -0.4f, -0.4f};
  static float [] out2= { -0.4f, -0.4f, +0.4f};
  static float [] out3= { -0.4f, +0.4f, -0.4f};

static float [] test1 = { -0.2f, -0.45f, +0.35f };
  static float [] test2 = { -0.33f, +0.41f, -0.38f };
  static float [] test3 = { +0.33f, -0.41f, -0.23f };

public static void main(String[] args) {
 Neural_1H nn = new Neural_1H(3, 3, 3);
 nn.addTrainingExample(in1, out1);
 nn.addTrainingExample(in2, out2);
 nn.addTrainingExample(in3, out3);
 for (int i=0; i<302; i++) {
 float error = nn.train();
 if ((i + 19) % 20 == 0) System.out.println("cycle " + i + " error is " + error);
 }
 test_recall(nn, test1);
 test_recall(nn, test2);
 test_recall(nn, test3);

nn.save("test.neural");
 Neural_1H nn2 = Neural_1H.Factory("test.neural");
 nn2.addTrainingExample(in1, out1);
 nn2.addTrainingExample(in2, out2);
 nn2.addTrainingExample(in3, out3);
 for (int i=0; i<102; i++) {
 float error = nn2.train();
 if ((i + 19) % 20 == 0) System.out.println("cycle " + i + " error is " + error);
 }
 test_recall(nn2, test1);
 test_recall(nn2, test2);
 test_recall(nn2, test3);
 }
 public static void test_recall(Neural_1H nn, float [] inputs) {
 float [] results = nn.recall(inputs);
 System.out.print("Test case: ");
 for (int i=0; i<inputs.length; i++) System.out.print(pp(inputs[i]) + " ");
 System.out.print(" results: ");
 for (int i=0; i<results.length; i++) System.out.print(pp(results[i]) + " ");
 System.out.println();
 }
 public static String pp(float x) {
 String s = new String("" + x + "00");
 int index = s.indexOf(".");
 if (index > -1) s = s.substring(0, index + 3);
 if (s.startsWith("-") == false) s = " " + s;
 return s;
 }
}

GUITest_2H$1.class

2000-05-07T16:00:00Z

GUITest_2H.class

2000-05-07T16:00:00Z

neural.jpr

2000-05-07T16:00:00Z

GUITest_2H.java

2000-04-18T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */

import javax.swing.*;
import java.awt.*;
import java.awt.event.*;

public class GUITest_2H extends JFrame {
  static float [] in1 = { -0.4f, -0.4f, +0.4f };
  static float [] in2 = { -0.4f, +0.4f, -0.4f };
  static float [] in3 = { +0.4f, -0.4f, -0.4f };

static float [] out1= { +0.4f, -0.4f, -0.4f};
  static float [] out2= { -0.4f, -0.4f, +0.4f};
  static float [] out3= { -0.4f, +0.4f, -0.4f};

static float [] test1 = { -0.2f, -0.45f, +0.35f };
  static float [] test2 = { -0.33f, +0.41f, -0.38f };
  static float [] test3 = { +0.33f, -0.41f, -0.23f };

Neural_2H nn = new Neural_2H(3, 3, 3, 3);
  Plot1DPanel inputPanel = new Plot1DPanel(3, -1.0f, 1.0f, nn.inputs);
  Plot1DPanel hidden1Panel = new Plot1DPanel(3, -1.0f, 1.0f, nn.hidden1);
  Plot1DPanel hidden2Panel = new Plot1DPanel(3, -1.0f, 1.0f, nn.hidden2);
  Plot1DPanel outputPanel = new Plot1DPanel(3, -1.0f, 1.0f, nn.outputs);
  Plot2DPanel w1Panel = new Plot2DPanel(3, 3, -1.0f, 1.0f, nn.W1);
  Plot2DPanel w2Panel = new Plot2DPanel(3, 3, -1.0f, 1.0f, nn.W2);
  Plot2DPanel w3Panel = new Plot2DPanel(3, 3, -1.0f, 1.0f, nn.W3);
  JButton jButton1 = new JButton();
  JLabel jLabel1 = new JLabel();
  JLabel jLabel2 = new JLabel();
  JLabel jLabel2b = new JLabel();
  JLabel jLabel3 = new JLabel();
  JLabel jLabel4 = new JLabel();
  JLabel jLabel4b = new JLabel();
  JLabel jLabel5 = new JLabel();

public GUITest_2H() {
    try {
    nn.addTrainingExample(in1, out1);
    nn.addTrainingExample(in2, out2);
    nn.addTrainingExample(in3, out3);
      jbInit();
      this.setSize(450, 450);
      this.setVisible(true);
    }
    catch(Exception e) {
      e.printStackTrace();
    }
  }

public static void main(String[] args) {
    GUITest_2H GUITest_1H1 = new GUITest_2H();
  }

private void jbInit() throws Exception {
    this.getContentPane().setLayout(null);
    inputPanel.setBounds(new Rectangle(5, 30, 400, 20));
    hidden1Panel.setBounds(new Rectangle(5, 138, 400, 20));
    hidden2Panel.setBounds(new Rectangle(5, 238, 400, 20));
    outputPanel.setBounds(new Rectangle(5, 340, 400, 20));
    w1Panel.setBounds(new Rectangle(160, 50, 60, 60));
    w2Panel.setBounds(new Rectangle(160, 158, 60, 60));
    w3Panel.setBounds(new Rectangle(160, 258, 60, 60));
    jButton1.setText("Rest and Run");
    jButton1.setBounds(new Rectangle(246, 380, 148, 28));
    jButton1.addMouseListener(new java.awt.event.MouseAdapter() {

public void mousePressed(MouseEvent e) {
        do_run_button(e);
      }
    });
    this.setDefaultCloseOperation(3);
    jLabel1.setText("Input neurons:");
    jLabel1.setBounds(new Rectangle(4, 3, 144, 19));
    jLabel2.setText("Hidden 1 neurons:");
    jLabel2.setBounds(new Rectangle(4, 102, 144, 19));
    jLabel2b.setText("Hidden 2 neurons:");
    jLabel2b.setBounds(new Rectangle(4, 202, 144, 19));
    jLabel3.setText("Output neurons:");
    jLabel3.setBounds(new Rectangle(4, 308, 240, 19));
    jLabel4.setText("input to hidden 1 weights");
    jLabel4.setBounds(new Rectangle(230, 80, 170, 19));
    jLabel4b.setText("hidden 1 to hidden 2 weights");
    jLabel4b.setBounds(new Rectangle(230, 180, 170, 19));
    jLabel5.setText("hidden 2 to output weights");
    jLabel5.setBounds(new Rectangle(230, 280, 170, 19));
    this.getContentPane().add(inputPanel, null);
    this.getContentPane().add(hidden1Panel, null);
    this.getContentPane().add(hidden2Panel, null);
    this.getContentPane().add(outputPanel, null);
    this.getContentPane().add(w1Panel, null);
    this.getContentPane().add(w2Panel, null);
    this.getContentPane().add(w3Panel, null);
    this.getContentPane().add(jButton1, null);
    this.getContentPane().add(jLabel1, null);
    this.getContentPane().add(jLabel2, null);
    this.getContentPane().add(jLabel2b, null);
    this.getContentPane().add(jLabel3, null);
    this.getContentPane().add(jLabel4, null);
    this.getContentPane().add(jLabel4b, null);
    this.getContentPane().add(jLabel5, null);
  }

void do_run_button(MouseEvent e) {
 Graphics g1 = inputPanel.getGraphics();
 Graphics g2 = hidden1Panel.getGraphics();
 Graphics g3 = hidden2Panel.getGraphics();
 Graphics g4 = outputPanel.getGraphics();
 Graphics g5 = w1Panel.getGraphics();
 Graphics g6 = w2Panel.getGraphics();
 Graphics g7 = w3Panel.getGraphics();
 for (int i=0; i<5600; i++) {
 float error = nn.train();
 //if ((i + 19) % 20 == 0) System.out.println("cycle " + i + " error is " + error);
 inputPanel.paint(g1);
 hidden1Panel.paint(g2);
 hidden2Panel.paint(g3);
 outputPanel.paint(g4);
 w1Panel.paint(g5);
 w2Panel.paint(g6);
 w3Panel.paint(g7);
 if (i > 5580) {
 try { Thread.sleep(300L); } catch (Exception e9) { }
 } else {
 try { Thread.sleep(10L); } catch (Exception e9) { }
 }
 }

}
}

GUITest_1H$1.class

2000-04-17T16:00:00Z

GUITest_1H.class

2000-04-17T16:00:00Z

Neural_1H.class

2000-04-17T16:00:00Z

Neural_2H.class

2000-04-17T16:00:00Z

Neural_2H.java

2000-04-17T16:00:00Z

/**
 * Title: Neural_2H
 * Description: Two hidden layer back propagtion neural network model
 * Copyright: Copyright (c) 2000 Mark Watson. Artisitic Open Source License
 * @author Mark Watson
 * @version 1.2
 */

import java.util.*;
import java.io.*;

class Neural_2H implements Serializable {

 protected int numInputs;
 protected int numHidden1;
 protected int numHidden2;
 protected int numOutputs;

 protected int numTraining;

 public float inputs[];
 protected float hidden1[];
 protected float hidden2[];
 public float outputs[];

 protected float W1[][];
 protected float W2[][];
 protected float W3[][];

 protected float output_errors[];
 protected float hidden1_errors[];
 protected float hidden2_errors[];

 transient protected Vector inputTraining = new Vector();
 transient protected Vector outputTraining = new Vector();

 Neural_2H(int num_in, int num_hidden1, int num_hidden2, int num_output) {
	numInputs = num_in;
	numHidden1 = num_hidden1;
	numHidden2 = num_hidden2;
	numOutputs = num_output;
	inputs = new float[numInputs];
	hidden1 = new float[numHidden1];
	hidden2 = new float[numHidden2];
	outputs = new float[numOutputs];
	W1 = new float[numInputs][numHidden1];
	W2 = new float[numHidden1][numHidden2];
	W3 = new float[numHidden2][numOutputs];
	randomizeWeights();

	output_errors = new float[numOutputs];
	hidden1_errors = new float[numHidden1];
	hidden2_errors = new float[numHidden2];
 }
 public void addTrainingExample(float [] inputs, float [] outputs) {
	if (inputs.length != numInputs || outputs.length != numOutputs) {
	 System.out.println("addTrainingExample(): array size is wrong");
	 return;
	}
	inputTraining.addElement(inputs);
	outputTraining.addElement(outputs);
 }

 public static Neural_2H Factory(String serialized_file_name) {
	Neural_2H nn = null;
	try {
	 InputStream ins = ClassLoader.getSystemResourceAsStream(serialized_file_name);
	 if (ins==null) {
		System.out.println("CachedExamples(): failed to open 'cache.dat' in JAR file");
		System.exit(1);
	 } else {
		ObjectInputStream p = new ObjectInputStream(ins);
		nn = (Neural_2H)p.readObject();
 nn.inputTraining = new Vector();
 nn.outputTraining = new Vector();
		ins.close();
	 }
	} catch (Exception e) {
	 e.printStackTrace();
	 return null;
	}
	return nn;
 }

 public void save(String file_name) {
	try {
	 FileOutputStream ostream = new FileOutputStream(file_name);
	 ObjectOutputStream p = new ObjectOutputStream(ostream);
	 p.writeObject(this);
	 p.flush();
	 ostream.close();
	} catch (Exception e) {
	 e.printStackTrace();
	}
 }

 public void randomizeWeights() {
 // Randomize weights here:
 for (int ii=0; ii<numInputs; ii++)
 for (int hh=0; hh<numHidden1; hh++)
 W1[ii][hh] =
 0.1f * (float)Math.random() - 0.05f;
 for (int ii=0; ii<numHidden1; ii++)
 for (int hh=0; hh<numHidden2; hh++)
 W2[ii][hh] =
 0.1f * (float)Math.random() - 0.05f;
 for (int hh=0; hh<numHidden2; hh++)
 for (int oo=0; oo<numOutputs; oo++)
 W3[hh][oo] =
 0.1f * (float)Math.random() - 0.05f;
 }

 public float [] recall(float [] in) {
 for (int i=0; i<numInputs; i++) inputs[i] = in[i];
 forwardPass();
 float [] ret = new float[numOutputs];
 for (int i=0; i<numOutputs; i++) ret[i] = outputs[i];
 return ret;
 }

 public void forwardPass() {
 int i, h, o;
 for (h=0; h<numHidden1; h++) {
 hidden1[h] = 0.0f;
 }
 for (h=0; h<numHidden2; h++) {
 hidden2[h] = 0.0f;
 }
 for (i=0; i<numInputs; i++) {
 for (h=0; h<numHidden1; h++) {
 hidden1[h] +=
 inputs[i] * W1[i][h];
 }
 }
 for (i=0; i<numHidden1; i++) {
 for (h=0; h<numHidden2; h++) {
 hidden2[h] +=
 hidden1[i] * W2[i][h];
 }
 }
 for (o=0; o<numOutputs; o++)
 outputs[o] = 0.0f;
 for (h=0; h<numHidden2; h++) {
 for (o=0; o<numOutputs; o++) {
 outputs[o] +=
 sigmoid(hidden2[h]) * W3[h][o];
 }
 }
 for (o=0; o<numOutputs; o++)
 outputs[o] = sigmoid(outputs[o]);
 }

 public float train() {
 return train(inputTraining, outputTraining);
 }

 private int current_example = 0;
 public float train(Vector ins, Vector v_outs) {
 int i, h, o;
 float error = 0.0f;
 int num_cases = ins.size();
 //for (int example=0; example<num_cases; example++) {
 // zero out error arrays:
 for (h=0; h<numHidden1; h++)
 hidden1_errors[h] = 0.0f;
 for (h=0; h<numHidden2; h++)
 hidden2_errors[h] = 0.0f;
 for (o=0; o<numOutputs; o++)
 output_errors[o] = 0.0f;
 // copy the input values:
 for (i=0; i<numInputs; i++) {
 inputs[i] = ((float [])ins.elementAt(current_example))[i];
 }
 // copy the output values:
 float [] outs = (float [])v_outs.elementAt(current_example);

 // perform a forward pass through the network:

 forwardPass();

 for (o=0; o<numOutputs; o++) {
 output_errors[o] =
 (outs[o] -
 outputs[o])
 *sigmoidP(outputs[o]);
 }
 for (h=0; h<numHidden2; h++) {
 hidden2_errors[h] = 0.0f;
 for (o=0; o<numOutputs; o++) {
 hidden2_errors[h] +=
 output_errors[o]*W3[h][o];
 }
 }
 for (h=0; h<numHidden1; h++) {
 hidden1_errors[h] = 0.0f;
 for (o=0; o<numHidden2; o++) {
 hidden1_errors[h] +=
 hidden2_errors[o]*W2[h][o];
 }
 }
 for (h=0; h<numHidden2; h++) {
 hidden2_errors[h] =
 hidden2_errors[h]*sigmoidP(hidden2[h]);
 }
 for (h=0; h<numHidden1; h++) {
 hidden1_errors[h] =
 hidden1_errors[h]*sigmoidP(hidden1[h]);
 }
 // update the hidden2 to output weights:
 for (o=0; o<numOutputs; o++) {
 for (h=0; h<numHidden2; h++) {
 W3[h][o] +=
 0.5 * output_errors[o] * hidden2[h];
 }
 }
 // update the hidden1 to hidden2 weights:
 for (o=0; o<numHidden2; o++) {
 for (h=0; h<numHidden1; h++) {
 W2[h][o] +=
 0.5 * hidden2_errors[o] * hidden1[h];
 }
 }
 // update the input to hidden1 weights:
 for (h=0; h<numHidden1; h++) {
 for (i=0; i<numInputs; i++) {
 W1[i][h] +=
 0.5 * hidden1_errors[h] * inputs[i];
 }
 }
 for (o=0; o<numOutputs; o++)
 error += Math.abs(output_errors[o]);
 //}
 current_example++;
 if (current_example >= num_cases) current_example = 0;
 return error;
 }

 protected float sigmoid(float x) {
 return
 (float)((1.0f/(1.0f+Math.exp((double)(-x))))-0.5f);
 }

 protected float sigmoidP(float x) {
 double z = sigmoid(x) + 0.5f;
 return (float)(z * (1.0f - z));
 }

}

Plot1DPanel.class

2000-04-17T16:00:00Z

Plot2DPanel.class

2000-04-17T16:00:00Z

Test_1H.class

2000-04-17T16:00:00Z

Test_2H.class

2000-04-17T16:00:00Z

GUITest_1H.java

2000-04-16T16:00:00Z

import javax.swing.*;
import java.awt.*;
import java.awt.event.*;
/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */

public class GUITest_1H extends JFrame {
  static float [] in1 = { -0.4f, -0.4f, +0.4f };
  static float [] in2 = { -0.4f, +0.4f, -0.4f };
  static float [] in3 = { +0.4f, -0.4f, -0.4f };

static float [] out1= { +0.4f, -0.4f, -0.4f};
  static float [] out2= { -0.4f, -0.4f, +0.4f};
  static float [] out3= { -0.4f, +0.4f, -0.4f};

static float [] test1 = { -0.2f, -0.45f, +0.35f };
  static float [] test2 = { -0.33f, +0.41f, -0.38f };
  static float [] test3 = { +0.33f, -0.41f, -0.23f };

Neural_1H nn = new Neural_1H(3, 3, 3);
  Plot1DPanel inputPanel = new Plot1DPanel(3, -1.0f, 1.0f, nn.inputs);
  Plot1DPanel hiddenPanel = new Plot1DPanel(3, -4.0f, 4.0f, nn.hidden);
  Plot1DPanel outputPanel = new Plot1DPanel(3, -1.0f, 1.0f, nn.outputs);
  Plot2DPanel w1Panel = new Plot2DPanel(3, 3, -1.0f, 1.0f, nn.W1);
  Plot2DPanel w2Panel = new Plot2DPanel(3, 3, -4.0f, 4.0f, nn.W2);
  JButton jButton1 = new JButton();
  JLabel jLabel1 = new JLabel();
  JLabel jLabel2 = new JLabel();
  JLabel jLabel3 = new JLabel();
  JLabel jLabel4 = new JLabel();
  JLabel jLabel5 = new JLabel();

public GUITest_1H() {
    try {
    nn.addTrainingExample(in1, out1);
    nn.addTrainingExample(in2, out2);
    nn.addTrainingExample(in3, out3);
      jbInit();
      this.setSize(450, 350);
      this.setVisible(true);
    }
    catch(Exception e) {
      e.printStackTrace();
    }
  }

public static void main(String[] args) {
    GUITest_1H GUITest_1H1 = new GUITest_1H();
  }

private void jbInit() throws Exception {
    this.getContentPane().setLayout(null);
    inputPanel.setBounds(new Rectangle(5, 30, 400, 20));
    hiddenPanel.setBounds(new Rectangle(5, 138, 400, 20));
    outputPanel.setBounds(new Rectangle(5, 240, 400, 20));
    w1Panel.setBounds(new Rectangle(200, 50, 60, 60));
    w2Panel.setBounds(new Rectangle(200, 158, 60, 60));
    jButton1.setText("Rest and Run");
    jButton1.setBounds(new Rectangle(246, 290, 148, 28));
    jButton1.addMouseListener(new java.awt.event.MouseAdapter() {

public void mousePressed(MouseEvent e) {
        do_run_button(e);
      }
    });
    this.setDefaultCloseOperation(3);
    jLabel1.setText("Input neurons:");
    jLabel1.setBounds(new Rectangle(4, 3, 144, 19));
    jLabel2.setText("Hidden neurons:");
    jLabel2.setBounds(new Rectangle(4, 102, 144, 19));
    jLabel3.setText("Output neurons:");
    jLabel3.setBounds(new Rectangle(4, 208, 240, 19));
    jLabel4.setText("input to hidden weights");
    jLabel4.setBounds(new Rectangle(270, 80, 150, 19));
    jLabel5.setText("hidden to output weights");
    jLabel5.setBounds(new Rectangle(270, 180, 150, 19));
    this.getContentPane().add(inputPanel, null);
    this.getContentPane().add(hiddenPanel, null);
    this.getContentPane().add(outputPanel, null);
    this.getContentPane().add(w1Panel, null);
    this.getContentPane().add(w2Panel, null);
    this.getContentPane().add(jButton1, null);
    this.getContentPane().add(jLabel1, null);
    this.getContentPane().add(jLabel2, null);
    this.getContentPane().add(jLabel3, null);
    this.getContentPane().add(jLabel4, null);
    this.getContentPane().add(jLabel5, null);
  }

void do_run_button(MouseEvent e) {
 Graphics g1 = inputPanel.getGraphics();
 Graphics g2 = hiddenPanel.getGraphics();
 Graphics g3 = outputPanel.getGraphics();
 Graphics g4 = w1Panel.getGraphics();
 Graphics g5 = w2Panel.getGraphics();
 for (int i=0; i<600; i++) {
 float error = nn.train();
 //if ((i + 19) % 20 == 0) System.out.println("cycle " + i + " error is " + error);
 inputPanel.paint(g1);
 hiddenPanel.paint(g2);
 outputPanel.paint(g3);
 w1Panel.paint(g4);
 w2Panel.paint(g5);
 if (i > 580) {
 try { Thread.sleep(500L); } catch (Exception e9) { }
 } else {
 try { Thread.sleep(30L); } catch (Exception e9) { }
 }
 }

}
}

Plot1DPanel.java

2000-04-16T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */
import javax.swing.*;
import java.awt.*;

public class Plot1DPanel extends java.awt.Canvas { // JPanel {

public Plot1DPanel(int num, float min, float max, float [] values) {
 super();
 this.num = num;
 this.min = min;
 this.max = max;
 this.values = values;
 colors = new Color[100];
 for (int i=0; i<100; i++) {
 float x = 1.0f - ((float)i) * 0.0096f;
 colors[i] = new Color(x, x, x);
 }
 }
 private int num;
 private float min, max;
 private float temp;
 private float [] values = null;
 private Color [] colors;
 //public void plot(float [] values) {
 //}
 public void paint(Graphics g) {
 if (values == null) return;
 int delta_width = this.getWidth() / num;
 int delta_height = this.getHeight() / num;
 for (int i=0; i<num; i++) {
 //System.out.println(this.toString() + ", values[" + i + "]=" + values[i]);
 temp = 100.0f * (values[i] - min) / (max - min);
 int ii = (int)temp;
 if (ii < 0) ii = 0;
 if (ii > 99) ii = 99;
 g.setColor(colors[ii]);
 g.fillRect(i * delta_width, 0, delta_width, delta_height);
 }
 }
}

Plot2DPanel.java

2000-04-16T16:00:00Z

/**
 * Title: 
 * Description: 
 * Copyright: Copyright (c) 
 * Company: 
 * @author
 * @version 1.0
 */
import javax.swing.*;
import java.awt.*;

public class Plot2DPanel extends java.awt.Canvas { // JPanel {

public Plot2DPanel(int num1, int num2, float min, float max, float [][] values) {
 super();
 this.num1 = num1;
 this.num2 = num2;
 this.min = min;
 this.max = max;
 this.values = values;
 colors = new Color[100];
 for (int i=0; i<100; i++) {
 float x = 1.0f - ((float)i) * 0.0096f;
 colors[i] = new Color(x, x, x);
 }
 }
 private int num1;
 private int num2;
 private float min, max;
 private float temp;
 private float [][] values = null;
 private Color [] colors;

public void paint(Graphics g) {
 if (values == null) return;
 int delta_width = this.getWidth() / num1;
 int delta_height = this.getHeight() / num2;
 for (int i=0; i<num1; i++) {
 for (int j=0; j<num2; j++) {
 //System.out.println(this.toString() + ", values[" + i + "]=" + values[i]);
 temp = 100.0f * (values[i][j] - min) / (max - min);
 int ii = (int)temp;
 if (ii < 0) ii = 0;
 if (ii > 99) ii = 99;
 g.setColor(colors[ii]);
 g.fillRect(i * delta_width, j * delta_height, delta_width, delta_height);
 }
 }
 }
}