Introduction to Neural Networks for Java, Session 6

Session Material

I present here my solution for program 1. You can use the link below to download an Eclipse project for my solution.

[My Solution for Program 1]

First you will need to create training data in a CSV file for Program 1.

Listing: CSV Input for Program 1

0,0,0,1,0,0
0,1,0,1,0,0
1,0,0,1,0,0
1,1,1,1,0,0
0,0,0,0,1,0
0,1,1,0,1,0
1,0,1,0,1,0
1,1,1,0,1,0
0,0,0,0,0,1
0,1,1,0,0,1
1,0,1,0,0,1
1,1,0,0,0,1

Solution for program 1.

Listing: Solution for Program 1

/**
 * This is an example file from the online course:
 * 
 * Introduction to Neural Networks with Java
 * 
 * http://www.heatonresearch.com/course/intro-neural-nets-java
 * 
 * This course is based on the following book:
 * 
 * Introduction to Neural Networks with Java, 2nd Edition
 * Copyright 2008 by Heaton Research, Inc. 
 * http://www.heatonresearch.com/books/java-neural-2/
 * 
 * ISBN13: 978-1-60439-008-7  	 
 * ISBN:   1-60439-008-5
 *   
 * This class is released under the:
 * GNU Lesser General Public License (LGPL)
 * http://www.gnu.org/copyleft/lesser.html
 */
package com.heatonresearch.course.introneuralnet.program1;

import java.io.IOException;

import com.heatonresearch.book.introneuralnet.common.ReadCSV;
import com.heatonresearch.book.introneuralnet.neural.feedforward.FeedforwardLayer;
import com.heatonresearch.book.introneuralnet.neural.feedforward.FeedforwardNetwork;
import com.heatonresearch.book.introneuralnet.neural.feedforward.train.Train;
import com.heatonresearch.book.introneuralnet.neural.feedforward.train.backpropagation.Backpropagation;

/**
 * This is my solution for Program 1.
 * 
 * @author Jeff Heaton
 * @version 1.0
 */
public class Program1 {
	static final int COLUMNS = 6;
	static final int OPP_AND = 0;
	static final int OPP_OR = 1;
	static final int OPP_XOR = 2;

	private FeedforwardNetwork network;
	private double[][] input;
	private double[][] ideal;

	public void createNetwork() {
		this.network = new FeedforwardNetwork();
		network.addLayer(new FeedforwardLayer(Program1.COLUMNS));
		network.addLayer(new FeedforwardLayer(3));
		network.addLayer(new FeedforwardLayer(1));
		network.reset();
	}

	public void train() {
		// train the neural network
		final Train train = new Backpropagation(network, this.input,
				this.ideal, 0.7, 0.8);

		int epoch = 1;

		do {
			train.iteration();
			System.out
					.println("Epoch #" + epoch + " Error:" + train.getError());
			epoch++;
		} while ((epoch < 5000) && (train.getError() > 0.001));
	}

	public double evaluate(double op1, double op2, int operation) {
		double[] input = new double[Program1.COLUMNS];
		input[0] = op1;
		input[1] = op2;

		// AND
		if (operation == Program1.OPP_AND)
			input[2] = 1.0;
		else
			input[2] = 0.0;

		// OR
		if (operation == Program1.OPP_OR)
			input[3] = 1.0;
		else
			input[3] = 0.0;

		// XOR
		if (operation == Program1.OPP_XOR)
			input[4] = 1.0;
		else
			input[4] = 0.0;

		double[] output = this.network.computeOutputs(input);
		return output[0];
	}

	public void load(String filename) throws IOException {
		int size = 0;

		// count the size of the file
		ReadCSV csv = new ReadCSV(filename, false);
		while (csv.next()) {
			size++;
		}
		csv.close();

		// allocate enough space
		this.input = new double[size][Program1.COLUMNS];
		this.ideal = new double[size][1];

		// now load it
		int index = 0;
		csv = new ReadCSV(filename, false);
		while (csv.next()) {
			this.input[index][0] = Double.parseDouble(csv.get(0));
			this.input[index][1] = Double.parseDouble(csv.get(1));
			this.input[index][2] = Double.parseDouble(csv.get(3));
			this.input[index][3] = Double.parseDouble(csv.get(4));
			this.input[index][4] = Double.parseDouble(csv.get(5));
			this.ideal[index][0] = Double.parseDouble(csv.get(2));
			index++;
		}
		csv.close();

	}

	public static void main(final String args[]) {
		try {
			Program1 prg = new Program1();
			prg.load("c:\\data\\logic.csv");
			prg.createNetwork();
			prg.train();
			System.out.println("1 AND 1 = "
					+ prg.evaluate(1, 1, Program1.OPP_AND));
			System.out.println("1 XOR 1 = "
					+ prg.evaluate(1, 1, Program1.OPP_XOR));
			System.out.println("1 AND 0 = "
					+ prg.evaluate(1, 0, Program1.OPP_AND));
			System.out.println("1 OR 1 = "
					+ prg.evaluate(1, 1, Program1.OPP_OR));
		} catch (Throwable t) {
			t.printStackTrace();
		}

	}
}

Videos for this Session

Video	Title
	Introduction to Neural Networks for Java(Class 6)