WidrowHoffLearner.java

/*
 * JASA Java Auction Simulator API
 * Copyright (C) 2013 Steve Phelps
 *
 * This program is free software; you can redistribute it and/or
 * modify it under the terms of the GNU General Public License as
 * published by the Free Software Foundation; either version 2 of
 * the License, or (at e;your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.
 * See the GNU General Public License for more details.
 */

package net.sourceforge.jabm.learning;

import java.io.Serializable;

import net.sourceforge.jabm.report.DataWriter;
import net.sourceforge.jabm.util.Prototypeable;
import cern.jet.random.AbstractContinousDistribution;
import cern.jet.random.Uniform;
import cern.jet.random.engine.RandomEngine;

/**
 * An implementation of the Widrow-Hoff learning algorithm for 1-dimensional
 * training sets.
 * 
 * @author Steve Phelps
 * @version $Revision: 306 $
 */
public class WidrowHoffLearner extends AbstractLearner implements
    MimicryLearner, Prototypeable, Serializable {

	/**
	 * The learning rate.
	 */
	protected double learningRate;

	/**
	 * The current output level.
	 */
	protected double currentOutput;

	/**
	 * The current amount of adjustment to the output.
	 */
	protected double delta;

	protected AbstractContinousDistribution randomParamDistribution;

	public static final double DEFAULT_LEARNING_RATE = 0.1;

	public WidrowHoffLearner(double learningRate, AbstractContinousDistribution randomParamDistribution) {
		this.randomParamDistribution = randomParamDistribution;
		this.learningRate = learningRate;
		initialise();
	}
	
	public WidrowHoffLearner(AbstractContinousDistribution randomParamDistribution) {
		this(DEFAULT_LEARNING_RATE, randomParamDistribution);
	}
	
	public WidrowHoffLearner() {
		this(null);
	}
	
	public WidrowHoffLearner(double learningRate, RandomEngine prng) {
		this(learningRate, new Uniform(0.1, 0.4, prng));
	}
//	
//	public WidrowHoffLearner(RandomEngine prng) {
//		this(DEFAULT_LEARNING_RATE, prng);
//	}

//	public void setup(ParameterDatabase parameters, Parameter base) {
//		super.setup(parameters, base);
//		learningRate = parameters.getDoubleWithDefault(base.push(P_LEARNINGRATE),
//		    new Parameter(P_DEF_BASE).push(P_LEARNINGRATE), DEFAULT_LEARNING_RATE);
//	}

	public void initialise() {
		delta = 0;
		currentOutput = 0;
	}

	public Object protoClone() {
		WidrowHoffLearner clone = new WidrowHoffLearner(learningRate, this.randomParamDistribution);
		return clone;
	}

	public double act() {
		return currentOutput;
	}

	public void train(double target) {
		currentOutput += delta(target);
	}

	public double delta(double target) {
		delta = learningRate * (target - currentOutput);
		return delta;
	}

	public void setOutputLevel(double currentOutput) {
		this.currentOutput = currentOutput;
	}

	public void dumpState(DataWriter out) {
		// TODO
	}

	public double getLearningDelta() {
		return delta;
	}

	public void reset() {
		initialise();
	}

	public void setLearningRate(double learningRate) {
		this.learningRate = learningRate;
	}

	public double getLearningRate() {
		return learningRate;
	}

	public void randomInitialise() {
		learningRate = randomParamDistribution.nextDouble();
	}
	
	public double getCurrentOutput() {
		return currentOutput;
	}
	
	public double getDelta() {
		return delta;
	}
	
	public AbstractContinousDistribution getRandomParamDistribution() {
		return randomParamDistribution;
	}

	public void setRandomParamDistribution(
			AbstractContinousDistribution randomParamDistribution) {
		this.randomParamDistribution = randomParamDistribution;
	}
	
	public String toString() {
		return "(" + getClass().getSimpleName() + " learningRate:" + learningRate
		    + ")";
	}
}