weka.classifiers.immune.airs.algorithm.AIRS2ParallelTrainer Maven / Gradle / Ivy

Go to download
Show more of this group Show more artifacts with this name
Show all versions of wekaclassalgos Show documentation
Fork of the following defunct sourceforge.net project: https://sourceforge.net/projects/wekaclassalgos/
There is a newer version: 2023.2.8
/*
 *   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 3 of the License, or
 *   (at 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.
 *
 *   You should have received a copy of the GNU General Public License
 *   along with this program.  If not, see .
 */

/*
 * Created on 8/01/2005
 *
 */
package weka.classifiers.immune.airs.algorithm;

import weka.classifiers.immune.airs.algorithm.merge.ConcatenateMerge;
import weka.classifiers.immune.airs.algorithm.merge.PruneMerge;
import weka.core.Instances;
import weka.filters.Filter;
import weka.filters.unsupervised.attribute.Normalize;

import java.util.LinkedList;
import java.util.Random;
import java.util.concurrent.CountDownLatch;

/**
 * Type: AIRS2ParallelTrainer

 * File: AIRS2ParallelTrainer.java

 * Date: 8/01/2005

 * 

 * Description:
 * 

 *
 * @author Jason Brownlee
 */
public class AIRS2ParallelTrainer implements AISTrainer {

  protected final double affinityThresholdScalar;

  protected final double clonalRate;

  protected final double hyperMutationRate;

  protected final double totalResources;

  protected final double stimulationThreshold;

  protected final int affinityThresholdNumInstances;

  protected final Random rand;

  protected final int memoryCellPoolInitialSize;

  protected final int kNN;

  // different means of merging
  public enum MERGE_MODE {
    CONCATENATE,
    PRUNE
  }

  // additional
  protected final int numThreads;

  protected final MERGE_MODE mergeMode;

  protected double affinityThreshold;

  protected MemoryCellMerger merger;

  protected CountDownLatch latch;

  protected String[] trainingSummaries;


  public AIRS2ParallelTrainer(
    double aAffinityThresholdScalar,
    double aClonalRate,
    double aHyperMutationRate,
    double aTotalResources,
    double aStimulationValue,
    int aNumInstancesAffinityThreshold,
    Random aRand,
    int aMemoryCellPoolInitialSize,
    int aKNN,
    int aNumThreads,
    MERGE_MODE aMergeMode) {
    affinityThresholdScalar = aAffinityThresholdScalar;
    clonalRate = aClonalRate;
    hyperMutationRate = aHyperMutationRate;
    totalResources = aTotalResources;
    stimulationThreshold = aStimulationValue;
    affinityThresholdNumInstances = aNumInstancesAffinityThreshold;
    rand = aRand;
    memoryCellPoolInitialSize = aMemoryCellPoolInitialSize;
    kNN = aKNN;
    // additional
    numThreads = aNumThreads;
    mergeMode = aMergeMode;
  }

  /**
   * @param aInstances
   * @return
   * @throws Exception
   */
  public AISModelClassifier train(Instances aInstances) throws Exception {
    // normalise the dataset
    Normalize normalise = new Normalize();
    normalise.setInputFormat(aInstances);
    Instances trainingSet = Filter.useFilter(aInstances, normalise);

    // calculate affinity threshold
    AffinityFunction affinityFunction = new AffinityFunction(trainingSet);
    affinityThreshold = Utils.calculateAffinityThreshold(trainingSet, affinityThresholdNumInstances, rand, affinityFunction);

    // prepare latch
    latch = new CountDownLatch(numThreads);

    // divide up dataset
    Instances[] instances = new Instances[numThreads];
    trainingSet.randomize(rand);
    int numPerThread = (int) Math.round((double) trainingSet.numInstances() / (double) numThreads);
    int offset = 0;
    for (int i = 0; i < instances.length; i++) {
      // check for last - give all remaining
      if (i == instances.length - 1) {
	instances[i] = new Instances(trainingSet, offset, trainingSet.numInstances() - offset);
      }
      else {
	instances[i] = new Instances(trainingSet, offset, numPerThread);
	offset += numPerThread;
      }
    }

    // prepare threads
    AIRSProcess[] threads = new AIRSProcess[numThreads];
    for (int i = 0; i < threads.length; i++) {
      // prep algorithm
      AIRS2Trainer algorithm = new AIRS2Trainer(
	affinityThresholdScalar,
	clonalRate,
	hyperMutationRate,
	totalResources,
	stimulationThreshold,
	affinityThresholdNumInstances,
	rand,
	memoryCellPoolInitialSize,
	kNN);
      // prepare algorithm - instances only needed for distance measure prep
      algorithm.algorithmPreperation(trainingSet);
      // create thread
      threads[i] = new AIRSProcess(algorithm, instances[i], normalise);
      // start the thread - don't need ref to it - got a latch
      Thread t = new Thread(threads[i], "AIRS2 Thread Number: " + i); // name for debug if required
      t.start();
    }

    // wait for all threads to complete
    latch.await();

    // collect memory cells
    LinkedList[] cells = new LinkedList[numThreads];
    trainingSummaries = new String[numThreads];
    for (int i = 0; i < threads.length; i++) {
      cells[i] = threads[i].getCells();
      trainingSummaries[i] = threads[i].getTrainingSummary();
    }

    // prepare classifier (merging)
    merger = getMeger();
    AISModelClassifier classifier = merger.mergeMemoryCells(cells, kNN, normalise, affinityFunction, trainingSet);
    return classifier;
  }


  protected MemoryCellMerger getMeger() {
    if (mergeMode == MERGE_MODE.CONCATENATE) {
      return new ConcatenateMerge();
    }
    else if (mergeMode == MERGE_MODE.PRUNE) {
      return new PruneMerge();
    }

    throw new RuntimeException("Unknown merge mode: " + mergeMode);
  }

  protected final class AIRSProcess implements Runnable {

    protected final AIRS2Trainer algorithm;

    protected final Instances instances;

    protected final Normalize normalise;

    protected AISModelClassifier classifier;

    public AIRSProcess(
      AIRS2Trainer aAlgorithm,
      Instances aInstances,
      Normalize aNormalise) {
      algorithm = aAlgorithm;
      instances = aInstances;
      normalise = aNormalise;
    }

    public void run() {
      // set the affinity threshold manually
      algorithm.setAffinityThreshold(affinityThreshold);
      // run training
      try {
	classifier = algorithm.internalTrain(instances, normalise);
      }
      catch (Exception e) {
	throw new RuntimeException("Failed to prepare classifier partition.", e);
      }
      finally {
	// finished
	latch.countDown();
      }
    }

    public LinkedList getCells() {
      return classifier.getModel().getCells();
    }

    public String getTrainingSummary() {
      return algorithm.getTrainingSummary();
    }
  }


  /**
   * @return
   */
  public String getTrainingSummary() {
    StringBuffer buffer = new StringBuffer();

    buffer.append(" - Parallel Training Summary - \n");
    buffer.append("Total Partitions: " + numThreads);
    buffer.append("\n\n");

    for (int i = 0; i < trainingSummaries.length; i++) {
      buffer.append(trainingSummaries[i]);
      buffer.append("\n");
    }

    return buffer.toString();
  }

}