Package org.encog.neural.prune

Source Code of org.encog.neural.prune.TestPruneSelective

/*
* Encog(tm) Core v3.3 - Java Version
* http://www.heatonresearch.com/encog/
* https://github.com/encog/encog-java-core
* Copyright 2008-2014 Heaton Research, Inc.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
*     http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*  
* For more information on Heaton Research copyrights, licenses
* and trademarks visit:
* http://www.heatonresearch.com/copyright
*/
package org.encog.neural.prune;

import junit.framework.TestCase;

import org.encog.ml.data.basic.BasicMLData;
import org.encog.neural.flat.FlatNetwork;
import org.encog.neural.networks.BasicNetwork;
import org.encog.neural.networks.XOR;
import org.encog.neural.networks.structure.NetworkCODEC;
import org.encog.util.simple.EncogUtility;
import org.junit.Assert;

public class TestPruneSelective extends TestCase {
 
  private BasicNetwork obtainNetwork()
  {
    BasicNetwork network = EncogUtility.simpleFeedForward(2,3,0,4,false);
    double[] weights = { 1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25 };
    NetworkCODEC.arrayToNetwork(weights, network);
   
    Assert.assertEquals(1.0, network.getWeight(1, 0, 0),0.01);   
    Assert.assertEquals(2.0, network.getWeight(1, 1, 0),0.01);
    Assert.assertEquals(3.0, network.getWeight(1, 2, 0),0.01);
    Assert.assertEquals(4.0, network.getWeight(1, 3, 0),0.01);
   
    Assert.assertEquals(5.0, network.getWeight(1, 0, 1),0.01);
    Assert.assertEquals(6.0, network.getWeight(1, 1, 1),0.01);
    Assert.assertEquals(7.0, network.getWeight(1, 2, 1),0.01);
    Assert.assertEquals(8.0, network.getWeight(1, 3, 1),0.01);
   
    Assert.assertEquals(9.0, network.getWeight(1, 0, 2),0.01);
    Assert.assertEquals(10.0, network.getWeight(1, 1, 2),0.01);
    Assert.assertEquals(11.0, network.getWeight(1, 2, 2),0.01);
    Assert.assertEquals(12.0, network.getWeight(1, 3, 2),0.01);
   
    Assert.assertEquals(13.0, network.getWeight(1, 0, 3),0.01);
    Assert.assertEquals(14.0, network.getWeight(1, 1, 3),0.01);
    Assert.assertEquals(15.0, network.getWeight(1, 2, 3),0.01);
    Assert.assertEquals(16.0, network.getWeight(1, 3, 3),0.01);
   
    Assert.assertEquals(17.0, network.getWeight(0, 0, 0),0.01);
    Assert.assertEquals(18.0, network.getWeight(0, 1, 0),0.01);
    Assert.assertEquals(19.0, network.getWeight(0, 2, 0),0.01);
    Assert.assertEquals(20.0, network.getWeight(0, 0, 1),0.01);
    Assert.assertEquals(21.0, network.getWeight(0, 1, 1),0.01);
    Assert.assertEquals(22.0, network.getWeight(0, 2, 1),0.01);
   
    Assert.assertEquals(20.0, network.getWeight(0, 0, 1),0.01);
    Assert.assertEquals(21.0, network.getWeight(0, 1, 1),0.01);
    Assert.assertEquals(22.0, network.getWeight(0, 2, 1),0.01);
   
    Assert.assertEquals(23.0, network.getWeight(0, 0, 2),0.01);
    Assert.assertEquals(24.0, network.getWeight(0, 1, 2),0.01);
    Assert.assertEquals(25.0, network.getWeight(0, 2, 2),0.01);

   
    return network;
  }
 
  private void checkWithModel(FlatNetwork model, FlatNetwork pruned)
  {
    Assert.assertEquals(model.getWeights().length, pruned.getWeights().length);
    Assert.assertArrayEquals(model.getContextTargetOffset(),pruned.getContextTargetOffset());
    Assert.assertArrayEquals(model.getContextTargetSize(),pruned.getContextTargetSize());
    Assert.assertArrayEquals(model.getLayerCounts(),pruned.getLayerCounts());
    Assert.assertArrayEquals(model.getLayerFeedCounts(),pruned.getLayerFeedCounts());
    Assert.assertArrayEquals(model.getLayerIndex(),pruned.getLayerIndex());
    Assert.assertEquals(model.getLayerOutput().length,pruned.getLayerOutput().length);
    Assert.assertArrayEquals(model.getWeightIndex(),pruned.getWeightIndex());
  }
 
  public void testPruneNeuronInput()
  {
    BasicNetwork network = obtainNetwork();
    Assert.assertEquals(2, network.getInputCount());
    PruneSelective prune = new PruneSelective(network);
    prune.prune(0, 1);
    Assert.assertEquals(22, network.encodedArrayLength());
    Assert.assertEquals(1,network.getLayerNeuronCount(0));
    Assert.assertEquals("1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,19,20,22,23,25", network.dumpWeights());
   
    BasicNetwork model = EncogUtility.simpleFeedForward(1,3,0,4,false);
    checkWithModel(model.getStructure().getFlat(),network.getStructure().getFlat());
    Assert.assertEquals(1, network.getInputCount());
  }
 
  public void testPruneNeuronHidden()
  {
    BasicNetwork network = obtainNetwork();
    PruneSelective prune = new PruneSelective(network);
    prune.prune(1, 1);
    Assert.assertEquals(18, network.encodedArrayLength());
    Assert.assertEquals(2,network.getLayerNeuronCount(1));
    Assert.assertEquals("1,3,4,5,7,8,9,11,12,13,15,16,17,18,19,23,24,25", network.dumpWeights());
   
    BasicNetwork model = EncogUtility.simpleFeedForward(2,2,0,4,false);
    checkWithModel(model.getStructure().getFlat(),network.getStructure().getFlat());   
  }
 
  public void testPruneNeuronOutput()
  {
    BasicNetwork network = obtainNetwork();
    Assert.assertEquals(4, network.getOutputCount());
    PruneSelective prune = new PruneSelective(network);
    prune.prune(2, 1);
    Assert.assertEquals(21, network.encodedArrayLength());
    Assert.assertEquals(3,network.getLayerNeuronCount(2));
    Assert.assertEquals("1,2,3,4,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25", network.dumpWeights());
   
    BasicNetwork model = EncogUtility.simpleFeedForward(2,3,0,3,false);
    checkWithModel(model.getStructure().getFlat(),network.getStructure().getFlat());
    Assert.assertEquals(3, network.getOutputCount());
  }
 
  public void testNeuronSignificance()
  {
    BasicNetwork network = obtainNetwork();   
    PruneSelective prune = new PruneSelective(network);
    double inputSig = prune.determineNeuronSignificance(0, 1);
    double hiddenSig = prune.determineNeuronSignificance(1, 1);
    double outputSig = prune.determineNeuronSignificance(2, 1);
    Assert.assertEquals(63.0, inputSig,0.01);
    Assert.assertEquals(95.0, hiddenSig,0.01);
    Assert.assertEquals(26.0, outputSig,0.01);
  }
 
  public void testIncreaseNeuronCountHidden()
  {
    BasicNetwork network = XOR.createTrainedXOR();
    Assert.assertTrue( XOR.verifyXOR(network, 0.10) );
    PruneSelective prune = new PruneSelective(network);
    prune.changeNeuronCount(1, 5);
   
    BasicNetwork model = EncogUtility.simpleFeedForward(2,5,0,1,false);
    checkWithModel(model.getStructure().getFlat(),network.getStructure().getFlat());
   
    Assert.assertTrue( XOR.verifyXOR(network, 0.10) );
  }
 
  public void testIncreaseNeuronCountHidden2()
  {
    BasicNetwork network = EncogUtility.simpleFeedForward(5,6,0,2,true);
    PruneSelective prune = new PruneSelective(network);
    prune.changeNeuronCount(1, 60);
   
    BasicMLData input = new BasicMLData(5);
    BasicNetwork model = EncogUtility.simpleFeedForward(5,60,0,2,true);
    checkWithModel(model.getStructure().getFlat(),network.getStructure().getFlat());
    model.compute(input);
    network.compute(input);
  }
 
  public void testRandomizeNeuronInput()
  {
    double[] d = { 0,0,0,0,0,0,0,0,0,0,0,0,0 };
    BasicNetwork network = EncogUtility.simpleFeedForward(2,3,0,1,false);
    NetworkCODEC.arrayToNetwork(d, network);
    PruneSelective prune = new PruneSelective(network);
    prune.randomizeNeuron(100, 100, 0,1);
    Assert.assertEquals("0,0,0,0,0,100,0,0,100,0,0,100,0", network.dumpWeights());
  }
 
  public void testRandomizeNeuronHidden()
  {
    double[] d = { 0,0,0,0,0,0,0,0,0,0,0,0,0 };
    BasicNetwork network = EncogUtility.simpleFeedForward(2,3,0,1,false);
    NetworkCODEC.arrayToNetwork(d, network);
    PruneSelective prune = new PruneSelective(network);
    prune.randomizeNeuron(100, 100, 1,1);
    Assert.assertEquals("0,100,0,0,0,0,0,100,100,100,0,0,0", network.dumpWeights());
  }
 
  public void testRandomizeNeuronOutput()
  {
    double[] d = { 0,0,0,0,0,0,0,0,0,0,0,0,0 };
    BasicNetwork network = EncogUtility.simpleFeedForward(2,3,0,1,false);
    NetworkCODEC.arrayToNetwork(d, network);
    PruneSelective prune = new PruneSelective(network);
    prune.randomizeNeuron(100, 100, 2,0);
    Assert.assertEquals("100,100,100,100,0,0,0,0,0,0,0,0,0", network.dumpWeights());
  }
}
TOP

Related Classes of org.encog.neural.prune.TestPruneSelective

TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.