/**
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You 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.
*/
package org.apache.mahout.classifier.rbm.model;
import java.io.IOException;
import java.util.Random;
import org.apache.hadoop.conf.Configuration;
import org.apache.hadoop.fs.FSDataInputStream;
import org.apache.hadoop.fs.FSDataOutputStream;
import org.apache.hadoop.fs.FileSystem;
import org.apache.hadoop.fs.Path;
import org.apache.mahout.classifier.rbm.layer.Layer;
import org.apache.mahout.common.ClassUtils;
import org.apache.mahout.math.DenseMatrix;
import org.apache.mahout.math.Matrix;
import org.apache.mahout.math.MatrixWritable;
import org.apache.mahout.math.Vector;
import com.google.common.io.Closeables;
/**
* The Class SimpleRBM consisting of a visible and hidden layer and a weight matrix that connects them.
*/
public class SimpleRBM extends RBMModel {
/** The weight matrix. */
protected Matrix weightMatrix; //rownumber is visible unit, column is hidden
/**
* Instantiates a new simple rbm.
*
* @param visibleLayer the visible layer
* @param hiddenLayer the hidden layer
*/
public SimpleRBM(Layer visibleLayer, Layer hiddenLayer) {
this.visibleLayer = visibleLayer;
this.hiddenLayer = hiddenLayer;
// initialize the random number generator
Random rand = new Random();
this.weightMatrix = new DenseMatrix(visibleLayer.getNeuronCount(), hiddenLayer.getNeuronCount());
//small random values chosen from a zero-mean Gaussian with
//a standard deviation of about 0.01
for (int i = 0; i < weightMatrix.columnSize(); i++) {
for (int j = 0; j < weightMatrix.rowSize(); j++) {
weightMatrix.set(j, i, rand.nextGaussian()/100);
}
}
}
/**
* Instantiates a new simple rbm.
*
* @param visibleLayer the visible layer
* @param hiddenLayer the hidden layer
* @param weightMatrix the weight matrix
*/
public SimpleRBM(Layer visibleLayer, Layer hiddenLayer, Matrix weightMatrix) {
this.visibleLayer = visibleLayer;
this.hiddenLayer = hiddenLayer;
this.weightMatrix = weightMatrix;
}
/**
* Gets the weight matrix.
*
* @return the weight matrix
*/
public Matrix getWeightMatrix() {
return weightMatrix;
}
/* (non-Javadoc)
* @see org.apache.mahout.classifier.rbm.model.RBMModel#serialize(org.apache.hadoop.fs.Path, org.apache.hadoop.conf.Configuration)
*/
@Override
public void serialize(Path output, Configuration conf) throws IOException {
FileSystem fs = output.getFileSystem(conf);
String rbmnr = conf.get("rbmnr");
FSDataOutputStream out = null;
if(rbmnr!=null)
out = fs.create(new Path(output, conf.get("rbmnr")), true);
else
out = fs.create(output, true);
try {
out.writeChars(this.getClass().getName()+" ");
out.writeChars(visibleLayer.getClass().getName()+" ");
out.writeChars(hiddenLayer.getClass().getName()+" ");
MatrixWritable.writeMatrix(out, weightMatrix);
} finally {
Closeables.closeQuietly(out);
}
}
/**
* Materialize.
*
* @param output path to serialize to
* @param conf the hadoop config
* @return the rbm
* @throws IOException Signals that an I/O exception has occurred.
*/
public static SimpleRBM materialize(Path output,Configuration conf) throws IOException {
FileSystem fs = output.getFileSystem(conf);
Matrix weightMatrix;
String visLayerType = "";
String hidLayerType = "";
FSDataInputStream in = fs.open(output);
try {
char chr;
while((chr=in.readChar())!=' ');
while((chr=in.readChar())!=' ')
visLayerType += chr;
while((chr=in.readChar())!=' ')
hidLayerType += chr;
weightMatrix = MatrixWritable.readMatrix(in);
} finally {
Closeables.closeQuietly(in);
}
Layer vl = ClassUtils.instantiateAs(visLayerType, Layer.class,new Class[]{int.class},new Object[]{weightMatrix.rowSize()});
Layer hl = ClassUtils.instantiateAs(hidLayerType, Layer.class,new Class[]{int.class},new Object[]{weightMatrix.columnSize()});
SimpleRBM rbm = new SimpleRBM(vl, hl);
rbm.setWeightMatrix(weightMatrix);
return rbm;
}
/* (non-Javadoc)
* @see org.apache.mahout.classifier.rbm.model.RBMModel#exciteHiddenLayer(double, boolean)
*/
public void exciteHiddenLayer(double inputFactor, boolean addInput) {
Matrix activations = visibleLayer.getTransposedActivations();
Matrix input = activations.times(weightMatrix).times(inputFactor);
if(addInput)
hiddenLayer.addInputs(input.viewRow(0));
else
hiddenLayer.setInputs(input.viewRow(0));
hiddenLayer.exciteNeurons();
}
/* (non-Javadoc)
* @see org.apache.mahout.classifier.rbm.model.RBMModel#exciteVisibleLayer(double, boolean)
*/
public void exciteVisibleLayer(double inputFactor, boolean addInput) {
Vector input = weightMatrix.times(getHiddenLayer().getActivations()).times(inputFactor);
if(addInput)
visibleLayer.addInputs(input);
else
visibleLayer.setInputs(input);
visibleLayer.exciteNeurons();
}
/**
* Sets the weight matrix.
*
* @param weightMatrix the new weight matrix
*/
public void setWeightMatrix(Matrix weightMatrix) {
this.weightMatrix = weightMatrix;
}
/* (non-Javadoc)
* @see org.apache.mahout.classifier.rbm.model.RBMModel#clone()
*/
@Override
public RBMModel clone() {
SimpleRBM rbm = new SimpleRBM(visibleLayer.clone(), hiddenLayer.clone());
rbm.weightMatrix = weightMatrix.clone();
return rbm;
}
}