Package ciir.umass.edu.learning.neuralnet

Class RankNet

java.lang.Object

ciir.umass.edu.learning.Ranker

ciir.umass.edu.learning.neuralnet.RankNet

Direct Known Subclasses:

LambdaRank, ListNet

public class RankNet
extends Ranker

Author:

vdang This class implements RankNet. C.J.C. Burges, T. Shaked, E. Renshaw, A. Lazier, M. Deeds, N. Hamilton and G. Hullender. Learning to rank using gradient descent. In Proc. of ICML, pages 89-96, 2005.

Field Summary

Fields

Modifier and Type	Field	Description
protected java.util.List<java.util.List<java.lang.Double>>	bestModelOnValidation
protected double	error
protected Layer	inputLayer
protected double	lastError
protected java.util.List<Layer>	layers
static double	learningRate
protected int	misorderedPairs
static int	nHiddenLayer
static int	nHiddenNodePerLayer
static int	nIteration
protected Layer	outputLayer
protected int	straightLoss
protected int	totalPairs

Fields inherited from class ciir.umass.edu.learning.Ranker

bestScoreOnValidationData, features, samples, scoreOnTrainingData, scorer, validationSamples, verbose

Constructor Summary

Constructors

Constructor	Description
RankNet()
RankNet(java.util.List<RankList> samples, int[] features, MetricScorer scorer)

Method Summary

Modifier and Type	Method	Description
protected void	addHiddenLayer(int size)
protected void	addInput(DataPoint p)	Auxiliary functions for pair-wise preference network learning.
protected void	batchBackPropagate(int[][] pairMap, float[][] pairWeight)
protected int[][]	batchFeedForward(RankList rl)
protected void	clearNeuronOutputs()
protected float[][]	computePairWeight(int[][] pairMap, RankList rl)
protected void	connect(int sourceLayer, int sourceNeuron, int targetLayer, int targetNeuron)
Ranker	createNew()
protected double	crossEntropy(double o1, double o2, double targetValue)
protected void	estimateLoss()
double	eval(DataPoint p)
void	init()	Main public functions
protected RankList	internalReorder(RankList rl)
void	learn()
void	loadFromString(java.lang.String fullText)
java.lang.String	model()
java.lang.String	name()
protected void	printNetworkConfig()	FOR DEBUGGING PURPOSE ONLY
void	printParameters()
protected void	printWeightVector()
protected void	propagate(int i)
protected void	restoreBestModelOnValidation()
protected void	saveBestModelOnValidation()	Model validation
protected void	setInputOutput(int nInput, int nOutput)	Setting up the Neural Network
protected void	setInputOutput(int nInput, int nOutput, int nType)
java.lang.String	toString()
protected void	wire()

Methods inherited from class ciir.umass.edu.learning.Ranker

copy, getFeatures, getScoreOnTrainingData, getScoreOnValidationData, PRINT, PRINT, PRINT_MEMORY_USAGE, PRINTLN, PRINTLN, PRINTTIME, rank, rank, save, setFeatures, setMetricScorer, setTrainingSet, setValidationSet

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Field Detail

nIteration

public static int nIteration

nHiddenLayer

public static int nHiddenLayer

nHiddenNodePerLayer

public static int nHiddenNodePerLayer

learningRate

public static double learningRate

layers

protected java.util.List<Layer> layers

inputLayer

protected Layer inputLayer

outputLayer

protected Layer outputLayer

bestModelOnValidation

protected java.util.List<java.util.List<java.lang.Double>> bestModelOnValidation

totalPairs

protected int totalPairs

misorderedPairs

protected int misorderedPairs

error

protected double error

lastError

protected double lastError

straightLoss

protected int straightLoss

Constructor Detail

RankNet

public RankNet()

RankNet

public RankNet(java.util.List<RankList> samples,
               int[] features,
               MetricScorer scorer)

Method Detail

setInputOutput

protected void setInputOutput(int nInput,
                              int nOutput)

Setting up the Neural Network

setInputOutput

protected void setInputOutput(int nInput,
                              int nOutput,
                              int nType)

addHiddenLayer

protected void addHiddenLayer(int size)

wire

protected void wire()

connect

protected void connect(int sourceLayer,
                       int sourceNeuron,
                       int targetLayer,
                       int targetNeuron)

addInput

protected void addInput(DataPoint p)

Auxiliary functions for pair-wise preference network learning.

propagate

protected void propagate(int i)

batchFeedForward

protected int[][] batchFeedForward(RankList rl)

batchBackPropagate

protected void batchBackPropagate(int[][] pairMap,
                                  float[][] pairWeight)

clearNeuronOutputs

protected void clearNeuronOutputs()

computePairWeight

protected float[][] computePairWeight(int[][] pairMap,
                                      RankList rl)

internalReorder

protected RankList internalReorder(RankList rl)

saveBestModelOnValidation

protected void saveBestModelOnValidation()

Model validation

restoreBestModelOnValidation

protected void restoreBestModelOnValidation()

crossEntropy

protected double crossEntropy(double o1,
                              double o2,
                              double targetValue)

estimateLoss

protected void estimateLoss()

init

public void init()

Main public functions

Specified by:

init in class Ranker

learn

public void learn()

Specified by:

learn in class Ranker

eval

public double eval(DataPoint p)

Overrides:

eval in class Ranker

createNew

public Ranker createNew()

Specified by:

createNew in class Ranker

toString

public java.lang.String toString()

Specified by:

toString in class Ranker

model

public java.lang.String model()

Specified by:

model in class Ranker

loadFromString

public void loadFromString(java.lang.String fullText)

Specified by:

loadFromString in class Ranker

printParameters

public void printParameters()

Specified by:

printParameters in class Ranker

name

public java.lang.String name()

Specified by:

name in class Ranker

printNetworkConfig

protected void printNetworkConfig()

FOR DEBUGGING PURPOSE ONLY

printWeightVector

protected void printWeightVector()