jsat.classifiers.neuralnetwork

Class DReDNetSimple

java.lang.Object

jsat.classifiers.neuralnetwork.DReDNetSimple

All Implemented Interfaces:

java.io.Serializable, java.lang.Cloneable, Classifier, Parameterized

public class DReDNetSimple
extends java.lang.Object
implements Classifier, Parameterized

This class provides a neural network based on Geoffrey Hinton's Deep Rectified Dropout Nets. It is parameterized to be "simpler" in that the default batch size and gradient updating method should require no tuning to get decent results

NOTE: Training neural networks is computationally expensive, you may want to consider a GPU implementation from another source.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
DReDNetSimple() Creates a new DRedNet that uses two hidden layers with 1024 neurons each.
DReDNetSimple(int... hiddenLayerSizes) Create a new DReDNet that uses the specified number of hidden layers.

Method Summary

Modifier and Type	Method and Description
CategoricalResults	classify(DataPoint data) Performs classification on the given data point.
DReDNetSimple	clone()
int	getBatchSize()
int	getEpochs()
int[]	getHiddenSizes()
void	setBatchSize(int batchSize) Sets the batch size for updates
void	setEpochs(int epochs) Sets the number of epochs to perform
void	setHiddenSizes(int[] hiddenSizes) Sets the hidden layer sizes for this network.
boolean	supportsWeightedData() Indicates whether the model knows how to train using weighted data points.
void	train(ClassificationDataSet dataSet, boolean parallel) Trains the classifier and constructs a model for classification using the given data set.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Methods inherited from interface jsat.classifiers.Classifier

train

Methods inherited from interface jsat.parameters.Parameterized

getParameter, getParameters

Constructor Detail

DReDNetSimple

public DReDNetSimple()

Creates a new DRedNet that uses two hidden layers with 1024 neurons each. A batch size of 256 and 100 epochs will be used.

DReDNetSimple

public DReDNetSimple(int... hiddenLayerSizes)

Create a new DReDNet that uses the specified number of hidden layers. A batch size of 256 and 100 epochs will be used.

Parameters:

hiddenLayerSizes - the length indicates the number of hidden layers, and the value in each index is the number of neurons in that layer

Method Detail

setHiddenSizes

public void setHiddenSizes(int[] hiddenSizes)

Sets the hidden layer sizes for this network. The size of the array is the number of hidden layers and the value in each index denotes the size of that layer.

Parameters:

hiddenSizes -

getHiddenSizes

public int[] getHiddenSizes()

Returns:

the array of hidden layer sizes

setBatchSize

public void setBatchSize(int batchSize)

Sets the batch size for updates

Parameters:

batchSize - the number of items to compute the gradient from

getBatchSize

public int getBatchSize()

Returns:

the number of data points to use for one gradient computation

setEpochs

public void setEpochs(int epochs)

Sets the number of epochs to perform

Parameters:

epochs - the number of training iterations through the whole data set

getEpochs

public int getEpochs()

Returns:

the number of training iterations through the data set

classify

public CategoricalResults classify(DataPoint data)

Description copied from interface: Classifier

Performs classification on the given data point.

Specified by:

classify in interface Classifier

Parameters:

data - the data point to classify

Returns:

the results of the classification.

train

public void train(ClassificationDataSet dataSet,
                  boolean parallel)

Description copied from interface: Classifier

Trains the classifier and constructs a model for classification using the given data set. If the training method knows how, it will used the threadPool to conduct training in parallel. This method will block until the training has completed.

Specified by:

train in interface Classifier

Parameters:

dataSet - the data set to train on

parallel - true if multiple threads should be used to train the model. false if it should be done in a single threaded manner.

supportsWeightedData

public boolean supportsWeightedData()

Description copied from interface: Classifier

Indicates whether the model knows how to train using weighted data points. If it does, the model will train assuming the weights. The values returned by this method may change depending on the parameters set for the model.

Specified by:

supportsWeightedData in interface Classifier

Returns:

true if the model supports weighted data, false otherwise

clone

public DReDNetSimple clone()

Specified by:

clone in interface Classifier

Overrides:

clone in class java.lang.Object