MLPAutoencoder

java.lang.Object
- weka.filters.Filter
- - weka.filters.SimpleFilter
  - - weka.filters.SimpleBatchFilter
    - - weka.filters.unsupervised.attribute.MLPAutoencoder

All Implemented Interfaces:

java.io.Serializable, weka.core.CapabilitiesHandler, weka.core.CapabilitiesIgnorer, weka.core.CommandlineRunnable, weka.core.OptionHandler, weka.core.RevisionHandler, weka.core.TechnicalInformationHandler, weka.filters.UnsupervisedFilter
```
public class MLPAutoencoder
extends weka.filters.SimpleBatchFilter
implements weka.filters.UnsupervisedFilter, weka.core.TechnicalInformationHandler
```
Implements an autoencoder with one hidden layer and tied weights using WEKA's Optimization class by minimizing the squared error plus a quadratic penalty (weight decay) with the BFGS method. Provides contractive autoencoder as an alternative to weight decay. Note that all attributes are standardized, including the target. There are several parameters. The lambda parameter is used to determine the penalty on the size of the weights. The number of hidden units can also be specified. Note that large numbers produce long training times. Finally, it is possible to use conjugate gradient descent rather than BFGS updates, which may be faster for cases with many parameters. To improve speed, an approximate version of the logistic function is used as the activation function. Also, if delta values in the backpropagation step are within the user-specified tolerance, the gradient is not updated for that particular instance, which saves some additional time. Paralled calculation of squared error and gradient is possible when multiple CPU cores are present. Data is split into batches and processed in separate threads in this case. Note that this only improves runtime for larger datasets. For more information on autoencoders, see

Salah Rifai, Pascal Vincent, Xavier Muller, Xavier Glorot, Yoshua Bengio: Contractive Auto-Encoders: Explicit Invariance During Feature Extraction. In: International Conference on Machine Learning, 833-840, 2011.
Valid options are:
```
 -N <int>
  Number of hidden units (default is 2).
 
```
```
 -L <double>
  Lambda factor for penalty on weights (default is 0.01).
 
```
```
 -O <double>
  Tolerance parameter for delta values (default is 1.0e-6).
 
```
```
 -G
  Use conjugate gradient descent (recommended for many attributes).
 
```
```
 -C
  Use contractive autoencoder instead of autoencoder with weight decay.
 
```
```
 -X
  Use exact sigmoid function rather than approximation.
 
```
```
 -F
  Output data in original space, so do not output reduced data.
 
```
```
 -P <int>
  The size of the thread pool, for example, the number of cores in the CPU. (default 1)
 
```
```
 -E <int>
  The number of threads to use, which should be >= size of thread pool. (default 1)
 
```
```
 -weights-file <filename>
  The file to write weight vectors to in ARFF format.
  (default: none)
 
```
```
 -S
  Whether to 0=normalize/1=standardize/2=neither. (default 1=standardize)
 
```
```
 -seed <int>
  The seed for the random number generator (default is 1).
 
```
Version:

$Revision: 9346 $

Author:

Eibe Frank (eibe@cs.waikato.ac.nz)

See Also:

Serialized Form

Field Summary

Fields
Modifier and Type Field and Description

static int FILTER_NONE

static int FILTER_NORMALIZE

static int FILTER_STANDARDIZE

static weka.core.Tag[] TAGS_FILTER

Fields
Modifier and Type	Field and Description
`static int`	`FILTER_NONE`
`static int`	`FILTER_NORMALIZE`
`static int`	`FILTER_STANDARDIZE`
`static weka.core.Tag[]`	`TAGS_FILTER`

Constructor Summary

Constructors
Constructor and Description

MLPAutoencoder()

Constructors
Constructor and Description
`MLPAutoencoder()`

Method Summary

All Methods Static Methods Instance Methods Concrete Methods
Modifier and Type	Method and Description
`java.lang.String`	`filterTypeTipText()` Returns the tip text for this property
`weka.core.Capabilities`	`getCapabilities()` Returns default capabilities of the filter.
`weka.core.SelectedTag`	`getFilterType()` Gets how the training data will be transformed.
`double`	`getLambda()` Gets the value of the lambda parameter.
`int`	`getNumFunctions()` Gets the number of functions.
`int`	`getNumThreads()` Gets the number of threads.
`java.lang.String[]`	`getOptions()` Gets the current settings of the Filter.
`boolean`	`getOutputInOriginalSpace()` Gets whether to use original space.
`int`	`getPoolSize()` Gets the number of threads.
`int`	`getSeed()` Gets the seed for the random number generations
`weka.core.TechnicalInformation`	`getTechnicalInformation()` Returns an instance of a TechnicalInformation object, containing detailed information about the technical background of this class, e.g., paper reference or book this class is based on.
`double`	`getTolerance()` Gets the tolerance parameter for the delta values.
`boolean`	`getUseCGD()` Gets whether to use CGD.
`boolean`	`getUseContractiveAutoencoder()` Gets whether to use ContractiveAutoencoder.
`boolean`	`getUseExactSigmoid()` Gets whether to use exact sigmoid.
`java.io.File`	`getWeightsFile()` Gets current weights file.
`java.lang.String`	`globalInfo()` This will return a string describing the filter.
`void`	`initFilter(weka.core.Instances data)` Builds the autoencoder network based on the given data.
`java.lang.String`	`lambdaTipText()`
`java.util.Enumeration<weka.core.Option>`	`listOptions()` Returns an enumeration describing the available options.
`static void`	`main(java.lang.String[] argv)` Main method to run the code from the command-line using the standard WEKA options.
`java.lang.String`	`numFunctionsTipText()`
`java.lang.String`	`numThreadsTipText()`
`java.lang.String`	`outputInOriginalSpaceTipText()`
`java.lang.String`	`poolSizeTipText()`
`java.lang.String`	`seedTipText()` Returns the tip text for this property
`void`	`setFilterType(weka.core.SelectedTag newType)` Sets how the training data will be transformed.
`void`	`setLambda(double newLambda)` Sets the value of the lambda parameter.
`void`	`setNumFunctions(int newNumFunctions)` Sets the number of functions.
`void`	`setNumThreads(int nT)` Sets the number of threads
`void`	`setOptions(java.lang.String[] options)` Parses a given list of options.
`void`	`setOutputInOriginalSpace(boolean newOutputInOriginalSpace)` Sets whether to use original space.
`void`	`setPoolSize(int nT)` Sets the number of threads
`void`	`setSeed(int seed)` Set the seed for random number generation.
`void`	`setTolerance(double newTolerance)` Sets the tolerance parameter for the delta values.
`void`	`setUseCGD(boolean newUseCGD)` Sets whether to use CGD.
`void`	`setUseContractiveAutoencoder(boolean newUseContractiveAutoencoder)` Sets whether to use ContractiveAutoencoder.
`void`	`setUseExactSigmoid(boolean newUseExactSigmoid)` Sets whether to use exact sigmoid.
`void`	`setWeightsFile(java.io.File value)` Sets the weights file to use.
`java.lang.String`	`toleranceTipText()`
`java.lang.String`	`useCGDTipText()`
`java.lang.String`	`useContractiveAutoencoderTipText()`
`java.lang.String`	`useExactSigmoidTipText()`
`java.lang.String`	`weightsFileTipText()` Returns the tip text for this property.

Methods inherited from class weka.filters.SimpleBatchFilter
allowAccessToFullInputFormat, batchFinished, input, input

Methods inherited from class weka.filters.SimpleFilter
setInputFormat

Methods inherited from class weka.filters.Filter
batchFilterFile, debugTipText, doNotCheckCapabilitiesTipText, filterFile, getCapabilities, getCopyOfInputFormat, getDebug, getDoNotCheckCapabilities, getOutputFormat, getRevision, isFirstBatchDone, isNewBatch, isOutputFormatDefined, makeCopies, makeCopy, mayRemoveInstanceAfterFirstBatchDone, numPendingOutput, output, outputPeek, postExecution, preExecution, run, runFilter, setDebug, setDoNotCheckCapabilities, toString, useFilter, wekaStaticWrapper

Methods inherited from class java.lang.Object
equals, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface weka.core.OptionHandler
makeCopy

- Field Detail
  - FILTER_NORMALIZE
```
public static final int FILTER_NORMALIZE
```
    See Also:
    
    Constant Field Values
  - FILTER_STANDARDIZE
```
public static final int FILTER_STANDARDIZE
```
    See Also:
    
    Constant Field Values
  - FILTER_NONE
```
public static final int FILTER_NONE
```
    See Also:
    
    Constant Field Values
  - TAGS_FILTER
```
public static final weka.core.Tag[] TAGS_FILTER
```
- Constructor Detail
  - MLPAutoencoder
```
public MLPAutoencoder()
```
- Method Detail
  - getCapabilities
```
public weka.core.Capabilities getCapabilities()
```
    Returns default capabilities of the filter.
    
    Specified by:
    
    getCapabilities in interface weka.core.CapabilitiesHandler
    
    Overrides:
    
    getCapabilities in class weka.filters.Filter
    
    Returns:
    
    the capabilities of this filter
  - initFilter
```
public void initFilter(weka.core.Instances data)
                throws java.lang.Exception
```
    Builds the autoencoder network based on the given data.
    
    Throws:
    
    java.lang.Exception
  - globalInfo
```
public java.lang.String globalInfo()
```
    This will return a string describing the filter.
    
    Specified by:
    
    globalInfo in class weka.filters.SimpleFilter
    
    Returns:
    
    The string.
  - getTechnicalInformation
```
public weka.core.TechnicalInformation getTechnicalInformation()
```
    Returns an instance of a TechnicalInformation object, containing detailed information about the technical background of this class, e.g., paper reference or book this class is based on.
    
    Specified by:
    
    getTechnicalInformation in interface weka.core.TechnicalInformationHandler
    
    Returns:
    
    the technical information about this class
  - seedTipText
```
public java.lang.String seedTipText()
```
    Returns the tip text for this property
    
    Returns:
    
    tip text for this property suitable for displaying in the explorer/experimenter gui
  - setSeed
```
public void setSeed(int seed)
```
    Set the seed for random number generation.
    
    Parameters:
    
    seed - the seed
  - getSeed
```
public int getSeed()
```
    Gets the seed for the random number generations
    
    Returns:
    
    the seed for the random number generation
  - toleranceTipText
```
public java.lang.String toleranceTipText()
```
    Returns:
    
    a string to describe the option
  - getTolerance
```
public double getTolerance()
```
    Gets the tolerance parameter for the delta values.
  - setTolerance
```
public void setTolerance(double newTolerance)
```
    Sets the tolerance parameter for the delta values.
  - numFunctionsTipText
```
public java.lang.String numFunctionsTipText()
```
    Returns:
    
    a string to describe the option
  - getNumFunctions
```
public int getNumFunctions()
```
    Gets the number of functions.
  - setNumFunctions
```
public void setNumFunctions(int newNumFunctions)
```
    Sets the number of functions.
  - lambdaTipText
```
public java.lang.String lambdaTipText()
```
    Returns:
    
    a string to describe the option
  - getLambda
```
public double getLambda()
```
    Gets the value of the lambda parameter.
  - setLambda
```
public void setLambda(double newLambda)
```
    Sets the value of the lambda parameter.
  - useCGDTipText
```
public java.lang.String useCGDTipText()
```
    Returns:
    
    a string to describe the option
  - getUseCGD
```
public boolean getUseCGD()
```
    Gets whether to use CGD.
  - setUseCGD
```
public void setUseCGD(boolean newUseCGD)
```
    Sets whether to use CGD.
  - useContractiveAutoencoderTipText
```
public java.lang.String useContractiveAutoencoderTipText()
```
    Returns:
    
    a string to describe the option
  - getUseContractiveAutoencoder
```
public boolean getUseContractiveAutoencoder()
```
    Gets whether to use ContractiveAutoencoder.
  - setUseContractiveAutoencoder
```
public void setUseContractiveAutoencoder(boolean newUseContractiveAutoencoder)
```
    Sets whether to use ContractiveAutoencoder.
  - useExactSigmoidTipText
```
public java.lang.String useExactSigmoidTipText()
```
    Returns:
    
    a string to describe the option
  - getUseExactSigmoid
```
public boolean getUseExactSigmoid()
```
    Gets whether to use exact sigmoid.
  - setUseExactSigmoid
```
public void setUseExactSigmoid(boolean newUseExactSigmoid)
```
    Sets whether to use exact sigmoid.
  - outputInOriginalSpaceTipText
```
public java.lang.String outputInOriginalSpaceTipText()
```
    Returns:
    
    a string to describe the option
  - getOutputInOriginalSpace
```
public boolean getOutputInOriginalSpace()
```
    Gets whether to use original space.
  - setOutputInOriginalSpace
```
public void setOutputInOriginalSpace(boolean newOutputInOriginalSpace)
```
    Sets whether to use original space.
  - numThreadsTipText
```
public java.lang.String numThreadsTipText()
```
    Returns:
    
    a string to describe the option
  - getNumThreads
```
public int getNumThreads()
```
    Gets the number of threads.
  - setNumThreads
```
public void setNumThreads(int nT)
```
    Sets the number of threads
  - poolSizeTipText
```
public java.lang.String poolSizeTipText()
```
    Returns:
    
    a string to describe the option
  - getPoolSize
```
public int getPoolSize()
```
    Gets the number of threads.
  - setPoolSize
```
public void setPoolSize(int nT)
```
    Sets the number of threads
  - weightsFileTipText
```
public java.lang.String weightsFileTipText()
```
    Returns the tip text for this property.
    
    Returns:
    
    tip text for this property suitable for displaying in the explorer/experimenter gui
  - getWeightsFile
```
public java.io.File getWeightsFile()
```
    Gets current weights file.
    
    Returns:
    
    the weights file.
  - setWeightsFile
```
public void setWeightsFile(java.io.File value)
```
    Sets the weights file to use.
    
    Parameters:
    
    value - the weights file.
  - filterTypeTipText
```
public java.lang.String filterTypeTipText()
```
    Returns the tip text for this property
    
    Returns:
    
    tip text for this property suitable for displaying in the explorer/experimenter gui
  - getFilterType
```
public weka.core.SelectedTag getFilterType()
```
    Gets how the training data will be transformed. Will be one of FILTER_NORMALIZE, FILTER_STANDARDIZE, FILTER_NONE.
    
    Returns:
    
    the filtering mode
  - setFilterType
```
public void setFilterType(weka.core.SelectedTag newType)
```
    Sets how the training data will be transformed. Should be one of FILTER_NORMALIZE, FILTER_STANDARDIZE, FILTER_NONE.
    
    Parameters:
    
    newType - the new filtering mode
  - listOptions
```
public java.util.Enumeration<weka.core.Option> listOptions()
```
    Returns an enumeration describing the available options.
    
    Specified by:
    
    listOptions in interface weka.core.OptionHandler
    
    Overrides:
    
    listOptions in class weka.filters.Filter
    
    Returns:
    
    an enumeration of all the available options.
  - setOptions
```
public void setOptions(java.lang.String[] options)
                throws java.lang.Exception
```
    Parses a given list of options.
    Valid options are:
```
 -N <int>
  Number of hidden units (default is 2).
 
```
```
 -L <double>
  Lambda factor for penalty on weights (default is 0.01).
 
```
```
 -O <double>
  Tolerance parameter for delta values (default is 1.0e-6).
 
```
```
 -G
  Use conjugate gradient descent (recommended for many attributes).
 
```
```
 -C
  Use contractive autoencoder instead of autoencoder with weight decay.
 
```
```
 -X
  Use exact sigmoid function rather than approximation.
 
```
```
 -F
  Output data in original space, so do not output reduced data.
 
```
```
 -P <int>
  The size of the thread pool, for example, the number of cores in the CPU. (default 1)
 
```
```
 -E <int>
  The number of threads to use, which should be >= size of thread pool. (default 1)
 
```
```
 -weights-file <filename>
  The file to write weight vectors to in ARFF format.
  (default: none)
 
```
```
 -S
  Whether to 0=normalize/1=standardize/2=neither. (default 1=standardize)
 
```
```
 -seed <int>
  The seed for the random number generator (default is 1).
 
```
    Options after -- are passed to the designated filter.
    Specified by:
    
    setOptions in interface weka.core.OptionHandler
    
    Overrides:
    
    setOptions in class weka.filters.Filter
    
    Parameters:
    
    options - the list of options as an array of strings
    
    Throws:
    
    java.lang.Exception - if an option is not supported
  - getOptions
```
public java.lang.String[] getOptions()
```
    Gets the current settings of the Filter.
    
    Specified by:
    
    getOptions in interface weka.core.OptionHandler
    
    Overrides:
    
    getOptions in class weka.filters.Filter
    
    Returns:
    
    an array of strings suitable for passing to setOptions
  - main
```
public static void main(java.lang.String[] argv)
```
    Main method to run the code from the command-line using the standard WEKA options.

Class MLPAutoencoder

Field Summary

Constructor Summary

Method Summary

Methods inherited from class weka.filters.SimpleBatchFilter

Methods inherited from class weka.filters.SimpleFilter

Methods inherited from class weka.filters.Filter

Methods inherited from class java.lang.Object

Methods inherited from interface weka.core.OptionHandler

Field Detail

FILTER_NORMALIZE

FILTER_STANDARDIZE

FILTER_NONE

TAGS_FILTER

Constructor Detail

MLPAutoencoder

Method Detail

getCapabilities

initFilter

globalInfo

getTechnicalInformation

seedTipText

setSeed

getSeed

toleranceTipText

getTolerance

setTolerance

numFunctionsTipText

getNumFunctions

setNumFunctions

lambdaTipText

getLambda

setLambda

useCGDTipText

getUseCGD

setUseCGD

useContractiveAutoencoderTipText

getUseContractiveAutoencoder

setUseContractiveAutoencoder

useExactSigmoidTipText

getUseExactSigmoid

setUseExactSigmoid

outputInOriginalSpaceTipText

getOutputInOriginalSpace

setOutputInOriginalSpace

numThreadsTipText

getNumThreads

setNumThreads

poolSizeTipText

getPoolSize

setPoolSize

weightsFileTipText

getWeightsFile

setWeightsFile

filterTypeTipText

getFilterType

setFilterType

listOptions

setOptions

getOptions

main