.TH "mlpack::nn::SparseAutoencoder< OptimizerType >" 3 "Tue Sep 9 2014" "Version 1.0.10" "MLPACK" \" -*- nroff -*-
.ad l
.nh
.SH NAME
mlpack::nn::SparseAutoencoder< OptimizerType > \- 
.PP
A sparse autoencoder is a neural network whose aim to learn compressed representations of the data, typically for dimensionality reduction, with a constraint on the activity of the neurons in the network\&.  

.SH SYNOPSIS
.br
.PP
.SS "Public Member Functions"

.in +1c
.ti -1c
.RI "\fBSparseAutoencoder\fP (const arma::mat &data, const size_t \fBvisibleSize\fP, const size_t \fBhiddenSize\fP, const double \fBlambda\fP=0\&.0001, const double \fBbeta\fP=3, const double \fBrho\fP=0\&.01)"
.br
.RI "\fIConstruct the sparse autoencoder model with the given training data\&. \fP"
.ti -1c
.RI "\fBSparseAutoencoder\fP (OptimizerType< \fBSparseAutoencoderFunction\fP > &optimizer)"
.br
.RI "\fIConstruct the sparse autoencoder model with the given training data\&. \fP"
.ti -1c
.RI "void \fBBeta\fP (const double b)"
.br
.RI "\fISets the KL divergence parameter\&. \fP"
.ti -1c
.RI "double \fBBeta\fP () const "
.br
.RI "\fIGets the KL divergence parameter\&. \fP"
.ti -1c
.RI "void \fBGetNewFeatures\fP (arma::mat &data, arma::mat &features)"
.br
.RI "\fITransforms the provided data into the representation learned by the sparse autoencoder\&. \fP"
.ti -1c
.RI "void \fBHiddenSize\fP (const size_t hidden)"
.br
.RI "\fISets size of the hidden layer\&. \fP"
.ti -1c
.RI "size_t \fBHiddenSize\fP () const "
.br
.RI "\fIGets the size of the hidden layer\&. \fP"
.ti -1c
.RI "void \fBLambda\fP (const double l)"
.br
.RI "\fISets the L2-regularization parameter\&. \fP"
.ti -1c
.RI "double \fBLambda\fP () const "
.br
.RI "\fIGets the L2-regularization parameter\&. \fP"
.ti -1c
.RI "void \fBRho\fP (const double r)"
.br
.RI "\fISets the sparsity parameter\&. \fP"
.ti -1c
.RI "double \fBRho\fP () const "
.br
.RI "\fIGets the sparsity parameter\&. \fP"
.ti -1c
.RI "void \fBSigmoid\fP (const arma::mat &x, arma::mat &output) const "
.br
.RI "\fIReturns the elementwise sigmoid of the passed matrix, where the sigmoid function of a real number 'x' is [1 / (1 + exp(-x))]\&. \fP"
.ti -1c
.RI "void \fBVisibleSize\fP (const size_t visible)"
.br
.RI "\fISets size of the visible layer\&. \fP"
.ti -1c
.RI "size_t \fBVisibleSize\fP () const "
.br
.RI "\fIGets size of the visible layer\&. \fP"
.in -1c
.SS "Private Attributes"

.in +1c
.ti -1c
.RI "double \fBbeta\fP"
.br
.RI "\fIKL divergence parameter\&. \fP"
.ti -1c
.RI "size_t \fBhiddenSize\fP"
.br
.RI "\fISize of the hidden layer\&. \fP"
.ti -1c
.RI "double \fBlambda\fP"
.br
.RI "\fIL2-regularization parameter\&. \fP"
.ti -1c
.RI "arma::mat \fBparameters\fP"
.br
.RI "\fIParameters after optimization\&. \fP"
.ti -1c
.RI "double \fBrho\fP"
.br
.RI "\fISparsity parameter\&. \fP"
.ti -1c
.RI "size_t \fBvisibleSize\fP"
.br
.RI "\fISize of the visible layer\&. \fP"
.in -1c
.SH "Detailed Description"
.PP 

.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS>class mlpack::nn::SparseAutoencoder< OptimizerType >"
A sparse autoencoder is a neural network whose aim to learn compressed representations of the data, typically for dimensionality reduction, with a constraint on the activity of the neurons in the network\&. 

Sparse autoencoders can be stacked together to learn a hierarchy of features, which provide a better representation of the data for classification\&. This is a method used in the recently developed field of deep learning\&. More technical details about the model can be found on the following webpage:
.PP
http://deeplearning.stanford.edu/wiki/index.php/UFLDL_Tutorial
.PP
An example of how to use the interface is shown below:
.PP
.PP
.nf
arma::mat data; // Data matrix\&.
const size_t vSize = 64; // Size of visible layer, depends on the data\&.
const size_t hSize = 25; // Size of hidden layer, depends on requirements\&.

// Train the model using default options\&.
SparseAutoencoder encoder1(data, vSize, hSize);

const size_t numBasis = 5; // Parameter required for L-BFGS algorithm\&.
const size_t numIterations = 100; // Maximum number of iterations\&.

// Use an instantiated optimizer for the training\&.
SparseAutoencoderFunction saf(data, vSize, hSize);
L_BFGS<SparseAutoencoderFunction> optimizer(saf, numBasis, numIterations);
SparseAutoencoder<L_BFGS> encoder2(optimizer);

arma::mat features1, features2; // Matrices for storing new representations\&.

// Get new representations from the trained models\&.
encoder1\&.GetNewFeatures(data, features1);
encoder2\&.GetNewFeatures(data, features2);
.fi
.PP
.PP
This implementation allows the use of arbitrary mlpack optimizers via the OptimizerType template parameter\&.
.PP
\fBTemplate Parameters:\fP
.RS 4
\fIOptimizerType\fP The optimizer to use; by default this is L-BFGS\&. Any mlpack optimizer can be used here\&. 
.RE
.PP

.PP
Definition at line 78 of file sparse_autoencoder\&.hpp\&.
.SH "Constructor & Destructor Documentation"
.PP 
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::\fBSparseAutoencoder\fP (const arma::mat &data, const size_tvisibleSize, const size_thiddenSize, const doublelambda = \fC0\&.0001\fP, const doublebeta = \fC3\fP, const doublerho = \fC0\&.01\fP)"

.PP
Construct the sparse autoencoder model with the given training data\&. This will train the model\&. The parameters 'lambda', 'beta' and 'rho' can be set optionally\&. Changing these parameters will have an effect on regularization and sparsity of the model\&.
.PP
\fBParameters:\fP
.RS 4
\fIdata\fP Input data with each column as one example\&. 
.br
\fIvisibleSize\fP Size of input vector expected at the visible layer\&. 
.br
\fIhiddenSize\fP Size of input vector expected at the hidden layer\&. 
.br
\fIlambda\fP L2-regularization parameter\&. 
.br
\fIbeta\fP KL divergence parameter\&. 
.br
\fIrho\fP Sparsity parameter\&. 
.RE
.PP

.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::\fBSparseAutoencoder\fP (OptimizerType< \fBSparseAutoencoderFunction\fP > &optimizer)"

.PP
Construct the sparse autoencoder model with the given training data\&. This will train the model\&. This overload takes an already instantiated optimizer and uses it to train the model\&. The optimizer should hold an instantiated \fBSparseAutoencoderFunction\fP object for the function to operate upon\&. This option should be preferred when the optimizer options are to be changed\&.
.PP
\fBParameters:\fP
.RS 4
\fIoptimizer\fP Instantiated optimizer with instantiated error function\&. 
.RE
.PP

.SH "Member Function Documentation"
.PP 
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Beta (const doubleb)\fC [inline]\fP"

.PP
Sets the KL divergence parameter\&. 
.PP
Definition at line 170 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::beta\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Beta () const\fC [inline]\fP"

.PP
Gets the KL divergence parameter\&. 
.PP
Definition at line 176 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::beta\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::GetNewFeatures (arma::mat &data, arma::mat &features)"

.PP
Transforms the provided data into the representation learned by the sparse autoencoder\&. The function basically performs a feedforward computation using the learned weights, and returns the hidden layer activations\&.
.PP
\fBParameters:\fP
.RS 4
\fIdata\fP Matrix of the provided data\&. 
.br
\fIfeatures\fP The hidden layer representation of the provided data\&. 
.RE
.PP

.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::HiddenSize (const size_thidden)\fC [inline]\fP"

.PP
Sets size of the hidden layer\&. 
.PP
Definition at line 146 of file sparse_autoencoder\&.hpp\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> size_t \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::HiddenSize () const\fC [inline]\fP"

.PP
Gets the size of the hidden layer\&. 
.PP
Definition at line 152 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::hiddenSize\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Lambda (const doublel)\fC [inline]\fP"

.PP
Sets the L2-regularization parameter\&. 
.PP
Definition at line 158 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::lambda\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Lambda () const\fC [inline]\fP"

.PP
Gets the L2-regularization parameter\&. 
.PP
Definition at line 164 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::lambda\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Rho (const doubler)\fC [inline]\fP"

.PP
Sets the sparsity parameter\&. 
.PP
Definition at line 182 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::rho\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Rho () const\fC [inline]\fP"

.PP
Gets the sparsity parameter\&. 
.PP
Definition at line 188 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::rho\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::Sigmoid (const arma::mat &x, arma::mat &output) const\fC [inline]\fP"

.PP
Returns the elementwise sigmoid of the passed matrix, where the sigmoid function of a real number 'x' is [1 / (1 + exp(-x))]\&. 
.PP
\fBParameters:\fP
.RS 4
\fIx\fP Matrix of real values for which we require the sigmoid activation\&. 
.RE
.PP

.PP
Definition at line 128 of file sparse_autoencoder\&.hpp\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> void \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::VisibleSize (const size_tvisible)\fC [inline]\fP"

.PP
Sets size of the visible layer\&. 
.PP
Definition at line 134 of file sparse_autoencoder\&.hpp\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> size_t \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::VisibleSize () const\fC [inline]\fP"

.PP
Gets size of the visible layer\&. 
.PP
Definition at line 140 of file sparse_autoencoder\&.hpp\&.
.PP
References mlpack::nn::SparseAutoencoder< OptimizerType >::visibleSize\&.
.SH "Member Data Documentation"
.PP 
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::beta\fC [private]\fP"

.PP
KL divergence parameter\&. 
.PP
Definition at line 203 of file sparse_autoencoder\&.hpp\&.
.PP
Referenced by mlpack::nn::SparseAutoencoder< OptimizerType >::Beta()\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> size_t \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::hiddenSize\fC [private]\fP"

.PP
Size of the hidden layer\&. 
.PP
Definition at line 199 of file sparse_autoencoder\&.hpp\&.
.PP
Referenced by mlpack::nn::SparseAutoencoder< OptimizerType >::HiddenSize()\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::lambda\fC [private]\fP"

.PP
L2-regularization parameter\&. 
.PP
Definition at line 201 of file sparse_autoencoder\&.hpp\&.
.PP
Referenced by mlpack::nn::SparseAutoencoder< OptimizerType >::Lambda()\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> arma::mat \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::parameters\fC [private]\fP"

.PP
Parameters after optimization\&. 
.PP
Definition at line 195 of file sparse_autoencoder\&.hpp\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> double \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::rho\fC [private]\fP"

.PP
Sparsity parameter\&. 
.PP
Definition at line 205 of file sparse_autoencoder\&.hpp\&.
.PP
Referenced by mlpack::nn::SparseAutoencoder< OptimizerType >::Rho()\&.
.SS "template<template< typename > class OptimizerType = mlpack::optimization::L_BFGS> size_t \fBmlpack::nn::SparseAutoencoder\fP< OptimizerType >::visibleSize\fC [private]\fP"

.PP
Size of the visible layer\&. 
.PP
Definition at line 197 of file sparse_autoencoder\&.hpp\&.
.PP
Referenced by mlpack::nn::SparseAutoencoder< OptimizerType >::VisibleSize()\&.

.SH "Author"
.PP 
Generated automatically by Doxygen for MLPACK from the source code\&.