SHOGUN  3.2.1
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Macros Groups Pages
List of all members | Public Member Functions | Public Attributes | Protected Member Functions | Protected Attributes
CDeepBeliefNetwork Class Reference

Detailed Description

A Deep Belief Network.

A Deep Belief Network [Hinton, 2006] is a multilayer probabilistic generative models. It consists of hidden layers and visible layers. The top hidden layer and the layer below it form a Restricted Boltzmann Machine. The rest of connections in the network are directed connections that go from a hidden layer into a visible layer or another hidden layer.

The network can be pre-trained by treating it as a stack of RBMs. Each hidden layer along with the layer below it form an RBM. Each RBM is then trained using (persistent) contrastive divergence. Pre-training often provides a good initialization for the network's parameters.

After pre-training, the parameters can be fine-tuned using a variant of the wake-sleep algorithm [Hinton, 2006].

The DBN can be used to initialize the parameters of a neural network using convert_to_neural_network().

Samples can be drawn from the model by starting with a random state for the top hidden layer, performing some steps of Gibbs sampling in the top RBM to obtain the states of the top hidden layer and then using those to infer the states of the lower layers using a down-pass.

Steps for using the DBN class:

Definition at line 91 of file DeepBeliefNetwork.h.

Inheritance diagram for CDeepBeliefNetwork:
Inheritance graph
[legend]

Public Member Functions

 CDeepBeliefNetwork ()
 CDeepBeliefNetwork (int32_t num_visible_units, ERBMVisibleUnitType unit_type=RBMVUT_BINARY)
virtual ~CDeepBeliefNetwork ()
virtual void add_hidden_layer (int32_t num_units)
virtual void initialize (float64_t sigma=0.01)
virtual void set_batch_size (int32_t batch_size)
virtual void pre_train (CDenseFeatures< float64_t > *features)
virtual void pre_train (int32_t index, CDenseFeatures< float64_t > *features)
virtual void train (CDenseFeatures< float64_t > *features)
virtual CDenseFeatures
< float64_t > * 
transform (CDenseFeatures< float64_t > *features, int32_t i=-1)
virtual CDenseFeatures
< float64_t > * 
sample (int32_t num_gibbs_steps=1, int32_t batch_size=1)
virtual void reset_chain ()
virtual CNeuralNetworkconvert_to_neural_network (CNeuralLayer *output_layer=NULL, float64_t sigma=0.01)
virtual SGMatrix< float64_tget_weights (int32_t index, SGVector< float64_t > p=SGVector< float64_t >())
virtual SGVector< float64_tget_biases (int32_t index, SGVector< float64_t > p=SGVector< float64_t >())
virtual const char * get_name () const
virtual CSGObjectshallow_copy () const
virtual CSGObjectdeep_copy () const
virtual bool is_generic (EPrimitiveType *generic) const
template<class T >
void set_generic ()
void unset_generic ()
virtual void print_serializable (const char *prefix="")
virtual bool save_serializable (CSerializableFile *file, const char *prefix="", int32_t param_version=Version::get_version_parameter())
virtual bool load_serializable (CSerializableFile *file, const char *prefix="", int32_t param_version=Version::get_version_parameter())
DynArray< TParameter * > * load_file_parameters (const SGParamInfo *param_info, int32_t file_version, CSerializableFile *file, const char *prefix="")
DynArray< TParameter * > * load_all_file_parameters (int32_t file_version, int32_t current_version, CSerializableFile *file, const char *prefix="")
void map_parameters (DynArray< TParameter * > *param_base, int32_t &base_version, DynArray< const SGParamInfo * > *target_param_infos)
void set_global_io (SGIO *io)
SGIOget_global_io ()
void set_global_parallel (Parallel *parallel)
Parallelget_global_parallel ()
void set_global_version (Version *version)
Versionget_global_version ()
SGStringList< char > get_modelsel_names ()
void print_modsel_params ()
char * get_modsel_param_descr (const char *param_name)
index_t get_modsel_param_index (const char *param_name)
void build_gradient_parameter_dictionary (CMap< TParameter *, CSGObject * > *dict)
virtual void update_parameter_hash ()
virtual bool parameter_hash_changed ()
virtual bool equals (CSGObject *other, float64_t accuracy=0.0, bool tolerant=false)
virtual CSGObjectclone ()

Public Attributes

SGVector< int32_t > pt_cd_num_steps
SGVector< bool > pt_cd_persistent
SGVector< bool > pt_cd_sample_visible
SGVector< float64_tpt_l2_coefficient
SGVector< float64_tpt_l1_coefficient
SGVector< int32_t > pt_monitoring_interval
SGVector< int32_t > pt_monitoring_method
SGVector< int32_t > pt_max_num_epochs
SGVector< int32_t > pt_gd_mini_batch_size
SGVector< float64_tpt_gd_learning_rate
SGVector< float64_tpt_gd_learning_rate_decay
SGVector< float64_tpt_gd_momentum
int32_t cd_num_steps
int32_t monitoring_interval
int32_t max_num_epochs
int32_t gd_mini_batch_size
float64_t gd_learning_rate
float64_t gd_learning_rate_decay
float64_t gd_momentum
SGIOio
Parallelparallel
Versionversion
Parameterm_parameters
Parameterm_model_selection_parameters
Parameterm_gradient_parameters
ParameterMapm_parameter_map
uint32_t m_hash

Protected Member Functions

virtual void down_step (int32_t index, SGVector< float64_t > params, SGMatrix< float64_t > input, SGMatrix< float64_t > result, bool sample_states=true)
virtual void up_step (int32_t index, SGVector< float64_t > params, SGMatrix< float64_t > input, SGMatrix< float64_t > result, bool sample_states=true)
virtual void wake_sleep (SGMatrix< float64_t > data, CRBM *top_rbm, SGMatrixList< float64_t > sleep_states, SGMatrixList< float64_t > wake_states, SGMatrixList< float64_t > psleep_states, SGMatrixList< float64_t > pwake_states, SGVector< float64_t > gen_params, SGVector< float64_t > rec_params, SGVector< float64_t > gen_gradients, SGVector< float64_t > rec_gradients)
virtual TParametermigrate (DynArray< TParameter * > *param_base, const SGParamInfo *target)
virtual void one_to_one_migration_prepare (DynArray< TParameter * > *param_base, const SGParamInfo *target, TParameter *&replacement, TParameter *&to_migrate, char *old_name=NULL)
virtual void load_serializable_pre () throw (ShogunException)
virtual void load_serializable_post () throw (ShogunException)
virtual void save_serializable_pre () throw (ShogunException)
virtual void save_serializable_post () throw (ShogunException)

Protected Attributes

ERBMVisibleUnitType m_visible_units_type
int32_t m_num_layers
CDynamicArray< int32_t > * m_layer_sizes
SGMatrixList< float64_tm_states
int32_t m_batch_size
SGVector< float64_tm_params
int32_t m_num_params
SGVector< int32_t > m_bias_index_offsets
SGVector< int32_t > m_weights_index_offsets
float64_t m_sigma

Constructor & Destructor Documentation

default constructor

Definition at line 51 of file DeepBeliefNetwork.cpp.

CDeepBeliefNetwork ( int32_t  num_visible_units,
ERBMVisibleUnitType  unit_type = RBMVUT_BINARY 
)

Creates a network with one layer of visible units

Parameters
num_visible_unitsNumber of visible units
unit_typeType of visible units

Definition at line 56 of file DeepBeliefNetwork.cpp.

~CDeepBeliefNetwork ( )
virtual

Definition at line 65 of file DeepBeliefNetwork.cpp.

Member Function Documentation

void add_hidden_layer ( int32_t  num_units)
virtual

Adds a layer of hidden units. The layer is connected to the layer that was added directly before it.

Parameters
num_unitsNumber of hidden units

Definition at line 70 of file DeepBeliefNetwork.cpp.

void build_gradient_parameter_dictionary ( CMap< TParameter *, CSGObject * > *  dict)
inherited

Builds a dictionary of all parameters in SGObject as well of those of SGObjects that are parameters of this object. Dictionary maps parameters to the objects that own them.

Parameters
dictdictionary of parameters to be built.

Definition at line 1243 of file SGObject.cpp.

CSGObject * clone ( )
virtualinherited

Creates a clone of the current object. This is done via recursively traversing all parameters, which corresponds to a deep copy. Calling equals on the cloned object always returns true although none of the memory of both objects overlaps.

Returns
an identical copy of the given object, which is disjoint in memory. NULL if the clone fails. Note that the returned object is SG_REF'ed

Definition at line 1360 of file SGObject.cpp.

CNeuralNetwork * convert_to_neural_network ( CNeuralLayer output_layer = NULL,
float64_t  sigma = 0.01 
)
virtual

Converts the DBN into a neural network with the same structure and parameters. The visible layer in the DBN is converted into a CNeuralInputLayer object, and the hidden layers are converted into CNeuralLogisticLayer objects. An output layer can also be stacked on top of the last hidden layer

Parameters
output_layerAn output layer
sigmaStandard deviation of the gaussian used to initialize the parameters of the output layer
Returns
Neural network inititialized using the DBN

Definition at line 360 of file DeepBeliefNetwork.cpp.

CSGObject * deep_copy ( ) const
virtualinherited

A deep copy. All the instance variables will also be copied.

Definition at line 200 of file SGObject.cpp.

void down_step ( int32_t  index,
SGVector< float64_t params,
SGMatrix< float64_t input,
SGMatrix< float64_t result,
bool  sample_states = true 
)
protectedvirtual

Computes the states of some layer using the states of the layer above it

Definition at line 393 of file DeepBeliefNetwork.cpp.

bool equals ( CSGObject other,
float64_t  accuracy = 0.0,
bool  tolerant = false 
)
virtualinherited

Recursively compares the current SGObject to another one. Compares all registered numerical parameters, recursion upon complex (SGObject) parameters. Does not compare pointers!

May be overwritten but please do with care! Should not be necessary in most cases.

Parameters
otherobject to compare with
accuracyaccuracy to use for comparison (optional)
tolerantallows linient check on float equality (within accuracy)
Returns
true if all parameters were equal, false if not

Definition at line 1264 of file SGObject.cpp.

SGVector< float64_t > get_biases ( int32_t  index,
SGVector< float64_t p = SGVector<float64_t>() 
)
virtual

Returns the bias vector of layer i

Parameters
indexLayer index
pIf specified, the bias vector is extracted from it instead of m_params

Definition at line 558 of file DeepBeliefNetwork.cpp.

SGIO * get_global_io ( )
inherited

get the io object

Returns
io object

Definition at line 237 of file SGObject.cpp.

Parallel * get_global_parallel ( )
inherited

get the parallel object

Returns
parallel object

Definition at line 278 of file SGObject.cpp.

Version * get_global_version ( )
inherited

get the version object

Returns
version object

Definition at line 291 of file SGObject.cpp.

SGStringList< char > get_modelsel_names ( )
inherited
Returns
vector of names of all parameters which are registered for model selection

Definition at line 1135 of file SGObject.cpp.

char * get_modsel_param_descr ( const char *  param_name)
inherited

Returns description of a given parameter string, if it exists. SG_ERROR otherwise

Parameters
param_namename of the parameter
Returns
description of the parameter

Definition at line 1159 of file SGObject.cpp.

index_t get_modsel_param_index ( const char *  param_name)
inherited

Returns index of model selection parameter with provided index

Parameters
param_namename of model selection parameter
Returns
index of model selection parameter with provided name, -1 if there is no such

Definition at line 1172 of file SGObject.cpp.

virtual const char* get_name ( ) const
virtual

Returns the name of the SGSerializable instance. It MUST BE the CLASS NAME without the prefixed `C'.

Returns
name of the SGSerializable

Implements CSGObject.

Definition at line 214 of file DeepBeliefNetwork.h.

SGMatrix< float64_t > get_weights ( int32_t  index,
SGVector< float64_t p = SGVector<float64_t>() 
)
virtual

Returns the weights matrix between layer i and i+1

Parameters
indexLayer index
pIf specified, the weight matrix is extracted from it instead of m_params

Definition at line 547 of file DeepBeliefNetwork.cpp.

void initialize ( float64_t  sigma = 0.01)
virtual

Initializes the DBN

Parameters
sigmaStandard deviation of the gaussian used to initialize the weights

Definition at line 76 of file DeepBeliefNetwork.cpp.

bool is_generic ( EPrimitiveType *  generic) const
virtualinherited

If the SGSerializable is a class template then TRUE will be returned and GENERIC is set to the type of the generic.

Parameters
genericset to the type of the generic if returning TRUE
Returns
TRUE if a class template.

Definition at line 297 of file SGObject.cpp.

DynArray< TParameter * > * load_all_file_parameters ( int32_t  file_version,
int32_t  current_version,
CSerializableFile file,
const char *  prefix = "" 
)
inherited

maps all parameters of this instance to the provided file version and loads all parameter data from the file into an array, which is sorted (basically calls load_file_parameter(...) for all parameters and puts all results into a sorted array)

Parameters
file_versionparameter version of the file
current_versionversion from which mapping begins (you want to use Version::get_version_parameter() for this in most cases)
filefile to load from
prefixprefix for members
Returns
(sorted) array of created TParameter instances with file data

Definition at line 704 of file SGObject.cpp.

DynArray< TParameter * > * load_file_parameters ( const SGParamInfo param_info,
int32_t  file_version,
CSerializableFile file,
const char *  prefix = "" 
)
inherited

loads some specified parameters from a file with a specified version The provided parameter info has a version which is recursively mapped until the file parameter version is reached. Note that there may be possibly multiple parameters in the mapping, therefore, a set of TParameter instances is returned

Parameters
param_infoinformation of parameter
file_versionparameter version of the file, must be <= provided parameter version
filefile to load from
prefixprefix for members
Returns
new array with TParameter instances with the attached data

Definition at line 545 of file SGObject.cpp.

bool load_serializable ( CSerializableFile file,
const char *  prefix = "",
int32_t  param_version = Version::get_version_parameter() 
)
virtualinherited

Load this object from file. If it will fail (returning FALSE) then this object will contain inconsistent data and should not be used!

Parameters
filewhere to load from
prefixprefix for members
param_version(optional) a parameter version different to (this is mainly for testing, better do not use)
Returns
TRUE if done, otherwise FALSE

Definition at line 374 of file SGObject.cpp.

void load_serializable_post ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to post-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::LOAD_SERIALIZABLE_POST is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel, CWeightedDegreePositionStringKernel, CList, CAlphabet, CLinearHMM, CGaussianKernel, CInverseMultiQuadricKernel, CCircularKernel, and CExponentialKernel.

Definition at line 1062 of file SGObject.cpp.

void load_serializable_pre ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to pre-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::LOAD_SERIALIZABLE_PRE is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CDynamicArray< T >, CDynamicArray< float64_t >, CDynamicArray< float32_t >, CDynamicArray< int32_t >, CDynamicArray< char >, CDynamicArray< bool >, and CDynamicObjectArray.

Definition at line 1057 of file SGObject.cpp.

void map_parameters ( DynArray< TParameter * > *  param_base,
int32_t &  base_version,
DynArray< const SGParamInfo * > *  target_param_infos 
)
inherited

Takes a set of TParameter instances (base) with a certain version and a set of target parameter infos and recursively maps the base level wise to the current version using CSGObject::migrate(...). The base is replaced. After this call, the base version containing parameters should be of same version/type as the initial target parameter infos. Note for this to work, the migrate methods and all the internal parameter mappings have to match

Parameters
param_baseset of TParameter instances that are mapped to the provided target parameter infos
base_versionversion of the parameter base
target_param_infosset of SGParamInfo instances that specify the target parameter base

Definition at line 742 of file SGObject.cpp.

TParameter * migrate ( DynArray< TParameter * > *  param_base,
const SGParamInfo target 
)
protectedvirtualinherited

creates a new TParameter instance, which contains migrated data from the version that is provided. The provided parameter data base is used for migration, this base is a collection of all parameter data of the previous version. Migration is done FROM the data in param_base TO the provided param info Migration is always one version step. Method has to be implemented in subclasses, if no match is found, base method has to be called.

If there is an element in the param_base which equals the target, a copy of the element is returned. This represents the case when nothing has changed and therefore, the migrate method is not overloaded in a subclass

Parameters
param_baseset of TParameter instances to use for migration
targetparameter info for the resulting TParameter
Returns
a new TParameter instance with migrated data from the base of the type which is specified by the target parameter

Definition at line 949 of file SGObject.cpp.

void one_to_one_migration_prepare ( DynArray< TParameter * > *  param_base,
const SGParamInfo target,
TParameter *&  replacement,
TParameter *&  to_migrate,
char *  old_name = NULL 
)
protectedvirtualinherited

This method prepares everything for a one-to-one parameter migration. One to one here means that only ONE element of the parameter base is needed for the migration (the one with the same name as the target). Data is allocated for the target (in the type as provided in the target SGParamInfo), and a corresponding new TParameter instance is written to replacement. The to_migrate pointer points to the single needed TParameter instance needed for migration. If a name change happened, the old name may be specified by old_name. In addition, the m_delete_data flag of to_migrate is set to true. So if you want to migrate data, the only thing to do after this call is converting the data in the m_parameter fields. If unsure how to use - have a look into an example for this. (base_migration_type_conversion.cpp for example)

Parameters
param_baseset of TParameter instances to use for migration
targetparameter info for the resulting TParameter
replacement(used as output) here the TParameter instance which is returned by migration is created into
to_migratethe only source that is used for migration
old_namewith this parameter, a name change may be specified

Definition at line 889 of file SGObject.cpp.

bool parameter_hash_changed ( )
virtualinherited
Returns
whether parameter combination has changed since last update

Definition at line 263 of file SGObject.cpp.

void pre_train ( CDenseFeatures< float64_t > *  features)
virtual

Pre-trains the DBN as a stack of RBMs

Parameters
featuresInput features. Should have as many features as the number of visible units in the DBN

Definition at line 149 of file DeepBeliefNetwork.cpp.

void pre_train ( int32_t  index,
CDenseFeatures< float64_t > *  features 
)
virtual

Pre-trains a single RBM

Parameters
indexIndex of the RBM
featuresInput features. Should have as many features as the total number of visible units in the DBN

Definition at line 159 of file DeepBeliefNetwork.cpp.

void print_modsel_params ( )
inherited

prints all parameter registered for model selection and their type

Definition at line 1111 of file SGObject.cpp.

void print_serializable ( const char *  prefix = "")
virtualinherited

prints registered parameters out

Parameters
prefixprefix for members

Definition at line 309 of file SGObject.cpp.

void reset_chain ( )
virtual

Resets the state of the markov chain used for sampling

Definition at line 352 of file DeepBeliefNetwork.cpp.

CDenseFeatures< float64_t > * sample ( int32_t  num_gibbs_steps = 1,
int32_t  batch_size = 1 
)
virtual

Draws samples from the marginal distribution of the visible units. The sampling starts from the values DBN's internal state and result of the sampling is stored there too.

Parameters
num_gibbs_stepsNumber of Gibbs sampling steps for the top RBM.
batch_sizeNumber of samples to be drawn. A seperate chain is used for each sample
Returns
Sampled states of the visible units

Definition at line 333 of file DeepBeliefNetwork.cpp.

bool save_serializable ( CSerializableFile file,
const char *  prefix = "",
int32_t  param_version = Version::get_version_parameter() 
)
virtualinherited

Save this object to file.

Parameters
filewhere to save the object; will be closed during returning if PREFIX is an empty string.
prefixprefix for members
param_version(optional) a parameter version different to (this is mainly for testing, better do not use)
Returns
TRUE if done, otherwise FALSE

Definition at line 315 of file SGObject.cpp.

void save_serializable_post ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to post-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::SAVE_SERIALIZABLE_POST is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel.

Definition at line 1072 of file SGObject.cpp.

void save_serializable_pre ( ) throw (ShogunException)
protectedvirtualinherited

Can (optionally) be overridden to pre-initialize some member variables which are not PARAMETER::ADD'ed. Make sure that at first the overridden method BASE_CLASS::SAVE_SERIALIZABLE_PRE is called.

Exceptions
ShogunExceptionwill be thrown if an error occurs.

Reimplemented in CKernel, CDynamicArray< T >, CDynamicArray< float64_t >, CDynamicArray< float32_t >, CDynamicArray< int32_t >, CDynamicArray< char >, CDynamicArray< bool >, and CDynamicObjectArray.

Definition at line 1067 of file SGObject.cpp.

void set_batch_size ( int32_t  batch_size)
virtual

Sets the number of train/test cases the RBM will deal with

Parameters
batch_sizeBatch size

Definition at line 135 of file DeepBeliefNetwork.cpp.

void set_generic< complex128_t > ( )
inherited

set generic type to T

Definition at line 42 of file SGObject.cpp.

void set_global_io ( SGIO io)
inherited

set the io object

Parameters
ioio object to use

Definition at line 230 of file SGObject.cpp.

void set_global_parallel ( Parallel parallel)
inherited

set the parallel object

Parameters
parallelparallel object to use

Definition at line 243 of file SGObject.cpp.

void set_global_version ( Version version)
inherited

set the version object

Parameters
versionversion object to use

Definition at line 284 of file SGObject.cpp.

CSGObject * shallow_copy ( ) const
virtualinherited

A shallow copy. All the SGObject instance variables will be simply assigned and SG_REF-ed.

Reimplemented in CGaussianKernel.

Definition at line 194 of file SGObject.cpp.

void train ( CDenseFeatures< float64_t > *  features)
virtual

Trains the DBN using the variant of the wake-sleep algorithm described in [A Fast Learning Algorithm for Deep Belief Nets, Hinton, 2006].

Parameters
featuresInput features. Should have as many features as the total number of visible units in the DBN

Definition at line 210 of file DeepBeliefNetwork.cpp.

CDenseFeatures< float64_t > * transform ( CDenseFeatures< float64_t > *  features,
int32_t  i = -1 
)
virtual

Applies the DBN as a features transformation

Forward-propagates the input features through the DBN and returns the Mean activations of the \( i^{th} \) hidden layer

Parameters
featuresInput features. Should have as many features as the number of visible units in the DBN
iIndex of the hidden layer. If -1, the activations of the last hidden layer is returned
Returns
Mean activations of the \( i^{th} \) hidden layer

Definition at line 316 of file DeepBeliefNetwork.cpp.

void unset_generic ( )
inherited

unset generic type

this has to be called in classes specializing a template class

Definition at line 304 of file SGObject.cpp.

void up_step ( int32_t  index,
SGVector< float64_t params,
SGMatrix< float64_t input,
SGMatrix< float64_t result,
bool  sample_states = true 
)
protectedvirtual

Computes the states of some layer using the states of the layer below it

Definition at line 443 of file DeepBeliefNetwork.cpp.

void update_parameter_hash ( )
virtualinherited

Updates the hash of current parameter combination

Definition at line 250 of file SGObject.cpp.

void wake_sleep ( SGMatrix< float64_t data,
CRBM top_rbm,
SGMatrixList< float64_t sleep_states,
SGMatrixList< float64_t wake_states,
SGMatrixList< float64_t psleep_states,
SGMatrixList< float64_t pwake_states,
SGVector< float64_t gen_params,
SGVector< float64_t rec_params,
SGVector< float64_t gen_gradients,
SGVector< float64_t rec_gradients 
)
protectedvirtual

Computes the gradients using the wake-sleep algorithm

Definition at line 473 of file DeepBeliefNetwork.cpp.

Member Data Documentation

int32_t cd_num_steps

Number of Gibbs sampling steps performed before each weight update during wake-sleep training. Default value is 1.

Definition at line 306 of file DeepBeliefNetwork.h.

float64_t gd_learning_rate

Gradient descent learning rate for wake-sleep training, defualt value 0.1

Definition at line 325 of file DeepBeliefNetwork.h.

float64_t gd_learning_rate_decay

Gradient descent learning rate decay for wake-sleep training. The learning rate is updated at each iteration i according to: alpha(i)=decay*alpha(i-1) Default value is 1.0 (no decay)

Definition at line 332 of file DeepBeliefNetwork.h.

int32_t gd_mini_batch_size

Size of the mini-batch used during gradient descent wake-sleep training, If 0 full-batch training is performed Default value is 0

Definition at line 322 of file DeepBeliefNetwork.h.

float64_t gd_momentum

gradient descent momentum multiplier for wake-sleep training

default value is 0.9

For more details on momentum, see this paper [Sutskever, 2013]

Definition at line 342 of file DeepBeliefNetwork.h.

SGIO* io
inherited

io

Definition at line 496 of file SGObject.h.

int32_t m_batch_size
protected

Number of train/test cases the network is currently dealing with

Definition at line 358 of file DeepBeliefNetwork.h.

SGVector<int32_t> m_bias_index_offsets
protected

Index at which the bias of each layer is stored in the parameters vector

Definition at line 367 of file DeepBeliefNetwork.h.

Parameter* m_gradient_parameters
inherited

parameters wrt which we can compute gradients

Definition at line 511 of file SGObject.h.

uint32_t m_hash
inherited

Hash of parameter values

Definition at line 517 of file SGObject.h.

CDynamicArray<int32_t>* m_layer_sizes
protected

Size of each layer

Definition at line 352 of file DeepBeliefNetwork.h.

Parameter* m_model_selection_parameters
inherited

model selection parameters

Definition at line 508 of file SGObject.h.

int32_t m_num_layers
protected

Number of layers

Definition at line 349 of file DeepBeliefNetwork.h.

int32_t m_num_params
protected

Number of parameters

Definition at line 364 of file DeepBeliefNetwork.h.

ParameterMap* m_parameter_map
inherited

map for different parameter versions

Definition at line 514 of file SGObject.h.

Parameter* m_parameters
inherited

parameters

Definition at line 505 of file SGObject.h.

SGVector<float64_t> m_params
protected

Parameters of the network

Definition at line 361 of file DeepBeliefNetwork.h.

float64_t m_sigma
protected

Standard deviation of the gaussian used to initialize the parameters

Definition at line 376 of file DeepBeliefNetwork.h.

SGMatrixList<float64_t> m_states
protected

States of each layer

Definition at line 355 of file DeepBeliefNetwork.h.

ERBMVisibleUnitType m_visible_units_type
protected

Type of the visible units

Definition at line 346 of file DeepBeliefNetwork.h.

SGVector<int32_t> m_weights_index_offsets
protected

Index at which the weights of each hidden layer is stored in the parameters vector

Definition at line 372 of file DeepBeliefNetwork.h.

int32_t max_num_epochs

Maximum number of iterations over the training set during wake-sleep training. Defualt value is 1

Definition at line 316 of file DeepBeliefNetwork.h.

int32_t monitoring_interval

Number of weight updates between each evaluation of the reconstruction error during wake-sleep training. Default value is 10.

Definition at line 311 of file DeepBeliefNetwork.h.

Parallel* parallel
inherited

parallel

Definition at line 499 of file SGObject.h.

SGVector<int32_t> pt_cd_num_steps

CRBM::cd_num_steps for pre-training each RBM. Default value is 1 for all RBMs

Definition at line 246 of file DeepBeliefNetwork.h.

SGVector<bool> pt_cd_persistent

CRBM::cd_persistent for pre-training each RBM. Default value is true for all RBMs

Definition at line 251 of file DeepBeliefNetwork.h.

SGVector<bool> pt_cd_sample_visible

CRBM::cd_sample_visible for pre-training each RBM. Default value is false for all RBMs

Definition at line 256 of file DeepBeliefNetwork.h.

SGVector<float64_t> pt_gd_learning_rate

CRBM::gd_learning_rate for pre-training each RBM. Default value is 0.1 for all RBMs

Definition at line 291 of file DeepBeliefNetwork.h.

SGVector<float64_t> pt_gd_learning_rate_decay

CRBM::gd_learning_rate_decay for pre-training each RBM. Default value is 1.0 for all RBMs

Definition at line 296 of file DeepBeliefNetwork.h.

SGVector<int32_t> pt_gd_mini_batch_size

CRBM::gd_mini_batch_size for pre-training each RBM. Default value is 0 for all RBMs

Definition at line 286 of file DeepBeliefNetwork.h.

SGVector<float64_t> pt_gd_momentum

CRBM::gd_momentum for pre-training each RBM. Default value is 0.9 for all RBMs

Definition at line 301 of file DeepBeliefNetwork.h.

SGVector<float64_t> pt_l1_coefficient

CRBM::l1_coefficient for pre-training each RBM. Default value is 0.0 for all RBMs

Definition at line 266 of file DeepBeliefNetwork.h.

SGVector<float64_t> pt_l2_coefficient

CRBM::l2_coefficient for pre-training each RBM. Default value is 0.0 for all RBMs

Definition at line 261 of file DeepBeliefNetwork.h.

SGVector<int32_t> pt_max_num_epochs

CRBM::max_num_epochs for pre-training each RBM. Default value is 1 for all RBMs

Definition at line 281 of file DeepBeliefNetwork.h.

SGVector<int32_t> pt_monitoring_interval

CRBM::monitoring_interval for pre-training each RBM. Default value is 10 for all RBMs

Definition at line 271 of file DeepBeliefNetwork.h.

SGVector<int32_t> pt_monitoring_method

CRBM::monitoring_method for pre-training each RBM. Default value is RBMMM_RECONSTRUCTION_ERROR for all RBMs

Definition at line 276 of file DeepBeliefNetwork.h.

Version* version
inherited

version

Definition at line 502 of file SGObject.h.


The documentation for this class was generated from the following files:

SHOGUN Machine Learning Toolbox - Documentation