Detailed Description

template<class ST>
class shogun::CSimpleFeatures< ST >

The class SimpleFeatures implements dense feature matrices.

The feature matrices are stored en-block in memory in fortran order, i.e. column-by-column, where a column denotes a feature vector.

There are get_num_vectors() many feature vectors, of dimension get_num_features(). To access a feature vector call get_feature_vector() and when you are done treating it call free_feature_vector(). While free_feature_vector() is a NOP in most cases feature vectors might have been generated on the fly (due to a number preprocessors being attached to them).

From this template class a number the following dense feature matrix types are used and supported:

bool matrix - CSimpleFeatures<bool>
8bit char matrix - CSimpleFeatures<char>
8bit Byte matrix - CSimpleFeatures<uint8_t>
16bit Integer matrix - CSimpleFeatures<int16_t>
16bit Word matrix - CSimpleFeatures<uint16_t>
32bit Integer matrix - CSimpleFeatures<int32_t>
32bit Unsigned Integer matrix - CSimpleFeatures<uint32_t>
32bit Float matrix - CSimpleFeatures<float32_t>
64bit Float matrix - CSimpleFeatures<float64_t>
64bit Float matrix in a file - CRealFileFeatures
64bit Tangent of posterior log-odds (TOP) features from HMM - CTOPFeatures
64bit Fisher Kernel (FK) features from HMM - CTOPFeatures
96bit Float matrix - CSimpleFeatures<floatmax_t>

Definition at line 65 of file SimpleFeatures.h.

Inheritance diagram for CSimpleFeatures< ST >:

[legend]

List of all members.

Public Member Functions
	CSimpleFeatures (int32_t size=0)
	CSimpleFeatures (const CSimpleFeatures &orig)
	CSimpleFeatures (SGMatrix< ST > matrix)
	CSimpleFeatures (ST *src, int32_t num_feat, int32_t num_vec)
	CSimpleFeatures (CFile *loader)
virtual CFeatures *	duplicate () const
virtual	~CSimpleFeatures ()
void	free_feature_matrix ()
void	free_features ()
ST *	get_feature_vector (int32_t num, int32_t &len, bool &dofree)
void	set_feature_vector (SGVector< ST > vector, int32_t num)
SGVector< ST >	get_feature_vector (int32_t num)
void	free_feature_vector (ST *feat_vec, int32_t num, bool dofree)
void	free_feature_vector (SGVector< ST > vec, int32_t num)
void	vector_subset (int32_t *idx, int32_t idx_len)
void	feature_subset (int32_t *idx, int32_t idx_len)
void	get_feature_matrix (ST *dst, int32_t num_feat, int32_t *num_vec)
SGMatrix< ST >	get_feature_matrix ()
void	set_feature_matrix (SGMatrix< ST > matrix)
ST *	get_feature_matrix (int32_t &num_feat, int32_t &num_vec)
CSimpleFeatures< ST > *	get_transposed ()
ST *	get_transposed (int32_t &num_feat, int32_t &num_vec)
virtual void	set_feature_matrix (ST *fm, int32_t num_feat, int32_t num_vec)
virtual void	copy_feature_matrix (SGMatrix< ST > src)
void	obtain_from_dot (CDotFeatures *df)
virtual bool	apply_preprocessor (bool force_preprocessing=false)
virtual int32_t	get_size ()
virtual int32_t	get_num_vectors () const
int32_t	get_num_features ()
void	set_num_features (int32_t num)
void	set_num_vectors (int32_t num)
void	initialize_cache ()
virtual EFeatureClass	get_feature_class ()
virtual EFeatureType	get_feature_type ()
virtual bool	reshape (int32_t p_num_features, int32_t p_num_vectors)
virtual int32_t	get_dim_feature_space () const
virtual float64_t	dot (int32_t vec_idx1, CDotFeatures *df, int32_t vec_idx2)
virtual float64_t	dense_dot (int32_t vec_idx1, const float64_t *vec2, int32_t vec2_len)
virtual void	add_to_dense_vec (float64_t alpha, int32_t vec_idx1, float64_t *vec2, int32_t vec2_len, bool abs_val=false)
virtual int32_t	get_nnz_features_for_vector (int32_t num)
virtual bool	Align_char_features (CStringFeatures< char > cf, CStringFeatures< char > Ref, float64_t gapCost)
virtual void	load (CFile *loader)
virtual void	save (CFile *saver)
virtual void *	get_feature_iterator (int32_t vector_index)
virtual bool	get_next_feature (int32_t &index, float64_t &value, void *iterator)
virtual void	free_feature_iterator (void *iterator)
virtual CFeatures *	copy_subset (SGVector< index_t > indices)
virtual const char *	get_name () const
Protected Member Functions
virtual ST *	compute_feature_vector (int32_t num, int32_t &len, ST *target=NULL)
Protected Attributes
int32_t	num_vectors
	number of vectors in cache
int32_t	num_features
	number of features in cache
ST *	feature_matrix
int32_t	feature_matrix_num_vectors
int32_t	feature_matrix_num_features
CCache< ST > *	feature_cache

Constructor & Destructor Documentation

CSimpleFeatures ( int32_t size = 0 )

constructor

Parameters:

size

cache size

Definition at line 72 of file SimpleFeatures.h.

CSimpleFeatures ( const CSimpleFeatures< ST > & orig )

copy constructor

Definition at line 75 of file SimpleFeatures.h.

CSimpleFeatures ( SGMatrix< ST > matrix )

constructor

Parameters:

matrix

feature matrix

Definition at line 89 of file SimpleFeatures.h.

CSimpleFeatures	(	ST *	src,
		int32_t	num_feat,
		int32_t	num_vec
	)

constructor

Parameters:

	src	feature matrix
	num_feat	number of features in matrix
	num_vec	number of vectors in matrix

Definition at line 102 of file SimpleFeatures.h.

CSimpleFeatures ( CFile * loader )

constructor loading features from file

Parameters:

loader

File object via which to load data

Definition at line 113 of file SimpleFeatures.h.

virtual ~CSimpleFeatures ( ) [virtual]

Definition at line 129 of file SimpleFeatures.h.

Member Function Documentation

virtual void add_to_dense_vec	(	float64_t	alpha,
		int32_t	vec_idx1,
		float64_t *	vec2,
		int32_t	vec2_len,
		bool	abs_val = `false`
	)			`[virtual]`

add vector 1 multiplied with alpha to dense vector2

possible with subset

Parameters:

	alpha	scalar alpha
	vec_idx1	index of first vector
	vec2	pointer to real valued vector
	vec2_len	length of real valued vector
	abs_val	if true add the absolute value

Implements CDotFeatures.

Definition at line 824 of file SimpleFeatures.h.

virtual bool Align_char_features	(	CStringFeatures< char > *	cf,
		CStringFeatures< char > *	Ref,
		float64_t	gapCost
	)			`[virtual]`

align char features

Parameters:

	cf	char features
	Ref	other char features
	gapCost	gap cost

Returns:: if aligning was successful

Definition at line 867 of file SimpleFeatures.h.

virtual bool apply_preprocessor ( bool force_preprocessing = false ) [virtual]

apply preprocessor

applies preprocessors to ALL features (subset removed before and restored afterwards)

not possible with subset

Parameters:

force_preprocessing

if preprocssing shall be forced

Returns:: if applying was successful

Definition at line 623 of file SimpleFeatures.h.

virtual ST* compute_feature_vector	(	int32_t	num,
		int32_t &	len,
		ST *	target = `NULL`
	)			`[protected, virtual]`

compute feature vector for sample num if target is set the vector is written to target len is returned by reference

NOT IMPLEMENTED!

Parameters:

	num	num
	len	len
	target

Returns:: feature vector

Reimplemented in CFKFeatures, CRealFileFeatures, and CTOPFeatures.

Definition at line 1008 of file SimpleFeatures.h.

virtual void copy_feature_matrix ( SGMatrix< ST > src ) [virtual]

copy feature matrix store copy of feature_matrix, where num_features is the column offset, and columns are linear in memory see below for definition of feature_matrix

not possible with subset

Parameters:

src

feature matrix to copy

Definition at line 559 of file SimpleFeatures.h.

virtual CFeatures* copy_subset ( SGVector< index_t > indices ) [virtual]

Creates a new CFeatures instance containing copies of the elements which are specified by the provided indices.

possible with subset

Parameters:

indices

indices of feature elements to copy

Returns:: new CFeatures instance with copies of feature data

Reimplemented from CFeatures.

Definition at line 978 of file SimpleFeatures.h.

virtual float64_t dense_dot	(	int32_t	vec_idx1,
		const float64_t *	vec2,
		int32_t	vec2_len
	)			`[virtual]`

compute dot product between vector1 and a dense vector

possible with subset TODO: where?

Parameters:

	vec_idx1	index of first vector
	vec2	pointer to real valued vector
	vec2_len	length of real valued vector

Implements CDotFeatures.

virtual float64_t dot	(	int32_t	vec_idx1,
		CDotFeatures *	df,
		int32_t	vec_idx2
	)			`[virtual]`

compute dot product between vector1 and vector2, appointed by their indices

possible with subset

Parameters:

	vec_idx1	index of first vector
	df	DotFeatures (of same kind) to compute dot product with
	vec_idx2	index of second vector

Implements CDotFeatures.

Definition at line 781 of file SimpleFeatures.h.

virtual CFeatures* duplicate ( ) const [virtual]

duplicate feature object

Returns:: feature object

Implements CFeatures.

Definition at line 124 of file SimpleFeatures.h.

void feature_subset	(	int32_t *	idx,
		int32_t	idx_len
	)

Extracts the features mentioned in idx and replaces them in feature matrix in place.

It does not resize the allocated memory block.

Not possible with subset.

Parameters:

	idx	index with features that shall remain in the feature matrix
	idx_len	length of the index

Note: assumes idx is sorted

Definition at line 365 of file SimpleFeatures.h.

virtual void free_feature_iterator ( void * iterator ) [virtual]

clean up iterator call this function with the iterator returned by get_first_feature

Parameters:

iterator

as returned by get_first_feature

Implements CDotFeatures.

Definition at line 960 of file SimpleFeatures.h.

void free_feature_matrix ( )

free feature matrix

Any subset is removed

Definition at line 135 of file SimpleFeatures.h.

void free_feature_vector	(	SGVector< ST >	vec,
		int32_t	num
	)

free feature vector

possible with subset

Parameters:

	vec	feature vector to free
	num	index in feature cache

Definition at line 303 of file SimpleFeatures.h.

void free_feature_vector	(	ST *	feat_vec,
		int32_t	num,
		bool	dofree
	)

free feature vector

possible with subset

Parameters:

	feat_vec	feature vector to free
	num	index in feature cache
	dofree	if vector should be really deleted

Definition at line 287 of file SimpleFeatures.h.

void free_features ( )

free feature matrix and cache

Any subset is removed

Definition at line 150 of file SimpleFeatures.h.

virtual int32_t get_dim_feature_space ( ) const [virtual]

obtain the dimensionality of the feature space

(not mix this up with the dimensionality of the input space, usually obtained via get_num_features())

Returns:: dimensionality

Implements CDotFeatures.

Definition at line 770 of file SimpleFeatures.h.

virtual EFeatureClass get_feature_class ( ) [virtual]

get feature class

Returns:: feature class SIMPLE

Implements CFeatures.

Definition at line 732 of file SimpleFeatures.h.

virtual void* get_feature_iterator ( int32_t vector_index ) [virtual]

iterate over the non-zero features

call get_feature_iterator first, followed by get_next_feature and free_feature_iterator to cleanup

possible with subset

Parameters:

vector_index

the index of the vector over whose components to iterate over

Returns:: feature iterator (to be passed to get_next_feature)

Implements CDotFeatures.

Definition at line 914 of file SimpleFeatures.h.

void get_feature_matrix	(	ST **	dst,
		int32_t *	num_feat,
		int32_t *	num_vec
	)

get a copy of the feature matrix num_feat,num_vectors are returned by reference

possible with subset

Parameters:

	dst	destination to store matrix in
	num_feat	number of features (rows of matrix)
	num_vec	number of vectors (columns of matrix)

Definition at line 404 of file SimpleFeatures.h.

SGMatrix<ST> get_feature_matrix ( )

Getter for feature matrix

subset is ignored

Returns:: matrix feature matrix

Definition at line 437 of file SimpleFeatures.h.

ST* get_feature_matrix	(	int32_t &	num_feat,
		int32_t &	num_vec
	)

get the pointer to the feature matrix num_feat,num_vectors are returned by reference

subset is ignored

Parameters:

	num_feat	number of features in matrix
	num_vec	number of vectors in matrix

Returns:: feature matrix

Definition at line 468 of file SimpleFeatures.h.

virtual EFeatureType get_feature_type ( ) [virtual]

get feature type

Returns:: templated feature type

Implements CFeatures.

ST* get_feature_vector	(	int32_t	num,
		int32_t &	len,
		bool &	dofree
	)

get feature vector for sample num from the matrix as it is if matrix is initialized, else return preprocessed compute_feature_vector (not implemented)

Parameters:

	num	index of feature vector
	len	length is returned by reference
	dofree	whether returned vector must be freed by caller via free_feature_vector

Returns:: feature vector

Definition at line 168 of file SimpleFeatures.h.

SGVector<ST> get_feature_vector ( int32_t num )

get feature vector num

possible with subset

Parameters:

num

index of vector

Returns:: feature vector

Definition at line 263 of file SimpleFeatures.h.

virtual const char* get_name ( void ) const [virtual]

Returns:: object name

Implements CSGObject.

Reimplemented in CFKFeatures, CRealFileFeatures, and CTOPFeatures.

Definition at line 994 of file SimpleFeatures.h.

virtual bool get_next_feature	(	int32_t &	index,
		float64_t &	value,
		void *	iterator
	)			`[virtual]`

iterate over the non-zero features

call this function with the iterator returned by get_first_feature and call free_feature_iterator to cleanup

possible with subset

Parameters:

	index	is returned by reference (-1 when not available)
	value	is returned by reference
	iterator	as returned by get_first_feature

Returns:: true if a new non-zero feature got returned

Implements CDotFeatures.

Definition at line 942 of file SimpleFeatures.h.

virtual int32_t get_nnz_features_for_vector ( int32_t num ) [virtual]

get number of non-zero features in vector

Parameters:

num

which vector

Returns:: number of non-zero features in vector

Implements CDotFeatures.

Definition at line 854 of file SimpleFeatures.h.

int32_t get_num_features ( )

get number of features (of possible subset)

Returns:: number of features

Definition at line 683 of file SimpleFeatures.h.

virtual int32_t get_num_vectors ( ) const [virtual]

get number of feature vectors

Returns:: number of feature vectors

Implements CFeatures.

Definition at line 674 of file SimpleFeatures.h.

virtual int32_t get_size ( ) [virtual]

get memory footprint of one feature

Returns:: memory footprint of one feature

Implements CFeatures.

Definition at line 668 of file SimpleFeatures.h.

ST* get_transposed	(	int32_t &	num_feat,
		int32_t &	num_vec
	)

compute and return the transpose of the feature matrix which will be prepocessed. num_feat, num_vectors are returned by reference caller has to clean up

possible with subset

Parameters:

	num_feat	number of features in matrix
	num_vec	number of vectors in matrix

Returns:: transposed sparse feature matrix

Definition at line 501 of file SimpleFeatures.h.

CSimpleFeatures<ST>* get_transposed ( )

get a transposed copy of the features

possible with subset

Returns:: transposed copy

Definition at line 481 of file SimpleFeatures.h.

void initialize_cache ( )

Initialize cache

not possible with subset

Definition at line 714 of file SimpleFeatures.h.

virtual void load ( CFile * loader ) [virtual]

load features from file

Parameters:

loader

File object via which to load data

Reimplemented from CFeatures.

void obtain_from_dot ( CDotFeatures * df )

obtain simple features from other dotfeatures

removes any subset before

Parameters:

df

dotfeatures to obtain features from

Definition at line 585 of file SimpleFeatures.h.

virtual bool reshape	(	int32_t	p_num_features,
		int32_t	p_num_vectors
	)			`[virtual]`

reshape

not possible with subset

Parameters:

	p_num_features	new number of features
	p_num_vectors	new number of vectors

Returns:: if reshaping was successful

Reimplemented from CFeatures.

Definition at line 748 of file SimpleFeatures.h.

virtual void save ( CFile * saver ) [virtual]

save features to file

Parameters:

saver

File object via which to save data

Reimplemented from CFeatures.

void set_feature_matrix ( SGMatrix< ST > matrix )

Setter for feature matrix

any subset is removed

Parameters:

matrix

feature matrix to set

Definition at line 448 of file SimpleFeatures.h.

virtual void set_feature_matrix	(	ST *	fm,
		int32_t	num_feat,
		int32_t	num_vec
	)			`[virtual]`

set feature matrix necessary to set feature_matrix, num_features, num_vectors, where num_features is the column offset, and columns are linear in memory see below for definition of feature_matrix

not possible with subset

Parameters:

	fm	feature matrix to se
	num_feat	number of features in matrix
	num_vec	number of vectors in matrix

Definition at line 535 of file SimpleFeatures.h.

void set_feature_vector	(	SGVector< ST >	vector,
		int32_t	num
	)

set feature vector num

possible with subset

Parameters:

	vector	vector
	num	index if vector to set

Definition at line 234 of file SimpleFeatures.h.

void set_num_features ( int32_t num )

set number of features

Parameters:

num

number to set

Definition at line 689 of file SimpleFeatures.h.

void set_num_vectors ( int32_t num )

set number of vectors

not possible with subset

Parameters:

num

number to set

Definition at line 701 of file SimpleFeatures.h.

void vector_subset	(	int32_t *	idx,
		int32_t	idx_len
	)

Extracts the feature vectors mentioned in idx and replaces them in feature matrix in place.

It does not resize the allocated memory block.

not possible with subset

Parameters:

	idx	index with examples that shall remain in the feature matrix
	idx_len	length of the index

Note: assumes idx is sorted

Definition at line 321 of file SimpleFeatures.h.

Member Data Documentation

CCache<ST>* feature_cache [protected]

feature cache

Definition at line 1058 of file SimpleFeatures.h.

ST* feature_matrix [protected]

Feature matrix and its associated number of vectors and features. Note that num_vectors / num_features above have the same sizes if feature_matrix != NULL

Definition at line 1049 of file SimpleFeatures.h.

int32_t feature_matrix_num_features [protected]

number of features in feature matrix

Definition at line 1055 of file SimpleFeatures.h.

int32_t feature_matrix_num_vectors [protected]

number of vectors in feature matrix

Definition at line 1052 of file SimpleFeatures.h.

int32_t num_features [protected]

number of features in cache

Definition at line 1043 of file SimpleFeatures.h.

int32_t num_vectors [protected]

number of vectors in cache

Definition at line 1040 of file SimpleFeatures.h.

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

SimpleFeatures.h

CSimpleFeatures< ST > Class Template Reference

Detailed Description

template<class ST> class shogun::CSimpleFeatures< ST >

Public Member Functions

Protected Member Functions

Protected Attributes

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation

template<class ST>
class shogun::CSimpleFeatures< ST >