Template class SparseFeatures implements sparse matrices.
Features are an array of TSparse, sorted w.r.t. vec_index (increasing) and withing same vec_index w.r.t. feat_index (increasing);
Sparse feature vectors can be accessed via get_sparse_feature_vector() and should be freed (this operation is a NOP in most cases) via free_sparse_feature_vector().
As this is a template class it can directly be used for different data types like sparse matrices of real valued, integer, byte etc type.
Definition at line 67 of file SparseFeatures.h.
Public Member Functions | |
CSparseFeatures (int32_t size=0) | |
CSparseFeatures (TSparse< ST > *src, int32_t num_feat, int32_t num_vec) | |
CSparseFeatures (ST *src, int32_t num_feat, int32_t num_vec) | |
CSparseFeatures (const CSparseFeatures &orig) | |
CSparseFeatures (char *fname) | |
virtual | ~CSparseFeatures () |
void | free_sparse_feature_matrix () |
void | free_sparse_features () |
virtual CFeatures * | duplicate () const |
ST * | get_full_feature_vector (int32_t num, int32_t &len) |
void | get_full_feature_vector (ST **dst, int32_t *len, int32_t num) |
virtual int32_t | get_nnz_features_for_vector (int32_t num) |
TSparseEntry< ST > * | get_sparse_feature_vector (int32_t num, int32_t &len, bool &vfree) |
ST | sparse_dot (ST alpha, TSparseEntry< ST > *avec, int32_t alen, TSparseEntry< ST > *bvec, int32_t blen) |
ST | dense_dot (ST alpha, int32_t num, ST *vec, int32_t dim, ST b) |
void | add_to_dense_vec (float64_t alpha, int32_t num, float64_t *vec, int32_t dim, bool abs_val=false) |
void | free_sparse_feature_vector (TSparseEntry< ST > *feat_vec, int32_t num, bool free) |
TSparse< ST > * | get_sparse_feature_matrix (int32_t &num_feat, int32_t &num_vec) |
void | get_sparse_feature_matrix (TSparse< ST > **dst, int32_t *num_feat, int32_t *num_vec, int64_t *nnz) |
void | clean_tsparse (TSparse< ST > *sfm, int32_t num_vec) |
TSparse< ST > * | get_transposed (int32_t &num_feat, int32_t &num_vec) |
virtual void | set_sparse_feature_matrix (TSparse< ST > *src, int32_t num_feat, int32_t num_vec) |
ST * | get_full_feature_matrix (int32_t &num_feat, int32_t &num_vec) |
void | get_full_feature_matrix (ST **dst, int32_t *num_feat, int32_t *num_vec) |
virtual bool | set_full_feature_matrix (ST *src, int32_t num_feat, int32_t num_vec) |
virtual bool | apply_preproc (bool force_preprocessing=false) |
virtual int32_t | get_size () |
bool | obtain_from_simple (CSimpleFeatures< ST > *sf) |
virtual int32_t | get_num_vectors () |
int32_t | get_num_features () |
int32_t | set_num_features (int32_t num) |
virtual EFeatureClass | get_feature_class () |
virtual EFeatureType | get_feature_type () |
void | free_feature_vector (TSparseEntry< ST > *feat_vec, int32_t num, bool free) |
int64_t | get_num_nonzero_entries () |
float64_t * | compute_squared (float64_t *sq) |
float64_t | compute_squared_norm (CSparseFeatures< float64_t > *lhs, float64_t *sq_lhs, int32_t idx_a, CSparseFeatures< float64_t > *rhs, float64_t *sq_rhs, int32_t idx_b) |
CLabels * | load_svmlight_file (char *fname) |
bool | write_svmlight_file (char *fname, CLabels *label) |
virtual int32_t | get_dim_feature_space () |
virtual float64_t | dot (int32_t vec_idx1, int32_t vec_idx2) |
virtual float64_t | dense_dot (int32_t vec_idx1, const float64_t *vec2, int32_t vec2_len) |
virtual const char * | get_name () const |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
template<> | |
EFeatureType | get_feature_type () |
Protected Member Functions | |
virtual TSparseEntry< ST > * | compute_sparse_feature_vector (int32_t num, int32_t &len, TSparseEntry< ST > *target=NULL) |
Protected Attributes | |
int32_t | num_vectors |
total number of vectors | |
int32_t | num_features |
total number of features | |
TSparse< ST > * | sparse_feature_matrix |
array of sparse vectors of size num_vectors | |
CCache< TSparseEntry< ST > > * | feature_cache |
CSparseFeatures< ST >::CSparseFeatures | ( | int32_t | size = 0 |
) |
CSparseFeatures< ST >::CSparseFeatures | ( | TSparse< ST > * | src, | |
int32_t | num_feat, | |||
int32_t | num_vec | |||
) |
convenience constructor that creates sparse features from the ones passed as argument
src | dense feature matrix | |
num_feat | number of features | |
num_vec | number of vectors |
Definition at line 86 of file SparseFeatures.h.
CSparseFeatures< ST >::CSparseFeatures | ( | ST * | src, | |
int32_t | num_feat, | |||
int32_t | num_vec | |||
) |
convenience constructor that creates sparse features from dense features
src | dense feature matrix | |
num_feat | number of features | |
num_vec | number of vectors |
Definition at line 100 of file SparseFeatures.h.
CSparseFeatures< ST >::CSparseFeatures | ( | const CSparseFeatures< ST > & | orig | ) |
copy constructor
Definition at line 108 of file SparseFeatures.h.
CSparseFeatures< ST >::CSparseFeatures | ( | char * | fname | ) |
constructor
fname | filename to load features from |
Definition at line 132 of file SparseFeatures.h.
virtual CSparseFeatures< ST >::~CSparseFeatures | ( | ) | [virtual] |
Definition at line 137 of file SparseFeatures.h.
void CSparseFeatures< ST >::add_to_dense_vec | ( | float64_t | alpha, | |
int32_t | num, | |||
float64_t * | vec, | |||
int32_t | dim, | |||
bool | abs_val = false | |||
) | [virtual] |
add a sparse feature vector onto a dense one dense+=alpha*sparse
alpha | scalar to multiply with | |
num | index of feature vector | |
vec | dense vector | |
dim | length of the dense vector | |
abs_val | if true, do dense+=alpha*abs(sparse) |
Implements CDotFeatures.
Definition at line 418 of file SparseFeatures.h.
virtual bool CSparseFeatures< ST >::apply_preproc | ( | bool | force_preprocessing = false |
) | [virtual] |
apply preprocessor
force_preprocessing | if preprocssing shall be forced |
Definition at line 754 of file SparseFeatures.h.
void CSparseFeatures< ST >::clean_tsparse | ( | TSparse< ST > * | sfm, | |
int32_t | num_vec | |||
) |
clean TSparse
sfm | sparse feature matrix | |
num_vec | number of vectors in matrix |
Definition at line 496 of file SparseFeatures.h.
virtual TSparseEntry<ST>* CSparseFeatures< ST >::compute_sparse_feature_vector | ( | int32_t | num, | |
int32_t & | len, | |||
TSparseEntry< 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!
num | num | |
len | len | |
target | target |
Definition at line 1246 of file SparseFeatures.h.
float64_t* CSparseFeatures< ST >::compute_squared | ( | float64_t * | sq | ) |
compute a^2 on all feature vectors
sq | the square for each vector is stored in here |
Definition at line 877 of file SparseFeatures.h.
float64_t CSparseFeatures< ST >::compute_squared_norm | ( | CSparseFeatures< float64_t > * | lhs, | |
float64_t * | sq_lhs, | |||
int32_t | idx_a, | |||
CSparseFeatures< float64_t > * | rhs, | |||
float64_t * | sq_rhs, | |||
int32_t | idx_b | |||
) |
compute (a-b)^2 (== a^2+b^2+2ab) usually called by kernels'/distances' compute functions works on two feature vectors, although it is a member of a single feature: can either be called by lhs or rhs.
lhs | left-hand side features | |
sq_lhs | squared values of left-hand side | |
idx_a | index of left-hand side's vector to compute | |
rhs | right-hand side features | |
sq_rhs | squared values of right-hand side | |
idx_b | index of right-hand side's vector to compute |
Definition at line 910 of file SparseFeatures.h.
virtual float64_t CSparseFeatures< ST >::dense_dot | ( | int32_t | vec_idx1, | |
const float64_t * | vec2, | |||
int32_t | vec2_len | |||
) | [virtual] |
compute dot product between vector1 and a dense vector
vec_idx1 | index of first vector | |
vec2 | pointer to real valued vector | |
vec2_len | length of real valued vector |
Implements CDotFeatures.
Definition at line 1211 of file SparseFeatures.h.
ST CSparseFeatures< ST >::dense_dot | ( | ST | alpha, | |
int32_t | num, | |||
ST * | vec, | |||
int32_t | dim, | |||
ST | b | |||
) |
compute the dot product between dense weights and a sparse feature vector alpha * sparse^T * w + b
alpha | scalar to multiply with | |
num | index of feature vector | |
vec | dense vector to compute dot product with | |
dim | length of the dense vector | |
b | bias |
Definition at line 389 of file SparseFeatures.h.
virtual float64_t CSparseFeatures< ST >::dot | ( | int32_t | vec_idx1, | |
int32_t | vec_idx2 | |||
) | [virtual] |
compute dot product between vector1 and vector2, appointed by their indices
vec_idx1 | index of first vector | |
vec_idx2 | index of second vector |
Implements CDotFeatures.
Definition at line 1190 of file SparseFeatures.h.
virtual CFeatures* CSparseFeatures< ST >::duplicate | ( | ) | const [virtual] |
duplicate feature object
Implements CFeatures.
Definition at line 167 of file SparseFeatures.h.
void CSparseFeatures< ST >::free_feature_vector | ( | TSparseEntry< ST > * | feat_vec, | |
int32_t | num, | |||
bool | free | |||
) |
free feature vector
feat_vec | feature vector to free | |
num | index of vector in cache | |
free | if vector really should be deleted |
Definition at line 850 of file SparseFeatures.h.
void CSparseFeatures< ST >::free_sparse_feature_matrix | ( | ) |
free sparse feature matrix
Definition at line 145 of file SparseFeatures.h.
void CSparseFeatures< ST >::free_sparse_feature_vector | ( | TSparseEntry< ST > * | feat_vec, | |
int32_t | num, | |||
bool | free | |||
) |
free sparse feature vector
feat_vec | feature vector to free | |
num | index of this vector in the cache | |
free | if vector should be really deleted |
Definition at line 450 of file SparseFeatures.h.
void CSparseFeatures< ST >::free_sparse_features | ( | ) |
free sparse feature matrix and cache
Definition at line 156 of file SparseFeatures.h.
virtual int32_t CSparseFeatures< ST >::get_dim_feature_space | ( | ) | [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())
Implements CDotFeatures.
Definition at line 1179 of file SparseFeatures.h.
virtual EFeatureClass CSparseFeatures< ST >::get_feature_class | ( | ) | [virtual] |
get feature class
Implements CFeatures.
Definition at line 836 of file SparseFeatures.h.
EFeatureType CSparseFeatures< floatmax_t >::get_feature_type | ( | ) | [virtual] |
get feature type the LONGREAL feature can deal with
Implements CFeatures.
Definition at line 1370 of file SparseFeatures.h.
EFeatureType CSparseFeatures< float64_t >::get_feature_type | ( | ) | [virtual] |
get feature type the DREAL feature can deal with
Implements CFeatures.
Definition at line 1361 of file SparseFeatures.h.
EFeatureType CSparseFeatures< float32_t >::get_feature_type | ( | ) | [virtual] |
get feature type the SHORTREAL feature can deal with
Implements CFeatures.
Definition at line 1352 of file SparseFeatures.h.
EFeatureType CSparseFeatures< uint64_t >::get_feature_type | ( | ) | [virtual] |
get feature type the ULONG feature can deal with
Implements CFeatures.
Definition at line 1343 of file SparseFeatures.h.
EFeatureType CSparseFeatures< int64_t >::get_feature_type | ( | ) | [virtual] |
get feature type the LONG feature can deal with
Implements CFeatures.
Definition at line 1334 of file SparseFeatures.h.
EFeatureType CSparseFeatures< uint32_t >::get_feature_type | ( | ) | [virtual] |
get feature type the UINT feature can deal with
Implements CFeatures.
Definition at line 1325 of file SparseFeatures.h.
EFeatureType CSparseFeatures< int32_t >::get_feature_type | ( | ) | [virtual] |
get feature type the INT feature can deal with
Implements CFeatures.
Definition at line 1316 of file SparseFeatures.h.
EFeatureType CSparseFeatures< uint16_t >::get_feature_type | ( | ) | [virtual] |
get feature type the WORD feature can deal with
Implements CFeatures.
Definition at line 1307 of file SparseFeatures.h.
EFeatureType CSparseFeatures< int16_t >::get_feature_type | ( | ) | [virtual] |
get feature type the SHORT feature can deal with
Implements CFeatures.
Definition at line 1298 of file SparseFeatures.h.
EFeatureType CSparseFeatures< uint8_t >::get_feature_type | ( | ) | [virtual] |
get feature type the BYTE feature can deal with
Implements CFeatures.
Definition at line 1289 of file SparseFeatures.h.
EFeatureType CSparseFeatures< char >::get_feature_type | ( | ) | [virtual] |
get feature type the CHAR feature can deal with
Implements CFeatures.
Definition at line 1280 of file SparseFeatures.h.
EFeatureType CSparseFeatures< bool >::get_feature_type | ( | ) | [virtual] |
get feature type the BOOL feature can deal with
Implements CFeatures.
Definition at line 1271 of file SparseFeatures.h.
virtual EFeatureType CSparseFeatures< ST >::get_feature_type | ( | ) | [virtual] |
void CSparseFeatures< ST >::get_full_feature_matrix | ( | ST ** | dst, | |
int32_t * | num_feat, | |||
int32_t * | num_vec | |||
) |
gets a copy of a full feature matrix (swig compatible) num_feat,num_vectors are returned by reference
dst | full feature matrix | |
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 630 of file SparseFeatures.h.
ST* CSparseFeatures< ST >::get_full_feature_matrix | ( | int32_t & | num_feat, | |
int32_t & | num_vec | |||
) |
gets a copy of a full feature matrix num_feat,num_vectors are returned by reference
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 595 of file SparseFeatures.h.
void CSparseFeatures< ST >::get_full_feature_vector | ( | ST ** | dst, | |
int32_t * | len, | |||
int32_t | num | |||
) |
get the fully expanded dense feature vector num
dst | feature vector | |
len | length is returned by reference | |
num | index of feature vector |
Definition at line 212 of file SparseFeatures.h.
ST* CSparseFeatures< ST >::get_full_feature_vector | ( | int32_t | num, | |
int32_t & | len | |||
) |
converts a sparse feature vector into a dense one preprocessed compute_feature_vector caller cleans up
num | index of feature vector | |
len | length is returned by reference |
Definition at line 180 of file SparseFeatures.h.
virtual const char* CSparseFeatures< ST >::get_name | ( | ) | const [virtual] |
virtual int32_t CSparseFeatures< ST >::get_nnz_features_for_vector | ( | int32_t | num | ) | [virtual] |
get number of non-zero features in vector
num | which vector |
Implements CDotFeatures.
Definition at line 244 of file SparseFeatures.h.
int32_t CSparseFeatures< ST >::get_num_features | ( | ) |
int64_t CSparseFeatures< ST >::get_num_nonzero_entries | ( | ) |
get number of non-zero entries in sparse feature matrix
Definition at line 863 of file SparseFeatures.h.
virtual int32_t CSparseFeatures< ST >::get_num_vectors | ( | ) | [virtual] |
get number of feature vectors
Implements CFeatures.
Definition at line 805 of file SparseFeatures.h.
virtual int32_t CSparseFeatures< ST >::get_size | ( | ) | [virtual] |
get memory footprint of one feature
Implements CFeatures.
Definition at line 784 of file SparseFeatures.h.
void CSparseFeatures< ST >::get_sparse_feature_matrix | ( | TSparse< ST > ** | dst, | |
int32_t * | num_feat, | |||
int32_t * | num_vec, | |||
int64_t * | nnz | |||
) |
get the pointer to the sparse feature matrix (swig compatible) num_feat,num_vectors are returned by reference
dst | feature matrix | |
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix | |
nnz | number of nonzero elements |
Definition at line 482 of file SparseFeatures.h.
TSparse<ST>* CSparseFeatures< ST >::get_sparse_feature_matrix | ( | int32_t & | num_feat, | |
int32_t & | num_vec | |||
) |
get the pointer to the sparse feature matrix num_feat,num_vectors are returned by reference
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 466 of file SparseFeatures.h.
TSparseEntry<ST>* CSparseFeatures< ST >::get_sparse_feature_vector | ( | int32_t | num, | |
int32_t & | len, | |||
bool & | vfree | |||
) |
get sparse feature vector for sample num from the matrix as it is if matrix is initialized, else return preprocessed compute_feature_vector
num | index of feature vector | |
len | number of sparse entries is returned by reference | |
vfree | whether returned vector must be freed by caller via free_sparse_feature_vector |
Definition at line 263 of file SparseFeatures.h.
TSparse<ST>* CSparseFeatures< ST >::get_transposed | ( | int32_t & | num_feat, | |
int32_t & | num_vec | |||
) |
compute and return the transpose of the sparse feature matrix which will be prepocessed. num_feat, num_vectors are returned by reference caller has to clean up
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 516 of file SparseFeatures.h.
CLabels* CSparseFeatures< ST >::load_svmlight_file | ( | char * | fname | ) |
load features from file
fname | filename to load from |
Definition at line 971 of file SparseFeatures.h.
bool CSparseFeatures< ST >::obtain_from_simple | ( | CSimpleFeatures< ST > * | sf | ) |
obtain sparse features from simple features
sf | simple features |
Definition at line 791 of file SparseFeatures.h.
virtual bool CSparseFeatures< ST >::set_full_feature_matrix | ( | ST * | src, | |
int32_t | num_feat, | |||
int32_t | num_vec | |||
) | [virtual] |
creates a sparse feature matrix from a full dense feature matrix necessary to set feature_matrix, num_features and num_vectors where num_features is the column offset, and columns are linear in memory see above for definition of sparse_feature_matrix
src | full feature matrix | |
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 665 of file SparseFeatures.h.
int32_t CSparseFeatures< ST >::set_num_features | ( | int32_t | num | ) |
set number of features
Sometimes when loading sparse features not all possible dimensions are used. This may pose a problem to classifiers when being applied to higher dimensional test-data. This function allows to artificially explode the feature space
num | the number of features, must be larger than the current number of features |
Definition at line 824 of file SparseFeatures.h.
virtual void CSparseFeatures< ST >::set_sparse_feature_matrix | ( | TSparse< ST > * | src, | |
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
src | new sparse feature matrix | |
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 579 of file SparseFeatures.h.
ST CSparseFeatures< ST >::sparse_dot | ( | ST | alpha, | |
TSparseEntry< ST > * | avec, | |||
int32_t | alen, | |||
TSparseEntry< ST > * | bvec, | |||
int32_t | blen | |||
) |
compute the dot product between two sparse feature vectors alpha * vec^T * vec
alpha | scalar to multiply with | |
avec | first sparse feature vector | |
alen | avec's length | |
bvec | second sparse feature vector | |
blen | bvec's length |
Definition at line 331 of file SparseFeatures.h.
bool CSparseFeatures< ST >::write_svmlight_file | ( | char * | fname, | |
CLabels * | label | |||
) |
write features to file using svm light format
fname | filename to write to | |
label | Label object (number of labels must correspond to number of features) |
Definition at line 1139 of file SparseFeatures.h.
CCache< TSparseEntry<ST> >* CSparseFeatures< ST >::feature_cache [protected] |
feature cache
Definition at line 1264 of file SparseFeatures.h.
int32_t CSparseFeatures< ST >::num_features [protected] |
total number of features
Definition at line 1258 of file SparseFeatures.h.
int32_t CSparseFeatures< ST >::num_vectors [protected] |
total number of vectors
Definition at line 1255 of file SparseFeatures.h.
TSparse<ST>* CSparseFeatures< ST >::sparse_feature_matrix [protected] |
array of sparse vectors of size num_vectors
Definition at line 1261 of file SparseFeatures.h.