Template class SparseFeatures implements sparse matrices.
Features are an array of SGSparseVector, 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.
(Partly) subset access is supported for this feature type. Simple use the (inherited) set_subset(), remove_subset() functions. If done, all calls that work with features are translated to the subset. See comments to find out whether it is supported for that method
Definition at line 52 of file SparseFeatures.h.
Public Member Functions | |
CSparseFeatures (int32_t size=0) | |
CSparseFeatures (SGSparseVector< ST > *src, int32_t num_feat, int32_t num_vec, bool copy=false) | |
CSparseFeatures (SGSparseMatrix< ST > sparse) | |
CSparseFeatures (SGMatrix< ST > dense) | |
CSparseFeatures (const CSparseFeatures &orig) | |
CSparseFeatures (CFile *loader) | |
virtual | ~CSparseFeatures () |
void | free_sparse_feature_matrix () |
void | free_sparse_features () |
virtual CFeatures * | duplicate () const |
ST | get_feature (int32_t num, int32_t index) |
ST * | get_full_feature_vector (int32_t num, int32_t &len) |
SGVector< ST > | get_full_feature_vector (int32_t num) |
virtual int32_t | get_nnz_features_for_vector (int32_t num) |
SGSparseVector< ST > | get_sparse_feature_vector (int32_t num) |
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 (SGSparseVector< ST > vec, int32_t num) |
SGSparseVector< ST > * | get_sparse_feature_matrix (int32_t &num_feat, int32_t &num_vec) |
SGSparseMatrix< ST > | get_sparse_feature_matrix () |
CSparseFeatures< ST > * | get_transposed () |
SGSparseVector< ST > * | get_transposed (int32_t &num_feat, int32_t &num_vec) |
void | set_sparse_feature_matrix (SGSparseMatrix< ST > sm) |
SGMatrix< ST > | get_full_feature_matrix () |
virtual bool | set_full_feature_matrix (SGMatrix< ST > full) |
virtual bool | apply_preprocessor (bool force_preprocessing=false) |
virtual int32_t | get_size () |
bool | obtain_from_simple (CSimpleFeatures< ST > *sf) |
virtual int32_t | get_num_vectors () const |
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 (SGSparseVector< ST > vec, int32_t num) |
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) |
void | load (CFile *loader) |
void | save (CFile *writer) |
CLabels * | load_svmlight_file (char *fname, bool do_sort_features=true) |
void | sort_features () |
bool | write_svmlight_file (char *fname, CLabels *label) |
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 * | 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 |
Static Public Member Functions | |
static ST | sparse_dot (ST alpha, SGSparseVectorEntry< ST > *avec, int32_t alen, SGSparseVectorEntry< ST > *bvec, int32_t blen) |
static void | clean_tsparse (SGSparseVector< ST > *sfm, int32_t num_vec) |
Protected Member Functions | |
virtual SGSparseVectorEntry< ST > * | compute_sparse_feature_vector (int32_t num, int32_t &len, SGSparseVectorEntry< ST > *target=NULL) |
Protected Attributes | |
int32_t | num_vectors |
total number of vectors | |
int32_t | num_features |
total number of features | |
SGSparseVector< ST > * | sparse_feature_matrix |
array of sparse vectors of size num_vectors | |
CCache< SGSparseVectorEntry < ST > > * | feature_cache |
CSparseFeatures | ( | int32_t | size = 0 |
) |
CSparseFeatures | ( | SGSparseVector< ST > * | src, | |
int32_t | num_feat, | |||
int32_t | num_vec, | |||
bool | copy = false | |||
) |
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 | |
copy | true to copy feature matrix |
Definition at line 20 of file SparseFeatures.cpp.
CSparseFeatures | ( | SGSparseMatrix< ST > | sparse | ) |
convenience constructor that creates sparse features from sparse features
sparse | sparse matrix |
Definition at line 42 of file SparseFeatures.cpp.
CSparseFeatures | ( | SGMatrix< ST > | dense | ) |
convenience constructor that creates sparse features from dense features
dense | dense feature matrix |
Definition at line 51 of file SparseFeatures.cpp.
CSparseFeatures | ( | const CSparseFeatures< ST > & | orig | ) |
copy constructor
Definition at line 60 of file SparseFeatures.cpp.
CSparseFeatures | ( | CFile * | loader | ) |
constructor loading features from file
loader | File object to load data from |
Definition at line 83 of file SparseFeatures.cpp.
~CSparseFeatures | ( | ) | [virtual] |
default destructor
Definition at line 92 of file SparseFeatures.cpp.
void 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
possible with subset
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 331 of file SparseFeatures.cpp.
bool apply_preprocessor | ( | bool | force_preprocessing = false |
) | [virtual] |
apply preprocessor
possible with subset
force_preprocessing | if preprocssing shall be forced |
Definition at line 596 of file SparseFeatures.cpp.
void clean_tsparse | ( | SGSparseVector< ST > * | sfm, | |
int32_t | num_vec | |||
) | [static] |
clean SGSparseVector
sfm | sparse feature matrix | |
num_vec | number of vectors in matrix |
Definition at line 394 of file SparseFeatures.cpp.
SGSparseVectorEntry< ST > * compute_sparse_feature_vector | ( | int32_t | num, | |
int32_t & | len, | |||
SGSparseVectorEntry< 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 1117 of file SparseFeatures.cpp.
compute a^2 on all feature vectors
possible with subset
sq | the square for each vector is stored in here |
Definition at line 676 of file SparseFeatures.cpp.
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 | |||
) |
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.
possible wiht subsets on 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 695 of file SparseFeatures.cpp.
Creates a new CFeatures instance containing copies of the elements which are specified by the provided indices.
indices | indices of feature elements to copy |
Reimplemented from CFeatures.
Definition at line 1092 of file SparseFeatures.cpp.
compute dot product between vector1 and a dense vector
possible with subset
vec_idx1 | index of first vector | |
vec2 | pointer to real valued vector | |
vec2_len | length of real valued vector |
Implements CDotFeatures.
Definition at line 1027 of file SparseFeatures.cpp.
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
possible with subset
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 310 of file SparseFeatures.cpp.
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 of this instance and of DotFeatures
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 1008 of file SparseFeatures.cpp.
CFeatures * duplicate | ( | ) | const [virtual] |
duplicate feature object
Implements CFeatures.
Definition at line 110 of file SparseFeatures.cpp.
void free_feature_iterator | ( | void * | iterator | ) | [virtual] |
clean up iterator call this function with the iterator returned by get_first_feature
iterator | as returned by get_first_feature |
Implements CDotFeatures.
Definition at line 1082 of file SparseFeatures.cpp.
void free_feature_vector | ( | SGSparseVector< ST > | vec, | |
int32_t | num | |||
) |
free feature vector
possible with subset
vec | feature vector to free | |
num | index of vector in cache |
Definition at line 658 of file SparseFeatures.cpp.
void free_sparse_feature_matrix | ( | ) |
void free_sparse_feature_vector | ( | SGSparseVector< ST > | vec, | |
int32_t | num | |||
) |
free sparse feature vector
possible with subset
vec | feature vector to free | |
num | index of this vector in the cache |
Definition at line 365 of file SparseFeatures.cpp.
void free_sparse_features | ( | ) |
free sparse feature matrix and cache
any subset is removed
Definition at line 104 of file SparseFeatures.cpp.
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())
Implements CDotFeatures.
Definition at line 1003 of file SparseFeatures.cpp.
ST get_feature | ( | int32_t | num, | |
int32_t | index | |||
) |
get a single feature
possible with subset
num | number of feature vector to retrieve | |
index | index of feature in this vector |
Definition at line 115 of file SparseFeatures.cpp.
EFeatureClass get_feature_class | ( | ) | [virtual] |
get feature class
Implements CFeatures.
Definition at line 653 of file SparseFeatures.cpp.
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
vector_index | the index of the vector over whose components to iterate over |
Implements CDotFeatures.
Definition at line 1050 of file SparseFeatures.cpp.
virtual EFeatureType get_feature_type | ( | ) | [virtual] |
SGMatrix< ST > get_full_feature_matrix | ( | ) |
gets a copy of a full feature matrix
possible with subset
Definition at line 477 of file SparseFeatures.cpp.
SGVector< ST > get_full_feature_vector | ( | int32_t | num | ) |
get the fully expanded dense feature vector num
num | index of feature vector |
Definition at line 160 of file SparseFeatures.cpp.
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 136 of file SparseFeatures.cpp.
virtual const char* get_name | ( | ) | const [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
index | is returned by reference (-1 when not available) | |
value | is returned by reference | |
iterator | as returned by get_first_feature |
Implements CDotFeatures.
Definition at line 1068 of file SparseFeatures.cpp.
int32_t 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 190 of file SparseFeatures.cpp.
int32_t get_num_features | ( | ) |
get number of features
Definition at line 640 of file SparseFeatures.cpp.
int64_t get_num_nonzero_entries | ( | ) |
get number of non-zero entries in sparse feature matrix
Definition at line 666 of file SparseFeatures.cpp.
int32_t get_num_vectors | ( | ) | const [virtual] |
get number of feature vectors, possibly of subset
Implements CFeatures.
Definition at line 635 of file SparseFeatures.cpp.
int32_t get_size | ( | ) | [virtual] |
get memory footprint of one feature
Implements CFeatures.
Definition at line 622 of file SparseFeatures.cpp.
SGSparseVector< 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
not possible with subset
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 373 of file SparseFeatures.cpp.
SGSparseMatrix< ST > get_sparse_feature_matrix | ( | ) |
get the sparse feature matrix
not possible with subset
Definition at line 384 of file SparseFeatures.cpp.
SGSparseVector< ST > get_sparse_feature_vector | ( | int32_t | num | ) |
get sparse feature vector for sample num from the matrix as it is if matrix is initialized, else return preprocessed compute_feature_vector
possible with subset
num | index of feature vector |
Definition at line 198 of file SparseFeatures.cpp.
SGSparseVector< 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
possible with subset
num_feat | number of features in matrix | |
num_vec | number of vectors in matrix |
Definition at line 413 of file SparseFeatures.cpp.
CSparseFeatures< ST > * get_transposed | ( | ) |
get a transposed copy of the features
possible with subset
Definition at line 405 of file SparseFeatures.cpp.
void load | ( | CFile * | loader | ) | [virtual] |
load features from file
any subset is removed before
loader | File object to load data from |
Reimplemented from CFeatures.
CLabels * load_svmlight_file | ( | char * | fname, | |
bool | do_sort_features = true | |||
) |
load features from file
any subset is removed before
fname | filename to load from | |
do_sort_features | if true features will be sorted to ensure they are in ascending order |
Definition at line 755 of file SparseFeatures.cpp.
bool obtain_from_simple | ( | CSimpleFeatures< ST > * | sf | ) |
obtain sparse features from simple features
subset on input is ignored, subset of this instance is removed
sf | simple features |
Definition at line 627 of file SparseFeatures.cpp.
void save | ( | CFile * | writer | ) | [virtual] |
save features to file
not possible with subset
writer | File object to write data to |
Reimplemented from CFeatures.
bool set_full_feature_matrix | ( | SGMatrix< ST > | full | ) | [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
any subset is removed before
full | full feature matrix |
Definition at line 506 of file SparseFeatures.cpp.
int32_t 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 645 of file SparseFeatures.cpp.
void set_sparse_feature_matrix | ( | SGSparseMatrix< ST > | sm | ) |
set sparse feature matrix
not possible with subset
sm | sparse feature matrix |
Definition at line 463 of file SparseFeatures.cpp.
void sort_features | ( | ) |
ensure that features occur in ascending order, only call when no preprocessors are attached
not possiblwe with subset
Definition at line 921 of file SparseFeatures.cpp.
ST sparse_dot | ( | ST | alpha, | |
SGSparseVectorEntry< ST > * | avec, | |||
int32_t | alen, | |||
SGSparseVectorEntry< ST > * | bvec, | |||
int32_t | blen | |||
) | [static] |
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 262 of file SparseFeatures.cpp.
bool write_svmlight_file | ( | char * | fname, | |
CLabels * | label | |||
) |
write features to file using svm light format
not possible with subset
fname | filename to write to | |
label | Label object (number of labels must correspond to number of features) |
Definition at line 966 of file SparseFeatures.cpp.
CCache< SGSparseVectorEntry<ST> >* feature_cache [protected] |
feature cache
Definition at line 557 of file SparseFeatures.h.
int32_t num_features [protected] |
total number of features
Definition at line 551 of file SparseFeatures.h.
int32_t num_vectors [protected] |
total number of vectors
Definition at line 548 of file SparseFeatures.h.
SGSparseVector<ST>* sparse_feature_matrix [protected] |
array of sparse vectors of size num_vectors
Definition at line 554 of file SparseFeatures.h.