Go to the documentation of this file.00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011 #ifndef _HASHEDWDFEATURES_H___
00012 #define _HASHEDWDFEATURES_H___
00013
00014 #include <shogun/lib/common.h>
00015 #include <shogun/features/DotFeatures.h>
00016 #include <shogun/features/StringFeatures.h>
00017 #include <shogun/lib/Hash.h>
00018
00019 namespace shogun
00020 {
00021 template<class ST> class CStringFeatures;
00022
00028 class CHashedWDFeatures: public CDotFeatures
00029 {
00030 public:
00032 CHashedWDFeatures();
00033
00042 CHashedWDFeatures(CStringFeatures<uint8_t>* str, int32_t start_order,
00043 int32_t order, int32_t from_order, int32_t hash_bits=12);
00044
00046 CHashedWDFeatures(const CHashedWDFeatures & orig);
00047
00049 virtual ~CHashedWDFeatures();
00050
00058 virtual int32_t get_dim_feature_space() const
00059 {
00060 return w_dim;
00061 }
00062
00070 virtual float64_t dot(int32_t vec_idx1, CDotFeatures* df, int32_t vec_idx2);
00071
00078 virtual float64_t dense_dot(int32_t vec_idx1, const float64_t* vec2,
00079 int32_t vec2_len);
00080
00089 virtual void add_to_dense_vec(float64_t alpha, int32_t vec_idx1,
00090 float64_t* vec2, int32_t vec2_len, bool abs_val=false);
00091
00097 virtual int32_t get_nnz_features_for_vector(int32_t num);
00098
00099 #ifndef DOXYGEN_SHOULD_SKIP_THIS
00100
00101 struct hashed_wd_feature_iterator
00102 {
00104 uint16_t* vec;
00106 int32_t vidx;
00108 int32_t vlen;
00110 bool vfree;
00111
00113 int32_t index;
00114
00115 };
00116 #endif
00117
00127 virtual void* get_feature_iterator(int32_t vector_index);
00128
00139 virtual bool get_next_feature(int32_t& index, float64_t& value,
00140 void* iterator);
00141
00147 virtual void free_feature_iterator(void* iterator);
00148
00153 virtual CFeatures* duplicate() const;
00154
00159 virtual EFeatureType get_feature_type() const
00160 {
00161 return F_UNKNOWN;
00162 }
00163
00168 virtual EFeatureClass get_feature_class() const
00169 {
00170 return C_WD;
00171 }
00172
00173 virtual int32_t get_num_vectors() const
00174 {
00175 return num_strings;
00176 }
00177
00178 virtual int32_t get_size() const
00179 {
00180 return sizeof(float64_t);
00181 }
00182
00185 void set_normalization_const(float64_t n=0);
00186
00188 inline float64_t get_normalization_const()
00189 {
00190 return normalization_const;
00191 }
00192
00194 virtual const char* get_name() const
00195 {
00196 return "HashedWDFeatures";
00197 }
00198
00199 protected:
00200
00202 void set_wd_weights();
00203
00204 protected:
00206 CStringFeatures<uint8_t>* strings;
00207
00209 int32_t degree;
00211 int32_t start_degree;
00213 int32_t from_degree;
00215 int32_t string_length;
00217 int32_t num_strings;
00219 int32_t alphabet_size;
00221 int32_t w_dim;
00223 int32_t partial_w_dim;
00225 float64_t* wd_weights;
00227 uint32_t mask;
00229 int32_t m_hash_bits;
00230
00232 float64_t normalization_const;
00233 };
00234 }
00235 #endif // _HASHEDWDFEATURES_H___