HashedWDFeaturesTransposed.h

Go to the documentation of this file.
00001 /*
00002  * This program is free software; you can redistribute it and/or modify
00003  * it under the terms of the GNU General Public License as published by
00004  * the Free Software Foundation; either version 3 of the License, or
00005  * (at your option) any later version.
00006  *
00007  * Written (W) 2010 Soeren Sonnenburg
00008  * Copyright (C) 2010 Berlin Institute of Technology
00009  */
00010 
00011 #ifndef _HASHEDWDFEATURESTRANSPOSED_H___
00012 #define _HASHEDWDFEATURESTRANSPOSED_H___
00013 
00014 #include "lib/common.h"
00015 #include "features/DotFeatures.h"
00016 #include "features/StringFeatures.h"
00017 #include "lib/Hash.h"
00018 
00019 namespace shogun
00020 {
00021 template <class ST> class CStringFeatures;
00022 
00028 class CHashedWDFeaturesTransposed : public CDotFeatures
00029 {
00030     public:
00032         CHashedWDFeaturesTransposed(void);
00033 
00042         CHashedWDFeaturesTransposed(CStringFeatures<uint8_t>* str, int32_t start_order,
00043                 int32_t order,  int32_t from_order, int32_t hash_bits=12);
00044 
00046         CHashedWDFeaturesTransposed(const CHashedWDFeaturesTransposed & orig);
00047 
00049         virtual ~CHashedWDFeaturesTransposed();
00050 
00058         inline virtual int32_t get_dim_feature_space()
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, int32_t vec2_len);
00079 
00091         virtual void dense_dot_range(float64_t* output, int32_t start, int32_t stop, float64_t* alphas, float64_t* vec, int32_t dim, float64_t b);
00092 
00104         virtual void dense_dot_range_subset(int32_t* sub_index, int32_t num, float64_t* output, float64_t* alphas, float64_t* vec, int32_t dim, float64_t b);
00105 
00106 
00115         virtual void add_to_dense_vec(float64_t alpha, int32_t vec_idx1, float64_t* vec2, int32_t vec2_len, bool abs_val=false);
00116 
00122         virtual inline int32_t get_nnz_features_for_vector(int32_t num)
00123         {
00124             return w_dim/alphabet_size;
00125         }
00126 
00131         virtual CFeatures* duplicate() const;
00132 
00137         inline virtual EFeatureType get_feature_type()
00138         {
00139             return F_UNKNOWN;
00140         }
00141 
00146         inline virtual EFeatureClass get_feature_class()
00147         {
00148             return C_WD;
00149         }
00150 
00151         inline virtual int32_t get_num_vectors()
00152         {
00153             return num_strings;
00154         }
00155 
00156         inline virtual int32_t get_size()
00157         {
00158             return sizeof(float64_t);
00159         }
00160 
00163         void set_normalization_const(float64_t n=0);
00164 
00166         inline float64_t get_normalization_const()
00167         {
00168             return normalization_const;
00169         }
00170 
00172         struct hashed_wd_transposed_feature_iterator
00173         {
00175             uint16_t* vec;
00177             int32_t vidx;
00179             int32_t vlen;
00181             bool vfree;
00182 
00184             int32_t index;
00185 
00186         };
00187 
00197         virtual void* get_feature_iterator(int32_t vector_index)
00198         {
00199             SG_NOTIMPLEMENTED;
00200             return NULL;
00201         }
00202 
00213         virtual bool get_next_feature(int32_t& index, float64_t& value, void* iterator)
00214         {
00215             SG_NOTIMPLEMENTED;
00216             return NULL;
00217         }
00218 
00224         virtual void free_feature_iterator(void* iterator)
00225         {
00226             SG_NOTIMPLEMENTED;
00227         }
00228 
00230         inline virtual const char* get_name() const { return "HashedWDFeaturesTransposed"; }
00231 
00232     protected:
00233 
00235         void set_wd_weights();
00236         static void* dense_dot_range_helper(void* p);
00237 
00238     protected:
00240         CStringFeatures<uint8_t>* strings;
00241 
00243         TString<uint8_t>* transposed_strings;
00244 
00246         int32_t degree;
00248         int32_t start_degree;
00250         int32_t from_degree;
00252         int32_t string_length;
00254         int32_t num_strings;
00256         int32_t alphabet_size;
00258         int32_t w_dim;
00260         int32_t partial_w_dim;
00262         float64_t* wd_weights;
00264         uint32_t mask;
00266         int32_t m_hash_bits;
00267 
00269         float64_t normalization_const;
00270 };
00271 }
00272 #endif // _HASHEDWDFEATURESTRANSPOSED_H___
 All Classes Namespaces Files Functions Variables Typedefs Enumerations Enumerator Friends Defines

SHOGUN Machine Learning Toolbox - Documentation