LocalityImprovedStringKernel.cpp
Go to the documentation of this file.00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012 #include "lib/common.h"
00013 #include "lib/io.h"
00014 #include "kernel/LocalityImprovedStringKernel.h"
00015 #include "features/StringFeatures.h"
00016
00017 CLocalityImprovedStringKernel::CLocalityImprovedStringKernel(
00018 int32_t size, int32_t l, int32_t id, int32_t od)
00019 : CStringKernel<char>(size), length(l), inner_degree(id), outer_degree(od)
00020 {
00021 SG_INFO( "LIK with parms: l=%d, id=%d, od=%d created!\n", l, id, od);
00022 }
00023
00024 CLocalityImprovedStringKernel::CLocalityImprovedStringKernel(
00025 CStringFeatures<char>* l, CStringFeatures<char>* r, int32_t len,
00026 int32_t id, int32_t od)
00027 : CStringKernel<char>(10), length(len), inner_degree(id), outer_degree(od)
00028 {
00029 SG_INFO( "LIK with parms: l=%d, id=%d, od=%d created!\n", len, id, od);
00030
00031 init(l, r);
00032 }
00033
00034 CLocalityImprovedStringKernel::~CLocalityImprovedStringKernel()
00035 {
00036 cleanup();
00037 }
00038
00039 bool CLocalityImprovedStringKernel::init(CFeatures* l, CFeatures* r)
00040 {
00041 CStringKernel<char>::init(l,r);
00042 return init_normalizer();
00043 }
00044
00045 bool CLocalityImprovedStringKernel::load_init(FILE* src)
00046 {
00047 return false;
00048 }
00049
00050 bool CLocalityImprovedStringKernel::save_init(FILE* dest)
00051 {
00052 return false;
00053 }
00054
00055 float64_t CLocalityImprovedStringKernel::compute(int32_t idx_a, int32_t idx_b)
00056 {
00057 int32_t alen, blen;
00058
00059 char* avec = ((CStringFeatures<char>*) lhs)->get_feature_vector(idx_a, alen);
00060 char* bvec = ((CStringFeatures<char>*) rhs)->get_feature_vector(idx_b, blen);
00061
00062 ASSERT(alen==blen && alen>0);
00063
00064 int32_t i,t;
00065 float64_t* match=new float64_t[alen];
00066
00067
00068 for (i = 0; i<alen; i++)
00069 match[i] = (avec[i] == bvec[i])? 1 : 0;
00070
00071 float64_t outer_sum = 0;
00072
00073 for (t = 0; t<alen-length; t++)
00074 {
00075 float64_t sum = 0;
00076 for (i = 0; i<length && t+i+length+1<alen; i++)
00077 sum += (i+1)*match[t+i]+(length-i)*match[t+i+length+1];
00078
00079 float64_t inner_sum = (sum + (length+1)*match[t+length]) / ((2*length+1)*(length+1));
00080 inner_sum = pow(inner_sum, inner_degree + 1);
00081 outer_sum += inner_sum;
00082 }
00083 delete[] match;
00084 return pow(outer_sum, outer_degree + 1);
00085 }