/**
* Counts features in a preliminary feature vector
* @param fv Valid feature vector
*/
static void count_feat(fvec_t *fv)
{
feat_t *p_dim = fv->dim;
float n = 0, *p_val = fv->val;
unsigned int i;
/* Loop over features */
for (i = 0; i < fv->len; i++) {
/* Skip zero values */
if (fabs(fv->val[i]) < 1e-12)
continue;
/* Check for duplicate dims */
if (i < fv->len - 1 && fv->dim[i] == fv->dim[i + 1]) {
n += fv->val[i];
} else {
*(p_dim++) = fv->dim[i];
*(p_val++) = fv->val[i] + n;
n = 0;
}
}
/* Update length */
fv->len = p_dim - fv->dim;
/* Reallocate memory */
fvec_realloc(fv);
}
/**
* Condense a feature vector by counting duplicate features.
* @param fv Feature vector
*/
static void fvec_condense(fvec_t *fv, embed_t e)
{
feat_t *p_dim = fv->dim;
float n = 0, *p_val = fv->val;
unsigned int i;
/* Loop over features */
for (i = 0; i < fv->len; i++) {
/* Skip zero values */
if (fabs(fv->val[i]) < 1e-12)
continue;
/* Check for duplicate dims */
if (i < fv->len - 1 && fv->dim[i] == fv->dim[i + 1]) {
switch (e) {
case EMBED_CNT:
n += fv->val[i];
break;
case EMBED_BIN:
n = (float) fmax(n, fv->val[i]);
break;
}
} else {
*(p_dim++) = fv->dim[i];
*(p_val++) = fv->val[i] + n;
n = 0;
}
}
/* Update length */
fv->len = p_dim - fv->dim;
fv->mem += fv->len * (sizeof(feat_t) + sizeof(float));
/* Reallocate memory */
fvec_realloc(fv);
}
