mirror of https://github.com/CGAL/cgal
Optimize training
This commit is contained in:
Simon Giraudot
parent
89b6c64cdf
commit
1255a9e794
|
|
@ -278,14 +278,17 @@ public:
|
|||
1. / (double)nb_trials_per_feature);
|
||||
|
||||
|
||||
double best_score = compute_worst_score(0.);
|
||||
double best_confidence = compute_worst_confidence(0.);
|
||||
double best_score = 0.;
|
||||
double best_confidence = 0.;
|
||||
boost::tie (best_confidence, best_score)
|
||||
= compute_worst_confidence_and_score (0., 0.);
|
||||
|
||||
CGAL_CLASSIFICATION_CERR << "TRAINING GLOBALLY: Best score evolution: " << std::endl;
|
||||
|
||||
CGAL_CLASSIFICATION_CERR << 100. * best_score << "% (found at initialization)" << std::endl;
|
||||
|
||||
std::size_t current_att_changed = 0;
|
||||
CGAL::Timer teff, tconf, tscore;
|
||||
for (std::size_t i = 0; i < att_used; ++ i)
|
||||
{
|
||||
while (att_train[current_att_changed].skipped)
|
||||
|
|
@ -302,7 +305,9 @@ public:
|
|||
continue;
|
||||
|
||||
m_classifier->feature(j)->set_weight(best_weights[j]);
|
||||
teff.start();
|
||||
estimate_feature_effect(m_classifier->feature(j));
|
||||
teff.stop();
|
||||
if (feature_useful(m_classifier->feature(j)))
|
||||
nb_used ++;
|
||||
}
|
||||
|
|
@ -312,11 +317,15 @@ public:
|
|||
current_att->set_weight(tr.wmin);
|
||||
for (std::size_t j = 0; j < nb_trials_per_feature; ++ j)
|
||||
{
|
||||
teff.start();
|
||||
estimate_feature_effect(current_att);
|
||||
teff.stop();
|
||||
|
||||
double worst_confidence = compute_worst_confidence(best_confidence);
|
||||
|
||||
double worst_score = compute_worst_score(best_score);
|
||||
tconf.start();
|
||||
double worst_confidence = 0., worst_score = 0.;
|
||||
boost::tie (worst_confidence, worst_score)
|
||||
= compute_worst_confidence_and_score (best_confidence, best_score);
|
||||
tconf.stop();
|
||||
|
||||
if (worst_score > best_score
|
||||
&& worst_confidence > best_confidence)
|
||||
|
|
@ -339,6 +348,10 @@ public:
|
|||
++ current_att_changed;
|
||||
}
|
||||
|
||||
std::cerr << "Estimation of effects = " << teff.time() << std::endl
|
||||
<< "Confidence computation = " << tconf.time() << std::endl
|
||||
<< "Score computation = " << tscore.time() << std::endl;
|
||||
|
||||
for (std::size_t i = 0; i < best_weights.size(); ++ i)
|
||||
{
|
||||
Feature_handle att = m_classifier->feature(i);
|
||||
|
|
@ -472,7 +485,7 @@ public:
|
|||
the sum of the true positives, of the false positives and of the
|
||||
false negatives.
|
||||
*/
|
||||
double IoU (Label_handle label) const
|
||||
double intersection_over_union (Label_handle label) const
|
||||
{
|
||||
std::size_t label_idx = (std::size_t)(-1);
|
||||
for (std::size_t i = 0; i < m_classifier->number_of_labels(); ++ i)
|
||||
|
|
@ -502,10 +515,10 @@ public:
|
|||
double mean_f1_score() const { return m_mean_f1; }
|
||||
|
||||
/*!
|
||||
\brief Returns the mean IoU of the training over all labels (see
|
||||
`IoU()`).
|
||||
\brief Returns the mean intersection over union of the training
|
||||
over all labels (see `intersection_over_union()`).
|
||||
*/
|
||||
double mean_IoU() const { return m_mean_iou; }
|
||||
double mean_intersection_over_union() const { return m_mean_iou; }
|
||||
|
||||
/// @}
|
||||
|
||||
|
|
@ -563,50 +576,15 @@ private:
|
|||
}
|
||||
}
|
||||
|
||||
|
||||
double compute_worst_score (double lower_bound)
|
||||
{
|
||||
double worst_score = 1.;
|
||||
for (std::size_t j = 0; j < m_classifier->number_of_labels(); ++ j)
|
||||
{
|
||||
std::size_t nb_okay = 0;
|
||||
for (std::size_t k = 0; k < m_training_sets[j].size(); ++ k)
|
||||
{
|
||||
std::size_t nb_class_best=0;
|
||||
double val_class_best = (std::numeric_limits<double>::max)();
|
||||
|
||||
for(std::size_t l = 0; l < m_classifier->number_of_labels(); ++ l)
|
||||
{
|
||||
double value = m_classifier->energy_of (m_classifier->label(l),
|
||||
m_training_sets[j][k]);
|
||||
|
||||
if(val_class_best > value)
|
||||
{
|
||||
val_class_best = value;
|
||||
nb_class_best = l;
|
||||
}
|
||||
}
|
||||
|
||||
if (nb_class_best == j)
|
||||
nb_okay ++;
|
||||
|
||||
}
|
||||
|
||||
double score = nb_okay / (double)(m_training_sets[j].size());
|
||||
if (score < worst_score)
|
||||
worst_score = score;
|
||||
if (worst_score < lower_bound)
|
||||
return worst_score;
|
||||
}
|
||||
return worst_score;
|
||||
}
|
||||
|
||||
double compute_worst_confidence (double lower_bound)
|
||||
std::pair<double, double> compute_worst_confidence_and_score (double lower_conf, double lower_score)
|
||||
{
|
||||
double worst_confidence = (std::numeric_limits<double>::max)();
|
||||
double worst_score = (std::numeric_limits<double>::max)();
|
||||
|
||||
for (std::size_t j = 0; j < m_classifier->number_of_labels(); ++ j)
|
||||
{
|
||||
double confidence = 0.;
|
||||
std::size_t nb_okay = 0;
|
||||
|
||||
for (std::size_t k = 0; k < m_training_sets[j].size(); ++ k)
|
||||
{
|
||||
|
|
@ -619,27 +597,24 @@ private:
|
|||
std::sort (values.begin(), values.end());
|
||||
|
||||
if (values[0].second == j)
|
||||
confidence += values[1].first - values[0].first;
|
||||
else
|
||||
{
|
||||
// for(std::size_t l = 0; l < values.size(); ++ l)
|
||||
// if (values[l].second == j)
|
||||
// {
|
||||
// confidence += values[0].first - values[l].first;
|
||||
// break;
|
||||
// }
|
||||
confidence += values[1].first - values[0].first;
|
||||
++ nb_okay;
|
||||
}
|
||||
|
||||
}
|
||||
|
||||
|
||||
double score = nb_okay / (double)(m_training_sets[j].size());
|
||||
confidence /= (double)(m_training_sets[j].size() * m_classifier->number_of_features());
|
||||
|
||||
if (confidence < worst_confidence)
|
||||
worst_confidence = confidence;
|
||||
if (worst_confidence < lower_bound)
|
||||
return worst_confidence;
|
||||
if (score < worst_score)
|
||||
worst_score = score;
|
||||
|
||||
if (worst_confidence < lower_conf || worst_score < lower_score)
|
||||
return std::make_pair (worst_confidence, worst_score);
|
||||
}
|
||||
return worst_confidence;
|
||||
return std::make_pair (worst_confidence, worst_score);
|
||||
}
|
||||
|
||||
bool feature_useful (Feature_handle att)
|
||||
|
|
|
|||
Loading…
Reference in New Issue