bs.h File Reference

#include <memory>

Go to the source code of this file.

Namespaces
	BS

Macros
#define	BS_TYPE_MEAN   0
#define	BS_TYPE_VOTE   1

Functions
LEARNER::base_learner *	bs_setup (VW::config::options_i &options, vw &all)
uint32_t	BS::weight_gen (std::shared_ptr< rand_state > &state)

Macro Definition Documentation

BS_TYPE_MEAN

#define BS_TYPE_MEAN   0

Definition at line 8 of file bs.h.

Referenced by bs_setup(), and predict_or_learn().

BS_TYPE_VOTE

#define BS_TYPE_VOTE   1

Definition at line 9 of file bs.h.

Referenced by bs_setup(), and predict_or_learn().

Function Documentation

bs_setup()

LEARNER::base_learner* bs_setup	(	VW::config::options_i &	options,
		vw &	all
	)

Definition at line 231 of file bs.cc.

References VW::config::option_group_definition::add(), add(), VW::config::options_i::add_and_parse(), LEARNER::as_singleline(), BS_TYPE_MEAN, BS_TYPE_VOTE, finish_example(), vw::get_random_state(), LEARNER::init_learner(), LEARNER::make_base(), VW::config::make_option(), LEARNER::learner< T, E >::set_finish_example(), setup_base(), and VW::config::options_i::was_supplied().

Referenced by parse_reductions().

{
  auto data = scoped_calloc_or_throw<bs>();
  std::string type_string("mean");
  option_group_definition new_options("Bootstrap");
  new_options.add(make_option("bootstrap", data->B).keep().help("k-way bootstrap by online importance resampling"))
      .add(make_option("bs_type", type_string).keep().help("prediction type {mean,vote}"));
  options.add_and_parse(new_options);

  if (!options.was_supplied("bootstrap"))
    return nullptr;

  data->ub = FLT_MAX;
  data->lb = -FLT_MAX;

  if (options.was_supplied("bs_type"))
  {
    if (type_string == "mean")
      data->bs_type = BS_TYPE_MEAN;
    else if (type_string == "vote")
      data->bs_type = BS_TYPE_VOTE;
    else
    {
      std::cerr << "warning: bs_type must be in {'mean','vote'}; resetting to mean." << std::endl;
      data->bs_type = BS_TYPE_MEAN;
    }
  }
  else  // by default use mean
    data->bs_type = BS_TYPE_MEAN;

  data->pred_vec = new std::vector<double>();
  data->pred_vec->reserve(data->B);
  data->all = &all;
  data->_random_state = all.get_random_state();

  learner<bs, example>& l = init_learner(
      data, as_singleline(setup_base(options, all)), predict_or_learn<true>, predict_or_learn<false>, data->B);
  l.set_finish_example(finish_example);

  return make_base(l);
}