gd_test.cpp

#include <mlpack/core.hpp>
#include <mlpack/methods/gmm/gmm.hpp>

PARAM_STRING_REQ("input_file", "Input dataset.", "i");

using namespace mlpack;
using namespace mlpack::distribution;
using namespace std;

int main(int argc, char** argv)
{
  CLI::ParseCommandLine(argc, argv);

  // Load the dataset.
  const string inputFile = CLI::GetParam<string>("input_file");
  arma::mat dataset;
  data::Load(inputFile, dataset, true);

  Timer::Start("estimate");
  GaussianDistribution g(dataset.n_rows);
  g.Estimate(dataset);
  Timer::Stop("estimate");

  // Make some random points and get their probability.
  arma::mat random;
  random.randu(dataset.n_rows, 100000);

  Timer::Start("probability_batch");
  arma::vec probabilities;
  g.Probability(random, probabilities);
  Timer::Stop("probability_batch");

  Timer::Start("probability_individual");
  for (size_t i = 0; i < random.n_cols; ++i)
    probabilities[i] = g.Probability(random.unsafe_col(i));
  Timer::Stop("probability_individual");

  // Generate random observations.
  Timer::Start("random");
  for (size_t i = 0; i < random.n_cols; ++i)
    random.col(i) = g.Random();
  Timer::Stop("random");

  // Do something kind of like GMM training.
  Timer::Start("gmm_training_imitation");
  for (size_t i = 0; i < 50; ++i)
  {
    g.Probability(dataset, probabilities);

    arma::mat covariance(dataset.n_rows, dataset.n_rows);
    covariance.randu();
    covariance = covariance.t() * covariance;
    g.Covariance(covariance);
  }
  Timer::Stop("gmm_training_imitation");
}