ra_search_rules.hpp

Go to the documentation of this file.

#ifndef MLPACK_METHODS_RANN_RA_SEARCH_RULES_HPP
#define MLPACK_METHODS_RANN_RA_SEARCH_RULES_HPP

#include <mlpack/core/tree/traversal_info.hpp>

namespace mlpack {
namespace neighbor {

template<typename SortPolicy, typename MetricType, typename TreeType>
class RASearchRules
{
 public:
  RASearchRules(const arma::mat& referenceSet,
                const arma::mat& querySet,
                const size_t k,
                MetricType& metric,
                const double tau = 5,
                const double alpha = 0.95,
                const bool naive = false,
                const bool sampleAtLeaves = false,
                const bool firstLeafExact = false,
                const size_t singleSampleLimit = 20,
                const bool sameSet = false);

  void GetResults(arma::Mat<size_t>& neighbors, arma::mat& distances);

  double BaseCase(const size_t queryIndex, const size_t referenceIndex);

  double Score(const size_t queryIndex, TreeType& referenceNode);

  double Score(const size_t queryIndex,
               TreeType& referenceNode,
               const double baseCaseResult);

  double Rescore(const size_t queryIndex,
                 TreeType& referenceNode,
                 const double oldScore);

  double Score(TreeType& queryNode, TreeType& referenceNode);

  double Score(TreeType& queryNode,
               TreeType& referenceNode,
               const double baseCaseResult);

  double Rescore(TreeType& queryNode,
                 TreeType& referenceNode,
                 const double oldScore);


  size_t NumDistComputations() { return numDistComputations; }
  size_t NumEffectiveSamples()
  {
    if (numSamplesMade.n_elem == 0)
      return 0;
    else
      return arma::sum(numSamplesMade);
  }

  typedef typename tree::TraversalInfo<TreeType> TraversalInfoType;

  const TraversalInfoType& TraversalInfo() const { return traversalInfo; }
  TraversalInfoType& TraversalInfo() { return traversalInfo; }

 private:
  const arma::mat& referenceSet;

  const arma::mat& querySet;

  typedef std::pair<double, size_t> Candidate;

  struct CandidateCmp {
    bool operator()(const Candidate& c1, const Candidate& c2)
    {
      return !SortPolicy::IsBetter(c2.first, c1.first);
    };
  };

  typedef std::priority_queue<Candidate, std::vector<Candidate>, CandidateCmp>
      CandidateList;

  std::vector<CandidateList> candidates;

  const size_t k;

  MetricType& metric;

  bool sampleAtLeaves;

  bool firstLeafExact;

  size_t singleSampleLimit;

  size_t numSamplesReqd;

  arma::Col<size_t> numSamplesMade;

  double samplingRatio;

  // TO REMOVE: just for testing
  size_t numDistComputations;

  bool sameSet;

  TraversalInfoType traversalInfo;

  void InsertNeighbor(const size_t queryIndex,
                      const size_t neighbor,
                      const double distance);

  double Score(const size_t queryIndex,
               TreeType& referenceNode,
               const double distance,
               const double bestDistance);

  double Score(TreeType& queryNode,
               TreeType& referenceNode,
               const double distance,
               const double bestDistance);

  static_assert(tree::TreeTraits<TreeType>::UniqueNumDescendants, "TreeType "
      "must provide a unique number of descendants points.");
}; // class RASearchRules

} // namespace neighbor
} // namespace mlpack

// Include implementation.
#include "ra_search_rules_impl.hpp"

#endif // MLPACK_METHODS_RANN_RA_SEARCH_RULES_HPP