Files
mlpack/src/mlpack/tests/knn_test.cpp
T

1156 lines
43 KiB
C++

/**
* @file knn_test.cpp
*
* Test file for KNN class.
*/
#include <mlpack/core.hpp>
#include <mlpack/methods/neighbor_search/neighbor_search.hpp>
#include <mlpack/methods/neighbor_search/unmap.hpp>
#include <mlpack/methods/neighbor_search/ns_model.hpp>
#include <mlpack/core/tree/cover_tree.hpp>
#include <mlpack/core/tree/example_tree.hpp>
#include <boost/test/unit_test.hpp>
#include "test_tools.hpp"
using namespace mlpack;
using namespace mlpack::neighbor;
using namespace mlpack::tree;
using namespace mlpack::metric;
using namespace mlpack::bound;
BOOST_AUTO_TEST_SUITE(KNNTest);
/**
* Test that Unmap() works in the dual-tree case (see unmap.hpp).
*/
BOOST_AUTO_TEST_CASE(DualTreeUnmapTest)
{
std::vector<size_t> refMap;
refMap.push_back(3);
refMap.push_back(4);
refMap.push_back(1);
refMap.push_back(2);
refMap.push_back(0);
std::vector<size_t> queryMap;
queryMap.push_back(2);
queryMap.push_back(0);
queryMap.push_back(4);
queryMap.push_back(3);
queryMap.push_back(1);
queryMap.push_back(5);
// Now generate some results. 6 queries, 5 references.
arma::Mat<size_t> neighbors("3 1 2 0 4;"
"1 0 2 3 4;"
"0 1 2 3 4;"
"4 1 0 3 2;"
"3 0 4 1 2;"
"3 0 4 1 2;");
neighbors = neighbors.t();
// Integer distances will work fine here.
arma::mat distances("3 1 2 0 4;"
"1 0 2 3 4;"
"0 1 2 3 4;"
"4 1 0 3 2;"
"3 0 4 1 2;"
"3 0 4 1 2;");
distances = distances.t();
// This is what the results should be when they are unmapped.
arma::Mat<size_t> correctNeighbors("4 3 1 2 0;"
"2 3 0 4 1;"
"2 4 1 3 0;"
"0 4 3 2 1;"
"3 4 1 2 0;"
"2 3 0 4 1;");
correctNeighbors = correctNeighbors.t();
arma::mat correctDistances("1 0 2 3 4;"
"3 0 4 1 2;"
"3 1 2 0 4;"
"4 1 0 3 2;"
"0 1 2 3 4;"
"3 0 4 1 2;");
correctDistances = correctDistances.t();
// Perform the unmapping.
arma::Mat<size_t> neighborsOut;
arma::mat distancesOut;
Unmap(neighbors, distances, refMap, queryMap, neighborsOut, distancesOut);
for (size_t i = 0; i < correctNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(neighborsOut[i], correctNeighbors[i]);
BOOST_REQUIRE_CLOSE(distancesOut[i], correctDistances[i], 1e-5);
}
// Now try taking the square root.
Unmap(neighbors, distances, refMap, queryMap, neighborsOut, distancesOut,
true);
for (size_t i = 0; i < correctNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(neighborsOut[i], correctNeighbors[i]);
BOOST_REQUIRE_CLOSE(distancesOut[i], sqrt(correctDistances[i]), 1e-5);
}
}
/**
* Check that Unmap() works in the single-tree case.
*/
BOOST_AUTO_TEST_CASE(SingleTreeUnmapTest)
{
std::vector<size_t> refMap;
refMap.push_back(3);
refMap.push_back(4);
refMap.push_back(1);
refMap.push_back(2);
refMap.push_back(0);
// Now generate some results. 6 queries, 5 references.
arma::Mat<size_t> neighbors("3 1 2 0 4;"
"1 0 2 3 4;"
"0 1 2 3 4;"
"4 1 0 3 2;"
"3 0 4 1 2;"
"3 0 4 1 2;");
neighbors = neighbors.t();
// Integer distances will work fine here.
arma::mat distances("3 1 2 0 4;"
"1 0 2 3 4;"
"0 1 2 3 4;"
"4 1 0 3 2;"
"3 0 4 1 2;"
"3 0 4 1 2;");
distances = distances.t();
// This is what the results should be when they are unmapped.
arma::Mat<size_t> correctNeighbors("2 4 1 3 0;"
"4 3 1 2 0;"
"3 4 1 2 0;"
"0 4 3 2 1;"
"2 3 0 4 1;"
"2 3 0 4 1;");
correctNeighbors = correctNeighbors.t();
arma::mat correctDistances = distances;
// Perform the unmapping.
arma::Mat<size_t> neighborsOut;
arma::mat distancesOut;
Unmap(neighbors, distances, refMap, neighborsOut, distancesOut);
for (size_t i = 0; i < correctNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(neighborsOut[i], correctNeighbors[i]);
BOOST_REQUIRE_CLOSE(distancesOut[i], correctDistances[i], 1e-5);
}
// Now try taking the square root.
Unmap(neighbors, distances, refMap, neighborsOut, distancesOut, true);
for (size_t i = 0; i < correctNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(neighborsOut[i], correctNeighbors[i]);
BOOST_REQUIRE_CLOSE(distancesOut[i], sqrt(correctDistances[i]), 1e-5);
}
}
/**
* Test that an empty KNN object will throw exceptions when Search() is
* called.
*/
BOOST_AUTO_TEST_CASE(EmptySearchTest)
{
KNN empty;
arma::mat dataset = arma::randu<arma::mat>(5, 100);
KNN::Tree queryTree(dataset);
arma::Mat<size_t> neighbors;
arma::mat distances;
BOOST_REQUIRE_THROW(empty.Search(dataset, 5, neighbors, distances),
std::invalid_argument);
BOOST_REQUIRE_THROW(empty.Search(5, neighbors, distances),
std::invalid_argument);
BOOST_REQUIRE_THROW(empty.Search(&queryTree, 5, neighbors, distances),
std::invalid_argument);
}
/**
* Test that when training is performed, the results are the same.
*/
BOOST_AUTO_TEST_CASE(TrainTest)
{
KNN empty;
arma::mat dataset = arma::randu<arma::mat>(5, 100);
KNN baseline(dataset);
arma::Mat<size_t> neighbors, baselineNeighbors;
arma::mat distances, baselineDistances;
empty.Train(dataset);
empty.Search(5, neighbors, distances);
baseline.Search(5, baselineNeighbors, baselineDistances);
BOOST_REQUIRE_EQUAL(neighbors.n_rows, baselineNeighbors.n_rows);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, baselineNeighbors.n_cols);
BOOST_REQUIRE_EQUAL(distances.n_rows, baselineDistances.n_rows);
BOOST_REQUIRE_EQUAL(distances.n_cols, baselineDistances.n_cols);
for (size_t i = 0; i < distances.n_elem; ++i)
{
if (std::abs(baselineDistances[i]) < 1e-5)
BOOST_REQUIRE_SMALL(distances[i], 1e-5);
else
BOOST_REQUIRE_CLOSE(distances[i], baselineDistances[i], 1e-5);
BOOST_REQUIRE_EQUAL(neighbors[i], baselineNeighbors[i]);
}
}
/**
* Test that when training is performed with a tree, the results are the same.
*/
BOOST_AUTO_TEST_CASE(TrainTreeTest)
{
KNN empty;
arma::mat dataset = arma::randu<arma::mat>(5, 100);
KNN baseline(dataset);
arma::Mat<size_t> neighbors, baselineNeighbors;
arma::mat distances, baselineDistances;
std::vector<size_t> oldFromNewReferences;
KNN::Tree tree(dataset, oldFromNewReferences);
empty.Train(&tree);
empty.Search(5, neighbors, distances);
baseline.Search(5, baselineNeighbors, baselineDistances);
BOOST_REQUIRE_EQUAL(neighbors.n_rows, baselineNeighbors.n_rows);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, baselineNeighbors.n_cols);
BOOST_REQUIRE_EQUAL(distances.n_rows, baselineDistances.n_rows);
BOOST_REQUIRE_EQUAL(distances.n_cols, baselineDistances.n_cols);
// We have to unmap the results.
arma::mat tmpDistances(distances.n_rows, distances.n_cols);
arma::Mat<size_t> tmpNeighbors(neighbors.n_rows, neighbors.n_cols);
for (size_t i = 0; i < distances.n_cols; ++i)
{
tmpDistances.col(oldFromNewReferences[i]) = distances.col(i);
for (size_t j = 0; j < distances.n_rows; ++j)
{
tmpNeighbors(j, oldFromNewReferences[i]) =
oldFromNewReferences[neighbors(j, i)];
}
}
for (size_t i = 0; i < distances.n_elem; ++i)
{
if (std::abs(baselineDistances[i]) < 1e-5)
BOOST_REQUIRE_SMALL(tmpDistances[i], 1e-5);
else
BOOST_REQUIRE_CLOSE(tmpDistances[i], baselineDistances[i], 1e-5);
BOOST_REQUIRE_EQUAL(tmpNeighbors[i], baselineNeighbors[i]);
}
}
/**
* Test that training with a tree throws an exception when in naive mode.
*/
BOOST_AUTO_TEST_CASE(NaiveTrainTreeTest)
{
KNN empty(true);
arma::mat dataset = arma::randu<arma::mat>(5, 100);
KNN::Tree tree(dataset);
BOOST_REQUIRE_THROW(empty.Train(&tree), std::invalid_argument);
}
/**
* Test that the rvalue reference move constructor works.
*/
BOOST_AUTO_TEST_CASE(MoveConstructorTest)
{
arma::mat dataset = arma::randu<arma::mat>(3, 200);
arma::mat copy(dataset);
KNN moveknn(std::move(copy));
KNN knn(dataset);
BOOST_REQUIRE_EQUAL(copy.n_elem, 0);
BOOST_REQUIRE_EQUAL(moveknn.ReferenceSet().n_rows, 3);
BOOST_REQUIRE_EQUAL(moveknn.ReferenceSet().n_cols, 200);
arma::mat moveDistances, distances;
arma::Mat<size_t> moveNeighbors, neighbors;
moveknn.Search(1, moveNeighbors, moveDistances);
knn.Search(1, neighbors, distances);
BOOST_REQUIRE_EQUAL(moveNeighbors.n_rows, neighbors.n_rows);
BOOST_REQUIRE_EQUAL(moveNeighbors.n_cols, neighbors.n_cols);
BOOST_REQUIRE_EQUAL(moveDistances.n_rows, distances.n_rows);
BOOST_REQUIRE_EQUAL(moveDistances.n_cols, distances.n_cols);
for (size_t i = 0; i < moveDistances.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(moveNeighbors[i], neighbors[i]);
if (std::abs(distances[i]) < 1e-5)
BOOST_REQUIRE_SMALL(moveDistances[i], 1e-5);
else
BOOST_REQUIRE_CLOSE(moveDistances[i], distances[i], 1e-5);
}
}
/**
* Test that the dataset can be retrained with the move Train() function.
*/
BOOST_AUTO_TEST_CASE(MoveTrainTest)
{
arma::mat dataset = arma::randu<arma::mat>(3, 200);
// Do it in tree mode, and in naive mode.
KNN knn;
knn.Train(std::move(dataset));
arma::mat distances;
arma::Mat<size_t> neighbors;
knn.Search(1, neighbors, distances);
BOOST_REQUIRE_EQUAL(dataset.n_elem, 0);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, 200);
BOOST_REQUIRE_EQUAL(distances.n_cols, 200);
dataset = arma::randu<arma::mat>(3, 300);
knn.Naive() = true;
knn.Train(std::move(dataset));
knn.Search(1, neighbors, distances);
BOOST_REQUIRE_EQUAL(dataset.n_elem, 0);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, 300);
BOOST_REQUIRE_EQUAL(distances.n_cols, 300);
}
/**
* Simple nearest-neighbors test with small, synthetic dataset. This is an
* exhaustive test, which checks that each method for performing the calculation
* (dual-tree, single-tree, naive) produces the correct results. An
* eleven-point dataset and the ten nearest neighbors are taken. The dataset is
* in one dimension for simplicity -- the correct functionality of distance
* functions is not tested here.
*/
BOOST_AUTO_TEST_CASE(ExhaustiveSyntheticTest)
{
// Set up our data.
arma::mat data(1, 11);
data[0] = 0.05; // Row addressing is unnecessary (they are all 0).
data[1] = 0.35;
data[2] = 0.15;
data[3] = 1.25;
data[4] = 5.05;
data[5] = -0.22;
data[6] = -2.00;
data[7] = -1.30;
data[8] = 0.45;
data[9] = 0.90;
data[10] = 1.00;
typedef KDTree<EuclideanDistance, NeighborSearchStat<NearestNeighborSort>,
arma::mat> TreeType;
// We will loop through three times, one for each method of performing the
// calculation.
std::vector<size_t> oldFromNew;
std::vector<size_t> newFromOld;
TreeType* tree = new TreeType(data, oldFromNew, newFromOld, 1);
for (int i = 0; i < 3; i++)
{
KNN* knn;
switch (i)
{
case 0: // Use the dual-tree method.
knn = new KNN(tree, false);
break;
case 1: // Use the single-tree method.
knn = new KNN(tree, true);
break;
case 2: // Use the naive method.
knn = new KNN(tree->Dataset(), true);
break;
}
// Now perform the actual calculation.
arma::Mat<size_t> neighbors;
arma::mat distances;
knn->Search(10, neighbors, distances);
// Now the exhaustive check for correctness. This will be long. We must
// also remember that the distances returned are squared distances. As a
// result, distance comparisons are written out as (distance * distance) for
// readability.
// Neighbors of point 0.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[0]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[0]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[0]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[0]), 0.27, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[0]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[0]), 0.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[0]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[0]), 0.40, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[0]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[0]), 0.85, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[0]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[0]), 0.95, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[0]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[0]), 1.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[0]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[0]), 1.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[0]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[0]), 2.05, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[0]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[0]), 5.00, 1e-5);
// Neighbors of point 1.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[1]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[1]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[1]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[1]), 0.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[1]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[1]), 0.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[1]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[1]), 0.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[1]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[1]), 0.57, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[1]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[1]), 0.65, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[1]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[1]), 0.90, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[1]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[1]), 1.65, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[1]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[1]), 2.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[1]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[1]), 4.70, 1e-5);
// Neighbors of point 2.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[2]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[2]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[2]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[2]), 0.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[2]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[2]), 0.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[2]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[2]), 0.37, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[2]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[2]), 0.75, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[2]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[2]), 0.85, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[2]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[2]), 1.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[2]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[2]), 1.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[2]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[2]), 2.15, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[2]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[2]), 4.90, 1e-5);
// Neighbors of point 3.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[3]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[3]), 0.25, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[3]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[3]), 0.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[3]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[3]), 0.80, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[3]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[3]), 0.90, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[3]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[3]), 1.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[3]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[3]), 1.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[3]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[3]), 1.47, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[3]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[3]), 2.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[3]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[3]), 3.25, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[3]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[3]), 3.80, 1e-5);
// Neighbors of point 4.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[4]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[4]), 3.80, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[4]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[4]), 4.05, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[4]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[4]), 4.15, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[4]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[4]), 4.60, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[4]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[4]), 4.70, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[4]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[4]), 4.90, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[4]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[4]), 5.00, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[4]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[4]), 5.27, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[4]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[4]), 6.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[4]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[4]), 7.05, 1e-5);
// Neighbors of point 5.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[5]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[5]), 0.27, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[5]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[5]), 0.37, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[5]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[5]), 0.57, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[5]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[5]), 0.67, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[5]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[5]), 1.08, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[5]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[5]), 1.12, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[5]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[5]), 1.22, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[5]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[5]), 1.47, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[5]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[5]), 1.78, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[5]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[5]), 5.27, 1e-5);
// Neighbors of point 6.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[6]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[6]), 0.70, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[6]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[6]), 1.78, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[6]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[6]), 2.05, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[6]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[6]), 2.15, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[6]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[6]), 2.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[6]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[6]), 2.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[6]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[6]), 2.90, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[6]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[6]), 3.00, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[6]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[6]), 3.25, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[6]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[6]), 7.05, 1e-5);
// Neighbors of point 7.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[7]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[7]), 0.70, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[7]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[7]), 1.08, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[7]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[7]), 1.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[7]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[7]), 1.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[7]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[7]), 1.65, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[7]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[7]), 1.75, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[7]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[7]), 2.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[7]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[7]), 2.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[7]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[7]), 2.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[7]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[7]), 6.35, 1e-5);
// Neighbors of point 8.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[8]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[8]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[8]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[8]), 0.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[8]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[8]), 0.40, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[8]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[8]), 0.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[8]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[8]), 0.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[8]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[8]), 0.67, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[8]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[8]), 0.80, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[8]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[8]), 1.75, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[8]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[8]), 2.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[8]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[8]), 4.60, 1e-5);
// Neighbors of point 9.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[9]), newFromOld[10]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[9]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[9]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[9]), 0.35, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[9]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[9]), 0.45, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[9]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[9]), 0.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[9]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[9]), 0.75, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[9]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[9]), 0.85, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[9]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[9]), 1.12, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[9]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[9]), 2.20, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[9]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[9]), 2.90, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[9]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[9]), 4.15, 1e-5);
// Neighbors of point 10.
BOOST_REQUIRE_EQUAL(neighbors(0, newFromOld[10]), newFromOld[9]);
BOOST_REQUIRE_CLOSE(distances(0, newFromOld[10]), 0.10, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(1, newFromOld[10]), newFromOld[3]);
BOOST_REQUIRE_CLOSE(distances(1, newFromOld[10]), 0.25, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(2, newFromOld[10]), newFromOld[8]);
BOOST_REQUIRE_CLOSE(distances(2, newFromOld[10]), 0.55, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(3, newFromOld[10]), newFromOld[1]);
BOOST_REQUIRE_CLOSE(distances(3, newFromOld[10]), 0.65, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(4, newFromOld[10]), newFromOld[2]);
BOOST_REQUIRE_CLOSE(distances(4, newFromOld[10]), 0.85, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(5, newFromOld[10]), newFromOld[0]);
BOOST_REQUIRE_CLOSE(distances(5, newFromOld[10]), 0.95, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(6, newFromOld[10]), newFromOld[5]);
BOOST_REQUIRE_CLOSE(distances(6, newFromOld[10]), 1.22, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(7, newFromOld[10]), newFromOld[7]);
BOOST_REQUIRE_CLOSE(distances(7, newFromOld[10]), 2.30, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(8, newFromOld[10]), newFromOld[6]);
BOOST_REQUIRE_CLOSE(distances(8, newFromOld[10]), 3.00, 1e-5);
BOOST_REQUIRE_EQUAL(neighbors(9, newFromOld[10]), newFromOld[4]);
BOOST_REQUIRE_CLOSE(distances(9, newFromOld[10]), 4.05, 1e-5);
// Clean the memory.
delete knn;
}
// Delete the tree.
delete tree;
}
/**
* Test the dual-tree nearest-neighbors method with the naive method. This
* uses both a query and reference dataset.
*
* Errors are produced if the results are not identical.
*/
BOOST_AUTO_TEST_CASE(DualTreeVsNaive1)
{
arma::mat dataset;
// Hard-coded filename: bad?
if (!data::Load("test_data_3_1000.csv", dataset))
BOOST_FAIL("Cannot load test dataset test_data_3_1000.csv!");
KNN knn(dataset);
KNN naive(dataset, true);
arma::Mat<size_t> neighborsTree;
arma::mat distancesTree;
knn.Search(dataset, 15, neighborsTree, distancesTree);
arma::Mat<size_t> neighborsNaive;
arma::mat distancesNaive;
naive.Search(dataset, 15, neighborsNaive, distancesNaive);
for (size_t i = 0; i < neighborsTree.n_elem; i++)
{
BOOST_REQUIRE_EQUAL(neighborsTree(i), neighborsNaive(i));
BOOST_REQUIRE_CLOSE(distancesTree(i), distancesNaive(i), 1e-5);
}
}
/**
* Test the dual-tree nearest-neighbors method with the naive method. This uses
* only a reference dataset.
*
* Errors are produced if the results are not identical.
*/
BOOST_AUTO_TEST_CASE(DualTreeVsNaive2)
{
arma::mat dataset;
// Hard-coded filename: bad?
// Code duplication: also bad!
if (!data::Load("test_data_3_1000.csv", dataset))
BOOST_FAIL("Cannot load test dataset test_data_3_1000.csv!");
KNN knn(dataset);
// Set naive mode.
KNN naive(dataset, true);
arma::Mat<size_t> neighborsTree;
arma::mat distancesTree;
knn.Search(15, neighborsTree, distancesTree);
arma::Mat<size_t> neighborsNaive;
arma::mat distancesNaive;
naive.Search(15, neighborsNaive, distancesNaive);
for (size_t i = 0; i < neighborsTree.n_elem; i++)
{
BOOST_REQUIRE_EQUAL(neighborsTree[i], neighborsNaive[i]);
BOOST_REQUIRE_CLOSE(distancesTree[i], distancesNaive[i], 1e-5);
}
}
/**
* Test the single-tree nearest-neighbors method with the naive method. This
* uses only a reference dataset.
*
* Errors are produced if the results are not identical.
*/
BOOST_AUTO_TEST_CASE(SingleTreeVsNaive)
{
arma::mat dataset;
// Hard-coded filename: bad?
// Code duplication: also bad!
if (!data::Load("test_data_3_1000.csv", dataset))
BOOST_FAIL("Cannot load test dataset test_data_3_1000.csv!");
KNN knn(dataset, false, true);
// Set up computation for naive mode.
KNN naive(dataset, true);
arma::Mat<size_t> neighborsTree;
arma::mat distancesTree;
knn.Search(15, neighborsTree, distancesTree);
arma::Mat<size_t> neighborsNaive;
arma::mat distancesNaive;
naive.Search(15, neighborsNaive, distancesNaive);
for (size_t i = 0; i < neighborsTree.n_elem; i++)
{
BOOST_REQUIRE_EQUAL(neighborsTree[i], neighborsNaive[i]);
BOOST_REQUIRE_CLOSE(distancesTree[i], distancesNaive[i], 1e-5);
}
}
/**
* Test the cover tree single-tree nearest-neighbors method against the naive
* method. This uses only a random reference dataset.
*
* Errors are produced if the results are not identical.
*/
BOOST_AUTO_TEST_CASE(SingleCoverTreeTest)
{
arma::mat data;
data.randu(75, 1000); // 75 dimensional, 1000 points.
StandardCoverTree<EuclideanDistance, NeighborSearchStat<NearestNeighborSort>,
arma::mat> tree(data);
NeighborSearch<NearestNeighborSort, LMetric<2>, arma::mat, StandardCoverTree>
coverTreeSearch(&tree, true);
KNN naive(data, true);
arma::Mat<size_t> coverTreeNeighbors;
arma::mat coverTreeDistances;
coverTreeSearch.Search(15, coverTreeNeighbors, coverTreeDistances);
arma::Mat<size_t> naiveNeighbors;
arma::mat naiveDistances;
naive.Search(15, naiveNeighbors, naiveDistances);
for (size_t i = 0; i < coverTreeNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(coverTreeNeighbors[i], naiveNeighbors[i]);
BOOST_REQUIRE_CLOSE(coverTreeDistances[i], naiveDistances[i], 1e-5);
}
}
/**
* Test the cover tree dual-tree nearest neighbors method against the naive
* method.
*/
BOOST_AUTO_TEST_CASE(DualCoverTreeTest)
{
arma::mat dataset;
data::Load("test_data_3_1000.csv", dataset);
KNN tree(dataset);
arma::Mat<size_t> kdNeighbors;
arma::mat kdDistances;
tree.Search(dataset, 5, kdNeighbors, kdDistances);
StandardCoverTree<EuclideanDistance, NeighborSearchStat<NearestNeighborSort>,
arma::mat> referenceTree(dataset);
NeighborSearch<NearestNeighborSort, EuclideanDistance, arma::mat,
StandardCoverTree> coverTreeSearch(&referenceTree);
arma::Mat<size_t> coverNeighbors;
arma::mat coverDistances;
coverTreeSearch.Search(&referenceTree, 5, coverNeighbors, coverDistances);
for (size_t i = 0; i < coverNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(coverNeighbors(i), kdNeighbors(i));
BOOST_REQUIRE_CLOSE(coverDistances(i), kdDistances(i), 1e-5);
}
}
/**
* Test the ball tree single-tree nearest-neighbors method against the naive
* method. This uses only a random reference dataset.
*
* Errors are produced if the results are not identical.
*/
BOOST_AUTO_TEST_CASE(SingleBallTreeTest)
{
arma::mat data;
data.randu(50, 300); // 50 dimensional, 300 points.
typedef BallTree<EuclideanDistance, NeighborSearchStat<NearestNeighborSort>,
arma::mat> TreeType;
TreeType tree(data);
// BinarySpaceTree modifies data. Use modified data to maintain the
// correspondance between points in the dataset for both methods. The order of
// query points in both methods should be same.
NeighborSearch<NearestNeighborSort, EuclideanDistance, arma::mat, BallTree>
ballTreeSearch(&tree, true);
KNN naive(tree.Dataset(), true);
arma::Mat<size_t> ballTreeNeighbors;
arma::mat ballTreeDistances;
ballTreeSearch.Search(2, ballTreeNeighbors, ballTreeDistances);
arma::Mat<size_t> naiveNeighbors;
arma::mat naiveDistances;
naive.Search(2, naiveNeighbors, naiveDistances);
for (size_t i = 0; i < ballTreeNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(ballTreeNeighbors[i], naiveNeighbors[i]);
BOOST_REQUIRE_CLOSE(ballTreeDistances[i], naiveDistances[i], 1e-5);
}
}
/**
* Test the ball tree dual-tree nearest neighbors method against the naive
* method.
*/
BOOST_AUTO_TEST_CASE(DualBallTreeTest)
{
arma::mat dataset;
data::Load("test_data_3_1000.csv", dataset);
KNN tree(dataset);
arma::Mat<size_t> kdNeighbors;
arma::mat kdDistances;
tree.Search(5, kdNeighbors, kdDistances);
NeighborSearch<NearestNeighborSort, EuclideanDistance, arma::mat, BallTree>
ballTreeSearch(dataset);
arma::Mat<size_t> ballNeighbors;
arma::mat ballDistances;
ballTreeSearch.Search(5, ballNeighbors, ballDistances);
for (size_t i = 0; i < ballNeighbors.n_elem; ++i)
{
BOOST_REQUIRE_EQUAL(ballNeighbors(i), kdNeighbors(i));
BOOST_REQUIRE_CLOSE(ballDistances(i), kdDistances(i), 1e-5);
}
}
/**
* Make sure sparse nearest neighbors works with kd trees.
*/
BOOST_AUTO_TEST_CASE(SparseKNNKDTreeTest)
{
// The dimensionality of these datasets must be high so that the probability
// of a completely empty point is very low. In this case, with dimensionality
// 70, the probability of all 70 dimensions being zero is 0.8^70 = 1.65e-7 in
// the reference set and 0.9^70 = 6.27e-4 in the query set.
arma::sp_mat queryDataset;
queryDataset.sprandu(70, 200, 0.2);
arma::sp_mat referenceDataset;
referenceDataset.sprandu(70, 500, 0.1);
arma::mat denseQuery(queryDataset);
arma::mat denseReference(referenceDataset);
typedef NeighborSearch<NearestNeighborSort, EuclideanDistance, arma::sp_mat,
KDTree> SparseKNN;
SparseKNN a(referenceDataset);
KNN naive(denseReference, true);
arma::mat sparseDistances;
arma::Mat<size_t> sparseNeighbors;
a.Search(queryDataset, 10, sparseNeighbors, sparseDistances);
arma::mat naiveDistances;
arma::Mat<size_t> naiveNeighbors;
naive.Search(denseQuery, 10, naiveNeighbors, naiveDistances);
for (size_t i = 0; i < naiveNeighbors.n_cols; ++i)
{
for (size_t j = 0; j < naiveNeighbors.n_rows; ++j)
{
BOOST_REQUIRE_EQUAL(naiveNeighbors(j, i), sparseNeighbors(j, i));
BOOST_REQUIRE_CLOSE(naiveDistances(j, i), sparseDistances(j, i), 1e-5);
}
}
}
/*
BOOST_AUTO_TEST_CASE(SparseKNNCoverTreeTest)
{
typedef CoverTree<LMetric<2, true>, FirstPointIsRoot,
NeighborSearchStat<NearestNeighborSort>, arma::sp_mat> SparseCoverTree;
// The dimensionality of these datasets must be high so that the probability
// of a completely empty point is very low. In this case, with dimensionality
// 70, the probability of all 70 dimensions being zero is 0.8^70 = 1.65e-7 in
// the reference set and 0.9^70 = 6.27e-4 in the query set.
arma::sp_mat queryDataset;
queryDataset.sprandu(50, 5000, 0.2);
arma::sp_mat referenceDataset;
referenceDataset.sprandu(50, 8000, 0.1);
arma::mat denseQuery(queryDataset);
arma::mat denseReference(referenceDataset);
typedef NeighborSearch<NearestNeighborSort, EuclideanDistance,
SparseCoverTree> SparseKNN;
arma::mat sparseDistances;
arma::Mat<size_t> sparseNeighbors;
a.Search(10, sparseNeighbors, sparseDistances);
arma::mat naiveDistances;
arma::Mat<size_t> naiveNeighbors;
naive.Search(10, naiveNeighbors, naiveDistances);
for (size_t i = 0; i < naiveNeighbors.n_cols; ++i)
{
for (size_t j = 0; j < naiveNeighbors.n_rows; ++j)
{
BOOST_REQUIRE_EQUAL(naiveNeighbors(j, i), sparseNeighbors(j, i));
BOOST_REQUIRE_CLOSE(naiveDistances(j, i), sparseDistances(j, i), 1e-5);
}
}
}
*/
BOOST_AUTO_TEST_CASE(KNNModelTest)
{
// Ensure that we can build an NSModel<NearestNeighborSearch> and get correct
// results.
typedef NSModel<NearestNeighborSort> KNNModel;
arma::mat queryData = arma::randu<arma::mat>(10, 50);
arma::mat referenceData = arma::randu<arma::mat>(10, 200);
// Build all the possible models.
KNNModel models[14];
models[0] = KNNModel(KNNModel::TreeTypes::KD_TREE, true);
models[1] = KNNModel(KNNModel::TreeTypes::KD_TREE, false);
models[2] = KNNModel(KNNModel::TreeTypes::COVER_TREE, true);
models[3] = KNNModel(KNNModel::TreeTypes::COVER_TREE, false);
models[4] = KNNModel(KNNModel::TreeTypes::R_TREE, true);
models[5] = KNNModel(KNNModel::TreeTypes::R_TREE, false);
models[6] = KNNModel(KNNModel::TreeTypes::R_STAR_TREE, true);
models[7] = KNNModel(KNNModel::TreeTypes::R_STAR_TREE, false);
models[8] = KNNModel(KNNModel::TreeTypes::X_TREE, true);
models[9] = KNNModel(KNNModel::TreeTypes::X_TREE, false);
models[10] = KNNModel(KNNModel::TreeTypes::BALL_TREE, true);
models[11] = KNNModel(KNNModel::TreeTypes::BALL_TREE, false);
models[12] = KNNModel(KNNModel::TreeTypes::HILBERT_R_TREE, true);
models[13] = KNNModel(KNNModel::TreeTypes::HILBERT_R_TREE, false);
for (size_t j = 0; j < 2; ++j)
{
// Get a baseline.
KNN knn(referenceData);
arma::Mat<size_t> baselineNeighbors;
arma::mat baselineDistances;
knn.Search(queryData, 3, baselineNeighbors, baselineDistances);
for (size_t i = 0; i < 14; ++i)
{
// We only have std::move() constructors so make a copy of our data.
arma::mat referenceCopy(referenceData);
arma::mat queryCopy(queryData);
if (j == 0)
models[i].BuildModel(std::move(referenceCopy), 20, false, false);
if (j == 1)
models[i].BuildModel(std::move(referenceCopy), 20, false, true);
if (j == 2)
models[i].BuildModel(std::move(referenceCopy), 20, true, false);
arma::Mat<size_t> neighbors;
arma::mat distances;
models[i].Search(std::move(queryCopy), 3, neighbors, distances);
BOOST_REQUIRE_EQUAL(neighbors.n_rows, baselineNeighbors.n_rows);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, baselineNeighbors.n_cols);
BOOST_REQUIRE_EQUAL(neighbors.n_elem, baselineNeighbors.n_elem);
BOOST_REQUIRE_EQUAL(distances.n_rows, baselineDistances.n_rows);
BOOST_REQUIRE_EQUAL(distances.n_cols, baselineDistances.n_cols);
BOOST_REQUIRE_EQUAL(distances.n_elem, baselineDistances.n_elem);
for (size_t k = 0; k < distances.n_elem; ++k)
{
BOOST_REQUIRE_EQUAL(neighbors[k], baselineNeighbors[k]);
if (std::abs(baselineDistances[k]) < 1e-5)
BOOST_REQUIRE_SMALL(distances[k], 1e-5);
else
BOOST_REQUIRE_CLOSE(distances[k], baselineDistances[k], 1e-5);
}
}
}
}
BOOST_AUTO_TEST_CASE(KNNModelMonochromaticTest)
{
// Ensure that we can build an NSModel<NearestNeighborSearch> and get correct
// results, in the case where the reference set is the same as the query set.
typedef NSModel<NearestNeighborSort> KNNModel;
arma::mat referenceData = arma::randu<arma::mat>(10, 200);
// Build all the possible models.
KNNModel models[14];
models[0] = KNNModel(KNNModel::TreeTypes::KD_TREE, true);
models[1] = KNNModel(KNNModel::TreeTypes::KD_TREE, false);
models[2] = KNNModel(KNNModel::TreeTypes::COVER_TREE, true);
models[3] = KNNModel(KNNModel::TreeTypes::COVER_TREE, false);
models[4] = KNNModel(KNNModel::TreeTypes::R_TREE, true);
models[5] = KNNModel(KNNModel::TreeTypes::R_TREE, false);
models[6] = KNNModel(KNNModel::TreeTypes::R_STAR_TREE, true);
models[7] = KNNModel(KNNModel::TreeTypes::R_STAR_TREE, false);
models[8] = KNNModel(KNNModel::TreeTypes::X_TREE, true);
models[9] = KNNModel(KNNModel::TreeTypes::X_TREE, false);
models[10] = KNNModel(KNNModel::TreeTypes::BALL_TREE, true);
models[11] = KNNModel(KNNModel::TreeTypes::BALL_TREE, false);
models[12] = KNNModel(KNNModel::TreeTypes::HILBERT_R_TREE, true);
models[13] = KNNModel(KNNModel::TreeTypes::HILBERT_R_TREE, false);
for (size_t j = 0; j < 2; ++j)
{
// Get a baseline.
KNN knn(referenceData);
arma::Mat<size_t> baselineNeighbors;
arma::mat baselineDistances;
knn.Search(3, baselineNeighbors, baselineDistances);
for (size_t i = 0; i < 14; ++i)
{
// We only have a std::move() constructor... so copy the data.
arma::mat referenceCopy(referenceData);
if (j == 0)
models[i].BuildModel(std::move(referenceCopy), 20, false, false);
if (j == 1)
models[i].BuildModel(std::move(referenceCopy), 20, false, true);
if (j == 2)
models[i].BuildModel(std::move(referenceCopy), 20, true, false);
arma::Mat<size_t> neighbors;
arma::mat distances;
models[i].Search(3, neighbors, distances);
BOOST_REQUIRE_EQUAL(neighbors.n_rows, baselineNeighbors.n_rows);
BOOST_REQUIRE_EQUAL(neighbors.n_cols, baselineNeighbors.n_cols);
BOOST_REQUIRE_EQUAL(neighbors.n_elem, baselineNeighbors.n_elem);
BOOST_REQUIRE_EQUAL(distances.n_rows, baselineDistances.n_rows);
BOOST_REQUIRE_EQUAL(distances.n_cols, baselineDistances.n_cols);
BOOST_REQUIRE_EQUAL(distances.n_elem, baselineDistances.n_elem);
for (size_t k = 0; k < distances.n_elem; ++k)
{
BOOST_REQUIRE_EQUAL(neighbors[k], baselineNeighbors[k]);
if (std::abs(baselineDistances[k]) < 1e-5)
BOOST_REQUIRE_SMALL(distances[k], 1e-5);
else
BOOST_REQUIRE_CLOSE(distances[k], baselineDistances[k], 1e-5);
}
}
}
}
/**
* If we search twice with the same reference tree, the bounds need to be reset
* before the second search. This test ensures that that happens, by making
* sure the number of scores and base cases are equivalent for each search.
*/
BOOST_AUTO_TEST_CASE(DoubleReferenceSearchTest)
{
arma::mat dataset = arma::randu<arma::mat>(5, 500);
KNN knn(std::move(dataset));
arma::mat distances, secondDistances;
arma::Mat<size_t> neighbors, secondNeighbors;
knn.Search(3, neighbors, distances);
size_t baseCases = knn.BaseCases();
size_t scores = knn.Scores();
knn.Search(3, secondNeighbors, secondDistances);
BOOST_REQUIRE_EQUAL(knn.BaseCases(), baseCases);
BOOST_REQUIRE_EQUAL(knn.Scores(), scores);
}
/**
* Make sure that the neighborPtr matrix isn't accidentally deleted.
* See issue #478.
*/
BOOST_AUTO_TEST_CASE(NeighborPtrDeleteTest)
{
arma::mat dataset = arma::randu<arma::mat>(5, 100);
// Build the tree ourselves.
std::vector<size_t> oldFromNewReferences;
KNN::Tree tree(dataset);
KNN knn(&tree);
// Now make a query set.
arma::mat queryset = arma::randu<arma::mat>(5, 50);
arma::mat distances;
arma::Mat<size_t> neighbors;
knn.Search(queryset, 3, neighbors, distances);
// These will (hopefully) fail is either the neighbors or the distances matrix
// has been accidentally deleted.
BOOST_REQUIRE_EQUAL(neighbors.n_cols, 50);
BOOST_REQUIRE_EQUAL(neighbors.n_rows, 3);
BOOST_REQUIRE_EQUAL(distances.n_cols, 50);
BOOST_REQUIRE_EQUAL(distances.n_rows, 3);
}
BOOST_AUTO_TEST_SUITE_END();