apaam/mlpack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Test RANN with cover trees.

Browse Source
This commit is contained in:
Ryan Curtin
2013-10-04 04:30:31 +00:00
parent 836332d526
commit a9b291485c
1 changed files with 143 additions and 11 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/mlpack/tests/allkrann_search_test.cpp
+143 -11
View File
@@ -7,6 +7,7 @@
#include <time.h>
#include <mlpack/core.hpp>
#include <mlpack/core/metrics/lmetric.hpp>
#include <mlpack/core/tree/cover_tree.hpp>
#include <boost/test/unit_test.hpp>
#include "old_boost_test_definitions.hpp"
@@ -22,7 +23,7 @@ BOOST_AUTO_TEST_SUITE(AllkRANNTest);
// Test AllkRANN in naive mode for exact results when the random seeds are set
// the same. This may not be the best test; if the implementation of RANN-RS
// gets random numbers in a different way, then this test might fail.
BOOST_AUTO_TEST_CASE(AllkRANNNaiveSearchExact)
BOOST_AUTO_TEST_CASE(NaiveSearchExact)
{
// First test on a small set.
arma::mat rdata(2, 10);
@@ -86,13 +87,13 @@ BOOST_AUTO_TEST_CASE(AllkRANNNaiveSearchExact)
// 3. Check the neighbor returned.
for (size_t i = 0; i < qdata.n_cols; i++)
{
BOOST_REQUIRE(neighbors(0, i) == rann[i]);
BOOST_REQUIRE_EQUAL(neighbors(0, i), rann[i]);
BOOST_REQUIRE_CLOSE(distances(0, i), rannDistances[i], 1e-5);
}
}
// Test the correctness and guarantees of AllkRANN when in naive mode.
BOOST_AUTO_TEST_CASE(AllkRANNNaiveGuaranteeTest)
BOOST_AUTO_TEST_CASE(NaiveGuaranteeTest)
{
arma::Mat<size_t> neighbors;
arma::mat distances;
@@ -143,12 +144,12 @@ BOOST_AUTO_TEST_CASE(AllkRANNNaiveGuaranteeTest)
// 5% of 100 queries is 5.
size_t maxNumQueriesFail = 6;
BOOST_REQUIRE(numQueriesFail < maxNumQueriesFail);
BOOST_REQUIRE_LT(numQueriesFail, maxNumQueriesFail);
}
// Test single-tree rank-approximate search (harder to test because of
// the randomness involved).
BOOST_AUTO_TEST_CASE(AllkRANNSingleTreeSearch)
BOOST_AUTO_TEST_CASE(SingleTreeSearch)
{
arma::mat refData;
arma::mat queryData;
@@ -202,12 +203,12 @@ BOOST_AUTO_TEST_CASE(AllkRANNSingleTreeSearch)
// 5% of 100 queries is 5.
size_t maxNumQueriesFail = 6;
BOOST_REQUIRE(numQueriesFail < maxNumQueriesFail);
BOOST_REQUIRE_LT(numQueriesFail, maxNumQueriesFail);
}
// Test dual-tree rank-approximate search (harder to test because of the
// randomness involved).
BOOST_AUTO_TEST_CASE(AllkRANNDualTreeSearch)
BOOST_AUTO_TEST_CASE(DualTreeSearch)
{
arma::mat refData;
arma::mat queryData;
@@ -262,12 +263,12 @@ BOOST_AUTO_TEST_CASE(AllkRANNDualTreeSearch)
// 5% of 100 queries is 5.
size_t maxNumQueriesFail = 6;
BOOST_REQUIRE(numQueriesFail < maxNumQueriesFail);
BOOST_REQUIRE_LT(numQueriesFail, maxNumQueriesFail);
}
// Test rank-approximate search with just a single dataset. These tests just
// ensure that the method runs okay.
BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetNaiveSearch)
BOOST_AUTO_TEST_CASE(SingleDatasetNaiveSearch)
{
arma::mat dataset(5, 2500);
dataset.randn();
@@ -287,7 +288,7 @@ BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetNaiveSearch)
// Test rank-approximate search with just a single dataset in single-tree mode.
// These tests just ensure that the method runs okay.
BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetSingleSearch)
BOOST_AUTO_TEST_CASE(SingleDatasetSingleSearch)
{
arma::mat dataset(5, 2500);
dataset.randn();
@@ -307,7 +308,7 @@ BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetSingleSearch)
// Test rank-approximate search with just a single dataset in dual-tree mode.
// These tests just ensure that the method runs okay.
BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetSearch)
BOOST_AUTO_TEST_CASE(SingleDatasetSearch)
{
arma::mat dataset(5, 2500);
dataset.randn();
@@ -324,4 +325,135 @@ BOOST_AUTO_TEST_CASE(AllkRANNSingleDatasetSearch)
BOOST_REQUIRE_EQUAL(distances.n_cols, 2500);
}
// Test single-tree rank-approximate search with cover trees.
BOOST_AUTO_TEST_CASE(SingleCoverTreeTest)
{
arma::mat refData;
arma::mat queryData;
data::Load("rann_test_r_3_900.csv", refData, true);
data::Load("rann_test_q_3_100.csv", queryData, true);
// Search for 1 rank-approximate nearest-neighbors in the top 30% of the point
// (rank error of 3).
arma::Mat<size_t> neighbors;
arma::mat distances;
typedef tree::CoverTree<metric::EuclideanDistance, tree::FirstPointIsRoot,
RAQueryStat<NearestNeighborSort> > TreeType;
typedef RASearch<NearestNeighborSort, metric::EuclideanDistance, TreeType>
RACoverTreeSearch;
TreeType refTree(refData);
RACoverTreeSearch tssRann(&refTree, NULL, refData, queryData, true);
// The relative ranks for the given query reference pair.
arma::Mat<size_t> qrRanks;
data::Load("rann_test_qr_ranks.csv", qrRanks, true, false); // No transpose.
size_t numRounds = 1000;
arma::Col<size_t> numSuccessRounds(queryData.n_cols);
numSuccessRounds.fill(0);
// 1% of 900 is 9, so the rank is expected to be less than 10.
size_t expectedRankErrorUB = 10;
for (size_t rounds = 0; rounds < numRounds; rounds++)
{
tssRann.Search(1, neighbors, distances, 1.0, 0.95, false, false, 5);
for (size_t i = 0; i < queryData.n_cols; i++)
if (qrRanks(i, neighbors(0, i)) < expectedRankErrorUB)
numSuccessRounds[i]++;
neighbors.reset();
distances.reset();
}
// Find the 95%-tile threshold so that 95% of the queries should pass this
// threshold.
size_t threshold = floor(numRounds *
(0.95 - (1.96 * sqrt(0.95 * 0.05 / numRounds))));
size_t numQueriesFail = 0;
for (size_t i = 0; i < queryData.n_cols; i++)
if (numSuccessRounds[i] < threshold)
numQueriesFail++;
Log::Warn << "RANN-TSS (cover tree): RANN guarantee fails on "
<< numQueriesFail << " queries." << endl;
// Assert that at most 5% of the queries fall out of this threshold.
// 5% of 100 queries is 5.
size_t maxNumQueriesFail = 6;
BOOST_REQUIRE_LT(numQueriesFail, maxNumQueriesFail);
}
// Test dual-tree rank-approximate search with cover trees.
BOOST_AUTO_TEST_CASE(DualCoverTreeTest)
{
arma::mat refData;
arma::mat queryData;
data::Load("rann_test_r_3_900.csv", refData, true);
data::Load("rann_test_q_3_100.csv", queryData, true);
// Search for 1 rank-approximate nearest-neighbors in the top 30% of the point
// (rank error of 3).
arma::Mat<size_t> neighbors;
arma::mat distances;
typedef tree::CoverTree<metric::EuclideanDistance, tree::FirstPointIsRoot,
RAQueryStat<NearestNeighborSort> > TreeType;
typedef RASearch<NearestNeighborSort, metric::EuclideanDistance, TreeType>
RACoverTreeSearch;
TreeType refTree(refData);
TreeType queryTree(queryData);
RACoverTreeSearch tsdRann(&refTree, &queryTree, refData, queryData, false);
arma::Mat<size_t> qrRanks;
data::Load("rann_test_qr_ranks.csv", qrRanks, true, false); // No transpose.
size_t numRounds = 1000;
arma::Col<size_t> numSuccessRounds(queryData.n_cols);
numSuccessRounds.fill(0);
// 1% of 900 is 9, so the rank is expected to be less than 10.
size_t expectedRankErrorUB = 10;
for (size_t rounds = 0; rounds < numRounds; rounds++)
{
tsdRann.Search(1, neighbors, distances, 1.0, 0.95, false, false, 5);
for (size_t i = 0; i < queryData.n_cols; i++)
if (qrRanks(i, neighbors(0, i)) < expectedRankErrorUB)
numSuccessRounds[i]++;
neighbors.reset();
distances.reset();
tsdRann.ResetQueryTree();
}
// Find the 95%-tile threshold so that 95% of the queries should pass this
// threshold.
size_t threshold = floor(numRounds *
(0.95 - (1.96 * sqrt(0.95 * 0.05 / numRounds))));
size_t numQueriesFail = 0;
for (size_t i = 0; i < queryData.n_cols; i++)
if (numSuccessRounds[i] < threshold)
numQueriesFail++;
Log::Warn << "RANN-TSD (cover tree): RANN guarantee fails on "
<< numQueriesFail << " queries." << endl;
// assert that at most 5% of the queries fall out of this threshold
// 5% of 100 queries is 5.
size_t maxNumQueriesFail = 6;
BOOST_REQUIRE_LT(numQueriesFail, maxNumQueriesFail);
}
BOOST_AUTO_TEST_SUITE_END();