[mlpack-git] master: Refactor allkfn program. (00eccfd)
gitdub at big.cc.gt.atl.ga.us
gitdub at big.cc.gt.atl.ga.us
Tue Oct 20 09:48:08 EDT 2015
Repository : https://github.com/mlpack/mlpack
On branch : master
Link : https://github.com/mlpack/mlpack/compare/67e0a132c7f62820c734eb508fe1bc83128a3e13...00eccfdb0d315de3d94bfa1da84cc1dc65c8af39
>---------------------------------------------------------------
commit 00eccfdb0d315de3d94bfa1da84cc1dc65c8af39
Author: Ryan Curtin <ryan at ratml.org>
Date: Tue Oct 20 13:41:35 2015 +0000
Refactor allkfn program.
>---------------------------------------------------------------
00eccfdb0d315de3d94bfa1da84cc1dc65c8af39
src/mlpack/methods/neighbor_search/allkfn_main.cpp | 317 +++++++++++----------
1 file changed, 162 insertions(+), 155 deletions(-)
diff --git a/src/mlpack/methods/neighbor_search/allkfn_main.cpp b/src/mlpack/methods/neighbor_search/allkfn_main.cpp
index 784faef..5bf2982 100644
--- a/src/mlpack/methods/neighbor_search/allkfn_main.cpp
+++ b/src/mlpack/methods/neighbor_search/allkfn_main.cpp
@@ -13,6 +13,7 @@
#include "neighbor_search.hpp"
#include "unmap.hpp"
+#include "ns_model.hpp"
using namespace std;
using namespace mlpack;
@@ -41,199 +42,205 @@ PROGRAM_INFO("All K-Furthest-Neighbors",
"corresponds to the distance between those two points.");
// Define our input parameters that this program will take.
-PARAM_STRING_REQ("reference_file", "File containing the reference dataset.",
- "r");
-PARAM_INT_REQ("k", "Number of furthest neighbors to find.", "k");
-PARAM_STRING_REQ("distances_file", "File to output distances into.", "d");
-PARAM_STRING_REQ("neighbors_file", "File to output neighbors into.", "n");
-
+PARAM_STRING("reference_file", "File containing the reference dataset.", "r",
+ "");
+PARAM_STRING("distances_file", "File to output distances into.", "d", "");
+PARAM_STRING("neighbors_file", "File to output neighbors into.", "n", "");
+
+// The option exists to load or save models.
+PARAM_STRING("input_model_file", "File containing pre-trained kFN model.", "m",
+ "");
+PARAM_STRING("output_model_file", "If specified, the kFN model will be saved to"
+ " the given file.", "M", "");
+
+// The user may specify a query file of query points and a number of furthest
+// neighbors to search for.
PARAM_STRING("query_file", "File containing query points (optional).", "q", "");
+PARAM_INT("k", "Number of furthest neighbors to find.", "k", 0);
+// The user may specify the type of tree to use, and a few pararmeters for tree
+// building.
+PARAM_STRING("tree_type", "Type of tree to use: 'kd', 'cover', 'r', 'r-star', "
+ "'ball'.", "t", "kd");
PARAM_INT("leaf_size", "Leaf size for tree building.", "l", 20);
+PARAM_FLAG("random_basis", "Before tree-building, project the data onto a "
+ "random orthogonal basis.", "R");
+PARAM_INT("seed", "Random seed (if 0, std::time(NULL) is used).", "s", 0);
+
+// Search settings.
PARAM_FLAG("naive", "If true, O(n^2) naive mode is used for computation.", "N");
PARAM_FLAG("single_mode", "If true, single-tree search is used (as opposed to "
"dual-tree search).", "s");
-PARAM_FLAG("r_tree", "If true, use an R-Tree to perform the search "
- "(experimental, may be slow.).", "T");
+
+// Convenience typedef.
+typedef NSModel<FurthestNeighborSort> KFNModel;
int main(int argc, char *argv[])
{
// Give CLI the command line parameters the user passed in.
CLI::ParseCommandLine(argc, argv);
- // Get all the parameters.
- string referenceFile = CLI::GetParam<string>("reference_file");
-
- string distancesFile = CLI::GetParam<string>("distances_file");
- string neighborsFile = CLI::GetParam<string>("neighbors_file");
-
- int lsInt = CLI::GetParam<int>("leaf_size");
-
- size_t k = CLI::GetParam<int>("k");
-
- bool naive = CLI::HasParam("naive");
- bool singleMode = CLI::HasParam("single_mode");
+ if (CLI::GetParam<int>("seed") != 0)
+ math::RandomSeed((size_t) CLI::GetParam<int>("seed"));
+ else
+ math::RandomSeed((size_t) std::time(NULL));
- arma::mat referenceData;
- arma::mat queryData; // So it doesn't go out of scope.
- data::Load(referenceFile, referenceData, true);
+ // A user cannot specify both reference data and a model.
+ if (CLI::HasParam("reference_file") && CLI::HasParam("input_model_file"))
+ Log::Fatal << "Only one of --reference_file (-r) or --input_model_file (-m)"
+ << " may be specified!" << endl;
- Log::Info << "Loaded reference data from '" << referenceFile << "' ("
- << referenceData.n_rows << " x " << referenceData.n_cols << ")." << endl;
+ // A user must specify one of them...
+ if (!CLI::HasParam("reference_file") && !CLI::HasParam("input_model_file"))
+ Log::Fatal << "No model specified (--input_model_file) and no reference "
+ << "data specified (--reference_file)! One must be provided." << endl;
- // Sanity check on k value: must be greater than 0, must be less than the
- // number of reference points.
- if (k > referenceData.n_cols)
+ if (CLI::HasParam("input_model_file"))
{
- Log::Fatal << "Invalid k: " << k << "; must be greater than 0 and less ";
- Log::Fatal << "than or equal to the number of reference points (";
- Log::Fatal << referenceData.n_cols << ")." << endl;
+ // Notify the user of parameters that will be ignored.
+ if (CLI::HasParam("tree_type"))
+ Log::Warn << "--tree_type (-t) will be ignored because --input_model_file"
+ << " is specified." << endl;
+ if (CLI::HasParam("leaf_size"))
+ Log::Warn << "--leaf_size (-l) will be ignored because --input_model_file"
+ << " is specified." << endl;
+ if (CLI::HasParam("random_basis"))
+ Log::Warn << "--random_basis (-R) will be ignored because "
+ << "--input_model_file is specified." << endl;
+ if (CLI::HasParam("naive"))
+ Log::Warn << "--naive (-N) will be ignored because --input_model_file is "
+ << "specified." << endl;
}
- if (CLI::GetParam<string>("query_file") != "")
- {
- string queryFile = CLI::GetParam<string>("query_file");
- data::Load(queryFile, queryData, true);
- }
+ // The user should give something to do...
+ if (!CLI::HasParam("k") && !CLI::HasParam("output_model_file"))
+ Log::Warn << "Neither -k nor --output_model_file are specified, so no "
+ << "results from this program will be saved!" << endl;
- // Sanity check on leaf size.
- if (lsInt < 0)
- {
- Log::Fatal << "Invalid leaf size: " << lsInt << ". Must be greater "
- "than or equal to 0." << endl;
- }
- size_t leafSize = lsInt;
+ // If the user specifies k but no output files, they should be warned.
+ if (CLI::HasParam("k") &&
+ !(CLI::HasParam("neighbors_file") || CLI::HasParam("distances_file")))
+ Log::Warn << "Neither --neighbors_file nor --distances_file is specified, "
+ << "so the furthest neighbor search results will not be saved!" << endl;
- // Naive mode overrides single mode.
- if (singleMode && naive)
- {
- Log::Warn << "--single_mode ignored because --naive is present." << endl;
- }
+ // If the user specifies output files but no k, they should be warned.
+ if ((CLI::HasParam("neighbors_file") || CLI::HasParam("distances_file")) &&
+ !CLI::HasParam("k"))
+ Log::Warn << "An output file for furthest neighbor search is given ("
+ << "--neighbors_file or --distances_file), but furthest neighbor search"
+ << " is not being performed because k (--k) is not specified!" << endl;
- arma::Mat<size_t> neighbors;
- arma::mat distances;
-
- if (naive)
+ // Sanity check on leaf size.
+ const int lsInt = CLI::GetParam<int>("leaf_size");
+ if (lsInt < 1)
+ Log::Fatal << "Invalid leaf size: " << lsInt << ". Must be greater than 0."
+ << endl;
+
+ // We either have to load the reference data, or we have to load the model.
+ NSModel<FurthestNeighborSort> kfn;
+ const bool naive = CLI::HasParam("naive");
+ const bool singleMode = CLI::HasParam("single_mode");
+ if (CLI::HasParam("reference_file"))
{
- AllkFN allkfn(referenceData, false, naive);
-
- if (CLI::HasParam("query_file"))
- allkfn.Search(queryData, k, neighbors, distances);
+ // Get all the parameters.
+ const string referenceFile = CLI::GetParam<string>("reference_file");
+ const string treeType = CLI::GetParam<string>("tree_type");
+ const bool randomBasis = CLI::HasParam("random_basis");
+
+ int tree = 0;
+ if (treeType == "kd")
+ tree = KFNModel::KD_TREE;
+ else if (treeType == "cover")
+ tree = KFNModel::COVER_TREE;
+ else if (treeType == "r")
+ tree = KFNModel::R_TREE;
+ else if (treeType == "r-star")
+ tree = KFNModel::R_STAR_TREE;
+ else if (treeType == "ball")
+ tree = KFNModel::BALL_TREE;
else
- allkfn.Search(k, neighbors, distances);
- }
- if (!CLI::HasParam("r_tree"))
- {
- // Use default kd-tree.
- std::vector<size_t> oldFromNewRefs;
+ Log::Fatal << "Unknown tree type '" << treeType << "'; valid choices are "
+ << "'kd', 'cover', 'r', 'r-star', and 'ball'." << endl;
- typedef KDTree<EuclideanDistance, NeighborSearchStat<FurthestNeighborSort>,
- arma::mat> TreeType;
+ kfn.TreeType() = tree;
+ kfn.RandomBasis() = randomBasis;
- // Build trees by hand, so we can save memory: if we pass a tree to
- // NeighborSearch, it does not copy the matrix.
- Log::Info << "Building reference tree..." << endl;
- Timer::Start("reference_tree_building");
- TreeType refTree(referenceData, oldFromNewRefs, leafSize);
- Timer::Stop("reference_tree_building");
+ arma::mat referenceSet;
+ data::Load(referenceFile, referenceSet, true);
- std::vector<size_t> oldFromNewQueries;
+ Log::Info << "Loaded reference data from '" << referenceFile << "' ("
+ << referenceSet.n_rows << " x " << referenceSet.n_cols << ")." << endl;
- AllkFN allkfn(&refTree, singleMode);
+ const size_t leafSize = (size_t) lsInt;
+ kfn.BuildModel(std::move(referenceSet), leafSize, naive, singleMode);
+ }
+ else
+ {
+ // Load the model from file.
+ const string inputModelFile = CLI::GetParam<string>("input_model_file");
+ data::Load(inputModelFile, "kfn_model", kfn, true); // Fatal on failure.
+
+ Log::Info << "Loaded kFN model from '" << inputModelFile << "' (trained on "
+ << kfn.Dataset().n_rows << "x" << kfn.Dataset().n_cols << " dataset)."
+ << endl;
+
+ // Adjust singleMode and naive if necessary.
+ if (CLI::HasParam("single_mode"))
+ kfn.SingleMode() = true;
+ if (CLI::HasParam("naive"))
+ kfn.Naive() = true;
+ if (CLI::HasParam("leaf_size"))
+ kfn.LeafSize() = (size_t) lsInt;
+ }
- arma::mat distancesOut(distances.n_rows, distances.n_cols);
- arma::Mat<size_t> neighborsOut(neighbors.n_rows, neighbors.n_cols);
+ // Perform search, if desired.
+ if (CLI::HasParam("k"))
+ {
+ const string queryFile = CLI::GetParam<string>("query_file");
+ const size_t k = (size_t) CLI::GetParam<int>("k");
- if (CLI::HasParam("query_file"))
+ arma::mat queryData;
+ if (queryFile != "")
{
- if (!singleMode)
- {
- // Build trees by hand, so we can save memory: if we pass a tree to
- // NeighborSearch, it does not copy the matrix.
- Log::Info << "Building query tree..." << endl;
- Timer::Start("tree_building");
- TreeType queryTree(queryData, oldFromNewQueries, leafSize);
- Timer::Stop("tree_building");
- Log::Info << "Tree built." << endl;
-
- Log::Info << "Computing " << k << " furthest neighbors..." << endl;
- allkfn.Search(&queryTree, k, neighborsOut, distancesOut);
- }
- else
- {
- Log::Info << "Computing " << k << " furthest neighbors..." << endl;
- allkfn.Search(queryData, k, neighborsOut, distancesOut);
- }
+ data::Load(queryFile, queryData, true);
+ Log::Info << "Loaded query data from '" << queryFile << "' ("
+ << queryData.n_rows << " x " << queryData.n_cols << ")." << endl;
}
- else
+
+ // Sanity check on k value: must be greater than 0, must be less than the
+ // number of reference points. Since it is unsigned, we only test the upper
+ // bound.
+ if (k > kfn.Dataset().n_cols)
{
- Log::Info << "Computing " << k << " furthest neighbors..." << endl;
- allkfn.Search(k, neighborsOut, distancesOut);
+ Log::Fatal << "Invalid k: " << k << "; must be greater than 0 and less "
+ << "than or equal to the number of reference points ("
+ << kfn.Dataset().n_cols << ")." << endl;
}
- Log::Info << "Neighbors computed." << endl;
+ // Naive mode overrides single mode.
+ if (singleMode && naive)
+ Log::Warn << "--single_mode ignored because --naive is present." << endl;
- // We have to map back to the original indices from before the tree
- // construction.
- Log::Info << "Re-mapping indices..." << endl;
+ // Now run the search.
+ arma::Mat<size_t> neighbors;
+ arma::mat distances;
- // Map the points back to their original locations.
- if ((CLI::GetParam<string>("query_file") != "") && !singleMode)
- Unmap(neighborsOut, distancesOut, oldFromNewRefs, oldFromNewQueries, neighbors,
- distances);
- else if ((CLI::GetParam<string>("query_file") != "") && singleMode)
- Unmap(neighborsOut, distancesOut, oldFromNewRefs, neighbors, distances);
+ if (CLI::HasParam("query_file"))
+ kfn.Search(std::move(queryData), k, neighbors, distances);
else
- Unmap(neighborsOut, distancesOut, oldFromNewRefs, oldFromNewRefs, neighbors,
- distances);
+ kfn.Search(k, neighbors, distances);
+ Log::Info << "Search complete." << endl;
+
+ // Save output, if desired.
+ if (CLI::HasParam("neighbors_file"))
+ data::Save(CLI::GetParam<string>("neighbors_file"), neighbors);
+ if (CLI::HasParam("distances_file"))
+ data::Save(CLI::GetParam<string>("distances_file"), distances);
}
- else
+
+ if (CLI::HasParam("output_model_file"))
{
- // Use the R tree.
- Log::Info << "Using R tree for furthest-neighbor calculation." << endl;
-
- // Convenience typedef.
- typedef RStarTree<EuclideanDistance,
- NeighborSearchStat<FurthestNeighborSort>, arma::mat> TreeType;
-
- // Build trees by hand, so we can save memory: if we pass a tree to
- // NeighborSearch, it does not copy the matrix.
- Log::Info << "Building reference tree..." << endl;
- Timer::Start("tree_building");
- TreeType refTree(referenceData, leafSize, leafSize * 0.4, 5, 2, 0);
- Timer::Stop("tree_building");
- Log::Info << "Tree built." << endl;
-
- typedef NeighborSearch<FurthestNeighborSort, EuclideanDistance, arma::mat,
- RStarTree> AllkFNType;
- AllkFNType allkfn(&refTree, singleMode);
-
- if (CLI::GetParam<string>("query_file") != "")
- {
- if (!singleMode)
- {
- Timer::Start("tree_building");
- TreeType queryTree(queryData, leafSize, leafSize * 0.4, 5, 2, 0);
- Timer::Stop("tree_building");
-
- Log::Info << "Computing " << k << " nearest neighbors..." << endl;
- allkfn.Search(&queryTree, k, neighbors, distances);
- }
- else
- {
- Log::Info << "Computing " << k << " nearest neighbors..." << endl;
- allkfn.Search(queryData, k, neighbors, distances);
- }
- }
- else
- {
- Log::Info << "Computing " << k << " nearest neighbors..." << endl;
- allkfn.Search(k, neighbors, distances);
- }
- Log::Info << "Neighbors computed." << endl;
+ const string outputModelFile = CLI::GetParam<string>("output_model_File");
+ data::Save(outputModelFile, "kfn_model", kfn);
}
-
- // Save output.
- data::Save(distancesFile, distances);
- data::Save(neighborsFile, neighbors);
}
More information about the mlpack-git
mailing list