Changeset 15842 for branches/2886_SymRegGrammarEnumeration/ExpressionClustering/Flann.cs

Timestamp:

03/15/18 10:41:20 (6 years ago)

Author:

gkronber

Message:

#2886: added clustering of functions and output of clusters, fixed bug in evaluation

File:

• branches/2886_SymRegGrammarEnumeration/ExpressionClustering/Flann.cs (modified) (3 diffs)

branches/2886_SymRegGrammarEnumeration/ExpressionClustering/Flann.cs

@@ -86 +86 @@
       public int checks;                /* how many leafs (features) to check in one search */
       public float cb_index;            /* cluster boundary index. Used when searching the kmeans tree */
-      public float eps; 
+      public float eps;
 
       /*  kdtree index parameters */
@@ -153 +153 @@
     [DllImport("flann-1.7.1.dll")]
     public static extern int flann_compute_cluster_centers(float[] dataset, int rows, int cols, int clusters, float[] result, ref FLANNParameters flann_params);
+
+    public static int FindClusters(List<double[]> dataset, out List<int> results, out List<double> distances, int nClusters) {
+      var _nRows = dataset.Count;
+      var _dists = new float[_nRows];
+      var _result = new int[_nRows];
+      var _dim = dataset.First().Length;
+      FLANNParameters p = DEFAULT_FLANN_PARAMETERS;
+      p.algorithm = flann_algorithm_t.FLANN_INDEX_LINEAR;
+      p.centers_init = flann_centers_init_t.FLANN_CENTERS_RANDOM;
+      p.target_precision = 0.9f;
+      p.log_level = flann_log_level_t.FLANN_LOG_INFO;
+      // copy training set
+      var _ds = new float[dataset.Count * _dim];
+      var i = 0;
+      foreach (var e in dataset) {
+        for (int d = 0; d < _dim; d++) {
+          _ds[i++] = (float)e[d];
+        }
+      }
+
+      flann_set_distance_type(flann_distance_t.FLANN_DIST_EUCLIDEAN, 0);
+
+      float[] centers = new float[nClusters * _dim];
+      int actualClusters = flann_compute_cluster_centers(_ds, rows: dataset.Count, cols: _dim, clusters: nClusters, result: centers, flann_params: ref p);
+
+      var res = flann_find_nearest_neighbors(centers, actualClusters, _dim, _ds, _nRows, _result, _dists, 1, ref p);
+
+
+      distances = _dists.Select(fi => (double)fi).ToList();
+      results = _result.ToList();
+      return res;
+    }
 
     public static int FindNearestNeighbours(List<double[]> dataset, List<double[]> queryset, out List<int> results, out List<double> distances, int nearestNeighbours = 3) {
@@ -176 +208 @@
       flann_set_distance_type(flann_distance_t.FLANN_DIST_EUCLIDEAN, 0);
 
-      int nClusters = 100;
-      float[] centers = new float[nClusters * _dim];
-      flann_compute_cluster_centers(_ds, rows: dataset.Count, cols: _dim, clusters: nClusters, result: centers, flann_params: ref p);
-
-      float speedup = -1.0f;
+      // int nClusters = 100;
+      // float[] centers = new float[nClusters * _dim];
+      // flann_compute_cluster_centers(_ds, rows: dataset.Count, cols: _dim, clusters: nClusters, result: centers, flann_params: ref p);
+
+
+      // for each point in the training set find the nearest cluster
+
+      // float speedup = -1.0f;
       // _ds must be a rows × cols matrix stored in row-major order (one feature on each row)                         
       //var index = flann_build_index(_ds, rows: dataset.Count, cols: _dim, speedup: ref speedup, flann_params: ref p);
-      
+
 
       // copy testset
       var _testset = new float[_tRows * _dim];
       i = 0;
-      for(int d = 0; d < _dim; d++) {
-        foreach(var e in queryset) {
+      for (int d = 0; d < _dim; d++) {
+        foreach (var e in queryset) {
           _testset[i++] = (float)e[d];
         }