Added code for rearranging / reinserting internal nodes.

Author: cmdevries

src/EMTree.cpp

@@ -88,8 +88,8 @@ void sigEMTreeCluster(vector<SVector<bool>*> &vectors) {
 
     // EMTree
     int depth = 3;
-    int iters = 2;
-    vector<int> nodeSizes = {100}; //{10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20};
+    int iters = 10;
+    vector<int> nodeSizes = {10}; //{10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20};
     for (int m : nodeSizes) {
         std::cout << "-------------------" << std::endl;
         EMTree<vecType, clustererType, distanceType, protoType> emt(m);
@@ -403,9 +403,9 @@ void journalPaperExperiments(vector<SVector<bool>*>& vectors) {
     typedef KMeans<vecType, seederType, distanceType, protoType> clustererType;
 
     // run TSVQ vs EM-tree convergence
-    if (false) {
+    if (true) {
         int depth = 3, m = 10;
-        int iterRange = 40; // test RMSE at 1 to maxiters iterations
+        int iterRange = 10; // test RMSE at 1 to maxiters iterations
 
         //TSVQ
         if (true) {
@@ -414,6 +414,7 @@ void journalPaperExperiments(vector<SVector<bool>*>& vectors) {
             vector<double> seconds;
             for (int maxiters = 1; maxiters <= iterRange; ++maxiters) {
                 boost::timer::auto_cpu_timer all;
+                srand(1234);
                 TSVQ<vecType, clustererType, distanceType, protoType> tsvq(m, depth, maxiters);
                 tsvq.cluster(vectors);
                 all.stop();
@@ -439,6 +440,7 @@ void journalPaperExperiments(vector<SVector<bool>*>& vectors) {
                     splits.push_back(m);
                 }
                 boost::timer::auto_cpu_timer all;
+                srand(1234);
                 EMTree<vecType, clustererType, distanceType, protoType> emt(m);
                 // seeding does first iteration
                 emt.seedSingleThreaded(vectors, splits);
@@ -489,7 +491,7 @@ void journalPaperExperiments(vector<SVector<bool>*>& vectors) {
         }
 
         //EM-tree
-        if (true) {
+        if (false) {
             int maxiters = 6;
             vector<double> rmse;
             vector<int> clusters;
@@ -616,7 +618,7 @@ int main(int argc, char** argv) {
         journalPaperExperiments(subset);
         //sigKTreeCluster(vectors);
         //sigTSVQCluster(vectors);
-        //sigEMTreeCluster(vectors);
+        //sigEMTreeCluster(subset);
         //testHistogram(vectors);
         //testMeanVersusNNSpeed(vectors);
         //testReadVectors();

src/EMTree.h

@@ -25,6 +25,7 @@ class EMTree {
     ProtoType _protoF;
 
     vector<T*> removed;
+    vector<Node<T>*> removedChildren;
 
     // Weights for prototype function (we don't have to use these)
     vector<int> weights;
@@ -206,6 +207,18 @@ class EMTree {
         removed.clear();
     }
 
+    void rearrangeInternal() {
+        for (int depth = 2; depth < getMaxLevelCount(); ++depth) {
+            removeDataInternal(_root, removed, removedChildren, depth);
+            for (int i = 0; i < removed.size(); i++) {
+                pushDownNoUpdateInternal(_root, removed[i], removedChildren[i], depth);
+            }
+            prune();
+            removed.clear();
+            removedChildren.clear();
+        }
+    }    
+
     int prune() {
         return prune(_root);
     }
@@ -416,6 +429,26 @@ void add(T *obj) {
             pushDownNoUpdate(nearestChild, vec);
         }
     }
+    
+    void pushDownNoUpdateInternal(Node<T> *n, T* key, Node<T>* child, int depth) {
+
+        //std::cout << "\n\tPushing down (no update) ...";
+
+        if (depth == 1) {
+            n->add(key, child); // Finished
+        } else { // It is an internal node.
+            // recurse via nearest neighbour cluster
+
+            vector<T*>& keys = n->getKeys();
+            vector<Node<T>*>& children = n->getChildren();
+
+            size_t nearest = nearestChild(key, keys);
+            Node<T> *nearestChild = children[nearest];
+
+            pushDownNoUpdateInternal(nearestChild, key, child, depth - 1);
+        }
+    }
+    
 
     /*
 SplitResult<T> pushDown(Node<T> *n, T *vec) {
@@ -591,7 +624,16 @@ SplitResult<T> pushDown(Node<T> *n, T *vec) {
             }
         }
     }
-
+    
+    void removeDataInternal(Node<T>* n, vector<T*>& keys, vector<Node<T>*>& children, int depth) {
+        if (depth == 1) {
+            n->removeData(keys, children);
+        } else {
+            for (Node<T>* child : n->getChildren()) {
+                removeDataInternal(child, keys, children, depth - 1);
+            }
+        }
+    }
 };